{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 3.9996846669294444, "eval_steps": 500, "global_step": 50736, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 7.883326763894364e-05, "grad_norm": 1.2909892797470093, "learning_rate": 0.0, "loss": 3.9869368076324463, "step": 1 }, { "epoch": 0.00015766653527788727, "grad_norm": 1.3863993883132935, "learning_rate": 2.5e-06, "loss": 3.9909722805023193, "step": 2 }, { "epoch": 0.0002364998029168309, "grad_norm": 1.4667969942092896, "learning_rate": 5e-06, "loss": 4.189632892608643, "step": 3 }, { "epoch": 0.00031533307055577455, "grad_norm": 1.4359663724899292, "learning_rate": 7.5e-06, "loss": 4.178802490234375, "step": 4 }, { "epoch": 0.00039416633819471815, "grad_norm": 1.4256248474121094, "learning_rate": 1e-05, "loss": 3.9096508026123047, "step": 5 }, { "epoch": 0.0004729996058336618, "grad_norm": 1.3901678323745728, "learning_rate": 1.25e-05, "loss": 3.9406092166900635, "step": 6 }, { "epoch": 0.0005518328734726055, "grad_norm": 1.378775715827942, "learning_rate": 1.5e-05, "loss": 3.9623005390167236, "step": 7 }, { "epoch": 0.0006306661411115491, "grad_norm": 1.276729941368103, "learning_rate": 1.7500000000000002e-05, "loss": 4.156357765197754, "step": 8 }, { "epoch": 0.0007094994087504927, "grad_norm": 1.2041534185409546, "learning_rate": 2e-05, "loss": 4.233007907867432, "step": 9 }, { "epoch": 0.0007883326763894363, "grad_norm": 1.2212412357330322, "learning_rate": 2.2499999999999998e-05, "loss": 3.996151924133301, "step": 10 }, { "epoch": 0.00086716594402838, "grad_norm": 1.3473256826400757, "learning_rate": 2.5e-05, "loss": 4.329695701599121, "step": 11 }, { "epoch": 0.0009459992116673236, "grad_norm": 1.1783545017242432, "learning_rate": 2.75e-05, "loss": 3.9792633056640625, "step": 12 }, { "epoch": 0.0010248324793062672, "grad_norm": 1.1675513982772827, "learning_rate": 3e-05, "loss": 3.984891414642334, "step": 13 }, { "epoch": 0.001103665746945211, "grad_norm": 1.5833405256271362, "learning_rate": 3.2500000000000004e-05, "loss": 4.024704456329346, "step": 14 }, { "epoch": 0.0011824990145841545, "grad_norm": 1.2529512643814087, "learning_rate": 3.5000000000000004e-05, "loss": 3.8372013568878174, "step": 15 }, { "epoch": 0.0012613322822230982, "grad_norm": 1.310129165649414, "learning_rate": 3.75e-05, "loss": 4.054625034332275, "step": 16 }, { "epoch": 0.0013401655498620417, "grad_norm": 1.9703868627548218, "learning_rate": 4e-05, "loss": 3.970207691192627, "step": 17 }, { "epoch": 0.0014189988175009854, "grad_norm": 1.57174551486969, "learning_rate": 4.25e-05, "loss": 3.873704433441162, "step": 18 }, { "epoch": 0.0014978320851399291, "grad_norm": 1.164918065071106, "learning_rate": 4.4999999999999996e-05, "loss": 3.981377601623535, "step": 19 }, { "epoch": 0.0015766653527788726, "grad_norm": 1.0862900018692017, "learning_rate": 4.75e-05, "loss": 3.799745559692383, "step": 20 }, { "epoch": 0.0016554986204178163, "grad_norm": 1.2039755582809448, "learning_rate": 5e-05, "loss": 3.958570957183838, "step": 21 }, { "epoch": 0.00173433188805676, "grad_norm": 0.98002690076828, "learning_rate": 5.25e-05, "loss": 4.017205715179443, "step": 22 }, { "epoch": 0.0018131651556957036, "grad_norm": 1.3787883520126343, "learning_rate": 5.5e-05, "loss": 3.8635618686676025, "step": 23 }, { "epoch": 0.0018919984233346473, "grad_norm": 1.0931342840194702, "learning_rate": 5.75e-05, "loss": 3.719726085662842, "step": 24 }, { "epoch": 0.001970831690973591, "grad_norm": 1.035198450088501, "learning_rate": 6e-05, "loss": 3.8367371559143066, "step": 25 }, { "epoch": 0.0020496649586125345, "grad_norm": 1.273508071899414, "learning_rate": 6.25e-05, "loss": 3.8539795875549316, "step": 26 }, { "epoch": 0.002128498226251478, "grad_norm": 1.0653773546218872, "learning_rate": 6.500000000000001e-05, "loss": 3.7471885681152344, "step": 27 }, { "epoch": 0.002207331493890422, "grad_norm": 1.2845909595489502, "learning_rate": 6.75e-05, "loss": 3.8054752349853516, "step": 28 }, { "epoch": 0.0022861647615293654, "grad_norm": 1.2351385354995728, "learning_rate": 7.000000000000001e-05, "loss": 3.714179039001465, "step": 29 }, { "epoch": 0.002364998029168309, "grad_norm": 1.2485263347625732, "learning_rate": 7.25e-05, "loss": 3.4024405479431152, "step": 30 }, { "epoch": 0.002443831296807253, "grad_norm": 1.122786045074463, "learning_rate": 7.5e-05, "loss": 3.608102321624756, "step": 31 }, { "epoch": 0.0025226645644461964, "grad_norm": 1.3537206649780273, "learning_rate": 7.75e-05, "loss": 3.6522512435913086, "step": 32 }, { "epoch": 0.00260149783208514, "grad_norm": 0.9483810663223267, "learning_rate": 8e-05, "loss": 3.834353446960449, "step": 33 }, { "epoch": 0.0026803310997240834, "grad_norm": 1.2498396635055542, "learning_rate": 8.25e-05, "loss": 3.8997678756713867, "step": 34 }, { "epoch": 0.0027591643673630273, "grad_norm": 1.304638147354126, "learning_rate": 8.5e-05, "loss": 3.6059463024139404, "step": 35 }, { "epoch": 0.002837997635001971, "grad_norm": 1.104451298713684, "learning_rate": 8.75e-05, "loss": 3.6989877223968506, "step": 36 }, { "epoch": 0.0029168309026409143, "grad_norm": 1.3763505220413208, "learning_rate": 8.999999999999999e-05, "loss": 4.0015058517456055, "step": 37 }, { "epoch": 0.0029956641702798582, "grad_norm": 1.0889540910720825, "learning_rate": 9.25e-05, "loss": 3.919745445251465, "step": 38 }, { "epoch": 0.0030744974379188017, "grad_norm": 1.083008885383606, "learning_rate": 9.5e-05, "loss": 3.7739315032958984, "step": 39 }, { "epoch": 0.0031533307055577452, "grad_norm": 1.0326296091079712, "learning_rate": 9.750000000000001e-05, "loss": 3.659557819366455, "step": 40 }, { "epoch": 0.003232163973196689, "grad_norm": 1.615552544593811, "learning_rate": 0.0001, "loss": 3.986082077026367, "step": 41 }, { "epoch": 0.0033109972408356327, "grad_norm": 1.0833969116210938, "learning_rate": 0.0001025, "loss": 3.6328365802764893, "step": 42 }, { "epoch": 0.003389830508474576, "grad_norm": 1.304260492324829, "learning_rate": 0.000105, "loss": 3.9040772914886475, "step": 43 }, { "epoch": 0.00346866377611352, "grad_norm": 1.4365555047988892, "learning_rate": 0.0001075, "loss": 3.4437508583068848, "step": 44 }, { "epoch": 0.0035474970437524636, "grad_norm": 1.3358991146087646, "learning_rate": 0.00011, "loss": 3.8140029907226562, "step": 45 }, { "epoch": 0.003626330311391407, "grad_norm": 1.0630435943603516, "learning_rate": 0.00011250000000000001, "loss": 3.594430923461914, "step": 46 }, { "epoch": 0.0037051635790303506, "grad_norm": 1.1728487014770508, "learning_rate": 0.000115, "loss": 3.619338035583496, "step": 47 }, { "epoch": 0.0037839968466692945, "grad_norm": 1.1206151247024536, "learning_rate": 0.0001175, "loss": 3.5354816913604736, "step": 48 }, { "epoch": 0.003862830114308238, "grad_norm": 1.000535011291504, "learning_rate": 0.00012, "loss": 3.5933759212493896, "step": 49 }, { "epoch": 0.003941663381947182, "grad_norm": 1.3078786134719849, "learning_rate": 0.0001225, "loss": 3.731964588165283, "step": 50 }, { "epoch": 0.0040204966495861255, "grad_norm": 1.34050714969635, "learning_rate": 0.000125, "loss": 3.473045825958252, "step": 51 }, { "epoch": 0.004099329917225069, "grad_norm": 1.1476699113845825, "learning_rate": 0.0001275, "loss": 3.452761173248291, "step": 52 }, { "epoch": 0.0041781631848640125, "grad_norm": 1.1733192205429077, "learning_rate": 0.00013000000000000002, "loss": 3.687279224395752, "step": 53 }, { "epoch": 0.004256996452502956, "grad_norm": 1.4254688024520874, "learning_rate": 0.00013250000000000002, "loss": 3.615694761276245, "step": 54 }, { "epoch": 0.0043358297201418995, "grad_norm": 1.0200062990188599, "learning_rate": 0.000135, "loss": 3.394174575805664, "step": 55 }, { "epoch": 0.004414662987780844, "grad_norm": 1.1072912216186523, "learning_rate": 0.0001375, "loss": 3.367204189300537, "step": 56 }, { "epoch": 0.004493496255419787, "grad_norm": 1.1724028587341309, "learning_rate": 0.00014000000000000001, "loss": 3.8871562480926514, "step": 57 }, { "epoch": 0.004572329523058731, "grad_norm": 1.1163767576217651, "learning_rate": 0.0001425, "loss": 3.899545192718506, "step": 58 }, { "epoch": 0.004651162790697674, "grad_norm": 1.1158010959625244, "learning_rate": 0.000145, "loss": 3.6541950702667236, "step": 59 }, { "epoch": 0.004729996058336618, "grad_norm": 1.1060293912887573, "learning_rate": 0.0001475, "loss": 3.3115267753601074, "step": 60 }, { "epoch": 0.004808829325975561, "grad_norm": 2.2628426551818848, "learning_rate": 0.00015, "loss": 3.4617648124694824, "step": 61 }, { "epoch": 0.004887662593614506, "grad_norm": 0.90595543384552, "learning_rate": 0.0001525, "loss": 3.794156789779663, "step": 62 }, { "epoch": 0.004966495861253449, "grad_norm": 1.1675336360931396, "learning_rate": 0.000155, "loss": 3.8264782428741455, "step": 63 }, { "epoch": 0.005045329128892393, "grad_norm": 1.3129647970199585, "learning_rate": 0.0001575, "loss": 3.51682186126709, "step": 64 }, { "epoch": 0.005124162396531336, "grad_norm": 1.1544660329818726, "learning_rate": 0.00016, "loss": 3.7952637672424316, "step": 65 }, { "epoch": 0.00520299566417028, "grad_norm": 1.066033124923706, "learning_rate": 0.00016250000000000002, "loss": 3.37300968170166, "step": 66 }, { "epoch": 0.005281828931809223, "grad_norm": 1.0589194297790527, "learning_rate": 0.000165, "loss": 3.4978339672088623, "step": 67 }, { "epoch": 0.005360662199448167, "grad_norm": 1.1001477241516113, "learning_rate": 0.0001675, "loss": 3.7671542167663574, "step": 68 }, { "epoch": 0.005439495467087111, "grad_norm": 1.3896911144256592, "learning_rate": 0.00017, "loss": 3.764815330505371, "step": 69 }, { "epoch": 0.005518328734726055, "grad_norm": 1.530462384223938, "learning_rate": 0.0001725, "loss": 3.4727628231048584, "step": 70 }, { "epoch": 0.005597162002364998, "grad_norm": 2.4215023517608643, "learning_rate": 0.000175, "loss": 3.7541236877441406, "step": 71 }, { "epoch": 0.005675995270003942, "grad_norm": 1.0200872421264648, "learning_rate": 0.0001775, "loss": 3.4105160236358643, "step": 72 }, { "epoch": 0.005754828537642885, "grad_norm": 1.5201122760772705, "learning_rate": 0.00017999999999999998, "loss": 3.37615966796875, "step": 73 }, { "epoch": 0.005833661805281829, "grad_norm": 1.0641369819641113, "learning_rate": 0.0001825, "loss": 3.580565929412842, "step": 74 }, { "epoch": 0.005912495072920773, "grad_norm": 1.01096510887146, "learning_rate": 0.000185, "loss": 3.487370491027832, "step": 75 }, { "epoch": 0.0059913283405597165, "grad_norm": 0.9361074566841125, "learning_rate": 0.0001875, "loss": 3.561272621154785, "step": 76 }, { "epoch": 0.00607016160819866, "grad_norm": 1.111008644104004, "learning_rate": 0.00019, "loss": 3.4263975620269775, "step": 77 }, { "epoch": 0.0061489948758376035, "grad_norm": 1.0749844312667847, "learning_rate": 0.00019250000000000002, "loss": 3.6591477394104004, "step": 78 }, { "epoch": 0.006227828143476547, "grad_norm": 1.2751796245574951, "learning_rate": 0.00019500000000000002, "loss": 3.478154420852661, "step": 79 }, { "epoch": 0.0063066614111154905, "grad_norm": 1.2468183040618896, "learning_rate": 0.0001975, "loss": 3.523364543914795, "step": 80 }, { "epoch": 0.006385494678754434, "grad_norm": 1.1697179079055786, "learning_rate": 0.0002, "loss": 3.445847988128662, "step": 81 }, { "epoch": 0.006464327946393378, "grad_norm": 2.212125062942505, "learning_rate": 0.00020250000000000002, "loss": 3.562182903289795, "step": 82 }, { "epoch": 0.006543161214032322, "grad_norm": 1.5526796579360962, "learning_rate": 0.000205, "loss": 3.2321901321411133, "step": 83 }, { "epoch": 0.006621994481671265, "grad_norm": 1.396242380142212, "learning_rate": 0.0002075, "loss": 3.514023780822754, "step": 84 }, { "epoch": 0.006700827749310209, "grad_norm": 1.202053427696228, "learning_rate": 0.00021, "loss": 3.436192750930786, "step": 85 }, { "epoch": 0.006779661016949152, "grad_norm": 1.1101396083831787, "learning_rate": 0.0002125, "loss": 3.7936761379241943, "step": 86 }, { "epoch": 0.006858494284588096, "grad_norm": 1.0608510971069336, "learning_rate": 0.000215, "loss": 3.6476383209228516, "step": 87 }, { "epoch": 0.00693732755222704, "grad_norm": 1.205814003944397, "learning_rate": 0.0002175, "loss": 3.6003289222717285, "step": 88 }, { "epoch": 0.007016160819865984, "grad_norm": 1.1523927450180054, "learning_rate": 0.00022, "loss": 3.7482104301452637, "step": 89 }, { "epoch": 0.007094994087504927, "grad_norm": 1.0030884742736816, "learning_rate": 0.00022250000000000001, "loss": 3.386308193206787, "step": 90 }, { "epoch": 0.007173827355143871, "grad_norm": 1.2629450559616089, "learning_rate": 0.00022500000000000002, "loss": 3.3911774158477783, "step": 91 }, { "epoch": 0.007252660622782814, "grad_norm": 1.1549445390701294, "learning_rate": 0.0002275, "loss": 3.335679531097412, "step": 92 }, { "epoch": 0.007331493890421758, "grad_norm": 1.1830787658691406, "learning_rate": 0.00023, "loss": 3.588998317718506, "step": 93 }, { "epoch": 0.007410327158060701, "grad_norm": 1.3316552639007568, "learning_rate": 0.0002325, "loss": 3.8189573287963867, "step": 94 }, { "epoch": 0.007489160425699646, "grad_norm": 1.1461470127105713, "learning_rate": 0.000235, "loss": 3.62258243560791, "step": 95 }, { "epoch": 0.007567993693338589, "grad_norm": 1.0605841875076294, "learning_rate": 0.0002375, "loss": 3.609536647796631, "step": 96 }, { "epoch": 0.007646826960977533, "grad_norm": 0.9710763692855835, "learning_rate": 0.00024, "loss": 3.7593331336975098, "step": 97 }, { "epoch": 0.007725660228616476, "grad_norm": 1.0779296159744263, "learning_rate": 0.00024249999999999999, "loss": 3.3635056018829346, "step": 98 }, { "epoch": 0.00780449349625542, "grad_norm": 1.1375508308410645, "learning_rate": 0.000245, "loss": 3.5905215740203857, "step": 99 }, { "epoch": 0.007883326763894364, "grad_norm": 1.319871425628662, "learning_rate": 0.0002475, "loss": 3.3687610626220703, "step": 100 }, { "epoch": 0.007962160031533307, "grad_norm": 1.5607830286026, "learning_rate": 0.00025, "loss": 3.3667688369750977, "step": 101 }, { "epoch": 0.008040993299172251, "grad_norm": 1.4140375852584839, "learning_rate": 0.0002525, "loss": 3.4200968742370605, "step": 102 }, { "epoch": 0.008119826566811194, "grad_norm": 1.2341362237930298, "learning_rate": 0.000255, "loss": 3.278961658477783, "step": 103 }, { "epoch": 0.008198659834450138, "grad_norm": 1.456739902496338, "learning_rate": 0.0002575, "loss": 3.7495884895324707, "step": 104 }, { "epoch": 0.008277493102089082, "grad_norm": 1.1423916816711426, "learning_rate": 0.00026000000000000003, "loss": 3.6319003105163574, "step": 105 }, { "epoch": 0.008356326369728025, "grad_norm": 1.310402274131775, "learning_rate": 0.00026250000000000004, "loss": 3.4967007637023926, "step": 106 }, { "epoch": 0.00843515963736697, "grad_norm": 1.0920919179916382, "learning_rate": 0.00026500000000000004, "loss": 3.5318260192871094, "step": 107 }, { "epoch": 0.008513992905005912, "grad_norm": 1.247283697128296, "learning_rate": 0.0002675, "loss": 3.6112823486328125, "step": 108 }, { "epoch": 0.008592826172644856, "grad_norm": 1.0105446577072144, "learning_rate": 0.00027, "loss": 3.454447031021118, "step": 109 }, { "epoch": 0.008671659440283799, "grad_norm": 1.1321001052856445, "learning_rate": 0.0002725, "loss": 3.6412012577056885, "step": 110 }, { "epoch": 0.008750492707922743, "grad_norm": 1.20563542842865, "learning_rate": 0.000275, "loss": 3.6129631996154785, "step": 111 }, { "epoch": 0.008829325975561688, "grad_norm": 1.0373088121414185, "learning_rate": 0.0002775, "loss": 3.4462108612060547, "step": 112 }, { "epoch": 0.00890815924320063, "grad_norm": 1.106246829032898, "learning_rate": 0.00028000000000000003, "loss": 3.184563159942627, "step": 113 }, { "epoch": 0.008986992510839575, "grad_norm": 1.2488908767700195, "learning_rate": 0.0002825, "loss": 3.292109966278076, "step": 114 }, { "epoch": 0.009065825778478517, "grad_norm": 1.1951904296875, "learning_rate": 0.000285, "loss": 3.5025501251220703, "step": 115 }, { "epoch": 0.009144659046117462, "grad_norm": 1.2820996046066284, "learning_rate": 0.0002875, "loss": 3.3829588890075684, "step": 116 }, { "epoch": 0.009223492313756404, "grad_norm": 1.1819391250610352, "learning_rate": 0.00029, "loss": 3.8084588050842285, "step": 117 }, { "epoch": 0.009302325581395349, "grad_norm": 1.3188508749008179, "learning_rate": 0.0002925, "loss": 3.5306808948516846, "step": 118 }, { "epoch": 0.009381158849034293, "grad_norm": 1.2339972257614136, "learning_rate": 0.000295, "loss": 3.6586480140686035, "step": 119 }, { "epoch": 0.009459992116673236, "grad_norm": 1.247235655784607, "learning_rate": 0.00029749999999999997, "loss": 3.6962318420410156, "step": 120 }, { "epoch": 0.00953882538431218, "grad_norm": 1.174729347229004, "learning_rate": 0.0003, "loss": 3.6624457836151123, "step": 121 }, { "epoch": 0.009617658651951123, "grad_norm": 1.087533712387085, "learning_rate": 0.0003025, "loss": 3.642052412033081, "step": 122 }, { "epoch": 0.009696491919590067, "grad_norm": 1.4550247192382812, "learning_rate": 0.000305, "loss": 3.536350965499878, "step": 123 }, { "epoch": 0.009775325187229011, "grad_norm": 2.2534549236297607, "learning_rate": 0.0003075, "loss": 3.426910400390625, "step": 124 }, { "epoch": 0.009854158454867954, "grad_norm": 1.3507254123687744, "learning_rate": 0.00031, "loss": 3.6407227516174316, "step": 125 }, { "epoch": 0.009932991722506898, "grad_norm": 1.1596534252166748, "learning_rate": 0.0003125, "loss": 3.462710380554199, "step": 126 }, { "epoch": 0.010011824990145841, "grad_norm": 1.2304898500442505, "learning_rate": 0.000315, "loss": 3.5708112716674805, "step": 127 }, { "epoch": 0.010090658257784785, "grad_norm": 1.1944444179534912, "learning_rate": 0.0003175, "loss": 3.371422052383423, "step": 128 }, { "epoch": 0.010169491525423728, "grad_norm": 1.0826646089553833, "learning_rate": 0.00032, "loss": 3.4359853267669678, "step": 129 }, { "epoch": 0.010248324793062672, "grad_norm": 1.662140130996704, "learning_rate": 0.00032250000000000003, "loss": 3.4368205070495605, "step": 130 }, { "epoch": 0.010327158060701617, "grad_norm": 1.3570330142974854, "learning_rate": 0.00032500000000000004, "loss": 3.3570237159729004, "step": 131 }, { "epoch": 0.01040599132834056, "grad_norm": 1.4127163887023926, "learning_rate": 0.00032750000000000005, "loss": 3.7564468383789062, "step": 132 }, { "epoch": 0.010484824595979504, "grad_norm": 1.1429404020309448, "learning_rate": 0.00033, "loss": 3.4063756465911865, "step": 133 }, { "epoch": 0.010563657863618446, "grad_norm": 1.2111010551452637, "learning_rate": 0.0003325, "loss": 3.433560371398926, "step": 134 }, { "epoch": 0.01064249113125739, "grad_norm": 1.280236840248108, "learning_rate": 0.000335, "loss": 3.406643867492676, "step": 135 }, { "epoch": 0.010721324398896333, "grad_norm": 1.2061959505081177, "learning_rate": 0.0003375, "loss": 3.329807758331299, "step": 136 }, { "epoch": 0.010800157666535278, "grad_norm": 1.1248111724853516, "learning_rate": 0.00034, "loss": 3.758631944656372, "step": 137 }, { "epoch": 0.010878990934174222, "grad_norm": 1.5204262733459473, "learning_rate": 0.00034250000000000003, "loss": 3.733189582824707, "step": 138 }, { "epoch": 0.010957824201813165, "grad_norm": 1.5611822605133057, "learning_rate": 0.000345, "loss": 3.4607648849487305, "step": 139 }, { "epoch": 0.01103665746945211, "grad_norm": 1.2811163663864136, "learning_rate": 0.0003475, "loss": 3.4818716049194336, "step": 140 }, { "epoch": 0.011115490737091052, "grad_norm": 1.2794173955917358, "learning_rate": 0.00035, "loss": 3.236426830291748, "step": 141 }, { "epoch": 0.011194324004729996, "grad_norm": 1.1741693019866943, "learning_rate": 0.0003525, "loss": 3.4416661262512207, "step": 142 }, { "epoch": 0.011273157272368939, "grad_norm": 1.3229241371154785, "learning_rate": 0.000355, "loss": 3.1270976066589355, "step": 143 }, { "epoch": 0.011351990540007883, "grad_norm": 1.194624900817871, "learning_rate": 0.0003575, "loss": 3.389202117919922, "step": 144 }, { "epoch": 0.011430823807646828, "grad_norm": 1.461098074913025, "learning_rate": 0.00035999999999999997, "loss": 3.2543914318084717, "step": 145 }, { "epoch": 0.01150965707528577, "grad_norm": 1.0703110694885254, "learning_rate": 0.0003625, "loss": 3.619788646697998, "step": 146 }, { "epoch": 0.011588490342924715, "grad_norm": 1.3262732028961182, "learning_rate": 0.000365, "loss": 3.106940269470215, "step": 147 }, { "epoch": 0.011667323610563657, "grad_norm": 1.237250804901123, "learning_rate": 0.0003675, "loss": 3.3565428256988525, "step": 148 }, { "epoch": 0.011746156878202602, "grad_norm": 1.1411231756210327, "learning_rate": 0.00037, "loss": 3.281879425048828, "step": 149 }, { "epoch": 0.011824990145841546, "grad_norm": 1.7046416997909546, "learning_rate": 0.0003725, "loss": 3.626103639602661, "step": 150 }, { "epoch": 0.011903823413480489, "grad_norm": 1.4786092042922974, "learning_rate": 0.000375, "loss": 3.5849802494049072, "step": 151 }, { "epoch": 0.011982656681119433, "grad_norm": 1.43594229221344, "learning_rate": 0.0003775, "loss": 3.8870186805725098, "step": 152 }, { "epoch": 0.012061489948758376, "grad_norm": 1.3139466047286987, "learning_rate": 0.00038, "loss": 3.3361353874206543, "step": 153 }, { "epoch": 0.01214032321639732, "grad_norm": 1.3537219762802124, "learning_rate": 0.00038250000000000003, "loss": 3.2331924438476562, "step": 154 }, { "epoch": 0.012219156484036263, "grad_norm": 1.184922218322754, "learning_rate": 0.00038500000000000003, "loss": 3.338902473449707, "step": 155 }, { "epoch": 0.012297989751675207, "grad_norm": 1.1843746900558472, "learning_rate": 0.00038750000000000004, "loss": 3.6676719188690186, "step": 156 }, { "epoch": 0.012376823019314151, "grad_norm": 1.110541820526123, "learning_rate": 0.00039000000000000005, "loss": 3.4757885932922363, "step": 157 }, { "epoch": 0.012455656286953094, "grad_norm": 1.461689829826355, "learning_rate": 0.0003925, "loss": 3.3666374683380127, "step": 158 }, { "epoch": 0.012534489554592038, "grad_norm": 1.1768327951431274, "learning_rate": 0.000395, "loss": 3.2368950843811035, "step": 159 }, { "epoch": 0.012613322822230981, "grad_norm": 1.364524006843567, "learning_rate": 0.0003975, "loss": 3.401289939880371, "step": 160 }, { "epoch": 0.012692156089869925, "grad_norm": 1.627847671508789, "learning_rate": 0.0004, "loss": 3.415161609649658, "step": 161 }, { "epoch": 0.012770989357508868, "grad_norm": 1.4825611114501953, "learning_rate": 0.0004025, "loss": 3.3412959575653076, "step": 162 }, { "epoch": 0.012849822625147812, "grad_norm": 1.1053380966186523, "learning_rate": 0.00040500000000000003, "loss": 3.5283262729644775, "step": 163 }, { "epoch": 0.012928655892786757, "grad_norm": 1.27605402469635, "learning_rate": 0.0004075, "loss": 3.532351016998291, "step": 164 }, { "epoch": 0.0130074891604257, "grad_norm": 1.2050412893295288, "learning_rate": 0.00041, "loss": 3.288633346557617, "step": 165 }, { "epoch": 0.013086322428064644, "grad_norm": 1.7435516119003296, "learning_rate": 0.0004125, "loss": 3.6211702823638916, "step": 166 }, { "epoch": 0.013165155695703586, "grad_norm": 1.3285155296325684, "learning_rate": 0.000415, "loss": 3.4822988510131836, "step": 167 }, { "epoch": 0.01324398896334253, "grad_norm": 1.0227549076080322, "learning_rate": 0.0004175, "loss": 3.4554615020751953, "step": 168 }, { "epoch": 0.013322822230981473, "grad_norm": 1.2771247625350952, "learning_rate": 0.00042, "loss": 3.552824020385742, "step": 169 }, { "epoch": 0.013401655498620418, "grad_norm": 1.1757771968841553, "learning_rate": 0.00042249999999999997, "loss": 3.2933852672576904, "step": 170 }, { "epoch": 0.013480488766259362, "grad_norm": 1.521738052368164, "learning_rate": 0.000425, "loss": 3.0136866569519043, "step": 171 }, { "epoch": 0.013559322033898305, "grad_norm": 1.488407015800476, "learning_rate": 0.0004275, "loss": 3.542475461959839, "step": 172 }, { "epoch": 0.013638155301537249, "grad_norm": 1.3312878608703613, "learning_rate": 0.00043, "loss": 3.1669998168945312, "step": 173 }, { "epoch": 0.013716988569176192, "grad_norm": 1.3878165483474731, "learning_rate": 0.0004325, "loss": 3.512331008911133, "step": 174 }, { "epoch": 0.013795821836815136, "grad_norm": 1.4363000392913818, "learning_rate": 0.000435, "loss": 3.6571438312530518, "step": 175 }, { "epoch": 0.01387465510445408, "grad_norm": 1.111497163772583, "learning_rate": 0.0004375, "loss": 3.500743865966797, "step": 176 }, { "epoch": 0.013953488372093023, "grad_norm": 1.0865037441253662, "learning_rate": 0.00044, "loss": 3.5135891437530518, "step": 177 }, { "epoch": 0.014032321639731967, "grad_norm": 1.4883099794387817, "learning_rate": 0.0004425, "loss": 3.1925530433654785, "step": 178 }, { "epoch": 0.01411115490737091, "grad_norm": 1.1942843198776245, "learning_rate": 0.00044500000000000003, "loss": 3.6066787242889404, "step": 179 }, { "epoch": 0.014189988175009854, "grad_norm": 1.156009316444397, "learning_rate": 0.00044750000000000004, "loss": 3.757991075515747, "step": 180 }, { "epoch": 0.014268821442648797, "grad_norm": 1.4009106159210205, "learning_rate": 0.00045000000000000004, "loss": 3.782160758972168, "step": 181 }, { "epoch": 0.014347654710287741, "grad_norm": 1.183206558227539, "learning_rate": 0.00045250000000000005, "loss": 3.1034622192382812, "step": 182 }, { "epoch": 0.014426487977926686, "grad_norm": 1.700610637664795, "learning_rate": 0.000455, "loss": 3.5950839519500732, "step": 183 }, { "epoch": 0.014505321245565628, "grad_norm": 1.2727079391479492, "learning_rate": 0.0004575, "loss": 3.673121452331543, "step": 184 }, { "epoch": 0.014584154513204573, "grad_norm": 1.3901525735855103, "learning_rate": 0.00046, "loss": 3.4256832599639893, "step": 185 }, { "epoch": 0.014662987780843515, "grad_norm": 1.3669480085372925, "learning_rate": 0.0004625, "loss": 3.8397409915924072, "step": 186 }, { "epoch": 0.01474182104848246, "grad_norm": 1.0236057043075562, "learning_rate": 0.000465, "loss": 3.8087282180786133, "step": 187 }, { "epoch": 0.014820654316121402, "grad_norm": 1.2547005414962769, "learning_rate": 0.00046750000000000003, "loss": 3.5526657104492188, "step": 188 }, { "epoch": 0.014899487583760347, "grad_norm": 1.1484493017196655, "learning_rate": 0.00047, "loss": 3.3630409240722656, "step": 189 }, { "epoch": 0.014978320851399291, "grad_norm": 1.2742552757263184, "learning_rate": 0.0004725, "loss": 3.4254164695739746, "step": 190 }, { "epoch": 0.015057154119038234, "grad_norm": 1.1838701963424683, "learning_rate": 0.000475, "loss": 3.7360355854034424, "step": 191 }, { "epoch": 0.015135987386677178, "grad_norm": 1.1551541090011597, "learning_rate": 0.0004775, "loss": 3.3967628479003906, "step": 192 }, { "epoch": 0.01521482065431612, "grad_norm": 1.3200480937957764, "learning_rate": 0.00048, "loss": 3.6513140201568604, "step": 193 }, { "epoch": 0.015293653921955065, "grad_norm": 1.3395854234695435, "learning_rate": 0.0004825, "loss": 3.8185977935791016, "step": 194 }, { "epoch": 0.015372487189594008, "grad_norm": 1.2479968070983887, "learning_rate": 0.00048499999999999997, "loss": 3.3903427124023438, "step": 195 }, { "epoch": 0.015451320457232952, "grad_norm": 1.2912198305130005, "learning_rate": 0.0004875, "loss": 3.4519152641296387, "step": 196 }, { "epoch": 0.015530153724871897, "grad_norm": 1.2837636470794678, "learning_rate": 0.00049, "loss": 3.696533679962158, "step": 197 }, { "epoch": 0.01560898699251084, "grad_norm": 1.5734386444091797, "learning_rate": 0.0004925, "loss": 3.340850591659546, "step": 198 }, { "epoch": 0.015687820260149782, "grad_norm": 1.2642221450805664, "learning_rate": 0.000495, "loss": 3.536836624145508, "step": 199 }, { "epoch": 0.015766653527788728, "grad_norm": 1.2482258081436157, "learning_rate": 0.0004975, "loss": 3.653921365737915, "step": 200 }, { "epoch": 0.01584548679542767, "grad_norm": 1.2682796716690063, "learning_rate": 0.0005, "loss": 3.6844587326049805, "step": 201 }, { "epoch": 0.015924320063066613, "grad_norm": 1.3594529628753662, "learning_rate": 0.0005, "loss": 3.6621241569519043, "step": 202 }, { "epoch": 0.01600315333070556, "grad_norm": 1.5811306238174438, "learning_rate": 0.0005, "loss": 3.245663642883301, "step": 203 }, { "epoch": 0.016081986598344502, "grad_norm": 1.3806195259094238, "learning_rate": 0.0005, "loss": 3.6963558197021484, "step": 204 }, { "epoch": 0.016160819865983445, "grad_norm": 1.349738597869873, "learning_rate": 0.0005, "loss": 3.5512802600860596, "step": 205 }, { "epoch": 0.016239653133622387, "grad_norm": 1.3612133264541626, "learning_rate": 0.0005, "loss": 3.45772123336792, "step": 206 }, { "epoch": 0.016318486401261333, "grad_norm": 1.3281705379486084, "learning_rate": 0.0005, "loss": 3.8212053775787354, "step": 207 }, { "epoch": 0.016397319668900276, "grad_norm": 1.5297672748565674, "learning_rate": 0.0005, "loss": 3.7062621116638184, "step": 208 }, { "epoch": 0.01647615293653922, "grad_norm": 1.391929268836975, "learning_rate": 0.0005, "loss": 3.2686517238616943, "step": 209 }, { "epoch": 0.016554986204178165, "grad_norm": 1.3146581649780273, "learning_rate": 0.0005, "loss": 4.005670547485352, "step": 210 }, { "epoch": 0.016633819471817107, "grad_norm": 1.278515100479126, "learning_rate": 0.0005, "loss": 3.37650990486145, "step": 211 }, { "epoch": 0.01671265273945605, "grad_norm": 1.5327140092849731, "learning_rate": 0.0005, "loss": 3.6881566047668457, "step": 212 }, { "epoch": 0.016791486007094993, "grad_norm": 1.465321660041809, "learning_rate": 0.0005, "loss": 3.3701698780059814, "step": 213 }, { "epoch": 0.01687031927473394, "grad_norm": 1.4118788242340088, "learning_rate": 0.0005, "loss": 3.452083110809326, "step": 214 }, { "epoch": 0.01694915254237288, "grad_norm": 1.2083232402801514, "learning_rate": 0.0005, "loss": 3.155740976333618, "step": 215 }, { "epoch": 0.017027985810011824, "grad_norm": 1.1926029920578003, "learning_rate": 0.0005, "loss": 3.3127808570861816, "step": 216 }, { "epoch": 0.01710681907765077, "grad_norm": 1.4807523488998413, "learning_rate": 0.0005, "loss": 3.8441073894500732, "step": 217 }, { "epoch": 0.017185652345289713, "grad_norm": 1.1789565086364746, "learning_rate": 0.0005, "loss": 3.408766269683838, "step": 218 }, { "epoch": 0.017264485612928655, "grad_norm": 1.1306606531143188, "learning_rate": 0.0005, "loss": 3.5224335193634033, "step": 219 }, { "epoch": 0.017343318880567598, "grad_norm": 1.4825248718261719, "learning_rate": 0.0005, "loss": 3.5410966873168945, "step": 220 }, { "epoch": 0.017422152148206544, "grad_norm": 1.4291383028030396, "learning_rate": 0.0005, "loss": 3.5274109840393066, "step": 221 }, { "epoch": 0.017500985415845487, "grad_norm": 1.6152513027191162, "learning_rate": 0.0005, "loss": 3.3067402839660645, "step": 222 }, { "epoch": 0.01757981868348443, "grad_norm": 1.4174582958221436, "learning_rate": 0.0005, "loss": 3.7738540172576904, "step": 223 }, { "epoch": 0.017658651951123375, "grad_norm": 1.1659225225448608, "learning_rate": 0.0005, "loss": 3.5388455390930176, "step": 224 }, { "epoch": 0.017737485218762318, "grad_norm": 1.440248727798462, "learning_rate": 0.0005, "loss": 3.3516433238983154, "step": 225 }, { "epoch": 0.01781631848640126, "grad_norm": 1.085769534111023, "learning_rate": 0.0005, "loss": 3.4208250045776367, "step": 226 }, { "epoch": 0.017895151754040203, "grad_norm": 1.2284679412841797, "learning_rate": 0.0005, "loss": 3.38759446144104, "step": 227 }, { "epoch": 0.01797398502167915, "grad_norm": 1.3823429346084595, "learning_rate": 0.0005, "loss": 3.529212474822998, "step": 228 }, { "epoch": 0.018052818289318092, "grad_norm": 2.075593948364258, "learning_rate": 0.0005, "loss": 3.6648125648498535, "step": 229 }, { "epoch": 0.018131651556957035, "grad_norm": 1.4033081531524658, "learning_rate": 0.0005, "loss": 3.367980480194092, "step": 230 }, { "epoch": 0.01821048482459598, "grad_norm": 1.752728819847107, "learning_rate": 0.0005, "loss": 3.6396470069885254, "step": 231 }, { "epoch": 0.018289318092234923, "grad_norm": 1.405388355255127, "learning_rate": 0.0005, "loss": 3.3189382553100586, "step": 232 }, { "epoch": 0.018368151359873866, "grad_norm": 1.1434738636016846, "learning_rate": 0.0005, "loss": 3.2853097915649414, "step": 233 }, { "epoch": 0.01844698462751281, "grad_norm": 1.6942532062530518, "learning_rate": 0.0005, "loss": 3.4409053325653076, "step": 234 }, { "epoch": 0.018525817895151755, "grad_norm": 1.2037899494171143, "learning_rate": 0.0005, "loss": 3.675380229949951, "step": 235 }, { "epoch": 0.018604651162790697, "grad_norm": 1.1243778467178345, "learning_rate": 0.0005, "loss": 3.590416669845581, "step": 236 }, { "epoch": 0.01868348443042964, "grad_norm": 1.0956485271453857, "learning_rate": 0.0005, "loss": 3.6363446712493896, "step": 237 }, { "epoch": 0.018762317698068586, "grad_norm": 1.4219588041305542, "learning_rate": 0.0005, "loss": 3.3461356163024902, "step": 238 }, { "epoch": 0.01884115096570753, "grad_norm": 1.2514739036560059, "learning_rate": 0.0005, "loss": 3.6878790855407715, "step": 239 }, { "epoch": 0.01891998423334647, "grad_norm": 1.0611553192138672, "learning_rate": 0.0005, "loss": 3.2454798221588135, "step": 240 }, { "epoch": 0.018998817500985418, "grad_norm": 1.2131600379943848, "learning_rate": 0.0005, "loss": 3.8220415115356445, "step": 241 }, { "epoch": 0.01907765076862436, "grad_norm": 1.4715874195098877, "learning_rate": 0.0005, "loss": 3.356499195098877, "step": 242 }, { "epoch": 0.019156484036263303, "grad_norm": 1.1965163946151733, "learning_rate": 0.0005, "loss": 3.3654680252075195, "step": 243 }, { "epoch": 0.019235317303902245, "grad_norm": 1.3859612941741943, "learning_rate": 0.0005, "loss": 3.337109088897705, "step": 244 }, { "epoch": 0.01931415057154119, "grad_norm": 1.4118480682373047, "learning_rate": 0.0005, "loss": 3.5113589763641357, "step": 245 }, { "epoch": 0.019392983839180134, "grad_norm": 1.904449701309204, "learning_rate": 0.0005, "loss": 3.521303415298462, "step": 246 }, { "epoch": 0.019471817106819077, "grad_norm": 1.2362396717071533, "learning_rate": 0.0005, "loss": 3.4035887718200684, "step": 247 }, { "epoch": 0.019550650374458023, "grad_norm": 1.479959487915039, "learning_rate": 0.0005, "loss": 3.095900535583496, "step": 248 }, { "epoch": 0.019629483642096966, "grad_norm": 1.1332054138183594, "learning_rate": 0.0005, "loss": 3.735197067260742, "step": 249 }, { "epoch": 0.019708316909735908, "grad_norm": 1.6566818952560425, "learning_rate": 0.0005, "loss": 3.6398515701293945, "step": 250 }, { "epoch": 0.01978715017737485, "grad_norm": 1.384452223777771, "learning_rate": 0.0005, "loss": 3.5554871559143066, "step": 251 }, { "epoch": 0.019865983445013797, "grad_norm": 1.9464489221572876, "learning_rate": 0.0005, "loss": 3.6155757904052734, "step": 252 }, { "epoch": 0.01994481671265274, "grad_norm": 1.3589204549789429, "learning_rate": 0.0005, "loss": 3.478684902191162, "step": 253 }, { "epoch": 0.020023649980291682, "grad_norm": 1.4145393371582031, "learning_rate": 0.0005, "loss": 3.2130606174468994, "step": 254 }, { "epoch": 0.02010248324793063, "grad_norm": 1.1960809230804443, "learning_rate": 0.0005, "loss": 3.540515422821045, "step": 255 }, { "epoch": 0.02018131651556957, "grad_norm": 1.1260367631912231, "learning_rate": 0.0005, "loss": 3.4193739891052246, "step": 256 }, { "epoch": 0.020260149783208514, "grad_norm": 1.1910934448242188, "learning_rate": 0.0005, "loss": 3.3379034996032715, "step": 257 }, { "epoch": 0.020338983050847456, "grad_norm": 1.2581212520599365, "learning_rate": 0.0005, "loss": 3.315595865249634, "step": 258 }, { "epoch": 0.020417816318486402, "grad_norm": 1.20155668258667, "learning_rate": 0.0005, "loss": 3.5569677352905273, "step": 259 }, { "epoch": 0.020496649586125345, "grad_norm": 1.151601791381836, "learning_rate": 0.0005, "loss": 3.5518693923950195, "step": 260 }, { "epoch": 0.020575482853764288, "grad_norm": 1.647775650024414, "learning_rate": 0.0005, "loss": 3.248203754425049, "step": 261 }, { "epoch": 0.020654316121403234, "grad_norm": 1.105270266532898, "learning_rate": 0.0005, "loss": 3.5662479400634766, "step": 262 }, { "epoch": 0.020733149389042176, "grad_norm": 1.1931555271148682, "learning_rate": 0.0005, "loss": 3.295649766921997, "step": 263 }, { "epoch": 0.02081198265668112, "grad_norm": 1.3942872285842896, "learning_rate": 0.0005, "loss": 3.70898175239563, "step": 264 }, { "epoch": 0.02089081592432006, "grad_norm": 1.8976624011993408, "learning_rate": 0.0005, "loss": 3.5300207138061523, "step": 265 }, { "epoch": 0.020969649191959008, "grad_norm": 1.348743200302124, "learning_rate": 0.0005, "loss": 3.3142361640930176, "step": 266 }, { "epoch": 0.02104848245959795, "grad_norm": 1.1862903833389282, "learning_rate": 0.0005, "loss": 3.526604413986206, "step": 267 }, { "epoch": 0.021127315727236893, "grad_norm": 1.2549365758895874, "learning_rate": 0.0005, "loss": 3.131690502166748, "step": 268 }, { "epoch": 0.02120614899487584, "grad_norm": 1.1820684671401978, "learning_rate": 0.0005, "loss": 3.6596996784210205, "step": 269 }, { "epoch": 0.02128498226251478, "grad_norm": 1.4724177122116089, "learning_rate": 0.0005, "loss": 3.305631399154663, "step": 270 }, { "epoch": 0.021363815530153724, "grad_norm": 1.3606109619140625, "learning_rate": 0.0005, "loss": 3.645080327987671, "step": 271 }, { "epoch": 0.021442648797792667, "grad_norm": 1.2325420379638672, "learning_rate": 0.0005, "loss": 3.8180437088012695, "step": 272 }, { "epoch": 0.021521482065431613, "grad_norm": 1.51113760471344, "learning_rate": 0.0005, "loss": 3.4486680030822754, "step": 273 }, { "epoch": 0.021600315333070556, "grad_norm": 1.1688392162322998, "learning_rate": 0.0005, "loss": 3.4090685844421387, "step": 274 }, { "epoch": 0.0216791486007095, "grad_norm": 1.068723440170288, "learning_rate": 0.0005, "loss": 3.863377809524536, "step": 275 }, { "epoch": 0.021757981868348444, "grad_norm": 1.1663106679916382, "learning_rate": 0.0005, "loss": 3.514554262161255, "step": 276 }, { "epoch": 0.021836815135987387, "grad_norm": 1.0072635412216187, "learning_rate": 0.0005, "loss": 3.6970701217651367, "step": 277 }, { "epoch": 0.02191564840362633, "grad_norm": 1.1624835729599, "learning_rate": 0.0005, "loss": 3.6668035984039307, "step": 278 }, { "epoch": 0.021994481671265272, "grad_norm": 1.224307894706726, "learning_rate": 0.0005, "loss": 3.6083998680114746, "step": 279 }, { "epoch": 0.02207331493890422, "grad_norm": 1.154725432395935, "learning_rate": 0.0005, "loss": 3.180368661880493, "step": 280 }, { "epoch": 0.02215214820654316, "grad_norm": 1.2179332971572876, "learning_rate": 0.0005, "loss": 3.382807970046997, "step": 281 }, { "epoch": 0.022230981474182104, "grad_norm": 1.1685839891433716, "learning_rate": 0.0005, "loss": 3.51487398147583, "step": 282 }, { "epoch": 0.02230981474182105, "grad_norm": 1.0267949104309082, "learning_rate": 0.0005, "loss": 3.1964406967163086, "step": 283 }, { "epoch": 0.022388648009459992, "grad_norm": 1.203006625175476, "learning_rate": 0.0005, "loss": 3.925713062286377, "step": 284 }, { "epoch": 0.022467481277098935, "grad_norm": 1.2236450910568237, "learning_rate": 0.0005, "loss": 3.115907669067383, "step": 285 }, { "epoch": 0.022546314544737878, "grad_norm": 1.2093987464904785, "learning_rate": 0.0005, "loss": 3.3921611309051514, "step": 286 }, { "epoch": 0.022625147812376824, "grad_norm": 1.0623291730880737, "learning_rate": 0.0005, "loss": 3.311068534851074, "step": 287 }, { "epoch": 0.022703981080015766, "grad_norm": 1.587517499923706, "learning_rate": 0.0005, "loss": 3.954165458679199, "step": 288 }, { "epoch": 0.02278281434765471, "grad_norm": 1.3286954164505005, "learning_rate": 0.0005, "loss": 3.4680023193359375, "step": 289 }, { "epoch": 0.022861647615293655, "grad_norm": 1.3520833253860474, "learning_rate": 0.0005, "loss": 3.8281235694885254, "step": 290 }, { "epoch": 0.022940480882932598, "grad_norm": 1.5785372257232666, "learning_rate": 0.0005, "loss": 3.47432541847229, "step": 291 }, { "epoch": 0.02301931415057154, "grad_norm": 1.6044633388519287, "learning_rate": 0.0005, "loss": 3.415471076965332, "step": 292 }, { "epoch": 0.023098147418210487, "grad_norm": 1.0853415727615356, "learning_rate": 0.0005, "loss": 3.516406536102295, "step": 293 }, { "epoch": 0.02317698068584943, "grad_norm": 1.3272557258605957, "learning_rate": 0.0005, "loss": 3.439403533935547, "step": 294 }, { "epoch": 0.023255813953488372, "grad_norm": 1.4184393882751465, "learning_rate": 0.0005, "loss": 3.546769857406616, "step": 295 }, { "epoch": 0.023334647221127314, "grad_norm": 1.4928053617477417, "learning_rate": 0.0005, "loss": 3.19215989112854, "step": 296 }, { "epoch": 0.02341348048876626, "grad_norm": 1.2886881828308105, "learning_rate": 0.0005, "loss": 3.430595874786377, "step": 297 }, { "epoch": 0.023492313756405203, "grad_norm": 1.4033921957015991, "learning_rate": 0.0005, "loss": 3.4842286109924316, "step": 298 }, { "epoch": 0.023571147024044146, "grad_norm": 1.183129906654358, "learning_rate": 0.0005, "loss": 3.571565628051758, "step": 299 }, { "epoch": 0.023649980291683092, "grad_norm": 1.5585912466049194, "learning_rate": 0.0005, "loss": 3.2015113830566406, "step": 300 }, { "epoch": 0.023728813559322035, "grad_norm": 0.9858217239379883, "learning_rate": 0.0005, "loss": 3.238356113433838, "step": 301 }, { "epoch": 0.023807646826960977, "grad_norm": 1.2393280267715454, "learning_rate": 0.0005, "loss": 3.210329294204712, "step": 302 }, { "epoch": 0.02388648009459992, "grad_norm": 1.2091089487075806, "learning_rate": 0.0005, "loss": 3.528986930847168, "step": 303 }, { "epoch": 0.023965313362238866, "grad_norm": 1.3089312314987183, "learning_rate": 0.0005, "loss": 3.365126371383667, "step": 304 }, { "epoch": 0.02404414662987781, "grad_norm": 1.5748966932296753, "learning_rate": 0.0005, "loss": 3.4917752742767334, "step": 305 }, { "epoch": 0.02412297989751675, "grad_norm": 1.3009737730026245, "learning_rate": 0.0005, "loss": 3.1314613819122314, "step": 306 }, { "epoch": 0.024201813165155697, "grad_norm": 1.0799438953399658, "learning_rate": 0.0005, "loss": 3.1833887100219727, "step": 307 }, { "epoch": 0.02428064643279464, "grad_norm": 1.2267457246780396, "learning_rate": 0.0005, "loss": 3.2530221939086914, "step": 308 }, { "epoch": 0.024359479700433583, "grad_norm": 1.2957037687301636, "learning_rate": 0.0005, "loss": 3.641814947128296, "step": 309 }, { "epoch": 0.024438312968072525, "grad_norm": 1.4257382154464722, "learning_rate": 0.0005, "loss": 3.6004672050476074, "step": 310 }, { "epoch": 0.02451714623571147, "grad_norm": 1.2300944328308105, "learning_rate": 0.0005, "loss": 3.824723243713379, "step": 311 }, { "epoch": 0.024595979503350414, "grad_norm": 1.2981353998184204, "learning_rate": 0.0005, "loss": 3.8369643688201904, "step": 312 }, { "epoch": 0.024674812770989357, "grad_norm": 1.3959726095199585, "learning_rate": 0.0005, "loss": 3.3043160438537598, "step": 313 }, { "epoch": 0.024753646038628303, "grad_norm": 2.3571395874023438, "learning_rate": 0.0005, "loss": 3.9934091567993164, "step": 314 }, { "epoch": 0.024832479306267245, "grad_norm": 1.2937111854553223, "learning_rate": 0.0005, "loss": 3.3966479301452637, "step": 315 }, { "epoch": 0.024911312573906188, "grad_norm": 1.1619800329208374, "learning_rate": 0.0005, "loss": 3.179720401763916, "step": 316 }, { "epoch": 0.02499014584154513, "grad_norm": 1.4487673044204712, "learning_rate": 0.0005, "loss": 3.5258374214172363, "step": 317 }, { "epoch": 0.025068979109184077, "grad_norm": 2.0019845962524414, "learning_rate": 0.0005, "loss": 3.408052921295166, "step": 318 }, { "epoch": 0.02514781237682302, "grad_norm": 1.286126732826233, "learning_rate": 0.0005, "loss": 3.5516605377197266, "step": 319 }, { "epoch": 0.025226645644461962, "grad_norm": 1.3143222332000732, "learning_rate": 0.0005, "loss": 3.694654703140259, "step": 320 }, { "epoch": 0.025305478912100908, "grad_norm": 1.673476219177246, "learning_rate": 0.0005, "loss": 3.5084383487701416, "step": 321 }, { "epoch": 0.02538431217973985, "grad_norm": 1.1377421617507935, "learning_rate": 0.0005, "loss": 3.615642547607422, "step": 322 }, { "epoch": 0.025463145447378793, "grad_norm": 1.5703206062316895, "learning_rate": 0.0005, "loss": 3.125418186187744, "step": 323 }, { "epoch": 0.025541978715017736, "grad_norm": 1.49063241481781, "learning_rate": 0.0005, "loss": 3.3448269367218018, "step": 324 }, { "epoch": 0.025620811982656682, "grad_norm": 1.2850857973098755, "learning_rate": 0.0005, "loss": 3.756791591644287, "step": 325 }, { "epoch": 0.025699645250295625, "grad_norm": 1.1900948286056519, "learning_rate": 0.0005, "loss": 3.122161865234375, "step": 326 }, { "epoch": 0.025778478517934567, "grad_norm": 1.534620761871338, "learning_rate": 0.0005, "loss": 3.458559274673462, "step": 327 }, { "epoch": 0.025857311785573513, "grad_norm": 1.217961072921753, "learning_rate": 0.0005, "loss": 3.5828657150268555, "step": 328 }, { "epoch": 0.025936145053212456, "grad_norm": 1.286255955696106, "learning_rate": 0.0005, "loss": 3.577082633972168, "step": 329 }, { "epoch": 0.0260149783208514, "grad_norm": 1.683424711227417, "learning_rate": 0.0005, "loss": 3.679690361022949, "step": 330 }, { "epoch": 0.02609381158849034, "grad_norm": 1.1375775337219238, "learning_rate": 0.0005, "loss": 3.3639066219329834, "step": 331 }, { "epoch": 0.026172644856129287, "grad_norm": 1.1834063529968262, "learning_rate": 0.0005, "loss": 3.569040536880493, "step": 332 }, { "epoch": 0.02625147812376823, "grad_norm": 1.445609211921692, "learning_rate": 0.0005, "loss": 3.5806798934936523, "step": 333 }, { "epoch": 0.026330311391407173, "grad_norm": 1.1964612007141113, "learning_rate": 0.0005, "loss": 3.4567246437072754, "step": 334 }, { "epoch": 0.02640914465904612, "grad_norm": 1.297935962677002, "learning_rate": 0.0005, "loss": 3.3632819652557373, "step": 335 }, { "epoch": 0.02648797792668506, "grad_norm": 1.3599333763122559, "learning_rate": 0.0005, "loss": 3.6123573780059814, "step": 336 }, { "epoch": 0.026566811194324004, "grad_norm": 1.1556501388549805, "learning_rate": 0.0005, "loss": 3.203195571899414, "step": 337 }, { "epoch": 0.026645644461962947, "grad_norm": 1.1639399528503418, "learning_rate": 0.0005, "loss": 3.4784891605377197, "step": 338 }, { "epoch": 0.026724477729601893, "grad_norm": 1.122413158416748, "learning_rate": 0.0005, "loss": 3.6085588932037354, "step": 339 }, { "epoch": 0.026803310997240835, "grad_norm": 1.147020936012268, "learning_rate": 0.0005, "loss": 3.8933568000793457, "step": 340 }, { "epoch": 0.026882144264879778, "grad_norm": 1.4959994554519653, "learning_rate": 0.0005, "loss": 3.715550422668457, "step": 341 }, { "epoch": 0.026960977532518724, "grad_norm": 1.4114956855773926, "learning_rate": 0.0005, "loss": 3.3332018852233887, "step": 342 }, { "epoch": 0.027039810800157667, "grad_norm": 1.2677092552185059, "learning_rate": 0.0005, "loss": 3.432974338531494, "step": 343 }, { "epoch": 0.02711864406779661, "grad_norm": 1.3483601808547974, "learning_rate": 0.0005, "loss": 3.3076164722442627, "step": 344 }, { "epoch": 0.027197477335435555, "grad_norm": 1.1506880521774292, "learning_rate": 0.0005, "loss": 3.7002758979797363, "step": 345 }, { "epoch": 0.027276310603074498, "grad_norm": 1.2017452716827393, "learning_rate": 0.0005, "loss": 3.3708207607269287, "step": 346 }, { "epoch": 0.02735514387071344, "grad_norm": 1.1788091659545898, "learning_rate": 0.0005, "loss": 3.4931657314300537, "step": 347 }, { "epoch": 0.027433977138352383, "grad_norm": 1.148366093635559, "learning_rate": 0.0005, "loss": 3.554046154022217, "step": 348 }, { "epoch": 0.02751281040599133, "grad_norm": 1.019975185394287, "learning_rate": 0.0005, "loss": 3.64616322517395, "step": 349 }, { "epoch": 0.027591643673630272, "grad_norm": 1.0739103555679321, "learning_rate": 0.0005, "loss": 3.2941017150878906, "step": 350 }, { "epoch": 0.027670476941269215, "grad_norm": 1.2092961072921753, "learning_rate": 0.0005, "loss": 2.8138275146484375, "step": 351 }, { "epoch": 0.02774931020890816, "grad_norm": 1.3703436851501465, "learning_rate": 0.0005, "loss": 3.4714436531066895, "step": 352 }, { "epoch": 0.027828143476547103, "grad_norm": 1.9658493995666504, "learning_rate": 0.0005, "loss": 3.5743374824523926, "step": 353 }, { "epoch": 0.027906976744186046, "grad_norm": 1.359309196472168, "learning_rate": 0.0005, "loss": 3.5717720985412598, "step": 354 }, { "epoch": 0.02798581001182499, "grad_norm": 1.2710325717926025, "learning_rate": 0.0005, "loss": 3.4169936180114746, "step": 355 }, { "epoch": 0.028064643279463935, "grad_norm": 1.5259253978729248, "learning_rate": 0.0005, "loss": 4.0657196044921875, "step": 356 }, { "epoch": 0.028143476547102877, "grad_norm": 1.4584317207336426, "learning_rate": 0.0005, "loss": 3.3397796154022217, "step": 357 }, { "epoch": 0.02822230981474182, "grad_norm": 1.284140944480896, "learning_rate": 0.0005, "loss": 3.155153751373291, "step": 358 }, { "epoch": 0.028301143082380766, "grad_norm": 1.0659091472625732, "learning_rate": 0.0005, "loss": 3.718365430831909, "step": 359 }, { "epoch": 0.02837997635001971, "grad_norm": 1.093007206916809, "learning_rate": 0.0005, "loss": 3.7117624282836914, "step": 360 }, { "epoch": 0.02845880961765865, "grad_norm": 1.7704641819000244, "learning_rate": 0.0005, "loss": 3.380169630050659, "step": 361 }, { "epoch": 0.028537642885297594, "grad_norm": 1.288331389427185, "learning_rate": 0.0005, "loss": 3.1053590774536133, "step": 362 }, { "epoch": 0.02861647615293654, "grad_norm": 1.3033514022827148, "learning_rate": 0.0005, "loss": 3.292210340499878, "step": 363 }, { "epoch": 0.028695309420575483, "grad_norm": 1.3344403505325317, "learning_rate": 0.0005, "loss": 3.3632290363311768, "step": 364 }, { "epoch": 0.028774142688214426, "grad_norm": 1.0968647003173828, "learning_rate": 0.0005, "loss": 3.258512020111084, "step": 365 }, { "epoch": 0.02885297595585337, "grad_norm": 1.174559473991394, "learning_rate": 0.0005, "loss": 3.6476593017578125, "step": 366 }, { "epoch": 0.028931809223492314, "grad_norm": 1.415581464767456, "learning_rate": 0.0005, "loss": 3.214468479156494, "step": 367 }, { "epoch": 0.029010642491131257, "grad_norm": 1.1866048574447632, "learning_rate": 0.0005, "loss": 3.4111881256103516, "step": 368 }, { "epoch": 0.0290894757587702, "grad_norm": 1.5665682554244995, "learning_rate": 0.0005, "loss": 3.376084804534912, "step": 369 }, { "epoch": 0.029168309026409146, "grad_norm": 1.3469092845916748, "learning_rate": 0.0005, "loss": 3.6000561714172363, "step": 370 }, { "epoch": 0.029247142294048088, "grad_norm": 2.8538570404052734, "learning_rate": 0.0005, "loss": 3.153810977935791, "step": 371 }, { "epoch": 0.02932597556168703, "grad_norm": 1.7493319511413574, "learning_rate": 0.0005, "loss": 3.852130889892578, "step": 372 }, { "epoch": 0.029404808829325977, "grad_norm": 1.4020522832870483, "learning_rate": 0.0005, "loss": 3.2911972999572754, "step": 373 }, { "epoch": 0.02948364209696492, "grad_norm": 1.2578651905059814, "learning_rate": 0.0005, "loss": 3.2023234367370605, "step": 374 }, { "epoch": 0.029562475364603862, "grad_norm": 1.3243436813354492, "learning_rate": 0.0005, "loss": 3.202449083328247, "step": 375 }, { "epoch": 0.029641308632242805, "grad_norm": 1.0396242141723633, "learning_rate": 0.0005, "loss": 3.397672176361084, "step": 376 }, { "epoch": 0.02972014189988175, "grad_norm": 1.4032177925109863, "learning_rate": 0.0005, "loss": 3.5673880577087402, "step": 377 }, { "epoch": 0.029798975167520694, "grad_norm": 1.36704421043396, "learning_rate": 0.0005, "loss": 3.4293782711029053, "step": 378 }, { "epoch": 0.029877808435159636, "grad_norm": 1.0609087944030762, "learning_rate": 0.0005, "loss": 3.4158501625061035, "step": 379 }, { "epoch": 0.029956641702798582, "grad_norm": 1.4512747526168823, "learning_rate": 0.0005, "loss": 3.1199443340301514, "step": 380 }, { "epoch": 0.030035474970437525, "grad_norm": 1.6618527173995972, "learning_rate": 0.0005, "loss": 3.5112247467041016, "step": 381 }, { "epoch": 0.030114308238076468, "grad_norm": 0.9831987619400024, "learning_rate": 0.0005, "loss": 3.137388229370117, "step": 382 }, { "epoch": 0.03019314150571541, "grad_norm": 1.2237682342529297, "learning_rate": 0.0005, "loss": 3.5751399993896484, "step": 383 }, { "epoch": 0.030271974773354356, "grad_norm": 1.1547911167144775, "learning_rate": 0.0005, "loss": 3.53130841255188, "step": 384 }, { "epoch": 0.0303508080409933, "grad_norm": 1.213491439819336, "learning_rate": 0.0005, "loss": 3.3491077423095703, "step": 385 }, { "epoch": 0.03042964130863224, "grad_norm": 1.2206385135650635, "learning_rate": 0.0005, "loss": 3.9102981090545654, "step": 386 }, { "epoch": 0.030508474576271188, "grad_norm": 1.2368849515914917, "learning_rate": 0.0005, "loss": 3.223499059677124, "step": 387 }, { "epoch": 0.03058730784391013, "grad_norm": 1.164564847946167, "learning_rate": 0.0005, "loss": 3.4701924324035645, "step": 388 }, { "epoch": 0.030666141111549073, "grad_norm": 1.2496337890625, "learning_rate": 0.0005, "loss": 3.522728443145752, "step": 389 }, { "epoch": 0.030744974379188016, "grad_norm": 1.1772087812423706, "learning_rate": 0.0005, "loss": 3.854550361633301, "step": 390 }, { "epoch": 0.03082380764682696, "grad_norm": 1.1452761888504028, "learning_rate": 0.0005, "loss": 3.4108967781066895, "step": 391 }, { "epoch": 0.030902640914465904, "grad_norm": 1.2091646194458008, "learning_rate": 0.0005, "loss": 3.1865596771240234, "step": 392 }, { "epoch": 0.030981474182104847, "grad_norm": 3.4727842807769775, "learning_rate": 0.0005, "loss": 3.558445453643799, "step": 393 }, { "epoch": 0.031060307449743793, "grad_norm": 1.158614158630371, "learning_rate": 0.0005, "loss": 3.463336944580078, "step": 394 }, { "epoch": 0.031139140717382736, "grad_norm": 1.2275310754776, "learning_rate": 0.0005, "loss": 3.3009114265441895, "step": 395 }, { "epoch": 0.03121797398502168, "grad_norm": 1.3312945365905762, "learning_rate": 0.0005, "loss": 2.9750733375549316, "step": 396 }, { "epoch": 0.031296807252660624, "grad_norm": 1.4253312349319458, "learning_rate": 0.0005, "loss": 3.317875862121582, "step": 397 }, { "epoch": 0.031375640520299564, "grad_norm": 1.2103908061981201, "learning_rate": 0.0005, "loss": 3.0903165340423584, "step": 398 }, { "epoch": 0.03145447378793851, "grad_norm": 1.610581636428833, "learning_rate": 0.0005, "loss": 2.8229947090148926, "step": 399 }, { "epoch": 0.031533307055577456, "grad_norm": 1.3847304582595825, "learning_rate": 0.0005, "loss": 3.5227837562561035, "step": 400 }, { "epoch": 0.031612140323216395, "grad_norm": 1.292847990989685, "learning_rate": 0.0005, "loss": 3.408275842666626, "step": 401 }, { "epoch": 0.03169097359085534, "grad_norm": 1.4252188205718994, "learning_rate": 0.0005, "loss": 3.3934412002563477, "step": 402 }, { "epoch": 0.03176980685849429, "grad_norm": 1.1216627359390259, "learning_rate": 0.0005, "loss": 2.968881607055664, "step": 403 }, { "epoch": 0.031848640126133226, "grad_norm": 1.1981056928634644, "learning_rate": 0.0005, "loss": 3.3417890071868896, "step": 404 }, { "epoch": 0.03192747339377217, "grad_norm": 1.0883185863494873, "learning_rate": 0.0005, "loss": 3.597031354904175, "step": 405 }, { "epoch": 0.03200630666141112, "grad_norm": 1.5217002630233765, "learning_rate": 0.0005, "loss": 3.565206289291382, "step": 406 }, { "epoch": 0.03208513992905006, "grad_norm": 1.2107211351394653, "learning_rate": 0.0005, "loss": 3.296560049057007, "step": 407 }, { "epoch": 0.032163973196689004, "grad_norm": 1.9717310667037964, "learning_rate": 0.0005, "loss": 3.499238967895508, "step": 408 }, { "epoch": 0.03224280646432794, "grad_norm": 2.2455050945281982, "learning_rate": 0.0005, "loss": 4.033890724182129, "step": 409 }, { "epoch": 0.03232163973196689, "grad_norm": 1.2446428537368774, "learning_rate": 0.0005, "loss": 3.4361326694488525, "step": 410 }, { "epoch": 0.032400472999605835, "grad_norm": 1.3062326908111572, "learning_rate": 0.0005, "loss": 3.4824960231781006, "step": 411 }, { "epoch": 0.032479306267244774, "grad_norm": 1.1465837955474854, "learning_rate": 0.0005, "loss": 3.331012725830078, "step": 412 }, { "epoch": 0.03255813953488372, "grad_norm": 1.0356196165084839, "learning_rate": 0.0005, "loss": 3.5794122219085693, "step": 413 }, { "epoch": 0.03263697280252267, "grad_norm": 1.0926153659820557, "learning_rate": 0.0005, "loss": 3.306978225708008, "step": 414 }, { "epoch": 0.032715806070161606, "grad_norm": 1.3285598754882812, "learning_rate": 0.0005, "loss": 3.5350372791290283, "step": 415 }, { "epoch": 0.03279463933780055, "grad_norm": 1.19300377368927, "learning_rate": 0.0005, "loss": 3.450086832046509, "step": 416 }, { "epoch": 0.0328734726054395, "grad_norm": 1.184708595275879, "learning_rate": 0.0005, "loss": 3.3872761726379395, "step": 417 }, { "epoch": 0.03295230587307844, "grad_norm": 1.2526798248291016, "learning_rate": 0.0005, "loss": 3.4365017414093018, "step": 418 }, { "epoch": 0.03303113914071738, "grad_norm": 1.1656689643859863, "learning_rate": 0.0005, "loss": 3.5658397674560547, "step": 419 }, { "epoch": 0.03310997240835633, "grad_norm": 1.0918724536895752, "learning_rate": 0.0005, "loss": 3.269357204437256, "step": 420 }, { "epoch": 0.03318880567599527, "grad_norm": 1.0859140157699585, "learning_rate": 0.0005, "loss": 3.3682072162628174, "step": 421 }, { "epoch": 0.033267638943634215, "grad_norm": 1.2509801387786865, "learning_rate": 0.0005, "loss": 3.3243730068206787, "step": 422 }, { "epoch": 0.03334647221127316, "grad_norm": 1.2508201599121094, "learning_rate": 0.0005, "loss": 3.5596933364868164, "step": 423 }, { "epoch": 0.0334253054789121, "grad_norm": 1.1539727449417114, "learning_rate": 0.0005, "loss": 3.3665332794189453, "step": 424 }, { "epoch": 0.033504138746551046, "grad_norm": 1.009252667427063, "learning_rate": 0.0005, "loss": 3.5386710166931152, "step": 425 }, { "epoch": 0.033582972014189985, "grad_norm": 1.0465344190597534, "learning_rate": 0.0005, "loss": 3.588407039642334, "step": 426 }, { "epoch": 0.03366180528182893, "grad_norm": 1.206168532371521, "learning_rate": 0.0005, "loss": 3.1672215461730957, "step": 427 }, { "epoch": 0.03374063854946788, "grad_norm": 1.2749841213226318, "learning_rate": 0.0005, "loss": 3.6611063480377197, "step": 428 }, { "epoch": 0.033819471817106816, "grad_norm": 1.2840204238891602, "learning_rate": 0.0005, "loss": 3.287320375442505, "step": 429 }, { "epoch": 0.03389830508474576, "grad_norm": 1.6099038124084473, "learning_rate": 0.0005, "loss": 3.39963960647583, "step": 430 }, { "epoch": 0.03397713835238471, "grad_norm": 1.1014355421066284, "learning_rate": 0.0005, "loss": 3.464049816131592, "step": 431 }, { "epoch": 0.03405597162002365, "grad_norm": 1.1098253726959229, "learning_rate": 0.0005, "loss": 3.38871431350708, "step": 432 }, { "epoch": 0.034134804887662594, "grad_norm": 1.4866275787353516, "learning_rate": 0.0005, "loss": 2.9545774459838867, "step": 433 }, { "epoch": 0.03421363815530154, "grad_norm": 1.1363483667373657, "learning_rate": 0.0005, "loss": 3.287071943283081, "step": 434 }, { "epoch": 0.03429247142294048, "grad_norm": 1.4835340976715088, "learning_rate": 0.0005, "loss": 3.2804360389709473, "step": 435 }, { "epoch": 0.034371304690579425, "grad_norm": 1.8433127403259277, "learning_rate": 0.0005, "loss": 3.33974289894104, "step": 436 }, { "epoch": 0.03445013795821837, "grad_norm": 1.33366060256958, "learning_rate": 0.0005, "loss": 3.2904481887817383, "step": 437 }, { "epoch": 0.03452897122585731, "grad_norm": 1.314056396484375, "learning_rate": 0.0005, "loss": 3.568887710571289, "step": 438 }, { "epoch": 0.03460780449349626, "grad_norm": 1.112310528755188, "learning_rate": 0.0005, "loss": 3.5068368911743164, "step": 439 }, { "epoch": 0.034686637761135196, "grad_norm": 1.3208391666412354, "learning_rate": 0.0005, "loss": 3.2219765186309814, "step": 440 }, { "epoch": 0.03476547102877414, "grad_norm": 1.2339404821395874, "learning_rate": 0.0005, "loss": 3.336988687515259, "step": 441 }, { "epoch": 0.03484430429641309, "grad_norm": 1.1641814708709717, "learning_rate": 0.0005, "loss": 3.275390625, "step": 442 }, { "epoch": 0.03492313756405203, "grad_norm": 1.3783477544784546, "learning_rate": 0.0005, "loss": 2.9249372482299805, "step": 443 }, { "epoch": 0.03500197083169097, "grad_norm": 1.3049006462097168, "learning_rate": 0.0005, "loss": 3.408329963684082, "step": 444 }, { "epoch": 0.03508080409932992, "grad_norm": 1.459725022315979, "learning_rate": 0.0005, "loss": 3.0917434692382812, "step": 445 }, { "epoch": 0.03515963736696886, "grad_norm": 1.0872540473937988, "learning_rate": 0.0005, "loss": 3.1775941848754883, "step": 446 }, { "epoch": 0.035238470634607805, "grad_norm": 1.2403359413146973, "learning_rate": 0.0005, "loss": 3.55122447013855, "step": 447 }, { "epoch": 0.03531730390224675, "grad_norm": 1.9231274127960205, "learning_rate": 0.0005, "loss": 3.4428863525390625, "step": 448 }, { "epoch": 0.03539613716988569, "grad_norm": 1.2275351285934448, "learning_rate": 0.0005, "loss": 3.076843023300171, "step": 449 }, { "epoch": 0.035474970437524636, "grad_norm": 1.291759967803955, "learning_rate": 0.0005, "loss": 3.5014381408691406, "step": 450 }, { "epoch": 0.03555380370516358, "grad_norm": 1.142245888710022, "learning_rate": 0.0005, "loss": 3.1777565479278564, "step": 451 }, { "epoch": 0.03563263697280252, "grad_norm": 1.426732063293457, "learning_rate": 0.0005, "loss": 3.144364356994629, "step": 452 }, { "epoch": 0.03571147024044147, "grad_norm": 1.3230739831924438, "learning_rate": 0.0005, "loss": 3.6016581058502197, "step": 453 }, { "epoch": 0.03579030350808041, "grad_norm": 1.1856499910354614, "learning_rate": 0.0005, "loss": 3.4254722595214844, "step": 454 }, { "epoch": 0.03586913677571935, "grad_norm": 1.0319135189056396, "learning_rate": 0.0005, "loss": 3.7783870697021484, "step": 455 }, { "epoch": 0.0359479700433583, "grad_norm": 1.0962822437286377, "learning_rate": 0.0005, "loss": 3.55379581451416, "step": 456 }, { "epoch": 0.03602680331099724, "grad_norm": 1.2576961517333984, "learning_rate": 0.0005, "loss": 3.3338770866394043, "step": 457 }, { "epoch": 0.036105636578636184, "grad_norm": 1.150810718536377, "learning_rate": 0.0005, "loss": 2.8909709453582764, "step": 458 }, { "epoch": 0.03618446984627513, "grad_norm": 1.4726202487945557, "learning_rate": 0.0005, "loss": 3.2704415321350098, "step": 459 }, { "epoch": 0.03626330311391407, "grad_norm": 1.2674211263656616, "learning_rate": 0.0005, "loss": 2.8898117542266846, "step": 460 }, { "epoch": 0.036342136381553015, "grad_norm": 1.0455418825149536, "learning_rate": 0.0005, "loss": 3.429502010345459, "step": 461 }, { "epoch": 0.03642096964919196, "grad_norm": 1.1845945119857788, "learning_rate": 0.0005, "loss": 3.539653778076172, "step": 462 }, { "epoch": 0.0364998029168309, "grad_norm": 1.2785578966140747, "learning_rate": 0.0005, "loss": 3.4828219413757324, "step": 463 }, { "epoch": 0.03657863618446985, "grad_norm": 3.3522987365722656, "learning_rate": 0.0005, "loss": 3.0017409324645996, "step": 464 }, { "epoch": 0.03665746945210879, "grad_norm": 1.1567721366882324, "learning_rate": 0.0005, "loss": 3.4062345027923584, "step": 465 }, { "epoch": 0.03673630271974773, "grad_norm": 1.3818527460098267, "learning_rate": 0.0005, "loss": 3.193614959716797, "step": 466 }, { "epoch": 0.03681513598738668, "grad_norm": 1.1275148391723633, "learning_rate": 0.0005, "loss": 3.547682762145996, "step": 467 }, { "epoch": 0.03689396925502562, "grad_norm": 1.9875763654708862, "learning_rate": 0.0005, "loss": 3.4177680015563965, "step": 468 }, { "epoch": 0.03697280252266456, "grad_norm": 1.4111087322235107, "learning_rate": 0.0005, "loss": 3.639385223388672, "step": 469 }, { "epoch": 0.03705163579030351, "grad_norm": 1.2753928899765015, "learning_rate": 0.0005, "loss": 3.640918731689453, "step": 470 }, { "epoch": 0.03713046905794245, "grad_norm": 1.2177393436431885, "learning_rate": 0.0005, "loss": 3.5221166610717773, "step": 471 }, { "epoch": 0.037209302325581395, "grad_norm": 1.105290174484253, "learning_rate": 0.0005, "loss": 3.030684471130371, "step": 472 }, { "epoch": 0.03728813559322034, "grad_norm": 1.3219718933105469, "learning_rate": 0.0005, "loss": 3.2669363021850586, "step": 473 }, { "epoch": 0.03736696886085928, "grad_norm": 2.6876938343048096, "learning_rate": 0.0005, "loss": 3.8205320835113525, "step": 474 }, { "epoch": 0.037445802128498226, "grad_norm": 1.8725392818450928, "learning_rate": 0.0005, "loss": 3.695891857147217, "step": 475 }, { "epoch": 0.03752463539613717, "grad_norm": 1.4101009368896484, "learning_rate": 0.0005, "loss": 3.4269449710845947, "step": 476 }, { "epoch": 0.03760346866377611, "grad_norm": 1.5436583757400513, "learning_rate": 0.0005, "loss": 2.8514175415039062, "step": 477 }, { "epoch": 0.03768230193141506, "grad_norm": 1.4508545398712158, "learning_rate": 0.0005, "loss": 3.6544268131256104, "step": 478 }, { "epoch": 0.037761135199054004, "grad_norm": 1.2842340469360352, "learning_rate": 0.0005, "loss": 3.536195755004883, "step": 479 }, { "epoch": 0.03783996846669294, "grad_norm": 2.201385498046875, "learning_rate": 0.0005, "loss": 3.543320655822754, "step": 480 }, { "epoch": 0.03791880173433189, "grad_norm": 1.559867262840271, "learning_rate": 0.0005, "loss": 3.120911121368408, "step": 481 }, { "epoch": 0.037997635001970835, "grad_norm": 1.308229923248291, "learning_rate": 0.0005, "loss": 3.619053602218628, "step": 482 }, { "epoch": 0.038076468269609774, "grad_norm": 1.1576933860778809, "learning_rate": 0.0005, "loss": 3.462218761444092, "step": 483 }, { "epoch": 0.03815530153724872, "grad_norm": 1.1318837404251099, "learning_rate": 0.0005, "loss": 3.4422810077667236, "step": 484 }, { "epoch": 0.03823413480488766, "grad_norm": 1.460220456123352, "learning_rate": 0.0005, "loss": 3.235718250274658, "step": 485 }, { "epoch": 0.038312968072526606, "grad_norm": 1.1556713581085205, "learning_rate": 0.0005, "loss": 3.3063340187072754, "step": 486 }, { "epoch": 0.03839180134016555, "grad_norm": 1.374734878540039, "learning_rate": 0.0005, "loss": 3.2719361782073975, "step": 487 }, { "epoch": 0.03847063460780449, "grad_norm": 1.302331805229187, "learning_rate": 0.0005, "loss": 3.5875163078308105, "step": 488 }, { "epoch": 0.03854946787544344, "grad_norm": 1.0745741128921509, "learning_rate": 0.0005, "loss": 3.359961748123169, "step": 489 }, { "epoch": 0.03862830114308238, "grad_norm": 1.0819042921066284, "learning_rate": 0.0005, "loss": 3.4435911178588867, "step": 490 }, { "epoch": 0.03870713441072132, "grad_norm": 1.2546721696853638, "learning_rate": 0.0005, "loss": 3.1480398178100586, "step": 491 }, { "epoch": 0.03878596767836027, "grad_norm": 1.27914559841156, "learning_rate": 0.0005, "loss": 3.588994026184082, "step": 492 }, { "epoch": 0.038864800945999214, "grad_norm": 1.1493897438049316, "learning_rate": 0.0005, "loss": 3.379101276397705, "step": 493 }, { "epoch": 0.038943634213638154, "grad_norm": 1.308172345161438, "learning_rate": 0.0005, "loss": 3.5348639488220215, "step": 494 }, { "epoch": 0.0390224674812771, "grad_norm": 1.2708184719085693, "learning_rate": 0.0005, "loss": 3.2783408164978027, "step": 495 }, { "epoch": 0.039101300748916046, "grad_norm": 1.2805733680725098, "learning_rate": 0.0005, "loss": 3.4284768104553223, "step": 496 }, { "epoch": 0.039180134016554985, "grad_norm": 1.3623733520507812, "learning_rate": 0.0005, "loss": 3.289419174194336, "step": 497 }, { "epoch": 0.03925896728419393, "grad_norm": 1.6790339946746826, "learning_rate": 0.0005, "loss": 3.2636876106262207, "step": 498 }, { "epoch": 0.03933780055183287, "grad_norm": 1.2092804908752441, "learning_rate": 0.0005, "loss": 3.3524904251098633, "step": 499 }, { "epoch": 0.039416633819471816, "grad_norm": 1.6130203008651733, "learning_rate": 0.0005, "loss": 3.17680025100708, "step": 500 }, { "epoch": 0.03949546708711076, "grad_norm": 1.1075069904327393, "learning_rate": 0.0005, "loss": 3.2065720558166504, "step": 501 }, { "epoch": 0.0395743003547497, "grad_norm": 1.309438705444336, "learning_rate": 0.0005, "loss": 3.2139928340911865, "step": 502 }, { "epoch": 0.03965313362238865, "grad_norm": 1.2194377183914185, "learning_rate": 0.0005, "loss": 3.5084028244018555, "step": 503 }, { "epoch": 0.039731966890027594, "grad_norm": 1.0720722675323486, "learning_rate": 0.0005, "loss": 3.6836800575256348, "step": 504 }, { "epoch": 0.03981080015766653, "grad_norm": 1.1673866510391235, "learning_rate": 0.0005, "loss": 3.405930995941162, "step": 505 }, { "epoch": 0.03988963342530548, "grad_norm": 1.3269739151000977, "learning_rate": 0.0005, "loss": 3.52073335647583, "step": 506 }, { "epoch": 0.039968466692944425, "grad_norm": 1.36161470413208, "learning_rate": 0.0005, "loss": 3.21189284324646, "step": 507 }, { "epoch": 0.040047299960583364, "grad_norm": 1.1456378698349, "learning_rate": 0.0005, "loss": 3.090243339538574, "step": 508 }, { "epoch": 0.04012613322822231, "grad_norm": 1.3395060300827026, "learning_rate": 0.0005, "loss": 3.5032482147216797, "step": 509 }, { "epoch": 0.04020496649586126, "grad_norm": 1.2203764915466309, "learning_rate": 0.0005, "loss": 3.2335848808288574, "step": 510 }, { "epoch": 0.040283799763500196, "grad_norm": 1.2103214263916016, "learning_rate": 0.0005, "loss": 3.330411195755005, "step": 511 }, { "epoch": 0.04036263303113914, "grad_norm": 1.4127817153930664, "learning_rate": 0.0005, "loss": 3.6802818775177, "step": 512 }, { "epoch": 0.04044146629877808, "grad_norm": 1.1737494468688965, "learning_rate": 0.0005, "loss": 3.6347904205322266, "step": 513 }, { "epoch": 0.04052029956641703, "grad_norm": 1.4023840427398682, "learning_rate": 0.0005, "loss": 3.4132137298583984, "step": 514 }, { "epoch": 0.04059913283405597, "grad_norm": 1.224610447883606, "learning_rate": 0.0005, "loss": 3.477529764175415, "step": 515 }, { "epoch": 0.04067796610169491, "grad_norm": 1.2315587997436523, "learning_rate": 0.0005, "loss": 3.1733131408691406, "step": 516 }, { "epoch": 0.04075679936933386, "grad_norm": 1.2826964855194092, "learning_rate": 0.0005, "loss": 3.2041573524475098, "step": 517 }, { "epoch": 0.040835632636972805, "grad_norm": 1.545418620109558, "learning_rate": 0.0005, "loss": 3.4795405864715576, "step": 518 }, { "epoch": 0.040914465904611744, "grad_norm": 1.126641035079956, "learning_rate": 0.0005, "loss": 2.9980533123016357, "step": 519 }, { "epoch": 0.04099329917225069, "grad_norm": 1.156040906906128, "learning_rate": 0.0005, "loss": 3.576308250427246, "step": 520 }, { "epoch": 0.041072132439889636, "grad_norm": 1.1742125749588013, "learning_rate": 0.0005, "loss": 3.3765053749084473, "step": 521 }, { "epoch": 0.041150965707528575, "grad_norm": 1.001039743423462, "learning_rate": 0.0005, "loss": 3.411174774169922, "step": 522 }, { "epoch": 0.04122979897516752, "grad_norm": 1.352500081062317, "learning_rate": 0.0005, "loss": 3.297353744506836, "step": 523 }, { "epoch": 0.04130863224280647, "grad_norm": 1.3233195543289185, "learning_rate": 0.0005, "loss": 3.011169910430908, "step": 524 }, { "epoch": 0.041387465510445406, "grad_norm": 2.0204501152038574, "learning_rate": 0.0005, "loss": 3.445763349533081, "step": 525 }, { "epoch": 0.04146629877808435, "grad_norm": 1.7270995378494263, "learning_rate": 0.0005, "loss": 3.4093384742736816, "step": 526 }, { "epoch": 0.0415451320457233, "grad_norm": 1.433883786201477, "learning_rate": 0.0005, "loss": 3.2995262145996094, "step": 527 }, { "epoch": 0.04162396531336224, "grad_norm": 1.4835256338119507, "learning_rate": 0.0005, "loss": 3.314955949783325, "step": 528 }, { "epoch": 0.041702798581001184, "grad_norm": 1.549500823020935, "learning_rate": 0.0005, "loss": 2.7338943481445312, "step": 529 }, { "epoch": 0.04178163184864012, "grad_norm": 1.9653165340423584, "learning_rate": 0.0005, "loss": 3.3398571014404297, "step": 530 }, { "epoch": 0.04186046511627907, "grad_norm": 1.1940393447875977, "learning_rate": 0.0005, "loss": 3.1216812133789062, "step": 531 }, { "epoch": 0.041939298383918015, "grad_norm": 1.3430439233779907, "learning_rate": 0.0005, "loss": 2.9417824745178223, "step": 532 }, { "epoch": 0.042018131651556954, "grad_norm": 1.131369948387146, "learning_rate": 0.0005, "loss": 3.6210341453552246, "step": 533 }, { "epoch": 0.0420969649191959, "grad_norm": 1.097426176071167, "learning_rate": 0.0005, "loss": 3.366222620010376, "step": 534 }, { "epoch": 0.04217579818683485, "grad_norm": 1.574958086013794, "learning_rate": 0.0005, "loss": 3.2850959300994873, "step": 535 }, { "epoch": 0.042254631454473786, "grad_norm": 1.1779263019561768, "learning_rate": 0.0005, "loss": 3.336498260498047, "step": 536 }, { "epoch": 0.04233346472211273, "grad_norm": 1.0711772441864014, "learning_rate": 0.0005, "loss": 3.263500690460205, "step": 537 }, { "epoch": 0.04241229798975168, "grad_norm": 2.884756326675415, "learning_rate": 0.0005, "loss": 3.887758255004883, "step": 538 }, { "epoch": 0.04249113125739062, "grad_norm": 1.3046224117279053, "learning_rate": 0.0005, "loss": 3.500455617904663, "step": 539 }, { "epoch": 0.04256996452502956, "grad_norm": 1.7359330654144287, "learning_rate": 0.0005, "loss": 3.3359761238098145, "step": 540 }, { "epoch": 0.04264879779266851, "grad_norm": 1.0994255542755127, "learning_rate": 0.0005, "loss": 3.223851203918457, "step": 541 }, { "epoch": 0.04272763106030745, "grad_norm": 1.3252286911010742, "learning_rate": 0.0005, "loss": 3.4684810638427734, "step": 542 }, { "epoch": 0.042806464327946395, "grad_norm": 1.3429324626922607, "learning_rate": 0.0005, "loss": 3.559609889984131, "step": 543 }, { "epoch": 0.042885297595585334, "grad_norm": 1.2324572801589966, "learning_rate": 0.0005, "loss": 3.5319180488586426, "step": 544 }, { "epoch": 0.04296413086322428, "grad_norm": 1.1949671506881714, "learning_rate": 0.0005, "loss": 3.3653364181518555, "step": 545 }, { "epoch": 0.043042964130863226, "grad_norm": 1.4169291257858276, "learning_rate": 0.0005, "loss": 3.6840569972991943, "step": 546 }, { "epoch": 0.043121797398502165, "grad_norm": 1.1708407402038574, "learning_rate": 0.0005, "loss": 3.007355213165283, "step": 547 }, { "epoch": 0.04320063066614111, "grad_norm": 1.3774759769439697, "learning_rate": 0.0005, "loss": 3.4584028720855713, "step": 548 }, { "epoch": 0.04327946393378006, "grad_norm": 1.103393793106079, "learning_rate": 0.0005, "loss": 3.303657054901123, "step": 549 }, { "epoch": 0.043358297201419, "grad_norm": 1.1124128103256226, "learning_rate": 0.0005, "loss": 3.491692543029785, "step": 550 }, { "epoch": 0.04343713046905794, "grad_norm": 1.35096275806427, "learning_rate": 0.0005, "loss": 3.2908589839935303, "step": 551 }, { "epoch": 0.04351596373669689, "grad_norm": 1.3302295207977295, "learning_rate": 0.0005, "loss": 3.324415683746338, "step": 552 }, { "epoch": 0.04359479700433583, "grad_norm": 1.293246865272522, "learning_rate": 0.0005, "loss": 3.5825142860412598, "step": 553 }, { "epoch": 0.043673630271974774, "grad_norm": 1.1372579336166382, "learning_rate": 0.0005, "loss": 3.5723347663879395, "step": 554 }, { "epoch": 0.04375246353961372, "grad_norm": 1.0924850702285767, "learning_rate": 0.0005, "loss": 3.573817014694214, "step": 555 }, { "epoch": 0.04383129680725266, "grad_norm": 1.3142914772033691, "learning_rate": 0.0005, "loss": 3.264627456665039, "step": 556 }, { "epoch": 0.043910130074891605, "grad_norm": 1.3795822858810425, "learning_rate": 0.0005, "loss": 3.307037591934204, "step": 557 }, { "epoch": 0.043988963342530545, "grad_norm": 1.2882578372955322, "learning_rate": 0.0005, "loss": 3.3157362937927246, "step": 558 }, { "epoch": 0.04406779661016949, "grad_norm": 1.2611340284347534, "learning_rate": 0.0005, "loss": 3.4383673667907715, "step": 559 }, { "epoch": 0.04414662987780844, "grad_norm": 1.5394314527511597, "learning_rate": 0.0005, "loss": 2.965608596801758, "step": 560 }, { "epoch": 0.044225463145447376, "grad_norm": 1.2024807929992676, "learning_rate": 0.0005, "loss": 3.410813808441162, "step": 561 }, { "epoch": 0.04430429641308632, "grad_norm": 1.4635016918182373, "learning_rate": 0.0005, "loss": 3.5318026542663574, "step": 562 }, { "epoch": 0.04438312968072527, "grad_norm": 1.459582805633545, "learning_rate": 0.0005, "loss": 3.6174702644348145, "step": 563 }, { "epoch": 0.04446196294836421, "grad_norm": 1.0609560012817383, "learning_rate": 0.0005, "loss": 3.067964792251587, "step": 564 }, { "epoch": 0.04454079621600315, "grad_norm": 1.088497281074524, "learning_rate": 0.0005, "loss": 3.529803991317749, "step": 565 }, { "epoch": 0.0446196294836421, "grad_norm": 1.0886354446411133, "learning_rate": 0.0005, "loss": 3.325190544128418, "step": 566 }, { "epoch": 0.04469846275128104, "grad_norm": 1.0846580266952515, "learning_rate": 0.0005, "loss": 3.173401355743408, "step": 567 }, { "epoch": 0.044777296018919985, "grad_norm": 1.384385347366333, "learning_rate": 0.0005, "loss": 3.1397597789764404, "step": 568 }, { "epoch": 0.04485612928655893, "grad_norm": 1.306830644607544, "learning_rate": 0.0005, "loss": 3.2339768409729004, "step": 569 }, { "epoch": 0.04493496255419787, "grad_norm": 1.6803805828094482, "learning_rate": 0.0005, "loss": 3.5295591354370117, "step": 570 }, { "epoch": 0.045013795821836816, "grad_norm": 1.2606583833694458, "learning_rate": 0.0005, "loss": 3.1832833290100098, "step": 571 }, { "epoch": 0.045092629089475755, "grad_norm": 1.1658039093017578, "learning_rate": 0.0005, "loss": 3.4282920360565186, "step": 572 }, { "epoch": 0.0451714623571147, "grad_norm": 1.2109153270721436, "learning_rate": 0.0005, "loss": 3.2137975692749023, "step": 573 }, { "epoch": 0.04525029562475365, "grad_norm": 1.0968244075775146, "learning_rate": 0.0005, "loss": 3.350623607635498, "step": 574 }, { "epoch": 0.04532912889239259, "grad_norm": 1.327101230621338, "learning_rate": 0.0005, "loss": 3.4273087978363037, "step": 575 }, { "epoch": 0.04540796216003153, "grad_norm": 1.1423691511154175, "learning_rate": 0.0005, "loss": 3.611020565032959, "step": 576 }, { "epoch": 0.04548679542767048, "grad_norm": 1.5174368619918823, "learning_rate": 0.0005, "loss": 3.0498433113098145, "step": 577 }, { "epoch": 0.04556562869530942, "grad_norm": 2.01898455619812, "learning_rate": 0.0005, "loss": 3.7370803356170654, "step": 578 }, { "epoch": 0.045644461962948364, "grad_norm": 1.2729299068450928, "learning_rate": 0.0005, "loss": 3.4487357139587402, "step": 579 }, { "epoch": 0.04572329523058731, "grad_norm": 1.1416113376617432, "learning_rate": 0.0005, "loss": 3.4044439792633057, "step": 580 }, { "epoch": 0.04580212849822625, "grad_norm": 1.1867350339889526, "learning_rate": 0.0005, "loss": 3.3460497856140137, "step": 581 }, { "epoch": 0.045880961765865196, "grad_norm": 1.1328703165054321, "learning_rate": 0.0005, "loss": 3.4852688312530518, "step": 582 }, { "epoch": 0.04595979503350414, "grad_norm": 1.1309475898742676, "learning_rate": 0.0005, "loss": 3.228696346282959, "step": 583 }, { "epoch": 0.04603862830114308, "grad_norm": 1.4721882343292236, "learning_rate": 0.0005, "loss": 2.948709487915039, "step": 584 }, { "epoch": 0.04611746156878203, "grad_norm": 1.7663055658340454, "learning_rate": 0.0005, "loss": 2.946345090866089, "step": 585 }, { "epoch": 0.04619629483642097, "grad_norm": 1.3349767923355103, "learning_rate": 0.0005, "loss": 2.9347870349884033, "step": 586 }, { "epoch": 0.04627512810405991, "grad_norm": 1.0787521600723267, "learning_rate": 0.0005, "loss": 3.414149522781372, "step": 587 }, { "epoch": 0.04635396137169886, "grad_norm": 1.2382749319076538, "learning_rate": 0.0005, "loss": 3.457331895828247, "step": 588 }, { "epoch": 0.0464327946393378, "grad_norm": 1.0293211936950684, "learning_rate": 0.0005, "loss": 3.2199316024780273, "step": 589 }, { "epoch": 0.046511627906976744, "grad_norm": 1.0938571691513062, "learning_rate": 0.0005, "loss": 3.404517889022827, "step": 590 }, { "epoch": 0.04659046117461569, "grad_norm": 1.2846306562423706, "learning_rate": 0.0005, "loss": 3.2495603561401367, "step": 591 }, { "epoch": 0.04666929444225463, "grad_norm": 1.164110541343689, "learning_rate": 0.0005, "loss": 3.264925718307495, "step": 592 }, { "epoch": 0.046748127709893575, "grad_norm": 1.3154577016830444, "learning_rate": 0.0005, "loss": 3.2691264152526855, "step": 593 }, { "epoch": 0.04682696097753252, "grad_norm": 1.488764762878418, "learning_rate": 0.0005, "loss": 3.149989128112793, "step": 594 }, { "epoch": 0.04690579424517146, "grad_norm": 1.003389596939087, "learning_rate": 0.0005, "loss": 3.2237236499786377, "step": 595 }, { "epoch": 0.046984627512810406, "grad_norm": 1.1196364164352417, "learning_rate": 0.0005, "loss": 3.6184308528900146, "step": 596 }, { "epoch": 0.04706346078044935, "grad_norm": 1.2757384777069092, "learning_rate": 0.0005, "loss": 3.1474575996398926, "step": 597 }, { "epoch": 0.04714229404808829, "grad_norm": 1.297890543937683, "learning_rate": 0.0005, "loss": 3.0413191318511963, "step": 598 }, { "epoch": 0.04722112731572724, "grad_norm": 1.2775177955627441, "learning_rate": 0.0005, "loss": 3.6362853050231934, "step": 599 }, { "epoch": 0.047299960583366184, "grad_norm": 1.2558778524398804, "learning_rate": 0.0005, "loss": 3.4423041343688965, "step": 600 }, { "epoch": 0.04737879385100512, "grad_norm": 2.1361916065216064, "learning_rate": 0.0005, "loss": 3.1247658729553223, "step": 601 }, { "epoch": 0.04745762711864407, "grad_norm": 1.031191349029541, "learning_rate": 0.0005, "loss": 3.194474220275879, "step": 602 }, { "epoch": 0.04753646038628301, "grad_norm": 1.3543239831924438, "learning_rate": 0.0005, "loss": 2.996751308441162, "step": 603 }, { "epoch": 0.047615293653921954, "grad_norm": 1.4955450296401978, "learning_rate": 0.0005, "loss": 3.1162099838256836, "step": 604 }, { "epoch": 0.0476941269215609, "grad_norm": 1.2984474897384644, "learning_rate": 0.0005, "loss": 2.951662063598633, "step": 605 }, { "epoch": 0.04777296018919984, "grad_norm": 1.254752278327942, "learning_rate": 0.0005, "loss": 3.4543566703796387, "step": 606 }, { "epoch": 0.047851793456838786, "grad_norm": 1.1711082458496094, "learning_rate": 0.0005, "loss": 3.170532703399658, "step": 607 }, { "epoch": 0.04793062672447773, "grad_norm": 1.3153057098388672, "learning_rate": 0.0005, "loss": 3.5072968006134033, "step": 608 }, { "epoch": 0.04800945999211667, "grad_norm": 1.1972317695617676, "learning_rate": 0.0005, "loss": 2.928518295288086, "step": 609 }, { "epoch": 0.04808829325975562, "grad_norm": 1.4460443258285522, "learning_rate": 0.0005, "loss": 3.2802698612213135, "step": 610 }, { "epoch": 0.04816712652739456, "grad_norm": 1.0507057905197144, "learning_rate": 0.0005, "loss": 3.2884652614593506, "step": 611 }, { "epoch": 0.0482459597950335, "grad_norm": 1.3766807317733765, "learning_rate": 0.0005, "loss": 3.1578259468078613, "step": 612 }, { "epoch": 0.04832479306267245, "grad_norm": 0.9693579077720642, "learning_rate": 0.0005, "loss": 3.2180275917053223, "step": 613 }, { "epoch": 0.048403626330311394, "grad_norm": 1.1619501113891602, "learning_rate": 0.0005, "loss": 3.3248231410980225, "step": 614 }, { "epoch": 0.048482459597950334, "grad_norm": 1.3559168577194214, "learning_rate": 0.0005, "loss": 3.4694533348083496, "step": 615 }, { "epoch": 0.04856129286558928, "grad_norm": 1.2397806644439697, "learning_rate": 0.0005, "loss": 3.4193367958068848, "step": 616 }, { "epoch": 0.04864012613322822, "grad_norm": 0.9798298478126526, "learning_rate": 0.0005, "loss": 3.098011016845703, "step": 617 }, { "epoch": 0.048718959400867165, "grad_norm": 1.5135929584503174, "learning_rate": 0.0005, "loss": 3.7665371894836426, "step": 618 }, { "epoch": 0.04879779266850611, "grad_norm": 1.2528756856918335, "learning_rate": 0.0005, "loss": 3.0714547634124756, "step": 619 }, { "epoch": 0.04887662593614505, "grad_norm": 1.0921695232391357, "learning_rate": 0.0005, "loss": 2.9808616638183594, "step": 620 }, { "epoch": 0.048955459203783996, "grad_norm": 1.530756950378418, "learning_rate": 0.0005, "loss": 3.2230224609375, "step": 621 }, { "epoch": 0.04903429247142294, "grad_norm": 1.3383375406265259, "learning_rate": 0.0005, "loss": 3.2234606742858887, "step": 622 }, { "epoch": 0.04911312573906188, "grad_norm": 1.5535293817520142, "learning_rate": 0.0005, "loss": 3.368701219558716, "step": 623 }, { "epoch": 0.04919195900670083, "grad_norm": 1.1964441537857056, "learning_rate": 0.0005, "loss": 3.3705453872680664, "step": 624 }, { "epoch": 0.049270792274339774, "grad_norm": 1.9129277467727661, "learning_rate": 0.0005, "loss": 3.3675832748413086, "step": 625 }, { "epoch": 0.04934962554197871, "grad_norm": 1.343531608581543, "learning_rate": 0.0005, "loss": 3.258057117462158, "step": 626 }, { "epoch": 0.04942845880961766, "grad_norm": 1.7178412675857544, "learning_rate": 0.0005, "loss": 3.624443769454956, "step": 627 }, { "epoch": 0.049507292077256605, "grad_norm": 1.65733802318573, "learning_rate": 0.0005, "loss": 2.8387608528137207, "step": 628 }, { "epoch": 0.049586125344895544, "grad_norm": 1.2563217878341675, "learning_rate": 0.0005, "loss": 3.3287343978881836, "step": 629 }, { "epoch": 0.04966495861253449, "grad_norm": 1.6125646829605103, "learning_rate": 0.0005, "loss": 3.345094680786133, "step": 630 }, { "epoch": 0.04974379188017344, "grad_norm": 1.1331838369369507, "learning_rate": 0.0005, "loss": 3.393915891647339, "step": 631 }, { "epoch": 0.049822625147812376, "grad_norm": 1.4334322214126587, "learning_rate": 0.0005, "loss": 3.0930352210998535, "step": 632 }, { "epoch": 0.04990145841545132, "grad_norm": 1.0835402011871338, "learning_rate": 0.0005, "loss": 3.2858033180236816, "step": 633 }, { "epoch": 0.04998029168309026, "grad_norm": 1.1923078298568726, "learning_rate": 0.0005, "loss": 3.205491304397583, "step": 634 }, { "epoch": 0.05005912495072921, "grad_norm": 1.0787242650985718, "learning_rate": 0.0005, "loss": 3.6104180812835693, "step": 635 }, { "epoch": 0.05013795821836815, "grad_norm": 1.0831382274627686, "learning_rate": 0.0005, "loss": 3.3803021907806396, "step": 636 }, { "epoch": 0.05021679148600709, "grad_norm": 1.0458101034164429, "learning_rate": 0.0005, "loss": 3.273881196975708, "step": 637 }, { "epoch": 0.05029562475364604, "grad_norm": 1.6194523572921753, "learning_rate": 0.0005, "loss": 3.328733444213867, "step": 638 }, { "epoch": 0.050374458021284985, "grad_norm": 1.0924760103225708, "learning_rate": 0.0005, "loss": 3.026151180267334, "step": 639 }, { "epoch": 0.050453291288923924, "grad_norm": 1.2581127882003784, "learning_rate": 0.0005, "loss": 3.1247200965881348, "step": 640 }, { "epoch": 0.05053212455656287, "grad_norm": 1.0933353900909424, "learning_rate": 0.0005, "loss": 3.2839598655700684, "step": 641 }, { "epoch": 0.050610957824201816, "grad_norm": 1.2360832691192627, "learning_rate": 0.0005, "loss": 3.2557973861694336, "step": 642 }, { "epoch": 0.050689791091840755, "grad_norm": 1.1726523637771606, "learning_rate": 0.0005, "loss": 3.02432918548584, "step": 643 }, { "epoch": 0.0507686243594797, "grad_norm": 1.1953811645507812, "learning_rate": 0.0005, "loss": 3.454766273498535, "step": 644 }, { "epoch": 0.05084745762711865, "grad_norm": 1.2466919422149658, "learning_rate": 0.0005, "loss": 2.964111089706421, "step": 645 }, { "epoch": 0.050926290894757587, "grad_norm": 1.2143456935882568, "learning_rate": 0.0005, "loss": 3.262254238128662, "step": 646 }, { "epoch": 0.05100512416239653, "grad_norm": 1.0732868909835815, "learning_rate": 0.0005, "loss": 3.0285091400146484, "step": 647 }, { "epoch": 0.05108395743003547, "grad_norm": 1.2476115226745605, "learning_rate": 0.0005, "loss": 3.3297882080078125, "step": 648 }, { "epoch": 0.05116279069767442, "grad_norm": 1.8578739166259766, "learning_rate": 0.0005, "loss": 3.3028392791748047, "step": 649 }, { "epoch": 0.051241623965313364, "grad_norm": 1.1863725185394287, "learning_rate": 0.0005, "loss": 3.214128017425537, "step": 650 }, { "epoch": 0.0513204572329523, "grad_norm": 1.1901631355285645, "learning_rate": 0.0005, "loss": 3.347227096557617, "step": 651 }, { "epoch": 0.05139929050059125, "grad_norm": 1.1496355533599854, "learning_rate": 0.0005, "loss": 3.455448627471924, "step": 652 }, { "epoch": 0.051478123768230195, "grad_norm": 1.0665861368179321, "learning_rate": 0.0005, "loss": 3.3511404991149902, "step": 653 }, { "epoch": 0.051556957035869135, "grad_norm": 1.1605504751205444, "learning_rate": 0.0005, "loss": 2.8153445720672607, "step": 654 }, { "epoch": 0.05163579030350808, "grad_norm": 1.0716941356658936, "learning_rate": 0.0005, "loss": 3.36814022064209, "step": 655 }, { "epoch": 0.05171462357114703, "grad_norm": 1.117663860321045, "learning_rate": 0.0005, "loss": 3.258676528930664, "step": 656 }, { "epoch": 0.051793456838785966, "grad_norm": 1.3856357336044312, "learning_rate": 0.0005, "loss": 3.605973958969116, "step": 657 }, { "epoch": 0.05187229010642491, "grad_norm": 1.1334861516952515, "learning_rate": 0.0005, "loss": 2.6506128311157227, "step": 658 }, { "epoch": 0.05195112337406386, "grad_norm": 1.3617310523986816, "learning_rate": 0.0005, "loss": 3.4632315635681152, "step": 659 }, { "epoch": 0.0520299566417028, "grad_norm": 1.13971745967865, "learning_rate": 0.0005, "loss": 3.2458741664886475, "step": 660 }, { "epoch": 0.05210878990934174, "grad_norm": 1.1515142917633057, "learning_rate": 0.0005, "loss": 3.5401735305786133, "step": 661 }, { "epoch": 0.05218762317698068, "grad_norm": 1.2330372333526611, "learning_rate": 0.0005, "loss": 3.367755889892578, "step": 662 }, { "epoch": 0.05226645644461963, "grad_norm": 1.1461677551269531, "learning_rate": 0.0005, "loss": 3.2617557048797607, "step": 663 }, { "epoch": 0.052345289712258575, "grad_norm": 1.461475133895874, "learning_rate": 0.0005, "loss": 3.296023368835449, "step": 664 }, { "epoch": 0.052424122979897514, "grad_norm": 1.3989259004592896, "learning_rate": 0.0005, "loss": 3.4530792236328125, "step": 665 }, { "epoch": 0.05250295624753646, "grad_norm": 1.8287458419799805, "learning_rate": 0.0005, "loss": 3.28825044631958, "step": 666 }, { "epoch": 0.052581789515175406, "grad_norm": 0.992625892162323, "learning_rate": 0.0005, "loss": 3.151820421218872, "step": 667 }, { "epoch": 0.052660622782814345, "grad_norm": 1.025199294090271, "learning_rate": 0.0005, "loss": 3.2754058837890625, "step": 668 }, { "epoch": 0.05273945605045329, "grad_norm": 1.5307706594467163, "learning_rate": 0.0005, "loss": 3.528667688369751, "step": 669 }, { "epoch": 0.05281828931809224, "grad_norm": 1.595569372177124, "learning_rate": 0.0005, "loss": 3.4608116149902344, "step": 670 }, { "epoch": 0.05289712258573118, "grad_norm": 1.4493260383605957, "learning_rate": 0.0005, "loss": 3.3914318084716797, "step": 671 }, { "epoch": 0.05297595585337012, "grad_norm": 1.4931493997573853, "learning_rate": 0.0005, "loss": 3.255354642868042, "step": 672 }, { "epoch": 0.05305478912100907, "grad_norm": 1.1108908653259277, "learning_rate": 0.0005, "loss": 3.1193137168884277, "step": 673 }, { "epoch": 0.05313362238864801, "grad_norm": 1.1484202146530151, "learning_rate": 0.0005, "loss": 3.3418753147125244, "step": 674 }, { "epoch": 0.053212455656286954, "grad_norm": 1.3731975555419922, "learning_rate": 0.0005, "loss": 3.0183310508728027, "step": 675 }, { "epoch": 0.05329128892392589, "grad_norm": 1.3762032985687256, "learning_rate": 0.0005, "loss": 3.383753538131714, "step": 676 }, { "epoch": 0.05337012219156484, "grad_norm": 2.534485101699829, "learning_rate": 0.0005, "loss": 3.3559532165527344, "step": 677 }, { "epoch": 0.053448955459203785, "grad_norm": 1.6349345445632935, "learning_rate": 0.0005, "loss": 3.534963369369507, "step": 678 }, { "epoch": 0.053527788726842725, "grad_norm": 1.880702257156372, "learning_rate": 0.0005, "loss": 3.1520628929138184, "step": 679 }, { "epoch": 0.05360662199448167, "grad_norm": 1.2173043489456177, "learning_rate": 0.0005, "loss": 3.3614273071289062, "step": 680 }, { "epoch": 0.05368545526212062, "grad_norm": 1.4106875658035278, "learning_rate": 0.0005, "loss": 3.334075450897217, "step": 681 }, { "epoch": 0.053764288529759556, "grad_norm": 1.1473875045776367, "learning_rate": 0.0005, "loss": 3.146206855773926, "step": 682 }, { "epoch": 0.0538431217973985, "grad_norm": 1.326805591583252, "learning_rate": 0.0005, "loss": 2.9667954444885254, "step": 683 }, { "epoch": 0.05392195506503745, "grad_norm": 1.2393527030944824, "learning_rate": 0.0005, "loss": 3.1302504539489746, "step": 684 }, { "epoch": 0.05400078833267639, "grad_norm": 1.334895133972168, "learning_rate": 0.0005, "loss": 3.109476327896118, "step": 685 }, { "epoch": 0.054079621600315333, "grad_norm": 1.2271188497543335, "learning_rate": 0.0005, "loss": 3.008117198944092, "step": 686 }, { "epoch": 0.05415845486795428, "grad_norm": 1.9430888891220093, "learning_rate": 0.0005, "loss": 3.1234309673309326, "step": 687 }, { "epoch": 0.05423728813559322, "grad_norm": 1.0936698913574219, "learning_rate": 0.0005, "loss": 3.282827854156494, "step": 688 }, { "epoch": 0.054316121403232165, "grad_norm": 1.1952059268951416, "learning_rate": 0.0005, "loss": 3.148775577545166, "step": 689 }, { "epoch": 0.05439495467087111, "grad_norm": 1.504250168800354, "learning_rate": 0.0005, "loss": 3.252624988555908, "step": 690 }, { "epoch": 0.05447378793851005, "grad_norm": 1.215872883796692, "learning_rate": 0.0005, "loss": 3.37072491645813, "step": 691 }, { "epoch": 0.054552621206148996, "grad_norm": 1.2499020099639893, "learning_rate": 0.0005, "loss": 3.266721248626709, "step": 692 }, { "epoch": 0.054631454473787935, "grad_norm": 1.2894091606140137, "learning_rate": 0.0005, "loss": 3.259817600250244, "step": 693 }, { "epoch": 0.05471028774142688, "grad_norm": 1.2467329502105713, "learning_rate": 0.0005, "loss": 3.37310791015625, "step": 694 }, { "epoch": 0.05478912100906583, "grad_norm": 1.285485029220581, "learning_rate": 0.0005, "loss": 3.385748863220215, "step": 695 }, { "epoch": 0.05486795427670477, "grad_norm": 1.1787294149398804, "learning_rate": 0.0005, "loss": 3.4450550079345703, "step": 696 }, { "epoch": 0.05494678754434371, "grad_norm": 1.2335996627807617, "learning_rate": 0.0005, "loss": 3.260061740875244, "step": 697 }, { "epoch": 0.05502562081198266, "grad_norm": 1.5032891035079956, "learning_rate": 0.0005, "loss": 3.5811662673950195, "step": 698 }, { "epoch": 0.0551044540796216, "grad_norm": 1.3364020586013794, "learning_rate": 0.0005, "loss": 3.529797077178955, "step": 699 }, { "epoch": 0.055183287347260544, "grad_norm": 1.2269526720046997, "learning_rate": 0.0005, "loss": 3.2592391967773438, "step": 700 }, { "epoch": 0.05526212061489949, "grad_norm": 1.1751422882080078, "learning_rate": 0.0005, "loss": 2.815464496612549, "step": 701 }, { "epoch": 0.05534095388253843, "grad_norm": 0.8912520408630371, "learning_rate": 0.0005, "loss": 3.4476990699768066, "step": 702 }, { "epoch": 0.055419787150177376, "grad_norm": 1.4409124851226807, "learning_rate": 0.0005, "loss": 3.0294764041900635, "step": 703 }, { "epoch": 0.05549862041781632, "grad_norm": 1.9564074277877808, "learning_rate": 0.0005, "loss": 2.909175157546997, "step": 704 }, { "epoch": 0.05557745368545526, "grad_norm": 1.1062296628952026, "learning_rate": 0.0005, "loss": 3.1892127990722656, "step": 705 }, { "epoch": 0.05565628695309421, "grad_norm": 1.2395284175872803, "learning_rate": 0.0005, "loss": 3.0871543884277344, "step": 706 }, { "epoch": 0.055735120220733146, "grad_norm": 1.1892108917236328, "learning_rate": 0.0005, "loss": 3.0462241172790527, "step": 707 }, { "epoch": 0.05581395348837209, "grad_norm": 1.8815339803695679, "learning_rate": 0.0005, "loss": 3.4752883911132812, "step": 708 }, { "epoch": 0.05589278675601104, "grad_norm": 1.2910900115966797, "learning_rate": 0.0005, "loss": 3.243994951248169, "step": 709 }, { "epoch": 0.05597162002364998, "grad_norm": 1.035852313041687, "learning_rate": 0.0005, "loss": 3.384941339492798, "step": 710 }, { "epoch": 0.056050453291288924, "grad_norm": 1.0435289144515991, "learning_rate": 0.0005, "loss": 3.296088218688965, "step": 711 }, { "epoch": 0.05612928655892787, "grad_norm": 1.2555207014083862, "learning_rate": 0.0005, "loss": 3.2532846927642822, "step": 712 }, { "epoch": 0.05620811982656681, "grad_norm": 1.3092801570892334, "learning_rate": 0.0005, "loss": 2.9994940757751465, "step": 713 }, { "epoch": 0.056286953094205755, "grad_norm": 1.2608872652053833, "learning_rate": 0.0005, "loss": 3.1538875102996826, "step": 714 }, { "epoch": 0.0563657863618447, "grad_norm": 1.2361127138137817, "learning_rate": 0.0005, "loss": 3.1608340740203857, "step": 715 }, { "epoch": 0.05644461962948364, "grad_norm": 1.3629308938980103, "learning_rate": 0.0005, "loss": 3.135941982269287, "step": 716 }, { "epoch": 0.056523452897122586, "grad_norm": 1.1818832159042358, "learning_rate": 0.0005, "loss": 2.8666200637817383, "step": 717 }, { "epoch": 0.05660228616476153, "grad_norm": 1.3303747177124023, "learning_rate": 0.0005, "loss": 3.465231418609619, "step": 718 }, { "epoch": 0.05668111943240047, "grad_norm": 1.4168767929077148, "learning_rate": 0.0005, "loss": 3.45102596282959, "step": 719 }, { "epoch": 0.05675995270003942, "grad_norm": 1.3206454515457153, "learning_rate": 0.0005, "loss": 3.2223291397094727, "step": 720 }, { "epoch": 0.05683878596767836, "grad_norm": 1.0420795679092407, "learning_rate": 0.0005, "loss": 3.1591923236846924, "step": 721 }, { "epoch": 0.0569176192353173, "grad_norm": 1.465225338935852, "learning_rate": 0.0005, "loss": 3.7620127201080322, "step": 722 }, { "epoch": 0.05699645250295625, "grad_norm": 1.0917035341262817, "learning_rate": 0.0005, "loss": 3.3141815662384033, "step": 723 }, { "epoch": 0.05707528577059519, "grad_norm": 1.0576244592666626, "learning_rate": 0.0005, "loss": 3.2864582538604736, "step": 724 }, { "epoch": 0.057154119038234134, "grad_norm": 1.125098705291748, "learning_rate": 0.0005, "loss": 3.1880855560302734, "step": 725 }, { "epoch": 0.05723295230587308, "grad_norm": 1.1765928268432617, "learning_rate": 0.0005, "loss": 3.3826937675476074, "step": 726 }, { "epoch": 0.05731178557351202, "grad_norm": 1.0925594568252563, "learning_rate": 0.0005, "loss": 3.1854629516601562, "step": 727 }, { "epoch": 0.057390618841150966, "grad_norm": 1.2504347562789917, "learning_rate": 0.0005, "loss": 2.5392823219299316, "step": 728 }, { "epoch": 0.05746945210878991, "grad_norm": 1.3188449144363403, "learning_rate": 0.0005, "loss": 3.242729902267456, "step": 729 }, { "epoch": 0.05754828537642885, "grad_norm": 1.116127848625183, "learning_rate": 0.0005, "loss": 3.155776023864746, "step": 730 }, { "epoch": 0.0576271186440678, "grad_norm": 1.2745417356491089, "learning_rate": 0.0005, "loss": 2.93058180809021, "step": 731 }, { "epoch": 0.05770595191170674, "grad_norm": 1.0157818794250488, "learning_rate": 0.0005, "loss": 3.044231653213501, "step": 732 }, { "epoch": 0.05778478517934568, "grad_norm": 1.2713931798934937, "learning_rate": 0.0005, "loss": 3.5161826610565186, "step": 733 }, { "epoch": 0.05786361844698463, "grad_norm": 1.1697665452957153, "learning_rate": 0.0005, "loss": 3.223219394683838, "step": 734 }, { "epoch": 0.057942451714623575, "grad_norm": 0.9445950984954834, "learning_rate": 0.0005, "loss": 3.384106159210205, "step": 735 }, { "epoch": 0.058021284982262514, "grad_norm": 1.3287028074264526, "learning_rate": 0.0005, "loss": 3.462470054626465, "step": 736 }, { "epoch": 0.05810011824990146, "grad_norm": 1.1431492567062378, "learning_rate": 0.0005, "loss": 3.029029369354248, "step": 737 }, { "epoch": 0.0581789515175404, "grad_norm": 1.2791996002197266, "learning_rate": 0.0005, "loss": 3.179541826248169, "step": 738 }, { "epoch": 0.058257784785179345, "grad_norm": 1.0414915084838867, "learning_rate": 0.0005, "loss": 3.2353954315185547, "step": 739 }, { "epoch": 0.05833661805281829, "grad_norm": 1.083288550376892, "learning_rate": 0.0005, "loss": 2.842074394226074, "step": 740 }, { "epoch": 0.05841545132045723, "grad_norm": 2.078544855117798, "learning_rate": 0.0005, "loss": 2.9708399772644043, "step": 741 }, { "epoch": 0.058494284588096176, "grad_norm": 1.2356534004211426, "learning_rate": 0.0005, "loss": 3.2913291454315186, "step": 742 }, { "epoch": 0.05857311785573512, "grad_norm": 1.4583487510681152, "learning_rate": 0.0005, "loss": 3.0814990997314453, "step": 743 }, { "epoch": 0.05865195112337406, "grad_norm": 1.304957389831543, "learning_rate": 0.0005, "loss": 3.031371831893921, "step": 744 }, { "epoch": 0.05873078439101301, "grad_norm": 1.0243489742279053, "learning_rate": 0.0005, "loss": 3.1846346855163574, "step": 745 }, { "epoch": 0.058809617658651954, "grad_norm": 1.3818371295928955, "learning_rate": 0.0005, "loss": 3.1572213172912598, "step": 746 }, { "epoch": 0.05888845092629089, "grad_norm": 1.0354775190353394, "learning_rate": 0.0005, "loss": 3.186891555786133, "step": 747 }, { "epoch": 0.05896728419392984, "grad_norm": 2.2753350734710693, "learning_rate": 0.0005, "loss": 3.680440902709961, "step": 748 }, { "epoch": 0.059046117461568785, "grad_norm": 1.258037805557251, "learning_rate": 0.0005, "loss": 2.8734991550445557, "step": 749 }, { "epoch": 0.059124950729207724, "grad_norm": 1.9946341514587402, "learning_rate": 0.0005, "loss": 2.965681314468384, "step": 750 }, { "epoch": 0.05920378399684667, "grad_norm": 1.1336818933486938, "learning_rate": 0.0005, "loss": 3.3700106143951416, "step": 751 }, { "epoch": 0.05928261726448561, "grad_norm": 1.034956932067871, "learning_rate": 0.0005, "loss": 3.268911361694336, "step": 752 }, { "epoch": 0.059361450532124556, "grad_norm": 1.2755690813064575, "learning_rate": 0.0005, "loss": 3.0688977241516113, "step": 753 }, { "epoch": 0.0594402837997635, "grad_norm": 1.5485811233520508, "learning_rate": 0.0005, "loss": 3.1337268352508545, "step": 754 }, { "epoch": 0.05951911706740244, "grad_norm": 1.4130138158798218, "learning_rate": 0.0005, "loss": 3.6177761554718018, "step": 755 }, { "epoch": 0.05959795033504139, "grad_norm": 1.4881139993667603, "learning_rate": 0.0005, "loss": 3.2025530338287354, "step": 756 }, { "epoch": 0.05967678360268033, "grad_norm": 1.2650874853134155, "learning_rate": 0.0005, "loss": 3.029672622680664, "step": 757 }, { "epoch": 0.05975561687031927, "grad_norm": 1.0861258506774902, "learning_rate": 0.0005, "loss": 3.068114757537842, "step": 758 }, { "epoch": 0.05983445013795822, "grad_norm": 1.0248973369598389, "learning_rate": 0.0005, "loss": 3.165593147277832, "step": 759 }, { "epoch": 0.059913283405597165, "grad_norm": 1.6353721618652344, "learning_rate": 0.0005, "loss": 3.1627774238586426, "step": 760 }, { "epoch": 0.059992116673236104, "grad_norm": 1.1087747812271118, "learning_rate": 0.0005, "loss": 3.263397455215454, "step": 761 }, { "epoch": 0.06007094994087505, "grad_norm": 1.1570932865142822, "learning_rate": 0.0005, "loss": 3.2604117393493652, "step": 762 }, { "epoch": 0.060149783208513996, "grad_norm": 2.28680682182312, "learning_rate": 0.0005, "loss": 3.015000820159912, "step": 763 }, { "epoch": 0.060228616476152935, "grad_norm": 1.576157808303833, "learning_rate": 0.0005, "loss": 3.281491279602051, "step": 764 }, { "epoch": 0.06030744974379188, "grad_norm": 1.4489054679870605, "learning_rate": 0.0005, "loss": 2.9361534118652344, "step": 765 }, { "epoch": 0.06038628301143082, "grad_norm": 1.3100696802139282, "learning_rate": 0.0005, "loss": 3.3476295471191406, "step": 766 }, { "epoch": 0.06046511627906977, "grad_norm": 1.1758580207824707, "learning_rate": 0.0005, "loss": 3.493936061859131, "step": 767 }, { "epoch": 0.06054394954670871, "grad_norm": 1.131361722946167, "learning_rate": 0.0005, "loss": 3.268935203552246, "step": 768 }, { "epoch": 0.06062278281434765, "grad_norm": 1.250918984413147, "learning_rate": 0.0005, "loss": 3.2786693572998047, "step": 769 }, { "epoch": 0.0607016160819866, "grad_norm": 1.364327311515808, "learning_rate": 0.0005, "loss": 3.340508222579956, "step": 770 }, { "epoch": 0.060780449349625544, "grad_norm": 1.5985991954803467, "learning_rate": 0.0005, "loss": 3.2466976642608643, "step": 771 }, { "epoch": 0.06085928261726448, "grad_norm": 1.20023775100708, "learning_rate": 0.0005, "loss": 3.1482419967651367, "step": 772 }, { "epoch": 0.06093811588490343, "grad_norm": 2.3212802410125732, "learning_rate": 0.0005, "loss": 3.3306899070739746, "step": 773 }, { "epoch": 0.061016949152542375, "grad_norm": 2.07165265083313, "learning_rate": 0.0005, "loss": 3.2213547229766846, "step": 774 }, { "epoch": 0.061095782420181315, "grad_norm": 1.2470247745513916, "learning_rate": 0.0005, "loss": 3.3743090629577637, "step": 775 }, { "epoch": 0.06117461568782026, "grad_norm": 1.253130555152893, "learning_rate": 0.0005, "loss": 3.2240262031555176, "step": 776 }, { "epoch": 0.06125344895545921, "grad_norm": 1.1536595821380615, "learning_rate": 0.0005, "loss": 3.4932172298431396, "step": 777 }, { "epoch": 0.061332282223098146, "grad_norm": 0.9988316297531128, "learning_rate": 0.0005, "loss": 3.0389623641967773, "step": 778 }, { "epoch": 0.06141111549073709, "grad_norm": 1.231165885925293, "learning_rate": 0.0005, "loss": 3.175915002822876, "step": 779 }, { "epoch": 0.06148994875837603, "grad_norm": 1.5379189252853394, "learning_rate": 0.0005, "loss": 3.350181818008423, "step": 780 }, { "epoch": 0.06156878202601498, "grad_norm": 1.0779262781143188, "learning_rate": 0.0005, "loss": 3.3428001403808594, "step": 781 }, { "epoch": 0.06164761529365392, "grad_norm": 1.2960174083709717, "learning_rate": 0.0005, "loss": 3.3621163368225098, "step": 782 }, { "epoch": 0.06172644856129286, "grad_norm": 4.520320892333984, "learning_rate": 0.0005, "loss": 2.8128082752227783, "step": 783 }, { "epoch": 0.06180528182893181, "grad_norm": 2.047140598297119, "learning_rate": 0.0005, "loss": 3.0457170009613037, "step": 784 }, { "epoch": 0.061884115096570755, "grad_norm": 1.2555248737335205, "learning_rate": 0.0005, "loss": 3.0264780521392822, "step": 785 }, { "epoch": 0.061962948364209694, "grad_norm": 1.1497668027877808, "learning_rate": 0.0005, "loss": 3.201875686645508, "step": 786 }, { "epoch": 0.06204178163184864, "grad_norm": 1.2771002054214478, "learning_rate": 0.0005, "loss": 3.196773052215576, "step": 787 }, { "epoch": 0.062120614899487586, "grad_norm": 1.1728899478912354, "learning_rate": 0.0005, "loss": 2.7971136569976807, "step": 788 }, { "epoch": 0.062199448167126525, "grad_norm": 1.2650017738342285, "learning_rate": 0.0005, "loss": 2.693507194519043, "step": 789 }, { "epoch": 0.06227828143476547, "grad_norm": 1.1554828882217407, "learning_rate": 0.0005, "loss": 3.344754219055176, "step": 790 }, { "epoch": 0.06235711470240442, "grad_norm": 1.5378334522247314, "learning_rate": 0.0005, "loss": 2.93575119972229, "step": 791 }, { "epoch": 0.06243594797004336, "grad_norm": 1.3950021266937256, "learning_rate": 0.0005, "loss": 2.917780876159668, "step": 792 }, { "epoch": 0.0625147812376823, "grad_norm": 1.1131256818771362, "learning_rate": 0.0005, "loss": 2.8882603645324707, "step": 793 }, { "epoch": 0.06259361450532125, "grad_norm": 1.3005752563476562, "learning_rate": 0.0005, "loss": 3.499013900756836, "step": 794 }, { "epoch": 0.06267244777296019, "grad_norm": 1.2493951320648193, "learning_rate": 0.0005, "loss": 3.4589180946350098, "step": 795 }, { "epoch": 0.06275128104059913, "grad_norm": 1.0870682001113892, "learning_rate": 0.0005, "loss": 3.0074045658111572, "step": 796 }, { "epoch": 0.06283011430823808, "grad_norm": 1.0957516431808472, "learning_rate": 0.0005, "loss": 3.0204124450683594, "step": 797 }, { "epoch": 0.06290894757587702, "grad_norm": 1.2527164220809937, "learning_rate": 0.0005, "loss": 3.2698569297790527, "step": 798 }, { "epoch": 0.06298778084351596, "grad_norm": 1.1448665857315063, "learning_rate": 0.0005, "loss": 3.0983192920684814, "step": 799 }, { "epoch": 0.06306661411115491, "grad_norm": 1.2374968528747559, "learning_rate": 0.0005, "loss": 3.1700448989868164, "step": 800 }, { "epoch": 0.06314544737879385, "grad_norm": 1.1687036752700806, "learning_rate": 0.0005, "loss": 3.5037953853607178, "step": 801 }, { "epoch": 0.06322428064643279, "grad_norm": 1.0661578178405762, "learning_rate": 0.0005, "loss": 3.3657922744750977, "step": 802 }, { "epoch": 0.06330311391407174, "grad_norm": 1.0421653985977173, "learning_rate": 0.0005, "loss": 3.5259251594543457, "step": 803 }, { "epoch": 0.06338194718171068, "grad_norm": 1.369493007659912, "learning_rate": 0.0005, "loss": 2.9728970527648926, "step": 804 }, { "epoch": 0.06346078044934962, "grad_norm": 1.3604567050933838, "learning_rate": 0.0005, "loss": 3.209859848022461, "step": 805 }, { "epoch": 0.06353961371698857, "grad_norm": 1.1579838991165161, "learning_rate": 0.0005, "loss": 3.224717140197754, "step": 806 }, { "epoch": 0.06361844698462751, "grad_norm": 1.7327277660369873, "learning_rate": 0.0005, "loss": 3.3356881141662598, "step": 807 }, { "epoch": 0.06369728025226645, "grad_norm": 1.2254114151000977, "learning_rate": 0.0005, "loss": 3.1215882301330566, "step": 808 }, { "epoch": 0.0637761135199054, "grad_norm": 1.1695728302001953, "learning_rate": 0.0005, "loss": 3.2933664321899414, "step": 809 }, { "epoch": 0.06385494678754434, "grad_norm": 1.0893580913543701, "learning_rate": 0.0005, "loss": 3.0097928047180176, "step": 810 }, { "epoch": 0.06393378005518328, "grad_norm": 1.029389500617981, "learning_rate": 0.0005, "loss": 3.3229422569274902, "step": 811 }, { "epoch": 0.06401261332282224, "grad_norm": 1.2160226106643677, "learning_rate": 0.0005, "loss": 3.1538596153259277, "step": 812 }, { "epoch": 0.06409144659046118, "grad_norm": 1.2306052446365356, "learning_rate": 0.0005, "loss": 3.0885982513427734, "step": 813 }, { "epoch": 0.06417027985810012, "grad_norm": 0.9646588563919067, "learning_rate": 0.0005, "loss": 3.9097838401794434, "step": 814 }, { "epoch": 0.06424911312573907, "grad_norm": 1.0505940914154053, "learning_rate": 0.0005, "loss": 3.2413229942321777, "step": 815 }, { "epoch": 0.06432794639337801, "grad_norm": 1.286840558052063, "learning_rate": 0.0005, "loss": 3.5195536613464355, "step": 816 }, { "epoch": 0.06440677966101695, "grad_norm": 1.0070300102233887, "learning_rate": 0.0005, "loss": 3.2525229454040527, "step": 817 }, { "epoch": 0.06448561292865589, "grad_norm": 1.267648696899414, "learning_rate": 0.0005, "loss": 3.2355103492736816, "step": 818 }, { "epoch": 0.06456444619629484, "grad_norm": 1.2634965181350708, "learning_rate": 0.0005, "loss": 3.002810478210449, "step": 819 }, { "epoch": 0.06464327946393378, "grad_norm": 1.2461286783218384, "learning_rate": 0.0005, "loss": 3.5027036666870117, "step": 820 }, { "epoch": 0.06472211273157272, "grad_norm": 1.3094098567962646, "learning_rate": 0.0005, "loss": 3.2733020782470703, "step": 821 }, { "epoch": 0.06480094599921167, "grad_norm": 1.2205564975738525, "learning_rate": 0.0005, "loss": 3.301213026046753, "step": 822 }, { "epoch": 0.06487977926685061, "grad_norm": 1.0752520561218262, "learning_rate": 0.0005, "loss": 3.3165643215179443, "step": 823 }, { "epoch": 0.06495861253448955, "grad_norm": 4.34546422958374, "learning_rate": 0.0005, "loss": 2.8333752155303955, "step": 824 }, { "epoch": 0.0650374458021285, "grad_norm": 1.8326221704483032, "learning_rate": 0.0005, "loss": 3.329925060272217, "step": 825 }, { "epoch": 0.06511627906976744, "grad_norm": 0.917302668094635, "learning_rate": 0.0005, "loss": 3.3077235221862793, "step": 826 }, { "epoch": 0.06519511233740638, "grad_norm": 1.081298589706421, "learning_rate": 0.0005, "loss": 3.103416919708252, "step": 827 }, { "epoch": 0.06527394560504533, "grad_norm": 1.1612070798873901, "learning_rate": 0.0005, "loss": 3.144096851348877, "step": 828 }, { "epoch": 0.06535277887268427, "grad_norm": 1.174181342124939, "learning_rate": 0.0005, "loss": 2.8308534622192383, "step": 829 }, { "epoch": 0.06543161214032321, "grad_norm": 0.8360102772712708, "learning_rate": 0.0005, "loss": 3.282762050628662, "step": 830 }, { "epoch": 0.06551044540796216, "grad_norm": 1.0840063095092773, "learning_rate": 0.0005, "loss": 3.1002912521362305, "step": 831 }, { "epoch": 0.0655892786756011, "grad_norm": 1.2906248569488525, "learning_rate": 0.0005, "loss": 3.251671314239502, "step": 832 }, { "epoch": 0.06566811194324004, "grad_norm": 1.3378770351409912, "learning_rate": 0.0005, "loss": 2.8527708053588867, "step": 833 }, { "epoch": 0.065746945210879, "grad_norm": 1.1485854387283325, "learning_rate": 0.0005, "loss": 3.02158784866333, "step": 834 }, { "epoch": 0.06582577847851794, "grad_norm": 1.2157154083251953, "learning_rate": 0.0005, "loss": 2.715120553970337, "step": 835 }, { "epoch": 0.06590461174615687, "grad_norm": 1.1206798553466797, "learning_rate": 0.0005, "loss": 3.280808925628662, "step": 836 }, { "epoch": 0.06598344501379583, "grad_norm": 1.1130961179733276, "learning_rate": 0.0005, "loss": 3.0136568546295166, "step": 837 }, { "epoch": 0.06606227828143477, "grad_norm": 1.4980478286743164, "learning_rate": 0.0005, "loss": 3.202317476272583, "step": 838 }, { "epoch": 0.0661411115490737, "grad_norm": 0.8976702094078064, "learning_rate": 0.0005, "loss": 3.208484172821045, "step": 839 }, { "epoch": 0.06621994481671266, "grad_norm": 1.1329405307769775, "learning_rate": 0.0005, "loss": 3.0839433670043945, "step": 840 }, { "epoch": 0.0662987780843516, "grad_norm": 1.415433406829834, "learning_rate": 0.0005, "loss": 3.1664037704467773, "step": 841 }, { "epoch": 0.06637761135199054, "grad_norm": 0.9994104504585266, "learning_rate": 0.0005, "loss": 3.106442928314209, "step": 842 }, { "epoch": 0.06645644461962949, "grad_norm": 1.9123039245605469, "learning_rate": 0.0005, "loss": 2.9298319816589355, "step": 843 }, { "epoch": 0.06653527788726843, "grad_norm": 1.3474290370941162, "learning_rate": 0.0005, "loss": 2.9375288486480713, "step": 844 }, { "epoch": 0.06661411115490737, "grad_norm": 1.0147053003311157, "learning_rate": 0.0005, "loss": 3.203993558883667, "step": 845 }, { "epoch": 0.06669294442254632, "grad_norm": 1.2915996313095093, "learning_rate": 0.0005, "loss": 3.373414993286133, "step": 846 }, { "epoch": 0.06677177769018526, "grad_norm": 1.1002376079559326, "learning_rate": 0.0005, "loss": 3.1784415245056152, "step": 847 }, { "epoch": 0.0668506109578242, "grad_norm": 1.1744657754898071, "learning_rate": 0.0005, "loss": 3.086735963821411, "step": 848 }, { "epoch": 0.06692944422546314, "grad_norm": 4.021763801574707, "learning_rate": 0.0005, "loss": 3.499570846557617, "step": 849 }, { "epoch": 0.06700827749310209, "grad_norm": 1.1262073516845703, "learning_rate": 0.0005, "loss": 3.0993943214416504, "step": 850 }, { "epoch": 0.06708711076074103, "grad_norm": 1.8941866159439087, "learning_rate": 0.0005, "loss": 2.8615670204162598, "step": 851 }, { "epoch": 0.06716594402837997, "grad_norm": 1.2111505270004272, "learning_rate": 0.0005, "loss": 2.672792434692383, "step": 852 }, { "epoch": 0.06724477729601892, "grad_norm": 1.361624836921692, "learning_rate": 0.0005, "loss": 3.192959785461426, "step": 853 }, { "epoch": 0.06732361056365786, "grad_norm": 1.7368800640106201, "learning_rate": 0.0005, "loss": 3.191533327102661, "step": 854 }, { "epoch": 0.0674024438312968, "grad_norm": 1.074380874633789, "learning_rate": 0.0005, "loss": 3.460824966430664, "step": 855 }, { "epoch": 0.06748127709893575, "grad_norm": 1.0732905864715576, "learning_rate": 0.0005, "loss": 2.7958664894104004, "step": 856 }, { "epoch": 0.0675601103665747, "grad_norm": 1.2812381982803345, "learning_rate": 0.0005, "loss": 3.221665859222412, "step": 857 }, { "epoch": 0.06763894363421363, "grad_norm": 1.2013728618621826, "learning_rate": 0.0005, "loss": 3.246583938598633, "step": 858 }, { "epoch": 0.06771777690185259, "grad_norm": 1.3102126121520996, "learning_rate": 0.0005, "loss": 3.6392033100128174, "step": 859 }, { "epoch": 0.06779661016949153, "grad_norm": 1.2025591135025024, "learning_rate": 0.0005, "loss": 3.4717800617218018, "step": 860 }, { "epoch": 0.06787544343713046, "grad_norm": 1.1169692277908325, "learning_rate": 0.0005, "loss": 3.278991222381592, "step": 861 }, { "epoch": 0.06795427670476942, "grad_norm": 1.405998706817627, "learning_rate": 0.0005, "loss": 3.1049957275390625, "step": 862 }, { "epoch": 0.06803310997240836, "grad_norm": 1.0360957384109497, "learning_rate": 0.0005, "loss": 3.1890158653259277, "step": 863 }, { "epoch": 0.0681119432400473, "grad_norm": 1.2594646215438843, "learning_rate": 0.0005, "loss": 3.4939327239990234, "step": 864 }, { "epoch": 0.06819077650768625, "grad_norm": 1.1367897987365723, "learning_rate": 0.0005, "loss": 3.2477049827575684, "step": 865 }, { "epoch": 0.06826960977532519, "grad_norm": 1.143682837486267, "learning_rate": 0.0005, "loss": 3.0855870246887207, "step": 866 }, { "epoch": 0.06834844304296413, "grad_norm": 1.2022114992141724, "learning_rate": 0.0005, "loss": 3.1816976070404053, "step": 867 }, { "epoch": 0.06842727631060308, "grad_norm": 1.2335261106491089, "learning_rate": 0.0005, "loss": 3.570833683013916, "step": 868 }, { "epoch": 0.06850610957824202, "grad_norm": 1.1869211196899414, "learning_rate": 0.0005, "loss": 3.344054698944092, "step": 869 }, { "epoch": 0.06858494284588096, "grad_norm": 1.3451533317565918, "learning_rate": 0.0005, "loss": 3.2416210174560547, "step": 870 }, { "epoch": 0.06866377611351991, "grad_norm": 1.6142551898956299, "learning_rate": 0.0005, "loss": 3.252108573913574, "step": 871 }, { "epoch": 0.06874260938115885, "grad_norm": 1.2289860248565674, "learning_rate": 0.0005, "loss": 3.3543753623962402, "step": 872 }, { "epoch": 0.06882144264879779, "grad_norm": 1.414291501045227, "learning_rate": 0.0005, "loss": 2.8436214923858643, "step": 873 }, { "epoch": 0.06890027591643674, "grad_norm": 1.2150654792785645, "learning_rate": 0.0005, "loss": 3.0666308403015137, "step": 874 }, { "epoch": 0.06897910918407568, "grad_norm": 1.1882822513580322, "learning_rate": 0.0005, "loss": 2.9070072174072266, "step": 875 }, { "epoch": 0.06905794245171462, "grad_norm": 0.9231125116348267, "learning_rate": 0.0005, "loss": 2.857083559036255, "step": 876 }, { "epoch": 0.06913677571935356, "grad_norm": 1.2122952938079834, "learning_rate": 0.0005, "loss": 3.167937755584717, "step": 877 }, { "epoch": 0.06921560898699251, "grad_norm": 0.9486527442932129, "learning_rate": 0.0005, "loss": 3.303269863128662, "step": 878 }, { "epoch": 0.06929444225463145, "grad_norm": 1.4468621015548706, "learning_rate": 0.0005, "loss": 2.893702983856201, "step": 879 }, { "epoch": 0.06937327552227039, "grad_norm": 1.1021040678024292, "learning_rate": 0.0005, "loss": 3.4536664485931396, "step": 880 }, { "epoch": 0.06945210878990934, "grad_norm": 1.3470629453659058, "learning_rate": 0.0005, "loss": 3.2079806327819824, "step": 881 }, { "epoch": 0.06953094205754828, "grad_norm": 1.2880066633224487, "learning_rate": 0.0005, "loss": 3.08547306060791, "step": 882 }, { "epoch": 0.06960977532518722, "grad_norm": 1.4935507774353027, "learning_rate": 0.0005, "loss": 3.3390188217163086, "step": 883 }, { "epoch": 0.06968860859282618, "grad_norm": 1.238304615020752, "learning_rate": 0.0005, "loss": 3.322190284729004, "step": 884 }, { "epoch": 0.06976744186046512, "grad_norm": 1.425707459449768, "learning_rate": 0.0005, "loss": 3.1317481994628906, "step": 885 }, { "epoch": 0.06984627512810405, "grad_norm": 2.054145574569702, "learning_rate": 0.0005, "loss": 3.176985502243042, "step": 886 }, { "epoch": 0.06992510839574301, "grad_norm": 1.397900938987732, "learning_rate": 0.0005, "loss": 3.006704092025757, "step": 887 }, { "epoch": 0.07000394166338195, "grad_norm": 1.140101671218872, "learning_rate": 0.0005, "loss": 3.2179861068725586, "step": 888 }, { "epoch": 0.07008277493102089, "grad_norm": 1.3137716054916382, "learning_rate": 0.0005, "loss": 3.030200958251953, "step": 889 }, { "epoch": 0.07016160819865984, "grad_norm": 1.24949312210083, "learning_rate": 0.0005, "loss": 3.1861491203308105, "step": 890 }, { "epoch": 0.07024044146629878, "grad_norm": 1.2081106901168823, "learning_rate": 0.0005, "loss": 3.0359461307525635, "step": 891 }, { "epoch": 0.07031927473393772, "grad_norm": 1.2633615732192993, "learning_rate": 0.0005, "loss": 3.2747108936309814, "step": 892 }, { "epoch": 0.07039810800157667, "grad_norm": 1.1803486347198486, "learning_rate": 0.0005, "loss": 2.9536309242248535, "step": 893 }, { "epoch": 0.07047694126921561, "grad_norm": 1.5596733093261719, "learning_rate": 0.0005, "loss": 3.4154791831970215, "step": 894 }, { "epoch": 0.07055577453685455, "grad_norm": 1.1881179809570312, "learning_rate": 0.0005, "loss": 2.8744258880615234, "step": 895 }, { "epoch": 0.0706346078044935, "grad_norm": 1.3868352174758911, "learning_rate": 0.0005, "loss": 3.135833740234375, "step": 896 }, { "epoch": 0.07071344107213244, "grad_norm": 1.1814906597137451, "learning_rate": 0.0005, "loss": 2.9982352256774902, "step": 897 }, { "epoch": 0.07079227433977138, "grad_norm": 1.1015738248825073, "learning_rate": 0.0005, "loss": 3.1765246391296387, "step": 898 }, { "epoch": 0.07087110760741033, "grad_norm": 1.2820159196853638, "learning_rate": 0.0005, "loss": 2.9477953910827637, "step": 899 }, { "epoch": 0.07094994087504927, "grad_norm": 1.2983763217926025, "learning_rate": 0.0005, "loss": 2.8726143836975098, "step": 900 }, { "epoch": 0.07102877414268821, "grad_norm": 1.3693171739578247, "learning_rate": 0.0005, "loss": 3.2234811782836914, "step": 901 }, { "epoch": 0.07110760741032716, "grad_norm": 1.0623842477798462, "learning_rate": 0.0005, "loss": 2.977499008178711, "step": 902 }, { "epoch": 0.0711864406779661, "grad_norm": 1.5958770513534546, "learning_rate": 0.0005, "loss": 3.0893917083740234, "step": 903 }, { "epoch": 0.07126527394560504, "grad_norm": 1.3785885572433472, "learning_rate": 0.0005, "loss": 3.629757881164551, "step": 904 }, { "epoch": 0.071344107213244, "grad_norm": 2.4800643920898438, "learning_rate": 0.0005, "loss": 2.9046995639801025, "step": 905 }, { "epoch": 0.07142294048088293, "grad_norm": 1.2692975997924805, "learning_rate": 0.0005, "loss": 3.4760687351226807, "step": 906 }, { "epoch": 0.07150177374852187, "grad_norm": 1.4601043462753296, "learning_rate": 0.0005, "loss": 2.9359054565429688, "step": 907 }, { "epoch": 0.07158060701616081, "grad_norm": 1.2195020914077759, "learning_rate": 0.0005, "loss": 3.254392147064209, "step": 908 }, { "epoch": 0.07165944028379977, "grad_norm": 1.3354181051254272, "learning_rate": 0.0005, "loss": 3.1120643615722656, "step": 909 }, { "epoch": 0.0717382735514387, "grad_norm": 1.4178307056427002, "learning_rate": 0.0005, "loss": 3.1129603385925293, "step": 910 }, { "epoch": 0.07181710681907764, "grad_norm": 1.4759554862976074, "learning_rate": 0.0005, "loss": 3.3910350799560547, "step": 911 }, { "epoch": 0.0718959400867166, "grad_norm": 1.1425200700759888, "learning_rate": 0.0005, "loss": 3.364078998565674, "step": 912 }, { "epoch": 0.07197477335435554, "grad_norm": 1.4007335901260376, "learning_rate": 0.0005, "loss": 2.9100911617279053, "step": 913 }, { "epoch": 0.07205360662199448, "grad_norm": 1.3810186386108398, "learning_rate": 0.0005, "loss": 3.2390100955963135, "step": 914 }, { "epoch": 0.07213243988963343, "grad_norm": 1.1849579811096191, "learning_rate": 0.0005, "loss": 3.3380024433135986, "step": 915 }, { "epoch": 0.07221127315727237, "grad_norm": 1.4535021781921387, "learning_rate": 0.0005, "loss": 3.313838005065918, "step": 916 }, { "epoch": 0.07229010642491131, "grad_norm": 1.3673361539840698, "learning_rate": 0.0005, "loss": 2.812779426574707, "step": 917 }, { "epoch": 0.07236893969255026, "grad_norm": 1.3407318592071533, "learning_rate": 0.0005, "loss": 2.9177024364471436, "step": 918 }, { "epoch": 0.0724477729601892, "grad_norm": 1.1351492404937744, "learning_rate": 0.0005, "loss": 2.9856550693511963, "step": 919 }, { "epoch": 0.07252660622782814, "grad_norm": 1.1536202430725098, "learning_rate": 0.0005, "loss": 2.820295572280884, "step": 920 }, { "epoch": 0.07260543949546709, "grad_norm": 1.5710443258285522, "learning_rate": 0.0005, "loss": 3.0183486938476562, "step": 921 }, { "epoch": 0.07268427276310603, "grad_norm": 1.155363917350769, "learning_rate": 0.0005, "loss": 3.228464126586914, "step": 922 }, { "epoch": 0.07276310603074497, "grad_norm": 1.4227153062820435, "learning_rate": 0.0005, "loss": 3.4586400985717773, "step": 923 }, { "epoch": 0.07284193929838392, "grad_norm": 1.3271557092666626, "learning_rate": 0.0005, "loss": 3.426854133605957, "step": 924 }, { "epoch": 0.07292077256602286, "grad_norm": 1.0941342115402222, "learning_rate": 0.0005, "loss": 3.1290416717529297, "step": 925 }, { "epoch": 0.0729996058336618, "grad_norm": 3.0507559776306152, "learning_rate": 0.0005, "loss": 3.2619402408599854, "step": 926 }, { "epoch": 0.07307843910130075, "grad_norm": 1.2622557878494263, "learning_rate": 0.0005, "loss": 3.412173271179199, "step": 927 }, { "epoch": 0.0731572723689397, "grad_norm": 1.0824135541915894, "learning_rate": 0.0005, "loss": 3.1465702056884766, "step": 928 }, { "epoch": 0.07323610563657863, "grad_norm": 1.068688988685608, "learning_rate": 0.0005, "loss": 3.366117238998413, "step": 929 }, { "epoch": 0.07331493890421759, "grad_norm": 1.0076663494110107, "learning_rate": 0.0005, "loss": 3.4844603538513184, "step": 930 }, { "epoch": 0.07339377217185653, "grad_norm": 1.1704378128051758, "learning_rate": 0.0005, "loss": 3.1863131523132324, "step": 931 }, { "epoch": 0.07347260543949546, "grad_norm": 1.2543449401855469, "learning_rate": 0.0005, "loss": 3.0637218952178955, "step": 932 }, { "epoch": 0.07355143870713442, "grad_norm": 1.1156996488571167, "learning_rate": 0.0005, "loss": 2.9954915046691895, "step": 933 }, { "epoch": 0.07363027197477336, "grad_norm": 1.1025476455688477, "learning_rate": 0.0005, "loss": 2.8030738830566406, "step": 934 }, { "epoch": 0.0737091052424123, "grad_norm": 1.5246942043304443, "learning_rate": 0.0005, "loss": 3.2209320068359375, "step": 935 }, { "epoch": 0.07378793851005123, "grad_norm": 1.879978895187378, "learning_rate": 0.0005, "loss": 3.122823715209961, "step": 936 }, { "epoch": 0.07386677177769019, "grad_norm": 1.336842656135559, "learning_rate": 0.0005, "loss": 3.2404205799102783, "step": 937 }, { "epoch": 0.07394560504532913, "grad_norm": 1.0944854021072388, "learning_rate": 0.0005, "loss": 2.961282253265381, "step": 938 }, { "epoch": 0.07402443831296807, "grad_norm": 1.0923861265182495, "learning_rate": 0.0005, "loss": 3.0685601234436035, "step": 939 }, { "epoch": 0.07410327158060702, "grad_norm": 1.610080361366272, "learning_rate": 0.0005, "loss": 3.194310426712036, "step": 940 }, { "epoch": 0.07418210484824596, "grad_norm": 1.231990098953247, "learning_rate": 0.0005, "loss": 3.294778823852539, "step": 941 }, { "epoch": 0.0742609381158849, "grad_norm": 0.9522135257720947, "learning_rate": 0.0005, "loss": 3.4212646484375, "step": 942 }, { "epoch": 0.07433977138352385, "grad_norm": 1.2698856592178345, "learning_rate": 0.0005, "loss": 3.2774219512939453, "step": 943 }, { "epoch": 0.07441860465116279, "grad_norm": 1.3310643434524536, "learning_rate": 0.0005, "loss": 2.9572620391845703, "step": 944 }, { "epoch": 0.07449743791880173, "grad_norm": 1.367033839225769, "learning_rate": 0.0005, "loss": 3.481034278869629, "step": 945 }, { "epoch": 0.07457627118644068, "grad_norm": 1.7504467964172363, "learning_rate": 0.0005, "loss": 3.210509777069092, "step": 946 }, { "epoch": 0.07465510445407962, "grad_norm": 1.170526385307312, "learning_rate": 0.0005, "loss": 2.755326747894287, "step": 947 }, { "epoch": 0.07473393772171856, "grad_norm": 1.308091163635254, "learning_rate": 0.0005, "loss": 3.1261041164398193, "step": 948 }, { "epoch": 0.07481277098935751, "grad_norm": 1.2882987260818481, "learning_rate": 0.0005, "loss": 3.311953544616699, "step": 949 }, { "epoch": 0.07489160425699645, "grad_norm": 1.3338488340377808, "learning_rate": 0.0005, "loss": 3.185431480407715, "step": 950 }, { "epoch": 0.07497043752463539, "grad_norm": 1.066916823387146, "learning_rate": 0.0005, "loss": 3.5376992225646973, "step": 951 }, { "epoch": 0.07504927079227434, "grad_norm": 1.0247313976287842, "learning_rate": 0.0005, "loss": 3.1565029621124268, "step": 952 }, { "epoch": 0.07512810405991328, "grad_norm": 1.343047857284546, "learning_rate": 0.0005, "loss": 3.0713398456573486, "step": 953 }, { "epoch": 0.07520693732755222, "grad_norm": 1.16220223903656, "learning_rate": 0.0005, "loss": 3.1811468601226807, "step": 954 }, { "epoch": 0.07528577059519118, "grad_norm": 1.1008790731430054, "learning_rate": 0.0005, "loss": 3.011115550994873, "step": 955 }, { "epoch": 0.07536460386283012, "grad_norm": 0.999744176864624, "learning_rate": 0.0005, "loss": 2.597677230834961, "step": 956 }, { "epoch": 0.07544343713046905, "grad_norm": 1.2994199991226196, "learning_rate": 0.0005, "loss": 3.0379085540771484, "step": 957 }, { "epoch": 0.07552227039810801, "grad_norm": 1.3879998922348022, "learning_rate": 0.0005, "loss": 2.899226665496826, "step": 958 }, { "epoch": 0.07560110366574695, "grad_norm": 1.4387339353561401, "learning_rate": 0.0005, "loss": 3.1452322006225586, "step": 959 }, { "epoch": 0.07567993693338589, "grad_norm": 1.021905541419983, "learning_rate": 0.0005, "loss": 3.0773229598999023, "step": 960 }, { "epoch": 0.07575877020102484, "grad_norm": 1.1793630123138428, "learning_rate": 0.0005, "loss": 2.8917083740234375, "step": 961 }, { "epoch": 0.07583760346866378, "grad_norm": 1.0778381824493408, "learning_rate": 0.0005, "loss": 3.0946409702301025, "step": 962 }, { "epoch": 0.07591643673630272, "grad_norm": 1.3607594966888428, "learning_rate": 0.0005, "loss": 3.1830649375915527, "step": 963 }, { "epoch": 0.07599527000394167, "grad_norm": 1.3151938915252686, "learning_rate": 0.0005, "loss": 3.2300214767456055, "step": 964 }, { "epoch": 0.07607410327158061, "grad_norm": 1.2030454874038696, "learning_rate": 0.0005, "loss": 3.532623767852783, "step": 965 }, { "epoch": 0.07615293653921955, "grad_norm": 1.0945987701416016, "learning_rate": 0.0005, "loss": 3.1923625469207764, "step": 966 }, { "epoch": 0.07623176980685849, "grad_norm": 1.4291070699691772, "learning_rate": 0.0005, "loss": 3.081881284713745, "step": 967 }, { "epoch": 0.07631060307449744, "grad_norm": 1.3683977127075195, "learning_rate": 0.0005, "loss": 3.1685094833374023, "step": 968 }, { "epoch": 0.07638943634213638, "grad_norm": 0.9578132033348083, "learning_rate": 0.0005, "loss": 3.104588270187378, "step": 969 }, { "epoch": 0.07646826960977532, "grad_norm": 1.3649729490280151, "learning_rate": 0.0005, "loss": 2.946930408477783, "step": 970 }, { "epoch": 0.07654710287741427, "grad_norm": 1.0479130744934082, "learning_rate": 0.0005, "loss": 3.1777570247650146, "step": 971 }, { "epoch": 0.07662593614505321, "grad_norm": 1.213371753692627, "learning_rate": 0.0005, "loss": 3.2274231910705566, "step": 972 }, { "epoch": 0.07670476941269215, "grad_norm": 1.4408212900161743, "learning_rate": 0.0005, "loss": 3.026638984680176, "step": 973 }, { "epoch": 0.0767836026803311, "grad_norm": 1.0784167051315308, "learning_rate": 0.0005, "loss": 3.360280752182007, "step": 974 }, { "epoch": 0.07686243594797004, "grad_norm": 1.0114718675613403, "learning_rate": 0.0005, "loss": 3.4194180965423584, "step": 975 }, { "epoch": 0.07694126921560898, "grad_norm": 1.1074601411819458, "learning_rate": 0.0005, "loss": 3.118553400039673, "step": 976 }, { "epoch": 0.07702010248324793, "grad_norm": 1.084316611289978, "learning_rate": 0.0005, "loss": 2.8928399085998535, "step": 977 }, { "epoch": 0.07709893575088687, "grad_norm": 1.078086256980896, "learning_rate": 0.0005, "loss": 3.1864583492279053, "step": 978 }, { "epoch": 0.07717776901852581, "grad_norm": 2.1299915313720703, "learning_rate": 0.0005, "loss": 3.06984806060791, "step": 979 }, { "epoch": 0.07725660228616477, "grad_norm": 1.1538772583007812, "learning_rate": 0.0005, "loss": 3.5705108642578125, "step": 980 }, { "epoch": 0.0773354355538037, "grad_norm": 1.2883645296096802, "learning_rate": 0.0005, "loss": 3.1134772300720215, "step": 981 }, { "epoch": 0.07741426882144264, "grad_norm": 1.2733180522918701, "learning_rate": 0.0005, "loss": 2.7787094116210938, "step": 982 }, { "epoch": 0.0774931020890816, "grad_norm": 1.1056376695632935, "learning_rate": 0.0005, "loss": 3.058624505996704, "step": 983 }, { "epoch": 0.07757193535672054, "grad_norm": 1.283794641494751, "learning_rate": 0.0005, "loss": 2.890331745147705, "step": 984 }, { "epoch": 0.07765076862435948, "grad_norm": 1.5299283266067505, "learning_rate": 0.0005, "loss": 3.173586368560791, "step": 985 }, { "epoch": 0.07772960189199843, "grad_norm": 1.3253647089004517, "learning_rate": 0.0005, "loss": 3.252891778945923, "step": 986 }, { "epoch": 0.07780843515963737, "grad_norm": 2.0436794757843018, "learning_rate": 0.0005, "loss": 2.9001219272613525, "step": 987 }, { "epoch": 0.07788726842727631, "grad_norm": 1.478078842163086, "learning_rate": 0.0005, "loss": 2.5080981254577637, "step": 988 }, { "epoch": 0.07796610169491526, "grad_norm": 1.3740044832229614, "learning_rate": 0.0005, "loss": 3.332953929901123, "step": 989 }, { "epoch": 0.0780449349625542, "grad_norm": 1.164228081703186, "learning_rate": 0.0005, "loss": 3.1132731437683105, "step": 990 }, { "epoch": 0.07812376823019314, "grad_norm": 1.37943696975708, "learning_rate": 0.0005, "loss": 3.036527633666992, "step": 991 }, { "epoch": 0.07820260149783209, "grad_norm": 1.1733503341674805, "learning_rate": 0.0005, "loss": 2.919379234313965, "step": 992 }, { "epoch": 0.07828143476547103, "grad_norm": 1.0557959079742432, "learning_rate": 0.0005, "loss": 3.441317081451416, "step": 993 }, { "epoch": 0.07836026803310997, "grad_norm": 1.2797067165374756, "learning_rate": 0.0005, "loss": 3.2321767807006836, "step": 994 }, { "epoch": 0.07843910130074892, "grad_norm": 1.0897945165634155, "learning_rate": 0.0005, "loss": 2.8040714263916016, "step": 995 }, { "epoch": 0.07851793456838786, "grad_norm": 1.6258233785629272, "learning_rate": 0.0005, "loss": 2.9194083213806152, "step": 996 }, { "epoch": 0.0785967678360268, "grad_norm": 1.9764398336410522, "learning_rate": 0.0005, "loss": 3.3017845153808594, "step": 997 }, { "epoch": 0.07867560110366574, "grad_norm": 1.2225991487503052, "learning_rate": 0.0005, "loss": 2.8507232666015625, "step": 998 }, { "epoch": 0.0787544343713047, "grad_norm": 1.2376258373260498, "learning_rate": 0.0005, "loss": 2.9852232933044434, "step": 999 }, { "epoch": 0.07883326763894363, "grad_norm": 1.0639050006866455, "learning_rate": 0.0005, "loss": 3.13029408454895, "step": 1000 }, { "epoch": 0.07891210090658257, "grad_norm": 1.2016143798828125, "learning_rate": 0.0005, "loss": 3.306429862976074, "step": 1001 }, { "epoch": 0.07899093417422152, "grad_norm": 1.1699377298355103, "learning_rate": 0.0005, "loss": 3.23884916305542, "step": 1002 }, { "epoch": 0.07906976744186046, "grad_norm": 1.1108462810516357, "learning_rate": 0.0005, "loss": 3.1126785278320312, "step": 1003 }, { "epoch": 0.0791486007094994, "grad_norm": 1.2897459268569946, "learning_rate": 0.0005, "loss": 3.3380441665649414, "step": 1004 }, { "epoch": 0.07922743397713836, "grad_norm": 1.1575543880462646, "learning_rate": 0.0005, "loss": 2.8489232063293457, "step": 1005 }, { "epoch": 0.0793062672447773, "grad_norm": 1.0051369667053223, "learning_rate": 0.0005, "loss": 2.886110782623291, "step": 1006 }, { "epoch": 0.07938510051241623, "grad_norm": 1.0554802417755127, "learning_rate": 0.0005, "loss": 3.1077942848205566, "step": 1007 }, { "epoch": 0.07946393378005519, "grad_norm": 1.3044250011444092, "learning_rate": 0.0005, "loss": 3.1363027095794678, "step": 1008 }, { "epoch": 0.07954276704769413, "grad_norm": 1.301828384399414, "learning_rate": 0.0005, "loss": 3.327144145965576, "step": 1009 }, { "epoch": 0.07962160031533307, "grad_norm": 1.428544521331787, "learning_rate": 0.0005, "loss": 3.1973965167999268, "step": 1010 }, { "epoch": 0.07970043358297202, "grad_norm": 1.595584511756897, "learning_rate": 0.0005, "loss": 3.216010332107544, "step": 1011 }, { "epoch": 0.07977926685061096, "grad_norm": 1.3708326816558838, "learning_rate": 0.0005, "loss": 3.063669204711914, "step": 1012 }, { "epoch": 0.0798581001182499, "grad_norm": 1.1710975170135498, "learning_rate": 0.0005, "loss": 2.9951107501983643, "step": 1013 }, { "epoch": 0.07993693338588885, "grad_norm": 2.240826368331909, "learning_rate": 0.0005, "loss": 3.3434691429138184, "step": 1014 }, { "epoch": 0.08001576665352779, "grad_norm": 1.234575867652893, "learning_rate": 0.0005, "loss": 2.935925006866455, "step": 1015 }, { "epoch": 0.08009459992116673, "grad_norm": 1.3338464498519897, "learning_rate": 0.0005, "loss": 3.0324275493621826, "step": 1016 }, { "epoch": 0.08017343318880568, "grad_norm": 1.158034324645996, "learning_rate": 0.0005, "loss": 2.7733113765716553, "step": 1017 }, { "epoch": 0.08025226645644462, "grad_norm": 1.8310140371322632, "learning_rate": 0.0005, "loss": 3.1053466796875, "step": 1018 }, { "epoch": 0.08033109972408356, "grad_norm": 1.6074248552322388, "learning_rate": 0.0005, "loss": 3.1823644638061523, "step": 1019 }, { "epoch": 0.08040993299172251, "grad_norm": 1.0683153867721558, "learning_rate": 0.0005, "loss": 3.359265089035034, "step": 1020 }, { "epoch": 0.08048876625936145, "grad_norm": 1.3034896850585938, "learning_rate": 0.0005, "loss": 3.1069886684417725, "step": 1021 }, { "epoch": 0.08056759952700039, "grad_norm": 1.3999178409576416, "learning_rate": 0.0005, "loss": 3.197096586227417, "step": 1022 }, { "epoch": 0.08064643279463934, "grad_norm": 1.1960713863372803, "learning_rate": 0.0005, "loss": 2.834275007247925, "step": 1023 }, { "epoch": 0.08072526606227828, "grad_norm": 1.2772685289382935, "learning_rate": 0.0005, "loss": 3.3945584297180176, "step": 1024 }, { "epoch": 0.08080409932991722, "grad_norm": 1.37058424949646, "learning_rate": 0.0005, "loss": 3.122969150543213, "step": 1025 }, { "epoch": 0.08088293259755616, "grad_norm": 1.372566819190979, "learning_rate": 0.0005, "loss": 2.7703604698181152, "step": 1026 }, { "epoch": 0.08096176586519511, "grad_norm": 1.0992217063903809, "learning_rate": 0.0005, "loss": 2.890817642211914, "step": 1027 }, { "epoch": 0.08104059913283405, "grad_norm": 1.0392658710479736, "learning_rate": 0.0005, "loss": 3.411524534225464, "step": 1028 }, { "epoch": 0.081119432400473, "grad_norm": 1.4833533763885498, "learning_rate": 0.0005, "loss": 3.023159980773926, "step": 1029 }, { "epoch": 0.08119826566811195, "grad_norm": 1.1987437009811401, "learning_rate": 0.0005, "loss": 3.4392032623291016, "step": 1030 }, { "epoch": 0.08127709893575089, "grad_norm": 1.1658298969268799, "learning_rate": 0.0005, "loss": 3.44301176071167, "step": 1031 }, { "epoch": 0.08135593220338982, "grad_norm": 2.557558536529541, "learning_rate": 0.0005, "loss": 2.9412641525268555, "step": 1032 }, { "epoch": 0.08143476547102878, "grad_norm": 1.2577357292175293, "learning_rate": 0.0005, "loss": 3.5655007362365723, "step": 1033 }, { "epoch": 0.08151359873866772, "grad_norm": 1.1145280599594116, "learning_rate": 0.0005, "loss": 3.343935251235962, "step": 1034 }, { "epoch": 0.08159243200630666, "grad_norm": 1.3343366384506226, "learning_rate": 0.0005, "loss": 3.015669107437134, "step": 1035 }, { "epoch": 0.08167126527394561, "grad_norm": 1.1890002489089966, "learning_rate": 0.0005, "loss": 2.925675868988037, "step": 1036 }, { "epoch": 0.08175009854158455, "grad_norm": 1.032173991203308, "learning_rate": 0.0005, "loss": 2.892131805419922, "step": 1037 }, { "epoch": 0.08182893180922349, "grad_norm": 1.2284785509109497, "learning_rate": 0.0005, "loss": 3.075679302215576, "step": 1038 }, { "epoch": 0.08190776507686244, "grad_norm": 1.121156096458435, "learning_rate": 0.0005, "loss": 3.077136516571045, "step": 1039 }, { "epoch": 0.08198659834450138, "grad_norm": 1.1981699466705322, "learning_rate": 0.0005, "loss": 3.4314002990722656, "step": 1040 }, { "epoch": 0.08206543161214032, "grad_norm": 1.0881319046020508, "learning_rate": 0.0005, "loss": 3.207157611846924, "step": 1041 }, { "epoch": 0.08214426487977927, "grad_norm": 1.1934243440628052, "learning_rate": 0.0005, "loss": 3.1947972774505615, "step": 1042 }, { "epoch": 0.08222309814741821, "grad_norm": 1.1312460899353027, "learning_rate": 0.0005, "loss": 2.9738621711730957, "step": 1043 }, { "epoch": 0.08230193141505715, "grad_norm": 1.3447847366333008, "learning_rate": 0.0005, "loss": 2.9973340034484863, "step": 1044 }, { "epoch": 0.0823807646826961, "grad_norm": 1.1484850645065308, "learning_rate": 0.0005, "loss": 3.4518532752990723, "step": 1045 }, { "epoch": 0.08245959795033504, "grad_norm": 1.5305362939834595, "learning_rate": 0.0005, "loss": 3.4463601112365723, "step": 1046 }, { "epoch": 0.08253843121797398, "grad_norm": 1.0515124797821045, "learning_rate": 0.0005, "loss": 3.3031344413757324, "step": 1047 }, { "epoch": 0.08261726448561293, "grad_norm": 1.0699748992919922, "learning_rate": 0.0005, "loss": 3.131286144256592, "step": 1048 }, { "epoch": 0.08269609775325187, "grad_norm": 1.2877318859100342, "learning_rate": 0.0005, "loss": 3.1413841247558594, "step": 1049 }, { "epoch": 0.08277493102089081, "grad_norm": 1.5138479471206665, "learning_rate": 0.0005, "loss": 2.8376832008361816, "step": 1050 }, { "epoch": 0.08285376428852977, "grad_norm": 1.1763598918914795, "learning_rate": 0.0005, "loss": 2.991037130355835, "step": 1051 }, { "epoch": 0.0829325975561687, "grad_norm": 1.1327147483825684, "learning_rate": 0.0005, "loss": 3.1451895236968994, "step": 1052 }, { "epoch": 0.08301143082380764, "grad_norm": 1.2434087991714478, "learning_rate": 0.0005, "loss": 3.4694182872772217, "step": 1053 }, { "epoch": 0.0830902640914466, "grad_norm": 2.0303354263305664, "learning_rate": 0.0005, "loss": 3.24540376663208, "step": 1054 }, { "epoch": 0.08316909735908554, "grad_norm": 1.0709563493728638, "learning_rate": 0.0005, "loss": 2.735656976699829, "step": 1055 }, { "epoch": 0.08324793062672448, "grad_norm": 1.1591917276382446, "learning_rate": 0.0005, "loss": 3.109431743621826, "step": 1056 }, { "epoch": 0.08332676389436341, "grad_norm": 1.0627613067626953, "learning_rate": 0.0005, "loss": 3.2251648902893066, "step": 1057 }, { "epoch": 0.08340559716200237, "grad_norm": 1.0343950986862183, "learning_rate": 0.0005, "loss": 3.213254451751709, "step": 1058 }, { "epoch": 0.08348443042964131, "grad_norm": 1.0812188386917114, "learning_rate": 0.0005, "loss": 3.2256338596343994, "step": 1059 }, { "epoch": 0.08356326369728025, "grad_norm": 1.5320290327072144, "learning_rate": 0.0005, "loss": 3.010056972503662, "step": 1060 }, { "epoch": 0.0836420969649192, "grad_norm": 1.237902283668518, "learning_rate": 0.0005, "loss": 3.11653733253479, "step": 1061 }, { "epoch": 0.08372093023255814, "grad_norm": 1.1566627025604248, "learning_rate": 0.0005, "loss": 3.1811766624450684, "step": 1062 }, { "epoch": 0.08379976350019708, "grad_norm": 1.4974268674850464, "learning_rate": 0.0005, "loss": 3.3219332695007324, "step": 1063 }, { "epoch": 0.08387859676783603, "grad_norm": 0.9272834658622742, "learning_rate": 0.0005, "loss": 2.986525058746338, "step": 1064 }, { "epoch": 0.08395743003547497, "grad_norm": 1.4614094495773315, "learning_rate": 0.0005, "loss": 2.950669288635254, "step": 1065 }, { "epoch": 0.08403626330311391, "grad_norm": 1.2426388263702393, "learning_rate": 0.0005, "loss": 3.129157543182373, "step": 1066 }, { "epoch": 0.08411509657075286, "grad_norm": 1.1661856174468994, "learning_rate": 0.0005, "loss": 3.201805830001831, "step": 1067 }, { "epoch": 0.0841939298383918, "grad_norm": 1.1791131496429443, "learning_rate": 0.0005, "loss": 3.2031736373901367, "step": 1068 }, { "epoch": 0.08427276310603074, "grad_norm": 1.1759933233261108, "learning_rate": 0.0005, "loss": 3.050448417663574, "step": 1069 }, { "epoch": 0.0843515963736697, "grad_norm": 0.9673504829406738, "learning_rate": 0.0005, "loss": 3.3011717796325684, "step": 1070 }, { "epoch": 0.08443042964130863, "grad_norm": 1.0411590337753296, "learning_rate": 0.0005, "loss": 2.878288984298706, "step": 1071 }, { "epoch": 0.08450926290894757, "grad_norm": 1.1355931758880615, "learning_rate": 0.0005, "loss": 2.7756712436676025, "step": 1072 }, { "epoch": 0.08458809617658652, "grad_norm": 1.2944172620773315, "learning_rate": 0.0005, "loss": 2.8177900314331055, "step": 1073 }, { "epoch": 0.08466692944422546, "grad_norm": 1.8331432342529297, "learning_rate": 0.0005, "loss": 3.331371307373047, "step": 1074 }, { "epoch": 0.0847457627118644, "grad_norm": 1.1777969598770142, "learning_rate": 0.0005, "loss": 3.4311141967773438, "step": 1075 }, { "epoch": 0.08482459597950336, "grad_norm": 1.3185285329818726, "learning_rate": 0.0005, "loss": 3.305233955383301, "step": 1076 }, { "epoch": 0.0849034292471423, "grad_norm": 1.3614827394485474, "learning_rate": 0.0005, "loss": 3.2321829795837402, "step": 1077 }, { "epoch": 0.08498226251478123, "grad_norm": 1.1656452417373657, "learning_rate": 0.0005, "loss": 3.32401704788208, "step": 1078 }, { "epoch": 0.08506109578242019, "grad_norm": 1.8012793064117432, "learning_rate": 0.0005, "loss": 2.8122692108154297, "step": 1079 }, { "epoch": 0.08513992905005913, "grad_norm": 1.0697658061981201, "learning_rate": 0.0005, "loss": 3.125065326690674, "step": 1080 }, { "epoch": 0.08521876231769807, "grad_norm": 1.6363316774368286, "learning_rate": 0.0005, "loss": 2.756337881088257, "step": 1081 }, { "epoch": 0.08529759558533702, "grad_norm": 1.2882808446884155, "learning_rate": 0.0005, "loss": 2.996391773223877, "step": 1082 }, { "epoch": 0.08537642885297596, "grad_norm": 1.101908564567566, "learning_rate": 0.0005, "loss": 2.9974887371063232, "step": 1083 }, { "epoch": 0.0854552621206149, "grad_norm": 1.3465548753738403, "learning_rate": 0.0005, "loss": 2.987428665161133, "step": 1084 }, { "epoch": 0.08553409538825384, "grad_norm": 1.3266652822494507, "learning_rate": 0.0005, "loss": 3.358883857727051, "step": 1085 }, { "epoch": 0.08561292865589279, "grad_norm": 1.1527392864227295, "learning_rate": 0.0005, "loss": 3.268697500228882, "step": 1086 }, { "epoch": 0.08569176192353173, "grad_norm": 0.9989957809448242, "learning_rate": 0.0005, "loss": 3.5150110721588135, "step": 1087 }, { "epoch": 0.08577059519117067, "grad_norm": 1.0066522359848022, "learning_rate": 0.0005, "loss": 2.813066005706787, "step": 1088 }, { "epoch": 0.08584942845880962, "grad_norm": 1.142954707145691, "learning_rate": 0.0005, "loss": 3.4508326053619385, "step": 1089 }, { "epoch": 0.08592826172644856, "grad_norm": 1.2050247192382812, "learning_rate": 0.0005, "loss": 3.2976057529449463, "step": 1090 }, { "epoch": 0.0860070949940875, "grad_norm": 1.2589716911315918, "learning_rate": 0.0005, "loss": 3.4350428581237793, "step": 1091 }, { "epoch": 0.08608592826172645, "grad_norm": 1.4289143085479736, "learning_rate": 0.0005, "loss": 3.0249128341674805, "step": 1092 }, { "epoch": 0.08616476152936539, "grad_norm": 1.1710901260375977, "learning_rate": 0.0005, "loss": 3.1161062717437744, "step": 1093 }, { "epoch": 0.08624359479700433, "grad_norm": 1.2228747606277466, "learning_rate": 0.0005, "loss": 3.518010139465332, "step": 1094 }, { "epoch": 0.08632242806464328, "grad_norm": 1.022835612297058, "learning_rate": 0.0005, "loss": 2.6985251903533936, "step": 1095 }, { "epoch": 0.08640126133228222, "grad_norm": 1.613154649734497, "learning_rate": 0.0005, "loss": 3.0326876640319824, "step": 1096 }, { "epoch": 0.08648009459992116, "grad_norm": 1.060630440711975, "learning_rate": 0.0005, "loss": 3.040642738342285, "step": 1097 }, { "epoch": 0.08655892786756011, "grad_norm": 1.0501972436904907, "learning_rate": 0.0005, "loss": 3.434267282485962, "step": 1098 }, { "epoch": 0.08663776113519905, "grad_norm": 1.3884506225585938, "learning_rate": 0.0005, "loss": 2.9396820068359375, "step": 1099 }, { "epoch": 0.086716594402838, "grad_norm": 1.1347695589065552, "learning_rate": 0.0005, "loss": 2.8728127479553223, "step": 1100 }, { "epoch": 0.08679542767047695, "grad_norm": 14.506518363952637, "learning_rate": 0.0005, "loss": 3.2759523391723633, "step": 1101 }, { "epoch": 0.08687426093811589, "grad_norm": 1.2038371562957764, "learning_rate": 0.0005, "loss": 2.8811988830566406, "step": 1102 }, { "epoch": 0.08695309420575482, "grad_norm": 1.7606310844421387, "learning_rate": 0.0005, "loss": 3.160069704055786, "step": 1103 }, { "epoch": 0.08703192747339378, "grad_norm": 1.2198244333267212, "learning_rate": 0.0005, "loss": 3.153092622756958, "step": 1104 }, { "epoch": 0.08711076074103272, "grad_norm": 1.333132028579712, "learning_rate": 0.0005, "loss": 3.084404706954956, "step": 1105 }, { "epoch": 0.08718959400867166, "grad_norm": 1.4581928253173828, "learning_rate": 0.0005, "loss": 3.4443442821502686, "step": 1106 }, { "epoch": 0.08726842727631061, "grad_norm": 1.3214185237884521, "learning_rate": 0.0005, "loss": 3.202038049697876, "step": 1107 }, { "epoch": 0.08734726054394955, "grad_norm": 1.078088641166687, "learning_rate": 0.0005, "loss": 3.3407111167907715, "step": 1108 }, { "epoch": 0.08742609381158849, "grad_norm": 1.1289641857147217, "learning_rate": 0.0005, "loss": 3.490628242492676, "step": 1109 }, { "epoch": 0.08750492707922744, "grad_norm": 1.046871542930603, "learning_rate": 0.0005, "loss": 2.925234794616699, "step": 1110 }, { "epoch": 0.08758376034686638, "grad_norm": 1.0994123220443726, "learning_rate": 0.0005, "loss": 2.7193901538848877, "step": 1111 }, { "epoch": 0.08766259361450532, "grad_norm": 1.1237962245941162, "learning_rate": 0.0005, "loss": 3.132009506225586, "step": 1112 }, { "epoch": 0.08774142688214427, "grad_norm": 8.665572166442871, "learning_rate": 0.0005, "loss": 3.4060139656066895, "step": 1113 }, { "epoch": 0.08782026014978321, "grad_norm": 1.558756947517395, "learning_rate": 0.0005, "loss": 3.465808391571045, "step": 1114 }, { "epoch": 0.08789909341742215, "grad_norm": 1.3271784782409668, "learning_rate": 0.0005, "loss": 3.134253978729248, "step": 1115 }, { "epoch": 0.08797792668506109, "grad_norm": 1.3142518997192383, "learning_rate": 0.0005, "loss": 3.2160146236419678, "step": 1116 }, { "epoch": 0.08805675995270004, "grad_norm": 1.0498778820037842, "learning_rate": 0.0005, "loss": 2.945786714553833, "step": 1117 }, { "epoch": 0.08813559322033898, "grad_norm": 1.1192703247070312, "learning_rate": 0.0005, "loss": 3.269731044769287, "step": 1118 }, { "epoch": 0.08821442648797792, "grad_norm": 1.2171969413757324, "learning_rate": 0.0005, "loss": 2.9625940322875977, "step": 1119 }, { "epoch": 0.08829325975561687, "grad_norm": 1.2573882341384888, "learning_rate": 0.0005, "loss": 3.513662576675415, "step": 1120 }, { "epoch": 0.08837209302325581, "grad_norm": 1.3275505304336548, "learning_rate": 0.0005, "loss": 3.2621045112609863, "step": 1121 }, { "epoch": 0.08845092629089475, "grad_norm": 0.995729923248291, "learning_rate": 0.0005, "loss": 3.1440420150756836, "step": 1122 }, { "epoch": 0.0885297595585337, "grad_norm": 0.9422194957733154, "learning_rate": 0.0005, "loss": 2.7506723403930664, "step": 1123 }, { "epoch": 0.08860859282617264, "grad_norm": 1.0668847560882568, "learning_rate": 0.0005, "loss": 2.7399048805236816, "step": 1124 }, { "epoch": 0.08868742609381158, "grad_norm": 0.9805517196655273, "learning_rate": 0.0005, "loss": 3.110975980758667, "step": 1125 }, { "epoch": 0.08876625936145054, "grad_norm": 1.3473470211029053, "learning_rate": 0.0005, "loss": 3.1267178058624268, "step": 1126 }, { "epoch": 0.08884509262908948, "grad_norm": 1.0974884033203125, "learning_rate": 0.0005, "loss": 3.1070363521575928, "step": 1127 }, { "epoch": 0.08892392589672841, "grad_norm": 1.4115636348724365, "learning_rate": 0.0005, "loss": 3.1245789527893066, "step": 1128 }, { "epoch": 0.08900275916436737, "grad_norm": 1.2120364904403687, "learning_rate": 0.0005, "loss": 2.8400204181671143, "step": 1129 }, { "epoch": 0.0890815924320063, "grad_norm": 1.2895394563674927, "learning_rate": 0.0005, "loss": 3.2597334384918213, "step": 1130 }, { "epoch": 0.08916042569964525, "grad_norm": 0.9334081411361694, "learning_rate": 0.0005, "loss": 2.9012598991394043, "step": 1131 }, { "epoch": 0.0892392589672842, "grad_norm": 1.0231503248214722, "learning_rate": 0.0005, "loss": 3.1355643272399902, "step": 1132 }, { "epoch": 0.08931809223492314, "grad_norm": 1.226396918296814, "learning_rate": 0.0005, "loss": 2.937458038330078, "step": 1133 }, { "epoch": 0.08939692550256208, "grad_norm": 1.5994168519973755, "learning_rate": 0.0005, "loss": 3.419086456298828, "step": 1134 }, { "epoch": 0.08947575877020103, "grad_norm": 1.4778223037719727, "learning_rate": 0.0005, "loss": 3.1390914916992188, "step": 1135 }, { "epoch": 0.08955459203783997, "grad_norm": 1.173162817955017, "learning_rate": 0.0005, "loss": 2.974642276763916, "step": 1136 }, { "epoch": 0.08963342530547891, "grad_norm": 1.1338704824447632, "learning_rate": 0.0005, "loss": 2.935983657836914, "step": 1137 }, { "epoch": 0.08971225857311786, "grad_norm": 0.9915338158607483, "learning_rate": 0.0005, "loss": 3.3026838302612305, "step": 1138 }, { "epoch": 0.0897910918407568, "grad_norm": 1.3002163171768188, "learning_rate": 0.0005, "loss": 3.263876438140869, "step": 1139 }, { "epoch": 0.08986992510839574, "grad_norm": 1.4094029664993286, "learning_rate": 0.0005, "loss": 2.779759407043457, "step": 1140 }, { "epoch": 0.0899487583760347, "grad_norm": 1.7763774394989014, "learning_rate": 0.0005, "loss": 3.2369179725646973, "step": 1141 }, { "epoch": 0.09002759164367363, "grad_norm": 1.1296310424804688, "learning_rate": 0.0005, "loss": 3.126315116882324, "step": 1142 }, { "epoch": 0.09010642491131257, "grad_norm": 3.428072452545166, "learning_rate": 0.0005, "loss": 3.477759599685669, "step": 1143 }, { "epoch": 0.09018525817895151, "grad_norm": 1.8087735176086426, "learning_rate": 0.0005, "loss": 3.120809555053711, "step": 1144 }, { "epoch": 0.09026409144659046, "grad_norm": 1.6561648845672607, "learning_rate": 0.0005, "loss": 2.924459457397461, "step": 1145 }, { "epoch": 0.0903429247142294, "grad_norm": 1.2037041187286377, "learning_rate": 0.0005, "loss": 3.0703787803649902, "step": 1146 }, { "epoch": 0.09042175798186834, "grad_norm": 1.2036504745483398, "learning_rate": 0.0005, "loss": 3.3124518394470215, "step": 1147 }, { "epoch": 0.0905005912495073, "grad_norm": 1.6794501543045044, "learning_rate": 0.0005, "loss": 3.028627395629883, "step": 1148 }, { "epoch": 0.09057942451714623, "grad_norm": 1.1323796510696411, "learning_rate": 0.0005, "loss": 2.9710025787353516, "step": 1149 }, { "epoch": 0.09065825778478517, "grad_norm": 1.0666440725326538, "learning_rate": 0.0005, "loss": 2.987440824508667, "step": 1150 }, { "epoch": 0.09073709105242413, "grad_norm": 1.3197555541992188, "learning_rate": 0.0005, "loss": 2.8906588554382324, "step": 1151 }, { "epoch": 0.09081592432006307, "grad_norm": 1.2480231523513794, "learning_rate": 0.0005, "loss": 3.2703559398651123, "step": 1152 }, { "epoch": 0.090894757587702, "grad_norm": 0.9457294940948486, "learning_rate": 0.0005, "loss": 3.00344181060791, "step": 1153 }, { "epoch": 0.09097359085534096, "grad_norm": 1.4412798881530762, "learning_rate": 0.0005, "loss": 3.0841405391693115, "step": 1154 }, { "epoch": 0.0910524241229799, "grad_norm": 1.0762170553207397, "learning_rate": 0.0005, "loss": 3.223844289779663, "step": 1155 }, { "epoch": 0.09113125739061884, "grad_norm": 1.1492173671722412, "learning_rate": 0.0005, "loss": 2.9728751182556152, "step": 1156 }, { "epoch": 0.09121009065825779, "grad_norm": 1.4792848825454712, "learning_rate": 0.0005, "loss": 3.1357836723327637, "step": 1157 }, { "epoch": 0.09128892392589673, "grad_norm": 1.334570050239563, "learning_rate": 0.0005, "loss": 3.0322837829589844, "step": 1158 }, { "epoch": 0.09136775719353567, "grad_norm": 1.5768284797668457, "learning_rate": 0.0005, "loss": 3.2459864616394043, "step": 1159 }, { "epoch": 0.09144659046117462, "grad_norm": 1.5494426488876343, "learning_rate": 0.0005, "loss": 2.8529348373413086, "step": 1160 }, { "epoch": 0.09152542372881356, "grad_norm": 1.001200795173645, "learning_rate": 0.0005, "loss": 3.031925916671753, "step": 1161 }, { "epoch": 0.0916042569964525, "grad_norm": 1.0531615018844604, "learning_rate": 0.0005, "loss": 2.935931444168091, "step": 1162 }, { "epoch": 0.09168309026409145, "grad_norm": 1.0839027166366577, "learning_rate": 0.0005, "loss": 3.1117701530456543, "step": 1163 }, { "epoch": 0.09176192353173039, "grad_norm": 1.2534749507904053, "learning_rate": 0.0005, "loss": 2.7523422241210938, "step": 1164 }, { "epoch": 0.09184075679936933, "grad_norm": 1.4415335655212402, "learning_rate": 0.0005, "loss": 3.1105167865753174, "step": 1165 }, { "epoch": 0.09191959006700828, "grad_norm": 1.1126060485839844, "learning_rate": 0.0005, "loss": 2.753082275390625, "step": 1166 }, { "epoch": 0.09199842333464722, "grad_norm": 1.152340292930603, "learning_rate": 0.0005, "loss": 2.9307265281677246, "step": 1167 }, { "epoch": 0.09207725660228616, "grad_norm": 1.0023020505905151, "learning_rate": 0.0005, "loss": 3.0299477577209473, "step": 1168 }, { "epoch": 0.09215608986992511, "grad_norm": 0.9428208470344543, "learning_rate": 0.0005, "loss": 3.195185661315918, "step": 1169 }, { "epoch": 0.09223492313756405, "grad_norm": 1.5101776123046875, "learning_rate": 0.0005, "loss": 3.2765917778015137, "step": 1170 }, { "epoch": 0.09231375640520299, "grad_norm": 1.107177495956421, "learning_rate": 0.0005, "loss": 2.899158477783203, "step": 1171 }, { "epoch": 0.09239258967284195, "grad_norm": 1.0458126068115234, "learning_rate": 0.0005, "loss": 3.3716304302215576, "step": 1172 }, { "epoch": 0.09247142294048089, "grad_norm": 1.0553195476531982, "learning_rate": 0.0005, "loss": 3.0471816062927246, "step": 1173 }, { "epoch": 0.09255025620811982, "grad_norm": 0.9890021681785583, "learning_rate": 0.0005, "loss": 3.1117136478424072, "step": 1174 }, { "epoch": 0.09262908947575876, "grad_norm": 0.9902506470680237, "learning_rate": 0.0005, "loss": 3.161365509033203, "step": 1175 }, { "epoch": 0.09270792274339772, "grad_norm": 1.2448147535324097, "learning_rate": 0.0005, "loss": 3.1684165000915527, "step": 1176 }, { "epoch": 0.09278675601103666, "grad_norm": 1.0191515684127808, "learning_rate": 0.0005, "loss": 3.0845398902893066, "step": 1177 }, { "epoch": 0.0928655892786756, "grad_norm": 1.164095401763916, "learning_rate": 0.0005, "loss": 2.9000251293182373, "step": 1178 }, { "epoch": 0.09294442254631455, "grad_norm": 1.3631091117858887, "learning_rate": 0.0005, "loss": 2.7280325889587402, "step": 1179 }, { "epoch": 0.09302325581395349, "grad_norm": 1.1976960897445679, "learning_rate": 0.0005, "loss": 2.757391929626465, "step": 1180 }, { "epoch": 0.09310208908159243, "grad_norm": 1.129598617553711, "learning_rate": 0.0005, "loss": 3.4632182121276855, "step": 1181 }, { "epoch": 0.09318092234923138, "grad_norm": 1.1587731838226318, "learning_rate": 0.0005, "loss": 3.0367798805236816, "step": 1182 }, { "epoch": 0.09325975561687032, "grad_norm": 1.0150620937347412, "learning_rate": 0.0005, "loss": 2.7978549003601074, "step": 1183 }, { "epoch": 0.09333858888450926, "grad_norm": 1.5791640281677246, "learning_rate": 0.0005, "loss": 2.842541217803955, "step": 1184 }, { "epoch": 0.09341742215214821, "grad_norm": 1.0919889211654663, "learning_rate": 0.0005, "loss": 2.920196056365967, "step": 1185 }, { "epoch": 0.09349625541978715, "grad_norm": 2.962249994277954, "learning_rate": 0.0005, "loss": 3.2373342514038086, "step": 1186 }, { "epoch": 0.09357508868742609, "grad_norm": 1.269404411315918, "learning_rate": 0.0005, "loss": 3.013361930847168, "step": 1187 }, { "epoch": 0.09365392195506504, "grad_norm": 1.4868251085281372, "learning_rate": 0.0005, "loss": 2.883091926574707, "step": 1188 }, { "epoch": 0.09373275522270398, "grad_norm": 1.2735530138015747, "learning_rate": 0.0005, "loss": 3.1555418968200684, "step": 1189 }, { "epoch": 0.09381158849034292, "grad_norm": 1.31119704246521, "learning_rate": 0.0005, "loss": 2.943917989730835, "step": 1190 }, { "epoch": 0.09389042175798187, "grad_norm": 2.419048309326172, "learning_rate": 0.0005, "loss": 2.9786672592163086, "step": 1191 }, { "epoch": 0.09396925502562081, "grad_norm": 1.545360803604126, "learning_rate": 0.0005, "loss": 3.1206140518188477, "step": 1192 }, { "epoch": 0.09404808829325975, "grad_norm": 1.4334959983825684, "learning_rate": 0.0005, "loss": 2.9161911010742188, "step": 1193 }, { "epoch": 0.0941269215608987, "grad_norm": 1.3192758560180664, "learning_rate": 0.0005, "loss": 3.081294059753418, "step": 1194 }, { "epoch": 0.09420575482853764, "grad_norm": 1.2689032554626465, "learning_rate": 0.0005, "loss": 2.994771957397461, "step": 1195 }, { "epoch": 0.09428458809617658, "grad_norm": 1.2419952154159546, "learning_rate": 0.0005, "loss": 2.9926624298095703, "step": 1196 }, { "epoch": 0.09436342136381554, "grad_norm": 1.4988898038864136, "learning_rate": 0.0005, "loss": 3.567835807800293, "step": 1197 }, { "epoch": 0.09444225463145448, "grad_norm": 1.4054135084152222, "learning_rate": 0.0005, "loss": 3.2574188709259033, "step": 1198 }, { "epoch": 0.09452108789909341, "grad_norm": 0.9514763951301575, "learning_rate": 0.0005, "loss": 2.6053671836853027, "step": 1199 }, { "epoch": 0.09459992116673237, "grad_norm": 1.1475886106491089, "learning_rate": 0.0005, "loss": 2.895188570022583, "step": 1200 }, { "epoch": 0.0946787544343713, "grad_norm": 1.2847248315811157, "learning_rate": 0.0005, "loss": 3.0215811729431152, "step": 1201 }, { "epoch": 0.09475758770201025, "grad_norm": 1.1154464483261108, "learning_rate": 0.0005, "loss": 3.085446357727051, "step": 1202 }, { "epoch": 0.0948364209696492, "grad_norm": 1.0463953018188477, "learning_rate": 0.0005, "loss": 2.9985086917877197, "step": 1203 }, { "epoch": 0.09491525423728814, "grad_norm": 1.2590192556381226, "learning_rate": 0.0005, "loss": 3.0876283645629883, "step": 1204 }, { "epoch": 0.09499408750492708, "grad_norm": 1.1719812154769897, "learning_rate": 0.0005, "loss": 3.139533519744873, "step": 1205 }, { "epoch": 0.09507292077256602, "grad_norm": 1.0675946474075317, "learning_rate": 0.0005, "loss": 2.843562126159668, "step": 1206 }, { "epoch": 0.09515175404020497, "grad_norm": 1.2281196117401123, "learning_rate": 0.0005, "loss": 2.9751782417297363, "step": 1207 }, { "epoch": 0.09523058730784391, "grad_norm": 1.136827826499939, "learning_rate": 0.0005, "loss": 2.808957099914551, "step": 1208 }, { "epoch": 0.09530942057548285, "grad_norm": 1.5429269075393677, "learning_rate": 0.0005, "loss": 3.051600694656372, "step": 1209 }, { "epoch": 0.0953882538431218, "grad_norm": 1.30561363697052, "learning_rate": 0.0005, "loss": 3.3268275260925293, "step": 1210 }, { "epoch": 0.09546708711076074, "grad_norm": 1.1979565620422363, "learning_rate": 0.0005, "loss": 2.7349610328674316, "step": 1211 }, { "epoch": 0.09554592037839968, "grad_norm": 0.9703543186187744, "learning_rate": 0.0005, "loss": 3.139110565185547, "step": 1212 }, { "epoch": 0.09562475364603863, "grad_norm": 0.960792601108551, "learning_rate": 0.0005, "loss": 3.11893367767334, "step": 1213 }, { "epoch": 0.09570358691367757, "grad_norm": 0.9436231851577759, "learning_rate": 0.0005, "loss": 2.851511001586914, "step": 1214 }, { "epoch": 0.09578242018131651, "grad_norm": 0.8776175379753113, "learning_rate": 0.0005, "loss": 2.7001328468322754, "step": 1215 }, { "epoch": 0.09586125344895546, "grad_norm": 1.0539400577545166, "learning_rate": 0.0005, "loss": 3.2762327194213867, "step": 1216 }, { "epoch": 0.0959400867165944, "grad_norm": 1.5159651041030884, "learning_rate": 0.0005, "loss": 2.847874879837036, "step": 1217 }, { "epoch": 0.09601891998423334, "grad_norm": 1.0833959579467773, "learning_rate": 0.0005, "loss": 2.848578453063965, "step": 1218 }, { "epoch": 0.0960977532518723, "grad_norm": 1.3430755138397217, "learning_rate": 0.0005, "loss": 3.1962943077087402, "step": 1219 }, { "epoch": 0.09617658651951123, "grad_norm": 0.8914755582809448, "learning_rate": 0.0005, "loss": 3.1967058181762695, "step": 1220 }, { "epoch": 0.09625541978715017, "grad_norm": 2.267906665802002, "learning_rate": 0.0005, "loss": 3.0426464080810547, "step": 1221 }, { "epoch": 0.09633425305478913, "grad_norm": 1.1431214809417725, "learning_rate": 0.0005, "loss": 2.9706029891967773, "step": 1222 }, { "epoch": 0.09641308632242807, "grad_norm": 1.0932525396347046, "learning_rate": 0.0005, "loss": 2.9453508853912354, "step": 1223 }, { "epoch": 0.096491919590067, "grad_norm": 0.985016405582428, "learning_rate": 0.0005, "loss": 2.514709949493408, "step": 1224 }, { "epoch": 0.09657075285770596, "grad_norm": 1.262444257736206, "learning_rate": 0.0005, "loss": 3.4191761016845703, "step": 1225 }, { "epoch": 0.0966495861253449, "grad_norm": 1.1364307403564453, "learning_rate": 0.0005, "loss": 2.9504141807556152, "step": 1226 }, { "epoch": 0.09672841939298384, "grad_norm": 1.0583449602127075, "learning_rate": 0.0005, "loss": 2.852410316467285, "step": 1227 }, { "epoch": 0.09680725266062279, "grad_norm": 1.1872377395629883, "learning_rate": 0.0005, "loss": 3.122502326965332, "step": 1228 }, { "epoch": 0.09688608592826173, "grad_norm": 1.056127905845642, "learning_rate": 0.0005, "loss": 2.9577131271362305, "step": 1229 }, { "epoch": 0.09696491919590067, "grad_norm": 0.9318512678146362, "learning_rate": 0.0005, "loss": 2.9786224365234375, "step": 1230 }, { "epoch": 0.09704375246353962, "grad_norm": 0.9366840720176697, "learning_rate": 0.0005, "loss": 3.0373220443725586, "step": 1231 }, { "epoch": 0.09712258573117856, "grad_norm": 1.09958016872406, "learning_rate": 0.0005, "loss": 3.0906715393066406, "step": 1232 }, { "epoch": 0.0972014189988175, "grad_norm": 1.2698945999145508, "learning_rate": 0.0005, "loss": 2.9590139389038086, "step": 1233 }, { "epoch": 0.09728025226645644, "grad_norm": 0.8813487887382507, "learning_rate": 0.0005, "loss": 3.076622724533081, "step": 1234 }, { "epoch": 0.09735908553409539, "grad_norm": 0.9471054673194885, "learning_rate": 0.0005, "loss": 3.1781647205352783, "step": 1235 }, { "epoch": 0.09743791880173433, "grad_norm": 1.02093505859375, "learning_rate": 0.0005, "loss": 3.1325387954711914, "step": 1236 }, { "epoch": 0.09751675206937327, "grad_norm": 1.278841257095337, "learning_rate": 0.0005, "loss": 3.0621726512908936, "step": 1237 }, { "epoch": 0.09759558533701222, "grad_norm": 1.0349315404891968, "learning_rate": 0.0005, "loss": 2.8760557174682617, "step": 1238 }, { "epoch": 0.09767441860465116, "grad_norm": 2.029972791671753, "learning_rate": 0.0005, "loss": 3.0448055267333984, "step": 1239 }, { "epoch": 0.0977532518722901, "grad_norm": 1.1891648769378662, "learning_rate": 0.0005, "loss": 3.2812082767486572, "step": 1240 }, { "epoch": 0.09783208513992905, "grad_norm": 1.3018139600753784, "learning_rate": 0.0005, "loss": 2.9953083992004395, "step": 1241 }, { "epoch": 0.09791091840756799, "grad_norm": 1.148863673210144, "learning_rate": 0.0005, "loss": 3.316427707672119, "step": 1242 }, { "epoch": 0.09798975167520693, "grad_norm": 1.0325634479522705, "learning_rate": 0.0005, "loss": 2.574453830718994, "step": 1243 }, { "epoch": 0.09806858494284589, "grad_norm": 1.032639980316162, "learning_rate": 0.0005, "loss": 2.9102909564971924, "step": 1244 }, { "epoch": 0.09814741821048482, "grad_norm": 1.2765061855316162, "learning_rate": 0.0005, "loss": 3.303880214691162, "step": 1245 }, { "epoch": 0.09822625147812376, "grad_norm": 1.3704135417938232, "learning_rate": 0.0005, "loss": 2.996060609817505, "step": 1246 }, { "epoch": 0.09830508474576272, "grad_norm": 1.1201854944229126, "learning_rate": 0.0005, "loss": 3.249419927597046, "step": 1247 }, { "epoch": 0.09838391801340166, "grad_norm": 1.1597718000411987, "learning_rate": 0.0005, "loss": 3.2108287811279297, "step": 1248 }, { "epoch": 0.0984627512810406, "grad_norm": 1.1173148155212402, "learning_rate": 0.0005, "loss": 2.524918556213379, "step": 1249 }, { "epoch": 0.09854158454867955, "grad_norm": 0.8914439082145691, "learning_rate": 0.0005, "loss": 2.8437180519104004, "step": 1250 }, { "epoch": 0.09862041781631849, "grad_norm": 1.1956579685211182, "learning_rate": 0.0005, "loss": 3.1845431327819824, "step": 1251 }, { "epoch": 0.09869925108395743, "grad_norm": 1.0767196416854858, "learning_rate": 0.0005, "loss": 2.914445400238037, "step": 1252 }, { "epoch": 0.09877808435159638, "grad_norm": 1.164122462272644, "learning_rate": 0.0005, "loss": 2.991990089416504, "step": 1253 }, { "epoch": 0.09885691761923532, "grad_norm": 1.4246882200241089, "learning_rate": 0.0005, "loss": 3.1672472953796387, "step": 1254 }, { "epoch": 0.09893575088687426, "grad_norm": 1.097917079925537, "learning_rate": 0.0005, "loss": 2.747563362121582, "step": 1255 }, { "epoch": 0.09901458415451321, "grad_norm": 1.0643573999404907, "learning_rate": 0.0005, "loss": 2.8321821689605713, "step": 1256 }, { "epoch": 0.09909341742215215, "grad_norm": 1.0525492429733276, "learning_rate": 0.0005, "loss": 3.162803888320923, "step": 1257 }, { "epoch": 0.09917225068979109, "grad_norm": 1.0420515537261963, "learning_rate": 0.0005, "loss": 3.3158199787139893, "step": 1258 }, { "epoch": 0.09925108395743004, "grad_norm": 1.0082377195358276, "learning_rate": 0.0005, "loss": 3.169555425643921, "step": 1259 }, { "epoch": 0.09932991722506898, "grad_norm": 1.4318830966949463, "learning_rate": 0.0005, "loss": 3.0709383487701416, "step": 1260 }, { "epoch": 0.09940875049270792, "grad_norm": 1.2740626335144043, "learning_rate": 0.0005, "loss": 3.1022729873657227, "step": 1261 }, { "epoch": 0.09948758376034687, "grad_norm": 1.2951278686523438, "learning_rate": 0.0005, "loss": 3.0501582622528076, "step": 1262 }, { "epoch": 0.09956641702798581, "grad_norm": 1.0142555236816406, "learning_rate": 0.0005, "loss": 3.3269529342651367, "step": 1263 }, { "epoch": 0.09964525029562475, "grad_norm": 1.0551929473876953, "learning_rate": 0.0005, "loss": 3.03818416595459, "step": 1264 }, { "epoch": 0.09972408356326369, "grad_norm": 1.213712453842163, "learning_rate": 0.0005, "loss": 2.9502570629119873, "step": 1265 }, { "epoch": 0.09980291683090264, "grad_norm": 1.010992407798767, "learning_rate": 0.0005, "loss": 2.880802631378174, "step": 1266 }, { "epoch": 0.09988175009854158, "grad_norm": 1.5254840850830078, "learning_rate": 0.0005, "loss": 3.097278118133545, "step": 1267 }, { "epoch": 0.09996058336618052, "grad_norm": 0.9995371103286743, "learning_rate": 0.0005, "loss": 3.0088906288146973, "step": 1268 }, { "epoch": 0.10003941663381948, "grad_norm": 1.3811920881271362, "learning_rate": 0.0005, "loss": 3.509111166000366, "step": 1269 }, { "epoch": 0.10011824990145841, "grad_norm": 1.3184595108032227, "learning_rate": 0.0005, "loss": 3.0030124187469482, "step": 1270 }, { "epoch": 0.10019708316909735, "grad_norm": 1.6346032619476318, "learning_rate": 0.0005, "loss": 3.2417843341827393, "step": 1271 }, { "epoch": 0.1002759164367363, "grad_norm": 1.162651538848877, "learning_rate": 0.0005, "loss": 3.086921215057373, "step": 1272 }, { "epoch": 0.10035474970437525, "grad_norm": 1.2379716634750366, "learning_rate": 0.0005, "loss": 3.376166343688965, "step": 1273 }, { "epoch": 0.10043358297201418, "grad_norm": 1.4248988628387451, "learning_rate": 0.0005, "loss": 2.9688243865966797, "step": 1274 }, { "epoch": 0.10051241623965314, "grad_norm": 1.5293540954589844, "learning_rate": 0.0005, "loss": 3.0628867149353027, "step": 1275 }, { "epoch": 0.10059124950729208, "grad_norm": 3.0667364597320557, "learning_rate": 0.0005, "loss": 2.7418627738952637, "step": 1276 }, { "epoch": 0.10067008277493102, "grad_norm": 0.9677996635437012, "learning_rate": 0.0005, "loss": 3.060722589492798, "step": 1277 }, { "epoch": 0.10074891604256997, "grad_norm": 0.9897066354751587, "learning_rate": 0.0005, "loss": 3.134047269821167, "step": 1278 }, { "epoch": 0.10082774931020891, "grad_norm": 1.0374752283096313, "learning_rate": 0.0005, "loss": 3.0531930923461914, "step": 1279 }, { "epoch": 0.10090658257784785, "grad_norm": 1.1105016469955444, "learning_rate": 0.0005, "loss": 3.041369915008545, "step": 1280 }, { "epoch": 0.1009854158454868, "grad_norm": 1.4749031066894531, "learning_rate": 0.0005, "loss": 2.779757022857666, "step": 1281 }, { "epoch": 0.10106424911312574, "grad_norm": 1.8786027431488037, "learning_rate": 0.0005, "loss": 3.035442352294922, "step": 1282 }, { "epoch": 0.10114308238076468, "grad_norm": 1.224363088607788, "learning_rate": 0.0005, "loss": 2.656346082687378, "step": 1283 }, { "epoch": 0.10122191564840363, "grad_norm": 1.0635366439819336, "learning_rate": 0.0005, "loss": 3.2732601165771484, "step": 1284 }, { "epoch": 0.10130074891604257, "grad_norm": 1.3066338300704956, "learning_rate": 0.0005, "loss": 3.2041990756988525, "step": 1285 }, { "epoch": 0.10137958218368151, "grad_norm": 1.1994484663009644, "learning_rate": 0.0005, "loss": 3.2214131355285645, "step": 1286 }, { "epoch": 0.10145841545132046, "grad_norm": 1.1889594793319702, "learning_rate": 0.0005, "loss": 3.09663987159729, "step": 1287 }, { "epoch": 0.1015372487189594, "grad_norm": 1.3378649950027466, "learning_rate": 0.0005, "loss": 2.787522792816162, "step": 1288 }, { "epoch": 0.10161608198659834, "grad_norm": 1.1262032985687256, "learning_rate": 0.0005, "loss": 2.6544363498687744, "step": 1289 }, { "epoch": 0.1016949152542373, "grad_norm": 1.1482683420181274, "learning_rate": 0.0005, "loss": 2.8344571590423584, "step": 1290 }, { "epoch": 0.10177374852187623, "grad_norm": 1.3627101182937622, "learning_rate": 0.0005, "loss": 3.4528465270996094, "step": 1291 }, { "epoch": 0.10185258178951517, "grad_norm": 0.9880310893058777, "learning_rate": 0.0005, "loss": 3.2656688690185547, "step": 1292 }, { "epoch": 0.10193141505715411, "grad_norm": 1.455518126487732, "learning_rate": 0.0005, "loss": 2.52476167678833, "step": 1293 }, { "epoch": 0.10201024832479307, "grad_norm": 1.4815887212753296, "learning_rate": 0.0005, "loss": 2.8519186973571777, "step": 1294 }, { "epoch": 0.102089081592432, "grad_norm": 1.3272813558578491, "learning_rate": 0.0005, "loss": 3.0296459197998047, "step": 1295 }, { "epoch": 0.10216791486007094, "grad_norm": 1.1933635473251343, "learning_rate": 0.0005, "loss": 3.058793067932129, "step": 1296 }, { "epoch": 0.1022467481277099, "grad_norm": 1.259922742843628, "learning_rate": 0.0005, "loss": 2.621760845184326, "step": 1297 }, { "epoch": 0.10232558139534884, "grad_norm": 1.1757739782333374, "learning_rate": 0.0005, "loss": 3.100008010864258, "step": 1298 }, { "epoch": 0.10240441466298777, "grad_norm": 1.1580607891082764, "learning_rate": 0.0005, "loss": 2.8063912391662598, "step": 1299 }, { "epoch": 0.10248324793062673, "grad_norm": 1.2573235034942627, "learning_rate": 0.0005, "loss": 3.021726131439209, "step": 1300 }, { "epoch": 0.10256208119826567, "grad_norm": 1.2903505563735962, "learning_rate": 0.0005, "loss": 3.082606077194214, "step": 1301 }, { "epoch": 0.1026409144659046, "grad_norm": 1.4229284524917603, "learning_rate": 0.0005, "loss": 3.166451930999756, "step": 1302 }, { "epoch": 0.10271974773354356, "grad_norm": 1.1663209199905396, "learning_rate": 0.0005, "loss": 2.9994630813598633, "step": 1303 }, { "epoch": 0.1027985810011825, "grad_norm": 1.0479140281677246, "learning_rate": 0.0005, "loss": 3.1778130531311035, "step": 1304 }, { "epoch": 0.10287741426882144, "grad_norm": 1.093858003616333, "learning_rate": 0.0005, "loss": 2.9144301414489746, "step": 1305 }, { "epoch": 0.10295624753646039, "grad_norm": 1.3069485425949097, "learning_rate": 0.0005, "loss": 3.013913869857788, "step": 1306 }, { "epoch": 0.10303508080409933, "grad_norm": 1.176064133644104, "learning_rate": 0.0005, "loss": 2.6382834911346436, "step": 1307 }, { "epoch": 0.10311391407173827, "grad_norm": 1.1623830795288086, "learning_rate": 0.0005, "loss": 2.7423009872436523, "step": 1308 }, { "epoch": 0.10319274733937722, "grad_norm": 1.295987844467163, "learning_rate": 0.0005, "loss": 3.2304797172546387, "step": 1309 }, { "epoch": 0.10327158060701616, "grad_norm": 0.9479731917381287, "learning_rate": 0.0005, "loss": 2.965336322784424, "step": 1310 }, { "epoch": 0.1033504138746551, "grad_norm": 1.1215977668762207, "learning_rate": 0.0005, "loss": 2.586212158203125, "step": 1311 }, { "epoch": 0.10342924714229405, "grad_norm": 1.114789605140686, "learning_rate": 0.0005, "loss": 2.725067615509033, "step": 1312 }, { "epoch": 0.10350808040993299, "grad_norm": 1.0610134601593018, "learning_rate": 0.0005, "loss": 2.989426612854004, "step": 1313 }, { "epoch": 0.10358691367757193, "grad_norm": 1.0845657587051392, "learning_rate": 0.0005, "loss": 2.6954598426818848, "step": 1314 }, { "epoch": 0.10366574694521088, "grad_norm": 1.1875123977661133, "learning_rate": 0.0005, "loss": 3.3332555294036865, "step": 1315 }, { "epoch": 0.10374458021284982, "grad_norm": 0.9864860773086548, "learning_rate": 0.0005, "loss": 3.0416908264160156, "step": 1316 }, { "epoch": 0.10382341348048876, "grad_norm": 1.3556883335113525, "learning_rate": 0.0005, "loss": 3.012624979019165, "step": 1317 }, { "epoch": 0.10390224674812772, "grad_norm": 1.1891729831695557, "learning_rate": 0.0005, "loss": 2.636064291000366, "step": 1318 }, { "epoch": 0.10398108001576666, "grad_norm": 1.138999342918396, "learning_rate": 0.0005, "loss": 2.851233959197998, "step": 1319 }, { "epoch": 0.1040599132834056, "grad_norm": 1.6123305559158325, "learning_rate": 0.0005, "loss": 3.0962326526641846, "step": 1320 }, { "epoch": 0.10413874655104455, "grad_norm": 1.6586353778839111, "learning_rate": 0.0005, "loss": 2.8843023777008057, "step": 1321 }, { "epoch": 0.10421757981868349, "grad_norm": 1.1115431785583496, "learning_rate": 0.0005, "loss": 2.7376978397369385, "step": 1322 }, { "epoch": 0.10429641308632243, "grad_norm": 0.9624471664428711, "learning_rate": 0.0005, "loss": 2.9591832160949707, "step": 1323 }, { "epoch": 0.10437524635396137, "grad_norm": 1.369929552078247, "learning_rate": 0.0005, "loss": 2.8705177307128906, "step": 1324 }, { "epoch": 0.10445407962160032, "grad_norm": 1.2753092050552368, "learning_rate": 0.0005, "loss": 3.0981204509735107, "step": 1325 }, { "epoch": 0.10453291288923926, "grad_norm": 1.3984099626541138, "learning_rate": 0.0005, "loss": 2.6348538398742676, "step": 1326 }, { "epoch": 0.1046117461568782, "grad_norm": 1.1465904712677002, "learning_rate": 0.0005, "loss": 3.1500158309936523, "step": 1327 }, { "epoch": 0.10469057942451715, "grad_norm": 1.3279350996017456, "learning_rate": 0.0005, "loss": 2.9271488189697266, "step": 1328 }, { "epoch": 0.10476941269215609, "grad_norm": 1.687501072883606, "learning_rate": 0.0005, "loss": 3.018829822540283, "step": 1329 }, { "epoch": 0.10484824595979503, "grad_norm": 1.7932679653167725, "learning_rate": 0.0005, "loss": 3.193477153778076, "step": 1330 }, { "epoch": 0.10492707922743398, "grad_norm": 1.265174150466919, "learning_rate": 0.0005, "loss": 2.971242904663086, "step": 1331 }, { "epoch": 0.10500591249507292, "grad_norm": 1.428533673286438, "learning_rate": 0.0005, "loss": 3.2092456817626953, "step": 1332 }, { "epoch": 0.10508474576271186, "grad_norm": 1.2668668031692505, "learning_rate": 0.0005, "loss": 2.919281482696533, "step": 1333 }, { "epoch": 0.10516357903035081, "grad_norm": 1.4209866523742676, "learning_rate": 0.0005, "loss": 3.135660171508789, "step": 1334 }, { "epoch": 0.10524241229798975, "grad_norm": 1.029517650604248, "learning_rate": 0.0005, "loss": 3.0235602855682373, "step": 1335 }, { "epoch": 0.10532124556562869, "grad_norm": 1.546088457107544, "learning_rate": 0.0005, "loss": 3.166921615600586, "step": 1336 }, { "epoch": 0.10540007883326764, "grad_norm": 1.686661720275879, "learning_rate": 0.0005, "loss": 2.892827033996582, "step": 1337 }, { "epoch": 0.10547891210090658, "grad_norm": 1.3403226137161255, "learning_rate": 0.0005, "loss": 3.0803613662719727, "step": 1338 }, { "epoch": 0.10555774536854552, "grad_norm": 1.2962629795074463, "learning_rate": 0.0005, "loss": 2.7761149406433105, "step": 1339 }, { "epoch": 0.10563657863618447, "grad_norm": 1.492487907409668, "learning_rate": 0.0005, "loss": 3.031073570251465, "step": 1340 }, { "epoch": 0.10571541190382341, "grad_norm": 1.2242681980133057, "learning_rate": 0.0005, "loss": 2.905747890472412, "step": 1341 }, { "epoch": 0.10579424517146235, "grad_norm": 1.0837377309799194, "learning_rate": 0.0005, "loss": 3.185822010040283, "step": 1342 }, { "epoch": 0.1058730784391013, "grad_norm": 0.903584897518158, "learning_rate": 0.0005, "loss": 2.6968724727630615, "step": 1343 }, { "epoch": 0.10595191170674025, "grad_norm": 1.2097190618515015, "learning_rate": 0.0005, "loss": 2.8694512844085693, "step": 1344 }, { "epoch": 0.10603074497437918, "grad_norm": 0.9452906250953674, "learning_rate": 0.0005, "loss": 2.9017105102539062, "step": 1345 }, { "epoch": 0.10610957824201814, "grad_norm": 1.3386433124542236, "learning_rate": 0.0005, "loss": 3.326029062271118, "step": 1346 }, { "epoch": 0.10618841150965708, "grad_norm": 2.0624310970306396, "learning_rate": 0.0005, "loss": 3.398277521133423, "step": 1347 }, { "epoch": 0.10626724477729602, "grad_norm": 1.1910476684570312, "learning_rate": 0.0005, "loss": 2.7920470237731934, "step": 1348 }, { "epoch": 0.10634607804493497, "grad_norm": 1.3108482360839844, "learning_rate": 0.0005, "loss": 2.907020330429077, "step": 1349 }, { "epoch": 0.10642491131257391, "grad_norm": 1.2858346700668335, "learning_rate": 0.0005, "loss": 3.058119058609009, "step": 1350 }, { "epoch": 0.10650374458021285, "grad_norm": 0.9716061353683472, "learning_rate": 0.0005, "loss": 3.0594072341918945, "step": 1351 }, { "epoch": 0.10658257784785179, "grad_norm": 1.3590245246887207, "learning_rate": 0.0005, "loss": 3.2811708450317383, "step": 1352 }, { "epoch": 0.10666141111549074, "grad_norm": 1.1328816413879395, "learning_rate": 0.0005, "loss": 2.990321159362793, "step": 1353 }, { "epoch": 0.10674024438312968, "grad_norm": 1.040809988975525, "learning_rate": 0.0005, "loss": 3.088465929031372, "step": 1354 }, { "epoch": 0.10681907765076862, "grad_norm": 1.0529749393463135, "learning_rate": 0.0005, "loss": 2.67929744720459, "step": 1355 }, { "epoch": 0.10689791091840757, "grad_norm": 1.1978211402893066, "learning_rate": 0.0005, "loss": 2.919551372528076, "step": 1356 }, { "epoch": 0.10697674418604651, "grad_norm": 0.9902403354644775, "learning_rate": 0.0005, "loss": 2.9569778442382812, "step": 1357 }, { "epoch": 0.10705557745368545, "grad_norm": 1.1016682386398315, "learning_rate": 0.0005, "loss": 2.8239688873291016, "step": 1358 }, { "epoch": 0.1071344107213244, "grad_norm": 1.149502158164978, "learning_rate": 0.0005, "loss": 3.1276328563690186, "step": 1359 }, { "epoch": 0.10721324398896334, "grad_norm": 1.589924693107605, "learning_rate": 0.0005, "loss": 2.784667491912842, "step": 1360 }, { "epoch": 0.10729207725660228, "grad_norm": 1.1790605783462524, "learning_rate": 0.0005, "loss": 3.484677791595459, "step": 1361 }, { "epoch": 0.10737091052424123, "grad_norm": 1.0483694076538086, "learning_rate": 0.0005, "loss": 3.0461530685424805, "step": 1362 }, { "epoch": 0.10744974379188017, "grad_norm": 0.9112489819526672, "learning_rate": 0.0005, "loss": 3.0475778579711914, "step": 1363 }, { "epoch": 0.10752857705951911, "grad_norm": 1.0645555257797241, "learning_rate": 0.0005, "loss": 2.7466845512390137, "step": 1364 }, { "epoch": 0.10760741032715807, "grad_norm": 1.2577975988388062, "learning_rate": 0.0005, "loss": 3.1440205574035645, "step": 1365 }, { "epoch": 0.107686243594797, "grad_norm": 1.3532558679580688, "learning_rate": 0.0005, "loss": 2.9531588554382324, "step": 1366 }, { "epoch": 0.10776507686243594, "grad_norm": 1.0669419765472412, "learning_rate": 0.0005, "loss": 2.809887409210205, "step": 1367 }, { "epoch": 0.1078439101300749, "grad_norm": 1.0491055250167847, "learning_rate": 0.0005, "loss": 2.8992056846618652, "step": 1368 }, { "epoch": 0.10792274339771384, "grad_norm": 1.0519968271255493, "learning_rate": 0.0005, "loss": 3.0378170013427734, "step": 1369 }, { "epoch": 0.10800157666535277, "grad_norm": 1.1430084705352783, "learning_rate": 0.0005, "loss": 2.966538429260254, "step": 1370 }, { "epoch": 0.10808040993299173, "grad_norm": 1.0211913585662842, "learning_rate": 0.0005, "loss": 2.859675407409668, "step": 1371 }, { "epoch": 0.10815924320063067, "grad_norm": 1.1620570421218872, "learning_rate": 0.0005, "loss": 2.6064910888671875, "step": 1372 }, { "epoch": 0.1082380764682696, "grad_norm": 1.293199896812439, "learning_rate": 0.0005, "loss": 3.2369589805603027, "step": 1373 }, { "epoch": 0.10831690973590856, "grad_norm": 1.2613338232040405, "learning_rate": 0.0005, "loss": 3.2209885120391846, "step": 1374 }, { "epoch": 0.1083957430035475, "grad_norm": 1.1778804063796997, "learning_rate": 0.0005, "loss": 2.8931217193603516, "step": 1375 }, { "epoch": 0.10847457627118644, "grad_norm": 1.5892796516418457, "learning_rate": 0.0005, "loss": 3.1843698024749756, "step": 1376 }, { "epoch": 0.10855340953882539, "grad_norm": 1.2083436250686646, "learning_rate": 0.0005, "loss": 3.2324819564819336, "step": 1377 }, { "epoch": 0.10863224280646433, "grad_norm": 1.0936079025268555, "learning_rate": 0.0005, "loss": 2.919307231903076, "step": 1378 }, { "epoch": 0.10871107607410327, "grad_norm": 0.9791391491889954, "learning_rate": 0.0005, "loss": 2.759709596633911, "step": 1379 }, { "epoch": 0.10878990934174222, "grad_norm": 1.3332964181900024, "learning_rate": 0.0005, "loss": 2.5852203369140625, "step": 1380 }, { "epoch": 0.10886874260938116, "grad_norm": 1.3418710231781006, "learning_rate": 0.0005, "loss": 3.4641830921173096, "step": 1381 }, { "epoch": 0.1089475758770201, "grad_norm": 2.2244975566864014, "learning_rate": 0.0005, "loss": 3.127821445465088, "step": 1382 }, { "epoch": 0.10902640914465904, "grad_norm": 1.0607632398605347, "learning_rate": 0.0005, "loss": 2.993448257446289, "step": 1383 }, { "epoch": 0.10910524241229799, "grad_norm": 1.0037246942520142, "learning_rate": 0.0005, "loss": 2.670053482055664, "step": 1384 }, { "epoch": 0.10918407567993693, "grad_norm": 1.2759188413619995, "learning_rate": 0.0005, "loss": 2.849719285964966, "step": 1385 }, { "epoch": 0.10926290894757587, "grad_norm": 1.069423794746399, "learning_rate": 0.0005, "loss": 3.0883374214172363, "step": 1386 }, { "epoch": 0.10934174221521482, "grad_norm": 1.4007899761199951, "learning_rate": 0.0005, "loss": 2.996890068054199, "step": 1387 }, { "epoch": 0.10942057548285376, "grad_norm": 1.0979173183441162, "learning_rate": 0.0005, "loss": 3.070232391357422, "step": 1388 }, { "epoch": 0.1094994087504927, "grad_norm": 1.3384164571762085, "learning_rate": 0.0005, "loss": 3.156991958618164, "step": 1389 }, { "epoch": 0.10957824201813166, "grad_norm": 1.0647451877593994, "learning_rate": 0.0005, "loss": 2.611534833908081, "step": 1390 }, { "epoch": 0.1096570752857706, "grad_norm": 1.127539038658142, "learning_rate": 0.0005, "loss": 3.168018341064453, "step": 1391 }, { "epoch": 0.10973590855340953, "grad_norm": 2.0397443771362305, "learning_rate": 0.0005, "loss": 3.150193691253662, "step": 1392 }, { "epoch": 0.10981474182104849, "grad_norm": 1.4825267791748047, "learning_rate": 0.0005, "loss": 2.9993224143981934, "step": 1393 }, { "epoch": 0.10989357508868743, "grad_norm": 1.470778465270996, "learning_rate": 0.0005, "loss": 3.194194793701172, "step": 1394 }, { "epoch": 0.10997240835632636, "grad_norm": 1.0849547386169434, "learning_rate": 0.0005, "loss": 2.6624903678894043, "step": 1395 }, { "epoch": 0.11005124162396532, "grad_norm": 1.3662480115890503, "learning_rate": 0.0005, "loss": 3.2772388458251953, "step": 1396 }, { "epoch": 0.11013007489160426, "grad_norm": 1.117218017578125, "learning_rate": 0.0005, "loss": 2.993060827255249, "step": 1397 }, { "epoch": 0.1102089081592432, "grad_norm": 1.4437320232391357, "learning_rate": 0.0005, "loss": 2.975511074066162, "step": 1398 }, { "epoch": 0.11028774142688215, "grad_norm": 1.199622392654419, "learning_rate": 0.0005, "loss": 2.6460747718811035, "step": 1399 }, { "epoch": 0.11036657469452109, "grad_norm": 1.1408854722976685, "learning_rate": 0.0005, "loss": 2.71604061126709, "step": 1400 }, { "epoch": 0.11044540796216003, "grad_norm": 0.9487289190292358, "learning_rate": 0.0005, "loss": 2.4397263526916504, "step": 1401 }, { "epoch": 0.11052424122979898, "grad_norm": 1.1418123245239258, "learning_rate": 0.0005, "loss": 3.178119421005249, "step": 1402 }, { "epoch": 0.11060307449743792, "grad_norm": 1.1330499649047852, "learning_rate": 0.0005, "loss": 2.9030344486236572, "step": 1403 }, { "epoch": 0.11068190776507686, "grad_norm": 0.9092229604721069, "learning_rate": 0.0005, "loss": 3.173326015472412, "step": 1404 }, { "epoch": 0.11076074103271581, "grad_norm": 1.1390280723571777, "learning_rate": 0.0005, "loss": 2.7829174995422363, "step": 1405 }, { "epoch": 0.11083957430035475, "grad_norm": 1.085639476776123, "learning_rate": 0.0005, "loss": 3.151188373565674, "step": 1406 }, { "epoch": 0.11091840756799369, "grad_norm": 1.0996670722961426, "learning_rate": 0.0005, "loss": 2.941206455230713, "step": 1407 }, { "epoch": 0.11099724083563264, "grad_norm": 1.0463584661483765, "learning_rate": 0.0005, "loss": 2.8480281829833984, "step": 1408 }, { "epoch": 0.11107607410327158, "grad_norm": 1.157405138015747, "learning_rate": 0.0005, "loss": 3.079491138458252, "step": 1409 }, { "epoch": 0.11115490737091052, "grad_norm": 1.1441259384155273, "learning_rate": 0.0005, "loss": 3.1034765243530273, "step": 1410 }, { "epoch": 0.11123374063854947, "grad_norm": 1.1094791889190674, "learning_rate": 0.0005, "loss": 2.7027716636657715, "step": 1411 }, { "epoch": 0.11131257390618841, "grad_norm": 1.189400553703308, "learning_rate": 0.0005, "loss": 2.8850183486938477, "step": 1412 }, { "epoch": 0.11139140717382735, "grad_norm": 1.1695352792739868, "learning_rate": 0.0005, "loss": 3.2147903442382812, "step": 1413 }, { "epoch": 0.11147024044146629, "grad_norm": 1.0542510747909546, "learning_rate": 0.0005, "loss": 3.218503952026367, "step": 1414 }, { "epoch": 0.11154907370910525, "grad_norm": 1.3018276691436768, "learning_rate": 0.0005, "loss": 2.99943470954895, "step": 1415 }, { "epoch": 0.11162790697674418, "grad_norm": 1.1326267719268799, "learning_rate": 0.0005, "loss": 2.8288683891296387, "step": 1416 }, { "epoch": 0.11170674024438312, "grad_norm": 0.9348523616790771, "learning_rate": 0.0005, "loss": 3.059218406677246, "step": 1417 }, { "epoch": 0.11178557351202208, "grad_norm": 1.3971987962722778, "learning_rate": 0.0005, "loss": 2.8972525596618652, "step": 1418 }, { "epoch": 0.11186440677966102, "grad_norm": 1.266766905784607, "learning_rate": 0.0005, "loss": 2.9024405479431152, "step": 1419 }, { "epoch": 0.11194324004729996, "grad_norm": 1.194997787475586, "learning_rate": 0.0005, "loss": 3.082520008087158, "step": 1420 }, { "epoch": 0.11202207331493891, "grad_norm": 2.176903486251831, "learning_rate": 0.0005, "loss": 2.9297971725463867, "step": 1421 }, { "epoch": 0.11210090658257785, "grad_norm": 1.4623756408691406, "learning_rate": 0.0005, "loss": 2.914973497390747, "step": 1422 }, { "epoch": 0.11217973985021679, "grad_norm": 1.4133344888687134, "learning_rate": 0.0005, "loss": 2.8341009616851807, "step": 1423 }, { "epoch": 0.11225857311785574, "grad_norm": 1.055319905281067, "learning_rate": 0.0005, "loss": 2.563159465789795, "step": 1424 }, { "epoch": 0.11233740638549468, "grad_norm": 1.4647682905197144, "learning_rate": 0.0005, "loss": 3.0744874477386475, "step": 1425 }, { "epoch": 0.11241623965313362, "grad_norm": 1.0179380178451538, "learning_rate": 0.0005, "loss": 2.9610016345977783, "step": 1426 }, { "epoch": 0.11249507292077257, "grad_norm": 1.13157320022583, "learning_rate": 0.0005, "loss": 2.9070563316345215, "step": 1427 }, { "epoch": 0.11257390618841151, "grad_norm": 1.7222859859466553, "learning_rate": 0.0005, "loss": 3.2079885005950928, "step": 1428 }, { "epoch": 0.11265273945605045, "grad_norm": 2.5556678771972656, "learning_rate": 0.0005, "loss": 2.7564516067504883, "step": 1429 }, { "epoch": 0.1127315727236894, "grad_norm": 1.139210820198059, "learning_rate": 0.0005, "loss": 3.3539247512817383, "step": 1430 }, { "epoch": 0.11281040599132834, "grad_norm": 1.0672234296798706, "learning_rate": 0.0005, "loss": 3.366514205932617, "step": 1431 }, { "epoch": 0.11288923925896728, "grad_norm": 0.8895896673202515, "learning_rate": 0.0005, "loss": 3.1126136779785156, "step": 1432 }, { "epoch": 0.11296807252660623, "grad_norm": 0.9566857814788818, "learning_rate": 0.0005, "loss": 3.052741050720215, "step": 1433 }, { "epoch": 0.11304690579424517, "grad_norm": 1.0554778575897217, "learning_rate": 0.0005, "loss": 3.0274696350097656, "step": 1434 }, { "epoch": 0.11312573906188411, "grad_norm": 1.2901105880737305, "learning_rate": 0.0005, "loss": 2.915128707885742, "step": 1435 }, { "epoch": 0.11320457232952306, "grad_norm": 1.4313477277755737, "learning_rate": 0.0005, "loss": 3.051048517227173, "step": 1436 }, { "epoch": 0.113283405597162, "grad_norm": 1.4478482007980347, "learning_rate": 0.0005, "loss": 3.0466647148132324, "step": 1437 }, { "epoch": 0.11336223886480094, "grad_norm": 1.04159414768219, "learning_rate": 0.0005, "loss": 3.1041035652160645, "step": 1438 }, { "epoch": 0.1134410721324399, "grad_norm": 1.1184637546539307, "learning_rate": 0.0005, "loss": 2.666254997253418, "step": 1439 }, { "epoch": 0.11351990540007884, "grad_norm": 1.0768907070159912, "learning_rate": 0.0005, "loss": 3.0024936199188232, "step": 1440 }, { "epoch": 0.11359873866771777, "grad_norm": 1.199966311454773, "learning_rate": 0.0005, "loss": 3.2063870429992676, "step": 1441 }, { "epoch": 0.11367757193535671, "grad_norm": 1.35801362991333, "learning_rate": 0.0005, "loss": 3.2787318229675293, "step": 1442 }, { "epoch": 0.11375640520299567, "grad_norm": 1.3204865455627441, "learning_rate": 0.0005, "loss": 2.5393266677856445, "step": 1443 }, { "epoch": 0.1138352384706346, "grad_norm": 1.0886415243148804, "learning_rate": 0.0005, "loss": 2.436861991882324, "step": 1444 }, { "epoch": 0.11391407173827355, "grad_norm": 1.6699501276016235, "learning_rate": 0.0005, "loss": 2.8716769218444824, "step": 1445 }, { "epoch": 0.1139929050059125, "grad_norm": 1.1629880666732788, "learning_rate": 0.0005, "loss": 3.27588152885437, "step": 1446 }, { "epoch": 0.11407173827355144, "grad_norm": 1.3219467401504517, "learning_rate": 0.0005, "loss": 3.1079659461975098, "step": 1447 }, { "epoch": 0.11415057154119038, "grad_norm": 1.2604103088378906, "learning_rate": 0.0005, "loss": 2.58443546295166, "step": 1448 }, { "epoch": 0.11422940480882933, "grad_norm": 1.0274490118026733, "learning_rate": 0.0005, "loss": 3.3963420391082764, "step": 1449 }, { "epoch": 0.11430823807646827, "grad_norm": 1.0245000123977661, "learning_rate": 0.0005, "loss": 3.0456619262695312, "step": 1450 }, { "epoch": 0.11438707134410721, "grad_norm": 1.084425926208496, "learning_rate": 0.0005, "loss": 3.0331692695617676, "step": 1451 }, { "epoch": 0.11446590461174616, "grad_norm": 1.1054295301437378, "learning_rate": 0.0005, "loss": 3.209963321685791, "step": 1452 }, { "epoch": 0.1145447378793851, "grad_norm": 1.0482696294784546, "learning_rate": 0.0005, "loss": 2.984748601913452, "step": 1453 }, { "epoch": 0.11462357114702404, "grad_norm": 1.1158874034881592, "learning_rate": 0.0005, "loss": 2.9372048377990723, "step": 1454 }, { "epoch": 0.11470240441466299, "grad_norm": 1.1218830347061157, "learning_rate": 0.0005, "loss": 3.053215503692627, "step": 1455 }, { "epoch": 0.11478123768230193, "grad_norm": 0.9622119069099426, "learning_rate": 0.0005, "loss": 2.904104709625244, "step": 1456 }, { "epoch": 0.11486007094994087, "grad_norm": 1.1983047723770142, "learning_rate": 0.0005, "loss": 2.651388168334961, "step": 1457 }, { "epoch": 0.11493890421757982, "grad_norm": 1.1226555109024048, "learning_rate": 0.0005, "loss": 2.972306966781616, "step": 1458 }, { "epoch": 0.11501773748521876, "grad_norm": 1.2221217155456543, "learning_rate": 0.0005, "loss": 2.4632725715637207, "step": 1459 }, { "epoch": 0.1150965707528577, "grad_norm": 1.2307605743408203, "learning_rate": 0.0005, "loss": 2.667745351791382, "step": 1460 }, { "epoch": 0.11517540402049666, "grad_norm": 1.6164121627807617, "learning_rate": 0.0005, "loss": 2.87197208404541, "step": 1461 }, { "epoch": 0.1152542372881356, "grad_norm": 1.352020025253296, "learning_rate": 0.0005, "loss": 3.2324209213256836, "step": 1462 }, { "epoch": 0.11533307055577453, "grad_norm": 1.1947664022445679, "learning_rate": 0.0005, "loss": 3.1804068088531494, "step": 1463 }, { "epoch": 0.11541190382341349, "grad_norm": 0.9755721688270569, "learning_rate": 0.0005, "loss": 2.9587812423706055, "step": 1464 }, { "epoch": 0.11549073709105243, "grad_norm": 1.229506492614746, "learning_rate": 0.0005, "loss": 3.032531261444092, "step": 1465 }, { "epoch": 0.11556957035869136, "grad_norm": 1.1830835342407227, "learning_rate": 0.0005, "loss": 3.08388614654541, "step": 1466 }, { "epoch": 0.11564840362633032, "grad_norm": 1.2102930545806885, "learning_rate": 0.0005, "loss": 2.693932056427002, "step": 1467 }, { "epoch": 0.11572723689396926, "grad_norm": 1.2238355875015259, "learning_rate": 0.0005, "loss": 3.2348265647888184, "step": 1468 }, { "epoch": 0.1158060701616082, "grad_norm": 1.1327226161956787, "learning_rate": 0.0005, "loss": 2.8169403076171875, "step": 1469 }, { "epoch": 0.11588490342924715, "grad_norm": 1.419603705406189, "learning_rate": 0.0005, "loss": 3.127197504043579, "step": 1470 }, { "epoch": 0.11596373669688609, "grad_norm": 1.3099534511566162, "learning_rate": 0.0005, "loss": 2.743279218673706, "step": 1471 }, { "epoch": 0.11604256996452503, "grad_norm": 1.2089719772338867, "learning_rate": 0.0005, "loss": 2.8730058670043945, "step": 1472 }, { "epoch": 0.11612140323216397, "grad_norm": 0.9571505784988403, "learning_rate": 0.0005, "loss": 3.2552332878112793, "step": 1473 }, { "epoch": 0.11620023649980292, "grad_norm": 1.5560600757598877, "learning_rate": 0.0005, "loss": 2.915142297744751, "step": 1474 }, { "epoch": 0.11627906976744186, "grad_norm": 0.9758757948875427, "learning_rate": 0.0005, "loss": 2.6164608001708984, "step": 1475 }, { "epoch": 0.1163579030350808, "grad_norm": 0.9835274815559387, "learning_rate": 0.0005, "loss": 3.0399136543273926, "step": 1476 }, { "epoch": 0.11643673630271975, "grad_norm": 1.0622485876083374, "learning_rate": 0.0005, "loss": 2.9085593223571777, "step": 1477 }, { "epoch": 0.11651556957035869, "grad_norm": 0.8728640675544739, "learning_rate": 0.0005, "loss": 2.836029291152954, "step": 1478 }, { "epoch": 0.11659440283799763, "grad_norm": 1.014827013015747, "learning_rate": 0.0005, "loss": 2.9165596961975098, "step": 1479 }, { "epoch": 0.11667323610563658, "grad_norm": 1.3163118362426758, "learning_rate": 0.0005, "loss": 3.1425113677978516, "step": 1480 }, { "epoch": 0.11675206937327552, "grad_norm": 1.4412394762039185, "learning_rate": 0.0005, "loss": 2.9864182472229004, "step": 1481 }, { "epoch": 0.11683090264091446, "grad_norm": 1.4612016677856445, "learning_rate": 0.0005, "loss": 2.89133882522583, "step": 1482 }, { "epoch": 0.11690973590855341, "grad_norm": 1.5273973941802979, "learning_rate": 0.0005, "loss": 2.9076409339904785, "step": 1483 }, { "epoch": 0.11698856917619235, "grad_norm": 1.0813740491867065, "learning_rate": 0.0005, "loss": 2.733497381210327, "step": 1484 }, { "epoch": 0.11706740244383129, "grad_norm": 1.5350110530853271, "learning_rate": 0.0005, "loss": 3.3150687217712402, "step": 1485 }, { "epoch": 0.11714623571147025, "grad_norm": 1.2152936458587646, "learning_rate": 0.0005, "loss": 3.060957908630371, "step": 1486 }, { "epoch": 0.11722506897910918, "grad_norm": 1.0230947732925415, "learning_rate": 0.0005, "loss": 2.8551242351531982, "step": 1487 }, { "epoch": 0.11730390224674812, "grad_norm": 1.2776871919631958, "learning_rate": 0.0005, "loss": 2.9819698333740234, "step": 1488 }, { "epoch": 0.11738273551438708, "grad_norm": 1.2442491054534912, "learning_rate": 0.0005, "loss": 2.8881759643554688, "step": 1489 }, { "epoch": 0.11746156878202602, "grad_norm": 1.0890130996704102, "learning_rate": 0.0005, "loss": 3.0107650756835938, "step": 1490 }, { "epoch": 0.11754040204966495, "grad_norm": 0.9325391054153442, "learning_rate": 0.0005, "loss": 2.6204733848571777, "step": 1491 }, { "epoch": 0.11761923531730391, "grad_norm": 0.9335503578186035, "learning_rate": 0.0005, "loss": 2.8749542236328125, "step": 1492 }, { "epoch": 0.11769806858494285, "grad_norm": 1.1195770502090454, "learning_rate": 0.0005, "loss": 2.823504686355591, "step": 1493 }, { "epoch": 0.11777690185258179, "grad_norm": 0.9898496866226196, "learning_rate": 0.0005, "loss": 3.159531354904175, "step": 1494 }, { "epoch": 0.11785573512022074, "grad_norm": 1.2189441919326782, "learning_rate": 0.0005, "loss": 2.7402572631835938, "step": 1495 }, { "epoch": 0.11793456838785968, "grad_norm": 1.6130354404449463, "learning_rate": 0.0005, "loss": 3.2675676345825195, "step": 1496 }, { "epoch": 0.11801340165549862, "grad_norm": 0.8742058277130127, "learning_rate": 0.0005, "loss": 2.9930334091186523, "step": 1497 }, { "epoch": 0.11809223492313757, "grad_norm": 1.3707417249679565, "learning_rate": 0.0005, "loss": 2.7028286457061768, "step": 1498 }, { "epoch": 0.11817106819077651, "grad_norm": 1.083534836769104, "learning_rate": 0.0005, "loss": 2.791172742843628, "step": 1499 }, { "epoch": 0.11824990145841545, "grad_norm": 1.3936697244644165, "learning_rate": 0.0005, "loss": 2.7059850692749023, "step": 1500 }, { "epoch": 0.11832873472605439, "grad_norm": 0.9889855980873108, "learning_rate": 0.0005, "loss": 2.9625401496887207, "step": 1501 }, { "epoch": 0.11840756799369334, "grad_norm": 0.9965956807136536, "learning_rate": 0.0005, "loss": 2.957655429840088, "step": 1502 }, { "epoch": 0.11848640126133228, "grad_norm": 0.9075914025306702, "learning_rate": 0.0005, "loss": 2.632150650024414, "step": 1503 }, { "epoch": 0.11856523452897122, "grad_norm": 1.424237847328186, "learning_rate": 0.0005, "loss": 3.2657828330993652, "step": 1504 }, { "epoch": 0.11864406779661017, "grad_norm": 1.0084084272384644, "learning_rate": 0.0005, "loss": 3.1337783336639404, "step": 1505 }, { "epoch": 0.11872290106424911, "grad_norm": 0.8244855403900146, "learning_rate": 0.0005, "loss": 2.665952444076538, "step": 1506 }, { "epoch": 0.11880173433188805, "grad_norm": 0.9361894726753235, "learning_rate": 0.0005, "loss": 2.659926414489746, "step": 1507 }, { "epoch": 0.118880567599527, "grad_norm": 1.1153348684310913, "learning_rate": 0.0005, "loss": 3.189767360687256, "step": 1508 }, { "epoch": 0.11895940086716594, "grad_norm": 1.0881571769714355, "learning_rate": 0.0005, "loss": 2.7129642963409424, "step": 1509 }, { "epoch": 0.11903823413480488, "grad_norm": 0.9597214460372925, "learning_rate": 0.0005, "loss": 2.8373641967773438, "step": 1510 }, { "epoch": 0.11911706740244384, "grad_norm": 1.042414665222168, "learning_rate": 0.0005, "loss": 3.0189943313598633, "step": 1511 }, { "epoch": 0.11919590067008277, "grad_norm": 1.263583779335022, "learning_rate": 0.0005, "loss": 2.845083236694336, "step": 1512 }, { "epoch": 0.11927473393772171, "grad_norm": 3.5249359607696533, "learning_rate": 0.0005, "loss": 2.723947525024414, "step": 1513 }, { "epoch": 0.11935356720536067, "grad_norm": 1.1884592771530151, "learning_rate": 0.0005, "loss": 3.0679264068603516, "step": 1514 }, { "epoch": 0.1194324004729996, "grad_norm": 1.0582244396209717, "learning_rate": 0.0005, "loss": 2.8885040283203125, "step": 1515 }, { "epoch": 0.11951123374063854, "grad_norm": 1.3937218189239502, "learning_rate": 0.0005, "loss": 2.8212738037109375, "step": 1516 }, { "epoch": 0.1195900670082775, "grad_norm": 1.0707519054412842, "learning_rate": 0.0005, "loss": 2.93806791305542, "step": 1517 }, { "epoch": 0.11966890027591644, "grad_norm": 1.1655994653701782, "learning_rate": 0.0005, "loss": 2.8515758514404297, "step": 1518 }, { "epoch": 0.11974773354355538, "grad_norm": 0.8780364990234375, "learning_rate": 0.0005, "loss": 3.2410287857055664, "step": 1519 }, { "epoch": 0.11982656681119433, "grad_norm": 1.1259841918945312, "learning_rate": 0.0005, "loss": 2.6241464614868164, "step": 1520 }, { "epoch": 0.11990540007883327, "grad_norm": 1.0003913640975952, "learning_rate": 0.0005, "loss": 2.687006950378418, "step": 1521 }, { "epoch": 0.11998423334647221, "grad_norm": 1.466145634651184, "learning_rate": 0.0005, "loss": 2.881392002105713, "step": 1522 }, { "epoch": 0.12006306661411116, "grad_norm": 1.3261598348617554, "learning_rate": 0.0005, "loss": 3.0265774726867676, "step": 1523 }, { "epoch": 0.1201418998817501, "grad_norm": 1.0093460083007812, "learning_rate": 0.0005, "loss": 2.3744821548461914, "step": 1524 }, { "epoch": 0.12022073314938904, "grad_norm": 1.4532004594802856, "learning_rate": 0.0005, "loss": 3.106579542160034, "step": 1525 }, { "epoch": 0.12029956641702799, "grad_norm": 1.2772338390350342, "learning_rate": 0.0005, "loss": 2.900585174560547, "step": 1526 }, { "epoch": 0.12037839968466693, "grad_norm": 1.0836741924285889, "learning_rate": 0.0005, "loss": 3.0405635833740234, "step": 1527 }, { "epoch": 0.12045723295230587, "grad_norm": 1.3114919662475586, "learning_rate": 0.0005, "loss": 2.9721434116363525, "step": 1528 }, { "epoch": 0.12053606621994482, "grad_norm": 1.1877005100250244, "learning_rate": 0.0005, "loss": 2.9028615951538086, "step": 1529 }, { "epoch": 0.12061489948758376, "grad_norm": 1.0551468133926392, "learning_rate": 0.0005, "loss": 2.2685580253601074, "step": 1530 }, { "epoch": 0.1206937327552227, "grad_norm": 1.1787660121917725, "learning_rate": 0.0005, "loss": 3.101198673248291, "step": 1531 }, { "epoch": 0.12077256602286164, "grad_norm": 1.0916376113891602, "learning_rate": 0.0005, "loss": 2.925426483154297, "step": 1532 }, { "epoch": 0.1208513992905006, "grad_norm": 1.0287889242172241, "learning_rate": 0.0005, "loss": 2.730313301086426, "step": 1533 }, { "epoch": 0.12093023255813953, "grad_norm": 0.8919782042503357, "learning_rate": 0.0005, "loss": 3.2646567821502686, "step": 1534 }, { "epoch": 0.12100906582577847, "grad_norm": 1.086782693862915, "learning_rate": 0.0005, "loss": 2.9696311950683594, "step": 1535 }, { "epoch": 0.12108789909341743, "grad_norm": 1.1025054454803467, "learning_rate": 0.0005, "loss": 2.855470895767212, "step": 1536 }, { "epoch": 0.12116673236105636, "grad_norm": 1.026066541671753, "learning_rate": 0.0005, "loss": 2.874544143676758, "step": 1537 }, { "epoch": 0.1212455656286953, "grad_norm": 1.1310619115829468, "learning_rate": 0.0005, "loss": 2.686995029449463, "step": 1538 }, { "epoch": 0.12132439889633426, "grad_norm": 1.1406526565551758, "learning_rate": 0.0005, "loss": 2.702394485473633, "step": 1539 }, { "epoch": 0.1214032321639732, "grad_norm": 1.2642992734909058, "learning_rate": 0.0005, "loss": 2.8082218170166016, "step": 1540 }, { "epoch": 0.12148206543161214, "grad_norm": 0.9892218708992004, "learning_rate": 0.0005, "loss": 2.850400447845459, "step": 1541 }, { "epoch": 0.12156089869925109, "grad_norm": 1.0921177864074707, "learning_rate": 0.0005, "loss": 3.094480514526367, "step": 1542 }, { "epoch": 0.12163973196689003, "grad_norm": 1.2794314622879028, "learning_rate": 0.0005, "loss": 2.983069896697998, "step": 1543 }, { "epoch": 0.12171856523452897, "grad_norm": 1.2353752851486206, "learning_rate": 0.0005, "loss": 2.9217584133148193, "step": 1544 }, { "epoch": 0.12179739850216792, "grad_norm": 1.1161431074142456, "learning_rate": 0.0005, "loss": 3.0958213806152344, "step": 1545 }, { "epoch": 0.12187623176980686, "grad_norm": 1.9977556467056274, "learning_rate": 0.0005, "loss": 2.798044443130493, "step": 1546 }, { "epoch": 0.1219550650374458, "grad_norm": 1.2114278078079224, "learning_rate": 0.0005, "loss": 2.820613384246826, "step": 1547 }, { "epoch": 0.12203389830508475, "grad_norm": 1.2110116481781006, "learning_rate": 0.0005, "loss": 3.333916187286377, "step": 1548 }, { "epoch": 0.12211273157272369, "grad_norm": 1.4004946947097778, "learning_rate": 0.0005, "loss": 2.834191083908081, "step": 1549 }, { "epoch": 0.12219156484036263, "grad_norm": 1.4032268524169922, "learning_rate": 0.0005, "loss": 2.6348047256469727, "step": 1550 }, { "epoch": 0.12227039810800158, "grad_norm": 1.4129785299301147, "learning_rate": 0.0005, "loss": 3.1250553131103516, "step": 1551 }, { "epoch": 0.12234923137564052, "grad_norm": 1.2414859533309937, "learning_rate": 0.0005, "loss": 2.9070732593536377, "step": 1552 }, { "epoch": 0.12242806464327946, "grad_norm": 0.9321284294128418, "learning_rate": 0.0005, "loss": 2.8873820304870605, "step": 1553 }, { "epoch": 0.12250689791091841, "grad_norm": 1.4325145483016968, "learning_rate": 0.0005, "loss": 2.853351354598999, "step": 1554 }, { "epoch": 0.12258573117855735, "grad_norm": 1.8126893043518066, "learning_rate": 0.0005, "loss": 2.7601046562194824, "step": 1555 }, { "epoch": 0.12266456444619629, "grad_norm": 1.055720567703247, "learning_rate": 0.0005, "loss": 3.013666868209839, "step": 1556 }, { "epoch": 0.12274339771383524, "grad_norm": 1.2095081806182861, "learning_rate": 0.0005, "loss": 2.918328285217285, "step": 1557 }, { "epoch": 0.12282223098147418, "grad_norm": 1.18082857131958, "learning_rate": 0.0005, "loss": 2.8935704231262207, "step": 1558 }, { "epoch": 0.12290106424911312, "grad_norm": 1.2749062776565552, "learning_rate": 0.0005, "loss": 2.9591026306152344, "step": 1559 }, { "epoch": 0.12297989751675206, "grad_norm": 1.1906416416168213, "learning_rate": 0.0005, "loss": 2.5415024757385254, "step": 1560 }, { "epoch": 0.12305873078439102, "grad_norm": 3.972587823867798, "learning_rate": 0.0005, "loss": 3.2104647159576416, "step": 1561 }, { "epoch": 0.12313756405202995, "grad_norm": 1.1139615774154663, "learning_rate": 0.0005, "loss": 3.067704916000366, "step": 1562 }, { "epoch": 0.1232163973196689, "grad_norm": 1.1662269830703735, "learning_rate": 0.0005, "loss": 2.9493234157562256, "step": 1563 }, { "epoch": 0.12329523058730785, "grad_norm": 0.9129340052604675, "learning_rate": 0.0005, "loss": 2.977095603942871, "step": 1564 }, { "epoch": 0.12337406385494679, "grad_norm": 1.0899467468261719, "learning_rate": 0.0005, "loss": 2.5421130657196045, "step": 1565 }, { "epoch": 0.12345289712258573, "grad_norm": 1.1362684965133667, "learning_rate": 0.0005, "loss": 2.5323290824890137, "step": 1566 }, { "epoch": 0.12353173039022468, "grad_norm": 1.0192161798477173, "learning_rate": 0.0005, "loss": 2.985523223876953, "step": 1567 }, { "epoch": 0.12361056365786362, "grad_norm": 1.1859986782073975, "learning_rate": 0.0005, "loss": 2.650402069091797, "step": 1568 }, { "epoch": 0.12368939692550256, "grad_norm": 1.1359716653823853, "learning_rate": 0.0005, "loss": 2.8850369453430176, "step": 1569 }, { "epoch": 0.12376823019314151, "grad_norm": 1.1156383752822876, "learning_rate": 0.0005, "loss": 2.768092632293701, "step": 1570 }, { "epoch": 0.12384706346078045, "grad_norm": 0.9677828550338745, "learning_rate": 0.0005, "loss": 2.58931303024292, "step": 1571 }, { "epoch": 0.12392589672841939, "grad_norm": 1.000115990638733, "learning_rate": 0.0005, "loss": 2.647099256515503, "step": 1572 }, { "epoch": 0.12400472999605834, "grad_norm": 1.2579423189163208, "learning_rate": 0.0005, "loss": 2.3095271587371826, "step": 1573 }, { "epoch": 0.12408356326369728, "grad_norm": 1.2846148014068604, "learning_rate": 0.0005, "loss": 2.9844443798065186, "step": 1574 }, { "epoch": 0.12416239653133622, "grad_norm": 1.0765808820724487, "learning_rate": 0.0005, "loss": 3.2292211055755615, "step": 1575 }, { "epoch": 0.12424122979897517, "grad_norm": 0.9902505278587341, "learning_rate": 0.0005, "loss": 2.766094923019409, "step": 1576 }, { "epoch": 0.12432006306661411, "grad_norm": 1.2251136302947998, "learning_rate": 0.0005, "loss": 3.0265512466430664, "step": 1577 }, { "epoch": 0.12439889633425305, "grad_norm": 1.8614345788955688, "learning_rate": 0.0005, "loss": 2.4969654083251953, "step": 1578 }, { "epoch": 0.124477729601892, "grad_norm": 1.1497461795806885, "learning_rate": 0.0005, "loss": 2.661708354949951, "step": 1579 }, { "epoch": 0.12455656286953094, "grad_norm": 0.9735224843025208, "learning_rate": 0.0005, "loss": 2.8514041900634766, "step": 1580 }, { "epoch": 0.12463539613716988, "grad_norm": 1.0297493934631348, "learning_rate": 0.0005, "loss": 2.6521525382995605, "step": 1581 }, { "epoch": 0.12471422940480884, "grad_norm": 1.0414855480194092, "learning_rate": 0.0005, "loss": 3.12109375, "step": 1582 }, { "epoch": 0.12479306267244777, "grad_norm": 0.9997009634971619, "learning_rate": 0.0005, "loss": 3.0118212699890137, "step": 1583 }, { "epoch": 0.12487189594008671, "grad_norm": 0.9712151288986206, "learning_rate": 0.0005, "loss": 2.8567464351654053, "step": 1584 }, { "epoch": 0.12495072920772567, "grad_norm": 1.0593351125717163, "learning_rate": 0.0005, "loss": 2.8521575927734375, "step": 1585 }, { "epoch": 0.1250295624753646, "grad_norm": 0.962336003780365, "learning_rate": 0.0005, "loss": 2.9366703033447266, "step": 1586 }, { "epoch": 0.12510839574300356, "grad_norm": 1.0887593030929565, "learning_rate": 0.0005, "loss": 3.3777968883514404, "step": 1587 }, { "epoch": 0.1251872290106425, "grad_norm": 1.053450107574463, "learning_rate": 0.0005, "loss": 2.955005645751953, "step": 1588 }, { "epoch": 0.12526606227828144, "grad_norm": 1.034590482711792, "learning_rate": 0.0005, "loss": 3.159834623336792, "step": 1589 }, { "epoch": 0.12534489554592038, "grad_norm": 1.0006910562515259, "learning_rate": 0.0005, "loss": 2.7697267532348633, "step": 1590 }, { "epoch": 0.12542372881355932, "grad_norm": 1.1725565195083618, "learning_rate": 0.0005, "loss": 3.203944683074951, "step": 1591 }, { "epoch": 0.12550256208119825, "grad_norm": 1.2075594663619995, "learning_rate": 0.0005, "loss": 2.90084171295166, "step": 1592 }, { "epoch": 0.12558139534883722, "grad_norm": 1.281184196472168, "learning_rate": 0.0005, "loss": 3.1173007488250732, "step": 1593 }, { "epoch": 0.12566022861647616, "grad_norm": 1.246846079826355, "learning_rate": 0.0005, "loss": 3.198415517807007, "step": 1594 }, { "epoch": 0.1257390618841151, "grad_norm": 1.2499728202819824, "learning_rate": 0.0005, "loss": 2.8894269466400146, "step": 1595 }, { "epoch": 0.12581789515175404, "grad_norm": 1.552024245262146, "learning_rate": 0.0005, "loss": 3.2975075244903564, "step": 1596 }, { "epoch": 0.12589672841939298, "grad_norm": 1.0391960144042969, "learning_rate": 0.0005, "loss": 2.929863929748535, "step": 1597 }, { "epoch": 0.12597556168703192, "grad_norm": 1.0652390718460083, "learning_rate": 0.0005, "loss": 3.080223560333252, "step": 1598 }, { "epoch": 0.12605439495467088, "grad_norm": 1.0786784887313843, "learning_rate": 0.0005, "loss": 2.695822238922119, "step": 1599 }, { "epoch": 0.12613322822230982, "grad_norm": 0.9936427474021912, "learning_rate": 0.0005, "loss": 2.675238847732544, "step": 1600 }, { "epoch": 0.12621206148994876, "grad_norm": 1.0450726747512817, "learning_rate": 0.0005, "loss": 3.099757432937622, "step": 1601 }, { "epoch": 0.1262908947575877, "grad_norm": 1.6062593460083008, "learning_rate": 0.0005, "loss": 3.16636061668396, "step": 1602 }, { "epoch": 0.12636972802522664, "grad_norm": 1.053587794303894, "learning_rate": 0.0005, "loss": 2.798535108566284, "step": 1603 }, { "epoch": 0.12644856129286558, "grad_norm": 1.162490725517273, "learning_rate": 0.0005, "loss": 2.6367526054382324, "step": 1604 }, { "epoch": 0.12652739456050452, "grad_norm": 1.2980544567108154, "learning_rate": 0.0005, "loss": 3.0074687004089355, "step": 1605 }, { "epoch": 0.12660622782814349, "grad_norm": 0.8542982339859009, "learning_rate": 0.0005, "loss": 2.5181198120117188, "step": 1606 }, { "epoch": 0.12668506109578243, "grad_norm": 1.2313241958618164, "learning_rate": 0.0005, "loss": 2.777064800262451, "step": 1607 }, { "epoch": 0.12676389436342136, "grad_norm": 1.0228341817855835, "learning_rate": 0.0005, "loss": 2.661759376525879, "step": 1608 }, { "epoch": 0.1268427276310603, "grad_norm": 1.4169108867645264, "learning_rate": 0.0005, "loss": 3.2735629081726074, "step": 1609 }, { "epoch": 0.12692156089869924, "grad_norm": 1.4074954986572266, "learning_rate": 0.0005, "loss": 2.416818141937256, "step": 1610 }, { "epoch": 0.12700039416633818, "grad_norm": 1.2159878015518188, "learning_rate": 0.0005, "loss": 3.2041029930114746, "step": 1611 }, { "epoch": 0.12707922743397715, "grad_norm": 1.08273446559906, "learning_rate": 0.0005, "loss": 2.73624849319458, "step": 1612 }, { "epoch": 0.1271580607016161, "grad_norm": 1.1848236322402954, "learning_rate": 0.0005, "loss": 2.993912696838379, "step": 1613 }, { "epoch": 0.12723689396925503, "grad_norm": 1.0352158546447754, "learning_rate": 0.0005, "loss": 3.0219764709472656, "step": 1614 }, { "epoch": 0.12731572723689397, "grad_norm": 1.1039884090423584, "learning_rate": 0.0005, "loss": 2.9744131565093994, "step": 1615 }, { "epoch": 0.1273945605045329, "grad_norm": 0.9986895322799683, "learning_rate": 0.0005, "loss": 3.0259203910827637, "step": 1616 }, { "epoch": 0.12747339377217184, "grad_norm": 1.7457815408706665, "learning_rate": 0.0005, "loss": 2.79831600189209, "step": 1617 }, { "epoch": 0.1275522270398108, "grad_norm": 0.9507664442062378, "learning_rate": 0.0005, "loss": 3.033644199371338, "step": 1618 }, { "epoch": 0.12763106030744975, "grad_norm": 5.261607646942139, "learning_rate": 0.0005, "loss": 3.536921977996826, "step": 1619 }, { "epoch": 0.1277098935750887, "grad_norm": 1.3656479120254517, "learning_rate": 0.0005, "loss": 2.891298770904541, "step": 1620 }, { "epoch": 0.12778872684272763, "grad_norm": 1.625416874885559, "learning_rate": 0.0005, "loss": 3.0467326641082764, "step": 1621 }, { "epoch": 0.12786756011036657, "grad_norm": 0.9669690728187561, "learning_rate": 0.0005, "loss": 3.170994520187378, "step": 1622 }, { "epoch": 0.1279463933780055, "grad_norm": 1.1232203245162964, "learning_rate": 0.0005, "loss": 2.5143308639526367, "step": 1623 }, { "epoch": 0.12802522664564447, "grad_norm": 1.2655338048934937, "learning_rate": 0.0005, "loss": 3.0093636512756348, "step": 1624 }, { "epoch": 0.1281040599132834, "grad_norm": 1.0340534448623657, "learning_rate": 0.0005, "loss": 3.020676612854004, "step": 1625 }, { "epoch": 0.12818289318092235, "grad_norm": 1.1447935104370117, "learning_rate": 0.0005, "loss": 2.9322423934936523, "step": 1626 }, { "epoch": 0.1282617264485613, "grad_norm": 1.3180662393569946, "learning_rate": 0.0005, "loss": 2.8578226566314697, "step": 1627 }, { "epoch": 0.12834055971620023, "grad_norm": 1.0726863145828247, "learning_rate": 0.0005, "loss": 2.692129135131836, "step": 1628 }, { "epoch": 0.12841939298383917, "grad_norm": 1.2827255725860596, "learning_rate": 0.0005, "loss": 2.7685868740081787, "step": 1629 }, { "epoch": 0.12849822625147814, "grad_norm": 0.9694007039070129, "learning_rate": 0.0005, "loss": 3.151923656463623, "step": 1630 }, { "epoch": 0.12857705951911708, "grad_norm": 1.2912497520446777, "learning_rate": 0.0005, "loss": 2.937666416168213, "step": 1631 }, { "epoch": 0.12865589278675602, "grad_norm": 1.0289207696914673, "learning_rate": 0.0005, "loss": 2.814614772796631, "step": 1632 }, { "epoch": 0.12873472605439495, "grad_norm": 1.0480414628982544, "learning_rate": 0.0005, "loss": 2.9206972122192383, "step": 1633 }, { "epoch": 0.1288135593220339, "grad_norm": 1.1492050886154175, "learning_rate": 0.0005, "loss": 2.80975341796875, "step": 1634 }, { "epoch": 0.12889239258967283, "grad_norm": 1.0185595750808716, "learning_rate": 0.0005, "loss": 2.7681467533111572, "step": 1635 }, { "epoch": 0.12897122585731177, "grad_norm": 1.127426028251648, "learning_rate": 0.0005, "loss": 3.1244382858276367, "step": 1636 }, { "epoch": 0.12905005912495074, "grad_norm": 1.611330270767212, "learning_rate": 0.0005, "loss": 3.0539865493774414, "step": 1637 }, { "epoch": 0.12912889239258968, "grad_norm": 1.1412122249603271, "learning_rate": 0.0005, "loss": 2.925535202026367, "step": 1638 }, { "epoch": 0.12920772566022862, "grad_norm": 1.1046485900878906, "learning_rate": 0.0005, "loss": 2.5958657264709473, "step": 1639 }, { "epoch": 0.12928655892786756, "grad_norm": 1.0139583349227905, "learning_rate": 0.0005, "loss": 3.162841320037842, "step": 1640 }, { "epoch": 0.1293653921955065, "grad_norm": 1.3273484706878662, "learning_rate": 0.0005, "loss": 2.8257856369018555, "step": 1641 }, { "epoch": 0.12944422546314543, "grad_norm": 0.9552870392799377, "learning_rate": 0.0005, "loss": 2.609668016433716, "step": 1642 }, { "epoch": 0.1295230587307844, "grad_norm": 1.139542818069458, "learning_rate": 0.0005, "loss": 2.75740647315979, "step": 1643 }, { "epoch": 0.12960189199842334, "grad_norm": 1.2494661808013916, "learning_rate": 0.0005, "loss": 2.3984920978546143, "step": 1644 }, { "epoch": 0.12968072526606228, "grad_norm": 1.1613328456878662, "learning_rate": 0.0005, "loss": 2.794361114501953, "step": 1645 }, { "epoch": 0.12975955853370122, "grad_norm": 1.0823179483413696, "learning_rate": 0.0005, "loss": 2.9116814136505127, "step": 1646 }, { "epoch": 0.12983839180134016, "grad_norm": 1.3175054788589478, "learning_rate": 0.0005, "loss": 2.92037296295166, "step": 1647 }, { "epoch": 0.1299172250689791, "grad_norm": 0.932970404624939, "learning_rate": 0.0005, "loss": 2.973797082901001, "step": 1648 }, { "epoch": 0.12999605833661806, "grad_norm": 1.5289336442947388, "learning_rate": 0.0005, "loss": 3.0693345069885254, "step": 1649 }, { "epoch": 0.130074891604257, "grad_norm": 1.3852788209915161, "learning_rate": 0.0005, "loss": 3.028978109359741, "step": 1650 }, { "epoch": 0.13015372487189594, "grad_norm": 1.0400325059890747, "learning_rate": 0.0005, "loss": 2.5287089347839355, "step": 1651 }, { "epoch": 0.13023255813953488, "grad_norm": 1.0737099647521973, "learning_rate": 0.0005, "loss": 2.910472869873047, "step": 1652 }, { "epoch": 0.13031139140717382, "grad_norm": 0.9821062684059143, "learning_rate": 0.0005, "loss": 2.8779139518737793, "step": 1653 }, { "epoch": 0.13039022467481276, "grad_norm": 1.157392978668213, "learning_rate": 0.0005, "loss": 2.853240489959717, "step": 1654 }, { "epoch": 0.13046905794245173, "grad_norm": 0.9777337908744812, "learning_rate": 0.0005, "loss": 2.755679130554199, "step": 1655 }, { "epoch": 0.13054789121009067, "grad_norm": 0.9441418051719666, "learning_rate": 0.0005, "loss": 3.2568788528442383, "step": 1656 }, { "epoch": 0.1306267244777296, "grad_norm": 1.0197652578353882, "learning_rate": 0.0005, "loss": 3.047058582305908, "step": 1657 }, { "epoch": 0.13070555774536854, "grad_norm": 0.9789628982543945, "learning_rate": 0.0005, "loss": 2.5245518684387207, "step": 1658 }, { "epoch": 0.13078439101300748, "grad_norm": 1.1914423704147339, "learning_rate": 0.0005, "loss": 2.856171131134033, "step": 1659 }, { "epoch": 0.13086322428064642, "grad_norm": 0.8722607493400574, "learning_rate": 0.0005, "loss": 3.29788875579834, "step": 1660 }, { "epoch": 0.1309420575482854, "grad_norm": 1.0144671201705933, "learning_rate": 0.0005, "loss": 2.9428935050964355, "step": 1661 }, { "epoch": 0.13102089081592433, "grad_norm": 1.0645145177841187, "learning_rate": 0.0005, "loss": 2.7049436569213867, "step": 1662 }, { "epoch": 0.13109972408356327, "grad_norm": 0.9503653049468994, "learning_rate": 0.0005, "loss": 3.1693902015686035, "step": 1663 }, { "epoch": 0.1311785573512022, "grad_norm": 1.2663482427597046, "learning_rate": 0.0005, "loss": 2.7546634674072266, "step": 1664 }, { "epoch": 0.13125739061884115, "grad_norm": 2.205841302871704, "learning_rate": 0.0005, "loss": 2.938115119934082, "step": 1665 }, { "epoch": 0.13133622388648009, "grad_norm": 1.0789885520935059, "learning_rate": 0.0005, "loss": 2.97658109664917, "step": 1666 }, { "epoch": 0.13141505715411902, "grad_norm": 0.9971449375152588, "learning_rate": 0.0005, "loss": 2.7936673164367676, "step": 1667 }, { "epoch": 0.131493890421758, "grad_norm": 1.094607949256897, "learning_rate": 0.0005, "loss": 2.880256414413452, "step": 1668 }, { "epoch": 0.13157272368939693, "grad_norm": 1.329376220703125, "learning_rate": 0.0005, "loss": 2.4199812412261963, "step": 1669 }, { "epoch": 0.13165155695703587, "grad_norm": 1.1322836875915527, "learning_rate": 0.0005, "loss": 2.8704264163970947, "step": 1670 }, { "epoch": 0.1317303902246748, "grad_norm": 1.1205573081970215, "learning_rate": 0.0005, "loss": 2.714057683944702, "step": 1671 }, { "epoch": 0.13180922349231375, "grad_norm": 1.0054770708084106, "learning_rate": 0.0005, "loss": 2.6855454444885254, "step": 1672 }, { "epoch": 0.1318880567599527, "grad_norm": 1.0073975324630737, "learning_rate": 0.0005, "loss": 2.4417405128479004, "step": 1673 }, { "epoch": 0.13196689002759165, "grad_norm": 1.3971024751663208, "learning_rate": 0.0005, "loss": 3.1283130645751953, "step": 1674 }, { "epoch": 0.1320457232952306, "grad_norm": 0.9881331324577332, "learning_rate": 0.0005, "loss": 2.79062557220459, "step": 1675 }, { "epoch": 0.13212455656286953, "grad_norm": 1.2587934732437134, "learning_rate": 0.0005, "loss": 2.936561107635498, "step": 1676 }, { "epoch": 0.13220338983050847, "grad_norm": 1.1867680549621582, "learning_rate": 0.0005, "loss": 2.6146187782287598, "step": 1677 }, { "epoch": 0.1322822230981474, "grad_norm": 0.9750652313232422, "learning_rate": 0.0005, "loss": 2.994478702545166, "step": 1678 }, { "epoch": 0.13236105636578635, "grad_norm": 1.2398251295089722, "learning_rate": 0.0005, "loss": 2.7689385414123535, "step": 1679 }, { "epoch": 0.13243988963342532, "grad_norm": 1.1812429428100586, "learning_rate": 0.0005, "loss": 2.8852391242980957, "step": 1680 }, { "epoch": 0.13251872290106426, "grad_norm": 0.9430133104324341, "learning_rate": 0.0005, "loss": 2.7539894580841064, "step": 1681 }, { "epoch": 0.1325975561687032, "grad_norm": 1.050681233406067, "learning_rate": 0.0005, "loss": 2.9914183616638184, "step": 1682 }, { "epoch": 0.13267638943634213, "grad_norm": 1.2015650272369385, "learning_rate": 0.0005, "loss": 2.873081922531128, "step": 1683 }, { "epoch": 0.13275522270398107, "grad_norm": 0.9602988958358765, "learning_rate": 0.0005, "loss": 2.784435272216797, "step": 1684 }, { "epoch": 0.13283405597162, "grad_norm": 1.3242932558059692, "learning_rate": 0.0005, "loss": 2.7607522010803223, "step": 1685 }, { "epoch": 0.13291288923925898, "grad_norm": 1.082060694694519, "learning_rate": 0.0005, "loss": 3.0290687084198, "step": 1686 }, { "epoch": 0.13299172250689792, "grad_norm": 0.9211710095405579, "learning_rate": 0.0005, "loss": 2.7421865463256836, "step": 1687 }, { "epoch": 0.13307055577453686, "grad_norm": 1.208199381828308, "learning_rate": 0.0005, "loss": 2.799792766571045, "step": 1688 }, { "epoch": 0.1331493890421758, "grad_norm": 0.873684823513031, "learning_rate": 0.0005, "loss": 2.884899377822876, "step": 1689 }, { "epoch": 0.13322822230981474, "grad_norm": 0.9105181694030762, "learning_rate": 0.0005, "loss": 2.6350369453430176, "step": 1690 }, { "epoch": 0.13330705557745368, "grad_norm": 1.2158585786819458, "learning_rate": 0.0005, "loss": 2.9978489875793457, "step": 1691 }, { "epoch": 0.13338588884509264, "grad_norm": 1.476320743560791, "learning_rate": 0.0005, "loss": 3.119534969329834, "step": 1692 }, { "epoch": 0.13346472211273158, "grad_norm": 0.9626395106315613, "learning_rate": 0.0005, "loss": 3.0425758361816406, "step": 1693 }, { "epoch": 0.13354355538037052, "grad_norm": 1.0878105163574219, "learning_rate": 0.0005, "loss": 3.313281536102295, "step": 1694 }, { "epoch": 0.13362238864800946, "grad_norm": 0.9948973059654236, "learning_rate": 0.0005, "loss": 3.168729305267334, "step": 1695 }, { "epoch": 0.1337012219156484, "grad_norm": 0.954817533493042, "learning_rate": 0.0005, "loss": 2.7821483612060547, "step": 1696 }, { "epoch": 0.13378005518328734, "grad_norm": 1.0820229053497314, "learning_rate": 0.0005, "loss": 2.7714834213256836, "step": 1697 }, { "epoch": 0.13385888845092628, "grad_norm": 1.29421865940094, "learning_rate": 0.0005, "loss": 3.0108866691589355, "step": 1698 }, { "epoch": 0.13393772171856524, "grad_norm": 1.2367292642593384, "learning_rate": 0.0005, "loss": 2.6595897674560547, "step": 1699 }, { "epoch": 0.13401655498620418, "grad_norm": 1.4191081523895264, "learning_rate": 0.0005, "loss": 2.978346824645996, "step": 1700 }, { "epoch": 0.13409538825384312, "grad_norm": 1.0546631813049316, "learning_rate": 0.0005, "loss": 2.7109627723693848, "step": 1701 }, { "epoch": 0.13417422152148206, "grad_norm": 1.2526838779449463, "learning_rate": 0.0005, "loss": 2.9048187732696533, "step": 1702 }, { "epoch": 0.134253054789121, "grad_norm": 1.1688300371170044, "learning_rate": 0.0005, "loss": 2.6560792922973633, "step": 1703 }, { "epoch": 0.13433188805675994, "grad_norm": 1.4144673347473145, "learning_rate": 0.0005, "loss": 2.7036616802215576, "step": 1704 }, { "epoch": 0.1344107213243989, "grad_norm": 1.2012325525283813, "learning_rate": 0.0005, "loss": 2.828432559967041, "step": 1705 }, { "epoch": 0.13448955459203785, "grad_norm": 1.9416087865829468, "learning_rate": 0.0005, "loss": 2.621090888977051, "step": 1706 }, { "epoch": 0.13456838785967679, "grad_norm": 1.5708869695663452, "learning_rate": 0.0005, "loss": 3.3492157459259033, "step": 1707 }, { "epoch": 0.13464722112731572, "grad_norm": 1.008966088294983, "learning_rate": 0.0005, "loss": 2.872086524963379, "step": 1708 }, { "epoch": 0.13472605439495466, "grad_norm": 1.2043722867965698, "learning_rate": 0.0005, "loss": 2.8121871948242188, "step": 1709 }, { "epoch": 0.1348048876625936, "grad_norm": 1.2279495000839233, "learning_rate": 0.0005, "loss": 3.039933204650879, "step": 1710 }, { "epoch": 0.13488372093023257, "grad_norm": 1.1094082593917847, "learning_rate": 0.0005, "loss": 3.0116114616394043, "step": 1711 }, { "epoch": 0.1349625541978715, "grad_norm": 1.0097063779830933, "learning_rate": 0.0005, "loss": 2.807342767715454, "step": 1712 }, { "epoch": 0.13504138746551045, "grad_norm": 1.0860413312911987, "learning_rate": 0.0005, "loss": 2.9742324352264404, "step": 1713 }, { "epoch": 0.1351202207331494, "grad_norm": 1.3295601606369019, "learning_rate": 0.0005, "loss": 2.7081079483032227, "step": 1714 }, { "epoch": 0.13519905400078833, "grad_norm": 0.95555180311203, "learning_rate": 0.0005, "loss": 2.75368595123291, "step": 1715 }, { "epoch": 0.13527788726842727, "grad_norm": 1.2550128698349, "learning_rate": 0.0005, "loss": 2.8314030170440674, "step": 1716 }, { "epoch": 0.13535672053606623, "grad_norm": 1.354425072669983, "learning_rate": 0.0005, "loss": 2.7921102046966553, "step": 1717 }, { "epoch": 0.13543555380370517, "grad_norm": 0.9918109178543091, "learning_rate": 0.0005, "loss": 2.759490489959717, "step": 1718 }, { "epoch": 0.1355143870713441, "grad_norm": 1.1149712800979614, "learning_rate": 0.0005, "loss": 3.0813283920288086, "step": 1719 }, { "epoch": 0.13559322033898305, "grad_norm": 1.0547645092010498, "learning_rate": 0.0005, "loss": 2.4275662899017334, "step": 1720 }, { "epoch": 0.135672053606622, "grad_norm": 1.1810967922210693, "learning_rate": 0.0005, "loss": 2.7592005729675293, "step": 1721 }, { "epoch": 0.13575088687426093, "grad_norm": 1.219088077545166, "learning_rate": 0.0005, "loss": 2.614795446395874, "step": 1722 }, { "epoch": 0.1358297201418999, "grad_norm": 1.398515224456787, "learning_rate": 0.0005, "loss": 3.168058156967163, "step": 1723 }, { "epoch": 0.13590855340953883, "grad_norm": 1.2492597103118896, "learning_rate": 0.0005, "loss": 2.9651598930358887, "step": 1724 }, { "epoch": 0.13598738667717777, "grad_norm": 0.9572410583496094, "learning_rate": 0.0005, "loss": 2.972810983657837, "step": 1725 }, { "epoch": 0.1360662199448167, "grad_norm": 1.17063570022583, "learning_rate": 0.0005, "loss": 2.693600654602051, "step": 1726 }, { "epoch": 0.13614505321245565, "grad_norm": 1.2693302631378174, "learning_rate": 0.0005, "loss": 2.8800461292266846, "step": 1727 }, { "epoch": 0.1362238864800946, "grad_norm": 1.1139767169952393, "learning_rate": 0.0005, "loss": 3.137040138244629, "step": 1728 }, { "epoch": 0.13630271974773353, "grad_norm": 1.0682860612869263, "learning_rate": 0.0005, "loss": 3.101653575897217, "step": 1729 }, { "epoch": 0.1363815530153725, "grad_norm": 1.1759798526763916, "learning_rate": 0.0005, "loss": 3.2279300689697266, "step": 1730 }, { "epoch": 0.13646038628301144, "grad_norm": 1.1070321798324585, "learning_rate": 0.0005, "loss": 2.7228479385375977, "step": 1731 }, { "epoch": 0.13653921955065038, "grad_norm": 0.9086763858795166, "learning_rate": 0.0005, "loss": 3.050550699234009, "step": 1732 }, { "epoch": 0.13661805281828932, "grad_norm": 0.9292863011360168, "learning_rate": 0.0005, "loss": 2.6476101875305176, "step": 1733 }, { "epoch": 0.13669688608592825, "grad_norm": 0.972629964351654, "learning_rate": 0.0005, "loss": 3.0060911178588867, "step": 1734 }, { "epoch": 0.1367757193535672, "grad_norm": 1.5517189502716064, "learning_rate": 0.0005, "loss": 3.033259630203247, "step": 1735 }, { "epoch": 0.13685455262120616, "grad_norm": 1.3255809545516968, "learning_rate": 0.0005, "loss": 2.580902338027954, "step": 1736 }, { "epoch": 0.1369333858888451, "grad_norm": 1.076076865196228, "learning_rate": 0.0005, "loss": 3.176163673400879, "step": 1737 }, { "epoch": 0.13701221915648404, "grad_norm": 0.9062820076942444, "learning_rate": 0.0005, "loss": 2.6472153663635254, "step": 1738 }, { "epoch": 0.13709105242412298, "grad_norm": 1.2890907526016235, "learning_rate": 0.0005, "loss": 2.85630464553833, "step": 1739 }, { "epoch": 0.13716988569176192, "grad_norm": 0.8745626211166382, "learning_rate": 0.0005, "loss": 2.7616682052612305, "step": 1740 }, { "epoch": 0.13724871895940086, "grad_norm": 1.236469030380249, "learning_rate": 0.0005, "loss": 3.1359376907348633, "step": 1741 }, { "epoch": 0.13732755222703982, "grad_norm": 1.1065921783447266, "learning_rate": 0.0005, "loss": 2.539182186126709, "step": 1742 }, { "epoch": 0.13740638549467876, "grad_norm": 0.9219560027122498, "learning_rate": 0.0005, "loss": 2.623783588409424, "step": 1743 }, { "epoch": 0.1374852187623177, "grad_norm": 0.9499765634536743, "learning_rate": 0.0005, "loss": 2.5465340614318848, "step": 1744 }, { "epoch": 0.13756405202995664, "grad_norm": 4.305906295776367, "learning_rate": 0.0005, "loss": 2.623396396636963, "step": 1745 }, { "epoch": 0.13764288529759558, "grad_norm": 1.2120091915130615, "learning_rate": 0.0005, "loss": 2.9949450492858887, "step": 1746 }, { "epoch": 0.13772171856523452, "grad_norm": 1.351003885269165, "learning_rate": 0.0005, "loss": 2.9877617359161377, "step": 1747 }, { "epoch": 0.13780055183287349, "grad_norm": 1.053305983543396, "learning_rate": 0.0005, "loss": 2.979727029800415, "step": 1748 }, { "epoch": 0.13787938510051242, "grad_norm": 1.1635524034500122, "learning_rate": 0.0005, "loss": 2.392331600189209, "step": 1749 }, { "epoch": 0.13795821836815136, "grad_norm": 1.0575958490371704, "learning_rate": 0.0005, "loss": 2.782444953918457, "step": 1750 }, { "epoch": 0.1380370516357903, "grad_norm": 1.174156665802002, "learning_rate": 0.0005, "loss": 3.1390602588653564, "step": 1751 }, { "epoch": 0.13811588490342924, "grad_norm": 1.7133574485778809, "learning_rate": 0.0005, "loss": 2.8418078422546387, "step": 1752 }, { "epoch": 0.13819471817106818, "grad_norm": 1.268619179725647, "learning_rate": 0.0005, "loss": 2.994295597076416, "step": 1753 }, { "epoch": 0.13827355143870712, "grad_norm": 1.0770248174667358, "learning_rate": 0.0005, "loss": 2.898499011993408, "step": 1754 }, { "epoch": 0.1383523847063461, "grad_norm": 1.5251233577728271, "learning_rate": 0.0005, "loss": 3.371042013168335, "step": 1755 }, { "epoch": 0.13843121797398503, "grad_norm": 1.0522005558013916, "learning_rate": 0.0005, "loss": 2.8727219104766846, "step": 1756 }, { "epoch": 0.13851005124162397, "grad_norm": 0.9201583862304688, "learning_rate": 0.0005, "loss": 2.7201170921325684, "step": 1757 }, { "epoch": 0.1385888845092629, "grad_norm": 1.1608823537826538, "learning_rate": 0.0005, "loss": 2.4803247451782227, "step": 1758 }, { "epoch": 0.13866771777690184, "grad_norm": 1.0033234357833862, "learning_rate": 0.0005, "loss": 2.8328166007995605, "step": 1759 }, { "epoch": 0.13874655104454078, "grad_norm": 1.094343662261963, "learning_rate": 0.0005, "loss": 2.796741008758545, "step": 1760 }, { "epoch": 0.13882538431217975, "grad_norm": 1.0457197427749634, "learning_rate": 0.0005, "loss": 3.0446951389312744, "step": 1761 }, { "epoch": 0.1389042175798187, "grad_norm": 1.1046502590179443, "learning_rate": 0.0005, "loss": 2.922976016998291, "step": 1762 }, { "epoch": 0.13898305084745763, "grad_norm": 1.6360039710998535, "learning_rate": 0.0005, "loss": 2.640903949737549, "step": 1763 }, { "epoch": 0.13906188411509657, "grad_norm": 1.0672463178634644, "learning_rate": 0.0005, "loss": 2.9185917377471924, "step": 1764 }, { "epoch": 0.1391407173827355, "grad_norm": 1.792411208152771, "learning_rate": 0.0005, "loss": 2.5803794860839844, "step": 1765 }, { "epoch": 0.13921955065037445, "grad_norm": 1.1969822645187378, "learning_rate": 0.0005, "loss": 2.6858086585998535, "step": 1766 }, { "epoch": 0.1392983839180134, "grad_norm": 0.9758456945419312, "learning_rate": 0.0005, "loss": 2.986234188079834, "step": 1767 }, { "epoch": 0.13937721718565235, "grad_norm": 1.2100119590759277, "learning_rate": 0.0005, "loss": 2.9536795616149902, "step": 1768 }, { "epoch": 0.1394560504532913, "grad_norm": 1.9811725616455078, "learning_rate": 0.0005, "loss": 2.9020843505859375, "step": 1769 }, { "epoch": 0.13953488372093023, "grad_norm": 1.0866483449935913, "learning_rate": 0.0005, "loss": 2.923783779144287, "step": 1770 }, { "epoch": 0.13961371698856917, "grad_norm": 1.3030906915664673, "learning_rate": 0.0005, "loss": 3.14605450630188, "step": 1771 }, { "epoch": 0.1396925502562081, "grad_norm": 1.1857975721359253, "learning_rate": 0.0005, "loss": 2.7044572830200195, "step": 1772 }, { "epoch": 0.13977138352384708, "grad_norm": 0.8775384426116943, "learning_rate": 0.0005, "loss": 2.4020121097564697, "step": 1773 }, { "epoch": 0.13985021679148602, "grad_norm": 0.9515107870101929, "learning_rate": 0.0005, "loss": 2.8994040489196777, "step": 1774 }, { "epoch": 0.13992905005912495, "grad_norm": 1.514914631843567, "learning_rate": 0.0005, "loss": 2.843987464904785, "step": 1775 }, { "epoch": 0.1400078833267639, "grad_norm": 1.081474781036377, "learning_rate": 0.0005, "loss": 2.5841329097747803, "step": 1776 }, { "epoch": 0.14008671659440283, "grad_norm": 1.0568716526031494, "learning_rate": 0.0005, "loss": 2.8816022872924805, "step": 1777 }, { "epoch": 0.14016554986204177, "grad_norm": 1.0105434656143188, "learning_rate": 0.0005, "loss": 2.8808844089508057, "step": 1778 }, { "epoch": 0.14024438312968074, "grad_norm": 1.1311531066894531, "learning_rate": 0.0005, "loss": 2.879823684692383, "step": 1779 }, { "epoch": 0.14032321639731968, "grad_norm": 1.2619349956512451, "learning_rate": 0.0005, "loss": 2.7642273902893066, "step": 1780 }, { "epoch": 0.14040204966495862, "grad_norm": 1.6120715141296387, "learning_rate": 0.0005, "loss": 2.584838390350342, "step": 1781 }, { "epoch": 0.14048088293259756, "grad_norm": 1.0216025114059448, "learning_rate": 0.0005, "loss": 3.042299747467041, "step": 1782 }, { "epoch": 0.1405597162002365, "grad_norm": 1.1719567775726318, "learning_rate": 0.0005, "loss": 3.053906202316284, "step": 1783 }, { "epoch": 0.14063854946787543, "grad_norm": 1.0160081386566162, "learning_rate": 0.0005, "loss": 3.16838002204895, "step": 1784 }, { "epoch": 0.14071738273551437, "grad_norm": 1.1012787818908691, "learning_rate": 0.0005, "loss": 2.8069567680358887, "step": 1785 }, { "epoch": 0.14079621600315334, "grad_norm": 1.160557508468628, "learning_rate": 0.0005, "loss": 2.590573787689209, "step": 1786 }, { "epoch": 0.14087504927079228, "grad_norm": 1.0063910484313965, "learning_rate": 0.0005, "loss": 3.0184476375579834, "step": 1787 }, { "epoch": 0.14095388253843122, "grad_norm": 2.171872138977051, "learning_rate": 0.0005, "loss": 2.7953391075134277, "step": 1788 }, { "epoch": 0.14103271580607016, "grad_norm": 0.9805051684379578, "learning_rate": 0.0005, "loss": 2.573307991027832, "step": 1789 }, { "epoch": 0.1411115490737091, "grad_norm": 3.4907386302948, "learning_rate": 0.0005, "loss": 2.8352067470550537, "step": 1790 }, { "epoch": 0.14119038234134804, "grad_norm": 1.4283174276351929, "learning_rate": 0.0005, "loss": 2.7544736862182617, "step": 1791 }, { "epoch": 0.141269215608987, "grad_norm": 1.1748989820480347, "learning_rate": 0.0005, "loss": 2.668288230895996, "step": 1792 }, { "epoch": 0.14134804887662594, "grad_norm": 1.3289871215820312, "learning_rate": 0.0005, "loss": 2.3376612663269043, "step": 1793 }, { "epoch": 0.14142688214426488, "grad_norm": 1.9303829669952393, "learning_rate": 0.0005, "loss": 2.9802613258361816, "step": 1794 }, { "epoch": 0.14150571541190382, "grad_norm": 1.4621610641479492, "learning_rate": 0.0005, "loss": 3.006744861602783, "step": 1795 }, { "epoch": 0.14158454867954276, "grad_norm": 1.0776309967041016, "learning_rate": 0.0005, "loss": 2.9490275382995605, "step": 1796 }, { "epoch": 0.1416633819471817, "grad_norm": 0.9584934115409851, "learning_rate": 0.0005, "loss": 2.5968940258026123, "step": 1797 }, { "epoch": 0.14174221521482067, "grad_norm": 1.7023580074310303, "learning_rate": 0.0005, "loss": 2.982880115509033, "step": 1798 }, { "epoch": 0.1418210484824596, "grad_norm": 1.1633491516113281, "learning_rate": 0.0005, "loss": 3.027505874633789, "step": 1799 }, { "epoch": 0.14189988175009854, "grad_norm": 1.1080503463745117, "learning_rate": 0.0005, "loss": 2.8558428287506104, "step": 1800 }, { "epoch": 0.14197871501773748, "grad_norm": 1.0059095621109009, "learning_rate": 0.0005, "loss": 2.7380239963531494, "step": 1801 }, { "epoch": 0.14205754828537642, "grad_norm": 0.9526697397232056, "learning_rate": 0.0005, "loss": 3.0298898220062256, "step": 1802 }, { "epoch": 0.14213638155301536, "grad_norm": 1.2949868440628052, "learning_rate": 0.0005, "loss": 3.2964444160461426, "step": 1803 }, { "epoch": 0.14221521482065433, "grad_norm": 1.0547411441802979, "learning_rate": 0.0005, "loss": 2.663827657699585, "step": 1804 }, { "epoch": 0.14229404808829327, "grad_norm": 1.402862787246704, "learning_rate": 0.0005, "loss": 3.008727550506592, "step": 1805 }, { "epoch": 0.1423728813559322, "grad_norm": 2.491114377975464, "learning_rate": 0.0005, "loss": 3.0772595405578613, "step": 1806 }, { "epoch": 0.14245171462357115, "grad_norm": 2.6022043228149414, "learning_rate": 0.0005, "loss": 2.679368734359741, "step": 1807 }, { "epoch": 0.14253054789121009, "grad_norm": 1.104771375656128, "learning_rate": 0.0005, "loss": 2.86348032951355, "step": 1808 }, { "epoch": 0.14260938115884902, "grad_norm": 1.214124321937561, "learning_rate": 0.0005, "loss": 2.436108112335205, "step": 1809 }, { "epoch": 0.142688214426488, "grad_norm": 1.2552441358566284, "learning_rate": 0.0005, "loss": 3.057016611099243, "step": 1810 }, { "epoch": 0.14276704769412693, "grad_norm": 1.073366641998291, "learning_rate": 0.0005, "loss": 3.0550851821899414, "step": 1811 }, { "epoch": 0.14284588096176587, "grad_norm": 0.9305434823036194, "learning_rate": 0.0005, "loss": 3.0761420726776123, "step": 1812 }, { "epoch": 0.1429247142294048, "grad_norm": 1.1262511014938354, "learning_rate": 0.0005, "loss": 2.7344436645507812, "step": 1813 }, { "epoch": 0.14300354749704375, "grad_norm": 1.1933274269104004, "learning_rate": 0.0005, "loss": 2.95027494430542, "step": 1814 }, { "epoch": 0.1430823807646827, "grad_norm": 1.4989718198776245, "learning_rate": 0.0005, "loss": 2.9569034576416016, "step": 1815 }, { "epoch": 0.14316121403232163, "grad_norm": 1.1039948463439941, "learning_rate": 0.0005, "loss": 2.8428988456726074, "step": 1816 }, { "epoch": 0.1432400472999606, "grad_norm": 1.0896700620651245, "learning_rate": 0.0005, "loss": 2.653676748275757, "step": 1817 }, { "epoch": 0.14331888056759953, "grad_norm": 0.9892072081565857, "learning_rate": 0.0005, "loss": 2.8283729553222656, "step": 1818 }, { "epoch": 0.14339771383523847, "grad_norm": 1.0345627069473267, "learning_rate": 0.0005, "loss": 2.889812469482422, "step": 1819 }, { "epoch": 0.1434765471028774, "grad_norm": 1.2472374439239502, "learning_rate": 0.0005, "loss": 2.9043569564819336, "step": 1820 }, { "epoch": 0.14355538037051635, "grad_norm": 1.17191743850708, "learning_rate": 0.0005, "loss": 2.88649845123291, "step": 1821 }, { "epoch": 0.1436342136381553, "grad_norm": 1.0804157257080078, "learning_rate": 0.0005, "loss": 2.1823806762695312, "step": 1822 }, { "epoch": 0.14371304690579426, "grad_norm": 1.1037657260894775, "learning_rate": 0.0005, "loss": 2.5156633853912354, "step": 1823 }, { "epoch": 0.1437918801734332, "grad_norm": 1.040097951889038, "learning_rate": 0.0005, "loss": 2.454925775527954, "step": 1824 }, { "epoch": 0.14387071344107213, "grad_norm": 1.2882030010223389, "learning_rate": 0.0005, "loss": 2.8645811080932617, "step": 1825 }, { "epoch": 0.14394954670871107, "grad_norm": 1.1760388612747192, "learning_rate": 0.0005, "loss": 3.029366970062256, "step": 1826 }, { "epoch": 0.14402837997635, "grad_norm": 1.1622450351715088, "learning_rate": 0.0005, "loss": 2.7712159156799316, "step": 1827 }, { "epoch": 0.14410721324398895, "grad_norm": 1.4654940366744995, "learning_rate": 0.0005, "loss": 2.566767454147339, "step": 1828 }, { "epoch": 0.14418604651162792, "grad_norm": 1.265681266784668, "learning_rate": 0.0005, "loss": 2.796346426010132, "step": 1829 }, { "epoch": 0.14426487977926686, "grad_norm": 0.7981548309326172, "learning_rate": 0.0005, "loss": 2.975027561187744, "step": 1830 }, { "epoch": 0.1443437130469058, "grad_norm": 1.0858150720596313, "learning_rate": 0.0005, "loss": 2.9394760131835938, "step": 1831 }, { "epoch": 0.14442254631454474, "grad_norm": 1.178911566734314, "learning_rate": 0.0005, "loss": 2.879945755004883, "step": 1832 }, { "epoch": 0.14450137958218368, "grad_norm": 1.144688367843628, "learning_rate": 0.0005, "loss": 2.9243781566619873, "step": 1833 }, { "epoch": 0.14458021284982261, "grad_norm": 1.4787752628326416, "learning_rate": 0.0005, "loss": 2.94765043258667, "step": 1834 }, { "epoch": 0.14465904611746158, "grad_norm": 0.967022716999054, "learning_rate": 0.0005, "loss": 2.617079257965088, "step": 1835 }, { "epoch": 0.14473787938510052, "grad_norm": 1.2734394073486328, "learning_rate": 0.0005, "loss": 3.0059022903442383, "step": 1836 }, { "epoch": 0.14481671265273946, "grad_norm": 1.3066164255142212, "learning_rate": 0.0005, "loss": 2.586483955383301, "step": 1837 }, { "epoch": 0.1448955459203784, "grad_norm": 1.2165179252624512, "learning_rate": 0.0005, "loss": 3.0177531242370605, "step": 1838 }, { "epoch": 0.14497437918801734, "grad_norm": 1.6378053426742554, "learning_rate": 0.0005, "loss": 2.4973626136779785, "step": 1839 }, { "epoch": 0.14505321245565628, "grad_norm": 2.118826389312744, "learning_rate": 0.0005, "loss": 2.9589881896972656, "step": 1840 }, { "epoch": 0.14513204572329524, "grad_norm": 1.0954396724700928, "learning_rate": 0.0005, "loss": 2.7598843574523926, "step": 1841 }, { "epoch": 0.14521087899093418, "grad_norm": 0.9838328957557678, "learning_rate": 0.0005, "loss": 2.643237590789795, "step": 1842 }, { "epoch": 0.14528971225857312, "grad_norm": 1.1374261379241943, "learning_rate": 0.0005, "loss": 2.5800347328186035, "step": 1843 }, { "epoch": 0.14536854552621206, "grad_norm": 0.8965299725532532, "learning_rate": 0.0005, "loss": 2.8348336219787598, "step": 1844 }, { "epoch": 0.145447378793851, "grad_norm": 1.0765725374221802, "learning_rate": 0.0005, "loss": 2.950108528137207, "step": 1845 }, { "epoch": 0.14552621206148994, "grad_norm": 1.128364086151123, "learning_rate": 0.0005, "loss": 2.5482988357543945, "step": 1846 }, { "epoch": 0.14560504532912888, "grad_norm": 3.155764102935791, "learning_rate": 0.0005, "loss": 2.5787124633789062, "step": 1847 }, { "epoch": 0.14568387859676785, "grad_norm": 1.2927584648132324, "learning_rate": 0.0005, "loss": 2.6866044998168945, "step": 1848 }, { "epoch": 0.14576271186440679, "grad_norm": 1.3012384176254272, "learning_rate": 0.0005, "loss": 3.093453884124756, "step": 1849 }, { "epoch": 0.14584154513204572, "grad_norm": 0.9110006093978882, "learning_rate": 0.0005, "loss": 2.626735210418701, "step": 1850 }, { "epoch": 0.14592037839968466, "grad_norm": 1.066279411315918, "learning_rate": 0.0005, "loss": 2.6557693481445312, "step": 1851 }, { "epoch": 0.1459992116673236, "grad_norm": 1.2638872861862183, "learning_rate": 0.0005, "loss": 2.5559020042419434, "step": 1852 }, { "epoch": 0.14607804493496254, "grad_norm": 1.2119925022125244, "learning_rate": 0.0005, "loss": 3.240020990371704, "step": 1853 }, { "epoch": 0.1461568782026015, "grad_norm": 1.136732578277588, "learning_rate": 0.0005, "loss": 2.8062386512756348, "step": 1854 }, { "epoch": 0.14623571147024045, "grad_norm": 1.2370365858078003, "learning_rate": 0.0005, "loss": 2.5455589294433594, "step": 1855 }, { "epoch": 0.1463145447378794, "grad_norm": 1.3877019882202148, "learning_rate": 0.0005, "loss": 2.847500801086426, "step": 1856 }, { "epoch": 0.14639337800551833, "grad_norm": 0.9443190097808838, "learning_rate": 0.0005, "loss": 2.876828193664551, "step": 1857 }, { "epoch": 0.14647221127315727, "grad_norm": 1.1658574342727661, "learning_rate": 0.0005, "loss": 2.489067316055298, "step": 1858 }, { "epoch": 0.1465510445407962, "grad_norm": 1.0656782388687134, "learning_rate": 0.0005, "loss": 3.0173041820526123, "step": 1859 }, { "epoch": 0.14662987780843517, "grad_norm": 1.0082669258117676, "learning_rate": 0.0005, "loss": 2.835348129272461, "step": 1860 }, { "epoch": 0.1467087110760741, "grad_norm": 1.0171277523040771, "learning_rate": 0.0005, "loss": 2.735750198364258, "step": 1861 }, { "epoch": 0.14678754434371305, "grad_norm": 1.112754464149475, "learning_rate": 0.0005, "loss": 2.6251580715179443, "step": 1862 }, { "epoch": 0.146866377611352, "grad_norm": 1.0092629194259644, "learning_rate": 0.0005, "loss": 3.2690818309783936, "step": 1863 }, { "epoch": 0.14694521087899093, "grad_norm": 0.983051598072052, "learning_rate": 0.0005, "loss": 2.356128692626953, "step": 1864 }, { "epoch": 0.14702404414662987, "grad_norm": 1.0257433652877808, "learning_rate": 0.0005, "loss": 2.8541269302368164, "step": 1865 }, { "epoch": 0.14710287741426883, "grad_norm": 1.1292409896850586, "learning_rate": 0.0005, "loss": 2.844078302383423, "step": 1866 }, { "epoch": 0.14718171068190777, "grad_norm": 2.0655641555786133, "learning_rate": 0.0005, "loss": 2.827791213989258, "step": 1867 }, { "epoch": 0.1472605439495467, "grad_norm": 1.2900694608688354, "learning_rate": 0.0005, "loss": 2.8041276931762695, "step": 1868 }, { "epoch": 0.14733937721718565, "grad_norm": 1.023074746131897, "learning_rate": 0.0005, "loss": 2.7516441345214844, "step": 1869 }, { "epoch": 0.1474182104848246, "grad_norm": 0.9802998304367065, "learning_rate": 0.0005, "loss": 2.970686435699463, "step": 1870 }, { "epoch": 0.14749704375246353, "grad_norm": 0.9377759099006653, "learning_rate": 0.0005, "loss": 2.595452308654785, "step": 1871 }, { "epoch": 0.14757587702010247, "grad_norm": 1.4988596439361572, "learning_rate": 0.0005, "loss": 2.9806301593780518, "step": 1872 }, { "epoch": 0.14765471028774144, "grad_norm": 1.0943453311920166, "learning_rate": 0.0005, "loss": 2.9433746337890625, "step": 1873 }, { "epoch": 0.14773354355538038, "grad_norm": 1.139263391494751, "learning_rate": 0.0005, "loss": 2.6829168796539307, "step": 1874 }, { "epoch": 0.14781237682301931, "grad_norm": 0.9290070533752441, "learning_rate": 0.0005, "loss": 2.3293590545654297, "step": 1875 }, { "epoch": 0.14789121009065825, "grad_norm": 1.0515639781951904, "learning_rate": 0.0005, "loss": 2.805360794067383, "step": 1876 }, { "epoch": 0.1479700433582972, "grad_norm": 1.1454744338989258, "learning_rate": 0.0005, "loss": 2.802572250366211, "step": 1877 }, { "epoch": 0.14804887662593613, "grad_norm": 1.392616629600525, "learning_rate": 0.0005, "loss": 3.3266334533691406, "step": 1878 }, { "epoch": 0.1481277098935751, "grad_norm": 0.9811335206031799, "learning_rate": 0.0005, "loss": 2.6435089111328125, "step": 1879 }, { "epoch": 0.14820654316121404, "grad_norm": 1.0866886377334595, "learning_rate": 0.0005, "loss": 2.8877644538879395, "step": 1880 }, { "epoch": 0.14828537642885298, "grad_norm": 1.5076874494552612, "learning_rate": 0.0005, "loss": 2.749673366546631, "step": 1881 }, { "epoch": 0.14836420969649192, "grad_norm": 1.038022518157959, "learning_rate": 0.0005, "loss": 2.972294807434082, "step": 1882 }, { "epoch": 0.14844304296413086, "grad_norm": 0.930609405040741, "learning_rate": 0.0005, "loss": 2.86433744430542, "step": 1883 }, { "epoch": 0.1485218762317698, "grad_norm": 1.456841230392456, "learning_rate": 0.0005, "loss": 2.6716084480285645, "step": 1884 }, { "epoch": 0.14860070949940876, "grad_norm": 1.080444574356079, "learning_rate": 0.0005, "loss": 2.7648773193359375, "step": 1885 }, { "epoch": 0.1486795427670477, "grad_norm": 1.1613514423370361, "learning_rate": 0.0005, "loss": 3.094876289367676, "step": 1886 }, { "epoch": 0.14875837603468664, "grad_norm": 1.2601648569107056, "learning_rate": 0.0005, "loss": 2.8409295082092285, "step": 1887 }, { "epoch": 0.14883720930232558, "grad_norm": 1.0111796855926514, "learning_rate": 0.0005, "loss": 2.7886977195739746, "step": 1888 }, { "epoch": 0.14891604256996452, "grad_norm": 0.9286566972732544, "learning_rate": 0.0005, "loss": 2.701531410217285, "step": 1889 }, { "epoch": 0.14899487583760346, "grad_norm": 0.8727455139160156, "learning_rate": 0.0005, "loss": 3.3565478324890137, "step": 1890 }, { "epoch": 0.14907370910524242, "grad_norm": 0.9612505435943604, "learning_rate": 0.0005, "loss": 2.8834590911865234, "step": 1891 }, { "epoch": 0.14915254237288136, "grad_norm": 1.0350896120071411, "learning_rate": 0.0005, "loss": 2.824312925338745, "step": 1892 }, { "epoch": 0.1492313756405203, "grad_norm": 0.946471631526947, "learning_rate": 0.0005, "loss": 2.8609070777893066, "step": 1893 }, { "epoch": 0.14931020890815924, "grad_norm": 1.3387866020202637, "learning_rate": 0.0005, "loss": 2.6725733280181885, "step": 1894 }, { "epoch": 0.14938904217579818, "grad_norm": 1.0727319717407227, "learning_rate": 0.0005, "loss": 2.7897019386291504, "step": 1895 }, { "epoch": 0.14946787544343712, "grad_norm": 1.1548668146133423, "learning_rate": 0.0005, "loss": 2.6402769088745117, "step": 1896 }, { "epoch": 0.1495467087110761, "grad_norm": 1.1306521892547607, "learning_rate": 0.0005, "loss": 2.8718581199645996, "step": 1897 }, { "epoch": 0.14962554197871503, "grad_norm": 1.1824883222579956, "learning_rate": 0.0005, "loss": 2.9022324085235596, "step": 1898 }, { "epoch": 0.14970437524635397, "grad_norm": 0.9500686526298523, "learning_rate": 0.0005, "loss": 2.624124050140381, "step": 1899 }, { "epoch": 0.1497832085139929, "grad_norm": 1.2770302295684814, "learning_rate": 0.0005, "loss": 3.3116979598999023, "step": 1900 }, { "epoch": 0.14986204178163184, "grad_norm": 1.1241511106491089, "learning_rate": 0.0005, "loss": 2.91922664642334, "step": 1901 }, { "epoch": 0.14994087504927078, "grad_norm": 0.9895510077476501, "learning_rate": 0.0005, "loss": 2.597586154937744, "step": 1902 }, { "epoch": 0.15001970831690972, "grad_norm": 1.4791409969329834, "learning_rate": 0.0005, "loss": 2.745314836502075, "step": 1903 }, { "epoch": 0.1500985415845487, "grad_norm": 1.1579582691192627, "learning_rate": 0.0005, "loss": 2.68113374710083, "step": 1904 }, { "epoch": 0.15017737485218763, "grad_norm": 1.4044888019561768, "learning_rate": 0.0005, "loss": 2.282658338546753, "step": 1905 }, { "epoch": 0.15025620811982657, "grad_norm": 1.277429461479187, "learning_rate": 0.0005, "loss": 2.984318256378174, "step": 1906 }, { "epoch": 0.1503350413874655, "grad_norm": 1.0104637145996094, "learning_rate": 0.0005, "loss": 2.8673524856567383, "step": 1907 }, { "epoch": 0.15041387465510445, "grad_norm": 0.9992188811302185, "learning_rate": 0.0005, "loss": 2.946249008178711, "step": 1908 }, { "epoch": 0.15049270792274339, "grad_norm": 1.0680952072143555, "learning_rate": 0.0005, "loss": 2.541774272918701, "step": 1909 }, { "epoch": 0.15057154119038235, "grad_norm": 1.1992614269256592, "learning_rate": 0.0005, "loss": 3.028080940246582, "step": 1910 }, { "epoch": 0.1506503744580213, "grad_norm": 0.8884642720222473, "learning_rate": 0.0005, "loss": 2.780183792114258, "step": 1911 }, { "epoch": 0.15072920772566023, "grad_norm": 1.287137746810913, "learning_rate": 0.0005, "loss": 2.805609703063965, "step": 1912 }, { "epoch": 0.15080804099329917, "grad_norm": 1.1435266733169556, "learning_rate": 0.0005, "loss": 2.7381529808044434, "step": 1913 }, { "epoch": 0.1508868742609381, "grad_norm": 1.4431551694869995, "learning_rate": 0.0005, "loss": 3.099433422088623, "step": 1914 }, { "epoch": 0.15096570752857705, "grad_norm": 1.1303131580352783, "learning_rate": 0.0005, "loss": 2.968789577484131, "step": 1915 }, { "epoch": 0.15104454079621601, "grad_norm": 1.1791225671768188, "learning_rate": 0.0005, "loss": 3.004091739654541, "step": 1916 }, { "epoch": 0.15112337406385495, "grad_norm": 1.4294936656951904, "learning_rate": 0.0005, "loss": 2.731205463409424, "step": 1917 }, { "epoch": 0.1512022073314939, "grad_norm": 1.2908005714416504, "learning_rate": 0.0005, "loss": 3.03298282623291, "step": 1918 }, { "epoch": 0.15128104059913283, "grad_norm": 1.1055082082748413, "learning_rate": 0.0005, "loss": 3.2102324962615967, "step": 1919 }, { "epoch": 0.15135987386677177, "grad_norm": 1.0782254934310913, "learning_rate": 0.0005, "loss": 2.4080092906951904, "step": 1920 }, { "epoch": 0.1514387071344107, "grad_norm": 1.0200334787368774, "learning_rate": 0.0005, "loss": 2.7714247703552246, "step": 1921 }, { "epoch": 0.15151754040204968, "grad_norm": 1.1293292045593262, "learning_rate": 0.0005, "loss": 2.8237967491149902, "step": 1922 }, { "epoch": 0.15159637366968862, "grad_norm": 1.1517891883850098, "learning_rate": 0.0005, "loss": 2.792011260986328, "step": 1923 }, { "epoch": 0.15167520693732756, "grad_norm": 1.1473197937011719, "learning_rate": 0.0005, "loss": 3.153195858001709, "step": 1924 }, { "epoch": 0.1517540402049665, "grad_norm": 1.1731210947036743, "learning_rate": 0.0005, "loss": 2.8697898387908936, "step": 1925 }, { "epoch": 0.15183287347260543, "grad_norm": 1.2290979623794556, "learning_rate": 0.0005, "loss": 2.7968103885650635, "step": 1926 }, { "epoch": 0.15191170674024437, "grad_norm": 1.0380862951278687, "learning_rate": 0.0005, "loss": 2.9118456840515137, "step": 1927 }, { "epoch": 0.15199054000788334, "grad_norm": 0.9132802486419678, "learning_rate": 0.0005, "loss": 2.6506638526916504, "step": 1928 }, { "epoch": 0.15206937327552228, "grad_norm": 1.5457085371017456, "learning_rate": 0.0005, "loss": 2.5933547019958496, "step": 1929 }, { "epoch": 0.15214820654316122, "grad_norm": 1.1370807886123657, "learning_rate": 0.0005, "loss": 3.0153136253356934, "step": 1930 }, { "epoch": 0.15222703981080016, "grad_norm": 0.9713404178619385, "learning_rate": 0.0005, "loss": 2.9401791095733643, "step": 1931 }, { "epoch": 0.1523058730784391, "grad_norm": 1.0671190023422241, "learning_rate": 0.0005, "loss": 2.92684268951416, "step": 1932 }, { "epoch": 0.15238470634607804, "grad_norm": 0.9885390400886536, "learning_rate": 0.0005, "loss": 3.020698070526123, "step": 1933 }, { "epoch": 0.15246353961371698, "grad_norm": 1.2393583059310913, "learning_rate": 0.0005, "loss": 2.890908718109131, "step": 1934 }, { "epoch": 0.15254237288135594, "grad_norm": 2.6744959354400635, "learning_rate": 0.0005, "loss": 3.0123789310455322, "step": 1935 }, { "epoch": 0.15262120614899488, "grad_norm": 0.9289911389350891, "learning_rate": 0.0005, "loss": 2.7157387733459473, "step": 1936 }, { "epoch": 0.15270003941663382, "grad_norm": 0.9821601510047913, "learning_rate": 0.0005, "loss": 2.6476097106933594, "step": 1937 }, { "epoch": 0.15277887268427276, "grad_norm": 0.8981399536132812, "learning_rate": 0.0005, "loss": 2.62716007232666, "step": 1938 }, { "epoch": 0.1528577059519117, "grad_norm": 1.469909429550171, "learning_rate": 0.0005, "loss": 2.90806245803833, "step": 1939 }, { "epoch": 0.15293653921955064, "grad_norm": 0.9567077159881592, "learning_rate": 0.0005, "loss": 2.6687281131744385, "step": 1940 }, { "epoch": 0.1530153724871896, "grad_norm": 1.7303224802017212, "learning_rate": 0.0005, "loss": 3.0873804092407227, "step": 1941 }, { "epoch": 0.15309420575482854, "grad_norm": 1.0349880456924438, "learning_rate": 0.0005, "loss": 2.543943405151367, "step": 1942 }, { "epoch": 0.15317303902246748, "grad_norm": 0.860159158706665, "learning_rate": 0.0005, "loss": 2.911067485809326, "step": 1943 }, { "epoch": 0.15325187229010642, "grad_norm": 1.0072039365768433, "learning_rate": 0.0005, "loss": 2.680786371231079, "step": 1944 }, { "epoch": 0.15333070555774536, "grad_norm": 1.6490061283111572, "learning_rate": 0.0005, "loss": 2.6198842525482178, "step": 1945 }, { "epoch": 0.1534095388253843, "grad_norm": 1.0367234945297241, "learning_rate": 0.0005, "loss": 2.7702441215515137, "step": 1946 }, { "epoch": 0.15348837209302327, "grad_norm": 1.4686229228973389, "learning_rate": 0.0005, "loss": 2.5608065128326416, "step": 1947 }, { "epoch": 0.1535672053606622, "grad_norm": 0.9626668691635132, "learning_rate": 0.0005, "loss": 3.1133813858032227, "step": 1948 }, { "epoch": 0.15364603862830115, "grad_norm": 1.3032653331756592, "learning_rate": 0.0005, "loss": 3.0485780239105225, "step": 1949 }, { "epoch": 0.15372487189594009, "grad_norm": 1.3493757247924805, "learning_rate": 0.0005, "loss": 2.8344249725341797, "step": 1950 }, { "epoch": 0.15380370516357902, "grad_norm": 1.0145372152328491, "learning_rate": 0.0005, "loss": 3.0096068382263184, "step": 1951 }, { "epoch": 0.15388253843121796, "grad_norm": 0.9150933027267456, "learning_rate": 0.0005, "loss": 2.5320053100585938, "step": 1952 }, { "epoch": 0.15396137169885693, "grad_norm": 0.958824634552002, "learning_rate": 0.0005, "loss": 2.7361061573028564, "step": 1953 }, { "epoch": 0.15404020496649587, "grad_norm": 1.1062673330307007, "learning_rate": 0.0005, "loss": 3.3397603034973145, "step": 1954 }, { "epoch": 0.1541190382341348, "grad_norm": 1.0894677639007568, "learning_rate": 0.0005, "loss": 3.037919044494629, "step": 1955 }, { "epoch": 0.15419787150177375, "grad_norm": 1.037037968635559, "learning_rate": 0.0005, "loss": 3.2491300106048584, "step": 1956 }, { "epoch": 0.1542767047694127, "grad_norm": 1.0586233139038086, "learning_rate": 0.0005, "loss": 2.6136393547058105, "step": 1957 }, { "epoch": 0.15435553803705163, "grad_norm": 0.9164761304855347, "learning_rate": 0.0005, "loss": 2.526604413986206, "step": 1958 }, { "epoch": 0.1544343713046906, "grad_norm": 1.0770986080169678, "learning_rate": 0.0005, "loss": 2.821202278137207, "step": 1959 }, { "epoch": 0.15451320457232953, "grad_norm": 1.200202465057373, "learning_rate": 0.0005, "loss": 2.657809257507324, "step": 1960 }, { "epoch": 0.15459203783996847, "grad_norm": 1.0521618127822876, "learning_rate": 0.0005, "loss": 2.7607264518737793, "step": 1961 }, { "epoch": 0.1546708711076074, "grad_norm": 1.0247180461883545, "learning_rate": 0.0005, "loss": 3.0111477375030518, "step": 1962 }, { "epoch": 0.15474970437524635, "grad_norm": 1.0663334131240845, "learning_rate": 0.0005, "loss": 3.3221755027770996, "step": 1963 }, { "epoch": 0.1548285376428853, "grad_norm": 0.9940852522850037, "learning_rate": 0.0005, "loss": 2.619853973388672, "step": 1964 }, { "epoch": 0.15490737091052423, "grad_norm": 0.9073441624641418, "learning_rate": 0.0005, "loss": 2.1965489387512207, "step": 1965 }, { "epoch": 0.1549862041781632, "grad_norm": 1.6671812534332275, "learning_rate": 0.0005, "loss": 3.1003897190093994, "step": 1966 }, { "epoch": 0.15506503744580213, "grad_norm": 1.6504164934158325, "learning_rate": 0.0005, "loss": 2.648120880126953, "step": 1967 }, { "epoch": 0.15514387071344107, "grad_norm": 0.9616149067878723, "learning_rate": 0.0005, "loss": 2.674172878265381, "step": 1968 }, { "epoch": 0.15522270398108, "grad_norm": 0.8869598507881165, "learning_rate": 0.0005, "loss": 2.692833185195923, "step": 1969 }, { "epoch": 0.15530153724871895, "grad_norm": 1.0084846019744873, "learning_rate": 0.0005, "loss": 2.537177801132202, "step": 1970 }, { "epoch": 0.1553803705163579, "grad_norm": 1.0423877239227295, "learning_rate": 0.0005, "loss": 2.5538668632507324, "step": 1971 }, { "epoch": 0.15545920378399686, "grad_norm": 1.005066990852356, "learning_rate": 0.0005, "loss": 2.9283292293548584, "step": 1972 }, { "epoch": 0.1555380370516358, "grad_norm": 1.1554697751998901, "learning_rate": 0.0005, "loss": 2.8480443954467773, "step": 1973 }, { "epoch": 0.15561687031927474, "grad_norm": 1.6214014291763306, "learning_rate": 0.0005, "loss": 2.9666361808776855, "step": 1974 }, { "epoch": 0.15569570358691368, "grad_norm": 1.665138602256775, "learning_rate": 0.0005, "loss": 2.925293445587158, "step": 1975 }, { "epoch": 0.15577453685455261, "grad_norm": 1.1264724731445312, "learning_rate": 0.0005, "loss": 2.989274024963379, "step": 1976 }, { "epoch": 0.15585337012219155, "grad_norm": 1.164776086807251, "learning_rate": 0.0005, "loss": 2.841794967651367, "step": 1977 }, { "epoch": 0.15593220338983052, "grad_norm": 0.8805086612701416, "learning_rate": 0.0005, "loss": 2.8207459449768066, "step": 1978 }, { "epoch": 0.15601103665746946, "grad_norm": 1.3013131618499756, "learning_rate": 0.0005, "loss": 2.901245355606079, "step": 1979 }, { "epoch": 0.1560898699251084, "grad_norm": 3.9883437156677246, "learning_rate": 0.0005, "loss": 2.8701565265655518, "step": 1980 }, { "epoch": 0.15616870319274734, "grad_norm": 1.1572976112365723, "learning_rate": 0.0005, "loss": 2.7956583499908447, "step": 1981 }, { "epoch": 0.15624753646038628, "grad_norm": 0.8979794383049011, "learning_rate": 0.0005, "loss": 2.903198480606079, "step": 1982 }, { "epoch": 0.15632636972802522, "grad_norm": 0.8368280529975891, "learning_rate": 0.0005, "loss": 2.4568099975585938, "step": 1983 }, { "epoch": 0.15640520299566418, "grad_norm": 1.05056893825531, "learning_rate": 0.0005, "loss": 2.613391399383545, "step": 1984 }, { "epoch": 0.15648403626330312, "grad_norm": 1.1779695749282837, "learning_rate": 0.0005, "loss": 3.3740100860595703, "step": 1985 }, { "epoch": 0.15656286953094206, "grad_norm": 1.093465805053711, "learning_rate": 0.0005, "loss": 2.607398509979248, "step": 1986 }, { "epoch": 0.156641702798581, "grad_norm": 1.2067581415176392, "learning_rate": 0.0005, "loss": 2.483138084411621, "step": 1987 }, { "epoch": 0.15672053606621994, "grad_norm": 0.9579700827598572, "learning_rate": 0.0005, "loss": 2.523282527923584, "step": 1988 }, { "epoch": 0.15679936933385888, "grad_norm": 1.3506708145141602, "learning_rate": 0.0005, "loss": 2.975785255432129, "step": 1989 }, { "epoch": 0.15687820260149785, "grad_norm": 0.9090879559516907, "learning_rate": 0.0005, "loss": 2.5605835914611816, "step": 1990 }, { "epoch": 0.15695703586913679, "grad_norm": 0.9083728194236755, "learning_rate": 0.0005, "loss": 2.8452072143554688, "step": 1991 }, { "epoch": 0.15703586913677572, "grad_norm": 1.0380617380142212, "learning_rate": 0.0005, "loss": 3.0909337997436523, "step": 1992 }, { "epoch": 0.15711470240441466, "grad_norm": 1.6451293230056763, "learning_rate": 0.0005, "loss": 2.6707663536071777, "step": 1993 }, { "epoch": 0.1571935356720536, "grad_norm": 1.0672826766967773, "learning_rate": 0.0005, "loss": 2.9725277423858643, "step": 1994 }, { "epoch": 0.15727236893969254, "grad_norm": 1.3396530151367188, "learning_rate": 0.0005, "loss": 2.6708760261535645, "step": 1995 }, { "epoch": 0.15735120220733148, "grad_norm": 1.1074594259262085, "learning_rate": 0.0005, "loss": 2.8342318534851074, "step": 1996 }, { "epoch": 0.15743003547497045, "grad_norm": 1.0151883363723755, "learning_rate": 0.0005, "loss": 2.4103569984436035, "step": 1997 }, { "epoch": 0.1575088687426094, "grad_norm": 1.788303256034851, "learning_rate": 0.0005, "loss": 2.9571354389190674, "step": 1998 }, { "epoch": 0.15758770201024833, "grad_norm": 1.2887996435165405, "learning_rate": 0.0005, "loss": 3.100512742996216, "step": 1999 }, { "epoch": 0.15766653527788727, "grad_norm": 0.9568374156951904, "learning_rate": 0.0005, "loss": 2.589623212814331, "step": 2000 }, { "epoch": 0.1577453685455262, "grad_norm": 1.0715333223342896, "learning_rate": 0.0005, "loss": 2.725552558898926, "step": 2001 }, { "epoch": 0.15782420181316514, "grad_norm": 0.95953369140625, "learning_rate": 0.0005, "loss": 2.5147812366485596, "step": 2002 }, { "epoch": 0.1579030350808041, "grad_norm": 1.0139397382736206, "learning_rate": 0.0005, "loss": 2.851327419281006, "step": 2003 }, { "epoch": 0.15798186834844305, "grad_norm": 1.0121384859085083, "learning_rate": 0.0005, "loss": 2.6558337211608887, "step": 2004 }, { "epoch": 0.158060701616082, "grad_norm": 1.2462832927703857, "learning_rate": 0.0005, "loss": 2.995410919189453, "step": 2005 }, { "epoch": 0.15813953488372093, "grad_norm": 0.9046538472175598, "learning_rate": 0.0005, "loss": 2.6819162368774414, "step": 2006 }, { "epoch": 0.15821836815135987, "grad_norm": 0.8238620758056641, "learning_rate": 0.0005, "loss": 2.976708173751831, "step": 2007 }, { "epoch": 0.1582972014189988, "grad_norm": 1.0077149868011475, "learning_rate": 0.0005, "loss": 2.865565299987793, "step": 2008 }, { "epoch": 0.15837603468663777, "grad_norm": 0.9489036202430725, "learning_rate": 0.0005, "loss": 2.8191781044006348, "step": 2009 }, { "epoch": 0.1584548679542767, "grad_norm": 0.8122431039810181, "learning_rate": 0.0005, "loss": 3.1171250343322754, "step": 2010 }, { "epoch": 0.15853370122191565, "grad_norm": 1.0350382328033447, "learning_rate": 0.0005, "loss": 2.9042110443115234, "step": 2011 }, { "epoch": 0.1586125344895546, "grad_norm": 1.2171852588653564, "learning_rate": 0.0005, "loss": 2.64117693901062, "step": 2012 }, { "epoch": 0.15869136775719353, "grad_norm": 0.9880216121673584, "learning_rate": 0.0005, "loss": 2.6849145889282227, "step": 2013 }, { "epoch": 0.15877020102483247, "grad_norm": 1.0774130821228027, "learning_rate": 0.0005, "loss": 2.8853750228881836, "step": 2014 }, { "epoch": 0.15884903429247144, "grad_norm": 1.232414960861206, "learning_rate": 0.0005, "loss": 2.718222141265869, "step": 2015 }, { "epoch": 0.15892786756011038, "grad_norm": 1.0338886976242065, "learning_rate": 0.0005, "loss": 3.0395054817199707, "step": 2016 }, { "epoch": 0.15900670082774931, "grad_norm": 1.4107160568237305, "learning_rate": 0.0005, "loss": 2.853005886077881, "step": 2017 }, { "epoch": 0.15908553409538825, "grad_norm": 1.1714507341384888, "learning_rate": 0.0005, "loss": 2.6053054332733154, "step": 2018 }, { "epoch": 0.1591643673630272, "grad_norm": 1.2619129419326782, "learning_rate": 0.0005, "loss": 2.6861331462860107, "step": 2019 }, { "epoch": 0.15924320063066613, "grad_norm": 1.2066408395767212, "learning_rate": 0.0005, "loss": 2.9651410579681396, "step": 2020 }, { "epoch": 0.15932203389830507, "grad_norm": 0.8796917796134949, "learning_rate": 0.0005, "loss": 2.905446767807007, "step": 2021 }, { "epoch": 0.15940086716594404, "grad_norm": 1.3625918626785278, "learning_rate": 0.0005, "loss": 2.5711517333984375, "step": 2022 }, { "epoch": 0.15947970043358298, "grad_norm": 1.0847899913787842, "learning_rate": 0.0005, "loss": 2.6536192893981934, "step": 2023 }, { "epoch": 0.15955853370122192, "grad_norm": 1.051171064376831, "learning_rate": 0.0005, "loss": 2.726438522338867, "step": 2024 }, { "epoch": 0.15963736696886086, "grad_norm": 1.1931884288787842, "learning_rate": 0.0005, "loss": 2.8256521224975586, "step": 2025 }, { "epoch": 0.1597162002364998, "grad_norm": 1.5851237773895264, "learning_rate": 0.0005, "loss": 2.2878360748291016, "step": 2026 }, { "epoch": 0.15979503350413873, "grad_norm": 0.9841295480728149, "learning_rate": 0.0005, "loss": 2.6369426250457764, "step": 2027 }, { "epoch": 0.1598738667717777, "grad_norm": 1.403476595878601, "learning_rate": 0.0005, "loss": 2.6861586570739746, "step": 2028 }, { "epoch": 0.15995270003941664, "grad_norm": 1.196219563484192, "learning_rate": 0.0005, "loss": 2.640657901763916, "step": 2029 }, { "epoch": 0.16003153330705558, "grad_norm": 1.022247314453125, "learning_rate": 0.0005, "loss": 2.6174585819244385, "step": 2030 }, { "epoch": 0.16011036657469452, "grad_norm": 1.4922515153884888, "learning_rate": 0.0005, "loss": 2.746391773223877, "step": 2031 }, { "epoch": 0.16018919984233346, "grad_norm": 1.0342175960540771, "learning_rate": 0.0005, "loss": 2.857358455657959, "step": 2032 }, { "epoch": 0.1602680331099724, "grad_norm": 0.884006917476654, "learning_rate": 0.0005, "loss": 2.7354953289031982, "step": 2033 }, { "epoch": 0.16034686637761136, "grad_norm": 0.9807911515235901, "learning_rate": 0.0005, "loss": 2.448270559310913, "step": 2034 }, { "epoch": 0.1604256996452503, "grad_norm": 1.03309166431427, "learning_rate": 0.0005, "loss": 2.7978742122650146, "step": 2035 }, { "epoch": 0.16050453291288924, "grad_norm": 0.967799961566925, "learning_rate": 0.0005, "loss": 2.69962215423584, "step": 2036 }, { "epoch": 0.16058336618052818, "grad_norm": 0.8869207501411438, "learning_rate": 0.0005, "loss": 2.7159335613250732, "step": 2037 }, { "epoch": 0.16066219944816712, "grad_norm": 0.9579550623893738, "learning_rate": 0.0005, "loss": 2.5381181240081787, "step": 2038 }, { "epoch": 0.16074103271580606, "grad_norm": 1.2105013132095337, "learning_rate": 0.0005, "loss": 2.6738336086273193, "step": 2039 }, { "epoch": 0.16081986598344503, "grad_norm": 1.0545423030853271, "learning_rate": 0.0005, "loss": 2.902113914489746, "step": 2040 }, { "epoch": 0.16089869925108397, "grad_norm": 0.9734444618225098, "learning_rate": 0.0005, "loss": 2.6382675170898438, "step": 2041 }, { "epoch": 0.1609775325187229, "grad_norm": 1.0956159830093384, "learning_rate": 0.0005, "loss": 2.6075527667999268, "step": 2042 }, { "epoch": 0.16105636578636184, "grad_norm": 1.2065401077270508, "learning_rate": 0.0005, "loss": 3.147862672805786, "step": 2043 }, { "epoch": 0.16113519905400078, "grad_norm": 0.9169648885726929, "learning_rate": 0.0005, "loss": 2.5804684162139893, "step": 2044 }, { "epoch": 0.16121403232163972, "grad_norm": 1.2507288455963135, "learning_rate": 0.0005, "loss": 2.4981846809387207, "step": 2045 }, { "epoch": 0.1612928655892787, "grad_norm": 1.3303139209747314, "learning_rate": 0.0005, "loss": 2.4637229442596436, "step": 2046 }, { "epoch": 0.16137169885691763, "grad_norm": 1.1150578260421753, "learning_rate": 0.0005, "loss": 3.033125877380371, "step": 2047 }, { "epoch": 0.16145053212455657, "grad_norm": 1.0467885732650757, "learning_rate": 0.0005, "loss": 2.6634674072265625, "step": 2048 }, { "epoch": 0.1615293653921955, "grad_norm": 0.911425769329071, "learning_rate": 0.0005, "loss": 2.9548234939575195, "step": 2049 }, { "epoch": 0.16160819865983445, "grad_norm": 1.1539791822433472, "learning_rate": 0.0005, "loss": 2.487834930419922, "step": 2050 }, { "epoch": 0.16168703192747338, "grad_norm": 0.9254512786865234, "learning_rate": 0.0005, "loss": 2.1546790599823, "step": 2051 }, { "epoch": 0.16176586519511232, "grad_norm": 1.5586209297180176, "learning_rate": 0.0005, "loss": 2.7790279388427734, "step": 2052 }, { "epoch": 0.1618446984627513, "grad_norm": 0.8620615005493164, "learning_rate": 0.0005, "loss": 2.813232421875, "step": 2053 }, { "epoch": 0.16192353173039023, "grad_norm": 0.9464796185493469, "learning_rate": 0.0005, "loss": 2.828792095184326, "step": 2054 }, { "epoch": 0.16200236499802917, "grad_norm": 1.0596349239349365, "learning_rate": 0.0005, "loss": 3.0034537315368652, "step": 2055 }, { "epoch": 0.1620811982656681, "grad_norm": 1.1260932683944702, "learning_rate": 0.0005, "loss": 2.9278311729431152, "step": 2056 }, { "epoch": 0.16216003153330705, "grad_norm": 1.2913014888763428, "learning_rate": 0.0005, "loss": 2.74442720413208, "step": 2057 }, { "epoch": 0.162238864800946, "grad_norm": 1.2415558099746704, "learning_rate": 0.0005, "loss": 2.5830774307250977, "step": 2058 }, { "epoch": 0.16231769806858495, "grad_norm": 0.9396257400512695, "learning_rate": 0.0005, "loss": 2.6751725673675537, "step": 2059 }, { "epoch": 0.1623965313362239, "grad_norm": 1.2141529321670532, "learning_rate": 0.0005, "loss": 3.080761432647705, "step": 2060 }, { "epoch": 0.16247536460386283, "grad_norm": 0.9284528493881226, "learning_rate": 0.0005, "loss": 3.1003875732421875, "step": 2061 }, { "epoch": 0.16255419787150177, "grad_norm": 1.0717732906341553, "learning_rate": 0.0005, "loss": 2.6913859844207764, "step": 2062 }, { "epoch": 0.1626330311391407, "grad_norm": 1.0120019912719727, "learning_rate": 0.0005, "loss": 2.751129150390625, "step": 2063 }, { "epoch": 0.16271186440677965, "grad_norm": 1.0258303880691528, "learning_rate": 0.0005, "loss": 2.736124277114868, "step": 2064 }, { "epoch": 0.16279069767441862, "grad_norm": 0.880850076675415, "learning_rate": 0.0005, "loss": 2.538116455078125, "step": 2065 }, { "epoch": 0.16286953094205756, "grad_norm": 1.5474820137023926, "learning_rate": 0.0005, "loss": 2.824401378631592, "step": 2066 }, { "epoch": 0.1629483642096965, "grad_norm": 1.3424768447875977, "learning_rate": 0.0005, "loss": 2.786210536956787, "step": 2067 }, { "epoch": 0.16302719747733543, "grad_norm": 1.0568357706069946, "learning_rate": 0.0005, "loss": 2.734684467315674, "step": 2068 }, { "epoch": 0.16310603074497437, "grad_norm": 1.316587209701538, "learning_rate": 0.0005, "loss": 2.864966869354248, "step": 2069 }, { "epoch": 0.1631848640126133, "grad_norm": 9.010010719299316, "learning_rate": 0.0005, "loss": 2.92759370803833, "step": 2070 }, { "epoch": 0.16326369728025228, "grad_norm": 1.0114662647247314, "learning_rate": 0.0005, "loss": 2.878004550933838, "step": 2071 }, { "epoch": 0.16334253054789122, "grad_norm": 0.9735981225967407, "learning_rate": 0.0005, "loss": 3.090169906616211, "step": 2072 }, { "epoch": 0.16342136381553016, "grad_norm": 1.1945501565933228, "learning_rate": 0.0005, "loss": 3.0483884811401367, "step": 2073 }, { "epoch": 0.1635001970831691, "grad_norm": 1.1976368427276611, "learning_rate": 0.0005, "loss": 2.474520683288574, "step": 2074 }, { "epoch": 0.16357903035080804, "grad_norm": 1.1040706634521484, "learning_rate": 0.0005, "loss": 2.7516632080078125, "step": 2075 }, { "epoch": 0.16365786361844697, "grad_norm": 1.1342493295669556, "learning_rate": 0.0005, "loss": 2.6257882118225098, "step": 2076 }, { "epoch": 0.16373669688608594, "grad_norm": 1.3744051456451416, "learning_rate": 0.0005, "loss": 2.9414665699005127, "step": 2077 }, { "epoch": 0.16381553015372488, "grad_norm": 1.174855351448059, "learning_rate": 0.0005, "loss": 2.6368119716644287, "step": 2078 }, { "epoch": 0.16389436342136382, "grad_norm": 1.023891806602478, "learning_rate": 0.0005, "loss": 2.900292158126831, "step": 2079 }, { "epoch": 0.16397319668900276, "grad_norm": 0.9634387493133545, "learning_rate": 0.0005, "loss": 2.6586217880249023, "step": 2080 }, { "epoch": 0.1640520299566417, "grad_norm": 1.166668176651001, "learning_rate": 0.0005, "loss": 2.951798439025879, "step": 2081 }, { "epoch": 0.16413086322428064, "grad_norm": 1.0673521757125854, "learning_rate": 0.0005, "loss": 2.90250301361084, "step": 2082 }, { "epoch": 0.16420969649191958, "grad_norm": 1.0847193002700806, "learning_rate": 0.0005, "loss": 2.8759384155273438, "step": 2083 }, { "epoch": 0.16428852975955854, "grad_norm": 1.0852495431900024, "learning_rate": 0.0005, "loss": 2.830444812774658, "step": 2084 }, { "epoch": 0.16436736302719748, "grad_norm": 1.0328584909439087, "learning_rate": 0.0005, "loss": 2.631713390350342, "step": 2085 }, { "epoch": 0.16444619629483642, "grad_norm": 1.1995272636413574, "learning_rate": 0.0005, "loss": 2.675785541534424, "step": 2086 }, { "epoch": 0.16452502956247536, "grad_norm": 1.4448999166488647, "learning_rate": 0.0005, "loss": 2.82411527633667, "step": 2087 }, { "epoch": 0.1646038628301143, "grad_norm": 1.0737388134002686, "learning_rate": 0.0005, "loss": 2.7733731269836426, "step": 2088 }, { "epoch": 0.16468269609775324, "grad_norm": 0.9488524198532104, "learning_rate": 0.0005, "loss": 2.642409324645996, "step": 2089 }, { "epoch": 0.1647615293653922, "grad_norm": 0.8331537842750549, "learning_rate": 0.0005, "loss": 2.735848903656006, "step": 2090 }, { "epoch": 0.16484036263303115, "grad_norm": 1.155297875404358, "learning_rate": 0.0005, "loss": 2.7830405235290527, "step": 2091 }, { "epoch": 0.16491919590067008, "grad_norm": 1.5412312746047974, "learning_rate": 0.0005, "loss": 2.889925479888916, "step": 2092 }, { "epoch": 0.16499802916830902, "grad_norm": 1.1030689477920532, "learning_rate": 0.0005, "loss": 2.544133424758911, "step": 2093 }, { "epoch": 0.16507686243594796, "grad_norm": 1.1668310165405273, "learning_rate": 0.0005, "loss": 2.605898857116699, "step": 2094 }, { "epoch": 0.1651556957035869, "grad_norm": 1.1418342590332031, "learning_rate": 0.0005, "loss": 2.703524112701416, "step": 2095 }, { "epoch": 0.16523452897122587, "grad_norm": 1.0241115093231201, "learning_rate": 0.0005, "loss": 2.939786434173584, "step": 2096 }, { "epoch": 0.1653133622388648, "grad_norm": 1.0179829597473145, "learning_rate": 0.0005, "loss": 2.7749524116516113, "step": 2097 }, { "epoch": 0.16539219550650375, "grad_norm": 0.8834307789802551, "learning_rate": 0.0005, "loss": 2.677675724029541, "step": 2098 }, { "epoch": 0.1654710287741427, "grad_norm": 0.9853507876396179, "learning_rate": 0.0005, "loss": 2.9158434867858887, "step": 2099 }, { "epoch": 0.16554986204178163, "grad_norm": 1.0389362573623657, "learning_rate": 0.0005, "loss": 2.8475968837738037, "step": 2100 }, { "epoch": 0.16562869530942056, "grad_norm": 1.183191180229187, "learning_rate": 0.0005, "loss": 2.837449550628662, "step": 2101 }, { "epoch": 0.16570752857705953, "grad_norm": 0.9668146967887878, "learning_rate": 0.0005, "loss": 2.976856231689453, "step": 2102 }, { "epoch": 0.16578636184469847, "grad_norm": 0.9919954538345337, "learning_rate": 0.0005, "loss": 2.8839826583862305, "step": 2103 }, { "epoch": 0.1658651951123374, "grad_norm": 1.020218849182129, "learning_rate": 0.0005, "loss": 2.613837242126465, "step": 2104 }, { "epoch": 0.16594402837997635, "grad_norm": 1.1288928985595703, "learning_rate": 0.0005, "loss": 2.5271682739257812, "step": 2105 }, { "epoch": 0.1660228616476153, "grad_norm": 1.1666706800460815, "learning_rate": 0.0005, "loss": 2.6664860248565674, "step": 2106 }, { "epoch": 0.16610169491525423, "grad_norm": 1.1567926406860352, "learning_rate": 0.0005, "loss": 2.7410240173339844, "step": 2107 }, { "epoch": 0.1661805281828932, "grad_norm": 1.0718361139297485, "learning_rate": 0.0005, "loss": 2.3900763988494873, "step": 2108 }, { "epoch": 0.16625936145053213, "grad_norm": 0.9739313721656799, "learning_rate": 0.0005, "loss": 2.8922321796417236, "step": 2109 }, { "epoch": 0.16633819471817107, "grad_norm": 1.1456031799316406, "learning_rate": 0.0005, "loss": 2.9475340843200684, "step": 2110 }, { "epoch": 0.16641702798581, "grad_norm": 1.1818472146987915, "learning_rate": 0.0005, "loss": 2.752695083618164, "step": 2111 }, { "epoch": 0.16649586125344895, "grad_norm": 1.385105848312378, "learning_rate": 0.0005, "loss": 2.7043635845184326, "step": 2112 }, { "epoch": 0.1665746945210879, "grad_norm": 0.9873515367507935, "learning_rate": 0.0005, "loss": 2.6417040824890137, "step": 2113 }, { "epoch": 0.16665352778872683, "grad_norm": 1.0481597185134888, "learning_rate": 0.0005, "loss": 2.5245213508605957, "step": 2114 }, { "epoch": 0.1667323610563658, "grad_norm": 1.2399451732635498, "learning_rate": 0.0005, "loss": 2.849048137664795, "step": 2115 }, { "epoch": 0.16681119432400474, "grad_norm": 0.9289021492004395, "learning_rate": 0.0005, "loss": 2.708198308944702, "step": 2116 }, { "epoch": 0.16689002759164367, "grad_norm": 1.1291754245758057, "learning_rate": 0.0005, "loss": 2.7837424278259277, "step": 2117 }, { "epoch": 0.16696886085928261, "grad_norm": 1.4666249752044678, "learning_rate": 0.0005, "loss": 3.003342390060425, "step": 2118 }, { "epoch": 0.16704769412692155, "grad_norm": 1.0390948057174683, "learning_rate": 0.0005, "loss": 2.8862054347991943, "step": 2119 }, { "epoch": 0.1671265273945605, "grad_norm": 0.9131105542182922, "learning_rate": 0.0005, "loss": 2.7250757217407227, "step": 2120 }, { "epoch": 0.16720536066219946, "grad_norm": 1.5832585096359253, "learning_rate": 0.0005, "loss": 2.8831005096435547, "step": 2121 }, { "epoch": 0.1672841939298384, "grad_norm": 1.1819777488708496, "learning_rate": 0.0005, "loss": 2.9435346126556396, "step": 2122 }, { "epoch": 0.16736302719747734, "grad_norm": 1.070996642112732, "learning_rate": 0.0005, "loss": 2.694852828979492, "step": 2123 }, { "epoch": 0.16744186046511628, "grad_norm": 1.1053335666656494, "learning_rate": 0.0005, "loss": 2.9864563941955566, "step": 2124 }, { "epoch": 0.16752069373275522, "grad_norm": 0.8737610578536987, "learning_rate": 0.0005, "loss": 2.717413902282715, "step": 2125 }, { "epoch": 0.16759952700039416, "grad_norm": 1.1106027364730835, "learning_rate": 0.0005, "loss": 2.8424692153930664, "step": 2126 }, { "epoch": 0.16767836026803312, "grad_norm": 0.9902650117874146, "learning_rate": 0.0005, "loss": 2.220917224884033, "step": 2127 }, { "epoch": 0.16775719353567206, "grad_norm": 0.9174429178237915, "learning_rate": 0.0005, "loss": 2.3716092109680176, "step": 2128 }, { "epoch": 0.167836026803311, "grad_norm": 1.421286702156067, "learning_rate": 0.0005, "loss": 2.4968698024749756, "step": 2129 }, { "epoch": 0.16791486007094994, "grad_norm": 1.182625412940979, "learning_rate": 0.0005, "loss": 2.5979511737823486, "step": 2130 }, { "epoch": 0.16799369333858888, "grad_norm": 0.9224263429641724, "learning_rate": 0.0005, "loss": 2.910146713256836, "step": 2131 }, { "epoch": 0.16807252660622782, "grad_norm": 1.0948731899261475, "learning_rate": 0.0005, "loss": 2.9595136642456055, "step": 2132 }, { "epoch": 0.16815135987386678, "grad_norm": 1.0281685590744019, "learning_rate": 0.0005, "loss": 2.57216739654541, "step": 2133 }, { "epoch": 0.16823019314150572, "grad_norm": 0.9943702816963196, "learning_rate": 0.0005, "loss": 2.827889919281006, "step": 2134 }, { "epoch": 0.16830902640914466, "grad_norm": 1.0157288312911987, "learning_rate": 0.0005, "loss": 2.898573398590088, "step": 2135 }, { "epoch": 0.1683878596767836, "grad_norm": 1.2833975553512573, "learning_rate": 0.0005, "loss": 2.6549367904663086, "step": 2136 }, { "epoch": 0.16846669294442254, "grad_norm": 1.0089726448059082, "learning_rate": 0.0005, "loss": 2.8488266468048096, "step": 2137 }, { "epoch": 0.16854552621206148, "grad_norm": 0.9695906639099121, "learning_rate": 0.0005, "loss": 2.8712780475616455, "step": 2138 }, { "epoch": 0.16862435947970042, "grad_norm": 1.4315212965011597, "learning_rate": 0.0005, "loss": 2.7474756240844727, "step": 2139 }, { "epoch": 0.1687031927473394, "grad_norm": 1.2359715700149536, "learning_rate": 0.0005, "loss": 2.8219776153564453, "step": 2140 }, { "epoch": 0.16878202601497833, "grad_norm": 1.1518372297286987, "learning_rate": 0.0005, "loss": 2.7560906410217285, "step": 2141 }, { "epoch": 0.16886085928261726, "grad_norm": 1.0841795206069946, "learning_rate": 0.0005, "loss": 2.0483903884887695, "step": 2142 }, { "epoch": 0.1689396925502562, "grad_norm": 1.2756993770599365, "learning_rate": 0.0005, "loss": 2.9779653549194336, "step": 2143 }, { "epoch": 0.16901852581789514, "grad_norm": 1.042048692703247, "learning_rate": 0.0005, "loss": 3.096444606781006, "step": 2144 }, { "epoch": 0.16909735908553408, "grad_norm": 1.3100324869155884, "learning_rate": 0.0005, "loss": 2.531996488571167, "step": 2145 }, { "epoch": 0.16917619235317305, "grad_norm": 1.1481261253356934, "learning_rate": 0.0005, "loss": 2.8010287284851074, "step": 2146 }, { "epoch": 0.169255025620812, "grad_norm": 1.1224435567855835, "learning_rate": 0.0005, "loss": 2.661220073699951, "step": 2147 }, { "epoch": 0.16933385888845093, "grad_norm": 1.1289689540863037, "learning_rate": 0.0005, "loss": 2.8149521350860596, "step": 2148 }, { "epoch": 0.16941269215608987, "grad_norm": 1.001178503036499, "learning_rate": 0.0005, "loss": 2.7252678871154785, "step": 2149 }, { "epoch": 0.1694915254237288, "grad_norm": 1.058695912361145, "learning_rate": 0.0005, "loss": 2.7742996215820312, "step": 2150 }, { "epoch": 0.16957035869136775, "grad_norm": 0.8052046895027161, "learning_rate": 0.0005, "loss": 2.291797637939453, "step": 2151 }, { "epoch": 0.1696491919590067, "grad_norm": 1.5597820281982422, "learning_rate": 0.0005, "loss": 2.8596458435058594, "step": 2152 }, { "epoch": 0.16972802522664565, "grad_norm": 1.2381258010864258, "learning_rate": 0.0005, "loss": 2.7814815044403076, "step": 2153 }, { "epoch": 0.1698068584942846, "grad_norm": 1.1856638193130493, "learning_rate": 0.0005, "loss": 2.5522422790527344, "step": 2154 }, { "epoch": 0.16988569176192353, "grad_norm": 0.9183720350265503, "learning_rate": 0.0005, "loss": 2.532561779022217, "step": 2155 }, { "epoch": 0.16996452502956247, "grad_norm": 1.1384086608886719, "learning_rate": 0.0005, "loss": 2.980926275253296, "step": 2156 }, { "epoch": 0.1700433582972014, "grad_norm": 1.5329152345657349, "learning_rate": 0.0005, "loss": 2.648416042327881, "step": 2157 }, { "epoch": 0.17012219156484037, "grad_norm": 0.8703145384788513, "learning_rate": 0.0005, "loss": 2.4525821208953857, "step": 2158 }, { "epoch": 0.17020102483247931, "grad_norm": 1.0866143703460693, "learning_rate": 0.0005, "loss": 2.6916751861572266, "step": 2159 }, { "epoch": 0.17027985810011825, "grad_norm": 1.3800897598266602, "learning_rate": 0.0005, "loss": 2.7892472743988037, "step": 2160 }, { "epoch": 0.1703586913677572, "grad_norm": 1.1050785779953003, "learning_rate": 0.0005, "loss": 2.853703022003174, "step": 2161 }, { "epoch": 0.17043752463539613, "grad_norm": 1.8649420738220215, "learning_rate": 0.0005, "loss": 2.994523048400879, "step": 2162 }, { "epoch": 0.17051635790303507, "grad_norm": 1.3728764057159424, "learning_rate": 0.0005, "loss": 3.028383731842041, "step": 2163 }, { "epoch": 0.17059519117067404, "grad_norm": 0.9231173396110535, "learning_rate": 0.0005, "loss": 2.7812209129333496, "step": 2164 }, { "epoch": 0.17067402443831298, "grad_norm": 1.122595191001892, "learning_rate": 0.0005, "loss": 2.502720594406128, "step": 2165 }, { "epoch": 0.17075285770595192, "grad_norm": 0.9392154812812805, "learning_rate": 0.0005, "loss": 2.97501277923584, "step": 2166 }, { "epoch": 0.17083169097359086, "grad_norm": 0.9636241793632507, "learning_rate": 0.0005, "loss": 2.534853458404541, "step": 2167 }, { "epoch": 0.1709105242412298, "grad_norm": 0.9490337371826172, "learning_rate": 0.0005, "loss": 2.470984935760498, "step": 2168 }, { "epoch": 0.17098935750886873, "grad_norm": 1.2772328853607178, "learning_rate": 0.0005, "loss": 2.546374797821045, "step": 2169 }, { "epoch": 0.17106819077650767, "grad_norm": 0.9656487107276917, "learning_rate": 0.0005, "loss": 2.954056978225708, "step": 2170 }, { "epoch": 0.17114702404414664, "grad_norm": 1.129634976387024, "learning_rate": 0.0005, "loss": 2.778284788131714, "step": 2171 }, { "epoch": 0.17122585731178558, "grad_norm": 1.0085153579711914, "learning_rate": 0.0005, "loss": 2.9316792488098145, "step": 2172 }, { "epoch": 0.17130469057942452, "grad_norm": 0.9025070667266846, "learning_rate": 0.0005, "loss": 2.7156739234924316, "step": 2173 }, { "epoch": 0.17138352384706346, "grad_norm": 0.9513170123100281, "learning_rate": 0.0005, "loss": 2.8429954051971436, "step": 2174 }, { "epoch": 0.1714623571147024, "grad_norm": 0.9536675810813904, "learning_rate": 0.0005, "loss": 2.776298999786377, "step": 2175 }, { "epoch": 0.17154119038234134, "grad_norm": 0.9838851094245911, "learning_rate": 0.0005, "loss": 2.479642391204834, "step": 2176 }, { "epoch": 0.1716200236499803, "grad_norm": 0.9832547307014465, "learning_rate": 0.0005, "loss": 2.739445924758911, "step": 2177 }, { "epoch": 0.17169885691761924, "grad_norm": 1.2694542407989502, "learning_rate": 0.0005, "loss": 2.8276920318603516, "step": 2178 }, { "epoch": 0.17177769018525818, "grad_norm": 1.2899099588394165, "learning_rate": 0.0005, "loss": 2.930635929107666, "step": 2179 }, { "epoch": 0.17185652345289712, "grad_norm": 0.7706491947174072, "learning_rate": 0.0005, "loss": 2.6605653762817383, "step": 2180 }, { "epoch": 0.17193535672053606, "grad_norm": 0.9375767111778259, "learning_rate": 0.0005, "loss": 2.8424429893493652, "step": 2181 }, { "epoch": 0.172014189988175, "grad_norm": 0.9692498445510864, "learning_rate": 0.0005, "loss": 3.0056192874908447, "step": 2182 }, { "epoch": 0.17209302325581396, "grad_norm": 1.4116376638412476, "learning_rate": 0.0005, "loss": 2.481875419616699, "step": 2183 }, { "epoch": 0.1721718565234529, "grad_norm": 1.0194876194000244, "learning_rate": 0.0005, "loss": 2.6617255210876465, "step": 2184 }, { "epoch": 0.17225068979109184, "grad_norm": 1.0165692567825317, "learning_rate": 0.0005, "loss": 3.0344436168670654, "step": 2185 }, { "epoch": 0.17232952305873078, "grad_norm": 1.0901488065719604, "learning_rate": 0.0005, "loss": 2.5469114780426025, "step": 2186 }, { "epoch": 0.17240835632636972, "grad_norm": 1.177277684211731, "learning_rate": 0.0005, "loss": 3.027879238128662, "step": 2187 }, { "epoch": 0.17248718959400866, "grad_norm": 0.9910203814506531, "learning_rate": 0.0005, "loss": 2.727799415588379, "step": 2188 }, { "epoch": 0.17256602286164763, "grad_norm": 1.0817621946334839, "learning_rate": 0.0005, "loss": 2.937532663345337, "step": 2189 }, { "epoch": 0.17264485612928657, "grad_norm": 1.611358404159546, "learning_rate": 0.0005, "loss": 2.740140438079834, "step": 2190 }, { "epoch": 0.1727236893969255, "grad_norm": 1.25057053565979, "learning_rate": 0.0005, "loss": 2.620332717895508, "step": 2191 }, { "epoch": 0.17280252266456445, "grad_norm": 0.9981539845466614, "learning_rate": 0.0005, "loss": 2.720926284790039, "step": 2192 }, { "epoch": 0.17288135593220338, "grad_norm": 1.1200776100158691, "learning_rate": 0.0005, "loss": 2.6915931701660156, "step": 2193 }, { "epoch": 0.17296018919984232, "grad_norm": 0.8463955521583557, "learning_rate": 0.0005, "loss": 2.61159610748291, "step": 2194 }, { "epoch": 0.1730390224674813, "grad_norm": 1.0797219276428223, "learning_rate": 0.0005, "loss": 3.1108884811401367, "step": 2195 }, { "epoch": 0.17311785573512023, "grad_norm": 0.9111573696136475, "learning_rate": 0.0005, "loss": 2.681175708770752, "step": 2196 }, { "epoch": 0.17319668900275917, "grad_norm": 0.9423704147338867, "learning_rate": 0.0005, "loss": 2.4629063606262207, "step": 2197 }, { "epoch": 0.1732755222703981, "grad_norm": 1.0554940700531006, "learning_rate": 0.0005, "loss": 2.7966015338897705, "step": 2198 }, { "epoch": 0.17335435553803705, "grad_norm": 0.8506665229797363, "learning_rate": 0.0005, "loss": 2.824566125869751, "step": 2199 }, { "epoch": 0.173433188805676, "grad_norm": 0.9487960338592529, "learning_rate": 0.0005, "loss": 2.5419697761535645, "step": 2200 }, { "epoch": 0.17351202207331493, "grad_norm": 1.1141211986541748, "learning_rate": 0.0005, "loss": 2.6286566257476807, "step": 2201 }, { "epoch": 0.1735908553409539, "grad_norm": 1.5403207540512085, "learning_rate": 0.0005, "loss": 2.7580604553222656, "step": 2202 }, { "epoch": 0.17366968860859283, "grad_norm": 1.157921552658081, "learning_rate": 0.0005, "loss": 2.7827529907226562, "step": 2203 }, { "epoch": 0.17374852187623177, "grad_norm": 1.2181528806686401, "learning_rate": 0.0005, "loss": 2.5800678730010986, "step": 2204 }, { "epoch": 0.1738273551438707, "grad_norm": 1.1398768424987793, "learning_rate": 0.0005, "loss": 3.194403648376465, "step": 2205 }, { "epoch": 0.17390618841150965, "grad_norm": 1.3057548999786377, "learning_rate": 0.0005, "loss": 2.951561450958252, "step": 2206 }, { "epoch": 0.1739850216791486, "grad_norm": 1.1613138914108276, "learning_rate": 0.0005, "loss": 2.633543014526367, "step": 2207 }, { "epoch": 0.17406385494678756, "grad_norm": 0.9666242599487305, "learning_rate": 0.0005, "loss": 2.5576720237731934, "step": 2208 }, { "epoch": 0.1741426882144265, "grad_norm": 1.0567982196807861, "learning_rate": 0.0005, "loss": 2.9106006622314453, "step": 2209 }, { "epoch": 0.17422152148206543, "grad_norm": 1.0249488353729248, "learning_rate": 0.0005, "loss": 2.5211987495422363, "step": 2210 }, { "epoch": 0.17430035474970437, "grad_norm": 1.0034868717193604, "learning_rate": 0.0005, "loss": 2.513118267059326, "step": 2211 }, { "epoch": 0.1743791880173433, "grad_norm": 1.5137943029403687, "learning_rate": 0.0005, "loss": 2.4160966873168945, "step": 2212 }, { "epoch": 0.17445802128498225, "grad_norm": 1.1029103994369507, "learning_rate": 0.0005, "loss": 2.4415721893310547, "step": 2213 }, { "epoch": 0.17453685455262122, "grad_norm": 0.9732348322868347, "learning_rate": 0.0005, "loss": 2.4817042350769043, "step": 2214 }, { "epoch": 0.17461568782026016, "grad_norm": 0.9026470184326172, "learning_rate": 0.0005, "loss": 2.9038162231445312, "step": 2215 }, { "epoch": 0.1746945210878991, "grad_norm": 1.331178069114685, "learning_rate": 0.0005, "loss": 2.586454391479492, "step": 2216 }, { "epoch": 0.17477335435553804, "grad_norm": 0.9032671451568604, "learning_rate": 0.0005, "loss": 2.484626054763794, "step": 2217 }, { "epoch": 0.17485218762317697, "grad_norm": 0.9734191298484802, "learning_rate": 0.0005, "loss": 2.5993621349334717, "step": 2218 }, { "epoch": 0.1749310208908159, "grad_norm": 0.9473301768302917, "learning_rate": 0.0005, "loss": 2.4644834995269775, "step": 2219 }, { "epoch": 0.17500985415845488, "grad_norm": 1.012037754058838, "learning_rate": 0.0005, "loss": 2.8128857612609863, "step": 2220 }, { "epoch": 0.17508868742609382, "grad_norm": 0.94283527135849, "learning_rate": 0.0005, "loss": 2.490245819091797, "step": 2221 }, { "epoch": 0.17516752069373276, "grad_norm": 1.2132196426391602, "learning_rate": 0.0005, "loss": 2.77657413482666, "step": 2222 }, { "epoch": 0.1752463539613717, "grad_norm": 1.2090981006622314, "learning_rate": 0.0005, "loss": 2.6231322288513184, "step": 2223 }, { "epoch": 0.17532518722901064, "grad_norm": 0.9021545648574829, "learning_rate": 0.0005, "loss": 2.69598388671875, "step": 2224 }, { "epoch": 0.17540402049664958, "grad_norm": 1.4262193441390991, "learning_rate": 0.0005, "loss": 2.585536241531372, "step": 2225 }, { "epoch": 0.17548285376428854, "grad_norm": 0.9721992611885071, "learning_rate": 0.0005, "loss": 2.1814756393432617, "step": 2226 }, { "epoch": 0.17556168703192748, "grad_norm": 0.9204886555671692, "learning_rate": 0.0005, "loss": 2.7714171409606934, "step": 2227 }, { "epoch": 0.17564052029956642, "grad_norm": 0.8752006888389587, "learning_rate": 0.0005, "loss": 2.5954368114471436, "step": 2228 }, { "epoch": 0.17571935356720536, "grad_norm": 1.004604458808899, "learning_rate": 0.0005, "loss": 2.501354217529297, "step": 2229 }, { "epoch": 0.1757981868348443, "grad_norm": 1.0195658206939697, "learning_rate": 0.0005, "loss": 2.576824903488159, "step": 2230 }, { "epoch": 0.17587702010248324, "grad_norm": 1.0690702199935913, "learning_rate": 0.0005, "loss": 2.764946222305298, "step": 2231 }, { "epoch": 0.17595585337012218, "grad_norm": 1.0647914409637451, "learning_rate": 0.0005, "loss": 2.475717544555664, "step": 2232 }, { "epoch": 0.17603468663776115, "grad_norm": 0.9436178207397461, "learning_rate": 0.0005, "loss": 2.6007943153381348, "step": 2233 }, { "epoch": 0.17611351990540008, "grad_norm": 0.8472604155540466, "learning_rate": 0.0005, "loss": 2.4686977863311768, "step": 2234 }, { "epoch": 0.17619235317303902, "grad_norm": 1.1759989261627197, "learning_rate": 0.0005, "loss": 2.569249153137207, "step": 2235 }, { "epoch": 0.17627118644067796, "grad_norm": 0.7505370378494263, "learning_rate": 0.0005, "loss": 2.5499138832092285, "step": 2236 }, { "epoch": 0.1763500197083169, "grad_norm": 1.5058883428573608, "learning_rate": 0.0005, "loss": 2.77004337310791, "step": 2237 }, { "epoch": 0.17642885297595584, "grad_norm": 0.9770467281341553, "learning_rate": 0.0005, "loss": 2.6569485664367676, "step": 2238 }, { "epoch": 0.1765076862435948, "grad_norm": 1.085196852684021, "learning_rate": 0.0005, "loss": 2.723008155822754, "step": 2239 }, { "epoch": 0.17658651951123375, "grad_norm": 0.9204864501953125, "learning_rate": 0.0005, "loss": 2.597195625305176, "step": 2240 }, { "epoch": 0.1766653527788727, "grad_norm": 1.131448745727539, "learning_rate": 0.0005, "loss": 2.5564370155334473, "step": 2241 }, { "epoch": 0.17674418604651163, "grad_norm": 1.0880661010742188, "learning_rate": 0.0005, "loss": 2.6273629665374756, "step": 2242 }, { "epoch": 0.17682301931415056, "grad_norm": 0.9176331162452698, "learning_rate": 0.0005, "loss": 2.7017195224761963, "step": 2243 }, { "epoch": 0.1769018525817895, "grad_norm": 1.0852246284484863, "learning_rate": 0.0005, "loss": 2.7188773155212402, "step": 2244 }, { "epoch": 0.17698068584942847, "grad_norm": 0.9677958488464355, "learning_rate": 0.0005, "loss": 3.115588903427124, "step": 2245 }, { "epoch": 0.1770595191170674, "grad_norm": 1.0193485021591187, "learning_rate": 0.0005, "loss": 2.6867175102233887, "step": 2246 }, { "epoch": 0.17713835238470635, "grad_norm": 0.8993326425552368, "learning_rate": 0.0005, "loss": 2.4244699478149414, "step": 2247 }, { "epoch": 0.1772171856523453, "grad_norm": 1.0311264991760254, "learning_rate": 0.0005, "loss": 2.9280519485473633, "step": 2248 }, { "epoch": 0.17729601891998423, "grad_norm": 1.775219440460205, "learning_rate": 0.0005, "loss": 3.231581687927246, "step": 2249 }, { "epoch": 0.17737485218762317, "grad_norm": 1.193231225013733, "learning_rate": 0.0005, "loss": 2.9232254028320312, "step": 2250 }, { "epoch": 0.17745368545526213, "grad_norm": 0.9755420684814453, "learning_rate": 0.0005, "loss": 2.651425361633301, "step": 2251 }, { "epoch": 0.17753251872290107, "grad_norm": 1.3944953680038452, "learning_rate": 0.0005, "loss": 2.7530884742736816, "step": 2252 }, { "epoch": 0.17761135199054, "grad_norm": 0.8215785622596741, "learning_rate": 0.0005, "loss": 2.850027561187744, "step": 2253 }, { "epoch": 0.17769018525817895, "grad_norm": 1.0068646669387817, "learning_rate": 0.0005, "loss": 2.8326919078826904, "step": 2254 }, { "epoch": 0.1777690185258179, "grad_norm": 0.9431862235069275, "learning_rate": 0.0005, "loss": 2.7986230850219727, "step": 2255 }, { "epoch": 0.17784785179345683, "grad_norm": 1.0336828231811523, "learning_rate": 0.0005, "loss": 2.778313159942627, "step": 2256 }, { "epoch": 0.1779266850610958, "grad_norm": 1.0485634803771973, "learning_rate": 0.0005, "loss": 2.722578287124634, "step": 2257 }, { "epoch": 0.17800551832873474, "grad_norm": 1.342901349067688, "learning_rate": 0.0005, "loss": 2.6186389923095703, "step": 2258 }, { "epoch": 0.17808435159637367, "grad_norm": 1.0629478693008423, "learning_rate": 0.0005, "loss": 2.79730224609375, "step": 2259 }, { "epoch": 0.1781631848640126, "grad_norm": 1.0486423969268799, "learning_rate": 0.0005, "loss": 2.9378066062927246, "step": 2260 }, { "epoch": 0.17824201813165155, "grad_norm": 1.0620821714401245, "learning_rate": 0.0005, "loss": 2.7964155673980713, "step": 2261 }, { "epoch": 0.1783208513992905, "grad_norm": 1.0918375253677368, "learning_rate": 0.0005, "loss": 2.777792453765869, "step": 2262 }, { "epoch": 0.17839968466692943, "grad_norm": 1.0377421379089355, "learning_rate": 0.0005, "loss": 2.468188762664795, "step": 2263 }, { "epoch": 0.1784785179345684, "grad_norm": 1.184008002281189, "learning_rate": 0.0005, "loss": 3.000497579574585, "step": 2264 }, { "epoch": 0.17855735120220734, "grad_norm": 1.0151917934417725, "learning_rate": 0.0005, "loss": 2.592787265777588, "step": 2265 }, { "epoch": 0.17863618446984628, "grad_norm": 1.2360639572143555, "learning_rate": 0.0005, "loss": 2.7661261558532715, "step": 2266 }, { "epoch": 0.17871501773748522, "grad_norm": 0.923212468624115, "learning_rate": 0.0005, "loss": 2.3597958087921143, "step": 2267 }, { "epoch": 0.17879385100512415, "grad_norm": 1.2592793703079224, "learning_rate": 0.0005, "loss": 2.503544330596924, "step": 2268 }, { "epoch": 0.1788726842727631, "grad_norm": 1.2912300825119019, "learning_rate": 0.0005, "loss": 2.8502583503723145, "step": 2269 }, { "epoch": 0.17895151754040206, "grad_norm": 1.079107642173767, "learning_rate": 0.0005, "loss": 2.596538782119751, "step": 2270 }, { "epoch": 0.179030350808041, "grad_norm": 1.1609286069869995, "learning_rate": 0.0005, "loss": 3.408015251159668, "step": 2271 }, { "epoch": 0.17910918407567994, "grad_norm": 1.2759164571762085, "learning_rate": 0.0005, "loss": 3.2578797340393066, "step": 2272 }, { "epoch": 0.17918801734331888, "grad_norm": 0.7790801525115967, "learning_rate": 0.0005, "loss": 2.610291004180908, "step": 2273 }, { "epoch": 0.17926685061095782, "grad_norm": 1.445273518562317, "learning_rate": 0.0005, "loss": 2.7099945545196533, "step": 2274 }, { "epoch": 0.17934568387859676, "grad_norm": 0.942294180393219, "learning_rate": 0.0005, "loss": 2.927377223968506, "step": 2275 }, { "epoch": 0.17942451714623572, "grad_norm": 0.8519986867904663, "learning_rate": 0.0005, "loss": 2.765247344970703, "step": 2276 }, { "epoch": 0.17950335041387466, "grad_norm": 0.7947522401809692, "learning_rate": 0.0005, "loss": 2.3631937503814697, "step": 2277 }, { "epoch": 0.1795821836815136, "grad_norm": 1.225218653678894, "learning_rate": 0.0005, "loss": 2.7963314056396484, "step": 2278 }, { "epoch": 0.17966101694915254, "grad_norm": 1.1356631517410278, "learning_rate": 0.0005, "loss": 2.8412392139434814, "step": 2279 }, { "epoch": 0.17973985021679148, "grad_norm": 1.0289844274520874, "learning_rate": 0.0005, "loss": 2.6706628799438477, "step": 2280 }, { "epoch": 0.17981868348443042, "grad_norm": 1.0518006086349487, "learning_rate": 0.0005, "loss": 2.450634002685547, "step": 2281 }, { "epoch": 0.1798975167520694, "grad_norm": 0.8774546384811401, "learning_rate": 0.0005, "loss": 2.2397072315216064, "step": 2282 }, { "epoch": 0.17997635001970833, "grad_norm": 1.4904707670211792, "learning_rate": 0.0005, "loss": 2.684253215789795, "step": 2283 }, { "epoch": 0.18005518328734726, "grad_norm": 1.019937515258789, "learning_rate": 0.0005, "loss": 2.6980056762695312, "step": 2284 }, { "epoch": 0.1801340165549862, "grad_norm": 1.0693223476409912, "learning_rate": 0.0005, "loss": 2.851945400238037, "step": 2285 }, { "epoch": 0.18021284982262514, "grad_norm": 1.1432245969772339, "learning_rate": 0.0005, "loss": 2.7473816871643066, "step": 2286 }, { "epoch": 0.18029168309026408, "grad_norm": 1.2426998615264893, "learning_rate": 0.0005, "loss": 2.6905791759490967, "step": 2287 }, { "epoch": 0.18037051635790302, "grad_norm": 0.9369615316390991, "learning_rate": 0.0005, "loss": 2.7748515605926514, "step": 2288 }, { "epoch": 0.180449349625542, "grad_norm": 1.1678804159164429, "learning_rate": 0.0005, "loss": 2.4486637115478516, "step": 2289 }, { "epoch": 0.18052818289318093, "grad_norm": 1.2805533409118652, "learning_rate": 0.0005, "loss": 2.7419705390930176, "step": 2290 }, { "epoch": 0.18060701616081987, "grad_norm": 1.01942777633667, "learning_rate": 0.0005, "loss": 3.008758783340454, "step": 2291 }, { "epoch": 0.1806858494284588, "grad_norm": 2.1138319969177246, "learning_rate": 0.0005, "loss": 2.8096766471862793, "step": 2292 }, { "epoch": 0.18076468269609774, "grad_norm": 1.504469394683838, "learning_rate": 0.0005, "loss": 2.866645336151123, "step": 2293 }, { "epoch": 0.18084351596373668, "grad_norm": 0.898655116558075, "learning_rate": 0.0005, "loss": 2.700028419494629, "step": 2294 }, { "epoch": 0.18092234923137565, "grad_norm": 1.146683931350708, "learning_rate": 0.0005, "loss": 2.6779749393463135, "step": 2295 }, { "epoch": 0.1810011824990146, "grad_norm": 1.3707631826400757, "learning_rate": 0.0005, "loss": 2.7819783687591553, "step": 2296 }, { "epoch": 0.18108001576665353, "grad_norm": 1.2755777835845947, "learning_rate": 0.0005, "loss": 2.676527976989746, "step": 2297 }, { "epoch": 0.18115884903429247, "grad_norm": 1.0705821514129639, "learning_rate": 0.0005, "loss": 2.997521162033081, "step": 2298 }, { "epoch": 0.1812376823019314, "grad_norm": 0.8102630972862244, "learning_rate": 0.0005, "loss": 2.2799015045166016, "step": 2299 }, { "epoch": 0.18131651556957035, "grad_norm": 0.940879762172699, "learning_rate": 0.0005, "loss": 3.1850533485412598, "step": 2300 }, { "epoch": 0.1813953488372093, "grad_norm": 0.9750786423683167, "learning_rate": 0.0005, "loss": 2.8068721294403076, "step": 2301 }, { "epoch": 0.18147418210484825, "grad_norm": 0.9477023482322693, "learning_rate": 0.0005, "loss": 2.5583105087280273, "step": 2302 }, { "epoch": 0.1815530153724872, "grad_norm": 1.0175966024398804, "learning_rate": 0.0005, "loss": 2.471951961517334, "step": 2303 }, { "epoch": 0.18163184864012613, "grad_norm": 1.2136741876602173, "learning_rate": 0.0005, "loss": 2.8672189712524414, "step": 2304 }, { "epoch": 0.18171068190776507, "grad_norm": 1.0416210889816284, "learning_rate": 0.0005, "loss": 2.618870735168457, "step": 2305 }, { "epoch": 0.181789515175404, "grad_norm": 0.9559871554374695, "learning_rate": 0.0005, "loss": 3.028887987136841, "step": 2306 }, { "epoch": 0.18186834844304298, "grad_norm": 0.9154088497161865, "learning_rate": 0.0005, "loss": 2.497997760772705, "step": 2307 }, { "epoch": 0.18194718171068192, "grad_norm": 0.8668755888938904, "learning_rate": 0.0005, "loss": 2.497039794921875, "step": 2308 }, { "epoch": 0.18202601497832085, "grad_norm": 0.999305784702301, "learning_rate": 0.0005, "loss": 2.6097428798675537, "step": 2309 }, { "epoch": 0.1821048482459598, "grad_norm": 0.9067691564559937, "learning_rate": 0.0005, "loss": 2.6263911724090576, "step": 2310 }, { "epoch": 0.18218368151359873, "grad_norm": 1.053699254989624, "learning_rate": 0.0005, "loss": 2.523216962814331, "step": 2311 }, { "epoch": 0.18226251478123767, "grad_norm": 1.1650066375732422, "learning_rate": 0.0005, "loss": 2.663867950439453, "step": 2312 }, { "epoch": 0.18234134804887664, "grad_norm": 0.9569174647331238, "learning_rate": 0.0005, "loss": 2.695514678955078, "step": 2313 }, { "epoch": 0.18242018131651558, "grad_norm": 0.9952144026756287, "learning_rate": 0.0005, "loss": 2.8212332725524902, "step": 2314 }, { "epoch": 0.18249901458415452, "grad_norm": 1.4802916049957275, "learning_rate": 0.0005, "loss": 2.6162667274475098, "step": 2315 }, { "epoch": 0.18257784785179346, "grad_norm": 1.05352783203125, "learning_rate": 0.0005, "loss": 2.5829474925994873, "step": 2316 }, { "epoch": 0.1826566811194324, "grad_norm": 1.058163046836853, "learning_rate": 0.0005, "loss": 2.652176856994629, "step": 2317 }, { "epoch": 0.18273551438707133, "grad_norm": 1.2576098442077637, "learning_rate": 0.0005, "loss": 2.5502328872680664, "step": 2318 }, { "epoch": 0.18281434765471027, "grad_norm": 0.9658716917037964, "learning_rate": 0.0005, "loss": 2.330636978149414, "step": 2319 }, { "epoch": 0.18289318092234924, "grad_norm": 1.0206031799316406, "learning_rate": 0.0005, "loss": 2.7988858222961426, "step": 2320 }, { "epoch": 0.18297201418998818, "grad_norm": 1.3315916061401367, "learning_rate": 0.0005, "loss": 2.5948472023010254, "step": 2321 }, { "epoch": 0.18305084745762712, "grad_norm": 0.9993289709091187, "learning_rate": 0.0005, "loss": 2.3883132934570312, "step": 2322 }, { "epoch": 0.18312968072526606, "grad_norm": 1.1751681566238403, "learning_rate": 0.0005, "loss": 2.56803035736084, "step": 2323 }, { "epoch": 0.183208513992905, "grad_norm": 0.9552611708641052, "learning_rate": 0.0005, "loss": 2.7136130332946777, "step": 2324 }, { "epoch": 0.18328734726054394, "grad_norm": 1.8037197589874268, "learning_rate": 0.0005, "loss": 2.7109313011169434, "step": 2325 }, { "epoch": 0.1833661805281829, "grad_norm": 1.200921893119812, "learning_rate": 0.0005, "loss": 2.6431005001068115, "step": 2326 }, { "epoch": 0.18344501379582184, "grad_norm": 0.8842598795890808, "learning_rate": 0.0005, "loss": 2.7901363372802734, "step": 2327 }, { "epoch": 0.18352384706346078, "grad_norm": 1.1299420595169067, "learning_rate": 0.0005, "loss": 2.7463650703430176, "step": 2328 }, { "epoch": 0.18360268033109972, "grad_norm": 0.9842706322669983, "learning_rate": 0.0005, "loss": 2.588191270828247, "step": 2329 }, { "epoch": 0.18368151359873866, "grad_norm": 1.047820806503296, "learning_rate": 0.0005, "loss": 2.5787723064422607, "step": 2330 }, { "epoch": 0.1837603468663776, "grad_norm": 0.9344037771224976, "learning_rate": 0.0005, "loss": 2.582033157348633, "step": 2331 }, { "epoch": 0.18383918013401657, "grad_norm": 1.1761504411697388, "learning_rate": 0.0005, "loss": 3.0036630630493164, "step": 2332 }, { "epoch": 0.1839180134016555, "grad_norm": 1.1417016983032227, "learning_rate": 0.0005, "loss": 2.4782676696777344, "step": 2333 }, { "epoch": 0.18399684666929444, "grad_norm": 1.2467349767684937, "learning_rate": 0.0005, "loss": 2.7620911598205566, "step": 2334 }, { "epoch": 0.18407567993693338, "grad_norm": 1.0760167837142944, "learning_rate": 0.0005, "loss": 2.7827014923095703, "step": 2335 }, { "epoch": 0.18415451320457232, "grad_norm": 1.027302622795105, "learning_rate": 0.0005, "loss": 3.182323455810547, "step": 2336 }, { "epoch": 0.18423334647221126, "grad_norm": 1.06200110912323, "learning_rate": 0.0005, "loss": 2.387758255004883, "step": 2337 }, { "epoch": 0.18431217973985023, "grad_norm": 0.8953685760498047, "learning_rate": 0.0005, "loss": 2.1546099185943604, "step": 2338 }, { "epoch": 0.18439101300748917, "grad_norm": 0.964486300945282, "learning_rate": 0.0005, "loss": 2.4613900184631348, "step": 2339 }, { "epoch": 0.1844698462751281, "grad_norm": 1.0783751010894775, "learning_rate": 0.0005, "loss": 2.7637922763824463, "step": 2340 }, { "epoch": 0.18454867954276705, "grad_norm": 2.7541544437408447, "learning_rate": 0.0005, "loss": 2.453248977661133, "step": 2341 }, { "epoch": 0.18462751281040599, "grad_norm": 1.0020729303359985, "learning_rate": 0.0005, "loss": 3.1716036796569824, "step": 2342 }, { "epoch": 0.18470634607804493, "grad_norm": 1.1928017139434814, "learning_rate": 0.0005, "loss": 2.762449026107788, "step": 2343 }, { "epoch": 0.1847851793456839, "grad_norm": 0.9684150815010071, "learning_rate": 0.0005, "loss": 2.486593723297119, "step": 2344 }, { "epoch": 0.18486401261332283, "grad_norm": 1.133639931678772, "learning_rate": 0.0005, "loss": 2.676957607269287, "step": 2345 }, { "epoch": 0.18494284588096177, "grad_norm": 1.238297462463379, "learning_rate": 0.0005, "loss": 2.8723621368408203, "step": 2346 }, { "epoch": 0.1850216791486007, "grad_norm": 0.9358422160148621, "learning_rate": 0.0005, "loss": 2.494450092315674, "step": 2347 }, { "epoch": 0.18510051241623965, "grad_norm": 0.9486033916473389, "learning_rate": 0.0005, "loss": 2.8086047172546387, "step": 2348 }, { "epoch": 0.1851793456838786, "grad_norm": 0.9581010341644287, "learning_rate": 0.0005, "loss": 2.9744081497192383, "step": 2349 }, { "epoch": 0.18525817895151753, "grad_norm": 1.256529450416565, "learning_rate": 0.0005, "loss": 2.458089590072632, "step": 2350 }, { "epoch": 0.1853370122191565, "grad_norm": 0.8593539595603943, "learning_rate": 0.0005, "loss": 2.515038013458252, "step": 2351 }, { "epoch": 0.18541584548679543, "grad_norm": 1.1450064182281494, "learning_rate": 0.0005, "loss": 2.6631011962890625, "step": 2352 }, { "epoch": 0.18549467875443437, "grad_norm": 1.066123604774475, "learning_rate": 0.0005, "loss": 2.7734785079956055, "step": 2353 }, { "epoch": 0.1855735120220733, "grad_norm": 0.9960803389549255, "learning_rate": 0.0005, "loss": 2.6026573181152344, "step": 2354 }, { "epoch": 0.18565234528971225, "grad_norm": 1.0478031635284424, "learning_rate": 0.0005, "loss": 2.408513069152832, "step": 2355 }, { "epoch": 0.1857311785573512, "grad_norm": 1.2200008630752563, "learning_rate": 0.0005, "loss": 2.6619842052459717, "step": 2356 }, { "epoch": 0.18581001182499016, "grad_norm": 0.8578428030014038, "learning_rate": 0.0005, "loss": 2.683842897415161, "step": 2357 }, { "epoch": 0.1858888450926291, "grad_norm": 1.0382875204086304, "learning_rate": 0.0005, "loss": 2.672900676727295, "step": 2358 }, { "epoch": 0.18596767836026803, "grad_norm": 1.0650529861450195, "learning_rate": 0.0005, "loss": 3.0110230445861816, "step": 2359 }, { "epoch": 0.18604651162790697, "grad_norm": 1.1040610074996948, "learning_rate": 0.0005, "loss": 2.8702244758605957, "step": 2360 }, { "epoch": 0.1861253448955459, "grad_norm": 1.0194286108016968, "learning_rate": 0.0005, "loss": 2.5293545722961426, "step": 2361 }, { "epoch": 0.18620417816318485, "grad_norm": 0.8522287011146545, "learning_rate": 0.0005, "loss": 2.718371868133545, "step": 2362 }, { "epoch": 0.18628301143082382, "grad_norm": 1.00639009475708, "learning_rate": 0.0005, "loss": 2.8530654907226562, "step": 2363 }, { "epoch": 0.18636184469846276, "grad_norm": 0.8423124551773071, "learning_rate": 0.0005, "loss": 2.6799333095550537, "step": 2364 }, { "epoch": 0.1864406779661017, "grad_norm": 0.9393455982208252, "learning_rate": 0.0005, "loss": 2.511272668838501, "step": 2365 }, { "epoch": 0.18651951123374064, "grad_norm": 1.1226112842559814, "learning_rate": 0.0005, "loss": 2.280374526977539, "step": 2366 }, { "epoch": 0.18659834450137958, "grad_norm": 1.198752999305725, "learning_rate": 0.0005, "loss": 2.849874496459961, "step": 2367 }, { "epoch": 0.18667717776901852, "grad_norm": 0.8683833479881287, "learning_rate": 0.0005, "loss": 2.502760171890259, "step": 2368 }, { "epoch": 0.18675601103665748, "grad_norm": 0.9201457500457764, "learning_rate": 0.0005, "loss": 2.6080498695373535, "step": 2369 }, { "epoch": 0.18683484430429642, "grad_norm": 0.828288197517395, "learning_rate": 0.0005, "loss": 2.2676472663879395, "step": 2370 }, { "epoch": 0.18691367757193536, "grad_norm": 0.9147785902023315, "learning_rate": 0.0005, "loss": 2.757587194442749, "step": 2371 }, { "epoch": 0.1869925108395743, "grad_norm": 0.942970871925354, "learning_rate": 0.0005, "loss": 2.4588451385498047, "step": 2372 }, { "epoch": 0.18707134410721324, "grad_norm": 0.9042602777481079, "learning_rate": 0.0005, "loss": 2.5306568145751953, "step": 2373 }, { "epoch": 0.18715017737485218, "grad_norm": 1.1202579736709595, "learning_rate": 0.0005, "loss": 2.8024635314941406, "step": 2374 }, { "epoch": 0.18722901064249114, "grad_norm": 0.8970156311988831, "learning_rate": 0.0005, "loss": 2.7833428382873535, "step": 2375 }, { "epoch": 0.18730784391013008, "grad_norm": 0.8490152359008789, "learning_rate": 0.0005, "loss": 2.4864377975463867, "step": 2376 }, { "epoch": 0.18738667717776902, "grad_norm": 0.878095805644989, "learning_rate": 0.0005, "loss": 2.3073525428771973, "step": 2377 }, { "epoch": 0.18746551044540796, "grad_norm": 1.1592655181884766, "learning_rate": 0.0005, "loss": 2.8958864212036133, "step": 2378 }, { "epoch": 0.1875443437130469, "grad_norm": 1.8616007566452026, "learning_rate": 0.0005, "loss": 2.8445334434509277, "step": 2379 }, { "epoch": 0.18762317698068584, "grad_norm": 0.8850615620613098, "learning_rate": 0.0005, "loss": 2.980963706970215, "step": 2380 }, { "epoch": 0.18770201024832478, "grad_norm": 1.241101622581482, "learning_rate": 0.0005, "loss": 3.1718432903289795, "step": 2381 }, { "epoch": 0.18778084351596375, "grad_norm": 1.335045337677002, "learning_rate": 0.0005, "loss": 2.825233221054077, "step": 2382 }, { "epoch": 0.18785967678360269, "grad_norm": 1.2039177417755127, "learning_rate": 0.0005, "loss": 2.671567916870117, "step": 2383 }, { "epoch": 0.18793851005124163, "grad_norm": 1.739733338356018, "learning_rate": 0.0005, "loss": 2.4958598613739014, "step": 2384 }, { "epoch": 0.18801734331888056, "grad_norm": 1.2788114547729492, "learning_rate": 0.0005, "loss": 2.7607669830322266, "step": 2385 }, { "epoch": 0.1880961765865195, "grad_norm": 1.0460553169250488, "learning_rate": 0.0005, "loss": 2.42380428314209, "step": 2386 }, { "epoch": 0.18817500985415844, "grad_norm": 1.0644012689590454, "learning_rate": 0.0005, "loss": 2.820725440979004, "step": 2387 }, { "epoch": 0.1882538431217974, "grad_norm": 0.8689565658569336, "learning_rate": 0.0005, "loss": 2.285491466522217, "step": 2388 }, { "epoch": 0.18833267638943635, "grad_norm": 1.0212135314941406, "learning_rate": 0.0005, "loss": 2.7885665893554688, "step": 2389 }, { "epoch": 0.1884115096570753, "grad_norm": 1.0351250171661377, "learning_rate": 0.0005, "loss": 2.2825865745544434, "step": 2390 }, { "epoch": 0.18849034292471423, "grad_norm": 1.1274422407150269, "learning_rate": 0.0005, "loss": 2.231828212738037, "step": 2391 }, { "epoch": 0.18856917619235317, "grad_norm": 1.0819287300109863, "learning_rate": 0.0005, "loss": 2.6531128883361816, "step": 2392 }, { "epoch": 0.1886480094599921, "grad_norm": 1.0232441425323486, "learning_rate": 0.0005, "loss": 2.5174031257629395, "step": 2393 }, { "epoch": 0.18872684272763107, "grad_norm": 1.214830994606018, "learning_rate": 0.0005, "loss": 2.7845616340637207, "step": 2394 }, { "epoch": 0.18880567599527, "grad_norm": 1.178692102432251, "learning_rate": 0.0005, "loss": 2.4483580589294434, "step": 2395 }, { "epoch": 0.18888450926290895, "grad_norm": 1.1661494970321655, "learning_rate": 0.0005, "loss": 2.494931221008301, "step": 2396 }, { "epoch": 0.1889633425305479, "grad_norm": 0.9948515892028809, "learning_rate": 0.0005, "loss": 2.7664546966552734, "step": 2397 }, { "epoch": 0.18904217579818683, "grad_norm": 1.0228278636932373, "learning_rate": 0.0005, "loss": 2.415935754776001, "step": 2398 }, { "epoch": 0.18912100906582577, "grad_norm": 0.9594892263412476, "learning_rate": 0.0005, "loss": 2.7119836807250977, "step": 2399 }, { "epoch": 0.18919984233346474, "grad_norm": 1.1137006282806396, "learning_rate": 0.0005, "loss": 2.567793846130371, "step": 2400 }, { "epoch": 0.18927867560110367, "grad_norm": 1.147626519203186, "learning_rate": 0.0005, "loss": 2.899252414703369, "step": 2401 }, { "epoch": 0.1893575088687426, "grad_norm": 1.0207709074020386, "learning_rate": 0.0005, "loss": 2.6342244148254395, "step": 2402 }, { "epoch": 0.18943634213638155, "grad_norm": 1.1680208444595337, "learning_rate": 0.0005, "loss": 2.6401772499084473, "step": 2403 }, { "epoch": 0.1895151754040205, "grad_norm": 1.0891178846359253, "learning_rate": 0.0005, "loss": 2.7868175506591797, "step": 2404 }, { "epoch": 0.18959400867165943, "grad_norm": 1.1003046035766602, "learning_rate": 0.0005, "loss": 2.861501693725586, "step": 2405 }, { "epoch": 0.1896728419392984, "grad_norm": 0.9495231509208679, "learning_rate": 0.0005, "loss": 2.674321174621582, "step": 2406 }, { "epoch": 0.18975167520693734, "grad_norm": 1.1561412811279297, "learning_rate": 0.0005, "loss": 2.6001157760620117, "step": 2407 }, { "epoch": 0.18983050847457628, "grad_norm": 0.9196112155914307, "learning_rate": 0.0005, "loss": 2.663233757019043, "step": 2408 }, { "epoch": 0.18990934174221522, "grad_norm": 1.01762855052948, "learning_rate": 0.0005, "loss": 2.7314517498016357, "step": 2409 }, { "epoch": 0.18998817500985415, "grad_norm": 0.9635135531425476, "learning_rate": 0.0005, "loss": 2.5037548542022705, "step": 2410 }, { "epoch": 0.1900670082774931, "grad_norm": 0.9670206308364868, "learning_rate": 0.0005, "loss": 2.430053234100342, "step": 2411 }, { "epoch": 0.19014584154513203, "grad_norm": 1.0591756105422974, "learning_rate": 0.0005, "loss": 2.2550387382507324, "step": 2412 }, { "epoch": 0.190224674812771, "grad_norm": 0.9063290357589722, "learning_rate": 0.0005, "loss": 2.5374162197113037, "step": 2413 }, { "epoch": 0.19030350808040994, "grad_norm": 0.9003044962882996, "learning_rate": 0.0005, "loss": 2.5424938201904297, "step": 2414 }, { "epoch": 0.19038234134804888, "grad_norm": 1.043797254562378, "learning_rate": 0.0005, "loss": 3.085576295852661, "step": 2415 }, { "epoch": 0.19046117461568782, "grad_norm": 0.8042542338371277, "learning_rate": 0.0005, "loss": 2.8081891536712646, "step": 2416 }, { "epoch": 0.19054000788332676, "grad_norm": 0.9239305853843689, "learning_rate": 0.0005, "loss": 2.7489376068115234, "step": 2417 }, { "epoch": 0.1906188411509657, "grad_norm": 1.005539059638977, "learning_rate": 0.0005, "loss": 3.0806384086608887, "step": 2418 }, { "epoch": 0.19069767441860466, "grad_norm": 0.9680427312850952, "learning_rate": 0.0005, "loss": 2.337660551071167, "step": 2419 }, { "epoch": 0.1907765076862436, "grad_norm": 0.9919313788414001, "learning_rate": 0.0005, "loss": 2.9426381587982178, "step": 2420 }, { "epoch": 0.19085534095388254, "grad_norm": 1.0967777967453003, "learning_rate": 0.0005, "loss": 2.6274144649505615, "step": 2421 }, { "epoch": 0.19093417422152148, "grad_norm": 1.1214097738265991, "learning_rate": 0.0005, "loss": 2.485013961791992, "step": 2422 }, { "epoch": 0.19101300748916042, "grad_norm": 1.019237995147705, "learning_rate": 0.0005, "loss": 2.6558408737182617, "step": 2423 }, { "epoch": 0.19109184075679936, "grad_norm": 0.9276807308197021, "learning_rate": 0.0005, "loss": 2.646721363067627, "step": 2424 }, { "epoch": 0.19117067402443833, "grad_norm": 1.0770560503005981, "learning_rate": 0.0005, "loss": 2.7146472930908203, "step": 2425 }, { "epoch": 0.19124950729207726, "grad_norm": 1.1015822887420654, "learning_rate": 0.0005, "loss": 2.7790865898132324, "step": 2426 }, { "epoch": 0.1913283405597162, "grad_norm": 0.7970889806747437, "learning_rate": 0.0005, "loss": 2.534234046936035, "step": 2427 }, { "epoch": 0.19140717382735514, "grad_norm": 0.8760112524032593, "learning_rate": 0.0005, "loss": 2.4913153648376465, "step": 2428 }, { "epoch": 0.19148600709499408, "grad_norm": 1.1087721586227417, "learning_rate": 0.0005, "loss": 2.913163423538208, "step": 2429 }, { "epoch": 0.19156484036263302, "grad_norm": 0.9055969715118408, "learning_rate": 0.0005, "loss": 2.5556721687316895, "step": 2430 }, { "epoch": 0.191643673630272, "grad_norm": 1.2412151098251343, "learning_rate": 0.0005, "loss": 2.87025785446167, "step": 2431 }, { "epoch": 0.19172250689791093, "grad_norm": 1.3139358758926392, "learning_rate": 0.0005, "loss": 2.8794925212860107, "step": 2432 }, { "epoch": 0.19180134016554987, "grad_norm": 0.816674530506134, "learning_rate": 0.0005, "loss": 2.3464102745056152, "step": 2433 }, { "epoch": 0.1918801734331888, "grad_norm": 1.0583889484405518, "learning_rate": 0.0005, "loss": 2.577080726623535, "step": 2434 }, { "epoch": 0.19195900670082774, "grad_norm": 1.0480490922927856, "learning_rate": 0.0005, "loss": 2.5472121238708496, "step": 2435 }, { "epoch": 0.19203783996846668, "grad_norm": 0.8256679177284241, "learning_rate": 0.0005, "loss": 2.1986136436462402, "step": 2436 }, { "epoch": 0.19211667323610562, "grad_norm": 0.9618512392044067, "learning_rate": 0.0005, "loss": 2.527754545211792, "step": 2437 }, { "epoch": 0.1921955065037446, "grad_norm": 1.0109405517578125, "learning_rate": 0.0005, "loss": 2.5104928016662598, "step": 2438 }, { "epoch": 0.19227433977138353, "grad_norm": 1.1588140726089478, "learning_rate": 0.0005, "loss": 3.0463385581970215, "step": 2439 }, { "epoch": 0.19235317303902247, "grad_norm": 0.8098783493041992, "learning_rate": 0.0005, "loss": 2.601257085800171, "step": 2440 }, { "epoch": 0.1924320063066614, "grad_norm": 1.5955299139022827, "learning_rate": 0.0005, "loss": 2.384565591812134, "step": 2441 }, { "epoch": 0.19251083957430035, "grad_norm": 1.2785089015960693, "learning_rate": 0.0005, "loss": 2.8703198432922363, "step": 2442 }, { "epoch": 0.19258967284193929, "grad_norm": 1.0532047748565674, "learning_rate": 0.0005, "loss": 2.416071891784668, "step": 2443 }, { "epoch": 0.19266850610957825, "grad_norm": 1.0425606966018677, "learning_rate": 0.0005, "loss": 2.7306504249572754, "step": 2444 }, { "epoch": 0.1927473393772172, "grad_norm": 0.9469169974327087, "learning_rate": 0.0005, "loss": 2.7907066345214844, "step": 2445 }, { "epoch": 0.19282617264485613, "grad_norm": 0.9832057356834412, "learning_rate": 0.0005, "loss": 2.4341163635253906, "step": 2446 }, { "epoch": 0.19290500591249507, "grad_norm": 0.8950257897377014, "learning_rate": 0.0005, "loss": 2.515902519226074, "step": 2447 }, { "epoch": 0.192983839180134, "grad_norm": 1.0437333583831787, "learning_rate": 0.0005, "loss": 2.6459531784057617, "step": 2448 }, { "epoch": 0.19306267244777295, "grad_norm": 1.0203205347061157, "learning_rate": 0.0005, "loss": 2.5220894813537598, "step": 2449 }, { "epoch": 0.19314150571541192, "grad_norm": 0.9791374206542969, "learning_rate": 0.0005, "loss": 2.511688709259033, "step": 2450 }, { "epoch": 0.19322033898305085, "grad_norm": 0.8399664163589478, "learning_rate": 0.0005, "loss": 2.7451021671295166, "step": 2451 }, { "epoch": 0.1932991722506898, "grad_norm": 1.1111396551132202, "learning_rate": 0.0005, "loss": 2.8369810581207275, "step": 2452 }, { "epoch": 0.19337800551832873, "grad_norm": 0.9827808737754822, "learning_rate": 0.0005, "loss": 2.7262630462646484, "step": 2453 }, { "epoch": 0.19345683878596767, "grad_norm": 1.0249123573303223, "learning_rate": 0.0005, "loss": 2.568096876144409, "step": 2454 }, { "epoch": 0.1935356720536066, "grad_norm": 1.030574917793274, "learning_rate": 0.0005, "loss": 2.6337757110595703, "step": 2455 }, { "epoch": 0.19361450532124558, "grad_norm": 0.9753539562225342, "learning_rate": 0.0005, "loss": 2.8977231979370117, "step": 2456 }, { "epoch": 0.19369333858888452, "grad_norm": 1.1447563171386719, "learning_rate": 0.0005, "loss": 2.479465961456299, "step": 2457 }, { "epoch": 0.19377217185652346, "grad_norm": 1.2520424127578735, "learning_rate": 0.0005, "loss": 2.8836677074432373, "step": 2458 }, { "epoch": 0.1938510051241624, "grad_norm": 1.0062942504882812, "learning_rate": 0.0005, "loss": 2.7463479042053223, "step": 2459 }, { "epoch": 0.19392983839180133, "grad_norm": 0.9892053604125977, "learning_rate": 0.0005, "loss": 2.686248302459717, "step": 2460 }, { "epoch": 0.19400867165944027, "grad_norm": 0.884956955909729, "learning_rate": 0.0005, "loss": 2.776073932647705, "step": 2461 }, { "epoch": 0.19408750492707924, "grad_norm": 1.0505714416503906, "learning_rate": 0.0005, "loss": 2.540072441101074, "step": 2462 }, { "epoch": 0.19416633819471818, "grad_norm": 8.078457832336426, "learning_rate": 0.0005, "loss": 3.1601834297180176, "step": 2463 }, { "epoch": 0.19424517146235712, "grad_norm": 0.922254204750061, "learning_rate": 0.0005, "loss": 2.8969244956970215, "step": 2464 }, { "epoch": 0.19432400472999606, "grad_norm": 0.9974682331085205, "learning_rate": 0.0005, "loss": 2.4934849739074707, "step": 2465 }, { "epoch": 0.194402837997635, "grad_norm": 0.8781672120094299, "learning_rate": 0.0005, "loss": 2.537794351577759, "step": 2466 }, { "epoch": 0.19448167126527394, "grad_norm": 0.7528958916664124, "learning_rate": 0.0005, "loss": 2.5549731254577637, "step": 2467 }, { "epoch": 0.19456050453291288, "grad_norm": 0.8797188401222229, "learning_rate": 0.0005, "loss": 2.822403907775879, "step": 2468 }, { "epoch": 0.19463933780055184, "grad_norm": 1.0063951015472412, "learning_rate": 0.0005, "loss": 2.4686226844787598, "step": 2469 }, { "epoch": 0.19471817106819078, "grad_norm": 1.072252631187439, "learning_rate": 0.0005, "loss": 3.107175350189209, "step": 2470 }, { "epoch": 0.19479700433582972, "grad_norm": 3.1175291538238525, "learning_rate": 0.0005, "loss": 2.6702938079833984, "step": 2471 }, { "epoch": 0.19487583760346866, "grad_norm": 1.1402277946472168, "learning_rate": 0.0005, "loss": 2.8606700897216797, "step": 2472 }, { "epoch": 0.1949546708711076, "grad_norm": 0.9956392645835876, "learning_rate": 0.0005, "loss": 2.710249423980713, "step": 2473 }, { "epoch": 0.19503350413874654, "grad_norm": 1.1129188537597656, "learning_rate": 0.0005, "loss": 2.948437452316284, "step": 2474 }, { "epoch": 0.1951123374063855, "grad_norm": 2.1646840572357178, "learning_rate": 0.0005, "loss": 2.5371034145355225, "step": 2475 }, { "epoch": 0.19519117067402444, "grad_norm": 0.8780826926231384, "learning_rate": 0.0005, "loss": 2.5990686416625977, "step": 2476 }, { "epoch": 0.19527000394166338, "grad_norm": 1.1331294775009155, "learning_rate": 0.0005, "loss": 2.7909951210021973, "step": 2477 }, { "epoch": 0.19534883720930232, "grad_norm": 0.9657599329948425, "learning_rate": 0.0005, "loss": 3.2087419033050537, "step": 2478 }, { "epoch": 0.19542767047694126, "grad_norm": 0.858260989189148, "learning_rate": 0.0005, "loss": 2.683053970336914, "step": 2479 }, { "epoch": 0.1955065037445802, "grad_norm": 1.2061443328857422, "learning_rate": 0.0005, "loss": 2.8546950817108154, "step": 2480 }, { "epoch": 0.19558533701221917, "grad_norm": 0.9024602770805359, "learning_rate": 0.0005, "loss": 2.150848150253296, "step": 2481 }, { "epoch": 0.1956641702798581, "grad_norm": 1.3323324918746948, "learning_rate": 0.0005, "loss": 2.697197914123535, "step": 2482 }, { "epoch": 0.19574300354749705, "grad_norm": 0.7898015975952148, "learning_rate": 0.0005, "loss": 2.4084115028381348, "step": 2483 }, { "epoch": 0.19582183681513599, "grad_norm": 1.025084137916565, "learning_rate": 0.0005, "loss": 2.9010863304138184, "step": 2484 }, { "epoch": 0.19590067008277492, "grad_norm": 0.8247458338737488, "learning_rate": 0.0005, "loss": 2.7293591499328613, "step": 2485 }, { "epoch": 0.19597950335041386, "grad_norm": 1.0002952814102173, "learning_rate": 0.0005, "loss": 2.6219089031219482, "step": 2486 }, { "epoch": 0.19605833661805283, "grad_norm": 0.8095434308052063, "learning_rate": 0.0005, "loss": 2.531818389892578, "step": 2487 }, { "epoch": 0.19613716988569177, "grad_norm": 1.1663185358047485, "learning_rate": 0.0005, "loss": 2.5916826725006104, "step": 2488 }, { "epoch": 0.1962160031533307, "grad_norm": 0.8555879592895508, "learning_rate": 0.0005, "loss": 2.7711238861083984, "step": 2489 }, { "epoch": 0.19629483642096965, "grad_norm": 0.815717875957489, "learning_rate": 0.0005, "loss": 2.823030471801758, "step": 2490 }, { "epoch": 0.1963736696886086, "grad_norm": 1.0075913667678833, "learning_rate": 0.0005, "loss": 2.5585520267486572, "step": 2491 }, { "epoch": 0.19645250295624753, "grad_norm": 1.1205532550811768, "learning_rate": 0.0005, "loss": 2.931424617767334, "step": 2492 }, { "epoch": 0.1965313362238865, "grad_norm": 0.9994161128997803, "learning_rate": 0.0005, "loss": 2.8179891109466553, "step": 2493 }, { "epoch": 0.19661016949152543, "grad_norm": 0.9588103890419006, "learning_rate": 0.0005, "loss": 2.657418727874756, "step": 2494 }, { "epoch": 0.19668900275916437, "grad_norm": 1.214605689048767, "learning_rate": 0.0005, "loss": 2.7736291885375977, "step": 2495 }, { "epoch": 0.1967678360268033, "grad_norm": 1.0353764295578003, "learning_rate": 0.0005, "loss": 2.638989210128784, "step": 2496 }, { "epoch": 0.19684666929444225, "grad_norm": 0.958460807800293, "learning_rate": 0.0005, "loss": 2.513265609741211, "step": 2497 }, { "epoch": 0.1969255025620812, "grad_norm": 0.85148686170578, "learning_rate": 0.0005, "loss": 2.735229253768921, "step": 2498 }, { "epoch": 0.19700433582972013, "grad_norm": 0.9048178195953369, "learning_rate": 0.0005, "loss": 2.404139518737793, "step": 2499 }, { "epoch": 0.1970831690973591, "grad_norm": 0.9280090928077698, "learning_rate": 0.0005, "loss": 2.751368999481201, "step": 2500 }, { "epoch": 0.19716200236499803, "grad_norm": 1.2748080492019653, "learning_rate": 0.0005, "loss": 3.05629301071167, "step": 2501 }, { "epoch": 0.19724083563263697, "grad_norm": 1.5365474224090576, "learning_rate": 0.0005, "loss": 2.604811668395996, "step": 2502 }, { "epoch": 0.1973196689002759, "grad_norm": 1.2482755184173584, "learning_rate": 0.0005, "loss": 2.6370420455932617, "step": 2503 }, { "epoch": 0.19739850216791485, "grad_norm": 1.0076308250427246, "learning_rate": 0.0005, "loss": 2.6333060264587402, "step": 2504 }, { "epoch": 0.1974773354355538, "grad_norm": 1.2190065383911133, "learning_rate": 0.0005, "loss": 2.496859073638916, "step": 2505 }, { "epoch": 0.19755616870319276, "grad_norm": 0.986742377281189, "learning_rate": 0.0005, "loss": 2.6206483840942383, "step": 2506 }, { "epoch": 0.1976350019708317, "grad_norm": 1.0855417251586914, "learning_rate": 0.0005, "loss": 2.8533802032470703, "step": 2507 }, { "epoch": 0.19771383523847064, "grad_norm": 0.8633438944816589, "learning_rate": 0.0005, "loss": 2.786648988723755, "step": 2508 }, { "epoch": 0.19779266850610958, "grad_norm": 0.8587032556533813, "learning_rate": 0.0005, "loss": 2.583740711212158, "step": 2509 }, { "epoch": 0.19787150177374851, "grad_norm": 1.0496612787246704, "learning_rate": 0.0005, "loss": 2.687518358230591, "step": 2510 }, { "epoch": 0.19795033504138745, "grad_norm": 1.2877854108810425, "learning_rate": 0.0005, "loss": 2.958827018737793, "step": 2511 }, { "epoch": 0.19802916830902642, "grad_norm": 1.1276991367340088, "learning_rate": 0.0005, "loss": 2.6223952770233154, "step": 2512 }, { "epoch": 0.19810800157666536, "grad_norm": 0.9614228010177612, "learning_rate": 0.0005, "loss": 2.7023415565490723, "step": 2513 }, { "epoch": 0.1981868348443043, "grad_norm": 1.0054811239242554, "learning_rate": 0.0005, "loss": 2.6940290927886963, "step": 2514 }, { "epoch": 0.19826566811194324, "grad_norm": 0.9528623223304749, "learning_rate": 0.0005, "loss": 2.432037115097046, "step": 2515 }, { "epoch": 0.19834450137958218, "grad_norm": 1.0256402492523193, "learning_rate": 0.0005, "loss": 2.7719149589538574, "step": 2516 }, { "epoch": 0.19842333464722112, "grad_norm": 0.8789752125740051, "learning_rate": 0.0005, "loss": 2.7717437744140625, "step": 2517 }, { "epoch": 0.19850216791486008, "grad_norm": 1.3696142435073853, "learning_rate": 0.0005, "loss": 2.74941349029541, "step": 2518 }, { "epoch": 0.19858100118249902, "grad_norm": 0.7793259620666504, "learning_rate": 0.0005, "loss": 2.676173210144043, "step": 2519 }, { "epoch": 0.19865983445013796, "grad_norm": 0.8371557593345642, "learning_rate": 0.0005, "loss": 2.578239917755127, "step": 2520 }, { "epoch": 0.1987386677177769, "grad_norm": 1.0318819284439087, "learning_rate": 0.0005, "loss": 2.4764068126678467, "step": 2521 }, { "epoch": 0.19881750098541584, "grad_norm": 1.386668086051941, "learning_rate": 0.0005, "loss": 2.737165927886963, "step": 2522 }, { "epoch": 0.19889633425305478, "grad_norm": 0.9469489455223083, "learning_rate": 0.0005, "loss": 2.398475170135498, "step": 2523 }, { "epoch": 0.19897516752069375, "grad_norm": 0.8656641244888306, "learning_rate": 0.0005, "loss": 2.565873146057129, "step": 2524 }, { "epoch": 0.19905400078833269, "grad_norm": 1.27346670627594, "learning_rate": 0.0005, "loss": 2.895312786102295, "step": 2525 }, { "epoch": 0.19913283405597162, "grad_norm": 1.257493019104004, "learning_rate": 0.0005, "loss": 2.5466980934143066, "step": 2526 }, { "epoch": 0.19921166732361056, "grad_norm": 0.9542024731636047, "learning_rate": 0.0005, "loss": 2.7185754776000977, "step": 2527 }, { "epoch": 0.1992905005912495, "grad_norm": 0.9652018547058105, "learning_rate": 0.0005, "loss": 2.532043695449829, "step": 2528 }, { "epoch": 0.19936933385888844, "grad_norm": 0.9373133182525635, "learning_rate": 0.0005, "loss": 2.512842893600464, "step": 2529 }, { "epoch": 0.19944816712652738, "grad_norm": 1.0185050964355469, "learning_rate": 0.0005, "loss": 2.1947031021118164, "step": 2530 }, { "epoch": 0.19952700039416635, "grad_norm": 1.4695628881454468, "learning_rate": 0.0005, "loss": 2.3892955780029297, "step": 2531 }, { "epoch": 0.1996058336618053, "grad_norm": 1.3470144271850586, "learning_rate": 0.0005, "loss": 2.508077621459961, "step": 2532 }, { "epoch": 0.19968466692944423, "grad_norm": 1.0884485244750977, "learning_rate": 0.0005, "loss": 2.685898780822754, "step": 2533 }, { "epoch": 0.19976350019708317, "grad_norm": 0.7967825531959534, "learning_rate": 0.0005, "loss": 2.3974668979644775, "step": 2534 }, { "epoch": 0.1998423334647221, "grad_norm": 0.962364673614502, "learning_rate": 0.0005, "loss": 2.5280728340148926, "step": 2535 }, { "epoch": 0.19992116673236104, "grad_norm": 1.1174241304397583, "learning_rate": 0.0005, "loss": 2.5478553771972656, "step": 2536 }, { "epoch": 0.2, "grad_norm": 0.9125129580497742, "learning_rate": 0.0005, "loss": 2.93860125541687, "step": 2537 }, { "epoch": 0.20007883326763895, "grad_norm": 0.949864387512207, "learning_rate": 0.0005, "loss": 2.730991840362549, "step": 2538 }, { "epoch": 0.2001576665352779, "grad_norm": 0.8202432990074158, "learning_rate": 0.0005, "loss": 2.677243709564209, "step": 2539 }, { "epoch": 0.20023649980291683, "grad_norm": 2.085073947906494, "learning_rate": 0.0005, "loss": 2.680115222930908, "step": 2540 }, { "epoch": 0.20031533307055577, "grad_norm": 1.0523724555969238, "learning_rate": 0.0005, "loss": 2.8508100509643555, "step": 2541 }, { "epoch": 0.2003941663381947, "grad_norm": 1.062444806098938, "learning_rate": 0.0005, "loss": 2.5812630653381348, "step": 2542 }, { "epoch": 0.20047299960583367, "grad_norm": 1.1212074756622314, "learning_rate": 0.0005, "loss": 2.4808571338653564, "step": 2543 }, { "epoch": 0.2005518328734726, "grad_norm": 1.0486114025115967, "learning_rate": 0.0005, "loss": 2.5098371505737305, "step": 2544 }, { "epoch": 0.20063066614111155, "grad_norm": 1.1200124025344849, "learning_rate": 0.0005, "loss": 2.7584614753723145, "step": 2545 }, { "epoch": 0.2007094994087505, "grad_norm": 1.1388381719589233, "learning_rate": 0.0005, "loss": 2.756577730178833, "step": 2546 }, { "epoch": 0.20078833267638943, "grad_norm": 0.9633463025093079, "learning_rate": 0.0005, "loss": 2.5189309120178223, "step": 2547 }, { "epoch": 0.20086716594402837, "grad_norm": 1.004089593887329, "learning_rate": 0.0005, "loss": 2.4873881340026855, "step": 2548 }, { "epoch": 0.20094599921166734, "grad_norm": 0.9025144577026367, "learning_rate": 0.0005, "loss": 2.731472969055176, "step": 2549 }, { "epoch": 0.20102483247930628, "grad_norm": 1.1555562019348145, "learning_rate": 0.0005, "loss": 2.679210901260376, "step": 2550 }, { "epoch": 0.20110366574694521, "grad_norm": 1.018526554107666, "learning_rate": 0.0005, "loss": 2.6355972290039062, "step": 2551 }, { "epoch": 0.20118249901458415, "grad_norm": 0.9001516699790955, "learning_rate": 0.0005, "loss": 2.6362104415893555, "step": 2552 }, { "epoch": 0.2012613322822231, "grad_norm": 1.0012215375900269, "learning_rate": 0.0005, "loss": 2.615786075592041, "step": 2553 }, { "epoch": 0.20134016554986203, "grad_norm": 1.0227208137512207, "learning_rate": 0.0005, "loss": 2.896394729614258, "step": 2554 }, { "epoch": 0.20141899881750097, "grad_norm": 0.9129611849784851, "learning_rate": 0.0005, "loss": 2.427722930908203, "step": 2555 }, { "epoch": 0.20149783208513994, "grad_norm": 0.976629376411438, "learning_rate": 0.0005, "loss": 2.6675424575805664, "step": 2556 }, { "epoch": 0.20157666535277888, "grad_norm": 0.8341677188873291, "learning_rate": 0.0005, "loss": 2.6822915077209473, "step": 2557 }, { "epoch": 0.20165549862041782, "grad_norm": 0.8719232678413391, "learning_rate": 0.0005, "loss": 2.6575870513916016, "step": 2558 }, { "epoch": 0.20173433188805676, "grad_norm": 1.1425106525421143, "learning_rate": 0.0005, "loss": 2.8020617961883545, "step": 2559 }, { "epoch": 0.2018131651556957, "grad_norm": 0.9795721173286438, "learning_rate": 0.0005, "loss": 3.0515522956848145, "step": 2560 }, { "epoch": 0.20189199842333463, "grad_norm": 0.914038360118866, "learning_rate": 0.0005, "loss": 2.5670461654663086, "step": 2561 }, { "epoch": 0.2019708316909736, "grad_norm": 0.8821324706077576, "learning_rate": 0.0005, "loss": 2.868488311767578, "step": 2562 }, { "epoch": 0.20204966495861254, "grad_norm": 0.86476731300354, "learning_rate": 0.0005, "loss": 2.6929197311401367, "step": 2563 }, { "epoch": 0.20212849822625148, "grad_norm": 1.3535422086715698, "learning_rate": 0.0005, "loss": 2.575638771057129, "step": 2564 }, { "epoch": 0.20220733149389042, "grad_norm": 1.0136573314666748, "learning_rate": 0.0005, "loss": 2.5510692596435547, "step": 2565 }, { "epoch": 0.20228616476152936, "grad_norm": 1.721966028213501, "learning_rate": 0.0005, "loss": 2.989952325820923, "step": 2566 }, { "epoch": 0.2023649980291683, "grad_norm": 1.3271865844726562, "learning_rate": 0.0005, "loss": 2.381803512573242, "step": 2567 }, { "epoch": 0.20244383129680726, "grad_norm": 1.0125641822814941, "learning_rate": 0.0005, "loss": 2.3579323291778564, "step": 2568 }, { "epoch": 0.2025226645644462, "grad_norm": 1.0003924369812012, "learning_rate": 0.0005, "loss": 2.6095755100250244, "step": 2569 }, { "epoch": 0.20260149783208514, "grad_norm": 1.068508505821228, "learning_rate": 0.0005, "loss": 2.769784688949585, "step": 2570 }, { "epoch": 0.20268033109972408, "grad_norm": 0.9386250972747803, "learning_rate": 0.0005, "loss": 2.4278807640075684, "step": 2571 }, { "epoch": 0.20275916436736302, "grad_norm": 0.997145414352417, "learning_rate": 0.0005, "loss": 2.482635974884033, "step": 2572 }, { "epoch": 0.20283799763500196, "grad_norm": 0.7689796686172485, "learning_rate": 0.0005, "loss": 2.1990208625793457, "step": 2573 }, { "epoch": 0.20291683090264093, "grad_norm": 1.1524635553359985, "learning_rate": 0.0005, "loss": 2.878643274307251, "step": 2574 }, { "epoch": 0.20299566417027987, "grad_norm": 1.0541030168533325, "learning_rate": 0.0005, "loss": 2.840127944946289, "step": 2575 }, { "epoch": 0.2030744974379188, "grad_norm": 0.958263099193573, "learning_rate": 0.0005, "loss": 2.7340502738952637, "step": 2576 }, { "epoch": 0.20315333070555774, "grad_norm": 1.3349180221557617, "learning_rate": 0.0005, "loss": 2.505471706390381, "step": 2577 }, { "epoch": 0.20323216397319668, "grad_norm": 1.9987167119979858, "learning_rate": 0.0005, "loss": 2.5278635025024414, "step": 2578 }, { "epoch": 0.20331099724083562, "grad_norm": 1.6053487062454224, "learning_rate": 0.0005, "loss": 2.793247699737549, "step": 2579 }, { "epoch": 0.2033898305084746, "grad_norm": 1.1670429706573486, "learning_rate": 0.0005, "loss": 2.5318167209625244, "step": 2580 }, { "epoch": 0.20346866377611353, "grad_norm": 1.0308245420455933, "learning_rate": 0.0005, "loss": 2.4351508617401123, "step": 2581 }, { "epoch": 0.20354749704375247, "grad_norm": 1.1011167764663696, "learning_rate": 0.0005, "loss": 2.4842309951782227, "step": 2582 }, { "epoch": 0.2036263303113914, "grad_norm": 0.7606783509254456, "learning_rate": 0.0005, "loss": 2.5157723426818848, "step": 2583 }, { "epoch": 0.20370516357903035, "grad_norm": 0.7523887157440186, "learning_rate": 0.0005, "loss": 2.556204080581665, "step": 2584 }, { "epoch": 0.20378399684666929, "grad_norm": 0.8294962644577026, "learning_rate": 0.0005, "loss": 2.392996311187744, "step": 2585 }, { "epoch": 0.20386283011430822, "grad_norm": 1.006245493888855, "learning_rate": 0.0005, "loss": 2.485562324523926, "step": 2586 }, { "epoch": 0.2039416633819472, "grad_norm": 0.9166111946105957, "learning_rate": 0.0005, "loss": 2.287966728210449, "step": 2587 }, { "epoch": 0.20402049664958613, "grad_norm": 0.9996941685676575, "learning_rate": 0.0005, "loss": 2.6810827255249023, "step": 2588 }, { "epoch": 0.20409932991722507, "grad_norm": 0.8842905163764954, "learning_rate": 0.0005, "loss": 2.424854278564453, "step": 2589 }, { "epoch": 0.204178163184864, "grad_norm": 1.104830026626587, "learning_rate": 0.0005, "loss": 2.6373114585876465, "step": 2590 }, { "epoch": 0.20425699645250295, "grad_norm": 0.8896054029464722, "learning_rate": 0.0005, "loss": 2.6659960746765137, "step": 2591 }, { "epoch": 0.2043358297201419, "grad_norm": 0.9823319911956787, "learning_rate": 0.0005, "loss": 2.839498996734619, "step": 2592 }, { "epoch": 0.20441466298778085, "grad_norm": 1.0090150833129883, "learning_rate": 0.0005, "loss": 2.402769088745117, "step": 2593 }, { "epoch": 0.2044934962554198, "grad_norm": 1.05147123336792, "learning_rate": 0.0005, "loss": 2.408534526824951, "step": 2594 }, { "epoch": 0.20457232952305873, "grad_norm": 0.9965627193450928, "learning_rate": 0.0005, "loss": 2.195120096206665, "step": 2595 }, { "epoch": 0.20465116279069767, "grad_norm": 0.9152048826217651, "learning_rate": 0.0005, "loss": 2.886568069458008, "step": 2596 }, { "epoch": 0.2047299960583366, "grad_norm": 1.2064365148544312, "learning_rate": 0.0005, "loss": 3.0972208976745605, "step": 2597 }, { "epoch": 0.20480882932597555, "grad_norm": 1.233710765838623, "learning_rate": 0.0005, "loss": 2.5846445560455322, "step": 2598 }, { "epoch": 0.20488766259361452, "grad_norm": 1.9126899242401123, "learning_rate": 0.0005, "loss": 2.4114787578582764, "step": 2599 }, { "epoch": 0.20496649586125346, "grad_norm": 0.6753955483436584, "learning_rate": 0.0005, "loss": 2.265244960784912, "step": 2600 }, { "epoch": 0.2050453291288924, "grad_norm": 0.9572881460189819, "learning_rate": 0.0005, "loss": 2.4803271293640137, "step": 2601 }, { "epoch": 0.20512416239653133, "grad_norm": 0.9540773630142212, "learning_rate": 0.0005, "loss": 2.307316303253174, "step": 2602 }, { "epoch": 0.20520299566417027, "grad_norm": 0.9315903186798096, "learning_rate": 0.0005, "loss": 2.501981735229492, "step": 2603 }, { "epoch": 0.2052818289318092, "grad_norm": 0.9944575428962708, "learning_rate": 0.0005, "loss": 2.790670394897461, "step": 2604 }, { "epoch": 0.20536066219944818, "grad_norm": 1.3383824825286865, "learning_rate": 0.0005, "loss": 2.649780750274658, "step": 2605 }, { "epoch": 0.20543949546708712, "grad_norm": 1.1953445672988892, "learning_rate": 0.0005, "loss": 2.5026583671569824, "step": 2606 }, { "epoch": 0.20551832873472606, "grad_norm": 1.176253318786621, "learning_rate": 0.0005, "loss": 2.726381778717041, "step": 2607 }, { "epoch": 0.205597162002365, "grad_norm": 1.1874897480010986, "learning_rate": 0.0005, "loss": 3.1021740436553955, "step": 2608 }, { "epoch": 0.20567599527000394, "grad_norm": 0.9606746435165405, "learning_rate": 0.0005, "loss": 2.323981285095215, "step": 2609 }, { "epoch": 0.20575482853764288, "grad_norm": 1.1684237718582153, "learning_rate": 0.0005, "loss": 2.694021224975586, "step": 2610 }, { "epoch": 0.20583366180528184, "grad_norm": 0.9704164266586304, "learning_rate": 0.0005, "loss": 2.585374116897583, "step": 2611 }, { "epoch": 0.20591249507292078, "grad_norm": 16.38308334350586, "learning_rate": 0.0005, "loss": 2.647721767425537, "step": 2612 }, { "epoch": 0.20599132834055972, "grad_norm": 1.3099992275238037, "learning_rate": 0.0005, "loss": 2.36411714553833, "step": 2613 }, { "epoch": 0.20607016160819866, "grad_norm": 1.2092262506484985, "learning_rate": 0.0005, "loss": 2.742389440536499, "step": 2614 }, { "epoch": 0.2061489948758376, "grad_norm": 0.9779444336891174, "learning_rate": 0.0005, "loss": 2.9865708351135254, "step": 2615 }, { "epoch": 0.20622782814347654, "grad_norm": 0.8163408041000366, "learning_rate": 0.0005, "loss": 2.4614756107330322, "step": 2616 }, { "epoch": 0.20630666141111548, "grad_norm": 1.0692740678787231, "learning_rate": 0.0005, "loss": 2.7471823692321777, "step": 2617 }, { "epoch": 0.20638549467875444, "grad_norm": 1.0461822748184204, "learning_rate": 0.0005, "loss": 2.903981924057007, "step": 2618 }, { "epoch": 0.20646432794639338, "grad_norm": 0.9836546182632446, "learning_rate": 0.0005, "loss": 2.519232988357544, "step": 2619 }, { "epoch": 0.20654316121403232, "grad_norm": 1.8423031568527222, "learning_rate": 0.0005, "loss": 2.8461804389953613, "step": 2620 }, { "epoch": 0.20662199448167126, "grad_norm": 1.1588701009750366, "learning_rate": 0.0005, "loss": 2.581383228302002, "step": 2621 }, { "epoch": 0.2067008277493102, "grad_norm": 0.9912558794021606, "learning_rate": 0.0005, "loss": 2.37037992477417, "step": 2622 }, { "epoch": 0.20677966101694914, "grad_norm": 0.8936465382575989, "learning_rate": 0.0005, "loss": 2.4949214458465576, "step": 2623 }, { "epoch": 0.2068584942845881, "grad_norm": 1.0305852890014648, "learning_rate": 0.0005, "loss": 2.6123194694519043, "step": 2624 }, { "epoch": 0.20693732755222705, "grad_norm": 1.1145139932632446, "learning_rate": 0.0005, "loss": 2.4383044242858887, "step": 2625 }, { "epoch": 0.20701616081986599, "grad_norm": 0.9150831699371338, "learning_rate": 0.0005, "loss": 2.8378078937530518, "step": 2626 }, { "epoch": 0.20709499408750492, "grad_norm": 1.1289117336273193, "learning_rate": 0.0005, "loss": 2.578462839126587, "step": 2627 }, { "epoch": 0.20717382735514386, "grad_norm": 1.1315722465515137, "learning_rate": 0.0005, "loss": 2.6355555057525635, "step": 2628 }, { "epoch": 0.2072526606227828, "grad_norm": 1.1040221452713013, "learning_rate": 0.0005, "loss": 2.4767162799835205, "step": 2629 }, { "epoch": 0.20733149389042177, "grad_norm": 1.0549529790878296, "learning_rate": 0.0005, "loss": 2.3984477519989014, "step": 2630 }, { "epoch": 0.2074103271580607, "grad_norm": 0.9259430170059204, "learning_rate": 0.0005, "loss": 2.69248628616333, "step": 2631 }, { "epoch": 0.20748916042569965, "grad_norm": 1.0807636976242065, "learning_rate": 0.0005, "loss": 2.5712196826934814, "step": 2632 }, { "epoch": 0.2075679936933386, "grad_norm": 0.894231379032135, "learning_rate": 0.0005, "loss": 2.5753555297851562, "step": 2633 }, { "epoch": 0.20764682696097753, "grad_norm": 0.9763141870498657, "learning_rate": 0.0005, "loss": 2.3520588874816895, "step": 2634 }, { "epoch": 0.20772566022861647, "grad_norm": 1.1333928108215332, "learning_rate": 0.0005, "loss": 3.0035288333892822, "step": 2635 }, { "epoch": 0.20780449349625543, "grad_norm": 1.8245834112167358, "learning_rate": 0.0005, "loss": 2.7840542793273926, "step": 2636 }, { "epoch": 0.20788332676389437, "grad_norm": 0.929559051990509, "learning_rate": 0.0005, "loss": 2.4209537506103516, "step": 2637 }, { "epoch": 0.2079621600315333, "grad_norm": 1.060905933380127, "learning_rate": 0.0005, "loss": 2.8892483711242676, "step": 2638 }, { "epoch": 0.20804099329917225, "grad_norm": 1.0602399110794067, "learning_rate": 0.0005, "loss": 2.648866891860962, "step": 2639 }, { "epoch": 0.2081198265668112, "grad_norm": 1.0444711446762085, "learning_rate": 0.0005, "loss": 2.6956958770751953, "step": 2640 }, { "epoch": 0.20819865983445013, "grad_norm": 1.0360047817230225, "learning_rate": 0.0005, "loss": 2.762674331665039, "step": 2641 }, { "epoch": 0.2082774931020891, "grad_norm": 0.9668130278587341, "learning_rate": 0.0005, "loss": 2.6674492359161377, "step": 2642 }, { "epoch": 0.20835632636972803, "grad_norm": 0.8620204925537109, "learning_rate": 0.0005, "loss": 2.5907135009765625, "step": 2643 }, { "epoch": 0.20843515963736697, "grad_norm": 1.0147007703781128, "learning_rate": 0.0005, "loss": 2.639761447906494, "step": 2644 }, { "epoch": 0.2085139929050059, "grad_norm": 1.0525506734848022, "learning_rate": 0.0005, "loss": 3.104618549346924, "step": 2645 }, { "epoch": 0.20859282617264485, "grad_norm": 0.9044104814529419, "learning_rate": 0.0005, "loss": 2.6506094932556152, "step": 2646 }, { "epoch": 0.2086716594402838, "grad_norm": 0.9249456524848938, "learning_rate": 0.0005, "loss": 2.951294422149658, "step": 2647 }, { "epoch": 0.20875049270792273, "grad_norm": 1.4686166048049927, "learning_rate": 0.0005, "loss": 2.6262221336364746, "step": 2648 }, { "epoch": 0.2088293259755617, "grad_norm": 1.112872838973999, "learning_rate": 0.0005, "loss": 2.4348032474517822, "step": 2649 }, { "epoch": 0.20890815924320064, "grad_norm": 1.3584997653961182, "learning_rate": 0.0005, "loss": 2.3140041828155518, "step": 2650 }, { "epoch": 0.20898699251083958, "grad_norm": 1.9063448905944824, "learning_rate": 0.0005, "loss": 2.8415470123291016, "step": 2651 }, { "epoch": 0.20906582577847851, "grad_norm": 1.2102482318878174, "learning_rate": 0.0005, "loss": 2.716561794281006, "step": 2652 }, { "epoch": 0.20914465904611745, "grad_norm": 1.1273577213287354, "learning_rate": 0.0005, "loss": 2.7610111236572266, "step": 2653 }, { "epoch": 0.2092234923137564, "grad_norm": 0.7750848531723022, "learning_rate": 0.0005, "loss": 2.680234670639038, "step": 2654 }, { "epoch": 0.20930232558139536, "grad_norm": 0.8443723917007446, "learning_rate": 0.0005, "loss": 2.583484172821045, "step": 2655 }, { "epoch": 0.2093811588490343, "grad_norm": 1.1253153085708618, "learning_rate": 0.0005, "loss": 2.410586357116699, "step": 2656 }, { "epoch": 0.20945999211667324, "grad_norm": 0.9535657167434692, "learning_rate": 0.0005, "loss": 2.258803367614746, "step": 2657 }, { "epoch": 0.20953882538431218, "grad_norm": 0.9340612888336182, "learning_rate": 0.0005, "loss": 2.654621124267578, "step": 2658 }, { "epoch": 0.20961765865195112, "grad_norm": 1.128711462020874, "learning_rate": 0.0005, "loss": 2.544597625732422, "step": 2659 }, { "epoch": 0.20969649191959006, "grad_norm": 1.0373684167861938, "learning_rate": 0.0005, "loss": 2.3281779289245605, "step": 2660 }, { "epoch": 0.20977532518722902, "grad_norm": 0.92118239402771, "learning_rate": 0.0005, "loss": 2.7186076641082764, "step": 2661 }, { "epoch": 0.20985415845486796, "grad_norm": 1.3161808252334595, "learning_rate": 0.0005, "loss": 2.7051315307617188, "step": 2662 }, { "epoch": 0.2099329917225069, "grad_norm": 0.6745904684066772, "learning_rate": 0.0005, "loss": 2.5440430641174316, "step": 2663 }, { "epoch": 0.21001182499014584, "grad_norm": 0.8798395395278931, "learning_rate": 0.0005, "loss": 2.511687755584717, "step": 2664 }, { "epoch": 0.21009065825778478, "grad_norm": 1.0483179092407227, "learning_rate": 0.0005, "loss": 2.618379592895508, "step": 2665 }, { "epoch": 0.21016949152542372, "grad_norm": 1.0136886835098267, "learning_rate": 0.0005, "loss": 2.8042678833007812, "step": 2666 }, { "epoch": 0.21024832479306269, "grad_norm": 0.9410913586616516, "learning_rate": 0.0005, "loss": 2.5258445739746094, "step": 2667 }, { "epoch": 0.21032715806070162, "grad_norm": 0.8447031378746033, "learning_rate": 0.0005, "loss": 2.542543411254883, "step": 2668 }, { "epoch": 0.21040599132834056, "grad_norm": 0.8503791093826294, "learning_rate": 0.0005, "loss": 2.6936850547790527, "step": 2669 }, { "epoch": 0.2104848245959795, "grad_norm": 0.9024385809898376, "learning_rate": 0.0005, "loss": 2.7029004096984863, "step": 2670 }, { "epoch": 0.21056365786361844, "grad_norm": 0.9080286622047424, "learning_rate": 0.0005, "loss": 2.6700823307037354, "step": 2671 }, { "epoch": 0.21064249113125738, "grad_norm": 1.0607643127441406, "learning_rate": 0.0005, "loss": 2.859527587890625, "step": 2672 }, { "epoch": 0.21072132439889635, "grad_norm": 1.0724101066589355, "learning_rate": 0.0005, "loss": 2.747743844985962, "step": 2673 }, { "epoch": 0.2108001576665353, "grad_norm": 1.0061780214309692, "learning_rate": 0.0005, "loss": 2.544548749923706, "step": 2674 }, { "epoch": 0.21087899093417423, "grad_norm": 1.303376317024231, "learning_rate": 0.0005, "loss": 2.4961130619049072, "step": 2675 }, { "epoch": 0.21095782420181317, "grad_norm": 0.9873676300048828, "learning_rate": 0.0005, "loss": 2.473243236541748, "step": 2676 }, { "epoch": 0.2110366574694521, "grad_norm": 0.912397563457489, "learning_rate": 0.0005, "loss": 2.5139055252075195, "step": 2677 }, { "epoch": 0.21111549073709104, "grad_norm": 1.0199341773986816, "learning_rate": 0.0005, "loss": 2.723654270172119, "step": 2678 }, { "epoch": 0.21119432400472998, "grad_norm": 0.8851656317710876, "learning_rate": 0.0005, "loss": 2.6846096515655518, "step": 2679 }, { "epoch": 0.21127315727236895, "grad_norm": 0.9910657405853271, "learning_rate": 0.0005, "loss": 3.0474720001220703, "step": 2680 }, { "epoch": 0.2113519905400079, "grad_norm": 1.2084732055664062, "learning_rate": 0.0005, "loss": 2.599503993988037, "step": 2681 }, { "epoch": 0.21143082380764683, "grad_norm": 0.8386585116386414, "learning_rate": 0.0005, "loss": 2.20290207862854, "step": 2682 }, { "epoch": 0.21150965707528577, "grad_norm": 0.8763902187347412, "learning_rate": 0.0005, "loss": 2.837454080581665, "step": 2683 }, { "epoch": 0.2115884903429247, "grad_norm": 0.9054470062255859, "learning_rate": 0.0005, "loss": 2.9508790969848633, "step": 2684 }, { "epoch": 0.21166732361056365, "grad_norm": 0.9000168442726135, "learning_rate": 0.0005, "loss": 2.8333468437194824, "step": 2685 }, { "epoch": 0.2117461568782026, "grad_norm": 1.2881122827529907, "learning_rate": 0.0005, "loss": 2.9246926307678223, "step": 2686 }, { "epoch": 0.21182499014584155, "grad_norm": 0.9100820422172546, "learning_rate": 0.0005, "loss": 2.49348783493042, "step": 2687 }, { "epoch": 0.2119038234134805, "grad_norm": 0.8722614645957947, "learning_rate": 0.0005, "loss": 2.659163236618042, "step": 2688 }, { "epoch": 0.21198265668111943, "grad_norm": 0.9403179287910461, "learning_rate": 0.0005, "loss": 2.513273239135742, "step": 2689 }, { "epoch": 0.21206148994875837, "grad_norm": 1.1141821146011353, "learning_rate": 0.0005, "loss": 2.691650867462158, "step": 2690 }, { "epoch": 0.2121403232163973, "grad_norm": 1.1417608261108398, "learning_rate": 0.0005, "loss": 2.56587553024292, "step": 2691 }, { "epoch": 0.21221915648403628, "grad_norm": 0.9215713739395142, "learning_rate": 0.0005, "loss": 2.49745512008667, "step": 2692 }, { "epoch": 0.21229798975167521, "grad_norm": 2.0072617530822754, "learning_rate": 0.0005, "loss": 2.6907691955566406, "step": 2693 }, { "epoch": 0.21237682301931415, "grad_norm": 1.8662010431289673, "learning_rate": 0.0005, "loss": 2.688735008239746, "step": 2694 }, { "epoch": 0.2124556562869531, "grad_norm": 1.0999537706375122, "learning_rate": 0.0005, "loss": 2.699470043182373, "step": 2695 }, { "epoch": 0.21253448955459203, "grad_norm": 1.0915042161941528, "learning_rate": 0.0005, "loss": 2.839846611022949, "step": 2696 }, { "epoch": 0.21261332282223097, "grad_norm": 0.9133226275444031, "learning_rate": 0.0005, "loss": 2.4259424209594727, "step": 2697 }, { "epoch": 0.21269215608986994, "grad_norm": 0.9287731051445007, "learning_rate": 0.0005, "loss": 2.5044214725494385, "step": 2698 }, { "epoch": 0.21277098935750888, "grad_norm": 0.9279826879501343, "learning_rate": 0.0005, "loss": 2.297562599182129, "step": 2699 }, { "epoch": 0.21284982262514782, "grad_norm": 1.1530632972717285, "learning_rate": 0.0005, "loss": 2.850693464279175, "step": 2700 }, { "epoch": 0.21292865589278676, "grad_norm": 0.9075629711151123, "learning_rate": 0.0005, "loss": 2.2149250507354736, "step": 2701 }, { "epoch": 0.2130074891604257, "grad_norm": 0.9478832483291626, "learning_rate": 0.0005, "loss": 2.4788317680358887, "step": 2702 }, { "epoch": 0.21308632242806463, "grad_norm": 1.078593134880066, "learning_rate": 0.0005, "loss": 2.657578945159912, "step": 2703 }, { "epoch": 0.21316515569570357, "grad_norm": 0.9252405166625977, "learning_rate": 0.0005, "loss": 2.5474557876586914, "step": 2704 }, { "epoch": 0.21324398896334254, "grad_norm": 0.8943978548049927, "learning_rate": 0.0005, "loss": 2.3554868698120117, "step": 2705 }, { "epoch": 0.21332282223098148, "grad_norm": 1.033301591873169, "learning_rate": 0.0005, "loss": 2.5819616317749023, "step": 2706 }, { "epoch": 0.21340165549862042, "grad_norm": 1.220017671585083, "learning_rate": 0.0005, "loss": 2.6150708198547363, "step": 2707 }, { "epoch": 0.21348048876625936, "grad_norm": 0.9306178092956543, "learning_rate": 0.0005, "loss": 2.5214953422546387, "step": 2708 }, { "epoch": 0.2135593220338983, "grad_norm": 0.9818108081817627, "learning_rate": 0.0005, "loss": 2.872243881225586, "step": 2709 }, { "epoch": 0.21363815530153724, "grad_norm": 0.9073784351348877, "learning_rate": 0.0005, "loss": 2.3372058868408203, "step": 2710 }, { "epoch": 0.2137169885691762, "grad_norm": 1.6741091012954712, "learning_rate": 0.0005, "loss": 2.81200909614563, "step": 2711 }, { "epoch": 0.21379582183681514, "grad_norm": 1.011991262435913, "learning_rate": 0.0005, "loss": 2.9123473167419434, "step": 2712 }, { "epoch": 0.21387465510445408, "grad_norm": 0.8965155482292175, "learning_rate": 0.0005, "loss": 2.803464412689209, "step": 2713 }, { "epoch": 0.21395348837209302, "grad_norm": 0.9040055871009827, "learning_rate": 0.0005, "loss": 2.472139596939087, "step": 2714 }, { "epoch": 0.21403232163973196, "grad_norm": 1.4664369821548462, "learning_rate": 0.0005, "loss": 2.709864616394043, "step": 2715 }, { "epoch": 0.2141111549073709, "grad_norm": 1.0178288221359253, "learning_rate": 0.0005, "loss": 2.734363555908203, "step": 2716 }, { "epoch": 0.21418998817500987, "grad_norm": 1.311861276626587, "learning_rate": 0.0005, "loss": 2.693391799926758, "step": 2717 }, { "epoch": 0.2142688214426488, "grad_norm": 1.0237056016921997, "learning_rate": 0.0005, "loss": 2.725980281829834, "step": 2718 }, { "epoch": 0.21434765471028774, "grad_norm": 1.0807859897613525, "learning_rate": 0.0005, "loss": 2.3970251083374023, "step": 2719 }, { "epoch": 0.21442648797792668, "grad_norm": 0.9808926582336426, "learning_rate": 0.0005, "loss": 2.9491686820983887, "step": 2720 }, { "epoch": 0.21450532124556562, "grad_norm": 1.0948083400726318, "learning_rate": 0.0005, "loss": 2.5230557918548584, "step": 2721 }, { "epoch": 0.21458415451320456, "grad_norm": 1.0196788311004639, "learning_rate": 0.0005, "loss": 2.640639305114746, "step": 2722 }, { "epoch": 0.21466298778084353, "grad_norm": 1.216009497642517, "learning_rate": 0.0005, "loss": 2.3810155391693115, "step": 2723 }, { "epoch": 0.21474182104848247, "grad_norm": 0.9505823254585266, "learning_rate": 0.0005, "loss": 2.4749693870544434, "step": 2724 }, { "epoch": 0.2148206543161214, "grad_norm": 1.0862587690353394, "learning_rate": 0.0005, "loss": 2.5430455207824707, "step": 2725 }, { "epoch": 0.21489948758376035, "grad_norm": 0.757129967212677, "learning_rate": 0.0005, "loss": 2.5568900108337402, "step": 2726 }, { "epoch": 0.21497832085139928, "grad_norm": 1.1092370748519897, "learning_rate": 0.0005, "loss": 2.791271924972534, "step": 2727 }, { "epoch": 0.21505715411903822, "grad_norm": 0.803605318069458, "learning_rate": 0.0005, "loss": 2.2895381450653076, "step": 2728 }, { "epoch": 0.2151359873866772, "grad_norm": 0.7812867760658264, "learning_rate": 0.0005, "loss": 2.528779983520508, "step": 2729 }, { "epoch": 0.21521482065431613, "grad_norm": 0.8633988499641418, "learning_rate": 0.0005, "loss": 2.653369903564453, "step": 2730 }, { "epoch": 0.21529365392195507, "grad_norm": 1.5892910957336426, "learning_rate": 0.0005, "loss": 2.7682318687438965, "step": 2731 }, { "epoch": 0.215372487189594, "grad_norm": 1.1090747117996216, "learning_rate": 0.0005, "loss": 2.6553754806518555, "step": 2732 }, { "epoch": 0.21545132045723295, "grad_norm": 0.9473001956939697, "learning_rate": 0.0005, "loss": 2.424773693084717, "step": 2733 }, { "epoch": 0.2155301537248719, "grad_norm": 1.0338646173477173, "learning_rate": 0.0005, "loss": 2.368055820465088, "step": 2734 }, { "epoch": 0.21560898699251083, "grad_norm": 0.8867125511169434, "learning_rate": 0.0005, "loss": 2.4370200634002686, "step": 2735 }, { "epoch": 0.2156878202601498, "grad_norm": 0.9167401194572449, "learning_rate": 0.0005, "loss": 2.5086288452148438, "step": 2736 }, { "epoch": 0.21576665352778873, "grad_norm": 0.8979668617248535, "learning_rate": 0.0005, "loss": 2.824514865875244, "step": 2737 }, { "epoch": 0.21584548679542767, "grad_norm": 1.2151743173599243, "learning_rate": 0.0005, "loss": 2.8923192024230957, "step": 2738 }, { "epoch": 0.2159243200630666, "grad_norm": 0.932807981967926, "learning_rate": 0.0005, "loss": 2.4548935890197754, "step": 2739 }, { "epoch": 0.21600315333070555, "grad_norm": 1.0218065977096558, "learning_rate": 0.0005, "loss": 2.7587623596191406, "step": 2740 }, { "epoch": 0.2160819865983445, "grad_norm": 0.8741108179092407, "learning_rate": 0.0005, "loss": 2.6394107341766357, "step": 2741 }, { "epoch": 0.21616081986598346, "grad_norm": 0.9494861960411072, "learning_rate": 0.0005, "loss": 2.656907081604004, "step": 2742 }, { "epoch": 0.2162396531336224, "grad_norm": 0.8806295394897461, "learning_rate": 0.0005, "loss": 2.835736036300659, "step": 2743 }, { "epoch": 0.21631848640126133, "grad_norm": 1.239945411682129, "learning_rate": 0.0005, "loss": 2.6965723037719727, "step": 2744 }, { "epoch": 0.21639731966890027, "grad_norm": 0.8865392804145813, "learning_rate": 0.0005, "loss": 2.722771167755127, "step": 2745 }, { "epoch": 0.2164761529365392, "grad_norm": 0.975391685962677, "learning_rate": 0.0005, "loss": 2.532341480255127, "step": 2746 }, { "epoch": 0.21655498620417815, "grad_norm": 1.3971644639968872, "learning_rate": 0.0005, "loss": 2.3966264724731445, "step": 2747 }, { "epoch": 0.21663381947181712, "grad_norm": 1.320173978805542, "learning_rate": 0.0005, "loss": 2.7348642349243164, "step": 2748 }, { "epoch": 0.21671265273945606, "grad_norm": 0.9572571516036987, "learning_rate": 0.0005, "loss": 2.536597967147827, "step": 2749 }, { "epoch": 0.216791486007095, "grad_norm": 0.8636579513549805, "learning_rate": 0.0005, "loss": 2.5577516555786133, "step": 2750 }, { "epoch": 0.21687031927473394, "grad_norm": 0.9148141741752625, "learning_rate": 0.0005, "loss": 2.6559066772460938, "step": 2751 }, { "epoch": 0.21694915254237288, "grad_norm": 0.8873129487037659, "learning_rate": 0.0005, "loss": 2.4400687217712402, "step": 2752 }, { "epoch": 0.21702798581001181, "grad_norm": 1.0180203914642334, "learning_rate": 0.0005, "loss": 2.86910343170166, "step": 2753 }, { "epoch": 0.21710681907765078, "grad_norm": 0.7233673930168152, "learning_rate": 0.0005, "loss": 2.2960829734802246, "step": 2754 }, { "epoch": 0.21718565234528972, "grad_norm": 0.9799016118049622, "learning_rate": 0.0005, "loss": 2.4047393798828125, "step": 2755 }, { "epoch": 0.21726448561292866, "grad_norm": 0.9123764038085938, "learning_rate": 0.0005, "loss": 2.4979145526885986, "step": 2756 }, { "epoch": 0.2173433188805676, "grad_norm": 1.0528019666671753, "learning_rate": 0.0005, "loss": 2.4536077976226807, "step": 2757 }, { "epoch": 0.21742215214820654, "grad_norm": 1.0269206762313843, "learning_rate": 0.0005, "loss": 2.5125763416290283, "step": 2758 }, { "epoch": 0.21750098541584548, "grad_norm": 0.8651137351989746, "learning_rate": 0.0005, "loss": 2.570509910583496, "step": 2759 }, { "epoch": 0.21757981868348444, "grad_norm": 0.8033015131950378, "learning_rate": 0.0005, "loss": 2.603238105773926, "step": 2760 }, { "epoch": 0.21765865195112338, "grad_norm": 1.3850680589675903, "learning_rate": 0.0005, "loss": 2.108907699584961, "step": 2761 }, { "epoch": 0.21773748521876232, "grad_norm": 1.0797247886657715, "learning_rate": 0.0005, "loss": 2.6364712715148926, "step": 2762 }, { "epoch": 0.21781631848640126, "grad_norm": 0.9517093896865845, "learning_rate": 0.0005, "loss": 2.2734131813049316, "step": 2763 }, { "epoch": 0.2178951517540402, "grad_norm": 0.9176563024520874, "learning_rate": 0.0005, "loss": 2.3845129013061523, "step": 2764 }, { "epoch": 0.21797398502167914, "grad_norm": 1.0162320137023926, "learning_rate": 0.0005, "loss": 2.7621166706085205, "step": 2765 }, { "epoch": 0.21805281828931808, "grad_norm": 1.4326390027999878, "learning_rate": 0.0005, "loss": 2.343047618865967, "step": 2766 }, { "epoch": 0.21813165155695705, "grad_norm": 1.2254387140274048, "learning_rate": 0.0005, "loss": 2.674630641937256, "step": 2767 }, { "epoch": 0.21821048482459598, "grad_norm": 1.1655421257019043, "learning_rate": 0.0005, "loss": 2.348212718963623, "step": 2768 }, { "epoch": 0.21828931809223492, "grad_norm": 1.0240718126296997, "learning_rate": 0.0005, "loss": 2.491443157196045, "step": 2769 }, { "epoch": 0.21836815135987386, "grad_norm": 0.8894004225730896, "learning_rate": 0.0005, "loss": 2.5699539184570312, "step": 2770 }, { "epoch": 0.2184469846275128, "grad_norm": 1.0692381858825684, "learning_rate": 0.0005, "loss": 2.860975742340088, "step": 2771 }, { "epoch": 0.21852581789515174, "grad_norm": 0.849958598613739, "learning_rate": 0.0005, "loss": 3.041282892227173, "step": 2772 }, { "epoch": 0.2186046511627907, "grad_norm": 0.8841503858566284, "learning_rate": 0.0005, "loss": 2.3603363037109375, "step": 2773 }, { "epoch": 0.21868348443042965, "grad_norm": 0.9201577305793762, "learning_rate": 0.0005, "loss": 2.4005212783813477, "step": 2774 }, { "epoch": 0.2187623176980686, "grad_norm": 1.9037612676620483, "learning_rate": 0.0005, "loss": 2.429389476776123, "step": 2775 }, { "epoch": 0.21884115096570753, "grad_norm": 1.1333096027374268, "learning_rate": 0.0005, "loss": 2.5073256492614746, "step": 2776 }, { "epoch": 0.21891998423334647, "grad_norm": 0.970121443271637, "learning_rate": 0.0005, "loss": 2.694934844970703, "step": 2777 }, { "epoch": 0.2189988175009854, "grad_norm": 0.8086928725242615, "learning_rate": 0.0005, "loss": 2.260491371154785, "step": 2778 }, { "epoch": 0.21907765076862437, "grad_norm": 0.7874186038970947, "learning_rate": 0.0005, "loss": 2.052989959716797, "step": 2779 }, { "epoch": 0.2191564840362633, "grad_norm": 1.00504469871521, "learning_rate": 0.0005, "loss": 2.479997396469116, "step": 2780 }, { "epoch": 0.21923531730390225, "grad_norm": 0.9120787382125854, "learning_rate": 0.0005, "loss": 2.6929640769958496, "step": 2781 }, { "epoch": 0.2193141505715412, "grad_norm": 0.9065794348716736, "learning_rate": 0.0005, "loss": 2.7689528465270996, "step": 2782 }, { "epoch": 0.21939298383918013, "grad_norm": 0.92899489402771, "learning_rate": 0.0005, "loss": 2.5684309005737305, "step": 2783 }, { "epoch": 0.21947181710681907, "grad_norm": 1.24668288230896, "learning_rate": 0.0005, "loss": 2.6535418033599854, "step": 2784 }, { "epoch": 0.21955065037445803, "grad_norm": 0.8511629104614258, "learning_rate": 0.0005, "loss": 2.4051575660705566, "step": 2785 }, { "epoch": 0.21962948364209697, "grad_norm": 1.005449652671814, "learning_rate": 0.0005, "loss": 2.659554958343506, "step": 2786 }, { "epoch": 0.2197083169097359, "grad_norm": 1.1603400707244873, "learning_rate": 0.0005, "loss": 2.417928695678711, "step": 2787 }, { "epoch": 0.21978715017737485, "grad_norm": 0.9767547249794006, "learning_rate": 0.0005, "loss": 2.5767228603363037, "step": 2788 }, { "epoch": 0.2198659834450138, "grad_norm": 1.136231541633606, "learning_rate": 0.0005, "loss": 2.7708635330200195, "step": 2789 }, { "epoch": 0.21994481671265273, "grad_norm": 1.2634700536727905, "learning_rate": 0.0005, "loss": 2.7567355632781982, "step": 2790 }, { "epoch": 0.2200236499802917, "grad_norm": 0.8930370211601257, "learning_rate": 0.0005, "loss": 2.3340563774108887, "step": 2791 }, { "epoch": 0.22010248324793064, "grad_norm": 1.0998014211654663, "learning_rate": 0.0005, "loss": 2.789428234100342, "step": 2792 }, { "epoch": 0.22018131651556958, "grad_norm": 1.0522228479385376, "learning_rate": 0.0005, "loss": 2.785378932952881, "step": 2793 }, { "epoch": 0.22026014978320851, "grad_norm": 1.3768069744110107, "learning_rate": 0.0005, "loss": 2.5046093463897705, "step": 2794 }, { "epoch": 0.22033898305084745, "grad_norm": 0.852019727230072, "learning_rate": 0.0005, "loss": 2.5624351501464844, "step": 2795 }, { "epoch": 0.2204178163184864, "grad_norm": 1.054810881614685, "learning_rate": 0.0005, "loss": 2.5463380813598633, "step": 2796 }, { "epoch": 0.22049664958612533, "grad_norm": 0.9006810784339905, "learning_rate": 0.0005, "loss": 2.384793519973755, "step": 2797 }, { "epoch": 0.2205754828537643, "grad_norm": 1.059069037437439, "learning_rate": 0.0005, "loss": 2.4101686477661133, "step": 2798 }, { "epoch": 0.22065431612140324, "grad_norm": 1.4362033605575562, "learning_rate": 0.0005, "loss": 2.5726394653320312, "step": 2799 }, { "epoch": 0.22073314938904218, "grad_norm": 0.8790398836135864, "learning_rate": 0.0005, "loss": 2.5703988075256348, "step": 2800 }, { "epoch": 0.22081198265668112, "grad_norm": 0.8204402327537537, "learning_rate": 0.0005, "loss": 2.6832962036132812, "step": 2801 }, { "epoch": 0.22089081592432006, "grad_norm": 0.8566075563430786, "learning_rate": 0.0005, "loss": 2.442162036895752, "step": 2802 }, { "epoch": 0.220969649191959, "grad_norm": 1.0057934522628784, "learning_rate": 0.0005, "loss": 2.6527292728424072, "step": 2803 }, { "epoch": 0.22104848245959796, "grad_norm": 0.9987114667892456, "learning_rate": 0.0005, "loss": 2.760533332824707, "step": 2804 }, { "epoch": 0.2211273157272369, "grad_norm": 1.0057340860366821, "learning_rate": 0.0005, "loss": 2.627711534500122, "step": 2805 }, { "epoch": 0.22120614899487584, "grad_norm": 0.9504619240760803, "learning_rate": 0.0005, "loss": 2.444096565246582, "step": 2806 }, { "epoch": 0.22128498226251478, "grad_norm": 0.9426950216293335, "learning_rate": 0.0005, "loss": 2.526477336883545, "step": 2807 }, { "epoch": 0.22136381553015372, "grad_norm": 0.9076807498931885, "learning_rate": 0.0005, "loss": 2.478039264678955, "step": 2808 }, { "epoch": 0.22144264879779266, "grad_norm": 1.030052900314331, "learning_rate": 0.0005, "loss": 2.7361605167388916, "step": 2809 }, { "epoch": 0.22152148206543162, "grad_norm": 1.0792882442474365, "learning_rate": 0.0005, "loss": 2.379031181335449, "step": 2810 }, { "epoch": 0.22160031533307056, "grad_norm": 0.7821901440620422, "learning_rate": 0.0005, "loss": 2.8105340003967285, "step": 2811 }, { "epoch": 0.2216791486007095, "grad_norm": 0.882205605506897, "learning_rate": 0.0005, "loss": 2.324171304702759, "step": 2812 }, { "epoch": 0.22175798186834844, "grad_norm": 0.7155893445014954, "learning_rate": 0.0005, "loss": 2.1776251792907715, "step": 2813 }, { "epoch": 0.22183681513598738, "grad_norm": 0.7509416341781616, "learning_rate": 0.0005, "loss": 2.348184585571289, "step": 2814 }, { "epoch": 0.22191564840362632, "grad_norm": 0.8969791531562805, "learning_rate": 0.0005, "loss": 2.645789623260498, "step": 2815 }, { "epoch": 0.2219944816712653, "grad_norm": 0.9888436794281006, "learning_rate": 0.0005, "loss": 2.3052029609680176, "step": 2816 }, { "epoch": 0.22207331493890423, "grad_norm": 1.0784369707107544, "learning_rate": 0.0005, "loss": 2.3889498710632324, "step": 2817 }, { "epoch": 0.22215214820654317, "grad_norm": 0.8771836161613464, "learning_rate": 0.0005, "loss": 2.507190704345703, "step": 2818 }, { "epoch": 0.2222309814741821, "grad_norm": 0.9264580607414246, "learning_rate": 0.0005, "loss": 2.5758161544799805, "step": 2819 }, { "epoch": 0.22230981474182104, "grad_norm": 0.8917271494865417, "learning_rate": 0.0005, "loss": 2.558854579925537, "step": 2820 }, { "epoch": 0.22238864800945998, "grad_norm": 0.8942017555236816, "learning_rate": 0.0005, "loss": 2.334484577178955, "step": 2821 }, { "epoch": 0.22246748127709895, "grad_norm": 0.9056081175804138, "learning_rate": 0.0005, "loss": 2.7884299755096436, "step": 2822 }, { "epoch": 0.2225463145447379, "grad_norm": 0.8282443881034851, "learning_rate": 0.0005, "loss": 2.6113600730895996, "step": 2823 }, { "epoch": 0.22262514781237683, "grad_norm": 0.8485841155052185, "learning_rate": 0.0005, "loss": 2.1056315898895264, "step": 2824 }, { "epoch": 0.22270398108001577, "grad_norm": 0.8956524729728699, "learning_rate": 0.0005, "loss": 2.646664619445801, "step": 2825 }, { "epoch": 0.2227828143476547, "grad_norm": 0.989365816116333, "learning_rate": 0.0005, "loss": 2.477468967437744, "step": 2826 }, { "epoch": 0.22286164761529365, "grad_norm": 1.2564367055892944, "learning_rate": 0.0005, "loss": 2.1970157623291016, "step": 2827 }, { "epoch": 0.22294048088293258, "grad_norm": 1.1421643495559692, "learning_rate": 0.0005, "loss": 2.489833354949951, "step": 2828 }, { "epoch": 0.22301931415057155, "grad_norm": 1.18460214138031, "learning_rate": 0.0005, "loss": 2.8119726181030273, "step": 2829 }, { "epoch": 0.2230981474182105, "grad_norm": 1.1422691345214844, "learning_rate": 0.0005, "loss": 2.4720637798309326, "step": 2830 }, { "epoch": 0.22317698068584943, "grad_norm": 0.83875972032547, "learning_rate": 0.0005, "loss": 2.547945976257324, "step": 2831 }, { "epoch": 0.22325581395348837, "grad_norm": 1.1380670070648193, "learning_rate": 0.0005, "loss": 2.747647285461426, "step": 2832 }, { "epoch": 0.2233346472211273, "grad_norm": 1.0210691690444946, "learning_rate": 0.0005, "loss": 2.81301212310791, "step": 2833 }, { "epoch": 0.22341348048876625, "grad_norm": 1.0659152269363403, "learning_rate": 0.0005, "loss": 2.5637893676757812, "step": 2834 }, { "epoch": 0.22349231375640521, "grad_norm": 0.8876428008079529, "learning_rate": 0.0005, "loss": 2.2767930030822754, "step": 2835 }, { "epoch": 0.22357114702404415, "grad_norm": 0.9723746180534363, "learning_rate": 0.0005, "loss": 2.67299747467041, "step": 2836 }, { "epoch": 0.2236499802916831, "grad_norm": 0.9448846578598022, "learning_rate": 0.0005, "loss": 2.7802059650421143, "step": 2837 }, { "epoch": 0.22372881355932203, "grad_norm": 1.223273515701294, "learning_rate": 0.0005, "loss": 2.3410744667053223, "step": 2838 }, { "epoch": 0.22380764682696097, "grad_norm": 0.9648638367652893, "learning_rate": 0.0005, "loss": 2.3617353439331055, "step": 2839 }, { "epoch": 0.2238864800945999, "grad_norm": 0.9322092533111572, "learning_rate": 0.0005, "loss": 2.082376718521118, "step": 2840 }, { "epoch": 0.22396531336223888, "grad_norm": 1.0928418636322021, "learning_rate": 0.0005, "loss": 2.4021987915039062, "step": 2841 }, { "epoch": 0.22404414662987782, "grad_norm": 0.9476019144058228, "learning_rate": 0.0005, "loss": 2.673806667327881, "step": 2842 }, { "epoch": 0.22412297989751676, "grad_norm": 1.0894641876220703, "learning_rate": 0.0005, "loss": 2.779736042022705, "step": 2843 }, { "epoch": 0.2242018131651557, "grad_norm": 0.8732802271842957, "learning_rate": 0.0005, "loss": 2.4367127418518066, "step": 2844 }, { "epoch": 0.22428064643279463, "grad_norm": 0.9831297993659973, "learning_rate": 0.0005, "loss": 2.2521495819091797, "step": 2845 }, { "epoch": 0.22435947970043357, "grad_norm": 0.8538446426391602, "learning_rate": 0.0005, "loss": 2.361696243286133, "step": 2846 }, { "epoch": 0.22443831296807254, "grad_norm": 1.0853627920150757, "learning_rate": 0.0005, "loss": 2.6773486137390137, "step": 2847 }, { "epoch": 0.22451714623571148, "grad_norm": 0.8158556818962097, "learning_rate": 0.0005, "loss": 2.4159750938415527, "step": 2848 }, { "epoch": 0.22459597950335042, "grad_norm": 0.7815194129943848, "learning_rate": 0.0005, "loss": 2.569129467010498, "step": 2849 }, { "epoch": 0.22467481277098936, "grad_norm": 0.8654152154922485, "learning_rate": 0.0005, "loss": 2.6571428775787354, "step": 2850 }, { "epoch": 0.2247536460386283, "grad_norm": 1.0135295391082764, "learning_rate": 0.0005, "loss": 2.488940477371216, "step": 2851 }, { "epoch": 0.22483247930626724, "grad_norm": 1.0245975255966187, "learning_rate": 0.0005, "loss": 2.9109268188476562, "step": 2852 }, { "epoch": 0.22491131257390617, "grad_norm": 1.000396490097046, "learning_rate": 0.0005, "loss": 2.5794565677642822, "step": 2853 }, { "epoch": 0.22499014584154514, "grad_norm": 0.911952793598175, "learning_rate": 0.0005, "loss": 2.7611918449401855, "step": 2854 }, { "epoch": 0.22506897910918408, "grad_norm": 0.9638902544975281, "learning_rate": 0.0005, "loss": 2.469231605529785, "step": 2855 }, { "epoch": 0.22514781237682302, "grad_norm": 1.1202555894851685, "learning_rate": 0.0005, "loss": 2.674229383468628, "step": 2856 }, { "epoch": 0.22522664564446196, "grad_norm": 0.8452120423316956, "learning_rate": 0.0005, "loss": 2.2972264289855957, "step": 2857 }, { "epoch": 0.2253054789121009, "grad_norm": 1.0469905138015747, "learning_rate": 0.0005, "loss": 2.4123520851135254, "step": 2858 }, { "epoch": 0.22538431217973984, "grad_norm": 0.8914953470230103, "learning_rate": 0.0005, "loss": 2.277949810028076, "step": 2859 }, { "epoch": 0.2254631454473788, "grad_norm": 1.1936904191970825, "learning_rate": 0.0005, "loss": 2.426039218902588, "step": 2860 }, { "epoch": 0.22554197871501774, "grad_norm": 1.2473787069320679, "learning_rate": 0.0005, "loss": 2.6553738117218018, "step": 2861 }, { "epoch": 0.22562081198265668, "grad_norm": 0.9891795516014099, "learning_rate": 0.0005, "loss": 2.498706579208374, "step": 2862 }, { "epoch": 0.22569964525029562, "grad_norm": 1.2751052379608154, "learning_rate": 0.0005, "loss": 2.40859317779541, "step": 2863 }, { "epoch": 0.22577847851793456, "grad_norm": 2.5063908100128174, "learning_rate": 0.0005, "loss": 2.8806381225585938, "step": 2864 }, { "epoch": 0.2258573117855735, "grad_norm": 1.1159344911575317, "learning_rate": 0.0005, "loss": 2.4240403175354004, "step": 2865 }, { "epoch": 0.22593614505321247, "grad_norm": 0.8630715608596802, "learning_rate": 0.0005, "loss": 2.209095001220703, "step": 2866 }, { "epoch": 0.2260149783208514, "grad_norm": 0.9937324523925781, "learning_rate": 0.0005, "loss": 2.407817840576172, "step": 2867 }, { "epoch": 0.22609381158849035, "grad_norm": 1.089374303817749, "learning_rate": 0.0005, "loss": 2.3816418647766113, "step": 2868 }, { "epoch": 0.22617264485612928, "grad_norm": 0.958264172077179, "learning_rate": 0.0005, "loss": 2.339756727218628, "step": 2869 }, { "epoch": 0.22625147812376822, "grad_norm": 0.7790486216545105, "learning_rate": 0.0005, "loss": 2.578864812850952, "step": 2870 }, { "epoch": 0.22633031139140716, "grad_norm": 1.5987411737442017, "learning_rate": 0.0005, "loss": 2.287208080291748, "step": 2871 }, { "epoch": 0.22640914465904613, "grad_norm": 0.8423597812652588, "learning_rate": 0.0005, "loss": 2.2809817790985107, "step": 2872 }, { "epoch": 0.22648797792668507, "grad_norm": 1.0262877941131592, "learning_rate": 0.0005, "loss": 2.2664666175842285, "step": 2873 }, { "epoch": 0.226566811194324, "grad_norm": 1.0235874652862549, "learning_rate": 0.0005, "loss": 2.316426992416382, "step": 2874 }, { "epoch": 0.22664564446196295, "grad_norm": 0.8005830645561218, "learning_rate": 0.0005, "loss": 2.2634477615356445, "step": 2875 }, { "epoch": 0.2267244777296019, "grad_norm": 0.9689998030662537, "learning_rate": 0.0005, "loss": 2.37149715423584, "step": 2876 }, { "epoch": 0.22680331099724083, "grad_norm": 1.0274394750595093, "learning_rate": 0.0005, "loss": 2.521000385284424, "step": 2877 }, { "epoch": 0.2268821442648798, "grad_norm": 1.076651692390442, "learning_rate": 0.0005, "loss": 2.612389087677002, "step": 2878 }, { "epoch": 0.22696097753251873, "grad_norm": 0.7880423069000244, "learning_rate": 0.0005, "loss": 2.365028142929077, "step": 2879 }, { "epoch": 0.22703981080015767, "grad_norm": 0.8524520993232727, "learning_rate": 0.0005, "loss": 2.400477886199951, "step": 2880 }, { "epoch": 0.2271186440677966, "grad_norm": 1.1230252981185913, "learning_rate": 0.0005, "loss": 2.559579849243164, "step": 2881 }, { "epoch": 0.22719747733543555, "grad_norm": 0.7612135410308838, "learning_rate": 0.0005, "loss": 2.4248597621917725, "step": 2882 }, { "epoch": 0.2272763106030745, "grad_norm": 1.021360158920288, "learning_rate": 0.0005, "loss": 2.4588334560394287, "step": 2883 }, { "epoch": 0.22735514387071343, "grad_norm": 1.2954307794570923, "learning_rate": 0.0005, "loss": 2.687699317932129, "step": 2884 }, { "epoch": 0.2274339771383524, "grad_norm": 0.8451679944992065, "learning_rate": 0.0005, "loss": 2.4504308700561523, "step": 2885 }, { "epoch": 0.22751281040599133, "grad_norm": 0.9691913723945618, "learning_rate": 0.0005, "loss": 3.0330915451049805, "step": 2886 }, { "epoch": 0.22759164367363027, "grad_norm": 0.8438127040863037, "learning_rate": 0.0005, "loss": 2.462635040283203, "step": 2887 }, { "epoch": 0.2276704769412692, "grad_norm": 0.8722970485687256, "learning_rate": 0.0005, "loss": 2.317995548248291, "step": 2888 }, { "epoch": 0.22774931020890815, "grad_norm": 1.0386340618133545, "learning_rate": 0.0005, "loss": 2.5500857830047607, "step": 2889 }, { "epoch": 0.2278281434765471, "grad_norm": 0.862892210483551, "learning_rate": 0.0005, "loss": 2.5780019760131836, "step": 2890 }, { "epoch": 0.22790697674418606, "grad_norm": 1.7895020246505737, "learning_rate": 0.0005, "loss": 2.3706092834472656, "step": 2891 }, { "epoch": 0.227985810011825, "grad_norm": 0.7879380583763123, "learning_rate": 0.0005, "loss": 2.77860164642334, "step": 2892 }, { "epoch": 0.22806464327946394, "grad_norm": 0.9000465869903564, "learning_rate": 0.0005, "loss": 2.7138819694519043, "step": 2893 }, { "epoch": 0.22814347654710287, "grad_norm": 1.0416632890701294, "learning_rate": 0.0005, "loss": 2.252229690551758, "step": 2894 }, { "epoch": 0.22822230981474181, "grad_norm": 0.9983274936676025, "learning_rate": 0.0005, "loss": 2.930561065673828, "step": 2895 }, { "epoch": 0.22830114308238075, "grad_norm": 1.2127203941345215, "learning_rate": 0.0005, "loss": 2.3488738536834717, "step": 2896 }, { "epoch": 0.22837997635001972, "grad_norm": 0.8615753650665283, "learning_rate": 0.0005, "loss": 2.224242925643921, "step": 2897 }, { "epoch": 0.22845880961765866, "grad_norm": 1.1066585779190063, "learning_rate": 0.0005, "loss": 2.360544204711914, "step": 2898 }, { "epoch": 0.2285376428852976, "grad_norm": 0.754987895488739, "learning_rate": 0.0005, "loss": 2.5566580295562744, "step": 2899 }, { "epoch": 0.22861647615293654, "grad_norm": 0.9671763181686401, "learning_rate": 0.0005, "loss": 2.433199405670166, "step": 2900 }, { "epoch": 0.22869530942057548, "grad_norm": 1.4503686428070068, "learning_rate": 0.0005, "loss": 2.221158742904663, "step": 2901 }, { "epoch": 0.22877414268821442, "grad_norm": 1.0906238555908203, "learning_rate": 0.0005, "loss": 2.5120043754577637, "step": 2902 }, { "epoch": 0.22885297595585338, "grad_norm": 0.8781293630599976, "learning_rate": 0.0005, "loss": 2.9359517097473145, "step": 2903 }, { "epoch": 0.22893180922349232, "grad_norm": 1.1646674871444702, "learning_rate": 0.0005, "loss": 2.6150410175323486, "step": 2904 }, { "epoch": 0.22901064249113126, "grad_norm": 0.9192079305648804, "learning_rate": 0.0005, "loss": 2.5195631980895996, "step": 2905 }, { "epoch": 0.2290894757587702, "grad_norm": 1.1354573965072632, "learning_rate": 0.0005, "loss": 2.7216362953186035, "step": 2906 }, { "epoch": 0.22916830902640914, "grad_norm": 0.8944800496101379, "learning_rate": 0.0005, "loss": 2.2204055786132812, "step": 2907 }, { "epoch": 0.22924714229404808, "grad_norm": 1.32254958152771, "learning_rate": 0.0005, "loss": 2.4612066745758057, "step": 2908 }, { "epoch": 0.22932597556168705, "grad_norm": 0.9761051535606384, "learning_rate": 0.0005, "loss": 2.7157540321350098, "step": 2909 }, { "epoch": 0.22940480882932598, "grad_norm": 1.1499606370925903, "learning_rate": 0.0005, "loss": 2.823911190032959, "step": 2910 }, { "epoch": 0.22948364209696492, "grad_norm": 1.1266422271728516, "learning_rate": 0.0005, "loss": 2.4121100902557373, "step": 2911 }, { "epoch": 0.22956247536460386, "grad_norm": 2.121929407119751, "learning_rate": 0.0005, "loss": 2.5877673625946045, "step": 2912 }, { "epoch": 0.2296413086322428, "grad_norm": 1.2410590648651123, "learning_rate": 0.0005, "loss": 2.7484145164489746, "step": 2913 }, { "epoch": 0.22972014189988174, "grad_norm": 1.1129757165908813, "learning_rate": 0.0005, "loss": 2.2772698402404785, "step": 2914 }, { "epoch": 0.22979897516752068, "grad_norm": 0.9886347055435181, "learning_rate": 0.0005, "loss": 2.5276741981506348, "step": 2915 }, { "epoch": 0.22987780843515965, "grad_norm": 0.7469438314437866, "learning_rate": 0.0005, "loss": 2.3968911170959473, "step": 2916 }, { "epoch": 0.2299566417027986, "grad_norm": 1.292676568031311, "learning_rate": 0.0005, "loss": 2.3202147483825684, "step": 2917 }, { "epoch": 0.23003547497043753, "grad_norm": 0.9192126989364624, "learning_rate": 0.0005, "loss": 2.747495174407959, "step": 2918 }, { "epoch": 0.23011430823807646, "grad_norm": 0.8706246018409729, "learning_rate": 0.0005, "loss": 2.3571527004241943, "step": 2919 }, { "epoch": 0.2301931415057154, "grad_norm": 1.0152428150177002, "learning_rate": 0.0005, "loss": 2.689870834350586, "step": 2920 }, { "epoch": 0.23027197477335434, "grad_norm": 1.06712806224823, "learning_rate": 0.0005, "loss": 2.500936985015869, "step": 2921 }, { "epoch": 0.2303508080409933, "grad_norm": 1.2392020225524902, "learning_rate": 0.0005, "loss": 2.6824049949645996, "step": 2922 }, { "epoch": 0.23042964130863225, "grad_norm": 0.8655687570571899, "learning_rate": 0.0005, "loss": 2.787334680557251, "step": 2923 }, { "epoch": 0.2305084745762712, "grad_norm": 1.0234105587005615, "learning_rate": 0.0005, "loss": 2.414081573486328, "step": 2924 }, { "epoch": 0.23058730784391013, "grad_norm": 0.9997438192367554, "learning_rate": 0.0005, "loss": 2.266005277633667, "step": 2925 }, { "epoch": 0.23066614111154907, "grad_norm": 1.0907572507858276, "learning_rate": 0.0005, "loss": 2.6631758213043213, "step": 2926 }, { "epoch": 0.230744974379188, "grad_norm": 0.9209237098693848, "learning_rate": 0.0005, "loss": 2.788844108581543, "step": 2927 }, { "epoch": 0.23082380764682697, "grad_norm": 0.972862720489502, "learning_rate": 0.0005, "loss": 2.7391090393066406, "step": 2928 }, { "epoch": 0.2309026409144659, "grad_norm": 0.7759242057800293, "learning_rate": 0.0005, "loss": 2.397922992706299, "step": 2929 }, { "epoch": 0.23098147418210485, "grad_norm": 0.9466118812561035, "learning_rate": 0.0005, "loss": 2.5105016231536865, "step": 2930 }, { "epoch": 0.2310603074497438, "grad_norm": 0.8757018446922302, "learning_rate": 0.0005, "loss": 2.2836692333221436, "step": 2931 }, { "epoch": 0.23113914071738273, "grad_norm": 0.8932138681411743, "learning_rate": 0.0005, "loss": 2.4467453956604004, "step": 2932 }, { "epoch": 0.23121797398502167, "grad_norm": 0.9626989364624023, "learning_rate": 0.0005, "loss": 2.448535203933716, "step": 2933 }, { "epoch": 0.23129680725266064, "grad_norm": 0.9825190901756287, "learning_rate": 0.0005, "loss": 2.289653778076172, "step": 2934 }, { "epoch": 0.23137564052029957, "grad_norm": 0.9809626936912537, "learning_rate": 0.0005, "loss": 2.2370071411132812, "step": 2935 }, { "epoch": 0.23145447378793851, "grad_norm": 1.0229626893997192, "learning_rate": 0.0005, "loss": 2.3688793182373047, "step": 2936 }, { "epoch": 0.23153330705557745, "grad_norm": 0.9337136745452881, "learning_rate": 0.0005, "loss": 2.255553722381592, "step": 2937 }, { "epoch": 0.2316121403232164, "grad_norm": 0.9206022024154663, "learning_rate": 0.0005, "loss": 2.15679669380188, "step": 2938 }, { "epoch": 0.23169097359085533, "grad_norm": 0.9496920108795166, "learning_rate": 0.0005, "loss": 2.4803431034088135, "step": 2939 }, { "epoch": 0.2317698068584943, "grad_norm": 1.0226902961730957, "learning_rate": 0.0005, "loss": 2.3623404502868652, "step": 2940 }, { "epoch": 0.23184864012613324, "grad_norm": 0.9631060361862183, "learning_rate": 0.0005, "loss": 2.6307179927825928, "step": 2941 }, { "epoch": 0.23192747339377218, "grad_norm": 0.9769289493560791, "learning_rate": 0.0005, "loss": 2.6286849975585938, "step": 2942 }, { "epoch": 0.23200630666141112, "grad_norm": 1.0468116998672485, "learning_rate": 0.0005, "loss": 2.609145402908325, "step": 2943 }, { "epoch": 0.23208513992905005, "grad_norm": 0.9732962846755981, "learning_rate": 0.0005, "loss": 2.4856796264648438, "step": 2944 }, { "epoch": 0.232163973196689, "grad_norm": 1.1225557327270508, "learning_rate": 0.0005, "loss": 2.9243569374084473, "step": 2945 }, { "epoch": 0.23224280646432793, "grad_norm": 1.0692838430404663, "learning_rate": 0.0005, "loss": 2.6384599208831787, "step": 2946 }, { "epoch": 0.2323216397319669, "grad_norm": 0.9118652939796448, "learning_rate": 0.0005, "loss": 2.7654268741607666, "step": 2947 }, { "epoch": 0.23240047299960584, "grad_norm": 2.3213753700256348, "learning_rate": 0.0005, "loss": 2.450589418411255, "step": 2948 }, { "epoch": 0.23247930626724478, "grad_norm": 1.2609648704528809, "learning_rate": 0.0005, "loss": 2.447934150695801, "step": 2949 }, { "epoch": 0.23255813953488372, "grad_norm": 1.01270592212677, "learning_rate": 0.0005, "loss": 2.63096284866333, "step": 2950 }, { "epoch": 0.23263697280252266, "grad_norm": 1.1128615140914917, "learning_rate": 0.0005, "loss": 2.676971435546875, "step": 2951 }, { "epoch": 0.2327158060701616, "grad_norm": 0.8167518973350525, "learning_rate": 0.0005, "loss": 2.532227039337158, "step": 2952 }, { "epoch": 0.23279463933780056, "grad_norm": 1.0336048603057861, "learning_rate": 0.0005, "loss": 2.469935655593872, "step": 2953 }, { "epoch": 0.2328734726054395, "grad_norm": 1.167673110961914, "learning_rate": 0.0005, "loss": 2.708339214324951, "step": 2954 }, { "epoch": 0.23295230587307844, "grad_norm": 0.8829301595687866, "learning_rate": 0.0005, "loss": 2.5666003227233887, "step": 2955 }, { "epoch": 0.23303113914071738, "grad_norm": 1.1245981454849243, "learning_rate": 0.0005, "loss": 2.3090736865997314, "step": 2956 }, { "epoch": 0.23310997240835632, "grad_norm": 0.955618143081665, "learning_rate": 0.0005, "loss": 2.5988152027130127, "step": 2957 }, { "epoch": 0.23318880567599526, "grad_norm": 0.7830253839492798, "learning_rate": 0.0005, "loss": 2.578111171722412, "step": 2958 }, { "epoch": 0.23326763894363423, "grad_norm": 0.8158147931098938, "learning_rate": 0.0005, "loss": 2.4715664386749268, "step": 2959 }, { "epoch": 0.23334647221127316, "grad_norm": 0.8228363394737244, "learning_rate": 0.0005, "loss": 2.7148678302764893, "step": 2960 }, { "epoch": 0.2334253054789121, "grad_norm": 1.0601972341537476, "learning_rate": 0.0005, "loss": 2.699653387069702, "step": 2961 }, { "epoch": 0.23350413874655104, "grad_norm": 0.7830919027328491, "learning_rate": 0.0005, "loss": 2.8088855743408203, "step": 2962 }, { "epoch": 0.23358297201418998, "grad_norm": 0.9723523855209351, "learning_rate": 0.0005, "loss": 2.6135265827178955, "step": 2963 }, { "epoch": 0.23366180528182892, "grad_norm": 1.027292013168335, "learning_rate": 0.0005, "loss": 2.849309206008911, "step": 2964 }, { "epoch": 0.2337406385494679, "grad_norm": 1.042324185371399, "learning_rate": 0.0005, "loss": 2.57220458984375, "step": 2965 }, { "epoch": 0.23381947181710683, "grad_norm": 0.8696467876434326, "learning_rate": 0.0005, "loss": 2.6623153686523438, "step": 2966 }, { "epoch": 0.23389830508474577, "grad_norm": 0.9269552230834961, "learning_rate": 0.0005, "loss": 2.419114351272583, "step": 2967 }, { "epoch": 0.2339771383523847, "grad_norm": 1.0591951608657837, "learning_rate": 0.0005, "loss": 2.658426523208618, "step": 2968 }, { "epoch": 0.23405597162002365, "grad_norm": 2.19917368888855, "learning_rate": 0.0005, "loss": 2.761119842529297, "step": 2969 }, { "epoch": 0.23413480488766258, "grad_norm": 0.8761427998542786, "learning_rate": 0.0005, "loss": 2.3304431438446045, "step": 2970 }, { "epoch": 0.23421363815530152, "grad_norm": 0.9791157841682434, "learning_rate": 0.0005, "loss": 2.328295946121216, "step": 2971 }, { "epoch": 0.2342924714229405, "grad_norm": 1.072414517402649, "learning_rate": 0.0005, "loss": 2.569359302520752, "step": 2972 }, { "epoch": 0.23437130469057943, "grad_norm": 0.9605534672737122, "learning_rate": 0.0005, "loss": 2.9362854957580566, "step": 2973 }, { "epoch": 0.23445013795821837, "grad_norm": 1.204398512840271, "learning_rate": 0.0005, "loss": 2.500023365020752, "step": 2974 }, { "epoch": 0.2345289712258573, "grad_norm": 0.8459603786468506, "learning_rate": 0.0005, "loss": 2.489551305770874, "step": 2975 }, { "epoch": 0.23460780449349625, "grad_norm": 1.1286466121673584, "learning_rate": 0.0005, "loss": 2.4733667373657227, "step": 2976 }, { "epoch": 0.2346866377611352, "grad_norm": 0.9142352938652039, "learning_rate": 0.0005, "loss": 2.4003334045410156, "step": 2977 }, { "epoch": 0.23476547102877415, "grad_norm": 0.9961639642715454, "learning_rate": 0.0005, "loss": 2.3148446083068848, "step": 2978 }, { "epoch": 0.2348443042964131, "grad_norm": 0.8928195238113403, "learning_rate": 0.0005, "loss": 2.5617995262145996, "step": 2979 }, { "epoch": 0.23492313756405203, "grad_norm": 1.1617909669876099, "learning_rate": 0.0005, "loss": 2.275507688522339, "step": 2980 }, { "epoch": 0.23500197083169097, "grad_norm": 0.9838461875915527, "learning_rate": 0.0005, "loss": 2.48831844329834, "step": 2981 }, { "epoch": 0.2350808040993299, "grad_norm": 0.8649630546569824, "learning_rate": 0.0005, "loss": 2.5452051162719727, "step": 2982 }, { "epoch": 0.23515963736696885, "grad_norm": 0.8808987140655518, "learning_rate": 0.0005, "loss": 2.6452817916870117, "step": 2983 }, { "epoch": 0.23523847063460782, "grad_norm": 0.9136822819709778, "learning_rate": 0.0005, "loss": 2.5921192169189453, "step": 2984 }, { "epoch": 0.23531730390224675, "grad_norm": 0.9820873737335205, "learning_rate": 0.0005, "loss": 2.7184934616088867, "step": 2985 }, { "epoch": 0.2353961371698857, "grad_norm": 0.8177094459533691, "learning_rate": 0.0005, "loss": 2.0152957439422607, "step": 2986 }, { "epoch": 0.23547497043752463, "grad_norm": 0.8988854885101318, "learning_rate": 0.0005, "loss": 2.496164321899414, "step": 2987 }, { "epoch": 0.23555380370516357, "grad_norm": 0.92451411485672, "learning_rate": 0.0005, "loss": 2.3498497009277344, "step": 2988 }, { "epoch": 0.2356326369728025, "grad_norm": 0.7557592988014221, "learning_rate": 0.0005, "loss": 2.3257579803466797, "step": 2989 }, { "epoch": 0.23571147024044148, "grad_norm": 1.0448650121688843, "learning_rate": 0.0005, "loss": 2.611675262451172, "step": 2990 }, { "epoch": 0.23579030350808042, "grad_norm": 0.7744594216346741, "learning_rate": 0.0005, "loss": 2.588068962097168, "step": 2991 }, { "epoch": 0.23586913677571936, "grad_norm": 0.9190212488174438, "learning_rate": 0.0005, "loss": 2.2188334465026855, "step": 2992 }, { "epoch": 0.2359479700433583, "grad_norm": 1.1343815326690674, "learning_rate": 0.0005, "loss": 2.725304126739502, "step": 2993 }, { "epoch": 0.23602680331099724, "grad_norm": 1.0435075759887695, "learning_rate": 0.0005, "loss": 2.373992919921875, "step": 2994 }, { "epoch": 0.23610563657863617, "grad_norm": 0.8399286270141602, "learning_rate": 0.0005, "loss": 2.6739659309387207, "step": 2995 }, { "epoch": 0.23618446984627514, "grad_norm": 0.904449462890625, "learning_rate": 0.0005, "loss": 2.606165885925293, "step": 2996 }, { "epoch": 0.23626330311391408, "grad_norm": 1.294633388519287, "learning_rate": 0.0005, "loss": 2.567086696624756, "step": 2997 }, { "epoch": 0.23634213638155302, "grad_norm": 1.8988795280456543, "learning_rate": 0.0005, "loss": 2.6564502716064453, "step": 2998 }, { "epoch": 0.23642096964919196, "grad_norm": 0.8919170498847961, "learning_rate": 0.0005, "loss": 2.404353380203247, "step": 2999 }, { "epoch": 0.2364998029168309, "grad_norm": 1.1175800561904907, "learning_rate": 0.0005, "loss": 2.3293824195861816, "step": 3000 }, { "epoch": 0.23657863618446984, "grad_norm": 1.299662709236145, "learning_rate": 0.0005, "loss": 2.7111799716949463, "step": 3001 }, { "epoch": 0.23665746945210878, "grad_norm": 0.9712973833084106, "learning_rate": 0.0005, "loss": 2.5139713287353516, "step": 3002 }, { "epoch": 0.23673630271974774, "grad_norm": 0.8509754538536072, "learning_rate": 0.0005, "loss": 2.4398252964019775, "step": 3003 }, { "epoch": 0.23681513598738668, "grad_norm": 1.0000144243240356, "learning_rate": 0.0005, "loss": 2.7336854934692383, "step": 3004 }, { "epoch": 0.23689396925502562, "grad_norm": 0.9443759918212891, "learning_rate": 0.0005, "loss": 2.538440704345703, "step": 3005 }, { "epoch": 0.23697280252266456, "grad_norm": 0.9406957626342773, "learning_rate": 0.0005, "loss": 2.6600770950317383, "step": 3006 }, { "epoch": 0.2370516357903035, "grad_norm": 1.088448166847229, "learning_rate": 0.0005, "loss": 2.9641337394714355, "step": 3007 }, { "epoch": 0.23713046905794244, "grad_norm": 0.9437215328216553, "learning_rate": 0.0005, "loss": 2.351804256439209, "step": 3008 }, { "epoch": 0.2372093023255814, "grad_norm": 1.146307349205017, "learning_rate": 0.0005, "loss": 2.8395187854766846, "step": 3009 }, { "epoch": 0.23728813559322035, "grad_norm": 0.9144481420516968, "learning_rate": 0.0005, "loss": 2.506544589996338, "step": 3010 }, { "epoch": 0.23736696886085928, "grad_norm": 0.8414566516876221, "learning_rate": 0.0005, "loss": 2.5275042057037354, "step": 3011 }, { "epoch": 0.23744580212849822, "grad_norm": 0.8234533667564392, "learning_rate": 0.0005, "loss": 2.4668986797332764, "step": 3012 }, { "epoch": 0.23752463539613716, "grad_norm": 0.8844017386436462, "learning_rate": 0.0005, "loss": 2.764681339263916, "step": 3013 }, { "epoch": 0.2376034686637761, "grad_norm": 1.022141456604004, "learning_rate": 0.0005, "loss": 2.719038486480713, "step": 3014 }, { "epoch": 0.23768230193141507, "grad_norm": 0.9016066789627075, "learning_rate": 0.0005, "loss": 2.537015199661255, "step": 3015 }, { "epoch": 0.237761135199054, "grad_norm": 0.9980467557907104, "learning_rate": 0.0005, "loss": 2.9885313510894775, "step": 3016 }, { "epoch": 0.23783996846669295, "grad_norm": 1.0457401275634766, "learning_rate": 0.0005, "loss": 2.3242197036743164, "step": 3017 }, { "epoch": 0.2379188017343319, "grad_norm": 0.8183895349502563, "learning_rate": 0.0005, "loss": 2.7567429542541504, "step": 3018 }, { "epoch": 0.23799763500197083, "grad_norm": 0.7417635321617126, "learning_rate": 0.0005, "loss": 2.359951972961426, "step": 3019 }, { "epoch": 0.23807646826960976, "grad_norm": 1.0898183584213257, "learning_rate": 0.0005, "loss": 2.6376171112060547, "step": 3020 }, { "epoch": 0.23815530153724873, "grad_norm": 0.8531420826911926, "learning_rate": 0.0005, "loss": 2.5603561401367188, "step": 3021 }, { "epoch": 0.23823413480488767, "grad_norm": 1.0605814456939697, "learning_rate": 0.0005, "loss": 2.622652292251587, "step": 3022 }, { "epoch": 0.2383129680725266, "grad_norm": 0.793324887752533, "learning_rate": 0.0005, "loss": 2.5824854373931885, "step": 3023 }, { "epoch": 0.23839180134016555, "grad_norm": 1.0206248760223389, "learning_rate": 0.0005, "loss": 2.5438761711120605, "step": 3024 }, { "epoch": 0.2384706346078045, "grad_norm": 1.1235260963439941, "learning_rate": 0.0005, "loss": 2.6703758239746094, "step": 3025 }, { "epoch": 0.23854946787544343, "grad_norm": 0.8580021858215332, "learning_rate": 0.0005, "loss": 2.422051191329956, "step": 3026 }, { "epoch": 0.2386283011430824, "grad_norm": 0.9659094214439392, "learning_rate": 0.0005, "loss": 2.286726951599121, "step": 3027 }, { "epoch": 0.23870713441072133, "grad_norm": 0.9630332589149475, "learning_rate": 0.0005, "loss": 2.2290706634521484, "step": 3028 }, { "epoch": 0.23878596767836027, "grad_norm": 0.9555943608283997, "learning_rate": 0.0005, "loss": 2.5655157566070557, "step": 3029 }, { "epoch": 0.2388648009459992, "grad_norm": 0.9825602769851685, "learning_rate": 0.0005, "loss": 2.5468907356262207, "step": 3030 }, { "epoch": 0.23894363421363815, "grad_norm": 0.9044749140739441, "learning_rate": 0.0005, "loss": 3.006455898284912, "step": 3031 }, { "epoch": 0.2390224674812771, "grad_norm": 1.244991421699524, "learning_rate": 0.0005, "loss": 2.563832998275757, "step": 3032 }, { "epoch": 0.23910130074891603, "grad_norm": 1.0936204195022583, "learning_rate": 0.0005, "loss": 2.7283358573913574, "step": 3033 }, { "epoch": 0.239180134016555, "grad_norm": 0.8105851411819458, "learning_rate": 0.0005, "loss": 2.5829708576202393, "step": 3034 }, { "epoch": 0.23925896728419394, "grad_norm": 1.07323157787323, "learning_rate": 0.0005, "loss": 2.527832269668579, "step": 3035 }, { "epoch": 0.23933780055183287, "grad_norm": 1.0381978750228882, "learning_rate": 0.0005, "loss": 2.540893077850342, "step": 3036 }, { "epoch": 0.2394166338194718, "grad_norm": 0.7912078499794006, "learning_rate": 0.0005, "loss": 2.742668390274048, "step": 3037 }, { "epoch": 0.23949546708711075, "grad_norm": 0.8192487359046936, "learning_rate": 0.0005, "loss": 2.6372265815734863, "step": 3038 }, { "epoch": 0.2395743003547497, "grad_norm": 0.9090244174003601, "learning_rate": 0.0005, "loss": 2.299360990524292, "step": 3039 }, { "epoch": 0.23965313362238866, "grad_norm": 0.8759653568267822, "learning_rate": 0.0005, "loss": 2.837047815322876, "step": 3040 }, { "epoch": 0.2397319668900276, "grad_norm": 0.8510549068450928, "learning_rate": 0.0005, "loss": 2.359104633331299, "step": 3041 }, { "epoch": 0.23981080015766654, "grad_norm": 0.7920495271682739, "learning_rate": 0.0005, "loss": 2.343033790588379, "step": 3042 }, { "epoch": 0.23988963342530548, "grad_norm": 0.9499079585075378, "learning_rate": 0.0005, "loss": 2.347561836242676, "step": 3043 }, { "epoch": 0.23996846669294442, "grad_norm": 0.8349394798278809, "learning_rate": 0.0005, "loss": 2.216712474822998, "step": 3044 }, { "epoch": 0.24004729996058335, "grad_norm": 0.9194797277450562, "learning_rate": 0.0005, "loss": 2.436549663543701, "step": 3045 }, { "epoch": 0.24012613322822232, "grad_norm": 1.057592511177063, "learning_rate": 0.0005, "loss": 2.564614772796631, "step": 3046 }, { "epoch": 0.24020496649586126, "grad_norm": 1.053277850151062, "learning_rate": 0.0005, "loss": 2.817136764526367, "step": 3047 }, { "epoch": 0.2402837997635002, "grad_norm": 0.9056686162948608, "learning_rate": 0.0005, "loss": 2.3409690856933594, "step": 3048 }, { "epoch": 0.24036263303113914, "grad_norm": 1.1534677743911743, "learning_rate": 0.0005, "loss": 2.545651435852051, "step": 3049 }, { "epoch": 0.24044146629877808, "grad_norm": 1.0722312927246094, "learning_rate": 0.0005, "loss": 2.6575849056243896, "step": 3050 }, { "epoch": 0.24052029956641702, "grad_norm": 1.039440631866455, "learning_rate": 0.0005, "loss": 2.4900100231170654, "step": 3051 }, { "epoch": 0.24059913283405598, "grad_norm": 0.9213700294494629, "learning_rate": 0.0005, "loss": 2.4983458518981934, "step": 3052 }, { "epoch": 0.24067796610169492, "grad_norm": 0.7905643582344055, "learning_rate": 0.0005, "loss": 2.3481974601745605, "step": 3053 }, { "epoch": 0.24075679936933386, "grad_norm": 0.7962742447853088, "learning_rate": 0.0005, "loss": 2.5092878341674805, "step": 3054 }, { "epoch": 0.2408356326369728, "grad_norm": 0.8278843760490417, "learning_rate": 0.0005, "loss": 2.4446024894714355, "step": 3055 }, { "epoch": 0.24091446590461174, "grad_norm": 1.98577082157135, "learning_rate": 0.0005, "loss": 2.348690986633301, "step": 3056 }, { "epoch": 0.24099329917225068, "grad_norm": 1.1349689960479736, "learning_rate": 0.0005, "loss": 2.403032064437866, "step": 3057 }, { "epoch": 0.24107213243988965, "grad_norm": 0.9032559990882874, "learning_rate": 0.0005, "loss": 2.811160087585449, "step": 3058 }, { "epoch": 0.2411509657075286, "grad_norm": 0.9676867127418518, "learning_rate": 0.0005, "loss": 2.705817699432373, "step": 3059 }, { "epoch": 0.24122979897516753, "grad_norm": 1.028438687324524, "learning_rate": 0.0005, "loss": 2.865161418914795, "step": 3060 }, { "epoch": 0.24130863224280646, "grad_norm": 1.1505966186523438, "learning_rate": 0.0005, "loss": 2.411102533340454, "step": 3061 }, { "epoch": 0.2413874655104454, "grad_norm": 1.2785097360610962, "learning_rate": 0.0005, "loss": 2.2250072956085205, "step": 3062 }, { "epoch": 0.24146629877808434, "grad_norm": 1.215009331703186, "learning_rate": 0.0005, "loss": 2.3741798400878906, "step": 3063 }, { "epoch": 0.24154513204572328, "grad_norm": 1.1327729225158691, "learning_rate": 0.0005, "loss": 2.442598819732666, "step": 3064 }, { "epoch": 0.24162396531336225, "grad_norm": 1.0550849437713623, "learning_rate": 0.0005, "loss": 2.7435660362243652, "step": 3065 }, { "epoch": 0.2417027985810012, "grad_norm": 1.0425792932510376, "learning_rate": 0.0005, "loss": 2.1046252250671387, "step": 3066 }, { "epoch": 0.24178163184864013, "grad_norm": 0.9977485537528992, "learning_rate": 0.0005, "loss": 2.3628125190734863, "step": 3067 }, { "epoch": 0.24186046511627907, "grad_norm": 0.9770811796188354, "learning_rate": 0.0005, "loss": 2.342592716217041, "step": 3068 }, { "epoch": 0.241939298383918, "grad_norm": 0.9482544660568237, "learning_rate": 0.0005, "loss": 2.3312129974365234, "step": 3069 }, { "epoch": 0.24201813165155694, "grad_norm": 1.1929422616958618, "learning_rate": 0.0005, "loss": 2.639347553253174, "step": 3070 }, { "epoch": 0.2420969649191959, "grad_norm": 5.228980541229248, "learning_rate": 0.0005, "loss": 2.053433895111084, "step": 3071 }, { "epoch": 0.24217579818683485, "grad_norm": 0.9421496987342834, "learning_rate": 0.0005, "loss": 2.6703970432281494, "step": 3072 }, { "epoch": 0.2422546314544738, "grad_norm": 0.8688445091247559, "learning_rate": 0.0005, "loss": 2.7308788299560547, "step": 3073 }, { "epoch": 0.24233346472211273, "grad_norm": 1.2864625453948975, "learning_rate": 0.0005, "loss": 2.0337367057800293, "step": 3074 }, { "epoch": 0.24241229798975167, "grad_norm": 0.8720927238464355, "learning_rate": 0.0005, "loss": 2.5787887573242188, "step": 3075 }, { "epoch": 0.2424911312573906, "grad_norm": 1.0451866388320923, "learning_rate": 0.0005, "loss": 2.379924774169922, "step": 3076 }, { "epoch": 0.24256996452502957, "grad_norm": 0.8860479593276978, "learning_rate": 0.0005, "loss": 2.6175732612609863, "step": 3077 }, { "epoch": 0.2426487977926685, "grad_norm": 0.7305130362510681, "learning_rate": 0.0005, "loss": 2.4714174270629883, "step": 3078 }, { "epoch": 0.24272763106030745, "grad_norm": 0.9667377471923828, "learning_rate": 0.0005, "loss": 2.7890329360961914, "step": 3079 }, { "epoch": 0.2428064643279464, "grad_norm": 0.9895792603492737, "learning_rate": 0.0005, "loss": 2.7976338863372803, "step": 3080 }, { "epoch": 0.24288529759558533, "grad_norm": 1.041471242904663, "learning_rate": 0.0005, "loss": 2.1720004081726074, "step": 3081 }, { "epoch": 0.24296413086322427, "grad_norm": 0.8862119913101196, "learning_rate": 0.0005, "loss": 2.6960537433624268, "step": 3082 }, { "epoch": 0.24304296413086324, "grad_norm": 0.7995978593826294, "learning_rate": 0.0005, "loss": 2.420438289642334, "step": 3083 }, { "epoch": 0.24312179739850218, "grad_norm": 1.2993963956832886, "learning_rate": 0.0005, "loss": 2.6513168811798096, "step": 3084 }, { "epoch": 0.24320063066614112, "grad_norm": 0.8802982568740845, "learning_rate": 0.0005, "loss": 2.258329391479492, "step": 3085 }, { "epoch": 0.24327946393378005, "grad_norm": 1.170333981513977, "learning_rate": 0.0005, "loss": 2.5121021270751953, "step": 3086 }, { "epoch": 0.243358297201419, "grad_norm": 1.01105535030365, "learning_rate": 0.0005, "loss": 2.4930360317230225, "step": 3087 }, { "epoch": 0.24343713046905793, "grad_norm": 1.3109992742538452, "learning_rate": 0.0005, "loss": 2.4837074279785156, "step": 3088 }, { "epoch": 0.2435159637366969, "grad_norm": 0.9554075002670288, "learning_rate": 0.0005, "loss": 2.763075590133667, "step": 3089 }, { "epoch": 0.24359479700433584, "grad_norm": 0.9861536026000977, "learning_rate": 0.0005, "loss": 2.267909526824951, "step": 3090 }, { "epoch": 0.24367363027197478, "grad_norm": 0.8929198384284973, "learning_rate": 0.0005, "loss": 2.1874685287475586, "step": 3091 }, { "epoch": 0.24375246353961372, "grad_norm": 0.9449867010116577, "learning_rate": 0.0005, "loss": 2.5375866889953613, "step": 3092 }, { "epoch": 0.24383129680725266, "grad_norm": 1.0342260599136353, "learning_rate": 0.0005, "loss": 2.537652015686035, "step": 3093 }, { "epoch": 0.2439101300748916, "grad_norm": 0.9916126132011414, "learning_rate": 0.0005, "loss": 2.485013484954834, "step": 3094 }, { "epoch": 0.24398896334253053, "grad_norm": 1.286288857460022, "learning_rate": 0.0005, "loss": 2.409755229949951, "step": 3095 }, { "epoch": 0.2440677966101695, "grad_norm": 1.3848917484283447, "learning_rate": 0.0005, "loss": 3.016080856323242, "step": 3096 }, { "epoch": 0.24414662987780844, "grad_norm": 1.0369089841842651, "learning_rate": 0.0005, "loss": 2.2622475624084473, "step": 3097 }, { "epoch": 0.24422546314544738, "grad_norm": 1.3752074241638184, "learning_rate": 0.0005, "loss": 3.2637157440185547, "step": 3098 }, { "epoch": 0.24430429641308632, "grad_norm": 0.9294754266738892, "learning_rate": 0.0005, "loss": 2.7690274715423584, "step": 3099 }, { "epoch": 0.24438312968072526, "grad_norm": 0.9566220045089722, "learning_rate": 0.0005, "loss": 2.2014334201812744, "step": 3100 }, { "epoch": 0.2444619629483642, "grad_norm": 1.0390158891677856, "learning_rate": 0.0005, "loss": 2.4799890518188477, "step": 3101 }, { "epoch": 0.24454079621600316, "grad_norm": 0.887715756893158, "learning_rate": 0.0005, "loss": 2.576634407043457, "step": 3102 }, { "epoch": 0.2446196294836421, "grad_norm": 1.574677586555481, "learning_rate": 0.0005, "loss": 2.90004301071167, "step": 3103 }, { "epoch": 0.24469846275128104, "grad_norm": 1.18546462059021, "learning_rate": 0.0005, "loss": 2.687934637069702, "step": 3104 }, { "epoch": 0.24477729601891998, "grad_norm": 1.1101140975952148, "learning_rate": 0.0005, "loss": 2.642306327819824, "step": 3105 }, { "epoch": 0.24485612928655892, "grad_norm": 1.012487769126892, "learning_rate": 0.0005, "loss": 2.321702003479004, "step": 3106 }, { "epoch": 0.24493496255419786, "grad_norm": 1.0213229656219482, "learning_rate": 0.0005, "loss": 2.7015607357025146, "step": 3107 }, { "epoch": 0.24501379582183683, "grad_norm": 1.0862435102462769, "learning_rate": 0.0005, "loss": 2.620810031890869, "step": 3108 }, { "epoch": 0.24509262908947577, "grad_norm": 0.8837053775787354, "learning_rate": 0.0005, "loss": 2.464472770690918, "step": 3109 }, { "epoch": 0.2451714623571147, "grad_norm": 1.007792353630066, "learning_rate": 0.0005, "loss": 2.741297483444214, "step": 3110 }, { "epoch": 0.24525029562475364, "grad_norm": 0.7539751529693604, "learning_rate": 0.0005, "loss": 2.527778387069702, "step": 3111 }, { "epoch": 0.24532912889239258, "grad_norm": 0.8489691019058228, "learning_rate": 0.0005, "loss": 2.2849559783935547, "step": 3112 }, { "epoch": 0.24540796216003152, "grad_norm": 1.0486091375350952, "learning_rate": 0.0005, "loss": 2.6640875339508057, "step": 3113 }, { "epoch": 0.2454867954276705, "grad_norm": 0.868053138256073, "learning_rate": 0.0005, "loss": 2.231480598449707, "step": 3114 }, { "epoch": 0.24556562869530943, "grad_norm": 0.8732636570930481, "learning_rate": 0.0005, "loss": 2.758824348449707, "step": 3115 }, { "epoch": 0.24564446196294837, "grad_norm": 0.9314684867858887, "learning_rate": 0.0005, "loss": 2.061418056488037, "step": 3116 }, { "epoch": 0.2457232952305873, "grad_norm": 1.0850145816802979, "learning_rate": 0.0005, "loss": 2.430996894836426, "step": 3117 }, { "epoch": 0.24580212849822625, "grad_norm": 0.8165575265884399, "learning_rate": 0.0005, "loss": 2.085977077484131, "step": 3118 }, { "epoch": 0.24588096176586519, "grad_norm": 0.8913048505783081, "learning_rate": 0.0005, "loss": 2.4488377571105957, "step": 3119 }, { "epoch": 0.24595979503350412, "grad_norm": 0.9539270401000977, "learning_rate": 0.0005, "loss": 2.596086025238037, "step": 3120 }, { "epoch": 0.2460386283011431, "grad_norm": 1.1350154876708984, "learning_rate": 0.0005, "loss": 2.225346088409424, "step": 3121 }, { "epoch": 0.24611746156878203, "grad_norm": 0.7997763156890869, "learning_rate": 0.0005, "loss": 2.4043638706207275, "step": 3122 }, { "epoch": 0.24619629483642097, "grad_norm": 1.0106030702590942, "learning_rate": 0.0005, "loss": 2.2162177562713623, "step": 3123 }, { "epoch": 0.2462751281040599, "grad_norm": 0.9227945804595947, "learning_rate": 0.0005, "loss": 2.906726837158203, "step": 3124 }, { "epoch": 0.24635396137169885, "grad_norm": 1.2032294273376465, "learning_rate": 0.0005, "loss": 2.57633900642395, "step": 3125 }, { "epoch": 0.2464327946393378, "grad_norm": 1.1824294328689575, "learning_rate": 0.0005, "loss": 2.1852383613586426, "step": 3126 }, { "epoch": 0.24651162790697675, "grad_norm": 1.0747109651565552, "learning_rate": 0.0005, "loss": 2.425288677215576, "step": 3127 }, { "epoch": 0.2465904611746157, "grad_norm": 0.801397442817688, "learning_rate": 0.0005, "loss": 2.6860716342926025, "step": 3128 }, { "epoch": 0.24666929444225463, "grad_norm": 0.8684742450714111, "learning_rate": 0.0005, "loss": 2.720311403274536, "step": 3129 }, { "epoch": 0.24674812770989357, "grad_norm": 0.9567261338233948, "learning_rate": 0.0005, "loss": 2.1804680824279785, "step": 3130 }, { "epoch": 0.2468269609775325, "grad_norm": 0.8523715138435364, "learning_rate": 0.0005, "loss": 2.430288791656494, "step": 3131 }, { "epoch": 0.24690579424517145, "grad_norm": 0.8672579526901245, "learning_rate": 0.0005, "loss": 2.418478012084961, "step": 3132 }, { "epoch": 0.24698462751281042, "grad_norm": 0.8551589846611023, "learning_rate": 0.0005, "loss": 2.516190528869629, "step": 3133 }, { "epoch": 0.24706346078044936, "grad_norm": 0.9791656732559204, "learning_rate": 0.0005, "loss": 2.521397590637207, "step": 3134 }, { "epoch": 0.2471422940480883, "grad_norm": 0.8899649381637573, "learning_rate": 0.0005, "loss": 2.748910903930664, "step": 3135 }, { "epoch": 0.24722112731572723, "grad_norm": 0.8718938827514648, "learning_rate": 0.0005, "loss": 2.554154872894287, "step": 3136 }, { "epoch": 0.24729996058336617, "grad_norm": 0.8944060206413269, "learning_rate": 0.0005, "loss": 2.724609851837158, "step": 3137 }, { "epoch": 0.2473787938510051, "grad_norm": 0.9811286330223083, "learning_rate": 0.0005, "loss": 2.331855535507202, "step": 3138 }, { "epoch": 0.24745762711864408, "grad_norm": 0.9984787106513977, "learning_rate": 0.0005, "loss": 2.6976494789123535, "step": 3139 }, { "epoch": 0.24753646038628302, "grad_norm": 0.7480726838111877, "learning_rate": 0.0005, "loss": 2.1102514266967773, "step": 3140 }, { "epoch": 0.24761529365392196, "grad_norm": 1.0926892757415771, "learning_rate": 0.0005, "loss": 2.3747944831848145, "step": 3141 }, { "epoch": 0.2476941269215609, "grad_norm": 1.0408332347869873, "learning_rate": 0.0005, "loss": 2.520897626876831, "step": 3142 }, { "epoch": 0.24777296018919984, "grad_norm": 0.9228712320327759, "learning_rate": 0.0005, "loss": 2.681619644165039, "step": 3143 }, { "epoch": 0.24785179345683878, "grad_norm": 0.8685837984085083, "learning_rate": 0.0005, "loss": 2.388462543487549, "step": 3144 }, { "epoch": 0.24793062672447774, "grad_norm": 1.1085388660430908, "learning_rate": 0.0005, "loss": 2.263063430786133, "step": 3145 }, { "epoch": 0.24800945999211668, "grad_norm": 0.9989566206932068, "learning_rate": 0.0005, "loss": 2.7941575050354004, "step": 3146 }, { "epoch": 0.24808829325975562, "grad_norm": 1.0470443964004517, "learning_rate": 0.0005, "loss": 2.3155264854431152, "step": 3147 }, { "epoch": 0.24816712652739456, "grad_norm": 1.5038410425186157, "learning_rate": 0.0005, "loss": 2.5360755920410156, "step": 3148 }, { "epoch": 0.2482459597950335, "grad_norm": 0.8046585321426392, "learning_rate": 0.0005, "loss": 2.4334397315979004, "step": 3149 }, { "epoch": 0.24832479306267244, "grad_norm": 1.067539930343628, "learning_rate": 0.0005, "loss": 2.5980024337768555, "step": 3150 }, { "epoch": 0.24840362633031138, "grad_norm": 1.18369460105896, "learning_rate": 0.0005, "loss": 2.8192050457000732, "step": 3151 }, { "epoch": 0.24848245959795034, "grad_norm": 0.8828449845314026, "learning_rate": 0.0005, "loss": 2.4101204872131348, "step": 3152 }, { "epoch": 0.24856129286558928, "grad_norm": 0.9460877180099487, "learning_rate": 0.0005, "loss": 2.132899761199951, "step": 3153 }, { "epoch": 0.24864012613322822, "grad_norm": 1.0363073348999023, "learning_rate": 0.0005, "loss": 2.7091288566589355, "step": 3154 }, { "epoch": 0.24871895940086716, "grad_norm": 0.7621098756790161, "learning_rate": 0.0005, "loss": 2.275830030441284, "step": 3155 }, { "epoch": 0.2487977926685061, "grad_norm": 1.2313754558563232, "learning_rate": 0.0005, "loss": 2.113570213317871, "step": 3156 }, { "epoch": 0.24887662593614504, "grad_norm": 0.9126701951026917, "learning_rate": 0.0005, "loss": 2.796659231185913, "step": 3157 }, { "epoch": 0.248955459203784, "grad_norm": 1.1240997314453125, "learning_rate": 0.0005, "loss": 2.4353575706481934, "step": 3158 }, { "epoch": 0.24903429247142295, "grad_norm": 0.8288122415542603, "learning_rate": 0.0005, "loss": 2.4251742362976074, "step": 3159 }, { "epoch": 0.24911312573906189, "grad_norm": 1.0597491264343262, "learning_rate": 0.0005, "loss": 2.33052134513855, "step": 3160 }, { "epoch": 0.24919195900670082, "grad_norm": 0.972439169883728, "learning_rate": 0.0005, "loss": 2.5996599197387695, "step": 3161 }, { "epoch": 0.24927079227433976, "grad_norm": 1.057792067527771, "learning_rate": 0.0005, "loss": 2.601925849914551, "step": 3162 }, { "epoch": 0.2493496255419787, "grad_norm": 1.1708343029022217, "learning_rate": 0.0005, "loss": 2.4076600074768066, "step": 3163 }, { "epoch": 0.24942845880961767, "grad_norm": 0.8422321677207947, "learning_rate": 0.0005, "loss": 2.4460339546203613, "step": 3164 }, { "epoch": 0.2495072920772566, "grad_norm": 0.7503582835197449, "learning_rate": 0.0005, "loss": 2.413975238800049, "step": 3165 }, { "epoch": 0.24958612534489555, "grad_norm": 1.1466773748397827, "learning_rate": 0.0005, "loss": 2.2461161613464355, "step": 3166 }, { "epoch": 0.2496649586125345, "grad_norm": 0.9140097498893738, "learning_rate": 0.0005, "loss": 2.54095721244812, "step": 3167 }, { "epoch": 0.24974379188017343, "grad_norm": 0.7810092568397522, "learning_rate": 0.0005, "loss": 2.5652217864990234, "step": 3168 }, { "epoch": 0.24982262514781237, "grad_norm": 1.527830958366394, "learning_rate": 0.0005, "loss": 2.2946176528930664, "step": 3169 }, { "epoch": 0.24990145841545133, "grad_norm": 0.8759078979492188, "learning_rate": 0.0005, "loss": 2.458449602127075, "step": 3170 }, { "epoch": 0.24998029168309027, "grad_norm": 1.141597032546997, "learning_rate": 0.0005, "loss": 2.754239082336426, "step": 3171 }, { "epoch": 0.2500591249507292, "grad_norm": 1.5685067176818848, "learning_rate": 0.0005, "loss": 2.509869337081909, "step": 3172 }, { "epoch": 0.25013795821836815, "grad_norm": 0.8918545842170715, "learning_rate": 0.0005, "loss": 2.4108195304870605, "step": 3173 }, { "epoch": 0.2502167914860071, "grad_norm": 0.8821497559547424, "learning_rate": 0.0005, "loss": 2.819000244140625, "step": 3174 }, { "epoch": 0.25029562475364603, "grad_norm": 0.8339043259620667, "learning_rate": 0.0005, "loss": 2.5261662006378174, "step": 3175 }, { "epoch": 0.250374458021285, "grad_norm": 0.8609758615493774, "learning_rate": 0.0005, "loss": 2.3348000049591064, "step": 3176 }, { "epoch": 0.2504532912889239, "grad_norm": 0.9502364993095398, "learning_rate": 0.0005, "loss": 2.3113937377929688, "step": 3177 }, { "epoch": 0.2505321245565629, "grad_norm": 0.8923535346984863, "learning_rate": 0.0005, "loss": 2.482658624649048, "step": 3178 }, { "epoch": 0.2506109578242018, "grad_norm": 1.0134626626968384, "learning_rate": 0.0005, "loss": 2.5525548458099365, "step": 3179 }, { "epoch": 0.25068979109184075, "grad_norm": 0.8955383896827698, "learning_rate": 0.0005, "loss": 2.5132837295532227, "step": 3180 }, { "epoch": 0.2507686243594797, "grad_norm": 1.2111080884933472, "learning_rate": 0.0005, "loss": 2.5129475593566895, "step": 3181 }, { "epoch": 0.25084745762711863, "grad_norm": 1.0796856880187988, "learning_rate": 0.0005, "loss": 2.41926908493042, "step": 3182 }, { "epoch": 0.2509262908947576, "grad_norm": 0.918317973613739, "learning_rate": 0.0005, "loss": 2.756809711456299, "step": 3183 }, { "epoch": 0.2510051241623965, "grad_norm": 1.6148124933242798, "learning_rate": 0.0005, "loss": 3.1108031272888184, "step": 3184 }, { "epoch": 0.2510839574300355, "grad_norm": 1.2567732334136963, "learning_rate": 0.0005, "loss": 2.7453012466430664, "step": 3185 }, { "epoch": 0.25116279069767444, "grad_norm": 1.0516387224197388, "learning_rate": 0.0005, "loss": 2.6348302364349365, "step": 3186 }, { "epoch": 0.25124162396531335, "grad_norm": 1.0571260452270508, "learning_rate": 0.0005, "loss": 2.7356934547424316, "step": 3187 }, { "epoch": 0.2513204572329523, "grad_norm": 1.0101486444473267, "learning_rate": 0.0005, "loss": 2.612165927886963, "step": 3188 }, { "epoch": 0.25139929050059123, "grad_norm": 0.9034597873687744, "learning_rate": 0.0005, "loss": 2.5939488410949707, "step": 3189 }, { "epoch": 0.2514781237682302, "grad_norm": 0.9119088649749756, "learning_rate": 0.0005, "loss": 2.2665956020355225, "step": 3190 }, { "epoch": 0.2515569570358691, "grad_norm": 0.7468304634094238, "learning_rate": 0.0005, "loss": 2.3723559379577637, "step": 3191 }, { "epoch": 0.2516357903035081, "grad_norm": 1.3170475959777832, "learning_rate": 0.0005, "loss": 2.390028953552246, "step": 3192 }, { "epoch": 0.25171462357114704, "grad_norm": 1.11244797706604, "learning_rate": 0.0005, "loss": 2.4778084754943848, "step": 3193 }, { "epoch": 0.25179345683878596, "grad_norm": 1.07865571975708, "learning_rate": 0.0005, "loss": 2.4254398345947266, "step": 3194 }, { "epoch": 0.2518722901064249, "grad_norm": 0.7941045165061951, "learning_rate": 0.0005, "loss": 2.7990548610687256, "step": 3195 }, { "epoch": 0.25195112337406383, "grad_norm": 0.8444695472717285, "learning_rate": 0.0005, "loss": 2.6789181232452393, "step": 3196 }, { "epoch": 0.2520299566417028, "grad_norm": 0.8984314203262329, "learning_rate": 0.0005, "loss": 2.55043363571167, "step": 3197 }, { "epoch": 0.25210878990934177, "grad_norm": 0.8881953358650208, "learning_rate": 0.0005, "loss": 2.128664016723633, "step": 3198 }, { "epoch": 0.2521876231769807, "grad_norm": 0.9012407064437866, "learning_rate": 0.0005, "loss": 2.2405905723571777, "step": 3199 }, { "epoch": 0.25226645644461965, "grad_norm": 1.0331010818481445, "learning_rate": 0.0005, "loss": 2.29587721824646, "step": 3200 }, { "epoch": 0.25234528971225856, "grad_norm": 0.9639036059379578, "learning_rate": 0.0005, "loss": 2.342085838317871, "step": 3201 }, { "epoch": 0.2524241229798975, "grad_norm": 0.8997994661331177, "learning_rate": 0.0005, "loss": 2.4997358322143555, "step": 3202 }, { "epoch": 0.25250295624753644, "grad_norm": 0.9490991234779358, "learning_rate": 0.0005, "loss": 2.1608757972717285, "step": 3203 }, { "epoch": 0.2525817895151754, "grad_norm": 1.175858974456787, "learning_rate": 0.0005, "loss": 2.462160110473633, "step": 3204 }, { "epoch": 0.25266062278281437, "grad_norm": 1.7790813446044922, "learning_rate": 0.0005, "loss": 2.5720419883728027, "step": 3205 }, { "epoch": 0.2527394560504533, "grad_norm": 0.8505237698554993, "learning_rate": 0.0005, "loss": 2.7489705085754395, "step": 3206 }, { "epoch": 0.25281828931809225, "grad_norm": 0.8816156387329102, "learning_rate": 0.0005, "loss": 2.4622244834899902, "step": 3207 }, { "epoch": 0.25289712258573116, "grad_norm": 0.8748038411140442, "learning_rate": 0.0005, "loss": 2.191404342651367, "step": 3208 }, { "epoch": 0.2529759558533701, "grad_norm": 1.4646186828613281, "learning_rate": 0.0005, "loss": 2.509690761566162, "step": 3209 }, { "epoch": 0.25305478912100904, "grad_norm": 1.0840106010437012, "learning_rate": 0.0005, "loss": 2.7577157020568848, "step": 3210 }, { "epoch": 0.253133622388648, "grad_norm": 0.8444401025772095, "learning_rate": 0.0005, "loss": 2.1327157020568848, "step": 3211 }, { "epoch": 0.25321245565628697, "grad_norm": 0.9622279405593872, "learning_rate": 0.0005, "loss": 2.39662766456604, "step": 3212 }, { "epoch": 0.2532912889239259, "grad_norm": 0.924445629119873, "learning_rate": 0.0005, "loss": 2.374476909637451, "step": 3213 }, { "epoch": 0.25337012219156485, "grad_norm": 1.1576697826385498, "learning_rate": 0.0005, "loss": 2.5350100994110107, "step": 3214 }, { "epoch": 0.25344895545920376, "grad_norm": 0.7374961376190186, "learning_rate": 0.0005, "loss": 2.0749893188476562, "step": 3215 }, { "epoch": 0.25352778872684273, "grad_norm": 0.9427622556686401, "learning_rate": 0.0005, "loss": 2.6788692474365234, "step": 3216 }, { "epoch": 0.2536066219944817, "grad_norm": 0.868455171585083, "learning_rate": 0.0005, "loss": 2.39021635055542, "step": 3217 }, { "epoch": 0.2536854552621206, "grad_norm": 0.9761685132980347, "learning_rate": 0.0005, "loss": 2.471677780151367, "step": 3218 }, { "epoch": 0.2537642885297596, "grad_norm": 0.9569480419158936, "learning_rate": 0.0005, "loss": 2.3556149005889893, "step": 3219 }, { "epoch": 0.2538431217973985, "grad_norm": 1.5134942531585693, "learning_rate": 0.0005, "loss": 2.797672748565674, "step": 3220 }, { "epoch": 0.25392195506503745, "grad_norm": 0.9784271717071533, "learning_rate": 0.0005, "loss": 2.579576015472412, "step": 3221 }, { "epoch": 0.25400078833267636, "grad_norm": 0.8614748120307922, "learning_rate": 0.0005, "loss": 2.629347562789917, "step": 3222 }, { "epoch": 0.25407962160031533, "grad_norm": 1.5850616693496704, "learning_rate": 0.0005, "loss": 2.778613567352295, "step": 3223 }, { "epoch": 0.2541584548679543, "grad_norm": 1.0151485204696655, "learning_rate": 0.0005, "loss": 2.2068934440612793, "step": 3224 }, { "epoch": 0.2542372881355932, "grad_norm": 0.8181979060173035, "learning_rate": 0.0005, "loss": 2.9726858139038086, "step": 3225 }, { "epoch": 0.2543161214032322, "grad_norm": 0.8664817214012146, "learning_rate": 0.0005, "loss": 2.0393857955932617, "step": 3226 }, { "epoch": 0.2543949546708711, "grad_norm": 0.99616938829422, "learning_rate": 0.0005, "loss": 2.1197307109832764, "step": 3227 }, { "epoch": 0.25447378793851005, "grad_norm": 1.3927215337753296, "learning_rate": 0.0005, "loss": 2.5996575355529785, "step": 3228 }, { "epoch": 0.254552621206149, "grad_norm": 1.1639262437820435, "learning_rate": 0.0005, "loss": 2.270350933074951, "step": 3229 }, { "epoch": 0.25463145447378793, "grad_norm": 0.9938727617263794, "learning_rate": 0.0005, "loss": 2.290323257446289, "step": 3230 }, { "epoch": 0.2547102877414269, "grad_norm": 1.1651585102081299, "learning_rate": 0.0005, "loss": 2.5524637699127197, "step": 3231 }, { "epoch": 0.2547891210090658, "grad_norm": 0.8534166216850281, "learning_rate": 0.0005, "loss": 2.367307662963867, "step": 3232 }, { "epoch": 0.2548679542767048, "grad_norm": 1.0111912488937378, "learning_rate": 0.0005, "loss": 2.522846221923828, "step": 3233 }, { "epoch": 0.2549467875443437, "grad_norm": 1.0530534982681274, "learning_rate": 0.0005, "loss": 2.0721192359924316, "step": 3234 }, { "epoch": 0.25502562081198266, "grad_norm": 1.3681191205978394, "learning_rate": 0.0005, "loss": 2.5064210891723633, "step": 3235 }, { "epoch": 0.2551044540796216, "grad_norm": 1.4768309593200684, "learning_rate": 0.0005, "loss": 2.7374467849731445, "step": 3236 }, { "epoch": 0.25518328734726053, "grad_norm": 0.8797965049743652, "learning_rate": 0.0005, "loss": 2.360413074493408, "step": 3237 }, { "epoch": 0.2552621206148995, "grad_norm": 0.9561537504196167, "learning_rate": 0.0005, "loss": 2.605501174926758, "step": 3238 }, { "epoch": 0.2553409538825384, "grad_norm": 0.9056525230407715, "learning_rate": 0.0005, "loss": 2.3404011726379395, "step": 3239 }, { "epoch": 0.2554197871501774, "grad_norm": 1.28548264503479, "learning_rate": 0.0005, "loss": 2.5850319862365723, "step": 3240 }, { "epoch": 0.2554986204178163, "grad_norm": 1.1039572954177856, "learning_rate": 0.0005, "loss": 2.724769115447998, "step": 3241 }, { "epoch": 0.25557745368545526, "grad_norm": 0.7788950800895691, "learning_rate": 0.0005, "loss": 2.421783447265625, "step": 3242 }, { "epoch": 0.2556562869530942, "grad_norm": 0.9570446014404297, "learning_rate": 0.0005, "loss": 2.369765281677246, "step": 3243 }, { "epoch": 0.25573512022073314, "grad_norm": 0.9804481267929077, "learning_rate": 0.0005, "loss": 2.4480552673339844, "step": 3244 }, { "epoch": 0.2558139534883721, "grad_norm": 0.7492142915725708, "learning_rate": 0.0005, "loss": 2.2410006523132324, "step": 3245 }, { "epoch": 0.255892786756011, "grad_norm": 0.8485122323036194, "learning_rate": 0.0005, "loss": 2.61220645904541, "step": 3246 }, { "epoch": 0.25597162002365, "grad_norm": 0.9974751472473145, "learning_rate": 0.0005, "loss": 2.6561226844787598, "step": 3247 }, { "epoch": 0.25605045329128895, "grad_norm": 0.8742047548294067, "learning_rate": 0.0005, "loss": 2.2985472679138184, "step": 3248 }, { "epoch": 0.25612928655892786, "grad_norm": 0.7393164038658142, "learning_rate": 0.0005, "loss": 2.313953161239624, "step": 3249 }, { "epoch": 0.2562081198265668, "grad_norm": 1.1897878646850586, "learning_rate": 0.0005, "loss": 2.5786595344543457, "step": 3250 }, { "epoch": 0.25628695309420574, "grad_norm": 0.8765383958816528, "learning_rate": 0.0005, "loss": 2.394995927810669, "step": 3251 }, { "epoch": 0.2563657863618447, "grad_norm": 0.8753107786178589, "learning_rate": 0.0005, "loss": 2.2965803146362305, "step": 3252 }, { "epoch": 0.2564446196294836, "grad_norm": 0.9375261664390564, "learning_rate": 0.0005, "loss": 2.7539172172546387, "step": 3253 }, { "epoch": 0.2565234528971226, "grad_norm": 0.8968144059181213, "learning_rate": 0.0005, "loss": 2.387910842895508, "step": 3254 }, { "epoch": 0.25660228616476155, "grad_norm": 0.9307904243469238, "learning_rate": 0.0005, "loss": 2.765410900115967, "step": 3255 }, { "epoch": 0.25668111943240046, "grad_norm": 0.8547713160514832, "learning_rate": 0.0005, "loss": 2.4206533432006836, "step": 3256 }, { "epoch": 0.25675995270003943, "grad_norm": 0.8392279744148254, "learning_rate": 0.0005, "loss": 2.348686695098877, "step": 3257 }, { "epoch": 0.25683878596767834, "grad_norm": 0.7949288487434387, "learning_rate": 0.0005, "loss": 2.529337167739868, "step": 3258 }, { "epoch": 0.2569176192353173, "grad_norm": 0.8452473282814026, "learning_rate": 0.0005, "loss": 2.576134204864502, "step": 3259 }, { "epoch": 0.2569964525029563, "grad_norm": 0.8367244005203247, "learning_rate": 0.0005, "loss": 2.147246837615967, "step": 3260 }, { "epoch": 0.2570752857705952, "grad_norm": 0.9693011045455933, "learning_rate": 0.0005, "loss": 2.227930784225464, "step": 3261 }, { "epoch": 0.25715411903823415, "grad_norm": 0.801139235496521, "learning_rate": 0.0005, "loss": 2.5204620361328125, "step": 3262 }, { "epoch": 0.25723295230587306, "grad_norm": 0.7963777184486389, "learning_rate": 0.0005, "loss": 2.4385998249053955, "step": 3263 }, { "epoch": 0.25731178557351203, "grad_norm": 0.8132331371307373, "learning_rate": 0.0005, "loss": 2.4746317863464355, "step": 3264 }, { "epoch": 0.25739061884115094, "grad_norm": 1.179927945137024, "learning_rate": 0.0005, "loss": 2.73921537399292, "step": 3265 }, { "epoch": 0.2574694521087899, "grad_norm": 1.0399366617202759, "learning_rate": 0.0005, "loss": 2.1359922885894775, "step": 3266 }, { "epoch": 0.2575482853764289, "grad_norm": 0.9237977266311646, "learning_rate": 0.0005, "loss": 2.591921329498291, "step": 3267 }, { "epoch": 0.2576271186440678, "grad_norm": 1.0333256721496582, "learning_rate": 0.0005, "loss": 2.6792712211608887, "step": 3268 }, { "epoch": 0.25770595191170675, "grad_norm": 0.9732509255409241, "learning_rate": 0.0005, "loss": 2.3286819458007812, "step": 3269 }, { "epoch": 0.25778478517934567, "grad_norm": 1.0669236183166504, "learning_rate": 0.0005, "loss": 2.227212429046631, "step": 3270 }, { "epoch": 0.25786361844698463, "grad_norm": 0.8856804370880127, "learning_rate": 0.0005, "loss": 2.935384750366211, "step": 3271 }, { "epoch": 0.25794245171462354, "grad_norm": 0.8813374042510986, "learning_rate": 0.0005, "loss": 2.051054000854492, "step": 3272 }, { "epoch": 0.2580212849822625, "grad_norm": 1.0390613079071045, "learning_rate": 0.0005, "loss": 2.5448951721191406, "step": 3273 }, { "epoch": 0.2581001182499015, "grad_norm": 1.0507673025131226, "learning_rate": 0.0005, "loss": 2.6829700469970703, "step": 3274 }, { "epoch": 0.2581789515175404, "grad_norm": 1.0234668254852295, "learning_rate": 0.0005, "loss": 2.5349416732788086, "step": 3275 }, { "epoch": 0.25825778478517936, "grad_norm": 1.1972068548202515, "learning_rate": 0.0005, "loss": 2.645791530609131, "step": 3276 }, { "epoch": 0.25833661805281827, "grad_norm": 0.8741170167922974, "learning_rate": 0.0005, "loss": 2.561946392059326, "step": 3277 }, { "epoch": 0.25841545132045723, "grad_norm": 0.9063804149627686, "learning_rate": 0.0005, "loss": 2.5400452613830566, "step": 3278 }, { "epoch": 0.2584942845880962, "grad_norm": 0.7810316681861877, "learning_rate": 0.0005, "loss": 2.2996482849121094, "step": 3279 }, { "epoch": 0.2585731178557351, "grad_norm": 0.7255467772483826, "learning_rate": 0.0005, "loss": 2.624441146850586, "step": 3280 }, { "epoch": 0.2586519511233741, "grad_norm": 0.8075140118598938, "learning_rate": 0.0005, "loss": 2.507047653198242, "step": 3281 }, { "epoch": 0.258730784391013, "grad_norm": 0.700920820236206, "learning_rate": 0.0005, "loss": 2.6540024280548096, "step": 3282 }, { "epoch": 0.25880961765865196, "grad_norm": 0.72508704662323, "learning_rate": 0.0005, "loss": 2.39713191986084, "step": 3283 }, { "epoch": 0.25888845092629087, "grad_norm": 0.9255033135414124, "learning_rate": 0.0005, "loss": 2.4292702674865723, "step": 3284 }, { "epoch": 0.25896728419392984, "grad_norm": 1.1881753206253052, "learning_rate": 0.0005, "loss": 2.243858575820923, "step": 3285 }, { "epoch": 0.2590461174615688, "grad_norm": 0.7381860613822937, "learning_rate": 0.0005, "loss": 2.0595014095306396, "step": 3286 }, { "epoch": 0.2591249507292077, "grad_norm": 0.7300698161125183, "learning_rate": 0.0005, "loss": 2.128809690475464, "step": 3287 }, { "epoch": 0.2592037839968467, "grad_norm": 0.8626882433891296, "learning_rate": 0.0005, "loss": 2.6142160892486572, "step": 3288 }, { "epoch": 0.2592826172644856, "grad_norm": 0.7937709093093872, "learning_rate": 0.0005, "loss": 2.427682399749756, "step": 3289 }, { "epoch": 0.25936145053212456, "grad_norm": 0.7296621799468994, "learning_rate": 0.0005, "loss": 2.452867031097412, "step": 3290 }, { "epoch": 0.2594402837997635, "grad_norm": 0.8510478138923645, "learning_rate": 0.0005, "loss": 2.5442802906036377, "step": 3291 }, { "epoch": 0.25951911706740244, "grad_norm": 0.8146489262580872, "learning_rate": 0.0005, "loss": 1.9566080570220947, "step": 3292 }, { "epoch": 0.2595979503350414, "grad_norm": 0.9252827763557434, "learning_rate": 0.0005, "loss": 2.3914175033569336, "step": 3293 }, { "epoch": 0.2596767836026803, "grad_norm": 1.031363606452942, "learning_rate": 0.0005, "loss": 2.309563159942627, "step": 3294 }, { "epoch": 0.2597556168703193, "grad_norm": 0.8962223529815674, "learning_rate": 0.0005, "loss": 2.2302777767181396, "step": 3295 }, { "epoch": 0.2598344501379582, "grad_norm": 0.9298590421676636, "learning_rate": 0.0005, "loss": 2.4654417037963867, "step": 3296 }, { "epoch": 0.25991328340559716, "grad_norm": 1.1380499601364136, "learning_rate": 0.0005, "loss": 2.4553120136260986, "step": 3297 }, { "epoch": 0.25999211667323613, "grad_norm": 0.9682663083076477, "learning_rate": 0.0005, "loss": 2.586325168609619, "step": 3298 }, { "epoch": 0.26007094994087504, "grad_norm": 0.9560713768005371, "learning_rate": 0.0005, "loss": 2.57958984375, "step": 3299 }, { "epoch": 0.260149783208514, "grad_norm": 0.9584992527961731, "learning_rate": 0.0005, "loss": 2.129202365875244, "step": 3300 }, { "epoch": 0.2602286164761529, "grad_norm": 0.8726485967636108, "learning_rate": 0.0005, "loss": 2.327054977416992, "step": 3301 }, { "epoch": 0.2603074497437919, "grad_norm": 0.9045419096946716, "learning_rate": 0.0005, "loss": 2.5648412704467773, "step": 3302 }, { "epoch": 0.2603862830114308, "grad_norm": 0.776070237159729, "learning_rate": 0.0005, "loss": 2.5467076301574707, "step": 3303 }, { "epoch": 0.26046511627906976, "grad_norm": 1.225157380104065, "learning_rate": 0.0005, "loss": 2.7995057106018066, "step": 3304 }, { "epoch": 0.26054394954670873, "grad_norm": 0.8244583606719971, "learning_rate": 0.0005, "loss": 2.488464832305908, "step": 3305 }, { "epoch": 0.26062278281434764, "grad_norm": 0.9387577772140503, "learning_rate": 0.0005, "loss": 2.5772628784179688, "step": 3306 }, { "epoch": 0.2607016160819866, "grad_norm": 1.0515985488891602, "learning_rate": 0.0005, "loss": 2.379181385040283, "step": 3307 }, { "epoch": 0.2607804493496255, "grad_norm": 0.8892161250114441, "learning_rate": 0.0005, "loss": 2.767669200897217, "step": 3308 }, { "epoch": 0.2608592826172645, "grad_norm": 0.9172317981719971, "learning_rate": 0.0005, "loss": 2.4935202598571777, "step": 3309 }, { "epoch": 0.26093811588490345, "grad_norm": 0.929521918296814, "learning_rate": 0.0005, "loss": 2.515343189239502, "step": 3310 }, { "epoch": 0.26101694915254237, "grad_norm": 0.9079726338386536, "learning_rate": 0.0005, "loss": 2.328078269958496, "step": 3311 }, { "epoch": 0.26109578242018133, "grad_norm": 0.9277470707893372, "learning_rate": 0.0005, "loss": 2.58613657951355, "step": 3312 }, { "epoch": 0.26117461568782024, "grad_norm": 0.8303861021995544, "learning_rate": 0.0005, "loss": 2.3572654724121094, "step": 3313 }, { "epoch": 0.2612534489554592, "grad_norm": 0.8438393473625183, "learning_rate": 0.0005, "loss": 2.6437020301818848, "step": 3314 }, { "epoch": 0.2613322822230981, "grad_norm": 0.8278083205223083, "learning_rate": 0.0005, "loss": 2.163297653198242, "step": 3315 }, { "epoch": 0.2614111154907371, "grad_norm": 0.8137400150299072, "learning_rate": 0.0005, "loss": 2.409071922302246, "step": 3316 }, { "epoch": 0.26148994875837606, "grad_norm": 1.1852829456329346, "learning_rate": 0.0005, "loss": 2.648235321044922, "step": 3317 }, { "epoch": 0.26156878202601497, "grad_norm": 0.8685473799705505, "learning_rate": 0.0005, "loss": 2.3902063369750977, "step": 3318 }, { "epoch": 0.26164761529365393, "grad_norm": 1.0059325695037842, "learning_rate": 0.0005, "loss": 2.670379638671875, "step": 3319 }, { "epoch": 0.26172644856129285, "grad_norm": 1.0163707733154297, "learning_rate": 0.0005, "loss": 2.6114487648010254, "step": 3320 }, { "epoch": 0.2618052818289318, "grad_norm": 1.1216788291931152, "learning_rate": 0.0005, "loss": 2.4217472076416016, "step": 3321 }, { "epoch": 0.2618841150965708, "grad_norm": 0.8180319666862488, "learning_rate": 0.0005, "loss": 2.3604378700256348, "step": 3322 }, { "epoch": 0.2619629483642097, "grad_norm": 0.8482187390327454, "learning_rate": 0.0005, "loss": 2.2945642471313477, "step": 3323 }, { "epoch": 0.26204178163184866, "grad_norm": 1.0296735763549805, "learning_rate": 0.0005, "loss": 2.25618839263916, "step": 3324 }, { "epoch": 0.26212061489948757, "grad_norm": 0.8778401017189026, "learning_rate": 0.0005, "loss": 2.757765293121338, "step": 3325 }, { "epoch": 0.26219944816712654, "grad_norm": 1.0455583333969116, "learning_rate": 0.0005, "loss": 2.5282089710235596, "step": 3326 }, { "epoch": 0.26227828143476545, "grad_norm": 0.9463562965393066, "learning_rate": 0.0005, "loss": 2.522263526916504, "step": 3327 }, { "epoch": 0.2623571147024044, "grad_norm": 0.9473645687103271, "learning_rate": 0.0005, "loss": 2.229280710220337, "step": 3328 }, { "epoch": 0.2624359479700434, "grad_norm": 1.0709336996078491, "learning_rate": 0.0005, "loss": 2.28299617767334, "step": 3329 }, { "epoch": 0.2625147812376823, "grad_norm": 0.8483102321624756, "learning_rate": 0.0005, "loss": 2.3909401893615723, "step": 3330 }, { "epoch": 0.26259361450532126, "grad_norm": 1.0340434312820435, "learning_rate": 0.0005, "loss": 2.5012269020080566, "step": 3331 }, { "epoch": 0.26267244777296017, "grad_norm": 0.7947579026222229, "learning_rate": 0.0005, "loss": 2.465669870376587, "step": 3332 }, { "epoch": 0.26275128104059914, "grad_norm": 0.8078390955924988, "learning_rate": 0.0005, "loss": 2.531693458557129, "step": 3333 }, { "epoch": 0.26283011430823805, "grad_norm": 0.9295218586921692, "learning_rate": 0.0005, "loss": 2.4977853298187256, "step": 3334 }, { "epoch": 0.262908947575877, "grad_norm": 0.8070359826087952, "learning_rate": 0.0005, "loss": 2.6960129737854004, "step": 3335 }, { "epoch": 0.262987780843516, "grad_norm": 0.8059617877006531, "learning_rate": 0.0005, "loss": 2.2464699745178223, "step": 3336 }, { "epoch": 0.2630666141111549, "grad_norm": 0.8619470596313477, "learning_rate": 0.0005, "loss": 2.3690741062164307, "step": 3337 }, { "epoch": 0.26314544737879386, "grad_norm": 0.9722385406494141, "learning_rate": 0.0005, "loss": 2.576626777648926, "step": 3338 }, { "epoch": 0.2632242806464328, "grad_norm": 0.9980414509773254, "learning_rate": 0.0005, "loss": 2.454524517059326, "step": 3339 }, { "epoch": 0.26330311391407174, "grad_norm": 0.86274254322052, "learning_rate": 0.0005, "loss": 2.3382487297058105, "step": 3340 }, { "epoch": 0.2633819471817107, "grad_norm": 0.9878090620040894, "learning_rate": 0.0005, "loss": 2.6443262100219727, "step": 3341 }, { "epoch": 0.2634607804493496, "grad_norm": 0.8772668838500977, "learning_rate": 0.0005, "loss": 2.687441349029541, "step": 3342 }, { "epoch": 0.2635396137169886, "grad_norm": 0.8394492268562317, "learning_rate": 0.0005, "loss": 2.4763894081115723, "step": 3343 }, { "epoch": 0.2636184469846275, "grad_norm": 0.9727105498313904, "learning_rate": 0.0005, "loss": 2.7850446701049805, "step": 3344 }, { "epoch": 0.26369728025226646, "grad_norm": 0.9831618666648865, "learning_rate": 0.0005, "loss": 2.7379300594329834, "step": 3345 }, { "epoch": 0.2637761135199054, "grad_norm": 0.8204821944236755, "learning_rate": 0.0005, "loss": 2.1898815631866455, "step": 3346 }, { "epoch": 0.26385494678754434, "grad_norm": 1.0743321180343628, "learning_rate": 0.0005, "loss": 2.9125068187713623, "step": 3347 }, { "epoch": 0.2639337800551833, "grad_norm": 1.1249816417694092, "learning_rate": 0.0005, "loss": 2.661299705505371, "step": 3348 }, { "epoch": 0.2640126133228222, "grad_norm": 0.8526836037635803, "learning_rate": 0.0005, "loss": 2.5785417556762695, "step": 3349 }, { "epoch": 0.2640914465904612, "grad_norm": 0.8626763224601746, "learning_rate": 0.0005, "loss": 2.5961029529571533, "step": 3350 }, { "epoch": 0.2641702798581001, "grad_norm": 1.1352897882461548, "learning_rate": 0.0005, "loss": 2.8801429271698, "step": 3351 }, { "epoch": 0.26424911312573907, "grad_norm": 1.107256293296814, "learning_rate": 0.0005, "loss": 2.428274393081665, "step": 3352 }, { "epoch": 0.26432794639337803, "grad_norm": 1.6783161163330078, "learning_rate": 0.0005, "loss": 2.3485748767852783, "step": 3353 }, { "epoch": 0.26440677966101694, "grad_norm": 0.7989725470542908, "learning_rate": 0.0005, "loss": 2.3086915016174316, "step": 3354 }, { "epoch": 0.2644856129286559, "grad_norm": 0.8712736964225769, "learning_rate": 0.0005, "loss": 2.3528077602386475, "step": 3355 }, { "epoch": 0.2645644461962948, "grad_norm": 0.7619342803955078, "learning_rate": 0.0005, "loss": 2.4806621074676514, "step": 3356 }, { "epoch": 0.2646432794639338, "grad_norm": 0.9612789750099182, "learning_rate": 0.0005, "loss": 2.2617015838623047, "step": 3357 }, { "epoch": 0.2647221127315727, "grad_norm": 1.1079754829406738, "learning_rate": 0.0005, "loss": 2.964733123779297, "step": 3358 }, { "epoch": 0.26480094599921167, "grad_norm": 0.7316728234291077, "learning_rate": 0.0005, "loss": 2.4839181900024414, "step": 3359 }, { "epoch": 0.26487977926685063, "grad_norm": 1.3482552766799927, "learning_rate": 0.0005, "loss": 2.3363149166107178, "step": 3360 }, { "epoch": 0.26495861253448955, "grad_norm": 0.9739624261856079, "learning_rate": 0.0005, "loss": 2.4777908325195312, "step": 3361 }, { "epoch": 0.2650374458021285, "grad_norm": 0.9135491847991943, "learning_rate": 0.0005, "loss": 2.3838658332824707, "step": 3362 }, { "epoch": 0.2651162790697674, "grad_norm": 1.2582406997680664, "learning_rate": 0.0005, "loss": 2.219236135482788, "step": 3363 }, { "epoch": 0.2651951123374064, "grad_norm": 0.8092764019966125, "learning_rate": 0.0005, "loss": 2.440129280090332, "step": 3364 }, { "epoch": 0.2652739456050453, "grad_norm": 0.9131902456283569, "learning_rate": 0.0005, "loss": 2.324859619140625, "step": 3365 }, { "epoch": 0.26535277887268427, "grad_norm": 1.239078402519226, "learning_rate": 0.0005, "loss": 2.4436824321746826, "step": 3366 }, { "epoch": 0.26543161214032324, "grad_norm": 2.006882429122925, "learning_rate": 0.0005, "loss": 2.488173484802246, "step": 3367 }, { "epoch": 0.26551044540796215, "grad_norm": 0.9756988286972046, "learning_rate": 0.0005, "loss": 2.303905963897705, "step": 3368 }, { "epoch": 0.2655892786756011, "grad_norm": 0.9734553098678589, "learning_rate": 0.0005, "loss": 2.754307508468628, "step": 3369 }, { "epoch": 0.26566811194324, "grad_norm": 0.9141581058502197, "learning_rate": 0.0005, "loss": 2.5068087577819824, "step": 3370 }, { "epoch": 0.265746945210879, "grad_norm": 0.8604581952095032, "learning_rate": 0.0005, "loss": 1.9829189777374268, "step": 3371 }, { "epoch": 0.26582577847851796, "grad_norm": 0.872081458568573, "learning_rate": 0.0005, "loss": 2.1710801124572754, "step": 3372 }, { "epoch": 0.26590461174615687, "grad_norm": 0.9019199013710022, "learning_rate": 0.0005, "loss": 2.582080364227295, "step": 3373 }, { "epoch": 0.26598344501379584, "grad_norm": 0.8393828272819519, "learning_rate": 0.0005, "loss": 2.281558036804199, "step": 3374 }, { "epoch": 0.26606227828143475, "grad_norm": 0.8349414467811584, "learning_rate": 0.0005, "loss": 2.3399834632873535, "step": 3375 }, { "epoch": 0.2661411115490737, "grad_norm": 0.9488534927368164, "learning_rate": 0.0005, "loss": 2.1660614013671875, "step": 3376 }, { "epoch": 0.26621994481671263, "grad_norm": 0.9148468971252441, "learning_rate": 0.0005, "loss": 2.6557321548461914, "step": 3377 }, { "epoch": 0.2662987780843516, "grad_norm": 0.9697790741920471, "learning_rate": 0.0005, "loss": 2.600558280944824, "step": 3378 }, { "epoch": 0.26637761135199056, "grad_norm": 1.057217001914978, "learning_rate": 0.0005, "loss": 2.6538686752319336, "step": 3379 }, { "epoch": 0.2664564446196295, "grad_norm": 0.9434530735015869, "learning_rate": 0.0005, "loss": 3.0131678581237793, "step": 3380 }, { "epoch": 0.26653527788726844, "grad_norm": 0.8032142519950867, "learning_rate": 0.0005, "loss": 2.464453935623169, "step": 3381 }, { "epoch": 0.26661411115490735, "grad_norm": 0.9318811893463135, "learning_rate": 0.0005, "loss": 2.485011577606201, "step": 3382 }, { "epoch": 0.2666929444225463, "grad_norm": 0.8752163052558899, "learning_rate": 0.0005, "loss": 2.364838123321533, "step": 3383 }, { "epoch": 0.2667717776901853, "grad_norm": 0.9208987951278687, "learning_rate": 0.0005, "loss": 2.643484592437744, "step": 3384 }, { "epoch": 0.2668506109578242, "grad_norm": 0.9766334891319275, "learning_rate": 0.0005, "loss": 2.8274688720703125, "step": 3385 }, { "epoch": 0.26692944422546316, "grad_norm": 0.9078258872032166, "learning_rate": 0.0005, "loss": 2.43277645111084, "step": 3386 }, { "epoch": 0.2670082774931021, "grad_norm": 0.7592792510986328, "learning_rate": 0.0005, "loss": 2.2554492950439453, "step": 3387 }, { "epoch": 0.26708711076074104, "grad_norm": 0.9278425574302673, "learning_rate": 0.0005, "loss": 2.353161334991455, "step": 3388 }, { "epoch": 0.26716594402837995, "grad_norm": 1.0080537796020508, "learning_rate": 0.0005, "loss": 2.629678726196289, "step": 3389 }, { "epoch": 0.2672447772960189, "grad_norm": 1.09190833568573, "learning_rate": 0.0005, "loss": 2.2801618576049805, "step": 3390 }, { "epoch": 0.2673236105636579, "grad_norm": 1.007863998413086, "learning_rate": 0.0005, "loss": 2.4661707878112793, "step": 3391 }, { "epoch": 0.2674024438312968, "grad_norm": 0.91923987865448, "learning_rate": 0.0005, "loss": 2.133026599884033, "step": 3392 }, { "epoch": 0.26748127709893577, "grad_norm": 1.219986915588379, "learning_rate": 0.0005, "loss": 2.5070667266845703, "step": 3393 }, { "epoch": 0.2675601103665747, "grad_norm": 0.7984216213226318, "learning_rate": 0.0005, "loss": 2.4330296516418457, "step": 3394 }, { "epoch": 0.26763894363421364, "grad_norm": 0.8803085088729858, "learning_rate": 0.0005, "loss": 2.376030683517456, "step": 3395 }, { "epoch": 0.26771777690185256, "grad_norm": 0.7147819399833679, "learning_rate": 0.0005, "loss": 2.415466785430908, "step": 3396 }, { "epoch": 0.2677966101694915, "grad_norm": 0.8830305337905884, "learning_rate": 0.0005, "loss": 2.492137908935547, "step": 3397 }, { "epoch": 0.2678754434371305, "grad_norm": 0.9751675724983215, "learning_rate": 0.0005, "loss": 2.2960457801818848, "step": 3398 }, { "epoch": 0.2679542767047694, "grad_norm": 0.9225991368293762, "learning_rate": 0.0005, "loss": 2.3419642448425293, "step": 3399 }, { "epoch": 0.26803310997240837, "grad_norm": 1.2837752103805542, "learning_rate": 0.0005, "loss": 2.4440431594848633, "step": 3400 }, { "epoch": 0.2681119432400473, "grad_norm": 1.2825853824615479, "learning_rate": 0.0005, "loss": 2.421337842941284, "step": 3401 }, { "epoch": 0.26819077650768625, "grad_norm": 1.1917005777359009, "learning_rate": 0.0005, "loss": 2.5540640354156494, "step": 3402 }, { "epoch": 0.2682696097753252, "grad_norm": 0.9930981993675232, "learning_rate": 0.0005, "loss": 2.6281301975250244, "step": 3403 }, { "epoch": 0.2683484430429641, "grad_norm": 0.731403648853302, "learning_rate": 0.0005, "loss": 2.5439059734344482, "step": 3404 }, { "epoch": 0.2684272763106031, "grad_norm": 1.1513837575912476, "learning_rate": 0.0005, "loss": 2.4228224754333496, "step": 3405 }, { "epoch": 0.268506109578242, "grad_norm": 0.8023396730422974, "learning_rate": 0.0005, "loss": 2.280244827270508, "step": 3406 }, { "epoch": 0.26858494284588097, "grad_norm": 0.7542794346809387, "learning_rate": 0.0005, "loss": 2.300828218460083, "step": 3407 }, { "epoch": 0.2686637761135199, "grad_norm": 0.7254005074501038, "learning_rate": 0.0005, "loss": 2.4900882244110107, "step": 3408 }, { "epoch": 0.26874260938115885, "grad_norm": 0.7796672582626343, "learning_rate": 0.0005, "loss": 2.382857322692871, "step": 3409 }, { "epoch": 0.2688214426487978, "grad_norm": 0.9457558393478394, "learning_rate": 0.0005, "loss": 2.185990333557129, "step": 3410 }, { "epoch": 0.2689002759164367, "grad_norm": 0.8148247003555298, "learning_rate": 0.0005, "loss": 2.1638131141662598, "step": 3411 }, { "epoch": 0.2689791091840757, "grad_norm": 0.7764817476272583, "learning_rate": 0.0005, "loss": 2.5295872688293457, "step": 3412 }, { "epoch": 0.2690579424517146, "grad_norm": 1.0152338743209839, "learning_rate": 0.0005, "loss": 2.265774965286255, "step": 3413 }, { "epoch": 0.26913677571935357, "grad_norm": 0.7965238094329834, "learning_rate": 0.0005, "loss": 2.4196228981018066, "step": 3414 }, { "epoch": 0.26921560898699254, "grad_norm": 0.9244599342346191, "learning_rate": 0.0005, "loss": 2.6645095348358154, "step": 3415 }, { "epoch": 0.26929444225463145, "grad_norm": 0.8675591349601746, "learning_rate": 0.0005, "loss": 2.398221015930176, "step": 3416 }, { "epoch": 0.2693732755222704, "grad_norm": 0.8982145190238953, "learning_rate": 0.0005, "loss": 2.104468584060669, "step": 3417 }, { "epoch": 0.26945210878990933, "grad_norm": 0.9771832823753357, "learning_rate": 0.0005, "loss": 2.0731911659240723, "step": 3418 }, { "epoch": 0.2695309420575483, "grad_norm": 0.9362913966178894, "learning_rate": 0.0005, "loss": 2.558650493621826, "step": 3419 }, { "epoch": 0.2696097753251872, "grad_norm": 0.8445902466773987, "learning_rate": 0.0005, "loss": 2.4293479919433594, "step": 3420 }, { "epoch": 0.2696886085928262, "grad_norm": 0.838721513748169, "learning_rate": 0.0005, "loss": 2.473068952560425, "step": 3421 }, { "epoch": 0.26976744186046514, "grad_norm": 1.0262117385864258, "learning_rate": 0.0005, "loss": 2.4857168197631836, "step": 3422 }, { "epoch": 0.26984627512810405, "grad_norm": 0.9782107472419739, "learning_rate": 0.0005, "loss": 2.471719741821289, "step": 3423 }, { "epoch": 0.269925108395743, "grad_norm": 0.6814362406730652, "learning_rate": 0.0005, "loss": 2.3051183223724365, "step": 3424 }, { "epoch": 0.27000394166338193, "grad_norm": 0.9026811122894287, "learning_rate": 0.0005, "loss": 2.1570708751678467, "step": 3425 }, { "epoch": 0.2700827749310209, "grad_norm": 0.9497970342636108, "learning_rate": 0.0005, "loss": 2.2819983959198, "step": 3426 }, { "epoch": 0.2701616081986598, "grad_norm": 1.2999558448791504, "learning_rate": 0.0005, "loss": 2.959282398223877, "step": 3427 }, { "epoch": 0.2702404414662988, "grad_norm": 0.878330647945404, "learning_rate": 0.0005, "loss": 2.4136271476745605, "step": 3428 }, { "epoch": 0.27031927473393774, "grad_norm": 0.9473031759262085, "learning_rate": 0.0005, "loss": 2.2740960121154785, "step": 3429 }, { "epoch": 0.27039810800157665, "grad_norm": 1.0061061382293701, "learning_rate": 0.0005, "loss": 2.838088035583496, "step": 3430 }, { "epoch": 0.2704769412692156, "grad_norm": 0.9786927700042725, "learning_rate": 0.0005, "loss": 2.3424441814422607, "step": 3431 }, { "epoch": 0.27055577453685453, "grad_norm": 0.9280362129211426, "learning_rate": 0.0005, "loss": 2.691023826599121, "step": 3432 }, { "epoch": 0.2706346078044935, "grad_norm": 0.8825889825820923, "learning_rate": 0.0005, "loss": 2.414092540740967, "step": 3433 }, { "epoch": 0.27071344107213247, "grad_norm": 0.8721573352813721, "learning_rate": 0.0005, "loss": 2.178818702697754, "step": 3434 }, { "epoch": 0.2707922743397714, "grad_norm": 1.0074481964111328, "learning_rate": 0.0005, "loss": 2.7132174968719482, "step": 3435 }, { "epoch": 0.27087110760741034, "grad_norm": 0.9343231320381165, "learning_rate": 0.0005, "loss": 2.6043169498443604, "step": 3436 }, { "epoch": 0.27094994087504926, "grad_norm": 1.251859188079834, "learning_rate": 0.0005, "loss": 2.482539653778076, "step": 3437 }, { "epoch": 0.2710287741426882, "grad_norm": 0.8946673274040222, "learning_rate": 0.0005, "loss": 2.2826035022735596, "step": 3438 }, { "epoch": 0.27110760741032713, "grad_norm": 1.2204043865203857, "learning_rate": 0.0005, "loss": 2.5001211166381836, "step": 3439 }, { "epoch": 0.2711864406779661, "grad_norm": 0.8240079283714294, "learning_rate": 0.0005, "loss": 2.3460640907287598, "step": 3440 }, { "epoch": 0.27126527394560507, "grad_norm": 0.9650371670722961, "learning_rate": 0.0005, "loss": 2.307307720184326, "step": 3441 }, { "epoch": 0.271344107213244, "grad_norm": 1.0592442750930786, "learning_rate": 0.0005, "loss": 2.4306859970092773, "step": 3442 }, { "epoch": 0.27142294048088295, "grad_norm": 1.1194357872009277, "learning_rate": 0.0005, "loss": 2.3644967079162598, "step": 3443 }, { "epoch": 0.27150177374852186, "grad_norm": 0.6770416498184204, "learning_rate": 0.0005, "loss": 2.2279653549194336, "step": 3444 }, { "epoch": 0.2715806070161608, "grad_norm": 0.7626627087593079, "learning_rate": 0.0005, "loss": 2.5144448280334473, "step": 3445 }, { "epoch": 0.2716594402837998, "grad_norm": 0.9002090096473694, "learning_rate": 0.0005, "loss": 2.410250186920166, "step": 3446 }, { "epoch": 0.2717382735514387, "grad_norm": 1.0754941701889038, "learning_rate": 0.0005, "loss": 2.901658058166504, "step": 3447 }, { "epoch": 0.27181710681907767, "grad_norm": 1.1504474878311157, "learning_rate": 0.0005, "loss": 2.7776708602905273, "step": 3448 }, { "epoch": 0.2718959400867166, "grad_norm": 0.952275812625885, "learning_rate": 0.0005, "loss": 2.6483993530273438, "step": 3449 }, { "epoch": 0.27197477335435555, "grad_norm": 0.9073514938354492, "learning_rate": 0.0005, "loss": 2.4782259464263916, "step": 3450 }, { "epoch": 0.27205360662199446, "grad_norm": 0.8813735842704773, "learning_rate": 0.0005, "loss": 2.5372910499572754, "step": 3451 }, { "epoch": 0.2721324398896334, "grad_norm": 1.4309908151626587, "learning_rate": 0.0005, "loss": 2.6087427139282227, "step": 3452 }, { "epoch": 0.2722112731572724, "grad_norm": 1.1859521865844727, "learning_rate": 0.0005, "loss": 2.589992046356201, "step": 3453 }, { "epoch": 0.2722901064249113, "grad_norm": 0.8821130990982056, "learning_rate": 0.0005, "loss": 2.2720773220062256, "step": 3454 }, { "epoch": 0.27236893969255027, "grad_norm": 0.936403214931488, "learning_rate": 0.0005, "loss": 2.6170334815979004, "step": 3455 }, { "epoch": 0.2724477729601892, "grad_norm": 0.7287675142288208, "learning_rate": 0.0005, "loss": 2.6229517459869385, "step": 3456 }, { "epoch": 0.27252660622782815, "grad_norm": 0.9023610949516296, "learning_rate": 0.0005, "loss": 2.3099141120910645, "step": 3457 }, { "epoch": 0.27260543949546706, "grad_norm": 1.0618596076965332, "learning_rate": 0.0005, "loss": 2.4679198265075684, "step": 3458 }, { "epoch": 0.27268427276310603, "grad_norm": 1.1988956928253174, "learning_rate": 0.0005, "loss": 2.4362175464630127, "step": 3459 }, { "epoch": 0.272763106030745, "grad_norm": 1.1036912202835083, "learning_rate": 0.0005, "loss": 2.3956546783447266, "step": 3460 }, { "epoch": 0.2728419392983839, "grad_norm": 0.8948495388031006, "learning_rate": 0.0005, "loss": 2.3809685707092285, "step": 3461 }, { "epoch": 0.2729207725660229, "grad_norm": 0.9637414813041687, "learning_rate": 0.0005, "loss": 2.5338335037231445, "step": 3462 }, { "epoch": 0.2729996058336618, "grad_norm": 0.9569233655929565, "learning_rate": 0.0005, "loss": 2.467278480529785, "step": 3463 }, { "epoch": 0.27307843910130075, "grad_norm": 0.9180746674537659, "learning_rate": 0.0005, "loss": 2.5873208045959473, "step": 3464 }, { "epoch": 0.2731572723689397, "grad_norm": 1.0093803405761719, "learning_rate": 0.0005, "loss": 2.374027729034424, "step": 3465 }, { "epoch": 0.27323610563657863, "grad_norm": 0.9194902777671814, "learning_rate": 0.0005, "loss": 2.077862501144409, "step": 3466 }, { "epoch": 0.2733149389042176, "grad_norm": 0.9538003206253052, "learning_rate": 0.0005, "loss": 2.581324815750122, "step": 3467 }, { "epoch": 0.2733937721718565, "grad_norm": 0.80986487865448, "learning_rate": 0.0005, "loss": 2.4835565090179443, "step": 3468 }, { "epoch": 0.2734726054394955, "grad_norm": 0.9431154727935791, "learning_rate": 0.0005, "loss": 2.0240225791931152, "step": 3469 }, { "epoch": 0.2735514387071344, "grad_norm": 1.0310444831848145, "learning_rate": 0.0005, "loss": 2.525404453277588, "step": 3470 }, { "epoch": 0.27363027197477335, "grad_norm": 1.3159749507904053, "learning_rate": 0.0005, "loss": 2.5798428058624268, "step": 3471 }, { "epoch": 0.2737091052424123, "grad_norm": 0.9995427131652832, "learning_rate": 0.0005, "loss": 2.276228427886963, "step": 3472 }, { "epoch": 0.27378793851005123, "grad_norm": 0.9591735601425171, "learning_rate": 0.0005, "loss": 2.494800329208374, "step": 3473 }, { "epoch": 0.2738667717776902, "grad_norm": 0.7342767715454102, "learning_rate": 0.0005, "loss": 2.605088233947754, "step": 3474 }, { "epoch": 0.2739456050453291, "grad_norm": 0.74836665391922, "learning_rate": 0.0005, "loss": 2.515105724334717, "step": 3475 }, { "epoch": 0.2740244383129681, "grad_norm": 0.9908217787742615, "learning_rate": 0.0005, "loss": 2.3368468284606934, "step": 3476 }, { "epoch": 0.274103271580607, "grad_norm": 0.8711658120155334, "learning_rate": 0.0005, "loss": 2.501849889755249, "step": 3477 }, { "epoch": 0.27418210484824596, "grad_norm": 0.7498374581336975, "learning_rate": 0.0005, "loss": 2.220898389816284, "step": 3478 }, { "epoch": 0.2742609381158849, "grad_norm": 1.1247512102127075, "learning_rate": 0.0005, "loss": 2.3405401706695557, "step": 3479 }, { "epoch": 0.27433977138352383, "grad_norm": 0.9352458715438843, "learning_rate": 0.0005, "loss": 2.7038350105285645, "step": 3480 }, { "epoch": 0.2744186046511628, "grad_norm": 0.9540475010871887, "learning_rate": 0.0005, "loss": 2.3051064014434814, "step": 3481 }, { "epoch": 0.2744974379188017, "grad_norm": 1.0711275339126587, "learning_rate": 0.0005, "loss": 2.372859477996826, "step": 3482 }, { "epoch": 0.2745762711864407, "grad_norm": 0.8486368656158447, "learning_rate": 0.0005, "loss": 2.308695077896118, "step": 3483 }, { "epoch": 0.27465510445407965, "grad_norm": 0.9818542003631592, "learning_rate": 0.0005, "loss": 2.474876642227173, "step": 3484 }, { "epoch": 0.27473393772171856, "grad_norm": 0.9052029252052307, "learning_rate": 0.0005, "loss": 2.8732500076293945, "step": 3485 }, { "epoch": 0.2748127709893575, "grad_norm": 0.7948331832885742, "learning_rate": 0.0005, "loss": 2.4030497074127197, "step": 3486 }, { "epoch": 0.27489160425699644, "grad_norm": 1.0541950464248657, "learning_rate": 0.0005, "loss": 2.508171558380127, "step": 3487 }, { "epoch": 0.2749704375246354, "grad_norm": 0.8737506866455078, "learning_rate": 0.0005, "loss": 2.294809341430664, "step": 3488 }, { "epoch": 0.2750492707922743, "grad_norm": 0.8529887795448303, "learning_rate": 0.0005, "loss": 2.211442708969116, "step": 3489 }, { "epoch": 0.2751281040599133, "grad_norm": 0.8375335335731506, "learning_rate": 0.0005, "loss": 2.1355931758880615, "step": 3490 }, { "epoch": 0.27520693732755225, "grad_norm": 0.8972581624984741, "learning_rate": 0.0005, "loss": 2.387904167175293, "step": 3491 }, { "epoch": 0.27528577059519116, "grad_norm": 0.7907663583755493, "learning_rate": 0.0005, "loss": 2.3334615230560303, "step": 3492 }, { "epoch": 0.2753646038628301, "grad_norm": 0.7932542562484741, "learning_rate": 0.0005, "loss": 2.257518768310547, "step": 3493 }, { "epoch": 0.27544343713046904, "grad_norm": 0.822773814201355, "learning_rate": 0.0005, "loss": 2.659249782562256, "step": 3494 }, { "epoch": 0.275522270398108, "grad_norm": 0.9085580706596375, "learning_rate": 0.0005, "loss": 2.5132389068603516, "step": 3495 }, { "epoch": 0.27560110366574697, "grad_norm": 0.9565507173538208, "learning_rate": 0.0005, "loss": 2.2837634086608887, "step": 3496 }, { "epoch": 0.2756799369333859, "grad_norm": 0.8085123896598816, "learning_rate": 0.0005, "loss": 2.2786476612091064, "step": 3497 }, { "epoch": 0.27575877020102485, "grad_norm": 1.0725996494293213, "learning_rate": 0.0005, "loss": 2.4532957077026367, "step": 3498 }, { "epoch": 0.27583760346866376, "grad_norm": 1.107703685760498, "learning_rate": 0.0005, "loss": 2.290802001953125, "step": 3499 }, { "epoch": 0.27591643673630273, "grad_norm": 1.10387122631073, "learning_rate": 0.0005, "loss": 2.5885157585144043, "step": 3500 }, { "epoch": 0.27599527000394164, "grad_norm": 1.1498911380767822, "learning_rate": 0.0005, "loss": 2.6200060844421387, "step": 3501 }, { "epoch": 0.2760741032715806, "grad_norm": 1.0832825899124146, "learning_rate": 0.0005, "loss": 2.3157858848571777, "step": 3502 }, { "epoch": 0.2761529365392196, "grad_norm": 0.8479926586151123, "learning_rate": 0.0005, "loss": 2.3108394145965576, "step": 3503 }, { "epoch": 0.2762317698068585, "grad_norm": 0.7302138209342957, "learning_rate": 0.0005, "loss": 2.553670883178711, "step": 3504 }, { "epoch": 0.27631060307449745, "grad_norm": 0.9934778213500977, "learning_rate": 0.0005, "loss": 2.6225404739379883, "step": 3505 }, { "epoch": 0.27638943634213636, "grad_norm": 1.1137012243270874, "learning_rate": 0.0005, "loss": 2.1776838302612305, "step": 3506 }, { "epoch": 0.27646826960977533, "grad_norm": 0.9420970678329468, "learning_rate": 0.0005, "loss": 2.3330001831054688, "step": 3507 }, { "epoch": 0.27654710287741424, "grad_norm": 0.8310918807983398, "learning_rate": 0.0005, "loss": 2.205519676208496, "step": 3508 }, { "epoch": 0.2766259361450532, "grad_norm": 1.152437448501587, "learning_rate": 0.0005, "loss": 2.783393383026123, "step": 3509 }, { "epoch": 0.2767047694126922, "grad_norm": 1.3684130907058716, "learning_rate": 0.0005, "loss": 2.3969593048095703, "step": 3510 }, { "epoch": 0.2767836026803311, "grad_norm": 0.9079890251159668, "learning_rate": 0.0005, "loss": 2.352116584777832, "step": 3511 }, { "epoch": 0.27686243594797005, "grad_norm": 0.7317957878112793, "learning_rate": 0.0005, "loss": 2.755878210067749, "step": 3512 }, { "epoch": 0.27694126921560897, "grad_norm": 1.1522915363311768, "learning_rate": 0.0005, "loss": 2.344175338745117, "step": 3513 }, { "epoch": 0.27702010248324793, "grad_norm": 0.9790273904800415, "learning_rate": 0.0005, "loss": 2.475787878036499, "step": 3514 }, { "epoch": 0.2770989357508869, "grad_norm": 0.9678095579147339, "learning_rate": 0.0005, "loss": 2.464435577392578, "step": 3515 }, { "epoch": 0.2771777690185258, "grad_norm": 0.8012865781784058, "learning_rate": 0.0005, "loss": 2.630415439605713, "step": 3516 }, { "epoch": 0.2772566022861648, "grad_norm": 0.7865713834762573, "learning_rate": 0.0005, "loss": 2.4326043128967285, "step": 3517 }, { "epoch": 0.2773354355538037, "grad_norm": 1.1032265424728394, "learning_rate": 0.0005, "loss": 2.4062817096710205, "step": 3518 }, { "epoch": 0.27741426882144266, "grad_norm": 3.2586605548858643, "learning_rate": 0.0005, "loss": 2.12378191947937, "step": 3519 }, { "epoch": 0.27749310208908157, "grad_norm": 0.8285626769065857, "learning_rate": 0.0005, "loss": 2.2686848640441895, "step": 3520 }, { "epoch": 0.27757193535672053, "grad_norm": 0.7928875088691711, "learning_rate": 0.0005, "loss": 2.006371021270752, "step": 3521 }, { "epoch": 0.2776507686243595, "grad_norm": 0.7207289934158325, "learning_rate": 0.0005, "loss": 2.285025119781494, "step": 3522 }, { "epoch": 0.2777296018919984, "grad_norm": 0.9127210378646851, "learning_rate": 0.0005, "loss": 2.7449541091918945, "step": 3523 }, { "epoch": 0.2778084351596374, "grad_norm": 1.4723104238510132, "learning_rate": 0.0005, "loss": 2.8701248168945312, "step": 3524 }, { "epoch": 0.2778872684272763, "grad_norm": 0.8561736345291138, "learning_rate": 0.0005, "loss": 2.333582878112793, "step": 3525 }, { "epoch": 0.27796610169491526, "grad_norm": 0.9059290885925293, "learning_rate": 0.0005, "loss": 2.482311725616455, "step": 3526 }, { "epoch": 0.2780449349625542, "grad_norm": 1.7870512008666992, "learning_rate": 0.0005, "loss": 2.5821783542633057, "step": 3527 }, { "epoch": 0.27812376823019314, "grad_norm": 2.555685520172119, "learning_rate": 0.0005, "loss": 2.5520567893981934, "step": 3528 }, { "epoch": 0.2782026014978321, "grad_norm": 1.0147532224655151, "learning_rate": 0.0005, "loss": 2.580376148223877, "step": 3529 }, { "epoch": 0.278281434765471, "grad_norm": 0.861088216304779, "learning_rate": 0.0005, "loss": 2.1908626556396484, "step": 3530 }, { "epoch": 0.27836026803311, "grad_norm": 0.9867201447486877, "learning_rate": 0.0005, "loss": 2.496619701385498, "step": 3531 }, { "epoch": 0.2784391013007489, "grad_norm": 0.967750072479248, "learning_rate": 0.0005, "loss": 2.6174588203430176, "step": 3532 }, { "epoch": 0.27851793456838786, "grad_norm": 0.9034680128097534, "learning_rate": 0.0005, "loss": 2.5119712352752686, "step": 3533 }, { "epoch": 0.2785967678360268, "grad_norm": 1.0264569520950317, "learning_rate": 0.0005, "loss": 2.365424156188965, "step": 3534 }, { "epoch": 0.27867560110366574, "grad_norm": 0.7584495544433594, "learning_rate": 0.0005, "loss": 2.154646635055542, "step": 3535 }, { "epoch": 0.2787544343713047, "grad_norm": 0.8320496678352356, "learning_rate": 0.0005, "loss": 2.3586976528167725, "step": 3536 }, { "epoch": 0.2788332676389436, "grad_norm": 0.8824335932731628, "learning_rate": 0.0005, "loss": 2.296970844268799, "step": 3537 }, { "epoch": 0.2789121009065826, "grad_norm": 0.9292646050453186, "learning_rate": 0.0005, "loss": 2.1594948768615723, "step": 3538 }, { "epoch": 0.2789909341742215, "grad_norm": 1.0416855812072754, "learning_rate": 0.0005, "loss": 2.442544937133789, "step": 3539 }, { "epoch": 0.27906976744186046, "grad_norm": 1.0006848573684692, "learning_rate": 0.0005, "loss": 2.211221218109131, "step": 3540 }, { "epoch": 0.27914860070949943, "grad_norm": 1.213729739189148, "learning_rate": 0.0005, "loss": 2.7315592765808105, "step": 3541 }, { "epoch": 0.27922743397713834, "grad_norm": 0.8435391783714294, "learning_rate": 0.0005, "loss": 2.323151111602783, "step": 3542 }, { "epoch": 0.2793062672447773, "grad_norm": 0.9555612802505493, "learning_rate": 0.0005, "loss": 2.275869131088257, "step": 3543 }, { "epoch": 0.2793851005124162, "grad_norm": 0.9097525477409363, "learning_rate": 0.0005, "loss": 2.3161497116088867, "step": 3544 }, { "epoch": 0.2794639337800552, "grad_norm": 0.8209591507911682, "learning_rate": 0.0005, "loss": 2.041785717010498, "step": 3545 }, { "epoch": 0.27954276704769415, "grad_norm": 0.9747030138969421, "learning_rate": 0.0005, "loss": 2.6194190979003906, "step": 3546 }, { "epoch": 0.27962160031533306, "grad_norm": 1.0678071975708008, "learning_rate": 0.0005, "loss": 2.5739340782165527, "step": 3547 }, { "epoch": 0.27970043358297203, "grad_norm": 0.7961319088935852, "learning_rate": 0.0005, "loss": 2.1383378505706787, "step": 3548 }, { "epoch": 0.27977926685061094, "grad_norm": 1.0815329551696777, "learning_rate": 0.0005, "loss": 2.2273359298706055, "step": 3549 }, { "epoch": 0.2798581001182499, "grad_norm": 1.0440720319747925, "learning_rate": 0.0005, "loss": 2.62310791015625, "step": 3550 }, { "epoch": 0.2799369333858888, "grad_norm": 1.0344326496124268, "learning_rate": 0.0005, "loss": 2.207760810852051, "step": 3551 }, { "epoch": 0.2800157666535278, "grad_norm": 0.9237789511680603, "learning_rate": 0.0005, "loss": 2.4489994049072266, "step": 3552 }, { "epoch": 0.28009459992116675, "grad_norm": 1.4592787027359009, "learning_rate": 0.0005, "loss": 2.511516571044922, "step": 3553 }, { "epoch": 0.28017343318880567, "grad_norm": 0.9006454944610596, "learning_rate": 0.0005, "loss": 2.4855544567108154, "step": 3554 }, { "epoch": 0.28025226645644463, "grad_norm": 1.3438626527786255, "learning_rate": 0.0005, "loss": 2.7832794189453125, "step": 3555 }, { "epoch": 0.28033109972408354, "grad_norm": 0.8771314024925232, "learning_rate": 0.0005, "loss": 2.4500489234924316, "step": 3556 }, { "epoch": 0.2804099329917225, "grad_norm": 0.8382759690284729, "learning_rate": 0.0005, "loss": 2.6237101554870605, "step": 3557 }, { "epoch": 0.2804887662593615, "grad_norm": 0.7708452343940735, "learning_rate": 0.0005, "loss": 2.533639430999756, "step": 3558 }, { "epoch": 0.2805675995270004, "grad_norm": 0.7889893651008606, "learning_rate": 0.0005, "loss": 2.4193763732910156, "step": 3559 }, { "epoch": 0.28064643279463936, "grad_norm": 0.9811109304428101, "learning_rate": 0.0005, "loss": 2.083402156829834, "step": 3560 }, { "epoch": 0.28072526606227827, "grad_norm": 0.7637153267860413, "learning_rate": 0.0005, "loss": 2.7890138626098633, "step": 3561 }, { "epoch": 0.28080409932991723, "grad_norm": 0.846784770488739, "learning_rate": 0.0005, "loss": 1.951620101928711, "step": 3562 }, { "epoch": 0.28088293259755615, "grad_norm": 1.0117279291152954, "learning_rate": 0.0005, "loss": 2.346583127975464, "step": 3563 }, { "epoch": 0.2809617658651951, "grad_norm": 1.0523985624313354, "learning_rate": 0.0005, "loss": 2.486574172973633, "step": 3564 }, { "epoch": 0.2810405991328341, "grad_norm": 0.932561457157135, "learning_rate": 0.0005, "loss": 2.4873337745666504, "step": 3565 }, { "epoch": 0.281119432400473, "grad_norm": 0.8178674578666687, "learning_rate": 0.0005, "loss": 2.33420729637146, "step": 3566 }, { "epoch": 0.28119826566811196, "grad_norm": 0.8951958417892456, "learning_rate": 0.0005, "loss": 2.1430578231811523, "step": 3567 }, { "epoch": 0.28127709893575087, "grad_norm": 0.8782473206520081, "learning_rate": 0.0005, "loss": 2.743922233581543, "step": 3568 }, { "epoch": 0.28135593220338984, "grad_norm": 0.9662783145904541, "learning_rate": 0.0005, "loss": 2.3697752952575684, "step": 3569 }, { "epoch": 0.28143476547102875, "grad_norm": 0.9781176447868347, "learning_rate": 0.0005, "loss": 2.6755552291870117, "step": 3570 }, { "epoch": 0.2815135987386677, "grad_norm": 1.010319709777832, "learning_rate": 0.0005, "loss": 2.3475828170776367, "step": 3571 }, { "epoch": 0.2815924320063067, "grad_norm": 0.8170123100280762, "learning_rate": 0.0005, "loss": 2.4563183784484863, "step": 3572 }, { "epoch": 0.2816712652739456, "grad_norm": 0.8757076859474182, "learning_rate": 0.0005, "loss": 2.2699134349823, "step": 3573 }, { "epoch": 0.28175009854158456, "grad_norm": 1.2846397161483765, "learning_rate": 0.0005, "loss": 2.276383399963379, "step": 3574 }, { "epoch": 0.28182893180922347, "grad_norm": 1.1537693738937378, "learning_rate": 0.0005, "loss": 2.222937822341919, "step": 3575 }, { "epoch": 0.28190776507686244, "grad_norm": 1.1210768222808838, "learning_rate": 0.0005, "loss": 2.46804141998291, "step": 3576 }, { "epoch": 0.2819865983445014, "grad_norm": 0.9542758464813232, "learning_rate": 0.0005, "loss": 2.2829315662384033, "step": 3577 }, { "epoch": 0.2820654316121403, "grad_norm": 0.9704506993293762, "learning_rate": 0.0005, "loss": 2.2942521572113037, "step": 3578 }, { "epoch": 0.2821442648797793, "grad_norm": 1.0914658308029175, "learning_rate": 0.0005, "loss": 2.34391713142395, "step": 3579 }, { "epoch": 0.2822230981474182, "grad_norm": 0.8662554025650024, "learning_rate": 0.0005, "loss": 2.7227492332458496, "step": 3580 }, { "epoch": 0.28230193141505716, "grad_norm": 0.8572208285331726, "learning_rate": 0.0005, "loss": 1.9905356168746948, "step": 3581 }, { "epoch": 0.2823807646826961, "grad_norm": 0.8330432176589966, "learning_rate": 0.0005, "loss": 2.1993072032928467, "step": 3582 }, { "epoch": 0.28245959795033504, "grad_norm": 0.8021578788757324, "learning_rate": 0.0005, "loss": 2.3386082649230957, "step": 3583 }, { "epoch": 0.282538431217974, "grad_norm": 1.073034644126892, "learning_rate": 0.0005, "loss": 2.616580009460449, "step": 3584 }, { "epoch": 0.2826172644856129, "grad_norm": 0.7947917580604553, "learning_rate": 0.0005, "loss": 2.5040149688720703, "step": 3585 }, { "epoch": 0.2826960977532519, "grad_norm": 1.0183063745498657, "learning_rate": 0.0005, "loss": 2.198193073272705, "step": 3586 }, { "epoch": 0.2827749310208908, "grad_norm": 0.8797093033790588, "learning_rate": 0.0005, "loss": 2.4148120880126953, "step": 3587 }, { "epoch": 0.28285376428852976, "grad_norm": 0.9502537846565247, "learning_rate": 0.0005, "loss": 2.3489232063293457, "step": 3588 }, { "epoch": 0.28293259755616873, "grad_norm": 1.1688904762268066, "learning_rate": 0.0005, "loss": 2.186736583709717, "step": 3589 }, { "epoch": 0.28301143082380764, "grad_norm": 0.9089594483375549, "learning_rate": 0.0005, "loss": 2.3254477977752686, "step": 3590 }, { "epoch": 0.2830902640914466, "grad_norm": 1.5830903053283691, "learning_rate": 0.0005, "loss": 2.581212043762207, "step": 3591 }, { "epoch": 0.2831690973590855, "grad_norm": 0.7680338621139526, "learning_rate": 0.0005, "loss": 2.2842092514038086, "step": 3592 }, { "epoch": 0.2832479306267245, "grad_norm": 0.9365275502204895, "learning_rate": 0.0005, "loss": 2.1414566040039062, "step": 3593 }, { "epoch": 0.2833267638943634, "grad_norm": 0.7984825372695923, "learning_rate": 0.0005, "loss": 2.3470458984375, "step": 3594 }, { "epoch": 0.28340559716200237, "grad_norm": 0.9091466069221497, "learning_rate": 0.0005, "loss": 2.098696708679199, "step": 3595 }, { "epoch": 0.28348443042964133, "grad_norm": 0.9904232025146484, "learning_rate": 0.0005, "loss": 2.448774814605713, "step": 3596 }, { "epoch": 0.28356326369728024, "grad_norm": 0.8147302865982056, "learning_rate": 0.0005, "loss": 2.427079677581787, "step": 3597 }, { "epoch": 0.2836420969649192, "grad_norm": 0.8619142770767212, "learning_rate": 0.0005, "loss": 2.5350341796875, "step": 3598 }, { "epoch": 0.2837209302325581, "grad_norm": 0.8328386545181274, "learning_rate": 0.0005, "loss": 2.2176058292388916, "step": 3599 }, { "epoch": 0.2837997635001971, "grad_norm": 1.0156755447387695, "learning_rate": 0.0005, "loss": 2.5437145233154297, "step": 3600 }, { "epoch": 0.283878596767836, "grad_norm": 0.9812365770339966, "learning_rate": 0.0005, "loss": 2.1289596557617188, "step": 3601 }, { "epoch": 0.28395743003547497, "grad_norm": 0.8823623657226562, "learning_rate": 0.0005, "loss": 2.4994964599609375, "step": 3602 }, { "epoch": 0.28403626330311393, "grad_norm": 0.8812546730041504, "learning_rate": 0.0005, "loss": 2.263607978820801, "step": 3603 }, { "epoch": 0.28411509657075285, "grad_norm": 0.9730786681175232, "learning_rate": 0.0005, "loss": 2.356193780899048, "step": 3604 }, { "epoch": 0.2841939298383918, "grad_norm": 0.9194872975349426, "learning_rate": 0.0005, "loss": 2.5969583988189697, "step": 3605 }, { "epoch": 0.2842727631060307, "grad_norm": 0.9052698016166687, "learning_rate": 0.0005, "loss": 2.6881821155548096, "step": 3606 }, { "epoch": 0.2843515963736697, "grad_norm": 0.7195021510124207, "learning_rate": 0.0005, "loss": 2.546243906021118, "step": 3607 }, { "epoch": 0.28443042964130866, "grad_norm": 0.9184260368347168, "learning_rate": 0.0005, "loss": 2.3714118003845215, "step": 3608 }, { "epoch": 0.28450926290894757, "grad_norm": 1.0067180395126343, "learning_rate": 0.0005, "loss": 2.5409393310546875, "step": 3609 }, { "epoch": 0.28458809617658654, "grad_norm": 1.2569470405578613, "learning_rate": 0.0005, "loss": 2.3045811653137207, "step": 3610 }, { "epoch": 0.28466692944422545, "grad_norm": 0.9725938439369202, "learning_rate": 0.0005, "loss": 2.2792718410491943, "step": 3611 }, { "epoch": 0.2847457627118644, "grad_norm": 0.9428415894508362, "learning_rate": 0.0005, "loss": 2.405463695526123, "step": 3612 }, { "epoch": 0.2848245959795033, "grad_norm": 0.8580836057662964, "learning_rate": 0.0005, "loss": 2.922431707382202, "step": 3613 }, { "epoch": 0.2849034292471423, "grad_norm": 1.0522874593734741, "learning_rate": 0.0005, "loss": 2.6967504024505615, "step": 3614 }, { "epoch": 0.28498226251478126, "grad_norm": 0.9897825121879578, "learning_rate": 0.0005, "loss": 2.2205207347869873, "step": 3615 }, { "epoch": 0.28506109578242017, "grad_norm": 0.7121339440345764, "learning_rate": 0.0005, "loss": 2.6591241359710693, "step": 3616 }, { "epoch": 0.28513992905005914, "grad_norm": 0.9189836978912354, "learning_rate": 0.0005, "loss": 1.9367313385009766, "step": 3617 }, { "epoch": 0.28521876231769805, "grad_norm": 0.8957257270812988, "learning_rate": 0.0005, "loss": 2.507472276687622, "step": 3618 }, { "epoch": 0.285297595585337, "grad_norm": 1.0649824142456055, "learning_rate": 0.0005, "loss": 2.724806785583496, "step": 3619 }, { "epoch": 0.285376428852976, "grad_norm": 0.9140908718109131, "learning_rate": 0.0005, "loss": 2.6776907444000244, "step": 3620 }, { "epoch": 0.2854552621206149, "grad_norm": 0.8927987217903137, "learning_rate": 0.0005, "loss": 3.0537285804748535, "step": 3621 }, { "epoch": 0.28553409538825386, "grad_norm": 0.8937910795211792, "learning_rate": 0.0005, "loss": 2.284567356109619, "step": 3622 }, { "epoch": 0.2856129286558928, "grad_norm": 0.9388891458511353, "learning_rate": 0.0005, "loss": 2.618424892425537, "step": 3623 }, { "epoch": 0.28569176192353174, "grad_norm": 0.9753729701042175, "learning_rate": 0.0005, "loss": 2.6596035957336426, "step": 3624 }, { "epoch": 0.28577059519117065, "grad_norm": 0.7972949743270874, "learning_rate": 0.0005, "loss": 2.1165313720703125, "step": 3625 }, { "epoch": 0.2858494284588096, "grad_norm": 0.7886601686477661, "learning_rate": 0.0005, "loss": 2.3653664588928223, "step": 3626 }, { "epoch": 0.2859282617264486, "grad_norm": 0.8971284627914429, "learning_rate": 0.0005, "loss": 2.34330415725708, "step": 3627 }, { "epoch": 0.2860070949940875, "grad_norm": 1.1001800298690796, "learning_rate": 0.0005, "loss": 2.391580104827881, "step": 3628 }, { "epoch": 0.28608592826172646, "grad_norm": 1.0441263914108276, "learning_rate": 0.0005, "loss": 2.3368959426879883, "step": 3629 }, { "epoch": 0.2861647615293654, "grad_norm": 0.829235315322876, "learning_rate": 0.0005, "loss": 2.1271772384643555, "step": 3630 }, { "epoch": 0.28624359479700434, "grad_norm": 0.8289174437522888, "learning_rate": 0.0005, "loss": 2.554827928543091, "step": 3631 }, { "epoch": 0.28632242806464325, "grad_norm": 0.724997341632843, "learning_rate": 0.0005, "loss": 2.3069748878479004, "step": 3632 }, { "epoch": 0.2864012613322822, "grad_norm": 1.1402379274368286, "learning_rate": 0.0005, "loss": 2.629316568374634, "step": 3633 }, { "epoch": 0.2864800945999212, "grad_norm": 0.942696213722229, "learning_rate": 0.0005, "loss": 2.642835855484009, "step": 3634 }, { "epoch": 0.2865589278675601, "grad_norm": 0.675948977470398, "learning_rate": 0.0005, "loss": 2.23604679107666, "step": 3635 }, { "epoch": 0.28663776113519907, "grad_norm": 0.7731327414512634, "learning_rate": 0.0005, "loss": 2.1862266063690186, "step": 3636 }, { "epoch": 0.286716594402838, "grad_norm": 0.8628929257392883, "learning_rate": 0.0005, "loss": 2.5033795833587646, "step": 3637 }, { "epoch": 0.28679542767047694, "grad_norm": 0.8144524097442627, "learning_rate": 0.0005, "loss": 2.062718629837036, "step": 3638 }, { "epoch": 0.2868742609381159, "grad_norm": 1.0034679174423218, "learning_rate": 0.0005, "loss": 2.437901020050049, "step": 3639 }, { "epoch": 0.2869530942057548, "grad_norm": 1.0121700763702393, "learning_rate": 0.0005, "loss": 2.391754150390625, "step": 3640 }, { "epoch": 0.2870319274733938, "grad_norm": 0.7940613627433777, "learning_rate": 0.0005, "loss": 2.230388879776001, "step": 3641 }, { "epoch": 0.2871107607410327, "grad_norm": 0.8504449129104614, "learning_rate": 0.0005, "loss": 2.372396945953369, "step": 3642 }, { "epoch": 0.28718959400867167, "grad_norm": 1.138357162475586, "learning_rate": 0.0005, "loss": 2.305323600769043, "step": 3643 }, { "epoch": 0.2872684272763106, "grad_norm": 0.9091107249259949, "learning_rate": 0.0005, "loss": 2.4692654609680176, "step": 3644 }, { "epoch": 0.28734726054394955, "grad_norm": 1.0043761730194092, "learning_rate": 0.0005, "loss": 2.629899501800537, "step": 3645 }, { "epoch": 0.2874260938115885, "grad_norm": 1.0769513845443726, "learning_rate": 0.0005, "loss": 2.3208131790161133, "step": 3646 }, { "epoch": 0.2875049270792274, "grad_norm": 1.0663197040557861, "learning_rate": 0.0005, "loss": 2.3171725273132324, "step": 3647 }, { "epoch": 0.2875837603468664, "grad_norm": 0.908843457698822, "learning_rate": 0.0005, "loss": 2.4970898628234863, "step": 3648 }, { "epoch": 0.2876625936145053, "grad_norm": 0.8683963418006897, "learning_rate": 0.0005, "loss": 2.6038379669189453, "step": 3649 }, { "epoch": 0.28774142688214427, "grad_norm": 0.9441118836402893, "learning_rate": 0.0005, "loss": 2.7628679275512695, "step": 3650 }, { "epoch": 0.28782026014978324, "grad_norm": 1.0567342042922974, "learning_rate": 0.0005, "loss": 2.1807737350463867, "step": 3651 }, { "epoch": 0.28789909341742215, "grad_norm": 0.8253458738327026, "learning_rate": 0.0005, "loss": 2.114238739013672, "step": 3652 }, { "epoch": 0.2879779266850611, "grad_norm": 1.3819215297698975, "learning_rate": 0.0005, "loss": 2.200373649597168, "step": 3653 }, { "epoch": 0.2880567599527, "grad_norm": 1.3549814224243164, "learning_rate": 0.0005, "loss": 2.517216205596924, "step": 3654 }, { "epoch": 0.288135593220339, "grad_norm": 1.3446049690246582, "learning_rate": 0.0005, "loss": 2.4839420318603516, "step": 3655 }, { "epoch": 0.2882144264879779, "grad_norm": 0.7829211354255676, "learning_rate": 0.0005, "loss": 2.6305551528930664, "step": 3656 }, { "epoch": 0.28829325975561687, "grad_norm": 1.3189935684204102, "learning_rate": 0.0005, "loss": 2.6282107830047607, "step": 3657 }, { "epoch": 0.28837209302325584, "grad_norm": 1.0255359411239624, "learning_rate": 0.0005, "loss": 2.15932297706604, "step": 3658 }, { "epoch": 0.28845092629089475, "grad_norm": 1.0049197673797607, "learning_rate": 0.0005, "loss": 2.173532009124756, "step": 3659 }, { "epoch": 0.2885297595585337, "grad_norm": 1.1210126876831055, "learning_rate": 0.0005, "loss": 2.521237850189209, "step": 3660 }, { "epoch": 0.2886085928261726, "grad_norm": 0.9130663871765137, "learning_rate": 0.0005, "loss": 2.410365581512451, "step": 3661 }, { "epoch": 0.2886874260938116, "grad_norm": 0.9424894452095032, "learning_rate": 0.0005, "loss": 2.3718366622924805, "step": 3662 }, { "epoch": 0.2887662593614505, "grad_norm": 0.9182581901550293, "learning_rate": 0.0005, "loss": 1.777908205986023, "step": 3663 }, { "epoch": 0.2888450926290895, "grad_norm": 1.4242678880691528, "learning_rate": 0.0005, "loss": 2.5772600173950195, "step": 3664 }, { "epoch": 0.28892392589672844, "grad_norm": 0.946256160736084, "learning_rate": 0.0005, "loss": 2.417600154876709, "step": 3665 }, { "epoch": 0.28900275916436735, "grad_norm": 0.963007390499115, "learning_rate": 0.0005, "loss": 2.6552703380584717, "step": 3666 }, { "epoch": 0.2890815924320063, "grad_norm": 0.9306038618087769, "learning_rate": 0.0005, "loss": 2.541369676589966, "step": 3667 }, { "epoch": 0.28916042569964523, "grad_norm": 1.0796658992767334, "learning_rate": 0.0005, "loss": 2.5432674884796143, "step": 3668 }, { "epoch": 0.2892392589672842, "grad_norm": 0.8308751583099365, "learning_rate": 0.0005, "loss": 2.6324462890625, "step": 3669 }, { "epoch": 0.28931809223492316, "grad_norm": 0.9824044108390808, "learning_rate": 0.0005, "loss": 2.175671100616455, "step": 3670 }, { "epoch": 0.2893969255025621, "grad_norm": 0.9693925380706787, "learning_rate": 0.0005, "loss": 2.601442337036133, "step": 3671 }, { "epoch": 0.28947575877020104, "grad_norm": 1.0082380771636963, "learning_rate": 0.0005, "loss": 2.4528398513793945, "step": 3672 }, { "epoch": 0.28955459203783995, "grad_norm": 0.8615622520446777, "learning_rate": 0.0005, "loss": 2.5855746269226074, "step": 3673 }, { "epoch": 0.2896334253054789, "grad_norm": 0.8666156530380249, "learning_rate": 0.0005, "loss": 2.1428775787353516, "step": 3674 }, { "epoch": 0.28971225857311783, "grad_norm": 0.9134936332702637, "learning_rate": 0.0005, "loss": 2.228020191192627, "step": 3675 }, { "epoch": 0.2897910918407568, "grad_norm": 1.1990848779678345, "learning_rate": 0.0005, "loss": 2.753561496734619, "step": 3676 }, { "epoch": 0.28986992510839577, "grad_norm": 0.876800000667572, "learning_rate": 0.0005, "loss": 2.524080276489258, "step": 3677 }, { "epoch": 0.2899487583760347, "grad_norm": 1.9092777967453003, "learning_rate": 0.0005, "loss": 2.420273780822754, "step": 3678 }, { "epoch": 0.29002759164367364, "grad_norm": 0.9252058863639832, "learning_rate": 0.0005, "loss": 2.4701437950134277, "step": 3679 }, { "epoch": 0.29010642491131255, "grad_norm": 0.7947568297386169, "learning_rate": 0.0005, "loss": 2.268772602081299, "step": 3680 }, { "epoch": 0.2901852581789515, "grad_norm": 1.0647677183151245, "learning_rate": 0.0005, "loss": 2.44260835647583, "step": 3681 }, { "epoch": 0.2902640914465905, "grad_norm": 0.9667165279388428, "learning_rate": 0.0005, "loss": 2.1643896102905273, "step": 3682 }, { "epoch": 0.2903429247142294, "grad_norm": 0.779683530330658, "learning_rate": 0.0005, "loss": 2.567103624343872, "step": 3683 }, { "epoch": 0.29042175798186837, "grad_norm": 0.8285120725631714, "learning_rate": 0.0005, "loss": 2.451190948486328, "step": 3684 }, { "epoch": 0.2905005912495073, "grad_norm": 0.989861786365509, "learning_rate": 0.0005, "loss": 2.7193403244018555, "step": 3685 }, { "epoch": 0.29057942451714625, "grad_norm": 1.0668936967849731, "learning_rate": 0.0005, "loss": 2.49222993850708, "step": 3686 }, { "epoch": 0.29065825778478516, "grad_norm": 0.8179040551185608, "learning_rate": 0.0005, "loss": 2.2704038619995117, "step": 3687 }, { "epoch": 0.2907370910524241, "grad_norm": 0.8103354573249817, "learning_rate": 0.0005, "loss": 2.0703554153442383, "step": 3688 }, { "epoch": 0.2908159243200631, "grad_norm": 1.0184133052825928, "learning_rate": 0.0005, "loss": 2.486711025238037, "step": 3689 }, { "epoch": 0.290894757587702, "grad_norm": 0.7954372763633728, "learning_rate": 0.0005, "loss": 2.1627321243286133, "step": 3690 }, { "epoch": 0.29097359085534097, "grad_norm": 0.7653738856315613, "learning_rate": 0.0005, "loss": 2.371652364730835, "step": 3691 }, { "epoch": 0.2910524241229799, "grad_norm": 0.8147855997085571, "learning_rate": 0.0005, "loss": 2.4237923622131348, "step": 3692 }, { "epoch": 0.29113125739061885, "grad_norm": 0.8971107602119446, "learning_rate": 0.0005, "loss": 2.3091747760772705, "step": 3693 }, { "epoch": 0.29121009065825776, "grad_norm": 1.0211026668548584, "learning_rate": 0.0005, "loss": 2.4158074855804443, "step": 3694 }, { "epoch": 0.2912889239258967, "grad_norm": 0.7507399916648865, "learning_rate": 0.0005, "loss": 2.5633654594421387, "step": 3695 }, { "epoch": 0.2913677571935357, "grad_norm": 1.0319355726242065, "learning_rate": 0.0005, "loss": 2.6787307262420654, "step": 3696 }, { "epoch": 0.2914465904611746, "grad_norm": 0.9607205390930176, "learning_rate": 0.0005, "loss": 2.168701648712158, "step": 3697 }, { "epoch": 0.29152542372881357, "grad_norm": 1.200281023979187, "learning_rate": 0.0005, "loss": 2.107150077819824, "step": 3698 }, { "epoch": 0.2916042569964525, "grad_norm": 1.0384926795959473, "learning_rate": 0.0005, "loss": 2.438987970352173, "step": 3699 }, { "epoch": 0.29168309026409145, "grad_norm": 1.6494196653366089, "learning_rate": 0.0005, "loss": 2.353895664215088, "step": 3700 }, { "epoch": 0.2917619235317304, "grad_norm": 1.0071942806243896, "learning_rate": 0.0005, "loss": 2.3059630393981934, "step": 3701 }, { "epoch": 0.2918407567993693, "grad_norm": 0.807665228843689, "learning_rate": 0.0005, "loss": 2.411436080932617, "step": 3702 }, { "epoch": 0.2919195900670083, "grad_norm": 0.8491439819335938, "learning_rate": 0.0005, "loss": 2.191878318786621, "step": 3703 }, { "epoch": 0.2919984233346472, "grad_norm": 0.7780987024307251, "learning_rate": 0.0005, "loss": 2.1543962955474854, "step": 3704 }, { "epoch": 0.2920772566022862, "grad_norm": 1.3151025772094727, "learning_rate": 0.0005, "loss": 2.564049243927002, "step": 3705 }, { "epoch": 0.2921560898699251, "grad_norm": 0.8086981773376465, "learning_rate": 0.0005, "loss": 2.5617175102233887, "step": 3706 }, { "epoch": 0.29223492313756405, "grad_norm": 0.886326253414154, "learning_rate": 0.0005, "loss": 2.547330379486084, "step": 3707 }, { "epoch": 0.292313756405203, "grad_norm": 0.8850374221801758, "learning_rate": 0.0005, "loss": 2.6054725646972656, "step": 3708 }, { "epoch": 0.29239258967284193, "grad_norm": 0.7654049396514893, "learning_rate": 0.0005, "loss": 2.2413716316223145, "step": 3709 }, { "epoch": 0.2924714229404809, "grad_norm": 0.7771105170249939, "learning_rate": 0.0005, "loss": 2.188117504119873, "step": 3710 }, { "epoch": 0.2925502562081198, "grad_norm": 0.8375852704048157, "learning_rate": 0.0005, "loss": 2.4683432579040527, "step": 3711 }, { "epoch": 0.2926290894757588, "grad_norm": 0.8993735313415527, "learning_rate": 0.0005, "loss": 2.4935269355773926, "step": 3712 }, { "epoch": 0.29270792274339774, "grad_norm": 0.9412141442298889, "learning_rate": 0.0005, "loss": 2.3224616050720215, "step": 3713 }, { "epoch": 0.29278675601103665, "grad_norm": 4.298461437225342, "learning_rate": 0.0005, "loss": 2.3507628440856934, "step": 3714 }, { "epoch": 0.2928655892786756, "grad_norm": 0.8949946761131287, "learning_rate": 0.0005, "loss": 2.430637836456299, "step": 3715 }, { "epoch": 0.29294442254631453, "grad_norm": 0.9810108542442322, "learning_rate": 0.0005, "loss": 2.365542411804199, "step": 3716 }, { "epoch": 0.2930232558139535, "grad_norm": 0.8565943837165833, "learning_rate": 0.0005, "loss": 2.4601969718933105, "step": 3717 }, { "epoch": 0.2931020890815924, "grad_norm": 1.6268113851547241, "learning_rate": 0.0005, "loss": 2.2045741081237793, "step": 3718 }, { "epoch": 0.2931809223492314, "grad_norm": 0.7451711297035217, "learning_rate": 0.0005, "loss": 2.3184561729431152, "step": 3719 }, { "epoch": 0.29325975561687034, "grad_norm": 0.8558937907218933, "learning_rate": 0.0005, "loss": 2.4659900665283203, "step": 3720 }, { "epoch": 0.29333858888450925, "grad_norm": 0.8924001455307007, "learning_rate": 0.0005, "loss": 2.3407444953918457, "step": 3721 }, { "epoch": 0.2934174221521482, "grad_norm": 0.9675099849700928, "learning_rate": 0.0005, "loss": 2.476384162902832, "step": 3722 }, { "epoch": 0.29349625541978713, "grad_norm": 2.3460521697998047, "learning_rate": 0.0005, "loss": 2.460635185241699, "step": 3723 }, { "epoch": 0.2935750886874261, "grad_norm": 1.0822677612304688, "learning_rate": 0.0005, "loss": 2.661437511444092, "step": 3724 }, { "epoch": 0.293653921955065, "grad_norm": 0.7747887372970581, "learning_rate": 0.0005, "loss": 2.0822293758392334, "step": 3725 }, { "epoch": 0.293732755222704, "grad_norm": 0.8793319463729858, "learning_rate": 0.0005, "loss": 2.3526716232299805, "step": 3726 }, { "epoch": 0.29381158849034295, "grad_norm": 0.8552988171577454, "learning_rate": 0.0005, "loss": 2.15358567237854, "step": 3727 }, { "epoch": 0.29389042175798186, "grad_norm": 1.0117547512054443, "learning_rate": 0.0005, "loss": 2.2522218227386475, "step": 3728 }, { "epoch": 0.2939692550256208, "grad_norm": 0.9937081933021545, "learning_rate": 0.0005, "loss": 2.2751379013061523, "step": 3729 }, { "epoch": 0.29404808829325974, "grad_norm": 0.8241045475006104, "learning_rate": 0.0005, "loss": 2.2167177200317383, "step": 3730 }, { "epoch": 0.2941269215608987, "grad_norm": 0.9223889112472534, "learning_rate": 0.0005, "loss": 2.190380573272705, "step": 3731 }, { "epoch": 0.29420575482853767, "grad_norm": 0.9877282977104187, "learning_rate": 0.0005, "loss": 2.3041834831237793, "step": 3732 }, { "epoch": 0.2942845880961766, "grad_norm": 0.820155918598175, "learning_rate": 0.0005, "loss": 2.4673454761505127, "step": 3733 }, { "epoch": 0.29436342136381555, "grad_norm": 0.993364155292511, "learning_rate": 0.0005, "loss": 2.429166316986084, "step": 3734 }, { "epoch": 0.29444225463145446, "grad_norm": 0.7833885550498962, "learning_rate": 0.0005, "loss": 2.4783411026000977, "step": 3735 }, { "epoch": 0.2945210878990934, "grad_norm": 1.129549503326416, "learning_rate": 0.0005, "loss": 2.1710710525512695, "step": 3736 }, { "epoch": 0.29459992116673234, "grad_norm": 0.8203480243682861, "learning_rate": 0.0005, "loss": 2.259453058242798, "step": 3737 }, { "epoch": 0.2946787544343713, "grad_norm": 0.9891650080680847, "learning_rate": 0.0005, "loss": 2.6561012268066406, "step": 3738 }, { "epoch": 0.29475758770201027, "grad_norm": 0.8578928112983704, "learning_rate": 0.0005, "loss": 2.650538921356201, "step": 3739 }, { "epoch": 0.2948364209696492, "grad_norm": 0.7887236475944519, "learning_rate": 0.0005, "loss": 2.5153892040252686, "step": 3740 }, { "epoch": 0.29491525423728815, "grad_norm": 0.7412446737289429, "learning_rate": 0.0005, "loss": 2.7703628540039062, "step": 3741 }, { "epoch": 0.29499408750492706, "grad_norm": 0.936485767364502, "learning_rate": 0.0005, "loss": 2.657597064971924, "step": 3742 }, { "epoch": 0.295072920772566, "grad_norm": 0.6835031509399414, "learning_rate": 0.0005, "loss": 2.093852996826172, "step": 3743 }, { "epoch": 0.29515175404020494, "grad_norm": 0.9958257079124451, "learning_rate": 0.0005, "loss": 2.510532855987549, "step": 3744 }, { "epoch": 0.2952305873078439, "grad_norm": 1.2319704294204712, "learning_rate": 0.0005, "loss": 2.1581296920776367, "step": 3745 }, { "epoch": 0.2953094205754829, "grad_norm": 0.9860330820083618, "learning_rate": 0.0005, "loss": 2.247124671936035, "step": 3746 }, { "epoch": 0.2953882538431218, "grad_norm": 0.7467780709266663, "learning_rate": 0.0005, "loss": 2.4675798416137695, "step": 3747 }, { "epoch": 0.29546708711076075, "grad_norm": 0.9271483421325684, "learning_rate": 0.0005, "loss": 2.1553940773010254, "step": 3748 }, { "epoch": 0.29554592037839966, "grad_norm": 0.8795661330223083, "learning_rate": 0.0005, "loss": 2.5739362239837646, "step": 3749 }, { "epoch": 0.29562475364603863, "grad_norm": 0.9023035764694214, "learning_rate": 0.0005, "loss": 2.262132167816162, "step": 3750 }, { "epoch": 0.2957035869136776, "grad_norm": 0.8781113028526306, "learning_rate": 0.0005, "loss": 2.4735257625579834, "step": 3751 }, { "epoch": 0.2957824201813165, "grad_norm": 0.8021232485771179, "learning_rate": 0.0005, "loss": 2.439786672592163, "step": 3752 }, { "epoch": 0.2958612534489555, "grad_norm": 0.7576350569725037, "learning_rate": 0.0005, "loss": 2.1824536323547363, "step": 3753 }, { "epoch": 0.2959400867165944, "grad_norm": 0.8943966627120972, "learning_rate": 0.0005, "loss": 2.529625177383423, "step": 3754 }, { "epoch": 0.29601891998423335, "grad_norm": 1.0645636320114136, "learning_rate": 0.0005, "loss": 2.1440634727478027, "step": 3755 }, { "epoch": 0.29609775325187226, "grad_norm": 2.653561592102051, "learning_rate": 0.0005, "loss": 2.283144950866699, "step": 3756 }, { "epoch": 0.29617658651951123, "grad_norm": 0.7346392273902893, "learning_rate": 0.0005, "loss": 2.3546252250671387, "step": 3757 }, { "epoch": 0.2962554197871502, "grad_norm": 0.8807170987129211, "learning_rate": 0.0005, "loss": 2.301135778427124, "step": 3758 }, { "epoch": 0.2963342530547891, "grad_norm": 0.7830122709274292, "learning_rate": 0.0005, "loss": 2.4103732109069824, "step": 3759 }, { "epoch": 0.2964130863224281, "grad_norm": 0.7751510143280029, "learning_rate": 0.0005, "loss": 2.5564486980438232, "step": 3760 }, { "epoch": 0.296491919590067, "grad_norm": 0.8636114001274109, "learning_rate": 0.0005, "loss": 2.6198220252990723, "step": 3761 }, { "epoch": 0.29657075285770595, "grad_norm": 0.8466020822525024, "learning_rate": 0.0005, "loss": 2.550164222717285, "step": 3762 }, { "epoch": 0.2966495861253449, "grad_norm": 0.8525172472000122, "learning_rate": 0.0005, "loss": 2.1457765102386475, "step": 3763 }, { "epoch": 0.29672841939298383, "grad_norm": 1.504265308380127, "learning_rate": 0.0005, "loss": 2.520327568054199, "step": 3764 }, { "epoch": 0.2968072526606228, "grad_norm": 0.7443748116493225, "learning_rate": 0.0005, "loss": 2.5039560794830322, "step": 3765 }, { "epoch": 0.2968860859282617, "grad_norm": 0.7943161129951477, "learning_rate": 0.0005, "loss": 2.55826473236084, "step": 3766 }, { "epoch": 0.2969649191959007, "grad_norm": 0.72699373960495, "learning_rate": 0.0005, "loss": 2.175631523132324, "step": 3767 }, { "epoch": 0.2970437524635396, "grad_norm": 0.8691933751106262, "learning_rate": 0.0005, "loss": 2.577897548675537, "step": 3768 }, { "epoch": 0.29712258573117856, "grad_norm": 0.8709051609039307, "learning_rate": 0.0005, "loss": 2.352550983428955, "step": 3769 }, { "epoch": 0.2972014189988175, "grad_norm": 0.9176205396652222, "learning_rate": 0.0005, "loss": 2.1756997108459473, "step": 3770 }, { "epoch": 0.29728025226645644, "grad_norm": 0.8428559899330139, "learning_rate": 0.0005, "loss": 2.138570547103882, "step": 3771 }, { "epoch": 0.2973590855340954, "grad_norm": 0.9927974343299866, "learning_rate": 0.0005, "loss": 2.2718734741210938, "step": 3772 }, { "epoch": 0.2974379188017343, "grad_norm": 0.9341039657592773, "learning_rate": 0.0005, "loss": 2.1787195205688477, "step": 3773 }, { "epoch": 0.2975167520693733, "grad_norm": 0.9889712929725647, "learning_rate": 0.0005, "loss": 2.35624361038208, "step": 3774 }, { "epoch": 0.2975955853370122, "grad_norm": 0.8043878078460693, "learning_rate": 0.0005, "loss": 2.4060921669006348, "step": 3775 }, { "epoch": 0.29767441860465116, "grad_norm": 0.7473666667938232, "learning_rate": 0.0005, "loss": 2.137913703918457, "step": 3776 }, { "epoch": 0.2977532518722901, "grad_norm": 0.7950655817985535, "learning_rate": 0.0005, "loss": 2.431793451309204, "step": 3777 }, { "epoch": 0.29783208513992904, "grad_norm": 0.9227300882339478, "learning_rate": 0.0005, "loss": 2.1982085704803467, "step": 3778 }, { "epoch": 0.297910918407568, "grad_norm": 0.8698317408561707, "learning_rate": 0.0005, "loss": 2.3361587524414062, "step": 3779 }, { "epoch": 0.2979897516752069, "grad_norm": 0.9408186078071594, "learning_rate": 0.0005, "loss": 2.1108555793762207, "step": 3780 }, { "epoch": 0.2980685849428459, "grad_norm": 0.8154129981994629, "learning_rate": 0.0005, "loss": 2.654942035675049, "step": 3781 }, { "epoch": 0.29814741821048485, "grad_norm": 1.3288809061050415, "learning_rate": 0.0005, "loss": 1.9658689498901367, "step": 3782 }, { "epoch": 0.29822625147812376, "grad_norm": 1.5242702960968018, "learning_rate": 0.0005, "loss": 2.3379416465759277, "step": 3783 }, { "epoch": 0.2983050847457627, "grad_norm": 0.8302115797996521, "learning_rate": 0.0005, "loss": 2.777231216430664, "step": 3784 }, { "epoch": 0.29838391801340164, "grad_norm": 0.8511932492256165, "learning_rate": 0.0005, "loss": 2.420628547668457, "step": 3785 }, { "epoch": 0.2984627512810406, "grad_norm": 1.2267284393310547, "learning_rate": 0.0005, "loss": 2.1339213848114014, "step": 3786 }, { "epoch": 0.2985415845486795, "grad_norm": 0.9474298357963562, "learning_rate": 0.0005, "loss": 2.6379833221435547, "step": 3787 }, { "epoch": 0.2986204178163185, "grad_norm": 0.8583858609199524, "learning_rate": 0.0005, "loss": 2.1559386253356934, "step": 3788 }, { "epoch": 0.29869925108395745, "grad_norm": 0.9481172561645508, "learning_rate": 0.0005, "loss": 2.1405839920043945, "step": 3789 }, { "epoch": 0.29877808435159636, "grad_norm": 1.0125830173492432, "learning_rate": 0.0005, "loss": 2.6949543952941895, "step": 3790 }, { "epoch": 0.29885691761923533, "grad_norm": 0.8357239365577698, "learning_rate": 0.0005, "loss": 2.5197720527648926, "step": 3791 }, { "epoch": 0.29893575088687424, "grad_norm": 1.0540440082550049, "learning_rate": 0.0005, "loss": 2.1735329627990723, "step": 3792 }, { "epoch": 0.2990145841545132, "grad_norm": 0.894795835018158, "learning_rate": 0.0005, "loss": 2.340334415435791, "step": 3793 }, { "epoch": 0.2990934174221522, "grad_norm": 0.8396936058998108, "learning_rate": 0.0005, "loss": 2.303842306137085, "step": 3794 }, { "epoch": 0.2991722506897911, "grad_norm": 1.4615048170089722, "learning_rate": 0.0005, "loss": 2.4319825172424316, "step": 3795 }, { "epoch": 0.29925108395743005, "grad_norm": 1.3344265222549438, "learning_rate": 0.0005, "loss": 2.3352789878845215, "step": 3796 }, { "epoch": 0.29932991722506896, "grad_norm": 1.171576976776123, "learning_rate": 0.0005, "loss": 2.602113723754883, "step": 3797 }, { "epoch": 0.29940875049270793, "grad_norm": 0.9687467217445374, "learning_rate": 0.0005, "loss": 2.7807259559631348, "step": 3798 }, { "epoch": 0.29948758376034684, "grad_norm": 0.7757846117019653, "learning_rate": 0.0005, "loss": 2.3433992862701416, "step": 3799 }, { "epoch": 0.2995664170279858, "grad_norm": 0.7990434169769287, "learning_rate": 0.0005, "loss": 2.3433773517608643, "step": 3800 }, { "epoch": 0.2996452502956248, "grad_norm": 1.098015546798706, "learning_rate": 0.0005, "loss": 2.199343204498291, "step": 3801 }, { "epoch": 0.2997240835632637, "grad_norm": 1.6350117921829224, "learning_rate": 0.0005, "loss": 2.655817985534668, "step": 3802 }, { "epoch": 0.29980291683090265, "grad_norm": 1.0020474195480347, "learning_rate": 0.0005, "loss": 2.3552651405334473, "step": 3803 }, { "epoch": 0.29988175009854157, "grad_norm": 0.90538090467453, "learning_rate": 0.0005, "loss": 2.7183377742767334, "step": 3804 }, { "epoch": 0.29996058336618053, "grad_norm": 0.9010961651802063, "learning_rate": 0.0005, "loss": 2.2848703861236572, "step": 3805 }, { "epoch": 0.30003941663381944, "grad_norm": 0.841389000415802, "learning_rate": 0.0005, "loss": 2.266136407852173, "step": 3806 }, { "epoch": 0.3001182499014584, "grad_norm": 7.100250244140625, "learning_rate": 0.0005, "loss": 2.636023998260498, "step": 3807 }, { "epoch": 0.3001970831690974, "grad_norm": 0.8682682514190674, "learning_rate": 0.0005, "loss": 2.4201767444610596, "step": 3808 }, { "epoch": 0.3002759164367363, "grad_norm": 0.8995018601417542, "learning_rate": 0.0005, "loss": 2.351036787033081, "step": 3809 }, { "epoch": 0.30035474970437526, "grad_norm": 0.7320342063903809, "learning_rate": 0.0005, "loss": 2.18693208694458, "step": 3810 }, { "epoch": 0.30043358297201417, "grad_norm": 0.7268723845481873, "learning_rate": 0.0005, "loss": 2.3583621978759766, "step": 3811 }, { "epoch": 0.30051241623965314, "grad_norm": 0.7993020415306091, "learning_rate": 0.0005, "loss": 2.5389673709869385, "step": 3812 }, { "epoch": 0.3005912495072921, "grad_norm": 0.7658891081809998, "learning_rate": 0.0005, "loss": 2.3473806381225586, "step": 3813 }, { "epoch": 0.300670082774931, "grad_norm": 0.9747141599655151, "learning_rate": 0.0005, "loss": 2.733356237411499, "step": 3814 }, { "epoch": 0.30074891604257, "grad_norm": 0.8422096371650696, "learning_rate": 0.0005, "loss": 1.861058235168457, "step": 3815 }, { "epoch": 0.3008277493102089, "grad_norm": 0.7245456576347351, "learning_rate": 0.0005, "loss": 2.4300661087036133, "step": 3816 }, { "epoch": 0.30090658257784786, "grad_norm": 0.8826534152030945, "learning_rate": 0.0005, "loss": 2.2434539794921875, "step": 3817 }, { "epoch": 0.30098541584548677, "grad_norm": 1.1909284591674805, "learning_rate": 0.0005, "loss": 2.263854742050171, "step": 3818 }, { "epoch": 0.30106424911312574, "grad_norm": 0.9918592572212219, "learning_rate": 0.0005, "loss": 2.460144519805908, "step": 3819 }, { "epoch": 0.3011430823807647, "grad_norm": 0.6988605856895447, "learning_rate": 0.0005, "loss": 2.2027809619903564, "step": 3820 }, { "epoch": 0.3012219156484036, "grad_norm": 1.0306217670440674, "learning_rate": 0.0005, "loss": 2.1530799865722656, "step": 3821 }, { "epoch": 0.3013007489160426, "grad_norm": 0.7120075225830078, "learning_rate": 0.0005, "loss": 2.2654542922973633, "step": 3822 }, { "epoch": 0.3013795821836815, "grad_norm": 0.9296073913574219, "learning_rate": 0.0005, "loss": 2.1088132858276367, "step": 3823 }, { "epoch": 0.30145841545132046, "grad_norm": 0.7533359527587891, "learning_rate": 0.0005, "loss": 2.4424726963043213, "step": 3824 }, { "epoch": 0.3015372487189594, "grad_norm": 0.8340731263160706, "learning_rate": 0.0005, "loss": 2.4965765476226807, "step": 3825 }, { "epoch": 0.30161608198659834, "grad_norm": 1.0962061882019043, "learning_rate": 0.0005, "loss": 2.212390422821045, "step": 3826 }, { "epoch": 0.3016949152542373, "grad_norm": 1.0002121925354004, "learning_rate": 0.0005, "loss": 2.5601682662963867, "step": 3827 }, { "epoch": 0.3017737485218762, "grad_norm": 0.8881319761276245, "learning_rate": 0.0005, "loss": 2.3487157821655273, "step": 3828 }, { "epoch": 0.3018525817895152, "grad_norm": 0.8182167410850525, "learning_rate": 0.0005, "loss": 2.541914939880371, "step": 3829 }, { "epoch": 0.3019314150571541, "grad_norm": 0.8300646543502808, "learning_rate": 0.0005, "loss": 2.347764253616333, "step": 3830 }, { "epoch": 0.30201024832479306, "grad_norm": 1.3314156532287598, "learning_rate": 0.0005, "loss": 2.2712509632110596, "step": 3831 }, { "epoch": 0.30208908159243203, "grad_norm": 1.1537295579910278, "learning_rate": 0.0005, "loss": 2.3152358531951904, "step": 3832 }, { "epoch": 0.30216791486007094, "grad_norm": 0.8585910797119141, "learning_rate": 0.0005, "loss": 2.657148838043213, "step": 3833 }, { "epoch": 0.3022467481277099, "grad_norm": 0.8249161839485168, "learning_rate": 0.0005, "loss": 2.5361785888671875, "step": 3834 }, { "epoch": 0.3023255813953488, "grad_norm": 0.9037429094314575, "learning_rate": 0.0005, "loss": 2.1889257431030273, "step": 3835 }, { "epoch": 0.3024044146629878, "grad_norm": 0.8432182669639587, "learning_rate": 0.0005, "loss": 2.5370049476623535, "step": 3836 }, { "epoch": 0.3024832479306267, "grad_norm": 0.7999249696731567, "learning_rate": 0.0005, "loss": 2.33760666847229, "step": 3837 }, { "epoch": 0.30256208119826566, "grad_norm": 0.9832265377044678, "learning_rate": 0.0005, "loss": 2.7204196453094482, "step": 3838 }, { "epoch": 0.30264091446590463, "grad_norm": 0.87233567237854, "learning_rate": 0.0005, "loss": 2.4110236167907715, "step": 3839 }, { "epoch": 0.30271974773354354, "grad_norm": 0.8701915740966797, "learning_rate": 0.0005, "loss": 2.205336332321167, "step": 3840 }, { "epoch": 0.3027985810011825, "grad_norm": 0.8545652627944946, "learning_rate": 0.0005, "loss": 2.343122720718384, "step": 3841 }, { "epoch": 0.3028774142688214, "grad_norm": 0.9759050011634827, "learning_rate": 0.0005, "loss": 2.5609471797943115, "step": 3842 }, { "epoch": 0.3029562475364604, "grad_norm": 0.8360735774040222, "learning_rate": 0.0005, "loss": 2.2871131896972656, "step": 3843 }, { "epoch": 0.30303508080409935, "grad_norm": 0.8324751257896423, "learning_rate": 0.0005, "loss": 2.3320608139038086, "step": 3844 }, { "epoch": 0.30311391407173827, "grad_norm": 0.8705811500549316, "learning_rate": 0.0005, "loss": 2.3189079761505127, "step": 3845 }, { "epoch": 0.30319274733937723, "grad_norm": 1.2807127237319946, "learning_rate": 0.0005, "loss": 2.2106781005859375, "step": 3846 }, { "epoch": 0.30327158060701614, "grad_norm": 1.5484601259231567, "learning_rate": 0.0005, "loss": 2.606717586517334, "step": 3847 }, { "epoch": 0.3033504138746551, "grad_norm": 0.9573257565498352, "learning_rate": 0.0005, "loss": 2.364434242248535, "step": 3848 }, { "epoch": 0.303429247142294, "grad_norm": 0.9152874946594238, "learning_rate": 0.0005, "loss": 2.2007241249084473, "step": 3849 }, { "epoch": 0.303508080409933, "grad_norm": 0.9097795486450195, "learning_rate": 0.0005, "loss": 2.0829553604125977, "step": 3850 }, { "epoch": 0.30358691367757196, "grad_norm": 0.8575879335403442, "learning_rate": 0.0005, "loss": 2.0854544639587402, "step": 3851 }, { "epoch": 0.30366574694521087, "grad_norm": 1.0709514617919922, "learning_rate": 0.0005, "loss": 2.3830413818359375, "step": 3852 }, { "epoch": 0.30374458021284984, "grad_norm": 0.8439760804176331, "learning_rate": 0.0005, "loss": 2.1367599964141846, "step": 3853 }, { "epoch": 0.30382341348048875, "grad_norm": 1.223519206047058, "learning_rate": 0.0005, "loss": 2.3779425621032715, "step": 3854 }, { "epoch": 0.3039022467481277, "grad_norm": 1.420275092124939, "learning_rate": 0.0005, "loss": 2.2133164405822754, "step": 3855 }, { "epoch": 0.3039810800157667, "grad_norm": 1.0787696838378906, "learning_rate": 0.0005, "loss": 2.4128775596618652, "step": 3856 }, { "epoch": 0.3040599132834056, "grad_norm": 0.9355255365371704, "learning_rate": 0.0005, "loss": 2.522094249725342, "step": 3857 }, { "epoch": 0.30413874655104456, "grad_norm": 0.7845360636711121, "learning_rate": 0.0005, "loss": 2.4884519577026367, "step": 3858 }, { "epoch": 0.30421757981868347, "grad_norm": 0.8631874322891235, "learning_rate": 0.0005, "loss": 2.604550361633301, "step": 3859 }, { "epoch": 0.30429641308632244, "grad_norm": 0.8217267394065857, "learning_rate": 0.0005, "loss": 2.165424108505249, "step": 3860 }, { "epoch": 0.30437524635396135, "grad_norm": 0.6902757883071899, "learning_rate": 0.0005, "loss": 2.6720149517059326, "step": 3861 }, { "epoch": 0.3044540796216003, "grad_norm": 1.498105764389038, "learning_rate": 0.0005, "loss": 2.4105749130249023, "step": 3862 }, { "epoch": 0.3045329128892393, "grad_norm": 0.827583909034729, "learning_rate": 0.0005, "loss": 2.3992741107940674, "step": 3863 }, { "epoch": 0.3046117461568782, "grad_norm": 1.126312255859375, "learning_rate": 0.0005, "loss": 2.2153923511505127, "step": 3864 }, { "epoch": 0.30469057942451716, "grad_norm": 0.9769221544265747, "learning_rate": 0.0005, "loss": 2.3167684078216553, "step": 3865 }, { "epoch": 0.30476941269215607, "grad_norm": 1.1286433935165405, "learning_rate": 0.0005, "loss": 2.3970093727111816, "step": 3866 }, { "epoch": 0.30484824595979504, "grad_norm": 0.9332139492034912, "learning_rate": 0.0005, "loss": 2.3106813430786133, "step": 3867 }, { "epoch": 0.30492707922743395, "grad_norm": 0.851387083530426, "learning_rate": 0.0005, "loss": 2.723203182220459, "step": 3868 }, { "epoch": 0.3050059124950729, "grad_norm": 0.9641399383544922, "learning_rate": 0.0005, "loss": 2.441800594329834, "step": 3869 }, { "epoch": 0.3050847457627119, "grad_norm": 0.9477606415748596, "learning_rate": 0.0005, "loss": 2.5390584468841553, "step": 3870 }, { "epoch": 0.3051635790303508, "grad_norm": 0.9831064939498901, "learning_rate": 0.0005, "loss": 2.212401866912842, "step": 3871 }, { "epoch": 0.30524241229798976, "grad_norm": 0.816434383392334, "learning_rate": 0.0005, "loss": 2.560612201690674, "step": 3872 }, { "epoch": 0.3053212455656287, "grad_norm": 0.8312962651252747, "learning_rate": 0.0005, "loss": 2.5211181640625, "step": 3873 }, { "epoch": 0.30540007883326764, "grad_norm": 1.0208097696304321, "learning_rate": 0.0005, "loss": 2.2987008094787598, "step": 3874 }, { "epoch": 0.3054789121009066, "grad_norm": 1.0930116176605225, "learning_rate": 0.0005, "loss": 2.561785936355591, "step": 3875 }, { "epoch": 0.3055577453685455, "grad_norm": 0.9882996082305908, "learning_rate": 0.0005, "loss": 2.3994410037994385, "step": 3876 }, { "epoch": 0.3056365786361845, "grad_norm": 0.7235465049743652, "learning_rate": 0.0005, "loss": 2.209320545196533, "step": 3877 }, { "epoch": 0.3057154119038234, "grad_norm": 0.8700662851333618, "learning_rate": 0.0005, "loss": 2.249319553375244, "step": 3878 }, { "epoch": 0.30579424517146236, "grad_norm": 0.7226818799972534, "learning_rate": 0.0005, "loss": 2.4620490074157715, "step": 3879 }, { "epoch": 0.3058730784391013, "grad_norm": 0.8852971196174622, "learning_rate": 0.0005, "loss": 2.554321765899658, "step": 3880 }, { "epoch": 0.30595191170674024, "grad_norm": 0.7913572192192078, "learning_rate": 0.0005, "loss": 2.5103812217712402, "step": 3881 }, { "epoch": 0.3060307449743792, "grad_norm": 1.0338759422302246, "learning_rate": 0.0005, "loss": 2.078023910522461, "step": 3882 }, { "epoch": 0.3061095782420181, "grad_norm": 1.0402382612228394, "learning_rate": 0.0005, "loss": 2.586812734603882, "step": 3883 }, { "epoch": 0.3061884115096571, "grad_norm": 0.9099810719490051, "learning_rate": 0.0005, "loss": 2.4280998706817627, "step": 3884 }, { "epoch": 0.306267244777296, "grad_norm": 1.012747049331665, "learning_rate": 0.0005, "loss": 2.222033977508545, "step": 3885 }, { "epoch": 0.30634607804493497, "grad_norm": 1.1828581094741821, "learning_rate": 0.0005, "loss": 2.651611804962158, "step": 3886 }, { "epoch": 0.30642491131257393, "grad_norm": 1.1792727708816528, "learning_rate": 0.0005, "loss": 1.9659556150436401, "step": 3887 }, { "epoch": 0.30650374458021284, "grad_norm": 1.4262940883636475, "learning_rate": 0.0005, "loss": 2.68534517288208, "step": 3888 }, { "epoch": 0.3065825778478518, "grad_norm": 0.776387631893158, "learning_rate": 0.0005, "loss": 2.1912474632263184, "step": 3889 }, { "epoch": 0.3066614111154907, "grad_norm": 0.9009765386581421, "learning_rate": 0.0005, "loss": 2.404231548309326, "step": 3890 }, { "epoch": 0.3067402443831297, "grad_norm": 0.8093684911727905, "learning_rate": 0.0005, "loss": 2.151843547821045, "step": 3891 }, { "epoch": 0.3068190776507686, "grad_norm": 0.8311958312988281, "learning_rate": 0.0005, "loss": 2.715319871902466, "step": 3892 }, { "epoch": 0.30689791091840757, "grad_norm": 0.8383795619010925, "learning_rate": 0.0005, "loss": 2.4746055603027344, "step": 3893 }, { "epoch": 0.30697674418604654, "grad_norm": 0.9893234968185425, "learning_rate": 0.0005, "loss": 2.4825096130371094, "step": 3894 }, { "epoch": 0.30705557745368545, "grad_norm": 0.9166406989097595, "learning_rate": 0.0005, "loss": 2.5376429557800293, "step": 3895 }, { "epoch": 0.3071344107213244, "grad_norm": 1.1141743659973145, "learning_rate": 0.0005, "loss": 2.1637284755706787, "step": 3896 }, { "epoch": 0.3072132439889633, "grad_norm": 1.6441460847854614, "learning_rate": 0.0005, "loss": 2.3707621097564697, "step": 3897 }, { "epoch": 0.3072920772566023, "grad_norm": 1.047243356704712, "learning_rate": 0.0005, "loss": 2.4082086086273193, "step": 3898 }, { "epoch": 0.3073709105242412, "grad_norm": 1.0738506317138672, "learning_rate": 0.0005, "loss": 2.391984462738037, "step": 3899 }, { "epoch": 0.30744974379188017, "grad_norm": 1.1275615692138672, "learning_rate": 0.0005, "loss": 2.4681715965270996, "step": 3900 }, { "epoch": 0.30752857705951914, "grad_norm": 1.1641631126403809, "learning_rate": 0.0005, "loss": 2.1800851821899414, "step": 3901 }, { "epoch": 0.30760741032715805, "grad_norm": 0.9139068722724915, "learning_rate": 0.0005, "loss": 2.110630750656128, "step": 3902 }, { "epoch": 0.307686243594797, "grad_norm": 0.9891606569290161, "learning_rate": 0.0005, "loss": 2.4188804626464844, "step": 3903 }, { "epoch": 0.3077650768624359, "grad_norm": 1.0101670026779175, "learning_rate": 0.0005, "loss": 2.3678970336914062, "step": 3904 }, { "epoch": 0.3078439101300749, "grad_norm": 0.727633535861969, "learning_rate": 0.0005, "loss": 1.9433435201644897, "step": 3905 }, { "epoch": 0.30792274339771386, "grad_norm": 1.0629533529281616, "learning_rate": 0.0005, "loss": 2.908973217010498, "step": 3906 }, { "epoch": 0.30800157666535277, "grad_norm": 1.0161471366882324, "learning_rate": 0.0005, "loss": 2.7554898262023926, "step": 3907 }, { "epoch": 0.30808040993299174, "grad_norm": 1.8411431312561035, "learning_rate": 0.0005, "loss": 2.8025498390197754, "step": 3908 }, { "epoch": 0.30815924320063065, "grad_norm": 0.8824642300605774, "learning_rate": 0.0005, "loss": 2.355642795562744, "step": 3909 }, { "epoch": 0.3082380764682696, "grad_norm": 1.0332049131393433, "learning_rate": 0.0005, "loss": 2.366230010986328, "step": 3910 }, { "epoch": 0.30831690973590853, "grad_norm": 0.9768374562263489, "learning_rate": 0.0005, "loss": 2.5377087593078613, "step": 3911 }, { "epoch": 0.3083957430035475, "grad_norm": 1.1365097761154175, "learning_rate": 0.0005, "loss": 2.4971916675567627, "step": 3912 }, { "epoch": 0.30847457627118646, "grad_norm": 0.927300751209259, "learning_rate": 0.0005, "loss": 2.3871238231658936, "step": 3913 }, { "epoch": 0.3085534095388254, "grad_norm": 0.9228167533874512, "learning_rate": 0.0005, "loss": 2.5356523990631104, "step": 3914 }, { "epoch": 0.30863224280646434, "grad_norm": 0.9252374172210693, "learning_rate": 0.0005, "loss": 2.3434548377990723, "step": 3915 }, { "epoch": 0.30871107607410325, "grad_norm": 0.7292805314064026, "learning_rate": 0.0005, "loss": 2.3955984115600586, "step": 3916 }, { "epoch": 0.3087899093417422, "grad_norm": 1.138960838317871, "learning_rate": 0.0005, "loss": 2.600790500640869, "step": 3917 }, { "epoch": 0.3088687426093812, "grad_norm": 0.7174670696258545, "learning_rate": 0.0005, "loss": 2.630856513977051, "step": 3918 }, { "epoch": 0.3089475758770201, "grad_norm": 0.9633723497390747, "learning_rate": 0.0005, "loss": 2.251466751098633, "step": 3919 }, { "epoch": 0.30902640914465906, "grad_norm": 1.4328070878982544, "learning_rate": 0.0005, "loss": 2.7697386741638184, "step": 3920 }, { "epoch": 0.309105242412298, "grad_norm": 0.8250853419303894, "learning_rate": 0.0005, "loss": 2.205789804458618, "step": 3921 }, { "epoch": 0.30918407567993694, "grad_norm": 0.8139668703079224, "learning_rate": 0.0005, "loss": 2.074862003326416, "step": 3922 }, { "epoch": 0.30926290894757585, "grad_norm": 1.2251546382904053, "learning_rate": 0.0005, "loss": 2.725210666656494, "step": 3923 }, { "epoch": 0.3093417422152148, "grad_norm": 0.9279176592826843, "learning_rate": 0.0005, "loss": 2.3861515522003174, "step": 3924 }, { "epoch": 0.3094205754828538, "grad_norm": 0.7385014891624451, "learning_rate": 0.0005, "loss": 2.0228500366210938, "step": 3925 }, { "epoch": 0.3094994087504927, "grad_norm": 0.8109090328216553, "learning_rate": 0.0005, "loss": 2.463338851928711, "step": 3926 }, { "epoch": 0.30957824201813167, "grad_norm": 1.1373012065887451, "learning_rate": 0.0005, "loss": 2.255908489227295, "step": 3927 }, { "epoch": 0.3096570752857706, "grad_norm": 0.7468518018722534, "learning_rate": 0.0005, "loss": 2.38120698928833, "step": 3928 }, { "epoch": 0.30973590855340954, "grad_norm": 0.8351835012435913, "learning_rate": 0.0005, "loss": 2.1935505867004395, "step": 3929 }, { "epoch": 0.30981474182104846, "grad_norm": 0.7715329527854919, "learning_rate": 0.0005, "loss": 2.123115062713623, "step": 3930 }, { "epoch": 0.3098935750886874, "grad_norm": 0.8691228032112122, "learning_rate": 0.0005, "loss": 2.2310447692871094, "step": 3931 }, { "epoch": 0.3099724083563264, "grad_norm": 0.9506332874298096, "learning_rate": 0.0005, "loss": 2.3716518878936768, "step": 3932 }, { "epoch": 0.3100512416239653, "grad_norm": 1.248738408088684, "learning_rate": 0.0005, "loss": 2.5453319549560547, "step": 3933 }, { "epoch": 0.31013007489160427, "grad_norm": 1.014594554901123, "learning_rate": 0.0005, "loss": 2.286226749420166, "step": 3934 }, { "epoch": 0.3102089081592432, "grad_norm": 0.9034395217895508, "learning_rate": 0.0005, "loss": 2.3381664752960205, "step": 3935 }, { "epoch": 0.31028774142688215, "grad_norm": 0.6857436895370483, "learning_rate": 0.0005, "loss": 2.5034399032592773, "step": 3936 }, { "epoch": 0.3103665746945211, "grad_norm": 0.9613934755325317, "learning_rate": 0.0005, "loss": 2.597594738006592, "step": 3937 }, { "epoch": 0.31044540796216, "grad_norm": 0.9802995920181274, "learning_rate": 0.0005, "loss": 2.41599702835083, "step": 3938 }, { "epoch": 0.310524241229799, "grad_norm": 0.7432453036308289, "learning_rate": 0.0005, "loss": 2.0855140686035156, "step": 3939 }, { "epoch": 0.3106030744974379, "grad_norm": 0.9425784945487976, "learning_rate": 0.0005, "loss": 2.5438265800476074, "step": 3940 }, { "epoch": 0.31068190776507687, "grad_norm": 0.8336599469184875, "learning_rate": 0.0005, "loss": 2.4586892127990723, "step": 3941 }, { "epoch": 0.3107607410327158, "grad_norm": 0.8783109784126282, "learning_rate": 0.0005, "loss": 2.1876440048217773, "step": 3942 }, { "epoch": 0.31083957430035475, "grad_norm": 0.948818027973175, "learning_rate": 0.0005, "loss": 1.901307225227356, "step": 3943 }, { "epoch": 0.3109184075679937, "grad_norm": 0.8127135038375854, "learning_rate": 0.0005, "loss": 2.403576612472534, "step": 3944 }, { "epoch": 0.3109972408356326, "grad_norm": 0.9760445356369019, "learning_rate": 0.0005, "loss": 2.412874460220337, "step": 3945 }, { "epoch": 0.3110760741032716, "grad_norm": 0.7993683815002441, "learning_rate": 0.0005, "loss": 2.3888397216796875, "step": 3946 }, { "epoch": 0.3111549073709105, "grad_norm": 0.8887633681297302, "learning_rate": 0.0005, "loss": 2.3668792247772217, "step": 3947 }, { "epoch": 0.31123374063854947, "grad_norm": 0.9990699291229248, "learning_rate": 0.0005, "loss": 2.421070098876953, "step": 3948 }, { "epoch": 0.31131257390618844, "grad_norm": 1.278748869895935, "learning_rate": 0.0005, "loss": 2.303999662399292, "step": 3949 }, { "epoch": 0.31139140717382735, "grad_norm": 0.9620633721351624, "learning_rate": 0.0005, "loss": 2.383225440979004, "step": 3950 }, { "epoch": 0.3114702404414663, "grad_norm": 0.8384878635406494, "learning_rate": 0.0005, "loss": 2.153082847595215, "step": 3951 }, { "epoch": 0.31154907370910523, "grad_norm": 1.0070356130599976, "learning_rate": 0.0005, "loss": 2.6289682388305664, "step": 3952 }, { "epoch": 0.3116279069767442, "grad_norm": 1.0558874607086182, "learning_rate": 0.0005, "loss": 2.2941360473632812, "step": 3953 }, { "epoch": 0.3117067402443831, "grad_norm": 1.2176839113235474, "learning_rate": 0.0005, "loss": 2.207784414291382, "step": 3954 }, { "epoch": 0.3117855735120221, "grad_norm": 0.7210554480552673, "learning_rate": 0.0005, "loss": 2.1213555335998535, "step": 3955 }, { "epoch": 0.31186440677966104, "grad_norm": 1.045502781867981, "learning_rate": 0.0005, "loss": 2.5165367126464844, "step": 3956 }, { "epoch": 0.31194324004729995, "grad_norm": 0.9070069193840027, "learning_rate": 0.0005, "loss": 2.473363161087036, "step": 3957 }, { "epoch": 0.3120220733149389, "grad_norm": 0.8612242341041565, "learning_rate": 0.0005, "loss": 2.590257167816162, "step": 3958 }, { "epoch": 0.31210090658257783, "grad_norm": 0.8525763750076294, "learning_rate": 0.0005, "loss": 2.3559699058532715, "step": 3959 }, { "epoch": 0.3121797398502168, "grad_norm": 0.9086276292800903, "learning_rate": 0.0005, "loss": 2.4262828826904297, "step": 3960 }, { "epoch": 0.3122585731178557, "grad_norm": 0.9219988584518433, "learning_rate": 0.0005, "loss": 2.4572482109069824, "step": 3961 }, { "epoch": 0.3123374063854947, "grad_norm": 0.9467925429344177, "learning_rate": 0.0005, "loss": 2.173835515975952, "step": 3962 }, { "epoch": 0.31241623965313364, "grad_norm": 1.1080981492996216, "learning_rate": 0.0005, "loss": 2.3727455139160156, "step": 3963 }, { "epoch": 0.31249507292077255, "grad_norm": 0.9495826363563538, "learning_rate": 0.0005, "loss": 2.5881006717681885, "step": 3964 }, { "epoch": 0.3125739061884115, "grad_norm": 0.9100613594055176, "learning_rate": 0.0005, "loss": 2.649956703186035, "step": 3965 }, { "epoch": 0.31265273945605043, "grad_norm": 0.7568554878234863, "learning_rate": 0.0005, "loss": 2.102700710296631, "step": 3966 }, { "epoch": 0.3127315727236894, "grad_norm": 1.070199728012085, "learning_rate": 0.0005, "loss": 2.4573423862457275, "step": 3967 }, { "epoch": 0.31281040599132837, "grad_norm": 0.7775909900665283, "learning_rate": 0.0005, "loss": 2.106785535812378, "step": 3968 }, { "epoch": 0.3128892392589673, "grad_norm": 0.83771812915802, "learning_rate": 0.0005, "loss": 2.936312198638916, "step": 3969 }, { "epoch": 0.31296807252660624, "grad_norm": 1.2218800783157349, "learning_rate": 0.0005, "loss": 2.5274994373321533, "step": 3970 }, { "epoch": 0.31304690579424516, "grad_norm": 1.291014552116394, "learning_rate": 0.0005, "loss": 2.7144603729248047, "step": 3971 }, { "epoch": 0.3131257390618841, "grad_norm": 0.8112363815307617, "learning_rate": 0.0005, "loss": 2.0867505073547363, "step": 3972 }, { "epoch": 0.31320457232952303, "grad_norm": 0.7616479396820068, "learning_rate": 0.0005, "loss": 2.0633840560913086, "step": 3973 }, { "epoch": 0.313283405597162, "grad_norm": 0.9723929762840271, "learning_rate": 0.0005, "loss": 2.250809669494629, "step": 3974 }, { "epoch": 0.31336223886480097, "grad_norm": 1.0045061111450195, "learning_rate": 0.0005, "loss": 2.232980251312256, "step": 3975 }, { "epoch": 0.3134410721324399, "grad_norm": 0.9634004831314087, "learning_rate": 0.0005, "loss": 2.4077982902526855, "step": 3976 }, { "epoch": 0.31351990540007885, "grad_norm": 0.8887698650360107, "learning_rate": 0.0005, "loss": 2.3637359142303467, "step": 3977 }, { "epoch": 0.31359873866771776, "grad_norm": 0.8388068675994873, "learning_rate": 0.0005, "loss": 2.315885066986084, "step": 3978 }, { "epoch": 0.3136775719353567, "grad_norm": 0.767788290977478, "learning_rate": 0.0005, "loss": 2.3099770545959473, "step": 3979 }, { "epoch": 0.3137564052029957, "grad_norm": 1.1123980283737183, "learning_rate": 0.0005, "loss": 2.3420326709747314, "step": 3980 }, { "epoch": 0.3138352384706346, "grad_norm": 0.8379848599433899, "learning_rate": 0.0005, "loss": 2.4434425830841064, "step": 3981 }, { "epoch": 0.31391407173827357, "grad_norm": 0.8716663718223572, "learning_rate": 0.0005, "loss": 2.102598190307617, "step": 3982 }, { "epoch": 0.3139929050059125, "grad_norm": 0.8915995359420776, "learning_rate": 0.0005, "loss": 2.2126963138580322, "step": 3983 }, { "epoch": 0.31407173827355145, "grad_norm": 0.8940473794937134, "learning_rate": 0.0005, "loss": 2.3898520469665527, "step": 3984 }, { "epoch": 0.31415057154119036, "grad_norm": 0.8944265246391296, "learning_rate": 0.0005, "loss": 2.2564358711242676, "step": 3985 }, { "epoch": 0.3142294048088293, "grad_norm": 0.8792874813079834, "learning_rate": 0.0005, "loss": 2.7257120609283447, "step": 3986 }, { "epoch": 0.3143082380764683, "grad_norm": 0.9480348825454712, "learning_rate": 0.0005, "loss": 2.373277425765991, "step": 3987 }, { "epoch": 0.3143870713441072, "grad_norm": 0.8151100277900696, "learning_rate": 0.0005, "loss": 1.9538953304290771, "step": 3988 }, { "epoch": 0.31446590461174617, "grad_norm": 0.8973047733306885, "learning_rate": 0.0005, "loss": 2.2649335861206055, "step": 3989 }, { "epoch": 0.3145447378793851, "grad_norm": 0.8190629482269287, "learning_rate": 0.0005, "loss": 2.5814740657806396, "step": 3990 }, { "epoch": 0.31462357114702405, "grad_norm": 0.8067668080329895, "learning_rate": 0.0005, "loss": 2.4220213890075684, "step": 3991 }, { "epoch": 0.31470240441466296, "grad_norm": 0.9741594791412354, "learning_rate": 0.0005, "loss": 2.2508485317230225, "step": 3992 }, { "epoch": 0.31478123768230193, "grad_norm": 2.3572404384613037, "learning_rate": 0.0005, "loss": 2.2476468086242676, "step": 3993 }, { "epoch": 0.3148600709499409, "grad_norm": 1.0194429159164429, "learning_rate": 0.0005, "loss": 2.415149211883545, "step": 3994 }, { "epoch": 0.3149389042175798, "grad_norm": 0.7133339643478394, "learning_rate": 0.0005, "loss": 2.242429256439209, "step": 3995 }, { "epoch": 0.3150177374852188, "grad_norm": 0.8982662558555603, "learning_rate": 0.0005, "loss": 2.428701639175415, "step": 3996 }, { "epoch": 0.3150965707528577, "grad_norm": 0.8154802918434143, "learning_rate": 0.0005, "loss": 2.425570011138916, "step": 3997 }, { "epoch": 0.31517540402049665, "grad_norm": 1.0147621631622314, "learning_rate": 0.0005, "loss": 2.1826894283294678, "step": 3998 }, { "epoch": 0.3152542372881356, "grad_norm": 0.9021548628807068, "learning_rate": 0.0005, "loss": 2.321403980255127, "step": 3999 }, { "epoch": 0.31533307055577453, "grad_norm": 1.0437809228897095, "learning_rate": 0.0005, "loss": 2.573711395263672, "step": 4000 }, { "epoch": 0.3154119038234135, "grad_norm": 1.111264705657959, "learning_rate": 0.0005, "loss": 2.5710396766662598, "step": 4001 }, { "epoch": 0.3154907370910524, "grad_norm": 1.2391703128814697, "learning_rate": 0.0005, "loss": 2.188819169998169, "step": 4002 }, { "epoch": 0.3155695703586914, "grad_norm": 1.0921493768692017, "learning_rate": 0.0005, "loss": 2.319700002670288, "step": 4003 }, { "epoch": 0.3156484036263303, "grad_norm": 0.8331060409545898, "learning_rate": 0.0005, "loss": 2.4800662994384766, "step": 4004 }, { "epoch": 0.31572723689396925, "grad_norm": 0.8836744427680969, "learning_rate": 0.0005, "loss": 2.196272850036621, "step": 4005 }, { "epoch": 0.3158060701616082, "grad_norm": 1.114844560623169, "learning_rate": 0.0005, "loss": 2.3099617958068848, "step": 4006 }, { "epoch": 0.31588490342924713, "grad_norm": 0.94858717918396, "learning_rate": 0.0005, "loss": 2.37465238571167, "step": 4007 }, { "epoch": 0.3159637366968861, "grad_norm": 0.8816660642623901, "learning_rate": 0.0005, "loss": 2.2750158309936523, "step": 4008 }, { "epoch": 0.316042569964525, "grad_norm": 0.9998138546943665, "learning_rate": 0.0005, "loss": 2.4614109992980957, "step": 4009 }, { "epoch": 0.316121403232164, "grad_norm": 0.7395707964897156, "learning_rate": 0.0005, "loss": 2.2378222942352295, "step": 4010 }, { "epoch": 0.3162002364998029, "grad_norm": 0.8213961720466614, "learning_rate": 0.0005, "loss": 2.3294267654418945, "step": 4011 }, { "epoch": 0.31627906976744186, "grad_norm": 0.9239826798439026, "learning_rate": 0.0005, "loss": 2.119121551513672, "step": 4012 }, { "epoch": 0.3163579030350808, "grad_norm": 0.8852728605270386, "learning_rate": 0.0005, "loss": 2.5645394325256348, "step": 4013 }, { "epoch": 0.31643673630271973, "grad_norm": 0.8342195153236389, "learning_rate": 0.0005, "loss": 2.3635597229003906, "step": 4014 }, { "epoch": 0.3165155695703587, "grad_norm": 0.8628013134002686, "learning_rate": 0.0005, "loss": 2.2641987800598145, "step": 4015 }, { "epoch": 0.3165944028379976, "grad_norm": 1.0164132118225098, "learning_rate": 0.0005, "loss": 1.9223734140396118, "step": 4016 }, { "epoch": 0.3166732361056366, "grad_norm": 0.7501916289329529, "learning_rate": 0.0005, "loss": 2.2512176036834717, "step": 4017 }, { "epoch": 0.31675206937327555, "grad_norm": 0.7097700238227844, "learning_rate": 0.0005, "loss": 2.036418914794922, "step": 4018 }, { "epoch": 0.31683090264091446, "grad_norm": 0.7797033786773682, "learning_rate": 0.0005, "loss": 2.172255754470825, "step": 4019 }, { "epoch": 0.3169097359085534, "grad_norm": 0.911017119884491, "learning_rate": 0.0005, "loss": 2.4163026809692383, "step": 4020 }, { "epoch": 0.31698856917619234, "grad_norm": 0.7585001587867737, "learning_rate": 0.0005, "loss": 1.9963841438293457, "step": 4021 }, { "epoch": 0.3170674024438313, "grad_norm": 1.1572386026382446, "learning_rate": 0.0005, "loss": 2.5906286239624023, "step": 4022 }, { "epoch": 0.3171462357114702, "grad_norm": 0.9278483390808105, "learning_rate": 0.0005, "loss": 2.251671075820923, "step": 4023 }, { "epoch": 0.3172250689791092, "grad_norm": 1.0492130517959595, "learning_rate": 0.0005, "loss": 2.3629989624023438, "step": 4024 }, { "epoch": 0.31730390224674815, "grad_norm": 0.8651338815689087, "learning_rate": 0.0005, "loss": 2.1778533458709717, "step": 4025 }, { "epoch": 0.31738273551438706, "grad_norm": 1.0342490673065186, "learning_rate": 0.0005, "loss": 2.1780498027801514, "step": 4026 }, { "epoch": 0.317461568782026, "grad_norm": 1.3477414846420288, "learning_rate": 0.0005, "loss": 2.7070398330688477, "step": 4027 }, { "epoch": 0.31754040204966494, "grad_norm": 0.9021597504615784, "learning_rate": 0.0005, "loss": 2.2933175563812256, "step": 4028 }, { "epoch": 0.3176192353173039, "grad_norm": 0.836517870426178, "learning_rate": 0.0005, "loss": 2.1869893074035645, "step": 4029 }, { "epoch": 0.31769806858494287, "grad_norm": 0.772655725479126, "learning_rate": 0.0005, "loss": 2.4724318981170654, "step": 4030 }, { "epoch": 0.3177769018525818, "grad_norm": 0.7456226348876953, "learning_rate": 0.0005, "loss": 2.1783201694488525, "step": 4031 }, { "epoch": 0.31785573512022075, "grad_norm": 0.8259648680686951, "learning_rate": 0.0005, "loss": 2.304898738861084, "step": 4032 }, { "epoch": 0.31793456838785966, "grad_norm": 1.1308716535568237, "learning_rate": 0.0005, "loss": 2.550994396209717, "step": 4033 }, { "epoch": 0.31801340165549863, "grad_norm": 1.0152878761291504, "learning_rate": 0.0005, "loss": 2.5183682441711426, "step": 4034 }, { "epoch": 0.31809223492313754, "grad_norm": 0.7995635867118835, "learning_rate": 0.0005, "loss": 2.5203285217285156, "step": 4035 }, { "epoch": 0.3181710681907765, "grad_norm": 0.838900625705719, "learning_rate": 0.0005, "loss": 2.1561551094055176, "step": 4036 }, { "epoch": 0.3182499014584155, "grad_norm": 1.0216768980026245, "learning_rate": 0.0005, "loss": 1.825283169746399, "step": 4037 }, { "epoch": 0.3183287347260544, "grad_norm": 0.8872069120407104, "learning_rate": 0.0005, "loss": 2.295844554901123, "step": 4038 }, { "epoch": 0.31840756799369335, "grad_norm": 0.9955423474311829, "learning_rate": 0.0005, "loss": 2.4391164779663086, "step": 4039 }, { "epoch": 0.31848640126133226, "grad_norm": 0.8370984792709351, "learning_rate": 0.0005, "loss": 2.781712532043457, "step": 4040 }, { "epoch": 0.31856523452897123, "grad_norm": 0.8184340596199036, "learning_rate": 0.0005, "loss": 2.46366286277771, "step": 4041 }, { "epoch": 0.31864406779661014, "grad_norm": 0.9075349569320679, "learning_rate": 0.0005, "loss": 2.2921533584594727, "step": 4042 }, { "epoch": 0.3187229010642491, "grad_norm": 0.7267945408821106, "learning_rate": 0.0005, "loss": 2.000422954559326, "step": 4043 }, { "epoch": 0.3188017343318881, "grad_norm": 0.7599294185638428, "learning_rate": 0.0005, "loss": 2.1084630489349365, "step": 4044 }, { "epoch": 0.318880567599527, "grad_norm": 0.7508640885353088, "learning_rate": 0.0005, "loss": 1.8901104927062988, "step": 4045 }, { "epoch": 0.31895940086716595, "grad_norm": 0.9132944941520691, "learning_rate": 0.0005, "loss": 2.290921926498413, "step": 4046 }, { "epoch": 0.31903823413480487, "grad_norm": 0.8930239081382751, "learning_rate": 0.0005, "loss": 2.3415887355804443, "step": 4047 }, { "epoch": 0.31911706740244383, "grad_norm": 1.0470056533813477, "learning_rate": 0.0005, "loss": 2.3238978385925293, "step": 4048 }, { "epoch": 0.3191959006700828, "grad_norm": 1.0076707601547241, "learning_rate": 0.0005, "loss": 2.142490863800049, "step": 4049 }, { "epoch": 0.3192747339377217, "grad_norm": 1.0436393022537231, "learning_rate": 0.0005, "loss": 2.175621271133423, "step": 4050 }, { "epoch": 0.3193535672053607, "grad_norm": 0.7997320890426636, "learning_rate": 0.0005, "loss": 2.16463565826416, "step": 4051 }, { "epoch": 0.3194324004729996, "grad_norm": 0.971836268901825, "learning_rate": 0.0005, "loss": 2.0887789726257324, "step": 4052 }, { "epoch": 0.31951123374063856, "grad_norm": 0.9251748323440552, "learning_rate": 0.0005, "loss": 2.4460909366607666, "step": 4053 }, { "epoch": 0.31959006700827747, "grad_norm": 0.754760205745697, "learning_rate": 0.0005, "loss": 2.2797622680664062, "step": 4054 }, { "epoch": 0.31966890027591643, "grad_norm": 0.8870622515678406, "learning_rate": 0.0005, "loss": 2.1417789459228516, "step": 4055 }, { "epoch": 0.3197477335435554, "grad_norm": 1.1508108377456665, "learning_rate": 0.0005, "loss": 2.642308235168457, "step": 4056 }, { "epoch": 0.3198265668111943, "grad_norm": 0.8887255787849426, "learning_rate": 0.0005, "loss": 2.2621312141418457, "step": 4057 }, { "epoch": 0.3199054000788333, "grad_norm": 0.7652175426483154, "learning_rate": 0.0005, "loss": 2.0890257358551025, "step": 4058 }, { "epoch": 0.3199842333464722, "grad_norm": 0.7338383793830872, "learning_rate": 0.0005, "loss": 2.3130130767822266, "step": 4059 }, { "epoch": 0.32006306661411116, "grad_norm": 0.7607131004333496, "learning_rate": 0.0005, "loss": 2.583648681640625, "step": 4060 }, { "epoch": 0.3201418998817501, "grad_norm": 0.9966383576393127, "learning_rate": 0.0005, "loss": 2.2239255905151367, "step": 4061 }, { "epoch": 0.32022073314938904, "grad_norm": 0.9585881233215332, "learning_rate": 0.0005, "loss": 2.345776081085205, "step": 4062 }, { "epoch": 0.320299566417028, "grad_norm": 0.9088306427001953, "learning_rate": 0.0005, "loss": 2.5944418907165527, "step": 4063 }, { "epoch": 0.3203783996846669, "grad_norm": 0.7469179630279541, "learning_rate": 0.0005, "loss": 2.584975004196167, "step": 4064 }, { "epoch": 0.3204572329523059, "grad_norm": 1.1041779518127441, "learning_rate": 0.0005, "loss": 2.4142496585845947, "step": 4065 }, { "epoch": 0.3205360662199448, "grad_norm": 0.9036282300949097, "learning_rate": 0.0005, "loss": 2.0703935623168945, "step": 4066 }, { "epoch": 0.32061489948758376, "grad_norm": 0.8211466073989868, "learning_rate": 0.0005, "loss": 2.252718687057495, "step": 4067 }, { "epoch": 0.3206937327552227, "grad_norm": 1.0209852457046509, "learning_rate": 0.0005, "loss": 2.299905300140381, "step": 4068 }, { "epoch": 0.32077256602286164, "grad_norm": 0.8141239881515503, "learning_rate": 0.0005, "loss": 2.042959213256836, "step": 4069 }, { "epoch": 0.3208513992905006, "grad_norm": 0.9951305985450745, "learning_rate": 0.0005, "loss": 2.2877378463745117, "step": 4070 }, { "epoch": 0.3209302325581395, "grad_norm": 1.0612565279006958, "learning_rate": 0.0005, "loss": 2.3656563758850098, "step": 4071 }, { "epoch": 0.3210090658257785, "grad_norm": 0.8851190209388733, "learning_rate": 0.0005, "loss": 2.1973624229431152, "step": 4072 }, { "epoch": 0.3210878990934174, "grad_norm": 0.7104393839836121, "learning_rate": 0.0005, "loss": 2.190389633178711, "step": 4073 }, { "epoch": 0.32116673236105636, "grad_norm": 1.1491597890853882, "learning_rate": 0.0005, "loss": 1.918642282485962, "step": 4074 }, { "epoch": 0.32124556562869533, "grad_norm": 1.105994462966919, "learning_rate": 0.0005, "loss": 2.0683975219726562, "step": 4075 }, { "epoch": 0.32132439889633424, "grad_norm": 0.8845444321632385, "learning_rate": 0.0005, "loss": 2.410421371459961, "step": 4076 }, { "epoch": 0.3214032321639732, "grad_norm": 0.8232694268226624, "learning_rate": 0.0005, "loss": 2.738762855529785, "step": 4077 }, { "epoch": 0.3214820654316121, "grad_norm": 1.0896645784378052, "learning_rate": 0.0005, "loss": 2.351006507873535, "step": 4078 }, { "epoch": 0.3215608986992511, "grad_norm": 1.1951491832733154, "learning_rate": 0.0005, "loss": 2.3001654148101807, "step": 4079 }, { "epoch": 0.32163973196689005, "grad_norm": 0.7200571894645691, "learning_rate": 0.0005, "loss": 2.253666877746582, "step": 4080 }, { "epoch": 0.32171856523452896, "grad_norm": 0.8882712125778198, "learning_rate": 0.0005, "loss": 2.668052911758423, "step": 4081 }, { "epoch": 0.32179739850216793, "grad_norm": 1.1220897436141968, "learning_rate": 0.0005, "loss": 2.317974090576172, "step": 4082 }, { "epoch": 0.32187623176980684, "grad_norm": 0.9821459650993347, "learning_rate": 0.0005, "loss": 2.641184091567993, "step": 4083 }, { "epoch": 0.3219550650374458, "grad_norm": 0.9151406288146973, "learning_rate": 0.0005, "loss": 2.3769123554229736, "step": 4084 }, { "epoch": 0.3220338983050847, "grad_norm": 1.2553786039352417, "learning_rate": 0.0005, "loss": 2.411932945251465, "step": 4085 }, { "epoch": 0.3221127315727237, "grad_norm": 0.6888280510902405, "learning_rate": 0.0005, "loss": 2.2079384326934814, "step": 4086 }, { "epoch": 0.32219156484036265, "grad_norm": 0.8381282091140747, "learning_rate": 0.0005, "loss": 2.15903902053833, "step": 4087 }, { "epoch": 0.32227039810800157, "grad_norm": 0.8642916679382324, "learning_rate": 0.0005, "loss": 2.4129886627197266, "step": 4088 }, { "epoch": 0.32234923137564053, "grad_norm": 1.0632351636886597, "learning_rate": 0.0005, "loss": 2.472045421600342, "step": 4089 }, { "epoch": 0.32242806464327944, "grad_norm": 0.9226319789886475, "learning_rate": 0.0005, "loss": 2.231356620788574, "step": 4090 }, { "epoch": 0.3225068979109184, "grad_norm": 0.9549286365509033, "learning_rate": 0.0005, "loss": 2.386427402496338, "step": 4091 }, { "epoch": 0.3225857311785574, "grad_norm": 1.435328722000122, "learning_rate": 0.0005, "loss": 2.1203653812408447, "step": 4092 }, { "epoch": 0.3226645644461963, "grad_norm": 0.7002387046813965, "learning_rate": 0.0005, "loss": 2.516327381134033, "step": 4093 }, { "epoch": 0.32274339771383526, "grad_norm": 0.8631134629249573, "learning_rate": 0.0005, "loss": 2.3081092834472656, "step": 4094 }, { "epoch": 0.32282223098147417, "grad_norm": 0.7623483538627625, "learning_rate": 0.0005, "loss": 2.3105340003967285, "step": 4095 }, { "epoch": 0.32290106424911313, "grad_norm": 1.129318356513977, "learning_rate": 0.0005, "loss": 2.1186516284942627, "step": 4096 }, { "epoch": 0.32297989751675205, "grad_norm": 0.8708477020263672, "learning_rate": 0.0005, "loss": 2.4707698822021484, "step": 4097 }, { "epoch": 0.323058730784391, "grad_norm": 0.8374446034431458, "learning_rate": 0.0005, "loss": 2.0782461166381836, "step": 4098 }, { "epoch": 0.32313756405203, "grad_norm": 1.003140926361084, "learning_rate": 0.0005, "loss": 2.2247886657714844, "step": 4099 }, { "epoch": 0.3232163973196689, "grad_norm": 1.194159984588623, "learning_rate": 0.0005, "loss": 2.3530068397521973, "step": 4100 }, { "epoch": 0.32329523058730786, "grad_norm": 0.8393344283103943, "learning_rate": 0.0005, "loss": 2.4036433696746826, "step": 4101 }, { "epoch": 0.32337406385494677, "grad_norm": 0.9509140253067017, "learning_rate": 0.0005, "loss": 2.1377182006835938, "step": 4102 }, { "epoch": 0.32345289712258574, "grad_norm": 1.006778359413147, "learning_rate": 0.0005, "loss": 2.4598588943481445, "step": 4103 }, { "epoch": 0.32353173039022465, "grad_norm": 0.917555034160614, "learning_rate": 0.0005, "loss": 2.12717342376709, "step": 4104 }, { "epoch": 0.3236105636578636, "grad_norm": 0.91770339012146, "learning_rate": 0.0005, "loss": 2.5367591381073, "step": 4105 }, { "epoch": 0.3236893969255026, "grad_norm": 1.1097360849380493, "learning_rate": 0.0005, "loss": 2.7537789344787598, "step": 4106 }, { "epoch": 0.3237682301931415, "grad_norm": 0.7282664775848389, "learning_rate": 0.0005, "loss": 2.007401704788208, "step": 4107 }, { "epoch": 0.32384706346078046, "grad_norm": 0.7887899875640869, "learning_rate": 0.0005, "loss": 2.4692513942718506, "step": 4108 }, { "epoch": 0.32392589672841937, "grad_norm": 0.9085406064987183, "learning_rate": 0.0005, "loss": 2.440173625946045, "step": 4109 }, { "epoch": 0.32400472999605834, "grad_norm": 0.9421018362045288, "learning_rate": 0.0005, "loss": 2.6386618614196777, "step": 4110 }, { "epoch": 0.3240835632636973, "grad_norm": 0.8495377898216248, "learning_rate": 0.0005, "loss": 2.3551528453826904, "step": 4111 }, { "epoch": 0.3241623965313362, "grad_norm": 0.8074390888214111, "learning_rate": 0.0005, "loss": 2.123695135116577, "step": 4112 }, { "epoch": 0.3242412297989752, "grad_norm": 1.358752965927124, "learning_rate": 0.0005, "loss": 2.249532699584961, "step": 4113 }, { "epoch": 0.3243200630666141, "grad_norm": 0.9268580079078674, "learning_rate": 0.0005, "loss": 2.3490991592407227, "step": 4114 }, { "epoch": 0.32439889633425306, "grad_norm": 0.7603405714035034, "learning_rate": 0.0005, "loss": 2.2296040058135986, "step": 4115 }, { "epoch": 0.324477729601892, "grad_norm": 0.7722649574279785, "learning_rate": 0.0005, "loss": 2.3288936614990234, "step": 4116 }, { "epoch": 0.32455656286953094, "grad_norm": 1.006981372833252, "learning_rate": 0.0005, "loss": 2.401552200317383, "step": 4117 }, { "epoch": 0.3246353961371699, "grad_norm": 0.787222146987915, "learning_rate": 0.0005, "loss": 2.4833178520202637, "step": 4118 }, { "epoch": 0.3247142294048088, "grad_norm": 0.727959394454956, "learning_rate": 0.0005, "loss": 1.9922242164611816, "step": 4119 }, { "epoch": 0.3247930626724478, "grad_norm": 0.7891901135444641, "learning_rate": 0.0005, "loss": 2.0690505504608154, "step": 4120 }, { "epoch": 0.3248718959400867, "grad_norm": 0.7437590956687927, "learning_rate": 0.0005, "loss": 2.167658805847168, "step": 4121 }, { "epoch": 0.32495072920772566, "grad_norm": 0.898516833782196, "learning_rate": 0.0005, "loss": 2.2840218544006348, "step": 4122 }, { "epoch": 0.32502956247536463, "grad_norm": 0.8334293961524963, "learning_rate": 0.0005, "loss": 2.432154417037964, "step": 4123 }, { "epoch": 0.32510839574300354, "grad_norm": 1.2370104789733887, "learning_rate": 0.0005, "loss": 2.086286783218384, "step": 4124 }, { "epoch": 0.3251872290106425, "grad_norm": 0.8173755407333374, "learning_rate": 0.0005, "loss": 2.2395853996276855, "step": 4125 }, { "epoch": 0.3252660622782814, "grad_norm": 0.7329492568969727, "learning_rate": 0.0005, "loss": 2.7489519119262695, "step": 4126 }, { "epoch": 0.3253448955459204, "grad_norm": 0.8339017033576965, "learning_rate": 0.0005, "loss": 2.6946587562561035, "step": 4127 }, { "epoch": 0.3254237288135593, "grad_norm": 0.7845823764801025, "learning_rate": 0.0005, "loss": 2.1419191360473633, "step": 4128 }, { "epoch": 0.32550256208119827, "grad_norm": 0.8709564208984375, "learning_rate": 0.0005, "loss": 2.734941244125366, "step": 4129 }, { "epoch": 0.32558139534883723, "grad_norm": 1.6732885837554932, "learning_rate": 0.0005, "loss": 2.105510711669922, "step": 4130 }, { "epoch": 0.32566022861647614, "grad_norm": 1.0726710557937622, "learning_rate": 0.0005, "loss": 2.295846462249756, "step": 4131 }, { "epoch": 0.3257390618841151, "grad_norm": 0.9160462021827698, "learning_rate": 0.0005, "loss": 2.360313892364502, "step": 4132 }, { "epoch": 0.325817895151754, "grad_norm": 0.9097951650619507, "learning_rate": 0.0005, "loss": 2.1004581451416016, "step": 4133 }, { "epoch": 0.325896728419393, "grad_norm": 1.1008206605911255, "learning_rate": 0.0005, "loss": 2.1906352043151855, "step": 4134 }, { "epoch": 0.3259755616870319, "grad_norm": 0.8389797806739807, "learning_rate": 0.0005, "loss": 2.0888967514038086, "step": 4135 }, { "epoch": 0.32605439495467087, "grad_norm": 0.8592799305915833, "learning_rate": 0.0005, "loss": 2.2009387016296387, "step": 4136 }, { "epoch": 0.32613322822230983, "grad_norm": 1.2333461046218872, "learning_rate": 0.0005, "loss": 2.096734046936035, "step": 4137 }, { "epoch": 0.32621206148994875, "grad_norm": 0.8167105913162231, "learning_rate": 0.0005, "loss": 1.911697268486023, "step": 4138 }, { "epoch": 0.3262908947575877, "grad_norm": 0.7426701784133911, "learning_rate": 0.0005, "loss": 2.257075786590576, "step": 4139 }, { "epoch": 0.3263697280252266, "grad_norm": 0.8383861184120178, "learning_rate": 0.0005, "loss": 2.182767629623413, "step": 4140 }, { "epoch": 0.3264485612928656, "grad_norm": 0.7563585042953491, "learning_rate": 0.0005, "loss": 2.0163068771362305, "step": 4141 }, { "epoch": 0.32652739456050456, "grad_norm": 0.9778832197189331, "learning_rate": 0.0005, "loss": 2.2841620445251465, "step": 4142 }, { "epoch": 0.32660622782814347, "grad_norm": 1.2151567935943604, "learning_rate": 0.0005, "loss": 2.2549190521240234, "step": 4143 }, { "epoch": 0.32668506109578244, "grad_norm": 0.8183861970901489, "learning_rate": 0.0005, "loss": 2.2764358520507812, "step": 4144 }, { "epoch": 0.32676389436342135, "grad_norm": 1.2386846542358398, "learning_rate": 0.0005, "loss": 2.1794865131378174, "step": 4145 }, { "epoch": 0.3268427276310603, "grad_norm": 1.0489892959594727, "learning_rate": 0.0005, "loss": 2.2542946338653564, "step": 4146 }, { "epoch": 0.3269215608986992, "grad_norm": 0.8469608426094055, "learning_rate": 0.0005, "loss": 2.19168758392334, "step": 4147 }, { "epoch": 0.3270003941663382, "grad_norm": 0.7380495667457581, "learning_rate": 0.0005, "loss": 1.9057666063308716, "step": 4148 }, { "epoch": 0.32707922743397716, "grad_norm": 0.981570839881897, "learning_rate": 0.0005, "loss": 2.29057240486145, "step": 4149 }, { "epoch": 0.32715806070161607, "grad_norm": 0.8723572492599487, "learning_rate": 0.0005, "loss": 2.384140729904175, "step": 4150 }, { "epoch": 0.32723689396925504, "grad_norm": 0.7758951783180237, "learning_rate": 0.0005, "loss": 2.169966220855713, "step": 4151 }, { "epoch": 0.32731572723689395, "grad_norm": 0.7838450074195862, "learning_rate": 0.0005, "loss": 1.9985551834106445, "step": 4152 }, { "epoch": 0.3273945605045329, "grad_norm": 0.9101659059524536, "learning_rate": 0.0005, "loss": 2.0829038619995117, "step": 4153 }, { "epoch": 0.3274733937721719, "grad_norm": 0.8689300417900085, "learning_rate": 0.0005, "loss": 2.3463988304138184, "step": 4154 }, { "epoch": 0.3275522270398108, "grad_norm": 1.0541563034057617, "learning_rate": 0.0005, "loss": 2.4855880737304688, "step": 4155 }, { "epoch": 0.32763106030744976, "grad_norm": 0.9686901569366455, "learning_rate": 0.0005, "loss": 2.1442723274230957, "step": 4156 }, { "epoch": 0.3277098935750887, "grad_norm": 0.7316643595695496, "learning_rate": 0.0005, "loss": 2.3003525733947754, "step": 4157 }, { "epoch": 0.32778872684272764, "grad_norm": 0.9574741721153259, "learning_rate": 0.0005, "loss": 2.0753376483917236, "step": 4158 }, { "epoch": 0.32786756011036655, "grad_norm": 1.2614779472351074, "learning_rate": 0.0005, "loss": 2.344176769256592, "step": 4159 }, { "epoch": 0.3279463933780055, "grad_norm": 0.9850409626960754, "learning_rate": 0.0005, "loss": 2.243496894836426, "step": 4160 }, { "epoch": 0.3280252266456445, "grad_norm": 0.9738880395889282, "learning_rate": 0.0005, "loss": 2.6322755813598633, "step": 4161 }, { "epoch": 0.3281040599132834, "grad_norm": 0.7896963357925415, "learning_rate": 0.0005, "loss": 2.512923240661621, "step": 4162 }, { "epoch": 0.32818289318092236, "grad_norm": 1.1451401710510254, "learning_rate": 0.0005, "loss": 2.5306177139282227, "step": 4163 }, { "epoch": 0.3282617264485613, "grad_norm": 1.632513403892517, "learning_rate": 0.0005, "loss": 2.154754161834717, "step": 4164 }, { "epoch": 0.32834055971620024, "grad_norm": 0.8058316707611084, "learning_rate": 0.0005, "loss": 2.34149432182312, "step": 4165 }, { "epoch": 0.32841939298383915, "grad_norm": 0.9766602516174316, "learning_rate": 0.0005, "loss": 2.4682564735412598, "step": 4166 }, { "epoch": 0.3284982262514781, "grad_norm": 0.972817599773407, "learning_rate": 0.0005, "loss": 2.4153358936309814, "step": 4167 }, { "epoch": 0.3285770595191171, "grad_norm": 0.8217610120773315, "learning_rate": 0.0005, "loss": 2.7075846195220947, "step": 4168 }, { "epoch": 0.328655892786756, "grad_norm": 0.8067740797996521, "learning_rate": 0.0005, "loss": 2.2899138927459717, "step": 4169 }, { "epoch": 0.32873472605439497, "grad_norm": 0.9552645683288574, "learning_rate": 0.0005, "loss": 2.3411779403686523, "step": 4170 }, { "epoch": 0.3288135593220339, "grad_norm": 0.8679389953613281, "learning_rate": 0.0005, "loss": 2.193909168243408, "step": 4171 }, { "epoch": 0.32889239258967284, "grad_norm": 0.9878363609313965, "learning_rate": 0.0005, "loss": 2.327864170074463, "step": 4172 }, { "epoch": 0.3289712258573118, "grad_norm": 0.7800814509391785, "learning_rate": 0.0005, "loss": 2.4977784156799316, "step": 4173 }, { "epoch": 0.3290500591249507, "grad_norm": 1.0099462270736694, "learning_rate": 0.0005, "loss": 2.3147189617156982, "step": 4174 }, { "epoch": 0.3291288923925897, "grad_norm": 0.894893229007721, "learning_rate": 0.0005, "loss": 2.1918482780456543, "step": 4175 }, { "epoch": 0.3292077256602286, "grad_norm": 1.4040107727050781, "learning_rate": 0.0005, "loss": 2.7518298625946045, "step": 4176 }, { "epoch": 0.32928655892786757, "grad_norm": 0.8024714589118958, "learning_rate": 0.0005, "loss": 2.311173439025879, "step": 4177 }, { "epoch": 0.3293653921955065, "grad_norm": 0.7937493920326233, "learning_rate": 0.0005, "loss": 1.9135183095932007, "step": 4178 }, { "epoch": 0.32944422546314545, "grad_norm": 1.0203566551208496, "learning_rate": 0.0005, "loss": 2.73891282081604, "step": 4179 }, { "epoch": 0.3295230587307844, "grad_norm": 0.8584861159324646, "learning_rate": 0.0005, "loss": 2.1504101753234863, "step": 4180 }, { "epoch": 0.3296018919984233, "grad_norm": 0.9088552594184875, "learning_rate": 0.0005, "loss": 2.1477410793304443, "step": 4181 }, { "epoch": 0.3296807252660623, "grad_norm": 1.0870776176452637, "learning_rate": 0.0005, "loss": 2.0089519023895264, "step": 4182 }, { "epoch": 0.3297595585337012, "grad_norm": 0.8975515961647034, "learning_rate": 0.0005, "loss": 2.332444429397583, "step": 4183 }, { "epoch": 0.32983839180134017, "grad_norm": 1.0247589349746704, "learning_rate": 0.0005, "loss": 2.4636785984039307, "step": 4184 }, { "epoch": 0.32991722506897914, "grad_norm": 1.062333106994629, "learning_rate": 0.0005, "loss": 2.2811508178710938, "step": 4185 }, { "epoch": 0.32999605833661805, "grad_norm": 0.9017295837402344, "learning_rate": 0.0005, "loss": 2.4208006858825684, "step": 4186 }, { "epoch": 0.330074891604257, "grad_norm": 0.9277485013008118, "learning_rate": 0.0005, "loss": 2.0554141998291016, "step": 4187 }, { "epoch": 0.3301537248718959, "grad_norm": 0.9185470938682556, "learning_rate": 0.0005, "loss": 2.387366771697998, "step": 4188 }, { "epoch": 0.3302325581395349, "grad_norm": 0.8023790717124939, "learning_rate": 0.0005, "loss": 2.355581045150757, "step": 4189 }, { "epoch": 0.3303113914071738, "grad_norm": 0.8403641581535339, "learning_rate": 0.0005, "loss": 2.4554529190063477, "step": 4190 }, { "epoch": 0.33039022467481277, "grad_norm": 0.7300953269004822, "learning_rate": 0.0005, "loss": 2.376903533935547, "step": 4191 }, { "epoch": 0.33046905794245174, "grad_norm": 1.038508415222168, "learning_rate": 0.0005, "loss": 2.556344509124756, "step": 4192 }, { "epoch": 0.33054789121009065, "grad_norm": 0.7306756377220154, "learning_rate": 0.0005, "loss": 2.1611239910125732, "step": 4193 }, { "epoch": 0.3306267244777296, "grad_norm": 0.8081608414649963, "learning_rate": 0.0005, "loss": 2.5137906074523926, "step": 4194 }, { "epoch": 0.33070555774536853, "grad_norm": 0.6883847117424011, "learning_rate": 0.0005, "loss": 2.214893341064453, "step": 4195 }, { "epoch": 0.3307843910130075, "grad_norm": 0.8307505249977112, "learning_rate": 0.0005, "loss": 2.2292871475219727, "step": 4196 }, { "epoch": 0.3308632242806464, "grad_norm": 5.589176177978516, "learning_rate": 0.0005, "loss": 2.850917339324951, "step": 4197 }, { "epoch": 0.3309420575482854, "grad_norm": 0.8265241384506226, "learning_rate": 0.0005, "loss": 2.241199493408203, "step": 4198 }, { "epoch": 0.33102089081592434, "grad_norm": 0.8452622294425964, "learning_rate": 0.0005, "loss": 2.3845038414001465, "step": 4199 }, { "epoch": 0.33109972408356325, "grad_norm": 0.8882436156272888, "learning_rate": 0.0005, "loss": 2.230912685394287, "step": 4200 }, { "epoch": 0.3311785573512022, "grad_norm": 0.7538256049156189, "learning_rate": 0.0005, "loss": 2.182507038116455, "step": 4201 }, { "epoch": 0.33125739061884113, "grad_norm": 0.822756290435791, "learning_rate": 0.0005, "loss": 2.184652090072632, "step": 4202 }, { "epoch": 0.3313362238864801, "grad_norm": 0.8252894878387451, "learning_rate": 0.0005, "loss": 2.1865956783294678, "step": 4203 }, { "epoch": 0.33141505715411906, "grad_norm": 0.8137761950492859, "learning_rate": 0.0005, "loss": 2.2720327377319336, "step": 4204 }, { "epoch": 0.331493890421758, "grad_norm": 1.4777233600616455, "learning_rate": 0.0005, "loss": 2.438462734222412, "step": 4205 }, { "epoch": 0.33157272368939694, "grad_norm": 0.7675942778587341, "learning_rate": 0.0005, "loss": 2.4269232749938965, "step": 4206 }, { "epoch": 0.33165155695703585, "grad_norm": 0.8445461988449097, "learning_rate": 0.0005, "loss": 2.3426895141601562, "step": 4207 }, { "epoch": 0.3317303902246748, "grad_norm": 2.0177435874938965, "learning_rate": 0.0005, "loss": 2.227255344390869, "step": 4208 }, { "epoch": 0.33180922349231373, "grad_norm": 1.0972685813903809, "learning_rate": 0.0005, "loss": 2.3852176666259766, "step": 4209 }, { "epoch": 0.3318880567599527, "grad_norm": 0.9881420135498047, "learning_rate": 0.0005, "loss": 2.1032304763793945, "step": 4210 }, { "epoch": 0.33196689002759167, "grad_norm": 1.2387404441833496, "learning_rate": 0.0005, "loss": 2.2374935150146484, "step": 4211 }, { "epoch": 0.3320457232952306, "grad_norm": 0.9473935961723328, "learning_rate": 0.0005, "loss": 2.151155948638916, "step": 4212 }, { "epoch": 0.33212455656286954, "grad_norm": 1.025619626045227, "learning_rate": 0.0005, "loss": 2.1856014728546143, "step": 4213 }, { "epoch": 0.33220338983050846, "grad_norm": 1.1854572296142578, "learning_rate": 0.0005, "loss": 2.753032684326172, "step": 4214 }, { "epoch": 0.3322822230981474, "grad_norm": 0.8462765216827393, "learning_rate": 0.0005, "loss": 2.239438533782959, "step": 4215 }, { "epoch": 0.3323610563657864, "grad_norm": 1.0097202062606812, "learning_rate": 0.0005, "loss": 2.4584197998046875, "step": 4216 }, { "epoch": 0.3324398896334253, "grad_norm": 0.8216413855552673, "learning_rate": 0.0005, "loss": 2.2021164894104004, "step": 4217 }, { "epoch": 0.33251872290106427, "grad_norm": 0.8016890287399292, "learning_rate": 0.0005, "loss": 2.0749073028564453, "step": 4218 }, { "epoch": 0.3325975561687032, "grad_norm": 0.7999099493026733, "learning_rate": 0.0005, "loss": 2.2850186824798584, "step": 4219 }, { "epoch": 0.33267638943634215, "grad_norm": 0.7161439061164856, "learning_rate": 0.0005, "loss": 2.5463266372680664, "step": 4220 }, { "epoch": 0.33275522270398106, "grad_norm": 1.7153418064117432, "learning_rate": 0.0005, "loss": 2.2268128395080566, "step": 4221 }, { "epoch": 0.33283405597162, "grad_norm": 0.8791934251785278, "learning_rate": 0.0005, "loss": 2.368467330932617, "step": 4222 }, { "epoch": 0.332912889239259, "grad_norm": 0.7045086026191711, "learning_rate": 0.0005, "loss": 2.3869218826293945, "step": 4223 }, { "epoch": 0.3329917225068979, "grad_norm": 1.3374342918395996, "learning_rate": 0.0005, "loss": 2.124661445617676, "step": 4224 }, { "epoch": 0.33307055577453687, "grad_norm": 0.771041989326477, "learning_rate": 0.0005, "loss": 2.6198511123657227, "step": 4225 }, { "epoch": 0.3331493890421758, "grad_norm": 0.9039906859397888, "learning_rate": 0.0005, "loss": 1.9462776184082031, "step": 4226 }, { "epoch": 0.33322822230981475, "grad_norm": 0.7678771018981934, "learning_rate": 0.0005, "loss": 1.9939192533493042, "step": 4227 }, { "epoch": 0.33330705557745366, "grad_norm": 1.2594468593597412, "learning_rate": 0.0005, "loss": 2.2921993732452393, "step": 4228 }, { "epoch": 0.3333858888450926, "grad_norm": 0.7508231401443481, "learning_rate": 0.0005, "loss": 1.836909294128418, "step": 4229 }, { "epoch": 0.3334647221127316, "grad_norm": 0.7520837187767029, "learning_rate": 0.0005, "loss": 2.0401291847229004, "step": 4230 }, { "epoch": 0.3335435553803705, "grad_norm": 0.674702525138855, "learning_rate": 0.0005, "loss": 1.9857003688812256, "step": 4231 }, { "epoch": 0.33362238864800947, "grad_norm": 0.6897671222686768, "learning_rate": 0.0005, "loss": 2.080019474029541, "step": 4232 }, { "epoch": 0.3337012219156484, "grad_norm": 1.127804160118103, "learning_rate": 0.0005, "loss": 2.129255533218384, "step": 4233 }, { "epoch": 0.33378005518328735, "grad_norm": 0.8660297989845276, "learning_rate": 0.0005, "loss": 2.466729164123535, "step": 4234 }, { "epoch": 0.3338588884509263, "grad_norm": 0.8160310387611389, "learning_rate": 0.0005, "loss": 2.3185102939605713, "step": 4235 }, { "epoch": 0.33393772171856523, "grad_norm": 0.7355316281318665, "learning_rate": 0.0005, "loss": 2.1753852367401123, "step": 4236 }, { "epoch": 0.3340165549862042, "grad_norm": 0.8464236259460449, "learning_rate": 0.0005, "loss": 2.368873119354248, "step": 4237 }, { "epoch": 0.3340953882538431, "grad_norm": 0.6407398581504822, "learning_rate": 0.0005, "loss": 2.1682372093200684, "step": 4238 }, { "epoch": 0.3341742215214821, "grad_norm": 0.9529276490211487, "learning_rate": 0.0005, "loss": 2.382084846496582, "step": 4239 }, { "epoch": 0.334253054789121, "grad_norm": 0.7999935746192932, "learning_rate": 0.0005, "loss": 2.4002678394317627, "step": 4240 }, { "epoch": 0.33433188805675995, "grad_norm": 0.8183249831199646, "learning_rate": 0.0005, "loss": 2.16549015045166, "step": 4241 }, { "epoch": 0.3344107213243989, "grad_norm": 0.8754058480262756, "learning_rate": 0.0005, "loss": 2.359651565551758, "step": 4242 }, { "epoch": 0.33448955459203783, "grad_norm": 0.9086723923683167, "learning_rate": 0.0005, "loss": 2.577193021774292, "step": 4243 }, { "epoch": 0.3345683878596768, "grad_norm": 0.8195057511329651, "learning_rate": 0.0005, "loss": 2.0293734073638916, "step": 4244 }, { "epoch": 0.3346472211273157, "grad_norm": 1.0761334896087646, "learning_rate": 0.0005, "loss": 2.519259452819824, "step": 4245 }, { "epoch": 0.3347260543949547, "grad_norm": 0.646371603012085, "learning_rate": 0.0005, "loss": 2.181689739227295, "step": 4246 }, { "epoch": 0.33480488766259364, "grad_norm": 1.0333184003829956, "learning_rate": 0.0005, "loss": 2.108093500137329, "step": 4247 }, { "epoch": 0.33488372093023255, "grad_norm": 0.8262304663658142, "learning_rate": 0.0005, "loss": 2.2571308612823486, "step": 4248 }, { "epoch": 0.3349625541978715, "grad_norm": 0.8942411541938782, "learning_rate": 0.0005, "loss": 2.156355857849121, "step": 4249 }, { "epoch": 0.33504138746551043, "grad_norm": 0.8652080297470093, "learning_rate": 0.0005, "loss": 2.029911756515503, "step": 4250 }, { "epoch": 0.3351202207331494, "grad_norm": 0.7335180044174194, "learning_rate": 0.0005, "loss": 2.1871285438537598, "step": 4251 }, { "epoch": 0.3351990540007883, "grad_norm": 1.0101510286331177, "learning_rate": 0.0005, "loss": 2.1472487449645996, "step": 4252 }, { "epoch": 0.3352778872684273, "grad_norm": 0.802278459072113, "learning_rate": 0.0005, "loss": 2.031506061553955, "step": 4253 }, { "epoch": 0.33535672053606624, "grad_norm": 0.8178927898406982, "learning_rate": 0.0005, "loss": 2.353245258331299, "step": 4254 }, { "epoch": 0.33543555380370516, "grad_norm": 0.8165950775146484, "learning_rate": 0.0005, "loss": 2.482572078704834, "step": 4255 }, { "epoch": 0.3355143870713441, "grad_norm": 0.7580184936523438, "learning_rate": 0.0005, "loss": 2.0331811904907227, "step": 4256 }, { "epoch": 0.33559322033898303, "grad_norm": 1.145548701286316, "learning_rate": 0.0005, "loss": 2.3657402992248535, "step": 4257 }, { "epoch": 0.335672053606622, "grad_norm": 0.7109836935997009, "learning_rate": 0.0005, "loss": 2.139439582824707, "step": 4258 }, { "epoch": 0.3357508868742609, "grad_norm": 0.9086513519287109, "learning_rate": 0.0005, "loss": 2.496185779571533, "step": 4259 }, { "epoch": 0.3358297201418999, "grad_norm": 0.7613256573677063, "learning_rate": 0.0005, "loss": 2.2598133087158203, "step": 4260 }, { "epoch": 0.33590855340953885, "grad_norm": 0.7420840263366699, "learning_rate": 0.0005, "loss": 2.035543441772461, "step": 4261 }, { "epoch": 0.33598738667717776, "grad_norm": 1.0496612787246704, "learning_rate": 0.0005, "loss": 2.3774056434631348, "step": 4262 }, { "epoch": 0.3360662199448167, "grad_norm": 0.7488998174667358, "learning_rate": 0.0005, "loss": 2.114964008331299, "step": 4263 }, { "epoch": 0.33614505321245564, "grad_norm": 0.8035982251167297, "learning_rate": 0.0005, "loss": 2.555039405822754, "step": 4264 }, { "epoch": 0.3362238864800946, "grad_norm": 1.0040218830108643, "learning_rate": 0.0005, "loss": 2.4205758571624756, "step": 4265 }, { "epoch": 0.33630271974773357, "grad_norm": 1.064121961593628, "learning_rate": 0.0005, "loss": 2.1383228302001953, "step": 4266 }, { "epoch": 0.3363815530153725, "grad_norm": 0.7915577292442322, "learning_rate": 0.0005, "loss": 2.22943115234375, "step": 4267 }, { "epoch": 0.33646038628301145, "grad_norm": 0.7451903223991394, "learning_rate": 0.0005, "loss": 2.4316775798797607, "step": 4268 }, { "epoch": 0.33653921955065036, "grad_norm": 0.7475740909576416, "learning_rate": 0.0005, "loss": 2.1171553134918213, "step": 4269 }, { "epoch": 0.3366180528182893, "grad_norm": 0.7658290266990662, "learning_rate": 0.0005, "loss": 2.447375774383545, "step": 4270 }, { "epoch": 0.33669688608592824, "grad_norm": 0.8331604599952698, "learning_rate": 0.0005, "loss": 2.629927635192871, "step": 4271 }, { "epoch": 0.3367757193535672, "grad_norm": 0.862824559211731, "learning_rate": 0.0005, "loss": 1.9279496669769287, "step": 4272 }, { "epoch": 0.33685455262120617, "grad_norm": 0.7007820010185242, "learning_rate": 0.0005, "loss": 2.299044132232666, "step": 4273 }, { "epoch": 0.3369333858888451, "grad_norm": 0.8713786005973816, "learning_rate": 0.0005, "loss": 2.171908140182495, "step": 4274 }, { "epoch": 0.33701221915648405, "grad_norm": 0.8643673658370972, "learning_rate": 0.0005, "loss": 1.8899117708206177, "step": 4275 }, { "epoch": 0.33709105242412296, "grad_norm": 0.6766954660415649, "learning_rate": 0.0005, "loss": 2.140613079071045, "step": 4276 }, { "epoch": 0.33716988569176193, "grad_norm": 0.7723100781440735, "learning_rate": 0.0005, "loss": 2.4120378494262695, "step": 4277 }, { "epoch": 0.33724871895940084, "grad_norm": 0.8714032769203186, "learning_rate": 0.0005, "loss": 2.22773814201355, "step": 4278 }, { "epoch": 0.3373275522270398, "grad_norm": 0.8739044666290283, "learning_rate": 0.0005, "loss": 2.483926773071289, "step": 4279 }, { "epoch": 0.3374063854946788, "grad_norm": 0.7668721675872803, "learning_rate": 0.0005, "loss": 2.0280652046203613, "step": 4280 }, { "epoch": 0.3374852187623177, "grad_norm": 1.3927931785583496, "learning_rate": 0.0005, "loss": 2.1503469944000244, "step": 4281 }, { "epoch": 0.33756405202995665, "grad_norm": 1.4540884494781494, "learning_rate": 0.0005, "loss": 2.2919585704803467, "step": 4282 }, { "epoch": 0.33764288529759556, "grad_norm": 0.8441020846366882, "learning_rate": 0.0005, "loss": 2.2822422981262207, "step": 4283 }, { "epoch": 0.33772171856523453, "grad_norm": 1.0647073984146118, "learning_rate": 0.0005, "loss": 2.2798454761505127, "step": 4284 }, { "epoch": 0.3378005518328735, "grad_norm": 0.8503660559654236, "learning_rate": 0.0005, "loss": 2.457955837249756, "step": 4285 }, { "epoch": 0.3378793851005124, "grad_norm": 0.795514702796936, "learning_rate": 0.0005, "loss": 2.313990592956543, "step": 4286 }, { "epoch": 0.3379582183681514, "grad_norm": 0.9200718402862549, "learning_rate": 0.0005, "loss": 2.2222187519073486, "step": 4287 }, { "epoch": 0.3380370516357903, "grad_norm": 1.0335725545883179, "learning_rate": 0.0005, "loss": 2.4312503337860107, "step": 4288 }, { "epoch": 0.33811588490342925, "grad_norm": 0.726789653301239, "learning_rate": 0.0005, "loss": 1.9994637966156006, "step": 4289 }, { "epoch": 0.33819471817106816, "grad_norm": 0.8705625534057617, "learning_rate": 0.0005, "loss": 2.3629300594329834, "step": 4290 }, { "epoch": 0.33827355143870713, "grad_norm": 0.8635729551315308, "learning_rate": 0.0005, "loss": 2.220221757888794, "step": 4291 }, { "epoch": 0.3383523847063461, "grad_norm": 0.8705141544342041, "learning_rate": 0.0005, "loss": 2.3560662269592285, "step": 4292 }, { "epoch": 0.338431217973985, "grad_norm": 0.8406426310539246, "learning_rate": 0.0005, "loss": 1.9886645078659058, "step": 4293 }, { "epoch": 0.338510051241624, "grad_norm": 0.8864599466323853, "learning_rate": 0.0005, "loss": 2.5980207920074463, "step": 4294 }, { "epoch": 0.3385888845092629, "grad_norm": 0.9173416495323181, "learning_rate": 0.0005, "loss": 2.44185733795166, "step": 4295 }, { "epoch": 0.33866771777690186, "grad_norm": 0.8606882691383362, "learning_rate": 0.0005, "loss": 2.3279781341552734, "step": 4296 }, { "epoch": 0.3387465510445408, "grad_norm": 0.9995918273925781, "learning_rate": 0.0005, "loss": 2.111734628677368, "step": 4297 }, { "epoch": 0.33882538431217973, "grad_norm": 0.7682708501815796, "learning_rate": 0.0005, "loss": 2.485635280609131, "step": 4298 }, { "epoch": 0.3389042175798187, "grad_norm": 0.9496160745620728, "learning_rate": 0.0005, "loss": 2.676211357116699, "step": 4299 }, { "epoch": 0.3389830508474576, "grad_norm": 0.69578617811203, "learning_rate": 0.0005, "loss": 2.20810604095459, "step": 4300 }, { "epoch": 0.3390618841150966, "grad_norm": 0.8768014907836914, "learning_rate": 0.0005, "loss": 2.3704071044921875, "step": 4301 }, { "epoch": 0.3391407173827355, "grad_norm": 0.8100217580795288, "learning_rate": 0.0005, "loss": 2.1759090423583984, "step": 4302 }, { "epoch": 0.33921955065037446, "grad_norm": 0.678010880947113, "learning_rate": 0.0005, "loss": 2.2590556144714355, "step": 4303 }, { "epoch": 0.3392983839180134, "grad_norm": 0.9710604548454285, "learning_rate": 0.0005, "loss": 2.301347017288208, "step": 4304 }, { "epoch": 0.33937721718565234, "grad_norm": 1.185620903968811, "learning_rate": 0.0005, "loss": 2.2495336532592773, "step": 4305 }, { "epoch": 0.3394560504532913, "grad_norm": 0.8340182900428772, "learning_rate": 0.0005, "loss": 2.1982603073120117, "step": 4306 }, { "epoch": 0.3395348837209302, "grad_norm": 0.7430475354194641, "learning_rate": 0.0005, "loss": 1.9146548509597778, "step": 4307 }, { "epoch": 0.3396137169885692, "grad_norm": 0.7194350957870483, "learning_rate": 0.0005, "loss": 2.267411231994629, "step": 4308 }, { "epoch": 0.3396925502562081, "grad_norm": 0.898097574710846, "learning_rate": 0.0005, "loss": 2.1814463138580322, "step": 4309 }, { "epoch": 0.33977138352384706, "grad_norm": 0.7814003825187683, "learning_rate": 0.0005, "loss": 2.3699545860290527, "step": 4310 }, { "epoch": 0.339850216791486, "grad_norm": 0.7827500104904175, "learning_rate": 0.0005, "loss": 2.1792373657226562, "step": 4311 }, { "epoch": 0.33992905005912494, "grad_norm": 0.6954042315483093, "learning_rate": 0.0005, "loss": 2.343357801437378, "step": 4312 }, { "epoch": 0.3400078833267639, "grad_norm": 0.7827900648117065, "learning_rate": 0.0005, "loss": 2.4207887649536133, "step": 4313 }, { "epoch": 0.3400867165944028, "grad_norm": 0.8244543671607971, "learning_rate": 0.0005, "loss": 1.9174684286117554, "step": 4314 }, { "epoch": 0.3401655498620418, "grad_norm": 0.9102217555046082, "learning_rate": 0.0005, "loss": 2.104386806488037, "step": 4315 }, { "epoch": 0.34024438312968075, "grad_norm": 0.8831731677055359, "learning_rate": 0.0005, "loss": 2.2891483306884766, "step": 4316 }, { "epoch": 0.34032321639731966, "grad_norm": 0.8582971096038818, "learning_rate": 0.0005, "loss": 2.21330189704895, "step": 4317 }, { "epoch": 0.34040204966495863, "grad_norm": 0.8782110214233398, "learning_rate": 0.0005, "loss": 2.54573130607605, "step": 4318 }, { "epoch": 0.34048088293259754, "grad_norm": 0.8228674530982971, "learning_rate": 0.0005, "loss": 2.0827605724334717, "step": 4319 }, { "epoch": 0.3405597162002365, "grad_norm": 0.8362197279930115, "learning_rate": 0.0005, "loss": 2.282166004180908, "step": 4320 }, { "epoch": 0.3406385494678754, "grad_norm": 0.8604477047920227, "learning_rate": 0.0005, "loss": 2.3380560874938965, "step": 4321 }, { "epoch": 0.3407173827355144, "grad_norm": 0.819715142250061, "learning_rate": 0.0005, "loss": 1.920466423034668, "step": 4322 }, { "epoch": 0.34079621600315335, "grad_norm": 0.8222600817680359, "learning_rate": 0.0005, "loss": 2.0087220668792725, "step": 4323 }, { "epoch": 0.34087504927079226, "grad_norm": 0.8887041807174683, "learning_rate": 0.0005, "loss": 2.583439350128174, "step": 4324 }, { "epoch": 0.34095388253843123, "grad_norm": 1.0518821477890015, "learning_rate": 0.0005, "loss": 2.703056812286377, "step": 4325 }, { "epoch": 0.34103271580607014, "grad_norm": 1.0307285785675049, "learning_rate": 0.0005, "loss": 2.1782593727111816, "step": 4326 }, { "epoch": 0.3411115490737091, "grad_norm": 0.8674636483192444, "learning_rate": 0.0005, "loss": 2.3803255558013916, "step": 4327 }, { "epoch": 0.3411903823413481, "grad_norm": 0.8006976842880249, "learning_rate": 0.0005, "loss": 2.300082206726074, "step": 4328 }, { "epoch": 0.341269215608987, "grad_norm": 0.816619336605072, "learning_rate": 0.0005, "loss": 2.3183507919311523, "step": 4329 }, { "epoch": 0.34134804887662595, "grad_norm": 0.9348520636558533, "learning_rate": 0.0005, "loss": 2.8131253719329834, "step": 4330 }, { "epoch": 0.34142688214426486, "grad_norm": 0.7935890555381775, "learning_rate": 0.0005, "loss": 2.486804246902466, "step": 4331 }, { "epoch": 0.34150571541190383, "grad_norm": 0.8448179960250854, "learning_rate": 0.0005, "loss": 2.420161247253418, "step": 4332 }, { "epoch": 0.34158454867954274, "grad_norm": 1.0032294988632202, "learning_rate": 0.0005, "loss": 2.406282663345337, "step": 4333 }, { "epoch": 0.3416633819471817, "grad_norm": 0.726253867149353, "learning_rate": 0.0005, "loss": 2.254633665084839, "step": 4334 }, { "epoch": 0.3417422152148207, "grad_norm": 1.0765024423599243, "learning_rate": 0.0005, "loss": 2.2242560386657715, "step": 4335 }, { "epoch": 0.3418210484824596, "grad_norm": 1.0761030912399292, "learning_rate": 0.0005, "loss": 2.3978304862976074, "step": 4336 }, { "epoch": 0.34189988175009856, "grad_norm": 0.8418175578117371, "learning_rate": 0.0005, "loss": 2.5095841884613037, "step": 4337 }, { "epoch": 0.34197871501773747, "grad_norm": 0.9842416048049927, "learning_rate": 0.0005, "loss": 2.3974390029907227, "step": 4338 }, { "epoch": 0.34205754828537643, "grad_norm": 0.778866171836853, "learning_rate": 0.0005, "loss": 2.1936163902282715, "step": 4339 }, { "epoch": 0.34213638155301535, "grad_norm": 1.0270754098892212, "learning_rate": 0.0005, "loss": 2.26849365234375, "step": 4340 }, { "epoch": 0.3422152148206543, "grad_norm": 0.8907473087310791, "learning_rate": 0.0005, "loss": 2.134068012237549, "step": 4341 }, { "epoch": 0.3422940480882933, "grad_norm": 0.8460687398910522, "learning_rate": 0.0005, "loss": 2.0346450805664062, "step": 4342 }, { "epoch": 0.3423728813559322, "grad_norm": 0.7139134407043457, "learning_rate": 0.0005, "loss": 2.286353349685669, "step": 4343 }, { "epoch": 0.34245171462357116, "grad_norm": 0.7632417678833008, "learning_rate": 0.0005, "loss": 1.931147813796997, "step": 4344 }, { "epoch": 0.34253054789121007, "grad_norm": 0.7691875696182251, "learning_rate": 0.0005, "loss": 2.639313220977783, "step": 4345 }, { "epoch": 0.34260938115884904, "grad_norm": 0.7675285935401917, "learning_rate": 0.0005, "loss": 2.2339048385620117, "step": 4346 }, { "epoch": 0.342688214426488, "grad_norm": 0.6872864961624146, "learning_rate": 0.0005, "loss": 2.2722558975219727, "step": 4347 }, { "epoch": 0.3427670476941269, "grad_norm": 0.760441243648529, "learning_rate": 0.0005, "loss": 2.5171475410461426, "step": 4348 }, { "epoch": 0.3428458809617659, "grad_norm": 0.801601767539978, "learning_rate": 0.0005, "loss": 2.226400852203369, "step": 4349 }, { "epoch": 0.3429247142294048, "grad_norm": 0.7514668107032776, "learning_rate": 0.0005, "loss": 2.233253002166748, "step": 4350 }, { "epoch": 0.34300354749704376, "grad_norm": 0.6208730936050415, "learning_rate": 0.0005, "loss": 1.7981209754943848, "step": 4351 }, { "epoch": 0.34308238076468267, "grad_norm": 0.8638427257537842, "learning_rate": 0.0005, "loss": 2.3611397743225098, "step": 4352 }, { "epoch": 0.34316121403232164, "grad_norm": 0.7074665427207947, "learning_rate": 0.0005, "loss": 1.979860782623291, "step": 4353 }, { "epoch": 0.3432400472999606, "grad_norm": 0.7905484437942505, "learning_rate": 0.0005, "loss": 2.3050482273101807, "step": 4354 }, { "epoch": 0.3433188805675995, "grad_norm": 0.7361218929290771, "learning_rate": 0.0005, "loss": 2.242072343826294, "step": 4355 }, { "epoch": 0.3433977138352385, "grad_norm": 0.8322104215621948, "learning_rate": 0.0005, "loss": 2.076167583465576, "step": 4356 }, { "epoch": 0.3434765471028774, "grad_norm": 0.8626072406768799, "learning_rate": 0.0005, "loss": 2.225795030593872, "step": 4357 }, { "epoch": 0.34355538037051636, "grad_norm": 0.914282500743866, "learning_rate": 0.0005, "loss": 2.6697580814361572, "step": 4358 }, { "epoch": 0.34363421363815533, "grad_norm": 0.6962729692459106, "learning_rate": 0.0005, "loss": 2.2912137508392334, "step": 4359 }, { "epoch": 0.34371304690579424, "grad_norm": 0.7592489123344421, "learning_rate": 0.0005, "loss": 2.1319756507873535, "step": 4360 }, { "epoch": 0.3437918801734332, "grad_norm": 0.8028894662857056, "learning_rate": 0.0005, "loss": 1.7858150005340576, "step": 4361 }, { "epoch": 0.3438707134410721, "grad_norm": 0.8726733326911926, "learning_rate": 0.0005, "loss": 1.9713943004608154, "step": 4362 }, { "epoch": 0.3439495467087111, "grad_norm": 0.9627441167831421, "learning_rate": 0.0005, "loss": 2.192011833190918, "step": 4363 }, { "epoch": 0.34402837997635, "grad_norm": 0.8753771781921387, "learning_rate": 0.0005, "loss": 2.419339895248413, "step": 4364 }, { "epoch": 0.34410721324398896, "grad_norm": 0.8804369568824768, "learning_rate": 0.0005, "loss": 1.9277560710906982, "step": 4365 }, { "epoch": 0.34418604651162793, "grad_norm": 0.8097967505455017, "learning_rate": 0.0005, "loss": 2.5229597091674805, "step": 4366 }, { "epoch": 0.34426487977926684, "grad_norm": 0.7250741720199585, "learning_rate": 0.0005, "loss": 2.362653970718384, "step": 4367 }, { "epoch": 0.3443437130469058, "grad_norm": 1.1533552408218384, "learning_rate": 0.0005, "loss": 2.4102821350097656, "step": 4368 }, { "epoch": 0.3444225463145447, "grad_norm": 0.9407368302345276, "learning_rate": 0.0005, "loss": 2.348289966583252, "step": 4369 }, { "epoch": 0.3445013795821837, "grad_norm": 1.0153849124908447, "learning_rate": 0.0005, "loss": 2.4516172409057617, "step": 4370 }, { "epoch": 0.3445802128498226, "grad_norm": 0.885084867477417, "learning_rate": 0.0005, "loss": 2.20261812210083, "step": 4371 }, { "epoch": 0.34465904611746156, "grad_norm": 0.8828091621398926, "learning_rate": 0.0005, "loss": 2.149228572845459, "step": 4372 }, { "epoch": 0.34473787938510053, "grad_norm": 0.8999118804931641, "learning_rate": 0.0005, "loss": 2.457728147506714, "step": 4373 }, { "epoch": 0.34481671265273944, "grad_norm": 1.180420994758606, "learning_rate": 0.0005, "loss": 2.1867661476135254, "step": 4374 }, { "epoch": 0.3448955459203784, "grad_norm": 0.8102383017539978, "learning_rate": 0.0005, "loss": 2.1904399394989014, "step": 4375 }, { "epoch": 0.3449743791880173, "grad_norm": 0.9120140671730042, "learning_rate": 0.0005, "loss": 2.3205440044403076, "step": 4376 }, { "epoch": 0.3450532124556563, "grad_norm": 0.7860575318336487, "learning_rate": 0.0005, "loss": 2.3424415588378906, "step": 4377 }, { "epoch": 0.34513204572329526, "grad_norm": 0.8676469922065735, "learning_rate": 0.0005, "loss": 2.537480592727661, "step": 4378 }, { "epoch": 0.34521087899093417, "grad_norm": 1.0181586742401123, "learning_rate": 0.0005, "loss": 2.533021926879883, "step": 4379 }, { "epoch": 0.34528971225857313, "grad_norm": 0.7881943583488464, "learning_rate": 0.0005, "loss": 2.4220476150512695, "step": 4380 }, { "epoch": 0.34536854552621205, "grad_norm": 1.0583685636520386, "learning_rate": 0.0005, "loss": 2.5215811729431152, "step": 4381 }, { "epoch": 0.345447378793851, "grad_norm": 0.9179670214653015, "learning_rate": 0.0005, "loss": 2.2942867279052734, "step": 4382 }, { "epoch": 0.3455262120614899, "grad_norm": 0.9093761444091797, "learning_rate": 0.0005, "loss": 2.293978691101074, "step": 4383 }, { "epoch": 0.3456050453291289, "grad_norm": 0.8799339532852173, "learning_rate": 0.0005, "loss": 2.1126770973205566, "step": 4384 }, { "epoch": 0.34568387859676786, "grad_norm": 0.76030033826828, "learning_rate": 0.0005, "loss": 1.9834774732589722, "step": 4385 }, { "epoch": 0.34576271186440677, "grad_norm": 0.6429919004440308, "learning_rate": 0.0005, "loss": 1.9873626232147217, "step": 4386 }, { "epoch": 0.34584154513204574, "grad_norm": 1.0083675384521484, "learning_rate": 0.0005, "loss": 2.230954647064209, "step": 4387 }, { "epoch": 0.34592037839968465, "grad_norm": 0.9365007877349854, "learning_rate": 0.0005, "loss": 2.2375564575195312, "step": 4388 }, { "epoch": 0.3459992116673236, "grad_norm": 0.7347977161407471, "learning_rate": 0.0005, "loss": 2.260199546813965, "step": 4389 }, { "epoch": 0.3460780449349626, "grad_norm": 0.901604413986206, "learning_rate": 0.0005, "loss": 2.0110321044921875, "step": 4390 }, { "epoch": 0.3461568782026015, "grad_norm": 0.9849783778190613, "learning_rate": 0.0005, "loss": 2.231703519821167, "step": 4391 }, { "epoch": 0.34623571147024046, "grad_norm": 0.8209254741668701, "learning_rate": 0.0005, "loss": 2.1680550575256348, "step": 4392 }, { "epoch": 0.34631454473787937, "grad_norm": 1.1314606666564941, "learning_rate": 0.0005, "loss": 2.016383409500122, "step": 4393 }, { "epoch": 0.34639337800551834, "grad_norm": 1.3179419040679932, "learning_rate": 0.0005, "loss": 2.5479376316070557, "step": 4394 }, { "epoch": 0.34647221127315725, "grad_norm": 0.7495944499969482, "learning_rate": 0.0005, "loss": 2.1934797763824463, "step": 4395 }, { "epoch": 0.3465510445407962, "grad_norm": 0.8811173439025879, "learning_rate": 0.0005, "loss": 2.3979978561401367, "step": 4396 }, { "epoch": 0.3466298778084352, "grad_norm": 0.8789743781089783, "learning_rate": 0.0005, "loss": 2.5215673446655273, "step": 4397 }, { "epoch": 0.3467087110760741, "grad_norm": 0.6944158673286438, "learning_rate": 0.0005, "loss": 2.2092270851135254, "step": 4398 }, { "epoch": 0.34678754434371306, "grad_norm": 0.7745226621627808, "learning_rate": 0.0005, "loss": 2.1725382804870605, "step": 4399 }, { "epoch": 0.346866377611352, "grad_norm": 0.7599490880966187, "learning_rate": 0.0005, "loss": 2.0680973529815674, "step": 4400 }, { "epoch": 0.34694521087899094, "grad_norm": 1.103739857673645, "learning_rate": 0.0005, "loss": 2.780034065246582, "step": 4401 }, { "epoch": 0.34702404414662985, "grad_norm": 0.852444052696228, "learning_rate": 0.0005, "loss": 2.822948694229126, "step": 4402 }, { "epoch": 0.3471028774142688, "grad_norm": 0.7445107102394104, "learning_rate": 0.0005, "loss": 2.2838587760925293, "step": 4403 }, { "epoch": 0.3471817106819078, "grad_norm": 0.8338779807090759, "learning_rate": 0.0005, "loss": 2.245349645614624, "step": 4404 }, { "epoch": 0.3472605439495467, "grad_norm": 0.9617253541946411, "learning_rate": 0.0005, "loss": 2.378164768218994, "step": 4405 }, { "epoch": 0.34733937721718566, "grad_norm": 0.8571138381958008, "learning_rate": 0.0005, "loss": 2.3386964797973633, "step": 4406 }, { "epoch": 0.3474182104848246, "grad_norm": 0.7707257270812988, "learning_rate": 0.0005, "loss": 1.9389281272888184, "step": 4407 }, { "epoch": 0.34749704375246354, "grad_norm": 0.8227323293685913, "learning_rate": 0.0005, "loss": 2.4622631072998047, "step": 4408 }, { "epoch": 0.3475758770201025, "grad_norm": 0.7856417894363403, "learning_rate": 0.0005, "loss": 2.1105799674987793, "step": 4409 }, { "epoch": 0.3476547102877414, "grad_norm": 2.356623888015747, "learning_rate": 0.0005, "loss": 2.1610913276672363, "step": 4410 }, { "epoch": 0.3477335435553804, "grad_norm": 0.7304190397262573, "learning_rate": 0.0005, "loss": 2.0771565437316895, "step": 4411 }, { "epoch": 0.3478123768230193, "grad_norm": 0.9755121469497681, "learning_rate": 0.0005, "loss": 2.211003303527832, "step": 4412 }, { "epoch": 0.34789121009065826, "grad_norm": 0.8368420004844666, "learning_rate": 0.0005, "loss": 2.060234785079956, "step": 4413 }, { "epoch": 0.3479700433582972, "grad_norm": 0.8896675705909729, "learning_rate": 0.0005, "loss": 2.1259853839874268, "step": 4414 }, { "epoch": 0.34804887662593614, "grad_norm": 0.6871895790100098, "learning_rate": 0.0005, "loss": 2.532595157623291, "step": 4415 }, { "epoch": 0.3481277098935751, "grad_norm": 0.880845308303833, "learning_rate": 0.0005, "loss": 2.121786594390869, "step": 4416 }, { "epoch": 0.348206543161214, "grad_norm": 0.69356369972229, "learning_rate": 0.0005, "loss": 2.0981318950653076, "step": 4417 }, { "epoch": 0.348285376428853, "grad_norm": 0.9631614089012146, "learning_rate": 0.0005, "loss": 2.1992650032043457, "step": 4418 }, { "epoch": 0.3483642096964919, "grad_norm": 0.8218686580657959, "learning_rate": 0.0005, "loss": 2.1640820503234863, "step": 4419 }, { "epoch": 0.34844304296413087, "grad_norm": 1.0028295516967773, "learning_rate": 0.0005, "loss": 2.3549437522888184, "step": 4420 }, { "epoch": 0.34852187623176983, "grad_norm": 1.4150564670562744, "learning_rate": 0.0005, "loss": 2.4841504096984863, "step": 4421 }, { "epoch": 0.34860070949940875, "grad_norm": 0.8334546089172363, "learning_rate": 0.0005, "loss": 2.2842392921447754, "step": 4422 }, { "epoch": 0.3486795427670477, "grad_norm": 0.7550429105758667, "learning_rate": 0.0005, "loss": 2.2108678817749023, "step": 4423 }, { "epoch": 0.3487583760346866, "grad_norm": 0.8257124423980713, "learning_rate": 0.0005, "loss": 2.2803573608398438, "step": 4424 }, { "epoch": 0.3488372093023256, "grad_norm": 0.8565625548362732, "learning_rate": 0.0005, "loss": 2.337381601333618, "step": 4425 }, { "epoch": 0.3489160425699645, "grad_norm": 0.7980043292045593, "learning_rate": 0.0005, "loss": 2.254148483276367, "step": 4426 }, { "epoch": 0.34899487583760347, "grad_norm": 0.9561946392059326, "learning_rate": 0.0005, "loss": 2.151254415512085, "step": 4427 }, { "epoch": 0.34907370910524244, "grad_norm": 0.7452016472816467, "learning_rate": 0.0005, "loss": 2.4644227027893066, "step": 4428 }, { "epoch": 0.34915254237288135, "grad_norm": 0.7380265593528748, "learning_rate": 0.0005, "loss": 2.097202777862549, "step": 4429 }, { "epoch": 0.3492313756405203, "grad_norm": 0.8990093469619751, "learning_rate": 0.0005, "loss": 2.333420991897583, "step": 4430 }, { "epoch": 0.3493102089081592, "grad_norm": 0.9194568395614624, "learning_rate": 0.0005, "loss": 2.1227216720581055, "step": 4431 }, { "epoch": 0.3493890421757982, "grad_norm": 1.3768787384033203, "learning_rate": 0.0005, "loss": 2.5531301498413086, "step": 4432 }, { "epoch": 0.3494678754434371, "grad_norm": 1.1862807273864746, "learning_rate": 0.0005, "loss": 2.4860172271728516, "step": 4433 }, { "epoch": 0.34954670871107607, "grad_norm": 1.0576765537261963, "learning_rate": 0.0005, "loss": 2.2472176551818848, "step": 4434 }, { "epoch": 0.34962554197871504, "grad_norm": 1.0346366167068481, "learning_rate": 0.0005, "loss": 2.556032657623291, "step": 4435 }, { "epoch": 0.34970437524635395, "grad_norm": 0.7875643372535706, "learning_rate": 0.0005, "loss": 1.6885366439819336, "step": 4436 }, { "epoch": 0.3497832085139929, "grad_norm": 0.7379118204116821, "learning_rate": 0.0005, "loss": 2.1097590923309326, "step": 4437 }, { "epoch": 0.3498620417816318, "grad_norm": 0.6680570840835571, "learning_rate": 0.0005, "loss": 2.2301583290100098, "step": 4438 }, { "epoch": 0.3499408750492708, "grad_norm": 0.9891995787620544, "learning_rate": 0.0005, "loss": 2.2056338787078857, "step": 4439 }, { "epoch": 0.35001970831690976, "grad_norm": 0.9371962547302246, "learning_rate": 0.0005, "loss": 2.1807875633239746, "step": 4440 }, { "epoch": 0.3500985415845487, "grad_norm": 1.0109070539474487, "learning_rate": 0.0005, "loss": 2.275975227355957, "step": 4441 }, { "epoch": 0.35017737485218764, "grad_norm": 0.693547248840332, "learning_rate": 0.0005, "loss": 1.994718074798584, "step": 4442 }, { "epoch": 0.35025620811982655, "grad_norm": 1.1818631887435913, "learning_rate": 0.0005, "loss": 2.268160104751587, "step": 4443 }, { "epoch": 0.3503350413874655, "grad_norm": 0.8730608224868774, "learning_rate": 0.0005, "loss": 2.4767308235168457, "step": 4444 }, { "epoch": 0.35041387465510443, "grad_norm": 0.80759197473526, "learning_rate": 0.0005, "loss": 2.729301691055298, "step": 4445 }, { "epoch": 0.3504927079227434, "grad_norm": 0.9332956075668335, "learning_rate": 0.0005, "loss": 2.2187020778656006, "step": 4446 }, { "epoch": 0.35057154119038236, "grad_norm": 1.0637791156768799, "learning_rate": 0.0005, "loss": 2.164371967315674, "step": 4447 }, { "epoch": 0.3506503744580213, "grad_norm": 0.811659038066864, "learning_rate": 0.0005, "loss": 2.414684534072876, "step": 4448 }, { "epoch": 0.35072920772566024, "grad_norm": 0.9177154302597046, "learning_rate": 0.0005, "loss": 2.235527992248535, "step": 4449 }, { "epoch": 0.35080804099329915, "grad_norm": 0.8518089652061462, "learning_rate": 0.0005, "loss": 2.3382155895233154, "step": 4450 }, { "epoch": 0.3508868742609381, "grad_norm": 0.8187662959098816, "learning_rate": 0.0005, "loss": 2.4735302925109863, "step": 4451 }, { "epoch": 0.3509657075285771, "grad_norm": 0.9073824882507324, "learning_rate": 0.0005, "loss": 2.3281116485595703, "step": 4452 }, { "epoch": 0.351044540796216, "grad_norm": 0.6890648007392883, "learning_rate": 0.0005, "loss": 1.978470802307129, "step": 4453 }, { "epoch": 0.35112337406385496, "grad_norm": 0.9522899985313416, "learning_rate": 0.0005, "loss": 1.9958910942077637, "step": 4454 }, { "epoch": 0.3512022073314939, "grad_norm": 0.7875922918319702, "learning_rate": 0.0005, "loss": 1.9852454662322998, "step": 4455 }, { "epoch": 0.35128104059913284, "grad_norm": 0.9341715574264526, "learning_rate": 0.0005, "loss": 1.996028184890747, "step": 4456 }, { "epoch": 0.35135987386677175, "grad_norm": 0.9913997054100037, "learning_rate": 0.0005, "loss": 2.6411633491516113, "step": 4457 }, { "epoch": 0.3514387071344107, "grad_norm": 0.720684826374054, "learning_rate": 0.0005, "loss": 2.2710301876068115, "step": 4458 }, { "epoch": 0.3515175404020497, "grad_norm": 0.927619993686676, "learning_rate": 0.0005, "loss": 2.623657703399658, "step": 4459 }, { "epoch": 0.3515963736696886, "grad_norm": 1.0119165182113647, "learning_rate": 0.0005, "loss": 2.2753772735595703, "step": 4460 }, { "epoch": 0.35167520693732757, "grad_norm": 1.0304092168807983, "learning_rate": 0.0005, "loss": 2.1353976726531982, "step": 4461 }, { "epoch": 0.3517540402049665, "grad_norm": 0.8142113089561462, "learning_rate": 0.0005, "loss": 2.263986587524414, "step": 4462 }, { "epoch": 0.35183287347260545, "grad_norm": 1.0852923393249512, "learning_rate": 0.0005, "loss": 1.9510786533355713, "step": 4463 }, { "epoch": 0.35191170674024436, "grad_norm": 1.2859944105148315, "learning_rate": 0.0005, "loss": 2.4092884063720703, "step": 4464 }, { "epoch": 0.3519905400078833, "grad_norm": 1.0190889835357666, "learning_rate": 0.0005, "loss": 2.142857313156128, "step": 4465 }, { "epoch": 0.3520693732755223, "grad_norm": 0.9543021321296692, "learning_rate": 0.0005, "loss": 2.2475225925445557, "step": 4466 }, { "epoch": 0.3521482065431612, "grad_norm": 0.9816914796829224, "learning_rate": 0.0005, "loss": 2.3032593727111816, "step": 4467 }, { "epoch": 0.35222703981080017, "grad_norm": 1.1063928604125977, "learning_rate": 0.0005, "loss": 2.017583131790161, "step": 4468 }, { "epoch": 0.3523058730784391, "grad_norm": 0.7856165766716003, "learning_rate": 0.0005, "loss": 2.1860976219177246, "step": 4469 }, { "epoch": 0.35238470634607805, "grad_norm": 0.8837853074073792, "learning_rate": 0.0005, "loss": 2.2979812622070312, "step": 4470 }, { "epoch": 0.352463539613717, "grad_norm": 0.8343512415885925, "learning_rate": 0.0005, "loss": 2.245270013809204, "step": 4471 }, { "epoch": 0.3525423728813559, "grad_norm": 0.8614963293075562, "learning_rate": 0.0005, "loss": 2.206364631652832, "step": 4472 }, { "epoch": 0.3526212061489949, "grad_norm": 0.8087760806083679, "learning_rate": 0.0005, "loss": 2.3370771408081055, "step": 4473 }, { "epoch": 0.3527000394166338, "grad_norm": 0.8533528447151184, "learning_rate": 0.0005, "loss": 2.3170251846313477, "step": 4474 }, { "epoch": 0.35277887268427277, "grad_norm": 1.0474519729614258, "learning_rate": 0.0005, "loss": 2.2261502742767334, "step": 4475 }, { "epoch": 0.3528577059519117, "grad_norm": 0.8000701665878296, "learning_rate": 0.0005, "loss": 2.164825916290283, "step": 4476 }, { "epoch": 0.35293653921955065, "grad_norm": 0.9783757925033569, "learning_rate": 0.0005, "loss": 2.2821457386016846, "step": 4477 }, { "epoch": 0.3530153724871896, "grad_norm": 0.9805205464363098, "learning_rate": 0.0005, "loss": 2.5951881408691406, "step": 4478 }, { "epoch": 0.3530942057548285, "grad_norm": 0.7593062520027161, "learning_rate": 0.0005, "loss": 2.4665346145629883, "step": 4479 }, { "epoch": 0.3531730390224675, "grad_norm": 1.0578430891036987, "learning_rate": 0.0005, "loss": 2.2601726055145264, "step": 4480 }, { "epoch": 0.3532518722901064, "grad_norm": 0.9873667359352112, "learning_rate": 0.0005, "loss": 2.319077253341675, "step": 4481 }, { "epoch": 0.3533307055577454, "grad_norm": 0.855517566204071, "learning_rate": 0.0005, "loss": 2.754642963409424, "step": 4482 }, { "epoch": 0.35340953882538434, "grad_norm": 0.8776902556419373, "learning_rate": 0.0005, "loss": 2.5797815322875977, "step": 4483 }, { "epoch": 0.35348837209302325, "grad_norm": 0.8058483600616455, "learning_rate": 0.0005, "loss": 2.2320096492767334, "step": 4484 }, { "epoch": 0.3535672053606622, "grad_norm": 0.9154753684997559, "learning_rate": 0.0005, "loss": 2.104046106338501, "step": 4485 }, { "epoch": 0.35364603862830113, "grad_norm": 0.8492897152900696, "learning_rate": 0.0005, "loss": 2.3064746856689453, "step": 4486 }, { "epoch": 0.3537248718959401, "grad_norm": 1.7547669410705566, "learning_rate": 0.0005, "loss": 2.3170952796936035, "step": 4487 }, { "epoch": 0.353803705163579, "grad_norm": 1.074547290802002, "learning_rate": 0.0005, "loss": 2.4268550872802734, "step": 4488 }, { "epoch": 0.353882538431218, "grad_norm": 0.8626855611801147, "learning_rate": 0.0005, "loss": 2.199821949005127, "step": 4489 }, { "epoch": 0.35396137169885694, "grad_norm": 0.7832263112068176, "learning_rate": 0.0005, "loss": 2.3778748512268066, "step": 4490 }, { "epoch": 0.35404020496649585, "grad_norm": 1.0443567037582397, "learning_rate": 0.0005, "loss": 2.2910401821136475, "step": 4491 }, { "epoch": 0.3541190382341348, "grad_norm": 0.9153346419334412, "learning_rate": 0.0005, "loss": 2.4521477222442627, "step": 4492 }, { "epoch": 0.35419787150177373, "grad_norm": 1.02101469039917, "learning_rate": 0.0005, "loss": 2.501162528991699, "step": 4493 }, { "epoch": 0.3542767047694127, "grad_norm": 0.90372234582901, "learning_rate": 0.0005, "loss": 2.006857395172119, "step": 4494 }, { "epoch": 0.3543555380370516, "grad_norm": 0.8475672006607056, "learning_rate": 0.0005, "loss": 2.2980527877807617, "step": 4495 }, { "epoch": 0.3544343713046906, "grad_norm": 0.9819989204406738, "learning_rate": 0.0005, "loss": 2.334990978240967, "step": 4496 }, { "epoch": 0.35451320457232954, "grad_norm": 1.6157147884368896, "learning_rate": 0.0005, "loss": 2.4986729621887207, "step": 4497 }, { "epoch": 0.35459203783996845, "grad_norm": 0.9415002465248108, "learning_rate": 0.0005, "loss": 2.0875678062438965, "step": 4498 }, { "epoch": 0.3546708711076074, "grad_norm": 1.07090163230896, "learning_rate": 0.0005, "loss": 2.6585097312927246, "step": 4499 }, { "epoch": 0.35474970437524633, "grad_norm": 0.8478979468345642, "learning_rate": 0.0005, "loss": 2.5321860313415527, "step": 4500 }, { "epoch": 0.3548285376428853, "grad_norm": 0.7855637073516846, "learning_rate": 0.0005, "loss": 2.2998054027557373, "step": 4501 }, { "epoch": 0.35490737091052427, "grad_norm": 0.993990421295166, "learning_rate": 0.0005, "loss": 2.1042096614837646, "step": 4502 }, { "epoch": 0.3549862041781632, "grad_norm": 0.8720332384109497, "learning_rate": 0.0005, "loss": 2.390352487564087, "step": 4503 }, { "epoch": 0.35506503744580215, "grad_norm": 0.7724331617355347, "learning_rate": 0.0005, "loss": 2.2012908458709717, "step": 4504 }, { "epoch": 0.35514387071344106, "grad_norm": 1.1828356981277466, "learning_rate": 0.0005, "loss": 2.6616575717926025, "step": 4505 }, { "epoch": 0.35522270398108, "grad_norm": 0.7514580488204956, "learning_rate": 0.0005, "loss": 2.04048228263855, "step": 4506 }, { "epoch": 0.35530153724871893, "grad_norm": 0.9661285877227783, "learning_rate": 0.0005, "loss": 2.706148147583008, "step": 4507 }, { "epoch": 0.3553803705163579, "grad_norm": 0.8064686059951782, "learning_rate": 0.0005, "loss": 2.367661237716675, "step": 4508 }, { "epoch": 0.35545920378399687, "grad_norm": 0.7406563758850098, "learning_rate": 0.0005, "loss": 2.026447296142578, "step": 4509 }, { "epoch": 0.3555380370516358, "grad_norm": 1.0311602354049683, "learning_rate": 0.0005, "loss": 2.377683639526367, "step": 4510 }, { "epoch": 0.35561687031927475, "grad_norm": 0.7358940243721008, "learning_rate": 0.0005, "loss": 2.043144702911377, "step": 4511 }, { "epoch": 0.35569570358691366, "grad_norm": 0.8176915049552917, "learning_rate": 0.0005, "loss": 2.1722159385681152, "step": 4512 }, { "epoch": 0.3557745368545526, "grad_norm": 0.8282971382141113, "learning_rate": 0.0005, "loss": 2.0336990356445312, "step": 4513 }, { "epoch": 0.3558533701221916, "grad_norm": 0.8190394639968872, "learning_rate": 0.0005, "loss": 2.2088382244110107, "step": 4514 }, { "epoch": 0.3559322033898305, "grad_norm": 1.4919782876968384, "learning_rate": 0.0005, "loss": 2.1053390502929688, "step": 4515 }, { "epoch": 0.35601103665746947, "grad_norm": 0.9358479380607605, "learning_rate": 0.0005, "loss": 2.1136817932128906, "step": 4516 }, { "epoch": 0.3560898699251084, "grad_norm": 1.0461974143981934, "learning_rate": 0.0005, "loss": 2.1206440925598145, "step": 4517 }, { "epoch": 0.35616870319274735, "grad_norm": 1.030521273612976, "learning_rate": 0.0005, "loss": 2.3212618827819824, "step": 4518 }, { "epoch": 0.35624753646038626, "grad_norm": 0.7624450922012329, "learning_rate": 0.0005, "loss": 2.1125597953796387, "step": 4519 }, { "epoch": 0.3563263697280252, "grad_norm": 0.7971547842025757, "learning_rate": 0.0005, "loss": 2.121537923812866, "step": 4520 }, { "epoch": 0.3564052029956642, "grad_norm": 0.8270801305770874, "learning_rate": 0.0005, "loss": 1.9926139116287231, "step": 4521 }, { "epoch": 0.3564840362633031, "grad_norm": 1.0108656883239746, "learning_rate": 0.0005, "loss": 2.39446759223938, "step": 4522 }, { "epoch": 0.3565628695309421, "grad_norm": 0.8639369010925293, "learning_rate": 0.0005, "loss": 2.5833933353424072, "step": 4523 }, { "epoch": 0.356641702798581, "grad_norm": 0.887739360332489, "learning_rate": 0.0005, "loss": 2.347743034362793, "step": 4524 }, { "epoch": 0.35672053606621995, "grad_norm": 0.9080387353897095, "learning_rate": 0.0005, "loss": 2.3475944995880127, "step": 4525 }, { "epoch": 0.35679936933385886, "grad_norm": 0.8206065893173218, "learning_rate": 0.0005, "loss": 2.5419225692749023, "step": 4526 }, { "epoch": 0.35687820260149783, "grad_norm": 0.9109181761741638, "learning_rate": 0.0005, "loss": 2.5731418132781982, "step": 4527 }, { "epoch": 0.3569570358691368, "grad_norm": 0.8594760298728943, "learning_rate": 0.0005, "loss": 2.501145839691162, "step": 4528 }, { "epoch": 0.3570358691367757, "grad_norm": 0.9531877040863037, "learning_rate": 0.0005, "loss": 2.1954052448272705, "step": 4529 }, { "epoch": 0.3571147024044147, "grad_norm": 1.1368600130081177, "learning_rate": 0.0005, "loss": 2.3311314582824707, "step": 4530 }, { "epoch": 0.3571935356720536, "grad_norm": 0.8391783237457275, "learning_rate": 0.0005, "loss": 2.288811206817627, "step": 4531 }, { "epoch": 0.35727236893969255, "grad_norm": 1.0781177282333374, "learning_rate": 0.0005, "loss": 2.257899284362793, "step": 4532 }, { "epoch": 0.3573512022073315, "grad_norm": 1.393004059791565, "learning_rate": 0.0005, "loss": 2.0608861446380615, "step": 4533 }, { "epoch": 0.35743003547497043, "grad_norm": 1.324872374534607, "learning_rate": 0.0005, "loss": 2.376493453979492, "step": 4534 }, { "epoch": 0.3575088687426094, "grad_norm": 0.8322539329528809, "learning_rate": 0.0005, "loss": 2.4845621585845947, "step": 4535 }, { "epoch": 0.3575877020102483, "grad_norm": 0.8041828274726868, "learning_rate": 0.0005, "loss": 2.1345198154449463, "step": 4536 }, { "epoch": 0.3576665352778873, "grad_norm": 1.0395424365997314, "learning_rate": 0.0005, "loss": 2.4896881580352783, "step": 4537 }, { "epoch": 0.3577453685455262, "grad_norm": 0.7395054697990417, "learning_rate": 0.0005, "loss": 1.9778382778167725, "step": 4538 }, { "epoch": 0.35782420181316515, "grad_norm": 0.9773373007774353, "learning_rate": 0.0005, "loss": 2.437594413757324, "step": 4539 }, { "epoch": 0.3579030350808041, "grad_norm": 1.0577311515808105, "learning_rate": 0.0005, "loss": 2.2646703720092773, "step": 4540 }, { "epoch": 0.35798186834844303, "grad_norm": 0.799174964427948, "learning_rate": 0.0005, "loss": 2.401104211807251, "step": 4541 }, { "epoch": 0.358060701616082, "grad_norm": 1.1559754610061646, "learning_rate": 0.0005, "loss": 2.2134246826171875, "step": 4542 }, { "epoch": 0.3581395348837209, "grad_norm": 0.7675278782844543, "learning_rate": 0.0005, "loss": 2.1625218391418457, "step": 4543 }, { "epoch": 0.3582183681513599, "grad_norm": 0.7476509809494019, "learning_rate": 0.0005, "loss": 1.8633155822753906, "step": 4544 }, { "epoch": 0.35829720141899885, "grad_norm": 0.7540692687034607, "learning_rate": 0.0005, "loss": 2.101935863494873, "step": 4545 }, { "epoch": 0.35837603468663776, "grad_norm": 0.8555936813354492, "learning_rate": 0.0005, "loss": 2.277491569519043, "step": 4546 }, { "epoch": 0.3584548679542767, "grad_norm": 0.6769229173660278, "learning_rate": 0.0005, "loss": 2.0477893352508545, "step": 4547 }, { "epoch": 0.35853370122191563, "grad_norm": 0.7915666103363037, "learning_rate": 0.0005, "loss": 2.5638251304626465, "step": 4548 }, { "epoch": 0.3586125344895546, "grad_norm": 0.7100366353988647, "learning_rate": 0.0005, "loss": 2.083676815032959, "step": 4549 }, { "epoch": 0.3586913677571935, "grad_norm": 0.6910125017166138, "learning_rate": 0.0005, "loss": 2.2006843090057373, "step": 4550 }, { "epoch": 0.3587702010248325, "grad_norm": 0.8418617844581604, "learning_rate": 0.0005, "loss": 2.1002306938171387, "step": 4551 }, { "epoch": 0.35884903429247145, "grad_norm": 2.9633755683898926, "learning_rate": 0.0005, "loss": 2.1465611457824707, "step": 4552 }, { "epoch": 0.35892786756011036, "grad_norm": 0.7902178764343262, "learning_rate": 0.0005, "loss": 2.4535317420959473, "step": 4553 }, { "epoch": 0.3590067008277493, "grad_norm": 0.7594088912010193, "learning_rate": 0.0005, "loss": 2.3654110431671143, "step": 4554 }, { "epoch": 0.35908553409538824, "grad_norm": 0.7999694347381592, "learning_rate": 0.0005, "loss": 2.0522818565368652, "step": 4555 }, { "epoch": 0.3591643673630272, "grad_norm": 1.4589112997055054, "learning_rate": 0.0005, "loss": 2.286320686340332, "step": 4556 }, { "epoch": 0.3592432006306661, "grad_norm": 0.7013243436813354, "learning_rate": 0.0005, "loss": 1.998037576675415, "step": 4557 }, { "epoch": 0.3593220338983051, "grad_norm": 0.9776773452758789, "learning_rate": 0.0005, "loss": 2.0318760871887207, "step": 4558 }, { "epoch": 0.35940086716594405, "grad_norm": 0.8609408140182495, "learning_rate": 0.0005, "loss": 2.225831985473633, "step": 4559 }, { "epoch": 0.35947970043358296, "grad_norm": 0.8832082748413086, "learning_rate": 0.0005, "loss": 2.166687250137329, "step": 4560 }, { "epoch": 0.3595585337012219, "grad_norm": 0.9460545778274536, "learning_rate": 0.0005, "loss": 2.2747802734375, "step": 4561 }, { "epoch": 0.35963736696886084, "grad_norm": 0.8075054287910461, "learning_rate": 0.0005, "loss": 2.632819175720215, "step": 4562 }, { "epoch": 0.3597162002364998, "grad_norm": 0.7242664694786072, "learning_rate": 0.0005, "loss": 2.349574089050293, "step": 4563 }, { "epoch": 0.3597950335041388, "grad_norm": 0.9032402634620667, "learning_rate": 0.0005, "loss": 2.4773082733154297, "step": 4564 }, { "epoch": 0.3598738667717777, "grad_norm": 0.7330467104911804, "learning_rate": 0.0005, "loss": 2.2115416526794434, "step": 4565 }, { "epoch": 0.35995270003941665, "grad_norm": 0.915713369846344, "learning_rate": 0.0005, "loss": 1.8595236539840698, "step": 4566 }, { "epoch": 0.36003153330705556, "grad_norm": 0.8222886323928833, "learning_rate": 0.0005, "loss": 2.205230474472046, "step": 4567 }, { "epoch": 0.36011036657469453, "grad_norm": 1.1058332920074463, "learning_rate": 0.0005, "loss": 2.3981707096099854, "step": 4568 }, { "epoch": 0.36018919984233344, "grad_norm": 1.0525158643722534, "learning_rate": 0.0005, "loss": 1.8120520114898682, "step": 4569 }, { "epoch": 0.3602680331099724, "grad_norm": 1.0205261707305908, "learning_rate": 0.0005, "loss": 2.1435766220092773, "step": 4570 }, { "epoch": 0.3603468663776114, "grad_norm": 0.8875319361686707, "learning_rate": 0.0005, "loss": 2.281332492828369, "step": 4571 }, { "epoch": 0.3604256996452503, "grad_norm": 0.8185949325561523, "learning_rate": 0.0005, "loss": 2.17388916015625, "step": 4572 }, { "epoch": 0.36050453291288925, "grad_norm": 0.808576226234436, "learning_rate": 0.0005, "loss": 2.4185116291046143, "step": 4573 }, { "epoch": 0.36058336618052816, "grad_norm": 1.0058178901672363, "learning_rate": 0.0005, "loss": 2.6049938201904297, "step": 4574 }, { "epoch": 0.36066219944816713, "grad_norm": 1.1193456649780273, "learning_rate": 0.0005, "loss": 2.3481950759887695, "step": 4575 }, { "epoch": 0.36074103271580604, "grad_norm": 0.7912054061889648, "learning_rate": 0.0005, "loss": 1.9611051082611084, "step": 4576 }, { "epoch": 0.360819865983445, "grad_norm": 1.0074671506881714, "learning_rate": 0.0005, "loss": 2.463383674621582, "step": 4577 }, { "epoch": 0.360898699251084, "grad_norm": 0.8070485591888428, "learning_rate": 0.0005, "loss": 2.4221835136413574, "step": 4578 }, { "epoch": 0.3609775325187229, "grad_norm": 0.7395592331886292, "learning_rate": 0.0005, "loss": 2.181565284729004, "step": 4579 }, { "epoch": 0.36105636578636185, "grad_norm": 1.40858793258667, "learning_rate": 0.0005, "loss": 2.068758964538574, "step": 4580 }, { "epoch": 0.36113519905400077, "grad_norm": 0.9426702857017517, "learning_rate": 0.0005, "loss": 2.3148038387298584, "step": 4581 }, { "epoch": 0.36121403232163973, "grad_norm": 0.8758218884468079, "learning_rate": 0.0005, "loss": 2.591313362121582, "step": 4582 }, { "epoch": 0.3612928655892787, "grad_norm": 0.9233219027519226, "learning_rate": 0.0005, "loss": 2.575059413909912, "step": 4583 }, { "epoch": 0.3613716988569176, "grad_norm": 0.7358710765838623, "learning_rate": 0.0005, "loss": 2.22709584236145, "step": 4584 }, { "epoch": 0.3614505321245566, "grad_norm": 0.7511298656463623, "learning_rate": 0.0005, "loss": 2.009988307952881, "step": 4585 }, { "epoch": 0.3615293653921955, "grad_norm": 1.4608808755874634, "learning_rate": 0.0005, "loss": 1.8874577283859253, "step": 4586 }, { "epoch": 0.36160819865983446, "grad_norm": 0.7081756591796875, "learning_rate": 0.0005, "loss": 2.044405221939087, "step": 4587 }, { "epoch": 0.36168703192747337, "grad_norm": 0.8059890866279602, "learning_rate": 0.0005, "loss": 2.114563465118408, "step": 4588 }, { "epoch": 0.36176586519511233, "grad_norm": 0.9844086170196533, "learning_rate": 0.0005, "loss": 2.3648414611816406, "step": 4589 }, { "epoch": 0.3618446984627513, "grad_norm": 0.7659456133842468, "learning_rate": 0.0005, "loss": 2.3401098251342773, "step": 4590 }, { "epoch": 0.3619235317303902, "grad_norm": 0.8833395838737488, "learning_rate": 0.0005, "loss": 2.3166699409484863, "step": 4591 }, { "epoch": 0.3620023649980292, "grad_norm": 0.841178834438324, "learning_rate": 0.0005, "loss": 1.9153937101364136, "step": 4592 }, { "epoch": 0.3620811982656681, "grad_norm": 0.9557263851165771, "learning_rate": 0.0005, "loss": 2.0429975986480713, "step": 4593 }, { "epoch": 0.36216003153330706, "grad_norm": 0.8328475952148438, "learning_rate": 0.0005, "loss": 1.6898820400238037, "step": 4594 }, { "epoch": 0.362238864800946, "grad_norm": 0.9284514784812927, "learning_rate": 0.0005, "loss": 2.3125452995300293, "step": 4595 }, { "epoch": 0.36231769806858494, "grad_norm": 0.8290981650352478, "learning_rate": 0.0005, "loss": 2.193995475769043, "step": 4596 }, { "epoch": 0.3623965313362239, "grad_norm": 0.7222711443901062, "learning_rate": 0.0005, "loss": 2.2725863456726074, "step": 4597 }, { "epoch": 0.3624753646038628, "grad_norm": 0.7451766729354858, "learning_rate": 0.0005, "loss": 2.2061562538146973, "step": 4598 }, { "epoch": 0.3625541978715018, "grad_norm": 0.8665896654129028, "learning_rate": 0.0005, "loss": 2.4324450492858887, "step": 4599 }, { "epoch": 0.3626330311391407, "grad_norm": 1.0575810670852661, "learning_rate": 0.0005, "loss": 2.0894527435302734, "step": 4600 }, { "epoch": 0.36271186440677966, "grad_norm": 0.692554235458374, "learning_rate": 0.0005, "loss": 2.3456153869628906, "step": 4601 }, { "epoch": 0.3627906976744186, "grad_norm": 0.791933536529541, "learning_rate": 0.0005, "loss": 2.2657341957092285, "step": 4602 }, { "epoch": 0.36286953094205754, "grad_norm": 1.1113497018814087, "learning_rate": 0.0005, "loss": 2.4543497562408447, "step": 4603 }, { "epoch": 0.3629483642096965, "grad_norm": 0.6635503172874451, "learning_rate": 0.0005, "loss": 2.118234634399414, "step": 4604 }, { "epoch": 0.3630271974773354, "grad_norm": 0.9770945310592651, "learning_rate": 0.0005, "loss": 2.3785109519958496, "step": 4605 }, { "epoch": 0.3631060307449744, "grad_norm": 1.0343654155731201, "learning_rate": 0.0005, "loss": 2.3019328117370605, "step": 4606 }, { "epoch": 0.3631848640126133, "grad_norm": 1.0323083400726318, "learning_rate": 0.0005, "loss": 2.2997071743011475, "step": 4607 }, { "epoch": 0.36326369728025226, "grad_norm": 0.9010106325149536, "learning_rate": 0.0005, "loss": 2.560062885284424, "step": 4608 }, { "epoch": 0.36334253054789123, "grad_norm": 0.8554013967514038, "learning_rate": 0.0005, "loss": 2.3035924434661865, "step": 4609 }, { "epoch": 0.36342136381553014, "grad_norm": 1.0370208024978638, "learning_rate": 0.0005, "loss": 2.3592405319213867, "step": 4610 }, { "epoch": 0.3635001970831691, "grad_norm": 0.6483105421066284, "learning_rate": 0.0005, "loss": 2.4124388694763184, "step": 4611 }, { "epoch": 0.363579030350808, "grad_norm": 0.7245023846626282, "learning_rate": 0.0005, "loss": 2.3209831714630127, "step": 4612 }, { "epoch": 0.363657863618447, "grad_norm": 0.7399799823760986, "learning_rate": 0.0005, "loss": 2.3501617908477783, "step": 4613 }, { "epoch": 0.36373669688608595, "grad_norm": 0.772767186164856, "learning_rate": 0.0005, "loss": 2.4109020233154297, "step": 4614 }, { "epoch": 0.36381553015372486, "grad_norm": 0.8099576234817505, "learning_rate": 0.0005, "loss": 2.4450411796569824, "step": 4615 }, { "epoch": 0.36389436342136383, "grad_norm": 0.9314303398132324, "learning_rate": 0.0005, "loss": 2.2499282360076904, "step": 4616 }, { "epoch": 0.36397319668900274, "grad_norm": 0.8125887513160706, "learning_rate": 0.0005, "loss": 2.2076969146728516, "step": 4617 }, { "epoch": 0.3640520299566417, "grad_norm": 0.7844932079315186, "learning_rate": 0.0005, "loss": 2.07940411567688, "step": 4618 }, { "epoch": 0.3641308632242806, "grad_norm": 0.925778865814209, "learning_rate": 0.0005, "loss": 2.1759889125823975, "step": 4619 }, { "epoch": 0.3642096964919196, "grad_norm": 0.65678870677948, "learning_rate": 0.0005, "loss": 2.2203493118286133, "step": 4620 }, { "epoch": 0.36428852975955855, "grad_norm": 1.1835423707962036, "learning_rate": 0.0005, "loss": 2.0743463039398193, "step": 4621 }, { "epoch": 0.36436736302719747, "grad_norm": 1.2403252124786377, "learning_rate": 0.0005, "loss": 2.0603950023651123, "step": 4622 }, { "epoch": 0.36444619629483643, "grad_norm": 1.0759313106536865, "learning_rate": 0.0005, "loss": 2.1240038871765137, "step": 4623 }, { "epoch": 0.36452502956247534, "grad_norm": 0.8325955867767334, "learning_rate": 0.0005, "loss": 2.109750270843506, "step": 4624 }, { "epoch": 0.3646038628301143, "grad_norm": 1.0311033725738525, "learning_rate": 0.0005, "loss": 2.0975799560546875, "step": 4625 }, { "epoch": 0.3646826960977533, "grad_norm": 0.8384554386138916, "learning_rate": 0.0005, "loss": 2.246969699859619, "step": 4626 }, { "epoch": 0.3647615293653922, "grad_norm": 0.9507786631584167, "learning_rate": 0.0005, "loss": 2.3365397453308105, "step": 4627 }, { "epoch": 0.36484036263303116, "grad_norm": 1.6616500616073608, "learning_rate": 0.0005, "loss": 2.5506882667541504, "step": 4628 }, { "epoch": 0.36491919590067007, "grad_norm": 0.6848528981208801, "learning_rate": 0.0005, "loss": 2.375394821166992, "step": 4629 }, { "epoch": 0.36499802916830903, "grad_norm": 0.8256076574325562, "learning_rate": 0.0005, "loss": 2.159397840499878, "step": 4630 }, { "epoch": 0.36507686243594795, "grad_norm": 0.6934673190116882, "learning_rate": 0.0005, "loss": 2.0392537117004395, "step": 4631 }, { "epoch": 0.3651556957035869, "grad_norm": 0.8220240473747253, "learning_rate": 0.0005, "loss": 2.374601364135742, "step": 4632 }, { "epoch": 0.3652345289712259, "grad_norm": 0.9832913875579834, "learning_rate": 0.0005, "loss": 2.476836681365967, "step": 4633 }, { "epoch": 0.3653133622388648, "grad_norm": 0.7298535108566284, "learning_rate": 0.0005, "loss": 1.6829895973205566, "step": 4634 }, { "epoch": 0.36539219550650376, "grad_norm": 1.2044397592544556, "learning_rate": 0.0005, "loss": 2.140085220336914, "step": 4635 }, { "epoch": 0.36547102877414267, "grad_norm": 0.9679599404335022, "learning_rate": 0.0005, "loss": 2.3241429328918457, "step": 4636 }, { "epoch": 0.36554986204178164, "grad_norm": 0.7866792678833008, "learning_rate": 0.0005, "loss": 2.165480136871338, "step": 4637 }, { "epoch": 0.36562869530942055, "grad_norm": 0.9278758764266968, "learning_rate": 0.0005, "loss": 2.115819215774536, "step": 4638 }, { "epoch": 0.3657075285770595, "grad_norm": 1.0566754341125488, "learning_rate": 0.0005, "loss": 2.317406177520752, "step": 4639 }, { "epoch": 0.3657863618446985, "grad_norm": 0.8974016308784485, "learning_rate": 0.0005, "loss": 2.164055824279785, "step": 4640 }, { "epoch": 0.3658651951123374, "grad_norm": 1.3423881530761719, "learning_rate": 0.0005, "loss": 2.216280460357666, "step": 4641 }, { "epoch": 0.36594402837997636, "grad_norm": 1.14641535282135, "learning_rate": 0.0005, "loss": 2.498936414718628, "step": 4642 }, { "epoch": 0.36602286164761527, "grad_norm": 1.3675581216812134, "learning_rate": 0.0005, "loss": 2.2729249000549316, "step": 4643 }, { "epoch": 0.36610169491525424, "grad_norm": 0.6671191453933716, "learning_rate": 0.0005, "loss": 2.18882417678833, "step": 4644 }, { "epoch": 0.3661805281828932, "grad_norm": 2.0310823917388916, "learning_rate": 0.0005, "loss": 2.458958625793457, "step": 4645 }, { "epoch": 0.3662593614505321, "grad_norm": 1.766332983970642, "learning_rate": 0.0005, "loss": 2.134366512298584, "step": 4646 }, { "epoch": 0.3663381947181711, "grad_norm": 1.385039210319519, "learning_rate": 0.0005, "loss": 1.8172218799591064, "step": 4647 }, { "epoch": 0.36641702798581, "grad_norm": 1.2239415645599365, "learning_rate": 0.0005, "loss": 2.426443099975586, "step": 4648 }, { "epoch": 0.36649586125344896, "grad_norm": 1.087660312652588, "learning_rate": 0.0005, "loss": 2.457744598388672, "step": 4649 }, { "epoch": 0.3665746945210879, "grad_norm": 0.8595335483551025, "learning_rate": 0.0005, "loss": 1.8587521314620972, "step": 4650 }, { "epoch": 0.36665352778872684, "grad_norm": 1.4146257638931274, "learning_rate": 0.0005, "loss": 2.194098949432373, "step": 4651 }, { "epoch": 0.3667323610563658, "grad_norm": 1.1040302515029907, "learning_rate": 0.0005, "loss": 2.130711555480957, "step": 4652 }, { "epoch": 0.3668111943240047, "grad_norm": 1.0904461145401, "learning_rate": 0.0005, "loss": 2.42291259765625, "step": 4653 }, { "epoch": 0.3668900275916437, "grad_norm": 1.3714656829833984, "learning_rate": 0.0005, "loss": 2.1727607250213623, "step": 4654 }, { "epoch": 0.3669688608592826, "grad_norm": 0.9474461674690247, "learning_rate": 0.0005, "loss": 2.242119312286377, "step": 4655 }, { "epoch": 0.36704769412692156, "grad_norm": 0.8334717750549316, "learning_rate": 0.0005, "loss": 1.7953486442565918, "step": 4656 }, { "epoch": 0.36712652739456053, "grad_norm": 0.8237254619598389, "learning_rate": 0.0005, "loss": 2.150737762451172, "step": 4657 }, { "epoch": 0.36720536066219944, "grad_norm": 0.8277129530906677, "learning_rate": 0.0005, "loss": 1.9484670162200928, "step": 4658 }, { "epoch": 0.3672841939298384, "grad_norm": 0.8326498866081238, "learning_rate": 0.0005, "loss": 2.058377265930176, "step": 4659 }, { "epoch": 0.3673630271974773, "grad_norm": 0.8207963705062866, "learning_rate": 0.0005, "loss": 2.4379007816314697, "step": 4660 }, { "epoch": 0.3674418604651163, "grad_norm": 1.09874427318573, "learning_rate": 0.0005, "loss": 2.3430960178375244, "step": 4661 }, { "epoch": 0.3675206937327552, "grad_norm": 0.9559383392333984, "learning_rate": 0.0005, "loss": 2.4007577896118164, "step": 4662 }, { "epoch": 0.36759952700039417, "grad_norm": 0.9097595810890198, "learning_rate": 0.0005, "loss": 2.6524384021759033, "step": 4663 }, { "epoch": 0.36767836026803313, "grad_norm": 0.8844671845436096, "learning_rate": 0.0005, "loss": 2.235903024673462, "step": 4664 }, { "epoch": 0.36775719353567204, "grad_norm": 0.5803987979888916, "learning_rate": 0.0005, "loss": 2.165787935256958, "step": 4665 }, { "epoch": 0.367836026803311, "grad_norm": 0.8164066672325134, "learning_rate": 0.0005, "loss": 2.329587936401367, "step": 4666 }, { "epoch": 0.3679148600709499, "grad_norm": 0.7980067729949951, "learning_rate": 0.0005, "loss": 1.9823956489562988, "step": 4667 }, { "epoch": 0.3679936933385889, "grad_norm": 0.7631111145019531, "learning_rate": 0.0005, "loss": 2.388720750808716, "step": 4668 }, { "epoch": 0.3680725266062278, "grad_norm": 1.0037593841552734, "learning_rate": 0.0005, "loss": 2.416099786758423, "step": 4669 }, { "epoch": 0.36815135987386677, "grad_norm": 0.8953880071640015, "learning_rate": 0.0005, "loss": 2.1853814125061035, "step": 4670 }, { "epoch": 0.36823019314150573, "grad_norm": 0.7904104590415955, "learning_rate": 0.0005, "loss": 2.1530470848083496, "step": 4671 }, { "epoch": 0.36830902640914465, "grad_norm": 0.7983020544052124, "learning_rate": 0.0005, "loss": 1.987244725227356, "step": 4672 }, { "epoch": 0.3683878596767836, "grad_norm": 0.7542317509651184, "learning_rate": 0.0005, "loss": 2.158087730407715, "step": 4673 }, { "epoch": 0.3684666929444225, "grad_norm": 0.7663019895553589, "learning_rate": 0.0005, "loss": 2.1457743644714355, "step": 4674 }, { "epoch": 0.3685455262120615, "grad_norm": 0.898314356803894, "learning_rate": 0.0005, "loss": 1.9960740804672241, "step": 4675 }, { "epoch": 0.36862435947970046, "grad_norm": 0.7565562129020691, "learning_rate": 0.0005, "loss": 1.9581223726272583, "step": 4676 }, { "epoch": 0.36870319274733937, "grad_norm": 0.7387588620185852, "learning_rate": 0.0005, "loss": 2.1804428100585938, "step": 4677 }, { "epoch": 0.36878202601497834, "grad_norm": 1.5146708488464355, "learning_rate": 0.0005, "loss": 2.2202396392822266, "step": 4678 }, { "epoch": 0.36886085928261725, "grad_norm": 1.1017237901687622, "learning_rate": 0.0005, "loss": 2.5198826789855957, "step": 4679 }, { "epoch": 0.3689396925502562, "grad_norm": 0.7405292391777039, "learning_rate": 0.0005, "loss": 2.0708985328674316, "step": 4680 }, { "epoch": 0.3690185258178951, "grad_norm": 1.312451720237732, "learning_rate": 0.0005, "loss": 2.3216795921325684, "step": 4681 }, { "epoch": 0.3690973590855341, "grad_norm": 0.9666385650634766, "learning_rate": 0.0005, "loss": 1.971503496170044, "step": 4682 }, { "epoch": 0.36917619235317306, "grad_norm": 0.978514552116394, "learning_rate": 0.0005, "loss": 2.434831142425537, "step": 4683 }, { "epoch": 0.36925502562081197, "grad_norm": 0.7895732522010803, "learning_rate": 0.0005, "loss": 2.2624688148498535, "step": 4684 }, { "epoch": 0.36933385888845094, "grad_norm": 1.01093590259552, "learning_rate": 0.0005, "loss": 2.1497116088867188, "step": 4685 }, { "epoch": 0.36941269215608985, "grad_norm": 0.9204006195068359, "learning_rate": 0.0005, "loss": 2.5185487270355225, "step": 4686 }, { "epoch": 0.3694915254237288, "grad_norm": 0.717889130115509, "learning_rate": 0.0005, "loss": 1.7660670280456543, "step": 4687 }, { "epoch": 0.3695703586913678, "grad_norm": 1.1557731628417969, "learning_rate": 0.0005, "loss": 2.5804238319396973, "step": 4688 }, { "epoch": 0.3696491919590067, "grad_norm": 0.9467589855194092, "learning_rate": 0.0005, "loss": 2.5381314754486084, "step": 4689 }, { "epoch": 0.36972802522664566, "grad_norm": 0.9878168702125549, "learning_rate": 0.0005, "loss": 2.345384120941162, "step": 4690 }, { "epoch": 0.3698068584942846, "grad_norm": 0.9899258613586426, "learning_rate": 0.0005, "loss": 2.476809501647949, "step": 4691 }, { "epoch": 0.36988569176192354, "grad_norm": 0.9033021330833435, "learning_rate": 0.0005, "loss": 2.11344575881958, "step": 4692 }, { "epoch": 0.36996452502956245, "grad_norm": 0.9656010866165161, "learning_rate": 0.0005, "loss": 2.1760191917419434, "step": 4693 }, { "epoch": 0.3700433582972014, "grad_norm": 0.8947020173072815, "learning_rate": 0.0005, "loss": 2.3776724338531494, "step": 4694 }, { "epoch": 0.3701221915648404, "grad_norm": 0.8412521481513977, "learning_rate": 0.0005, "loss": 2.043463706970215, "step": 4695 }, { "epoch": 0.3702010248324793, "grad_norm": 0.8179986476898193, "learning_rate": 0.0005, "loss": 2.1136960983276367, "step": 4696 }, { "epoch": 0.37027985810011826, "grad_norm": 0.861450731754303, "learning_rate": 0.0005, "loss": 1.8735039234161377, "step": 4697 }, { "epoch": 0.3703586913677572, "grad_norm": 0.963437557220459, "learning_rate": 0.0005, "loss": 2.366791248321533, "step": 4698 }, { "epoch": 0.37043752463539614, "grad_norm": 1.3286744356155396, "learning_rate": 0.0005, "loss": 2.1744637489318848, "step": 4699 }, { "epoch": 0.37051635790303505, "grad_norm": 0.8096798658370972, "learning_rate": 0.0005, "loss": 2.207315444946289, "step": 4700 }, { "epoch": 0.370595191170674, "grad_norm": 0.7510271072387695, "learning_rate": 0.0005, "loss": 2.018444776535034, "step": 4701 }, { "epoch": 0.370674024438313, "grad_norm": 0.7397879362106323, "learning_rate": 0.0005, "loss": 2.1447930335998535, "step": 4702 }, { "epoch": 0.3707528577059519, "grad_norm": 0.6771042346954346, "learning_rate": 0.0005, "loss": 2.0787384510040283, "step": 4703 }, { "epoch": 0.37083169097359087, "grad_norm": 0.7749902009963989, "learning_rate": 0.0005, "loss": 2.4033102989196777, "step": 4704 }, { "epoch": 0.3709105242412298, "grad_norm": 0.7323293685913086, "learning_rate": 0.0005, "loss": 2.027772903442383, "step": 4705 }, { "epoch": 0.37098935750886874, "grad_norm": 0.9543929100036621, "learning_rate": 0.0005, "loss": 1.9525707960128784, "step": 4706 }, { "epoch": 0.3710681907765077, "grad_norm": 0.9739181399345398, "learning_rate": 0.0005, "loss": 2.394925355911255, "step": 4707 }, { "epoch": 0.3711470240441466, "grad_norm": 0.7362783551216125, "learning_rate": 0.0005, "loss": 2.37556791305542, "step": 4708 }, { "epoch": 0.3712258573117856, "grad_norm": 0.794199526309967, "learning_rate": 0.0005, "loss": 2.4004011154174805, "step": 4709 }, { "epoch": 0.3713046905794245, "grad_norm": 1.0187374353408813, "learning_rate": 0.0005, "loss": 2.613938808441162, "step": 4710 }, { "epoch": 0.37138352384706347, "grad_norm": 0.819523811340332, "learning_rate": 0.0005, "loss": 2.3451595306396484, "step": 4711 }, { "epoch": 0.3714623571147024, "grad_norm": 0.6841616034507751, "learning_rate": 0.0005, "loss": 2.127887010574341, "step": 4712 }, { "epoch": 0.37154119038234135, "grad_norm": 0.8907955884933472, "learning_rate": 0.0005, "loss": 2.356776237487793, "step": 4713 }, { "epoch": 0.3716200236499803, "grad_norm": 0.736594557762146, "learning_rate": 0.0005, "loss": 2.2016026973724365, "step": 4714 }, { "epoch": 0.3716988569176192, "grad_norm": 0.8132147789001465, "learning_rate": 0.0005, "loss": 2.2374703884124756, "step": 4715 }, { "epoch": 0.3717776901852582, "grad_norm": 0.9086300134658813, "learning_rate": 0.0005, "loss": 2.2903785705566406, "step": 4716 }, { "epoch": 0.3718565234528971, "grad_norm": 0.8873487710952759, "learning_rate": 0.0005, "loss": 2.2610886096954346, "step": 4717 }, { "epoch": 0.37193535672053607, "grad_norm": 1.1091409921646118, "learning_rate": 0.0005, "loss": 2.1952147483825684, "step": 4718 }, { "epoch": 0.37201418998817504, "grad_norm": 0.7733641862869263, "learning_rate": 0.0005, "loss": 2.3530380725860596, "step": 4719 }, { "epoch": 0.37209302325581395, "grad_norm": 0.8812749981880188, "learning_rate": 0.0005, "loss": 2.4318861961364746, "step": 4720 }, { "epoch": 0.3721718565234529, "grad_norm": 0.7516034245491028, "learning_rate": 0.0005, "loss": 2.00496768951416, "step": 4721 }, { "epoch": 0.3722506897910918, "grad_norm": 1.262711763381958, "learning_rate": 0.0005, "loss": 2.0841100215911865, "step": 4722 }, { "epoch": 0.3723295230587308, "grad_norm": 0.7625051736831665, "learning_rate": 0.0005, "loss": 2.4545559883117676, "step": 4723 }, { "epoch": 0.3724083563263697, "grad_norm": 0.7659446597099304, "learning_rate": 0.0005, "loss": 2.101621627807617, "step": 4724 }, { "epoch": 0.37248718959400867, "grad_norm": 0.752388596534729, "learning_rate": 0.0005, "loss": 2.1849448680877686, "step": 4725 }, { "epoch": 0.37256602286164764, "grad_norm": 2.0297114849090576, "learning_rate": 0.0005, "loss": 2.162423610687256, "step": 4726 }, { "epoch": 0.37264485612928655, "grad_norm": 0.7801057696342468, "learning_rate": 0.0005, "loss": 2.3734357357025146, "step": 4727 }, { "epoch": 0.3727236893969255, "grad_norm": 0.8858425617218018, "learning_rate": 0.0005, "loss": 2.221357822418213, "step": 4728 }, { "epoch": 0.37280252266456443, "grad_norm": 0.8357223272323608, "learning_rate": 0.0005, "loss": 2.2022547721862793, "step": 4729 }, { "epoch": 0.3728813559322034, "grad_norm": 0.7151485085487366, "learning_rate": 0.0005, "loss": 2.087594985961914, "step": 4730 }, { "epoch": 0.3729601891998423, "grad_norm": 1.2814654111862183, "learning_rate": 0.0005, "loss": 2.0688540935516357, "step": 4731 }, { "epoch": 0.3730390224674813, "grad_norm": 0.8503655791282654, "learning_rate": 0.0005, "loss": 1.9298089742660522, "step": 4732 }, { "epoch": 0.37311785573512024, "grad_norm": 5.974271774291992, "learning_rate": 0.0005, "loss": 2.4777069091796875, "step": 4733 }, { "epoch": 0.37319668900275915, "grad_norm": 0.6912168264389038, "learning_rate": 0.0005, "loss": 2.0509910583496094, "step": 4734 }, { "epoch": 0.3732755222703981, "grad_norm": 1.4067432880401611, "learning_rate": 0.0005, "loss": 2.347114086151123, "step": 4735 }, { "epoch": 0.37335435553803703, "grad_norm": 1.0385751724243164, "learning_rate": 0.0005, "loss": 2.402735710144043, "step": 4736 }, { "epoch": 0.373433188805676, "grad_norm": 0.9553209543228149, "learning_rate": 0.0005, "loss": 1.9639208316802979, "step": 4737 }, { "epoch": 0.37351202207331496, "grad_norm": 0.932416558265686, "learning_rate": 0.0005, "loss": 2.2012481689453125, "step": 4738 }, { "epoch": 0.3735908553409539, "grad_norm": 0.8550033569335938, "learning_rate": 0.0005, "loss": 2.136700391769409, "step": 4739 }, { "epoch": 0.37366968860859284, "grad_norm": 0.8821911215782166, "learning_rate": 0.0005, "loss": 2.2716455459594727, "step": 4740 }, { "epoch": 0.37374852187623175, "grad_norm": 0.7679721713066101, "learning_rate": 0.0005, "loss": 2.308466672897339, "step": 4741 }, { "epoch": 0.3738273551438707, "grad_norm": 1.4301936626434326, "learning_rate": 0.0005, "loss": 2.4725797176361084, "step": 4742 }, { "epoch": 0.37390618841150963, "grad_norm": 1.2491940259933472, "learning_rate": 0.0005, "loss": 2.2534093856811523, "step": 4743 }, { "epoch": 0.3739850216791486, "grad_norm": 1.1281001567840576, "learning_rate": 0.0005, "loss": 2.1523284912109375, "step": 4744 }, { "epoch": 0.37406385494678757, "grad_norm": 0.8182841539382935, "learning_rate": 0.0005, "loss": 2.5178189277648926, "step": 4745 }, { "epoch": 0.3741426882144265, "grad_norm": 0.6735888719558716, "learning_rate": 0.0005, "loss": 2.103381395339966, "step": 4746 }, { "epoch": 0.37422152148206544, "grad_norm": 0.8493959903717041, "learning_rate": 0.0005, "loss": 2.2009053230285645, "step": 4747 }, { "epoch": 0.37430035474970436, "grad_norm": 1.0267926454544067, "learning_rate": 0.0005, "loss": 2.4303042888641357, "step": 4748 }, { "epoch": 0.3743791880173433, "grad_norm": 0.9026446342468262, "learning_rate": 0.0005, "loss": 2.150191068649292, "step": 4749 }, { "epoch": 0.3744580212849823, "grad_norm": 1.0181660652160645, "learning_rate": 0.0005, "loss": 2.254394769668579, "step": 4750 }, { "epoch": 0.3745368545526212, "grad_norm": 1.049787163734436, "learning_rate": 0.0005, "loss": 2.210092306137085, "step": 4751 }, { "epoch": 0.37461568782026017, "grad_norm": 0.9056318998336792, "learning_rate": 0.0005, "loss": 1.94831383228302, "step": 4752 }, { "epoch": 0.3746945210878991, "grad_norm": 0.7821081280708313, "learning_rate": 0.0005, "loss": 2.582414388656616, "step": 4753 }, { "epoch": 0.37477335435553805, "grad_norm": 0.8146700859069824, "learning_rate": 0.0005, "loss": 2.1355934143066406, "step": 4754 }, { "epoch": 0.37485218762317696, "grad_norm": 0.7426967620849609, "learning_rate": 0.0005, "loss": 2.546306610107422, "step": 4755 }, { "epoch": 0.3749310208908159, "grad_norm": 0.7487889528274536, "learning_rate": 0.0005, "loss": 2.202286958694458, "step": 4756 }, { "epoch": 0.3750098541584549, "grad_norm": 0.820871114730835, "learning_rate": 0.0005, "loss": 2.1504502296447754, "step": 4757 }, { "epoch": 0.3750886874260938, "grad_norm": 5.580800533294678, "learning_rate": 0.0005, "loss": 2.1548140048980713, "step": 4758 }, { "epoch": 0.37516752069373277, "grad_norm": 0.8876654505729675, "learning_rate": 0.0005, "loss": 2.566910743713379, "step": 4759 }, { "epoch": 0.3752463539613717, "grad_norm": 0.7053182125091553, "learning_rate": 0.0005, "loss": 2.1161046028137207, "step": 4760 }, { "epoch": 0.37532518722901065, "grad_norm": 0.5952621102333069, "learning_rate": 0.0005, "loss": 2.387404441833496, "step": 4761 }, { "epoch": 0.37540402049664956, "grad_norm": 1.0687527656555176, "learning_rate": 0.0005, "loss": 2.1730494499206543, "step": 4762 }, { "epoch": 0.3754828537642885, "grad_norm": 0.9268488883972168, "learning_rate": 0.0005, "loss": 2.243462085723877, "step": 4763 }, { "epoch": 0.3755616870319275, "grad_norm": 0.8987555503845215, "learning_rate": 0.0005, "loss": 2.328885555267334, "step": 4764 }, { "epoch": 0.3756405202995664, "grad_norm": 0.7203240394592285, "learning_rate": 0.0005, "loss": 2.2300195693969727, "step": 4765 }, { "epoch": 0.37571935356720537, "grad_norm": 0.6621392965316772, "learning_rate": 0.0005, "loss": 1.9429705142974854, "step": 4766 }, { "epoch": 0.3757981868348443, "grad_norm": 0.6414642333984375, "learning_rate": 0.0005, "loss": 2.1649975776672363, "step": 4767 }, { "epoch": 0.37587702010248325, "grad_norm": 1.200095772743225, "learning_rate": 0.0005, "loss": 2.4125771522521973, "step": 4768 }, { "epoch": 0.3759558533701222, "grad_norm": 0.8110928535461426, "learning_rate": 0.0005, "loss": 2.2100934982299805, "step": 4769 }, { "epoch": 0.37603468663776113, "grad_norm": 0.8456200361251831, "learning_rate": 0.0005, "loss": 2.1193642616271973, "step": 4770 }, { "epoch": 0.3761135199054001, "grad_norm": 0.8861733078956604, "learning_rate": 0.0005, "loss": 1.7284489870071411, "step": 4771 }, { "epoch": 0.376192353173039, "grad_norm": 0.7982100248336792, "learning_rate": 0.0005, "loss": 2.2269301414489746, "step": 4772 }, { "epoch": 0.376271186440678, "grad_norm": 0.8814553618431091, "learning_rate": 0.0005, "loss": 2.1995277404785156, "step": 4773 }, { "epoch": 0.3763500197083169, "grad_norm": 0.6866291165351868, "learning_rate": 0.0005, "loss": 2.3423595428466797, "step": 4774 }, { "epoch": 0.37642885297595585, "grad_norm": 0.7548454999923706, "learning_rate": 0.0005, "loss": 2.231943130493164, "step": 4775 }, { "epoch": 0.3765076862435948, "grad_norm": 0.7420712113380432, "learning_rate": 0.0005, "loss": 2.4744138717651367, "step": 4776 }, { "epoch": 0.37658651951123373, "grad_norm": 0.8002915978431702, "learning_rate": 0.0005, "loss": 2.347486972808838, "step": 4777 }, { "epoch": 0.3766653527788727, "grad_norm": 0.9302638173103333, "learning_rate": 0.0005, "loss": 2.611182928085327, "step": 4778 }, { "epoch": 0.3767441860465116, "grad_norm": 0.7711660861968994, "learning_rate": 0.0005, "loss": 2.317281484603882, "step": 4779 }, { "epoch": 0.3768230193141506, "grad_norm": 0.8761409521102905, "learning_rate": 0.0005, "loss": 1.9825942516326904, "step": 4780 }, { "epoch": 0.37690185258178954, "grad_norm": 1.0190032720565796, "learning_rate": 0.0005, "loss": 2.582918643951416, "step": 4781 }, { "epoch": 0.37698068584942845, "grad_norm": 0.7712819576263428, "learning_rate": 0.0005, "loss": 2.1174063682556152, "step": 4782 }, { "epoch": 0.3770595191170674, "grad_norm": 1.1617989540100098, "learning_rate": 0.0005, "loss": 2.362692356109619, "step": 4783 }, { "epoch": 0.37713835238470633, "grad_norm": 0.9911831617355347, "learning_rate": 0.0005, "loss": 2.4026103019714355, "step": 4784 }, { "epoch": 0.3772171856523453, "grad_norm": 0.8934951424598694, "learning_rate": 0.0005, "loss": 2.286616325378418, "step": 4785 }, { "epoch": 0.3772960189199842, "grad_norm": 0.7493199706077576, "learning_rate": 0.0005, "loss": 2.034943103790283, "step": 4786 }, { "epoch": 0.3773748521876232, "grad_norm": 0.7437785267829895, "learning_rate": 0.0005, "loss": 1.9743456840515137, "step": 4787 }, { "epoch": 0.37745368545526214, "grad_norm": 1.171339511871338, "learning_rate": 0.0005, "loss": 2.418240785598755, "step": 4788 }, { "epoch": 0.37753251872290106, "grad_norm": 1.26819908618927, "learning_rate": 0.0005, "loss": 2.2338907718658447, "step": 4789 }, { "epoch": 0.37761135199054, "grad_norm": 0.8488458395004272, "learning_rate": 0.0005, "loss": 2.4495253562927246, "step": 4790 }, { "epoch": 0.37769018525817893, "grad_norm": 0.7090034484863281, "learning_rate": 0.0005, "loss": 2.2415106296539307, "step": 4791 }, { "epoch": 0.3777690185258179, "grad_norm": 1.1050900220870972, "learning_rate": 0.0005, "loss": 2.274179697036743, "step": 4792 }, { "epoch": 0.3778478517934568, "grad_norm": 0.9082356095314026, "learning_rate": 0.0005, "loss": 2.120988368988037, "step": 4793 }, { "epoch": 0.3779266850610958, "grad_norm": 0.6137245893478394, "learning_rate": 0.0005, "loss": 2.291085720062256, "step": 4794 }, { "epoch": 0.37800551832873475, "grad_norm": 0.7616752982139587, "learning_rate": 0.0005, "loss": 1.852242350578308, "step": 4795 }, { "epoch": 0.37808435159637366, "grad_norm": 0.7578268647193909, "learning_rate": 0.0005, "loss": 2.058176040649414, "step": 4796 }, { "epoch": 0.3781631848640126, "grad_norm": 0.7737611532211304, "learning_rate": 0.0005, "loss": 2.145554542541504, "step": 4797 }, { "epoch": 0.37824201813165154, "grad_norm": 0.890489935874939, "learning_rate": 0.0005, "loss": 2.2248401641845703, "step": 4798 }, { "epoch": 0.3783208513992905, "grad_norm": 0.7462796568870544, "learning_rate": 0.0005, "loss": 2.0703189373016357, "step": 4799 }, { "epoch": 0.37839968466692947, "grad_norm": 0.7203043699264526, "learning_rate": 0.0005, "loss": 2.2437775135040283, "step": 4800 }, { "epoch": 0.3784785179345684, "grad_norm": 0.7019189596176147, "learning_rate": 0.0005, "loss": 2.141765594482422, "step": 4801 }, { "epoch": 0.37855735120220735, "grad_norm": 0.8606092929840088, "learning_rate": 0.0005, "loss": 2.38497257232666, "step": 4802 }, { "epoch": 0.37863618446984626, "grad_norm": 1.0948481559753418, "learning_rate": 0.0005, "loss": 2.233806610107422, "step": 4803 }, { "epoch": 0.3787150177374852, "grad_norm": 0.8665494918823242, "learning_rate": 0.0005, "loss": 1.8804199695587158, "step": 4804 }, { "epoch": 0.37879385100512414, "grad_norm": 1.1526354551315308, "learning_rate": 0.0005, "loss": 2.1673383712768555, "step": 4805 }, { "epoch": 0.3788726842727631, "grad_norm": 1.1307505369186401, "learning_rate": 0.0005, "loss": 2.3378376960754395, "step": 4806 }, { "epoch": 0.37895151754040207, "grad_norm": 0.8420827984809875, "learning_rate": 0.0005, "loss": 2.0424797534942627, "step": 4807 }, { "epoch": 0.379030350808041, "grad_norm": 0.8650415539741516, "learning_rate": 0.0005, "loss": 2.1657919883728027, "step": 4808 }, { "epoch": 0.37910918407567995, "grad_norm": 0.6779599785804749, "learning_rate": 0.0005, "loss": 2.10132098197937, "step": 4809 }, { "epoch": 0.37918801734331886, "grad_norm": 0.8969871401786804, "learning_rate": 0.0005, "loss": 2.257467746734619, "step": 4810 }, { "epoch": 0.37926685061095783, "grad_norm": 0.77347731590271, "learning_rate": 0.0005, "loss": 2.357908248901367, "step": 4811 }, { "epoch": 0.3793456838785968, "grad_norm": 0.95139080286026, "learning_rate": 0.0005, "loss": 2.135622262954712, "step": 4812 }, { "epoch": 0.3794245171462357, "grad_norm": 0.8129999041557312, "learning_rate": 0.0005, "loss": 2.2759666442871094, "step": 4813 }, { "epoch": 0.3795033504138747, "grad_norm": 0.9102080464363098, "learning_rate": 0.0005, "loss": 2.094815492630005, "step": 4814 }, { "epoch": 0.3795821836815136, "grad_norm": 0.842985987663269, "learning_rate": 0.0005, "loss": 2.3372693061828613, "step": 4815 }, { "epoch": 0.37966101694915255, "grad_norm": 0.8353831768035889, "learning_rate": 0.0005, "loss": 2.3298473358154297, "step": 4816 }, { "epoch": 0.37973985021679146, "grad_norm": 0.8430817723274231, "learning_rate": 0.0005, "loss": 2.3203325271606445, "step": 4817 }, { "epoch": 0.37981868348443043, "grad_norm": 0.8984256386756897, "learning_rate": 0.0005, "loss": 2.2228541374206543, "step": 4818 }, { "epoch": 0.3798975167520694, "grad_norm": 0.7747560143470764, "learning_rate": 0.0005, "loss": 2.2004075050354004, "step": 4819 }, { "epoch": 0.3799763500197083, "grad_norm": 0.7835522294044495, "learning_rate": 0.0005, "loss": 2.2636754512786865, "step": 4820 }, { "epoch": 0.3800551832873473, "grad_norm": 0.8325902223587036, "learning_rate": 0.0005, "loss": 2.2922921180725098, "step": 4821 }, { "epoch": 0.3801340165549862, "grad_norm": 0.6707316040992737, "learning_rate": 0.0005, "loss": 1.982075572013855, "step": 4822 }, { "epoch": 0.38021284982262515, "grad_norm": 0.8706952929496765, "learning_rate": 0.0005, "loss": 2.373988628387451, "step": 4823 }, { "epoch": 0.38029168309026407, "grad_norm": 0.7453814148902893, "learning_rate": 0.0005, "loss": 2.3900182247161865, "step": 4824 }, { "epoch": 0.38037051635790303, "grad_norm": 0.8149365782737732, "learning_rate": 0.0005, "loss": 2.4850165843963623, "step": 4825 }, { "epoch": 0.380449349625542, "grad_norm": 0.9177526235580444, "learning_rate": 0.0005, "loss": 2.270049571990967, "step": 4826 }, { "epoch": 0.3805281828931809, "grad_norm": 0.7353161573410034, "learning_rate": 0.0005, "loss": 2.1355857849121094, "step": 4827 }, { "epoch": 0.3806070161608199, "grad_norm": 0.8646506667137146, "learning_rate": 0.0005, "loss": 2.3104376792907715, "step": 4828 }, { "epoch": 0.3806858494284588, "grad_norm": 1.0139206647872925, "learning_rate": 0.0005, "loss": 2.5210633277893066, "step": 4829 }, { "epoch": 0.38076468269609776, "grad_norm": 0.8788626194000244, "learning_rate": 0.0005, "loss": 2.166245698928833, "step": 4830 }, { "epoch": 0.3808435159637367, "grad_norm": 0.8317779898643494, "learning_rate": 0.0005, "loss": 2.14078426361084, "step": 4831 }, { "epoch": 0.38092234923137563, "grad_norm": 1.1912599802017212, "learning_rate": 0.0005, "loss": 2.446951389312744, "step": 4832 }, { "epoch": 0.3810011824990146, "grad_norm": 0.8037182092666626, "learning_rate": 0.0005, "loss": 2.261664628982544, "step": 4833 }, { "epoch": 0.3810800157666535, "grad_norm": 0.7669972777366638, "learning_rate": 0.0005, "loss": 2.0945987701416016, "step": 4834 }, { "epoch": 0.3811588490342925, "grad_norm": 0.8641537427902222, "learning_rate": 0.0005, "loss": 2.335444688796997, "step": 4835 }, { "epoch": 0.3812376823019314, "grad_norm": 0.854434609413147, "learning_rate": 0.0005, "loss": 2.614811420440674, "step": 4836 }, { "epoch": 0.38131651556957036, "grad_norm": 0.6249685883522034, "learning_rate": 0.0005, "loss": 2.1249494552612305, "step": 4837 }, { "epoch": 0.3813953488372093, "grad_norm": 0.7936680316925049, "learning_rate": 0.0005, "loss": 2.2288546562194824, "step": 4838 }, { "epoch": 0.38147418210484824, "grad_norm": 0.9496831297874451, "learning_rate": 0.0005, "loss": 2.3130269050598145, "step": 4839 }, { "epoch": 0.3815530153724872, "grad_norm": 0.7957252264022827, "learning_rate": 0.0005, "loss": 2.3012428283691406, "step": 4840 }, { "epoch": 0.3816318486401261, "grad_norm": 0.9785687923431396, "learning_rate": 0.0005, "loss": 2.357941150665283, "step": 4841 }, { "epoch": 0.3817106819077651, "grad_norm": 0.7894613742828369, "learning_rate": 0.0005, "loss": 1.9554318189620972, "step": 4842 }, { "epoch": 0.381789515175404, "grad_norm": 0.9225161671638489, "learning_rate": 0.0005, "loss": 2.0558433532714844, "step": 4843 }, { "epoch": 0.38186834844304296, "grad_norm": 0.9908376932144165, "learning_rate": 0.0005, "loss": 2.317471504211426, "step": 4844 }, { "epoch": 0.3819471817106819, "grad_norm": 0.9069491624832153, "learning_rate": 0.0005, "loss": 2.0462417602539062, "step": 4845 }, { "epoch": 0.38202601497832084, "grad_norm": 0.8681067228317261, "learning_rate": 0.0005, "loss": 2.2987637519836426, "step": 4846 }, { "epoch": 0.3821048482459598, "grad_norm": 1.0000615119934082, "learning_rate": 0.0005, "loss": 2.246410608291626, "step": 4847 }, { "epoch": 0.3821836815135987, "grad_norm": 0.8119431138038635, "learning_rate": 0.0005, "loss": 2.073716640472412, "step": 4848 }, { "epoch": 0.3822625147812377, "grad_norm": 1.0161750316619873, "learning_rate": 0.0005, "loss": 2.2440686225891113, "step": 4849 }, { "epoch": 0.38234134804887665, "grad_norm": 0.6502408981323242, "learning_rate": 0.0005, "loss": 2.0806150436401367, "step": 4850 }, { "epoch": 0.38242018131651556, "grad_norm": 0.9462538361549377, "learning_rate": 0.0005, "loss": 2.307363271713257, "step": 4851 }, { "epoch": 0.38249901458415453, "grad_norm": 0.8408336639404297, "learning_rate": 0.0005, "loss": 2.198719024658203, "step": 4852 }, { "epoch": 0.38257784785179344, "grad_norm": 0.7997454404830933, "learning_rate": 0.0005, "loss": 2.3170111179351807, "step": 4853 }, { "epoch": 0.3826566811194324, "grad_norm": 0.7938427329063416, "learning_rate": 0.0005, "loss": 2.0523695945739746, "step": 4854 }, { "epoch": 0.3827355143870713, "grad_norm": 0.9574416279792786, "learning_rate": 0.0005, "loss": 2.1272482872009277, "step": 4855 }, { "epoch": 0.3828143476547103, "grad_norm": 0.9029686450958252, "learning_rate": 0.0005, "loss": 2.283914566040039, "step": 4856 }, { "epoch": 0.38289318092234925, "grad_norm": 1.0940440893173218, "learning_rate": 0.0005, "loss": 2.4491636753082275, "step": 4857 }, { "epoch": 0.38297201418998816, "grad_norm": 0.8922966122627258, "learning_rate": 0.0005, "loss": 2.35441255569458, "step": 4858 }, { "epoch": 0.38305084745762713, "grad_norm": 0.6993593573570251, "learning_rate": 0.0005, "loss": 1.8458698987960815, "step": 4859 }, { "epoch": 0.38312968072526604, "grad_norm": 0.9307324886322021, "learning_rate": 0.0005, "loss": 2.113483428955078, "step": 4860 }, { "epoch": 0.383208513992905, "grad_norm": 0.7822681665420532, "learning_rate": 0.0005, "loss": 1.844022274017334, "step": 4861 }, { "epoch": 0.383287347260544, "grad_norm": 1.1079392433166504, "learning_rate": 0.0005, "loss": 2.2342309951782227, "step": 4862 }, { "epoch": 0.3833661805281829, "grad_norm": 0.8983864784240723, "learning_rate": 0.0005, "loss": 2.0746665000915527, "step": 4863 }, { "epoch": 0.38344501379582185, "grad_norm": 0.6987885236740112, "learning_rate": 0.0005, "loss": 2.228440761566162, "step": 4864 }, { "epoch": 0.38352384706346077, "grad_norm": 0.8024245500564575, "learning_rate": 0.0005, "loss": 2.094449520111084, "step": 4865 }, { "epoch": 0.38360268033109973, "grad_norm": 0.9485614895820618, "learning_rate": 0.0005, "loss": 2.359077215194702, "step": 4866 }, { "epoch": 0.38368151359873864, "grad_norm": 0.6703504920005798, "learning_rate": 0.0005, "loss": 2.139610767364502, "step": 4867 }, { "epoch": 0.3837603468663776, "grad_norm": 0.7731342911720276, "learning_rate": 0.0005, "loss": 2.3082971572875977, "step": 4868 }, { "epoch": 0.3838391801340166, "grad_norm": 0.6868301033973694, "learning_rate": 0.0005, "loss": 2.58709716796875, "step": 4869 }, { "epoch": 0.3839180134016555, "grad_norm": 0.8907373547554016, "learning_rate": 0.0005, "loss": 2.5231828689575195, "step": 4870 }, { "epoch": 0.38399684666929446, "grad_norm": 1.045267105102539, "learning_rate": 0.0005, "loss": 2.3968420028686523, "step": 4871 }, { "epoch": 0.38407567993693337, "grad_norm": 0.8297731280326843, "learning_rate": 0.0005, "loss": 2.1256165504455566, "step": 4872 }, { "epoch": 0.38415451320457233, "grad_norm": 0.7400527596473694, "learning_rate": 0.0005, "loss": 2.1548657417297363, "step": 4873 }, { "epoch": 0.38423334647221125, "grad_norm": 0.8853329420089722, "learning_rate": 0.0005, "loss": 2.081627607345581, "step": 4874 }, { "epoch": 0.3843121797398502, "grad_norm": 0.8311012387275696, "learning_rate": 0.0005, "loss": 2.028759002685547, "step": 4875 }, { "epoch": 0.3843910130074892, "grad_norm": 0.6857603788375854, "learning_rate": 0.0005, "loss": 1.9067275524139404, "step": 4876 }, { "epoch": 0.3844698462751281, "grad_norm": 0.8444660305976868, "learning_rate": 0.0005, "loss": 2.0026516914367676, "step": 4877 }, { "epoch": 0.38454867954276706, "grad_norm": 0.980273425579071, "learning_rate": 0.0005, "loss": 2.273221969604492, "step": 4878 }, { "epoch": 0.38462751281040597, "grad_norm": 0.7967492341995239, "learning_rate": 0.0005, "loss": 2.368840456008911, "step": 4879 }, { "epoch": 0.38470634607804494, "grad_norm": 0.9761157035827637, "learning_rate": 0.0005, "loss": 2.250804901123047, "step": 4880 }, { "epoch": 0.3847851793456839, "grad_norm": 1.0976455211639404, "learning_rate": 0.0005, "loss": 2.0988636016845703, "step": 4881 }, { "epoch": 0.3848640126133228, "grad_norm": 0.9226869940757751, "learning_rate": 0.0005, "loss": 2.14196515083313, "step": 4882 }, { "epoch": 0.3849428458809618, "grad_norm": 1.1656547784805298, "learning_rate": 0.0005, "loss": 1.9526712894439697, "step": 4883 }, { "epoch": 0.3850216791486007, "grad_norm": 1.1580731868743896, "learning_rate": 0.0005, "loss": 2.4823694229125977, "step": 4884 }, { "epoch": 0.38510051241623966, "grad_norm": 0.8120155334472656, "learning_rate": 0.0005, "loss": 2.29148268699646, "step": 4885 }, { "epoch": 0.38517934568387857, "grad_norm": 0.6605699062347412, "learning_rate": 0.0005, "loss": 2.049201726913452, "step": 4886 }, { "epoch": 0.38525817895151754, "grad_norm": 0.8537171483039856, "learning_rate": 0.0005, "loss": 2.3358399868011475, "step": 4887 }, { "epoch": 0.3853370122191565, "grad_norm": 0.8308089375495911, "learning_rate": 0.0005, "loss": 2.0091357231140137, "step": 4888 }, { "epoch": 0.3854158454867954, "grad_norm": 1.067623257637024, "learning_rate": 0.0005, "loss": 2.1807851791381836, "step": 4889 }, { "epoch": 0.3854946787544344, "grad_norm": 0.8882347345352173, "learning_rate": 0.0005, "loss": 2.448413372039795, "step": 4890 }, { "epoch": 0.3855735120220733, "grad_norm": 0.8155529499053955, "learning_rate": 0.0005, "loss": 2.061058521270752, "step": 4891 }, { "epoch": 0.38565234528971226, "grad_norm": 0.7768350839614868, "learning_rate": 0.0005, "loss": 2.494481086730957, "step": 4892 }, { "epoch": 0.38573117855735123, "grad_norm": 0.9924904108047485, "learning_rate": 0.0005, "loss": 2.3917155265808105, "step": 4893 }, { "epoch": 0.38581001182499014, "grad_norm": 0.9932873845100403, "learning_rate": 0.0005, "loss": 2.422588348388672, "step": 4894 }, { "epoch": 0.3858888450926291, "grad_norm": 0.8028660416603088, "learning_rate": 0.0005, "loss": 2.2024600505828857, "step": 4895 }, { "epoch": 0.385967678360268, "grad_norm": 0.6862390041351318, "learning_rate": 0.0005, "loss": 2.2120208740234375, "step": 4896 }, { "epoch": 0.386046511627907, "grad_norm": 0.8116962909698486, "learning_rate": 0.0005, "loss": 1.9059332609176636, "step": 4897 }, { "epoch": 0.3861253448955459, "grad_norm": 0.9951727390289307, "learning_rate": 0.0005, "loss": 1.9559952020645142, "step": 4898 }, { "epoch": 0.38620417816318486, "grad_norm": 1.055473804473877, "learning_rate": 0.0005, "loss": 2.300382137298584, "step": 4899 }, { "epoch": 0.38628301143082383, "grad_norm": 2.095970392227173, "learning_rate": 0.0005, "loss": 2.255553960800171, "step": 4900 }, { "epoch": 0.38636184469846274, "grad_norm": 0.7481924891471863, "learning_rate": 0.0005, "loss": 2.0700948238372803, "step": 4901 }, { "epoch": 0.3864406779661017, "grad_norm": 0.8027042746543884, "learning_rate": 0.0005, "loss": 2.3182873725891113, "step": 4902 }, { "epoch": 0.3865195112337406, "grad_norm": 0.7716585993766785, "learning_rate": 0.0005, "loss": 2.2073111534118652, "step": 4903 }, { "epoch": 0.3865983445013796, "grad_norm": 1.026533603668213, "learning_rate": 0.0005, "loss": 2.0606517791748047, "step": 4904 }, { "epoch": 0.3866771777690185, "grad_norm": 0.804589033126831, "learning_rate": 0.0005, "loss": 2.135725259780884, "step": 4905 }, { "epoch": 0.38675601103665747, "grad_norm": 1.1305855512619019, "learning_rate": 0.0005, "loss": 2.3437769412994385, "step": 4906 }, { "epoch": 0.38683484430429643, "grad_norm": 1.0421342849731445, "learning_rate": 0.0005, "loss": 2.2137668132781982, "step": 4907 }, { "epoch": 0.38691367757193534, "grad_norm": 0.8130959272384644, "learning_rate": 0.0005, "loss": 2.1747262477874756, "step": 4908 }, { "epoch": 0.3869925108395743, "grad_norm": 0.9793137907981873, "learning_rate": 0.0005, "loss": 2.2798991203308105, "step": 4909 }, { "epoch": 0.3870713441072132, "grad_norm": 0.9304959177970886, "learning_rate": 0.0005, "loss": 2.614813804626465, "step": 4910 }, { "epoch": 0.3871501773748522, "grad_norm": 0.8443928956985474, "learning_rate": 0.0005, "loss": 2.199692487716675, "step": 4911 }, { "epoch": 0.38722901064249116, "grad_norm": 0.62420254945755, "learning_rate": 0.0005, "loss": 2.1364645957946777, "step": 4912 }, { "epoch": 0.38730784391013007, "grad_norm": 0.8300524353981018, "learning_rate": 0.0005, "loss": 2.614119529724121, "step": 4913 }, { "epoch": 0.38738667717776903, "grad_norm": 0.6455122232437134, "learning_rate": 0.0005, "loss": 1.9920867681503296, "step": 4914 }, { "epoch": 0.38746551044540795, "grad_norm": 0.8683711290359497, "learning_rate": 0.0005, "loss": 2.2591793537139893, "step": 4915 }, { "epoch": 0.3875443437130469, "grad_norm": 0.9198518395423889, "learning_rate": 0.0005, "loss": 2.0259673595428467, "step": 4916 }, { "epoch": 0.3876231769806858, "grad_norm": 1.065020203590393, "learning_rate": 0.0005, "loss": 2.2202093601226807, "step": 4917 }, { "epoch": 0.3877020102483248, "grad_norm": 0.7913913130760193, "learning_rate": 0.0005, "loss": 2.178628444671631, "step": 4918 }, { "epoch": 0.38778084351596376, "grad_norm": 0.8117746114730835, "learning_rate": 0.0005, "loss": 1.940975308418274, "step": 4919 }, { "epoch": 0.38785967678360267, "grad_norm": 0.9708519577980042, "learning_rate": 0.0005, "loss": 2.23781418800354, "step": 4920 }, { "epoch": 0.38793851005124164, "grad_norm": 0.8846243023872375, "learning_rate": 0.0005, "loss": 2.289862632751465, "step": 4921 }, { "epoch": 0.38801734331888055, "grad_norm": 0.9441101551055908, "learning_rate": 0.0005, "loss": 2.267289400100708, "step": 4922 }, { "epoch": 0.3880961765865195, "grad_norm": 0.9826088547706604, "learning_rate": 0.0005, "loss": 2.428537607192993, "step": 4923 }, { "epoch": 0.3881750098541585, "grad_norm": 1.0033622980117798, "learning_rate": 0.0005, "loss": 2.7049384117126465, "step": 4924 }, { "epoch": 0.3882538431217974, "grad_norm": 0.7488973140716553, "learning_rate": 0.0005, "loss": 1.97734534740448, "step": 4925 }, { "epoch": 0.38833267638943636, "grad_norm": 1.0512471199035645, "learning_rate": 0.0005, "loss": 2.556102752685547, "step": 4926 }, { "epoch": 0.38841150965707527, "grad_norm": 0.7703791856765747, "learning_rate": 0.0005, "loss": 2.143836736679077, "step": 4927 }, { "epoch": 0.38849034292471424, "grad_norm": 0.9567985534667969, "learning_rate": 0.0005, "loss": 2.4863882064819336, "step": 4928 }, { "epoch": 0.38856917619235315, "grad_norm": 0.8295222520828247, "learning_rate": 0.0005, "loss": 2.1505889892578125, "step": 4929 }, { "epoch": 0.3886480094599921, "grad_norm": 0.9583556652069092, "learning_rate": 0.0005, "loss": 2.0370354652404785, "step": 4930 }, { "epoch": 0.3887268427276311, "grad_norm": 0.8404065370559692, "learning_rate": 0.0005, "loss": 2.263948678970337, "step": 4931 }, { "epoch": 0.38880567599527, "grad_norm": 1.2718415260314941, "learning_rate": 0.0005, "loss": 2.2105872631073, "step": 4932 }, { "epoch": 0.38888450926290896, "grad_norm": 0.7935903668403625, "learning_rate": 0.0005, "loss": 1.9673601388931274, "step": 4933 }, { "epoch": 0.3889633425305479, "grad_norm": 0.6725834012031555, "learning_rate": 0.0005, "loss": 2.136803150177002, "step": 4934 }, { "epoch": 0.38904217579818684, "grad_norm": 0.9310417175292969, "learning_rate": 0.0005, "loss": 2.5983328819274902, "step": 4935 }, { "epoch": 0.38912100906582575, "grad_norm": 0.8719510436058044, "learning_rate": 0.0005, "loss": 2.2472147941589355, "step": 4936 }, { "epoch": 0.3891998423334647, "grad_norm": 0.7177674770355225, "learning_rate": 0.0005, "loss": 2.3730790615081787, "step": 4937 }, { "epoch": 0.3892786756011037, "grad_norm": 0.9870800375938416, "learning_rate": 0.0005, "loss": 2.261230707168579, "step": 4938 }, { "epoch": 0.3893575088687426, "grad_norm": 0.772304117679596, "learning_rate": 0.0005, "loss": 2.2720301151275635, "step": 4939 }, { "epoch": 0.38943634213638156, "grad_norm": 1.032426357269287, "learning_rate": 0.0005, "loss": 2.196582794189453, "step": 4940 }, { "epoch": 0.3895151754040205, "grad_norm": 0.761943519115448, "learning_rate": 0.0005, "loss": 2.279109001159668, "step": 4941 }, { "epoch": 0.38959400867165944, "grad_norm": 1.2369821071624756, "learning_rate": 0.0005, "loss": 2.7261557579040527, "step": 4942 }, { "epoch": 0.3896728419392984, "grad_norm": 1.244960904121399, "learning_rate": 0.0005, "loss": 2.5500152111053467, "step": 4943 }, { "epoch": 0.3897516752069373, "grad_norm": 0.7135141491889954, "learning_rate": 0.0005, "loss": 2.128413200378418, "step": 4944 }, { "epoch": 0.3898305084745763, "grad_norm": 0.782278835773468, "learning_rate": 0.0005, "loss": 2.234861373901367, "step": 4945 }, { "epoch": 0.3899093417422152, "grad_norm": 0.7192114591598511, "learning_rate": 0.0005, "loss": 2.236921787261963, "step": 4946 }, { "epoch": 0.38998817500985417, "grad_norm": 0.7788824439048767, "learning_rate": 0.0005, "loss": 2.294093132019043, "step": 4947 }, { "epoch": 0.3900670082774931, "grad_norm": 0.6960503458976746, "learning_rate": 0.0005, "loss": 2.2905778884887695, "step": 4948 }, { "epoch": 0.39014584154513204, "grad_norm": 0.8293249011039734, "learning_rate": 0.0005, "loss": 2.221508741378784, "step": 4949 }, { "epoch": 0.390224674812771, "grad_norm": 1.399992823600769, "learning_rate": 0.0005, "loss": 2.14675235748291, "step": 4950 }, { "epoch": 0.3903035080804099, "grad_norm": 0.7498987317085266, "learning_rate": 0.0005, "loss": 2.1259396076202393, "step": 4951 }, { "epoch": 0.3903823413480489, "grad_norm": 0.7894922494888306, "learning_rate": 0.0005, "loss": 2.259640693664551, "step": 4952 }, { "epoch": 0.3904611746156878, "grad_norm": 0.6916839480400085, "learning_rate": 0.0005, "loss": 2.038539409637451, "step": 4953 }, { "epoch": 0.39054000788332677, "grad_norm": 0.8960065245628357, "learning_rate": 0.0005, "loss": 2.1785848140716553, "step": 4954 }, { "epoch": 0.39061884115096573, "grad_norm": 0.8974778056144714, "learning_rate": 0.0005, "loss": 2.5279078483581543, "step": 4955 }, { "epoch": 0.39069767441860465, "grad_norm": 0.9898285865783691, "learning_rate": 0.0005, "loss": 2.2836360931396484, "step": 4956 }, { "epoch": 0.3907765076862436, "grad_norm": 6.471831798553467, "learning_rate": 0.0005, "loss": 2.185053825378418, "step": 4957 }, { "epoch": 0.3908553409538825, "grad_norm": 1.2877731323242188, "learning_rate": 0.0005, "loss": 2.5198919773101807, "step": 4958 }, { "epoch": 0.3909341742215215, "grad_norm": 1.127615213394165, "learning_rate": 0.0005, "loss": 2.3861961364746094, "step": 4959 }, { "epoch": 0.3910130074891604, "grad_norm": 0.6420276165008545, "learning_rate": 0.0005, "loss": 2.091341018676758, "step": 4960 }, { "epoch": 0.39109184075679937, "grad_norm": 0.8367733359336853, "learning_rate": 0.0005, "loss": 1.962547779083252, "step": 4961 }, { "epoch": 0.39117067402443834, "grad_norm": 0.9549483060836792, "learning_rate": 0.0005, "loss": 2.1579694747924805, "step": 4962 }, { "epoch": 0.39124950729207725, "grad_norm": 0.8872567415237427, "learning_rate": 0.0005, "loss": 2.3256750106811523, "step": 4963 }, { "epoch": 0.3913283405597162, "grad_norm": 0.594849169254303, "learning_rate": 0.0005, "loss": 1.9802165031433105, "step": 4964 }, { "epoch": 0.3914071738273551, "grad_norm": 1.1375023126602173, "learning_rate": 0.0005, "loss": 2.3147058486938477, "step": 4965 }, { "epoch": 0.3914860070949941, "grad_norm": 0.731351375579834, "learning_rate": 0.0005, "loss": 2.021279811859131, "step": 4966 }, { "epoch": 0.391564840362633, "grad_norm": 0.8662339448928833, "learning_rate": 0.0005, "loss": 2.366422176361084, "step": 4967 }, { "epoch": 0.39164367363027197, "grad_norm": 1.0267900228500366, "learning_rate": 0.0005, "loss": 2.4151077270507812, "step": 4968 }, { "epoch": 0.39172250689791094, "grad_norm": 0.816585123538971, "learning_rate": 0.0005, "loss": 2.368614435195923, "step": 4969 }, { "epoch": 0.39180134016554985, "grad_norm": 0.9841475486755371, "learning_rate": 0.0005, "loss": 2.4684271812438965, "step": 4970 }, { "epoch": 0.3918801734331888, "grad_norm": 1.1422169208526611, "learning_rate": 0.0005, "loss": 2.2650928497314453, "step": 4971 }, { "epoch": 0.39195900670082773, "grad_norm": 0.9214304089546204, "learning_rate": 0.0005, "loss": 2.009439468383789, "step": 4972 }, { "epoch": 0.3920378399684667, "grad_norm": 0.6864163875579834, "learning_rate": 0.0005, "loss": 1.9859789609909058, "step": 4973 }, { "epoch": 0.39211667323610566, "grad_norm": 0.7061984539031982, "learning_rate": 0.0005, "loss": 2.33563232421875, "step": 4974 }, { "epoch": 0.3921955065037446, "grad_norm": 0.8865677714347839, "learning_rate": 0.0005, "loss": 2.122534990310669, "step": 4975 }, { "epoch": 0.39227433977138354, "grad_norm": 0.798652172088623, "learning_rate": 0.0005, "loss": 2.347975254058838, "step": 4976 }, { "epoch": 0.39235317303902245, "grad_norm": 0.6772968769073486, "learning_rate": 0.0005, "loss": 2.031525135040283, "step": 4977 }, { "epoch": 0.3924320063066614, "grad_norm": 1.0529284477233887, "learning_rate": 0.0005, "loss": 2.418522834777832, "step": 4978 }, { "epoch": 0.39251083957430033, "grad_norm": 0.7841979265213013, "learning_rate": 0.0005, "loss": 2.1005382537841797, "step": 4979 }, { "epoch": 0.3925896728419393, "grad_norm": 0.7383213043212891, "learning_rate": 0.0005, "loss": 2.341060161590576, "step": 4980 }, { "epoch": 0.39266850610957826, "grad_norm": 0.6536068916320801, "learning_rate": 0.0005, "loss": 2.4313769340515137, "step": 4981 }, { "epoch": 0.3927473393772172, "grad_norm": 1.0241254568099976, "learning_rate": 0.0005, "loss": 2.1322412490844727, "step": 4982 }, { "epoch": 0.39282617264485614, "grad_norm": 0.8027800917625427, "learning_rate": 0.0005, "loss": 1.9574297666549683, "step": 4983 }, { "epoch": 0.39290500591249505, "grad_norm": 0.7639032006263733, "learning_rate": 0.0005, "loss": 2.0135486125946045, "step": 4984 }, { "epoch": 0.392983839180134, "grad_norm": 0.738789439201355, "learning_rate": 0.0005, "loss": 1.989742636680603, "step": 4985 }, { "epoch": 0.393062672447773, "grad_norm": 0.6886469125747681, "learning_rate": 0.0005, "loss": 2.2120683193206787, "step": 4986 }, { "epoch": 0.3931415057154119, "grad_norm": 0.768947184085846, "learning_rate": 0.0005, "loss": 2.0057015419006348, "step": 4987 }, { "epoch": 0.39322033898305087, "grad_norm": 0.7479013800621033, "learning_rate": 0.0005, "loss": 2.3207263946533203, "step": 4988 }, { "epoch": 0.3932991722506898, "grad_norm": 0.9191554188728333, "learning_rate": 0.0005, "loss": 2.492034673690796, "step": 4989 }, { "epoch": 0.39337800551832874, "grad_norm": 0.8706396222114563, "learning_rate": 0.0005, "loss": 2.4028186798095703, "step": 4990 }, { "epoch": 0.39345683878596766, "grad_norm": 0.9912456274032593, "learning_rate": 0.0005, "loss": 2.1155052185058594, "step": 4991 }, { "epoch": 0.3935356720536066, "grad_norm": 0.9879803657531738, "learning_rate": 0.0005, "loss": 2.305577039718628, "step": 4992 }, { "epoch": 0.3936145053212456, "grad_norm": 0.8164455890655518, "learning_rate": 0.0005, "loss": 2.1996426582336426, "step": 4993 }, { "epoch": 0.3936933385888845, "grad_norm": 0.8088383078575134, "learning_rate": 0.0005, "loss": 2.5807571411132812, "step": 4994 }, { "epoch": 0.39377217185652347, "grad_norm": 0.8806400299072266, "learning_rate": 0.0005, "loss": 2.460484504699707, "step": 4995 }, { "epoch": 0.3938510051241624, "grad_norm": 0.8226709961891174, "learning_rate": 0.0005, "loss": 2.4140093326568604, "step": 4996 }, { "epoch": 0.39392983839180135, "grad_norm": 0.89757239818573, "learning_rate": 0.0005, "loss": 1.971149206161499, "step": 4997 }, { "epoch": 0.39400867165944026, "grad_norm": 1.0626451969146729, "learning_rate": 0.0005, "loss": 1.8707950115203857, "step": 4998 }, { "epoch": 0.3940875049270792, "grad_norm": 4.316755771636963, "learning_rate": 0.0005, "loss": 2.745405912399292, "step": 4999 }, { "epoch": 0.3941663381947182, "grad_norm": 1.0369668006896973, "learning_rate": 0.0005, "loss": 2.3306870460510254, "step": 5000 }, { "epoch": 0.3942451714623571, "grad_norm": 0.8531708121299744, "learning_rate": 0.0005, "loss": 2.0526294708251953, "step": 5001 }, { "epoch": 0.39432400472999607, "grad_norm": 0.7186673879623413, "learning_rate": 0.0005, "loss": 1.9573326110839844, "step": 5002 }, { "epoch": 0.394402837997635, "grad_norm": 1.0117714405059814, "learning_rate": 0.0005, "loss": 2.084829092025757, "step": 5003 }, { "epoch": 0.39448167126527395, "grad_norm": 0.9852591753005981, "learning_rate": 0.0005, "loss": 2.041138172149658, "step": 5004 }, { "epoch": 0.3945605045329129, "grad_norm": 0.9836575388908386, "learning_rate": 0.0005, "loss": 2.1631720066070557, "step": 5005 }, { "epoch": 0.3946393378005518, "grad_norm": 0.8405408263206482, "learning_rate": 0.0005, "loss": 2.346255302429199, "step": 5006 }, { "epoch": 0.3947181710681908, "grad_norm": 0.6564434766769409, "learning_rate": 0.0005, "loss": 2.0904695987701416, "step": 5007 }, { "epoch": 0.3947970043358297, "grad_norm": 0.8476855158805847, "learning_rate": 0.0005, "loss": 2.4381892681121826, "step": 5008 }, { "epoch": 0.39487583760346867, "grad_norm": 0.7691852450370789, "learning_rate": 0.0005, "loss": 2.5732405185699463, "step": 5009 }, { "epoch": 0.3949546708711076, "grad_norm": 0.8816739320755005, "learning_rate": 0.0005, "loss": 2.522517681121826, "step": 5010 }, { "epoch": 0.39503350413874655, "grad_norm": 1.0281999111175537, "learning_rate": 0.0005, "loss": 2.2962288856506348, "step": 5011 }, { "epoch": 0.3951123374063855, "grad_norm": 0.7954680919647217, "learning_rate": 0.0005, "loss": 2.118481397628784, "step": 5012 }, { "epoch": 0.39519117067402443, "grad_norm": 0.7746585607528687, "learning_rate": 0.0005, "loss": 2.0391478538513184, "step": 5013 }, { "epoch": 0.3952700039416634, "grad_norm": 0.9094809889793396, "learning_rate": 0.0005, "loss": 1.8110754489898682, "step": 5014 }, { "epoch": 0.3953488372093023, "grad_norm": 0.6900218725204468, "learning_rate": 0.0005, "loss": 2.0949275493621826, "step": 5015 }, { "epoch": 0.3954276704769413, "grad_norm": 0.680334746837616, "learning_rate": 0.0005, "loss": 2.029620409011841, "step": 5016 }, { "epoch": 0.39550650374458024, "grad_norm": 0.7775824666023254, "learning_rate": 0.0005, "loss": 2.114332437515259, "step": 5017 }, { "epoch": 0.39558533701221915, "grad_norm": 0.8663257956504822, "learning_rate": 0.0005, "loss": 2.3849644660949707, "step": 5018 }, { "epoch": 0.3956641702798581, "grad_norm": 0.8239548206329346, "learning_rate": 0.0005, "loss": 1.9896427392959595, "step": 5019 }, { "epoch": 0.39574300354749703, "grad_norm": 0.9226107597351074, "learning_rate": 0.0005, "loss": 2.4934146404266357, "step": 5020 }, { "epoch": 0.395821836815136, "grad_norm": 0.8562825322151184, "learning_rate": 0.0005, "loss": 2.426358699798584, "step": 5021 }, { "epoch": 0.3959006700827749, "grad_norm": 0.6500740647315979, "learning_rate": 0.0005, "loss": 2.0860514640808105, "step": 5022 }, { "epoch": 0.3959795033504139, "grad_norm": 0.7313438653945923, "learning_rate": 0.0005, "loss": 2.223219394683838, "step": 5023 }, { "epoch": 0.39605833661805284, "grad_norm": 0.7730783820152283, "learning_rate": 0.0005, "loss": 1.9646780490875244, "step": 5024 }, { "epoch": 0.39613716988569175, "grad_norm": 0.7589247226715088, "learning_rate": 0.0005, "loss": 2.1794357299804688, "step": 5025 }, { "epoch": 0.3962160031533307, "grad_norm": 0.8535398840904236, "learning_rate": 0.0005, "loss": 2.1834607124328613, "step": 5026 }, { "epoch": 0.39629483642096963, "grad_norm": 0.7725093364715576, "learning_rate": 0.0005, "loss": 2.451044797897339, "step": 5027 }, { "epoch": 0.3963736696886086, "grad_norm": 0.6746149659156799, "learning_rate": 0.0005, "loss": 2.049121141433716, "step": 5028 }, { "epoch": 0.3964525029562475, "grad_norm": 0.7270202040672302, "learning_rate": 0.0005, "loss": 2.277052402496338, "step": 5029 }, { "epoch": 0.3965313362238865, "grad_norm": 0.7644716501235962, "learning_rate": 0.0005, "loss": 2.22666072845459, "step": 5030 }, { "epoch": 0.39661016949152544, "grad_norm": 2.145831346511841, "learning_rate": 0.0005, "loss": 2.1968979835510254, "step": 5031 }, { "epoch": 0.39668900275916436, "grad_norm": 0.912771463394165, "learning_rate": 0.0005, "loss": 2.117062568664551, "step": 5032 }, { "epoch": 0.3967678360268033, "grad_norm": 0.8024474382400513, "learning_rate": 0.0005, "loss": 2.2958014011383057, "step": 5033 }, { "epoch": 0.39684666929444223, "grad_norm": 0.8976792097091675, "learning_rate": 0.0005, "loss": 2.4046199321746826, "step": 5034 }, { "epoch": 0.3969255025620812, "grad_norm": 0.8567783832550049, "learning_rate": 0.0005, "loss": 2.340949296951294, "step": 5035 }, { "epoch": 0.39700433582972017, "grad_norm": 1.0533031225204468, "learning_rate": 0.0005, "loss": 2.142064094543457, "step": 5036 }, { "epoch": 0.3970831690973591, "grad_norm": 0.8166460394859314, "learning_rate": 0.0005, "loss": 2.1961779594421387, "step": 5037 }, { "epoch": 0.39716200236499805, "grad_norm": 0.9688296318054199, "learning_rate": 0.0005, "loss": 2.2054929733276367, "step": 5038 }, { "epoch": 0.39724083563263696, "grad_norm": 1.1173056364059448, "learning_rate": 0.0005, "loss": 2.3288822174072266, "step": 5039 }, { "epoch": 0.3973196689002759, "grad_norm": 0.8256306648254395, "learning_rate": 0.0005, "loss": 2.2103958129882812, "step": 5040 }, { "epoch": 0.39739850216791484, "grad_norm": 0.6457046270370483, "learning_rate": 0.0005, "loss": 2.2806501388549805, "step": 5041 }, { "epoch": 0.3974773354355538, "grad_norm": 0.7724252939224243, "learning_rate": 0.0005, "loss": 2.163483142852783, "step": 5042 }, { "epoch": 0.39755616870319277, "grad_norm": 0.7531334161758423, "learning_rate": 0.0005, "loss": 2.2705700397491455, "step": 5043 }, { "epoch": 0.3976350019708317, "grad_norm": 0.9911941289901733, "learning_rate": 0.0005, "loss": 2.2899208068847656, "step": 5044 }, { "epoch": 0.39771383523847065, "grad_norm": 0.9833418130874634, "learning_rate": 0.0005, "loss": 2.257018566131592, "step": 5045 }, { "epoch": 0.39779266850610956, "grad_norm": 0.9172472357749939, "learning_rate": 0.0005, "loss": 2.4868221282958984, "step": 5046 }, { "epoch": 0.3978715017737485, "grad_norm": 0.9030329585075378, "learning_rate": 0.0005, "loss": 2.065981388092041, "step": 5047 }, { "epoch": 0.3979503350413875, "grad_norm": 0.8837336301803589, "learning_rate": 0.0005, "loss": 2.059495210647583, "step": 5048 }, { "epoch": 0.3980291683090264, "grad_norm": 0.7393807768821716, "learning_rate": 0.0005, "loss": 2.0906753540039062, "step": 5049 }, { "epoch": 0.39810800157666537, "grad_norm": 0.8171359300613403, "learning_rate": 0.0005, "loss": 2.2886853218078613, "step": 5050 }, { "epoch": 0.3981868348443043, "grad_norm": 1.0548185110092163, "learning_rate": 0.0005, "loss": 2.0652732849121094, "step": 5051 }, { "epoch": 0.39826566811194325, "grad_norm": 0.8322126865386963, "learning_rate": 0.0005, "loss": 2.223629951477051, "step": 5052 }, { "epoch": 0.39834450137958216, "grad_norm": 0.6971533298492432, "learning_rate": 0.0005, "loss": 2.0771260261535645, "step": 5053 }, { "epoch": 0.39842333464722113, "grad_norm": 0.8085646033287048, "learning_rate": 0.0005, "loss": 2.2817723751068115, "step": 5054 }, { "epoch": 0.3985021679148601, "grad_norm": 0.8376708626747131, "learning_rate": 0.0005, "loss": 2.58607816696167, "step": 5055 }, { "epoch": 0.398581001182499, "grad_norm": 1.2184909582138062, "learning_rate": 0.0005, "loss": 2.542196750640869, "step": 5056 }, { "epoch": 0.398659834450138, "grad_norm": 1.0058984756469727, "learning_rate": 0.0005, "loss": 2.132077217102051, "step": 5057 }, { "epoch": 0.3987386677177769, "grad_norm": 0.8436896204948425, "learning_rate": 0.0005, "loss": 2.3506851196289062, "step": 5058 }, { "epoch": 0.39881750098541585, "grad_norm": 0.8378556370735168, "learning_rate": 0.0005, "loss": 2.1160929203033447, "step": 5059 }, { "epoch": 0.39889633425305476, "grad_norm": 0.8760175704956055, "learning_rate": 0.0005, "loss": 1.972981333732605, "step": 5060 }, { "epoch": 0.39897516752069373, "grad_norm": 0.7209435105323792, "learning_rate": 0.0005, "loss": 2.223736047744751, "step": 5061 }, { "epoch": 0.3990540007883327, "grad_norm": 0.8035697340965271, "learning_rate": 0.0005, "loss": 1.943629264831543, "step": 5062 }, { "epoch": 0.3991328340559716, "grad_norm": 0.8745662569999695, "learning_rate": 0.0005, "loss": 2.1624298095703125, "step": 5063 }, { "epoch": 0.3992116673236106, "grad_norm": 0.7606039047241211, "learning_rate": 0.0005, "loss": 2.2814502716064453, "step": 5064 }, { "epoch": 0.3992905005912495, "grad_norm": 0.8777943849563599, "learning_rate": 0.0005, "loss": 2.380842685699463, "step": 5065 }, { "epoch": 0.39936933385888845, "grad_norm": 0.9852864146232605, "learning_rate": 0.0005, "loss": 2.168935775756836, "step": 5066 }, { "epoch": 0.3994481671265274, "grad_norm": 0.9036074876785278, "learning_rate": 0.0005, "loss": 2.33097505569458, "step": 5067 }, { "epoch": 0.39952700039416633, "grad_norm": 0.9326032996177673, "learning_rate": 0.0005, "loss": 2.325040102005005, "step": 5068 }, { "epoch": 0.3996058336618053, "grad_norm": 0.6603683829307556, "learning_rate": 0.0005, "loss": 1.9366751909255981, "step": 5069 }, { "epoch": 0.3996846669294442, "grad_norm": 0.8317791819572449, "learning_rate": 0.0005, "loss": 2.140070915222168, "step": 5070 }, { "epoch": 0.3997635001970832, "grad_norm": 0.7425040006637573, "learning_rate": 0.0005, "loss": 2.3631463050842285, "step": 5071 }, { "epoch": 0.3998423334647221, "grad_norm": 0.7488522529602051, "learning_rate": 0.0005, "loss": 2.214785099029541, "step": 5072 }, { "epoch": 0.39992116673236106, "grad_norm": 0.759699285030365, "learning_rate": 0.0005, "loss": 2.19761061668396, "step": 5073 }, { "epoch": 0.4, "grad_norm": 1.3587555885314941, "learning_rate": 0.0005, "loss": 2.455761671066284, "step": 5074 }, { "epoch": 0.40007883326763893, "grad_norm": 0.7223493456840515, "learning_rate": 0.0005, "loss": 2.4860165119171143, "step": 5075 }, { "epoch": 0.4001576665352779, "grad_norm": 0.7294080257415771, "learning_rate": 0.0005, "loss": 2.2055439949035645, "step": 5076 }, { "epoch": 0.4002364998029168, "grad_norm": 0.8494081497192383, "learning_rate": 0.0005, "loss": 2.202160358428955, "step": 5077 }, { "epoch": 0.4003153330705558, "grad_norm": 0.8257840275764465, "learning_rate": 0.0005, "loss": 1.9188283681869507, "step": 5078 }, { "epoch": 0.40039416633819475, "grad_norm": 0.7476503849029541, "learning_rate": 0.0005, "loss": 2.3393421173095703, "step": 5079 }, { "epoch": 0.40047299960583366, "grad_norm": 0.7713143229484558, "learning_rate": 0.0005, "loss": 2.0856480598449707, "step": 5080 }, { "epoch": 0.4005518328734726, "grad_norm": 0.958388090133667, "learning_rate": 0.0005, "loss": 2.1799135208129883, "step": 5081 }, { "epoch": 0.40063066614111154, "grad_norm": 0.8434625267982483, "learning_rate": 0.0005, "loss": 2.2107644081115723, "step": 5082 }, { "epoch": 0.4007094994087505, "grad_norm": 0.7895044088363647, "learning_rate": 0.0005, "loss": 2.0961756706237793, "step": 5083 }, { "epoch": 0.4007883326763894, "grad_norm": 0.7806116342544556, "learning_rate": 0.0005, "loss": 2.1248867511749268, "step": 5084 }, { "epoch": 0.4008671659440284, "grad_norm": 0.7741579413414001, "learning_rate": 0.0005, "loss": 2.1798110008239746, "step": 5085 }, { "epoch": 0.40094599921166735, "grad_norm": 0.5705351233482361, "learning_rate": 0.0005, "loss": 2.0175206661224365, "step": 5086 }, { "epoch": 0.40102483247930626, "grad_norm": 0.8121646642684937, "learning_rate": 0.0005, "loss": 2.3834667205810547, "step": 5087 }, { "epoch": 0.4011036657469452, "grad_norm": 0.7307389378547668, "learning_rate": 0.0005, "loss": 2.2332632541656494, "step": 5088 }, { "epoch": 0.40118249901458414, "grad_norm": 0.6724919080734253, "learning_rate": 0.0005, "loss": 2.113332748413086, "step": 5089 }, { "epoch": 0.4012613322822231, "grad_norm": 0.6254923343658447, "learning_rate": 0.0005, "loss": 2.218653440475464, "step": 5090 }, { "epoch": 0.401340165549862, "grad_norm": 0.7493970990180969, "learning_rate": 0.0005, "loss": 2.0391087532043457, "step": 5091 }, { "epoch": 0.401418998817501, "grad_norm": 0.6904385089874268, "learning_rate": 0.0005, "loss": 2.131270408630371, "step": 5092 }, { "epoch": 0.40149783208513995, "grad_norm": 0.7719884514808655, "learning_rate": 0.0005, "loss": 2.349058151245117, "step": 5093 }, { "epoch": 0.40157666535277886, "grad_norm": 0.8029124736785889, "learning_rate": 0.0005, "loss": 2.089317798614502, "step": 5094 }, { "epoch": 0.40165549862041783, "grad_norm": 0.8656196594238281, "learning_rate": 0.0005, "loss": 2.1991162300109863, "step": 5095 }, { "epoch": 0.40173433188805674, "grad_norm": 0.7568539977073669, "learning_rate": 0.0005, "loss": 2.518378257751465, "step": 5096 }, { "epoch": 0.4018131651556957, "grad_norm": 0.8244537115097046, "learning_rate": 0.0005, "loss": 2.238715887069702, "step": 5097 }, { "epoch": 0.4018919984233347, "grad_norm": 0.793151319026947, "learning_rate": 0.0005, "loss": 2.2125394344329834, "step": 5098 }, { "epoch": 0.4019708316909736, "grad_norm": 0.6524155735969543, "learning_rate": 0.0005, "loss": 2.2656421661376953, "step": 5099 }, { "epoch": 0.40204966495861255, "grad_norm": 0.9964621067047119, "learning_rate": 0.0005, "loss": 2.1672372817993164, "step": 5100 }, { "epoch": 0.40212849822625146, "grad_norm": 0.8588401079177856, "learning_rate": 0.0005, "loss": 2.3388843536376953, "step": 5101 }, { "epoch": 0.40220733149389043, "grad_norm": 0.7395530939102173, "learning_rate": 0.0005, "loss": 2.3226380348205566, "step": 5102 }, { "epoch": 0.40228616476152934, "grad_norm": 0.9833006858825684, "learning_rate": 0.0005, "loss": 2.365893840789795, "step": 5103 }, { "epoch": 0.4023649980291683, "grad_norm": 0.8146090507507324, "learning_rate": 0.0005, "loss": 2.0288965702056885, "step": 5104 }, { "epoch": 0.4024438312968073, "grad_norm": 0.77292799949646, "learning_rate": 0.0005, "loss": 2.280550956726074, "step": 5105 }, { "epoch": 0.4025226645644462, "grad_norm": 1.004500150680542, "learning_rate": 0.0005, "loss": 2.3958821296691895, "step": 5106 }, { "epoch": 0.40260149783208515, "grad_norm": 0.9256649613380432, "learning_rate": 0.0005, "loss": 2.258114814758301, "step": 5107 }, { "epoch": 0.40268033109972406, "grad_norm": 0.7577342391014099, "learning_rate": 0.0005, "loss": 2.1391587257385254, "step": 5108 }, { "epoch": 0.40275916436736303, "grad_norm": 0.815213680267334, "learning_rate": 0.0005, "loss": 2.0977301597595215, "step": 5109 }, { "epoch": 0.40283799763500194, "grad_norm": 0.7838099598884583, "learning_rate": 0.0005, "loss": 2.155803918838501, "step": 5110 }, { "epoch": 0.4029168309026409, "grad_norm": 0.9006372690200806, "learning_rate": 0.0005, "loss": 2.154231071472168, "step": 5111 }, { "epoch": 0.4029956641702799, "grad_norm": 0.6864951252937317, "learning_rate": 0.0005, "loss": 1.970710277557373, "step": 5112 }, { "epoch": 0.4030744974379188, "grad_norm": 0.6408382654190063, "learning_rate": 0.0005, "loss": 2.02640700340271, "step": 5113 }, { "epoch": 0.40315333070555776, "grad_norm": 0.6860344409942627, "learning_rate": 0.0005, "loss": 2.084367275238037, "step": 5114 }, { "epoch": 0.40323216397319667, "grad_norm": 0.7571201920509338, "learning_rate": 0.0005, "loss": 2.0179898738861084, "step": 5115 }, { "epoch": 0.40331099724083563, "grad_norm": 0.7464425563812256, "learning_rate": 0.0005, "loss": 2.2148277759552, "step": 5116 }, { "epoch": 0.4033898305084746, "grad_norm": 0.5933861136436462, "learning_rate": 0.0005, "loss": 2.001016616821289, "step": 5117 }, { "epoch": 0.4034686637761135, "grad_norm": 0.7084435820579529, "learning_rate": 0.0005, "loss": 2.1921515464782715, "step": 5118 }, { "epoch": 0.4035474970437525, "grad_norm": 0.6961441040039062, "learning_rate": 0.0005, "loss": 2.085435390472412, "step": 5119 }, { "epoch": 0.4036263303113914, "grad_norm": 0.6338987350463867, "learning_rate": 0.0005, "loss": 2.3113999366760254, "step": 5120 }, { "epoch": 0.40370516357903036, "grad_norm": 0.8561065793037415, "learning_rate": 0.0005, "loss": 2.4097886085510254, "step": 5121 }, { "epoch": 0.40378399684666927, "grad_norm": 0.785887598991394, "learning_rate": 0.0005, "loss": 2.2709097862243652, "step": 5122 }, { "epoch": 0.40386283011430824, "grad_norm": 0.7095122933387756, "learning_rate": 0.0005, "loss": 2.164763927459717, "step": 5123 }, { "epoch": 0.4039416633819472, "grad_norm": 0.7936539053916931, "learning_rate": 0.0005, "loss": 2.481654167175293, "step": 5124 }, { "epoch": 0.4040204966495861, "grad_norm": 0.7711783051490784, "learning_rate": 0.0005, "loss": 2.045689582824707, "step": 5125 }, { "epoch": 0.4040993299172251, "grad_norm": 0.8214617967605591, "learning_rate": 0.0005, "loss": 1.8539527654647827, "step": 5126 }, { "epoch": 0.404178163184864, "grad_norm": 0.7555902004241943, "learning_rate": 0.0005, "loss": 2.0597705841064453, "step": 5127 }, { "epoch": 0.40425699645250296, "grad_norm": 0.811551570892334, "learning_rate": 0.0005, "loss": 2.287384271621704, "step": 5128 }, { "epoch": 0.4043358297201419, "grad_norm": 1.4131613969802856, "learning_rate": 0.0005, "loss": 2.4351487159729004, "step": 5129 }, { "epoch": 0.40441466298778084, "grad_norm": 0.6852754354476929, "learning_rate": 0.0005, "loss": 2.3597326278686523, "step": 5130 }, { "epoch": 0.4044934962554198, "grad_norm": 0.8008916974067688, "learning_rate": 0.0005, "loss": 2.328747272491455, "step": 5131 }, { "epoch": 0.4045723295230587, "grad_norm": 0.7464451193809509, "learning_rate": 0.0005, "loss": 2.281923770904541, "step": 5132 }, { "epoch": 0.4046511627906977, "grad_norm": 0.8182990550994873, "learning_rate": 0.0005, "loss": 2.4158096313476562, "step": 5133 }, { "epoch": 0.4047299960583366, "grad_norm": 0.7644513845443726, "learning_rate": 0.0005, "loss": 2.0172722339630127, "step": 5134 }, { "epoch": 0.40480882932597556, "grad_norm": 0.6939759254455566, "learning_rate": 0.0005, "loss": 2.403724431991577, "step": 5135 }, { "epoch": 0.40488766259361453, "grad_norm": 0.7815477848052979, "learning_rate": 0.0005, "loss": 2.394773006439209, "step": 5136 }, { "epoch": 0.40496649586125344, "grad_norm": 2.819884777069092, "learning_rate": 0.0005, "loss": 2.2323946952819824, "step": 5137 }, { "epoch": 0.4050453291288924, "grad_norm": 0.832139790058136, "learning_rate": 0.0005, "loss": 2.0003926753997803, "step": 5138 }, { "epoch": 0.4051241623965313, "grad_norm": 0.7998437881469727, "learning_rate": 0.0005, "loss": 2.0569324493408203, "step": 5139 }, { "epoch": 0.4052029956641703, "grad_norm": 0.902834415435791, "learning_rate": 0.0005, "loss": 2.0974435806274414, "step": 5140 }, { "epoch": 0.4052818289318092, "grad_norm": 0.7670913934707642, "learning_rate": 0.0005, "loss": 2.1889915466308594, "step": 5141 }, { "epoch": 0.40536066219944816, "grad_norm": 1.136899471282959, "learning_rate": 0.0005, "loss": 2.0635128021240234, "step": 5142 }, { "epoch": 0.40543949546708713, "grad_norm": 0.6485269665718079, "learning_rate": 0.0005, "loss": 1.9632757902145386, "step": 5143 }, { "epoch": 0.40551832873472604, "grad_norm": 0.6472072005271912, "learning_rate": 0.0005, "loss": 1.86977219581604, "step": 5144 }, { "epoch": 0.405597162002365, "grad_norm": 0.7849631905555725, "learning_rate": 0.0005, "loss": 2.2343716621398926, "step": 5145 }, { "epoch": 0.4056759952700039, "grad_norm": 0.9479857683181763, "learning_rate": 0.0005, "loss": 2.1932220458984375, "step": 5146 }, { "epoch": 0.4057548285376429, "grad_norm": 1.9430491924285889, "learning_rate": 0.0005, "loss": 2.052797794342041, "step": 5147 }, { "epoch": 0.40583366180528185, "grad_norm": 0.8691158890724182, "learning_rate": 0.0005, "loss": 2.4438729286193848, "step": 5148 }, { "epoch": 0.40591249507292076, "grad_norm": 0.7390284538269043, "learning_rate": 0.0005, "loss": 1.9055825471878052, "step": 5149 }, { "epoch": 0.40599132834055973, "grad_norm": 1.114168405532837, "learning_rate": 0.0005, "loss": 2.6856257915496826, "step": 5150 }, { "epoch": 0.40607016160819864, "grad_norm": 0.6896492838859558, "learning_rate": 0.0005, "loss": 2.0248966217041016, "step": 5151 }, { "epoch": 0.4061489948758376, "grad_norm": 0.8479079604148865, "learning_rate": 0.0005, "loss": 2.0325958728790283, "step": 5152 }, { "epoch": 0.4062278281434765, "grad_norm": 0.7685619592666626, "learning_rate": 0.0005, "loss": 2.1419057846069336, "step": 5153 }, { "epoch": 0.4063066614111155, "grad_norm": 0.7111914753913879, "learning_rate": 0.0005, "loss": 2.086958408355713, "step": 5154 }, { "epoch": 0.40638549467875446, "grad_norm": 0.7533344030380249, "learning_rate": 0.0005, "loss": 2.350701332092285, "step": 5155 }, { "epoch": 0.40646432794639337, "grad_norm": 0.9566806554794312, "learning_rate": 0.0005, "loss": 2.380251884460449, "step": 5156 }, { "epoch": 0.40654316121403233, "grad_norm": 0.9124642610549927, "learning_rate": 0.0005, "loss": 2.1906580924987793, "step": 5157 }, { "epoch": 0.40662199448167125, "grad_norm": 0.9412301778793335, "learning_rate": 0.0005, "loss": 2.1833200454711914, "step": 5158 }, { "epoch": 0.4067008277493102, "grad_norm": 0.694901168346405, "learning_rate": 0.0005, "loss": 2.0898044109344482, "step": 5159 }, { "epoch": 0.4067796610169492, "grad_norm": 0.7309814691543579, "learning_rate": 0.0005, "loss": 2.20141863822937, "step": 5160 }, { "epoch": 0.4068584942845881, "grad_norm": 0.8033947944641113, "learning_rate": 0.0005, "loss": 2.2746329307556152, "step": 5161 }, { "epoch": 0.40693732755222706, "grad_norm": 0.866167426109314, "learning_rate": 0.0005, "loss": 2.265364170074463, "step": 5162 }, { "epoch": 0.40701616081986597, "grad_norm": 0.7872490882873535, "learning_rate": 0.0005, "loss": 2.104170322418213, "step": 5163 }, { "epoch": 0.40709499408750494, "grad_norm": 1.1366589069366455, "learning_rate": 0.0005, "loss": 1.9452543258666992, "step": 5164 }, { "epoch": 0.40717382735514385, "grad_norm": 0.9405073523521423, "learning_rate": 0.0005, "loss": 2.3966076374053955, "step": 5165 }, { "epoch": 0.4072526606227828, "grad_norm": 0.6971275210380554, "learning_rate": 0.0005, "loss": 2.1176953315734863, "step": 5166 }, { "epoch": 0.4073314938904218, "grad_norm": 1.4246309995651245, "learning_rate": 0.0005, "loss": 2.2386844158172607, "step": 5167 }, { "epoch": 0.4074103271580607, "grad_norm": 0.7732844948768616, "learning_rate": 0.0005, "loss": 2.04278564453125, "step": 5168 }, { "epoch": 0.40748916042569966, "grad_norm": 0.7944534420967102, "learning_rate": 0.0005, "loss": 2.4062252044677734, "step": 5169 }, { "epoch": 0.40756799369333857, "grad_norm": 0.7007161378860474, "learning_rate": 0.0005, "loss": 1.805535078048706, "step": 5170 }, { "epoch": 0.40764682696097754, "grad_norm": 0.7569386959075928, "learning_rate": 0.0005, "loss": 2.371511220932007, "step": 5171 }, { "epoch": 0.40772566022861645, "grad_norm": 0.7168270945549011, "learning_rate": 0.0005, "loss": 2.3523683547973633, "step": 5172 }, { "epoch": 0.4078044934962554, "grad_norm": 1.3958686590194702, "learning_rate": 0.0005, "loss": 2.651233196258545, "step": 5173 }, { "epoch": 0.4078833267638944, "grad_norm": 0.8437748551368713, "learning_rate": 0.0005, "loss": 2.4796245098114014, "step": 5174 }, { "epoch": 0.4079621600315333, "grad_norm": 0.9986200332641602, "learning_rate": 0.0005, "loss": 2.0925769805908203, "step": 5175 }, { "epoch": 0.40804099329917226, "grad_norm": 0.8710203766822815, "learning_rate": 0.0005, "loss": 2.0497946739196777, "step": 5176 }, { "epoch": 0.4081198265668112, "grad_norm": 0.9249043464660645, "learning_rate": 0.0005, "loss": 2.220533847808838, "step": 5177 }, { "epoch": 0.40819865983445014, "grad_norm": 0.7076131105422974, "learning_rate": 0.0005, "loss": 2.224276304244995, "step": 5178 }, { "epoch": 0.4082774931020891, "grad_norm": 0.9113644361495972, "learning_rate": 0.0005, "loss": 2.1816630363464355, "step": 5179 }, { "epoch": 0.408356326369728, "grad_norm": 0.7402561902999878, "learning_rate": 0.0005, "loss": 2.1509976387023926, "step": 5180 }, { "epoch": 0.408435159637367, "grad_norm": 1.1293377876281738, "learning_rate": 0.0005, "loss": 1.9205303192138672, "step": 5181 }, { "epoch": 0.4085139929050059, "grad_norm": 0.7898808717727661, "learning_rate": 0.0005, "loss": 2.4672954082489014, "step": 5182 }, { "epoch": 0.40859282617264486, "grad_norm": 0.8249049186706543, "learning_rate": 0.0005, "loss": 2.1738219261169434, "step": 5183 }, { "epoch": 0.4086716594402838, "grad_norm": 0.9897887706756592, "learning_rate": 0.0005, "loss": 2.0902674198150635, "step": 5184 }, { "epoch": 0.40875049270792274, "grad_norm": 0.9511328339576721, "learning_rate": 0.0005, "loss": 1.9830174446105957, "step": 5185 }, { "epoch": 0.4088293259755617, "grad_norm": 0.9679328203201294, "learning_rate": 0.0005, "loss": 2.045257329940796, "step": 5186 }, { "epoch": 0.4089081592432006, "grad_norm": 0.903188169002533, "learning_rate": 0.0005, "loss": 2.007570743560791, "step": 5187 }, { "epoch": 0.4089869925108396, "grad_norm": 0.7926276922225952, "learning_rate": 0.0005, "loss": 1.9877488613128662, "step": 5188 }, { "epoch": 0.4090658257784785, "grad_norm": 0.827358603477478, "learning_rate": 0.0005, "loss": 2.244835615158081, "step": 5189 }, { "epoch": 0.40914465904611746, "grad_norm": 0.8984667658805847, "learning_rate": 0.0005, "loss": 2.0927748680114746, "step": 5190 }, { "epoch": 0.40922349231375643, "grad_norm": 0.9141483306884766, "learning_rate": 0.0005, "loss": 2.4368062019348145, "step": 5191 }, { "epoch": 0.40930232558139534, "grad_norm": 0.6564611196517944, "learning_rate": 0.0005, "loss": 2.013556480407715, "step": 5192 }, { "epoch": 0.4093811588490343, "grad_norm": 0.7599799036979675, "learning_rate": 0.0005, "loss": 2.321582555770874, "step": 5193 }, { "epoch": 0.4094599921166732, "grad_norm": 0.7516236901283264, "learning_rate": 0.0005, "loss": 2.0630099773406982, "step": 5194 }, { "epoch": 0.4095388253843122, "grad_norm": 0.7421900629997253, "learning_rate": 0.0005, "loss": 2.369048595428467, "step": 5195 }, { "epoch": 0.4096176586519511, "grad_norm": 0.7129167914390564, "learning_rate": 0.0005, "loss": 2.209341049194336, "step": 5196 }, { "epoch": 0.40969649191959007, "grad_norm": 1.5748257637023926, "learning_rate": 0.0005, "loss": 2.0487048625946045, "step": 5197 }, { "epoch": 0.40977532518722903, "grad_norm": 0.8225797414779663, "learning_rate": 0.0005, "loss": 2.202091693878174, "step": 5198 }, { "epoch": 0.40985415845486795, "grad_norm": 0.7191761136054993, "learning_rate": 0.0005, "loss": 2.225858688354492, "step": 5199 }, { "epoch": 0.4099329917225069, "grad_norm": 0.7343851327896118, "learning_rate": 0.0005, "loss": 1.8977564573287964, "step": 5200 }, { "epoch": 0.4100118249901458, "grad_norm": 0.797100841999054, "learning_rate": 0.0005, "loss": 2.037489414215088, "step": 5201 }, { "epoch": 0.4100906582577848, "grad_norm": 0.7025439143180847, "learning_rate": 0.0005, "loss": 2.0255861282348633, "step": 5202 }, { "epoch": 0.4101694915254237, "grad_norm": 0.7193772792816162, "learning_rate": 0.0005, "loss": 2.357849359512329, "step": 5203 }, { "epoch": 0.41024832479306267, "grad_norm": 0.8273177146911621, "learning_rate": 0.0005, "loss": 2.585857391357422, "step": 5204 }, { "epoch": 0.41032715806070164, "grad_norm": 1.1102980375289917, "learning_rate": 0.0005, "loss": 2.575456380844116, "step": 5205 }, { "epoch": 0.41040599132834055, "grad_norm": 0.6546550989151001, "learning_rate": 0.0005, "loss": 2.333918809890747, "step": 5206 }, { "epoch": 0.4104848245959795, "grad_norm": 0.7389711141586304, "learning_rate": 0.0005, "loss": 2.2846455574035645, "step": 5207 }, { "epoch": 0.4105636578636184, "grad_norm": 0.8716476559638977, "learning_rate": 0.0005, "loss": 2.0080878734588623, "step": 5208 }, { "epoch": 0.4106424911312574, "grad_norm": 0.739624559879303, "learning_rate": 0.0005, "loss": 1.9170453548431396, "step": 5209 }, { "epoch": 0.41072132439889636, "grad_norm": 0.6844218373298645, "learning_rate": 0.0005, "loss": 2.441587209701538, "step": 5210 }, { "epoch": 0.41080015766653527, "grad_norm": 0.8249683380126953, "learning_rate": 0.0005, "loss": 2.2929186820983887, "step": 5211 }, { "epoch": 0.41087899093417424, "grad_norm": 0.8374485373497009, "learning_rate": 0.0005, "loss": 2.0966732501983643, "step": 5212 }, { "epoch": 0.41095782420181315, "grad_norm": 0.8781185150146484, "learning_rate": 0.0005, "loss": 1.8995661735534668, "step": 5213 }, { "epoch": 0.4110366574694521, "grad_norm": 0.7153488993644714, "learning_rate": 0.0005, "loss": 2.1543965339660645, "step": 5214 }, { "epoch": 0.411115490737091, "grad_norm": 0.7791061401367188, "learning_rate": 0.0005, "loss": 2.486278772354126, "step": 5215 }, { "epoch": 0.41119432400473, "grad_norm": 0.7744368314743042, "learning_rate": 0.0005, "loss": 2.217924118041992, "step": 5216 }, { "epoch": 0.41127315727236896, "grad_norm": 0.7563591003417969, "learning_rate": 0.0005, "loss": 2.056905508041382, "step": 5217 }, { "epoch": 0.4113519905400079, "grad_norm": 0.8010368943214417, "learning_rate": 0.0005, "loss": 2.318467617034912, "step": 5218 }, { "epoch": 0.41143082380764684, "grad_norm": 0.9346733093261719, "learning_rate": 0.0005, "loss": 2.0358810424804688, "step": 5219 }, { "epoch": 0.41150965707528575, "grad_norm": 0.7311986684799194, "learning_rate": 0.0005, "loss": 2.1555864810943604, "step": 5220 }, { "epoch": 0.4115884903429247, "grad_norm": 0.6474584937095642, "learning_rate": 0.0005, "loss": 2.1209421157836914, "step": 5221 }, { "epoch": 0.4116673236105637, "grad_norm": 0.7272340655326843, "learning_rate": 0.0005, "loss": 2.0748963356018066, "step": 5222 }, { "epoch": 0.4117461568782026, "grad_norm": 0.780712902545929, "learning_rate": 0.0005, "loss": 2.6036510467529297, "step": 5223 }, { "epoch": 0.41182499014584156, "grad_norm": 0.7840883135795593, "learning_rate": 0.0005, "loss": 2.234161138534546, "step": 5224 }, { "epoch": 0.4119038234134805, "grad_norm": 0.685147762298584, "learning_rate": 0.0005, "loss": 2.061598777770996, "step": 5225 }, { "epoch": 0.41198265668111944, "grad_norm": 0.8294315934181213, "learning_rate": 0.0005, "loss": 2.3087449073791504, "step": 5226 }, { "epoch": 0.41206148994875835, "grad_norm": 0.9392314553260803, "learning_rate": 0.0005, "loss": 1.937800407409668, "step": 5227 }, { "epoch": 0.4121403232163973, "grad_norm": 0.7976779937744141, "learning_rate": 0.0005, "loss": 2.3080058097839355, "step": 5228 }, { "epoch": 0.4122191564840363, "grad_norm": 0.6637188792228699, "learning_rate": 0.0005, "loss": 1.8291661739349365, "step": 5229 }, { "epoch": 0.4122979897516752, "grad_norm": 4.515519142150879, "learning_rate": 0.0005, "loss": 2.300020456314087, "step": 5230 }, { "epoch": 0.41237682301931416, "grad_norm": 1.2487742900848389, "learning_rate": 0.0005, "loss": 2.3569955825805664, "step": 5231 }, { "epoch": 0.4124556562869531, "grad_norm": 0.9669857621192932, "learning_rate": 0.0005, "loss": 1.9196199178695679, "step": 5232 }, { "epoch": 0.41253448955459204, "grad_norm": 0.7209745645523071, "learning_rate": 0.0005, "loss": 1.7977938652038574, "step": 5233 }, { "epoch": 0.41261332282223095, "grad_norm": 0.7983418107032776, "learning_rate": 0.0005, "loss": 2.378626585006714, "step": 5234 }, { "epoch": 0.4126921560898699, "grad_norm": 0.8599093556404114, "learning_rate": 0.0005, "loss": 2.1988728046417236, "step": 5235 }, { "epoch": 0.4127709893575089, "grad_norm": 1.2284926176071167, "learning_rate": 0.0005, "loss": 2.0972232818603516, "step": 5236 }, { "epoch": 0.4128498226251478, "grad_norm": 0.7861712574958801, "learning_rate": 0.0005, "loss": 2.548801898956299, "step": 5237 }, { "epoch": 0.41292865589278677, "grad_norm": 0.9314821362495422, "learning_rate": 0.0005, "loss": 2.3258535861968994, "step": 5238 }, { "epoch": 0.4130074891604257, "grad_norm": 0.8164387941360474, "learning_rate": 0.0005, "loss": 2.110325813293457, "step": 5239 }, { "epoch": 0.41308632242806465, "grad_norm": 0.8945828080177307, "learning_rate": 0.0005, "loss": 2.060279369354248, "step": 5240 }, { "epoch": 0.4131651556957036, "grad_norm": 1.1029531955718994, "learning_rate": 0.0005, "loss": 2.2037930488586426, "step": 5241 }, { "epoch": 0.4132439889633425, "grad_norm": 0.7077769041061401, "learning_rate": 0.0005, "loss": 2.074258804321289, "step": 5242 }, { "epoch": 0.4133228222309815, "grad_norm": 0.8063687086105347, "learning_rate": 0.0005, "loss": 1.9789862632751465, "step": 5243 }, { "epoch": 0.4134016554986204, "grad_norm": 0.8994413018226624, "learning_rate": 0.0005, "loss": 2.283116102218628, "step": 5244 }, { "epoch": 0.41348048876625937, "grad_norm": 0.945802628993988, "learning_rate": 0.0005, "loss": 2.373063564300537, "step": 5245 }, { "epoch": 0.4135593220338983, "grad_norm": 1.0665279626846313, "learning_rate": 0.0005, "loss": 2.1915791034698486, "step": 5246 }, { "epoch": 0.41363815530153725, "grad_norm": 0.7977834343910217, "learning_rate": 0.0005, "loss": 2.3284497261047363, "step": 5247 }, { "epoch": 0.4137169885691762, "grad_norm": 1.1356641054153442, "learning_rate": 0.0005, "loss": 2.775458812713623, "step": 5248 }, { "epoch": 0.4137958218368151, "grad_norm": 0.9721737504005432, "learning_rate": 0.0005, "loss": 2.0109996795654297, "step": 5249 }, { "epoch": 0.4138746551044541, "grad_norm": 12.237715721130371, "learning_rate": 0.0005, "loss": 2.1696503162384033, "step": 5250 }, { "epoch": 0.413953488372093, "grad_norm": 0.7833325862884521, "learning_rate": 0.0005, "loss": 1.8561729192733765, "step": 5251 }, { "epoch": 0.41403232163973197, "grad_norm": 0.777529239654541, "learning_rate": 0.0005, "loss": 2.1182076930999756, "step": 5252 }, { "epoch": 0.41411115490737094, "grad_norm": 0.7359645366668701, "learning_rate": 0.0005, "loss": 2.1995601654052734, "step": 5253 }, { "epoch": 0.41418998817500985, "grad_norm": 0.7304294109344482, "learning_rate": 0.0005, "loss": 1.7537970542907715, "step": 5254 }, { "epoch": 0.4142688214426488, "grad_norm": 0.8461727499961853, "learning_rate": 0.0005, "loss": 2.1522300243377686, "step": 5255 }, { "epoch": 0.4143476547102877, "grad_norm": 0.9448644518852234, "learning_rate": 0.0005, "loss": 2.150599479675293, "step": 5256 }, { "epoch": 0.4144264879779267, "grad_norm": 0.9448061585426331, "learning_rate": 0.0005, "loss": 2.038684368133545, "step": 5257 }, { "epoch": 0.4145053212455656, "grad_norm": 0.6686558723449707, "learning_rate": 0.0005, "loss": 2.3775367736816406, "step": 5258 }, { "epoch": 0.4145841545132046, "grad_norm": 1.0235459804534912, "learning_rate": 0.0005, "loss": 2.2502455711364746, "step": 5259 }, { "epoch": 0.41466298778084354, "grad_norm": 0.9997145533561707, "learning_rate": 0.0005, "loss": 2.553814172744751, "step": 5260 }, { "epoch": 0.41474182104848245, "grad_norm": 0.8182546496391296, "learning_rate": 0.0005, "loss": 2.047367572784424, "step": 5261 }, { "epoch": 0.4148206543161214, "grad_norm": 0.7661029100418091, "learning_rate": 0.0005, "loss": 2.3319389820098877, "step": 5262 }, { "epoch": 0.41489948758376033, "grad_norm": 0.7484336495399475, "learning_rate": 0.0005, "loss": 2.047126054763794, "step": 5263 }, { "epoch": 0.4149783208513993, "grad_norm": 1.0721205472946167, "learning_rate": 0.0005, "loss": 2.0391221046447754, "step": 5264 }, { "epoch": 0.4150571541190382, "grad_norm": 0.8756629228591919, "learning_rate": 0.0005, "loss": 2.1257753372192383, "step": 5265 }, { "epoch": 0.4151359873866772, "grad_norm": 0.7163436412811279, "learning_rate": 0.0005, "loss": 2.074481725692749, "step": 5266 }, { "epoch": 0.41521482065431614, "grad_norm": 0.7645373940467834, "learning_rate": 0.0005, "loss": 1.983177661895752, "step": 5267 }, { "epoch": 0.41529365392195505, "grad_norm": 0.7000550031661987, "learning_rate": 0.0005, "loss": 2.2946507930755615, "step": 5268 }, { "epoch": 0.415372487189594, "grad_norm": 0.8277298808097839, "learning_rate": 0.0005, "loss": 2.6434056758880615, "step": 5269 }, { "epoch": 0.41545132045723293, "grad_norm": 0.878454864025116, "learning_rate": 0.0005, "loss": 2.1933634281158447, "step": 5270 }, { "epoch": 0.4155301537248719, "grad_norm": 0.7379858493804932, "learning_rate": 0.0005, "loss": 1.9379582405090332, "step": 5271 }, { "epoch": 0.41560898699251086, "grad_norm": 0.780091404914856, "learning_rate": 0.0005, "loss": 2.038555145263672, "step": 5272 }, { "epoch": 0.4156878202601498, "grad_norm": 0.6765053868293762, "learning_rate": 0.0005, "loss": 1.8517913818359375, "step": 5273 }, { "epoch": 0.41576665352778874, "grad_norm": 0.8530063033103943, "learning_rate": 0.0005, "loss": 2.2493438720703125, "step": 5274 }, { "epoch": 0.41584548679542765, "grad_norm": 0.7058964967727661, "learning_rate": 0.0005, "loss": 2.2441203594207764, "step": 5275 }, { "epoch": 0.4159243200630666, "grad_norm": 0.7059004306793213, "learning_rate": 0.0005, "loss": 2.5291543006896973, "step": 5276 }, { "epoch": 0.41600315333070553, "grad_norm": 0.7382227778434753, "learning_rate": 0.0005, "loss": 1.749955654144287, "step": 5277 }, { "epoch": 0.4160819865983445, "grad_norm": 0.8017101883888245, "learning_rate": 0.0005, "loss": 2.4427919387817383, "step": 5278 }, { "epoch": 0.41616081986598347, "grad_norm": 0.9575508832931519, "learning_rate": 0.0005, "loss": 2.3004839420318604, "step": 5279 }, { "epoch": 0.4162396531336224, "grad_norm": 0.9719989895820618, "learning_rate": 0.0005, "loss": 2.4376163482666016, "step": 5280 }, { "epoch": 0.41631848640126135, "grad_norm": 0.9474101066589355, "learning_rate": 0.0005, "loss": 2.3613054752349854, "step": 5281 }, { "epoch": 0.41639731966890026, "grad_norm": 0.8547304272651672, "learning_rate": 0.0005, "loss": 2.1981348991394043, "step": 5282 }, { "epoch": 0.4164761529365392, "grad_norm": 0.6951717138290405, "learning_rate": 0.0005, "loss": 2.085157632827759, "step": 5283 }, { "epoch": 0.4165549862041782, "grad_norm": 0.8152608275413513, "learning_rate": 0.0005, "loss": 2.144104242324829, "step": 5284 }, { "epoch": 0.4166338194718171, "grad_norm": 0.5927978157997131, "learning_rate": 0.0005, "loss": 1.8952609300613403, "step": 5285 }, { "epoch": 0.41671265273945607, "grad_norm": 0.7240961790084839, "learning_rate": 0.0005, "loss": 2.007401943206787, "step": 5286 }, { "epoch": 0.416791486007095, "grad_norm": 0.9481016397476196, "learning_rate": 0.0005, "loss": 2.1394031047821045, "step": 5287 }, { "epoch": 0.41687031927473395, "grad_norm": 0.7578754425048828, "learning_rate": 0.0005, "loss": 1.817147135734558, "step": 5288 }, { "epoch": 0.41694915254237286, "grad_norm": 1.045069932937622, "learning_rate": 0.0005, "loss": 2.4709312915802, "step": 5289 }, { "epoch": 0.4170279858100118, "grad_norm": 0.7201942205429077, "learning_rate": 0.0005, "loss": 2.0416178703308105, "step": 5290 }, { "epoch": 0.4171068190776508, "grad_norm": 0.943597137928009, "learning_rate": 0.0005, "loss": 2.123213291168213, "step": 5291 }, { "epoch": 0.4171856523452897, "grad_norm": 0.7906286716461182, "learning_rate": 0.0005, "loss": 1.9966614246368408, "step": 5292 }, { "epoch": 0.41726448561292867, "grad_norm": 0.8439268469810486, "learning_rate": 0.0005, "loss": 2.2459473609924316, "step": 5293 }, { "epoch": 0.4173433188805676, "grad_norm": 1.7472693920135498, "learning_rate": 0.0005, "loss": 2.656126022338867, "step": 5294 }, { "epoch": 0.41742215214820655, "grad_norm": 0.7048131227493286, "learning_rate": 0.0005, "loss": 1.9885532855987549, "step": 5295 }, { "epoch": 0.41750098541584546, "grad_norm": 1.109616756439209, "learning_rate": 0.0005, "loss": 2.10274600982666, "step": 5296 }, { "epoch": 0.4175798186834844, "grad_norm": 0.6721999049186707, "learning_rate": 0.0005, "loss": 1.9612668752670288, "step": 5297 }, { "epoch": 0.4176586519511234, "grad_norm": 1.0101807117462158, "learning_rate": 0.0005, "loss": 2.2696571350097656, "step": 5298 }, { "epoch": 0.4177374852187623, "grad_norm": 0.8500837087631226, "learning_rate": 0.0005, "loss": 2.1273105144500732, "step": 5299 }, { "epoch": 0.4178163184864013, "grad_norm": 0.9489452838897705, "learning_rate": 0.0005, "loss": 2.272467613220215, "step": 5300 }, { "epoch": 0.4178951517540402, "grad_norm": 0.9153518080711365, "learning_rate": 0.0005, "loss": 1.9980467557907104, "step": 5301 }, { "epoch": 0.41797398502167915, "grad_norm": 0.7210026383399963, "learning_rate": 0.0005, "loss": 2.1243653297424316, "step": 5302 }, { "epoch": 0.4180528182893181, "grad_norm": 0.7805284857749939, "learning_rate": 0.0005, "loss": 1.8722174167633057, "step": 5303 }, { "epoch": 0.41813165155695703, "grad_norm": 0.8697298765182495, "learning_rate": 0.0005, "loss": 2.0329971313476562, "step": 5304 }, { "epoch": 0.418210484824596, "grad_norm": 0.8243052363395691, "learning_rate": 0.0005, "loss": 2.3504979610443115, "step": 5305 }, { "epoch": 0.4182893180922349, "grad_norm": 0.7559241652488708, "learning_rate": 0.0005, "loss": 2.1847689151763916, "step": 5306 }, { "epoch": 0.4183681513598739, "grad_norm": 0.6949223279953003, "learning_rate": 0.0005, "loss": 2.029085636138916, "step": 5307 }, { "epoch": 0.4184469846275128, "grad_norm": 0.7609473466873169, "learning_rate": 0.0005, "loss": 1.9619355201721191, "step": 5308 }, { "epoch": 0.41852581789515175, "grad_norm": 0.7227525115013123, "learning_rate": 0.0005, "loss": 2.2845892906188965, "step": 5309 }, { "epoch": 0.4186046511627907, "grad_norm": 1.0229281187057495, "learning_rate": 0.0005, "loss": 2.1333835124969482, "step": 5310 }, { "epoch": 0.41868348443042963, "grad_norm": 0.8124006390571594, "learning_rate": 0.0005, "loss": 2.173799514770508, "step": 5311 }, { "epoch": 0.4187623176980686, "grad_norm": 0.844591498374939, "learning_rate": 0.0005, "loss": 2.4236958026885986, "step": 5312 }, { "epoch": 0.4188411509657075, "grad_norm": 0.8642271161079407, "learning_rate": 0.0005, "loss": 2.2955009937286377, "step": 5313 }, { "epoch": 0.4189199842333465, "grad_norm": 0.7998703122138977, "learning_rate": 0.0005, "loss": 2.118853807449341, "step": 5314 }, { "epoch": 0.41899881750098544, "grad_norm": 0.7476187348365784, "learning_rate": 0.0005, "loss": 2.1372995376586914, "step": 5315 }, { "epoch": 0.41907765076862435, "grad_norm": 0.6834120750427246, "learning_rate": 0.0005, "loss": 2.183788299560547, "step": 5316 }, { "epoch": 0.4191564840362633, "grad_norm": 0.7712175250053406, "learning_rate": 0.0005, "loss": 2.175448417663574, "step": 5317 }, { "epoch": 0.41923531730390223, "grad_norm": 0.9596613049507141, "learning_rate": 0.0005, "loss": 2.028317451477051, "step": 5318 }, { "epoch": 0.4193141505715412, "grad_norm": 0.7265832424163818, "learning_rate": 0.0005, "loss": 2.0585103034973145, "step": 5319 }, { "epoch": 0.4193929838391801, "grad_norm": 0.819325864315033, "learning_rate": 0.0005, "loss": 2.4412431716918945, "step": 5320 }, { "epoch": 0.4194718171068191, "grad_norm": 0.8250051736831665, "learning_rate": 0.0005, "loss": 2.090108633041382, "step": 5321 }, { "epoch": 0.41955065037445805, "grad_norm": 0.5949156880378723, "learning_rate": 0.0005, "loss": 2.1693105697631836, "step": 5322 }, { "epoch": 0.41962948364209696, "grad_norm": 0.6687288284301758, "learning_rate": 0.0005, "loss": 1.9720525741577148, "step": 5323 }, { "epoch": 0.4197083169097359, "grad_norm": 0.6439929604530334, "learning_rate": 0.0005, "loss": 1.8086531162261963, "step": 5324 }, { "epoch": 0.41978715017737483, "grad_norm": 0.8053128719329834, "learning_rate": 0.0005, "loss": 2.1309103965759277, "step": 5325 }, { "epoch": 0.4198659834450138, "grad_norm": 0.7662195563316345, "learning_rate": 0.0005, "loss": 2.2476325035095215, "step": 5326 }, { "epoch": 0.4199448167126527, "grad_norm": 0.710724413394928, "learning_rate": 0.0005, "loss": 2.1174819469451904, "step": 5327 }, { "epoch": 0.4200236499802917, "grad_norm": 0.7306267023086548, "learning_rate": 0.0005, "loss": 1.773437738418579, "step": 5328 }, { "epoch": 0.42010248324793065, "grad_norm": 0.8909225463867188, "learning_rate": 0.0005, "loss": 2.289616584777832, "step": 5329 }, { "epoch": 0.42018131651556956, "grad_norm": 0.7204698920249939, "learning_rate": 0.0005, "loss": 2.314131498336792, "step": 5330 }, { "epoch": 0.4202601497832085, "grad_norm": 0.873271644115448, "learning_rate": 0.0005, "loss": 2.3119053840637207, "step": 5331 }, { "epoch": 0.42033898305084744, "grad_norm": 0.756477415561676, "learning_rate": 0.0005, "loss": 1.9012298583984375, "step": 5332 }, { "epoch": 0.4204178163184864, "grad_norm": 0.6472693681716919, "learning_rate": 0.0005, "loss": 1.7658532857894897, "step": 5333 }, { "epoch": 0.42049664958612537, "grad_norm": 1.4124573469161987, "learning_rate": 0.0005, "loss": 2.2482285499572754, "step": 5334 }, { "epoch": 0.4205754828537643, "grad_norm": 0.950758159160614, "learning_rate": 0.0005, "loss": 1.8330159187316895, "step": 5335 }, { "epoch": 0.42065431612140325, "grad_norm": 0.8873202800750732, "learning_rate": 0.0005, "loss": 2.560985565185547, "step": 5336 }, { "epoch": 0.42073314938904216, "grad_norm": 0.8036081194877625, "learning_rate": 0.0005, "loss": 2.198517322540283, "step": 5337 }, { "epoch": 0.4208119826566811, "grad_norm": 0.7299888730049133, "learning_rate": 0.0005, "loss": 1.8426049947738647, "step": 5338 }, { "epoch": 0.42089081592432004, "grad_norm": 0.7149014472961426, "learning_rate": 0.0005, "loss": 2.188896656036377, "step": 5339 }, { "epoch": 0.420969649191959, "grad_norm": 0.8011002540588379, "learning_rate": 0.0005, "loss": 2.3694276809692383, "step": 5340 }, { "epoch": 0.421048482459598, "grad_norm": 0.7588143348693848, "learning_rate": 0.0005, "loss": 1.9589643478393555, "step": 5341 }, { "epoch": 0.4211273157272369, "grad_norm": 0.9439340233802795, "learning_rate": 0.0005, "loss": 2.373779296875, "step": 5342 }, { "epoch": 0.42120614899487585, "grad_norm": 1.0769561529159546, "learning_rate": 0.0005, "loss": 2.3368098735809326, "step": 5343 }, { "epoch": 0.42128498226251476, "grad_norm": 0.809670090675354, "learning_rate": 0.0005, "loss": 1.9911102056503296, "step": 5344 }, { "epoch": 0.42136381553015373, "grad_norm": 0.7471521496772766, "learning_rate": 0.0005, "loss": 1.8946189880371094, "step": 5345 }, { "epoch": 0.4214426487977927, "grad_norm": 0.8435695171356201, "learning_rate": 0.0005, "loss": 2.3466835021972656, "step": 5346 }, { "epoch": 0.4215214820654316, "grad_norm": 0.8118247389793396, "learning_rate": 0.0005, "loss": 2.1136889457702637, "step": 5347 }, { "epoch": 0.4216003153330706, "grad_norm": 0.7918103933334351, "learning_rate": 0.0005, "loss": 2.40649676322937, "step": 5348 }, { "epoch": 0.4216791486007095, "grad_norm": 0.8353357315063477, "learning_rate": 0.0005, "loss": 2.145327091217041, "step": 5349 }, { "epoch": 0.42175798186834845, "grad_norm": 0.6354847550392151, "learning_rate": 0.0005, "loss": 2.123988628387451, "step": 5350 }, { "epoch": 0.42183681513598736, "grad_norm": 0.7949324250221252, "learning_rate": 0.0005, "loss": 2.300095796585083, "step": 5351 }, { "epoch": 0.42191564840362633, "grad_norm": 0.6581177115440369, "learning_rate": 0.0005, "loss": 2.2591238021850586, "step": 5352 }, { "epoch": 0.4219944816712653, "grad_norm": 0.9876916408538818, "learning_rate": 0.0005, "loss": 2.4414830207824707, "step": 5353 }, { "epoch": 0.4220733149389042, "grad_norm": 0.8337066173553467, "learning_rate": 0.0005, "loss": 2.471296787261963, "step": 5354 }, { "epoch": 0.4221521482065432, "grad_norm": 0.8831012845039368, "learning_rate": 0.0005, "loss": 2.3024981021881104, "step": 5355 }, { "epoch": 0.4222309814741821, "grad_norm": 0.7153449654579163, "learning_rate": 0.0005, "loss": 2.2489185333251953, "step": 5356 }, { "epoch": 0.42230981474182105, "grad_norm": 0.8739980459213257, "learning_rate": 0.0005, "loss": 2.0440011024475098, "step": 5357 }, { "epoch": 0.42238864800945997, "grad_norm": 0.720363438129425, "learning_rate": 0.0005, "loss": 1.9930871725082397, "step": 5358 }, { "epoch": 0.42246748127709893, "grad_norm": 0.9132615327835083, "learning_rate": 0.0005, "loss": 2.2534751892089844, "step": 5359 }, { "epoch": 0.4225463145447379, "grad_norm": 0.6429372429847717, "learning_rate": 0.0005, "loss": 2.0121278762817383, "step": 5360 }, { "epoch": 0.4226251478123768, "grad_norm": 0.8588346242904663, "learning_rate": 0.0005, "loss": 2.3598203659057617, "step": 5361 }, { "epoch": 0.4227039810800158, "grad_norm": 0.7727931141853333, "learning_rate": 0.0005, "loss": 2.308528184890747, "step": 5362 }, { "epoch": 0.4227828143476547, "grad_norm": 0.726091742515564, "learning_rate": 0.0005, "loss": 2.2783443927764893, "step": 5363 }, { "epoch": 0.42286164761529366, "grad_norm": 0.9735282063484192, "learning_rate": 0.0005, "loss": 2.0271286964416504, "step": 5364 }, { "epoch": 0.4229404808829326, "grad_norm": 0.9160600900650024, "learning_rate": 0.0005, "loss": 2.0305838584899902, "step": 5365 }, { "epoch": 0.42301931415057153, "grad_norm": 0.6987941265106201, "learning_rate": 0.0005, "loss": 2.1777632236480713, "step": 5366 }, { "epoch": 0.4230981474182105, "grad_norm": 0.6992496252059937, "learning_rate": 0.0005, "loss": 2.0645527839660645, "step": 5367 }, { "epoch": 0.4231769806858494, "grad_norm": 0.8284111022949219, "learning_rate": 0.0005, "loss": 2.305604934692383, "step": 5368 }, { "epoch": 0.4232558139534884, "grad_norm": 0.759779691696167, "learning_rate": 0.0005, "loss": 2.0031745433807373, "step": 5369 }, { "epoch": 0.4233346472211273, "grad_norm": 0.6747938394546509, "learning_rate": 0.0005, "loss": 1.9500360488891602, "step": 5370 }, { "epoch": 0.42341348048876626, "grad_norm": 0.7891080975532532, "learning_rate": 0.0005, "loss": 2.069932460784912, "step": 5371 }, { "epoch": 0.4234923137564052, "grad_norm": 0.8489547967910767, "learning_rate": 0.0005, "loss": 2.067021369934082, "step": 5372 }, { "epoch": 0.42357114702404414, "grad_norm": 0.9451838731765747, "learning_rate": 0.0005, "loss": 2.1218910217285156, "step": 5373 }, { "epoch": 0.4236499802916831, "grad_norm": 0.8930579423904419, "learning_rate": 0.0005, "loss": 2.1899027824401855, "step": 5374 }, { "epoch": 0.423728813559322, "grad_norm": 0.7153492569923401, "learning_rate": 0.0005, "loss": 2.5292437076568604, "step": 5375 }, { "epoch": 0.423807646826961, "grad_norm": 1.1981898546218872, "learning_rate": 0.0005, "loss": 2.2781119346618652, "step": 5376 }, { "epoch": 0.42388648009459995, "grad_norm": 0.642795741558075, "learning_rate": 0.0005, "loss": 2.0091335773468018, "step": 5377 }, { "epoch": 0.42396531336223886, "grad_norm": 0.7774488925933838, "learning_rate": 0.0005, "loss": 2.329824447631836, "step": 5378 }, { "epoch": 0.4240441466298778, "grad_norm": 0.9531291127204895, "learning_rate": 0.0005, "loss": 2.1059112548828125, "step": 5379 }, { "epoch": 0.42412297989751674, "grad_norm": 0.7424734830856323, "learning_rate": 0.0005, "loss": 1.9824272394180298, "step": 5380 }, { "epoch": 0.4242018131651557, "grad_norm": 0.8399268984794617, "learning_rate": 0.0005, "loss": 2.2067160606384277, "step": 5381 }, { "epoch": 0.4242806464327946, "grad_norm": 0.7236492037773132, "learning_rate": 0.0005, "loss": 2.2805914878845215, "step": 5382 }, { "epoch": 0.4243594797004336, "grad_norm": 0.7269864082336426, "learning_rate": 0.0005, "loss": 2.262482166290283, "step": 5383 }, { "epoch": 0.42443831296807255, "grad_norm": 0.6949905753135681, "learning_rate": 0.0005, "loss": 2.0460281372070312, "step": 5384 }, { "epoch": 0.42451714623571146, "grad_norm": 0.8764810562133789, "learning_rate": 0.0005, "loss": 1.867128610610962, "step": 5385 }, { "epoch": 0.42459597950335043, "grad_norm": 0.6224263906478882, "learning_rate": 0.0005, "loss": 2.1184492111206055, "step": 5386 }, { "epoch": 0.42467481277098934, "grad_norm": 0.7727214097976685, "learning_rate": 0.0005, "loss": 2.251554012298584, "step": 5387 }, { "epoch": 0.4247536460386283, "grad_norm": 1.0241200923919678, "learning_rate": 0.0005, "loss": 2.3704941272735596, "step": 5388 }, { "epoch": 0.4248324793062672, "grad_norm": 0.8314588665962219, "learning_rate": 0.0005, "loss": 1.795351505279541, "step": 5389 }, { "epoch": 0.4249113125739062, "grad_norm": 0.8069274425506592, "learning_rate": 0.0005, "loss": 2.04681396484375, "step": 5390 }, { "epoch": 0.42499014584154515, "grad_norm": 0.8324095010757446, "learning_rate": 0.0005, "loss": 2.1695947647094727, "step": 5391 }, { "epoch": 0.42506897910918406, "grad_norm": 0.7424857020378113, "learning_rate": 0.0005, "loss": 1.9762299060821533, "step": 5392 }, { "epoch": 0.42514781237682303, "grad_norm": 0.6639153361320496, "learning_rate": 0.0005, "loss": 2.2553882598876953, "step": 5393 }, { "epoch": 0.42522664564446194, "grad_norm": 0.8478643894195557, "learning_rate": 0.0005, "loss": 2.3769540786743164, "step": 5394 }, { "epoch": 0.4253054789121009, "grad_norm": 0.7482966184616089, "learning_rate": 0.0005, "loss": 2.289885997772217, "step": 5395 }, { "epoch": 0.4253843121797399, "grad_norm": 1.0221366882324219, "learning_rate": 0.0005, "loss": 2.0351858139038086, "step": 5396 }, { "epoch": 0.4254631454473788, "grad_norm": 0.7454598546028137, "learning_rate": 0.0005, "loss": 2.161616802215576, "step": 5397 }, { "epoch": 0.42554197871501775, "grad_norm": 0.6469958424568176, "learning_rate": 0.0005, "loss": 1.9580612182617188, "step": 5398 }, { "epoch": 0.42562081198265667, "grad_norm": 0.7288214564323425, "learning_rate": 0.0005, "loss": 2.0247576236724854, "step": 5399 }, { "epoch": 0.42569964525029563, "grad_norm": 0.9465596079826355, "learning_rate": 0.0005, "loss": 2.530449628829956, "step": 5400 }, { "epoch": 0.42577847851793454, "grad_norm": 1.7540405988693237, "learning_rate": 0.0005, "loss": 2.4910340309143066, "step": 5401 }, { "epoch": 0.4258573117855735, "grad_norm": 0.8424733877182007, "learning_rate": 0.0005, "loss": 2.482809543609619, "step": 5402 }, { "epoch": 0.4259361450532125, "grad_norm": 0.884675920009613, "learning_rate": 0.0005, "loss": 2.049809217453003, "step": 5403 }, { "epoch": 0.4260149783208514, "grad_norm": 0.7541864514350891, "learning_rate": 0.0005, "loss": 2.226238965988159, "step": 5404 }, { "epoch": 0.42609381158849036, "grad_norm": 1.4276825189590454, "learning_rate": 0.0005, "loss": 2.093780994415283, "step": 5405 }, { "epoch": 0.42617264485612927, "grad_norm": 0.8012533187866211, "learning_rate": 0.0005, "loss": 2.2638540267944336, "step": 5406 }, { "epoch": 0.42625147812376823, "grad_norm": 0.7673342227935791, "learning_rate": 0.0005, "loss": 2.1756207942962646, "step": 5407 }, { "epoch": 0.42633031139140715, "grad_norm": 0.9624249935150146, "learning_rate": 0.0005, "loss": 1.9729317426681519, "step": 5408 }, { "epoch": 0.4264091446590461, "grad_norm": 0.6759654879570007, "learning_rate": 0.0005, "loss": 1.841889500617981, "step": 5409 }, { "epoch": 0.4264879779266851, "grad_norm": 1.0534132719039917, "learning_rate": 0.0005, "loss": 2.386812686920166, "step": 5410 }, { "epoch": 0.426566811194324, "grad_norm": 0.8456099033355713, "learning_rate": 0.0005, "loss": 2.7024495601654053, "step": 5411 }, { "epoch": 0.42664564446196296, "grad_norm": 1.0082588195800781, "learning_rate": 0.0005, "loss": 2.1240899562835693, "step": 5412 }, { "epoch": 0.42672447772960187, "grad_norm": 0.7538738250732422, "learning_rate": 0.0005, "loss": 2.2460389137268066, "step": 5413 }, { "epoch": 0.42680331099724084, "grad_norm": 1.7969123125076294, "learning_rate": 0.0005, "loss": 1.9661778211593628, "step": 5414 }, { "epoch": 0.4268821442648798, "grad_norm": 0.9361697435379028, "learning_rate": 0.0005, "loss": 2.059123992919922, "step": 5415 }, { "epoch": 0.4269609775325187, "grad_norm": 0.9590519666671753, "learning_rate": 0.0005, "loss": 2.1104607582092285, "step": 5416 }, { "epoch": 0.4270398108001577, "grad_norm": 0.6613437533378601, "learning_rate": 0.0005, "loss": 1.8886096477508545, "step": 5417 }, { "epoch": 0.4271186440677966, "grad_norm": 0.8423597812652588, "learning_rate": 0.0005, "loss": 2.3574018478393555, "step": 5418 }, { "epoch": 0.42719747733543556, "grad_norm": 0.6517740488052368, "learning_rate": 0.0005, "loss": 2.0846054553985596, "step": 5419 }, { "epoch": 0.42727631060307447, "grad_norm": 0.8688225746154785, "learning_rate": 0.0005, "loss": 2.259154796600342, "step": 5420 }, { "epoch": 0.42735514387071344, "grad_norm": 0.9004712700843811, "learning_rate": 0.0005, "loss": 2.0847527980804443, "step": 5421 }, { "epoch": 0.4274339771383524, "grad_norm": 0.7038086652755737, "learning_rate": 0.0005, "loss": 2.262080192565918, "step": 5422 }, { "epoch": 0.4275128104059913, "grad_norm": 0.7991819381713867, "learning_rate": 0.0005, "loss": 2.0760605335235596, "step": 5423 }, { "epoch": 0.4275916436736303, "grad_norm": 1.2042888402938843, "learning_rate": 0.0005, "loss": 2.2058563232421875, "step": 5424 }, { "epoch": 0.4276704769412692, "grad_norm": 0.6857059001922607, "learning_rate": 0.0005, "loss": 2.1549875736236572, "step": 5425 }, { "epoch": 0.42774931020890816, "grad_norm": 0.7380450367927551, "learning_rate": 0.0005, "loss": 2.1138103008270264, "step": 5426 }, { "epoch": 0.42782814347654713, "grad_norm": 0.7313492894172668, "learning_rate": 0.0005, "loss": 2.459656238555908, "step": 5427 }, { "epoch": 0.42790697674418604, "grad_norm": 1.5818800926208496, "learning_rate": 0.0005, "loss": 2.37703800201416, "step": 5428 }, { "epoch": 0.427985810011825, "grad_norm": 0.8648122549057007, "learning_rate": 0.0005, "loss": 2.1963424682617188, "step": 5429 }, { "epoch": 0.4280646432794639, "grad_norm": 0.6191871762275696, "learning_rate": 0.0005, "loss": 2.110419988632202, "step": 5430 }, { "epoch": 0.4281434765471029, "grad_norm": 0.7285412549972534, "learning_rate": 0.0005, "loss": 1.9100109338760376, "step": 5431 }, { "epoch": 0.4282223098147418, "grad_norm": 1.3777703046798706, "learning_rate": 0.0005, "loss": 2.3643341064453125, "step": 5432 }, { "epoch": 0.42830114308238076, "grad_norm": 0.7715537548065186, "learning_rate": 0.0005, "loss": 2.5113673210144043, "step": 5433 }, { "epoch": 0.42837997635001973, "grad_norm": 1.0561652183532715, "learning_rate": 0.0005, "loss": 2.2824645042419434, "step": 5434 }, { "epoch": 0.42845880961765864, "grad_norm": 0.6920111775398254, "learning_rate": 0.0005, "loss": 2.152332305908203, "step": 5435 }, { "epoch": 0.4285376428852976, "grad_norm": 0.7822750806808472, "learning_rate": 0.0005, "loss": 2.125659465789795, "step": 5436 }, { "epoch": 0.4286164761529365, "grad_norm": 0.7414671182632446, "learning_rate": 0.0005, "loss": 2.173340082168579, "step": 5437 }, { "epoch": 0.4286953094205755, "grad_norm": 0.7088406682014465, "learning_rate": 0.0005, "loss": 2.241816282272339, "step": 5438 }, { "epoch": 0.4287741426882144, "grad_norm": 0.7468655109405518, "learning_rate": 0.0005, "loss": 2.3419015407562256, "step": 5439 }, { "epoch": 0.42885297595585337, "grad_norm": 0.682339072227478, "learning_rate": 0.0005, "loss": 1.989567756652832, "step": 5440 }, { "epoch": 0.42893180922349233, "grad_norm": 0.7936223745346069, "learning_rate": 0.0005, "loss": 2.063458204269409, "step": 5441 }, { "epoch": 0.42901064249113124, "grad_norm": 0.8361531496047974, "learning_rate": 0.0005, "loss": 2.0408432483673096, "step": 5442 }, { "epoch": 0.4290894757587702, "grad_norm": 0.6286973357200623, "learning_rate": 0.0005, "loss": 1.9867948293685913, "step": 5443 }, { "epoch": 0.4291683090264091, "grad_norm": 0.6500899791717529, "learning_rate": 0.0005, "loss": 2.0862417221069336, "step": 5444 }, { "epoch": 0.4292471422940481, "grad_norm": 0.611491858959198, "learning_rate": 0.0005, "loss": 1.6902804374694824, "step": 5445 }, { "epoch": 0.42932597556168706, "grad_norm": 0.6213261485099792, "learning_rate": 0.0005, "loss": 1.8632704019546509, "step": 5446 }, { "epoch": 0.42940480882932597, "grad_norm": 0.8967099189758301, "learning_rate": 0.0005, "loss": 1.8868906497955322, "step": 5447 }, { "epoch": 0.42948364209696493, "grad_norm": 0.7647256851196289, "learning_rate": 0.0005, "loss": 2.4923486709594727, "step": 5448 }, { "epoch": 0.42956247536460385, "grad_norm": 0.7188752293586731, "learning_rate": 0.0005, "loss": 2.3263697624206543, "step": 5449 }, { "epoch": 0.4296413086322428, "grad_norm": 0.886204183101654, "learning_rate": 0.0005, "loss": 1.770539402961731, "step": 5450 }, { "epoch": 0.4297201418998817, "grad_norm": 0.9337040185928345, "learning_rate": 0.0005, "loss": 2.494016170501709, "step": 5451 }, { "epoch": 0.4297989751675207, "grad_norm": 0.751958966255188, "learning_rate": 0.0005, "loss": 1.9890425205230713, "step": 5452 }, { "epoch": 0.42987780843515966, "grad_norm": 0.7781546711921692, "learning_rate": 0.0005, "loss": 2.218691349029541, "step": 5453 }, { "epoch": 0.42995664170279857, "grad_norm": 0.9002652168273926, "learning_rate": 0.0005, "loss": 2.2100424766540527, "step": 5454 }, { "epoch": 0.43003547497043754, "grad_norm": 0.7051560878753662, "learning_rate": 0.0005, "loss": 2.050800323486328, "step": 5455 }, { "epoch": 0.43011430823807645, "grad_norm": 0.8477756381034851, "learning_rate": 0.0005, "loss": 2.4485044479370117, "step": 5456 }, { "epoch": 0.4301931415057154, "grad_norm": 1.0536118745803833, "learning_rate": 0.0005, "loss": 2.1613550186157227, "step": 5457 }, { "epoch": 0.4302719747733544, "grad_norm": 0.74080491065979, "learning_rate": 0.0005, "loss": 2.2321617603302, "step": 5458 }, { "epoch": 0.4303508080409933, "grad_norm": 0.9301217794418335, "learning_rate": 0.0005, "loss": 2.163083553314209, "step": 5459 }, { "epoch": 0.43042964130863226, "grad_norm": 0.5386443138122559, "learning_rate": 0.0005, "loss": 2.031689167022705, "step": 5460 }, { "epoch": 0.43050847457627117, "grad_norm": 0.5821754336357117, "learning_rate": 0.0005, "loss": 2.13486385345459, "step": 5461 }, { "epoch": 0.43058730784391014, "grad_norm": 0.7450363636016846, "learning_rate": 0.0005, "loss": 2.1648826599121094, "step": 5462 }, { "epoch": 0.43066614111154905, "grad_norm": 0.6910276412963867, "learning_rate": 0.0005, "loss": 2.2694408893585205, "step": 5463 }, { "epoch": 0.430744974379188, "grad_norm": 0.7680632472038269, "learning_rate": 0.0005, "loss": 2.1216657161712646, "step": 5464 }, { "epoch": 0.430823807646827, "grad_norm": 1.1369071006774902, "learning_rate": 0.0005, "loss": 2.2671778202056885, "step": 5465 }, { "epoch": 0.4309026409144659, "grad_norm": 0.7277780175209045, "learning_rate": 0.0005, "loss": 2.003462553024292, "step": 5466 }, { "epoch": 0.43098147418210486, "grad_norm": 0.9290866255760193, "learning_rate": 0.0005, "loss": 2.154970407485962, "step": 5467 }, { "epoch": 0.4310603074497438, "grad_norm": 0.8346081972122192, "learning_rate": 0.0005, "loss": 2.0651769638061523, "step": 5468 }, { "epoch": 0.43113914071738274, "grad_norm": 0.9631778597831726, "learning_rate": 0.0005, "loss": 2.4817466735839844, "step": 5469 }, { "epoch": 0.43121797398502165, "grad_norm": 0.8213978409767151, "learning_rate": 0.0005, "loss": 2.130277156829834, "step": 5470 }, { "epoch": 0.4312968072526606, "grad_norm": 0.7341722249984741, "learning_rate": 0.0005, "loss": 2.3324146270751953, "step": 5471 }, { "epoch": 0.4313756405202996, "grad_norm": 0.7197587490081787, "learning_rate": 0.0005, "loss": 2.2368061542510986, "step": 5472 }, { "epoch": 0.4314544737879385, "grad_norm": 0.7225975394248962, "learning_rate": 0.0005, "loss": 2.0290114879608154, "step": 5473 }, { "epoch": 0.43153330705557746, "grad_norm": 1.2004313468933105, "learning_rate": 0.0005, "loss": 2.306405782699585, "step": 5474 }, { "epoch": 0.4316121403232164, "grad_norm": 1.055243730545044, "learning_rate": 0.0005, "loss": 2.2475454807281494, "step": 5475 }, { "epoch": 0.43169097359085534, "grad_norm": 0.6530787944793701, "learning_rate": 0.0005, "loss": 1.8778977394104004, "step": 5476 }, { "epoch": 0.4317698068584943, "grad_norm": 0.7835645079612732, "learning_rate": 0.0005, "loss": 2.5015461444854736, "step": 5477 }, { "epoch": 0.4318486401261332, "grad_norm": 0.7386176586151123, "learning_rate": 0.0005, "loss": 1.8042508363723755, "step": 5478 }, { "epoch": 0.4319274733937722, "grad_norm": 0.6533234119415283, "learning_rate": 0.0005, "loss": 2.0637476444244385, "step": 5479 }, { "epoch": 0.4320063066614111, "grad_norm": 0.7680717706680298, "learning_rate": 0.0005, "loss": 1.9397168159484863, "step": 5480 }, { "epoch": 0.43208513992905007, "grad_norm": 0.8816759586334229, "learning_rate": 0.0005, "loss": 2.416965961456299, "step": 5481 }, { "epoch": 0.432163973196689, "grad_norm": 0.7191063165664673, "learning_rate": 0.0005, "loss": 2.2647602558135986, "step": 5482 }, { "epoch": 0.43224280646432794, "grad_norm": 0.7111219167709351, "learning_rate": 0.0005, "loss": 2.190497875213623, "step": 5483 }, { "epoch": 0.4323216397319669, "grad_norm": 0.706217885017395, "learning_rate": 0.0005, "loss": 2.122163772583008, "step": 5484 }, { "epoch": 0.4324004729996058, "grad_norm": 0.7509106993675232, "learning_rate": 0.0005, "loss": 2.167407512664795, "step": 5485 }, { "epoch": 0.4324793062672448, "grad_norm": 0.7874780893325806, "learning_rate": 0.0005, "loss": 2.1278374195098877, "step": 5486 }, { "epoch": 0.4325581395348837, "grad_norm": 0.7693070769309998, "learning_rate": 0.0005, "loss": 2.0530760288238525, "step": 5487 }, { "epoch": 0.43263697280252267, "grad_norm": 0.7575477361679077, "learning_rate": 0.0005, "loss": 2.3843941688537598, "step": 5488 }, { "epoch": 0.43271580607016163, "grad_norm": 0.9547325372695923, "learning_rate": 0.0005, "loss": 2.140752077102661, "step": 5489 }, { "epoch": 0.43279463933780055, "grad_norm": 2.358363628387451, "learning_rate": 0.0005, "loss": 2.329526424407959, "step": 5490 }, { "epoch": 0.4328734726054395, "grad_norm": 0.7768775820732117, "learning_rate": 0.0005, "loss": 2.0612053871154785, "step": 5491 }, { "epoch": 0.4329523058730784, "grad_norm": 0.6274071931838989, "learning_rate": 0.0005, "loss": 2.1635141372680664, "step": 5492 }, { "epoch": 0.4330311391407174, "grad_norm": 0.8046379089355469, "learning_rate": 0.0005, "loss": 2.2222490310668945, "step": 5493 }, { "epoch": 0.4331099724083563, "grad_norm": 0.8042774200439453, "learning_rate": 0.0005, "loss": 2.3304150104522705, "step": 5494 }, { "epoch": 0.43318880567599527, "grad_norm": 0.9051457047462463, "learning_rate": 0.0005, "loss": 1.861386775970459, "step": 5495 }, { "epoch": 0.43326763894363424, "grad_norm": 0.9659410715103149, "learning_rate": 0.0005, "loss": 2.213444709777832, "step": 5496 }, { "epoch": 0.43334647221127315, "grad_norm": 0.8682563304901123, "learning_rate": 0.0005, "loss": 2.1395325660705566, "step": 5497 }, { "epoch": 0.4334253054789121, "grad_norm": 0.6558504104614258, "learning_rate": 0.0005, "loss": 1.9684487581253052, "step": 5498 }, { "epoch": 0.433504138746551, "grad_norm": 0.9440422654151917, "learning_rate": 0.0005, "loss": 2.3528807163238525, "step": 5499 }, { "epoch": 0.43358297201419, "grad_norm": 0.7584413886070251, "learning_rate": 0.0005, "loss": 2.008554458618164, "step": 5500 }, { "epoch": 0.4336618052818289, "grad_norm": 0.8285616040229797, "learning_rate": 0.0005, "loss": 2.0608413219451904, "step": 5501 }, { "epoch": 0.43374063854946787, "grad_norm": 0.9356375932693481, "learning_rate": 0.0005, "loss": 2.63295316696167, "step": 5502 }, { "epoch": 0.43381947181710684, "grad_norm": 0.777155876159668, "learning_rate": 0.0005, "loss": 2.230548143386841, "step": 5503 }, { "epoch": 0.43389830508474575, "grad_norm": 0.9178016781806946, "learning_rate": 0.0005, "loss": 1.9200012683868408, "step": 5504 }, { "epoch": 0.4339771383523847, "grad_norm": 0.8643798232078552, "learning_rate": 0.0005, "loss": 2.0670206546783447, "step": 5505 }, { "epoch": 0.43405597162002363, "grad_norm": 1.0684541463851929, "learning_rate": 0.0005, "loss": 2.682321071624756, "step": 5506 }, { "epoch": 0.4341348048876626, "grad_norm": 0.6471952795982361, "learning_rate": 0.0005, "loss": 1.9601731300354004, "step": 5507 }, { "epoch": 0.43421363815530156, "grad_norm": 0.7264888286590576, "learning_rate": 0.0005, "loss": 1.9508907794952393, "step": 5508 }, { "epoch": 0.4342924714229405, "grad_norm": 0.9522725939750671, "learning_rate": 0.0005, "loss": 1.8956142663955688, "step": 5509 }, { "epoch": 0.43437130469057944, "grad_norm": 1.0823872089385986, "learning_rate": 0.0005, "loss": 2.2768795490264893, "step": 5510 }, { "epoch": 0.43445013795821835, "grad_norm": 0.9577037692070007, "learning_rate": 0.0005, "loss": 2.3172366619110107, "step": 5511 }, { "epoch": 0.4345289712258573, "grad_norm": 0.8583095073699951, "learning_rate": 0.0005, "loss": 1.9445726871490479, "step": 5512 }, { "epoch": 0.43460780449349623, "grad_norm": 0.7561057806015015, "learning_rate": 0.0005, "loss": 1.9470560550689697, "step": 5513 }, { "epoch": 0.4346866377611352, "grad_norm": 0.7670291662216187, "learning_rate": 0.0005, "loss": 2.6091015338897705, "step": 5514 }, { "epoch": 0.43476547102877416, "grad_norm": 0.9394797086715698, "learning_rate": 0.0005, "loss": 2.0769355297088623, "step": 5515 }, { "epoch": 0.4348443042964131, "grad_norm": 0.6538980007171631, "learning_rate": 0.0005, "loss": 2.057021379470825, "step": 5516 }, { "epoch": 0.43492313756405204, "grad_norm": 0.7558830380439758, "learning_rate": 0.0005, "loss": 2.0677237510681152, "step": 5517 }, { "epoch": 0.43500197083169095, "grad_norm": 0.7701875567436218, "learning_rate": 0.0005, "loss": 2.075559139251709, "step": 5518 }, { "epoch": 0.4350808040993299, "grad_norm": 0.825631856918335, "learning_rate": 0.0005, "loss": 1.777994990348816, "step": 5519 }, { "epoch": 0.4351596373669689, "grad_norm": 0.928331732749939, "learning_rate": 0.0005, "loss": 2.0379226207733154, "step": 5520 }, { "epoch": 0.4352384706346078, "grad_norm": 0.7779189944267273, "learning_rate": 0.0005, "loss": 2.152174472808838, "step": 5521 }, { "epoch": 0.43531730390224677, "grad_norm": 1.028089165687561, "learning_rate": 0.0005, "loss": 1.8115565776824951, "step": 5522 }, { "epoch": 0.4353961371698857, "grad_norm": 0.7701312303543091, "learning_rate": 0.0005, "loss": 2.3572821617126465, "step": 5523 }, { "epoch": 0.43547497043752464, "grad_norm": 0.7240291237831116, "learning_rate": 0.0005, "loss": 1.9315180778503418, "step": 5524 }, { "epoch": 0.43555380370516356, "grad_norm": 0.7598422169685364, "learning_rate": 0.0005, "loss": 2.141890048980713, "step": 5525 }, { "epoch": 0.4356326369728025, "grad_norm": 0.7549421191215515, "learning_rate": 0.0005, "loss": 2.366246461868286, "step": 5526 }, { "epoch": 0.4357114702404415, "grad_norm": 0.6754224300384521, "learning_rate": 0.0005, "loss": 2.225423812866211, "step": 5527 }, { "epoch": 0.4357903035080804, "grad_norm": 0.7667760252952576, "learning_rate": 0.0005, "loss": 2.4150209426879883, "step": 5528 }, { "epoch": 0.43586913677571937, "grad_norm": 0.8533955216407776, "learning_rate": 0.0005, "loss": 2.012284278869629, "step": 5529 }, { "epoch": 0.4359479700433583, "grad_norm": 0.9074493050575256, "learning_rate": 0.0005, "loss": 2.233433246612549, "step": 5530 }, { "epoch": 0.43602680331099725, "grad_norm": 0.9093543291091919, "learning_rate": 0.0005, "loss": 1.792149305343628, "step": 5531 }, { "epoch": 0.43610563657863616, "grad_norm": 0.8016659021377563, "learning_rate": 0.0005, "loss": 1.88339364528656, "step": 5532 }, { "epoch": 0.4361844698462751, "grad_norm": 1.293699860572815, "learning_rate": 0.0005, "loss": 1.9826946258544922, "step": 5533 }, { "epoch": 0.4362633031139141, "grad_norm": 0.8499022722244263, "learning_rate": 0.0005, "loss": 2.0437655448913574, "step": 5534 }, { "epoch": 0.436342136381553, "grad_norm": 0.7333089113235474, "learning_rate": 0.0005, "loss": 2.1699421405792236, "step": 5535 }, { "epoch": 0.43642096964919197, "grad_norm": 0.694879949092865, "learning_rate": 0.0005, "loss": 2.0855250358581543, "step": 5536 }, { "epoch": 0.4364998029168309, "grad_norm": 0.6848726868629456, "learning_rate": 0.0005, "loss": 2.3233742713928223, "step": 5537 }, { "epoch": 0.43657863618446985, "grad_norm": 0.8343655467033386, "learning_rate": 0.0005, "loss": 2.296319007873535, "step": 5538 }, { "epoch": 0.4366574694521088, "grad_norm": 1.33335280418396, "learning_rate": 0.0005, "loss": 1.9985463619232178, "step": 5539 }, { "epoch": 0.4367363027197477, "grad_norm": 0.8533251881599426, "learning_rate": 0.0005, "loss": 2.0671160221099854, "step": 5540 }, { "epoch": 0.4368151359873867, "grad_norm": 0.8666346073150635, "learning_rate": 0.0005, "loss": 2.2517647743225098, "step": 5541 }, { "epoch": 0.4368939692550256, "grad_norm": 0.9379417300224304, "learning_rate": 0.0005, "loss": 2.232494831085205, "step": 5542 }, { "epoch": 0.43697280252266457, "grad_norm": 0.7787320017814636, "learning_rate": 0.0005, "loss": 2.107504367828369, "step": 5543 }, { "epoch": 0.4370516357903035, "grad_norm": 1.196872353553772, "learning_rate": 0.0005, "loss": 2.2855443954467773, "step": 5544 }, { "epoch": 0.43713046905794245, "grad_norm": 0.8859530687332153, "learning_rate": 0.0005, "loss": 2.2267353534698486, "step": 5545 }, { "epoch": 0.4372093023255814, "grad_norm": 0.7461510896682739, "learning_rate": 0.0005, "loss": 1.995384931564331, "step": 5546 }, { "epoch": 0.43728813559322033, "grad_norm": 0.7932303547859192, "learning_rate": 0.0005, "loss": 2.2525291442871094, "step": 5547 }, { "epoch": 0.4373669688608593, "grad_norm": 0.7225112915039062, "learning_rate": 0.0005, "loss": 1.9308878183364868, "step": 5548 }, { "epoch": 0.4374458021284982, "grad_norm": 0.7317027449607849, "learning_rate": 0.0005, "loss": 1.8596868515014648, "step": 5549 }, { "epoch": 0.4375246353961372, "grad_norm": 0.764611542224884, "learning_rate": 0.0005, "loss": 2.1239867210388184, "step": 5550 }, { "epoch": 0.43760346866377614, "grad_norm": 0.9302797913551331, "learning_rate": 0.0005, "loss": 2.2158310413360596, "step": 5551 }, { "epoch": 0.43768230193141505, "grad_norm": 0.9301177859306335, "learning_rate": 0.0005, "loss": 2.249678611755371, "step": 5552 }, { "epoch": 0.437761135199054, "grad_norm": 0.7902008891105652, "learning_rate": 0.0005, "loss": 2.112609386444092, "step": 5553 }, { "epoch": 0.43783996846669293, "grad_norm": 1.0031253099441528, "learning_rate": 0.0005, "loss": 2.04443621635437, "step": 5554 }, { "epoch": 0.4379188017343319, "grad_norm": 0.8358235955238342, "learning_rate": 0.0005, "loss": 1.7530934810638428, "step": 5555 }, { "epoch": 0.4379976350019708, "grad_norm": 0.9067978858947754, "learning_rate": 0.0005, "loss": 1.9924869537353516, "step": 5556 }, { "epoch": 0.4380764682696098, "grad_norm": 0.8360188007354736, "learning_rate": 0.0005, "loss": 2.5524110794067383, "step": 5557 }, { "epoch": 0.43815530153724874, "grad_norm": 0.8658453822135925, "learning_rate": 0.0005, "loss": 2.3046624660491943, "step": 5558 }, { "epoch": 0.43823413480488765, "grad_norm": 0.9689998030662537, "learning_rate": 0.0005, "loss": 1.9758222103118896, "step": 5559 }, { "epoch": 0.4383129680725266, "grad_norm": 0.8087487816810608, "learning_rate": 0.0005, "loss": 1.9699019193649292, "step": 5560 }, { "epoch": 0.43839180134016553, "grad_norm": 4.90369987487793, "learning_rate": 0.0005, "loss": 2.9918084144592285, "step": 5561 }, { "epoch": 0.4384706346078045, "grad_norm": 0.9476282596588135, "learning_rate": 0.0005, "loss": 2.192643642425537, "step": 5562 }, { "epoch": 0.4385494678754434, "grad_norm": 0.9275436401367188, "learning_rate": 0.0005, "loss": 2.080282211303711, "step": 5563 }, { "epoch": 0.4386283011430824, "grad_norm": 0.8986785411834717, "learning_rate": 0.0005, "loss": 2.6499147415161133, "step": 5564 }, { "epoch": 0.43870713441072134, "grad_norm": 0.872594952583313, "learning_rate": 0.0005, "loss": 2.2188682556152344, "step": 5565 }, { "epoch": 0.43878596767836026, "grad_norm": 1.3123313188552856, "learning_rate": 0.0005, "loss": 2.361514091491699, "step": 5566 }, { "epoch": 0.4388648009459992, "grad_norm": 0.7466089129447937, "learning_rate": 0.0005, "loss": 2.290750026702881, "step": 5567 }, { "epoch": 0.43894363421363813, "grad_norm": 0.8476835489273071, "learning_rate": 0.0005, "loss": 2.2515311241149902, "step": 5568 }, { "epoch": 0.4390224674812771, "grad_norm": 0.8535008430480957, "learning_rate": 0.0005, "loss": 2.398678779602051, "step": 5569 }, { "epoch": 0.43910130074891607, "grad_norm": 1.0523306131362915, "learning_rate": 0.0005, "loss": 2.054886817932129, "step": 5570 }, { "epoch": 0.439180134016555, "grad_norm": 0.689387321472168, "learning_rate": 0.0005, "loss": 1.8573014736175537, "step": 5571 }, { "epoch": 0.43925896728419395, "grad_norm": 0.6704949736595154, "learning_rate": 0.0005, "loss": 1.9271728992462158, "step": 5572 }, { "epoch": 0.43933780055183286, "grad_norm": 0.6618713736534119, "learning_rate": 0.0005, "loss": 1.9820811748504639, "step": 5573 }, { "epoch": 0.4394166338194718, "grad_norm": 0.6958328485488892, "learning_rate": 0.0005, "loss": 1.8253552913665771, "step": 5574 }, { "epoch": 0.43949546708711074, "grad_norm": 0.7227240800857544, "learning_rate": 0.0005, "loss": 2.0364990234375, "step": 5575 }, { "epoch": 0.4395743003547497, "grad_norm": 1.0786290168762207, "learning_rate": 0.0005, "loss": 2.1678028106689453, "step": 5576 }, { "epoch": 0.43965313362238867, "grad_norm": 0.6535278558731079, "learning_rate": 0.0005, "loss": 2.057880401611328, "step": 5577 }, { "epoch": 0.4397319668900276, "grad_norm": 0.9615755677223206, "learning_rate": 0.0005, "loss": 2.136718511581421, "step": 5578 }, { "epoch": 0.43981080015766655, "grad_norm": 0.8335837721824646, "learning_rate": 0.0005, "loss": 2.2134881019592285, "step": 5579 }, { "epoch": 0.43988963342530546, "grad_norm": 0.7228657007217407, "learning_rate": 0.0005, "loss": 1.9361909627914429, "step": 5580 }, { "epoch": 0.4399684666929444, "grad_norm": 0.7872079014778137, "learning_rate": 0.0005, "loss": 2.290688991546631, "step": 5581 }, { "epoch": 0.4400472999605834, "grad_norm": 0.7255532145500183, "learning_rate": 0.0005, "loss": 2.0268123149871826, "step": 5582 }, { "epoch": 0.4401261332282223, "grad_norm": 0.710199236869812, "learning_rate": 0.0005, "loss": 2.189418315887451, "step": 5583 }, { "epoch": 0.44020496649586127, "grad_norm": 0.8069461584091187, "learning_rate": 0.0005, "loss": 1.9636033773422241, "step": 5584 }, { "epoch": 0.4402837997635002, "grad_norm": 0.7482892274856567, "learning_rate": 0.0005, "loss": 2.165142059326172, "step": 5585 }, { "epoch": 0.44036263303113915, "grad_norm": 0.755458652973175, "learning_rate": 0.0005, "loss": 2.3614747524261475, "step": 5586 }, { "epoch": 0.44044146629877806, "grad_norm": 0.9881508946418762, "learning_rate": 0.0005, "loss": 2.148759603500366, "step": 5587 }, { "epoch": 0.44052029956641703, "grad_norm": 0.8374794721603394, "learning_rate": 0.0005, "loss": 2.030235528945923, "step": 5588 }, { "epoch": 0.440599132834056, "grad_norm": 0.6438601016998291, "learning_rate": 0.0005, "loss": 2.1100881099700928, "step": 5589 }, { "epoch": 0.4406779661016949, "grad_norm": 0.7603535056114197, "learning_rate": 0.0005, "loss": 1.9416860342025757, "step": 5590 }, { "epoch": 0.4407567993693339, "grad_norm": 0.7358622550964355, "learning_rate": 0.0005, "loss": 1.9987258911132812, "step": 5591 }, { "epoch": 0.4408356326369728, "grad_norm": 0.7078192234039307, "learning_rate": 0.0005, "loss": 1.851302146911621, "step": 5592 }, { "epoch": 0.44091446590461175, "grad_norm": 0.9663969278335571, "learning_rate": 0.0005, "loss": 1.9879752397537231, "step": 5593 }, { "epoch": 0.44099329917225066, "grad_norm": 0.7661881446838379, "learning_rate": 0.0005, "loss": 2.372985363006592, "step": 5594 }, { "epoch": 0.44107213243988963, "grad_norm": 0.6825140118598938, "learning_rate": 0.0005, "loss": 2.041546106338501, "step": 5595 }, { "epoch": 0.4411509657075286, "grad_norm": 0.696234941482544, "learning_rate": 0.0005, "loss": 2.1468300819396973, "step": 5596 }, { "epoch": 0.4412297989751675, "grad_norm": 0.9547156095504761, "learning_rate": 0.0005, "loss": 2.3360507488250732, "step": 5597 }, { "epoch": 0.4413086322428065, "grad_norm": 0.7488545775413513, "learning_rate": 0.0005, "loss": 2.266021728515625, "step": 5598 }, { "epoch": 0.4413874655104454, "grad_norm": 1.0693151950836182, "learning_rate": 0.0005, "loss": 2.1627235412597656, "step": 5599 }, { "epoch": 0.44146629877808435, "grad_norm": 0.9104386568069458, "learning_rate": 0.0005, "loss": 2.2459094524383545, "step": 5600 }, { "epoch": 0.4415451320457233, "grad_norm": 0.8074135184288025, "learning_rate": 0.0005, "loss": 1.7987070083618164, "step": 5601 }, { "epoch": 0.44162396531336223, "grad_norm": 0.7208414077758789, "learning_rate": 0.0005, "loss": 2.219008207321167, "step": 5602 }, { "epoch": 0.4417027985810012, "grad_norm": 0.6638484597206116, "learning_rate": 0.0005, "loss": 2.173856735229492, "step": 5603 }, { "epoch": 0.4417816318486401, "grad_norm": 0.7465078830718994, "learning_rate": 0.0005, "loss": 2.0206286907196045, "step": 5604 }, { "epoch": 0.4418604651162791, "grad_norm": 0.674835741519928, "learning_rate": 0.0005, "loss": 1.887779712677002, "step": 5605 }, { "epoch": 0.441939298383918, "grad_norm": 0.8218427896499634, "learning_rate": 0.0005, "loss": 2.1489486694335938, "step": 5606 }, { "epoch": 0.44201813165155696, "grad_norm": 0.9129899740219116, "learning_rate": 0.0005, "loss": 2.2872471809387207, "step": 5607 }, { "epoch": 0.4420969649191959, "grad_norm": 0.7483484148979187, "learning_rate": 0.0005, "loss": 2.1212573051452637, "step": 5608 }, { "epoch": 0.44217579818683483, "grad_norm": 0.7654059529304504, "learning_rate": 0.0005, "loss": 2.0983762741088867, "step": 5609 }, { "epoch": 0.4422546314544738, "grad_norm": 0.907893717288971, "learning_rate": 0.0005, "loss": 2.1669015884399414, "step": 5610 }, { "epoch": 0.4423334647221127, "grad_norm": 0.8976231217384338, "learning_rate": 0.0005, "loss": 1.9193236827850342, "step": 5611 }, { "epoch": 0.4424122979897517, "grad_norm": 0.6516030430793762, "learning_rate": 0.0005, "loss": 1.8700451850891113, "step": 5612 }, { "epoch": 0.44249113125739065, "grad_norm": 0.7700786590576172, "learning_rate": 0.0005, "loss": 2.4004392623901367, "step": 5613 }, { "epoch": 0.44256996452502956, "grad_norm": 0.7877932786941528, "learning_rate": 0.0005, "loss": 2.3315234184265137, "step": 5614 }, { "epoch": 0.4426487977926685, "grad_norm": 0.763690710067749, "learning_rate": 0.0005, "loss": 2.450719118118286, "step": 5615 }, { "epoch": 0.44272763106030744, "grad_norm": 0.9224606156349182, "learning_rate": 0.0005, "loss": 1.8861076831817627, "step": 5616 }, { "epoch": 0.4428064643279464, "grad_norm": 5.064005374908447, "learning_rate": 0.0005, "loss": 2.073432445526123, "step": 5617 }, { "epoch": 0.4428852975955853, "grad_norm": 0.82819664478302, "learning_rate": 0.0005, "loss": 2.2371060848236084, "step": 5618 }, { "epoch": 0.4429641308632243, "grad_norm": 0.9668128490447998, "learning_rate": 0.0005, "loss": 2.4195985794067383, "step": 5619 }, { "epoch": 0.44304296413086325, "grad_norm": 0.8683441877365112, "learning_rate": 0.0005, "loss": 2.1867756843566895, "step": 5620 }, { "epoch": 0.44312179739850216, "grad_norm": 0.6307071447372437, "learning_rate": 0.0005, "loss": 2.254807710647583, "step": 5621 }, { "epoch": 0.4432006306661411, "grad_norm": 0.7830531597137451, "learning_rate": 0.0005, "loss": 2.465895414352417, "step": 5622 }, { "epoch": 0.44327946393378004, "grad_norm": 1.1411930322647095, "learning_rate": 0.0005, "loss": 2.0272984504699707, "step": 5623 }, { "epoch": 0.443358297201419, "grad_norm": 0.7492795586585999, "learning_rate": 0.0005, "loss": 1.9015361070632935, "step": 5624 }, { "epoch": 0.4434371304690579, "grad_norm": 0.70859295129776, "learning_rate": 0.0005, "loss": 2.1261048316955566, "step": 5625 }, { "epoch": 0.4435159637366969, "grad_norm": 0.7487939596176147, "learning_rate": 0.0005, "loss": 2.3170862197875977, "step": 5626 }, { "epoch": 0.44359479700433585, "grad_norm": 1.0680711269378662, "learning_rate": 0.0005, "loss": 2.106323480606079, "step": 5627 }, { "epoch": 0.44367363027197476, "grad_norm": 0.7196305990219116, "learning_rate": 0.0005, "loss": 1.9936316013336182, "step": 5628 }, { "epoch": 0.44375246353961373, "grad_norm": 0.8767620325088501, "learning_rate": 0.0005, "loss": 2.101172924041748, "step": 5629 }, { "epoch": 0.44383129680725264, "grad_norm": 0.7393118143081665, "learning_rate": 0.0005, "loss": 2.274486780166626, "step": 5630 }, { "epoch": 0.4439101300748916, "grad_norm": 0.7440823316574097, "learning_rate": 0.0005, "loss": 2.224137306213379, "step": 5631 }, { "epoch": 0.4439889633425306, "grad_norm": 0.7877005934715271, "learning_rate": 0.0005, "loss": 2.057433605194092, "step": 5632 }, { "epoch": 0.4440677966101695, "grad_norm": 0.809868335723877, "learning_rate": 0.0005, "loss": 2.395019769668579, "step": 5633 }, { "epoch": 0.44414662987780845, "grad_norm": 0.8319374918937683, "learning_rate": 0.0005, "loss": 2.0681259632110596, "step": 5634 }, { "epoch": 0.44422546314544736, "grad_norm": 0.7507580518722534, "learning_rate": 0.0005, "loss": 2.054093360900879, "step": 5635 }, { "epoch": 0.44430429641308633, "grad_norm": 0.7295622825622559, "learning_rate": 0.0005, "loss": 2.081515312194824, "step": 5636 }, { "epoch": 0.44438312968072524, "grad_norm": 0.7206539511680603, "learning_rate": 0.0005, "loss": 2.105560779571533, "step": 5637 }, { "epoch": 0.4444619629483642, "grad_norm": 0.84336918592453, "learning_rate": 0.0005, "loss": 2.1763916015625, "step": 5638 }, { "epoch": 0.4445407962160032, "grad_norm": 0.7212508916854858, "learning_rate": 0.0005, "loss": 2.004335880279541, "step": 5639 }, { "epoch": 0.4446196294836421, "grad_norm": 0.7755146622657776, "learning_rate": 0.0005, "loss": 2.339212417602539, "step": 5640 }, { "epoch": 0.44469846275128105, "grad_norm": 0.7569099068641663, "learning_rate": 0.0005, "loss": 2.153040885925293, "step": 5641 }, { "epoch": 0.44477729601891997, "grad_norm": 0.6504946351051331, "learning_rate": 0.0005, "loss": 1.9309332370758057, "step": 5642 }, { "epoch": 0.44485612928655893, "grad_norm": 0.6371791362762451, "learning_rate": 0.0005, "loss": 1.8619518280029297, "step": 5643 }, { "epoch": 0.4449349625541979, "grad_norm": 0.6926081776618958, "learning_rate": 0.0005, "loss": 2.1902616024017334, "step": 5644 }, { "epoch": 0.4450137958218368, "grad_norm": 0.766113817691803, "learning_rate": 0.0005, "loss": 2.2581143379211426, "step": 5645 }, { "epoch": 0.4450926290894758, "grad_norm": 0.6927595138549805, "learning_rate": 0.0005, "loss": 2.256822347640991, "step": 5646 }, { "epoch": 0.4451714623571147, "grad_norm": 0.7588585019111633, "learning_rate": 0.0005, "loss": 2.2890920639038086, "step": 5647 }, { "epoch": 0.44525029562475366, "grad_norm": 0.9584424495697021, "learning_rate": 0.0005, "loss": 2.2206313610076904, "step": 5648 }, { "epoch": 0.44532912889239257, "grad_norm": 0.7721008062362671, "learning_rate": 0.0005, "loss": 2.0583977699279785, "step": 5649 }, { "epoch": 0.44540796216003153, "grad_norm": 1.199359655380249, "learning_rate": 0.0005, "loss": 2.125565767288208, "step": 5650 }, { "epoch": 0.4454867954276705, "grad_norm": 0.8860439658164978, "learning_rate": 0.0005, "loss": 2.065021514892578, "step": 5651 }, { "epoch": 0.4455656286953094, "grad_norm": 0.7325164079666138, "learning_rate": 0.0005, "loss": 2.1669859886169434, "step": 5652 }, { "epoch": 0.4456444619629484, "grad_norm": 0.8741803169250488, "learning_rate": 0.0005, "loss": 1.9844977855682373, "step": 5653 }, { "epoch": 0.4457232952305873, "grad_norm": 0.7405385971069336, "learning_rate": 0.0005, "loss": 2.0968103408813477, "step": 5654 }, { "epoch": 0.44580212849822626, "grad_norm": 0.7859671115875244, "learning_rate": 0.0005, "loss": 2.3611819744110107, "step": 5655 }, { "epoch": 0.44588096176586517, "grad_norm": 0.6983577609062195, "learning_rate": 0.0005, "loss": 2.3939311504364014, "step": 5656 }, { "epoch": 0.44595979503350414, "grad_norm": 0.817106306552887, "learning_rate": 0.0005, "loss": 2.203627109527588, "step": 5657 }, { "epoch": 0.4460386283011431, "grad_norm": 0.8407450318336487, "learning_rate": 0.0005, "loss": 2.057826042175293, "step": 5658 }, { "epoch": 0.446117461568782, "grad_norm": 0.7645785808563232, "learning_rate": 0.0005, "loss": 1.8769207000732422, "step": 5659 }, { "epoch": 0.446196294836421, "grad_norm": 0.7087584733963013, "learning_rate": 0.0005, "loss": 2.225477695465088, "step": 5660 }, { "epoch": 0.4462751281040599, "grad_norm": 0.8875786662101746, "learning_rate": 0.0005, "loss": 2.2389674186706543, "step": 5661 }, { "epoch": 0.44635396137169886, "grad_norm": 0.7128922343254089, "learning_rate": 0.0005, "loss": 2.244194984436035, "step": 5662 }, { "epoch": 0.4464327946393378, "grad_norm": 0.7964428663253784, "learning_rate": 0.0005, "loss": 2.0870156288146973, "step": 5663 }, { "epoch": 0.44651162790697674, "grad_norm": 0.7327782511711121, "learning_rate": 0.0005, "loss": 1.794633150100708, "step": 5664 }, { "epoch": 0.4465904611746157, "grad_norm": 0.767849862575531, "learning_rate": 0.0005, "loss": 2.4143404960632324, "step": 5665 }, { "epoch": 0.4466692944422546, "grad_norm": 0.8067911863327026, "learning_rate": 0.0005, "loss": 2.191042900085449, "step": 5666 }, { "epoch": 0.4467481277098936, "grad_norm": 0.7487733364105225, "learning_rate": 0.0005, "loss": 1.697016954421997, "step": 5667 }, { "epoch": 0.4468269609775325, "grad_norm": 0.6076229810714722, "learning_rate": 0.0005, "loss": 1.925220012664795, "step": 5668 }, { "epoch": 0.44690579424517146, "grad_norm": 0.6095839738845825, "learning_rate": 0.0005, "loss": 2.0775082111358643, "step": 5669 }, { "epoch": 0.44698462751281043, "grad_norm": 0.8160870671272278, "learning_rate": 0.0005, "loss": 1.9298224449157715, "step": 5670 }, { "epoch": 0.44706346078044934, "grad_norm": 0.7722778916358948, "learning_rate": 0.0005, "loss": 2.029832601547241, "step": 5671 }, { "epoch": 0.4471422940480883, "grad_norm": 0.759099543094635, "learning_rate": 0.0005, "loss": 2.068342924118042, "step": 5672 }, { "epoch": 0.4472211273157272, "grad_norm": 0.9010778665542603, "learning_rate": 0.0005, "loss": 1.9097861051559448, "step": 5673 }, { "epoch": 0.4472999605833662, "grad_norm": 0.7406754493713379, "learning_rate": 0.0005, "loss": 2.023120164871216, "step": 5674 }, { "epoch": 0.4473787938510051, "grad_norm": 0.9375815391540527, "learning_rate": 0.0005, "loss": 2.178311347961426, "step": 5675 }, { "epoch": 0.44745762711864406, "grad_norm": 0.7963021397590637, "learning_rate": 0.0005, "loss": 1.9796621799468994, "step": 5676 }, { "epoch": 0.44753646038628303, "grad_norm": 0.9040088653564453, "learning_rate": 0.0005, "loss": 2.1174559593200684, "step": 5677 }, { "epoch": 0.44761529365392194, "grad_norm": 0.7551902532577515, "learning_rate": 0.0005, "loss": 2.044107675552368, "step": 5678 }, { "epoch": 0.4476941269215609, "grad_norm": 0.6677297949790955, "learning_rate": 0.0005, "loss": 2.089486837387085, "step": 5679 }, { "epoch": 0.4477729601891998, "grad_norm": 0.775694727897644, "learning_rate": 0.0005, "loss": 1.8994134664535522, "step": 5680 }, { "epoch": 0.4478517934568388, "grad_norm": 0.7039572596549988, "learning_rate": 0.0005, "loss": 2.1594085693359375, "step": 5681 }, { "epoch": 0.44793062672447775, "grad_norm": 0.9197142720222473, "learning_rate": 0.0005, "loss": 2.0850861072540283, "step": 5682 }, { "epoch": 0.44800945999211667, "grad_norm": 0.6552700996398926, "learning_rate": 0.0005, "loss": 1.9568790197372437, "step": 5683 }, { "epoch": 0.44808829325975563, "grad_norm": 0.792214035987854, "learning_rate": 0.0005, "loss": 1.9183762073516846, "step": 5684 }, { "epoch": 0.44816712652739454, "grad_norm": 1.4446461200714111, "learning_rate": 0.0005, "loss": 1.9405419826507568, "step": 5685 }, { "epoch": 0.4482459597950335, "grad_norm": 0.648640513420105, "learning_rate": 0.0005, "loss": 2.0560214519500732, "step": 5686 }, { "epoch": 0.4483247930626724, "grad_norm": 0.7140499949455261, "learning_rate": 0.0005, "loss": 2.2941205501556396, "step": 5687 }, { "epoch": 0.4484036263303114, "grad_norm": 0.7223594188690186, "learning_rate": 0.0005, "loss": 2.1535611152648926, "step": 5688 }, { "epoch": 0.44848245959795036, "grad_norm": 0.7324228286743164, "learning_rate": 0.0005, "loss": 2.3798654079437256, "step": 5689 }, { "epoch": 0.44856129286558927, "grad_norm": 0.7305378913879395, "learning_rate": 0.0005, "loss": 2.127906084060669, "step": 5690 }, { "epoch": 0.44864012613322823, "grad_norm": 0.7421197295188904, "learning_rate": 0.0005, "loss": 1.9204179048538208, "step": 5691 }, { "epoch": 0.44871895940086715, "grad_norm": 0.6813128590583801, "learning_rate": 0.0005, "loss": 2.022848129272461, "step": 5692 }, { "epoch": 0.4487977926685061, "grad_norm": 0.9009110331535339, "learning_rate": 0.0005, "loss": 2.490774631500244, "step": 5693 }, { "epoch": 0.4488766259361451, "grad_norm": 0.9640554189682007, "learning_rate": 0.0005, "loss": 1.9951562881469727, "step": 5694 }, { "epoch": 0.448955459203784, "grad_norm": 0.934436023235321, "learning_rate": 0.0005, "loss": 2.010392665863037, "step": 5695 }, { "epoch": 0.44903429247142296, "grad_norm": 0.8219857811927795, "learning_rate": 0.0005, "loss": 2.175304651260376, "step": 5696 }, { "epoch": 0.44911312573906187, "grad_norm": 0.8675054907798767, "learning_rate": 0.0005, "loss": 2.1285204887390137, "step": 5697 }, { "epoch": 0.44919195900670084, "grad_norm": 0.7625871300697327, "learning_rate": 0.0005, "loss": 2.308279275894165, "step": 5698 }, { "epoch": 0.44927079227433975, "grad_norm": 1.2096763849258423, "learning_rate": 0.0005, "loss": 2.2796521186828613, "step": 5699 }, { "epoch": 0.4493496255419787, "grad_norm": 0.6564116477966309, "learning_rate": 0.0005, "loss": 2.201061725616455, "step": 5700 }, { "epoch": 0.4494284588096177, "grad_norm": 0.930730402469635, "learning_rate": 0.0005, "loss": 2.1075692176818848, "step": 5701 }, { "epoch": 0.4495072920772566, "grad_norm": 0.8230409026145935, "learning_rate": 0.0005, "loss": 2.2238173484802246, "step": 5702 }, { "epoch": 0.44958612534489556, "grad_norm": 1.0486148595809937, "learning_rate": 0.0005, "loss": 2.1434226036071777, "step": 5703 }, { "epoch": 0.44966495861253447, "grad_norm": 0.7574014663696289, "learning_rate": 0.0005, "loss": 2.0257344245910645, "step": 5704 }, { "epoch": 0.44974379188017344, "grad_norm": 0.9461992383003235, "learning_rate": 0.0005, "loss": 2.0586957931518555, "step": 5705 }, { "epoch": 0.44982262514781235, "grad_norm": 0.7451009750366211, "learning_rate": 0.0005, "loss": 2.186196804046631, "step": 5706 }, { "epoch": 0.4499014584154513, "grad_norm": 0.7795181274414062, "learning_rate": 0.0005, "loss": 1.9413602352142334, "step": 5707 }, { "epoch": 0.4499802916830903, "grad_norm": 0.7285429835319519, "learning_rate": 0.0005, "loss": 2.135132074356079, "step": 5708 }, { "epoch": 0.4500591249507292, "grad_norm": 0.7140355706214905, "learning_rate": 0.0005, "loss": 2.1061103343963623, "step": 5709 }, { "epoch": 0.45013795821836816, "grad_norm": 0.655647337436676, "learning_rate": 0.0005, "loss": 2.0977630615234375, "step": 5710 }, { "epoch": 0.4502167914860071, "grad_norm": 0.9120708703994751, "learning_rate": 0.0005, "loss": 2.361771583557129, "step": 5711 }, { "epoch": 0.45029562475364604, "grad_norm": 0.8962976336479187, "learning_rate": 0.0005, "loss": 2.3695778846740723, "step": 5712 }, { "epoch": 0.450374458021285, "grad_norm": 1.542189359664917, "learning_rate": 0.0005, "loss": 2.331773281097412, "step": 5713 }, { "epoch": 0.4504532912889239, "grad_norm": 0.7908520102500916, "learning_rate": 0.0005, "loss": 2.113774061203003, "step": 5714 }, { "epoch": 0.4505321245565629, "grad_norm": 0.7159728407859802, "learning_rate": 0.0005, "loss": 1.8972089290618896, "step": 5715 }, { "epoch": 0.4506109578242018, "grad_norm": 0.9671239256858826, "learning_rate": 0.0005, "loss": 2.020780563354492, "step": 5716 }, { "epoch": 0.45068979109184076, "grad_norm": 0.6599705219268799, "learning_rate": 0.0005, "loss": 2.0166962146759033, "step": 5717 }, { "epoch": 0.4507686243594797, "grad_norm": 0.7299333810806274, "learning_rate": 0.0005, "loss": 1.8536367416381836, "step": 5718 }, { "epoch": 0.45084745762711864, "grad_norm": 0.774875819683075, "learning_rate": 0.0005, "loss": 2.1147804260253906, "step": 5719 }, { "epoch": 0.4509262908947576, "grad_norm": 0.691982090473175, "learning_rate": 0.0005, "loss": 1.892152190208435, "step": 5720 }, { "epoch": 0.4510051241623965, "grad_norm": 0.8029156923294067, "learning_rate": 0.0005, "loss": 2.0485010147094727, "step": 5721 }, { "epoch": 0.4510839574300355, "grad_norm": 1.1560781002044678, "learning_rate": 0.0005, "loss": 2.0534214973449707, "step": 5722 }, { "epoch": 0.4511627906976744, "grad_norm": 0.9299892783164978, "learning_rate": 0.0005, "loss": 1.7774872779846191, "step": 5723 }, { "epoch": 0.45124162396531337, "grad_norm": 0.8686760067939758, "learning_rate": 0.0005, "loss": 2.3006017208099365, "step": 5724 }, { "epoch": 0.45132045723295233, "grad_norm": 0.8825153708457947, "learning_rate": 0.0005, "loss": 2.1994128227233887, "step": 5725 }, { "epoch": 0.45139929050059124, "grad_norm": 1.2891044616699219, "learning_rate": 0.0005, "loss": 2.1424357891082764, "step": 5726 }, { "epoch": 0.4514781237682302, "grad_norm": 0.7625544667243958, "learning_rate": 0.0005, "loss": 2.103541612625122, "step": 5727 }, { "epoch": 0.4515569570358691, "grad_norm": 0.7382210493087769, "learning_rate": 0.0005, "loss": 2.0926122665405273, "step": 5728 }, { "epoch": 0.4516357903035081, "grad_norm": 1.0005125999450684, "learning_rate": 0.0005, "loss": 2.2254035472869873, "step": 5729 }, { "epoch": 0.451714623571147, "grad_norm": 0.7814415693283081, "learning_rate": 0.0005, "loss": 1.812387228012085, "step": 5730 }, { "epoch": 0.45179345683878597, "grad_norm": 0.6747971177101135, "learning_rate": 0.0005, "loss": 2.304603099822998, "step": 5731 }, { "epoch": 0.45187229010642493, "grad_norm": 0.8670118451118469, "learning_rate": 0.0005, "loss": 2.310124635696411, "step": 5732 }, { "epoch": 0.45195112337406385, "grad_norm": 0.9692051410675049, "learning_rate": 0.0005, "loss": 2.4136524200439453, "step": 5733 }, { "epoch": 0.4520299566417028, "grad_norm": 0.9438790082931519, "learning_rate": 0.0005, "loss": 2.192368268966675, "step": 5734 }, { "epoch": 0.4521087899093417, "grad_norm": 0.7472353577613831, "learning_rate": 0.0005, "loss": 2.4194068908691406, "step": 5735 }, { "epoch": 0.4521876231769807, "grad_norm": 0.6499298214912415, "learning_rate": 0.0005, "loss": 2.046205759048462, "step": 5736 }, { "epoch": 0.4522664564446196, "grad_norm": 0.9652872681617737, "learning_rate": 0.0005, "loss": 2.1320412158966064, "step": 5737 }, { "epoch": 0.45234528971225857, "grad_norm": 0.9335271120071411, "learning_rate": 0.0005, "loss": 2.209261894226074, "step": 5738 }, { "epoch": 0.45242412297989754, "grad_norm": 0.6398860812187195, "learning_rate": 0.0005, "loss": 2.1402790546417236, "step": 5739 }, { "epoch": 0.45250295624753645, "grad_norm": 0.8786141872406006, "learning_rate": 0.0005, "loss": 2.0820400714874268, "step": 5740 }, { "epoch": 0.4525817895151754, "grad_norm": 0.7735904455184937, "learning_rate": 0.0005, "loss": 2.0804967880249023, "step": 5741 }, { "epoch": 0.4526606227828143, "grad_norm": 0.7430741190910339, "learning_rate": 0.0005, "loss": 2.0472798347473145, "step": 5742 }, { "epoch": 0.4527394560504533, "grad_norm": 0.7954118847846985, "learning_rate": 0.0005, "loss": 2.2632761001586914, "step": 5743 }, { "epoch": 0.45281828931809226, "grad_norm": 0.7163712382316589, "learning_rate": 0.0005, "loss": 1.8169077634811401, "step": 5744 }, { "epoch": 0.45289712258573117, "grad_norm": 0.6605684757232666, "learning_rate": 0.0005, "loss": 2.1768264770507812, "step": 5745 }, { "epoch": 0.45297595585337014, "grad_norm": 0.8334497809410095, "learning_rate": 0.0005, "loss": 2.3376715183258057, "step": 5746 }, { "epoch": 0.45305478912100905, "grad_norm": 0.6526357531547546, "learning_rate": 0.0005, "loss": 2.2176332473754883, "step": 5747 }, { "epoch": 0.453133622388648, "grad_norm": 0.9771761894226074, "learning_rate": 0.0005, "loss": 2.2377278804779053, "step": 5748 }, { "epoch": 0.4532124556562869, "grad_norm": 0.7932350039482117, "learning_rate": 0.0005, "loss": 1.9997354745864868, "step": 5749 }, { "epoch": 0.4532912889239259, "grad_norm": 0.8370984792709351, "learning_rate": 0.0005, "loss": 2.2197978496551514, "step": 5750 }, { "epoch": 0.45337012219156486, "grad_norm": 0.7913294434547424, "learning_rate": 0.0005, "loss": 2.2430813312530518, "step": 5751 }, { "epoch": 0.4534489554592038, "grad_norm": 0.977168619632721, "learning_rate": 0.0005, "loss": 2.204482316970825, "step": 5752 }, { "epoch": 0.45352778872684274, "grad_norm": 0.7374755144119263, "learning_rate": 0.0005, "loss": 2.028590202331543, "step": 5753 }, { "epoch": 0.45360662199448165, "grad_norm": 0.8494960069656372, "learning_rate": 0.0005, "loss": 1.905666708946228, "step": 5754 }, { "epoch": 0.4536854552621206, "grad_norm": 0.7678709030151367, "learning_rate": 0.0005, "loss": 1.8289023637771606, "step": 5755 }, { "epoch": 0.4537642885297596, "grad_norm": 0.6754419207572937, "learning_rate": 0.0005, "loss": 2.0365262031555176, "step": 5756 }, { "epoch": 0.4538431217973985, "grad_norm": 0.6829531788825989, "learning_rate": 0.0005, "loss": 2.129075050354004, "step": 5757 }, { "epoch": 0.45392195506503746, "grad_norm": 1.0692262649536133, "learning_rate": 0.0005, "loss": 2.235244035720825, "step": 5758 }, { "epoch": 0.4540007883326764, "grad_norm": 0.5960254669189453, "learning_rate": 0.0005, "loss": 1.9216340780258179, "step": 5759 }, { "epoch": 0.45407962160031534, "grad_norm": 0.9034645557403564, "learning_rate": 0.0005, "loss": 2.238182544708252, "step": 5760 }, { "epoch": 0.45415845486795425, "grad_norm": 0.6802783012390137, "learning_rate": 0.0005, "loss": 1.987762212753296, "step": 5761 }, { "epoch": 0.4542372881355932, "grad_norm": 0.690782368183136, "learning_rate": 0.0005, "loss": 2.2559971809387207, "step": 5762 }, { "epoch": 0.4543161214032322, "grad_norm": 0.728869616985321, "learning_rate": 0.0005, "loss": 2.0317866802215576, "step": 5763 }, { "epoch": 0.4543949546708711, "grad_norm": 0.7780988812446594, "learning_rate": 0.0005, "loss": 2.1822447776794434, "step": 5764 }, { "epoch": 0.45447378793851007, "grad_norm": 0.6582409739494324, "learning_rate": 0.0005, "loss": 1.9559543132781982, "step": 5765 }, { "epoch": 0.454552621206149, "grad_norm": 1.3373267650604248, "learning_rate": 0.0005, "loss": 2.0773584842681885, "step": 5766 }, { "epoch": 0.45463145447378794, "grad_norm": 0.7085114121437073, "learning_rate": 0.0005, "loss": 2.3420662879943848, "step": 5767 }, { "epoch": 0.45471028774142686, "grad_norm": 0.8543407320976257, "learning_rate": 0.0005, "loss": 2.6271719932556152, "step": 5768 }, { "epoch": 0.4547891210090658, "grad_norm": 0.9548184871673584, "learning_rate": 0.0005, "loss": 2.41644287109375, "step": 5769 }, { "epoch": 0.4548679542767048, "grad_norm": 0.9458824992179871, "learning_rate": 0.0005, "loss": 2.2594590187072754, "step": 5770 }, { "epoch": 0.4549467875443437, "grad_norm": 1.065619945526123, "learning_rate": 0.0005, "loss": 2.225390672683716, "step": 5771 }, { "epoch": 0.45502562081198267, "grad_norm": 0.7048435807228088, "learning_rate": 0.0005, "loss": 2.070652961730957, "step": 5772 }, { "epoch": 0.4551044540796216, "grad_norm": 0.8147085905075073, "learning_rate": 0.0005, "loss": 2.2900328636169434, "step": 5773 }, { "epoch": 0.45518328734726055, "grad_norm": 0.8475753664970398, "learning_rate": 0.0005, "loss": 2.0894646644592285, "step": 5774 }, { "epoch": 0.4552621206148995, "grad_norm": 0.7392250895500183, "learning_rate": 0.0005, "loss": 2.1472325325012207, "step": 5775 }, { "epoch": 0.4553409538825384, "grad_norm": 0.8924118876457214, "learning_rate": 0.0005, "loss": 1.9168223142623901, "step": 5776 }, { "epoch": 0.4554197871501774, "grad_norm": 0.586413562297821, "learning_rate": 0.0005, "loss": 2.133216619491577, "step": 5777 }, { "epoch": 0.4554986204178163, "grad_norm": 0.75078284740448, "learning_rate": 0.0005, "loss": 2.076857805252075, "step": 5778 }, { "epoch": 0.45557745368545527, "grad_norm": 0.7990244626998901, "learning_rate": 0.0005, "loss": 2.263437032699585, "step": 5779 }, { "epoch": 0.4556562869530942, "grad_norm": 0.7172836065292358, "learning_rate": 0.0005, "loss": 1.9563286304473877, "step": 5780 }, { "epoch": 0.45573512022073315, "grad_norm": 0.8419123291969299, "learning_rate": 0.0005, "loss": 2.518533706665039, "step": 5781 }, { "epoch": 0.4558139534883721, "grad_norm": 0.6662079095840454, "learning_rate": 0.0005, "loss": 1.7365260124206543, "step": 5782 }, { "epoch": 0.455892786756011, "grad_norm": 0.7403051257133484, "learning_rate": 0.0005, "loss": 2.12127685546875, "step": 5783 }, { "epoch": 0.45597162002365, "grad_norm": 0.6479083895683289, "learning_rate": 0.0005, "loss": 1.9886116981506348, "step": 5784 }, { "epoch": 0.4560504532912889, "grad_norm": 1.319411039352417, "learning_rate": 0.0005, "loss": 1.8184545040130615, "step": 5785 }, { "epoch": 0.45612928655892787, "grad_norm": 0.7755622863769531, "learning_rate": 0.0005, "loss": 2.1356418132781982, "step": 5786 }, { "epoch": 0.45620811982656684, "grad_norm": 0.7718808650970459, "learning_rate": 0.0005, "loss": 1.868659496307373, "step": 5787 }, { "epoch": 0.45628695309420575, "grad_norm": 0.6678516864776611, "learning_rate": 0.0005, "loss": 2.2182250022888184, "step": 5788 }, { "epoch": 0.4563657863618447, "grad_norm": 1.0167508125305176, "learning_rate": 0.0005, "loss": 2.61399507522583, "step": 5789 }, { "epoch": 0.45644461962948363, "grad_norm": 0.9269980192184448, "learning_rate": 0.0005, "loss": 2.139021635055542, "step": 5790 }, { "epoch": 0.4565234528971226, "grad_norm": 0.7371542453765869, "learning_rate": 0.0005, "loss": 2.1133594512939453, "step": 5791 }, { "epoch": 0.4566022861647615, "grad_norm": 1.0369173288345337, "learning_rate": 0.0005, "loss": 2.118968963623047, "step": 5792 }, { "epoch": 0.4566811194324005, "grad_norm": 0.8069349527359009, "learning_rate": 0.0005, "loss": 2.056861162185669, "step": 5793 }, { "epoch": 0.45675995270003944, "grad_norm": 0.667759895324707, "learning_rate": 0.0005, "loss": 1.9914648532867432, "step": 5794 }, { "epoch": 0.45683878596767835, "grad_norm": 0.8201262950897217, "learning_rate": 0.0005, "loss": 2.1097397804260254, "step": 5795 }, { "epoch": 0.4569176192353173, "grad_norm": 0.8494722247123718, "learning_rate": 0.0005, "loss": 2.0134525299072266, "step": 5796 }, { "epoch": 0.45699645250295623, "grad_norm": 1.2404335737228394, "learning_rate": 0.0005, "loss": 2.3105320930480957, "step": 5797 }, { "epoch": 0.4570752857705952, "grad_norm": 0.7034474015235901, "learning_rate": 0.0005, "loss": 2.072373390197754, "step": 5798 }, { "epoch": 0.4571541190382341, "grad_norm": 1.1940382719039917, "learning_rate": 0.0005, "loss": 1.913881540298462, "step": 5799 }, { "epoch": 0.4572329523058731, "grad_norm": 0.8164752125740051, "learning_rate": 0.0005, "loss": 1.9746724367141724, "step": 5800 }, { "epoch": 0.45731178557351204, "grad_norm": 0.7563211917877197, "learning_rate": 0.0005, "loss": 1.8611204624176025, "step": 5801 }, { "epoch": 0.45739061884115095, "grad_norm": 0.8618016242980957, "learning_rate": 0.0005, "loss": 2.147503137588501, "step": 5802 }, { "epoch": 0.4574694521087899, "grad_norm": 0.7644109129905701, "learning_rate": 0.0005, "loss": 1.9857325553894043, "step": 5803 }, { "epoch": 0.45754828537642883, "grad_norm": 0.9070976972579956, "learning_rate": 0.0005, "loss": 2.1851935386657715, "step": 5804 }, { "epoch": 0.4576271186440678, "grad_norm": 0.8115745186805725, "learning_rate": 0.0005, "loss": 2.061638355255127, "step": 5805 }, { "epoch": 0.45770595191170677, "grad_norm": 0.8721157908439636, "learning_rate": 0.0005, "loss": 2.186826229095459, "step": 5806 }, { "epoch": 0.4577847851793457, "grad_norm": 0.6884402632713318, "learning_rate": 0.0005, "loss": 2.2046899795532227, "step": 5807 }, { "epoch": 0.45786361844698464, "grad_norm": 0.8292841911315918, "learning_rate": 0.0005, "loss": 1.9527943134307861, "step": 5808 }, { "epoch": 0.45794245171462356, "grad_norm": 0.7331587672233582, "learning_rate": 0.0005, "loss": 2.1221375465393066, "step": 5809 }, { "epoch": 0.4580212849822625, "grad_norm": 0.8305934071540833, "learning_rate": 0.0005, "loss": 2.2574706077575684, "step": 5810 }, { "epoch": 0.45810011824990143, "grad_norm": 1.0523425340652466, "learning_rate": 0.0005, "loss": 2.0036239624023438, "step": 5811 }, { "epoch": 0.4581789515175404, "grad_norm": 1.0577415227890015, "learning_rate": 0.0005, "loss": 2.075971841812134, "step": 5812 }, { "epoch": 0.45825778478517937, "grad_norm": 0.589751660823822, "learning_rate": 0.0005, "loss": 1.7325763702392578, "step": 5813 }, { "epoch": 0.4583366180528183, "grad_norm": 0.7092217206954956, "learning_rate": 0.0005, "loss": 2.3439249992370605, "step": 5814 }, { "epoch": 0.45841545132045725, "grad_norm": 1.058251142501831, "learning_rate": 0.0005, "loss": 2.2172293663024902, "step": 5815 }, { "epoch": 0.45849428458809616, "grad_norm": 0.6745464205741882, "learning_rate": 0.0005, "loss": 1.9367127418518066, "step": 5816 }, { "epoch": 0.4585731178557351, "grad_norm": 0.8575301170349121, "learning_rate": 0.0005, "loss": 2.1900858879089355, "step": 5817 }, { "epoch": 0.4586519511233741, "grad_norm": 0.701921820640564, "learning_rate": 0.0005, "loss": 1.840636134147644, "step": 5818 }, { "epoch": 0.458730784391013, "grad_norm": 0.8672274351119995, "learning_rate": 0.0005, "loss": 2.1502134799957275, "step": 5819 }, { "epoch": 0.45880961765865197, "grad_norm": 1.0765336751937866, "learning_rate": 0.0005, "loss": 2.0908737182617188, "step": 5820 }, { "epoch": 0.4588884509262909, "grad_norm": 0.9699724316596985, "learning_rate": 0.0005, "loss": 2.405067205429077, "step": 5821 }, { "epoch": 0.45896728419392985, "grad_norm": 0.7212780714035034, "learning_rate": 0.0005, "loss": 1.9609291553497314, "step": 5822 }, { "epoch": 0.45904611746156876, "grad_norm": 0.7757715582847595, "learning_rate": 0.0005, "loss": 1.9210550785064697, "step": 5823 }, { "epoch": 0.4591249507292077, "grad_norm": 0.8996360898017883, "learning_rate": 0.0005, "loss": 2.078705072402954, "step": 5824 }, { "epoch": 0.4592037839968467, "grad_norm": 0.6509512662887573, "learning_rate": 0.0005, "loss": 1.928419828414917, "step": 5825 }, { "epoch": 0.4592826172644856, "grad_norm": 0.9864317774772644, "learning_rate": 0.0005, "loss": 2.0394394397735596, "step": 5826 }, { "epoch": 0.45936145053212457, "grad_norm": 1.287353277206421, "learning_rate": 0.0005, "loss": 2.1751368045806885, "step": 5827 }, { "epoch": 0.4594402837997635, "grad_norm": 0.8113353252410889, "learning_rate": 0.0005, "loss": 2.2223525047302246, "step": 5828 }, { "epoch": 0.45951911706740245, "grad_norm": 0.8726862668991089, "learning_rate": 0.0005, "loss": 2.137907028198242, "step": 5829 }, { "epoch": 0.45959795033504136, "grad_norm": 0.7005771994590759, "learning_rate": 0.0005, "loss": 2.076425790786743, "step": 5830 }, { "epoch": 0.45967678360268033, "grad_norm": 0.7546130418777466, "learning_rate": 0.0005, "loss": 2.056842565536499, "step": 5831 }, { "epoch": 0.4597556168703193, "grad_norm": 0.7464792132377625, "learning_rate": 0.0005, "loss": 2.185199737548828, "step": 5832 }, { "epoch": 0.4598344501379582, "grad_norm": 0.9886377453804016, "learning_rate": 0.0005, "loss": 2.099156379699707, "step": 5833 }, { "epoch": 0.4599132834055972, "grad_norm": 0.6780949831008911, "learning_rate": 0.0005, "loss": 2.10798716545105, "step": 5834 }, { "epoch": 0.4599921166732361, "grad_norm": 1.0582985877990723, "learning_rate": 0.0005, "loss": 1.9962095022201538, "step": 5835 }, { "epoch": 0.46007094994087505, "grad_norm": 0.7307746410369873, "learning_rate": 0.0005, "loss": 2.0079164505004883, "step": 5836 }, { "epoch": 0.460149783208514, "grad_norm": 0.6667842268943787, "learning_rate": 0.0005, "loss": 1.9286181926727295, "step": 5837 }, { "epoch": 0.46022861647615293, "grad_norm": 0.906774640083313, "learning_rate": 0.0005, "loss": 2.082336902618408, "step": 5838 }, { "epoch": 0.4603074497437919, "grad_norm": 0.9732469320297241, "learning_rate": 0.0005, "loss": 2.4016530513763428, "step": 5839 }, { "epoch": 0.4603862830114308, "grad_norm": 0.7219259738922119, "learning_rate": 0.0005, "loss": 2.092465400695801, "step": 5840 }, { "epoch": 0.4604651162790698, "grad_norm": 0.940788745880127, "learning_rate": 0.0005, "loss": 2.4286670684814453, "step": 5841 }, { "epoch": 0.4605439495467087, "grad_norm": 0.818163275718689, "learning_rate": 0.0005, "loss": 1.9393094778060913, "step": 5842 }, { "epoch": 0.46062278281434765, "grad_norm": 0.8046512603759766, "learning_rate": 0.0005, "loss": 2.0676984786987305, "step": 5843 }, { "epoch": 0.4607016160819866, "grad_norm": 0.6661481261253357, "learning_rate": 0.0005, "loss": 2.09584903717041, "step": 5844 }, { "epoch": 0.46078044934962553, "grad_norm": 0.7771867513656616, "learning_rate": 0.0005, "loss": 2.1695942878723145, "step": 5845 }, { "epoch": 0.4608592826172645, "grad_norm": 0.8304514288902283, "learning_rate": 0.0005, "loss": 2.1149802207946777, "step": 5846 }, { "epoch": 0.4609381158849034, "grad_norm": 0.7161229848861694, "learning_rate": 0.0005, "loss": 2.1540980339050293, "step": 5847 }, { "epoch": 0.4610169491525424, "grad_norm": 0.6907753348350525, "learning_rate": 0.0005, "loss": 2.112309217453003, "step": 5848 }, { "epoch": 0.46109578242018134, "grad_norm": 0.7825407981872559, "learning_rate": 0.0005, "loss": 2.2199997901916504, "step": 5849 }, { "epoch": 0.46117461568782026, "grad_norm": 0.6832786798477173, "learning_rate": 0.0005, "loss": 1.9673210382461548, "step": 5850 }, { "epoch": 0.4612534489554592, "grad_norm": 0.6561490297317505, "learning_rate": 0.0005, "loss": 2.215484380722046, "step": 5851 }, { "epoch": 0.46133228222309813, "grad_norm": 0.7407476305961609, "learning_rate": 0.0005, "loss": 2.057600975036621, "step": 5852 }, { "epoch": 0.4614111154907371, "grad_norm": 0.7889396548271179, "learning_rate": 0.0005, "loss": 2.2073752880096436, "step": 5853 }, { "epoch": 0.461489948758376, "grad_norm": 0.8015267848968506, "learning_rate": 0.0005, "loss": 2.243460178375244, "step": 5854 }, { "epoch": 0.461568782026015, "grad_norm": 0.9507186412811279, "learning_rate": 0.0005, "loss": 2.2699975967407227, "step": 5855 }, { "epoch": 0.46164761529365395, "grad_norm": 0.823050320148468, "learning_rate": 0.0005, "loss": 1.6011583805084229, "step": 5856 }, { "epoch": 0.46172644856129286, "grad_norm": 0.8598644733428955, "learning_rate": 0.0005, "loss": 1.924119472503662, "step": 5857 }, { "epoch": 0.4618052818289318, "grad_norm": 0.6945767998695374, "learning_rate": 0.0005, "loss": 2.2707653045654297, "step": 5858 }, { "epoch": 0.46188411509657074, "grad_norm": 1.1709736585617065, "learning_rate": 0.0005, "loss": 2.08713960647583, "step": 5859 }, { "epoch": 0.4619629483642097, "grad_norm": 0.8520591259002686, "learning_rate": 0.0005, "loss": 2.169219493865967, "step": 5860 }, { "epoch": 0.4620417816318486, "grad_norm": 0.982821524143219, "learning_rate": 0.0005, "loss": 1.8586149215698242, "step": 5861 }, { "epoch": 0.4621206148994876, "grad_norm": 1.3172030448913574, "learning_rate": 0.0005, "loss": 1.9352445602416992, "step": 5862 }, { "epoch": 0.46219944816712655, "grad_norm": 0.8199349641799927, "learning_rate": 0.0005, "loss": 2.1684532165527344, "step": 5863 }, { "epoch": 0.46227828143476546, "grad_norm": 0.8080886006355286, "learning_rate": 0.0005, "loss": 2.079651355743408, "step": 5864 }, { "epoch": 0.4623571147024044, "grad_norm": 1.1800543069839478, "learning_rate": 0.0005, "loss": 2.545780658721924, "step": 5865 }, { "epoch": 0.46243594797004334, "grad_norm": 0.8991209268569946, "learning_rate": 0.0005, "loss": 2.022726058959961, "step": 5866 }, { "epoch": 0.4625147812376823, "grad_norm": 0.6098579168319702, "learning_rate": 0.0005, "loss": 1.9077861309051514, "step": 5867 }, { "epoch": 0.46259361450532127, "grad_norm": 0.8513306975364685, "learning_rate": 0.0005, "loss": 2.0532760620117188, "step": 5868 }, { "epoch": 0.4626724477729602, "grad_norm": 0.7823258638381958, "learning_rate": 0.0005, "loss": 1.8653607368469238, "step": 5869 }, { "epoch": 0.46275128104059915, "grad_norm": 0.7473210096359253, "learning_rate": 0.0005, "loss": 2.4248905181884766, "step": 5870 }, { "epoch": 0.46283011430823806, "grad_norm": 0.7873032093048096, "learning_rate": 0.0005, "loss": 2.1179585456848145, "step": 5871 }, { "epoch": 0.46290894757587703, "grad_norm": 0.9147257804870605, "learning_rate": 0.0005, "loss": 2.160668134689331, "step": 5872 }, { "epoch": 0.46298778084351594, "grad_norm": 0.7420557737350464, "learning_rate": 0.0005, "loss": 2.034719944000244, "step": 5873 }, { "epoch": 0.4630666141111549, "grad_norm": 0.7292804718017578, "learning_rate": 0.0005, "loss": 2.2085154056549072, "step": 5874 }, { "epoch": 0.4631454473787939, "grad_norm": 0.7597225308418274, "learning_rate": 0.0005, "loss": 2.1737914085388184, "step": 5875 }, { "epoch": 0.4632242806464328, "grad_norm": 0.7488322854042053, "learning_rate": 0.0005, "loss": 2.2442359924316406, "step": 5876 }, { "epoch": 0.46330311391407175, "grad_norm": 0.754414975643158, "learning_rate": 0.0005, "loss": 1.9299668073654175, "step": 5877 }, { "epoch": 0.46338194718171066, "grad_norm": 1.043850302696228, "learning_rate": 0.0005, "loss": 1.9987409114837646, "step": 5878 }, { "epoch": 0.46346078044934963, "grad_norm": 0.8967310786247253, "learning_rate": 0.0005, "loss": 1.9906277656555176, "step": 5879 }, { "epoch": 0.4635396137169886, "grad_norm": 0.6660097241401672, "learning_rate": 0.0005, "loss": 2.1684107780456543, "step": 5880 }, { "epoch": 0.4636184469846275, "grad_norm": 0.8069327473640442, "learning_rate": 0.0005, "loss": 1.9566209316253662, "step": 5881 }, { "epoch": 0.4636972802522665, "grad_norm": 0.8971702456474304, "learning_rate": 0.0005, "loss": 2.0253982543945312, "step": 5882 }, { "epoch": 0.4637761135199054, "grad_norm": 0.9751272797584534, "learning_rate": 0.0005, "loss": 1.9889355897903442, "step": 5883 }, { "epoch": 0.46385494678754435, "grad_norm": 1.0212998390197754, "learning_rate": 0.0005, "loss": 2.2140133380889893, "step": 5884 }, { "epoch": 0.46393378005518326, "grad_norm": 0.661095917224884, "learning_rate": 0.0005, "loss": 1.9852360486984253, "step": 5885 }, { "epoch": 0.46401261332282223, "grad_norm": 0.6314801573753357, "learning_rate": 0.0005, "loss": 1.8410369157791138, "step": 5886 }, { "epoch": 0.4640914465904612, "grad_norm": 0.7466867566108704, "learning_rate": 0.0005, "loss": 2.217766046524048, "step": 5887 }, { "epoch": 0.4641702798581001, "grad_norm": 0.6961286067962646, "learning_rate": 0.0005, "loss": 2.1557559967041016, "step": 5888 }, { "epoch": 0.4642491131257391, "grad_norm": 0.8693527579307556, "learning_rate": 0.0005, "loss": 2.022597074508667, "step": 5889 }, { "epoch": 0.464327946393378, "grad_norm": 0.958732545375824, "learning_rate": 0.0005, "loss": 2.0628089904785156, "step": 5890 }, { "epoch": 0.46440677966101696, "grad_norm": 0.8463954329490662, "learning_rate": 0.0005, "loss": 2.0258851051330566, "step": 5891 }, { "epoch": 0.46448561292865587, "grad_norm": 0.8363407850265503, "learning_rate": 0.0005, "loss": 2.292872905731201, "step": 5892 }, { "epoch": 0.46456444619629483, "grad_norm": 1.0383059978485107, "learning_rate": 0.0005, "loss": 2.208601713180542, "step": 5893 }, { "epoch": 0.4646432794639338, "grad_norm": 0.6933984756469727, "learning_rate": 0.0005, "loss": 1.9637668132781982, "step": 5894 }, { "epoch": 0.4647221127315727, "grad_norm": 0.6968057155609131, "learning_rate": 0.0005, "loss": 2.1707606315612793, "step": 5895 }, { "epoch": 0.4648009459992117, "grad_norm": 0.9692963361740112, "learning_rate": 0.0005, "loss": 2.2014999389648438, "step": 5896 }, { "epoch": 0.4648797792668506, "grad_norm": 1.105703353881836, "learning_rate": 0.0005, "loss": 1.8514435291290283, "step": 5897 }, { "epoch": 0.46495861253448956, "grad_norm": 1.001900553703308, "learning_rate": 0.0005, "loss": 2.1731443405151367, "step": 5898 }, { "epoch": 0.4650374458021285, "grad_norm": 0.8561297655105591, "learning_rate": 0.0005, "loss": 2.416727066040039, "step": 5899 }, { "epoch": 0.46511627906976744, "grad_norm": 0.6097352504730225, "learning_rate": 0.0005, "loss": 1.8828133344650269, "step": 5900 }, { "epoch": 0.4651951123374064, "grad_norm": 0.936747133731842, "learning_rate": 0.0005, "loss": 1.9668887853622437, "step": 5901 }, { "epoch": 0.4652739456050453, "grad_norm": 0.7223634123802185, "learning_rate": 0.0005, "loss": 1.9096705913543701, "step": 5902 }, { "epoch": 0.4653527788726843, "grad_norm": 0.6152540445327759, "learning_rate": 0.0005, "loss": 2.001601219177246, "step": 5903 }, { "epoch": 0.4654316121403232, "grad_norm": 0.7722352147102356, "learning_rate": 0.0005, "loss": 1.6985280513763428, "step": 5904 }, { "epoch": 0.46551044540796216, "grad_norm": 0.7071764469146729, "learning_rate": 0.0005, "loss": 2.3857967853546143, "step": 5905 }, { "epoch": 0.4655892786756011, "grad_norm": 0.7271634340286255, "learning_rate": 0.0005, "loss": 2.0651984214782715, "step": 5906 }, { "epoch": 0.46566811194324004, "grad_norm": 0.8535984754562378, "learning_rate": 0.0005, "loss": 1.9224591255187988, "step": 5907 }, { "epoch": 0.465746945210879, "grad_norm": 0.8147041201591492, "learning_rate": 0.0005, "loss": 2.17134165763855, "step": 5908 }, { "epoch": 0.4658257784785179, "grad_norm": 0.8126394748687744, "learning_rate": 0.0005, "loss": 2.1160073280334473, "step": 5909 }, { "epoch": 0.4659046117461569, "grad_norm": 0.6014012694358826, "learning_rate": 0.0005, "loss": 1.884669542312622, "step": 5910 }, { "epoch": 0.46598344501379585, "grad_norm": 0.9639891982078552, "learning_rate": 0.0005, "loss": 2.356801986694336, "step": 5911 }, { "epoch": 0.46606227828143476, "grad_norm": 0.807908296585083, "learning_rate": 0.0005, "loss": 1.8371267318725586, "step": 5912 }, { "epoch": 0.46614111154907373, "grad_norm": 0.737722635269165, "learning_rate": 0.0005, "loss": 1.8491458892822266, "step": 5913 }, { "epoch": 0.46621994481671264, "grad_norm": 0.7342734336853027, "learning_rate": 0.0005, "loss": 1.894320011138916, "step": 5914 }, { "epoch": 0.4662987780843516, "grad_norm": 0.6921334266662598, "learning_rate": 0.0005, "loss": 2.101994514465332, "step": 5915 }, { "epoch": 0.4663776113519905, "grad_norm": 0.8330889344215393, "learning_rate": 0.0005, "loss": 2.058663845062256, "step": 5916 }, { "epoch": 0.4664564446196295, "grad_norm": 0.8769908547401428, "learning_rate": 0.0005, "loss": 2.1235876083374023, "step": 5917 }, { "epoch": 0.46653527788726845, "grad_norm": 0.8533422946929932, "learning_rate": 0.0005, "loss": 2.1197915077209473, "step": 5918 }, { "epoch": 0.46661411115490736, "grad_norm": 0.7560228109359741, "learning_rate": 0.0005, "loss": 2.1616599559783936, "step": 5919 }, { "epoch": 0.46669294442254633, "grad_norm": 0.7251061797142029, "learning_rate": 0.0005, "loss": 2.134211778640747, "step": 5920 }, { "epoch": 0.46677177769018524, "grad_norm": 0.9102327227592468, "learning_rate": 0.0005, "loss": 2.1704230308532715, "step": 5921 }, { "epoch": 0.4668506109578242, "grad_norm": 0.8915872573852539, "learning_rate": 0.0005, "loss": 2.1349399089813232, "step": 5922 }, { "epoch": 0.4669294442254631, "grad_norm": 0.8272286653518677, "learning_rate": 0.0005, "loss": 1.998133659362793, "step": 5923 }, { "epoch": 0.4670082774931021, "grad_norm": 0.7043827772140503, "learning_rate": 0.0005, "loss": 2.177635669708252, "step": 5924 }, { "epoch": 0.46708711076074105, "grad_norm": 0.7344229817390442, "learning_rate": 0.0005, "loss": 1.983656644821167, "step": 5925 }, { "epoch": 0.46716594402837996, "grad_norm": 0.6967781782150269, "learning_rate": 0.0005, "loss": 2.316621780395508, "step": 5926 }, { "epoch": 0.46724477729601893, "grad_norm": 0.949730396270752, "learning_rate": 0.0005, "loss": 2.068556785583496, "step": 5927 }, { "epoch": 0.46732361056365784, "grad_norm": 0.748136043548584, "learning_rate": 0.0005, "loss": 2.3264119625091553, "step": 5928 }, { "epoch": 0.4674024438312968, "grad_norm": 0.6751865744590759, "learning_rate": 0.0005, "loss": 2.418914318084717, "step": 5929 }, { "epoch": 0.4674812770989358, "grad_norm": 0.6191279888153076, "learning_rate": 0.0005, "loss": 2.1128859519958496, "step": 5930 }, { "epoch": 0.4675601103665747, "grad_norm": 0.9621669054031372, "learning_rate": 0.0005, "loss": 2.454404354095459, "step": 5931 }, { "epoch": 0.46763894363421366, "grad_norm": 0.924508273601532, "learning_rate": 0.0005, "loss": 2.505336284637451, "step": 5932 }, { "epoch": 0.46771777690185257, "grad_norm": 0.6904059052467346, "learning_rate": 0.0005, "loss": 2.1350815296173096, "step": 5933 }, { "epoch": 0.46779661016949153, "grad_norm": 0.7598994970321655, "learning_rate": 0.0005, "loss": 2.3013577461242676, "step": 5934 }, { "epoch": 0.46787544343713044, "grad_norm": 0.8208487629890442, "learning_rate": 0.0005, "loss": 1.877061128616333, "step": 5935 }, { "epoch": 0.4679542767047694, "grad_norm": 0.8427968621253967, "learning_rate": 0.0005, "loss": 1.913210153579712, "step": 5936 }, { "epoch": 0.4680331099724084, "grad_norm": 0.8956680297851562, "learning_rate": 0.0005, "loss": 2.065077781677246, "step": 5937 }, { "epoch": 0.4681119432400473, "grad_norm": 0.7368064522743225, "learning_rate": 0.0005, "loss": 2.2556726932525635, "step": 5938 }, { "epoch": 0.46819077650768626, "grad_norm": 0.757015585899353, "learning_rate": 0.0005, "loss": 2.395543336868286, "step": 5939 }, { "epoch": 0.46826960977532517, "grad_norm": 0.8146267533302307, "learning_rate": 0.0005, "loss": 2.0052711963653564, "step": 5940 }, { "epoch": 0.46834844304296414, "grad_norm": 0.7502194046974182, "learning_rate": 0.0005, "loss": 1.7730073928833008, "step": 5941 }, { "epoch": 0.46842727631060305, "grad_norm": 0.7986584901809692, "learning_rate": 0.0005, "loss": 2.2114646434783936, "step": 5942 }, { "epoch": 0.468506109578242, "grad_norm": 0.8996094465255737, "learning_rate": 0.0005, "loss": 2.0573644638061523, "step": 5943 }, { "epoch": 0.468584942845881, "grad_norm": 0.8575459122657776, "learning_rate": 0.0005, "loss": 2.2009143829345703, "step": 5944 }, { "epoch": 0.4686637761135199, "grad_norm": 0.8522474765777588, "learning_rate": 0.0005, "loss": 1.7811747789382935, "step": 5945 }, { "epoch": 0.46874260938115886, "grad_norm": 0.676013708114624, "learning_rate": 0.0005, "loss": 2.323904275894165, "step": 5946 }, { "epoch": 0.46882144264879777, "grad_norm": 0.7155892252922058, "learning_rate": 0.0005, "loss": 2.0427663326263428, "step": 5947 }, { "epoch": 0.46890027591643674, "grad_norm": 0.8376377820968628, "learning_rate": 0.0005, "loss": 2.046309232711792, "step": 5948 }, { "epoch": 0.4689791091840757, "grad_norm": 0.6779383420944214, "learning_rate": 0.0005, "loss": 2.0703821182250977, "step": 5949 }, { "epoch": 0.4690579424517146, "grad_norm": 2.0925753116607666, "learning_rate": 0.0005, "loss": 2.570420026779175, "step": 5950 }, { "epoch": 0.4691367757193536, "grad_norm": 0.8364335894584656, "learning_rate": 0.0005, "loss": 2.393505811691284, "step": 5951 }, { "epoch": 0.4692156089869925, "grad_norm": 0.7888864874839783, "learning_rate": 0.0005, "loss": 2.2634997367858887, "step": 5952 }, { "epoch": 0.46929444225463146, "grad_norm": 0.6485593914985657, "learning_rate": 0.0005, "loss": 1.9245015382766724, "step": 5953 }, { "epoch": 0.4693732755222704, "grad_norm": 1.0678497552871704, "learning_rate": 0.0005, "loss": 2.3044986724853516, "step": 5954 }, { "epoch": 0.46945210878990934, "grad_norm": 0.8752442002296448, "learning_rate": 0.0005, "loss": 2.3598999977111816, "step": 5955 }, { "epoch": 0.4695309420575483, "grad_norm": 0.6865941286087036, "learning_rate": 0.0005, "loss": 2.2458083629608154, "step": 5956 }, { "epoch": 0.4696097753251872, "grad_norm": 0.6701242327690125, "learning_rate": 0.0005, "loss": 2.2367289066314697, "step": 5957 }, { "epoch": 0.4696886085928262, "grad_norm": 0.8318026661872864, "learning_rate": 0.0005, "loss": 2.5465965270996094, "step": 5958 }, { "epoch": 0.4697674418604651, "grad_norm": 0.784492015838623, "learning_rate": 0.0005, "loss": 1.8987603187561035, "step": 5959 }, { "epoch": 0.46984627512810406, "grad_norm": 0.6609687209129333, "learning_rate": 0.0005, "loss": 2.070709705352783, "step": 5960 }, { "epoch": 0.46992510839574303, "grad_norm": 0.6698420643806458, "learning_rate": 0.0005, "loss": 2.067561626434326, "step": 5961 }, { "epoch": 0.47000394166338194, "grad_norm": 0.719680666923523, "learning_rate": 0.0005, "loss": 2.0844664573669434, "step": 5962 }, { "epoch": 0.4700827749310209, "grad_norm": 0.9659118056297302, "learning_rate": 0.0005, "loss": 2.2027788162231445, "step": 5963 }, { "epoch": 0.4701616081986598, "grad_norm": 0.7127230763435364, "learning_rate": 0.0005, "loss": 2.2335758209228516, "step": 5964 }, { "epoch": 0.4702404414662988, "grad_norm": 0.8059515953063965, "learning_rate": 0.0005, "loss": 2.1371984481811523, "step": 5965 }, { "epoch": 0.4703192747339377, "grad_norm": 0.7487753629684448, "learning_rate": 0.0005, "loss": 2.530117988586426, "step": 5966 }, { "epoch": 0.47039810800157666, "grad_norm": 0.7182380557060242, "learning_rate": 0.0005, "loss": 2.191161632537842, "step": 5967 }, { "epoch": 0.47047694126921563, "grad_norm": 0.6452466249465942, "learning_rate": 0.0005, "loss": 1.941196084022522, "step": 5968 }, { "epoch": 0.47055577453685454, "grad_norm": 0.7491137385368347, "learning_rate": 0.0005, "loss": 2.048466682434082, "step": 5969 }, { "epoch": 0.4706346078044935, "grad_norm": 0.7303741574287415, "learning_rate": 0.0005, "loss": 2.3744683265686035, "step": 5970 }, { "epoch": 0.4707134410721324, "grad_norm": 0.7063186764717102, "learning_rate": 0.0005, "loss": 2.1232104301452637, "step": 5971 }, { "epoch": 0.4707922743397714, "grad_norm": 0.7804533243179321, "learning_rate": 0.0005, "loss": 1.864971399307251, "step": 5972 }, { "epoch": 0.4708711076074103, "grad_norm": 0.697094738483429, "learning_rate": 0.0005, "loss": 2.1607303619384766, "step": 5973 }, { "epoch": 0.47094994087504927, "grad_norm": 0.8234975934028625, "learning_rate": 0.0005, "loss": 2.444467782974243, "step": 5974 }, { "epoch": 0.47102877414268823, "grad_norm": 0.8207083940505981, "learning_rate": 0.0005, "loss": 1.929311990737915, "step": 5975 }, { "epoch": 0.47110760741032714, "grad_norm": 0.775415301322937, "learning_rate": 0.0005, "loss": 2.2306108474731445, "step": 5976 }, { "epoch": 0.4711864406779661, "grad_norm": 0.8799971342086792, "learning_rate": 0.0005, "loss": 2.200751543045044, "step": 5977 }, { "epoch": 0.471265273945605, "grad_norm": 0.7591375112533569, "learning_rate": 0.0005, "loss": 2.2198679447174072, "step": 5978 }, { "epoch": 0.471344107213244, "grad_norm": 0.7761450409889221, "learning_rate": 0.0005, "loss": 2.2425389289855957, "step": 5979 }, { "epoch": 0.47142294048088296, "grad_norm": 0.6865121126174927, "learning_rate": 0.0005, "loss": 2.220780849456787, "step": 5980 }, { "epoch": 0.47150177374852187, "grad_norm": 0.5981858968734741, "learning_rate": 0.0005, "loss": 2.039653778076172, "step": 5981 }, { "epoch": 0.47158060701616084, "grad_norm": 0.8000566363334656, "learning_rate": 0.0005, "loss": 1.9553359746932983, "step": 5982 }, { "epoch": 0.47165944028379975, "grad_norm": 0.5971846580505371, "learning_rate": 0.0005, "loss": 2.0322699546813965, "step": 5983 }, { "epoch": 0.4717382735514387, "grad_norm": 0.7094132900238037, "learning_rate": 0.0005, "loss": 2.2956416606903076, "step": 5984 }, { "epoch": 0.4718171068190776, "grad_norm": 0.6564924120903015, "learning_rate": 0.0005, "loss": 2.31650447845459, "step": 5985 }, { "epoch": 0.4718959400867166, "grad_norm": 0.5798284411430359, "learning_rate": 0.0005, "loss": 1.7828855514526367, "step": 5986 }, { "epoch": 0.47197477335435556, "grad_norm": 0.8511265516281128, "learning_rate": 0.0005, "loss": 2.175271987915039, "step": 5987 }, { "epoch": 0.47205360662199447, "grad_norm": 0.9288597106933594, "learning_rate": 0.0005, "loss": 2.1033694744110107, "step": 5988 }, { "epoch": 0.47213243988963344, "grad_norm": 0.6495901942253113, "learning_rate": 0.0005, "loss": 1.9745978116989136, "step": 5989 }, { "epoch": 0.47221127315727235, "grad_norm": 0.6904810667037964, "learning_rate": 0.0005, "loss": 1.946323037147522, "step": 5990 }, { "epoch": 0.4722901064249113, "grad_norm": 0.795929491519928, "learning_rate": 0.0005, "loss": 2.122326374053955, "step": 5991 }, { "epoch": 0.4723689396925503, "grad_norm": 0.7631859183311462, "learning_rate": 0.0005, "loss": 1.8757877349853516, "step": 5992 }, { "epoch": 0.4724477729601892, "grad_norm": 0.8710039854049683, "learning_rate": 0.0005, "loss": 1.8713736534118652, "step": 5993 }, { "epoch": 0.47252660622782816, "grad_norm": 1.042441964149475, "learning_rate": 0.0005, "loss": 2.3737711906433105, "step": 5994 }, { "epoch": 0.4726054394954671, "grad_norm": 0.7012948989868164, "learning_rate": 0.0005, "loss": 2.1355693340301514, "step": 5995 }, { "epoch": 0.47268427276310604, "grad_norm": 0.8818148970603943, "learning_rate": 0.0005, "loss": 2.246500253677368, "step": 5996 }, { "epoch": 0.47276310603074495, "grad_norm": 0.7619019746780396, "learning_rate": 0.0005, "loss": 2.043566942214966, "step": 5997 }, { "epoch": 0.4728419392983839, "grad_norm": 0.6504283547401428, "learning_rate": 0.0005, "loss": 2.013225555419922, "step": 5998 }, { "epoch": 0.4729207725660229, "grad_norm": 0.9130555391311646, "learning_rate": 0.0005, "loss": 2.272836208343506, "step": 5999 }, { "epoch": 0.4729996058336618, "grad_norm": 0.6866353154182434, "learning_rate": 0.0005, "loss": 2.3432865142822266, "step": 6000 }, { "epoch": 0.47307843910130076, "grad_norm": 0.6201152801513672, "learning_rate": 0.0005, "loss": 2.169743061065674, "step": 6001 }, { "epoch": 0.4731572723689397, "grad_norm": 0.8614976406097412, "learning_rate": 0.0005, "loss": 2.119560956954956, "step": 6002 }, { "epoch": 0.47323610563657864, "grad_norm": 0.6918981671333313, "learning_rate": 0.0005, "loss": 1.8664871454238892, "step": 6003 }, { "epoch": 0.47331493890421755, "grad_norm": 0.8318985104560852, "learning_rate": 0.0005, "loss": 2.2424488067626953, "step": 6004 }, { "epoch": 0.4733937721718565, "grad_norm": 0.6625984907150269, "learning_rate": 0.0005, "loss": 2.2102279663085938, "step": 6005 }, { "epoch": 0.4734726054394955, "grad_norm": 0.8985819220542908, "learning_rate": 0.0005, "loss": 2.089218854904175, "step": 6006 }, { "epoch": 0.4735514387071344, "grad_norm": 0.711482584476471, "learning_rate": 0.0005, "loss": 2.06500244140625, "step": 6007 }, { "epoch": 0.47363027197477336, "grad_norm": 1.1224994659423828, "learning_rate": 0.0005, "loss": 2.31526517868042, "step": 6008 }, { "epoch": 0.4737091052424123, "grad_norm": 0.9717295169830322, "learning_rate": 0.0005, "loss": 2.0442142486572266, "step": 6009 }, { "epoch": 0.47378793851005124, "grad_norm": 0.880034863948822, "learning_rate": 0.0005, "loss": 2.3168725967407227, "step": 6010 }, { "epoch": 0.4738667717776902, "grad_norm": 1.218376874923706, "learning_rate": 0.0005, "loss": 1.983931303024292, "step": 6011 }, { "epoch": 0.4739456050453291, "grad_norm": 0.7766829133033752, "learning_rate": 0.0005, "loss": 2.002206802368164, "step": 6012 }, { "epoch": 0.4740244383129681, "grad_norm": 0.6350474953651428, "learning_rate": 0.0005, "loss": 2.0028669834136963, "step": 6013 }, { "epoch": 0.474103271580607, "grad_norm": 0.6186656355857849, "learning_rate": 0.0005, "loss": 1.9937572479248047, "step": 6014 }, { "epoch": 0.47418210484824597, "grad_norm": 0.9055896401405334, "learning_rate": 0.0005, "loss": 1.841078281402588, "step": 6015 }, { "epoch": 0.4742609381158849, "grad_norm": 0.7005674839019775, "learning_rate": 0.0005, "loss": 1.8914721012115479, "step": 6016 }, { "epoch": 0.47433977138352384, "grad_norm": 0.7226890921592712, "learning_rate": 0.0005, "loss": 2.002701759338379, "step": 6017 }, { "epoch": 0.4744186046511628, "grad_norm": 0.7602525949478149, "learning_rate": 0.0005, "loss": 1.8577632904052734, "step": 6018 }, { "epoch": 0.4744974379188017, "grad_norm": 0.6594354510307312, "learning_rate": 0.0005, "loss": 2.0593554973602295, "step": 6019 }, { "epoch": 0.4745762711864407, "grad_norm": 0.992542564868927, "learning_rate": 0.0005, "loss": 2.2717247009277344, "step": 6020 }, { "epoch": 0.4746551044540796, "grad_norm": 0.9163529276847839, "learning_rate": 0.0005, "loss": 2.298443555831909, "step": 6021 }, { "epoch": 0.47473393772171857, "grad_norm": 0.8354882597923279, "learning_rate": 0.0005, "loss": 2.511368989944458, "step": 6022 }, { "epoch": 0.47481277098935754, "grad_norm": 0.7400940656661987, "learning_rate": 0.0005, "loss": 1.840693473815918, "step": 6023 }, { "epoch": 0.47489160425699645, "grad_norm": 0.8568270802497864, "learning_rate": 0.0005, "loss": 1.8393402099609375, "step": 6024 }, { "epoch": 0.4749704375246354, "grad_norm": 0.6945070028305054, "learning_rate": 0.0005, "loss": 1.9563475847244263, "step": 6025 }, { "epoch": 0.4750492707922743, "grad_norm": 0.8846202492713928, "learning_rate": 0.0005, "loss": 2.544334888458252, "step": 6026 }, { "epoch": 0.4751281040599133, "grad_norm": 0.8618253469467163, "learning_rate": 0.0005, "loss": 1.9274526834487915, "step": 6027 }, { "epoch": 0.4752069373275522, "grad_norm": 0.7485884428024292, "learning_rate": 0.0005, "loss": 2.317793607711792, "step": 6028 }, { "epoch": 0.47528577059519117, "grad_norm": 0.842446506023407, "learning_rate": 0.0005, "loss": 2.044219970703125, "step": 6029 }, { "epoch": 0.47536460386283014, "grad_norm": 0.7713102698326111, "learning_rate": 0.0005, "loss": 2.3812739849090576, "step": 6030 }, { "epoch": 0.47544343713046905, "grad_norm": 0.8028557896614075, "learning_rate": 0.0005, "loss": 2.1203413009643555, "step": 6031 }, { "epoch": 0.475522270398108, "grad_norm": 0.8523238897323608, "learning_rate": 0.0005, "loss": 2.106743574142456, "step": 6032 }, { "epoch": 0.4756011036657469, "grad_norm": 1.3243052959442139, "learning_rate": 0.0005, "loss": 2.1229896545410156, "step": 6033 }, { "epoch": 0.4756799369333859, "grad_norm": 0.750977635383606, "learning_rate": 0.0005, "loss": 2.008322238922119, "step": 6034 }, { "epoch": 0.4757587702010248, "grad_norm": 0.7950063347816467, "learning_rate": 0.0005, "loss": 1.8788118362426758, "step": 6035 }, { "epoch": 0.4758376034686638, "grad_norm": 0.8129026293754578, "learning_rate": 0.0005, "loss": 2.016587734222412, "step": 6036 }, { "epoch": 0.47591643673630274, "grad_norm": 0.89902263879776, "learning_rate": 0.0005, "loss": 2.2173404693603516, "step": 6037 }, { "epoch": 0.47599527000394165, "grad_norm": 1.0749316215515137, "learning_rate": 0.0005, "loss": 2.252735137939453, "step": 6038 }, { "epoch": 0.4760741032715806, "grad_norm": 0.9266635775566101, "learning_rate": 0.0005, "loss": 2.396869421005249, "step": 6039 }, { "epoch": 0.47615293653921953, "grad_norm": 0.7535120844841003, "learning_rate": 0.0005, "loss": 2.111142635345459, "step": 6040 }, { "epoch": 0.4762317698068585, "grad_norm": 0.8177496194839478, "learning_rate": 0.0005, "loss": 2.001410484313965, "step": 6041 }, { "epoch": 0.47631060307449746, "grad_norm": 0.6475705504417419, "learning_rate": 0.0005, "loss": 1.955190896987915, "step": 6042 }, { "epoch": 0.4763894363421364, "grad_norm": 1.053757667541504, "learning_rate": 0.0005, "loss": 2.114932060241699, "step": 6043 }, { "epoch": 0.47646826960977534, "grad_norm": 0.7868731617927551, "learning_rate": 0.0005, "loss": 2.074709177017212, "step": 6044 }, { "epoch": 0.47654710287741425, "grad_norm": 0.7978662252426147, "learning_rate": 0.0005, "loss": 2.066542625427246, "step": 6045 }, { "epoch": 0.4766259361450532, "grad_norm": 0.7766115069389343, "learning_rate": 0.0005, "loss": 2.361675500869751, "step": 6046 }, { "epoch": 0.47670476941269213, "grad_norm": 0.7375943660736084, "learning_rate": 0.0005, "loss": 2.0275113582611084, "step": 6047 }, { "epoch": 0.4767836026803311, "grad_norm": 0.9022160172462463, "learning_rate": 0.0005, "loss": 2.1091341972351074, "step": 6048 }, { "epoch": 0.47686243594797006, "grad_norm": 0.7503889799118042, "learning_rate": 0.0005, "loss": 2.197866439819336, "step": 6049 }, { "epoch": 0.476941269215609, "grad_norm": 0.6630483865737915, "learning_rate": 0.0005, "loss": 1.9732741117477417, "step": 6050 }, { "epoch": 0.47702010248324794, "grad_norm": 0.8733715415000916, "learning_rate": 0.0005, "loss": 2.066333055496216, "step": 6051 }, { "epoch": 0.47709893575088685, "grad_norm": 0.819506824016571, "learning_rate": 0.0005, "loss": 1.99871826171875, "step": 6052 }, { "epoch": 0.4771777690185258, "grad_norm": 0.6601189970970154, "learning_rate": 0.0005, "loss": 2.1408729553222656, "step": 6053 }, { "epoch": 0.4772566022861648, "grad_norm": 0.62352454662323, "learning_rate": 0.0005, "loss": 1.8690485954284668, "step": 6054 }, { "epoch": 0.4773354355538037, "grad_norm": 0.7640159726142883, "learning_rate": 0.0005, "loss": 1.9330573081970215, "step": 6055 }, { "epoch": 0.47741426882144267, "grad_norm": 0.8558868169784546, "learning_rate": 0.0005, "loss": 2.113307476043701, "step": 6056 }, { "epoch": 0.4774931020890816, "grad_norm": 0.7224906086921692, "learning_rate": 0.0005, "loss": 2.231673240661621, "step": 6057 }, { "epoch": 0.47757193535672054, "grad_norm": 0.7109651565551758, "learning_rate": 0.0005, "loss": 2.3874146938323975, "step": 6058 }, { "epoch": 0.47765076862435946, "grad_norm": 0.9782761335372925, "learning_rate": 0.0005, "loss": 2.247387409210205, "step": 6059 }, { "epoch": 0.4777296018919984, "grad_norm": 0.7241505980491638, "learning_rate": 0.0005, "loss": 2.1715307235717773, "step": 6060 }, { "epoch": 0.4778084351596374, "grad_norm": 0.6191240549087524, "learning_rate": 0.0005, "loss": 1.5919826030731201, "step": 6061 }, { "epoch": 0.4778872684272763, "grad_norm": 0.6779581904411316, "learning_rate": 0.0005, "loss": 2.354823589324951, "step": 6062 }, { "epoch": 0.47796610169491527, "grad_norm": 0.6982458829879761, "learning_rate": 0.0005, "loss": 2.1288321018218994, "step": 6063 }, { "epoch": 0.4780449349625542, "grad_norm": 0.8128506541252136, "learning_rate": 0.0005, "loss": 2.1773793697357178, "step": 6064 }, { "epoch": 0.47812376823019315, "grad_norm": 0.9244840145111084, "learning_rate": 0.0005, "loss": 2.165800094604492, "step": 6065 }, { "epoch": 0.47820260149783206, "grad_norm": 0.6654143333435059, "learning_rate": 0.0005, "loss": 2.0513086318969727, "step": 6066 }, { "epoch": 0.478281434765471, "grad_norm": 0.6944580674171448, "learning_rate": 0.0005, "loss": 2.2045888900756836, "step": 6067 }, { "epoch": 0.47836026803311, "grad_norm": 0.8379547595977783, "learning_rate": 0.0005, "loss": 2.2848243713378906, "step": 6068 }, { "epoch": 0.4784391013007489, "grad_norm": 0.790023684501648, "learning_rate": 0.0005, "loss": 2.2001588344573975, "step": 6069 }, { "epoch": 0.47851793456838787, "grad_norm": 0.8336043357849121, "learning_rate": 0.0005, "loss": 1.9002301692962646, "step": 6070 }, { "epoch": 0.4785967678360268, "grad_norm": 0.9132195711135864, "learning_rate": 0.0005, "loss": 1.8249623775482178, "step": 6071 }, { "epoch": 0.47867560110366575, "grad_norm": 1.5344854593276978, "learning_rate": 0.0005, "loss": 2.1949052810668945, "step": 6072 }, { "epoch": 0.4787544343713047, "grad_norm": 0.7461217045783997, "learning_rate": 0.0005, "loss": 2.239701986312866, "step": 6073 }, { "epoch": 0.4788332676389436, "grad_norm": 0.8346728682518005, "learning_rate": 0.0005, "loss": 2.254577159881592, "step": 6074 }, { "epoch": 0.4789121009065826, "grad_norm": 0.764660656452179, "learning_rate": 0.0005, "loss": 2.045070171356201, "step": 6075 }, { "epoch": 0.4789909341742215, "grad_norm": 0.665938675403595, "learning_rate": 0.0005, "loss": 1.7537789344787598, "step": 6076 }, { "epoch": 0.4790697674418605, "grad_norm": 0.8018065094947815, "learning_rate": 0.0005, "loss": 1.9821321964263916, "step": 6077 }, { "epoch": 0.4791486007094994, "grad_norm": 0.9270064830780029, "learning_rate": 0.0005, "loss": 2.1068849563598633, "step": 6078 }, { "epoch": 0.47922743397713835, "grad_norm": 0.7490548491477966, "learning_rate": 0.0005, "loss": 1.8183801174163818, "step": 6079 }, { "epoch": 0.4793062672447773, "grad_norm": 1.0680477619171143, "learning_rate": 0.0005, "loss": 2.501175880432129, "step": 6080 }, { "epoch": 0.47938510051241623, "grad_norm": 0.6618924140930176, "learning_rate": 0.0005, "loss": 2.2420578002929688, "step": 6081 }, { "epoch": 0.4794639337800552, "grad_norm": 0.7480191588401794, "learning_rate": 0.0005, "loss": 2.3315916061401367, "step": 6082 }, { "epoch": 0.4795427670476941, "grad_norm": 0.9333640933036804, "learning_rate": 0.0005, "loss": 1.9855742454528809, "step": 6083 }, { "epoch": 0.4796216003153331, "grad_norm": 0.7426320910453796, "learning_rate": 0.0005, "loss": 2.1526787281036377, "step": 6084 }, { "epoch": 0.47970043358297204, "grad_norm": 0.7064155340194702, "learning_rate": 0.0005, "loss": 1.9715386629104614, "step": 6085 }, { "epoch": 0.47977926685061095, "grad_norm": 0.8047029972076416, "learning_rate": 0.0005, "loss": 2.021794557571411, "step": 6086 }, { "epoch": 0.4798581001182499, "grad_norm": 0.6127332448959351, "learning_rate": 0.0005, "loss": 2.0486528873443604, "step": 6087 }, { "epoch": 0.47993693338588883, "grad_norm": 0.7618416547775269, "learning_rate": 0.0005, "loss": 2.07724666595459, "step": 6088 }, { "epoch": 0.4800157666535278, "grad_norm": 0.7653990983963013, "learning_rate": 0.0005, "loss": 2.0176784992218018, "step": 6089 }, { "epoch": 0.4800945999211667, "grad_norm": 0.6659051775932312, "learning_rate": 0.0005, "loss": 2.3331265449523926, "step": 6090 }, { "epoch": 0.4801734331888057, "grad_norm": 0.8430803418159485, "learning_rate": 0.0005, "loss": 2.1467599868774414, "step": 6091 }, { "epoch": 0.48025226645644464, "grad_norm": 0.6315040588378906, "learning_rate": 0.0005, "loss": 2.055532932281494, "step": 6092 }, { "epoch": 0.48033109972408355, "grad_norm": 0.8662042617797852, "learning_rate": 0.0005, "loss": 2.2427725791931152, "step": 6093 }, { "epoch": 0.4804099329917225, "grad_norm": 0.6455965042114258, "learning_rate": 0.0005, "loss": 1.8733603954315186, "step": 6094 }, { "epoch": 0.48048876625936143, "grad_norm": 0.8550066947937012, "learning_rate": 0.0005, "loss": 1.842365026473999, "step": 6095 }, { "epoch": 0.4805675995270004, "grad_norm": 0.8287873864173889, "learning_rate": 0.0005, "loss": 2.1344637870788574, "step": 6096 }, { "epoch": 0.4806464327946393, "grad_norm": 0.6512097716331482, "learning_rate": 0.0005, "loss": 1.943285584449768, "step": 6097 }, { "epoch": 0.4807252660622783, "grad_norm": 0.7978892922401428, "learning_rate": 0.0005, "loss": 2.1906368732452393, "step": 6098 }, { "epoch": 0.48080409932991724, "grad_norm": 1.0080327987670898, "learning_rate": 0.0005, "loss": 2.052537679672241, "step": 6099 }, { "epoch": 0.48088293259755616, "grad_norm": 0.7020682692527771, "learning_rate": 0.0005, "loss": 1.8672428131103516, "step": 6100 }, { "epoch": 0.4809617658651951, "grad_norm": 0.6790408492088318, "learning_rate": 0.0005, "loss": 2.311319589614868, "step": 6101 }, { "epoch": 0.48104059913283403, "grad_norm": 0.6582736372947693, "learning_rate": 0.0005, "loss": 1.8628379106521606, "step": 6102 }, { "epoch": 0.481119432400473, "grad_norm": 0.850816547870636, "learning_rate": 0.0005, "loss": 2.2808008193969727, "step": 6103 }, { "epoch": 0.48119826566811197, "grad_norm": 0.6627414226531982, "learning_rate": 0.0005, "loss": 2.034346342086792, "step": 6104 }, { "epoch": 0.4812770989357509, "grad_norm": 0.8894122838973999, "learning_rate": 0.0005, "loss": 2.2898101806640625, "step": 6105 }, { "epoch": 0.48135593220338985, "grad_norm": 10.918322563171387, "learning_rate": 0.0005, "loss": 1.8985198736190796, "step": 6106 }, { "epoch": 0.48143476547102876, "grad_norm": 0.7449589967727661, "learning_rate": 0.0005, "loss": 2.0463976860046387, "step": 6107 }, { "epoch": 0.4815135987386677, "grad_norm": 1.127840280532837, "learning_rate": 0.0005, "loss": 2.1580941677093506, "step": 6108 }, { "epoch": 0.48159243200630664, "grad_norm": 0.8486847281455994, "learning_rate": 0.0005, "loss": 2.1951284408569336, "step": 6109 }, { "epoch": 0.4816712652739456, "grad_norm": 0.7022362947463989, "learning_rate": 0.0005, "loss": 1.7318785190582275, "step": 6110 }, { "epoch": 0.48175009854158457, "grad_norm": 0.655464768409729, "learning_rate": 0.0005, "loss": 2.1941030025482178, "step": 6111 }, { "epoch": 0.4818289318092235, "grad_norm": 1.0176564455032349, "learning_rate": 0.0005, "loss": 2.086956024169922, "step": 6112 }, { "epoch": 0.48190776507686245, "grad_norm": 0.8235392570495605, "learning_rate": 0.0005, "loss": 1.8808894157409668, "step": 6113 }, { "epoch": 0.48198659834450136, "grad_norm": 0.7177219986915588, "learning_rate": 0.0005, "loss": 1.9571177959442139, "step": 6114 }, { "epoch": 0.4820654316121403, "grad_norm": 0.8989542722702026, "learning_rate": 0.0005, "loss": 2.2980499267578125, "step": 6115 }, { "epoch": 0.4821442648797793, "grad_norm": 0.7545682787895203, "learning_rate": 0.0005, "loss": 2.089552640914917, "step": 6116 }, { "epoch": 0.4822230981474182, "grad_norm": 0.8262479305267334, "learning_rate": 0.0005, "loss": 2.294506788253784, "step": 6117 }, { "epoch": 0.4823019314150572, "grad_norm": 0.9219067692756653, "learning_rate": 0.0005, "loss": 1.9123228788375854, "step": 6118 }, { "epoch": 0.4823807646826961, "grad_norm": 0.730721116065979, "learning_rate": 0.0005, "loss": 2.3995747566223145, "step": 6119 }, { "epoch": 0.48245959795033505, "grad_norm": 0.7607832551002502, "learning_rate": 0.0005, "loss": 2.038409471511841, "step": 6120 }, { "epoch": 0.48253843121797396, "grad_norm": 0.846230149269104, "learning_rate": 0.0005, "loss": 2.202916145324707, "step": 6121 }, { "epoch": 0.48261726448561293, "grad_norm": 0.7430521845817566, "learning_rate": 0.0005, "loss": 2.072014808654785, "step": 6122 }, { "epoch": 0.4826960977532519, "grad_norm": 0.6597007513046265, "learning_rate": 0.0005, "loss": 2.4150218963623047, "step": 6123 }, { "epoch": 0.4827749310208908, "grad_norm": 0.7174841165542603, "learning_rate": 0.0005, "loss": 2.421067476272583, "step": 6124 }, { "epoch": 0.4828537642885298, "grad_norm": 0.5777460932731628, "learning_rate": 0.0005, "loss": 2.2904715538024902, "step": 6125 }, { "epoch": 0.4829325975561687, "grad_norm": 0.9124022722244263, "learning_rate": 0.0005, "loss": 2.2389683723449707, "step": 6126 }, { "epoch": 0.48301143082380765, "grad_norm": 0.9738324284553528, "learning_rate": 0.0005, "loss": 2.3887746334075928, "step": 6127 }, { "epoch": 0.48309026409144656, "grad_norm": 0.7866266965866089, "learning_rate": 0.0005, "loss": 2.107842445373535, "step": 6128 }, { "epoch": 0.48316909735908553, "grad_norm": 0.7301947474479675, "learning_rate": 0.0005, "loss": 2.112515449523926, "step": 6129 }, { "epoch": 0.4832479306267245, "grad_norm": 0.7578866481781006, "learning_rate": 0.0005, "loss": 1.9591469764709473, "step": 6130 }, { "epoch": 0.4833267638943634, "grad_norm": 0.7376816868782043, "learning_rate": 0.0005, "loss": 2.1472601890563965, "step": 6131 }, { "epoch": 0.4834055971620024, "grad_norm": 0.6949366331100464, "learning_rate": 0.0005, "loss": 2.0344181060791016, "step": 6132 }, { "epoch": 0.4834844304296413, "grad_norm": 0.6581720113754272, "learning_rate": 0.0005, "loss": 1.9954763650894165, "step": 6133 }, { "epoch": 0.48356326369728025, "grad_norm": 0.7502775192260742, "learning_rate": 0.0005, "loss": 1.8951258659362793, "step": 6134 }, { "epoch": 0.4836420969649192, "grad_norm": 0.5747241973876953, "learning_rate": 0.0005, "loss": 1.8607637882232666, "step": 6135 }, { "epoch": 0.48372093023255813, "grad_norm": 0.728421688079834, "learning_rate": 0.0005, "loss": 2.226428985595703, "step": 6136 }, { "epoch": 0.4837997635001971, "grad_norm": 0.6421897411346436, "learning_rate": 0.0005, "loss": 2.148902654647827, "step": 6137 }, { "epoch": 0.483878596767836, "grad_norm": 0.7295370101928711, "learning_rate": 0.0005, "loss": 2.0132272243499756, "step": 6138 }, { "epoch": 0.483957430035475, "grad_norm": 0.8645273447036743, "learning_rate": 0.0005, "loss": 2.2312541007995605, "step": 6139 }, { "epoch": 0.4840362633031139, "grad_norm": 0.7516099810600281, "learning_rate": 0.0005, "loss": 2.3027284145355225, "step": 6140 }, { "epoch": 0.48411509657075286, "grad_norm": 0.7747275233268738, "learning_rate": 0.0005, "loss": 2.5761914253234863, "step": 6141 }, { "epoch": 0.4841939298383918, "grad_norm": 0.8190990090370178, "learning_rate": 0.0005, "loss": 2.2054128646850586, "step": 6142 }, { "epoch": 0.48427276310603073, "grad_norm": 0.7855672836303711, "learning_rate": 0.0005, "loss": 1.991410493850708, "step": 6143 }, { "epoch": 0.4843515963736697, "grad_norm": 0.7087606191635132, "learning_rate": 0.0005, "loss": 1.9702067375183105, "step": 6144 }, { "epoch": 0.4844304296413086, "grad_norm": 0.6990965604782104, "learning_rate": 0.0005, "loss": 2.136385679244995, "step": 6145 }, { "epoch": 0.4845092629089476, "grad_norm": 0.8081690669059753, "learning_rate": 0.0005, "loss": 2.163867950439453, "step": 6146 }, { "epoch": 0.48458809617658655, "grad_norm": 0.6497622728347778, "learning_rate": 0.0005, "loss": 1.8880960941314697, "step": 6147 }, { "epoch": 0.48466692944422546, "grad_norm": 0.8350061178207397, "learning_rate": 0.0005, "loss": 1.9384007453918457, "step": 6148 }, { "epoch": 0.4847457627118644, "grad_norm": 0.7939273715019226, "learning_rate": 0.0005, "loss": 2.0116324424743652, "step": 6149 }, { "epoch": 0.48482459597950334, "grad_norm": 0.7656359672546387, "learning_rate": 0.0005, "loss": 2.001217842102051, "step": 6150 }, { "epoch": 0.4849034292471423, "grad_norm": 0.5937368869781494, "learning_rate": 0.0005, "loss": 1.836553931236267, "step": 6151 }, { "epoch": 0.4849822625147812, "grad_norm": 0.8921777009963989, "learning_rate": 0.0005, "loss": 2.5095369815826416, "step": 6152 }, { "epoch": 0.4850610957824202, "grad_norm": 0.6932396292686462, "learning_rate": 0.0005, "loss": 2.2327184677124023, "step": 6153 }, { "epoch": 0.48513992905005915, "grad_norm": 0.8056274056434631, "learning_rate": 0.0005, "loss": 2.3271961212158203, "step": 6154 }, { "epoch": 0.48521876231769806, "grad_norm": 0.7297812104225159, "learning_rate": 0.0005, "loss": 2.303095817565918, "step": 6155 }, { "epoch": 0.485297595585337, "grad_norm": 0.884975254535675, "learning_rate": 0.0005, "loss": 2.03898024559021, "step": 6156 }, { "epoch": 0.48537642885297594, "grad_norm": 0.7662782669067383, "learning_rate": 0.0005, "loss": 2.0297040939331055, "step": 6157 }, { "epoch": 0.4854552621206149, "grad_norm": 0.9570702314376831, "learning_rate": 0.0005, "loss": 2.246480703353882, "step": 6158 }, { "epoch": 0.4855340953882538, "grad_norm": 0.7836255431175232, "learning_rate": 0.0005, "loss": 2.321652889251709, "step": 6159 }, { "epoch": 0.4856129286558928, "grad_norm": 0.7321615219116211, "learning_rate": 0.0005, "loss": 2.3031275272369385, "step": 6160 }, { "epoch": 0.48569176192353175, "grad_norm": 1.083320140838623, "learning_rate": 0.0005, "loss": 2.1090822219848633, "step": 6161 }, { "epoch": 0.48577059519117066, "grad_norm": 0.7854934930801392, "learning_rate": 0.0005, "loss": 1.9922902584075928, "step": 6162 }, { "epoch": 0.48584942845880963, "grad_norm": 0.7868211269378662, "learning_rate": 0.0005, "loss": 1.8831151723861694, "step": 6163 }, { "epoch": 0.48592826172644854, "grad_norm": 0.781467854976654, "learning_rate": 0.0005, "loss": 2.1318247318267822, "step": 6164 }, { "epoch": 0.4860070949940875, "grad_norm": 0.7463970184326172, "learning_rate": 0.0005, "loss": 1.8520305156707764, "step": 6165 }, { "epoch": 0.4860859282617265, "grad_norm": 0.872862696647644, "learning_rate": 0.0005, "loss": 2.449709177017212, "step": 6166 }, { "epoch": 0.4861647615293654, "grad_norm": 0.6697245240211487, "learning_rate": 0.0005, "loss": 2.047396659851074, "step": 6167 }, { "epoch": 0.48624359479700435, "grad_norm": 0.8821247816085815, "learning_rate": 0.0005, "loss": 2.1570520401000977, "step": 6168 }, { "epoch": 0.48632242806464326, "grad_norm": 1.3032294511795044, "learning_rate": 0.0005, "loss": 2.150317430496216, "step": 6169 }, { "epoch": 0.48640126133228223, "grad_norm": 0.7212545275688171, "learning_rate": 0.0005, "loss": 2.1546127796173096, "step": 6170 }, { "epoch": 0.48648009459992114, "grad_norm": 0.7933535575866699, "learning_rate": 0.0005, "loss": 2.127371311187744, "step": 6171 }, { "epoch": 0.4865589278675601, "grad_norm": 0.9530965089797974, "learning_rate": 0.0005, "loss": 2.0139102935791016, "step": 6172 }, { "epoch": 0.4866377611351991, "grad_norm": 0.615286648273468, "learning_rate": 0.0005, "loss": 2.0971343517303467, "step": 6173 }, { "epoch": 0.486716594402838, "grad_norm": 0.7400515079498291, "learning_rate": 0.0005, "loss": 1.8612425327301025, "step": 6174 }, { "epoch": 0.48679542767047695, "grad_norm": 0.8957095146179199, "learning_rate": 0.0005, "loss": 1.9534012079238892, "step": 6175 }, { "epoch": 0.48687426093811587, "grad_norm": 0.81638103723526, "learning_rate": 0.0005, "loss": 2.2120883464813232, "step": 6176 }, { "epoch": 0.48695309420575483, "grad_norm": 0.8371590971946716, "learning_rate": 0.0005, "loss": 2.6372222900390625, "step": 6177 }, { "epoch": 0.4870319274733938, "grad_norm": 0.8636767268180847, "learning_rate": 0.0005, "loss": 1.9811677932739258, "step": 6178 }, { "epoch": 0.4871107607410327, "grad_norm": 0.6232184767723083, "learning_rate": 0.0005, "loss": 2.0051372051239014, "step": 6179 }, { "epoch": 0.4871895940086717, "grad_norm": 0.6300936937332153, "learning_rate": 0.0005, "loss": 2.103358745574951, "step": 6180 }, { "epoch": 0.4872684272763106, "grad_norm": 0.7210426330566406, "learning_rate": 0.0005, "loss": 1.7569375038146973, "step": 6181 }, { "epoch": 0.48734726054394956, "grad_norm": 0.5719450116157532, "learning_rate": 0.0005, "loss": 2.05875301361084, "step": 6182 }, { "epoch": 0.48742609381158847, "grad_norm": 0.6611929535865784, "learning_rate": 0.0005, "loss": 2.0402238368988037, "step": 6183 }, { "epoch": 0.48750492707922743, "grad_norm": 0.5881970524787903, "learning_rate": 0.0005, "loss": 2.3973400592803955, "step": 6184 }, { "epoch": 0.4875837603468664, "grad_norm": 0.6865686774253845, "learning_rate": 0.0005, "loss": 1.889713168144226, "step": 6185 }, { "epoch": 0.4876625936145053, "grad_norm": 0.8483547568321228, "learning_rate": 0.0005, "loss": 2.045459270477295, "step": 6186 }, { "epoch": 0.4877414268821443, "grad_norm": 0.8159933090209961, "learning_rate": 0.0005, "loss": 1.896444320678711, "step": 6187 }, { "epoch": 0.4878202601497832, "grad_norm": 0.9703830480575562, "learning_rate": 0.0005, "loss": 1.9507853984832764, "step": 6188 }, { "epoch": 0.48789909341742216, "grad_norm": 0.7298377156257629, "learning_rate": 0.0005, "loss": 1.636018991470337, "step": 6189 }, { "epoch": 0.48797792668506107, "grad_norm": 0.6409031748771667, "learning_rate": 0.0005, "loss": 1.742830514907837, "step": 6190 }, { "epoch": 0.48805675995270004, "grad_norm": 1.0092508792877197, "learning_rate": 0.0005, "loss": 1.9635201692581177, "step": 6191 }, { "epoch": 0.488135593220339, "grad_norm": 0.7816257476806641, "learning_rate": 0.0005, "loss": 2.265505790710449, "step": 6192 }, { "epoch": 0.4882144264879779, "grad_norm": 0.798008143901825, "learning_rate": 0.0005, "loss": 2.0912437438964844, "step": 6193 }, { "epoch": 0.4882932597556169, "grad_norm": 0.9140622615814209, "learning_rate": 0.0005, "loss": 2.3621630668640137, "step": 6194 }, { "epoch": 0.4883720930232558, "grad_norm": 0.7662726640701294, "learning_rate": 0.0005, "loss": 2.1996681690216064, "step": 6195 }, { "epoch": 0.48845092629089476, "grad_norm": 0.731513500213623, "learning_rate": 0.0005, "loss": 2.0890650749206543, "step": 6196 }, { "epoch": 0.4885297595585337, "grad_norm": 0.7022278904914856, "learning_rate": 0.0005, "loss": 2.2635931968688965, "step": 6197 }, { "epoch": 0.48860859282617264, "grad_norm": 0.647158145904541, "learning_rate": 0.0005, "loss": 2.01544189453125, "step": 6198 }, { "epoch": 0.4886874260938116, "grad_norm": 0.8577305674552917, "learning_rate": 0.0005, "loss": 2.083911418914795, "step": 6199 }, { "epoch": 0.4887662593614505, "grad_norm": 0.7195108532905579, "learning_rate": 0.0005, "loss": 2.1424524784088135, "step": 6200 }, { "epoch": 0.4888450926290895, "grad_norm": 0.6897614598274231, "learning_rate": 0.0005, "loss": 2.0297183990478516, "step": 6201 }, { "epoch": 0.4889239258967284, "grad_norm": 0.7113688588142395, "learning_rate": 0.0005, "loss": 2.4005656242370605, "step": 6202 }, { "epoch": 0.48900275916436736, "grad_norm": 0.9588683843612671, "learning_rate": 0.0005, "loss": 1.7431731224060059, "step": 6203 }, { "epoch": 0.48908159243200633, "grad_norm": 0.8227931261062622, "learning_rate": 0.0005, "loss": 2.2961041927337646, "step": 6204 }, { "epoch": 0.48916042569964524, "grad_norm": 0.7509339451789856, "learning_rate": 0.0005, "loss": 1.8369708061218262, "step": 6205 }, { "epoch": 0.4892392589672842, "grad_norm": 0.733521044254303, "learning_rate": 0.0005, "loss": 1.9028714895248413, "step": 6206 }, { "epoch": 0.4893180922349231, "grad_norm": 0.6752141714096069, "learning_rate": 0.0005, "loss": 2.136909246444702, "step": 6207 }, { "epoch": 0.4893969255025621, "grad_norm": 0.6215294599533081, "learning_rate": 0.0005, "loss": 2.0051403045654297, "step": 6208 }, { "epoch": 0.489475758770201, "grad_norm": 0.7651437520980835, "learning_rate": 0.0005, "loss": 2.1105518341064453, "step": 6209 }, { "epoch": 0.48955459203783996, "grad_norm": 1.0211446285247803, "learning_rate": 0.0005, "loss": 2.036362409591675, "step": 6210 }, { "epoch": 0.48963342530547893, "grad_norm": 0.7290043234825134, "learning_rate": 0.0005, "loss": 2.2569379806518555, "step": 6211 }, { "epoch": 0.48971225857311784, "grad_norm": 0.7194265127182007, "learning_rate": 0.0005, "loss": 2.194409132003784, "step": 6212 }, { "epoch": 0.4897910918407568, "grad_norm": 9.788618087768555, "learning_rate": 0.0005, "loss": 2.2681655883789062, "step": 6213 }, { "epoch": 0.4898699251083957, "grad_norm": 0.9110327363014221, "learning_rate": 0.0005, "loss": 2.128821611404419, "step": 6214 }, { "epoch": 0.4899487583760347, "grad_norm": 1.1020444631576538, "learning_rate": 0.0005, "loss": 2.0779733657836914, "step": 6215 }, { "epoch": 0.49002759164367365, "grad_norm": 1.0843780040740967, "learning_rate": 0.0005, "loss": 2.121764898300171, "step": 6216 }, { "epoch": 0.49010642491131257, "grad_norm": 0.6869621276855469, "learning_rate": 0.0005, "loss": 2.2800958156585693, "step": 6217 }, { "epoch": 0.49018525817895153, "grad_norm": 0.8416841626167297, "learning_rate": 0.0005, "loss": 2.1283555030822754, "step": 6218 }, { "epoch": 0.49026409144659044, "grad_norm": 0.8394928574562073, "learning_rate": 0.0005, "loss": 1.7388414144515991, "step": 6219 }, { "epoch": 0.4903429247142294, "grad_norm": 0.9484405517578125, "learning_rate": 0.0005, "loss": 2.0347394943237305, "step": 6220 }, { "epoch": 0.4904217579818683, "grad_norm": 0.9218053817749023, "learning_rate": 0.0005, "loss": 2.1929268836975098, "step": 6221 }, { "epoch": 0.4905005912495073, "grad_norm": 0.7000519037246704, "learning_rate": 0.0005, "loss": 2.0266060829162598, "step": 6222 }, { "epoch": 0.49057942451714626, "grad_norm": 0.6931633353233337, "learning_rate": 0.0005, "loss": 1.9999442100524902, "step": 6223 }, { "epoch": 0.49065825778478517, "grad_norm": 0.7547497749328613, "learning_rate": 0.0005, "loss": 2.0729331970214844, "step": 6224 }, { "epoch": 0.49073709105242413, "grad_norm": 0.7689460515975952, "learning_rate": 0.0005, "loss": 1.9815797805786133, "step": 6225 }, { "epoch": 0.49081592432006305, "grad_norm": 0.7096936106681824, "learning_rate": 0.0005, "loss": 2.1038365364074707, "step": 6226 }, { "epoch": 0.490894757587702, "grad_norm": 0.8129679560661316, "learning_rate": 0.0005, "loss": 2.1254019737243652, "step": 6227 }, { "epoch": 0.490973590855341, "grad_norm": 1.0562137365341187, "learning_rate": 0.0005, "loss": 1.8266654014587402, "step": 6228 }, { "epoch": 0.4910524241229799, "grad_norm": 0.9056048393249512, "learning_rate": 0.0005, "loss": 2.0528573989868164, "step": 6229 }, { "epoch": 0.49113125739061886, "grad_norm": 0.6481890082359314, "learning_rate": 0.0005, "loss": 2.038233518600464, "step": 6230 }, { "epoch": 0.49121009065825777, "grad_norm": 0.6447815299034119, "learning_rate": 0.0005, "loss": 2.0936760902404785, "step": 6231 }, { "epoch": 0.49128892392589674, "grad_norm": 0.7284529805183411, "learning_rate": 0.0005, "loss": 2.020638942718506, "step": 6232 }, { "epoch": 0.49136775719353565, "grad_norm": 0.8731178045272827, "learning_rate": 0.0005, "loss": 2.253371000289917, "step": 6233 }, { "epoch": 0.4914465904611746, "grad_norm": 0.753470242023468, "learning_rate": 0.0005, "loss": 2.0492751598358154, "step": 6234 }, { "epoch": 0.4915254237288136, "grad_norm": 0.9811033010482788, "learning_rate": 0.0005, "loss": 1.9167307615280151, "step": 6235 }, { "epoch": 0.4916042569964525, "grad_norm": 0.8715121150016785, "learning_rate": 0.0005, "loss": 2.1621322631835938, "step": 6236 }, { "epoch": 0.49168309026409146, "grad_norm": 0.7179142832756042, "learning_rate": 0.0005, "loss": 2.3578908443450928, "step": 6237 }, { "epoch": 0.49176192353173037, "grad_norm": 0.7337979674339294, "learning_rate": 0.0005, "loss": 2.352952718734741, "step": 6238 }, { "epoch": 0.49184075679936934, "grad_norm": 0.6089599132537842, "learning_rate": 0.0005, "loss": 1.7365686893463135, "step": 6239 }, { "epoch": 0.49191959006700825, "grad_norm": 0.7781942486763, "learning_rate": 0.0005, "loss": 1.8782528638839722, "step": 6240 }, { "epoch": 0.4919984233346472, "grad_norm": 0.6551744341850281, "learning_rate": 0.0005, "loss": 1.6541078090667725, "step": 6241 }, { "epoch": 0.4920772566022862, "grad_norm": 1.2736717462539673, "learning_rate": 0.0005, "loss": 2.3199586868286133, "step": 6242 }, { "epoch": 0.4921560898699251, "grad_norm": 0.689804196357727, "learning_rate": 0.0005, "loss": 1.9719129800796509, "step": 6243 }, { "epoch": 0.49223492313756406, "grad_norm": 1.0178277492523193, "learning_rate": 0.0005, "loss": 1.7613184452056885, "step": 6244 }, { "epoch": 0.492313756405203, "grad_norm": 0.7390385270118713, "learning_rate": 0.0005, "loss": 1.9785754680633545, "step": 6245 }, { "epoch": 0.49239258967284194, "grad_norm": 5.080535888671875, "learning_rate": 0.0005, "loss": 2.015932559967041, "step": 6246 }, { "epoch": 0.4924714229404809, "grad_norm": 0.8546715974807739, "learning_rate": 0.0005, "loss": 2.2535147666931152, "step": 6247 }, { "epoch": 0.4925502562081198, "grad_norm": 0.7733351588249207, "learning_rate": 0.0005, "loss": 2.059969425201416, "step": 6248 }, { "epoch": 0.4926290894757588, "grad_norm": 0.9166790843009949, "learning_rate": 0.0005, "loss": 2.3150928020477295, "step": 6249 }, { "epoch": 0.4927079227433977, "grad_norm": 0.8748522400856018, "learning_rate": 0.0005, "loss": 2.3727967739105225, "step": 6250 }, { "epoch": 0.49278675601103666, "grad_norm": 0.6632548570632935, "learning_rate": 0.0005, "loss": 2.151576280593872, "step": 6251 }, { "epoch": 0.4928655892786756, "grad_norm": 0.8089959621429443, "learning_rate": 0.0005, "loss": 2.1254830360412598, "step": 6252 }, { "epoch": 0.49294442254631454, "grad_norm": 0.8368954062461853, "learning_rate": 0.0005, "loss": 1.9488874673843384, "step": 6253 }, { "epoch": 0.4930232558139535, "grad_norm": 0.6819337010383606, "learning_rate": 0.0005, "loss": 2.020141124725342, "step": 6254 }, { "epoch": 0.4931020890815924, "grad_norm": 0.7998510599136353, "learning_rate": 0.0005, "loss": 2.081468343734741, "step": 6255 }, { "epoch": 0.4931809223492314, "grad_norm": 0.8035281896591187, "learning_rate": 0.0005, "loss": 2.0784549713134766, "step": 6256 }, { "epoch": 0.4932597556168703, "grad_norm": 0.7364303469657898, "learning_rate": 0.0005, "loss": 2.0501558780670166, "step": 6257 }, { "epoch": 0.49333858888450927, "grad_norm": 0.7789298892021179, "learning_rate": 0.0005, "loss": 2.2139406204223633, "step": 6258 }, { "epoch": 0.49341742215214823, "grad_norm": 0.7188334465026855, "learning_rate": 0.0005, "loss": 2.0551581382751465, "step": 6259 }, { "epoch": 0.49349625541978714, "grad_norm": 0.728101909160614, "learning_rate": 0.0005, "loss": 2.173211097717285, "step": 6260 }, { "epoch": 0.4935750886874261, "grad_norm": 0.7004189491271973, "learning_rate": 0.0005, "loss": 1.9307336807250977, "step": 6261 }, { "epoch": 0.493653921955065, "grad_norm": 0.7232697010040283, "learning_rate": 0.0005, "loss": 2.056797742843628, "step": 6262 }, { "epoch": 0.493732755222704, "grad_norm": 0.6560314893722534, "learning_rate": 0.0005, "loss": 2.2747247219085693, "step": 6263 }, { "epoch": 0.4938115884903429, "grad_norm": 0.7261777520179749, "learning_rate": 0.0005, "loss": 1.6484918594360352, "step": 6264 }, { "epoch": 0.49389042175798187, "grad_norm": 0.7471207976341248, "learning_rate": 0.0005, "loss": 1.9830728769302368, "step": 6265 }, { "epoch": 0.49396925502562083, "grad_norm": 0.8014674186706543, "learning_rate": 0.0005, "loss": 2.290769338607788, "step": 6266 }, { "epoch": 0.49404808829325975, "grad_norm": 1.024863362312317, "learning_rate": 0.0005, "loss": 1.9814012050628662, "step": 6267 }, { "epoch": 0.4941269215608987, "grad_norm": 0.7817201614379883, "learning_rate": 0.0005, "loss": 1.9759547710418701, "step": 6268 }, { "epoch": 0.4942057548285376, "grad_norm": 0.8211438059806824, "learning_rate": 0.0005, "loss": 2.324204444885254, "step": 6269 }, { "epoch": 0.4942845880961766, "grad_norm": 0.8600711226463318, "learning_rate": 0.0005, "loss": 2.197200059890747, "step": 6270 }, { "epoch": 0.4943634213638155, "grad_norm": 0.71937096118927, "learning_rate": 0.0005, "loss": 1.7840123176574707, "step": 6271 }, { "epoch": 0.49444225463145447, "grad_norm": 0.6933271884918213, "learning_rate": 0.0005, "loss": 2.1262495517730713, "step": 6272 }, { "epoch": 0.49452108789909344, "grad_norm": 0.7266911864280701, "learning_rate": 0.0005, "loss": 1.8876657485961914, "step": 6273 }, { "epoch": 0.49459992116673235, "grad_norm": 0.7062659859657288, "learning_rate": 0.0005, "loss": 2.284414052963257, "step": 6274 }, { "epoch": 0.4946787544343713, "grad_norm": 0.8236706256866455, "learning_rate": 0.0005, "loss": 2.0455262660980225, "step": 6275 }, { "epoch": 0.4947575877020102, "grad_norm": 0.8896453976631165, "learning_rate": 0.0005, "loss": 2.112234354019165, "step": 6276 }, { "epoch": 0.4948364209696492, "grad_norm": 7.467959403991699, "learning_rate": 0.0005, "loss": 1.9016882181167603, "step": 6277 }, { "epoch": 0.49491525423728816, "grad_norm": 0.760385274887085, "learning_rate": 0.0005, "loss": 2.0107977390289307, "step": 6278 }, { "epoch": 0.49499408750492707, "grad_norm": 0.8526378870010376, "learning_rate": 0.0005, "loss": 2.1739320755004883, "step": 6279 }, { "epoch": 0.49507292077256604, "grad_norm": 0.8074870705604553, "learning_rate": 0.0005, "loss": 2.035083770751953, "step": 6280 }, { "epoch": 0.49515175404020495, "grad_norm": 0.7563087940216064, "learning_rate": 0.0005, "loss": 2.358790874481201, "step": 6281 }, { "epoch": 0.4952305873078439, "grad_norm": 0.6868990659713745, "learning_rate": 0.0005, "loss": 1.901199221611023, "step": 6282 }, { "epoch": 0.49530942057548283, "grad_norm": 0.7703144550323486, "learning_rate": 0.0005, "loss": 2.2600951194763184, "step": 6283 }, { "epoch": 0.4953882538431218, "grad_norm": 0.6985622048377991, "learning_rate": 0.0005, "loss": 1.7738232612609863, "step": 6284 }, { "epoch": 0.49546708711076076, "grad_norm": 0.8407099843025208, "learning_rate": 0.0005, "loss": 2.2664666175842285, "step": 6285 }, { "epoch": 0.4955459203783997, "grad_norm": 0.6465562582015991, "learning_rate": 0.0005, "loss": 1.8999241590499878, "step": 6286 }, { "epoch": 0.49562475364603864, "grad_norm": 0.8399977684020996, "learning_rate": 0.0005, "loss": 1.993030309677124, "step": 6287 }, { "epoch": 0.49570358691367755, "grad_norm": 0.7016215920448303, "learning_rate": 0.0005, "loss": 2.161210536956787, "step": 6288 }, { "epoch": 0.4957824201813165, "grad_norm": 0.6986257433891296, "learning_rate": 0.0005, "loss": 1.9505950212478638, "step": 6289 }, { "epoch": 0.4958612534489555, "grad_norm": 1.0350162982940674, "learning_rate": 0.0005, "loss": 2.1518383026123047, "step": 6290 }, { "epoch": 0.4959400867165944, "grad_norm": 0.7512651085853577, "learning_rate": 0.0005, "loss": 1.957564353942871, "step": 6291 }, { "epoch": 0.49601891998423336, "grad_norm": 0.6641942262649536, "learning_rate": 0.0005, "loss": 1.8454391956329346, "step": 6292 }, { "epoch": 0.4960977532518723, "grad_norm": 0.7139710187911987, "learning_rate": 0.0005, "loss": 2.0757884979248047, "step": 6293 }, { "epoch": 0.49617658651951124, "grad_norm": 0.5271614193916321, "learning_rate": 0.0005, "loss": 1.8962005376815796, "step": 6294 }, { "epoch": 0.49625541978715015, "grad_norm": 0.9764026403427124, "learning_rate": 0.0005, "loss": 1.7916889190673828, "step": 6295 }, { "epoch": 0.4963342530547891, "grad_norm": 0.6712157130241394, "learning_rate": 0.0005, "loss": 2.1329901218414307, "step": 6296 }, { "epoch": 0.4964130863224281, "grad_norm": 0.8252001404762268, "learning_rate": 0.0005, "loss": 2.1289329528808594, "step": 6297 }, { "epoch": 0.496491919590067, "grad_norm": 0.6874616742134094, "learning_rate": 0.0005, "loss": 2.2083053588867188, "step": 6298 }, { "epoch": 0.49657075285770597, "grad_norm": 0.7862403392791748, "learning_rate": 0.0005, "loss": 2.1200404167175293, "step": 6299 }, { "epoch": 0.4966495861253449, "grad_norm": 0.6154536008834839, "learning_rate": 0.0005, "loss": 1.9512629508972168, "step": 6300 }, { "epoch": 0.49672841939298384, "grad_norm": 0.8000721335411072, "learning_rate": 0.0005, "loss": 1.9070807695388794, "step": 6301 }, { "epoch": 0.49680725266062276, "grad_norm": 0.9721888303756714, "learning_rate": 0.0005, "loss": 2.072467565536499, "step": 6302 }, { "epoch": 0.4968860859282617, "grad_norm": 0.8207035660743713, "learning_rate": 0.0005, "loss": 2.176180839538574, "step": 6303 }, { "epoch": 0.4969649191959007, "grad_norm": 1.8556686639785767, "learning_rate": 0.0005, "loss": 2.102121353149414, "step": 6304 }, { "epoch": 0.4970437524635396, "grad_norm": 0.8230851888656616, "learning_rate": 0.0005, "loss": 2.025763750076294, "step": 6305 }, { "epoch": 0.49712258573117857, "grad_norm": 0.7880061268806458, "learning_rate": 0.0005, "loss": 2.094743251800537, "step": 6306 }, { "epoch": 0.4972014189988175, "grad_norm": 0.6176831126213074, "learning_rate": 0.0005, "loss": 1.799651861190796, "step": 6307 }, { "epoch": 0.49728025226645645, "grad_norm": 0.6828396320343018, "learning_rate": 0.0005, "loss": 2.017594575881958, "step": 6308 }, { "epoch": 0.4973590855340954, "grad_norm": 0.8427488207817078, "learning_rate": 0.0005, "loss": 2.042123317718506, "step": 6309 }, { "epoch": 0.4974379188017343, "grad_norm": 0.7861418128013611, "learning_rate": 0.0005, "loss": 2.2649741172790527, "step": 6310 }, { "epoch": 0.4975167520693733, "grad_norm": 0.6249112486839294, "learning_rate": 0.0005, "loss": 2.260132312774658, "step": 6311 }, { "epoch": 0.4975955853370122, "grad_norm": 0.7073264718055725, "learning_rate": 0.0005, "loss": 2.025623083114624, "step": 6312 }, { "epoch": 0.49767441860465117, "grad_norm": 0.8131003975868225, "learning_rate": 0.0005, "loss": 2.2315597534179688, "step": 6313 }, { "epoch": 0.4977532518722901, "grad_norm": 0.7649245262145996, "learning_rate": 0.0005, "loss": 2.124119281768799, "step": 6314 }, { "epoch": 0.49783208513992905, "grad_norm": 0.7364455461502075, "learning_rate": 0.0005, "loss": 2.154938220977783, "step": 6315 }, { "epoch": 0.497910918407568, "grad_norm": 0.5841551423072815, "learning_rate": 0.0005, "loss": 1.9208592176437378, "step": 6316 }, { "epoch": 0.4979897516752069, "grad_norm": 0.7747642397880554, "learning_rate": 0.0005, "loss": 2.081836700439453, "step": 6317 }, { "epoch": 0.4980685849428459, "grad_norm": 0.7386011481285095, "learning_rate": 0.0005, "loss": 2.1177384853363037, "step": 6318 }, { "epoch": 0.4981474182104848, "grad_norm": 0.7277663946151733, "learning_rate": 0.0005, "loss": 1.9717869758605957, "step": 6319 }, { "epoch": 0.49822625147812377, "grad_norm": 0.7401934862136841, "learning_rate": 0.0005, "loss": 1.998666524887085, "step": 6320 }, { "epoch": 0.49830508474576274, "grad_norm": 0.7558733224868774, "learning_rate": 0.0005, "loss": 2.2458834648132324, "step": 6321 }, { "epoch": 0.49838391801340165, "grad_norm": 0.5691420435905457, "learning_rate": 0.0005, "loss": 1.7720413208007812, "step": 6322 }, { "epoch": 0.4984627512810406, "grad_norm": 0.8290870189666748, "learning_rate": 0.0005, "loss": 1.708032250404358, "step": 6323 }, { "epoch": 0.49854158454867953, "grad_norm": 0.8034535050392151, "learning_rate": 0.0005, "loss": 2.379073143005371, "step": 6324 }, { "epoch": 0.4986204178163185, "grad_norm": 0.899395763874054, "learning_rate": 0.0005, "loss": 2.071485996246338, "step": 6325 }, { "epoch": 0.4986992510839574, "grad_norm": 0.7150313258171082, "learning_rate": 0.0005, "loss": 2.1682684421539307, "step": 6326 }, { "epoch": 0.4987780843515964, "grad_norm": 0.7420378923416138, "learning_rate": 0.0005, "loss": 1.9008088111877441, "step": 6327 }, { "epoch": 0.49885691761923534, "grad_norm": 0.8821872472763062, "learning_rate": 0.0005, "loss": 1.948421835899353, "step": 6328 }, { "epoch": 0.49893575088687425, "grad_norm": 0.7945290207862854, "learning_rate": 0.0005, "loss": 2.1114959716796875, "step": 6329 }, { "epoch": 0.4990145841545132, "grad_norm": 0.6937982439994812, "learning_rate": 0.0005, "loss": 1.8228026628494263, "step": 6330 }, { "epoch": 0.49909341742215213, "grad_norm": 0.727499783039093, "learning_rate": 0.0005, "loss": 2.0590946674346924, "step": 6331 }, { "epoch": 0.4991722506897911, "grad_norm": 0.8509840369224548, "learning_rate": 0.0005, "loss": 2.138643503189087, "step": 6332 }, { "epoch": 0.49925108395743, "grad_norm": 0.7261499762535095, "learning_rate": 0.0005, "loss": 2.287153720855713, "step": 6333 }, { "epoch": 0.499329917225069, "grad_norm": 0.6118324398994446, "learning_rate": 0.0005, "loss": 1.8133260011672974, "step": 6334 }, { "epoch": 0.49940875049270794, "grad_norm": 1.1594703197479248, "learning_rate": 0.0005, "loss": 2.0613441467285156, "step": 6335 }, { "epoch": 0.49948758376034685, "grad_norm": 0.6913710832595825, "learning_rate": 0.0005, "loss": 2.341956377029419, "step": 6336 }, { "epoch": 0.4995664170279858, "grad_norm": 0.7829994559288025, "learning_rate": 0.0005, "loss": 2.3549563884735107, "step": 6337 }, { "epoch": 0.49964525029562473, "grad_norm": 0.9092851281166077, "learning_rate": 0.0005, "loss": 2.225888729095459, "step": 6338 }, { "epoch": 0.4997240835632637, "grad_norm": 0.9123308062553406, "learning_rate": 0.0005, "loss": 2.051058053970337, "step": 6339 }, { "epoch": 0.49980291683090267, "grad_norm": 0.7787347435951233, "learning_rate": 0.0005, "loss": 2.287963390350342, "step": 6340 }, { "epoch": 0.4998817500985416, "grad_norm": 0.6865640878677368, "learning_rate": 0.0005, "loss": 1.998483419418335, "step": 6341 }, { "epoch": 0.49996058336618054, "grad_norm": 0.7934045195579529, "learning_rate": 0.0005, "loss": 2.049726724624634, "step": 6342 }, { "epoch": 0.5000394166338195, "grad_norm": 0.5865219235420227, "learning_rate": 0.0005, "loss": 1.8948585987091064, "step": 6343 }, { "epoch": 0.5001182499014584, "grad_norm": 0.6414729952812195, "learning_rate": 0.0005, "loss": 2.0507514476776123, "step": 6344 }, { "epoch": 0.5001970831690974, "grad_norm": 0.7310541272163391, "learning_rate": 0.0005, "loss": 2.213383197784424, "step": 6345 }, { "epoch": 0.5002759164367363, "grad_norm": 0.7082070708274841, "learning_rate": 0.0005, "loss": 2.109410047531128, "step": 6346 }, { "epoch": 0.5003547497043752, "grad_norm": 1.1153918504714966, "learning_rate": 0.0005, "loss": 2.0901618003845215, "step": 6347 }, { "epoch": 0.5004335829720142, "grad_norm": 0.6534603834152222, "learning_rate": 0.0005, "loss": 2.193077564239502, "step": 6348 }, { "epoch": 0.5005124162396531, "grad_norm": 0.7538958787918091, "learning_rate": 0.0005, "loss": 2.233980655670166, "step": 6349 }, { "epoch": 0.5005912495072921, "grad_norm": 0.768614649772644, "learning_rate": 0.0005, "loss": 2.125157594680786, "step": 6350 }, { "epoch": 0.500670082774931, "grad_norm": 0.7231115102767944, "learning_rate": 0.0005, "loss": 2.1358065605163574, "step": 6351 }, { "epoch": 0.50074891604257, "grad_norm": 0.8335344195365906, "learning_rate": 0.0005, "loss": 2.0062355995178223, "step": 6352 }, { "epoch": 0.5008277493102089, "grad_norm": 1.0280754566192627, "learning_rate": 0.0005, "loss": 2.2726876735687256, "step": 6353 }, { "epoch": 0.5009065825778478, "grad_norm": 0.5762625336647034, "learning_rate": 0.0005, "loss": 1.9378596544265747, "step": 6354 }, { "epoch": 0.5009854158454868, "grad_norm": 0.706963062286377, "learning_rate": 0.0005, "loss": 1.8329319953918457, "step": 6355 }, { "epoch": 0.5010642491131257, "grad_norm": 0.7665640115737915, "learning_rate": 0.0005, "loss": 2.0063791275024414, "step": 6356 }, { "epoch": 0.5011430823807647, "grad_norm": 0.8306924700737, "learning_rate": 0.0005, "loss": 1.9590842723846436, "step": 6357 }, { "epoch": 0.5012219156484036, "grad_norm": 0.7051384449005127, "learning_rate": 0.0005, "loss": 2.0514631271362305, "step": 6358 }, { "epoch": 0.5013007489160426, "grad_norm": 0.6536523699760437, "learning_rate": 0.0005, "loss": 2.022451877593994, "step": 6359 }, { "epoch": 0.5013795821836815, "grad_norm": 0.9997424483299255, "learning_rate": 0.0005, "loss": 2.2154250144958496, "step": 6360 }, { "epoch": 0.5014584154513204, "grad_norm": 0.6435181498527527, "learning_rate": 0.0005, "loss": 2.299168586730957, "step": 6361 }, { "epoch": 0.5015372487189594, "grad_norm": 0.5928472280502319, "learning_rate": 0.0005, "loss": 2.0532288551330566, "step": 6362 }, { "epoch": 0.5016160819865984, "grad_norm": 0.7839056849479675, "learning_rate": 0.0005, "loss": 2.008486270904541, "step": 6363 }, { "epoch": 0.5016949152542373, "grad_norm": 0.9207819104194641, "learning_rate": 0.0005, "loss": 1.9839117527008057, "step": 6364 }, { "epoch": 0.5017737485218763, "grad_norm": 0.6752997040748596, "learning_rate": 0.0005, "loss": 2.1294915676116943, "step": 6365 }, { "epoch": 0.5018525817895152, "grad_norm": 0.6340699791908264, "learning_rate": 0.0005, "loss": 2.188979387283325, "step": 6366 }, { "epoch": 0.5019314150571541, "grad_norm": 0.6578187942504883, "learning_rate": 0.0005, "loss": 2.234154462814331, "step": 6367 }, { "epoch": 0.502010248324793, "grad_norm": 0.7128844261169434, "learning_rate": 0.0005, "loss": 1.780545711517334, "step": 6368 }, { "epoch": 0.502089081592432, "grad_norm": 0.6708013415336609, "learning_rate": 0.0005, "loss": 1.6763067245483398, "step": 6369 }, { "epoch": 0.502167914860071, "grad_norm": 1.0215215682983398, "learning_rate": 0.0005, "loss": 2.0589659214019775, "step": 6370 }, { "epoch": 0.5022467481277099, "grad_norm": 1.0206868648529053, "learning_rate": 0.0005, "loss": 2.1375019550323486, "step": 6371 }, { "epoch": 0.5023255813953489, "grad_norm": 0.8559842705726624, "learning_rate": 0.0005, "loss": 2.170856237411499, "step": 6372 }, { "epoch": 0.5024044146629878, "grad_norm": 0.7436604499816895, "learning_rate": 0.0005, "loss": 1.9257571697235107, "step": 6373 }, { "epoch": 0.5024832479306267, "grad_norm": 1.8505698442459106, "learning_rate": 0.0005, "loss": 2.066542148590088, "step": 6374 }, { "epoch": 0.5025620811982656, "grad_norm": 0.6505256295204163, "learning_rate": 0.0005, "loss": 2.0999810695648193, "step": 6375 }, { "epoch": 0.5026409144659046, "grad_norm": 0.679142415523529, "learning_rate": 0.0005, "loss": 1.8344486951828003, "step": 6376 }, { "epoch": 0.5027197477335436, "grad_norm": 0.8165237903594971, "learning_rate": 0.0005, "loss": 2.092489004135132, "step": 6377 }, { "epoch": 0.5027985810011825, "grad_norm": 1.1269153356552124, "learning_rate": 0.0005, "loss": 2.2530198097229004, "step": 6378 }, { "epoch": 0.5028774142688215, "grad_norm": 0.7506309747695923, "learning_rate": 0.0005, "loss": 1.977217197418213, "step": 6379 }, { "epoch": 0.5029562475364604, "grad_norm": 0.7123488783836365, "learning_rate": 0.0005, "loss": 2.0464115142822266, "step": 6380 }, { "epoch": 0.5030350808040993, "grad_norm": 0.9371535181999207, "learning_rate": 0.0005, "loss": 1.8143401145935059, "step": 6381 }, { "epoch": 0.5031139140717382, "grad_norm": 0.775357723236084, "learning_rate": 0.0005, "loss": 2.1833863258361816, "step": 6382 }, { "epoch": 0.5031927473393772, "grad_norm": 0.6782098412513733, "learning_rate": 0.0005, "loss": 1.9784992933273315, "step": 6383 }, { "epoch": 0.5032715806070162, "grad_norm": 0.9258432388305664, "learning_rate": 0.0005, "loss": 2.2265756130218506, "step": 6384 }, { "epoch": 0.5033504138746551, "grad_norm": 0.6435133814811707, "learning_rate": 0.0005, "loss": 1.9575648307800293, "step": 6385 }, { "epoch": 0.5034292471422941, "grad_norm": 0.6887959241867065, "learning_rate": 0.0005, "loss": 2.2173564434051514, "step": 6386 }, { "epoch": 0.503508080409933, "grad_norm": 0.7579021453857422, "learning_rate": 0.0005, "loss": 2.086916923522949, "step": 6387 }, { "epoch": 0.5035869136775719, "grad_norm": 0.684141218662262, "learning_rate": 0.0005, "loss": 2.08553409576416, "step": 6388 }, { "epoch": 0.5036657469452108, "grad_norm": 0.7137749791145325, "learning_rate": 0.0005, "loss": 1.797681450843811, "step": 6389 }, { "epoch": 0.5037445802128498, "grad_norm": 0.754926860332489, "learning_rate": 0.0005, "loss": 1.9849532842636108, "step": 6390 }, { "epoch": 0.5038234134804888, "grad_norm": 0.7076828479766846, "learning_rate": 0.0005, "loss": 1.9526362419128418, "step": 6391 }, { "epoch": 0.5039022467481277, "grad_norm": 0.595946192741394, "learning_rate": 0.0005, "loss": 1.899662733078003, "step": 6392 }, { "epoch": 0.5039810800157667, "grad_norm": 0.7654907703399658, "learning_rate": 0.0005, "loss": 2.0764079093933105, "step": 6393 }, { "epoch": 0.5040599132834056, "grad_norm": 0.6820142269134521, "learning_rate": 0.0005, "loss": 2.347886562347412, "step": 6394 }, { "epoch": 0.5041387465510445, "grad_norm": 0.6741540431976318, "learning_rate": 0.0005, "loss": 1.8892943859100342, "step": 6395 }, { "epoch": 0.5042175798186835, "grad_norm": 0.8393210172653198, "learning_rate": 0.0005, "loss": 1.9377210140228271, "step": 6396 }, { "epoch": 0.5042964130863224, "grad_norm": 0.8135398626327515, "learning_rate": 0.0005, "loss": 1.9055863618850708, "step": 6397 }, { "epoch": 0.5043752463539614, "grad_norm": 0.7960635423660278, "learning_rate": 0.0005, "loss": 2.1756222248077393, "step": 6398 }, { "epoch": 0.5044540796216003, "grad_norm": 0.5657373070716858, "learning_rate": 0.0005, "loss": 1.7219388484954834, "step": 6399 }, { "epoch": 0.5045329128892393, "grad_norm": 0.7911690473556519, "learning_rate": 0.0005, "loss": 2.04524564743042, "step": 6400 }, { "epoch": 0.5046117461568782, "grad_norm": 0.6405468583106995, "learning_rate": 0.0005, "loss": 2.1313061714172363, "step": 6401 }, { "epoch": 0.5046905794245171, "grad_norm": 0.7510327696800232, "learning_rate": 0.0005, "loss": 2.0583319664001465, "step": 6402 }, { "epoch": 0.5047694126921561, "grad_norm": 0.6676178574562073, "learning_rate": 0.0005, "loss": 1.867452621459961, "step": 6403 }, { "epoch": 0.504848245959795, "grad_norm": 0.8123084306716919, "learning_rate": 0.0005, "loss": 2.035755157470703, "step": 6404 }, { "epoch": 0.504927079227434, "grad_norm": 0.9796284437179565, "learning_rate": 0.0005, "loss": 1.8395427465438843, "step": 6405 }, { "epoch": 0.5050059124950729, "grad_norm": 1.0315648317337036, "learning_rate": 0.0005, "loss": 2.2956924438476562, "step": 6406 }, { "epoch": 0.5050847457627119, "grad_norm": 0.6646860837936401, "learning_rate": 0.0005, "loss": 2.0416159629821777, "step": 6407 }, { "epoch": 0.5051635790303508, "grad_norm": 0.7247145771980286, "learning_rate": 0.0005, "loss": 1.8903727531433105, "step": 6408 }, { "epoch": 0.5052424122979897, "grad_norm": 0.6330117583274841, "learning_rate": 0.0005, "loss": 1.8395966291427612, "step": 6409 }, { "epoch": 0.5053212455656287, "grad_norm": 0.7837880849838257, "learning_rate": 0.0005, "loss": 2.057281255722046, "step": 6410 }, { "epoch": 0.5054000788332677, "grad_norm": 0.5427443385124207, "learning_rate": 0.0005, "loss": 1.831387996673584, "step": 6411 }, { "epoch": 0.5054789121009066, "grad_norm": 0.6838268637657166, "learning_rate": 0.0005, "loss": 2.130528450012207, "step": 6412 }, { "epoch": 0.5055577453685455, "grad_norm": 0.8612074851989746, "learning_rate": 0.0005, "loss": 2.183145523071289, "step": 6413 }, { "epoch": 0.5056365786361845, "grad_norm": 0.733439564704895, "learning_rate": 0.0005, "loss": 1.9420499801635742, "step": 6414 }, { "epoch": 0.5057154119038234, "grad_norm": 0.6833911538124084, "learning_rate": 0.0005, "loss": 2.1106040477752686, "step": 6415 }, { "epoch": 0.5057942451714623, "grad_norm": 0.7249692678451538, "learning_rate": 0.0005, "loss": 2.072701930999756, "step": 6416 }, { "epoch": 0.5058730784391013, "grad_norm": 0.9143070578575134, "learning_rate": 0.0005, "loss": 2.2802059650421143, "step": 6417 }, { "epoch": 0.5059519117067403, "grad_norm": 1.6217948198318481, "learning_rate": 0.0005, "loss": 2.3108222484588623, "step": 6418 }, { "epoch": 0.5060307449743792, "grad_norm": 0.6026843786239624, "learning_rate": 0.0005, "loss": 1.8186557292938232, "step": 6419 }, { "epoch": 0.5061095782420181, "grad_norm": 1.148271083831787, "learning_rate": 0.0005, "loss": 2.2475857734680176, "step": 6420 }, { "epoch": 0.5061884115096571, "grad_norm": 0.692494809627533, "learning_rate": 0.0005, "loss": 1.825892448425293, "step": 6421 }, { "epoch": 0.506267244777296, "grad_norm": 0.6483763456344604, "learning_rate": 0.0005, "loss": 2.224853992462158, "step": 6422 }, { "epoch": 0.5063460780449349, "grad_norm": 0.8824981451034546, "learning_rate": 0.0005, "loss": 2.127112865447998, "step": 6423 }, { "epoch": 0.5064249113125739, "grad_norm": 0.6847624182701111, "learning_rate": 0.0005, "loss": 1.9441704750061035, "step": 6424 }, { "epoch": 0.5065037445802129, "grad_norm": 0.978126049041748, "learning_rate": 0.0005, "loss": 1.8173784017562866, "step": 6425 }, { "epoch": 0.5065825778478518, "grad_norm": 0.9190197587013245, "learning_rate": 0.0005, "loss": 2.5056378841400146, "step": 6426 }, { "epoch": 0.5066614111154908, "grad_norm": 0.7262889742851257, "learning_rate": 0.0005, "loss": 2.1594672203063965, "step": 6427 }, { "epoch": 0.5067402443831297, "grad_norm": 0.8474106788635254, "learning_rate": 0.0005, "loss": 1.9265997409820557, "step": 6428 }, { "epoch": 0.5068190776507686, "grad_norm": 1.0677337646484375, "learning_rate": 0.0005, "loss": 2.1264564990997314, "step": 6429 }, { "epoch": 0.5068979109184075, "grad_norm": 0.951948881149292, "learning_rate": 0.0005, "loss": 2.062106132507324, "step": 6430 }, { "epoch": 0.5069767441860465, "grad_norm": 1.0198986530303955, "learning_rate": 0.0005, "loss": 2.07450008392334, "step": 6431 }, { "epoch": 0.5070555774536855, "grad_norm": 0.7838196158409119, "learning_rate": 0.0005, "loss": 2.0379533767700195, "step": 6432 }, { "epoch": 0.5071344107213244, "grad_norm": 0.7141558527946472, "learning_rate": 0.0005, "loss": 2.344402313232422, "step": 6433 }, { "epoch": 0.5072132439889634, "grad_norm": 0.7533993124961853, "learning_rate": 0.0005, "loss": 2.196699857711792, "step": 6434 }, { "epoch": 0.5072920772566023, "grad_norm": 0.8598559498786926, "learning_rate": 0.0005, "loss": 1.892289400100708, "step": 6435 }, { "epoch": 0.5073709105242412, "grad_norm": 0.566533088684082, "learning_rate": 0.0005, "loss": 1.6870956420898438, "step": 6436 }, { "epoch": 0.5074497437918801, "grad_norm": 0.6587040424346924, "learning_rate": 0.0005, "loss": 2.105604887008667, "step": 6437 }, { "epoch": 0.5075285770595191, "grad_norm": 0.6713230609893799, "learning_rate": 0.0005, "loss": 1.8810880184173584, "step": 6438 }, { "epoch": 0.5076074103271581, "grad_norm": 0.6981633305549622, "learning_rate": 0.0005, "loss": 2.3118653297424316, "step": 6439 }, { "epoch": 0.507686243594797, "grad_norm": 0.7724265456199646, "learning_rate": 0.0005, "loss": 2.2289657592773438, "step": 6440 }, { "epoch": 0.507765076862436, "grad_norm": 0.9056540727615356, "learning_rate": 0.0005, "loss": 1.9085586071014404, "step": 6441 }, { "epoch": 0.5078439101300749, "grad_norm": 0.8715552091598511, "learning_rate": 0.0005, "loss": 1.727980613708496, "step": 6442 }, { "epoch": 0.5079227433977138, "grad_norm": 0.911389946937561, "learning_rate": 0.0005, "loss": 1.670344352722168, "step": 6443 }, { "epoch": 0.5080015766653527, "grad_norm": 0.8272135853767395, "learning_rate": 0.0005, "loss": 2.3996589183807373, "step": 6444 }, { "epoch": 0.5080804099329918, "grad_norm": 0.9115672707557678, "learning_rate": 0.0005, "loss": 1.9940338134765625, "step": 6445 }, { "epoch": 0.5081592432006307, "grad_norm": 0.8198719620704651, "learning_rate": 0.0005, "loss": 2.245227813720703, "step": 6446 }, { "epoch": 0.5082380764682696, "grad_norm": 0.6997912526130676, "learning_rate": 0.0005, "loss": 2.3612594604492188, "step": 6447 }, { "epoch": 0.5083169097359086, "grad_norm": 0.769831120967865, "learning_rate": 0.0005, "loss": 2.268742084503174, "step": 6448 }, { "epoch": 0.5083957430035475, "grad_norm": 1.1158438920974731, "learning_rate": 0.0005, "loss": 2.0878031253814697, "step": 6449 }, { "epoch": 0.5084745762711864, "grad_norm": 1.287609338760376, "learning_rate": 0.0005, "loss": 2.5590476989746094, "step": 6450 }, { "epoch": 0.5085534095388253, "grad_norm": 0.828968346118927, "learning_rate": 0.0005, "loss": 2.0086050033569336, "step": 6451 }, { "epoch": 0.5086322428064644, "grad_norm": 0.6147163510322571, "learning_rate": 0.0005, "loss": 1.9042946100234985, "step": 6452 }, { "epoch": 0.5087110760741033, "grad_norm": 0.780308723449707, "learning_rate": 0.0005, "loss": 1.8911641836166382, "step": 6453 }, { "epoch": 0.5087899093417422, "grad_norm": 0.8993359804153442, "learning_rate": 0.0005, "loss": 1.8766965866088867, "step": 6454 }, { "epoch": 0.5088687426093812, "grad_norm": 0.6211287379264832, "learning_rate": 0.0005, "loss": 2.120293140411377, "step": 6455 }, { "epoch": 0.5089475758770201, "grad_norm": 1.0457509756088257, "learning_rate": 0.0005, "loss": 1.9019787311553955, "step": 6456 }, { "epoch": 0.509026409144659, "grad_norm": 0.8484043478965759, "learning_rate": 0.0005, "loss": 1.9209926128387451, "step": 6457 }, { "epoch": 0.509105242412298, "grad_norm": 0.6760343313217163, "learning_rate": 0.0005, "loss": 2.176349639892578, "step": 6458 }, { "epoch": 0.509184075679937, "grad_norm": 0.7039400339126587, "learning_rate": 0.0005, "loss": 2.129697799682617, "step": 6459 }, { "epoch": 0.5092629089475759, "grad_norm": 0.7758392095565796, "learning_rate": 0.0005, "loss": 1.9330781698226929, "step": 6460 }, { "epoch": 0.5093417422152148, "grad_norm": 0.7104073762893677, "learning_rate": 0.0005, "loss": 2.083528518676758, "step": 6461 }, { "epoch": 0.5094205754828538, "grad_norm": 0.7994508743286133, "learning_rate": 0.0005, "loss": 1.8626646995544434, "step": 6462 }, { "epoch": 0.5094994087504927, "grad_norm": 0.8247259855270386, "learning_rate": 0.0005, "loss": 2.23394775390625, "step": 6463 }, { "epoch": 0.5095782420181316, "grad_norm": 0.7931819558143616, "learning_rate": 0.0005, "loss": 2.3264687061309814, "step": 6464 }, { "epoch": 0.5096570752857706, "grad_norm": 0.6970381736755371, "learning_rate": 0.0005, "loss": 2.1444246768951416, "step": 6465 }, { "epoch": 0.5097359085534096, "grad_norm": 1.0586477518081665, "learning_rate": 0.0005, "loss": 2.1101913452148438, "step": 6466 }, { "epoch": 0.5098147418210485, "grad_norm": 0.7921530604362488, "learning_rate": 0.0005, "loss": 2.282803535461426, "step": 6467 }, { "epoch": 0.5098935750886874, "grad_norm": 0.7527603507041931, "learning_rate": 0.0005, "loss": 2.04876708984375, "step": 6468 }, { "epoch": 0.5099724083563264, "grad_norm": 0.7679764032363892, "learning_rate": 0.0005, "loss": 2.178783655166626, "step": 6469 }, { "epoch": 0.5100512416239653, "grad_norm": 0.7692520618438721, "learning_rate": 0.0005, "loss": 1.9857127666473389, "step": 6470 }, { "epoch": 0.5101300748916042, "grad_norm": 0.968051552772522, "learning_rate": 0.0005, "loss": 2.1008460521698, "step": 6471 }, { "epoch": 0.5102089081592432, "grad_norm": 0.744173526763916, "learning_rate": 0.0005, "loss": 1.8919938802719116, "step": 6472 }, { "epoch": 0.5102877414268822, "grad_norm": 0.7904074788093567, "learning_rate": 0.0005, "loss": 1.9545955657958984, "step": 6473 }, { "epoch": 0.5103665746945211, "grad_norm": 1.0641188621520996, "learning_rate": 0.0005, "loss": 1.7511415481567383, "step": 6474 }, { "epoch": 0.51044540796216, "grad_norm": 0.655511200428009, "learning_rate": 0.0005, "loss": 2.397031307220459, "step": 6475 }, { "epoch": 0.510524241229799, "grad_norm": 0.6838549971580505, "learning_rate": 0.0005, "loss": 1.9496750831604004, "step": 6476 }, { "epoch": 0.5106030744974379, "grad_norm": 0.7053768038749695, "learning_rate": 0.0005, "loss": 1.867710828781128, "step": 6477 }, { "epoch": 0.5106819077650768, "grad_norm": 0.7898997068405151, "learning_rate": 0.0005, "loss": 2.2379655838012695, "step": 6478 }, { "epoch": 0.5107607410327158, "grad_norm": 0.657128632068634, "learning_rate": 0.0005, "loss": 1.815213918685913, "step": 6479 }, { "epoch": 0.5108395743003548, "grad_norm": 0.7434514164924622, "learning_rate": 0.0005, "loss": 2.1711652278900146, "step": 6480 }, { "epoch": 0.5109184075679937, "grad_norm": 0.8848334550857544, "learning_rate": 0.0005, "loss": 2.176983594894409, "step": 6481 }, { "epoch": 0.5109972408356326, "grad_norm": 0.7859110832214355, "learning_rate": 0.0005, "loss": 2.031428575515747, "step": 6482 }, { "epoch": 0.5110760741032716, "grad_norm": 0.6719304323196411, "learning_rate": 0.0005, "loss": 2.115111827850342, "step": 6483 }, { "epoch": 0.5111549073709105, "grad_norm": 0.6789891719818115, "learning_rate": 0.0005, "loss": 1.9587719440460205, "step": 6484 }, { "epoch": 0.5112337406385494, "grad_norm": 0.8525139093399048, "learning_rate": 0.0005, "loss": 2.2538185119628906, "step": 6485 }, { "epoch": 0.5113125739061885, "grad_norm": 0.680577278137207, "learning_rate": 0.0005, "loss": 2.228508472442627, "step": 6486 }, { "epoch": 0.5113914071738274, "grad_norm": 0.813039243221283, "learning_rate": 0.0005, "loss": 2.2510385513305664, "step": 6487 }, { "epoch": 0.5114702404414663, "grad_norm": 0.8591915369033813, "learning_rate": 0.0005, "loss": 2.1122422218322754, "step": 6488 }, { "epoch": 0.5115490737091053, "grad_norm": 0.8084139227867126, "learning_rate": 0.0005, "loss": 2.010197162628174, "step": 6489 }, { "epoch": 0.5116279069767442, "grad_norm": 0.7313998937606812, "learning_rate": 0.0005, "loss": 1.727496862411499, "step": 6490 }, { "epoch": 0.5117067402443831, "grad_norm": 0.9578163027763367, "learning_rate": 0.0005, "loss": 1.8366427421569824, "step": 6491 }, { "epoch": 0.511785573512022, "grad_norm": 0.9313250184059143, "learning_rate": 0.0005, "loss": 2.103240489959717, "step": 6492 }, { "epoch": 0.511864406779661, "grad_norm": 0.7813620567321777, "learning_rate": 0.0005, "loss": 2.094430446624756, "step": 6493 }, { "epoch": 0.5119432400473, "grad_norm": 0.6871518492698669, "learning_rate": 0.0005, "loss": 1.946660041809082, "step": 6494 }, { "epoch": 0.5120220733149389, "grad_norm": 1.0529229640960693, "learning_rate": 0.0005, "loss": 1.9811331033706665, "step": 6495 }, { "epoch": 0.5121009065825779, "grad_norm": 0.7847253084182739, "learning_rate": 0.0005, "loss": 2.13299822807312, "step": 6496 }, { "epoch": 0.5121797398502168, "grad_norm": 2.014125108718872, "learning_rate": 0.0005, "loss": 2.4834628105163574, "step": 6497 }, { "epoch": 0.5122585731178557, "grad_norm": 0.7053129076957703, "learning_rate": 0.0005, "loss": 1.8928303718566895, "step": 6498 }, { "epoch": 0.5123374063854946, "grad_norm": 0.9460268616676331, "learning_rate": 0.0005, "loss": 2.300698757171631, "step": 6499 }, { "epoch": 0.5124162396531337, "grad_norm": 0.6193488240242004, "learning_rate": 0.0005, "loss": 2.2762627601623535, "step": 6500 }, { "epoch": 0.5124950729207726, "grad_norm": 0.7640928626060486, "learning_rate": 0.0005, "loss": 2.0227904319763184, "step": 6501 }, { "epoch": 0.5125739061884115, "grad_norm": 0.752444326877594, "learning_rate": 0.0005, "loss": 2.0720887184143066, "step": 6502 }, { "epoch": 0.5126527394560505, "grad_norm": 0.9607972502708435, "learning_rate": 0.0005, "loss": 1.8835843801498413, "step": 6503 }, { "epoch": 0.5127315727236894, "grad_norm": 0.7589804530143738, "learning_rate": 0.0005, "loss": 2.5115270614624023, "step": 6504 }, { "epoch": 0.5128104059913283, "grad_norm": 0.8771670460700989, "learning_rate": 0.0005, "loss": 2.0264291763305664, "step": 6505 }, { "epoch": 0.5128892392589672, "grad_norm": 0.9188981652259827, "learning_rate": 0.0005, "loss": 2.2616350650787354, "step": 6506 }, { "epoch": 0.5129680725266063, "grad_norm": 0.7134186625480652, "learning_rate": 0.0005, "loss": 2.39163875579834, "step": 6507 }, { "epoch": 0.5130469057942452, "grad_norm": 0.7081385850906372, "learning_rate": 0.0005, "loss": 2.2929487228393555, "step": 6508 }, { "epoch": 0.5131257390618841, "grad_norm": 0.8276671171188354, "learning_rate": 0.0005, "loss": 2.2609293460845947, "step": 6509 }, { "epoch": 0.5132045723295231, "grad_norm": 0.500478208065033, "learning_rate": 0.0005, "loss": 1.4428225755691528, "step": 6510 }, { "epoch": 0.513283405597162, "grad_norm": 0.7282401919364929, "learning_rate": 0.0005, "loss": 1.8280491828918457, "step": 6511 }, { "epoch": 0.5133622388648009, "grad_norm": 0.7283053398132324, "learning_rate": 0.0005, "loss": 1.9659054279327393, "step": 6512 }, { "epoch": 0.5134410721324398, "grad_norm": 0.7698034048080444, "learning_rate": 0.0005, "loss": 1.9688098430633545, "step": 6513 }, { "epoch": 0.5135199054000789, "grad_norm": 0.6783542633056641, "learning_rate": 0.0005, "loss": 2.2612292766571045, "step": 6514 }, { "epoch": 0.5135987386677178, "grad_norm": 0.7465468049049377, "learning_rate": 0.0005, "loss": 2.122372627258301, "step": 6515 }, { "epoch": 0.5136775719353567, "grad_norm": 0.6949198246002197, "learning_rate": 0.0005, "loss": 1.8365299701690674, "step": 6516 }, { "epoch": 0.5137564052029957, "grad_norm": 0.623304009437561, "learning_rate": 0.0005, "loss": 2.1621246337890625, "step": 6517 }, { "epoch": 0.5138352384706346, "grad_norm": 0.6647234559059143, "learning_rate": 0.0005, "loss": 1.926844596862793, "step": 6518 }, { "epoch": 0.5139140717382735, "grad_norm": 0.7328523993492126, "learning_rate": 0.0005, "loss": 1.8817185163497925, "step": 6519 }, { "epoch": 0.5139929050059125, "grad_norm": 0.7573676109313965, "learning_rate": 0.0005, "loss": 2.076643466949463, "step": 6520 }, { "epoch": 0.5140717382735515, "grad_norm": 0.7497263550758362, "learning_rate": 0.0005, "loss": 2.1041676998138428, "step": 6521 }, { "epoch": 0.5141505715411904, "grad_norm": 0.7080767750740051, "learning_rate": 0.0005, "loss": 1.940640926361084, "step": 6522 }, { "epoch": 0.5142294048088293, "grad_norm": 0.6306454539299011, "learning_rate": 0.0005, "loss": 2.255707025527954, "step": 6523 }, { "epoch": 0.5143082380764683, "grad_norm": 0.6747228503227234, "learning_rate": 0.0005, "loss": 1.7923911809921265, "step": 6524 }, { "epoch": 0.5143870713441072, "grad_norm": 0.6766449809074402, "learning_rate": 0.0005, "loss": 2.2202982902526855, "step": 6525 }, { "epoch": 0.5144659046117461, "grad_norm": 0.970451831817627, "learning_rate": 0.0005, "loss": 2.4072837829589844, "step": 6526 }, { "epoch": 0.5145447378793852, "grad_norm": 0.7609847784042358, "learning_rate": 0.0005, "loss": 1.9402308464050293, "step": 6527 }, { "epoch": 0.5146235711470241, "grad_norm": 0.7848509550094604, "learning_rate": 0.0005, "loss": 2.140730381011963, "step": 6528 }, { "epoch": 0.514702404414663, "grad_norm": 0.8825620412826538, "learning_rate": 0.0005, "loss": 1.9850609302520752, "step": 6529 }, { "epoch": 0.5147812376823019, "grad_norm": 0.6968220472335815, "learning_rate": 0.0005, "loss": 1.8452465534210205, "step": 6530 }, { "epoch": 0.5148600709499409, "grad_norm": 0.6550359129905701, "learning_rate": 0.0005, "loss": 1.7846980094909668, "step": 6531 }, { "epoch": 0.5149389042175798, "grad_norm": 0.58058762550354, "learning_rate": 0.0005, "loss": 2.046278476715088, "step": 6532 }, { "epoch": 0.5150177374852187, "grad_norm": 0.5943891406059265, "learning_rate": 0.0005, "loss": 1.9383761882781982, "step": 6533 }, { "epoch": 0.5150965707528578, "grad_norm": 0.8825696110725403, "learning_rate": 0.0005, "loss": 2.103050708770752, "step": 6534 }, { "epoch": 0.5151754040204967, "grad_norm": 0.8268117904663086, "learning_rate": 0.0005, "loss": 2.153047800064087, "step": 6535 }, { "epoch": 0.5152542372881356, "grad_norm": 0.591609001159668, "learning_rate": 0.0005, "loss": 1.9985519647598267, "step": 6536 }, { "epoch": 0.5153330705557745, "grad_norm": 0.8185657858848572, "learning_rate": 0.0005, "loss": 2.1092710494995117, "step": 6537 }, { "epoch": 0.5154119038234135, "grad_norm": 0.831178605556488, "learning_rate": 0.0005, "loss": 2.3378665447235107, "step": 6538 }, { "epoch": 0.5154907370910524, "grad_norm": 0.7170799970626831, "learning_rate": 0.0005, "loss": 2.424027919769287, "step": 6539 }, { "epoch": 0.5155695703586913, "grad_norm": 0.705841064453125, "learning_rate": 0.0005, "loss": 1.868146538734436, "step": 6540 }, { "epoch": 0.5156484036263304, "grad_norm": 0.7801923155784607, "learning_rate": 0.0005, "loss": 2.2242043018341064, "step": 6541 }, { "epoch": 0.5157272368939693, "grad_norm": 0.7460983395576477, "learning_rate": 0.0005, "loss": 2.1594762802124023, "step": 6542 }, { "epoch": 0.5158060701616082, "grad_norm": 0.6330990195274353, "learning_rate": 0.0005, "loss": 2.0861964225769043, "step": 6543 }, { "epoch": 0.5158849034292471, "grad_norm": 0.6362169981002808, "learning_rate": 0.0005, "loss": 1.946758508682251, "step": 6544 }, { "epoch": 0.5159637366968861, "grad_norm": 0.7401531934738159, "learning_rate": 0.0005, "loss": 1.901939034461975, "step": 6545 }, { "epoch": 0.516042569964525, "grad_norm": 0.6541084051132202, "learning_rate": 0.0005, "loss": 2.011183261871338, "step": 6546 }, { "epoch": 0.5161214032321639, "grad_norm": 0.7914556264877319, "learning_rate": 0.0005, "loss": 2.1774516105651855, "step": 6547 }, { "epoch": 0.516200236499803, "grad_norm": 0.6168916821479797, "learning_rate": 0.0005, "loss": 1.7780702114105225, "step": 6548 }, { "epoch": 0.5162790697674419, "grad_norm": 1.1313235759735107, "learning_rate": 0.0005, "loss": 1.8399767875671387, "step": 6549 }, { "epoch": 0.5163579030350808, "grad_norm": 0.8432016372680664, "learning_rate": 0.0005, "loss": 2.1267452239990234, "step": 6550 }, { "epoch": 0.5164367363027198, "grad_norm": 0.6126102209091187, "learning_rate": 0.0005, "loss": 1.9984171390533447, "step": 6551 }, { "epoch": 0.5165155695703587, "grad_norm": 0.6948968172073364, "learning_rate": 0.0005, "loss": 2.037381887435913, "step": 6552 }, { "epoch": 0.5165944028379976, "grad_norm": 0.8441342711448669, "learning_rate": 0.0005, "loss": 2.224574565887451, "step": 6553 }, { "epoch": 0.5166732361056365, "grad_norm": 1.5562207698822021, "learning_rate": 0.0005, "loss": 2.2572922706604004, "step": 6554 }, { "epoch": 0.5167520693732756, "grad_norm": 0.7029499411582947, "learning_rate": 0.0005, "loss": 2.099818468093872, "step": 6555 }, { "epoch": 0.5168309026409145, "grad_norm": 0.7140821814537048, "learning_rate": 0.0005, "loss": 2.2072558403015137, "step": 6556 }, { "epoch": 0.5169097359085534, "grad_norm": 0.6842448115348816, "learning_rate": 0.0005, "loss": 1.8833248615264893, "step": 6557 }, { "epoch": 0.5169885691761924, "grad_norm": 0.6985316276550293, "learning_rate": 0.0005, "loss": 2.249551773071289, "step": 6558 }, { "epoch": 0.5170674024438313, "grad_norm": 0.7199367880821228, "learning_rate": 0.0005, "loss": 2.05430006980896, "step": 6559 }, { "epoch": 0.5171462357114702, "grad_norm": 0.6461794376373291, "learning_rate": 0.0005, "loss": 1.9166337251663208, "step": 6560 }, { "epoch": 0.5172250689791091, "grad_norm": 0.6699815988540649, "learning_rate": 0.0005, "loss": 2.0360119342803955, "step": 6561 }, { "epoch": 0.5173039022467482, "grad_norm": 0.7045876383781433, "learning_rate": 0.0005, "loss": 1.897579550743103, "step": 6562 }, { "epoch": 0.5173827355143871, "grad_norm": 0.5544307827949524, "learning_rate": 0.0005, "loss": 1.7733795642852783, "step": 6563 }, { "epoch": 0.517461568782026, "grad_norm": 0.6764855980873108, "learning_rate": 0.0005, "loss": 1.966745376586914, "step": 6564 }, { "epoch": 0.517540402049665, "grad_norm": 0.8308161497116089, "learning_rate": 0.0005, "loss": 2.2393970489501953, "step": 6565 }, { "epoch": 0.5176192353173039, "grad_norm": 0.7632996439933777, "learning_rate": 0.0005, "loss": 2.2272582054138184, "step": 6566 }, { "epoch": 0.5176980685849428, "grad_norm": 0.6616365909576416, "learning_rate": 0.0005, "loss": 1.9660615921020508, "step": 6567 }, { "epoch": 0.5177769018525817, "grad_norm": 0.7444329261779785, "learning_rate": 0.0005, "loss": 2.085319995880127, "step": 6568 }, { "epoch": 0.5178557351202208, "grad_norm": 0.7102121114730835, "learning_rate": 0.0005, "loss": 2.0788190364837646, "step": 6569 }, { "epoch": 0.5179345683878597, "grad_norm": 0.6870746612548828, "learning_rate": 0.0005, "loss": 2.069143772125244, "step": 6570 }, { "epoch": 0.5180134016554986, "grad_norm": 0.9221516251564026, "learning_rate": 0.0005, "loss": 1.9691758155822754, "step": 6571 }, { "epoch": 0.5180922349231376, "grad_norm": 0.5291075110435486, "learning_rate": 0.0005, "loss": 1.825040340423584, "step": 6572 }, { "epoch": 0.5181710681907765, "grad_norm": 0.6997559070587158, "learning_rate": 0.0005, "loss": 2.0655570030212402, "step": 6573 }, { "epoch": 0.5182499014584154, "grad_norm": 0.9671095013618469, "learning_rate": 0.0005, "loss": 2.1408677101135254, "step": 6574 }, { "epoch": 0.5183287347260543, "grad_norm": 0.7821630835533142, "learning_rate": 0.0005, "loss": 2.1941957473754883, "step": 6575 }, { "epoch": 0.5184075679936934, "grad_norm": 0.7734060287475586, "learning_rate": 0.0005, "loss": 2.053720235824585, "step": 6576 }, { "epoch": 0.5184864012613323, "grad_norm": 1.0038877725601196, "learning_rate": 0.0005, "loss": 2.0153121948242188, "step": 6577 }, { "epoch": 0.5185652345289712, "grad_norm": 0.6732265949249268, "learning_rate": 0.0005, "loss": 1.8292996883392334, "step": 6578 }, { "epoch": 0.5186440677966102, "grad_norm": 0.7770448327064514, "learning_rate": 0.0005, "loss": 2.0673346519470215, "step": 6579 }, { "epoch": 0.5187229010642491, "grad_norm": 0.7035406827926636, "learning_rate": 0.0005, "loss": 2.1014559268951416, "step": 6580 }, { "epoch": 0.518801734331888, "grad_norm": 0.7958241701126099, "learning_rate": 0.0005, "loss": 2.255582332611084, "step": 6581 }, { "epoch": 0.518880567599527, "grad_norm": 1.4066216945648193, "learning_rate": 0.0005, "loss": 2.2479329109191895, "step": 6582 }, { "epoch": 0.518959400867166, "grad_norm": 0.6053043603897095, "learning_rate": 0.0005, "loss": 1.9444249868392944, "step": 6583 }, { "epoch": 0.5190382341348049, "grad_norm": 0.6711348295211792, "learning_rate": 0.0005, "loss": 2.0161490440368652, "step": 6584 }, { "epoch": 0.5191170674024438, "grad_norm": 0.7387116551399231, "learning_rate": 0.0005, "loss": 2.2990074157714844, "step": 6585 }, { "epoch": 0.5191959006700828, "grad_norm": 0.6765607595443726, "learning_rate": 0.0005, "loss": 2.278622627258301, "step": 6586 }, { "epoch": 0.5192747339377217, "grad_norm": 0.789633572101593, "learning_rate": 0.0005, "loss": 1.910495638847351, "step": 6587 }, { "epoch": 0.5193535672053606, "grad_norm": 0.7070184350013733, "learning_rate": 0.0005, "loss": 2.204124927520752, "step": 6588 }, { "epoch": 0.5194324004729997, "grad_norm": 0.6847938895225525, "learning_rate": 0.0005, "loss": 1.9344019889831543, "step": 6589 }, { "epoch": 0.5195112337406386, "grad_norm": 0.737322211265564, "learning_rate": 0.0005, "loss": 1.9007999897003174, "step": 6590 }, { "epoch": 0.5195900670082775, "grad_norm": 0.7250795960426331, "learning_rate": 0.0005, "loss": 2.1766107082366943, "step": 6591 }, { "epoch": 0.5196689002759164, "grad_norm": 0.9165570735931396, "learning_rate": 0.0005, "loss": 1.859363317489624, "step": 6592 }, { "epoch": 0.5197477335435554, "grad_norm": 0.8421022295951843, "learning_rate": 0.0005, "loss": 2.1092495918273926, "step": 6593 }, { "epoch": 0.5198265668111943, "grad_norm": 0.7521007657051086, "learning_rate": 0.0005, "loss": 2.079296112060547, "step": 6594 }, { "epoch": 0.5199054000788332, "grad_norm": 0.7864177823066711, "learning_rate": 0.0005, "loss": 1.936572790145874, "step": 6595 }, { "epoch": 0.5199842333464723, "grad_norm": 0.7141655087471008, "learning_rate": 0.0005, "loss": 2.180762529373169, "step": 6596 }, { "epoch": 0.5200630666141112, "grad_norm": 1.0862922668457031, "learning_rate": 0.0005, "loss": 1.5152556896209717, "step": 6597 }, { "epoch": 0.5201418998817501, "grad_norm": 0.6746960878372192, "learning_rate": 0.0005, "loss": 2.0106518268585205, "step": 6598 }, { "epoch": 0.520220733149389, "grad_norm": 0.829291820526123, "learning_rate": 0.0005, "loss": 1.9242432117462158, "step": 6599 }, { "epoch": 0.520299566417028, "grad_norm": 0.7693510055541992, "learning_rate": 0.0005, "loss": 2.18949556350708, "step": 6600 }, { "epoch": 0.5203783996846669, "grad_norm": 0.7543879151344299, "learning_rate": 0.0005, "loss": 2.012148857116699, "step": 6601 }, { "epoch": 0.5204572329523058, "grad_norm": 0.7832505106925964, "learning_rate": 0.0005, "loss": 2.1009039878845215, "step": 6602 }, { "epoch": 0.5205360662199449, "grad_norm": 0.7518325448036194, "learning_rate": 0.0005, "loss": 1.9931795597076416, "step": 6603 }, { "epoch": 0.5206148994875838, "grad_norm": 0.8042719960212708, "learning_rate": 0.0005, "loss": 2.0869390964508057, "step": 6604 }, { "epoch": 0.5206937327552227, "grad_norm": 0.6263353824615479, "learning_rate": 0.0005, "loss": 2.1808063983917236, "step": 6605 }, { "epoch": 0.5207725660228616, "grad_norm": 0.7890037894248962, "learning_rate": 0.0005, "loss": 2.073815107345581, "step": 6606 }, { "epoch": 0.5208513992905006, "grad_norm": 0.7121945023536682, "learning_rate": 0.0005, "loss": 1.9678175449371338, "step": 6607 }, { "epoch": 0.5209302325581395, "grad_norm": 0.6892842650413513, "learning_rate": 0.0005, "loss": 2.1948726177215576, "step": 6608 }, { "epoch": 0.5210090658257784, "grad_norm": 0.7453395128250122, "learning_rate": 0.0005, "loss": 2.2521920204162598, "step": 6609 }, { "epoch": 0.5210878990934175, "grad_norm": 0.5915964841842651, "learning_rate": 0.0005, "loss": 1.7812153100967407, "step": 6610 }, { "epoch": 0.5211667323610564, "grad_norm": 0.8183128833770752, "learning_rate": 0.0005, "loss": 2.339200019836426, "step": 6611 }, { "epoch": 0.5212455656286953, "grad_norm": 0.7509436011314392, "learning_rate": 0.0005, "loss": 2.122335433959961, "step": 6612 }, { "epoch": 0.5213243988963343, "grad_norm": 0.6787338852882385, "learning_rate": 0.0005, "loss": 1.8490573167800903, "step": 6613 }, { "epoch": 0.5214032321639732, "grad_norm": 0.7246100902557373, "learning_rate": 0.0005, "loss": 1.9948196411132812, "step": 6614 }, { "epoch": 0.5214820654316121, "grad_norm": 0.9690107703208923, "learning_rate": 0.0005, "loss": 2.4484381675720215, "step": 6615 }, { "epoch": 0.521560898699251, "grad_norm": 0.7422318458557129, "learning_rate": 0.0005, "loss": 2.068909168243408, "step": 6616 }, { "epoch": 0.5216397319668901, "grad_norm": 0.8827736377716064, "learning_rate": 0.0005, "loss": 2.2073845863342285, "step": 6617 }, { "epoch": 0.521718565234529, "grad_norm": 0.617599368095398, "learning_rate": 0.0005, "loss": 1.9952223300933838, "step": 6618 }, { "epoch": 0.5217973985021679, "grad_norm": 0.716539740562439, "learning_rate": 0.0005, "loss": 1.9265971183776855, "step": 6619 }, { "epoch": 0.5218762317698069, "grad_norm": 0.8793630599975586, "learning_rate": 0.0005, "loss": 2.088515520095825, "step": 6620 }, { "epoch": 0.5219550650374458, "grad_norm": 0.6849331259727478, "learning_rate": 0.0005, "loss": 2.240337371826172, "step": 6621 }, { "epoch": 0.5220338983050847, "grad_norm": 0.8641734719276428, "learning_rate": 0.0005, "loss": 2.200424909591675, "step": 6622 }, { "epoch": 0.5221127315727236, "grad_norm": 1.1069401502609253, "learning_rate": 0.0005, "loss": 2.086395740509033, "step": 6623 }, { "epoch": 0.5221915648403627, "grad_norm": 0.7892831563949585, "learning_rate": 0.0005, "loss": 2.4188761711120605, "step": 6624 }, { "epoch": 0.5222703981080016, "grad_norm": 0.6837638020515442, "learning_rate": 0.0005, "loss": 2.004136562347412, "step": 6625 }, { "epoch": 0.5223492313756405, "grad_norm": 0.6169276237487793, "learning_rate": 0.0005, "loss": 1.825134515762329, "step": 6626 }, { "epoch": 0.5224280646432795, "grad_norm": 1.3777111768722534, "learning_rate": 0.0005, "loss": 1.9668769836425781, "step": 6627 }, { "epoch": 0.5225068979109184, "grad_norm": 2.8083930015563965, "learning_rate": 0.0005, "loss": 2.1145453453063965, "step": 6628 }, { "epoch": 0.5225857311785573, "grad_norm": 0.6259323358535767, "learning_rate": 0.0005, "loss": 2.0934646129608154, "step": 6629 }, { "epoch": 0.5226645644461962, "grad_norm": 2.4670393466949463, "learning_rate": 0.0005, "loss": 2.2062530517578125, "step": 6630 }, { "epoch": 0.5227433977138353, "grad_norm": 0.6413877606391907, "learning_rate": 0.0005, "loss": 2.0577526092529297, "step": 6631 }, { "epoch": 0.5228222309814742, "grad_norm": 0.7321548461914062, "learning_rate": 0.0005, "loss": 1.6971596479415894, "step": 6632 }, { "epoch": 0.5229010642491131, "grad_norm": 0.756787896156311, "learning_rate": 0.0005, "loss": 1.8127782344818115, "step": 6633 }, { "epoch": 0.5229798975167521, "grad_norm": 0.6101874113082886, "learning_rate": 0.0005, "loss": 1.8895875215530396, "step": 6634 }, { "epoch": 0.523058730784391, "grad_norm": 0.7026779651641846, "learning_rate": 0.0005, "loss": 1.9519152641296387, "step": 6635 }, { "epoch": 0.5231375640520299, "grad_norm": 0.764797568321228, "learning_rate": 0.0005, "loss": 2.298779010772705, "step": 6636 }, { "epoch": 0.5232163973196688, "grad_norm": 0.7381085157394409, "learning_rate": 0.0005, "loss": 1.8450899124145508, "step": 6637 }, { "epoch": 0.5232952305873079, "grad_norm": 0.7563802599906921, "learning_rate": 0.0005, "loss": 2.25243878364563, "step": 6638 }, { "epoch": 0.5233740638549468, "grad_norm": 0.6378979682922363, "learning_rate": 0.0005, "loss": 1.7909111976623535, "step": 6639 }, { "epoch": 0.5234528971225857, "grad_norm": 0.7419382929801941, "learning_rate": 0.0005, "loss": 1.8332206010818481, "step": 6640 }, { "epoch": 0.5235317303902247, "grad_norm": 0.7459235787391663, "learning_rate": 0.0005, "loss": 2.0026142597198486, "step": 6641 }, { "epoch": 0.5236105636578636, "grad_norm": 0.8025335073471069, "learning_rate": 0.0005, "loss": 2.345533609390259, "step": 6642 }, { "epoch": 0.5236893969255025, "grad_norm": 0.7449461221694946, "learning_rate": 0.0005, "loss": 2.1178765296936035, "step": 6643 }, { "epoch": 0.5237682301931416, "grad_norm": 0.9999982118606567, "learning_rate": 0.0005, "loss": 2.1352648735046387, "step": 6644 }, { "epoch": 0.5238470634607805, "grad_norm": 0.7414854764938354, "learning_rate": 0.0005, "loss": 2.0480170249938965, "step": 6645 }, { "epoch": 0.5239258967284194, "grad_norm": 0.6977322101593018, "learning_rate": 0.0005, "loss": 1.8670910596847534, "step": 6646 }, { "epoch": 0.5240047299960583, "grad_norm": 0.574394941329956, "learning_rate": 0.0005, "loss": 2.1333460807800293, "step": 6647 }, { "epoch": 0.5240835632636973, "grad_norm": 0.8043791651725769, "learning_rate": 0.0005, "loss": 2.0602378845214844, "step": 6648 }, { "epoch": 0.5241623965313362, "grad_norm": 1.100476861000061, "learning_rate": 0.0005, "loss": 2.1538193225860596, "step": 6649 }, { "epoch": 0.5242412297989751, "grad_norm": 0.6926333904266357, "learning_rate": 0.0005, "loss": 2.2152323722839355, "step": 6650 }, { "epoch": 0.5243200630666142, "grad_norm": 0.8390545845031738, "learning_rate": 0.0005, "loss": 2.05435848236084, "step": 6651 }, { "epoch": 0.5243988963342531, "grad_norm": 0.7393633723258972, "learning_rate": 0.0005, "loss": 2.0436244010925293, "step": 6652 }, { "epoch": 0.524477729601892, "grad_norm": 0.8587251901626587, "learning_rate": 0.0005, "loss": 1.7886182069778442, "step": 6653 }, { "epoch": 0.5245565628695309, "grad_norm": 0.6991569995880127, "learning_rate": 0.0005, "loss": 2.2636170387268066, "step": 6654 }, { "epoch": 0.5246353961371699, "grad_norm": 0.642979621887207, "learning_rate": 0.0005, "loss": 1.8658299446105957, "step": 6655 }, { "epoch": 0.5247142294048088, "grad_norm": 1.4800447225570679, "learning_rate": 0.0005, "loss": 2.092325210571289, "step": 6656 }, { "epoch": 0.5247930626724477, "grad_norm": 0.7935178875923157, "learning_rate": 0.0005, "loss": 2.210442543029785, "step": 6657 }, { "epoch": 0.5248718959400868, "grad_norm": 0.6715022921562195, "learning_rate": 0.0005, "loss": 1.914886474609375, "step": 6658 }, { "epoch": 0.5249507292077257, "grad_norm": 0.7434694766998291, "learning_rate": 0.0005, "loss": 2.0596728324890137, "step": 6659 }, { "epoch": 0.5250295624753646, "grad_norm": 0.694715678691864, "learning_rate": 0.0005, "loss": 2.236635208129883, "step": 6660 }, { "epoch": 0.5251083957430035, "grad_norm": 0.6107905507087708, "learning_rate": 0.0005, "loss": 1.9630595445632935, "step": 6661 }, { "epoch": 0.5251872290106425, "grad_norm": 0.5830665826797485, "learning_rate": 0.0005, "loss": 2.057981014251709, "step": 6662 }, { "epoch": 0.5252660622782814, "grad_norm": 0.6196030974388123, "learning_rate": 0.0005, "loss": 2.066025733947754, "step": 6663 }, { "epoch": 0.5253448955459203, "grad_norm": 0.879101037979126, "learning_rate": 0.0005, "loss": 2.2800674438476562, "step": 6664 }, { "epoch": 0.5254237288135594, "grad_norm": 0.7975561022758484, "learning_rate": 0.0005, "loss": 2.054518938064575, "step": 6665 }, { "epoch": 0.5255025620811983, "grad_norm": 0.9801653027534485, "learning_rate": 0.0005, "loss": 2.2888712882995605, "step": 6666 }, { "epoch": 0.5255813953488372, "grad_norm": 0.6919838190078735, "learning_rate": 0.0005, "loss": 2.1769301891326904, "step": 6667 }, { "epoch": 0.5256602286164761, "grad_norm": 0.7606837153434753, "learning_rate": 0.0005, "loss": 1.915602207183838, "step": 6668 }, { "epoch": 0.5257390618841151, "grad_norm": 0.8940503001213074, "learning_rate": 0.0005, "loss": 1.963742971420288, "step": 6669 }, { "epoch": 0.525817895151754, "grad_norm": 1.1402621269226074, "learning_rate": 0.0005, "loss": 2.020010232925415, "step": 6670 }, { "epoch": 0.525896728419393, "grad_norm": 0.7238412499427795, "learning_rate": 0.0005, "loss": 2.2816109657287598, "step": 6671 }, { "epoch": 0.525975561687032, "grad_norm": 0.797213613986969, "learning_rate": 0.0005, "loss": 2.2404403686523438, "step": 6672 }, { "epoch": 0.5260543949546709, "grad_norm": 0.7507100105285645, "learning_rate": 0.0005, "loss": 2.2832605838775635, "step": 6673 }, { "epoch": 0.5261332282223098, "grad_norm": 0.7016539573669434, "learning_rate": 0.0005, "loss": 2.190977096557617, "step": 6674 }, { "epoch": 0.5262120614899488, "grad_norm": 0.7826810479164124, "learning_rate": 0.0005, "loss": 2.0831046104431152, "step": 6675 }, { "epoch": 0.5262908947575877, "grad_norm": 0.7115052342414856, "learning_rate": 0.0005, "loss": 1.8199622631072998, "step": 6676 }, { "epoch": 0.5263697280252266, "grad_norm": 0.7369949817657471, "learning_rate": 0.0005, "loss": 1.9332847595214844, "step": 6677 }, { "epoch": 0.5264485612928655, "grad_norm": 0.822803258895874, "learning_rate": 0.0005, "loss": 2.1534132957458496, "step": 6678 }, { "epoch": 0.5265273945605046, "grad_norm": 0.8365658521652222, "learning_rate": 0.0005, "loss": 2.130797863006592, "step": 6679 }, { "epoch": 0.5266062278281435, "grad_norm": 0.6458798050880432, "learning_rate": 0.0005, "loss": 2.05193829536438, "step": 6680 }, { "epoch": 0.5266850610957824, "grad_norm": 0.770647406578064, "learning_rate": 0.0005, "loss": 2.264799118041992, "step": 6681 }, { "epoch": 0.5267638943634214, "grad_norm": 0.6307328939437866, "learning_rate": 0.0005, "loss": 1.8482236862182617, "step": 6682 }, { "epoch": 0.5268427276310603, "grad_norm": 0.7378534078598022, "learning_rate": 0.0005, "loss": 1.8996542692184448, "step": 6683 }, { "epoch": 0.5269215608986992, "grad_norm": 0.7170873880386353, "learning_rate": 0.0005, "loss": 2.2004082202911377, "step": 6684 }, { "epoch": 0.5270003941663381, "grad_norm": 0.6957203149795532, "learning_rate": 0.0005, "loss": 1.8480420112609863, "step": 6685 }, { "epoch": 0.5270792274339772, "grad_norm": 0.5857036709785461, "learning_rate": 0.0005, "loss": 1.9891722202301025, "step": 6686 }, { "epoch": 0.5271580607016161, "grad_norm": 0.8817178010940552, "learning_rate": 0.0005, "loss": 1.959040880203247, "step": 6687 }, { "epoch": 0.527236893969255, "grad_norm": 0.7890664339065552, "learning_rate": 0.0005, "loss": 2.217550039291382, "step": 6688 }, { "epoch": 0.527315727236894, "grad_norm": 1.1326496601104736, "learning_rate": 0.0005, "loss": 2.211988687515259, "step": 6689 }, { "epoch": 0.5273945605045329, "grad_norm": 0.6989772915840149, "learning_rate": 0.0005, "loss": 2.1352877616882324, "step": 6690 }, { "epoch": 0.5274733937721718, "grad_norm": 0.7080333828926086, "learning_rate": 0.0005, "loss": 2.2260684967041016, "step": 6691 }, { "epoch": 0.5275522270398108, "grad_norm": 0.6532138586044312, "learning_rate": 0.0005, "loss": 2.174365520477295, "step": 6692 }, { "epoch": 0.5276310603074498, "grad_norm": 0.5791553854942322, "learning_rate": 0.0005, "loss": 1.9044889211654663, "step": 6693 }, { "epoch": 0.5277098935750887, "grad_norm": 0.7429763674736023, "learning_rate": 0.0005, "loss": 2.116863965988159, "step": 6694 }, { "epoch": 0.5277887268427276, "grad_norm": 0.5111766457557678, "learning_rate": 0.0005, "loss": 1.8712695837020874, "step": 6695 }, { "epoch": 0.5278675601103666, "grad_norm": 0.6264615654945374, "learning_rate": 0.0005, "loss": 1.8830392360687256, "step": 6696 }, { "epoch": 0.5279463933780055, "grad_norm": 0.6318603157997131, "learning_rate": 0.0005, "loss": 2.1148695945739746, "step": 6697 }, { "epoch": 0.5280252266456444, "grad_norm": 0.8152506351470947, "learning_rate": 0.0005, "loss": 2.2418031692504883, "step": 6698 }, { "epoch": 0.5281040599132834, "grad_norm": 0.8275061845779419, "learning_rate": 0.0005, "loss": 2.2068209648132324, "step": 6699 }, { "epoch": 0.5281828931809224, "grad_norm": 0.6520290374755859, "learning_rate": 0.0005, "loss": 1.8163163661956787, "step": 6700 }, { "epoch": 0.5282617264485613, "grad_norm": 0.6804658770561218, "learning_rate": 0.0005, "loss": 2.131593704223633, "step": 6701 }, { "epoch": 0.5283405597162002, "grad_norm": 1.4947885274887085, "learning_rate": 0.0005, "loss": 2.2901034355163574, "step": 6702 }, { "epoch": 0.5284193929838392, "grad_norm": 0.8279228210449219, "learning_rate": 0.0005, "loss": 2.420083999633789, "step": 6703 }, { "epoch": 0.5284982262514781, "grad_norm": 0.833588719367981, "learning_rate": 0.0005, "loss": 2.132841110229492, "step": 6704 }, { "epoch": 0.528577059519117, "grad_norm": 0.8555593490600586, "learning_rate": 0.0005, "loss": 2.23384952545166, "step": 6705 }, { "epoch": 0.5286558927867561, "grad_norm": 0.7690750360488892, "learning_rate": 0.0005, "loss": 2.333841323852539, "step": 6706 }, { "epoch": 0.528734726054395, "grad_norm": 1.6301287412643433, "learning_rate": 0.0005, "loss": 1.7999441623687744, "step": 6707 }, { "epoch": 0.5288135593220339, "grad_norm": 0.7936424016952515, "learning_rate": 0.0005, "loss": 1.8929170370101929, "step": 6708 }, { "epoch": 0.5288923925896728, "grad_norm": 0.8807873129844666, "learning_rate": 0.0005, "loss": 1.9413141012191772, "step": 6709 }, { "epoch": 0.5289712258573118, "grad_norm": 0.8060272336006165, "learning_rate": 0.0005, "loss": 2.133606433868408, "step": 6710 }, { "epoch": 0.5290500591249507, "grad_norm": 0.6581811308860779, "learning_rate": 0.0005, "loss": 2.189936637878418, "step": 6711 }, { "epoch": 0.5291288923925896, "grad_norm": 0.7807177901268005, "learning_rate": 0.0005, "loss": 1.8117406368255615, "step": 6712 }, { "epoch": 0.5292077256602287, "grad_norm": 1.1963469982147217, "learning_rate": 0.0005, "loss": 1.9212439060211182, "step": 6713 }, { "epoch": 0.5292865589278676, "grad_norm": 0.7342074513435364, "learning_rate": 0.0005, "loss": 1.8973822593688965, "step": 6714 }, { "epoch": 0.5293653921955065, "grad_norm": 1.5413020849227905, "learning_rate": 0.0005, "loss": 1.7475659847259521, "step": 6715 }, { "epoch": 0.5294442254631454, "grad_norm": 0.8627899289131165, "learning_rate": 0.0005, "loss": 1.9210710525512695, "step": 6716 }, { "epoch": 0.5295230587307844, "grad_norm": 1.015063762664795, "learning_rate": 0.0005, "loss": 2.2540125846862793, "step": 6717 }, { "epoch": 0.5296018919984233, "grad_norm": 1.133630394935608, "learning_rate": 0.0005, "loss": 2.218627452850342, "step": 6718 }, { "epoch": 0.5296807252660622, "grad_norm": 1.0671411752700806, "learning_rate": 0.0005, "loss": 2.198166608810425, "step": 6719 }, { "epoch": 0.5297595585337013, "grad_norm": 2.244680643081665, "learning_rate": 0.0005, "loss": 2.019216537475586, "step": 6720 }, { "epoch": 0.5298383918013402, "grad_norm": 0.9861341714859009, "learning_rate": 0.0005, "loss": 2.2355756759643555, "step": 6721 }, { "epoch": 0.5299172250689791, "grad_norm": 0.7442020177841187, "learning_rate": 0.0005, "loss": 2.116614818572998, "step": 6722 }, { "epoch": 0.529996058336618, "grad_norm": 0.8975839018821716, "learning_rate": 0.0005, "loss": 1.9732301235198975, "step": 6723 }, { "epoch": 0.530074891604257, "grad_norm": 1.0904244184494019, "learning_rate": 0.0005, "loss": 2.2967214584350586, "step": 6724 }, { "epoch": 0.5301537248718959, "grad_norm": 0.8168929815292358, "learning_rate": 0.0005, "loss": 2.0693931579589844, "step": 6725 }, { "epoch": 0.5302325581395348, "grad_norm": 0.5829454660415649, "learning_rate": 0.0005, "loss": 2.0746994018554688, "step": 6726 }, { "epoch": 0.5303113914071739, "grad_norm": 0.7367590665817261, "learning_rate": 0.0005, "loss": 2.0858426094055176, "step": 6727 }, { "epoch": 0.5303902246748128, "grad_norm": 0.8151957988739014, "learning_rate": 0.0005, "loss": 1.8498073816299438, "step": 6728 }, { "epoch": 0.5304690579424517, "grad_norm": 0.8774442672729492, "learning_rate": 0.0005, "loss": 1.7596690654754639, "step": 6729 }, { "epoch": 0.5305478912100906, "grad_norm": 0.587738573551178, "learning_rate": 0.0005, "loss": 1.8518468141555786, "step": 6730 }, { "epoch": 0.5306267244777296, "grad_norm": 0.8366149663925171, "learning_rate": 0.0005, "loss": 1.8164222240447998, "step": 6731 }, { "epoch": 0.5307055577453685, "grad_norm": 0.730644941329956, "learning_rate": 0.0005, "loss": 2.1402931213378906, "step": 6732 }, { "epoch": 0.5307843910130075, "grad_norm": 0.9322410821914673, "learning_rate": 0.0005, "loss": 1.7432501316070557, "step": 6733 }, { "epoch": 0.5308632242806465, "grad_norm": 0.7497172355651855, "learning_rate": 0.0005, "loss": 2.060620069503784, "step": 6734 }, { "epoch": 0.5309420575482854, "grad_norm": 0.8694497346878052, "learning_rate": 0.0005, "loss": 2.228642463684082, "step": 6735 }, { "epoch": 0.5310208908159243, "grad_norm": 0.7903329133987427, "learning_rate": 0.0005, "loss": 2.113309144973755, "step": 6736 }, { "epoch": 0.5310997240835633, "grad_norm": 0.6882847547531128, "learning_rate": 0.0005, "loss": 2.181084156036377, "step": 6737 }, { "epoch": 0.5311785573512022, "grad_norm": 0.6987554430961609, "learning_rate": 0.0005, "loss": 1.985177993774414, "step": 6738 }, { "epoch": 0.5312573906188411, "grad_norm": 1.0013620853424072, "learning_rate": 0.0005, "loss": 1.921337604522705, "step": 6739 }, { "epoch": 0.53133622388648, "grad_norm": 0.8616316318511963, "learning_rate": 0.0005, "loss": 1.9135136604309082, "step": 6740 }, { "epoch": 0.5314150571541191, "grad_norm": 0.5910910964012146, "learning_rate": 0.0005, "loss": 2.102097511291504, "step": 6741 }, { "epoch": 0.531493890421758, "grad_norm": 0.6222331523895264, "learning_rate": 0.0005, "loss": 2.2219247817993164, "step": 6742 }, { "epoch": 0.5315727236893969, "grad_norm": 0.8685970306396484, "learning_rate": 0.0005, "loss": 1.9551626443862915, "step": 6743 }, { "epoch": 0.5316515569570359, "grad_norm": 0.7799015045166016, "learning_rate": 0.0005, "loss": 2.487682342529297, "step": 6744 }, { "epoch": 0.5317303902246748, "grad_norm": 0.6259810328483582, "learning_rate": 0.0005, "loss": 2.0142321586608887, "step": 6745 }, { "epoch": 0.5318092234923137, "grad_norm": 0.79015052318573, "learning_rate": 0.0005, "loss": 1.9110889434814453, "step": 6746 }, { "epoch": 0.5318880567599527, "grad_norm": 0.9646888375282288, "learning_rate": 0.0005, "loss": 2.2307987213134766, "step": 6747 }, { "epoch": 0.5319668900275917, "grad_norm": 0.8898813724517822, "learning_rate": 0.0005, "loss": 1.9319496154785156, "step": 6748 }, { "epoch": 0.5320457232952306, "grad_norm": 0.5802428722381592, "learning_rate": 0.0005, "loss": 1.8789091110229492, "step": 6749 }, { "epoch": 0.5321245565628695, "grad_norm": 1.0040992498397827, "learning_rate": 0.0005, "loss": 1.7881712913513184, "step": 6750 }, { "epoch": 0.5322033898305085, "grad_norm": 0.6882926821708679, "learning_rate": 0.0005, "loss": 2.109471321105957, "step": 6751 }, { "epoch": 0.5322822230981474, "grad_norm": 0.6659057140350342, "learning_rate": 0.0005, "loss": 1.860172152519226, "step": 6752 }, { "epoch": 0.5323610563657863, "grad_norm": 0.6445650458335876, "learning_rate": 0.0005, "loss": 1.9691078662872314, "step": 6753 }, { "epoch": 0.5324398896334253, "grad_norm": 0.7598139643669128, "learning_rate": 0.0005, "loss": 1.9835357666015625, "step": 6754 }, { "epoch": 0.5325187229010643, "grad_norm": 0.6774816513061523, "learning_rate": 0.0005, "loss": 1.8243391513824463, "step": 6755 }, { "epoch": 0.5325975561687032, "grad_norm": 0.8399122953414917, "learning_rate": 0.0005, "loss": 2.0127429962158203, "step": 6756 }, { "epoch": 0.5326763894363421, "grad_norm": 0.7109445333480835, "learning_rate": 0.0005, "loss": 1.8342504501342773, "step": 6757 }, { "epoch": 0.5327552227039811, "grad_norm": 0.817852795124054, "learning_rate": 0.0005, "loss": 1.941719651222229, "step": 6758 }, { "epoch": 0.53283405597162, "grad_norm": 0.715671181678772, "learning_rate": 0.0005, "loss": 1.836242437362671, "step": 6759 }, { "epoch": 0.532912889239259, "grad_norm": 0.8123165965080261, "learning_rate": 0.0005, "loss": 2.0624430179595947, "step": 6760 }, { "epoch": 0.5329917225068979, "grad_norm": 0.781125009059906, "learning_rate": 0.0005, "loss": 2.0534698963165283, "step": 6761 }, { "epoch": 0.5330705557745369, "grad_norm": 0.7560190558433533, "learning_rate": 0.0005, "loss": 1.9029231071472168, "step": 6762 }, { "epoch": 0.5331493890421758, "grad_norm": 0.8140107989311218, "learning_rate": 0.0005, "loss": 1.8319928646087646, "step": 6763 }, { "epoch": 0.5332282223098147, "grad_norm": 0.7046973705291748, "learning_rate": 0.0005, "loss": 1.9081534147262573, "step": 6764 }, { "epoch": 0.5333070555774537, "grad_norm": 0.7259675860404968, "learning_rate": 0.0005, "loss": 2.041034698486328, "step": 6765 }, { "epoch": 0.5333858888450926, "grad_norm": 0.9214619994163513, "learning_rate": 0.0005, "loss": 1.9580403566360474, "step": 6766 }, { "epoch": 0.5334647221127315, "grad_norm": 0.5567334294319153, "learning_rate": 0.0005, "loss": 1.7866077423095703, "step": 6767 }, { "epoch": 0.5335435553803706, "grad_norm": 0.8049611449241638, "learning_rate": 0.0005, "loss": 2.335695743560791, "step": 6768 }, { "epoch": 0.5336223886480095, "grad_norm": 0.8773139715194702, "learning_rate": 0.0005, "loss": 1.8759164810180664, "step": 6769 }, { "epoch": 0.5337012219156484, "grad_norm": 1.0237843990325928, "learning_rate": 0.0005, "loss": 2.254051685333252, "step": 6770 }, { "epoch": 0.5337800551832873, "grad_norm": 0.7054091691970825, "learning_rate": 0.0005, "loss": 2.124091148376465, "step": 6771 }, { "epoch": 0.5338588884509263, "grad_norm": 0.8389642238616943, "learning_rate": 0.0005, "loss": 1.962587594985962, "step": 6772 }, { "epoch": 0.5339377217185652, "grad_norm": 0.722431480884552, "learning_rate": 0.0005, "loss": 2.061117172241211, "step": 6773 }, { "epoch": 0.5340165549862042, "grad_norm": 1.0758042335510254, "learning_rate": 0.0005, "loss": 1.9988408088684082, "step": 6774 }, { "epoch": 0.5340953882538432, "grad_norm": 0.8915878534317017, "learning_rate": 0.0005, "loss": 1.8820065259933472, "step": 6775 }, { "epoch": 0.5341742215214821, "grad_norm": 0.8106533288955688, "learning_rate": 0.0005, "loss": 1.993850588798523, "step": 6776 }, { "epoch": 0.534253054789121, "grad_norm": 0.8183537721633911, "learning_rate": 0.0005, "loss": 2.0085840225219727, "step": 6777 }, { "epoch": 0.5343318880567599, "grad_norm": 0.7311915159225464, "learning_rate": 0.0005, "loss": 1.8874573707580566, "step": 6778 }, { "epoch": 0.5344107213243989, "grad_norm": 0.6909571290016174, "learning_rate": 0.0005, "loss": 2.1298632621765137, "step": 6779 }, { "epoch": 0.5344895545920378, "grad_norm": 0.8279744982719421, "learning_rate": 0.0005, "loss": 2.0121748447418213, "step": 6780 }, { "epoch": 0.5345683878596768, "grad_norm": 0.8219069242477417, "learning_rate": 0.0005, "loss": 1.818532109260559, "step": 6781 }, { "epoch": 0.5346472211273158, "grad_norm": 0.852444589138031, "learning_rate": 0.0005, "loss": 2.1957626342773438, "step": 6782 }, { "epoch": 0.5347260543949547, "grad_norm": 0.8072472810745239, "learning_rate": 0.0005, "loss": 2.150972604751587, "step": 6783 }, { "epoch": 0.5348048876625936, "grad_norm": 0.8836286664009094, "learning_rate": 0.0005, "loss": 2.3005616664886475, "step": 6784 }, { "epoch": 0.5348837209302325, "grad_norm": 0.8229070901870728, "learning_rate": 0.0005, "loss": 1.8502812385559082, "step": 6785 }, { "epoch": 0.5349625541978715, "grad_norm": 0.8107569217681885, "learning_rate": 0.0005, "loss": 2.170903205871582, "step": 6786 }, { "epoch": 0.5350413874655104, "grad_norm": 1.5154740810394287, "learning_rate": 0.0005, "loss": 1.8681721687316895, "step": 6787 }, { "epoch": 0.5351202207331494, "grad_norm": 1.0197170972824097, "learning_rate": 0.0005, "loss": 1.934059500694275, "step": 6788 }, { "epoch": 0.5351990540007884, "grad_norm": 1.008080005645752, "learning_rate": 0.0005, "loss": 2.1291422843933105, "step": 6789 }, { "epoch": 0.5352778872684273, "grad_norm": 0.6649060249328613, "learning_rate": 0.0005, "loss": 2.083568572998047, "step": 6790 }, { "epoch": 0.5353567205360662, "grad_norm": 0.7669535875320435, "learning_rate": 0.0005, "loss": 1.9679360389709473, "step": 6791 }, { "epoch": 0.5354355538037051, "grad_norm": 0.7787926197052002, "learning_rate": 0.0005, "loss": 2.1840317249298096, "step": 6792 }, { "epoch": 0.5355143870713441, "grad_norm": 0.6472585201263428, "learning_rate": 0.0005, "loss": 2.117995023727417, "step": 6793 }, { "epoch": 0.535593220338983, "grad_norm": 0.8447850346565247, "learning_rate": 0.0005, "loss": 2.4998104572296143, "step": 6794 }, { "epoch": 0.535672053606622, "grad_norm": 0.743776261806488, "learning_rate": 0.0005, "loss": 2.0493412017822266, "step": 6795 }, { "epoch": 0.535750886874261, "grad_norm": 0.900614857673645, "learning_rate": 0.0005, "loss": 1.9534556865692139, "step": 6796 }, { "epoch": 0.5358297201418999, "grad_norm": 0.6854509711265564, "learning_rate": 0.0005, "loss": 2.340898036956787, "step": 6797 }, { "epoch": 0.5359085534095388, "grad_norm": 0.753595769405365, "learning_rate": 0.0005, "loss": 1.8848602771759033, "step": 6798 }, { "epoch": 0.5359873866771778, "grad_norm": 0.6764273643493652, "learning_rate": 0.0005, "loss": 1.938559651374817, "step": 6799 }, { "epoch": 0.5360662199448167, "grad_norm": 0.722976803779602, "learning_rate": 0.0005, "loss": 2.1402347087860107, "step": 6800 }, { "epoch": 0.5361450532124556, "grad_norm": 0.6912752389907837, "learning_rate": 0.0005, "loss": 1.9650835990905762, "step": 6801 }, { "epoch": 0.5362238864800946, "grad_norm": 0.9435974955558777, "learning_rate": 0.0005, "loss": 1.7818841934204102, "step": 6802 }, { "epoch": 0.5363027197477336, "grad_norm": 0.7012559175491333, "learning_rate": 0.0005, "loss": 2.130903959274292, "step": 6803 }, { "epoch": 0.5363815530153725, "grad_norm": 0.9243645668029785, "learning_rate": 0.0005, "loss": 2.237888813018799, "step": 6804 }, { "epoch": 0.5364603862830114, "grad_norm": 0.7955995202064514, "learning_rate": 0.0005, "loss": 2.1691575050354004, "step": 6805 }, { "epoch": 0.5365392195506504, "grad_norm": 0.7690231800079346, "learning_rate": 0.0005, "loss": 2.1837258338928223, "step": 6806 }, { "epoch": 0.5366180528182893, "grad_norm": 0.8270469903945923, "learning_rate": 0.0005, "loss": 1.8850218057632446, "step": 6807 }, { "epoch": 0.5366968860859282, "grad_norm": 1.154311180114746, "learning_rate": 0.0005, "loss": 1.7900217771530151, "step": 6808 }, { "epoch": 0.5367757193535672, "grad_norm": 0.6375277042388916, "learning_rate": 0.0005, "loss": 2.0801172256469727, "step": 6809 }, { "epoch": 0.5368545526212062, "grad_norm": 0.9103530645370483, "learning_rate": 0.0005, "loss": 1.9618879556655884, "step": 6810 }, { "epoch": 0.5369333858888451, "grad_norm": 0.6775336861610413, "learning_rate": 0.0005, "loss": 2.1115522384643555, "step": 6811 }, { "epoch": 0.537012219156484, "grad_norm": 0.6852522492408752, "learning_rate": 0.0005, "loss": 2.2181763648986816, "step": 6812 }, { "epoch": 0.537091052424123, "grad_norm": 0.5805981755256653, "learning_rate": 0.0005, "loss": 1.8458659648895264, "step": 6813 }, { "epoch": 0.5371698856917619, "grad_norm": 0.6250248551368713, "learning_rate": 0.0005, "loss": 2.207430839538574, "step": 6814 }, { "epoch": 0.5372487189594009, "grad_norm": 0.7520530819892883, "learning_rate": 0.0005, "loss": 2.3421359062194824, "step": 6815 }, { "epoch": 0.5373275522270398, "grad_norm": 0.760859489440918, "learning_rate": 0.0005, "loss": 2.09025239944458, "step": 6816 }, { "epoch": 0.5374063854946788, "grad_norm": 0.7094698548316956, "learning_rate": 0.0005, "loss": 1.9821488857269287, "step": 6817 }, { "epoch": 0.5374852187623177, "grad_norm": 1.0412614345550537, "learning_rate": 0.0005, "loss": 1.9954156875610352, "step": 6818 }, { "epoch": 0.5375640520299566, "grad_norm": 0.8752338290214539, "learning_rate": 0.0005, "loss": 2.1210317611694336, "step": 6819 }, { "epoch": 0.5376428852975956, "grad_norm": 0.8861441612243652, "learning_rate": 0.0005, "loss": 1.9879246950149536, "step": 6820 }, { "epoch": 0.5377217185652345, "grad_norm": 0.7848621606826782, "learning_rate": 0.0005, "loss": 2.083249092102051, "step": 6821 }, { "epoch": 0.5378005518328735, "grad_norm": 0.7755754590034485, "learning_rate": 0.0005, "loss": 1.837254285812378, "step": 6822 }, { "epoch": 0.5378793851005124, "grad_norm": 0.8528449535369873, "learning_rate": 0.0005, "loss": 2.444918632507324, "step": 6823 }, { "epoch": 0.5379582183681514, "grad_norm": 1.1729187965393066, "learning_rate": 0.0005, "loss": 2.0998191833496094, "step": 6824 }, { "epoch": 0.5380370516357903, "grad_norm": 0.7785987854003906, "learning_rate": 0.0005, "loss": 1.8995825052261353, "step": 6825 }, { "epoch": 0.5381158849034292, "grad_norm": 0.7139005661010742, "learning_rate": 0.0005, "loss": 2.1427390575408936, "step": 6826 }, { "epoch": 0.5381947181710682, "grad_norm": 0.6014544367790222, "learning_rate": 0.0005, "loss": 1.846469759941101, "step": 6827 }, { "epoch": 0.5382735514387071, "grad_norm": 0.8315635919570923, "learning_rate": 0.0005, "loss": 1.9604488611221313, "step": 6828 }, { "epoch": 0.538352384706346, "grad_norm": 0.8256328701972961, "learning_rate": 0.0005, "loss": 2.2711710929870605, "step": 6829 }, { "epoch": 0.5384312179739851, "grad_norm": 0.9635756611824036, "learning_rate": 0.0005, "loss": 1.9544661045074463, "step": 6830 }, { "epoch": 0.538510051241624, "grad_norm": 0.6371930241584778, "learning_rate": 0.0005, "loss": 1.954953908920288, "step": 6831 }, { "epoch": 0.5385888845092629, "grad_norm": 0.8373253345489502, "learning_rate": 0.0005, "loss": 2.1177425384521484, "step": 6832 }, { "epoch": 0.5386677177769018, "grad_norm": 0.7046528458595276, "learning_rate": 0.0005, "loss": 1.9655320644378662, "step": 6833 }, { "epoch": 0.5387465510445408, "grad_norm": 0.7175527811050415, "learning_rate": 0.0005, "loss": 2.234771251678467, "step": 6834 }, { "epoch": 0.5388253843121797, "grad_norm": 0.6826691031455994, "learning_rate": 0.0005, "loss": 2.1156797409057617, "step": 6835 }, { "epoch": 0.5389042175798187, "grad_norm": 0.758400559425354, "learning_rate": 0.0005, "loss": 1.8352888822555542, "step": 6836 }, { "epoch": 0.5389830508474577, "grad_norm": 0.7565062046051025, "learning_rate": 0.0005, "loss": 2.0947647094726562, "step": 6837 }, { "epoch": 0.5390618841150966, "grad_norm": 0.6747809052467346, "learning_rate": 0.0005, "loss": 2.046200752258301, "step": 6838 }, { "epoch": 0.5391407173827355, "grad_norm": 0.7380421161651611, "learning_rate": 0.0005, "loss": 2.0820059776306152, "step": 6839 }, { "epoch": 0.5392195506503744, "grad_norm": 1.0935684442520142, "learning_rate": 0.0005, "loss": 1.9906761646270752, "step": 6840 }, { "epoch": 0.5392983839180134, "grad_norm": 0.7836950421333313, "learning_rate": 0.0005, "loss": 2.091320514678955, "step": 6841 }, { "epoch": 0.5393772171856523, "grad_norm": 0.9296519160270691, "learning_rate": 0.0005, "loss": 2.1167030334472656, "step": 6842 }, { "epoch": 0.5394560504532913, "grad_norm": 0.8999133110046387, "learning_rate": 0.0005, "loss": 2.033647060394287, "step": 6843 }, { "epoch": 0.5395348837209303, "grad_norm": 0.7025803327560425, "learning_rate": 0.0005, "loss": 2.047407627105713, "step": 6844 }, { "epoch": 0.5396137169885692, "grad_norm": 0.8426165580749512, "learning_rate": 0.0005, "loss": 2.083728075027466, "step": 6845 }, { "epoch": 0.5396925502562081, "grad_norm": 0.6703872084617615, "learning_rate": 0.0005, "loss": 2.1616430282592773, "step": 6846 }, { "epoch": 0.539771383523847, "grad_norm": 0.7711074352264404, "learning_rate": 0.0005, "loss": 2.107428550720215, "step": 6847 }, { "epoch": 0.539850216791486, "grad_norm": 0.7471078634262085, "learning_rate": 0.0005, "loss": 2.10408878326416, "step": 6848 }, { "epoch": 0.539929050059125, "grad_norm": 0.9584313035011292, "learning_rate": 0.0005, "loss": 2.284062385559082, "step": 6849 }, { "epoch": 0.5400078833267639, "grad_norm": 0.8580319285392761, "learning_rate": 0.0005, "loss": 2.2184770107269287, "step": 6850 }, { "epoch": 0.5400867165944029, "grad_norm": 0.7707950472831726, "learning_rate": 0.0005, "loss": 2.4683308601379395, "step": 6851 }, { "epoch": 0.5401655498620418, "grad_norm": 0.9681329131126404, "learning_rate": 0.0005, "loss": 2.034451961517334, "step": 6852 }, { "epoch": 0.5402443831296807, "grad_norm": 0.627593457698822, "learning_rate": 0.0005, "loss": 1.8520848751068115, "step": 6853 }, { "epoch": 0.5403232163973196, "grad_norm": 0.6319739818572998, "learning_rate": 0.0005, "loss": 1.9116851091384888, "step": 6854 }, { "epoch": 0.5404020496649586, "grad_norm": 0.8584330677986145, "learning_rate": 0.0005, "loss": 1.923453688621521, "step": 6855 }, { "epoch": 0.5404808829325976, "grad_norm": 0.9968644976615906, "learning_rate": 0.0005, "loss": 2.1022770404815674, "step": 6856 }, { "epoch": 0.5405597162002365, "grad_norm": 0.9550208449363708, "learning_rate": 0.0005, "loss": 1.9949136972427368, "step": 6857 }, { "epoch": 0.5406385494678755, "grad_norm": 0.8315478563308716, "learning_rate": 0.0005, "loss": 1.9656803607940674, "step": 6858 }, { "epoch": 0.5407173827355144, "grad_norm": 0.6483157873153687, "learning_rate": 0.0005, "loss": 2.0920352935791016, "step": 6859 }, { "epoch": 0.5407962160031533, "grad_norm": 1.403592824935913, "learning_rate": 0.0005, "loss": 2.310023784637451, "step": 6860 }, { "epoch": 0.5408750492707923, "grad_norm": 0.7710642218589783, "learning_rate": 0.0005, "loss": 2.1848716735839844, "step": 6861 }, { "epoch": 0.5409538825384312, "grad_norm": 0.7122172713279724, "learning_rate": 0.0005, "loss": 1.8684735298156738, "step": 6862 }, { "epoch": 0.5410327158060702, "grad_norm": 0.8675169944763184, "learning_rate": 0.0005, "loss": 2.3927624225616455, "step": 6863 }, { "epoch": 0.5411115490737091, "grad_norm": 0.8802556395530701, "learning_rate": 0.0005, "loss": 2.1380691528320312, "step": 6864 }, { "epoch": 0.5411903823413481, "grad_norm": 0.7734774947166443, "learning_rate": 0.0005, "loss": 2.006296157836914, "step": 6865 }, { "epoch": 0.541269215608987, "grad_norm": 0.6715797781944275, "learning_rate": 0.0005, "loss": 2.2547364234924316, "step": 6866 }, { "epoch": 0.5413480488766259, "grad_norm": 0.8779233694076538, "learning_rate": 0.0005, "loss": 1.7352641820907593, "step": 6867 }, { "epoch": 0.5414268821442649, "grad_norm": 1.6582945585250854, "learning_rate": 0.0005, "loss": 1.7477997541427612, "step": 6868 }, { "epoch": 0.5415057154119038, "grad_norm": 0.9882705211639404, "learning_rate": 0.0005, "loss": 2.4413540363311768, "step": 6869 }, { "epoch": 0.5415845486795428, "grad_norm": 0.7131158709526062, "learning_rate": 0.0005, "loss": 2.0533487796783447, "step": 6870 }, { "epoch": 0.5416633819471817, "grad_norm": 0.6221247315406799, "learning_rate": 0.0005, "loss": 2.001697540283203, "step": 6871 }, { "epoch": 0.5417422152148207, "grad_norm": 0.8157238364219666, "learning_rate": 0.0005, "loss": 2.097165584564209, "step": 6872 }, { "epoch": 0.5418210484824596, "grad_norm": 0.6773784756660461, "learning_rate": 0.0005, "loss": 2.1382832527160645, "step": 6873 }, { "epoch": 0.5418998817500985, "grad_norm": 0.7327675819396973, "learning_rate": 0.0005, "loss": 2.0060973167419434, "step": 6874 }, { "epoch": 0.5419787150177375, "grad_norm": 0.791810929775238, "learning_rate": 0.0005, "loss": 1.9868335723876953, "step": 6875 }, { "epoch": 0.5420575482853764, "grad_norm": 1.1104463338851929, "learning_rate": 0.0005, "loss": 1.9977977275848389, "step": 6876 }, { "epoch": 0.5421363815530154, "grad_norm": 0.7820784449577332, "learning_rate": 0.0005, "loss": 2.1120481491088867, "step": 6877 }, { "epoch": 0.5422152148206543, "grad_norm": 0.6737146973609924, "learning_rate": 0.0005, "loss": 1.946129322052002, "step": 6878 }, { "epoch": 0.5422940480882933, "grad_norm": 0.9176114201545715, "learning_rate": 0.0005, "loss": 2.108633518218994, "step": 6879 }, { "epoch": 0.5423728813559322, "grad_norm": 0.652513325214386, "learning_rate": 0.0005, "loss": 1.9233802556991577, "step": 6880 }, { "epoch": 0.5424517146235711, "grad_norm": 0.6475906372070312, "learning_rate": 0.0005, "loss": 1.9437174797058105, "step": 6881 }, { "epoch": 0.5425305478912101, "grad_norm": 0.8550512194633484, "learning_rate": 0.0005, "loss": 2.130683422088623, "step": 6882 }, { "epoch": 0.542609381158849, "grad_norm": 0.7163300514221191, "learning_rate": 0.0005, "loss": 2.1838974952697754, "step": 6883 }, { "epoch": 0.542688214426488, "grad_norm": 1.210755705833435, "learning_rate": 0.0005, "loss": 1.9631233215332031, "step": 6884 }, { "epoch": 0.5427670476941269, "grad_norm": 0.7504674196243286, "learning_rate": 0.0005, "loss": 1.9422725439071655, "step": 6885 }, { "epoch": 0.5428458809617659, "grad_norm": 0.767214834690094, "learning_rate": 0.0005, "loss": 1.8978935480117798, "step": 6886 }, { "epoch": 0.5429247142294048, "grad_norm": 0.6809638142585754, "learning_rate": 0.0005, "loss": 1.891648530960083, "step": 6887 }, { "epoch": 0.5430035474970437, "grad_norm": 0.7278615236282349, "learning_rate": 0.0005, "loss": 1.7727614641189575, "step": 6888 }, { "epoch": 0.5430823807646827, "grad_norm": 0.8663883805274963, "learning_rate": 0.0005, "loss": 2.1250555515289307, "step": 6889 }, { "epoch": 0.5431612140323216, "grad_norm": 1.0227657556533813, "learning_rate": 0.0005, "loss": 2.0934314727783203, "step": 6890 }, { "epoch": 0.5432400472999606, "grad_norm": 0.9256550073623657, "learning_rate": 0.0005, "loss": 1.906933307647705, "step": 6891 }, { "epoch": 0.5433188805675996, "grad_norm": 0.7045082449913025, "learning_rate": 0.0005, "loss": 2.227320671081543, "step": 6892 }, { "epoch": 0.5433977138352385, "grad_norm": 0.8264241814613342, "learning_rate": 0.0005, "loss": 1.918727159500122, "step": 6893 }, { "epoch": 0.5434765471028774, "grad_norm": 0.9504369497299194, "learning_rate": 0.0005, "loss": 1.9517836570739746, "step": 6894 }, { "epoch": 0.5435553803705163, "grad_norm": 0.9047439098358154, "learning_rate": 0.0005, "loss": 2.0637872219085693, "step": 6895 }, { "epoch": 0.5436342136381553, "grad_norm": 0.8042513728141785, "learning_rate": 0.0005, "loss": 1.9874515533447266, "step": 6896 }, { "epoch": 0.5437130469057943, "grad_norm": 0.605313777923584, "learning_rate": 0.0005, "loss": 1.823440670967102, "step": 6897 }, { "epoch": 0.5437918801734332, "grad_norm": 0.63932204246521, "learning_rate": 0.0005, "loss": 1.9665889739990234, "step": 6898 }, { "epoch": 0.5438707134410722, "grad_norm": 0.7801187038421631, "learning_rate": 0.0005, "loss": 2.2545368671417236, "step": 6899 }, { "epoch": 0.5439495467087111, "grad_norm": 0.6008855700492859, "learning_rate": 0.0005, "loss": 2.1692914962768555, "step": 6900 }, { "epoch": 0.54402837997635, "grad_norm": 0.7542037963867188, "learning_rate": 0.0005, "loss": 1.963324785232544, "step": 6901 }, { "epoch": 0.5441072132439889, "grad_norm": 0.6217300891876221, "learning_rate": 0.0005, "loss": 2.001901626586914, "step": 6902 }, { "epoch": 0.5441860465116279, "grad_norm": 0.6660045981407166, "learning_rate": 0.0005, "loss": 1.9075883626937866, "step": 6903 }, { "epoch": 0.5442648797792669, "grad_norm": 0.6076797842979431, "learning_rate": 0.0005, "loss": 1.8614877462387085, "step": 6904 }, { "epoch": 0.5443437130469058, "grad_norm": 0.754605770111084, "learning_rate": 0.0005, "loss": 2.1473631858825684, "step": 6905 }, { "epoch": 0.5444225463145448, "grad_norm": 0.6198930740356445, "learning_rate": 0.0005, "loss": 1.936955213546753, "step": 6906 }, { "epoch": 0.5445013795821837, "grad_norm": 0.6881319284439087, "learning_rate": 0.0005, "loss": 1.7936723232269287, "step": 6907 }, { "epoch": 0.5445802128498226, "grad_norm": 0.8897431492805481, "learning_rate": 0.0005, "loss": 2.12770938873291, "step": 6908 }, { "epoch": 0.5446590461174615, "grad_norm": 0.6759089827537537, "learning_rate": 0.0005, "loss": 2.2849628925323486, "step": 6909 }, { "epoch": 0.5447378793851005, "grad_norm": 0.9570245742797852, "learning_rate": 0.0005, "loss": 2.138399600982666, "step": 6910 }, { "epoch": 0.5448167126527395, "grad_norm": 0.5867388248443604, "learning_rate": 0.0005, "loss": 1.8590099811553955, "step": 6911 }, { "epoch": 0.5448955459203784, "grad_norm": 0.7532191276550293, "learning_rate": 0.0005, "loss": 1.9109165668487549, "step": 6912 }, { "epoch": 0.5449743791880174, "grad_norm": 0.6738330125808716, "learning_rate": 0.0005, "loss": 2.051910877227783, "step": 6913 }, { "epoch": 0.5450532124556563, "grad_norm": 0.7336247563362122, "learning_rate": 0.0005, "loss": 2.0599513053894043, "step": 6914 }, { "epoch": 0.5451320457232952, "grad_norm": 0.6870585680007935, "learning_rate": 0.0005, "loss": 2.107999563217163, "step": 6915 }, { "epoch": 0.5452108789909341, "grad_norm": 0.6679915189743042, "learning_rate": 0.0005, "loss": 1.8037046194076538, "step": 6916 }, { "epoch": 0.5452897122585731, "grad_norm": 0.7191512584686279, "learning_rate": 0.0005, "loss": 2.0622425079345703, "step": 6917 }, { "epoch": 0.5453685455262121, "grad_norm": 0.6806748509407043, "learning_rate": 0.0005, "loss": 2.0947959423065186, "step": 6918 }, { "epoch": 0.545447378793851, "grad_norm": 0.6521394848823547, "learning_rate": 0.0005, "loss": 1.9434900283813477, "step": 6919 }, { "epoch": 0.54552621206149, "grad_norm": 0.8829811811447144, "learning_rate": 0.0005, "loss": 2.3024613857269287, "step": 6920 }, { "epoch": 0.5456050453291289, "grad_norm": 0.6438430547714233, "learning_rate": 0.0005, "loss": 2.067405939102173, "step": 6921 }, { "epoch": 0.5456838785967678, "grad_norm": 0.6625534296035767, "learning_rate": 0.0005, "loss": 2.3185811042785645, "step": 6922 }, { "epoch": 0.5457627118644067, "grad_norm": 0.8339265584945679, "learning_rate": 0.0005, "loss": 2.1339645385742188, "step": 6923 }, { "epoch": 0.5458415451320457, "grad_norm": 0.6500088572502136, "learning_rate": 0.0005, "loss": 2.0741686820983887, "step": 6924 }, { "epoch": 0.5459203783996847, "grad_norm": 0.5428372025489807, "learning_rate": 0.0005, "loss": 2.013294219970703, "step": 6925 }, { "epoch": 0.5459992116673236, "grad_norm": 0.6539401412010193, "learning_rate": 0.0005, "loss": 1.9874396324157715, "step": 6926 }, { "epoch": 0.5460780449349626, "grad_norm": 0.9532127380371094, "learning_rate": 0.0005, "loss": 2.0004231929779053, "step": 6927 }, { "epoch": 0.5461568782026015, "grad_norm": 0.7972294092178345, "learning_rate": 0.0005, "loss": 2.139758586883545, "step": 6928 }, { "epoch": 0.5462357114702404, "grad_norm": 1.7058899402618408, "learning_rate": 0.0005, "loss": 1.8429274559020996, "step": 6929 }, { "epoch": 0.5463145447378794, "grad_norm": 0.5676557421684265, "learning_rate": 0.0005, "loss": 1.9983634948730469, "step": 6930 }, { "epoch": 0.5463933780055183, "grad_norm": 0.6577631235122681, "learning_rate": 0.0005, "loss": 1.7514406442642212, "step": 6931 }, { "epoch": 0.5464722112731573, "grad_norm": 0.788063108921051, "learning_rate": 0.0005, "loss": 2.346672773361206, "step": 6932 }, { "epoch": 0.5465510445407962, "grad_norm": 0.7445555925369263, "learning_rate": 0.0005, "loss": 1.9072613716125488, "step": 6933 }, { "epoch": 0.5466298778084352, "grad_norm": 0.6731053590774536, "learning_rate": 0.0005, "loss": 2.101861000061035, "step": 6934 }, { "epoch": 0.5467087110760741, "grad_norm": 0.6560541391372681, "learning_rate": 0.0005, "loss": 2.3535988330841064, "step": 6935 }, { "epoch": 0.546787544343713, "grad_norm": 0.6255850195884705, "learning_rate": 0.0005, "loss": 2.1470351219177246, "step": 6936 }, { "epoch": 0.546866377611352, "grad_norm": 0.659070611000061, "learning_rate": 0.0005, "loss": 1.940346121788025, "step": 6937 }, { "epoch": 0.546945210878991, "grad_norm": 0.7754195928573608, "learning_rate": 0.0005, "loss": 2.1747167110443115, "step": 6938 }, { "epoch": 0.5470240441466299, "grad_norm": 0.8321529626846313, "learning_rate": 0.0005, "loss": 1.8242905139923096, "step": 6939 }, { "epoch": 0.5471028774142688, "grad_norm": 0.6451398134231567, "learning_rate": 0.0005, "loss": 2.059593439102173, "step": 6940 }, { "epoch": 0.5471817106819078, "grad_norm": 0.5723848938941956, "learning_rate": 0.0005, "loss": 1.8002333641052246, "step": 6941 }, { "epoch": 0.5472605439495467, "grad_norm": 0.6809789538383484, "learning_rate": 0.0005, "loss": 2.081948757171631, "step": 6942 }, { "epoch": 0.5473393772171856, "grad_norm": 0.6647675037384033, "learning_rate": 0.0005, "loss": 1.784217119216919, "step": 6943 }, { "epoch": 0.5474182104848246, "grad_norm": 0.7599614262580872, "learning_rate": 0.0005, "loss": 1.824894905090332, "step": 6944 }, { "epoch": 0.5474970437524636, "grad_norm": 0.8483279943466187, "learning_rate": 0.0005, "loss": 2.423992156982422, "step": 6945 }, { "epoch": 0.5475758770201025, "grad_norm": 0.578178346157074, "learning_rate": 0.0005, "loss": 2.030937671661377, "step": 6946 }, { "epoch": 0.5476547102877414, "grad_norm": 0.6896923184394836, "learning_rate": 0.0005, "loss": 1.9151036739349365, "step": 6947 }, { "epoch": 0.5477335435553804, "grad_norm": 0.7183994650840759, "learning_rate": 0.0005, "loss": 1.9996678829193115, "step": 6948 }, { "epoch": 0.5478123768230193, "grad_norm": 0.7710168361663818, "learning_rate": 0.0005, "loss": 2.2074108123779297, "step": 6949 }, { "epoch": 0.5478912100906582, "grad_norm": 0.7267532348632812, "learning_rate": 0.0005, "loss": 2.020766258239746, "step": 6950 }, { "epoch": 0.5479700433582972, "grad_norm": 0.7948284149169922, "learning_rate": 0.0005, "loss": 2.0023016929626465, "step": 6951 }, { "epoch": 0.5480488766259362, "grad_norm": 0.6468748450279236, "learning_rate": 0.0005, "loss": 2.180391788482666, "step": 6952 }, { "epoch": 0.5481277098935751, "grad_norm": 0.5297496318817139, "learning_rate": 0.0005, "loss": 1.5731251239776611, "step": 6953 }, { "epoch": 0.548206543161214, "grad_norm": 0.6524786949157715, "learning_rate": 0.0005, "loss": 2.156641960144043, "step": 6954 }, { "epoch": 0.548285376428853, "grad_norm": 0.5903908610343933, "learning_rate": 0.0005, "loss": 1.5286493301391602, "step": 6955 }, { "epoch": 0.5483642096964919, "grad_norm": 0.6983802318572998, "learning_rate": 0.0005, "loss": 1.8294864892959595, "step": 6956 }, { "epoch": 0.5484430429641308, "grad_norm": 0.7607172727584839, "learning_rate": 0.0005, "loss": 1.968656301498413, "step": 6957 }, { "epoch": 0.5485218762317698, "grad_norm": 0.6906482577323914, "learning_rate": 0.0005, "loss": 2.1644814014434814, "step": 6958 }, { "epoch": 0.5486007094994088, "grad_norm": 0.6687275171279907, "learning_rate": 0.0005, "loss": 1.8575644493103027, "step": 6959 }, { "epoch": 0.5486795427670477, "grad_norm": 1.2455532550811768, "learning_rate": 0.0005, "loss": 2.034397840499878, "step": 6960 }, { "epoch": 0.5487583760346867, "grad_norm": 0.798221230506897, "learning_rate": 0.0005, "loss": 1.9089939594268799, "step": 6961 }, { "epoch": 0.5488372093023256, "grad_norm": 0.5839728713035583, "learning_rate": 0.0005, "loss": 1.7458453178405762, "step": 6962 }, { "epoch": 0.5489160425699645, "grad_norm": 0.7300747036933899, "learning_rate": 0.0005, "loss": 2.3481884002685547, "step": 6963 }, { "epoch": 0.5489948758376034, "grad_norm": 0.7087581157684326, "learning_rate": 0.0005, "loss": 2.1168758869171143, "step": 6964 }, { "epoch": 0.5490737091052424, "grad_norm": 0.6330181956291199, "learning_rate": 0.0005, "loss": 2.140828847885132, "step": 6965 }, { "epoch": 0.5491525423728814, "grad_norm": 1.0739021301269531, "learning_rate": 0.0005, "loss": 2.4401907920837402, "step": 6966 }, { "epoch": 0.5492313756405203, "grad_norm": 0.6712499856948853, "learning_rate": 0.0005, "loss": 1.7391231060028076, "step": 6967 }, { "epoch": 0.5493102089081593, "grad_norm": 0.8030682802200317, "learning_rate": 0.0005, "loss": 1.9072498083114624, "step": 6968 }, { "epoch": 0.5493890421757982, "grad_norm": 0.8308392763137817, "learning_rate": 0.0005, "loss": 2.1762804985046387, "step": 6969 }, { "epoch": 0.5494678754434371, "grad_norm": 0.9249051809310913, "learning_rate": 0.0005, "loss": 2.2133936882019043, "step": 6970 }, { "epoch": 0.549546708711076, "grad_norm": 0.7958228588104248, "learning_rate": 0.0005, "loss": 2.2463557720184326, "step": 6971 }, { "epoch": 0.549625541978715, "grad_norm": 0.7120843529701233, "learning_rate": 0.0005, "loss": 2.0025553703308105, "step": 6972 }, { "epoch": 0.549704375246354, "grad_norm": 1.0039877891540527, "learning_rate": 0.0005, "loss": 1.8984670639038086, "step": 6973 }, { "epoch": 0.5497832085139929, "grad_norm": 0.8937810063362122, "learning_rate": 0.0005, "loss": 2.1964797973632812, "step": 6974 }, { "epoch": 0.5498620417816319, "grad_norm": 0.6950346827507019, "learning_rate": 0.0005, "loss": 1.9339783191680908, "step": 6975 }, { "epoch": 0.5499408750492708, "grad_norm": 0.7052532434463501, "learning_rate": 0.0005, "loss": 1.8597469329833984, "step": 6976 }, { "epoch": 0.5500197083169097, "grad_norm": 0.7239198684692383, "learning_rate": 0.0005, "loss": 2.0149881839752197, "step": 6977 }, { "epoch": 0.5500985415845486, "grad_norm": 0.8070123791694641, "learning_rate": 0.0005, "loss": 2.0074684619903564, "step": 6978 }, { "epoch": 0.5501773748521877, "grad_norm": 0.7670649886131287, "learning_rate": 0.0005, "loss": 1.9647172689437866, "step": 6979 }, { "epoch": 0.5502562081198266, "grad_norm": 0.7112013697624207, "learning_rate": 0.0005, "loss": 2.040527820587158, "step": 6980 }, { "epoch": 0.5503350413874655, "grad_norm": 0.6959348320960999, "learning_rate": 0.0005, "loss": 1.772411584854126, "step": 6981 }, { "epoch": 0.5504138746551045, "grad_norm": 0.6963090300559998, "learning_rate": 0.0005, "loss": 2.0212771892547607, "step": 6982 }, { "epoch": 0.5504927079227434, "grad_norm": 0.6221656799316406, "learning_rate": 0.0005, "loss": 1.8695402145385742, "step": 6983 }, { "epoch": 0.5505715411903823, "grad_norm": 1.1560006141662598, "learning_rate": 0.0005, "loss": 2.247105836868286, "step": 6984 }, { "epoch": 0.5506503744580212, "grad_norm": 1.5962845087051392, "learning_rate": 0.0005, "loss": 1.9866597652435303, "step": 6985 }, { "epoch": 0.5507292077256603, "grad_norm": 0.9897283911705017, "learning_rate": 0.0005, "loss": 2.099613666534424, "step": 6986 }, { "epoch": 0.5508080409932992, "grad_norm": 0.9559526443481445, "learning_rate": 0.0005, "loss": 2.0165951251983643, "step": 6987 }, { "epoch": 0.5508868742609381, "grad_norm": 1.5882933139801025, "learning_rate": 0.0005, "loss": 2.342939853668213, "step": 6988 }, { "epoch": 0.5509657075285771, "grad_norm": 0.7424855828285217, "learning_rate": 0.0005, "loss": 2.018430233001709, "step": 6989 }, { "epoch": 0.551044540796216, "grad_norm": 0.8892948031425476, "learning_rate": 0.0005, "loss": 2.490832805633545, "step": 6990 }, { "epoch": 0.5511233740638549, "grad_norm": 0.7419086694717407, "learning_rate": 0.0005, "loss": 1.6776151657104492, "step": 6991 }, { "epoch": 0.5512022073314939, "grad_norm": 0.7252477407455444, "learning_rate": 0.0005, "loss": 2.263409376144409, "step": 6992 }, { "epoch": 0.5512810405991329, "grad_norm": 0.8854888677597046, "learning_rate": 0.0005, "loss": 2.350209951400757, "step": 6993 }, { "epoch": 0.5513598738667718, "grad_norm": 0.8374346494674683, "learning_rate": 0.0005, "loss": 2.0453691482543945, "step": 6994 }, { "epoch": 0.5514387071344107, "grad_norm": 0.7200425267219543, "learning_rate": 0.0005, "loss": 2.1866865158081055, "step": 6995 }, { "epoch": 0.5515175404020497, "grad_norm": 0.8443692922592163, "learning_rate": 0.0005, "loss": 2.255359172821045, "step": 6996 }, { "epoch": 0.5515963736696886, "grad_norm": 0.705954372882843, "learning_rate": 0.0005, "loss": 1.9598686695098877, "step": 6997 }, { "epoch": 0.5516752069373275, "grad_norm": 0.8520089387893677, "learning_rate": 0.0005, "loss": 2.0583200454711914, "step": 6998 }, { "epoch": 0.5517540402049665, "grad_norm": 0.8031508326530457, "learning_rate": 0.0005, "loss": 2.107187509536743, "step": 6999 }, { "epoch": 0.5518328734726055, "grad_norm": 0.8956488966941833, "learning_rate": 0.0005, "loss": 2.0950303077697754, "step": 7000 }, { "epoch": 0.5519117067402444, "grad_norm": 0.8055446147918701, "learning_rate": 0.0005, "loss": 1.9395768642425537, "step": 7001 }, { "epoch": 0.5519905400078833, "grad_norm": 0.7308546304702759, "learning_rate": 0.0005, "loss": 2.2683002948760986, "step": 7002 }, { "epoch": 0.5520693732755223, "grad_norm": 0.7762198448181152, "learning_rate": 0.0005, "loss": 2.045012950897217, "step": 7003 }, { "epoch": 0.5521482065431612, "grad_norm": 0.8373662829399109, "learning_rate": 0.0005, "loss": 2.1019649505615234, "step": 7004 }, { "epoch": 0.5522270398108001, "grad_norm": 0.7795401215553284, "learning_rate": 0.0005, "loss": 1.7557810544967651, "step": 7005 }, { "epoch": 0.5523058730784391, "grad_norm": 0.9440579414367676, "learning_rate": 0.0005, "loss": 2.318168878555298, "step": 7006 }, { "epoch": 0.5523847063460781, "grad_norm": 0.9834153652191162, "learning_rate": 0.0005, "loss": 2.3295493125915527, "step": 7007 }, { "epoch": 0.552463539613717, "grad_norm": 0.6635832786560059, "learning_rate": 0.0005, "loss": 1.984362244606018, "step": 7008 }, { "epoch": 0.5525423728813559, "grad_norm": 0.6579563617706299, "learning_rate": 0.0005, "loss": 2.153533935546875, "step": 7009 }, { "epoch": 0.5526212061489949, "grad_norm": 0.8319841027259827, "learning_rate": 0.0005, "loss": 1.6610004901885986, "step": 7010 }, { "epoch": 0.5527000394166338, "grad_norm": 1.0518782138824463, "learning_rate": 0.0005, "loss": 1.7370948791503906, "step": 7011 }, { "epoch": 0.5527788726842727, "grad_norm": 0.5672541856765747, "learning_rate": 0.0005, "loss": 1.9178683757781982, "step": 7012 }, { "epoch": 0.5528577059519117, "grad_norm": 0.5434147119522095, "learning_rate": 0.0005, "loss": 1.9017919301986694, "step": 7013 }, { "epoch": 0.5529365392195507, "grad_norm": 1.0902864933013916, "learning_rate": 0.0005, "loss": 2.122678279876709, "step": 7014 }, { "epoch": 0.5530153724871896, "grad_norm": 0.7515530586242676, "learning_rate": 0.0005, "loss": 1.8408527374267578, "step": 7015 }, { "epoch": 0.5530942057548285, "grad_norm": 0.663739025592804, "learning_rate": 0.0005, "loss": 1.9646645784378052, "step": 7016 }, { "epoch": 0.5531730390224675, "grad_norm": 0.5990853905677795, "learning_rate": 0.0005, "loss": 2.157010555267334, "step": 7017 }, { "epoch": 0.5532518722901064, "grad_norm": 0.7443552613258362, "learning_rate": 0.0005, "loss": 2.109614372253418, "step": 7018 }, { "epoch": 0.5533307055577453, "grad_norm": 0.7598957419395447, "learning_rate": 0.0005, "loss": 1.846020221710205, "step": 7019 }, { "epoch": 0.5534095388253844, "grad_norm": 0.8404775857925415, "learning_rate": 0.0005, "loss": 1.888386607170105, "step": 7020 }, { "epoch": 0.5534883720930233, "grad_norm": 0.6287032961845398, "learning_rate": 0.0005, "loss": 2.1748299598693848, "step": 7021 }, { "epoch": 0.5535672053606622, "grad_norm": 0.7812638878822327, "learning_rate": 0.0005, "loss": 2.052018642425537, "step": 7022 }, { "epoch": 0.5536460386283012, "grad_norm": 0.9819993376731873, "learning_rate": 0.0005, "loss": 2.297874927520752, "step": 7023 }, { "epoch": 0.5537248718959401, "grad_norm": 1.0793451070785522, "learning_rate": 0.0005, "loss": 1.6601983308792114, "step": 7024 }, { "epoch": 0.553803705163579, "grad_norm": 0.827719509601593, "learning_rate": 0.0005, "loss": 2.078706741333008, "step": 7025 }, { "epoch": 0.5538825384312179, "grad_norm": 0.6849817633628845, "learning_rate": 0.0005, "loss": 2.0608744621276855, "step": 7026 }, { "epoch": 0.553961371698857, "grad_norm": 0.5249306559562683, "learning_rate": 0.0005, "loss": 1.7647905349731445, "step": 7027 }, { "epoch": 0.5540402049664959, "grad_norm": 0.7940905094146729, "learning_rate": 0.0005, "loss": 2.0502235889434814, "step": 7028 }, { "epoch": 0.5541190382341348, "grad_norm": 1.0219075679779053, "learning_rate": 0.0005, "loss": 2.2245965003967285, "step": 7029 }, { "epoch": 0.5541978715017738, "grad_norm": 0.8868881464004517, "learning_rate": 0.0005, "loss": 2.201143503189087, "step": 7030 }, { "epoch": 0.5542767047694127, "grad_norm": 0.7181439995765686, "learning_rate": 0.0005, "loss": 2.0546226501464844, "step": 7031 }, { "epoch": 0.5543555380370516, "grad_norm": 0.7131105661392212, "learning_rate": 0.0005, "loss": 1.7952609062194824, "step": 7032 }, { "epoch": 0.5544343713046905, "grad_norm": 0.6906564831733704, "learning_rate": 0.0005, "loss": 2.0704591274261475, "step": 7033 }, { "epoch": 0.5545132045723296, "grad_norm": 0.6234737038612366, "learning_rate": 0.0005, "loss": 1.6513463258743286, "step": 7034 }, { "epoch": 0.5545920378399685, "grad_norm": 0.8420196771621704, "learning_rate": 0.0005, "loss": 2.0595595836639404, "step": 7035 }, { "epoch": 0.5546708711076074, "grad_norm": 1.1934038400650024, "learning_rate": 0.0005, "loss": 2.1807937622070312, "step": 7036 }, { "epoch": 0.5547497043752464, "grad_norm": 0.8281866908073425, "learning_rate": 0.0005, "loss": 1.7975081205368042, "step": 7037 }, { "epoch": 0.5548285376428853, "grad_norm": 0.7578425407409668, "learning_rate": 0.0005, "loss": 1.9809609651565552, "step": 7038 }, { "epoch": 0.5549073709105242, "grad_norm": 0.8746256232261658, "learning_rate": 0.0005, "loss": 1.891708493232727, "step": 7039 }, { "epoch": 0.5549862041781631, "grad_norm": 0.8060150742530823, "learning_rate": 0.0005, "loss": 2.164851665496826, "step": 7040 }, { "epoch": 0.5550650374458022, "grad_norm": 0.8040370941162109, "learning_rate": 0.0005, "loss": 2.1169559955596924, "step": 7041 }, { "epoch": 0.5551438707134411, "grad_norm": 0.8401045203208923, "learning_rate": 0.0005, "loss": 2.187516450881958, "step": 7042 }, { "epoch": 0.55522270398108, "grad_norm": 0.8711281418800354, "learning_rate": 0.0005, "loss": 2.101442813873291, "step": 7043 }, { "epoch": 0.555301537248719, "grad_norm": 0.8013982772827148, "learning_rate": 0.0005, "loss": 2.0112195014953613, "step": 7044 }, { "epoch": 0.5553803705163579, "grad_norm": 0.7634455561637878, "learning_rate": 0.0005, "loss": 2.0669524669647217, "step": 7045 }, { "epoch": 0.5554592037839968, "grad_norm": 0.8047888875007629, "learning_rate": 0.0005, "loss": 2.064761161804199, "step": 7046 }, { "epoch": 0.5555380370516357, "grad_norm": 0.7439361214637756, "learning_rate": 0.0005, "loss": 1.939265251159668, "step": 7047 }, { "epoch": 0.5556168703192748, "grad_norm": 0.6198379993438721, "learning_rate": 0.0005, "loss": 1.9129154682159424, "step": 7048 }, { "epoch": 0.5556957035869137, "grad_norm": 0.6594248414039612, "learning_rate": 0.0005, "loss": 1.8301246166229248, "step": 7049 }, { "epoch": 0.5557745368545526, "grad_norm": 0.6121141910552979, "learning_rate": 0.0005, "loss": 1.9709665775299072, "step": 7050 }, { "epoch": 0.5558533701221916, "grad_norm": 0.7731350660324097, "learning_rate": 0.0005, "loss": 1.843765139579773, "step": 7051 }, { "epoch": 0.5559322033898305, "grad_norm": 0.6953184604644775, "learning_rate": 0.0005, "loss": 1.6597139835357666, "step": 7052 }, { "epoch": 0.5560110366574694, "grad_norm": 0.6931308507919312, "learning_rate": 0.0005, "loss": 1.9822278022766113, "step": 7053 }, { "epoch": 0.5560898699251084, "grad_norm": 1.376166582107544, "learning_rate": 0.0005, "loss": 1.841270923614502, "step": 7054 }, { "epoch": 0.5561687031927474, "grad_norm": 0.6373425722122192, "learning_rate": 0.0005, "loss": 1.6638281345367432, "step": 7055 }, { "epoch": 0.5562475364603863, "grad_norm": 0.5963247418403625, "learning_rate": 0.0005, "loss": 2.0864908695220947, "step": 7056 }, { "epoch": 0.5563263697280252, "grad_norm": 0.61235511302948, "learning_rate": 0.0005, "loss": 2.0386247634887695, "step": 7057 }, { "epoch": 0.5564052029956642, "grad_norm": 0.7939590215682983, "learning_rate": 0.0005, "loss": 1.8664485216140747, "step": 7058 }, { "epoch": 0.5564840362633031, "grad_norm": 0.7871153354644775, "learning_rate": 0.0005, "loss": 1.9976246356964111, "step": 7059 }, { "epoch": 0.556562869530942, "grad_norm": 0.6208968162536621, "learning_rate": 0.0005, "loss": 2.0145041942596436, "step": 7060 }, { "epoch": 0.556641702798581, "grad_norm": 0.8640124797821045, "learning_rate": 0.0005, "loss": 2.0510969161987305, "step": 7061 }, { "epoch": 0.55672053606622, "grad_norm": 0.8045404553413391, "learning_rate": 0.0005, "loss": 1.996360182762146, "step": 7062 }, { "epoch": 0.5567993693338589, "grad_norm": 0.6732845306396484, "learning_rate": 0.0005, "loss": 2.023909330368042, "step": 7063 }, { "epoch": 0.5568782026014978, "grad_norm": 0.9630001187324524, "learning_rate": 0.0005, "loss": 1.7599319219589233, "step": 7064 }, { "epoch": 0.5569570358691368, "grad_norm": 0.686759889125824, "learning_rate": 0.0005, "loss": 1.7758859395980835, "step": 7065 }, { "epoch": 0.5570358691367757, "grad_norm": 0.7695621252059937, "learning_rate": 0.0005, "loss": 2.3064560890197754, "step": 7066 }, { "epoch": 0.5571147024044146, "grad_norm": 0.777127742767334, "learning_rate": 0.0005, "loss": 2.1519808769226074, "step": 7067 }, { "epoch": 0.5571935356720537, "grad_norm": 1.1219476461410522, "learning_rate": 0.0005, "loss": 2.350554943084717, "step": 7068 }, { "epoch": 0.5572723689396926, "grad_norm": 0.5482181310653687, "learning_rate": 0.0005, "loss": 1.9581904411315918, "step": 7069 }, { "epoch": 0.5573512022073315, "grad_norm": 0.6826770901679993, "learning_rate": 0.0005, "loss": 2.097052574157715, "step": 7070 }, { "epoch": 0.5574300354749704, "grad_norm": 0.6871082782745361, "learning_rate": 0.0005, "loss": 1.9079844951629639, "step": 7071 }, { "epoch": 0.5575088687426094, "grad_norm": 0.6251689791679382, "learning_rate": 0.0005, "loss": 2.102302312850952, "step": 7072 }, { "epoch": 0.5575877020102483, "grad_norm": 0.6356125473976135, "learning_rate": 0.0005, "loss": 1.8486084938049316, "step": 7073 }, { "epoch": 0.5576665352778872, "grad_norm": 0.6782674193382263, "learning_rate": 0.0005, "loss": 2.025928020477295, "step": 7074 }, { "epoch": 0.5577453685455263, "grad_norm": 0.6211305260658264, "learning_rate": 0.0005, "loss": 1.8402713537216187, "step": 7075 }, { "epoch": 0.5578242018131652, "grad_norm": 1.0135409832000732, "learning_rate": 0.0005, "loss": 2.0873796939849854, "step": 7076 }, { "epoch": 0.5579030350808041, "grad_norm": 0.840990424156189, "learning_rate": 0.0005, "loss": 2.150963544845581, "step": 7077 }, { "epoch": 0.557981868348443, "grad_norm": 0.7497901320457458, "learning_rate": 0.0005, "loss": 2.0462913513183594, "step": 7078 }, { "epoch": 0.558060701616082, "grad_norm": 0.6625010371208191, "learning_rate": 0.0005, "loss": 2.119959831237793, "step": 7079 }, { "epoch": 0.5581395348837209, "grad_norm": 0.623797595500946, "learning_rate": 0.0005, "loss": 1.765160083770752, "step": 7080 }, { "epoch": 0.5582183681513598, "grad_norm": 0.7251706719398499, "learning_rate": 0.0005, "loss": 1.9774634838104248, "step": 7081 }, { "epoch": 0.5582972014189989, "grad_norm": 0.9700659513473511, "learning_rate": 0.0005, "loss": 2.127131938934326, "step": 7082 }, { "epoch": 0.5583760346866378, "grad_norm": 0.7018996477127075, "learning_rate": 0.0005, "loss": 1.9698340892791748, "step": 7083 }, { "epoch": 0.5584548679542767, "grad_norm": 0.6807276606559753, "learning_rate": 0.0005, "loss": 2.003476619720459, "step": 7084 }, { "epoch": 0.5585337012219157, "grad_norm": 0.7308688163757324, "learning_rate": 0.0005, "loss": 2.0622591972351074, "step": 7085 }, { "epoch": 0.5586125344895546, "grad_norm": 0.6135713458061218, "learning_rate": 0.0005, "loss": 2.11197829246521, "step": 7086 }, { "epoch": 0.5586913677571935, "grad_norm": 0.6853305101394653, "learning_rate": 0.0005, "loss": 2.186314105987549, "step": 7087 }, { "epoch": 0.5587702010248324, "grad_norm": 0.6533026099205017, "learning_rate": 0.0005, "loss": 2.100142478942871, "step": 7088 }, { "epoch": 0.5588490342924715, "grad_norm": 1.1022076606750488, "learning_rate": 0.0005, "loss": 1.8049840927124023, "step": 7089 }, { "epoch": 0.5589278675601104, "grad_norm": 0.847603440284729, "learning_rate": 0.0005, "loss": 2.595362424850464, "step": 7090 }, { "epoch": 0.5590067008277493, "grad_norm": 0.677352786064148, "learning_rate": 0.0005, "loss": 2.1535747051239014, "step": 7091 }, { "epoch": 0.5590855340953883, "grad_norm": 0.5670561790466309, "learning_rate": 0.0005, "loss": 1.7659823894500732, "step": 7092 }, { "epoch": 0.5591643673630272, "grad_norm": 0.9228956699371338, "learning_rate": 0.0005, "loss": 2.4025938510894775, "step": 7093 }, { "epoch": 0.5592432006306661, "grad_norm": 1.128955364227295, "learning_rate": 0.0005, "loss": 1.8301531076431274, "step": 7094 }, { "epoch": 0.559322033898305, "grad_norm": 0.6681258082389832, "learning_rate": 0.0005, "loss": 1.9962714910507202, "step": 7095 }, { "epoch": 0.5594008671659441, "grad_norm": 0.6630362272262573, "learning_rate": 0.0005, "loss": 1.94966459274292, "step": 7096 }, { "epoch": 0.559479700433583, "grad_norm": 0.8412531018257141, "learning_rate": 0.0005, "loss": 2.1072678565979004, "step": 7097 }, { "epoch": 0.5595585337012219, "grad_norm": 0.7824633717536926, "learning_rate": 0.0005, "loss": 2.2329275608062744, "step": 7098 }, { "epoch": 0.5596373669688609, "grad_norm": 0.6345484852790833, "learning_rate": 0.0005, "loss": 1.958041787147522, "step": 7099 }, { "epoch": 0.5597162002364998, "grad_norm": 0.743863046169281, "learning_rate": 0.0005, "loss": 2.2398593425750732, "step": 7100 }, { "epoch": 0.5597950335041387, "grad_norm": 0.8407720923423767, "learning_rate": 0.0005, "loss": 2.197981834411621, "step": 7101 }, { "epoch": 0.5598738667717776, "grad_norm": 0.6107566356658936, "learning_rate": 0.0005, "loss": 1.8064823150634766, "step": 7102 }, { "epoch": 0.5599527000394167, "grad_norm": 1.2141004800796509, "learning_rate": 0.0005, "loss": 2.035881280899048, "step": 7103 }, { "epoch": 0.5600315333070556, "grad_norm": 1.0307745933532715, "learning_rate": 0.0005, "loss": 2.291688919067383, "step": 7104 }, { "epoch": 0.5601103665746945, "grad_norm": 1.0330382585525513, "learning_rate": 0.0005, "loss": 2.1529152393341064, "step": 7105 }, { "epoch": 0.5601891998423335, "grad_norm": 0.7215076684951782, "learning_rate": 0.0005, "loss": 2.134774684906006, "step": 7106 }, { "epoch": 0.5602680331099724, "grad_norm": 1.0431352853775024, "learning_rate": 0.0005, "loss": 2.3280932903289795, "step": 7107 }, { "epoch": 0.5603468663776113, "grad_norm": 0.8804777264595032, "learning_rate": 0.0005, "loss": 2.3306169509887695, "step": 7108 }, { "epoch": 0.5604256996452502, "grad_norm": 0.9854072332382202, "learning_rate": 0.0005, "loss": 2.0992279052734375, "step": 7109 }, { "epoch": 0.5605045329128893, "grad_norm": 0.8334928750991821, "learning_rate": 0.0005, "loss": 2.206376075744629, "step": 7110 }, { "epoch": 0.5605833661805282, "grad_norm": 0.5090861916542053, "learning_rate": 0.0005, "loss": 2.096320152282715, "step": 7111 }, { "epoch": 0.5606621994481671, "grad_norm": 0.6865389347076416, "learning_rate": 0.0005, "loss": 1.8176332712173462, "step": 7112 }, { "epoch": 0.5607410327158061, "grad_norm": 0.6756151914596558, "learning_rate": 0.0005, "loss": 2.129903793334961, "step": 7113 }, { "epoch": 0.560819865983445, "grad_norm": 0.9188939332962036, "learning_rate": 0.0005, "loss": 2.3781423568725586, "step": 7114 }, { "epoch": 0.5608986992510839, "grad_norm": 0.6438277363777161, "learning_rate": 0.0005, "loss": 1.7830610275268555, "step": 7115 }, { "epoch": 0.560977532518723, "grad_norm": 0.7644033432006836, "learning_rate": 0.0005, "loss": 1.985870361328125, "step": 7116 }, { "epoch": 0.5610563657863619, "grad_norm": 0.7723543643951416, "learning_rate": 0.0005, "loss": 2.0287716388702393, "step": 7117 }, { "epoch": 0.5611351990540008, "grad_norm": 0.9497147798538208, "learning_rate": 0.0005, "loss": 2.176939010620117, "step": 7118 }, { "epoch": 0.5612140323216397, "grad_norm": 1.1413228511810303, "learning_rate": 0.0005, "loss": 1.958866000175476, "step": 7119 }, { "epoch": 0.5612928655892787, "grad_norm": 1.058180332183838, "learning_rate": 0.0005, "loss": 2.0941238403320312, "step": 7120 }, { "epoch": 0.5613716988569176, "grad_norm": 0.7299509048461914, "learning_rate": 0.0005, "loss": 2.0545692443847656, "step": 7121 }, { "epoch": 0.5614505321245565, "grad_norm": 0.6358416676521301, "learning_rate": 0.0005, "loss": 1.9642970561981201, "step": 7122 }, { "epoch": 0.5615293653921956, "grad_norm": 1.0099726915359497, "learning_rate": 0.0005, "loss": 2.027726650238037, "step": 7123 }, { "epoch": 0.5616081986598345, "grad_norm": 0.6735865473747253, "learning_rate": 0.0005, "loss": 2.022716999053955, "step": 7124 }, { "epoch": 0.5616870319274734, "grad_norm": 0.6625887751579285, "learning_rate": 0.0005, "loss": 2.194211483001709, "step": 7125 }, { "epoch": 0.5617658651951123, "grad_norm": 0.7519993185997009, "learning_rate": 0.0005, "loss": 2.007894277572632, "step": 7126 }, { "epoch": 0.5618446984627513, "grad_norm": 0.941746175289154, "learning_rate": 0.0005, "loss": 1.8539915084838867, "step": 7127 }, { "epoch": 0.5619235317303902, "grad_norm": 1.0250933170318604, "learning_rate": 0.0005, "loss": 1.9842467308044434, "step": 7128 }, { "epoch": 0.5620023649980291, "grad_norm": 0.6518638134002686, "learning_rate": 0.0005, "loss": 1.8782289028167725, "step": 7129 }, { "epoch": 0.5620811982656682, "grad_norm": 0.635904848575592, "learning_rate": 0.0005, "loss": 1.980825662612915, "step": 7130 }, { "epoch": 0.5621600315333071, "grad_norm": 0.6065253019332886, "learning_rate": 0.0005, "loss": 1.930870532989502, "step": 7131 }, { "epoch": 0.562238864800946, "grad_norm": 0.6511166095733643, "learning_rate": 0.0005, "loss": 1.9255685806274414, "step": 7132 }, { "epoch": 0.5623176980685849, "grad_norm": 0.5958336591720581, "learning_rate": 0.0005, "loss": 1.76760995388031, "step": 7133 }, { "epoch": 0.5623965313362239, "grad_norm": 0.7596738338470459, "learning_rate": 0.0005, "loss": 2.082411527633667, "step": 7134 }, { "epoch": 0.5624753646038628, "grad_norm": 0.6162387132644653, "learning_rate": 0.0005, "loss": 2.094810724258423, "step": 7135 }, { "epoch": 0.5625541978715017, "grad_norm": 0.5308334827423096, "learning_rate": 0.0005, "loss": 1.747053861618042, "step": 7136 }, { "epoch": 0.5626330311391408, "grad_norm": 0.6340474486351013, "learning_rate": 0.0005, "loss": 2.1250686645507812, "step": 7137 }, { "epoch": 0.5627118644067797, "grad_norm": 0.6859188079833984, "learning_rate": 0.0005, "loss": 1.9196643829345703, "step": 7138 }, { "epoch": 0.5627906976744186, "grad_norm": 1.0292028188705444, "learning_rate": 0.0005, "loss": 1.8154945373535156, "step": 7139 }, { "epoch": 0.5628695309420575, "grad_norm": 0.7736685276031494, "learning_rate": 0.0005, "loss": 2.3550095558166504, "step": 7140 }, { "epoch": 0.5629483642096965, "grad_norm": 0.9030925035476685, "learning_rate": 0.0005, "loss": 2.08794903755188, "step": 7141 }, { "epoch": 0.5630271974773354, "grad_norm": 0.5959968566894531, "learning_rate": 0.0005, "loss": 2.0044949054718018, "step": 7142 }, { "epoch": 0.5631060307449743, "grad_norm": 0.721038818359375, "learning_rate": 0.0005, "loss": 1.7702419757843018, "step": 7143 }, { "epoch": 0.5631848640126134, "grad_norm": 0.7017906308174133, "learning_rate": 0.0005, "loss": 1.7628428936004639, "step": 7144 }, { "epoch": 0.5632636972802523, "grad_norm": 0.7073724269866943, "learning_rate": 0.0005, "loss": 1.9704673290252686, "step": 7145 }, { "epoch": 0.5633425305478912, "grad_norm": 0.6541709899902344, "learning_rate": 0.0005, "loss": 2.034179925918579, "step": 7146 }, { "epoch": 0.5634213638155302, "grad_norm": 1.1006207466125488, "learning_rate": 0.0005, "loss": 1.8058558702468872, "step": 7147 }, { "epoch": 0.5635001970831691, "grad_norm": 0.7960951328277588, "learning_rate": 0.0005, "loss": 1.869642972946167, "step": 7148 }, { "epoch": 0.563579030350808, "grad_norm": 0.7943262457847595, "learning_rate": 0.0005, "loss": 2.140192985534668, "step": 7149 }, { "epoch": 0.5636578636184469, "grad_norm": 0.6260942220687866, "learning_rate": 0.0005, "loss": 1.9520540237426758, "step": 7150 }, { "epoch": 0.563736696886086, "grad_norm": 0.6648343205451965, "learning_rate": 0.0005, "loss": 1.9470634460449219, "step": 7151 }, { "epoch": 0.5638155301537249, "grad_norm": 0.6436957716941833, "learning_rate": 0.0005, "loss": 1.8867945671081543, "step": 7152 }, { "epoch": 0.5638943634213638, "grad_norm": 0.9034432172775269, "learning_rate": 0.0005, "loss": 1.9995691776275635, "step": 7153 }, { "epoch": 0.5639731966890028, "grad_norm": 0.7258438467979431, "learning_rate": 0.0005, "loss": 1.9020318984985352, "step": 7154 }, { "epoch": 0.5640520299566417, "grad_norm": 0.5761942863464355, "learning_rate": 0.0005, "loss": 1.7310616970062256, "step": 7155 }, { "epoch": 0.5641308632242806, "grad_norm": 0.8416871428489685, "learning_rate": 0.0005, "loss": 1.91204833984375, "step": 7156 }, { "epoch": 0.5642096964919195, "grad_norm": 1.0562297105789185, "learning_rate": 0.0005, "loss": 2.373777151107788, "step": 7157 }, { "epoch": 0.5642885297595586, "grad_norm": 0.720992386341095, "learning_rate": 0.0005, "loss": 2.042649507522583, "step": 7158 }, { "epoch": 0.5643673630271975, "grad_norm": 1.0672909021377563, "learning_rate": 0.0005, "loss": 1.7896307706832886, "step": 7159 }, { "epoch": 0.5644461962948364, "grad_norm": 0.7038190364837646, "learning_rate": 0.0005, "loss": 1.9102718830108643, "step": 7160 }, { "epoch": 0.5645250295624754, "grad_norm": 1.0477739572525024, "learning_rate": 0.0005, "loss": 1.987891435623169, "step": 7161 }, { "epoch": 0.5646038628301143, "grad_norm": 0.9589520692825317, "learning_rate": 0.0005, "loss": 2.092085838317871, "step": 7162 }, { "epoch": 0.5646826960977532, "grad_norm": 1.4473313093185425, "learning_rate": 0.0005, "loss": 2.206543207168579, "step": 7163 }, { "epoch": 0.5647615293653921, "grad_norm": 0.6588624119758606, "learning_rate": 0.0005, "loss": 2.006016254425049, "step": 7164 }, { "epoch": 0.5648403626330312, "grad_norm": 0.608857274055481, "learning_rate": 0.0005, "loss": 1.8708367347717285, "step": 7165 }, { "epoch": 0.5649191959006701, "grad_norm": 0.707084059715271, "learning_rate": 0.0005, "loss": 2.1290698051452637, "step": 7166 }, { "epoch": 0.564998029168309, "grad_norm": 0.7218282222747803, "learning_rate": 0.0005, "loss": 1.9544827938079834, "step": 7167 }, { "epoch": 0.565076862435948, "grad_norm": 0.6174373626708984, "learning_rate": 0.0005, "loss": 1.9219435453414917, "step": 7168 }, { "epoch": 0.5651556957035869, "grad_norm": 0.8291468620300293, "learning_rate": 0.0005, "loss": 1.9793484210968018, "step": 7169 }, { "epoch": 0.5652345289712258, "grad_norm": 1.7333507537841797, "learning_rate": 0.0005, "loss": 2.273690700531006, "step": 7170 }, { "epoch": 0.5653133622388647, "grad_norm": 0.7671072483062744, "learning_rate": 0.0005, "loss": 2.040407180786133, "step": 7171 }, { "epoch": 0.5653921955065038, "grad_norm": 0.697902500629425, "learning_rate": 0.0005, "loss": 1.9415916204452515, "step": 7172 }, { "epoch": 0.5654710287741427, "grad_norm": 0.6632758378982544, "learning_rate": 0.0005, "loss": 1.7881393432617188, "step": 7173 }, { "epoch": 0.5655498620417816, "grad_norm": 0.7314077615737915, "learning_rate": 0.0005, "loss": 1.8361808061599731, "step": 7174 }, { "epoch": 0.5656286953094206, "grad_norm": 0.6773676872253418, "learning_rate": 0.0005, "loss": 1.9892323017120361, "step": 7175 }, { "epoch": 0.5657075285770595, "grad_norm": 0.6497194170951843, "learning_rate": 0.0005, "loss": 1.9902327060699463, "step": 7176 }, { "epoch": 0.5657863618446984, "grad_norm": 0.6815332174301147, "learning_rate": 0.0005, "loss": 2.101713180541992, "step": 7177 }, { "epoch": 0.5658651951123375, "grad_norm": 0.8482638597488403, "learning_rate": 0.0005, "loss": 2.217622756958008, "step": 7178 }, { "epoch": 0.5659440283799764, "grad_norm": 0.6378570199012756, "learning_rate": 0.0005, "loss": 1.9469082355499268, "step": 7179 }, { "epoch": 0.5660228616476153, "grad_norm": 0.9524834156036377, "learning_rate": 0.0005, "loss": 2.2052063941955566, "step": 7180 }, { "epoch": 0.5661016949152542, "grad_norm": 0.6217897534370422, "learning_rate": 0.0005, "loss": 1.8304334878921509, "step": 7181 }, { "epoch": 0.5661805281828932, "grad_norm": 0.7337222695350647, "learning_rate": 0.0005, "loss": 1.978403091430664, "step": 7182 }, { "epoch": 0.5662593614505321, "grad_norm": 0.7838826775550842, "learning_rate": 0.0005, "loss": 1.9794142246246338, "step": 7183 }, { "epoch": 0.566338194718171, "grad_norm": 3.052971839904785, "learning_rate": 0.0005, "loss": 1.8281686305999756, "step": 7184 }, { "epoch": 0.5664170279858101, "grad_norm": 0.8419924378395081, "learning_rate": 0.0005, "loss": 1.990943431854248, "step": 7185 }, { "epoch": 0.566495861253449, "grad_norm": 0.7275838851928711, "learning_rate": 0.0005, "loss": 1.9104804992675781, "step": 7186 }, { "epoch": 0.5665746945210879, "grad_norm": 0.8450642824172974, "learning_rate": 0.0005, "loss": 1.9627349376678467, "step": 7187 }, { "epoch": 0.5666535277887268, "grad_norm": 0.8182643055915833, "learning_rate": 0.0005, "loss": 2.1881327629089355, "step": 7188 }, { "epoch": 0.5667323610563658, "grad_norm": 0.825054943561554, "learning_rate": 0.0005, "loss": 2.0643057823181152, "step": 7189 }, { "epoch": 0.5668111943240047, "grad_norm": 2.351330041885376, "learning_rate": 0.0005, "loss": 2.163797616958618, "step": 7190 }, { "epoch": 0.5668900275916436, "grad_norm": 0.7821138501167297, "learning_rate": 0.0005, "loss": 1.8055343627929688, "step": 7191 }, { "epoch": 0.5669688608592827, "grad_norm": 0.7546339631080627, "learning_rate": 0.0005, "loss": 2.2230820655822754, "step": 7192 }, { "epoch": 0.5670476941269216, "grad_norm": 0.9100866913795471, "learning_rate": 0.0005, "loss": 2.0717613697052, "step": 7193 }, { "epoch": 0.5671265273945605, "grad_norm": 0.8858117461204529, "learning_rate": 0.0005, "loss": 1.9798660278320312, "step": 7194 }, { "epoch": 0.5672053606621994, "grad_norm": 0.7227451801300049, "learning_rate": 0.0005, "loss": 2.0360422134399414, "step": 7195 }, { "epoch": 0.5672841939298384, "grad_norm": 0.6866704225540161, "learning_rate": 0.0005, "loss": 1.9001743793487549, "step": 7196 }, { "epoch": 0.5673630271974773, "grad_norm": 0.8113340139389038, "learning_rate": 0.0005, "loss": 2.0961294174194336, "step": 7197 }, { "epoch": 0.5674418604651162, "grad_norm": 0.6711322069168091, "learning_rate": 0.0005, "loss": 1.963693380355835, "step": 7198 }, { "epoch": 0.5675206937327553, "grad_norm": 0.7973064184188843, "learning_rate": 0.0005, "loss": 1.6781446933746338, "step": 7199 }, { "epoch": 0.5675995270003942, "grad_norm": 0.9616238474845886, "learning_rate": 0.0005, "loss": 1.929955244064331, "step": 7200 }, { "epoch": 0.5676783602680331, "grad_norm": 0.6648404002189636, "learning_rate": 0.0005, "loss": 2.278479814529419, "step": 7201 }, { "epoch": 0.567757193535672, "grad_norm": 0.5746452808380127, "learning_rate": 0.0005, "loss": 1.831319808959961, "step": 7202 }, { "epoch": 0.567836026803311, "grad_norm": 0.9376950263977051, "learning_rate": 0.0005, "loss": 2.189807415008545, "step": 7203 }, { "epoch": 0.5679148600709499, "grad_norm": 0.6289094686508179, "learning_rate": 0.0005, "loss": 1.79920494556427, "step": 7204 }, { "epoch": 0.5679936933385888, "grad_norm": 0.6465677618980408, "learning_rate": 0.0005, "loss": 1.6368746757507324, "step": 7205 }, { "epoch": 0.5680725266062279, "grad_norm": 0.8584327697753906, "learning_rate": 0.0005, "loss": 2.252951145172119, "step": 7206 }, { "epoch": 0.5681513598738668, "grad_norm": 1.2631033658981323, "learning_rate": 0.0005, "loss": 2.2286534309387207, "step": 7207 }, { "epoch": 0.5682301931415057, "grad_norm": 0.9706670641899109, "learning_rate": 0.0005, "loss": 2.0816245079040527, "step": 7208 }, { "epoch": 0.5683090264091447, "grad_norm": 0.7633174657821655, "learning_rate": 0.0005, "loss": 1.93446946144104, "step": 7209 }, { "epoch": 0.5683878596767836, "grad_norm": 0.7945253252983093, "learning_rate": 0.0005, "loss": 1.897226095199585, "step": 7210 }, { "epoch": 0.5684666929444225, "grad_norm": 0.9037158489227295, "learning_rate": 0.0005, "loss": 2.1185057163238525, "step": 7211 }, { "epoch": 0.5685455262120614, "grad_norm": 0.8852322697639465, "learning_rate": 0.0005, "loss": 2.226557493209839, "step": 7212 }, { "epoch": 0.5686243594797005, "grad_norm": 0.9734306335449219, "learning_rate": 0.0005, "loss": 2.0781826972961426, "step": 7213 }, { "epoch": 0.5687031927473394, "grad_norm": 0.7097362279891968, "learning_rate": 0.0005, "loss": 2.2608320713043213, "step": 7214 }, { "epoch": 0.5687820260149783, "grad_norm": 0.8097958564758301, "learning_rate": 0.0005, "loss": 1.9102660417556763, "step": 7215 }, { "epoch": 0.5688608592826173, "grad_norm": 0.7459951043128967, "learning_rate": 0.0005, "loss": 2.118635654449463, "step": 7216 }, { "epoch": 0.5689396925502562, "grad_norm": 0.5222042202949524, "learning_rate": 0.0005, "loss": 1.901375412940979, "step": 7217 }, { "epoch": 0.5690185258178951, "grad_norm": 0.7508121132850647, "learning_rate": 0.0005, "loss": 2.147613286972046, "step": 7218 }, { "epoch": 0.569097359085534, "grad_norm": 1.0392223596572876, "learning_rate": 0.0005, "loss": 1.9697788953781128, "step": 7219 }, { "epoch": 0.5691761923531731, "grad_norm": 0.7484208345413208, "learning_rate": 0.0005, "loss": 2.3037619590759277, "step": 7220 }, { "epoch": 0.569255025620812, "grad_norm": 0.7310312986373901, "learning_rate": 0.0005, "loss": 1.7597565650939941, "step": 7221 }, { "epoch": 0.5693338588884509, "grad_norm": 0.7127040028572083, "learning_rate": 0.0005, "loss": 1.913132667541504, "step": 7222 }, { "epoch": 0.5694126921560899, "grad_norm": 0.6935757398605347, "learning_rate": 0.0005, "loss": 1.7326464653015137, "step": 7223 }, { "epoch": 0.5694915254237288, "grad_norm": 0.6168098449707031, "learning_rate": 0.0005, "loss": 2.1566991806030273, "step": 7224 }, { "epoch": 0.5695703586913677, "grad_norm": 0.6384440064430237, "learning_rate": 0.0005, "loss": 1.7519328594207764, "step": 7225 }, { "epoch": 0.5696491919590067, "grad_norm": 0.6227549314498901, "learning_rate": 0.0005, "loss": 1.7305487394332886, "step": 7226 }, { "epoch": 0.5697280252266457, "grad_norm": 0.7912190556526184, "learning_rate": 0.0005, "loss": 1.906797170639038, "step": 7227 }, { "epoch": 0.5698068584942846, "grad_norm": 0.7371187806129456, "learning_rate": 0.0005, "loss": 1.654822587966919, "step": 7228 }, { "epoch": 0.5698856917619235, "grad_norm": 0.6801797151565552, "learning_rate": 0.0005, "loss": 2.1599321365356445, "step": 7229 }, { "epoch": 0.5699645250295625, "grad_norm": 0.7125924229621887, "learning_rate": 0.0005, "loss": 2.008955478668213, "step": 7230 }, { "epoch": 0.5700433582972014, "grad_norm": 0.8601839542388916, "learning_rate": 0.0005, "loss": 2.210599422454834, "step": 7231 }, { "epoch": 0.5701221915648403, "grad_norm": 0.6403365731239319, "learning_rate": 0.0005, "loss": 1.9512009620666504, "step": 7232 }, { "epoch": 0.5702010248324793, "grad_norm": 0.6632677316665649, "learning_rate": 0.0005, "loss": 1.9852650165557861, "step": 7233 }, { "epoch": 0.5702798581001183, "grad_norm": 0.6096879243850708, "learning_rate": 0.0005, "loss": 1.8797169923782349, "step": 7234 }, { "epoch": 0.5703586913677572, "grad_norm": 0.5860570073127747, "learning_rate": 0.0005, "loss": 1.84175705909729, "step": 7235 }, { "epoch": 0.5704375246353961, "grad_norm": 0.7204781174659729, "learning_rate": 0.0005, "loss": 2.0733909606933594, "step": 7236 }, { "epoch": 0.5705163579030351, "grad_norm": 0.5683663487434387, "learning_rate": 0.0005, "loss": 1.9385902881622314, "step": 7237 }, { "epoch": 0.570595191170674, "grad_norm": 0.8185559511184692, "learning_rate": 0.0005, "loss": 2.22456955909729, "step": 7238 }, { "epoch": 0.5706740244383129, "grad_norm": 0.6701527833938599, "learning_rate": 0.0005, "loss": 2.06687068939209, "step": 7239 }, { "epoch": 0.570752857705952, "grad_norm": 0.8556898236274719, "learning_rate": 0.0005, "loss": 1.9547452926635742, "step": 7240 }, { "epoch": 0.5708316909735909, "grad_norm": 0.9590785503387451, "learning_rate": 0.0005, "loss": 2.133150339126587, "step": 7241 }, { "epoch": 0.5709105242412298, "grad_norm": 1.1888842582702637, "learning_rate": 0.0005, "loss": 2.0173208713531494, "step": 7242 }, { "epoch": 0.5709893575088687, "grad_norm": 0.6406176686286926, "learning_rate": 0.0005, "loss": 1.8251280784606934, "step": 7243 }, { "epoch": 0.5710681907765077, "grad_norm": 0.7479535341262817, "learning_rate": 0.0005, "loss": 2.12619686126709, "step": 7244 }, { "epoch": 0.5711470240441466, "grad_norm": 0.6327852606773376, "learning_rate": 0.0005, "loss": 1.9432761669158936, "step": 7245 }, { "epoch": 0.5712258573117855, "grad_norm": 0.8145601153373718, "learning_rate": 0.0005, "loss": 2.053366184234619, "step": 7246 }, { "epoch": 0.5713046905794246, "grad_norm": 0.7006847858428955, "learning_rate": 0.0005, "loss": 1.8716294765472412, "step": 7247 }, { "epoch": 0.5713835238470635, "grad_norm": 0.6307551264762878, "learning_rate": 0.0005, "loss": 1.9519559144973755, "step": 7248 }, { "epoch": 0.5714623571147024, "grad_norm": 0.7702152729034424, "learning_rate": 0.0005, "loss": 2.1740469932556152, "step": 7249 }, { "epoch": 0.5715411903823413, "grad_norm": 0.6485022902488708, "learning_rate": 0.0005, "loss": 2.150695323944092, "step": 7250 }, { "epoch": 0.5716200236499803, "grad_norm": 0.7497897744178772, "learning_rate": 0.0005, "loss": 2.2256433963775635, "step": 7251 }, { "epoch": 0.5716988569176192, "grad_norm": 0.9960232973098755, "learning_rate": 0.0005, "loss": 1.9089574813842773, "step": 7252 }, { "epoch": 0.5717776901852581, "grad_norm": 0.8689825534820557, "learning_rate": 0.0005, "loss": 1.9139195680618286, "step": 7253 }, { "epoch": 0.5718565234528972, "grad_norm": 0.6583952307701111, "learning_rate": 0.0005, "loss": 2.142030715942383, "step": 7254 }, { "epoch": 0.5719353567205361, "grad_norm": 0.7891757488250732, "learning_rate": 0.0005, "loss": 1.8811944723129272, "step": 7255 }, { "epoch": 0.572014189988175, "grad_norm": 1.1650649309158325, "learning_rate": 0.0005, "loss": 2.2429263591766357, "step": 7256 }, { "epoch": 0.5720930232558139, "grad_norm": 0.6552821397781372, "learning_rate": 0.0005, "loss": 1.9767847061157227, "step": 7257 }, { "epoch": 0.5721718565234529, "grad_norm": 0.7827199697494507, "learning_rate": 0.0005, "loss": 2.0692341327667236, "step": 7258 }, { "epoch": 0.5722506897910918, "grad_norm": 0.7335914969444275, "learning_rate": 0.0005, "loss": 1.8968009948730469, "step": 7259 }, { "epoch": 0.5723295230587307, "grad_norm": 0.677226722240448, "learning_rate": 0.0005, "loss": 1.7797282934188843, "step": 7260 }, { "epoch": 0.5724083563263698, "grad_norm": 0.7340863347053528, "learning_rate": 0.0005, "loss": 1.8015697002410889, "step": 7261 }, { "epoch": 0.5724871895940087, "grad_norm": 0.6552121043205261, "learning_rate": 0.0005, "loss": 1.975207805633545, "step": 7262 }, { "epoch": 0.5725660228616476, "grad_norm": 0.5723315477371216, "learning_rate": 0.0005, "loss": 1.6592364311218262, "step": 7263 }, { "epoch": 0.5726448561292865, "grad_norm": 0.6072494387626648, "learning_rate": 0.0005, "loss": 2.0725345611572266, "step": 7264 }, { "epoch": 0.5727236893969255, "grad_norm": 0.7320535182952881, "learning_rate": 0.0005, "loss": 2.1023008823394775, "step": 7265 }, { "epoch": 0.5728025226645644, "grad_norm": 0.658835768699646, "learning_rate": 0.0005, "loss": 1.970767617225647, "step": 7266 }, { "epoch": 0.5728813559322034, "grad_norm": 0.762559175491333, "learning_rate": 0.0005, "loss": 1.9513497352600098, "step": 7267 }, { "epoch": 0.5729601891998424, "grad_norm": 0.8043250441551208, "learning_rate": 0.0005, "loss": 2.239436626434326, "step": 7268 }, { "epoch": 0.5730390224674813, "grad_norm": 0.6155537962913513, "learning_rate": 0.0005, "loss": 1.8907922506332397, "step": 7269 }, { "epoch": 0.5731178557351202, "grad_norm": 0.6011226773262024, "learning_rate": 0.0005, "loss": 1.8963345289230347, "step": 7270 }, { "epoch": 0.5731966890027592, "grad_norm": 0.8688202500343323, "learning_rate": 0.0005, "loss": 2.0554113388061523, "step": 7271 }, { "epoch": 0.5732755222703981, "grad_norm": 0.7597020864486694, "learning_rate": 0.0005, "loss": 2.04121732711792, "step": 7272 }, { "epoch": 0.573354355538037, "grad_norm": 1.0633925199508667, "learning_rate": 0.0005, "loss": 1.9170892238616943, "step": 7273 }, { "epoch": 0.573433188805676, "grad_norm": 0.8912485241889954, "learning_rate": 0.0005, "loss": 2.0612404346466064, "step": 7274 }, { "epoch": 0.573512022073315, "grad_norm": 0.7389342188835144, "learning_rate": 0.0005, "loss": 1.7538261413574219, "step": 7275 }, { "epoch": 0.5735908553409539, "grad_norm": 0.7901861071586609, "learning_rate": 0.0005, "loss": 1.8693712949752808, "step": 7276 }, { "epoch": 0.5736696886085928, "grad_norm": 0.7992045283317566, "learning_rate": 0.0005, "loss": 2.0970799922943115, "step": 7277 }, { "epoch": 0.5737485218762318, "grad_norm": 0.7841617465019226, "learning_rate": 0.0005, "loss": 2.159942626953125, "step": 7278 }, { "epoch": 0.5738273551438707, "grad_norm": 0.9141185879707336, "learning_rate": 0.0005, "loss": 2.3338522911071777, "step": 7279 }, { "epoch": 0.5739061884115096, "grad_norm": 0.7008412480354309, "learning_rate": 0.0005, "loss": 1.839477300643921, "step": 7280 }, { "epoch": 0.5739850216791486, "grad_norm": 0.7690760493278503, "learning_rate": 0.0005, "loss": 1.938830018043518, "step": 7281 }, { "epoch": 0.5740638549467876, "grad_norm": 0.8390436172485352, "learning_rate": 0.0005, "loss": 2.0586347579956055, "step": 7282 }, { "epoch": 0.5741426882144265, "grad_norm": 0.7262389659881592, "learning_rate": 0.0005, "loss": 2.015714645385742, "step": 7283 }, { "epoch": 0.5742215214820654, "grad_norm": 0.7617836594581604, "learning_rate": 0.0005, "loss": 1.8754388093948364, "step": 7284 }, { "epoch": 0.5743003547497044, "grad_norm": 0.6467767357826233, "learning_rate": 0.0005, "loss": 2.0574302673339844, "step": 7285 }, { "epoch": 0.5743791880173433, "grad_norm": 0.6632921099662781, "learning_rate": 0.0005, "loss": 1.7202138900756836, "step": 7286 }, { "epoch": 0.5744580212849822, "grad_norm": 0.6380833387374878, "learning_rate": 0.0005, "loss": 2.120180606842041, "step": 7287 }, { "epoch": 0.5745368545526212, "grad_norm": 0.5198236703872681, "learning_rate": 0.0005, "loss": 1.6926648616790771, "step": 7288 }, { "epoch": 0.5746156878202602, "grad_norm": 0.7402539253234863, "learning_rate": 0.0005, "loss": 2.0644993782043457, "step": 7289 }, { "epoch": 0.5746945210878991, "grad_norm": 0.8742323517799377, "learning_rate": 0.0005, "loss": 2.0275402069091797, "step": 7290 }, { "epoch": 0.574773354355538, "grad_norm": 0.8330410718917847, "learning_rate": 0.0005, "loss": 1.980168104171753, "step": 7291 }, { "epoch": 0.574852187623177, "grad_norm": 0.9868296980857849, "learning_rate": 0.0005, "loss": 1.7748298645019531, "step": 7292 }, { "epoch": 0.5749310208908159, "grad_norm": 0.6584125757217407, "learning_rate": 0.0005, "loss": 1.816413402557373, "step": 7293 }, { "epoch": 0.5750098541584548, "grad_norm": 1.1727513074874878, "learning_rate": 0.0005, "loss": 1.911195993423462, "step": 7294 }, { "epoch": 0.5750886874260938, "grad_norm": 0.8353362679481506, "learning_rate": 0.0005, "loss": 2.0467758178710938, "step": 7295 }, { "epoch": 0.5751675206937328, "grad_norm": 0.6569873094558716, "learning_rate": 0.0005, "loss": 2.0832316875457764, "step": 7296 }, { "epoch": 0.5752463539613717, "grad_norm": 0.9094858169555664, "learning_rate": 0.0005, "loss": 2.0643224716186523, "step": 7297 }, { "epoch": 0.5753251872290106, "grad_norm": 0.7043681144714355, "learning_rate": 0.0005, "loss": 2.565600872039795, "step": 7298 }, { "epoch": 0.5754040204966496, "grad_norm": 0.7397662401199341, "learning_rate": 0.0005, "loss": 1.932183027267456, "step": 7299 }, { "epoch": 0.5754828537642885, "grad_norm": 0.7719001770019531, "learning_rate": 0.0005, "loss": 1.8000599145889282, "step": 7300 }, { "epoch": 0.5755616870319274, "grad_norm": 0.6271357536315918, "learning_rate": 0.0005, "loss": 1.899575114250183, "step": 7301 }, { "epoch": 0.5756405202995665, "grad_norm": 0.8207060694694519, "learning_rate": 0.0005, "loss": 2.125443458557129, "step": 7302 }, { "epoch": 0.5757193535672054, "grad_norm": 0.7140509486198425, "learning_rate": 0.0005, "loss": 1.9435306787490845, "step": 7303 }, { "epoch": 0.5757981868348443, "grad_norm": 0.8414044380187988, "learning_rate": 0.0005, "loss": 2.0895814895629883, "step": 7304 }, { "epoch": 0.5758770201024832, "grad_norm": 0.6863530874252319, "learning_rate": 0.0005, "loss": 1.9650118350982666, "step": 7305 }, { "epoch": 0.5759558533701222, "grad_norm": 0.6778491139411926, "learning_rate": 0.0005, "loss": 1.9188427925109863, "step": 7306 }, { "epoch": 0.5760346866377611, "grad_norm": 0.6519877910614014, "learning_rate": 0.0005, "loss": 2.1466383934020996, "step": 7307 }, { "epoch": 0.5761135199054, "grad_norm": 1.222556710243225, "learning_rate": 0.0005, "loss": 2.1172261238098145, "step": 7308 }, { "epoch": 0.5761923531730391, "grad_norm": 0.9285964965820312, "learning_rate": 0.0005, "loss": 1.9930341243743896, "step": 7309 }, { "epoch": 0.576271186440678, "grad_norm": 0.7243059277534485, "learning_rate": 0.0005, "loss": 1.8944706916809082, "step": 7310 }, { "epoch": 0.5763500197083169, "grad_norm": 0.7169192433357239, "learning_rate": 0.0005, "loss": 2.180042266845703, "step": 7311 }, { "epoch": 0.5764288529759558, "grad_norm": 0.8511126041412354, "learning_rate": 0.0005, "loss": 2.270982265472412, "step": 7312 }, { "epoch": 0.5765076862435948, "grad_norm": 0.6738325953483582, "learning_rate": 0.0005, "loss": 2.122150421142578, "step": 7313 }, { "epoch": 0.5765865195112337, "grad_norm": 3.408372640609741, "learning_rate": 0.0005, "loss": 1.9263551235198975, "step": 7314 }, { "epoch": 0.5766653527788727, "grad_norm": 0.6457016468048096, "learning_rate": 0.0005, "loss": 2.199328660964966, "step": 7315 }, { "epoch": 0.5767441860465117, "grad_norm": 0.62241131067276, "learning_rate": 0.0005, "loss": 1.9109784364700317, "step": 7316 }, { "epoch": 0.5768230193141506, "grad_norm": 0.6648703217506409, "learning_rate": 0.0005, "loss": 2.1341333389282227, "step": 7317 }, { "epoch": 0.5769018525817895, "grad_norm": 0.6348555684089661, "learning_rate": 0.0005, "loss": 1.9633920192718506, "step": 7318 }, { "epoch": 0.5769806858494284, "grad_norm": 0.8901347517967224, "learning_rate": 0.0005, "loss": 2.351797580718994, "step": 7319 }, { "epoch": 0.5770595191170674, "grad_norm": 1.03125, "learning_rate": 0.0005, "loss": 2.300942897796631, "step": 7320 }, { "epoch": 0.5771383523847063, "grad_norm": 0.8430801630020142, "learning_rate": 0.0005, "loss": 1.8864740133285522, "step": 7321 }, { "epoch": 0.5772171856523453, "grad_norm": 0.6788723468780518, "learning_rate": 0.0005, "loss": 1.968241810798645, "step": 7322 }, { "epoch": 0.5772960189199843, "grad_norm": 0.6621561646461487, "learning_rate": 0.0005, "loss": 1.864535927772522, "step": 7323 }, { "epoch": 0.5773748521876232, "grad_norm": 0.8037974834442139, "learning_rate": 0.0005, "loss": 1.9248859882354736, "step": 7324 }, { "epoch": 0.5774536854552621, "grad_norm": 0.7670780420303345, "learning_rate": 0.0005, "loss": 1.8921750783920288, "step": 7325 }, { "epoch": 0.577532518722901, "grad_norm": 0.603976845741272, "learning_rate": 0.0005, "loss": 1.8441872596740723, "step": 7326 }, { "epoch": 0.57761135199054, "grad_norm": 0.9216703176498413, "learning_rate": 0.0005, "loss": 2.2733421325683594, "step": 7327 }, { "epoch": 0.577690185258179, "grad_norm": 0.6820099949836731, "learning_rate": 0.0005, "loss": 2.2236621379852295, "step": 7328 }, { "epoch": 0.5777690185258179, "grad_norm": 0.5117199420928955, "learning_rate": 0.0005, "loss": 2.1875159740448, "step": 7329 }, { "epoch": 0.5778478517934569, "grad_norm": 0.7634373903274536, "learning_rate": 0.0005, "loss": 1.9432945251464844, "step": 7330 }, { "epoch": 0.5779266850610958, "grad_norm": 0.8121824264526367, "learning_rate": 0.0005, "loss": 2.0999763011932373, "step": 7331 }, { "epoch": 0.5780055183287347, "grad_norm": 0.6813707947731018, "learning_rate": 0.0005, "loss": 1.9979504346847534, "step": 7332 }, { "epoch": 0.5780843515963737, "grad_norm": 0.7807489037513733, "learning_rate": 0.0005, "loss": 1.9536900520324707, "step": 7333 }, { "epoch": 0.5781631848640126, "grad_norm": 0.6366038918495178, "learning_rate": 0.0005, "loss": 2.0274477005004883, "step": 7334 }, { "epoch": 0.5782420181316515, "grad_norm": 0.7532272934913635, "learning_rate": 0.0005, "loss": 1.7934964895248413, "step": 7335 }, { "epoch": 0.5783208513992905, "grad_norm": 0.6784632205963135, "learning_rate": 0.0005, "loss": 1.9727498292922974, "step": 7336 }, { "epoch": 0.5783996846669295, "grad_norm": 0.7190457582473755, "learning_rate": 0.0005, "loss": 1.7881481647491455, "step": 7337 }, { "epoch": 0.5784785179345684, "grad_norm": 0.7125281095504761, "learning_rate": 0.0005, "loss": 1.8492639064788818, "step": 7338 }, { "epoch": 0.5785573512022073, "grad_norm": 0.7366788983345032, "learning_rate": 0.0005, "loss": 1.9665652513504028, "step": 7339 }, { "epoch": 0.5786361844698463, "grad_norm": 0.688294529914856, "learning_rate": 0.0005, "loss": 2.169100522994995, "step": 7340 }, { "epoch": 0.5787150177374852, "grad_norm": 0.8185492753982544, "learning_rate": 0.0005, "loss": 1.9090229272842407, "step": 7341 }, { "epoch": 0.5787938510051241, "grad_norm": 0.6873346567153931, "learning_rate": 0.0005, "loss": 2.000683546066284, "step": 7342 }, { "epoch": 0.5788726842727631, "grad_norm": 0.7215948700904846, "learning_rate": 0.0005, "loss": 2.1156229972839355, "step": 7343 }, { "epoch": 0.5789515175404021, "grad_norm": 1.0560990571975708, "learning_rate": 0.0005, "loss": 2.1469950675964355, "step": 7344 }, { "epoch": 0.579030350808041, "grad_norm": 0.7109767198562622, "learning_rate": 0.0005, "loss": 1.6052498817443848, "step": 7345 }, { "epoch": 0.5791091840756799, "grad_norm": 0.7174134254455566, "learning_rate": 0.0005, "loss": 1.9563133716583252, "step": 7346 }, { "epoch": 0.5791880173433189, "grad_norm": 0.7185845971107483, "learning_rate": 0.0005, "loss": 1.9364025592803955, "step": 7347 }, { "epoch": 0.5792668506109578, "grad_norm": 0.7993655204772949, "learning_rate": 0.0005, "loss": 1.971056342124939, "step": 7348 }, { "epoch": 0.5793456838785968, "grad_norm": 0.6212140321731567, "learning_rate": 0.0005, "loss": 1.9491446018218994, "step": 7349 }, { "epoch": 0.5794245171462357, "grad_norm": 0.95302414894104, "learning_rate": 0.0005, "loss": 1.9708701372146606, "step": 7350 }, { "epoch": 0.5795033504138747, "grad_norm": 0.8333086371421814, "learning_rate": 0.0005, "loss": 1.9891307353973389, "step": 7351 }, { "epoch": 0.5795821836815136, "grad_norm": 0.8936498761177063, "learning_rate": 0.0005, "loss": 2.13364577293396, "step": 7352 }, { "epoch": 0.5796610169491525, "grad_norm": 0.6639046669006348, "learning_rate": 0.0005, "loss": 1.7495726346969604, "step": 7353 }, { "epoch": 0.5797398502167915, "grad_norm": 1.0112013816833496, "learning_rate": 0.0005, "loss": 2.165726661682129, "step": 7354 }, { "epoch": 0.5798186834844304, "grad_norm": 3.580893039703369, "learning_rate": 0.0005, "loss": 1.7899079322814941, "step": 7355 }, { "epoch": 0.5798975167520694, "grad_norm": 0.7326050996780396, "learning_rate": 0.0005, "loss": 2.039559841156006, "step": 7356 }, { "epoch": 0.5799763500197083, "grad_norm": 0.8122267723083496, "learning_rate": 0.0005, "loss": 2.0339202880859375, "step": 7357 }, { "epoch": 0.5800551832873473, "grad_norm": 0.7102697491645813, "learning_rate": 0.0005, "loss": 1.9816588163375854, "step": 7358 }, { "epoch": 0.5801340165549862, "grad_norm": 0.6739342212677002, "learning_rate": 0.0005, "loss": 1.8879238367080688, "step": 7359 }, { "epoch": 0.5802128498226251, "grad_norm": 0.648720383644104, "learning_rate": 0.0005, "loss": 2.213945150375366, "step": 7360 }, { "epoch": 0.5802916830902641, "grad_norm": 0.7210912108421326, "learning_rate": 0.0005, "loss": 2.031172275543213, "step": 7361 }, { "epoch": 0.580370516357903, "grad_norm": 0.6136062145233154, "learning_rate": 0.0005, "loss": 2.4814682006835938, "step": 7362 }, { "epoch": 0.580449349625542, "grad_norm": 0.5723323822021484, "learning_rate": 0.0005, "loss": 1.750929355621338, "step": 7363 }, { "epoch": 0.580528182893181, "grad_norm": 0.825310468673706, "learning_rate": 0.0005, "loss": 2.07121205329895, "step": 7364 }, { "epoch": 0.5806070161608199, "grad_norm": 0.8248792290687561, "learning_rate": 0.0005, "loss": 1.9701998233795166, "step": 7365 }, { "epoch": 0.5806858494284588, "grad_norm": 1.1218416690826416, "learning_rate": 0.0005, "loss": 2.594111204147339, "step": 7366 }, { "epoch": 0.5807646826960977, "grad_norm": 0.7508931159973145, "learning_rate": 0.0005, "loss": 2.0345237255096436, "step": 7367 }, { "epoch": 0.5808435159637367, "grad_norm": 0.8332595825195312, "learning_rate": 0.0005, "loss": 2.0463805198669434, "step": 7368 }, { "epoch": 0.5809223492313756, "grad_norm": 0.7477124333381653, "learning_rate": 0.0005, "loss": 1.9708280563354492, "step": 7369 }, { "epoch": 0.5810011824990146, "grad_norm": 0.6039433479309082, "learning_rate": 0.0005, "loss": 1.7575643062591553, "step": 7370 }, { "epoch": 0.5810800157666536, "grad_norm": 0.5821584463119507, "learning_rate": 0.0005, "loss": 1.7620489597320557, "step": 7371 }, { "epoch": 0.5811588490342925, "grad_norm": 0.7351267337799072, "learning_rate": 0.0005, "loss": 1.849616527557373, "step": 7372 }, { "epoch": 0.5812376823019314, "grad_norm": 0.7537945508956909, "learning_rate": 0.0005, "loss": 1.7710952758789062, "step": 7373 }, { "epoch": 0.5813165155695703, "grad_norm": 0.7319613099098206, "learning_rate": 0.0005, "loss": 1.8116369247436523, "step": 7374 }, { "epoch": 0.5813953488372093, "grad_norm": 0.7544410824775696, "learning_rate": 0.0005, "loss": 1.9519968032836914, "step": 7375 }, { "epoch": 0.5814741821048482, "grad_norm": 0.5440706610679626, "learning_rate": 0.0005, "loss": 1.8295974731445312, "step": 7376 }, { "epoch": 0.5815530153724872, "grad_norm": 0.8451952934265137, "learning_rate": 0.0005, "loss": 2.1955318450927734, "step": 7377 }, { "epoch": 0.5816318486401262, "grad_norm": 0.8793179392814636, "learning_rate": 0.0005, "loss": 2.0136754512786865, "step": 7378 }, { "epoch": 0.5817106819077651, "grad_norm": 0.72365802526474, "learning_rate": 0.0005, "loss": 1.9838563203811646, "step": 7379 }, { "epoch": 0.581789515175404, "grad_norm": 0.698235273361206, "learning_rate": 0.0005, "loss": 2.139981746673584, "step": 7380 }, { "epoch": 0.5818683484430429, "grad_norm": 0.7697486877441406, "learning_rate": 0.0005, "loss": 2.0290915966033936, "step": 7381 }, { "epoch": 0.5819471817106819, "grad_norm": 0.628812313079834, "learning_rate": 0.0005, "loss": 1.9435904026031494, "step": 7382 }, { "epoch": 0.5820260149783208, "grad_norm": 0.8249680995941162, "learning_rate": 0.0005, "loss": 2.055746078491211, "step": 7383 }, { "epoch": 0.5821048482459598, "grad_norm": 0.822452962398529, "learning_rate": 0.0005, "loss": 1.9751849174499512, "step": 7384 }, { "epoch": 0.5821836815135988, "grad_norm": 0.6028671264648438, "learning_rate": 0.0005, "loss": 2.0295164585113525, "step": 7385 }, { "epoch": 0.5822625147812377, "grad_norm": 1.0411219596862793, "learning_rate": 0.0005, "loss": 2.1861042976379395, "step": 7386 }, { "epoch": 0.5823413480488766, "grad_norm": 0.7263089418411255, "learning_rate": 0.0005, "loss": 1.9128308296203613, "step": 7387 }, { "epoch": 0.5824201813165155, "grad_norm": 0.7098044157028198, "learning_rate": 0.0005, "loss": 1.9458365440368652, "step": 7388 }, { "epoch": 0.5824990145841545, "grad_norm": 1.9023916721343994, "learning_rate": 0.0005, "loss": 2.1691713333129883, "step": 7389 }, { "epoch": 0.5825778478517935, "grad_norm": 0.5830426812171936, "learning_rate": 0.0005, "loss": 1.9881346225738525, "step": 7390 }, { "epoch": 0.5826566811194324, "grad_norm": 0.7134573459625244, "learning_rate": 0.0005, "loss": 1.8306026458740234, "step": 7391 }, { "epoch": 0.5827355143870714, "grad_norm": 0.6890772581100464, "learning_rate": 0.0005, "loss": 2.52313232421875, "step": 7392 }, { "epoch": 0.5828143476547103, "grad_norm": 0.6649017930030823, "learning_rate": 0.0005, "loss": 1.8498557806015015, "step": 7393 }, { "epoch": 0.5828931809223492, "grad_norm": 0.6299012899398804, "learning_rate": 0.0005, "loss": 1.7730035781860352, "step": 7394 }, { "epoch": 0.5829720141899882, "grad_norm": 1.174783706665039, "learning_rate": 0.0005, "loss": 2.00797700881958, "step": 7395 }, { "epoch": 0.5830508474576271, "grad_norm": 0.7483850121498108, "learning_rate": 0.0005, "loss": 2.2014527320861816, "step": 7396 }, { "epoch": 0.583129680725266, "grad_norm": 0.9448046088218689, "learning_rate": 0.0005, "loss": 2.3418428897857666, "step": 7397 }, { "epoch": 0.583208513992905, "grad_norm": 0.7191468477249146, "learning_rate": 0.0005, "loss": 1.9484238624572754, "step": 7398 }, { "epoch": 0.583287347260544, "grad_norm": 0.6472828388214111, "learning_rate": 0.0005, "loss": 2.220625877380371, "step": 7399 }, { "epoch": 0.5833661805281829, "grad_norm": 0.6071100234985352, "learning_rate": 0.0005, "loss": 1.9230225086212158, "step": 7400 }, { "epoch": 0.5834450137958218, "grad_norm": 0.7205000519752502, "learning_rate": 0.0005, "loss": 1.9495923519134521, "step": 7401 }, { "epoch": 0.5835238470634608, "grad_norm": 0.7530670762062073, "learning_rate": 0.0005, "loss": 1.9286264181137085, "step": 7402 }, { "epoch": 0.5836026803310997, "grad_norm": 0.5657660961151123, "learning_rate": 0.0005, "loss": 1.8562171459197998, "step": 7403 }, { "epoch": 0.5836815135987387, "grad_norm": 0.5848467946052551, "learning_rate": 0.0005, "loss": 1.9330705404281616, "step": 7404 }, { "epoch": 0.5837603468663776, "grad_norm": 0.6633527278900146, "learning_rate": 0.0005, "loss": 2.0116591453552246, "step": 7405 }, { "epoch": 0.5838391801340166, "grad_norm": 0.5795087218284607, "learning_rate": 0.0005, "loss": 2.122354030609131, "step": 7406 }, { "epoch": 0.5839180134016555, "grad_norm": 1.0438140630722046, "learning_rate": 0.0005, "loss": 1.9807000160217285, "step": 7407 }, { "epoch": 0.5839968466692944, "grad_norm": 0.7867108583450317, "learning_rate": 0.0005, "loss": 1.9532470703125, "step": 7408 }, { "epoch": 0.5840756799369334, "grad_norm": 0.6935163140296936, "learning_rate": 0.0005, "loss": 1.8119478225708008, "step": 7409 }, { "epoch": 0.5841545132045723, "grad_norm": 0.8286797404289246, "learning_rate": 0.0005, "loss": 1.9212876558303833, "step": 7410 }, { "epoch": 0.5842333464722113, "grad_norm": 0.6812551617622375, "learning_rate": 0.0005, "loss": 1.8334110975265503, "step": 7411 }, { "epoch": 0.5843121797398502, "grad_norm": 0.7066707015037537, "learning_rate": 0.0005, "loss": 2.2957301139831543, "step": 7412 }, { "epoch": 0.5843910130074892, "grad_norm": 0.6825031638145447, "learning_rate": 0.0005, "loss": 1.915177583694458, "step": 7413 }, { "epoch": 0.5844698462751281, "grad_norm": 1.122648000717163, "learning_rate": 0.0005, "loss": 1.828707218170166, "step": 7414 }, { "epoch": 0.584548679542767, "grad_norm": 0.7184036374092102, "learning_rate": 0.0005, "loss": 1.968482494354248, "step": 7415 }, { "epoch": 0.584627512810406, "grad_norm": 1.0607037544250488, "learning_rate": 0.0005, "loss": 2.0546393394470215, "step": 7416 }, { "epoch": 0.584706346078045, "grad_norm": 0.809438169002533, "learning_rate": 0.0005, "loss": 1.8023262023925781, "step": 7417 }, { "epoch": 0.5847851793456839, "grad_norm": 0.7092893719673157, "learning_rate": 0.0005, "loss": 1.849149227142334, "step": 7418 }, { "epoch": 0.5848640126133228, "grad_norm": 0.8628074526786804, "learning_rate": 0.0005, "loss": 2.1344943046569824, "step": 7419 }, { "epoch": 0.5849428458809618, "grad_norm": 0.7161156535148621, "learning_rate": 0.0005, "loss": 1.8043246269226074, "step": 7420 }, { "epoch": 0.5850216791486007, "grad_norm": 0.6500622034072876, "learning_rate": 0.0005, "loss": 1.7900807857513428, "step": 7421 }, { "epoch": 0.5851005124162396, "grad_norm": 0.6058142781257629, "learning_rate": 0.0005, "loss": 1.8501386642456055, "step": 7422 }, { "epoch": 0.5851793456838786, "grad_norm": 0.8679923415184021, "learning_rate": 0.0005, "loss": 2.0558507442474365, "step": 7423 }, { "epoch": 0.5852581789515175, "grad_norm": 0.6302821636199951, "learning_rate": 0.0005, "loss": 2.010913848876953, "step": 7424 }, { "epoch": 0.5853370122191565, "grad_norm": 0.7937870621681213, "learning_rate": 0.0005, "loss": 2.078003406524658, "step": 7425 }, { "epoch": 0.5854158454867955, "grad_norm": 0.6289886832237244, "learning_rate": 0.0005, "loss": 1.9435054063796997, "step": 7426 }, { "epoch": 0.5854946787544344, "grad_norm": 0.6649936437606812, "learning_rate": 0.0005, "loss": 1.8512067794799805, "step": 7427 }, { "epoch": 0.5855735120220733, "grad_norm": 0.8005658388137817, "learning_rate": 0.0005, "loss": 1.8653556108474731, "step": 7428 }, { "epoch": 0.5856523452897122, "grad_norm": 0.6994357705116272, "learning_rate": 0.0005, "loss": 2.1743078231811523, "step": 7429 }, { "epoch": 0.5857311785573512, "grad_norm": 0.6511334776878357, "learning_rate": 0.0005, "loss": 2.134678840637207, "step": 7430 }, { "epoch": 0.5858100118249902, "grad_norm": 0.7922676801681519, "learning_rate": 0.0005, "loss": 1.9459218978881836, "step": 7431 }, { "epoch": 0.5858888450926291, "grad_norm": 0.7891363501548767, "learning_rate": 0.0005, "loss": 1.9710066318511963, "step": 7432 }, { "epoch": 0.5859676783602681, "grad_norm": 0.9748491644859314, "learning_rate": 0.0005, "loss": 1.7081897258758545, "step": 7433 }, { "epoch": 0.586046511627907, "grad_norm": 0.6385135650634766, "learning_rate": 0.0005, "loss": 1.944148302078247, "step": 7434 }, { "epoch": 0.5861253448955459, "grad_norm": 0.6465703248977661, "learning_rate": 0.0005, "loss": 2.1449360847473145, "step": 7435 }, { "epoch": 0.5862041781631848, "grad_norm": 0.5847073197364807, "learning_rate": 0.0005, "loss": 1.969970941543579, "step": 7436 }, { "epoch": 0.5862830114308238, "grad_norm": 1.0638158321380615, "learning_rate": 0.0005, "loss": 1.8186789751052856, "step": 7437 }, { "epoch": 0.5863618446984628, "grad_norm": 0.9299233555793762, "learning_rate": 0.0005, "loss": 2.082296848297119, "step": 7438 }, { "epoch": 0.5864406779661017, "grad_norm": 0.8539411425590515, "learning_rate": 0.0005, "loss": 1.996080756187439, "step": 7439 }, { "epoch": 0.5865195112337407, "grad_norm": 0.7641857266426086, "learning_rate": 0.0005, "loss": 2.1070897579193115, "step": 7440 }, { "epoch": 0.5865983445013796, "grad_norm": 0.6711196899414062, "learning_rate": 0.0005, "loss": 1.7062654495239258, "step": 7441 }, { "epoch": 0.5866771777690185, "grad_norm": 1.1991710662841797, "learning_rate": 0.0005, "loss": 1.8868706226348877, "step": 7442 }, { "epoch": 0.5867560110366574, "grad_norm": 0.6817166209220886, "learning_rate": 0.0005, "loss": 2.2182679176330566, "step": 7443 }, { "epoch": 0.5868348443042964, "grad_norm": 0.6579222679138184, "learning_rate": 0.0005, "loss": 2.118974208831787, "step": 7444 }, { "epoch": 0.5869136775719354, "grad_norm": 0.7068800330162048, "learning_rate": 0.0005, "loss": 2.2521557807922363, "step": 7445 }, { "epoch": 0.5869925108395743, "grad_norm": 0.7217605113983154, "learning_rate": 0.0005, "loss": 2.0543556213378906, "step": 7446 }, { "epoch": 0.5870713441072133, "grad_norm": 0.826414167881012, "learning_rate": 0.0005, "loss": 2.1485118865966797, "step": 7447 }, { "epoch": 0.5871501773748522, "grad_norm": 0.8378903269767761, "learning_rate": 0.0005, "loss": 2.249263048171997, "step": 7448 }, { "epoch": 0.5872290106424911, "grad_norm": 0.723210871219635, "learning_rate": 0.0005, "loss": 1.8867312669754028, "step": 7449 }, { "epoch": 0.58730784391013, "grad_norm": 0.8344254493713379, "learning_rate": 0.0005, "loss": 1.988027811050415, "step": 7450 }, { "epoch": 0.587386677177769, "grad_norm": 0.8858373761177063, "learning_rate": 0.0005, "loss": 2.165009021759033, "step": 7451 }, { "epoch": 0.587465510445408, "grad_norm": 0.6584643721580505, "learning_rate": 0.0005, "loss": 1.8205593824386597, "step": 7452 }, { "epoch": 0.5875443437130469, "grad_norm": 0.5837174654006958, "learning_rate": 0.0005, "loss": 1.6056455373764038, "step": 7453 }, { "epoch": 0.5876231769806859, "grad_norm": 0.5731794238090515, "learning_rate": 0.0005, "loss": 1.9354872703552246, "step": 7454 }, { "epoch": 0.5877020102483248, "grad_norm": 0.9282258152961731, "learning_rate": 0.0005, "loss": 1.8676396608352661, "step": 7455 }, { "epoch": 0.5877808435159637, "grad_norm": 0.7764774560928345, "learning_rate": 0.0005, "loss": 2.2985451221466064, "step": 7456 }, { "epoch": 0.5878596767836026, "grad_norm": 0.7102818489074707, "learning_rate": 0.0005, "loss": 1.6489895582199097, "step": 7457 }, { "epoch": 0.5879385100512416, "grad_norm": 0.9124100208282471, "learning_rate": 0.0005, "loss": 2.393120050430298, "step": 7458 }, { "epoch": 0.5880173433188806, "grad_norm": 0.8651835322380066, "learning_rate": 0.0005, "loss": 2.089231252670288, "step": 7459 }, { "epoch": 0.5880961765865195, "grad_norm": 0.7837098836898804, "learning_rate": 0.0005, "loss": 1.9144374132156372, "step": 7460 }, { "epoch": 0.5881750098541585, "grad_norm": 0.6228205561637878, "learning_rate": 0.0005, "loss": 1.8167804479599, "step": 7461 }, { "epoch": 0.5882538431217974, "grad_norm": 0.732013463973999, "learning_rate": 0.0005, "loss": 2.1114583015441895, "step": 7462 }, { "epoch": 0.5883326763894363, "grad_norm": 0.805246889591217, "learning_rate": 0.0005, "loss": 2.054986000061035, "step": 7463 }, { "epoch": 0.5884115096570753, "grad_norm": 0.7351343035697937, "learning_rate": 0.0005, "loss": 2.0548758506774902, "step": 7464 }, { "epoch": 0.5884903429247142, "grad_norm": 0.671595573425293, "learning_rate": 0.0005, "loss": 1.884580135345459, "step": 7465 }, { "epoch": 0.5885691761923532, "grad_norm": 0.8125171661376953, "learning_rate": 0.0005, "loss": 2.0167250633239746, "step": 7466 }, { "epoch": 0.5886480094599921, "grad_norm": 1.1229581832885742, "learning_rate": 0.0005, "loss": 2.2213056087493896, "step": 7467 }, { "epoch": 0.5887268427276311, "grad_norm": 0.8056671023368835, "learning_rate": 0.0005, "loss": 2.0953328609466553, "step": 7468 }, { "epoch": 0.58880567599527, "grad_norm": 0.8363956212997437, "learning_rate": 0.0005, "loss": 1.99948251247406, "step": 7469 }, { "epoch": 0.5888845092629089, "grad_norm": 0.8177576661109924, "learning_rate": 0.0005, "loss": 1.8112828731536865, "step": 7470 }, { "epoch": 0.5889633425305479, "grad_norm": 1.0491015911102295, "learning_rate": 0.0005, "loss": 2.1061453819274902, "step": 7471 }, { "epoch": 0.5890421757981869, "grad_norm": 0.9522292017936707, "learning_rate": 0.0005, "loss": 2.1756350994110107, "step": 7472 }, { "epoch": 0.5891210090658258, "grad_norm": 0.8466187119483948, "learning_rate": 0.0005, "loss": 2.117800712585449, "step": 7473 }, { "epoch": 0.5891998423334647, "grad_norm": 0.46801233291625977, "learning_rate": 0.0005, "loss": 1.6990296840667725, "step": 7474 }, { "epoch": 0.5892786756011037, "grad_norm": 0.7751872539520264, "learning_rate": 0.0005, "loss": 1.8740887641906738, "step": 7475 }, { "epoch": 0.5893575088687426, "grad_norm": 0.6867125034332275, "learning_rate": 0.0005, "loss": 2.1373040676116943, "step": 7476 }, { "epoch": 0.5894363421363815, "grad_norm": 0.7240882515907288, "learning_rate": 0.0005, "loss": 1.8530036211013794, "step": 7477 }, { "epoch": 0.5895151754040205, "grad_norm": 0.7230568528175354, "learning_rate": 0.0005, "loss": 2.2821924686431885, "step": 7478 }, { "epoch": 0.5895940086716595, "grad_norm": 0.743337869644165, "learning_rate": 0.0005, "loss": 2.111985683441162, "step": 7479 }, { "epoch": 0.5896728419392984, "grad_norm": 0.5212400555610657, "learning_rate": 0.0005, "loss": 1.8204479217529297, "step": 7480 }, { "epoch": 0.5897516752069373, "grad_norm": 0.7355701327323914, "learning_rate": 0.0005, "loss": 1.8068487644195557, "step": 7481 }, { "epoch": 0.5898305084745763, "grad_norm": 0.6329627633094788, "learning_rate": 0.0005, "loss": 1.8994197845458984, "step": 7482 }, { "epoch": 0.5899093417422152, "grad_norm": 0.6948001384735107, "learning_rate": 0.0005, "loss": 2.027508020401001, "step": 7483 }, { "epoch": 0.5899881750098541, "grad_norm": 0.7010281682014465, "learning_rate": 0.0005, "loss": 2.0903003215789795, "step": 7484 }, { "epoch": 0.5900670082774931, "grad_norm": 0.7441275119781494, "learning_rate": 0.0005, "loss": 1.8486689329147339, "step": 7485 }, { "epoch": 0.590145841545132, "grad_norm": 0.7089042663574219, "learning_rate": 0.0005, "loss": 1.9737889766693115, "step": 7486 }, { "epoch": 0.590224674812771, "grad_norm": 0.6866204142570496, "learning_rate": 0.0005, "loss": 2.020559787750244, "step": 7487 }, { "epoch": 0.5903035080804099, "grad_norm": 0.6556402444839478, "learning_rate": 0.0005, "loss": 1.8766450881958008, "step": 7488 }, { "epoch": 0.5903823413480489, "grad_norm": 0.7484536170959473, "learning_rate": 0.0005, "loss": 2.095040798187256, "step": 7489 }, { "epoch": 0.5904611746156878, "grad_norm": 0.7958198189735413, "learning_rate": 0.0005, "loss": 2.0440337657928467, "step": 7490 }, { "epoch": 0.5905400078833267, "grad_norm": 2.1447649002075195, "learning_rate": 0.0005, "loss": 1.988274097442627, "step": 7491 }, { "epoch": 0.5906188411509657, "grad_norm": 0.7723760604858398, "learning_rate": 0.0005, "loss": 2.130016326904297, "step": 7492 }, { "epoch": 0.5906976744186047, "grad_norm": 0.6949961185455322, "learning_rate": 0.0005, "loss": 1.9242730140686035, "step": 7493 }, { "epoch": 0.5907765076862436, "grad_norm": 0.7108324766159058, "learning_rate": 0.0005, "loss": 1.9855080842971802, "step": 7494 }, { "epoch": 0.5908553409538826, "grad_norm": 0.8120949268341064, "learning_rate": 0.0005, "loss": 2.1049115657806396, "step": 7495 }, { "epoch": 0.5909341742215215, "grad_norm": 0.7033634781837463, "learning_rate": 0.0005, "loss": 2.1042747497558594, "step": 7496 }, { "epoch": 0.5910130074891604, "grad_norm": 0.7144129276275635, "learning_rate": 0.0005, "loss": 1.5884987115859985, "step": 7497 }, { "epoch": 0.5910918407567993, "grad_norm": 0.829643964767456, "learning_rate": 0.0005, "loss": 2.074788808822632, "step": 7498 }, { "epoch": 0.5911706740244383, "grad_norm": 0.5098941326141357, "learning_rate": 0.0005, "loss": 1.6840558052062988, "step": 7499 }, { "epoch": 0.5912495072920773, "grad_norm": 0.6142054200172424, "learning_rate": 0.0005, "loss": 2.092237949371338, "step": 7500 }, { "epoch": 0.5913283405597162, "grad_norm": 0.6424511075019836, "learning_rate": 0.0005, "loss": 2.126990556716919, "step": 7501 }, { "epoch": 0.5914071738273552, "grad_norm": 0.778072714805603, "learning_rate": 0.0005, "loss": 2.261807918548584, "step": 7502 }, { "epoch": 0.5914860070949941, "grad_norm": 0.6573628783226013, "learning_rate": 0.0005, "loss": 2.128356695175171, "step": 7503 }, { "epoch": 0.591564840362633, "grad_norm": 0.8250551223754883, "learning_rate": 0.0005, "loss": 1.7139761447906494, "step": 7504 }, { "epoch": 0.5916436736302719, "grad_norm": 5.7981858253479, "learning_rate": 0.0005, "loss": 2.3888678550720215, "step": 7505 }, { "epoch": 0.591722506897911, "grad_norm": 0.7156029343605042, "learning_rate": 0.0005, "loss": 2.0209732055664062, "step": 7506 }, { "epoch": 0.5918013401655499, "grad_norm": 0.6903124451637268, "learning_rate": 0.0005, "loss": 2.0488202571868896, "step": 7507 }, { "epoch": 0.5918801734331888, "grad_norm": 0.8029569387435913, "learning_rate": 0.0005, "loss": 2.0708932876586914, "step": 7508 }, { "epoch": 0.5919590067008278, "grad_norm": 0.7853533029556274, "learning_rate": 0.0005, "loss": 2.1342272758483887, "step": 7509 }, { "epoch": 0.5920378399684667, "grad_norm": 0.7441530823707581, "learning_rate": 0.0005, "loss": 2.0378408432006836, "step": 7510 }, { "epoch": 0.5921166732361056, "grad_norm": 0.7568268179893494, "learning_rate": 0.0005, "loss": 2.1786975860595703, "step": 7511 }, { "epoch": 0.5921955065037445, "grad_norm": 0.7436584234237671, "learning_rate": 0.0005, "loss": 2.274439811706543, "step": 7512 }, { "epoch": 0.5922743397713836, "grad_norm": 0.6721987724304199, "learning_rate": 0.0005, "loss": 2.2347195148468018, "step": 7513 }, { "epoch": 0.5923531730390225, "grad_norm": 0.7419882416725159, "learning_rate": 0.0005, "loss": 2.5782485008239746, "step": 7514 }, { "epoch": 0.5924320063066614, "grad_norm": 0.7243111729621887, "learning_rate": 0.0005, "loss": 1.8470842838287354, "step": 7515 }, { "epoch": 0.5925108395743004, "grad_norm": 0.6630018949508667, "learning_rate": 0.0005, "loss": 2.1795992851257324, "step": 7516 }, { "epoch": 0.5925896728419393, "grad_norm": 0.7276042699813843, "learning_rate": 0.0005, "loss": 2.154811143875122, "step": 7517 }, { "epoch": 0.5926685061095782, "grad_norm": 0.6482138633728027, "learning_rate": 0.0005, "loss": 1.9533950090408325, "step": 7518 }, { "epoch": 0.5927473393772171, "grad_norm": 0.8276026248931885, "learning_rate": 0.0005, "loss": 1.855907678604126, "step": 7519 }, { "epoch": 0.5928261726448562, "grad_norm": 0.7574939131736755, "learning_rate": 0.0005, "loss": 1.7215365171432495, "step": 7520 }, { "epoch": 0.5929050059124951, "grad_norm": 0.6896957159042358, "learning_rate": 0.0005, "loss": 1.9425631761550903, "step": 7521 }, { "epoch": 0.592983839180134, "grad_norm": 0.8069627285003662, "learning_rate": 0.0005, "loss": 1.7831627130508423, "step": 7522 }, { "epoch": 0.593062672447773, "grad_norm": 0.69040846824646, "learning_rate": 0.0005, "loss": 2.0048675537109375, "step": 7523 }, { "epoch": 0.5931415057154119, "grad_norm": 0.6526136994361877, "learning_rate": 0.0005, "loss": 2.131688117980957, "step": 7524 }, { "epoch": 0.5932203389830508, "grad_norm": 0.6811690926551819, "learning_rate": 0.0005, "loss": 2.2464499473571777, "step": 7525 }, { "epoch": 0.5932991722506898, "grad_norm": 0.8078383207321167, "learning_rate": 0.0005, "loss": 1.9129879474639893, "step": 7526 }, { "epoch": 0.5933780055183288, "grad_norm": 0.6991943717002869, "learning_rate": 0.0005, "loss": 2.184854507446289, "step": 7527 }, { "epoch": 0.5934568387859677, "grad_norm": 0.8084227442741394, "learning_rate": 0.0005, "loss": 2.141481399536133, "step": 7528 }, { "epoch": 0.5935356720536066, "grad_norm": 0.6932190656661987, "learning_rate": 0.0005, "loss": 2.018934488296509, "step": 7529 }, { "epoch": 0.5936145053212456, "grad_norm": 0.5925294160842896, "learning_rate": 0.0005, "loss": 1.8100863695144653, "step": 7530 }, { "epoch": 0.5936933385888845, "grad_norm": 0.8188559412956238, "learning_rate": 0.0005, "loss": 2.063966751098633, "step": 7531 }, { "epoch": 0.5937721718565234, "grad_norm": 0.9179404973983765, "learning_rate": 0.0005, "loss": 1.840120792388916, "step": 7532 }, { "epoch": 0.5938510051241624, "grad_norm": 0.8492512702941895, "learning_rate": 0.0005, "loss": 1.9942963123321533, "step": 7533 }, { "epoch": 0.5939298383918014, "grad_norm": 0.790327787399292, "learning_rate": 0.0005, "loss": 2.2464818954467773, "step": 7534 }, { "epoch": 0.5940086716594403, "grad_norm": 0.96760493516922, "learning_rate": 0.0005, "loss": 2.1565117835998535, "step": 7535 }, { "epoch": 0.5940875049270792, "grad_norm": 0.7871758937835693, "learning_rate": 0.0005, "loss": 2.1930928230285645, "step": 7536 }, { "epoch": 0.5941663381947182, "grad_norm": 0.7822115421295166, "learning_rate": 0.0005, "loss": 2.2951340675354004, "step": 7537 }, { "epoch": 0.5942451714623571, "grad_norm": 0.7269631624221802, "learning_rate": 0.0005, "loss": 2.2234737873077393, "step": 7538 }, { "epoch": 0.594324004729996, "grad_norm": 0.800719141960144, "learning_rate": 0.0005, "loss": 2.349001169204712, "step": 7539 }, { "epoch": 0.594402837997635, "grad_norm": 0.7387000322341919, "learning_rate": 0.0005, "loss": 1.816707730293274, "step": 7540 }, { "epoch": 0.594481671265274, "grad_norm": 0.6040659546852112, "learning_rate": 0.0005, "loss": 1.8483006954193115, "step": 7541 }, { "epoch": 0.5945605045329129, "grad_norm": 0.6679539680480957, "learning_rate": 0.0005, "loss": 1.7000004053115845, "step": 7542 }, { "epoch": 0.5946393378005518, "grad_norm": 1.4304144382476807, "learning_rate": 0.0005, "loss": 1.852620244026184, "step": 7543 }, { "epoch": 0.5947181710681908, "grad_norm": 1.0371894836425781, "learning_rate": 0.0005, "loss": 2.0533463954925537, "step": 7544 }, { "epoch": 0.5947970043358297, "grad_norm": 0.7536638379096985, "learning_rate": 0.0005, "loss": 1.8322720527648926, "step": 7545 }, { "epoch": 0.5948758376034686, "grad_norm": 0.6572755575180054, "learning_rate": 0.0005, "loss": 1.9072952270507812, "step": 7546 }, { "epoch": 0.5949546708711076, "grad_norm": 0.7183701992034912, "learning_rate": 0.0005, "loss": 1.8157830238342285, "step": 7547 }, { "epoch": 0.5950335041387466, "grad_norm": 0.6178100109100342, "learning_rate": 0.0005, "loss": 2.0773468017578125, "step": 7548 }, { "epoch": 0.5951123374063855, "grad_norm": 0.7619871497154236, "learning_rate": 0.0005, "loss": 2.254707098007202, "step": 7549 }, { "epoch": 0.5951911706740244, "grad_norm": 0.8179551362991333, "learning_rate": 0.0005, "loss": 1.849760890007019, "step": 7550 }, { "epoch": 0.5952700039416634, "grad_norm": 0.7083888053894043, "learning_rate": 0.0005, "loss": 1.8785367012023926, "step": 7551 }, { "epoch": 0.5953488372093023, "grad_norm": 0.8435075283050537, "learning_rate": 0.0005, "loss": 1.9661688804626465, "step": 7552 }, { "epoch": 0.5954276704769412, "grad_norm": 0.8491793274879456, "learning_rate": 0.0005, "loss": 2.1974453926086426, "step": 7553 }, { "epoch": 0.5955065037445803, "grad_norm": 0.8756760954856873, "learning_rate": 0.0005, "loss": 2.21976375579834, "step": 7554 }, { "epoch": 0.5955853370122192, "grad_norm": 0.5850735902786255, "learning_rate": 0.0005, "loss": 1.6178960800170898, "step": 7555 }, { "epoch": 0.5956641702798581, "grad_norm": 0.7232041954994202, "learning_rate": 0.0005, "loss": 2.0425822734832764, "step": 7556 }, { "epoch": 0.5957430035474971, "grad_norm": 0.6446658968925476, "learning_rate": 0.0005, "loss": 1.9563863277435303, "step": 7557 }, { "epoch": 0.595821836815136, "grad_norm": 0.6757242679595947, "learning_rate": 0.0005, "loss": 2.354896306991577, "step": 7558 }, { "epoch": 0.5959006700827749, "grad_norm": 0.7020974159240723, "learning_rate": 0.0005, "loss": 2.383481025695801, "step": 7559 }, { "epoch": 0.5959795033504138, "grad_norm": 1.0238453149795532, "learning_rate": 0.0005, "loss": 1.870142936706543, "step": 7560 }, { "epoch": 0.5960583366180529, "grad_norm": 0.7379862666130066, "learning_rate": 0.0005, "loss": 1.650283932685852, "step": 7561 }, { "epoch": 0.5961371698856918, "grad_norm": 0.7627502679824829, "learning_rate": 0.0005, "loss": 1.9459888935089111, "step": 7562 }, { "epoch": 0.5962160031533307, "grad_norm": 0.7205128073692322, "learning_rate": 0.0005, "loss": 2.1993112564086914, "step": 7563 }, { "epoch": 0.5962948364209697, "grad_norm": 0.6799890398979187, "learning_rate": 0.0005, "loss": 2.093412160873413, "step": 7564 }, { "epoch": 0.5963736696886086, "grad_norm": 0.6840742230415344, "learning_rate": 0.0005, "loss": 1.6445574760437012, "step": 7565 }, { "epoch": 0.5964525029562475, "grad_norm": 0.6670187711715698, "learning_rate": 0.0005, "loss": 1.7359685897827148, "step": 7566 }, { "epoch": 0.5965313362238864, "grad_norm": 0.7120644450187683, "learning_rate": 0.0005, "loss": 1.998565673828125, "step": 7567 }, { "epoch": 0.5966101694915255, "grad_norm": 0.8839901089668274, "learning_rate": 0.0005, "loss": 1.9894862174987793, "step": 7568 }, { "epoch": 0.5966890027591644, "grad_norm": 1.6378164291381836, "learning_rate": 0.0005, "loss": 2.1893081665039062, "step": 7569 }, { "epoch": 0.5967678360268033, "grad_norm": 0.5949457883834839, "learning_rate": 0.0005, "loss": 1.7654476165771484, "step": 7570 }, { "epoch": 0.5968466692944423, "grad_norm": 1.073104739189148, "learning_rate": 0.0005, "loss": 2.0223543643951416, "step": 7571 }, { "epoch": 0.5969255025620812, "grad_norm": 1.1045546531677246, "learning_rate": 0.0005, "loss": 2.299926280975342, "step": 7572 }, { "epoch": 0.5970043358297201, "grad_norm": 0.6898534893989563, "learning_rate": 0.0005, "loss": 2.10282564163208, "step": 7573 }, { "epoch": 0.597083169097359, "grad_norm": 0.8171335458755493, "learning_rate": 0.0005, "loss": 2.0894217491149902, "step": 7574 }, { "epoch": 0.5971620023649981, "grad_norm": 0.7444638609886169, "learning_rate": 0.0005, "loss": 2.058269500732422, "step": 7575 }, { "epoch": 0.597240835632637, "grad_norm": 0.8908458352088928, "learning_rate": 0.0005, "loss": 2.0802769660949707, "step": 7576 }, { "epoch": 0.5973196689002759, "grad_norm": 0.912244975566864, "learning_rate": 0.0005, "loss": 2.1581974029541016, "step": 7577 }, { "epoch": 0.5973985021679149, "grad_norm": 0.7789884209632874, "learning_rate": 0.0005, "loss": 2.226217746734619, "step": 7578 }, { "epoch": 0.5974773354355538, "grad_norm": 0.815192699432373, "learning_rate": 0.0005, "loss": 1.919287085533142, "step": 7579 }, { "epoch": 0.5975561687031927, "grad_norm": 0.8633619546890259, "learning_rate": 0.0005, "loss": 2.0505282878875732, "step": 7580 }, { "epoch": 0.5976350019708316, "grad_norm": 0.711729109287262, "learning_rate": 0.0005, "loss": 2.1801540851593018, "step": 7581 }, { "epoch": 0.5977138352384707, "grad_norm": 0.5614473819732666, "learning_rate": 0.0005, "loss": 1.715698003768921, "step": 7582 }, { "epoch": 0.5977926685061096, "grad_norm": 1.5609627962112427, "learning_rate": 0.0005, "loss": 2.4311487674713135, "step": 7583 }, { "epoch": 0.5978715017737485, "grad_norm": 0.7379286885261536, "learning_rate": 0.0005, "loss": 1.8369330167770386, "step": 7584 }, { "epoch": 0.5979503350413875, "grad_norm": 1.8054475784301758, "learning_rate": 0.0005, "loss": 2.225897789001465, "step": 7585 }, { "epoch": 0.5980291683090264, "grad_norm": 0.623668372631073, "learning_rate": 0.0005, "loss": 2.1826577186584473, "step": 7586 }, { "epoch": 0.5981080015766653, "grad_norm": 0.9283813238143921, "learning_rate": 0.0005, "loss": 2.2617263793945312, "step": 7587 }, { "epoch": 0.5981868348443043, "grad_norm": 0.6349437236785889, "learning_rate": 0.0005, "loss": 1.919159173965454, "step": 7588 }, { "epoch": 0.5982656681119433, "grad_norm": 0.6246135234832764, "learning_rate": 0.0005, "loss": 1.9737924337387085, "step": 7589 }, { "epoch": 0.5983445013795822, "grad_norm": 0.7903486490249634, "learning_rate": 0.0005, "loss": 2.1199235916137695, "step": 7590 }, { "epoch": 0.5984233346472211, "grad_norm": 0.5915944576263428, "learning_rate": 0.0005, "loss": 2.0045037269592285, "step": 7591 }, { "epoch": 0.5985021679148601, "grad_norm": 0.6950168013572693, "learning_rate": 0.0005, "loss": 1.922754168510437, "step": 7592 }, { "epoch": 0.598581001182499, "grad_norm": 0.5967329740524292, "learning_rate": 0.0005, "loss": 2.0456113815307617, "step": 7593 }, { "epoch": 0.5986598344501379, "grad_norm": 0.6837626099586487, "learning_rate": 0.0005, "loss": 1.8796837329864502, "step": 7594 }, { "epoch": 0.598738667717777, "grad_norm": 0.9361146688461304, "learning_rate": 0.0005, "loss": 2.1977877616882324, "step": 7595 }, { "epoch": 0.5988175009854159, "grad_norm": 0.6315705180168152, "learning_rate": 0.0005, "loss": 1.7100894451141357, "step": 7596 }, { "epoch": 0.5988963342530548, "grad_norm": 0.8738213777542114, "learning_rate": 0.0005, "loss": 2.4621877670288086, "step": 7597 }, { "epoch": 0.5989751675206937, "grad_norm": 0.6031028628349304, "learning_rate": 0.0005, "loss": 2.062246799468994, "step": 7598 }, { "epoch": 0.5990540007883327, "grad_norm": 0.697933554649353, "learning_rate": 0.0005, "loss": 1.664591670036316, "step": 7599 }, { "epoch": 0.5991328340559716, "grad_norm": 0.6555274724960327, "learning_rate": 0.0005, "loss": 1.9774361848831177, "step": 7600 }, { "epoch": 0.5992116673236105, "grad_norm": 0.6120737791061401, "learning_rate": 0.0005, "loss": 1.9681246280670166, "step": 7601 }, { "epoch": 0.5992905005912496, "grad_norm": 0.6226893663406372, "learning_rate": 0.0005, "loss": 1.7655482292175293, "step": 7602 }, { "epoch": 0.5993693338588885, "grad_norm": 0.6248326301574707, "learning_rate": 0.0005, "loss": 1.7512729167938232, "step": 7603 }, { "epoch": 0.5994481671265274, "grad_norm": 0.6033955216407776, "learning_rate": 0.0005, "loss": 1.6883716583251953, "step": 7604 }, { "epoch": 0.5995270003941663, "grad_norm": 0.5821585059165955, "learning_rate": 0.0005, "loss": 1.9459872245788574, "step": 7605 }, { "epoch": 0.5996058336618053, "grad_norm": 0.6212997436523438, "learning_rate": 0.0005, "loss": 1.6200188398361206, "step": 7606 }, { "epoch": 0.5996846669294442, "grad_norm": 0.7755148410797119, "learning_rate": 0.0005, "loss": 1.8399078845977783, "step": 7607 }, { "epoch": 0.5997635001970831, "grad_norm": 0.6521704196929932, "learning_rate": 0.0005, "loss": 1.865135669708252, "step": 7608 }, { "epoch": 0.5998423334647222, "grad_norm": 0.7136890888214111, "learning_rate": 0.0005, "loss": 2.028733253479004, "step": 7609 }, { "epoch": 0.5999211667323611, "grad_norm": 0.6311696767807007, "learning_rate": 0.0005, "loss": 1.883496880531311, "step": 7610 }, { "epoch": 0.6, "grad_norm": 0.7093007564544678, "learning_rate": 0.0005, "loss": 1.651885747909546, "step": 7611 }, { "epoch": 0.6000788332676389, "grad_norm": 0.82331782579422, "learning_rate": 0.0005, "loss": 2.0218074321746826, "step": 7612 }, { "epoch": 0.6001576665352779, "grad_norm": 0.6851481795310974, "learning_rate": 0.0005, "loss": 2.0419864654541016, "step": 7613 }, { "epoch": 0.6002364998029168, "grad_norm": 0.8003920316696167, "learning_rate": 0.0005, "loss": 2.0713870525360107, "step": 7614 }, { "epoch": 0.6003153330705557, "grad_norm": 0.847341001033783, "learning_rate": 0.0005, "loss": 1.9902729988098145, "step": 7615 }, { "epoch": 0.6003941663381948, "grad_norm": 0.8224207162857056, "learning_rate": 0.0005, "loss": 1.7663154602050781, "step": 7616 }, { "epoch": 0.6004729996058337, "grad_norm": 0.7536410093307495, "learning_rate": 0.0005, "loss": 2.201953887939453, "step": 7617 }, { "epoch": 0.6005518328734726, "grad_norm": 0.7094388604164124, "learning_rate": 0.0005, "loss": 2.1148624420166016, "step": 7618 }, { "epoch": 0.6006306661411116, "grad_norm": 0.7983142137527466, "learning_rate": 0.0005, "loss": 1.8141025304794312, "step": 7619 }, { "epoch": 0.6007094994087505, "grad_norm": 0.668379545211792, "learning_rate": 0.0005, "loss": 2.043060302734375, "step": 7620 }, { "epoch": 0.6007883326763894, "grad_norm": 0.647246241569519, "learning_rate": 0.0005, "loss": 2.1726527214050293, "step": 7621 }, { "epoch": 0.6008671659440283, "grad_norm": 0.6101401448249817, "learning_rate": 0.0005, "loss": 1.9297922849655151, "step": 7622 }, { "epoch": 0.6009459992116674, "grad_norm": 0.7778668999671936, "learning_rate": 0.0005, "loss": 2.013861894607544, "step": 7623 }, { "epoch": 0.6010248324793063, "grad_norm": 0.5813465118408203, "learning_rate": 0.0005, "loss": 1.7974706888198853, "step": 7624 }, { "epoch": 0.6011036657469452, "grad_norm": 0.6635475158691406, "learning_rate": 0.0005, "loss": 2.0439846515655518, "step": 7625 }, { "epoch": 0.6011824990145842, "grad_norm": 0.745154082775116, "learning_rate": 0.0005, "loss": 2.015066146850586, "step": 7626 }, { "epoch": 0.6012613322822231, "grad_norm": 0.6658248901367188, "learning_rate": 0.0005, "loss": 1.8648138046264648, "step": 7627 }, { "epoch": 0.601340165549862, "grad_norm": 0.6016044616699219, "learning_rate": 0.0005, "loss": 1.8706920146942139, "step": 7628 }, { "epoch": 0.6014189988175009, "grad_norm": 0.6643432974815369, "learning_rate": 0.0005, "loss": 1.8510065078735352, "step": 7629 }, { "epoch": 0.60149783208514, "grad_norm": 0.5848356485366821, "learning_rate": 0.0005, "loss": 1.8691749572753906, "step": 7630 }, { "epoch": 0.6015766653527789, "grad_norm": 0.769328236579895, "learning_rate": 0.0005, "loss": 2.0433967113494873, "step": 7631 }, { "epoch": 0.6016554986204178, "grad_norm": 0.5907840728759766, "learning_rate": 0.0005, "loss": 1.7169418334960938, "step": 7632 }, { "epoch": 0.6017343318880568, "grad_norm": 0.7156495451927185, "learning_rate": 0.0005, "loss": 2.0142154693603516, "step": 7633 }, { "epoch": 0.6018131651556957, "grad_norm": 3.253047466278076, "learning_rate": 0.0005, "loss": 2.0591063499450684, "step": 7634 }, { "epoch": 0.6018919984233346, "grad_norm": 0.6018629670143127, "learning_rate": 0.0005, "loss": 1.589669942855835, "step": 7635 }, { "epoch": 0.6019708316909735, "grad_norm": 0.7475101947784424, "learning_rate": 0.0005, "loss": 1.787588357925415, "step": 7636 }, { "epoch": 0.6020496649586126, "grad_norm": 0.7814127206802368, "learning_rate": 0.0005, "loss": 2.043552875518799, "step": 7637 }, { "epoch": 0.6021284982262515, "grad_norm": 0.8004441261291504, "learning_rate": 0.0005, "loss": 2.0352044105529785, "step": 7638 }, { "epoch": 0.6022073314938904, "grad_norm": 0.6323158740997314, "learning_rate": 0.0005, "loss": 1.849981665611267, "step": 7639 }, { "epoch": 0.6022861647615294, "grad_norm": 0.7937262654304504, "learning_rate": 0.0005, "loss": 2.2339062690734863, "step": 7640 }, { "epoch": 0.6023649980291683, "grad_norm": 0.8397883772850037, "learning_rate": 0.0005, "loss": 1.7666563987731934, "step": 7641 }, { "epoch": 0.6024438312968072, "grad_norm": 0.7088157534599304, "learning_rate": 0.0005, "loss": 1.8214311599731445, "step": 7642 }, { "epoch": 0.6025226645644461, "grad_norm": 0.7129523754119873, "learning_rate": 0.0005, "loss": 1.844214677810669, "step": 7643 }, { "epoch": 0.6026014978320852, "grad_norm": 0.6603196263313293, "learning_rate": 0.0005, "loss": 1.7817332744598389, "step": 7644 }, { "epoch": 0.6026803310997241, "grad_norm": 0.7461372017860413, "learning_rate": 0.0005, "loss": 2.018169403076172, "step": 7645 }, { "epoch": 0.602759164367363, "grad_norm": 0.6145225167274475, "learning_rate": 0.0005, "loss": 1.7323024272918701, "step": 7646 }, { "epoch": 0.602837997635002, "grad_norm": 0.7470778226852417, "learning_rate": 0.0005, "loss": 2.023266077041626, "step": 7647 }, { "epoch": 0.6029168309026409, "grad_norm": 0.5743195414543152, "learning_rate": 0.0005, "loss": 1.5364251136779785, "step": 7648 }, { "epoch": 0.6029956641702798, "grad_norm": 0.8495063781738281, "learning_rate": 0.0005, "loss": 1.8745437860488892, "step": 7649 }, { "epoch": 0.6030744974379189, "grad_norm": 0.948910117149353, "learning_rate": 0.0005, "loss": 2.1413888931274414, "step": 7650 }, { "epoch": 0.6031533307055578, "grad_norm": 0.8430313467979431, "learning_rate": 0.0005, "loss": 2.3508353233337402, "step": 7651 }, { "epoch": 0.6032321639731967, "grad_norm": 0.7102949023246765, "learning_rate": 0.0005, "loss": 2.3382997512817383, "step": 7652 }, { "epoch": 0.6033109972408356, "grad_norm": 0.9486050605773926, "learning_rate": 0.0005, "loss": 1.8433310985565186, "step": 7653 }, { "epoch": 0.6033898305084746, "grad_norm": 0.7572183609008789, "learning_rate": 0.0005, "loss": 1.865426778793335, "step": 7654 }, { "epoch": 0.6034686637761135, "grad_norm": 0.8338025212287903, "learning_rate": 0.0005, "loss": 2.122995615005493, "step": 7655 }, { "epoch": 0.6035474970437524, "grad_norm": 0.7118244171142578, "learning_rate": 0.0005, "loss": 2.018141031265259, "step": 7656 }, { "epoch": 0.6036263303113915, "grad_norm": 0.7111934423446655, "learning_rate": 0.0005, "loss": 2.1568188667297363, "step": 7657 }, { "epoch": 0.6037051635790304, "grad_norm": 0.7263141870498657, "learning_rate": 0.0005, "loss": 1.91080904006958, "step": 7658 }, { "epoch": 0.6037839968466693, "grad_norm": 0.6225332617759705, "learning_rate": 0.0005, "loss": 2.074303388595581, "step": 7659 }, { "epoch": 0.6038628301143082, "grad_norm": 0.6758707165718079, "learning_rate": 0.0005, "loss": 1.7504234313964844, "step": 7660 }, { "epoch": 0.6039416633819472, "grad_norm": 0.5468757748603821, "learning_rate": 0.0005, "loss": 2.0014102458953857, "step": 7661 }, { "epoch": 0.6040204966495861, "grad_norm": 0.5655372142791748, "learning_rate": 0.0005, "loss": 1.716697335243225, "step": 7662 }, { "epoch": 0.604099329917225, "grad_norm": 0.6264396905899048, "learning_rate": 0.0005, "loss": 1.7290561199188232, "step": 7663 }, { "epoch": 0.6041781631848641, "grad_norm": 0.741911768913269, "learning_rate": 0.0005, "loss": 2.1917355060577393, "step": 7664 }, { "epoch": 0.604256996452503, "grad_norm": 0.6415625214576721, "learning_rate": 0.0005, "loss": 1.9340094327926636, "step": 7665 }, { "epoch": 0.6043358297201419, "grad_norm": 1.0834054946899414, "learning_rate": 0.0005, "loss": 2.1388258934020996, "step": 7666 }, { "epoch": 0.6044146629877808, "grad_norm": 1.1421691179275513, "learning_rate": 0.0005, "loss": 1.803344964981079, "step": 7667 }, { "epoch": 0.6044934962554198, "grad_norm": 0.7633721828460693, "learning_rate": 0.0005, "loss": 2.0162787437438965, "step": 7668 }, { "epoch": 0.6045723295230587, "grad_norm": 0.5019429326057434, "learning_rate": 0.0005, "loss": 1.5364646911621094, "step": 7669 }, { "epoch": 0.6046511627906976, "grad_norm": 1.5856144428253174, "learning_rate": 0.0005, "loss": 1.6382062435150146, "step": 7670 }, { "epoch": 0.6047299960583367, "grad_norm": 0.7290878891944885, "learning_rate": 0.0005, "loss": 2.048007011413574, "step": 7671 }, { "epoch": 0.6048088293259756, "grad_norm": 0.6541798710823059, "learning_rate": 0.0005, "loss": 1.9116514921188354, "step": 7672 }, { "epoch": 0.6048876625936145, "grad_norm": 0.6937924027442932, "learning_rate": 0.0005, "loss": 1.8792481422424316, "step": 7673 }, { "epoch": 0.6049664958612534, "grad_norm": 0.8850037455558777, "learning_rate": 0.0005, "loss": 2.0132384300231934, "step": 7674 }, { "epoch": 0.6050453291288924, "grad_norm": 0.6450008749961853, "learning_rate": 0.0005, "loss": 1.787233591079712, "step": 7675 }, { "epoch": 0.6051241623965313, "grad_norm": 0.6780219674110413, "learning_rate": 0.0005, "loss": 1.7651160955429077, "step": 7676 }, { "epoch": 0.6052029956641702, "grad_norm": 0.7204359173774719, "learning_rate": 0.0005, "loss": 2.1244754791259766, "step": 7677 }, { "epoch": 0.6052818289318093, "grad_norm": 0.6103798151016235, "learning_rate": 0.0005, "loss": 2.179250717163086, "step": 7678 }, { "epoch": 0.6053606621994482, "grad_norm": 0.8550133109092712, "learning_rate": 0.0005, "loss": 1.98098623752594, "step": 7679 }, { "epoch": 0.6054394954670871, "grad_norm": 0.7485162615776062, "learning_rate": 0.0005, "loss": 2.0209293365478516, "step": 7680 }, { "epoch": 0.6055183287347261, "grad_norm": 0.9963369965553284, "learning_rate": 0.0005, "loss": 1.9450609683990479, "step": 7681 }, { "epoch": 0.605597162002365, "grad_norm": 0.6655940413475037, "learning_rate": 0.0005, "loss": 1.8791531324386597, "step": 7682 }, { "epoch": 0.6056759952700039, "grad_norm": 0.8128695487976074, "learning_rate": 0.0005, "loss": 2.094151020050049, "step": 7683 }, { "epoch": 0.6057548285376428, "grad_norm": 0.8185688853263855, "learning_rate": 0.0005, "loss": 1.9943019151687622, "step": 7684 }, { "epoch": 0.6058336618052819, "grad_norm": 0.699759304523468, "learning_rate": 0.0005, "loss": 1.7333108186721802, "step": 7685 }, { "epoch": 0.6059124950729208, "grad_norm": 0.7435595393180847, "learning_rate": 0.0005, "loss": 2.0585715770721436, "step": 7686 }, { "epoch": 0.6059913283405597, "grad_norm": 0.7962642908096313, "learning_rate": 0.0005, "loss": 2.006835699081421, "step": 7687 }, { "epoch": 0.6060701616081987, "grad_norm": 0.6410244703292847, "learning_rate": 0.0005, "loss": 1.5991005897521973, "step": 7688 }, { "epoch": 0.6061489948758376, "grad_norm": 0.7911721467971802, "learning_rate": 0.0005, "loss": 1.8007320165634155, "step": 7689 }, { "epoch": 0.6062278281434765, "grad_norm": 1.2926419973373413, "learning_rate": 0.0005, "loss": 2.257965564727783, "step": 7690 }, { "epoch": 0.6063066614111154, "grad_norm": 0.6257904171943665, "learning_rate": 0.0005, "loss": 1.675400733947754, "step": 7691 }, { "epoch": 0.6063854946787545, "grad_norm": 0.7098769545555115, "learning_rate": 0.0005, "loss": 2.5102391242980957, "step": 7692 }, { "epoch": 0.6064643279463934, "grad_norm": 0.952411413192749, "learning_rate": 0.0005, "loss": 2.0449025630950928, "step": 7693 }, { "epoch": 0.6065431612140323, "grad_norm": 0.5977345705032349, "learning_rate": 0.0005, "loss": 2.0939927101135254, "step": 7694 }, { "epoch": 0.6066219944816713, "grad_norm": 0.6365813612937927, "learning_rate": 0.0005, "loss": 2.1798181533813477, "step": 7695 }, { "epoch": 0.6067008277493102, "grad_norm": 0.6827133893966675, "learning_rate": 0.0005, "loss": 2.081265449523926, "step": 7696 }, { "epoch": 0.6067796610169491, "grad_norm": 0.8038467168807983, "learning_rate": 0.0005, "loss": 1.8500925302505493, "step": 7697 }, { "epoch": 0.606858494284588, "grad_norm": 0.665000319480896, "learning_rate": 0.0005, "loss": 1.8206576108932495, "step": 7698 }, { "epoch": 0.6069373275522271, "grad_norm": 0.6234785914421082, "learning_rate": 0.0005, "loss": 2.114902973175049, "step": 7699 }, { "epoch": 0.607016160819866, "grad_norm": 0.7816980481147766, "learning_rate": 0.0005, "loss": 1.9150105714797974, "step": 7700 }, { "epoch": 0.6070949940875049, "grad_norm": 0.6439060568809509, "learning_rate": 0.0005, "loss": 1.990065097808838, "step": 7701 }, { "epoch": 0.6071738273551439, "grad_norm": 0.9609915614128113, "learning_rate": 0.0005, "loss": 2.124816417694092, "step": 7702 }, { "epoch": 0.6072526606227828, "grad_norm": 0.6657870411872864, "learning_rate": 0.0005, "loss": 2.2119765281677246, "step": 7703 }, { "epoch": 0.6073314938904217, "grad_norm": 0.576016902923584, "learning_rate": 0.0005, "loss": 1.8456827402114868, "step": 7704 }, { "epoch": 0.6074103271580606, "grad_norm": 0.6570245027542114, "learning_rate": 0.0005, "loss": 1.868696928024292, "step": 7705 }, { "epoch": 0.6074891604256997, "grad_norm": 0.580137312412262, "learning_rate": 0.0005, "loss": 1.9211325645446777, "step": 7706 }, { "epoch": 0.6075679936933386, "grad_norm": 0.7469223141670227, "learning_rate": 0.0005, "loss": 1.9214173555374146, "step": 7707 }, { "epoch": 0.6076468269609775, "grad_norm": 1.9802755117416382, "learning_rate": 0.0005, "loss": 2.180016040802002, "step": 7708 }, { "epoch": 0.6077256602286165, "grad_norm": 0.9129128456115723, "learning_rate": 0.0005, "loss": 2.2531368732452393, "step": 7709 }, { "epoch": 0.6078044934962554, "grad_norm": 0.6821150779724121, "learning_rate": 0.0005, "loss": 2.0422983169555664, "step": 7710 }, { "epoch": 0.6078833267638943, "grad_norm": 0.7244832515716553, "learning_rate": 0.0005, "loss": 2.0897927284240723, "step": 7711 }, { "epoch": 0.6079621600315334, "grad_norm": 0.6638339161872864, "learning_rate": 0.0005, "loss": 1.8905441761016846, "step": 7712 }, { "epoch": 0.6080409932991723, "grad_norm": 0.6410540342330933, "learning_rate": 0.0005, "loss": 1.7351908683776855, "step": 7713 }, { "epoch": 0.6081198265668112, "grad_norm": 0.7083589434623718, "learning_rate": 0.0005, "loss": 2.016907215118408, "step": 7714 }, { "epoch": 0.6081986598344501, "grad_norm": 0.6300932168960571, "learning_rate": 0.0005, "loss": 2.0757646560668945, "step": 7715 }, { "epoch": 0.6082774931020891, "grad_norm": 0.6470307111740112, "learning_rate": 0.0005, "loss": 1.7274436950683594, "step": 7716 }, { "epoch": 0.608356326369728, "grad_norm": 0.8111466765403748, "learning_rate": 0.0005, "loss": 2.046154022216797, "step": 7717 }, { "epoch": 0.6084351596373669, "grad_norm": 0.5199065804481506, "learning_rate": 0.0005, "loss": 1.8104050159454346, "step": 7718 }, { "epoch": 0.608513992905006, "grad_norm": 0.5774438977241516, "learning_rate": 0.0005, "loss": 1.915070652961731, "step": 7719 }, { "epoch": 0.6085928261726449, "grad_norm": 1.4293317794799805, "learning_rate": 0.0005, "loss": 2.095909833908081, "step": 7720 }, { "epoch": 0.6086716594402838, "grad_norm": 0.8000852465629578, "learning_rate": 0.0005, "loss": 2.041419744491577, "step": 7721 }, { "epoch": 0.6087504927079227, "grad_norm": 0.7917728424072266, "learning_rate": 0.0005, "loss": 2.0959343910217285, "step": 7722 }, { "epoch": 0.6088293259755617, "grad_norm": 0.6636001467704773, "learning_rate": 0.0005, "loss": 2.1019039154052734, "step": 7723 }, { "epoch": 0.6089081592432006, "grad_norm": 0.7722229361534119, "learning_rate": 0.0005, "loss": 2.018620014190674, "step": 7724 }, { "epoch": 0.6089869925108395, "grad_norm": 1.2078980207443237, "learning_rate": 0.0005, "loss": 2.224292755126953, "step": 7725 }, { "epoch": 0.6090658257784786, "grad_norm": 0.8109756112098694, "learning_rate": 0.0005, "loss": 2.2095487117767334, "step": 7726 }, { "epoch": 0.6091446590461175, "grad_norm": 0.801239013671875, "learning_rate": 0.0005, "loss": 1.8065061569213867, "step": 7727 }, { "epoch": 0.6092234923137564, "grad_norm": 0.6832506656646729, "learning_rate": 0.0005, "loss": 1.8870302438735962, "step": 7728 }, { "epoch": 0.6093023255813953, "grad_norm": 0.7298136353492737, "learning_rate": 0.0005, "loss": 1.9744362831115723, "step": 7729 }, { "epoch": 0.6093811588490343, "grad_norm": 0.5882171988487244, "learning_rate": 0.0005, "loss": 1.933664083480835, "step": 7730 }, { "epoch": 0.6094599921166732, "grad_norm": 0.6135546565055847, "learning_rate": 0.0005, "loss": 2.008697748184204, "step": 7731 }, { "epoch": 0.6095388253843121, "grad_norm": 0.7733432650566101, "learning_rate": 0.0005, "loss": 1.9158825874328613, "step": 7732 }, { "epoch": 0.6096176586519512, "grad_norm": 0.8343362808227539, "learning_rate": 0.0005, "loss": 1.8441729545593262, "step": 7733 }, { "epoch": 0.6096964919195901, "grad_norm": 0.8029317855834961, "learning_rate": 0.0005, "loss": 1.9659624099731445, "step": 7734 }, { "epoch": 0.609775325187229, "grad_norm": 0.7532074451446533, "learning_rate": 0.0005, "loss": 2.1667704582214355, "step": 7735 }, { "epoch": 0.6098541584548679, "grad_norm": 0.5515525341033936, "learning_rate": 0.0005, "loss": 1.9138762950897217, "step": 7736 }, { "epoch": 0.6099329917225069, "grad_norm": 0.6744198799133301, "learning_rate": 0.0005, "loss": 1.9850715398788452, "step": 7737 }, { "epoch": 0.6100118249901458, "grad_norm": 0.8183950781822205, "learning_rate": 0.0005, "loss": 2.000845432281494, "step": 7738 }, { "epoch": 0.6100906582577847, "grad_norm": 0.7833641171455383, "learning_rate": 0.0005, "loss": 1.8561674356460571, "step": 7739 }, { "epoch": 0.6101694915254238, "grad_norm": 0.833048403263092, "learning_rate": 0.0005, "loss": 1.8487921953201294, "step": 7740 }, { "epoch": 0.6102483247930627, "grad_norm": 0.8020120859146118, "learning_rate": 0.0005, "loss": 2.274838447570801, "step": 7741 }, { "epoch": 0.6103271580607016, "grad_norm": 0.754845142364502, "learning_rate": 0.0005, "loss": 2.1332550048828125, "step": 7742 }, { "epoch": 0.6104059913283406, "grad_norm": 0.7133682370185852, "learning_rate": 0.0005, "loss": 2.0245590209960938, "step": 7743 }, { "epoch": 0.6104848245959795, "grad_norm": 0.6957107186317444, "learning_rate": 0.0005, "loss": 2.089465856552124, "step": 7744 }, { "epoch": 0.6105636578636184, "grad_norm": 0.599891185760498, "learning_rate": 0.0005, "loss": 1.8867719173431396, "step": 7745 }, { "epoch": 0.6106424911312573, "grad_norm": 0.7658184170722961, "learning_rate": 0.0005, "loss": 1.846992015838623, "step": 7746 }, { "epoch": 0.6107213243988964, "grad_norm": 0.6404123902320862, "learning_rate": 0.0005, "loss": 2.038999557495117, "step": 7747 }, { "epoch": 0.6108001576665353, "grad_norm": 0.6132237911224365, "learning_rate": 0.0005, "loss": 1.7475697994232178, "step": 7748 }, { "epoch": 0.6108789909341742, "grad_norm": 0.6688417196273804, "learning_rate": 0.0005, "loss": 1.8370821475982666, "step": 7749 }, { "epoch": 0.6109578242018132, "grad_norm": 0.6515703201293945, "learning_rate": 0.0005, "loss": 1.9897944927215576, "step": 7750 }, { "epoch": 0.6110366574694521, "grad_norm": 0.6645998358726501, "learning_rate": 0.0005, "loss": 2.255876302719116, "step": 7751 }, { "epoch": 0.611115490737091, "grad_norm": 1.0633336305618286, "learning_rate": 0.0005, "loss": 2.137587070465088, "step": 7752 }, { "epoch": 0.61119432400473, "grad_norm": 0.6458523273468018, "learning_rate": 0.0005, "loss": 1.8000354766845703, "step": 7753 }, { "epoch": 0.611273157272369, "grad_norm": 0.7292439937591553, "learning_rate": 0.0005, "loss": 1.9992187023162842, "step": 7754 }, { "epoch": 0.6113519905400079, "grad_norm": 0.8151102662086487, "learning_rate": 0.0005, "loss": 1.8517532348632812, "step": 7755 }, { "epoch": 0.6114308238076468, "grad_norm": 0.7597056031227112, "learning_rate": 0.0005, "loss": 1.8427132368087769, "step": 7756 }, { "epoch": 0.6115096570752858, "grad_norm": 0.8817107081413269, "learning_rate": 0.0005, "loss": 1.8987066745758057, "step": 7757 }, { "epoch": 0.6115884903429247, "grad_norm": 0.9377971291542053, "learning_rate": 0.0005, "loss": 2.221357822418213, "step": 7758 }, { "epoch": 0.6116673236105636, "grad_norm": 0.5756919980049133, "learning_rate": 0.0005, "loss": 1.6894726753234863, "step": 7759 }, { "epoch": 0.6117461568782026, "grad_norm": 0.5979618430137634, "learning_rate": 0.0005, "loss": 2.186648368835449, "step": 7760 }, { "epoch": 0.6118249901458416, "grad_norm": 0.6509280800819397, "learning_rate": 0.0005, "loss": 1.9914780855178833, "step": 7761 }, { "epoch": 0.6119038234134805, "grad_norm": 0.9359279870986938, "learning_rate": 0.0005, "loss": 2.5768463611602783, "step": 7762 }, { "epoch": 0.6119826566811194, "grad_norm": 0.8313981294631958, "learning_rate": 0.0005, "loss": 2.3386855125427246, "step": 7763 }, { "epoch": 0.6120614899487584, "grad_norm": 0.9800131916999817, "learning_rate": 0.0005, "loss": 2.247929334640503, "step": 7764 }, { "epoch": 0.6121403232163973, "grad_norm": 0.8457232713699341, "learning_rate": 0.0005, "loss": 1.7792432308197021, "step": 7765 }, { "epoch": 0.6122191564840362, "grad_norm": 0.8108575344085693, "learning_rate": 0.0005, "loss": 1.9442307949066162, "step": 7766 }, { "epoch": 0.6122979897516752, "grad_norm": 0.894809365272522, "learning_rate": 0.0005, "loss": 1.7588691711425781, "step": 7767 }, { "epoch": 0.6123768230193142, "grad_norm": 0.6506929993629456, "learning_rate": 0.0005, "loss": 1.5907522439956665, "step": 7768 }, { "epoch": 0.6124556562869531, "grad_norm": 0.6421696543693542, "learning_rate": 0.0005, "loss": 2.016158103942871, "step": 7769 }, { "epoch": 0.612534489554592, "grad_norm": 0.765733540058136, "learning_rate": 0.0005, "loss": 2.1380462646484375, "step": 7770 }, { "epoch": 0.612613322822231, "grad_norm": 0.7075734734535217, "learning_rate": 0.0005, "loss": 1.894810676574707, "step": 7771 }, { "epoch": 0.6126921560898699, "grad_norm": 0.6485433578491211, "learning_rate": 0.0005, "loss": 1.9435532093048096, "step": 7772 }, { "epoch": 0.6127709893575088, "grad_norm": 0.708578884601593, "learning_rate": 0.0005, "loss": 2.2244467735290527, "step": 7773 }, { "epoch": 0.6128498226251479, "grad_norm": 0.7899301648139954, "learning_rate": 0.0005, "loss": 1.753993034362793, "step": 7774 }, { "epoch": 0.6129286558927868, "grad_norm": 0.6806843280792236, "learning_rate": 0.0005, "loss": 2.0342190265655518, "step": 7775 }, { "epoch": 0.6130074891604257, "grad_norm": 0.8290082216262817, "learning_rate": 0.0005, "loss": 1.8722681999206543, "step": 7776 }, { "epoch": 0.6130863224280646, "grad_norm": 0.7454754710197449, "learning_rate": 0.0005, "loss": 1.9326825141906738, "step": 7777 }, { "epoch": 0.6131651556957036, "grad_norm": 0.6577631235122681, "learning_rate": 0.0005, "loss": 1.8235775232315063, "step": 7778 }, { "epoch": 0.6132439889633425, "grad_norm": 0.7163167595863342, "learning_rate": 0.0005, "loss": 2.012331485748291, "step": 7779 }, { "epoch": 0.6133228222309814, "grad_norm": 0.767912745475769, "learning_rate": 0.0005, "loss": 1.87567138671875, "step": 7780 }, { "epoch": 0.6134016554986205, "grad_norm": 0.7566722631454468, "learning_rate": 0.0005, "loss": 2.0376598834991455, "step": 7781 }, { "epoch": 0.6134804887662594, "grad_norm": 0.6285141110420227, "learning_rate": 0.0005, "loss": 2.0880966186523438, "step": 7782 }, { "epoch": 0.6135593220338983, "grad_norm": 1.4921396970748901, "learning_rate": 0.0005, "loss": 2.6057605743408203, "step": 7783 }, { "epoch": 0.6136381553015372, "grad_norm": 0.9095253348350525, "learning_rate": 0.0005, "loss": 1.9943995475769043, "step": 7784 }, { "epoch": 0.6137169885691762, "grad_norm": 0.7815788388252258, "learning_rate": 0.0005, "loss": 2.297590970993042, "step": 7785 }, { "epoch": 0.6137958218368151, "grad_norm": 0.5537441968917847, "learning_rate": 0.0005, "loss": 1.610684871673584, "step": 7786 }, { "epoch": 0.613874655104454, "grad_norm": 2.073590040206909, "learning_rate": 0.0005, "loss": 1.970185399055481, "step": 7787 }, { "epoch": 0.6139534883720931, "grad_norm": 1.0627998113632202, "learning_rate": 0.0005, "loss": 1.9686832427978516, "step": 7788 }, { "epoch": 0.614032321639732, "grad_norm": 0.5705973505973816, "learning_rate": 0.0005, "loss": 1.6899563074111938, "step": 7789 }, { "epoch": 0.6141111549073709, "grad_norm": 0.6536740660667419, "learning_rate": 0.0005, "loss": 1.6258713006973267, "step": 7790 }, { "epoch": 0.6141899881750098, "grad_norm": 0.7418392896652222, "learning_rate": 0.0005, "loss": 1.8479878902435303, "step": 7791 }, { "epoch": 0.6142688214426488, "grad_norm": 0.7425833940505981, "learning_rate": 0.0005, "loss": 2.062695026397705, "step": 7792 }, { "epoch": 0.6143476547102877, "grad_norm": 0.6599012017250061, "learning_rate": 0.0005, "loss": 1.9912168979644775, "step": 7793 }, { "epoch": 0.6144264879779266, "grad_norm": 0.758401095867157, "learning_rate": 0.0005, "loss": 2.1965394020080566, "step": 7794 }, { "epoch": 0.6145053212455657, "grad_norm": 0.8315044641494751, "learning_rate": 0.0005, "loss": 2.3814337253570557, "step": 7795 }, { "epoch": 0.6145841545132046, "grad_norm": 0.7642269730567932, "learning_rate": 0.0005, "loss": 1.8479700088500977, "step": 7796 }, { "epoch": 0.6146629877808435, "grad_norm": 0.5507708787918091, "learning_rate": 0.0005, "loss": 1.9212383031845093, "step": 7797 }, { "epoch": 0.6147418210484824, "grad_norm": 0.7125229835510254, "learning_rate": 0.0005, "loss": 2.1160402297973633, "step": 7798 }, { "epoch": 0.6148206543161214, "grad_norm": 0.5310502052307129, "learning_rate": 0.0005, "loss": 1.8527991771697998, "step": 7799 }, { "epoch": 0.6148994875837603, "grad_norm": 0.7354638576507568, "learning_rate": 0.0005, "loss": 2.0748939514160156, "step": 7800 }, { "epoch": 0.6149783208513993, "grad_norm": 0.6895526647567749, "learning_rate": 0.0005, "loss": 2.166382074356079, "step": 7801 }, { "epoch": 0.6150571541190383, "grad_norm": 0.5918729901313782, "learning_rate": 0.0005, "loss": 1.7205716371536255, "step": 7802 }, { "epoch": 0.6151359873866772, "grad_norm": 0.5552065372467041, "learning_rate": 0.0005, "loss": 1.7428547143936157, "step": 7803 }, { "epoch": 0.6152148206543161, "grad_norm": 0.7661557793617249, "learning_rate": 0.0005, "loss": 1.9952218532562256, "step": 7804 }, { "epoch": 0.6152936539219551, "grad_norm": 0.7200088500976562, "learning_rate": 0.0005, "loss": 1.9399781227111816, "step": 7805 }, { "epoch": 0.615372487189594, "grad_norm": 0.7638854384422302, "learning_rate": 0.0005, "loss": 2.1217103004455566, "step": 7806 }, { "epoch": 0.6154513204572329, "grad_norm": 0.6011865139007568, "learning_rate": 0.0005, "loss": 1.7708643674850464, "step": 7807 }, { "epoch": 0.6155301537248719, "grad_norm": 0.8094801306724548, "learning_rate": 0.0005, "loss": 1.8281883001327515, "step": 7808 }, { "epoch": 0.6156089869925109, "grad_norm": 0.6551589369773865, "learning_rate": 0.0005, "loss": 2.070758819580078, "step": 7809 }, { "epoch": 0.6156878202601498, "grad_norm": 0.8532541394233704, "learning_rate": 0.0005, "loss": 1.7330410480499268, "step": 7810 }, { "epoch": 0.6157666535277887, "grad_norm": 0.6977929472923279, "learning_rate": 0.0005, "loss": 2.2501227855682373, "step": 7811 }, { "epoch": 0.6158454867954277, "grad_norm": 0.9660869836807251, "learning_rate": 0.0005, "loss": 2.201078176498413, "step": 7812 }, { "epoch": 0.6159243200630666, "grad_norm": 0.6675713062286377, "learning_rate": 0.0005, "loss": 2.0842437744140625, "step": 7813 }, { "epoch": 0.6160031533307055, "grad_norm": 0.687201201915741, "learning_rate": 0.0005, "loss": 2.208110809326172, "step": 7814 }, { "epoch": 0.6160819865983445, "grad_norm": 0.6187863945960999, "learning_rate": 0.0005, "loss": 2.025428295135498, "step": 7815 }, { "epoch": 0.6161608198659835, "grad_norm": 0.6944299936294556, "learning_rate": 0.0005, "loss": 1.9265927076339722, "step": 7816 }, { "epoch": 0.6162396531336224, "grad_norm": 0.7797004580497742, "learning_rate": 0.0005, "loss": 2.071260452270508, "step": 7817 }, { "epoch": 0.6163184864012613, "grad_norm": 0.8556925058364868, "learning_rate": 0.0005, "loss": 2.3566317558288574, "step": 7818 }, { "epoch": 0.6163973196689003, "grad_norm": 0.7089732885360718, "learning_rate": 0.0005, "loss": 2.00415301322937, "step": 7819 }, { "epoch": 0.6164761529365392, "grad_norm": 0.675103485584259, "learning_rate": 0.0005, "loss": 2.095262050628662, "step": 7820 }, { "epoch": 0.6165549862041781, "grad_norm": 0.7314499616622925, "learning_rate": 0.0005, "loss": 1.9142799377441406, "step": 7821 }, { "epoch": 0.6166338194718171, "grad_norm": 0.5948259234428406, "learning_rate": 0.0005, "loss": 1.7353301048278809, "step": 7822 }, { "epoch": 0.6167126527394561, "grad_norm": 0.7571884393692017, "learning_rate": 0.0005, "loss": 2.1162185668945312, "step": 7823 }, { "epoch": 0.616791486007095, "grad_norm": 0.668750524520874, "learning_rate": 0.0005, "loss": 1.8792376518249512, "step": 7824 }, { "epoch": 0.6168703192747339, "grad_norm": 0.6350967288017273, "learning_rate": 0.0005, "loss": 2.2356951236724854, "step": 7825 }, { "epoch": 0.6169491525423729, "grad_norm": 0.6597984433174133, "learning_rate": 0.0005, "loss": 1.9166970252990723, "step": 7826 }, { "epoch": 0.6170279858100118, "grad_norm": 0.7592209577560425, "learning_rate": 0.0005, "loss": 2.126133441925049, "step": 7827 }, { "epoch": 0.6171068190776507, "grad_norm": 0.6967306733131409, "learning_rate": 0.0005, "loss": 2.05888295173645, "step": 7828 }, { "epoch": 0.6171856523452897, "grad_norm": 0.7825565338134766, "learning_rate": 0.0005, "loss": 1.9221209287643433, "step": 7829 }, { "epoch": 0.6172644856129287, "grad_norm": 0.7995684742927551, "learning_rate": 0.0005, "loss": 2.2940900325775146, "step": 7830 }, { "epoch": 0.6173433188805676, "grad_norm": 0.8684821128845215, "learning_rate": 0.0005, "loss": 2.1263322830200195, "step": 7831 }, { "epoch": 0.6174221521482065, "grad_norm": 0.7411788702011108, "learning_rate": 0.0005, "loss": 2.0364654064178467, "step": 7832 }, { "epoch": 0.6175009854158455, "grad_norm": 0.8947806358337402, "learning_rate": 0.0005, "loss": 1.9172730445861816, "step": 7833 }, { "epoch": 0.6175798186834844, "grad_norm": 0.7604368329048157, "learning_rate": 0.0005, "loss": 2.0296573638916016, "step": 7834 }, { "epoch": 0.6176586519511233, "grad_norm": 0.6839663982391357, "learning_rate": 0.0005, "loss": 2.0064878463745117, "step": 7835 }, { "epoch": 0.6177374852187624, "grad_norm": 1.1340364217758179, "learning_rate": 0.0005, "loss": 2.067448616027832, "step": 7836 }, { "epoch": 0.6178163184864013, "grad_norm": 1.1534202098846436, "learning_rate": 0.0005, "loss": 1.8034124374389648, "step": 7837 }, { "epoch": 0.6178951517540402, "grad_norm": 0.7142545580863953, "learning_rate": 0.0005, "loss": 2.197706699371338, "step": 7838 }, { "epoch": 0.6179739850216791, "grad_norm": 0.8007799983024597, "learning_rate": 0.0005, "loss": 1.9475256204605103, "step": 7839 }, { "epoch": 0.6180528182893181, "grad_norm": 0.7628846168518066, "learning_rate": 0.0005, "loss": 2.035494565963745, "step": 7840 }, { "epoch": 0.618131651556957, "grad_norm": 0.6869367957115173, "learning_rate": 0.0005, "loss": 2.041874885559082, "step": 7841 }, { "epoch": 0.618210484824596, "grad_norm": 0.6279162168502808, "learning_rate": 0.0005, "loss": 1.6292444467544556, "step": 7842 }, { "epoch": 0.618289318092235, "grad_norm": 0.7054912447929382, "learning_rate": 0.0005, "loss": 1.732378363609314, "step": 7843 }, { "epoch": 0.6183681513598739, "grad_norm": 0.7965982556343079, "learning_rate": 0.0005, "loss": 1.8801201581954956, "step": 7844 }, { "epoch": 0.6184469846275128, "grad_norm": 0.8931815028190613, "learning_rate": 0.0005, "loss": 2.0503430366516113, "step": 7845 }, { "epoch": 0.6185258178951517, "grad_norm": 0.6982487440109253, "learning_rate": 0.0005, "loss": 1.9803707599639893, "step": 7846 }, { "epoch": 0.6186046511627907, "grad_norm": 0.8096261620521545, "learning_rate": 0.0005, "loss": 2.067854881286621, "step": 7847 }, { "epoch": 0.6186834844304296, "grad_norm": 0.8186665177345276, "learning_rate": 0.0005, "loss": 2.143268585205078, "step": 7848 }, { "epoch": 0.6187623176980686, "grad_norm": 0.7616584897041321, "learning_rate": 0.0005, "loss": 2.1046791076660156, "step": 7849 }, { "epoch": 0.6188411509657076, "grad_norm": 0.5894580483436584, "learning_rate": 0.0005, "loss": 1.661630630493164, "step": 7850 }, { "epoch": 0.6189199842333465, "grad_norm": 0.7092646360397339, "learning_rate": 0.0005, "loss": 2.0990381240844727, "step": 7851 }, { "epoch": 0.6189988175009854, "grad_norm": 0.6973830461502075, "learning_rate": 0.0005, "loss": 2.0938329696655273, "step": 7852 }, { "epoch": 0.6190776507686243, "grad_norm": 0.6812970638275146, "learning_rate": 0.0005, "loss": 1.684657335281372, "step": 7853 }, { "epoch": 0.6191564840362633, "grad_norm": 0.7669672966003418, "learning_rate": 0.0005, "loss": 1.9774043560028076, "step": 7854 }, { "epoch": 0.6192353173039022, "grad_norm": 0.739186704158783, "learning_rate": 0.0005, "loss": 1.8692493438720703, "step": 7855 }, { "epoch": 0.6193141505715412, "grad_norm": 0.6185120940208435, "learning_rate": 0.0005, "loss": 2.1564583778381348, "step": 7856 }, { "epoch": 0.6193929838391802, "grad_norm": 0.5816425085067749, "learning_rate": 0.0005, "loss": 1.6487278938293457, "step": 7857 }, { "epoch": 0.6194718171068191, "grad_norm": 0.8710684180259705, "learning_rate": 0.0005, "loss": 1.9692586660385132, "step": 7858 }, { "epoch": 0.619550650374458, "grad_norm": 0.7659351229667664, "learning_rate": 0.0005, "loss": 1.868116855621338, "step": 7859 }, { "epoch": 0.6196294836420969, "grad_norm": 0.5864940285682678, "learning_rate": 0.0005, "loss": 1.8691350221633911, "step": 7860 }, { "epoch": 0.6197083169097359, "grad_norm": 0.7586115002632141, "learning_rate": 0.0005, "loss": 1.8220239877700806, "step": 7861 }, { "epoch": 0.6197871501773748, "grad_norm": 0.6404208540916443, "learning_rate": 0.0005, "loss": 2.0156233310699463, "step": 7862 }, { "epoch": 0.6198659834450138, "grad_norm": 0.6766176819801331, "learning_rate": 0.0005, "loss": 2.08469557762146, "step": 7863 }, { "epoch": 0.6199448167126528, "grad_norm": 0.675993025302887, "learning_rate": 0.0005, "loss": 1.8581756353378296, "step": 7864 }, { "epoch": 0.6200236499802917, "grad_norm": 0.6423795223236084, "learning_rate": 0.0005, "loss": 2.0414745807647705, "step": 7865 }, { "epoch": 0.6201024832479306, "grad_norm": 0.7072766423225403, "learning_rate": 0.0005, "loss": 2.144038438796997, "step": 7866 }, { "epoch": 0.6201813165155696, "grad_norm": 0.8251670598983765, "learning_rate": 0.0005, "loss": 1.8938097953796387, "step": 7867 }, { "epoch": 0.6202601497832085, "grad_norm": 0.6382574439048767, "learning_rate": 0.0005, "loss": 1.7451848983764648, "step": 7868 }, { "epoch": 0.6203389830508474, "grad_norm": 0.6205853223800659, "learning_rate": 0.0005, "loss": 1.9297631978988647, "step": 7869 }, { "epoch": 0.6204178163184864, "grad_norm": 0.7393835186958313, "learning_rate": 0.0005, "loss": 1.6972764730453491, "step": 7870 }, { "epoch": 0.6204966495861254, "grad_norm": 0.6492877006530762, "learning_rate": 0.0005, "loss": 1.624358892440796, "step": 7871 }, { "epoch": 0.6205754828537643, "grad_norm": 0.6642666459083557, "learning_rate": 0.0005, "loss": 2.1119866371154785, "step": 7872 }, { "epoch": 0.6206543161214032, "grad_norm": 0.6985116600990295, "learning_rate": 0.0005, "loss": 2.251923084259033, "step": 7873 }, { "epoch": 0.6207331493890422, "grad_norm": 0.779819905757904, "learning_rate": 0.0005, "loss": 2.280179023742676, "step": 7874 }, { "epoch": 0.6208119826566811, "grad_norm": 0.616561770439148, "learning_rate": 0.0005, "loss": 1.7709842920303345, "step": 7875 }, { "epoch": 0.62089081592432, "grad_norm": 0.6321876049041748, "learning_rate": 0.0005, "loss": 1.9029160737991333, "step": 7876 }, { "epoch": 0.620969649191959, "grad_norm": 0.5569067597389221, "learning_rate": 0.0005, "loss": 1.7618165016174316, "step": 7877 }, { "epoch": 0.621048482459598, "grad_norm": 0.7529124021530151, "learning_rate": 0.0005, "loss": 1.875410556793213, "step": 7878 }, { "epoch": 0.6211273157272369, "grad_norm": 0.6989037394523621, "learning_rate": 0.0005, "loss": 1.993054747581482, "step": 7879 }, { "epoch": 0.6212061489948758, "grad_norm": 0.842589795589447, "learning_rate": 0.0005, "loss": 1.9284337759017944, "step": 7880 }, { "epoch": 0.6212849822625148, "grad_norm": 0.6687557697296143, "learning_rate": 0.0005, "loss": 1.8515090942382812, "step": 7881 }, { "epoch": 0.6213638155301537, "grad_norm": 0.6295689344406128, "learning_rate": 0.0005, "loss": 2.0475199222564697, "step": 7882 }, { "epoch": 0.6214426487977927, "grad_norm": 0.7118863463401794, "learning_rate": 0.0005, "loss": 2.3173913955688477, "step": 7883 }, { "epoch": 0.6215214820654316, "grad_norm": 0.7397261261940002, "learning_rate": 0.0005, "loss": 2.0026214122772217, "step": 7884 }, { "epoch": 0.6216003153330706, "grad_norm": 0.7036344408988953, "learning_rate": 0.0005, "loss": 2.2866873741149902, "step": 7885 }, { "epoch": 0.6216791486007095, "grad_norm": 0.6403814554214478, "learning_rate": 0.0005, "loss": 1.886625051498413, "step": 7886 }, { "epoch": 0.6217579818683484, "grad_norm": 1.008912205696106, "learning_rate": 0.0005, "loss": 1.8256477117538452, "step": 7887 }, { "epoch": 0.6218368151359874, "grad_norm": 0.862349808216095, "learning_rate": 0.0005, "loss": 1.7618975639343262, "step": 7888 }, { "epoch": 0.6219156484036263, "grad_norm": 0.6734375357627869, "learning_rate": 0.0005, "loss": 1.866590142250061, "step": 7889 }, { "epoch": 0.6219944816712653, "grad_norm": 0.7844937443733215, "learning_rate": 0.0005, "loss": 1.8442366123199463, "step": 7890 }, { "epoch": 0.6220733149389042, "grad_norm": 0.6481330394744873, "learning_rate": 0.0005, "loss": 1.904129981994629, "step": 7891 }, { "epoch": 0.6221521482065432, "grad_norm": 0.6336777210235596, "learning_rate": 0.0005, "loss": 1.8258881568908691, "step": 7892 }, { "epoch": 0.6222309814741821, "grad_norm": 0.750697135925293, "learning_rate": 0.0005, "loss": 1.9658756256103516, "step": 7893 }, { "epoch": 0.622309814741821, "grad_norm": 0.6090053915977478, "learning_rate": 0.0005, "loss": 1.8434092998504639, "step": 7894 }, { "epoch": 0.62238864800946, "grad_norm": 0.5454962253570557, "learning_rate": 0.0005, "loss": 2.015233039855957, "step": 7895 }, { "epoch": 0.6224674812770989, "grad_norm": 0.643890917301178, "learning_rate": 0.0005, "loss": 1.9155128002166748, "step": 7896 }, { "epoch": 0.6225463145447379, "grad_norm": 1.1609450578689575, "learning_rate": 0.0005, "loss": 1.974050521850586, "step": 7897 }, { "epoch": 0.6226251478123769, "grad_norm": 0.7352462410926819, "learning_rate": 0.0005, "loss": 1.9848374128341675, "step": 7898 }, { "epoch": 0.6227039810800158, "grad_norm": 0.6224905848503113, "learning_rate": 0.0005, "loss": 1.9650801420211792, "step": 7899 }, { "epoch": 0.6227828143476547, "grad_norm": 1.0002009868621826, "learning_rate": 0.0005, "loss": 1.9414732456207275, "step": 7900 }, { "epoch": 0.6228616476152936, "grad_norm": 0.8108724355697632, "learning_rate": 0.0005, "loss": 2.224179267883301, "step": 7901 }, { "epoch": 0.6229404808829326, "grad_norm": 0.7538080215454102, "learning_rate": 0.0005, "loss": 2.019711494445801, "step": 7902 }, { "epoch": 0.6230193141505715, "grad_norm": 0.9522905349731445, "learning_rate": 0.0005, "loss": 2.3169665336608887, "step": 7903 }, { "epoch": 0.6230981474182105, "grad_norm": 0.6430705785751343, "learning_rate": 0.0005, "loss": 1.875999927520752, "step": 7904 }, { "epoch": 0.6231769806858495, "grad_norm": 0.7382544279098511, "learning_rate": 0.0005, "loss": 1.9617544412612915, "step": 7905 }, { "epoch": 0.6232558139534884, "grad_norm": 0.8529183864593506, "learning_rate": 0.0005, "loss": 2.186418056488037, "step": 7906 }, { "epoch": 0.6233346472211273, "grad_norm": 0.6077398061752319, "learning_rate": 0.0005, "loss": 1.6556861400604248, "step": 7907 }, { "epoch": 0.6234134804887662, "grad_norm": 0.6108144521713257, "learning_rate": 0.0005, "loss": 2.0909650325775146, "step": 7908 }, { "epoch": 0.6234923137564052, "grad_norm": 1.2005680799484253, "learning_rate": 0.0005, "loss": 1.974328637123108, "step": 7909 }, { "epoch": 0.6235711470240441, "grad_norm": 0.5887544751167297, "learning_rate": 0.0005, "loss": 1.888201355934143, "step": 7910 }, { "epoch": 0.6236499802916831, "grad_norm": 0.8072364330291748, "learning_rate": 0.0005, "loss": 1.9652434587478638, "step": 7911 }, { "epoch": 0.6237288135593221, "grad_norm": 0.91798335313797, "learning_rate": 0.0005, "loss": 2.3010122776031494, "step": 7912 }, { "epoch": 0.623807646826961, "grad_norm": 0.7833639979362488, "learning_rate": 0.0005, "loss": 1.7774568796157837, "step": 7913 }, { "epoch": 0.6238864800945999, "grad_norm": 0.6144251227378845, "learning_rate": 0.0005, "loss": 2.028879165649414, "step": 7914 }, { "epoch": 0.6239653133622388, "grad_norm": 0.6421524882316589, "learning_rate": 0.0005, "loss": 1.8266397714614868, "step": 7915 }, { "epoch": 0.6240441466298778, "grad_norm": 0.7593070864677429, "learning_rate": 0.0005, "loss": 1.7870569229125977, "step": 7916 }, { "epoch": 0.6241229798975167, "grad_norm": 0.6602139472961426, "learning_rate": 0.0005, "loss": 1.8130649328231812, "step": 7917 }, { "epoch": 0.6242018131651557, "grad_norm": 0.8007247447967529, "learning_rate": 0.0005, "loss": 2.065190315246582, "step": 7918 }, { "epoch": 0.6242806464327947, "grad_norm": 0.705246090888977, "learning_rate": 0.0005, "loss": 2.0955393314361572, "step": 7919 }, { "epoch": 0.6243594797004336, "grad_norm": 1.156091570854187, "learning_rate": 0.0005, "loss": 1.9652016162872314, "step": 7920 }, { "epoch": 0.6244383129680725, "grad_norm": 0.5826235413551331, "learning_rate": 0.0005, "loss": 2.0860605239868164, "step": 7921 }, { "epoch": 0.6245171462357114, "grad_norm": 0.8372243642807007, "learning_rate": 0.0005, "loss": 1.7542965412139893, "step": 7922 }, { "epoch": 0.6245959795033504, "grad_norm": 0.7172059416770935, "learning_rate": 0.0005, "loss": 1.9186517000198364, "step": 7923 }, { "epoch": 0.6246748127709894, "grad_norm": 0.7313645482063293, "learning_rate": 0.0005, "loss": 2.2309484481811523, "step": 7924 }, { "epoch": 0.6247536460386283, "grad_norm": 0.7060967087745667, "learning_rate": 0.0005, "loss": 1.9199644327163696, "step": 7925 }, { "epoch": 0.6248324793062673, "grad_norm": 0.6590510010719299, "learning_rate": 0.0005, "loss": 2.014692783355713, "step": 7926 }, { "epoch": 0.6249113125739062, "grad_norm": 0.6286236643791199, "learning_rate": 0.0005, "loss": 1.8606714010238647, "step": 7927 }, { "epoch": 0.6249901458415451, "grad_norm": 0.9105036854743958, "learning_rate": 0.0005, "loss": 1.9405016899108887, "step": 7928 }, { "epoch": 0.6250689791091841, "grad_norm": 0.7481178045272827, "learning_rate": 0.0005, "loss": 1.7916661500930786, "step": 7929 }, { "epoch": 0.625147812376823, "grad_norm": 0.7626765370368958, "learning_rate": 0.0005, "loss": 1.9889295101165771, "step": 7930 }, { "epoch": 0.625226645644462, "grad_norm": 0.644576907157898, "learning_rate": 0.0005, "loss": 2.146566390991211, "step": 7931 }, { "epoch": 0.6253054789121009, "grad_norm": 0.5958629250526428, "learning_rate": 0.0005, "loss": 2.100435256958008, "step": 7932 }, { "epoch": 0.6253843121797399, "grad_norm": 0.6388061046600342, "learning_rate": 0.0005, "loss": 1.952974557876587, "step": 7933 }, { "epoch": 0.6254631454473788, "grad_norm": 0.7548669576644897, "learning_rate": 0.0005, "loss": 1.8843576908111572, "step": 7934 }, { "epoch": 0.6255419787150177, "grad_norm": 0.7185406684875488, "learning_rate": 0.0005, "loss": 1.7374579906463623, "step": 7935 }, { "epoch": 0.6256208119826567, "grad_norm": 0.6481385231018066, "learning_rate": 0.0005, "loss": 1.8911477327346802, "step": 7936 }, { "epoch": 0.6256996452502956, "grad_norm": 0.7730635404586792, "learning_rate": 0.0005, "loss": 1.9488792419433594, "step": 7937 }, { "epoch": 0.6257784785179346, "grad_norm": 0.5876734852790833, "learning_rate": 0.0005, "loss": 1.8357669115066528, "step": 7938 }, { "epoch": 0.6258573117855735, "grad_norm": 0.6029050350189209, "learning_rate": 0.0005, "loss": 1.8771213293075562, "step": 7939 }, { "epoch": 0.6259361450532125, "grad_norm": 0.6540886163711548, "learning_rate": 0.0005, "loss": 1.7830495834350586, "step": 7940 }, { "epoch": 0.6260149783208514, "grad_norm": 0.5615285634994507, "learning_rate": 0.0005, "loss": 1.9642128944396973, "step": 7941 }, { "epoch": 0.6260938115884903, "grad_norm": 0.7242400050163269, "learning_rate": 0.0005, "loss": 2.0956220626831055, "step": 7942 }, { "epoch": 0.6261726448561293, "grad_norm": 0.693418562412262, "learning_rate": 0.0005, "loss": 1.8145332336425781, "step": 7943 }, { "epoch": 0.6262514781237682, "grad_norm": 0.5957058668136597, "learning_rate": 0.0005, "loss": 2.099759101867676, "step": 7944 }, { "epoch": 0.6263303113914072, "grad_norm": 0.5548215508460999, "learning_rate": 0.0005, "loss": 1.6776244640350342, "step": 7945 }, { "epoch": 0.6264091446590461, "grad_norm": 0.7383337616920471, "learning_rate": 0.0005, "loss": 1.8518704175949097, "step": 7946 }, { "epoch": 0.6264879779266851, "grad_norm": 0.6905514597892761, "learning_rate": 0.0005, "loss": 2.275932550430298, "step": 7947 }, { "epoch": 0.626566811194324, "grad_norm": 0.6715543866157532, "learning_rate": 0.0005, "loss": 1.812477707862854, "step": 7948 }, { "epoch": 0.6266456444619629, "grad_norm": 0.835199236869812, "learning_rate": 0.0005, "loss": 1.8643383979797363, "step": 7949 }, { "epoch": 0.6267244777296019, "grad_norm": 0.7073977589607239, "learning_rate": 0.0005, "loss": 2.101778507232666, "step": 7950 }, { "epoch": 0.6268033109972408, "grad_norm": 0.6449385285377502, "learning_rate": 0.0005, "loss": 1.9481371641159058, "step": 7951 }, { "epoch": 0.6268821442648798, "grad_norm": 0.6145459413528442, "learning_rate": 0.0005, "loss": 1.8875168561935425, "step": 7952 }, { "epoch": 0.6269609775325187, "grad_norm": 0.5980973243713379, "learning_rate": 0.0005, "loss": 1.8757365942001343, "step": 7953 }, { "epoch": 0.6270398108001577, "grad_norm": 0.685809850692749, "learning_rate": 0.0005, "loss": 1.8964511156082153, "step": 7954 }, { "epoch": 0.6271186440677966, "grad_norm": 0.6836221218109131, "learning_rate": 0.0005, "loss": 1.8218753337860107, "step": 7955 }, { "epoch": 0.6271974773354355, "grad_norm": 1.022123098373413, "learning_rate": 0.0005, "loss": 1.8883068561553955, "step": 7956 }, { "epoch": 0.6272763106030745, "grad_norm": 0.6672208905220032, "learning_rate": 0.0005, "loss": 1.961833119392395, "step": 7957 }, { "epoch": 0.6273551438707134, "grad_norm": 0.7122788429260254, "learning_rate": 0.0005, "loss": 1.9026074409484863, "step": 7958 }, { "epoch": 0.6274339771383524, "grad_norm": 0.671273946762085, "learning_rate": 0.0005, "loss": 2.0614845752716064, "step": 7959 }, { "epoch": 0.6275128104059914, "grad_norm": 0.6307066082954407, "learning_rate": 0.0005, "loss": 2.0014753341674805, "step": 7960 }, { "epoch": 0.6275916436736303, "grad_norm": 0.6670500040054321, "learning_rate": 0.0005, "loss": 1.9440765380859375, "step": 7961 }, { "epoch": 0.6276704769412692, "grad_norm": 0.7191141247749329, "learning_rate": 0.0005, "loss": 2.2454657554626465, "step": 7962 }, { "epoch": 0.6277493102089081, "grad_norm": 0.6792474389076233, "learning_rate": 0.0005, "loss": 1.8896570205688477, "step": 7963 }, { "epoch": 0.6278281434765471, "grad_norm": 0.6188526153564453, "learning_rate": 0.0005, "loss": 1.7073357105255127, "step": 7964 }, { "epoch": 0.627906976744186, "grad_norm": 0.7526249289512634, "learning_rate": 0.0005, "loss": 2.0150980949401855, "step": 7965 }, { "epoch": 0.627985810011825, "grad_norm": 0.6515436172485352, "learning_rate": 0.0005, "loss": 1.6699154376983643, "step": 7966 }, { "epoch": 0.628064643279464, "grad_norm": 0.677460253238678, "learning_rate": 0.0005, "loss": 1.865581750869751, "step": 7967 }, { "epoch": 0.6281434765471029, "grad_norm": 0.8070083856582642, "learning_rate": 0.0005, "loss": 2.201796531677246, "step": 7968 }, { "epoch": 0.6282223098147418, "grad_norm": 0.6422936320304871, "learning_rate": 0.0005, "loss": 1.8123226165771484, "step": 7969 }, { "epoch": 0.6283011430823807, "grad_norm": 0.6821611523628235, "learning_rate": 0.0005, "loss": 1.918182134628296, "step": 7970 }, { "epoch": 0.6283799763500197, "grad_norm": 0.7525246143341064, "learning_rate": 0.0005, "loss": 2.2881970405578613, "step": 7971 }, { "epoch": 0.6284588096176587, "grad_norm": 1.073047399520874, "learning_rate": 0.0005, "loss": 1.9894603490829468, "step": 7972 }, { "epoch": 0.6285376428852976, "grad_norm": 0.9014083743095398, "learning_rate": 0.0005, "loss": 1.939136028289795, "step": 7973 }, { "epoch": 0.6286164761529366, "grad_norm": 0.7044394016265869, "learning_rate": 0.0005, "loss": 1.904030442237854, "step": 7974 }, { "epoch": 0.6286953094205755, "grad_norm": 0.7360153794288635, "learning_rate": 0.0005, "loss": 2.2367215156555176, "step": 7975 }, { "epoch": 0.6287741426882144, "grad_norm": 0.6486571431159973, "learning_rate": 0.0005, "loss": 1.8219002485275269, "step": 7976 }, { "epoch": 0.6288529759558533, "grad_norm": 0.9065880179405212, "learning_rate": 0.0005, "loss": 2.0006327629089355, "step": 7977 }, { "epoch": 0.6289318092234923, "grad_norm": 0.7068051099777222, "learning_rate": 0.0005, "loss": 1.861035943031311, "step": 7978 }, { "epoch": 0.6290106424911313, "grad_norm": 0.6389260292053223, "learning_rate": 0.0005, "loss": 2.0054538249969482, "step": 7979 }, { "epoch": 0.6290894757587702, "grad_norm": 0.670844554901123, "learning_rate": 0.0005, "loss": 1.9617613554000854, "step": 7980 }, { "epoch": 0.6291683090264092, "grad_norm": 0.7910451889038086, "learning_rate": 0.0005, "loss": 2.0800468921661377, "step": 7981 }, { "epoch": 0.6292471422940481, "grad_norm": 0.6164593696594238, "learning_rate": 0.0005, "loss": 1.973865270614624, "step": 7982 }, { "epoch": 0.629325975561687, "grad_norm": 0.7666727900505066, "learning_rate": 0.0005, "loss": 2.0526349544525146, "step": 7983 }, { "epoch": 0.6294048088293259, "grad_norm": 0.6720907688140869, "learning_rate": 0.0005, "loss": 2.139984607696533, "step": 7984 }, { "epoch": 0.629483642096965, "grad_norm": 0.6277740001678467, "learning_rate": 0.0005, "loss": 1.7213212251663208, "step": 7985 }, { "epoch": 0.6295624753646039, "grad_norm": 0.8410170078277588, "learning_rate": 0.0005, "loss": 2.18735408782959, "step": 7986 }, { "epoch": 0.6296413086322428, "grad_norm": 0.6558997631072998, "learning_rate": 0.0005, "loss": 2.055779457092285, "step": 7987 }, { "epoch": 0.6297201418998818, "grad_norm": 0.6392652988433838, "learning_rate": 0.0005, "loss": 1.9615445137023926, "step": 7988 }, { "epoch": 0.6297989751675207, "grad_norm": 0.6064314842224121, "learning_rate": 0.0005, "loss": 1.973820686340332, "step": 7989 }, { "epoch": 0.6298778084351596, "grad_norm": 0.802216112613678, "learning_rate": 0.0005, "loss": 2.0026562213897705, "step": 7990 }, { "epoch": 0.6299566417027986, "grad_norm": 0.6504228115081787, "learning_rate": 0.0005, "loss": 1.9702740907669067, "step": 7991 }, { "epoch": 0.6300354749704375, "grad_norm": 0.7397983074188232, "learning_rate": 0.0005, "loss": 1.9692901372909546, "step": 7992 }, { "epoch": 0.6301143082380765, "grad_norm": 0.7196206450462341, "learning_rate": 0.0005, "loss": 2.0515031814575195, "step": 7993 }, { "epoch": 0.6301931415057154, "grad_norm": 0.7439385652542114, "learning_rate": 0.0005, "loss": 1.9366453886032104, "step": 7994 }, { "epoch": 0.6302719747733544, "grad_norm": 0.6863455176353455, "learning_rate": 0.0005, "loss": 1.9727381467819214, "step": 7995 }, { "epoch": 0.6303508080409933, "grad_norm": 0.6365745067596436, "learning_rate": 0.0005, "loss": 2.1428346633911133, "step": 7996 }, { "epoch": 0.6304296413086322, "grad_norm": 0.9650020003318787, "learning_rate": 0.0005, "loss": 2.1493523120880127, "step": 7997 }, { "epoch": 0.6305084745762712, "grad_norm": 0.520074188709259, "learning_rate": 0.0005, "loss": 1.7763652801513672, "step": 7998 }, { "epoch": 0.6305873078439101, "grad_norm": 0.6317260265350342, "learning_rate": 0.0005, "loss": 1.9316531419754028, "step": 7999 }, { "epoch": 0.6306661411115491, "grad_norm": 0.7732667326927185, "learning_rate": 0.0005, "loss": 2.016343593597412, "step": 8000 }, { "epoch": 0.630744974379188, "grad_norm": 1.4159334897994995, "learning_rate": 0.0005, "loss": 2.067133903503418, "step": 8001 }, { "epoch": 0.630823807646827, "grad_norm": 0.6015551090240479, "learning_rate": 0.0005, "loss": 1.8224109411239624, "step": 8002 }, { "epoch": 0.6309026409144659, "grad_norm": 0.5650236010551453, "learning_rate": 0.0005, "loss": 1.8336066007614136, "step": 8003 }, { "epoch": 0.6309814741821048, "grad_norm": 1.1045403480529785, "learning_rate": 0.0005, "loss": 1.7677160501480103, "step": 8004 }, { "epoch": 0.6310603074497438, "grad_norm": 0.5859236717224121, "learning_rate": 0.0005, "loss": 1.9367061853408813, "step": 8005 }, { "epoch": 0.6311391407173828, "grad_norm": 0.7911934852600098, "learning_rate": 0.0005, "loss": 2.0635976791381836, "step": 8006 }, { "epoch": 0.6312179739850217, "grad_norm": 0.627652645111084, "learning_rate": 0.0005, "loss": 1.9893416166305542, "step": 8007 }, { "epoch": 0.6312968072526606, "grad_norm": 0.5670990943908691, "learning_rate": 0.0005, "loss": 1.916915774345398, "step": 8008 }, { "epoch": 0.6313756405202996, "grad_norm": 0.9727463126182556, "learning_rate": 0.0005, "loss": 2.0177059173583984, "step": 8009 }, { "epoch": 0.6314544737879385, "grad_norm": 0.6209296584129333, "learning_rate": 0.0005, "loss": 1.8988697528839111, "step": 8010 }, { "epoch": 0.6315333070555774, "grad_norm": 0.9577561020851135, "learning_rate": 0.0005, "loss": 1.9656245708465576, "step": 8011 }, { "epoch": 0.6316121403232164, "grad_norm": 0.8774923086166382, "learning_rate": 0.0005, "loss": 1.8344852924346924, "step": 8012 }, { "epoch": 0.6316909735908554, "grad_norm": 0.6049273610115051, "learning_rate": 0.0005, "loss": 1.666628122329712, "step": 8013 }, { "epoch": 0.6317698068584943, "grad_norm": 0.6167004704475403, "learning_rate": 0.0005, "loss": 1.8213082551956177, "step": 8014 }, { "epoch": 0.6318486401261332, "grad_norm": 0.8241738677024841, "learning_rate": 0.0005, "loss": 1.956248164176941, "step": 8015 }, { "epoch": 0.6319274733937722, "grad_norm": 0.8688602447509766, "learning_rate": 0.0005, "loss": 1.9207489490509033, "step": 8016 }, { "epoch": 0.6320063066614111, "grad_norm": 0.6314960718154907, "learning_rate": 0.0005, "loss": 1.8188066482543945, "step": 8017 }, { "epoch": 0.63208513992905, "grad_norm": 0.6076542735099792, "learning_rate": 0.0005, "loss": 1.6634697914123535, "step": 8018 }, { "epoch": 0.632163973196689, "grad_norm": 0.7283501029014587, "learning_rate": 0.0005, "loss": 1.9155802726745605, "step": 8019 }, { "epoch": 0.632242806464328, "grad_norm": 0.9659722447395325, "learning_rate": 0.0005, "loss": 2.0523040294647217, "step": 8020 }, { "epoch": 0.6323216397319669, "grad_norm": 1.0601626634597778, "learning_rate": 0.0005, "loss": 1.7754361629486084, "step": 8021 }, { "epoch": 0.6324004729996058, "grad_norm": 0.6475048065185547, "learning_rate": 0.0005, "loss": 2.045124053955078, "step": 8022 }, { "epoch": 0.6324793062672448, "grad_norm": 0.5912690162658691, "learning_rate": 0.0005, "loss": 1.7644782066345215, "step": 8023 }, { "epoch": 0.6325581395348837, "grad_norm": 0.8009953498840332, "learning_rate": 0.0005, "loss": 1.8400328159332275, "step": 8024 }, { "epoch": 0.6326369728025226, "grad_norm": 0.5324380993843079, "learning_rate": 0.0005, "loss": 1.9385132789611816, "step": 8025 }, { "epoch": 0.6327158060701616, "grad_norm": 0.7074233293533325, "learning_rate": 0.0005, "loss": 1.866342544555664, "step": 8026 }, { "epoch": 0.6327946393378006, "grad_norm": 0.649309515953064, "learning_rate": 0.0005, "loss": 2.0355913639068604, "step": 8027 }, { "epoch": 0.6328734726054395, "grad_norm": 1.501896619796753, "learning_rate": 0.0005, "loss": 1.9943208694458008, "step": 8028 }, { "epoch": 0.6329523058730785, "grad_norm": 0.6329449415206909, "learning_rate": 0.0005, "loss": 1.7894717454910278, "step": 8029 }, { "epoch": 0.6330311391407174, "grad_norm": 1.0826430320739746, "learning_rate": 0.0005, "loss": 2.082603931427002, "step": 8030 }, { "epoch": 0.6331099724083563, "grad_norm": 0.8938332200050354, "learning_rate": 0.0005, "loss": 2.0007033348083496, "step": 8031 }, { "epoch": 0.6331888056759952, "grad_norm": 0.7636827230453491, "learning_rate": 0.0005, "loss": 2.428173303604126, "step": 8032 }, { "epoch": 0.6332676389436342, "grad_norm": 0.8951476812362671, "learning_rate": 0.0005, "loss": 1.8348569869995117, "step": 8033 }, { "epoch": 0.6333464722112732, "grad_norm": 0.8090243935585022, "learning_rate": 0.0005, "loss": 1.9028971195220947, "step": 8034 }, { "epoch": 0.6334253054789121, "grad_norm": 0.6723201870918274, "learning_rate": 0.0005, "loss": 1.6854259967803955, "step": 8035 }, { "epoch": 0.6335041387465511, "grad_norm": 0.5682494044303894, "learning_rate": 0.0005, "loss": 1.7673661708831787, "step": 8036 }, { "epoch": 0.63358297201419, "grad_norm": 0.666979968547821, "learning_rate": 0.0005, "loss": 1.9081931114196777, "step": 8037 }, { "epoch": 0.6336618052818289, "grad_norm": 0.5355668067932129, "learning_rate": 0.0005, "loss": 1.9116344451904297, "step": 8038 }, { "epoch": 0.6337406385494678, "grad_norm": 0.8906972408294678, "learning_rate": 0.0005, "loss": 2.010059356689453, "step": 8039 }, { "epoch": 0.6338194718171068, "grad_norm": 0.6713274121284485, "learning_rate": 0.0005, "loss": 1.763986587524414, "step": 8040 }, { "epoch": 0.6338983050847458, "grad_norm": 0.589422345161438, "learning_rate": 0.0005, "loss": 1.9065324068069458, "step": 8041 }, { "epoch": 0.6339771383523847, "grad_norm": 0.7685873508453369, "learning_rate": 0.0005, "loss": 2.2119579315185547, "step": 8042 }, { "epoch": 0.6340559716200237, "grad_norm": 0.7933538556098938, "learning_rate": 0.0005, "loss": 1.9126238822937012, "step": 8043 }, { "epoch": 0.6341348048876626, "grad_norm": 0.6636650562286377, "learning_rate": 0.0005, "loss": 2.107583999633789, "step": 8044 }, { "epoch": 0.6342136381553015, "grad_norm": 1.2774330377578735, "learning_rate": 0.0005, "loss": 2.1823360919952393, "step": 8045 }, { "epoch": 0.6342924714229404, "grad_norm": 0.6949257254600525, "learning_rate": 0.0005, "loss": 1.8713977336883545, "step": 8046 }, { "epoch": 0.6343713046905795, "grad_norm": 0.8548584580421448, "learning_rate": 0.0005, "loss": 2.087230682373047, "step": 8047 }, { "epoch": 0.6344501379582184, "grad_norm": 0.6645081043243408, "learning_rate": 0.0005, "loss": 2.027813196182251, "step": 8048 }, { "epoch": 0.6345289712258573, "grad_norm": 2.4586873054504395, "learning_rate": 0.0005, "loss": 2.0163769721984863, "step": 8049 }, { "epoch": 0.6346078044934963, "grad_norm": 0.7084152698516846, "learning_rate": 0.0005, "loss": 1.8740828037261963, "step": 8050 }, { "epoch": 0.6346866377611352, "grad_norm": 0.7255890369415283, "learning_rate": 0.0005, "loss": 2.2542800903320312, "step": 8051 }, { "epoch": 0.6347654710287741, "grad_norm": 0.8703327775001526, "learning_rate": 0.0005, "loss": 2.063040256500244, "step": 8052 }, { "epoch": 0.634844304296413, "grad_norm": 0.717380940914154, "learning_rate": 0.0005, "loss": 2.3448450565338135, "step": 8053 }, { "epoch": 0.634923137564052, "grad_norm": 0.7567645311355591, "learning_rate": 0.0005, "loss": 1.7826355695724487, "step": 8054 }, { "epoch": 0.635001970831691, "grad_norm": 0.6014901399612427, "learning_rate": 0.0005, "loss": 1.7124205827713013, "step": 8055 }, { "epoch": 0.6350808040993299, "grad_norm": 0.6826550960540771, "learning_rate": 0.0005, "loss": 2.2472422122955322, "step": 8056 }, { "epoch": 0.6351596373669689, "grad_norm": 0.7335334420204163, "learning_rate": 0.0005, "loss": 2.0974836349487305, "step": 8057 }, { "epoch": 0.6352384706346078, "grad_norm": 0.6221922039985657, "learning_rate": 0.0005, "loss": 1.8894469738006592, "step": 8058 }, { "epoch": 0.6353173039022467, "grad_norm": 0.6533574461936951, "learning_rate": 0.0005, "loss": 1.9726712703704834, "step": 8059 }, { "epoch": 0.6353961371698857, "grad_norm": 1.6360359191894531, "learning_rate": 0.0005, "loss": 2.158224582672119, "step": 8060 }, { "epoch": 0.6354749704375247, "grad_norm": 0.7363919615745544, "learning_rate": 0.0005, "loss": 1.912322759628296, "step": 8061 }, { "epoch": 0.6355538037051636, "grad_norm": 0.660290002822876, "learning_rate": 0.0005, "loss": 1.9028183221817017, "step": 8062 }, { "epoch": 0.6356326369728025, "grad_norm": 0.7833247184753418, "learning_rate": 0.0005, "loss": 2.1063272953033447, "step": 8063 }, { "epoch": 0.6357114702404415, "grad_norm": 0.7624629139900208, "learning_rate": 0.0005, "loss": 2.178663730621338, "step": 8064 }, { "epoch": 0.6357903035080804, "grad_norm": 0.565091609954834, "learning_rate": 0.0005, "loss": 1.7756178379058838, "step": 8065 }, { "epoch": 0.6358691367757193, "grad_norm": 0.6692348718643188, "learning_rate": 0.0005, "loss": 2.0386037826538086, "step": 8066 }, { "epoch": 0.6359479700433583, "grad_norm": 0.8642399907112122, "learning_rate": 0.0005, "loss": 2.03474760055542, "step": 8067 }, { "epoch": 0.6360268033109973, "grad_norm": 0.6326513886451721, "learning_rate": 0.0005, "loss": 2.0069942474365234, "step": 8068 }, { "epoch": 0.6361056365786362, "grad_norm": 0.6733282804489136, "learning_rate": 0.0005, "loss": 1.9720993041992188, "step": 8069 }, { "epoch": 0.6361844698462751, "grad_norm": 0.6544129848480225, "learning_rate": 0.0005, "loss": 1.8995453119277954, "step": 8070 }, { "epoch": 0.6362633031139141, "grad_norm": 0.619870126247406, "learning_rate": 0.0005, "loss": 2.2588000297546387, "step": 8071 }, { "epoch": 0.636342136381553, "grad_norm": 0.6854743361473083, "learning_rate": 0.0005, "loss": 2.0609538555145264, "step": 8072 }, { "epoch": 0.6364209696491919, "grad_norm": 0.6249291300773621, "learning_rate": 0.0005, "loss": 1.9903719425201416, "step": 8073 }, { "epoch": 0.636499802916831, "grad_norm": 0.7033778429031372, "learning_rate": 0.0005, "loss": 1.9657241106033325, "step": 8074 }, { "epoch": 0.6365786361844699, "grad_norm": 0.7230069637298584, "learning_rate": 0.0005, "loss": 1.9026066064834595, "step": 8075 }, { "epoch": 0.6366574694521088, "grad_norm": 0.6215095520019531, "learning_rate": 0.0005, "loss": 2.05250883102417, "step": 8076 }, { "epoch": 0.6367363027197477, "grad_norm": 0.6314377188682556, "learning_rate": 0.0005, "loss": 1.8375625610351562, "step": 8077 }, { "epoch": 0.6368151359873867, "grad_norm": 1.7437098026275635, "learning_rate": 0.0005, "loss": 2.0119712352752686, "step": 8078 }, { "epoch": 0.6368939692550256, "grad_norm": 0.7210612297058105, "learning_rate": 0.0005, "loss": 1.9372780323028564, "step": 8079 }, { "epoch": 0.6369728025226645, "grad_norm": 0.5893077850341797, "learning_rate": 0.0005, "loss": 1.807631492614746, "step": 8080 }, { "epoch": 0.6370516357903035, "grad_norm": 0.6406525373458862, "learning_rate": 0.0005, "loss": 1.9782795906066895, "step": 8081 }, { "epoch": 0.6371304690579425, "grad_norm": 0.7212018370628357, "learning_rate": 0.0005, "loss": 1.789022445678711, "step": 8082 }, { "epoch": 0.6372093023255814, "grad_norm": 0.7719012498855591, "learning_rate": 0.0005, "loss": 1.867842674255371, "step": 8083 }, { "epoch": 0.6372881355932203, "grad_norm": 0.6006649732589722, "learning_rate": 0.0005, "loss": 2.056044578552246, "step": 8084 }, { "epoch": 0.6373669688608593, "grad_norm": 0.6167123913764954, "learning_rate": 0.0005, "loss": 2.188058853149414, "step": 8085 }, { "epoch": 0.6374458021284982, "grad_norm": 0.6160167455673218, "learning_rate": 0.0005, "loss": 2.1974644660949707, "step": 8086 }, { "epoch": 0.6375246353961371, "grad_norm": 3.5535213947296143, "learning_rate": 0.0005, "loss": 2.28725528717041, "step": 8087 }, { "epoch": 0.6376034686637762, "grad_norm": 0.8711931705474854, "learning_rate": 0.0005, "loss": 1.956334114074707, "step": 8088 }, { "epoch": 0.6376823019314151, "grad_norm": 0.7897620797157288, "learning_rate": 0.0005, "loss": 2.204512119293213, "step": 8089 }, { "epoch": 0.637761135199054, "grad_norm": 0.658521831035614, "learning_rate": 0.0005, "loss": 1.7766904830932617, "step": 8090 }, { "epoch": 0.637839968466693, "grad_norm": 0.7655677199363708, "learning_rate": 0.0005, "loss": 2.006073474884033, "step": 8091 }, { "epoch": 0.6379188017343319, "grad_norm": 0.6752690672874451, "learning_rate": 0.0005, "loss": 1.7648143768310547, "step": 8092 }, { "epoch": 0.6379976350019708, "grad_norm": 0.6936795711517334, "learning_rate": 0.0005, "loss": 1.7663187980651855, "step": 8093 }, { "epoch": 0.6380764682696097, "grad_norm": 0.9555419087409973, "learning_rate": 0.0005, "loss": 2.07212495803833, "step": 8094 }, { "epoch": 0.6381553015372488, "grad_norm": 0.6656067967414856, "learning_rate": 0.0005, "loss": 2.082404613494873, "step": 8095 }, { "epoch": 0.6382341348048877, "grad_norm": 0.6426931619644165, "learning_rate": 0.0005, "loss": 2.090467929840088, "step": 8096 }, { "epoch": 0.6383129680725266, "grad_norm": 0.8091958165168762, "learning_rate": 0.0005, "loss": 2.2227864265441895, "step": 8097 }, { "epoch": 0.6383918013401656, "grad_norm": 0.7286891341209412, "learning_rate": 0.0005, "loss": 2.055060863494873, "step": 8098 }, { "epoch": 0.6384706346078045, "grad_norm": 0.6885048747062683, "learning_rate": 0.0005, "loss": 1.92728590965271, "step": 8099 }, { "epoch": 0.6385494678754434, "grad_norm": 0.8240960836410522, "learning_rate": 0.0005, "loss": 1.8715935945510864, "step": 8100 }, { "epoch": 0.6386283011430823, "grad_norm": 0.6415705680847168, "learning_rate": 0.0005, "loss": 1.993984580039978, "step": 8101 }, { "epoch": 0.6387071344107214, "grad_norm": 0.7813156247138977, "learning_rate": 0.0005, "loss": 2.05991530418396, "step": 8102 }, { "epoch": 0.6387859676783603, "grad_norm": 0.6678496599197388, "learning_rate": 0.0005, "loss": 1.9167323112487793, "step": 8103 }, { "epoch": 0.6388648009459992, "grad_norm": 0.7675507068634033, "learning_rate": 0.0005, "loss": 1.9446609020233154, "step": 8104 }, { "epoch": 0.6389436342136382, "grad_norm": 0.7212470769882202, "learning_rate": 0.0005, "loss": 1.7872803211212158, "step": 8105 }, { "epoch": 0.6390224674812771, "grad_norm": 0.646408200263977, "learning_rate": 0.0005, "loss": 2.0630016326904297, "step": 8106 }, { "epoch": 0.639101300748916, "grad_norm": 0.823733389377594, "learning_rate": 0.0005, "loss": 1.8553245067596436, "step": 8107 }, { "epoch": 0.6391801340165549, "grad_norm": 0.6833798289299011, "learning_rate": 0.0005, "loss": 1.8999199867248535, "step": 8108 }, { "epoch": 0.639258967284194, "grad_norm": 0.7076898813247681, "learning_rate": 0.0005, "loss": 1.9463253021240234, "step": 8109 }, { "epoch": 0.6393378005518329, "grad_norm": 0.8038158416748047, "learning_rate": 0.0005, "loss": 1.9529927968978882, "step": 8110 }, { "epoch": 0.6394166338194718, "grad_norm": 0.686625063419342, "learning_rate": 0.0005, "loss": 1.928532361984253, "step": 8111 }, { "epoch": 0.6394954670871108, "grad_norm": 0.6449020504951477, "learning_rate": 0.0005, "loss": 2.134519577026367, "step": 8112 }, { "epoch": 0.6395743003547497, "grad_norm": 0.640824556350708, "learning_rate": 0.0005, "loss": 1.8773472309112549, "step": 8113 }, { "epoch": 0.6396531336223886, "grad_norm": 0.6782270669937134, "learning_rate": 0.0005, "loss": 1.6930859088897705, "step": 8114 }, { "epoch": 0.6397319668900275, "grad_norm": 0.6093318462371826, "learning_rate": 0.0005, "loss": 2.1217033863067627, "step": 8115 }, { "epoch": 0.6398108001576666, "grad_norm": 0.5909536480903625, "learning_rate": 0.0005, "loss": 1.7716705799102783, "step": 8116 }, { "epoch": 0.6398896334253055, "grad_norm": 0.7314227223396301, "learning_rate": 0.0005, "loss": 2.3487071990966797, "step": 8117 }, { "epoch": 0.6399684666929444, "grad_norm": 0.6041474938392639, "learning_rate": 0.0005, "loss": 1.7540313005447388, "step": 8118 }, { "epoch": 0.6400472999605834, "grad_norm": 1.1739892959594727, "learning_rate": 0.0005, "loss": 2.048879384994507, "step": 8119 }, { "epoch": 0.6401261332282223, "grad_norm": 2.121445655822754, "learning_rate": 0.0005, "loss": 2.150632381439209, "step": 8120 }, { "epoch": 0.6402049664958612, "grad_norm": 0.7590043544769287, "learning_rate": 0.0005, "loss": 1.9836270809173584, "step": 8121 }, { "epoch": 0.6402837997635002, "grad_norm": 0.7256433963775635, "learning_rate": 0.0005, "loss": 1.6368770599365234, "step": 8122 }, { "epoch": 0.6403626330311392, "grad_norm": 0.6499599814414978, "learning_rate": 0.0005, "loss": 2.0675048828125, "step": 8123 }, { "epoch": 0.6404414662987781, "grad_norm": 0.7311793565750122, "learning_rate": 0.0005, "loss": 2.110227108001709, "step": 8124 }, { "epoch": 0.640520299566417, "grad_norm": 0.5910196304321289, "learning_rate": 0.0005, "loss": 1.8849174976348877, "step": 8125 }, { "epoch": 0.640599132834056, "grad_norm": 0.6065644025802612, "learning_rate": 0.0005, "loss": 1.99261474609375, "step": 8126 }, { "epoch": 0.6406779661016949, "grad_norm": 0.8551362156867981, "learning_rate": 0.0005, "loss": 2.0989766120910645, "step": 8127 }, { "epoch": 0.6407567993693338, "grad_norm": 0.6411532759666443, "learning_rate": 0.0005, "loss": 1.7118239402770996, "step": 8128 }, { "epoch": 0.6408356326369729, "grad_norm": 0.6941370964050293, "learning_rate": 0.0005, "loss": 1.8422166109085083, "step": 8129 }, { "epoch": 0.6409144659046118, "grad_norm": 0.6663609147071838, "learning_rate": 0.0005, "loss": 2.01920747756958, "step": 8130 }, { "epoch": 0.6409932991722507, "grad_norm": 0.6602540016174316, "learning_rate": 0.0005, "loss": 1.9670342206954956, "step": 8131 }, { "epoch": 0.6410721324398896, "grad_norm": 0.5632356405258179, "learning_rate": 0.0005, "loss": 1.9799163341522217, "step": 8132 }, { "epoch": 0.6411509657075286, "grad_norm": 0.7367009520530701, "learning_rate": 0.0005, "loss": 1.767988681793213, "step": 8133 }, { "epoch": 0.6412297989751675, "grad_norm": 0.6025672554969788, "learning_rate": 0.0005, "loss": 1.547196388244629, "step": 8134 }, { "epoch": 0.6413086322428064, "grad_norm": 0.7392664551734924, "learning_rate": 0.0005, "loss": 2.0753774642944336, "step": 8135 }, { "epoch": 0.6413874655104455, "grad_norm": 0.7255401015281677, "learning_rate": 0.0005, "loss": 2.030829429626465, "step": 8136 }, { "epoch": 0.6414662987780844, "grad_norm": 0.8419644832611084, "learning_rate": 0.0005, "loss": 2.153409481048584, "step": 8137 }, { "epoch": 0.6415451320457233, "grad_norm": 0.7345002889633179, "learning_rate": 0.0005, "loss": 1.884939193725586, "step": 8138 }, { "epoch": 0.6416239653133622, "grad_norm": 0.8101310133934021, "learning_rate": 0.0005, "loss": 2.1880857944488525, "step": 8139 }, { "epoch": 0.6417027985810012, "grad_norm": 0.8813620209693909, "learning_rate": 0.0005, "loss": 1.9126346111297607, "step": 8140 }, { "epoch": 0.6417816318486401, "grad_norm": 0.7090926766395569, "learning_rate": 0.0005, "loss": 2.400653839111328, "step": 8141 }, { "epoch": 0.641860465116279, "grad_norm": 0.5850433111190796, "learning_rate": 0.0005, "loss": 1.8191242218017578, "step": 8142 }, { "epoch": 0.6419392983839181, "grad_norm": 0.7959818243980408, "learning_rate": 0.0005, "loss": 2.132990837097168, "step": 8143 }, { "epoch": 0.642018131651557, "grad_norm": 0.7825940847396851, "learning_rate": 0.0005, "loss": 1.8088757991790771, "step": 8144 }, { "epoch": 0.6420969649191959, "grad_norm": 0.8344801068305969, "learning_rate": 0.0005, "loss": 1.9983713626861572, "step": 8145 }, { "epoch": 0.6421757981868348, "grad_norm": 0.6598977446556091, "learning_rate": 0.0005, "loss": 2.1056277751922607, "step": 8146 }, { "epoch": 0.6422546314544738, "grad_norm": 0.5649371147155762, "learning_rate": 0.0005, "loss": 1.7283098697662354, "step": 8147 }, { "epoch": 0.6423334647221127, "grad_norm": 0.6996720433235168, "learning_rate": 0.0005, "loss": 2.472553014755249, "step": 8148 }, { "epoch": 0.6424122979897516, "grad_norm": 0.6912568211555481, "learning_rate": 0.0005, "loss": 1.8114902973175049, "step": 8149 }, { "epoch": 0.6424911312573907, "grad_norm": 0.6652288436889648, "learning_rate": 0.0005, "loss": 1.528122901916504, "step": 8150 }, { "epoch": 0.6425699645250296, "grad_norm": 0.7278953790664673, "learning_rate": 0.0005, "loss": 1.770901083946228, "step": 8151 }, { "epoch": 0.6426487977926685, "grad_norm": 0.7948997020721436, "learning_rate": 0.0005, "loss": 1.8755834102630615, "step": 8152 }, { "epoch": 0.6427276310603075, "grad_norm": 0.7185433506965637, "learning_rate": 0.0005, "loss": 2.0994787216186523, "step": 8153 }, { "epoch": 0.6428064643279464, "grad_norm": 0.7902995347976685, "learning_rate": 0.0005, "loss": 2.089456796646118, "step": 8154 }, { "epoch": 0.6428852975955853, "grad_norm": 0.6879950761795044, "learning_rate": 0.0005, "loss": 1.9946943521499634, "step": 8155 }, { "epoch": 0.6429641308632242, "grad_norm": 0.746030867099762, "learning_rate": 0.0005, "loss": 2.2992491722106934, "step": 8156 }, { "epoch": 0.6430429641308633, "grad_norm": 0.8112070560455322, "learning_rate": 0.0005, "loss": 2.108351230621338, "step": 8157 }, { "epoch": 0.6431217973985022, "grad_norm": 0.6244791746139526, "learning_rate": 0.0005, "loss": 1.8575440645217896, "step": 8158 }, { "epoch": 0.6432006306661411, "grad_norm": 0.6629212498664856, "learning_rate": 0.0005, "loss": 1.7824089527130127, "step": 8159 }, { "epoch": 0.6432794639337801, "grad_norm": 0.750413715839386, "learning_rate": 0.0005, "loss": 1.977848768234253, "step": 8160 }, { "epoch": 0.643358297201419, "grad_norm": 0.7409748435020447, "learning_rate": 0.0005, "loss": 1.8767335414886475, "step": 8161 }, { "epoch": 0.6434371304690579, "grad_norm": 1.0010972023010254, "learning_rate": 0.0005, "loss": 1.8961683511734009, "step": 8162 }, { "epoch": 0.6435159637366968, "grad_norm": 0.6528675556182861, "learning_rate": 0.0005, "loss": 1.746232509613037, "step": 8163 }, { "epoch": 0.6435947970043359, "grad_norm": 0.7425708174705505, "learning_rate": 0.0005, "loss": 2.079472541809082, "step": 8164 }, { "epoch": 0.6436736302719748, "grad_norm": 0.7138482928276062, "learning_rate": 0.0005, "loss": 2.207624912261963, "step": 8165 }, { "epoch": 0.6437524635396137, "grad_norm": 0.6119658946990967, "learning_rate": 0.0005, "loss": 1.9082067012786865, "step": 8166 }, { "epoch": 0.6438312968072527, "grad_norm": 0.5792962908744812, "learning_rate": 0.0005, "loss": 1.7345983982086182, "step": 8167 }, { "epoch": 0.6439101300748916, "grad_norm": 0.7558194994926453, "learning_rate": 0.0005, "loss": 1.799170732498169, "step": 8168 }, { "epoch": 0.6439889633425305, "grad_norm": 0.7663613557815552, "learning_rate": 0.0005, "loss": 2.1892824172973633, "step": 8169 }, { "epoch": 0.6440677966101694, "grad_norm": 0.7413734197616577, "learning_rate": 0.0005, "loss": 1.9179773330688477, "step": 8170 }, { "epoch": 0.6441466298778085, "grad_norm": 1.312962293624878, "learning_rate": 0.0005, "loss": 1.8340470790863037, "step": 8171 }, { "epoch": 0.6442254631454474, "grad_norm": 0.8130562901496887, "learning_rate": 0.0005, "loss": 2.113590717315674, "step": 8172 }, { "epoch": 0.6443042964130863, "grad_norm": 0.7978177070617676, "learning_rate": 0.0005, "loss": 1.9955103397369385, "step": 8173 }, { "epoch": 0.6443831296807253, "grad_norm": 0.6943460702896118, "learning_rate": 0.0005, "loss": 2.0846588611602783, "step": 8174 }, { "epoch": 0.6444619629483642, "grad_norm": 0.7168995141983032, "learning_rate": 0.0005, "loss": 2.074125289916992, "step": 8175 }, { "epoch": 0.6445407962160031, "grad_norm": 0.7801298499107361, "learning_rate": 0.0005, "loss": 2.025169610977173, "step": 8176 }, { "epoch": 0.644619629483642, "grad_norm": 0.6802406907081604, "learning_rate": 0.0005, "loss": 1.8763823509216309, "step": 8177 }, { "epoch": 0.6446984627512811, "grad_norm": 0.8398820161819458, "learning_rate": 0.0005, "loss": 2.0075132846832275, "step": 8178 }, { "epoch": 0.64477729601892, "grad_norm": 0.5932385325431824, "learning_rate": 0.0005, "loss": 2.0922248363494873, "step": 8179 }, { "epoch": 0.6448561292865589, "grad_norm": 0.9306070804595947, "learning_rate": 0.0005, "loss": 1.8655387163162231, "step": 8180 }, { "epoch": 0.6449349625541979, "grad_norm": 0.9022364616394043, "learning_rate": 0.0005, "loss": 1.7964086532592773, "step": 8181 }, { "epoch": 0.6450137958218368, "grad_norm": 0.8066693544387817, "learning_rate": 0.0005, "loss": 1.8654634952545166, "step": 8182 }, { "epoch": 0.6450926290894757, "grad_norm": 0.67714923620224, "learning_rate": 0.0005, "loss": 1.8528454303741455, "step": 8183 }, { "epoch": 0.6451714623571148, "grad_norm": 0.8685013651847839, "learning_rate": 0.0005, "loss": 1.9124367237091064, "step": 8184 }, { "epoch": 0.6452502956247537, "grad_norm": 0.5569533705711365, "learning_rate": 0.0005, "loss": 1.6715104579925537, "step": 8185 }, { "epoch": 0.6453291288923926, "grad_norm": 0.7861062288284302, "learning_rate": 0.0005, "loss": 2.021529197692871, "step": 8186 }, { "epoch": 0.6454079621600315, "grad_norm": 0.658187985420227, "learning_rate": 0.0005, "loss": 2.200018882751465, "step": 8187 }, { "epoch": 0.6454867954276705, "grad_norm": 0.7153075933456421, "learning_rate": 0.0005, "loss": 2.168292999267578, "step": 8188 }, { "epoch": 0.6455656286953094, "grad_norm": 0.6344463229179382, "learning_rate": 0.0005, "loss": 1.7996461391448975, "step": 8189 }, { "epoch": 0.6456444619629483, "grad_norm": 0.6205416917800903, "learning_rate": 0.0005, "loss": 2.079413890838623, "step": 8190 }, { "epoch": 0.6457232952305874, "grad_norm": 0.595081627368927, "learning_rate": 0.0005, "loss": 1.9731706380844116, "step": 8191 }, { "epoch": 0.6458021284982263, "grad_norm": 0.9471068382263184, "learning_rate": 0.0005, "loss": 2.1495161056518555, "step": 8192 }, { "epoch": 0.6458809617658652, "grad_norm": 0.5787721276283264, "learning_rate": 0.0005, "loss": 1.8878084421157837, "step": 8193 }, { "epoch": 0.6459597950335041, "grad_norm": 0.6749175190925598, "learning_rate": 0.0005, "loss": 2.033703327178955, "step": 8194 }, { "epoch": 0.6460386283011431, "grad_norm": 0.8074145317077637, "learning_rate": 0.0005, "loss": 2.3109943866729736, "step": 8195 }, { "epoch": 0.646117461568782, "grad_norm": 0.7526283860206604, "learning_rate": 0.0005, "loss": 1.8246726989746094, "step": 8196 }, { "epoch": 0.6461962948364209, "grad_norm": 0.6642453670501709, "learning_rate": 0.0005, "loss": 1.978135347366333, "step": 8197 }, { "epoch": 0.64627512810406, "grad_norm": 0.6299697756767273, "learning_rate": 0.0005, "loss": 2.1749236583709717, "step": 8198 }, { "epoch": 0.6463539613716989, "grad_norm": 0.6653615832328796, "learning_rate": 0.0005, "loss": 2.1664717197418213, "step": 8199 }, { "epoch": 0.6464327946393378, "grad_norm": 0.7533428072929382, "learning_rate": 0.0005, "loss": 1.968038558959961, "step": 8200 }, { "epoch": 0.6465116279069767, "grad_norm": 0.659822404384613, "learning_rate": 0.0005, "loss": 2.2042784690856934, "step": 8201 }, { "epoch": 0.6465904611746157, "grad_norm": 0.755220353603363, "learning_rate": 0.0005, "loss": 1.7559170722961426, "step": 8202 }, { "epoch": 0.6466692944422546, "grad_norm": 0.5824292898178101, "learning_rate": 0.0005, "loss": 1.7499467134475708, "step": 8203 }, { "epoch": 0.6467481277098935, "grad_norm": 1.845077395439148, "learning_rate": 0.0005, "loss": 1.9075465202331543, "step": 8204 }, { "epoch": 0.6468269609775326, "grad_norm": 0.705975353717804, "learning_rate": 0.0005, "loss": 1.9634437561035156, "step": 8205 }, { "epoch": 0.6469057942451715, "grad_norm": 0.6928243637084961, "learning_rate": 0.0005, "loss": 1.958971381187439, "step": 8206 }, { "epoch": 0.6469846275128104, "grad_norm": 0.929805338382721, "learning_rate": 0.0005, "loss": 1.8945367336273193, "step": 8207 }, { "epoch": 0.6470634607804493, "grad_norm": 0.5178948640823364, "learning_rate": 0.0005, "loss": 1.5903007984161377, "step": 8208 }, { "epoch": 0.6471422940480883, "grad_norm": 0.8864415287971497, "learning_rate": 0.0005, "loss": 1.8417831659317017, "step": 8209 }, { "epoch": 0.6472211273157272, "grad_norm": 0.7113584280014038, "learning_rate": 0.0005, "loss": 1.9445505142211914, "step": 8210 }, { "epoch": 0.6472999605833661, "grad_norm": 0.7000446319580078, "learning_rate": 0.0005, "loss": 1.9883843660354614, "step": 8211 }, { "epoch": 0.6473787938510052, "grad_norm": 0.8571217060089111, "learning_rate": 0.0005, "loss": 1.9208389520645142, "step": 8212 }, { "epoch": 0.6474576271186441, "grad_norm": 0.6935285925865173, "learning_rate": 0.0005, "loss": 1.6078979969024658, "step": 8213 }, { "epoch": 0.647536460386283, "grad_norm": 0.7925455570220947, "learning_rate": 0.0005, "loss": 1.760308027267456, "step": 8214 }, { "epoch": 0.647615293653922, "grad_norm": 0.7373059988021851, "learning_rate": 0.0005, "loss": 1.917776346206665, "step": 8215 }, { "epoch": 0.6476941269215609, "grad_norm": 0.7092247605323792, "learning_rate": 0.0005, "loss": 1.982851505279541, "step": 8216 }, { "epoch": 0.6477729601891998, "grad_norm": 0.7048954963684082, "learning_rate": 0.0005, "loss": 1.944828987121582, "step": 8217 }, { "epoch": 0.6478517934568387, "grad_norm": 0.6544216871261597, "learning_rate": 0.0005, "loss": 1.979628086090088, "step": 8218 }, { "epoch": 0.6479306267244778, "grad_norm": 0.7103351950645447, "learning_rate": 0.0005, "loss": 2.0115256309509277, "step": 8219 }, { "epoch": 0.6480094599921167, "grad_norm": 0.7187413573265076, "learning_rate": 0.0005, "loss": 2.081777334213257, "step": 8220 }, { "epoch": 0.6480882932597556, "grad_norm": 0.7328738570213318, "learning_rate": 0.0005, "loss": 2.068952798843384, "step": 8221 }, { "epoch": 0.6481671265273946, "grad_norm": 0.6411043405532837, "learning_rate": 0.0005, "loss": 2.0558178424835205, "step": 8222 }, { "epoch": 0.6482459597950335, "grad_norm": 0.6727764010429382, "learning_rate": 0.0005, "loss": 1.821988821029663, "step": 8223 }, { "epoch": 0.6483247930626724, "grad_norm": 0.686969518661499, "learning_rate": 0.0005, "loss": 1.8718124628067017, "step": 8224 }, { "epoch": 0.6484036263303113, "grad_norm": 0.5454424023628235, "learning_rate": 0.0005, "loss": 1.7898683547973633, "step": 8225 }, { "epoch": 0.6484824595979504, "grad_norm": 0.7905969023704529, "learning_rate": 0.0005, "loss": 2.1143672466278076, "step": 8226 }, { "epoch": 0.6485612928655893, "grad_norm": 0.7420629262924194, "learning_rate": 0.0005, "loss": 1.9910863637924194, "step": 8227 }, { "epoch": 0.6486401261332282, "grad_norm": 0.6517621874809265, "learning_rate": 0.0005, "loss": 2.072442054748535, "step": 8228 }, { "epoch": 0.6487189594008672, "grad_norm": 0.7413787245750427, "learning_rate": 0.0005, "loss": 2.1216578483581543, "step": 8229 }, { "epoch": 0.6487977926685061, "grad_norm": 0.728662371635437, "learning_rate": 0.0005, "loss": 1.9730281829833984, "step": 8230 }, { "epoch": 0.648876625936145, "grad_norm": 0.8412355184555054, "learning_rate": 0.0005, "loss": 2.0143725872039795, "step": 8231 }, { "epoch": 0.648955459203784, "grad_norm": 0.6081428527832031, "learning_rate": 0.0005, "loss": 1.871425986289978, "step": 8232 }, { "epoch": 0.649034292471423, "grad_norm": 0.8496191501617432, "learning_rate": 0.0005, "loss": 2.095402479171753, "step": 8233 }, { "epoch": 0.6491131257390619, "grad_norm": 0.6442078351974487, "learning_rate": 0.0005, "loss": 1.9240951538085938, "step": 8234 }, { "epoch": 0.6491919590067008, "grad_norm": 0.6663504838943481, "learning_rate": 0.0005, "loss": 1.7855033874511719, "step": 8235 }, { "epoch": 0.6492707922743398, "grad_norm": 0.75440913438797, "learning_rate": 0.0005, "loss": 1.885035514831543, "step": 8236 }, { "epoch": 0.6493496255419787, "grad_norm": 0.5759526491165161, "learning_rate": 0.0005, "loss": 1.8215599060058594, "step": 8237 }, { "epoch": 0.6494284588096176, "grad_norm": 0.7580419182777405, "learning_rate": 0.0005, "loss": 2.366694927215576, "step": 8238 }, { "epoch": 0.6495072920772565, "grad_norm": 0.5640077590942383, "learning_rate": 0.0005, "loss": 2.130509853363037, "step": 8239 }, { "epoch": 0.6495861253448956, "grad_norm": 0.8375074863433838, "learning_rate": 0.0005, "loss": 1.8554885387420654, "step": 8240 }, { "epoch": 0.6496649586125345, "grad_norm": 1.0361690521240234, "learning_rate": 0.0005, "loss": 2.280252456665039, "step": 8241 }, { "epoch": 0.6497437918801734, "grad_norm": 0.9529671669006348, "learning_rate": 0.0005, "loss": 2.198631763458252, "step": 8242 }, { "epoch": 0.6498226251478124, "grad_norm": 0.7060918211936951, "learning_rate": 0.0005, "loss": 1.8941315412521362, "step": 8243 }, { "epoch": 0.6499014584154513, "grad_norm": 1.1930011510849, "learning_rate": 0.0005, "loss": 1.884899616241455, "step": 8244 }, { "epoch": 0.6499802916830902, "grad_norm": 0.5291358828544617, "learning_rate": 0.0005, "loss": 1.935210943222046, "step": 8245 }, { "epoch": 0.6500591249507293, "grad_norm": 0.7010288834571838, "learning_rate": 0.0005, "loss": 1.9528687000274658, "step": 8246 }, { "epoch": 0.6501379582183682, "grad_norm": 0.6832571029663086, "learning_rate": 0.0005, "loss": 1.9291925430297852, "step": 8247 }, { "epoch": 0.6502167914860071, "grad_norm": 0.790170431137085, "learning_rate": 0.0005, "loss": 2.065375328063965, "step": 8248 }, { "epoch": 0.650295624753646, "grad_norm": 0.6316516995429993, "learning_rate": 0.0005, "loss": 2.1152451038360596, "step": 8249 }, { "epoch": 0.650374458021285, "grad_norm": 0.5755611658096313, "learning_rate": 0.0005, "loss": 1.862477421760559, "step": 8250 }, { "epoch": 0.6504532912889239, "grad_norm": 0.8292490839958191, "learning_rate": 0.0005, "loss": 1.9661041498184204, "step": 8251 }, { "epoch": 0.6505321245565628, "grad_norm": 0.9380815029144287, "learning_rate": 0.0005, "loss": 2.021749973297119, "step": 8252 }, { "epoch": 0.6506109578242019, "grad_norm": 0.7655206918716431, "learning_rate": 0.0005, "loss": 2.16951322555542, "step": 8253 }, { "epoch": 0.6506897910918408, "grad_norm": 0.6864801049232483, "learning_rate": 0.0005, "loss": 1.8137034177780151, "step": 8254 }, { "epoch": 0.6507686243594797, "grad_norm": 0.8712167143821716, "learning_rate": 0.0005, "loss": 1.910344123840332, "step": 8255 }, { "epoch": 0.6508474576271186, "grad_norm": 0.5831003785133362, "learning_rate": 0.0005, "loss": 2.029073715209961, "step": 8256 }, { "epoch": 0.6509262908947576, "grad_norm": 0.6952813863754272, "learning_rate": 0.0005, "loss": 1.992331624031067, "step": 8257 }, { "epoch": 0.6510051241623965, "grad_norm": 0.8667104840278625, "learning_rate": 0.0005, "loss": 1.8807904720306396, "step": 8258 }, { "epoch": 0.6510839574300354, "grad_norm": 0.6697434782981873, "learning_rate": 0.0005, "loss": 2.1959590911865234, "step": 8259 }, { "epoch": 0.6511627906976745, "grad_norm": 0.7437328100204468, "learning_rate": 0.0005, "loss": 1.9375455379486084, "step": 8260 }, { "epoch": 0.6512416239653134, "grad_norm": 0.5294200778007507, "learning_rate": 0.0005, "loss": 1.8393728733062744, "step": 8261 }, { "epoch": 0.6513204572329523, "grad_norm": 0.6296746134757996, "learning_rate": 0.0005, "loss": 1.9873366355895996, "step": 8262 }, { "epoch": 0.6513992905005912, "grad_norm": 0.6365728974342346, "learning_rate": 0.0005, "loss": 1.7504723072052002, "step": 8263 }, { "epoch": 0.6514781237682302, "grad_norm": 0.8221620321273804, "learning_rate": 0.0005, "loss": 1.62754225730896, "step": 8264 }, { "epoch": 0.6515569570358691, "grad_norm": 1.436550498008728, "learning_rate": 0.0005, "loss": 1.9766559600830078, "step": 8265 }, { "epoch": 0.651635790303508, "grad_norm": 0.8363021612167358, "learning_rate": 0.0005, "loss": 2.252194881439209, "step": 8266 }, { "epoch": 0.6517146235711471, "grad_norm": 0.976979672908783, "learning_rate": 0.0005, "loss": 2.245389461517334, "step": 8267 }, { "epoch": 0.651793456838786, "grad_norm": 0.8197925686836243, "learning_rate": 0.0005, "loss": 1.9888056516647339, "step": 8268 }, { "epoch": 0.6518722901064249, "grad_norm": 1.445988655090332, "learning_rate": 0.0005, "loss": 2.128049850463867, "step": 8269 }, { "epoch": 0.6519511233740638, "grad_norm": 0.8900904655456543, "learning_rate": 0.0005, "loss": 2.091015338897705, "step": 8270 }, { "epoch": 0.6520299566417028, "grad_norm": 0.7276207208633423, "learning_rate": 0.0005, "loss": 1.9294663667678833, "step": 8271 }, { "epoch": 0.6521087899093417, "grad_norm": 0.7485138177871704, "learning_rate": 0.0005, "loss": 2.154442310333252, "step": 8272 }, { "epoch": 0.6521876231769806, "grad_norm": 0.9634320735931396, "learning_rate": 0.0005, "loss": 2.138162136077881, "step": 8273 }, { "epoch": 0.6522664564446197, "grad_norm": 0.6452769041061401, "learning_rate": 0.0005, "loss": 1.9517183303833008, "step": 8274 }, { "epoch": 0.6523452897122586, "grad_norm": 0.7637542486190796, "learning_rate": 0.0005, "loss": 2.213101387023926, "step": 8275 }, { "epoch": 0.6524241229798975, "grad_norm": 0.8095252513885498, "learning_rate": 0.0005, "loss": 1.9715819358825684, "step": 8276 }, { "epoch": 0.6525029562475365, "grad_norm": 1.0079110860824585, "learning_rate": 0.0005, "loss": 2.066330909729004, "step": 8277 }, { "epoch": 0.6525817895151754, "grad_norm": 0.6422763466835022, "learning_rate": 0.0005, "loss": 1.7293052673339844, "step": 8278 }, { "epoch": 0.6526606227828143, "grad_norm": 0.8802218437194824, "learning_rate": 0.0005, "loss": 1.9852383136749268, "step": 8279 }, { "epoch": 0.6527394560504532, "grad_norm": 1.0456626415252686, "learning_rate": 0.0005, "loss": 1.9614651203155518, "step": 8280 }, { "epoch": 0.6528182893180923, "grad_norm": 0.8427013158798218, "learning_rate": 0.0005, "loss": 2.069493055343628, "step": 8281 }, { "epoch": 0.6528971225857312, "grad_norm": 0.8287612199783325, "learning_rate": 0.0005, "loss": 2.0301992893218994, "step": 8282 }, { "epoch": 0.6529759558533701, "grad_norm": 0.6930299401283264, "learning_rate": 0.0005, "loss": 1.7652409076690674, "step": 8283 }, { "epoch": 0.6530547891210091, "grad_norm": 0.995013415813446, "learning_rate": 0.0005, "loss": 1.7764906883239746, "step": 8284 }, { "epoch": 0.653133622388648, "grad_norm": 0.6656351089477539, "learning_rate": 0.0005, "loss": 2.1566848754882812, "step": 8285 }, { "epoch": 0.6532124556562869, "grad_norm": 0.7733866572380066, "learning_rate": 0.0005, "loss": 1.9491236209869385, "step": 8286 }, { "epoch": 0.6532912889239259, "grad_norm": 0.6987186074256897, "learning_rate": 0.0005, "loss": 1.804183006286621, "step": 8287 }, { "epoch": 0.6533701221915649, "grad_norm": 0.5843937993049622, "learning_rate": 0.0005, "loss": 1.7043598890304565, "step": 8288 }, { "epoch": 0.6534489554592038, "grad_norm": 0.5888324975967407, "learning_rate": 0.0005, "loss": 1.7372074127197266, "step": 8289 }, { "epoch": 0.6535277887268427, "grad_norm": 0.5797258615493774, "learning_rate": 0.0005, "loss": 1.9192979335784912, "step": 8290 }, { "epoch": 0.6536066219944817, "grad_norm": 0.9300929307937622, "learning_rate": 0.0005, "loss": 1.9738590717315674, "step": 8291 }, { "epoch": 0.6536854552621206, "grad_norm": 0.7230942249298096, "learning_rate": 0.0005, "loss": 2.128136396408081, "step": 8292 }, { "epoch": 0.6537642885297595, "grad_norm": 0.8685794472694397, "learning_rate": 0.0005, "loss": 1.8771507740020752, "step": 8293 }, { "epoch": 0.6538431217973985, "grad_norm": 0.7885186672210693, "learning_rate": 0.0005, "loss": 1.891607403755188, "step": 8294 }, { "epoch": 0.6539219550650375, "grad_norm": 0.8685685396194458, "learning_rate": 0.0005, "loss": 1.9540040493011475, "step": 8295 }, { "epoch": 0.6540007883326764, "grad_norm": 0.5722512006759644, "learning_rate": 0.0005, "loss": 1.7355575561523438, "step": 8296 }, { "epoch": 0.6540796216003153, "grad_norm": 0.6670389175415039, "learning_rate": 0.0005, "loss": 1.9868876934051514, "step": 8297 }, { "epoch": 0.6541584548679543, "grad_norm": 0.8825467228889465, "learning_rate": 0.0005, "loss": 2.275831699371338, "step": 8298 }, { "epoch": 0.6542372881355932, "grad_norm": 0.7060750722885132, "learning_rate": 0.0005, "loss": 2.2640762329101562, "step": 8299 }, { "epoch": 0.6543161214032321, "grad_norm": 0.6476631760597229, "learning_rate": 0.0005, "loss": 1.8764758110046387, "step": 8300 }, { "epoch": 0.654394954670871, "grad_norm": 0.6679727435112, "learning_rate": 0.0005, "loss": 1.9309308528900146, "step": 8301 }, { "epoch": 0.6544737879385101, "grad_norm": 0.7417203783988953, "learning_rate": 0.0005, "loss": 2.2359485626220703, "step": 8302 }, { "epoch": 0.654552621206149, "grad_norm": 0.6621909141540527, "learning_rate": 0.0005, "loss": 1.742078185081482, "step": 8303 }, { "epoch": 0.6546314544737879, "grad_norm": 0.6400102376937866, "learning_rate": 0.0005, "loss": 1.9688425064086914, "step": 8304 }, { "epoch": 0.6547102877414269, "grad_norm": 0.566774845123291, "learning_rate": 0.0005, "loss": 2.029157876968384, "step": 8305 }, { "epoch": 0.6547891210090658, "grad_norm": 0.7729645371437073, "learning_rate": 0.0005, "loss": 1.833348035812378, "step": 8306 }, { "epoch": 0.6548679542767047, "grad_norm": 0.7125519514083862, "learning_rate": 0.0005, "loss": 2.23317813873291, "step": 8307 }, { "epoch": 0.6549467875443438, "grad_norm": 0.6098889708518982, "learning_rate": 0.0005, "loss": 1.9080471992492676, "step": 8308 }, { "epoch": 0.6550256208119827, "grad_norm": 0.6698272228240967, "learning_rate": 0.0005, "loss": 1.84706711769104, "step": 8309 }, { "epoch": 0.6551044540796216, "grad_norm": 0.8639217615127563, "learning_rate": 0.0005, "loss": 1.902500867843628, "step": 8310 }, { "epoch": 0.6551832873472605, "grad_norm": 1.0192015171051025, "learning_rate": 0.0005, "loss": 2.0857205390930176, "step": 8311 }, { "epoch": 0.6552621206148995, "grad_norm": 0.8356777429580688, "learning_rate": 0.0005, "loss": 1.999332308769226, "step": 8312 }, { "epoch": 0.6553409538825384, "grad_norm": 0.6609906554222107, "learning_rate": 0.0005, "loss": 2.004539966583252, "step": 8313 }, { "epoch": 0.6554197871501773, "grad_norm": 0.8097979426383972, "learning_rate": 0.0005, "loss": 1.9812073707580566, "step": 8314 }, { "epoch": 0.6554986204178164, "grad_norm": 0.8831878304481506, "learning_rate": 0.0005, "loss": 2.0471391677856445, "step": 8315 }, { "epoch": 0.6555774536854553, "grad_norm": 1.0566693544387817, "learning_rate": 0.0005, "loss": 2.401332378387451, "step": 8316 }, { "epoch": 0.6556562869530942, "grad_norm": 0.6009207963943481, "learning_rate": 0.0005, "loss": 1.8493576049804688, "step": 8317 }, { "epoch": 0.6557351202207331, "grad_norm": 0.6978276371955872, "learning_rate": 0.0005, "loss": 2.021300792694092, "step": 8318 }, { "epoch": 0.6558139534883721, "grad_norm": 0.8375896215438843, "learning_rate": 0.0005, "loss": 2.0510685443878174, "step": 8319 }, { "epoch": 0.655892786756011, "grad_norm": 0.8128892779350281, "learning_rate": 0.0005, "loss": 2.05705189704895, "step": 8320 }, { "epoch": 0.65597162002365, "grad_norm": 0.6401622295379639, "learning_rate": 0.0005, "loss": 1.8326420783996582, "step": 8321 }, { "epoch": 0.656050453291289, "grad_norm": 0.5673272609710693, "learning_rate": 0.0005, "loss": 1.707188367843628, "step": 8322 }, { "epoch": 0.6561292865589279, "grad_norm": 0.7816467881202698, "learning_rate": 0.0005, "loss": 2.5717813968658447, "step": 8323 }, { "epoch": 0.6562081198265668, "grad_norm": 0.5706167221069336, "learning_rate": 0.0005, "loss": 1.7955231666564941, "step": 8324 }, { "epoch": 0.6562869530942057, "grad_norm": 0.8508036732673645, "learning_rate": 0.0005, "loss": 1.9120900630950928, "step": 8325 }, { "epoch": 0.6563657863618447, "grad_norm": 0.7848405838012695, "learning_rate": 0.0005, "loss": 1.9166852235794067, "step": 8326 }, { "epoch": 0.6564446196294836, "grad_norm": 0.7356013059616089, "learning_rate": 0.0005, "loss": 2.051516056060791, "step": 8327 }, { "epoch": 0.6565234528971226, "grad_norm": 0.7354854941368103, "learning_rate": 0.0005, "loss": 2.106327533721924, "step": 8328 }, { "epoch": 0.6566022861647616, "grad_norm": 0.8648977279663086, "learning_rate": 0.0005, "loss": 2.272918224334717, "step": 8329 }, { "epoch": 0.6566811194324005, "grad_norm": 0.733055055141449, "learning_rate": 0.0005, "loss": 1.9229673147201538, "step": 8330 }, { "epoch": 0.6567599527000394, "grad_norm": 0.6257708072662354, "learning_rate": 0.0005, "loss": 1.9225925207138062, "step": 8331 }, { "epoch": 0.6568387859676783, "grad_norm": 0.6645185947418213, "learning_rate": 0.0005, "loss": 2.295177936553955, "step": 8332 }, { "epoch": 0.6569176192353173, "grad_norm": 0.8062751293182373, "learning_rate": 0.0005, "loss": 2.127068281173706, "step": 8333 }, { "epoch": 0.6569964525029562, "grad_norm": 0.5366919040679932, "learning_rate": 0.0005, "loss": 1.493201494216919, "step": 8334 }, { "epoch": 0.6570752857705952, "grad_norm": 0.5926831364631653, "learning_rate": 0.0005, "loss": 2.0180296897888184, "step": 8335 }, { "epoch": 0.6571541190382342, "grad_norm": 2.0064010620117188, "learning_rate": 0.0005, "loss": 2.004284381866455, "step": 8336 }, { "epoch": 0.6572329523058731, "grad_norm": 0.7903807759284973, "learning_rate": 0.0005, "loss": 1.9059727191925049, "step": 8337 }, { "epoch": 0.657311785573512, "grad_norm": 0.7845794558525085, "learning_rate": 0.0005, "loss": 1.9302566051483154, "step": 8338 }, { "epoch": 0.657390618841151, "grad_norm": 1.7909456491470337, "learning_rate": 0.0005, "loss": 2.250176429748535, "step": 8339 }, { "epoch": 0.6574694521087899, "grad_norm": 0.7512203454971313, "learning_rate": 0.0005, "loss": 1.9401938915252686, "step": 8340 }, { "epoch": 0.6575482853764288, "grad_norm": 0.7169385552406311, "learning_rate": 0.0005, "loss": 1.8779456615447998, "step": 8341 }, { "epoch": 0.6576271186440678, "grad_norm": 0.7638134360313416, "learning_rate": 0.0005, "loss": 1.9358891248703003, "step": 8342 }, { "epoch": 0.6577059519117068, "grad_norm": 1.2214959859848022, "learning_rate": 0.0005, "loss": 2.186026096343994, "step": 8343 }, { "epoch": 0.6577847851793457, "grad_norm": 0.7422903180122375, "learning_rate": 0.0005, "loss": 2.0235936641693115, "step": 8344 }, { "epoch": 0.6578636184469846, "grad_norm": 0.6453390717506409, "learning_rate": 0.0005, "loss": 1.8841265439987183, "step": 8345 }, { "epoch": 0.6579424517146236, "grad_norm": 0.7855357527732849, "learning_rate": 0.0005, "loss": 2.032907485961914, "step": 8346 }, { "epoch": 0.6580212849822625, "grad_norm": 0.6494960784912109, "learning_rate": 0.0005, "loss": 1.8651089668273926, "step": 8347 }, { "epoch": 0.6581001182499014, "grad_norm": 0.5913901329040527, "learning_rate": 0.0005, "loss": 2.242647171020508, "step": 8348 }, { "epoch": 0.6581789515175404, "grad_norm": 0.9371767044067383, "learning_rate": 0.0005, "loss": 2.243602752685547, "step": 8349 }, { "epoch": 0.6582577847851794, "grad_norm": 0.7448204159736633, "learning_rate": 0.0005, "loss": 1.8680270910263062, "step": 8350 }, { "epoch": 0.6583366180528183, "grad_norm": 0.656268298625946, "learning_rate": 0.0005, "loss": 1.8279457092285156, "step": 8351 }, { "epoch": 0.6584154513204572, "grad_norm": 0.6687211394309998, "learning_rate": 0.0005, "loss": 1.6182234287261963, "step": 8352 }, { "epoch": 0.6584942845880962, "grad_norm": 0.6783250570297241, "learning_rate": 0.0005, "loss": 2.0297698974609375, "step": 8353 }, { "epoch": 0.6585731178557351, "grad_norm": 0.6900233626365662, "learning_rate": 0.0005, "loss": 1.9922857284545898, "step": 8354 }, { "epoch": 0.658651951123374, "grad_norm": 0.6358187198638916, "learning_rate": 0.0005, "loss": 2.0388054847717285, "step": 8355 }, { "epoch": 0.658730784391013, "grad_norm": 0.7534788846969604, "learning_rate": 0.0005, "loss": 1.9587914943695068, "step": 8356 }, { "epoch": 0.658809617658652, "grad_norm": 1.073050618171692, "learning_rate": 0.0005, "loss": 2.2332539558410645, "step": 8357 }, { "epoch": 0.6588884509262909, "grad_norm": 0.5988805294036865, "learning_rate": 0.0005, "loss": 1.6284043788909912, "step": 8358 }, { "epoch": 0.6589672841939298, "grad_norm": 0.7440950274467468, "learning_rate": 0.0005, "loss": 2.1374685764312744, "step": 8359 }, { "epoch": 0.6590461174615688, "grad_norm": 0.6883450150489807, "learning_rate": 0.0005, "loss": 1.91408109664917, "step": 8360 }, { "epoch": 0.6591249507292077, "grad_norm": 0.8356339931488037, "learning_rate": 0.0005, "loss": 2.056298017501831, "step": 8361 }, { "epoch": 0.6592037839968466, "grad_norm": 0.9650540351867676, "learning_rate": 0.0005, "loss": 2.2323994636535645, "step": 8362 }, { "epoch": 0.6592826172644856, "grad_norm": 0.8994888663291931, "learning_rate": 0.0005, "loss": 1.9625120162963867, "step": 8363 }, { "epoch": 0.6593614505321246, "grad_norm": 0.7060121297836304, "learning_rate": 0.0005, "loss": 2.0065293312072754, "step": 8364 }, { "epoch": 0.6594402837997635, "grad_norm": 1.9948135614395142, "learning_rate": 0.0005, "loss": 2.022599935531616, "step": 8365 }, { "epoch": 0.6595191170674024, "grad_norm": 0.5586662888526917, "learning_rate": 0.0005, "loss": 1.9368488788604736, "step": 8366 }, { "epoch": 0.6595979503350414, "grad_norm": 0.606752872467041, "learning_rate": 0.0005, "loss": 2.026315689086914, "step": 8367 }, { "epoch": 0.6596767836026803, "grad_norm": 0.6775895953178406, "learning_rate": 0.0005, "loss": 1.6964024305343628, "step": 8368 }, { "epoch": 0.6597556168703193, "grad_norm": 0.6697570085525513, "learning_rate": 0.0005, "loss": 2.0794267654418945, "step": 8369 }, { "epoch": 0.6598344501379583, "grad_norm": 0.6470810174942017, "learning_rate": 0.0005, "loss": 1.7267032861709595, "step": 8370 }, { "epoch": 0.6599132834055972, "grad_norm": 0.6504761576652527, "learning_rate": 0.0005, "loss": 2.006063461303711, "step": 8371 }, { "epoch": 0.6599921166732361, "grad_norm": 0.8002340793609619, "learning_rate": 0.0005, "loss": 1.722888469696045, "step": 8372 }, { "epoch": 0.660070949940875, "grad_norm": 0.7512356042861938, "learning_rate": 0.0005, "loss": 1.8215394020080566, "step": 8373 }, { "epoch": 0.660149783208514, "grad_norm": 0.9166261553764343, "learning_rate": 0.0005, "loss": 1.960858702659607, "step": 8374 }, { "epoch": 0.6602286164761529, "grad_norm": 1.0455337762832642, "learning_rate": 0.0005, "loss": 1.8280220031738281, "step": 8375 }, { "epoch": 0.6603074497437919, "grad_norm": 1.3316620588302612, "learning_rate": 0.0005, "loss": 2.1686477661132812, "step": 8376 }, { "epoch": 0.6603862830114309, "grad_norm": 0.7653486132621765, "learning_rate": 0.0005, "loss": 2.0417280197143555, "step": 8377 }, { "epoch": 0.6604651162790698, "grad_norm": 0.663846492767334, "learning_rate": 0.0005, "loss": 2.085932731628418, "step": 8378 }, { "epoch": 0.6605439495467087, "grad_norm": 0.6814785599708557, "learning_rate": 0.0005, "loss": 1.792227029800415, "step": 8379 }, { "epoch": 0.6606227828143476, "grad_norm": 0.6420809030532837, "learning_rate": 0.0005, "loss": 2.0443921089172363, "step": 8380 }, { "epoch": 0.6607016160819866, "grad_norm": 0.8612914681434631, "learning_rate": 0.0005, "loss": 1.9590001106262207, "step": 8381 }, { "epoch": 0.6607804493496255, "grad_norm": 0.6612874269485474, "learning_rate": 0.0005, "loss": 1.8544917106628418, "step": 8382 }, { "epoch": 0.6608592826172645, "grad_norm": 0.5908232927322388, "learning_rate": 0.0005, "loss": 1.5572426319122314, "step": 8383 }, { "epoch": 0.6609381158849035, "grad_norm": 0.686673104763031, "learning_rate": 0.0005, "loss": 2.0145316123962402, "step": 8384 }, { "epoch": 0.6610169491525424, "grad_norm": 0.7149165272712708, "learning_rate": 0.0005, "loss": 1.9309654235839844, "step": 8385 }, { "epoch": 0.6610957824201813, "grad_norm": 0.7754611372947693, "learning_rate": 0.0005, "loss": 2.075455665588379, "step": 8386 }, { "epoch": 0.6611746156878202, "grad_norm": 0.7609901428222656, "learning_rate": 0.0005, "loss": 1.837531328201294, "step": 8387 }, { "epoch": 0.6612534489554592, "grad_norm": 0.6894577741622925, "learning_rate": 0.0005, "loss": 1.85026216506958, "step": 8388 }, { "epoch": 0.6613322822230981, "grad_norm": 0.7508984804153442, "learning_rate": 0.0005, "loss": 1.8694097995758057, "step": 8389 }, { "epoch": 0.6614111154907371, "grad_norm": 0.7833923101425171, "learning_rate": 0.0005, "loss": 1.9198722839355469, "step": 8390 }, { "epoch": 0.6614899487583761, "grad_norm": 0.7816600799560547, "learning_rate": 0.0005, "loss": 1.9310007095336914, "step": 8391 }, { "epoch": 0.661568782026015, "grad_norm": 0.9235304594039917, "learning_rate": 0.0005, "loss": 1.918362021446228, "step": 8392 }, { "epoch": 0.6616476152936539, "grad_norm": 0.6505020260810852, "learning_rate": 0.0005, "loss": 1.949289321899414, "step": 8393 }, { "epoch": 0.6617264485612928, "grad_norm": 0.7560827136039734, "learning_rate": 0.0005, "loss": 2.234829902648926, "step": 8394 }, { "epoch": 0.6618052818289318, "grad_norm": 0.7250648736953735, "learning_rate": 0.0005, "loss": 2.172102689743042, "step": 8395 }, { "epoch": 0.6618841150965707, "grad_norm": 0.9032857418060303, "learning_rate": 0.0005, "loss": 1.8769867420196533, "step": 8396 }, { "epoch": 0.6619629483642097, "grad_norm": 0.8605602979660034, "learning_rate": 0.0005, "loss": 1.8862627744674683, "step": 8397 }, { "epoch": 0.6620417816318487, "grad_norm": 0.7177006602287292, "learning_rate": 0.0005, "loss": 1.923527717590332, "step": 8398 }, { "epoch": 0.6621206148994876, "grad_norm": 0.9196759462356567, "learning_rate": 0.0005, "loss": 1.96419095993042, "step": 8399 }, { "epoch": 0.6621994481671265, "grad_norm": 0.7546095252037048, "learning_rate": 0.0005, "loss": 2.098249912261963, "step": 8400 }, { "epoch": 0.6622782814347655, "grad_norm": 0.6561224460601807, "learning_rate": 0.0005, "loss": 2.017982244491577, "step": 8401 }, { "epoch": 0.6623571147024044, "grad_norm": 0.5938795804977417, "learning_rate": 0.0005, "loss": 2.045365810394287, "step": 8402 }, { "epoch": 0.6624359479700433, "grad_norm": 0.7414180636405945, "learning_rate": 0.0005, "loss": 1.9357092380523682, "step": 8403 }, { "epoch": 0.6625147812376823, "grad_norm": 0.8551433086395264, "learning_rate": 0.0005, "loss": 1.885655403137207, "step": 8404 }, { "epoch": 0.6625936145053213, "grad_norm": 0.6857073307037354, "learning_rate": 0.0005, "loss": 2.0712714195251465, "step": 8405 }, { "epoch": 0.6626724477729602, "grad_norm": 0.6788223385810852, "learning_rate": 0.0005, "loss": 1.8681495189666748, "step": 8406 }, { "epoch": 0.6627512810405991, "grad_norm": 1.1118183135986328, "learning_rate": 0.0005, "loss": 2.1027188301086426, "step": 8407 }, { "epoch": 0.6628301143082381, "grad_norm": 0.674104630947113, "learning_rate": 0.0005, "loss": 1.8276787996292114, "step": 8408 }, { "epoch": 0.662908947575877, "grad_norm": 1.082872986793518, "learning_rate": 0.0005, "loss": 1.9043093919754028, "step": 8409 }, { "epoch": 0.662987780843516, "grad_norm": 0.6807162761688232, "learning_rate": 0.0005, "loss": 1.869584083557129, "step": 8410 }, { "epoch": 0.6630666141111549, "grad_norm": 0.5228207111358643, "learning_rate": 0.0005, "loss": 1.808721899986267, "step": 8411 }, { "epoch": 0.6631454473787939, "grad_norm": 0.6201748847961426, "learning_rate": 0.0005, "loss": 1.770728588104248, "step": 8412 }, { "epoch": 0.6632242806464328, "grad_norm": 0.7431924343109131, "learning_rate": 0.0005, "loss": 1.9600284099578857, "step": 8413 }, { "epoch": 0.6633031139140717, "grad_norm": 0.8415871262550354, "learning_rate": 0.0005, "loss": 2.0542211532592773, "step": 8414 }, { "epoch": 0.6633819471817107, "grad_norm": 0.621480405330658, "learning_rate": 0.0005, "loss": 1.9755065441131592, "step": 8415 }, { "epoch": 0.6634607804493496, "grad_norm": 0.6512006521224976, "learning_rate": 0.0005, "loss": 1.8976693153381348, "step": 8416 }, { "epoch": 0.6635396137169886, "grad_norm": 0.8449742197990417, "learning_rate": 0.0005, "loss": 1.9674479961395264, "step": 8417 }, { "epoch": 0.6636184469846275, "grad_norm": 1.1711523532867432, "learning_rate": 0.0005, "loss": 2.2610297203063965, "step": 8418 }, { "epoch": 0.6636972802522665, "grad_norm": 0.8569939136505127, "learning_rate": 0.0005, "loss": 1.9376225471496582, "step": 8419 }, { "epoch": 0.6637761135199054, "grad_norm": 1.599104881286621, "learning_rate": 0.0005, "loss": 2.0641098022460938, "step": 8420 }, { "epoch": 0.6638549467875443, "grad_norm": 0.5985324382781982, "learning_rate": 0.0005, "loss": 1.9527151584625244, "step": 8421 }, { "epoch": 0.6639337800551833, "grad_norm": 0.6873892545700073, "learning_rate": 0.0005, "loss": 1.8939454555511475, "step": 8422 }, { "epoch": 0.6640126133228222, "grad_norm": 0.6855416297912598, "learning_rate": 0.0005, "loss": 1.9234607219696045, "step": 8423 }, { "epoch": 0.6640914465904612, "grad_norm": 0.7130889892578125, "learning_rate": 0.0005, "loss": 1.7369540929794312, "step": 8424 }, { "epoch": 0.6641702798581001, "grad_norm": 0.6933546662330627, "learning_rate": 0.0005, "loss": 2.044827938079834, "step": 8425 }, { "epoch": 0.6642491131257391, "grad_norm": 0.6229465007781982, "learning_rate": 0.0005, "loss": 1.5830270051956177, "step": 8426 }, { "epoch": 0.664327946393378, "grad_norm": 0.6218013763427734, "learning_rate": 0.0005, "loss": 1.7785327434539795, "step": 8427 }, { "epoch": 0.6644067796610169, "grad_norm": 0.731010377407074, "learning_rate": 0.0005, "loss": 2.044260263442993, "step": 8428 }, { "epoch": 0.6644856129286559, "grad_norm": 0.6386950016021729, "learning_rate": 0.0005, "loss": 1.9926948547363281, "step": 8429 }, { "epoch": 0.6645644461962948, "grad_norm": 0.6818119883537292, "learning_rate": 0.0005, "loss": 1.7584048509597778, "step": 8430 }, { "epoch": 0.6646432794639338, "grad_norm": 0.6881605386734009, "learning_rate": 0.0005, "loss": 2.1367928981781006, "step": 8431 }, { "epoch": 0.6647221127315728, "grad_norm": 0.6804190278053284, "learning_rate": 0.0005, "loss": 2.3622541427612305, "step": 8432 }, { "epoch": 0.6648009459992117, "grad_norm": 1.3251328468322754, "learning_rate": 0.0005, "loss": 1.8677681684494019, "step": 8433 }, { "epoch": 0.6648797792668506, "grad_norm": 11.193155288696289, "learning_rate": 0.0005, "loss": 2.083178758621216, "step": 8434 }, { "epoch": 0.6649586125344895, "grad_norm": 0.7240884900093079, "learning_rate": 0.0005, "loss": 2.1547493934631348, "step": 8435 }, { "epoch": 0.6650374458021285, "grad_norm": 0.6842703819274902, "learning_rate": 0.0005, "loss": 1.7068257331848145, "step": 8436 }, { "epoch": 0.6651162790697674, "grad_norm": 0.6983543634414673, "learning_rate": 0.0005, "loss": 2.334871292114258, "step": 8437 }, { "epoch": 0.6651951123374064, "grad_norm": 0.6484246253967285, "learning_rate": 0.0005, "loss": 2.017402172088623, "step": 8438 }, { "epoch": 0.6652739456050454, "grad_norm": 0.7786164879798889, "learning_rate": 0.0005, "loss": 2.1695122718811035, "step": 8439 }, { "epoch": 0.6653527788726843, "grad_norm": 0.9178446531295776, "learning_rate": 0.0005, "loss": 1.8987414836883545, "step": 8440 }, { "epoch": 0.6654316121403232, "grad_norm": 0.7595052719116211, "learning_rate": 0.0005, "loss": 2.308894634246826, "step": 8441 }, { "epoch": 0.6655104454079621, "grad_norm": 0.685346245765686, "learning_rate": 0.0005, "loss": 1.737722635269165, "step": 8442 }, { "epoch": 0.6655892786756011, "grad_norm": 0.8403093814849854, "learning_rate": 0.0005, "loss": 1.7461957931518555, "step": 8443 }, { "epoch": 0.66566811194324, "grad_norm": 0.6143761277198792, "learning_rate": 0.0005, "loss": 2.0676279067993164, "step": 8444 }, { "epoch": 0.665746945210879, "grad_norm": 0.5368073582649231, "learning_rate": 0.0005, "loss": 1.7567710876464844, "step": 8445 }, { "epoch": 0.665825778478518, "grad_norm": 0.7366176843643188, "learning_rate": 0.0005, "loss": 1.9426908493041992, "step": 8446 }, { "epoch": 0.6659046117461569, "grad_norm": 0.7150968909263611, "learning_rate": 0.0005, "loss": 1.9589080810546875, "step": 8447 }, { "epoch": 0.6659834450137958, "grad_norm": 0.6442666053771973, "learning_rate": 0.0005, "loss": 1.712874174118042, "step": 8448 }, { "epoch": 0.6660622782814347, "grad_norm": 0.5816052556037903, "learning_rate": 0.0005, "loss": 1.8729368448257446, "step": 8449 }, { "epoch": 0.6661411115490737, "grad_norm": 0.7065220475196838, "learning_rate": 0.0005, "loss": 1.7286851406097412, "step": 8450 }, { "epoch": 0.6662199448167127, "grad_norm": 0.8048073649406433, "learning_rate": 0.0005, "loss": 1.9178917407989502, "step": 8451 }, { "epoch": 0.6662987780843516, "grad_norm": 0.6196715831756592, "learning_rate": 0.0005, "loss": 2.0928897857666016, "step": 8452 }, { "epoch": 0.6663776113519906, "grad_norm": 0.7654058933258057, "learning_rate": 0.0005, "loss": 1.7824888229370117, "step": 8453 }, { "epoch": 0.6664564446196295, "grad_norm": 0.7507015466690063, "learning_rate": 0.0005, "loss": 2.093438148498535, "step": 8454 }, { "epoch": 0.6665352778872684, "grad_norm": 0.8421732783317566, "learning_rate": 0.0005, "loss": 1.9466091394424438, "step": 8455 }, { "epoch": 0.6666141111549073, "grad_norm": 0.6114482879638672, "learning_rate": 0.0005, "loss": 1.8466655015945435, "step": 8456 }, { "epoch": 0.6666929444225463, "grad_norm": 0.709430456161499, "learning_rate": 0.0005, "loss": 1.9382799863815308, "step": 8457 }, { "epoch": 0.6667717776901853, "grad_norm": 0.6340858936309814, "learning_rate": 0.0005, "loss": 1.9290779829025269, "step": 8458 }, { "epoch": 0.6668506109578242, "grad_norm": 0.6860564351081848, "learning_rate": 0.0005, "loss": 1.9184459447860718, "step": 8459 }, { "epoch": 0.6669294442254632, "grad_norm": 0.8149225115776062, "learning_rate": 0.0005, "loss": 2.1056456565856934, "step": 8460 }, { "epoch": 0.6670082774931021, "grad_norm": 0.579887866973877, "learning_rate": 0.0005, "loss": 1.835756778717041, "step": 8461 }, { "epoch": 0.667087110760741, "grad_norm": 0.7597453594207764, "learning_rate": 0.0005, "loss": 1.986711025238037, "step": 8462 }, { "epoch": 0.66716594402838, "grad_norm": 0.8508546948432922, "learning_rate": 0.0005, "loss": 1.7257722616195679, "step": 8463 }, { "epoch": 0.6672447772960189, "grad_norm": 0.7118097543716431, "learning_rate": 0.0005, "loss": 1.8747340440750122, "step": 8464 }, { "epoch": 0.6673236105636579, "grad_norm": 0.596816897392273, "learning_rate": 0.0005, "loss": 1.6532249450683594, "step": 8465 }, { "epoch": 0.6674024438312968, "grad_norm": 0.5924704670906067, "learning_rate": 0.0005, "loss": 1.9188084602355957, "step": 8466 }, { "epoch": 0.6674812770989358, "grad_norm": 0.5363520979881287, "learning_rate": 0.0005, "loss": 1.7373836040496826, "step": 8467 }, { "epoch": 0.6675601103665747, "grad_norm": 1.891788363456726, "learning_rate": 0.0005, "loss": 2.2174744606018066, "step": 8468 }, { "epoch": 0.6676389436342136, "grad_norm": 0.8523077368736267, "learning_rate": 0.0005, "loss": 2.2650444507598877, "step": 8469 }, { "epoch": 0.6677177769018526, "grad_norm": 0.6456308960914612, "learning_rate": 0.0005, "loss": 1.7802393436431885, "step": 8470 }, { "epoch": 0.6677966101694915, "grad_norm": 0.6995793581008911, "learning_rate": 0.0005, "loss": 1.7170536518096924, "step": 8471 }, { "epoch": 0.6678754434371305, "grad_norm": 0.6345504522323608, "learning_rate": 0.0005, "loss": 1.8157575130462646, "step": 8472 }, { "epoch": 0.6679542767047694, "grad_norm": 0.7956648468971252, "learning_rate": 0.0005, "loss": 1.9291024208068848, "step": 8473 }, { "epoch": 0.6680331099724084, "grad_norm": 0.7659640312194824, "learning_rate": 0.0005, "loss": 1.763246774673462, "step": 8474 }, { "epoch": 0.6681119432400473, "grad_norm": 0.6802865266799927, "learning_rate": 0.0005, "loss": 2.1358237266540527, "step": 8475 }, { "epoch": 0.6681907765076862, "grad_norm": 0.6504387855529785, "learning_rate": 0.0005, "loss": 2.2422690391540527, "step": 8476 }, { "epoch": 0.6682696097753252, "grad_norm": 0.9479255080223083, "learning_rate": 0.0005, "loss": 2.0830585956573486, "step": 8477 }, { "epoch": 0.6683484430429641, "grad_norm": 0.7805399894714355, "learning_rate": 0.0005, "loss": 2.162569046020508, "step": 8478 }, { "epoch": 0.6684272763106031, "grad_norm": 0.6967340707778931, "learning_rate": 0.0005, "loss": 1.9543181657791138, "step": 8479 }, { "epoch": 0.668506109578242, "grad_norm": 0.6569582223892212, "learning_rate": 0.0005, "loss": 1.943098545074463, "step": 8480 }, { "epoch": 0.668584942845881, "grad_norm": 0.7351756691932678, "learning_rate": 0.0005, "loss": 1.9628024101257324, "step": 8481 }, { "epoch": 0.6686637761135199, "grad_norm": 0.6938802599906921, "learning_rate": 0.0005, "loss": 1.8963429927825928, "step": 8482 }, { "epoch": 0.6687426093811588, "grad_norm": 0.71184241771698, "learning_rate": 0.0005, "loss": 1.8543057441711426, "step": 8483 }, { "epoch": 0.6688214426487978, "grad_norm": 0.6666020750999451, "learning_rate": 0.0005, "loss": 2.0674006938934326, "step": 8484 }, { "epoch": 0.6689002759164367, "grad_norm": 0.6547889113426208, "learning_rate": 0.0005, "loss": 1.8215402364730835, "step": 8485 }, { "epoch": 0.6689791091840757, "grad_norm": 0.7246688008308411, "learning_rate": 0.0005, "loss": 1.9578806161880493, "step": 8486 }, { "epoch": 0.6690579424517146, "grad_norm": 0.6829432249069214, "learning_rate": 0.0005, "loss": 2.145852565765381, "step": 8487 }, { "epoch": 0.6691367757193536, "grad_norm": 0.7196440100669861, "learning_rate": 0.0005, "loss": 2.0091772079467773, "step": 8488 }, { "epoch": 0.6692156089869925, "grad_norm": 0.6715716123580933, "learning_rate": 0.0005, "loss": 2.0714573860168457, "step": 8489 }, { "epoch": 0.6692944422546314, "grad_norm": 0.6391919255256653, "learning_rate": 0.0005, "loss": 1.804256796836853, "step": 8490 }, { "epoch": 0.6693732755222704, "grad_norm": 0.5757808089256287, "learning_rate": 0.0005, "loss": 1.9322971105575562, "step": 8491 }, { "epoch": 0.6694521087899094, "grad_norm": 0.6140109896659851, "learning_rate": 0.0005, "loss": 1.7907778024673462, "step": 8492 }, { "epoch": 0.6695309420575483, "grad_norm": 0.8100420236587524, "learning_rate": 0.0005, "loss": 1.8082239627838135, "step": 8493 }, { "epoch": 0.6696097753251873, "grad_norm": 0.7217358946800232, "learning_rate": 0.0005, "loss": 2.0024001598358154, "step": 8494 }, { "epoch": 0.6696886085928262, "grad_norm": 0.704264223575592, "learning_rate": 0.0005, "loss": 2.1587438583374023, "step": 8495 }, { "epoch": 0.6697674418604651, "grad_norm": 0.5894622206687927, "learning_rate": 0.0005, "loss": 1.813929796218872, "step": 8496 }, { "epoch": 0.669846275128104, "grad_norm": 0.6189675331115723, "learning_rate": 0.0005, "loss": 1.848943829536438, "step": 8497 }, { "epoch": 0.669925108395743, "grad_norm": 0.7415697574615479, "learning_rate": 0.0005, "loss": 1.8368377685546875, "step": 8498 }, { "epoch": 0.670003941663382, "grad_norm": 0.5279144048690796, "learning_rate": 0.0005, "loss": 1.7271674871444702, "step": 8499 }, { "epoch": 0.6700827749310209, "grad_norm": 0.6034676432609558, "learning_rate": 0.0005, "loss": 1.8922522068023682, "step": 8500 }, { "epoch": 0.6701616081986599, "grad_norm": 0.5737296938896179, "learning_rate": 0.0005, "loss": 1.982319712638855, "step": 8501 }, { "epoch": 0.6702404414662988, "grad_norm": 0.9091891050338745, "learning_rate": 0.0005, "loss": 2.068242073059082, "step": 8502 }, { "epoch": 0.6703192747339377, "grad_norm": 0.7980204820632935, "learning_rate": 0.0005, "loss": 2.271005153656006, "step": 8503 }, { "epoch": 0.6703981080015766, "grad_norm": 0.8782872557640076, "learning_rate": 0.0005, "loss": 2.17099666595459, "step": 8504 }, { "epoch": 0.6704769412692156, "grad_norm": 0.6234171390533447, "learning_rate": 0.0005, "loss": 1.6502959728240967, "step": 8505 }, { "epoch": 0.6705557745368546, "grad_norm": 0.7208744287490845, "learning_rate": 0.0005, "loss": 1.9389865398406982, "step": 8506 }, { "epoch": 0.6706346078044935, "grad_norm": 0.753692626953125, "learning_rate": 0.0005, "loss": 1.9185065031051636, "step": 8507 }, { "epoch": 0.6707134410721325, "grad_norm": 0.5726492404937744, "learning_rate": 0.0005, "loss": 1.7936846017837524, "step": 8508 }, { "epoch": 0.6707922743397714, "grad_norm": 0.7906981706619263, "learning_rate": 0.0005, "loss": 1.9977904558181763, "step": 8509 }, { "epoch": 0.6708711076074103, "grad_norm": 0.740553617477417, "learning_rate": 0.0005, "loss": 2.023756742477417, "step": 8510 }, { "epoch": 0.6709499408750492, "grad_norm": 0.795475959777832, "learning_rate": 0.0005, "loss": 2.0257837772369385, "step": 8511 }, { "epoch": 0.6710287741426882, "grad_norm": 0.6383272409439087, "learning_rate": 0.0005, "loss": 1.9849838018417358, "step": 8512 }, { "epoch": 0.6711076074103272, "grad_norm": 0.6684447526931763, "learning_rate": 0.0005, "loss": 1.8365910053253174, "step": 8513 }, { "epoch": 0.6711864406779661, "grad_norm": 0.8069792985916138, "learning_rate": 0.0005, "loss": 2.1570918560028076, "step": 8514 }, { "epoch": 0.6712652739456051, "grad_norm": 0.839213490486145, "learning_rate": 0.0005, "loss": 2.028261423110962, "step": 8515 }, { "epoch": 0.671344107213244, "grad_norm": 0.6658850908279419, "learning_rate": 0.0005, "loss": 1.8569642305374146, "step": 8516 }, { "epoch": 0.6714229404808829, "grad_norm": 0.779927670955658, "learning_rate": 0.0005, "loss": 1.9367835521697998, "step": 8517 }, { "epoch": 0.6715017737485218, "grad_norm": 0.5960057377815247, "learning_rate": 0.0005, "loss": 1.9699301719665527, "step": 8518 }, { "epoch": 0.6715806070161608, "grad_norm": 0.6496012210845947, "learning_rate": 0.0005, "loss": 1.8888297080993652, "step": 8519 }, { "epoch": 0.6716594402837998, "grad_norm": 0.6201934218406677, "learning_rate": 0.0005, "loss": 1.7282260656356812, "step": 8520 }, { "epoch": 0.6717382735514387, "grad_norm": 0.6859639883041382, "learning_rate": 0.0005, "loss": 2.1780996322631836, "step": 8521 }, { "epoch": 0.6718171068190777, "grad_norm": 0.600490927696228, "learning_rate": 0.0005, "loss": 1.9220352172851562, "step": 8522 }, { "epoch": 0.6718959400867166, "grad_norm": 0.7602946162223816, "learning_rate": 0.0005, "loss": 1.936082363128662, "step": 8523 }, { "epoch": 0.6719747733543555, "grad_norm": 0.5868539214134216, "learning_rate": 0.0005, "loss": 1.7920260429382324, "step": 8524 }, { "epoch": 0.6720536066219945, "grad_norm": 0.5893965363502502, "learning_rate": 0.0005, "loss": 1.7906291484832764, "step": 8525 }, { "epoch": 0.6721324398896334, "grad_norm": 0.9444671869277954, "learning_rate": 0.0005, "loss": 1.9180073738098145, "step": 8526 }, { "epoch": 0.6722112731572724, "grad_norm": 0.6678837537765503, "learning_rate": 0.0005, "loss": 1.8742170333862305, "step": 8527 }, { "epoch": 0.6722901064249113, "grad_norm": 0.6625818014144897, "learning_rate": 0.0005, "loss": 1.7318520545959473, "step": 8528 }, { "epoch": 0.6723689396925503, "grad_norm": 0.6761202812194824, "learning_rate": 0.0005, "loss": 1.9322258234024048, "step": 8529 }, { "epoch": 0.6724477729601892, "grad_norm": 0.8264444470405579, "learning_rate": 0.0005, "loss": 1.9933291673660278, "step": 8530 }, { "epoch": 0.6725266062278281, "grad_norm": 0.9086037278175354, "learning_rate": 0.0005, "loss": 2.296633720397949, "step": 8531 }, { "epoch": 0.6726054394954671, "grad_norm": 0.6563221216201782, "learning_rate": 0.0005, "loss": 2.2541942596435547, "step": 8532 }, { "epoch": 0.672684272763106, "grad_norm": 0.8169102668762207, "learning_rate": 0.0005, "loss": 1.8578479290008545, "step": 8533 }, { "epoch": 0.672763106030745, "grad_norm": 0.5569620728492737, "learning_rate": 0.0005, "loss": 1.8967105150222778, "step": 8534 }, { "epoch": 0.6728419392983839, "grad_norm": 0.7903403043746948, "learning_rate": 0.0005, "loss": 2.0625739097595215, "step": 8535 }, { "epoch": 0.6729207725660229, "grad_norm": 0.790250837802887, "learning_rate": 0.0005, "loss": 1.9130396842956543, "step": 8536 }, { "epoch": 0.6729996058336618, "grad_norm": 0.795436680316925, "learning_rate": 0.0005, "loss": 2.0251145362854004, "step": 8537 }, { "epoch": 0.6730784391013007, "grad_norm": 0.7196424007415771, "learning_rate": 0.0005, "loss": 1.9620568752288818, "step": 8538 }, { "epoch": 0.6731572723689397, "grad_norm": 0.6688858866691589, "learning_rate": 0.0005, "loss": 2.03956937789917, "step": 8539 }, { "epoch": 0.6732361056365787, "grad_norm": 0.6651943325996399, "learning_rate": 0.0005, "loss": 1.7130441665649414, "step": 8540 }, { "epoch": 0.6733149389042176, "grad_norm": 0.7496647834777832, "learning_rate": 0.0005, "loss": 2.1864166259765625, "step": 8541 }, { "epoch": 0.6733937721718565, "grad_norm": 0.702559232711792, "learning_rate": 0.0005, "loss": 2.079049825668335, "step": 8542 }, { "epoch": 0.6734726054394955, "grad_norm": 0.63485187292099, "learning_rate": 0.0005, "loss": 1.8757247924804688, "step": 8543 }, { "epoch": 0.6735514387071344, "grad_norm": 0.7392326593399048, "learning_rate": 0.0005, "loss": 1.9449949264526367, "step": 8544 }, { "epoch": 0.6736302719747733, "grad_norm": 0.6355005502700806, "learning_rate": 0.0005, "loss": 2.046063184738159, "step": 8545 }, { "epoch": 0.6737091052424123, "grad_norm": 0.6310865879058838, "learning_rate": 0.0005, "loss": 1.8214699029922485, "step": 8546 }, { "epoch": 0.6737879385100513, "grad_norm": 0.7831661105155945, "learning_rate": 0.0005, "loss": 2.2610228061676025, "step": 8547 }, { "epoch": 0.6738667717776902, "grad_norm": 0.676266610622406, "learning_rate": 0.0005, "loss": 2.1136622428894043, "step": 8548 }, { "epoch": 0.6739456050453291, "grad_norm": 0.5823510885238647, "learning_rate": 0.0005, "loss": 1.925546646118164, "step": 8549 }, { "epoch": 0.6740244383129681, "grad_norm": 0.6475022435188293, "learning_rate": 0.0005, "loss": 1.9562151432037354, "step": 8550 }, { "epoch": 0.674103271580607, "grad_norm": 0.5672389268875122, "learning_rate": 0.0005, "loss": 1.800599217414856, "step": 8551 }, { "epoch": 0.6741821048482459, "grad_norm": 0.8811495304107666, "learning_rate": 0.0005, "loss": 2.0707216262817383, "step": 8552 }, { "epoch": 0.674260938115885, "grad_norm": 0.6440484523773193, "learning_rate": 0.0005, "loss": 2.099863290786743, "step": 8553 }, { "epoch": 0.6743397713835239, "grad_norm": 0.7121887803077698, "learning_rate": 0.0005, "loss": 1.8697434663772583, "step": 8554 }, { "epoch": 0.6744186046511628, "grad_norm": 0.7205542325973511, "learning_rate": 0.0005, "loss": 2.075195550918579, "step": 8555 }, { "epoch": 0.6744974379188017, "grad_norm": 0.6843655109405518, "learning_rate": 0.0005, "loss": 1.775890827178955, "step": 8556 }, { "epoch": 0.6745762711864407, "grad_norm": 0.5682156682014465, "learning_rate": 0.0005, "loss": 2.0817933082580566, "step": 8557 }, { "epoch": 0.6746551044540796, "grad_norm": 0.7869980335235596, "learning_rate": 0.0005, "loss": 2.110135555267334, "step": 8558 }, { "epoch": 0.6747339377217185, "grad_norm": 0.9457343816757202, "learning_rate": 0.0005, "loss": 2.1894800662994385, "step": 8559 }, { "epoch": 0.6748127709893575, "grad_norm": 0.6914669871330261, "learning_rate": 0.0005, "loss": 1.929782509803772, "step": 8560 }, { "epoch": 0.6748916042569965, "grad_norm": 0.6077049374580383, "learning_rate": 0.0005, "loss": 1.7373921871185303, "step": 8561 }, { "epoch": 0.6749704375246354, "grad_norm": 0.7083054184913635, "learning_rate": 0.0005, "loss": 2.327157735824585, "step": 8562 }, { "epoch": 0.6750492707922744, "grad_norm": 0.6688942909240723, "learning_rate": 0.0005, "loss": 2.174077033996582, "step": 8563 }, { "epoch": 0.6751281040599133, "grad_norm": 0.57797771692276, "learning_rate": 0.0005, "loss": 1.7863576412200928, "step": 8564 }, { "epoch": 0.6752069373275522, "grad_norm": 0.6576065421104431, "learning_rate": 0.0005, "loss": 1.822540044784546, "step": 8565 }, { "epoch": 0.6752857705951911, "grad_norm": 0.6542556881904602, "learning_rate": 0.0005, "loss": 1.7725613117218018, "step": 8566 }, { "epoch": 0.6753646038628301, "grad_norm": 0.6818370223045349, "learning_rate": 0.0005, "loss": 2.039755344390869, "step": 8567 }, { "epoch": 0.6754434371304691, "grad_norm": 0.7244836688041687, "learning_rate": 0.0005, "loss": 1.9133777618408203, "step": 8568 }, { "epoch": 0.675522270398108, "grad_norm": 0.573406994342804, "learning_rate": 0.0005, "loss": 1.8054728507995605, "step": 8569 }, { "epoch": 0.675601103665747, "grad_norm": 0.6361267566680908, "learning_rate": 0.0005, "loss": 1.9183399677276611, "step": 8570 }, { "epoch": 0.6756799369333859, "grad_norm": 0.8258230686187744, "learning_rate": 0.0005, "loss": 1.6711900234222412, "step": 8571 }, { "epoch": 0.6757587702010248, "grad_norm": 0.7424345016479492, "learning_rate": 0.0005, "loss": 2.202921152114868, "step": 8572 }, { "epoch": 0.6758376034686637, "grad_norm": 0.5486292243003845, "learning_rate": 0.0005, "loss": 1.7977330684661865, "step": 8573 }, { "epoch": 0.6759164367363028, "grad_norm": 0.5721737742424011, "learning_rate": 0.0005, "loss": 1.9192529916763306, "step": 8574 }, { "epoch": 0.6759952700039417, "grad_norm": 0.843104362487793, "learning_rate": 0.0005, "loss": 2.1950793266296387, "step": 8575 }, { "epoch": 0.6760741032715806, "grad_norm": 0.7597920894622803, "learning_rate": 0.0005, "loss": 1.9975427389144897, "step": 8576 }, { "epoch": 0.6761529365392196, "grad_norm": 0.7641933560371399, "learning_rate": 0.0005, "loss": 2.2739975452423096, "step": 8577 }, { "epoch": 0.6762317698068585, "grad_norm": 0.6502341032028198, "learning_rate": 0.0005, "loss": 1.842904806137085, "step": 8578 }, { "epoch": 0.6763106030744974, "grad_norm": 0.9280173182487488, "learning_rate": 0.0005, "loss": 1.7925667762756348, "step": 8579 }, { "epoch": 0.6763894363421363, "grad_norm": 0.5927907824516296, "learning_rate": 0.0005, "loss": 1.786948561668396, "step": 8580 }, { "epoch": 0.6764682696097754, "grad_norm": 0.958641529083252, "learning_rate": 0.0005, "loss": 1.9094865322113037, "step": 8581 }, { "epoch": 0.6765471028774143, "grad_norm": 0.6185813546180725, "learning_rate": 0.0005, "loss": 1.8659565448760986, "step": 8582 }, { "epoch": 0.6766259361450532, "grad_norm": 0.932476818561554, "learning_rate": 0.0005, "loss": 1.8545228242874146, "step": 8583 }, { "epoch": 0.6767047694126922, "grad_norm": 0.9397087097167969, "learning_rate": 0.0005, "loss": 2.1449131965637207, "step": 8584 }, { "epoch": 0.6767836026803311, "grad_norm": 0.6033342480659485, "learning_rate": 0.0005, "loss": 1.7550984621047974, "step": 8585 }, { "epoch": 0.67686243594797, "grad_norm": 0.6026788949966431, "learning_rate": 0.0005, "loss": 2.1453118324279785, "step": 8586 }, { "epoch": 0.6769412692156089, "grad_norm": 0.7030408978462219, "learning_rate": 0.0005, "loss": 1.9757890701293945, "step": 8587 }, { "epoch": 0.677020102483248, "grad_norm": 0.6075047850608826, "learning_rate": 0.0005, "loss": 1.7438359260559082, "step": 8588 }, { "epoch": 0.6770989357508869, "grad_norm": 0.6584635376930237, "learning_rate": 0.0005, "loss": 1.9280203580856323, "step": 8589 }, { "epoch": 0.6771777690185258, "grad_norm": 0.6684239506721497, "learning_rate": 0.0005, "loss": 1.6580250263214111, "step": 8590 }, { "epoch": 0.6772566022861648, "grad_norm": 0.5830010771751404, "learning_rate": 0.0005, "loss": 1.8058669567108154, "step": 8591 }, { "epoch": 0.6773354355538037, "grad_norm": 0.5871631503105164, "learning_rate": 0.0005, "loss": 1.6465078592300415, "step": 8592 }, { "epoch": 0.6774142688214426, "grad_norm": 0.6412233710289001, "learning_rate": 0.0005, "loss": 1.725164532661438, "step": 8593 }, { "epoch": 0.6774931020890816, "grad_norm": 0.8784490823745728, "learning_rate": 0.0005, "loss": 2.021143913269043, "step": 8594 }, { "epoch": 0.6775719353567206, "grad_norm": 0.6024890542030334, "learning_rate": 0.0005, "loss": 1.7837103605270386, "step": 8595 }, { "epoch": 0.6776507686243595, "grad_norm": 0.6390483379364014, "learning_rate": 0.0005, "loss": 1.9112192392349243, "step": 8596 }, { "epoch": 0.6777296018919984, "grad_norm": 0.6678439974784851, "learning_rate": 0.0005, "loss": 1.8624252080917358, "step": 8597 }, { "epoch": 0.6778084351596374, "grad_norm": 0.6960686445236206, "learning_rate": 0.0005, "loss": 1.8783409595489502, "step": 8598 }, { "epoch": 0.6778872684272763, "grad_norm": 0.6529090404510498, "learning_rate": 0.0005, "loss": 1.8526631593704224, "step": 8599 }, { "epoch": 0.6779661016949152, "grad_norm": 0.8052026629447937, "learning_rate": 0.0005, "loss": 1.8895787000656128, "step": 8600 }, { "epoch": 0.6780449349625542, "grad_norm": 0.7622575163841248, "learning_rate": 0.0005, "loss": 1.771226406097412, "step": 8601 }, { "epoch": 0.6781237682301932, "grad_norm": 0.7503442168235779, "learning_rate": 0.0005, "loss": 1.938664197921753, "step": 8602 }, { "epoch": 0.6782026014978321, "grad_norm": 0.8418692946434021, "learning_rate": 0.0005, "loss": 2.0983633995056152, "step": 8603 }, { "epoch": 0.678281434765471, "grad_norm": 0.8483644723892212, "learning_rate": 0.0005, "loss": 1.8545055389404297, "step": 8604 }, { "epoch": 0.67836026803311, "grad_norm": 0.6586797833442688, "learning_rate": 0.0005, "loss": 2.025773525238037, "step": 8605 }, { "epoch": 0.6784391013007489, "grad_norm": 0.595202624797821, "learning_rate": 0.0005, "loss": 2.0171585083007812, "step": 8606 }, { "epoch": 0.6785179345683878, "grad_norm": 0.8338581323623657, "learning_rate": 0.0005, "loss": 2.1697802543640137, "step": 8607 }, { "epoch": 0.6785967678360268, "grad_norm": 0.8077709674835205, "learning_rate": 0.0005, "loss": 1.92388916015625, "step": 8608 }, { "epoch": 0.6786756011036658, "grad_norm": 0.5962253212928772, "learning_rate": 0.0005, "loss": 1.769134283065796, "step": 8609 }, { "epoch": 0.6787544343713047, "grad_norm": 0.768866777420044, "learning_rate": 0.0005, "loss": 2.2107834815979004, "step": 8610 }, { "epoch": 0.6788332676389436, "grad_norm": 0.6343615055084229, "learning_rate": 0.0005, "loss": 2.0454354286193848, "step": 8611 }, { "epoch": 0.6789121009065826, "grad_norm": 0.481851190328598, "learning_rate": 0.0005, "loss": 1.8115885257720947, "step": 8612 }, { "epoch": 0.6789909341742215, "grad_norm": 0.678600549697876, "learning_rate": 0.0005, "loss": 1.8843053579330444, "step": 8613 }, { "epoch": 0.6790697674418604, "grad_norm": 0.7148755788803101, "learning_rate": 0.0005, "loss": 1.6986433267593384, "step": 8614 }, { "epoch": 0.6791486007094995, "grad_norm": 0.5919039249420166, "learning_rate": 0.0005, "loss": 1.7433497905731201, "step": 8615 }, { "epoch": 0.6792274339771384, "grad_norm": 0.5844440460205078, "learning_rate": 0.0005, "loss": 1.7506955862045288, "step": 8616 }, { "epoch": 0.6793062672447773, "grad_norm": 0.6942524909973145, "learning_rate": 0.0005, "loss": 1.9523097276687622, "step": 8617 }, { "epoch": 0.6793851005124162, "grad_norm": 1.3262940645217896, "learning_rate": 0.0005, "loss": 2.0930869579315186, "step": 8618 }, { "epoch": 0.6794639337800552, "grad_norm": 0.7537075877189636, "learning_rate": 0.0005, "loss": 1.9725853204727173, "step": 8619 }, { "epoch": 0.6795427670476941, "grad_norm": 1.0865583419799805, "learning_rate": 0.0005, "loss": 2.081362724304199, "step": 8620 }, { "epoch": 0.679621600315333, "grad_norm": 0.8223584890365601, "learning_rate": 0.0005, "loss": 1.7761025428771973, "step": 8621 }, { "epoch": 0.679700433582972, "grad_norm": 0.7853056192398071, "learning_rate": 0.0005, "loss": 1.806278944015503, "step": 8622 }, { "epoch": 0.679779266850611, "grad_norm": 0.7531580328941345, "learning_rate": 0.0005, "loss": 2.188516855239868, "step": 8623 }, { "epoch": 0.6798581001182499, "grad_norm": 0.8056167364120483, "learning_rate": 0.0005, "loss": 2.2642369270324707, "step": 8624 }, { "epoch": 0.6799369333858889, "grad_norm": 0.7009548544883728, "learning_rate": 0.0005, "loss": 1.9283467531204224, "step": 8625 }, { "epoch": 0.6800157666535278, "grad_norm": 0.8345093131065369, "learning_rate": 0.0005, "loss": 2.006535053253174, "step": 8626 }, { "epoch": 0.6800945999211667, "grad_norm": 0.7410575151443481, "learning_rate": 0.0005, "loss": 1.7320115566253662, "step": 8627 }, { "epoch": 0.6801734331888056, "grad_norm": 0.7012538313865662, "learning_rate": 0.0005, "loss": 1.841556429862976, "step": 8628 }, { "epoch": 0.6802522664564447, "grad_norm": 0.7442522048950195, "learning_rate": 0.0005, "loss": 1.8640587329864502, "step": 8629 }, { "epoch": 0.6803310997240836, "grad_norm": 0.7672918438911438, "learning_rate": 0.0005, "loss": 1.7418384552001953, "step": 8630 }, { "epoch": 0.6804099329917225, "grad_norm": 0.6833810806274414, "learning_rate": 0.0005, "loss": 2.0775113105773926, "step": 8631 }, { "epoch": 0.6804887662593615, "grad_norm": 0.6693474650382996, "learning_rate": 0.0005, "loss": 1.9984166622161865, "step": 8632 }, { "epoch": 0.6805675995270004, "grad_norm": 0.8574925065040588, "learning_rate": 0.0005, "loss": 1.9870412349700928, "step": 8633 }, { "epoch": 0.6806464327946393, "grad_norm": 1.1818901300430298, "learning_rate": 0.0005, "loss": 1.83259916305542, "step": 8634 }, { "epoch": 0.6807252660622782, "grad_norm": 0.6486817598342896, "learning_rate": 0.0005, "loss": 1.7234543561935425, "step": 8635 }, { "epoch": 0.6808040993299173, "grad_norm": 0.6748758554458618, "learning_rate": 0.0005, "loss": 1.9120014905929565, "step": 8636 }, { "epoch": 0.6808829325975562, "grad_norm": 0.7730270028114319, "learning_rate": 0.0005, "loss": 1.9420390129089355, "step": 8637 }, { "epoch": 0.6809617658651951, "grad_norm": 0.745000958442688, "learning_rate": 0.0005, "loss": 2.0632784366607666, "step": 8638 }, { "epoch": 0.6810405991328341, "grad_norm": 0.7371770143508911, "learning_rate": 0.0005, "loss": 1.8029552698135376, "step": 8639 }, { "epoch": 0.681119432400473, "grad_norm": 0.6738818883895874, "learning_rate": 0.0005, "loss": 2.0124166011810303, "step": 8640 }, { "epoch": 0.6811982656681119, "grad_norm": 0.6850194334983826, "learning_rate": 0.0005, "loss": 2.283459186553955, "step": 8641 }, { "epoch": 0.6812770989357508, "grad_norm": 0.8987077474594116, "learning_rate": 0.0005, "loss": 1.8573832511901855, "step": 8642 }, { "epoch": 0.6813559322033899, "grad_norm": 0.6523581743240356, "learning_rate": 0.0005, "loss": 1.7893338203430176, "step": 8643 }, { "epoch": 0.6814347654710288, "grad_norm": 0.6142168045043945, "learning_rate": 0.0005, "loss": 1.7701045274734497, "step": 8644 }, { "epoch": 0.6815135987386677, "grad_norm": 0.7145549654960632, "learning_rate": 0.0005, "loss": 1.7846457958221436, "step": 8645 }, { "epoch": 0.6815924320063067, "grad_norm": 0.5310472846031189, "learning_rate": 0.0005, "loss": 1.7850520610809326, "step": 8646 }, { "epoch": 0.6816712652739456, "grad_norm": 0.726791262626648, "learning_rate": 0.0005, "loss": 1.939948558807373, "step": 8647 }, { "epoch": 0.6817500985415845, "grad_norm": 0.8153197169303894, "learning_rate": 0.0005, "loss": 2.2966630458831787, "step": 8648 }, { "epoch": 0.6818289318092234, "grad_norm": 0.657659649848938, "learning_rate": 0.0005, "loss": 1.9219003915786743, "step": 8649 }, { "epoch": 0.6819077650768625, "grad_norm": 0.5779261589050293, "learning_rate": 0.0005, "loss": 1.5363181829452515, "step": 8650 }, { "epoch": 0.6819865983445014, "grad_norm": 1.2785723209381104, "learning_rate": 0.0005, "loss": 2.0113275051116943, "step": 8651 }, { "epoch": 0.6820654316121403, "grad_norm": 0.8272299766540527, "learning_rate": 0.0005, "loss": 2.0915160179138184, "step": 8652 }, { "epoch": 0.6821442648797793, "grad_norm": 0.7953646779060364, "learning_rate": 0.0005, "loss": 2.411990165710449, "step": 8653 }, { "epoch": 0.6822230981474182, "grad_norm": 0.7268038988113403, "learning_rate": 0.0005, "loss": 1.8308409452438354, "step": 8654 }, { "epoch": 0.6823019314150571, "grad_norm": 0.6458531022071838, "learning_rate": 0.0005, "loss": 1.8742146492004395, "step": 8655 }, { "epoch": 0.6823807646826962, "grad_norm": 0.7104936838150024, "learning_rate": 0.0005, "loss": 1.9009506702423096, "step": 8656 }, { "epoch": 0.6824595979503351, "grad_norm": 1.0361380577087402, "learning_rate": 0.0005, "loss": 1.7184937000274658, "step": 8657 }, { "epoch": 0.682538431217974, "grad_norm": 1.0349805355072021, "learning_rate": 0.0005, "loss": 1.9366612434387207, "step": 8658 }, { "epoch": 0.6826172644856129, "grad_norm": 0.6769363880157471, "learning_rate": 0.0005, "loss": 1.9382538795471191, "step": 8659 }, { "epoch": 0.6826960977532519, "grad_norm": 0.7299816608428955, "learning_rate": 0.0005, "loss": 1.7876005172729492, "step": 8660 }, { "epoch": 0.6827749310208908, "grad_norm": 0.6543339490890503, "learning_rate": 0.0005, "loss": 1.990053653717041, "step": 8661 }, { "epoch": 0.6828537642885297, "grad_norm": 0.6199028491973877, "learning_rate": 0.0005, "loss": 1.7459391355514526, "step": 8662 }, { "epoch": 0.6829325975561688, "grad_norm": 0.5579183101654053, "learning_rate": 0.0005, "loss": 1.7845258712768555, "step": 8663 }, { "epoch": 0.6830114308238077, "grad_norm": 0.668214738368988, "learning_rate": 0.0005, "loss": 1.8011029958724976, "step": 8664 }, { "epoch": 0.6830902640914466, "grad_norm": 0.8127542734146118, "learning_rate": 0.0005, "loss": 1.9101593494415283, "step": 8665 }, { "epoch": 0.6831690973590855, "grad_norm": 0.6237632632255554, "learning_rate": 0.0005, "loss": 1.9096686840057373, "step": 8666 }, { "epoch": 0.6832479306267245, "grad_norm": 0.64863121509552, "learning_rate": 0.0005, "loss": 1.8836231231689453, "step": 8667 }, { "epoch": 0.6833267638943634, "grad_norm": 0.8058998584747314, "learning_rate": 0.0005, "loss": 2.1244428157806396, "step": 8668 }, { "epoch": 0.6834055971620023, "grad_norm": 0.5230349898338318, "learning_rate": 0.0005, "loss": 1.7630383968353271, "step": 8669 }, { "epoch": 0.6834844304296414, "grad_norm": 0.7378169894218445, "learning_rate": 0.0005, "loss": 2.068213939666748, "step": 8670 }, { "epoch": 0.6835632636972803, "grad_norm": 0.9104118943214417, "learning_rate": 0.0005, "loss": 2.139063835144043, "step": 8671 }, { "epoch": 0.6836420969649192, "grad_norm": 0.6391451358795166, "learning_rate": 0.0005, "loss": 2.025796413421631, "step": 8672 }, { "epoch": 0.6837209302325581, "grad_norm": 0.6862768530845642, "learning_rate": 0.0005, "loss": 1.8167593479156494, "step": 8673 }, { "epoch": 0.6837997635001971, "grad_norm": 0.7206098437309265, "learning_rate": 0.0005, "loss": 2.078362226486206, "step": 8674 }, { "epoch": 0.683878596767836, "grad_norm": 0.5901284217834473, "learning_rate": 0.0005, "loss": 1.8606407642364502, "step": 8675 }, { "epoch": 0.6839574300354749, "grad_norm": 0.7215868234634399, "learning_rate": 0.0005, "loss": 2.008883476257324, "step": 8676 }, { "epoch": 0.684036263303114, "grad_norm": 0.6023926138877869, "learning_rate": 0.0005, "loss": 2.0703866481781006, "step": 8677 }, { "epoch": 0.6841150965707529, "grad_norm": 0.8342561721801758, "learning_rate": 0.0005, "loss": 2.014181137084961, "step": 8678 }, { "epoch": 0.6841939298383918, "grad_norm": 0.6162523627281189, "learning_rate": 0.0005, "loss": 1.6817317008972168, "step": 8679 }, { "epoch": 0.6842727631060307, "grad_norm": 0.6961566209793091, "learning_rate": 0.0005, "loss": 1.8017607927322388, "step": 8680 }, { "epoch": 0.6843515963736697, "grad_norm": 0.6696818470954895, "learning_rate": 0.0005, "loss": 1.816842794418335, "step": 8681 }, { "epoch": 0.6844304296413086, "grad_norm": 0.767470121383667, "learning_rate": 0.0005, "loss": 1.9229199886322021, "step": 8682 }, { "epoch": 0.6845092629089475, "grad_norm": 0.6755869388580322, "learning_rate": 0.0005, "loss": 1.6227259635925293, "step": 8683 }, { "epoch": 0.6845880961765866, "grad_norm": 0.7170653939247131, "learning_rate": 0.0005, "loss": 2.121123790740967, "step": 8684 }, { "epoch": 0.6846669294442255, "grad_norm": 1.0714701414108276, "learning_rate": 0.0005, "loss": 2.029236316680908, "step": 8685 }, { "epoch": 0.6847457627118644, "grad_norm": 0.7361412644386292, "learning_rate": 0.0005, "loss": 1.9151957035064697, "step": 8686 }, { "epoch": 0.6848245959795034, "grad_norm": 0.8553542494773865, "learning_rate": 0.0005, "loss": 2.21732759475708, "step": 8687 }, { "epoch": 0.6849034292471423, "grad_norm": 0.7463082075119019, "learning_rate": 0.0005, "loss": 2.083547592163086, "step": 8688 }, { "epoch": 0.6849822625147812, "grad_norm": 1.092117428779602, "learning_rate": 0.0005, "loss": 1.653913140296936, "step": 8689 }, { "epoch": 0.6850610957824201, "grad_norm": 0.5221986770629883, "learning_rate": 0.0005, "loss": 1.82805335521698, "step": 8690 }, { "epoch": 0.6851399290500592, "grad_norm": 0.7362056970596313, "learning_rate": 0.0005, "loss": 1.8690048456192017, "step": 8691 }, { "epoch": 0.6852187623176981, "grad_norm": 0.6812586188316345, "learning_rate": 0.0005, "loss": 2.046363353729248, "step": 8692 }, { "epoch": 0.685297595585337, "grad_norm": 0.5712913870811462, "learning_rate": 0.0005, "loss": 1.7229118347167969, "step": 8693 }, { "epoch": 0.685376428852976, "grad_norm": 0.8312872052192688, "learning_rate": 0.0005, "loss": 2.183504819869995, "step": 8694 }, { "epoch": 0.6854552621206149, "grad_norm": 0.6574113965034485, "learning_rate": 0.0005, "loss": 1.9122743606567383, "step": 8695 }, { "epoch": 0.6855340953882538, "grad_norm": 0.6319741010665894, "learning_rate": 0.0005, "loss": 1.5683751106262207, "step": 8696 }, { "epoch": 0.6856129286558927, "grad_norm": 0.7248690724372864, "learning_rate": 0.0005, "loss": 1.897550106048584, "step": 8697 }, { "epoch": 0.6856917619235318, "grad_norm": 0.67668217420578, "learning_rate": 0.0005, "loss": 2.059725284576416, "step": 8698 }, { "epoch": 0.6857705951911707, "grad_norm": 0.6375545859336853, "learning_rate": 0.0005, "loss": 1.9737060070037842, "step": 8699 }, { "epoch": 0.6858494284588096, "grad_norm": 0.726442277431488, "learning_rate": 0.0005, "loss": 1.9855380058288574, "step": 8700 }, { "epoch": 0.6859282617264486, "grad_norm": 0.6968805193901062, "learning_rate": 0.0005, "loss": 1.9892805814743042, "step": 8701 }, { "epoch": 0.6860070949940875, "grad_norm": 0.6208008527755737, "learning_rate": 0.0005, "loss": 1.9490115642547607, "step": 8702 }, { "epoch": 0.6860859282617264, "grad_norm": 0.6522743701934814, "learning_rate": 0.0005, "loss": 1.9106736183166504, "step": 8703 }, { "epoch": 0.6861647615293653, "grad_norm": 0.654243528842926, "learning_rate": 0.0005, "loss": 1.9983742237091064, "step": 8704 }, { "epoch": 0.6862435947970044, "grad_norm": 0.6616128087043762, "learning_rate": 0.0005, "loss": 1.8000503778457642, "step": 8705 }, { "epoch": 0.6863224280646433, "grad_norm": 0.7290578484535217, "learning_rate": 0.0005, "loss": 1.8532450199127197, "step": 8706 }, { "epoch": 0.6864012613322822, "grad_norm": 0.5886913537979126, "learning_rate": 0.0005, "loss": 1.6857702732086182, "step": 8707 }, { "epoch": 0.6864800945999212, "grad_norm": 0.7145200967788696, "learning_rate": 0.0005, "loss": 2.1330652236938477, "step": 8708 }, { "epoch": 0.6865589278675601, "grad_norm": 0.7409058809280396, "learning_rate": 0.0005, "loss": 2.058671474456787, "step": 8709 }, { "epoch": 0.686637761135199, "grad_norm": 0.7088483572006226, "learning_rate": 0.0005, "loss": 1.93672776222229, "step": 8710 }, { "epoch": 0.6867165944028379, "grad_norm": 0.8666973114013672, "learning_rate": 0.0005, "loss": 2.0193395614624023, "step": 8711 }, { "epoch": 0.686795427670477, "grad_norm": 0.7337602972984314, "learning_rate": 0.0005, "loss": 1.9535170793533325, "step": 8712 }, { "epoch": 0.6868742609381159, "grad_norm": 1.0752637386322021, "learning_rate": 0.0005, "loss": 1.8604422807693481, "step": 8713 }, { "epoch": 0.6869530942057548, "grad_norm": 0.9664540886878967, "learning_rate": 0.0005, "loss": 2.0846688747406006, "step": 8714 }, { "epoch": 0.6870319274733938, "grad_norm": 0.8302044868469238, "learning_rate": 0.0005, "loss": 2.0805697441101074, "step": 8715 }, { "epoch": 0.6871107607410327, "grad_norm": 0.8686814904212952, "learning_rate": 0.0005, "loss": 1.9887402057647705, "step": 8716 }, { "epoch": 0.6871895940086716, "grad_norm": 0.7788152694702148, "learning_rate": 0.0005, "loss": 1.8533883094787598, "step": 8717 }, { "epoch": 0.6872684272763107, "grad_norm": 0.6267239451408386, "learning_rate": 0.0005, "loss": 2.2003324031829834, "step": 8718 }, { "epoch": 0.6873472605439496, "grad_norm": 0.5692688822746277, "learning_rate": 0.0005, "loss": 1.81269371509552, "step": 8719 }, { "epoch": 0.6874260938115885, "grad_norm": 0.7351714968681335, "learning_rate": 0.0005, "loss": 1.9621145725250244, "step": 8720 }, { "epoch": 0.6875049270792274, "grad_norm": 0.9148641228675842, "learning_rate": 0.0005, "loss": 1.8867661952972412, "step": 8721 }, { "epoch": 0.6875837603468664, "grad_norm": 0.825027585029602, "learning_rate": 0.0005, "loss": 2.0450363159179688, "step": 8722 }, { "epoch": 0.6876625936145053, "grad_norm": 0.5804219245910645, "learning_rate": 0.0005, "loss": 1.7417247295379639, "step": 8723 }, { "epoch": 0.6877414268821442, "grad_norm": 0.7190540432929993, "learning_rate": 0.0005, "loss": 2.018407106399536, "step": 8724 }, { "epoch": 0.6878202601497833, "grad_norm": 0.665303647518158, "learning_rate": 0.0005, "loss": 1.671442985534668, "step": 8725 }, { "epoch": 0.6878990934174222, "grad_norm": 0.6408644914627075, "learning_rate": 0.0005, "loss": 1.6063249111175537, "step": 8726 }, { "epoch": 0.6879779266850611, "grad_norm": 0.6196833252906799, "learning_rate": 0.0005, "loss": 1.8299510478973389, "step": 8727 }, { "epoch": 0.6880567599527, "grad_norm": 0.6841453313827515, "learning_rate": 0.0005, "loss": 2.075669288635254, "step": 8728 }, { "epoch": 0.688135593220339, "grad_norm": 0.7032492756843567, "learning_rate": 0.0005, "loss": 1.8891923427581787, "step": 8729 }, { "epoch": 0.6882144264879779, "grad_norm": 0.7191810011863708, "learning_rate": 0.0005, "loss": 1.8589465618133545, "step": 8730 }, { "epoch": 0.6882932597556168, "grad_norm": 0.609432578086853, "learning_rate": 0.0005, "loss": 1.8613271713256836, "step": 8731 }, { "epoch": 0.6883720930232559, "grad_norm": 0.8323217630386353, "learning_rate": 0.0005, "loss": 2.0871787071228027, "step": 8732 }, { "epoch": 0.6884509262908948, "grad_norm": 0.988520622253418, "learning_rate": 0.0005, "loss": 1.6864373683929443, "step": 8733 }, { "epoch": 0.6885297595585337, "grad_norm": 0.8323594927787781, "learning_rate": 0.0005, "loss": 2.3613574504852295, "step": 8734 }, { "epoch": 0.6886085928261726, "grad_norm": 0.826377272605896, "learning_rate": 0.0005, "loss": 2.339378833770752, "step": 8735 }, { "epoch": 0.6886874260938116, "grad_norm": 0.7861857414245605, "learning_rate": 0.0005, "loss": 1.7363760471343994, "step": 8736 }, { "epoch": 0.6887662593614505, "grad_norm": 0.8435484766960144, "learning_rate": 0.0005, "loss": 2.2521986961364746, "step": 8737 }, { "epoch": 0.6888450926290894, "grad_norm": 0.6038731932640076, "learning_rate": 0.0005, "loss": 1.8938947916030884, "step": 8738 }, { "epoch": 0.6889239258967285, "grad_norm": 0.6508927941322327, "learning_rate": 0.0005, "loss": 2.064061164855957, "step": 8739 }, { "epoch": 0.6890027591643674, "grad_norm": 0.6005999445915222, "learning_rate": 0.0005, "loss": 1.8565597534179688, "step": 8740 }, { "epoch": 0.6890815924320063, "grad_norm": 0.6716495752334595, "learning_rate": 0.0005, "loss": 1.8243530988693237, "step": 8741 }, { "epoch": 0.6891604256996452, "grad_norm": 0.7859533429145813, "learning_rate": 0.0005, "loss": 1.8453326225280762, "step": 8742 }, { "epoch": 0.6892392589672842, "grad_norm": 0.665876567363739, "learning_rate": 0.0005, "loss": 1.9734004735946655, "step": 8743 }, { "epoch": 0.6893180922349231, "grad_norm": 0.734544038772583, "learning_rate": 0.0005, "loss": 1.9655735492706299, "step": 8744 }, { "epoch": 0.689396925502562, "grad_norm": 0.8647028207778931, "learning_rate": 0.0005, "loss": 2.3436572551727295, "step": 8745 }, { "epoch": 0.6894757587702011, "grad_norm": 0.637925922870636, "learning_rate": 0.0005, "loss": 1.954982042312622, "step": 8746 }, { "epoch": 0.68955459203784, "grad_norm": 1.0981543064117432, "learning_rate": 0.0005, "loss": 2.3999712467193604, "step": 8747 }, { "epoch": 0.6896334253054789, "grad_norm": 0.59820157289505, "learning_rate": 0.0005, "loss": 1.766838550567627, "step": 8748 }, { "epoch": 0.6897122585731179, "grad_norm": 0.7337178587913513, "learning_rate": 0.0005, "loss": 2.136291027069092, "step": 8749 }, { "epoch": 0.6897910918407568, "grad_norm": 0.6704249382019043, "learning_rate": 0.0005, "loss": 1.8526697158813477, "step": 8750 }, { "epoch": 0.6898699251083957, "grad_norm": 1.098075032234192, "learning_rate": 0.0005, "loss": 2.4198033809661865, "step": 8751 }, { "epoch": 0.6899487583760346, "grad_norm": 0.6893532872200012, "learning_rate": 0.0005, "loss": 1.8086750507354736, "step": 8752 }, { "epoch": 0.6900275916436737, "grad_norm": 0.6167333126068115, "learning_rate": 0.0005, "loss": 2.090949535369873, "step": 8753 }, { "epoch": 0.6901064249113126, "grad_norm": 0.5307744145393372, "learning_rate": 0.0005, "loss": 1.7716853618621826, "step": 8754 }, { "epoch": 0.6901852581789515, "grad_norm": 0.7035183906555176, "learning_rate": 0.0005, "loss": 1.9328519105911255, "step": 8755 }, { "epoch": 0.6902640914465905, "grad_norm": 0.6235803365707397, "learning_rate": 0.0005, "loss": 2.060572385787964, "step": 8756 }, { "epoch": 0.6903429247142294, "grad_norm": 0.6307493448257446, "learning_rate": 0.0005, "loss": 1.6559364795684814, "step": 8757 }, { "epoch": 0.6904217579818683, "grad_norm": 0.6333315968513489, "learning_rate": 0.0005, "loss": 1.7237945795059204, "step": 8758 }, { "epoch": 0.6905005912495072, "grad_norm": 1.3608254194259644, "learning_rate": 0.0005, "loss": 1.9024231433868408, "step": 8759 }, { "epoch": 0.6905794245171463, "grad_norm": 1.5020910501480103, "learning_rate": 0.0005, "loss": 2.0049235820770264, "step": 8760 }, { "epoch": 0.6906582577847852, "grad_norm": 0.5160411596298218, "learning_rate": 0.0005, "loss": 1.8351502418518066, "step": 8761 }, { "epoch": 0.6907370910524241, "grad_norm": 0.6207759976387024, "learning_rate": 0.0005, "loss": 2.1625356674194336, "step": 8762 }, { "epoch": 0.6908159243200631, "grad_norm": 1.2470096349716187, "learning_rate": 0.0005, "loss": 1.8467692136764526, "step": 8763 }, { "epoch": 0.690894757587702, "grad_norm": 0.7710334062576294, "learning_rate": 0.0005, "loss": 1.746527910232544, "step": 8764 }, { "epoch": 0.6909735908553409, "grad_norm": 0.6390781998634338, "learning_rate": 0.0005, "loss": 1.9984776973724365, "step": 8765 }, { "epoch": 0.6910524241229798, "grad_norm": 0.8869658708572388, "learning_rate": 0.0005, "loss": 1.9123269319534302, "step": 8766 }, { "epoch": 0.6911312573906189, "grad_norm": 0.6496580243110657, "learning_rate": 0.0005, "loss": 1.9240647554397583, "step": 8767 }, { "epoch": 0.6912100906582578, "grad_norm": 0.5893158316612244, "learning_rate": 0.0005, "loss": 1.688184380531311, "step": 8768 }, { "epoch": 0.6912889239258967, "grad_norm": 0.7582855820655823, "learning_rate": 0.0005, "loss": 1.8468825817108154, "step": 8769 }, { "epoch": 0.6913677571935357, "grad_norm": 1.1813806295394897, "learning_rate": 0.0005, "loss": 2.043870210647583, "step": 8770 }, { "epoch": 0.6914465904611746, "grad_norm": 0.6080225110054016, "learning_rate": 0.0005, "loss": 1.690434217453003, "step": 8771 }, { "epoch": 0.6915254237288135, "grad_norm": 0.622336745262146, "learning_rate": 0.0005, "loss": 1.9491424560546875, "step": 8772 }, { "epoch": 0.6916042569964524, "grad_norm": 0.7329972386360168, "learning_rate": 0.0005, "loss": 1.9990015029907227, "step": 8773 }, { "epoch": 0.6916830902640915, "grad_norm": 1.1467535495758057, "learning_rate": 0.0005, "loss": 1.9906668663024902, "step": 8774 }, { "epoch": 0.6917619235317304, "grad_norm": 0.7474154829978943, "learning_rate": 0.0005, "loss": 1.7889043092727661, "step": 8775 }, { "epoch": 0.6918407567993693, "grad_norm": 0.6365373134613037, "learning_rate": 0.0005, "loss": 1.954040288925171, "step": 8776 }, { "epoch": 0.6919195900670083, "grad_norm": 0.8149355053901672, "learning_rate": 0.0005, "loss": 2.1376824378967285, "step": 8777 }, { "epoch": 0.6919984233346472, "grad_norm": 0.5443808436393738, "learning_rate": 0.0005, "loss": 1.798332929611206, "step": 8778 }, { "epoch": 0.6920772566022861, "grad_norm": 0.6061779856681824, "learning_rate": 0.0005, "loss": 1.9418522119522095, "step": 8779 }, { "epoch": 0.6921560898699252, "grad_norm": 0.6732588410377502, "learning_rate": 0.0005, "loss": 1.7834705114364624, "step": 8780 }, { "epoch": 0.6922349231375641, "grad_norm": 0.6314563155174255, "learning_rate": 0.0005, "loss": 1.7094842195510864, "step": 8781 }, { "epoch": 0.692313756405203, "grad_norm": 0.5828714370727539, "learning_rate": 0.0005, "loss": 2.0194129943847656, "step": 8782 }, { "epoch": 0.6923925896728419, "grad_norm": 0.7208450436592102, "learning_rate": 0.0005, "loss": 2.071533203125, "step": 8783 }, { "epoch": 0.6924714229404809, "grad_norm": 0.6181646585464478, "learning_rate": 0.0005, "loss": 1.9162909984588623, "step": 8784 }, { "epoch": 0.6925502562081198, "grad_norm": 0.6081052422523499, "learning_rate": 0.0005, "loss": 2.1464319229125977, "step": 8785 }, { "epoch": 0.6926290894757587, "grad_norm": 0.5444908142089844, "learning_rate": 0.0005, "loss": 1.6109473705291748, "step": 8786 }, { "epoch": 0.6927079227433978, "grad_norm": 0.5988839268684387, "learning_rate": 0.0005, "loss": 1.6247243881225586, "step": 8787 }, { "epoch": 0.6927867560110367, "grad_norm": 0.6633831858634949, "learning_rate": 0.0005, "loss": 2.0951032638549805, "step": 8788 }, { "epoch": 0.6928655892786756, "grad_norm": 0.71506667137146, "learning_rate": 0.0005, "loss": 1.9126002788543701, "step": 8789 }, { "epoch": 0.6929444225463145, "grad_norm": 1.0128569602966309, "learning_rate": 0.0005, "loss": 1.8260459899902344, "step": 8790 }, { "epoch": 0.6930232558139535, "grad_norm": 0.5909336805343628, "learning_rate": 0.0005, "loss": 1.9248937368392944, "step": 8791 }, { "epoch": 0.6931020890815924, "grad_norm": 0.6495074033737183, "learning_rate": 0.0005, "loss": 1.906332015991211, "step": 8792 }, { "epoch": 0.6931809223492313, "grad_norm": 0.6908571124076843, "learning_rate": 0.0005, "loss": 2.0524282455444336, "step": 8793 }, { "epoch": 0.6932597556168704, "grad_norm": 0.6035875678062439, "learning_rate": 0.0005, "loss": 1.8706015348434448, "step": 8794 }, { "epoch": 0.6933385888845093, "grad_norm": 0.7297312021255493, "learning_rate": 0.0005, "loss": 2.056438684463501, "step": 8795 }, { "epoch": 0.6934174221521482, "grad_norm": 0.602843165397644, "learning_rate": 0.0005, "loss": 1.805237054824829, "step": 8796 }, { "epoch": 0.6934962554197871, "grad_norm": 0.6131585240364075, "learning_rate": 0.0005, "loss": 1.917876124382019, "step": 8797 }, { "epoch": 0.6935750886874261, "grad_norm": 0.7222874164581299, "learning_rate": 0.0005, "loss": 2.0336873531341553, "step": 8798 }, { "epoch": 0.693653921955065, "grad_norm": 0.7700352668762207, "learning_rate": 0.0005, "loss": 1.9996320009231567, "step": 8799 }, { "epoch": 0.693732755222704, "grad_norm": 0.5944440364837646, "learning_rate": 0.0005, "loss": 1.9642753601074219, "step": 8800 }, { "epoch": 0.693811588490343, "grad_norm": 0.7216354608535767, "learning_rate": 0.0005, "loss": 1.9488153457641602, "step": 8801 }, { "epoch": 0.6938904217579819, "grad_norm": 0.6147357225418091, "learning_rate": 0.0005, "loss": 1.811815857887268, "step": 8802 }, { "epoch": 0.6939692550256208, "grad_norm": 0.6447511315345764, "learning_rate": 0.0005, "loss": 1.836345911026001, "step": 8803 }, { "epoch": 0.6940480882932597, "grad_norm": 0.6929115653038025, "learning_rate": 0.0005, "loss": 2.112799882888794, "step": 8804 }, { "epoch": 0.6941269215608987, "grad_norm": 1.047156572341919, "learning_rate": 0.0005, "loss": 1.954828143119812, "step": 8805 }, { "epoch": 0.6942057548285376, "grad_norm": 0.6816280484199524, "learning_rate": 0.0005, "loss": 2.087379217147827, "step": 8806 }, { "epoch": 0.6942845880961765, "grad_norm": 0.6314884424209595, "learning_rate": 0.0005, "loss": 2.0651650428771973, "step": 8807 }, { "epoch": 0.6943634213638156, "grad_norm": 0.9166470766067505, "learning_rate": 0.0005, "loss": 1.8329720497131348, "step": 8808 }, { "epoch": 0.6944422546314545, "grad_norm": 0.6472055315971375, "learning_rate": 0.0005, "loss": 1.8225724697113037, "step": 8809 }, { "epoch": 0.6945210878990934, "grad_norm": 0.7708163261413574, "learning_rate": 0.0005, "loss": 1.7629477977752686, "step": 8810 }, { "epoch": 0.6945999211667324, "grad_norm": 0.9097344279289246, "learning_rate": 0.0005, "loss": 2.0273385047912598, "step": 8811 }, { "epoch": 0.6946787544343713, "grad_norm": 0.6557726263999939, "learning_rate": 0.0005, "loss": 1.8169517517089844, "step": 8812 }, { "epoch": 0.6947575877020102, "grad_norm": 0.9683692455291748, "learning_rate": 0.0005, "loss": 2.1898257732391357, "step": 8813 }, { "epoch": 0.6948364209696491, "grad_norm": 0.6480376720428467, "learning_rate": 0.0005, "loss": 1.927323579788208, "step": 8814 }, { "epoch": 0.6949152542372882, "grad_norm": 0.6990041136741638, "learning_rate": 0.0005, "loss": 2.018920421600342, "step": 8815 }, { "epoch": 0.6949940875049271, "grad_norm": 1.0163763761520386, "learning_rate": 0.0005, "loss": 2.066743850708008, "step": 8816 }, { "epoch": 0.695072920772566, "grad_norm": 0.6704509258270264, "learning_rate": 0.0005, "loss": 1.8865684270858765, "step": 8817 }, { "epoch": 0.695151754040205, "grad_norm": 0.5613265037536621, "learning_rate": 0.0005, "loss": 1.891613483428955, "step": 8818 }, { "epoch": 0.6952305873078439, "grad_norm": 0.8103727698326111, "learning_rate": 0.0005, "loss": 2.298954963684082, "step": 8819 }, { "epoch": 0.6953094205754828, "grad_norm": 0.6856748461723328, "learning_rate": 0.0005, "loss": 1.8203219175338745, "step": 8820 }, { "epoch": 0.6953882538431218, "grad_norm": 0.7757614254951477, "learning_rate": 0.0005, "loss": 1.8291668891906738, "step": 8821 }, { "epoch": 0.6954670871107608, "grad_norm": 0.6742920279502869, "learning_rate": 0.0005, "loss": 1.5524424314498901, "step": 8822 }, { "epoch": 0.6955459203783997, "grad_norm": 0.5902875661849976, "learning_rate": 0.0005, "loss": 1.7416397333145142, "step": 8823 }, { "epoch": 0.6956247536460386, "grad_norm": 0.887919545173645, "learning_rate": 0.0005, "loss": 1.7972962856292725, "step": 8824 }, { "epoch": 0.6957035869136776, "grad_norm": 0.6723894476890564, "learning_rate": 0.0005, "loss": 2.0345139503479004, "step": 8825 }, { "epoch": 0.6957824201813165, "grad_norm": 0.8737958073616028, "learning_rate": 0.0005, "loss": 2.1705915927886963, "step": 8826 }, { "epoch": 0.6958612534489554, "grad_norm": 0.6225427389144897, "learning_rate": 0.0005, "loss": 1.6805059909820557, "step": 8827 }, { "epoch": 0.6959400867165944, "grad_norm": 0.5992065668106079, "learning_rate": 0.0005, "loss": 1.742931842803955, "step": 8828 }, { "epoch": 0.6960189199842334, "grad_norm": 0.8391276597976685, "learning_rate": 0.0005, "loss": 2.0671114921569824, "step": 8829 }, { "epoch": 0.6960977532518723, "grad_norm": 0.6642206907272339, "learning_rate": 0.0005, "loss": 1.4741382598876953, "step": 8830 }, { "epoch": 0.6961765865195112, "grad_norm": 0.8690774440765381, "learning_rate": 0.0005, "loss": 2.2198166847229004, "step": 8831 }, { "epoch": 0.6962554197871502, "grad_norm": 0.6537166237831116, "learning_rate": 0.0005, "loss": 1.7502386569976807, "step": 8832 }, { "epoch": 0.6963342530547891, "grad_norm": 0.6809489727020264, "learning_rate": 0.0005, "loss": 1.9081758260726929, "step": 8833 }, { "epoch": 0.696413086322428, "grad_norm": 0.8991751074790955, "learning_rate": 0.0005, "loss": 1.9068278074264526, "step": 8834 }, { "epoch": 0.696491919590067, "grad_norm": 0.5964224338531494, "learning_rate": 0.0005, "loss": 1.8074235916137695, "step": 8835 }, { "epoch": 0.696570752857706, "grad_norm": 0.9264720678329468, "learning_rate": 0.0005, "loss": 2.015537738800049, "step": 8836 }, { "epoch": 0.6966495861253449, "grad_norm": 0.7257335186004639, "learning_rate": 0.0005, "loss": 1.7075960636138916, "step": 8837 }, { "epoch": 0.6967284193929838, "grad_norm": 0.7112572193145752, "learning_rate": 0.0005, "loss": 2.0343925952911377, "step": 8838 }, { "epoch": 0.6968072526606228, "grad_norm": 0.625937819480896, "learning_rate": 0.0005, "loss": 1.5951740741729736, "step": 8839 }, { "epoch": 0.6968860859282617, "grad_norm": 0.610831081867218, "learning_rate": 0.0005, "loss": 1.8567607402801514, "step": 8840 }, { "epoch": 0.6969649191959006, "grad_norm": 0.6716991066932678, "learning_rate": 0.0005, "loss": 2.085606813430786, "step": 8841 }, { "epoch": 0.6970437524635397, "grad_norm": 0.6689303517341614, "learning_rate": 0.0005, "loss": 1.802391529083252, "step": 8842 }, { "epoch": 0.6971225857311786, "grad_norm": 0.8228468298912048, "learning_rate": 0.0005, "loss": 1.7591819763183594, "step": 8843 }, { "epoch": 0.6972014189988175, "grad_norm": 0.6553555130958557, "learning_rate": 0.0005, "loss": 2.0659074783325195, "step": 8844 }, { "epoch": 0.6972802522664564, "grad_norm": 0.6370279788970947, "learning_rate": 0.0005, "loss": 2.0280258655548096, "step": 8845 }, { "epoch": 0.6973590855340954, "grad_norm": 0.7418977618217468, "learning_rate": 0.0005, "loss": 1.8388203382492065, "step": 8846 }, { "epoch": 0.6974379188017343, "grad_norm": 0.5471371412277222, "learning_rate": 0.0005, "loss": 1.963726282119751, "step": 8847 }, { "epoch": 0.6975167520693732, "grad_norm": 0.6896178722381592, "learning_rate": 0.0005, "loss": 1.9478068351745605, "step": 8848 }, { "epoch": 0.6975955853370123, "grad_norm": 0.6839373707771301, "learning_rate": 0.0005, "loss": 2.008716344833374, "step": 8849 }, { "epoch": 0.6976744186046512, "grad_norm": 0.5626472234725952, "learning_rate": 0.0005, "loss": 1.9825706481933594, "step": 8850 }, { "epoch": 0.6977532518722901, "grad_norm": 0.6645696759223938, "learning_rate": 0.0005, "loss": 1.7830402851104736, "step": 8851 }, { "epoch": 0.697832085139929, "grad_norm": 0.7052987813949585, "learning_rate": 0.0005, "loss": 2.0533671379089355, "step": 8852 }, { "epoch": 0.697910918407568, "grad_norm": 0.7400262951850891, "learning_rate": 0.0005, "loss": 1.9862157106399536, "step": 8853 }, { "epoch": 0.6979897516752069, "grad_norm": 0.7310228943824768, "learning_rate": 0.0005, "loss": 1.9894590377807617, "step": 8854 }, { "epoch": 0.6980685849428458, "grad_norm": 0.6922510862350464, "learning_rate": 0.0005, "loss": 1.8558655977249146, "step": 8855 }, { "epoch": 0.6981474182104849, "grad_norm": 0.651345431804657, "learning_rate": 0.0005, "loss": 2.03564453125, "step": 8856 }, { "epoch": 0.6982262514781238, "grad_norm": 0.7591719031333923, "learning_rate": 0.0005, "loss": 1.8547189235687256, "step": 8857 }, { "epoch": 0.6983050847457627, "grad_norm": 0.7914844751358032, "learning_rate": 0.0005, "loss": 1.6635701656341553, "step": 8858 }, { "epoch": 0.6983839180134016, "grad_norm": 0.6908377408981323, "learning_rate": 0.0005, "loss": 2.0041487216949463, "step": 8859 }, { "epoch": 0.6984627512810406, "grad_norm": 0.7159443497657776, "learning_rate": 0.0005, "loss": 1.9037055969238281, "step": 8860 }, { "epoch": 0.6985415845486795, "grad_norm": 0.6419534683227539, "learning_rate": 0.0005, "loss": 1.8547627925872803, "step": 8861 }, { "epoch": 0.6986204178163185, "grad_norm": 0.5328836441040039, "learning_rate": 0.0005, "loss": 1.7287107706069946, "step": 8862 }, { "epoch": 0.6986992510839575, "grad_norm": 0.8109490871429443, "learning_rate": 0.0005, "loss": 1.8522788286209106, "step": 8863 }, { "epoch": 0.6987780843515964, "grad_norm": 0.6556570529937744, "learning_rate": 0.0005, "loss": 1.8428586721420288, "step": 8864 }, { "epoch": 0.6988569176192353, "grad_norm": 0.7076923847198486, "learning_rate": 0.0005, "loss": 2.073737144470215, "step": 8865 }, { "epoch": 0.6989357508868742, "grad_norm": 0.6405500769615173, "learning_rate": 0.0005, "loss": 2.0033891201019287, "step": 8866 }, { "epoch": 0.6990145841545132, "grad_norm": 0.6899377703666687, "learning_rate": 0.0005, "loss": 1.9976307153701782, "step": 8867 }, { "epoch": 0.6990934174221521, "grad_norm": 0.659552276134491, "learning_rate": 0.0005, "loss": 1.944791555404663, "step": 8868 }, { "epoch": 0.699172250689791, "grad_norm": 0.6443501114845276, "learning_rate": 0.0005, "loss": 1.9815188646316528, "step": 8869 }, { "epoch": 0.6992510839574301, "grad_norm": 0.6411625742912292, "learning_rate": 0.0005, "loss": 1.8599903583526611, "step": 8870 }, { "epoch": 0.699329917225069, "grad_norm": 0.6854770183563232, "learning_rate": 0.0005, "loss": 1.6329621076583862, "step": 8871 }, { "epoch": 0.6994087504927079, "grad_norm": 0.6612970232963562, "learning_rate": 0.0005, "loss": 2.0787808895111084, "step": 8872 }, { "epoch": 0.6994875837603469, "grad_norm": 0.6048598289489746, "learning_rate": 0.0005, "loss": 1.835554599761963, "step": 8873 }, { "epoch": 0.6995664170279858, "grad_norm": 0.6689056158065796, "learning_rate": 0.0005, "loss": 1.739050269126892, "step": 8874 }, { "epoch": 0.6996452502956247, "grad_norm": 0.7460542321205139, "learning_rate": 0.0005, "loss": 1.9241211414337158, "step": 8875 }, { "epoch": 0.6997240835632637, "grad_norm": 0.6228158473968506, "learning_rate": 0.0005, "loss": 1.7971150875091553, "step": 8876 }, { "epoch": 0.6998029168309027, "grad_norm": 0.7178689241409302, "learning_rate": 0.0005, "loss": 1.9449663162231445, "step": 8877 }, { "epoch": 0.6998817500985416, "grad_norm": 0.7103215456008911, "learning_rate": 0.0005, "loss": 1.5233354568481445, "step": 8878 }, { "epoch": 0.6999605833661805, "grad_norm": 0.599052906036377, "learning_rate": 0.0005, "loss": 1.7583715915679932, "step": 8879 }, { "epoch": 0.7000394166338195, "grad_norm": 0.8837611079216003, "learning_rate": 0.0005, "loss": 2.133420944213867, "step": 8880 }, { "epoch": 0.7001182499014584, "grad_norm": 0.6605651378631592, "learning_rate": 0.0005, "loss": 2.0920095443725586, "step": 8881 }, { "epoch": 0.7001970831690973, "grad_norm": 0.5974389910697937, "learning_rate": 0.0005, "loss": 1.9097704887390137, "step": 8882 }, { "epoch": 0.7002759164367363, "grad_norm": 0.880386233329773, "learning_rate": 0.0005, "loss": 2.221682071685791, "step": 8883 }, { "epoch": 0.7003547497043753, "grad_norm": 0.7637114524841309, "learning_rate": 0.0005, "loss": 1.950932502746582, "step": 8884 }, { "epoch": 0.7004335829720142, "grad_norm": 0.6110139489173889, "learning_rate": 0.0005, "loss": 1.8428800106048584, "step": 8885 }, { "epoch": 0.7005124162396531, "grad_norm": 0.9974817037582397, "learning_rate": 0.0005, "loss": 1.7648260593414307, "step": 8886 }, { "epoch": 0.7005912495072921, "grad_norm": 0.5879464149475098, "learning_rate": 0.0005, "loss": 1.7212293148040771, "step": 8887 }, { "epoch": 0.700670082774931, "grad_norm": 0.7036622166633606, "learning_rate": 0.0005, "loss": 1.9534002542495728, "step": 8888 }, { "epoch": 0.70074891604257, "grad_norm": 2.743593692779541, "learning_rate": 0.0005, "loss": 1.8608899116516113, "step": 8889 }, { "epoch": 0.7008277493102089, "grad_norm": 0.696598470211029, "learning_rate": 0.0005, "loss": 2.01849627494812, "step": 8890 }, { "epoch": 0.7009065825778479, "grad_norm": 0.7186076641082764, "learning_rate": 0.0005, "loss": 1.882962703704834, "step": 8891 }, { "epoch": 0.7009854158454868, "grad_norm": 0.6908323168754578, "learning_rate": 0.0005, "loss": 2.113779306411743, "step": 8892 }, { "epoch": 0.7010642491131257, "grad_norm": 0.704079270362854, "learning_rate": 0.0005, "loss": 2.112855911254883, "step": 8893 }, { "epoch": 0.7011430823807647, "grad_norm": 0.6859497427940369, "learning_rate": 0.0005, "loss": 2.0585694313049316, "step": 8894 }, { "epoch": 0.7012219156484036, "grad_norm": 0.8215508460998535, "learning_rate": 0.0005, "loss": 1.9298698902130127, "step": 8895 }, { "epoch": 0.7013007489160425, "grad_norm": 0.601547122001648, "learning_rate": 0.0005, "loss": 2.1829001903533936, "step": 8896 }, { "epoch": 0.7013795821836815, "grad_norm": 0.7646875381469727, "learning_rate": 0.0005, "loss": 1.7619396448135376, "step": 8897 }, { "epoch": 0.7014584154513205, "grad_norm": 0.5972719788551331, "learning_rate": 0.0005, "loss": 1.9071080684661865, "step": 8898 }, { "epoch": 0.7015372487189594, "grad_norm": 0.6510870456695557, "learning_rate": 0.0005, "loss": 1.85164213180542, "step": 8899 }, { "epoch": 0.7016160819865983, "grad_norm": 0.6338783502578735, "learning_rate": 0.0005, "loss": 1.7717474699020386, "step": 8900 }, { "epoch": 0.7016949152542373, "grad_norm": 0.6680553555488586, "learning_rate": 0.0005, "loss": 1.9742857217788696, "step": 8901 }, { "epoch": 0.7017737485218762, "grad_norm": 0.7491831183433533, "learning_rate": 0.0005, "loss": 2.067103385925293, "step": 8902 }, { "epoch": 0.7018525817895152, "grad_norm": 0.7306076884269714, "learning_rate": 0.0005, "loss": 1.67429518699646, "step": 8903 }, { "epoch": 0.7019314150571542, "grad_norm": 0.6238167881965637, "learning_rate": 0.0005, "loss": 1.818739414215088, "step": 8904 }, { "epoch": 0.7020102483247931, "grad_norm": 0.7873718738555908, "learning_rate": 0.0005, "loss": 1.792830228805542, "step": 8905 }, { "epoch": 0.702089081592432, "grad_norm": 0.6282179951667786, "learning_rate": 0.0005, "loss": 1.5388867855072021, "step": 8906 }, { "epoch": 0.7021679148600709, "grad_norm": 0.7436443567276001, "learning_rate": 0.0005, "loss": 2.24519681930542, "step": 8907 }, { "epoch": 0.7022467481277099, "grad_norm": 0.7949690818786621, "learning_rate": 0.0005, "loss": 2.2097318172454834, "step": 8908 }, { "epoch": 0.7023255813953488, "grad_norm": 0.7024117708206177, "learning_rate": 0.0005, "loss": 1.6218504905700684, "step": 8909 }, { "epoch": 0.7024044146629878, "grad_norm": 1.955763816833496, "learning_rate": 0.0005, "loss": 1.894101858139038, "step": 8910 }, { "epoch": 0.7024832479306268, "grad_norm": 0.6276135444641113, "learning_rate": 0.0005, "loss": 1.7213796377182007, "step": 8911 }, { "epoch": 0.7025620811982657, "grad_norm": 0.6724201440811157, "learning_rate": 0.0005, "loss": 1.8024017810821533, "step": 8912 }, { "epoch": 0.7026409144659046, "grad_norm": 0.5671404600143433, "learning_rate": 0.0005, "loss": 2.1921043395996094, "step": 8913 }, { "epoch": 0.7027197477335435, "grad_norm": 0.7376483082771301, "learning_rate": 0.0005, "loss": 1.8912091255187988, "step": 8914 }, { "epoch": 0.7027985810011825, "grad_norm": 0.6744471192359924, "learning_rate": 0.0005, "loss": 1.805349588394165, "step": 8915 }, { "epoch": 0.7028774142688214, "grad_norm": 0.658440113067627, "learning_rate": 0.0005, "loss": 1.885523796081543, "step": 8916 }, { "epoch": 0.7029562475364604, "grad_norm": 0.7245202660560608, "learning_rate": 0.0005, "loss": 1.8587620258331299, "step": 8917 }, { "epoch": 0.7030350808040994, "grad_norm": 0.5843244194984436, "learning_rate": 0.0005, "loss": 1.9477665424346924, "step": 8918 }, { "epoch": 0.7031139140717383, "grad_norm": 0.6567580699920654, "learning_rate": 0.0005, "loss": 1.9805423021316528, "step": 8919 }, { "epoch": 0.7031927473393772, "grad_norm": 0.6032101511955261, "learning_rate": 0.0005, "loss": 2.016153573989868, "step": 8920 }, { "epoch": 0.7032715806070161, "grad_norm": 0.6413899064064026, "learning_rate": 0.0005, "loss": 1.8511848449707031, "step": 8921 }, { "epoch": 0.7033504138746551, "grad_norm": 0.6164105534553528, "learning_rate": 0.0005, "loss": 1.6141935586929321, "step": 8922 }, { "epoch": 0.703429247142294, "grad_norm": 0.5337650775909424, "learning_rate": 0.0005, "loss": 1.888688325881958, "step": 8923 }, { "epoch": 0.703508080409933, "grad_norm": 0.6208983659744263, "learning_rate": 0.0005, "loss": 1.9848898649215698, "step": 8924 }, { "epoch": 0.703586913677572, "grad_norm": 0.6693817973136902, "learning_rate": 0.0005, "loss": 1.8094855546951294, "step": 8925 }, { "epoch": 0.7036657469452109, "grad_norm": 0.6922783851623535, "learning_rate": 0.0005, "loss": 1.8455926179885864, "step": 8926 }, { "epoch": 0.7037445802128498, "grad_norm": 0.7463017702102661, "learning_rate": 0.0005, "loss": 1.929701566696167, "step": 8927 }, { "epoch": 0.7038234134804887, "grad_norm": 0.709496259689331, "learning_rate": 0.0005, "loss": 1.929421067237854, "step": 8928 }, { "epoch": 0.7039022467481277, "grad_norm": 0.5867365002632141, "learning_rate": 0.0005, "loss": 1.803855299949646, "step": 8929 }, { "epoch": 0.7039810800157666, "grad_norm": 0.7347686290740967, "learning_rate": 0.0005, "loss": 1.8440940380096436, "step": 8930 }, { "epoch": 0.7040599132834056, "grad_norm": 0.5698645710945129, "learning_rate": 0.0005, "loss": 1.9595873355865479, "step": 8931 }, { "epoch": 0.7041387465510446, "grad_norm": 0.6159088015556335, "learning_rate": 0.0005, "loss": 1.8518974781036377, "step": 8932 }, { "epoch": 0.7042175798186835, "grad_norm": 0.7080986499786377, "learning_rate": 0.0005, "loss": 1.8560543060302734, "step": 8933 }, { "epoch": 0.7042964130863224, "grad_norm": 0.829309344291687, "learning_rate": 0.0005, "loss": 1.9242587089538574, "step": 8934 }, { "epoch": 0.7043752463539614, "grad_norm": 0.5033401846885681, "learning_rate": 0.0005, "loss": 1.5777276754379272, "step": 8935 }, { "epoch": 0.7044540796216003, "grad_norm": 0.6764448285102844, "learning_rate": 0.0005, "loss": 1.9208943843841553, "step": 8936 }, { "epoch": 0.7045329128892392, "grad_norm": 0.8254282474517822, "learning_rate": 0.0005, "loss": 1.6329176425933838, "step": 8937 }, { "epoch": 0.7046117461568782, "grad_norm": 0.6970812082290649, "learning_rate": 0.0005, "loss": 1.889047622680664, "step": 8938 }, { "epoch": 0.7046905794245172, "grad_norm": 0.8048964738845825, "learning_rate": 0.0005, "loss": 2.0396385192871094, "step": 8939 }, { "epoch": 0.7047694126921561, "grad_norm": 0.8855869174003601, "learning_rate": 0.0005, "loss": 2.0190625190734863, "step": 8940 }, { "epoch": 0.704848245959795, "grad_norm": 0.6994805932044983, "learning_rate": 0.0005, "loss": 2.080789089202881, "step": 8941 }, { "epoch": 0.704927079227434, "grad_norm": 0.8560994863510132, "learning_rate": 0.0005, "loss": 2.2347683906555176, "step": 8942 }, { "epoch": 0.7050059124950729, "grad_norm": 0.6777955889701843, "learning_rate": 0.0005, "loss": 1.7967442274093628, "step": 8943 }, { "epoch": 0.7050847457627119, "grad_norm": 0.6978964805603027, "learning_rate": 0.0005, "loss": 1.894280195236206, "step": 8944 }, { "epoch": 0.7051635790303508, "grad_norm": 0.6401827931404114, "learning_rate": 0.0005, "loss": 1.744279384613037, "step": 8945 }, { "epoch": 0.7052424122979898, "grad_norm": 0.620496392250061, "learning_rate": 0.0005, "loss": 1.809579610824585, "step": 8946 }, { "epoch": 0.7053212455656287, "grad_norm": 0.6711276769638062, "learning_rate": 0.0005, "loss": 1.6342244148254395, "step": 8947 }, { "epoch": 0.7054000788332676, "grad_norm": 0.7322826385498047, "learning_rate": 0.0005, "loss": 1.7901698350906372, "step": 8948 }, { "epoch": 0.7054789121009066, "grad_norm": 0.64076828956604, "learning_rate": 0.0005, "loss": 1.994867205619812, "step": 8949 }, { "epoch": 0.7055577453685455, "grad_norm": 0.6009416580200195, "learning_rate": 0.0005, "loss": 1.9284172058105469, "step": 8950 }, { "epoch": 0.7056365786361845, "grad_norm": 0.6763923168182373, "learning_rate": 0.0005, "loss": 1.9182308912277222, "step": 8951 }, { "epoch": 0.7057154119038234, "grad_norm": 0.6649597883224487, "learning_rate": 0.0005, "loss": 1.6325997114181519, "step": 8952 }, { "epoch": 0.7057942451714624, "grad_norm": 0.6120514869689941, "learning_rate": 0.0005, "loss": 1.9017194509506226, "step": 8953 }, { "epoch": 0.7058730784391013, "grad_norm": 0.8646584749221802, "learning_rate": 0.0005, "loss": 1.7716782093048096, "step": 8954 }, { "epoch": 0.7059519117067402, "grad_norm": 0.7548441886901855, "learning_rate": 0.0005, "loss": 1.93280029296875, "step": 8955 }, { "epoch": 0.7060307449743792, "grad_norm": 0.49123528599739075, "learning_rate": 0.0005, "loss": 1.5928456783294678, "step": 8956 }, { "epoch": 0.7061095782420181, "grad_norm": 0.648051917552948, "learning_rate": 0.0005, "loss": 1.9972829818725586, "step": 8957 }, { "epoch": 0.706188411509657, "grad_norm": 0.5564116835594177, "learning_rate": 0.0005, "loss": 1.9623188972473145, "step": 8958 }, { "epoch": 0.706267244777296, "grad_norm": 0.6690625548362732, "learning_rate": 0.0005, "loss": 1.9509520530700684, "step": 8959 }, { "epoch": 0.706346078044935, "grad_norm": 0.629300594329834, "learning_rate": 0.0005, "loss": 1.786903738975525, "step": 8960 }, { "epoch": 0.7064249113125739, "grad_norm": 0.7845571041107178, "learning_rate": 0.0005, "loss": 2.014260768890381, "step": 8961 }, { "epoch": 0.7065037445802128, "grad_norm": 0.7331199645996094, "learning_rate": 0.0005, "loss": 1.8144288063049316, "step": 8962 }, { "epoch": 0.7065825778478518, "grad_norm": 0.6693704128265381, "learning_rate": 0.0005, "loss": 1.8344831466674805, "step": 8963 }, { "epoch": 0.7066614111154907, "grad_norm": 0.8602104783058167, "learning_rate": 0.0005, "loss": 1.96268892288208, "step": 8964 }, { "epoch": 0.7067402443831297, "grad_norm": 0.7730624079704285, "learning_rate": 0.0005, "loss": 1.916576862335205, "step": 8965 }, { "epoch": 0.7068190776507687, "grad_norm": 0.548543393611908, "learning_rate": 0.0005, "loss": 1.5689666271209717, "step": 8966 }, { "epoch": 0.7068979109184076, "grad_norm": 0.6867972612380981, "learning_rate": 0.0005, "loss": 1.692934513092041, "step": 8967 }, { "epoch": 0.7069767441860465, "grad_norm": 0.6845465302467346, "learning_rate": 0.0005, "loss": 1.791621208190918, "step": 8968 }, { "epoch": 0.7070555774536854, "grad_norm": 0.701603889465332, "learning_rate": 0.0005, "loss": 1.8460744619369507, "step": 8969 }, { "epoch": 0.7071344107213244, "grad_norm": 0.7976036667823792, "learning_rate": 0.0005, "loss": 1.8656870126724243, "step": 8970 }, { "epoch": 0.7072132439889633, "grad_norm": 0.6437585949897766, "learning_rate": 0.0005, "loss": 1.9114229679107666, "step": 8971 }, { "epoch": 0.7072920772566023, "grad_norm": 0.6819480061531067, "learning_rate": 0.0005, "loss": 1.9152684211730957, "step": 8972 }, { "epoch": 0.7073709105242413, "grad_norm": 0.5911652445793152, "learning_rate": 0.0005, "loss": 1.8420977592468262, "step": 8973 }, { "epoch": 0.7074497437918802, "grad_norm": 0.6912791132926941, "learning_rate": 0.0005, "loss": 1.9617009162902832, "step": 8974 }, { "epoch": 0.7075285770595191, "grad_norm": 0.6140013933181763, "learning_rate": 0.0005, "loss": 1.932973861694336, "step": 8975 }, { "epoch": 0.707607410327158, "grad_norm": 1.2095928192138672, "learning_rate": 0.0005, "loss": 2.1465470790863037, "step": 8976 }, { "epoch": 0.707686243594797, "grad_norm": 0.6085275411605835, "learning_rate": 0.0005, "loss": 1.6907477378845215, "step": 8977 }, { "epoch": 0.707765076862436, "grad_norm": 3.2843849658966064, "learning_rate": 0.0005, "loss": 1.8696426153182983, "step": 8978 }, { "epoch": 0.7078439101300749, "grad_norm": 0.6264610290527344, "learning_rate": 0.0005, "loss": 1.7803382873535156, "step": 8979 }, { "epoch": 0.7079227433977139, "grad_norm": 0.6566851735115051, "learning_rate": 0.0005, "loss": 1.811126947402954, "step": 8980 }, { "epoch": 0.7080015766653528, "grad_norm": 0.7353314757347107, "learning_rate": 0.0005, "loss": 1.818058729171753, "step": 8981 }, { "epoch": 0.7080804099329917, "grad_norm": 0.6187954545021057, "learning_rate": 0.0005, "loss": 1.7035441398620605, "step": 8982 }, { "epoch": 0.7081592432006306, "grad_norm": 0.5694604516029358, "learning_rate": 0.0005, "loss": 1.8042980432510376, "step": 8983 }, { "epoch": 0.7082380764682696, "grad_norm": 0.7988401651382446, "learning_rate": 0.0005, "loss": 2.2723708152770996, "step": 8984 }, { "epoch": 0.7083169097359086, "grad_norm": 0.7042355537414551, "learning_rate": 0.0005, "loss": 2.02681040763855, "step": 8985 }, { "epoch": 0.7083957430035475, "grad_norm": 0.7207702398300171, "learning_rate": 0.0005, "loss": 2.161247491836548, "step": 8986 }, { "epoch": 0.7084745762711865, "grad_norm": 0.7216571569442749, "learning_rate": 0.0005, "loss": 1.8732044696807861, "step": 8987 }, { "epoch": 0.7085534095388254, "grad_norm": 0.736642599105835, "learning_rate": 0.0005, "loss": 1.937469244003296, "step": 8988 }, { "epoch": 0.7086322428064643, "grad_norm": 0.8041325211524963, "learning_rate": 0.0005, "loss": 2.1770613193511963, "step": 8989 }, { "epoch": 0.7087110760741032, "grad_norm": 0.5535585284233093, "learning_rate": 0.0005, "loss": 1.8117706775665283, "step": 8990 }, { "epoch": 0.7087899093417422, "grad_norm": 0.7764073610305786, "learning_rate": 0.0005, "loss": 1.999823808670044, "step": 8991 }, { "epoch": 0.7088687426093812, "grad_norm": 0.6687220335006714, "learning_rate": 0.0005, "loss": 2.0056843757629395, "step": 8992 }, { "epoch": 0.7089475758770201, "grad_norm": 0.7479552030563354, "learning_rate": 0.0005, "loss": 2.1083922386169434, "step": 8993 }, { "epoch": 0.7090264091446591, "grad_norm": 0.5812981128692627, "learning_rate": 0.0005, "loss": 1.6796610355377197, "step": 8994 }, { "epoch": 0.709105242412298, "grad_norm": 0.7208630442619324, "learning_rate": 0.0005, "loss": 2.0713632106781006, "step": 8995 }, { "epoch": 0.7091840756799369, "grad_norm": 0.6374859809875488, "learning_rate": 0.0005, "loss": 2.0882627964019775, "step": 8996 }, { "epoch": 0.7092629089475759, "grad_norm": 0.6082772612571716, "learning_rate": 0.0005, "loss": 1.9751067161560059, "step": 8997 }, { "epoch": 0.7093417422152148, "grad_norm": 0.7387077212333679, "learning_rate": 0.0005, "loss": 2.322505474090576, "step": 8998 }, { "epoch": 0.7094205754828538, "grad_norm": 0.7384523153305054, "learning_rate": 0.0005, "loss": 1.9688851833343506, "step": 8999 }, { "epoch": 0.7094994087504927, "grad_norm": 0.6184685230255127, "learning_rate": 0.0005, "loss": 2.0130274295806885, "step": 9000 }, { "epoch": 0.7095782420181317, "grad_norm": 0.6008473038673401, "learning_rate": 0.0005, "loss": 1.8573012351989746, "step": 9001 }, { "epoch": 0.7096570752857706, "grad_norm": 0.8580577969551086, "learning_rate": 0.0005, "loss": 1.998940348625183, "step": 9002 }, { "epoch": 0.7097359085534095, "grad_norm": 0.817044734954834, "learning_rate": 0.0005, "loss": 1.8419322967529297, "step": 9003 }, { "epoch": 0.7098147418210485, "grad_norm": 0.660850465297699, "learning_rate": 0.0005, "loss": 1.815832495689392, "step": 9004 }, { "epoch": 0.7098935750886874, "grad_norm": 0.7558153867721558, "learning_rate": 0.0005, "loss": 2.0981252193450928, "step": 9005 }, { "epoch": 0.7099724083563264, "grad_norm": 0.7021241784095764, "learning_rate": 0.0005, "loss": 2.251403331756592, "step": 9006 }, { "epoch": 0.7100512416239653, "grad_norm": 0.9511860013008118, "learning_rate": 0.0005, "loss": 1.9171643257141113, "step": 9007 }, { "epoch": 0.7101300748916043, "grad_norm": 0.7848994135856628, "learning_rate": 0.0005, "loss": 2.154378652572632, "step": 9008 }, { "epoch": 0.7102089081592432, "grad_norm": 0.6903116703033447, "learning_rate": 0.0005, "loss": 1.7750132083892822, "step": 9009 }, { "epoch": 0.7102877414268821, "grad_norm": 0.9027793407440186, "learning_rate": 0.0005, "loss": 1.833825707435608, "step": 9010 }, { "epoch": 0.7103665746945211, "grad_norm": 0.7508420348167419, "learning_rate": 0.0005, "loss": 1.98518967628479, "step": 9011 }, { "epoch": 0.71044540796216, "grad_norm": 0.6326285004615784, "learning_rate": 0.0005, "loss": 1.796751856803894, "step": 9012 }, { "epoch": 0.710524241229799, "grad_norm": 1.2111929655075073, "learning_rate": 0.0005, "loss": 2.2019147872924805, "step": 9013 }, { "epoch": 0.7106030744974379, "grad_norm": 0.5952425599098206, "learning_rate": 0.0005, "loss": 1.7575129270553589, "step": 9014 }, { "epoch": 0.7106819077650769, "grad_norm": 0.7393332123756409, "learning_rate": 0.0005, "loss": 1.8639633655548096, "step": 9015 }, { "epoch": 0.7107607410327158, "grad_norm": 0.7062744498252869, "learning_rate": 0.0005, "loss": 1.6392985582351685, "step": 9016 }, { "epoch": 0.7108395743003547, "grad_norm": 0.8886156678199768, "learning_rate": 0.0005, "loss": 1.7958933115005493, "step": 9017 }, { "epoch": 0.7109184075679937, "grad_norm": 0.6272728443145752, "learning_rate": 0.0005, "loss": 1.9258089065551758, "step": 9018 }, { "epoch": 0.7109972408356326, "grad_norm": 0.8407267332077026, "learning_rate": 0.0005, "loss": 2.0561470985412598, "step": 9019 }, { "epoch": 0.7110760741032716, "grad_norm": 0.6458300352096558, "learning_rate": 0.0005, "loss": 1.7025678157806396, "step": 9020 }, { "epoch": 0.7111549073709105, "grad_norm": 0.6611862778663635, "learning_rate": 0.0005, "loss": 2.114241600036621, "step": 9021 }, { "epoch": 0.7112337406385495, "grad_norm": 0.7420319318771362, "learning_rate": 0.0005, "loss": 1.8447099924087524, "step": 9022 }, { "epoch": 0.7113125739061884, "grad_norm": 0.7053461670875549, "learning_rate": 0.0005, "loss": 2.0296123027801514, "step": 9023 }, { "epoch": 0.7113914071738273, "grad_norm": 0.8033475875854492, "learning_rate": 0.0005, "loss": 1.826629877090454, "step": 9024 }, { "epoch": 0.7114702404414663, "grad_norm": 0.7199105620384216, "learning_rate": 0.0005, "loss": 1.984470248222351, "step": 9025 }, { "epoch": 0.7115490737091053, "grad_norm": 1.1110926866531372, "learning_rate": 0.0005, "loss": 1.8160288333892822, "step": 9026 }, { "epoch": 0.7116279069767442, "grad_norm": 0.8286452889442444, "learning_rate": 0.0005, "loss": 1.7809720039367676, "step": 9027 }, { "epoch": 0.7117067402443832, "grad_norm": 0.7707359194755554, "learning_rate": 0.0005, "loss": 2.0334343910217285, "step": 9028 }, { "epoch": 0.7117855735120221, "grad_norm": 1.2130154371261597, "learning_rate": 0.0005, "loss": 1.815622329711914, "step": 9029 }, { "epoch": 0.711864406779661, "grad_norm": 0.7079350352287292, "learning_rate": 0.0005, "loss": 1.9198042154312134, "step": 9030 }, { "epoch": 0.7119432400472999, "grad_norm": 0.7212936878204346, "learning_rate": 0.0005, "loss": 2.1123600006103516, "step": 9031 }, { "epoch": 0.7120220733149389, "grad_norm": 0.6851325035095215, "learning_rate": 0.0005, "loss": 1.6777229309082031, "step": 9032 }, { "epoch": 0.7121009065825779, "grad_norm": 0.6283901333808899, "learning_rate": 0.0005, "loss": 1.6806309223175049, "step": 9033 }, { "epoch": 0.7121797398502168, "grad_norm": 0.6485600471496582, "learning_rate": 0.0005, "loss": 1.7046234607696533, "step": 9034 }, { "epoch": 0.7122585731178558, "grad_norm": 0.6039959788322449, "learning_rate": 0.0005, "loss": 1.93104887008667, "step": 9035 }, { "epoch": 0.7123374063854947, "grad_norm": 0.6003027558326721, "learning_rate": 0.0005, "loss": 1.771875023841858, "step": 9036 }, { "epoch": 0.7124162396531336, "grad_norm": 0.6216831207275391, "learning_rate": 0.0005, "loss": 2.159397840499878, "step": 9037 }, { "epoch": 0.7124950729207725, "grad_norm": 0.7368359565734863, "learning_rate": 0.0005, "loss": 1.8500263690948486, "step": 9038 }, { "epoch": 0.7125739061884115, "grad_norm": 1.1045726537704468, "learning_rate": 0.0005, "loss": 1.983144998550415, "step": 9039 }, { "epoch": 0.7126527394560505, "grad_norm": 0.6328017115592957, "learning_rate": 0.0005, "loss": 2.0468149185180664, "step": 9040 }, { "epoch": 0.7127315727236894, "grad_norm": 2.1781396865844727, "learning_rate": 0.0005, "loss": 1.7822505235671997, "step": 9041 }, { "epoch": 0.7128104059913284, "grad_norm": 0.6866794228553772, "learning_rate": 0.0005, "loss": 2.02239727973938, "step": 9042 }, { "epoch": 0.7128892392589673, "grad_norm": 0.7871918082237244, "learning_rate": 0.0005, "loss": 1.7860143184661865, "step": 9043 }, { "epoch": 0.7129680725266062, "grad_norm": 0.8053306937217712, "learning_rate": 0.0005, "loss": 1.8629136085510254, "step": 9044 }, { "epoch": 0.7130469057942451, "grad_norm": 1.2055312395095825, "learning_rate": 0.0005, "loss": 1.9280567169189453, "step": 9045 }, { "epoch": 0.7131257390618841, "grad_norm": 0.6685540676116943, "learning_rate": 0.0005, "loss": 1.8358899354934692, "step": 9046 }, { "epoch": 0.7132045723295231, "grad_norm": 0.5850504040718079, "learning_rate": 0.0005, "loss": 1.9655344486236572, "step": 9047 }, { "epoch": 0.713283405597162, "grad_norm": 0.6444545984268188, "learning_rate": 0.0005, "loss": 1.5848071575164795, "step": 9048 }, { "epoch": 0.713362238864801, "grad_norm": 0.5720649361610413, "learning_rate": 0.0005, "loss": 1.8529225587844849, "step": 9049 }, { "epoch": 0.7134410721324399, "grad_norm": 0.6774404644966125, "learning_rate": 0.0005, "loss": 1.9637646675109863, "step": 9050 }, { "epoch": 0.7135199054000788, "grad_norm": 0.6250064373016357, "learning_rate": 0.0005, "loss": 1.6759086847305298, "step": 9051 }, { "epoch": 0.7135987386677177, "grad_norm": 0.8104651570320129, "learning_rate": 0.0005, "loss": 1.804384469985962, "step": 9052 }, { "epoch": 0.7136775719353567, "grad_norm": 0.7084808945655823, "learning_rate": 0.0005, "loss": 1.8040952682495117, "step": 9053 }, { "epoch": 0.7137564052029957, "grad_norm": 0.7093406319618225, "learning_rate": 0.0005, "loss": 2.186579704284668, "step": 9054 }, { "epoch": 0.7138352384706346, "grad_norm": 0.6090840697288513, "learning_rate": 0.0005, "loss": 1.8878775835037231, "step": 9055 }, { "epoch": 0.7139140717382736, "grad_norm": 0.5364607572555542, "learning_rate": 0.0005, "loss": 1.8202768564224243, "step": 9056 }, { "epoch": 0.7139929050059125, "grad_norm": 0.6966667771339417, "learning_rate": 0.0005, "loss": 2.162843942642212, "step": 9057 }, { "epoch": 0.7140717382735514, "grad_norm": 0.9087989330291748, "learning_rate": 0.0005, "loss": 1.8430137634277344, "step": 9058 }, { "epoch": 0.7141505715411904, "grad_norm": 0.6638909578323364, "learning_rate": 0.0005, "loss": 1.9999542236328125, "step": 9059 }, { "epoch": 0.7142294048088293, "grad_norm": 0.6052001118659973, "learning_rate": 0.0005, "loss": 1.7730785608291626, "step": 9060 }, { "epoch": 0.7143082380764683, "grad_norm": 0.6514628529548645, "learning_rate": 0.0005, "loss": 1.900782823562622, "step": 9061 }, { "epoch": 0.7143870713441072, "grad_norm": 0.7376337647438049, "learning_rate": 0.0005, "loss": 1.8672958612442017, "step": 9062 }, { "epoch": 0.7144659046117462, "grad_norm": 0.7503199577331543, "learning_rate": 0.0005, "loss": 1.5739600658416748, "step": 9063 }, { "epoch": 0.7145447378793851, "grad_norm": 0.8191638588905334, "learning_rate": 0.0005, "loss": 1.9507626295089722, "step": 9064 }, { "epoch": 0.714623571147024, "grad_norm": 0.6567510962486267, "learning_rate": 0.0005, "loss": 2.013500690460205, "step": 9065 }, { "epoch": 0.714702404414663, "grad_norm": 0.7085049152374268, "learning_rate": 0.0005, "loss": 1.5414458513259888, "step": 9066 }, { "epoch": 0.714781237682302, "grad_norm": 0.6356458067893982, "learning_rate": 0.0005, "loss": 1.8434696197509766, "step": 9067 }, { "epoch": 0.7148600709499409, "grad_norm": 0.7003029584884644, "learning_rate": 0.0005, "loss": 2.0682296752929688, "step": 9068 }, { "epoch": 0.7149389042175798, "grad_norm": 0.618234395980835, "learning_rate": 0.0005, "loss": 1.6952121257781982, "step": 9069 }, { "epoch": 0.7150177374852188, "grad_norm": 0.5963043570518494, "learning_rate": 0.0005, "loss": 1.7880938053131104, "step": 9070 }, { "epoch": 0.7150965707528577, "grad_norm": 0.678164541721344, "learning_rate": 0.0005, "loss": 2.0060949325561523, "step": 9071 }, { "epoch": 0.7151754040204966, "grad_norm": 0.6117436289787292, "learning_rate": 0.0005, "loss": 1.7452526092529297, "step": 9072 }, { "epoch": 0.7152542372881356, "grad_norm": 0.8138479590415955, "learning_rate": 0.0005, "loss": 2.1544246673583984, "step": 9073 }, { "epoch": 0.7153330705557746, "grad_norm": 0.7130839228630066, "learning_rate": 0.0005, "loss": 1.8422306776046753, "step": 9074 }, { "epoch": 0.7154119038234135, "grad_norm": 0.6775808930397034, "learning_rate": 0.0005, "loss": 1.638984203338623, "step": 9075 }, { "epoch": 0.7154907370910524, "grad_norm": 0.658139169216156, "learning_rate": 0.0005, "loss": 1.769113302230835, "step": 9076 }, { "epoch": 0.7155695703586914, "grad_norm": 0.5907267332077026, "learning_rate": 0.0005, "loss": 1.746962547302246, "step": 9077 }, { "epoch": 0.7156484036263303, "grad_norm": 0.7602522373199463, "learning_rate": 0.0005, "loss": 1.9332001209259033, "step": 9078 }, { "epoch": 0.7157272368939692, "grad_norm": 0.7584087252616882, "learning_rate": 0.0005, "loss": 1.9297442436218262, "step": 9079 }, { "epoch": 0.7158060701616082, "grad_norm": 0.512758731842041, "learning_rate": 0.0005, "loss": 1.6903424263000488, "step": 9080 }, { "epoch": 0.7158849034292472, "grad_norm": 0.5786159634590149, "learning_rate": 0.0005, "loss": 1.7900322675704956, "step": 9081 }, { "epoch": 0.7159637366968861, "grad_norm": 0.6116799712181091, "learning_rate": 0.0005, "loss": 1.7503628730773926, "step": 9082 }, { "epoch": 0.716042569964525, "grad_norm": 0.7252975702285767, "learning_rate": 0.0005, "loss": 1.960749864578247, "step": 9083 }, { "epoch": 0.716121403232164, "grad_norm": 0.8880739808082581, "learning_rate": 0.0005, "loss": 1.921026349067688, "step": 9084 }, { "epoch": 0.7162002364998029, "grad_norm": 0.799583911895752, "learning_rate": 0.0005, "loss": 1.9269559383392334, "step": 9085 }, { "epoch": 0.7162790697674418, "grad_norm": 0.5507819652557373, "learning_rate": 0.0005, "loss": 1.7716878652572632, "step": 9086 }, { "epoch": 0.7163579030350808, "grad_norm": 0.7330897450447083, "learning_rate": 0.0005, "loss": 2.0707790851593018, "step": 9087 }, { "epoch": 0.7164367363027198, "grad_norm": 0.6396144032478333, "learning_rate": 0.0005, "loss": 1.7530694007873535, "step": 9088 }, { "epoch": 0.7165155695703587, "grad_norm": 0.6336062550544739, "learning_rate": 0.0005, "loss": 1.8830153942108154, "step": 9089 }, { "epoch": 0.7165944028379977, "grad_norm": 0.44931143522262573, "learning_rate": 0.0005, "loss": 1.5704193115234375, "step": 9090 }, { "epoch": 0.7166732361056366, "grad_norm": 0.6877802610397339, "learning_rate": 0.0005, "loss": 2.012302875518799, "step": 9091 }, { "epoch": 0.7167520693732755, "grad_norm": 0.8831125497817993, "learning_rate": 0.0005, "loss": 2.0895297527313232, "step": 9092 }, { "epoch": 0.7168309026409144, "grad_norm": 0.8000476956367493, "learning_rate": 0.0005, "loss": 1.669079065322876, "step": 9093 }, { "epoch": 0.7169097359085534, "grad_norm": 0.5975538492202759, "learning_rate": 0.0005, "loss": 1.7508877515792847, "step": 9094 }, { "epoch": 0.7169885691761924, "grad_norm": 0.7545892596244812, "learning_rate": 0.0005, "loss": 1.9449269771575928, "step": 9095 }, { "epoch": 0.7170674024438313, "grad_norm": 0.7806311845779419, "learning_rate": 0.0005, "loss": 1.9924033880233765, "step": 9096 }, { "epoch": 0.7171462357114703, "grad_norm": 0.7010353803634644, "learning_rate": 0.0005, "loss": 2.0681681632995605, "step": 9097 }, { "epoch": 0.7172250689791092, "grad_norm": 0.8377106189727783, "learning_rate": 0.0005, "loss": 1.934471607208252, "step": 9098 }, { "epoch": 0.7173039022467481, "grad_norm": 0.5442004203796387, "learning_rate": 0.0005, "loss": 1.6628625392913818, "step": 9099 }, { "epoch": 0.717382735514387, "grad_norm": 0.6691781878471375, "learning_rate": 0.0005, "loss": 1.8320720195770264, "step": 9100 }, { "epoch": 0.717461568782026, "grad_norm": 0.5681731700897217, "learning_rate": 0.0005, "loss": 1.8707175254821777, "step": 9101 }, { "epoch": 0.717540402049665, "grad_norm": 0.8807938098907471, "learning_rate": 0.0005, "loss": 1.7973625659942627, "step": 9102 }, { "epoch": 0.7176192353173039, "grad_norm": 0.7666869759559631, "learning_rate": 0.0005, "loss": 2.13889741897583, "step": 9103 }, { "epoch": 0.7176980685849429, "grad_norm": 0.8859459161758423, "learning_rate": 0.0005, "loss": 2.0140419006347656, "step": 9104 }, { "epoch": 0.7177769018525818, "grad_norm": 0.7270392775535583, "learning_rate": 0.0005, "loss": 1.9066739082336426, "step": 9105 }, { "epoch": 0.7178557351202207, "grad_norm": 0.7113624215126038, "learning_rate": 0.0005, "loss": 1.8290836811065674, "step": 9106 }, { "epoch": 0.7179345683878596, "grad_norm": 0.7106273174285889, "learning_rate": 0.0005, "loss": 2.003389835357666, "step": 9107 }, { "epoch": 0.7180134016554987, "grad_norm": 0.7051187753677368, "learning_rate": 0.0005, "loss": 1.7411210536956787, "step": 9108 }, { "epoch": 0.7180922349231376, "grad_norm": 1.1681753396987915, "learning_rate": 0.0005, "loss": 2.0528974533081055, "step": 9109 }, { "epoch": 0.7181710681907765, "grad_norm": 0.6289944052696228, "learning_rate": 0.0005, "loss": 1.9099452495574951, "step": 9110 }, { "epoch": 0.7182499014584155, "grad_norm": 0.7109825015068054, "learning_rate": 0.0005, "loss": 1.9417181015014648, "step": 9111 }, { "epoch": 0.7183287347260544, "grad_norm": 0.6861571073532104, "learning_rate": 0.0005, "loss": 1.8482041358947754, "step": 9112 }, { "epoch": 0.7184075679936933, "grad_norm": 0.5702254772186279, "learning_rate": 0.0005, "loss": 1.7330718040466309, "step": 9113 }, { "epoch": 0.7184864012613322, "grad_norm": 0.5809434652328491, "learning_rate": 0.0005, "loss": 1.9788930416107178, "step": 9114 }, { "epoch": 0.7185652345289713, "grad_norm": 0.9661300778388977, "learning_rate": 0.0005, "loss": 2.213404893875122, "step": 9115 }, { "epoch": 0.7186440677966102, "grad_norm": 0.7522345185279846, "learning_rate": 0.0005, "loss": 1.8224681615829468, "step": 9116 }, { "epoch": 0.7187229010642491, "grad_norm": 0.7236948013305664, "learning_rate": 0.0005, "loss": 1.8953291177749634, "step": 9117 }, { "epoch": 0.7188017343318881, "grad_norm": 0.6138215661048889, "learning_rate": 0.0005, "loss": 1.838881015777588, "step": 9118 }, { "epoch": 0.718880567599527, "grad_norm": 0.7445434331893921, "learning_rate": 0.0005, "loss": 2.188218593597412, "step": 9119 }, { "epoch": 0.7189594008671659, "grad_norm": 0.7342599034309387, "learning_rate": 0.0005, "loss": 2.2013931274414062, "step": 9120 }, { "epoch": 0.7190382341348048, "grad_norm": 0.6173650026321411, "learning_rate": 0.0005, "loss": 1.9023351669311523, "step": 9121 }, { "epoch": 0.7191170674024439, "grad_norm": 0.6578730344772339, "learning_rate": 0.0005, "loss": 1.9832087755203247, "step": 9122 }, { "epoch": 0.7191959006700828, "grad_norm": 0.8837140202522278, "learning_rate": 0.0005, "loss": 2.318786382675171, "step": 9123 }, { "epoch": 0.7192747339377217, "grad_norm": 0.7041317224502563, "learning_rate": 0.0005, "loss": 1.6869094371795654, "step": 9124 }, { "epoch": 0.7193535672053607, "grad_norm": 0.772716760635376, "learning_rate": 0.0005, "loss": 2.029101848602295, "step": 9125 }, { "epoch": 0.7194324004729996, "grad_norm": 0.6444029808044434, "learning_rate": 0.0005, "loss": 1.9674816131591797, "step": 9126 }, { "epoch": 0.7195112337406385, "grad_norm": 0.7928563952445984, "learning_rate": 0.0005, "loss": 2.0106852054595947, "step": 9127 }, { "epoch": 0.7195900670082775, "grad_norm": 0.6481205224990845, "learning_rate": 0.0005, "loss": 1.8867526054382324, "step": 9128 }, { "epoch": 0.7196689002759165, "grad_norm": 0.6736655235290527, "learning_rate": 0.0005, "loss": 1.9631303548812866, "step": 9129 }, { "epoch": 0.7197477335435554, "grad_norm": 0.6857545971870422, "learning_rate": 0.0005, "loss": 1.9374480247497559, "step": 9130 }, { "epoch": 0.7198265668111943, "grad_norm": 0.5990808010101318, "learning_rate": 0.0005, "loss": 2.002887725830078, "step": 9131 }, { "epoch": 0.7199054000788333, "grad_norm": 0.6099865436553955, "learning_rate": 0.0005, "loss": 2.0119130611419678, "step": 9132 }, { "epoch": 0.7199842333464722, "grad_norm": 0.7086707353591919, "learning_rate": 0.0005, "loss": 1.755299687385559, "step": 9133 }, { "epoch": 0.7200630666141111, "grad_norm": 0.7320718765258789, "learning_rate": 0.0005, "loss": 1.7497183084487915, "step": 9134 }, { "epoch": 0.7201418998817501, "grad_norm": 1.1224048137664795, "learning_rate": 0.0005, "loss": 1.9570448398590088, "step": 9135 }, { "epoch": 0.7202207331493891, "grad_norm": 0.648246169090271, "learning_rate": 0.0005, "loss": 1.6066863536834717, "step": 9136 }, { "epoch": 0.720299566417028, "grad_norm": 0.7403075695037842, "learning_rate": 0.0005, "loss": 1.9738408327102661, "step": 9137 }, { "epoch": 0.7203783996846669, "grad_norm": 0.7755053043365479, "learning_rate": 0.0005, "loss": 2.16160249710083, "step": 9138 }, { "epoch": 0.7204572329523059, "grad_norm": 2.6527650356292725, "learning_rate": 0.0005, "loss": 2.0390074253082275, "step": 9139 }, { "epoch": 0.7205360662199448, "grad_norm": 0.7530584931373596, "learning_rate": 0.0005, "loss": 2.011415958404541, "step": 9140 }, { "epoch": 0.7206148994875837, "grad_norm": 0.622187077999115, "learning_rate": 0.0005, "loss": 1.7939780950546265, "step": 9141 }, { "epoch": 0.7206937327552227, "grad_norm": 0.7944179177284241, "learning_rate": 0.0005, "loss": 1.6396043300628662, "step": 9142 }, { "epoch": 0.7207725660228617, "grad_norm": 0.7338907718658447, "learning_rate": 0.0005, "loss": 2.0567290782928467, "step": 9143 }, { "epoch": 0.7208513992905006, "grad_norm": 0.648578405380249, "learning_rate": 0.0005, "loss": 1.936786413192749, "step": 9144 }, { "epoch": 0.7209302325581395, "grad_norm": 0.703314483165741, "learning_rate": 0.0005, "loss": 1.7805581092834473, "step": 9145 }, { "epoch": 0.7210090658257785, "grad_norm": 0.705288827419281, "learning_rate": 0.0005, "loss": 1.9091802835464478, "step": 9146 }, { "epoch": 0.7210878990934174, "grad_norm": 1.3375366926193237, "learning_rate": 0.0005, "loss": 2.1744163036346436, "step": 9147 }, { "epoch": 0.7211667323610563, "grad_norm": 0.8203662633895874, "learning_rate": 0.0005, "loss": 1.797156810760498, "step": 9148 }, { "epoch": 0.7212455656286954, "grad_norm": 0.6662830114364624, "learning_rate": 0.0005, "loss": 2.08599591255188, "step": 9149 }, { "epoch": 0.7213243988963343, "grad_norm": 0.5988807678222656, "learning_rate": 0.0005, "loss": 2.0274598598480225, "step": 9150 }, { "epoch": 0.7214032321639732, "grad_norm": 0.5252552032470703, "learning_rate": 0.0005, "loss": 2.1666951179504395, "step": 9151 }, { "epoch": 0.7214820654316121, "grad_norm": 0.6408218145370483, "learning_rate": 0.0005, "loss": 1.7781651020050049, "step": 9152 }, { "epoch": 0.7215608986992511, "grad_norm": 0.6716012954711914, "learning_rate": 0.0005, "loss": 2.285550117492676, "step": 9153 }, { "epoch": 0.72163973196689, "grad_norm": 0.5968449711799622, "learning_rate": 0.0005, "loss": 1.8948105573654175, "step": 9154 }, { "epoch": 0.7217185652345289, "grad_norm": 0.6809960007667542, "learning_rate": 0.0005, "loss": 1.8683676719665527, "step": 9155 }, { "epoch": 0.721797398502168, "grad_norm": 0.7013782262802124, "learning_rate": 0.0005, "loss": 1.7766215801239014, "step": 9156 }, { "epoch": 0.7218762317698069, "grad_norm": 0.7739149332046509, "learning_rate": 0.0005, "loss": 1.968279480934143, "step": 9157 }, { "epoch": 0.7219550650374458, "grad_norm": 0.5407289862632751, "learning_rate": 0.0005, "loss": 1.7242122888565063, "step": 9158 }, { "epoch": 0.7220338983050848, "grad_norm": 0.7608084678649902, "learning_rate": 0.0005, "loss": 1.7206916809082031, "step": 9159 }, { "epoch": 0.7221127315727237, "grad_norm": 1.3508877754211426, "learning_rate": 0.0005, "loss": 2.2628121376037598, "step": 9160 }, { "epoch": 0.7221915648403626, "grad_norm": 0.6148934364318848, "learning_rate": 0.0005, "loss": 1.811471939086914, "step": 9161 }, { "epoch": 0.7222703981080015, "grad_norm": 0.6698945760726929, "learning_rate": 0.0005, "loss": 1.6820063591003418, "step": 9162 }, { "epoch": 0.7223492313756406, "grad_norm": 0.5890193581581116, "learning_rate": 0.0005, "loss": 1.5687222480773926, "step": 9163 }, { "epoch": 0.7224280646432795, "grad_norm": 0.6092206239700317, "learning_rate": 0.0005, "loss": 1.455243468284607, "step": 9164 }, { "epoch": 0.7225068979109184, "grad_norm": 0.6784874200820923, "learning_rate": 0.0005, "loss": 1.9707964658737183, "step": 9165 }, { "epoch": 0.7225857311785574, "grad_norm": 2.1146080493927, "learning_rate": 0.0005, "loss": 2.079360246658325, "step": 9166 }, { "epoch": 0.7226645644461963, "grad_norm": 0.6205592751502991, "learning_rate": 0.0005, "loss": 1.8313943147659302, "step": 9167 }, { "epoch": 0.7227433977138352, "grad_norm": 0.6705474257469177, "learning_rate": 0.0005, "loss": 1.8486336469650269, "step": 9168 }, { "epoch": 0.7228222309814741, "grad_norm": 0.5164082050323486, "learning_rate": 0.0005, "loss": 1.823287010192871, "step": 9169 }, { "epoch": 0.7229010642491132, "grad_norm": 0.6677771210670471, "learning_rate": 0.0005, "loss": 1.808037519454956, "step": 9170 }, { "epoch": 0.7229798975167521, "grad_norm": 0.5666815638542175, "learning_rate": 0.0005, "loss": 1.8760323524475098, "step": 9171 }, { "epoch": 0.723058730784391, "grad_norm": 0.586421012878418, "learning_rate": 0.0005, "loss": 1.8239831924438477, "step": 9172 }, { "epoch": 0.72313756405203, "grad_norm": 0.5882858037948608, "learning_rate": 0.0005, "loss": 1.9119080305099487, "step": 9173 }, { "epoch": 0.7232163973196689, "grad_norm": 0.7144732475280762, "learning_rate": 0.0005, "loss": 2.126288890838623, "step": 9174 }, { "epoch": 0.7232952305873078, "grad_norm": 0.7834759950637817, "learning_rate": 0.0005, "loss": 2.0106234550476074, "step": 9175 }, { "epoch": 0.7233740638549467, "grad_norm": 0.7632516026496887, "learning_rate": 0.0005, "loss": 2.2622451782226562, "step": 9176 }, { "epoch": 0.7234528971225858, "grad_norm": 0.6859428882598877, "learning_rate": 0.0005, "loss": 1.84165358543396, "step": 9177 }, { "epoch": 0.7235317303902247, "grad_norm": 1.0894452333450317, "learning_rate": 0.0005, "loss": 1.814507007598877, "step": 9178 }, { "epoch": 0.7236105636578636, "grad_norm": 0.5310894250869751, "learning_rate": 0.0005, "loss": 1.8738524913787842, "step": 9179 }, { "epoch": 0.7236893969255026, "grad_norm": 0.6352155208587646, "learning_rate": 0.0005, "loss": 2.090054750442505, "step": 9180 }, { "epoch": 0.7237682301931415, "grad_norm": 0.5678902864456177, "learning_rate": 0.0005, "loss": 2.134382724761963, "step": 9181 }, { "epoch": 0.7238470634607804, "grad_norm": 0.6992247104644775, "learning_rate": 0.0005, "loss": 1.9547717571258545, "step": 9182 }, { "epoch": 0.7239258967284193, "grad_norm": 0.7276099324226379, "learning_rate": 0.0005, "loss": 2.0893406867980957, "step": 9183 }, { "epoch": 0.7240047299960584, "grad_norm": 0.9320024847984314, "learning_rate": 0.0005, "loss": 1.7518529891967773, "step": 9184 }, { "epoch": 0.7240835632636973, "grad_norm": 0.7102754712104797, "learning_rate": 0.0005, "loss": 1.9072201251983643, "step": 9185 }, { "epoch": 0.7241623965313362, "grad_norm": 0.6678953766822815, "learning_rate": 0.0005, "loss": 1.6465024948120117, "step": 9186 }, { "epoch": 0.7242412297989752, "grad_norm": 0.777271568775177, "learning_rate": 0.0005, "loss": 1.6377025842666626, "step": 9187 }, { "epoch": 0.7243200630666141, "grad_norm": 0.6475114822387695, "learning_rate": 0.0005, "loss": 1.8497936725616455, "step": 9188 }, { "epoch": 0.724398896334253, "grad_norm": 0.816404402256012, "learning_rate": 0.0005, "loss": 2.0755887031555176, "step": 9189 }, { "epoch": 0.724477729601892, "grad_norm": 0.6246086955070496, "learning_rate": 0.0005, "loss": 1.722692847251892, "step": 9190 }, { "epoch": 0.724556562869531, "grad_norm": 0.6239805817604065, "learning_rate": 0.0005, "loss": 2.073604106903076, "step": 9191 }, { "epoch": 0.7246353961371699, "grad_norm": 0.529111385345459, "learning_rate": 0.0005, "loss": 1.6456329822540283, "step": 9192 }, { "epoch": 0.7247142294048088, "grad_norm": 0.5755679607391357, "learning_rate": 0.0005, "loss": 1.9748305082321167, "step": 9193 }, { "epoch": 0.7247930626724478, "grad_norm": 0.6635373830795288, "learning_rate": 0.0005, "loss": 1.984920620918274, "step": 9194 }, { "epoch": 0.7248718959400867, "grad_norm": 0.5406182408332825, "learning_rate": 0.0005, "loss": 2.0081310272216797, "step": 9195 }, { "epoch": 0.7249507292077256, "grad_norm": 0.6269696950912476, "learning_rate": 0.0005, "loss": 2.035898447036743, "step": 9196 }, { "epoch": 0.7250295624753647, "grad_norm": 0.6225789189338684, "learning_rate": 0.0005, "loss": 2.0482053756713867, "step": 9197 }, { "epoch": 0.7251083957430036, "grad_norm": 0.5318544507026672, "learning_rate": 0.0005, "loss": 1.884030818939209, "step": 9198 }, { "epoch": 0.7251872290106425, "grad_norm": 0.7302780151367188, "learning_rate": 0.0005, "loss": 1.8078062534332275, "step": 9199 }, { "epoch": 0.7252660622782814, "grad_norm": 0.6203171014785767, "learning_rate": 0.0005, "loss": 1.8865020275115967, "step": 9200 }, { "epoch": 0.7253448955459204, "grad_norm": 0.6446621417999268, "learning_rate": 0.0005, "loss": 1.8146026134490967, "step": 9201 }, { "epoch": 0.7254237288135593, "grad_norm": 0.5778706669807434, "learning_rate": 0.0005, "loss": 1.727292537689209, "step": 9202 }, { "epoch": 0.7255025620811982, "grad_norm": 0.585108757019043, "learning_rate": 0.0005, "loss": 1.6958857774734497, "step": 9203 }, { "epoch": 0.7255813953488373, "grad_norm": 0.7869667410850525, "learning_rate": 0.0005, "loss": 2.054356575012207, "step": 9204 }, { "epoch": 0.7256602286164762, "grad_norm": 0.5820947885513306, "learning_rate": 0.0005, "loss": 1.6531181335449219, "step": 9205 }, { "epoch": 0.7257390618841151, "grad_norm": 0.6645328402519226, "learning_rate": 0.0005, "loss": 1.629195213317871, "step": 9206 }, { "epoch": 0.725817895151754, "grad_norm": 0.7737454175949097, "learning_rate": 0.0005, "loss": 1.9206736087799072, "step": 9207 }, { "epoch": 0.725896728419393, "grad_norm": 0.6677359342575073, "learning_rate": 0.0005, "loss": 2.0026919841766357, "step": 9208 }, { "epoch": 0.7259755616870319, "grad_norm": 0.8163275718688965, "learning_rate": 0.0005, "loss": 2.0394675731658936, "step": 9209 }, { "epoch": 0.7260543949546708, "grad_norm": 0.852433443069458, "learning_rate": 0.0005, "loss": 2.156156539916992, "step": 9210 }, { "epoch": 0.7261332282223099, "grad_norm": 0.6539653539657593, "learning_rate": 0.0005, "loss": 1.8848497867584229, "step": 9211 }, { "epoch": 0.7262120614899488, "grad_norm": 0.6293109655380249, "learning_rate": 0.0005, "loss": 1.8090461492538452, "step": 9212 }, { "epoch": 0.7262908947575877, "grad_norm": 0.6116794347763062, "learning_rate": 0.0005, "loss": 1.8886730670928955, "step": 9213 }, { "epoch": 0.7263697280252266, "grad_norm": 0.6331881880760193, "learning_rate": 0.0005, "loss": 1.8858267068862915, "step": 9214 }, { "epoch": 0.7264485612928656, "grad_norm": 0.7041909098625183, "learning_rate": 0.0005, "loss": 1.8572415113449097, "step": 9215 }, { "epoch": 0.7265273945605045, "grad_norm": 0.6758946180343628, "learning_rate": 0.0005, "loss": 1.8499945402145386, "step": 9216 }, { "epoch": 0.7266062278281434, "grad_norm": 0.661801815032959, "learning_rate": 0.0005, "loss": 1.872406005859375, "step": 9217 }, { "epoch": 0.7266850610957825, "grad_norm": 0.7866308689117432, "learning_rate": 0.0005, "loss": 1.874739408493042, "step": 9218 }, { "epoch": 0.7267638943634214, "grad_norm": 0.6695553064346313, "learning_rate": 0.0005, "loss": 1.7279857397079468, "step": 9219 }, { "epoch": 0.7268427276310603, "grad_norm": 0.9124420881271362, "learning_rate": 0.0005, "loss": 1.9639097452163696, "step": 9220 }, { "epoch": 0.7269215608986993, "grad_norm": 0.702796995639801, "learning_rate": 0.0005, "loss": 2.0938494205474854, "step": 9221 }, { "epoch": 0.7270003941663382, "grad_norm": 0.7466312646865845, "learning_rate": 0.0005, "loss": 1.7383677959442139, "step": 9222 }, { "epoch": 0.7270792274339771, "grad_norm": 0.6487343907356262, "learning_rate": 0.0005, "loss": 1.724494218826294, "step": 9223 }, { "epoch": 0.727158060701616, "grad_norm": 0.7380297183990479, "learning_rate": 0.0005, "loss": 1.8726019859313965, "step": 9224 }, { "epoch": 0.7272368939692551, "grad_norm": 0.7723127007484436, "learning_rate": 0.0005, "loss": 2.0416386127471924, "step": 9225 }, { "epoch": 0.727315727236894, "grad_norm": 0.6916347146034241, "learning_rate": 0.0005, "loss": 2.075620174407959, "step": 9226 }, { "epoch": 0.7273945605045329, "grad_norm": 0.701336145401001, "learning_rate": 0.0005, "loss": 2.0146660804748535, "step": 9227 }, { "epoch": 0.7274733937721719, "grad_norm": 0.6941640377044678, "learning_rate": 0.0005, "loss": 1.6829098463058472, "step": 9228 }, { "epoch": 0.7275522270398108, "grad_norm": 0.6882567405700684, "learning_rate": 0.0005, "loss": 2.1464967727661133, "step": 9229 }, { "epoch": 0.7276310603074497, "grad_norm": 0.69140625, "learning_rate": 0.0005, "loss": 2.0793280601501465, "step": 9230 }, { "epoch": 0.7277098935750886, "grad_norm": 0.7236253023147583, "learning_rate": 0.0005, "loss": 1.8436520099639893, "step": 9231 }, { "epoch": 0.7277887268427277, "grad_norm": 0.6923153400421143, "learning_rate": 0.0005, "loss": 2.1518630981445312, "step": 9232 }, { "epoch": 0.7278675601103666, "grad_norm": 1.2251912355422974, "learning_rate": 0.0005, "loss": 2.0422089099884033, "step": 9233 }, { "epoch": 0.7279463933780055, "grad_norm": 0.9096042513847351, "learning_rate": 0.0005, "loss": 1.8771710395812988, "step": 9234 }, { "epoch": 0.7280252266456445, "grad_norm": 0.7284536957740784, "learning_rate": 0.0005, "loss": 1.9916077852249146, "step": 9235 }, { "epoch": 0.7281040599132834, "grad_norm": 0.6336915493011475, "learning_rate": 0.0005, "loss": 1.813010811805725, "step": 9236 }, { "epoch": 0.7281828931809223, "grad_norm": 0.7554452419281006, "learning_rate": 0.0005, "loss": 1.8374959230422974, "step": 9237 }, { "epoch": 0.7282617264485612, "grad_norm": 0.6589020490646362, "learning_rate": 0.0005, "loss": 1.802243709564209, "step": 9238 }, { "epoch": 0.7283405597162003, "grad_norm": 0.6439796090126038, "learning_rate": 0.0005, "loss": 2.009364604949951, "step": 9239 }, { "epoch": 0.7284193929838392, "grad_norm": 0.6604204177856445, "learning_rate": 0.0005, "loss": 2.087946653366089, "step": 9240 }, { "epoch": 0.7284982262514781, "grad_norm": 0.684755265712738, "learning_rate": 0.0005, "loss": 2.1476776599884033, "step": 9241 }, { "epoch": 0.7285770595191171, "grad_norm": 0.7723010182380676, "learning_rate": 0.0005, "loss": 1.9076249599456787, "step": 9242 }, { "epoch": 0.728655892786756, "grad_norm": 0.6219289898872375, "learning_rate": 0.0005, "loss": 1.8534420728683472, "step": 9243 }, { "epoch": 0.7287347260543949, "grad_norm": 0.5667567849159241, "learning_rate": 0.0005, "loss": 1.7941197156906128, "step": 9244 }, { "epoch": 0.7288135593220338, "grad_norm": 0.6024348735809326, "learning_rate": 0.0005, "loss": 1.663787841796875, "step": 9245 }, { "epoch": 0.7288923925896729, "grad_norm": 0.7072734832763672, "learning_rate": 0.0005, "loss": 1.8972561359405518, "step": 9246 }, { "epoch": 0.7289712258573118, "grad_norm": 0.8052246570587158, "learning_rate": 0.0005, "loss": 2.048527956008911, "step": 9247 }, { "epoch": 0.7290500591249507, "grad_norm": 0.607955813407898, "learning_rate": 0.0005, "loss": 2.0506110191345215, "step": 9248 }, { "epoch": 0.7291288923925897, "grad_norm": 0.623384416103363, "learning_rate": 0.0005, "loss": 1.845408320426941, "step": 9249 }, { "epoch": 0.7292077256602286, "grad_norm": 0.8751408457756042, "learning_rate": 0.0005, "loss": 2.0774619579315186, "step": 9250 }, { "epoch": 0.7292865589278675, "grad_norm": 0.617188036441803, "learning_rate": 0.0005, "loss": 1.8187553882598877, "step": 9251 }, { "epoch": 0.7293653921955066, "grad_norm": 2.5564887523651123, "learning_rate": 0.0005, "loss": 1.8016409873962402, "step": 9252 }, { "epoch": 0.7294442254631455, "grad_norm": 0.883312463760376, "learning_rate": 0.0005, "loss": 2.1258726119995117, "step": 9253 }, { "epoch": 0.7295230587307844, "grad_norm": 0.7274292707443237, "learning_rate": 0.0005, "loss": 2.0485599040985107, "step": 9254 }, { "epoch": 0.7296018919984233, "grad_norm": 0.6198480725288391, "learning_rate": 0.0005, "loss": 2.003571033477783, "step": 9255 }, { "epoch": 0.7296807252660623, "grad_norm": 0.9079511761665344, "learning_rate": 0.0005, "loss": 2.2400519847869873, "step": 9256 }, { "epoch": 0.7297595585337012, "grad_norm": 0.5944641828536987, "learning_rate": 0.0005, "loss": 2.0327794551849365, "step": 9257 }, { "epoch": 0.7298383918013401, "grad_norm": 0.6694498062133789, "learning_rate": 0.0005, "loss": 1.947149395942688, "step": 9258 }, { "epoch": 0.7299172250689792, "grad_norm": 0.651549756526947, "learning_rate": 0.0005, "loss": 1.8075251579284668, "step": 9259 }, { "epoch": 0.7299960583366181, "grad_norm": 0.7179029583930969, "learning_rate": 0.0005, "loss": 1.8397003412246704, "step": 9260 }, { "epoch": 0.730074891604257, "grad_norm": 0.6529034972190857, "learning_rate": 0.0005, "loss": 2.1029884815216064, "step": 9261 }, { "epoch": 0.7301537248718959, "grad_norm": 0.7384169101715088, "learning_rate": 0.0005, "loss": 1.9254512786865234, "step": 9262 }, { "epoch": 0.7302325581395349, "grad_norm": 0.5955119132995605, "learning_rate": 0.0005, "loss": 1.647581934928894, "step": 9263 }, { "epoch": 0.7303113914071738, "grad_norm": 0.6594130992889404, "learning_rate": 0.0005, "loss": 2.0979132652282715, "step": 9264 }, { "epoch": 0.7303902246748127, "grad_norm": 0.6386796832084656, "learning_rate": 0.0005, "loss": 1.6309599876403809, "step": 9265 }, { "epoch": 0.7304690579424518, "grad_norm": 0.7871642112731934, "learning_rate": 0.0005, "loss": 1.7723091840744019, "step": 9266 }, { "epoch": 0.7305478912100907, "grad_norm": 0.5960090756416321, "learning_rate": 0.0005, "loss": 1.8033452033996582, "step": 9267 }, { "epoch": 0.7306267244777296, "grad_norm": 0.6642274260520935, "learning_rate": 0.0005, "loss": 1.7772847414016724, "step": 9268 }, { "epoch": 0.7307055577453685, "grad_norm": 0.5574668645858765, "learning_rate": 0.0005, "loss": 1.9719375371932983, "step": 9269 }, { "epoch": 0.7307843910130075, "grad_norm": 0.7780315279960632, "learning_rate": 0.0005, "loss": 1.9721921682357788, "step": 9270 }, { "epoch": 0.7308632242806464, "grad_norm": 0.6496181488037109, "learning_rate": 0.0005, "loss": 1.9073245525360107, "step": 9271 }, { "epoch": 0.7309420575482853, "grad_norm": 0.7109174728393555, "learning_rate": 0.0005, "loss": 1.8269866704940796, "step": 9272 }, { "epoch": 0.7310208908159244, "grad_norm": 0.521573543548584, "learning_rate": 0.0005, "loss": 1.688244104385376, "step": 9273 }, { "epoch": 0.7310997240835633, "grad_norm": 0.5715900659561157, "learning_rate": 0.0005, "loss": 1.8024870157241821, "step": 9274 }, { "epoch": 0.7311785573512022, "grad_norm": 0.5565628409385681, "learning_rate": 0.0005, "loss": 1.6540955305099487, "step": 9275 }, { "epoch": 0.7312573906188411, "grad_norm": 0.5963089466094971, "learning_rate": 0.0005, "loss": 1.7955420017242432, "step": 9276 }, { "epoch": 0.7313362238864801, "grad_norm": 0.6327680349349976, "learning_rate": 0.0005, "loss": 1.752040147781372, "step": 9277 }, { "epoch": 0.731415057154119, "grad_norm": 0.6209014654159546, "learning_rate": 0.0005, "loss": 1.967177391052246, "step": 9278 }, { "epoch": 0.7314938904217579, "grad_norm": 0.4974265694618225, "learning_rate": 0.0005, "loss": 1.6130280494689941, "step": 9279 }, { "epoch": 0.731572723689397, "grad_norm": 0.661117672920227, "learning_rate": 0.0005, "loss": 1.6549468040466309, "step": 9280 }, { "epoch": 0.7316515569570359, "grad_norm": 0.9393086433410645, "learning_rate": 0.0005, "loss": 1.8642261028289795, "step": 9281 }, { "epoch": 0.7317303902246748, "grad_norm": 0.6482251286506653, "learning_rate": 0.0005, "loss": 2.199115037918091, "step": 9282 }, { "epoch": 0.7318092234923138, "grad_norm": 2.297667980194092, "learning_rate": 0.0005, "loss": 1.847386360168457, "step": 9283 }, { "epoch": 0.7318880567599527, "grad_norm": 0.673198401927948, "learning_rate": 0.0005, "loss": 2.083526372909546, "step": 9284 }, { "epoch": 0.7319668900275916, "grad_norm": 0.6447394490242004, "learning_rate": 0.0005, "loss": 2.02809739112854, "step": 9285 }, { "epoch": 0.7320457232952305, "grad_norm": 0.5910573601722717, "learning_rate": 0.0005, "loss": 1.8471155166625977, "step": 9286 }, { "epoch": 0.7321245565628696, "grad_norm": 0.6269771456718445, "learning_rate": 0.0005, "loss": 1.9257879257202148, "step": 9287 }, { "epoch": 0.7322033898305085, "grad_norm": 0.748071014881134, "learning_rate": 0.0005, "loss": 1.9466692209243774, "step": 9288 }, { "epoch": 0.7322822230981474, "grad_norm": 0.7456823587417603, "learning_rate": 0.0005, "loss": 2.0872879028320312, "step": 9289 }, { "epoch": 0.7323610563657864, "grad_norm": 0.6865342855453491, "learning_rate": 0.0005, "loss": 1.8183655738830566, "step": 9290 }, { "epoch": 0.7324398896334253, "grad_norm": 1.7621053457260132, "learning_rate": 0.0005, "loss": 1.92844820022583, "step": 9291 }, { "epoch": 0.7325187229010642, "grad_norm": 0.7711586356163025, "learning_rate": 0.0005, "loss": 1.8159961700439453, "step": 9292 }, { "epoch": 0.7325975561687031, "grad_norm": 0.7484322786331177, "learning_rate": 0.0005, "loss": 1.9292103052139282, "step": 9293 }, { "epoch": 0.7326763894363422, "grad_norm": 0.6802356243133545, "learning_rate": 0.0005, "loss": 1.8082082271575928, "step": 9294 }, { "epoch": 0.7327552227039811, "grad_norm": 0.8134555816650391, "learning_rate": 0.0005, "loss": 1.7980425357818604, "step": 9295 }, { "epoch": 0.73283405597162, "grad_norm": 0.719700813293457, "learning_rate": 0.0005, "loss": 1.8275178670883179, "step": 9296 }, { "epoch": 0.732912889239259, "grad_norm": 0.7126932144165039, "learning_rate": 0.0005, "loss": 2.0204057693481445, "step": 9297 }, { "epoch": 0.7329917225068979, "grad_norm": 0.6955311298370361, "learning_rate": 0.0005, "loss": 1.7538230419158936, "step": 9298 }, { "epoch": 0.7330705557745368, "grad_norm": 0.628836989402771, "learning_rate": 0.0005, "loss": 1.792870044708252, "step": 9299 }, { "epoch": 0.7331493890421757, "grad_norm": 1.0254037380218506, "learning_rate": 0.0005, "loss": 2.1388893127441406, "step": 9300 }, { "epoch": 0.7332282223098148, "grad_norm": 0.6011600494384766, "learning_rate": 0.0005, "loss": 2.177943229675293, "step": 9301 }, { "epoch": 0.7333070555774537, "grad_norm": 0.7428950667381287, "learning_rate": 0.0005, "loss": 1.665197491645813, "step": 9302 }, { "epoch": 0.7333858888450926, "grad_norm": 0.664896547794342, "learning_rate": 0.0005, "loss": 1.874908685684204, "step": 9303 }, { "epoch": 0.7334647221127316, "grad_norm": 0.6655921339988708, "learning_rate": 0.0005, "loss": 1.6772016286849976, "step": 9304 }, { "epoch": 0.7335435553803705, "grad_norm": 0.7035356163978577, "learning_rate": 0.0005, "loss": 2.249755382537842, "step": 9305 }, { "epoch": 0.7336223886480094, "grad_norm": 0.7008165121078491, "learning_rate": 0.0005, "loss": 2.102942943572998, "step": 9306 }, { "epoch": 0.7337012219156483, "grad_norm": 0.7763233780860901, "learning_rate": 0.0005, "loss": 1.7659902572631836, "step": 9307 }, { "epoch": 0.7337800551832874, "grad_norm": 0.7838553786277771, "learning_rate": 0.0005, "loss": 1.96608304977417, "step": 9308 }, { "epoch": 0.7338588884509263, "grad_norm": 0.9047229886054993, "learning_rate": 0.0005, "loss": 2.2390661239624023, "step": 9309 }, { "epoch": 0.7339377217185652, "grad_norm": 0.715875506401062, "learning_rate": 0.0005, "loss": 2.078465461730957, "step": 9310 }, { "epoch": 0.7340165549862042, "grad_norm": 0.8194119930267334, "learning_rate": 0.0005, "loss": 2.2171201705932617, "step": 9311 }, { "epoch": 0.7340953882538431, "grad_norm": 0.6442577838897705, "learning_rate": 0.0005, "loss": 1.9141653776168823, "step": 9312 }, { "epoch": 0.734174221521482, "grad_norm": 0.7497192621231079, "learning_rate": 0.0005, "loss": 1.7111520767211914, "step": 9313 }, { "epoch": 0.7342530547891211, "grad_norm": 0.6444454193115234, "learning_rate": 0.0005, "loss": 1.7171436548233032, "step": 9314 }, { "epoch": 0.73433188805676, "grad_norm": 0.5085998773574829, "learning_rate": 0.0005, "loss": 1.9374969005584717, "step": 9315 }, { "epoch": 0.7344107213243989, "grad_norm": 0.5928830504417419, "learning_rate": 0.0005, "loss": 2.167095184326172, "step": 9316 }, { "epoch": 0.7344895545920378, "grad_norm": 0.6628264784812927, "learning_rate": 0.0005, "loss": 1.819027066230774, "step": 9317 }, { "epoch": 0.7345683878596768, "grad_norm": 0.68562912940979, "learning_rate": 0.0005, "loss": 1.9842840433120728, "step": 9318 }, { "epoch": 0.7346472211273157, "grad_norm": 0.7468436360359192, "learning_rate": 0.0005, "loss": 1.936884880065918, "step": 9319 }, { "epoch": 0.7347260543949546, "grad_norm": 0.5909259915351868, "learning_rate": 0.0005, "loss": 1.827455759048462, "step": 9320 }, { "epoch": 0.7348048876625937, "grad_norm": 0.9623626470565796, "learning_rate": 0.0005, "loss": 1.777940273284912, "step": 9321 }, { "epoch": 0.7348837209302326, "grad_norm": 0.6518658995628357, "learning_rate": 0.0005, "loss": 1.9404507875442505, "step": 9322 }, { "epoch": 0.7349625541978715, "grad_norm": 0.6388916969299316, "learning_rate": 0.0005, "loss": 1.7215594053268433, "step": 9323 }, { "epoch": 0.7350413874655104, "grad_norm": 0.7139084339141846, "learning_rate": 0.0005, "loss": 1.740221381187439, "step": 9324 }, { "epoch": 0.7351202207331494, "grad_norm": 1.106285810470581, "learning_rate": 0.0005, "loss": 2.0100769996643066, "step": 9325 }, { "epoch": 0.7351990540007883, "grad_norm": 0.6923117637634277, "learning_rate": 0.0005, "loss": 1.8020999431610107, "step": 9326 }, { "epoch": 0.7352778872684272, "grad_norm": 0.6867321133613586, "learning_rate": 0.0005, "loss": 1.6327259540557861, "step": 9327 }, { "epoch": 0.7353567205360663, "grad_norm": 0.6651520133018494, "learning_rate": 0.0005, "loss": 1.9783014059066772, "step": 9328 }, { "epoch": 0.7354355538037052, "grad_norm": 0.523807942867279, "learning_rate": 0.0005, "loss": 1.6676099300384521, "step": 9329 }, { "epoch": 0.7355143870713441, "grad_norm": 0.6219662427902222, "learning_rate": 0.0005, "loss": 1.7360100746154785, "step": 9330 }, { "epoch": 0.735593220338983, "grad_norm": 0.7754186987876892, "learning_rate": 0.0005, "loss": 1.9898430109024048, "step": 9331 }, { "epoch": 0.735672053606622, "grad_norm": 0.8002005815505981, "learning_rate": 0.0005, "loss": 1.9728385210037231, "step": 9332 }, { "epoch": 0.7357508868742609, "grad_norm": 0.6767207384109497, "learning_rate": 0.0005, "loss": 2.2906250953674316, "step": 9333 }, { "epoch": 0.7358297201418998, "grad_norm": 0.7000625729560852, "learning_rate": 0.0005, "loss": 2.120560646057129, "step": 9334 }, { "epoch": 0.7359085534095389, "grad_norm": 0.6460463404655457, "learning_rate": 0.0005, "loss": 1.680589199066162, "step": 9335 }, { "epoch": 0.7359873866771778, "grad_norm": 0.6831854581832886, "learning_rate": 0.0005, "loss": 1.7242716550827026, "step": 9336 }, { "epoch": 0.7360662199448167, "grad_norm": 0.8595014214515686, "learning_rate": 0.0005, "loss": 2.180544137954712, "step": 9337 }, { "epoch": 0.7361450532124556, "grad_norm": 0.6680096387863159, "learning_rate": 0.0005, "loss": 1.8224265575408936, "step": 9338 }, { "epoch": 0.7362238864800946, "grad_norm": 0.718492865562439, "learning_rate": 0.0005, "loss": 2.251499652862549, "step": 9339 }, { "epoch": 0.7363027197477335, "grad_norm": 0.9291606545448303, "learning_rate": 0.0005, "loss": 1.6696076393127441, "step": 9340 }, { "epoch": 0.7363815530153724, "grad_norm": 0.6137329339981079, "learning_rate": 0.0005, "loss": 1.725028157234192, "step": 9341 }, { "epoch": 0.7364603862830115, "grad_norm": 0.788347601890564, "learning_rate": 0.0005, "loss": 1.8838155269622803, "step": 9342 }, { "epoch": 0.7365392195506504, "grad_norm": 0.8166630268096924, "learning_rate": 0.0005, "loss": 2.076749801635742, "step": 9343 }, { "epoch": 0.7366180528182893, "grad_norm": 1.011088252067566, "learning_rate": 0.0005, "loss": 1.7262744903564453, "step": 9344 }, { "epoch": 0.7366968860859283, "grad_norm": 1.17681884765625, "learning_rate": 0.0005, "loss": 2.182227611541748, "step": 9345 }, { "epoch": 0.7367757193535672, "grad_norm": 0.5574020743370056, "learning_rate": 0.0005, "loss": 1.6923418045043945, "step": 9346 }, { "epoch": 0.7368545526212061, "grad_norm": 0.7923884987831116, "learning_rate": 0.0005, "loss": 1.994381308555603, "step": 9347 }, { "epoch": 0.736933385888845, "grad_norm": 1.0350980758666992, "learning_rate": 0.0005, "loss": 1.8780813217163086, "step": 9348 }, { "epoch": 0.7370122191564841, "grad_norm": 0.7955427765846252, "learning_rate": 0.0005, "loss": 2.036726951599121, "step": 9349 }, { "epoch": 0.737091052424123, "grad_norm": 0.935761570930481, "learning_rate": 0.0005, "loss": 2.036334276199341, "step": 9350 }, { "epoch": 0.7371698856917619, "grad_norm": 0.8311247825622559, "learning_rate": 0.0005, "loss": 2.143333911895752, "step": 9351 }, { "epoch": 0.7372487189594009, "grad_norm": 0.6372998356819153, "learning_rate": 0.0005, "loss": 1.7276607751846313, "step": 9352 }, { "epoch": 0.7373275522270398, "grad_norm": 0.6347121596336365, "learning_rate": 0.0005, "loss": 1.9017786979675293, "step": 9353 }, { "epoch": 0.7374063854946787, "grad_norm": 0.6760802865028381, "learning_rate": 0.0005, "loss": 2.0410261154174805, "step": 9354 }, { "epoch": 0.7374852187623177, "grad_norm": 0.777618408203125, "learning_rate": 0.0005, "loss": 1.9358339309692383, "step": 9355 }, { "epoch": 0.7375640520299567, "grad_norm": 0.5410312414169312, "learning_rate": 0.0005, "loss": 1.8622875213623047, "step": 9356 }, { "epoch": 0.7376428852975956, "grad_norm": 0.721366286277771, "learning_rate": 0.0005, "loss": 2.041903018951416, "step": 9357 }, { "epoch": 0.7377217185652345, "grad_norm": 0.6612979769706726, "learning_rate": 0.0005, "loss": 1.6221637725830078, "step": 9358 }, { "epoch": 0.7378005518328735, "grad_norm": 0.6274705529212952, "learning_rate": 0.0005, "loss": 1.9835309982299805, "step": 9359 }, { "epoch": 0.7378793851005124, "grad_norm": 0.751263439655304, "learning_rate": 0.0005, "loss": 1.928181529045105, "step": 9360 }, { "epoch": 0.7379582183681513, "grad_norm": 0.6786234974861145, "learning_rate": 0.0005, "loss": 1.722631812095642, "step": 9361 }, { "epoch": 0.7380370516357903, "grad_norm": 0.6935789585113525, "learning_rate": 0.0005, "loss": 1.8259354829788208, "step": 9362 }, { "epoch": 0.7381158849034293, "grad_norm": 0.6719855070114136, "learning_rate": 0.0005, "loss": 2.090113878250122, "step": 9363 }, { "epoch": 0.7381947181710682, "grad_norm": 0.7219762206077576, "learning_rate": 0.0005, "loss": 2.007392168045044, "step": 9364 }, { "epoch": 0.7382735514387071, "grad_norm": 0.8585512042045593, "learning_rate": 0.0005, "loss": 1.9698632955551147, "step": 9365 }, { "epoch": 0.7383523847063461, "grad_norm": 0.5778297185897827, "learning_rate": 0.0005, "loss": 1.80032479763031, "step": 9366 }, { "epoch": 0.738431217973985, "grad_norm": 0.8195244669914246, "learning_rate": 0.0005, "loss": 1.9891992807388306, "step": 9367 }, { "epoch": 0.7385100512416239, "grad_norm": 0.657261312007904, "learning_rate": 0.0005, "loss": 1.9693405628204346, "step": 9368 }, { "epoch": 0.7385888845092629, "grad_norm": 0.6056134104728699, "learning_rate": 0.0005, "loss": 1.9751005172729492, "step": 9369 }, { "epoch": 0.7386677177769019, "grad_norm": 0.7972062230110168, "learning_rate": 0.0005, "loss": 2.014038562774658, "step": 9370 }, { "epoch": 0.7387465510445408, "grad_norm": 0.6183061003684998, "learning_rate": 0.0005, "loss": 1.8128588199615479, "step": 9371 }, { "epoch": 0.7388253843121797, "grad_norm": 3.031938076019287, "learning_rate": 0.0005, "loss": 2.967801094055176, "step": 9372 }, { "epoch": 0.7389042175798187, "grad_norm": 0.671689510345459, "learning_rate": 0.0005, "loss": 1.730024814605713, "step": 9373 }, { "epoch": 0.7389830508474576, "grad_norm": 1.0256588459014893, "learning_rate": 0.0005, "loss": 2.198366641998291, "step": 9374 }, { "epoch": 0.7390618841150965, "grad_norm": 0.6721063256263733, "learning_rate": 0.0005, "loss": 1.9346482753753662, "step": 9375 }, { "epoch": 0.7391407173827356, "grad_norm": 0.6331116557121277, "learning_rate": 0.0005, "loss": 1.7808444499969482, "step": 9376 }, { "epoch": 0.7392195506503745, "grad_norm": 0.6057696342468262, "learning_rate": 0.0005, "loss": 2.007404327392578, "step": 9377 }, { "epoch": 0.7392983839180134, "grad_norm": 1.4609225988388062, "learning_rate": 0.0005, "loss": 2.108362913131714, "step": 9378 }, { "epoch": 0.7393772171856523, "grad_norm": 0.7261483669281006, "learning_rate": 0.0005, "loss": 1.619896411895752, "step": 9379 }, { "epoch": 0.7394560504532913, "grad_norm": 0.6014655828475952, "learning_rate": 0.0005, "loss": 1.8504137992858887, "step": 9380 }, { "epoch": 0.7395348837209302, "grad_norm": 0.6415349245071411, "learning_rate": 0.0005, "loss": 2.3506925106048584, "step": 9381 }, { "epoch": 0.7396137169885691, "grad_norm": 0.7369745969772339, "learning_rate": 0.0005, "loss": 1.76716947555542, "step": 9382 }, { "epoch": 0.7396925502562082, "grad_norm": 0.6488073468208313, "learning_rate": 0.0005, "loss": 1.6211354732513428, "step": 9383 }, { "epoch": 0.7397713835238471, "grad_norm": 0.6249187588691711, "learning_rate": 0.0005, "loss": 1.8889377117156982, "step": 9384 }, { "epoch": 0.739850216791486, "grad_norm": 0.677157461643219, "learning_rate": 0.0005, "loss": 2.001084804534912, "step": 9385 }, { "epoch": 0.7399290500591249, "grad_norm": 0.6606568694114685, "learning_rate": 0.0005, "loss": 1.8375847339630127, "step": 9386 }, { "epoch": 0.7400078833267639, "grad_norm": 0.7176454663276672, "learning_rate": 0.0005, "loss": 1.8770360946655273, "step": 9387 }, { "epoch": 0.7400867165944028, "grad_norm": 0.634096086025238, "learning_rate": 0.0005, "loss": 1.955171823501587, "step": 9388 }, { "epoch": 0.7401655498620417, "grad_norm": 0.5811935663223267, "learning_rate": 0.0005, "loss": 1.916268229484558, "step": 9389 }, { "epoch": 0.7402443831296808, "grad_norm": 0.6632620096206665, "learning_rate": 0.0005, "loss": 1.7357149124145508, "step": 9390 }, { "epoch": 0.7403232163973197, "grad_norm": 0.6158890724182129, "learning_rate": 0.0005, "loss": 1.548090934753418, "step": 9391 }, { "epoch": 0.7404020496649586, "grad_norm": 0.6841351985931396, "learning_rate": 0.0005, "loss": 2.240352153778076, "step": 9392 }, { "epoch": 0.7404808829325975, "grad_norm": 0.6358786225318909, "learning_rate": 0.0005, "loss": 1.858986258506775, "step": 9393 }, { "epoch": 0.7405597162002365, "grad_norm": 0.5860949754714966, "learning_rate": 0.0005, "loss": 1.8498295545578003, "step": 9394 }, { "epoch": 0.7406385494678754, "grad_norm": 0.6979439854621887, "learning_rate": 0.0005, "loss": 1.7796804904937744, "step": 9395 }, { "epoch": 0.7407173827355144, "grad_norm": 0.929480254650116, "learning_rate": 0.0005, "loss": 1.7653121948242188, "step": 9396 }, { "epoch": 0.7407962160031534, "grad_norm": 0.5215401649475098, "learning_rate": 0.0005, "loss": 1.692287564277649, "step": 9397 }, { "epoch": 0.7408750492707923, "grad_norm": 0.6242921948432922, "learning_rate": 0.0005, "loss": 1.7481112480163574, "step": 9398 }, { "epoch": 0.7409538825384312, "grad_norm": 0.6243659853935242, "learning_rate": 0.0005, "loss": 2.0437347888946533, "step": 9399 }, { "epoch": 0.7410327158060701, "grad_norm": 0.6949470639228821, "learning_rate": 0.0005, "loss": 1.5450202226638794, "step": 9400 }, { "epoch": 0.7411115490737091, "grad_norm": 0.6421172022819519, "learning_rate": 0.0005, "loss": 1.7978309392929077, "step": 9401 }, { "epoch": 0.741190382341348, "grad_norm": 0.7640712857246399, "learning_rate": 0.0005, "loss": 1.8627731800079346, "step": 9402 }, { "epoch": 0.741269215608987, "grad_norm": 0.5950422286987305, "learning_rate": 0.0005, "loss": 1.8194870948791504, "step": 9403 }, { "epoch": 0.741348048876626, "grad_norm": 0.6296285390853882, "learning_rate": 0.0005, "loss": 1.8618298768997192, "step": 9404 }, { "epoch": 0.7414268821442649, "grad_norm": 0.6076847910881042, "learning_rate": 0.0005, "loss": 2.0402026176452637, "step": 9405 }, { "epoch": 0.7415057154119038, "grad_norm": 0.6331121921539307, "learning_rate": 0.0005, "loss": 1.9780962467193604, "step": 9406 }, { "epoch": 0.7415845486795428, "grad_norm": 0.7686463594436646, "learning_rate": 0.0005, "loss": 1.9764399528503418, "step": 9407 }, { "epoch": 0.7416633819471817, "grad_norm": 0.6587954163551331, "learning_rate": 0.0005, "loss": 2.127603769302368, "step": 9408 }, { "epoch": 0.7417422152148206, "grad_norm": 0.7151138782501221, "learning_rate": 0.0005, "loss": 1.7045577764511108, "step": 9409 }, { "epoch": 0.7418210484824596, "grad_norm": 0.6737441420555115, "learning_rate": 0.0005, "loss": 1.8013054132461548, "step": 9410 }, { "epoch": 0.7418998817500986, "grad_norm": 0.5130723714828491, "learning_rate": 0.0005, "loss": 1.8705371618270874, "step": 9411 }, { "epoch": 0.7419787150177375, "grad_norm": 0.5868210196495056, "learning_rate": 0.0005, "loss": 1.8955929279327393, "step": 9412 }, { "epoch": 0.7420575482853764, "grad_norm": 0.6405282020568848, "learning_rate": 0.0005, "loss": 2.083127975463867, "step": 9413 }, { "epoch": 0.7421363815530154, "grad_norm": 1.0602688789367676, "learning_rate": 0.0005, "loss": 1.940481424331665, "step": 9414 }, { "epoch": 0.7422152148206543, "grad_norm": 0.6163985729217529, "learning_rate": 0.0005, "loss": 1.869401216506958, "step": 9415 }, { "epoch": 0.7422940480882932, "grad_norm": 0.7802676558494568, "learning_rate": 0.0005, "loss": 2.074805498123169, "step": 9416 }, { "epoch": 0.7423728813559322, "grad_norm": 0.7036586403846741, "learning_rate": 0.0005, "loss": 1.9488688707351685, "step": 9417 }, { "epoch": 0.7424517146235712, "grad_norm": 0.7142435312271118, "learning_rate": 0.0005, "loss": 2.1045289039611816, "step": 9418 }, { "epoch": 0.7425305478912101, "grad_norm": 0.6797157526016235, "learning_rate": 0.0005, "loss": 1.8975210189819336, "step": 9419 }, { "epoch": 0.742609381158849, "grad_norm": 0.6535142660140991, "learning_rate": 0.0005, "loss": 1.8210071325302124, "step": 9420 }, { "epoch": 0.742688214426488, "grad_norm": 0.6172381043434143, "learning_rate": 0.0005, "loss": 1.8476154804229736, "step": 9421 }, { "epoch": 0.7427670476941269, "grad_norm": 0.5951502919197083, "learning_rate": 0.0005, "loss": 1.7878386974334717, "step": 9422 }, { "epoch": 0.7428458809617658, "grad_norm": 0.8341038823127747, "learning_rate": 0.0005, "loss": 2.055222988128662, "step": 9423 }, { "epoch": 0.7429247142294048, "grad_norm": 0.6393533945083618, "learning_rate": 0.0005, "loss": 1.738761305809021, "step": 9424 }, { "epoch": 0.7430035474970438, "grad_norm": 0.7115991711616516, "learning_rate": 0.0005, "loss": 1.8755028247833252, "step": 9425 }, { "epoch": 0.7430823807646827, "grad_norm": 0.981277585029602, "learning_rate": 0.0005, "loss": 1.709221363067627, "step": 9426 }, { "epoch": 0.7431612140323216, "grad_norm": 0.8188973069190979, "learning_rate": 0.0005, "loss": 1.7964303493499756, "step": 9427 }, { "epoch": 0.7432400472999606, "grad_norm": 0.9960445165634155, "learning_rate": 0.0005, "loss": 2.042606830596924, "step": 9428 }, { "epoch": 0.7433188805675995, "grad_norm": 0.6750003099441528, "learning_rate": 0.0005, "loss": 1.6218504905700684, "step": 9429 }, { "epoch": 0.7433977138352384, "grad_norm": 0.8220221400260925, "learning_rate": 0.0005, "loss": 2.0969417095184326, "step": 9430 }, { "epoch": 0.7434765471028774, "grad_norm": 0.6630042791366577, "learning_rate": 0.0005, "loss": 1.8055369853973389, "step": 9431 }, { "epoch": 0.7435553803705164, "grad_norm": 0.5545400381088257, "learning_rate": 0.0005, "loss": 1.9182159900665283, "step": 9432 }, { "epoch": 0.7436342136381553, "grad_norm": 0.6724621057510376, "learning_rate": 0.0005, "loss": 2.285407066345215, "step": 9433 }, { "epoch": 0.7437130469057942, "grad_norm": 0.7336643934249878, "learning_rate": 0.0005, "loss": 1.9845901727676392, "step": 9434 }, { "epoch": 0.7437918801734332, "grad_norm": 0.7529267072677612, "learning_rate": 0.0005, "loss": 2.0279219150543213, "step": 9435 }, { "epoch": 0.7438707134410721, "grad_norm": 0.6473482847213745, "learning_rate": 0.0005, "loss": 1.8692783117294312, "step": 9436 }, { "epoch": 0.743949546708711, "grad_norm": 0.7713811993598938, "learning_rate": 0.0005, "loss": 1.9075895547866821, "step": 9437 }, { "epoch": 0.7440283799763501, "grad_norm": 0.6847334504127502, "learning_rate": 0.0005, "loss": 1.8680477142333984, "step": 9438 }, { "epoch": 0.744107213243989, "grad_norm": 0.8041982650756836, "learning_rate": 0.0005, "loss": 2.132695198059082, "step": 9439 }, { "epoch": 0.7441860465116279, "grad_norm": 0.6636675000190735, "learning_rate": 0.0005, "loss": 2.2217774391174316, "step": 9440 }, { "epoch": 0.7442648797792668, "grad_norm": 0.6225996613502502, "learning_rate": 0.0005, "loss": 1.7063970565795898, "step": 9441 }, { "epoch": 0.7443437130469058, "grad_norm": 0.9043843746185303, "learning_rate": 0.0005, "loss": 2.106266975402832, "step": 9442 }, { "epoch": 0.7444225463145447, "grad_norm": 0.7803217768669128, "learning_rate": 0.0005, "loss": 1.9295918941497803, "step": 9443 }, { "epoch": 0.7445013795821837, "grad_norm": 0.8032199740409851, "learning_rate": 0.0005, "loss": 1.8918747901916504, "step": 9444 }, { "epoch": 0.7445802128498227, "grad_norm": 0.6163333058357239, "learning_rate": 0.0005, "loss": 1.7276020050048828, "step": 9445 }, { "epoch": 0.7446590461174616, "grad_norm": 0.5947667956352234, "learning_rate": 0.0005, "loss": 1.670823097229004, "step": 9446 }, { "epoch": 0.7447378793851005, "grad_norm": 0.6234520077705383, "learning_rate": 0.0005, "loss": 1.8488085269927979, "step": 9447 }, { "epoch": 0.7448167126527394, "grad_norm": 0.8129595518112183, "learning_rate": 0.0005, "loss": 1.945617437362671, "step": 9448 }, { "epoch": 0.7448955459203784, "grad_norm": 0.70818692445755, "learning_rate": 0.0005, "loss": 2.0425772666931152, "step": 9449 }, { "epoch": 0.7449743791880173, "grad_norm": 0.8825316429138184, "learning_rate": 0.0005, "loss": 1.8023018836975098, "step": 9450 }, { "epoch": 0.7450532124556563, "grad_norm": 0.5812469124794006, "learning_rate": 0.0005, "loss": 2.0896167755126953, "step": 9451 }, { "epoch": 0.7451320457232953, "grad_norm": 0.6321719288825989, "learning_rate": 0.0005, "loss": 1.6862314939498901, "step": 9452 }, { "epoch": 0.7452108789909342, "grad_norm": 0.7495690584182739, "learning_rate": 0.0005, "loss": 1.7010905742645264, "step": 9453 }, { "epoch": 0.7452897122585731, "grad_norm": 0.6294728517532349, "learning_rate": 0.0005, "loss": 1.8974385261535645, "step": 9454 }, { "epoch": 0.745368545526212, "grad_norm": 0.7153623700141907, "learning_rate": 0.0005, "loss": 1.7092225551605225, "step": 9455 }, { "epoch": 0.745447378793851, "grad_norm": 0.586287796497345, "learning_rate": 0.0005, "loss": 1.916864275932312, "step": 9456 }, { "epoch": 0.74552621206149, "grad_norm": 0.8471535444259644, "learning_rate": 0.0005, "loss": 2.1238956451416016, "step": 9457 }, { "epoch": 0.7456050453291289, "grad_norm": 0.5566790699958801, "learning_rate": 0.0005, "loss": 1.9395959377288818, "step": 9458 }, { "epoch": 0.7456838785967679, "grad_norm": 0.601482629776001, "learning_rate": 0.0005, "loss": 1.809140920639038, "step": 9459 }, { "epoch": 0.7457627118644068, "grad_norm": 0.8390362858772278, "learning_rate": 0.0005, "loss": 1.7957050800323486, "step": 9460 }, { "epoch": 0.7458415451320457, "grad_norm": 0.8408161997795105, "learning_rate": 0.0005, "loss": 1.9412693977355957, "step": 9461 }, { "epoch": 0.7459203783996846, "grad_norm": 0.6349329352378845, "learning_rate": 0.0005, "loss": 1.7223780155181885, "step": 9462 }, { "epoch": 0.7459992116673236, "grad_norm": 0.7049526572227478, "learning_rate": 0.0005, "loss": 1.798529863357544, "step": 9463 }, { "epoch": 0.7460780449349625, "grad_norm": 0.7503758072853088, "learning_rate": 0.0005, "loss": 1.8833870887756348, "step": 9464 }, { "epoch": 0.7461568782026015, "grad_norm": 0.6127842664718628, "learning_rate": 0.0005, "loss": 1.768512487411499, "step": 9465 }, { "epoch": 0.7462357114702405, "grad_norm": 0.7770593166351318, "learning_rate": 0.0005, "loss": 1.7461004257202148, "step": 9466 }, { "epoch": 0.7463145447378794, "grad_norm": 0.7181020975112915, "learning_rate": 0.0005, "loss": 1.941760540008545, "step": 9467 }, { "epoch": 0.7463933780055183, "grad_norm": 0.7387993931770325, "learning_rate": 0.0005, "loss": 1.858577013015747, "step": 9468 }, { "epoch": 0.7464722112731573, "grad_norm": 0.7565706372261047, "learning_rate": 0.0005, "loss": 2.0132288932800293, "step": 9469 }, { "epoch": 0.7465510445407962, "grad_norm": 0.6837952136993408, "learning_rate": 0.0005, "loss": 1.7050447463989258, "step": 9470 }, { "epoch": 0.7466298778084351, "grad_norm": 0.7890509366989136, "learning_rate": 0.0005, "loss": 1.7850022315979004, "step": 9471 }, { "epoch": 0.7467087110760741, "grad_norm": 0.7154546976089478, "learning_rate": 0.0005, "loss": 1.8183085918426514, "step": 9472 }, { "epoch": 0.7467875443437131, "grad_norm": 0.9724082350730896, "learning_rate": 0.0005, "loss": 2.1149282455444336, "step": 9473 }, { "epoch": 0.746866377611352, "grad_norm": 0.7877157926559448, "learning_rate": 0.0005, "loss": 1.875256061553955, "step": 9474 }, { "epoch": 0.7469452108789909, "grad_norm": 0.8293591737747192, "learning_rate": 0.0005, "loss": 1.7433507442474365, "step": 9475 }, { "epoch": 0.7470240441466299, "grad_norm": 0.7069013714790344, "learning_rate": 0.0005, "loss": 2.0698132514953613, "step": 9476 }, { "epoch": 0.7471028774142688, "grad_norm": 0.9460427761077881, "learning_rate": 0.0005, "loss": 2.1798977851867676, "step": 9477 }, { "epoch": 0.7471817106819078, "grad_norm": 0.7561049461364746, "learning_rate": 0.0005, "loss": 1.785122036933899, "step": 9478 }, { "epoch": 0.7472605439495467, "grad_norm": 0.8396790027618408, "learning_rate": 0.0005, "loss": 1.8882427215576172, "step": 9479 }, { "epoch": 0.7473393772171857, "grad_norm": 0.6637889742851257, "learning_rate": 0.0005, "loss": 1.9099292755126953, "step": 9480 }, { "epoch": 0.7474182104848246, "grad_norm": 0.9719801545143127, "learning_rate": 0.0005, "loss": 2.0569005012512207, "step": 9481 }, { "epoch": 0.7474970437524635, "grad_norm": 0.9107517004013062, "learning_rate": 0.0005, "loss": 1.7441656589508057, "step": 9482 }, { "epoch": 0.7475758770201025, "grad_norm": 0.7859469652175903, "learning_rate": 0.0005, "loss": 1.8328557014465332, "step": 9483 }, { "epoch": 0.7476547102877414, "grad_norm": 0.7045490145683289, "learning_rate": 0.0005, "loss": 2.123648166656494, "step": 9484 }, { "epoch": 0.7477335435553804, "grad_norm": 0.6018396019935608, "learning_rate": 0.0005, "loss": 1.746429443359375, "step": 9485 }, { "epoch": 0.7478123768230193, "grad_norm": 0.669033408164978, "learning_rate": 0.0005, "loss": 1.8897383213043213, "step": 9486 }, { "epoch": 0.7478912100906583, "grad_norm": 0.801735520362854, "learning_rate": 0.0005, "loss": 2.0370168685913086, "step": 9487 }, { "epoch": 0.7479700433582972, "grad_norm": 0.6456570625305176, "learning_rate": 0.0005, "loss": 1.7446281909942627, "step": 9488 }, { "epoch": 0.7480488766259361, "grad_norm": 0.5549982786178589, "learning_rate": 0.0005, "loss": 1.6473777294158936, "step": 9489 }, { "epoch": 0.7481277098935751, "grad_norm": 0.7107434272766113, "learning_rate": 0.0005, "loss": 1.897111415863037, "step": 9490 }, { "epoch": 0.748206543161214, "grad_norm": 5.938976764678955, "learning_rate": 0.0005, "loss": 1.6969655752182007, "step": 9491 }, { "epoch": 0.748285376428853, "grad_norm": 0.5220922827720642, "learning_rate": 0.0005, "loss": 1.8385577201843262, "step": 9492 }, { "epoch": 0.7483642096964919, "grad_norm": 0.5780250430107117, "learning_rate": 0.0005, "loss": 1.818175196647644, "step": 9493 }, { "epoch": 0.7484430429641309, "grad_norm": 0.5331944823265076, "learning_rate": 0.0005, "loss": 1.7276332378387451, "step": 9494 }, { "epoch": 0.7485218762317698, "grad_norm": 0.6907032132148743, "learning_rate": 0.0005, "loss": 1.9447065591812134, "step": 9495 }, { "epoch": 0.7486007094994087, "grad_norm": 0.8424057364463806, "learning_rate": 0.0005, "loss": 2.136902332305908, "step": 9496 }, { "epoch": 0.7486795427670477, "grad_norm": 0.8648026585578918, "learning_rate": 0.0005, "loss": 1.9466395378112793, "step": 9497 }, { "epoch": 0.7487583760346866, "grad_norm": 0.6391685605049133, "learning_rate": 0.0005, "loss": 1.925161600112915, "step": 9498 }, { "epoch": 0.7488372093023256, "grad_norm": 0.662695586681366, "learning_rate": 0.0005, "loss": 1.8023536205291748, "step": 9499 }, { "epoch": 0.7489160425699646, "grad_norm": 0.793180525302887, "learning_rate": 0.0005, "loss": 1.640744686126709, "step": 9500 }, { "epoch": 0.7489948758376035, "grad_norm": 0.9535254240036011, "learning_rate": 0.0005, "loss": 2.102997064590454, "step": 9501 }, { "epoch": 0.7490737091052424, "grad_norm": 0.6445890665054321, "learning_rate": 0.0005, "loss": 2.212419033050537, "step": 9502 }, { "epoch": 0.7491525423728813, "grad_norm": 0.6137818694114685, "learning_rate": 0.0005, "loss": 2.037705183029175, "step": 9503 }, { "epoch": 0.7492313756405203, "grad_norm": 0.5884995460510254, "learning_rate": 0.0005, "loss": 1.7128162384033203, "step": 9504 }, { "epoch": 0.7493102089081592, "grad_norm": 0.5914112329483032, "learning_rate": 0.0005, "loss": 1.9285888671875, "step": 9505 }, { "epoch": 0.7493890421757982, "grad_norm": 0.705950915813446, "learning_rate": 0.0005, "loss": 2.0667779445648193, "step": 9506 }, { "epoch": 0.7494678754434372, "grad_norm": 0.6747382283210754, "learning_rate": 0.0005, "loss": 1.864668846130371, "step": 9507 }, { "epoch": 0.7495467087110761, "grad_norm": 0.7928435802459717, "learning_rate": 0.0005, "loss": 1.948056697845459, "step": 9508 }, { "epoch": 0.749625541978715, "grad_norm": 0.7316873073577881, "learning_rate": 0.0005, "loss": 1.980168342590332, "step": 9509 }, { "epoch": 0.7497043752463539, "grad_norm": 0.718750536441803, "learning_rate": 0.0005, "loss": 1.8734393119812012, "step": 9510 }, { "epoch": 0.7497832085139929, "grad_norm": 0.8118669390678406, "learning_rate": 0.0005, "loss": 1.797528624534607, "step": 9511 }, { "epoch": 0.7498620417816318, "grad_norm": 0.9346274733543396, "learning_rate": 0.0005, "loss": 1.934675931930542, "step": 9512 }, { "epoch": 0.7499408750492708, "grad_norm": 0.5939444899559021, "learning_rate": 0.0005, "loss": 1.7789413928985596, "step": 9513 }, { "epoch": 0.7500197083169098, "grad_norm": 0.6857068538665771, "learning_rate": 0.0005, "loss": 1.764451265335083, "step": 9514 }, { "epoch": 0.7500985415845487, "grad_norm": 0.6143438220024109, "learning_rate": 0.0005, "loss": 1.9802302122116089, "step": 9515 }, { "epoch": 0.7501773748521876, "grad_norm": 0.7378582954406738, "learning_rate": 0.0005, "loss": 1.8777971267700195, "step": 9516 }, { "epoch": 0.7502562081198265, "grad_norm": 0.8611502051353455, "learning_rate": 0.0005, "loss": 1.8627986907958984, "step": 9517 }, { "epoch": 0.7503350413874655, "grad_norm": 0.6738184094429016, "learning_rate": 0.0005, "loss": 2.2769322395324707, "step": 9518 }, { "epoch": 0.7504138746551045, "grad_norm": 0.5980923771858215, "learning_rate": 0.0005, "loss": 2.1356635093688965, "step": 9519 }, { "epoch": 0.7504927079227434, "grad_norm": 0.6586607694625854, "learning_rate": 0.0005, "loss": 1.8829160928726196, "step": 9520 }, { "epoch": 0.7505715411903824, "grad_norm": 0.7256748080253601, "learning_rate": 0.0005, "loss": 1.930692434310913, "step": 9521 }, { "epoch": 0.7506503744580213, "grad_norm": 0.7280701398849487, "learning_rate": 0.0005, "loss": 1.9734035730361938, "step": 9522 }, { "epoch": 0.7507292077256602, "grad_norm": 0.6641029715538025, "learning_rate": 0.0005, "loss": 1.8553969860076904, "step": 9523 }, { "epoch": 0.7508080409932991, "grad_norm": 0.6494272947311401, "learning_rate": 0.0005, "loss": 1.854858636856079, "step": 9524 }, { "epoch": 0.7508868742609381, "grad_norm": 0.632293701171875, "learning_rate": 0.0005, "loss": 2.1403696537017822, "step": 9525 }, { "epoch": 0.750965707528577, "grad_norm": 0.7157517075538635, "learning_rate": 0.0005, "loss": 1.9962966442108154, "step": 9526 }, { "epoch": 0.751044540796216, "grad_norm": 0.6785523295402527, "learning_rate": 0.0005, "loss": 1.8660458326339722, "step": 9527 }, { "epoch": 0.751123374063855, "grad_norm": 0.6640981435775757, "learning_rate": 0.0005, "loss": 1.776436448097229, "step": 9528 }, { "epoch": 0.7512022073314939, "grad_norm": 0.7512786388397217, "learning_rate": 0.0005, "loss": 1.7350490093231201, "step": 9529 }, { "epoch": 0.7512810405991328, "grad_norm": 0.5327550768852234, "learning_rate": 0.0005, "loss": 1.9253917932510376, "step": 9530 }, { "epoch": 0.7513598738667718, "grad_norm": 0.5296353697776794, "learning_rate": 0.0005, "loss": 1.795757532119751, "step": 9531 }, { "epoch": 0.7514387071344107, "grad_norm": 0.7461438775062561, "learning_rate": 0.0005, "loss": 1.7903352975845337, "step": 9532 }, { "epoch": 0.7515175404020497, "grad_norm": 0.6780737042427063, "learning_rate": 0.0005, "loss": 1.9021544456481934, "step": 9533 }, { "epoch": 0.7515963736696886, "grad_norm": 0.8810786008834839, "learning_rate": 0.0005, "loss": 2.147088050842285, "step": 9534 }, { "epoch": 0.7516752069373276, "grad_norm": 0.5851035714149475, "learning_rate": 0.0005, "loss": 1.885951042175293, "step": 9535 }, { "epoch": 0.7517540402049665, "grad_norm": 0.6352204084396362, "learning_rate": 0.0005, "loss": 1.9053399562835693, "step": 9536 }, { "epoch": 0.7518328734726054, "grad_norm": 0.7132212519645691, "learning_rate": 0.0005, "loss": 1.8079464435577393, "step": 9537 }, { "epoch": 0.7519117067402444, "grad_norm": 0.6654767990112305, "learning_rate": 0.0005, "loss": 1.8498594760894775, "step": 9538 }, { "epoch": 0.7519905400078833, "grad_norm": 0.7197739481925964, "learning_rate": 0.0005, "loss": 1.9643663167953491, "step": 9539 }, { "epoch": 0.7520693732755223, "grad_norm": 0.6011229753494263, "learning_rate": 0.0005, "loss": 2.2138619422912598, "step": 9540 }, { "epoch": 0.7521482065431612, "grad_norm": 0.7549465894699097, "learning_rate": 0.0005, "loss": 1.7690341472625732, "step": 9541 }, { "epoch": 0.7522270398108002, "grad_norm": 0.7769417762756348, "learning_rate": 0.0005, "loss": 2.1240386962890625, "step": 9542 }, { "epoch": 0.7523058730784391, "grad_norm": 0.7738850116729736, "learning_rate": 0.0005, "loss": 2.11641788482666, "step": 9543 }, { "epoch": 0.752384706346078, "grad_norm": 1.0134069919586182, "learning_rate": 0.0005, "loss": 1.887312889099121, "step": 9544 }, { "epoch": 0.752463539613717, "grad_norm": 0.8920906186103821, "learning_rate": 0.0005, "loss": 1.8243080377578735, "step": 9545 }, { "epoch": 0.752542372881356, "grad_norm": 0.735536515712738, "learning_rate": 0.0005, "loss": 1.910412311553955, "step": 9546 }, { "epoch": 0.7526212061489949, "grad_norm": 0.8010821342468262, "learning_rate": 0.0005, "loss": 1.9463162422180176, "step": 9547 }, { "epoch": 0.7527000394166338, "grad_norm": 0.6073998808860779, "learning_rate": 0.0005, "loss": 1.8432667255401611, "step": 9548 }, { "epoch": 0.7527788726842728, "grad_norm": 0.7809390425682068, "learning_rate": 0.0005, "loss": 2.093147039413452, "step": 9549 }, { "epoch": 0.7528577059519117, "grad_norm": 0.6733795404434204, "learning_rate": 0.0005, "loss": 1.8083831071853638, "step": 9550 }, { "epoch": 0.7529365392195506, "grad_norm": 0.7204479575157166, "learning_rate": 0.0005, "loss": 1.9627957344055176, "step": 9551 }, { "epoch": 0.7530153724871896, "grad_norm": 0.768606424331665, "learning_rate": 0.0005, "loss": 1.9803524017333984, "step": 9552 }, { "epoch": 0.7530942057548285, "grad_norm": 0.6190979480743408, "learning_rate": 0.0005, "loss": 2.1522953510284424, "step": 9553 }, { "epoch": 0.7531730390224675, "grad_norm": 0.6495590209960938, "learning_rate": 0.0005, "loss": 2.1017751693725586, "step": 9554 }, { "epoch": 0.7532518722901064, "grad_norm": 0.6613282561302185, "learning_rate": 0.0005, "loss": 2.0207693576812744, "step": 9555 }, { "epoch": 0.7533307055577454, "grad_norm": 0.7618421912193298, "learning_rate": 0.0005, "loss": 1.8343504667282104, "step": 9556 }, { "epoch": 0.7534095388253843, "grad_norm": 0.7562319040298462, "learning_rate": 0.0005, "loss": 1.9362322092056274, "step": 9557 }, { "epoch": 0.7534883720930232, "grad_norm": 0.764877438545227, "learning_rate": 0.0005, "loss": 1.9544928073883057, "step": 9558 }, { "epoch": 0.7535672053606622, "grad_norm": 0.7298498749732971, "learning_rate": 0.0005, "loss": 1.693576455116272, "step": 9559 }, { "epoch": 0.7536460386283012, "grad_norm": 0.6820181608200073, "learning_rate": 0.0005, "loss": 1.8865869045257568, "step": 9560 }, { "epoch": 0.7537248718959401, "grad_norm": 0.8165816068649292, "learning_rate": 0.0005, "loss": 2.1870639324188232, "step": 9561 }, { "epoch": 0.7538037051635791, "grad_norm": 0.5898927450180054, "learning_rate": 0.0005, "loss": 2.104133129119873, "step": 9562 }, { "epoch": 0.753882538431218, "grad_norm": 0.5957799553871155, "learning_rate": 0.0005, "loss": 1.7773364782333374, "step": 9563 }, { "epoch": 0.7539613716988569, "grad_norm": 0.6504945755004883, "learning_rate": 0.0005, "loss": 2.129768133163452, "step": 9564 }, { "epoch": 0.7540402049664958, "grad_norm": 0.573940098285675, "learning_rate": 0.0005, "loss": 1.6927297115325928, "step": 9565 }, { "epoch": 0.7541190382341348, "grad_norm": 0.7184739708900452, "learning_rate": 0.0005, "loss": 1.9211963415145874, "step": 9566 }, { "epoch": 0.7541978715017738, "grad_norm": 0.6323495507240295, "learning_rate": 0.0005, "loss": 1.6747725009918213, "step": 9567 }, { "epoch": 0.7542767047694127, "grad_norm": 0.7963364124298096, "learning_rate": 0.0005, "loss": 1.9392635822296143, "step": 9568 }, { "epoch": 0.7543555380370517, "grad_norm": 0.6712313294410706, "learning_rate": 0.0005, "loss": 1.9985604286193848, "step": 9569 }, { "epoch": 0.7544343713046906, "grad_norm": 1.0103485584259033, "learning_rate": 0.0005, "loss": 2.1508758068084717, "step": 9570 }, { "epoch": 0.7545132045723295, "grad_norm": 0.7430096864700317, "learning_rate": 0.0005, "loss": 1.9507296085357666, "step": 9571 }, { "epoch": 0.7545920378399684, "grad_norm": 0.6597070693969727, "learning_rate": 0.0005, "loss": 1.401767611503601, "step": 9572 }, { "epoch": 0.7546708711076074, "grad_norm": 0.6738054752349854, "learning_rate": 0.0005, "loss": 1.6562291383743286, "step": 9573 }, { "epoch": 0.7547497043752464, "grad_norm": 0.815472424030304, "learning_rate": 0.0005, "loss": 1.9696390628814697, "step": 9574 }, { "epoch": 0.7548285376428853, "grad_norm": 0.704196572303772, "learning_rate": 0.0005, "loss": 2.2813448905944824, "step": 9575 }, { "epoch": 0.7549073709105243, "grad_norm": 0.5858284831047058, "learning_rate": 0.0005, "loss": 1.7282801866531372, "step": 9576 }, { "epoch": 0.7549862041781632, "grad_norm": 0.5848071575164795, "learning_rate": 0.0005, "loss": 1.8343675136566162, "step": 9577 }, { "epoch": 0.7550650374458021, "grad_norm": 1.3203685283660889, "learning_rate": 0.0005, "loss": 1.6772453784942627, "step": 9578 }, { "epoch": 0.755143870713441, "grad_norm": 3.0391130447387695, "learning_rate": 0.0005, "loss": 2.0429892539978027, "step": 9579 }, { "epoch": 0.75522270398108, "grad_norm": 0.7923193573951721, "learning_rate": 0.0005, "loss": 1.850250482559204, "step": 9580 }, { "epoch": 0.755301537248719, "grad_norm": 0.7694003582000732, "learning_rate": 0.0005, "loss": 2.160785436630249, "step": 9581 }, { "epoch": 0.7553803705163579, "grad_norm": 0.773237943649292, "learning_rate": 0.0005, "loss": 1.7861156463623047, "step": 9582 }, { "epoch": 0.7554592037839969, "grad_norm": 0.6849798560142517, "learning_rate": 0.0005, "loss": 1.9469640254974365, "step": 9583 }, { "epoch": 0.7555380370516358, "grad_norm": 0.927285373210907, "learning_rate": 0.0005, "loss": 2.1330370903015137, "step": 9584 }, { "epoch": 0.7556168703192747, "grad_norm": 0.624186098575592, "learning_rate": 0.0005, "loss": 1.669286847114563, "step": 9585 }, { "epoch": 0.7556957035869136, "grad_norm": 0.8701735734939575, "learning_rate": 0.0005, "loss": 2.0168309211730957, "step": 9586 }, { "epoch": 0.7557745368545526, "grad_norm": 0.7397705316543579, "learning_rate": 0.0005, "loss": 1.7827117443084717, "step": 9587 }, { "epoch": 0.7558533701221916, "grad_norm": 0.8716285228729248, "learning_rate": 0.0005, "loss": 1.8682594299316406, "step": 9588 }, { "epoch": 0.7559322033898305, "grad_norm": 0.714983344078064, "learning_rate": 0.0005, "loss": 1.87761390209198, "step": 9589 }, { "epoch": 0.7560110366574695, "grad_norm": 0.7139299511909485, "learning_rate": 0.0005, "loss": 2.080296516418457, "step": 9590 }, { "epoch": 0.7560898699251084, "grad_norm": 0.70094233751297, "learning_rate": 0.0005, "loss": 1.9591768980026245, "step": 9591 }, { "epoch": 0.7561687031927473, "grad_norm": 0.746668815612793, "learning_rate": 0.0005, "loss": 1.847060203552246, "step": 9592 }, { "epoch": 0.7562475364603863, "grad_norm": 0.6540949940681458, "learning_rate": 0.0005, "loss": 1.9988700151443481, "step": 9593 }, { "epoch": 0.7563263697280252, "grad_norm": 0.5942267179489136, "learning_rate": 0.0005, "loss": 1.9348453283309937, "step": 9594 }, { "epoch": 0.7564052029956642, "grad_norm": 0.6450237035751343, "learning_rate": 0.0005, "loss": 1.85713791847229, "step": 9595 }, { "epoch": 0.7564840362633031, "grad_norm": 0.8797244429588318, "learning_rate": 0.0005, "loss": 1.6179471015930176, "step": 9596 }, { "epoch": 0.7565628695309421, "grad_norm": 0.6022152304649353, "learning_rate": 0.0005, "loss": 1.6800189018249512, "step": 9597 }, { "epoch": 0.756641702798581, "grad_norm": 0.7267169952392578, "learning_rate": 0.0005, "loss": 2.064171314239502, "step": 9598 }, { "epoch": 0.7567205360662199, "grad_norm": 0.5726354718208313, "learning_rate": 0.0005, "loss": 1.7729719877243042, "step": 9599 }, { "epoch": 0.7567993693338589, "grad_norm": 0.6954489350318909, "learning_rate": 0.0005, "loss": 2.016983985900879, "step": 9600 }, { "epoch": 0.7568782026014979, "grad_norm": 0.5841681361198425, "learning_rate": 0.0005, "loss": 2.0636816024780273, "step": 9601 }, { "epoch": 0.7569570358691368, "grad_norm": 0.701827883720398, "learning_rate": 0.0005, "loss": 1.769761085510254, "step": 9602 }, { "epoch": 0.7570358691367757, "grad_norm": 0.8084259033203125, "learning_rate": 0.0005, "loss": 1.9987496137619019, "step": 9603 }, { "epoch": 0.7571147024044147, "grad_norm": 0.5427303314208984, "learning_rate": 0.0005, "loss": 1.8430888652801514, "step": 9604 }, { "epoch": 0.7571935356720536, "grad_norm": 0.6766587495803833, "learning_rate": 0.0005, "loss": 2.0262038707733154, "step": 9605 }, { "epoch": 0.7572723689396925, "grad_norm": 0.655891478061676, "learning_rate": 0.0005, "loss": 1.6579537391662598, "step": 9606 }, { "epoch": 0.7573512022073315, "grad_norm": 0.6347250938415527, "learning_rate": 0.0005, "loss": 1.889564037322998, "step": 9607 }, { "epoch": 0.7574300354749705, "grad_norm": 0.6304407715797424, "learning_rate": 0.0005, "loss": 1.9345462322235107, "step": 9608 }, { "epoch": 0.7575088687426094, "grad_norm": 0.6335510611534119, "learning_rate": 0.0005, "loss": 1.7708381414413452, "step": 9609 }, { "epoch": 0.7575877020102483, "grad_norm": 0.6951086521148682, "learning_rate": 0.0005, "loss": 1.7172313928604126, "step": 9610 }, { "epoch": 0.7576665352778873, "grad_norm": 0.8636806607246399, "learning_rate": 0.0005, "loss": 2.0383591651916504, "step": 9611 }, { "epoch": 0.7577453685455262, "grad_norm": 0.6039167046546936, "learning_rate": 0.0005, "loss": 1.833572268486023, "step": 9612 }, { "epoch": 0.7578242018131651, "grad_norm": 0.6308411955833435, "learning_rate": 0.0005, "loss": 1.6216437816619873, "step": 9613 }, { "epoch": 0.7579030350808041, "grad_norm": 0.8220205307006836, "learning_rate": 0.0005, "loss": 2.007143020629883, "step": 9614 }, { "epoch": 0.757981868348443, "grad_norm": 0.670795738697052, "learning_rate": 0.0005, "loss": 2.179070472717285, "step": 9615 }, { "epoch": 0.758060701616082, "grad_norm": 0.6371515393257141, "learning_rate": 0.0005, "loss": 1.792628288269043, "step": 9616 }, { "epoch": 0.7581395348837209, "grad_norm": 0.7872962951660156, "learning_rate": 0.0005, "loss": 1.939854621887207, "step": 9617 }, { "epoch": 0.7582183681513599, "grad_norm": 0.7967979311943054, "learning_rate": 0.0005, "loss": 1.9826958179473877, "step": 9618 }, { "epoch": 0.7582972014189988, "grad_norm": 0.8993197679519653, "learning_rate": 0.0005, "loss": 1.92897367477417, "step": 9619 }, { "epoch": 0.7583760346866377, "grad_norm": 0.5434950590133667, "learning_rate": 0.0005, "loss": 1.8244569301605225, "step": 9620 }, { "epoch": 0.7584548679542767, "grad_norm": 0.6604675054550171, "learning_rate": 0.0005, "loss": 1.6196095943450928, "step": 9621 }, { "epoch": 0.7585337012219157, "grad_norm": 0.7030808925628662, "learning_rate": 0.0005, "loss": 2.0119121074676514, "step": 9622 }, { "epoch": 0.7586125344895546, "grad_norm": 0.6454131603240967, "learning_rate": 0.0005, "loss": 2.0392565727233887, "step": 9623 }, { "epoch": 0.7586913677571936, "grad_norm": 0.5673903226852417, "learning_rate": 0.0005, "loss": 2.205927848815918, "step": 9624 }, { "epoch": 0.7587702010248325, "grad_norm": 0.7989629507064819, "learning_rate": 0.0005, "loss": 2.017622232437134, "step": 9625 }, { "epoch": 0.7588490342924714, "grad_norm": 0.6853219866752625, "learning_rate": 0.0005, "loss": 1.9286818504333496, "step": 9626 }, { "epoch": 0.7589278675601103, "grad_norm": 0.6876373291015625, "learning_rate": 0.0005, "loss": 2.1334118843078613, "step": 9627 }, { "epoch": 0.7590067008277493, "grad_norm": 0.8902524709701538, "learning_rate": 0.0005, "loss": 2.1079869270324707, "step": 9628 }, { "epoch": 0.7590855340953883, "grad_norm": 0.6126108169555664, "learning_rate": 0.0005, "loss": 1.8498473167419434, "step": 9629 }, { "epoch": 0.7591643673630272, "grad_norm": 0.68550705909729, "learning_rate": 0.0005, "loss": 1.8302351236343384, "step": 9630 }, { "epoch": 0.7592432006306662, "grad_norm": 0.6760510206222534, "learning_rate": 0.0005, "loss": 2.03532338142395, "step": 9631 }, { "epoch": 0.7593220338983051, "grad_norm": 0.5628787875175476, "learning_rate": 0.0005, "loss": 2.0102362632751465, "step": 9632 }, { "epoch": 0.759400867165944, "grad_norm": 0.7442917823791504, "learning_rate": 0.0005, "loss": 1.868760347366333, "step": 9633 }, { "epoch": 0.7594797004335829, "grad_norm": 0.6568204760551453, "learning_rate": 0.0005, "loss": 1.9493765830993652, "step": 9634 }, { "epoch": 0.759558533701222, "grad_norm": 0.921615719795227, "learning_rate": 0.0005, "loss": 1.7800426483154297, "step": 9635 }, { "epoch": 0.7596373669688609, "grad_norm": 0.5987340807914734, "learning_rate": 0.0005, "loss": 1.7773187160491943, "step": 9636 }, { "epoch": 0.7597162002364998, "grad_norm": 0.6707016229629517, "learning_rate": 0.0005, "loss": 1.7529408931732178, "step": 9637 }, { "epoch": 0.7597950335041388, "grad_norm": 0.5714256167411804, "learning_rate": 0.0005, "loss": 1.8159184455871582, "step": 9638 }, { "epoch": 0.7598738667717777, "grad_norm": 0.9020292162895203, "learning_rate": 0.0005, "loss": 1.8538321256637573, "step": 9639 }, { "epoch": 0.7599527000394166, "grad_norm": 0.7602828145027161, "learning_rate": 0.0005, "loss": 1.8638525009155273, "step": 9640 }, { "epoch": 0.7600315333070555, "grad_norm": 0.6262767910957336, "learning_rate": 0.0005, "loss": 2.0784711837768555, "step": 9641 }, { "epoch": 0.7601103665746946, "grad_norm": 0.7141523361206055, "learning_rate": 0.0005, "loss": 2.1678342819213867, "step": 9642 }, { "epoch": 0.7601891998423335, "grad_norm": 0.6800837516784668, "learning_rate": 0.0005, "loss": 2.066469669342041, "step": 9643 }, { "epoch": 0.7602680331099724, "grad_norm": 0.7436042428016663, "learning_rate": 0.0005, "loss": 1.763553261756897, "step": 9644 }, { "epoch": 0.7603468663776114, "grad_norm": 0.6118333339691162, "learning_rate": 0.0005, "loss": 1.909172534942627, "step": 9645 }, { "epoch": 0.7604256996452503, "grad_norm": 0.6194436550140381, "learning_rate": 0.0005, "loss": 1.9112025499343872, "step": 9646 }, { "epoch": 0.7605045329128892, "grad_norm": 0.5914410352706909, "learning_rate": 0.0005, "loss": 1.8141851425170898, "step": 9647 }, { "epoch": 0.7605833661805281, "grad_norm": 0.6360521912574768, "learning_rate": 0.0005, "loss": 1.8041465282440186, "step": 9648 }, { "epoch": 0.7606621994481672, "grad_norm": 0.7153316736221313, "learning_rate": 0.0005, "loss": 1.8116202354431152, "step": 9649 }, { "epoch": 0.7607410327158061, "grad_norm": 0.6238380074501038, "learning_rate": 0.0005, "loss": 1.8116369247436523, "step": 9650 }, { "epoch": 0.760819865983445, "grad_norm": 0.4694344699382782, "learning_rate": 0.0005, "loss": 1.7786222696304321, "step": 9651 }, { "epoch": 0.760898699251084, "grad_norm": 0.6643207669258118, "learning_rate": 0.0005, "loss": 2.117783546447754, "step": 9652 }, { "epoch": 0.7609775325187229, "grad_norm": 0.8549937009811401, "learning_rate": 0.0005, "loss": 1.6621979475021362, "step": 9653 }, { "epoch": 0.7610563657863618, "grad_norm": 0.6263688802719116, "learning_rate": 0.0005, "loss": 1.8774510622024536, "step": 9654 }, { "epoch": 0.7611351990540007, "grad_norm": 0.6383907198905945, "learning_rate": 0.0005, "loss": 1.7596070766448975, "step": 9655 }, { "epoch": 0.7612140323216398, "grad_norm": 2.814974546432495, "learning_rate": 0.0005, "loss": 1.8262147903442383, "step": 9656 }, { "epoch": 0.7612928655892787, "grad_norm": 0.6817873120307922, "learning_rate": 0.0005, "loss": 2.0533275604248047, "step": 9657 }, { "epoch": 0.7613716988569176, "grad_norm": 0.5802472233772278, "learning_rate": 0.0005, "loss": 1.8637752532958984, "step": 9658 }, { "epoch": 0.7614505321245566, "grad_norm": 0.6312976479530334, "learning_rate": 0.0005, "loss": 1.854755163192749, "step": 9659 }, { "epoch": 0.7615293653921955, "grad_norm": 0.7173134684562683, "learning_rate": 0.0005, "loss": 2.287905216217041, "step": 9660 }, { "epoch": 0.7616081986598344, "grad_norm": 0.5678176283836365, "learning_rate": 0.0005, "loss": 1.8444759845733643, "step": 9661 }, { "epoch": 0.7616870319274734, "grad_norm": 0.6860433220863342, "learning_rate": 0.0005, "loss": 2.1236515045166016, "step": 9662 }, { "epoch": 0.7617658651951124, "grad_norm": 0.7786016464233398, "learning_rate": 0.0005, "loss": 1.9497430324554443, "step": 9663 }, { "epoch": 0.7618446984627513, "grad_norm": 0.7542541027069092, "learning_rate": 0.0005, "loss": 1.8465874195098877, "step": 9664 }, { "epoch": 0.7619235317303902, "grad_norm": 0.7568122744560242, "learning_rate": 0.0005, "loss": 1.8021042346954346, "step": 9665 }, { "epoch": 0.7620023649980292, "grad_norm": 0.5667560696601868, "learning_rate": 0.0005, "loss": 1.9161782264709473, "step": 9666 }, { "epoch": 0.7620811982656681, "grad_norm": 0.5600212216377258, "learning_rate": 0.0005, "loss": 2.0042030811309814, "step": 9667 }, { "epoch": 0.762160031533307, "grad_norm": 0.7440314292907715, "learning_rate": 0.0005, "loss": 2.0534262657165527, "step": 9668 }, { "epoch": 0.762238864800946, "grad_norm": 0.5803859233856201, "learning_rate": 0.0005, "loss": 1.5572291612625122, "step": 9669 }, { "epoch": 0.762317698068585, "grad_norm": 0.6532506346702576, "learning_rate": 0.0005, "loss": 1.9918714761734009, "step": 9670 }, { "epoch": 0.7623965313362239, "grad_norm": 0.5541456341743469, "learning_rate": 0.0005, "loss": 1.9277994632720947, "step": 9671 }, { "epoch": 0.7624753646038628, "grad_norm": 0.8744319677352905, "learning_rate": 0.0005, "loss": 1.8048338890075684, "step": 9672 }, { "epoch": 0.7625541978715018, "grad_norm": 0.8299760818481445, "learning_rate": 0.0005, "loss": 1.7933722734451294, "step": 9673 }, { "epoch": 0.7626330311391407, "grad_norm": 0.5036723017692566, "learning_rate": 0.0005, "loss": 1.6992597579956055, "step": 9674 }, { "epoch": 0.7627118644067796, "grad_norm": 0.7398864030838013, "learning_rate": 0.0005, "loss": 1.8427711725234985, "step": 9675 }, { "epoch": 0.7627906976744186, "grad_norm": 0.7526012659072876, "learning_rate": 0.0005, "loss": 2.016098976135254, "step": 9676 }, { "epoch": 0.7628695309420576, "grad_norm": 0.8867815136909485, "learning_rate": 0.0005, "loss": 2.189685106277466, "step": 9677 }, { "epoch": 0.7629483642096965, "grad_norm": 0.6711008548736572, "learning_rate": 0.0005, "loss": 2.0422730445861816, "step": 9678 }, { "epoch": 0.7630271974773354, "grad_norm": 0.660627007484436, "learning_rate": 0.0005, "loss": 1.8819873332977295, "step": 9679 }, { "epoch": 0.7631060307449744, "grad_norm": 0.6975915431976318, "learning_rate": 0.0005, "loss": 1.9362132549285889, "step": 9680 }, { "epoch": 0.7631848640126133, "grad_norm": 0.7712237238883972, "learning_rate": 0.0005, "loss": 1.8711851835250854, "step": 9681 }, { "epoch": 0.7632636972802522, "grad_norm": 0.7060180902481079, "learning_rate": 0.0005, "loss": 1.94450044631958, "step": 9682 }, { "epoch": 0.7633425305478913, "grad_norm": 0.7078857421875, "learning_rate": 0.0005, "loss": 1.8005127906799316, "step": 9683 }, { "epoch": 0.7634213638155302, "grad_norm": 0.8629178404808044, "learning_rate": 0.0005, "loss": 1.9150084257125854, "step": 9684 }, { "epoch": 0.7635001970831691, "grad_norm": 0.8699003458023071, "learning_rate": 0.0005, "loss": 1.9824395179748535, "step": 9685 }, { "epoch": 0.763579030350808, "grad_norm": 0.6686098575592041, "learning_rate": 0.0005, "loss": 2.026090621948242, "step": 9686 }, { "epoch": 0.763657863618447, "grad_norm": 0.5891409516334534, "learning_rate": 0.0005, "loss": 1.5523978471755981, "step": 9687 }, { "epoch": 0.7637366968860859, "grad_norm": 0.5805684924125671, "learning_rate": 0.0005, "loss": 1.9426040649414062, "step": 9688 }, { "epoch": 0.7638155301537248, "grad_norm": 0.8146694302558899, "learning_rate": 0.0005, "loss": 2.0443081855773926, "step": 9689 }, { "epoch": 0.7638943634213639, "grad_norm": 0.904092013835907, "learning_rate": 0.0005, "loss": 2.006895065307617, "step": 9690 }, { "epoch": 0.7639731966890028, "grad_norm": 1.1325128078460693, "learning_rate": 0.0005, "loss": 1.8958780765533447, "step": 9691 }, { "epoch": 0.7640520299566417, "grad_norm": 0.7354027628898621, "learning_rate": 0.0005, "loss": 1.8694127798080444, "step": 9692 }, { "epoch": 0.7641308632242807, "grad_norm": 1.0618773698806763, "learning_rate": 0.0005, "loss": 1.8286492824554443, "step": 9693 }, { "epoch": 0.7642096964919196, "grad_norm": 0.800609290599823, "learning_rate": 0.0005, "loss": 1.8704166412353516, "step": 9694 }, { "epoch": 0.7642885297595585, "grad_norm": 0.7318792939186096, "learning_rate": 0.0005, "loss": 2.105508327484131, "step": 9695 }, { "epoch": 0.7643673630271974, "grad_norm": 0.7464516162872314, "learning_rate": 0.0005, "loss": 1.9751474857330322, "step": 9696 }, { "epoch": 0.7644461962948365, "grad_norm": 0.5617482662200928, "learning_rate": 0.0005, "loss": 1.8033347129821777, "step": 9697 }, { "epoch": 0.7645250295624754, "grad_norm": 1.1398109197616577, "learning_rate": 0.0005, "loss": 1.9685636758804321, "step": 9698 }, { "epoch": 0.7646038628301143, "grad_norm": 0.6497039794921875, "learning_rate": 0.0005, "loss": 2.030517578125, "step": 9699 }, { "epoch": 0.7646826960977533, "grad_norm": 0.6965346932411194, "learning_rate": 0.0005, "loss": 1.8496379852294922, "step": 9700 }, { "epoch": 0.7647615293653922, "grad_norm": 0.6573374271392822, "learning_rate": 0.0005, "loss": 1.728206992149353, "step": 9701 }, { "epoch": 0.7648403626330311, "grad_norm": 0.633629560470581, "learning_rate": 0.0005, "loss": 1.5814461708068848, "step": 9702 }, { "epoch": 0.76491919590067, "grad_norm": 0.8019252419471741, "learning_rate": 0.0005, "loss": 1.8745179176330566, "step": 9703 }, { "epoch": 0.7649980291683091, "grad_norm": 0.7862555980682373, "learning_rate": 0.0005, "loss": 1.9882121086120605, "step": 9704 }, { "epoch": 0.765076862435948, "grad_norm": 1.0568619966506958, "learning_rate": 0.0005, "loss": 1.8174552917480469, "step": 9705 }, { "epoch": 0.7651556957035869, "grad_norm": 0.7181916832923889, "learning_rate": 0.0005, "loss": 1.7082964181900024, "step": 9706 }, { "epoch": 0.7652345289712259, "grad_norm": 0.5542250275611877, "learning_rate": 0.0005, "loss": 1.917541265487671, "step": 9707 }, { "epoch": 0.7653133622388648, "grad_norm": 0.5857599377632141, "learning_rate": 0.0005, "loss": 1.8012535572052002, "step": 9708 }, { "epoch": 0.7653921955065037, "grad_norm": 0.5560348033905029, "learning_rate": 0.0005, "loss": 1.9312853813171387, "step": 9709 }, { "epoch": 0.7654710287741426, "grad_norm": 0.7162504196166992, "learning_rate": 0.0005, "loss": 2.007331371307373, "step": 9710 }, { "epoch": 0.7655498620417817, "grad_norm": 0.6547414660453796, "learning_rate": 0.0005, "loss": 1.821505069732666, "step": 9711 }, { "epoch": 0.7656286953094206, "grad_norm": 0.6919723749160767, "learning_rate": 0.0005, "loss": 1.8100835084915161, "step": 9712 }, { "epoch": 0.7657075285770595, "grad_norm": 0.7707840204238892, "learning_rate": 0.0005, "loss": 1.7723143100738525, "step": 9713 }, { "epoch": 0.7657863618446985, "grad_norm": 0.9673981666564941, "learning_rate": 0.0005, "loss": 2.1099917888641357, "step": 9714 }, { "epoch": 0.7658651951123374, "grad_norm": 0.6191896200180054, "learning_rate": 0.0005, "loss": 1.9399466514587402, "step": 9715 }, { "epoch": 0.7659440283799763, "grad_norm": 0.690001904964447, "learning_rate": 0.0005, "loss": 2.063109874725342, "step": 9716 }, { "epoch": 0.7660228616476152, "grad_norm": 0.7301979660987854, "learning_rate": 0.0005, "loss": 2.2054386138916016, "step": 9717 }, { "epoch": 0.7661016949152543, "grad_norm": 0.5403182506561279, "learning_rate": 0.0005, "loss": 2.0209357738494873, "step": 9718 }, { "epoch": 0.7661805281828932, "grad_norm": 0.6064553260803223, "learning_rate": 0.0005, "loss": 1.993525743484497, "step": 9719 }, { "epoch": 0.7662593614505321, "grad_norm": 0.47742995619773865, "learning_rate": 0.0005, "loss": 1.5661178827285767, "step": 9720 }, { "epoch": 0.7663381947181711, "grad_norm": 0.5962353348731995, "learning_rate": 0.0005, "loss": 1.822866678237915, "step": 9721 }, { "epoch": 0.76641702798581, "grad_norm": 0.6955959796905518, "learning_rate": 0.0005, "loss": 1.6823229789733887, "step": 9722 }, { "epoch": 0.7664958612534489, "grad_norm": 0.6194590330123901, "learning_rate": 0.0005, "loss": 1.5914082527160645, "step": 9723 }, { "epoch": 0.766574694521088, "grad_norm": 0.5451357960700989, "learning_rate": 0.0005, "loss": 1.6739646196365356, "step": 9724 }, { "epoch": 0.7666535277887269, "grad_norm": 0.5707087516784668, "learning_rate": 0.0005, "loss": 1.793026089668274, "step": 9725 }, { "epoch": 0.7667323610563658, "grad_norm": 0.6786473393440247, "learning_rate": 0.0005, "loss": 2.0465965270996094, "step": 9726 }, { "epoch": 0.7668111943240047, "grad_norm": 0.6164670586585999, "learning_rate": 0.0005, "loss": 1.9379620552062988, "step": 9727 }, { "epoch": 0.7668900275916437, "grad_norm": 0.6140705347061157, "learning_rate": 0.0005, "loss": 1.8914735317230225, "step": 9728 }, { "epoch": 0.7669688608592826, "grad_norm": 0.6042851805686951, "learning_rate": 0.0005, "loss": 1.7186022996902466, "step": 9729 }, { "epoch": 0.7670476941269215, "grad_norm": 0.5971127152442932, "learning_rate": 0.0005, "loss": 1.7549480199813843, "step": 9730 }, { "epoch": 0.7671265273945606, "grad_norm": 0.7073298096656799, "learning_rate": 0.0005, "loss": 2.076423168182373, "step": 9731 }, { "epoch": 0.7672053606621995, "grad_norm": 0.5404890775680542, "learning_rate": 0.0005, "loss": 2.07729434967041, "step": 9732 }, { "epoch": 0.7672841939298384, "grad_norm": 0.959205687046051, "learning_rate": 0.0005, "loss": 1.9803440570831299, "step": 9733 }, { "epoch": 0.7673630271974773, "grad_norm": 0.769873321056366, "learning_rate": 0.0005, "loss": 2.047978639602661, "step": 9734 }, { "epoch": 0.7674418604651163, "grad_norm": 0.7931831479072571, "learning_rate": 0.0005, "loss": 1.67959463596344, "step": 9735 }, { "epoch": 0.7675206937327552, "grad_norm": 0.6519039273262024, "learning_rate": 0.0005, "loss": 1.912780523300171, "step": 9736 }, { "epoch": 0.7675995270003941, "grad_norm": 0.650351345539093, "learning_rate": 0.0005, "loss": 1.9707987308502197, "step": 9737 }, { "epoch": 0.7676783602680332, "grad_norm": 0.7801169157028198, "learning_rate": 0.0005, "loss": 1.8794708251953125, "step": 9738 }, { "epoch": 0.7677571935356721, "grad_norm": 0.6632278561592102, "learning_rate": 0.0005, "loss": 1.9545316696166992, "step": 9739 }, { "epoch": 0.767836026803311, "grad_norm": 0.6481016278266907, "learning_rate": 0.0005, "loss": 1.9888954162597656, "step": 9740 }, { "epoch": 0.7679148600709499, "grad_norm": 0.9004318714141846, "learning_rate": 0.0005, "loss": 1.9415903091430664, "step": 9741 }, { "epoch": 0.7679936933385889, "grad_norm": 0.616181492805481, "learning_rate": 0.0005, "loss": 1.759347677230835, "step": 9742 }, { "epoch": 0.7680725266062278, "grad_norm": 1.2357765436172485, "learning_rate": 0.0005, "loss": 1.8417823314666748, "step": 9743 }, { "epoch": 0.7681513598738667, "grad_norm": 0.5868474841117859, "learning_rate": 0.0005, "loss": 1.875890851020813, "step": 9744 }, { "epoch": 0.7682301931415058, "grad_norm": 0.6322609186172485, "learning_rate": 0.0005, "loss": 2.077667236328125, "step": 9745 }, { "epoch": 0.7683090264091447, "grad_norm": 0.6294344663619995, "learning_rate": 0.0005, "loss": 1.6201884746551514, "step": 9746 }, { "epoch": 0.7683878596767836, "grad_norm": 1.1432386636734009, "learning_rate": 0.0005, "loss": 1.881658673286438, "step": 9747 }, { "epoch": 0.7684666929444225, "grad_norm": 0.5866641402244568, "learning_rate": 0.0005, "loss": 1.8480191230773926, "step": 9748 }, { "epoch": 0.7685455262120615, "grad_norm": 0.6638316512107849, "learning_rate": 0.0005, "loss": 1.7823107242584229, "step": 9749 }, { "epoch": 0.7686243594797004, "grad_norm": 0.664998471736908, "learning_rate": 0.0005, "loss": 1.996128797531128, "step": 9750 }, { "epoch": 0.7687031927473393, "grad_norm": 0.702552318572998, "learning_rate": 0.0005, "loss": 2.3110578060150146, "step": 9751 }, { "epoch": 0.7687820260149784, "grad_norm": 0.5975324511528015, "learning_rate": 0.0005, "loss": 1.8988640308380127, "step": 9752 }, { "epoch": 0.7688608592826173, "grad_norm": 0.6880884170532227, "learning_rate": 0.0005, "loss": 1.8022197484970093, "step": 9753 }, { "epoch": 0.7689396925502562, "grad_norm": 0.5957682132720947, "learning_rate": 0.0005, "loss": 1.8821649551391602, "step": 9754 }, { "epoch": 0.7690185258178952, "grad_norm": 0.6008805632591248, "learning_rate": 0.0005, "loss": 2.062976598739624, "step": 9755 }, { "epoch": 0.7690973590855341, "grad_norm": 0.5674881935119629, "learning_rate": 0.0005, "loss": 1.9103922843933105, "step": 9756 }, { "epoch": 0.769176192353173, "grad_norm": 0.6310634613037109, "learning_rate": 0.0005, "loss": 1.9216983318328857, "step": 9757 }, { "epoch": 0.7692550256208119, "grad_norm": 0.530458927154541, "learning_rate": 0.0005, "loss": 1.8577739000320435, "step": 9758 }, { "epoch": 0.769333858888451, "grad_norm": 0.6267334222793579, "learning_rate": 0.0005, "loss": 1.8464019298553467, "step": 9759 }, { "epoch": 0.7694126921560899, "grad_norm": 0.6037139892578125, "learning_rate": 0.0005, "loss": 1.9843683242797852, "step": 9760 }, { "epoch": 0.7694915254237288, "grad_norm": 0.6300432682037354, "learning_rate": 0.0005, "loss": 1.9836825132369995, "step": 9761 }, { "epoch": 0.7695703586913678, "grad_norm": 0.7719058394432068, "learning_rate": 0.0005, "loss": 1.874476671218872, "step": 9762 }, { "epoch": 0.7696491919590067, "grad_norm": 0.6887869238853455, "learning_rate": 0.0005, "loss": 1.907083511352539, "step": 9763 }, { "epoch": 0.7697280252266456, "grad_norm": 0.7270832657814026, "learning_rate": 0.0005, "loss": 1.983835220336914, "step": 9764 }, { "epoch": 0.7698068584942845, "grad_norm": 0.6550728678703308, "learning_rate": 0.0005, "loss": 2.1223673820495605, "step": 9765 }, { "epoch": 0.7698856917619236, "grad_norm": 0.7435579299926758, "learning_rate": 0.0005, "loss": 2.012751817703247, "step": 9766 }, { "epoch": 0.7699645250295625, "grad_norm": 0.702879011631012, "learning_rate": 0.0005, "loss": 2.119410514831543, "step": 9767 }, { "epoch": 0.7700433582972014, "grad_norm": 0.744400680065155, "learning_rate": 0.0005, "loss": 2.3010640144348145, "step": 9768 }, { "epoch": 0.7701221915648404, "grad_norm": 0.7417020201683044, "learning_rate": 0.0005, "loss": 2.11368989944458, "step": 9769 }, { "epoch": 0.7702010248324793, "grad_norm": 0.7380865812301636, "learning_rate": 0.0005, "loss": 1.5423521995544434, "step": 9770 }, { "epoch": 0.7702798581001182, "grad_norm": 0.6627405285835266, "learning_rate": 0.0005, "loss": 1.8085886240005493, "step": 9771 }, { "epoch": 0.7703586913677571, "grad_norm": 0.7403175234794617, "learning_rate": 0.0005, "loss": 1.7855358123779297, "step": 9772 }, { "epoch": 0.7704375246353962, "grad_norm": 0.6789031624794006, "learning_rate": 0.0005, "loss": 1.5782499313354492, "step": 9773 }, { "epoch": 0.7705163579030351, "grad_norm": 0.7204723954200745, "learning_rate": 0.0005, "loss": 2.0700786113739014, "step": 9774 }, { "epoch": 0.770595191170674, "grad_norm": 0.7261170148849487, "learning_rate": 0.0005, "loss": 2.0677549839019775, "step": 9775 }, { "epoch": 0.770674024438313, "grad_norm": 0.7551696300506592, "learning_rate": 0.0005, "loss": 1.9977850914001465, "step": 9776 }, { "epoch": 0.7707528577059519, "grad_norm": 0.6824832558631897, "learning_rate": 0.0005, "loss": 1.8109681606292725, "step": 9777 }, { "epoch": 0.7708316909735908, "grad_norm": 0.702302098274231, "learning_rate": 0.0005, "loss": 1.9892181158065796, "step": 9778 }, { "epoch": 0.7709105242412297, "grad_norm": 0.6502611041069031, "learning_rate": 0.0005, "loss": 2.5095176696777344, "step": 9779 }, { "epoch": 0.7709893575088688, "grad_norm": 0.6498104333877563, "learning_rate": 0.0005, "loss": 1.9180184602737427, "step": 9780 }, { "epoch": 0.7710681907765077, "grad_norm": 0.6583666205406189, "learning_rate": 0.0005, "loss": 2.04402494430542, "step": 9781 }, { "epoch": 0.7711470240441466, "grad_norm": 0.8832109570503235, "learning_rate": 0.0005, "loss": 1.8561344146728516, "step": 9782 }, { "epoch": 0.7712258573117856, "grad_norm": 0.609546959400177, "learning_rate": 0.0005, "loss": 1.9060269594192505, "step": 9783 }, { "epoch": 0.7713046905794245, "grad_norm": 0.75091552734375, "learning_rate": 0.0005, "loss": 1.746770977973938, "step": 9784 }, { "epoch": 0.7713835238470634, "grad_norm": 0.6929930448532104, "learning_rate": 0.0005, "loss": 1.7205684185028076, "step": 9785 }, { "epoch": 0.7714623571147025, "grad_norm": 0.630841076374054, "learning_rate": 0.0005, "loss": 2.1283836364746094, "step": 9786 }, { "epoch": 0.7715411903823414, "grad_norm": 0.6556524038314819, "learning_rate": 0.0005, "loss": 1.810777187347412, "step": 9787 }, { "epoch": 0.7716200236499803, "grad_norm": 0.6758262515068054, "learning_rate": 0.0005, "loss": 1.8425248861312866, "step": 9788 }, { "epoch": 0.7716988569176192, "grad_norm": 0.6696859002113342, "learning_rate": 0.0005, "loss": 1.9688096046447754, "step": 9789 }, { "epoch": 0.7717776901852582, "grad_norm": 0.6164277791976929, "learning_rate": 0.0005, "loss": 1.9468357563018799, "step": 9790 }, { "epoch": 0.7718565234528971, "grad_norm": 0.7695998549461365, "learning_rate": 0.0005, "loss": 1.9093589782714844, "step": 9791 }, { "epoch": 0.771935356720536, "grad_norm": 0.6992902159690857, "learning_rate": 0.0005, "loss": 1.8519585132598877, "step": 9792 }, { "epoch": 0.7720141899881751, "grad_norm": 0.6938392519950867, "learning_rate": 0.0005, "loss": 2.0060648918151855, "step": 9793 }, { "epoch": 0.772093023255814, "grad_norm": 0.6317853927612305, "learning_rate": 0.0005, "loss": 1.6120134592056274, "step": 9794 }, { "epoch": 0.7721718565234529, "grad_norm": 0.690429151058197, "learning_rate": 0.0005, "loss": 1.9602833986282349, "step": 9795 }, { "epoch": 0.7722506897910918, "grad_norm": 0.8060763478279114, "learning_rate": 0.0005, "loss": 1.728285551071167, "step": 9796 }, { "epoch": 0.7723295230587308, "grad_norm": 0.5818153023719788, "learning_rate": 0.0005, "loss": 1.9087185859680176, "step": 9797 }, { "epoch": 0.7724083563263697, "grad_norm": 0.7989526987075806, "learning_rate": 0.0005, "loss": 1.7328133583068848, "step": 9798 }, { "epoch": 0.7724871895940086, "grad_norm": 0.5840961933135986, "learning_rate": 0.0005, "loss": 1.8726427555084229, "step": 9799 }, { "epoch": 0.7725660228616477, "grad_norm": 0.6440554261207581, "learning_rate": 0.0005, "loss": 2.1491293907165527, "step": 9800 }, { "epoch": 0.7726448561292866, "grad_norm": 0.7809963226318359, "learning_rate": 0.0005, "loss": 1.9820995330810547, "step": 9801 }, { "epoch": 0.7727236893969255, "grad_norm": 0.6442262530326843, "learning_rate": 0.0005, "loss": 1.961634874343872, "step": 9802 }, { "epoch": 0.7728025226645644, "grad_norm": 0.6808914542198181, "learning_rate": 0.0005, "loss": 1.7757840156555176, "step": 9803 }, { "epoch": 0.7728813559322034, "grad_norm": 0.6167582273483276, "learning_rate": 0.0005, "loss": 1.7532069683074951, "step": 9804 }, { "epoch": 0.7729601891998423, "grad_norm": 0.6222337484359741, "learning_rate": 0.0005, "loss": 2.0464210510253906, "step": 9805 }, { "epoch": 0.7730390224674812, "grad_norm": 0.6252005100250244, "learning_rate": 0.0005, "loss": 2.1442809104919434, "step": 9806 }, { "epoch": 0.7731178557351203, "grad_norm": 0.8967965245246887, "learning_rate": 0.0005, "loss": 1.8263540267944336, "step": 9807 }, { "epoch": 0.7731966890027592, "grad_norm": 0.5866290330886841, "learning_rate": 0.0005, "loss": 1.7379074096679688, "step": 9808 }, { "epoch": 0.7732755222703981, "grad_norm": 0.7394431233406067, "learning_rate": 0.0005, "loss": 1.7822563648223877, "step": 9809 }, { "epoch": 0.773354355538037, "grad_norm": 0.6296077966690063, "learning_rate": 0.0005, "loss": 1.9684733152389526, "step": 9810 }, { "epoch": 0.773433188805676, "grad_norm": 0.7452313899993896, "learning_rate": 0.0005, "loss": 1.8840395212173462, "step": 9811 }, { "epoch": 0.7735120220733149, "grad_norm": 0.8666424751281738, "learning_rate": 0.0005, "loss": 2.2148449420928955, "step": 9812 }, { "epoch": 0.7735908553409538, "grad_norm": 0.6061679720878601, "learning_rate": 0.0005, "loss": 1.8974056243896484, "step": 9813 }, { "epoch": 0.7736696886085929, "grad_norm": 0.7828044891357422, "learning_rate": 0.0005, "loss": 2.051445484161377, "step": 9814 }, { "epoch": 0.7737485218762318, "grad_norm": 0.7474690675735474, "learning_rate": 0.0005, "loss": 1.8019825220108032, "step": 9815 }, { "epoch": 0.7738273551438707, "grad_norm": 0.7089780569076538, "learning_rate": 0.0005, "loss": 1.9032399654388428, "step": 9816 }, { "epoch": 0.7739061884115097, "grad_norm": 0.6207769513130188, "learning_rate": 0.0005, "loss": 1.995471715927124, "step": 9817 }, { "epoch": 0.7739850216791486, "grad_norm": 0.6460766196250916, "learning_rate": 0.0005, "loss": 2.264031410217285, "step": 9818 }, { "epoch": 0.7740638549467875, "grad_norm": 1.0736887454986572, "learning_rate": 0.0005, "loss": 2.153808116912842, "step": 9819 }, { "epoch": 0.7741426882144264, "grad_norm": 0.8085726499557495, "learning_rate": 0.0005, "loss": 1.7322016954421997, "step": 9820 }, { "epoch": 0.7742215214820655, "grad_norm": 0.6555004119873047, "learning_rate": 0.0005, "loss": 1.7932199239730835, "step": 9821 }, { "epoch": 0.7743003547497044, "grad_norm": 0.8114264607429504, "learning_rate": 0.0005, "loss": 1.9787925481796265, "step": 9822 }, { "epoch": 0.7743791880173433, "grad_norm": 1.0373077392578125, "learning_rate": 0.0005, "loss": 1.9230499267578125, "step": 9823 }, { "epoch": 0.7744580212849823, "grad_norm": 1.137292504310608, "learning_rate": 0.0005, "loss": 1.835513949394226, "step": 9824 }, { "epoch": 0.7745368545526212, "grad_norm": 0.5560681819915771, "learning_rate": 0.0005, "loss": 1.8052340745925903, "step": 9825 }, { "epoch": 0.7746156878202601, "grad_norm": 0.7312155961990356, "learning_rate": 0.0005, "loss": 1.7974731922149658, "step": 9826 }, { "epoch": 0.774694521087899, "grad_norm": 0.7438182234764099, "learning_rate": 0.0005, "loss": 1.7120492458343506, "step": 9827 }, { "epoch": 0.7747733543555381, "grad_norm": 0.7360633611679077, "learning_rate": 0.0005, "loss": 1.9704939126968384, "step": 9828 }, { "epoch": 0.774852187623177, "grad_norm": 0.5970469117164612, "learning_rate": 0.0005, "loss": 2.008744716644287, "step": 9829 }, { "epoch": 0.7749310208908159, "grad_norm": 0.6270855069160461, "learning_rate": 0.0005, "loss": 2.0787901878356934, "step": 9830 }, { "epoch": 0.7750098541584549, "grad_norm": 0.5988455414772034, "learning_rate": 0.0005, "loss": 1.7347261905670166, "step": 9831 }, { "epoch": 0.7750886874260938, "grad_norm": 0.818187952041626, "learning_rate": 0.0005, "loss": 1.7855437994003296, "step": 9832 }, { "epoch": 0.7751675206937327, "grad_norm": 0.6134910583496094, "learning_rate": 0.0005, "loss": 1.9538826942443848, "step": 9833 }, { "epoch": 0.7752463539613716, "grad_norm": 0.769573986530304, "learning_rate": 0.0005, "loss": 1.8040199279785156, "step": 9834 }, { "epoch": 0.7753251872290107, "grad_norm": 0.7385703921318054, "learning_rate": 0.0005, "loss": 1.8695656061172485, "step": 9835 }, { "epoch": 0.7754040204966496, "grad_norm": 0.6209280490875244, "learning_rate": 0.0005, "loss": 1.7138564586639404, "step": 9836 }, { "epoch": 0.7754828537642885, "grad_norm": 0.7672067284584045, "learning_rate": 0.0005, "loss": 2.1203832626342773, "step": 9837 }, { "epoch": 0.7755616870319275, "grad_norm": 0.6232805252075195, "learning_rate": 0.0005, "loss": 1.9218149185180664, "step": 9838 }, { "epoch": 0.7756405202995664, "grad_norm": 0.5944527983665466, "learning_rate": 0.0005, "loss": 1.7430393695831299, "step": 9839 }, { "epoch": 0.7757193535672053, "grad_norm": 0.6011990308761597, "learning_rate": 0.0005, "loss": 1.5354267358779907, "step": 9840 }, { "epoch": 0.7757981868348443, "grad_norm": 0.799647867679596, "learning_rate": 0.0005, "loss": 1.738051414489746, "step": 9841 }, { "epoch": 0.7758770201024833, "grad_norm": 0.7033988237380981, "learning_rate": 0.0005, "loss": 1.8200956583023071, "step": 9842 }, { "epoch": 0.7759558533701222, "grad_norm": 0.48953649401664734, "learning_rate": 0.0005, "loss": 1.6314663887023926, "step": 9843 }, { "epoch": 0.7760346866377611, "grad_norm": 0.7906949520111084, "learning_rate": 0.0005, "loss": 1.8732846975326538, "step": 9844 }, { "epoch": 0.7761135199054001, "grad_norm": 0.7586579322814941, "learning_rate": 0.0005, "loss": 2.2093863487243652, "step": 9845 }, { "epoch": 0.776192353173039, "grad_norm": 1.3917230367660522, "learning_rate": 0.0005, "loss": 2.1666717529296875, "step": 9846 }, { "epoch": 0.7762711864406779, "grad_norm": 0.9359334111213684, "learning_rate": 0.0005, "loss": 2.0285487174987793, "step": 9847 }, { "epoch": 0.776350019708317, "grad_norm": 0.6447242498397827, "learning_rate": 0.0005, "loss": 2.0147719383239746, "step": 9848 }, { "epoch": 0.7764288529759559, "grad_norm": 0.6453753113746643, "learning_rate": 0.0005, "loss": 1.8079922199249268, "step": 9849 }, { "epoch": 0.7765076862435948, "grad_norm": 0.7285075187683105, "learning_rate": 0.0005, "loss": 1.7928707599639893, "step": 9850 }, { "epoch": 0.7765865195112337, "grad_norm": 0.6478874683380127, "learning_rate": 0.0005, "loss": 1.833906650543213, "step": 9851 }, { "epoch": 0.7766653527788727, "grad_norm": 0.608046293258667, "learning_rate": 0.0005, "loss": 1.8691623210906982, "step": 9852 }, { "epoch": 0.7767441860465116, "grad_norm": 1.035213589668274, "learning_rate": 0.0005, "loss": 1.8809003829956055, "step": 9853 }, { "epoch": 0.7768230193141505, "grad_norm": 0.7742915153503418, "learning_rate": 0.0005, "loss": 2.0944437980651855, "step": 9854 }, { "epoch": 0.7769018525817896, "grad_norm": 0.8745570182800293, "learning_rate": 0.0005, "loss": 1.8526643514633179, "step": 9855 }, { "epoch": 0.7769806858494285, "grad_norm": 0.7259846329689026, "learning_rate": 0.0005, "loss": 1.8816351890563965, "step": 9856 }, { "epoch": 0.7770595191170674, "grad_norm": 0.5883885025978088, "learning_rate": 0.0005, "loss": 1.9310702085494995, "step": 9857 }, { "epoch": 0.7771383523847063, "grad_norm": 1.0442904233932495, "learning_rate": 0.0005, "loss": 2.237438201904297, "step": 9858 }, { "epoch": 0.7772171856523453, "grad_norm": 0.7561408281326294, "learning_rate": 0.0005, "loss": 1.591854214668274, "step": 9859 }, { "epoch": 0.7772960189199842, "grad_norm": 0.7320520877838135, "learning_rate": 0.0005, "loss": 1.8786845207214355, "step": 9860 }, { "epoch": 0.7773748521876231, "grad_norm": 0.7377865314483643, "learning_rate": 0.0005, "loss": 1.8033926486968994, "step": 9861 }, { "epoch": 0.7774536854552622, "grad_norm": 0.7002078294754028, "learning_rate": 0.0005, "loss": 1.8474974632263184, "step": 9862 }, { "epoch": 0.7775325187229011, "grad_norm": 0.7867648601531982, "learning_rate": 0.0005, "loss": 1.951645851135254, "step": 9863 }, { "epoch": 0.77761135199054, "grad_norm": 0.5432989001274109, "learning_rate": 0.0005, "loss": 1.8854670524597168, "step": 9864 }, { "epoch": 0.7776901852581789, "grad_norm": 0.5997778177261353, "learning_rate": 0.0005, "loss": 1.931732177734375, "step": 9865 }, { "epoch": 0.7777690185258179, "grad_norm": 0.6021782159805298, "learning_rate": 0.0005, "loss": 1.8755890130996704, "step": 9866 }, { "epoch": 0.7778478517934568, "grad_norm": 1.2276618480682373, "learning_rate": 0.0005, "loss": 2.2194337844848633, "step": 9867 }, { "epoch": 0.7779266850610957, "grad_norm": 0.8412491083145142, "learning_rate": 0.0005, "loss": 2.344970226287842, "step": 9868 }, { "epoch": 0.7780055183287348, "grad_norm": 0.5854480266571045, "learning_rate": 0.0005, "loss": 1.8856761455535889, "step": 9869 }, { "epoch": 0.7780843515963737, "grad_norm": 0.5953001976013184, "learning_rate": 0.0005, "loss": 1.563359260559082, "step": 9870 }, { "epoch": 0.7781631848640126, "grad_norm": 0.6223945617675781, "learning_rate": 0.0005, "loss": 1.9336663484573364, "step": 9871 }, { "epoch": 0.7782420181316515, "grad_norm": 0.6300014853477478, "learning_rate": 0.0005, "loss": 2.0183849334716797, "step": 9872 }, { "epoch": 0.7783208513992905, "grad_norm": 0.5810673832893372, "learning_rate": 0.0005, "loss": 1.6115000247955322, "step": 9873 }, { "epoch": 0.7783996846669294, "grad_norm": 2.345125198364258, "learning_rate": 0.0005, "loss": 2.101762533187866, "step": 9874 }, { "epoch": 0.7784785179345683, "grad_norm": 0.5834037661552429, "learning_rate": 0.0005, "loss": 1.8210092782974243, "step": 9875 }, { "epoch": 0.7785573512022074, "grad_norm": 0.5768101811408997, "learning_rate": 0.0005, "loss": 2.255248785018921, "step": 9876 }, { "epoch": 0.7786361844698463, "grad_norm": 0.6796313524246216, "learning_rate": 0.0005, "loss": 1.8893063068389893, "step": 9877 }, { "epoch": 0.7787150177374852, "grad_norm": 0.5323329567909241, "learning_rate": 0.0005, "loss": 1.8606367111206055, "step": 9878 }, { "epoch": 0.7787938510051242, "grad_norm": 0.5491125583648682, "learning_rate": 0.0005, "loss": 1.6425244808197021, "step": 9879 }, { "epoch": 0.7788726842727631, "grad_norm": 0.9004731774330139, "learning_rate": 0.0005, "loss": 1.8760676383972168, "step": 9880 }, { "epoch": 0.778951517540402, "grad_norm": 0.7782180905342102, "learning_rate": 0.0005, "loss": 1.8647600412368774, "step": 9881 }, { "epoch": 0.779030350808041, "grad_norm": 0.7545580863952637, "learning_rate": 0.0005, "loss": 1.6068849563598633, "step": 9882 }, { "epoch": 0.77910918407568, "grad_norm": 0.692423403263092, "learning_rate": 0.0005, "loss": 1.9478037357330322, "step": 9883 }, { "epoch": 0.7791880173433189, "grad_norm": 0.5740602016448975, "learning_rate": 0.0005, "loss": 1.814913034439087, "step": 9884 }, { "epoch": 0.7792668506109578, "grad_norm": 0.7450970411300659, "learning_rate": 0.0005, "loss": 1.73137629032135, "step": 9885 }, { "epoch": 0.7793456838785968, "grad_norm": 0.6758357286453247, "learning_rate": 0.0005, "loss": 1.9589002132415771, "step": 9886 }, { "epoch": 0.7794245171462357, "grad_norm": 0.7106594443321228, "learning_rate": 0.0005, "loss": 2.0888004302978516, "step": 9887 }, { "epoch": 0.7795033504138746, "grad_norm": 0.6422126293182373, "learning_rate": 0.0005, "loss": 1.8455747365951538, "step": 9888 }, { "epoch": 0.7795821836815136, "grad_norm": 0.5223023295402527, "learning_rate": 0.0005, "loss": 1.7635152339935303, "step": 9889 }, { "epoch": 0.7796610169491526, "grad_norm": 0.5944274067878723, "learning_rate": 0.0005, "loss": 1.8614434003829956, "step": 9890 }, { "epoch": 0.7797398502167915, "grad_norm": 0.6914060115814209, "learning_rate": 0.0005, "loss": 1.7677958011627197, "step": 9891 }, { "epoch": 0.7798186834844304, "grad_norm": 0.8111676573753357, "learning_rate": 0.0005, "loss": 1.9977047443389893, "step": 9892 }, { "epoch": 0.7798975167520694, "grad_norm": 0.7558987736701965, "learning_rate": 0.0005, "loss": 1.911595106124878, "step": 9893 }, { "epoch": 0.7799763500197083, "grad_norm": 0.6656980514526367, "learning_rate": 0.0005, "loss": 2.1458892822265625, "step": 9894 }, { "epoch": 0.7800551832873472, "grad_norm": 0.6441869735717773, "learning_rate": 0.0005, "loss": 1.8504623174667358, "step": 9895 }, { "epoch": 0.7801340165549862, "grad_norm": 0.641650915145874, "learning_rate": 0.0005, "loss": 1.8662071228027344, "step": 9896 }, { "epoch": 0.7802128498226252, "grad_norm": 0.6392648816108704, "learning_rate": 0.0005, "loss": 1.9092868566513062, "step": 9897 }, { "epoch": 0.7802916830902641, "grad_norm": 0.8311995267868042, "learning_rate": 0.0005, "loss": 1.8996987342834473, "step": 9898 }, { "epoch": 0.780370516357903, "grad_norm": 0.5397886633872986, "learning_rate": 0.0005, "loss": 1.8353041410446167, "step": 9899 }, { "epoch": 0.780449349625542, "grad_norm": 0.6853272318840027, "learning_rate": 0.0005, "loss": 2.148509979248047, "step": 9900 }, { "epoch": 0.7805281828931809, "grad_norm": 0.8460965752601624, "learning_rate": 0.0005, "loss": 2.0893776416778564, "step": 9901 }, { "epoch": 0.7806070161608198, "grad_norm": 0.6366903185844421, "learning_rate": 0.0005, "loss": 2.0683717727661133, "step": 9902 }, { "epoch": 0.7806858494284588, "grad_norm": 0.8637062311172485, "learning_rate": 0.0005, "loss": 1.9589496850967407, "step": 9903 }, { "epoch": 0.7807646826960978, "grad_norm": 0.8298429250717163, "learning_rate": 0.0005, "loss": 2.0727922916412354, "step": 9904 }, { "epoch": 0.7808435159637367, "grad_norm": 0.6634124517440796, "learning_rate": 0.0005, "loss": 1.8025290966033936, "step": 9905 }, { "epoch": 0.7809223492313756, "grad_norm": 2.2317895889282227, "learning_rate": 0.0005, "loss": 2.0476255416870117, "step": 9906 }, { "epoch": 0.7810011824990146, "grad_norm": 0.9988524317741394, "learning_rate": 0.0005, "loss": 2.1936652660369873, "step": 9907 }, { "epoch": 0.7810800157666535, "grad_norm": 0.6310136318206787, "learning_rate": 0.0005, "loss": 1.9341471195220947, "step": 9908 }, { "epoch": 0.7811588490342924, "grad_norm": 0.6149312853813171, "learning_rate": 0.0005, "loss": 1.7832413911819458, "step": 9909 }, { "epoch": 0.7812376823019315, "grad_norm": 0.9282742142677307, "learning_rate": 0.0005, "loss": 1.6278517246246338, "step": 9910 }, { "epoch": 0.7813165155695704, "grad_norm": 0.8218213319778442, "learning_rate": 0.0005, "loss": 2.087501049041748, "step": 9911 }, { "epoch": 0.7813953488372093, "grad_norm": 0.6698494553565979, "learning_rate": 0.0005, "loss": 2.1089508533477783, "step": 9912 }, { "epoch": 0.7814741821048482, "grad_norm": 0.7752707600593567, "learning_rate": 0.0005, "loss": 1.8193368911743164, "step": 9913 }, { "epoch": 0.7815530153724872, "grad_norm": 0.5647933483123779, "learning_rate": 0.0005, "loss": 1.7170109748840332, "step": 9914 }, { "epoch": 0.7816318486401261, "grad_norm": 0.7423956990242004, "learning_rate": 0.0005, "loss": 2.1239748001098633, "step": 9915 }, { "epoch": 0.781710681907765, "grad_norm": 0.6511154770851135, "learning_rate": 0.0005, "loss": 2.0719282627105713, "step": 9916 }, { "epoch": 0.7817895151754041, "grad_norm": 0.7728721499443054, "learning_rate": 0.0005, "loss": 1.891204833984375, "step": 9917 }, { "epoch": 0.781868348443043, "grad_norm": 0.6833807229995728, "learning_rate": 0.0005, "loss": 1.9449114799499512, "step": 9918 }, { "epoch": 0.7819471817106819, "grad_norm": 0.5804338455200195, "learning_rate": 0.0005, "loss": 1.9118891954421997, "step": 9919 }, { "epoch": 0.7820260149783208, "grad_norm": 0.5397234559059143, "learning_rate": 0.0005, "loss": 1.9364097118377686, "step": 9920 }, { "epoch": 0.7821048482459598, "grad_norm": 0.53081214427948, "learning_rate": 0.0005, "loss": 1.9428491592407227, "step": 9921 }, { "epoch": 0.7821836815135987, "grad_norm": 0.5735167860984802, "learning_rate": 0.0005, "loss": 1.9737768173217773, "step": 9922 }, { "epoch": 0.7822625147812377, "grad_norm": 0.6454995274543762, "learning_rate": 0.0005, "loss": 1.6194865703582764, "step": 9923 }, { "epoch": 0.7823413480488767, "grad_norm": 0.6506761312484741, "learning_rate": 0.0005, "loss": 1.8720779418945312, "step": 9924 }, { "epoch": 0.7824201813165156, "grad_norm": 0.6472958326339722, "learning_rate": 0.0005, "loss": 1.6471951007843018, "step": 9925 }, { "epoch": 0.7824990145841545, "grad_norm": 0.6264646053314209, "learning_rate": 0.0005, "loss": 1.8398933410644531, "step": 9926 }, { "epoch": 0.7825778478517934, "grad_norm": 0.6388620734214783, "learning_rate": 0.0005, "loss": 1.7722594738006592, "step": 9927 }, { "epoch": 0.7826566811194324, "grad_norm": 0.5719149708747864, "learning_rate": 0.0005, "loss": 1.6788113117218018, "step": 9928 }, { "epoch": 0.7827355143870713, "grad_norm": 0.5474898219108582, "learning_rate": 0.0005, "loss": 1.6824513673782349, "step": 9929 }, { "epoch": 0.7828143476547103, "grad_norm": 0.5623157620429993, "learning_rate": 0.0005, "loss": 2.0141830444335938, "step": 9930 }, { "epoch": 0.7828931809223493, "grad_norm": 0.7499642968177795, "learning_rate": 0.0005, "loss": 1.9538949728012085, "step": 9931 }, { "epoch": 0.7829720141899882, "grad_norm": 0.6533353924751282, "learning_rate": 0.0005, "loss": 1.789383888244629, "step": 9932 }, { "epoch": 0.7830508474576271, "grad_norm": 0.9451440572738647, "learning_rate": 0.0005, "loss": 1.9435231685638428, "step": 9933 }, { "epoch": 0.783129680725266, "grad_norm": 0.665052056312561, "learning_rate": 0.0005, "loss": 1.727158546447754, "step": 9934 }, { "epoch": 0.783208513992905, "grad_norm": 0.5732452273368835, "learning_rate": 0.0005, "loss": 1.8069674968719482, "step": 9935 }, { "epoch": 0.7832873472605439, "grad_norm": 0.6877955794334412, "learning_rate": 0.0005, "loss": 2.022718906402588, "step": 9936 }, { "epoch": 0.7833661805281829, "grad_norm": 0.7179865837097168, "learning_rate": 0.0005, "loss": 1.901234745979309, "step": 9937 }, { "epoch": 0.7834450137958219, "grad_norm": 0.7157789468765259, "learning_rate": 0.0005, "loss": 1.8190999031066895, "step": 9938 }, { "epoch": 0.7835238470634608, "grad_norm": 0.7296658158302307, "learning_rate": 0.0005, "loss": 2.127664566040039, "step": 9939 }, { "epoch": 0.7836026803310997, "grad_norm": 0.6645367741584778, "learning_rate": 0.0005, "loss": 1.973416805267334, "step": 9940 }, { "epoch": 0.7836815135987387, "grad_norm": 1.0175226926803589, "learning_rate": 0.0005, "loss": 2.1703901290893555, "step": 9941 }, { "epoch": 0.7837603468663776, "grad_norm": 0.6350244283676147, "learning_rate": 0.0005, "loss": 1.716379165649414, "step": 9942 }, { "epoch": 0.7838391801340165, "grad_norm": 0.5973676443099976, "learning_rate": 0.0005, "loss": 1.8971084356307983, "step": 9943 }, { "epoch": 0.7839180134016555, "grad_norm": 0.8703056573867798, "learning_rate": 0.0005, "loss": 2.006580352783203, "step": 9944 }, { "epoch": 0.7839968466692945, "grad_norm": 0.8337137699127197, "learning_rate": 0.0005, "loss": 2.1492161750793457, "step": 9945 }, { "epoch": 0.7840756799369334, "grad_norm": 0.7590520977973938, "learning_rate": 0.0005, "loss": 2.095707893371582, "step": 9946 }, { "epoch": 0.7841545132045723, "grad_norm": 0.519766628742218, "learning_rate": 0.0005, "loss": 1.8377397060394287, "step": 9947 }, { "epoch": 0.7842333464722113, "grad_norm": 0.6900141835212708, "learning_rate": 0.0005, "loss": 1.5838806629180908, "step": 9948 }, { "epoch": 0.7843121797398502, "grad_norm": 0.7950006127357483, "learning_rate": 0.0005, "loss": 1.7973781824111938, "step": 9949 }, { "epoch": 0.7843910130074891, "grad_norm": 0.6152257323265076, "learning_rate": 0.0005, "loss": 2.0338096618652344, "step": 9950 }, { "epoch": 0.7844698462751281, "grad_norm": 0.5182784795761108, "learning_rate": 0.0005, "loss": 1.6353651285171509, "step": 9951 }, { "epoch": 0.7845486795427671, "grad_norm": 0.503068208694458, "learning_rate": 0.0005, "loss": 1.6500935554504395, "step": 9952 }, { "epoch": 0.784627512810406, "grad_norm": 0.6883622407913208, "learning_rate": 0.0005, "loss": 1.8536412715911865, "step": 9953 }, { "epoch": 0.7847063460780449, "grad_norm": 0.575686514377594, "learning_rate": 0.0005, "loss": 1.930901050567627, "step": 9954 }, { "epoch": 0.7847851793456839, "grad_norm": 0.5809673070907593, "learning_rate": 0.0005, "loss": 1.805862545967102, "step": 9955 }, { "epoch": 0.7848640126133228, "grad_norm": 0.6719465255737305, "learning_rate": 0.0005, "loss": 1.6926617622375488, "step": 9956 }, { "epoch": 0.7849428458809617, "grad_norm": 0.7083708643913269, "learning_rate": 0.0005, "loss": 2.0570895671844482, "step": 9957 }, { "epoch": 0.7850216791486007, "grad_norm": 0.5864545106887817, "learning_rate": 0.0005, "loss": 1.6876461505889893, "step": 9958 }, { "epoch": 0.7851005124162397, "grad_norm": 0.6501890420913696, "learning_rate": 0.0005, "loss": 2.144038200378418, "step": 9959 }, { "epoch": 0.7851793456838786, "grad_norm": 0.550570011138916, "learning_rate": 0.0005, "loss": 1.8288242816925049, "step": 9960 }, { "epoch": 0.7852581789515175, "grad_norm": 0.6509894728660583, "learning_rate": 0.0005, "loss": 1.7637338638305664, "step": 9961 }, { "epoch": 0.7853370122191565, "grad_norm": 0.6545818448066711, "learning_rate": 0.0005, "loss": 2.103717088699341, "step": 9962 }, { "epoch": 0.7854158454867954, "grad_norm": 0.6175153255462646, "learning_rate": 0.0005, "loss": 2.0878167152404785, "step": 9963 }, { "epoch": 0.7854946787544344, "grad_norm": 0.6884578466415405, "learning_rate": 0.0005, "loss": 1.921337366104126, "step": 9964 }, { "epoch": 0.7855735120220733, "grad_norm": 0.7198188304901123, "learning_rate": 0.0005, "loss": 2.1074206829071045, "step": 9965 }, { "epoch": 0.7856523452897123, "grad_norm": 0.5768089294433594, "learning_rate": 0.0005, "loss": 1.9061479568481445, "step": 9966 }, { "epoch": 0.7857311785573512, "grad_norm": 0.5545753836631775, "learning_rate": 0.0005, "loss": 1.7304742336273193, "step": 9967 }, { "epoch": 0.7858100118249901, "grad_norm": 0.716414749622345, "learning_rate": 0.0005, "loss": 1.5745939016342163, "step": 9968 }, { "epoch": 0.7858888450926291, "grad_norm": 1.2920477390289307, "learning_rate": 0.0005, "loss": 1.9808882474899292, "step": 9969 }, { "epoch": 0.785967678360268, "grad_norm": 0.5832154154777527, "learning_rate": 0.0005, "loss": 1.9942735433578491, "step": 9970 }, { "epoch": 0.786046511627907, "grad_norm": 0.6796560287475586, "learning_rate": 0.0005, "loss": 1.7332758903503418, "step": 9971 }, { "epoch": 0.786125344895546, "grad_norm": 0.5781319737434387, "learning_rate": 0.0005, "loss": 2.060586929321289, "step": 9972 }, { "epoch": 0.7862041781631849, "grad_norm": 0.8502489924430847, "learning_rate": 0.0005, "loss": 2.06783390045166, "step": 9973 }, { "epoch": 0.7862830114308238, "grad_norm": 0.7197983264923096, "learning_rate": 0.0005, "loss": 1.912956953048706, "step": 9974 }, { "epoch": 0.7863618446984627, "grad_norm": 0.7320321202278137, "learning_rate": 0.0005, "loss": 1.7538286447525024, "step": 9975 }, { "epoch": 0.7864406779661017, "grad_norm": 0.6995599865913391, "learning_rate": 0.0005, "loss": 1.677340030670166, "step": 9976 }, { "epoch": 0.7865195112337406, "grad_norm": 0.78448486328125, "learning_rate": 0.0005, "loss": 2.022557258605957, "step": 9977 }, { "epoch": 0.7865983445013796, "grad_norm": 0.7196556329727173, "learning_rate": 0.0005, "loss": 1.815420389175415, "step": 9978 }, { "epoch": 0.7866771777690186, "grad_norm": 0.6936919093132019, "learning_rate": 0.0005, "loss": 1.939823031425476, "step": 9979 }, { "epoch": 0.7867560110366575, "grad_norm": 0.7277519106864929, "learning_rate": 0.0005, "loss": 2.19107723236084, "step": 9980 }, { "epoch": 0.7868348443042964, "grad_norm": 0.4840737581253052, "learning_rate": 0.0005, "loss": 1.6501752138137817, "step": 9981 }, { "epoch": 0.7869136775719353, "grad_norm": 0.6188724637031555, "learning_rate": 0.0005, "loss": 1.8089110851287842, "step": 9982 }, { "epoch": 0.7869925108395743, "grad_norm": 0.811853289604187, "learning_rate": 0.0005, "loss": 1.7272273302078247, "step": 9983 }, { "epoch": 0.7870713441072132, "grad_norm": 1.9520602226257324, "learning_rate": 0.0005, "loss": 1.9402122497558594, "step": 9984 }, { "epoch": 0.7871501773748522, "grad_norm": 0.5757530927658081, "learning_rate": 0.0005, "loss": 1.822036862373352, "step": 9985 }, { "epoch": 0.7872290106424912, "grad_norm": 1.2975250482559204, "learning_rate": 0.0005, "loss": 1.7205579280853271, "step": 9986 }, { "epoch": 0.7873078439101301, "grad_norm": 0.6839802265167236, "learning_rate": 0.0005, "loss": 1.861891508102417, "step": 9987 }, { "epoch": 0.787386677177769, "grad_norm": 0.6991479396820068, "learning_rate": 0.0005, "loss": 1.737650752067566, "step": 9988 }, { "epoch": 0.7874655104454079, "grad_norm": 0.6467998623847961, "learning_rate": 0.0005, "loss": 1.8787579536437988, "step": 9989 }, { "epoch": 0.7875443437130469, "grad_norm": 0.6810709238052368, "learning_rate": 0.0005, "loss": 1.9591724872589111, "step": 9990 }, { "epoch": 0.7876231769806858, "grad_norm": 4.283895015716553, "learning_rate": 0.0005, "loss": 1.7740428447723389, "step": 9991 }, { "epoch": 0.7877020102483248, "grad_norm": 0.672652542591095, "learning_rate": 0.0005, "loss": 1.9655108451843262, "step": 9992 }, { "epoch": 0.7877808435159638, "grad_norm": 0.7832207083702087, "learning_rate": 0.0005, "loss": 1.974393606185913, "step": 9993 }, { "epoch": 0.7878596767836027, "grad_norm": 0.6178276538848877, "learning_rate": 0.0005, "loss": 1.821828842163086, "step": 9994 }, { "epoch": 0.7879385100512416, "grad_norm": 0.6629186272621155, "learning_rate": 0.0005, "loss": 1.8528238534927368, "step": 9995 }, { "epoch": 0.7880173433188805, "grad_norm": 0.6378860473632812, "learning_rate": 0.0005, "loss": 1.739995002746582, "step": 9996 }, { "epoch": 0.7880961765865195, "grad_norm": 0.6491075754165649, "learning_rate": 0.0005, "loss": 2.010505437850952, "step": 9997 }, { "epoch": 0.7881750098541584, "grad_norm": 0.6249232888221741, "learning_rate": 0.0005, "loss": 1.9921848773956299, "step": 9998 }, { "epoch": 0.7882538431217974, "grad_norm": 0.5913081169128418, "learning_rate": 0.0005, "loss": 1.554072618484497, "step": 9999 }, { "epoch": 0.7883326763894364, "grad_norm": 0.667809784412384, "learning_rate": 0.0005, "loss": 1.8131394386291504, "step": 10000 }, { "epoch": 0.7884115096570753, "grad_norm": 0.8016331195831299, "learning_rate": 0.0005, "loss": 2.319016933441162, "step": 10001 }, { "epoch": 0.7884903429247142, "grad_norm": 0.8572549223899841, "learning_rate": 0.0005, "loss": 1.971320629119873, "step": 10002 }, { "epoch": 0.7885691761923532, "grad_norm": 0.6814366579055786, "learning_rate": 0.0005, "loss": 1.8716819286346436, "step": 10003 }, { "epoch": 0.7886480094599921, "grad_norm": 1.1437658071517944, "learning_rate": 0.0005, "loss": 1.7527942657470703, "step": 10004 }, { "epoch": 0.788726842727631, "grad_norm": 1.1017186641693115, "learning_rate": 0.0005, "loss": 2.2745726108551025, "step": 10005 }, { "epoch": 0.78880567599527, "grad_norm": 0.6317563652992249, "learning_rate": 0.0005, "loss": 1.7224042415618896, "step": 10006 }, { "epoch": 0.788884509262909, "grad_norm": 0.627469539642334, "learning_rate": 0.0005, "loss": 1.6506354808807373, "step": 10007 }, { "epoch": 0.7889633425305479, "grad_norm": 0.7561095356941223, "learning_rate": 0.0005, "loss": 2.2896909713745117, "step": 10008 }, { "epoch": 0.7890421757981868, "grad_norm": 0.7583928108215332, "learning_rate": 0.0005, "loss": 2.0815324783325195, "step": 10009 }, { "epoch": 0.7891210090658258, "grad_norm": 0.6996136903762817, "learning_rate": 0.0005, "loss": 1.9663318395614624, "step": 10010 }, { "epoch": 0.7891998423334647, "grad_norm": 0.6168032288551331, "learning_rate": 0.0005, "loss": 1.9281280040740967, "step": 10011 }, { "epoch": 0.7892786756011037, "grad_norm": 3.3983118534088135, "learning_rate": 0.0005, "loss": 2.0193138122558594, "step": 10012 }, { "epoch": 0.7893575088687426, "grad_norm": 0.7010294795036316, "learning_rate": 0.0005, "loss": 1.7399916648864746, "step": 10013 }, { "epoch": 0.7894363421363816, "grad_norm": 0.7797930836677551, "learning_rate": 0.0005, "loss": 1.6754732131958008, "step": 10014 }, { "epoch": 0.7895151754040205, "grad_norm": 0.6483454704284668, "learning_rate": 0.0005, "loss": 1.867774248123169, "step": 10015 }, { "epoch": 0.7895940086716594, "grad_norm": 0.7345425486564636, "learning_rate": 0.0005, "loss": 1.7611058950424194, "step": 10016 }, { "epoch": 0.7896728419392984, "grad_norm": 0.7685381174087524, "learning_rate": 0.0005, "loss": 1.9337204694747925, "step": 10017 }, { "epoch": 0.7897516752069373, "grad_norm": 0.7012912631034851, "learning_rate": 0.0005, "loss": 1.9706746339797974, "step": 10018 }, { "epoch": 0.7898305084745763, "grad_norm": 0.5905911326408386, "learning_rate": 0.0005, "loss": 2.024399995803833, "step": 10019 }, { "epoch": 0.7899093417422152, "grad_norm": 0.6315476298332214, "learning_rate": 0.0005, "loss": 2.169713020324707, "step": 10020 }, { "epoch": 0.7899881750098542, "grad_norm": 0.819206953048706, "learning_rate": 0.0005, "loss": 2.1198599338531494, "step": 10021 }, { "epoch": 0.7900670082774931, "grad_norm": 0.6733411550521851, "learning_rate": 0.0005, "loss": 1.9420651197433472, "step": 10022 }, { "epoch": 0.790145841545132, "grad_norm": 0.7630260586738586, "learning_rate": 0.0005, "loss": 1.7542744874954224, "step": 10023 }, { "epoch": 0.790224674812771, "grad_norm": 0.6650546193122864, "learning_rate": 0.0005, "loss": 1.939131259918213, "step": 10024 }, { "epoch": 0.7903035080804099, "grad_norm": 0.6583081483840942, "learning_rate": 0.0005, "loss": 2.069889545440674, "step": 10025 }, { "epoch": 0.7903823413480489, "grad_norm": 0.5082564949989319, "learning_rate": 0.0005, "loss": 1.6034023761749268, "step": 10026 }, { "epoch": 0.7904611746156878, "grad_norm": 0.6349079608917236, "learning_rate": 0.0005, "loss": 1.7810090780258179, "step": 10027 }, { "epoch": 0.7905400078833268, "grad_norm": 0.6348554491996765, "learning_rate": 0.0005, "loss": 1.7862526178359985, "step": 10028 }, { "epoch": 0.7906188411509657, "grad_norm": 0.8015027642250061, "learning_rate": 0.0005, "loss": 2.0977745056152344, "step": 10029 }, { "epoch": 0.7906976744186046, "grad_norm": 0.8913885354995728, "learning_rate": 0.0005, "loss": 2.1252565383911133, "step": 10030 }, { "epoch": 0.7907765076862436, "grad_norm": 0.5671066045761108, "learning_rate": 0.0005, "loss": 1.5447759628295898, "step": 10031 }, { "epoch": 0.7908553409538825, "grad_norm": 0.52008056640625, "learning_rate": 0.0005, "loss": 1.5614022016525269, "step": 10032 }, { "epoch": 0.7909341742215215, "grad_norm": 0.5765251517295837, "learning_rate": 0.0005, "loss": 1.7741258144378662, "step": 10033 }, { "epoch": 0.7910130074891605, "grad_norm": 0.5606542825698853, "learning_rate": 0.0005, "loss": 1.8150718212127686, "step": 10034 }, { "epoch": 0.7910918407567994, "grad_norm": 0.7445987462997437, "learning_rate": 0.0005, "loss": 1.6747968196868896, "step": 10035 }, { "epoch": 0.7911706740244383, "grad_norm": 0.6255777478218079, "learning_rate": 0.0005, "loss": 2.0296554565429688, "step": 10036 }, { "epoch": 0.7912495072920772, "grad_norm": 0.8397960662841797, "learning_rate": 0.0005, "loss": 2.0739879608154297, "step": 10037 }, { "epoch": 0.7913283405597162, "grad_norm": 0.7877931594848633, "learning_rate": 0.0005, "loss": 1.7331421375274658, "step": 10038 }, { "epoch": 0.7914071738273551, "grad_norm": 0.6328072547912598, "learning_rate": 0.0005, "loss": 1.8168190717697144, "step": 10039 }, { "epoch": 0.7914860070949941, "grad_norm": 4.4184794425964355, "learning_rate": 0.0005, "loss": 2.067521572113037, "step": 10040 }, { "epoch": 0.7915648403626331, "grad_norm": 0.6598876714706421, "learning_rate": 0.0005, "loss": 2.0572891235351562, "step": 10041 }, { "epoch": 0.791643673630272, "grad_norm": 0.6154906749725342, "learning_rate": 0.0005, "loss": 1.681797981262207, "step": 10042 }, { "epoch": 0.7917225068979109, "grad_norm": 0.668335497379303, "learning_rate": 0.0005, "loss": 1.9612772464752197, "step": 10043 }, { "epoch": 0.7918013401655498, "grad_norm": 0.733022153377533, "learning_rate": 0.0005, "loss": 1.726622223854065, "step": 10044 }, { "epoch": 0.7918801734331888, "grad_norm": 0.6488445997238159, "learning_rate": 0.0005, "loss": 1.6794356107711792, "step": 10045 }, { "epoch": 0.7919590067008278, "grad_norm": 0.6206173300743103, "learning_rate": 0.0005, "loss": 1.7180355787277222, "step": 10046 }, { "epoch": 0.7920378399684667, "grad_norm": 0.5684302449226379, "learning_rate": 0.0005, "loss": 1.8891818523406982, "step": 10047 }, { "epoch": 0.7921166732361057, "grad_norm": 0.7386744618415833, "learning_rate": 0.0005, "loss": 2.072328567504883, "step": 10048 }, { "epoch": 0.7921955065037446, "grad_norm": 0.8259344696998596, "learning_rate": 0.0005, "loss": 2.3649709224700928, "step": 10049 }, { "epoch": 0.7922743397713835, "grad_norm": 0.5626376271247864, "learning_rate": 0.0005, "loss": 1.7691419124603271, "step": 10050 }, { "epoch": 0.7923531730390224, "grad_norm": 0.9665438532829285, "learning_rate": 0.0005, "loss": 2.0190372467041016, "step": 10051 }, { "epoch": 0.7924320063066614, "grad_norm": 0.6427826881408691, "learning_rate": 0.0005, "loss": 2.0646729469299316, "step": 10052 }, { "epoch": 0.7925108395743004, "grad_norm": 0.6820741891860962, "learning_rate": 0.0005, "loss": 1.736153244972229, "step": 10053 }, { "epoch": 0.7925896728419393, "grad_norm": 0.9149224162101746, "learning_rate": 0.0005, "loss": 1.781864881515503, "step": 10054 }, { "epoch": 0.7926685061095783, "grad_norm": 0.7484353184700012, "learning_rate": 0.0005, "loss": 1.7526004314422607, "step": 10055 }, { "epoch": 0.7927473393772172, "grad_norm": 0.8077864646911621, "learning_rate": 0.0005, "loss": 2.10123348236084, "step": 10056 }, { "epoch": 0.7928261726448561, "grad_norm": 0.8122225999832153, "learning_rate": 0.0005, "loss": 1.7753217220306396, "step": 10057 }, { "epoch": 0.792905005912495, "grad_norm": 0.7918270230293274, "learning_rate": 0.0005, "loss": 1.9749300479888916, "step": 10058 }, { "epoch": 0.792983839180134, "grad_norm": 0.5465977191925049, "learning_rate": 0.0005, "loss": 1.8690471649169922, "step": 10059 }, { "epoch": 0.793062672447773, "grad_norm": 0.6230088472366333, "learning_rate": 0.0005, "loss": 1.990936517715454, "step": 10060 }, { "epoch": 0.7931415057154119, "grad_norm": 0.7132392525672913, "learning_rate": 0.0005, "loss": 1.948716640472412, "step": 10061 }, { "epoch": 0.7932203389830509, "grad_norm": 0.9206492304801941, "learning_rate": 0.0005, "loss": 2.0167622566223145, "step": 10062 }, { "epoch": 0.7932991722506898, "grad_norm": 0.739793062210083, "learning_rate": 0.0005, "loss": 1.811539649963379, "step": 10063 }, { "epoch": 0.7933780055183287, "grad_norm": 0.5895393490791321, "learning_rate": 0.0005, "loss": 1.8063201904296875, "step": 10064 }, { "epoch": 0.7934568387859677, "grad_norm": 0.8154612183570862, "learning_rate": 0.0005, "loss": 1.8730871677398682, "step": 10065 }, { "epoch": 0.7935356720536066, "grad_norm": 0.6077893376350403, "learning_rate": 0.0005, "loss": 2.1002204418182373, "step": 10066 }, { "epoch": 0.7936145053212456, "grad_norm": 0.6937012076377869, "learning_rate": 0.0005, "loss": 1.744985818862915, "step": 10067 }, { "epoch": 0.7936933385888845, "grad_norm": 0.7400424480438232, "learning_rate": 0.0005, "loss": 1.8179914951324463, "step": 10068 }, { "epoch": 0.7937721718565235, "grad_norm": 0.6168462038040161, "learning_rate": 0.0005, "loss": 1.721529483795166, "step": 10069 }, { "epoch": 0.7938510051241624, "grad_norm": 0.8222773671150208, "learning_rate": 0.0005, "loss": 2.2430152893066406, "step": 10070 }, { "epoch": 0.7939298383918013, "grad_norm": 0.7541828751564026, "learning_rate": 0.0005, "loss": 1.8794560432434082, "step": 10071 }, { "epoch": 0.7940086716594403, "grad_norm": 1.0958313941955566, "learning_rate": 0.0005, "loss": 1.926184058189392, "step": 10072 }, { "epoch": 0.7940875049270792, "grad_norm": 0.7057438492774963, "learning_rate": 0.0005, "loss": 1.9314353466033936, "step": 10073 }, { "epoch": 0.7941663381947182, "grad_norm": 1.2013572454452515, "learning_rate": 0.0005, "loss": 1.7468241453170776, "step": 10074 }, { "epoch": 0.7942451714623571, "grad_norm": 0.6763855814933777, "learning_rate": 0.0005, "loss": 2.0220208168029785, "step": 10075 }, { "epoch": 0.7943240047299961, "grad_norm": 0.7065508365631104, "learning_rate": 0.0005, "loss": 1.7746573686599731, "step": 10076 }, { "epoch": 0.794402837997635, "grad_norm": 1.0342681407928467, "learning_rate": 0.0005, "loss": 2.189373016357422, "step": 10077 }, { "epoch": 0.7944816712652739, "grad_norm": 0.7193498015403748, "learning_rate": 0.0005, "loss": 1.6672649383544922, "step": 10078 }, { "epoch": 0.7945605045329129, "grad_norm": 0.625390350818634, "learning_rate": 0.0005, "loss": 1.7639304399490356, "step": 10079 }, { "epoch": 0.7946393378005518, "grad_norm": 0.5849116444587708, "learning_rate": 0.0005, "loss": 1.9291478395462036, "step": 10080 }, { "epoch": 0.7947181710681908, "grad_norm": 0.6402416825294495, "learning_rate": 0.0005, "loss": 1.8677273988723755, "step": 10081 }, { "epoch": 0.7947970043358297, "grad_norm": 0.7486133575439453, "learning_rate": 0.0005, "loss": 1.8800368309020996, "step": 10082 }, { "epoch": 0.7948758376034687, "grad_norm": 0.6493003964424133, "learning_rate": 0.0005, "loss": 1.6744189262390137, "step": 10083 }, { "epoch": 0.7949546708711076, "grad_norm": 0.7016822099685669, "learning_rate": 0.0005, "loss": 1.8836064338684082, "step": 10084 }, { "epoch": 0.7950335041387465, "grad_norm": 0.9019405841827393, "learning_rate": 0.0005, "loss": 2.0035762786865234, "step": 10085 }, { "epoch": 0.7951123374063855, "grad_norm": 0.8710477352142334, "learning_rate": 0.0005, "loss": 1.8862228393554688, "step": 10086 }, { "epoch": 0.7951911706740245, "grad_norm": 0.6516849994659424, "learning_rate": 0.0005, "loss": 1.6615020036697388, "step": 10087 }, { "epoch": 0.7952700039416634, "grad_norm": 0.7374396324157715, "learning_rate": 0.0005, "loss": 2.0242984294891357, "step": 10088 }, { "epoch": 0.7953488372093023, "grad_norm": 0.5439767241477966, "learning_rate": 0.0005, "loss": 1.6907398700714111, "step": 10089 }, { "epoch": 0.7954276704769413, "grad_norm": 0.6809948682785034, "learning_rate": 0.0005, "loss": 2.1115708351135254, "step": 10090 }, { "epoch": 0.7955065037445802, "grad_norm": 0.6942641735076904, "learning_rate": 0.0005, "loss": 1.8307279348373413, "step": 10091 }, { "epoch": 0.7955853370122191, "grad_norm": 0.652752697467804, "learning_rate": 0.0005, "loss": 1.8021554946899414, "step": 10092 }, { "epoch": 0.7956641702798581, "grad_norm": 0.6342136263847351, "learning_rate": 0.0005, "loss": 2.121025800704956, "step": 10093 }, { "epoch": 0.795743003547497, "grad_norm": 0.4950277507305145, "learning_rate": 0.0005, "loss": 1.7071430683135986, "step": 10094 }, { "epoch": 0.795821836815136, "grad_norm": 0.8112697601318359, "learning_rate": 0.0005, "loss": 1.917161226272583, "step": 10095 }, { "epoch": 0.795900670082775, "grad_norm": 0.662248969078064, "learning_rate": 0.0005, "loss": 1.845322608947754, "step": 10096 }, { "epoch": 0.7959795033504139, "grad_norm": 0.810053288936615, "learning_rate": 0.0005, "loss": 1.8284094333648682, "step": 10097 }, { "epoch": 0.7960583366180528, "grad_norm": 0.7183748483657837, "learning_rate": 0.0005, "loss": 1.8837251663208008, "step": 10098 }, { "epoch": 0.7961371698856917, "grad_norm": 0.6211414933204651, "learning_rate": 0.0005, "loss": 1.772481918334961, "step": 10099 }, { "epoch": 0.7962160031533307, "grad_norm": 0.5489468574523926, "learning_rate": 0.0005, "loss": 2.0470380783081055, "step": 10100 }, { "epoch": 0.7962948364209697, "grad_norm": 0.676941454410553, "learning_rate": 0.0005, "loss": 1.9081809520721436, "step": 10101 }, { "epoch": 0.7963736696886086, "grad_norm": 0.7156005501747131, "learning_rate": 0.0005, "loss": 1.9380115270614624, "step": 10102 }, { "epoch": 0.7964525029562476, "grad_norm": 0.7335159182548523, "learning_rate": 0.0005, "loss": 1.7425410747528076, "step": 10103 }, { "epoch": 0.7965313362238865, "grad_norm": 0.7713028192520142, "learning_rate": 0.0005, "loss": 1.9051545858383179, "step": 10104 }, { "epoch": 0.7966101694915254, "grad_norm": 0.7134608626365662, "learning_rate": 0.0005, "loss": 2.03300142288208, "step": 10105 }, { "epoch": 0.7966890027591643, "grad_norm": 0.7845785617828369, "learning_rate": 0.0005, "loss": 1.925041913986206, "step": 10106 }, { "epoch": 0.7967678360268033, "grad_norm": 0.589975118637085, "learning_rate": 0.0005, "loss": 2.0871710777282715, "step": 10107 }, { "epoch": 0.7968466692944423, "grad_norm": 0.6634731292724609, "learning_rate": 0.0005, "loss": 1.8834185600280762, "step": 10108 }, { "epoch": 0.7969255025620812, "grad_norm": 0.5787336826324463, "learning_rate": 0.0005, "loss": 1.82682466506958, "step": 10109 }, { "epoch": 0.7970043358297202, "grad_norm": 0.6629981994628906, "learning_rate": 0.0005, "loss": 1.9273501634597778, "step": 10110 }, { "epoch": 0.7970831690973591, "grad_norm": 0.6634460091590881, "learning_rate": 0.0005, "loss": 1.6836930513381958, "step": 10111 }, { "epoch": 0.797162002364998, "grad_norm": 0.6612284779548645, "learning_rate": 0.0005, "loss": 1.8452801704406738, "step": 10112 }, { "epoch": 0.7972408356326369, "grad_norm": 0.566770076751709, "learning_rate": 0.0005, "loss": 1.6797676086425781, "step": 10113 }, { "epoch": 0.797319668900276, "grad_norm": 0.9533722996711731, "learning_rate": 0.0005, "loss": 1.9163436889648438, "step": 10114 }, { "epoch": 0.7973985021679149, "grad_norm": 0.6389642953872681, "learning_rate": 0.0005, "loss": 2.02101993560791, "step": 10115 }, { "epoch": 0.7974773354355538, "grad_norm": 0.7584618330001831, "learning_rate": 0.0005, "loss": 1.9551336765289307, "step": 10116 }, { "epoch": 0.7975561687031928, "grad_norm": 0.6099403500556946, "learning_rate": 0.0005, "loss": 2.196323871612549, "step": 10117 }, { "epoch": 0.7976350019708317, "grad_norm": 0.5986718535423279, "learning_rate": 0.0005, "loss": 1.7736849784851074, "step": 10118 }, { "epoch": 0.7977138352384706, "grad_norm": 0.7475489974021912, "learning_rate": 0.0005, "loss": 1.657238245010376, "step": 10119 }, { "epoch": 0.7977926685061095, "grad_norm": 0.5859628319740295, "learning_rate": 0.0005, "loss": 1.879140853881836, "step": 10120 }, { "epoch": 0.7978715017737485, "grad_norm": 0.6943318247795105, "learning_rate": 0.0005, "loss": 1.8291090726852417, "step": 10121 }, { "epoch": 0.7979503350413875, "grad_norm": 0.7127925157546997, "learning_rate": 0.0005, "loss": 1.7805081605911255, "step": 10122 }, { "epoch": 0.7980291683090264, "grad_norm": 0.6061229705810547, "learning_rate": 0.0005, "loss": 1.8495268821716309, "step": 10123 }, { "epoch": 0.7981080015766654, "grad_norm": 0.9323229789733887, "learning_rate": 0.0005, "loss": 1.944502592086792, "step": 10124 }, { "epoch": 0.7981868348443043, "grad_norm": 0.7064144015312195, "learning_rate": 0.0005, "loss": 2.0658607482910156, "step": 10125 }, { "epoch": 0.7982656681119432, "grad_norm": 0.5760524272918701, "learning_rate": 0.0005, "loss": 1.6451054811477661, "step": 10126 }, { "epoch": 0.7983445013795822, "grad_norm": 0.6561356782913208, "learning_rate": 0.0005, "loss": 1.9401321411132812, "step": 10127 }, { "epoch": 0.7984233346472212, "grad_norm": 0.7332654595375061, "learning_rate": 0.0005, "loss": 2.0738680362701416, "step": 10128 }, { "epoch": 0.7985021679148601, "grad_norm": 0.8350367546081543, "learning_rate": 0.0005, "loss": 1.9193952083587646, "step": 10129 }, { "epoch": 0.798581001182499, "grad_norm": 0.5516589879989624, "learning_rate": 0.0005, "loss": 1.6845521926879883, "step": 10130 }, { "epoch": 0.798659834450138, "grad_norm": 0.5102859735488892, "learning_rate": 0.0005, "loss": 1.7021963596343994, "step": 10131 }, { "epoch": 0.7987386677177769, "grad_norm": 0.6076673269271851, "learning_rate": 0.0005, "loss": 1.9514579772949219, "step": 10132 }, { "epoch": 0.7988175009854158, "grad_norm": 0.6467166543006897, "learning_rate": 0.0005, "loss": 1.8576579093933105, "step": 10133 }, { "epoch": 0.7988963342530548, "grad_norm": 0.6445178389549255, "learning_rate": 0.0005, "loss": 2.02227520942688, "step": 10134 }, { "epoch": 0.7989751675206938, "grad_norm": 0.7042179703712463, "learning_rate": 0.0005, "loss": 1.6288001537322998, "step": 10135 }, { "epoch": 0.7990540007883327, "grad_norm": 0.6797757744789124, "learning_rate": 0.0005, "loss": 2.031031608581543, "step": 10136 }, { "epoch": 0.7991328340559716, "grad_norm": 0.7443244457244873, "learning_rate": 0.0005, "loss": 1.5208513736724854, "step": 10137 }, { "epoch": 0.7992116673236106, "grad_norm": 0.6762390732765198, "learning_rate": 0.0005, "loss": 2.021712303161621, "step": 10138 }, { "epoch": 0.7992905005912495, "grad_norm": 0.6964069604873657, "learning_rate": 0.0005, "loss": 1.633188247680664, "step": 10139 }, { "epoch": 0.7993693338588884, "grad_norm": 0.6324161887168884, "learning_rate": 0.0005, "loss": 1.7948033809661865, "step": 10140 }, { "epoch": 0.7994481671265274, "grad_norm": 0.6891744136810303, "learning_rate": 0.0005, "loss": 1.7598233222961426, "step": 10141 }, { "epoch": 0.7995270003941664, "grad_norm": 0.6113512516021729, "learning_rate": 0.0005, "loss": 1.829998254776001, "step": 10142 }, { "epoch": 0.7996058336618053, "grad_norm": 0.6830698847770691, "learning_rate": 0.0005, "loss": 2.0298991203308105, "step": 10143 }, { "epoch": 0.7996846669294442, "grad_norm": 0.6979404091835022, "learning_rate": 0.0005, "loss": 1.8189172744750977, "step": 10144 }, { "epoch": 0.7997635001970832, "grad_norm": 0.6762039661407471, "learning_rate": 0.0005, "loss": 2.164142608642578, "step": 10145 }, { "epoch": 0.7998423334647221, "grad_norm": 0.7207293510437012, "learning_rate": 0.0005, "loss": 1.8518998622894287, "step": 10146 }, { "epoch": 0.799921166732361, "grad_norm": 0.6931858062744141, "learning_rate": 0.0005, "loss": 2.0087318420410156, "step": 10147 }, { "epoch": 0.8, "grad_norm": 0.654856264591217, "learning_rate": 0.0005, "loss": 1.9099688529968262, "step": 10148 }, { "epoch": 0.800078833267639, "grad_norm": 0.6377786993980408, "learning_rate": 0.0005, "loss": 2.0060720443725586, "step": 10149 }, { "epoch": 0.8001576665352779, "grad_norm": 0.5982405543327332, "learning_rate": 0.0005, "loss": 1.7468196153640747, "step": 10150 }, { "epoch": 0.8002364998029168, "grad_norm": 0.6043877601623535, "learning_rate": 0.0005, "loss": 2.0154566764831543, "step": 10151 }, { "epoch": 0.8003153330705558, "grad_norm": 0.7027552723884583, "learning_rate": 0.0005, "loss": 1.8972225189208984, "step": 10152 }, { "epoch": 0.8003941663381947, "grad_norm": 0.8331918716430664, "learning_rate": 0.0005, "loss": 1.7966840267181396, "step": 10153 }, { "epoch": 0.8004729996058336, "grad_norm": 0.6126558184623718, "learning_rate": 0.0005, "loss": 1.6018290519714355, "step": 10154 }, { "epoch": 0.8005518328734726, "grad_norm": 4.159585475921631, "learning_rate": 0.0005, "loss": 2.1135406494140625, "step": 10155 }, { "epoch": 0.8006306661411116, "grad_norm": 0.6376009583473206, "learning_rate": 0.0005, "loss": 1.6461148262023926, "step": 10156 }, { "epoch": 0.8007094994087505, "grad_norm": 0.7400861978530884, "learning_rate": 0.0005, "loss": 2.0464928150177, "step": 10157 }, { "epoch": 0.8007883326763895, "grad_norm": 0.5301267504692078, "learning_rate": 0.0005, "loss": 1.5769785642623901, "step": 10158 }, { "epoch": 0.8008671659440284, "grad_norm": 0.6383369565010071, "learning_rate": 0.0005, "loss": 1.8696231842041016, "step": 10159 }, { "epoch": 0.8009459992116673, "grad_norm": 0.606197714805603, "learning_rate": 0.0005, "loss": 1.874979853630066, "step": 10160 }, { "epoch": 0.8010248324793062, "grad_norm": 0.6416091918945312, "learning_rate": 0.0005, "loss": 1.8050332069396973, "step": 10161 }, { "epoch": 0.8011036657469452, "grad_norm": 0.5717421174049377, "learning_rate": 0.0005, "loss": 1.6893978118896484, "step": 10162 }, { "epoch": 0.8011824990145842, "grad_norm": 0.5718833208084106, "learning_rate": 0.0005, "loss": 1.860152006149292, "step": 10163 }, { "epoch": 0.8012613322822231, "grad_norm": 0.5548736453056335, "learning_rate": 0.0005, "loss": 1.6642539501190186, "step": 10164 }, { "epoch": 0.8013401655498621, "grad_norm": 0.49001091718673706, "learning_rate": 0.0005, "loss": 1.8079752922058105, "step": 10165 }, { "epoch": 0.801418998817501, "grad_norm": 0.5964950919151306, "learning_rate": 0.0005, "loss": 1.6492141485214233, "step": 10166 }, { "epoch": 0.8014978320851399, "grad_norm": 1.4483946561813354, "learning_rate": 0.0005, "loss": 2.025672435760498, "step": 10167 }, { "epoch": 0.8015766653527788, "grad_norm": 0.5802063941955566, "learning_rate": 0.0005, "loss": 1.7679221630096436, "step": 10168 }, { "epoch": 0.8016554986204179, "grad_norm": 0.5360571146011353, "learning_rate": 0.0005, "loss": 1.9998923540115356, "step": 10169 }, { "epoch": 0.8017343318880568, "grad_norm": 0.7119500041007996, "learning_rate": 0.0005, "loss": 1.9967622756958008, "step": 10170 }, { "epoch": 0.8018131651556957, "grad_norm": 0.5564509034156799, "learning_rate": 0.0005, "loss": 1.839604139328003, "step": 10171 }, { "epoch": 0.8018919984233347, "grad_norm": 0.8177748918533325, "learning_rate": 0.0005, "loss": 1.8871524333953857, "step": 10172 }, { "epoch": 0.8019708316909736, "grad_norm": 0.6830393671989441, "learning_rate": 0.0005, "loss": 1.8488004207611084, "step": 10173 }, { "epoch": 0.8020496649586125, "grad_norm": 0.5769047141075134, "learning_rate": 0.0005, "loss": 2.0509843826293945, "step": 10174 }, { "epoch": 0.8021284982262514, "grad_norm": 0.7253319025039673, "learning_rate": 0.0005, "loss": 1.7556589841842651, "step": 10175 }, { "epoch": 0.8022073314938905, "grad_norm": 0.6696925163269043, "learning_rate": 0.0005, "loss": 1.5911656618118286, "step": 10176 }, { "epoch": 0.8022861647615294, "grad_norm": 0.5777558088302612, "learning_rate": 0.0005, "loss": 1.7640283107757568, "step": 10177 }, { "epoch": 0.8023649980291683, "grad_norm": 0.7975102066993713, "learning_rate": 0.0005, "loss": 1.986182689666748, "step": 10178 }, { "epoch": 0.8024438312968073, "grad_norm": 0.6396498084068298, "learning_rate": 0.0005, "loss": 1.7692482471466064, "step": 10179 }, { "epoch": 0.8025226645644462, "grad_norm": 0.7574316263198853, "learning_rate": 0.0005, "loss": 2.0484790802001953, "step": 10180 }, { "epoch": 0.8026014978320851, "grad_norm": 0.6903258562088013, "learning_rate": 0.0005, "loss": 1.5952011346817017, "step": 10181 }, { "epoch": 0.802680331099724, "grad_norm": 0.8093411326408386, "learning_rate": 0.0005, "loss": 1.778954267501831, "step": 10182 }, { "epoch": 0.802759164367363, "grad_norm": 0.6230021119117737, "learning_rate": 0.0005, "loss": 1.8160332441329956, "step": 10183 }, { "epoch": 0.802837997635002, "grad_norm": 0.7062439918518066, "learning_rate": 0.0005, "loss": 1.6496188640594482, "step": 10184 }, { "epoch": 0.8029168309026409, "grad_norm": 0.7345288395881653, "learning_rate": 0.0005, "loss": 2.0108275413513184, "step": 10185 }, { "epoch": 0.8029956641702799, "grad_norm": 0.6936807632446289, "learning_rate": 0.0005, "loss": 1.6968727111816406, "step": 10186 }, { "epoch": 0.8030744974379188, "grad_norm": 1.0387461185455322, "learning_rate": 0.0005, "loss": 2.7462158203125, "step": 10187 }, { "epoch": 0.8031533307055577, "grad_norm": 1.830946922302246, "learning_rate": 0.0005, "loss": 2.015448570251465, "step": 10188 }, { "epoch": 0.8032321639731967, "grad_norm": 0.5731258392333984, "learning_rate": 0.0005, "loss": 1.564621090888977, "step": 10189 }, { "epoch": 0.8033109972408357, "grad_norm": 0.8418495059013367, "learning_rate": 0.0005, "loss": 1.923640489578247, "step": 10190 }, { "epoch": 0.8033898305084746, "grad_norm": 1.036604881286621, "learning_rate": 0.0005, "loss": 1.988736867904663, "step": 10191 }, { "epoch": 0.8034686637761135, "grad_norm": 0.5865787267684937, "learning_rate": 0.0005, "loss": 1.707292914390564, "step": 10192 }, { "epoch": 0.8035474970437525, "grad_norm": 0.6338862776756287, "learning_rate": 0.0005, "loss": 1.779612421989441, "step": 10193 }, { "epoch": 0.8036263303113914, "grad_norm": 0.6395756602287292, "learning_rate": 0.0005, "loss": 1.8124957084655762, "step": 10194 }, { "epoch": 0.8037051635790303, "grad_norm": 1.5687557458877563, "learning_rate": 0.0005, "loss": 1.8111215829849243, "step": 10195 }, { "epoch": 0.8037839968466693, "grad_norm": 0.6829225420951843, "learning_rate": 0.0005, "loss": 1.636752963066101, "step": 10196 }, { "epoch": 0.8038628301143083, "grad_norm": 1.530887484550476, "learning_rate": 0.0005, "loss": 2.1367950439453125, "step": 10197 }, { "epoch": 0.8039416633819472, "grad_norm": 0.5516893267631531, "learning_rate": 0.0005, "loss": 1.7038135528564453, "step": 10198 }, { "epoch": 0.8040204966495861, "grad_norm": 0.6139196753501892, "learning_rate": 0.0005, "loss": 1.9660496711730957, "step": 10199 }, { "epoch": 0.8040993299172251, "grad_norm": 0.547343373298645, "learning_rate": 0.0005, "loss": 1.7539666891098022, "step": 10200 }, { "epoch": 0.804178163184864, "grad_norm": 0.6647371649742126, "learning_rate": 0.0005, "loss": 1.762692928314209, "step": 10201 }, { "epoch": 0.8042569964525029, "grad_norm": 0.5560408234596252, "learning_rate": 0.0005, "loss": 1.858231782913208, "step": 10202 }, { "epoch": 0.804335829720142, "grad_norm": 0.6452599167823792, "learning_rate": 0.0005, "loss": 1.836357593536377, "step": 10203 }, { "epoch": 0.8044146629877809, "grad_norm": 0.6023591160774231, "learning_rate": 0.0005, "loss": 2.172974109649658, "step": 10204 }, { "epoch": 0.8044934962554198, "grad_norm": 0.683203399181366, "learning_rate": 0.0005, "loss": 2.0194525718688965, "step": 10205 }, { "epoch": 0.8045723295230587, "grad_norm": 0.9185241460800171, "learning_rate": 0.0005, "loss": 1.9760539531707764, "step": 10206 }, { "epoch": 0.8046511627906977, "grad_norm": 0.543978750705719, "learning_rate": 0.0005, "loss": 1.8885258436203003, "step": 10207 }, { "epoch": 0.8047299960583366, "grad_norm": 0.5395861268043518, "learning_rate": 0.0005, "loss": 1.6909677982330322, "step": 10208 }, { "epoch": 0.8048088293259755, "grad_norm": 0.6469770669937134, "learning_rate": 0.0005, "loss": 1.8338391780853271, "step": 10209 }, { "epoch": 0.8048876625936146, "grad_norm": 0.932448148727417, "learning_rate": 0.0005, "loss": 1.9241650104522705, "step": 10210 }, { "epoch": 0.8049664958612535, "grad_norm": 1.5516631603240967, "learning_rate": 0.0005, "loss": 2.1253137588500977, "step": 10211 }, { "epoch": 0.8050453291288924, "grad_norm": 0.6699617505073547, "learning_rate": 0.0005, "loss": 1.913634181022644, "step": 10212 }, { "epoch": 0.8051241623965313, "grad_norm": 0.7393274903297424, "learning_rate": 0.0005, "loss": 1.947312831878662, "step": 10213 }, { "epoch": 0.8052029956641703, "grad_norm": 0.5663841366767883, "learning_rate": 0.0005, "loss": 1.7667298316955566, "step": 10214 }, { "epoch": 0.8052818289318092, "grad_norm": 0.7034714818000793, "learning_rate": 0.0005, "loss": 2.0454354286193848, "step": 10215 }, { "epoch": 0.8053606621994481, "grad_norm": 0.6418812274932861, "learning_rate": 0.0005, "loss": 1.9273076057434082, "step": 10216 }, { "epoch": 0.8054394954670872, "grad_norm": 0.6429540514945984, "learning_rate": 0.0005, "loss": 1.8207746744155884, "step": 10217 }, { "epoch": 0.8055183287347261, "grad_norm": 0.5337487459182739, "learning_rate": 0.0005, "loss": 1.875537633895874, "step": 10218 }, { "epoch": 0.805597162002365, "grad_norm": 0.6473302841186523, "learning_rate": 0.0005, "loss": 1.9966645240783691, "step": 10219 }, { "epoch": 0.8056759952700039, "grad_norm": 0.8091647028923035, "learning_rate": 0.0005, "loss": 2.01987886428833, "step": 10220 }, { "epoch": 0.8057548285376429, "grad_norm": 0.5726760625839233, "learning_rate": 0.0005, "loss": 1.5842196941375732, "step": 10221 }, { "epoch": 0.8058336618052818, "grad_norm": 0.8780947923660278, "learning_rate": 0.0005, "loss": 1.9063743352890015, "step": 10222 }, { "epoch": 0.8059124950729207, "grad_norm": 0.7437862753868103, "learning_rate": 0.0005, "loss": 1.9935047626495361, "step": 10223 }, { "epoch": 0.8059913283405598, "grad_norm": 1.4139509201049805, "learning_rate": 0.0005, "loss": 2.137777090072632, "step": 10224 }, { "epoch": 0.8060701616081987, "grad_norm": 0.5817727446556091, "learning_rate": 0.0005, "loss": 1.9480701684951782, "step": 10225 }, { "epoch": 0.8061489948758376, "grad_norm": 0.61297208070755, "learning_rate": 0.0005, "loss": 1.798162817955017, "step": 10226 }, { "epoch": 0.8062278281434766, "grad_norm": 0.7873659133911133, "learning_rate": 0.0005, "loss": 1.9577605724334717, "step": 10227 }, { "epoch": 0.8063066614111155, "grad_norm": 0.6542823910713196, "learning_rate": 0.0005, "loss": 1.617134690284729, "step": 10228 }, { "epoch": 0.8063854946787544, "grad_norm": 0.6568942666053772, "learning_rate": 0.0005, "loss": 2.0018157958984375, "step": 10229 }, { "epoch": 0.8064643279463933, "grad_norm": 0.624214768409729, "learning_rate": 0.0005, "loss": 1.8312822580337524, "step": 10230 }, { "epoch": 0.8065431612140324, "grad_norm": 0.6191061735153198, "learning_rate": 0.0005, "loss": 1.6838711500167847, "step": 10231 }, { "epoch": 0.8066219944816713, "grad_norm": 0.7306764125823975, "learning_rate": 0.0005, "loss": 2.0556488037109375, "step": 10232 }, { "epoch": 0.8067008277493102, "grad_norm": 0.7254113554954529, "learning_rate": 0.0005, "loss": 1.738940954208374, "step": 10233 }, { "epoch": 0.8067796610169492, "grad_norm": 0.5677376985549927, "learning_rate": 0.0005, "loss": 1.73617422580719, "step": 10234 }, { "epoch": 0.8068584942845881, "grad_norm": 0.6175091862678528, "learning_rate": 0.0005, "loss": 1.9055910110473633, "step": 10235 }, { "epoch": 0.806937327552227, "grad_norm": 0.652140736579895, "learning_rate": 0.0005, "loss": 1.5620090961456299, "step": 10236 }, { "epoch": 0.8070161608198659, "grad_norm": 0.7581016421318054, "learning_rate": 0.0005, "loss": 2.270683765411377, "step": 10237 }, { "epoch": 0.807094994087505, "grad_norm": 0.9464313387870789, "learning_rate": 0.0005, "loss": 1.944315791130066, "step": 10238 }, { "epoch": 0.8071738273551439, "grad_norm": 0.7957989573478699, "learning_rate": 0.0005, "loss": 2.1905713081359863, "step": 10239 }, { "epoch": 0.8072526606227828, "grad_norm": 0.5253204107284546, "learning_rate": 0.0005, "loss": 1.905442237854004, "step": 10240 }, { "epoch": 0.8073314938904218, "grad_norm": 0.6074486970901489, "learning_rate": 0.0005, "loss": 1.9627677202224731, "step": 10241 }, { "epoch": 0.8074103271580607, "grad_norm": 0.5802618861198425, "learning_rate": 0.0005, "loss": 1.7475333213806152, "step": 10242 }, { "epoch": 0.8074891604256996, "grad_norm": 0.6254745125770569, "learning_rate": 0.0005, "loss": 1.6011468172073364, "step": 10243 }, { "epoch": 0.8075679936933385, "grad_norm": 0.5944474935531616, "learning_rate": 0.0005, "loss": 1.802202582359314, "step": 10244 }, { "epoch": 0.8076468269609776, "grad_norm": 0.8293747901916504, "learning_rate": 0.0005, "loss": 1.7444255352020264, "step": 10245 }, { "epoch": 0.8077256602286165, "grad_norm": 0.7552623152732849, "learning_rate": 0.0005, "loss": 1.8461337089538574, "step": 10246 }, { "epoch": 0.8078044934962554, "grad_norm": 0.5753352046012878, "learning_rate": 0.0005, "loss": 1.7197351455688477, "step": 10247 }, { "epoch": 0.8078833267638944, "grad_norm": 0.6571676135063171, "learning_rate": 0.0005, "loss": 1.7449243068695068, "step": 10248 }, { "epoch": 0.8079621600315333, "grad_norm": 0.7275256514549255, "learning_rate": 0.0005, "loss": 1.941666603088379, "step": 10249 }, { "epoch": 0.8080409932991722, "grad_norm": 0.5781152844429016, "learning_rate": 0.0005, "loss": 1.845287561416626, "step": 10250 }, { "epoch": 0.8081198265668111, "grad_norm": 0.6175826191902161, "learning_rate": 0.0005, "loss": 1.7041256427764893, "step": 10251 }, { "epoch": 0.8081986598344502, "grad_norm": 0.6924351453781128, "learning_rate": 0.0005, "loss": 2.0472941398620605, "step": 10252 }, { "epoch": 0.8082774931020891, "grad_norm": 0.6139202117919922, "learning_rate": 0.0005, "loss": 1.9057700634002686, "step": 10253 }, { "epoch": 0.808356326369728, "grad_norm": 0.550475001335144, "learning_rate": 0.0005, "loss": 1.4838292598724365, "step": 10254 }, { "epoch": 0.808435159637367, "grad_norm": 0.7179314494132996, "learning_rate": 0.0005, "loss": 2.111476421356201, "step": 10255 }, { "epoch": 0.8085139929050059, "grad_norm": 0.6839327812194824, "learning_rate": 0.0005, "loss": 2.0281882286071777, "step": 10256 }, { "epoch": 0.8085928261726448, "grad_norm": 0.5861836075782776, "learning_rate": 0.0005, "loss": 1.7323344945907593, "step": 10257 }, { "epoch": 0.8086716594402839, "grad_norm": 0.7347720265388489, "learning_rate": 0.0005, "loss": 1.9723155498504639, "step": 10258 }, { "epoch": 0.8087504927079228, "grad_norm": 0.6099910736083984, "learning_rate": 0.0005, "loss": 1.8395106792449951, "step": 10259 }, { "epoch": 0.8088293259755617, "grad_norm": 0.6967657804489136, "learning_rate": 0.0005, "loss": 1.9967654943466187, "step": 10260 }, { "epoch": 0.8089081592432006, "grad_norm": 0.6467057466506958, "learning_rate": 0.0005, "loss": 1.6560423374176025, "step": 10261 }, { "epoch": 0.8089869925108396, "grad_norm": 0.6651995182037354, "learning_rate": 0.0005, "loss": 1.818180799484253, "step": 10262 }, { "epoch": 0.8090658257784785, "grad_norm": 0.6502604484558105, "learning_rate": 0.0005, "loss": 2.091109275817871, "step": 10263 }, { "epoch": 0.8091446590461174, "grad_norm": 0.5625911355018616, "learning_rate": 0.0005, "loss": 1.719103217124939, "step": 10264 }, { "epoch": 0.8092234923137565, "grad_norm": 0.7189836502075195, "learning_rate": 0.0005, "loss": 1.9513258934020996, "step": 10265 }, { "epoch": 0.8093023255813954, "grad_norm": 0.6550812125205994, "learning_rate": 0.0005, "loss": 1.7909209728240967, "step": 10266 }, { "epoch": 0.8093811588490343, "grad_norm": 0.668787956237793, "learning_rate": 0.0005, "loss": 1.7860946655273438, "step": 10267 }, { "epoch": 0.8094599921166732, "grad_norm": 0.6877173781394958, "learning_rate": 0.0005, "loss": 1.7044157981872559, "step": 10268 }, { "epoch": 0.8095388253843122, "grad_norm": 0.7973102331161499, "learning_rate": 0.0005, "loss": 1.832763433456421, "step": 10269 }, { "epoch": 0.8096176586519511, "grad_norm": 0.6035998463630676, "learning_rate": 0.0005, "loss": 1.5896036624908447, "step": 10270 }, { "epoch": 0.80969649191959, "grad_norm": 0.8007937073707581, "learning_rate": 0.0005, "loss": 1.6433948278427124, "step": 10271 }, { "epoch": 0.8097753251872291, "grad_norm": 1.1511083841323853, "learning_rate": 0.0005, "loss": 2.0495200157165527, "step": 10272 }, { "epoch": 0.809854158454868, "grad_norm": 0.7138436436653137, "learning_rate": 0.0005, "loss": 2.2325000762939453, "step": 10273 }, { "epoch": 0.8099329917225069, "grad_norm": 0.900679886341095, "learning_rate": 0.0005, "loss": 1.8890870809555054, "step": 10274 }, { "epoch": 0.8100118249901458, "grad_norm": 0.5927530527114868, "learning_rate": 0.0005, "loss": 1.6987184286117554, "step": 10275 }, { "epoch": 0.8100906582577848, "grad_norm": 0.7842803001403809, "learning_rate": 0.0005, "loss": 1.9581444263458252, "step": 10276 }, { "epoch": 0.8101694915254237, "grad_norm": 0.5969504117965698, "learning_rate": 0.0005, "loss": 1.7107924222946167, "step": 10277 }, { "epoch": 0.8102483247930626, "grad_norm": 0.7391321063041687, "learning_rate": 0.0005, "loss": 1.8747278451919556, "step": 10278 }, { "epoch": 0.8103271580607017, "grad_norm": 0.5962021946907043, "learning_rate": 0.0005, "loss": 2.094252586364746, "step": 10279 }, { "epoch": 0.8104059913283406, "grad_norm": 1.0856784582138062, "learning_rate": 0.0005, "loss": 1.8725876808166504, "step": 10280 }, { "epoch": 0.8104848245959795, "grad_norm": 0.8648874759674072, "learning_rate": 0.0005, "loss": 1.8495713472366333, "step": 10281 }, { "epoch": 0.8105636578636184, "grad_norm": 0.763118326663971, "learning_rate": 0.0005, "loss": 2.147754192352295, "step": 10282 }, { "epoch": 0.8106424911312574, "grad_norm": 0.7159836292266846, "learning_rate": 0.0005, "loss": 1.8308448791503906, "step": 10283 }, { "epoch": 0.8107213243988963, "grad_norm": 0.7309080958366394, "learning_rate": 0.0005, "loss": 1.904660701751709, "step": 10284 }, { "epoch": 0.8108001576665352, "grad_norm": 0.7212557196617126, "learning_rate": 0.0005, "loss": 1.9319705963134766, "step": 10285 }, { "epoch": 0.8108789909341743, "grad_norm": 0.6689227819442749, "learning_rate": 0.0005, "loss": 2.1375465393066406, "step": 10286 }, { "epoch": 0.8109578242018132, "grad_norm": 0.6588737368583679, "learning_rate": 0.0005, "loss": 1.9402170181274414, "step": 10287 }, { "epoch": 0.8110366574694521, "grad_norm": 0.5345249176025391, "learning_rate": 0.0005, "loss": 1.421384572982788, "step": 10288 }, { "epoch": 0.8111154907370911, "grad_norm": 0.7455499768257141, "learning_rate": 0.0005, "loss": 1.9461369514465332, "step": 10289 }, { "epoch": 0.81119432400473, "grad_norm": 0.7245396971702576, "learning_rate": 0.0005, "loss": 1.877344012260437, "step": 10290 }, { "epoch": 0.8112731572723689, "grad_norm": 0.5858057141304016, "learning_rate": 0.0005, "loss": 1.848118782043457, "step": 10291 }, { "epoch": 0.8113519905400078, "grad_norm": 0.6302492618560791, "learning_rate": 0.0005, "loss": 1.6874237060546875, "step": 10292 }, { "epoch": 0.8114308238076469, "grad_norm": 0.5514097213745117, "learning_rate": 0.0005, "loss": 1.9320541620254517, "step": 10293 }, { "epoch": 0.8115096570752858, "grad_norm": 0.684113085269928, "learning_rate": 0.0005, "loss": 1.6359996795654297, "step": 10294 }, { "epoch": 0.8115884903429247, "grad_norm": 0.694005012512207, "learning_rate": 0.0005, "loss": 1.8487364053726196, "step": 10295 }, { "epoch": 0.8116673236105637, "grad_norm": 0.6318814754486084, "learning_rate": 0.0005, "loss": 2.1231160163879395, "step": 10296 }, { "epoch": 0.8117461568782026, "grad_norm": 0.6533340215682983, "learning_rate": 0.0005, "loss": 1.835360050201416, "step": 10297 }, { "epoch": 0.8118249901458415, "grad_norm": 0.6099222302436829, "learning_rate": 0.0005, "loss": 2.0044350624084473, "step": 10298 }, { "epoch": 0.8119038234134804, "grad_norm": 0.5823947787284851, "learning_rate": 0.0005, "loss": 2.081602096557617, "step": 10299 }, { "epoch": 0.8119826566811195, "grad_norm": 0.7971334457397461, "learning_rate": 0.0005, "loss": 1.9261209964752197, "step": 10300 }, { "epoch": 0.8120614899487584, "grad_norm": 0.6402869820594788, "learning_rate": 0.0005, "loss": 1.5494723320007324, "step": 10301 }, { "epoch": 0.8121403232163973, "grad_norm": 0.6486212015151978, "learning_rate": 0.0005, "loss": 2.0075368881225586, "step": 10302 }, { "epoch": 0.8122191564840363, "grad_norm": 0.8076770305633545, "learning_rate": 0.0005, "loss": 2.079641819000244, "step": 10303 }, { "epoch": 0.8122979897516752, "grad_norm": 0.6441473960876465, "learning_rate": 0.0005, "loss": 1.7654139995574951, "step": 10304 }, { "epoch": 0.8123768230193141, "grad_norm": 0.6831235289573669, "learning_rate": 0.0005, "loss": 1.741208553314209, "step": 10305 }, { "epoch": 0.812455656286953, "grad_norm": 0.6882385611534119, "learning_rate": 0.0005, "loss": 1.9200289249420166, "step": 10306 }, { "epoch": 0.8125344895545921, "grad_norm": 0.5754812359809875, "learning_rate": 0.0005, "loss": 2.0325522422790527, "step": 10307 }, { "epoch": 0.812613322822231, "grad_norm": 0.5646578669548035, "learning_rate": 0.0005, "loss": 1.6517155170440674, "step": 10308 }, { "epoch": 0.8126921560898699, "grad_norm": 0.6773731708526611, "learning_rate": 0.0005, "loss": 1.993030309677124, "step": 10309 }, { "epoch": 0.8127709893575089, "grad_norm": 0.6261622309684753, "learning_rate": 0.0005, "loss": 1.9797027111053467, "step": 10310 }, { "epoch": 0.8128498226251478, "grad_norm": 0.6221500635147095, "learning_rate": 0.0005, "loss": 2.0653486251831055, "step": 10311 }, { "epoch": 0.8129286558927867, "grad_norm": 0.7445411086082458, "learning_rate": 0.0005, "loss": 1.916618824005127, "step": 10312 }, { "epoch": 0.8130074891604256, "grad_norm": 0.733810544013977, "learning_rate": 0.0005, "loss": 1.9091100692749023, "step": 10313 }, { "epoch": 0.8130863224280647, "grad_norm": 0.768422544002533, "learning_rate": 0.0005, "loss": 2.016235113143921, "step": 10314 }, { "epoch": 0.8131651556957036, "grad_norm": 0.7169650197029114, "learning_rate": 0.0005, "loss": 2.116206169128418, "step": 10315 }, { "epoch": 0.8132439889633425, "grad_norm": 0.652172327041626, "learning_rate": 0.0005, "loss": 1.8635817766189575, "step": 10316 }, { "epoch": 0.8133228222309815, "grad_norm": 0.7159938812255859, "learning_rate": 0.0005, "loss": 1.9943759441375732, "step": 10317 }, { "epoch": 0.8134016554986204, "grad_norm": 0.7279742360115051, "learning_rate": 0.0005, "loss": 1.6209745407104492, "step": 10318 }, { "epoch": 0.8134804887662593, "grad_norm": 0.5490813851356506, "learning_rate": 0.0005, "loss": 1.8381354808807373, "step": 10319 }, { "epoch": 0.8135593220338984, "grad_norm": 0.8989583253860474, "learning_rate": 0.0005, "loss": 1.9087848663330078, "step": 10320 }, { "epoch": 0.8136381553015373, "grad_norm": 0.7006810307502747, "learning_rate": 0.0005, "loss": 1.7001738548278809, "step": 10321 }, { "epoch": 0.8137169885691762, "grad_norm": 1.7750918865203857, "learning_rate": 0.0005, "loss": 1.862262487411499, "step": 10322 }, { "epoch": 0.8137958218368151, "grad_norm": 0.626507580280304, "learning_rate": 0.0005, "loss": 1.8219950199127197, "step": 10323 }, { "epoch": 0.8138746551044541, "grad_norm": 0.6640881896018982, "learning_rate": 0.0005, "loss": 1.9168014526367188, "step": 10324 }, { "epoch": 0.813953488372093, "grad_norm": 0.7055761814117432, "learning_rate": 0.0005, "loss": 1.9436049461364746, "step": 10325 }, { "epoch": 0.8140323216397319, "grad_norm": 0.7343440055847168, "learning_rate": 0.0005, "loss": 1.8036558628082275, "step": 10326 }, { "epoch": 0.814111154907371, "grad_norm": 0.8903201222419739, "learning_rate": 0.0005, "loss": 1.8049061298370361, "step": 10327 }, { "epoch": 0.8141899881750099, "grad_norm": 0.6532639861106873, "learning_rate": 0.0005, "loss": 1.725052833557129, "step": 10328 }, { "epoch": 0.8142688214426488, "grad_norm": 0.6989489197731018, "learning_rate": 0.0005, "loss": 1.8808244466781616, "step": 10329 }, { "epoch": 0.8143476547102877, "grad_norm": 0.642649233341217, "learning_rate": 0.0005, "loss": 1.8447973728179932, "step": 10330 }, { "epoch": 0.8144264879779267, "grad_norm": 0.7329175472259521, "learning_rate": 0.0005, "loss": 1.902578353881836, "step": 10331 }, { "epoch": 0.8145053212455656, "grad_norm": 0.9880040287971497, "learning_rate": 0.0005, "loss": 1.8432440757751465, "step": 10332 }, { "epoch": 0.8145841545132045, "grad_norm": 0.5974106788635254, "learning_rate": 0.0005, "loss": 1.870619773864746, "step": 10333 }, { "epoch": 0.8146629877808436, "grad_norm": 1.099688172340393, "learning_rate": 0.0005, "loss": 1.8963305950164795, "step": 10334 }, { "epoch": 0.8147418210484825, "grad_norm": 0.6390479803085327, "learning_rate": 0.0005, "loss": 1.8468412160873413, "step": 10335 }, { "epoch": 0.8148206543161214, "grad_norm": 0.7753957509994507, "learning_rate": 0.0005, "loss": 2.130946636199951, "step": 10336 }, { "epoch": 0.8148994875837603, "grad_norm": 0.6642528176307678, "learning_rate": 0.0005, "loss": 2.0041728019714355, "step": 10337 }, { "epoch": 0.8149783208513993, "grad_norm": 0.7178678512573242, "learning_rate": 0.0005, "loss": 1.746283769607544, "step": 10338 }, { "epoch": 0.8150571541190382, "grad_norm": 0.6519806385040283, "learning_rate": 0.0005, "loss": 2.0215988159179688, "step": 10339 }, { "epoch": 0.8151359873866771, "grad_norm": 1.0301567316055298, "learning_rate": 0.0005, "loss": 1.9714012145996094, "step": 10340 }, { "epoch": 0.8152148206543162, "grad_norm": 0.6765994429588318, "learning_rate": 0.0005, "loss": 1.8040213584899902, "step": 10341 }, { "epoch": 0.8152936539219551, "grad_norm": 0.5474951863288879, "learning_rate": 0.0005, "loss": 1.8465044498443604, "step": 10342 }, { "epoch": 0.815372487189594, "grad_norm": 0.5945013761520386, "learning_rate": 0.0005, "loss": 1.4956517219543457, "step": 10343 }, { "epoch": 0.8154513204572329, "grad_norm": 0.7036038637161255, "learning_rate": 0.0005, "loss": 1.8856089115142822, "step": 10344 }, { "epoch": 0.8155301537248719, "grad_norm": 0.5899015069007874, "learning_rate": 0.0005, "loss": 1.6870157718658447, "step": 10345 }, { "epoch": 0.8156089869925108, "grad_norm": 0.96673983335495, "learning_rate": 0.0005, "loss": 1.8588435649871826, "step": 10346 }, { "epoch": 0.8156878202601497, "grad_norm": 0.5846199989318848, "learning_rate": 0.0005, "loss": 1.6650059223175049, "step": 10347 }, { "epoch": 0.8157666535277888, "grad_norm": 0.6508434414863586, "learning_rate": 0.0005, "loss": 1.8758478164672852, "step": 10348 }, { "epoch": 0.8158454867954277, "grad_norm": 0.6685184836387634, "learning_rate": 0.0005, "loss": 1.863072395324707, "step": 10349 }, { "epoch": 0.8159243200630666, "grad_norm": 0.47137969732284546, "learning_rate": 0.0005, "loss": 1.3759400844573975, "step": 10350 }, { "epoch": 0.8160031533307056, "grad_norm": 0.49202191829681396, "learning_rate": 0.0005, "loss": 1.6718082427978516, "step": 10351 }, { "epoch": 0.8160819865983445, "grad_norm": 0.6063372492790222, "learning_rate": 0.0005, "loss": 1.782845139503479, "step": 10352 }, { "epoch": 0.8161608198659834, "grad_norm": 0.5240353345870972, "learning_rate": 0.0005, "loss": 1.8546043634414673, "step": 10353 }, { "epoch": 0.8162396531336223, "grad_norm": 0.6548762321472168, "learning_rate": 0.0005, "loss": 2.0228445529937744, "step": 10354 }, { "epoch": 0.8163184864012614, "grad_norm": 0.7071240544319153, "learning_rate": 0.0005, "loss": 1.6852729320526123, "step": 10355 }, { "epoch": 0.8163973196689003, "grad_norm": 0.5846965312957764, "learning_rate": 0.0005, "loss": 1.933760643005371, "step": 10356 }, { "epoch": 0.8164761529365392, "grad_norm": 0.7517383694648743, "learning_rate": 0.0005, "loss": 1.7144027948379517, "step": 10357 }, { "epoch": 0.8165549862041782, "grad_norm": 0.6680002808570862, "learning_rate": 0.0005, "loss": 2.110260486602783, "step": 10358 }, { "epoch": 0.8166338194718171, "grad_norm": 0.5947278738021851, "learning_rate": 0.0005, "loss": 1.965559482574463, "step": 10359 }, { "epoch": 0.816712652739456, "grad_norm": 0.5546758770942688, "learning_rate": 0.0005, "loss": 1.738469123840332, "step": 10360 }, { "epoch": 0.816791486007095, "grad_norm": 0.592871367931366, "learning_rate": 0.0005, "loss": 1.8892040252685547, "step": 10361 }, { "epoch": 0.816870319274734, "grad_norm": 0.6505705714225769, "learning_rate": 0.0005, "loss": 1.9051042795181274, "step": 10362 }, { "epoch": 0.8169491525423729, "grad_norm": 0.5665449500083923, "learning_rate": 0.0005, "loss": 1.798288106918335, "step": 10363 }, { "epoch": 0.8170279858100118, "grad_norm": 0.6189061403274536, "learning_rate": 0.0005, "loss": 2.01316499710083, "step": 10364 }, { "epoch": 0.8171068190776508, "grad_norm": 0.7174134850502014, "learning_rate": 0.0005, "loss": 1.8049827814102173, "step": 10365 }, { "epoch": 0.8171856523452897, "grad_norm": 0.6885195374488831, "learning_rate": 0.0005, "loss": 1.9355106353759766, "step": 10366 }, { "epoch": 0.8172644856129286, "grad_norm": 0.5166360139846802, "learning_rate": 0.0005, "loss": 1.7896249294281006, "step": 10367 }, { "epoch": 0.8173433188805675, "grad_norm": 0.6031656265258789, "learning_rate": 0.0005, "loss": 2.0602951049804688, "step": 10368 }, { "epoch": 0.8174221521482066, "grad_norm": 0.6259276270866394, "learning_rate": 0.0005, "loss": 1.7410080432891846, "step": 10369 }, { "epoch": 0.8175009854158455, "grad_norm": 0.8942043781280518, "learning_rate": 0.0005, "loss": 1.876682996749878, "step": 10370 }, { "epoch": 0.8175798186834844, "grad_norm": 0.7339830994606018, "learning_rate": 0.0005, "loss": 1.9535075426101685, "step": 10371 }, { "epoch": 0.8176586519511234, "grad_norm": 0.8530148267745972, "learning_rate": 0.0005, "loss": 1.899878740310669, "step": 10372 }, { "epoch": 0.8177374852187623, "grad_norm": 0.552130937576294, "learning_rate": 0.0005, "loss": 1.6854907274246216, "step": 10373 }, { "epoch": 0.8178163184864012, "grad_norm": 0.682909369468689, "learning_rate": 0.0005, "loss": 1.7321964502334595, "step": 10374 }, { "epoch": 0.8178951517540402, "grad_norm": 0.7658067941665649, "learning_rate": 0.0005, "loss": 1.7358182668685913, "step": 10375 }, { "epoch": 0.8179739850216792, "grad_norm": 0.5939306616783142, "learning_rate": 0.0005, "loss": 1.8192360401153564, "step": 10376 }, { "epoch": 0.8180528182893181, "grad_norm": 0.725644588470459, "learning_rate": 0.0005, "loss": 2.210176467895508, "step": 10377 }, { "epoch": 0.818131651556957, "grad_norm": 0.6114161014556885, "learning_rate": 0.0005, "loss": 1.740811824798584, "step": 10378 }, { "epoch": 0.818210484824596, "grad_norm": 0.6145317554473877, "learning_rate": 0.0005, "loss": 1.853095293045044, "step": 10379 }, { "epoch": 0.8182893180922349, "grad_norm": 0.6283078193664551, "learning_rate": 0.0005, "loss": 2.0621697902679443, "step": 10380 }, { "epoch": 0.8183681513598738, "grad_norm": 0.7028976082801819, "learning_rate": 0.0005, "loss": 1.804787278175354, "step": 10381 }, { "epoch": 0.8184469846275129, "grad_norm": 0.8050670027732849, "learning_rate": 0.0005, "loss": 2.109203338623047, "step": 10382 }, { "epoch": 0.8185258178951518, "grad_norm": 0.6253171563148499, "learning_rate": 0.0005, "loss": 1.7398768663406372, "step": 10383 }, { "epoch": 0.8186046511627907, "grad_norm": 0.719923734664917, "learning_rate": 0.0005, "loss": 1.9460811614990234, "step": 10384 }, { "epoch": 0.8186834844304296, "grad_norm": 0.6097255945205688, "learning_rate": 0.0005, "loss": 1.697511076927185, "step": 10385 }, { "epoch": 0.8187623176980686, "grad_norm": 0.7830270528793335, "learning_rate": 0.0005, "loss": 1.8810559511184692, "step": 10386 }, { "epoch": 0.8188411509657075, "grad_norm": 0.6173020005226135, "learning_rate": 0.0005, "loss": 1.8813436031341553, "step": 10387 }, { "epoch": 0.8189199842333464, "grad_norm": 0.904902458190918, "learning_rate": 0.0005, "loss": 1.8769413232803345, "step": 10388 }, { "epoch": 0.8189988175009855, "grad_norm": 0.6768904328346252, "learning_rate": 0.0005, "loss": 2.040982246398926, "step": 10389 }, { "epoch": 0.8190776507686244, "grad_norm": 0.58603835105896, "learning_rate": 0.0005, "loss": 1.8030756711959839, "step": 10390 }, { "epoch": 0.8191564840362633, "grad_norm": 0.7138760685920715, "learning_rate": 0.0005, "loss": 2.1828651428222656, "step": 10391 }, { "epoch": 0.8192353173039022, "grad_norm": 0.6366111040115356, "learning_rate": 0.0005, "loss": 1.920366883277893, "step": 10392 }, { "epoch": 0.8193141505715412, "grad_norm": 0.7090716361999512, "learning_rate": 0.0005, "loss": 1.9255385398864746, "step": 10393 }, { "epoch": 0.8193929838391801, "grad_norm": 0.6245114803314209, "learning_rate": 0.0005, "loss": 1.905432939529419, "step": 10394 }, { "epoch": 0.819471817106819, "grad_norm": 0.5143139958381653, "learning_rate": 0.0005, "loss": 1.7215843200683594, "step": 10395 }, { "epoch": 0.8195506503744581, "grad_norm": 0.7377063035964966, "learning_rate": 0.0005, "loss": 1.9177100658416748, "step": 10396 }, { "epoch": 0.819629483642097, "grad_norm": 0.7358789443969727, "learning_rate": 0.0005, "loss": 1.7139474153518677, "step": 10397 }, { "epoch": 0.8197083169097359, "grad_norm": 2.487450122833252, "learning_rate": 0.0005, "loss": 1.8775403499603271, "step": 10398 }, { "epoch": 0.8197871501773748, "grad_norm": 0.6898779273033142, "learning_rate": 0.0005, "loss": 2.113762855529785, "step": 10399 }, { "epoch": 0.8198659834450138, "grad_norm": 0.7251031398773193, "learning_rate": 0.0005, "loss": 1.9541287422180176, "step": 10400 }, { "epoch": 0.8199448167126527, "grad_norm": 0.5955016016960144, "learning_rate": 0.0005, "loss": 1.976848840713501, "step": 10401 }, { "epoch": 0.8200236499802916, "grad_norm": 0.5435850024223328, "learning_rate": 0.0005, "loss": 1.8329496383666992, "step": 10402 }, { "epoch": 0.8201024832479307, "grad_norm": 0.6052641868591309, "learning_rate": 0.0005, "loss": 1.951744556427002, "step": 10403 }, { "epoch": 0.8201813165155696, "grad_norm": 0.6504499912261963, "learning_rate": 0.0005, "loss": 1.9076963663101196, "step": 10404 }, { "epoch": 0.8202601497832085, "grad_norm": 0.490427166223526, "learning_rate": 0.0005, "loss": 1.8942571878433228, "step": 10405 }, { "epoch": 0.8203389830508474, "grad_norm": 0.7310695052146912, "learning_rate": 0.0005, "loss": 2.1581532955169678, "step": 10406 }, { "epoch": 0.8204178163184864, "grad_norm": 0.9682796001434326, "learning_rate": 0.0005, "loss": 2.068530559539795, "step": 10407 }, { "epoch": 0.8204966495861253, "grad_norm": 0.7772119045257568, "learning_rate": 0.0005, "loss": 2.029223918914795, "step": 10408 }, { "epoch": 0.8205754828537642, "grad_norm": 0.542522668838501, "learning_rate": 0.0005, "loss": 1.7541215419769287, "step": 10409 }, { "epoch": 0.8206543161214033, "grad_norm": 0.6665462255477905, "learning_rate": 0.0005, "loss": 2.1506149768829346, "step": 10410 }, { "epoch": 0.8207331493890422, "grad_norm": 0.920274019241333, "learning_rate": 0.0005, "loss": 2.2939534187316895, "step": 10411 }, { "epoch": 0.8208119826566811, "grad_norm": 0.6646876335144043, "learning_rate": 0.0005, "loss": 1.7297312021255493, "step": 10412 }, { "epoch": 0.8208908159243201, "grad_norm": 0.7066044807434082, "learning_rate": 0.0005, "loss": 1.8166449069976807, "step": 10413 }, { "epoch": 0.820969649191959, "grad_norm": 0.6268971562385559, "learning_rate": 0.0005, "loss": 1.881648063659668, "step": 10414 }, { "epoch": 0.8210484824595979, "grad_norm": 0.5166982412338257, "learning_rate": 0.0005, "loss": 1.8705439567565918, "step": 10415 }, { "epoch": 0.8211273157272369, "grad_norm": 0.6481226682662964, "learning_rate": 0.0005, "loss": 2.0301382541656494, "step": 10416 }, { "epoch": 0.8212061489948759, "grad_norm": 0.8067310452461243, "learning_rate": 0.0005, "loss": 2.1473140716552734, "step": 10417 }, { "epoch": 0.8212849822625148, "grad_norm": 0.5631579756736755, "learning_rate": 0.0005, "loss": 1.678206205368042, "step": 10418 }, { "epoch": 0.8213638155301537, "grad_norm": 0.6492827534675598, "learning_rate": 0.0005, "loss": 2.001770496368408, "step": 10419 }, { "epoch": 0.8214426487977927, "grad_norm": 0.6256643533706665, "learning_rate": 0.0005, "loss": 1.8561575412750244, "step": 10420 }, { "epoch": 0.8215214820654316, "grad_norm": 0.7606669068336487, "learning_rate": 0.0005, "loss": 1.9910968542099, "step": 10421 }, { "epoch": 0.8216003153330705, "grad_norm": 0.7262014746665955, "learning_rate": 0.0005, "loss": 1.8188780546188354, "step": 10422 }, { "epoch": 0.8216791486007095, "grad_norm": 0.7133756279945374, "learning_rate": 0.0005, "loss": 2.016099214553833, "step": 10423 }, { "epoch": 0.8217579818683485, "grad_norm": 0.7315927147865295, "learning_rate": 0.0005, "loss": 1.9167797565460205, "step": 10424 }, { "epoch": 0.8218368151359874, "grad_norm": 0.6227834224700928, "learning_rate": 0.0005, "loss": 1.6926482915878296, "step": 10425 }, { "epoch": 0.8219156484036263, "grad_norm": 0.5699467062950134, "learning_rate": 0.0005, "loss": 1.6851625442504883, "step": 10426 }, { "epoch": 0.8219944816712653, "grad_norm": 0.6351961493492126, "learning_rate": 0.0005, "loss": 2.0875470638275146, "step": 10427 }, { "epoch": 0.8220733149389042, "grad_norm": 0.6407554745674133, "learning_rate": 0.0005, "loss": 1.8455476760864258, "step": 10428 }, { "epoch": 0.8221521482065431, "grad_norm": 1.170484185218811, "learning_rate": 0.0005, "loss": 1.8462693691253662, "step": 10429 }, { "epoch": 0.822230981474182, "grad_norm": 0.6232242584228516, "learning_rate": 0.0005, "loss": 1.8393570184707642, "step": 10430 }, { "epoch": 0.8223098147418211, "grad_norm": 1.1445503234863281, "learning_rate": 0.0005, "loss": 1.848766803741455, "step": 10431 }, { "epoch": 0.82238864800946, "grad_norm": 0.7826279997825623, "learning_rate": 0.0005, "loss": 1.6555602550506592, "step": 10432 }, { "epoch": 0.8224674812770989, "grad_norm": 0.8504858016967773, "learning_rate": 0.0005, "loss": 1.8944329023361206, "step": 10433 }, { "epoch": 0.8225463145447379, "grad_norm": 0.7147566676139832, "learning_rate": 0.0005, "loss": 1.8914381265640259, "step": 10434 }, { "epoch": 0.8226251478123768, "grad_norm": 0.5917763113975525, "learning_rate": 0.0005, "loss": 2.098008155822754, "step": 10435 }, { "epoch": 0.8227039810800157, "grad_norm": 0.6800616383552551, "learning_rate": 0.0005, "loss": 2.0804848670959473, "step": 10436 }, { "epoch": 0.8227828143476547, "grad_norm": 0.6122899651527405, "learning_rate": 0.0005, "loss": 1.6589176654815674, "step": 10437 }, { "epoch": 0.8228616476152937, "grad_norm": 0.573069155216217, "learning_rate": 0.0005, "loss": 1.9944580793380737, "step": 10438 }, { "epoch": 0.8229404808829326, "grad_norm": 0.7133736610412598, "learning_rate": 0.0005, "loss": 2.060291290283203, "step": 10439 }, { "epoch": 0.8230193141505715, "grad_norm": 0.8031381368637085, "learning_rate": 0.0005, "loss": 1.994912028312683, "step": 10440 }, { "epoch": 0.8230981474182105, "grad_norm": 0.6044841408729553, "learning_rate": 0.0005, "loss": 1.8341436386108398, "step": 10441 }, { "epoch": 0.8231769806858494, "grad_norm": 0.743929386138916, "learning_rate": 0.0005, "loss": 1.7383519411087036, "step": 10442 }, { "epoch": 0.8232558139534883, "grad_norm": 0.623686671257019, "learning_rate": 0.0005, "loss": 1.6307461261749268, "step": 10443 }, { "epoch": 0.8233346472211274, "grad_norm": 0.6947680711746216, "learning_rate": 0.0005, "loss": 1.8780746459960938, "step": 10444 }, { "epoch": 0.8234134804887663, "grad_norm": 0.6660517454147339, "learning_rate": 0.0005, "loss": 1.920000433921814, "step": 10445 }, { "epoch": 0.8234923137564052, "grad_norm": 0.7518771886825562, "learning_rate": 0.0005, "loss": 1.89395010471344, "step": 10446 }, { "epoch": 0.8235711470240441, "grad_norm": 0.6158276796340942, "learning_rate": 0.0005, "loss": 1.8209580183029175, "step": 10447 }, { "epoch": 0.8236499802916831, "grad_norm": 0.5651091933250427, "learning_rate": 0.0005, "loss": 1.8011009693145752, "step": 10448 }, { "epoch": 0.823728813559322, "grad_norm": 0.8538320660591125, "learning_rate": 0.0005, "loss": 1.853305697441101, "step": 10449 }, { "epoch": 0.823807646826961, "grad_norm": 0.7151476144790649, "learning_rate": 0.0005, "loss": 1.7053923606872559, "step": 10450 }, { "epoch": 0.8238864800946, "grad_norm": 0.7197484374046326, "learning_rate": 0.0005, "loss": 1.884638786315918, "step": 10451 }, { "epoch": 0.8239653133622389, "grad_norm": 0.8336031436920166, "learning_rate": 0.0005, "loss": 2.0443520545959473, "step": 10452 }, { "epoch": 0.8240441466298778, "grad_norm": 0.6771779656410217, "learning_rate": 0.0005, "loss": 1.8449010848999023, "step": 10453 }, { "epoch": 0.8241229798975167, "grad_norm": 0.5390148758888245, "learning_rate": 0.0005, "loss": 1.720711350440979, "step": 10454 }, { "epoch": 0.8242018131651557, "grad_norm": 0.6547183990478516, "learning_rate": 0.0005, "loss": 1.8684313297271729, "step": 10455 }, { "epoch": 0.8242806464327946, "grad_norm": 0.6330966949462891, "learning_rate": 0.0005, "loss": 1.817164659500122, "step": 10456 }, { "epoch": 0.8243594797004336, "grad_norm": 0.727542519569397, "learning_rate": 0.0005, "loss": 1.8093465566635132, "step": 10457 }, { "epoch": 0.8244383129680726, "grad_norm": 0.5793325304985046, "learning_rate": 0.0005, "loss": 1.999369502067566, "step": 10458 }, { "epoch": 0.8245171462357115, "grad_norm": 0.6323398947715759, "learning_rate": 0.0005, "loss": 1.6544551849365234, "step": 10459 }, { "epoch": 0.8245959795033504, "grad_norm": 0.5425798296928406, "learning_rate": 0.0005, "loss": 1.760735034942627, "step": 10460 }, { "epoch": 0.8246748127709893, "grad_norm": 0.5446764230728149, "learning_rate": 0.0005, "loss": 1.7316067218780518, "step": 10461 }, { "epoch": 0.8247536460386283, "grad_norm": 0.6725013256072998, "learning_rate": 0.0005, "loss": 1.8420332670211792, "step": 10462 }, { "epoch": 0.8248324793062672, "grad_norm": 0.6197083592414856, "learning_rate": 0.0005, "loss": 1.7606289386749268, "step": 10463 }, { "epoch": 0.8249113125739062, "grad_norm": 0.9138935804367065, "learning_rate": 0.0005, "loss": 1.9311645030975342, "step": 10464 }, { "epoch": 0.8249901458415452, "grad_norm": 0.8418394327163696, "learning_rate": 0.0005, "loss": 1.9325897693634033, "step": 10465 }, { "epoch": 0.8250689791091841, "grad_norm": 1.168365478515625, "learning_rate": 0.0005, "loss": 1.9002795219421387, "step": 10466 }, { "epoch": 0.825147812376823, "grad_norm": 0.8153954148292542, "learning_rate": 0.0005, "loss": 1.9773409366607666, "step": 10467 }, { "epoch": 0.8252266456444619, "grad_norm": 0.9903894066810608, "learning_rate": 0.0005, "loss": 1.988349199295044, "step": 10468 }, { "epoch": 0.8253054789121009, "grad_norm": 0.6974439024925232, "learning_rate": 0.0005, "loss": 1.897043228149414, "step": 10469 }, { "epoch": 0.8253843121797398, "grad_norm": 0.6796897649765015, "learning_rate": 0.0005, "loss": 2.1615869998931885, "step": 10470 }, { "epoch": 0.8254631454473788, "grad_norm": 0.7760595679283142, "learning_rate": 0.0005, "loss": 1.8143534660339355, "step": 10471 }, { "epoch": 0.8255419787150178, "grad_norm": 1.6457303762435913, "learning_rate": 0.0005, "loss": 1.707255482673645, "step": 10472 }, { "epoch": 0.8256208119826567, "grad_norm": 0.7323176264762878, "learning_rate": 0.0005, "loss": 1.9343645572662354, "step": 10473 }, { "epoch": 0.8256996452502956, "grad_norm": 0.7020061016082764, "learning_rate": 0.0005, "loss": 1.7025161981582642, "step": 10474 }, { "epoch": 0.8257784785179346, "grad_norm": 0.6525555849075317, "learning_rate": 0.0005, "loss": 2.0243260860443115, "step": 10475 }, { "epoch": 0.8258573117855735, "grad_norm": 0.7897815108299255, "learning_rate": 0.0005, "loss": 1.9295339584350586, "step": 10476 }, { "epoch": 0.8259361450532124, "grad_norm": 0.6797564625740051, "learning_rate": 0.0005, "loss": 1.799241065979004, "step": 10477 }, { "epoch": 0.8260149783208514, "grad_norm": 0.6203755140304565, "learning_rate": 0.0005, "loss": 1.827072262763977, "step": 10478 }, { "epoch": 0.8260938115884904, "grad_norm": 0.4925023317337036, "learning_rate": 0.0005, "loss": 1.7122904062271118, "step": 10479 }, { "epoch": 0.8261726448561293, "grad_norm": 0.6679466962814331, "learning_rate": 0.0005, "loss": 1.8282945156097412, "step": 10480 }, { "epoch": 0.8262514781237682, "grad_norm": 0.6237309575080872, "learning_rate": 0.0005, "loss": 1.8042538166046143, "step": 10481 }, { "epoch": 0.8263303113914072, "grad_norm": 0.5660620331764221, "learning_rate": 0.0005, "loss": 1.7326443195343018, "step": 10482 }, { "epoch": 0.8264091446590461, "grad_norm": 0.6977839469909668, "learning_rate": 0.0005, "loss": 2.006521701812744, "step": 10483 }, { "epoch": 0.826487977926685, "grad_norm": 0.8344002366065979, "learning_rate": 0.0005, "loss": 2.07907772064209, "step": 10484 }, { "epoch": 0.826566811194324, "grad_norm": 0.6497340202331543, "learning_rate": 0.0005, "loss": 1.7769556045532227, "step": 10485 }, { "epoch": 0.826645644461963, "grad_norm": 0.7346502542495728, "learning_rate": 0.0005, "loss": 1.7150837182998657, "step": 10486 }, { "epoch": 0.8267244777296019, "grad_norm": 0.718494176864624, "learning_rate": 0.0005, "loss": 1.7598354816436768, "step": 10487 }, { "epoch": 0.8268033109972408, "grad_norm": 0.5807303190231323, "learning_rate": 0.0005, "loss": 1.866091251373291, "step": 10488 }, { "epoch": 0.8268821442648798, "grad_norm": 0.6334530115127563, "learning_rate": 0.0005, "loss": 2.1069531440734863, "step": 10489 }, { "epoch": 0.8269609775325187, "grad_norm": 0.5442454218864441, "learning_rate": 0.0005, "loss": 1.7476392984390259, "step": 10490 }, { "epoch": 0.8270398108001576, "grad_norm": 0.6799169182777405, "learning_rate": 0.0005, "loss": 1.5718719959259033, "step": 10491 }, { "epoch": 0.8271186440677966, "grad_norm": 0.5972808003425598, "learning_rate": 0.0005, "loss": 1.8705470561981201, "step": 10492 }, { "epoch": 0.8271974773354356, "grad_norm": 0.5648763179779053, "learning_rate": 0.0005, "loss": 1.6946022510528564, "step": 10493 }, { "epoch": 0.8272763106030745, "grad_norm": 0.5927114486694336, "learning_rate": 0.0005, "loss": 1.5810223817825317, "step": 10494 }, { "epoch": 0.8273551438707134, "grad_norm": 0.7702100276947021, "learning_rate": 0.0005, "loss": 1.5612382888793945, "step": 10495 }, { "epoch": 0.8274339771383524, "grad_norm": 0.6499118208885193, "learning_rate": 0.0005, "loss": 2.0448906421661377, "step": 10496 }, { "epoch": 0.8275128104059913, "grad_norm": 0.6007353067398071, "learning_rate": 0.0005, "loss": 1.8461178541183472, "step": 10497 }, { "epoch": 0.8275916436736303, "grad_norm": 0.5436953902244568, "learning_rate": 0.0005, "loss": 1.5593973398208618, "step": 10498 }, { "epoch": 0.8276704769412692, "grad_norm": 0.7733659744262695, "learning_rate": 0.0005, "loss": 2.0434865951538086, "step": 10499 }, { "epoch": 0.8277493102089082, "grad_norm": 0.6331778764724731, "learning_rate": 0.0005, "loss": 1.8832478523254395, "step": 10500 }, { "epoch": 0.8278281434765471, "grad_norm": 0.6487700343132019, "learning_rate": 0.0005, "loss": 1.90689218044281, "step": 10501 }, { "epoch": 0.827906976744186, "grad_norm": 0.5642850399017334, "learning_rate": 0.0005, "loss": 1.9112907648086548, "step": 10502 }, { "epoch": 0.827985810011825, "grad_norm": 0.5650310516357422, "learning_rate": 0.0005, "loss": 1.95862877368927, "step": 10503 }, { "epoch": 0.8280646432794639, "grad_norm": 0.5529904961585999, "learning_rate": 0.0005, "loss": 1.5379456281661987, "step": 10504 }, { "epoch": 0.8281434765471029, "grad_norm": 0.7148826122283936, "learning_rate": 0.0005, "loss": 1.9241451025009155, "step": 10505 }, { "epoch": 0.8282223098147419, "grad_norm": 0.8234978914260864, "learning_rate": 0.0005, "loss": 1.9559561014175415, "step": 10506 }, { "epoch": 0.8283011430823808, "grad_norm": 0.8184162378311157, "learning_rate": 0.0005, "loss": 1.8542362451553345, "step": 10507 }, { "epoch": 0.8283799763500197, "grad_norm": 0.6099265217781067, "learning_rate": 0.0005, "loss": 2.1250500679016113, "step": 10508 }, { "epoch": 0.8284588096176586, "grad_norm": 0.48291394114494324, "learning_rate": 0.0005, "loss": 1.7221826314926147, "step": 10509 }, { "epoch": 0.8285376428852976, "grad_norm": 0.7307543158531189, "learning_rate": 0.0005, "loss": 1.8554733991622925, "step": 10510 }, { "epoch": 0.8286164761529365, "grad_norm": 0.4772869348526001, "learning_rate": 0.0005, "loss": 1.5574002265930176, "step": 10511 }, { "epoch": 0.8286953094205755, "grad_norm": 0.562661349773407, "learning_rate": 0.0005, "loss": 1.9012670516967773, "step": 10512 }, { "epoch": 0.8287741426882145, "grad_norm": 0.5119149088859558, "learning_rate": 0.0005, "loss": 1.402078628540039, "step": 10513 }, { "epoch": 0.8288529759558534, "grad_norm": 0.5722540616989136, "learning_rate": 0.0005, "loss": 1.7483456134796143, "step": 10514 }, { "epoch": 0.8289318092234923, "grad_norm": 0.7875526547431946, "learning_rate": 0.0005, "loss": 1.9143654108047485, "step": 10515 }, { "epoch": 0.8290106424911312, "grad_norm": 0.6086776852607727, "learning_rate": 0.0005, "loss": 1.829224944114685, "step": 10516 }, { "epoch": 0.8290894757587702, "grad_norm": 0.5517969727516174, "learning_rate": 0.0005, "loss": 1.8781602382659912, "step": 10517 }, { "epoch": 0.8291683090264091, "grad_norm": 0.5625207424163818, "learning_rate": 0.0005, "loss": 1.6635377407073975, "step": 10518 }, { "epoch": 0.8292471422940481, "grad_norm": 0.6271605491638184, "learning_rate": 0.0005, "loss": 2.1219873428344727, "step": 10519 }, { "epoch": 0.8293259755616871, "grad_norm": 0.5888684391975403, "learning_rate": 0.0005, "loss": 1.899185299873352, "step": 10520 }, { "epoch": 0.829404808829326, "grad_norm": 0.7016578912734985, "learning_rate": 0.0005, "loss": 1.9277880191802979, "step": 10521 }, { "epoch": 0.8294836420969649, "grad_norm": 0.7136979103088379, "learning_rate": 0.0005, "loss": 1.738002896308899, "step": 10522 }, { "epoch": 0.8295624753646038, "grad_norm": 0.6935098171234131, "learning_rate": 0.0005, "loss": 1.8582097291946411, "step": 10523 }, { "epoch": 0.8296413086322428, "grad_norm": 0.7006714344024658, "learning_rate": 0.0005, "loss": 2.0471324920654297, "step": 10524 }, { "epoch": 0.8297201418998817, "grad_norm": 0.5937922596931458, "learning_rate": 0.0005, "loss": 1.6903905868530273, "step": 10525 }, { "epoch": 0.8297989751675207, "grad_norm": 0.619698703289032, "learning_rate": 0.0005, "loss": 1.7271943092346191, "step": 10526 }, { "epoch": 0.8298778084351597, "grad_norm": 0.6041438579559326, "learning_rate": 0.0005, "loss": 1.6087088584899902, "step": 10527 }, { "epoch": 0.8299566417027986, "grad_norm": 0.6463608145713806, "learning_rate": 0.0005, "loss": 1.8367081880569458, "step": 10528 }, { "epoch": 0.8300354749704375, "grad_norm": 0.6071917414665222, "learning_rate": 0.0005, "loss": 1.8465584516525269, "step": 10529 }, { "epoch": 0.8301143082380764, "grad_norm": 0.6843395233154297, "learning_rate": 0.0005, "loss": 1.9857028722763062, "step": 10530 }, { "epoch": 0.8301931415057154, "grad_norm": 0.7794920206069946, "learning_rate": 0.0005, "loss": 1.900130271911621, "step": 10531 }, { "epoch": 0.8302719747733543, "grad_norm": 0.5822708606719971, "learning_rate": 0.0005, "loss": 1.9322261810302734, "step": 10532 }, { "epoch": 0.8303508080409933, "grad_norm": 0.6237813234329224, "learning_rate": 0.0005, "loss": 2.007719039916992, "step": 10533 }, { "epoch": 0.8304296413086323, "grad_norm": 0.6669094562530518, "learning_rate": 0.0005, "loss": 1.7865251302719116, "step": 10534 }, { "epoch": 0.8305084745762712, "grad_norm": 0.7101410031318665, "learning_rate": 0.0005, "loss": 2.050304412841797, "step": 10535 }, { "epoch": 0.8305873078439101, "grad_norm": 0.5938491821289062, "learning_rate": 0.0005, "loss": 1.9575526714324951, "step": 10536 }, { "epoch": 0.8306661411115491, "grad_norm": 0.5918595790863037, "learning_rate": 0.0005, "loss": 1.85591459274292, "step": 10537 }, { "epoch": 0.830744974379188, "grad_norm": 1.1347821950912476, "learning_rate": 0.0005, "loss": 1.9826451539993286, "step": 10538 }, { "epoch": 0.830823807646827, "grad_norm": 0.698243260383606, "learning_rate": 0.0005, "loss": 1.8151352405548096, "step": 10539 }, { "epoch": 0.8309026409144659, "grad_norm": 0.7563011050224304, "learning_rate": 0.0005, "loss": 1.7588176727294922, "step": 10540 }, { "epoch": 0.8309814741821049, "grad_norm": 0.9091121554374695, "learning_rate": 0.0005, "loss": 2.314042091369629, "step": 10541 }, { "epoch": 0.8310603074497438, "grad_norm": 0.5182684659957886, "learning_rate": 0.0005, "loss": 1.5927152633666992, "step": 10542 }, { "epoch": 0.8311391407173827, "grad_norm": 0.5609098076820374, "learning_rate": 0.0005, "loss": 1.7069075107574463, "step": 10543 }, { "epoch": 0.8312179739850217, "grad_norm": 0.5642565488815308, "learning_rate": 0.0005, "loss": 1.4850518703460693, "step": 10544 }, { "epoch": 0.8312968072526606, "grad_norm": 0.7680564522743225, "learning_rate": 0.0005, "loss": 2.2485601902008057, "step": 10545 }, { "epoch": 0.8313756405202996, "grad_norm": 0.637474775314331, "learning_rate": 0.0005, "loss": 1.670344352722168, "step": 10546 }, { "epoch": 0.8314544737879385, "grad_norm": 1.0996781587600708, "learning_rate": 0.0005, "loss": 1.9274959564208984, "step": 10547 }, { "epoch": 0.8315333070555775, "grad_norm": 0.704212486743927, "learning_rate": 0.0005, "loss": 2.0986814498901367, "step": 10548 }, { "epoch": 0.8316121403232164, "grad_norm": 0.6783899664878845, "learning_rate": 0.0005, "loss": 2.005267381668091, "step": 10549 }, { "epoch": 0.8316909735908553, "grad_norm": 0.5500155091285706, "learning_rate": 0.0005, "loss": 1.7391059398651123, "step": 10550 }, { "epoch": 0.8317698068584943, "grad_norm": 0.6281704902648926, "learning_rate": 0.0005, "loss": 1.826127529144287, "step": 10551 }, { "epoch": 0.8318486401261332, "grad_norm": 0.728752613067627, "learning_rate": 0.0005, "loss": 1.9208004474639893, "step": 10552 }, { "epoch": 0.8319274733937722, "grad_norm": 0.5926421880722046, "learning_rate": 0.0005, "loss": 2.018882989883423, "step": 10553 }, { "epoch": 0.8320063066614111, "grad_norm": 0.8977496027946472, "learning_rate": 0.0005, "loss": 1.813583493232727, "step": 10554 }, { "epoch": 0.8320851399290501, "grad_norm": 0.6346114277839661, "learning_rate": 0.0005, "loss": 1.926708459854126, "step": 10555 }, { "epoch": 0.832163973196689, "grad_norm": 0.7075062394142151, "learning_rate": 0.0005, "loss": 1.8555192947387695, "step": 10556 }, { "epoch": 0.8322428064643279, "grad_norm": 0.6445000171661377, "learning_rate": 0.0005, "loss": 1.73188054561615, "step": 10557 }, { "epoch": 0.8323216397319669, "grad_norm": 0.6856288313865662, "learning_rate": 0.0005, "loss": 1.7292101383209229, "step": 10558 }, { "epoch": 0.8324004729996058, "grad_norm": 0.7957462668418884, "learning_rate": 0.0005, "loss": 1.776795744895935, "step": 10559 }, { "epoch": 0.8324793062672448, "grad_norm": 0.7247533798217773, "learning_rate": 0.0005, "loss": 1.7124065160751343, "step": 10560 }, { "epoch": 0.8325581395348837, "grad_norm": 0.8376884460449219, "learning_rate": 0.0005, "loss": 2.265965700149536, "step": 10561 }, { "epoch": 0.8326369728025227, "grad_norm": 0.5912513136863708, "learning_rate": 0.0005, "loss": 1.5599169731140137, "step": 10562 }, { "epoch": 0.8327158060701616, "grad_norm": 0.6367140412330627, "learning_rate": 0.0005, "loss": 1.826533555984497, "step": 10563 }, { "epoch": 0.8327946393378005, "grad_norm": 0.81217360496521, "learning_rate": 0.0005, "loss": 1.4929485321044922, "step": 10564 }, { "epoch": 0.8328734726054395, "grad_norm": 0.690794050693512, "learning_rate": 0.0005, "loss": 1.74399995803833, "step": 10565 }, { "epoch": 0.8329523058730784, "grad_norm": 0.6844632625579834, "learning_rate": 0.0005, "loss": 1.9071131944656372, "step": 10566 }, { "epoch": 0.8330311391407174, "grad_norm": 0.6818488240242004, "learning_rate": 0.0005, "loss": 2.0583696365356445, "step": 10567 }, { "epoch": 0.8331099724083564, "grad_norm": 0.6448789834976196, "learning_rate": 0.0005, "loss": 1.8870419263839722, "step": 10568 }, { "epoch": 0.8331888056759953, "grad_norm": 0.6366636157035828, "learning_rate": 0.0005, "loss": 2.010594367980957, "step": 10569 }, { "epoch": 0.8332676389436342, "grad_norm": 0.6617685556411743, "learning_rate": 0.0005, "loss": 1.993743896484375, "step": 10570 }, { "epoch": 0.8333464722112731, "grad_norm": 0.49398696422576904, "learning_rate": 0.0005, "loss": 1.5656189918518066, "step": 10571 }, { "epoch": 0.8334253054789121, "grad_norm": 0.6687254309654236, "learning_rate": 0.0005, "loss": 1.922275424003601, "step": 10572 }, { "epoch": 0.833504138746551, "grad_norm": 0.6228387951850891, "learning_rate": 0.0005, "loss": 1.7813560962677002, "step": 10573 }, { "epoch": 0.83358297201419, "grad_norm": 0.5559341907501221, "learning_rate": 0.0005, "loss": 1.8017535209655762, "step": 10574 }, { "epoch": 0.833661805281829, "grad_norm": 0.6393284797668457, "learning_rate": 0.0005, "loss": 2.0479671955108643, "step": 10575 }, { "epoch": 0.8337406385494679, "grad_norm": 0.5350298285484314, "learning_rate": 0.0005, "loss": 1.9270503520965576, "step": 10576 }, { "epoch": 0.8338194718171068, "grad_norm": 0.8863551020622253, "learning_rate": 0.0005, "loss": 2.091446876525879, "step": 10577 }, { "epoch": 0.8338983050847457, "grad_norm": 0.5633636713027954, "learning_rate": 0.0005, "loss": 2.0208098888397217, "step": 10578 }, { "epoch": 0.8339771383523847, "grad_norm": 0.7498865723609924, "learning_rate": 0.0005, "loss": 2.026198387145996, "step": 10579 }, { "epoch": 0.8340559716200237, "grad_norm": 1.043056845664978, "learning_rate": 0.0005, "loss": 2.1363205909729004, "step": 10580 }, { "epoch": 0.8341348048876626, "grad_norm": 0.6442359685897827, "learning_rate": 0.0005, "loss": 1.7416489124298096, "step": 10581 }, { "epoch": 0.8342136381553016, "grad_norm": 0.6604309678077698, "learning_rate": 0.0005, "loss": 1.7012250423431396, "step": 10582 }, { "epoch": 0.8342924714229405, "grad_norm": 0.836093008518219, "learning_rate": 0.0005, "loss": 2.037041664123535, "step": 10583 }, { "epoch": 0.8343713046905794, "grad_norm": 0.6492248773574829, "learning_rate": 0.0005, "loss": 1.7750054597854614, "step": 10584 }, { "epoch": 0.8344501379582183, "grad_norm": 0.595893383026123, "learning_rate": 0.0005, "loss": 1.5440526008605957, "step": 10585 }, { "epoch": 0.8345289712258573, "grad_norm": 0.6705156564712524, "learning_rate": 0.0005, "loss": 1.9912033081054688, "step": 10586 }, { "epoch": 0.8346078044934963, "grad_norm": 0.7657786011695862, "learning_rate": 0.0005, "loss": 1.791301965713501, "step": 10587 }, { "epoch": 0.8346866377611352, "grad_norm": 0.6489401459693909, "learning_rate": 0.0005, "loss": 1.8823466300964355, "step": 10588 }, { "epoch": 0.8347654710287742, "grad_norm": 0.5935500860214233, "learning_rate": 0.0005, "loss": 2.0246033668518066, "step": 10589 }, { "epoch": 0.8348443042964131, "grad_norm": 0.5817825794219971, "learning_rate": 0.0005, "loss": 1.768031120300293, "step": 10590 }, { "epoch": 0.834923137564052, "grad_norm": 0.5957558751106262, "learning_rate": 0.0005, "loss": 1.854487419128418, "step": 10591 }, { "epoch": 0.8350019708316909, "grad_norm": 0.64564049243927, "learning_rate": 0.0005, "loss": 1.9771441221237183, "step": 10592 }, { "epoch": 0.8350808040993299, "grad_norm": 0.6602985262870789, "learning_rate": 0.0005, "loss": 2.011542558670044, "step": 10593 }, { "epoch": 0.8351596373669689, "grad_norm": 0.6902225613594055, "learning_rate": 0.0005, "loss": 1.546213150024414, "step": 10594 }, { "epoch": 0.8352384706346078, "grad_norm": 0.6001030206680298, "learning_rate": 0.0005, "loss": 1.93650484085083, "step": 10595 }, { "epoch": 0.8353173039022468, "grad_norm": 0.5282344818115234, "learning_rate": 0.0005, "loss": 1.804025650024414, "step": 10596 }, { "epoch": 0.8353961371698857, "grad_norm": 0.6054744124412537, "learning_rate": 0.0005, "loss": 1.8308249711990356, "step": 10597 }, { "epoch": 0.8354749704375246, "grad_norm": 0.7171506285667419, "learning_rate": 0.0005, "loss": 2.1027722358703613, "step": 10598 }, { "epoch": 0.8355538037051636, "grad_norm": 0.6260239481925964, "learning_rate": 0.0005, "loss": 2.036578416824341, "step": 10599 }, { "epoch": 0.8356326369728025, "grad_norm": 0.5673593282699585, "learning_rate": 0.0005, "loss": 1.7807393074035645, "step": 10600 }, { "epoch": 0.8357114702404415, "grad_norm": 0.5765479207038879, "learning_rate": 0.0005, "loss": 1.8634817600250244, "step": 10601 }, { "epoch": 0.8357903035080804, "grad_norm": 0.683987557888031, "learning_rate": 0.0005, "loss": 1.959392786026001, "step": 10602 }, { "epoch": 0.8358691367757194, "grad_norm": 0.5593666434288025, "learning_rate": 0.0005, "loss": 1.530369758605957, "step": 10603 }, { "epoch": 0.8359479700433583, "grad_norm": 0.5489460229873657, "learning_rate": 0.0005, "loss": 1.8100169897079468, "step": 10604 }, { "epoch": 0.8360268033109972, "grad_norm": 0.6664132475852966, "learning_rate": 0.0005, "loss": 1.9266690015792847, "step": 10605 }, { "epoch": 0.8361056365786362, "grad_norm": 0.5706784725189209, "learning_rate": 0.0005, "loss": 1.8876876831054688, "step": 10606 }, { "epoch": 0.8361844698462751, "grad_norm": 0.5712887048721313, "learning_rate": 0.0005, "loss": 1.5754663944244385, "step": 10607 }, { "epoch": 0.8362633031139141, "grad_norm": 0.6276536583900452, "learning_rate": 0.0005, "loss": 1.584047794342041, "step": 10608 }, { "epoch": 0.836342136381553, "grad_norm": 0.705479621887207, "learning_rate": 0.0005, "loss": 1.9491207599639893, "step": 10609 }, { "epoch": 0.836420969649192, "grad_norm": 0.6222658157348633, "learning_rate": 0.0005, "loss": 2.0248680114746094, "step": 10610 }, { "epoch": 0.8364998029168309, "grad_norm": 0.8248553276062012, "learning_rate": 0.0005, "loss": 2.054399013519287, "step": 10611 }, { "epoch": 0.8365786361844698, "grad_norm": 0.7814981937408447, "learning_rate": 0.0005, "loss": 1.8822689056396484, "step": 10612 }, { "epoch": 0.8366574694521088, "grad_norm": 0.5970431566238403, "learning_rate": 0.0005, "loss": 1.8689181804656982, "step": 10613 }, { "epoch": 0.8367363027197477, "grad_norm": 0.6134517192840576, "learning_rate": 0.0005, "loss": 1.9462209939956665, "step": 10614 }, { "epoch": 0.8368151359873867, "grad_norm": 0.6932109594345093, "learning_rate": 0.0005, "loss": 1.8073152303695679, "step": 10615 }, { "epoch": 0.8368939692550256, "grad_norm": 1.2042860984802246, "learning_rate": 0.0005, "loss": 1.9631493091583252, "step": 10616 }, { "epoch": 0.8369728025226646, "grad_norm": 0.791350781917572, "learning_rate": 0.0005, "loss": 1.7295821905136108, "step": 10617 }, { "epoch": 0.8370516357903035, "grad_norm": 0.6347309947013855, "learning_rate": 0.0005, "loss": 1.7141196727752686, "step": 10618 }, { "epoch": 0.8371304690579424, "grad_norm": 0.7040022015571594, "learning_rate": 0.0005, "loss": 2.030752658843994, "step": 10619 }, { "epoch": 0.8372093023255814, "grad_norm": 0.6586161255836487, "learning_rate": 0.0005, "loss": 1.918851375579834, "step": 10620 }, { "epoch": 0.8372881355932204, "grad_norm": 0.6877442598342896, "learning_rate": 0.0005, "loss": 1.9004181623458862, "step": 10621 }, { "epoch": 0.8373669688608593, "grad_norm": 0.758690357208252, "learning_rate": 0.0005, "loss": 1.7796804904937744, "step": 10622 }, { "epoch": 0.8374458021284982, "grad_norm": 0.6304671168327332, "learning_rate": 0.0005, "loss": 2.0641841888427734, "step": 10623 }, { "epoch": 0.8375246353961372, "grad_norm": 0.568165123462677, "learning_rate": 0.0005, "loss": 1.8930414915084839, "step": 10624 }, { "epoch": 0.8376034686637761, "grad_norm": 0.8061226010322571, "learning_rate": 0.0005, "loss": 1.9542350769042969, "step": 10625 }, { "epoch": 0.837682301931415, "grad_norm": 0.5979214310646057, "learning_rate": 0.0005, "loss": 1.5363565683364868, "step": 10626 }, { "epoch": 0.837761135199054, "grad_norm": 0.5979094505310059, "learning_rate": 0.0005, "loss": 1.615400791168213, "step": 10627 }, { "epoch": 0.837839968466693, "grad_norm": 0.6612571477890015, "learning_rate": 0.0005, "loss": 1.7296347618103027, "step": 10628 }, { "epoch": 0.8379188017343319, "grad_norm": 0.9596099853515625, "learning_rate": 0.0005, "loss": 1.8472856283187866, "step": 10629 }, { "epoch": 0.8379976350019709, "grad_norm": 0.7616423964500427, "learning_rate": 0.0005, "loss": 1.7336900234222412, "step": 10630 }, { "epoch": 0.8380764682696098, "grad_norm": 1.1890971660614014, "learning_rate": 0.0005, "loss": 1.87497079372406, "step": 10631 }, { "epoch": 0.8381553015372487, "grad_norm": 0.7393597960472107, "learning_rate": 0.0005, "loss": 2.089663028717041, "step": 10632 }, { "epoch": 0.8382341348048876, "grad_norm": 0.6128467917442322, "learning_rate": 0.0005, "loss": 1.9532606601715088, "step": 10633 }, { "epoch": 0.8383129680725266, "grad_norm": 0.6097343564033508, "learning_rate": 0.0005, "loss": 1.653450608253479, "step": 10634 }, { "epoch": 0.8383918013401656, "grad_norm": 1.0475460290908813, "learning_rate": 0.0005, "loss": 1.8192930221557617, "step": 10635 }, { "epoch": 0.8384706346078045, "grad_norm": 0.7153878808021545, "learning_rate": 0.0005, "loss": 2.407784938812256, "step": 10636 }, { "epoch": 0.8385494678754435, "grad_norm": 0.735256552696228, "learning_rate": 0.0005, "loss": 2.235023021697998, "step": 10637 }, { "epoch": 0.8386283011430824, "grad_norm": 0.546592116355896, "learning_rate": 0.0005, "loss": 1.6519505977630615, "step": 10638 }, { "epoch": 0.8387071344107213, "grad_norm": 0.7493112087249756, "learning_rate": 0.0005, "loss": 2.0574541091918945, "step": 10639 }, { "epoch": 0.8387859676783602, "grad_norm": 0.6904661059379578, "learning_rate": 0.0005, "loss": 1.8856723308563232, "step": 10640 }, { "epoch": 0.8388648009459992, "grad_norm": 0.7462922930717468, "learning_rate": 0.0005, "loss": 1.7940402030944824, "step": 10641 }, { "epoch": 0.8389436342136382, "grad_norm": 0.6069934964179993, "learning_rate": 0.0005, "loss": 1.783630132675171, "step": 10642 }, { "epoch": 0.8390224674812771, "grad_norm": 0.7699776291847229, "learning_rate": 0.0005, "loss": 1.8893948793411255, "step": 10643 }, { "epoch": 0.8391013007489161, "grad_norm": 0.7701184749603271, "learning_rate": 0.0005, "loss": 2.205474376678467, "step": 10644 }, { "epoch": 0.839180134016555, "grad_norm": 0.6536039113998413, "learning_rate": 0.0005, "loss": 1.9915980100631714, "step": 10645 }, { "epoch": 0.8392589672841939, "grad_norm": 0.6358144879341125, "learning_rate": 0.0005, "loss": 1.8911190032958984, "step": 10646 }, { "epoch": 0.8393378005518328, "grad_norm": 0.6717963218688965, "learning_rate": 0.0005, "loss": 1.7135436534881592, "step": 10647 }, { "epoch": 0.8394166338194718, "grad_norm": 1.3622050285339355, "learning_rate": 0.0005, "loss": 1.9711241722106934, "step": 10648 }, { "epoch": 0.8394954670871108, "grad_norm": 0.7037469744682312, "learning_rate": 0.0005, "loss": 1.8324568271636963, "step": 10649 }, { "epoch": 0.8395743003547497, "grad_norm": 0.5507345795631409, "learning_rate": 0.0005, "loss": 1.6894161701202393, "step": 10650 }, { "epoch": 0.8396531336223887, "grad_norm": 0.5848253965377808, "learning_rate": 0.0005, "loss": 1.8227046728134155, "step": 10651 }, { "epoch": 0.8397319668900276, "grad_norm": 0.6470548510551453, "learning_rate": 0.0005, "loss": 2.0726993083953857, "step": 10652 }, { "epoch": 0.8398108001576665, "grad_norm": 0.6719205379486084, "learning_rate": 0.0005, "loss": 1.8764011859893799, "step": 10653 }, { "epoch": 0.8398896334253054, "grad_norm": 0.7582356929779053, "learning_rate": 0.0005, "loss": 1.913275957107544, "step": 10654 }, { "epoch": 0.8399684666929444, "grad_norm": 0.6029213070869446, "learning_rate": 0.0005, "loss": 1.933470368385315, "step": 10655 }, { "epoch": 0.8400472999605834, "grad_norm": 0.8941866159439087, "learning_rate": 0.0005, "loss": 1.6678001880645752, "step": 10656 }, { "epoch": 0.8401261332282223, "grad_norm": 0.6979048848152161, "learning_rate": 0.0005, "loss": 1.997736930847168, "step": 10657 }, { "epoch": 0.8402049664958613, "grad_norm": 0.6911380290985107, "learning_rate": 0.0005, "loss": 2.0887908935546875, "step": 10658 }, { "epoch": 0.8402837997635002, "grad_norm": 0.6805702447891235, "learning_rate": 0.0005, "loss": 1.8044414520263672, "step": 10659 }, { "epoch": 0.8403626330311391, "grad_norm": 0.6037561297416687, "learning_rate": 0.0005, "loss": 1.763953447341919, "step": 10660 }, { "epoch": 0.8404414662987781, "grad_norm": 0.5597310662269592, "learning_rate": 0.0005, "loss": 1.7323708534240723, "step": 10661 }, { "epoch": 0.840520299566417, "grad_norm": 0.9299106597900391, "learning_rate": 0.0005, "loss": 2.3939695358276367, "step": 10662 }, { "epoch": 0.840599132834056, "grad_norm": 0.729485809803009, "learning_rate": 0.0005, "loss": 1.9624450206756592, "step": 10663 }, { "epoch": 0.8406779661016949, "grad_norm": 0.8451427817344666, "learning_rate": 0.0005, "loss": 2.026632308959961, "step": 10664 }, { "epoch": 0.8407567993693339, "grad_norm": 0.5456540584564209, "learning_rate": 0.0005, "loss": 1.8136627674102783, "step": 10665 }, { "epoch": 0.8408356326369728, "grad_norm": 0.6503391861915588, "learning_rate": 0.0005, "loss": 1.8239617347717285, "step": 10666 }, { "epoch": 0.8409144659046117, "grad_norm": 0.770385205745697, "learning_rate": 0.0005, "loss": 1.7550084590911865, "step": 10667 }, { "epoch": 0.8409932991722507, "grad_norm": 0.7692108154296875, "learning_rate": 0.0005, "loss": 1.8655842542648315, "step": 10668 }, { "epoch": 0.8410721324398897, "grad_norm": 0.5836448073387146, "learning_rate": 0.0005, "loss": 1.8786284923553467, "step": 10669 }, { "epoch": 0.8411509657075286, "grad_norm": 0.5449267029762268, "learning_rate": 0.0005, "loss": 1.7225332260131836, "step": 10670 }, { "epoch": 0.8412297989751675, "grad_norm": 0.7112568020820618, "learning_rate": 0.0005, "loss": 1.9682066440582275, "step": 10671 }, { "epoch": 0.8413086322428065, "grad_norm": 0.5125686526298523, "learning_rate": 0.0005, "loss": 1.8610928058624268, "step": 10672 }, { "epoch": 0.8413874655104454, "grad_norm": 0.7204345464706421, "learning_rate": 0.0005, "loss": 1.7396427392959595, "step": 10673 }, { "epoch": 0.8414662987780843, "grad_norm": 0.6276339888572693, "learning_rate": 0.0005, "loss": 1.6337831020355225, "step": 10674 }, { "epoch": 0.8415451320457233, "grad_norm": 0.73542720079422, "learning_rate": 0.0005, "loss": 1.8195712566375732, "step": 10675 }, { "epoch": 0.8416239653133623, "grad_norm": 0.667781412601471, "learning_rate": 0.0005, "loss": 1.8577791452407837, "step": 10676 }, { "epoch": 0.8417027985810012, "grad_norm": 0.6500980257987976, "learning_rate": 0.0005, "loss": 1.780718207359314, "step": 10677 }, { "epoch": 0.8417816318486401, "grad_norm": 0.6668098568916321, "learning_rate": 0.0005, "loss": 1.740416407585144, "step": 10678 }, { "epoch": 0.8418604651162791, "grad_norm": 0.5723721385002136, "learning_rate": 0.0005, "loss": 1.923279047012329, "step": 10679 }, { "epoch": 0.841939298383918, "grad_norm": 0.6523297429084778, "learning_rate": 0.0005, "loss": 1.6121118068695068, "step": 10680 }, { "epoch": 0.8420181316515569, "grad_norm": 0.6289618611335754, "learning_rate": 0.0005, "loss": 1.7839502096176147, "step": 10681 }, { "epoch": 0.842096964919196, "grad_norm": 0.6741948127746582, "learning_rate": 0.0005, "loss": 2.254732608795166, "step": 10682 }, { "epoch": 0.8421757981868349, "grad_norm": 0.5451548099517822, "learning_rate": 0.0005, "loss": 1.7527060508728027, "step": 10683 }, { "epoch": 0.8422546314544738, "grad_norm": 0.618404746055603, "learning_rate": 0.0005, "loss": 1.9481215476989746, "step": 10684 }, { "epoch": 0.8423334647221127, "grad_norm": 0.584328293800354, "learning_rate": 0.0005, "loss": 1.5568528175354004, "step": 10685 }, { "epoch": 0.8424122979897517, "grad_norm": 0.7532247304916382, "learning_rate": 0.0005, "loss": 2.0481557846069336, "step": 10686 }, { "epoch": 0.8424911312573906, "grad_norm": 0.7028487324714661, "learning_rate": 0.0005, "loss": 1.7846574783325195, "step": 10687 }, { "epoch": 0.8425699645250295, "grad_norm": 0.6521657109260559, "learning_rate": 0.0005, "loss": 1.9289724826812744, "step": 10688 }, { "epoch": 0.8426487977926685, "grad_norm": 0.6318284273147583, "learning_rate": 0.0005, "loss": 1.9348771572113037, "step": 10689 }, { "epoch": 0.8427276310603075, "grad_norm": 0.7215744256973267, "learning_rate": 0.0005, "loss": 1.6969068050384521, "step": 10690 }, { "epoch": 0.8428064643279464, "grad_norm": 0.7691812515258789, "learning_rate": 0.0005, "loss": 1.9137890338897705, "step": 10691 }, { "epoch": 0.8428852975955854, "grad_norm": 0.5903488397598267, "learning_rate": 0.0005, "loss": 1.7556246519088745, "step": 10692 }, { "epoch": 0.8429641308632243, "grad_norm": 1.4577291011810303, "learning_rate": 0.0005, "loss": 1.6633775234222412, "step": 10693 }, { "epoch": 0.8430429641308632, "grad_norm": 1.0240148305892944, "learning_rate": 0.0005, "loss": 2.0628607273101807, "step": 10694 }, { "epoch": 0.8431217973985021, "grad_norm": 0.922695517539978, "learning_rate": 0.0005, "loss": 1.674034833908081, "step": 10695 }, { "epoch": 0.8432006306661411, "grad_norm": 0.7109664678573608, "learning_rate": 0.0005, "loss": 1.8191335201263428, "step": 10696 }, { "epoch": 0.8432794639337801, "grad_norm": 0.6772646903991699, "learning_rate": 0.0005, "loss": 1.8442738056182861, "step": 10697 }, { "epoch": 0.843358297201419, "grad_norm": 0.7321011424064636, "learning_rate": 0.0005, "loss": 2.0876576900482178, "step": 10698 }, { "epoch": 0.843437130469058, "grad_norm": 0.6261601448059082, "learning_rate": 0.0005, "loss": 2.1892573833465576, "step": 10699 }, { "epoch": 0.8435159637366969, "grad_norm": 0.7520914077758789, "learning_rate": 0.0005, "loss": 1.8854053020477295, "step": 10700 }, { "epoch": 0.8435947970043358, "grad_norm": 0.7089800834655762, "learning_rate": 0.0005, "loss": 1.847210168838501, "step": 10701 }, { "epoch": 0.8436736302719747, "grad_norm": 0.6881296634674072, "learning_rate": 0.0005, "loss": 2.059335231781006, "step": 10702 }, { "epoch": 0.8437524635396138, "grad_norm": 0.7247923016548157, "learning_rate": 0.0005, "loss": 1.910964846611023, "step": 10703 }, { "epoch": 0.8438312968072527, "grad_norm": 0.8963965177536011, "learning_rate": 0.0005, "loss": 2.0592041015625, "step": 10704 }, { "epoch": 0.8439101300748916, "grad_norm": 0.6273907423019409, "learning_rate": 0.0005, "loss": 1.6060779094696045, "step": 10705 }, { "epoch": 0.8439889633425306, "grad_norm": 0.8461928963661194, "learning_rate": 0.0005, "loss": 2.1715564727783203, "step": 10706 }, { "epoch": 0.8440677966101695, "grad_norm": 0.6942427754402161, "learning_rate": 0.0005, "loss": 1.5220222473144531, "step": 10707 }, { "epoch": 0.8441466298778084, "grad_norm": 0.601650595664978, "learning_rate": 0.0005, "loss": 2.0149199962615967, "step": 10708 }, { "epoch": 0.8442254631454473, "grad_norm": 0.8040207624435425, "learning_rate": 0.0005, "loss": 1.7152574062347412, "step": 10709 }, { "epoch": 0.8443042964130864, "grad_norm": 0.7638383507728577, "learning_rate": 0.0005, "loss": 1.883744478225708, "step": 10710 }, { "epoch": 0.8443831296807253, "grad_norm": 0.7438653707504272, "learning_rate": 0.0005, "loss": 1.96929931640625, "step": 10711 }, { "epoch": 0.8444619629483642, "grad_norm": 0.8102585673332214, "learning_rate": 0.0005, "loss": 1.8081037998199463, "step": 10712 }, { "epoch": 0.8445407962160032, "grad_norm": 0.6463583707809448, "learning_rate": 0.0005, "loss": 1.8874807357788086, "step": 10713 }, { "epoch": 0.8446196294836421, "grad_norm": 0.6367742419242859, "learning_rate": 0.0005, "loss": 1.7292759418487549, "step": 10714 }, { "epoch": 0.844698462751281, "grad_norm": 0.6291674971580505, "learning_rate": 0.0005, "loss": 1.9198384284973145, "step": 10715 }, { "epoch": 0.8447772960189199, "grad_norm": 0.7778330445289612, "learning_rate": 0.0005, "loss": 2.141252279281616, "step": 10716 }, { "epoch": 0.844856129286559, "grad_norm": 0.5360024571418762, "learning_rate": 0.0005, "loss": 1.765002727508545, "step": 10717 }, { "epoch": 0.8449349625541979, "grad_norm": 0.7491407990455627, "learning_rate": 0.0005, "loss": 1.525766372680664, "step": 10718 }, { "epoch": 0.8450137958218368, "grad_norm": 0.6924151182174683, "learning_rate": 0.0005, "loss": 1.6311390399932861, "step": 10719 }, { "epoch": 0.8450926290894758, "grad_norm": 0.7656903266906738, "learning_rate": 0.0005, "loss": 1.9910047054290771, "step": 10720 }, { "epoch": 0.8451714623571147, "grad_norm": 0.641974925994873, "learning_rate": 0.0005, "loss": 1.8817224502563477, "step": 10721 }, { "epoch": 0.8452502956247536, "grad_norm": 0.8773479461669922, "learning_rate": 0.0005, "loss": 2.1239726543426514, "step": 10722 }, { "epoch": 0.8453291288923926, "grad_norm": 0.7913196682929993, "learning_rate": 0.0005, "loss": 1.793283462524414, "step": 10723 }, { "epoch": 0.8454079621600316, "grad_norm": 0.7578370571136475, "learning_rate": 0.0005, "loss": 1.9369041919708252, "step": 10724 }, { "epoch": 0.8454867954276705, "grad_norm": 0.8803998827934265, "learning_rate": 0.0005, "loss": 1.8429431915283203, "step": 10725 }, { "epoch": 0.8455656286953094, "grad_norm": 0.7082814574241638, "learning_rate": 0.0005, "loss": 1.9222111701965332, "step": 10726 }, { "epoch": 0.8456444619629484, "grad_norm": 0.6506069898605347, "learning_rate": 0.0005, "loss": 1.6409342288970947, "step": 10727 }, { "epoch": 0.8457232952305873, "grad_norm": 0.5436004400253296, "learning_rate": 0.0005, "loss": 1.6505910158157349, "step": 10728 }, { "epoch": 0.8458021284982262, "grad_norm": 0.7005735039710999, "learning_rate": 0.0005, "loss": 1.7069828510284424, "step": 10729 }, { "epoch": 0.8458809617658652, "grad_norm": 0.7751452922821045, "learning_rate": 0.0005, "loss": 1.9309916496276855, "step": 10730 }, { "epoch": 0.8459597950335042, "grad_norm": 0.6909346580505371, "learning_rate": 0.0005, "loss": 1.8380773067474365, "step": 10731 }, { "epoch": 0.8460386283011431, "grad_norm": 0.7009491920471191, "learning_rate": 0.0005, "loss": 1.7730252742767334, "step": 10732 }, { "epoch": 0.846117461568782, "grad_norm": 0.6600309610366821, "learning_rate": 0.0005, "loss": 1.9075264930725098, "step": 10733 }, { "epoch": 0.846196294836421, "grad_norm": 0.6607075333595276, "learning_rate": 0.0005, "loss": 1.913441777229309, "step": 10734 }, { "epoch": 0.8462751281040599, "grad_norm": 0.8265632390975952, "learning_rate": 0.0005, "loss": 1.8686599731445312, "step": 10735 }, { "epoch": 0.8463539613716988, "grad_norm": 0.7049846053123474, "learning_rate": 0.0005, "loss": 1.9953036308288574, "step": 10736 }, { "epoch": 0.8464327946393378, "grad_norm": 0.6379351019859314, "learning_rate": 0.0005, "loss": 1.7706942558288574, "step": 10737 }, { "epoch": 0.8465116279069768, "grad_norm": 0.6994251608848572, "learning_rate": 0.0005, "loss": 2.0473358631134033, "step": 10738 }, { "epoch": 0.8465904611746157, "grad_norm": 0.8324824571609497, "learning_rate": 0.0005, "loss": 1.787534236907959, "step": 10739 }, { "epoch": 0.8466692944422546, "grad_norm": 0.5798301696777344, "learning_rate": 0.0005, "loss": 1.554358959197998, "step": 10740 }, { "epoch": 0.8467481277098936, "grad_norm": 0.6870864033699036, "learning_rate": 0.0005, "loss": 1.7323899269104004, "step": 10741 }, { "epoch": 0.8468269609775325, "grad_norm": 0.8077320456504822, "learning_rate": 0.0005, "loss": 1.7121020555496216, "step": 10742 }, { "epoch": 0.8469057942451714, "grad_norm": 0.7892671227455139, "learning_rate": 0.0005, "loss": 1.8291527032852173, "step": 10743 }, { "epoch": 0.8469846275128105, "grad_norm": 0.5814428329467773, "learning_rate": 0.0005, "loss": 1.7399835586547852, "step": 10744 }, { "epoch": 0.8470634607804494, "grad_norm": 0.718880295753479, "learning_rate": 0.0005, "loss": 1.7796013355255127, "step": 10745 }, { "epoch": 0.8471422940480883, "grad_norm": 0.7102035284042358, "learning_rate": 0.0005, "loss": 1.6688284873962402, "step": 10746 }, { "epoch": 0.8472211273157272, "grad_norm": 0.6438635587692261, "learning_rate": 0.0005, "loss": 1.673483967781067, "step": 10747 }, { "epoch": 0.8472999605833662, "grad_norm": 0.6867705583572388, "learning_rate": 0.0005, "loss": 1.5859103202819824, "step": 10748 }, { "epoch": 0.8473787938510051, "grad_norm": 0.6034401655197144, "learning_rate": 0.0005, "loss": 1.8854951858520508, "step": 10749 }, { "epoch": 0.847457627118644, "grad_norm": 0.6144939064979553, "learning_rate": 0.0005, "loss": 1.6871891021728516, "step": 10750 }, { "epoch": 0.847536460386283, "grad_norm": 0.5343065857887268, "learning_rate": 0.0005, "loss": 1.7316131591796875, "step": 10751 }, { "epoch": 0.847615293653922, "grad_norm": 0.9290215969085693, "learning_rate": 0.0005, "loss": 1.6629177331924438, "step": 10752 }, { "epoch": 0.8476941269215609, "grad_norm": 0.6493815183639526, "learning_rate": 0.0005, "loss": 1.6826181411743164, "step": 10753 }, { "epoch": 0.8477729601891999, "grad_norm": 0.6602986454963684, "learning_rate": 0.0005, "loss": 1.8179423809051514, "step": 10754 }, { "epoch": 0.8478517934568388, "grad_norm": 0.7536755800247192, "learning_rate": 0.0005, "loss": 1.744909405708313, "step": 10755 }, { "epoch": 0.8479306267244777, "grad_norm": 0.7342852354049683, "learning_rate": 0.0005, "loss": 1.9061824083328247, "step": 10756 }, { "epoch": 0.8480094599921166, "grad_norm": 1.2057991027832031, "learning_rate": 0.0005, "loss": 2.21328067779541, "step": 10757 }, { "epoch": 0.8480882932597557, "grad_norm": 0.7505393624305725, "learning_rate": 0.0005, "loss": 1.7901040315628052, "step": 10758 }, { "epoch": 0.8481671265273946, "grad_norm": 0.7403426170349121, "learning_rate": 0.0005, "loss": 2.2915730476379395, "step": 10759 }, { "epoch": 0.8482459597950335, "grad_norm": 0.8484962582588196, "learning_rate": 0.0005, "loss": 1.8762235641479492, "step": 10760 }, { "epoch": 0.8483247930626725, "grad_norm": 0.6246187090873718, "learning_rate": 0.0005, "loss": 2.086304187774658, "step": 10761 }, { "epoch": 0.8484036263303114, "grad_norm": 0.7224816083908081, "learning_rate": 0.0005, "loss": 1.8534648418426514, "step": 10762 }, { "epoch": 0.8484824595979503, "grad_norm": 0.5704164505004883, "learning_rate": 0.0005, "loss": 1.9564318656921387, "step": 10763 }, { "epoch": 0.8485612928655892, "grad_norm": 0.7327892780303955, "learning_rate": 0.0005, "loss": 1.8021953105926514, "step": 10764 }, { "epoch": 0.8486401261332283, "grad_norm": 0.7395235896110535, "learning_rate": 0.0005, "loss": 1.8420720100402832, "step": 10765 }, { "epoch": 0.8487189594008672, "grad_norm": 0.577225923538208, "learning_rate": 0.0005, "loss": 1.6935558319091797, "step": 10766 }, { "epoch": 0.8487977926685061, "grad_norm": 0.7518944144248962, "learning_rate": 0.0005, "loss": 1.9562008380889893, "step": 10767 }, { "epoch": 0.8488766259361451, "grad_norm": 0.6332230567932129, "learning_rate": 0.0005, "loss": 1.857848882675171, "step": 10768 }, { "epoch": 0.848955459203784, "grad_norm": 0.5626236200332642, "learning_rate": 0.0005, "loss": 1.628310203552246, "step": 10769 }, { "epoch": 0.8490342924714229, "grad_norm": 0.8386937975883484, "learning_rate": 0.0005, "loss": 1.937524437904358, "step": 10770 }, { "epoch": 0.8491131257390618, "grad_norm": 0.6318687200546265, "learning_rate": 0.0005, "loss": 1.9435374736785889, "step": 10771 }, { "epoch": 0.8491919590067009, "grad_norm": 0.8136398196220398, "learning_rate": 0.0005, "loss": 2.008770704269409, "step": 10772 }, { "epoch": 0.8492707922743398, "grad_norm": 0.6107156276702881, "learning_rate": 0.0005, "loss": 1.9343127012252808, "step": 10773 }, { "epoch": 0.8493496255419787, "grad_norm": 0.5892302393913269, "learning_rate": 0.0005, "loss": 1.688395619392395, "step": 10774 }, { "epoch": 0.8494284588096177, "grad_norm": 0.9081904888153076, "learning_rate": 0.0005, "loss": 2.066786289215088, "step": 10775 }, { "epoch": 0.8495072920772566, "grad_norm": 0.6018631458282471, "learning_rate": 0.0005, "loss": 1.5712924003601074, "step": 10776 }, { "epoch": 0.8495861253448955, "grad_norm": 0.5916231870651245, "learning_rate": 0.0005, "loss": 1.7541162967681885, "step": 10777 }, { "epoch": 0.8496649586125344, "grad_norm": 0.6775339841842651, "learning_rate": 0.0005, "loss": 2.03230619430542, "step": 10778 }, { "epoch": 0.8497437918801735, "grad_norm": 0.6293484568595886, "learning_rate": 0.0005, "loss": 1.9339909553527832, "step": 10779 }, { "epoch": 0.8498226251478124, "grad_norm": 0.6278170943260193, "learning_rate": 0.0005, "loss": 1.7611446380615234, "step": 10780 }, { "epoch": 0.8499014584154513, "grad_norm": 13.839205741882324, "learning_rate": 0.0005, "loss": 1.5875191688537598, "step": 10781 }, { "epoch": 0.8499802916830903, "grad_norm": 0.5279918909072876, "learning_rate": 0.0005, "loss": 1.9597773551940918, "step": 10782 }, { "epoch": 0.8500591249507292, "grad_norm": 0.6725324392318726, "learning_rate": 0.0005, "loss": 1.885702133178711, "step": 10783 }, { "epoch": 0.8501379582183681, "grad_norm": 0.7181377410888672, "learning_rate": 0.0005, "loss": 1.811863899230957, "step": 10784 }, { "epoch": 0.850216791486007, "grad_norm": 0.5747455358505249, "learning_rate": 0.0005, "loss": 1.988573670387268, "step": 10785 }, { "epoch": 0.8502956247536461, "grad_norm": 1.0737180709838867, "learning_rate": 0.0005, "loss": 1.921647310256958, "step": 10786 }, { "epoch": 0.850374458021285, "grad_norm": 0.5783663392066956, "learning_rate": 0.0005, "loss": 1.8088399171829224, "step": 10787 }, { "epoch": 0.8504532912889239, "grad_norm": 1.3448386192321777, "learning_rate": 0.0005, "loss": 2.050844669342041, "step": 10788 }, { "epoch": 0.8505321245565629, "grad_norm": 0.7308090329170227, "learning_rate": 0.0005, "loss": 1.7769255638122559, "step": 10789 }, { "epoch": 0.8506109578242018, "grad_norm": 0.5648853182792664, "learning_rate": 0.0005, "loss": 1.8071422576904297, "step": 10790 }, { "epoch": 0.8506897910918407, "grad_norm": 0.6536906361579895, "learning_rate": 0.0005, "loss": 2.141787528991699, "step": 10791 }, { "epoch": 0.8507686243594798, "grad_norm": 0.7052187323570251, "learning_rate": 0.0005, "loss": 2.1665430068969727, "step": 10792 }, { "epoch": 0.8508474576271187, "grad_norm": 0.5636386871337891, "learning_rate": 0.0005, "loss": 1.8404356241226196, "step": 10793 }, { "epoch": 0.8509262908947576, "grad_norm": 0.6271618604660034, "learning_rate": 0.0005, "loss": 1.8084266185760498, "step": 10794 }, { "epoch": 0.8510051241623965, "grad_norm": 0.4974965453147888, "learning_rate": 0.0005, "loss": 1.7581021785736084, "step": 10795 }, { "epoch": 0.8510839574300355, "grad_norm": 0.6180289387702942, "learning_rate": 0.0005, "loss": 1.6992220878601074, "step": 10796 }, { "epoch": 0.8511627906976744, "grad_norm": 0.5370598435401917, "learning_rate": 0.0005, "loss": 1.7962416410446167, "step": 10797 }, { "epoch": 0.8512416239653133, "grad_norm": 0.9920796155929565, "learning_rate": 0.0005, "loss": 1.8555982112884521, "step": 10798 }, { "epoch": 0.8513204572329524, "grad_norm": 0.6215249300003052, "learning_rate": 0.0005, "loss": 1.7592859268188477, "step": 10799 }, { "epoch": 0.8513992905005913, "grad_norm": 0.784356415271759, "learning_rate": 0.0005, "loss": 1.7172561883926392, "step": 10800 }, { "epoch": 0.8514781237682302, "grad_norm": 0.6201134324073792, "learning_rate": 0.0005, "loss": 1.9230769872665405, "step": 10801 }, { "epoch": 0.8515569570358691, "grad_norm": 0.6042521595954895, "learning_rate": 0.0005, "loss": 1.8705596923828125, "step": 10802 }, { "epoch": 0.8516357903035081, "grad_norm": 0.5333616733551025, "learning_rate": 0.0005, "loss": 1.8683669567108154, "step": 10803 }, { "epoch": 0.851714623571147, "grad_norm": 0.6637266278266907, "learning_rate": 0.0005, "loss": 1.8554810285568237, "step": 10804 }, { "epoch": 0.8517934568387859, "grad_norm": 0.6153160929679871, "learning_rate": 0.0005, "loss": 1.7290866374969482, "step": 10805 }, { "epoch": 0.851872290106425, "grad_norm": 0.7930431365966797, "learning_rate": 0.0005, "loss": 1.8370707035064697, "step": 10806 }, { "epoch": 0.8519511233740639, "grad_norm": 0.736497700214386, "learning_rate": 0.0005, "loss": 1.954465627670288, "step": 10807 }, { "epoch": 0.8520299566417028, "grad_norm": 0.6049705743789673, "learning_rate": 0.0005, "loss": 1.844396948814392, "step": 10808 }, { "epoch": 0.8521087899093417, "grad_norm": 0.6939821839332581, "learning_rate": 0.0005, "loss": 2.11021089553833, "step": 10809 }, { "epoch": 0.8521876231769807, "grad_norm": 0.8285228610038757, "learning_rate": 0.0005, "loss": 2.0898780822753906, "step": 10810 }, { "epoch": 0.8522664564446196, "grad_norm": 0.542055606842041, "learning_rate": 0.0005, "loss": 1.7354276180267334, "step": 10811 }, { "epoch": 0.8523452897122585, "grad_norm": 0.5029675364494324, "learning_rate": 0.0005, "loss": 2.069082260131836, "step": 10812 }, { "epoch": 0.8524241229798976, "grad_norm": 0.6724139451980591, "learning_rate": 0.0005, "loss": 1.8124313354492188, "step": 10813 }, { "epoch": 0.8525029562475365, "grad_norm": 0.7783294916152954, "learning_rate": 0.0005, "loss": 1.7961654663085938, "step": 10814 }, { "epoch": 0.8525817895151754, "grad_norm": 0.537260890007019, "learning_rate": 0.0005, "loss": 1.7439203262329102, "step": 10815 }, { "epoch": 0.8526606227828143, "grad_norm": 0.7770602107048035, "learning_rate": 0.0005, "loss": 1.8434131145477295, "step": 10816 }, { "epoch": 0.8527394560504533, "grad_norm": 0.7117513418197632, "learning_rate": 0.0005, "loss": 1.9237585067749023, "step": 10817 }, { "epoch": 0.8528182893180922, "grad_norm": 0.5454321503639221, "learning_rate": 0.0005, "loss": 1.5459383726119995, "step": 10818 }, { "epoch": 0.8528971225857311, "grad_norm": 0.693088173866272, "learning_rate": 0.0005, "loss": 1.8404393196105957, "step": 10819 }, { "epoch": 0.8529759558533702, "grad_norm": 0.7418949007987976, "learning_rate": 0.0005, "loss": 1.8304214477539062, "step": 10820 }, { "epoch": 0.8530547891210091, "grad_norm": 0.4946625828742981, "learning_rate": 0.0005, "loss": 1.7961143255233765, "step": 10821 }, { "epoch": 0.853133622388648, "grad_norm": 0.6759390830993652, "learning_rate": 0.0005, "loss": 1.7490090131759644, "step": 10822 }, { "epoch": 0.853212455656287, "grad_norm": 0.5692172050476074, "learning_rate": 0.0005, "loss": 1.964889645576477, "step": 10823 }, { "epoch": 0.8532912889239259, "grad_norm": 0.644793689250946, "learning_rate": 0.0005, "loss": 1.8473546504974365, "step": 10824 }, { "epoch": 0.8533701221915648, "grad_norm": 0.6496919393539429, "learning_rate": 0.0005, "loss": 1.7236225605010986, "step": 10825 }, { "epoch": 0.8534489554592037, "grad_norm": 0.49692708253860474, "learning_rate": 0.0005, "loss": 1.7203575372695923, "step": 10826 }, { "epoch": 0.8535277887268428, "grad_norm": 0.5613380670547485, "learning_rate": 0.0005, "loss": 1.7235009670257568, "step": 10827 }, { "epoch": 0.8536066219944817, "grad_norm": 0.6624119281768799, "learning_rate": 0.0005, "loss": 1.8806239366531372, "step": 10828 }, { "epoch": 0.8536854552621206, "grad_norm": 0.665294349193573, "learning_rate": 0.0005, "loss": 1.9219064712524414, "step": 10829 }, { "epoch": 0.8537642885297596, "grad_norm": 0.9092952013015747, "learning_rate": 0.0005, "loss": 1.7725391387939453, "step": 10830 }, { "epoch": 0.8538431217973985, "grad_norm": 0.7030458450317383, "learning_rate": 0.0005, "loss": 1.9302136898040771, "step": 10831 }, { "epoch": 0.8539219550650374, "grad_norm": 0.6516076326370239, "learning_rate": 0.0005, "loss": 1.5869256258010864, "step": 10832 }, { "epoch": 0.8540007883326763, "grad_norm": 0.801878035068512, "learning_rate": 0.0005, "loss": 2.046215057373047, "step": 10833 }, { "epoch": 0.8540796216003154, "grad_norm": 0.8011929988861084, "learning_rate": 0.0005, "loss": 1.9147847890853882, "step": 10834 }, { "epoch": 0.8541584548679543, "grad_norm": 0.6580336689949036, "learning_rate": 0.0005, "loss": 1.7866164445877075, "step": 10835 }, { "epoch": 0.8542372881355932, "grad_norm": 0.6294060945510864, "learning_rate": 0.0005, "loss": 1.8765511512756348, "step": 10836 }, { "epoch": 0.8543161214032322, "grad_norm": 0.6233863830566406, "learning_rate": 0.0005, "loss": 1.7113722562789917, "step": 10837 }, { "epoch": 0.8543949546708711, "grad_norm": 0.6098200678825378, "learning_rate": 0.0005, "loss": 1.8715344667434692, "step": 10838 }, { "epoch": 0.85447378793851, "grad_norm": 0.6406558752059937, "learning_rate": 0.0005, "loss": 1.7081087827682495, "step": 10839 }, { "epoch": 0.8545526212061489, "grad_norm": 0.5869911909103394, "learning_rate": 0.0005, "loss": 2.0831191539764404, "step": 10840 }, { "epoch": 0.854631454473788, "grad_norm": 0.6150075793266296, "learning_rate": 0.0005, "loss": 1.7005945444107056, "step": 10841 }, { "epoch": 0.8547102877414269, "grad_norm": 0.6309293508529663, "learning_rate": 0.0005, "loss": 1.8856604099273682, "step": 10842 }, { "epoch": 0.8547891210090658, "grad_norm": 0.6464264988899231, "learning_rate": 0.0005, "loss": 1.8518221378326416, "step": 10843 }, { "epoch": 0.8548679542767048, "grad_norm": 0.7954972386360168, "learning_rate": 0.0005, "loss": 1.880295753479004, "step": 10844 }, { "epoch": 0.8549467875443437, "grad_norm": 0.5258210301399231, "learning_rate": 0.0005, "loss": 1.7691576480865479, "step": 10845 }, { "epoch": 0.8550256208119826, "grad_norm": 0.9251082539558411, "learning_rate": 0.0005, "loss": 1.9830294847488403, "step": 10846 }, { "epoch": 0.8551044540796215, "grad_norm": 0.6898059844970703, "learning_rate": 0.0005, "loss": 1.8894755840301514, "step": 10847 }, { "epoch": 0.8551832873472606, "grad_norm": 0.5700519680976868, "learning_rate": 0.0005, "loss": 1.7762150764465332, "step": 10848 }, { "epoch": 0.8552621206148995, "grad_norm": 0.7709536552429199, "learning_rate": 0.0005, "loss": 1.8024349212646484, "step": 10849 }, { "epoch": 0.8553409538825384, "grad_norm": 0.6792388558387756, "learning_rate": 0.0005, "loss": 2.0629966259002686, "step": 10850 }, { "epoch": 0.8554197871501774, "grad_norm": 0.6836187839508057, "learning_rate": 0.0005, "loss": 1.7537751197814941, "step": 10851 }, { "epoch": 0.8554986204178163, "grad_norm": 0.6403771638870239, "learning_rate": 0.0005, "loss": 1.855696678161621, "step": 10852 }, { "epoch": 0.8555774536854552, "grad_norm": 0.770372211933136, "learning_rate": 0.0005, "loss": 1.8750087022781372, "step": 10853 }, { "epoch": 0.8556562869530943, "grad_norm": 0.5404126048088074, "learning_rate": 0.0005, "loss": 1.5354315042495728, "step": 10854 }, { "epoch": 0.8557351202207332, "grad_norm": 0.6376833915710449, "learning_rate": 0.0005, "loss": 1.9732569456100464, "step": 10855 }, { "epoch": 0.8558139534883721, "grad_norm": 0.5585441589355469, "learning_rate": 0.0005, "loss": 1.7832908630371094, "step": 10856 }, { "epoch": 0.855892786756011, "grad_norm": 0.576744019985199, "learning_rate": 0.0005, "loss": 1.6983978748321533, "step": 10857 }, { "epoch": 0.85597162002365, "grad_norm": 0.6941421031951904, "learning_rate": 0.0005, "loss": 1.8043959140777588, "step": 10858 }, { "epoch": 0.8560504532912889, "grad_norm": 0.5598097443580627, "learning_rate": 0.0005, "loss": 1.971810221672058, "step": 10859 }, { "epoch": 0.8561292865589278, "grad_norm": 0.6003239154815674, "learning_rate": 0.0005, "loss": 1.7990747690200806, "step": 10860 }, { "epoch": 0.8562081198265669, "grad_norm": 0.7447678446769714, "learning_rate": 0.0005, "loss": 1.8719825744628906, "step": 10861 }, { "epoch": 0.8562869530942058, "grad_norm": 0.5707050561904907, "learning_rate": 0.0005, "loss": 1.750500202178955, "step": 10862 }, { "epoch": 0.8563657863618447, "grad_norm": 0.5899686813354492, "learning_rate": 0.0005, "loss": 1.5500179529190063, "step": 10863 }, { "epoch": 0.8564446196294836, "grad_norm": 0.6320222020149231, "learning_rate": 0.0005, "loss": 1.5582735538482666, "step": 10864 }, { "epoch": 0.8565234528971226, "grad_norm": 0.6280597448348999, "learning_rate": 0.0005, "loss": 1.726547360420227, "step": 10865 }, { "epoch": 0.8566022861647615, "grad_norm": 1.1111937761306763, "learning_rate": 0.0005, "loss": 1.9410820007324219, "step": 10866 }, { "epoch": 0.8566811194324004, "grad_norm": 0.6523169279098511, "learning_rate": 0.0005, "loss": 1.996545433998108, "step": 10867 }, { "epoch": 0.8567599527000395, "grad_norm": 0.6004152894020081, "learning_rate": 0.0005, "loss": 1.888498306274414, "step": 10868 }, { "epoch": 0.8568387859676784, "grad_norm": 0.6605027914047241, "learning_rate": 0.0005, "loss": 1.7548909187316895, "step": 10869 }, { "epoch": 0.8569176192353173, "grad_norm": 0.5537862181663513, "learning_rate": 0.0005, "loss": 1.693164348602295, "step": 10870 }, { "epoch": 0.8569964525029562, "grad_norm": 0.7334237694740295, "learning_rate": 0.0005, "loss": 2.2275657653808594, "step": 10871 }, { "epoch": 0.8570752857705952, "grad_norm": 0.5975823402404785, "learning_rate": 0.0005, "loss": 1.8089163303375244, "step": 10872 }, { "epoch": 0.8571541190382341, "grad_norm": 0.693657636642456, "learning_rate": 0.0005, "loss": 1.827239990234375, "step": 10873 }, { "epoch": 0.857232952305873, "grad_norm": 0.6162561178207397, "learning_rate": 0.0005, "loss": 1.7425280809402466, "step": 10874 }, { "epoch": 0.8573117855735121, "grad_norm": 0.6587256789207458, "learning_rate": 0.0005, "loss": 2.348578929901123, "step": 10875 }, { "epoch": 0.857390618841151, "grad_norm": 0.5482460260391235, "learning_rate": 0.0005, "loss": 1.8175603151321411, "step": 10876 }, { "epoch": 0.8574694521087899, "grad_norm": 0.512476921081543, "learning_rate": 0.0005, "loss": 1.754446029663086, "step": 10877 }, { "epoch": 0.8575482853764288, "grad_norm": 0.6715736389160156, "learning_rate": 0.0005, "loss": 1.6299850940704346, "step": 10878 }, { "epoch": 0.8576271186440678, "grad_norm": 0.6604553461074829, "learning_rate": 0.0005, "loss": 1.9326443672180176, "step": 10879 }, { "epoch": 0.8577059519117067, "grad_norm": 0.5301081538200378, "learning_rate": 0.0005, "loss": 1.7571440935134888, "step": 10880 }, { "epoch": 0.8577847851793456, "grad_norm": 0.6814805865287781, "learning_rate": 0.0005, "loss": 1.9936764240264893, "step": 10881 }, { "epoch": 0.8578636184469847, "grad_norm": 0.5545506477355957, "learning_rate": 0.0005, "loss": 1.8055896759033203, "step": 10882 }, { "epoch": 0.8579424517146236, "grad_norm": 0.6578049659729004, "learning_rate": 0.0005, "loss": 1.6817615032196045, "step": 10883 }, { "epoch": 0.8580212849822625, "grad_norm": 0.7227363586425781, "learning_rate": 0.0005, "loss": 1.7888777256011963, "step": 10884 }, { "epoch": 0.8581001182499015, "grad_norm": 1.8110629320144653, "learning_rate": 0.0005, "loss": 1.9182937145233154, "step": 10885 }, { "epoch": 0.8581789515175404, "grad_norm": 0.8317458033561707, "learning_rate": 0.0005, "loss": 1.6668856143951416, "step": 10886 }, { "epoch": 0.8582577847851793, "grad_norm": 0.6892325282096863, "learning_rate": 0.0005, "loss": 2.0795235633850098, "step": 10887 }, { "epoch": 0.8583366180528182, "grad_norm": 0.9952433705329895, "learning_rate": 0.0005, "loss": 2.138488292694092, "step": 10888 }, { "epoch": 0.8584154513204573, "grad_norm": 0.644352912902832, "learning_rate": 0.0005, "loss": 1.7288295030593872, "step": 10889 }, { "epoch": 0.8584942845880962, "grad_norm": 0.6097816228866577, "learning_rate": 0.0005, "loss": 1.767846703529358, "step": 10890 }, { "epoch": 0.8585731178557351, "grad_norm": 0.7113772630691528, "learning_rate": 0.0005, "loss": 2.0542263984680176, "step": 10891 }, { "epoch": 0.8586519511233741, "grad_norm": 0.6143508553504944, "learning_rate": 0.0005, "loss": 1.7445902824401855, "step": 10892 }, { "epoch": 0.858730784391013, "grad_norm": 0.8254122734069824, "learning_rate": 0.0005, "loss": 1.9404635429382324, "step": 10893 }, { "epoch": 0.8588096176586519, "grad_norm": 0.7182203531265259, "learning_rate": 0.0005, "loss": 1.9708222150802612, "step": 10894 }, { "epoch": 0.8588884509262908, "grad_norm": 0.6343135237693787, "learning_rate": 0.0005, "loss": 1.77861750125885, "step": 10895 }, { "epoch": 0.8589672841939299, "grad_norm": 0.7945162653923035, "learning_rate": 0.0005, "loss": 1.8970718383789062, "step": 10896 }, { "epoch": 0.8590461174615688, "grad_norm": 0.8062782287597656, "learning_rate": 0.0005, "loss": 2.15450382232666, "step": 10897 }, { "epoch": 0.8591249507292077, "grad_norm": 0.7579550743103027, "learning_rate": 0.0005, "loss": 1.9207566976547241, "step": 10898 }, { "epoch": 0.8592037839968467, "grad_norm": 0.6469972729682922, "learning_rate": 0.0005, "loss": 1.6606171131134033, "step": 10899 }, { "epoch": 0.8592826172644856, "grad_norm": 0.7150862812995911, "learning_rate": 0.0005, "loss": 1.747056245803833, "step": 10900 }, { "epoch": 0.8593614505321245, "grad_norm": 0.6538676619529724, "learning_rate": 0.0005, "loss": 1.8509806394577026, "step": 10901 }, { "epoch": 0.8594402837997634, "grad_norm": 0.5611709952354431, "learning_rate": 0.0005, "loss": 1.900644063949585, "step": 10902 }, { "epoch": 0.8595191170674025, "grad_norm": 0.4746920168399811, "learning_rate": 0.0005, "loss": 1.7417957782745361, "step": 10903 }, { "epoch": 0.8595979503350414, "grad_norm": 1.055834412574768, "learning_rate": 0.0005, "loss": 1.7302460670471191, "step": 10904 }, { "epoch": 0.8596767836026803, "grad_norm": 0.6477896571159363, "learning_rate": 0.0005, "loss": 1.7880845069885254, "step": 10905 }, { "epoch": 0.8597556168703193, "grad_norm": 0.5018757581710815, "learning_rate": 0.0005, "loss": 1.6013085842132568, "step": 10906 }, { "epoch": 0.8598344501379582, "grad_norm": 0.5532771944999695, "learning_rate": 0.0005, "loss": 1.7252607345581055, "step": 10907 }, { "epoch": 0.8599132834055971, "grad_norm": 0.7314910888671875, "learning_rate": 0.0005, "loss": 1.7617321014404297, "step": 10908 }, { "epoch": 0.859992116673236, "grad_norm": 0.9786425232887268, "learning_rate": 0.0005, "loss": 1.9103658199310303, "step": 10909 }, { "epoch": 0.8600709499408751, "grad_norm": 0.5219215154647827, "learning_rate": 0.0005, "loss": 1.815738558769226, "step": 10910 }, { "epoch": 0.860149783208514, "grad_norm": 0.7270509600639343, "learning_rate": 0.0005, "loss": 1.8748397827148438, "step": 10911 }, { "epoch": 0.8602286164761529, "grad_norm": 0.8322158455848694, "learning_rate": 0.0005, "loss": 1.9923198223114014, "step": 10912 }, { "epoch": 0.8603074497437919, "grad_norm": 0.654731273651123, "learning_rate": 0.0005, "loss": 1.908827543258667, "step": 10913 }, { "epoch": 0.8603862830114308, "grad_norm": 0.6574465036392212, "learning_rate": 0.0005, "loss": 1.8726979494094849, "step": 10914 }, { "epoch": 0.8604651162790697, "grad_norm": 0.8381742238998413, "learning_rate": 0.0005, "loss": 2.0610299110412598, "step": 10915 }, { "epoch": 0.8605439495467088, "grad_norm": 0.7120850682258606, "learning_rate": 0.0005, "loss": 1.6297013759613037, "step": 10916 }, { "epoch": 0.8606227828143477, "grad_norm": 0.8495756983757019, "learning_rate": 0.0005, "loss": 1.877672553062439, "step": 10917 }, { "epoch": 0.8607016160819866, "grad_norm": 0.732624351978302, "learning_rate": 0.0005, "loss": 1.9183926582336426, "step": 10918 }, { "epoch": 0.8607804493496255, "grad_norm": 0.8271675109863281, "learning_rate": 0.0005, "loss": 1.709836483001709, "step": 10919 }, { "epoch": 0.8608592826172645, "grad_norm": 0.5805216431617737, "learning_rate": 0.0005, "loss": 1.6799322366714478, "step": 10920 }, { "epoch": 0.8609381158849034, "grad_norm": 0.725286066532135, "learning_rate": 0.0005, "loss": 1.8726475238800049, "step": 10921 }, { "epoch": 0.8610169491525423, "grad_norm": 0.6053407788276672, "learning_rate": 0.0005, "loss": 1.8278861045837402, "step": 10922 }, { "epoch": 0.8610957824201814, "grad_norm": 0.54362952709198, "learning_rate": 0.0005, "loss": 1.4166545867919922, "step": 10923 }, { "epoch": 0.8611746156878203, "grad_norm": 0.5730987191200256, "learning_rate": 0.0005, "loss": 1.6673684120178223, "step": 10924 }, { "epoch": 0.8612534489554592, "grad_norm": 0.6512162089347839, "learning_rate": 0.0005, "loss": 1.939113974571228, "step": 10925 }, { "epoch": 0.8613322822230981, "grad_norm": 0.5683672428131104, "learning_rate": 0.0005, "loss": 1.71653151512146, "step": 10926 }, { "epoch": 0.8614111154907371, "grad_norm": 0.6237955689430237, "learning_rate": 0.0005, "loss": 1.9463368654251099, "step": 10927 }, { "epoch": 0.861489948758376, "grad_norm": 0.6717649102210999, "learning_rate": 0.0005, "loss": 1.8934545516967773, "step": 10928 }, { "epoch": 0.861568782026015, "grad_norm": 0.8453453779220581, "learning_rate": 0.0005, "loss": 1.7044123411178589, "step": 10929 }, { "epoch": 0.861647615293654, "grad_norm": 0.5831731557846069, "learning_rate": 0.0005, "loss": 1.6984211206436157, "step": 10930 }, { "epoch": 0.8617264485612929, "grad_norm": 0.6475604176521301, "learning_rate": 0.0005, "loss": 1.9868135452270508, "step": 10931 }, { "epoch": 0.8618052818289318, "grad_norm": 0.5918003916740417, "learning_rate": 0.0005, "loss": 1.8279331922531128, "step": 10932 }, { "epoch": 0.8618841150965707, "grad_norm": 0.7453907132148743, "learning_rate": 0.0005, "loss": 1.9088730812072754, "step": 10933 }, { "epoch": 0.8619629483642097, "grad_norm": 0.9081973433494568, "learning_rate": 0.0005, "loss": 2.1226871013641357, "step": 10934 }, { "epoch": 0.8620417816318486, "grad_norm": 0.626147985458374, "learning_rate": 0.0005, "loss": 1.8204517364501953, "step": 10935 }, { "epoch": 0.8621206148994875, "grad_norm": 0.5792678594589233, "learning_rate": 0.0005, "loss": 1.8056163787841797, "step": 10936 }, { "epoch": 0.8621994481671266, "grad_norm": 0.5343708395957947, "learning_rate": 0.0005, "loss": 1.904533863067627, "step": 10937 }, { "epoch": 0.8622782814347655, "grad_norm": 0.697441816329956, "learning_rate": 0.0005, "loss": 1.8424150943756104, "step": 10938 }, { "epoch": 0.8623571147024044, "grad_norm": 0.5667994618415833, "learning_rate": 0.0005, "loss": 1.6963220834732056, "step": 10939 }, { "epoch": 0.8624359479700433, "grad_norm": 0.6716017723083496, "learning_rate": 0.0005, "loss": 1.9153742790222168, "step": 10940 }, { "epoch": 0.8625147812376823, "grad_norm": 0.6615665555000305, "learning_rate": 0.0005, "loss": 2.1233181953430176, "step": 10941 }, { "epoch": 0.8625936145053212, "grad_norm": 0.48560917377471924, "learning_rate": 0.0005, "loss": 1.764925479888916, "step": 10942 }, { "epoch": 0.8626724477729601, "grad_norm": 0.7256820201873779, "learning_rate": 0.0005, "loss": 2.0828309059143066, "step": 10943 }, { "epoch": 0.8627512810405992, "grad_norm": 0.6442727446556091, "learning_rate": 0.0005, "loss": 1.6730518341064453, "step": 10944 }, { "epoch": 0.8628301143082381, "grad_norm": 0.7347531914710999, "learning_rate": 0.0005, "loss": 1.7853727340698242, "step": 10945 }, { "epoch": 0.862908947575877, "grad_norm": 0.7609979510307312, "learning_rate": 0.0005, "loss": 1.8923653364181519, "step": 10946 }, { "epoch": 0.862987780843516, "grad_norm": 0.5877893567085266, "learning_rate": 0.0005, "loss": 2.001016139984131, "step": 10947 }, { "epoch": 0.8630666141111549, "grad_norm": 0.7066425085067749, "learning_rate": 0.0005, "loss": 1.8979541063308716, "step": 10948 }, { "epoch": 0.8631454473787938, "grad_norm": 0.7315638065338135, "learning_rate": 0.0005, "loss": 1.7998895645141602, "step": 10949 }, { "epoch": 0.8632242806464328, "grad_norm": 0.5640417337417603, "learning_rate": 0.0005, "loss": 1.558164358139038, "step": 10950 }, { "epoch": 0.8633031139140718, "grad_norm": 0.7305045127868652, "learning_rate": 0.0005, "loss": 1.8378084897994995, "step": 10951 }, { "epoch": 0.8633819471817107, "grad_norm": 0.8859153389930725, "learning_rate": 0.0005, "loss": 1.8918184041976929, "step": 10952 }, { "epoch": 0.8634607804493496, "grad_norm": 0.8154594302177429, "learning_rate": 0.0005, "loss": 1.908111572265625, "step": 10953 }, { "epoch": 0.8635396137169886, "grad_norm": 0.7523348927497864, "learning_rate": 0.0005, "loss": 1.9009428024291992, "step": 10954 }, { "epoch": 0.8636184469846275, "grad_norm": 0.6605863571166992, "learning_rate": 0.0005, "loss": 1.7706365585327148, "step": 10955 }, { "epoch": 0.8636972802522664, "grad_norm": 0.753574013710022, "learning_rate": 0.0005, "loss": 1.8694064617156982, "step": 10956 }, { "epoch": 0.8637761135199054, "grad_norm": 0.559638500213623, "learning_rate": 0.0005, "loss": 1.7622467279434204, "step": 10957 }, { "epoch": 0.8638549467875444, "grad_norm": 0.8141286969184875, "learning_rate": 0.0005, "loss": 1.7860145568847656, "step": 10958 }, { "epoch": 0.8639337800551833, "grad_norm": 0.8193605542182922, "learning_rate": 0.0005, "loss": 1.8026741743087769, "step": 10959 }, { "epoch": 0.8640126133228222, "grad_norm": 0.8567485809326172, "learning_rate": 0.0005, "loss": 1.9975149631500244, "step": 10960 }, { "epoch": 0.8640914465904612, "grad_norm": 0.6397842168807983, "learning_rate": 0.0005, "loss": 1.5915462970733643, "step": 10961 }, { "epoch": 0.8641702798581001, "grad_norm": 0.6887772679328918, "learning_rate": 0.0005, "loss": 1.9245867729187012, "step": 10962 }, { "epoch": 0.864249113125739, "grad_norm": 0.7835848331451416, "learning_rate": 0.0005, "loss": 1.690565824508667, "step": 10963 }, { "epoch": 0.864327946393378, "grad_norm": 0.7943428754806519, "learning_rate": 0.0005, "loss": 1.9133481979370117, "step": 10964 }, { "epoch": 0.864406779661017, "grad_norm": 0.5429421067237854, "learning_rate": 0.0005, "loss": 1.719849705696106, "step": 10965 }, { "epoch": 0.8644856129286559, "grad_norm": 0.5435956120491028, "learning_rate": 0.0005, "loss": 1.6704304218292236, "step": 10966 }, { "epoch": 0.8645644461962948, "grad_norm": 0.7550603747367859, "learning_rate": 0.0005, "loss": 1.9645888805389404, "step": 10967 }, { "epoch": 0.8646432794639338, "grad_norm": 0.6104381084442139, "learning_rate": 0.0005, "loss": 1.8007314205169678, "step": 10968 }, { "epoch": 0.8647221127315727, "grad_norm": 0.5964451432228088, "learning_rate": 0.0005, "loss": 1.9275636672973633, "step": 10969 }, { "epoch": 0.8648009459992116, "grad_norm": 0.6166847348213196, "learning_rate": 0.0005, "loss": 1.6911479234695435, "step": 10970 }, { "epoch": 0.8648797792668506, "grad_norm": 0.6457399725914001, "learning_rate": 0.0005, "loss": 2.278834581375122, "step": 10971 }, { "epoch": 0.8649586125344896, "grad_norm": 0.6278932690620422, "learning_rate": 0.0005, "loss": 1.9229365587234497, "step": 10972 }, { "epoch": 0.8650374458021285, "grad_norm": 0.5499104857444763, "learning_rate": 0.0005, "loss": 1.6295270919799805, "step": 10973 }, { "epoch": 0.8651162790697674, "grad_norm": 0.539484441280365, "learning_rate": 0.0005, "loss": 1.5085382461547852, "step": 10974 }, { "epoch": 0.8651951123374064, "grad_norm": 0.7026931643486023, "learning_rate": 0.0005, "loss": 1.5602622032165527, "step": 10975 }, { "epoch": 0.8652739456050453, "grad_norm": 0.5833602547645569, "learning_rate": 0.0005, "loss": 1.7655010223388672, "step": 10976 }, { "epoch": 0.8653527788726842, "grad_norm": 0.7675616145133972, "learning_rate": 0.0005, "loss": 2.0987982749938965, "step": 10977 }, { "epoch": 0.8654316121403233, "grad_norm": 0.7075190544128418, "learning_rate": 0.0005, "loss": 1.7824631929397583, "step": 10978 }, { "epoch": 0.8655104454079622, "grad_norm": 0.7141557335853577, "learning_rate": 0.0005, "loss": 1.7815791368484497, "step": 10979 }, { "epoch": 0.8655892786756011, "grad_norm": 0.6067956686019897, "learning_rate": 0.0005, "loss": 2.050457715988159, "step": 10980 }, { "epoch": 0.86566811194324, "grad_norm": 0.5421156287193298, "learning_rate": 0.0005, "loss": 1.6721632480621338, "step": 10981 }, { "epoch": 0.865746945210879, "grad_norm": 0.7460564374923706, "learning_rate": 0.0005, "loss": 1.983759880065918, "step": 10982 }, { "epoch": 0.8658257784785179, "grad_norm": 0.6603270769119263, "learning_rate": 0.0005, "loss": 1.7947711944580078, "step": 10983 }, { "epoch": 0.8659046117461568, "grad_norm": 1.0968016386032104, "learning_rate": 0.0005, "loss": 1.7244118452072144, "step": 10984 }, { "epoch": 0.8659834450137959, "grad_norm": 0.5427020788192749, "learning_rate": 0.0005, "loss": 1.7236127853393555, "step": 10985 }, { "epoch": 0.8660622782814348, "grad_norm": 0.5934039354324341, "learning_rate": 0.0005, "loss": 1.9521124362945557, "step": 10986 }, { "epoch": 0.8661411115490737, "grad_norm": 0.5909937620162964, "learning_rate": 0.0005, "loss": 1.601130723953247, "step": 10987 }, { "epoch": 0.8662199448167126, "grad_norm": 0.6769326329231262, "learning_rate": 0.0005, "loss": 1.7908909320831299, "step": 10988 }, { "epoch": 0.8662987780843516, "grad_norm": 0.6104797124862671, "learning_rate": 0.0005, "loss": 1.9028395414352417, "step": 10989 }, { "epoch": 0.8663776113519905, "grad_norm": 0.5150168538093567, "learning_rate": 0.0005, "loss": 1.8460986614227295, "step": 10990 }, { "epoch": 0.8664564446196295, "grad_norm": 0.49584856629371643, "learning_rate": 0.0005, "loss": 1.55316162109375, "step": 10991 }, { "epoch": 0.8665352778872685, "grad_norm": 0.7430627346038818, "learning_rate": 0.0005, "loss": 1.8243038654327393, "step": 10992 }, { "epoch": 0.8666141111549074, "grad_norm": 0.5588799118995667, "learning_rate": 0.0005, "loss": 1.7484922409057617, "step": 10993 }, { "epoch": 0.8666929444225463, "grad_norm": 0.656844437122345, "learning_rate": 0.0005, "loss": 1.6170886754989624, "step": 10994 }, { "epoch": 0.8667717776901852, "grad_norm": 0.8011026382446289, "learning_rate": 0.0005, "loss": 1.663285493850708, "step": 10995 }, { "epoch": 0.8668506109578242, "grad_norm": 0.6335368156433105, "learning_rate": 0.0005, "loss": 1.9726486206054688, "step": 10996 }, { "epoch": 0.8669294442254631, "grad_norm": 0.9218173623085022, "learning_rate": 0.0005, "loss": 1.9276068210601807, "step": 10997 }, { "epoch": 0.867008277493102, "grad_norm": 0.6681466102600098, "learning_rate": 0.0005, "loss": 1.8865976333618164, "step": 10998 }, { "epoch": 0.8670871107607411, "grad_norm": 0.7447180151939392, "learning_rate": 0.0005, "loss": 1.6889028549194336, "step": 10999 }, { "epoch": 0.86716594402838, "grad_norm": 0.6021137833595276, "learning_rate": 0.0005, "loss": 1.8171106576919556, "step": 11000 }, { "epoch": 0.8672447772960189, "grad_norm": 0.6213676929473877, "learning_rate": 0.0005, "loss": 1.6229729652404785, "step": 11001 }, { "epoch": 0.8673236105636578, "grad_norm": 0.8790631294250488, "learning_rate": 0.0005, "loss": 1.8094247579574585, "step": 11002 }, { "epoch": 0.8674024438312968, "grad_norm": 0.7026717066764832, "learning_rate": 0.0005, "loss": 1.9037749767303467, "step": 11003 }, { "epoch": 0.8674812770989357, "grad_norm": 0.8344844579696655, "learning_rate": 0.0005, "loss": 1.9263901710510254, "step": 11004 }, { "epoch": 0.8675601103665747, "grad_norm": 0.710562527179718, "learning_rate": 0.0005, "loss": 1.9249051809310913, "step": 11005 }, { "epoch": 0.8676389436342137, "grad_norm": 0.6983678936958313, "learning_rate": 0.0005, "loss": 1.997401237487793, "step": 11006 }, { "epoch": 0.8677177769018526, "grad_norm": 0.5963241457939148, "learning_rate": 0.0005, "loss": 1.9009449481964111, "step": 11007 }, { "epoch": 0.8677966101694915, "grad_norm": 0.5720390677452087, "learning_rate": 0.0005, "loss": 1.8833564519882202, "step": 11008 }, { "epoch": 0.8678754434371305, "grad_norm": 0.47452571988105774, "learning_rate": 0.0005, "loss": 1.6710535287857056, "step": 11009 }, { "epoch": 0.8679542767047694, "grad_norm": 0.760710597038269, "learning_rate": 0.0005, "loss": 1.896140456199646, "step": 11010 }, { "epoch": 0.8680331099724083, "grad_norm": 0.6658653616905212, "learning_rate": 0.0005, "loss": 1.891474962234497, "step": 11011 }, { "epoch": 0.8681119432400473, "grad_norm": 0.6744153499603271, "learning_rate": 0.0005, "loss": 1.9204370975494385, "step": 11012 }, { "epoch": 0.8681907765076863, "grad_norm": 0.601228654384613, "learning_rate": 0.0005, "loss": 1.815791130065918, "step": 11013 }, { "epoch": 0.8682696097753252, "grad_norm": 0.5898275971412659, "learning_rate": 0.0005, "loss": 1.7581136226654053, "step": 11014 }, { "epoch": 0.8683484430429641, "grad_norm": 0.6207999587059021, "learning_rate": 0.0005, "loss": 1.8683056831359863, "step": 11015 }, { "epoch": 0.8684272763106031, "grad_norm": 1.1034741401672363, "learning_rate": 0.0005, "loss": 1.9490141868591309, "step": 11016 }, { "epoch": 0.868506109578242, "grad_norm": 0.7471820116043091, "learning_rate": 0.0005, "loss": 1.9832205772399902, "step": 11017 }, { "epoch": 0.868584942845881, "grad_norm": 1.006019115447998, "learning_rate": 0.0005, "loss": 2.1088356971740723, "step": 11018 }, { "epoch": 0.8686637761135199, "grad_norm": 0.5535745024681091, "learning_rate": 0.0005, "loss": 1.5912342071533203, "step": 11019 }, { "epoch": 0.8687426093811589, "grad_norm": 0.6276733875274658, "learning_rate": 0.0005, "loss": 1.7002143859863281, "step": 11020 }, { "epoch": 0.8688214426487978, "grad_norm": 0.6147332191467285, "learning_rate": 0.0005, "loss": 1.8861534595489502, "step": 11021 }, { "epoch": 0.8689002759164367, "grad_norm": 5.945161819458008, "learning_rate": 0.0005, "loss": 1.8599324226379395, "step": 11022 }, { "epoch": 0.8689791091840757, "grad_norm": 0.5874007344245911, "learning_rate": 0.0005, "loss": 1.5366523265838623, "step": 11023 }, { "epoch": 0.8690579424517146, "grad_norm": 0.5270107984542847, "learning_rate": 0.0005, "loss": 1.7623741626739502, "step": 11024 }, { "epoch": 0.8691367757193535, "grad_norm": 0.7293107509613037, "learning_rate": 0.0005, "loss": 2.0383944511413574, "step": 11025 }, { "epoch": 0.8692156089869925, "grad_norm": 0.5180811882019043, "learning_rate": 0.0005, "loss": 1.7736331224441528, "step": 11026 }, { "epoch": 0.8692944422546315, "grad_norm": 0.8423336148262024, "learning_rate": 0.0005, "loss": 2.057938814163208, "step": 11027 }, { "epoch": 0.8693732755222704, "grad_norm": 0.6722580790519714, "learning_rate": 0.0005, "loss": 1.768612027168274, "step": 11028 }, { "epoch": 0.8694521087899093, "grad_norm": 0.6550602316856384, "learning_rate": 0.0005, "loss": 1.8822290897369385, "step": 11029 }, { "epoch": 0.8695309420575483, "grad_norm": 0.750981330871582, "learning_rate": 0.0005, "loss": 1.9581100940704346, "step": 11030 }, { "epoch": 0.8696097753251872, "grad_norm": 0.5944566130638123, "learning_rate": 0.0005, "loss": 1.8525116443634033, "step": 11031 }, { "epoch": 0.8696886085928262, "grad_norm": 0.5862103700637817, "learning_rate": 0.0005, "loss": 1.7696205377578735, "step": 11032 }, { "epoch": 0.8697674418604651, "grad_norm": 0.6883090734481812, "learning_rate": 0.0005, "loss": 1.7113127708435059, "step": 11033 }, { "epoch": 0.8698462751281041, "grad_norm": 0.9029700756072998, "learning_rate": 0.0005, "loss": 1.9418530464172363, "step": 11034 }, { "epoch": 0.869925108395743, "grad_norm": 0.7856699228286743, "learning_rate": 0.0005, "loss": 2.035781145095825, "step": 11035 }, { "epoch": 0.8700039416633819, "grad_norm": 0.567001223564148, "learning_rate": 0.0005, "loss": 1.6548599004745483, "step": 11036 }, { "epoch": 0.8700827749310209, "grad_norm": 0.5468209385871887, "learning_rate": 0.0005, "loss": 1.55702543258667, "step": 11037 }, { "epoch": 0.8701616081986598, "grad_norm": 0.582763671875, "learning_rate": 0.0005, "loss": 1.7694592475891113, "step": 11038 }, { "epoch": 0.8702404414662988, "grad_norm": 0.6643790602684021, "learning_rate": 0.0005, "loss": 1.5484986305236816, "step": 11039 }, { "epoch": 0.8703192747339378, "grad_norm": 0.6800456047058105, "learning_rate": 0.0005, "loss": 2.0862061977386475, "step": 11040 }, { "epoch": 0.8703981080015767, "grad_norm": 0.7076618075370789, "learning_rate": 0.0005, "loss": 2.059699535369873, "step": 11041 }, { "epoch": 0.8704769412692156, "grad_norm": 0.7144508361816406, "learning_rate": 0.0005, "loss": 1.7707250118255615, "step": 11042 }, { "epoch": 0.8705557745368545, "grad_norm": 0.7315049767494202, "learning_rate": 0.0005, "loss": 1.8088465929031372, "step": 11043 }, { "epoch": 0.8706346078044935, "grad_norm": 0.7697799205780029, "learning_rate": 0.0005, "loss": 1.9384963512420654, "step": 11044 }, { "epoch": 0.8707134410721324, "grad_norm": 0.6139425039291382, "learning_rate": 0.0005, "loss": 1.9255948066711426, "step": 11045 }, { "epoch": 0.8707922743397714, "grad_norm": 0.7339931726455688, "learning_rate": 0.0005, "loss": 1.9042491912841797, "step": 11046 }, { "epoch": 0.8708711076074104, "grad_norm": 0.6077285408973694, "learning_rate": 0.0005, "loss": 1.810038685798645, "step": 11047 }, { "epoch": 0.8709499408750493, "grad_norm": 0.7343231439590454, "learning_rate": 0.0005, "loss": 1.6919724941253662, "step": 11048 }, { "epoch": 0.8710287741426882, "grad_norm": 0.5799630284309387, "learning_rate": 0.0005, "loss": 1.785384178161621, "step": 11049 }, { "epoch": 0.8711076074103271, "grad_norm": 0.5249469876289368, "learning_rate": 0.0005, "loss": 1.5782530307769775, "step": 11050 }, { "epoch": 0.8711864406779661, "grad_norm": 0.5627151727676392, "learning_rate": 0.0005, "loss": 1.7016854286193848, "step": 11051 }, { "epoch": 0.871265273945605, "grad_norm": 0.5386794805526733, "learning_rate": 0.0005, "loss": 1.9256621599197388, "step": 11052 }, { "epoch": 0.871344107213244, "grad_norm": 0.7102028727531433, "learning_rate": 0.0005, "loss": 1.680938959121704, "step": 11053 }, { "epoch": 0.871422940480883, "grad_norm": 0.6560521125793457, "learning_rate": 0.0005, "loss": 1.9102742671966553, "step": 11054 }, { "epoch": 0.8715017737485219, "grad_norm": 2.6257810592651367, "learning_rate": 0.0005, "loss": 1.6669886112213135, "step": 11055 }, { "epoch": 0.8715806070161608, "grad_norm": 0.897358775138855, "learning_rate": 0.0005, "loss": 1.710001826286316, "step": 11056 }, { "epoch": 0.8716594402837997, "grad_norm": 0.6487656235694885, "learning_rate": 0.0005, "loss": 1.727815866470337, "step": 11057 }, { "epoch": 0.8717382735514387, "grad_norm": 0.6999585032463074, "learning_rate": 0.0005, "loss": 1.7534549236297607, "step": 11058 }, { "epoch": 0.8718171068190776, "grad_norm": 0.6897155046463013, "learning_rate": 0.0005, "loss": 2.132730007171631, "step": 11059 }, { "epoch": 0.8718959400867166, "grad_norm": 0.7202165722846985, "learning_rate": 0.0005, "loss": 2.0319769382476807, "step": 11060 }, { "epoch": 0.8719747733543556, "grad_norm": 0.619536280632019, "learning_rate": 0.0005, "loss": 1.8090002536773682, "step": 11061 }, { "epoch": 0.8720536066219945, "grad_norm": 0.5550589561462402, "learning_rate": 0.0005, "loss": 2.04553484916687, "step": 11062 }, { "epoch": 0.8721324398896334, "grad_norm": 0.6047603487968445, "learning_rate": 0.0005, "loss": 2.1939001083374023, "step": 11063 }, { "epoch": 0.8722112731572723, "grad_norm": 0.5718801021575928, "learning_rate": 0.0005, "loss": 1.899505615234375, "step": 11064 }, { "epoch": 0.8722901064249113, "grad_norm": 0.9570212364196777, "learning_rate": 0.0005, "loss": 1.8671284914016724, "step": 11065 }, { "epoch": 0.8723689396925502, "grad_norm": 0.9388786554336548, "learning_rate": 0.0005, "loss": 1.7550652027130127, "step": 11066 }, { "epoch": 0.8724477729601892, "grad_norm": 0.687267005443573, "learning_rate": 0.0005, "loss": 2.0206832885742188, "step": 11067 }, { "epoch": 0.8725266062278282, "grad_norm": 0.7604491114616394, "learning_rate": 0.0005, "loss": 1.9202828407287598, "step": 11068 }, { "epoch": 0.8726054394954671, "grad_norm": 0.8010371923446655, "learning_rate": 0.0005, "loss": 1.7450436353683472, "step": 11069 }, { "epoch": 0.872684272763106, "grad_norm": 0.5589393973350525, "learning_rate": 0.0005, "loss": 1.701600432395935, "step": 11070 }, { "epoch": 0.872763106030745, "grad_norm": 0.6756139397621155, "learning_rate": 0.0005, "loss": 1.9250545501708984, "step": 11071 }, { "epoch": 0.8728419392983839, "grad_norm": 0.5074424147605896, "learning_rate": 0.0005, "loss": 1.6073273420333862, "step": 11072 }, { "epoch": 0.8729207725660229, "grad_norm": 0.6608985662460327, "learning_rate": 0.0005, "loss": 1.870753526687622, "step": 11073 }, { "epoch": 0.8729996058336618, "grad_norm": 0.575061023235321, "learning_rate": 0.0005, "loss": 1.7729843854904175, "step": 11074 }, { "epoch": 0.8730784391013008, "grad_norm": 0.8583476543426514, "learning_rate": 0.0005, "loss": 1.8947170972824097, "step": 11075 }, { "epoch": 0.8731572723689397, "grad_norm": 0.5735119581222534, "learning_rate": 0.0005, "loss": 1.9168866872787476, "step": 11076 }, { "epoch": 0.8732361056365786, "grad_norm": 0.8294088840484619, "learning_rate": 0.0005, "loss": 1.8039698600769043, "step": 11077 }, { "epoch": 0.8733149389042176, "grad_norm": 0.6121046543121338, "learning_rate": 0.0005, "loss": 1.6490862369537354, "step": 11078 }, { "epoch": 0.8733937721718565, "grad_norm": 0.7022924423217773, "learning_rate": 0.0005, "loss": 1.7820446491241455, "step": 11079 }, { "epoch": 0.8734726054394955, "grad_norm": 0.6776008605957031, "learning_rate": 0.0005, "loss": 1.967680811882019, "step": 11080 }, { "epoch": 0.8735514387071344, "grad_norm": 0.6089568734169006, "learning_rate": 0.0005, "loss": 1.8564027547836304, "step": 11081 }, { "epoch": 0.8736302719747734, "grad_norm": 0.5106855630874634, "learning_rate": 0.0005, "loss": 1.7721035480499268, "step": 11082 }, { "epoch": 0.8737091052424123, "grad_norm": 0.5431917309761047, "learning_rate": 0.0005, "loss": 1.9464975595474243, "step": 11083 }, { "epoch": 0.8737879385100512, "grad_norm": 0.6040152907371521, "learning_rate": 0.0005, "loss": 1.925213098526001, "step": 11084 }, { "epoch": 0.8738667717776902, "grad_norm": 0.5601497292518616, "learning_rate": 0.0005, "loss": 1.4376821517944336, "step": 11085 }, { "epoch": 0.8739456050453291, "grad_norm": 0.5578122735023499, "learning_rate": 0.0005, "loss": 1.6099791526794434, "step": 11086 }, { "epoch": 0.874024438312968, "grad_norm": 0.7405868172645569, "learning_rate": 0.0005, "loss": 1.8191261291503906, "step": 11087 }, { "epoch": 0.874103271580607, "grad_norm": 0.5855607390403748, "learning_rate": 0.0005, "loss": 1.7084095478057861, "step": 11088 }, { "epoch": 0.874182104848246, "grad_norm": 0.7127414345741272, "learning_rate": 0.0005, "loss": 1.9660102128982544, "step": 11089 }, { "epoch": 0.8742609381158849, "grad_norm": 0.6639841794967651, "learning_rate": 0.0005, "loss": 1.6690013408660889, "step": 11090 }, { "epoch": 0.8743397713835238, "grad_norm": 0.5722383856773376, "learning_rate": 0.0005, "loss": 1.773115873336792, "step": 11091 }, { "epoch": 0.8744186046511628, "grad_norm": 0.6315729022026062, "learning_rate": 0.0005, "loss": 1.5389320850372314, "step": 11092 }, { "epoch": 0.8744974379188017, "grad_norm": 0.6338496208190918, "learning_rate": 0.0005, "loss": 2.0393643379211426, "step": 11093 }, { "epoch": 0.8745762711864407, "grad_norm": 0.685623824596405, "learning_rate": 0.0005, "loss": 1.9090256690979004, "step": 11094 }, { "epoch": 0.8746551044540796, "grad_norm": 0.6931763291358948, "learning_rate": 0.0005, "loss": 1.8249599933624268, "step": 11095 }, { "epoch": 0.8747339377217186, "grad_norm": 0.6960217356681824, "learning_rate": 0.0005, "loss": 2.2430622577667236, "step": 11096 }, { "epoch": 0.8748127709893575, "grad_norm": 0.7465780973434448, "learning_rate": 0.0005, "loss": 1.7528183460235596, "step": 11097 }, { "epoch": 0.8748916042569964, "grad_norm": 0.5874951481819153, "learning_rate": 0.0005, "loss": 1.8125361204147339, "step": 11098 }, { "epoch": 0.8749704375246354, "grad_norm": 0.758129358291626, "learning_rate": 0.0005, "loss": 2.1340622901916504, "step": 11099 }, { "epoch": 0.8750492707922743, "grad_norm": 0.5762280821800232, "learning_rate": 0.0005, "loss": 1.691664218902588, "step": 11100 }, { "epoch": 0.8751281040599133, "grad_norm": 0.5799843668937683, "learning_rate": 0.0005, "loss": 1.6719080209732056, "step": 11101 }, { "epoch": 0.8752069373275523, "grad_norm": 0.6407623887062073, "learning_rate": 0.0005, "loss": 1.665661096572876, "step": 11102 }, { "epoch": 0.8752857705951912, "grad_norm": 0.6115953326225281, "learning_rate": 0.0005, "loss": 1.8834573030471802, "step": 11103 }, { "epoch": 0.8753646038628301, "grad_norm": 0.7439048290252686, "learning_rate": 0.0005, "loss": 1.9076474905014038, "step": 11104 }, { "epoch": 0.875443437130469, "grad_norm": 0.5907498002052307, "learning_rate": 0.0005, "loss": 1.8528673648834229, "step": 11105 }, { "epoch": 0.875522270398108, "grad_norm": 0.6113381385803223, "learning_rate": 0.0005, "loss": 1.685685157775879, "step": 11106 }, { "epoch": 0.875601103665747, "grad_norm": 0.6627318859100342, "learning_rate": 0.0005, "loss": 2.0184340476989746, "step": 11107 }, { "epoch": 0.8756799369333859, "grad_norm": 0.6973775029182434, "learning_rate": 0.0005, "loss": 1.9341686964035034, "step": 11108 }, { "epoch": 0.8757587702010249, "grad_norm": 0.7734078168869019, "learning_rate": 0.0005, "loss": 2.1091532707214355, "step": 11109 }, { "epoch": 0.8758376034686638, "grad_norm": 0.611633837223053, "learning_rate": 0.0005, "loss": 1.8277946710586548, "step": 11110 }, { "epoch": 0.8759164367363027, "grad_norm": 0.5357246398925781, "learning_rate": 0.0005, "loss": 1.7683231830596924, "step": 11111 }, { "epoch": 0.8759952700039416, "grad_norm": 0.7353929281234741, "learning_rate": 0.0005, "loss": 1.8773359060287476, "step": 11112 }, { "epoch": 0.8760741032715806, "grad_norm": 0.5565480589866638, "learning_rate": 0.0005, "loss": 1.5165011882781982, "step": 11113 }, { "epoch": 0.8761529365392196, "grad_norm": 0.6387001276016235, "learning_rate": 0.0005, "loss": 1.5807485580444336, "step": 11114 }, { "epoch": 0.8762317698068585, "grad_norm": 0.6791170835494995, "learning_rate": 0.0005, "loss": 1.8594961166381836, "step": 11115 }, { "epoch": 0.8763106030744975, "grad_norm": 0.665382981300354, "learning_rate": 0.0005, "loss": 1.6850252151489258, "step": 11116 }, { "epoch": 0.8763894363421364, "grad_norm": 0.7419476509094238, "learning_rate": 0.0005, "loss": 1.7763984203338623, "step": 11117 }, { "epoch": 0.8764682696097753, "grad_norm": 0.6269413828849792, "learning_rate": 0.0005, "loss": 1.739573359489441, "step": 11118 }, { "epoch": 0.8765471028774142, "grad_norm": 0.716291069984436, "learning_rate": 0.0005, "loss": 1.970030426979065, "step": 11119 }, { "epoch": 0.8766259361450532, "grad_norm": 0.6616815328598022, "learning_rate": 0.0005, "loss": 1.7774931192398071, "step": 11120 }, { "epoch": 0.8767047694126922, "grad_norm": 0.7651381492614746, "learning_rate": 0.0005, "loss": 1.8479869365692139, "step": 11121 }, { "epoch": 0.8767836026803311, "grad_norm": 0.6749985814094543, "learning_rate": 0.0005, "loss": 2.0319271087646484, "step": 11122 }, { "epoch": 0.8768624359479701, "grad_norm": 0.5844978094100952, "learning_rate": 0.0005, "loss": 1.8666374683380127, "step": 11123 }, { "epoch": 0.876941269215609, "grad_norm": 0.7997537851333618, "learning_rate": 0.0005, "loss": 1.442497968673706, "step": 11124 }, { "epoch": 0.8770201024832479, "grad_norm": 0.6427596211433411, "learning_rate": 0.0005, "loss": 1.8354721069335938, "step": 11125 }, { "epoch": 0.8770989357508868, "grad_norm": 0.5610995292663574, "learning_rate": 0.0005, "loss": 1.8806791305541992, "step": 11126 }, { "epoch": 0.8771777690185258, "grad_norm": 0.5659012794494629, "learning_rate": 0.0005, "loss": 1.9922568798065186, "step": 11127 }, { "epoch": 0.8772566022861648, "grad_norm": 0.7842775583267212, "learning_rate": 0.0005, "loss": 2.1415488719940186, "step": 11128 }, { "epoch": 0.8773354355538037, "grad_norm": 0.7462314367294312, "learning_rate": 0.0005, "loss": 1.9493217468261719, "step": 11129 }, { "epoch": 0.8774142688214427, "grad_norm": 0.7817524671554565, "learning_rate": 0.0005, "loss": 2.1071090698242188, "step": 11130 }, { "epoch": 0.8774931020890816, "grad_norm": 0.6885939240455627, "learning_rate": 0.0005, "loss": 2.021347761154175, "step": 11131 }, { "epoch": 0.8775719353567205, "grad_norm": 0.6942852735519409, "learning_rate": 0.0005, "loss": 1.999716877937317, "step": 11132 }, { "epoch": 0.8776507686243595, "grad_norm": 0.7723423838615417, "learning_rate": 0.0005, "loss": 1.9337021112442017, "step": 11133 }, { "epoch": 0.8777296018919984, "grad_norm": 0.5710580945014954, "learning_rate": 0.0005, "loss": 1.8827853202819824, "step": 11134 }, { "epoch": 0.8778084351596374, "grad_norm": 0.7679268717765808, "learning_rate": 0.0005, "loss": 2.118459463119507, "step": 11135 }, { "epoch": 0.8778872684272763, "grad_norm": 0.6227681636810303, "learning_rate": 0.0005, "loss": 1.9960589408874512, "step": 11136 }, { "epoch": 0.8779661016949153, "grad_norm": 0.5049387216567993, "learning_rate": 0.0005, "loss": 1.7089269161224365, "step": 11137 }, { "epoch": 0.8780449349625542, "grad_norm": 0.4835253655910492, "learning_rate": 0.0005, "loss": 1.6492167711257935, "step": 11138 }, { "epoch": 0.8781237682301931, "grad_norm": 0.5122486352920532, "learning_rate": 0.0005, "loss": 1.5245835781097412, "step": 11139 }, { "epoch": 0.8782026014978321, "grad_norm": 0.6520735025405884, "learning_rate": 0.0005, "loss": 1.7251592874526978, "step": 11140 }, { "epoch": 0.878281434765471, "grad_norm": 0.6721542477607727, "learning_rate": 0.0005, "loss": 1.944681167602539, "step": 11141 }, { "epoch": 0.87836026803311, "grad_norm": 0.6136579513549805, "learning_rate": 0.0005, "loss": 1.9159209728240967, "step": 11142 }, { "epoch": 0.8784391013007489, "grad_norm": 0.5681377053260803, "learning_rate": 0.0005, "loss": 1.6430591344833374, "step": 11143 }, { "epoch": 0.8785179345683879, "grad_norm": 0.6108636260032654, "learning_rate": 0.0005, "loss": 1.8969879150390625, "step": 11144 }, { "epoch": 0.8785967678360268, "grad_norm": 0.5820021033287048, "learning_rate": 0.0005, "loss": 1.9443557262420654, "step": 11145 }, { "epoch": 0.8786756011036657, "grad_norm": 0.6363447904586792, "learning_rate": 0.0005, "loss": 1.8179314136505127, "step": 11146 }, { "epoch": 0.8787544343713047, "grad_norm": 0.8042029142379761, "learning_rate": 0.0005, "loss": 1.9587289094924927, "step": 11147 }, { "epoch": 0.8788332676389436, "grad_norm": 0.6854690909385681, "learning_rate": 0.0005, "loss": 2.2159740924835205, "step": 11148 }, { "epoch": 0.8789121009065826, "grad_norm": 0.581342339515686, "learning_rate": 0.0005, "loss": 1.7247101068496704, "step": 11149 }, { "epoch": 0.8789909341742215, "grad_norm": 0.5138601064682007, "learning_rate": 0.0005, "loss": 1.690562129020691, "step": 11150 }, { "epoch": 0.8790697674418605, "grad_norm": 0.602813184261322, "learning_rate": 0.0005, "loss": 1.863248348236084, "step": 11151 }, { "epoch": 0.8791486007094994, "grad_norm": 0.6087700724601746, "learning_rate": 0.0005, "loss": 1.9924780130386353, "step": 11152 }, { "epoch": 0.8792274339771383, "grad_norm": 1.0472992658615112, "learning_rate": 0.0005, "loss": 1.8893963098526, "step": 11153 }, { "epoch": 0.8793062672447773, "grad_norm": 0.6055338382720947, "learning_rate": 0.0005, "loss": 1.8446053266525269, "step": 11154 }, { "epoch": 0.8793851005124163, "grad_norm": 0.7229956984519958, "learning_rate": 0.0005, "loss": 2.1582846641540527, "step": 11155 }, { "epoch": 0.8794639337800552, "grad_norm": 0.6783524751663208, "learning_rate": 0.0005, "loss": 1.7341816425323486, "step": 11156 }, { "epoch": 0.8795427670476941, "grad_norm": 0.706221878528595, "learning_rate": 0.0005, "loss": 1.9889174699783325, "step": 11157 }, { "epoch": 0.8796216003153331, "grad_norm": 0.5167839527130127, "learning_rate": 0.0005, "loss": 1.8876309394836426, "step": 11158 }, { "epoch": 0.879700433582972, "grad_norm": 0.5471898317337036, "learning_rate": 0.0005, "loss": 1.7964240312576294, "step": 11159 }, { "epoch": 0.8797792668506109, "grad_norm": 0.7587172389030457, "learning_rate": 0.0005, "loss": 1.9528298377990723, "step": 11160 }, { "epoch": 0.8798581001182499, "grad_norm": 0.5064942240715027, "learning_rate": 0.0005, "loss": 1.7976160049438477, "step": 11161 }, { "epoch": 0.8799369333858889, "grad_norm": 0.6153081059455872, "learning_rate": 0.0005, "loss": 2.2342169284820557, "step": 11162 }, { "epoch": 0.8800157666535278, "grad_norm": 0.6201756000518799, "learning_rate": 0.0005, "loss": 2.014646053314209, "step": 11163 }, { "epoch": 0.8800945999211668, "grad_norm": 0.5614802241325378, "learning_rate": 0.0005, "loss": 1.98654305934906, "step": 11164 }, { "epoch": 0.8801734331888057, "grad_norm": 0.5571019649505615, "learning_rate": 0.0005, "loss": 1.6207669973373413, "step": 11165 }, { "epoch": 0.8802522664564446, "grad_norm": 1.0244972705841064, "learning_rate": 0.0005, "loss": 1.7994847297668457, "step": 11166 }, { "epoch": 0.8803310997240835, "grad_norm": 0.6811331510543823, "learning_rate": 0.0005, "loss": 1.875058889389038, "step": 11167 }, { "epoch": 0.8804099329917225, "grad_norm": 0.7045281529426575, "learning_rate": 0.0005, "loss": 1.800302505493164, "step": 11168 }, { "epoch": 0.8804887662593615, "grad_norm": 0.5884195566177368, "learning_rate": 0.0005, "loss": 1.708937168121338, "step": 11169 }, { "epoch": 0.8805675995270004, "grad_norm": 0.7255246639251709, "learning_rate": 0.0005, "loss": 1.8463984727859497, "step": 11170 }, { "epoch": 0.8806464327946394, "grad_norm": 0.6281893253326416, "learning_rate": 0.0005, "loss": 1.794570803642273, "step": 11171 }, { "epoch": 0.8807252660622783, "grad_norm": 0.698771059513092, "learning_rate": 0.0005, "loss": 1.8178192377090454, "step": 11172 }, { "epoch": 0.8808040993299172, "grad_norm": 0.915637195110321, "learning_rate": 0.0005, "loss": 1.9413634538650513, "step": 11173 }, { "epoch": 0.8808829325975561, "grad_norm": 0.6601623296737671, "learning_rate": 0.0005, "loss": 1.6887205839157104, "step": 11174 }, { "epoch": 0.8809617658651951, "grad_norm": 0.6671756505966187, "learning_rate": 0.0005, "loss": 1.7545913457870483, "step": 11175 }, { "epoch": 0.8810405991328341, "grad_norm": 0.7565961480140686, "learning_rate": 0.0005, "loss": 1.4849517345428467, "step": 11176 }, { "epoch": 0.881119432400473, "grad_norm": 0.7183915376663208, "learning_rate": 0.0005, "loss": 1.8951120376586914, "step": 11177 }, { "epoch": 0.881198265668112, "grad_norm": 0.705604076385498, "learning_rate": 0.0005, "loss": 1.8608226776123047, "step": 11178 }, { "epoch": 0.8812770989357509, "grad_norm": 0.5370259284973145, "learning_rate": 0.0005, "loss": 1.6758241653442383, "step": 11179 }, { "epoch": 0.8813559322033898, "grad_norm": 0.7361024618148804, "learning_rate": 0.0005, "loss": 2.2071614265441895, "step": 11180 }, { "epoch": 0.8814347654710287, "grad_norm": 0.7689661383628845, "learning_rate": 0.0005, "loss": 1.9172375202178955, "step": 11181 }, { "epoch": 0.8815135987386677, "grad_norm": 0.5516561269760132, "learning_rate": 0.0005, "loss": 1.7376493215560913, "step": 11182 }, { "epoch": 0.8815924320063067, "grad_norm": 0.7632742524147034, "learning_rate": 0.0005, "loss": 2.0821433067321777, "step": 11183 }, { "epoch": 0.8816712652739456, "grad_norm": 0.6993881464004517, "learning_rate": 0.0005, "loss": 1.7541282176971436, "step": 11184 }, { "epoch": 0.8817500985415846, "grad_norm": 0.5657574534416199, "learning_rate": 0.0005, "loss": 1.8476989269256592, "step": 11185 }, { "epoch": 0.8818289318092235, "grad_norm": 0.6601195931434631, "learning_rate": 0.0005, "loss": 1.89180588722229, "step": 11186 }, { "epoch": 0.8819077650768624, "grad_norm": 0.5372526049613953, "learning_rate": 0.0005, "loss": 1.8078196048736572, "step": 11187 }, { "epoch": 0.8819865983445013, "grad_norm": 0.5776539444923401, "learning_rate": 0.0005, "loss": 1.7066597938537598, "step": 11188 }, { "epoch": 0.8820654316121403, "grad_norm": 0.6790067553520203, "learning_rate": 0.0005, "loss": 1.6374692916870117, "step": 11189 }, { "epoch": 0.8821442648797793, "grad_norm": 0.5727586150169373, "learning_rate": 0.0005, "loss": 1.612369179725647, "step": 11190 }, { "epoch": 0.8822230981474182, "grad_norm": 0.4524347484111786, "learning_rate": 0.0005, "loss": 1.6586570739746094, "step": 11191 }, { "epoch": 0.8823019314150572, "grad_norm": 0.7889916896820068, "learning_rate": 0.0005, "loss": 2.121034622192383, "step": 11192 }, { "epoch": 0.8823807646826961, "grad_norm": 0.5787528157234192, "learning_rate": 0.0005, "loss": 1.6788229942321777, "step": 11193 }, { "epoch": 0.882459597950335, "grad_norm": 0.9295852780342102, "learning_rate": 0.0005, "loss": 2.0947465896606445, "step": 11194 }, { "epoch": 0.882538431217974, "grad_norm": 0.8182554244995117, "learning_rate": 0.0005, "loss": 2.088449478149414, "step": 11195 }, { "epoch": 0.882617264485613, "grad_norm": 0.6228364109992981, "learning_rate": 0.0005, "loss": 1.7943763732910156, "step": 11196 }, { "epoch": 0.8826960977532519, "grad_norm": 0.8085106611251831, "learning_rate": 0.0005, "loss": 1.951704502105713, "step": 11197 }, { "epoch": 0.8827749310208908, "grad_norm": 0.5163233876228333, "learning_rate": 0.0005, "loss": 1.7349345684051514, "step": 11198 }, { "epoch": 0.8828537642885298, "grad_norm": 0.543513298034668, "learning_rate": 0.0005, "loss": 1.6020678281784058, "step": 11199 }, { "epoch": 0.8829325975561687, "grad_norm": 0.561224639415741, "learning_rate": 0.0005, "loss": 1.5610225200653076, "step": 11200 }, { "epoch": 0.8830114308238076, "grad_norm": 0.5600796937942505, "learning_rate": 0.0005, "loss": 1.691896677017212, "step": 11201 }, { "epoch": 0.8830902640914466, "grad_norm": 0.6012566685676575, "learning_rate": 0.0005, "loss": 1.8634297847747803, "step": 11202 }, { "epoch": 0.8831690973590856, "grad_norm": 0.5495597124099731, "learning_rate": 0.0005, "loss": 1.8148257732391357, "step": 11203 }, { "epoch": 0.8832479306267245, "grad_norm": 0.9372612833976746, "learning_rate": 0.0005, "loss": 2.1180338859558105, "step": 11204 }, { "epoch": 0.8833267638943634, "grad_norm": 0.6898351311683655, "learning_rate": 0.0005, "loss": 1.8486382961273193, "step": 11205 }, { "epoch": 0.8834055971620024, "grad_norm": 0.6473827362060547, "learning_rate": 0.0005, "loss": 1.9963551759719849, "step": 11206 }, { "epoch": 0.8834844304296413, "grad_norm": 0.7056037187576294, "learning_rate": 0.0005, "loss": 1.6996538639068604, "step": 11207 }, { "epoch": 0.8835632636972802, "grad_norm": 0.7234231233596802, "learning_rate": 0.0005, "loss": 2.0971152782440186, "step": 11208 }, { "epoch": 0.8836420969649192, "grad_norm": 0.7601068019866943, "learning_rate": 0.0005, "loss": 1.9735989570617676, "step": 11209 }, { "epoch": 0.8837209302325582, "grad_norm": 0.7463284134864807, "learning_rate": 0.0005, "loss": 1.966918706893921, "step": 11210 }, { "epoch": 0.8837997635001971, "grad_norm": 0.586125910282135, "learning_rate": 0.0005, "loss": 2.0977115631103516, "step": 11211 }, { "epoch": 0.883878596767836, "grad_norm": 0.7676091194152832, "learning_rate": 0.0005, "loss": 1.854446530342102, "step": 11212 }, { "epoch": 0.883957430035475, "grad_norm": 0.6822498440742493, "learning_rate": 0.0005, "loss": 2.159636974334717, "step": 11213 }, { "epoch": 0.8840362633031139, "grad_norm": 0.7007235288619995, "learning_rate": 0.0005, "loss": 1.7354910373687744, "step": 11214 }, { "epoch": 0.8841150965707528, "grad_norm": 0.7595242857933044, "learning_rate": 0.0005, "loss": 1.7511661052703857, "step": 11215 }, { "epoch": 0.8841939298383918, "grad_norm": 0.5290014743804932, "learning_rate": 0.0005, "loss": 1.8743383884429932, "step": 11216 }, { "epoch": 0.8842727631060308, "grad_norm": 0.8676373958587646, "learning_rate": 0.0005, "loss": 1.7720258235931396, "step": 11217 }, { "epoch": 0.8843515963736697, "grad_norm": 0.5727275013923645, "learning_rate": 0.0005, "loss": 1.9738717079162598, "step": 11218 }, { "epoch": 0.8844304296413086, "grad_norm": 0.6074027419090271, "learning_rate": 0.0005, "loss": 1.847098469734192, "step": 11219 }, { "epoch": 0.8845092629089476, "grad_norm": 0.6338500380516052, "learning_rate": 0.0005, "loss": 1.7934578657150269, "step": 11220 }, { "epoch": 0.8845880961765865, "grad_norm": 0.8753687739372253, "learning_rate": 0.0005, "loss": 2.11602520942688, "step": 11221 }, { "epoch": 0.8846669294442254, "grad_norm": 0.6425017714500427, "learning_rate": 0.0005, "loss": 1.652822494506836, "step": 11222 }, { "epoch": 0.8847457627118644, "grad_norm": 0.722257673740387, "learning_rate": 0.0005, "loss": 1.9254834651947021, "step": 11223 }, { "epoch": 0.8848245959795034, "grad_norm": 0.6959419846534729, "learning_rate": 0.0005, "loss": 1.8657777309417725, "step": 11224 }, { "epoch": 0.8849034292471423, "grad_norm": 0.7643623352050781, "learning_rate": 0.0005, "loss": 1.6931730508804321, "step": 11225 }, { "epoch": 0.8849822625147813, "grad_norm": 0.7856998443603516, "learning_rate": 0.0005, "loss": 1.9032583236694336, "step": 11226 }, { "epoch": 0.8850610957824202, "grad_norm": 0.7419236302375793, "learning_rate": 0.0005, "loss": 1.8847408294677734, "step": 11227 }, { "epoch": 0.8851399290500591, "grad_norm": 0.6945250630378723, "learning_rate": 0.0005, "loss": 1.8137459754943848, "step": 11228 }, { "epoch": 0.885218762317698, "grad_norm": 0.6896947026252747, "learning_rate": 0.0005, "loss": 1.6312369108200073, "step": 11229 }, { "epoch": 0.885297595585337, "grad_norm": 0.7834482789039612, "learning_rate": 0.0005, "loss": 1.9266221523284912, "step": 11230 }, { "epoch": 0.885376428852976, "grad_norm": 0.5858837962150574, "learning_rate": 0.0005, "loss": 1.5906894207000732, "step": 11231 }, { "epoch": 0.8854552621206149, "grad_norm": 0.6217517256736755, "learning_rate": 0.0005, "loss": 1.8477799892425537, "step": 11232 }, { "epoch": 0.8855340953882539, "grad_norm": 0.5472264289855957, "learning_rate": 0.0005, "loss": 2.0286765098571777, "step": 11233 }, { "epoch": 0.8856129286558928, "grad_norm": 0.9515649676322937, "learning_rate": 0.0005, "loss": 2.161054849624634, "step": 11234 }, { "epoch": 0.8856917619235317, "grad_norm": 0.7768858671188354, "learning_rate": 0.0005, "loss": 1.9356054067611694, "step": 11235 }, { "epoch": 0.8857705951911706, "grad_norm": 0.7896724939346313, "learning_rate": 0.0005, "loss": 1.7424020767211914, "step": 11236 }, { "epoch": 0.8858494284588097, "grad_norm": 0.7041078209877014, "learning_rate": 0.0005, "loss": 1.652170181274414, "step": 11237 }, { "epoch": 0.8859282617264486, "grad_norm": 0.6272445321083069, "learning_rate": 0.0005, "loss": 1.7778866291046143, "step": 11238 }, { "epoch": 0.8860070949940875, "grad_norm": 0.5797969698905945, "learning_rate": 0.0005, "loss": 1.8426002264022827, "step": 11239 }, { "epoch": 0.8860859282617265, "grad_norm": 0.8702223896980286, "learning_rate": 0.0005, "loss": 1.6681183576583862, "step": 11240 }, { "epoch": 0.8861647615293654, "grad_norm": 0.8012522459030151, "learning_rate": 0.0005, "loss": 2.06095027923584, "step": 11241 }, { "epoch": 0.8862435947970043, "grad_norm": 0.5775155425071716, "learning_rate": 0.0005, "loss": 1.6729419231414795, "step": 11242 }, { "epoch": 0.8863224280646432, "grad_norm": 0.6675479412078857, "learning_rate": 0.0005, "loss": 1.922729730606079, "step": 11243 }, { "epoch": 0.8864012613322823, "grad_norm": 0.6791771650314331, "learning_rate": 0.0005, "loss": 1.834863543510437, "step": 11244 }, { "epoch": 0.8864800945999212, "grad_norm": 0.6630333662033081, "learning_rate": 0.0005, "loss": 2.0604212284088135, "step": 11245 }, { "epoch": 0.8865589278675601, "grad_norm": 0.7151852250099182, "learning_rate": 0.0005, "loss": 1.680361270904541, "step": 11246 }, { "epoch": 0.8866377611351991, "grad_norm": 0.5919637084007263, "learning_rate": 0.0005, "loss": 1.9658596515655518, "step": 11247 }, { "epoch": 0.886716594402838, "grad_norm": 0.6236903071403503, "learning_rate": 0.0005, "loss": 1.7910454273223877, "step": 11248 }, { "epoch": 0.8867954276704769, "grad_norm": 0.6867380738258362, "learning_rate": 0.0005, "loss": 1.9394607543945312, "step": 11249 }, { "epoch": 0.8868742609381158, "grad_norm": 0.6984199285507202, "learning_rate": 0.0005, "loss": 1.811610221862793, "step": 11250 }, { "epoch": 0.8869530942057549, "grad_norm": 0.6943473815917969, "learning_rate": 0.0005, "loss": 1.960552453994751, "step": 11251 }, { "epoch": 0.8870319274733938, "grad_norm": 0.6361351013183594, "learning_rate": 0.0005, "loss": 1.704775333404541, "step": 11252 }, { "epoch": 0.8871107607410327, "grad_norm": 0.6571155786514282, "learning_rate": 0.0005, "loss": 1.687664270401001, "step": 11253 }, { "epoch": 0.8871895940086717, "grad_norm": 0.8534155488014221, "learning_rate": 0.0005, "loss": 1.5573983192443848, "step": 11254 }, { "epoch": 0.8872684272763106, "grad_norm": 0.6042873859405518, "learning_rate": 0.0005, "loss": 1.5065499544143677, "step": 11255 }, { "epoch": 0.8873472605439495, "grad_norm": 0.8027812838554382, "learning_rate": 0.0005, "loss": 1.9597450494766235, "step": 11256 }, { "epoch": 0.8874260938115885, "grad_norm": 0.5937493443489075, "learning_rate": 0.0005, "loss": 1.8763427734375, "step": 11257 }, { "epoch": 0.8875049270792275, "grad_norm": 0.6195068955421448, "learning_rate": 0.0005, "loss": 1.7051398754119873, "step": 11258 }, { "epoch": 0.8875837603468664, "grad_norm": 0.6484401226043701, "learning_rate": 0.0005, "loss": 1.9324266910552979, "step": 11259 }, { "epoch": 0.8876625936145053, "grad_norm": 0.6931671500205994, "learning_rate": 0.0005, "loss": 1.7776682376861572, "step": 11260 }, { "epoch": 0.8877414268821443, "grad_norm": 0.47758498787879944, "learning_rate": 0.0005, "loss": 1.7640050649642944, "step": 11261 }, { "epoch": 0.8878202601497832, "grad_norm": 0.7989630699157715, "learning_rate": 0.0005, "loss": 2.0930776596069336, "step": 11262 }, { "epoch": 0.8878990934174221, "grad_norm": 0.7149235606193542, "learning_rate": 0.0005, "loss": 1.9548615217208862, "step": 11263 }, { "epoch": 0.8879779266850611, "grad_norm": 2.1805763244628906, "learning_rate": 0.0005, "loss": 2.0738463401794434, "step": 11264 }, { "epoch": 0.8880567599527001, "grad_norm": 0.6172289252281189, "learning_rate": 0.0005, "loss": 1.6902484893798828, "step": 11265 }, { "epoch": 0.888135593220339, "grad_norm": 0.7455790638923645, "learning_rate": 0.0005, "loss": 2.0132033824920654, "step": 11266 }, { "epoch": 0.8882144264879779, "grad_norm": 0.6914110779762268, "learning_rate": 0.0005, "loss": 1.890216588973999, "step": 11267 }, { "epoch": 0.8882932597556169, "grad_norm": 0.6425818800926208, "learning_rate": 0.0005, "loss": 1.832761526107788, "step": 11268 }, { "epoch": 0.8883720930232558, "grad_norm": 0.7715330719947815, "learning_rate": 0.0005, "loss": 1.7727216482162476, "step": 11269 }, { "epoch": 0.8884509262908947, "grad_norm": 0.6407915949821472, "learning_rate": 0.0005, "loss": 1.8104429244995117, "step": 11270 }, { "epoch": 0.8885297595585337, "grad_norm": 0.809935986995697, "learning_rate": 0.0005, "loss": 1.8669835329055786, "step": 11271 }, { "epoch": 0.8886085928261727, "grad_norm": 0.9020791053771973, "learning_rate": 0.0005, "loss": 1.9602749347686768, "step": 11272 }, { "epoch": 0.8886874260938116, "grad_norm": 0.5452032685279846, "learning_rate": 0.0005, "loss": 1.7070512771606445, "step": 11273 }, { "epoch": 0.8887662593614505, "grad_norm": 0.5743933916091919, "learning_rate": 0.0005, "loss": 1.9638144969940186, "step": 11274 }, { "epoch": 0.8888450926290895, "grad_norm": 0.6291778087615967, "learning_rate": 0.0005, "loss": 1.9285411834716797, "step": 11275 }, { "epoch": 0.8889239258967284, "grad_norm": 0.8051819801330566, "learning_rate": 0.0005, "loss": 1.9424799680709839, "step": 11276 }, { "epoch": 0.8890027591643673, "grad_norm": 0.6144729256629944, "learning_rate": 0.0005, "loss": 1.8630170822143555, "step": 11277 }, { "epoch": 0.8890815924320064, "grad_norm": 0.7309758067131042, "learning_rate": 0.0005, "loss": 1.7637479305267334, "step": 11278 }, { "epoch": 0.8891604256996453, "grad_norm": 0.7018097639083862, "learning_rate": 0.0005, "loss": 1.7297520637512207, "step": 11279 }, { "epoch": 0.8892392589672842, "grad_norm": 0.9929654598236084, "learning_rate": 0.0005, "loss": 1.7721842527389526, "step": 11280 }, { "epoch": 0.8893180922349231, "grad_norm": 0.5617247819900513, "learning_rate": 0.0005, "loss": 2.077571153640747, "step": 11281 }, { "epoch": 0.8893969255025621, "grad_norm": 0.6525714993476868, "learning_rate": 0.0005, "loss": 2.132967948913574, "step": 11282 }, { "epoch": 0.889475758770201, "grad_norm": 0.6618795990943909, "learning_rate": 0.0005, "loss": 1.875950574874878, "step": 11283 }, { "epoch": 0.8895545920378399, "grad_norm": 0.6066217422485352, "learning_rate": 0.0005, "loss": 1.7457671165466309, "step": 11284 }, { "epoch": 0.889633425305479, "grad_norm": 0.6549685597419739, "learning_rate": 0.0005, "loss": 1.882882833480835, "step": 11285 }, { "epoch": 0.8897122585731179, "grad_norm": 0.6856088042259216, "learning_rate": 0.0005, "loss": 2.1990609169006348, "step": 11286 }, { "epoch": 0.8897910918407568, "grad_norm": 0.9227926731109619, "learning_rate": 0.0005, "loss": 1.9973974227905273, "step": 11287 }, { "epoch": 0.8898699251083958, "grad_norm": 0.6044324636459351, "learning_rate": 0.0005, "loss": 1.7161920070648193, "step": 11288 }, { "epoch": 0.8899487583760347, "grad_norm": 0.6440989971160889, "learning_rate": 0.0005, "loss": 1.8234069347381592, "step": 11289 }, { "epoch": 0.8900275916436736, "grad_norm": 0.6607451438903809, "learning_rate": 0.0005, "loss": 1.7742903232574463, "step": 11290 }, { "epoch": 0.8901064249113125, "grad_norm": 0.7881723046302795, "learning_rate": 0.0005, "loss": 1.7990410327911377, "step": 11291 }, { "epoch": 0.8901852581789516, "grad_norm": 0.6487064957618713, "learning_rate": 0.0005, "loss": 1.6315650939941406, "step": 11292 }, { "epoch": 0.8902640914465905, "grad_norm": 0.5941152572631836, "learning_rate": 0.0005, "loss": 2.028273582458496, "step": 11293 }, { "epoch": 0.8903429247142294, "grad_norm": 0.7230865955352783, "learning_rate": 0.0005, "loss": 1.6849714517593384, "step": 11294 }, { "epoch": 0.8904217579818684, "grad_norm": 0.5765455961227417, "learning_rate": 0.0005, "loss": 1.8912073373794556, "step": 11295 }, { "epoch": 0.8905005912495073, "grad_norm": 0.6234105825424194, "learning_rate": 0.0005, "loss": 1.847465991973877, "step": 11296 }, { "epoch": 0.8905794245171462, "grad_norm": 0.6846083998680115, "learning_rate": 0.0005, "loss": 2.065925359725952, "step": 11297 }, { "epoch": 0.8906582577847851, "grad_norm": 0.7281930446624756, "learning_rate": 0.0005, "loss": 1.7766191959381104, "step": 11298 }, { "epoch": 0.8907370910524242, "grad_norm": 0.6117414236068726, "learning_rate": 0.0005, "loss": 1.7526111602783203, "step": 11299 }, { "epoch": 0.8908159243200631, "grad_norm": 0.6072653532028198, "learning_rate": 0.0005, "loss": 1.6839323043823242, "step": 11300 }, { "epoch": 0.890894757587702, "grad_norm": 0.5399062633514404, "learning_rate": 0.0005, "loss": 1.816225528717041, "step": 11301 }, { "epoch": 0.890973590855341, "grad_norm": 0.6527709364891052, "learning_rate": 0.0005, "loss": 1.8629765510559082, "step": 11302 }, { "epoch": 0.8910524241229799, "grad_norm": 0.5793249607086182, "learning_rate": 0.0005, "loss": 1.839771032333374, "step": 11303 }, { "epoch": 0.8911312573906188, "grad_norm": 0.663854718208313, "learning_rate": 0.0005, "loss": 1.9176394939422607, "step": 11304 }, { "epoch": 0.8912100906582577, "grad_norm": 0.5533884763717651, "learning_rate": 0.0005, "loss": 1.7498555183410645, "step": 11305 }, { "epoch": 0.8912889239258968, "grad_norm": 0.4694667160511017, "learning_rate": 0.0005, "loss": 1.6741284132003784, "step": 11306 }, { "epoch": 0.8913677571935357, "grad_norm": 0.7510668039321899, "learning_rate": 0.0005, "loss": 2.219043731689453, "step": 11307 }, { "epoch": 0.8914465904611746, "grad_norm": 0.7672231793403625, "learning_rate": 0.0005, "loss": 1.8960158824920654, "step": 11308 }, { "epoch": 0.8915254237288136, "grad_norm": 0.6371926069259644, "learning_rate": 0.0005, "loss": 2.090991497039795, "step": 11309 }, { "epoch": 0.8916042569964525, "grad_norm": 0.49137505888938904, "learning_rate": 0.0005, "loss": 1.819127082824707, "step": 11310 }, { "epoch": 0.8916830902640914, "grad_norm": 0.6672195792198181, "learning_rate": 0.0005, "loss": 1.9813947677612305, "step": 11311 }, { "epoch": 0.8917619235317303, "grad_norm": 0.5792180299758911, "learning_rate": 0.0005, "loss": 1.777355670928955, "step": 11312 }, { "epoch": 0.8918407567993694, "grad_norm": 1.2425615787506104, "learning_rate": 0.0005, "loss": 2.073333263397217, "step": 11313 }, { "epoch": 0.8919195900670083, "grad_norm": 0.603335976600647, "learning_rate": 0.0005, "loss": 1.8989148139953613, "step": 11314 }, { "epoch": 0.8919984233346472, "grad_norm": 0.6485317349433899, "learning_rate": 0.0005, "loss": 1.8118988275527954, "step": 11315 }, { "epoch": 0.8920772566022862, "grad_norm": 0.6343483328819275, "learning_rate": 0.0005, "loss": 1.8711707592010498, "step": 11316 }, { "epoch": 0.8921560898699251, "grad_norm": 0.5689440369606018, "learning_rate": 0.0005, "loss": 1.8452184200286865, "step": 11317 }, { "epoch": 0.892234923137564, "grad_norm": 0.4972347617149353, "learning_rate": 0.0005, "loss": 1.5625542402267456, "step": 11318 }, { "epoch": 0.8923137564052029, "grad_norm": 0.685588002204895, "learning_rate": 0.0005, "loss": 1.970818042755127, "step": 11319 }, { "epoch": 0.892392589672842, "grad_norm": 0.6581182479858398, "learning_rate": 0.0005, "loss": 1.8563008308410645, "step": 11320 }, { "epoch": 0.8924714229404809, "grad_norm": 0.7627583146095276, "learning_rate": 0.0005, "loss": 2.0990142822265625, "step": 11321 }, { "epoch": 0.8925502562081198, "grad_norm": 0.5152308940887451, "learning_rate": 0.0005, "loss": 1.8888466358184814, "step": 11322 }, { "epoch": 0.8926290894757588, "grad_norm": 0.5642611384391785, "learning_rate": 0.0005, "loss": 1.7080228328704834, "step": 11323 }, { "epoch": 0.8927079227433977, "grad_norm": 0.5918394327163696, "learning_rate": 0.0005, "loss": 1.6546130180358887, "step": 11324 }, { "epoch": 0.8927867560110366, "grad_norm": 0.7675516605377197, "learning_rate": 0.0005, "loss": 1.9742224216461182, "step": 11325 }, { "epoch": 0.8928655892786757, "grad_norm": 0.648819088935852, "learning_rate": 0.0005, "loss": 1.9016326665878296, "step": 11326 }, { "epoch": 0.8929444225463146, "grad_norm": 0.7604578137397766, "learning_rate": 0.0005, "loss": 1.6731942892074585, "step": 11327 }, { "epoch": 0.8930232558139535, "grad_norm": 0.6435709595680237, "learning_rate": 0.0005, "loss": 1.7961872816085815, "step": 11328 }, { "epoch": 0.8931020890815924, "grad_norm": 0.9696373343467712, "learning_rate": 0.0005, "loss": 1.8133409023284912, "step": 11329 }, { "epoch": 0.8931809223492314, "grad_norm": 0.742562472820282, "learning_rate": 0.0005, "loss": 1.9102766513824463, "step": 11330 }, { "epoch": 0.8932597556168703, "grad_norm": 0.9122843742370605, "learning_rate": 0.0005, "loss": 1.7119346857070923, "step": 11331 }, { "epoch": 0.8933385888845092, "grad_norm": 0.6819433569908142, "learning_rate": 0.0005, "loss": 1.9390795230865479, "step": 11332 }, { "epoch": 0.8934174221521483, "grad_norm": 0.6196334362030029, "learning_rate": 0.0005, "loss": 2.0131890773773193, "step": 11333 }, { "epoch": 0.8934962554197872, "grad_norm": 0.6951673030853271, "learning_rate": 0.0005, "loss": 1.886695384979248, "step": 11334 }, { "epoch": 0.8935750886874261, "grad_norm": 0.9494037628173828, "learning_rate": 0.0005, "loss": 1.8942341804504395, "step": 11335 }, { "epoch": 0.893653921955065, "grad_norm": 0.7460700869560242, "learning_rate": 0.0005, "loss": 1.9108107089996338, "step": 11336 }, { "epoch": 0.893732755222704, "grad_norm": 0.6364762783050537, "learning_rate": 0.0005, "loss": 1.7118327617645264, "step": 11337 }, { "epoch": 0.8938115884903429, "grad_norm": 0.4928935468196869, "learning_rate": 0.0005, "loss": 1.8000316619873047, "step": 11338 }, { "epoch": 0.8938904217579818, "grad_norm": 0.6085902452468872, "learning_rate": 0.0005, "loss": 1.9768656492233276, "step": 11339 }, { "epoch": 0.8939692550256209, "grad_norm": 0.6454700231552124, "learning_rate": 0.0005, "loss": 1.761433720588684, "step": 11340 }, { "epoch": 0.8940480882932598, "grad_norm": 0.9616930484771729, "learning_rate": 0.0005, "loss": 1.7204697132110596, "step": 11341 }, { "epoch": 0.8941269215608987, "grad_norm": 0.761963427066803, "learning_rate": 0.0005, "loss": 1.8227828741073608, "step": 11342 }, { "epoch": 0.8942057548285376, "grad_norm": 0.8669356107711792, "learning_rate": 0.0005, "loss": 1.9484713077545166, "step": 11343 }, { "epoch": 0.8942845880961766, "grad_norm": 0.7268620133399963, "learning_rate": 0.0005, "loss": 2.1153101921081543, "step": 11344 }, { "epoch": 0.8943634213638155, "grad_norm": 0.6060914397239685, "learning_rate": 0.0005, "loss": 1.7588300704956055, "step": 11345 }, { "epoch": 0.8944422546314544, "grad_norm": 0.9945561289787292, "learning_rate": 0.0005, "loss": 2.0300590991973877, "step": 11346 }, { "epoch": 0.8945210878990935, "grad_norm": 0.5451644062995911, "learning_rate": 0.0005, "loss": 1.8012163639068604, "step": 11347 }, { "epoch": 0.8945999211667324, "grad_norm": 0.8680224418640137, "learning_rate": 0.0005, "loss": 2.203523635864258, "step": 11348 }, { "epoch": 0.8946787544343713, "grad_norm": 0.8135586977005005, "learning_rate": 0.0005, "loss": 1.540766954421997, "step": 11349 }, { "epoch": 0.8947575877020102, "grad_norm": 0.8708769679069519, "learning_rate": 0.0005, "loss": 2.0202972888946533, "step": 11350 }, { "epoch": 0.8948364209696492, "grad_norm": 1.1169825792312622, "learning_rate": 0.0005, "loss": 2.0355429649353027, "step": 11351 }, { "epoch": 0.8949152542372881, "grad_norm": 0.6718869209289551, "learning_rate": 0.0005, "loss": 1.9690709114074707, "step": 11352 }, { "epoch": 0.894994087504927, "grad_norm": 0.6287291049957275, "learning_rate": 0.0005, "loss": 1.9273972511291504, "step": 11353 }, { "epoch": 0.8950729207725661, "grad_norm": 0.7176713943481445, "learning_rate": 0.0005, "loss": 1.8717148303985596, "step": 11354 }, { "epoch": 0.895151754040205, "grad_norm": 0.6513767242431641, "learning_rate": 0.0005, "loss": 1.9624757766723633, "step": 11355 }, { "epoch": 0.8952305873078439, "grad_norm": 0.6488184332847595, "learning_rate": 0.0005, "loss": 1.888075351715088, "step": 11356 }, { "epoch": 0.8953094205754829, "grad_norm": 0.767796516418457, "learning_rate": 0.0005, "loss": 1.8174214363098145, "step": 11357 }, { "epoch": 0.8953882538431218, "grad_norm": 0.6633116006851196, "learning_rate": 0.0005, "loss": 2.0150914192199707, "step": 11358 }, { "epoch": 0.8954670871107607, "grad_norm": 0.8399151563644409, "learning_rate": 0.0005, "loss": 1.9446032047271729, "step": 11359 }, { "epoch": 0.8955459203783996, "grad_norm": 0.7485116124153137, "learning_rate": 0.0005, "loss": 1.8196271657943726, "step": 11360 }, { "epoch": 0.8956247536460387, "grad_norm": 0.5561650991439819, "learning_rate": 0.0005, "loss": 1.5551341772079468, "step": 11361 }, { "epoch": 0.8957035869136776, "grad_norm": 0.7379795908927917, "learning_rate": 0.0005, "loss": 1.6204028129577637, "step": 11362 }, { "epoch": 0.8957824201813165, "grad_norm": 0.8155869245529175, "learning_rate": 0.0005, "loss": 1.8125975131988525, "step": 11363 }, { "epoch": 0.8958612534489555, "grad_norm": 0.7369685769081116, "learning_rate": 0.0005, "loss": 1.8102085590362549, "step": 11364 }, { "epoch": 0.8959400867165944, "grad_norm": 0.7686203122138977, "learning_rate": 0.0005, "loss": 1.6840945482254028, "step": 11365 }, { "epoch": 0.8960189199842333, "grad_norm": 0.7650321125984192, "learning_rate": 0.0005, "loss": 1.893237590789795, "step": 11366 }, { "epoch": 0.8960977532518722, "grad_norm": 0.7674878239631653, "learning_rate": 0.0005, "loss": 2.045438766479492, "step": 11367 }, { "epoch": 0.8961765865195113, "grad_norm": 0.7233961820602417, "learning_rate": 0.0005, "loss": 1.7479350566864014, "step": 11368 }, { "epoch": 0.8962554197871502, "grad_norm": 0.7133005857467651, "learning_rate": 0.0005, "loss": 2.052690267562866, "step": 11369 }, { "epoch": 0.8963342530547891, "grad_norm": 0.8065017461776733, "learning_rate": 0.0005, "loss": 1.9509626626968384, "step": 11370 }, { "epoch": 0.8964130863224281, "grad_norm": 0.652157187461853, "learning_rate": 0.0005, "loss": 1.6500880718231201, "step": 11371 }, { "epoch": 0.896491919590067, "grad_norm": 0.5577069520950317, "learning_rate": 0.0005, "loss": 1.5038514137268066, "step": 11372 }, { "epoch": 0.8965707528577059, "grad_norm": 0.5320397615432739, "learning_rate": 0.0005, "loss": 1.8447035551071167, "step": 11373 }, { "epoch": 0.8966495861253448, "grad_norm": 0.4761612117290497, "learning_rate": 0.0005, "loss": 1.5477745532989502, "step": 11374 }, { "epoch": 0.8967284193929839, "grad_norm": 0.6984400153160095, "learning_rate": 0.0005, "loss": 1.786097526550293, "step": 11375 }, { "epoch": 0.8968072526606228, "grad_norm": 0.6136370301246643, "learning_rate": 0.0005, "loss": 1.9724500179290771, "step": 11376 }, { "epoch": 0.8968860859282617, "grad_norm": 0.63223797082901, "learning_rate": 0.0005, "loss": 1.7537087202072144, "step": 11377 }, { "epoch": 0.8969649191959007, "grad_norm": 0.5599949955940247, "learning_rate": 0.0005, "loss": 1.966230869293213, "step": 11378 }, { "epoch": 0.8970437524635396, "grad_norm": 0.5654624700546265, "learning_rate": 0.0005, "loss": 2.0059022903442383, "step": 11379 }, { "epoch": 0.8971225857311785, "grad_norm": 0.5025176405906677, "learning_rate": 0.0005, "loss": 1.7310867309570312, "step": 11380 }, { "epoch": 0.8972014189988174, "grad_norm": 0.6833561658859253, "learning_rate": 0.0005, "loss": 1.8655840158462524, "step": 11381 }, { "epoch": 0.8972802522664565, "grad_norm": 0.7970373034477234, "learning_rate": 0.0005, "loss": 1.7782173156738281, "step": 11382 }, { "epoch": 0.8973590855340954, "grad_norm": 0.5095146298408508, "learning_rate": 0.0005, "loss": 1.719452142715454, "step": 11383 }, { "epoch": 0.8974379188017343, "grad_norm": 0.634371280670166, "learning_rate": 0.0005, "loss": 2.0128393173217773, "step": 11384 }, { "epoch": 0.8975167520693733, "grad_norm": 0.691902756690979, "learning_rate": 0.0005, "loss": 1.6808576583862305, "step": 11385 }, { "epoch": 0.8975955853370122, "grad_norm": 0.7272119522094727, "learning_rate": 0.0005, "loss": 1.9011462926864624, "step": 11386 }, { "epoch": 0.8976744186046511, "grad_norm": 0.5470457673072815, "learning_rate": 0.0005, "loss": 1.78836989402771, "step": 11387 }, { "epoch": 0.8977532518722902, "grad_norm": 0.4969526529312134, "learning_rate": 0.0005, "loss": 1.856162667274475, "step": 11388 }, { "epoch": 0.8978320851399291, "grad_norm": 0.6573765277862549, "learning_rate": 0.0005, "loss": 1.7403149604797363, "step": 11389 }, { "epoch": 0.897910918407568, "grad_norm": 0.5252629518508911, "learning_rate": 0.0005, "loss": 2.0547022819519043, "step": 11390 }, { "epoch": 0.8979897516752069, "grad_norm": 0.9406096339225769, "learning_rate": 0.0005, "loss": 1.6741151809692383, "step": 11391 }, { "epoch": 0.8980685849428459, "grad_norm": 0.6485819816589355, "learning_rate": 0.0005, "loss": 1.9602487087249756, "step": 11392 }, { "epoch": 0.8981474182104848, "grad_norm": 0.5714802742004395, "learning_rate": 0.0005, "loss": 1.8333370685577393, "step": 11393 }, { "epoch": 0.8982262514781237, "grad_norm": 0.626883864402771, "learning_rate": 0.0005, "loss": 1.6601868867874146, "step": 11394 }, { "epoch": 0.8983050847457628, "grad_norm": 0.7387762069702148, "learning_rate": 0.0005, "loss": 2.0500731468200684, "step": 11395 }, { "epoch": 0.8983839180134017, "grad_norm": 0.6967203617095947, "learning_rate": 0.0005, "loss": 1.9184629917144775, "step": 11396 }, { "epoch": 0.8984627512810406, "grad_norm": 0.7002344727516174, "learning_rate": 0.0005, "loss": 1.864801287651062, "step": 11397 }, { "epoch": 0.8985415845486795, "grad_norm": 0.835706353187561, "learning_rate": 0.0005, "loss": 1.815508484840393, "step": 11398 }, { "epoch": 0.8986204178163185, "grad_norm": 0.7494478821754456, "learning_rate": 0.0005, "loss": 2.0348775386810303, "step": 11399 }, { "epoch": 0.8986992510839574, "grad_norm": 0.8322300910949707, "learning_rate": 0.0005, "loss": 2.0771656036376953, "step": 11400 }, { "epoch": 0.8987780843515963, "grad_norm": 0.5930215120315552, "learning_rate": 0.0005, "loss": 1.9446601867675781, "step": 11401 }, { "epoch": 0.8988569176192354, "grad_norm": 0.7370638847351074, "learning_rate": 0.0005, "loss": 1.9944109916687012, "step": 11402 }, { "epoch": 0.8989357508868743, "grad_norm": 0.6079489588737488, "learning_rate": 0.0005, "loss": 1.8042062520980835, "step": 11403 }, { "epoch": 0.8990145841545132, "grad_norm": 0.5535160899162292, "learning_rate": 0.0005, "loss": 1.721152424812317, "step": 11404 }, { "epoch": 0.8990934174221521, "grad_norm": 0.6392717361450195, "learning_rate": 0.0005, "loss": 1.9324615001678467, "step": 11405 }, { "epoch": 0.8991722506897911, "grad_norm": 0.6958793997764587, "learning_rate": 0.0005, "loss": 2.1705586910247803, "step": 11406 }, { "epoch": 0.89925108395743, "grad_norm": 0.8123486638069153, "learning_rate": 0.0005, "loss": 2.2002594470977783, "step": 11407 }, { "epoch": 0.8993299172250689, "grad_norm": 0.6661878824234009, "learning_rate": 0.0005, "loss": 1.6748504638671875, "step": 11408 }, { "epoch": 0.899408750492708, "grad_norm": 0.5793383121490479, "learning_rate": 0.0005, "loss": 1.8304896354675293, "step": 11409 }, { "epoch": 0.8994875837603469, "grad_norm": 0.6180418729782104, "learning_rate": 0.0005, "loss": 1.8020919561386108, "step": 11410 }, { "epoch": 0.8995664170279858, "grad_norm": 1.049468994140625, "learning_rate": 0.0005, "loss": 2.120173454284668, "step": 11411 }, { "epoch": 0.8996452502956247, "grad_norm": 0.6327284574508667, "learning_rate": 0.0005, "loss": 1.959733009338379, "step": 11412 }, { "epoch": 0.8997240835632637, "grad_norm": 0.7215591073036194, "learning_rate": 0.0005, "loss": 1.9013437032699585, "step": 11413 }, { "epoch": 0.8998029168309026, "grad_norm": 0.6183239817619324, "learning_rate": 0.0005, "loss": 1.713784098625183, "step": 11414 }, { "epoch": 0.8998817500985415, "grad_norm": 0.7799504995346069, "learning_rate": 0.0005, "loss": 2.049325466156006, "step": 11415 }, { "epoch": 0.8999605833661806, "grad_norm": 0.5186173915863037, "learning_rate": 0.0005, "loss": 1.5675735473632812, "step": 11416 }, { "epoch": 0.9000394166338195, "grad_norm": 0.7698858380317688, "learning_rate": 0.0005, "loss": 2.0404012203216553, "step": 11417 }, { "epoch": 0.9001182499014584, "grad_norm": 0.5545641779899597, "learning_rate": 0.0005, "loss": 1.9514707326889038, "step": 11418 }, { "epoch": 0.9001970831690974, "grad_norm": 0.5796684622764587, "learning_rate": 0.0005, "loss": 1.6339327096939087, "step": 11419 }, { "epoch": 0.9002759164367363, "grad_norm": 0.6473045349121094, "learning_rate": 0.0005, "loss": 1.9172638654708862, "step": 11420 }, { "epoch": 0.9003547497043752, "grad_norm": 0.5481053590774536, "learning_rate": 0.0005, "loss": 1.9075509309768677, "step": 11421 }, { "epoch": 0.9004335829720141, "grad_norm": 0.6086079478263855, "learning_rate": 0.0005, "loss": 1.9417637586593628, "step": 11422 }, { "epoch": 0.9005124162396532, "grad_norm": 0.6923940181732178, "learning_rate": 0.0005, "loss": 2.0715649127960205, "step": 11423 }, { "epoch": 0.9005912495072921, "grad_norm": 0.751220703125, "learning_rate": 0.0005, "loss": 1.97133207321167, "step": 11424 }, { "epoch": 0.900670082774931, "grad_norm": 0.6810780763626099, "learning_rate": 0.0005, "loss": 1.7540068626403809, "step": 11425 }, { "epoch": 0.90074891604257, "grad_norm": 0.828855037689209, "learning_rate": 0.0005, "loss": 1.7534173727035522, "step": 11426 }, { "epoch": 0.9008277493102089, "grad_norm": 0.7261730432510376, "learning_rate": 0.0005, "loss": 1.9287686347961426, "step": 11427 }, { "epoch": 0.9009065825778478, "grad_norm": 0.6754130721092224, "learning_rate": 0.0005, "loss": 1.747414231300354, "step": 11428 }, { "epoch": 0.9009854158454867, "grad_norm": 0.5898585319519043, "learning_rate": 0.0005, "loss": 1.7754069566726685, "step": 11429 }, { "epoch": 0.9010642491131258, "grad_norm": 0.6397625803947449, "learning_rate": 0.0005, "loss": 1.7819732427597046, "step": 11430 }, { "epoch": 0.9011430823807647, "grad_norm": 0.7601812481880188, "learning_rate": 0.0005, "loss": 1.6521631479263306, "step": 11431 }, { "epoch": 0.9012219156484036, "grad_norm": 0.7452226877212524, "learning_rate": 0.0005, "loss": 2.0989508628845215, "step": 11432 }, { "epoch": 0.9013007489160426, "grad_norm": 0.6079995632171631, "learning_rate": 0.0005, "loss": 1.8917405605316162, "step": 11433 }, { "epoch": 0.9013795821836815, "grad_norm": 0.6083856821060181, "learning_rate": 0.0005, "loss": 1.7567558288574219, "step": 11434 }, { "epoch": 0.9014584154513204, "grad_norm": 0.5740156173706055, "learning_rate": 0.0005, "loss": 1.8586581945419312, "step": 11435 }, { "epoch": 0.9015372487189593, "grad_norm": 0.6364583969116211, "learning_rate": 0.0005, "loss": 1.8501074314117432, "step": 11436 }, { "epoch": 0.9016160819865984, "grad_norm": 0.6802034378051758, "learning_rate": 0.0005, "loss": 1.8991272449493408, "step": 11437 }, { "epoch": 0.9016949152542373, "grad_norm": 1.0753408670425415, "learning_rate": 0.0005, "loss": 2.060695171356201, "step": 11438 }, { "epoch": 0.9017737485218762, "grad_norm": 0.6859268546104431, "learning_rate": 0.0005, "loss": 1.619333267211914, "step": 11439 }, { "epoch": 0.9018525817895152, "grad_norm": 0.7466437816619873, "learning_rate": 0.0005, "loss": 1.762214183807373, "step": 11440 }, { "epoch": 0.9019314150571541, "grad_norm": 0.5475311875343323, "learning_rate": 0.0005, "loss": 1.893770456314087, "step": 11441 }, { "epoch": 0.902010248324793, "grad_norm": 0.9164003133773804, "learning_rate": 0.0005, "loss": 2.116034507751465, "step": 11442 }, { "epoch": 0.902089081592432, "grad_norm": 0.7814232707023621, "learning_rate": 0.0005, "loss": 1.9971790313720703, "step": 11443 }, { "epoch": 0.902167914860071, "grad_norm": 0.6787496209144592, "learning_rate": 0.0005, "loss": 1.759857416152954, "step": 11444 }, { "epoch": 0.9022467481277099, "grad_norm": 0.7382428050041199, "learning_rate": 0.0005, "loss": 1.6859636306762695, "step": 11445 }, { "epoch": 0.9023255813953488, "grad_norm": 0.6801466941833496, "learning_rate": 0.0005, "loss": 1.9373852014541626, "step": 11446 }, { "epoch": 0.9024044146629878, "grad_norm": 0.6167446374893188, "learning_rate": 0.0005, "loss": 1.769317388534546, "step": 11447 }, { "epoch": 0.9024832479306267, "grad_norm": 0.7251893877983093, "learning_rate": 0.0005, "loss": 1.7204140424728394, "step": 11448 }, { "epoch": 0.9025620811982656, "grad_norm": 0.8415836691856384, "learning_rate": 0.0005, "loss": 1.5588955879211426, "step": 11449 }, { "epoch": 0.9026409144659047, "grad_norm": 0.5807324051856995, "learning_rate": 0.0005, "loss": 2.066183567047119, "step": 11450 }, { "epoch": 0.9027197477335436, "grad_norm": 0.6235368251800537, "learning_rate": 0.0005, "loss": 1.919067144393921, "step": 11451 }, { "epoch": 0.9027985810011825, "grad_norm": 0.687635064125061, "learning_rate": 0.0005, "loss": 1.8496127128601074, "step": 11452 }, { "epoch": 0.9028774142688214, "grad_norm": 0.5730286836624146, "learning_rate": 0.0005, "loss": 1.4140219688415527, "step": 11453 }, { "epoch": 0.9029562475364604, "grad_norm": 1.0616594552993774, "learning_rate": 0.0005, "loss": 2.3369617462158203, "step": 11454 }, { "epoch": 0.9030350808040993, "grad_norm": 0.5607295632362366, "learning_rate": 0.0005, "loss": 1.6038010120391846, "step": 11455 }, { "epoch": 0.9031139140717382, "grad_norm": 0.6694273352622986, "learning_rate": 0.0005, "loss": 1.892712116241455, "step": 11456 }, { "epoch": 0.9031927473393773, "grad_norm": 0.6110194325447083, "learning_rate": 0.0005, "loss": 1.873566746711731, "step": 11457 }, { "epoch": 0.9032715806070162, "grad_norm": 0.7278068661689758, "learning_rate": 0.0005, "loss": 1.7079449892044067, "step": 11458 }, { "epoch": 0.9033504138746551, "grad_norm": 0.6367216110229492, "learning_rate": 0.0005, "loss": 1.8479599952697754, "step": 11459 }, { "epoch": 0.903429247142294, "grad_norm": 0.7466517090797424, "learning_rate": 0.0005, "loss": 1.7642364501953125, "step": 11460 }, { "epoch": 0.903508080409933, "grad_norm": 0.6372365355491638, "learning_rate": 0.0005, "loss": 1.6892590522766113, "step": 11461 }, { "epoch": 0.9035869136775719, "grad_norm": 0.5973851680755615, "learning_rate": 0.0005, "loss": 1.4526896476745605, "step": 11462 }, { "epoch": 0.9036657469452108, "grad_norm": 0.852333128452301, "learning_rate": 0.0005, "loss": 2.0472912788391113, "step": 11463 }, { "epoch": 0.9037445802128499, "grad_norm": 0.6925153136253357, "learning_rate": 0.0005, "loss": 1.709317684173584, "step": 11464 }, { "epoch": 0.9038234134804888, "grad_norm": 0.6979454159736633, "learning_rate": 0.0005, "loss": 1.9108490943908691, "step": 11465 }, { "epoch": 0.9039022467481277, "grad_norm": 0.5314859747886658, "learning_rate": 0.0005, "loss": 1.773363471031189, "step": 11466 }, { "epoch": 0.9039810800157666, "grad_norm": 0.5661711692810059, "learning_rate": 0.0005, "loss": 1.9656709432601929, "step": 11467 }, { "epoch": 0.9040599132834056, "grad_norm": 0.6991778612136841, "learning_rate": 0.0005, "loss": 2.0270535945892334, "step": 11468 }, { "epoch": 0.9041387465510445, "grad_norm": 0.7079845070838928, "learning_rate": 0.0005, "loss": 2.0013718605041504, "step": 11469 }, { "epoch": 0.9042175798186834, "grad_norm": 0.7674297094345093, "learning_rate": 0.0005, "loss": 1.7028205394744873, "step": 11470 }, { "epoch": 0.9042964130863225, "grad_norm": 1.028371810913086, "learning_rate": 0.0005, "loss": 2.026455879211426, "step": 11471 }, { "epoch": 0.9043752463539614, "grad_norm": 0.6750798225402832, "learning_rate": 0.0005, "loss": 1.8799939155578613, "step": 11472 }, { "epoch": 0.9044540796216003, "grad_norm": 0.8012058138847351, "learning_rate": 0.0005, "loss": 1.6594773530960083, "step": 11473 }, { "epoch": 0.9045329128892392, "grad_norm": 0.8549631237983704, "learning_rate": 0.0005, "loss": 1.7763155698776245, "step": 11474 }, { "epoch": 0.9046117461568782, "grad_norm": 0.6637629270553589, "learning_rate": 0.0005, "loss": 2.0419130325317383, "step": 11475 }, { "epoch": 0.9046905794245171, "grad_norm": 0.6980300545692444, "learning_rate": 0.0005, "loss": 1.6439095735549927, "step": 11476 }, { "epoch": 0.904769412692156, "grad_norm": 0.7412062883377075, "learning_rate": 0.0005, "loss": 2.2354955673217773, "step": 11477 }, { "epoch": 0.9048482459597951, "grad_norm": 0.6180593967437744, "learning_rate": 0.0005, "loss": 1.8815007209777832, "step": 11478 }, { "epoch": 0.904927079227434, "grad_norm": 0.7187169790267944, "learning_rate": 0.0005, "loss": 1.6384834051132202, "step": 11479 }, { "epoch": 0.9050059124950729, "grad_norm": 0.6522760987281799, "learning_rate": 0.0005, "loss": 1.8462227582931519, "step": 11480 }, { "epoch": 0.9050847457627119, "grad_norm": 0.5376715064048767, "learning_rate": 0.0005, "loss": 1.6168999671936035, "step": 11481 }, { "epoch": 0.9051635790303508, "grad_norm": 0.5953806042671204, "learning_rate": 0.0005, "loss": 1.6628921031951904, "step": 11482 }, { "epoch": 0.9052424122979897, "grad_norm": 0.6462514996528625, "learning_rate": 0.0005, "loss": 1.7513322830200195, "step": 11483 }, { "epoch": 0.9053212455656287, "grad_norm": 0.598845362663269, "learning_rate": 0.0005, "loss": 2.073395013809204, "step": 11484 }, { "epoch": 0.9054000788332677, "grad_norm": 0.6917313933372498, "learning_rate": 0.0005, "loss": 1.787381649017334, "step": 11485 }, { "epoch": 0.9054789121009066, "grad_norm": 0.7609949707984924, "learning_rate": 0.0005, "loss": 1.8676023483276367, "step": 11486 }, { "epoch": 0.9055577453685455, "grad_norm": 0.6531338691711426, "learning_rate": 0.0005, "loss": 1.8655786514282227, "step": 11487 }, { "epoch": 0.9056365786361845, "grad_norm": 0.8256713151931763, "learning_rate": 0.0005, "loss": 1.8558247089385986, "step": 11488 }, { "epoch": 0.9057154119038234, "grad_norm": 0.5917832255363464, "learning_rate": 0.0005, "loss": 1.900895595550537, "step": 11489 }, { "epoch": 0.9057942451714623, "grad_norm": 0.5870068073272705, "learning_rate": 0.0005, "loss": 1.7639508247375488, "step": 11490 }, { "epoch": 0.9058730784391013, "grad_norm": 0.5589222311973572, "learning_rate": 0.0005, "loss": 1.6251652240753174, "step": 11491 }, { "epoch": 0.9059519117067403, "grad_norm": 0.6137932538986206, "learning_rate": 0.0005, "loss": 1.8218755722045898, "step": 11492 }, { "epoch": 0.9060307449743792, "grad_norm": 0.6419891119003296, "learning_rate": 0.0005, "loss": 1.814923882484436, "step": 11493 }, { "epoch": 0.9061095782420181, "grad_norm": 0.877469003200531, "learning_rate": 0.0005, "loss": 1.6378145217895508, "step": 11494 }, { "epoch": 0.9061884115096571, "grad_norm": 0.6393672227859497, "learning_rate": 0.0005, "loss": 1.9475594758987427, "step": 11495 }, { "epoch": 0.906267244777296, "grad_norm": 0.634463906288147, "learning_rate": 0.0005, "loss": 1.8692419528961182, "step": 11496 }, { "epoch": 0.9063460780449349, "grad_norm": 0.5490925908088684, "learning_rate": 0.0005, "loss": 1.9385404586791992, "step": 11497 }, { "epoch": 0.9064249113125739, "grad_norm": 0.5717808604240417, "learning_rate": 0.0005, "loss": 1.850284218788147, "step": 11498 }, { "epoch": 0.9065037445802129, "grad_norm": 0.60957932472229, "learning_rate": 0.0005, "loss": 1.7280981540679932, "step": 11499 }, { "epoch": 0.9065825778478518, "grad_norm": 0.6979996562004089, "learning_rate": 0.0005, "loss": 1.9068279266357422, "step": 11500 }, { "epoch": 0.9066614111154907, "grad_norm": 0.732028067111969, "learning_rate": 0.0005, "loss": 1.8436529636383057, "step": 11501 }, { "epoch": 0.9067402443831297, "grad_norm": 0.5943363308906555, "learning_rate": 0.0005, "loss": 1.937735676765442, "step": 11502 }, { "epoch": 0.9068190776507686, "grad_norm": 0.5919404625892639, "learning_rate": 0.0005, "loss": 1.7888877391815186, "step": 11503 }, { "epoch": 0.9068979109184075, "grad_norm": 0.7035335302352905, "learning_rate": 0.0005, "loss": 1.9148080348968506, "step": 11504 }, { "epoch": 0.9069767441860465, "grad_norm": 0.7125320434570312, "learning_rate": 0.0005, "loss": 1.6694667339324951, "step": 11505 }, { "epoch": 0.9070555774536855, "grad_norm": 0.7566222548484802, "learning_rate": 0.0005, "loss": 1.9184869527816772, "step": 11506 }, { "epoch": 0.9071344107213244, "grad_norm": 0.5478129982948303, "learning_rate": 0.0005, "loss": 1.6778833866119385, "step": 11507 }, { "epoch": 0.9072132439889633, "grad_norm": 0.7958002090454102, "learning_rate": 0.0005, "loss": 2.0181655883789062, "step": 11508 }, { "epoch": 0.9072920772566023, "grad_norm": 0.5162912011146545, "learning_rate": 0.0005, "loss": 1.7496511936187744, "step": 11509 }, { "epoch": 0.9073709105242412, "grad_norm": 0.6975655555725098, "learning_rate": 0.0005, "loss": 1.9937243461608887, "step": 11510 }, { "epoch": 0.9074497437918801, "grad_norm": 0.49750107526779175, "learning_rate": 0.0005, "loss": 1.6194584369659424, "step": 11511 }, { "epoch": 0.9075285770595192, "grad_norm": 0.5199116468429565, "learning_rate": 0.0005, "loss": 1.7956814765930176, "step": 11512 }, { "epoch": 0.9076074103271581, "grad_norm": 0.6378533244132996, "learning_rate": 0.0005, "loss": 1.6593046188354492, "step": 11513 }, { "epoch": 0.907686243594797, "grad_norm": 0.5471298098564148, "learning_rate": 0.0005, "loss": 1.8172106742858887, "step": 11514 }, { "epoch": 0.9077650768624359, "grad_norm": 0.6029770970344543, "learning_rate": 0.0005, "loss": 1.7251168489456177, "step": 11515 }, { "epoch": 0.9078439101300749, "grad_norm": 0.5699765682220459, "learning_rate": 0.0005, "loss": 1.7733078002929688, "step": 11516 }, { "epoch": 0.9079227433977138, "grad_norm": 0.7714288234710693, "learning_rate": 0.0005, "loss": 2.00169038772583, "step": 11517 }, { "epoch": 0.9080015766653527, "grad_norm": 0.6871557831764221, "learning_rate": 0.0005, "loss": 1.975029468536377, "step": 11518 }, { "epoch": 0.9080804099329918, "grad_norm": 0.4991752803325653, "learning_rate": 0.0005, "loss": 1.5139634609222412, "step": 11519 }, { "epoch": 0.9081592432006307, "grad_norm": 0.5903025269508362, "learning_rate": 0.0005, "loss": 1.7297601699829102, "step": 11520 }, { "epoch": 0.9082380764682696, "grad_norm": 0.6361173391342163, "learning_rate": 0.0005, "loss": 1.7134665250778198, "step": 11521 }, { "epoch": 0.9083169097359085, "grad_norm": 0.734546959400177, "learning_rate": 0.0005, "loss": 1.6392567157745361, "step": 11522 }, { "epoch": 0.9083957430035475, "grad_norm": 0.5634956359863281, "learning_rate": 0.0005, "loss": 1.8552780151367188, "step": 11523 }, { "epoch": 0.9084745762711864, "grad_norm": 0.6354663968086243, "learning_rate": 0.0005, "loss": 1.8345296382904053, "step": 11524 }, { "epoch": 0.9085534095388254, "grad_norm": 0.7126505970954895, "learning_rate": 0.0005, "loss": 1.7786065340042114, "step": 11525 }, { "epoch": 0.9086322428064644, "grad_norm": 0.5852231383323669, "learning_rate": 0.0005, "loss": 2.148292303085327, "step": 11526 }, { "epoch": 0.9087110760741033, "grad_norm": 0.6772783398628235, "learning_rate": 0.0005, "loss": 1.731231927871704, "step": 11527 }, { "epoch": 0.9087899093417422, "grad_norm": 0.6074910759925842, "learning_rate": 0.0005, "loss": 1.836285948753357, "step": 11528 }, { "epoch": 0.9088687426093811, "grad_norm": 0.7321765422821045, "learning_rate": 0.0005, "loss": 1.9117379188537598, "step": 11529 }, { "epoch": 0.9089475758770201, "grad_norm": 0.5373765230178833, "learning_rate": 0.0005, "loss": 1.7927260398864746, "step": 11530 }, { "epoch": 0.909026409144659, "grad_norm": 0.5104963779449463, "learning_rate": 0.0005, "loss": 1.7044017314910889, "step": 11531 }, { "epoch": 0.909105242412298, "grad_norm": 0.5022671818733215, "learning_rate": 0.0005, "loss": 1.7995753288269043, "step": 11532 }, { "epoch": 0.909184075679937, "grad_norm": 0.548234224319458, "learning_rate": 0.0005, "loss": 2.0285444259643555, "step": 11533 }, { "epoch": 0.9092629089475759, "grad_norm": 0.6146717667579651, "learning_rate": 0.0005, "loss": 1.7885499000549316, "step": 11534 }, { "epoch": 0.9093417422152148, "grad_norm": 0.5976431965827942, "learning_rate": 0.0005, "loss": 1.7951362133026123, "step": 11535 }, { "epoch": 0.9094205754828537, "grad_norm": 0.5708363652229309, "learning_rate": 0.0005, "loss": 1.867814540863037, "step": 11536 }, { "epoch": 0.9094994087504927, "grad_norm": 0.5158214569091797, "learning_rate": 0.0005, "loss": 1.762041449546814, "step": 11537 }, { "epoch": 0.9095782420181316, "grad_norm": 0.6460022926330566, "learning_rate": 0.0005, "loss": 1.7157278060913086, "step": 11538 }, { "epoch": 0.9096570752857706, "grad_norm": 0.5303933024406433, "learning_rate": 0.0005, "loss": 1.8207643032073975, "step": 11539 }, { "epoch": 0.9097359085534096, "grad_norm": 0.7421061396598816, "learning_rate": 0.0005, "loss": 2.175262451171875, "step": 11540 }, { "epoch": 0.9098147418210485, "grad_norm": 0.5868029594421387, "learning_rate": 0.0005, "loss": 1.735673189163208, "step": 11541 }, { "epoch": 0.9098935750886874, "grad_norm": 0.6014599204063416, "learning_rate": 0.0005, "loss": 1.9984850883483887, "step": 11542 }, { "epoch": 0.9099724083563264, "grad_norm": 0.6604465246200562, "learning_rate": 0.0005, "loss": 1.7073042392730713, "step": 11543 }, { "epoch": 0.9100512416239653, "grad_norm": 0.6819690465927124, "learning_rate": 0.0005, "loss": 1.8037933111190796, "step": 11544 }, { "epoch": 0.9101300748916042, "grad_norm": 0.551024854183197, "learning_rate": 0.0005, "loss": 1.9143213033676147, "step": 11545 }, { "epoch": 0.9102089081592432, "grad_norm": 0.5354592204093933, "learning_rate": 0.0005, "loss": 1.6857728958129883, "step": 11546 }, { "epoch": 0.9102877414268822, "grad_norm": 0.9627084136009216, "learning_rate": 0.0005, "loss": 1.8545567989349365, "step": 11547 }, { "epoch": 0.9103665746945211, "grad_norm": 0.5879049897193909, "learning_rate": 0.0005, "loss": 2.0165655612945557, "step": 11548 }, { "epoch": 0.91044540796216, "grad_norm": 0.6756230592727661, "learning_rate": 0.0005, "loss": 1.9073777198791504, "step": 11549 }, { "epoch": 0.910524241229799, "grad_norm": 0.753149151802063, "learning_rate": 0.0005, "loss": 1.6652860641479492, "step": 11550 }, { "epoch": 0.9106030744974379, "grad_norm": 0.9152052998542786, "learning_rate": 0.0005, "loss": 1.860170841217041, "step": 11551 }, { "epoch": 0.9106819077650768, "grad_norm": 0.516312837600708, "learning_rate": 0.0005, "loss": 1.3170595169067383, "step": 11552 }, { "epoch": 0.9107607410327158, "grad_norm": 0.6233481168746948, "learning_rate": 0.0005, "loss": 1.5597360134124756, "step": 11553 }, { "epoch": 0.9108395743003548, "grad_norm": 0.6098544597625732, "learning_rate": 0.0005, "loss": 1.8415288925170898, "step": 11554 }, { "epoch": 0.9109184075679937, "grad_norm": 0.6919965147972107, "learning_rate": 0.0005, "loss": 1.5948303937911987, "step": 11555 }, { "epoch": 0.9109972408356326, "grad_norm": 0.5800155997276306, "learning_rate": 0.0005, "loss": 1.685329794883728, "step": 11556 }, { "epoch": 0.9110760741032716, "grad_norm": 0.5675588250160217, "learning_rate": 0.0005, "loss": 1.8557515144348145, "step": 11557 }, { "epoch": 0.9111549073709105, "grad_norm": 0.722663938999176, "learning_rate": 0.0005, "loss": 1.8727171421051025, "step": 11558 }, { "epoch": 0.9112337406385494, "grad_norm": 0.6345788836479187, "learning_rate": 0.0005, "loss": 1.7721641063690186, "step": 11559 }, { "epoch": 0.9113125739061884, "grad_norm": 0.7147472500801086, "learning_rate": 0.0005, "loss": 1.7390789985656738, "step": 11560 }, { "epoch": 0.9113914071738274, "grad_norm": 0.6639487743377686, "learning_rate": 0.0005, "loss": 1.7096161842346191, "step": 11561 }, { "epoch": 0.9114702404414663, "grad_norm": 0.564862072467804, "learning_rate": 0.0005, "loss": 1.8733407258987427, "step": 11562 }, { "epoch": 0.9115490737091052, "grad_norm": 0.6198612451553345, "learning_rate": 0.0005, "loss": 1.928518533706665, "step": 11563 }, { "epoch": 0.9116279069767442, "grad_norm": 0.7114989161491394, "learning_rate": 0.0005, "loss": 1.83713960647583, "step": 11564 }, { "epoch": 0.9117067402443831, "grad_norm": 0.6255855560302734, "learning_rate": 0.0005, "loss": 1.60379958152771, "step": 11565 }, { "epoch": 0.911785573512022, "grad_norm": 0.6315047144889832, "learning_rate": 0.0005, "loss": 1.580796480178833, "step": 11566 }, { "epoch": 0.911864406779661, "grad_norm": 0.6531157493591309, "learning_rate": 0.0005, "loss": 1.827453851699829, "step": 11567 }, { "epoch": 0.9119432400473, "grad_norm": 0.546174168586731, "learning_rate": 0.0005, "loss": 1.8089081048965454, "step": 11568 }, { "epoch": 0.9120220733149389, "grad_norm": 0.5567180514335632, "learning_rate": 0.0005, "loss": 1.7636127471923828, "step": 11569 }, { "epoch": 0.9121009065825778, "grad_norm": 0.7009392976760864, "learning_rate": 0.0005, "loss": 1.7725000381469727, "step": 11570 }, { "epoch": 0.9121797398502168, "grad_norm": 0.5059404373168945, "learning_rate": 0.0005, "loss": 1.3347110748291016, "step": 11571 }, { "epoch": 0.9122585731178557, "grad_norm": 0.5324031114578247, "learning_rate": 0.0005, "loss": 1.5523970127105713, "step": 11572 }, { "epoch": 0.9123374063854947, "grad_norm": 0.6270380616188049, "learning_rate": 0.0005, "loss": 1.7790945768356323, "step": 11573 }, { "epoch": 0.9124162396531337, "grad_norm": 0.6884292960166931, "learning_rate": 0.0005, "loss": 2.1356844902038574, "step": 11574 }, { "epoch": 0.9124950729207726, "grad_norm": 0.7147566080093384, "learning_rate": 0.0005, "loss": 1.57076895236969, "step": 11575 }, { "epoch": 0.9125739061884115, "grad_norm": 0.5809904932975769, "learning_rate": 0.0005, "loss": 1.7627681493759155, "step": 11576 }, { "epoch": 0.9126527394560504, "grad_norm": 0.6109252572059631, "learning_rate": 0.0005, "loss": 1.492180585861206, "step": 11577 }, { "epoch": 0.9127315727236894, "grad_norm": 0.7226533889770508, "learning_rate": 0.0005, "loss": 1.9955490827560425, "step": 11578 }, { "epoch": 0.9128104059913283, "grad_norm": 0.4836318790912628, "learning_rate": 0.0005, "loss": 1.7683017253875732, "step": 11579 }, { "epoch": 0.9128892392589673, "grad_norm": 0.6148911714553833, "learning_rate": 0.0005, "loss": 1.8497958183288574, "step": 11580 }, { "epoch": 0.9129680725266063, "grad_norm": 0.5570179224014282, "learning_rate": 0.0005, "loss": 1.9366251230239868, "step": 11581 }, { "epoch": 0.9130469057942452, "grad_norm": 0.6664530038833618, "learning_rate": 0.0005, "loss": 1.8888118267059326, "step": 11582 }, { "epoch": 0.9131257390618841, "grad_norm": 0.6870473027229309, "learning_rate": 0.0005, "loss": 1.6884961128234863, "step": 11583 }, { "epoch": 0.913204572329523, "grad_norm": 0.7823158502578735, "learning_rate": 0.0005, "loss": 2.3394689559936523, "step": 11584 }, { "epoch": 0.913283405597162, "grad_norm": 0.8308542966842651, "learning_rate": 0.0005, "loss": 1.9244452714920044, "step": 11585 }, { "epoch": 0.913362238864801, "grad_norm": 0.5497216582298279, "learning_rate": 0.0005, "loss": 1.7301924228668213, "step": 11586 }, { "epoch": 0.9134410721324399, "grad_norm": 0.7074354887008667, "learning_rate": 0.0005, "loss": 1.7798058986663818, "step": 11587 }, { "epoch": 0.9135199054000789, "grad_norm": 0.5774028897285461, "learning_rate": 0.0005, "loss": 1.8693941831588745, "step": 11588 }, { "epoch": 0.9135987386677178, "grad_norm": 0.6903212666511536, "learning_rate": 0.0005, "loss": 1.6773927211761475, "step": 11589 }, { "epoch": 0.9136775719353567, "grad_norm": 0.6813832521438599, "learning_rate": 0.0005, "loss": 1.865344762802124, "step": 11590 }, { "epoch": 0.9137564052029956, "grad_norm": 0.8029569983482361, "learning_rate": 0.0005, "loss": 1.8394020795822144, "step": 11591 }, { "epoch": 0.9138352384706346, "grad_norm": 0.4691865146160126, "learning_rate": 0.0005, "loss": 1.6609715223312378, "step": 11592 }, { "epoch": 0.9139140717382735, "grad_norm": 0.6123281717300415, "learning_rate": 0.0005, "loss": 1.772922158241272, "step": 11593 }, { "epoch": 0.9139929050059125, "grad_norm": 0.6384981870651245, "learning_rate": 0.0005, "loss": 1.7351574897766113, "step": 11594 }, { "epoch": 0.9140717382735515, "grad_norm": 0.6279686689376831, "learning_rate": 0.0005, "loss": 2.1283252239227295, "step": 11595 }, { "epoch": 0.9141505715411904, "grad_norm": 0.5316555500030518, "learning_rate": 0.0005, "loss": 1.7187798023223877, "step": 11596 }, { "epoch": 0.9142294048088293, "grad_norm": 0.5748708248138428, "learning_rate": 0.0005, "loss": 1.8816208839416504, "step": 11597 }, { "epoch": 0.9143082380764682, "grad_norm": 0.7279051542282104, "learning_rate": 0.0005, "loss": 2.0811660289764404, "step": 11598 }, { "epoch": 0.9143870713441072, "grad_norm": 0.48608213663101196, "learning_rate": 0.0005, "loss": 1.7227017879486084, "step": 11599 }, { "epoch": 0.9144659046117461, "grad_norm": 0.5353703498840332, "learning_rate": 0.0005, "loss": 1.7604060173034668, "step": 11600 }, { "epoch": 0.9145447378793851, "grad_norm": 0.591590166091919, "learning_rate": 0.0005, "loss": 1.775905966758728, "step": 11601 }, { "epoch": 0.9146235711470241, "grad_norm": 0.584112286567688, "learning_rate": 0.0005, "loss": 2.0168161392211914, "step": 11602 }, { "epoch": 0.914702404414663, "grad_norm": 0.6487283706665039, "learning_rate": 0.0005, "loss": 1.7392581701278687, "step": 11603 }, { "epoch": 0.9147812376823019, "grad_norm": 0.7275392413139343, "learning_rate": 0.0005, "loss": 1.9698379039764404, "step": 11604 }, { "epoch": 0.9148600709499409, "grad_norm": 0.6705190539360046, "learning_rate": 0.0005, "loss": 1.985732078552246, "step": 11605 }, { "epoch": 0.9149389042175798, "grad_norm": 0.5737301707267761, "learning_rate": 0.0005, "loss": 1.6494393348693848, "step": 11606 }, { "epoch": 0.9150177374852188, "grad_norm": 0.6459156274795532, "learning_rate": 0.0005, "loss": 1.6560578346252441, "step": 11607 }, { "epoch": 0.9150965707528577, "grad_norm": 0.8962375521659851, "learning_rate": 0.0005, "loss": 1.7523599863052368, "step": 11608 }, { "epoch": 0.9151754040204967, "grad_norm": 0.6293601989746094, "learning_rate": 0.0005, "loss": 1.7546100616455078, "step": 11609 }, { "epoch": 0.9152542372881356, "grad_norm": 0.5993903875350952, "learning_rate": 0.0005, "loss": 1.7748863697052002, "step": 11610 }, { "epoch": 0.9153330705557745, "grad_norm": 0.7637306451797485, "learning_rate": 0.0005, "loss": 1.8790264129638672, "step": 11611 }, { "epoch": 0.9154119038234135, "grad_norm": 0.6271835565567017, "learning_rate": 0.0005, "loss": 1.7604708671569824, "step": 11612 }, { "epoch": 0.9154907370910524, "grad_norm": 0.7126160264015198, "learning_rate": 0.0005, "loss": 1.7370439767837524, "step": 11613 }, { "epoch": 0.9155695703586914, "grad_norm": 0.6912739276885986, "learning_rate": 0.0005, "loss": 2.0241305828094482, "step": 11614 }, { "epoch": 0.9156484036263303, "grad_norm": 0.5823034048080444, "learning_rate": 0.0005, "loss": 1.7180507183074951, "step": 11615 }, { "epoch": 0.9157272368939693, "grad_norm": 0.8654314279556274, "learning_rate": 0.0005, "loss": 1.7686729431152344, "step": 11616 }, { "epoch": 0.9158060701616082, "grad_norm": 0.5156548619270325, "learning_rate": 0.0005, "loss": 1.854313611984253, "step": 11617 }, { "epoch": 0.9158849034292471, "grad_norm": 0.656613290309906, "learning_rate": 0.0005, "loss": 1.7152800559997559, "step": 11618 }, { "epoch": 0.9159637366968861, "grad_norm": 0.7051109075546265, "learning_rate": 0.0005, "loss": 1.9243810176849365, "step": 11619 }, { "epoch": 0.916042569964525, "grad_norm": 0.92943274974823, "learning_rate": 0.0005, "loss": 1.8891198635101318, "step": 11620 }, { "epoch": 0.916121403232164, "grad_norm": 0.6985666155815125, "learning_rate": 0.0005, "loss": 1.8454325199127197, "step": 11621 }, { "epoch": 0.9162002364998029, "grad_norm": 0.4921969473361969, "learning_rate": 0.0005, "loss": 1.6212708950042725, "step": 11622 }, { "epoch": 0.9162790697674419, "grad_norm": 0.6851573586463928, "learning_rate": 0.0005, "loss": 1.942376971244812, "step": 11623 }, { "epoch": 0.9163579030350808, "grad_norm": 0.6799976825714111, "learning_rate": 0.0005, "loss": 1.6861172914505005, "step": 11624 }, { "epoch": 0.9164367363027197, "grad_norm": 0.5884646773338318, "learning_rate": 0.0005, "loss": 1.976027250289917, "step": 11625 }, { "epoch": 0.9165155695703587, "grad_norm": 0.891810953617096, "learning_rate": 0.0005, "loss": 2.2418434619903564, "step": 11626 }, { "epoch": 0.9165944028379976, "grad_norm": 0.568275511264801, "learning_rate": 0.0005, "loss": 1.7434754371643066, "step": 11627 }, { "epoch": 0.9166732361056366, "grad_norm": 0.7829170227050781, "learning_rate": 0.0005, "loss": 1.9091633558273315, "step": 11628 }, { "epoch": 0.9167520693732755, "grad_norm": 0.6429881453514099, "learning_rate": 0.0005, "loss": 1.6757138967514038, "step": 11629 }, { "epoch": 0.9168309026409145, "grad_norm": 0.6843387484550476, "learning_rate": 0.0005, "loss": 1.7837930917739868, "step": 11630 }, { "epoch": 0.9169097359085534, "grad_norm": 0.5506452322006226, "learning_rate": 0.0005, "loss": 1.44899582862854, "step": 11631 }, { "epoch": 0.9169885691761923, "grad_norm": 0.7397261261940002, "learning_rate": 0.0005, "loss": 1.8919782638549805, "step": 11632 }, { "epoch": 0.9170674024438313, "grad_norm": 0.5287863612174988, "learning_rate": 0.0005, "loss": 1.7930073738098145, "step": 11633 }, { "epoch": 0.9171462357114702, "grad_norm": 1.008283019065857, "learning_rate": 0.0005, "loss": 1.8766107559204102, "step": 11634 }, { "epoch": 0.9172250689791092, "grad_norm": 0.7096644639968872, "learning_rate": 0.0005, "loss": 1.6723288297653198, "step": 11635 }, { "epoch": 0.9173039022467482, "grad_norm": 0.6416416168212891, "learning_rate": 0.0005, "loss": 1.6793352365493774, "step": 11636 }, { "epoch": 0.9173827355143871, "grad_norm": 0.6387192010879517, "learning_rate": 0.0005, "loss": 1.9443061351776123, "step": 11637 }, { "epoch": 0.917461568782026, "grad_norm": 0.7717581391334534, "learning_rate": 0.0005, "loss": 1.779202938079834, "step": 11638 }, { "epoch": 0.9175404020496649, "grad_norm": 0.5977578163146973, "learning_rate": 0.0005, "loss": 1.702056884765625, "step": 11639 }, { "epoch": 0.9176192353173039, "grad_norm": 0.6173940896987915, "learning_rate": 0.0005, "loss": 2.135173797607422, "step": 11640 }, { "epoch": 0.9176980685849428, "grad_norm": 0.6138042211532593, "learning_rate": 0.0005, "loss": 2.017902374267578, "step": 11641 }, { "epoch": 0.9177769018525818, "grad_norm": 0.7791653275489807, "learning_rate": 0.0005, "loss": 1.794572353363037, "step": 11642 }, { "epoch": 0.9178557351202208, "grad_norm": 0.4862363934516907, "learning_rate": 0.0005, "loss": 1.5853267908096313, "step": 11643 }, { "epoch": 0.9179345683878597, "grad_norm": 0.7020629644393921, "learning_rate": 0.0005, "loss": 1.676573634147644, "step": 11644 }, { "epoch": 0.9180134016554986, "grad_norm": 0.6549175977706909, "learning_rate": 0.0005, "loss": 1.8905174732208252, "step": 11645 }, { "epoch": 0.9180922349231375, "grad_norm": 0.9412369728088379, "learning_rate": 0.0005, "loss": 2.4650049209594727, "step": 11646 }, { "epoch": 0.9181710681907765, "grad_norm": 0.5264995098114014, "learning_rate": 0.0005, "loss": 1.4331181049346924, "step": 11647 }, { "epoch": 0.9182499014584155, "grad_norm": 0.6135927438735962, "learning_rate": 0.0005, "loss": 1.810927152633667, "step": 11648 }, { "epoch": 0.9183287347260544, "grad_norm": 0.8743124604225159, "learning_rate": 0.0005, "loss": 1.6948578357696533, "step": 11649 }, { "epoch": 0.9184075679936934, "grad_norm": 0.6133935451507568, "learning_rate": 0.0005, "loss": 1.6203105449676514, "step": 11650 }, { "epoch": 0.9184864012613323, "grad_norm": 0.7837606072425842, "learning_rate": 0.0005, "loss": 1.8368148803710938, "step": 11651 }, { "epoch": 0.9185652345289712, "grad_norm": 0.5711126327514648, "learning_rate": 0.0005, "loss": 1.8450998067855835, "step": 11652 }, { "epoch": 0.9186440677966101, "grad_norm": 0.7785565853118896, "learning_rate": 0.0005, "loss": 1.9117915630340576, "step": 11653 }, { "epoch": 0.9187229010642491, "grad_norm": 0.5409201979637146, "learning_rate": 0.0005, "loss": 1.9239002466201782, "step": 11654 }, { "epoch": 0.918801734331888, "grad_norm": 0.7458576560020447, "learning_rate": 0.0005, "loss": 1.8756136894226074, "step": 11655 }, { "epoch": 0.918880567599527, "grad_norm": 0.5871217846870422, "learning_rate": 0.0005, "loss": 1.843057632446289, "step": 11656 }, { "epoch": 0.918959400867166, "grad_norm": 0.8204489946365356, "learning_rate": 0.0005, "loss": 1.901955246925354, "step": 11657 }, { "epoch": 0.9190382341348049, "grad_norm": 0.5748139023780823, "learning_rate": 0.0005, "loss": 1.653939962387085, "step": 11658 }, { "epoch": 0.9191170674024438, "grad_norm": 0.7472779750823975, "learning_rate": 0.0005, "loss": 1.8531718254089355, "step": 11659 }, { "epoch": 0.9191959006700827, "grad_norm": 0.5885127782821655, "learning_rate": 0.0005, "loss": 1.890769600868225, "step": 11660 }, { "epoch": 0.9192747339377217, "grad_norm": 0.5645451545715332, "learning_rate": 0.0005, "loss": 1.9027600288391113, "step": 11661 }, { "epoch": 0.9193535672053607, "grad_norm": 0.6568878889083862, "learning_rate": 0.0005, "loss": 1.7932738065719604, "step": 11662 }, { "epoch": 0.9194324004729996, "grad_norm": 0.7014934420585632, "learning_rate": 0.0005, "loss": 2.31453537940979, "step": 11663 }, { "epoch": 0.9195112337406386, "grad_norm": 0.7862191200256348, "learning_rate": 0.0005, "loss": 2.219816207885742, "step": 11664 }, { "epoch": 0.9195900670082775, "grad_norm": 0.558460533618927, "learning_rate": 0.0005, "loss": 1.9074742794036865, "step": 11665 }, { "epoch": 0.9196689002759164, "grad_norm": 0.6116070747375488, "learning_rate": 0.0005, "loss": 1.6861462593078613, "step": 11666 }, { "epoch": 0.9197477335435554, "grad_norm": 0.5527525544166565, "learning_rate": 0.0005, "loss": 1.9232444763183594, "step": 11667 }, { "epoch": 0.9198265668111943, "grad_norm": 0.6159908175468445, "learning_rate": 0.0005, "loss": 1.9265788793563843, "step": 11668 }, { "epoch": 0.9199054000788333, "grad_norm": 0.6611000895500183, "learning_rate": 0.0005, "loss": 1.6995041370391846, "step": 11669 }, { "epoch": 0.9199842333464722, "grad_norm": 0.6105436086654663, "learning_rate": 0.0005, "loss": 2.004166841506958, "step": 11670 }, { "epoch": 0.9200630666141112, "grad_norm": 0.5259067416191101, "learning_rate": 0.0005, "loss": 1.7956246137619019, "step": 11671 }, { "epoch": 0.9201418998817501, "grad_norm": 0.794177770614624, "learning_rate": 0.0005, "loss": 1.950434923171997, "step": 11672 }, { "epoch": 0.920220733149389, "grad_norm": 0.5284155607223511, "learning_rate": 0.0005, "loss": 1.6058646440505981, "step": 11673 }, { "epoch": 0.920299566417028, "grad_norm": 0.6137681007385254, "learning_rate": 0.0005, "loss": 1.930189609527588, "step": 11674 }, { "epoch": 0.920378399684667, "grad_norm": 0.6330748200416565, "learning_rate": 0.0005, "loss": 1.922804832458496, "step": 11675 }, { "epoch": 0.9204572329523059, "grad_norm": 0.7754462957382202, "learning_rate": 0.0005, "loss": 1.8421659469604492, "step": 11676 }, { "epoch": 0.9205360662199448, "grad_norm": 0.7686214447021484, "learning_rate": 0.0005, "loss": 2.0768699645996094, "step": 11677 }, { "epoch": 0.9206148994875838, "grad_norm": 0.6428587436676025, "learning_rate": 0.0005, "loss": 1.685788631439209, "step": 11678 }, { "epoch": 0.9206937327552227, "grad_norm": 0.664417564868927, "learning_rate": 0.0005, "loss": 1.7680916786193848, "step": 11679 }, { "epoch": 0.9207725660228616, "grad_norm": 0.7324604392051697, "learning_rate": 0.0005, "loss": 2.0732316970825195, "step": 11680 }, { "epoch": 0.9208513992905006, "grad_norm": 0.7556031942367554, "learning_rate": 0.0005, "loss": 1.9927117824554443, "step": 11681 }, { "epoch": 0.9209302325581395, "grad_norm": 0.6276370882987976, "learning_rate": 0.0005, "loss": 2.0747480392456055, "step": 11682 }, { "epoch": 0.9210090658257785, "grad_norm": 0.7361471652984619, "learning_rate": 0.0005, "loss": 1.9928460121154785, "step": 11683 }, { "epoch": 0.9210878990934174, "grad_norm": 0.6503912806510925, "learning_rate": 0.0005, "loss": 1.7171871662139893, "step": 11684 }, { "epoch": 0.9211667323610564, "grad_norm": 0.6011273264884949, "learning_rate": 0.0005, "loss": 1.8714585304260254, "step": 11685 }, { "epoch": 0.9212455656286953, "grad_norm": 0.6166703701019287, "learning_rate": 0.0005, "loss": 1.503212809562683, "step": 11686 }, { "epoch": 0.9213243988963342, "grad_norm": 0.6949473023414612, "learning_rate": 0.0005, "loss": 1.5611159801483154, "step": 11687 }, { "epoch": 0.9214032321639732, "grad_norm": 0.5420535206794739, "learning_rate": 0.0005, "loss": 1.6399407386779785, "step": 11688 }, { "epoch": 0.9214820654316122, "grad_norm": 0.6862473487854004, "learning_rate": 0.0005, "loss": 1.7239536046981812, "step": 11689 }, { "epoch": 0.9215608986992511, "grad_norm": 0.5597370266914368, "learning_rate": 0.0005, "loss": 1.8207398653030396, "step": 11690 }, { "epoch": 0.92163973196689, "grad_norm": 0.5944098830223083, "learning_rate": 0.0005, "loss": 1.899059534072876, "step": 11691 }, { "epoch": 0.921718565234529, "grad_norm": 0.7655537128448486, "learning_rate": 0.0005, "loss": 1.8811482191085815, "step": 11692 }, { "epoch": 0.9217973985021679, "grad_norm": 0.49467581510543823, "learning_rate": 0.0005, "loss": 1.8165830373764038, "step": 11693 }, { "epoch": 0.9218762317698068, "grad_norm": 0.6981571912765503, "learning_rate": 0.0005, "loss": 1.7529383897781372, "step": 11694 }, { "epoch": 0.9219550650374458, "grad_norm": 0.5053221583366394, "learning_rate": 0.0005, "loss": 1.737255334854126, "step": 11695 }, { "epoch": 0.9220338983050848, "grad_norm": 0.749805212020874, "learning_rate": 0.0005, "loss": 2.026132345199585, "step": 11696 }, { "epoch": 0.9221127315727237, "grad_norm": 0.5598634481430054, "learning_rate": 0.0005, "loss": 1.3551709651947021, "step": 11697 }, { "epoch": 0.9221915648403627, "grad_norm": 0.6524683237075806, "learning_rate": 0.0005, "loss": 1.864228367805481, "step": 11698 }, { "epoch": 0.9222703981080016, "grad_norm": 0.5979815125465393, "learning_rate": 0.0005, "loss": 1.8662546873092651, "step": 11699 }, { "epoch": 0.9223492313756405, "grad_norm": 0.566249668598175, "learning_rate": 0.0005, "loss": 1.797123670578003, "step": 11700 }, { "epoch": 0.9224280646432794, "grad_norm": 0.5854669213294983, "learning_rate": 0.0005, "loss": 1.717658519744873, "step": 11701 }, { "epoch": 0.9225068979109184, "grad_norm": 0.6484285593032837, "learning_rate": 0.0005, "loss": 1.7377599477767944, "step": 11702 }, { "epoch": 0.9225857311785574, "grad_norm": 0.7494165301322937, "learning_rate": 0.0005, "loss": 1.7641775608062744, "step": 11703 }, { "epoch": 0.9226645644461963, "grad_norm": 0.592620313167572, "learning_rate": 0.0005, "loss": 1.579888105392456, "step": 11704 }, { "epoch": 0.9227433977138353, "grad_norm": 0.5988560914993286, "learning_rate": 0.0005, "loss": 1.6134487390518188, "step": 11705 }, { "epoch": 0.9228222309814742, "grad_norm": 0.6449071764945984, "learning_rate": 0.0005, "loss": 1.6702262163162231, "step": 11706 }, { "epoch": 0.9229010642491131, "grad_norm": 0.6460862755775452, "learning_rate": 0.0005, "loss": 1.9071803092956543, "step": 11707 }, { "epoch": 0.922979897516752, "grad_norm": 0.9197508096694946, "learning_rate": 0.0005, "loss": 2.1303019523620605, "step": 11708 }, { "epoch": 0.923058730784391, "grad_norm": 0.5665455460548401, "learning_rate": 0.0005, "loss": 2.060150623321533, "step": 11709 }, { "epoch": 0.92313756405203, "grad_norm": 0.779328465461731, "learning_rate": 0.0005, "loss": 1.8832041025161743, "step": 11710 }, { "epoch": 0.9232163973196689, "grad_norm": 0.6232807040214539, "learning_rate": 0.0005, "loss": 1.7050261497497559, "step": 11711 }, { "epoch": 0.9232952305873079, "grad_norm": 0.7987861037254333, "learning_rate": 0.0005, "loss": 2.052236557006836, "step": 11712 }, { "epoch": 0.9233740638549468, "grad_norm": 0.636489987373352, "learning_rate": 0.0005, "loss": 1.7368921041488647, "step": 11713 }, { "epoch": 0.9234528971225857, "grad_norm": 0.7123619914054871, "learning_rate": 0.0005, "loss": 1.666930913925171, "step": 11714 }, { "epoch": 0.9235317303902246, "grad_norm": 0.6516141891479492, "learning_rate": 0.0005, "loss": 1.891728162765503, "step": 11715 }, { "epoch": 0.9236105636578636, "grad_norm": 0.6871906518936157, "learning_rate": 0.0005, "loss": 1.8428997993469238, "step": 11716 }, { "epoch": 0.9236893969255026, "grad_norm": 0.6699273586273193, "learning_rate": 0.0005, "loss": 1.8521208763122559, "step": 11717 }, { "epoch": 0.9237682301931415, "grad_norm": 0.5160558223724365, "learning_rate": 0.0005, "loss": 1.8060485124588013, "step": 11718 }, { "epoch": 0.9238470634607805, "grad_norm": 0.5620858073234558, "learning_rate": 0.0005, "loss": 1.9042996168136597, "step": 11719 }, { "epoch": 0.9239258967284194, "grad_norm": 0.5592208504676819, "learning_rate": 0.0005, "loss": 1.5521970987319946, "step": 11720 }, { "epoch": 0.9240047299960583, "grad_norm": 0.7506398558616638, "learning_rate": 0.0005, "loss": 1.6291053295135498, "step": 11721 }, { "epoch": 0.9240835632636972, "grad_norm": 1.3051671981811523, "learning_rate": 0.0005, "loss": 1.955834150314331, "step": 11722 }, { "epoch": 0.9241623965313362, "grad_norm": 0.6001757383346558, "learning_rate": 0.0005, "loss": 1.8055386543273926, "step": 11723 }, { "epoch": 0.9242412297989752, "grad_norm": 0.7513509392738342, "learning_rate": 0.0005, "loss": 2.084800958633423, "step": 11724 }, { "epoch": 0.9243200630666141, "grad_norm": 0.4739287495613098, "learning_rate": 0.0005, "loss": 1.8419227600097656, "step": 11725 }, { "epoch": 0.9243988963342531, "grad_norm": 0.6543392539024353, "learning_rate": 0.0005, "loss": 1.9043669700622559, "step": 11726 }, { "epoch": 0.924477729601892, "grad_norm": 0.6587169766426086, "learning_rate": 0.0005, "loss": 2.0430750846862793, "step": 11727 }, { "epoch": 0.9245565628695309, "grad_norm": 0.7868324518203735, "learning_rate": 0.0005, "loss": 1.7394962310791016, "step": 11728 }, { "epoch": 0.9246353961371699, "grad_norm": 0.6676275730133057, "learning_rate": 0.0005, "loss": 1.8670272827148438, "step": 11729 }, { "epoch": 0.9247142294048089, "grad_norm": 0.5621129870414734, "learning_rate": 0.0005, "loss": 1.5162241458892822, "step": 11730 }, { "epoch": 0.9247930626724478, "grad_norm": 0.6298297643661499, "learning_rate": 0.0005, "loss": 1.9153413772583008, "step": 11731 }, { "epoch": 0.9248718959400867, "grad_norm": 0.8796188235282898, "learning_rate": 0.0005, "loss": 1.991605281829834, "step": 11732 }, { "epoch": 0.9249507292077257, "grad_norm": 0.6562861800193787, "learning_rate": 0.0005, "loss": 2.1877827644348145, "step": 11733 }, { "epoch": 0.9250295624753646, "grad_norm": 0.6373978853225708, "learning_rate": 0.0005, "loss": 1.92452073097229, "step": 11734 }, { "epoch": 0.9251083957430035, "grad_norm": 0.5481263995170593, "learning_rate": 0.0005, "loss": 1.864302158355713, "step": 11735 }, { "epoch": 0.9251872290106425, "grad_norm": 1.0913289785385132, "learning_rate": 0.0005, "loss": 1.8786240816116333, "step": 11736 }, { "epoch": 0.9252660622782815, "grad_norm": 0.6413065195083618, "learning_rate": 0.0005, "loss": 1.8481322526931763, "step": 11737 }, { "epoch": 0.9253448955459204, "grad_norm": 0.6940206289291382, "learning_rate": 0.0005, "loss": 1.8490631580352783, "step": 11738 }, { "epoch": 0.9254237288135593, "grad_norm": 0.6011402606964111, "learning_rate": 0.0005, "loss": 1.5382497310638428, "step": 11739 }, { "epoch": 0.9255025620811983, "grad_norm": 0.6352910399436951, "learning_rate": 0.0005, "loss": 1.669490933418274, "step": 11740 }, { "epoch": 0.9255813953488372, "grad_norm": 0.5104634165763855, "learning_rate": 0.0005, "loss": 1.8535032272338867, "step": 11741 }, { "epoch": 0.9256602286164761, "grad_norm": 0.7170076966285706, "learning_rate": 0.0005, "loss": 1.8547747135162354, "step": 11742 }, { "epoch": 0.9257390618841151, "grad_norm": 0.5105710029602051, "learning_rate": 0.0005, "loss": 1.893724799156189, "step": 11743 }, { "epoch": 0.9258178951517541, "grad_norm": 0.5383216738700867, "learning_rate": 0.0005, "loss": 2.0072665214538574, "step": 11744 }, { "epoch": 0.925896728419393, "grad_norm": 0.6573003530502319, "learning_rate": 0.0005, "loss": 1.7173964977264404, "step": 11745 }, { "epoch": 0.9259755616870319, "grad_norm": 0.6727480888366699, "learning_rate": 0.0005, "loss": 1.9760780334472656, "step": 11746 }, { "epoch": 0.9260543949546709, "grad_norm": 0.7522886395454407, "learning_rate": 0.0005, "loss": 2.01764178276062, "step": 11747 }, { "epoch": 0.9261332282223098, "grad_norm": 0.6639924645423889, "learning_rate": 0.0005, "loss": 1.7509543895721436, "step": 11748 }, { "epoch": 0.9262120614899487, "grad_norm": 0.6898094415664673, "learning_rate": 0.0005, "loss": 1.8535712957382202, "step": 11749 }, { "epoch": 0.9262908947575877, "grad_norm": 0.6645023226737976, "learning_rate": 0.0005, "loss": 1.8202310800552368, "step": 11750 }, { "epoch": 0.9263697280252267, "grad_norm": 0.6323335766792297, "learning_rate": 0.0005, "loss": 1.974142074584961, "step": 11751 }, { "epoch": 0.9264485612928656, "grad_norm": 0.5838978886604309, "learning_rate": 0.0005, "loss": 1.6220743656158447, "step": 11752 }, { "epoch": 0.9265273945605045, "grad_norm": 0.8510795831680298, "learning_rate": 0.0005, "loss": 1.6449278593063354, "step": 11753 }, { "epoch": 0.9266062278281435, "grad_norm": 0.6658623814582825, "learning_rate": 0.0005, "loss": 1.6661646366119385, "step": 11754 }, { "epoch": 0.9266850610957824, "grad_norm": 0.6351900696754456, "learning_rate": 0.0005, "loss": 1.8523756265640259, "step": 11755 }, { "epoch": 0.9267638943634213, "grad_norm": 0.6734033823013306, "learning_rate": 0.0005, "loss": 1.7872556447982788, "step": 11756 }, { "epoch": 0.9268427276310603, "grad_norm": 0.6932073831558228, "learning_rate": 0.0005, "loss": 1.9126856327056885, "step": 11757 }, { "epoch": 0.9269215608986993, "grad_norm": 0.7511993050575256, "learning_rate": 0.0005, "loss": 1.7860257625579834, "step": 11758 }, { "epoch": 0.9270003941663382, "grad_norm": 0.609956681728363, "learning_rate": 0.0005, "loss": 2.010080099105835, "step": 11759 }, { "epoch": 0.9270792274339772, "grad_norm": 0.5597667694091797, "learning_rate": 0.0005, "loss": 1.9248418807983398, "step": 11760 }, { "epoch": 0.9271580607016161, "grad_norm": 0.8075128197669983, "learning_rate": 0.0005, "loss": 1.7497918605804443, "step": 11761 }, { "epoch": 0.927236893969255, "grad_norm": 0.5260193347930908, "learning_rate": 0.0005, "loss": 1.506680965423584, "step": 11762 }, { "epoch": 0.9273157272368939, "grad_norm": 0.5232863426208496, "learning_rate": 0.0005, "loss": 1.6912400722503662, "step": 11763 }, { "epoch": 0.927394560504533, "grad_norm": 0.6510034799575806, "learning_rate": 0.0005, "loss": 1.955521821975708, "step": 11764 }, { "epoch": 0.9274733937721719, "grad_norm": 0.6309261918067932, "learning_rate": 0.0005, "loss": 1.5647435188293457, "step": 11765 }, { "epoch": 0.9275522270398108, "grad_norm": 0.8839739561080933, "learning_rate": 0.0005, "loss": 2.052232265472412, "step": 11766 }, { "epoch": 0.9276310603074498, "grad_norm": 0.7179363369941711, "learning_rate": 0.0005, "loss": 1.8092787265777588, "step": 11767 }, { "epoch": 0.9277098935750887, "grad_norm": 0.5543674230575562, "learning_rate": 0.0005, "loss": 2.063675880432129, "step": 11768 }, { "epoch": 0.9277887268427276, "grad_norm": 0.617184579372406, "learning_rate": 0.0005, "loss": 1.7680611610412598, "step": 11769 }, { "epoch": 0.9278675601103665, "grad_norm": 0.6268219351768494, "learning_rate": 0.0005, "loss": 1.648241400718689, "step": 11770 }, { "epoch": 0.9279463933780056, "grad_norm": 0.6679320335388184, "learning_rate": 0.0005, "loss": 2.132535934448242, "step": 11771 }, { "epoch": 0.9280252266456445, "grad_norm": 0.530750036239624, "learning_rate": 0.0005, "loss": 1.611093282699585, "step": 11772 }, { "epoch": 0.9281040599132834, "grad_norm": 0.49461889266967773, "learning_rate": 0.0005, "loss": 1.4729595184326172, "step": 11773 }, { "epoch": 0.9281828931809224, "grad_norm": 0.6326369047164917, "learning_rate": 0.0005, "loss": 1.7035568952560425, "step": 11774 }, { "epoch": 0.9282617264485613, "grad_norm": 0.623119056224823, "learning_rate": 0.0005, "loss": 1.667754054069519, "step": 11775 }, { "epoch": 0.9283405597162002, "grad_norm": 0.64737468957901, "learning_rate": 0.0005, "loss": 1.961017370223999, "step": 11776 }, { "epoch": 0.9284193929838391, "grad_norm": 0.5704742670059204, "learning_rate": 0.0005, "loss": 1.6974334716796875, "step": 11777 }, { "epoch": 0.9284982262514782, "grad_norm": 0.6948301196098328, "learning_rate": 0.0005, "loss": 1.6966471672058105, "step": 11778 }, { "epoch": 0.9285770595191171, "grad_norm": 0.6677987575531006, "learning_rate": 0.0005, "loss": 1.9874759912490845, "step": 11779 }, { "epoch": 0.928655892786756, "grad_norm": 0.6119590401649475, "learning_rate": 0.0005, "loss": 1.697221040725708, "step": 11780 }, { "epoch": 0.928734726054395, "grad_norm": 0.6194531321525574, "learning_rate": 0.0005, "loss": 1.7608425617218018, "step": 11781 }, { "epoch": 0.9288135593220339, "grad_norm": 0.5492265820503235, "learning_rate": 0.0005, "loss": 1.7192294597625732, "step": 11782 }, { "epoch": 0.9288923925896728, "grad_norm": 0.5842873454093933, "learning_rate": 0.0005, "loss": 1.7096691131591797, "step": 11783 }, { "epoch": 0.9289712258573117, "grad_norm": 0.7455161809921265, "learning_rate": 0.0005, "loss": 1.9746323823928833, "step": 11784 }, { "epoch": 0.9290500591249508, "grad_norm": 2.6745307445526123, "learning_rate": 0.0005, "loss": 1.8385422229766846, "step": 11785 }, { "epoch": 0.9291288923925897, "grad_norm": 0.8289634585380554, "learning_rate": 0.0005, "loss": 1.801072597503662, "step": 11786 }, { "epoch": 0.9292077256602286, "grad_norm": 0.6480610370635986, "learning_rate": 0.0005, "loss": 2.102478265762329, "step": 11787 }, { "epoch": 0.9292865589278676, "grad_norm": 0.6341183185577393, "learning_rate": 0.0005, "loss": 1.933265209197998, "step": 11788 }, { "epoch": 0.9293653921955065, "grad_norm": 0.5659371614456177, "learning_rate": 0.0005, "loss": 1.797839879989624, "step": 11789 }, { "epoch": 0.9294442254631454, "grad_norm": 0.6042575240135193, "learning_rate": 0.0005, "loss": 1.7137811183929443, "step": 11790 }, { "epoch": 0.9295230587307844, "grad_norm": 0.6639628410339355, "learning_rate": 0.0005, "loss": 2.07136607170105, "step": 11791 }, { "epoch": 0.9296018919984234, "grad_norm": 0.5681605935096741, "learning_rate": 0.0005, "loss": 1.7669126987457275, "step": 11792 }, { "epoch": 0.9296807252660623, "grad_norm": 0.6293914318084717, "learning_rate": 0.0005, "loss": 1.5536952018737793, "step": 11793 }, { "epoch": 0.9297595585337012, "grad_norm": 0.583259642124176, "learning_rate": 0.0005, "loss": 1.810204029083252, "step": 11794 }, { "epoch": 0.9298383918013402, "grad_norm": 0.5705042481422424, "learning_rate": 0.0005, "loss": 1.9542198181152344, "step": 11795 }, { "epoch": 0.9299172250689791, "grad_norm": 0.5440019965171814, "learning_rate": 0.0005, "loss": 1.9230409860610962, "step": 11796 }, { "epoch": 0.929996058336618, "grad_norm": 0.8085290789604187, "learning_rate": 0.0005, "loss": 1.7064354419708252, "step": 11797 }, { "epoch": 0.930074891604257, "grad_norm": 1.1368333101272583, "learning_rate": 0.0005, "loss": 2.247493267059326, "step": 11798 }, { "epoch": 0.930153724871896, "grad_norm": 0.6118596792221069, "learning_rate": 0.0005, "loss": 1.7308706045150757, "step": 11799 }, { "epoch": 0.9302325581395349, "grad_norm": 0.6369962096214294, "learning_rate": 0.0005, "loss": 1.6674803495407104, "step": 11800 }, { "epoch": 0.9303113914071738, "grad_norm": 0.8392461538314819, "learning_rate": 0.0005, "loss": 1.9682598114013672, "step": 11801 }, { "epoch": 0.9303902246748128, "grad_norm": 0.6891761422157288, "learning_rate": 0.0005, "loss": 1.5941941738128662, "step": 11802 }, { "epoch": 0.9304690579424517, "grad_norm": 0.6916091442108154, "learning_rate": 0.0005, "loss": 2.083325147628784, "step": 11803 }, { "epoch": 0.9305478912100906, "grad_norm": 0.8881251215934753, "learning_rate": 0.0005, "loss": 1.7674543857574463, "step": 11804 }, { "epoch": 0.9306267244777296, "grad_norm": 0.5910885334014893, "learning_rate": 0.0005, "loss": 1.803503394126892, "step": 11805 }, { "epoch": 0.9307055577453686, "grad_norm": 0.6997777223587036, "learning_rate": 0.0005, "loss": 1.8855584859848022, "step": 11806 }, { "epoch": 0.9307843910130075, "grad_norm": 0.5514601469039917, "learning_rate": 0.0005, "loss": 1.9419629573822021, "step": 11807 }, { "epoch": 0.9308632242806464, "grad_norm": 0.6711503267288208, "learning_rate": 0.0005, "loss": 1.814619541168213, "step": 11808 }, { "epoch": 0.9309420575482854, "grad_norm": 0.6768834590911865, "learning_rate": 0.0005, "loss": 2.0310328006744385, "step": 11809 }, { "epoch": 0.9310208908159243, "grad_norm": 0.6148741245269775, "learning_rate": 0.0005, "loss": 1.8344310522079468, "step": 11810 }, { "epoch": 0.9310997240835632, "grad_norm": 0.5384054183959961, "learning_rate": 0.0005, "loss": 1.7128729820251465, "step": 11811 }, { "epoch": 0.9311785573512023, "grad_norm": 0.5892127156257629, "learning_rate": 0.0005, "loss": 1.8646736145019531, "step": 11812 }, { "epoch": 0.9312573906188412, "grad_norm": 0.7580090761184692, "learning_rate": 0.0005, "loss": 1.9040439128875732, "step": 11813 }, { "epoch": 0.9313362238864801, "grad_norm": 0.7546406984329224, "learning_rate": 0.0005, "loss": 1.885408639907837, "step": 11814 }, { "epoch": 0.931415057154119, "grad_norm": 0.900594174861908, "learning_rate": 0.0005, "loss": 1.978864073753357, "step": 11815 }, { "epoch": 0.931493890421758, "grad_norm": 0.8201185464859009, "learning_rate": 0.0005, "loss": 1.9858421087265015, "step": 11816 }, { "epoch": 0.9315727236893969, "grad_norm": 0.6739137768745422, "learning_rate": 0.0005, "loss": 1.894073247909546, "step": 11817 }, { "epoch": 0.9316515569570358, "grad_norm": 0.7110694646835327, "learning_rate": 0.0005, "loss": 1.8328458070755005, "step": 11818 }, { "epoch": 0.9317303902246749, "grad_norm": 0.6960161924362183, "learning_rate": 0.0005, "loss": 1.7095718383789062, "step": 11819 }, { "epoch": 0.9318092234923138, "grad_norm": 0.6561651825904846, "learning_rate": 0.0005, "loss": 1.911726951599121, "step": 11820 }, { "epoch": 0.9318880567599527, "grad_norm": 0.6210526823997498, "learning_rate": 0.0005, "loss": 1.6118665933609009, "step": 11821 }, { "epoch": 0.9319668900275917, "grad_norm": 0.9762800335884094, "learning_rate": 0.0005, "loss": 1.730027198791504, "step": 11822 }, { "epoch": 0.9320457232952306, "grad_norm": 0.686276912689209, "learning_rate": 0.0005, "loss": 2.0109658241271973, "step": 11823 }, { "epoch": 0.9321245565628695, "grad_norm": 0.605945885181427, "learning_rate": 0.0005, "loss": 1.6832945346832275, "step": 11824 }, { "epoch": 0.9322033898305084, "grad_norm": 0.6914631724357605, "learning_rate": 0.0005, "loss": 1.9762747287750244, "step": 11825 }, { "epoch": 0.9322822230981475, "grad_norm": 0.6205142140388489, "learning_rate": 0.0005, "loss": 1.6436214447021484, "step": 11826 }, { "epoch": 0.9323610563657864, "grad_norm": 0.5721641182899475, "learning_rate": 0.0005, "loss": 1.8153841495513916, "step": 11827 }, { "epoch": 0.9324398896334253, "grad_norm": 0.5957128405570984, "learning_rate": 0.0005, "loss": 1.8070131540298462, "step": 11828 }, { "epoch": 0.9325187229010643, "grad_norm": 0.6183828711509705, "learning_rate": 0.0005, "loss": 1.8140122890472412, "step": 11829 }, { "epoch": 0.9325975561687032, "grad_norm": 0.5862563848495483, "learning_rate": 0.0005, "loss": 1.7630980014801025, "step": 11830 }, { "epoch": 0.9326763894363421, "grad_norm": 0.8656314611434937, "learning_rate": 0.0005, "loss": 1.8005249500274658, "step": 11831 }, { "epoch": 0.932755222703981, "grad_norm": 0.5658913254737854, "learning_rate": 0.0005, "loss": 1.8365812301635742, "step": 11832 }, { "epoch": 0.9328340559716201, "grad_norm": 0.6304697394371033, "learning_rate": 0.0005, "loss": 1.7232853174209595, "step": 11833 }, { "epoch": 0.932912889239259, "grad_norm": 0.6575843095779419, "learning_rate": 0.0005, "loss": 2.083965539932251, "step": 11834 }, { "epoch": 0.9329917225068979, "grad_norm": 0.6710783243179321, "learning_rate": 0.0005, "loss": 1.6123108863830566, "step": 11835 }, { "epoch": 0.9330705557745369, "grad_norm": 0.6125486493110657, "learning_rate": 0.0005, "loss": 1.9720784425735474, "step": 11836 }, { "epoch": 0.9331493890421758, "grad_norm": 1.985022783279419, "learning_rate": 0.0005, "loss": 1.9196853637695312, "step": 11837 }, { "epoch": 0.9332282223098147, "grad_norm": 0.7255325317382812, "learning_rate": 0.0005, "loss": 1.7704899311065674, "step": 11838 }, { "epoch": 0.9333070555774536, "grad_norm": 0.6422814726829529, "learning_rate": 0.0005, "loss": 1.74650239944458, "step": 11839 }, { "epoch": 0.9333858888450927, "grad_norm": 0.502720057964325, "learning_rate": 0.0005, "loss": 1.4504905939102173, "step": 11840 }, { "epoch": 0.9334647221127316, "grad_norm": 0.7153573036193848, "learning_rate": 0.0005, "loss": 1.7137486934661865, "step": 11841 }, { "epoch": 0.9335435553803705, "grad_norm": 0.7221958637237549, "learning_rate": 0.0005, "loss": 1.9140313863754272, "step": 11842 }, { "epoch": 0.9336223886480095, "grad_norm": 0.6548260450363159, "learning_rate": 0.0005, "loss": 1.8492401838302612, "step": 11843 }, { "epoch": 0.9337012219156484, "grad_norm": 0.5552295446395874, "learning_rate": 0.0005, "loss": 1.7552742958068848, "step": 11844 }, { "epoch": 0.9337800551832873, "grad_norm": 0.7645944356918335, "learning_rate": 0.0005, "loss": 1.9676709175109863, "step": 11845 }, { "epoch": 0.9338588884509262, "grad_norm": 0.597312331199646, "learning_rate": 0.0005, "loss": 1.8593239784240723, "step": 11846 }, { "epoch": 0.9339377217185653, "grad_norm": 0.680240273475647, "learning_rate": 0.0005, "loss": 1.8185510635375977, "step": 11847 }, { "epoch": 0.9340165549862042, "grad_norm": 0.5687347650527954, "learning_rate": 0.0005, "loss": 1.769507646560669, "step": 11848 }, { "epoch": 0.9340953882538431, "grad_norm": 1.0254809856414795, "learning_rate": 0.0005, "loss": 1.9788278341293335, "step": 11849 }, { "epoch": 0.9341742215214821, "grad_norm": 0.9231231212615967, "learning_rate": 0.0005, "loss": 1.8486639261245728, "step": 11850 }, { "epoch": 0.934253054789121, "grad_norm": 1.0309175252914429, "learning_rate": 0.0005, "loss": 1.897905707359314, "step": 11851 }, { "epoch": 0.9343318880567599, "grad_norm": 0.5507082343101501, "learning_rate": 0.0005, "loss": 1.6292390823364258, "step": 11852 }, { "epoch": 0.934410721324399, "grad_norm": 0.5645356774330139, "learning_rate": 0.0005, "loss": 1.7046806812286377, "step": 11853 }, { "epoch": 0.9344895545920379, "grad_norm": 0.5575624108314514, "learning_rate": 0.0005, "loss": 1.538313388824463, "step": 11854 }, { "epoch": 0.9345683878596768, "grad_norm": 0.7110762596130371, "learning_rate": 0.0005, "loss": 1.653233528137207, "step": 11855 }, { "epoch": 0.9346472211273157, "grad_norm": 0.7180546522140503, "learning_rate": 0.0005, "loss": 2.0458767414093018, "step": 11856 }, { "epoch": 0.9347260543949547, "grad_norm": 0.6506874561309814, "learning_rate": 0.0005, "loss": 1.908468246459961, "step": 11857 }, { "epoch": 0.9348048876625936, "grad_norm": 0.6226797103881836, "learning_rate": 0.0005, "loss": 1.5025644302368164, "step": 11858 }, { "epoch": 0.9348837209302325, "grad_norm": 0.6443557143211365, "learning_rate": 0.0005, "loss": 1.8986308574676514, "step": 11859 }, { "epoch": 0.9349625541978716, "grad_norm": 0.5356549620628357, "learning_rate": 0.0005, "loss": 1.6125190258026123, "step": 11860 }, { "epoch": 0.9350413874655105, "grad_norm": 0.8013724088668823, "learning_rate": 0.0005, "loss": 1.9977078437805176, "step": 11861 }, { "epoch": 0.9351202207331494, "grad_norm": 0.72828209400177, "learning_rate": 0.0005, "loss": 1.839169979095459, "step": 11862 }, { "epoch": 0.9351990540007883, "grad_norm": 0.5959152579307556, "learning_rate": 0.0005, "loss": 2.085175037384033, "step": 11863 }, { "epoch": 0.9352778872684273, "grad_norm": 0.6131113767623901, "learning_rate": 0.0005, "loss": 1.4431424140930176, "step": 11864 }, { "epoch": 0.9353567205360662, "grad_norm": 0.669998824596405, "learning_rate": 0.0005, "loss": 1.82765793800354, "step": 11865 }, { "epoch": 0.9354355538037051, "grad_norm": 0.6227160096168518, "learning_rate": 0.0005, "loss": 1.8150081634521484, "step": 11866 }, { "epoch": 0.9355143870713442, "grad_norm": 0.6716116070747375, "learning_rate": 0.0005, "loss": 2.2010459899902344, "step": 11867 }, { "epoch": 0.9355932203389831, "grad_norm": 0.5477124452590942, "learning_rate": 0.0005, "loss": 1.742507815361023, "step": 11868 }, { "epoch": 0.935672053606622, "grad_norm": 0.5870775580406189, "learning_rate": 0.0005, "loss": 1.7987351417541504, "step": 11869 }, { "epoch": 0.9357508868742609, "grad_norm": 0.6752505302429199, "learning_rate": 0.0005, "loss": 1.8537747859954834, "step": 11870 }, { "epoch": 0.9358297201418999, "grad_norm": 0.653796374797821, "learning_rate": 0.0005, "loss": 1.9068126678466797, "step": 11871 }, { "epoch": 0.9359085534095388, "grad_norm": 0.9703697562217712, "learning_rate": 0.0005, "loss": 1.8557884693145752, "step": 11872 }, { "epoch": 0.9359873866771777, "grad_norm": 0.5378881692886353, "learning_rate": 0.0005, "loss": 1.7576870918273926, "step": 11873 }, { "epoch": 0.9360662199448168, "grad_norm": 0.6872445344924927, "learning_rate": 0.0005, "loss": 1.6095517873764038, "step": 11874 }, { "epoch": 0.9361450532124557, "grad_norm": 0.5398042798042297, "learning_rate": 0.0005, "loss": 1.624502420425415, "step": 11875 }, { "epoch": 0.9362238864800946, "grad_norm": 0.646535336971283, "learning_rate": 0.0005, "loss": 1.8660616874694824, "step": 11876 }, { "epoch": 0.9363027197477335, "grad_norm": 0.4997541904449463, "learning_rate": 0.0005, "loss": 1.7691385746002197, "step": 11877 }, { "epoch": 0.9363815530153725, "grad_norm": 0.7872259020805359, "learning_rate": 0.0005, "loss": 1.9028339385986328, "step": 11878 }, { "epoch": 0.9364603862830114, "grad_norm": 0.6904066801071167, "learning_rate": 0.0005, "loss": 1.7343480587005615, "step": 11879 }, { "epoch": 0.9365392195506503, "grad_norm": 0.5234655141830444, "learning_rate": 0.0005, "loss": 1.7982065677642822, "step": 11880 }, { "epoch": 0.9366180528182894, "grad_norm": 0.6475844383239746, "learning_rate": 0.0005, "loss": 2.049776315689087, "step": 11881 }, { "epoch": 0.9366968860859283, "grad_norm": 0.9848289489746094, "learning_rate": 0.0005, "loss": 2.0870375633239746, "step": 11882 }, { "epoch": 0.9367757193535672, "grad_norm": 0.7313181757926941, "learning_rate": 0.0005, "loss": 1.8704636096954346, "step": 11883 }, { "epoch": 0.9368545526212061, "grad_norm": 0.7614474892616272, "learning_rate": 0.0005, "loss": 1.8217623233795166, "step": 11884 }, { "epoch": 0.9369333858888451, "grad_norm": 0.6099168062210083, "learning_rate": 0.0005, "loss": 1.7064341306686401, "step": 11885 }, { "epoch": 0.937012219156484, "grad_norm": 0.7625274658203125, "learning_rate": 0.0005, "loss": 2.0239336490631104, "step": 11886 }, { "epoch": 0.9370910524241229, "grad_norm": 0.9789429903030396, "learning_rate": 0.0005, "loss": 2.0408592224121094, "step": 11887 }, { "epoch": 0.937169885691762, "grad_norm": 0.5880967378616333, "learning_rate": 0.0005, "loss": 1.6364107131958008, "step": 11888 }, { "epoch": 0.9372487189594009, "grad_norm": 0.7609308958053589, "learning_rate": 0.0005, "loss": 1.6865906715393066, "step": 11889 }, { "epoch": 0.9373275522270398, "grad_norm": 0.5676504969596863, "learning_rate": 0.0005, "loss": 1.9077450037002563, "step": 11890 }, { "epoch": 0.9374063854946788, "grad_norm": 0.7055366039276123, "learning_rate": 0.0005, "loss": 2.1134746074676514, "step": 11891 }, { "epoch": 0.9374852187623177, "grad_norm": 0.5907406806945801, "learning_rate": 0.0005, "loss": 1.7634282112121582, "step": 11892 }, { "epoch": 0.9375640520299566, "grad_norm": 0.586661159992218, "learning_rate": 0.0005, "loss": 1.7593268156051636, "step": 11893 }, { "epoch": 0.9376428852975955, "grad_norm": 0.6135310530662537, "learning_rate": 0.0005, "loss": 1.9309499263763428, "step": 11894 }, { "epoch": 0.9377217185652346, "grad_norm": 0.693389892578125, "learning_rate": 0.0005, "loss": 1.7428385019302368, "step": 11895 }, { "epoch": 0.9378005518328735, "grad_norm": 0.5899178385734558, "learning_rate": 0.0005, "loss": 1.8978179693222046, "step": 11896 }, { "epoch": 0.9378793851005124, "grad_norm": 0.6092788577079773, "learning_rate": 0.0005, "loss": 1.8690578937530518, "step": 11897 }, { "epoch": 0.9379582183681514, "grad_norm": 0.6053569912910461, "learning_rate": 0.0005, "loss": 1.9062232971191406, "step": 11898 }, { "epoch": 0.9380370516357903, "grad_norm": 0.7091948390007019, "learning_rate": 0.0005, "loss": 1.8540538549423218, "step": 11899 }, { "epoch": 0.9381158849034292, "grad_norm": 0.5572349429130554, "learning_rate": 0.0005, "loss": 1.7983241081237793, "step": 11900 }, { "epoch": 0.9381947181710681, "grad_norm": 0.8351790904998779, "learning_rate": 0.0005, "loss": 2.056230068206787, "step": 11901 }, { "epoch": 0.9382735514387072, "grad_norm": 0.681461751461029, "learning_rate": 0.0005, "loss": 2.070329189300537, "step": 11902 }, { "epoch": 0.9383523847063461, "grad_norm": 0.6058235764503479, "learning_rate": 0.0005, "loss": 1.8460972309112549, "step": 11903 }, { "epoch": 0.938431217973985, "grad_norm": 0.7315847873687744, "learning_rate": 0.0005, "loss": 2.1477713584899902, "step": 11904 }, { "epoch": 0.938510051241624, "grad_norm": 0.625580370426178, "learning_rate": 0.0005, "loss": 1.9231748580932617, "step": 11905 }, { "epoch": 0.9385888845092629, "grad_norm": 0.7806404232978821, "learning_rate": 0.0005, "loss": 1.7356960773468018, "step": 11906 }, { "epoch": 0.9386677177769018, "grad_norm": 0.653266429901123, "learning_rate": 0.0005, "loss": 1.9608533382415771, "step": 11907 }, { "epoch": 0.9387465510445407, "grad_norm": 0.5088934302330017, "learning_rate": 0.0005, "loss": 1.794631004333496, "step": 11908 }, { "epoch": 0.9388253843121798, "grad_norm": 0.6477068066596985, "learning_rate": 0.0005, "loss": 2.0650582313537598, "step": 11909 }, { "epoch": 0.9389042175798187, "grad_norm": 0.4838598966598511, "learning_rate": 0.0005, "loss": 1.5086067914962769, "step": 11910 }, { "epoch": 0.9389830508474576, "grad_norm": 0.6441128253936768, "learning_rate": 0.0005, "loss": 2.0100388526916504, "step": 11911 }, { "epoch": 0.9390618841150966, "grad_norm": 0.5184757113456726, "learning_rate": 0.0005, "loss": 1.6642861366271973, "step": 11912 }, { "epoch": 0.9391407173827355, "grad_norm": 0.7410992980003357, "learning_rate": 0.0005, "loss": 2.166834831237793, "step": 11913 }, { "epoch": 0.9392195506503744, "grad_norm": 0.49023720622062683, "learning_rate": 0.0005, "loss": 1.8844058513641357, "step": 11914 }, { "epoch": 0.9392983839180133, "grad_norm": 0.582018256187439, "learning_rate": 0.0005, "loss": 2.0202584266662598, "step": 11915 }, { "epoch": 0.9393772171856524, "grad_norm": 0.6224114298820496, "learning_rate": 0.0005, "loss": 2.032614231109619, "step": 11916 }, { "epoch": 0.9394560504532913, "grad_norm": 0.6959503293037415, "learning_rate": 0.0005, "loss": 1.9318076372146606, "step": 11917 }, { "epoch": 0.9395348837209302, "grad_norm": 0.7161077260971069, "learning_rate": 0.0005, "loss": 2.0737452507019043, "step": 11918 }, { "epoch": 0.9396137169885692, "grad_norm": 0.8832194209098816, "learning_rate": 0.0005, "loss": 1.5668537616729736, "step": 11919 }, { "epoch": 0.9396925502562081, "grad_norm": 0.6629918217658997, "learning_rate": 0.0005, "loss": 1.7289870977401733, "step": 11920 }, { "epoch": 0.939771383523847, "grad_norm": 0.6255016922950745, "learning_rate": 0.0005, "loss": 1.8758773803710938, "step": 11921 }, { "epoch": 0.9398502167914861, "grad_norm": 0.5143685936927795, "learning_rate": 0.0005, "loss": 1.5666584968566895, "step": 11922 }, { "epoch": 0.939929050059125, "grad_norm": 0.6233640313148499, "learning_rate": 0.0005, "loss": 1.6688549518585205, "step": 11923 }, { "epoch": 0.9400078833267639, "grad_norm": 0.6860897541046143, "learning_rate": 0.0005, "loss": 1.9234368801116943, "step": 11924 }, { "epoch": 0.9400867165944028, "grad_norm": 1.105316400527954, "learning_rate": 0.0005, "loss": 2.027937650680542, "step": 11925 }, { "epoch": 0.9401655498620418, "grad_norm": 0.6774095296859741, "learning_rate": 0.0005, "loss": 1.658325433731079, "step": 11926 }, { "epoch": 0.9402443831296807, "grad_norm": 0.6401793360710144, "learning_rate": 0.0005, "loss": 1.9111685752868652, "step": 11927 }, { "epoch": 0.9403232163973196, "grad_norm": 0.6738514304161072, "learning_rate": 0.0005, "loss": 1.7198845148086548, "step": 11928 }, { "epoch": 0.9404020496649587, "grad_norm": 0.765850841999054, "learning_rate": 0.0005, "loss": 2.090390682220459, "step": 11929 }, { "epoch": 0.9404808829325976, "grad_norm": 0.8137714266777039, "learning_rate": 0.0005, "loss": 1.8201630115509033, "step": 11930 }, { "epoch": 0.9405597162002365, "grad_norm": 0.8063976764678955, "learning_rate": 0.0005, "loss": 2.073525905609131, "step": 11931 }, { "epoch": 0.9406385494678754, "grad_norm": 0.7891557216644287, "learning_rate": 0.0005, "loss": 1.7498550415039062, "step": 11932 }, { "epoch": 0.9407173827355144, "grad_norm": 0.9485548138618469, "learning_rate": 0.0005, "loss": 1.6396164894104004, "step": 11933 }, { "epoch": 0.9407962160031533, "grad_norm": 0.8270146250724792, "learning_rate": 0.0005, "loss": 1.7196165323257446, "step": 11934 }, { "epoch": 0.9408750492707922, "grad_norm": 0.681107759475708, "learning_rate": 0.0005, "loss": 1.9577313661575317, "step": 11935 }, { "epoch": 0.9409538825384313, "grad_norm": 0.7193945646286011, "learning_rate": 0.0005, "loss": 1.7389109134674072, "step": 11936 }, { "epoch": 0.9410327158060702, "grad_norm": 0.5878910422325134, "learning_rate": 0.0005, "loss": 1.5731849670410156, "step": 11937 }, { "epoch": 0.9411115490737091, "grad_norm": 0.6244120001792908, "learning_rate": 0.0005, "loss": 1.7248034477233887, "step": 11938 }, { "epoch": 0.941190382341348, "grad_norm": 0.603855550289154, "learning_rate": 0.0005, "loss": 1.7760142087936401, "step": 11939 }, { "epoch": 0.941269215608987, "grad_norm": 0.7521638870239258, "learning_rate": 0.0005, "loss": 1.85760498046875, "step": 11940 }, { "epoch": 0.9413480488766259, "grad_norm": 0.9292328953742981, "learning_rate": 0.0005, "loss": 1.8689460754394531, "step": 11941 }, { "epoch": 0.9414268821442648, "grad_norm": 0.5736239552497864, "learning_rate": 0.0005, "loss": 1.8995634317398071, "step": 11942 }, { "epoch": 0.9415057154119039, "grad_norm": 0.6528165340423584, "learning_rate": 0.0005, "loss": 1.8606078624725342, "step": 11943 }, { "epoch": 0.9415845486795428, "grad_norm": 0.5965094566345215, "learning_rate": 0.0005, "loss": 1.8536591529846191, "step": 11944 }, { "epoch": 0.9416633819471817, "grad_norm": 0.524298906326294, "learning_rate": 0.0005, "loss": 1.8835656642913818, "step": 11945 }, { "epoch": 0.9417422152148206, "grad_norm": 1.1694438457489014, "learning_rate": 0.0005, "loss": 1.9425671100616455, "step": 11946 }, { "epoch": 0.9418210484824596, "grad_norm": 1.0145925283432007, "learning_rate": 0.0005, "loss": 1.8455836772918701, "step": 11947 }, { "epoch": 0.9418998817500985, "grad_norm": 0.6214426159858704, "learning_rate": 0.0005, "loss": 1.8015310764312744, "step": 11948 }, { "epoch": 0.9419787150177374, "grad_norm": 0.5647615790367126, "learning_rate": 0.0005, "loss": 1.62892484664917, "step": 11949 }, { "epoch": 0.9420575482853765, "grad_norm": 0.5265786647796631, "learning_rate": 0.0005, "loss": 1.6400196552276611, "step": 11950 }, { "epoch": 0.9421363815530154, "grad_norm": 0.5908535718917847, "learning_rate": 0.0005, "loss": 1.8454296588897705, "step": 11951 }, { "epoch": 0.9422152148206543, "grad_norm": 0.6472607254981995, "learning_rate": 0.0005, "loss": 1.486964464187622, "step": 11952 }, { "epoch": 0.9422940480882933, "grad_norm": 0.6448205709457397, "learning_rate": 0.0005, "loss": 1.8296966552734375, "step": 11953 }, { "epoch": 0.9423728813559322, "grad_norm": 1.250543475151062, "learning_rate": 0.0005, "loss": 1.6725294589996338, "step": 11954 }, { "epoch": 0.9424517146235711, "grad_norm": 0.7359920144081116, "learning_rate": 0.0005, "loss": 1.571315050125122, "step": 11955 }, { "epoch": 0.94253054789121, "grad_norm": 0.5593740940093994, "learning_rate": 0.0005, "loss": 1.8029193878173828, "step": 11956 }, { "epoch": 0.9426093811588491, "grad_norm": 0.7198768258094788, "learning_rate": 0.0005, "loss": 1.947839617729187, "step": 11957 }, { "epoch": 0.942688214426488, "grad_norm": 0.6274557113647461, "learning_rate": 0.0005, "loss": 1.7532758712768555, "step": 11958 }, { "epoch": 0.9427670476941269, "grad_norm": 0.8586755990982056, "learning_rate": 0.0005, "loss": 1.8444262742996216, "step": 11959 }, { "epoch": 0.9428458809617659, "grad_norm": 0.8724877238273621, "learning_rate": 0.0005, "loss": 1.937464714050293, "step": 11960 }, { "epoch": 0.9429247142294048, "grad_norm": 0.7203596830368042, "learning_rate": 0.0005, "loss": 1.8863269090652466, "step": 11961 }, { "epoch": 0.9430035474970437, "grad_norm": 0.623302161693573, "learning_rate": 0.0005, "loss": 1.7016215324401855, "step": 11962 }, { "epoch": 0.9430823807646826, "grad_norm": 0.5654927492141724, "learning_rate": 0.0005, "loss": 1.9388973712921143, "step": 11963 }, { "epoch": 0.9431612140323217, "grad_norm": 0.968311607837677, "learning_rate": 0.0005, "loss": 1.609975814819336, "step": 11964 }, { "epoch": 0.9432400472999606, "grad_norm": 0.6711878776550293, "learning_rate": 0.0005, "loss": 1.5690944194793701, "step": 11965 }, { "epoch": 0.9433188805675995, "grad_norm": 0.6913467645645142, "learning_rate": 0.0005, "loss": 1.8455654382705688, "step": 11966 }, { "epoch": 0.9433977138352385, "grad_norm": 0.7088236212730408, "learning_rate": 0.0005, "loss": 1.8404252529144287, "step": 11967 }, { "epoch": 0.9434765471028774, "grad_norm": 0.7778558731079102, "learning_rate": 0.0005, "loss": 1.9993298053741455, "step": 11968 }, { "epoch": 0.9435553803705163, "grad_norm": 0.6327707171440125, "learning_rate": 0.0005, "loss": 1.9857757091522217, "step": 11969 }, { "epoch": 0.9436342136381553, "grad_norm": 0.7207748889923096, "learning_rate": 0.0005, "loss": 1.8389129638671875, "step": 11970 }, { "epoch": 0.9437130469057943, "grad_norm": 0.6934857964515686, "learning_rate": 0.0005, "loss": 1.7633893489837646, "step": 11971 }, { "epoch": 0.9437918801734332, "grad_norm": 0.5802167057991028, "learning_rate": 0.0005, "loss": 1.8174514770507812, "step": 11972 }, { "epoch": 0.9438707134410721, "grad_norm": 0.6355030536651611, "learning_rate": 0.0005, "loss": 1.8356834650039673, "step": 11973 }, { "epoch": 0.9439495467087111, "grad_norm": 0.7183297872543335, "learning_rate": 0.0005, "loss": 2.047400712966919, "step": 11974 }, { "epoch": 0.94402837997635, "grad_norm": 0.768134355545044, "learning_rate": 0.0005, "loss": 1.9070544242858887, "step": 11975 }, { "epoch": 0.9441072132439889, "grad_norm": 0.6409363150596619, "learning_rate": 0.0005, "loss": 1.772437572479248, "step": 11976 }, { "epoch": 0.9441860465116279, "grad_norm": 0.6577056050300598, "learning_rate": 0.0005, "loss": 2.1760411262512207, "step": 11977 }, { "epoch": 0.9442648797792669, "grad_norm": 0.6239019632339478, "learning_rate": 0.0005, "loss": 2.036635160446167, "step": 11978 }, { "epoch": 0.9443437130469058, "grad_norm": 0.6543570160865784, "learning_rate": 0.0005, "loss": 1.9266765117645264, "step": 11979 }, { "epoch": 0.9444225463145447, "grad_norm": 0.7401684522628784, "learning_rate": 0.0005, "loss": 1.886704921722412, "step": 11980 }, { "epoch": 0.9445013795821837, "grad_norm": 0.4687421917915344, "learning_rate": 0.0005, "loss": 1.7147092819213867, "step": 11981 }, { "epoch": 0.9445802128498226, "grad_norm": 0.6845884919166565, "learning_rate": 0.0005, "loss": 1.7992254495620728, "step": 11982 }, { "epoch": 0.9446590461174615, "grad_norm": 0.7375138401985168, "learning_rate": 0.0005, "loss": 1.8677210807800293, "step": 11983 }, { "epoch": 0.9447378793851006, "grad_norm": 0.7616721391677856, "learning_rate": 0.0005, "loss": 2.0052905082702637, "step": 11984 }, { "epoch": 0.9448167126527395, "grad_norm": 0.7320252656936646, "learning_rate": 0.0005, "loss": 1.8236595392227173, "step": 11985 }, { "epoch": 0.9448955459203784, "grad_norm": 0.7703680992126465, "learning_rate": 0.0005, "loss": 1.8358793258666992, "step": 11986 }, { "epoch": 0.9449743791880173, "grad_norm": 0.9357755780220032, "learning_rate": 0.0005, "loss": 1.795275330543518, "step": 11987 }, { "epoch": 0.9450532124556563, "grad_norm": 0.8342528343200684, "learning_rate": 0.0005, "loss": 1.9140164852142334, "step": 11988 }, { "epoch": 0.9451320457232952, "grad_norm": 0.6202064752578735, "learning_rate": 0.0005, "loss": 1.9358153343200684, "step": 11989 }, { "epoch": 0.9452108789909341, "grad_norm": 0.6967918872833252, "learning_rate": 0.0005, "loss": 2.0248122215270996, "step": 11990 }, { "epoch": 0.9452897122585732, "grad_norm": 0.6699209213256836, "learning_rate": 0.0005, "loss": 1.8426904678344727, "step": 11991 }, { "epoch": 0.9453685455262121, "grad_norm": 0.8996866345405579, "learning_rate": 0.0005, "loss": 1.9179737567901611, "step": 11992 }, { "epoch": 0.945447378793851, "grad_norm": 0.6350829005241394, "learning_rate": 0.0005, "loss": 1.917327880859375, "step": 11993 }, { "epoch": 0.9455262120614899, "grad_norm": 0.5446997880935669, "learning_rate": 0.0005, "loss": 1.654941439628601, "step": 11994 }, { "epoch": 0.9456050453291289, "grad_norm": 0.5424375534057617, "learning_rate": 0.0005, "loss": 1.8780426979064941, "step": 11995 }, { "epoch": 0.9456838785967678, "grad_norm": 0.6031180024147034, "learning_rate": 0.0005, "loss": 1.8657221794128418, "step": 11996 }, { "epoch": 0.9457627118644067, "grad_norm": 0.5316861271858215, "learning_rate": 0.0005, "loss": 1.57322096824646, "step": 11997 }, { "epoch": 0.9458415451320458, "grad_norm": 0.6849039196968079, "learning_rate": 0.0005, "loss": 2.451957941055298, "step": 11998 }, { "epoch": 0.9459203783996847, "grad_norm": 0.6586533784866333, "learning_rate": 0.0005, "loss": 1.956477403640747, "step": 11999 }, { "epoch": 0.9459992116673236, "grad_norm": 0.6963156461715698, "learning_rate": 0.0005, "loss": 1.9873608350753784, "step": 12000 }, { "epoch": 0.9460780449349625, "grad_norm": 0.5530774593353271, "learning_rate": 0.0005, "loss": 1.7337019443511963, "step": 12001 }, { "epoch": 0.9461568782026015, "grad_norm": 0.6521455645561218, "learning_rate": 0.0005, "loss": 1.901859164237976, "step": 12002 }, { "epoch": 0.9462357114702404, "grad_norm": 0.5206519365310669, "learning_rate": 0.0005, "loss": 1.5279959440231323, "step": 12003 }, { "epoch": 0.9463145447378793, "grad_norm": 0.6602674722671509, "learning_rate": 0.0005, "loss": 2.0795669555664062, "step": 12004 }, { "epoch": 0.9463933780055184, "grad_norm": 0.5269892811775208, "learning_rate": 0.0005, "loss": 1.8587942123413086, "step": 12005 }, { "epoch": 0.9464722112731573, "grad_norm": 2.030626058578491, "learning_rate": 0.0005, "loss": 2.029085636138916, "step": 12006 }, { "epoch": 0.9465510445407962, "grad_norm": 0.574324905872345, "learning_rate": 0.0005, "loss": 1.6362721920013428, "step": 12007 }, { "epoch": 0.9466298778084351, "grad_norm": 0.5613451600074768, "learning_rate": 0.0005, "loss": 1.8000417947769165, "step": 12008 }, { "epoch": 0.9467087110760741, "grad_norm": 0.6576301455497742, "learning_rate": 0.0005, "loss": 2.0051426887512207, "step": 12009 }, { "epoch": 0.946787544343713, "grad_norm": 0.6232117414474487, "learning_rate": 0.0005, "loss": 1.5719668865203857, "step": 12010 }, { "epoch": 0.946866377611352, "grad_norm": 0.5160341858863831, "learning_rate": 0.0005, "loss": 1.6142699718475342, "step": 12011 }, { "epoch": 0.946945210878991, "grad_norm": 0.6247581839561462, "learning_rate": 0.0005, "loss": 1.8003625869750977, "step": 12012 }, { "epoch": 0.9470240441466299, "grad_norm": 0.8782171010971069, "learning_rate": 0.0005, "loss": 1.7832788228988647, "step": 12013 }, { "epoch": 0.9471028774142688, "grad_norm": 0.7944473624229431, "learning_rate": 0.0005, "loss": 1.97401762008667, "step": 12014 }, { "epoch": 0.9471817106819078, "grad_norm": 0.6362853050231934, "learning_rate": 0.0005, "loss": 1.6929823160171509, "step": 12015 }, { "epoch": 0.9472605439495467, "grad_norm": 0.7279049754142761, "learning_rate": 0.0005, "loss": 1.686263918876648, "step": 12016 }, { "epoch": 0.9473393772171856, "grad_norm": 0.5614390969276428, "learning_rate": 0.0005, "loss": 1.9287397861480713, "step": 12017 }, { "epoch": 0.9474182104848246, "grad_norm": 0.598564624786377, "learning_rate": 0.0005, "loss": 1.6540379524230957, "step": 12018 }, { "epoch": 0.9474970437524636, "grad_norm": 0.5948309302330017, "learning_rate": 0.0005, "loss": 1.5192737579345703, "step": 12019 }, { "epoch": 0.9475758770201025, "grad_norm": 0.7713953852653503, "learning_rate": 0.0005, "loss": 1.8410255908966064, "step": 12020 }, { "epoch": 0.9476547102877414, "grad_norm": 0.6577714681625366, "learning_rate": 0.0005, "loss": 1.7083370685577393, "step": 12021 }, { "epoch": 0.9477335435553804, "grad_norm": 0.6686899662017822, "learning_rate": 0.0005, "loss": 1.772970199584961, "step": 12022 }, { "epoch": 0.9478123768230193, "grad_norm": 0.5756083726882935, "learning_rate": 0.0005, "loss": 1.753293514251709, "step": 12023 }, { "epoch": 0.9478912100906582, "grad_norm": 0.6180533170700073, "learning_rate": 0.0005, "loss": 1.9821040630340576, "step": 12024 }, { "epoch": 0.9479700433582972, "grad_norm": 1.413960337638855, "learning_rate": 0.0005, "loss": 1.8564833402633667, "step": 12025 }, { "epoch": 0.9480488766259362, "grad_norm": 0.5588462352752686, "learning_rate": 0.0005, "loss": 1.7066690921783447, "step": 12026 }, { "epoch": 0.9481277098935751, "grad_norm": 0.5779609680175781, "learning_rate": 0.0005, "loss": 1.918161392211914, "step": 12027 }, { "epoch": 0.948206543161214, "grad_norm": 0.7179232835769653, "learning_rate": 0.0005, "loss": 2.127772331237793, "step": 12028 }, { "epoch": 0.948285376428853, "grad_norm": 0.677756667137146, "learning_rate": 0.0005, "loss": 1.9648571014404297, "step": 12029 }, { "epoch": 0.9483642096964919, "grad_norm": 0.6482381224632263, "learning_rate": 0.0005, "loss": 1.7231274843215942, "step": 12030 }, { "epoch": 0.9484430429641308, "grad_norm": 0.8585371375083923, "learning_rate": 0.0005, "loss": 1.9943333864212036, "step": 12031 }, { "epoch": 0.9485218762317698, "grad_norm": 0.6247181296348572, "learning_rate": 0.0005, "loss": 1.988189697265625, "step": 12032 }, { "epoch": 0.9486007094994088, "grad_norm": 0.7149267196655273, "learning_rate": 0.0005, "loss": 2.0060951709747314, "step": 12033 }, { "epoch": 0.9486795427670477, "grad_norm": 0.5091618895530701, "learning_rate": 0.0005, "loss": 1.7087702751159668, "step": 12034 }, { "epoch": 0.9487583760346866, "grad_norm": 0.9006410837173462, "learning_rate": 0.0005, "loss": 1.6885790824890137, "step": 12035 }, { "epoch": 0.9488372093023256, "grad_norm": 0.6637842655181885, "learning_rate": 0.0005, "loss": 1.939150094985962, "step": 12036 }, { "epoch": 0.9489160425699645, "grad_norm": 0.6798787713050842, "learning_rate": 0.0005, "loss": 2.1179025173187256, "step": 12037 }, { "epoch": 0.9489948758376034, "grad_norm": 0.5638744831085205, "learning_rate": 0.0005, "loss": 1.6672022342681885, "step": 12038 }, { "epoch": 0.9490737091052424, "grad_norm": 0.5207383036613464, "learning_rate": 0.0005, "loss": 1.534853219985962, "step": 12039 }, { "epoch": 0.9491525423728814, "grad_norm": 0.6143492460250854, "learning_rate": 0.0005, "loss": 2.0227270126342773, "step": 12040 }, { "epoch": 0.9492313756405203, "grad_norm": 0.6201773881912231, "learning_rate": 0.0005, "loss": 2.0514326095581055, "step": 12041 }, { "epoch": 0.9493102089081592, "grad_norm": 0.5806398391723633, "learning_rate": 0.0005, "loss": 1.6471871137619019, "step": 12042 }, { "epoch": 0.9493890421757982, "grad_norm": 0.7100990414619446, "learning_rate": 0.0005, "loss": 1.7840561866760254, "step": 12043 }, { "epoch": 0.9494678754434371, "grad_norm": 0.6736432313919067, "learning_rate": 0.0005, "loss": 1.5730760097503662, "step": 12044 }, { "epoch": 0.949546708711076, "grad_norm": 0.6592764258384705, "learning_rate": 0.0005, "loss": 1.667489767074585, "step": 12045 }, { "epoch": 0.9496255419787151, "grad_norm": 0.7540794610977173, "learning_rate": 0.0005, "loss": 1.8331745862960815, "step": 12046 }, { "epoch": 0.949704375246354, "grad_norm": 0.5779742002487183, "learning_rate": 0.0005, "loss": 1.696373462677002, "step": 12047 }, { "epoch": 0.9497832085139929, "grad_norm": 0.7049471139907837, "learning_rate": 0.0005, "loss": 1.9674925804138184, "step": 12048 }, { "epoch": 0.9498620417816318, "grad_norm": 0.5323022603988647, "learning_rate": 0.0005, "loss": 1.648020625114441, "step": 12049 }, { "epoch": 0.9499408750492708, "grad_norm": 0.8301557302474976, "learning_rate": 0.0005, "loss": 2.043461561203003, "step": 12050 }, { "epoch": 0.9500197083169097, "grad_norm": 0.7685629725456238, "learning_rate": 0.0005, "loss": 2.0056958198547363, "step": 12051 }, { "epoch": 0.9500985415845487, "grad_norm": 0.7811840772628784, "learning_rate": 0.0005, "loss": 2.149822235107422, "step": 12052 }, { "epoch": 0.9501773748521877, "grad_norm": 0.7234789133071899, "learning_rate": 0.0005, "loss": 1.9308836460113525, "step": 12053 }, { "epoch": 0.9502562081198266, "grad_norm": 0.6521314382553101, "learning_rate": 0.0005, "loss": 1.4979256391525269, "step": 12054 }, { "epoch": 0.9503350413874655, "grad_norm": 0.6394228935241699, "learning_rate": 0.0005, "loss": 1.5524768829345703, "step": 12055 }, { "epoch": 0.9504138746551044, "grad_norm": 0.6710900068283081, "learning_rate": 0.0005, "loss": 2.0165271759033203, "step": 12056 }, { "epoch": 0.9504927079227434, "grad_norm": 0.645646333694458, "learning_rate": 0.0005, "loss": 1.767356514930725, "step": 12057 }, { "epoch": 0.9505715411903823, "grad_norm": 0.6944862008094788, "learning_rate": 0.0005, "loss": 1.921367883682251, "step": 12058 }, { "epoch": 0.9506503744580213, "grad_norm": 0.6208862066268921, "learning_rate": 0.0005, "loss": 1.7727890014648438, "step": 12059 }, { "epoch": 0.9507292077256603, "grad_norm": 0.628838300704956, "learning_rate": 0.0005, "loss": 1.915444254875183, "step": 12060 }, { "epoch": 0.9508080409932992, "grad_norm": 0.5564094185829163, "learning_rate": 0.0005, "loss": 1.7156081199645996, "step": 12061 }, { "epoch": 0.9508868742609381, "grad_norm": 0.5252388119697571, "learning_rate": 0.0005, "loss": 1.606412410736084, "step": 12062 }, { "epoch": 0.950965707528577, "grad_norm": 0.7567915916442871, "learning_rate": 0.0005, "loss": 1.8105533123016357, "step": 12063 }, { "epoch": 0.951044540796216, "grad_norm": 0.6731853485107422, "learning_rate": 0.0005, "loss": 1.750468373298645, "step": 12064 }, { "epoch": 0.9511233740638549, "grad_norm": 0.6615139842033386, "learning_rate": 0.0005, "loss": 2.0023303031921387, "step": 12065 }, { "epoch": 0.9512022073314939, "grad_norm": 0.6246493458747864, "learning_rate": 0.0005, "loss": 1.5255658626556396, "step": 12066 }, { "epoch": 0.9512810405991329, "grad_norm": 0.8952435851097107, "learning_rate": 0.0005, "loss": 1.9208765029907227, "step": 12067 }, { "epoch": 0.9513598738667718, "grad_norm": 0.6371081471443176, "learning_rate": 0.0005, "loss": 1.8292971849441528, "step": 12068 }, { "epoch": 0.9514387071344107, "grad_norm": 0.6590877175331116, "learning_rate": 0.0005, "loss": 1.9893531799316406, "step": 12069 }, { "epoch": 0.9515175404020496, "grad_norm": 0.6490070223808289, "learning_rate": 0.0005, "loss": 1.768691897392273, "step": 12070 }, { "epoch": 0.9515963736696886, "grad_norm": 1.0897847414016724, "learning_rate": 0.0005, "loss": 2.157048225402832, "step": 12071 }, { "epoch": 0.9516752069373275, "grad_norm": 0.6263425946235657, "learning_rate": 0.0005, "loss": 1.8673880100250244, "step": 12072 }, { "epoch": 0.9517540402049665, "grad_norm": 0.8311329483985901, "learning_rate": 0.0005, "loss": 1.9699232578277588, "step": 12073 }, { "epoch": 0.9518328734726055, "grad_norm": 0.5620693564414978, "learning_rate": 0.0005, "loss": 1.6105308532714844, "step": 12074 }, { "epoch": 0.9519117067402444, "grad_norm": 0.5833069086074829, "learning_rate": 0.0005, "loss": 1.67276930809021, "step": 12075 }, { "epoch": 0.9519905400078833, "grad_norm": 0.6911810636520386, "learning_rate": 0.0005, "loss": 1.6472814083099365, "step": 12076 }, { "epoch": 0.9520693732755223, "grad_norm": 0.6654730439186096, "learning_rate": 0.0005, "loss": 1.7290832996368408, "step": 12077 }, { "epoch": 0.9521482065431612, "grad_norm": 0.6213081479072571, "learning_rate": 0.0005, "loss": 1.6910854578018188, "step": 12078 }, { "epoch": 0.9522270398108001, "grad_norm": 0.6232923865318298, "learning_rate": 0.0005, "loss": 1.7020806074142456, "step": 12079 }, { "epoch": 0.9523058730784391, "grad_norm": 0.4973296821117401, "learning_rate": 0.0005, "loss": 1.674864649772644, "step": 12080 }, { "epoch": 0.9523847063460781, "grad_norm": 0.7712442278862, "learning_rate": 0.0005, "loss": 1.833564281463623, "step": 12081 }, { "epoch": 0.952463539613717, "grad_norm": 0.574712872505188, "learning_rate": 0.0005, "loss": 1.615666389465332, "step": 12082 }, { "epoch": 0.9525423728813559, "grad_norm": 0.6971684694290161, "learning_rate": 0.0005, "loss": 1.851590871810913, "step": 12083 }, { "epoch": 0.9526212061489949, "grad_norm": 1.125603199005127, "learning_rate": 0.0005, "loss": 1.9111833572387695, "step": 12084 }, { "epoch": 0.9527000394166338, "grad_norm": 0.9226773381233215, "learning_rate": 0.0005, "loss": 1.8320276737213135, "step": 12085 }, { "epoch": 0.9527788726842727, "grad_norm": 0.6285414695739746, "learning_rate": 0.0005, "loss": 1.8441717624664307, "step": 12086 }, { "epoch": 0.9528577059519117, "grad_norm": 0.6833418011665344, "learning_rate": 0.0005, "loss": 1.848166584968567, "step": 12087 }, { "epoch": 0.9529365392195507, "grad_norm": 0.69693922996521, "learning_rate": 0.0005, "loss": 1.8956787586212158, "step": 12088 }, { "epoch": 0.9530153724871896, "grad_norm": 0.6775310039520264, "learning_rate": 0.0005, "loss": 1.889615535736084, "step": 12089 }, { "epoch": 0.9530942057548285, "grad_norm": 0.7231348156929016, "learning_rate": 0.0005, "loss": 1.738590955734253, "step": 12090 }, { "epoch": 0.9531730390224675, "grad_norm": 0.5382523536682129, "learning_rate": 0.0005, "loss": 1.8064241409301758, "step": 12091 }, { "epoch": 0.9532518722901064, "grad_norm": 0.6069351434707642, "learning_rate": 0.0005, "loss": 2.0056276321411133, "step": 12092 }, { "epoch": 0.9533307055577454, "grad_norm": 0.6000885367393494, "learning_rate": 0.0005, "loss": 1.8836276531219482, "step": 12093 }, { "epoch": 0.9534095388253843, "grad_norm": 0.6042110919952393, "learning_rate": 0.0005, "loss": 1.6313481330871582, "step": 12094 }, { "epoch": 0.9534883720930233, "grad_norm": 0.6543102860450745, "learning_rate": 0.0005, "loss": 1.841051459312439, "step": 12095 }, { "epoch": 0.9535672053606622, "grad_norm": 0.6702331900596619, "learning_rate": 0.0005, "loss": 1.8079078197479248, "step": 12096 }, { "epoch": 0.9536460386283011, "grad_norm": 0.6281411647796631, "learning_rate": 0.0005, "loss": 1.7852859497070312, "step": 12097 }, { "epoch": 0.9537248718959401, "grad_norm": 0.9597216248512268, "learning_rate": 0.0005, "loss": 1.9771277904510498, "step": 12098 }, { "epoch": 0.953803705163579, "grad_norm": 0.9658043384552002, "learning_rate": 0.0005, "loss": 1.936078667640686, "step": 12099 }, { "epoch": 0.953882538431218, "grad_norm": 0.6050565838813782, "learning_rate": 0.0005, "loss": 1.720434546470642, "step": 12100 }, { "epoch": 0.9539613716988569, "grad_norm": 0.7008838653564453, "learning_rate": 0.0005, "loss": 2.012728691101074, "step": 12101 }, { "epoch": 0.9540402049664959, "grad_norm": 0.6140643358230591, "learning_rate": 0.0005, "loss": 1.8623919486999512, "step": 12102 }, { "epoch": 0.9541190382341348, "grad_norm": 0.7328381538391113, "learning_rate": 0.0005, "loss": 1.8609631061553955, "step": 12103 }, { "epoch": 0.9541978715017737, "grad_norm": 0.6084979176521301, "learning_rate": 0.0005, "loss": 1.8344147205352783, "step": 12104 }, { "epoch": 0.9542767047694127, "grad_norm": 0.5946957468986511, "learning_rate": 0.0005, "loss": 1.9472583532333374, "step": 12105 }, { "epoch": 0.9543555380370516, "grad_norm": 0.6125467419624329, "learning_rate": 0.0005, "loss": 1.8933229446411133, "step": 12106 }, { "epoch": 0.9544343713046906, "grad_norm": 0.8442271947860718, "learning_rate": 0.0005, "loss": 1.9227253198623657, "step": 12107 }, { "epoch": 0.9545132045723296, "grad_norm": 1.5415149927139282, "learning_rate": 0.0005, "loss": 1.9405021667480469, "step": 12108 }, { "epoch": 0.9545920378399685, "grad_norm": 0.622738242149353, "learning_rate": 0.0005, "loss": 1.5577400922775269, "step": 12109 }, { "epoch": 0.9546708711076074, "grad_norm": 0.7508772015571594, "learning_rate": 0.0005, "loss": 1.8234758377075195, "step": 12110 }, { "epoch": 0.9547497043752463, "grad_norm": 0.6520843505859375, "learning_rate": 0.0005, "loss": 1.6789193153381348, "step": 12111 }, { "epoch": 0.9548285376428853, "grad_norm": 0.7065758109092712, "learning_rate": 0.0005, "loss": 2.0150485038757324, "step": 12112 }, { "epoch": 0.9549073709105242, "grad_norm": 0.711794912815094, "learning_rate": 0.0005, "loss": 1.6974985599517822, "step": 12113 }, { "epoch": 0.9549862041781632, "grad_norm": 0.6634053587913513, "learning_rate": 0.0005, "loss": 1.7476763725280762, "step": 12114 }, { "epoch": 0.9550650374458022, "grad_norm": 0.818228542804718, "learning_rate": 0.0005, "loss": 1.937305212020874, "step": 12115 }, { "epoch": 0.9551438707134411, "grad_norm": 0.7667527794837952, "learning_rate": 0.0005, "loss": 2.053027629852295, "step": 12116 }, { "epoch": 0.95522270398108, "grad_norm": 0.5890988111495972, "learning_rate": 0.0005, "loss": 1.8070716857910156, "step": 12117 }, { "epoch": 0.9553015372487189, "grad_norm": 0.8994041085243225, "learning_rate": 0.0005, "loss": 2.07291841506958, "step": 12118 }, { "epoch": 0.9553803705163579, "grad_norm": 0.5867421627044678, "learning_rate": 0.0005, "loss": 1.878104567527771, "step": 12119 }, { "epoch": 0.9554592037839968, "grad_norm": 0.6438695788383484, "learning_rate": 0.0005, "loss": 1.5812902450561523, "step": 12120 }, { "epoch": 0.9555380370516358, "grad_norm": 0.601626992225647, "learning_rate": 0.0005, "loss": 1.8226196765899658, "step": 12121 }, { "epoch": 0.9556168703192748, "grad_norm": 0.7715703845024109, "learning_rate": 0.0005, "loss": 1.7732553482055664, "step": 12122 }, { "epoch": 0.9556957035869137, "grad_norm": 0.7597951889038086, "learning_rate": 0.0005, "loss": 2.099975347518921, "step": 12123 }, { "epoch": 0.9557745368545526, "grad_norm": 0.5004699230194092, "learning_rate": 0.0005, "loss": 1.7649308443069458, "step": 12124 }, { "epoch": 0.9558533701221915, "grad_norm": 0.4728747308254242, "learning_rate": 0.0005, "loss": 1.6378307342529297, "step": 12125 }, { "epoch": 0.9559322033898305, "grad_norm": 0.667583167552948, "learning_rate": 0.0005, "loss": 2.143355131149292, "step": 12126 }, { "epoch": 0.9560110366574694, "grad_norm": 0.5770732760429382, "learning_rate": 0.0005, "loss": 1.7745769023895264, "step": 12127 }, { "epoch": 0.9560898699251084, "grad_norm": 0.7107782959938049, "learning_rate": 0.0005, "loss": 1.765578031539917, "step": 12128 }, { "epoch": 0.9561687031927474, "grad_norm": 0.7137196063995361, "learning_rate": 0.0005, "loss": 2.2942724227905273, "step": 12129 }, { "epoch": 0.9562475364603863, "grad_norm": 0.5232632756233215, "learning_rate": 0.0005, "loss": 1.956620454788208, "step": 12130 }, { "epoch": 0.9563263697280252, "grad_norm": 0.5779112577438354, "learning_rate": 0.0005, "loss": 1.725768804550171, "step": 12131 }, { "epoch": 0.9564052029956641, "grad_norm": 0.8243470191955566, "learning_rate": 0.0005, "loss": 1.7624452114105225, "step": 12132 }, { "epoch": 0.9564840362633031, "grad_norm": 0.640383243560791, "learning_rate": 0.0005, "loss": 1.9384609460830688, "step": 12133 }, { "epoch": 0.956562869530942, "grad_norm": 0.6299882531166077, "learning_rate": 0.0005, "loss": 2.045471668243408, "step": 12134 }, { "epoch": 0.956641702798581, "grad_norm": 0.912535548210144, "learning_rate": 0.0005, "loss": 1.847074270248413, "step": 12135 }, { "epoch": 0.95672053606622, "grad_norm": 0.622043788433075, "learning_rate": 0.0005, "loss": 1.6360316276550293, "step": 12136 }, { "epoch": 0.9567993693338589, "grad_norm": 0.6050410866737366, "learning_rate": 0.0005, "loss": 1.5901844501495361, "step": 12137 }, { "epoch": 0.9568782026014978, "grad_norm": 0.6607770919799805, "learning_rate": 0.0005, "loss": 1.518254280090332, "step": 12138 }, { "epoch": 0.9569570358691368, "grad_norm": 0.7253471612930298, "learning_rate": 0.0005, "loss": 1.852184534072876, "step": 12139 }, { "epoch": 0.9570358691367757, "grad_norm": 0.7085413336753845, "learning_rate": 0.0005, "loss": 1.9128139019012451, "step": 12140 }, { "epoch": 0.9571147024044147, "grad_norm": 0.6282076835632324, "learning_rate": 0.0005, "loss": 1.784616231918335, "step": 12141 }, { "epoch": 0.9571935356720536, "grad_norm": 0.5419542193412781, "learning_rate": 0.0005, "loss": 1.839633584022522, "step": 12142 }, { "epoch": 0.9572723689396926, "grad_norm": 0.6033021211624146, "learning_rate": 0.0005, "loss": 1.7976171970367432, "step": 12143 }, { "epoch": 0.9573512022073315, "grad_norm": 0.6689469814300537, "learning_rate": 0.0005, "loss": 1.7978112697601318, "step": 12144 }, { "epoch": 0.9574300354749704, "grad_norm": 0.546928346157074, "learning_rate": 0.0005, "loss": 1.7799136638641357, "step": 12145 }, { "epoch": 0.9575088687426094, "grad_norm": 0.5430269241333008, "learning_rate": 0.0005, "loss": 1.9260203838348389, "step": 12146 }, { "epoch": 0.9575877020102483, "grad_norm": 0.5577376484870911, "learning_rate": 0.0005, "loss": 1.4062023162841797, "step": 12147 }, { "epoch": 0.9576665352778873, "grad_norm": 0.5292969942092896, "learning_rate": 0.0005, "loss": 1.923171877861023, "step": 12148 }, { "epoch": 0.9577453685455262, "grad_norm": 0.5890215039253235, "learning_rate": 0.0005, "loss": 1.8115427494049072, "step": 12149 }, { "epoch": 0.9578242018131652, "grad_norm": 0.6772845983505249, "learning_rate": 0.0005, "loss": 1.7015384435653687, "step": 12150 }, { "epoch": 0.9579030350808041, "grad_norm": 0.7864693999290466, "learning_rate": 0.0005, "loss": 2.070751905441284, "step": 12151 }, { "epoch": 0.957981868348443, "grad_norm": 0.5912469625473022, "learning_rate": 0.0005, "loss": 1.7946454286575317, "step": 12152 }, { "epoch": 0.958060701616082, "grad_norm": 0.7775011658668518, "learning_rate": 0.0005, "loss": 1.6591155529022217, "step": 12153 }, { "epoch": 0.958139534883721, "grad_norm": 0.5785068273544312, "learning_rate": 0.0005, "loss": 1.7702158689498901, "step": 12154 }, { "epoch": 0.9582183681513599, "grad_norm": 0.6703113913536072, "learning_rate": 0.0005, "loss": 1.872291922569275, "step": 12155 }, { "epoch": 0.9582972014189988, "grad_norm": 0.6136072874069214, "learning_rate": 0.0005, "loss": 1.9414278268814087, "step": 12156 }, { "epoch": 0.9583760346866378, "grad_norm": 0.5187790393829346, "learning_rate": 0.0005, "loss": 1.6766287088394165, "step": 12157 }, { "epoch": 0.9584548679542767, "grad_norm": 0.58220374584198, "learning_rate": 0.0005, "loss": 2.1174545288085938, "step": 12158 }, { "epoch": 0.9585337012219156, "grad_norm": 1.81294584274292, "learning_rate": 0.0005, "loss": 1.8177675008773804, "step": 12159 }, { "epoch": 0.9586125344895546, "grad_norm": 0.6773900985717773, "learning_rate": 0.0005, "loss": 1.6790647506713867, "step": 12160 }, { "epoch": 0.9586913677571935, "grad_norm": 0.6063616275787354, "learning_rate": 0.0005, "loss": 1.8133187294006348, "step": 12161 }, { "epoch": 0.9587702010248325, "grad_norm": 0.7772536277770996, "learning_rate": 0.0005, "loss": 1.8804302215576172, "step": 12162 }, { "epoch": 0.9588490342924714, "grad_norm": 0.6202031373977661, "learning_rate": 0.0005, "loss": 1.737037181854248, "step": 12163 }, { "epoch": 0.9589278675601104, "grad_norm": 0.4941852390766144, "learning_rate": 0.0005, "loss": 1.5719404220581055, "step": 12164 }, { "epoch": 0.9590067008277493, "grad_norm": 0.5901998281478882, "learning_rate": 0.0005, "loss": 1.6604987382888794, "step": 12165 }, { "epoch": 0.9590855340953882, "grad_norm": 0.8301244974136353, "learning_rate": 0.0005, "loss": 1.8807591199874878, "step": 12166 }, { "epoch": 0.9591643673630272, "grad_norm": 0.6479819416999817, "learning_rate": 0.0005, "loss": 2.0844902992248535, "step": 12167 }, { "epoch": 0.9592432006306661, "grad_norm": 0.6051045656204224, "learning_rate": 0.0005, "loss": 2.091768741607666, "step": 12168 }, { "epoch": 0.9593220338983051, "grad_norm": 0.7716818451881409, "learning_rate": 0.0005, "loss": 1.6908369064331055, "step": 12169 }, { "epoch": 0.9594008671659441, "grad_norm": 0.6304624080657959, "learning_rate": 0.0005, "loss": 1.884592056274414, "step": 12170 }, { "epoch": 0.959479700433583, "grad_norm": 0.5178719162940979, "learning_rate": 0.0005, "loss": 1.8118271827697754, "step": 12171 }, { "epoch": 0.9595585337012219, "grad_norm": 0.6372894048690796, "learning_rate": 0.0005, "loss": 1.8747731447219849, "step": 12172 }, { "epoch": 0.9596373669688608, "grad_norm": 0.4961526095867157, "learning_rate": 0.0005, "loss": 1.597059726715088, "step": 12173 }, { "epoch": 0.9597162002364998, "grad_norm": 0.6013271808624268, "learning_rate": 0.0005, "loss": 1.829584002494812, "step": 12174 }, { "epoch": 0.9597950335041388, "grad_norm": 0.5686814188957214, "learning_rate": 0.0005, "loss": 1.7185486555099487, "step": 12175 }, { "epoch": 0.9598738667717777, "grad_norm": 0.5142695307731628, "learning_rate": 0.0005, "loss": 1.8277134895324707, "step": 12176 }, { "epoch": 0.9599527000394167, "grad_norm": 0.6577163338661194, "learning_rate": 0.0005, "loss": 1.8545217514038086, "step": 12177 }, { "epoch": 0.9600315333070556, "grad_norm": 0.8367303609848022, "learning_rate": 0.0005, "loss": 1.821903944015503, "step": 12178 }, { "epoch": 0.9601103665746945, "grad_norm": 0.6601249575614929, "learning_rate": 0.0005, "loss": 2.1129393577575684, "step": 12179 }, { "epoch": 0.9601891998423334, "grad_norm": 0.676586389541626, "learning_rate": 0.0005, "loss": 1.8092447519302368, "step": 12180 }, { "epoch": 0.9602680331099724, "grad_norm": 0.7235445976257324, "learning_rate": 0.0005, "loss": 1.9723403453826904, "step": 12181 }, { "epoch": 0.9603468663776114, "grad_norm": 0.5891346335411072, "learning_rate": 0.0005, "loss": 1.7473063468933105, "step": 12182 }, { "epoch": 0.9604256996452503, "grad_norm": 0.5966372489929199, "learning_rate": 0.0005, "loss": 1.7843656539916992, "step": 12183 }, { "epoch": 0.9605045329128893, "grad_norm": 0.5296210050582886, "learning_rate": 0.0005, "loss": 1.6049246788024902, "step": 12184 }, { "epoch": 0.9605833661805282, "grad_norm": 0.7388749718666077, "learning_rate": 0.0005, "loss": 1.6931967735290527, "step": 12185 }, { "epoch": 0.9606621994481671, "grad_norm": 0.6254679560661316, "learning_rate": 0.0005, "loss": 1.8901655673980713, "step": 12186 }, { "epoch": 0.960741032715806, "grad_norm": 0.7564817070960999, "learning_rate": 0.0005, "loss": 1.7422131299972534, "step": 12187 }, { "epoch": 0.960819865983445, "grad_norm": 0.6750915050506592, "learning_rate": 0.0005, "loss": 1.9819108247756958, "step": 12188 }, { "epoch": 0.960898699251084, "grad_norm": 0.5700365900993347, "learning_rate": 0.0005, "loss": 2.079679489135742, "step": 12189 }, { "epoch": 0.9609775325187229, "grad_norm": 0.5887787938117981, "learning_rate": 0.0005, "loss": 1.9172773361206055, "step": 12190 }, { "epoch": 0.9610563657863619, "grad_norm": 0.5660619139671326, "learning_rate": 0.0005, "loss": 1.8014277219772339, "step": 12191 }, { "epoch": 0.9611351990540008, "grad_norm": 0.5907054543495178, "learning_rate": 0.0005, "loss": 1.7757935523986816, "step": 12192 }, { "epoch": 0.9612140323216397, "grad_norm": 0.6482387781143188, "learning_rate": 0.0005, "loss": 1.834904432296753, "step": 12193 }, { "epoch": 0.9612928655892786, "grad_norm": 0.6767908930778503, "learning_rate": 0.0005, "loss": 1.7141473293304443, "step": 12194 }, { "epoch": 0.9613716988569176, "grad_norm": 0.9425580501556396, "learning_rate": 0.0005, "loss": 1.7035629749298096, "step": 12195 }, { "epoch": 0.9614505321245566, "grad_norm": 0.5249836444854736, "learning_rate": 0.0005, "loss": 1.7699240446090698, "step": 12196 }, { "epoch": 0.9615293653921955, "grad_norm": 0.7425432205200195, "learning_rate": 0.0005, "loss": 1.547767162322998, "step": 12197 }, { "epoch": 0.9616081986598345, "grad_norm": 0.8657558560371399, "learning_rate": 0.0005, "loss": 1.8799480199813843, "step": 12198 }, { "epoch": 0.9616870319274734, "grad_norm": 0.731078565120697, "learning_rate": 0.0005, "loss": 1.9056153297424316, "step": 12199 }, { "epoch": 0.9617658651951123, "grad_norm": 1.1730636358261108, "learning_rate": 0.0005, "loss": 1.7569775581359863, "step": 12200 }, { "epoch": 0.9618446984627513, "grad_norm": 1.0666096210479736, "learning_rate": 0.0005, "loss": 1.8936012983322144, "step": 12201 }, { "epoch": 0.9619235317303902, "grad_norm": 0.5894647240638733, "learning_rate": 0.0005, "loss": 1.5954651832580566, "step": 12202 }, { "epoch": 0.9620023649980292, "grad_norm": 0.5347822904586792, "learning_rate": 0.0005, "loss": 1.7238625288009644, "step": 12203 }, { "epoch": 0.9620811982656681, "grad_norm": 0.7655786871910095, "learning_rate": 0.0005, "loss": 1.8041281700134277, "step": 12204 }, { "epoch": 0.9621600315333071, "grad_norm": 0.6024687886238098, "learning_rate": 0.0005, "loss": 1.7843070030212402, "step": 12205 }, { "epoch": 0.962238864800946, "grad_norm": 0.5596131682395935, "learning_rate": 0.0005, "loss": 1.8706669807434082, "step": 12206 }, { "epoch": 0.9623176980685849, "grad_norm": 0.6352346539497375, "learning_rate": 0.0005, "loss": 2.1040406227111816, "step": 12207 }, { "epoch": 0.9623965313362239, "grad_norm": 0.628007709980011, "learning_rate": 0.0005, "loss": 1.7678779363632202, "step": 12208 }, { "epoch": 0.9624753646038628, "grad_norm": 0.5455201864242554, "learning_rate": 0.0005, "loss": 1.5648016929626465, "step": 12209 }, { "epoch": 0.9625541978715018, "grad_norm": 0.6802771091461182, "learning_rate": 0.0005, "loss": 1.748313069343567, "step": 12210 }, { "epoch": 0.9626330311391407, "grad_norm": 0.640718936920166, "learning_rate": 0.0005, "loss": 1.7781779766082764, "step": 12211 }, { "epoch": 0.9627118644067797, "grad_norm": 0.7027872204780579, "learning_rate": 0.0005, "loss": 1.6930831670761108, "step": 12212 }, { "epoch": 0.9627906976744186, "grad_norm": 0.5926321744918823, "learning_rate": 0.0005, "loss": 1.8328471183776855, "step": 12213 }, { "epoch": 0.9628695309420575, "grad_norm": 0.5412847995758057, "learning_rate": 0.0005, "loss": 1.7603646516799927, "step": 12214 }, { "epoch": 0.9629483642096965, "grad_norm": 0.7212463021278381, "learning_rate": 0.0005, "loss": 1.314534068107605, "step": 12215 }, { "epoch": 0.9630271974773355, "grad_norm": 0.5873795747756958, "learning_rate": 0.0005, "loss": 1.8041106462478638, "step": 12216 }, { "epoch": 0.9631060307449744, "grad_norm": 0.7587001919746399, "learning_rate": 0.0005, "loss": 1.8324717283248901, "step": 12217 }, { "epoch": 0.9631848640126133, "grad_norm": 0.6625997424125671, "learning_rate": 0.0005, "loss": 1.9893996715545654, "step": 12218 }, { "epoch": 0.9632636972802523, "grad_norm": 0.614715039730072, "learning_rate": 0.0005, "loss": 1.9332034587860107, "step": 12219 }, { "epoch": 0.9633425305478912, "grad_norm": 0.7308074235916138, "learning_rate": 0.0005, "loss": 1.8153001070022583, "step": 12220 }, { "epoch": 0.9634213638155301, "grad_norm": 0.6234843730926514, "learning_rate": 0.0005, "loss": 1.652766227722168, "step": 12221 }, { "epoch": 0.9635001970831691, "grad_norm": 0.695702075958252, "learning_rate": 0.0005, "loss": 1.8416519165039062, "step": 12222 }, { "epoch": 0.963579030350808, "grad_norm": 0.6142617464065552, "learning_rate": 0.0005, "loss": 2.0220537185668945, "step": 12223 }, { "epoch": 0.963657863618447, "grad_norm": 0.6580385565757751, "learning_rate": 0.0005, "loss": 1.62356436252594, "step": 12224 }, { "epoch": 0.9637366968860859, "grad_norm": 0.5229710340499878, "learning_rate": 0.0005, "loss": 1.9077951908111572, "step": 12225 }, { "epoch": 0.9638155301537249, "grad_norm": 0.6085696220397949, "learning_rate": 0.0005, "loss": 1.9208937883377075, "step": 12226 }, { "epoch": 0.9638943634213638, "grad_norm": 0.6718611121177673, "learning_rate": 0.0005, "loss": 1.9804201126098633, "step": 12227 }, { "epoch": 0.9639731966890027, "grad_norm": 0.5741734504699707, "learning_rate": 0.0005, "loss": 1.6579921245574951, "step": 12228 }, { "epoch": 0.9640520299566417, "grad_norm": 0.6273475289344788, "learning_rate": 0.0005, "loss": 1.6663904190063477, "step": 12229 }, { "epoch": 0.9641308632242807, "grad_norm": 0.5287464261054993, "learning_rate": 0.0005, "loss": 1.9123228788375854, "step": 12230 }, { "epoch": 0.9642096964919196, "grad_norm": 0.6198311448097229, "learning_rate": 0.0005, "loss": 1.996415138244629, "step": 12231 }, { "epoch": 0.9642885297595586, "grad_norm": 0.7277414798736572, "learning_rate": 0.0005, "loss": 1.6750421524047852, "step": 12232 }, { "epoch": 0.9643673630271975, "grad_norm": 0.6031604409217834, "learning_rate": 0.0005, "loss": 2.03706955909729, "step": 12233 }, { "epoch": 0.9644461962948364, "grad_norm": 0.6335852742195129, "learning_rate": 0.0005, "loss": 1.774829626083374, "step": 12234 }, { "epoch": 0.9645250295624753, "grad_norm": 0.597221314907074, "learning_rate": 0.0005, "loss": 1.5827854871749878, "step": 12235 }, { "epoch": 0.9646038628301143, "grad_norm": 0.6287409067153931, "learning_rate": 0.0005, "loss": 1.9331761598587036, "step": 12236 }, { "epoch": 0.9646826960977533, "grad_norm": 0.6201039552688599, "learning_rate": 0.0005, "loss": 1.7595863342285156, "step": 12237 }, { "epoch": 0.9647615293653922, "grad_norm": 0.6113021373748779, "learning_rate": 0.0005, "loss": 1.941840410232544, "step": 12238 }, { "epoch": 0.9648403626330312, "grad_norm": 0.5920991897583008, "learning_rate": 0.0005, "loss": 1.720324993133545, "step": 12239 }, { "epoch": 0.9649191959006701, "grad_norm": 0.6207271814346313, "learning_rate": 0.0005, "loss": 1.844012975692749, "step": 12240 }, { "epoch": 0.964998029168309, "grad_norm": 0.5620027184486389, "learning_rate": 0.0005, "loss": 2.0048367977142334, "step": 12241 }, { "epoch": 0.9650768624359479, "grad_norm": 0.7090928554534912, "learning_rate": 0.0005, "loss": 1.7728779315948486, "step": 12242 }, { "epoch": 0.965155695703587, "grad_norm": 0.554735004901886, "learning_rate": 0.0005, "loss": 1.7953356504440308, "step": 12243 }, { "epoch": 0.9652345289712259, "grad_norm": 0.46774137020111084, "learning_rate": 0.0005, "loss": 1.3705673217773438, "step": 12244 }, { "epoch": 0.9653133622388648, "grad_norm": 0.5520761609077454, "learning_rate": 0.0005, "loss": 1.6778703927993774, "step": 12245 }, { "epoch": 0.9653921955065038, "grad_norm": 0.5752168893814087, "learning_rate": 0.0005, "loss": 2.0043530464172363, "step": 12246 }, { "epoch": 0.9654710287741427, "grad_norm": 1.6678110361099243, "learning_rate": 0.0005, "loss": 2.037938356399536, "step": 12247 }, { "epoch": 0.9655498620417816, "grad_norm": 0.7814998030662537, "learning_rate": 0.0005, "loss": 2.0421862602233887, "step": 12248 }, { "epoch": 0.9656286953094205, "grad_norm": 0.6315165758132935, "learning_rate": 0.0005, "loss": 1.969071626663208, "step": 12249 }, { "epoch": 0.9657075285770595, "grad_norm": 0.5540602803230286, "learning_rate": 0.0005, "loss": 1.5616816282272339, "step": 12250 }, { "epoch": 0.9657863618446985, "grad_norm": 0.7359117865562439, "learning_rate": 0.0005, "loss": 2.020695686340332, "step": 12251 }, { "epoch": 0.9658651951123374, "grad_norm": 0.6666514873504639, "learning_rate": 0.0005, "loss": 1.9713680744171143, "step": 12252 }, { "epoch": 0.9659440283799764, "grad_norm": 0.7184674739837646, "learning_rate": 0.0005, "loss": 1.8101905584335327, "step": 12253 }, { "epoch": 0.9660228616476153, "grad_norm": 0.6673750877380371, "learning_rate": 0.0005, "loss": 1.9439496994018555, "step": 12254 }, { "epoch": 0.9661016949152542, "grad_norm": 0.6295291781425476, "learning_rate": 0.0005, "loss": 1.7550275325775146, "step": 12255 }, { "epoch": 0.9661805281828931, "grad_norm": 0.6899886727333069, "learning_rate": 0.0005, "loss": 1.9476362466812134, "step": 12256 }, { "epoch": 0.9662593614505322, "grad_norm": 0.839077353477478, "learning_rate": 0.0005, "loss": 1.7983720302581787, "step": 12257 }, { "epoch": 0.9663381947181711, "grad_norm": 0.5976513028144836, "learning_rate": 0.0005, "loss": 1.6568048000335693, "step": 12258 }, { "epoch": 0.96641702798581, "grad_norm": 0.5947893857955933, "learning_rate": 0.0005, "loss": 2.011725425720215, "step": 12259 }, { "epoch": 0.966495861253449, "grad_norm": 0.6505144238471985, "learning_rate": 0.0005, "loss": 1.6801245212554932, "step": 12260 }, { "epoch": 0.9665746945210879, "grad_norm": 0.8383302092552185, "learning_rate": 0.0005, "loss": 1.986088752746582, "step": 12261 }, { "epoch": 0.9666535277887268, "grad_norm": 0.5552211403846741, "learning_rate": 0.0005, "loss": 1.9228075742721558, "step": 12262 }, { "epoch": 0.9667323610563658, "grad_norm": 0.6791725158691406, "learning_rate": 0.0005, "loss": 1.924437403678894, "step": 12263 }, { "epoch": 0.9668111943240048, "grad_norm": 0.759989857673645, "learning_rate": 0.0005, "loss": 1.641162395477295, "step": 12264 }, { "epoch": 0.9668900275916437, "grad_norm": 0.6762863397598267, "learning_rate": 0.0005, "loss": 1.7190933227539062, "step": 12265 }, { "epoch": 0.9669688608592826, "grad_norm": 0.7547929883003235, "learning_rate": 0.0005, "loss": 1.7729442119598389, "step": 12266 }, { "epoch": 0.9670476941269216, "grad_norm": 0.6791649460792542, "learning_rate": 0.0005, "loss": 1.8067435026168823, "step": 12267 }, { "epoch": 0.9671265273945605, "grad_norm": 0.7044896483421326, "learning_rate": 0.0005, "loss": 1.7869656085968018, "step": 12268 }, { "epoch": 0.9672053606621994, "grad_norm": 0.6781332492828369, "learning_rate": 0.0005, "loss": 1.6604042053222656, "step": 12269 }, { "epoch": 0.9672841939298384, "grad_norm": 1.035780906677246, "learning_rate": 0.0005, "loss": 1.7299686670303345, "step": 12270 }, { "epoch": 0.9673630271974774, "grad_norm": 0.6593203544616699, "learning_rate": 0.0005, "loss": 1.6586723327636719, "step": 12271 }, { "epoch": 0.9674418604651163, "grad_norm": 0.5553435683250427, "learning_rate": 0.0005, "loss": 1.5487744808197021, "step": 12272 }, { "epoch": 0.9675206937327552, "grad_norm": 0.5120639801025391, "learning_rate": 0.0005, "loss": 1.7330639362335205, "step": 12273 }, { "epoch": 0.9675995270003942, "grad_norm": 0.7921207547187805, "learning_rate": 0.0005, "loss": 1.8600720167160034, "step": 12274 }, { "epoch": 0.9676783602680331, "grad_norm": 0.6972181797027588, "learning_rate": 0.0005, "loss": 2.0955729484558105, "step": 12275 }, { "epoch": 0.967757193535672, "grad_norm": 0.5678662657737732, "learning_rate": 0.0005, "loss": 1.5689994096755981, "step": 12276 }, { "epoch": 0.967836026803311, "grad_norm": 0.5922690033912659, "learning_rate": 0.0005, "loss": 1.7658253908157349, "step": 12277 }, { "epoch": 0.96791486007095, "grad_norm": 0.7331373691558838, "learning_rate": 0.0005, "loss": 1.4348039627075195, "step": 12278 }, { "epoch": 0.9679936933385889, "grad_norm": 1.0034526586532593, "learning_rate": 0.0005, "loss": 1.9673075675964355, "step": 12279 }, { "epoch": 0.9680725266062278, "grad_norm": 0.6516888737678528, "learning_rate": 0.0005, "loss": 1.7404937744140625, "step": 12280 }, { "epoch": 0.9681513598738668, "grad_norm": 0.6667100787162781, "learning_rate": 0.0005, "loss": 1.7549080848693848, "step": 12281 }, { "epoch": 0.9682301931415057, "grad_norm": 0.6879061460494995, "learning_rate": 0.0005, "loss": 1.5522539615631104, "step": 12282 }, { "epoch": 0.9683090264091446, "grad_norm": 0.5890297293663025, "learning_rate": 0.0005, "loss": 1.7863296270370483, "step": 12283 }, { "epoch": 0.9683878596767836, "grad_norm": 0.5835069417953491, "learning_rate": 0.0005, "loss": 1.723567247390747, "step": 12284 }, { "epoch": 0.9684666929444226, "grad_norm": 0.6414235234260559, "learning_rate": 0.0005, "loss": 1.8463091850280762, "step": 12285 }, { "epoch": 0.9685455262120615, "grad_norm": 0.8525949716567993, "learning_rate": 0.0005, "loss": 2.0484604835510254, "step": 12286 }, { "epoch": 0.9686243594797004, "grad_norm": 0.743553876876831, "learning_rate": 0.0005, "loss": 1.8418244123458862, "step": 12287 }, { "epoch": 0.9687031927473394, "grad_norm": 0.5968447923660278, "learning_rate": 0.0005, "loss": 1.7618544101715088, "step": 12288 }, { "epoch": 0.9687820260149783, "grad_norm": 0.6706159710884094, "learning_rate": 0.0005, "loss": 1.4786463975906372, "step": 12289 }, { "epoch": 0.9688608592826172, "grad_norm": 0.7358986735343933, "learning_rate": 0.0005, "loss": 1.88834547996521, "step": 12290 }, { "epoch": 0.9689396925502562, "grad_norm": 0.5317319631576538, "learning_rate": 0.0005, "loss": 1.9005182981491089, "step": 12291 }, { "epoch": 0.9690185258178952, "grad_norm": 0.6003445386886597, "learning_rate": 0.0005, "loss": 1.7545177936553955, "step": 12292 }, { "epoch": 0.9690973590855341, "grad_norm": 0.7446880340576172, "learning_rate": 0.0005, "loss": 1.950110912322998, "step": 12293 }, { "epoch": 0.9691761923531731, "grad_norm": 1.106857180595398, "learning_rate": 0.0005, "loss": 1.94889235496521, "step": 12294 }, { "epoch": 0.969255025620812, "grad_norm": 0.7423416972160339, "learning_rate": 0.0005, "loss": 1.741053819656372, "step": 12295 }, { "epoch": 0.9693338588884509, "grad_norm": 0.5452178120613098, "learning_rate": 0.0005, "loss": 1.8782594203948975, "step": 12296 }, { "epoch": 0.9694126921560898, "grad_norm": 0.5214489102363586, "learning_rate": 0.0005, "loss": 1.8525912761688232, "step": 12297 }, { "epoch": 0.9694915254237289, "grad_norm": 0.6407910585403442, "learning_rate": 0.0005, "loss": 1.8468575477600098, "step": 12298 }, { "epoch": 0.9695703586913678, "grad_norm": 0.5823267102241516, "learning_rate": 0.0005, "loss": 1.8051555156707764, "step": 12299 }, { "epoch": 0.9696491919590067, "grad_norm": 0.4995437562465668, "learning_rate": 0.0005, "loss": 1.6641771793365479, "step": 12300 }, { "epoch": 0.9697280252266457, "grad_norm": 0.6588416695594788, "learning_rate": 0.0005, "loss": 1.748908519744873, "step": 12301 }, { "epoch": 0.9698068584942846, "grad_norm": 0.5736010074615479, "learning_rate": 0.0005, "loss": 1.6668283939361572, "step": 12302 }, { "epoch": 0.9698856917619235, "grad_norm": 0.6369046568870544, "learning_rate": 0.0005, "loss": 1.7396633625030518, "step": 12303 }, { "epoch": 0.9699645250295624, "grad_norm": 0.5856796503067017, "learning_rate": 0.0005, "loss": 1.7154595851898193, "step": 12304 }, { "epoch": 0.9700433582972015, "grad_norm": 0.5168479084968567, "learning_rate": 0.0005, "loss": 1.7937488555908203, "step": 12305 }, { "epoch": 0.9701221915648404, "grad_norm": 0.7382676005363464, "learning_rate": 0.0005, "loss": 1.7611823081970215, "step": 12306 }, { "epoch": 0.9702010248324793, "grad_norm": 0.7022562623023987, "learning_rate": 0.0005, "loss": 1.928567886352539, "step": 12307 }, { "epoch": 0.9702798581001183, "grad_norm": 0.4907553791999817, "learning_rate": 0.0005, "loss": 1.4816393852233887, "step": 12308 }, { "epoch": 0.9703586913677572, "grad_norm": 0.6532677412033081, "learning_rate": 0.0005, "loss": 1.6259143352508545, "step": 12309 }, { "epoch": 0.9704375246353961, "grad_norm": 0.6188883185386658, "learning_rate": 0.0005, "loss": 1.7282496690750122, "step": 12310 }, { "epoch": 0.970516357903035, "grad_norm": 0.4779253602027893, "learning_rate": 0.0005, "loss": 1.5975239276885986, "step": 12311 }, { "epoch": 0.970595191170674, "grad_norm": 0.7764217257499695, "learning_rate": 0.0005, "loss": 2.0259621143341064, "step": 12312 }, { "epoch": 0.970674024438313, "grad_norm": 0.574699878692627, "learning_rate": 0.0005, "loss": 1.7947065830230713, "step": 12313 }, { "epoch": 0.9707528577059519, "grad_norm": 0.6924034953117371, "learning_rate": 0.0005, "loss": 1.643944263458252, "step": 12314 }, { "epoch": 0.9708316909735909, "grad_norm": 0.7918136715888977, "learning_rate": 0.0005, "loss": 1.7387194633483887, "step": 12315 }, { "epoch": 0.9709105242412298, "grad_norm": 1.056547999382019, "learning_rate": 0.0005, "loss": 1.7796567678451538, "step": 12316 }, { "epoch": 0.9709893575088687, "grad_norm": 0.767993688583374, "learning_rate": 0.0005, "loss": 2.1261839866638184, "step": 12317 }, { "epoch": 0.9710681907765076, "grad_norm": 0.9704846739768982, "learning_rate": 0.0005, "loss": 1.8173012733459473, "step": 12318 }, { "epoch": 0.9711470240441467, "grad_norm": 0.6477787494659424, "learning_rate": 0.0005, "loss": 1.777018427848816, "step": 12319 }, { "epoch": 0.9712258573117856, "grad_norm": 0.7160539031028748, "learning_rate": 0.0005, "loss": 1.9148552417755127, "step": 12320 }, { "epoch": 0.9713046905794245, "grad_norm": 0.5712782144546509, "learning_rate": 0.0005, "loss": 1.8209620714187622, "step": 12321 }, { "epoch": 0.9713835238470635, "grad_norm": 0.7001696825027466, "learning_rate": 0.0005, "loss": 1.8679149150848389, "step": 12322 }, { "epoch": 0.9714623571147024, "grad_norm": 0.6128641366958618, "learning_rate": 0.0005, "loss": 1.653687834739685, "step": 12323 }, { "epoch": 0.9715411903823413, "grad_norm": 0.49790695309638977, "learning_rate": 0.0005, "loss": 1.647004246711731, "step": 12324 }, { "epoch": 0.9716200236499803, "grad_norm": 0.8714274168014526, "learning_rate": 0.0005, "loss": 1.8277039527893066, "step": 12325 }, { "epoch": 0.9716988569176193, "grad_norm": 0.9186640977859497, "learning_rate": 0.0005, "loss": 1.835472583770752, "step": 12326 }, { "epoch": 0.9717776901852582, "grad_norm": 0.527840793132782, "learning_rate": 0.0005, "loss": 1.6173434257507324, "step": 12327 }, { "epoch": 0.9718565234528971, "grad_norm": 0.6221011877059937, "learning_rate": 0.0005, "loss": 1.8936591148376465, "step": 12328 }, { "epoch": 0.9719353567205361, "grad_norm": 0.6009587049484253, "learning_rate": 0.0005, "loss": 1.9300693273544312, "step": 12329 }, { "epoch": 0.972014189988175, "grad_norm": 0.7019836902618408, "learning_rate": 0.0005, "loss": 1.631054401397705, "step": 12330 }, { "epoch": 0.9720930232558139, "grad_norm": 0.7687069773674011, "learning_rate": 0.0005, "loss": 1.9044852256774902, "step": 12331 }, { "epoch": 0.972171856523453, "grad_norm": 0.8447601199150085, "learning_rate": 0.0005, "loss": 2.0061299800872803, "step": 12332 }, { "epoch": 0.9722506897910919, "grad_norm": 0.5978487730026245, "learning_rate": 0.0005, "loss": 1.7094063758850098, "step": 12333 }, { "epoch": 0.9723295230587308, "grad_norm": 0.5580906867980957, "learning_rate": 0.0005, "loss": 1.556701421737671, "step": 12334 }, { "epoch": 0.9724083563263697, "grad_norm": 0.5915252566337585, "learning_rate": 0.0005, "loss": 1.825047492980957, "step": 12335 }, { "epoch": 0.9724871895940087, "grad_norm": 0.7613011002540588, "learning_rate": 0.0005, "loss": 1.6762020587921143, "step": 12336 }, { "epoch": 0.9725660228616476, "grad_norm": 0.745249330997467, "learning_rate": 0.0005, "loss": 1.9137779474258423, "step": 12337 }, { "epoch": 0.9726448561292865, "grad_norm": 0.7025601267814636, "learning_rate": 0.0005, "loss": 1.9437891244888306, "step": 12338 }, { "epoch": 0.9727236893969256, "grad_norm": 1.0506428480148315, "learning_rate": 0.0005, "loss": 1.901761531829834, "step": 12339 }, { "epoch": 0.9728025226645645, "grad_norm": 0.7971819043159485, "learning_rate": 0.0005, "loss": 1.566359519958496, "step": 12340 }, { "epoch": 0.9728813559322034, "grad_norm": 0.5170961022377014, "learning_rate": 0.0005, "loss": 1.7768025398254395, "step": 12341 }, { "epoch": 0.9729601891998423, "grad_norm": 0.6805590391159058, "learning_rate": 0.0005, "loss": 1.7392010688781738, "step": 12342 }, { "epoch": 0.9730390224674813, "grad_norm": 0.7532625198364258, "learning_rate": 0.0005, "loss": 1.8184136152267456, "step": 12343 }, { "epoch": 0.9731178557351202, "grad_norm": 0.7048110365867615, "learning_rate": 0.0005, "loss": 1.9408013820648193, "step": 12344 }, { "epoch": 0.9731966890027591, "grad_norm": 0.7602077126502991, "learning_rate": 0.0005, "loss": 1.9054229259490967, "step": 12345 }, { "epoch": 0.9732755222703982, "grad_norm": 0.5136536359786987, "learning_rate": 0.0005, "loss": 1.706742763519287, "step": 12346 }, { "epoch": 0.9733543555380371, "grad_norm": 0.8350491523742676, "learning_rate": 0.0005, "loss": 1.9149823188781738, "step": 12347 }, { "epoch": 0.973433188805676, "grad_norm": 0.6227098703384399, "learning_rate": 0.0005, "loss": 1.6880223751068115, "step": 12348 }, { "epoch": 0.9735120220733149, "grad_norm": 0.571623682975769, "learning_rate": 0.0005, "loss": 1.8708069324493408, "step": 12349 }, { "epoch": 0.9735908553409539, "grad_norm": 0.6681063175201416, "learning_rate": 0.0005, "loss": 1.952918529510498, "step": 12350 }, { "epoch": 0.9736696886085928, "grad_norm": 0.6474777460098267, "learning_rate": 0.0005, "loss": 1.9556684494018555, "step": 12351 }, { "epoch": 0.9737485218762317, "grad_norm": 0.5848963856697083, "learning_rate": 0.0005, "loss": 1.7022093534469604, "step": 12352 }, { "epoch": 0.9738273551438708, "grad_norm": 0.593431293964386, "learning_rate": 0.0005, "loss": 1.6482805013656616, "step": 12353 }, { "epoch": 0.9739061884115097, "grad_norm": 0.5623369216918945, "learning_rate": 0.0005, "loss": 1.7868947982788086, "step": 12354 }, { "epoch": 0.9739850216791486, "grad_norm": 0.5630538463592529, "learning_rate": 0.0005, "loss": 1.7397403717041016, "step": 12355 }, { "epoch": 0.9740638549467876, "grad_norm": 0.7181075811386108, "learning_rate": 0.0005, "loss": 1.9466602802276611, "step": 12356 }, { "epoch": 0.9741426882144265, "grad_norm": 0.4975186884403229, "learning_rate": 0.0005, "loss": 1.770988941192627, "step": 12357 }, { "epoch": 0.9742215214820654, "grad_norm": 0.5634474754333496, "learning_rate": 0.0005, "loss": 1.9293763637542725, "step": 12358 }, { "epoch": 0.9743003547497043, "grad_norm": 0.533232569694519, "learning_rate": 0.0005, "loss": 1.8865680694580078, "step": 12359 }, { "epoch": 0.9743791880173434, "grad_norm": 0.5593099594116211, "learning_rate": 0.0005, "loss": 1.7603399753570557, "step": 12360 }, { "epoch": 0.9744580212849823, "grad_norm": 0.5686172842979431, "learning_rate": 0.0005, "loss": 1.8039227724075317, "step": 12361 }, { "epoch": 0.9745368545526212, "grad_norm": 0.4652913510799408, "learning_rate": 0.0005, "loss": 1.7600104808807373, "step": 12362 }, { "epoch": 0.9746156878202602, "grad_norm": 0.5294466614723206, "learning_rate": 0.0005, "loss": 1.927474021911621, "step": 12363 }, { "epoch": 0.9746945210878991, "grad_norm": 0.5959737300872803, "learning_rate": 0.0005, "loss": 1.5222856998443604, "step": 12364 }, { "epoch": 0.974773354355538, "grad_norm": 0.5986389517784119, "learning_rate": 0.0005, "loss": 1.7753410339355469, "step": 12365 }, { "epoch": 0.9748521876231769, "grad_norm": 0.6642602682113647, "learning_rate": 0.0005, "loss": 1.771979570388794, "step": 12366 }, { "epoch": 0.974931020890816, "grad_norm": 0.8562294244766235, "learning_rate": 0.0005, "loss": 1.8522967100143433, "step": 12367 }, { "epoch": 0.9750098541584549, "grad_norm": 0.5514124631881714, "learning_rate": 0.0005, "loss": 1.6051709651947021, "step": 12368 }, { "epoch": 0.9750886874260938, "grad_norm": 0.6524640321731567, "learning_rate": 0.0005, "loss": 1.7596298456192017, "step": 12369 }, { "epoch": 0.9751675206937328, "grad_norm": 0.6384069323539734, "learning_rate": 0.0005, "loss": 1.8849393129348755, "step": 12370 }, { "epoch": 0.9752463539613717, "grad_norm": 0.7628166079521179, "learning_rate": 0.0005, "loss": 2.13651704788208, "step": 12371 }, { "epoch": 0.9753251872290106, "grad_norm": 0.5053326487541199, "learning_rate": 0.0005, "loss": 1.7541204690933228, "step": 12372 }, { "epoch": 0.9754040204966495, "grad_norm": 0.7724643349647522, "learning_rate": 0.0005, "loss": 1.7973445653915405, "step": 12373 }, { "epoch": 0.9754828537642886, "grad_norm": 1.0715636014938354, "learning_rate": 0.0005, "loss": 1.7713587284088135, "step": 12374 }, { "epoch": 0.9755616870319275, "grad_norm": 0.7009636163711548, "learning_rate": 0.0005, "loss": 1.9096314907073975, "step": 12375 }, { "epoch": 0.9756405202995664, "grad_norm": 0.6600708365440369, "learning_rate": 0.0005, "loss": 1.9125230312347412, "step": 12376 }, { "epoch": 0.9757193535672054, "grad_norm": 0.641528308391571, "learning_rate": 0.0005, "loss": 2.0375146865844727, "step": 12377 }, { "epoch": 0.9757981868348443, "grad_norm": 0.6034340858459473, "learning_rate": 0.0005, "loss": 1.6698194742202759, "step": 12378 }, { "epoch": 0.9758770201024832, "grad_norm": 0.5914421081542969, "learning_rate": 0.0005, "loss": 1.5453181266784668, "step": 12379 }, { "epoch": 0.9759558533701221, "grad_norm": 0.6779699921607971, "learning_rate": 0.0005, "loss": 2.0400102138519287, "step": 12380 }, { "epoch": 0.9760346866377612, "grad_norm": 0.6500849723815918, "learning_rate": 0.0005, "loss": 1.8638601303100586, "step": 12381 }, { "epoch": 0.9761135199054001, "grad_norm": 0.7954310774803162, "learning_rate": 0.0005, "loss": 1.759164571762085, "step": 12382 }, { "epoch": 0.976192353173039, "grad_norm": 0.9708516001701355, "learning_rate": 0.0005, "loss": 1.676373839378357, "step": 12383 }, { "epoch": 0.976271186440678, "grad_norm": 0.5687318444252014, "learning_rate": 0.0005, "loss": 1.8236720561981201, "step": 12384 }, { "epoch": 0.9763500197083169, "grad_norm": 0.6276861429214478, "learning_rate": 0.0005, "loss": 1.5245356559753418, "step": 12385 }, { "epoch": 0.9764288529759558, "grad_norm": 0.6287492513656616, "learning_rate": 0.0005, "loss": 1.8275249004364014, "step": 12386 }, { "epoch": 0.9765076862435949, "grad_norm": 0.6178817749023438, "learning_rate": 0.0005, "loss": 1.8878865242004395, "step": 12387 }, { "epoch": 0.9765865195112338, "grad_norm": 0.601876437664032, "learning_rate": 0.0005, "loss": 1.8314976692199707, "step": 12388 }, { "epoch": 0.9766653527788727, "grad_norm": 0.5990644097328186, "learning_rate": 0.0005, "loss": 1.8050482273101807, "step": 12389 }, { "epoch": 0.9767441860465116, "grad_norm": 0.5450709462165833, "learning_rate": 0.0005, "loss": 1.9046432971954346, "step": 12390 }, { "epoch": 0.9768230193141506, "grad_norm": 0.6661045551300049, "learning_rate": 0.0005, "loss": 1.4773271083831787, "step": 12391 }, { "epoch": 0.9769018525817895, "grad_norm": 0.5837551355361938, "learning_rate": 0.0005, "loss": 1.9472014904022217, "step": 12392 }, { "epoch": 0.9769806858494284, "grad_norm": 0.653218150138855, "learning_rate": 0.0005, "loss": 1.7371407747268677, "step": 12393 }, { "epoch": 0.9770595191170675, "grad_norm": 0.49779242277145386, "learning_rate": 0.0005, "loss": 1.7190115451812744, "step": 12394 }, { "epoch": 0.9771383523847064, "grad_norm": 0.6274150013923645, "learning_rate": 0.0005, "loss": 1.664586067199707, "step": 12395 }, { "epoch": 0.9772171856523453, "grad_norm": 0.6088120341300964, "learning_rate": 0.0005, "loss": 1.8010716438293457, "step": 12396 }, { "epoch": 0.9772960189199842, "grad_norm": 0.6953388452529907, "learning_rate": 0.0005, "loss": 1.6783392429351807, "step": 12397 }, { "epoch": 0.9773748521876232, "grad_norm": 0.6708552837371826, "learning_rate": 0.0005, "loss": 1.8431837558746338, "step": 12398 }, { "epoch": 0.9774536854552621, "grad_norm": 0.6870183944702148, "learning_rate": 0.0005, "loss": 1.9585896730422974, "step": 12399 }, { "epoch": 0.977532518722901, "grad_norm": 0.5892333388328552, "learning_rate": 0.0005, "loss": 1.854934573173523, "step": 12400 }, { "epoch": 0.9776113519905401, "grad_norm": 0.8035091757774353, "learning_rate": 0.0005, "loss": 1.9451327323913574, "step": 12401 }, { "epoch": 0.977690185258179, "grad_norm": 0.5021173357963562, "learning_rate": 0.0005, "loss": 1.769713044166565, "step": 12402 }, { "epoch": 0.9777690185258179, "grad_norm": 0.8203129172325134, "learning_rate": 0.0005, "loss": 1.7853648662567139, "step": 12403 }, { "epoch": 0.9778478517934568, "grad_norm": 0.5300228595733643, "learning_rate": 0.0005, "loss": 1.7628333568572998, "step": 12404 }, { "epoch": 0.9779266850610958, "grad_norm": 0.6674106121063232, "learning_rate": 0.0005, "loss": 1.7721364498138428, "step": 12405 }, { "epoch": 0.9780055183287347, "grad_norm": 0.6555780172348022, "learning_rate": 0.0005, "loss": 1.938126802444458, "step": 12406 }, { "epoch": 0.9780843515963736, "grad_norm": 0.6040320992469788, "learning_rate": 0.0005, "loss": 1.832406997680664, "step": 12407 }, { "epoch": 0.9781631848640127, "grad_norm": 0.6736385226249695, "learning_rate": 0.0005, "loss": 1.8467588424682617, "step": 12408 }, { "epoch": 0.9782420181316516, "grad_norm": 0.5884456634521484, "learning_rate": 0.0005, "loss": 1.8098270893096924, "step": 12409 }, { "epoch": 0.9783208513992905, "grad_norm": 0.6323280930519104, "learning_rate": 0.0005, "loss": 1.780060052871704, "step": 12410 }, { "epoch": 0.9783996846669294, "grad_norm": 0.5971824526786804, "learning_rate": 0.0005, "loss": 1.8256614208221436, "step": 12411 }, { "epoch": 0.9784785179345684, "grad_norm": 0.5868951678276062, "learning_rate": 0.0005, "loss": 1.8166890144348145, "step": 12412 }, { "epoch": 0.9785573512022073, "grad_norm": 0.6727818250656128, "learning_rate": 0.0005, "loss": 1.8658456802368164, "step": 12413 }, { "epoch": 0.9786361844698462, "grad_norm": 0.7585088014602661, "learning_rate": 0.0005, "loss": 1.9731202125549316, "step": 12414 }, { "epoch": 0.9787150177374853, "grad_norm": 0.9351969957351685, "learning_rate": 0.0005, "loss": 2.1193251609802246, "step": 12415 }, { "epoch": 0.9787938510051242, "grad_norm": 0.7630964517593384, "learning_rate": 0.0005, "loss": 1.8488457202911377, "step": 12416 }, { "epoch": 0.9788726842727631, "grad_norm": 0.6587035059928894, "learning_rate": 0.0005, "loss": 1.545419692993164, "step": 12417 }, { "epoch": 0.978951517540402, "grad_norm": 0.714815616607666, "learning_rate": 0.0005, "loss": 1.7648118734359741, "step": 12418 }, { "epoch": 0.979030350808041, "grad_norm": 0.6213857531547546, "learning_rate": 0.0005, "loss": 1.534016728401184, "step": 12419 }, { "epoch": 0.9791091840756799, "grad_norm": 0.5507486462593079, "learning_rate": 0.0005, "loss": 1.5132110118865967, "step": 12420 }, { "epoch": 0.9791880173433188, "grad_norm": 0.7641070485115051, "learning_rate": 0.0005, "loss": 1.9478325843811035, "step": 12421 }, { "epoch": 0.9792668506109579, "grad_norm": 0.5403378009796143, "learning_rate": 0.0005, "loss": 1.8414756059646606, "step": 12422 }, { "epoch": 0.9793456838785968, "grad_norm": 0.5758997797966003, "learning_rate": 0.0005, "loss": 1.7022809982299805, "step": 12423 }, { "epoch": 0.9794245171462357, "grad_norm": 0.5343966484069824, "learning_rate": 0.0005, "loss": 1.8230319023132324, "step": 12424 }, { "epoch": 0.9795033504138747, "grad_norm": 0.5569162964820862, "learning_rate": 0.0005, "loss": 1.6062705516815186, "step": 12425 }, { "epoch": 0.9795821836815136, "grad_norm": 0.5959395170211792, "learning_rate": 0.0005, "loss": 1.6309703588485718, "step": 12426 }, { "epoch": 0.9796610169491525, "grad_norm": 1.222723126411438, "learning_rate": 0.0005, "loss": 2.213409423828125, "step": 12427 }, { "epoch": 0.9797398502167914, "grad_norm": 0.6544755101203918, "learning_rate": 0.0005, "loss": 1.9406251907348633, "step": 12428 }, { "epoch": 0.9798186834844305, "grad_norm": 0.5918619632720947, "learning_rate": 0.0005, "loss": 1.8600258827209473, "step": 12429 }, { "epoch": 0.9798975167520694, "grad_norm": 0.5649130344390869, "learning_rate": 0.0005, "loss": 1.7815316915512085, "step": 12430 }, { "epoch": 0.9799763500197083, "grad_norm": 0.6165576577186584, "learning_rate": 0.0005, "loss": 1.587428092956543, "step": 12431 }, { "epoch": 0.9800551832873473, "grad_norm": 0.7845364809036255, "learning_rate": 0.0005, "loss": 2.170982837677002, "step": 12432 }, { "epoch": 0.9801340165549862, "grad_norm": 0.5506788492202759, "learning_rate": 0.0005, "loss": 1.8354196548461914, "step": 12433 }, { "epoch": 0.9802128498226251, "grad_norm": 0.6234975457191467, "learning_rate": 0.0005, "loss": 1.5728899240493774, "step": 12434 }, { "epoch": 0.980291683090264, "grad_norm": 0.581449031829834, "learning_rate": 0.0005, "loss": 1.870469093322754, "step": 12435 }, { "epoch": 0.9803705163579031, "grad_norm": 0.6364299654960632, "learning_rate": 0.0005, "loss": 1.8787267208099365, "step": 12436 }, { "epoch": 0.980449349625542, "grad_norm": 0.8585332036018372, "learning_rate": 0.0005, "loss": 1.9439765214920044, "step": 12437 }, { "epoch": 0.9805281828931809, "grad_norm": 0.5625652074813843, "learning_rate": 0.0005, "loss": 1.7185287475585938, "step": 12438 }, { "epoch": 0.9806070161608199, "grad_norm": 0.6184680461883545, "learning_rate": 0.0005, "loss": 2.001927375793457, "step": 12439 }, { "epoch": 0.9806858494284588, "grad_norm": 0.5284099578857422, "learning_rate": 0.0005, "loss": 1.5652811527252197, "step": 12440 }, { "epoch": 0.9807646826960977, "grad_norm": 0.5056220293045044, "learning_rate": 0.0005, "loss": 1.5474658012390137, "step": 12441 }, { "epoch": 0.9808435159637366, "grad_norm": 0.5362501740455627, "learning_rate": 0.0005, "loss": 1.8257588148117065, "step": 12442 }, { "epoch": 0.9809223492313757, "grad_norm": 0.6732532382011414, "learning_rate": 0.0005, "loss": 1.7305831909179688, "step": 12443 }, { "epoch": 0.9810011824990146, "grad_norm": 0.6162886023521423, "learning_rate": 0.0005, "loss": 1.811568021774292, "step": 12444 }, { "epoch": 0.9810800157666535, "grad_norm": 0.8794226050376892, "learning_rate": 0.0005, "loss": 2.1472160816192627, "step": 12445 }, { "epoch": 0.9811588490342925, "grad_norm": 0.614498496055603, "learning_rate": 0.0005, "loss": 2.0648231506347656, "step": 12446 }, { "epoch": 0.9812376823019314, "grad_norm": 0.5343084931373596, "learning_rate": 0.0005, "loss": 1.8136694431304932, "step": 12447 }, { "epoch": 0.9813165155695703, "grad_norm": 0.4951323866844177, "learning_rate": 0.0005, "loss": 1.688073754310608, "step": 12448 }, { "epoch": 0.9813953488372092, "grad_norm": 0.7456504106521606, "learning_rate": 0.0005, "loss": 2.0643792152404785, "step": 12449 }, { "epoch": 0.9814741821048483, "grad_norm": 0.487907737493515, "learning_rate": 0.0005, "loss": 1.8798086643218994, "step": 12450 }, { "epoch": 0.9815530153724872, "grad_norm": 0.4896201491355896, "learning_rate": 0.0005, "loss": 1.6405417919158936, "step": 12451 }, { "epoch": 0.9816318486401261, "grad_norm": 0.519051194190979, "learning_rate": 0.0005, "loss": 1.6244971752166748, "step": 12452 }, { "epoch": 0.9817106819077651, "grad_norm": 0.7231850624084473, "learning_rate": 0.0005, "loss": 1.8388147354125977, "step": 12453 }, { "epoch": 0.981789515175404, "grad_norm": 0.5149953961372375, "learning_rate": 0.0005, "loss": 1.7519810199737549, "step": 12454 }, { "epoch": 0.9818683484430429, "grad_norm": 0.6055126786231995, "learning_rate": 0.0005, "loss": 1.5392320156097412, "step": 12455 }, { "epoch": 0.981947181710682, "grad_norm": 0.5362564921379089, "learning_rate": 0.0005, "loss": 1.7866655588150024, "step": 12456 }, { "epoch": 0.9820260149783209, "grad_norm": 0.7436782121658325, "learning_rate": 0.0005, "loss": 1.843348503112793, "step": 12457 }, { "epoch": 0.9821048482459598, "grad_norm": 0.441071480512619, "learning_rate": 0.0005, "loss": 1.5364930629730225, "step": 12458 }, { "epoch": 0.9821836815135987, "grad_norm": 0.604056179523468, "learning_rate": 0.0005, "loss": 1.6638461351394653, "step": 12459 }, { "epoch": 0.9822625147812377, "grad_norm": 0.6249485015869141, "learning_rate": 0.0005, "loss": 1.932807207107544, "step": 12460 }, { "epoch": 0.9823413480488766, "grad_norm": 0.6562855243682861, "learning_rate": 0.0005, "loss": 1.7470741271972656, "step": 12461 }, { "epoch": 0.9824201813165155, "grad_norm": 0.5374111533164978, "learning_rate": 0.0005, "loss": 1.7330647706985474, "step": 12462 }, { "epoch": 0.9824990145841546, "grad_norm": 0.7661451697349548, "learning_rate": 0.0005, "loss": 2.094712734222412, "step": 12463 }, { "epoch": 0.9825778478517935, "grad_norm": 0.5416646599769592, "learning_rate": 0.0005, "loss": 1.7827651500701904, "step": 12464 }, { "epoch": 0.9826566811194324, "grad_norm": 0.5338551998138428, "learning_rate": 0.0005, "loss": 1.6645426750183105, "step": 12465 }, { "epoch": 0.9827355143870713, "grad_norm": 0.7737277746200562, "learning_rate": 0.0005, "loss": 1.9733870029449463, "step": 12466 }, { "epoch": 0.9828143476547103, "grad_norm": 0.6134556531906128, "learning_rate": 0.0005, "loss": 1.8830764293670654, "step": 12467 }, { "epoch": 0.9828931809223492, "grad_norm": 0.555975615978241, "learning_rate": 0.0005, "loss": 1.9619271755218506, "step": 12468 }, { "epoch": 0.9829720141899881, "grad_norm": 0.563356876373291, "learning_rate": 0.0005, "loss": 2.190901756286621, "step": 12469 }, { "epoch": 0.9830508474576272, "grad_norm": 0.7216205596923828, "learning_rate": 0.0005, "loss": 1.9410157203674316, "step": 12470 }, { "epoch": 0.9831296807252661, "grad_norm": 0.7173407673835754, "learning_rate": 0.0005, "loss": 1.9739007949829102, "step": 12471 }, { "epoch": 0.983208513992905, "grad_norm": 0.5929383039474487, "learning_rate": 0.0005, "loss": 1.7165913581848145, "step": 12472 }, { "epoch": 0.9832873472605439, "grad_norm": 0.5888204574584961, "learning_rate": 0.0005, "loss": 1.7190170288085938, "step": 12473 }, { "epoch": 0.9833661805281829, "grad_norm": 0.5741307735443115, "learning_rate": 0.0005, "loss": 1.723568081855774, "step": 12474 }, { "epoch": 0.9834450137958218, "grad_norm": 0.6027041673660278, "learning_rate": 0.0005, "loss": 1.7455296516418457, "step": 12475 }, { "epoch": 0.9835238470634607, "grad_norm": 0.7533004283905029, "learning_rate": 0.0005, "loss": 1.8742932081222534, "step": 12476 }, { "epoch": 0.9836026803310998, "grad_norm": 0.7100922465324402, "learning_rate": 0.0005, "loss": 1.8709230422973633, "step": 12477 }, { "epoch": 0.9836815135987387, "grad_norm": 1.9665110111236572, "learning_rate": 0.0005, "loss": 2.0273075103759766, "step": 12478 }, { "epoch": 0.9837603468663776, "grad_norm": 0.8675937056541443, "learning_rate": 0.0005, "loss": 2.139955997467041, "step": 12479 }, { "epoch": 0.9838391801340165, "grad_norm": 0.6506417989730835, "learning_rate": 0.0005, "loss": 1.9493110179901123, "step": 12480 }, { "epoch": 0.9839180134016555, "grad_norm": 0.7058857679367065, "learning_rate": 0.0005, "loss": 1.7646409273147583, "step": 12481 }, { "epoch": 0.9839968466692944, "grad_norm": 0.7024737000465393, "learning_rate": 0.0005, "loss": 2.0768537521362305, "step": 12482 }, { "epoch": 0.9840756799369333, "grad_norm": 0.7101999521255493, "learning_rate": 0.0005, "loss": 1.9587959051132202, "step": 12483 }, { "epoch": 0.9841545132045724, "grad_norm": 0.5931584239006042, "learning_rate": 0.0005, "loss": 1.6774994134902954, "step": 12484 }, { "epoch": 0.9842333464722113, "grad_norm": 0.6974093317985535, "learning_rate": 0.0005, "loss": 1.777090311050415, "step": 12485 }, { "epoch": 0.9843121797398502, "grad_norm": 0.5645046830177307, "learning_rate": 0.0005, "loss": 1.7327055931091309, "step": 12486 }, { "epoch": 0.9843910130074892, "grad_norm": 0.6747199892997742, "learning_rate": 0.0005, "loss": 1.835540533065796, "step": 12487 }, { "epoch": 0.9844698462751281, "grad_norm": 0.6812103390693665, "learning_rate": 0.0005, "loss": 1.6432323455810547, "step": 12488 }, { "epoch": 0.984548679542767, "grad_norm": 0.766654908657074, "learning_rate": 0.0005, "loss": 1.5988503694534302, "step": 12489 }, { "epoch": 0.984627512810406, "grad_norm": 0.6681013703346252, "learning_rate": 0.0005, "loss": 1.752163290977478, "step": 12490 }, { "epoch": 0.984706346078045, "grad_norm": 0.6443748474121094, "learning_rate": 0.0005, "loss": 1.6826845407485962, "step": 12491 }, { "epoch": 0.9847851793456839, "grad_norm": 0.7295851111412048, "learning_rate": 0.0005, "loss": 2.036809206008911, "step": 12492 }, { "epoch": 0.9848640126133228, "grad_norm": 0.6306408047676086, "learning_rate": 0.0005, "loss": 1.8293465375900269, "step": 12493 }, { "epoch": 0.9849428458809618, "grad_norm": 0.8098365664482117, "learning_rate": 0.0005, "loss": 1.8975696563720703, "step": 12494 }, { "epoch": 0.9850216791486007, "grad_norm": 0.6968610882759094, "learning_rate": 0.0005, "loss": 2.147322177886963, "step": 12495 }, { "epoch": 0.9851005124162396, "grad_norm": 0.6101803183555603, "learning_rate": 0.0005, "loss": 1.7208826541900635, "step": 12496 }, { "epoch": 0.9851793456838785, "grad_norm": 0.5373969674110413, "learning_rate": 0.0005, "loss": 1.6888160705566406, "step": 12497 }, { "epoch": 0.9852581789515176, "grad_norm": 0.576948881149292, "learning_rate": 0.0005, "loss": 1.6007531881332397, "step": 12498 }, { "epoch": 0.9853370122191565, "grad_norm": 0.6725378632545471, "learning_rate": 0.0005, "loss": 1.6867058277130127, "step": 12499 }, { "epoch": 0.9854158454867954, "grad_norm": 0.5520244240760803, "learning_rate": 0.0005, "loss": 1.748913049697876, "step": 12500 }, { "epoch": 0.9854946787544344, "grad_norm": 0.6165617108345032, "learning_rate": 0.0005, "loss": 1.443497896194458, "step": 12501 }, { "epoch": 0.9855735120220733, "grad_norm": 0.5312541127204895, "learning_rate": 0.0005, "loss": 1.389369010925293, "step": 12502 }, { "epoch": 0.9856523452897122, "grad_norm": 0.5875048041343689, "learning_rate": 0.0005, "loss": 1.9857091903686523, "step": 12503 }, { "epoch": 0.9857311785573512, "grad_norm": 0.5781737565994263, "learning_rate": 0.0005, "loss": 1.7555463314056396, "step": 12504 }, { "epoch": 0.9858100118249902, "grad_norm": 0.7100691199302673, "learning_rate": 0.0005, "loss": 2.066924571990967, "step": 12505 }, { "epoch": 0.9858888450926291, "grad_norm": 0.5350795388221741, "learning_rate": 0.0005, "loss": 1.7578611373901367, "step": 12506 }, { "epoch": 0.985967678360268, "grad_norm": 0.6190508008003235, "learning_rate": 0.0005, "loss": 1.8941199779510498, "step": 12507 }, { "epoch": 0.986046511627907, "grad_norm": 0.4935775399208069, "learning_rate": 0.0005, "loss": 1.6642322540283203, "step": 12508 }, { "epoch": 0.9861253448955459, "grad_norm": 0.6262416839599609, "learning_rate": 0.0005, "loss": 1.7319796085357666, "step": 12509 }, { "epoch": 0.9862041781631848, "grad_norm": 0.6110016703605652, "learning_rate": 0.0005, "loss": 1.495500087738037, "step": 12510 }, { "epoch": 0.9862830114308238, "grad_norm": 0.6140185594558716, "learning_rate": 0.0005, "loss": 1.703547477722168, "step": 12511 }, { "epoch": 0.9863618446984628, "grad_norm": 1.0519944429397583, "learning_rate": 0.0005, "loss": 1.6584843397140503, "step": 12512 }, { "epoch": 0.9864406779661017, "grad_norm": 1.2857515811920166, "learning_rate": 0.0005, "loss": 1.6459124088287354, "step": 12513 }, { "epoch": 0.9865195112337406, "grad_norm": 0.7478846907615662, "learning_rate": 0.0005, "loss": 1.7044181823730469, "step": 12514 }, { "epoch": 0.9865983445013796, "grad_norm": 0.5886471271514893, "learning_rate": 0.0005, "loss": 1.631568431854248, "step": 12515 }, { "epoch": 0.9866771777690185, "grad_norm": 0.6630094647407532, "learning_rate": 0.0005, "loss": 1.6153972148895264, "step": 12516 }, { "epoch": 0.9867560110366574, "grad_norm": 1.3224736452102661, "learning_rate": 0.0005, "loss": 1.703702688217163, "step": 12517 }, { "epoch": 0.9868348443042965, "grad_norm": 0.8840603828430176, "learning_rate": 0.0005, "loss": 1.8942084312438965, "step": 12518 }, { "epoch": 0.9869136775719354, "grad_norm": 0.8007849454879761, "learning_rate": 0.0005, "loss": 1.8917772769927979, "step": 12519 }, { "epoch": 0.9869925108395743, "grad_norm": 0.8108237385749817, "learning_rate": 0.0005, "loss": 2.2665019035339355, "step": 12520 }, { "epoch": 0.9870713441072132, "grad_norm": 0.6003510355949402, "learning_rate": 0.0005, "loss": 1.710741400718689, "step": 12521 }, { "epoch": 0.9871501773748522, "grad_norm": 0.8976725935935974, "learning_rate": 0.0005, "loss": 1.865462303161621, "step": 12522 }, { "epoch": 0.9872290106424911, "grad_norm": 0.6295019388198853, "learning_rate": 0.0005, "loss": 1.585730791091919, "step": 12523 }, { "epoch": 0.98730784391013, "grad_norm": 0.6524795293807983, "learning_rate": 0.0005, "loss": 1.9068288803100586, "step": 12524 }, { "epoch": 0.9873866771777691, "grad_norm": 0.6623415946960449, "learning_rate": 0.0005, "loss": 1.747004747390747, "step": 12525 }, { "epoch": 0.987465510445408, "grad_norm": 0.6206986308097839, "learning_rate": 0.0005, "loss": 1.8367377519607544, "step": 12526 }, { "epoch": 0.9875443437130469, "grad_norm": 0.7346264719963074, "learning_rate": 0.0005, "loss": 1.5606043338775635, "step": 12527 }, { "epoch": 0.9876231769806858, "grad_norm": 0.6589102745056152, "learning_rate": 0.0005, "loss": 1.847705364227295, "step": 12528 }, { "epoch": 0.9877020102483248, "grad_norm": 0.5222496390342712, "learning_rate": 0.0005, "loss": 1.7994647026062012, "step": 12529 }, { "epoch": 0.9877808435159637, "grad_norm": 0.5311022996902466, "learning_rate": 0.0005, "loss": 1.8728463649749756, "step": 12530 }, { "epoch": 0.9878596767836026, "grad_norm": 0.6026343703269958, "learning_rate": 0.0005, "loss": 1.5202487707138062, "step": 12531 }, { "epoch": 0.9879385100512417, "grad_norm": 0.587706446647644, "learning_rate": 0.0005, "loss": 1.8591315746307373, "step": 12532 }, { "epoch": 0.9880173433188806, "grad_norm": 0.7643780708312988, "learning_rate": 0.0005, "loss": 1.932222843170166, "step": 12533 }, { "epoch": 0.9880961765865195, "grad_norm": 0.7876136898994446, "learning_rate": 0.0005, "loss": 1.820217251777649, "step": 12534 }, { "epoch": 0.9881750098541584, "grad_norm": 0.7220777869224548, "learning_rate": 0.0005, "loss": 1.6749522686004639, "step": 12535 }, { "epoch": 0.9882538431217974, "grad_norm": 0.7406767010688782, "learning_rate": 0.0005, "loss": 1.94535493850708, "step": 12536 }, { "epoch": 0.9883326763894363, "grad_norm": 0.6580744385719299, "learning_rate": 0.0005, "loss": 1.8111310005187988, "step": 12537 }, { "epoch": 0.9884115096570752, "grad_norm": 0.5993771553039551, "learning_rate": 0.0005, "loss": 1.837872862815857, "step": 12538 }, { "epoch": 0.9884903429247143, "grad_norm": 0.6929386854171753, "learning_rate": 0.0005, "loss": 1.7913532257080078, "step": 12539 }, { "epoch": 0.9885691761923532, "grad_norm": 0.7097368836402893, "learning_rate": 0.0005, "loss": 1.8557931184768677, "step": 12540 }, { "epoch": 0.9886480094599921, "grad_norm": 0.5162842869758606, "learning_rate": 0.0005, "loss": 1.565943956375122, "step": 12541 }, { "epoch": 0.988726842727631, "grad_norm": 1.9138237237930298, "learning_rate": 0.0005, "loss": 1.5936634540557861, "step": 12542 }, { "epoch": 0.98880567599527, "grad_norm": 0.6549236178398132, "learning_rate": 0.0005, "loss": 1.8358211517333984, "step": 12543 }, { "epoch": 0.9888845092629089, "grad_norm": 0.7157192230224609, "learning_rate": 0.0005, "loss": 1.8505001068115234, "step": 12544 }, { "epoch": 0.9889633425305479, "grad_norm": 0.5580142140388489, "learning_rate": 0.0005, "loss": 1.697204828262329, "step": 12545 }, { "epoch": 0.9890421757981869, "grad_norm": 0.6309951543807983, "learning_rate": 0.0005, "loss": 1.7413872480392456, "step": 12546 }, { "epoch": 0.9891210090658258, "grad_norm": 0.5519231557846069, "learning_rate": 0.0005, "loss": 1.7458043098449707, "step": 12547 }, { "epoch": 0.9891998423334647, "grad_norm": 0.6972495317459106, "learning_rate": 0.0005, "loss": 1.6939780712127686, "step": 12548 }, { "epoch": 0.9892786756011037, "grad_norm": 0.6596774458885193, "learning_rate": 0.0005, "loss": 1.8061693906784058, "step": 12549 }, { "epoch": 0.9893575088687426, "grad_norm": 0.7083027958869934, "learning_rate": 0.0005, "loss": 2.0667967796325684, "step": 12550 }, { "epoch": 0.9894363421363815, "grad_norm": 0.668195366859436, "learning_rate": 0.0005, "loss": 2.0766119956970215, "step": 12551 }, { "epoch": 0.9895151754040205, "grad_norm": 0.6071955561637878, "learning_rate": 0.0005, "loss": 1.8120931386947632, "step": 12552 }, { "epoch": 0.9895940086716595, "grad_norm": 0.5848173499107361, "learning_rate": 0.0005, "loss": 1.770229697227478, "step": 12553 }, { "epoch": 0.9896728419392984, "grad_norm": 0.816150963306427, "learning_rate": 0.0005, "loss": 2.062044620513916, "step": 12554 }, { "epoch": 0.9897516752069373, "grad_norm": 0.6458075046539307, "learning_rate": 0.0005, "loss": 1.9529390335083008, "step": 12555 }, { "epoch": 0.9898305084745763, "grad_norm": 0.6778939962387085, "learning_rate": 0.0005, "loss": 1.8642449378967285, "step": 12556 }, { "epoch": 0.9899093417422152, "grad_norm": 0.7366519570350647, "learning_rate": 0.0005, "loss": 1.7522170543670654, "step": 12557 }, { "epoch": 0.9899881750098541, "grad_norm": 0.5518277287483215, "learning_rate": 0.0005, "loss": 1.526111125946045, "step": 12558 }, { "epoch": 0.990067008277493, "grad_norm": 0.6147257685661316, "learning_rate": 0.0005, "loss": 1.7650158405303955, "step": 12559 }, { "epoch": 0.9901458415451321, "grad_norm": 0.6152243614196777, "learning_rate": 0.0005, "loss": 1.8753442764282227, "step": 12560 }, { "epoch": 0.990224674812771, "grad_norm": 0.6526151299476624, "learning_rate": 0.0005, "loss": 1.7946827411651611, "step": 12561 }, { "epoch": 0.9903035080804099, "grad_norm": 0.5803068280220032, "learning_rate": 0.0005, "loss": 1.63485586643219, "step": 12562 }, { "epoch": 0.9903823413480489, "grad_norm": 0.6677099466323853, "learning_rate": 0.0005, "loss": 1.8861076831817627, "step": 12563 }, { "epoch": 0.9904611746156878, "grad_norm": 0.5652481317520142, "learning_rate": 0.0005, "loss": 1.547911286354065, "step": 12564 }, { "epoch": 0.9905400078833267, "grad_norm": 0.7461881637573242, "learning_rate": 0.0005, "loss": 1.932950735092163, "step": 12565 }, { "epoch": 0.9906188411509657, "grad_norm": 0.5990780591964722, "learning_rate": 0.0005, "loss": 1.7796286344528198, "step": 12566 }, { "epoch": 0.9906976744186047, "grad_norm": 0.6764493584632874, "learning_rate": 0.0005, "loss": 1.7192857265472412, "step": 12567 }, { "epoch": 0.9907765076862436, "grad_norm": 0.6726474761962891, "learning_rate": 0.0005, "loss": 1.9649550914764404, "step": 12568 }, { "epoch": 0.9908553409538825, "grad_norm": 0.5552232265472412, "learning_rate": 0.0005, "loss": 1.8530771732330322, "step": 12569 }, { "epoch": 0.9909341742215215, "grad_norm": 0.5199099779129028, "learning_rate": 0.0005, "loss": 1.5053640604019165, "step": 12570 }, { "epoch": 0.9910130074891604, "grad_norm": 1.1084295511245728, "learning_rate": 0.0005, "loss": 1.6572920083999634, "step": 12571 }, { "epoch": 0.9910918407567993, "grad_norm": 0.7123056650161743, "learning_rate": 0.0005, "loss": 2.0443854331970215, "step": 12572 }, { "epoch": 0.9911706740244383, "grad_norm": 0.5708014369010925, "learning_rate": 0.0005, "loss": 1.956298828125, "step": 12573 }, { "epoch": 0.9912495072920773, "grad_norm": 0.4754754900932312, "learning_rate": 0.0005, "loss": 1.795450210571289, "step": 12574 }, { "epoch": 0.9913283405597162, "grad_norm": 0.5433998703956604, "learning_rate": 0.0005, "loss": 1.9053258895874023, "step": 12575 }, { "epoch": 0.9914071738273551, "grad_norm": 0.616651713848114, "learning_rate": 0.0005, "loss": 1.7399381399154663, "step": 12576 }, { "epoch": 0.9914860070949941, "grad_norm": 0.5376543402671814, "learning_rate": 0.0005, "loss": 1.7627944946289062, "step": 12577 }, { "epoch": 0.991564840362633, "grad_norm": 0.5092484951019287, "learning_rate": 0.0005, "loss": 1.547722578048706, "step": 12578 }, { "epoch": 0.991643673630272, "grad_norm": 0.5448411107063293, "learning_rate": 0.0005, "loss": 1.7957359552383423, "step": 12579 }, { "epoch": 0.991722506897911, "grad_norm": 0.5859760046005249, "learning_rate": 0.0005, "loss": 1.7388317584991455, "step": 12580 }, { "epoch": 0.9918013401655499, "grad_norm": 0.5439971089363098, "learning_rate": 0.0005, "loss": 1.6410043239593506, "step": 12581 }, { "epoch": 0.9918801734331888, "grad_norm": 0.5470061302185059, "learning_rate": 0.0005, "loss": 1.6598035097122192, "step": 12582 }, { "epoch": 0.9919590067008277, "grad_norm": 0.6130142211914062, "learning_rate": 0.0005, "loss": 1.7525310516357422, "step": 12583 }, { "epoch": 0.9920378399684667, "grad_norm": 0.6138441562652588, "learning_rate": 0.0005, "loss": 1.8266878128051758, "step": 12584 }, { "epoch": 0.9921166732361056, "grad_norm": 0.6384431719779968, "learning_rate": 0.0005, "loss": 1.9596974849700928, "step": 12585 }, { "epoch": 0.9921955065037446, "grad_norm": 0.827504575252533, "learning_rate": 0.0005, "loss": 1.9750372171401978, "step": 12586 }, { "epoch": 0.9922743397713836, "grad_norm": 0.5750557780265808, "learning_rate": 0.0005, "loss": 1.7446870803833008, "step": 12587 }, { "epoch": 0.9923531730390225, "grad_norm": 0.591456949710846, "learning_rate": 0.0005, "loss": 1.598036527633667, "step": 12588 }, { "epoch": 0.9924320063066614, "grad_norm": 0.5284631848335266, "learning_rate": 0.0005, "loss": 1.7796496152877808, "step": 12589 }, { "epoch": 0.9925108395743003, "grad_norm": 0.70888352394104, "learning_rate": 0.0005, "loss": 1.6635595560073853, "step": 12590 }, { "epoch": 0.9925896728419393, "grad_norm": 0.5812563896179199, "learning_rate": 0.0005, "loss": 1.548755407333374, "step": 12591 }, { "epoch": 0.9926685061095782, "grad_norm": 0.8973672389984131, "learning_rate": 0.0005, "loss": 2.089284896850586, "step": 12592 }, { "epoch": 0.9927473393772172, "grad_norm": 0.5014591813087463, "learning_rate": 0.0005, "loss": 1.5220377445220947, "step": 12593 }, { "epoch": 0.9928261726448562, "grad_norm": 0.7155680656433105, "learning_rate": 0.0005, "loss": 1.9629788398742676, "step": 12594 }, { "epoch": 0.9929050059124951, "grad_norm": 0.6069255471229553, "learning_rate": 0.0005, "loss": 1.7305320501327515, "step": 12595 }, { "epoch": 0.992983839180134, "grad_norm": 0.5685803294181824, "learning_rate": 0.0005, "loss": 1.6637990474700928, "step": 12596 }, { "epoch": 0.9930626724477729, "grad_norm": 0.5595672726631165, "learning_rate": 0.0005, "loss": 1.7101621627807617, "step": 12597 }, { "epoch": 0.9931415057154119, "grad_norm": 0.5948522090911865, "learning_rate": 0.0005, "loss": 1.7221157550811768, "step": 12598 }, { "epoch": 0.9932203389830508, "grad_norm": 0.5693272352218628, "learning_rate": 0.0005, "loss": 1.6613624095916748, "step": 12599 }, { "epoch": 0.9932991722506898, "grad_norm": 0.7192005515098572, "learning_rate": 0.0005, "loss": 1.8349177837371826, "step": 12600 }, { "epoch": 0.9933780055183288, "grad_norm": 0.582300066947937, "learning_rate": 0.0005, "loss": 1.620833158493042, "step": 12601 }, { "epoch": 0.9934568387859677, "grad_norm": 0.7732320427894592, "learning_rate": 0.0005, "loss": 1.8184641599655151, "step": 12602 }, { "epoch": 0.9935356720536066, "grad_norm": 0.5951229333877563, "learning_rate": 0.0005, "loss": 1.8612563610076904, "step": 12603 }, { "epoch": 0.9936145053212455, "grad_norm": 0.6792368292808533, "learning_rate": 0.0005, "loss": 1.711949348449707, "step": 12604 }, { "epoch": 0.9936933385888845, "grad_norm": 0.703927218914032, "learning_rate": 0.0005, "loss": 1.8171672821044922, "step": 12605 }, { "epoch": 0.9937721718565234, "grad_norm": 0.6624727845191956, "learning_rate": 0.0005, "loss": 1.9167003631591797, "step": 12606 }, { "epoch": 0.9938510051241624, "grad_norm": 0.5817882418632507, "learning_rate": 0.0005, "loss": 1.8364394903182983, "step": 12607 }, { "epoch": 0.9939298383918014, "grad_norm": 12.671679496765137, "learning_rate": 0.0005, "loss": 2.0326340198516846, "step": 12608 }, { "epoch": 0.9940086716594403, "grad_norm": 0.6628590822219849, "learning_rate": 0.0005, "loss": 1.6795985698699951, "step": 12609 }, { "epoch": 0.9940875049270792, "grad_norm": 0.5829019546508789, "learning_rate": 0.0005, "loss": 1.8553982973098755, "step": 12610 }, { "epoch": 0.9941663381947182, "grad_norm": 0.9221625924110413, "learning_rate": 0.0005, "loss": 2.423628091812134, "step": 12611 }, { "epoch": 0.9942451714623571, "grad_norm": 0.5315333604812622, "learning_rate": 0.0005, "loss": 1.4706611633300781, "step": 12612 }, { "epoch": 0.994324004729996, "grad_norm": 0.6668756008148193, "learning_rate": 0.0005, "loss": 1.7512128353118896, "step": 12613 }, { "epoch": 0.994402837997635, "grad_norm": 0.598342776298523, "learning_rate": 0.0005, "loss": 1.759777545928955, "step": 12614 }, { "epoch": 0.994481671265274, "grad_norm": 0.6139475703239441, "learning_rate": 0.0005, "loss": 1.8709769248962402, "step": 12615 }, { "epoch": 0.9945605045329129, "grad_norm": 0.6865462064743042, "learning_rate": 0.0005, "loss": 1.7774147987365723, "step": 12616 }, { "epoch": 0.9946393378005518, "grad_norm": 0.635132908821106, "learning_rate": 0.0005, "loss": 1.985363245010376, "step": 12617 }, { "epoch": 0.9947181710681908, "grad_norm": 0.5916011333465576, "learning_rate": 0.0005, "loss": 1.642242431640625, "step": 12618 }, { "epoch": 0.9947970043358297, "grad_norm": 0.44606325030326843, "learning_rate": 0.0005, "loss": 1.68644380569458, "step": 12619 }, { "epoch": 0.9948758376034686, "grad_norm": 0.6302103996276855, "learning_rate": 0.0005, "loss": 1.7112088203430176, "step": 12620 }, { "epoch": 0.9949546708711076, "grad_norm": 0.6002242565155029, "learning_rate": 0.0005, "loss": 1.7891957759857178, "step": 12621 }, { "epoch": 0.9950335041387466, "grad_norm": 0.6393771767616272, "learning_rate": 0.0005, "loss": 1.6047370433807373, "step": 12622 }, { "epoch": 0.9951123374063855, "grad_norm": 0.6556588411331177, "learning_rate": 0.0005, "loss": 1.846303105354309, "step": 12623 }, { "epoch": 0.9951911706740244, "grad_norm": 0.6805118918418884, "learning_rate": 0.0005, "loss": 1.91605806350708, "step": 12624 }, { "epoch": 0.9952700039416634, "grad_norm": 1.3315900564193726, "learning_rate": 0.0005, "loss": 1.8954524993896484, "step": 12625 }, { "epoch": 0.9953488372093023, "grad_norm": 0.6283223032951355, "learning_rate": 0.0005, "loss": 1.576022744178772, "step": 12626 }, { "epoch": 0.9954276704769413, "grad_norm": 0.553683340549469, "learning_rate": 0.0005, "loss": 1.9000805616378784, "step": 12627 }, { "epoch": 0.9955065037445802, "grad_norm": 0.741512656211853, "learning_rate": 0.0005, "loss": 1.9799381494522095, "step": 12628 }, { "epoch": 0.9955853370122192, "grad_norm": 0.5301133990287781, "learning_rate": 0.0005, "loss": 1.6899007558822632, "step": 12629 }, { "epoch": 0.9956641702798581, "grad_norm": 0.5061066150665283, "learning_rate": 0.0005, "loss": 1.7482244968414307, "step": 12630 }, { "epoch": 0.995743003547497, "grad_norm": 0.578086256980896, "learning_rate": 0.0005, "loss": 1.701045274734497, "step": 12631 }, { "epoch": 0.995821836815136, "grad_norm": 0.6761948466300964, "learning_rate": 0.0005, "loss": 1.928593635559082, "step": 12632 }, { "epoch": 0.9959006700827749, "grad_norm": 0.6039466261863708, "learning_rate": 0.0005, "loss": 1.98073148727417, "step": 12633 }, { "epoch": 0.9959795033504139, "grad_norm": 0.5278012156486511, "learning_rate": 0.0005, "loss": 1.6428860425949097, "step": 12634 }, { "epoch": 0.9960583366180528, "grad_norm": 0.5758330821990967, "learning_rate": 0.0005, "loss": 1.85360586643219, "step": 12635 }, { "epoch": 0.9961371698856918, "grad_norm": 0.6939254999160767, "learning_rate": 0.0005, "loss": 1.825273036956787, "step": 12636 }, { "epoch": 0.9962160031533307, "grad_norm": 0.634772002696991, "learning_rate": 0.0005, "loss": 1.977358341217041, "step": 12637 }, { "epoch": 0.9962948364209696, "grad_norm": 0.6377807855606079, "learning_rate": 0.0005, "loss": 2.031996726989746, "step": 12638 }, { "epoch": 0.9963736696886086, "grad_norm": 0.4905216693878174, "learning_rate": 0.0005, "loss": 1.4708826541900635, "step": 12639 }, { "epoch": 0.9964525029562475, "grad_norm": 0.6020634770393372, "learning_rate": 0.0005, "loss": 1.6490620374679565, "step": 12640 }, { "epoch": 0.9965313362238865, "grad_norm": 1.0073745250701904, "learning_rate": 0.0005, "loss": 1.8811993598937988, "step": 12641 }, { "epoch": 0.9966101694915255, "grad_norm": 0.5462333559989929, "learning_rate": 0.0005, "loss": 1.8427238464355469, "step": 12642 }, { "epoch": 0.9966890027591644, "grad_norm": 0.6023728251457214, "learning_rate": 0.0005, "loss": 1.706129789352417, "step": 12643 }, { "epoch": 0.9967678360268033, "grad_norm": 0.6006491780281067, "learning_rate": 0.0005, "loss": 1.8308260440826416, "step": 12644 }, { "epoch": 0.9968466692944422, "grad_norm": 0.6424300670623779, "learning_rate": 0.0005, "loss": 1.717355489730835, "step": 12645 }, { "epoch": 0.9969255025620812, "grad_norm": 0.5046111345291138, "learning_rate": 0.0005, "loss": 1.699350118637085, "step": 12646 }, { "epoch": 0.9970043358297201, "grad_norm": 0.7047225832939148, "learning_rate": 0.0005, "loss": 1.7961149215698242, "step": 12647 }, { "epoch": 0.9970831690973591, "grad_norm": 0.7774263024330139, "learning_rate": 0.0005, "loss": 1.9600634574890137, "step": 12648 }, { "epoch": 0.9971620023649981, "grad_norm": 0.5682125091552734, "learning_rate": 0.0005, "loss": 1.7084205150604248, "step": 12649 }, { "epoch": 0.997240835632637, "grad_norm": 0.6149210929870605, "learning_rate": 0.0005, "loss": 1.8789387941360474, "step": 12650 }, { "epoch": 0.9973196689002759, "grad_norm": 0.8005303740501404, "learning_rate": 0.0005, "loss": 1.9696071147918701, "step": 12651 }, { "epoch": 0.9973985021679148, "grad_norm": 0.6763851046562195, "learning_rate": 0.0005, "loss": 1.7521324157714844, "step": 12652 }, { "epoch": 0.9974773354355538, "grad_norm": 0.5283495783805847, "learning_rate": 0.0005, "loss": 1.5603410005569458, "step": 12653 }, { "epoch": 0.9975561687031927, "grad_norm": 0.7012802958488464, "learning_rate": 0.0005, "loss": 1.9438960552215576, "step": 12654 }, { "epoch": 0.9976350019708317, "grad_norm": 0.7534082531929016, "learning_rate": 0.0005, "loss": 1.7372181415557861, "step": 12655 }, { "epoch": 0.9977138352384707, "grad_norm": 0.7764307856559753, "learning_rate": 0.0005, "loss": 1.8954582214355469, "step": 12656 }, { "epoch": 0.9977926685061096, "grad_norm": 0.7176365852355957, "learning_rate": 0.0005, "loss": 1.9408960342407227, "step": 12657 }, { "epoch": 0.9978715017737485, "grad_norm": 0.6599109768867493, "learning_rate": 0.0005, "loss": 1.8414418697357178, "step": 12658 }, { "epoch": 0.9979503350413874, "grad_norm": 0.7728519439697266, "learning_rate": 0.0005, "loss": 1.8942158222198486, "step": 12659 }, { "epoch": 0.9980291683090264, "grad_norm": 0.7326244711875916, "learning_rate": 0.0005, "loss": 1.7460476160049438, "step": 12660 }, { "epoch": 0.9981080015766653, "grad_norm": 0.56809002161026, "learning_rate": 0.0005, "loss": 1.7794848680496216, "step": 12661 }, { "epoch": 0.9981868348443043, "grad_norm": 0.7895651459693909, "learning_rate": 0.0005, "loss": 1.7668631076812744, "step": 12662 }, { "epoch": 0.9982656681119433, "grad_norm": 0.7021631002426147, "learning_rate": 0.0005, "loss": 1.7928295135498047, "step": 12663 }, { "epoch": 0.9983445013795822, "grad_norm": 0.5203651785850525, "learning_rate": 0.0005, "loss": 1.8569045066833496, "step": 12664 }, { "epoch": 0.9984233346472211, "grad_norm": 0.623939037322998, "learning_rate": 0.0005, "loss": 1.6451354026794434, "step": 12665 }, { "epoch": 0.99850216791486, "grad_norm": 0.4341806173324585, "learning_rate": 0.0005, "loss": 1.696606159210205, "step": 12666 }, { "epoch": 0.998581001182499, "grad_norm": 0.6796512603759766, "learning_rate": 0.0005, "loss": 1.9478778839111328, "step": 12667 }, { "epoch": 0.998659834450138, "grad_norm": 0.7157154083251953, "learning_rate": 0.0005, "loss": 1.7061891555786133, "step": 12668 }, { "epoch": 0.9987386677177769, "grad_norm": 1.2700780630111694, "learning_rate": 0.0005, "loss": 2.229727268218994, "step": 12669 }, { "epoch": 0.9988175009854159, "grad_norm": 0.7543469071388245, "learning_rate": 0.0005, "loss": 1.976012945175171, "step": 12670 }, { "epoch": 0.9988963342530548, "grad_norm": 0.638501763343811, "learning_rate": 0.0005, "loss": 1.707757830619812, "step": 12671 }, { "epoch": 0.9989751675206937, "grad_norm": 0.9731208682060242, "learning_rate": 0.0005, "loss": 1.812985897064209, "step": 12672 }, { "epoch": 0.9990540007883327, "grad_norm": 0.6701606512069702, "learning_rate": 0.0005, "loss": 1.5293262004852295, "step": 12673 }, { "epoch": 0.9991328340559716, "grad_norm": 0.8771061301231384, "learning_rate": 0.0005, "loss": 1.7404425144195557, "step": 12674 }, { "epoch": 0.9992116673236106, "grad_norm": 0.6326627731323242, "learning_rate": 0.0005, "loss": 1.92451012134552, "step": 12675 }, { "epoch": 0.9992905005912495, "grad_norm": 3.907426118850708, "learning_rate": 0.0005, "loss": 1.9644246101379395, "step": 12676 }, { "epoch": 0.9993693338588885, "grad_norm": 0.5835637450218201, "learning_rate": 0.0005, "loss": 1.5266361236572266, "step": 12677 }, { "epoch": 0.9994481671265274, "grad_norm": 0.6322555541992188, "learning_rate": 0.0005, "loss": 1.8334741592407227, "step": 12678 }, { "epoch": 0.9995270003941663, "grad_norm": 0.5761070251464844, "learning_rate": 0.0005, "loss": 1.730844259262085, "step": 12679 }, { "epoch": 0.9996058336618053, "grad_norm": 0.5789262056350708, "learning_rate": 0.0005, "loss": 1.5340659618377686, "step": 12680 }, { "epoch": 0.9996846669294442, "grad_norm": 0.5844905376434326, "learning_rate": 0.0005, "loss": 1.6845386028289795, "step": 12681 }, { "epoch": 0.9997635001970832, "grad_norm": 0.5939300656318665, "learning_rate": 0.0005, "loss": 1.7379059791564941, "step": 12682 }, { "epoch": 0.9998423334647221, "grad_norm": 0.7240161895751953, "learning_rate": 0.0005, "loss": 1.7474159002304077, "step": 12683 }, { "epoch": 0.9999211667323611, "grad_norm": 0.7295636534690857, "learning_rate": 0.0005, "loss": 1.8308302164077759, "step": 12684 }, { "epoch": 1.0, "grad_norm": 0.6492947936058044, "learning_rate": 0.0005, "loss": 1.6515228748321533, "step": 12685 }, { "epoch": 1.000078833267639, "grad_norm": 0.7117326855659485, "learning_rate": 0.0005, "loss": 1.8795130252838135, "step": 12686 }, { "epoch": 1.0001576665352778, "grad_norm": 0.581716775894165, "learning_rate": 0.0005, "loss": 1.584263801574707, "step": 12687 }, { "epoch": 1.0002364998029167, "grad_norm": 0.5734694600105286, "learning_rate": 0.0005, "loss": 1.9581117630004883, "step": 12688 }, { "epoch": 1.0003153330705559, "grad_norm": 0.5830156207084656, "learning_rate": 0.0005, "loss": 1.9562597274780273, "step": 12689 }, { "epoch": 1.0003941663381948, "grad_norm": 0.6658135056495667, "learning_rate": 0.0005, "loss": 1.7992362976074219, "step": 12690 }, { "epoch": 1.0004729996058337, "grad_norm": 0.589397132396698, "learning_rate": 0.0005, "loss": 1.8647538423538208, "step": 12691 }, { "epoch": 1.0005518328734726, "grad_norm": 0.521004319190979, "learning_rate": 0.0005, "loss": 1.7530707120895386, "step": 12692 }, { "epoch": 1.0006306661411115, "grad_norm": 0.6848132610321045, "learning_rate": 0.0005, "loss": 1.5264705419540405, "step": 12693 }, { "epoch": 1.0007094994087504, "grad_norm": 0.5317161679267883, "learning_rate": 0.0005, "loss": 1.6576476097106934, "step": 12694 }, { "epoch": 1.0007883326763893, "grad_norm": 0.5356929898262024, "learning_rate": 0.0005, "loss": 1.6232527494430542, "step": 12695 }, { "epoch": 1.0008671659440285, "grad_norm": 0.8294675350189209, "learning_rate": 0.0005, "loss": 1.6905596256256104, "step": 12696 }, { "epoch": 1.0009459992116674, "grad_norm": 0.5167319178581238, "learning_rate": 0.0005, "loss": 1.6725554466247559, "step": 12697 }, { "epoch": 1.0010248324793063, "grad_norm": 0.48319628834724426, "learning_rate": 0.0005, "loss": 1.5296528339385986, "step": 12698 }, { "epoch": 1.0011036657469452, "grad_norm": 0.7130226492881775, "learning_rate": 0.0005, "loss": 1.8057920932769775, "step": 12699 }, { "epoch": 1.0011824990145841, "grad_norm": 0.7876770496368408, "learning_rate": 0.0005, "loss": 2.010535478591919, "step": 12700 }, { "epoch": 1.001261332282223, "grad_norm": 0.5317705273628235, "learning_rate": 0.0005, "loss": 1.6268494129180908, "step": 12701 }, { "epoch": 1.001340165549862, "grad_norm": 2.4501867294311523, "learning_rate": 0.0005, "loss": 1.6354562044143677, "step": 12702 }, { "epoch": 1.001418998817501, "grad_norm": 0.5232968926429749, "learning_rate": 0.0005, "loss": 1.5482282638549805, "step": 12703 }, { "epoch": 1.00149783208514, "grad_norm": 0.5245763063430786, "learning_rate": 0.0005, "loss": 1.7128114700317383, "step": 12704 }, { "epoch": 1.001576665352779, "grad_norm": 0.47225311398506165, "learning_rate": 0.0005, "loss": 1.778111457824707, "step": 12705 }, { "epoch": 1.0016554986204178, "grad_norm": 0.7087892889976501, "learning_rate": 0.0005, "loss": 1.8148188591003418, "step": 12706 }, { "epoch": 1.0017343318880567, "grad_norm": 0.6507248282432556, "learning_rate": 0.0005, "loss": 1.7710416316986084, "step": 12707 }, { "epoch": 1.0018131651556956, "grad_norm": 0.5106654763221741, "learning_rate": 0.0005, "loss": 1.649633526802063, "step": 12708 }, { "epoch": 1.0018919984233345, "grad_norm": 0.5700387954711914, "learning_rate": 0.0005, "loss": 1.6617448329925537, "step": 12709 }, { "epoch": 1.0019708316909737, "grad_norm": 0.4801177382469177, "learning_rate": 0.0005, "loss": 1.4858543872833252, "step": 12710 }, { "epoch": 1.0020496649586126, "grad_norm": 0.5151267647743225, "learning_rate": 0.0005, "loss": 1.688516616821289, "step": 12711 }, { "epoch": 1.0021284982262515, "grad_norm": 5.02023983001709, "learning_rate": 0.0005, "loss": 1.870823621749878, "step": 12712 }, { "epoch": 1.0022073314938904, "grad_norm": 0.525831937789917, "learning_rate": 0.0005, "loss": 1.645626425743103, "step": 12713 }, { "epoch": 1.0022861647615293, "grad_norm": 0.6620439291000366, "learning_rate": 0.0005, "loss": 1.8005871772766113, "step": 12714 }, { "epoch": 1.0023649980291682, "grad_norm": 0.5697690844535828, "learning_rate": 0.0005, "loss": 1.8668380975723267, "step": 12715 }, { "epoch": 1.0024438312968074, "grad_norm": 0.521748960018158, "learning_rate": 0.0005, "loss": 1.78280770778656, "step": 12716 }, { "epoch": 1.0025226645644463, "grad_norm": 0.7308337092399597, "learning_rate": 0.0005, "loss": 1.9217853546142578, "step": 12717 }, { "epoch": 1.0026014978320852, "grad_norm": 0.6828036904335022, "learning_rate": 0.0005, "loss": 1.837904930114746, "step": 12718 }, { "epoch": 1.002680331099724, "grad_norm": 0.614086389541626, "learning_rate": 0.0005, "loss": 1.8915939331054688, "step": 12719 }, { "epoch": 1.002759164367363, "grad_norm": 0.4962795674800873, "learning_rate": 0.0005, "loss": 1.395573377609253, "step": 12720 }, { "epoch": 1.002837997635002, "grad_norm": 0.3990088403224945, "learning_rate": 0.0005, "loss": 1.420436143875122, "step": 12721 }, { "epoch": 1.0029168309026408, "grad_norm": 0.46272292733192444, "learning_rate": 0.0005, "loss": 1.79116952419281, "step": 12722 }, { "epoch": 1.00299566417028, "grad_norm": 0.5623204708099365, "learning_rate": 0.0005, "loss": 1.7912564277648926, "step": 12723 }, { "epoch": 1.0030744974379189, "grad_norm": 0.5429496765136719, "learning_rate": 0.0005, "loss": 1.749731421470642, "step": 12724 }, { "epoch": 1.0031533307055578, "grad_norm": 0.7000612020492554, "learning_rate": 0.0005, "loss": 1.8068534135818481, "step": 12725 }, { "epoch": 1.0032321639731967, "grad_norm": 0.7441208958625793, "learning_rate": 0.0005, "loss": 2.0215461254119873, "step": 12726 }, { "epoch": 1.0033109972408356, "grad_norm": 0.4929634928703308, "learning_rate": 0.0005, "loss": 1.8633726835250854, "step": 12727 }, { "epoch": 1.0033898305084745, "grad_norm": 0.4284701943397522, "learning_rate": 0.0005, "loss": 1.5105313062667847, "step": 12728 }, { "epoch": 1.0034686637761134, "grad_norm": 0.48766273260116577, "learning_rate": 0.0005, "loss": 1.5990979671478271, "step": 12729 }, { "epoch": 1.0035474970437526, "grad_norm": 0.5961654186248779, "learning_rate": 0.0005, "loss": 1.8294113874435425, "step": 12730 }, { "epoch": 1.0036263303113915, "grad_norm": 0.8238674402236938, "learning_rate": 0.0005, "loss": 1.6640641689300537, "step": 12731 }, { "epoch": 1.0037051635790304, "grad_norm": 0.5020102262496948, "learning_rate": 0.0005, "loss": 1.6172840595245361, "step": 12732 }, { "epoch": 1.0037839968466693, "grad_norm": 0.5319114327430725, "learning_rate": 0.0005, "loss": 1.4670324325561523, "step": 12733 }, { "epoch": 1.0038628301143082, "grad_norm": 0.6663116812705994, "learning_rate": 0.0005, "loss": 1.6262454986572266, "step": 12734 }, { "epoch": 1.0039416633819471, "grad_norm": 0.6095593571662903, "learning_rate": 0.0005, "loss": 1.792175531387329, "step": 12735 }, { "epoch": 1.004020496649586, "grad_norm": 0.5647503733634949, "learning_rate": 0.0005, "loss": 1.896082878112793, "step": 12736 }, { "epoch": 1.0040993299172252, "grad_norm": 0.5050663352012634, "learning_rate": 0.0005, "loss": 1.7930657863616943, "step": 12737 }, { "epoch": 1.004178163184864, "grad_norm": 0.5750671625137329, "learning_rate": 0.0005, "loss": 1.6580970287322998, "step": 12738 }, { "epoch": 1.004256996452503, "grad_norm": 0.5867236256599426, "learning_rate": 0.0005, "loss": 1.7725894451141357, "step": 12739 }, { "epoch": 1.004335829720142, "grad_norm": 0.5117447972297668, "learning_rate": 0.0005, "loss": 1.7121795415878296, "step": 12740 }, { "epoch": 1.0044146629877808, "grad_norm": 0.6401829719543457, "learning_rate": 0.0005, "loss": 1.592557668685913, "step": 12741 }, { "epoch": 1.0044934962554197, "grad_norm": 0.6354086995124817, "learning_rate": 0.0005, "loss": 1.716160774230957, "step": 12742 }, { "epoch": 1.0045723295230586, "grad_norm": 0.5421633720397949, "learning_rate": 0.0005, "loss": 1.7646255493164062, "step": 12743 }, { "epoch": 1.0046511627906978, "grad_norm": 0.49507632851600647, "learning_rate": 0.0005, "loss": 1.712310791015625, "step": 12744 }, { "epoch": 1.0047299960583367, "grad_norm": 0.529744029045105, "learning_rate": 0.0005, "loss": 1.5798733234405518, "step": 12745 }, { "epoch": 1.0048088293259756, "grad_norm": 0.5807482600212097, "learning_rate": 0.0005, "loss": 1.7663052082061768, "step": 12746 }, { "epoch": 1.0048876625936145, "grad_norm": 0.6184395551681519, "learning_rate": 0.0005, "loss": 1.605460286140442, "step": 12747 }, { "epoch": 1.0049664958612534, "grad_norm": 0.832381546497345, "learning_rate": 0.0005, "loss": 1.669700026512146, "step": 12748 }, { "epoch": 1.0050453291288923, "grad_norm": 0.6629236340522766, "learning_rate": 0.0005, "loss": 1.9032366275787354, "step": 12749 }, { "epoch": 1.0051241623965312, "grad_norm": 0.6285645365715027, "learning_rate": 0.0005, "loss": 1.8306901454925537, "step": 12750 }, { "epoch": 1.0052029956641704, "grad_norm": 0.5443150401115417, "learning_rate": 0.0005, "loss": 1.982412338256836, "step": 12751 }, { "epoch": 1.0052818289318093, "grad_norm": 0.5396905541419983, "learning_rate": 0.0005, "loss": 1.7449474334716797, "step": 12752 }, { "epoch": 1.0053606621994482, "grad_norm": 0.660599946975708, "learning_rate": 0.0005, "loss": 1.938918948173523, "step": 12753 }, { "epoch": 1.005439495467087, "grad_norm": 0.6328840851783752, "learning_rate": 0.0005, "loss": 1.607088327407837, "step": 12754 }, { "epoch": 1.005518328734726, "grad_norm": 0.4818970859050751, "learning_rate": 0.0005, "loss": 1.7458209991455078, "step": 12755 }, { "epoch": 1.005597162002365, "grad_norm": 0.5148895978927612, "learning_rate": 0.0005, "loss": 1.5525057315826416, "step": 12756 }, { "epoch": 1.0056759952700038, "grad_norm": 0.7319002747535706, "learning_rate": 0.0005, "loss": 1.548279047012329, "step": 12757 }, { "epoch": 1.005754828537643, "grad_norm": 0.5830951929092407, "learning_rate": 0.0005, "loss": 1.709469199180603, "step": 12758 }, { "epoch": 1.0058336618052819, "grad_norm": 0.5325725674629211, "learning_rate": 0.0005, "loss": 1.6277847290039062, "step": 12759 }, { "epoch": 1.0059124950729208, "grad_norm": 0.5162160396575928, "learning_rate": 0.0005, "loss": 1.5874481201171875, "step": 12760 }, { "epoch": 1.0059913283405597, "grad_norm": 0.6565716862678528, "learning_rate": 0.0005, "loss": 1.769312858581543, "step": 12761 }, { "epoch": 1.0060701616081986, "grad_norm": 0.33802253007888794, "learning_rate": 0.0005, "loss": 1.2660844326019287, "step": 12762 }, { "epoch": 1.0061489948758375, "grad_norm": 0.5419055819511414, "learning_rate": 0.0005, "loss": 1.6427857875823975, "step": 12763 }, { "epoch": 1.0062278281434764, "grad_norm": 0.4893801510334015, "learning_rate": 0.0005, "loss": 1.6437766551971436, "step": 12764 }, { "epoch": 1.0063066614111156, "grad_norm": 0.7767627239227295, "learning_rate": 0.0005, "loss": 1.662085771560669, "step": 12765 }, { "epoch": 1.0063854946787545, "grad_norm": 0.48976898193359375, "learning_rate": 0.0005, "loss": 1.6925172805786133, "step": 12766 }, { "epoch": 1.0064643279463934, "grad_norm": 0.524459719657898, "learning_rate": 0.0005, "loss": 1.6470736265182495, "step": 12767 }, { "epoch": 1.0065431612140323, "grad_norm": 0.8486864566802979, "learning_rate": 0.0005, "loss": 1.690453290939331, "step": 12768 }, { "epoch": 1.0066219944816712, "grad_norm": 0.7153468132019043, "learning_rate": 0.0005, "loss": 1.7183282375335693, "step": 12769 }, { "epoch": 1.0067008277493101, "grad_norm": 0.49503403902053833, "learning_rate": 0.0005, "loss": 1.5116804838180542, "step": 12770 }, { "epoch": 1.006779661016949, "grad_norm": 0.5463730692863464, "learning_rate": 0.0005, "loss": 2.0208871364593506, "step": 12771 }, { "epoch": 1.0068584942845882, "grad_norm": 0.5468184947967529, "learning_rate": 0.0005, "loss": 1.686381220817566, "step": 12772 }, { "epoch": 1.006937327552227, "grad_norm": 0.6452279090881348, "learning_rate": 0.0005, "loss": 1.7833452224731445, "step": 12773 }, { "epoch": 1.007016160819866, "grad_norm": 0.5338138937950134, "learning_rate": 0.0005, "loss": 1.618098258972168, "step": 12774 }, { "epoch": 1.007094994087505, "grad_norm": 0.4870446026325226, "learning_rate": 0.0005, "loss": 1.6028413772583008, "step": 12775 }, { "epoch": 1.0071738273551438, "grad_norm": 0.5407032370567322, "learning_rate": 0.0005, "loss": 1.6669974327087402, "step": 12776 }, { "epoch": 1.0072526606227827, "grad_norm": 0.4277613162994385, "learning_rate": 0.0005, "loss": 1.5304450988769531, "step": 12777 }, { "epoch": 1.0073314938904216, "grad_norm": 0.5760273933410645, "learning_rate": 0.0005, "loss": 1.5315219163894653, "step": 12778 }, { "epoch": 1.0074103271580608, "grad_norm": 0.6180756092071533, "learning_rate": 0.0005, "loss": 1.8959720134735107, "step": 12779 }, { "epoch": 1.0074891604256997, "grad_norm": 0.5279597043991089, "learning_rate": 0.0005, "loss": 1.5853767395019531, "step": 12780 }, { "epoch": 1.0075679936933386, "grad_norm": 0.6110203266143799, "learning_rate": 0.0005, "loss": 1.7320668697357178, "step": 12781 }, { "epoch": 1.0076468269609775, "grad_norm": 0.532432496547699, "learning_rate": 0.0005, "loss": 1.7213618755340576, "step": 12782 }, { "epoch": 1.0077256602286164, "grad_norm": 0.4764426648616791, "learning_rate": 0.0005, "loss": 1.3014949560165405, "step": 12783 }, { "epoch": 1.0078044934962553, "grad_norm": 0.6459869742393494, "learning_rate": 0.0005, "loss": 1.798537015914917, "step": 12784 }, { "epoch": 1.0078833267638945, "grad_norm": 0.6324913501739502, "learning_rate": 0.0005, "loss": 1.8325693607330322, "step": 12785 }, { "epoch": 1.0079621600315334, "grad_norm": 0.5586258172988892, "learning_rate": 0.0005, "loss": 1.766862154006958, "step": 12786 }, { "epoch": 1.0080409932991723, "grad_norm": 0.5211542844772339, "learning_rate": 0.0005, "loss": 1.7386925220489502, "step": 12787 }, { "epoch": 1.0081198265668112, "grad_norm": 0.591874361038208, "learning_rate": 0.0005, "loss": 1.7440345287322998, "step": 12788 }, { "epoch": 1.0081986598344501, "grad_norm": 0.5344576835632324, "learning_rate": 0.0005, "loss": 1.5252615213394165, "step": 12789 }, { "epoch": 1.008277493102089, "grad_norm": 0.5323571562767029, "learning_rate": 0.0005, "loss": 1.7537637948989868, "step": 12790 }, { "epoch": 1.008356326369728, "grad_norm": 0.48486030101776123, "learning_rate": 0.0005, "loss": 1.447812795639038, "step": 12791 }, { "epoch": 1.008435159637367, "grad_norm": 0.6021968126296997, "learning_rate": 0.0005, "loss": 1.5603176355361938, "step": 12792 }, { "epoch": 1.008513992905006, "grad_norm": 0.4813899099826813, "learning_rate": 0.0005, "loss": 1.4532313346862793, "step": 12793 }, { "epoch": 1.008592826172645, "grad_norm": 0.6198461055755615, "learning_rate": 0.0005, "loss": 1.9747536182403564, "step": 12794 }, { "epoch": 1.0086716594402838, "grad_norm": 0.49826133251190186, "learning_rate": 0.0005, "loss": 1.58830988407135, "step": 12795 }, { "epoch": 1.0087504927079227, "grad_norm": 0.4389275312423706, "learning_rate": 0.0005, "loss": 1.6491925716400146, "step": 12796 }, { "epoch": 1.0088293259755616, "grad_norm": 0.552125871181488, "learning_rate": 0.0005, "loss": 1.7093031406402588, "step": 12797 }, { "epoch": 1.0089081592432005, "grad_norm": 0.506664514541626, "learning_rate": 0.0005, "loss": 1.9186186790466309, "step": 12798 }, { "epoch": 1.0089869925108397, "grad_norm": 0.4580901265144348, "learning_rate": 0.0005, "loss": 1.7166759967803955, "step": 12799 }, { "epoch": 1.0090658257784786, "grad_norm": 0.5867875814437866, "learning_rate": 0.0005, "loss": 1.5977880954742432, "step": 12800 }, { "epoch": 1.0091446590461175, "grad_norm": 0.5599817633628845, "learning_rate": 0.0005, "loss": 1.7626149654388428, "step": 12801 }, { "epoch": 1.0092234923137564, "grad_norm": 0.6206426024436951, "learning_rate": 0.0005, "loss": 1.5537281036376953, "step": 12802 }, { "epoch": 1.0093023255813953, "grad_norm": 0.596619725227356, "learning_rate": 0.0005, "loss": 1.8089148998260498, "step": 12803 }, { "epoch": 1.0093811588490342, "grad_norm": 0.5004245042800903, "learning_rate": 0.0005, "loss": 1.6295456886291504, "step": 12804 }, { "epoch": 1.0094599921166731, "grad_norm": 0.6295502781867981, "learning_rate": 0.0005, "loss": 1.7803051471710205, "step": 12805 }, { "epoch": 1.0095388253843123, "grad_norm": 0.5425712466239929, "learning_rate": 0.0005, "loss": 1.761500597000122, "step": 12806 }, { "epoch": 1.0096176586519512, "grad_norm": 0.4994536340236664, "learning_rate": 0.0005, "loss": 1.5712456703186035, "step": 12807 }, { "epoch": 1.00969649191959, "grad_norm": 0.5376179814338684, "learning_rate": 0.0005, "loss": 1.777616024017334, "step": 12808 }, { "epoch": 1.009775325187229, "grad_norm": 0.5237816572189331, "learning_rate": 0.0005, "loss": 1.5366511344909668, "step": 12809 }, { "epoch": 1.009854158454868, "grad_norm": 0.7202628254890442, "learning_rate": 0.0005, "loss": 1.574802041053772, "step": 12810 }, { "epoch": 1.0099329917225068, "grad_norm": 0.49988889694213867, "learning_rate": 0.0005, "loss": 1.6462215185165405, "step": 12811 }, { "epoch": 1.0100118249901457, "grad_norm": 0.4869120717048645, "learning_rate": 0.0005, "loss": 1.6439197063446045, "step": 12812 }, { "epoch": 1.0100906582577849, "grad_norm": 0.6321154236793518, "learning_rate": 0.0005, "loss": 1.7142624855041504, "step": 12813 }, { "epoch": 1.0101694915254238, "grad_norm": 0.5459350943565369, "learning_rate": 0.0005, "loss": 1.6853677034378052, "step": 12814 }, { "epoch": 1.0102483247930627, "grad_norm": 0.5793962478637695, "learning_rate": 0.0005, "loss": 1.6093366146087646, "step": 12815 }, { "epoch": 1.0103271580607016, "grad_norm": 0.6002652645111084, "learning_rate": 0.0005, "loss": 1.5141420364379883, "step": 12816 }, { "epoch": 1.0104059913283405, "grad_norm": 0.5898426175117493, "learning_rate": 0.0005, "loss": 1.739447832107544, "step": 12817 }, { "epoch": 1.0104848245959794, "grad_norm": 0.5814378261566162, "learning_rate": 0.0005, "loss": 1.641129493713379, "step": 12818 }, { "epoch": 1.0105636578636183, "grad_norm": 0.563989520072937, "learning_rate": 0.0005, "loss": 1.5502009391784668, "step": 12819 }, { "epoch": 1.0106424911312575, "grad_norm": 0.5236315727233887, "learning_rate": 0.0005, "loss": 1.5812960863113403, "step": 12820 }, { "epoch": 1.0107213243988964, "grad_norm": 0.5186370015144348, "learning_rate": 0.0005, "loss": 1.6209850311279297, "step": 12821 }, { "epoch": 1.0108001576665353, "grad_norm": 0.6953315138816833, "learning_rate": 0.0005, "loss": 1.5197724103927612, "step": 12822 }, { "epoch": 1.0108789909341742, "grad_norm": 0.874299943447113, "learning_rate": 0.0005, "loss": 1.7042371034622192, "step": 12823 }, { "epoch": 1.0109578242018131, "grad_norm": 0.8014461994171143, "learning_rate": 0.0005, "loss": 1.8619358539581299, "step": 12824 }, { "epoch": 1.011036657469452, "grad_norm": 0.4364451467990875, "learning_rate": 0.0005, "loss": 1.5675292015075684, "step": 12825 }, { "epoch": 1.011115490737091, "grad_norm": 0.4695291519165039, "learning_rate": 0.0005, "loss": 1.5621435642242432, "step": 12826 }, { "epoch": 1.01119432400473, "grad_norm": 0.48663055896759033, "learning_rate": 0.0005, "loss": 1.5882792472839355, "step": 12827 }, { "epoch": 1.011273157272369, "grad_norm": 0.6377779841423035, "learning_rate": 0.0005, "loss": 1.790574550628662, "step": 12828 }, { "epoch": 1.011351990540008, "grad_norm": 0.5968761444091797, "learning_rate": 0.0005, "loss": 1.7493704557418823, "step": 12829 }, { "epoch": 1.0114308238076468, "grad_norm": 0.5510304570198059, "learning_rate": 0.0005, "loss": 1.7433453798294067, "step": 12830 }, { "epoch": 1.0115096570752857, "grad_norm": 0.47392329573631287, "learning_rate": 0.0005, "loss": 1.649721622467041, "step": 12831 }, { "epoch": 1.0115884903429246, "grad_norm": 0.5938378572463989, "learning_rate": 0.0005, "loss": 1.693169355392456, "step": 12832 }, { "epoch": 1.0116673236105636, "grad_norm": 0.5576387047767639, "learning_rate": 0.0005, "loss": 1.5257008075714111, "step": 12833 }, { "epoch": 1.0117461568782027, "grad_norm": 0.47545889019966125, "learning_rate": 0.0005, "loss": 1.7228515148162842, "step": 12834 }, { "epoch": 1.0118249901458416, "grad_norm": 0.6453250050544739, "learning_rate": 0.0005, "loss": 1.6874847412109375, "step": 12835 }, { "epoch": 1.0119038234134805, "grad_norm": 0.5636612772941589, "learning_rate": 0.0005, "loss": 1.5876593589782715, "step": 12836 }, { "epoch": 1.0119826566811194, "grad_norm": 0.47850877046585083, "learning_rate": 0.0005, "loss": 1.6420313119888306, "step": 12837 }, { "epoch": 1.0120614899487583, "grad_norm": 0.6018263697624207, "learning_rate": 0.0005, "loss": 1.6314247846603394, "step": 12838 }, { "epoch": 1.0121403232163972, "grad_norm": 0.7244160771369934, "learning_rate": 0.0005, "loss": 1.7616547346115112, "step": 12839 }, { "epoch": 1.0122191564840362, "grad_norm": 0.6606260538101196, "learning_rate": 0.0005, "loss": 1.7434391975402832, "step": 12840 }, { "epoch": 1.0122979897516753, "grad_norm": 0.5120312571525574, "learning_rate": 0.0005, "loss": 1.6123653650283813, "step": 12841 }, { "epoch": 1.0123768230193142, "grad_norm": 0.5717610716819763, "learning_rate": 0.0005, "loss": 1.6920796632766724, "step": 12842 }, { "epoch": 1.012455656286953, "grad_norm": 0.6808404922485352, "learning_rate": 0.0005, "loss": 1.7142772674560547, "step": 12843 }, { "epoch": 1.012534489554592, "grad_norm": 0.5933465957641602, "learning_rate": 0.0005, "loss": 1.5907105207443237, "step": 12844 }, { "epoch": 1.012613322822231, "grad_norm": 0.6221289038658142, "learning_rate": 0.0005, "loss": 1.5661275386810303, "step": 12845 }, { "epoch": 1.0126921560898698, "grad_norm": 0.4761563539505005, "learning_rate": 0.0005, "loss": 1.7390064001083374, "step": 12846 }, { "epoch": 1.012770989357509, "grad_norm": 0.6276479959487915, "learning_rate": 0.0005, "loss": 1.82320237159729, "step": 12847 }, { "epoch": 1.0128498226251479, "grad_norm": 0.5725457072257996, "learning_rate": 0.0005, "loss": 1.7296768426895142, "step": 12848 }, { "epoch": 1.0129286558927868, "grad_norm": 0.558558464050293, "learning_rate": 0.0005, "loss": 1.5224337577819824, "step": 12849 }, { "epoch": 1.0130074891604257, "grad_norm": 0.5502772927284241, "learning_rate": 0.0005, "loss": 1.7912007570266724, "step": 12850 }, { "epoch": 1.0130863224280646, "grad_norm": 0.567550539970398, "learning_rate": 0.0005, "loss": 1.6925642490386963, "step": 12851 }, { "epoch": 1.0131651556957035, "grad_norm": 0.5044676065444946, "learning_rate": 0.0005, "loss": 1.723309874534607, "step": 12852 }, { "epoch": 1.0132439889633424, "grad_norm": 0.6452385187149048, "learning_rate": 0.0005, "loss": 1.741152048110962, "step": 12853 }, { "epoch": 1.0133228222309816, "grad_norm": 0.5249278545379639, "learning_rate": 0.0005, "loss": 1.678356409072876, "step": 12854 }, { "epoch": 1.0134016554986205, "grad_norm": 0.4938535690307617, "learning_rate": 0.0005, "loss": 1.562430500984192, "step": 12855 }, { "epoch": 1.0134804887662594, "grad_norm": 0.6630310416221619, "learning_rate": 0.0005, "loss": 1.7487268447875977, "step": 12856 }, { "epoch": 1.0135593220338983, "grad_norm": 0.5067946314811707, "learning_rate": 0.0005, "loss": 1.7911189794540405, "step": 12857 }, { "epoch": 1.0136381553015372, "grad_norm": 0.5523273944854736, "learning_rate": 0.0005, "loss": 1.406306505203247, "step": 12858 }, { "epoch": 1.0137169885691761, "grad_norm": 0.5744993090629578, "learning_rate": 0.0005, "loss": 1.7361862659454346, "step": 12859 }, { "epoch": 1.013795821836815, "grad_norm": 0.534295916557312, "learning_rate": 0.0005, "loss": 1.5939724445343018, "step": 12860 }, { "epoch": 1.0138746551044542, "grad_norm": 0.5300690531730652, "learning_rate": 0.0005, "loss": 1.5438536405563354, "step": 12861 }, { "epoch": 1.013953488372093, "grad_norm": 0.5042603611946106, "learning_rate": 0.0005, "loss": 1.662304401397705, "step": 12862 }, { "epoch": 1.014032321639732, "grad_norm": 0.7071676254272461, "learning_rate": 0.0005, "loss": 1.9118001461029053, "step": 12863 }, { "epoch": 1.014111154907371, "grad_norm": 0.542687714099884, "learning_rate": 0.0005, "loss": 1.7164177894592285, "step": 12864 }, { "epoch": 1.0141899881750098, "grad_norm": 0.6137215495109558, "learning_rate": 0.0005, "loss": 1.9052720069885254, "step": 12865 }, { "epoch": 1.0142688214426487, "grad_norm": 0.3925498127937317, "learning_rate": 0.0005, "loss": 1.5745234489440918, "step": 12866 }, { "epoch": 1.0143476547102876, "grad_norm": 0.5454870462417603, "learning_rate": 0.0005, "loss": 1.717992901802063, "step": 12867 }, { "epoch": 1.0144264879779268, "grad_norm": 0.4244868755340576, "learning_rate": 0.0005, "loss": 1.5802081823349, "step": 12868 }, { "epoch": 1.0145053212455657, "grad_norm": 0.516120433807373, "learning_rate": 0.0005, "loss": 1.8035112619400024, "step": 12869 }, { "epoch": 1.0145841545132046, "grad_norm": 0.7088428139686584, "learning_rate": 0.0005, "loss": 1.7592988014221191, "step": 12870 }, { "epoch": 1.0146629877808435, "grad_norm": 0.5593705177307129, "learning_rate": 0.0005, "loss": 1.7009531259536743, "step": 12871 }, { "epoch": 1.0147418210484824, "grad_norm": 0.5969935655593872, "learning_rate": 0.0005, "loss": 1.843205213546753, "step": 12872 }, { "epoch": 1.0148206543161213, "grad_norm": 0.5100752115249634, "learning_rate": 0.0005, "loss": 1.8959529399871826, "step": 12873 }, { "epoch": 1.0148994875837603, "grad_norm": 0.48664674162864685, "learning_rate": 0.0005, "loss": 1.566413402557373, "step": 12874 }, { "epoch": 1.0149783208513994, "grad_norm": 0.6019432544708252, "learning_rate": 0.0005, "loss": 1.7762587070465088, "step": 12875 }, { "epoch": 1.0150571541190383, "grad_norm": 0.5422535538673401, "learning_rate": 0.0005, "loss": 1.6612242460250854, "step": 12876 }, { "epoch": 1.0151359873866772, "grad_norm": 0.48812335729599, "learning_rate": 0.0005, "loss": 1.638979434967041, "step": 12877 }, { "epoch": 1.0152148206543161, "grad_norm": 0.5733979344367981, "learning_rate": 0.0005, "loss": 1.8998150825500488, "step": 12878 }, { "epoch": 1.015293653921955, "grad_norm": 0.9633660316467285, "learning_rate": 0.0005, "loss": 1.6400656700134277, "step": 12879 }, { "epoch": 1.015372487189594, "grad_norm": 0.570738673210144, "learning_rate": 0.0005, "loss": 1.6333236694335938, "step": 12880 }, { "epoch": 1.0154513204572329, "grad_norm": 0.6652533411979675, "learning_rate": 0.0005, "loss": 1.775564432144165, "step": 12881 }, { "epoch": 1.015530153724872, "grad_norm": 0.6587675213813782, "learning_rate": 0.0005, "loss": 1.6847052574157715, "step": 12882 }, { "epoch": 1.015608986992511, "grad_norm": 0.49440979957580566, "learning_rate": 0.0005, "loss": 1.7432700395584106, "step": 12883 }, { "epoch": 1.0156878202601498, "grad_norm": 0.48666515946388245, "learning_rate": 0.0005, "loss": 1.5456185340881348, "step": 12884 }, { "epoch": 1.0157666535277887, "grad_norm": 0.8057880401611328, "learning_rate": 0.0005, "loss": 1.9548165798187256, "step": 12885 }, { "epoch": 1.0158454867954276, "grad_norm": 0.523446798324585, "learning_rate": 0.0005, "loss": 1.7048476934432983, "step": 12886 }, { "epoch": 1.0159243200630665, "grad_norm": 0.5961158275604248, "learning_rate": 0.0005, "loss": 1.5337848663330078, "step": 12887 }, { "epoch": 1.0160031533307055, "grad_norm": 0.47014185786247253, "learning_rate": 0.0005, "loss": 1.7473218441009521, "step": 12888 }, { "epoch": 1.0160819865983446, "grad_norm": 0.5198342800140381, "learning_rate": 0.0005, "loss": 1.5608515739440918, "step": 12889 }, { "epoch": 1.0161608198659835, "grad_norm": 0.6867214441299438, "learning_rate": 0.0005, "loss": 1.6445558071136475, "step": 12890 }, { "epoch": 1.0162396531336224, "grad_norm": 0.5881261825561523, "learning_rate": 0.0005, "loss": 1.6664278507232666, "step": 12891 }, { "epoch": 1.0163184864012613, "grad_norm": 0.5647026300430298, "learning_rate": 0.0005, "loss": 1.7246886491775513, "step": 12892 }, { "epoch": 1.0163973196689002, "grad_norm": 0.5628154873847961, "learning_rate": 0.0005, "loss": 1.4938836097717285, "step": 12893 }, { "epoch": 1.0164761529365391, "grad_norm": 0.5009053349494934, "learning_rate": 0.0005, "loss": 1.6652092933654785, "step": 12894 }, { "epoch": 1.016554986204178, "grad_norm": 0.7293728590011597, "learning_rate": 0.0005, "loss": 1.811124563217163, "step": 12895 }, { "epoch": 1.0166338194718172, "grad_norm": 0.9043456315994263, "learning_rate": 0.0005, "loss": 1.5452684164047241, "step": 12896 }, { "epoch": 1.016712652739456, "grad_norm": 0.6095166802406311, "learning_rate": 0.0005, "loss": 1.9054895639419556, "step": 12897 }, { "epoch": 1.016791486007095, "grad_norm": 0.4657185971736908, "learning_rate": 0.0005, "loss": 1.5048325061798096, "step": 12898 }, { "epoch": 1.016870319274734, "grad_norm": 0.44192731380462646, "learning_rate": 0.0005, "loss": 1.5888612270355225, "step": 12899 }, { "epoch": 1.0169491525423728, "grad_norm": 0.6758037209510803, "learning_rate": 0.0005, "loss": 1.695172905921936, "step": 12900 }, { "epoch": 1.0170279858100117, "grad_norm": 0.48539674282073975, "learning_rate": 0.0005, "loss": 1.7218068838119507, "step": 12901 }, { "epoch": 1.0171068190776507, "grad_norm": 0.5480556488037109, "learning_rate": 0.0005, "loss": 1.7097474336624146, "step": 12902 }, { "epoch": 1.0171856523452898, "grad_norm": 0.5770822167396545, "learning_rate": 0.0005, "loss": 1.87847900390625, "step": 12903 }, { "epoch": 1.0172644856129287, "grad_norm": 0.5185340642929077, "learning_rate": 0.0005, "loss": 1.7142624855041504, "step": 12904 }, { "epoch": 1.0173433188805676, "grad_norm": 0.6312534809112549, "learning_rate": 0.0005, "loss": 1.7151081562042236, "step": 12905 }, { "epoch": 1.0174221521482065, "grad_norm": 0.5431466102600098, "learning_rate": 0.0005, "loss": 1.679699182510376, "step": 12906 }, { "epoch": 1.0175009854158454, "grad_norm": 0.6263130903244019, "learning_rate": 0.0005, "loss": 1.6380953788757324, "step": 12907 }, { "epoch": 1.0175798186834843, "grad_norm": 0.7001510262489319, "learning_rate": 0.0005, "loss": 1.6661357879638672, "step": 12908 }, { "epoch": 1.0176586519511235, "grad_norm": 0.7122490406036377, "learning_rate": 0.0005, "loss": 2.0330891609191895, "step": 12909 }, { "epoch": 1.0177374852187624, "grad_norm": 0.7996482253074646, "learning_rate": 0.0005, "loss": 1.5762913227081299, "step": 12910 }, { "epoch": 1.0178163184864013, "grad_norm": 0.6653999090194702, "learning_rate": 0.0005, "loss": 1.7486698627471924, "step": 12911 }, { "epoch": 1.0178951517540402, "grad_norm": 0.48922237753868103, "learning_rate": 0.0005, "loss": 1.5473417043685913, "step": 12912 }, { "epoch": 1.0179739850216791, "grad_norm": 0.5885183215141296, "learning_rate": 0.0005, "loss": 1.5104320049285889, "step": 12913 }, { "epoch": 1.018052818289318, "grad_norm": 0.5736448168754578, "learning_rate": 0.0005, "loss": 1.7013916969299316, "step": 12914 }, { "epoch": 1.018131651556957, "grad_norm": 0.4589400589466095, "learning_rate": 0.0005, "loss": 1.5990327596664429, "step": 12915 }, { "epoch": 1.018210484824596, "grad_norm": 0.5114646553993225, "learning_rate": 0.0005, "loss": 1.8207855224609375, "step": 12916 }, { "epoch": 1.018289318092235, "grad_norm": 0.5199742317199707, "learning_rate": 0.0005, "loss": 1.8558170795440674, "step": 12917 }, { "epoch": 1.018368151359874, "grad_norm": 0.7751470804214478, "learning_rate": 0.0005, "loss": 1.8503735065460205, "step": 12918 }, { "epoch": 1.0184469846275128, "grad_norm": 0.4955967962741852, "learning_rate": 0.0005, "loss": 1.7452282905578613, "step": 12919 }, { "epoch": 1.0185258178951517, "grad_norm": 0.6358543634414673, "learning_rate": 0.0005, "loss": 1.6869227886199951, "step": 12920 }, { "epoch": 1.0186046511627906, "grad_norm": 0.4896043837070465, "learning_rate": 0.0005, "loss": 1.5027304887771606, "step": 12921 }, { "epoch": 1.0186834844304296, "grad_norm": 0.5332189798355103, "learning_rate": 0.0005, "loss": 1.675444483757019, "step": 12922 }, { "epoch": 1.0187623176980687, "grad_norm": 0.5188814997673035, "learning_rate": 0.0005, "loss": 1.8546987771987915, "step": 12923 }, { "epoch": 1.0188411509657076, "grad_norm": 0.5664215683937073, "learning_rate": 0.0005, "loss": 1.8449501991271973, "step": 12924 }, { "epoch": 1.0189199842333465, "grad_norm": 1.1964902877807617, "learning_rate": 0.0005, "loss": 1.6838972568511963, "step": 12925 }, { "epoch": 1.0189988175009854, "grad_norm": 0.4798499643802643, "learning_rate": 0.0005, "loss": 1.6224253177642822, "step": 12926 }, { "epoch": 1.0190776507686243, "grad_norm": 0.5651968121528625, "learning_rate": 0.0005, "loss": 1.8940008878707886, "step": 12927 }, { "epoch": 1.0191564840362632, "grad_norm": 1.3074913024902344, "learning_rate": 0.0005, "loss": 1.5861790180206299, "step": 12928 }, { "epoch": 1.0192353173039022, "grad_norm": 0.44464191794395447, "learning_rate": 0.0005, "loss": 1.7612806558609009, "step": 12929 }, { "epoch": 1.0193141505715413, "grad_norm": 0.47379589080810547, "learning_rate": 0.0005, "loss": 1.7535269260406494, "step": 12930 }, { "epoch": 1.0193929838391802, "grad_norm": 0.6892691254615784, "learning_rate": 0.0005, "loss": 2.024704933166504, "step": 12931 }, { "epoch": 1.0194718171068191, "grad_norm": 0.5628840923309326, "learning_rate": 0.0005, "loss": 1.825399398803711, "step": 12932 }, { "epoch": 1.019550650374458, "grad_norm": 0.5860928893089294, "learning_rate": 0.0005, "loss": 1.620241403579712, "step": 12933 }, { "epoch": 1.019629483642097, "grad_norm": 0.5703926086425781, "learning_rate": 0.0005, "loss": 1.7231693267822266, "step": 12934 }, { "epoch": 1.0197083169097358, "grad_norm": 0.5221336483955383, "learning_rate": 0.0005, "loss": 1.5322909355163574, "step": 12935 }, { "epoch": 1.0197871501773748, "grad_norm": 0.5994620323181152, "learning_rate": 0.0005, "loss": 1.5952396392822266, "step": 12936 }, { "epoch": 1.019865983445014, "grad_norm": 0.535720944404602, "learning_rate": 0.0005, "loss": 1.940143346786499, "step": 12937 }, { "epoch": 1.0199448167126528, "grad_norm": 0.5588040351867676, "learning_rate": 0.0005, "loss": 1.8292678594589233, "step": 12938 }, { "epoch": 1.0200236499802917, "grad_norm": 0.5726542472839355, "learning_rate": 0.0005, "loss": 1.6298179626464844, "step": 12939 }, { "epoch": 1.0201024832479306, "grad_norm": 0.5548467040061951, "learning_rate": 0.0005, "loss": 1.6379811763763428, "step": 12940 }, { "epoch": 1.0201813165155695, "grad_norm": 0.880950391292572, "learning_rate": 0.0005, "loss": 1.8261443376541138, "step": 12941 }, { "epoch": 1.0202601497832084, "grad_norm": 0.5053159594535828, "learning_rate": 0.0005, "loss": 1.6385104656219482, "step": 12942 }, { "epoch": 1.0203389830508474, "grad_norm": 0.6348488330841064, "learning_rate": 0.0005, "loss": 2.0102171897888184, "step": 12943 }, { "epoch": 1.0204178163184865, "grad_norm": 0.5160261392593384, "learning_rate": 0.0005, "loss": 1.6107244491577148, "step": 12944 }, { "epoch": 1.0204966495861254, "grad_norm": 0.5538614392280579, "learning_rate": 0.0005, "loss": 1.8174817562103271, "step": 12945 }, { "epoch": 1.0205754828537643, "grad_norm": 0.49673089385032654, "learning_rate": 0.0005, "loss": 1.5648397207260132, "step": 12946 }, { "epoch": 1.0206543161214032, "grad_norm": 1.1711809635162354, "learning_rate": 0.0005, "loss": 1.8666532039642334, "step": 12947 }, { "epoch": 1.0207331493890421, "grad_norm": 0.4710139334201813, "learning_rate": 0.0005, "loss": 1.542872428894043, "step": 12948 }, { "epoch": 1.020811982656681, "grad_norm": 0.49233579635620117, "learning_rate": 0.0005, "loss": 1.3800709247589111, "step": 12949 }, { "epoch": 1.02089081592432, "grad_norm": 0.5399828553199768, "learning_rate": 0.0005, "loss": 1.679444670677185, "step": 12950 }, { "epoch": 1.020969649191959, "grad_norm": 0.6472378969192505, "learning_rate": 0.0005, "loss": 1.9152384996414185, "step": 12951 }, { "epoch": 1.021048482459598, "grad_norm": 0.4819762110710144, "learning_rate": 0.0005, "loss": 1.5978765487670898, "step": 12952 }, { "epoch": 1.021127315727237, "grad_norm": 0.5490145087242126, "learning_rate": 0.0005, "loss": 1.5870089530944824, "step": 12953 }, { "epoch": 1.0212061489948758, "grad_norm": 0.6067835688591003, "learning_rate": 0.0005, "loss": 1.7711988687515259, "step": 12954 }, { "epoch": 1.0212849822625147, "grad_norm": 0.5451778769493103, "learning_rate": 0.0005, "loss": 1.4475575685501099, "step": 12955 }, { "epoch": 1.0213638155301537, "grad_norm": 0.7391431331634521, "learning_rate": 0.0005, "loss": 1.8819279670715332, "step": 12956 }, { "epoch": 1.0214426487977926, "grad_norm": 0.5277689099311829, "learning_rate": 0.0005, "loss": 1.8653475046157837, "step": 12957 }, { "epoch": 1.0215214820654317, "grad_norm": 1.0208340883255005, "learning_rate": 0.0005, "loss": 1.6752076148986816, "step": 12958 }, { "epoch": 1.0216003153330706, "grad_norm": 0.46007177233695984, "learning_rate": 0.0005, "loss": 1.5286765098571777, "step": 12959 }, { "epoch": 1.0216791486007095, "grad_norm": 0.6746207475662231, "learning_rate": 0.0005, "loss": 1.5280003547668457, "step": 12960 }, { "epoch": 1.0217579818683484, "grad_norm": 1.7751379013061523, "learning_rate": 0.0005, "loss": 2.129080295562744, "step": 12961 }, { "epoch": 1.0218368151359873, "grad_norm": 0.579120934009552, "learning_rate": 0.0005, "loss": 1.7648794651031494, "step": 12962 }, { "epoch": 1.0219156484036263, "grad_norm": 0.5505395531654358, "learning_rate": 0.0005, "loss": 1.830177664756775, "step": 12963 }, { "epoch": 1.0219944816712652, "grad_norm": 0.49655085802078247, "learning_rate": 0.0005, "loss": 1.774639368057251, "step": 12964 }, { "epoch": 1.0220733149389043, "grad_norm": 0.5433403253555298, "learning_rate": 0.0005, "loss": 1.7372418642044067, "step": 12965 }, { "epoch": 1.0221521482065432, "grad_norm": 0.5329174399375916, "learning_rate": 0.0005, "loss": 1.5841628313064575, "step": 12966 }, { "epoch": 1.0222309814741821, "grad_norm": 0.848680317401886, "learning_rate": 0.0005, "loss": 1.925429105758667, "step": 12967 }, { "epoch": 1.022309814741821, "grad_norm": 0.542518675327301, "learning_rate": 0.0005, "loss": 1.7174205780029297, "step": 12968 }, { "epoch": 1.02238864800946, "grad_norm": 0.6036101579666138, "learning_rate": 0.0005, "loss": 1.7698228359222412, "step": 12969 }, { "epoch": 1.0224674812770989, "grad_norm": 0.6147660613059998, "learning_rate": 0.0005, "loss": 1.7080715894699097, "step": 12970 }, { "epoch": 1.022546314544738, "grad_norm": 0.66041499376297, "learning_rate": 0.0005, "loss": 1.7843296527862549, "step": 12971 }, { "epoch": 1.022625147812377, "grad_norm": 0.6190462708473206, "learning_rate": 0.0005, "loss": 1.7003263235092163, "step": 12972 }, { "epoch": 1.0227039810800158, "grad_norm": 0.7002339959144592, "learning_rate": 0.0005, "loss": 1.565671443939209, "step": 12973 }, { "epoch": 1.0227828143476547, "grad_norm": 0.45207861065864563, "learning_rate": 0.0005, "loss": 1.582268238067627, "step": 12974 }, { "epoch": 1.0228616476152936, "grad_norm": 0.6642864942550659, "learning_rate": 0.0005, "loss": 1.7417895793914795, "step": 12975 }, { "epoch": 1.0229404808829325, "grad_norm": 0.5731069445610046, "learning_rate": 0.0005, "loss": 1.6734673976898193, "step": 12976 }, { "epoch": 1.0230193141505715, "grad_norm": 0.5271856188774109, "learning_rate": 0.0005, "loss": 1.489973545074463, "step": 12977 }, { "epoch": 1.0230981474182106, "grad_norm": 0.5848885774612427, "learning_rate": 0.0005, "loss": 1.5757238864898682, "step": 12978 }, { "epoch": 1.0231769806858495, "grad_norm": 0.5871674418449402, "learning_rate": 0.0005, "loss": 1.6597847938537598, "step": 12979 }, { "epoch": 1.0232558139534884, "grad_norm": 0.740760087966919, "learning_rate": 0.0005, "loss": 1.603543758392334, "step": 12980 }, { "epoch": 1.0233346472211273, "grad_norm": 0.6874999403953552, "learning_rate": 0.0005, "loss": 1.7743449211120605, "step": 12981 }, { "epoch": 1.0234134804887662, "grad_norm": 0.5335138440132141, "learning_rate": 0.0005, "loss": 1.8148865699768066, "step": 12982 }, { "epoch": 1.0234923137564051, "grad_norm": 0.5060628056526184, "learning_rate": 0.0005, "loss": 1.6644961833953857, "step": 12983 }, { "epoch": 1.023571147024044, "grad_norm": 0.4285849332809448, "learning_rate": 0.0005, "loss": 1.5366902351379395, "step": 12984 }, { "epoch": 1.0236499802916832, "grad_norm": 0.4873327910900116, "learning_rate": 0.0005, "loss": 1.7729140520095825, "step": 12985 }, { "epoch": 1.023728813559322, "grad_norm": 0.6250122785568237, "learning_rate": 0.0005, "loss": 1.7934192419052124, "step": 12986 }, { "epoch": 1.023807646826961, "grad_norm": 0.5075249075889587, "learning_rate": 0.0005, "loss": 1.6249184608459473, "step": 12987 }, { "epoch": 1.0238864800946, "grad_norm": 0.48242878913879395, "learning_rate": 0.0005, "loss": 1.5857555866241455, "step": 12988 }, { "epoch": 1.0239653133622388, "grad_norm": 0.5123632550239563, "learning_rate": 0.0005, "loss": 1.4962698221206665, "step": 12989 }, { "epoch": 1.0240441466298777, "grad_norm": 0.5917131304740906, "learning_rate": 0.0005, "loss": 1.5433223247528076, "step": 12990 }, { "epoch": 1.0241229798975167, "grad_norm": 0.5418887138366699, "learning_rate": 0.0005, "loss": 1.5567405223846436, "step": 12991 }, { "epoch": 1.0242018131651558, "grad_norm": 0.47095370292663574, "learning_rate": 0.0005, "loss": 1.6662647724151611, "step": 12992 }, { "epoch": 1.0242806464327947, "grad_norm": 0.5841354131698608, "learning_rate": 0.0005, "loss": 1.7697988748550415, "step": 12993 }, { "epoch": 1.0243594797004336, "grad_norm": 0.5293015241622925, "learning_rate": 0.0005, "loss": 1.6648529767990112, "step": 12994 }, { "epoch": 1.0244383129680725, "grad_norm": 0.7224124073982239, "learning_rate": 0.0005, "loss": 1.805346131324768, "step": 12995 }, { "epoch": 1.0245171462357114, "grad_norm": 0.5307212471961975, "learning_rate": 0.0005, "loss": 1.7121299505233765, "step": 12996 }, { "epoch": 1.0245959795033504, "grad_norm": 0.5367142558097839, "learning_rate": 0.0005, "loss": 1.7898874282836914, "step": 12997 }, { "epoch": 1.0246748127709893, "grad_norm": 0.5635845065116882, "learning_rate": 0.0005, "loss": 1.6074810028076172, "step": 12998 }, { "epoch": 1.0247536460386284, "grad_norm": 0.7814607620239258, "learning_rate": 0.0005, "loss": 1.5456100702285767, "step": 12999 }, { "epoch": 1.0248324793062673, "grad_norm": 0.49582764506340027, "learning_rate": 0.0005, "loss": 1.9096083641052246, "step": 13000 }, { "epoch": 1.0249113125739062, "grad_norm": 0.5114544034004211, "learning_rate": 0.0005, "loss": 1.7127330303192139, "step": 13001 }, { "epoch": 1.0249901458415451, "grad_norm": 0.5587915778160095, "learning_rate": 0.0005, "loss": 1.6925370693206787, "step": 13002 }, { "epoch": 1.025068979109184, "grad_norm": 0.5139444470405579, "learning_rate": 0.0005, "loss": 1.7018418312072754, "step": 13003 }, { "epoch": 1.025147812376823, "grad_norm": 0.5533376932144165, "learning_rate": 0.0005, "loss": 1.7604502439498901, "step": 13004 }, { "epoch": 1.0252266456444619, "grad_norm": 0.5809431672096252, "learning_rate": 0.0005, "loss": 1.7518247365951538, "step": 13005 }, { "epoch": 1.025305478912101, "grad_norm": 0.4853513836860657, "learning_rate": 0.0005, "loss": 1.6404235363006592, "step": 13006 }, { "epoch": 1.02538431217974, "grad_norm": 0.536821722984314, "learning_rate": 0.0005, "loss": 1.7320164442062378, "step": 13007 }, { "epoch": 1.0254631454473788, "grad_norm": 0.6078463196754456, "learning_rate": 0.0005, "loss": 1.4917047023773193, "step": 13008 }, { "epoch": 1.0255419787150177, "grad_norm": 0.4838433265686035, "learning_rate": 0.0005, "loss": 1.835161805152893, "step": 13009 }, { "epoch": 1.0256208119826566, "grad_norm": 0.5228469967842102, "learning_rate": 0.0005, "loss": 1.701222538948059, "step": 13010 }, { "epoch": 1.0256996452502956, "grad_norm": 0.6798361539840698, "learning_rate": 0.0005, "loss": 1.5790692567825317, "step": 13011 }, { "epoch": 1.0257784785179345, "grad_norm": 0.5760246515274048, "learning_rate": 0.0005, "loss": 1.5267404317855835, "step": 13012 }, { "epoch": 1.0258573117855736, "grad_norm": 0.6303824782371521, "learning_rate": 0.0005, "loss": 1.6889899969100952, "step": 13013 }, { "epoch": 1.0259361450532125, "grad_norm": 0.6466764807701111, "learning_rate": 0.0005, "loss": 1.6785396337509155, "step": 13014 }, { "epoch": 1.0260149783208514, "grad_norm": 0.5383630990982056, "learning_rate": 0.0005, "loss": 1.8870999813079834, "step": 13015 }, { "epoch": 1.0260938115884903, "grad_norm": 0.6669480800628662, "learning_rate": 0.0005, "loss": 2.024970769882202, "step": 13016 }, { "epoch": 1.0261726448561292, "grad_norm": 0.5756479501724243, "learning_rate": 0.0005, "loss": 1.784471035003662, "step": 13017 }, { "epoch": 1.0262514781237682, "grad_norm": 0.5092270374298096, "learning_rate": 0.0005, "loss": 1.3345890045166016, "step": 13018 }, { "epoch": 1.026330311391407, "grad_norm": 0.7404310703277588, "learning_rate": 0.0005, "loss": 1.8158830404281616, "step": 13019 }, { "epoch": 1.0264091446590462, "grad_norm": 0.5138136744499207, "learning_rate": 0.0005, "loss": 1.5145163536071777, "step": 13020 }, { "epoch": 1.0264879779266851, "grad_norm": 0.504957377910614, "learning_rate": 0.0005, "loss": 1.7144203186035156, "step": 13021 }, { "epoch": 1.026566811194324, "grad_norm": 0.47616976499557495, "learning_rate": 0.0005, "loss": 1.6452527046203613, "step": 13022 }, { "epoch": 1.026645644461963, "grad_norm": 0.6721386313438416, "learning_rate": 0.0005, "loss": 2.018125534057617, "step": 13023 }, { "epoch": 1.0267244777296018, "grad_norm": 0.6517605781555176, "learning_rate": 0.0005, "loss": 1.8861498832702637, "step": 13024 }, { "epoch": 1.0268033109972408, "grad_norm": 0.5156795978546143, "learning_rate": 0.0005, "loss": 1.8595654964447021, "step": 13025 }, { "epoch": 1.0268821442648797, "grad_norm": 0.5581664443016052, "learning_rate": 0.0005, "loss": 1.7559425830841064, "step": 13026 }, { "epoch": 1.0269609775325188, "grad_norm": 0.6088144779205322, "learning_rate": 0.0005, "loss": 1.7413685321807861, "step": 13027 }, { "epoch": 1.0270398108001577, "grad_norm": 0.4628487825393677, "learning_rate": 0.0005, "loss": 1.7071175575256348, "step": 13028 }, { "epoch": 1.0271186440677966, "grad_norm": 0.495898574590683, "learning_rate": 0.0005, "loss": 1.614898443222046, "step": 13029 }, { "epoch": 1.0271974773354355, "grad_norm": 0.676791250705719, "learning_rate": 0.0005, "loss": 1.6863911151885986, "step": 13030 }, { "epoch": 1.0272763106030744, "grad_norm": 0.4455007314682007, "learning_rate": 0.0005, "loss": 1.8150699138641357, "step": 13031 }, { "epoch": 1.0273551438707134, "grad_norm": 0.5782683491706848, "learning_rate": 0.0005, "loss": 1.450678825378418, "step": 13032 }, { "epoch": 1.0274339771383523, "grad_norm": 0.5020875930786133, "learning_rate": 0.0005, "loss": 1.8599146604537964, "step": 13033 }, { "epoch": 1.0275128104059914, "grad_norm": 0.5912691950798035, "learning_rate": 0.0005, "loss": 1.761553168296814, "step": 13034 }, { "epoch": 1.0275916436736303, "grad_norm": 0.5320544242858887, "learning_rate": 0.0005, "loss": 1.618135690689087, "step": 13035 }, { "epoch": 1.0276704769412692, "grad_norm": 0.9263684153556824, "learning_rate": 0.0005, "loss": 1.9382829666137695, "step": 13036 }, { "epoch": 1.0277493102089081, "grad_norm": 1.1219351291656494, "learning_rate": 0.0005, "loss": 1.9557241201400757, "step": 13037 }, { "epoch": 1.027828143476547, "grad_norm": 0.6294212937355042, "learning_rate": 0.0005, "loss": 1.7733123302459717, "step": 13038 }, { "epoch": 1.027906976744186, "grad_norm": 0.5693590641021729, "learning_rate": 0.0005, "loss": 1.7870702743530273, "step": 13039 }, { "epoch": 1.027985810011825, "grad_norm": 0.5173965096473694, "learning_rate": 0.0005, "loss": 1.6167423725128174, "step": 13040 }, { "epoch": 1.028064643279464, "grad_norm": 0.6073510646820068, "learning_rate": 0.0005, "loss": 1.870598554611206, "step": 13041 }, { "epoch": 1.028143476547103, "grad_norm": 0.5339252352714539, "learning_rate": 0.0005, "loss": 1.6830456256866455, "step": 13042 }, { "epoch": 1.0282223098147418, "grad_norm": 0.6162847280502319, "learning_rate": 0.0005, "loss": 1.5446555614471436, "step": 13043 }, { "epoch": 1.0283011430823807, "grad_norm": 0.608456552028656, "learning_rate": 0.0005, "loss": 2.031938314437866, "step": 13044 }, { "epoch": 1.0283799763500197, "grad_norm": 0.4691295921802521, "learning_rate": 0.0005, "loss": 1.9637360572814941, "step": 13045 }, { "epoch": 1.0284588096176586, "grad_norm": 0.6405089497566223, "learning_rate": 0.0005, "loss": 1.7356585264205933, "step": 13046 }, { "epoch": 1.0285376428852977, "grad_norm": 0.5099338293075562, "learning_rate": 0.0005, "loss": 1.695562720298767, "step": 13047 }, { "epoch": 1.0286164761529366, "grad_norm": 0.5433681607246399, "learning_rate": 0.0005, "loss": 1.6488182544708252, "step": 13048 }, { "epoch": 1.0286953094205755, "grad_norm": 0.4351987838745117, "learning_rate": 0.0005, "loss": 1.6142933368682861, "step": 13049 }, { "epoch": 1.0287741426882144, "grad_norm": 0.6500259637832642, "learning_rate": 0.0005, "loss": 1.628091812133789, "step": 13050 }, { "epoch": 1.0288529759558533, "grad_norm": 0.6524912118911743, "learning_rate": 0.0005, "loss": 1.8337032794952393, "step": 13051 }, { "epoch": 1.0289318092234923, "grad_norm": 0.5474245548248291, "learning_rate": 0.0005, "loss": 1.502750039100647, "step": 13052 }, { "epoch": 1.0290106424911312, "grad_norm": 0.750873863697052, "learning_rate": 0.0005, "loss": 1.5196552276611328, "step": 13053 }, { "epoch": 1.0290894757587703, "grad_norm": 0.7309288382530212, "learning_rate": 0.0005, "loss": 1.930774450302124, "step": 13054 }, { "epoch": 1.0291683090264092, "grad_norm": 2.2167539596557617, "learning_rate": 0.0005, "loss": 1.5908187627792358, "step": 13055 }, { "epoch": 1.0292471422940481, "grad_norm": 0.5777267813682556, "learning_rate": 0.0005, "loss": 1.660507321357727, "step": 13056 }, { "epoch": 1.029325975561687, "grad_norm": 0.7174568176269531, "learning_rate": 0.0005, "loss": 1.8167941570281982, "step": 13057 }, { "epoch": 1.029404808829326, "grad_norm": 0.5290317535400391, "learning_rate": 0.0005, "loss": 1.8803967237472534, "step": 13058 }, { "epoch": 1.0294836420969649, "grad_norm": 0.578967809677124, "learning_rate": 0.0005, "loss": 1.746488332748413, "step": 13059 }, { "epoch": 1.0295624753646038, "grad_norm": 0.47788310050964355, "learning_rate": 0.0005, "loss": 1.7562198638916016, "step": 13060 }, { "epoch": 1.029641308632243, "grad_norm": 0.5067899823188782, "learning_rate": 0.0005, "loss": 1.5953750610351562, "step": 13061 }, { "epoch": 1.0297201418998818, "grad_norm": 0.6974384784698486, "learning_rate": 0.0005, "loss": 1.9508421421051025, "step": 13062 }, { "epoch": 1.0297989751675207, "grad_norm": 0.7079967260360718, "learning_rate": 0.0005, "loss": 1.7572346925735474, "step": 13063 }, { "epoch": 1.0298778084351596, "grad_norm": 0.5175374150276184, "learning_rate": 0.0005, "loss": 1.7721056938171387, "step": 13064 }, { "epoch": 1.0299566417027985, "grad_norm": 0.5930699706077576, "learning_rate": 0.0005, "loss": 1.6458157300949097, "step": 13065 }, { "epoch": 1.0300354749704375, "grad_norm": 0.5322570204734802, "learning_rate": 0.0005, "loss": 1.5591596364974976, "step": 13066 }, { "epoch": 1.0301143082380764, "grad_norm": 0.8560011386871338, "learning_rate": 0.0005, "loss": 1.650411605834961, "step": 13067 }, { "epoch": 1.0301931415057155, "grad_norm": 0.5428878664970398, "learning_rate": 0.0005, "loss": 1.7211155891418457, "step": 13068 }, { "epoch": 1.0302719747733544, "grad_norm": 1.6204770803451538, "learning_rate": 0.0005, "loss": 1.5753920078277588, "step": 13069 }, { "epoch": 1.0303508080409933, "grad_norm": 0.5728191137313843, "learning_rate": 0.0005, "loss": 1.6555365324020386, "step": 13070 }, { "epoch": 1.0304296413086322, "grad_norm": 0.48788341879844666, "learning_rate": 0.0005, "loss": 1.6726596355438232, "step": 13071 }, { "epoch": 1.0305084745762711, "grad_norm": 0.5856693983078003, "learning_rate": 0.0005, "loss": 1.8023247718811035, "step": 13072 }, { "epoch": 1.03058730784391, "grad_norm": 0.4657604694366455, "learning_rate": 0.0005, "loss": 1.6700084209442139, "step": 13073 }, { "epoch": 1.030666141111549, "grad_norm": 0.7456731796264648, "learning_rate": 0.0005, "loss": 1.994001865386963, "step": 13074 }, { "epoch": 1.030744974379188, "grad_norm": 0.49420151114463806, "learning_rate": 0.0005, "loss": 1.6825408935546875, "step": 13075 }, { "epoch": 1.030823807646827, "grad_norm": 0.5128656029701233, "learning_rate": 0.0005, "loss": 1.6196976900100708, "step": 13076 }, { "epoch": 1.030902640914466, "grad_norm": 0.7466577887535095, "learning_rate": 0.0005, "loss": 1.786927342414856, "step": 13077 }, { "epoch": 1.0309814741821048, "grad_norm": 0.5107924342155457, "learning_rate": 0.0005, "loss": 1.6555980443954468, "step": 13078 }, { "epoch": 1.0310603074497438, "grad_norm": 0.6342727541923523, "learning_rate": 0.0005, "loss": 1.8191955089569092, "step": 13079 }, { "epoch": 1.0311391407173827, "grad_norm": 0.5651176571846008, "learning_rate": 0.0005, "loss": 1.7446208000183105, "step": 13080 }, { "epoch": 1.0312179739850216, "grad_norm": 0.5844842791557312, "learning_rate": 0.0005, "loss": 1.8026716709136963, "step": 13081 }, { "epoch": 1.0312968072526607, "grad_norm": 0.4561960995197296, "learning_rate": 0.0005, "loss": 1.6059794425964355, "step": 13082 }, { "epoch": 1.0313756405202996, "grad_norm": 0.680454671382904, "learning_rate": 0.0005, "loss": 1.706356406211853, "step": 13083 }, { "epoch": 1.0314544737879385, "grad_norm": 0.6245647072792053, "learning_rate": 0.0005, "loss": 1.8696749210357666, "step": 13084 }, { "epoch": 1.0315333070555774, "grad_norm": 0.5863022804260254, "learning_rate": 0.0005, "loss": 1.7979705333709717, "step": 13085 }, { "epoch": 1.0316121403232164, "grad_norm": 0.5841493010520935, "learning_rate": 0.0005, "loss": 1.6558403968811035, "step": 13086 }, { "epoch": 1.0316909735908553, "grad_norm": 0.4321652352809906, "learning_rate": 0.0005, "loss": 1.6848015785217285, "step": 13087 }, { "epoch": 1.0317698068584944, "grad_norm": 0.49574223160743713, "learning_rate": 0.0005, "loss": 1.8622907400131226, "step": 13088 }, { "epoch": 1.0318486401261333, "grad_norm": 0.6603681445121765, "learning_rate": 0.0005, "loss": 1.6621246337890625, "step": 13089 }, { "epoch": 1.0319274733937722, "grad_norm": 0.7463895082473755, "learning_rate": 0.0005, "loss": 1.922244668006897, "step": 13090 }, { "epoch": 1.0320063066614111, "grad_norm": 0.6491203904151917, "learning_rate": 0.0005, "loss": 1.5135478973388672, "step": 13091 }, { "epoch": 1.03208513992905, "grad_norm": 2.627295970916748, "learning_rate": 0.0005, "loss": 1.8639057874679565, "step": 13092 }, { "epoch": 1.032163973196689, "grad_norm": 0.8637574315071106, "learning_rate": 0.0005, "loss": 1.5927858352661133, "step": 13093 }, { "epoch": 1.0322428064643279, "grad_norm": 0.718712329864502, "learning_rate": 0.0005, "loss": 1.8810851573944092, "step": 13094 }, { "epoch": 1.032321639731967, "grad_norm": 1.2640111446380615, "learning_rate": 0.0005, "loss": 1.9960691928863525, "step": 13095 }, { "epoch": 1.032400472999606, "grad_norm": 0.6950739026069641, "learning_rate": 0.0005, "loss": 1.8974463939666748, "step": 13096 }, { "epoch": 1.0324793062672448, "grad_norm": 0.6677806377410889, "learning_rate": 0.0005, "loss": 1.9019238948822021, "step": 13097 }, { "epoch": 1.0325581395348837, "grad_norm": 0.6126833558082581, "learning_rate": 0.0005, "loss": 1.7134959697723389, "step": 13098 }, { "epoch": 1.0326369728025226, "grad_norm": 0.6232456564903259, "learning_rate": 0.0005, "loss": 1.9077526330947876, "step": 13099 }, { "epoch": 1.0327158060701616, "grad_norm": 0.751410961151123, "learning_rate": 0.0005, "loss": 1.7654061317443848, "step": 13100 }, { "epoch": 1.0327946393378005, "grad_norm": 0.5187357068061829, "learning_rate": 0.0005, "loss": 1.766315221786499, "step": 13101 }, { "epoch": 1.0328734726054396, "grad_norm": 0.6386058330535889, "learning_rate": 0.0005, "loss": 1.6237940788269043, "step": 13102 }, { "epoch": 1.0329523058730785, "grad_norm": 0.4376908242702484, "learning_rate": 0.0005, "loss": 1.463376522064209, "step": 13103 }, { "epoch": 1.0330311391407174, "grad_norm": 0.5547128319740295, "learning_rate": 0.0005, "loss": 1.8357453346252441, "step": 13104 }, { "epoch": 1.0331099724083563, "grad_norm": 0.5932473540306091, "learning_rate": 0.0005, "loss": 1.6777907609939575, "step": 13105 }, { "epoch": 1.0331888056759952, "grad_norm": 0.6252790689468384, "learning_rate": 0.0005, "loss": 1.460770845413208, "step": 13106 }, { "epoch": 1.0332676389436342, "grad_norm": 0.4622552692890167, "learning_rate": 0.0005, "loss": 1.4520251750946045, "step": 13107 }, { "epoch": 1.033346472211273, "grad_norm": 0.5193429589271545, "learning_rate": 0.0005, "loss": 1.8333402872085571, "step": 13108 }, { "epoch": 1.0334253054789122, "grad_norm": 0.6734906435012817, "learning_rate": 0.0005, "loss": 1.8963356018066406, "step": 13109 }, { "epoch": 1.0335041387465511, "grad_norm": 0.5692622661590576, "learning_rate": 0.0005, "loss": 1.6717019081115723, "step": 13110 }, { "epoch": 1.03358297201419, "grad_norm": 0.45771563053131104, "learning_rate": 0.0005, "loss": 1.453153371810913, "step": 13111 }, { "epoch": 1.033661805281829, "grad_norm": 0.6896628737449646, "learning_rate": 0.0005, "loss": 1.6875925064086914, "step": 13112 }, { "epoch": 1.0337406385494678, "grad_norm": 0.5610089898109436, "learning_rate": 0.0005, "loss": 1.7554069757461548, "step": 13113 }, { "epoch": 1.0338194718171068, "grad_norm": 0.5715593099594116, "learning_rate": 0.0005, "loss": 1.7721165418624878, "step": 13114 }, { "epoch": 1.0338983050847457, "grad_norm": 0.5886994004249573, "learning_rate": 0.0005, "loss": 1.80006742477417, "step": 13115 }, { "epoch": 1.0339771383523848, "grad_norm": 0.6227253079414368, "learning_rate": 0.0005, "loss": 1.7533071041107178, "step": 13116 }, { "epoch": 1.0340559716200237, "grad_norm": 0.7238116264343262, "learning_rate": 0.0005, "loss": 1.826775312423706, "step": 13117 }, { "epoch": 1.0341348048876626, "grad_norm": 0.537441611289978, "learning_rate": 0.0005, "loss": 1.764725685119629, "step": 13118 }, { "epoch": 1.0342136381553015, "grad_norm": 0.5109566450119019, "learning_rate": 0.0005, "loss": 1.7048077583312988, "step": 13119 }, { "epoch": 1.0342924714229405, "grad_norm": 0.7033576965332031, "learning_rate": 0.0005, "loss": 1.7871341705322266, "step": 13120 }, { "epoch": 1.0343713046905794, "grad_norm": 0.5835492014884949, "learning_rate": 0.0005, "loss": 1.7533001899719238, "step": 13121 }, { "epoch": 1.0344501379582183, "grad_norm": 0.6865378618240356, "learning_rate": 0.0005, "loss": 1.8465564250946045, "step": 13122 }, { "epoch": 1.0345289712258574, "grad_norm": 0.6063058972358704, "learning_rate": 0.0005, "loss": 1.583439588546753, "step": 13123 }, { "epoch": 1.0346078044934963, "grad_norm": 0.7513446807861328, "learning_rate": 0.0005, "loss": 1.6037678718566895, "step": 13124 }, { "epoch": 1.0346866377611352, "grad_norm": 0.4195346236228943, "learning_rate": 0.0005, "loss": 1.6738104820251465, "step": 13125 }, { "epoch": 1.0347654710287741, "grad_norm": 0.5933294892311096, "learning_rate": 0.0005, "loss": 1.8347134590148926, "step": 13126 }, { "epoch": 1.034844304296413, "grad_norm": 0.4838194251060486, "learning_rate": 0.0005, "loss": 1.5967974662780762, "step": 13127 }, { "epoch": 1.034923137564052, "grad_norm": 0.662904679775238, "learning_rate": 0.0005, "loss": 1.4433854818344116, "step": 13128 }, { "epoch": 1.0350019708316909, "grad_norm": 0.6465095281600952, "learning_rate": 0.0005, "loss": 1.740211009979248, "step": 13129 }, { "epoch": 1.03508080409933, "grad_norm": 0.5511054992675781, "learning_rate": 0.0005, "loss": 1.6045548915863037, "step": 13130 }, { "epoch": 1.035159637366969, "grad_norm": 0.699962854385376, "learning_rate": 0.0005, "loss": 1.7468457221984863, "step": 13131 }, { "epoch": 1.0352384706346078, "grad_norm": 0.6675087213516235, "learning_rate": 0.0005, "loss": 1.7526109218597412, "step": 13132 }, { "epoch": 1.0353173039022467, "grad_norm": 0.7153405547142029, "learning_rate": 0.0005, "loss": 1.8446629047393799, "step": 13133 }, { "epoch": 1.0353961371698857, "grad_norm": 0.6140467524528503, "learning_rate": 0.0005, "loss": 1.6963491439819336, "step": 13134 }, { "epoch": 1.0354749704375246, "grad_norm": 0.59980309009552, "learning_rate": 0.0005, "loss": 2.020505428314209, "step": 13135 }, { "epoch": 1.0355538037051635, "grad_norm": 0.5983839631080627, "learning_rate": 0.0005, "loss": 1.6234246492385864, "step": 13136 }, { "epoch": 1.0356326369728026, "grad_norm": 0.8157930970191956, "learning_rate": 0.0005, "loss": 1.6361920833587646, "step": 13137 }, { "epoch": 1.0357114702404415, "grad_norm": 0.5110871195793152, "learning_rate": 0.0005, "loss": 1.6589175462722778, "step": 13138 }, { "epoch": 1.0357903035080804, "grad_norm": 0.7526003122329712, "learning_rate": 0.0005, "loss": 1.8508849143981934, "step": 13139 }, { "epoch": 1.0358691367757193, "grad_norm": 0.5814292430877686, "learning_rate": 0.0005, "loss": 1.6533384323120117, "step": 13140 }, { "epoch": 1.0359479700433583, "grad_norm": 0.6874909996986389, "learning_rate": 0.0005, "loss": 1.571319341659546, "step": 13141 }, { "epoch": 1.0360268033109972, "grad_norm": 0.5306835770606995, "learning_rate": 0.0005, "loss": 1.6949764490127563, "step": 13142 }, { "epoch": 1.036105636578636, "grad_norm": 0.6471834182739258, "learning_rate": 0.0005, "loss": 1.9426242113113403, "step": 13143 }, { "epoch": 1.0361844698462752, "grad_norm": 0.5145131349563599, "learning_rate": 0.0005, "loss": 1.9046636819839478, "step": 13144 }, { "epoch": 1.0362633031139141, "grad_norm": 0.5828177332878113, "learning_rate": 0.0005, "loss": 1.8078052997589111, "step": 13145 }, { "epoch": 1.036342136381553, "grad_norm": 0.6130172610282898, "learning_rate": 0.0005, "loss": 1.9480524063110352, "step": 13146 }, { "epoch": 1.036420969649192, "grad_norm": 0.5965791940689087, "learning_rate": 0.0005, "loss": 1.7439775466918945, "step": 13147 }, { "epoch": 1.0364998029168309, "grad_norm": 0.6842475533485413, "learning_rate": 0.0005, "loss": 1.703392744064331, "step": 13148 }, { "epoch": 1.0365786361844698, "grad_norm": 0.6583107709884644, "learning_rate": 0.0005, "loss": 1.4996023178100586, "step": 13149 }, { "epoch": 1.0366574694521087, "grad_norm": 0.583888590335846, "learning_rate": 0.0005, "loss": 1.7772828340530396, "step": 13150 }, { "epoch": 1.0367363027197478, "grad_norm": 0.7679954171180725, "learning_rate": 0.0005, "loss": 1.939866304397583, "step": 13151 }, { "epoch": 1.0368151359873867, "grad_norm": 0.6298708915710449, "learning_rate": 0.0005, "loss": 1.8320637941360474, "step": 13152 }, { "epoch": 1.0368939692550256, "grad_norm": 0.6722590327262878, "learning_rate": 0.0005, "loss": 1.901940107345581, "step": 13153 }, { "epoch": 1.0369728025226645, "grad_norm": 0.5293821096420288, "learning_rate": 0.0005, "loss": 1.7689483165740967, "step": 13154 }, { "epoch": 1.0370516357903035, "grad_norm": 0.5525546669960022, "learning_rate": 0.0005, "loss": 1.8825347423553467, "step": 13155 }, { "epoch": 1.0371304690579424, "grad_norm": 0.5219606757164001, "learning_rate": 0.0005, "loss": 1.5639444589614868, "step": 13156 }, { "epoch": 1.0372093023255813, "grad_norm": 0.5719616413116455, "learning_rate": 0.0005, "loss": 1.8576557636260986, "step": 13157 }, { "epoch": 1.0372881355932204, "grad_norm": 0.6172844171524048, "learning_rate": 0.0005, "loss": 1.9492013454437256, "step": 13158 }, { "epoch": 1.0373669688608593, "grad_norm": 0.6672769784927368, "learning_rate": 0.0005, "loss": 2.16316819190979, "step": 13159 }, { "epoch": 1.0374458021284982, "grad_norm": 0.5028641223907471, "learning_rate": 0.0005, "loss": 1.5882110595703125, "step": 13160 }, { "epoch": 1.0375246353961372, "grad_norm": 0.6420232057571411, "learning_rate": 0.0005, "loss": 1.7696071863174438, "step": 13161 }, { "epoch": 1.037603468663776, "grad_norm": 0.553852915763855, "learning_rate": 0.0005, "loss": 1.7309753894805908, "step": 13162 }, { "epoch": 1.037682301931415, "grad_norm": 0.5301490426063538, "learning_rate": 0.0005, "loss": 1.6686229705810547, "step": 13163 }, { "epoch": 1.037761135199054, "grad_norm": 0.46516838669776917, "learning_rate": 0.0005, "loss": 1.536070466041565, "step": 13164 }, { "epoch": 1.037839968466693, "grad_norm": 0.6281406283378601, "learning_rate": 0.0005, "loss": 1.7940990924835205, "step": 13165 }, { "epoch": 1.037918801734332, "grad_norm": 0.9108172059059143, "learning_rate": 0.0005, "loss": 1.6485002040863037, "step": 13166 }, { "epoch": 1.0379976350019708, "grad_norm": 0.6677516102790833, "learning_rate": 0.0005, "loss": 1.75741708278656, "step": 13167 }, { "epoch": 1.0380764682696098, "grad_norm": 0.6341899633407593, "learning_rate": 0.0005, "loss": 1.772462248802185, "step": 13168 }, { "epoch": 1.0381553015372487, "grad_norm": 0.4475862681865692, "learning_rate": 0.0005, "loss": 1.496077299118042, "step": 13169 }, { "epoch": 1.0382341348048876, "grad_norm": 0.5420595407485962, "learning_rate": 0.0005, "loss": 1.3450396060943604, "step": 13170 }, { "epoch": 1.0383129680725267, "grad_norm": 0.4865090250968933, "learning_rate": 0.0005, "loss": 1.599663496017456, "step": 13171 }, { "epoch": 1.0383918013401656, "grad_norm": 0.7816340327262878, "learning_rate": 0.0005, "loss": 1.7587776184082031, "step": 13172 }, { "epoch": 1.0384706346078045, "grad_norm": 0.5300713777542114, "learning_rate": 0.0005, "loss": 1.6551589965820312, "step": 13173 }, { "epoch": 1.0385494678754434, "grad_norm": 0.5270292162895203, "learning_rate": 0.0005, "loss": 1.7288568019866943, "step": 13174 }, { "epoch": 1.0386283011430824, "grad_norm": 0.5608175992965698, "learning_rate": 0.0005, "loss": 1.8087078332901, "step": 13175 }, { "epoch": 1.0387071344107213, "grad_norm": 0.7511230707168579, "learning_rate": 0.0005, "loss": 1.813761830329895, "step": 13176 }, { "epoch": 1.0387859676783602, "grad_norm": 0.597070574760437, "learning_rate": 0.0005, "loss": 1.813324213027954, "step": 13177 }, { "epoch": 1.0388648009459993, "grad_norm": 0.5563604831695557, "learning_rate": 0.0005, "loss": 1.6476314067840576, "step": 13178 }, { "epoch": 1.0389436342136382, "grad_norm": 0.633111298084259, "learning_rate": 0.0005, "loss": 1.9397484064102173, "step": 13179 }, { "epoch": 1.0390224674812771, "grad_norm": 0.5292466282844543, "learning_rate": 0.0005, "loss": 1.6706956624984741, "step": 13180 }, { "epoch": 1.039101300748916, "grad_norm": 0.6341202259063721, "learning_rate": 0.0005, "loss": 1.7990102767944336, "step": 13181 }, { "epoch": 1.039180134016555, "grad_norm": 0.6736276149749756, "learning_rate": 0.0005, "loss": 1.6647701263427734, "step": 13182 }, { "epoch": 1.0392589672841939, "grad_norm": 0.5919846296310425, "learning_rate": 0.0005, "loss": 1.7047979831695557, "step": 13183 }, { "epoch": 1.0393378005518328, "grad_norm": 0.6172394752502441, "learning_rate": 0.0005, "loss": 1.6527049541473389, "step": 13184 }, { "epoch": 1.039416633819472, "grad_norm": 0.5814763903617859, "learning_rate": 0.0005, "loss": 1.7059446573257446, "step": 13185 }, { "epoch": 1.0394954670871108, "grad_norm": 0.5527219176292419, "learning_rate": 0.0005, "loss": 1.700402855873108, "step": 13186 }, { "epoch": 1.0395743003547497, "grad_norm": 0.4966641366481781, "learning_rate": 0.0005, "loss": 1.598196029663086, "step": 13187 }, { "epoch": 1.0396531336223886, "grad_norm": 0.6521430015563965, "learning_rate": 0.0005, "loss": 1.5480334758758545, "step": 13188 }, { "epoch": 1.0397319668900276, "grad_norm": 0.42045459151268005, "learning_rate": 0.0005, "loss": 1.5292017459869385, "step": 13189 }, { "epoch": 1.0398108001576665, "grad_norm": 0.6989414095878601, "learning_rate": 0.0005, "loss": 1.663482666015625, "step": 13190 }, { "epoch": 1.0398896334253054, "grad_norm": 0.7521051168441772, "learning_rate": 0.0005, "loss": 1.9703974723815918, "step": 13191 }, { "epoch": 1.0399684666929445, "grad_norm": 0.5576485395431519, "learning_rate": 0.0005, "loss": 1.9060978889465332, "step": 13192 }, { "epoch": 1.0400472999605834, "grad_norm": 0.5682029724121094, "learning_rate": 0.0005, "loss": 1.780043601989746, "step": 13193 }, { "epoch": 1.0401261332282223, "grad_norm": 0.5274089574813843, "learning_rate": 0.0005, "loss": 1.45737624168396, "step": 13194 }, { "epoch": 1.0402049664958612, "grad_norm": 0.5639030933380127, "learning_rate": 0.0005, "loss": 1.354828119277954, "step": 13195 }, { "epoch": 1.0402837997635002, "grad_norm": 0.6412230134010315, "learning_rate": 0.0005, "loss": 1.5910909175872803, "step": 13196 }, { "epoch": 1.040362633031139, "grad_norm": 0.5576053261756897, "learning_rate": 0.0005, "loss": 1.496435284614563, "step": 13197 }, { "epoch": 1.040441466298778, "grad_norm": 0.5738696455955505, "learning_rate": 0.0005, "loss": 1.624240756034851, "step": 13198 }, { "epoch": 1.0405202995664171, "grad_norm": 0.5495827794075012, "learning_rate": 0.0005, "loss": 1.5308243036270142, "step": 13199 }, { "epoch": 1.040599132834056, "grad_norm": 0.5564860105514526, "learning_rate": 0.0005, "loss": 1.8374812602996826, "step": 13200 }, { "epoch": 1.040677966101695, "grad_norm": 0.5869741439819336, "learning_rate": 0.0005, "loss": 1.7675695419311523, "step": 13201 }, { "epoch": 1.0407567993693339, "grad_norm": 0.6040613055229187, "learning_rate": 0.0005, "loss": 1.8487050533294678, "step": 13202 }, { "epoch": 1.0408356326369728, "grad_norm": 0.6778818964958191, "learning_rate": 0.0005, "loss": 1.756797432899475, "step": 13203 }, { "epoch": 1.0409144659046117, "grad_norm": 0.4985543191432953, "learning_rate": 0.0005, "loss": 1.7496247291564941, "step": 13204 }, { "epoch": 1.0409932991722506, "grad_norm": 0.6299685835838318, "learning_rate": 0.0005, "loss": 1.506786823272705, "step": 13205 }, { "epoch": 1.0410721324398897, "grad_norm": 0.5641809105873108, "learning_rate": 0.0005, "loss": 1.592700481414795, "step": 13206 }, { "epoch": 1.0411509657075286, "grad_norm": 0.7950090169906616, "learning_rate": 0.0005, "loss": 1.7967363595962524, "step": 13207 }, { "epoch": 1.0412297989751675, "grad_norm": 0.5532976984977722, "learning_rate": 0.0005, "loss": 1.9675703048706055, "step": 13208 }, { "epoch": 1.0413086322428065, "grad_norm": 0.5414875745773315, "learning_rate": 0.0005, "loss": 1.6571156978607178, "step": 13209 }, { "epoch": 1.0413874655104454, "grad_norm": 0.6252766251564026, "learning_rate": 0.0005, "loss": 1.5729036331176758, "step": 13210 }, { "epoch": 1.0414662987780843, "grad_norm": 0.591647744178772, "learning_rate": 0.0005, "loss": 1.61418879032135, "step": 13211 }, { "epoch": 1.0415451320457234, "grad_norm": 0.5453519225120544, "learning_rate": 0.0005, "loss": 1.7243239879608154, "step": 13212 }, { "epoch": 1.0416239653133623, "grad_norm": 0.5705095529556274, "learning_rate": 0.0005, "loss": 1.7050633430480957, "step": 13213 }, { "epoch": 1.0417027985810012, "grad_norm": 0.5287531018257141, "learning_rate": 0.0005, "loss": 1.8263546228408813, "step": 13214 }, { "epoch": 1.0417816318486401, "grad_norm": 0.7812236547470093, "learning_rate": 0.0005, "loss": 1.6455552577972412, "step": 13215 }, { "epoch": 1.041860465116279, "grad_norm": 0.4504851698875427, "learning_rate": 0.0005, "loss": 1.5945937633514404, "step": 13216 }, { "epoch": 1.041939298383918, "grad_norm": 0.6396479606628418, "learning_rate": 0.0005, "loss": 1.7237406969070435, "step": 13217 }, { "epoch": 1.0420181316515569, "grad_norm": 0.5806246995925903, "learning_rate": 0.0005, "loss": 1.711165189743042, "step": 13218 }, { "epoch": 1.042096964919196, "grad_norm": 0.7132092118263245, "learning_rate": 0.0005, "loss": 1.456195592880249, "step": 13219 }, { "epoch": 1.042175798186835, "grad_norm": 0.5866623520851135, "learning_rate": 0.0005, "loss": 1.7910971641540527, "step": 13220 }, { "epoch": 1.0422546314544738, "grad_norm": 0.5786961317062378, "learning_rate": 0.0005, "loss": 2.0232906341552734, "step": 13221 }, { "epoch": 1.0423334647221127, "grad_norm": 0.528862714767456, "learning_rate": 0.0005, "loss": 1.653158187866211, "step": 13222 }, { "epoch": 1.0424122979897517, "grad_norm": 0.6138260960578918, "learning_rate": 0.0005, "loss": 1.7387771606445312, "step": 13223 }, { "epoch": 1.0424911312573906, "grad_norm": 1.5966718196868896, "learning_rate": 0.0005, "loss": 1.676106572151184, "step": 13224 }, { "epoch": 1.0425699645250295, "grad_norm": 0.5691272616386414, "learning_rate": 0.0005, "loss": 1.6467088460922241, "step": 13225 }, { "epoch": 1.0426487977926686, "grad_norm": 0.6270583868026733, "learning_rate": 0.0005, "loss": 1.673252820968628, "step": 13226 }, { "epoch": 1.0427276310603075, "grad_norm": 0.5540144443511963, "learning_rate": 0.0005, "loss": 1.7643518447875977, "step": 13227 }, { "epoch": 1.0428064643279464, "grad_norm": 0.6037807464599609, "learning_rate": 0.0005, "loss": 1.882033109664917, "step": 13228 }, { "epoch": 1.0428852975955853, "grad_norm": 0.44268712401390076, "learning_rate": 0.0005, "loss": 1.4517059326171875, "step": 13229 }, { "epoch": 1.0429641308632243, "grad_norm": 0.533217191696167, "learning_rate": 0.0005, "loss": 1.6651694774627686, "step": 13230 }, { "epoch": 1.0430429641308632, "grad_norm": 0.589100182056427, "learning_rate": 0.0005, "loss": 1.9062485694885254, "step": 13231 }, { "epoch": 1.043121797398502, "grad_norm": 0.5458572506904602, "learning_rate": 0.0005, "loss": 1.6886383295059204, "step": 13232 }, { "epoch": 1.0432006306661412, "grad_norm": 0.43537572026252747, "learning_rate": 0.0005, "loss": 1.3611518144607544, "step": 13233 }, { "epoch": 1.0432794639337801, "grad_norm": 0.570300817489624, "learning_rate": 0.0005, "loss": 1.6357629299163818, "step": 13234 }, { "epoch": 1.043358297201419, "grad_norm": 0.6038179397583008, "learning_rate": 0.0005, "loss": 1.5969277620315552, "step": 13235 }, { "epoch": 1.043437130469058, "grad_norm": 0.7116793394088745, "learning_rate": 0.0005, "loss": 1.7952353954315186, "step": 13236 }, { "epoch": 1.0435159637366969, "grad_norm": 0.525528609752655, "learning_rate": 0.0005, "loss": 1.5895991325378418, "step": 13237 }, { "epoch": 1.0435947970043358, "grad_norm": 0.5426629185676575, "learning_rate": 0.0005, "loss": 2.071113109588623, "step": 13238 }, { "epoch": 1.0436736302719747, "grad_norm": 0.5638309121131897, "learning_rate": 0.0005, "loss": 1.53562593460083, "step": 13239 }, { "epoch": 1.0437524635396138, "grad_norm": 0.5193949341773987, "learning_rate": 0.0005, "loss": 1.605208396911621, "step": 13240 }, { "epoch": 1.0438312968072527, "grad_norm": 0.676393449306488, "learning_rate": 0.0005, "loss": 1.9016461372375488, "step": 13241 }, { "epoch": 1.0439101300748916, "grad_norm": 0.6794238686561584, "learning_rate": 0.0005, "loss": 1.8703341484069824, "step": 13242 }, { "epoch": 1.0439889633425306, "grad_norm": 0.47310304641723633, "learning_rate": 0.0005, "loss": 1.5549403429031372, "step": 13243 }, { "epoch": 1.0440677966101695, "grad_norm": 0.5051347613334656, "learning_rate": 0.0005, "loss": 1.4844543933868408, "step": 13244 }, { "epoch": 1.0441466298778084, "grad_norm": 0.5364636182785034, "learning_rate": 0.0005, "loss": 1.898083209991455, "step": 13245 }, { "epoch": 1.0442254631454473, "grad_norm": 0.5586405396461487, "learning_rate": 0.0005, "loss": 1.6556689739227295, "step": 13246 }, { "epoch": 1.0443042964130864, "grad_norm": 0.4912817180156708, "learning_rate": 0.0005, "loss": 1.7472586631774902, "step": 13247 }, { "epoch": 1.0443831296807253, "grad_norm": 0.5990015268325806, "learning_rate": 0.0005, "loss": 1.699564814567566, "step": 13248 }, { "epoch": 1.0444619629483642, "grad_norm": 0.5161863565444946, "learning_rate": 0.0005, "loss": 1.5573127269744873, "step": 13249 }, { "epoch": 1.0445407962160032, "grad_norm": 0.7245948314666748, "learning_rate": 0.0005, "loss": 1.7827250957489014, "step": 13250 }, { "epoch": 1.044619629483642, "grad_norm": 0.5310068130493164, "learning_rate": 0.0005, "loss": 1.637559413909912, "step": 13251 }, { "epoch": 1.044698462751281, "grad_norm": 0.7415225505828857, "learning_rate": 0.0005, "loss": 1.8463410139083862, "step": 13252 }, { "epoch": 1.0447772960189199, "grad_norm": 0.7980534434318542, "learning_rate": 0.0005, "loss": 1.669128656387329, "step": 13253 }, { "epoch": 1.044856129286559, "grad_norm": 0.6116705536842346, "learning_rate": 0.0005, "loss": 1.7433444261550903, "step": 13254 }, { "epoch": 1.044934962554198, "grad_norm": 0.4575263559818268, "learning_rate": 0.0005, "loss": 1.6101471185684204, "step": 13255 }, { "epoch": 1.0450137958218368, "grad_norm": 0.5693133473396301, "learning_rate": 0.0005, "loss": 1.6685750484466553, "step": 13256 }, { "epoch": 1.0450926290894758, "grad_norm": 0.5045129656791687, "learning_rate": 0.0005, "loss": 1.539527177810669, "step": 13257 }, { "epoch": 1.0451714623571147, "grad_norm": 0.5002594590187073, "learning_rate": 0.0005, "loss": 1.6880154609680176, "step": 13258 }, { "epoch": 1.0452502956247536, "grad_norm": 0.548330545425415, "learning_rate": 0.0005, "loss": 1.7129846811294556, "step": 13259 }, { "epoch": 1.0453291288923925, "grad_norm": 0.564075767993927, "learning_rate": 0.0005, "loss": 1.8212809562683105, "step": 13260 }, { "epoch": 1.0454079621600316, "grad_norm": 0.5266327857971191, "learning_rate": 0.0005, "loss": 1.7632980346679688, "step": 13261 }, { "epoch": 1.0454867954276705, "grad_norm": 0.5845702886581421, "learning_rate": 0.0005, "loss": 1.9958449602127075, "step": 13262 }, { "epoch": 1.0455656286953094, "grad_norm": 1.9375925064086914, "learning_rate": 0.0005, "loss": 1.7607389688491821, "step": 13263 }, { "epoch": 1.0456444619629484, "grad_norm": 0.5667065382003784, "learning_rate": 0.0005, "loss": 1.4728152751922607, "step": 13264 }, { "epoch": 1.0457232952305873, "grad_norm": 0.4961667060852051, "learning_rate": 0.0005, "loss": 1.5551068782806396, "step": 13265 }, { "epoch": 1.0458021284982262, "grad_norm": 0.5239778161048889, "learning_rate": 0.0005, "loss": 1.6074388027191162, "step": 13266 }, { "epoch": 1.045880961765865, "grad_norm": 0.5803062319755554, "learning_rate": 0.0005, "loss": 1.5630438327789307, "step": 13267 }, { "epoch": 1.0459597950335042, "grad_norm": 0.564570426940918, "learning_rate": 0.0005, "loss": 1.617147445678711, "step": 13268 }, { "epoch": 1.0460386283011431, "grad_norm": 0.6342430710792542, "learning_rate": 0.0005, "loss": 1.7067736387252808, "step": 13269 }, { "epoch": 1.046117461568782, "grad_norm": 1.0071128606796265, "learning_rate": 0.0005, "loss": 1.776200532913208, "step": 13270 }, { "epoch": 1.046196294836421, "grad_norm": 0.5725826025009155, "learning_rate": 0.0005, "loss": 1.9448895454406738, "step": 13271 }, { "epoch": 1.0462751281040599, "grad_norm": 0.685160756111145, "learning_rate": 0.0005, "loss": 2.0109972953796387, "step": 13272 }, { "epoch": 1.0463539613716988, "grad_norm": 0.7908613681793213, "learning_rate": 0.0005, "loss": 1.9595284461975098, "step": 13273 }, { "epoch": 1.0464327946393377, "grad_norm": 0.5522803068161011, "learning_rate": 0.0005, "loss": 1.8078056573867798, "step": 13274 }, { "epoch": 1.0465116279069768, "grad_norm": 0.4466158151626587, "learning_rate": 0.0005, "loss": 1.5473922491073608, "step": 13275 }, { "epoch": 1.0465904611746157, "grad_norm": 0.4830526113510132, "learning_rate": 0.0005, "loss": 1.554047703742981, "step": 13276 }, { "epoch": 1.0466692944422546, "grad_norm": 0.570507287979126, "learning_rate": 0.0005, "loss": 1.7376782894134521, "step": 13277 }, { "epoch": 1.0467481277098936, "grad_norm": 0.5136055946350098, "learning_rate": 0.0005, "loss": 1.6755061149597168, "step": 13278 }, { "epoch": 1.0468269609775325, "grad_norm": 0.5632473826408386, "learning_rate": 0.0005, "loss": 1.629929542541504, "step": 13279 }, { "epoch": 1.0469057942451714, "grad_norm": 0.6309940218925476, "learning_rate": 0.0005, "loss": 1.6907005310058594, "step": 13280 }, { "epoch": 1.0469846275128103, "grad_norm": 0.5447558164596558, "learning_rate": 0.0005, "loss": 1.9642893075942993, "step": 13281 }, { "epoch": 1.0470634607804494, "grad_norm": 0.47284916043281555, "learning_rate": 0.0005, "loss": 1.521153211593628, "step": 13282 }, { "epoch": 1.0471422940480883, "grad_norm": 0.6072766780853271, "learning_rate": 0.0005, "loss": 1.8082218170166016, "step": 13283 }, { "epoch": 1.0472211273157273, "grad_norm": 0.4191284775733948, "learning_rate": 0.0005, "loss": 1.5933557748794556, "step": 13284 }, { "epoch": 1.0472999605833662, "grad_norm": 0.5126131176948547, "learning_rate": 0.0005, "loss": 1.6573851108551025, "step": 13285 }, { "epoch": 1.047378793851005, "grad_norm": 0.5462143421173096, "learning_rate": 0.0005, "loss": 1.9115030765533447, "step": 13286 }, { "epoch": 1.047457627118644, "grad_norm": 0.5434253215789795, "learning_rate": 0.0005, "loss": 1.551012396812439, "step": 13287 }, { "epoch": 1.047536460386283, "grad_norm": 0.4823831617832184, "learning_rate": 0.0005, "loss": 1.6014268398284912, "step": 13288 }, { "epoch": 1.047615293653922, "grad_norm": 0.48880434036254883, "learning_rate": 0.0005, "loss": 1.6437311172485352, "step": 13289 }, { "epoch": 1.047694126921561, "grad_norm": 0.4797038733959198, "learning_rate": 0.0005, "loss": 1.655632495880127, "step": 13290 }, { "epoch": 1.0477729601891999, "grad_norm": 0.6458621025085449, "learning_rate": 0.0005, "loss": 1.7974934577941895, "step": 13291 }, { "epoch": 1.0478517934568388, "grad_norm": 0.4956298768520355, "learning_rate": 0.0005, "loss": 1.763427734375, "step": 13292 }, { "epoch": 1.0479306267244777, "grad_norm": 0.4990523159503937, "learning_rate": 0.0005, "loss": 1.696333408355713, "step": 13293 }, { "epoch": 1.0480094599921166, "grad_norm": 0.6435927748680115, "learning_rate": 0.0005, "loss": 1.7143380641937256, "step": 13294 }, { "epoch": 1.0480882932597557, "grad_norm": 0.4785163104534149, "learning_rate": 0.0005, "loss": 1.6902213096618652, "step": 13295 }, { "epoch": 1.0481671265273946, "grad_norm": 0.48560941219329834, "learning_rate": 0.0005, "loss": 1.6234294176101685, "step": 13296 }, { "epoch": 1.0482459597950335, "grad_norm": 0.5022635459899902, "learning_rate": 0.0005, "loss": 1.588093638420105, "step": 13297 }, { "epoch": 1.0483247930626725, "grad_norm": 0.5216768383979797, "learning_rate": 0.0005, "loss": 1.4766943454742432, "step": 13298 }, { "epoch": 1.0484036263303114, "grad_norm": 0.5080408453941345, "learning_rate": 0.0005, "loss": 1.667608618736267, "step": 13299 }, { "epoch": 1.0484824595979503, "grad_norm": 0.658304750919342, "learning_rate": 0.0005, "loss": 1.7174403667449951, "step": 13300 }, { "epoch": 1.0485612928655892, "grad_norm": 0.49398866295814514, "learning_rate": 0.0005, "loss": 1.6528913974761963, "step": 13301 }, { "epoch": 1.0486401261332283, "grad_norm": 0.621850311756134, "learning_rate": 0.0005, "loss": 1.6600013971328735, "step": 13302 }, { "epoch": 1.0487189594008672, "grad_norm": 0.5182613134384155, "learning_rate": 0.0005, "loss": 1.5901302099227905, "step": 13303 }, { "epoch": 1.0487977926685061, "grad_norm": 0.5313428640365601, "learning_rate": 0.0005, "loss": 1.7374556064605713, "step": 13304 }, { "epoch": 1.048876625936145, "grad_norm": 0.5145926475524902, "learning_rate": 0.0005, "loss": 1.5447094440460205, "step": 13305 }, { "epoch": 1.048955459203784, "grad_norm": 0.5121238231658936, "learning_rate": 0.0005, "loss": 1.840120553970337, "step": 13306 }, { "epoch": 1.0490342924714229, "grad_norm": 0.47162720561027527, "learning_rate": 0.0005, "loss": 1.5358508825302124, "step": 13307 }, { "epoch": 1.0491131257390618, "grad_norm": 0.7250525951385498, "learning_rate": 0.0005, "loss": 1.6603446006774902, "step": 13308 }, { "epoch": 1.049191959006701, "grad_norm": 0.4611268639564514, "learning_rate": 0.0005, "loss": 1.6233309507369995, "step": 13309 }, { "epoch": 1.0492707922743398, "grad_norm": 0.5826339721679688, "learning_rate": 0.0005, "loss": 1.7394574880599976, "step": 13310 }, { "epoch": 1.0493496255419787, "grad_norm": 0.6952164173126221, "learning_rate": 0.0005, "loss": 1.7620913982391357, "step": 13311 }, { "epoch": 1.0494284588096177, "grad_norm": 0.5000464916229248, "learning_rate": 0.0005, "loss": 1.5698556900024414, "step": 13312 }, { "epoch": 1.0495072920772566, "grad_norm": 0.5068764686584473, "learning_rate": 0.0005, "loss": 1.7735934257507324, "step": 13313 }, { "epoch": 1.0495861253448955, "grad_norm": 0.5453595519065857, "learning_rate": 0.0005, "loss": 1.5998493432998657, "step": 13314 }, { "epoch": 1.0496649586125344, "grad_norm": 0.5197749733924866, "learning_rate": 0.0005, "loss": 1.810964822769165, "step": 13315 }, { "epoch": 1.0497437918801735, "grad_norm": 0.4146251678466797, "learning_rate": 0.0005, "loss": 1.5712710618972778, "step": 13316 }, { "epoch": 1.0498226251478124, "grad_norm": 0.5286675095558167, "learning_rate": 0.0005, "loss": 1.8104541301727295, "step": 13317 }, { "epoch": 1.0499014584154513, "grad_norm": 0.5246681571006775, "learning_rate": 0.0005, "loss": 1.7867088317871094, "step": 13318 }, { "epoch": 1.0499802916830903, "grad_norm": 0.5918377041816711, "learning_rate": 0.0005, "loss": 1.7941087484359741, "step": 13319 }, { "epoch": 1.0500591249507292, "grad_norm": 0.5014297366142273, "learning_rate": 0.0005, "loss": 1.6019587516784668, "step": 13320 }, { "epoch": 1.050137958218368, "grad_norm": 0.6102501749992371, "learning_rate": 0.0005, "loss": 1.7515010833740234, "step": 13321 }, { "epoch": 1.050216791486007, "grad_norm": 0.6099991202354431, "learning_rate": 0.0005, "loss": 1.5308747291564941, "step": 13322 }, { "epoch": 1.0502956247536461, "grad_norm": 0.5607995390892029, "learning_rate": 0.0005, "loss": 1.9483649730682373, "step": 13323 }, { "epoch": 1.050374458021285, "grad_norm": 0.6344029307365417, "learning_rate": 0.0005, "loss": 1.8074331283569336, "step": 13324 }, { "epoch": 1.050453291288924, "grad_norm": 1.5285512208938599, "learning_rate": 0.0005, "loss": 1.608553171157837, "step": 13325 }, { "epoch": 1.0505321245565629, "grad_norm": 0.5757984519004822, "learning_rate": 0.0005, "loss": 1.5517653226852417, "step": 13326 }, { "epoch": 1.0506109578242018, "grad_norm": 0.5605365633964539, "learning_rate": 0.0005, "loss": 1.9047081470489502, "step": 13327 }, { "epoch": 1.0506897910918407, "grad_norm": 0.5986006259918213, "learning_rate": 0.0005, "loss": 1.853549599647522, "step": 13328 }, { "epoch": 1.0507686243594796, "grad_norm": 0.5457937121391296, "learning_rate": 0.0005, "loss": 1.8691837787628174, "step": 13329 }, { "epoch": 1.0508474576271187, "grad_norm": 0.409486323595047, "learning_rate": 0.0005, "loss": 1.5572113990783691, "step": 13330 }, { "epoch": 1.0509262908947576, "grad_norm": 0.5310108661651611, "learning_rate": 0.0005, "loss": 1.6014569997787476, "step": 13331 }, { "epoch": 1.0510051241623966, "grad_norm": 0.5291109085083008, "learning_rate": 0.0005, "loss": 1.577406644821167, "step": 13332 }, { "epoch": 1.0510839574300355, "grad_norm": 0.65732342004776, "learning_rate": 0.0005, "loss": 1.7192445993423462, "step": 13333 }, { "epoch": 1.0511627906976744, "grad_norm": 0.5479196310043335, "learning_rate": 0.0005, "loss": 1.6357841491699219, "step": 13334 }, { "epoch": 1.0512416239653133, "grad_norm": 0.5393666625022888, "learning_rate": 0.0005, "loss": 1.6446834802627563, "step": 13335 }, { "epoch": 1.0513204572329522, "grad_norm": 0.4875918924808502, "learning_rate": 0.0005, "loss": 1.4941213130950928, "step": 13336 }, { "epoch": 1.0513992905005913, "grad_norm": 0.5559951663017273, "learning_rate": 0.0005, "loss": 1.9119192361831665, "step": 13337 }, { "epoch": 1.0514781237682302, "grad_norm": 0.5199026465415955, "learning_rate": 0.0005, "loss": 1.670724868774414, "step": 13338 }, { "epoch": 1.0515569570358692, "grad_norm": 0.5881038904190063, "learning_rate": 0.0005, "loss": 1.5767550468444824, "step": 13339 }, { "epoch": 1.051635790303508, "grad_norm": 1.0243425369262695, "learning_rate": 0.0005, "loss": 1.5642359256744385, "step": 13340 }, { "epoch": 1.051714623571147, "grad_norm": 0.6175920367240906, "learning_rate": 0.0005, "loss": 1.6749684810638428, "step": 13341 }, { "epoch": 1.051793456838786, "grad_norm": 0.6553532481193542, "learning_rate": 0.0005, "loss": 2.074028253555298, "step": 13342 }, { "epoch": 1.051872290106425, "grad_norm": 1.3089629411697388, "learning_rate": 0.0005, "loss": 1.6363211870193481, "step": 13343 }, { "epoch": 1.051951123374064, "grad_norm": 0.5255804657936096, "learning_rate": 0.0005, "loss": 1.780484676361084, "step": 13344 }, { "epoch": 1.0520299566417028, "grad_norm": 0.7153865098953247, "learning_rate": 0.0005, "loss": 2.1315441131591797, "step": 13345 }, { "epoch": 1.0521087899093418, "grad_norm": 0.642914891242981, "learning_rate": 0.0005, "loss": 1.6563074588775635, "step": 13346 }, { "epoch": 1.0521876231769807, "grad_norm": 0.854539155960083, "learning_rate": 0.0005, "loss": 1.6902530193328857, "step": 13347 }, { "epoch": 1.0522664564446196, "grad_norm": 0.5967704057693481, "learning_rate": 0.0005, "loss": 1.656790852546692, "step": 13348 }, { "epoch": 1.0523452897122585, "grad_norm": 0.5847855806350708, "learning_rate": 0.0005, "loss": 1.8536975383758545, "step": 13349 }, { "epoch": 1.0524241229798976, "grad_norm": 0.5589514374732971, "learning_rate": 0.0005, "loss": 1.5901694297790527, "step": 13350 }, { "epoch": 1.0525029562475365, "grad_norm": 0.531482994556427, "learning_rate": 0.0005, "loss": 1.756924033164978, "step": 13351 }, { "epoch": 1.0525817895151754, "grad_norm": 0.49980178475379944, "learning_rate": 0.0005, "loss": 1.8490324020385742, "step": 13352 }, { "epoch": 1.0526606227828144, "grad_norm": 0.6432878971099854, "learning_rate": 0.0005, "loss": 1.8372676372528076, "step": 13353 }, { "epoch": 1.0527394560504533, "grad_norm": 0.4864765703678131, "learning_rate": 0.0005, "loss": 1.6826964616775513, "step": 13354 }, { "epoch": 1.0528182893180922, "grad_norm": 0.6117452383041382, "learning_rate": 0.0005, "loss": 1.7223365306854248, "step": 13355 }, { "epoch": 1.052897122585731, "grad_norm": 0.8836718797683716, "learning_rate": 0.0005, "loss": 1.8171368837356567, "step": 13356 }, { "epoch": 1.0529759558533702, "grad_norm": 0.6338281631469727, "learning_rate": 0.0005, "loss": 1.7074519395828247, "step": 13357 }, { "epoch": 1.0530547891210091, "grad_norm": 0.7489244341850281, "learning_rate": 0.0005, "loss": 1.6490956544876099, "step": 13358 }, { "epoch": 1.053133622388648, "grad_norm": 0.556447446346283, "learning_rate": 0.0005, "loss": 1.6157646179199219, "step": 13359 }, { "epoch": 1.053212455656287, "grad_norm": 0.7292153239250183, "learning_rate": 0.0005, "loss": 2.042405366897583, "step": 13360 }, { "epoch": 1.0532912889239259, "grad_norm": 0.47497469186782837, "learning_rate": 0.0005, "loss": 1.5423190593719482, "step": 13361 }, { "epoch": 1.0533701221915648, "grad_norm": 0.9285709857940674, "learning_rate": 0.0005, "loss": 1.6283527612686157, "step": 13362 }, { "epoch": 1.0534489554592037, "grad_norm": 0.5062013864517212, "learning_rate": 0.0005, "loss": 1.6092052459716797, "step": 13363 }, { "epoch": 1.0535277887268428, "grad_norm": 0.4915492534637451, "learning_rate": 0.0005, "loss": 1.6876015663146973, "step": 13364 }, { "epoch": 1.0536066219944817, "grad_norm": 0.800880491733551, "learning_rate": 0.0005, "loss": 1.7982574701309204, "step": 13365 }, { "epoch": 1.0536854552621207, "grad_norm": 0.6111340522766113, "learning_rate": 0.0005, "loss": 1.757390022277832, "step": 13366 }, { "epoch": 1.0537642885297596, "grad_norm": 0.5229191780090332, "learning_rate": 0.0005, "loss": 1.5633268356323242, "step": 13367 }, { "epoch": 1.0538431217973985, "grad_norm": 0.7275505065917969, "learning_rate": 0.0005, "loss": 1.8483788967132568, "step": 13368 }, { "epoch": 1.0539219550650374, "grad_norm": 0.6574406623840332, "learning_rate": 0.0005, "loss": 1.915914535522461, "step": 13369 }, { "epoch": 1.0540007883326763, "grad_norm": 0.5341038107872009, "learning_rate": 0.0005, "loss": 1.7986321449279785, "step": 13370 }, { "epoch": 1.0540796216003154, "grad_norm": 0.511225700378418, "learning_rate": 0.0005, "loss": 1.765920639038086, "step": 13371 }, { "epoch": 1.0541584548679543, "grad_norm": 0.6592176556587219, "learning_rate": 0.0005, "loss": 1.9275058507919312, "step": 13372 }, { "epoch": 1.0542372881355933, "grad_norm": 0.633665919303894, "learning_rate": 0.0005, "loss": 1.9837089776992798, "step": 13373 }, { "epoch": 1.0543161214032322, "grad_norm": 0.6421405076980591, "learning_rate": 0.0005, "loss": 1.7520445585250854, "step": 13374 }, { "epoch": 1.054394954670871, "grad_norm": 0.6086478233337402, "learning_rate": 0.0005, "loss": 1.6927415132522583, "step": 13375 }, { "epoch": 1.05447378793851, "grad_norm": 0.6355472207069397, "learning_rate": 0.0005, "loss": 1.602257490158081, "step": 13376 }, { "epoch": 1.054552621206149, "grad_norm": 0.6044941544532776, "learning_rate": 0.0005, "loss": 1.6858104467391968, "step": 13377 }, { "epoch": 1.054631454473788, "grad_norm": 0.4920119047164917, "learning_rate": 0.0005, "loss": 1.610876441001892, "step": 13378 }, { "epoch": 1.054710287741427, "grad_norm": 0.46967577934265137, "learning_rate": 0.0005, "loss": 1.511326551437378, "step": 13379 }, { "epoch": 1.0547891210090659, "grad_norm": 0.541896641254425, "learning_rate": 0.0005, "loss": 1.6055281162261963, "step": 13380 }, { "epoch": 1.0548679542767048, "grad_norm": 0.6500384211540222, "learning_rate": 0.0005, "loss": 1.7742102146148682, "step": 13381 }, { "epoch": 1.0549467875443437, "grad_norm": 0.5243141055107117, "learning_rate": 0.0005, "loss": 1.7013294696807861, "step": 13382 }, { "epoch": 1.0550256208119826, "grad_norm": 0.6080385446548462, "learning_rate": 0.0005, "loss": 1.5967904329299927, "step": 13383 }, { "epoch": 1.0551044540796215, "grad_norm": 0.788105845451355, "learning_rate": 0.0005, "loss": 1.9496872425079346, "step": 13384 }, { "epoch": 1.0551832873472606, "grad_norm": 0.5939327478408813, "learning_rate": 0.0005, "loss": 1.668226718902588, "step": 13385 }, { "epoch": 1.0552621206148995, "grad_norm": 0.6039755940437317, "learning_rate": 0.0005, "loss": 1.5813335180282593, "step": 13386 }, { "epoch": 1.0553409538825385, "grad_norm": 0.5321123003959656, "learning_rate": 0.0005, "loss": 1.6257017850875854, "step": 13387 }, { "epoch": 1.0554197871501774, "grad_norm": 0.7715131640434265, "learning_rate": 0.0005, "loss": 1.6436090469360352, "step": 13388 }, { "epoch": 1.0554986204178163, "grad_norm": 0.6184648871421814, "learning_rate": 0.0005, "loss": 1.7882177829742432, "step": 13389 }, { "epoch": 1.0555774536854552, "grad_norm": 0.6683810949325562, "learning_rate": 0.0005, "loss": 1.515042781829834, "step": 13390 }, { "epoch": 1.055656286953094, "grad_norm": 0.5370205044746399, "learning_rate": 0.0005, "loss": 1.6013636589050293, "step": 13391 }, { "epoch": 1.0557351202207332, "grad_norm": 0.5620766878128052, "learning_rate": 0.0005, "loss": 1.5580902099609375, "step": 13392 }, { "epoch": 1.0558139534883721, "grad_norm": 0.558558464050293, "learning_rate": 0.0005, "loss": 1.814183235168457, "step": 13393 }, { "epoch": 1.055892786756011, "grad_norm": 0.514988362789154, "learning_rate": 0.0005, "loss": 1.7923904657363892, "step": 13394 }, { "epoch": 1.05597162002365, "grad_norm": 0.5821977853775024, "learning_rate": 0.0005, "loss": 1.7894093990325928, "step": 13395 }, { "epoch": 1.0560504532912889, "grad_norm": 0.5341989398002625, "learning_rate": 0.0005, "loss": 1.6733100414276123, "step": 13396 }, { "epoch": 1.0561292865589278, "grad_norm": 0.48730331659317017, "learning_rate": 0.0005, "loss": 1.7142548561096191, "step": 13397 }, { "epoch": 1.0562081198265667, "grad_norm": 0.5824174284934998, "learning_rate": 0.0005, "loss": 1.4540542364120483, "step": 13398 }, { "epoch": 1.0562869530942058, "grad_norm": 0.5432743430137634, "learning_rate": 0.0005, "loss": 1.5953515768051147, "step": 13399 }, { "epoch": 1.0563657863618447, "grad_norm": 0.5675034523010254, "learning_rate": 0.0005, "loss": 1.5905836820602417, "step": 13400 }, { "epoch": 1.0564446196294837, "grad_norm": 0.5885457396507263, "learning_rate": 0.0005, "loss": 1.6913915872573853, "step": 13401 }, { "epoch": 1.0565234528971226, "grad_norm": 0.6688988208770752, "learning_rate": 0.0005, "loss": 1.6323115825653076, "step": 13402 }, { "epoch": 1.0566022861647615, "grad_norm": 0.520232617855072, "learning_rate": 0.0005, "loss": 1.565029501914978, "step": 13403 }, { "epoch": 1.0566811194324004, "grad_norm": 0.5008198022842407, "learning_rate": 0.0005, "loss": 1.707648515701294, "step": 13404 }, { "epoch": 1.0567599527000393, "grad_norm": 1.135636329650879, "learning_rate": 0.0005, "loss": 1.5643854141235352, "step": 13405 }, { "epoch": 1.0568387859676784, "grad_norm": 0.46677425503730774, "learning_rate": 0.0005, "loss": 1.4090995788574219, "step": 13406 }, { "epoch": 1.0569176192353174, "grad_norm": 0.48672839999198914, "learning_rate": 0.0005, "loss": 1.548403263092041, "step": 13407 }, { "epoch": 1.0569964525029563, "grad_norm": 0.544252872467041, "learning_rate": 0.0005, "loss": 1.5919872522354126, "step": 13408 }, { "epoch": 1.0570752857705952, "grad_norm": 0.5721133351325989, "learning_rate": 0.0005, "loss": 1.823323130607605, "step": 13409 }, { "epoch": 1.057154119038234, "grad_norm": 0.46875089406967163, "learning_rate": 0.0005, "loss": 1.5892505645751953, "step": 13410 }, { "epoch": 1.057232952305873, "grad_norm": 0.5161632299423218, "learning_rate": 0.0005, "loss": 1.6598597764968872, "step": 13411 }, { "epoch": 1.057311785573512, "grad_norm": 0.6137933135032654, "learning_rate": 0.0005, "loss": 1.694608449935913, "step": 13412 }, { "epoch": 1.057390618841151, "grad_norm": 0.5797489881515503, "learning_rate": 0.0005, "loss": 1.7890578508377075, "step": 13413 }, { "epoch": 1.05746945210879, "grad_norm": 0.585590660572052, "learning_rate": 0.0005, "loss": 1.6540943384170532, "step": 13414 }, { "epoch": 1.0575482853764289, "grad_norm": 0.5634309649467468, "learning_rate": 0.0005, "loss": 1.597744107246399, "step": 13415 }, { "epoch": 1.0576271186440678, "grad_norm": 0.6250518560409546, "learning_rate": 0.0005, "loss": 1.6631956100463867, "step": 13416 }, { "epoch": 1.0577059519117067, "grad_norm": 0.5144924521446228, "learning_rate": 0.0005, "loss": 1.6858829259872437, "step": 13417 }, { "epoch": 1.0577847851793456, "grad_norm": 0.44921380281448364, "learning_rate": 0.0005, "loss": 1.5633883476257324, "step": 13418 }, { "epoch": 1.0578636184469847, "grad_norm": 0.5122542977333069, "learning_rate": 0.0005, "loss": 1.6086056232452393, "step": 13419 }, { "epoch": 1.0579424517146236, "grad_norm": 0.5746581554412842, "learning_rate": 0.0005, "loss": 1.7003605365753174, "step": 13420 }, { "epoch": 1.0580212849822626, "grad_norm": 1.5580883026123047, "learning_rate": 0.0005, "loss": 1.6510801315307617, "step": 13421 }, { "epoch": 1.0581001182499015, "grad_norm": 0.5238170623779297, "learning_rate": 0.0005, "loss": 1.7385107278823853, "step": 13422 }, { "epoch": 1.0581789515175404, "grad_norm": 0.5807369351387024, "learning_rate": 0.0005, "loss": 1.9575001001358032, "step": 13423 }, { "epoch": 1.0582577847851793, "grad_norm": 0.5404176115989685, "learning_rate": 0.0005, "loss": 1.549396276473999, "step": 13424 }, { "epoch": 1.0583366180528182, "grad_norm": 0.6421748399734497, "learning_rate": 0.0005, "loss": 1.7594339847564697, "step": 13425 }, { "epoch": 1.0584154513204573, "grad_norm": 0.6488949060440063, "learning_rate": 0.0005, "loss": 1.582692265510559, "step": 13426 }, { "epoch": 1.0584942845880962, "grad_norm": 0.4792097508907318, "learning_rate": 0.0005, "loss": 1.4599418640136719, "step": 13427 }, { "epoch": 1.0585731178557352, "grad_norm": 0.5984530448913574, "learning_rate": 0.0005, "loss": 1.7383661270141602, "step": 13428 }, { "epoch": 1.058651951123374, "grad_norm": 0.5739016532897949, "learning_rate": 0.0005, "loss": 1.743638277053833, "step": 13429 }, { "epoch": 1.058730784391013, "grad_norm": 0.903240442276001, "learning_rate": 0.0005, "loss": 1.6500000953674316, "step": 13430 }, { "epoch": 1.058809617658652, "grad_norm": 0.5674660801887512, "learning_rate": 0.0005, "loss": 1.8158297538757324, "step": 13431 }, { "epoch": 1.0588884509262908, "grad_norm": 0.5474959015846252, "learning_rate": 0.0005, "loss": 1.6554423570632935, "step": 13432 }, { "epoch": 1.05896728419393, "grad_norm": 0.734148383140564, "learning_rate": 0.0005, "loss": 1.6390833854675293, "step": 13433 }, { "epoch": 1.0590461174615688, "grad_norm": 0.6160300970077515, "learning_rate": 0.0005, "loss": 1.671439290046692, "step": 13434 }, { "epoch": 1.0591249507292078, "grad_norm": 0.7401288151741028, "learning_rate": 0.0005, "loss": 1.8891963958740234, "step": 13435 }, { "epoch": 1.0592037839968467, "grad_norm": 0.5308393836021423, "learning_rate": 0.0005, "loss": 1.899192452430725, "step": 13436 }, { "epoch": 1.0592826172644856, "grad_norm": 0.5482363700866699, "learning_rate": 0.0005, "loss": 1.8122320175170898, "step": 13437 }, { "epoch": 1.0593614505321245, "grad_norm": 0.5391284823417664, "learning_rate": 0.0005, "loss": 1.9414609670639038, "step": 13438 }, { "epoch": 1.0594402837997634, "grad_norm": 4.494253158569336, "learning_rate": 0.0005, "loss": 1.728804349899292, "step": 13439 }, { "epoch": 1.0595191170674025, "grad_norm": 0.6380442976951599, "learning_rate": 0.0005, "loss": 1.554333209991455, "step": 13440 }, { "epoch": 1.0595979503350414, "grad_norm": 0.4821554124355316, "learning_rate": 0.0005, "loss": 1.6779463291168213, "step": 13441 }, { "epoch": 1.0596767836026804, "grad_norm": 0.5350962281227112, "learning_rate": 0.0005, "loss": 1.8715739250183105, "step": 13442 }, { "epoch": 1.0597556168703193, "grad_norm": 0.9884201288223267, "learning_rate": 0.0005, "loss": 1.5951347351074219, "step": 13443 }, { "epoch": 1.0598344501379582, "grad_norm": 0.5609971880912781, "learning_rate": 0.0005, "loss": 1.9316298961639404, "step": 13444 }, { "epoch": 1.059913283405597, "grad_norm": 0.6750773787498474, "learning_rate": 0.0005, "loss": 1.8188531398773193, "step": 13445 }, { "epoch": 1.059992116673236, "grad_norm": 0.5921214818954468, "learning_rate": 0.0005, "loss": 1.5942230224609375, "step": 13446 }, { "epoch": 1.0600709499408751, "grad_norm": 0.7849864959716797, "learning_rate": 0.0005, "loss": 1.717246651649475, "step": 13447 }, { "epoch": 1.060149783208514, "grad_norm": 0.6008761525154114, "learning_rate": 0.0005, "loss": 1.706974744796753, "step": 13448 }, { "epoch": 1.060228616476153, "grad_norm": 0.782253623008728, "learning_rate": 0.0005, "loss": 1.9202346801757812, "step": 13449 }, { "epoch": 1.0603074497437919, "grad_norm": 0.5101867318153381, "learning_rate": 0.0005, "loss": 1.726175308227539, "step": 13450 }, { "epoch": 1.0603862830114308, "grad_norm": 0.6300123929977417, "learning_rate": 0.0005, "loss": 1.7717962265014648, "step": 13451 }, { "epoch": 1.0604651162790697, "grad_norm": 0.46467944979667664, "learning_rate": 0.0005, "loss": 1.530075192451477, "step": 13452 }, { "epoch": 1.0605439495467086, "grad_norm": 0.5101496577262878, "learning_rate": 0.0005, "loss": 1.655624508857727, "step": 13453 }, { "epoch": 1.0606227828143477, "grad_norm": 0.6745526194572449, "learning_rate": 0.0005, "loss": 1.6377615928649902, "step": 13454 }, { "epoch": 1.0607016160819867, "grad_norm": 0.5697922706604004, "learning_rate": 0.0005, "loss": 1.756420612335205, "step": 13455 }, { "epoch": 1.0607804493496256, "grad_norm": 0.5077976584434509, "learning_rate": 0.0005, "loss": 1.5407016277313232, "step": 13456 }, { "epoch": 1.0608592826172645, "grad_norm": 0.5575256943702698, "learning_rate": 0.0005, "loss": 1.776059865951538, "step": 13457 }, { "epoch": 1.0609381158849034, "grad_norm": 0.5716343522071838, "learning_rate": 0.0005, "loss": 1.9512499570846558, "step": 13458 }, { "epoch": 1.0610169491525423, "grad_norm": 0.6293484568595886, "learning_rate": 0.0005, "loss": 1.7520544528961182, "step": 13459 }, { "epoch": 1.0610957824201812, "grad_norm": 0.5935997366905212, "learning_rate": 0.0005, "loss": 1.7168176174163818, "step": 13460 }, { "epoch": 1.0611746156878203, "grad_norm": 0.55080246925354, "learning_rate": 0.0005, "loss": 1.5972857475280762, "step": 13461 }, { "epoch": 1.0612534489554593, "grad_norm": 0.5268682241439819, "learning_rate": 0.0005, "loss": 1.4481279850006104, "step": 13462 }, { "epoch": 1.0613322822230982, "grad_norm": 0.4798765182495117, "learning_rate": 0.0005, "loss": 1.5853054523468018, "step": 13463 }, { "epoch": 1.061411115490737, "grad_norm": 0.49785903096199036, "learning_rate": 0.0005, "loss": 1.6283215284347534, "step": 13464 }, { "epoch": 1.061489948758376, "grad_norm": 0.6116218566894531, "learning_rate": 0.0005, "loss": 1.752484679222107, "step": 13465 }, { "epoch": 1.061568782026015, "grad_norm": 0.540626585483551, "learning_rate": 0.0005, "loss": 1.9387526512145996, "step": 13466 }, { "epoch": 1.061647615293654, "grad_norm": 0.6296195983886719, "learning_rate": 0.0005, "loss": 1.6973278522491455, "step": 13467 }, { "epoch": 1.061726448561293, "grad_norm": 0.5342658758163452, "learning_rate": 0.0005, "loss": 1.7683991193771362, "step": 13468 }, { "epoch": 1.0618052818289319, "grad_norm": 0.8796437978744507, "learning_rate": 0.0005, "loss": 1.8546690940856934, "step": 13469 }, { "epoch": 1.0618841150965708, "grad_norm": 0.516319751739502, "learning_rate": 0.0005, "loss": 1.7220512628555298, "step": 13470 }, { "epoch": 1.0619629483642097, "grad_norm": 0.5088516473770142, "learning_rate": 0.0005, "loss": 1.3984266519546509, "step": 13471 }, { "epoch": 1.0620417816318486, "grad_norm": 0.6257545351982117, "learning_rate": 0.0005, "loss": 1.6942546367645264, "step": 13472 }, { "epoch": 1.0621206148994875, "grad_norm": 0.5182271599769592, "learning_rate": 0.0005, "loss": 1.7306993007659912, "step": 13473 }, { "epoch": 1.0621994481671266, "grad_norm": 0.45859819650650024, "learning_rate": 0.0005, "loss": 1.6769814491271973, "step": 13474 }, { "epoch": 1.0622782814347655, "grad_norm": 0.5119176506996155, "learning_rate": 0.0005, "loss": 1.6527471542358398, "step": 13475 }, { "epoch": 1.0623571147024045, "grad_norm": 0.5330623388290405, "learning_rate": 0.0005, "loss": 1.8394179344177246, "step": 13476 }, { "epoch": 1.0624359479700434, "grad_norm": 0.7160366773605347, "learning_rate": 0.0005, "loss": 1.585710048675537, "step": 13477 }, { "epoch": 1.0625147812376823, "grad_norm": 0.5631256699562073, "learning_rate": 0.0005, "loss": 1.7013570070266724, "step": 13478 }, { "epoch": 1.0625936145053212, "grad_norm": 0.5550729632377625, "learning_rate": 0.0005, "loss": 1.598585605621338, "step": 13479 }, { "epoch": 1.06267244777296, "grad_norm": 0.6369578838348389, "learning_rate": 0.0005, "loss": 1.8238826990127563, "step": 13480 }, { "epoch": 1.0627512810405992, "grad_norm": 0.533954918384552, "learning_rate": 0.0005, "loss": 1.7520103454589844, "step": 13481 }, { "epoch": 1.0628301143082381, "grad_norm": 0.567753791809082, "learning_rate": 0.0005, "loss": 1.5791072845458984, "step": 13482 }, { "epoch": 1.062908947575877, "grad_norm": 0.7497128844261169, "learning_rate": 0.0005, "loss": 1.688634991645813, "step": 13483 }, { "epoch": 1.062987780843516, "grad_norm": 0.517972469329834, "learning_rate": 0.0005, "loss": 1.5684542655944824, "step": 13484 }, { "epoch": 1.0630666141111549, "grad_norm": 0.5197839140892029, "learning_rate": 0.0005, "loss": 1.6377241611480713, "step": 13485 }, { "epoch": 1.0631454473787938, "grad_norm": 0.6008304357528687, "learning_rate": 0.0005, "loss": 1.522895336151123, "step": 13486 }, { "epoch": 1.0632242806464327, "grad_norm": 0.6184144616127014, "learning_rate": 0.0005, "loss": 1.91493558883667, "step": 13487 }, { "epoch": 1.0633031139140718, "grad_norm": 0.5387498736381531, "learning_rate": 0.0005, "loss": 1.6469557285308838, "step": 13488 }, { "epoch": 1.0633819471817108, "grad_norm": 0.48853498697280884, "learning_rate": 0.0005, "loss": 1.4724738597869873, "step": 13489 }, { "epoch": 1.0634607804493497, "grad_norm": 0.5482469201087952, "learning_rate": 0.0005, "loss": 1.6428415775299072, "step": 13490 }, { "epoch": 1.0635396137169886, "grad_norm": 0.5891803503036499, "learning_rate": 0.0005, "loss": 1.4801521301269531, "step": 13491 }, { "epoch": 1.0636184469846275, "grad_norm": 0.8630189895629883, "learning_rate": 0.0005, "loss": 1.86670982837677, "step": 13492 }, { "epoch": 1.0636972802522664, "grad_norm": 0.5122170448303223, "learning_rate": 0.0005, "loss": 1.5724949836730957, "step": 13493 }, { "epoch": 1.0637761135199053, "grad_norm": 0.5015780925750732, "learning_rate": 0.0005, "loss": 1.7413196563720703, "step": 13494 }, { "epoch": 1.0638549467875444, "grad_norm": 0.4768752157688141, "learning_rate": 0.0005, "loss": 1.7176909446716309, "step": 13495 }, { "epoch": 1.0639337800551834, "grad_norm": 0.5434554219245911, "learning_rate": 0.0005, "loss": 1.7959909439086914, "step": 13496 }, { "epoch": 1.0640126133228223, "grad_norm": 0.5069762468338013, "learning_rate": 0.0005, "loss": 1.40602707862854, "step": 13497 }, { "epoch": 1.0640914465904612, "grad_norm": 0.585979700088501, "learning_rate": 0.0005, "loss": 1.6599249839782715, "step": 13498 }, { "epoch": 1.0641702798581, "grad_norm": 0.7826510071754456, "learning_rate": 0.0005, "loss": 1.8295496702194214, "step": 13499 }, { "epoch": 1.064249113125739, "grad_norm": 0.5566679239273071, "learning_rate": 0.0005, "loss": 1.7496256828308105, "step": 13500 }, { "epoch": 1.064327946393378, "grad_norm": 0.5472040176391602, "learning_rate": 0.0005, "loss": 1.6342352628707886, "step": 13501 }, { "epoch": 1.064406779661017, "grad_norm": 0.6772241592407227, "learning_rate": 0.0005, "loss": 1.9788538217544556, "step": 13502 }, { "epoch": 1.064485612928656, "grad_norm": 0.6155622601509094, "learning_rate": 0.0005, "loss": 1.5987638235092163, "step": 13503 }, { "epoch": 1.0645644461962949, "grad_norm": 0.5372052192687988, "learning_rate": 0.0005, "loss": 1.7514173984527588, "step": 13504 }, { "epoch": 1.0646432794639338, "grad_norm": 0.6720326542854309, "learning_rate": 0.0005, "loss": 1.9381604194641113, "step": 13505 }, { "epoch": 1.0647221127315727, "grad_norm": 0.48746374249458313, "learning_rate": 0.0005, "loss": 1.6254478693008423, "step": 13506 }, { "epoch": 1.0648009459992116, "grad_norm": 0.5759307742118835, "learning_rate": 0.0005, "loss": 2.0280966758728027, "step": 13507 }, { "epoch": 1.0648797792668505, "grad_norm": 0.5334249138832092, "learning_rate": 0.0005, "loss": 1.8863763809204102, "step": 13508 }, { "epoch": 1.0649586125344896, "grad_norm": 0.5513377785682678, "learning_rate": 0.0005, "loss": 1.6539278030395508, "step": 13509 }, { "epoch": 1.0650374458021286, "grad_norm": 0.5331655740737915, "learning_rate": 0.0005, "loss": 1.7808698415756226, "step": 13510 }, { "epoch": 1.0651162790697675, "grad_norm": 0.4605885446071625, "learning_rate": 0.0005, "loss": 1.6640326976776123, "step": 13511 }, { "epoch": 1.0651951123374064, "grad_norm": 0.48351359367370605, "learning_rate": 0.0005, "loss": 1.7986016273498535, "step": 13512 }, { "epoch": 1.0652739456050453, "grad_norm": 0.6019052267074585, "learning_rate": 0.0005, "loss": 1.7247872352600098, "step": 13513 }, { "epoch": 1.0653527788726842, "grad_norm": 0.5613637566566467, "learning_rate": 0.0005, "loss": 1.74908447265625, "step": 13514 }, { "epoch": 1.0654316121403231, "grad_norm": 0.5597377419471741, "learning_rate": 0.0005, "loss": 1.9334933757781982, "step": 13515 }, { "epoch": 1.0655104454079622, "grad_norm": 0.553067147731781, "learning_rate": 0.0005, "loss": 1.5857857465744019, "step": 13516 }, { "epoch": 1.0655892786756012, "grad_norm": 0.5050188302993774, "learning_rate": 0.0005, "loss": 1.5026239156723022, "step": 13517 }, { "epoch": 1.06566811194324, "grad_norm": 0.42550384998321533, "learning_rate": 0.0005, "loss": 1.6008095741271973, "step": 13518 }, { "epoch": 1.065746945210879, "grad_norm": 0.5438492298126221, "learning_rate": 0.0005, "loss": 1.6443498134613037, "step": 13519 }, { "epoch": 1.065825778478518, "grad_norm": 0.5332252383232117, "learning_rate": 0.0005, "loss": 1.621553659439087, "step": 13520 }, { "epoch": 1.0659046117461568, "grad_norm": 0.5618301630020142, "learning_rate": 0.0005, "loss": 1.617440104484558, "step": 13521 }, { "epoch": 1.0659834450137957, "grad_norm": 0.5451810956001282, "learning_rate": 0.0005, "loss": 1.5626628398895264, "step": 13522 }, { "epoch": 1.0660622782814348, "grad_norm": 0.5257537364959717, "learning_rate": 0.0005, "loss": 1.5889922380447388, "step": 13523 }, { "epoch": 1.0661411115490738, "grad_norm": 0.6052573919296265, "learning_rate": 0.0005, "loss": 1.8615360260009766, "step": 13524 }, { "epoch": 1.0662199448167127, "grad_norm": 0.5487276315689087, "learning_rate": 0.0005, "loss": 1.5371739864349365, "step": 13525 }, { "epoch": 1.0662987780843516, "grad_norm": 0.626615583896637, "learning_rate": 0.0005, "loss": 1.668562889099121, "step": 13526 }, { "epoch": 1.0663776113519905, "grad_norm": 0.6352393627166748, "learning_rate": 0.0005, "loss": 1.5702093839645386, "step": 13527 }, { "epoch": 1.0664564446196294, "grad_norm": 0.7128547430038452, "learning_rate": 0.0005, "loss": 1.912441611289978, "step": 13528 }, { "epoch": 1.0665352778872683, "grad_norm": 0.5766035914421082, "learning_rate": 0.0005, "loss": 1.6916706562042236, "step": 13529 }, { "epoch": 1.0666141111549075, "grad_norm": 0.5401145815849304, "learning_rate": 0.0005, "loss": 1.7576186656951904, "step": 13530 }, { "epoch": 1.0666929444225464, "grad_norm": 50.44214630126953, "learning_rate": 0.0005, "loss": 1.736607551574707, "step": 13531 }, { "epoch": 1.0667717776901853, "grad_norm": 0.5641458630561829, "learning_rate": 0.0005, "loss": 1.5559526681900024, "step": 13532 }, { "epoch": 1.0668506109578242, "grad_norm": 0.5189776420593262, "learning_rate": 0.0005, "loss": 1.3616342544555664, "step": 13533 }, { "epoch": 1.066929444225463, "grad_norm": 0.5623258352279663, "learning_rate": 0.0005, "loss": 1.7282378673553467, "step": 13534 }, { "epoch": 1.067008277493102, "grad_norm": 0.6173083782196045, "learning_rate": 0.0005, "loss": 1.6807539463043213, "step": 13535 }, { "epoch": 1.067087110760741, "grad_norm": 0.5247048139572144, "learning_rate": 0.0005, "loss": 1.4742319583892822, "step": 13536 }, { "epoch": 1.06716594402838, "grad_norm": 0.5186307430267334, "learning_rate": 0.0005, "loss": 1.8050222396850586, "step": 13537 }, { "epoch": 1.067244777296019, "grad_norm": 1.1046290397644043, "learning_rate": 0.0005, "loss": 1.825463056564331, "step": 13538 }, { "epoch": 1.0673236105636579, "grad_norm": 0.6752568483352661, "learning_rate": 0.0005, "loss": 1.7545597553253174, "step": 13539 }, { "epoch": 1.0674024438312968, "grad_norm": 0.6534644961357117, "learning_rate": 0.0005, "loss": 1.7396743297576904, "step": 13540 }, { "epoch": 1.0674812770989357, "grad_norm": 0.6130356788635254, "learning_rate": 0.0005, "loss": 1.6639362573623657, "step": 13541 }, { "epoch": 1.0675601103665746, "grad_norm": 0.6297160387039185, "learning_rate": 0.0005, "loss": 1.5711588859558105, "step": 13542 }, { "epoch": 1.0676389436342135, "grad_norm": 0.8286089301109314, "learning_rate": 0.0005, "loss": 1.777134895324707, "step": 13543 }, { "epoch": 1.0677177769018527, "grad_norm": 0.5482945442199707, "learning_rate": 0.0005, "loss": 1.5764660835266113, "step": 13544 }, { "epoch": 1.0677966101694916, "grad_norm": 0.6047588586807251, "learning_rate": 0.0005, "loss": 1.8906205892562866, "step": 13545 }, { "epoch": 1.0678754434371305, "grad_norm": 0.5916316509246826, "learning_rate": 0.0005, "loss": 1.9678184986114502, "step": 13546 }, { "epoch": 1.0679542767047694, "grad_norm": 0.5348604321479797, "learning_rate": 0.0005, "loss": 1.7544045448303223, "step": 13547 }, { "epoch": 1.0680331099724083, "grad_norm": 0.673819899559021, "learning_rate": 0.0005, "loss": 1.5893023014068604, "step": 13548 }, { "epoch": 1.0681119432400472, "grad_norm": 0.5505779385566711, "learning_rate": 0.0005, "loss": 1.698699712753296, "step": 13549 }, { "epoch": 1.0681907765076863, "grad_norm": 0.5606470704078674, "learning_rate": 0.0005, "loss": 1.9459706544876099, "step": 13550 }, { "epoch": 1.0682696097753253, "grad_norm": 0.5200503468513489, "learning_rate": 0.0005, "loss": 1.7255754470825195, "step": 13551 }, { "epoch": 1.0683484430429642, "grad_norm": 0.5742192268371582, "learning_rate": 0.0005, "loss": 1.8702040910720825, "step": 13552 }, { "epoch": 1.068427276310603, "grad_norm": 0.5890740752220154, "learning_rate": 0.0005, "loss": 1.7296384572982788, "step": 13553 }, { "epoch": 1.068506109578242, "grad_norm": 0.5747410655021667, "learning_rate": 0.0005, "loss": 1.6050753593444824, "step": 13554 }, { "epoch": 1.068584942845881, "grad_norm": 0.5538831949234009, "learning_rate": 0.0005, "loss": 1.6038007736206055, "step": 13555 }, { "epoch": 1.0686637761135198, "grad_norm": 0.5228196978569031, "learning_rate": 0.0005, "loss": 1.7479166984558105, "step": 13556 }, { "epoch": 1.068742609381159, "grad_norm": 0.4814293086528778, "learning_rate": 0.0005, "loss": 1.6846373081207275, "step": 13557 }, { "epoch": 1.0688214426487979, "grad_norm": 0.4992999732494354, "learning_rate": 0.0005, "loss": 1.8338191509246826, "step": 13558 }, { "epoch": 1.0689002759164368, "grad_norm": 0.5699171423912048, "learning_rate": 0.0005, "loss": 1.718122959136963, "step": 13559 }, { "epoch": 1.0689791091840757, "grad_norm": 0.5265280604362488, "learning_rate": 0.0005, "loss": 1.6370620727539062, "step": 13560 }, { "epoch": 1.0690579424517146, "grad_norm": 0.7110734581947327, "learning_rate": 0.0005, "loss": 1.9090864658355713, "step": 13561 }, { "epoch": 1.0691367757193535, "grad_norm": 0.555332362651825, "learning_rate": 0.0005, "loss": 1.8649401664733887, "step": 13562 }, { "epoch": 1.0692156089869924, "grad_norm": 0.5541017651557922, "learning_rate": 0.0005, "loss": 1.7626264095306396, "step": 13563 }, { "epoch": 1.0692944422546315, "grad_norm": 0.5657234787940979, "learning_rate": 0.0005, "loss": 1.881622314453125, "step": 13564 }, { "epoch": 1.0693732755222705, "grad_norm": 0.46917566657066345, "learning_rate": 0.0005, "loss": 1.3524856567382812, "step": 13565 }, { "epoch": 1.0694521087899094, "grad_norm": 0.5633916258811951, "learning_rate": 0.0005, "loss": 1.7601673603057861, "step": 13566 }, { "epoch": 1.0695309420575483, "grad_norm": 0.7548848986625671, "learning_rate": 0.0005, "loss": 1.807917833328247, "step": 13567 }, { "epoch": 1.0696097753251872, "grad_norm": 0.5209828019142151, "learning_rate": 0.0005, "loss": 1.7340748310089111, "step": 13568 }, { "epoch": 1.069688608592826, "grad_norm": 0.5251085162162781, "learning_rate": 0.0005, "loss": 1.6895980834960938, "step": 13569 }, { "epoch": 1.069767441860465, "grad_norm": 0.5742053985595703, "learning_rate": 0.0005, "loss": 1.8660951852798462, "step": 13570 }, { "epoch": 1.0698462751281042, "grad_norm": 0.5340659618377686, "learning_rate": 0.0005, "loss": 1.5182446241378784, "step": 13571 }, { "epoch": 1.069925108395743, "grad_norm": 0.5814655423164368, "learning_rate": 0.0005, "loss": 1.743844985961914, "step": 13572 }, { "epoch": 1.070003941663382, "grad_norm": 0.6059851050376892, "learning_rate": 0.0005, "loss": 1.6854329109191895, "step": 13573 }, { "epoch": 1.0700827749310209, "grad_norm": 0.6021251082420349, "learning_rate": 0.0005, "loss": 1.6887843608856201, "step": 13574 }, { "epoch": 1.0701616081986598, "grad_norm": 0.4790485203266144, "learning_rate": 0.0005, "loss": 1.507476806640625, "step": 13575 }, { "epoch": 1.0702404414662987, "grad_norm": 0.4939042627811432, "learning_rate": 0.0005, "loss": 1.5035126209259033, "step": 13576 }, { "epoch": 1.0703192747339376, "grad_norm": 0.5646658539772034, "learning_rate": 0.0005, "loss": 1.4925284385681152, "step": 13577 }, { "epoch": 1.0703981080015768, "grad_norm": 0.6452912092208862, "learning_rate": 0.0005, "loss": 1.6572632789611816, "step": 13578 }, { "epoch": 1.0704769412692157, "grad_norm": 0.6321020722389221, "learning_rate": 0.0005, "loss": 1.7930840253829956, "step": 13579 }, { "epoch": 1.0705557745368546, "grad_norm": 0.4519067704677582, "learning_rate": 0.0005, "loss": 1.341817855834961, "step": 13580 }, { "epoch": 1.0706346078044935, "grad_norm": 0.618242621421814, "learning_rate": 0.0005, "loss": 1.7476885318756104, "step": 13581 }, { "epoch": 1.0707134410721324, "grad_norm": 0.7997574210166931, "learning_rate": 0.0005, "loss": 1.6691035032272339, "step": 13582 }, { "epoch": 1.0707922743397713, "grad_norm": 0.5744231939315796, "learning_rate": 0.0005, "loss": 1.6459662914276123, "step": 13583 }, { "epoch": 1.0708711076074104, "grad_norm": 0.6679100394248962, "learning_rate": 0.0005, "loss": 1.5918208360671997, "step": 13584 }, { "epoch": 1.0709499408750494, "grad_norm": 0.5208094120025635, "learning_rate": 0.0005, "loss": 1.6694496870040894, "step": 13585 }, { "epoch": 1.0710287741426883, "grad_norm": 0.5332810282707214, "learning_rate": 0.0005, "loss": 1.8413310050964355, "step": 13586 }, { "epoch": 1.0711076074103272, "grad_norm": 0.5528799891471863, "learning_rate": 0.0005, "loss": 1.6921679973602295, "step": 13587 }, { "epoch": 1.071186440677966, "grad_norm": 0.6287891268730164, "learning_rate": 0.0005, "loss": 1.6123828887939453, "step": 13588 }, { "epoch": 1.071265273945605, "grad_norm": 0.7505897879600525, "learning_rate": 0.0005, "loss": 1.7550501823425293, "step": 13589 }, { "epoch": 1.071344107213244, "grad_norm": 0.6202657222747803, "learning_rate": 0.0005, "loss": 1.6523568630218506, "step": 13590 }, { "epoch": 1.071422940480883, "grad_norm": 0.4025111496448517, "learning_rate": 0.0005, "loss": 1.4982211589813232, "step": 13591 }, { "epoch": 1.071501773748522, "grad_norm": 0.7193833589553833, "learning_rate": 0.0005, "loss": 1.8271440267562866, "step": 13592 }, { "epoch": 1.0715806070161609, "grad_norm": 0.6018887162208557, "learning_rate": 0.0005, "loss": 1.8310163021087646, "step": 13593 }, { "epoch": 1.0716594402837998, "grad_norm": 0.4384104609489441, "learning_rate": 0.0005, "loss": 1.404446005821228, "step": 13594 }, { "epoch": 1.0717382735514387, "grad_norm": 0.45528602600097656, "learning_rate": 0.0005, "loss": 1.6453702449798584, "step": 13595 }, { "epoch": 1.0718171068190776, "grad_norm": 0.6270642876625061, "learning_rate": 0.0005, "loss": 1.518406629562378, "step": 13596 }, { "epoch": 1.0718959400867165, "grad_norm": 0.6530649065971375, "learning_rate": 0.0005, "loss": 1.7856882810592651, "step": 13597 }, { "epoch": 1.0719747733543556, "grad_norm": 0.4904923737049103, "learning_rate": 0.0005, "loss": 1.5796983242034912, "step": 13598 }, { "epoch": 1.0720536066219946, "grad_norm": 0.5629155039787292, "learning_rate": 0.0005, "loss": 1.5367224216461182, "step": 13599 }, { "epoch": 1.0721324398896335, "grad_norm": 0.5849433541297913, "learning_rate": 0.0005, "loss": 1.8337825536727905, "step": 13600 }, { "epoch": 1.0722112731572724, "grad_norm": 0.45904359221458435, "learning_rate": 0.0005, "loss": 1.4580078125, "step": 13601 }, { "epoch": 1.0722901064249113, "grad_norm": 0.5155127644538879, "learning_rate": 0.0005, "loss": 1.7936458587646484, "step": 13602 }, { "epoch": 1.0723689396925502, "grad_norm": 0.41519781947135925, "learning_rate": 0.0005, "loss": 1.5346791744232178, "step": 13603 }, { "epoch": 1.0724477729601891, "grad_norm": 0.6725053191184998, "learning_rate": 0.0005, "loss": 1.6324905157089233, "step": 13604 }, { "epoch": 1.0725266062278282, "grad_norm": 0.5937532186508179, "learning_rate": 0.0005, "loss": 1.8268611431121826, "step": 13605 }, { "epoch": 1.0726054394954672, "grad_norm": 0.581476628780365, "learning_rate": 0.0005, "loss": 1.5574809312820435, "step": 13606 }, { "epoch": 1.072684272763106, "grad_norm": 0.44822901487350464, "learning_rate": 0.0005, "loss": 1.5413551330566406, "step": 13607 }, { "epoch": 1.072763106030745, "grad_norm": 0.5082523822784424, "learning_rate": 0.0005, "loss": 1.5772755146026611, "step": 13608 }, { "epoch": 1.072841939298384, "grad_norm": 0.5329713821411133, "learning_rate": 0.0005, "loss": 1.6685681343078613, "step": 13609 }, { "epoch": 1.0729207725660228, "grad_norm": 0.602357029914856, "learning_rate": 0.0005, "loss": 1.8219600915908813, "step": 13610 }, { "epoch": 1.0729996058336617, "grad_norm": 0.583686113357544, "learning_rate": 0.0005, "loss": 1.5168371200561523, "step": 13611 }, { "epoch": 1.0730784391013009, "grad_norm": 0.6060814261436462, "learning_rate": 0.0005, "loss": 1.6038548946380615, "step": 13612 }, { "epoch": 1.0731572723689398, "grad_norm": 0.4489723742008209, "learning_rate": 0.0005, "loss": 1.6431832313537598, "step": 13613 }, { "epoch": 1.0732361056365787, "grad_norm": 0.7839450836181641, "learning_rate": 0.0005, "loss": 1.8437204360961914, "step": 13614 }, { "epoch": 1.0733149389042176, "grad_norm": 0.437735915184021, "learning_rate": 0.0005, "loss": 1.679964542388916, "step": 13615 }, { "epoch": 1.0733937721718565, "grad_norm": 0.5511022806167603, "learning_rate": 0.0005, "loss": 1.73982834815979, "step": 13616 }, { "epoch": 1.0734726054394954, "grad_norm": 0.5275190472602844, "learning_rate": 0.0005, "loss": 1.7528607845306396, "step": 13617 }, { "epoch": 1.0735514387071343, "grad_norm": 0.5623526573181152, "learning_rate": 0.0005, "loss": 1.716681957244873, "step": 13618 }, { "epoch": 1.0736302719747735, "grad_norm": 0.6513161659240723, "learning_rate": 0.0005, "loss": 1.7747626304626465, "step": 13619 }, { "epoch": 1.0737091052424124, "grad_norm": 0.639546275138855, "learning_rate": 0.0005, "loss": 1.7670905590057373, "step": 13620 }, { "epoch": 1.0737879385100513, "grad_norm": 0.6297661662101746, "learning_rate": 0.0005, "loss": 1.8650903701782227, "step": 13621 }, { "epoch": 1.0738667717776902, "grad_norm": 0.5615915656089783, "learning_rate": 0.0005, "loss": 1.711982250213623, "step": 13622 }, { "epoch": 1.073945605045329, "grad_norm": 0.6761948466300964, "learning_rate": 0.0005, "loss": 1.5658867359161377, "step": 13623 }, { "epoch": 1.074024438312968, "grad_norm": 0.47928211092948914, "learning_rate": 0.0005, "loss": 1.575456976890564, "step": 13624 }, { "epoch": 1.074103271580607, "grad_norm": 0.5550615191459656, "learning_rate": 0.0005, "loss": 1.9656821489334106, "step": 13625 }, { "epoch": 1.074182104848246, "grad_norm": 0.604586660861969, "learning_rate": 0.0005, "loss": 1.9769399166107178, "step": 13626 }, { "epoch": 1.074260938115885, "grad_norm": 0.45807552337646484, "learning_rate": 0.0005, "loss": 1.6505444049835205, "step": 13627 }, { "epoch": 1.0743397713835239, "grad_norm": 0.6850214600563049, "learning_rate": 0.0005, "loss": 1.7505435943603516, "step": 13628 }, { "epoch": 1.0744186046511628, "grad_norm": 0.47617748379707336, "learning_rate": 0.0005, "loss": 1.529272198677063, "step": 13629 }, { "epoch": 1.0744974379188017, "grad_norm": 0.746522843837738, "learning_rate": 0.0005, "loss": 2.1050338745117188, "step": 13630 }, { "epoch": 1.0745762711864406, "grad_norm": 1.4368607997894287, "learning_rate": 0.0005, "loss": 1.8372673988342285, "step": 13631 }, { "epoch": 1.0746551044540795, "grad_norm": 0.5171843767166138, "learning_rate": 0.0005, "loss": 1.6341054439544678, "step": 13632 }, { "epoch": 1.0747339377217187, "grad_norm": 0.6104775071144104, "learning_rate": 0.0005, "loss": 1.793626070022583, "step": 13633 }, { "epoch": 1.0748127709893576, "grad_norm": 0.6518983244895935, "learning_rate": 0.0005, "loss": 1.7037229537963867, "step": 13634 }, { "epoch": 1.0748916042569965, "grad_norm": 0.9449499249458313, "learning_rate": 0.0005, "loss": 1.758079171180725, "step": 13635 }, { "epoch": 1.0749704375246354, "grad_norm": 0.4818759858608246, "learning_rate": 0.0005, "loss": 1.5650057792663574, "step": 13636 }, { "epoch": 1.0750492707922743, "grad_norm": 0.5176905989646912, "learning_rate": 0.0005, "loss": 1.540034532546997, "step": 13637 }, { "epoch": 1.0751281040599132, "grad_norm": 0.6724596619606018, "learning_rate": 0.0005, "loss": 1.7137181758880615, "step": 13638 }, { "epoch": 1.0752069373275521, "grad_norm": 0.6574404239654541, "learning_rate": 0.0005, "loss": 1.6529271602630615, "step": 13639 }, { "epoch": 1.0752857705951913, "grad_norm": 0.5878896117210388, "learning_rate": 0.0005, "loss": 1.8622058629989624, "step": 13640 }, { "epoch": 1.0753646038628302, "grad_norm": 0.4444858431816101, "learning_rate": 0.0005, "loss": 1.594102144241333, "step": 13641 }, { "epoch": 1.075443437130469, "grad_norm": 0.6555593013763428, "learning_rate": 0.0005, "loss": 1.5761520862579346, "step": 13642 }, { "epoch": 1.075522270398108, "grad_norm": 0.8979542255401611, "learning_rate": 0.0005, "loss": 1.8788695335388184, "step": 13643 }, { "epoch": 1.075601103665747, "grad_norm": 0.9123912453651428, "learning_rate": 0.0005, "loss": 1.482478141784668, "step": 13644 }, { "epoch": 1.0756799369333858, "grad_norm": 0.6529093980789185, "learning_rate": 0.0005, "loss": 1.7704739570617676, "step": 13645 }, { "epoch": 1.0757587702010247, "grad_norm": 0.8790276646614075, "learning_rate": 0.0005, "loss": 1.7669233083724976, "step": 13646 }, { "epoch": 1.0758376034686639, "grad_norm": 0.5135601162910461, "learning_rate": 0.0005, "loss": 1.6444649696350098, "step": 13647 }, { "epoch": 1.0759164367363028, "grad_norm": 0.6686716079711914, "learning_rate": 0.0005, "loss": 1.924568772315979, "step": 13648 }, { "epoch": 1.0759952700039417, "grad_norm": 0.638759195804596, "learning_rate": 0.0005, "loss": 1.5379797220230103, "step": 13649 }, { "epoch": 1.0760741032715806, "grad_norm": 1.1706387996673584, "learning_rate": 0.0005, "loss": 1.8992338180541992, "step": 13650 }, { "epoch": 1.0761529365392195, "grad_norm": 0.6240952610969543, "learning_rate": 0.0005, "loss": 1.872004508972168, "step": 13651 }, { "epoch": 1.0762317698068584, "grad_norm": 0.5086461305618286, "learning_rate": 0.0005, "loss": 1.6865196228027344, "step": 13652 }, { "epoch": 1.0763106030744973, "grad_norm": 0.5216138958930969, "learning_rate": 0.0005, "loss": 1.482576847076416, "step": 13653 }, { "epoch": 1.0763894363421365, "grad_norm": 0.4933859407901764, "learning_rate": 0.0005, "loss": 1.716140627861023, "step": 13654 }, { "epoch": 1.0764682696097754, "grad_norm": 0.5554332137107849, "learning_rate": 0.0005, "loss": 1.9157099723815918, "step": 13655 }, { "epoch": 1.0765471028774143, "grad_norm": 0.4553031325340271, "learning_rate": 0.0005, "loss": 1.646812915802002, "step": 13656 }, { "epoch": 1.0766259361450532, "grad_norm": 0.5338940024375916, "learning_rate": 0.0005, "loss": 1.652827262878418, "step": 13657 }, { "epoch": 1.076704769412692, "grad_norm": 0.5001242756843567, "learning_rate": 0.0005, "loss": 1.6155611276626587, "step": 13658 }, { "epoch": 1.076783602680331, "grad_norm": 0.6168420314788818, "learning_rate": 0.0005, "loss": 1.560347080230713, "step": 13659 }, { "epoch": 1.07686243594797, "grad_norm": 0.5030704736709595, "learning_rate": 0.0005, "loss": 1.3210346698760986, "step": 13660 }, { "epoch": 1.076941269215609, "grad_norm": 0.5614242553710938, "learning_rate": 0.0005, "loss": 1.8748462200164795, "step": 13661 }, { "epoch": 1.077020102483248, "grad_norm": 0.4750452935695648, "learning_rate": 0.0005, "loss": 1.6597979068756104, "step": 13662 }, { "epoch": 1.0770989357508869, "grad_norm": 0.5258659720420837, "learning_rate": 0.0005, "loss": 1.9321277141571045, "step": 13663 }, { "epoch": 1.0771777690185258, "grad_norm": 0.5671402215957642, "learning_rate": 0.0005, "loss": 1.7647554874420166, "step": 13664 }, { "epoch": 1.0772566022861647, "grad_norm": 0.459838330745697, "learning_rate": 0.0005, "loss": 1.3904812335968018, "step": 13665 }, { "epoch": 1.0773354355538036, "grad_norm": 0.5437631607055664, "learning_rate": 0.0005, "loss": 1.6455209255218506, "step": 13666 }, { "epoch": 1.0774142688214425, "grad_norm": 0.45881882309913635, "learning_rate": 0.0005, "loss": 1.5906100273132324, "step": 13667 }, { "epoch": 1.0774931020890817, "grad_norm": 1.4002766609191895, "learning_rate": 0.0005, "loss": 1.7149873971939087, "step": 13668 }, { "epoch": 1.0775719353567206, "grad_norm": 0.5171582102775574, "learning_rate": 0.0005, "loss": 1.64145827293396, "step": 13669 }, { "epoch": 1.0776507686243595, "grad_norm": 0.5009943842887878, "learning_rate": 0.0005, "loss": 1.5660184621810913, "step": 13670 }, { "epoch": 1.0777296018919984, "grad_norm": 0.6528684496879578, "learning_rate": 0.0005, "loss": 1.889267921447754, "step": 13671 }, { "epoch": 1.0778084351596373, "grad_norm": 0.6023707985877991, "learning_rate": 0.0005, "loss": 1.7191327810287476, "step": 13672 }, { "epoch": 1.0778872684272762, "grad_norm": 0.5368633270263672, "learning_rate": 0.0005, "loss": 1.5993027687072754, "step": 13673 }, { "epoch": 1.0779661016949154, "grad_norm": 0.48767349123954773, "learning_rate": 0.0005, "loss": 1.6156978607177734, "step": 13674 }, { "epoch": 1.0780449349625543, "grad_norm": 0.8661410808563232, "learning_rate": 0.0005, "loss": 1.8227670192718506, "step": 13675 }, { "epoch": 1.0781237682301932, "grad_norm": 0.6926280856132507, "learning_rate": 0.0005, "loss": 1.570631742477417, "step": 13676 }, { "epoch": 1.078202601497832, "grad_norm": 0.8375327587127686, "learning_rate": 0.0005, "loss": 2.028978109359741, "step": 13677 }, { "epoch": 1.078281434765471, "grad_norm": 0.7055156826972961, "learning_rate": 0.0005, "loss": 1.6718521118164062, "step": 13678 }, { "epoch": 1.07836026803311, "grad_norm": 0.6233785152435303, "learning_rate": 0.0005, "loss": 1.6271072626113892, "step": 13679 }, { "epoch": 1.0784391013007488, "grad_norm": 0.5770809650421143, "learning_rate": 0.0005, "loss": 1.7376829385757446, "step": 13680 }, { "epoch": 1.078517934568388, "grad_norm": 0.7063153386116028, "learning_rate": 0.0005, "loss": 1.744520664215088, "step": 13681 }, { "epoch": 1.0785967678360269, "grad_norm": 0.5197502970695496, "learning_rate": 0.0005, "loss": 1.9262664318084717, "step": 13682 }, { "epoch": 1.0786756011036658, "grad_norm": 0.6297504305839539, "learning_rate": 0.0005, "loss": 1.9480868577957153, "step": 13683 }, { "epoch": 1.0787544343713047, "grad_norm": 0.5253872871398926, "learning_rate": 0.0005, "loss": 1.5432335138320923, "step": 13684 }, { "epoch": 1.0788332676389436, "grad_norm": 0.5504809021949768, "learning_rate": 0.0005, "loss": 1.7571243047714233, "step": 13685 }, { "epoch": 1.0789121009065825, "grad_norm": 0.5895737409591675, "learning_rate": 0.0005, "loss": 1.6339141130447388, "step": 13686 }, { "epoch": 1.0789909341742214, "grad_norm": 0.5627643465995789, "learning_rate": 0.0005, "loss": 1.661910057067871, "step": 13687 }, { "epoch": 1.0790697674418606, "grad_norm": 0.5419149994850159, "learning_rate": 0.0005, "loss": 1.4269862174987793, "step": 13688 }, { "epoch": 1.0791486007094995, "grad_norm": 0.6033362150192261, "learning_rate": 0.0005, "loss": 1.8060581684112549, "step": 13689 }, { "epoch": 1.0792274339771384, "grad_norm": 0.5444309115409851, "learning_rate": 0.0005, "loss": 1.6591635942459106, "step": 13690 }, { "epoch": 1.0793062672447773, "grad_norm": 0.5560516715049744, "learning_rate": 0.0005, "loss": 1.7141406536102295, "step": 13691 }, { "epoch": 1.0793851005124162, "grad_norm": 0.6868231892585754, "learning_rate": 0.0005, "loss": 1.6224693059921265, "step": 13692 }, { "epoch": 1.0794639337800551, "grad_norm": 0.5224133729934692, "learning_rate": 0.0005, "loss": 1.5469729900360107, "step": 13693 }, { "epoch": 1.079542767047694, "grad_norm": 0.5774931311607361, "learning_rate": 0.0005, "loss": 1.8448067903518677, "step": 13694 }, { "epoch": 1.0796216003153332, "grad_norm": 0.6675964593887329, "learning_rate": 0.0005, "loss": 1.6901973485946655, "step": 13695 }, { "epoch": 1.079700433582972, "grad_norm": 0.6120389103889465, "learning_rate": 0.0005, "loss": 1.7824854850769043, "step": 13696 }, { "epoch": 1.079779266850611, "grad_norm": 0.6651740074157715, "learning_rate": 0.0005, "loss": 1.7284908294677734, "step": 13697 }, { "epoch": 1.07985810011825, "grad_norm": 0.6779504418373108, "learning_rate": 0.0005, "loss": 1.875638723373413, "step": 13698 }, { "epoch": 1.0799369333858888, "grad_norm": 0.6190659999847412, "learning_rate": 0.0005, "loss": 1.6725082397460938, "step": 13699 }, { "epoch": 1.0800157666535277, "grad_norm": 0.7397318482398987, "learning_rate": 0.0005, "loss": 1.6776458024978638, "step": 13700 }, { "epoch": 1.0800945999211666, "grad_norm": 0.7470884919166565, "learning_rate": 0.0005, "loss": 1.733129620552063, "step": 13701 }, { "epoch": 1.0801734331888058, "grad_norm": 0.5298351049423218, "learning_rate": 0.0005, "loss": 1.8979803323745728, "step": 13702 }, { "epoch": 1.0802522664564447, "grad_norm": 0.5825279951095581, "learning_rate": 0.0005, "loss": 1.930649995803833, "step": 13703 }, { "epoch": 1.0803310997240836, "grad_norm": 0.5367222428321838, "learning_rate": 0.0005, "loss": 1.7590105533599854, "step": 13704 }, { "epoch": 1.0804099329917225, "grad_norm": 0.5130618214607239, "learning_rate": 0.0005, "loss": 2.019212245941162, "step": 13705 }, { "epoch": 1.0804887662593614, "grad_norm": 0.5214599370956421, "learning_rate": 0.0005, "loss": 1.7231251001358032, "step": 13706 }, { "epoch": 1.0805675995270003, "grad_norm": 0.547233521938324, "learning_rate": 0.0005, "loss": 1.7146763801574707, "step": 13707 }, { "epoch": 1.0806464327946395, "grad_norm": 0.5925426483154297, "learning_rate": 0.0005, "loss": 1.884345293045044, "step": 13708 }, { "epoch": 1.0807252660622784, "grad_norm": 0.4670857787132263, "learning_rate": 0.0005, "loss": 1.485282301902771, "step": 13709 }, { "epoch": 1.0808040993299173, "grad_norm": 0.559419572353363, "learning_rate": 0.0005, "loss": 1.8374930620193481, "step": 13710 }, { "epoch": 1.0808829325975562, "grad_norm": 0.5113118290901184, "learning_rate": 0.0005, "loss": 1.5136610269546509, "step": 13711 }, { "epoch": 1.080961765865195, "grad_norm": 0.6153789758682251, "learning_rate": 0.0005, "loss": 1.803213357925415, "step": 13712 }, { "epoch": 1.081040599132834, "grad_norm": 0.5498733520507812, "learning_rate": 0.0005, "loss": 1.7293262481689453, "step": 13713 }, { "epoch": 1.081119432400473, "grad_norm": 0.5725846290588379, "learning_rate": 0.0005, "loss": 1.7374427318572998, "step": 13714 }, { "epoch": 1.081198265668112, "grad_norm": 0.652224063873291, "learning_rate": 0.0005, "loss": 2.0162174701690674, "step": 13715 }, { "epoch": 1.081277098935751, "grad_norm": 2.1501638889312744, "learning_rate": 0.0005, "loss": 1.9327985048294067, "step": 13716 }, { "epoch": 1.0813559322033899, "grad_norm": 0.7531702518463135, "learning_rate": 0.0005, "loss": 1.4654422998428345, "step": 13717 }, { "epoch": 1.0814347654710288, "grad_norm": 0.5481653213500977, "learning_rate": 0.0005, "loss": 1.7098867893218994, "step": 13718 }, { "epoch": 1.0815135987386677, "grad_norm": 0.7586999535560608, "learning_rate": 0.0005, "loss": 1.7124273777008057, "step": 13719 }, { "epoch": 1.0815924320063066, "grad_norm": 0.8862889409065247, "learning_rate": 0.0005, "loss": 2.047236919403076, "step": 13720 }, { "epoch": 1.0816712652739455, "grad_norm": 0.46335017681121826, "learning_rate": 0.0005, "loss": 1.3881652355194092, "step": 13721 }, { "epoch": 1.0817500985415847, "grad_norm": 0.5121864676475525, "learning_rate": 0.0005, "loss": 1.4509742259979248, "step": 13722 }, { "epoch": 1.0818289318092236, "grad_norm": 0.6152992844581604, "learning_rate": 0.0005, "loss": 1.8958661556243896, "step": 13723 }, { "epoch": 1.0819077650768625, "grad_norm": 0.6074045300483704, "learning_rate": 0.0005, "loss": 1.4977076053619385, "step": 13724 }, { "epoch": 1.0819865983445014, "grad_norm": 0.5088282823562622, "learning_rate": 0.0005, "loss": 1.651139736175537, "step": 13725 }, { "epoch": 1.0820654316121403, "grad_norm": 0.4769277572631836, "learning_rate": 0.0005, "loss": 1.2794092893600464, "step": 13726 }, { "epoch": 1.0821442648797792, "grad_norm": 0.4502770006656647, "learning_rate": 0.0005, "loss": 1.5106829404830933, "step": 13727 }, { "epoch": 1.0822230981474181, "grad_norm": 0.5194819569587708, "learning_rate": 0.0005, "loss": 1.8123959302902222, "step": 13728 }, { "epoch": 1.0823019314150573, "grad_norm": 0.5452989339828491, "learning_rate": 0.0005, "loss": 1.4778964519500732, "step": 13729 }, { "epoch": 1.0823807646826962, "grad_norm": 0.455161988735199, "learning_rate": 0.0005, "loss": 1.7706036567687988, "step": 13730 }, { "epoch": 1.082459597950335, "grad_norm": 0.5648624300956726, "learning_rate": 0.0005, "loss": 1.5140992403030396, "step": 13731 }, { "epoch": 1.082538431217974, "grad_norm": 0.6852217316627502, "learning_rate": 0.0005, "loss": 1.7279322147369385, "step": 13732 }, { "epoch": 1.082617264485613, "grad_norm": 0.5058507323265076, "learning_rate": 0.0005, "loss": 1.941741704940796, "step": 13733 }, { "epoch": 1.0826960977532518, "grad_norm": 0.5577330589294434, "learning_rate": 0.0005, "loss": 1.7270634174346924, "step": 13734 }, { "epoch": 1.0827749310208907, "grad_norm": 0.5551367998123169, "learning_rate": 0.0005, "loss": 1.8283888101577759, "step": 13735 }, { "epoch": 1.0828537642885299, "grad_norm": 0.469330757856369, "learning_rate": 0.0005, "loss": 1.7075165510177612, "step": 13736 }, { "epoch": 1.0829325975561688, "grad_norm": 0.5278681516647339, "learning_rate": 0.0005, "loss": 1.900343656539917, "step": 13737 }, { "epoch": 1.0830114308238077, "grad_norm": 0.6309038996696472, "learning_rate": 0.0005, "loss": 1.6513820886611938, "step": 13738 }, { "epoch": 1.0830902640914466, "grad_norm": 0.5173075795173645, "learning_rate": 0.0005, "loss": 1.6487057209014893, "step": 13739 }, { "epoch": 1.0831690973590855, "grad_norm": 0.9425436854362488, "learning_rate": 0.0005, "loss": 1.833172082901001, "step": 13740 }, { "epoch": 1.0832479306267244, "grad_norm": 0.7781128883361816, "learning_rate": 0.0005, "loss": 1.5853767395019531, "step": 13741 }, { "epoch": 1.0833267638943633, "grad_norm": 0.4890540838241577, "learning_rate": 0.0005, "loss": 1.6850922107696533, "step": 13742 }, { "epoch": 1.0834055971620025, "grad_norm": 0.510970950126648, "learning_rate": 0.0005, "loss": 1.667189598083496, "step": 13743 }, { "epoch": 1.0834844304296414, "grad_norm": 0.5347151756286621, "learning_rate": 0.0005, "loss": 1.728124737739563, "step": 13744 }, { "epoch": 1.0835632636972803, "grad_norm": 0.5895126461982727, "learning_rate": 0.0005, "loss": 1.7151894569396973, "step": 13745 }, { "epoch": 1.0836420969649192, "grad_norm": 0.4902716875076294, "learning_rate": 0.0005, "loss": 1.5029265880584717, "step": 13746 }, { "epoch": 1.083720930232558, "grad_norm": 0.8985428214073181, "learning_rate": 0.0005, "loss": 1.696695327758789, "step": 13747 }, { "epoch": 1.083799763500197, "grad_norm": 0.5331404805183411, "learning_rate": 0.0005, "loss": 1.8226045370101929, "step": 13748 }, { "epoch": 1.083878596767836, "grad_norm": 0.5209709405899048, "learning_rate": 0.0005, "loss": 1.7650563716888428, "step": 13749 }, { "epoch": 1.083957430035475, "grad_norm": 0.9190455675125122, "learning_rate": 0.0005, "loss": 1.9994115829467773, "step": 13750 }, { "epoch": 1.084036263303114, "grad_norm": 0.719884991645813, "learning_rate": 0.0005, "loss": 1.8910808563232422, "step": 13751 }, { "epoch": 1.084115096570753, "grad_norm": 0.7132506966590881, "learning_rate": 0.0005, "loss": 1.9598698616027832, "step": 13752 }, { "epoch": 1.0841939298383918, "grad_norm": 0.5999653339385986, "learning_rate": 0.0005, "loss": 1.865218162536621, "step": 13753 }, { "epoch": 1.0842727631060307, "grad_norm": 0.7166880369186401, "learning_rate": 0.0005, "loss": 1.6709855794906616, "step": 13754 }, { "epoch": 1.0843515963736696, "grad_norm": 0.5036819577217102, "learning_rate": 0.0005, "loss": 1.6274436712265015, "step": 13755 }, { "epoch": 1.0844304296413085, "grad_norm": 0.5264765024185181, "learning_rate": 0.0005, "loss": 1.6553012132644653, "step": 13756 }, { "epoch": 1.0845092629089477, "grad_norm": 0.5881864428520203, "learning_rate": 0.0005, "loss": 1.6624388694763184, "step": 13757 }, { "epoch": 1.0845880961765866, "grad_norm": 0.6537392735481262, "learning_rate": 0.0005, "loss": 1.8214970827102661, "step": 13758 }, { "epoch": 1.0846669294442255, "grad_norm": 0.796275794506073, "learning_rate": 0.0005, "loss": 2.095668315887451, "step": 13759 }, { "epoch": 1.0847457627118644, "grad_norm": 1.025785207748413, "learning_rate": 0.0005, "loss": 1.8466989994049072, "step": 13760 }, { "epoch": 1.0848245959795033, "grad_norm": 0.6716512441635132, "learning_rate": 0.0005, "loss": 1.5837143659591675, "step": 13761 }, { "epoch": 1.0849034292471422, "grad_norm": 0.6092441082000732, "learning_rate": 0.0005, "loss": 1.770878553390503, "step": 13762 }, { "epoch": 1.0849822625147811, "grad_norm": 0.5364110469818115, "learning_rate": 0.0005, "loss": 1.5734870433807373, "step": 13763 }, { "epoch": 1.0850610957824203, "grad_norm": 0.5559357404708862, "learning_rate": 0.0005, "loss": 1.6950581073760986, "step": 13764 }, { "epoch": 1.0851399290500592, "grad_norm": 0.4908505082130432, "learning_rate": 0.0005, "loss": 1.7446444034576416, "step": 13765 }, { "epoch": 1.085218762317698, "grad_norm": 0.6216492652893066, "learning_rate": 0.0005, "loss": 1.6169122457504272, "step": 13766 }, { "epoch": 1.085297595585337, "grad_norm": 0.4440411925315857, "learning_rate": 0.0005, "loss": 1.4402976036071777, "step": 13767 }, { "epoch": 1.085376428852976, "grad_norm": 0.5630261301994324, "learning_rate": 0.0005, "loss": 1.588510513305664, "step": 13768 }, { "epoch": 1.0854552621206148, "grad_norm": 0.5251868963241577, "learning_rate": 0.0005, "loss": 1.6302809715270996, "step": 13769 }, { "epoch": 1.0855340953882537, "grad_norm": 0.6232468485832214, "learning_rate": 0.0005, "loss": 1.7136600017547607, "step": 13770 }, { "epoch": 1.0856129286558929, "grad_norm": 0.528556227684021, "learning_rate": 0.0005, "loss": 1.4670082330703735, "step": 13771 }, { "epoch": 1.0856917619235318, "grad_norm": 0.6626797318458557, "learning_rate": 0.0005, "loss": 1.7562837600708008, "step": 13772 }, { "epoch": 1.0857705951911707, "grad_norm": 0.5232129096984863, "learning_rate": 0.0005, "loss": 1.5316654443740845, "step": 13773 }, { "epoch": 1.0858494284588096, "grad_norm": 0.7017101645469666, "learning_rate": 0.0005, "loss": 1.9404741525650024, "step": 13774 }, { "epoch": 1.0859282617264485, "grad_norm": 0.5357726216316223, "learning_rate": 0.0005, "loss": 1.5036976337432861, "step": 13775 }, { "epoch": 1.0860070949940874, "grad_norm": 0.5623042583465576, "learning_rate": 0.0005, "loss": 1.564867615699768, "step": 13776 }, { "epoch": 1.0860859282617263, "grad_norm": 0.5469633340835571, "learning_rate": 0.0005, "loss": 1.8409876823425293, "step": 13777 }, { "epoch": 1.0861647615293655, "grad_norm": 0.5839430093765259, "learning_rate": 0.0005, "loss": 1.8700084686279297, "step": 13778 }, { "epoch": 1.0862435947970044, "grad_norm": 0.6059000492095947, "learning_rate": 0.0005, "loss": 1.674286961555481, "step": 13779 }, { "epoch": 1.0863224280646433, "grad_norm": 0.5233036279678345, "learning_rate": 0.0005, "loss": 1.8826205730438232, "step": 13780 }, { "epoch": 1.0864012613322822, "grad_norm": 0.5627710223197937, "learning_rate": 0.0005, "loss": 1.68135404586792, "step": 13781 }, { "epoch": 1.0864800945999211, "grad_norm": 0.4707038700580597, "learning_rate": 0.0005, "loss": 1.75001859664917, "step": 13782 }, { "epoch": 1.08655892786756, "grad_norm": 0.5327612161636353, "learning_rate": 0.0005, "loss": 1.6734552383422852, "step": 13783 }, { "epoch": 1.086637761135199, "grad_norm": 0.6175418496131897, "learning_rate": 0.0005, "loss": 1.8511122465133667, "step": 13784 }, { "epoch": 1.086716594402838, "grad_norm": 0.4841333031654358, "learning_rate": 0.0005, "loss": 1.5964875221252441, "step": 13785 }, { "epoch": 1.086795427670477, "grad_norm": 0.5069947242736816, "learning_rate": 0.0005, "loss": 1.7204110622406006, "step": 13786 }, { "epoch": 1.086874260938116, "grad_norm": 0.5797964334487915, "learning_rate": 0.0005, "loss": 1.6548057794570923, "step": 13787 }, { "epoch": 1.0869530942057548, "grad_norm": 0.5755504965782166, "learning_rate": 0.0005, "loss": 1.7947161197662354, "step": 13788 }, { "epoch": 1.0870319274733937, "grad_norm": 0.488389253616333, "learning_rate": 0.0005, "loss": 1.6339144706726074, "step": 13789 }, { "epoch": 1.0871107607410326, "grad_norm": 0.7143346667289734, "learning_rate": 0.0005, "loss": 2.0283682346343994, "step": 13790 }, { "epoch": 1.0871895940086715, "grad_norm": 0.397157222032547, "learning_rate": 0.0005, "loss": 1.4522831439971924, "step": 13791 }, { "epoch": 1.0872684272763107, "grad_norm": 0.5869820713996887, "learning_rate": 0.0005, "loss": 1.7070960998535156, "step": 13792 }, { "epoch": 1.0873472605439496, "grad_norm": 0.5798458456993103, "learning_rate": 0.0005, "loss": 1.5224839448928833, "step": 13793 }, { "epoch": 1.0874260938115885, "grad_norm": 0.6384881138801575, "learning_rate": 0.0005, "loss": 1.9392035007476807, "step": 13794 }, { "epoch": 1.0875049270792274, "grad_norm": 0.5377897620201111, "learning_rate": 0.0005, "loss": 1.667665719985962, "step": 13795 }, { "epoch": 1.0875837603468663, "grad_norm": 0.5342070460319519, "learning_rate": 0.0005, "loss": 1.7437455654144287, "step": 13796 }, { "epoch": 1.0876625936145052, "grad_norm": 0.6059144139289856, "learning_rate": 0.0005, "loss": 1.5938564538955688, "step": 13797 }, { "epoch": 1.0877414268821444, "grad_norm": 0.6663393974304199, "learning_rate": 0.0005, "loss": 1.7771453857421875, "step": 13798 }, { "epoch": 1.0878202601497833, "grad_norm": 0.9476544857025146, "learning_rate": 0.0005, "loss": 1.6804072856903076, "step": 13799 }, { "epoch": 1.0878990934174222, "grad_norm": 0.5225136876106262, "learning_rate": 0.0005, "loss": 1.8001165390014648, "step": 13800 }, { "epoch": 1.087977926685061, "grad_norm": 1.0791258811950684, "learning_rate": 0.0005, "loss": 1.8297489881515503, "step": 13801 }, { "epoch": 1.0880567599527, "grad_norm": 0.5045891404151917, "learning_rate": 0.0005, "loss": 1.5950467586517334, "step": 13802 }, { "epoch": 1.088135593220339, "grad_norm": 0.5278923511505127, "learning_rate": 0.0005, "loss": 1.6328972578048706, "step": 13803 }, { "epoch": 1.0882144264879778, "grad_norm": 0.643142819404602, "learning_rate": 0.0005, "loss": 1.8246580362319946, "step": 13804 }, { "epoch": 1.088293259755617, "grad_norm": 0.4509749412536621, "learning_rate": 0.0005, "loss": 1.570207118988037, "step": 13805 }, { "epoch": 1.0883720930232559, "grad_norm": 0.5619410872459412, "learning_rate": 0.0005, "loss": 1.5598633289337158, "step": 13806 }, { "epoch": 1.0884509262908948, "grad_norm": 0.6334952712059021, "learning_rate": 0.0005, "loss": 1.7822726964950562, "step": 13807 }, { "epoch": 1.0885297595585337, "grad_norm": 0.5861926674842834, "learning_rate": 0.0005, "loss": 1.653239369392395, "step": 13808 }, { "epoch": 1.0886085928261726, "grad_norm": 0.5704270005226135, "learning_rate": 0.0005, "loss": 1.581781029701233, "step": 13809 }, { "epoch": 1.0886874260938115, "grad_norm": 0.493808388710022, "learning_rate": 0.0005, "loss": 1.5401462316513062, "step": 13810 }, { "epoch": 1.0887662593614504, "grad_norm": 0.569737434387207, "learning_rate": 0.0005, "loss": 1.7220162153244019, "step": 13811 }, { "epoch": 1.0888450926290896, "grad_norm": 0.6257238984107971, "learning_rate": 0.0005, "loss": 1.9019677639007568, "step": 13812 }, { "epoch": 1.0889239258967285, "grad_norm": 0.7161325812339783, "learning_rate": 0.0005, "loss": 1.962862491607666, "step": 13813 }, { "epoch": 1.0890027591643674, "grad_norm": 0.5192984938621521, "learning_rate": 0.0005, "loss": 1.689601182937622, "step": 13814 }, { "epoch": 1.0890815924320063, "grad_norm": 0.5359362363815308, "learning_rate": 0.0005, "loss": 1.7963135242462158, "step": 13815 }, { "epoch": 1.0891604256996452, "grad_norm": 0.6004175543785095, "learning_rate": 0.0005, "loss": 1.7604211568832397, "step": 13816 }, { "epoch": 1.0892392589672841, "grad_norm": 0.5370796322822571, "learning_rate": 0.0005, "loss": 2.163684368133545, "step": 13817 }, { "epoch": 1.089318092234923, "grad_norm": 0.5216787457466125, "learning_rate": 0.0005, "loss": 1.7439467906951904, "step": 13818 }, { "epoch": 1.0893969255025622, "grad_norm": 0.4890236258506775, "learning_rate": 0.0005, "loss": 1.4681787490844727, "step": 13819 }, { "epoch": 1.089475758770201, "grad_norm": 0.5773103833198547, "learning_rate": 0.0005, "loss": 1.710085391998291, "step": 13820 }, { "epoch": 1.08955459203784, "grad_norm": 0.7338376045227051, "learning_rate": 0.0005, "loss": 1.6324222087860107, "step": 13821 }, { "epoch": 1.089633425305479, "grad_norm": 0.690589427947998, "learning_rate": 0.0005, "loss": 1.4906327724456787, "step": 13822 }, { "epoch": 1.0897122585731178, "grad_norm": 0.5230172872543335, "learning_rate": 0.0005, "loss": 1.6973541975021362, "step": 13823 }, { "epoch": 1.0897910918407567, "grad_norm": 0.6111791729927063, "learning_rate": 0.0005, "loss": 1.8026630878448486, "step": 13824 }, { "epoch": 1.0898699251083956, "grad_norm": 0.5505195260047913, "learning_rate": 0.0005, "loss": 1.6423120498657227, "step": 13825 }, { "epoch": 1.0899487583760348, "grad_norm": 0.5053861737251282, "learning_rate": 0.0005, "loss": 1.6693835258483887, "step": 13826 }, { "epoch": 1.0900275916436737, "grad_norm": 0.4686300456523895, "learning_rate": 0.0005, "loss": 1.698003888130188, "step": 13827 }, { "epoch": 1.0901064249113126, "grad_norm": 0.5071134567260742, "learning_rate": 0.0005, "loss": 1.6732317209243774, "step": 13828 }, { "epoch": 1.0901852581789515, "grad_norm": 0.533397376537323, "learning_rate": 0.0005, "loss": 1.741415023803711, "step": 13829 }, { "epoch": 1.0902640914465904, "grad_norm": 0.8841724395751953, "learning_rate": 0.0005, "loss": 1.5059523582458496, "step": 13830 }, { "epoch": 1.0903429247142293, "grad_norm": 0.5085527896881104, "learning_rate": 0.0005, "loss": 1.9190746545791626, "step": 13831 }, { "epoch": 1.0904217579818682, "grad_norm": 0.5600855350494385, "learning_rate": 0.0005, "loss": 1.6855183839797974, "step": 13832 }, { "epoch": 1.0905005912495074, "grad_norm": 0.6293860673904419, "learning_rate": 0.0005, "loss": 1.6968705654144287, "step": 13833 }, { "epoch": 1.0905794245171463, "grad_norm": 0.5324727296829224, "learning_rate": 0.0005, "loss": 1.5891129970550537, "step": 13834 }, { "epoch": 1.0906582577847852, "grad_norm": 0.6067538261413574, "learning_rate": 0.0005, "loss": 1.4716206789016724, "step": 13835 }, { "epoch": 1.0907370910524241, "grad_norm": 0.46366775035858154, "learning_rate": 0.0005, "loss": 1.5892188549041748, "step": 13836 }, { "epoch": 1.090815924320063, "grad_norm": 0.5566190481185913, "learning_rate": 0.0005, "loss": 1.8872631788253784, "step": 13837 }, { "epoch": 1.090894757587702, "grad_norm": 0.6746429204940796, "learning_rate": 0.0005, "loss": 1.5830867290496826, "step": 13838 }, { "epoch": 1.090973590855341, "grad_norm": 0.473873108625412, "learning_rate": 0.0005, "loss": 1.5847055912017822, "step": 13839 }, { "epoch": 1.09105242412298, "grad_norm": 0.547014594078064, "learning_rate": 0.0005, "loss": 1.962186336517334, "step": 13840 }, { "epoch": 1.091131257390619, "grad_norm": 0.5750038027763367, "learning_rate": 0.0005, "loss": 1.5536249876022339, "step": 13841 }, { "epoch": 1.0912100906582578, "grad_norm": 0.5795025825500488, "learning_rate": 0.0005, "loss": 1.938209056854248, "step": 13842 }, { "epoch": 1.0912889239258967, "grad_norm": 0.5534141063690186, "learning_rate": 0.0005, "loss": 1.5824674367904663, "step": 13843 }, { "epoch": 1.0913677571935356, "grad_norm": 0.4798521399497986, "learning_rate": 0.0005, "loss": 1.6605851650238037, "step": 13844 }, { "epoch": 1.0914465904611745, "grad_norm": 0.47576117515563965, "learning_rate": 0.0005, "loss": 1.5551531314849854, "step": 13845 }, { "epoch": 1.0915254237288137, "grad_norm": 0.6661534309387207, "learning_rate": 0.0005, "loss": 1.9085118770599365, "step": 13846 }, { "epoch": 1.0916042569964526, "grad_norm": 0.6152422428131104, "learning_rate": 0.0005, "loss": 1.8775458335876465, "step": 13847 }, { "epoch": 1.0916830902640915, "grad_norm": 0.4601272940635681, "learning_rate": 0.0005, "loss": 1.4475836753845215, "step": 13848 }, { "epoch": 1.0917619235317304, "grad_norm": 0.4891617000102997, "learning_rate": 0.0005, "loss": 1.618438720703125, "step": 13849 }, { "epoch": 1.0918407567993693, "grad_norm": 0.4360329806804657, "learning_rate": 0.0005, "loss": 1.5630838871002197, "step": 13850 }, { "epoch": 1.0919195900670082, "grad_norm": 0.464500367641449, "learning_rate": 0.0005, "loss": 1.718658447265625, "step": 13851 }, { "epoch": 1.0919984233346471, "grad_norm": 0.5192052125930786, "learning_rate": 0.0005, "loss": 1.7441930770874023, "step": 13852 }, { "epoch": 1.0920772566022863, "grad_norm": 0.48138877749443054, "learning_rate": 0.0005, "loss": 1.7468594312667847, "step": 13853 }, { "epoch": 1.0921560898699252, "grad_norm": 0.5693536400794983, "learning_rate": 0.0005, "loss": 1.5720374584197998, "step": 13854 }, { "epoch": 1.092234923137564, "grad_norm": 0.4672485589981079, "learning_rate": 0.0005, "loss": 1.7417659759521484, "step": 13855 }, { "epoch": 1.092313756405203, "grad_norm": 0.6374363899230957, "learning_rate": 0.0005, "loss": 1.4353042840957642, "step": 13856 }, { "epoch": 1.092392589672842, "grad_norm": 0.5475602746009827, "learning_rate": 0.0005, "loss": 1.6920571327209473, "step": 13857 }, { "epoch": 1.0924714229404808, "grad_norm": 0.47700944542884827, "learning_rate": 0.0005, "loss": 1.5853800773620605, "step": 13858 }, { "epoch": 1.0925502562081197, "grad_norm": 0.6929528713226318, "learning_rate": 0.0005, "loss": 1.670475721359253, "step": 13859 }, { "epoch": 1.0926290894757589, "grad_norm": 0.5764536261558533, "learning_rate": 0.0005, "loss": 1.6043039560317993, "step": 13860 }, { "epoch": 1.0927079227433978, "grad_norm": 0.4978554844856262, "learning_rate": 0.0005, "loss": 1.8854621648788452, "step": 13861 }, { "epoch": 1.0927867560110367, "grad_norm": 0.5472878217697144, "learning_rate": 0.0005, "loss": 1.85335111618042, "step": 13862 }, { "epoch": 1.0928655892786756, "grad_norm": 0.5168014764785767, "learning_rate": 0.0005, "loss": 1.5570114850997925, "step": 13863 }, { "epoch": 1.0929444225463145, "grad_norm": 0.6567767858505249, "learning_rate": 0.0005, "loss": 1.6161673069000244, "step": 13864 }, { "epoch": 1.0930232558139534, "grad_norm": 0.454787939786911, "learning_rate": 0.0005, "loss": 1.617741346359253, "step": 13865 }, { "epoch": 1.0931020890815923, "grad_norm": 0.5617341995239258, "learning_rate": 0.0005, "loss": 1.6081459522247314, "step": 13866 }, { "epoch": 1.0931809223492315, "grad_norm": 0.6010182499885559, "learning_rate": 0.0005, "loss": 1.8808084726333618, "step": 13867 }, { "epoch": 1.0932597556168704, "grad_norm": 0.5411809682846069, "learning_rate": 0.0005, "loss": 1.4178011417388916, "step": 13868 }, { "epoch": 1.0933385888845093, "grad_norm": 0.5143337249755859, "learning_rate": 0.0005, "loss": 1.6871066093444824, "step": 13869 }, { "epoch": 1.0934174221521482, "grad_norm": 0.5436952710151672, "learning_rate": 0.0005, "loss": 1.9684078693389893, "step": 13870 }, { "epoch": 1.0934962554197871, "grad_norm": 0.5098973512649536, "learning_rate": 0.0005, "loss": 1.6089550256729126, "step": 13871 }, { "epoch": 1.093575088687426, "grad_norm": 0.6288309097290039, "learning_rate": 0.0005, "loss": 1.9394309520721436, "step": 13872 }, { "epoch": 1.093653921955065, "grad_norm": 0.39874738454818726, "learning_rate": 0.0005, "loss": 1.5063731670379639, "step": 13873 }, { "epoch": 1.093732755222704, "grad_norm": 0.543832540512085, "learning_rate": 0.0005, "loss": 1.6947648525238037, "step": 13874 }, { "epoch": 1.093811588490343, "grad_norm": 0.611406683921814, "learning_rate": 0.0005, "loss": 1.764857530593872, "step": 13875 }, { "epoch": 1.093890421757982, "grad_norm": 0.48466581106185913, "learning_rate": 0.0005, "loss": 1.7016360759735107, "step": 13876 }, { "epoch": 1.0939692550256208, "grad_norm": 0.5191201567649841, "learning_rate": 0.0005, "loss": 1.6632484197616577, "step": 13877 }, { "epoch": 1.0940480882932597, "grad_norm": 0.4918072521686554, "learning_rate": 0.0005, "loss": 1.5558888912200928, "step": 13878 }, { "epoch": 1.0941269215608986, "grad_norm": 0.5392233729362488, "learning_rate": 0.0005, "loss": 1.7176958322525024, "step": 13879 }, { "epoch": 1.0942057548285375, "grad_norm": 0.6684110760688782, "learning_rate": 0.0005, "loss": 1.668743371963501, "step": 13880 }, { "epoch": 1.0942845880961767, "grad_norm": 0.5905817747116089, "learning_rate": 0.0005, "loss": 1.7555679082870483, "step": 13881 }, { "epoch": 1.0943634213638156, "grad_norm": 0.5879877805709839, "learning_rate": 0.0005, "loss": 1.9914181232452393, "step": 13882 }, { "epoch": 1.0944422546314545, "grad_norm": 0.5250833034515381, "learning_rate": 0.0005, "loss": 1.7917736768722534, "step": 13883 }, { "epoch": 1.0945210878990934, "grad_norm": 0.7611029744148254, "learning_rate": 0.0005, "loss": 1.7401679754257202, "step": 13884 }, { "epoch": 1.0945999211667323, "grad_norm": 0.6197764277458191, "learning_rate": 0.0005, "loss": 1.7715961933135986, "step": 13885 }, { "epoch": 1.0946787544343712, "grad_norm": 0.5063101053237915, "learning_rate": 0.0005, "loss": 1.6906323432922363, "step": 13886 }, { "epoch": 1.0947575877020101, "grad_norm": 0.5934268832206726, "learning_rate": 0.0005, "loss": 2.0479702949523926, "step": 13887 }, { "epoch": 1.0948364209696493, "grad_norm": 0.6529024243354797, "learning_rate": 0.0005, "loss": 1.8207635879516602, "step": 13888 }, { "epoch": 1.0949152542372882, "grad_norm": 0.462460458278656, "learning_rate": 0.0005, "loss": 1.7430715560913086, "step": 13889 }, { "epoch": 1.094994087504927, "grad_norm": 0.625630795955658, "learning_rate": 0.0005, "loss": 1.564803123474121, "step": 13890 }, { "epoch": 1.095072920772566, "grad_norm": 0.4663325548171997, "learning_rate": 0.0005, "loss": 1.3289597034454346, "step": 13891 }, { "epoch": 1.095151754040205, "grad_norm": 0.4679498076438904, "learning_rate": 0.0005, "loss": 1.5725324153900146, "step": 13892 }, { "epoch": 1.0952305873078438, "grad_norm": 0.4765869379043579, "learning_rate": 0.0005, "loss": 1.514388084411621, "step": 13893 }, { "epoch": 1.0953094205754828, "grad_norm": 0.6460517048835754, "learning_rate": 0.0005, "loss": 1.7909961938858032, "step": 13894 }, { "epoch": 1.0953882538431219, "grad_norm": 0.678114652633667, "learning_rate": 0.0005, "loss": 1.980210542678833, "step": 13895 }, { "epoch": 1.0954670871107608, "grad_norm": 0.609857439994812, "learning_rate": 0.0005, "loss": 1.777290940284729, "step": 13896 }, { "epoch": 1.0955459203783997, "grad_norm": 0.5405455827713013, "learning_rate": 0.0005, "loss": 1.6644127368927002, "step": 13897 }, { "epoch": 1.0956247536460386, "grad_norm": 0.5211985111236572, "learning_rate": 0.0005, "loss": 1.6373738050460815, "step": 13898 }, { "epoch": 1.0957035869136775, "grad_norm": 0.6081262826919556, "learning_rate": 0.0005, "loss": 1.677741289138794, "step": 13899 }, { "epoch": 1.0957824201813164, "grad_norm": 0.6180880665779114, "learning_rate": 0.0005, "loss": 1.7056418657302856, "step": 13900 }, { "epoch": 1.0958612534489554, "grad_norm": 0.4527357816696167, "learning_rate": 0.0005, "loss": 1.47212553024292, "step": 13901 }, { "epoch": 1.0959400867165945, "grad_norm": 0.47353416681289673, "learning_rate": 0.0005, "loss": 1.553724765777588, "step": 13902 }, { "epoch": 1.0960189199842334, "grad_norm": 0.5541364550590515, "learning_rate": 0.0005, "loss": 1.7878484725952148, "step": 13903 }, { "epoch": 1.0960977532518723, "grad_norm": 0.4910883605480194, "learning_rate": 0.0005, "loss": 1.8737435340881348, "step": 13904 }, { "epoch": 1.0961765865195112, "grad_norm": 0.5318499207496643, "learning_rate": 0.0005, "loss": 2.0392394065856934, "step": 13905 }, { "epoch": 1.0962554197871501, "grad_norm": 0.47517797350883484, "learning_rate": 0.0005, "loss": 1.8548924922943115, "step": 13906 }, { "epoch": 1.096334253054789, "grad_norm": 0.5546130537986755, "learning_rate": 0.0005, "loss": 1.705635666847229, "step": 13907 }, { "epoch": 1.096413086322428, "grad_norm": 0.550374448299408, "learning_rate": 0.0005, "loss": 1.763878583908081, "step": 13908 }, { "epoch": 1.096491919590067, "grad_norm": 0.48308175802230835, "learning_rate": 0.0005, "loss": 1.512919545173645, "step": 13909 }, { "epoch": 1.096570752857706, "grad_norm": 0.6288777589797974, "learning_rate": 0.0005, "loss": 1.773921012878418, "step": 13910 }, { "epoch": 1.096649586125345, "grad_norm": 0.5175527930259705, "learning_rate": 0.0005, "loss": 1.6473954916000366, "step": 13911 }, { "epoch": 1.0967284193929838, "grad_norm": 0.5983279943466187, "learning_rate": 0.0005, "loss": 1.582542061805725, "step": 13912 }, { "epoch": 1.0968072526606227, "grad_norm": 0.5469558238983154, "learning_rate": 0.0005, "loss": 1.8703241348266602, "step": 13913 }, { "epoch": 1.0968860859282616, "grad_norm": 0.6228156089782715, "learning_rate": 0.0005, "loss": 1.8466551303863525, "step": 13914 }, { "epoch": 1.0969649191959006, "grad_norm": 0.5823207497596741, "learning_rate": 0.0005, "loss": 1.7307794094085693, "step": 13915 }, { "epoch": 1.0970437524635397, "grad_norm": 0.4693434238433838, "learning_rate": 0.0005, "loss": 1.7762163877487183, "step": 13916 }, { "epoch": 1.0971225857311786, "grad_norm": 0.529890239238739, "learning_rate": 0.0005, "loss": 1.68073570728302, "step": 13917 }, { "epoch": 1.0972014189988175, "grad_norm": 0.5468106865882874, "learning_rate": 0.0005, "loss": 1.4416100978851318, "step": 13918 }, { "epoch": 1.0972802522664564, "grad_norm": 0.5100714564323425, "learning_rate": 0.0005, "loss": 1.5471317768096924, "step": 13919 }, { "epoch": 1.0973590855340953, "grad_norm": 0.4842197299003601, "learning_rate": 0.0005, "loss": 1.5183048248291016, "step": 13920 }, { "epoch": 1.0974379188017342, "grad_norm": 0.5600854754447937, "learning_rate": 0.0005, "loss": 1.5448923110961914, "step": 13921 }, { "epoch": 1.0975167520693732, "grad_norm": 0.5136591196060181, "learning_rate": 0.0005, "loss": 1.7364234924316406, "step": 13922 }, { "epoch": 1.0975955853370123, "grad_norm": 0.5732925534248352, "learning_rate": 0.0005, "loss": 1.763556957244873, "step": 13923 }, { "epoch": 1.0976744186046512, "grad_norm": 0.8043040037155151, "learning_rate": 0.0005, "loss": 1.877805471420288, "step": 13924 }, { "epoch": 1.0977532518722901, "grad_norm": 0.8992496728897095, "learning_rate": 0.0005, "loss": 1.7897655963897705, "step": 13925 }, { "epoch": 1.097832085139929, "grad_norm": 0.6772200465202332, "learning_rate": 0.0005, "loss": 1.8800550699234009, "step": 13926 }, { "epoch": 1.097910918407568, "grad_norm": 0.5084061026573181, "learning_rate": 0.0005, "loss": 1.6186894178390503, "step": 13927 }, { "epoch": 1.0979897516752068, "grad_norm": 0.5259662866592407, "learning_rate": 0.0005, "loss": 1.5422379970550537, "step": 13928 }, { "epoch": 1.098068584942846, "grad_norm": 0.595073401927948, "learning_rate": 0.0005, "loss": 1.574399471282959, "step": 13929 }, { "epoch": 1.098147418210485, "grad_norm": 0.6799827218055725, "learning_rate": 0.0005, "loss": 1.9502851963043213, "step": 13930 }, { "epoch": 1.0982262514781238, "grad_norm": 1.0160342454910278, "learning_rate": 0.0005, "loss": 1.7757813930511475, "step": 13931 }, { "epoch": 1.0983050847457627, "grad_norm": 0.6674513220787048, "learning_rate": 0.0005, "loss": 1.6835699081420898, "step": 13932 }, { "epoch": 1.0983839180134016, "grad_norm": 0.5267891883850098, "learning_rate": 0.0005, "loss": 1.7822463512420654, "step": 13933 }, { "epoch": 1.0984627512810405, "grad_norm": 0.5327855944633484, "learning_rate": 0.0005, "loss": 1.8029260635375977, "step": 13934 }, { "epoch": 1.0985415845486795, "grad_norm": 0.5546032190322876, "learning_rate": 0.0005, "loss": 1.6123319864273071, "step": 13935 }, { "epoch": 1.0986204178163186, "grad_norm": 0.5868673920631409, "learning_rate": 0.0005, "loss": 1.505487084388733, "step": 13936 }, { "epoch": 1.0986992510839575, "grad_norm": 0.5907269716262817, "learning_rate": 0.0005, "loss": 1.6448640823364258, "step": 13937 }, { "epoch": 1.0987780843515964, "grad_norm": 0.45607632398605347, "learning_rate": 0.0005, "loss": 1.8219735622406006, "step": 13938 }, { "epoch": 1.0988569176192353, "grad_norm": 0.6307852268218994, "learning_rate": 0.0005, "loss": 1.7012431621551514, "step": 13939 }, { "epoch": 1.0989357508868742, "grad_norm": 0.487275093793869, "learning_rate": 0.0005, "loss": 1.6274468898773193, "step": 13940 }, { "epoch": 1.0990145841545131, "grad_norm": 0.5377055406570435, "learning_rate": 0.0005, "loss": 1.7634797096252441, "step": 13941 }, { "epoch": 1.099093417422152, "grad_norm": 0.5502238869667053, "learning_rate": 0.0005, "loss": 1.7316582202911377, "step": 13942 }, { "epoch": 1.0991722506897912, "grad_norm": 0.564627468585968, "learning_rate": 0.0005, "loss": 1.659306287765503, "step": 13943 }, { "epoch": 1.09925108395743, "grad_norm": 0.542304277420044, "learning_rate": 0.0005, "loss": 1.6011844873428345, "step": 13944 }, { "epoch": 1.099329917225069, "grad_norm": 0.6886412501335144, "learning_rate": 0.0005, "loss": 1.5293807983398438, "step": 13945 }, { "epoch": 1.099408750492708, "grad_norm": 0.5669482350349426, "learning_rate": 0.0005, "loss": 1.664381980895996, "step": 13946 }, { "epoch": 1.0994875837603468, "grad_norm": 0.6246986389160156, "learning_rate": 0.0005, "loss": 1.5386955738067627, "step": 13947 }, { "epoch": 1.0995664170279857, "grad_norm": 0.5022936463356018, "learning_rate": 0.0005, "loss": 1.6661593914031982, "step": 13948 }, { "epoch": 1.0996452502956247, "grad_norm": 0.7096837759017944, "learning_rate": 0.0005, "loss": 1.745138168334961, "step": 13949 }, { "epoch": 1.0997240835632638, "grad_norm": 0.5942827463150024, "learning_rate": 0.0005, "loss": 1.7189226150512695, "step": 13950 }, { "epoch": 1.0998029168309027, "grad_norm": 0.5427598357200623, "learning_rate": 0.0005, "loss": 1.5425469875335693, "step": 13951 }, { "epoch": 1.0998817500985416, "grad_norm": 0.5463884472846985, "learning_rate": 0.0005, "loss": 1.5945693254470825, "step": 13952 }, { "epoch": 1.0999605833661805, "grad_norm": 0.6897771954536438, "learning_rate": 0.0005, "loss": 1.7769508361816406, "step": 13953 }, { "epoch": 1.1000394166338194, "grad_norm": 0.5263572335243225, "learning_rate": 0.0005, "loss": 1.6267006397247314, "step": 13954 }, { "epoch": 1.1001182499014583, "grad_norm": 0.5386734008789062, "learning_rate": 0.0005, "loss": 1.9147121906280518, "step": 13955 }, { "epoch": 1.1001970831690973, "grad_norm": 0.6447956562042236, "learning_rate": 0.0005, "loss": 1.9455060958862305, "step": 13956 }, { "epoch": 1.1002759164367364, "grad_norm": 0.5616393685340881, "learning_rate": 0.0005, "loss": 1.667581558227539, "step": 13957 }, { "epoch": 1.1003547497043753, "grad_norm": 0.5784801244735718, "learning_rate": 0.0005, "loss": 1.8368277549743652, "step": 13958 }, { "epoch": 1.1004335829720142, "grad_norm": 0.5364838242530823, "learning_rate": 0.0005, "loss": 1.6504603624343872, "step": 13959 }, { "epoch": 1.1005124162396531, "grad_norm": 0.5282788276672363, "learning_rate": 0.0005, "loss": 1.7346059083938599, "step": 13960 }, { "epoch": 1.100591249507292, "grad_norm": 0.494192510843277, "learning_rate": 0.0005, "loss": 1.5517492294311523, "step": 13961 }, { "epoch": 1.100670082774931, "grad_norm": 0.5234783887863159, "learning_rate": 0.0005, "loss": 1.6265079975128174, "step": 13962 }, { "epoch": 1.10074891604257, "grad_norm": 0.9669865965843201, "learning_rate": 0.0005, "loss": 1.6803200244903564, "step": 13963 }, { "epoch": 1.100827749310209, "grad_norm": 0.6153643131256104, "learning_rate": 0.0005, "loss": 1.8670985698699951, "step": 13964 }, { "epoch": 1.100906582577848, "grad_norm": 0.4974161684513092, "learning_rate": 0.0005, "loss": 1.434674859046936, "step": 13965 }, { "epoch": 1.1009854158454868, "grad_norm": 0.7035824656486511, "learning_rate": 0.0005, "loss": 1.6370805501937866, "step": 13966 }, { "epoch": 1.1010642491131257, "grad_norm": 0.7777389287948608, "learning_rate": 0.0005, "loss": 1.7555073499679565, "step": 13967 }, { "epoch": 1.1011430823807646, "grad_norm": 0.5860784649848938, "learning_rate": 0.0005, "loss": 1.7096831798553467, "step": 13968 }, { "epoch": 1.1012219156484035, "grad_norm": 0.537481427192688, "learning_rate": 0.0005, "loss": 1.6503702402114868, "step": 13969 }, { "epoch": 1.1013007489160427, "grad_norm": 1.066070318222046, "learning_rate": 0.0005, "loss": 1.8230007886886597, "step": 13970 }, { "epoch": 1.1013795821836816, "grad_norm": 0.627884030342102, "learning_rate": 0.0005, "loss": 1.8440184593200684, "step": 13971 }, { "epoch": 1.1014584154513205, "grad_norm": 0.5353026986122131, "learning_rate": 0.0005, "loss": 1.6363883018493652, "step": 13972 }, { "epoch": 1.1015372487189594, "grad_norm": 0.5856590867042542, "learning_rate": 0.0005, "loss": 1.8927412033081055, "step": 13973 }, { "epoch": 1.1016160819865983, "grad_norm": 0.612211287021637, "learning_rate": 0.0005, "loss": 1.7491685152053833, "step": 13974 }, { "epoch": 1.1016949152542372, "grad_norm": 0.783684492111206, "learning_rate": 0.0005, "loss": 1.9730334281921387, "step": 13975 }, { "epoch": 1.1017737485218762, "grad_norm": 0.5289786458015442, "learning_rate": 0.0005, "loss": 1.7747377157211304, "step": 13976 }, { "epoch": 1.1018525817895153, "grad_norm": 0.6381410956382751, "learning_rate": 0.0005, "loss": 1.946211338043213, "step": 13977 }, { "epoch": 1.1019314150571542, "grad_norm": 0.671933650970459, "learning_rate": 0.0005, "loss": 1.568819522857666, "step": 13978 }, { "epoch": 1.102010248324793, "grad_norm": 0.5273049473762512, "learning_rate": 0.0005, "loss": 1.6744282245635986, "step": 13979 }, { "epoch": 1.102089081592432, "grad_norm": 0.5854420065879822, "learning_rate": 0.0005, "loss": 1.793602466583252, "step": 13980 }, { "epoch": 1.102167914860071, "grad_norm": 0.7038111686706543, "learning_rate": 0.0005, "loss": 1.7746795415878296, "step": 13981 }, { "epoch": 1.1022467481277098, "grad_norm": 0.5391398072242737, "learning_rate": 0.0005, "loss": 1.6159098148345947, "step": 13982 }, { "epoch": 1.1023255813953488, "grad_norm": 0.7168143391609192, "learning_rate": 0.0005, "loss": 1.74257493019104, "step": 13983 }, { "epoch": 1.1024044146629879, "grad_norm": 0.6316627860069275, "learning_rate": 0.0005, "loss": 1.969759225845337, "step": 13984 }, { "epoch": 1.1024832479306268, "grad_norm": 0.5100290775299072, "learning_rate": 0.0005, "loss": 1.8283671140670776, "step": 13985 }, { "epoch": 1.1025620811982657, "grad_norm": 0.9048227667808533, "learning_rate": 0.0005, "loss": 1.8666795492172241, "step": 13986 }, { "epoch": 1.1026409144659046, "grad_norm": 0.5046235918998718, "learning_rate": 0.0005, "loss": 1.6935482025146484, "step": 13987 }, { "epoch": 1.1027197477335435, "grad_norm": 0.623696506023407, "learning_rate": 0.0005, "loss": 1.9791754484176636, "step": 13988 }, { "epoch": 1.1027985810011824, "grad_norm": 0.6514818668365479, "learning_rate": 0.0005, "loss": 1.5783720016479492, "step": 13989 }, { "epoch": 1.1028774142688214, "grad_norm": 0.536359965801239, "learning_rate": 0.0005, "loss": 1.5102165937423706, "step": 13990 }, { "epoch": 1.1029562475364605, "grad_norm": 0.5731738805770874, "learning_rate": 0.0005, "loss": 1.7475459575653076, "step": 13991 }, { "epoch": 1.1030350808040994, "grad_norm": 0.9733299612998962, "learning_rate": 0.0005, "loss": 1.7658357620239258, "step": 13992 }, { "epoch": 1.1031139140717383, "grad_norm": 0.5759283304214478, "learning_rate": 0.0005, "loss": 1.736502766609192, "step": 13993 }, { "epoch": 1.1031927473393772, "grad_norm": 0.5472170114517212, "learning_rate": 0.0005, "loss": 1.6969122886657715, "step": 13994 }, { "epoch": 1.1032715806070161, "grad_norm": 0.5315292477607727, "learning_rate": 0.0005, "loss": 1.7112724781036377, "step": 13995 }, { "epoch": 1.103350413874655, "grad_norm": 0.884128749370575, "learning_rate": 0.0005, "loss": 1.9450024366378784, "step": 13996 }, { "epoch": 1.103429247142294, "grad_norm": 0.5121734738349915, "learning_rate": 0.0005, "loss": 1.7424206733703613, "step": 13997 }, { "epoch": 1.103508080409933, "grad_norm": 0.6148339509963989, "learning_rate": 0.0005, "loss": 1.705973744392395, "step": 13998 }, { "epoch": 1.103586913677572, "grad_norm": 0.4090328514575958, "learning_rate": 0.0005, "loss": 1.4480304718017578, "step": 13999 }, { "epoch": 1.103665746945211, "grad_norm": 0.5235453248023987, "learning_rate": 0.0005, "loss": 1.5642826557159424, "step": 14000 }, { "epoch": 1.1037445802128498, "grad_norm": 0.5634087324142456, "learning_rate": 0.0005, "loss": 1.7698638439178467, "step": 14001 }, { "epoch": 1.1038234134804887, "grad_norm": 0.6508038640022278, "learning_rate": 0.0005, "loss": 1.497715950012207, "step": 14002 }, { "epoch": 1.1039022467481276, "grad_norm": 0.6541269421577454, "learning_rate": 0.0005, "loss": 1.8553073406219482, "step": 14003 }, { "epoch": 1.1039810800157666, "grad_norm": 0.5857685208320618, "learning_rate": 0.0005, "loss": 1.6588151454925537, "step": 14004 }, { "epoch": 1.1040599132834057, "grad_norm": 0.8765740394592285, "learning_rate": 0.0005, "loss": 1.5893522500991821, "step": 14005 }, { "epoch": 1.1041387465510446, "grad_norm": 0.7795305252075195, "learning_rate": 0.0005, "loss": 1.7752336263656616, "step": 14006 }, { "epoch": 1.1042175798186835, "grad_norm": 0.6743353009223938, "learning_rate": 0.0005, "loss": 1.7011137008666992, "step": 14007 }, { "epoch": 1.1042964130863224, "grad_norm": 0.5305758118629456, "learning_rate": 0.0005, "loss": 1.487815260887146, "step": 14008 }, { "epoch": 1.1043752463539613, "grad_norm": 0.5659968256950378, "learning_rate": 0.0005, "loss": 1.7461113929748535, "step": 14009 }, { "epoch": 1.1044540796216002, "grad_norm": 0.45557901263237, "learning_rate": 0.0005, "loss": 1.491415023803711, "step": 14010 }, { "epoch": 1.1045329128892392, "grad_norm": 0.5773646235466003, "learning_rate": 0.0005, "loss": 1.6821446418762207, "step": 14011 }, { "epoch": 1.1046117461568783, "grad_norm": 0.5197925567626953, "learning_rate": 0.0005, "loss": 1.7697489261627197, "step": 14012 }, { "epoch": 1.1046905794245172, "grad_norm": 0.5655046701431274, "learning_rate": 0.0005, "loss": 1.9108271598815918, "step": 14013 }, { "epoch": 1.1047694126921561, "grad_norm": 0.6154970526695251, "learning_rate": 0.0005, "loss": 1.8731868267059326, "step": 14014 }, { "epoch": 1.104848245959795, "grad_norm": 0.6022217273712158, "learning_rate": 0.0005, "loss": 1.5480294227600098, "step": 14015 }, { "epoch": 1.104927079227434, "grad_norm": 0.524147093296051, "learning_rate": 0.0005, "loss": 1.5860741138458252, "step": 14016 }, { "epoch": 1.1050059124950729, "grad_norm": 0.5629016160964966, "learning_rate": 0.0005, "loss": 1.9214423894882202, "step": 14017 }, { "epoch": 1.1050847457627118, "grad_norm": 0.4577442407608032, "learning_rate": 0.0005, "loss": 1.4795458316802979, "step": 14018 }, { "epoch": 1.105163579030351, "grad_norm": 0.5477046370506287, "learning_rate": 0.0005, "loss": 1.7405576705932617, "step": 14019 }, { "epoch": 1.1052424122979898, "grad_norm": 0.5850414037704468, "learning_rate": 0.0005, "loss": 1.831117868423462, "step": 14020 }, { "epoch": 1.1053212455656287, "grad_norm": 0.5734711289405823, "learning_rate": 0.0005, "loss": 1.487316370010376, "step": 14021 }, { "epoch": 1.1054000788332676, "grad_norm": 0.6985531449317932, "learning_rate": 0.0005, "loss": 1.643988847732544, "step": 14022 }, { "epoch": 1.1054789121009065, "grad_norm": 0.47227194905281067, "learning_rate": 0.0005, "loss": 1.7374082803726196, "step": 14023 }, { "epoch": 1.1055577453685455, "grad_norm": 0.6935197114944458, "learning_rate": 0.0005, "loss": 1.7526237964630127, "step": 14024 }, { "epoch": 1.1056365786361844, "grad_norm": 0.44341304898262024, "learning_rate": 0.0005, "loss": 1.5649135112762451, "step": 14025 }, { "epoch": 1.1057154119038235, "grad_norm": 0.48512887954711914, "learning_rate": 0.0005, "loss": 1.5922467708587646, "step": 14026 }, { "epoch": 1.1057942451714624, "grad_norm": 0.6909276843070984, "learning_rate": 0.0005, "loss": 1.7875081300735474, "step": 14027 }, { "epoch": 1.1058730784391013, "grad_norm": 0.5448755025863647, "learning_rate": 0.0005, "loss": 1.71803879737854, "step": 14028 }, { "epoch": 1.1059519117067402, "grad_norm": 0.4360259473323822, "learning_rate": 0.0005, "loss": 1.5168739557266235, "step": 14029 }, { "epoch": 1.1060307449743791, "grad_norm": 0.5786203742027283, "learning_rate": 0.0005, "loss": 1.9970673322677612, "step": 14030 }, { "epoch": 1.106109578242018, "grad_norm": 0.43280819058418274, "learning_rate": 0.0005, "loss": 1.6547654867172241, "step": 14031 }, { "epoch": 1.106188411509657, "grad_norm": 0.6185824275016785, "learning_rate": 0.0005, "loss": 1.689733624458313, "step": 14032 }, { "epoch": 1.106267244777296, "grad_norm": 0.5924859642982483, "learning_rate": 0.0005, "loss": 1.7218918800354004, "step": 14033 }, { "epoch": 1.106346078044935, "grad_norm": 0.6221556663513184, "learning_rate": 0.0005, "loss": 1.5984803438186646, "step": 14034 }, { "epoch": 1.106424911312574, "grad_norm": 0.41968196630477905, "learning_rate": 0.0005, "loss": 1.6248798370361328, "step": 14035 }, { "epoch": 1.1065037445802128, "grad_norm": 0.5125490427017212, "learning_rate": 0.0005, "loss": 1.6532635688781738, "step": 14036 }, { "epoch": 1.1065825778478517, "grad_norm": 0.536715030670166, "learning_rate": 0.0005, "loss": 1.8031198978424072, "step": 14037 }, { "epoch": 1.1066614111154907, "grad_norm": 0.6221043467521667, "learning_rate": 0.0005, "loss": 1.8315327167510986, "step": 14038 }, { "epoch": 1.1067402443831296, "grad_norm": 0.5367494225502014, "learning_rate": 0.0005, "loss": 1.8629083633422852, "step": 14039 }, { "epoch": 1.1068190776507687, "grad_norm": 0.40307629108428955, "learning_rate": 0.0005, "loss": 1.606896996498108, "step": 14040 }, { "epoch": 1.1068979109184076, "grad_norm": 0.4873870313167572, "learning_rate": 0.0005, "loss": 1.5755136013031006, "step": 14041 }, { "epoch": 1.1069767441860465, "grad_norm": 0.5528654456138611, "learning_rate": 0.0005, "loss": 1.7052078247070312, "step": 14042 }, { "epoch": 1.1070555774536854, "grad_norm": 0.5167166590690613, "learning_rate": 0.0005, "loss": 1.73722243309021, "step": 14043 }, { "epoch": 1.1071344107213243, "grad_norm": 0.5780044794082642, "learning_rate": 0.0005, "loss": 1.8584527969360352, "step": 14044 }, { "epoch": 1.1072132439889633, "grad_norm": 0.633338212966919, "learning_rate": 0.0005, "loss": 1.5999155044555664, "step": 14045 }, { "epoch": 1.1072920772566022, "grad_norm": 0.4188178777694702, "learning_rate": 0.0005, "loss": 1.5073986053466797, "step": 14046 }, { "epoch": 1.1073709105242413, "grad_norm": 1.1573289632797241, "learning_rate": 0.0005, "loss": 1.7299275398254395, "step": 14047 }, { "epoch": 1.1074497437918802, "grad_norm": 0.5297644734382629, "learning_rate": 0.0005, "loss": 1.629817008972168, "step": 14048 }, { "epoch": 1.1075285770595191, "grad_norm": 0.6205182075500488, "learning_rate": 0.0005, "loss": 1.9101357460021973, "step": 14049 }, { "epoch": 1.107607410327158, "grad_norm": 0.8632319569587708, "learning_rate": 0.0005, "loss": 1.5356297492980957, "step": 14050 }, { "epoch": 1.107686243594797, "grad_norm": 0.8173779845237732, "learning_rate": 0.0005, "loss": 1.613213300704956, "step": 14051 }, { "epoch": 1.1077650768624359, "grad_norm": 0.5101737380027771, "learning_rate": 0.0005, "loss": 1.6479476690292358, "step": 14052 }, { "epoch": 1.107843910130075, "grad_norm": 0.6042264699935913, "learning_rate": 0.0005, "loss": 1.5616610050201416, "step": 14053 }, { "epoch": 1.107922743397714, "grad_norm": 0.49922215938568115, "learning_rate": 0.0005, "loss": 1.5938154458999634, "step": 14054 }, { "epoch": 1.1080015766653528, "grad_norm": 0.6541813611984253, "learning_rate": 0.0005, "loss": 1.511103630065918, "step": 14055 }, { "epoch": 1.1080804099329917, "grad_norm": 0.5848384499549866, "learning_rate": 0.0005, "loss": 1.7803646326065063, "step": 14056 }, { "epoch": 1.1081592432006306, "grad_norm": 0.4760589599609375, "learning_rate": 0.0005, "loss": 1.5198352336883545, "step": 14057 }, { "epoch": 1.1082380764682696, "grad_norm": 0.6212893724441528, "learning_rate": 0.0005, "loss": 1.5972034931182861, "step": 14058 }, { "epoch": 1.1083169097359085, "grad_norm": 0.6510147452354431, "learning_rate": 0.0005, "loss": 1.8315997123718262, "step": 14059 }, { "epoch": 1.1083957430035476, "grad_norm": 0.7276162505149841, "learning_rate": 0.0005, "loss": 1.8722646236419678, "step": 14060 }, { "epoch": 1.1084745762711865, "grad_norm": 0.5448331832885742, "learning_rate": 0.0005, "loss": 1.6710963249206543, "step": 14061 }, { "epoch": 1.1085534095388254, "grad_norm": 0.7308654189109802, "learning_rate": 0.0005, "loss": 1.6941964626312256, "step": 14062 }, { "epoch": 1.1086322428064643, "grad_norm": 0.564210832118988, "learning_rate": 0.0005, "loss": 1.5521786212921143, "step": 14063 }, { "epoch": 1.1087110760741032, "grad_norm": 0.6073871850967407, "learning_rate": 0.0005, "loss": 1.8173232078552246, "step": 14064 }, { "epoch": 1.1087899093417422, "grad_norm": 0.5114420652389526, "learning_rate": 0.0005, "loss": 1.658531904220581, "step": 14065 }, { "epoch": 1.108868742609381, "grad_norm": 0.6866733431816101, "learning_rate": 0.0005, "loss": 1.7845813035964966, "step": 14066 }, { "epoch": 1.1089475758770202, "grad_norm": 0.569064736366272, "learning_rate": 0.0005, "loss": 1.6902987957000732, "step": 14067 }, { "epoch": 1.109026409144659, "grad_norm": 0.5260558724403381, "learning_rate": 0.0005, "loss": 1.6484768390655518, "step": 14068 }, { "epoch": 1.109105242412298, "grad_norm": 0.539430558681488, "learning_rate": 0.0005, "loss": 1.6805064678192139, "step": 14069 }, { "epoch": 1.109184075679937, "grad_norm": 0.5407657623291016, "learning_rate": 0.0005, "loss": 1.4917060136795044, "step": 14070 }, { "epoch": 1.1092629089475758, "grad_norm": 0.5741651058197021, "learning_rate": 0.0005, "loss": 1.5330371856689453, "step": 14071 }, { "epoch": 1.1093417422152148, "grad_norm": 0.5631844997406006, "learning_rate": 0.0005, "loss": 1.8226211071014404, "step": 14072 }, { "epoch": 1.1094205754828537, "grad_norm": 0.5994653701782227, "learning_rate": 0.0005, "loss": 1.6070204973220825, "step": 14073 }, { "epoch": 1.1094994087504928, "grad_norm": 0.5059816241264343, "learning_rate": 0.0005, "loss": 1.5914700031280518, "step": 14074 }, { "epoch": 1.1095782420181317, "grad_norm": 0.5284014940261841, "learning_rate": 0.0005, "loss": 1.4651882648468018, "step": 14075 }, { "epoch": 1.1096570752857706, "grad_norm": 0.528197169303894, "learning_rate": 0.0005, "loss": 1.7890944480895996, "step": 14076 }, { "epoch": 1.1097359085534095, "grad_norm": 0.4346868097782135, "learning_rate": 0.0005, "loss": 1.5566842555999756, "step": 14077 }, { "epoch": 1.1098147418210484, "grad_norm": 0.6318454146385193, "learning_rate": 0.0005, "loss": 1.6064881086349487, "step": 14078 }, { "epoch": 1.1098935750886874, "grad_norm": 0.5699800252914429, "learning_rate": 0.0005, "loss": 1.746584177017212, "step": 14079 }, { "epoch": 1.1099724083563263, "grad_norm": 0.5012221932411194, "learning_rate": 0.0005, "loss": 1.6938881874084473, "step": 14080 }, { "epoch": 1.1100512416239654, "grad_norm": 0.5390148758888245, "learning_rate": 0.0005, "loss": 1.8418811559677124, "step": 14081 }, { "epoch": 1.1101300748916043, "grad_norm": 0.5128952860832214, "learning_rate": 0.0005, "loss": 1.4937968254089355, "step": 14082 }, { "epoch": 1.1102089081592432, "grad_norm": 0.6465287208557129, "learning_rate": 0.0005, "loss": 1.722963809967041, "step": 14083 }, { "epoch": 1.1102877414268821, "grad_norm": 0.5509558916091919, "learning_rate": 0.0005, "loss": 1.7075324058532715, "step": 14084 }, { "epoch": 1.110366574694521, "grad_norm": 0.5497403144836426, "learning_rate": 0.0005, "loss": 1.7963316440582275, "step": 14085 }, { "epoch": 1.11044540796216, "grad_norm": 0.5834263563156128, "learning_rate": 0.0005, "loss": 1.8973814249038696, "step": 14086 }, { "epoch": 1.110524241229799, "grad_norm": 0.5129197835922241, "learning_rate": 0.0005, "loss": 1.7749195098876953, "step": 14087 }, { "epoch": 1.110603074497438, "grad_norm": 0.4621793031692505, "learning_rate": 0.0005, "loss": 1.695656180381775, "step": 14088 }, { "epoch": 1.110681907765077, "grad_norm": 0.5082611441612244, "learning_rate": 0.0005, "loss": 1.5880236625671387, "step": 14089 }, { "epoch": 1.1107607410327158, "grad_norm": 0.4929165840148926, "learning_rate": 0.0005, "loss": 1.5911498069763184, "step": 14090 }, { "epoch": 1.1108395743003547, "grad_norm": 0.6112847328186035, "learning_rate": 0.0005, "loss": 1.8832508325576782, "step": 14091 }, { "epoch": 1.1109184075679936, "grad_norm": 0.479808509349823, "learning_rate": 0.0005, "loss": 1.5741519927978516, "step": 14092 }, { "epoch": 1.1109972408356326, "grad_norm": 0.5967225432395935, "learning_rate": 0.0005, "loss": 1.8990066051483154, "step": 14093 }, { "epoch": 1.1110760741032717, "grad_norm": 0.5408843755722046, "learning_rate": 0.0005, "loss": 1.7665014266967773, "step": 14094 }, { "epoch": 1.1111549073709106, "grad_norm": 0.6210362911224365, "learning_rate": 0.0005, "loss": 1.7252187728881836, "step": 14095 }, { "epoch": 1.1112337406385495, "grad_norm": 0.5568244457244873, "learning_rate": 0.0005, "loss": 1.6078855991363525, "step": 14096 }, { "epoch": 1.1113125739061884, "grad_norm": 0.6712204813957214, "learning_rate": 0.0005, "loss": 1.7429049015045166, "step": 14097 }, { "epoch": 1.1113914071738273, "grad_norm": 1.2652219533920288, "learning_rate": 0.0005, "loss": 1.68451988697052, "step": 14098 }, { "epoch": 1.1114702404414663, "grad_norm": 0.5341976881027222, "learning_rate": 0.0005, "loss": 1.6553184986114502, "step": 14099 }, { "epoch": 1.1115490737091052, "grad_norm": 0.521737277507782, "learning_rate": 0.0005, "loss": 1.7064697742462158, "step": 14100 }, { "epoch": 1.1116279069767443, "grad_norm": 0.6279134154319763, "learning_rate": 0.0005, "loss": 1.5554804801940918, "step": 14101 }, { "epoch": 1.1117067402443832, "grad_norm": 0.5265687704086304, "learning_rate": 0.0005, "loss": 1.8226137161254883, "step": 14102 }, { "epoch": 1.1117855735120221, "grad_norm": 0.5587771534919739, "learning_rate": 0.0005, "loss": 1.635056734085083, "step": 14103 }, { "epoch": 1.111864406779661, "grad_norm": 0.5366615056991577, "learning_rate": 0.0005, "loss": 1.3784313201904297, "step": 14104 }, { "epoch": 1.1119432400473, "grad_norm": 0.5564573407173157, "learning_rate": 0.0005, "loss": 1.5509669780731201, "step": 14105 }, { "epoch": 1.1120220733149389, "grad_norm": 0.41390568017959595, "learning_rate": 0.0005, "loss": 1.4314656257629395, "step": 14106 }, { "epoch": 1.1121009065825778, "grad_norm": 0.426910400390625, "learning_rate": 0.0005, "loss": 1.3651349544525146, "step": 14107 }, { "epoch": 1.112179739850217, "grad_norm": 0.5468862652778625, "learning_rate": 0.0005, "loss": 1.6179540157318115, "step": 14108 }, { "epoch": 1.1122585731178558, "grad_norm": 0.4990772008895874, "learning_rate": 0.0005, "loss": 1.6722707748413086, "step": 14109 }, { "epoch": 1.1123374063854947, "grad_norm": 0.48876693844795227, "learning_rate": 0.0005, "loss": 1.7051655054092407, "step": 14110 }, { "epoch": 1.1124162396531336, "grad_norm": 0.48348215222358704, "learning_rate": 0.0005, "loss": 1.6678717136383057, "step": 14111 }, { "epoch": 1.1124950729207725, "grad_norm": 0.6593735814094543, "learning_rate": 0.0005, "loss": 1.7026801109313965, "step": 14112 }, { "epoch": 1.1125739061884115, "grad_norm": 0.6126489043235779, "learning_rate": 0.0005, "loss": 1.8146425485610962, "step": 14113 }, { "epoch": 1.1126527394560504, "grad_norm": 0.4784631133079529, "learning_rate": 0.0005, "loss": 1.6110117435455322, "step": 14114 }, { "epoch": 1.1127315727236895, "grad_norm": 0.6179253458976746, "learning_rate": 0.0005, "loss": 1.5924227237701416, "step": 14115 }, { "epoch": 1.1128104059913284, "grad_norm": 0.7511506676673889, "learning_rate": 0.0005, "loss": 1.6399285793304443, "step": 14116 }, { "epoch": 1.1128892392589673, "grad_norm": 0.5576344728469849, "learning_rate": 0.0005, "loss": 1.4817888736724854, "step": 14117 }, { "epoch": 1.1129680725266062, "grad_norm": 0.5288722515106201, "learning_rate": 0.0005, "loss": 1.550968050956726, "step": 14118 }, { "epoch": 1.1130469057942451, "grad_norm": 0.431018590927124, "learning_rate": 0.0005, "loss": 1.545426845550537, "step": 14119 }, { "epoch": 1.113125739061884, "grad_norm": 0.5261527895927429, "learning_rate": 0.0005, "loss": 1.6720448732376099, "step": 14120 }, { "epoch": 1.113204572329523, "grad_norm": 0.5505680441856384, "learning_rate": 0.0005, "loss": 1.6422514915466309, "step": 14121 }, { "epoch": 1.113283405597162, "grad_norm": 0.6551259756088257, "learning_rate": 0.0005, "loss": 2.038947343826294, "step": 14122 }, { "epoch": 1.113362238864801, "grad_norm": 0.5087364315986633, "learning_rate": 0.0005, "loss": 1.6264591217041016, "step": 14123 }, { "epoch": 1.11344107213244, "grad_norm": 0.5396706461906433, "learning_rate": 0.0005, "loss": 1.7786164283752441, "step": 14124 }, { "epoch": 1.1135199054000788, "grad_norm": 0.7127491235733032, "learning_rate": 0.0005, "loss": 1.8029518127441406, "step": 14125 }, { "epoch": 1.1135987386677177, "grad_norm": 0.5915989875793457, "learning_rate": 0.0005, "loss": 1.8092230558395386, "step": 14126 }, { "epoch": 1.1136775719353567, "grad_norm": 0.4722263514995575, "learning_rate": 0.0005, "loss": 1.6760289669036865, "step": 14127 }, { "epoch": 1.1137564052029956, "grad_norm": 0.48447385430336, "learning_rate": 0.0005, "loss": 1.7510292530059814, "step": 14128 }, { "epoch": 1.1138352384706347, "grad_norm": 0.5471850037574768, "learning_rate": 0.0005, "loss": 1.5508480072021484, "step": 14129 }, { "epoch": 1.1139140717382736, "grad_norm": 0.5066393613815308, "learning_rate": 0.0005, "loss": 1.7130258083343506, "step": 14130 }, { "epoch": 1.1139929050059125, "grad_norm": 0.48519718647003174, "learning_rate": 0.0005, "loss": 1.4937336444854736, "step": 14131 }, { "epoch": 1.1140717382735514, "grad_norm": 0.5087928771972656, "learning_rate": 0.0005, "loss": 1.4636033773422241, "step": 14132 }, { "epoch": 1.1141505715411903, "grad_norm": 0.5851499438285828, "learning_rate": 0.0005, "loss": 1.8171497583389282, "step": 14133 }, { "epoch": 1.1142294048088293, "grad_norm": 0.5513944625854492, "learning_rate": 0.0005, "loss": 1.9414602518081665, "step": 14134 }, { "epoch": 1.1143082380764682, "grad_norm": 0.47255271673202515, "learning_rate": 0.0005, "loss": 1.6300454139709473, "step": 14135 }, { "epoch": 1.1143870713441073, "grad_norm": 0.581919252872467, "learning_rate": 0.0005, "loss": 1.6533327102661133, "step": 14136 }, { "epoch": 1.1144659046117462, "grad_norm": 0.6141672730445862, "learning_rate": 0.0005, "loss": 1.7184381484985352, "step": 14137 }, { "epoch": 1.1145447378793851, "grad_norm": 0.5039578676223755, "learning_rate": 0.0005, "loss": 1.5221309661865234, "step": 14138 }, { "epoch": 1.114623571147024, "grad_norm": 0.4794485867023468, "learning_rate": 0.0005, "loss": 1.657241702079773, "step": 14139 }, { "epoch": 1.114702404414663, "grad_norm": 0.4701903462409973, "learning_rate": 0.0005, "loss": 1.6478519439697266, "step": 14140 }, { "epoch": 1.1147812376823019, "grad_norm": 0.6496021747589111, "learning_rate": 0.0005, "loss": 1.7465293407440186, "step": 14141 }, { "epoch": 1.1148600709499408, "grad_norm": 0.60770183801651, "learning_rate": 0.0005, "loss": 1.7388081550598145, "step": 14142 }, { "epoch": 1.11493890421758, "grad_norm": 0.690835177898407, "learning_rate": 0.0005, "loss": 1.7911514043807983, "step": 14143 }, { "epoch": 1.1150177374852188, "grad_norm": 0.6568775773048401, "learning_rate": 0.0005, "loss": 1.8661223649978638, "step": 14144 }, { "epoch": 1.1150965707528577, "grad_norm": 0.7220645546913147, "learning_rate": 0.0005, "loss": 1.8258188962936401, "step": 14145 }, { "epoch": 1.1151754040204966, "grad_norm": 0.7486995458602905, "learning_rate": 0.0005, "loss": 1.9182231426239014, "step": 14146 }, { "epoch": 1.1152542372881356, "grad_norm": 0.4826543629169464, "learning_rate": 0.0005, "loss": 1.589048147201538, "step": 14147 }, { "epoch": 1.1153330705557745, "grad_norm": 1.8275901079177856, "learning_rate": 0.0005, "loss": 1.6714428663253784, "step": 14148 }, { "epoch": 1.1154119038234134, "grad_norm": 0.5188758969306946, "learning_rate": 0.0005, "loss": 1.3483691215515137, "step": 14149 }, { "epoch": 1.1154907370910525, "grad_norm": 0.5496590733528137, "learning_rate": 0.0005, "loss": 1.8325121402740479, "step": 14150 }, { "epoch": 1.1155695703586914, "grad_norm": 0.6269723773002625, "learning_rate": 0.0005, "loss": 1.6667739152908325, "step": 14151 }, { "epoch": 1.1156484036263303, "grad_norm": 0.5576602220535278, "learning_rate": 0.0005, "loss": 1.6016254425048828, "step": 14152 }, { "epoch": 1.1157272368939692, "grad_norm": 0.519395112991333, "learning_rate": 0.0005, "loss": 1.552657127380371, "step": 14153 }, { "epoch": 1.1158060701616082, "grad_norm": 0.5261523127555847, "learning_rate": 0.0005, "loss": 1.986201286315918, "step": 14154 }, { "epoch": 1.115884903429247, "grad_norm": 0.48913639783859253, "learning_rate": 0.0005, "loss": 1.3980580568313599, "step": 14155 }, { "epoch": 1.115963736696886, "grad_norm": 0.6205279231071472, "learning_rate": 0.0005, "loss": 1.7639946937561035, "step": 14156 }, { "epoch": 1.116042569964525, "grad_norm": 0.5576470494270325, "learning_rate": 0.0005, "loss": 1.7316292524337769, "step": 14157 }, { "epoch": 1.116121403232164, "grad_norm": 0.5964882373809814, "learning_rate": 0.0005, "loss": 1.7141001224517822, "step": 14158 }, { "epoch": 1.116200236499803, "grad_norm": 0.5931586623191833, "learning_rate": 0.0005, "loss": 1.7864432334899902, "step": 14159 }, { "epoch": 1.1162790697674418, "grad_norm": 0.7298742532730103, "learning_rate": 0.0005, "loss": 1.5085417032241821, "step": 14160 }, { "epoch": 1.1163579030350808, "grad_norm": 0.6979408264160156, "learning_rate": 0.0005, "loss": 1.976531744003296, "step": 14161 }, { "epoch": 1.1164367363027197, "grad_norm": 0.4977814257144928, "learning_rate": 0.0005, "loss": 1.546210765838623, "step": 14162 }, { "epoch": 1.1165155695703586, "grad_norm": 0.537792980670929, "learning_rate": 0.0005, "loss": 1.5956885814666748, "step": 14163 }, { "epoch": 1.1165944028379977, "grad_norm": 0.6321941018104553, "learning_rate": 0.0005, "loss": 1.7252475023269653, "step": 14164 }, { "epoch": 1.1166732361056366, "grad_norm": 0.6763855218887329, "learning_rate": 0.0005, "loss": 1.7815099954605103, "step": 14165 }, { "epoch": 1.1167520693732755, "grad_norm": 0.7112977504730225, "learning_rate": 0.0005, "loss": 1.9636203050613403, "step": 14166 }, { "epoch": 1.1168309026409144, "grad_norm": 0.4918462038040161, "learning_rate": 0.0005, "loss": 1.5705280303955078, "step": 14167 }, { "epoch": 1.1169097359085534, "grad_norm": 0.5377641916275024, "learning_rate": 0.0005, "loss": 1.5746320486068726, "step": 14168 }, { "epoch": 1.1169885691761923, "grad_norm": 0.91446852684021, "learning_rate": 0.0005, "loss": 1.50261390209198, "step": 14169 }, { "epoch": 1.1170674024438312, "grad_norm": 0.5319551825523376, "learning_rate": 0.0005, "loss": 1.7231106758117676, "step": 14170 }, { "epoch": 1.1171462357114703, "grad_norm": 0.5471636652946472, "learning_rate": 0.0005, "loss": 1.7231597900390625, "step": 14171 }, { "epoch": 1.1172250689791092, "grad_norm": 0.5643971562385559, "learning_rate": 0.0005, "loss": 1.7598563432693481, "step": 14172 }, { "epoch": 1.1173039022467481, "grad_norm": 0.5945037007331848, "learning_rate": 0.0005, "loss": 1.7632815837860107, "step": 14173 }, { "epoch": 1.117382735514387, "grad_norm": 0.5630118250846863, "learning_rate": 0.0005, "loss": 1.6875134706497192, "step": 14174 }, { "epoch": 1.117461568782026, "grad_norm": 0.6691963076591492, "learning_rate": 0.0005, "loss": 1.5764201879501343, "step": 14175 }, { "epoch": 1.1175404020496649, "grad_norm": 0.5092188119888306, "learning_rate": 0.0005, "loss": 1.5672235488891602, "step": 14176 }, { "epoch": 1.117619235317304, "grad_norm": 0.522921085357666, "learning_rate": 0.0005, "loss": 1.579479694366455, "step": 14177 }, { "epoch": 1.117698068584943, "grad_norm": 0.6097573041915894, "learning_rate": 0.0005, "loss": 1.7986531257629395, "step": 14178 }, { "epoch": 1.1177769018525818, "grad_norm": 0.6328471302986145, "learning_rate": 0.0005, "loss": 1.6892695426940918, "step": 14179 }, { "epoch": 1.1178557351202207, "grad_norm": 0.6394714117050171, "learning_rate": 0.0005, "loss": 1.7187378406524658, "step": 14180 }, { "epoch": 1.1179345683878597, "grad_norm": 0.5136781930923462, "learning_rate": 0.0005, "loss": 1.6872475147247314, "step": 14181 }, { "epoch": 1.1180134016554986, "grad_norm": 0.4786726236343384, "learning_rate": 0.0005, "loss": 1.6016523838043213, "step": 14182 }, { "epoch": 1.1180922349231375, "grad_norm": 1.68412446975708, "learning_rate": 0.0005, "loss": 1.6118450164794922, "step": 14183 }, { "epoch": 1.1181710681907766, "grad_norm": 0.5824786424636841, "learning_rate": 0.0005, "loss": 1.6401125192642212, "step": 14184 }, { "epoch": 1.1182499014584155, "grad_norm": 0.505612850189209, "learning_rate": 0.0005, "loss": 1.6465160846710205, "step": 14185 }, { "epoch": 1.1183287347260544, "grad_norm": 0.5804927349090576, "learning_rate": 0.0005, "loss": 1.7065978050231934, "step": 14186 }, { "epoch": 1.1184075679936933, "grad_norm": 0.4791603982448578, "learning_rate": 0.0005, "loss": 1.6201878786087036, "step": 14187 }, { "epoch": 1.1184864012613323, "grad_norm": 1.0046701431274414, "learning_rate": 0.0005, "loss": 1.7647669315338135, "step": 14188 }, { "epoch": 1.1185652345289712, "grad_norm": 0.5530943870544434, "learning_rate": 0.0005, "loss": 1.6684556007385254, "step": 14189 }, { "epoch": 1.11864406779661, "grad_norm": 5.936802387237549, "learning_rate": 0.0005, "loss": 1.693925380706787, "step": 14190 }, { "epoch": 1.1187229010642492, "grad_norm": 0.4526240825653076, "learning_rate": 0.0005, "loss": 1.4952973127365112, "step": 14191 }, { "epoch": 1.1188017343318881, "grad_norm": 0.47926437854766846, "learning_rate": 0.0005, "loss": 1.7351093292236328, "step": 14192 }, { "epoch": 1.118880567599527, "grad_norm": 0.6058984994888306, "learning_rate": 0.0005, "loss": 1.5551538467407227, "step": 14193 }, { "epoch": 1.118959400867166, "grad_norm": 0.551226794719696, "learning_rate": 0.0005, "loss": 1.4531126022338867, "step": 14194 }, { "epoch": 1.1190382341348049, "grad_norm": 0.46711447834968567, "learning_rate": 0.0005, "loss": 1.5600433349609375, "step": 14195 }, { "epoch": 1.1191170674024438, "grad_norm": 2.175809621810913, "learning_rate": 0.0005, "loss": 1.6357288360595703, "step": 14196 }, { "epoch": 1.1191959006700827, "grad_norm": 0.5289312601089478, "learning_rate": 0.0005, "loss": 1.7796375751495361, "step": 14197 }, { "epoch": 1.1192747339377218, "grad_norm": 0.645768404006958, "learning_rate": 0.0005, "loss": 1.5330381393432617, "step": 14198 }, { "epoch": 1.1193535672053607, "grad_norm": 0.5397375226020813, "learning_rate": 0.0005, "loss": 1.5772075653076172, "step": 14199 }, { "epoch": 1.1194324004729996, "grad_norm": 0.6146400570869446, "learning_rate": 0.0005, "loss": 1.6265411376953125, "step": 14200 }, { "epoch": 1.1195112337406385, "grad_norm": 0.5940685868263245, "learning_rate": 0.0005, "loss": 1.6223509311676025, "step": 14201 }, { "epoch": 1.1195900670082775, "grad_norm": 0.5274398922920227, "learning_rate": 0.0005, "loss": 1.5733468532562256, "step": 14202 }, { "epoch": 1.1196689002759164, "grad_norm": 0.6329453587532043, "learning_rate": 0.0005, "loss": 1.597826361656189, "step": 14203 }, { "epoch": 1.1197477335435553, "grad_norm": 0.47708821296691895, "learning_rate": 0.0005, "loss": 1.530733585357666, "step": 14204 }, { "epoch": 1.1198265668111944, "grad_norm": 0.484029620885849, "learning_rate": 0.0005, "loss": 1.6739192008972168, "step": 14205 }, { "epoch": 1.1199054000788333, "grad_norm": 0.8320679664611816, "learning_rate": 0.0005, "loss": 1.9547200202941895, "step": 14206 }, { "epoch": 1.1199842333464722, "grad_norm": 0.47570937871932983, "learning_rate": 0.0005, "loss": 1.5756659507751465, "step": 14207 }, { "epoch": 1.1200630666141111, "grad_norm": 0.5789851546287537, "learning_rate": 0.0005, "loss": 1.6883864402770996, "step": 14208 }, { "epoch": 1.12014189988175, "grad_norm": 0.517194926738739, "learning_rate": 0.0005, "loss": 1.5859405994415283, "step": 14209 }, { "epoch": 1.120220733149389, "grad_norm": 0.5395468473434448, "learning_rate": 0.0005, "loss": 1.6663750410079956, "step": 14210 }, { "epoch": 1.120299566417028, "grad_norm": 0.6863217353820801, "learning_rate": 0.0005, "loss": 1.6785166263580322, "step": 14211 }, { "epoch": 1.120378399684667, "grad_norm": 0.544524073600769, "learning_rate": 0.0005, "loss": 1.8846474885940552, "step": 14212 }, { "epoch": 1.120457232952306, "grad_norm": 0.5649818181991577, "learning_rate": 0.0005, "loss": 1.7679892778396606, "step": 14213 }, { "epoch": 1.1205360662199448, "grad_norm": 0.5393256545066833, "learning_rate": 0.0005, "loss": 1.5181413888931274, "step": 14214 }, { "epoch": 1.1206148994875837, "grad_norm": 0.6284219026565552, "learning_rate": 0.0005, "loss": 1.6359131336212158, "step": 14215 }, { "epoch": 1.1206937327552227, "grad_norm": 0.552455723285675, "learning_rate": 0.0005, "loss": 1.8730486631393433, "step": 14216 }, { "epoch": 1.1207725660228616, "grad_norm": 0.4889737665653229, "learning_rate": 0.0005, "loss": 1.6851627826690674, "step": 14217 }, { "epoch": 1.1208513992905007, "grad_norm": 0.48912253975868225, "learning_rate": 0.0005, "loss": 1.7105008363723755, "step": 14218 }, { "epoch": 1.1209302325581396, "grad_norm": 0.8338659405708313, "learning_rate": 0.0005, "loss": 1.8772774934768677, "step": 14219 }, { "epoch": 1.1210090658257785, "grad_norm": 0.5753021836280823, "learning_rate": 0.0005, "loss": 1.864778995513916, "step": 14220 }, { "epoch": 1.1210878990934174, "grad_norm": 0.4993921220302582, "learning_rate": 0.0005, "loss": 1.5897705554962158, "step": 14221 }, { "epoch": 1.1211667323610564, "grad_norm": 0.4787963032722473, "learning_rate": 0.0005, "loss": 1.5618408918380737, "step": 14222 }, { "epoch": 1.1212455656286953, "grad_norm": 0.5067249536514282, "learning_rate": 0.0005, "loss": 1.5563464164733887, "step": 14223 }, { "epoch": 1.1213243988963342, "grad_norm": 0.6046735644340515, "learning_rate": 0.0005, "loss": 1.663964033126831, "step": 14224 }, { "epoch": 1.1214032321639733, "grad_norm": 0.5825841426849365, "learning_rate": 0.0005, "loss": 1.5705630779266357, "step": 14225 }, { "epoch": 1.1214820654316122, "grad_norm": 0.7077191472053528, "learning_rate": 0.0005, "loss": 1.5640151500701904, "step": 14226 }, { "epoch": 1.1215608986992511, "grad_norm": 0.546785295009613, "learning_rate": 0.0005, "loss": 1.7770254611968994, "step": 14227 }, { "epoch": 1.12163973196689, "grad_norm": 0.47519564628601074, "learning_rate": 0.0005, "loss": 1.5073589086532593, "step": 14228 }, { "epoch": 1.121718565234529, "grad_norm": 0.604467511177063, "learning_rate": 0.0005, "loss": 1.6815705299377441, "step": 14229 }, { "epoch": 1.1217973985021679, "grad_norm": 0.5192369818687439, "learning_rate": 0.0005, "loss": 1.847084403038025, "step": 14230 }, { "epoch": 1.1218762317698068, "grad_norm": 0.6565727591514587, "learning_rate": 0.0005, "loss": 1.8412089347839355, "step": 14231 }, { "epoch": 1.121955065037446, "grad_norm": 0.45373621582984924, "learning_rate": 0.0005, "loss": 1.6254345178604126, "step": 14232 }, { "epoch": 1.1220338983050848, "grad_norm": 0.507694661617279, "learning_rate": 0.0005, "loss": 1.6756892204284668, "step": 14233 }, { "epoch": 1.1221127315727237, "grad_norm": 0.776747465133667, "learning_rate": 0.0005, "loss": 1.9507148265838623, "step": 14234 }, { "epoch": 1.1221915648403626, "grad_norm": 0.6380991339683533, "learning_rate": 0.0005, "loss": 1.8601274490356445, "step": 14235 }, { "epoch": 1.1222703981080016, "grad_norm": 0.48775872588157654, "learning_rate": 0.0005, "loss": 1.6265313625335693, "step": 14236 }, { "epoch": 1.1223492313756405, "grad_norm": 0.5503037571907043, "learning_rate": 0.0005, "loss": 1.7453234195709229, "step": 14237 }, { "epoch": 1.1224280646432794, "grad_norm": 0.5823713541030884, "learning_rate": 0.0005, "loss": 1.802199125289917, "step": 14238 }, { "epoch": 1.1225068979109185, "grad_norm": 0.5753485560417175, "learning_rate": 0.0005, "loss": 1.8334290981292725, "step": 14239 }, { "epoch": 1.1225857311785574, "grad_norm": 0.5318880081176758, "learning_rate": 0.0005, "loss": 1.7831110954284668, "step": 14240 }, { "epoch": 1.1226645644461963, "grad_norm": 0.5817672610282898, "learning_rate": 0.0005, "loss": 1.6858282089233398, "step": 14241 }, { "epoch": 1.1227433977138352, "grad_norm": 0.6424186825752258, "learning_rate": 0.0005, "loss": 1.6658759117126465, "step": 14242 }, { "epoch": 1.1228222309814742, "grad_norm": 0.5946460962295532, "learning_rate": 0.0005, "loss": 1.6142387390136719, "step": 14243 }, { "epoch": 1.122901064249113, "grad_norm": 0.6112849116325378, "learning_rate": 0.0005, "loss": 1.9839465618133545, "step": 14244 }, { "epoch": 1.122979897516752, "grad_norm": 0.5653818249702454, "learning_rate": 0.0005, "loss": 1.6190614700317383, "step": 14245 }, { "epoch": 1.1230587307843911, "grad_norm": 0.5113243460655212, "learning_rate": 0.0005, "loss": 1.6705121994018555, "step": 14246 }, { "epoch": 1.12313756405203, "grad_norm": 0.5543169379234314, "learning_rate": 0.0005, "loss": 1.9343891143798828, "step": 14247 }, { "epoch": 1.123216397319669, "grad_norm": 0.501844048500061, "learning_rate": 0.0005, "loss": 1.6095823049545288, "step": 14248 }, { "epoch": 1.1232952305873078, "grad_norm": 0.8892223238945007, "learning_rate": 0.0005, "loss": 1.6742727756500244, "step": 14249 }, { "epoch": 1.1233740638549468, "grad_norm": 0.46954217553138733, "learning_rate": 0.0005, "loss": 1.7054884433746338, "step": 14250 }, { "epoch": 1.1234528971225857, "grad_norm": 0.9280720949172974, "learning_rate": 0.0005, "loss": 1.763139247894287, "step": 14251 }, { "epoch": 1.1235317303902246, "grad_norm": 0.5157151222229004, "learning_rate": 0.0005, "loss": 1.7762311697006226, "step": 14252 }, { "epoch": 1.1236105636578637, "grad_norm": 0.5939319133758545, "learning_rate": 0.0005, "loss": 1.5633267164230347, "step": 14253 }, { "epoch": 1.1236893969255026, "grad_norm": 0.4635508358478546, "learning_rate": 0.0005, "loss": 1.827622890472412, "step": 14254 }, { "epoch": 1.1237682301931415, "grad_norm": 0.5804438591003418, "learning_rate": 0.0005, "loss": 1.8325095176696777, "step": 14255 }, { "epoch": 1.1238470634607804, "grad_norm": 0.49737006425857544, "learning_rate": 0.0005, "loss": 1.5880780220031738, "step": 14256 }, { "epoch": 1.1239258967284194, "grad_norm": 0.45037198066711426, "learning_rate": 0.0005, "loss": 1.7411574125289917, "step": 14257 }, { "epoch": 1.1240047299960583, "grad_norm": 0.5319235324859619, "learning_rate": 0.0005, "loss": 1.6013922691345215, "step": 14258 }, { "epoch": 1.1240835632636972, "grad_norm": 0.5285080671310425, "learning_rate": 0.0005, "loss": 1.4218738079071045, "step": 14259 }, { "epoch": 1.1241623965313363, "grad_norm": 0.5834307074546814, "learning_rate": 0.0005, "loss": 1.7041207551956177, "step": 14260 }, { "epoch": 1.1242412297989752, "grad_norm": 0.7113031148910522, "learning_rate": 0.0005, "loss": 1.8303396701812744, "step": 14261 }, { "epoch": 1.1243200630666141, "grad_norm": 0.6158239245414734, "learning_rate": 0.0005, "loss": 1.7532844543457031, "step": 14262 }, { "epoch": 1.124398896334253, "grad_norm": 0.4973711669445038, "learning_rate": 0.0005, "loss": 1.720464825630188, "step": 14263 }, { "epoch": 1.124477729601892, "grad_norm": 0.5833487510681152, "learning_rate": 0.0005, "loss": 1.777909517288208, "step": 14264 }, { "epoch": 1.1245565628695309, "grad_norm": 0.5474655032157898, "learning_rate": 0.0005, "loss": 1.6285045146942139, "step": 14265 }, { "epoch": 1.1246353961371698, "grad_norm": 0.5476786494255066, "learning_rate": 0.0005, "loss": 1.6396408081054688, "step": 14266 }, { "epoch": 1.124714229404809, "grad_norm": 0.6056952476501465, "learning_rate": 0.0005, "loss": 1.7620396614074707, "step": 14267 }, { "epoch": 1.1247930626724478, "grad_norm": 0.5058304667472839, "learning_rate": 0.0005, "loss": 1.3814340829849243, "step": 14268 }, { "epoch": 1.1248718959400867, "grad_norm": 0.5499755144119263, "learning_rate": 0.0005, "loss": 1.8534841537475586, "step": 14269 }, { "epoch": 1.1249507292077257, "grad_norm": 0.5191597938537598, "learning_rate": 0.0005, "loss": 1.7715647220611572, "step": 14270 }, { "epoch": 1.1250295624753646, "grad_norm": 0.4585578441619873, "learning_rate": 0.0005, "loss": 1.809067964553833, "step": 14271 }, { "epoch": 1.1251083957430035, "grad_norm": 0.4760035574436188, "learning_rate": 0.0005, "loss": 1.5763486623764038, "step": 14272 }, { "epoch": 1.1251872290106424, "grad_norm": 0.6156794428825378, "learning_rate": 0.0005, "loss": 1.7707972526550293, "step": 14273 }, { "epoch": 1.1252660622782815, "grad_norm": 0.7344043254852295, "learning_rate": 0.0005, "loss": 1.4164011478424072, "step": 14274 }, { "epoch": 1.1253448955459204, "grad_norm": 0.5889252424240112, "learning_rate": 0.0005, "loss": 1.7426249980926514, "step": 14275 }, { "epoch": 1.1254237288135593, "grad_norm": 0.5321017503738403, "learning_rate": 0.0005, "loss": 1.850119709968567, "step": 14276 }, { "epoch": 1.1255025620811983, "grad_norm": 0.5643163919448853, "learning_rate": 0.0005, "loss": 1.9417645931243896, "step": 14277 }, { "epoch": 1.1255813953488372, "grad_norm": 0.6483258605003357, "learning_rate": 0.0005, "loss": 1.6575050354003906, "step": 14278 }, { "epoch": 1.125660228616476, "grad_norm": 0.6448652148246765, "learning_rate": 0.0005, "loss": 1.3502742052078247, "step": 14279 }, { "epoch": 1.125739061884115, "grad_norm": 0.5580830574035645, "learning_rate": 0.0005, "loss": 1.5430848598480225, "step": 14280 }, { "epoch": 1.1258178951517541, "grad_norm": 0.5605588555335999, "learning_rate": 0.0005, "loss": 1.6738206148147583, "step": 14281 }, { "epoch": 1.125896728419393, "grad_norm": 0.5137878060340881, "learning_rate": 0.0005, "loss": 1.6201014518737793, "step": 14282 }, { "epoch": 1.125975561687032, "grad_norm": 0.5775144100189209, "learning_rate": 0.0005, "loss": 1.4173791408538818, "step": 14283 }, { "epoch": 1.1260543949546709, "grad_norm": 0.5285016894340515, "learning_rate": 0.0005, "loss": 1.5343117713928223, "step": 14284 }, { "epoch": 1.1261332282223098, "grad_norm": 0.43931150436401367, "learning_rate": 0.0005, "loss": 1.5932034254074097, "step": 14285 }, { "epoch": 1.1262120614899487, "grad_norm": 0.5457548499107361, "learning_rate": 0.0005, "loss": 1.626935601234436, "step": 14286 }, { "epoch": 1.1262908947575876, "grad_norm": 0.5341464877128601, "learning_rate": 0.0005, "loss": 1.7910600900650024, "step": 14287 }, { "epoch": 1.1263697280252267, "grad_norm": 0.48961517214775085, "learning_rate": 0.0005, "loss": 1.7033371925354004, "step": 14288 }, { "epoch": 1.1264485612928656, "grad_norm": 0.5300148725509644, "learning_rate": 0.0005, "loss": 1.6582236289978027, "step": 14289 }, { "epoch": 1.1265273945605045, "grad_norm": 0.617404043674469, "learning_rate": 0.0005, "loss": 1.6915123462677002, "step": 14290 }, { "epoch": 1.1266062278281435, "grad_norm": 0.6198391318321228, "learning_rate": 0.0005, "loss": 1.6047712564468384, "step": 14291 }, { "epoch": 1.1266850610957824, "grad_norm": 0.5013235211372375, "learning_rate": 0.0005, "loss": 1.710239052772522, "step": 14292 }, { "epoch": 1.1267638943634213, "grad_norm": 0.6227364540100098, "learning_rate": 0.0005, "loss": 1.6717970371246338, "step": 14293 }, { "epoch": 1.1268427276310602, "grad_norm": 0.8473483920097351, "learning_rate": 0.0005, "loss": 1.6148542165756226, "step": 14294 }, { "epoch": 1.1269215608986993, "grad_norm": 0.4835701584815979, "learning_rate": 0.0005, "loss": 1.40259850025177, "step": 14295 }, { "epoch": 1.1270003941663382, "grad_norm": 0.8001574873924255, "learning_rate": 0.0005, "loss": 1.9724010229110718, "step": 14296 }, { "epoch": 1.1270792274339771, "grad_norm": 0.5186448097229004, "learning_rate": 0.0005, "loss": 1.5287344455718994, "step": 14297 }, { "epoch": 1.127158060701616, "grad_norm": 0.6413453817367554, "learning_rate": 0.0005, "loss": 1.9132941961288452, "step": 14298 }, { "epoch": 1.127236893969255, "grad_norm": 0.6134053468704224, "learning_rate": 0.0005, "loss": 1.992562174797058, "step": 14299 }, { "epoch": 1.1273157272368939, "grad_norm": 0.5863289833068848, "learning_rate": 0.0005, "loss": 1.5296117067337036, "step": 14300 }, { "epoch": 1.1273945605045328, "grad_norm": 0.4934126138687134, "learning_rate": 0.0005, "loss": 1.5423710346221924, "step": 14301 }, { "epoch": 1.127473393772172, "grad_norm": 0.706976592540741, "learning_rate": 0.0005, "loss": 1.659794807434082, "step": 14302 }, { "epoch": 1.1275522270398108, "grad_norm": 0.9172369837760925, "learning_rate": 0.0005, "loss": 1.770162582397461, "step": 14303 }, { "epoch": 1.1276310603074498, "grad_norm": 0.6448965072631836, "learning_rate": 0.0005, "loss": 1.836951732635498, "step": 14304 }, { "epoch": 1.1277098935750887, "grad_norm": 0.8493574857711792, "learning_rate": 0.0005, "loss": 1.9233819246292114, "step": 14305 }, { "epoch": 1.1277887268427276, "grad_norm": 0.6537180542945862, "learning_rate": 0.0005, "loss": 1.5227501392364502, "step": 14306 }, { "epoch": 1.1278675601103665, "grad_norm": 0.6352951526641846, "learning_rate": 0.0005, "loss": 1.53401780128479, "step": 14307 }, { "epoch": 1.1279463933780054, "grad_norm": 1.3031408786773682, "learning_rate": 0.0005, "loss": 1.7083450555801392, "step": 14308 }, { "epoch": 1.1280252266456445, "grad_norm": 0.5737544298171997, "learning_rate": 0.0005, "loss": 1.737685203552246, "step": 14309 }, { "epoch": 1.1281040599132834, "grad_norm": 0.5424536466598511, "learning_rate": 0.0005, "loss": 1.7366262674331665, "step": 14310 }, { "epoch": 1.1281828931809224, "grad_norm": 0.6900746822357178, "learning_rate": 0.0005, "loss": 1.5155991315841675, "step": 14311 }, { "epoch": 1.1282617264485613, "grad_norm": 0.5433887243270874, "learning_rate": 0.0005, "loss": 1.76297926902771, "step": 14312 }, { "epoch": 1.1283405597162002, "grad_norm": 0.6717103123664856, "learning_rate": 0.0005, "loss": 1.6238799095153809, "step": 14313 }, { "epoch": 1.128419392983839, "grad_norm": 0.7724665999412537, "learning_rate": 0.0005, "loss": 2.0753085613250732, "step": 14314 }, { "epoch": 1.1284982262514782, "grad_norm": 0.6617595553398132, "learning_rate": 0.0005, "loss": 2.036515235900879, "step": 14315 }, { "epoch": 1.1285770595191171, "grad_norm": 0.8552690744400024, "learning_rate": 0.0005, "loss": 1.5413129329681396, "step": 14316 }, { "epoch": 1.128655892786756, "grad_norm": 0.54496169090271, "learning_rate": 0.0005, "loss": 1.6544629335403442, "step": 14317 }, { "epoch": 1.128734726054395, "grad_norm": 0.5194554328918457, "learning_rate": 0.0005, "loss": 1.3818689584732056, "step": 14318 }, { "epoch": 1.1288135593220339, "grad_norm": 0.5025997161865234, "learning_rate": 0.0005, "loss": 1.6229455471038818, "step": 14319 }, { "epoch": 1.1288923925896728, "grad_norm": 0.5598267316818237, "learning_rate": 0.0005, "loss": 1.8016996383666992, "step": 14320 }, { "epoch": 1.1289712258573117, "grad_norm": 0.5852110981941223, "learning_rate": 0.0005, "loss": 1.6821647882461548, "step": 14321 }, { "epoch": 1.1290500591249508, "grad_norm": 0.5820730328559875, "learning_rate": 0.0005, "loss": 1.8371342420578003, "step": 14322 }, { "epoch": 1.1291288923925897, "grad_norm": 0.5570569634437561, "learning_rate": 0.0005, "loss": 1.687706708908081, "step": 14323 }, { "epoch": 1.1292077256602286, "grad_norm": 0.5749927163124084, "learning_rate": 0.0005, "loss": 1.9675371646881104, "step": 14324 }, { "epoch": 1.1292865589278676, "grad_norm": 0.5169310569763184, "learning_rate": 0.0005, "loss": 1.667234182357788, "step": 14325 }, { "epoch": 1.1293653921955065, "grad_norm": 0.6179091334342957, "learning_rate": 0.0005, "loss": 1.8473279476165771, "step": 14326 }, { "epoch": 1.1294442254631454, "grad_norm": 0.43583789467811584, "learning_rate": 0.0005, "loss": 1.4973527193069458, "step": 14327 }, { "epoch": 1.1295230587307845, "grad_norm": 0.8284884691238403, "learning_rate": 0.0005, "loss": 1.8878090381622314, "step": 14328 }, { "epoch": 1.1296018919984234, "grad_norm": 0.6845372915267944, "learning_rate": 0.0005, "loss": 1.8093591928482056, "step": 14329 }, { "epoch": 1.1296807252660623, "grad_norm": 0.5713828802108765, "learning_rate": 0.0005, "loss": 1.8258367776870728, "step": 14330 }, { "epoch": 1.1297595585337012, "grad_norm": 0.602464497089386, "learning_rate": 0.0005, "loss": 1.6816973686218262, "step": 14331 }, { "epoch": 1.1298383918013402, "grad_norm": 0.49095821380615234, "learning_rate": 0.0005, "loss": 1.545393705368042, "step": 14332 }, { "epoch": 1.129917225068979, "grad_norm": 0.5238255858421326, "learning_rate": 0.0005, "loss": 1.5388715267181396, "step": 14333 }, { "epoch": 1.129996058336618, "grad_norm": 0.4722045063972473, "learning_rate": 0.0005, "loss": 1.450751781463623, "step": 14334 }, { "epoch": 1.1300748916042571, "grad_norm": 0.6508039832115173, "learning_rate": 0.0005, "loss": 1.6240670680999756, "step": 14335 }, { "epoch": 1.130153724871896, "grad_norm": 0.5533376932144165, "learning_rate": 0.0005, "loss": 1.6167657375335693, "step": 14336 }, { "epoch": 1.130232558139535, "grad_norm": 0.5601659417152405, "learning_rate": 0.0005, "loss": 1.6040927171707153, "step": 14337 }, { "epoch": 1.1303113914071738, "grad_norm": 0.622469961643219, "learning_rate": 0.0005, "loss": 1.7529911994934082, "step": 14338 }, { "epoch": 1.1303902246748128, "grad_norm": 0.6269214749336243, "learning_rate": 0.0005, "loss": 1.8637416362762451, "step": 14339 }, { "epoch": 1.1304690579424517, "grad_norm": 0.6705576777458191, "learning_rate": 0.0005, "loss": 1.577754020690918, "step": 14340 }, { "epoch": 1.1305478912100906, "grad_norm": 0.6698470711708069, "learning_rate": 0.0005, "loss": 1.6810587644577026, "step": 14341 }, { "epoch": 1.1306267244777297, "grad_norm": 0.5946546792984009, "learning_rate": 0.0005, "loss": 1.7444167137145996, "step": 14342 }, { "epoch": 1.1307055577453686, "grad_norm": 0.7369880080223083, "learning_rate": 0.0005, "loss": 1.6687791347503662, "step": 14343 }, { "epoch": 1.1307843910130075, "grad_norm": 0.5260802507400513, "learning_rate": 0.0005, "loss": 1.8931632041931152, "step": 14344 }, { "epoch": 1.1308632242806465, "grad_norm": 2.1471307277679443, "learning_rate": 0.0005, "loss": 1.5721590518951416, "step": 14345 }, { "epoch": 1.1309420575482854, "grad_norm": 0.7230892777442932, "learning_rate": 0.0005, "loss": 1.9290679693222046, "step": 14346 }, { "epoch": 1.1310208908159243, "grad_norm": 0.46360883116722107, "learning_rate": 0.0005, "loss": 1.6424636840820312, "step": 14347 }, { "epoch": 1.1310997240835632, "grad_norm": 0.41956549882888794, "learning_rate": 0.0005, "loss": 1.5697396993637085, "step": 14348 }, { "epoch": 1.1311785573512023, "grad_norm": 0.5510677695274353, "learning_rate": 0.0005, "loss": 1.7036515474319458, "step": 14349 }, { "epoch": 1.1312573906188412, "grad_norm": 0.5216786861419678, "learning_rate": 0.0005, "loss": 1.7613248825073242, "step": 14350 }, { "epoch": 1.1313362238864801, "grad_norm": 0.6447511911392212, "learning_rate": 0.0005, "loss": 1.9011516571044922, "step": 14351 }, { "epoch": 1.131415057154119, "grad_norm": 0.7446997165679932, "learning_rate": 0.0005, "loss": 1.699052333831787, "step": 14352 }, { "epoch": 1.131493890421758, "grad_norm": 0.536590576171875, "learning_rate": 0.0005, "loss": 1.5758178234100342, "step": 14353 }, { "epoch": 1.1315727236893969, "grad_norm": 0.87401282787323, "learning_rate": 0.0005, "loss": 1.7203662395477295, "step": 14354 }, { "epoch": 1.1316515569570358, "grad_norm": 0.6183731555938721, "learning_rate": 0.0005, "loss": 1.7978265285491943, "step": 14355 }, { "epoch": 1.131730390224675, "grad_norm": 0.6935025453567505, "learning_rate": 0.0005, "loss": 1.5856890678405762, "step": 14356 }, { "epoch": 1.1318092234923138, "grad_norm": 0.7012208700180054, "learning_rate": 0.0005, "loss": 1.7160193920135498, "step": 14357 }, { "epoch": 1.1318880567599527, "grad_norm": 0.5525352358818054, "learning_rate": 0.0005, "loss": 1.6492843627929688, "step": 14358 }, { "epoch": 1.1319668900275917, "grad_norm": 0.48651206493377686, "learning_rate": 0.0005, "loss": 1.6322414875030518, "step": 14359 }, { "epoch": 1.1320457232952306, "grad_norm": 0.5485055446624756, "learning_rate": 0.0005, "loss": 1.4919471740722656, "step": 14360 }, { "epoch": 1.1321245565628695, "grad_norm": 0.5549324154853821, "learning_rate": 0.0005, "loss": 1.6236801147460938, "step": 14361 }, { "epoch": 1.1322033898305084, "grad_norm": 0.5945490002632141, "learning_rate": 0.0005, "loss": 1.635251522064209, "step": 14362 }, { "epoch": 1.1322822230981475, "grad_norm": 0.5301374197006226, "learning_rate": 0.0005, "loss": 1.5815882682800293, "step": 14363 }, { "epoch": 1.1323610563657864, "grad_norm": 0.5669935345649719, "learning_rate": 0.0005, "loss": 1.8815476894378662, "step": 14364 }, { "epoch": 1.1324398896334253, "grad_norm": 0.6952226758003235, "learning_rate": 0.0005, "loss": 1.835657000541687, "step": 14365 }, { "epoch": 1.1325187229010643, "grad_norm": 0.7342749834060669, "learning_rate": 0.0005, "loss": 1.7426042556762695, "step": 14366 }, { "epoch": 1.1325975561687032, "grad_norm": 0.5031358599662781, "learning_rate": 0.0005, "loss": 1.573430061340332, "step": 14367 }, { "epoch": 1.132676389436342, "grad_norm": 0.5076251029968262, "learning_rate": 0.0005, "loss": 1.727089285850525, "step": 14368 }, { "epoch": 1.132755222703981, "grad_norm": 0.5689321160316467, "learning_rate": 0.0005, "loss": 1.8465709686279297, "step": 14369 }, { "epoch": 1.1328340559716201, "grad_norm": 0.5063073635101318, "learning_rate": 0.0005, "loss": 1.5829322338104248, "step": 14370 }, { "epoch": 1.132912889239259, "grad_norm": 0.499978244304657, "learning_rate": 0.0005, "loss": 1.6119645833969116, "step": 14371 }, { "epoch": 1.132991722506898, "grad_norm": 0.5753496885299683, "learning_rate": 0.0005, "loss": 1.6771361827850342, "step": 14372 }, { "epoch": 1.1330705557745369, "grad_norm": 0.7459123134613037, "learning_rate": 0.0005, "loss": 1.6691501140594482, "step": 14373 }, { "epoch": 1.1331493890421758, "grad_norm": 0.5472405552864075, "learning_rate": 0.0005, "loss": 1.5591949224472046, "step": 14374 }, { "epoch": 1.1332282223098147, "grad_norm": 0.7757577300071716, "learning_rate": 0.0005, "loss": 1.5930907726287842, "step": 14375 }, { "epoch": 1.1333070555774536, "grad_norm": 0.5202301740646362, "learning_rate": 0.0005, "loss": 1.6273713111877441, "step": 14376 }, { "epoch": 1.1333858888450927, "grad_norm": 0.4896254241466522, "learning_rate": 0.0005, "loss": 1.612817645072937, "step": 14377 }, { "epoch": 1.1334647221127316, "grad_norm": 0.4601401090621948, "learning_rate": 0.0005, "loss": 1.426095724105835, "step": 14378 }, { "epoch": 1.1335435553803705, "grad_norm": 1.124431848526001, "learning_rate": 0.0005, "loss": 1.950628638267517, "step": 14379 }, { "epoch": 1.1336223886480095, "grad_norm": 0.5409936904907227, "learning_rate": 0.0005, "loss": 1.8073580265045166, "step": 14380 }, { "epoch": 1.1337012219156484, "grad_norm": 0.9285182356834412, "learning_rate": 0.0005, "loss": 1.8786237239837646, "step": 14381 }, { "epoch": 1.1337800551832873, "grad_norm": 0.6318385601043701, "learning_rate": 0.0005, "loss": 1.6885242462158203, "step": 14382 }, { "epoch": 1.1338588884509262, "grad_norm": 0.569372832775116, "learning_rate": 0.0005, "loss": 1.8304400444030762, "step": 14383 }, { "epoch": 1.1339377217185653, "grad_norm": 0.5584993362426758, "learning_rate": 0.0005, "loss": 1.832497000694275, "step": 14384 }, { "epoch": 1.1340165549862042, "grad_norm": 0.5617901086807251, "learning_rate": 0.0005, "loss": 1.6775248050689697, "step": 14385 }, { "epoch": 1.1340953882538432, "grad_norm": 0.5970054268836975, "learning_rate": 0.0005, "loss": 1.5917946100234985, "step": 14386 }, { "epoch": 1.134174221521482, "grad_norm": 0.578534722328186, "learning_rate": 0.0005, "loss": 1.6262342929840088, "step": 14387 }, { "epoch": 1.134253054789121, "grad_norm": 0.6996297836303711, "learning_rate": 0.0005, "loss": 1.8266302347183228, "step": 14388 }, { "epoch": 1.1343318880567599, "grad_norm": 1.4327516555786133, "learning_rate": 0.0005, "loss": 1.8211877346038818, "step": 14389 }, { "epoch": 1.1344107213243988, "grad_norm": 0.6378213763237, "learning_rate": 0.0005, "loss": 1.8147282600402832, "step": 14390 }, { "epoch": 1.134489554592038, "grad_norm": 0.6846408247947693, "learning_rate": 0.0005, "loss": 1.9317506551742554, "step": 14391 }, { "epoch": 1.1345683878596768, "grad_norm": 0.7220199108123779, "learning_rate": 0.0005, "loss": 1.9739893674850464, "step": 14392 }, { "epoch": 1.1346472211273158, "grad_norm": 0.9106923341751099, "learning_rate": 0.0005, "loss": 1.7019894123077393, "step": 14393 }, { "epoch": 1.1347260543949547, "grad_norm": 0.5603052973747253, "learning_rate": 0.0005, "loss": 1.7595590353012085, "step": 14394 }, { "epoch": 1.1348048876625936, "grad_norm": 0.4177681505680084, "learning_rate": 0.0005, "loss": 1.3729565143585205, "step": 14395 }, { "epoch": 1.1348837209302325, "grad_norm": 0.5189051628112793, "learning_rate": 0.0005, "loss": 1.6289420127868652, "step": 14396 }, { "epoch": 1.1349625541978714, "grad_norm": 0.5274521708488464, "learning_rate": 0.0005, "loss": 1.454470157623291, "step": 14397 }, { "epoch": 1.1350413874655105, "grad_norm": 0.6769077181816101, "learning_rate": 0.0005, "loss": 1.618679404258728, "step": 14398 }, { "epoch": 1.1351202207331494, "grad_norm": 0.6012132167816162, "learning_rate": 0.0005, "loss": 1.7139142751693726, "step": 14399 }, { "epoch": 1.1351990540007884, "grad_norm": 0.5208827257156372, "learning_rate": 0.0005, "loss": 1.7581794261932373, "step": 14400 }, { "epoch": 1.1352778872684273, "grad_norm": 0.5443966388702393, "learning_rate": 0.0005, "loss": 1.7490782737731934, "step": 14401 }, { "epoch": 1.1353567205360662, "grad_norm": 1.5954338312149048, "learning_rate": 0.0005, "loss": 1.946718454360962, "step": 14402 }, { "epoch": 1.135435553803705, "grad_norm": 0.5128312110900879, "learning_rate": 0.0005, "loss": 1.5729608535766602, "step": 14403 }, { "epoch": 1.135514387071344, "grad_norm": 0.547673225402832, "learning_rate": 0.0005, "loss": 1.7019438743591309, "step": 14404 }, { "epoch": 1.1355932203389831, "grad_norm": 0.45805105566978455, "learning_rate": 0.0005, "loss": 1.7310796976089478, "step": 14405 }, { "epoch": 1.135672053606622, "grad_norm": 1.175755262374878, "learning_rate": 0.0005, "loss": 1.4766271114349365, "step": 14406 }, { "epoch": 1.135750886874261, "grad_norm": 0.6728258728981018, "learning_rate": 0.0005, "loss": 1.9017789363861084, "step": 14407 }, { "epoch": 1.1358297201418999, "grad_norm": 0.58509361743927, "learning_rate": 0.0005, "loss": 1.6995265483856201, "step": 14408 }, { "epoch": 1.1359085534095388, "grad_norm": 0.5061493515968323, "learning_rate": 0.0005, "loss": 1.452659249305725, "step": 14409 }, { "epoch": 1.1359873866771777, "grad_norm": 0.676017701625824, "learning_rate": 0.0005, "loss": 1.8124868869781494, "step": 14410 }, { "epoch": 1.1360662199448166, "grad_norm": 0.6797938942909241, "learning_rate": 0.0005, "loss": 1.6557482481002808, "step": 14411 }, { "epoch": 1.1361450532124557, "grad_norm": 0.5792561769485474, "learning_rate": 0.0005, "loss": 1.6661014556884766, "step": 14412 }, { "epoch": 1.1362238864800946, "grad_norm": 0.47906023263931274, "learning_rate": 0.0005, "loss": 1.5867350101470947, "step": 14413 }, { "epoch": 1.1363027197477336, "grad_norm": 0.5652987957000732, "learning_rate": 0.0005, "loss": 1.7355234622955322, "step": 14414 }, { "epoch": 1.1363815530153725, "grad_norm": 0.5677073001861572, "learning_rate": 0.0005, "loss": 2.0211400985717773, "step": 14415 }, { "epoch": 1.1364603862830114, "grad_norm": 0.6284754872322083, "learning_rate": 0.0005, "loss": 1.8631293773651123, "step": 14416 }, { "epoch": 1.1365392195506503, "grad_norm": 0.43538016080856323, "learning_rate": 0.0005, "loss": 1.6800858974456787, "step": 14417 }, { "epoch": 1.1366180528182892, "grad_norm": 0.5837237238883972, "learning_rate": 0.0005, "loss": 1.753164529800415, "step": 14418 }, { "epoch": 1.1366968860859283, "grad_norm": 0.5286058783531189, "learning_rate": 0.0005, "loss": 1.7048823833465576, "step": 14419 }, { "epoch": 1.1367757193535672, "grad_norm": 0.7698415517807007, "learning_rate": 0.0005, "loss": 1.6330132484436035, "step": 14420 }, { "epoch": 1.1368545526212062, "grad_norm": 0.6962748765945435, "learning_rate": 0.0005, "loss": 1.6411254405975342, "step": 14421 }, { "epoch": 1.136933385888845, "grad_norm": 0.538262665271759, "learning_rate": 0.0005, "loss": 1.5604853630065918, "step": 14422 }, { "epoch": 1.137012219156484, "grad_norm": 0.6092243194580078, "learning_rate": 0.0005, "loss": 1.9056110382080078, "step": 14423 }, { "epoch": 1.137091052424123, "grad_norm": 0.5786088705062866, "learning_rate": 0.0005, "loss": 1.6548547744750977, "step": 14424 }, { "epoch": 1.1371698856917618, "grad_norm": 0.5512592792510986, "learning_rate": 0.0005, "loss": 1.483675479888916, "step": 14425 }, { "epoch": 1.137248718959401, "grad_norm": 0.683123767375946, "learning_rate": 0.0005, "loss": 1.762235164642334, "step": 14426 }, { "epoch": 1.1373275522270399, "grad_norm": 0.5780425071716309, "learning_rate": 0.0005, "loss": 1.7352991104125977, "step": 14427 }, { "epoch": 1.1374063854946788, "grad_norm": 0.627882719039917, "learning_rate": 0.0005, "loss": 1.5884549617767334, "step": 14428 }, { "epoch": 1.1374852187623177, "grad_norm": 0.5754303932189941, "learning_rate": 0.0005, "loss": 1.6395776271820068, "step": 14429 }, { "epoch": 1.1375640520299566, "grad_norm": 0.6557706594467163, "learning_rate": 0.0005, "loss": 1.6467503309249878, "step": 14430 }, { "epoch": 1.1376428852975955, "grad_norm": 0.6664376258850098, "learning_rate": 0.0005, "loss": 1.5128697156906128, "step": 14431 }, { "epoch": 1.1377217185652344, "grad_norm": 0.5428707599639893, "learning_rate": 0.0005, "loss": 1.7832257747650146, "step": 14432 }, { "epoch": 1.1378005518328735, "grad_norm": 0.6166155338287354, "learning_rate": 0.0005, "loss": 1.752064824104309, "step": 14433 }, { "epoch": 1.1378793851005125, "grad_norm": 0.6197337508201599, "learning_rate": 0.0005, "loss": 1.7042698860168457, "step": 14434 }, { "epoch": 1.1379582183681514, "grad_norm": 0.5495155453681946, "learning_rate": 0.0005, "loss": 1.8381140232086182, "step": 14435 }, { "epoch": 1.1380370516357903, "grad_norm": 0.6186150908470154, "learning_rate": 0.0005, "loss": 1.5113990306854248, "step": 14436 }, { "epoch": 1.1381158849034292, "grad_norm": 0.6329721212387085, "learning_rate": 0.0005, "loss": 1.917337417602539, "step": 14437 }, { "epoch": 1.138194718171068, "grad_norm": 0.5442931652069092, "learning_rate": 0.0005, "loss": 1.8428410291671753, "step": 14438 }, { "epoch": 1.138273551438707, "grad_norm": 0.4716603457927704, "learning_rate": 0.0005, "loss": 1.6889243125915527, "step": 14439 }, { "epoch": 1.1383523847063461, "grad_norm": 0.5099309682846069, "learning_rate": 0.0005, "loss": 1.610795497894287, "step": 14440 }, { "epoch": 1.138431217973985, "grad_norm": 0.5341246724128723, "learning_rate": 0.0005, "loss": 1.8653193712234497, "step": 14441 }, { "epoch": 1.138510051241624, "grad_norm": 0.6297435760498047, "learning_rate": 0.0005, "loss": 2.1223959922790527, "step": 14442 }, { "epoch": 1.1385888845092629, "grad_norm": 0.7834682464599609, "learning_rate": 0.0005, "loss": 1.8757028579711914, "step": 14443 }, { "epoch": 1.1386677177769018, "grad_norm": 0.6400464177131653, "learning_rate": 0.0005, "loss": 1.698299765586853, "step": 14444 }, { "epoch": 1.1387465510445407, "grad_norm": 0.451619416475296, "learning_rate": 0.0005, "loss": 1.6017112731933594, "step": 14445 }, { "epoch": 1.1388253843121798, "grad_norm": 0.6160362362861633, "learning_rate": 0.0005, "loss": 1.9560538530349731, "step": 14446 }, { "epoch": 1.1389042175798187, "grad_norm": 0.628783643245697, "learning_rate": 0.0005, "loss": 1.6588263511657715, "step": 14447 }, { "epoch": 1.1389830508474577, "grad_norm": 0.4862578809261322, "learning_rate": 0.0005, "loss": 1.722381830215454, "step": 14448 }, { "epoch": 1.1390618841150966, "grad_norm": 0.6362446546554565, "learning_rate": 0.0005, "loss": 1.5877622365951538, "step": 14449 }, { "epoch": 1.1391407173827355, "grad_norm": 0.6142780184745789, "learning_rate": 0.0005, "loss": 1.8128690719604492, "step": 14450 }, { "epoch": 1.1392195506503744, "grad_norm": 0.707761287689209, "learning_rate": 0.0005, "loss": 1.500121831893921, "step": 14451 }, { "epoch": 1.1392983839180135, "grad_norm": 0.5445098280906677, "learning_rate": 0.0005, "loss": 1.6711370944976807, "step": 14452 }, { "epoch": 1.1393772171856524, "grad_norm": 0.46379390358924866, "learning_rate": 0.0005, "loss": 1.6497011184692383, "step": 14453 }, { "epoch": 1.1394560504532913, "grad_norm": 0.5842791199684143, "learning_rate": 0.0005, "loss": 1.523270606994629, "step": 14454 }, { "epoch": 1.1395348837209303, "grad_norm": 0.5155230760574341, "learning_rate": 0.0005, "loss": 1.6738756895065308, "step": 14455 }, { "epoch": 1.1396137169885692, "grad_norm": 0.5247166752815247, "learning_rate": 0.0005, "loss": 1.8140621185302734, "step": 14456 }, { "epoch": 1.139692550256208, "grad_norm": 0.9897801280021667, "learning_rate": 0.0005, "loss": 1.6511874198913574, "step": 14457 }, { "epoch": 1.139771383523847, "grad_norm": 0.5480172038078308, "learning_rate": 0.0005, "loss": 1.8834593296051025, "step": 14458 }, { "epoch": 1.1398502167914861, "grad_norm": 0.5761409401893616, "learning_rate": 0.0005, "loss": 1.6571646928787231, "step": 14459 }, { "epoch": 1.139929050059125, "grad_norm": 0.5698851346969604, "learning_rate": 0.0005, "loss": 1.7205232381820679, "step": 14460 }, { "epoch": 1.140007883326764, "grad_norm": 0.5089299082756042, "learning_rate": 0.0005, "loss": 1.5231455564498901, "step": 14461 }, { "epoch": 1.1400867165944029, "grad_norm": 0.6671081781387329, "learning_rate": 0.0005, "loss": 1.6665159463882446, "step": 14462 }, { "epoch": 1.1401655498620418, "grad_norm": 0.5592573881149292, "learning_rate": 0.0005, "loss": 1.7307488918304443, "step": 14463 }, { "epoch": 1.1402443831296807, "grad_norm": 0.510949969291687, "learning_rate": 0.0005, "loss": 1.6462624073028564, "step": 14464 }, { "epoch": 1.1403232163973196, "grad_norm": 0.7662904262542725, "learning_rate": 0.0005, "loss": 1.7204183340072632, "step": 14465 }, { "epoch": 1.1404020496649587, "grad_norm": 0.4513677954673767, "learning_rate": 0.0005, "loss": 1.576723337173462, "step": 14466 }, { "epoch": 1.1404808829325976, "grad_norm": 0.5651301741600037, "learning_rate": 0.0005, "loss": 1.8204972743988037, "step": 14467 }, { "epoch": 1.1405597162002366, "grad_norm": 0.5996485352516174, "learning_rate": 0.0005, "loss": 1.5908946990966797, "step": 14468 }, { "epoch": 1.1406385494678755, "grad_norm": 0.8339952826499939, "learning_rate": 0.0005, "loss": 1.617056131362915, "step": 14469 }, { "epoch": 1.1407173827355144, "grad_norm": 0.6355894207954407, "learning_rate": 0.0005, "loss": 1.7869595289230347, "step": 14470 }, { "epoch": 1.1407962160031533, "grad_norm": 0.9252794981002808, "learning_rate": 0.0005, "loss": 1.8505845069885254, "step": 14471 }, { "epoch": 1.1408750492707922, "grad_norm": 0.7425115704536438, "learning_rate": 0.0005, "loss": 1.6190354824066162, "step": 14472 }, { "epoch": 1.1409538825384313, "grad_norm": 0.5195939540863037, "learning_rate": 0.0005, "loss": 1.5365383625030518, "step": 14473 }, { "epoch": 1.1410327158060702, "grad_norm": 0.8605195879936218, "learning_rate": 0.0005, "loss": 1.7619365453720093, "step": 14474 }, { "epoch": 1.1411115490737092, "grad_norm": 0.5772062540054321, "learning_rate": 0.0005, "loss": 1.5654877424240112, "step": 14475 }, { "epoch": 1.141190382341348, "grad_norm": 1.059694528579712, "learning_rate": 0.0005, "loss": 1.572176456451416, "step": 14476 }, { "epoch": 1.141269215608987, "grad_norm": 0.6311451196670532, "learning_rate": 0.0005, "loss": 1.6954240798950195, "step": 14477 }, { "epoch": 1.1413480488766259, "grad_norm": 0.7783624529838562, "learning_rate": 0.0005, "loss": 1.6793503761291504, "step": 14478 }, { "epoch": 1.1414268821442648, "grad_norm": 0.5725017786026001, "learning_rate": 0.0005, "loss": 1.7638729810714722, "step": 14479 }, { "epoch": 1.141505715411904, "grad_norm": 0.528713047504425, "learning_rate": 0.0005, "loss": 1.9139153957366943, "step": 14480 }, { "epoch": 1.1415845486795428, "grad_norm": 0.5850708484649658, "learning_rate": 0.0005, "loss": 1.7643802165985107, "step": 14481 }, { "epoch": 1.1416633819471818, "grad_norm": 0.6465645432472229, "learning_rate": 0.0005, "loss": 1.6304821968078613, "step": 14482 }, { "epoch": 1.1417422152148207, "grad_norm": 0.5456634163856506, "learning_rate": 0.0005, "loss": 1.7356324195861816, "step": 14483 }, { "epoch": 1.1418210484824596, "grad_norm": 0.48104214668273926, "learning_rate": 0.0005, "loss": 1.6860548257827759, "step": 14484 }, { "epoch": 1.1418998817500985, "grad_norm": 0.6549045443534851, "learning_rate": 0.0005, "loss": 1.626781702041626, "step": 14485 }, { "epoch": 1.1419787150177374, "grad_norm": 1.936081886291504, "learning_rate": 0.0005, "loss": 1.7718793153762817, "step": 14486 }, { "epoch": 1.1420575482853765, "grad_norm": 0.4697158634662628, "learning_rate": 0.0005, "loss": 1.5918383598327637, "step": 14487 }, { "epoch": 1.1421363815530154, "grad_norm": 0.5082311034202576, "learning_rate": 0.0005, "loss": 1.7277944087982178, "step": 14488 }, { "epoch": 1.1422152148206544, "grad_norm": 0.5454304814338684, "learning_rate": 0.0005, "loss": 1.4541581869125366, "step": 14489 }, { "epoch": 1.1422940480882933, "grad_norm": 0.6497122645378113, "learning_rate": 0.0005, "loss": 1.6490309238433838, "step": 14490 }, { "epoch": 1.1423728813559322, "grad_norm": 0.6431276202201843, "learning_rate": 0.0005, "loss": 1.7722456455230713, "step": 14491 }, { "epoch": 1.142451714623571, "grad_norm": 0.5904833674430847, "learning_rate": 0.0005, "loss": 1.7247941493988037, "step": 14492 }, { "epoch": 1.14253054789121, "grad_norm": 0.6166362166404724, "learning_rate": 0.0005, "loss": 1.6522037982940674, "step": 14493 }, { "epoch": 1.1426093811588491, "grad_norm": 0.4967200458049774, "learning_rate": 0.0005, "loss": 1.3082274198532104, "step": 14494 }, { "epoch": 1.142688214426488, "grad_norm": 0.8357533812522888, "learning_rate": 0.0005, "loss": 1.6868360042572021, "step": 14495 }, { "epoch": 1.142767047694127, "grad_norm": 0.5583243370056152, "learning_rate": 0.0005, "loss": 1.7738440036773682, "step": 14496 }, { "epoch": 1.1428458809617659, "grad_norm": 0.6997463703155518, "learning_rate": 0.0005, "loss": 1.7552365064620972, "step": 14497 }, { "epoch": 1.1429247142294048, "grad_norm": 0.5365268588066101, "learning_rate": 0.0005, "loss": 1.5298179388046265, "step": 14498 }, { "epoch": 1.1430035474970437, "grad_norm": 1.2424205541610718, "learning_rate": 0.0005, "loss": 1.696305513381958, "step": 14499 }, { "epoch": 1.1430823807646826, "grad_norm": 0.893881618976593, "learning_rate": 0.0005, "loss": 1.7699769735336304, "step": 14500 }, { "epoch": 1.1431612140323217, "grad_norm": 0.5269955992698669, "learning_rate": 0.0005, "loss": 1.6658942699432373, "step": 14501 }, { "epoch": 1.1432400472999606, "grad_norm": 0.6586158275604248, "learning_rate": 0.0005, "loss": 1.9355348348617554, "step": 14502 }, { "epoch": 1.1433188805675996, "grad_norm": 0.7031962871551514, "learning_rate": 0.0005, "loss": 1.8153750896453857, "step": 14503 }, { "epoch": 1.1433977138352385, "grad_norm": 0.5636985301971436, "learning_rate": 0.0005, "loss": 1.7071833610534668, "step": 14504 }, { "epoch": 1.1434765471028774, "grad_norm": 0.6625503897666931, "learning_rate": 0.0005, "loss": 1.6899094581604004, "step": 14505 }, { "epoch": 1.1435553803705163, "grad_norm": 0.5333245396614075, "learning_rate": 0.0005, "loss": 1.620161771774292, "step": 14506 }, { "epoch": 1.1436342136381552, "grad_norm": 0.5333598256111145, "learning_rate": 0.0005, "loss": 1.6230809688568115, "step": 14507 }, { "epoch": 1.1437130469057943, "grad_norm": 0.6711152791976929, "learning_rate": 0.0005, "loss": 1.511676549911499, "step": 14508 }, { "epoch": 1.1437918801734333, "grad_norm": 0.489249587059021, "learning_rate": 0.0005, "loss": 1.645127534866333, "step": 14509 }, { "epoch": 1.1438707134410722, "grad_norm": 0.6222441792488098, "learning_rate": 0.0005, "loss": 1.7260969877243042, "step": 14510 }, { "epoch": 1.143949546708711, "grad_norm": 0.5911145210266113, "learning_rate": 0.0005, "loss": 1.7097501754760742, "step": 14511 }, { "epoch": 1.14402837997635, "grad_norm": 0.4335083067417145, "learning_rate": 0.0005, "loss": 1.493668794631958, "step": 14512 }, { "epoch": 1.144107213243989, "grad_norm": 0.47449710965156555, "learning_rate": 0.0005, "loss": 1.6548577547073364, "step": 14513 }, { "epoch": 1.1441860465116278, "grad_norm": 0.5628563761711121, "learning_rate": 0.0005, "loss": 1.7193708419799805, "step": 14514 }, { "epoch": 1.144264879779267, "grad_norm": 0.7381434440612793, "learning_rate": 0.0005, "loss": 1.5767937898635864, "step": 14515 }, { "epoch": 1.1443437130469059, "grad_norm": 0.45839929580688477, "learning_rate": 0.0005, "loss": 1.6414825916290283, "step": 14516 }, { "epoch": 1.1444225463145448, "grad_norm": 0.5735212564468384, "learning_rate": 0.0005, "loss": 1.9308037757873535, "step": 14517 }, { "epoch": 1.1445013795821837, "grad_norm": 0.5108332633972168, "learning_rate": 0.0005, "loss": 1.8256787061691284, "step": 14518 }, { "epoch": 1.1445802128498226, "grad_norm": 0.607320249080658, "learning_rate": 0.0005, "loss": 1.767987847328186, "step": 14519 }, { "epoch": 1.1446590461174615, "grad_norm": 0.6840664148330688, "learning_rate": 0.0005, "loss": 1.9158406257629395, "step": 14520 }, { "epoch": 1.1447378793851004, "grad_norm": 0.4665023386478424, "learning_rate": 0.0005, "loss": 1.4522576332092285, "step": 14521 }, { "epoch": 1.1448167126527395, "grad_norm": 0.6347650289535522, "learning_rate": 0.0005, "loss": 1.7250633239746094, "step": 14522 }, { "epoch": 1.1448955459203785, "grad_norm": 0.5000835061073303, "learning_rate": 0.0005, "loss": 1.6469707489013672, "step": 14523 }, { "epoch": 1.1449743791880174, "grad_norm": 0.5668047666549683, "learning_rate": 0.0005, "loss": 1.5968997478485107, "step": 14524 }, { "epoch": 1.1450532124556563, "grad_norm": 0.6447446942329407, "learning_rate": 0.0005, "loss": 1.646979570388794, "step": 14525 }, { "epoch": 1.1451320457232952, "grad_norm": 0.483571320772171, "learning_rate": 0.0005, "loss": 1.493890404701233, "step": 14526 }, { "epoch": 1.145210878990934, "grad_norm": 0.4458810091018677, "learning_rate": 0.0005, "loss": 1.3670958280563354, "step": 14527 }, { "epoch": 1.145289712258573, "grad_norm": 0.5054314732551575, "learning_rate": 0.0005, "loss": 1.6829404830932617, "step": 14528 }, { "epoch": 1.1453685455262121, "grad_norm": 0.5028937458992004, "learning_rate": 0.0005, "loss": 1.7762157917022705, "step": 14529 }, { "epoch": 1.145447378793851, "grad_norm": 0.573641300201416, "learning_rate": 0.0005, "loss": 1.664488434791565, "step": 14530 }, { "epoch": 1.14552621206149, "grad_norm": 0.7353740334510803, "learning_rate": 0.0005, "loss": 1.7670727968215942, "step": 14531 }, { "epoch": 1.1456050453291289, "grad_norm": 0.48925304412841797, "learning_rate": 0.0005, "loss": 1.5177010297775269, "step": 14532 }, { "epoch": 1.1456838785967678, "grad_norm": 0.6026087403297424, "learning_rate": 0.0005, "loss": 1.826795220375061, "step": 14533 }, { "epoch": 1.1457627118644067, "grad_norm": 0.4203391969203949, "learning_rate": 0.0005, "loss": 1.4057831764221191, "step": 14534 }, { "epoch": 1.1458415451320456, "grad_norm": 0.6239122152328491, "learning_rate": 0.0005, "loss": 1.8567042350769043, "step": 14535 }, { "epoch": 1.1459203783996847, "grad_norm": 0.5510665774345398, "learning_rate": 0.0005, "loss": 1.7079813480377197, "step": 14536 }, { "epoch": 1.1459992116673237, "grad_norm": 0.5662954449653625, "learning_rate": 0.0005, "loss": 1.8321895599365234, "step": 14537 }, { "epoch": 1.1460780449349626, "grad_norm": 0.6629450917243958, "learning_rate": 0.0005, "loss": 1.76055109500885, "step": 14538 }, { "epoch": 1.1461568782026015, "grad_norm": 0.5535005927085876, "learning_rate": 0.0005, "loss": 1.8184446096420288, "step": 14539 }, { "epoch": 1.1462357114702404, "grad_norm": 1.679733157157898, "learning_rate": 0.0005, "loss": 1.6335570812225342, "step": 14540 }, { "epoch": 1.1463145447378793, "grad_norm": 0.5818712711334229, "learning_rate": 0.0005, "loss": 1.8406355381011963, "step": 14541 }, { "epoch": 1.1463933780055182, "grad_norm": 0.46820077300071716, "learning_rate": 0.0005, "loss": 1.7732577323913574, "step": 14542 }, { "epoch": 1.1464722112731573, "grad_norm": 0.5074489712715149, "learning_rate": 0.0005, "loss": 1.7170168161392212, "step": 14543 }, { "epoch": 1.1465510445407963, "grad_norm": 0.79007488489151, "learning_rate": 0.0005, "loss": 1.6248631477355957, "step": 14544 }, { "epoch": 1.1466298778084352, "grad_norm": 0.59185791015625, "learning_rate": 0.0005, "loss": 1.8279516696929932, "step": 14545 }, { "epoch": 1.146708711076074, "grad_norm": 0.6287750601768494, "learning_rate": 0.0005, "loss": 1.7043377161026, "step": 14546 }, { "epoch": 1.146787544343713, "grad_norm": 0.4241739809513092, "learning_rate": 0.0005, "loss": 1.4893076419830322, "step": 14547 }, { "epoch": 1.146866377611352, "grad_norm": 0.5578091740608215, "learning_rate": 0.0005, "loss": 1.7537715435028076, "step": 14548 }, { "epoch": 1.1469452108789908, "grad_norm": 0.7135096192359924, "learning_rate": 0.0005, "loss": 1.7048616409301758, "step": 14549 }, { "epoch": 1.14702404414663, "grad_norm": 0.5531341433525085, "learning_rate": 0.0005, "loss": 1.9288179874420166, "step": 14550 }, { "epoch": 1.1471028774142689, "grad_norm": 0.7598443031311035, "learning_rate": 0.0005, "loss": 1.8795199394226074, "step": 14551 }, { "epoch": 1.1471817106819078, "grad_norm": 0.538419246673584, "learning_rate": 0.0005, "loss": 1.3767938613891602, "step": 14552 }, { "epoch": 1.1472605439495467, "grad_norm": 0.7156845927238464, "learning_rate": 0.0005, "loss": 1.767683982849121, "step": 14553 }, { "epoch": 1.1473393772171856, "grad_norm": 0.6162684559822083, "learning_rate": 0.0005, "loss": 1.9504024982452393, "step": 14554 }, { "epoch": 1.1474182104848245, "grad_norm": 0.4681345224380493, "learning_rate": 0.0005, "loss": 1.5341622829437256, "step": 14555 }, { "epoch": 1.1474970437524634, "grad_norm": 0.5886788368225098, "learning_rate": 0.0005, "loss": 1.7989411354064941, "step": 14556 }, { "epoch": 1.1475758770201026, "grad_norm": 0.6712591052055359, "learning_rate": 0.0005, "loss": 1.676140546798706, "step": 14557 }, { "epoch": 1.1476547102877415, "grad_norm": 0.5465651154518127, "learning_rate": 0.0005, "loss": 1.696852684020996, "step": 14558 }, { "epoch": 1.1477335435553804, "grad_norm": 0.7271455526351929, "learning_rate": 0.0005, "loss": 1.7248308658599854, "step": 14559 }, { "epoch": 1.1478123768230193, "grad_norm": 0.4426192045211792, "learning_rate": 0.0005, "loss": 1.5775854587554932, "step": 14560 }, { "epoch": 1.1478912100906582, "grad_norm": 0.5141662359237671, "learning_rate": 0.0005, "loss": 1.8594903945922852, "step": 14561 }, { "epoch": 1.147970043358297, "grad_norm": 0.48202887177467346, "learning_rate": 0.0005, "loss": 1.6181461811065674, "step": 14562 }, { "epoch": 1.148048876625936, "grad_norm": 1.6238150596618652, "learning_rate": 0.0005, "loss": 1.6427847146987915, "step": 14563 }, { "epoch": 1.1481277098935752, "grad_norm": 0.8364775776863098, "learning_rate": 0.0005, "loss": 1.6007874011993408, "step": 14564 }, { "epoch": 1.148206543161214, "grad_norm": 0.5212916135787964, "learning_rate": 0.0005, "loss": 1.7585093975067139, "step": 14565 }, { "epoch": 1.148285376428853, "grad_norm": 0.7193771004676819, "learning_rate": 0.0005, "loss": 1.4876575469970703, "step": 14566 }, { "epoch": 1.1483642096964919, "grad_norm": 0.5922109484672546, "learning_rate": 0.0005, "loss": 1.6782821416854858, "step": 14567 }, { "epoch": 1.1484430429641308, "grad_norm": 0.6192021369934082, "learning_rate": 0.0005, "loss": 1.7299927473068237, "step": 14568 }, { "epoch": 1.1485218762317697, "grad_norm": 0.6677246689796448, "learning_rate": 0.0005, "loss": 1.8193774223327637, "step": 14569 }, { "epoch": 1.1486007094994088, "grad_norm": 0.574213981628418, "learning_rate": 0.0005, "loss": 1.706408977508545, "step": 14570 }, { "epoch": 1.1486795427670478, "grad_norm": 0.5588791966438293, "learning_rate": 0.0005, "loss": 1.622691035270691, "step": 14571 }, { "epoch": 1.1487583760346867, "grad_norm": 0.5784635543823242, "learning_rate": 0.0005, "loss": 1.687161922454834, "step": 14572 }, { "epoch": 1.1488372093023256, "grad_norm": 0.5441659092903137, "learning_rate": 0.0005, "loss": 1.649069905281067, "step": 14573 }, { "epoch": 1.1489160425699645, "grad_norm": 0.4953846335411072, "learning_rate": 0.0005, "loss": 1.6355448961257935, "step": 14574 }, { "epoch": 1.1489948758376034, "grad_norm": 0.6225398182868958, "learning_rate": 0.0005, "loss": 1.7308168411254883, "step": 14575 }, { "epoch": 1.1490737091052425, "grad_norm": 0.4711378514766693, "learning_rate": 0.0005, "loss": 1.6589572429656982, "step": 14576 }, { "epoch": 1.1491525423728814, "grad_norm": 0.5422899723052979, "learning_rate": 0.0005, "loss": 1.7123010158538818, "step": 14577 }, { "epoch": 1.1492313756405204, "grad_norm": 0.5310273766517639, "learning_rate": 0.0005, "loss": 1.7226769924163818, "step": 14578 }, { "epoch": 1.1493102089081593, "grad_norm": 0.544792652130127, "learning_rate": 0.0005, "loss": 1.6452093124389648, "step": 14579 }, { "epoch": 1.1493890421757982, "grad_norm": 0.5147685408592224, "learning_rate": 0.0005, "loss": 1.6209843158721924, "step": 14580 }, { "epoch": 1.149467875443437, "grad_norm": 0.6397682428359985, "learning_rate": 0.0005, "loss": 1.7215354442596436, "step": 14581 }, { "epoch": 1.149546708711076, "grad_norm": 0.5421988368034363, "learning_rate": 0.0005, "loss": 1.5927304029464722, "step": 14582 }, { "epoch": 1.1496255419787151, "grad_norm": 0.5659898519515991, "learning_rate": 0.0005, "loss": 1.840348243713379, "step": 14583 }, { "epoch": 1.149704375246354, "grad_norm": 0.48967400193214417, "learning_rate": 0.0005, "loss": 1.5982301235198975, "step": 14584 }, { "epoch": 1.149783208513993, "grad_norm": 0.5429301261901855, "learning_rate": 0.0005, "loss": 1.6049022674560547, "step": 14585 }, { "epoch": 1.1498620417816319, "grad_norm": 1.645992636680603, "learning_rate": 0.0005, "loss": 1.6895849704742432, "step": 14586 }, { "epoch": 1.1499408750492708, "grad_norm": 0.5364757776260376, "learning_rate": 0.0005, "loss": 1.7676770687103271, "step": 14587 }, { "epoch": 1.1500197083169097, "grad_norm": 0.4635642468929291, "learning_rate": 0.0005, "loss": 1.6649811267852783, "step": 14588 }, { "epoch": 1.1500985415845486, "grad_norm": 0.5377112030982971, "learning_rate": 0.0005, "loss": 1.7702255249023438, "step": 14589 }, { "epoch": 1.1501773748521877, "grad_norm": 0.5999036431312561, "learning_rate": 0.0005, "loss": 1.7448872327804565, "step": 14590 }, { "epoch": 1.1502562081198267, "grad_norm": 0.48287978768348694, "learning_rate": 0.0005, "loss": 1.7813758850097656, "step": 14591 }, { "epoch": 1.1503350413874656, "grad_norm": 0.5812130570411682, "learning_rate": 0.0005, "loss": 1.4733757972717285, "step": 14592 }, { "epoch": 1.1504138746551045, "grad_norm": 0.5443636775016785, "learning_rate": 0.0005, "loss": 1.6384555101394653, "step": 14593 }, { "epoch": 1.1504927079227434, "grad_norm": 0.6228477358818054, "learning_rate": 0.0005, "loss": 1.6371452808380127, "step": 14594 }, { "epoch": 1.1505715411903823, "grad_norm": 1.3009922504425049, "learning_rate": 0.0005, "loss": 1.7498180866241455, "step": 14595 }, { "epoch": 1.1506503744580212, "grad_norm": 0.8594830632209778, "learning_rate": 0.0005, "loss": 1.8949474096298218, "step": 14596 }, { "epoch": 1.1507292077256603, "grad_norm": 0.5184760689735413, "learning_rate": 0.0005, "loss": 1.6462359428405762, "step": 14597 }, { "epoch": 1.1508080409932993, "grad_norm": 0.6186419129371643, "learning_rate": 0.0005, "loss": 1.8213614225387573, "step": 14598 }, { "epoch": 1.1508868742609382, "grad_norm": 0.4996068775653839, "learning_rate": 0.0005, "loss": 1.6279772520065308, "step": 14599 }, { "epoch": 1.150965707528577, "grad_norm": 0.5669642686843872, "learning_rate": 0.0005, "loss": 1.798153281211853, "step": 14600 }, { "epoch": 1.151044540796216, "grad_norm": 0.6230382323265076, "learning_rate": 0.0005, "loss": 1.5761950016021729, "step": 14601 }, { "epoch": 1.151123374063855, "grad_norm": 0.581515371799469, "learning_rate": 0.0005, "loss": 1.5555078983306885, "step": 14602 }, { "epoch": 1.1512022073314938, "grad_norm": 0.8757210969924927, "learning_rate": 0.0005, "loss": 1.8824830055236816, "step": 14603 }, { "epoch": 1.151281040599133, "grad_norm": 0.611115038394928, "learning_rate": 0.0005, "loss": 1.7452726364135742, "step": 14604 }, { "epoch": 1.1513598738667719, "grad_norm": 0.6714422702789307, "learning_rate": 0.0005, "loss": 1.855777382850647, "step": 14605 }, { "epoch": 1.1514387071344108, "grad_norm": 0.5097957849502563, "learning_rate": 0.0005, "loss": 1.695155382156372, "step": 14606 }, { "epoch": 1.1515175404020497, "grad_norm": 0.5653910040855408, "learning_rate": 0.0005, "loss": 1.6217398643493652, "step": 14607 }, { "epoch": 1.1515963736696886, "grad_norm": 0.5183175802230835, "learning_rate": 0.0005, "loss": 1.825869083404541, "step": 14608 }, { "epoch": 1.1516752069373275, "grad_norm": 0.4713838994503021, "learning_rate": 0.0005, "loss": 1.4511792659759521, "step": 14609 }, { "epoch": 1.1517540402049664, "grad_norm": 0.563296914100647, "learning_rate": 0.0005, "loss": 1.7328169345855713, "step": 14610 }, { "epoch": 1.1518328734726055, "grad_norm": 0.49170780181884766, "learning_rate": 0.0005, "loss": 1.5536288022994995, "step": 14611 }, { "epoch": 1.1519117067402445, "grad_norm": 0.5398159027099609, "learning_rate": 0.0005, "loss": 1.5128655433654785, "step": 14612 }, { "epoch": 1.1519905400078834, "grad_norm": 0.6810095906257629, "learning_rate": 0.0005, "loss": 1.8857399225234985, "step": 14613 }, { "epoch": 1.1520693732755223, "grad_norm": 0.45546525716781616, "learning_rate": 0.0005, "loss": 1.6728225946426392, "step": 14614 }, { "epoch": 1.1521482065431612, "grad_norm": 0.4555807411670685, "learning_rate": 0.0005, "loss": 1.5661413669586182, "step": 14615 }, { "epoch": 1.1522270398108, "grad_norm": 0.4520138204097748, "learning_rate": 0.0005, "loss": 1.5044426918029785, "step": 14616 }, { "epoch": 1.152305873078439, "grad_norm": 0.5021979212760925, "learning_rate": 0.0005, "loss": 1.610966682434082, "step": 14617 }, { "epoch": 1.1523847063460781, "grad_norm": 0.5075367093086243, "learning_rate": 0.0005, "loss": 1.6719468832015991, "step": 14618 }, { "epoch": 1.152463539613717, "grad_norm": 0.5736755728721619, "learning_rate": 0.0005, "loss": 1.4906466007232666, "step": 14619 }, { "epoch": 1.152542372881356, "grad_norm": 0.5280910134315491, "learning_rate": 0.0005, "loss": 1.6913814544677734, "step": 14620 }, { "epoch": 1.1526212061489949, "grad_norm": 0.632494330406189, "learning_rate": 0.0005, "loss": 1.5422098636627197, "step": 14621 }, { "epoch": 1.1527000394166338, "grad_norm": 0.49153932929039, "learning_rate": 0.0005, "loss": 1.676903247833252, "step": 14622 }, { "epoch": 1.1527788726842727, "grad_norm": 1.3189170360565186, "learning_rate": 0.0005, "loss": 1.8413845300674438, "step": 14623 }, { "epoch": 1.1528577059519116, "grad_norm": 0.5260983109474182, "learning_rate": 0.0005, "loss": 1.5551104545593262, "step": 14624 }, { "epoch": 1.1529365392195507, "grad_norm": 0.450486421585083, "learning_rate": 0.0005, "loss": 1.5135010480880737, "step": 14625 }, { "epoch": 1.1530153724871897, "grad_norm": 0.44204452633857727, "learning_rate": 0.0005, "loss": 1.670264482498169, "step": 14626 }, { "epoch": 1.1530942057548286, "grad_norm": 0.45091840624809265, "learning_rate": 0.0005, "loss": 1.653825283050537, "step": 14627 }, { "epoch": 1.1531730390224675, "grad_norm": 0.5539184212684631, "learning_rate": 0.0005, "loss": 1.5659741163253784, "step": 14628 }, { "epoch": 1.1532518722901064, "grad_norm": 0.4800357520580292, "learning_rate": 0.0005, "loss": 1.406240463256836, "step": 14629 }, { "epoch": 1.1533307055577453, "grad_norm": 0.4385070502758026, "learning_rate": 0.0005, "loss": 1.4532831907272339, "step": 14630 }, { "epoch": 1.1534095388253842, "grad_norm": 0.4912656843662262, "learning_rate": 0.0005, "loss": 1.8150014877319336, "step": 14631 }, { "epoch": 1.1534883720930234, "grad_norm": 0.6468206644058228, "learning_rate": 0.0005, "loss": 1.7622871398925781, "step": 14632 }, { "epoch": 1.1535672053606623, "grad_norm": 0.8680338859558105, "learning_rate": 0.0005, "loss": 1.755699872970581, "step": 14633 }, { "epoch": 1.1536460386283012, "grad_norm": 0.4416022002696991, "learning_rate": 0.0005, "loss": 1.4690375328063965, "step": 14634 }, { "epoch": 1.15372487189594, "grad_norm": 0.5561847686767578, "learning_rate": 0.0005, "loss": 1.7193468809127808, "step": 14635 }, { "epoch": 1.153803705163579, "grad_norm": 0.6556582450866699, "learning_rate": 0.0005, "loss": 1.6737360954284668, "step": 14636 }, { "epoch": 1.153882538431218, "grad_norm": 0.5433633923530579, "learning_rate": 0.0005, "loss": 1.6224040985107422, "step": 14637 }, { "epoch": 1.1539613716988568, "grad_norm": 0.5665510296821594, "learning_rate": 0.0005, "loss": 1.8478195667266846, "step": 14638 }, { "epoch": 1.154040204966496, "grad_norm": 0.5006211996078491, "learning_rate": 0.0005, "loss": 1.6686036586761475, "step": 14639 }, { "epoch": 1.1541190382341349, "grad_norm": 0.4982791244983673, "learning_rate": 0.0005, "loss": 1.460686206817627, "step": 14640 }, { "epoch": 1.1541978715017738, "grad_norm": 0.5250362157821655, "learning_rate": 0.0005, "loss": 1.7731990814208984, "step": 14641 }, { "epoch": 1.1542767047694127, "grad_norm": 0.7764790654182434, "learning_rate": 0.0005, "loss": 1.783207654953003, "step": 14642 }, { "epoch": 1.1543555380370516, "grad_norm": 0.8331155180931091, "learning_rate": 0.0005, "loss": 1.5080552101135254, "step": 14643 }, { "epoch": 1.1544343713046905, "grad_norm": 0.4989929497241974, "learning_rate": 0.0005, "loss": 1.542539119720459, "step": 14644 }, { "epoch": 1.1545132045723294, "grad_norm": 0.5368719696998596, "learning_rate": 0.0005, "loss": 1.870762825012207, "step": 14645 }, { "epoch": 1.1545920378399686, "grad_norm": 0.518969714641571, "learning_rate": 0.0005, "loss": 1.4758607149124146, "step": 14646 }, { "epoch": 1.1546708711076075, "grad_norm": 0.5676249861717224, "learning_rate": 0.0005, "loss": 1.691765546798706, "step": 14647 }, { "epoch": 1.1547497043752464, "grad_norm": 0.5691429376602173, "learning_rate": 0.0005, "loss": 1.788435935974121, "step": 14648 }, { "epoch": 1.1548285376428853, "grad_norm": 0.501891016960144, "learning_rate": 0.0005, "loss": 1.4815280437469482, "step": 14649 }, { "epoch": 1.1549073709105242, "grad_norm": 0.4685541093349457, "learning_rate": 0.0005, "loss": 1.596003532409668, "step": 14650 }, { "epoch": 1.1549862041781631, "grad_norm": 0.5906623005867004, "learning_rate": 0.0005, "loss": 1.5430933237075806, "step": 14651 }, { "epoch": 1.155065037445802, "grad_norm": 0.5075674057006836, "learning_rate": 0.0005, "loss": 1.485203504562378, "step": 14652 }, { "epoch": 1.1551438707134412, "grad_norm": 0.46747899055480957, "learning_rate": 0.0005, "loss": 1.6416977643966675, "step": 14653 }, { "epoch": 1.15522270398108, "grad_norm": 0.6364209055900574, "learning_rate": 0.0005, "loss": 1.5772777795791626, "step": 14654 }, { "epoch": 1.155301537248719, "grad_norm": 0.5491989254951477, "learning_rate": 0.0005, "loss": 1.7026607990264893, "step": 14655 }, { "epoch": 1.155380370516358, "grad_norm": 0.5574931502342224, "learning_rate": 0.0005, "loss": 1.6614267826080322, "step": 14656 }, { "epoch": 1.1554592037839968, "grad_norm": 0.5965660214424133, "learning_rate": 0.0005, "loss": 1.6528568267822266, "step": 14657 }, { "epoch": 1.1555380370516357, "grad_norm": 0.6924660205841064, "learning_rate": 0.0005, "loss": 1.6864306926727295, "step": 14658 }, { "epoch": 1.1556168703192746, "grad_norm": 0.4460349678993225, "learning_rate": 0.0005, "loss": 1.79201078414917, "step": 14659 }, { "epoch": 1.1556957035869138, "grad_norm": 0.5672601461410522, "learning_rate": 0.0005, "loss": 1.8472181558609009, "step": 14660 }, { "epoch": 1.1557745368545527, "grad_norm": 0.5579609274864197, "learning_rate": 0.0005, "loss": 1.6238844394683838, "step": 14661 }, { "epoch": 1.1558533701221916, "grad_norm": 0.5268244743347168, "learning_rate": 0.0005, "loss": 1.7928872108459473, "step": 14662 }, { "epoch": 1.1559322033898305, "grad_norm": 0.6067347526550293, "learning_rate": 0.0005, "loss": 1.7647979259490967, "step": 14663 }, { "epoch": 1.1560110366574694, "grad_norm": 0.4901074469089508, "learning_rate": 0.0005, "loss": 1.713270902633667, "step": 14664 }, { "epoch": 1.1560898699251083, "grad_norm": 0.619029700756073, "learning_rate": 0.0005, "loss": 1.6120920181274414, "step": 14665 }, { "epoch": 1.1561687031927472, "grad_norm": 0.640764594078064, "learning_rate": 0.0005, "loss": 1.7171357870101929, "step": 14666 }, { "epoch": 1.1562475364603864, "grad_norm": 0.7474042773246765, "learning_rate": 0.0005, "loss": 1.7225472927093506, "step": 14667 }, { "epoch": 1.1563263697280253, "grad_norm": 0.4782135486602783, "learning_rate": 0.0005, "loss": 1.6026945114135742, "step": 14668 }, { "epoch": 1.1564052029956642, "grad_norm": 0.5457425713539124, "learning_rate": 0.0005, "loss": 1.8274776935577393, "step": 14669 }, { "epoch": 1.156484036263303, "grad_norm": 0.6343937516212463, "learning_rate": 0.0005, "loss": 1.835207462310791, "step": 14670 }, { "epoch": 1.156562869530942, "grad_norm": 0.5906991958618164, "learning_rate": 0.0005, "loss": 1.881126880645752, "step": 14671 }, { "epoch": 1.156641702798581, "grad_norm": 0.5180050134658813, "learning_rate": 0.0005, "loss": 1.5327616930007935, "step": 14672 }, { "epoch": 1.1567205360662198, "grad_norm": 0.7257353663444519, "learning_rate": 0.0005, "loss": 1.9245216846466064, "step": 14673 }, { "epoch": 1.156799369333859, "grad_norm": 0.6579846739768982, "learning_rate": 0.0005, "loss": 1.39967679977417, "step": 14674 }, { "epoch": 1.1568782026014979, "grad_norm": 0.5830433368682861, "learning_rate": 0.0005, "loss": 1.6157519817352295, "step": 14675 }, { "epoch": 1.1569570358691368, "grad_norm": 0.5480793118476868, "learning_rate": 0.0005, "loss": 1.8578177690505981, "step": 14676 }, { "epoch": 1.1570358691367757, "grad_norm": 0.6987666487693787, "learning_rate": 0.0005, "loss": 1.7934174537658691, "step": 14677 }, { "epoch": 1.1571147024044146, "grad_norm": 0.5859096050262451, "learning_rate": 0.0005, "loss": 1.6477208137512207, "step": 14678 }, { "epoch": 1.1571935356720535, "grad_norm": 0.5619818568229675, "learning_rate": 0.0005, "loss": 1.6362769603729248, "step": 14679 }, { "epoch": 1.1572723689396924, "grad_norm": 0.7548298835754395, "learning_rate": 0.0005, "loss": 2.017918825149536, "step": 14680 }, { "epoch": 1.1573512022073316, "grad_norm": 0.5257420539855957, "learning_rate": 0.0005, "loss": 1.4260756969451904, "step": 14681 }, { "epoch": 1.1574300354749705, "grad_norm": 0.4882514774799347, "learning_rate": 0.0005, "loss": 1.5035046339035034, "step": 14682 }, { "epoch": 1.1575088687426094, "grad_norm": 0.5454621315002441, "learning_rate": 0.0005, "loss": 1.7578918933868408, "step": 14683 }, { "epoch": 1.1575877020102483, "grad_norm": 0.4927426874637604, "learning_rate": 0.0005, "loss": 1.6888034343719482, "step": 14684 }, { "epoch": 1.1576665352778872, "grad_norm": 0.5578814148902893, "learning_rate": 0.0005, "loss": 1.8397092819213867, "step": 14685 }, { "epoch": 1.1577453685455261, "grad_norm": 0.5341482758522034, "learning_rate": 0.0005, "loss": 1.9364991188049316, "step": 14686 }, { "epoch": 1.157824201813165, "grad_norm": 0.5735177993774414, "learning_rate": 0.0005, "loss": 1.6679972410202026, "step": 14687 }, { "epoch": 1.1579030350808042, "grad_norm": 0.5945784449577332, "learning_rate": 0.0005, "loss": 1.6154754161834717, "step": 14688 }, { "epoch": 1.157981868348443, "grad_norm": 0.5039581656455994, "learning_rate": 0.0005, "loss": 1.7569844722747803, "step": 14689 }, { "epoch": 1.158060701616082, "grad_norm": 0.46163812279701233, "learning_rate": 0.0005, "loss": 1.2958672046661377, "step": 14690 }, { "epoch": 1.158139534883721, "grad_norm": 0.6345112323760986, "learning_rate": 0.0005, "loss": 1.700168490409851, "step": 14691 }, { "epoch": 1.1582183681513598, "grad_norm": 0.5805102586746216, "learning_rate": 0.0005, "loss": 1.650911569595337, "step": 14692 }, { "epoch": 1.1582972014189987, "grad_norm": 0.5318769216537476, "learning_rate": 0.0005, "loss": 1.7615498304367065, "step": 14693 }, { "epoch": 1.1583760346866379, "grad_norm": 0.5357584953308105, "learning_rate": 0.0005, "loss": 1.5727322101593018, "step": 14694 }, { "epoch": 1.1584548679542768, "grad_norm": 0.527919590473175, "learning_rate": 0.0005, "loss": 1.8589346408843994, "step": 14695 }, { "epoch": 1.1585337012219157, "grad_norm": 0.5802293419837952, "learning_rate": 0.0005, "loss": 1.8184757232666016, "step": 14696 }, { "epoch": 1.1586125344895546, "grad_norm": 0.6292651891708374, "learning_rate": 0.0005, "loss": 1.7443852424621582, "step": 14697 }, { "epoch": 1.1586913677571935, "grad_norm": 0.6534966826438904, "learning_rate": 0.0005, "loss": 1.8194737434387207, "step": 14698 }, { "epoch": 1.1587702010248324, "grad_norm": 0.4969765841960907, "learning_rate": 0.0005, "loss": 1.7627005577087402, "step": 14699 }, { "epoch": 1.1588490342924715, "grad_norm": 0.5309556126594543, "learning_rate": 0.0005, "loss": 1.9352372884750366, "step": 14700 }, { "epoch": 1.1589278675601105, "grad_norm": 0.5985321402549744, "learning_rate": 0.0005, "loss": 1.8061537742614746, "step": 14701 }, { "epoch": 1.1590067008277494, "grad_norm": 0.5496582984924316, "learning_rate": 0.0005, "loss": 1.522454857826233, "step": 14702 }, { "epoch": 1.1590855340953883, "grad_norm": 0.5929880738258362, "learning_rate": 0.0005, "loss": 1.671036958694458, "step": 14703 }, { "epoch": 1.1591643673630272, "grad_norm": 0.6046112179756165, "learning_rate": 0.0005, "loss": 1.7000415325164795, "step": 14704 }, { "epoch": 1.159243200630666, "grad_norm": 0.6181857585906982, "learning_rate": 0.0005, "loss": 1.777328610420227, "step": 14705 }, { "epoch": 1.159322033898305, "grad_norm": 0.6311547756195068, "learning_rate": 0.0005, "loss": 1.6578819751739502, "step": 14706 }, { "epoch": 1.1594008671659441, "grad_norm": 0.5936295390129089, "learning_rate": 0.0005, "loss": 1.5758090019226074, "step": 14707 }, { "epoch": 1.159479700433583, "grad_norm": 0.5050267577171326, "learning_rate": 0.0005, "loss": 1.6557352542877197, "step": 14708 }, { "epoch": 1.159558533701222, "grad_norm": 0.9288801550865173, "learning_rate": 0.0005, "loss": 1.5789463520050049, "step": 14709 }, { "epoch": 1.1596373669688609, "grad_norm": 0.5570785999298096, "learning_rate": 0.0005, "loss": 1.7109769582748413, "step": 14710 }, { "epoch": 1.1597162002364998, "grad_norm": 0.5069451332092285, "learning_rate": 0.0005, "loss": 1.532057762145996, "step": 14711 }, { "epoch": 1.1597950335041387, "grad_norm": 0.6245222687721252, "learning_rate": 0.0005, "loss": 1.5837492942810059, "step": 14712 }, { "epoch": 1.1598738667717776, "grad_norm": 2.9035422801971436, "learning_rate": 0.0005, "loss": 1.8906549215316772, "step": 14713 }, { "epoch": 1.1599527000394168, "grad_norm": 0.5415701866149902, "learning_rate": 0.0005, "loss": 1.4735487699508667, "step": 14714 }, { "epoch": 1.1600315333070557, "grad_norm": 0.5030486583709717, "learning_rate": 0.0005, "loss": 1.4814000129699707, "step": 14715 }, { "epoch": 1.1601103665746946, "grad_norm": 0.4683551490306854, "learning_rate": 0.0005, "loss": 1.5003716945648193, "step": 14716 }, { "epoch": 1.1601891998423335, "grad_norm": 0.453823447227478, "learning_rate": 0.0005, "loss": 1.595111608505249, "step": 14717 }, { "epoch": 1.1602680331099724, "grad_norm": 0.5645297169685364, "learning_rate": 0.0005, "loss": 1.691041350364685, "step": 14718 }, { "epoch": 1.1603468663776113, "grad_norm": 0.5606017112731934, "learning_rate": 0.0005, "loss": 1.7723976373672485, "step": 14719 }, { "epoch": 1.1604256996452502, "grad_norm": 0.9009813070297241, "learning_rate": 0.0005, "loss": 1.8036415576934814, "step": 14720 }, { "epoch": 1.1605045329128894, "grad_norm": 0.5344095826148987, "learning_rate": 0.0005, "loss": 1.8326504230499268, "step": 14721 }, { "epoch": 1.1605833661805283, "grad_norm": 0.512924313545227, "learning_rate": 0.0005, "loss": 1.609892725944519, "step": 14722 }, { "epoch": 1.1606621994481672, "grad_norm": 0.5729672312736511, "learning_rate": 0.0005, "loss": 1.8300799131393433, "step": 14723 }, { "epoch": 1.160741032715806, "grad_norm": 0.5687638521194458, "learning_rate": 0.0005, "loss": 1.8057851791381836, "step": 14724 }, { "epoch": 1.160819865983445, "grad_norm": 0.5880171656608582, "learning_rate": 0.0005, "loss": 1.6361067295074463, "step": 14725 }, { "epoch": 1.160898699251084, "grad_norm": 0.5885128378868103, "learning_rate": 0.0005, "loss": 1.6359031200408936, "step": 14726 }, { "epoch": 1.1609775325187228, "grad_norm": 0.6107869148254395, "learning_rate": 0.0005, "loss": 1.5578572750091553, "step": 14727 }, { "epoch": 1.161056365786362, "grad_norm": 0.4832933247089386, "learning_rate": 0.0005, "loss": 1.679553747177124, "step": 14728 }, { "epoch": 1.1611351990540009, "grad_norm": 0.5064546465873718, "learning_rate": 0.0005, "loss": 1.6083896160125732, "step": 14729 }, { "epoch": 1.1612140323216398, "grad_norm": 0.5798136591911316, "learning_rate": 0.0005, "loss": 1.6780613660812378, "step": 14730 }, { "epoch": 1.1612928655892787, "grad_norm": 0.4641168415546417, "learning_rate": 0.0005, "loss": 1.6907519102096558, "step": 14731 }, { "epoch": 1.1613716988569176, "grad_norm": 0.6914163827896118, "learning_rate": 0.0005, "loss": 1.930525541305542, "step": 14732 }, { "epoch": 1.1614505321245565, "grad_norm": 0.4276849031448364, "learning_rate": 0.0005, "loss": 1.4454796314239502, "step": 14733 }, { "epoch": 1.1615293653921954, "grad_norm": 0.5399702787399292, "learning_rate": 0.0005, "loss": 1.9374197721481323, "step": 14734 }, { "epoch": 1.1616081986598346, "grad_norm": 0.5780925750732422, "learning_rate": 0.0005, "loss": 1.8179101943969727, "step": 14735 }, { "epoch": 1.1616870319274735, "grad_norm": 0.624245285987854, "learning_rate": 0.0005, "loss": 1.8399113416671753, "step": 14736 }, { "epoch": 1.1617658651951124, "grad_norm": 0.7026976943016052, "learning_rate": 0.0005, "loss": 1.868089199066162, "step": 14737 }, { "epoch": 1.1618446984627513, "grad_norm": 0.45832082629203796, "learning_rate": 0.0005, "loss": 1.471720576286316, "step": 14738 }, { "epoch": 1.1619235317303902, "grad_norm": 0.46927914023399353, "learning_rate": 0.0005, "loss": 1.7305231094360352, "step": 14739 }, { "epoch": 1.1620023649980291, "grad_norm": 0.5148837566375732, "learning_rate": 0.0005, "loss": 1.5382070541381836, "step": 14740 }, { "epoch": 1.162081198265668, "grad_norm": 0.9384694695472717, "learning_rate": 0.0005, "loss": 1.6413123607635498, "step": 14741 }, { "epoch": 1.1621600315333072, "grad_norm": 0.5098440051078796, "learning_rate": 0.0005, "loss": 1.6225470304489136, "step": 14742 }, { "epoch": 1.162238864800946, "grad_norm": 0.6165545582771301, "learning_rate": 0.0005, "loss": 1.6475790739059448, "step": 14743 }, { "epoch": 1.162317698068585, "grad_norm": 0.47167515754699707, "learning_rate": 0.0005, "loss": 1.5677764415740967, "step": 14744 }, { "epoch": 1.162396531336224, "grad_norm": 0.558968186378479, "learning_rate": 0.0005, "loss": 1.6628220081329346, "step": 14745 }, { "epoch": 1.1624753646038628, "grad_norm": 0.6087949872016907, "learning_rate": 0.0005, "loss": 1.731780767440796, "step": 14746 }, { "epoch": 1.1625541978715017, "grad_norm": 0.5889904499053955, "learning_rate": 0.0005, "loss": 1.82318115234375, "step": 14747 }, { "epoch": 1.1626330311391406, "grad_norm": 0.5955399870872498, "learning_rate": 0.0005, "loss": 1.806214690208435, "step": 14748 }, { "epoch": 1.1627118644067798, "grad_norm": 0.5899330377578735, "learning_rate": 0.0005, "loss": 1.7520718574523926, "step": 14749 }, { "epoch": 1.1627906976744187, "grad_norm": 0.5763498544692993, "learning_rate": 0.0005, "loss": 1.444105863571167, "step": 14750 }, { "epoch": 1.1628695309420576, "grad_norm": 0.5925334692001343, "learning_rate": 0.0005, "loss": 1.7202777862548828, "step": 14751 }, { "epoch": 1.1629483642096965, "grad_norm": 0.5621956586837769, "learning_rate": 0.0005, "loss": 1.9906889200210571, "step": 14752 }, { "epoch": 1.1630271974773354, "grad_norm": 0.7489275932312012, "learning_rate": 0.0005, "loss": 1.9667785167694092, "step": 14753 }, { "epoch": 1.1631060307449743, "grad_norm": 0.4577459394931793, "learning_rate": 0.0005, "loss": 1.5055932998657227, "step": 14754 }, { "epoch": 1.1631848640126132, "grad_norm": 0.7572917938232422, "learning_rate": 0.0005, "loss": 1.738396406173706, "step": 14755 }, { "epoch": 1.1632636972802524, "grad_norm": 0.5952538251876831, "learning_rate": 0.0005, "loss": 1.8934369087219238, "step": 14756 }, { "epoch": 1.1633425305478913, "grad_norm": 0.5032907724380493, "learning_rate": 0.0005, "loss": 1.5804808139801025, "step": 14757 }, { "epoch": 1.1634213638155302, "grad_norm": 0.668984055519104, "learning_rate": 0.0005, "loss": 1.9323277473449707, "step": 14758 }, { "epoch": 1.163500197083169, "grad_norm": 0.4922429025173187, "learning_rate": 0.0005, "loss": 1.7832894325256348, "step": 14759 }, { "epoch": 1.163579030350808, "grad_norm": 0.5943044424057007, "learning_rate": 0.0005, "loss": 1.529077172279358, "step": 14760 }, { "epoch": 1.163657863618447, "grad_norm": 0.45017942786216736, "learning_rate": 0.0005, "loss": 1.4933865070343018, "step": 14761 }, { "epoch": 1.1637366968860858, "grad_norm": 0.4318336546421051, "learning_rate": 0.0005, "loss": 1.6097638607025146, "step": 14762 }, { "epoch": 1.163815530153725, "grad_norm": 0.44713321328163147, "learning_rate": 0.0005, "loss": 1.6505234241485596, "step": 14763 }, { "epoch": 1.1638943634213639, "grad_norm": 0.5319219827651978, "learning_rate": 0.0005, "loss": 1.6713571548461914, "step": 14764 }, { "epoch": 1.1639731966890028, "grad_norm": 0.6666430234909058, "learning_rate": 0.0005, "loss": 1.7379217147827148, "step": 14765 }, { "epoch": 1.1640520299566417, "grad_norm": 0.46906328201293945, "learning_rate": 0.0005, "loss": 1.7731832265853882, "step": 14766 }, { "epoch": 1.1641308632242806, "grad_norm": 0.630176842212677, "learning_rate": 0.0005, "loss": 1.673170804977417, "step": 14767 }, { "epoch": 1.1642096964919195, "grad_norm": 0.40770089626312256, "learning_rate": 0.0005, "loss": 1.5669002532958984, "step": 14768 }, { "epoch": 1.1642885297595584, "grad_norm": 0.5134689807891846, "learning_rate": 0.0005, "loss": 1.7670462131500244, "step": 14769 }, { "epoch": 1.1643673630271976, "grad_norm": 0.5202974081039429, "learning_rate": 0.0005, "loss": 1.706383466720581, "step": 14770 }, { "epoch": 1.1644461962948365, "grad_norm": 0.5863208770751953, "learning_rate": 0.0005, "loss": 1.8283360004425049, "step": 14771 }, { "epoch": 1.1645250295624754, "grad_norm": 0.47042199969291687, "learning_rate": 0.0005, "loss": 1.666610598564148, "step": 14772 }, { "epoch": 1.1646038628301143, "grad_norm": 0.5000407695770264, "learning_rate": 0.0005, "loss": 1.8149135112762451, "step": 14773 }, { "epoch": 1.1646826960977532, "grad_norm": 0.6542022824287415, "learning_rate": 0.0005, "loss": 1.8068400621414185, "step": 14774 }, { "epoch": 1.1647615293653921, "grad_norm": 0.535125195980072, "learning_rate": 0.0005, "loss": 1.5281941890716553, "step": 14775 }, { "epoch": 1.164840362633031, "grad_norm": 0.5396705865859985, "learning_rate": 0.0005, "loss": 1.8483874797821045, "step": 14776 }, { "epoch": 1.1649191959006702, "grad_norm": 0.7147555947303772, "learning_rate": 0.0005, "loss": 2.128573417663574, "step": 14777 }, { "epoch": 1.164998029168309, "grad_norm": 0.647392213344574, "learning_rate": 0.0005, "loss": 1.7723973989486694, "step": 14778 }, { "epoch": 1.165076862435948, "grad_norm": 0.7066801190376282, "learning_rate": 0.0005, "loss": 1.4245333671569824, "step": 14779 }, { "epoch": 1.165155695703587, "grad_norm": 0.5633702874183655, "learning_rate": 0.0005, "loss": 1.721675157546997, "step": 14780 }, { "epoch": 1.1652345289712258, "grad_norm": 0.6600817441940308, "learning_rate": 0.0005, "loss": 1.5957841873168945, "step": 14781 }, { "epoch": 1.1653133622388647, "grad_norm": 0.5266275405883789, "learning_rate": 0.0005, "loss": 1.7158293724060059, "step": 14782 }, { "epoch": 1.1653921955065036, "grad_norm": 0.7986271381378174, "learning_rate": 0.0005, "loss": 1.876487135887146, "step": 14783 }, { "epoch": 1.1654710287741428, "grad_norm": 0.48764610290527344, "learning_rate": 0.0005, "loss": 1.5445773601531982, "step": 14784 }, { "epoch": 1.1655498620417817, "grad_norm": 1.1194628477096558, "learning_rate": 0.0005, "loss": 1.4717340469360352, "step": 14785 }, { "epoch": 1.1656286953094206, "grad_norm": 0.5806705355644226, "learning_rate": 0.0005, "loss": 1.678279995918274, "step": 14786 }, { "epoch": 1.1657075285770595, "grad_norm": 0.50586998462677, "learning_rate": 0.0005, "loss": 1.734464168548584, "step": 14787 }, { "epoch": 1.1657863618446984, "grad_norm": 0.5304727554321289, "learning_rate": 0.0005, "loss": 1.8060557842254639, "step": 14788 }, { "epoch": 1.1658651951123373, "grad_norm": 0.5776223540306091, "learning_rate": 0.0005, "loss": 1.5974336862564087, "step": 14789 }, { "epoch": 1.1659440283799762, "grad_norm": 0.9287005662918091, "learning_rate": 0.0005, "loss": 1.8799009323120117, "step": 14790 }, { "epoch": 1.1660228616476154, "grad_norm": 0.4629890024662018, "learning_rate": 0.0005, "loss": 1.509528398513794, "step": 14791 }, { "epoch": 1.1661016949152543, "grad_norm": 0.5739027261734009, "learning_rate": 0.0005, "loss": 1.5487087965011597, "step": 14792 }, { "epoch": 1.1661805281828932, "grad_norm": 0.5976693630218506, "learning_rate": 0.0005, "loss": 1.7421430349349976, "step": 14793 }, { "epoch": 1.166259361450532, "grad_norm": 0.6057692766189575, "learning_rate": 0.0005, "loss": 1.660538911819458, "step": 14794 }, { "epoch": 1.166338194718171, "grad_norm": 0.546769917011261, "learning_rate": 0.0005, "loss": 1.5512335300445557, "step": 14795 }, { "epoch": 1.16641702798581, "grad_norm": 2.563098192214966, "learning_rate": 0.0005, "loss": 1.50984525680542, "step": 14796 }, { "epoch": 1.1664958612534488, "grad_norm": 0.49255087971687317, "learning_rate": 0.0005, "loss": 1.6028945446014404, "step": 14797 }, { "epoch": 1.166574694521088, "grad_norm": 0.5833403468132019, "learning_rate": 0.0005, "loss": 1.431780219078064, "step": 14798 }, { "epoch": 1.1666535277887269, "grad_norm": 0.7004661560058594, "learning_rate": 0.0005, "loss": 1.8002784252166748, "step": 14799 }, { "epoch": 1.1667323610563658, "grad_norm": 0.7311219573020935, "learning_rate": 0.0005, "loss": 1.7390248775482178, "step": 14800 }, { "epoch": 1.1668111943240047, "grad_norm": 0.5628801584243774, "learning_rate": 0.0005, "loss": 1.5359346866607666, "step": 14801 }, { "epoch": 1.1668900275916436, "grad_norm": 0.5099985003471375, "learning_rate": 0.0005, "loss": 1.5832982063293457, "step": 14802 }, { "epoch": 1.1669688608592825, "grad_norm": 0.6251236200332642, "learning_rate": 0.0005, "loss": 1.6210236549377441, "step": 14803 }, { "epoch": 1.1670476941269214, "grad_norm": 0.6436808109283447, "learning_rate": 0.0005, "loss": 2.057828903198242, "step": 14804 }, { "epoch": 1.1671265273945606, "grad_norm": 0.5943015217781067, "learning_rate": 0.0005, "loss": 2.0447609424591064, "step": 14805 }, { "epoch": 1.1672053606621995, "grad_norm": 0.4539163112640381, "learning_rate": 0.0005, "loss": 1.4323146343231201, "step": 14806 }, { "epoch": 1.1672841939298384, "grad_norm": 0.4846813678741455, "learning_rate": 0.0005, "loss": 1.5433073043823242, "step": 14807 }, { "epoch": 1.1673630271974773, "grad_norm": 0.7331525683403015, "learning_rate": 0.0005, "loss": 1.567083477973938, "step": 14808 }, { "epoch": 1.1674418604651162, "grad_norm": 0.8254988193511963, "learning_rate": 0.0005, "loss": 1.9417407512664795, "step": 14809 }, { "epoch": 1.1675206937327551, "grad_norm": 0.5231937170028687, "learning_rate": 0.0005, "loss": 1.7080225944519043, "step": 14810 }, { "epoch": 1.167599527000394, "grad_norm": 0.534320056438446, "learning_rate": 0.0005, "loss": 1.9711754322052002, "step": 14811 }, { "epoch": 1.1676783602680332, "grad_norm": 0.6015343070030212, "learning_rate": 0.0005, "loss": 1.7443331480026245, "step": 14812 }, { "epoch": 1.167757193535672, "grad_norm": 0.6063361167907715, "learning_rate": 0.0005, "loss": 1.8399856090545654, "step": 14813 }, { "epoch": 1.167836026803311, "grad_norm": 0.49083802103996277, "learning_rate": 0.0005, "loss": 1.5760889053344727, "step": 14814 }, { "epoch": 1.16791486007095, "grad_norm": 0.6481942534446716, "learning_rate": 0.0005, "loss": 1.8293756246566772, "step": 14815 }, { "epoch": 1.1679936933385888, "grad_norm": 0.8077271580696106, "learning_rate": 0.0005, "loss": 1.9971847534179688, "step": 14816 }, { "epoch": 1.1680725266062277, "grad_norm": 0.7277699112892151, "learning_rate": 0.0005, "loss": 1.7237303256988525, "step": 14817 }, { "epoch": 1.1681513598738669, "grad_norm": 0.5167619585990906, "learning_rate": 0.0005, "loss": 1.8071463108062744, "step": 14818 }, { "epoch": 1.1682301931415058, "grad_norm": 0.5389079451560974, "learning_rate": 0.0005, "loss": 1.6785255670547485, "step": 14819 }, { "epoch": 1.1683090264091447, "grad_norm": 0.61678147315979, "learning_rate": 0.0005, "loss": 1.8967479467391968, "step": 14820 }, { "epoch": 1.1683878596767836, "grad_norm": 0.4891737997531891, "learning_rate": 0.0005, "loss": 1.8614369630813599, "step": 14821 }, { "epoch": 1.1684666929444225, "grad_norm": 0.49665367603302, "learning_rate": 0.0005, "loss": 1.4990813732147217, "step": 14822 }, { "epoch": 1.1685455262120614, "grad_norm": 0.5356688499450684, "learning_rate": 0.0005, "loss": 1.5757025480270386, "step": 14823 }, { "epoch": 1.1686243594797003, "grad_norm": 0.681460976600647, "learning_rate": 0.0005, "loss": 1.80338716506958, "step": 14824 }, { "epoch": 1.1687031927473395, "grad_norm": 0.5081955194473267, "learning_rate": 0.0005, "loss": 1.5290768146514893, "step": 14825 }, { "epoch": 1.1687820260149784, "grad_norm": 0.481284499168396, "learning_rate": 0.0005, "loss": 1.7250871658325195, "step": 14826 }, { "epoch": 1.1688608592826173, "grad_norm": 0.5578591823577881, "learning_rate": 0.0005, "loss": 1.6242802143096924, "step": 14827 }, { "epoch": 1.1689396925502562, "grad_norm": 0.5071128606796265, "learning_rate": 0.0005, "loss": 1.6710326671600342, "step": 14828 }, { "epoch": 1.1690185258178951, "grad_norm": 0.4893966317176819, "learning_rate": 0.0005, "loss": 1.8300714492797852, "step": 14829 }, { "epoch": 1.169097359085534, "grad_norm": 0.542374849319458, "learning_rate": 0.0005, "loss": 1.8094085454940796, "step": 14830 }, { "epoch": 1.1691761923531732, "grad_norm": 0.5879952311515808, "learning_rate": 0.0005, "loss": 1.848506212234497, "step": 14831 }, { "epoch": 1.169255025620812, "grad_norm": 0.6179021000862122, "learning_rate": 0.0005, "loss": 1.7868183851242065, "step": 14832 }, { "epoch": 1.169333858888451, "grad_norm": 0.5501428246498108, "learning_rate": 0.0005, "loss": 1.617438793182373, "step": 14833 }, { "epoch": 1.16941269215609, "grad_norm": 0.5382853150367737, "learning_rate": 0.0005, "loss": 2.0236382484436035, "step": 14834 }, { "epoch": 1.1694915254237288, "grad_norm": 0.5627709627151489, "learning_rate": 0.0005, "loss": 1.792283296585083, "step": 14835 }, { "epoch": 1.1695703586913677, "grad_norm": 0.45582321286201477, "learning_rate": 0.0005, "loss": 1.5140111446380615, "step": 14836 }, { "epoch": 1.1696491919590066, "grad_norm": 0.45384180545806885, "learning_rate": 0.0005, "loss": 1.5810351371765137, "step": 14837 }, { "epoch": 1.1697280252266458, "grad_norm": 0.5164992213249207, "learning_rate": 0.0005, "loss": 1.627272605895996, "step": 14838 }, { "epoch": 1.1698068584942847, "grad_norm": 0.5104542374610901, "learning_rate": 0.0005, "loss": 1.5811338424682617, "step": 14839 }, { "epoch": 1.1698856917619236, "grad_norm": 0.5997002124786377, "learning_rate": 0.0005, "loss": 1.520737648010254, "step": 14840 }, { "epoch": 1.1699645250295625, "grad_norm": 0.5520023703575134, "learning_rate": 0.0005, "loss": 1.6488583087921143, "step": 14841 }, { "epoch": 1.1700433582972014, "grad_norm": 2.1772994995117188, "learning_rate": 0.0005, "loss": 1.6640726327896118, "step": 14842 }, { "epoch": 1.1701221915648403, "grad_norm": 0.5359391570091248, "learning_rate": 0.0005, "loss": 1.896761417388916, "step": 14843 }, { "epoch": 1.1702010248324792, "grad_norm": 0.536802351474762, "learning_rate": 0.0005, "loss": 1.629617691040039, "step": 14844 }, { "epoch": 1.1702798581001184, "grad_norm": 1.0691792964935303, "learning_rate": 0.0005, "loss": 1.876215934753418, "step": 14845 }, { "epoch": 1.1703586913677573, "grad_norm": 0.5979534983634949, "learning_rate": 0.0005, "loss": 1.9094133377075195, "step": 14846 }, { "epoch": 1.1704375246353962, "grad_norm": 0.5898892283439636, "learning_rate": 0.0005, "loss": 1.7851121425628662, "step": 14847 }, { "epoch": 1.170516357903035, "grad_norm": 0.6506102085113525, "learning_rate": 0.0005, "loss": 1.491865634918213, "step": 14848 }, { "epoch": 1.170595191170674, "grad_norm": 0.6229507327079773, "learning_rate": 0.0005, "loss": 1.5862071514129639, "step": 14849 }, { "epoch": 1.170674024438313, "grad_norm": 0.7329683899879456, "learning_rate": 0.0005, "loss": 1.616201639175415, "step": 14850 }, { "epoch": 1.1707528577059518, "grad_norm": 0.5893822312355042, "learning_rate": 0.0005, "loss": 1.757781744003296, "step": 14851 }, { "epoch": 1.170831690973591, "grad_norm": 0.5688644647598267, "learning_rate": 0.0005, "loss": 1.653821587562561, "step": 14852 }, { "epoch": 1.1709105242412299, "grad_norm": 0.5209906697273254, "learning_rate": 0.0005, "loss": 1.727931022644043, "step": 14853 }, { "epoch": 1.1709893575088688, "grad_norm": 0.5690051913261414, "learning_rate": 0.0005, "loss": 1.5524405241012573, "step": 14854 }, { "epoch": 1.1710681907765077, "grad_norm": 0.5778870582580566, "learning_rate": 0.0005, "loss": 1.8801312446594238, "step": 14855 }, { "epoch": 1.1711470240441466, "grad_norm": 0.5508738160133362, "learning_rate": 0.0005, "loss": 1.8317210674285889, "step": 14856 }, { "epoch": 1.1712258573117855, "grad_norm": 0.4388248324394226, "learning_rate": 0.0005, "loss": 1.5019619464874268, "step": 14857 }, { "epoch": 1.1713046905794244, "grad_norm": 0.69437175989151, "learning_rate": 0.0005, "loss": 1.5116117000579834, "step": 14858 }, { "epoch": 1.1713835238470636, "grad_norm": 0.5289844870567322, "learning_rate": 0.0005, "loss": 1.7985572814941406, "step": 14859 }, { "epoch": 1.1714623571147025, "grad_norm": 0.6428233981132507, "learning_rate": 0.0005, "loss": 1.7918660640716553, "step": 14860 }, { "epoch": 1.1715411903823414, "grad_norm": 0.5569784641265869, "learning_rate": 0.0005, "loss": 1.7608299255371094, "step": 14861 }, { "epoch": 1.1716200236499803, "grad_norm": 0.5860293507575989, "learning_rate": 0.0005, "loss": 1.7358896732330322, "step": 14862 }, { "epoch": 1.1716988569176192, "grad_norm": 0.6926673054695129, "learning_rate": 0.0005, "loss": 1.8722567558288574, "step": 14863 }, { "epoch": 1.1717776901852581, "grad_norm": 0.5504025816917419, "learning_rate": 0.0005, "loss": 1.693844199180603, "step": 14864 }, { "epoch": 1.171856523452897, "grad_norm": 0.4906632602214813, "learning_rate": 0.0005, "loss": 1.548705816268921, "step": 14865 }, { "epoch": 1.1719353567205362, "grad_norm": 0.5271687507629395, "learning_rate": 0.0005, "loss": 1.6645598411560059, "step": 14866 }, { "epoch": 1.172014189988175, "grad_norm": 0.48650607466697693, "learning_rate": 0.0005, "loss": 1.7052693367004395, "step": 14867 }, { "epoch": 1.172093023255814, "grad_norm": 0.6012440323829651, "learning_rate": 0.0005, "loss": 1.870011568069458, "step": 14868 }, { "epoch": 1.172171856523453, "grad_norm": 0.6478787660598755, "learning_rate": 0.0005, "loss": 1.6826132535934448, "step": 14869 }, { "epoch": 1.1722506897910918, "grad_norm": 0.4842686355113983, "learning_rate": 0.0005, "loss": 1.8172656297683716, "step": 14870 }, { "epoch": 1.1723295230587307, "grad_norm": 0.5489843487739563, "learning_rate": 0.0005, "loss": 1.7270147800445557, "step": 14871 }, { "epoch": 1.1724083563263696, "grad_norm": 0.5644809007644653, "learning_rate": 0.0005, "loss": 1.5981037616729736, "step": 14872 }, { "epoch": 1.1724871895940088, "grad_norm": 0.5247041583061218, "learning_rate": 0.0005, "loss": 1.4758763313293457, "step": 14873 }, { "epoch": 1.1725660228616477, "grad_norm": 0.5486081838607788, "learning_rate": 0.0005, "loss": 1.6637368202209473, "step": 14874 }, { "epoch": 1.1726448561292866, "grad_norm": 0.5308844447135925, "learning_rate": 0.0005, "loss": 1.6973612308502197, "step": 14875 }, { "epoch": 1.1727236893969255, "grad_norm": 0.5416121482849121, "learning_rate": 0.0005, "loss": 1.729861855506897, "step": 14876 }, { "epoch": 1.1728025226645644, "grad_norm": 0.5323233008384705, "learning_rate": 0.0005, "loss": 1.5463371276855469, "step": 14877 }, { "epoch": 1.1728813559322033, "grad_norm": 0.5646855235099792, "learning_rate": 0.0005, "loss": 1.7410112619400024, "step": 14878 }, { "epoch": 1.1729601891998422, "grad_norm": 0.5395973324775696, "learning_rate": 0.0005, "loss": 1.5818684101104736, "step": 14879 }, { "epoch": 1.1730390224674814, "grad_norm": 0.5420454144477844, "learning_rate": 0.0005, "loss": 1.9105474948883057, "step": 14880 }, { "epoch": 1.1731178557351203, "grad_norm": 0.5437034368515015, "learning_rate": 0.0005, "loss": 1.647279143333435, "step": 14881 }, { "epoch": 1.1731966890027592, "grad_norm": 0.5024282336235046, "learning_rate": 0.0005, "loss": 1.609689712524414, "step": 14882 }, { "epoch": 1.173275522270398, "grad_norm": 0.48374319076538086, "learning_rate": 0.0005, "loss": 1.7984763383865356, "step": 14883 }, { "epoch": 1.173354355538037, "grad_norm": 0.582013726234436, "learning_rate": 0.0005, "loss": 1.6290948390960693, "step": 14884 }, { "epoch": 1.173433188805676, "grad_norm": 0.4700412452220917, "learning_rate": 0.0005, "loss": 1.583152174949646, "step": 14885 }, { "epoch": 1.1735120220733148, "grad_norm": 0.5591966509819031, "learning_rate": 0.0005, "loss": 1.9710984230041504, "step": 14886 }, { "epoch": 1.173590855340954, "grad_norm": 0.5340425372123718, "learning_rate": 0.0005, "loss": 1.3663041591644287, "step": 14887 }, { "epoch": 1.1736696886085929, "grad_norm": 0.5013870596885681, "learning_rate": 0.0005, "loss": 1.6480472087860107, "step": 14888 }, { "epoch": 1.1737485218762318, "grad_norm": 0.5904043912887573, "learning_rate": 0.0005, "loss": 1.8354051113128662, "step": 14889 }, { "epoch": 1.1738273551438707, "grad_norm": 0.4823387563228607, "learning_rate": 0.0005, "loss": 1.7117624282836914, "step": 14890 }, { "epoch": 1.1739061884115096, "grad_norm": 0.5461208820343018, "learning_rate": 0.0005, "loss": 2.0452499389648438, "step": 14891 }, { "epoch": 1.1739850216791485, "grad_norm": 0.5051956176757812, "learning_rate": 0.0005, "loss": 1.6995341777801514, "step": 14892 }, { "epoch": 1.1740638549467874, "grad_norm": 0.6342728137969971, "learning_rate": 0.0005, "loss": 1.7079178094863892, "step": 14893 }, { "epoch": 1.1741426882144266, "grad_norm": 0.7103026509284973, "learning_rate": 0.0005, "loss": 1.899221420288086, "step": 14894 }, { "epoch": 1.1742215214820655, "grad_norm": 0.5628573298454285, "learning_rate": 0.0005, "loss": 1.7667628526687622, "step": 14895 }, { "epoch": 1.1743003547497044, "grad_norm": 0.5435157418251038, "learning_rate": 0.0005, "loss": 1.6834585666656494, "step": 14896 }, { "epoch": 1.1743791880173433, "grad_norm": 0.5501986145973206, "learning_rate": 0.0005, "loss": 1.482956886291504, "step": 14897 }, { "epoch": 1.1744580212849822, "grad_norm": 0.533971905708313, "learning_rate": 0.0005, "loss": 1.4393494129180908, "step": 14898 }, { "epoch": 1.1745368545526211, "grad_norm": 0.6317647099494934, "learning_rate": 0.0005, "loss": 1.7479510307312012, "step": 14899 }, { "epoch": 1.17461568782026, "grad_norm": 0.5139126181602478, "learning_rate": 0.0005, "loss": 1.861828088760376, "step": 14900 }, { "epoch": 1.1746945210878992, "grad_norm": 0.5430934429168701, "learning_rate": 0.0005, "loss": 1.6462205648422241, "step": 14901 }, { "epoch": 1.174773354355538, "grad_norm": 0.7858989238739014, "learning_rate": 0.0005, "loss": 1.5654850006103516, "step": 14902 }, { "epoch": 1.174852187623177, "grad_norm": 0.6404993534088135, "learning_rate": 0.0005, "loss": 1.9264843463897705, "step": 14903 }, { "epoch": 1.174931020890816, "grad_norm": 0.8226879239082336, "learning_rate": 0.0005, "loss": 1.7983016967773438, "step": 14904 }, { "epoch": 1.1750098541584548, "grad_norm": 0.9306414127349854, "learning_rate": 0.0005, "loss": 1.651810884475708, "step": 14905 }, { "epoch": 1.1750886874260937, "grad_norm": 0.4778617322444916, "learning_rate": 0.0005, "loss": 1.5547382831573486, "step": 14906 }, { "epoch": 1.1751675206937326, "grad_norm": 3.847215175628662, "learning_rate": 0.0005, "loss": 1.5815048217773438, "step": 14907 }, { "epoch": 1.1752463539613718, "grad_norm": 0.8204733729362488, "learning_rate": 0.0005, "loss": 1.8938655853271484, "step": 14908 }, { "epoch": 1.1753251872290107, "grad_norm": 0.5384106636047363, "learning_rate": 0.0005, "loss": 1.8768017292022705, "step": 14909 }, { "epoch": 1.1754040204966496, "grad_norm": 1.163092017173767, "learning_rate": 0.0005, "loss": 1.6556625366210938, "step": 14910 }, { "epoch": 1.1754828537642885, "grad_norm": 0.5722174048423767, "learning_rate": 0.0005, "loss": 1.6340755224227905, "step": 14911 }, { "epoch": 1.1755616870319274, "grad_norm": 0.6464200615882874, "learning_rate": 0.0005, "loss": 1.633277177810669, "step": 14912 }, { "epoch": 1.1756405202995663, "grad_norm": 0.5606274008750916, "learning_rate": 0.0005, "loss": 1.6479461193084717, "step": 14913 }, { "epoch": 1.1757193535672052, "grad_norm": 0.49046581983566284, "learning_rate": 0.0005, "loss": 1.5298995971679688, "step": 14914 }, { "epoch": 1.1757981868348444, "grad_norm": 1.2200961112976074, "learning_rate": 0.0005, "loss": 1.7729363441467285, "step": 14915 }, { "epoch": 1.1758770201024833, "grad_norm": 0.6058907508850098, "learning_rate": 0.0005, "loss": 1.6443188190460205, "step": 14916 }, { "epoch": 1.1759558533701222, "grad_norm": 0.6593195796012878, "learning_rate": 0.0005, "loss": 1.6632015705108643, "step": 14917 }, { "epoch": 1.1760346866377611, "grad_norm": 0.8311710357666016, "learning_rate": 0.0005, "loss": 1.8741481304168701, "step": 14918 }, { "epoch": 1.1761135199054, "grad_norm": 0.6314473152160645, "learning_rate": 0.0005, "loss": 1.8372788429260254, "step": 14919 }, { "epoch": 1.176192353173039, "grad_norm": 0.6963744759559631, "learning_rate": 0.0005, "loss": 1.8546178340911865, "step": 14920 }, { "epoch": 1.1762711864406779, "grad_norm": 0.549301028251648, "learning_rate": 0.0005, "loss": 1.575016736984253, "step": 14921 }, { "epoch": 1.176350019708317, "grad_norm": 0.4924038052558899, "learning_rate": 0.0005, "loss": 1.5960592031478882, "step": 14922 }, { "epoch": 1.176428852975956, "grad_norm": 0.5588400959968567, "learning_rate": 0.0005, "loss": 1.9068269729614258, "step": 14923 }, { "epoch": 1.1765076862435948, "grad_norm": 0.6562055945396423, "learning_rate": 0.0005, "loss": 1.6700069904327393, "step": 14924 }, { "epoch": 1.1765865195112337, "grad_norm": 0.6237421035766602, "learning_rate": 0.0005, "loss": 2.037928819656372, "step": 14925 }, { "epoch": 1.1766653527788726, "grad_norm": 0.5831493139266968, "learning_rate": 0.0005, "loss": 1.7882648706436157, "step": 14926 }, { "epoch": 1.1767441860465115, "grad_norm": 0.441476970911026, "learning_rate": 0.0005, "loss": 1.4317917823791504, "step": 14927 }, { "epoch": 1.1768230193141505, "grad_norm": 0.5002341866493225, "learning_rate": 0.0005, "loss": 1.7509374618530273, "step": 14928 }, { "epoch": 1.1769018525817896, "grad_norm": 0.5297947525978088, "learning_rate": 0.0005, "loss": 1.586513638496399, "step": 14929 }, { "epoch": 1.1769806858494285, "grad_norm": 0.5944413542747498, "learning_rate": 0.0005, "loss": 1.7986655235290527, "step": 14930 }, { "epoch": 1.1770595191170674, "grad_norm": 0.45504775643348694, "learning_rate": 0.0005, "loss": 1.4668656587600708, "step": 14931 }, { "epoch": 1.1771383523847063, "grad_norm": 0.5802126526832581, "learning_rate": 0.0005, "loss": 1.6153111457824707, "step": 14932 }, { "epoch": 1.1772171856523452, "grad_norm": 0.5328615307807922, "learning_rate": 0.0005, "loss": 1.8173667192459106, "step": 14933 }, { "epoch": 1.1772960189199841, "grad_norm": 0.5911689400672913, "learning_rate": 0.0005, "loss": 1.7517906427383423, "step": 14934 }, { "epoch": 1.177374852187623, "grad_norm": 0.5873813629150391, "learning_rate": 0.0005, "loss": 2.020512342453003, "step": 14935 }, { "epoch": 1.1774536854552622, "grad_norm": 1.9162184000015259, "learning_rate": 0.0005, "loss": 1.8270970582962036, "step": 14936 }, { "epoch": 1.177532518722901, "grad_norm": 0.47491639852523804, "learning_rate": 0.0005, "loss": 1.4487038850784302, "step": 14937 }, { "epoch": 1.17761135199054, "grad_norm": 0.6356821656227112, "learning_rate": 0.0005, "loss": 1.8980604410171509, "step": 14938 }, { "epoch": 1.177690185258179, "grad_norm": 0.7555972337722778, "learning_rate": 0.0005, "loss": 1.8746540546417236, "step": 14939 }, { "epoch": 1.1777690185258178, "grad_norm": 0.5519038438796997, "learning_rate": 0.0005, "loss": 1.5591106414794922, "step": 14940 }, { "epoch": 1.1778478517934567, "grad_norm": 0.7330582141876221, "learning_rate": 0.0005, "loss": 1.7546755075454712, "step": 14941 }, { "epoch": 1.1779266850610959, "grad_norm": 0.5912792682647705, "learning_rate": 0.0005, "loss": 1.7361948490142822, "step": 14942 }, { "epoch": 1.1780055183287348, "grad_norm": 0.6486424803733826, "learning_rate": 0.0005, "loss": 1.7424023151397705, "step": 14943 }, { "epoch": 1.1780843515963737, "grad_norm": 0.5896940231323242, "learning_rate": 0.0005, "loss": 1.5787267684936523, "step": 14944 }, { "epoch": 1.1781631848640126, "grad_norm": 0.515076220035553, "learning_rate": 0.0005, "loss": 1.450797438621521, "step": 14945 }, { "epoch": 1.1782420181316515, "grad_norm": 0.6196977496147156, "learning_rate": 0.0005, "loss": 1.9448049068450928, "step": 14946 }, { "epoch": 1.1783208513992904, "grad_norm": 0.6103847026824951, "learning_rate": 0.0005, "loss": 1.8227453231811523, "step": 14947 }, { "epoch": 1.1783996846669293, "grad_norm": 0.46594226360321045, "learning_rate": 0.0005, "loss": 1.7282501459121704, "step": 14948 }, { "epoch": 1.1784785179345685, "grad_norm": 0.47506341338157654, "learning_rate": 0.0005, "loss": 1.5887235403060913, "step": 14949 }, { "epoch": 1.1785573512022074, "grad_norm": 0.5910429954528809, "learning_rate": 0.0005, "loss": 1.6551361083984375, "step": 14950 }, { "epoch": 1.1786361844698463, "grad_norm": 0.5549733638763428, "learning_rate": 0.0005, "loss": 1.4978384971618652, "step": 14951 }, { "epoch": 1.1787150177374852, "grad_norm": 0.8019834160804749, "learning_rate": 0.0005, "loss": 1.8464257717132568, "step": 14952 }, { "epoch": 1.1787938510051241, "grad_norm": 0.5862184166908264, "learning_rate": 0.0005, "loss": 1.6559271812438965, "step": 14953 }, { "epoch": 1.178872684272763, "grad_norm": 0.5084710717201233, "learning_rate": 0.0005, "loss": 1.3122717142105103, "step": 14954 }, { "epoch": 1.1789515175404022, "grad_norm": 0.8037705421447754, "learning_rate": 0.0005, "loss": 1.8012757301330566, "step": 14955 }, { "epoch": 1.179030350808041, "grad_norm": 0.5725706219673157, "learning_rate": 0.0005, "loss": 1.5646400451660156, "step": 14956 }, { "epoch": 1.17910918407568, "grad_norm": 0.7585146427154541, "learning_rate": 0.0005, "loss": 1.6008081436157227, "step": 14957 }, { "epoch": 1.179188017343319, "grad_norm": 0.5338521599769592, "learning_rate": 0.0005, "loss": 1.813563346862793, "step": 14958 }, { "epoch": 1.1792668506109578, "grad_norm": 0.538343608379364, "learning_rate": 0.0005, "loss": 1.689941167831421, "step": 14959 }, { "epoch": 1.1793456838785967, "grad_norm": 0.5296321511268616, "learning_rate": 0.0005, "loss": 1.5865763425827026, "step": 14960 }, { "epoch": 1.1794245171462356, "grad_norm": 0.5761128664016724, "learning_rate": 0.0005, "loss": 1.7321419715881348, "step": 14961 }, { "epoch": 1.1795033504138748, "grad_norm": 0.5474506616592407, "learning_rate": 0.0005, "loss": 1.7376184463500977, "step": 14962 }, { "epoch": 1.1795821836815137, "grad_norm": 0.493787556886673, "learning_rate": 0.0005, "loss": 1.6297799348831177, "step": 14963 }, { "epoch": 1.1796610169491526, "grad_norm": 0.49807634949684143, "learning_rate": 0.0005, "loss": 1.5006457567214966, "step": 14964 }, { "epoch": 1.1797398502167915, "grad_norm": 0.6829431056976318, "learning_rate": 0.0005, "loss": 1.8572862148284912, "step": 14965 }, { "epoch": 1.1798186834844304, "grad_norm": 0.5295193791389465, "learning_rate": 0.0005, "loss": 1.7498500347137451, "step": 14966 }, { "epoch": 1.1798975167520693, "grad_norm": 0.5218766927719116, "learning_rate": 0.0005, "loss": 1.6099417209625244, "step": 14967 }, { "epoch": 1.1799763500197082, "grad_norm": 0.5599879026412964, "learning_rate": 0.0005, "loss": 1.621617078781128, "step": 14968 }, { "epoch": 1.1800551832873474, "grad_norm": 0.6709167957305908, "learning_rate": 0.0005, "loss": 1.7063672542572021, "step": 14969 }, { "epoch": 1.1801340165549863, "grad_norm": 0.6447254419326782, "learning_rate": 0.0005, "loss": 1.6504836082458496, "step": 14970 }, { "epoch": 1.1802128498226252, "grad_norm": 0.46828240156173706, "learning_rate": 0.0005, "loss": 1.6549761295318604, "step": 14971 }, { "epoch": 1.180291683090264, "grad_norm": 0.6440352201461792, "learning_rate": 0.0005, "loss": 1.766568899154663, "step": 14972 }, { "epoch": 1.180370516357903, "grad_norm": 0.5618107318878174, "learning_rate": 0.0005, "loss": 1.8182812929153442, "step": 14973 }, { "epoch": 1.180449349625542, "grad_norm": 0.5576437711715698, "learning_rate": 0.0005, "loss": 1.6050752401351929, "step": 14974 }, { "epoch": 1.1805281828931808, "grad_norm": 0.5430149435997009, "learning_rate": 0.0005, "loss": 1.643483281135559, "step": 14975 }, { "epoch": 1.18060701616082, "grad_norm": 0.5588108897209167, "learning_rate": 0.0005, "loss": 1.490631341934204, "step": 14976 }, { "epoch": 1.1806858494284589, "grad_norm": 0.6631270051002502, "learning_rate": 0.0005, "loss": 1.740851879119873, "step": 14977 }, { "epoch": 1.1807646826960978, "grad_norm": 0.5246522426605225, "learning_rate": 0.0005, "loss": 1.4639558792114258, "step": 14978 }, { "epoch": 1.1808435159637367, "grad_norm": 0.5331841707229614, "learning_rate": 0.0005, "loss": 1.5866189002990723, "step": 14979 }, { "epoch": 1.1809223492313756, "grad_norm": 0.5216993689537048, "learning_rate": 0.0005, "loss": 1.717353105545044, "step": 14980 }, { "epoch": 1.1810011824990145, "grad_norm": 0.4379534125328064, "learning_rate": 0.0005, "loss": 1.4377495050430298, "step": 14981 }, { "epoch": 1.1810800157666534, "grad_norm": 0.5555115938186646, "learning_rate": 0.0005, "loss": 1.6286771297454834, "step": 14982 }, { "epoch": 1.1811588490342926, "grad_norm": 0.5111465454101562, "learning_rate": 0.0005, "loss": 1.5475308895111084, "step": 14983 }, { "epoch": 1.1812376823019315, "grad_norm": 0.44597694277763367, "learning_rate": 0.0005, "loss": 1.3895469903945923, "step": 14984 }, { "epoch": 1.1813165155695704, "grad_norm": 0.5851807594299316, "learning_rate": 0.0005, "loss": 1.6578781604766846, "step": 14985 }, { "epoch": 1.1813953488372093, "grad_norm": 0.5495781302452087, "learning_rate": 0.0005, "loss": 1.97707200050354, "step": 14986 }, { "epoch": 1.1814741821048482, "grad_norm": 0.5493658185005188, "learning_rate": 0.0005, "loss": 1.5240713357925415, "step": 14987 }, { "epoch": 1.1815530153724871, "grad_norm": 0.5286668539047241, "learning_rate": 0.0005, "loss": 1.6487149000167847, "step": 14988 }, { "epoch": 1.181631848640126, "grad_norm": 0.6009783744812012, "learning_rate": 0.0005, "loss": 1.6418570280075073, "step": 14989 }, { "epoch": 1.1817106819077652, "grad_norm": 0.7822855114936829, "learning_rate": 0.0005, "loss": 1.8239802122116089, "step": 14990 }, { "epoch": 1.181789515175404, "grad_norm": 0.4848598539829254, "learning_rate": 0.0005, "loss": 1.548224687576294, "step": 14991 }, { "epoch": 1.181868348443043, "grad_norm": 0.5643460750579834, "learning_rate": 0.0005, "loss": 1.4203240871429443, "step": 14992 }, { "epoch": 1.181947181710682, "grad_norm": 0.6719220876693726, "learning_rate": 0.0005, "loss": 1.7204562425613403, "step": 14993 }, { "epoch": 1.1820260149783208, "grad_norm": 0.6068660020828247, "learning_rate": 0.0005, "loss": 1.5025936365127563, "step": 14994 }, { "epoch": 1.1821048482459597, "grad_norm": 0.614340603351593, "learning_rate": 0.0005, "loss": 1.766420602798462, "step": 14995 }, { "epoch": 1.1821836815135986, "grad_norm": 0.9588218927383423, "learning_rate": 0.0005, "loss": 1.7560702562332153, "step": 14996 }, { "epoch": 1.1822625147812378, "grad_norm": 0.7416885495185852, "learning_rate": 0.0005, "loss": 1.7268368005752563, "step": 14997 }, { "epoch": 1.1823413480488767, "grad_norm": 0.6478714942932129, "learning_rate": 0.0005, "loss": 1.5615575313568115, "step": 14998 }, { "epoch": 1.1824201813165156, "grad_norm": 0.6365734338760376, "learning_rate": 0.0005, "loss": 1.7558249235153198, "step": 14999 }, { "epoch": 1.1824990145841545, "grad_norm": 0.5276303887367249, "learning_rate": 0.0005, "loss": 1.5750203132629395, "step": 15000 }, { "epoch": 1.1825778478517934, "grad_norm": 0.5637723803520203, "learning_rate": 0.0005, "loss": 1.8138096332550049, "step": 15001 }, { "epoch": 1.1826566811194323, "grad_norm": 0.49935463070869446, "learning_rate": 0.0005, "loss": 1.5872786045074463, "step": 15002 }, { "epoch": 1.1827355143870713, "grad_norm": 0.4788287281990051, "learning_rate": 0.0005, "loss": 1.4203577041625977, "step": 15003 }, { "epoch": 1.1828143476547104, "grad_norm": 0.7095337510108948, "learning_rate": 0.0005, "loss": 1.939183235168457, "step": 15004 }, { "epoch": 1.1828931809223493, "grad_norm": 0.6035265922546387, "learning_rate": 0.0005, "loss": 1.8021354675292969, "step": 15005 }, { "epoch": 1.1829720141899882, "grad_norm": 0.5997790694236755, "learning_rate": 0.0005, "loss": 1.6374659538269043, "step": 15006 }, { "epoch": 1.1830508474576271, "grad_norm": 0.5499025583267212, "learning_rate": 0.0005, "loss": 1.93712317943573, "step": 15007 }, { "epoch": 1.183129680725266, "grad_norm": 0.5795935988426208, "learning_rate": 0.0005, "loss": 1.6376423835754395, "step": 15008 }, { "epoch": 1.183208513992905, "grad_norm": 0.4891674518585205, "learning_rate": 0.0005, "loss": 1.5791560411453247, "step": 15009 }, { "epoch": 1.1832873472605439, "grad_norm": 0.5299795269966125, "learning_rate": 0.0005, "loss": 1.7612532377243042, "step": 15010 }, { "epoch": 1.183366180528183, "grad_norm": 0.44582420587539673, "learning_rate": 0.0005, "loss": 1.65611732006073, "step": 15011 }, { "epoch": 1.183445013795822, "grad_norm": 0.5499736070632935, "learning_rate": 0.0005, "loss": 1.6855677366256714, "step": 15012 }, { "epoch": 1.1835238470634608, "grad_norm": 0.763435959815979, "learning_rate": 0.0005, "loss": 1.7087252140045166, "step": 15013 }, { "epoch": 1.1836026803310997, "grad_norm": 0.5852077007293701, "learning_rate": 0.0005, "loss": 1.6647789478302002, "step": 15014 }, { "epoch": 1.1836815135987386, "grad_norm": 0.541266143321991, "learning_rate": 0.0005, "loss": 1.6935062408447266, "step": 15015 }, { "epoch": 1.1837603468663775, "grad_norm": 0.515537440776825, "learning_rate": 0.0005, "loss": 1.5533990859985352, "step": 15016 }, { "epoch": 1.1838391801340165, "grad_norm": 0.6101832389831543, "learning_rate": 0.0005, "loss": 1.6909968852996826, "step": 15017 }, { "epoch": 1.1839180134016556, "grad_norm": 0.6199068427085876, "learning_rate": 0.0005, "loss": 1.7283844947814941, "step": 15018 }, { "epoch": 1.1839968466692945, "grad_norm": 0.5088817477226257, "learning_rate": 0.0005, "loss": 1.6143420934677124, "step": 15019 }, { "epoch": 1.1840756799369334, "grad_norm": 0.6568331122398376, "learning_rate": 0.0005, "loss": 1.5024051666259766, "step": 15020 }, { "epoch": 1.1841545132045723, "grad_norm": 0.6134170889854431, "learning_rate": 0.0005, "loss": 1.5582427978515625, "step": 15021 }, { "epoch": 1.1842333464722112, "grad_norm": 0.6567361950874329, "learning_rate": 0.0005, "loss": 1.6252305507659912, "step": 15022 }, { "epoch": 1.1843121797398501, "grad_norm": 0.627960741519928, "learning_rate": 0.0005, "loss": 1.9185632467269897, "step": 15023 }, { "epoch": 1.184391013007489, "grad_norm": 0.4854506850242615, "learning_rate": 0.0005, "loss": 1.5909209251403809, "step": 15024 }, { "epoch": 1.1844698462751282, "grad_norm": 0.5030145645141602, "learning_rate": 0.0005, "loss": 1.7705930471420288, "step": 15025 }, { "epoch": 1.184548679542767, "grad_norm": 0.5433763861656189, "learning_rate": 0.0005, "loss": 1.916089415550232, "step": 15026 }, { "epoch": 1.184627512810406, "grad_norm": 0.5130103826522827, "learning_rate": 0.0005, "loss": 1.590423345565796, "step": 15027 }, { "epoch": 1.184706346078045, "grad_norm": 0.6243157982826233, "learning_rate": 0.0005, "loss": 1.7193126678466797, "step": 15028 }, { "epoch": 1.1847851793456838, "grad_norm": 0.50975501537323, "learning_rate": 0.0005, "loss": 1.6525933742523193, "step": 15029 }, { "epoch": 1.1848640126133227, "grad_norm": 0.5089796781539917, "learning_rate": 0.0005, "loss": 1.359140396118164, "step": 15030 }, { "epoch": 1.1849428458809617, "grad_norm": 0.6005828976631165, "learning_rate": 0.0005, "loss": 1.6160764694213867, "step": 15031 }, { "epoch": 1.1850216791486008, "grad_norm": 0.5112869739532471, "learning_rate": 0.0005, "loss": 1.412285566329956, "step": 15032 }, { "epoch": 1.1851005124162397, "grad_norm": 0.4996456503868103, "learning_rate": 0.0005, "loss": 1.6224204301834106, "step": 15033 }, { "epoch": 1.1851793456838786, "grad_norm": 0.6219953298568726, "learning_rate": 0.0005, "loss": 1.706979513168335, "step": 15034 }, { "epoch": 1.1852581789515175, "grad_norm": 0.7622995376586914, "learning_rate": 0.0005, "loss": 1.7510695457458496, "step": 15035 }, { "epoch": 1.1853370122191564, "grad_norm": 0.4735175371170044, "learning_rate": 0.0005, "loss": 1.5020368099212646, "step": 15036 }, { "epoch": 1.1854158454867953, "grad_norm": 0.487993061542511, "learning_rate": 0.0005, "loss": 1.5703084468841553, "step": 15037 }, { "epoch": 1.1854946787544343, "grad_norm": 0.572178304195404, "learning_rate": 0.0005, "loss": 1.8130369186401367, "step": 15038 }, { "epoch": 1.1855735120220734, "grad_norm": 0.7094960808753967, "learning_rate": 0.0005, "loss": 1.7823729515075684, "step": 15039 }, { "epoch": 1.1856523452897123, "grad_norm": 0.5522006750106812, "learning_rate": 0.0005, "loss": 1.7415108680725098, "step": 15040 }, { "epoch": 1.1857311785573512, "grad_norm": 0.7573570013046265, "learning_rate": 0.0005, "loss": 1.593096375465393, "step": 15041 }, { "epoch": 1.1858100118249901, "grad_norm": 0.5861647129058838, "learning_rate": 0.0005, "loss": 1.737415075302124, "step": 15042 }, { "epoch": 1.185888845092629, "grad_norm": 0.4611140489578247, "learning_rate": 0.0005, "loss": 1.5354222059249878, "step": 15043 }, { "epoch": 1.185967678360268, "grad_norm": 0.5152342319488525, "learning_rate": 0.0005, "loss": 1.633906602859497, "step": 15044 }, { "epoch": 1.1860465116279069, "grad_norm": 0.6425832509994507, "learning_rate": 0.0005, "loss": 1.7709972858428955, "step": 15045 }, { "epoch": 1.186125344895546, "grad_norm": 0.5040218234062195, "learning_rate": 0.0005, "loss": 1.6495615243911743, "step": 15046 }, { "epoch": 1.186204178163185, "grad_norm": 0.5517052412033081, "learning_rate": 0.0005, "loss": 1.5025224685668945, "step": 15047 }, { "epoch": 1.1862830114308238, "grad_norm": 0.4676332175731659, "learning_rate": 0.0005, "loss": 1.729489803314209, "step": 15048 }, { "epoch": 1.1863618446984627, "grad_norm": 0.5351168513298035, "learning_rate": 0.0005, "loss": 1.7341216802597046, "step": 15049 }, { "epoch": 1.1864406779661016, "grad_norm": 0.5472707748413086, "learning_rate": 0.0005, "loss": 1.5747941732406616, "step": 15050 }, { "epoch": 1.1865195112337406, "grad_norm": 0.9892175793647766, "learning_rate": 0.0005, "loss": 1.656839370727539, "step": 15051 }, { "epoch": 1.1865983445013795, "grad_norm": 0.5504602789878845, "learning_rate": 0.0005, "loss": 1.5922670364379883, "step": 15052 }, { "epoch": 1.1866771777690186, "grad_norm": 0.7155340909957886, "learning_rate": 0.0005, "loss": 1.856553554534912, "step": 15053 }, { "epoch": 1.1867560110366575, "grad_norm": 0.617402970790863, "learning_rate": 0.0005, "loss": 1.5938549041748047, "step": 15054 }, { "epoch": 1.1868348443042964, "grad_norm": 0.49500295519828796, "learning_rate": 0.0005, "loss": 1.448211431503296, "step": 15055 }, { "epoch": 1.1869136775719353, "grad_norm": 0.4725003242492676, "learning_rate": 0.0005, "loss": 1.6422436237335205, "step": 15056 }, { "epoch": 1.1869925108395742, "grad_norm": 0.5444419980049133, "learning_rate": 0.0005, "loss": 1.5289077758789062, "step": 15057 }, { "epoch": 1.1870713441072132, "grad_norm": 0.5181954503059387, "learning_rate": 0.0005, "loss": 1.4797452688217163, "step": 15058 }, { "epoch": 1.187150177374852, "grad_norm": 0.5509660243988037, "learning_rate": 0.0005, "loss": 1.6187305450439453, "step": 15059 }, { "epoch": 1.1872290106424912, "grad_norm": 0.5048860907554626, "learning_rate": 0.0005, "loss": 1.6310951709747314, "step": 15060 }, { "epoch": 1.1873078439101301, "grad_norm": 0.7365735173225403, "learning_rate": 0.0005, "loss": 1.6853526830673218, "step": 15061 }, { "epoch": 1.187386677177769, "grad_norm": 0.5845659375190735, "learning_rate": 0.0005, "loss": 1.9330365657806396, "step": 15062 }, { "epoch": 1.187465510445408, "grad_norm": 0.49966585636138916, "learning_rate": 0.0005, "loss": 1.5455803871154785, "step": 15063 }, { "epoch": 1.1875443437130468, "grad_norm": 0.6423776745796204, "learning_rate": 0.0005, "loss": 2.047694683074951, "step": 15064 }, { "epoch": 1.1876231769806858, "grad_norm": 0.5115677118301392, "learning_rate": 0.0005, "loss": 1.5072057247161865, "step": 15065 }, { "epoch": 1.1877020102483247, "grad_norm": 0.5467529296875, "learning_rate": 0.0005, "loss": 1.5405447483062744, "step": 15066 }, { "epoch": 1.1877808435159638, "grad_norm": 0.7943440675735474, "learning_rate": 0.0005, "loss": 1.9362976551055908, "step": 15067 }, { "epoch": 1.1878596767836027, "grad_norm": 0.6418803334236145, "learning_rate": 0.0005, "loss": 1.8233983516693115, "step": 15068 }, { "epoch": 1.1879385100512416, "grad_norm": 0.444244921207428, "learning_rate": 0.0005, "loss": 1.4340810775756836, "step": 15069 }, { "epoch": 1.1880173433188805, "grad_norm": 0.5919808149337769, "learning_rate": 0.0005, "loss": 1.7362335920333862, "step": 15070 }, { "epoch": 1.1880961765865194, "grad_norm": 0.5900271534919739, "learning_rate": 0.0005, "loss": 1.6886091232299805, "step": 15071 }, { "epoch": 1.1881750098541584, "grad_norm": 0.6470315456390381, "learning_rate": 0.0005, "loss": 1.8226242065429688, "step": 15072 }, { "epoch": 1.1882538431217975, "grad_norm": 0.5221322774887085, "learning_rate": 0.0005, "loss": 1.7370092868804932, "step": 15073 }, { "epoch": 1.1883326763894364, "grad_norm": 0.5665398836135864, "learning_rate": 0.0005, "loss": 1.8952029943466187, "step": 15074 }, { "epoch": 1.1884115096570753, "grad_norm": 0.6791867017745972, "learning_rate": 0.0005, "loss": 1.6549307107925415, "step": 15075 }, { "epoch": 1.1884903429247142, "grad_norm": 0.4842994511127472, "learning_rate": 0.0005, "loss": 1.4608306884765625, "step": 15076 }, { "epoch": 1.1885691761923531, "grad_norm": 0.49949324131011963, "learning_rate": 0.0005, "loss": 1.7312196493148804, "step": 15077 }, { "epoch": 1.188648009459992, "grad_norm": 0.5660809874534607, "learning_rate": 0.0005, "loss": 1.8519649505615234, "step": 15078 }, { "epoch": 1.1887268427276312, "grad_norm": 0.5380091667175293, "learning_rate": 0.0005, "loss": 1.5021214485168457, "step": 15079 }, { "epoch": 1.18880567599527, "grad_norm": 0.49159952998161316, "learning_rate": 0.0005, "loss": 1.872867226600647, "step": 15080 }, { "epoch": 1.188884509262909, "grad_norm": 0.4439111053943634, "learning_rate": 0.0005, "loss": 1.5865354537963867, "step": 15081 }, { "epoch": 1.188963342530548, "grad_norm": 0.543453574180603, "learning_rate": 0.0005, "loss": 1.74824857711792, "step": 15082 }, { "epoch": 1.1890421757981868, "grad_norm": 0.6412876844406128, "learning_rate": 0.0005, "loss": 1.8878365755081177, "step": 15083 }, { "epoch": 1.1891210090658257, "grad_norm": 0.5238459706306458, "learning_rate": 0.0005, "loss": 1.7179386615753174, "step": 15084 }, { "epoch": 1.1891998423334647, "grad_norm": 0.45478278398513794, "learning_rate": 0.0005, "loss": 1.4879257678985596, "step": 15085 }, { "epoch": 1.1892786756011038, "grad_norm": 0.48517829179763794, "learning_rate": 0.0005, "loss": 1.7580900192260742, "step": 15086 }, { "epoch": 1.1893575088687427, "grad_norm": 0.4554067850112915, "learning_rate": 0.0005, "loss": 1.689294457435608, "step": 15087 }, { "epoch": 1.1894363421363816, "grad_norm": 0.5834798812866211, "learning_rate": 0.0005, "loss": 1.6316723823547363, "step": 15088 }, { "epoch": 1.1895151754040205, "grad_norm": 0.6897408366203308, "learning_rate": 0.0005, "loss": 1.9167603254318237, "step": 15089 }, { "epoch": 1.1895940086716594, "grad_norm": 0.5220935344696045, "learning_rate": 0.0005, "loss": 1.765803337097168, "step": 15090 }, { "epoch": 1.1896728419392983, "grad_norm": 0.6774111390113831, "learning_rate": 0.0005, "loss": 1.5550137758255005, "step": 15091 }, { "epoch": 1.1897516752069373, "grad_norm": 0.710086464881897, "learning_rate": 0.0005, "loss": 1.6136764287948608, "step": 15092 }, { "epoch": 1.1898305084745764, "grad_norm": 0.7266132235527039, "learning_rate": 0.0005, "loss": 1.6501060724258423, "step": 15093 }, { "epoch": 1.1899093417422153, "grad_norm": 0.67974454164505, "learning_rate": 0.0005, "loss": 1.732811450958252, "step": 15094 }, { "epoch": 1.1899881750098542, "grad_norm": 0.5941579937934875, "learning_rate": 0.0005, "loss": 1.8985002040863037, "step": 15095 }, { "epoch": 1.1900670082774931, "grad_norm": 0.5062357783317566, "learning_rate": 0.0005, "loss": 1.5150877237319946, "step": 15096 }, { "epoch": 1.190145841545132, "grad_norm": 0.45392316579818726, "learning_rate": 0.0005, "loss": 1.5920419692993164, "step": 15097 }, { "epoch": 1.190224674812771, "grad_norm": 0.5131552815437317, "learning_rate": 0.0005, "loss": 1.7145380973815918, "step": 15098 }, { "epoch": 1.1903035080804099, "grad_norm": 0.7247618436813354, "learning_rate": 0.0005, "loss": 1.891869306564331, "step": 15099 }, { "epoch": 1.190382341348049, "grad_norm": 0.48627981543540955, "learning_rate": 0.0005, "loss": 1.5983190536499023, "step": 15100 }, { "epoch": 1.190461174615688, "grad_norm": 0.6163986325263977, "learning_rate": 0.0005, "loss": 1.7575355768203735, "step": 15101 }, { "epoch": 1.1905400078833268, "grad_norm": 0.5888614654541016, "learning_rate": 0.0005, "loss": 1.813715934753418, "step": 15102 }, { "epoch": 1.1906188411509657, "grad_norm": 0.7450884580612183, "learning_rate": 0.0005, "loss": 1.6580325365066528, "step": 15103 }, { "epoch": 1.1906976744186046, "grad_norm": 0.6158527731895447, "learning_rate": 0.0005, "loss": 1.5581562519073486, "step": 15104 }, { "epoch": 1.1907765076862435, "grad_norm": 0.5236246585845947, "learning_rate": 0.0005, "loss": 1.5958573818206787, "step": 15105 }, { "epoch": 1.1908553409538825, "grad_norm": 0.5339880585670471, "learning_rate": 0.0005, "loss": 1.7648802995681763, "step": 15106 }, { "epoch": 1.1909341742215216, "grad_norm": 0.4306437075138092, "learning_rate": 0.0005, "loss": 1.407364010810852, "step": 15107 }, { "epoch": 1.1910130074891605, "grad_norm": 0.5090740323066711, "learning_rate": 0.0005, "loss": 1.7642977237701416, "step": 15108 }, { "epoch": 1.1910918407567994, "grad_norm": 0.5810378193855286, "learning_rate": 0.0005, "loss": 1.7641940116882324, "step": 15109 }, { "epoch": 1.1911706740244383, "grad_norm": 0.5452317595481873, "learning_rate": 0.0005, "loss": 1.5564945936203003, "step": 15110 }, { "epoch": 1.1912495072920772, "grad_norm": 0.5088919997215271, "learning_rate": 0.0005, "loss": 1.7883769273757935, "step": 15111 }, { "epoch": 1.1913283405597161, "grad_norm": 0.5026547312736511, "learning_rate": 0.0005, "loss": 1.698697805404663, "step": 15112 }, { "epoch": 1.191407173827355, "grad_norm": 0.44624072313308716, "learning_rate": 0.0005, "loss": 1.5454132556915283, "step": 15113 }, { "epoch": 1.1914860070949942, "grad_norm": 0.5195713639259338, "learning_rate": 0.0005, "loss": 1.647562026977539, "step": 15114 }, { "epoch": 1.191564840362633, "grad_norm": 0.539138913154602, "learning_rate": 0.0005, "loss": 1.854417085647583, "step": 15115 }, { "epoch": 1.191643673630272, "grad_norm": 0.4857242703437805, "learning_rate": 0.0005, "loss": 1.754119634628296, "step": 15116 }, { "epoch": 1.191722506897911, "grad_norm": 0.535013735294342, "learning_rate": 0.0005, "loss": 1.572321891784668, "step": 15117 }, { "epoch": 1.1918013401655498, "grad_norm": 0.4790973961353302, "learning_rate": 0.0005, "loss": 1.5875158309936523, "step": 15118 }, { "epoch": 1.1918801734331887, "grad_norm": 0.6052520275115967, "learning_rate": 0.0005, "loss": 1.7832584381103516, "step": 15119 }, { "epoch": 1.1919590067008277, "grad_norm": 0.511671781539917, "learning_rate": 0.0005, "loss": 1.561112880706787, "step": 15120 }, { "epoch": 1.1920378399684668, "grad_norm": 0.7710080146789551, "learning_rate": 0.0005, "loss": 2.044147491455078, "step": 15121 }, { "epoch": 1.1921166732361057, "grad_norm": 0.5559218525886536, "learning_rate": 0.0005, "loss": 1.6094099283218384, "step": 15122 }, { "epoch": 1.1921955065037446, "grad_norm": 0.5970556139945984, "learning_rate": 0.0005, "loss": 1.505329966545105, "step": 15123 }, { "epoch": 1.1922743397713835, "grad_norm": 0.5508680939674377, "learning_rate": 0.0005, "loss": 1.70583176612854, "step": 15124 }, { "epoch": 1.1923531730390224, "grad_norm": 0.6284942626953125, "learning_rate": 0.0005, "loss": 1.5662652254104614, "step": 15125 }, { "epoch": 1.1924320063066614, "grad_norm": 0.5883733034133911, "learning_rate": 0.0005, "loss": 1.7769393920898438, "step": 15126 }, { "epoch": 1.1925108395743003, "grad_norm": 0.5115776658058167, "learning_rate": 0.0005, "loss": 1.670569658279419, "step": 15127 }, { "epoch": 1.1925896728419394, "grad_norm": 0.5212023854255676, "learning_rate": 0.0005, "loss": 1.7919492721557617, "step": 15128 }, { "epoch": 1.1926685061095783, "grad_norm": 0.4779689610004425, "learning_rate": 0.0005, "loss": 1.5506792068481445, "step": 15129 }, { "epoch": 1.1927473393772172, "grad_norm": 0.5182217955589294, "learning_rate": 0.0005, "loss": 1.4580622911453247, "step": 15130 }, { "epoch": 1.1928261726448561, "grad_norm": 0.5226972699165344, "learning_rate": 0.0005, "loss": 1.715412974357605, "step": 15131 }, { "epoch": 1.192905005912495, "grad_norm": 0.5312500596046448, "learning_rate": 0.0005, "loss": 1.886388897895813, "step": 15132 }, { "epoch": 1.192983839180134, "grad_norm": 0.47275859117507935, "learning_rate": 0.0005, "loss": 1.6996458768844604, "step": 15133 }, { "epoch": 1.1930626724477729, "grad_norm": 0.5760577917098999, "learning_rate": 0.0005, "loss": 1.7381739616394043, "step": 15134 }, { "epoch": 1.193141505715412, "grad_norm": 0.564145565032959, "learning_rate": 0.0005, "loss": 1.6761113405227661, "step": 15135 }, { "epoch": 1.193220338983051, "grad_norm": 0.544167160987854, "learning_rate": 0.0005, "loss": 1.5823991298675537, "step": 15136 }, { "epoch": 1.1932991722506898, "grad_norm": 0.5800890922546387, "learning_rate": 0.0005, "loss": 1.7304041385650635, "step": 15137 }, { "epoch": 1.1933780055183287, "grad_norm": 0.5986903309822083, "learning_rate": 0.0005, "loss": 1.8772296905517578, "step": 15138 }, { "epoch": 1.1934568387859676, "grad_norm": 0.559023380279541, "learning_rate": 0.0005, "loss": 1.5543100833892822, "step": 15139 }, { "epoch": 1.1935356720536066, "grad_norm": 0.531122088432312, "learning_rate": 0.0005, "loss": 1.9200170040130615, "step": 15140 }, { "epoch": 1.1936145053212455, "grad_norm": 0.5438970923423767, "learning_rate": 0.0005, "loss": 1.7159521579742432, "step": 15141 }, { "epoch": 1.1936933385888846, "grad_norm": 0.5858630537986755, "learning_rate": 0.0005, "loss": 1.8047168254852295, "step": 15142 }, { "epoch": 1.1937721718565235, "grad_norm": 0.4937656819820404, "learning_rate": 0.0005, "loss": 1.604087233543396, "step": 15143 }, { "epoch": 1.1938510051241624, "grad_norm": 0.41084572672843933, "learning_rate": 0.0005, "loss": 1.5748934745788574, "step": 15144 }, { "epoch": 1.1939298383918013, "grad_norm": 0.5174111127853394, "learning_rate": 0.0005, "loss": 1.5274039506912231, "step": 15145 }, { "epoch": 1.1940086716594402, "grad_norm": 0.5015656352043152, "learning_rate": 0.0005, "loss": 1.7618348598480225, "step": 15146 }, { "epoch": 1.1940875049270792, "grad_norm": 0.5190155506134033, "learning_rate": 0.0005, "loss": 1.6492908000946045, "step": 15147 }, { "epoch": 1.194166338194718, "grad_norm": 0.573250949382782, "learning_rate": 0.0005, "loss": 1.5829484462738037, "step": 15148 }, { "epoch": 1.1942451714623572, "grad_norm": 0.4657633900642395, "learning_rate": 0.0005, "loss": 1.616637945175171, "step": 15149 }, { "epoch": 1.1943240047299961, "grad_norm": 0.5096322894096375, "learning_rate": 0.0005, "loss": 1.9159541130065918, "step": 15150 }, { "epoch": 1.194402837997635, "grad_norm": 0.5036360025405884, "learning_rate": 0.0005, "loss": 1.8205162286758423, "step": 15151 }, { "epoch": 1.194481671265274, "grad_norm": 0.7144911885261536, "learning_rate": 0.0005, "loss": 1.5904301404953003, "step": 15152 }, { "epoch": 1.1945605045329128, "grad_norm": 0.5092399716377258, "learning_rate": 0.0005, "loss": 1.5866317749023438, "step": 15153 }, { "epoch": 1.1946393378005518, "grad_norm": 0.5569769144058228, "learning_rate": 0.0005, "loss": 1.5447297096252441, "step": 15154 }, { "epoch": 1.1947181710681907, "grad_norm": 0.6764218211174011, "learning_rate": 0.0005, "loss": 1.9740182161331177, "step": 15155 }, { "epoch": 1.1947970043358298, "grad_norm": 0.4989326000213623, "learning_rate": 0.0005, "loss": 1.638939380645752, "step": 15156 }, { "epoch": 1.1948758376034687, "grad_norm": 0.6120710968971252, "learning_rate": 0.0005, "loss": 1.7619473934173584, "step": 15157 }, { "epoch": 1.1949546708711076, "grad_norm": 0.49991634488105774, "learning_rate": 0.0005, "loss": 1.6833665370941162, "step": 15158 }, { "epoch": 1.1950335041387465, "grad_norm": 0.5560547113418579, "learning_rate": 0.0005, "loss": 1.5952868461608887, "step": 15159 }, { "epoch": 1.1951123374063854, "grad_norm": 0.6633109450340271, "learning_rate": 0.0005, "loss": 1.963578462600708, "step": 15160 }, { "epoch": 1.1951911706740244, "grad_norm": 0.4310299754142761, "learning_rate": 0.0005, "loss": 1.5841442346572876, "step": 15161 }, { "epoch": 1.1952700039416633, "grad_norm": 0.4819820523262024, "learning_rate": 0.0005, "loss": 1.6047570705413818, "step": 15162 }, { "epoch": 1.1953488372093024, "grad_norm": 0.5482916831970215, "learning_rate": 0.0005, "loss": 1.610577940940857, "step": 15163 }, { "epoch": 1.1954276704769413, "grad_norm": 0.6596310138702393, "learning_rate": 0.0005, "loss": 1.7023684978485107, "step": 15164 }, { "epoch": 1.1955065037445802, "grad_norm": 0.5039756894111633, "learning_rate": 0.0005, "loss": 1.7082432508468628, "step": 15165 }, { "epoch": 1.1955853370122191, "grad_norm": 0.5479620099067688, "learning_rate": 0.0005, "loss": 1.921120285987854, "step": 15166 }, { "epoch": 1.195664170279858, "grad_norm": 0.6028782725334167, "learning_rate": 0.0005, "loss": 1.8048069477081299, "step": 15167 }, { "epoch": 1.195743003547497, "grad_norm": 0.4473978579044342, "learning_rate": 0.0005, "loss": 1.658847451210022, "step": 15168 }, { "epoch": 1.1958218368151359, "grad_norm": 0.578493058681488, "learning_rate": 0.0005, "loss": 1.5368072986602783, "step": 15169 }, { "epoch": 1.195900670082775, "grad_norm": 0.5015849471092224, "learning_rate": 0.0005, "loss": 1.7513319253921509, "step": 15170 }, { "epoch": 1.195979503350414, "grad_norm": 0.516312301158905, "learning_rate": 0.0005, "loss": 1.4643909931182861, "step": 15171 }, { "epoch": 1.1960583366180528, "grad_norm": 0.5122430920600891, "learning_rate": 0.0005, "loss": 1.5138155221939087, "step": 15172 }, { "epoch": 1.1961371698856917, "grad_norm": 0.4266965985298157, "learning_rate": 0.0005, "loss": 1.382295846939087, "step": 15173 }, { "epoch": 1.1962160031533307, "grad_norm": 0.5606334209442139, "learning_rate": 0.0005, "loss": 1.7289581298828125, "step": 15174 }, { "epoch": 1.1962948364209696, "grad_norm": 0.5726836323738098, "learning_rate": 0.0005, "loss": 1.9841359853744507, "step": 15175 }, { "epoch": 1.1963736696886085, "grad_norm": 0.503987193107605, "learning_rate": 0.0005, "loss": 1.5857045650482178, "step": 15176 }, { "epoch": 1.1964525029562476, "grad_norm": 0.4368279278278351, "learning_rate": 0.0005, "loss": 1.621433138847351, "step": 15177 }, { "epoch": 1.1965313362238865, "grad_norm": 0.46974480152130127, "learning_rate": 0.0005, "loss": 1.6172233819961548, "step": 15178 }, { "epoch": 1.1966101694915254, "grad_norm": 0.7596850991249084, "learning_rate": 0.0005, "loss": 1.5085136890411377, "step": 15179 }, { "epoch": 1.1966890027591643, "grad_norm": 0.4269702434539795, "learning_rate": 0.0005, "loss": 1.5901274681091309, "step": 15180 }, { "epoch": 1.1967678360268033, "grad_norm": 0.6256317496299744, "learning_rate": 0.0005, "loss": 1.7999945878982544, "step": 15181 }, { "epoch": 1.1968466692944422, "grad_norm": 0.576167643070221, "learning_rate": 0.0005, "loss": 1.671827793121338, "step": 15182 }, { "epoch": 1.196925502562081, "grad_norm": 0.715121865272522, "learning_rate": 0.0005, "loss": 1.8064777851104736, "step": 15183 }, { "epoch": 1.1970043358297202, "grad_norm": 0.6141430735588074, "learning_rate": 0.0005, "loss": 1.6748156547546387, "step": 15184 }, { "epoch": 1.1970831690973591, "grad_norm": 0.5916896462440491, "learning_rate": 0.0005, "loss": 1.6256365776062012, "step": 15185 }, { "epoch": 1.197162002364998, "grad_norm": 0.5947795510292053, "learning_rate": 0.0005, "loss": 1.8284624814987183, "step": 15186 }, { "epoch": 1.197240835632637, "grad_norm": 0.5703900456428528, "learning_rate": 0.0005, "loss": 1.6639404296875, "step": 15187 }, { "epoch": 1.1973196689002759, "grad_norm": 0.5720299482345581, "learning_rate": 0.0005, "loss": 1.8461285829544067, "step": 15188 }, { "epoch": 1.1973985021679148, "grad_norm": 0.5867312550544739, "learning_rate": 0.0005, "loss": 1.6798467636108398, "step": 15189 }, { "epoch": 1.1974773354355537, "grad_norm": 0.5216468572616577, "learning_rate": 0.0005, "loss": 1.9057543277740479, "step": 15190 }, { "epoch": 1.1975561687031928, "grad_norm": 0.47777050733566284, "learning_rate": 0.0005, "loss": 1.5911870002746582, "step": 15191 }, { "epoch": 1.1976350019708317, "grad_norm": 0.5505812764167786, "learning_rate": 0.0005, "loss": 1.8973989486694336, "step": 15192 }, { "epoch": 1.1977138352384706, "grad_norm": 0.7034669518470764, "learning_rate": 0.0005, "loss": 1.8897902965545654, "step": 15193 }, { "epoch": 1.1977926685061095, "grad_norm": 0.45915487408638, "learning_rate": 0.0005, "loss": 1.4969407320022583, "step": 15194 }, { "epoch": 1.1978715017737485, "grad_norm": 0.6080670356750488, "learning_rate": 0.0005, "loss": 1.9230836629867554, "step": 15195 }, { "epoch": 1.1979503350413874, "grad_norm": 0.5777031779289246, "learning_rate": 0.0005, "loss": 1.796811580657959, "step": 15196 }, { "epoch": 1.1980291683090265, "grad_norm": 0.4896765649318695, "learning_rate": 0.0005, "loss": 1.7451057434082031, "step": 15197 }, { "epoch": 1.1981080015766654, "grad_norm": 0.5518513321876526, "learning_rate": 0.0005, "loss": 1.6629035472869873, "step": 15198 }, { "epoch": 1.1981868348443043, "grad_norm": 0.47603610157966614, "learning_rate": 0.0005, "loss": 1.5882691144943237, "step": 15199 }, { "epoch": 1.1982656681119432, "grad_norm": 0.5155795812606812, "learning_rate": 0.0005, "loss": 1.7014273405075073, "step": 15200 }, { "epoch": 1.1983445013795821, "grad_norm": 0.581057608127594, "learning_rate": 0.0005, "loss": 1.6351206302642822, "step": 15201 }, { "epoch": 1.198423334647221, "grad_norm": 0.47850921750068665, "learning_rate": 0.0005, "loss": 1.4710407257080078, "step": 15202 }, { "epoch": 1.1985021679148602, "grad_norm": 0.6139301657676697, "learning_rate": 0.0005, "loss": 1.882239580154419, "step": 15203 }, { "epoch": 1.198581001182499, "grad_norm": 0.5611553192138672, "learning_rate": 0.0005, "loss": 1.6774771213531494, "step": 15204 }, { "epoch": 1.198659834450138, "grad_norm": 0.5111081004142761, "learning_rate": 0.0005, "loss": 1.6386834383010864, "step": 15205 }, { "epoch": 1.198738667717777, "grad_norm": 0.547588586807251, "learning_rate": 0.0005, "loss": 1.8409842252731323, "step": 15206 }, { "epoch": 1.1988175009854158, "grad_norm": 0.46864911913871765, "learning_rate": 0.0005, "loss": 1.527409553527832, "step": 15207 }, { "epoch": 1.1988963342530548, "grad_norm": 0.8202710151672363, "learning_rate": 0.0005, "loss": 1.709385633468628, "step": 15208 }, { "epoch": 1.1989751675206937, "grad_norm": 0.4524272084236145, "learning_rate": 0.0005, "loss": 1.5705922842025757, "step": 15209 }, { "epoch": 1.1990540007883328, "grad_norm": 0.44927671551704407, "learning_rate": 0.0005, "loss": 1.5120664834976196, "step": 15210 }, { "epoch": 1.1991328340559717, "grad_norm": 0.5929425954818726, "learning_rate": 0.0005, "loss": 1.6770586967468262, "step": 15211 }, { "epoch": 1.1992116673236106, "grad_norm": 0.5846718549728394, "learning_rate": 0.0005, "loss": 1.716797113418579, "step": 15212 }, { "epoch": 1.1992905005912495, "grad_norm": 0.5951944589614868, "learning_rate": 0.0005, "loss": 1.7034633159637451, "step": 15213 }, { "epoch": 1.1993693338588884, "grad_norm": 0.5379964113235474, "learning_rate": 0.0005, "loss": 1.5850999355316162, "step": 15214 }, { "epoch": 1.1994481671265274, "grad_norm": 0.5430234670639038, "learning_rate": 0.0005, "loss": 1.499298334121704, "step": 15215 }, { "epoch": 1.1995270003941663, "grad_norm": 0.5503233671188354, "learning_rate": 0.0005, "loss": 1.555996060371399, "step": 15216 }, { "epoch": 1.1996058336618054, "grad_norm": 0.7128013372421265, "learning_rate": 0.0005, "loss": 1.6498687267303467, "step": 15217 }, { "epoch": 1.1996846669294443, "grad_norm": 0.6538295149803162, "learning_rate": 0.0005, "loss": 1.7469208240509033, "step": 15218 }, { "epoch": 1.1997635001970832, "grad_norm": 0.5080063939094543, "learning_rate": 0.0005, "loss": 1.6243832111358643, "step": 15219 }, { "epoch": 1.1998423334647221, "grad_norm": 0.5357194542884827, "learning_rate": 0.0005, "loss": 1.6812392473220825, "step": 15220 }, { "epoch": 1.199921166732361, "grad_norm": 1.928136944770813, "learning_rate": 0.0005, "loss": 1.6782739162445068, "step": 15221 }, { "epoch": 1.2, "grad_norm": 0.5857486128807068, "learning_rate": 0.0005, "loss": 1.8506321907043457, "step": 15222 }, { "epoch": 1.2000788332676389, "grad_norm": 0.8118435144424438, "learning_rate": 0.0005, "loss": 1.7100775241851807, "step": 15223 }, { "epoch": 1.200157666535278, "grad_norm": 0.6216061115264893, "learning_rate": 0.0005, "loss": 1.7328290939331055, "step": 15224 }, { "epoch": 1.200236499802917, "grad_norm": 0.6584612727165222, "learning_rate": 0.0005, "loss": 1.703381061553955, "step": 15225 }, { "epoch": 1.2003153330705558, "grad_norm": 0.8216317296028137, "learning_rate": 0.0005, "loss": 1.9353573322296143, "step": 15226 }, { "epoch": 1.2003941663381947, "grad_norm": 0.6764179468154907, "learning_rate": 0.0005, "loss": 1.7482893466949463, "step": 15227 }, { "epoch": 1.2004729996058336, "grad_norm": 0.606603741645813, "learning_rate": 0.0005, "loss": 1.758500337600708, "step": 15228 }, { "epoch": 1.2005518328734726, "grad_norm": 0.5915757417678833, "learning_rate": 0.0005, "loss": 1.729130506515503, "step": 15229 }, { "epoch": 1.2006306661411115, "grad_norm": 0.6889644265174866, "learning_rate": 0.0005, "loss": 1.9877912998199463, "step": 15230 }, { "epoch": 1.2007094994087506, "grad_norm": 0.5856956243515015, "learning_rate": 0.0005, "loss": 1.6541129350662231, "step": 15231 }, { "epoch": 1.2007883326763895, "grad_norm": 0.6736074686050415, "learning_rate": 0.0005, "loss": 1.9433315992355347, "step": 15232 }, { "epoch": 1.2008671659440284, "grad_norm": 0.587721586227417, "learning_rate": 0.0005, "loss": 2.0334668159484863, "step": 15233 }, { "epoch": 1.2009459992116673, "grad_norm": 0.6735603213310242, "learning_rate": 0.0005, "loss": 1.6688785552978516, "step": 15234 }, { "epoch": 1.2010248324793062, "grad_norm": 0.5746515393257141, "learning_rate": 0.0005, "loss": 1.7738069295883179, "step": 15235 }, { "epoch": 1.2011036657469452, "grad_norm": 0.6512685418128967, "learning_rate": 0.0005, "loss": 1.8469765186309814, "step": 15236 }, { "epoch": 1.201182499014584, "grad_norm": 0.7196995615959167, "learning_rate": 0.0005, "loss": 1.75456702709198, "step": 15237 }, { "epoch": 1.2012613322822232, "grad_norm": 0.4658198058605194, "learning_rate": 0.0005, "loss": 1.567425012588501, "step": 15238 }, { "epoch": 1.2013401655498621, "grad_norm": 0.40903252363204956, "learning_rate": 0.0005, "loss": 1.4680026769638062, "step": 15239 }, { "epoch": 1.201418998817501, "grad_norm": 0.6440495848655701, "learning_rate": 0.0005, "loss": 1.791093111038208, "step": 15240 }, { "epoch": 1.20149783208514, "grad_norm": 0.6134178042411804, "learning_rate": 0.0005, "loss": 1.8633722066879272, "step": 15241 }, { "epoch": 1.2015766653527788, "grad_norm": 0.4918627440929413, "learning_rate": 0.0005, "loss": 1.5765957832336426, "step": 15242 }, { "epoch": 1.2016554986204178, "grad_norm": 0.4916958808898926, "learning_rate": 0.0005, "loss": 1.670360803604126, "step": 15243 }, { "epoch": 1.2017343318880567, "grad_norm": 0.6120978593826294, "learning_rate": 0.0005, "loss": 1.4926748275756836, "step": 15244 }, { "epoch": 1.2018131651556958, "grad_norm": 0.5991610884666443, "learning_rate": 0.0005, "loss": 1.8407137393951416, "step": 15245 }, { "epoch": 1.2018919984233347, "grad_norm": 0.5161502361297607, "learning_rate": 0.0005, "loss": 1.5915104150772095, "step": 15246 }, { "epoch": 1.2019708316909736, "grad_norm": 0.5038031935691833, "learning_rate": 0.0005, "loss": 1.6483622789382935, "step": 15247 }, { "epoch": 1.2020496649586125, "grad_norm": 0.6035229563713074, "learning_rate": 0.0005, "loss": 1.5879709720611572, "step": 15248 }, { "epoch": 1.2021284982262515, "grad_norm": 0.4574185907840729, "learning_rate": 0.0005, "loss": 1.601757526397705, "step": 15249 }, { "epoch": 1.2022073314938904, "grad_norm": 0.6455079913139343, "learning_rate": 0.0005, "loss": 1.783783197402954, "step": 15250 }, { "epoch": 1.2022861647615293, "grad_norm": 0.5240138173103333, "learning_rate": 0.0005, "loss": 1.6876839399337769, "step": 15251 }, { "epoch": 1.2023649980291684, "grad_norm": 0.4774777293205261, "learning_rate": 0.0005, "loss": 1.5640017986297607, "step": 15252 }, { "epoch": 1.2024438312968073, "grad_norm": 0.669806182384491, "learning_rate": 0.0005, "loss": 1.771620750427246, "step": 15253 }, { "epoch": 1.2025226645644462, "grad_norm": 0.5460696220397949, "learning_rate": 0.0005, "loss": 1.463376760482788, "step": 15254 }, { "epoch": 1.2026014978320851, "grad_norm": 0.483695924282074, "learning_rate": 0.0005, "loss": 1.615607500076294, "step": 15255 }, { "epoch": 1.202680331099724, "grad_norm": 0.41977035999298096, "learning_rate": 0.0005, "loss": 1.609989881515503, "step": 15256 }, { "epoch": 1.202759164367363, "grad_norm": 0.5168651342391968, "learning_rate": 0.0005, "loss": 1.7986066341400146, "step": 15257 }, { "epoch": 1.2028379976350019, "grad_norm": 0.7346487045288086, "learning_rate": 0.0005, "loss": 1.8233147859573364, "step": 15258 }, { "epoch": 1.202916830902641, "grad_norm": 0.45772793889045715, "learning_rate": 0.0005, "loss": 1.6791986227035522, "step": 15259 }, { "epoch": 1.20299566417028, "grad_norm": 0.5044553279876709, "learning_rate": 0.0005, "loss": 1.420743703842163, "step": 15260 }, { "epoch": 1.2030744974379188, "grad_norm": 0.7628927230834961, "learning_rate": 0.0005, "loss": 1.7106764316558838, "step": 15261 }, { "epoch": 1.2031533307055577, "grad_norm": 0.6068065762519836, "learning_rate": 0.0005, "loss": 1.4811468124389648, "step": 15262 }, { "epoch": 1.2032321639731967, "grad_norm": 0.6061273813247681, "learning_rate": 0.0005, "loss": 1.6129292249679565, "step": 15263 }, { "epoch": 1.2033109972408356, "grad_norm": 0.6262735724449158, "learning_rate": 0.0005, "loss": 1.6951417922973633, "step": 15264 }, { "epoch": 1.2033898305084745, "grad_norm": 0.6015650629997253, "learning_rate": 0.0005, "loss": 1.8190257549285889, "step": 15265 }, { "epoch": 1.2034686637761136, "grad_norm": 0.6200754642486572, "learning_rate": 0.0005, "loss": 1.625292420387268, "step": 15266 }, { "epoch": 1.2035474970437525, "grad_norm": 0.5887545943260193, "learning_rate": 0.0005, "loss": 1.6856344938278198, "step": 15267 }, { "epoch": 1.2036263303113914, "grad_norm": 0.7617027163505554, "learning_rate": 0.0005, "loss": 1.7014503479003906, "step": 15268 }, { "epoch": 1.2037051635790303, "grad_norm": 0.49392062425613403, "learning_rate": 0.0005, "loss": 1.5410284996032715, "step": 15269 }, { "epoch": 1.2037839968466693, "grad_norm": 0.5172895789146423, "learning_rate": 0.0005, "loss": 1.5825940370559692, "step": 15270 }, { "epoch": 1.2038628301143082, "grad_norm": 0.4476257264614105, "learning_rate": 0.0005, "loss": 1.6189124584197998, "step": 15271 }, { "epoch": 1.203941663381947, "grad_norm": 0.45799264311790466, "learning_rate": 0.0005, "loss": 1.6641350984573364, "step": 15272 }, { "epoch": 1.2040204966495862, "grad_norm": 0.5310491919517517, "learning_rate": 0.0005, "loss": 1.690891981124878, "step": 15273 }, { "epoch": 1.2040993299172251, "grad_norm": 0.5992605090141296, "learning_rate": 0.0005, "loss": 1.7099794149398804, "step": 15274 }, { "epoch": 1.204178163184864, "grad_norm": 0.5686150193214417, "learning_rate": 0.0005, "loss": 1.7189226150512695, "step": 15275 }, { "epoch": 1.204256996452503, "grad_norm": 0.5682621002197266, "learning_rate": 0.0005, "loss": 1.7165879011154175, "step": 15276 }, { "epoch": 1.2043358297201419, "grad_norm": 0.6447309851646423, "learning_rate": 0.0005, "loss": 1.8208611011505127, "step": 15277 }, { "epoch": 1.2044146629877808, "grad_norm": 1.1975148916244507, "learning_rate": 0.0005, "loss": 1.9414695501327515, "step": 15278 }, { "epoch": 1.2044934962554197, "grad_norm": 0.5476824045181274, "learning_rate": 0.0005, "loss": 1.6588563919067383, "step": 15279 }, { "epoch": 1.2045723295230588, "grad_norm": 0.5049151182174683, "learning_rate": 0.0005, "loss": 1.4741636514663696, "step": 15280 }, { "epoch": 1.2046511627906977, "grad_norm": 0.4931791126728058, "learning_rate": 0.0005, "loss": 1.734095811843872, "step": 15281 }, { "epoch": 1.2047299960583366, "grad_norm": 0.5401196479797363, "learning_rate": 0.0005, "loss": 1.4744443893432617, "step": 15282 }, { "epoch": 1.2048088293259755, "grad_norm": 0.4984266459941864, "learning_rate": 0.0005, "loss": 1.6345126628875732, "step": 15283 }, { "epoch": 1.2048876625936145, "grad_norm": 0.6270520091056824, "learning_rate": 0.0005, "loss": 1.7194130420684814, "step": 15284 }, { "epoch": 1.2049664958612534, "grad_norm": 0.8722665309906006, "learning_rate": 0.0005, "loss": 1.5077908039093018, "step": 15285 }, { "epoch": 1.2050453291288923, "grad_norm": 0.6475991606712341, "learning_rate": 0.0005, "loss": 1.775547742843628, "step": 15286 }, { "epoch": 1.2051241623965314, "grad_norm": 0.550436794757843, "learning_rate": 0.0005, "loss": 1.7092777490615845, "step": 15287 }, { "epoch": 1.2052029956641703, "grad_norm": 1.3686106204986572, "learning_rate": 0.0005, "loss": 1.7152223587036133, "step": 15288 }, { "epoch": 1.2052818289318092, "grad_norm": 0.46579721570014954, "learning_rate": 0.0005, "loss": 1.489506483078003, "step": 15289 }, { "epoch": 1.2053606621994482, "grad_norm": 0.5063746571540833, "learning_rate": 0.0005, "loss": 1.496906042098999, "step": 15290 }, { "epoch": 1.205439495467087, "grad_norm": 0.5703736543655396, "learning_rate": 0.0005, "loss": 1.915440559387207, "step": 15291 }, { "epoch": 1.205518328734726, "grad_norm": 0.5212112665176392, "learning_rate": 0.0005, "loss": 1.79888916015625, "step": 15292 }, { "epoch": 1.2055971620023649, "grad_norm": 0.6020947098731995, "learning_rate": 0.0005, "loss": 1.977226972579956, "step": 15293 }, { "epoch": 1.205675995270004, "grad_norm": 0.732244610786438, "learning_rate": 0.0005, "loss": 1.5803349018096924, "step": 15294 }, { "epoch": 1.205754828537643, "grad_norm": 0.5278403759002686, "learning_rate": 0.0005, "loss": 1.746299386024475, "step": 15295 }, { "epoch": 1.2058336618052818, "grad_norm": 0.5315293669700623, "learning_rate": 0.0005, "loss": 1.591713309288025, "step": 15296 }, { "epoch": 1.2059124950729208, "grad_norm": 0.5343469977378845, "learning_rate": 0.0005, "loss": 1.7379686832427979, "step": 15297 }, { "epoch": 1.2059913283405597, "grad_norm": 0.5857917070388794, "learning_rate": 0.0005, "loss": 1.8420099020004272, "step": 15298 }, { "epoch": 1.2060701616081986, "grad_norm": 0.5033384561538696, "learning_rate": 0.0005, "loss": 1.443197250366211, "step": 15299 }, { "epoch": 1.2061489948758375, "grad_norm": 0.4962122440338135, "learning_rate": 0.0005, "loss": 1.6592681407928467, "step": 15300 }, { "epoch": 1.2062278281434766, "grad_norm": 0.6123422384262085, "learning_rate": 0.0005, "loss": 1.828629732131958, "step": 15301 }, { "epoch": 1.2063066614111155, "grad_norm": 0.5835451483726501, "learning_rate": 0.0005, "loss": 1.5859975814819336, "step": 15302 }, { "epoch": 1.2063854946787544, "grad_norm": 0.639141857624054, "learning_rate": 0.0005, "loss": 1.8053197860717773, "step": 15303 }, { "epoch": 1.2064643279463934, "grad_norm": 0.695186972618103, "learning_rate": 0.0005, "loss": 1.6488533020019531, "step": 15304 }, { "epoch": 1.2065431612140323, "grad_norm": 0.4961484968662262, "learning_rate": 0.0005, "loss": 1.6167683601379395, "step": 15305 }, { "epoch": 1.2066219944816712, "grad_norm": 0.47236427664756775, "learning_rate": 0.0005, "loss": 1.5787602663040161, "step": 15306 }, { "epoch": 1.20670082774931, "grad_norm": 0.4541153907775879, "learning_rate": 0.0005, "loss": 1.7446646690368652, "step": 15307 }, { "epoch": 1.2067796610169492, "grad_norm": 0.5564960837364197, "learning_rate": 0.0005, "loss": 1.3736708164215088, "step": 15308 }, { "epoch": 1.2068584942845881, "grad_norm": 0.5038495063781738, "learning_rate": 0.0005, "loss": 1.701087236404419, "step": 15309 }, { "epoch": 1.206937327552227, "grad_norm": 0.5415472388267517, "learning_rate": 0.0005, "loss": 1.716801643371582, "step": 15310 }, { "epoch": 1.207016160819866, "grad_norm": 0.48757204413414, "learning_rate": 0.0005, "loss": 1.7256743907928467, "step": 15311 }, { "epoch": 1.2070949940875049, "grad_norm": 0.5754101276397705, "learning_rate": 0.0005, "loss": 1.6745225191116333, "step": 15312 }, { "epoch": 1.2071738273551438, "grad_norm": 0.5277220010757446, "learning_rate": 0.0005, "loss": 1.694301962852478, "step": 15313 }, { "epoch": 1.2072526606227827, "grad_norm": 0.5066428184509277, "learning_rate": 0.0005, "loss": 1.5375691652297974, "step": 15314 }, { "epoch": 1.2073314938904218, "grad_norm": 0.6408808827400208, "learning_rate": 0.0005, "loss": 1.9433810710906982, "step": 15315 }, { "epoch": 1.2074103271580607, "grad_norm": 0.6923020482063293, "learning_rate": 0.0005, "loss": 1.7898492813110352, "step": 15316 }, { "epoch": 1.2074891604256996, "grad_norm": 0.9119395017623901, "learning_rate": 0.0005, "loss": 1.743180513381958, "step": 15317 }, { "epoch": 1.2075679936933386, "grad_norm": 0.48751580715179443, "learning_rate": 0.0005, "loss": 1.6584287881851196, "step": 15318 }, { "epoch": 1.2076468269609775, "grad_norm": 0.5149477124214172, "learning_rate": 0.0005, "loss": 1.7122573852539062, "step": 15319 }, { "epoch": 1.2077256602286164, "grad_norm": 0.608051598072052, "learning_rate": 0.0005, "loss": 1.913581371307373, "step": 15320 }, { "epoch": 1.2078044934962555, "grad_norm": 0.5828972458839417, "learning_rate": 0.0005, "loss": 1.9952386617660522, "step": 15321 }, { "epoch": 1.2078833267638944, "grad_norm": 0.5210216641426086, "learning_rate": 0.0005, "loss": 1.659365177154541, "step": 15322 }, { "epoch": 1.2079621600315333, "grad_norm": 0.6359607577323914, "learning_rate": 0.0005, "loss": 1.6767873764038086, "step": 15323 }, { "epoch": 1.2080409932991722, "grad_norm": 0.5296566486358643, "learning_rate": 0.0005, "loss": 1.9186365604400635, "step": 15324 }, { "epoch": 1.2081198265668112, "grad_norm": 0.7210936546325684, "learning_rate": 0.0005, "loss": 1.7603108882904053, "step": 15325 }, { "epoch": 1.20819865983445, "grad_norm": 0.5358119606971741, "learning_rate": 0.0005, "loss": 1.8366129398345947, "step": 15326 }, { "epoch": 1.2082774931020892, "grad_norm": 0.6146848201751709, "learning_rate": 0.0005, "loss": 1.8664460182189941, "step": 15327 }, { "epoch": 1.2083563263697281, "grad_norm": 0.6809008717536926, "learning_rate": 0.0005, "loss": 1.6006985902786255, "step": 15328 }, { "epoch": 1.208435159637367, "grad_norm": 0.555163562297821, "learning_rate": 0.0005, "loss": 1.6678943634033203, "step": 15329 }, { "epoch": 1.208513992905006, "grad_norm": 0.6626529097557068, "learning_rate": 0.0005, "loss": 1.806680679321289, "step": 15330 }, { "epoch": 1.2085928261726449, "grad_norm": 0.5442817211151123, "learning_rate": 0.0005, "loss": 1.5947028398513794, "step": 15331 }, { "epoch": 1.2086716594402838, "grad_norm": 0.4985570013523102, "learning_rate": 0.0005, "loss": 1.5966131687164307, "step": 15332 }, { "epoch": 1.2087504927079227, "grad_norm": 0.5004648566246033, "learning_rate": 0.0005, "loss": 1.6386679410934448, "step": 15333 }, { "epoch": 1.2088293259755618, "grad_norm": 0.8386752605438232, "learning_rate": 0.0005, "loss": 1.5399409532546997, "step": 15334 }, { "epoch": 1.2089081592432007, "grad_norm": 0.619915783405304, "learning_rate": 0.0005, "loss": 1.67742919921875, "step": 15335 }, { "epoch": 1.2089869925108396, "grad_norm": 0.5449030995368958, "learning_rate": 0.0005, "loss": 1.6338896751403809, "step": 15336 }, { "epoch": 1.2090658257784785, "grad_norm": 0.44279834628105164, "learning_rate": 0.0005, "loss": 1.3420069217681885, "step": 15337 }, { "epoch": 1.2091446590461175, "grad_norm": 0.5392820835113525, "learning_rate": 0.0005, "loss": 1.7851814031600952, "step": 15338 }, { "epoch": 1.2092234923137564, "grad_norm": 0.653617799282074, "learning_rate": 0.0005, "loss": 1.8243716955184937, "step": 15339 }, { "epoch": 1.2093023255813953, "grad_norm": 0.9095037579536438, "learning_rate": 0.0005, "loss": 1.674147129058838, "step": 15340 }, { "epoch": 1.2093811588490344, "grad_norm": 0.5375209450721741, "learning_rate": 0.0005, "loss": 1.6854665279388428, "step": 15341 }, { "epoch": 1.2094599921166733, "grad_norm": 0.6971024870872498, "learning_rate": 0.0005, "loss": 1.769379734992981, "step": 15342 }, { "epoch": 1.2095388253843122, "grad_norm": 0.6183691024780273, "learning_rate": 0.0005, "loss": 1.700358271598816, "step": 15343 }, { "epoch": 1.2096176586519511, "grad_norm": 0.74290931224823, "learning_rate": 0.0005, "loss": 1.943479061126709, "step": 15344 }, { "epoch": 1.20969649191959, "grad_norm": 1.1092184782028198, "learning_rate": 0.0005, "loss": 1.647542119026184, "step": 15345 }, { "epoch": 1.209775325187229, "grad_norm": 0.49397245049476624, "learning_rate": 0.0005, "loss": 1.52268385887146, "step": 15346 }, { "epoch": 1.2098541584548679, "grad_norm": 0.5862923264503479, "learning_rate": 0.0005, "loss": 1.7362022399902344, "step": 15347 }, { "epoch": 1.209932991722507, "grad_norm": 0.4743283987045288, "learning_rate": 0.0005, "loss": 1.6937744617462158, "step": 15348 }, { "epoch": 1.210011824990146, "grad_norm": 0.7029910683631897, "learning_rate": 0.0005, "loss": 1.7693767547607422, "step": 15349 }, { "epoch": 1.2100906582577848, "grad_norm": 0.715964674949646, "learning_rate": 0.0005, "loss": 1.9337722063064575, "step": 15350 }, { "epoch": 1.2101694915254237, "grad_norm": 0.5965186953544617, "learning_rate": 0.0005, "loss": 1.819394826889038, "step": 15351 }, { "epoch": 1.2102483247930627, "grad_norm": 0.4832349717617035, "learning_rate": 0.0005, "loss": 1.7820154428482056, "step": 15352 }, { "epoch": 1.2103271580607016, "grad_norm": 0.47322702407836914, "learning_rate": 0.0005, "loss": 1.585465908050537, "step": 15353 }, { "epoch": 1.2104059913283405, "grad_norm": 2.0568439960479736, "learning_rate": 0.0005, "loss": 1.690140962600708, "step": 15354 }, { "epoch": 1.2104848245959796, "grad_norm": 0.547853410243988, "learning_rate": 0.0005, "loss": 1.7652766704559326, "step": 15355 }, { "epoch": 1.2105636578636185, "grad_norm": 0.5577595829963684, "learning_rate": 0.0005, "loss": 1.798136830329895, "step": 15356 }, { "epoch": 1.2106424911312574, "grad_norm": 0.5738134980201721, "learning_rate": 0.0005, "loss": 1.9337682723999023, "step": 15357 }, { "epoch": 1.2107213243988963, "grad_norm": 0.6037928462028503, "learning_rate": 0.0005, "loss": 2.0706958770751953, "step": 15358 }, { "epoch": 1.2108001576665353, "grad_norm": 0.37919938564300537, "learning_rate": 0.0005, "loss": 1.2403044700622559, "step": 15359 }, { "epoch": 1.2108789909341742, "grad_norm": 0.8093419075012207, "learning_rate": 0.0005, "loss": 1.9471734762191772, "step": 15360 }, { "epoch": 1.210957824201813, "grad_norm": 0.5251843333244324, "learning_rate": 0.0005, "loss": 1.772080898284912, "step": 15361 }, { "epoch": 1.2110366574694522, "grad_norm": 0.5221585631370544, "learning_rate": 0.0005, "loss": 1.5193217992782593, "step": 15362 }, { "epoch": 1.2111154907370911, "grad_norm": 0.7401081323623657, "learning_rate": 0.0005, "loss": 1.909031867980957, "step": 15363 }, { "epoch": 1.21119432400473, "grad_norm": 0.5715287923812866, "learning_rate": 0.0005, "loss": 1.7513337135314941, "step": 15364 }, { "epoch": 1.211273157272369, "grad_norm": 0.5263634920120239, "learning_rate": 0.0005, "loss": 1.61033296585083, "step": 15365 }, { "epoch": 1.2113519905400079, "grad_norm": 0.547789454460144, "learning_rate": 0.0005, "loss": 1.7042200565338135, "step": 15366 }, { "epoch": 1.2114308238076468, "grad_norm": 0.739872395992279, "learning_rate": 0.0005, "loss": 1.4118233919143677, "step": 15367 }, { "epoch": 1.2115096570752857, "grad_norm": 0.7764166593551636, "learning_rate": 0.0005, "loss": 1.7487156391143799, "step": 15368 }, { "epoch": 1.2115884903429248, "grad_norm": 0.6095437407493591, "learning_rate": 0.0005, "loss": 1.7880592346191406, "step": 15369 }, { "epoch": 1.2116673236105637, "grad_norm": 0.6757930517196655, "learning_rate": 0.0005, "loss": 1.8232522010803223, "step": 15370 }, { "epoch": 1.2117461568782026, "grad_norm": 0.5972622632980347, "learning_rate": 0.0005, "loss": 1.4390864372253418, "step": 15371 }, { "epoch": 1.2118249901458416, "grad_norm": 0.6326949596405029, "learning_rate": 0.0005, "loss": 1.547756552696228, "step": 15372 }, { "epoch": 1.2119038234134805, "grad_norm": 0.7854399681091309, "learning_rate": 0.0005, "loss": 1.766884684562683, "step": 15373 }, { "epoch": 1.2119826566811194, "grad_norm": 0.5308806300163269, "learning_rate": 0.0005, "loss": 1.443404197692871, "step": 15374 }, { "epoch": 1.2120614899487583, "grad_norm": 0.6605204939842224, "learning_rate": 0.0005, "loss": 1.7657161951065063, "step": 15375 }, { "epoch": 1.2121403232163974, "grad_norm": 0.656840443611145, "learning_rate": 0.0005, "loss": 1.7573622465133667, "step": 15376 }, { "epoch": 1.2122191564840363, "grad_norm": 0.8295089602470398, "learning_rate": 0.0005, "loss": 1.785379409790039, "step": 15377 }, { "epoch": 1.2122979897516752, "grad_norm": 0.5412781834602356, "learning_rate": 0.0005, "loss": 1.5980944633483887, "step": 15378 }, { "epoch": 1.2123768230193142, "grad_norm": 0.6724934577941895, "learning_rate": 0.0005, "loss": 1.854155421257019, "step": 15379 }, { "epoch": 1.212455656286953, "grad_norm": 0.5477112531661987, "learning_rate": 0.0005, "loss": 1.7856460809707642, "step": 15380 }, { "epoch": 1.212534489554592, "grad_norm": 0.6187506318092346, "learning_rate": 0.0005, "loss": 1.6437326669692993, "step": 15381 }, { "epoch": 1.2126133228222309, "grad_norm": 0.5960636138916016, "learning_rate": 0.0005, "loss": 1.6722332239151, "step": 15382 }, { "epoch": 1.21269215608987, "grad_norm": 0.6167498826980591, "learning_rate": 0.0005, "loss": 1.694344401359558, "step": 15383 }, { "epoch": 1.212770989357509, "grad_norm": 0.6801584362983704, "learning_rate": 0.0005, "loss": 1.9990609884262085, "step": 15384 }, { "epoch": 1.2128498226251478, "grad_norm": 0.47675058245658875, "learning_rate": 0.0005, "loss": 1.5407789945602417, "step": 15385 }, { "epoch": 1.2129286558927868, "grad_norm": 0.5029101371765137, "learning_rate": 0.0005, "loss": 1.6901860237121582, "step": 15386 }, { "epoch": 1.2130074891604257, "grad_norm": 0.8664703369140625, "learning_rate": 0.0005, "loss": 1.5331554412841797, "step": 15387 }, { "epoch": 1.2130863224280646, "grad_norm": 0.7454119324684143, "learning_rate": 0.0005, "loss": 1.6153972148895264, "step": 15388 }, { "epoch": 1.2131651556957035, "grad_norm": 2.0856497287750244, "learning_rate": 0.0005, "loss": 1.8073066473007202, "step": 15389 }, { "epoch": 1.2132439889633426, "grad_norm": 0.6130282282829285, "learning_rate": 0.0005, "loss": 1.6222877502441406, "step": 15390 }, { "epoch": 1.2133228222309815, "grad_norm": 0.7530216574668884, "learning_rate": 0.0005, "loss": 1.6875832080841064, "step": 15391 }, { "epoch": 1.2134016554986204, "grad_norm": 0.6090624928474426, "learning_rate": 0.0005, "loss": 1.8161977529525757, "step": 15392 }, { "epoch": 1.2134804887662594, "grad_norm": 0.5576527118682861, "learning_rate": 0.0005, "loss": 1.5905730724334717, "step": 15393 }, { "epoch": 1.2135593220338983, "grad_norm": 0.5345066785812378, "learning_rate": 0.0005, "loss": 1.5965255498886108, "step": 15394 }, { "epoch": 1.2136381553015372, "grad_norm": 0.6734365820884705, "learning_rate": 0.0005, "loss": 1.7963106632232666, "step": 15395 }, { "epoch": 1.213716988569176, "grad_norm": 0.6728298664093018, "learning_rate": 0.0005, "loss": 1.8068647384643555, "step": 15396 }, { "epoch": 1.2137958218368152, "grad_norm": 0.4819977879524231, "learning_rate": 0.0005, "loss": 1.6000534296035767, "step": 15397 }, { "epoch": 1.2138746551044541, "grad_norm": 0.5214294791221619, "learning_rate": 0.0005, "loss": 1.6643354892730713, "step": 15398 }, { "epoch": 1.213953488372093, "grad_norm": 0.5766260623931885, "learning_rate": 0.0005, "loss": 1.5916597843170166, "step": 15399 }, { "epoch": 1.214032321639732, "grad_norm": 0.568316638469696, "learning_rate": 0.0005, "loss": 1.588585376739502, "step": 15400 }, { "epoch": 1.2141111549073709, "grad_norm": 0.6050885915756226, "learning_rate": 0.0005, "loss": 1.8303635120391846, "step": 15401 }, { "epoch": 1.2141899881750098, "grad_norm": 0.7613165378570557, "learning_rate": 0.0005, "loss": 1.826209545135498, "step": 15402 }, { "epoch": 1.2142688214426487, "grad_norm": 0.5834946036338806, "learning_rate": 0.0005, "loss": 1.6936662197113037, "step": 15403 }, { "epoch": 1.2143476547102878, "grad_norm": 0.570122480392456, "learning_rate": 0.0005, "loss": 1.5414698123931885, "step": 15404 }, { "epoch": 1.2144264879779267, "grad_norm": 0.6121641397476196, "learning_rate": 0.0005, "loss": 1.5607190132141113, "step": 15405 }, { "epoch": 1.2145053212455656, "grad_norm": 0.45634374022483826, "learning_rate": 0.0005, "loss": 1.6539703607559204, "step": 15406 }, { "epoch": 1.2145841545132046, "grad_norm": 0.5073204040527344, "learning_rate": 0.0005, "loss": 1.738818883895874, "step": 15407 }, { "epoch": 1.2146629877808435, "grad_norm": 0.6901116967201233, "learning_rate": 0.0005, "loss": 1.6395686864852905, "step": 15408 }, { "epoch": 1.2147418210484824, "grad_norm": 0.4322959780693054, "learning_rate": 0.0005, "loss": 1.501537799835205, "step": 15409 }, { "epoch": 1.2148206543161213, "grad_norm": 0.582693874835968, "learning_rate": 0.0005, "loss": 1.693469762802124, "step": 15410 }, { "epoch": 1.2148994875837604, "grad_norm": 0.48575878143310547, "learning_rate": 0.0005, "loss": 1.428318738937378, "step": 15411 }, { "epoch": 1.2149783208513993, "grad_norm": 0.6709820032119751, "learning_rate": 0.0005, "loss": 1.7355296611785889, "step": 15412 }, { "epoch": 1.2150571541190383, "grad_norm": 0.48386818170547485, "learning_rate": 0.0005, "loss": 1.6974577903747559, "step": 15413 }, { "epoch": 1.2151359873866772, "grad_norm": 0.4721604287624359, "learning_rate": 0.0005, "loss": 1.7572507858276367, "step": 15414 }, { "epoch": 1.215214820654316, "grad_norm": 0.509026288986206, "learning_rate": 0.0005, "loss": 1.8127801418304443, "step": 15415 }, { "epoch": 1.215293653921955, "grad_norm": 0.48964327573776245, "learning_rate": 0.0005, "loss": 1.6415499448776245, "step": 15416 }, { "epoch": 1.215372487189594, "grad_norm": 0.4068465530872345, "learning_rate": 0.0005, "loss": 1.4946331977844238, "step": 15417 }, { "epoch": 1.215451320457233, "grad_norm": 0.4873611629009247, "learning_rate": 0.0005, "loss": 1.5597875118255615, "step": 15418 }, { "epoch": 1.215530153724872, "grad_norm": 0.6033602952957153, "learning_rate": 0.0005, "loss": 1.6281588077545166, "step": 15419 }, { "epoch": 1.2156089869925109, "grad_norm": 0.4458174407482147, "learning_rate": 0.0005, "loss": 1.5151522159576416, "step": 15420 }, { "epoch": 1.2156878202601498, "grad_norm": 0.49028339982032776, "learning_rate": 0.0005, "loss": 1.5488808155059814, "step": 15421 }, { "epoch": 1.2157666535277887, "grad_norm": 0.6122227311134338, "learning_rate": 0.0005, "loss": 1.9345307350158691, "step": 15422 }, { "epoch": 1.2158454867954276, "grad_norm": 0.42569494247436523, "learning_rate": 0.0005, "loss": 1.6004819869995117, "step": 15423 }, { "epoch": 1.2159243200630665, "grad_norm": 0.5426173806190491, "learning_rate": 0.0005, "loss": 1.8100842237472534, "step": 15424 }, { "epoch": 1.2160031533307056, "grad_norm": 0.4217762351036072, "learning_rate": 0.0005, "loss": 1.401275396347046, "step": 15425 }, { "epoch": 1.2160819865983445, "grad_norm": 0.5347508192062378, "learning_rate": 0.0005, "loss": 1.5564309358596802, "step": 15426 }, { "epoch": 1.2161608198659835, "grad_norm": 0.523810625076294, "learning_rate": 0.0005, "loss": 1.6115432977676392, "step": 15427 }, { "epoch": 1.2162396531336224, "grad_norm": 0.644193172454834, "learning_rate": 0.0005, "loss": 1.898345708847046, "step": 15428 }, { "epoch": 1.2163184864012613, "grad_norm": 0.5200145840644836, "learning_rate": 0.0005, "loss": 1.7742053270339966, "step": 15429 }, { "epoch": 1.2163973196689002, "grad_norm": 0.5339886546134949, "learning_rate": 0.0005, "loss": 1.6627964973449707, "step": 15430 }, { "epoch": 1.216476152936539, "grad_norm": 0.45033353567123413, "learning_rate": 0.0005, "loss": 1.5737485885620117, "step": 15431 }, { "epoch": 1.2165549862041782, "grad_norm": 0.5385597348213196, "learning_rate": 0.0005, "loss": 1.566375970840454, "step": 15432 }, { "epoch": 1.2166338194718171, "grad_norm": 0.6056596636772156, "learning_rate": 0.0005, "loss": 1.6505719423294067, "step": 15433 }, { "epoch": 1.216712652739456, "grad_norm": 0.5678758025169373, "learning_rate": 0.0005, "loss": 1.5608994960784912, "step": 15434 }, { "epoch": 1.216791486007095, "grad_norm": 0.5104716420173645, "learning_rate": 0.0005, "loss": 1.5889768600463867, "step": 15435 }, { "epoch": 1.2168703192747339, "grad_norm": 0.6561195254325867, "learning_rate": 0.0005, "loss": 1.6992807388305664, "step": 15436 }, { "epoch": 1.2169491525423728, "grad_norm": 0.5407997369766235, "learning_rate": 0.0005, "loss": 1.6787532567977905, "step": 15437 }, { "epoch": 1.2170279858100117, "grad_norm": 0.6117703914642334, "learning_rate": 0.0005, "loss": 1.597711443901062, "step": 15438 }, { "epoch": 1.2171068190776508, "grad_norm": 0.49127861857414246, "learning_rate": 0.0005, "loss": 1.6716270446777344, "step": 15439 }, { "epoch": 1.2171856523452897, "grad_norm": 1.0047461986541748, "learning_rate": 0.0005, "loss": 1.6806862354278564, "step": 15440 }, { "epoch": 1.2172644856129287, "grad_norm": 0.4570590555667877, "learning_rate": 0.0005, "loss": 1.4918488264083862, "step": 15441 }, { "epoch": 1.2173433188805676, "grad_norm": 0.5642048120498657, "learning_rate": 0.0005, "loss": 1.6648890972137451, "step": 15442 }, { "epoch": 1.2174221521482065, "grad_norm": 0.6918135285377502, "learning_rate": 0.0005, "loss": 1.562756896018982, "step": 15443 }, { "epoch": 1.2175009854158454, "grad_norm": 0.531454861164093, "learning_rate": 0.0005, "loss": 1.4868518114089966, "step": 15444 }, { "epoch": 1.2175798186834845, "grad_norm": 0.642273485660553, "learning_rate": 0.0005, "loss": 1.9599283933639526, "step": 15445 }, { "epoch": 1.2176586519511234, "grad_norm": 0.49154841899871826, "learning_rate": 0.0005, "loss": 1.4344508647918701, "step": 15446 }, { "epoch": 1.2177374852187623, "grad_norm": 0.6169136166572571, "learning_rate": 0.0005, "loss": 1.6671442985534668, "step": 15447 }, { "epoch": 1.2178163184864013, "grad_norm": 0.4313957989215851, "learning_rate": 0.0005, "loss": 1.577108383178711, "step": 15448 }, { "epoch": 1.2178951517540402, "grad_norm": 0.5777519941329956, "learning_rate": 0.0005, "loss": 1.6489334106445312, "step": 15449 }, { "epoch": 1.217973985021679, "grad_norm": 0.4485616385936737, "learning_rate": 0.0005, "loss": 1.5409431457519531, "step": 15450 }, { "epoch": 1.218052818289318, "grad_norm": 0.7108526229858398, "learning_rate": 0.0005, "loss": 1.7169557809829712, "step": 15451 }, { "epoch": 1.2181316515569571, "grad_norm": 0.5443329811096191, "learning_rate": 0.0005, "loss": 1.610209345817566, "step": 15452 }, { "epoch": 1.218210484824596, "grad_norm": 0.6513167023658752, "learning_rate": 0.0005, "loss": 1.6712229251861572, "step": 15453 }, { "epoch": 1.218289318092235, "grad_norm": 0.8445947170257568, "learning_rate": 0.0005, "loss": 1.733769416809082, "step": 15454 }, { "epoch": 1.2183681513598739, "grad_norm": 0.5621665716171265, "learning_rate": 0.0005, "loss": 1.5797659158706665, "step": 15455 }, { "epoch": 1.2184469846275128, "grad_norm": 0.6021209955215454, "learning_rate": 0.0005, "loss": 1.7985820770263672, "step": 15456 }, { "epoch": 1.2185258178951517, "grad_norm": 0.4673624336719513, "learning_rate": 0.0005, "loss": 1.623201847076416, "step": 15457 }, { "epoch": 1.2186046511627908, "grad_norm": 0.7644620537757874, "learning_rate": 0.0005, "loss": 1.8609917163848877, "step": 15458 }, { "epoch": 1.2186834844304297, "grad_norm": 0.5048531889915466, "learning_rate": 0.0005, "loss": 1.448301076889038, "step": 15459 }, { "epoch": 1.2187623176980686, "grad_norm": 0.4448542892932892, "learning_rate": 0.0005, "loss": 1.5070182085037231, "step": 15460 }, { "epoch": 1.2188411509657076, "grad_norm": 0.5524616241455078, "learning_rate": 0.0005, "loss": 1.6930701732635498, "step": 15461 }, { "epoch": 1.2189199842333465, "grad_norm": 0.43108969926834106, "learning_rate": 0.0005, "loss": 1.5100219249725342, "step": 15462 }, { "epoch": 1.2189988175009854, "grad_norm": 0.49916547536849976, "learning_rate": 0.0005, "loss": 1.4242932796478271, "step": 15463 }, { "epoch": 1.2190776507686243, "grad_norm": 0.5559829473495483, "learning_rate": 0.0005, "loss": 1.5316004753112793, "step": 15464 }, { "epoch": 1.2191564840362634, "grad_norm": 0.5896731019020081, "learning_rate": 0.0005, "loss": 1.618375539779663, "step": 15465 }, { "epoch": 1.2192353173039023, "grad_norm": 0.5054067373275757, "learning_rate": 0.0005, "loss": 1.6137453317642212, "step": 15466 }, { "epoch": 1.2193141505715412, "grad_norm": 0.63897705078125, "learning_rate": 0.0005, "loss": 1.722415566444397, "step": 15467 }, { "epoch": 1.2193929838391802, "grad_norm": 0.6242159008979797, "learning_rate": 0.0005, "loss": 1.6102275848388672, "step": 15468 }, { "epoch": 1.219471817106819, "grad_norm": 0.7826116681098938, "learning_rate": 0.0005, "loss": 1.515087604522705, "step": 15469 }, { "epoch": 1.219550650374458, "grad_norm": 0.6305211782455444, "learning_rate": 0.0005, "loss": 1.8051071166992188, "step": 15470 }, { "epoch": 1.219629483642097, "grad_norm": 0.5245979428291321, "learning_rate": 0.0005, "loss": 1.6173348426818848, "step": 15471 }, { "epoch": 1.219708316909736, "grad_norm": 0.4486229717731476, "learning_rate": 0.0005, "loss": 1.4709914922714233, "step": 15472 }, { "epoch": 1.219787150177375, "grad_norm": 0.5399969220161438, "learning_rate": 0.0005, "loss": 1.8277534246444702, "step": 15473 }, { "epoch": 1.2198659834450138, "grad_norm": 0.6166089773178101, "learning_rate": 0.0005, "loss": 1.7589221000671387, "step": 15474 }, { "epoch": 1.2199448167126528, "grad_norm": 0.5410759449005127, "learning_rate": 0.0005, "loss": 1.6929525136947632, "step": 15475 }, { "epoch": 1.2200236499802917, "grad_norm": 0.502708375453949, "learning_rate": 0.0005, "loss": 1.5147846937179565, "step": 15476 }, { "epoch": 1.2201024832479306, "grad_norm": 1.3637025356292725, "learning_rate": 0.0005, "loss": 1.7457058429718018, "step": 15477 }, { "epoch": 1.2201813165155695, "grad_norm": 0.6220483779907227, "learning_rate": 0.0005, "loss": 1.5865603685379028, "step": 15478 }, { "epoch": 1.2202601497832086, "grad_norm": 0.6258396506309509, "learning_rate": 0.0005, "loss": 1.7743258476257324, "step": 15479 }, { "epoch": 1.2203389830508475, "grad_norm": 0.6615880131721497, "learning_rate": 0.0005, "loss": 1.6426208019256592, "step": 15480 }, { "epoch": 1.2204178163184864, "grad_norm": 0.7429782748222351, "learning_rate": 0.0005, "loss": 1.8012579679489136, "step": 15481 }, { "epoch": 1.2204966495861254, "grad_norm": 0.5282760858535767, "learning_rate": 0.0005, "loss": 1.4038951396942139, "step": 15482 }, { "epoch": 1.2205754828537643, "grad_norm": 0.5180088877677917, "learning_rate": 0.0005, "loss": 1.6639137268066406, "step": 15483 }, { "epoch": 1.2206543161214032, "grad_norm": 0.7550874948501587, "learning_rate": 0.0005, "loss": 1.5765984058380127, "step": 15484 }, { "epoch": 1.220733149389042, "grad_norm": 0.6076250672340393, "learning_rate": 0.0005, "loss": 1.6804262399673462, "step": 15485 }, { "epoch": 1.2208119826566812, "grad_norm": 0.4890037775039673, "learning_rate": 0.0005, "loss": 1.4494355916976929, "step": 15486 }, { "epoch": 1.2208908159243201, "grad_norm": 0.6042231321334839, "learning_rate": 0.0005, "loss": 1.9094789028167725, "step": 15487 }, { "epoch": 1.220969649191959, "grad_norm": 0.7909720540046692, "learning_rate": 0.0005, "loss": 1.8238626718521118, "step": 15488 }, { "epoch": 1.221048482459598, "grad_norm": 0.5345808863639832, "learning_rate": 0.0005, "loss": 1.7389605045318604, "step": 15489 }, { "epoch": 1.2211273157272369, "grad_norm": 0.5706446766853333, "learning_rate": 0.0005, "loss": 1.6612348556518555, "step": 15490 }, { "epoch": 1.2212061489948758, "grad_norm": 0.6959536075592041, "learning_rate": 0.0005, "loss": 1.8230180740356445, "step": 15491 }, { "epoch": 1.2212849822625147, "grad_norm": 0.6248745918273926, "learning_rate": 0.0005, "loss": 1.8222694396972656, "step": 15492 }, { "epoch": 1.2213638155301538, "grad_norm": 0.8361755013465881, "learning_rate": 0.0005, "loss": 1.6551132202148438, "step": 15493 }, { "epoch": 1.2214426487977927, "grad_norm": 0.5806564092636108, "learning_rate": 0.0005, "loss": 1.8408218622207642, "step": 15494 }, { "epoch": 1.2215214820654317, "grad_norm": 0.5036678314208984, "learning_rate": 0.0005, "loss": 1.568590760231018, "step": 15495 }, { "epoch": 1.2216003153330706, "grad_norm": 0.5062741637229919, "learning_rate": 0.0005, "loss": 1.4722418785095215, "step": 15496 }, { "epoch": 1.2216791486007095, "grad_norm": 0.541355311870575, "learning_rate": 0.0005, "loss": 1.7051033973693848, "step": 15497 }, { "epoch": 1.2217579818683484, "grad_norm": 0.5528529286384583, "learning_rate": 0.0005, "loss": 1.7881133556365967, "step": 15498 }, { "epoch": 1.2218368151359873, "grad_norm": 0.534929633140564, "learning_rate": 0.0005, "loss": 1.8048206567764282, "step": 15499 }, { "epoch": 1.2219156484036264, "grad_norm": 0.49881646037101746, "learning_rate": 0.0005, "loss": 1.5501958131790161, "step": 15500 }, { "epoch": 1.2219944816712653, "grad_norm": 0.5551921129226685, "learning_rate": 0.0005, "loss": 1.5053942203521729, "step": 15501 }, { "epoch": 1.2220733149389043, "grad_norm": 0.4542054533958435, "learning_rate": 0.0005, "loss": 1.4500038623809814, "step": 15502 }, { "epoch": 1.2221521482065432, "grad_norm": 0.8229964375495911, "learning_rate": 0.0005, "loss": 1.7266005277633667, "step": 15503 }, { "epoch": 1.222230981474182, "grad_norm": 0.508608341217041, "learning_rate": 0.0005, "loss": 1.5053274631500244, "step": 15504 }, { "epoch": 1.222309814741821, "grad_norm": 0.4273674488067627, "learning_rate": 0.0005, "loss": 1.457770824432373, "step": 15505 }, { "epoch": 1.22238864800946, "grad_norm": 0.6351158022880554, "learning_rate": 0.0005, "loss": 1.6778696775436401, "step": 15506 }, { "epoch": 1.222467481277099, "grad_norm": 0.7204036116600037, "learning_rate": 0.0005, "loss": 1.9977729320526123, "step": 15507 }, { "epoch": 1.222546314544738, "grad_norm": 0.5636265873908997, "learning_rate": 0.0005, "loss": 1.8328807353973389, "step": 15508 }, { "epoch": 1.2226251478123769, "grad_norm": 0.5308933258056641, "learning_rate": 0.0005, "loss": 1.5577747821807861, "step": 15509 }, { "epoch": 1.2227039810800158, "grad_norm": 0.5330300331115723, "learning_rate": 0.0005, "loss": 1.955057978630066, "step": 15510 }, { "epoch": 1.2227828143476547, "grad_norm": 0.6075654029846191, "learning_rate": 0.0005, "loss": 1.9042057991027832, "step": 15511 }, { "epoch": 1.2228616476152936, "grad_norm": 0.5406442880630493, "learning_rate": 0.0005, "loss": 1.8974134922027588, "step": 15512 }, { "epoch": 1.2229404808829325, "grad_norm": 2.1282901763916016, "learning_rate": 0.0005, "loss": 1.6532180309295654, "step": 15513 }, { "epoch": 1.2230193141505716, "grad_norm": 0.6206493973731995, "learning_rate": 0.0005, "loss": 1.79414701461792, "step": 15514 }, { "epoch": 1.2230981474182105, "grad_norm": 0.5269594788551331, "learning_rate": 0.0005, "loss": 1.5855172872543335, "step": 15515 }, { "epoch": 1.2231769806858495, "grad_norm": 0.5641602277755737, "learning_rate": 0.0005, "loss": 1.6921674013137817, "step": 15516 }, { "epoch": 1.2232558139534884, "grad_norm": 0.7124523520469666, "learning_rate": 0.0005, "loss": 1.5047731399536133, "step": 15517 }, { "epoch": 1.2233346472211273, "grad_norm": 0.5576900243759155, "learning_rate": 0.0005, "loss": 1.527940034866333, "step": 15518 }, { "epoch": 1.2234134804887662, "grad_norm": 0.5822327136993408, "learning_rate": 0.0005, "loss": 1.6527225971221924, "step": 15519 }, { "epoch": 1.223492313756405, "grad_norm": 0.6169250011444092, "learning_rate": 0.0005, "loss": 1.6723861694335938, "step": 15520 }, { "epoch": 1.2235711470240442, "grad_norm": 0.6225999593734741, "learning_rate": 0.0005, "loss": 1.6102349758148193, "step": 15521 }, { "epoch": 1.2236499802916831, "grad_norm": 0.6961748003959656, "learning_rate": 0.0005, "loss": 1.78102707862854, "step": 15522 }, { "epoch": 1.223728813559322, "grad_norm": 0.49596577882766724, "learning_rate": 0.0005, "loss": 1.6724354028701782, "step": 15523 }, { "epoch": 1.223807646826961, "grad_norm": 0.7185206413269043, "learning_rate": 0.0005, "loss": 1.518758773803711, "step": 15524 }, { "epoch": 1.2238864800945999, "grad_norm": 0.48717138171195984, "learning_rate": 0.0005, "loss": 1.6473863124847412, "step": 15525 }, { "epoch": 1.2239653133622388, "grad_norm": 0.6318318843841553, "learning_rate": 0.0005, "loss": 1.7395291328430176, "step": 15526 }, { "epoch": 1.2240441466298777, "grad_norm": 0.5318425893783569, "learning_rate": 0.0005, "loss": 1.563077688217163, "step": 15527 }, { "epoch": 1.2241229798975168, "grad_norm": 0.5340604186058044, "learning_rate": 0.0005, "loss": 1.4416782855987549, "step": 15528 }, { "epoch": 1.2242018131651557, "grad_norm": 0.5290209054946899, "learning_rate": 0.0005, "loss": 1.7036128044128418, "step": 15529 }, { "epoch": 1.2242806464327947, "grad_norm": 0.811394989490509, "learning_rate": 0.0005, "loss": 1.8441884517669678, "step": 15530 }, { "epoch": 1.2243594797004336, "grad_norm": 0.6452702283859253, "learning_rate": 0.0005, "loss": 1.7644715309143066, "step": 15531 }, { "epoch": 1.2244383129680725, "grad_norm": 0.5320075750350952, "learning_rate": 0.0005, "loss": 1.8108055591583252, "step": 15532 }, { "epoch": 1.2245171462357114, "grad_norm": 0.5544009208679199, "learning_rate": 0.0005, "loss": 1.914544701576233, "step": 15533 }, { "epoch": 1.2245959795033503, "grad_norm": 0.500809907913208, "learning_rate": 0.0005, "loss": 1.6467806100845337, "step": 15534 }, { "epoch": 1.2246748127709894, "grad_norm": 0.4846452474594116, "learning_rate": 0.0005, "loss": 1.5091334581375122, "step": 15535 }, { "epoch": 1.2247536460386284, "grad_norm": 0.5357967615127563, "learning_rate": 0.0005, "loss": 2.030057907104492, "step": 15536 }, { "epoch": 1.2248324793062673, "grad_norm": 0.5331000089645386, "learning_rate": 0.0005, "loss": 1.6317421197891235, "step": 15537 }, { "epoch": 1.2249113125739062, "grad_norm": 0.5410763025283813, "learning_rate": 0.0005, "loss": 1.9937717914581299, "step": 15538 }, { "epoch": 1.224990145841545, "grad_norm": 0.6154453754425049, "learning_rate": 0.0005, "loss": 1.8019182682037354, "step": 15539 }, { "epoch": 1.225068979109184, "grad_norm": 0.505126953125, "learning_rate": 0.0005, "loss": 1.7022106647491455, "step": 15540 }, { "epoch": 1.225147812376823, "grad_norm": 0.6068640351295471, "learning_rate": 0.0005, "loss": 1.800572156906128, "step": 15541 }, { "epoch": 1.225226645644462, "grad_norm": 0.6727757453918457, "learning_rate": 0.0005, "loss": 1.7070868015289307, "step": 15542 }, { "epoch": 1.225305478912101, "grad_norm": 0.5317418575286865, "learning_rate": 0.0005, "loss": 1.5868330001831055, "step": 15543 }, { "epoch": 1.2253843121797399, "grad_norm": 0.4592323303222656, "learning_rate": 0.0005, "loss": 1.5491385459899902, "step": 15544 }, { "epoch": 1.2254631454473788, "grad_norm": 0.43360278010368347, "learning_rate": 0.0005, "loss": 1.472320556640625, "step": 15545 }, { "epoch": 1.2255419787150177, "grad_norm": 0.4818079471588135, "learning_rate": 0.0005, "loss": 1.5685274600982666, "step": 15546 }, { "epoch": 1.2256208119826566, "grad_norm": 0.8190422058105469, "learning_rate": 0.0005, "loss": 1.769265055656433, "step": 15547 }, { "epoch": 1.2256996452502955, "grad_norm": 0.5142362713813782, "learning_rate": 0.0005, "loss": 1.6464321613311768, "step": 15548 }, { "epoch": 1.2257784785179346, "grad_norm": 0.5055639147758484, "learning_rate": 0.0005, "loss": 1.6969468593597412, "step": 15549 }, { "epoch": 1.2258573117855736, "grad_norm": 0.5550550222396851, "learning_rate": 0.0005, "loss": 1.6125366687774658, "step": 15550 }, { "epoch": 1.2259361450532125, "grad_norm": 0.6027385592460632, "learning_rate": 0.0005, "loss": 1.594346046447754, "step": 15551 }, { "epoch": 1.2260149783208514, "grad_norm": 0.5422868728637695, "learning_rate": 0.0005, "loss": 1.584097146987915, "step": 15552 }, { "epoch": 1.2260938115884903, "grad_norm": 0.5064677000045776, "learning_rate": 0.0005, "loss": 1.5383331775665283, "step": 15553 }, { "epoch": 1.2261726448561292, "grad_norm": 0.5041961073875427, "learning_rate": 0.0005, "loss": 1.5496654510498047, "step": 15554 }, { "epoch": 1.2262514781237681, "grad_norm": 0.49385157227516174, "learning_rate": 0.0005, "loss": 1.6459871530532837, "step": 15555 }, { "epoch": 1.2263303113914072, "grad_norm": 0.5817968249320984, "learning_rate": 0.0005, "loss": 1.6841850280761719, "step": 15556 }, { "epoch": 1.2264091446590462, "grad_norm": 0.5505383014678955, "learning_rate": 0.0005, "loss": 1.5986354351043701, "step": 15557 }, { "epoch": 1.226487977926685, "grad_norm": 0.539481520652771, "learning_rate": 0.0005, "loss": 1.6888874769210815, "step": 15558 }, { "epoch": 1.226566811194324, "grad_norm": 0.5064769983291626, "learning_rate": 0.0005, "loss": 1.773051142692566, "step": 15559 }, { "epoch": 1.226645644461963, "grad_norm": 0.7085024118423462, "learning_rate": 0.0005, "loss": 1.6889864206314087, "step": 15560 }, { "epoch": 1.2267244777296018, "grad_norm": 0.5380257368087769, "learning_rate": 0.0005, "loss": 1.6878269910812378, "step": 15561 }, { "epoch": 1.2268033109972407, "grad_norm": 0.5616385340690613, "learning_rate": 0.0005, "loss": 1.8137900829315186, "step": 15562 }, { "epoch": 1.2268821442648798, "grad_norm": 0.4609825909137726, "learning_rate": 0.0005, "loss": 1.4826948642730713, "step": 15563 }, { "epoch": 1.2269609775325188, "grad_norm": 0.5167688131332397, "learning_rate": 0.0005, "loss": 1.6591813564300537, "step": 15564 }, { "epoch": 1.2270398108001577, "grad_norm": 0.6323534846305847, "learning_rate": 0.0005, "loss": 1.6009185314178467, "step": 15565 }, { "epoch": 1.2271186440677966, "grad_norm": 0.4861752390861511, "learning_rate": 0.0005, "loss": 1.633821725845337, "step": 15566 }, { "epoch": 1.2271974773354355, "grad_norm": 0.646930456161499, "learning_rate": 0.0005, "loss": 1.4742443561553955, "step": 15567 }, { "epoch": 1.2272763106030744, "grad_norm": 0.41595229506492615, "learning_rate": 0.0005, "loss": 1.477277398109436, "step": 15568 }, { "epoch": 1.2273551438707133, "grad_norm": 0.497366338968277, "learning_rate": 0.0005, "loss": 1.546821117401123, "step": 15569 }, { "epoch": 1.2274339771383524, "grad_norm": 0.5716981887817383, "learning_rate": 0.0005, "loss": 1.6258809566497803, "step": 15570 }, { "epoch": 1.2275128104059914, "grad_norm": 0.47864821553230286, "learning_rate": 0.0005, "loss": 1.5692678689956665, "step": 15571 }, { "epoch": 1.2275916436736303, "grad_norm": 0.526311457157135, "learning_rate": 0.0005, "loss": 1.5213087797164917, "step": 15572 }, { "epoch": 1.2276704769412692, "grad_norm": 0.7683601379394531, "learning_rate": 0.0005, "loss": 1.701702356338501, "step": 15573 }, { "epoch": 1.227749310208908, "grad_norm": 0.5012270212173462, "learning_rate": 0.0005, "loss": 1.590179443359375, "step": 15574 }, { "epoch": 1.227828143476547, "grad_norm": 1.0953929424285889, "learning_rate": 0.0005, "loss": 1.8576973676681519, "step": 15575 }, { "epoch": 1.2279069767441861, "grad_norm": 0.5505748987197876, "learning_rate": 0.0005, "loss": 1.6739128828048706, "step": 15576 }, { "epoch": 1.227985810011825, "grad_norm": 0.5985298752784729, "learning_rate": 0.0005, "loss": 1.7048137187957764, "step": 15577 }, { "epoch": 1.228064643279464, "grad_norm": 0.5383005142211914, "learning_rate": 0.0005, "loss": 1.6712754964828491, "step": 15578 }, { "epoch": 1.2281434765471029, "grad_norm": 0.5153027176856995, "learning_rate": 0.0005, "loss": 1.685826063156128, "step": 15579 }, { "epoch": 1.2282223098147418, "grad_norm": 0.4931628108024597, "learning_rate": 0.0005, "loss": 1.6001014709472656, "step": 15580 }, { "epoch": 1.2283011430823807, "grad_norm": 0.537674605846405, "learning_rate": 0.0005, "loss": 1.6859432458877563, "step": 15581 }, { "epoch": 1.2283799763500198, "grad_norm": 0.6268202662467957, "learning_rate": 0.0005, "loss": 1.731447458267212, "step": 15582 }, { "epoch": 1.2284588096176587, "grad_norm": 0.5735504031181335, "learning_rate": 0.0005, "loss": 1.645665168762207, "step": 15583 }, { "epoch": 1.2285376428852977, "grad_norm": 0.4465550482273102, "learning_rate": 0.0005, "loss": 1.5020678043365479, "step": 15584 }, { "epoch": 1.2286164761529366, "grad_norm": 0.7464560270309448, "learning_rate": 0.0005, "loss": 1.5302363634109497, "step": 15585 }, { "epoch": 1.2286953094205755, "grad_norm": 0.5637055039405823, "learning_rate": 0.0005, "loss": 1.8385813236236572, "step": 15586 }, { "epoch": 1.2287741426882144, "grad_norm": 0.5594606399536133, "learning_rate": 0.0005, "loss": 1.817036747932434, "step": 15587 }, { "epoch": 1.2288529759558533, "grad_norm": 0.5750634074211121, "learning_rate": 0.0005, "loss": 1.6512863636016846, "step": 15588 }, { "epoch": 1.2289318092234924, "grad_norm": 0.6216475963592529, "learning_rate": 0.0005, "loss": 1.6855584383010864, "step": 15589 }, { "epoch": 1.2290106424911313, "grad_norm": 0.6365160942077637, "learning_rate": 0.0005, "loss": 1.5985376834869385, "step": 15590 }, { "epoch": 1.2290894757587703, "grad_norm": 0.5850534439086914, "learning_rate": 0.0005, "loss": 1.8248740434646606, "step": 15591 }, { "epoch": 1.2291683090264092, "grad_norm": 0.6071105003356934, "learning_rate": 0.0005, "loss": 1.660627841949463, "step": 15592 }, { "epoch": 1.229247142294048, "grad_norm": 0.6423181295394897, "learning_rate": 0.0005, "loss": 1.7495845556259155, "step": 15593 }, { "epoch": 1.229325975561687, "grad_norm": 0.5087584257125854, "learning_rate": 0.0005, "loss": 1.624342441558838, "step": 15594 }, { "epoch": 1.229404808829326, "grad_norm": 0.470279335975647, "learning_rate": 0.0005, "loss": 1.4813783168792725, "step": 15595 }, { "epoch": 1.229483642096965, "grad_norm": 0.5473673343658447, "learning_rate": 0.0005, "loss": 1.6823337078094482, "step": 15596 }, { "epoch": 1.229562475364604, "grad_norm": 0.6595911979675293, "learning_rate": 0.0005, "loss": 1.842409372329712, "step": 15597 }, { "epoch": 1.2296413086322429, "grad_norm": 0.5124372243881226, "learning_rate": 0.0005, "loss": 1.4195890426635742, "step": 15598 }, { "epoch": 1.2297201418998818, "grad_norm": 0.48838260769844055, "learning_rate": 0.0005, "loss": 1.6247891187667847, "step": 15599 }, { "epoch": 1.2297989751675207, "grad_norm": 0.4173114597797394, "learning_rate": 0.0005, "loss": 1.4892305135726929, "step": 15600 }, { "epoch": 1.2298778084351596, "grad_norm": 0.5788822174072266, "learning_rate": 0.0005, "loss": 1.6962461471557617, "step": 15601 }, { "epoch": 1.2299566417027985, "grad_norm": 0.4434886574745178, "learning_rate": 0.0005, "loss": 1.6959178447723389, "step": 15602 }, { "epoch": 1.2300354749704376, "grad_norm": 0.6787210702896118, "learning_rate": 0.0005, "loss": 1.5561795234680176, "step": 15603 }, { "epoch": 1.2301143082380765, "grad_norm": 0.588450014591217, "learning_rate": 0.0005, "loss": 1.8263182640075684, "step": 15604 }, { "epoch": 1.2301931415057155, "grad_norm": 0.54229336977005, "learning_rate": 0.0005, "loss": 1.7055690288543701, "step": 15605 }, { "epoch": 1.2302719747733544, "grad_norm": 0.9589024186134338, "learning_rate": 0.0005, "loss": 1.7000638246536255, "step": 15606 }, { "epoch": 1.2303508080409933, "grad_norm": 0.5634742379188538, "learning_rate": 0.0005, "loss": 1.7486813068389893, "step": 15607 }, { "epoch": 1.2304296413086322, "grad_norm": 0.5316523313522339, "learning_rate": 0.0005, "loss": 1.7796121835708618, "step": 15608 }, { "epoch": 1.230508474576271, "grad_norm": 0.6712679862976074, "learning_rate": 0.0005, "loss": 1.8727571964263916, "step": 15609 }, { "epoch": 1.2305873078439102, "grad_norm": 0.5893130898475647, "learning_rate": 0.0005, "loss": 1.5917956829071045, "step": 15610 }, { "epoch": 1.2306661411115492, "grad_norm": 0.7868282198905945, "learning_rate": 0.0005, "loss": 1.861222505569458, "step": 15611 }, { "epoch": 1.230744974379188, "grad_norm": 0.7434920072555542, "learning_rate": 0.0005, "loss": 1.8366949558258057, "step": 15612 }, { "epoch": 1.230823807646827, "grad_norm": 0.5649845600128174, "learning_rate": 0.0005, "loss": 1.8839342594146729, "step": 15613 }, { "epoch": 1.2309026409144659, "grad_norm": 0.6759474873542786, "learning_rate": 0.0005, "loss": 1.6265093088150024, "step": 15614 }, { "epoch": 1.2309814741821048, "grad_norm": 0.5244393348693848, "learning_rate": 0.0005, "loss": 1.7990751266479492, "step": 15615 }, { "epoch": 1.2310603074497437, "grad_norm": 0.5443044900894165, "learning_rate": 0.0005, "loss": 1.6907074451446533, "step": 15616 }, { "epoch": 1.2311391407173828, "grad_norm": 0.5623918175697327, "learning_rate": 0.0005, "loss": 1.4162029027938843, "step": 15617 }, { "epoch": 1.2312179739850218, "grad_norm": 0.5718559622764587, "learning_rate": 0.0005, "loss": 1.5534546375274658, "step": 15618 }, { "epoch": 1.2312968072526607, "grad_norm": 0.6104839444160461, "learning_rate": 0.0005, "loss": 1.5874192714691162, "step": 15619 }, { "epoch": 1.2313756405202996, "grad_norm": 0.4429222345352173, "learning_rate": 0.0005, "loss": 1.7056888341903687, "step": 15620 }, { "epoch": 1.2314544737879385, "grad_norm": 0.6051164865493774, "learning_rate": 0.0005, "loss": 1.7798391580581665, "step": 15621 }, { "epoch": 1.2315333070555774, "grad_norm": 0.5347055792808533, "learning_rate": 0.0005, "loss": 1.556023120880127, "step": 15622 }, { "epoch": 1.2316121403232163, "grad_norm": 0.5499024391174316, "learning_rate": 0.0005, "loss": 1.5412464141845703, "step": 15623 }, { "epoch": 1.2316909735908554, "grad_norm": 0.5500909090042114, "learning_rate": 0.0005, "loss": 1.6872986555099487, "step": 15624 }, { "epoch": 1.2317698068584944, "grad_norm": 0.6949670910835266, "learning_rate": 0.0005, "loss": 1.8273413181304932, "step": 15625 }, { "epoch": 1.2318486401261333, "grad_norm": 0.4719126224517822, "learning_rate": 0.0005, "loss": 1.7228840589523315, "step": 15626 }, { "epoch": 1.2319274733937722, "grad_norm": 0.545304000377655, "learning_rate": 0.0005, "loss": 1.7034261226654053, "step": 15627 }, { "epoch": 1.232006306661411, "grad_norm": 0.5171966552734375, "learning_rate": 0.0005, "loss": 1.7098308801651, "step": 15628 }, { "epoch": 1.23208513992905, "grad_norm": 0.4870721697807312, "learning_rate": 0.0005, "loss": 1.7132254838943481, "step": 15629 }, { "epoch": 1.232163973196689, "grad_norm": 0.5918062329292297, "learning_rate": 0.0005, "loss": 1.5815178155899048, "step": 15630 }, { "epoch": 1.232242806464328, "grad_norm": 0.6588762998580933, "learning_rate": 0.0005, "loss": 1.6573598384857178, "step": 15631 }, { "epoch": 1.232321639731967, "grad_norm": 0.6084684133529663, "learning_rate": 0.0005, "loss": 1.5619949102401733, "step": 15632 }, { "epoch": 1.2324004729996059, "grad_norm": 0.6370214223861694, "learning_rate": 0.0005, "loss": 1.5433350801467896, "step": 15633 }, { "epoch": 1.2324793062672448, "grad_norm": 0.5907426476478577, "learning_rate": 0.0005, "loss": 1.550118327140808, "step": 15634 }, { "epoch": 1.2325581395348837, "grad_norm": 0.6432005763053894, "learning_rate": 0.0005, "loss": 1.9507642984390259, "step": 15635 }, { "epoch": 1.2326369728025226, "grad_norm": 0.5743098258972168, "learning_rate": 0.0005, "loss": 1.8011603355407715, "step": 15636 }, { "epoch": 1.2327158060701615, "grad_norm": 0.5198805928230286, "learning_rate": 0.0005, "loss": 1.6144136190414429, "step": 15637 }, { "epoch": 1.2327946393378006, "grad_norm": 0.4406694173812866, "learning_rate": 0.0005, "loss": 1.5265753269195557, "step": 15638 }, { "epoch": 1.2328734726054396, "grad_norm": 0.5849698781967163, "learning_rate": 0.0005, "loss": 1.5225331783294678, "step": 15639 }, { "epoch": 1.2329523058730785, "grad_norm": 0.5159674286842346, "learning_rate": 0.0005, "loss": 1.7020106315612793, "step": 15640 }, { "epoch": 1.2330311391407174, "grad_norm": 0.8055472373962402, "learning_rate": 0.0005, "loss": 1.7361531257629395, "step": 15641 }, { "epoch": 1.2331099724083563, "grad_norm": 0.5030842423439026, "learning_rate": 0.0005, "loss": 1.7594618797302246, "step": 15642 }, { "epoch": 1.2331888056759952, "grad_norm": 0.6209436058998108, "learning_rate": 0.0005, "loss": 1.5484631061553955, "step": 15643 }, { "epoch": 1.2332676389436341, "grad_norm": 0.4850277304649353, "learning_rate": 0.0005, "loss": 1.7559609413146973, "step": 15644 }, { "epoch": 1.2333464722112732, "grad_norm": 0.4758710563182831, "learning_rate": 0.0005, "loss": 1.558556079864502, "step": 15645 }, { "epoch": 1.2334253054789122, "grad_norm": 0.686992883682251, "learning_rate": 0.0005, "loss": 1.6846774816513062, "step": 15646 }, { "epoch": 1.233504138746551, "grad_norm": 0.47531425952911377, "learning_rate": 0.0005, "loss": 1.6519138813018799, "step": 15647 }, { "epoch": 1.23358297201419, "grad_norm": 0.4911077916622162, "learning_rate": 0.0005, "loss": 1.581405758857727, "step": 15648 }, { "epoch": 1.233661805281829, "grad_norm": 0.48960840702056885, "learning_rate": 0.0005, "loss": 1.770256757736206, "step": 15649 }, { "epoch": 1.2337406385494678, "grad_norm": 0.6038901209831238, "learning_rate": 0.0005, "loss": 1.71484375, "step": 15650 }, { "epoch": 1.2338194718171067, "grad_norm": 0.5989807844161987, "learning_rate": 0.0005, "loss": 1.7087045907974243, "step": 15651 }, { "epoch": 1.2338983050847459, "grad_norm": 0.8696545958518982, "learning_rate": 0.0005, "loss": 1.7474706172943115, "step": 15652 }, { "epoch": 1.2339771383523848, "grad_norm": 0.47677990794181824, "learning_rate": 0.0005, "loss": 1.6083317995071411, "step": 15653 }, { "epoch": 1.2340559716200237, "grad_norm": 0.582740068435669, "learning_rate": 0.0005, "loss": 1.420417070388794, "step": 15654 }, { "epoch": 1.2341348048876626, "grad_norm": 0.5840864181518555, "learning_rate": 0.0005, "loss": 1.7844064235687256, "step": 15655 }, { "epoch": 1.2342136381553015, "grad_norm": 0.6604684591293335, "learning_rate": 0.0005, "loss": 1.7002291679382324, "step": 15656 }, { "epoch": 1.2342924714229404, "grad_norm": 0.8655408620834351, "learning_rate": 0.0005, "loss": 1.7380398511886597, "step": 15657 }, { "epoch": 1.2343713046905793, "grad_norm": 0.4755207598209381, "learning_rate": 0.0005, "loss": 1.6160664558410645, "step": 15658 }, { "epoch": 1.2344501379582185, "grad_norm": 0.6315718293190002, "learning_rate": 0.0005, "loss": 1.6039633750915527, "step": 15659 }, { "epoch": 1.2345289712258574, "grad_norm": 0.5800605416297913, "learning_rate": 0.0005, "loss": 1.5599327087402344, "step": 15660 }, { "epoch": 1.2346078044934963, "grad_norm": 0.7592897415161133, "learning_rate": 0.0005, "loss": 1.5408134460449219, "step": 15661 }, { "epoch": 1.2346866377611352, "grad_norm": 0.5929234623908997, "learning_rate": 0.0005, "loss": 1.8196730613708496, "step": 15662 }, { "epoch": 1.234765471028774, "grad_norm": 0.7244356870651245, "learning_rate": 0.0005, "loss": 1.8454267978668213, "step": 15663 }, { "epoch": 1.234844304296413, "grad_norm": 0.6313714981079102, "learning_rate": 0.0005, "loss": 1.8953083753585815, "step": 15664 }, { "epoch": 1.234923137564052, "grad_norm": 0.4519643187522888, "learning_rate": 0.0005, "loss": 1.5630848407745361, "step": 15665 }, { "epoch": 1.235001970831691, "grad_norm": 0.5317373275756836, "learning_rate": 0.0005, "loss": 1.531064748764038, "step": 15666 }, { "epoch": 1.23508080409933, "grad_norm": 0.5335253477096558, "learning_rate": 0.0005, "loss": 1.6277151107788086, "step": 15667 }, { "epoch": 1.2351596373669689, "grad_norm": 0.45908844470977783, "learning_rate": 0.0005, "loss": 1.6001813411712646, "step": 15668 }, { "epoch": 1.2352384706346078, "grad_norm": 0.6228585243225098, "learning_rate": 0.0005, "loss": 1.587522268295288, "step": 15669 }, { "epoch": 1.2353173039022467, "grad_norm": 0.8319181203842163, "learning_rate": 0.0005, "loss": 1.8191791772842407, "step": 15670 }, { "epoch": 1.2353961371698856, "grad_norm": 0.4793608784675598, "learning_rate": 0.0005, "loss": 1.6287307739257812, "step": 15671 }, { "epoch": 1.2354749704375245, "grad_norm": 0.5364194512367249, "learning_rate": 0.0005, "loss": 1.6346588134765625, "step": 15672 }, { "epoch": 1.2355538037051637, "grad_norm": 2.3170292377471924, "learning_rate": 0.0005, "loss": 1.8171656131744385, "step": 15673 }, { "epoch": 1.2356326369728026, "grad_norm": 0.6763073205947876, "learning_rate": 0.0005, "loss": 1.6306538581848145, "step": 15674 }, { "epoch": 1.2357114702404415, "grad_norm": 0.6310574412345886, "learning_rate": 0.0005, "loss": 1.614259123802185, "step": 15675 }, { "epoch": 1.2357903035080804, "grad_norm": 0.6057565212249756, "learning_rate": 0.0005, "loss": 1.5278229713439941, "step": 15676 }, { "epoch": 1.2358691367757193, "grad_norm": 0.5440430641174316, "learning_rate": 0.0005, "loss": 1.5270235538482666, "step": 15677 }, { "epoch": 1.2359479700433582, "grad_norm": 0.5433061718940735, "learning_rate": 0.0005, "loss": 1.5609465837478638, "step": 15678 }, { "epoch": 1.2360268033109971, "grad_norm": 0.4837104082107544, "learning_rate": 0.0005, "loss": 1.662259578704834, "step": 15679 }, { "epoch": 1.2361056365786363, "grad_norm": 0.5816364884376526, "learning_rate": 0.0005, "loss": 1.8590662479400635, "step": 15680 }, { "epoch": 1.2361844698462752, "grad_norm": 0.5357201099395752, "learning_rate": 0.0005, "loss": 1.4999783039093018, "step": 15681 }, { "epoch": 1.236263303113914, "grad_norm": 0.4741368293762207, "learning_rate": 0.0005, "loss": 1.4958460330963135, "step": 15682 }, { "epoch": 1.236342136381553, "grad_norm": 0.5348310470581055, "learning_rate": 0.0005, "loss": 1.5268884897232056, "step": 15683 }, { "epoch": 1.236420969649192, "grad_norm": 0.6244857311248779, "learning_rate": 0.0005, "loss": 1.5540878772735596, "step": 15684 }, { "epoch": 1.2364998029168308, "grad_norm": 0.5267787575721741, "learning_rate": 0.0005, "loss": 1.4877985715866089, "step": 15685 }, { "epoch": 1.2365786361844697, "grad_norm": 1.3352186679840088, "learning_rate": 0.0005, "loss": 1.7795639038085938, "step": 15686 }, { "epoch": 1.2366574694521089, "grad_norm": 0.4589032530784607, "learning_rate": 0.0005, "loss": 1.5073561668395996, "step": 15687 }, { "epoch": 1.2367363027197478, "grad_norm": 0.5401099920272827, "learning_rate": 0.0005, "loss": 1.8823878765106201, "step": 15688 }, { "epoch": 1.2368151359873867, "grad_norm": 0.7525220513343811, "learning_rate": 0.0005, "loss": 1.8386731147766113, "step": 15689 }, { "epoch": 1.2368939692550256, "grad_norm": 0.488457053899765, "learning_rate": 0.0005, "loss": 1.6574612855911255, "step": 15690 }, { "epoch": 1.2369728025226645, "grad_norm": 0.6279475092887878, "learning_rate": 0.0005, "loss": 1.7582186460494995, "step": 15691 }, { "epoch": 1.2370516357903034, "grad_norm": 0.5236470103263855, "learning_rate": 0.0005, "loss": 1.5536024570465088, "step": 15692 }, { "epoch": 1.2371304690579423, "grad_norm": 0.5660252571105957, "learning_rate": 0.0005, "loss": 1.5966438055038452, "step": 15693 }, { "epoch": 1.2372093023255815, "grad_norm": 0.49802839756011963, "learning_rate": 0.0005, "loss": 1.6238614320755005, "step": 15694 }, { "epoch": 1.2372881355932204, "grad_norm": 0.45097988843917847, "learning_rate": 0.0005, "loss": 1.6049857139587402, "step": 15695 }, { "epoch": 1.2373669688608593, "grad_norm": 0.5234255790710449, "learning_rate": 0.0005, "loss": 1.653451681137085, "step": 15696 }, { "epoch": 1.2374458021284982, "grad_norm": 0.5334762334823608, "learning_rate": 0.0005, "loss": 1.584718942642212, "step": 15697 }, { "epoch": 1.237524635396137, "grad_norm": 0.4691975712776184, "learning_rate": 0.0005, "loss": 1.5804786682128906, "step": 15698 }, { "epoch": 1.237603468663776, "grad_norm": 0.5367788076400757, "learning_rate": 0.0005, "loss": 1.8562922477722168, "step": 15699 }, { "epoch": 1.2376823019314152, "grad_norm": 0.6064496636390686, "learning_rate": 0.0005, "loss": 1.9631751775741577, "step": 15700 }, { "epoch": 1.237761135199054, "grad_norm": 0.46768462657928467, "learning_rate": 0.0005, "loss": 1.4925771951675415, "step": 15701 }, { "epoch": 1.237839968466693, "grad_norm": 1.0809317827224731, "learning_rate": 0.0005, "loss": 1.6611769199371338, "step": 15702 }, { "epoch": 1.2379188017343319, "grad_norm": 0.5762224197387695, "learning_rate": 0.0005, "loss": 1.8059306144714355, "step": 15703 }, { "epoch": 1.2379976350019708, "grad_norm": 0.5099462270736694, "learning_rate": 0.0005, "loss": 1.6720614433288574, "step": 15704 }, { "epoch": 1.2380764682696097, "grad_norm": 0.6660698056221008, "learning_rate": 0.0005, "loss": 1.6778547763824463, "step": 15705 }, { "epoch": 1.2381553015372488, "grad_norm": 0.5709065198898315, "learning_rate": 0.0005, "loss": 1.7553579807281494, "step": 15706 }, { "epoch": 1.2382341348048878, "grad_norm": 0.5709912776947021, "learning_rate": 0.0005, "loss": 1.6387213468551636, "step": 15707 }, { "epoch": 1.2383129680725267, "grad_norm": 0.5474280714988708, "learning_rate": 0.0005, "loss": 1.5882644653320312, "step": 15708 }, { "epoch": 1.2383918013401656, "grad_norm": 0.530032753944397, "learning_rate": 0.0005, "loss": 1.5345982313156128, "step": 15709 }, { "epoch": 1.2384706346078045, "grad_norm": 0.6586294770240784, "learning_rate": 0.0005, "loss": 1.737717628479004, "step": 15710 }, { "epoch": 1.2385494678754434, "grad_norm": 0.5206442475318909, "learning_rate": 0.0005, "loss": 1.521625280380249, "step": 15711 }, { "epoch": 1.2386283011430823, "grad_norm": 0.5307389497756958, "learning_rate": 0.0005, "loss": 1.5162250995635986, "step": 15712 }, { "epoch": 1.2387071344107214, "grad_norm": 0.5048028826713562, "learning_rate": 0.0005, "loss": 1.6701934337615967, "step": 15713 }, { "epoch": 1.2387859676783604, "grad_norm": 0.506649911403656, "learning_rate": 0.0005, "loss": 1.6605499982833862, "step": 15714 }, { "epoch": 1.2388648009459993, "grad_norm": 0.39846187829971313, "learning_rate": 0.0005, "loss": 1.5850558280944824, "step": 15715 }, { "epoch": 1.2389436342136382, "grad_norm": 0.4476587474346161, "learning_rate": 0.0005, "loss": 1.4489567279815674, "step": 15716 }, { "epoch": 1.239022467481277, "grad_norm": 0.5020514726638794, "learning_rate": 0.0005, "loss": 1.5509432554244995, "step": 15717 }, { "epoch": 1.239101300748916, "grad_norm": 0.5010282397270203, "learning_rate": 0.0005, "loss": 1.645681619644165, "step": 15718 }, { "epoch": 1.239180134016555, "grad_norm": 0.522217333316803, "learning_rate": 0.0005, "loss": 1.638647198677063, "step": 15719 }, { "epoch": 1.239258967284194, "grad_norm": 0.6079864501953125, "learning_rate": 0.0005, "loss": 1.8505218029022217, "step": 15720 }, { "epoch": 1.239337800551833, "grad_norm": 0.560928463935852, "learning_rate": 0.0005, "loss": 1.5735212564468384, "step": 15721 }, { "epoch": 1.2394166338194719, "grad_norm": 0.5645549297332764, "learning_rate": 0.0005, "loss": 1.7470136880874634, "step": 15722 }, { "epoch": 1.2394954670871108, "grad_norm": 0.5536320805549622, "learning_rate": 0.0005, "loss": 1.6404080390930176, "step": 15723 }, { "epoch": 1.2395743003547497, "grad_norm": 0.4689537584781647, "learning_rate": 0.0005, "loss": 1.6273257732391357, "step": 15724 }, { "epoch": 1.2396531336223886, "grad_norm": 0.3814515173435211, "learning_rate": 0.0005, "loss": 1.32160222530365, "step": 15725 }, { "epoch": 1.2397319668900275, "grad_norm": 0.5062263607978821, "learning_rate": 0.0005, "loss": 1.551652431488037, "step": 15726 }, { "epoch": 1.2398108001576666, "grad_norm": 1.1014376878738403, "learning_rate": 0.0005, "loss": 1.7063935995101929, "step": 15727 }, { "epoch": 1.2398896334253056, "grad_norm": 0.46435967087745667, "learning_rate": 0.0005, "loss": 1.6036584377288818, "step": 15728 }, { "epoch": 1.2399684666929445, "grad_norm": 0.530300498008728, "learning_rate": 0.0005, "loss": 1.7203538417816162, "step": 15729 }, { "epoch": 1.2400472999605834, "grad_norm": 0.546351969242096, "learning_rate": 0.0005, "loss": 1.5351903438568115, "step": 15730 }, { "epoch": 1.2401261332282223, "grad_norm": 0.4778505265712738, "learning_rate": 0.0005, "loss": 1.737020492553711, "step": 15731 }, { "epoch": 1.2402049664958612, "grad_norm": 0.62996906042099, "learning_rate": 0.0005, "loss": 1.8733991384506226, "step": 15732 }, { "epoch": 1.2402837997635001, "grad_norm": 0.5099615454673767, "learning_rate": 0.0005, "loss": 1.4968290328979492, "step": 15733 }, { "epoch": 1.2403626330311393, "grad_norm": 0.47336751222610474, "learning_rate": 0.0005, "loss": 1.693542718887329, "step": 15734 }, { "epoch": 1.2404414662987782, "grad_norm": 0.46949347853660583, "learning_rate": 0.0005, "loss": 1.7200077772140503, "step": 15735 }, { "epoch": 1.240520299566417, "grad_norm": 0.6305859088897705, "learning_rate": 0.0005, "loss": 1.7290525436401367, "step": 15736 }, { "epoch": 1.240599132834056, "grad_norm": 0.4021303355693817, "learning_rate": 0.0005, "loss": 1.4711476564407349, "step": 15737 }, { "epoch": 1.240677966101695, "grad_norm": 0.5444663166999817, "learning_rate": 0.0005, "loss": 1.7879836559295654, "step": 15738 }, { "epoch": 1.2407567993693338, "grad_norm": 0.5785897374153137, "learning_rate": 0.0005, "loss": 1.6358225345611572, "step": 15739 }, { "epoch": 1.2408356326369727, "grad_norm": 0.60039222240448, "learning_rate": 0.0005, "loss": 1.610013484954834, "step": 15740 }, { "epoch": 1.2409144659046119, "grad_norm": 0.5100374817848206, "learning_rate": 0.0005, "loss": 1.5968135595321655, "step": 15741 }, { "epoch": 1.2409932991722508, "grad_norm": 0.48160016536712646, "learning_rate": 0.0005, "loss": 1.653388500213623, "step": 15742 }, { "epoch": 1.2410721324398897, "grad_norm": 0.8604482412338257, "learning_rate": 0.0005, "loss": 1.8353981971740723, "step": 15743 }, { "epoch": 1.2411509657075286, "grad_norm": 0.46424612402915955, "learning_rate": 0.0005, "loss": 1.5857925415039062, "step": 15744 }, { "epoch": 1.2412297989751675, "grad_norm": 1.0053662061691284, "learning_rate": 0.0005, "loss": 1.6571605205535889, "step": 15745 }, { "epoch": 1.2413086322428064, "grad_norm": 0.46649169921875, "learning_rate": 0.0005, "loss": 1.70134699344635, "step": 15746 }, { "epoch": 1.2413874655104453, "grad_norm": 0.4207659661769867, "learning_rate": 0.0005, "loss": 1.5943922996520996, "step": 15747 }, { "epoch": 1.2414662987780845, "grad_norm": 0.4760202169418335, "learning_rate": 0.0005, "loss": 1.4869859218597412, "step": 15748 }, { "epoch": 1.2415451320457234, "grad_norm": 0.6558659076690674, "learning_rate": 0.0005, "loss": 1.7831768989562988, "step": 15749 }, { "epoch": 1.2416239653133623, "grad_norm": 0.6010317206382751, "learning_rate": 0.0005, "loss": 1.7758394479751587, "step": 15750 }, { "epoch": 1.2417027985810012, "grad_norm": 0.6765087842941284, "learning_rate": 0.0005, "loss": 1.5983386039733887, "step": 15751 }, { "epoch": 1.24178163184864, "grad_norm": 0.8458071351051331, "learning_rate": 0.0005, "loss": 1.8077958822250366, "step": 15752 }, { "epoch": 1.241860465116279, "grad_norm": 0.5755757689476013, "learning_rate": 0.0005, "loss": 1.5749671459197998, "step": 15753 }, { "epoch": 1.241939298383918, "grad_norm": 0.5907924771308899, "learning_rate": 0.0005, "loss": 1.8248999118804932, "step": 15754 }, { "epoch": 1.242018131651557, "grad_norm": 0.8665474057197571, "learning_rate": 0.0005, "loss": 2.0054824352264404, "step": 15755 }, { "epoch": 1.242096964919196, "grad_norm": 1.0395965576171875, "learning_rate": 0.0005, "loss": 1.79360032081604, "step": 15756 }, { "epoch": 1.2421757981868349, "grad_norm": 0.6628696322441101, "learning_rate": 0.0005, "loss": 1.826930284500122, "step": 15757 }, { "epoch": 1.2422546314544738, "grad_norm": 0.44609928131103516, "learning_rate": 0.0005, "loss": 1.557183027267456, "step": 15758 }, { "epoch": 1.2423334647221127, "grad_norm": 0.6485400795936584, "learning_rate": 0.0005, "loss": 1.709204912185669, "step": 15759 }, { "epoch": 1.2424122979897516, "grad_norm": 0.5092734694480896, "learning_rate": 0.0005, "loss": 1.6448934078216553, "step": 15760 }, { "epoch": 1.2424911312573905, "grad_norm": 0.5376051068305969, "learning_rate": 0.0005, "loss": 1.6923587322235107, "step": 15761 }, { "epoch": 1.2425699645250297, "grad_norm": 0.675588846206665, "learning_rate": 0.0005, "loss": 1.802890658378601, "step": 15762 }, { "epoch": 1.2426487977926686, "grad_norm": 0.7445474863052368, "learning_rate": 0.0005, "loss": 1.625887155532837, "step": 15763 }, { "epoch": 1.2427276310603075, "grad_norm": 0.45432811975479126, "learning_rate": 0.0005, "loss": 1.500600814819336, "step": 15764 }, { "epoch": 1.2428064643279464, "grad_norm": 0.44421273469924927, "learning_rate": 0.0005, "loss": 1.3190923929214478, "step": 15765 }, { "epoch": 1.2428852975955853, "grad_norm": 0.457021564245224, "learning_rate": 0.0005, "loss": 1.3701927661895752, "step": 15766 }, { "epoch": 1.2429641308632242, "grad_norm": 0.5659327507019043, "learning_rate": 0.0005, "loss": 1.4880452156066895, "step": 15767 }, { "epoch": 1.2430429641308631, "grad_norm": 0.5579295754432678, "learning_rate": 0.0005, "loss": 1.5007811784744263, "step": 15768 }, { "epoch": 1.2431217973985023, "grad_norm": 0.5427930355072021, "learning_rate": 0.0005, "loss": 1.6558208465576172, "step": 15769 }, { "epoch": 1.2432006306661412, "grad_norm": 0.5210928916931152, "learning_rate": 0.0005, "loss": 1.8763434886932373, "step": 15770 }, { "epoch": 1.24327946393378, "grad_norm": 0.650664746761322, "learning_rate": 0.0005, "loss": 1.8661508560180664, "step": 15771 }, { "epoch": 1.243358297201419, "grad_norm": 0.5491171479225159, "learning_rate": 0.0005, "loss": 1.9140970706939697, "step": 15772 }, { "epoch": 1.243437130469058, "grad_norm": 0.7111063003540039, "learning_rate": 0.0005, "loss": 1.7507750988006592, "step": 15773 }, { "epoch": 1.2435159637366968, "grad_norm": 1.0750478506088257, "learning_rate": 0.0005, "loss": 2.104679584503174, "step": 15774 }, { "epoch": 1.2435947970043357, "grad_norm": 0.49621015787124634, "learning_rate": 0.0005, "loss": 1.641198992729187, "step": 15775 }, { "epoch": 1.2436736302719749, "grad_norm": 0.9255581498146057, "learning_rate": 0.0005, "loss": 1.676773190498352, "step": 15776 }, { "epoch": 1.2437524635396138, "grad_norm": 1.0188206434249878, "learning_rate": 0.0005, "loss": 1.853655219078064, "step": 15777 }, { "epoch": 1.2438312968072527, "grad_norm": 0.6110467314720154, "learning_rate": 0.0005, "loss": 1.7289636135101318, "step": 15778 }, { "epoch": 1.2439101300748916, "grad_norm": 0.7685739398002625, "learning_rate": 0.0005, "loss": 1.554271936416626, "step": 15779 }, { "epoch": 1.2439889633425305, "grad_norm": 0.49243074655532837, "learning_rate": 0.0005, "loss": 1.7753922939300537, "step": 15780 }, { "epoch": 1.2440677966101694, "grad_norm": 0.6043607592582703, "learning_rate": 0.0005, "loss": 1.659911870956421, "step": 15781 }, { "epoch": 1.2441466298778083, "grad_norm": 0.5208516716957092, "learning_rate": 0.0005, "loss": 1.6827083826065063, "step": 15782 }, { "epoch": 1.2442254631454475, "grad_norm": 1.0420341491699219, "learning_rate": 0.0005, "loss": 1.7497987747192383, "step": 15783 }, { "epoch": 1.2443042964130864, "grad_norm": 0.5341644287109375, "learning_rate": 0.0005, "loss": 1.6344980001449585, "step": 15784 }, { "epoch": 1.2443831296807253, "grad_norm": 0.7037890553474426, "learning_rate": 0.0005, "loss": 1.7724114656448364, "step": 15785 }, { "epoch": 1.2444619629483642, "grad_norm": 0.494212806224823, "learning_rate": 0.0005, "loss": 1.6776586771011353, "step": 15786 }, { "epoch": 1.244540796216003, "grad_norm": 0.5871871113777161, "learning_rate": 0.0005, "loss": 1.7126353979110718, "step": 15787 }, { "epoch": 1.244619629483642, "grad_norm": 0.4197465181350708, "learning_rate": 0.0005, "loss": 1.5373823642730713, "step": 15788 }, { "epoch": 1.244698462751281, "grad_norm": 0.46317431330680847, "learning_rate": 0.0005, "loss": 1.450871467590332, "step": 15789 }, { "epoch": 1.24477729601892, "grad_norm": 0.5485199093818665, "learning_rate": 0.0005, "loss": 1.6064015626907349, "step": 15790 }, { "epoch": 1.244856129286559, "grad_norm": 0.6950797438621521, "learning_rate": 0.0005, "loss": 1.7165828943252563, "step": 15791 }, { "epoch": 1.2449349625541979, "grad_norm": 0.49704328179359436, "learning_rate": 0.0005, "loss": 1.6928410530090332, "step": 15792 }, { "epoch": 1.2450137958218368, "grad_norm": 0.68964022397995, "learning_rate": 0.0005, "loss": 1.800309181213379, "step": 15793 }, { "epoch": 1.2450926290894757, "grad_norm": 0.519105076789856, "learning_rate": 0.0005, "loss": 1.5611456632614136, "step": 15794 }, { "epoch": 1.2451714623571146, "grad_norm": 0.49811840057373047, "learning_rate": 0.0005, "loss": 1.8288991451263428, "step": 15795 }, { "epoch": 1.2452502956247535, "grad_norm": 0.4890497624874115, "learning_rate": 0.0005, "loss": 1.5926704406738281, "step": 15796 }, { "epoch": 1.2453291288923927, "grad_norm": 0.6393155455589294, "learning_rate": 0.0005, "loss": 1.6912353038787842, "step": 15797 }, { "epoch": 1.2454079621600316, "grad_norm": 0.5732468962669373, "learning_rate": 0.0005, "loss": 1.6454200744628906, "step": 15798 }, { "epoch": 1.2454867954276705, "grad_norm": 0.6000556945800781, "learning_rate": 0.0005, "loss": 1.7229256629943848, "step": 15799 }, { "epoch": 1.2455656286953094, "grad_norm": 0.6099276542663574, "learning_rate": 0.0005, "loss": 1.4638007879257202, "step": 15800 }, { "epoch": 1.2456444619629483, "grad_norm": 0.4792404770851135, "learning_rate": 0.0005, "loss": 1.6050443649291992, "step": 15801 }, { "epoch": 1.2457232952305872, "grad_norm": 0.5431691408157349, "learning_rate": 0.0005, "loss": 1.7700775861740112, "step": 15802 }, { "epoch": 1.2458021284982261, "grad_norm": 0.5503525733947754, "learning_rate": 0.0005, "loss": 1.6034713983535767, "step": 15803 }, { "epoch": 1.2458809617658653, "grad_norm": 0.550026535987854, "learning_rate": 0.0005, "loss": 1.6676881313323975, "step": 15804 }, { "epoch": 1.2459597950335042, "grad_norm": 0.5562528967857361, "learning_rate": 0.0005, "loss": 1.5357145071029663, "step": 15805 }, { "epoch": 1.246038628301143, "grad_norm": 0.4910193383693695, "learning_rate": 0.0005, "loss": 1.6389133930206299, "step": 15806 }, { "epoch": 1.246117461568782, "grad_norm": 0.5801909565925598, "learning_rate": 0.0005, "loss": 1.4936615228652954, "step": 15807 }, { "epoch": 1.246196294836421, "grad_norm": 0.5893033146858215, "learning_rate": 0.0005, "loss": 1.8513445854187012, "step": 15808 }, { "epoch": 1.2462751281040598, "grad_norm": 0.8606426119804382, "learning_rate": 0.0005, "loss": 1.5666768550872803, "step": 15809 }, { "epoch": 1.2463539613716987, "grad_norm": 0.517105221748352, "learning_rate": 0.0005, "loss": 1.741908073425293, "step": 15810 }, { "epoch": 1.2464327946393379, "grad_norm": 0.5001320242881775, "learning_rate": 0.0005, "loss": 1.694293737411499, "step": 15811 }, { "epoch": 1.2465116279069768, "grad_norm": 0.6311666369438171, "learning_rate": 0.0005, "loss": 1.6525650024414062, "step": 15812 }, { "epoch": 1.2465904611746157, "grad_norm": 0.6347431540489197, "learning_rate": 0.0005, "loss": 1.6245052814483643, "step": 15813 }, { "epoch": 1.2466692944422546, "grad_norm": 0.6253892183303833, "learning_rate": 0.0005, "loss": 1.7179292440414429, "step": 15814 }, { "epoch": 1.2467481277098935, "grad_norm": 0.7195687890052795, "learning_rate": 0.0005, "loss": 1.5424996614456177, "step": 15815 }, { "epoch": 1.2468269609775324, "grad_norm": 0.5339179635047913, "learning_rate": 0.0005, "loss": 1.7703278064727783, "step": 15816 }, { "epoch": 1.2469057942451713, "grad_norm": 0.5369883179664612, "learning_rate": 0.0005, "loss": 1.544820785522461, "step": 15817 }, { "epoch": 1.2469846275128105, "grad_norm": 0.5706118941307068, "learning_rate": 0.0005, "loss": 1.518175721168518, "step": 15818 }, { "epoch": 1.2470634607804494, "grad_norm": 0.4828546643257141, "learning_rate": 0.0005, "loss": 1.5781934261322021, "step": 15819 }, { "epoch": 1.2471422940480883, "grad_norm": 0.5460571646690369, "learning_rate": 0.0005, "loss": 1.5828741788864136, "step": 15820 }, { "epoch": 1.2472211273157272, "grad_norm": 0.596621036529541, "learning_rate": 0.0005, "loss": 1.50253164768219, "step": 15821 }, { "epoch": 1.2472999605833661, "grad_norm": 0.4887222349643707, "learning_rate": 0.0005, "loss": 1.7129868268966675, "step": 15822 }, { "epoch": 1.247378793851005, "grad_norm": 0.5783732533454895, "learning_rate": 0.0005, "loss": 2.0649232864379883, "step": 15823 }, { "epoch": 1.2474576271186442, "grad_norm": 0.5802816152572632, "learning_rate": 0.0005, "loss": 1.811110258102417, "step": 15824 }, { "epoch": 1.247536460386283, "grad_norm": 0.5292932987213135, "learning_rate": 0.0005, "loss": 1.9388480186462402, "step": 15825 }, { "epoch": 1.247615293653922, "grad_norm": 0.6391376852989197, "learning_rate": 0.0005, "loss": 1.854636788368225, "step": 15826 }, { "epoch": 1.247694126921561, "grad_norm": 0.602183997631073, "learning_rate": 0.0005, "loss": 1.8835117816925049, "step": 15827 }, { "epoch": 1.2477729601891998, "grad_norm": 0.8226170539855957, "learning_rate": 0.0005, "loss": 1.7000218629837036, "step": 15828 }, { "epoch": 1.2478517934568387, "grad_norm": 0.5668759942054749, "learning_rate": 0.0005, "loss": 1.7856364250183105, "step": 15829 }, { "epoch": 1.2479306267244779, "grad_norm": 0.5081961750984192, "learning_rate": 0.0005, "loss": 1.4540352821350098, "step": 15830 }, { "epoch": 1.2480094599921168, "grad_norm": 0.5363002419471741, "learning_rate": 0.0005, "loss": 1.745004415512085, "step": 15831 }, { "epoch": 1.2480882932597557, "grad_norm": 0.48051685094833374, "learning_rate": 0.0005, "loss": 1.5651004314422607, "step": 15832 }, { "epoch": 1.2481671265273946, "grad_norm": 0.667708158493042, "learning_rate": 0.0005, "loss": 1.5682944059371948, "step": 15833 }, { "epoch": 1.2482459597950335, "grad_norm": 0.5282391309738159, "learning_rate": 0.0005, "loss": 1.769182562828064, "step": 15834 }, { "epoch": 1.2483247930626724, "grad_norm": 0.5486899614334106, "learning_rate": 0.0005, "loss": 1.685738444328308, "step": 15835 }, { "epoch": 1.2484036263303113, "grad_norm": 0.47276315093040466, "learning_rate": 0.0005, "loss": 1.3460071086883545, "step": 15836 }, { "epoch": 1.2484824595979505, "grad_norm": 0.508560836315155, "learning_rate": 0.0005, "loss": 1.6855214834213257, "step": 15837 }, { "epoch": 1.2485612928655894, "grad_norm": 0.6816402077674866, "learning_rate": 0.0005, "loss": 1.9044216871261597, "step": 15838 }, { "epoch": 1.2486401261332283, "grad_norm": 0.5623419880867004, "learning_rate": 0.0005, "loss": 1.6790924072265625, "step": 15839 }, { "epoch": 1.2487189594008672, "grad_norm": 0.4737342298030853, "learning_rate": 0.0005, "loss": 1.644406795501709, "step": 15840 }, { "epoch": 1.248797792668506, "grad_norm": 0.7018925547599792, "learning_rate": 0.0005, "loss": 1.6389548778533936, "step": 15841 }, { "epoch": 1.248876625936145, "grad_norm": 0.5942711234092712, "learning_rate": 0.0005, "loss": 1.6048228740692139, "step": 15842 }, { "epoch": 1.248955459203784, "grad_norm": 0.4270505905151367, "learning_rate": 0.0005, "loss": 1.4471595287322998, "step": 15843 }, { "epoch": 1.249034292471423, "grad_norm": 0.6403487920761108, "learning_rate": 0.0005, "loss": 1.611452579498291, "step": 15844 }, { "epoch": 1.249113125739062, "grad_norm": 0.5279291272163391, "learning_rate": 0.0005, "loss": 1.527353048324585, "step": 15845 }, { "epoch": 1.2491919590067009, "grad_norm": 0.47053515911102295, "learning_rate": 0.0005, "loss": 1.7468607425689697, "step": 15846 }, { "epoch": 1.2492707922743398, "grad_norm": 0.8135877251625061, "learning_rate": 0.0005, "loss": 1.6505073308944702, "step": 15847 }, { "epoch": 1.2493496255419787, "grad_norm": 1.3120591640472412, "learning_rate": 0.0005, "loss": 1.8311409950256348, "step": 15848 }, { "epoch": 1.2494284588096176, "grad_norm": 0.5109254717826843, "learning_rate": 0.0005, "loss": 1.6125149726867676, "step": 15849 }, { "epoch": 1.2495072920772565, "grad_norm": 0.5159115791320801, "learning_rate": 0.0005, "loss": 1.5747251510620117, "step": 15850 }, { "epoch": 1.2495861253448957, "grad_norm": 0.6175613403320312, "learning_rate": 0.0005, "loss": 1.6815178394317627, "step": 15851 }, { "epoch": 1.2496649586125346, "grad_norm": 0.5489767789840698, "learning_rate": 0.0005, "loss": 1.8422937393188477, "step": 15852 }, { "epoch": 1.2497437918801735, "grad_norm": 0.49882155656814575, "learning_rate": 0.0005, "loss": 1.3557826280593872, "step": 15853 }, { "epoch": 1.2498226251478124, "grad_norm": 0.6604541540145874, "learning_rate": 0.0005, "loss": 1.658585548400879, "step": 15854 }, { "epoch": 1.2499014584154513, "grad_norm": 0.4979846775531769, "learning_rate": 0.0005, "loss": 1.6734769344329834, "step": 15855 }, { "epoch": 1.2499802916830902, "grad_norm": 0.8441175222396851, "learning_rate": 0.0005, "loss": 1.656220555305481, "step": 15856 }, { "epoch": 1.2500591249507291, "grad_norm": 0.46963390707969666, "learning_rate": 0.0005, "loss": 1.7489521503448486, "step": 15857 }, { "epoch": 1.2501379582183683, "grad_norm": 0.4838217794895172, "learning_rate": 0.0005, "loss": 1.581945538520813, "step": 15858 }, { "epoch": 1.2502167914860072, "grad_norm": 0.6810618042945862, "learning_rate": 0.0005, "loss": 1.9144078493118286, "step": 15859 }, { "epoch": 1.250295624753646, "grad_norm": 0.7084276676177979, "learning_rate": 0.0005, "loss": 1.7565884590148926, "step": 15860 }, { "epoch": 1.250374458021285, "grad_norm": 0.5300784111022949, "learning_rate": 0.0005, "loss": 1.939124345779419, "step": 15861 }, { "epoch": 1.250453291288924, "grad_norm": 0.6259430050849915, "learning_rate": 0.0005, "loss": 1.870566725730896, "step": 15862 }, { "epoch": 1.2505321245565628, "grad_norm": 0.5269721746444702, "learning_rate": 0.0005, "loss": 1.5061696767807007, "step": 15863 }, { "epoch": 1.2506109578242017, "grad_norm": 0.5436665415763855, "learning_rate": 0.0005, "loss": 1.5913219451904297, "step": 15864 }, { "epoch": 1.2506897910918409, "grad_norm": 0.5858231782913208, "learning_rate": 0.0005, "loss": 1.5407774448394775, "step": 15865 }, { "epoch": 1.2507686243594798, "grad_norm": 0.6590050458908081, "learning_rate": 0.0005, "loss": 1.7329976558685303, "step": 15866 }, { "epoch": 1.2508474576271187, "grad_norm": 0.5244808197021484, "learning_rate": 0.0005, "loss": 1.4636421203613281, "step": 15867 }, { "epoch": 1.2509262908947576, "grad_norm": 0.4805212914943695, "learning_rate": 0.0005, "loss": 1.5822975635528564, "step": 15868 }, { "epoch": 1.2510051241623965, "grad_norm": 0.5111709833145142, "learning_rate": 0.0005, "loss": 1.6128075122833252, "step": 15869 }, { "epoch": 1.2510839574300354, "grad_norm": 0.5748624205589294, "learning_rate": 0.0005, "loss": 1.6224164962768555, "step": 15870 }, { "epoch": 1.2511627906976743, "grad_norm": 0.529780387878418, "learning_rate": 0.0005, "loss": 1.7203631401062012, "step": 15871 }, { "epoch": 1.2512416239653135, "grad_norm": 0.501426100730896, "learning_rate": 0.0005, "loss": 1.7701345682144165, "step": 15872 }, { "epoch": 1.2513204572329524, "grad_norm": 0.5185605883598328, "learning_rate": 0.0005, "loss": 1.6902403831481934, "step": 15873 }, { "epoch": 1.2513992905005913, "grad_norm": 0.4461478590965271, "learning_rate": 0.0005, "loss": 1.6676976680755615, "step": 15874 }, { "epoch": 1.2514781237682302, "grad_norm": 0.39233681559562683, "learning_rate": 0.0005, "loss": 1.4355723857879639, "step": 15875 }, { "epoch": 1.251556957035869, "grad_norm": 0.5208204984664917, "learning_rate": 0.0005, "loss": 1.5400733947753906, "step": 15876 }, { "epoch": 1.251635790303508, "grad_norm": 0.6098079085350037, "learning_rate": 0.0005, "loss": 1.8319635391235352, "step": 15877 }, { "epoch": 1.251714623571147, "grad_norm": 0.4353399872779846, "learning_rate": 0.0005, "loss": 1.635209560394287, "step": 15878 }, { "epoch": 1.251793456838786, "grad_norm": 0.4966816306114197, "learning_rate": 0.0005, "loss": 1.4220221042633057, "step": 15879 }, { "epoch": 1.251872290106425, "grad_norm": 0.43533003330230713, "learning_rate": 0.0005, "loss": 1.5887081623077393, "step": 15880 }, { "epoch": 1.251951123374064, "grad_norm": 0.4317392408847809, "learning_rate": 0.0005, "loss": 1.6379879713058472, "step": 15881 }, { "epoch": 1.2520299566417028, "grad_norm": 0.6816631555557251, "learning_rate": 0.0005, "loss": 2.0190563201904297, "step": 15882 }, { "epoch": 1.2521087899093417, "grad_norm": 0.5779871940612793, "learning_rate": 0.0005, "loss": 1.6838717460632324, "step": 15883 }, { "epoch": 1.2521876231769806, "grad_norm": 0.48079776763916016, "learning_rate": 0.0005, "loss": 1.7171491384506226, "step": 15884 }, { "epoch": 1.2522664564446195, "grad_norm": 0.4964282512664795, "learning_rate": 0.0005, "loss": 1.6911661624908447, "step": 15885 }, { "epoch": 1.2523452897122587, "grad_norm": 0.5084666013717651, "learning_rate": 0.0005, "loss": 1.8001279830932617, "step": 15886 }, { "epoch": 1.2524241229798976, "grad_norm": 0.48177802562713623, "learning_rate": 0.0005, "loss": 1.701951503753662, "step": 15887 }, { "epoch": 1.2525029562475365, "grad_norm": 0.44445571303367615, "learning_rate": 0.0005, "loss": 1.5405421257019043, "step": 15888 }, { "epoch": 1.2525817895151754, "grad_norm": 0.5114111304283142, "learning_rate": 0.0005, "loss": 1.7051911354064941, "step": 15889 }, { "epoch": 1.2526606227828143, "grad_norm": 0.5304768085479736, "learning_rate": 0.0005, "loss": 1.7556744813919067, "step": 15890 }, { "epoch": 1.2527394560504532, "grad_norm": 0.6324511766433716, "learning_rate": 0.0005, "loss": 1.5899195671081543, "step": 15891 }, { "epoch": 1.2528182893180921, "grad_norm": 0.49719831347465515, "learning_rate": 0.0005, "loss": 1.722327709197998, "step": 15892 }, { "epoch": 1.2528971225857313, "grad_norm": 0.5323262810707092, "learning_rate": 0.0005, "loss": 1.5498085021972656, "step": 15893 }, { "epoch": 1.2529759558533702, "grad_norm": 0.5839340686798096, "learning_rate": 0.0005, "loss": 1.7517640590667725, "step": 15894 }, { "epoch": 1.253054789121009, "grad_norm": 0.4127492308616638, "learning_rate": 0.0005, "loss": 1.435644268989563, "step": 15895 }, { "epoch": 1.253133622388648, "grad_norm": 0.5466283559799194, "learning_rate": 0.0005, "loss": 1.7365208864212036, "step": 15896 }, { "epoch": 1.253212455656287, "grad_norm": 0.4905604124069214, "learning_rate": 0.0005, "loss": 1.6622706651687622, "step": 15897 }, { "epoch": 1.2532912889239258, "grad_norm": 0.6670861840248108, "learning_rate": 0.0005, "loss": 1.7980072498321533, "step": 15898 }, { "epoch": 1.2533701221915647, "grad_norm": 0.5381441712379456, "learning_rate": 0.0005, "loss": 1.5819532871246338, "step": 15899 }, { "epoch": 1.2534489554592039, "grad_norm": 0.5222118496894836, "learning_rate": 0.0005, "loss": 1.5404000282287598, "step": 15900 }, { "epoch": 1.2535277887268428, "grad_norm": 0.5390132069587708, "learning_rate": 0.0005, "loss": 1.6550958156585693, "step": 15901 }, { "epoch": 1.2536066219944817, "grad_norm": 0.48541000485420227, "learning_rate": 0.0005, "loss": 1.694891333580017, "step": 15902 }, { "epoch": 1.2536854552621206, "grad_norm": 0.5702087879180908, "learning_rate": 0.0005, "loss": 1.8211379051208496, "step": 15903 }, { "epoch": 1.2537642885297595, "grad_norm": 0.5121960639953613, "learning_rate": 0.0005, "loss": 1.5787075757980347, "step": 15904 }, { "epoch": 1.2538431217973984, "grad_norm": 0.5365370512008667, "learning_rate": 0.0005, "loss": 1.7102190256118774, "step": 15905 }, { "epoch": 1.2539219550650373, "grad_norm": 0.5384899973869324, "learning_rate": 0.0005, "loss": 1.778303623199463, "step": 15906 }, { "epoch": 1.2540007883326765, "grad_norm": 0.46527108550071716, "learning_rate": 0.0005, "loss": 1.6189641952514648, "step": 15907 }, { "epoch": 1.2540796216003154, "grad_norm": 0.619164764881134, "learning_rate": 0.0005, "loss": 1.6871899366378784, "step": 15908 }, { "epoch": 1.2541584548679543, "grad_norm": 0.5323622822761536, "learning_rate": 0.0005, "loss": 1.7138490676879883, "step": 15909 }, { "epoch": 1.2542372881355932, "grad_norm": 0.5366505980491638, "learning_rate": 0.0005, "loss": 1.830230712890625, "step": 15910 }, { "epoch": 1.2543161214032321, "grad_norm": 0.5990465879440308, "learning_rate": 0.0005, "loss": 1.6888420581817627, "step": 15911 }, { "epoch": 1.254394954670871, "grad_norm": 0.45762526988983154, "learning_rate": 0.0005, "loss": 1.6777831315994263, "step": 15912 }, { "epoch": 1.25447378793851, "grad_norm": 0.761273980140686, "learning_rate": 0.0005, "loss": 1.9296964406967163, "step": 15913 }, { "epoch": 1.254552621206149, "grad_norm": 0.4991605579853058, "learning_rate": 0.0005, "loss": 1.6315524578094482, "step": 15914 }, { "epoch": 1.254631454473788, "grad_norm": 0.6811697483062744, "learning_rate": 0.0005, "loss": 1.8742811679840088, "step": 15915 }, { "epoch": 1.254710287741427, "grad_norm": 0.4650741517543793, "learning_rate": 0.0005, "loss": 1.4158129692077637, "step": 15916 }, { "epoch": 1.2547891210090658, "grad_norm": 0.6655911207199097, "learning_rate": 0.0005, "loss": 1.6859493255615234, "step": 15917 }, { "epoch": 1.2548679542767047, "grad_norm": 0.6081767082214355, "learning_rate": 0.0005, "loss": 1.7128242254257202, "step": 15918 }, { "epoch": 1.2549467875443436, "grad_norm": 0.6719064712524414, "learning_rate": 0.0005, "loss": 1.537210464477539, "step": 15919 }, { "epoch": 1.2550256208119825, "grad_norm": 0.46322324872016907, "learning_rate": 0.0005, "loss": 1.545272946357727, "step": 15920 }, { "epoch": 1.2551044540796217, "grad_norm": 0.6618010401725769, "learning_rate": 0.0005, "loss": 1.7662442922592163, "step": 15921 }, { "epoch": 1.2551832873472606, "grad_norm": 0.5900012850761414, "learning_rate": 0.0005, "loss": 1.3906521797180176, "step": 15922 }, { "epoch": 1.2552621206148995, "grad_norm": 0.49009838700294495, "learning_rate": 0.0005, "loss": 1.6203885078430176, "step": 15923 }, { "epoch": 1.2553409538825384, "grad_norm": 0.622157096862793, "learning_rate": 0.0005, "loss": 1.6945043802261353, "step": 15924 }, { "epoch": 1.2554197871501773, "grad_norm": 0.6957057118415833, "learning_rate": 0.0005, "loss": 1.8268659114837646, "step": 15925 }, { "epoch": 1.2554986204178162, "grad_norm": 0.60182785987854, "learning_rate": 0.0005, "loss": 1.7159514427185059, "step": 15926 }, { "epoch": 1.2555774536854551, "grad_norm": 0.5538005232810974, "learning_rate": 0.0005, "loss": 1.663076639175415, "step": 15927 }, { "epoch": 1.2556562869530943, "grad_norm": 0.5032464861869812, "learning_rate": 0.0005, "loss": 1.7517470121383667, "step": 15928 }, { "epoch": 1.2557351202207332, "grad_norm": 0.5306207537651062, "learning_rate": 0.0005, "loss": 1.438770055770874, "step": 15929 }, { "epoch": 1.255813953488372, "grad_norm": 0.6609837412834167, "learning_rate": 0.0005, "loss": 1.67897629737854, "step": 15930 }, { "epoch": 1.255892786756011, "grad_norm": 0.57890784740448, "learning_rate": 0.0005, "loss": 1.6705302000045776, "step": 15931 }, { "epoch": 1.25597162002365, "grad_norm": 0.5383132100105286, "learning_rate": 0.0005, "loss": 1.666025161743164, "step": 15932 }, { "epoch": 1.256050453291289, "grad_norm": 0.5277935266494751, "learning_rate": 0.0005, "loss": 1.6483020782470703, "step": 15933 }, { "epoch": 1.2561292865589277, "grad_norm": 0.5629367828369141, "learning_rate": 0.0005, "loss": 1.7048768997192383, "step": 15934 }, { "epoch": 1.2562081198265669, "grad_norm": 0.5455548167228699, "learning_rate": 0.0005, "loss": 1.6330987215042114, "step": 15935 }, { "epoch": 1.2562869530942058, "grad_norm": 0.6711118221282959, "learning_rate": 0.0005, "loss": 1.75020432472229, "step": 15936 }, { "epoch": 1.2563657863618447, "grad_norm": 0.504727303981781, "learning_rate": 0.0005, "loss": 1.6256364583969116, "step": 15937 }, { "epoch": 1.2564446196294836, "grad_norm": 0.44084683060646057, "learning_rate": 0.0005, "loss": 1.38969087600708, "step": 15938 }, { "epoch": 1.2565234528971225, "grad_norm": 0.4479779899120331, "learning_rate": 0.0005, "loss": 1.4695346355438232, "step": 15939 }, { "epoch": 1.2566022861647617, "grad_norm": 0.4650697112083435, "learning_rate": 0.0005, "loss": 1.57722806930542, "step": 15940 }, { "epoch": 1.2566811194324004, "grad_norm": 0.4939112067222595, "learning_rate": 0.0005, "loss": 1.6787426471710205, "step": 15941 }, { "epoch": 1.2567599527000395, "grad_norm": 0.6565028429031372, "learning_rate": 0.0005, "loss": 1.6241297721862793, "step": 15942 }, { "epoch": 1.2568387859676784, "grad_norm": 0.5776252746582031, "learning_rate": 0.0005, "loss": 1.6444475650787354, "step": 15943 }, { "epoch": 1.2569176192353173, "grad_norm": 0.5467838048934937, "learning_rate": 0.0005, "loss": 1.6741547584533691, "step": 15944 }, { "epoch": 1.2569964525029562, "grad_norm": 0.6417495012283325, "learning_rate": 0.0005, "loss": 1.7274055480957031, "step": 15945 }, { "epoch": 1.2570752857705951, "grad_norm": 0.6263626217842102, "learning_rate": 0.0005, "loss": 1.709894061088562, "step": 15946 }, { "epoch": 1.2571541190382343, "grad_norm": 0.45897719264030457, "learning_rate": 0.0005, "loss": 1.5101959705352783, "step": 15947 }, { "epoch": 1.257232952305873, "grad_norm": 0.6076292991638184, "learning_rate": 0.0005, "loss": 1.5536342859268188, "step": 15948 }, { "epoch": 1.257311785573512, "grad_norm": 0.5667453408241272, "learning_rate": 0.0005, "loss": 1.6359665393829346, "step": 15949 }, { "epoch": 1.257390618841151, "grad_norm": 0.5388630628585815, "learning_rate": 0.0005, "loss": 1.8696606159210205, "step": 15950 }, { "epoch": 1.25746945210879, "grad_norm": 0.5590030550956726, "learning_rate": 0.0005, "loss": 1.3822723627090454, "step": 15951 }, { "epoch": 1.2575482853764288, "grad_norm": 0.7700631618499756, "learning_rate": 0.0005, "loss": 1.6259956359863281, "step": 15952 }, { "epoch": 1.2576271186440677, "grad_norm": 0.5014906525611877, "learning_rate": 0.0005, "loss": 1.5355957746505737, "step": 15953 }, { "epoch": 1.2577059519117069, "grad_norm": 0.5959880352020264, "learning_rate": 0.0005, "loss": 1.7736468315124512, "step": 15954 }, { "epoch": 1.2577847851793456, "grad_norm": 0.5073497891426086, "learning_rate": 0.0005, "loss": 1.6074509620666504, "step": 15955 }, { "epoch": 1.2578636184469847, "grad_norm": 0.5775490999221802, "learning_rate": 0.0005, "loss": 1.6443564891815186, "step": 15956 }, { "epoch": 1.2579424517146236, "grad_norm": 0.609246551990509, "learning_rate": 0.0005, "loss": 1.7017028331756592, "step": 15957 }, { "epoch": 1.2580212849822625, "grad_norm": 0.6368516087532043, "learning_rate": 0.0005, "loss": 1.845171332359314, "step": 15958 }, { "epoch": 1.2581001182499014, "grad_norm": 0.5700616240501404, "learning_rate": 0.0005, "loss": 1.4839918613433838, "step": 15959 }, { "epoch": 1.2581789515175403, "grad_norm": 0.6477546095848083, "learning_rate": 0.0005, "loss": 1.775803565979004, "step": 15960 }, { "epoch": 1.2582577847851795, "grad_norm": 0.6464781761169434, "learning_rate": 0.0005, "loss": 1.7650511264801025, "step": 15961 }, { "epoch": 1.2583366180528182, "grad_norm": 0.5596324801445007, "learning_rate": 0.0005, "loss": 1.6124048233032227, "step": 15962 }, { "epoch": 1.2584154513204573, "grad_norm": 0.8081305623054504, "learning_rate": 0.0005, "loss": 1.7097206115722656, "step": 15963 }, { "epoch": 1.2584942845880962, "grad_norm": 0.681527316570282, "learning_rate": 0.0005, "loss": 1.567905306816101, "step": 15964 }, { "epoch": 1.2585731178557351, "grad_norm": 0.5194681286811829, "learning_rate": 0.0005, "loss": 1.5260629653930664, "step": 15965 }, { "epoch": 1.258651951123374, "grad_norm": 0.5155999660491943, "learning_rate": 0.0005, "loss": 1.6033096313476562, "step": 15966 }, { "epoch": 1.258730784391013, "grad_norm": 0.679924488067627, "learning_rate": 0.0005, "loss": 1.4986300468444824, "step": 15967 }, { "epoch": 1.258809617658652, "grad_norm": 0.5786062479019165, "learning_rate": 0.0005, "loss": 2.126729965209961, "step": 15968 }, { "epoch": 1.2588884509262908, "grad_norm": 0.5849636793136597, "learning_rate": 0.0005, "loss": 1.5846740007400513, "step": 15969 }, { "epoch": 1.25896728419393, "grad_norm": 0.6393915414810181, "learning_rate": 0.0005, "loss": 1.5515516996383667, "step": 15970 }, { "epoch": 1.2590461174615688, "grad_norm": 0.4999786615371704, "learning_rate": 0.0005, "loss": 1.5951364040374756, "step": 15971 }, { "epoch": 1.2591249507292077, "grad_norm": 0.4281655251979828, "learning_rate": 0.0005, "loss": 1.4317986965179443, "step": 15972 }, { "epoch": 1.2592037839968466, "grad_norm": 0.49887096881866455, "learning_rate": 0.0005, "loss": 1.629561424255371, "step": 15973 }, { "epoch": 1.2592826172644855, "grad_norm": 0.7407338619232178, "learning_rate": 0.0005, "loss": 1.7338895797729492, "step": 15974 }, { "epoch": 1.2593614505321247, "grad_norm": 0.7388477325439453, "learning_rate": 0.0005, "loss": 1.568720817565918, "step": 15975 }, { "epoch": 1.2594402837997636, "grad_norm": 0.583737313747406, "learning_rate": 0.0005, "loss": 1.5651919841766357, "step": 15976 }, { "epoch": 1.2595191170674025, "grad_norm": 0.8034011721611023, "learning_rate": 0.0005, "loss": 1.694131851196289, "step": 15977 }, { "epoch": 1.2595979503350414, "grad_norm": 0.5352872014045715, "learning_rate": 0.0005, "loss": 1.7906066179275513, "step": 15978 }, { "epoch": 1.2596767836026803, "grad_norm": 0.45246225595474243, "learning_rate": 0.0005, "loss": 1.464379906654358, "step": 15979 }, { "epoch": 1.2597556168703192, "grad_norm": 1.1770676374435425, "learning_rate": 0.0005, "loss": 1.55515456199646, "step": 15980 }, { "epoch": 1.2598344501379581, "grad_norm": 0.4606248736381531, "learning_rate": 0.0005, "loss": 1.3260560035705566, "step": 15981 }, { "epoch": 1.2599132834055973, "grad_norm": 0.4775690734386444, "learning_rate": 0.0005, "loss": 1.5665534734725952, "step": 15982 }, { "epoch": 1.2599921166732362, "grad_norm": 0.44815322756767273, "learning_rate": 0.0005, "loss": 1.5316412448883057, "step": 15983 }, { "epoch": 1.260070949940875, "grad_norm": 0.6891964077949524, "learning_rate": 0.0005, "loss": 2.0392932891845703, "step": 15984 }, { "epoch": 1.260149783208514, "grad_norm": 0.9000562429428101, "learning_rate": 0.0005, "loss": 1.688301682472229, "step": 15985 }, { "epoch": 1.260228616476153, "grad_norm": 0.5810025334358215, "learning_rate": 0.0005, "loss": 1.7152063846588135, "step": 15986 }, { "epoch": 1.2603074497437918, "grad_norm": 0.5426386594772339, "learning_rate": 0.0005, "loss": 1.5588037967681885, "step": 15987 }, { "epoch": 1.2603862830114307, "grad_norm": 0.6223267912864685, "learning_rate": 0.0005, "loss": 1.92201566696167, "step": 15988 }, { "epoch": 1.2604651162790699, "grad_norm": 0.6345322728157043, "learning_rate": 0.0005, "loss": 1.8415067195892334, "step": 15989 }, { "epoch": 1.2605439495467088, "grad_norm": 0.568854570388794, "learning_rate": 0.0005, "loss": 2.021897315979004, "step": 15990 }, { "epoch": 1.2606227828143477, "grad_norm": 0.5292080640792847, "learning_rate": 0.0005, "loss": 1.5342400074005127, "step": 15991 }, { "epoch": 1.2607016160819866, "grad_norm": 0.5130759477615356, "learning_rate": 0.0005, "loss": 1.757705569267273, "step": 15992 }, { "epoch": 1.2607804493496255, "grad_norm": 0.4533393383026123, "learning_rate": 0.0005, "loss": 1.7176024913787842, "step": 15993 }, { "epoch": 1.2608592826172644, "grad_norm": 0.5760734677314758, "learning_rate": 0.0005, "loss": 1.82896089553833, "step": 15994 }, { "epoch": 1.2609381158849033, "grad_norm": 0.6266669631004333, "learning_rate": 0.0005, "loss": 1.5679512023925781, "step": 15995 }, { "epoch": 1.2610169491525425, "grad_norm": 0.41611528396606445, "learning_rate": 0.0005, "loss": 1.481817364692688, "step": 15996 }, { "epoch": 1.2610957824201814, "grad_norm": 0.5537511706352234, "learning_rate": 0.0005, "loss": 1.6786887645721436, "step": 15997 }, { "epoch": 1.2611746156878203, "grad_norm": 0.619317352771759, "learning_rate": 0.0005, "loss": 1.7287755012512207, "step": 15998 }, { "epoch": 1.2612534489554592, "grad_norm": 0.5844945311546326, "learning_rate": 0.0005, "loss": 1.6348235607147217, "step": 15999 }, { "epoch": 1.2613322822230981, "grad_norm": 0.5441606640815735, "learning_rate": 0.0005, "loss": 1.7748637199401855, "step": 16000 }, { "epoch": 1.261411115490737, "grad_norm": 0.5441505908966064, "learning_rate": 0.0005, "loss": 1.8706567287445068, "step": 16001 }, { "epoch": 1.261489948758376, "grad_norm": 0.5163463950157166, "learning_rate": 0.0005, "loss": 1.688374638557434, "step": 16002 }, { "epoch": 1.261568782026015, "grad_norm": 0.630737841129303, "learning_rate": 0.0005, "loss": 1.6798762083053589, "step": 16003 }, { "epoch": 1.261647615293654, "grad_norm": 0.6580555438995361, "learning_rate": 0.0005, "loss": 1.84474515914917, "step": 16004 }, { "epoch": 1.261726448561293, "grad_norm": 0.7568377256393433, "learning_rate": 0.0005, "loss": 1.7608414888381958, "step": 16005 }, { "epoch": 1.2618052818289318, "grad_norm": 0.7123785018920898, "learning_rate": 0.0005, "loss": 1.8608546257019043, "step": 16006 }, { "epoch": 1.2618841150965707, "grad_norm": 0.578754186630249, "learning_rate": 0.0005, "loss": 1.8175517320632935, "step": 16007 }, { "epoch": 1.2619629483642096, "grad_norm": 1.000507116317749, "learning_rate": 0.0005, "loss": 1.6627588272094727, "step": 16008 }, { "epoch": 1.2620417816318485, "grad_norm": 0.8867602348327637, "learning_rate": 0.0005, "loss": 1.7246251106262207, "step": 16009 }, { "epoch": 1.2621206148994877, "grad_norm": 0.587487518787384, "learning_rate": 0.0005, "loss": 1.7321662902832031, "step": 16010 }, { "epoch": 1.2621994481671266, "grad_norm": 0.41989389061927795, "learning_rate": 0.0005, "loss": 1.4850178956985474, "step": 16011 }, { "epoch": 1.2622782814347655, "grad_norm": 0.5689435601234436, "learning_rate": 0.0005, "loss": 1.7283579111099243, "step": 16012 }, { "epoch": 1.2623571147024044, "grad_norm": 0.5681312084197998, "learning_rate": 0.0005, "loss": 1.6002986431121826, "step": 16013 }, { "epoch": 1.2624359479700433, "grad_norm": 0.7785550951957703, "learning_rate": 0.0005, "loss": 1.716403841972351, "step": 16014 }, { "epoch": 1.2625147812376822, "grad_norm": 0.6166669726371765, "learning_rate": 0.0005, "loss": 1.4624319076538086, "step": 16015 }, { "epoch": 1.2625936145053211, "grad_norm": 0.47418856620788574, "learning_rate": 0.0005, "loss": 1.4779753684997559, "step": 16016 }, { "epoch": 1.2626724477729603, "grad_norm": 0.5256221294403076, "learning_rate": 0.0005, "loss": 1.6253609657287598, "step": 16017 }, { "epoch": 1.2627512810405992, "grad_norm": 0.4925445020198822, "learning_rate": 0.0005, "loss": 1.3548493385314941, "step": 16018 }, { "epoch": 1.262830114308238, "grad_norm": 0.7953348159790039, "learning_rate": 0.0005, "loss": 1.8225480318069458, "step": 16019 }, { "epoch": 1.262908947575877, "grad_norm": 0.6088740825653076, "learning_rate": 0.0005, "loss": 1.5147128105163574, "step": 16020 }, { "epoch": 1.262987780843516, "grad_norm": 0.5160344839096069, "learning_rate": 0.0005, "loss": 1.7922816276550293, "step": 16021 }, { "epoch": 1.2630666141111548, "grad_norm": 0.5187093019485474, "learning_rate": 0.0005, "loss": 1.6557979583740234, "step": 16022 }, { "epoch": 1.2631454473787938, "grad_norm": 0.5541436672210693, "learning_rate": 0.0005, "loss": 1.5617117881774902, "step": 16023 }, { "epoch": 1.2632242806464329, "grad_norm": 0.4870927929878235, "learning_rate": 0.0005, "loss": 1.7222988605499268, "step": 16024 }, { "epoch": 1.2633031139140718, "grad_norm": 0.561786949634552, "learning_rate": 0.0005, "loss": 1.8085832595825195, "step": 16025 }, { "epoch": 1.2633819471817107, "grad_norm": 0.4225577712059021, "learning_rate": 0.0005, "loss": 1.470259189605713, "step": 16026 }, { "epoch": 1.2634607804493496, "grad_norm": 0.5060257315635681, "learning_rate": 0.0005, "loss": 1.5396687984466553, "step": 16027 }, { "epoch": 1.2635396137169885, "grad_norm": 0.5525215268135071, "learning_rate": 0.0005, "loss": 1.6099094152450562, "step": 16028 }, { "epoch": 1.2636184469846274, "grad_norm": 0.8196669220924377, "learning_rate": 0.0005, "loss": 1.9774572849273682, "step": 16029 }, { "epoch": 1.2636972802522664, "grad_norm": 0.5321429967880249, "learning_rate": 0.0005, "loss": 1.5888938903808594, "step": 16030 }, { "epoch": 1.2637761135199055, "grad_norm": 0.5092148184776306, "learning_rate": 0.0005, "loss": 1.7359445095062256, "step": 16031 }, { "epoch": 1.2638549467875444, "grad_norm": 0.5861032009124756, "learning_rate": 0.0005, "loss": 1.5492515563964844, "step": 16032 }, { "epoch": 1.2639337800551833, "grad_norm": 0.6642863154411316, "learning_rate": 0.0005, "loss": 1.6678863763809204, "step": 16033 }, { "epoch": 1.2640126133228222, "grad_norm": 0.6275978684425354, "learning_rate": 0.0005, "loss": 1.7489134073257446, "step": 16034 }, { "epoch": 1.2640914465904611, "grad_norm": 0.4934603273868561, "learning_rate": 0.0005, "loss": 1.5473456382751465, "step": 16035 }, { "epoch": 1.2641702798581, "grad_norm": 0.5047363638877869, "learning_rate": 0.0005, "loss": 1.4761992692947388, "step": 16036 }, { "epoch": 1.264249113125739, "grad_norm": 0.49521970748901367, "learning_rate": 0.0005, "loss": 1.788286566734314, "step": 16037 }, { "epoch": 1.264327946393378, "grad_norm": 0.568166196346283, "learning_rate": 0.0005, "loss": 1.4882304668426514, "step": 16038 }, { "epoch": 1.264406779661017, "grad_norm": 0.5842240452766418, "learning_rate": 0.0005, "loss": 1.8199148178100586, "step": 16039 }, { "epoch": 1.264485612928656, "grad_norm": 0.48504579067230225, "learning_rate": 0.0005, "loss": 1.6807975769042969, "step": 16040 }, { "epoch": 1.2645644461962948, "grad_norm": 0.6779853105545044, "learning_rate": 0.0005, "loss": 1.7434751987457275, "step": 16041 }, { "epoch": 1.2646432794639337, "grad_norm": 0.626899003982544, "learning_rate": 0.0005, "loss": 1.5424728393554688, "step": 16042 }, { "epoch": 1.2647221127315726, "grad_norm": 0.9093732237815857, "learning_rate": 0.0005, "loss": 1.8427472114562988, "step": 16043 }, { "epoch": 1.2648009459992116, "grad_norm": 0.6241005063056946, "learning_rate": 0.0005, "loss": 1.729488492012024, "step": 16044 }, { "epoch": 1.2648797792668507, "grad_norm": 0.5866352319717407, "learning_rate": 0.0005, "loss": 1.4686914682388306, "step": 16045 }, { "epoch": 1.2649586125344896, "grad_norm": 0.472743958234787, "learning_rate": 0.0005, "loss": 1.6477937698364258, "step": 16046 }, { "epoch": 1.2650374458021285, "grad_norm": 0.70343416929245, "learning_rate": 0.0005, "loss": 1.6861062049865723, "step": 16047 }, { "epoch": 1.2651162790697674, "grad_norm": 0.48376771807670593, "learning_rate": 0.0005, "loss": 1.5199753046035767, "step": 16048 }, { "epoch": 1.2651951123374063, "grad_norm": 0.6480627059936523, "learning_rate": 0.0005, "loss": 1.6873116493225098, "step": 16049 }, { "epoch": 1.2652739456050452, "grad_norm": 0.5899717807769775, "learning_rate": 0.0005, "loss": 1.733057975769043, "step": 16050 }, { "epoch": 1.2653527788726842, "grad_norm": 0.45196887850761414, "learning_rate": 0.0005, "loss": 1.4805798530578613, "step": 16051 }, { "epoch": 1.2654316121403233, "grad_norm": 2.8355298042297363, "learning_rate": 0.0005, "loss": 1.50431227684021, "step": 16052 }, { "epoch": 1.2655104454079622, "grad_norm": 0.7127467393875122, "learning_rate": 0.0005, "loss": 1.4947396516799927, "step": 16053 }, { "epoch": 1.2655892786756011, "grad_norm": 0.5198074579238892, "learning_rate": 0.0005, "loss": 1.6906826496124268, "step": 16054 }, { "epoch": 1.26566811194324, "grad_norm": 0.5377006530761719, "learning_rate": 0.0005, "loss": 1.7343121767044067, "step": 16055 }, { "epoch": 1.265746945210879, "grad_norm": 0.5405824184417725, "learning_rate": 0.0005, "loss": 1.524294376373291, "step": 16056 }, { "epoch": 1.265825778478518, "grad_norm": 0.5504980683326721, "learning_rate": 0.0005, "loss": 1.7552403211593628, "step": 16057 }, { "epoch": 1.2659046117461568, "grad_norm": 0.5891038775444031, "learning_rate": 0.0005, "loss": 1.8429291248321533, "step": 16058 }, { "epoch": 1.265983445013796, "grad_norm": 0.6459779143333435, "learning_rate": 0.0005, "loss": 1.7168755531311035, "step": 16059 }, { "epoch": 1.2660622782814348, "grad_norm": 0.5498571395874023, "learning_rate": 0.0005, "loss": 1.6257156133651733, "step": 16060 }, { "epoch": 1.2661411115490737, "grad_norm": 0.652683436870575, "learning_rate": 0.0005, "loss": 1.7321573495864868, "step": 16061 }, { "epoch": 1.2662199448167126, "grad_norm": 0.5210015177726746, "learning_rate": 0.0005, "loss": 1.6315937042236328, "step": 16062 }, { "epoch": 1.2662987780843515, "grad_norm": 0.48277437686920166, "learning_rate": 0.0005, "loss": 1.6384994983673096, "step": 16063 }, { "epoch": 1.2663776113519907, "grad_norm": 0.5129207372665405, "learning_rate": 0.0005, "loss": 1.6580240726470947, "step": 16064 }, { "epoch": 1.2664564446196294, "grad_norm": 0.7475224733352661, "learning_rate": 0.0005, "loss": 1.9414536952972412, "step": 16065 }, { "epoch": 1.2665352778872685, "grad_norm": 0.5563601851463318, "learning_rate": 0.0005, "loss": 1.7529535293579102, "step": 16066 }, { "epoch": 1.2666141111549074, "grad_norm": 0.4961113929748535, "learning_rate": 0.0005, "loss": 1.717550277709961, "step": 16067 }, { "epoch": 1.2666929444225463, "grad_norm": 0.9003609418869019, "learning_rate": 0.0005, "loss": 1.8358356952667236, "step": 16068 }, { "epoch": 1.2667717776901852, "grad_norm": 0.5105368494987488, "learning_rate": 0.0005, "loss": 1.7480549812316895, "step": 16069 }, { "epoch": 1.2668506109578241, "grad_norm": 0.4485400319099426, "learning_rate": 0.0005, "loss": 1.7154847383499146, "step": 16070 }, { "epoch": 1.2669294442254633, "grad_norm": 0.5080695152282715, "learning_rate": 0.0005, "loss": 1.8136526346206665, "step": 16071 }, { "epoch": 1.267008277493102, "grad_norm": 0.6384178400039673, "learning_rate": 0.0005, "loss": 1.6256822347640991, "step": 16072 }, { "epoch": 1.267087110760741, "grad_norm": 0.6450077891349792, "learning_rate": 0.0005, "loss": 1.8885116577148438, "step": 16073 }, { "epoch": 1.26716594402838, "grad_norm": 0.6323207020759583, "learning_rate": 0.0005, "loss": 1.7109432220458984, "step": 16074 }, { "epoch": 1.267244777296019, "grad_norm": 0.787061870098114, "learning_rate": 0.0005, "loss": 1.982516884803772, "step": 16075 }, { "epoch": 1.2673236105636578, "grad_norm": 0.5548886060714722, "learning_rate": 0.0005, "loss": 1.6913213729858398, "step": 16076 }, { "epoch": 1.2674024438312967, "grad_norm": 0.5965359210968018, "learning_rate": 0.0005, "loss": 1.7461273670196533, "step": 16077 }, { "epoch": 1.2674812770989359, "grad_norm": 0.5157556533813477, "learning_rate": 0.0005, "loss": 1.765403151512146, "step": 16078 }, { "epoch": 1.2675601103665746, "grad_norm": 0.6026372313499451, "learning_rate": 0.0005, "loss": 1.4729112386703491, "step": 16079 }, { "epoch": 1.2676389436342137, "grad_norm": 0.6058189272880554, "learning_rate": 0.0005, "loss": 1.7634536027908325, "step": 16080 }, { "epoch": 1.2677177769018526, "grad_norm": 0.5179236531257629, "learning_rate": 0.0005, "loss": 1.6109951734542847, "step": 16081 }, { "epoch": 1.2677966101694915, "grad_norm": 0.4497359097003937, "learning_rate": 0.0005, "loss": 1.4952335357666016, "step": 16082 }, { "epoch": 1.2678754434371304, "grad_norm": 0.541038990020752, "learning_rate": 0.0005, "loss": 1.4998323917388916, "step": 16083 }, { "epoch": 1.2679542767047693, "grad_norm": 0.4354187846183777, "learning_rate": 0.0005, "loss": 1.3849749565124512, "step": 16084 }, { "epoch": 1.2680331099724085, "grad_norm": 0.6090059876441956, "learning_rate": 0.0005, "loss": 1.8119192123413086, "step": 16085 }, { "epoch": 1.2681119432400472, "grad_norm": 0.5265938639640808, "learning_rate": 0.0005, "loss": 1.6171236038208008, "step": 16086 }, { "epoch": 1.2681907765076863, "grad_norm": 0.5545985698699951, "learning_rate": 0.0005, "loss": 1.787416696548462, "step": 16087 }, { "epoch": 1.2682696097753252, "grad_norm": 0.4507065415382385, "learning_rate": 0.0005, "loss": 1.5899739265441895, "step": 16088 }, { "epoch": 1.2683484430429641, "grad_norm": 0.5522058010101318, "learning_rate": 0.0005, "loss": 1.7475115060806274, "step": 16089 }, { "epoch": 1.268427276310603, "grad_norm": 0.5252842307090759, "learning_rate": 0.0005, "loss": 1.6406424045562744, "step": 16090 }, { "epoch": 1.268506109578242, "grad_norm": 0.5268976092338562, "learning_rate": 0.0005, "loss": 1.6072967052459717, "step": 16091 }, { "epoch": 1.268584942845881, "grad_norm": 0.6442657709121704, "learning_rate": 0.0005, "loss": 1.7110319137573242, "step": 16092 }, { "epoch": 1.2686637761135198, "grad_norm": 0.5520563721656799, "learning_rate": 0.0005, "loss": 1.7351744174957275, "step": 16093 }, { "epoch": 1.268742609381159, "grad_norm": 0.6472369432449341, "learning_rate": 0.0005, "loss": 1.421374797821045, "step": 16094 }, { "epoch": 1.2688214426487978, "grad_norm": 0.5908406972885132, "learning_rate": 0.0005, "loss": 1.6602025032043457, "step": 16095 }, { "epoch": 1.2689002759164367, "grad_norm": 0.47055691480636597, "learning_rate": 0.0005, "loss": 1.8071889877319336, "step": 16096 }, { "epoch": 1.2689791091840756, "grad_norm": 0.43658795952796936, "learning_rate": 0.0005, "loss": 1.5940406322479248, "step": 16097 }, { "epoch": 1.2690579424517145, "grad_norm": 0.5610352754592896, "learning_rate": 0.0005, "loss": 1.5812203884124756, "step": 16098 }, { "epoch": 1.2691367757193537, "grad_norm": 0.44222113490104675, "learning_rate": 0.0005, "loss": 1.4984471797943115, "step": 16099 }, { "epoch": 1.2692156089869926, "grad_norm": 0.5598527789115906, "learning_rate": 0.0005, "loss": 1.5282869338989258, "step": 16100 }, { "epoch": 1.2692944422546315, "grad_norm": 0.5004889369010925, "learning_rate": 0.0005, "loss": 1.6935433149337769, "step": 16101 }, { "epoch": 1.2693732755222704, "grad_norm": 0.48560830950737, "learning_rate": 0.0005, "loss": 1.3422296047210693, "step": 16102 }, { "epoch": 1.2694521087899093, "grad_norm": 0.5119836330413818, "learning_rate": 0.0005, "loss": 1.7403357028961182, "step": 16103 }, { "epoch": 1.2695309420575482, "grad_norm": 0.47648391127586365, "learning_rate": 0.0005, "loss": 1.6056771278381348, "step": 16104 }, { "epoch": 1.2696097753251872, "grad_norm": 0.6393008828163147, "learning_rate": 0.0005, "loss": 1.9129559993743896, "step": 16105 }, { "epoch": 1.2696886085928263, "grad_norm": 0.7191959619522095, "learning_rate": 0.0005, "loss": 1.3693690299987793, "step": 16106 }, { "epoch": 1.2697674418604652, "grad_norm": 0.6104875206947327, "learning_rate": 0.0005, "loss": 1.7303431034088135, "step": 16107 }, { "epoch": 1.269846275128104, "grad_norm": 0.4757913649082184, "learning_rate": 0.0005, "loss": 1.4961731433868408, "step": 16108 }, { "epoch": 1.269925108395743, "grad_norm": 0.5780622959136963, "learning_rate": 0.0005, "loss": 1.6457481384277344, "step": 16109 }, { "epoch": 1.270003941663382, "grad_norm": 0.529335618019104, "learning_rate": 0.0005, "loss": 1.5939090251922607, "step": 16110 }, { "epoch": 1.2700827749310208, "grad_norm": 0.6572418808937073, "learning_rate": 0.0005, "loss": 1.6469354629516602, "step": 16111 }, { "epoch": 1.2701616081986598, "grad_norm": 0.6192021369934082, "learning_rate": 0.0005, "loss": 1.6897976398468018, "step": 16112 }, { "epoch": 1.2702404414662989, "grad_norm": 0.46355319023132324, "learning_rate": 0.0005, "loss": 1.5351874828338623, "step": 16113 }, { "epoch": 1.2703192747339378, "grad_norm": 0.7418138384819031, "learning_rate": 0.0005, "loss": 1.745164394378662, "step": 16114 }, { "epoch": 1.2703981080015767, "grad_norm": 0.6036005020141602, "learning_rate": 0.0005, "loss": 1.5325942039489746, "step": 16115 }, { "epoch": 1.2704769412692156, "grad_norm": 0.6169306039810181, "learning_rate": 0.0005, "loss": 1.7097671031951904, "step": 16116 }, { "epoch": 1.2705557745368545, "grad_norm": 0.5235539078712463, "learning_rate": 0.0005, "loss": 1.5592596530914307, "step": 16117 }, { "epoch": 1.2706346078044934, "grad_norm": 0.5391539931297302, "learning_rate": 0.0005, "loss": 1.6448886394500732, "step": 16118 }, { "epoch": 1.2707134410721324, "grad_norm": 0.4743286371231079, "learning_rate": 0.0005, "loss": 1.5747517347335815, "step": 16119 }, { "epoch": 1.2707922743397715, "grad_norm": 0.6324219107627869, "learning_rate": 0.0005, "loss": 1.8019951581954956, "step": 16120 }, { "epoch": 1.2708711076074104, "grad_norm": 0.5686339139938354, "learning_rate": 0.0005, "loss": 1.7560397386550903, "step": 16121 }, { "epoch": 1.2709499408750493, "grad_norm": 0.6729896068572998, "learning_rate": 0.0005, "loss": 1.6087992191314697, "step": 16122 }, { "epoch": 1.2710287741426882, "grad_norm": 0.6010687351226807, "learning_rate": 0.0005, "loss": 1.8441452980041504, "step": 16123 }, { "epoch": 1.2711076074103271, "grad_norm": 0.586886465549469, "learning_rate": 0.0005, "loss": 1.7967323064804077, "step": 16124 }, { "epoch": 1.271186440677966, "grad_norm": 0.673886239528656, "learning_rate": 0.0005, "loss": 1.7454664707183838, "step": 16125 }, { "epoch": 1.271265273945605, "grad_norm": 0.5083664059638977, "learning_rate": 0.0005, "loss": 1.6079758405685425, "step": 16126 }, { "epoch": 1.271344107213244, "grad_norm": 0.5851960182189941, "learning_rate": 0.0005, "loss": 1.7731809616088867, "step": 16127 }, { "epoch": 1.271422940480883, "grad_norm": 0.5030792951583862, "learning_rate": 0.0005, "loss": 1.5575275421142578, "step": 16128 }, { "epoch": 1.271501773748522, "grad_norm": 0.6851491332054138, "learning_rate": 0.0005, "loss": 1.3732118606567383, "step": 16129 }, { "epoch": 1.2715806070161608, "grad_norm": 0.8840955495834351, "learning_rate": 0.0005, "loss": 1.4832487106323242, "step": 16130 }, { "epoch": 1.2716594402837997, "grad_norm": 0.4634357690811157, "learning_rate": 0.0005, "loss": 1.5400221347808838, "step": 16131 }, { "epoch": 1.2717382735514386, "grad_norm": 0.4712198078632355, "learning_rate": 0.0005, "loss": 1.4388397932052612, "step": 16132 }, { "epoch": 1.2718171068190776, "grad_norm": 0.5787768959999084, "learning_rate": 0.0005, "loss": 1.8265281915664673, "step": 16133 }, { "epoch": 1.2718959400867167, "grad_norm": 0.49125558137893677, "learning_rate": 0.0005, "loss": 1.431089162826538, "step": 16134 }, { "epoch": 1.2719747733543556, "grad_norm": 0.7843859195709229, "learning_rate": 0.0005, "loss": 1.7654492855072021, "step": 16135 }, { "epoch": 1.2720536066219945, "grad_norm": 0.570080578327179, "learning_rate": 0.0005, "loss": 1.649754285812378, "step": 16136 }, { "epoch": 1.2721324398896334, "grad_norm": 0.6238524317741394, "learning_rate": 0.0005, "loss": 1.6556503772735596, "step": 16137 }, { "epoch": 1.2722112731572723, "grad_norm": 0.6181821823120117, "learning_rate": 0.0005, "loss": 1.6029937267303467, "step": 16138 }, { "epoch": 1.2722901064249112, "grad_norm": 0.5314601063728333, "learning_rate": 0.0005, "loss": 1.7996807098388672, "step": 16139 }, { "epoch": 1.2723689396925502, "grad_norm": 0.5402716994285583, "learning_rate": 0.0005, "loss": 1.7508025169372559, "step": 16140 }, { "epoch": 1.2724477729601893, "grad_norm": 0.4959122836589813, "learning_rate": 0.0005, "loss": 1.5414485931396484, "step": 16141 }, { "epoch": 1.2725266062278282, "grad_norm": 0.6347779035568237, "learning_rate": 0.0005, "loss": 1.8484784364700317, "step": 16142 }, { "epoch": 1.2726054394954671, "grad_norm": 0.7187505960464478, "learning_rate": 0.0005, "loss": 1.6728448867797852, "step": 16143 }, { "epoch": 1.272684272763106, "grad_norm": 0.6694938540458679, "learning_rate": 0.0005, "loss": 1.947218656539917, "step": 16144 }, { "epoch": 1.272763106030745, "grad_norm": 0.6280667185783386, "learning_rate": 0.0005, "loss": 1.7680411338806152, "step": 16145 }, { "epoch": 1.2728419392983839, "grad_norm": 0.5391851663589478, "learning_rate": 0.0005, "loss": 1.4500842094421387, "step": 16146 }, { "epoch": 1.2729207725660228, "grad_norm": 0.604754626750946, "learning_rate": 0.0005, "loss": 1.7628637552261353, "step": 16147 }, { "epoch": 1.272999605833662, "grad_norm": 4.848252773284912, "learning_rate": 0.0005, "loss": 1.7552143335342407, "step": 16148 }, { "epoch": 1.2730784391013008, "grad_norm": 0.5162718892097473, "learning_rate": 0.0005, "loss": 1.6387816667556763, "step": 16149 }, { "epoch": 1.2731572723689397, "grad_norm": 0.5126772522926331, "learning_rate": 0.0005, "loss": 1.6166789531707764, "step": 16150 }, { "epoch": 1.2732361056365786, "grad_norm": 0.5984975695610046, "learning_rate": 0.0005, "loss": 1.6295115947723389, "step": 16151 }, { "epoch": 1.2733149389042175, "grad_norm": 0.5531330108642578, "learning_rate": 0.0005, "loss": 1.5155683755874634, "step": 16152 }, { "epoch": 1.2733937721718565, "grad_norm": 0.5833953022956848, "learning_rate": 0.0005, "loss": 1.4604586362838745, "step": 16153 }, { "epoch": 1.2734726054394954, "grad_norm": 0.545599102973938, "learning_rate": 0.0005, "loss": 1.7776046991348267, "step": 16154 }, { "epoch": 1.2735514387071345, "grad_norm": 0.656997799873352, "learning_rate": 0.0005, "loss": 1.6762913465499878, "step": 16155 }, { "epoch": 1.2736302719747734, "grad_norm": 0.4864376187324524, "learning_rate": 0.0005, "loss": 1.6489585638046265, "step": 16156 }, { "epoch": 1.2737091052424123, "grad_norm": 0.49645230174064636, "learning_rate": 0.0005, "loss": 1.6082077026367188, "step": 16157 }, { "epoch": 1.2737879385100512, "grad_norm": 0.76308274269104, "learning_rate": 0.0005, "loss": 1.8432880640029907, "step": 16158 }, { "epoch": 1.2738667717776901, "grad_norm": 0.5098764896392822, "learning_rate": 0.0005, "loss": 1.3280415534973145, "step": 16159 }, { "epoch": 1.273945605045329, "grad_norm": 0.5895016193389893, "learning_rate": 0.0005, "loss": 1.5748412609100342, "step": 16160 }, { "epoch": 1.274024438312968, "grad_norm": 0.6254615187644958, "learning_rate": 0.0005, "loss": 1.600121259689331, "step": 16161 }, { "epoch": 1.274103271580607, "grad_norm": 0.6622096300125122, "learning_rate": 0.0005, "loss": 1.7094753980636597, "step": 16162 }, { "epoch": 1.274182104848246, "grad_norm": 0.6476050019264221, "learning_rate": 0.0005, "loss": 1.8238502740859985, "step": 16163 }, { "epoch": 1.274260938115885, "grad_norm": 0.6936165690422058, "learning_rate": 0.0005, "loss": 1.678799033164978, "step": 16164 }, { "epoch": 1.2743397713835238, "grad_norm": 0.8762518763542175, "learning_rate": 0.0005, "loss": 1.5564758777618408, "step": 16165 }, { "epoch": 1.2744186046511627, "grad_norm": 0.6245739459991455, "learning_rate": 0.0005, "loss": 1.6157840490341187, "step": 16166 }, { "epoch": 1.2744974379188017, "grad_norm": 0.5041972398757935, "learning_rate": 0.0005, "loss": 1.5510828495025635, "step": 16167 }, { "epoch": 1.2745762711864406, "grad_norm": 0.6845785975456238, "learning_rate": 0.0005, "loss": 1.760995864868164, "step": 16168 }, { "epoch": 1.2746551044540797, "grad_norm": 0.7803280353546143, "learning_rate": 0.0005, "loss": 1.6289911270141602, "step": 16169 }, { "epoch": 1.2747339377217186, "grad_norm": 0.673867404460907, "learning_rate": 0.0005, "loss": 1.7967007160186768, "step": 16170 }, { "epoch": 1.2748127709893575, "grad_norm": 0.6386691331863403, "learning_rate": 0.0005, "loss": 1.7393476963043213, "step": 16171 }, { "epoch": 1.2748916042569964, "grad_norm": 0.5585483908653259, "learning_rate": 0.0005, "loss": 1.6172819137573242, "step": 16172 }, { "epoch": 1.2749704375246353, "grad_norm": 0.6508100032806396, "learning_rate": 0.0005, "loss": 1.889472484588623, "step": 16173 }, { "epoch": 1.2750492707922743, "grad_norm": 0.47682613134384155, "learning_rate": 0.0005, "loss": 1.699253797531128, "step": 16174 }, { "epoch": 1.2751281040599132, "grad_norm": 0.6358739733695984, "learning_rate": 0.0005, "loss": 1.6642961502075195, "step": 16175 }, { "epoch": 1.2752069373275523, "grad_norm": 0.6186701655387878, "learning_rate": 0.0005, "loss": 1.516200065612793, "step": 16176 }, { "epoch": 1.2752857705951912, "grad_norm": 0.717319905757904, "learning_rate": 0.0005, "loss": 1.8671882152557373, "step": 16177 }, { "epoch": 1.2753646038628301, "grad_norm": 0.5790184140205383, "learning_rate": 0.0005, "loss": 1.7856969833374023, "step": 16178 }, { "epoch": 1.275443437130469, "grad_norm": 0.5405283570289612, "learning_rate": 0.0005, "loss": 1.5554425716400146, "step": 16179 }, { "epoch": 1.275522270398108, "grad_norm": 0.5417516827583313, "learning_rate": 0.0005, "loss": 1.707385778427124, "step": 16180 }, { "epoch": 1.275601103665747, "grad_norm": 0.5951985716819763, "learning_rate": 0.0005, "loss": 1.817333459854126, "step": 16181 }, { "epoch": 1.2756799369333858, "grad_norm": 0.5680277943611145, "learning_rate": 0.0005, "loss": 1.732590675354004, "step": 16182 }, { "epoch": 1.275758770201025, "grad_norm": 0.5907093286514282, "learning_rate": 0.0005, "loss": 1.904380202293396, "step": 16183 }, { "epoch": 1.2758376034686638, "grad_norm": 0.5298436880111694, "learning_rate": 0.0005, "loss": 1.7310338020324707, "step": 16184 }, { "epoch": 1.2759164367363027, "grad_norm": 0.6131483316421509, "learning_rate": 0.0005, "loss": 1.6908655166625977, "step": 16185 }, { "epoch": 1.2759952700039416, "grad_norm": 0.6489503979682922, "learning_rate": 0.0005, "loss": 1.798210620880127, "step": 16186 }, { "epoch": 1.2760741032715806, "grad_norm": 0.5089479088783264, "learning_rate": 0.0005, "loss": 1.8093503713607788, "step": 16187 }, { "epoch": 1.2761529365392197, "grad_norm": 0.5618636012077332, "learning_rate": 0.0005, "loss": 1.6836555004119873, "step": 16188 }, { "epoch": 1.2762317698068584, "grad_norm": 0.4367291033267975, "learning_rate": 0.0005, "loss": 1.5768200159072876, "step": 16189 }, { "epoch": 1.2763106030744975, "grad_norm": 0.9241332411766052, "learning_rate": 0.0005, "loss": 1.7810449600219727, "step": 16190 }, { "epoch": 1.2763894363421364, "grad_norm": 0.6471385955810547, "learning_rate": 0.0005, "loss": 1.8902692794799805, "step": 16191 }, { "epoch": 1.2764682696097753, "grad_norm": 0.5187317728996277, "learning_rate": 0.0005, "loss": 1.7423794269561768, "step": 16192 }, { "epoch": 1.2765471028774142, "grad_norm": 0.6757825016975403, "learning_rate": 0.0005, "loss": 1.872384786605835, "step": 16193 }, { "epoch": 1.2766259361450532, "grad_norm": 0.4703502953052521, "learning_rate": 0.0005, "loss": 1.6150221824645996, "step": 16194 }, { "epoch": 1.2767047694126923, "grad_norm": 0.48808935284614563, "learning_rate": 0.0005, "loss": 1.6786797046661377, "step": 16195 }, { "epoch": 1.276783602680331, "grad_norm": 0.5023089051246643, "learning_rate": 0.0005, "loss": 1.74703049659729, "step": 16196 }, { "epoch": 1.27686243594797, "grad_norm": 0.5462737083435059, "learning_rate": 0.0005, "loss": 1.870656967163086, "step": 16197 }, { "epoch": 1.276941269215609, "grad_norm": 0.5119497179985046, "learning_rate": 0.0005, "loss": 1.6386961936950684, "step": 16198 }, { "epoch": 1.277020102483248, "grad_norm": 0.6002764105796814, "learning_rate": 0.0005, "loss": 1.8946332931518555, "step": 16199 }, { "epoch": 1.2770989357508868, "grad_norm": 0.40279078483581543, "learning_rate": 0.0005, "loss": 1.5458096265792847, "step": 16200 }, { "epoch": 1.2771777690185258, "grad_norm": 0.7141332030296326, "learning_rate": 0.0005, "loss": 1.94463312625885, "step": 16201 }, { "epoch": 1.2772566022861649, "grad_norm": 0.5034751892089844, "learning_rate": 0.0005, "loss": 1.622035026550293, "step": 16202 }, { "epoch": 1.2773354355538036, "grad_norm": 0.522134006023407, "learning_rate": 0.0005, "loss": 1.7031772136688232, "step": 16203 }, { "epoch": 1.2774142688214427, "grad_norm": 0.5491982698440552, "learning_rate": 0.0005, "loss": 1.5482900142669678, "step": 16204 }, { "epoch": 1.2774931020890816, "grad_norm": 0.5811529755592346, "learning_rate": 0.0005, "loss": 1.6289401054382324, "step": 16205 }, { "epoch": 1.2775719353567205, "grad_norm": 0.943046510219574, "learning_rate": 0.0005, "loss": 1.8981826305389404, "step": 16206 }, { "epoch": 1.2776507686243594, "grad_norm": 0.45606574416160583, "learning_rate": 0.0005, "loss": 1.5769932270050049, "step": 16207 }, { "epoch": 1.2777296018919984, "grad_norm": 0.5504775047302246, "learning_rate": 0.0005, "loss": 1.6283588409423828, "step": 16208 }, { "epoch": 1.2778084351596375, "grad_norm": 0.538217306137085, "learning_rate": 0.0005, "loss": 1.5129482746124268, "step": 16209 }, { "epoch": 1.2778872684272762, "grad_norm": 0.6118860244750977, "learning_rate": 0.0005, "loss": 1.6027963161468506, "step": 16210 }, { "epoch": 1.2779661016949153, "grad_norm": 0.6119711399078369, "learning_rate": 0.0005, "loss": 1.6786091327667236, "step": 16211 }, { "epoch": 1.2780449349625542, "grad_norm": 0.46137532591819763, "learning_rate": 0.0005, "loss": 1.5416291952133179, "step": 16212 }, { "epoch": 1.2781237682301931, "grad_norm": 0.5577031970024109, "learning_rate": 0.0005, "loss": 1.670820713043213, "step": 16213 }, { "epoch": 1.278202601497832, "grad_norm": 0.5137978792190552, "learning_rate": 0.0005, "loss": 1.7389552593231201, "step": 16214 }, { "epoch": 1.278281434765471, "grad_norm": 0.5388693809509277, "learning_rate": 0.0005, "loss": 1.6544605493545532, "step": 16215 }, { "epoch": 1.27836026803311, "grad_norm": 0.5440279841423035, "learning_rate": 0.0005, "loss": 1.6675450801849365, "step": 16216 }, { "epoch": 1.2784391013007488, "grad_norm": 0.577179491519928, "learning_rate": 0.0005, "loss": 1.5938212871551514, "step": 16217 }, { "epoch": 1.278517934568388, "grad_norm": 0.6072812676429749, "learning_rate": 0.0005, "loss": 1.691739797592163, "step": 16218 }, { "epoch": 1.2785967678360268, "grad_norm": 0.6427597999572754, "learning_rate": 0.0005, "loss": 1.915257453918457, "step": 16219 }, { "epoch": 1.2786756011036657, "grad_norm": 0.7832884788513184, "learning_rate": 0.0005, "loss": 1.7991957664489746, "step": 16220 }, { "epoch": 1.2787544343713046, "grad_norm": 0.5444676876068115, "learning_rate": 0.0005, "loss": 1.481771469116211, "step": 16221 }, { "epoch": 1.2788332676389436, "grad_norm": 0.5799961090087891, "learning_rate": 0.0005, "loss": 1.7672898769378662, "step": 16222 }, { "epoch": 1.2789121009065827, "grad_norm": 0.6100388169288635, "learning_rate": 0.0005, "loss": 1.6503369808197021, "step": 16223 }, { "epoch": 1.2789909341742214, "grad_norm": 0.4878090023994446, "learning_rate": 0.0005, "loss": 1.5874804258346558, "step": 16224 }, { "epoch": 1.2790697674418605, "grad_norm": 0.5195603370666504, "learning_rate": 0.0005, "loss": 1.4936034679412842, "step": 16225 }, { "epoch": 1.2791486007094994, "grad_norm": 2.3605024814605713, "learning_rate": 0.0005, "loss": 1.7556874752044678, "step": 16226 }, { "epoch": 1.2792274339771383, "grad_norm": 0.6007657647132874, "learning_rate": 0.0005, "loss": 1.7107536792755127, "step": 16227 }, { "epoch": 1.2793062672447773, "grad_norm": 0.6845259070396423, "learning_rate": 0.0005, "loss": 1.467603325843811, "step": 16228 }, { "epoch": 1.2793851005124162, "grad_norm": 0.5049802660942078, "learning_rate": 0.0005, "loss": 1.5410720109939575, "step": 16229 }, { "epoch": 1.2794639337800553, "grad_norm": 0.5403023958206177, "learning_rate": 0.0005, "loss": 1.7983672618865967, "step": 16230 }, { "epoch": 1.2795427670476942, "grad_norm": 0.6532866358757019, "learning_rate": 0.0005, "loss": 1.6423176527023315, "step": 16231 }, { "epoch": 1.2796216003153331, "grad_norm": 0.5866442918777466, "learning_rate": 0.0005, "loss": 1.4703290462493896, "step": 16232 }, { "epoch": 1.279700433582972, "grad_norm": 0.5194730758666992, "learning_rate": 0.0005, "loss": 1.8107833862304688, "step": 16233 }, { "epoch": 1.279779266850611, "grad_norm": 0.44363638758659363, "learning_rate": 0.0005, "loss": 1.7159392833709717, "step": 16234 }, { "epoch": 1.2798581001182499, "grad_norm": 0.7876383066177368, "learning_rate": 0.0005, "loss": 1.6045787334442139, "step": 16235 }, { "epoch": 1.2799369333858888, "grad_norm": 0.5927329659461975, "learning_rate": 0.0005, "loss": 1.6003974676132202, "step": 16236 }, { "epoch": 1.280015766653528, "grad_norm": 0.45131149888038635, "learning_rate": 0.0005, "loss": 1.545769214630127, "step": 16237 }, { "epoch": 1.2800945999211668, "grad_norm": 0.6988897919654846, "learning_rate": 0.0005, "loss": 1.8509001731872559, "step": 16238 }, { "epoch": 1.2801734331888057, "grad_norm": 0.5627400279045105, "learning_rate": 0.0005, "loss": 1.7932422161102295, "step": 16239 }, { "epoch": 1.2802522664564446, "grad_norm": 0.6062765121459961, "learning_rate": 0.0005, "loss": 1.6881046295166016, "step": 16240 }, { "epoch": 1.2803310997240835, "grad_norm": 0.53563392162323, "learning_rate": 0.0005, "loss": 1.85140061378479, "step": 16241 }, { "epoch": 1.2804099329917225, "grad_norm": 0.6940979957580566, "learning_rate": 0.0005, "loss": 1.7510547637939453, "step": 16242 }, { "epoch": 1.2804887662593614, "grad_norm": 0.4870188236236572, "learning_rate": 0.0005, "loss": 1.5861537456512451, "step": 16243 }, { "epoch": 1.2805675995270005, "grad_norm": 0.5718528032302856, "learning_rate": 0.0005, "loss": 1.405195951461792, "step": 16244 }, { "epoch": 1.2806464327946394, "grad_norm": 0.4071103632450104, "learning_rate": 0.0005, "loss": 1.4925270080566406, "step": 16245 }, { "epoch": 1.2807252660622783, "grad_norm": 0.5333779454231262, "learning_rate": 0.0005, "loss": 1.380239725112915, "step": 16246 }, { "epoch": 1.2808040993299172, "grad_norm": 0.677919328212738, "learning_rate": 0.0005, "loss": 1.6799825429916382, "step": 16247 }, { "epoch": 1.2808829325975561, "grad_norm": 0.461843341588974, "learning_rate": 0.0005, "loss": 1.6604827642440796, "step": 16248 }, { "epoch": 1.280961765865195, "grad_norm": 0.492857962846756, "learning_rate": 0.0005, "loss": 1.7521764039993286, "step": 16249 }, { "epoch": 1.281040599132834, "grad_norm": 0.72981858253479, "learning_rate": 0.0005, "loss": 1.5999572277069092, "step": 16250 }, { "epoch": 1.281119432400473, "grad_norm": 0.43159711360931396, "learning_rate": 0.0005, "loss": 1.425980567932129, "step": 16251 }, { "epoch": 1.281198265668112, "grad_norm": 0.546534538269043, "learning_rate": 0.0005, "loss": 1.7128591537475586, "step": 16252 }, { "epoch": 1.281277098935751, "grad_norm": 0.5153809785842896, "learning_rate": 0.0005, "loss": 1.4371623992919922, "step": 16253 }, { "epoch": 1.2813559322033898, "grad_norm": 0.6197747588157654, "learning_rate": 0.0005, "loss": 1.8811986446380615, "step": 16254 }, { "epoch": 1.2814347654710287, "grad_norm": 0.5624358654022217, "learning_rate": 0.0005, "loss": 1.6963286399841309, "step": 16255 }, { "epoch": 1.2815135987386677, "grad_norm": 0.4882468283176422, "learning_rate": 0.0005, "loss": 1.492785930633545, "step": 16256 }, { "epoch": 1.2815924320063066, "grad_norm": 0.519934892654419, "learning_rate": 0.0005, "loss": 1.5043964385986328, "step": 16257 }, { "epoch": 1.2816712652739457, "grad_norm": 0.5474907159805298, "learning_rate": 0.0005, "loss": 1.6600360870361328, "step": 16258 }, { "epoch": 1.2817500985415846, "grad_norm": 0.49467962980270386, "learning_rate": 0.0005, "loss": 1.5375185012817383, "step": 16259 }, { "epoch": 1.2818289318092235, "grad_norm": 0.4765951931476593, "learning_rate": 0.0005, "loss": 1.7265803813934326, "step": 16260 }, { "epoch": 1.2819077650768624, "grad_norm": 0.4398925304412842, "learning_rate": 0.0005, "loss": 1.6477628946304321, "step": 16261 }, { "epoch": 1.2819865983445013, "grad_norm": 0.7108867168426514, "learning_rate": 0.0005, "loss": 2.0089802742004395, "step": 16262 }, { "epoch": 1.2820654316121403, "grad_norm": 0.5476586222648621, "learning_rate": 0.0005, "loss": 1.8330988883972168, "step": 16263 }, { "epoch": 1.2821442648797792, "grad_norm": 0.5062450766563416, "learning_rate": 0.0005, "loss": 1.8355598449707031, "step": 16264 }, { "epoch": 1.2822230981474183, "grad_norm": 0.5444106459617615, "learning_rate": 0.0005, "loss": 1.847428798675537, "step": 16265 }, { "epoch": 1.2823019314150572, "grad_norm": 0.6006224751472473, "learning_rate": 0.0005, "loss": 1.6838443279266357, "step": 16266 }, { "epoch": 1.2823807646826961, "grad_norm": 0.6114441156387329, "learning_rate": 0.0005, "loss": 2.0514743328094482, "step": 16267 }, { "epoch": 1.282459597950335, "grad_norm": 0.6641794443130493, "learning_rate": 0.0005, "loss": 1.9860243797302246, "step": 16268 }, { "epoch": 1.282538431217974, "grad_norm": 0.5176475644111633, "learning_rate": 0.0005, "loss": 1.6824917793273926, "step": 16269 }, { "epoch": 1.2826172644856129, "grad_norm": 0.5953558087348938, "learning_rate": 0.0005, "loss": 1.6516969203948975, "step": 16270 }, { "epoch": 1.2826960977532518, "grad_norm": 0.5616496205329895, "learning_rate": 0.0005, "loss": 1.9089410305023193, "step": 16271 }, { "epoch": 1.282774931020891, "grad_norm": 0.53795325756073, "learning_rate": 0.0005, "loss": 1.4812426567077637, "step": 16272 }, { "epoch": 1.2828537642885298, "grad_norm": 0.710115373134613, "learning_rate": 0.0005, "loss": 1.63662588596344, "step": 16273 }, { "epoch": 1.2829325975561687, "grad_norm": 0.6487851142883301, "learning_rate": 0.0005, "loss": 1.6479616165161133, "step": 16274 }, { "epoch": 1.2830114308238076, "grad_norm": 0.5023473501205444, "learning_rate": 0.0005, "loss": 1.5286741256713867, "step": 16275 }, { "epoch": 1.2830902640914466, "grad_norm": 0.5709382891654968, "learning_rate": 0.0005, "loss": 1.848450779914856, "step": 16276 }, { "epoch": 1.2831690973590855, "grad_norm": 0.458336740732193, "learning_rate": 0.0005, "loss": 1.5512821674346924, "step": 16277 }, { "epoch": 1.2832479306267244, "grad_norm": 0.5564455389976501, "learning_rate": 0.0005, "loss": 1.8512420654296875, "step": 16278 }, { "epoch": 1.2833267638943635, "grad_norm": 0.4867623448371887, "learning_rate": 0.0005, "loss": 1.9586576223373413, "step": 16279 }, { "epoch": 1.2834055971620024, "grad_norm": 0.5089395046234131, "learning_rate": 0.0005, "loss": 1.6726336479187012, "step": 16280 }, { "epoch": 1.2834844304296413, "grad_norm": 0.471529096364975, "learning_rate": 0.0005, "loss": 1.7695274353027344, "step": 16281 }, { "epoch": 1.2835632636972802, "grad_norm": 0.4925093948841095, "learning_rate": 0.0005, "loss": 1.63776695728302, "step": 16282 }, { "epoch": 1.2836420969649192, "grad_norm": 0.5504893660545349, "learning_rate": 0.0005, "loss": 1.4888222217559814, "step": 16283 }, { "epoch": 1.283720930232558, "grad_norm": 0.520899772644043, "learning_rate": 0.0005, "loss": 1.605201005935669, "step": 16284 }, { "epoch": 1.283799763500197, "grad_norm": 0.5917043089866638, "learning_rate": 0.0005, "loss": 1.6348557472229004, "step": 16285 }, { "epoch": 1.2838785967678361, "grad_norm": 0.5931698679924011, "learning_rate": 0.0005, "loss": 1.7183349132537842, "step": 16286 }, { "epoch": 1.283957430035475, "grad_norm": 0.5083533525466919, "learning_rate": 0.0005, "loss": 1.7547187805175781, "step": 16287 }, { "epoch": 1.284036263303114, "grad_norm": 0.6420134902000427, "learning_rate": 0.0005, "loss": 1.7665650844573975, "step": 16288 }, { "epoch": 1.2841150965707528, "grad_norm": 0.6785754561424255, "learning_rate": 0.0005, "loss": 2.004054069519043, "step": 16289 }, { "epoch": 1.2841939298383918, "grad_norm": 0.6200540065765381, "learning_rate": 0.0005, "loss": 1.928299903869629, "step": 16290 }, { "epoch": 1.2842727631060307, "grad_norm": 0.7417194247245789, "learning_rate": 0.0005, "loss": 1.7567546367645264, "step": 16291 }, { "epoch": 1.2843515963736696, "grad_norm": 0.5509018898010254, "learning_rate": 0.0005, "loss": 1.5949398279190063, "step": 16292 }, { "epoch": 1.2844304296413087, "grad_norm": 0.5870612859725952, "learning_rate": 0.0005, "loss": 1.6863288879394531, "step": 16293 }, { "epoch": 1.2845092629089476, "grad_norm": 0.6715487241744995, "learning_rate": 0.0005, "loss": 1.6591564416885376, "step": 16294 }, { "epoch": 1.2845880961765865, "grad_norm": 0.5645233988761902, "learning_rate": 0.0005, "loss": 1.4968582391738892, "step": 16295 }, { "epoch": 1.2846669294442254, "grad_norm": 0.8973541259765625, "learning_rate": 0.0005, "loss": 1.9057681560516357, "step": 16296 }, { "epoch": 1.2847457627118644, "grad_norm": 0.5243551731109619, "learning_rate": 0.0005, "loss": 1.5026133060455322, "step": 16297 }, { "epoch": 1.2848245959795033, "grad_norm": 0.488538533449173, "learning_rate": 0.0005, "loss": 1.9108762741088867, "step": 16298 }, { "epoch": 1.2849034292471422, "grad_norm": 0.5778293609619141, "learning_rate": 0.0005, "loss": 1.5107871294021606, "step": 16299 }, { "epoch": 1.2849822625147813, "grad_norm": 0.5583764314651489, "learning_rate": 0.0005, "loss": 1.7950725555419922, "step": 16300 }, { "epoch": 1.2850610957824202, "grad_norm": 0.5349593758583069, "learning_rate": 0.0005, "loss": 1.536675214767456, "step": 16301 }, { "epoch": 1.2851399290500591, "grad_norm": 0.6515592932701111, "learning_rate": 0.0005, "loss": 1.4579188823699951, "step": 16302 }, { "epoch": 1.285218762317698, "grad_norm": 0.577098548412323, "learning_rate": 0.0005, "loss": 1.6450722217559814, "step": 16303 }, { "epoch": 1.285297595585337, "grad_norm": 0.5366072654724121, "learning_rate": 0.0005, "loss": 1.5557523965835571, "step": 16304 }, { "epoch": 1.285376428852976, "grad_norm": 0.6110992431640625, "learning_rate": 0.0005, "loss": 1.8285627365112305, "step": 16305 }, { "epoch": 1.2854552621206148, "grad_norm": 0.6787968873977661, "learning_rate": 0.0005, "loss": 1.768758773803711, "step": 16306 }, { "epoch": 1.285534095388254, "grad_norm": 0.5091232061386108, "learning_rate": 0.0005, "loss": 1.5995378494262695, "step": 16307 }, { "epoch": 1.2856129286558928, "grad_norm": 0.45053619146347046, "learning_rate": 0.0005, "loss": 1.51140296459198, "step": 16308 }, { "epoch": 1.2856917619235317, "grad_norm": 0.7188429236412048, "learning_rate": 0.0005, "loss": 1.6939144134521484, "step": 16309 }, { "epoch": 1.2857705951911707, "grad_norm": 0.5852843523025513, "learning_rate": 0.0005, "loss": 1.6196222305297852, "step": 16310 }, { "epoch": 1.2858494284588096, "grad_norm": 0.5137341022491455, "learning_rate": 0.0005, "loss": 1.4629523754119873, "step": 16311 }, { "epoch": 1.2859282617264487, "grad_norm": 0.5571936368942261, "learning_rate": 0.0005, "loss": 1.8798692226409912, "step": 16312 }, { "epoch": 1.2860070949940874, "grad_norm": 0.4217562973499298, "learning_rate": 0.0005, "loss": 1.5574886798858643, "step": 16313 }, { "epoch": 1.2860859282617265, "grad_norm": 0.5444017648696899, "learning_rate": 0.0005, "loss": 1.6303184032440186, "step": 16314 }, { "epoch": 1.2861647615293654, "grad_norm": 0.6279269456863403, "learning_rate": 0.0005, "loss": 1.892164945602417, "step": 16315 }, { "epoch": 1.2862435947970043, "grad_norm": 0.5524676442146301, "learning_rate": 0.0005, "loss": 1.6944606304168701, "step": 16316 }, { "epoch": 1.2863224280646433, "grad_norm": 1.3035467863082886, "learning_rate": 0.0005, "loss": 1.6869343519210815, "step": 16317 }, { "epoch": 1.2864012613322822, "grad_norm": 0.6170306205749512, "learning_rate": 0.0005, "loss": 1.6833571195602417, "step": 16318 }, { "epoch": 1.2864800945999213, "grad_norm": 0.5265322327613831, "learning_rate": 0.0005, "loss": 1.661128044128418, "step": 16319 }, { "epoch": 1.28655892786756, "grad_norm": 0.4883571267127991, "learning_rate": 0.0005, "loss": 1.5870976448059082, "step": 16320 }, { "epoch": 1.2866377611351991, "grad_norm": 16.01200294494629, "learning_rate": 0.0005, "loss": 1.850595235824585, "step": 16321 }, { "epoch": 1.286716594402838, "grad_norm": 0.6847409605979919, "learning_rate": 0.0005, "loss": 1.782341718673706, "step": 16322 }, { "epoch": 1.286795427670477, "grad_norm": 0.5781711935997009, "learning_rate": 0.0005, "loss": 1.5247520208358765, "step": 16323 }, { "epoch": 1.2868742609381159, "grad_norm": 0.6852307915687561, "learning_rate": 0.0005, "loss": 1.6536154747009277, "step": 16324 }, { "epoch": 1.2869530942057548, "grad_norm": 0.6650007367134094, "learning_rate": 0.0005, "loss": 1.71049165725708, "step": 16325 }, { "epoch": 1.287031927473394, "grad_norm": 0.5325497388839722, "learning_rate": 0.0005, "loss": 1.6407721042633057, "step": 16326 }, { "epoch": 1.2871107607410326, "grad_norm": 0.5374517440795898, "learning_rate": 0.0005, "loss": 1.5519330501556396, "step": 16327 }, { "epoch": 1.2871895940086717, "grad_norm": 0.6400114893913269, "learning_rate": 0.0005, "loss": 1.6941099166870117, "step": 16328 }, { "epoch": 1.2872684272763106, "grad_norm": 0.545672595500946, "learning_rate": 0.0005, "loss": 1.7248737812042236, "step": 16329 }, { "epoch": 1.2873472605439495, "grad_norm": 0.795964241027832, "learning_rate": 0.0005, "loss": 1.8578765392303467, "step": 16330 }, { "epoch": 1.2874260938115885, "grad_norm": 0.5245641469955444, "learning_rate": 0.0005, "loss": 1.6607024669647217, "step": 16331 }, { "epoch": 1.2875049270792274, "grad_norm": 0.5061404705047607, "learning_rate": 0.0005, "loss": 1.6393086910247803, "step": 16332 }, { "epoch": 1.2875837603468665, "grad_norm": 0.5224198698997498, "learning_rate": 0.0005, "loss": 1.6905443668365479, "step": 16333 }, { "epoch": 1.2876625936145052, "grad_norm": 0.6106587052345276, "learning_rate": 0.0005, "loss": 1.8652927875518799, "step": 16334 }, { "epoch": 1.2877414268821443, "grad_norm": 0.5873441696166992, "learning_rate": 0.0005, "loss": 1.847324252128601, "step": 16335 }, { "epoch": 1.2878202601497832, "grad_norm": 0.47459957003593445, "learning_rate": 0.0005, "loss": 1.6776840686798096, "step": 16336 }, { "epoch": 1.2878990934174221, "grad_norm": 0.5023598074913025, "learning_rate": 0.0005, "loss": 1.5768334865570068, "step": 16337 }, { "epoch": 1.287977926685061, "grad_norm": 0.5916839241981506, "learning_rate": 0.0005, "loss": 1.8881845474243164, "step": 16338 }, { "epoch": 1.2880567599527, "grad_norm": 0.5908645987510681, "learning_rate": 0.0005, "loss": 1.7768484354019165, "step": 16339 }, { "epoch": 1.288135593220339, "grad_norm": 0.5737666487693787, "learning_rate": 0.0005, "loss": 1.717142105102539, "step": 16340 }, { "epoch": 1.2882144264879778, "grad_norm": 0.5212497711181641, "learning_rate": 0.0005, "loss": 1.6373107433319092, "step": 16341 }, { "epoch": 1.288293259755617, "grad_norm": 0.5014066100120544, "learning_rate": 0.0005, "loss": 1.6207921504974365, "step": 16342 }, { "epoch": 1.2883720930232558, "grad_norm": 0.552494466304779, "learning_rate": 0.0005, "loss": 1.7736167907714844, "step": 16343 }, { "epoch": 1.2884509262908947, "grad_norm": 0.5983884334564209, "learning_rate": 0.0005, "loss": 1.7661296129226685, "step": 16344 }, { "epoch": 1.2885297595585337, "grad_norm": 0.4881661832332611, "learning_rate": 0.0005, "loss": 1.682194471359253, "step": 16345 }, { "epoch": 1.2886085928261726, "grad_norm": 0.6152004599571228, "learning_rate": 0.0005, "loss": 1.6013237237930298, "step": 16346 }, { "epoch": 1.2886874260938117, "grad_norm": 0.7592200040817261, "learning_rate": 0.0005, "loss": 1.335212230682373, "step": 16347 }, { "epoch": 1.2887662593614504, "grad_norm": 0.544599711894989, "learning_rate": 0.0005, "loss": 1.687145709991455, "step": 16348 }, { "epoch": 1.2888450926290895, "grad_norm": 0.5504790544509888, "learning_rate": 0.0005, "loss": 1.651847243309021, "step": 16349 }, { "epoch": 1.2889239258967284, "grad_norm": 0.7547469139099121, "learning_rate": 0.0005, "loss": 1.9596636295318604, "step": 16350 }, { "epoch": 1.2890027591643674, "grad_norm": 0.5437620878219604, "learning_rate": 0.0005, "loss": 1.679384469985962, "step": 16351 }, { "epoch": 1.2890815924320063, "grad_norm": 0.5621730089187622, "learning_rate": 0.0005, "loss": 1.741833209991455, "step": 16352 }, { "epoch": 1.2891604256996452, "grad_norm": 0.6213497519493103, "learning_rate": 0.0005, "loss": 1.7524869441986084, "step": 16353 }, { "epoch": 1.2892392589672843, "grad_norm": 0.5001117587089539, "learning_rate": 0.0005, "loss": 1.583225965499878, "step": 16354 }, { "epoch": 1.2893180922349232, "grad_norm": 0.48096296191215515, "learning_rate": 0.0005, "loss": 1.5556178092956543, "step": 16355 }, { "epoch": 1.2893969255025621, "grad_norm": 0.49297192692756653, "learning_rate": 0.0005, "loss": 1.798917531967163, "step": 16356 }, { "epoch": 1.289475758770201, "grad_norm": 0.50123530626297, "learning_rate": 0.0005, "loss": 1.5139377117156982, "step": 16357 }, { "epoch": 1.28955459203784, "grad_norm": 0.5978092551231384, "learning_rate": 0.0005, "loss": 1.4569578170776367, "step": 16358 }, { "epoch": 1.2896334253054789, "grad_norm": 0.5744048953056335, "learning_rate": 0.0005, "loss": 1.5262960195541382, "step": 16359 }, { "epoch": 1.2897122585731178, "grad_norm": 0.4145336449146271, "learning_rate": 0.0005, "loss": 1.7156143188476562, "step": 16360 }, { "epoch": 1.289791091840757, "grad_norm": 0.5830739140510559, "learning_rate": 0.0005, "loss": 1.5830750465393066, "step": 16361 }, { "epoch": 1.2898699251083958, "grad_norm": 0.5337801575660706, "learning_rate": 0.0005, "loss": 1.9249563217163086, "step": 16362 }, { "epoch": 1.2899487583760347, "grad_norm": 2.14082670211792, "learning_rate": 0.0005, "loss": 1.516862392425537, "step": 16363 }, { "epoch": 1.2900275916436736, "grad_norm": 0.5674865245819092, "learning_rate": 0.0005, "loss": 1.6296546459197998, "step": 16364 }, { "epoch": 1.2901064249113126, "grad_norm": 0.5189149379730225, "learning_rate": 0.0005, "loss": 1.8481019735336304, "step": 16365 }, { "epoch": 1.2901852581789515, "grad_norm": 0.5039810538291931, "learning_rate": 0.0005, "loss": 1.7560245990753174, "step": 16366 }, { "epoch": 1.2902640914465904, "grad_norm": 0.5233887434005737, "learning_rate": 0.0005, "loss": 1.6696453094482422, "step": 16367 }, { "epoch": 1.2903429247142295, "grad_norm": 0.546486496925354, "learning_rate": 0.0005, "loss": 1.4451799392700195, "step": 16368 }, { "epoch": 1.2904217579818684, "grad_norm": 0.6476457715034485, "learning_rate": 0.0005, "loss": 1.9304240942001343, "step": 16369 }, { "epoch": 1.2905005912495073, "grad_norm": 0.5469260811805725, "learning_rate": 0.0005, "loss": 1.7339287996292114, "step": 16370 }, { "epoch": 1.2905794245171462, "grad_norm": 0.5124191045761108, "learning_rate": 0.0005, "loss": 1.68705415725708, "step": 16371 }, { "epoch": 1.2906582577847852, "grad_norm": 0.4956658184528351, "learning_rate": 0.0005, "loss": 1.5321204662322998, "step": 16372 }, { "epoch": 1.290737091052424, "grad_norm": 0.4795651435852051, "learning_rate": 0.0005, "loss": 1.7360751628875732, "step": 16373 }, { "epoch": 1.290815924320063, "grad_norm": 0.513078510761261, "learning_rate": 0.0005, "loss": 1.5783958435058594, "step": 16374 }, { "epoch": 1.2908947575877021, "grad_norm": 0.507374107837677, "learning_rate": 0.0005, "loss": 1.5364799499511719, "step": 16375 }, { "epoch": 1.290973590855341, "grad_norm": 0.553907036781311, "learning_rate": 0.0005, "loss": 1.6056790351867676, "step": 16376 }, { "epoch": 1.29105242412298, "grad_norm": 0.5542507767677307, "learning_rate": 0.0005, "loss": 1.792806625366211, "step": 16377 }, { "epoch": 1.2911312573906188, "grad_norm": 0.6021363735198975, "learning_rate": 0.0005, "loss": 1.8984241485595703, "step": 16378 }, { "epoch": 1.2912100906582578, "grad_norm": 0.4914703369140625, "learning_rate": 0.0005, "loss": 1.6284854412078857, "step": 16379 }, { "epoch": 1.2912889239258967, "grad_norm": 0.7228968739509583, "learning_rate": 0.0005, "loss": 1.8153311014175415, "step": 16380 }, { "epoch": 1.2913677571935356, "grad_norm": 0.48294320702552795, "learning_rate": 0.0005, "loss": 1.6997984647750854, "step": 16381 }, { "epoch": 1.2914465904611747, "grad_norm": 0.655809223651886, "learning_rate": 0.0005, "loss": 1.8074619770050049, "step": 16382 }, { "epoch": 1.2915254237288136, "grad_norm": 0.6176903247833252, "learning_rate": 0.0005, "loss": 1.8537839651107788, "step": 16383 }, { "epoch": 1.2916042569964525, "grad_norm": 0.47038641571998596, "learning_rate": 0.0005, "loss": 1.5543361902236938, "step": 16384 }, { "epoch": 1.2916830902640914, "grad_norm": 0.5295878648757935, "learning_rate": 0.0005, "loss": 1.62215256690979, "step": 16385 }, { "epoch": 1.2917619235317304, "grad_norm": 0.7638161778450012, "learning_rate": 0.0005, "loss": 1.7898728847503662, "step": 16386 }, { "epoch": 1.2918407567993693, "grad_norm": 0.44959357380867004, "learning_rate": 0.0005, "loss": 1.6504837274551392, "step": 16387 }, { "epoch": 1.2919195900670082, "grad_norm": 0.5720979571342468, "learning_rate": 0.0005, "loss": 1.6445108652114868, "step": 16388 }, { "epoch": 1.2919984233346473, "grad_norm": 0.4887177050113678, "learning_rate": 0.0005, "loss": 1.4288362264633179, "step": 16389 }, { "epoch": 1.2920772566022862, "grad_norm": 0.506751298904419, "learning_rate": 0.0005, "loss": 1.5750030279159546, "step": 16390 }, { "epoch": 1.2921560898699251, "grad_norm": 0.5923897624015808, "learning_rate": 0.0005, "loss": 1.545784831047058, "step": 16391 }, { "epoch": 1.292234923137564, "grad_norm": 0.4974520802497864, "learning_rate": 0.0005, "loss": 1.7456648349761963, "step": 16392 }, { "epoch": 1.292313756405203, "grad_norm": 0.5558609366416931, "learning_rate": 0.0005, "loss": 1.779484510421753, "step": 16393 }, { "epoch": 1.2923925896728419, "grad_norm": 0.5514588952064514, "learning_rate": 0.0005, "loss": 1.5295531749725342, "step": 16394 }, { "epoch": 1.2924714229404808, "grad_norm": 0.632116973400116, "learning_rate": 0.0005, "loss": 1.739088773727417, "step": 16395 }, { "epoch": 1.29255025620812, "grad_norm": 0.44967493414878845, "learning_rate": 0.0005, "loss": 1.4308230876922607, "step": 16396 }, { "epoch": 1.2926290894757588, "grad_norm": 0.45518702268600464, "learning_rate": 0.0005, "loss": 1.5509322881698608, "step": 16397 }, { "epoch": 1.2927079227433977, "grad_norm": 0.5505265593528748, "learning_rate": 0.0005, "loss": 1.5830917358398438, "step": 16398 }, { "epoch": 1.2927867560110367, "grad_norm": 0.6765403747558594, "learning_rate": 0.0005, "loss": 1.7184020280838013, "step": 16399 }, { "epoch": 1.2928655892786756, "grad_norm": 0.6030421257019043, "learning_rate": 0.0005, "loss": 1.6660385131835938, "step": 16400 }, { "epoch": 1.2929444225463145, "grad_norm": 0.5019797086715698, "learning_rate": 0.0005, "loss": 1.6643917560577393, "step": 16401 }, { "epoch": 1.2930232558139534, "grad_norm": 0.5535996556282043, "learning_rate": 0.0005, "loss": 1.6006593704223633, "step": 16402 }, { "epoch": 1.2931020890815925, "grad_norm": 0.47863271832466125, "learning_rate": 0.0005, "loss": 1.619985818862915, "step": 16403 }, { "epoch": 1.2931809223492314, "grad_norm": 0.48397934436798096, "learning_rate": 0.0005, "loss": 1.3899691104888916, "step": 16404 }, { "epoch": 1.2932597556168703, "grad_norm": 0.5347735285758972, "learning_rate": 0.0005, "loss": 1.817284107208252, "step": 16405 }, { "epoch": 1.2933385888845093, "grad_norm": 0.5725525617599487, "learning_rate": 0.0005, "loss": 1.498193383216858, "step": 16406 }, { "epoch": 1.2934174221521482, "grad_norm": 0.6080936789512634, "learning_rate": 0.0005, "loss": 1.570991039276123, "step": 16407 }, { "epoch": 1.293496255419787, "grad_norm": 0.5793853998184204, "learning_rate": 0.0005, "loss": 1.8931143283843994, "step": 16408 }, { "epoch": 1.293575088687426, "grad_norm": 0.5053015947341919, "learning_rate": 0.0005, "loss": 1.5646929740905762, "step": 16409 }, { "epoch": 1.2936539219550651, "grad_norm": 0.5315813422203064, "learning_rate": 0.0005, "loss": 1.5751423835754395, "step": 16410 }, { "epoch": 1.293732755222704, "grad_norm": 0.7644834518432617, "learning_rate": 0.0005, "loss": 1.7360477447509766, "step": 16411 }, { "epoch": 1.293811588490343, "grad_norm": 0.5947962403297424, "learning_rate": 0.0005, "loss": 1.5742120742797852, "step": 16412 }, { "epoch": 1.2938904217579819, "grad_norm": 0.48568660020828247, "learning_rate": 0.0005, "loss": 1.4982690811157227, "step": 16413 }, { "epoch": 1.2939692550256208, "grad_norm": 0.563490092754364, "learning_rate": 0.0005, "loss": 1.5753848552703857, "step": 16414 }, { "epoch": 1.2940480882932597, "grad_norm": 0.5152623057365417, "learning_rate": 0.0005, "loss": 1.5425347089767456, "step": 16415 }, { "epoch": 1.2941269215608986, "grad_norm": 0.46243906021118164, "learning_rate": 0.0005, "loss": 1.353243350982666, "step": 16416 }, { "epoch": 1.2942057548285377, "grad_norm": 0.5337766408920288, "learning_rate": 0.0005, "loss": 1.6349900960922241, "step": 16417 }, { "epoch": 1.2942845880961766, "grad_norm": 0.8215436339378357, "learning_rate": 0.0005, "loss": 1.8140833377838135, "step": 16418 }, { "epoch": 1.2943634213638155, "grad_norm": 0.5969615578651428, "learning_rate": 0.0005, "loss": 1.8535529375076294, "step": 16419 }, { "epoch": 1.2944422546314545, "grad_norm": 0.6097906231880188, "learning_rate": 0.0005, "loss": 1.6660380363464355, "step": 16420 }, { "epoch": 1.2945210878990934, "grad_norm": 0.5097244381904602, "learning_rate": 0.0005, "loss": 1.6855031251907349, "step": 16421 }, { "epoch": 1.2945999211667323, "grad_norm": 0.6213039755821228, "learning_rate": 0.0005, "loss": 1.8037577867507935, "step": 16422 }, { "epoch": 1.2946787544343712, "grad_norm": 0.4582936763763428, "learning_rate": 0.0005, "loss": 1.6120946407318115, "step": 16423 }, { "epoch": 1.2947575877020103, "grad_norm": 0.4898034632205963, "learning_rate": 0.0005, "loss": 1.4575724601745605, "step": 16424 }, { "epoch": 1.2948364209696492, "grad_norm": 0.5378584861755371, "learning_rate": 0.0005, "loss": 1.5437235832214355, "step": 16425 }, { "epoch": 1.2949152542372881, "grad_norm": 0.4376225173473358, "learning_rate": 0.0005, "loss": 1.6075105667114258, "step": 16426 }, { "epoch": 1.294994087504927, "grad_norm": 0.7222475409507751, "learning_rate": 0.0005, "loss": 1.793383240699768, "step": 16427 }, { "epoch": 1.295072920772566, "grad_norm": 0.4641386866569519, "learning_rate": 0.0005, "loss": 1.601355791091919, "step": 16428 }, { "epoch": 1.2951517540402049, "grad_norm": 0.44347435235977173, "learning_rate": 0.0005, "loss": 1.6170252561569214, "step": 16429 }, { "epoch": 1.2952305873078438, "grad_norm": 0.7955272197723389, "learning_rate": 0.0005, "loss": 1.7373125553131104, "step": 16430 }, { "epoch": 1.295309420575483, "grad_norm": 0.5845488905906677, "learning_rate": 0.0005, "loss": 1.6789546012878418, "step": 16431 }, { "epoch": 1.2953882538431218, "grad_norm": 0.6171836256980896, "learning_rate": 0.0005, "loss": 1.5892338752746582, "step": 16432 }, { "epoch": 1.2954670871107608, "grad_norm": 0.6681563854217529, "learning_rate": 0.0005, "loss": 1.7925530672073364, "step": 16433 }, { "epoch": 1.2955459203783997, "grad_norm": 0.7145748138427734, "learning_rate": 0.0005, "loss": 1.6790666580200195, "step": 16434 }, { "epoch": 1.2956247536460386, "grad_norm": 0.9547610282897949, "learning_rate": 0.0005, "loss": 1.617977499961853, "step": 16435 }, { "epoch": 1.2957035869136777, "grad_norm": 0.552197277545929, "learning_rate": 0.0005, "loss": 1.6040855646133423, "step": 16436 }, { "epoch": 1.2957824201813164, "grad_norm": 0.5074076652526855, "learning_rate": 0.0005, "loss": 1.8262825012207031, "step": 16437 }, { "epoch": 1.2958612534489555, "grad_norm": 0.6118460893630981, "learning_rate": 0.0005, "loss": 1.7807928323745728, "step": 16438 }, { "epoch": 1.2959400867165944, "grad_norm": 0.5310564041137695, "learning_rate": 0.0005, "loss": 1.598615288734436, "step": 16439 }, { "epoch": 1.2960189199842334, "grad_norm": 0.5470342040061951, "learning_rate": 0.0005, "loss": 1.6861672401428223, "step": 16440 }, { "epoch": 1.2960977532518723, "grad_norm": 0.5586937665939331, "learning_rate": 0.0005, "loss": 1.7444407939910889, "step": 16441 }, { "epoch": 1.2961765865195112, "grad_norm": 0.5084741711616516, "learning_rate": 0.0005, "loss": 1.684358835220337, "step": 16442 }, { "epoch": 1.2962554197871503, "grad_norm": 0.608423113822937, "learning_rate": 0.0005, "loss": 1.5308525562286377, "step": 16443 }, { "epoch": 1.296334253054789, "grad_norm": 0.4958801865577698, "learning_rate": 0.0005, "loss": 1.585374116897583, "step": 16444 }, { "epoch": 1.2964130863224281, "grad_norm": 0.49316102266311646, "learning_rate": 0.0005, "loss": 1.7207884788513184, "step": 16445 }, { "epoch": 1.296491919590067, "grad_norm": 0.6603737473487854, "learning_rate": 0.0005, "loss": 1.5887523889541626, "step": 16446 }, { "epoch": 1.296570752857706, "grad_norm": 0.6745709180831909, "learning_rate": 0.0005, "loss": 1.8270115852355957, "step": 16447 }, { "epoch": 1.2966495861253449, "grad_norm": 0.6898952722549438, "learning_rate": 0.0005, "loss": 1.8393398523330688, "step": 16448 }, { "epoch": 1.2967284193929838, "grad_norm": 0.656644880771637, "learning_rate": 0.0005, "loss": 1.8570177555084229, "step": 16449 }, { "epoch": 1.296807252660623, "grad_norm": 0.5532902479171753, "learning_rate": 0.0005, "loss": 1.8526921272277832, "step": 16450 }, { "epoch": 1.2968860859282616, "grad_norm": 0.5031512379646301, "learning_rate": 0.0005, "loss": 1.7040010690689087, "step": 16451 }, { "epoch": 1.2969649191959007, "grad_norm": 0.5265820026397705, "learning_rate": 0.0005, "loss": 1.4999265670776367, "step": 16452 }, { "epoch": 1.2970437524635396, "grad_norm": 0.6039464473724365, "learning_rate": 0.0005, "loss": 1.7987321615219116, "step": 16453 }, { "epoch": 1.2971225857311786, "grad_norm": 0.5303980112075806, "learning_rate": 0.0005, "loss": 1.4994250535964966, "step": 16454 }, { "epoch": 1.2972014189988175, "grad_norm": 0.7804281115531921, "learning_rate": 0.0005, "loss": 1.609020709991455, "step": 16455 }, { "epoch": 1.2972802522664564, "grad_norm": 0.536757230758667, "learning_rate": 0.0005, "loss": 1.8602068424224854, "step": 16456 }, { "epoch": 1.2973590855340955, "grad_norm": 0.5023003816604614, "learning_rate": 0.0005, "loss": 1.5678857564926147, "step": 16457 }, { "epoch": 1.2974379188017342, "grad_norm": 1.1141043901443481, "learning_rate": 0.0005, "loss": 1.6477506160736084, "step": 16458 }, { "epoch": 1.2975167520693733, "grad_norm": 0.541527509689331, "learning_rate": 0.0005, "loss": 1.6401910781860352, "step": 16459 }, { "epoch": 1.2975955853370122, "grad_norm": 0.6550244092941284, "learning_rate": 0.0005, "loss": 1.7138041257858276, "step": 16460 }, { "epoch": 1.2976744186046512, "grad_norm": 0.5426862835884094, "learning_rate": 0.0005, "loss": 1.6892063617706299, "step": 16461 }, { "epoch": 1.29775325187229, "grad_norm": 0.4943287670612335, "learning_rate": 0.0005, "loss": 1.6366794109344482, "step": 16462 }, { "epoch": 1.297832085139929, "grad_norm": 0.47967082262039185, "learning_rate": 0.0005, "loss": 1.7338883876800537, "step": 16463 }, { "epoch": 1.2979109184075681, "grad_norm": 0.5015673041343689, "learning_rate": 0.0005, "loss": 1.7489317655563354, "step": 16464 }, { "epoch": 1.2979897516752068, "grad_norm": 0.5420117378234863, "learning_rate": 0.0005, "loss": 1.5318331718444824, "step": 16465 }, { "epoch": 1.298068584942846, "grad_norm": 0.46765533089637756, "learning_rate": 0.0005, "loss": 1.7042843103408813, "step": 16466 }, { "epoch": 1.2981474182104848, "grad_norm": 0.5175180435180664, "learning_rate": 0.0005, "loss": 1.4765293598175049, "step": 16467 }, { "epoch": 1.2982262514781238, "grad_norm": 0.5217573046684265, "learning_rate": 0.0005, "loss": 1.6932954788208008, "step": 16468 }, { "epoch": 1.2983050847457627, "grad_norm": 0.5953305959701538, "learning_rate": 0.0005, "loss": 1.5005356073379517, "step": 16469 }, { "epoch": 1.2983839180134016, "grad_norm": 0.5686867237091064, "learning_rate": 0.0005, "loss": 1.7587065696716309, "step": 16470 }, { "epoch": 1.2984627512810407, "grad_norm": 0.5459917187690735, "learning_rate": 0.0005, "loss": 1.8121557235717773, "step": 16471 }, { "epoch": 1.2985415845486794, "grad_norm": 0.5470423102378845, "learning_rate": 0.0005, "loss": 1.7487107515335083, "step": 16472 }, { "epoch": 1.2986204178163185, "grad_norm": 0.47488197684288025, "learning_rate": 0.0005, "loss": 1.7290961742401123, "step": 16473 }, { "epoch": 1.2986992510839575, "grad_norm": 0.5551976561546326, "learning_rate": 0.0005, "loss": 1.5554641485214233, "step": 16474 }, { "epoch": 1.2987780843515964, "grad_norm": 0.5308830738067627, "learning_rate": 0.0005, "loss": 1.4924910068511963, "step": 16475 }, { "epoch": 1.2988569176192353, "grad_norm": 0.48452940583229065, "learning_rate": 0.0005, "loss": 1.656510353088379, "step": 16476 }, { "epoch": 1.2989357508868742, "grad_norm": 0.5547462105751038, "learning_rate": 0.0005, "loss": 1.7337956428527832, "step": 16477 }, { "epoch": 1.2990145841545133, "grad_norm": 0.5447864532470703, "learning_rate": 0.0005, "loss": 1.7663178443908691, "step": 16478 }, { "epoch": 1.2990934174221522, "grad_norm": 0.6184216141700745, "learning_rate": 0.0005, "loss": 1.488558292388916, "step": 16479 }, { "epoch": 1.2991722506897911, "grad_norm": 0.508578896522522, "learning_rate": 0.0005, "loss": 1.6802784204483032, "step": 16480 }, { "epoch": 1.29925108395743, "grad_norm": 0.5732950568199158, "learning_rate": 0.0005, "loss": 1.8662163019180298, "step": 16481 }, { "epoch": 1.299329917225069, "grad_norm": 0.6635144352912903, "learning_rate": 0.0005, "loss": 1.6510305404663086, "step": 16482 }, { "epoch": 1.2994087504927079, "grad_norm": 0.5297288298606873, "learning_rate": 0.0005, "loss": 1.7481062412261963, "step": 16483 }, { "epoch": 1.2994875837603468, "grad_norm": 0.6482883095741272, "learning_rate": 0.0005, "loss": 1.9993966817855835, "step": 16484 }, { "epoch": 1.299566417027986, "grad_norm": 0.5399617552757263, "learning_rate": 0.0005, "loss": 1.676367998123169, "step": 16485 }, { "epoch": 1.2996452502956248, "grad_norm": 0.5612931251525879, "learning_rate": 0.0005, "loss": 1.77274751663208, "step": 16486 }, { "epoch": 1.2997240835632637, "grad_norm": 0.7505900859832764, "learning_rate": 0.0005, "loss": 1.517871379852295, "step": 16487 }, { "epoch": 1.2998029168309027, "grad_norm": 0.7119584083557129, "learning_rate": 0.0005, "loss": 1.5842676162719727, "step": 16488 }, { "epoch": 1.2998817500985416, "grad_norm": 0.5336191058158875, "learning_rate": 0.0005, "loss": 1.6681289672851562, "step": 16489 }, { "epoch": 1.2999605833661805, "grad_norm": 0.597244918346405, "learning_rate": 0.0005, "loss": 1.770667314529419, "step": 16490 }, { "epoch": 1.3000394166338194, "grad_norm": 0.5836706161499023, "learning_rate": 0.0005, "loss": 1.8489633798599243, "step": 16491 }, { "epoch": 1.3001182499014585, "grad_norm": 0.5647438168525696, "learning_rate": 0.0005, "loss": 1.8109641075134277, "step": 16492 }, { "epoch": 1.3001970831690974, "grad_norm": 5.088593482971191, "learning_rate": 0.0005, "loss": 1.5451948642730713, "step": 16493 }, { "epoch": 1.3002759164367363, "grad_norm": 0.6209520697593689, "learning_rate": 0.0005, "loss": 1.5262287855148315, "step": 16494 }, { "epoch": 1.3003547497043753, "grad_norm": 0.48073604702949524, "learning_rate": 0.0005, "loss": 1.5958945751190186, "step": 16495 }, { "epoch": 1.3004335829720142, "grad_norm": 0.43343666195869446, "learning_rate": 0.0005, "loss": 1.4667503833770752, "step": 16496 }, { "epoch": 1.300512416239653, "grad_norm": 0.5582642555236816, "learning_rate": 0.0005, "loss": 1.5558621883392334, "step": 16497 }, { "epoch": 1.300591249507292, "grad_norm": 0.465750515460968, "learning_rate": 0.0005, "loss": 1.6559756994247437, "step": 16498 }, { "epoch": 1.3006700827749311, "grad_norm": 0.5355920195579529, "learning_rate": 0.0005, "loss": 1.5059690475463867, "step": 16499 }, { "epoch": 1.30074891604257, "grad_norm": 0.5853081941604614, "learning_rate": 0.0005, "loss": 1.5787513256072998, "step": 16500 }, { "epoch": 1.300827749310209, "grad_norm": 0.6450321674346924, "learning_rate": 0.0005, "loss": 1.804452896118164, "step": 16501 }, { "epoch": 1.3009065825778479, "grad_norm": 0.5236570239067078, "learning_rate": 0.0005, "loss": 1.642075538635254, "step": 16502 }, { "epoch": 1.3009854158454868, "grad_norm": 0.514921247959137, "learning_rate": 0.0005, "loss": 1.5272529125213623, "step": 16503 }, { "epoch": 1.3010642491131257, "grad_norm": 0.6028572916984558, "learning_rate": 0.0005, "loss": 1.6308190822601318, "step": 16504 }, { "epoch": 1.3011430823807646, "grad_norm": 0.5548992156982422, "learning_rate": 0.0005, "loss": 1.6880813837051392, "step": 16505 }, { "epoch": 1.3012219156484037, "grad_norm": 0.5676562190055847, "learning_rate": 0.0005, "loss": 1.6671855449676514, "step": 16506 }, { "epoch": 1.3013007489160426, "grad_norm": 0.5255846381187439, "learning_rate": 0.0005, "loss": 1.6387789249420166, "step": 16507 }, { "epoch": 1.3013795821836815, "grad_norm": 0.7272902131080627, "learning_rate": 0.0005, "loss": 1.769747257232666, "step": 16508 }, { "epoch": 1.3014584154513205, "grad_norm": 0.6549646854400635, "learning_rate": 0.0005, "loss": 1.888757586479187, "step": 16509 }, { "epoch": 1.3015372487189594, "grad_norm": 0.6260742545127869, "learning_rate": 0.0005, "loss": 1.6701679229736328, "step": 16510 }, { "epoch": 1.3016160819865983, "grad_norm": 0.7238055467605591, "learning_rate": 0.0005, "loss": 1.9897434711456299, "step": 16511 }, { "epoch": 1.3016949152542372, "grad_norm": 0.5741068124771118, "learning_rate": 0.0005, "loss": 1.6457746028900146, "step": 16512 }, { "epoch": 1.3017737485218763, "grad_norm": 0.5521218180656433, "learning_rate": 0.0005, "loss": 1.68754243850708, "step": 16513 }, { "epoch": 1.3018525817895152, "grad_norm": 0.5568358898162842, "learning_rate": 0.0005, "loss": 1.6557074785232544, "step": 16514 }, { "epoch": 1.3019314150571542, "grad_norm": 0.7782950401306152, "learning_rate": 0.0005, "loss": 1.8793107271194458, "step": 16515 }, { "epoch": 1.302010248324793, "grad_norm": 0.49223360419273376, "learning_rate": 0.0005, "loss": 1.5591340065002441, "step": 16516 }, { "epoch": 1.302089081592432, "grad_norm": 0.6507790684700012, "learning_rate": 0.0005, "loss": 1.8560645580291748, "step": 16517 }, { "epoch": 1.3021679148600709, "grad_norm": 0.6846305131912231, "learning_rate": 0.0005, "loss": 1.8896962404251099, "step": 16518 }, { "epoch": 1.3022467481277098, "grad_norm": 0.5893647074699402, "learning_rate": 0.0005, "loss": 1.7395493984222412, "step": 16519 }, { "epoch": 1.302325581395349, "grad_norm": 0.5063155889511108, "learning_rate": 0.0005, "loss": 1.709012508392334, "step": 16520 }, { "epoch": 1.3024044146629878, "grad_norm": 0.5247663259506226, "learning_rate": 0.0005, "loss": 1.5221636295318604, "step": 16521 }, { "epoch": 1.3024832479306268, "grad_norm": 1.0296905040740967, "learning_rate": 0.0005, "loss": 1.7163246870040894, "step": 16522 }, { "epoch": 1.3025620811982657, "grad_norm": 0.6706163287162781, "learning_rate": 0.0005, "loss": 1.7006111145019531, "step": 16523 }, { "epoch": 1.3026409144659046, "grad_norm": 0.5633531212806702, "learning_rate": 0.0005, "loss": 1.687646746635437, "step": 16524 }, { "epoch": 1.3027197477335435, "grad_norm": 0.6498264074325562, "learning_rate": 0.0005, "loss": 1.768587589263916, "step": 16525 }, { "epoch": 1.3027985810011824, "grad_norm": 0.5771600008010864, "learning_rate": 0.0005, "loss": 1.698974609375, "step": 16526 }, { "epoch": 1.3028774142688215, "grad_norm": 0.522480309009552, "learning_rate": 0.0005, "loss": 1.4988393783569336, "step": 16527 }, { "epoch": 1.3029562475364604, "grad_norm": 0.5199746489524841, "learning_rate": 0.0005, "loss": 1.594977855682373, "step": 16528 }, { "epoch": 1.3030350808040994, "grad_norm": 0.6729859709739685, "learning_rate": 0.0005, "loss": 1.567960262298584, "step": 16529 }, { "epoch": 1.3031139140717383, "grad_norm": 0.5088810920715332, "learning_rate": 0.0005, "loss": 1.4770948886871338, "step": 16530 }, { "epoch": 1.3031927473393772, "grad_norm": 0.6297950148582458, "learning_rate": 0.0005, "loss": 1.7180917263031006, "step": 16531 }, { "epoch": 1.303271580607016, "grad_norm": 0.4420780837535858, "learning_rate": 0.0005, "loss": 1.4431679248809814, "step": 16532 }, { "epoch": 1.303350413874655, "grad_norm": 0.5419628024101257, "learning_rate": 0.0005, "loss": 1.872476577758789, "step": 16533 }, { "epoch": 1.3034292471422941, "grad_norm": 0.571617603302002, "learning_rate": 0.0005, "loss": 1.7049431800842285, "step": 16534 }, { "epoch": 1.303508080409933, "grad_norm": 0.6496443152427673, "learning_rate": 0.0005, "loss": 1.7814576625823975, "step": 16535 }, { "epoch": 1.303586913677572, "grad_norm": 0.5768800377845764, "learning_rate": 0.0005, "loss": 1.8656210899353027, "step": 16536 }, { "epoch": 1.3036657469452109, "grad_norm": 0.5643390417098999, "learning_rate": 0.0005, "loss": 1.5223551988601685, "step": 16537 }, { "epoch": 1.3037445802128498, "grad_norm": 0.6196240186691284, "learning_rate": 0.0005, "loss": 1.518166422843933, "step": 16538 }, { "epoch": 1.3038234134804887, "grad_norm": 0.5675488710403442, "learning_rate": 0.0005, "loss": 1.7160744667053223, "step": 16539 }, { "epoch": 1.3039022467481276, "grad_norm": 0.46216922998428345, "learning_rate": 0.0005, "loss": 1.648374080657959, "step": 16540 }, { "epoch": 1.3039810800157667, "grad_norm": 0.5199292898178101, "learning_rate": 0.0005, "loss": 1.6295766830444336, "step": 16541 }, { "epoch": 1.3040599132834056, "grad_norm": 0.4014875292778015, "learning_rate": 0.0005, "loss": 1.538421392440796, "step": 16542 }, { "epoch": 1.3041387465510446, "grad_norm": 0.6437225341796875, "learning_rate": 0.0005, "loss": 1.9294307231903076, "step": 16543 }, { "epoch": 1.3042175798186835, "grad_norm": 0.5470113754272461, "learning_rate": 0.0005, "loss": 1.6357855796813965, "step": 16544 }, { "epoch": 1.3042964130863224, "grad_norm": 0.43244385719299316, "learning_rate": 0.0005, "loss": 1.5685502290725708, "step": 16545 }, { "epoch": 1.3043752463539613, "grad_norm": 0.5897993445396423, "learning_rate": 0.0005, "loss": 1.6375069618225098, "step": 16546 }, { "epoch": 1.3044540796216002, "grad_norm": 0.6005218625068665, "learning_rate": 0.0005, "loss": 1.6696141958236694, "step": 16547 }, { "epoch": 1.3045329128892393, "grad_norm": 0.5736899971961975, "learning_rate": 0.0005, "loss": 1.6508105993270874, "step": 16548 }, { "epoch": 1.3046117461568782, "grad_norm": 0.5103191137313843, "learning_rate": 0.0005, "loss": 1.8238182067871094, "step": 16549 }, { "epoch": 1.3046905794245172, "grad_norm": 0.5729617476463318, "learning_rate": 0.0005, "loss": 1.650259256362915, "step": 16550 }, { "epoch": 1.304769412692156, "grad_norm": 0.4562944173812866, "learning_rate": 0.0005, "loss": 1.7299106121063232, "step": 16551 }, { "epoch": 1.304848245959795, "grad_norm": 0.5080835819244385, "learning_rate": 0.0005, "loss": 1.6574668884277344, "step": 16552 }, { "epoch": 1.304927079227434, "grad_norm": 0.6735987663269043, "learning_rate": 0.0005, "loss": 1.6159980297088623, "step": 16553 }, { "epoch": 1.3050059124950728, "grad_norm": 0.8239535689353943, "learning_rate": 0.0005, "loss": 1.5423527956008911, "step": 16554 }, { "epoch": 1.305084745762712, "grad_norm": 0.5630528330802917, "learning_rate": 0.0005, "loss": 1.839966058731079, "step": 16555 }, { "epoch": 1.3051635790303509, "grad_norm": 0.5009517669677734, "learning_rate": 0.0005, "loss": 1.525788426399231, "step": 16556 }, { "epoch": 1.3052424122979898, "grad_norm": 0.6878223419189453, "learning_rate": 0.0005, "loss": 1.9717459678649902, "step": 16557 }, { "epoch": 1.3053212455656287, "grad_norm": 0.7346689105033875, "learning_rate": 0.0005, "loss": 1.5360021591186523, "step": 16558 }, { "epoch": 1.3054000788332676, "grad_norm": 0.5045962333679199, "learning_rate": 0.0005, "loss": 1.4146320819854736, "step": 16559 }, { "epoch": 1.3054789121009067, "grad_norm": 0.6038016676902771, "learning_rate": 0.0005, "loss": 1.464311122894287, "step": 16560 }, { "epoch": 1.3055577453685454, "grad_norm": 0.5650172829627991, "learning_rate": 0.0005, "loss": 1.6847140789031982, "step": 16561 }, { "epoch": 1.3056365786361845, "grad_norm": 0.529643714427948, "learning_rate": 0.0005, "loss": 1.7571660280227661, "step": 16562 }, { "epoch": 1.3057154119038235, "grad_norm": 0.6200976967811584, "learning_rate": 0.0005, "loss": 1.655521273612976, "step": 16563 }, { "epoch": 1.3057942451714624, "grad_norm": 0.6499121785163879, "learning_rate": 0.0005, "loss": 1.673454999923706, "step": 16564 }, { "epoch": 1.3058730784391013, "grad_norm": 3.9890737533569336, "learning_rate": 0.0005, "loss": 1.6139161586761475, "step": 16565 }, { "epoch": 1.3059519117067402, "grad_norm": 0.5173966288566589, "learning_rate": 0.0005, "loss": 1.5399249792099, "step": 16566 }, { "epoch": 1.3060307449743793, "grad_norm": 0.5042251944541931, "learning_rate": 0.0005, "loss": 1.6701536178588867, "step": 16567 }, { "epoch": 1.306109578242018, "grad_norm": 0.5478310585021973, "learning_rate": 0.0005, "loss": 1.568279504776001, "step": 16568 }, { "epoch": 1.3061884115096571, "grad_norm": 0.5145899653434753, "learning_rate": 0.0005, "loss": 1.423869252204895, "step": 16569 }, { "epoch": 1.306267244777296, "grad_norm": 0.506247341632843, "learning_rate": 0.0005, "loss": 1.8938766717910767, "step": 16570 }, { "epoch": 1.306346078044935, "grad_norm": 0.4863179922103882, "learning_rate": 0.0005, "loss": 1.421806812286377, "step": 16571 }, { "epoch": 1.3064249113125739, "grad_norm": 0.47562646865844727, "learning_rate": 0.0005, "loss": 1.7217822074890137, "step": 16572 }, { "epoch": 1.3065037445802128, "grad_norm": 0.5107959508895874, "learning_rate": 0.0005, "loss": 1.557779312133789, "step": 16573 }, { "epoch": 1.306582577847852, "grad_norm": 0.49538227915763855, "learning_rate": 0.0005, "loss": 1.6010743379592896, "step": 16574 }, { "epoch": 1.3066614111154906, "grad_norm": 0.6216931939125061, "learning_rate": 0.0005, "loss": 1.660902976989746, "step": 16575 }, { "epoch": 1.3067402443831297, "grad_norm": 0.7445635795593262, "learning_rate": 0.0005, "loss": 1.89307701587677, "step": 16576 }, { "epoch": 1.3068190776507687, "grad_norm": 0.5279483199119568, "learning_rate": 0.0005, "loss": 1.4990439414978027, "step": 16577 }, { "epoch": 1.3068979109184076, "grad_norm": 0.6056724190711975, "learning_rate": 0.0005, "loss": 1.8303966522216797, "step": 16578 }, { "epoch": 1.3069767441860465, "grad_norm": 0.7372734546661377, "learning_rate": 0.0005, "loss": 1.5223283767700195, "step": 16579 }, { "epoch": 1.3070555774536854, "grad_norm": 0.6665064692497253, "learning_rate": 0.0005, "loss": 1.618194818496704, "step": 16580 }, { "epoch": 1.3071344107213245, "grad_norm": 0.5886832475662231, "learning_rate": 0.0005, "loss": 1.7088483572006226, "step": 16581 }, { "epoch": 1.3072132439889632, "grad_norm": 0.5469506978988647, "learning_rate": 0.0005, "loss": 1.7335875034332275, "step": 16582 }, { "epoch": 1.3072920772566023, "grad_norm": 0.5555254817008972, "learning_rate": 0.0005, "loss": 1.583950400352478, "step": 16583 }, { "epoch": 1.3073709105242413, "grad_norm": 0.5100095868110657, "learning_rate": 0.0005, "loss": 1.766172170639038, "step": 16584 }, { "epoch": 1.3074497437918802, "grad_norm": 0.4176933169364929, "learning_rate": 0.0005, "loss": 1.4703989028930664, "step": 16585 }, { "epoch": 1.307528577059519, "grad_norm": 0.6181454658508301, "learning_rate": 0.0005, "loss": 1.5452044010162354, "step": 16586 }, { "epoch": 1.307607410327158, "grad_norm": 0.7191962003707886, "learning_rate": 0.0005, "loss": 1.6487678289413452, "step": 16587 }, { "epoch": 1.3076862435947971, "grad_norm": 0.5528697371482849, "learning_rate": 0.0005, "loss": 1.698054552078247, "step": 16588 }, { "epoch": 1.3077650768624358, "grad_norm": 0.5218310952186584, "learning_rate": 0.0005, "loss": 1.5895963907241821, "step": 16589 }, { "epoch": 1.307843910130075, "grad_norm": 0.4797791540622711, "learning_rate": 0.0005, "loss": 1.6479978561401367, "step": 16590 }, { "epoch": 1.3079227433977139, "grad_norm": 0.5100741982460022, "learning_rate": 0.0005, "loss": 1.4108035564422607, "step": 16591 }, { "epoch": 1.3080015766653528, "grad_norm": 0.6479815244674683, "learning_rate": 0.0005, "loss": 1.5493690967559814, "step": 16592 }, { "epoch": 1.3080804099329917, "grad_norm": 0.49943360686302185, "learning_rate": 0.0005, "loss": 1.727186918258667, "step": 16593 }, { "epoch": 1.3081592432006306, "grad_norm": 0.5094314813613892, "learning_rate": 0.0005, "loss": 1.721055030822754, "step": 16594 }, { "epoch": 1.3082380764682697, "grad_norm": 0.6375869512557983, "learning_rate": 0.0005, "loss": 1.4606609344482422, "step": 16595 }, { "epoch": 1.3083169097359084, "grad_norm": 0.7032809853553772, "learning_rate": 0.0005, "loss": 1.8193601369857788, "step": 16596 }, { "epoch": 1.3083957430035476, "grad_norm": 0.8763865232467651, "learning_rate": 0.0005, "loss": 1.6711679697036743, "step": 16597 }, { "epoch": 1.3084745762711865, "grad_norm": 0.6953641176223755, "learning_rate": 0.0005, "loss": 1.7241363525390625, "step": 16598 }, { "epoch": 1.3085534095388254, "grad_norm": 0.6242073774337769, "learning_rate": 0.0005, "loss": 1.738696575164795, "step": 16599 }, { "epoch": 1.3086322428064643, "grad_norm": 0.6926101446151733, "learning_rate": 0.0005, "loss": 1.8260849714279175, "step": 16600 }, { "epoch": 1.3087110760741032, "grad_norm": 0.6419817805290222, "learning_rate": 0.0005, "loss": 1.8582664728164673, "step": 16601 }, { "epoch": 1.3087899093417423, "grad_norm": 0.5027973055839539, "learning_rate": 0.0005, "loss": 1.5909812450408936, "step": 16602 }, { "epoch": 1.3088687426093812, "grad_norm": 0.5886250734329224, "learning_rate": 0.0005, "loss": 1.7745368480682373, "step": 16603 }, { "epoch": 1.3089475758770202, "grad_norm": 0.5512305498123169, "learning_rate": 0.0005, "loss": 1.7886945009231567, "step": 16604 }, { "epoch": 1.309026409144659, "grad_norm": 0.5837420225143433, "learning_rate": 0.0005, "loss": 1.7286512851715088, "step": 16605 }, { "epoch": 1.309105242412298, "grad_norm": 0.5237575769424438, "learning_rate": 0.0005, "loss": 1.6171908378601074, "step": 16606 }, { "epoch": 1.3091840756799369, "grad_norm": 0.487793505191803, "learning_rate": 0.0005, "loss": 1.7827109098434448, "step": 16607 }, { "epoch": 1.3092629089475758, "grad_norm": 0.7591241598129272, "learning_rate": 0.0005, "loss": 2.0887060165405273, "step": 16608 }, { "epoch": 1.309341742215215, "grad_norm": 0.5040878057479858, "learning_rate": 0.0005, "loss": 1.5447137355804443, "step": 16609 }, { "epoch": 1.3094205754828538, "grad_norm": 0.5211352109909058, "learning_rate": 0.0005, "loss": 1.7354092597961426, "step": 16610 }, { "epoch": 1.3094994087504928, "grad_norm": 0.5727187395095825, "learning_rate": 0.0005, "loss": 1.5411767959594727, "step": 16611 }, { "epoch": 1.3095782420181317, "grad_norm": 0.5982409119606018, "learning_rate": 0.0005, "loss": 1.7263747453689575, "step": 16612 }, { "epoch": 1.3096570752857706, "grad_norm": 0.732428252696991, "learning_rate": 0.0005, "loss": 1.706735372543335, "step": 16613 }, { "epoch": 1.3097359085534095, "grad_norm": 0.4998161196708679, "learning_rate": 0.0005, "loss": 1.5176231861114502, "step": 16614 }, { "epoch": 1.3098147418210484, "grad_norm": 0.49868059158325195, "learning_rate": 0.0005, "loss": 1.4826295375823975, "step": 16615 }, { "epoch": 1.3098935750886875, "grad_norm": 0.5986575484275818, "learning_rate": 0.0005, "loss": 1.9196970462799072, "step": 16616 }, { "epoch": 1.3099724083563264, "grad_norm": 0.4196893870830536, "learning_rate": 0.0005, "loss": 1.5752184391021729, "step": 16617 }, { "epoch": 1.3100512416239654, "grad_norm": 0.6125084161758423, "learning_rate": 0.0005, "loss": 1.940888524055481, "step": 16618 }, { "epoch": 1.3101300748916043, "grad_norm": 0.5007608532905579, "learning_rate": 0.0005, "loss": 1.8693033456802368, "step": 16619 }, { "epoch": 1.3102089081592432, "grad_norm": 0.646122395992279, "learning_rate": 0.0005, "loss": 1.7109228372573853, "step": 16620 }, { "epoch": 1.310287741426882, "grad_norm": 0.6651207804679871, "learning_rate": 0.0005, "loss": 1.4960167407989502, "step": 16621 }, { "epoch": 1.310366574694521, "grad_norm": 2.206305503845215, "learning_rate": 0.0005, "loss": 1.7059805393218994, "step": 16622 }, { "epoch": 1.3104454079621601, "grad_norm": 0.5540555715560913, "learning_rate": 0.0005, "loss": 1.9402765035629272, "step": 16623 }, { "epoch": 1.310524241229799, "grad_norm": 0.4421805739402771, "learning_rate": 0.0005, "loss": 1.5092287063598633, "step": 16624 }, { "epoch": 1.310603074497438, "grad_norm": 0.7199657559394836, "learning_rate": 0.0005, "loss": 1.9401928186416626, "step": 16625 }, { "epoch": 1.3106819077650769, "grad_norm": 0.6066240668296814, "learning_rate": 0.0005, "loss": 1.8666657209396362, "step": 16626 }, { "epoch": 1.3107607410327158, "grad_norm": 0.6084010601043701, "learning_rate": 0.0005, "loss": 1.7166403532028198, "step": 16627 }, { "epoch": 1.3108395743003547, "grad_norm": 0.5405679941177368, "learning_rate": 0.0005, "loss": 1.6326208114624023, "step": 16628 }, { "epoch": 1.3109184075679936, "grad_norm": 0.5162632465362549, "learning_rate": 0.0005, "loss": 1.7063720226287842, "step": 16629 }, { "epoch": 1.3109972408356327, "grad_norm": 0.596144437789917, "learning_rate": 0.0005, "loss": 1.4245121479034424, "step": 16630 }, { "epoch": 1.3110760741032716, "grad_norm": 0.7228094935417175, "learning_rate": 0.0005, "loss": 1.573995590209961, "step": 16631 }, { "epoch": 1.3111549073709106, "grad_norm": 0.5435023903846741, "learning_rate": 0.0005, "loss": 1.482508897781372, "step": 16632 }, { "epoch": 1.3112337406385495, "grad_norm": 0.8037675619125366, "learning_rate": 0.0005, "loss": 1.7378551959991455, "step": 16633 }, { "epoch": 1.3113125739061884, "grad_norm": 0.5962668061256409, "learning_rate": 0.0005, "loss": 1.8482437133789062, "step": 16634 }, { "epoch": 1.3113914071738273, "grad_norm": 0.6777794361114502, "learning_rate": 0.0005, "loss": 1.525664210319519, "step": 16635 }, { "epoch": 1.3114702404414662, "grad_norm": 0.5971822738647461, "learning_rate": 0.0005, "loss": 1.698179841041565, "step": 16636 }, { "epoch": 1.3115490737091053, "grad_norm": 0.5022454261779785, "learning_rate": 0.0005, "loss": 1.5575604438781738, "step": 16637 }, { "epoch": 1.3116279069767443, "grad_norm": 0.5539846420288086, "learning_rate": 0.0005, "loss": 1.713181972503662, "step": 16638 }, { "epoch": 1.3117067402443832, "grad_norm": 0.5097240805625916, "learning_rate": 0.0005, "loss": 1.6696898937225342, "step": 16639 }, { "epoch": 1.311785573512022, "grad_norm": 0.5339061617851257, "learning_rate": 0.0005, "loss": 1.8644388914108276, "step": 16640 }, { "epoch": 1.311864406779661, "grad_norm": 0.6139029860496521, "learning_rate": 0.0005, "loss": 1.6941134929656982, "step": 16641 }, { "epoch": 1.3119432400473, "grad_norm": 0.5091960430145264, "learning_rate": 0.0005, "loss": 1.8465219736099243, "step": 16642 }, { "epoch": 1.3120220733149388, "grad_norm": 0.5333038568496704, "learning_rate": 0.0005, "loss": 1.665558934211731, "step": 16643 }, { "epoch": 1.312100906582578, "grad_norm": 0.6055859327316284, "learning_rate": 0.0005, "loss": 1.8511773347854614, "step": 16644 }, { "epoch": 1.3121797398502169, "grad_norm": 0.784895122051239, "learning_rate": 0.0005, "loss": 1.8212404251098633, "step": 16645 }, { "epoch": 1.3122585731178558, "grad_norm": 0.4485912024974823, "learning_rate": 0.0005, "loss": 1.7403913736343384, "step": 16646 }, { "epoch": 1.3123374063854947, "grad_norm": 0.6623035073280334, "learning_rate": 0.0005, "loss": 1.666365146636963, "step": 16647 }, { "epoch": 1.3124162396531336, "grad_norm": 1.2001259326934814, "learning_rate": 0.0005, "loss": 1.5860655307769775, "step": 16648 }, { "epoch": 1.3124950729207725, "grad_norm": 0.8683874011039734, "learning_rate": 0.0005, "loss": 1.6656209230422974, "step": 16649 }, { "epoch": 1.3125739061884114, "grad_norm": 0.5003255605697632, "learning_rate": 0.0005, "loss": 1.7110265493392944, "step": 16650 }, { "epoch": 1.3126527394560505, "grad_norm": 0.5385138392448425, "learning_rate": 0.0005, "loss": 1.4787790775299072, "step": 16651 }, { "epoch": 1.3127315727236895, "grad_norm": 0.7497560381889343, "learning_rate": 0.0005, "loss": 1.7945960760116577, "step": 16652 }, { "epoch": 1.3128104059913284, "grad_norm": 0.49923956394195557, "learning_rate": 0.0005, "loss": 1.497973084449768, "step": 16653 }, { "epoch": 1.3128892392589673, "grad_norm": 0.8232337832450867, "learning_rate": 0.0005, "loss": 1.8458209037780762, "step": 16654 }, { "epoch": 1.3129680725266062, "grad_norm": 0.6094875931739807, "learning_rate": 0.0005, "loss": 1.685855507850647, "step": 16655 }, { "epoch": 1.313046905794245, "grad_norm": 0.5826997756958008, "learning_rate": 0.0005, "loss": 1.8595292568206787, "step": 16656 }, { "epoch": 1.313125739061884, "grad_norm": 0.5489538311958313, "learning_rate": 0.0005, "loss": 1.594176173210144, "step": 16657 }, { "epoch": 1.3132045723295231, "grad_norm": 0.6022722125053406, "learning_rate": 0.0005, "loss": 1.5955724716186523, "step": 16658 }, { "epoch": 1.313283405597162, "grad_norm": 0.6553894281387329, "learning_rate": 0.0005, "loss": 1.8597298860549927, "step": 16659 }, { "epoch": 1.313362238864801, "grad_norm": 0.6644474267959595, "learning_rate": 0.0005, "loss": 1.9188807010650635, "step": 16660 }, { "epoch": 1.3134410721324399, "grad_norm": 0.6770210266113281, "learning_rate": 0.0005, "loss": 1.442176103591919, "step": 16661 }, { "epoch": 1.3135199054000788, "grad_norm": 0.5648154616355896, "learning_rate": 0.0005, "loss": 1.893348217010498, "step": 16662 }, { "epoch": 1.3135987386677177, "grad_norm": 0.5438832640647888, "learning_rate": 0.0005, "loss": 1.56477689743042, "step": 16663 }, { "epoch": 1.3136775719353566, "grad_norm": 0.6381976008415222, "learning_rate": 0.0005, "loss": 1.8810147047042847, "step": 16664 }, { "epoch": 1.3137564052029957, "grad_norm": 0.7280592918395996, "learning_rate": 0.0005, "loss": 1.515371561050415, "step": 16665 }, { "epoch": 1.3138352384706347, "grad_norm": 0.7047384977340698, "learning_rate": 0.0005, "loss": 1.5093464851379395, "step": 16666 }, { "epoch": 1.3139140717382736, "grad_norm": 0.6413060426712036, "learning_rate": 0.0005, "loss": 1.8692033290863037, "step": 16667 }, { "epoch": 1.3139929050059125, "grad_norm": 0.7540603876113892, "learning_rate": 0.0005, "loss": 1.4459385871887207, "step": 16668 }, { "epoch": 1.3140717382735514, "grad_norm": 0.5471228361129761, "learning_rate": 0.0005, "loss": 1.7770627737045288, "step": 16669 }, { "epoch": 1.3141505715411903, "grad_norm": 0.44628459215164185, "learning_rate": 0.0005, "loss": 1.523271083831787, "step": 16670 }, { "epoch": 1.3142294048088292, "grad_norm": 0.5928093791007996, "learning_rate": 0.0005, "loss": 1.597336769104004, "step": 16671 }, { "epoch": 1.3143082380764683, "grad_norm": 0.5215874910354614, "learning_rate": 0.0005, "loss": 1.61525297164917, "step": 16672 }, { "epoch": 1.3143870713441073, "grad_norm": 0.48446494340896606, "learning_rate": 0.0005, "loss": 1.4196012020111084, "step": 16673 }, { "epoch": 1.3144659046117462, "grad_norm": 0.5644163489341736, "learning_rate": 0.0005, "loss": 1.428243637084961, "step": 16674 }, { "epoch": 1.314544737879385, "grad_norm": 0.6448827981948853, "learning_rate": 0.0005, "loss": 1.6610112190246582, "step": 16675 }, { "epoch": 1.314623571147024, "grad_norm": 0.5049019455909729, "learning_rate": 0.0005, "loss": 1.5893402099609375, "step": 16676 }, { "epoch": 1.314702404414663, "grad_norm": 0.47859570384025574, "learning_rate": 0.0005, "loss": 1.7123732566833496, "step": 16677 }, { "epoch": 1.3147812376823018, "grad_norm": 0.5405650734901428, "learning_rate": 0.0005, "loss": 1.7705460786819458, "step": 16678 }, { "epoch": 1.314860070949941, "grad_norm": 0.568510890007019, "learning_rate": 0.0005, "loss": 1.9928337335586548, "step": 16679 }, { "epoch": 1.3149389042175799, "grad_norm": 0.4851703941822052, "learning_rate": 0.0005, "loss": 1.6606045961380005, "step": 16680 }, { "epoch": 1.3150177374852188, "grad_norm": 0.6187222599983215, "learning_rate": 0.0005, "loss": 1.5483779907226562, "step": 16681 }, { "epoch": 1.3150965707528577, "grad_norm": 0.6087377667427063, "learning_rate": 0.0005, "loss": 1.6221262216567993, "step": 16682 }, { "epoch": 1.3151754040204966, "grad_norm": 0.8063441514968872, "learning_rate": 0.0005, "loss": 1.810450553894043, "step": 16683 }, { "epoch": 1.3152542372881357, "grad_norm": 0.5697421431541443, "learning_rate": 0.0005, "loss": 1.7310439348220825, "step": 16684 }, { "epoch": 1.3153330705557744, "grad_norm": 0.6323398947715759, "learning_rate": 0.0005, "loss": 1.9074532985687256, "step": 16685 }, { "epoch": 1.3154119038234136, "grad_norm": 0.6482712626457214, "learning_rate": 0.0005, "loss": 1.6849172115325928, "step": 16686 }, { "epoch": 1.3154907370910525, "grad_norm": 0.6209237575531006, "learning_rate": 0.0005, "loss": 1.6298425197601318, "step": 16687 }, { "epoch": 1.3155695703586914, "grad_norm": 0.7118633389472961, "learning_rate": 0.0005, "loss": 1.9301910400390625, "step": 16688 }, { "epoch": 1.3156484036263303, "grad_norm": 0.4224746525287628, "learning_rate": 0.0005, "loss": 1.5231997966766357, "step": 16689 }, { "epoch": 1.3157272368939692, "grad_norm": 1.1981730461120605, "learning_rate": 0.0005, "loss": 1.593988299369812, "step": 16690 }, { "epoch": 1.3158060701616083, "grad_norm": 0.5773099064826965, "learning_rate": 0.0005, "loss": 1.6851361989974976, "step": 16691 }, { "epoch": 1.315884903429247, "grad_norm": 0.6785696744918823, "learning_rate": 0.0005, "loss": 1.720953345298767, "step": 16692 }, { "epoch": 1.3159637366968862, "grad_norm": 0.5639362335205078, "learning_rate": 0.0005, "loss": 1.7977805137634277, "step": 16693 }, { "epoch": 1.316042569964525, "grad_norm": 0.6256957054138184, "learning_rate": 0.0005, "loss": 1.6156866550445557, "step": 16694 }, { "epoch": 1.316121403232164, "grad_norm": 0.6094821095466614, "learning_rate": 0.0005, "loss": 1.6458173990249634, "step": 16695 }, { "epoch": 1.316200236499803, "grad_norm": 0.5613052248954773, "learning_rate": 0.0005, "loss": 1.4888229370117188, "step": 16696 }, { "epoch": 1.3162790697674418, "grad_norm": 0.4766487777233124, "learning_rate": 0.0005, "loss": 1.6421244144439697, "step": 16697 }, { "epoch": 1.316357903035081, "grad_norm": 0.49651604890823364, "learning_rate": 0.0005, "loss": 1.7399530410766602, "step": 16698 }, { "epoch": 1.3164367363027196, "grad_norm": 0.6917082667350769, "learning_rate": 0.0005, "loss": 1.713232398033142, "step": 16699 }, { "epoch": 1.3165155695703588, "grad_norm": 0.44610080122947693, "learning_rate": 0.0005, "loss": 1.7527947425842285, "step": 16700 }, { "epoch": 1.3165944028379977, "grad_norm": 0.42604124546051025, "learning_rate": 0.0005, "loss": 1.4369032382965088, "step": 16701 }, { "epoch": 1.3166732361056366, "grad_norm": 0.7160279750823975, "learning_rate": 0.0005, "loss": 1.5831408500671387, "step": 16702 }, { "epoch": 1.3167520693732755, "grad_norm": 0.5896918773651123, "learning_rate": 0.0005, "loss": 1.7455857992172241, "step": 16703 }, { "epoch": 1.3168309026409144, "grad_norm": 0.6204319000244141, "learning_rate": 0.0005, "loss": 1.7091777324676514, "step": 16704 }, { "epoch": 1.3169097359085535, "grad_norm": 0.605360209941864, "learning_rate": 0.0005, "loss": 1.7820734977722168, "step": 16705 }, { "epoch": 1.3169885691761922, "grad_norm": 0.6566907167434692, "learning_rate": 0.0005, "loss": 1.702817678451538, "step": 16706 }, { "epoch": 1.3170674024438314, "grad_norm": 0.535261332988739, "learning_rate": 0.0005, "loss": 1.6797630786895752, "step": 16707 }, { "epoch": 1.3171462357114703, "grad_norm": 0.5517454147338867, "learning_rate": 0.0005, "loss": 1.5980010032653809, "step": 16708 }, { "epoch": 1.3172250689791092, "grad_norm": 0.5481560826301575, "learning_rate": 0.0005, "loss": 1.7068183422088623, "step": 16709 }, { "epoch": 1.317303902246748, "grad_norm": 0.6457581520080566, "learning_rate": 0.0005, "loss": 2.026808738708496, "step": 16710 }, { "epoch": 1.317382735514387, "grad_norm": 2.263676881790161, "learning_rate": 0.0005, "loss": 1.663221836090088, "step": 16711 }, { "epoch": 1.3174615687820261, "grad_norm": 0.5083995461463928, "learning_rate": 0.0005, "loss": 1.4998042583465576, "step": 16712 }, { "epoch": 1.3175404020496648, "grad_norm": 0.438507616519928, "learning_rate": 0.0005, "loss": 1.3776003122329712, "step": 16713 }, { "epoch": 1.317619235317304, "grad_norm": 0.6272575855255127, "learning_rate": 0.0005, "loss": 1.700657844543457, "step": 16714 }, { "epoch": 1.3176980685849429, "grad_norm": 0.4721764326095581, "learning_rate": 0.0005, "loss": 1.5637822151184082, "step": 16715 }, { "epoch": 1.3177769018525818, "grad_norm": 0.41582417488098145, "learning_rate": 0.0005, "loss": 1.4861624240875244, "step": 16716 }, { "epoch": 1.3178557351202207, "grad_norm": 0.48871174454689026, "learning_rate": 0.0005, "loss": 1.5543246269226074, "step": 16717 }, { "epoch": 1.3179345683878596, "grad_norm": 0.6734890341758728, "learning_rate": 0.0005, "loss": 1.7324113845825195, "step": 16718 }, { "epoch": 1.3180134016554987, "grad_norm": 0.6545780301094055, "learning_rate": 0.0005, "loss": 1.5383408069610596, "step": 16719 }, { "epoch": 1.3180922349231374, "grad_norm": 0.5415906310081482, "learning_rate": 0.0005, "loss": 1.71968674659729, "step": 16720 }, { "epoch": 1.3181710681907766, "grad_norm": 0.6009842753410339, "learning_rate": 0.0005, "loss": 1.7696988582611084, "step": 16721 }, { "epoch": 1.3182499014584155, "grad_norm": 0.7641175985336304, "learning_rate": 0.0005, "loss": 1.8380297422409058, "step": 16722 }, { "epoch": 1.3183287347260544, "grad_norm": 0.8183141350746155, "learning_rate": 0.0005, "loss": 1.7317540645599365, "step": 16723 }, { "epoch": 1.3184075679936933, "grad_norm": 0.5280996561050415, "learning_rate": 0.0005, "loss": 1.6290662288665771, "step": 16724 }, { "epoch": 1.3184864012613322, "grad_norm": 0.5050414204597473, "learning_rate": 0.0005, "loss": 1.4215255975723267, "step": 16725 }, { "epoch": 1.3185652345289713, "grad_norm": 0.49139752984046936, "learning_rate": 0.0005, "loss": 1.6462184190750122, "step": 16726 }, { "epoch": 1.31864406779661, "grad_norm": 0.6301082968711853, "learning_rate": 0.0005, "loss": 1.4619309902191162, "step": 16727 }, { "epoch": 1.3187229010642492, "grad_norm": 0.6160288453102112, "learning_rate": 0.0005, "loss": 1.7397583723068237, "step": 16728 }, { "epoch": 1.318801734331888, "grad_norm": 0.48595353960990906, "learning_rate": 0.0005, "loss": 1.7331385612487793, "step": 16729 }, { "epoch": 1.318880567599527, "grad_norm": 0.5839007496833801, "learning_rate": 0.0005, "loss": 1.5871925354003906, "step": 16730 }, { "epoch": 1.318959400867166, "grad_norm": 0.48852694034576416, "learning_rate": 0.0005, "loss": 1.4643925428390503, "step": 16731 }, { "epoch": 1.3190382341348048, "grad_norm": 0.48715147376060486, "learning_rate": 0.0005, "loss": 1.6484471559524536, "step": 16732 }, { "epoch": 1.319117067402444, "grad_norm": 0.7714029550552368, "learning_rate": 0.0005, "loss": 1.8601442575454712, "step": 16733 }, { "epoch": 1.3191959006700829, "grad_norm": 0.5682331323623657, "learning_rate": 0.0005, "loss": 1.5414267778396606, "step": 16734 }, { "epoch": 1.3192747339377218, "grad_norm": 0.4956636130809784, "learning_rate": 0.0005, "loss": 1.665165901184082, "step": 16735 }, { "epoch": 1.3193535672053607, "grad_norm": 0.6176039576530457, "learning_rate": 0.0005, "loss": 1.9476044178009033, "step": 16736 }, { "epoch": 1.3194324004729996, "grad_norm": 0.6972343921661377, "learning_rate": 0.0005, "loss": 1.835931420326233, "step": 16737 }, { "epoch": 1.3195112337406385, "grad_norm": 0.5394236445426941, "learning_rate": 0.0005, "loss": 1.829131841659546, "step": 16738 }, { "epoch": 1.3195900670082774, "grad_norm": 0.5993794798851013, "learning_rate": 0.0005, "loss": 1.6413655281066895, "step": 16739 }, { "epoch": 1.3196689002759165, "grad_norm": 0.5259270071983337, "learning_rate": 0.0005, "loss": 1.6629034280776978, "step": 16740 }, { "epoch": 1.3197477335435555, "grad_norm": 0.7676050662994385, "learning_rate": 0.0005, "loss": 1.5294246673583984, "step": 16741 }, { "epoch": 1.3198265668111944, "grad_norm": 0.5002383589744568, "learning_rate": 0.0005, "loss": 1.5468320846557617, "step": 16742 }, { "epoch": 1.3199054000788333, "grad_norm": 0.7437489032745361, "learning_rate": 0.0005, "loss": 1.6549335718154907, "step": 16743 }, { "epoch": 1.3199842333464722, "grad_norm": 0.6835600733757019, "learning_rate": 0.0005, "loss": 1.9594498872756958, "step": 16744 }, { "epoch": 1.320063066614111, "grad_norm": 0.4448634684085846, "learning_rate": 0.0005, "loss": 1.4879028797149658, "step": 16745 }, { "epoch": 1.32014189988175, "grad_norm": 0.6107032299041748, "learning_rate": 0.0005, "loss": 1.4723846912384033, "step": 16746 }, { "epoch": 1.3202207331493891, "grad_norm": 0.6014159917831421, "learning_rate": 0.0005, "loss": 1.622450351715088, "step": 16747 }, { "epoch": 1.320299566417028, "grad_norm": 0.40803271532058716, "learning_rate": 0.0005, "loss": 1.2787905931472778, "step": 16748 }, { "epoch": 1.320378399684667, "grad_norm": 0.5224656462669373, "learning_rate": 0.0005, "loss": 1.577352523803711, "step": 16749 }, { "epoch": 1.3204572329523059, "grad_norm": 0.5063985586166382, "learning_rate": 0.0005, "loss": 1.7112131118774414, "step": 16750 }, { "epoch": 1.3205360662199448, "grad_norm": 0.5190714597702026, "learning_rate": 0.0005, "loss": 1.9019834995269775, "step": 16751 }, { "epoch": 1.3206148994875837, "grad_norm": 0.4882906973361969, "learning_rate": 0.0005, "loss": 1.7260639667510986, "step": 16752 }, { "epoch": 1.3206937327552226, "grad_norm": 0.8163085579872131, "learning_rate": 0.0005, "loss": 1.7467749118804932, "step": 16753 }, { "epoch": 1.3207725660228617, "grad_norm": 0.5373263955116272, "learning_rate": 0.0005, "loss": 1.7654995918273926, "step": 16754 }, { "epoch": 1.3208513992905007, "grad_norm": 0.47130507230758667, "learning_rate": 0.0005, "loss": 1.5939147472381592, "step": 16755 }, { "epoch": 1.3209302325581396, "grad_norm": 0.5334415435791016, "learning_rate": 0.0005, "loss": 1.7675673961639404, "step": 16756 }, { "epoch": 1.3210090658257785, "grad_norm": 0.824087381362915, "learning_rate": 0.0005, "loss": 1.9141281843185425, "step": 16757 }, { "epoch": 1.3210878990934174, "grad_norm": 0.5354638695716858, "learning_rate": 0.0005, "loss": 1.7266192436218262, "step": 16758 }, { "epoch": 1.3211667323610563, "grad_norm": 0.5252808332443237, "learning_rate": 0.0005, "loss": 1.7441802024841309, "step": 16759 }, { "epoch": 1.3212455656286952, "grad_norm": 0.6215066909790039, "learning_rate": 0.0005, "loss": 1.67069673538208, "step": 16760 }, { "epoch": 1.3213243988963344, "grad_norm": 0.5768830180168152, "learning_rate": 0.0005, "loss": 1.646294116973877, "step": 16761 }, { "epoch": 1.3214032321639733, "grad_norm": 0.638879120349884, "learning_rate": 0.0005, "loss": 1.5992462635040283, "step": 16762 }, { "epoch": 1.3214820654316122, "grad_norm": 0.5547172427177429, "learning_rate": 0.0005, "loss": 1.6381011009216309, "step": 16763 }, { "epoch": 1.321560898699251, "grad_norm": 0.5697636008262634, "learning_rate": 0.0005, "loss": 1.5451924800872803, "step": 16764 }, { "epoch": 1.32163973196689, "grad_norm": 0.5715596675872803, "learning_rate": 0.0005, "loss": 1.5372912883758545, "step": 16765 }, { "epoch": 1.321718565234529, "grad_norm": 0.6050769090652466, "learning_rate": 0.0005, "loss": 1.5502830743789673, "step": 16766 }, { "epoch": 1.3217973985021678, "grad_norm": 0.5944322347640991, "learning_rate": 0.0005, "loss": 1.7351295948028564, "step": 16767 }, { "epoch": 1.321876231769807, "grad_norm": 1.9913785457611084, "learning_rate": 0.0005, "loss": 1.5368303060531616, "step": 16768 }, { "epoch": 1.3219550650374459, "grad_norm": 0.5778361558914185, "learning_rate": 0.0005, "loss": 1.6888482570648193, "step": 16769 }, { "epoch": 1.3220338983050848, "grad_norm": 0.5044400095939636, "learning_rate": 0.0005, "loss": 1.4147791862487793, "step": 16770 }, { "epoch": 1.3221127315727237, "grad_norm": 0.5929719805717468, "learning_rate": 0.0005, "loss": 1.4928522109985352, "step": 16771 }, { "epoch": 1.3221915648403626, "grad_norm": 0.5931583046913147, "learning_rate": 0.0005, "loss": 1.7983345985412598, "step": 16772 }, { "epoch": 1.3222703981080015, "grad_norm": 0.5762350559234619, "learning_rate": 0.0005, "loss": 1.664167881011963, "step": 16773 }, { "epoch": 1.3223492313756404, "grad_norm": 0.5220169425010681, "learning_rate": 0.0005, "loss": 1.5256061553955078, "step": 16774 }, { "epoch": 1.3224280646432796, "grad_norm": 0.4415367841720581, "learning_rate": 0.0005, "loss": 1.5875673294067383, "step": 16775 }, { "epoch": 1.3225068979109185, "grad_norm": 0.5652298331260681, "learning_rate": 0.0005, "loss": 1.4356626272201538, "step": 16776 }, { "epoch": 1.3225857311785574, "grad_norm": 0.5429994463920593, "learning_rate": 0.0005, "loss": 1.6671262979507446, "step": 16777 }, { "epoch": 1.3226645644461963, "grad_norm": 0.5803901553153992, "learning_rate": 0.0005, "loss": 1.7634673118591309, "step": 16778 }, { "epoch": 1.3227433977138352, "grad_norm": 0.5339213013648987, "learning_rate": 0.0005, "loss": 1.6070383787155151, "step": 16779 }, { "epoch": 1.3228222309814741, "grad_norm": 0.46246013045310974, "learning_rate": 0.0005, "loss": 1.706247329711914, "step": 16780 }, { "epoch": 1.322901064249113, "grad_norm": 0.6437537670135498, "learning_rate": 0.0005, "loss": 1.686279296875, "step": 16781 }, { "epoch": 1.3229798975167522, "grad_norm": 0.513616681098938, "learning_rate": 0.0005, "loss": 1.760676622390747, "step": 16782 }, { "epoch": 1.323058730784391, "grad_norm": 0.574687123298645, "learning_rate": 0.0005, "loss": 1.6793253421783447, "step": 16783 }, { "epoch": 1.32313756405203, "grad_norm": 0.47685450315475464, "learning_rate": 0.0005, "loss": 1.7890105247497559, "step": 16784 }, { "epoch": 1.323216397319669, "grad_norm": 0.547264814376831, "learning_rate": 0.0005, "loss": 1.402160882949829, "step": 16785 }, { "epoch": 1.3232952305873078, "grad_norm": 0.5340713858604431, "learning_rate": 0.0005, "loss": 1.7000219821929932, "step": 16786 }, { "epoch": 1.3233740638549467, "grad_norm": 0.6307125091552734, "learning_rate": 0.0005, "loss": 1.9551217555999756, "step": 16787 }, { "epoch": 1.3234528971225856, "grad_norm": 0.57465660572052, "learning_rate": 0.0005, "loss": 1.4423367977142334, "step": 16788 }, { "epoch": 1.3235317303902248, "grad_norm": 0.5546050071716309, "learning_rate": 0.0005, "loss": 1.7557697296142578, "step": 16789 }, { "epoch": 1.3236105636578637, "grad_norm": 0.5546057820320129, "learning_rate": 0.0005, "loss": 1.652890920639038, "step": 16790 }, { "epoch": 1.3236893969255026, "grad_norm": 0.6898985505104065, "learning_rate": 0.0005, "loss": 1.805091142654419, "step": 16791 }, { "epoch": 1.3237682301931415, "grad_norm": 0.47238999605178833, "learning_rate": 0.0005, "loss": 1.5456537008285522, "step": 16792 }, { "epoch": 1.3238470634607804, "grad_norm": 0.5018516778945923, "learning_rate": 0.0005, "loss": 1.7697452306747437, "step": 16793 }, { "epoch": 1.3239258967284193, "grad_norm": 0.44843366742134094, "learning_rate": 0.0005, "loss": 1.4963762760162354, "step": 16794 }, { "epoch": 1.3240047299960582, "grad_norm": 0.8322166800498962, "learning_rate": 0.0005, "loss": 1.708306074142456, "step": 16795 }, { "epoch": 1.3240835632636974, "grad_norm": 0.5363776087760925, "learning_rate": 0.0005, "loss": 1.5219454765319824, "step": 16796 }, { "epoch": 1.3241623965313363, "grad_norm": 0.5869677662849426, "learning_rate": 0.0005, "loss": 1.8256809711456299, "step": 16797 }, { "epoch": 1.3242412297989752, "grad_norm": 0.47025567293167114, "learning_rate": 0.0005, "loss": 1.760319471359253, "step": 16798 }, { "epoch": 1.324320063066614, "grad_norm": 0.608018159866333, "learning_rate": 0.0005, "loss": 1.7855017185211182, "step": 16799 }, { "epoch": 1.324398896334253, "grad_norm": 0.47039079666137695, "learning_rate": 0.0005, "loss": 1.5226850509643555, "step": 16800 }, { "epoch": 1.324477729601892, "grad_norm": 0.475553423166275, "learning_rate": 0.0005, "loss": 1.6439012289047241, "step": 16801 }, { "epoch": 1.3245565628695308, "grad_norm": 0.5089041590690613, "learning_rate": 0.0005, "loss": 1.5541305541992188, "step": 16802 }, { "epoch": 1.32463539613717, "grad_norm": 0.4738022983074188, "learning_rate": 0.0005, "loss": 1.4300649166107178, "step": 16803 }, { "epoch": 1.3247142294048089, "grad_norm": 0.5083917379379272, "learning_rate": 0.0005, "loss": 1.7161636352539062, "step": 16804 }, { "epoch": 1.3247930626724478, "grad_norm": 0.580518901348114, "learning_rate": 0.0005, "loss": 1.7492939233779907, "step": 16805 }, { "epoch": 1.3248718959400867, "grad_norm": 0.5487970113754272, "learning_rate": 0.0005, "loss": 1.7617934942245483, "step": 16806 }, { "epoch": 1.3249507292077256, "grad_norm": 0.5184162259101868, "learning_rate": 0.0005, "loss": 1.646524429321289, "step": 16807 }, { "epoch": 1.3250295624753647, "grad_norm": 0.5183872580528259, "learning_rate": 0.0005, "loss": 2.005709409713745, "step": 16808 }, { "epoch": 1.3251083957430034, "grad_norm": 0.5230003595352173, "learning_rate": 0.0005, "loss": 1.4420892000198364, "step": 16809 }, { "epoch": 1.3251872290106426, "grad_norm": 0.6097807288169861, "learning_rate": 0.0005, "loss": 1.6181716918945312, "step": 16810 }, { "epoch": 1.3252660622782815, "grad_norm": 0.499877393245697, "learning_rate": 0.0005, "loss": 1.627171277999878, "step": 16811 }, { "epoch": 1.3253448955459204, "grad_norm": 0.6127349734306335, "learning_rate": 0.0005, "loss": 1.6559593677520752, "step": 16812 }, { "epoch": 1.3254237288135593, "grad_norm": 0.5413045287132263, "learning_rate": 0.0005, "loss": 1.8717494010925293, "step": 16813 }, { "epoch": 1.3255025620811982, "grad_norm": 0.6018269062042236, "learning_rate": 0.0005, "loss": 1.8331788778305054, "step": 16814 }, { "epoch": 1.3255813953488373, "grad_norm": 0.5345142483711243, "learning_rate": 0.0005, "loss": 1.9462515115737915, "step": 16815 }, { "epoch": 1.325660228616476, "grad_norm": 0.41976282000541687, "learning_rate": 0.0005, "loss": 1.4939872026443481, "step": 16816 }, { "epoch": 1.3257390618841152, "grad_norm": 0.43187204003334045, "learning_rate": 0.0005, "loss": 1.8509223461151123, "step": 16817 }, { "epoch": 1.325817895151754, "grad_norm": 0.5569939017295837, "learning_rate": 0.0005, "loss": 1.6781096458435059, "step": 16818 }, { "epoch": 1.325896728419393, "grad_norm": 0.5042058825492859, "learning_rate": 0.0005, "loss": 1.6180026531219482, "step": 16819 }, { "epoch": 1.325975561687032, "grad_norm": 0.5056485533714294, "learning_rate": 0.0005, "loss": 1.5297470092773438, "step": 16820 }, { "epoch": 1.3260543949546708, "grad_norm": 0.5365154147148132, "learning_rate": 0.0005, "loss": 1.546904444694519, "step": 16821 }, { "epoch": 1.32613322822231, "grad_norm": 0.6119964122772217, "learning_rate": 0.0005, "loss": 1.6894288063049316, "step": 16822 }, { "epoch": 1.3262120614899486, "grad_norm": 0.5953901410102844, "learning_rate": 0.0005, "loss": 1.651496171951294, "step": 16823 }, { "epoch": 1.3262908947575878, "grad_norm": 0.4972142279148102, "learning_rate": 0.0005, "loss": 1.6957197189331055, "step": 16824 }, { "epoch": 1.3263697280252267, "grad_norm": 0.5229928493499756, "learning_rate": 0.0005, "loss": 1.4943721294403076, "step": 16825 }, { "epoch": 1.3264485612928656, "grad_norm": 0.5777267217636108, "learning_rate": 0.0005, "loss": 1.7348546981811523, "step": 16826 }, { "epoch": 1.3265273945605045, "grad_norm": 0.5109955072402954, "learning_rate": 0.0005, "loss": 1.5697846412658691, "step": 16827 }, { "epoch": 1.3266062278281434, "grad_norm": 0.5814977288246155, "learning_rate": 0.0005, "loss": 1.5600001811981201, "step": 16828 }, { "epoch": 1.3266850610957825, "grad_norm": 0.5664435029029846, "learning_rate": 0.0005, "loss": 1.672690987586975, "step": 16829 }, { "epoch": 1.3267638943634212, "grad_norm": 0.4300754964351654, "learning_rate": 0.0005, "loss": 1.7127751111984253, "step": 16830 }, { "epoch": 1.3268427276310604, "grad_norm": 0.5608718395233154, "learning_rate": 0.0005, "loss": 1.6206718683242798, "step": 16831 }, { "epoch": 1.3269215608986993, "grad_norm": 0.5098638534545898, "learning_rate": 0.0005, "loss": 1.561793327331543, "step": 16832 }, { "epoch": 1.3270003941663382, "grad_norm": 0.5409768223762512, "learning_rate": 0.0005, "loss": 1.6201884746551514, "step": 16833 }, { "epoch": 1.327079227433977, "grad_norm": 0.6156073808670044, "learning_rate": 0.0005, "loss": 1.731727957725525, "step": 16834 }, { "epoch": 1.327158060701616, "grad_norm": 0.5582153797149658, "learning_rate": 0.0005, "loss": 1.6611876487731934, "step": 16835 }, { "epoch": 1.3272368939692551, "grad_norm": 0.5825337171554565, "learning_rate": 0.0005, "loss": 1.548032522201538, "step": 16836 }, { "epoch": 1.3273157272368938, "grad_norm": 0.8570287227630615, "learning_rate": 0.0005, "loss": 1.423209309577942, "step": 16837 }, { "epoch": 1.327394560504533, "grad_norm": 0.46538329124450684, "learning_rate": 0.0005, "loss": 1.5731427669525146, "step": 16838 }, { "epoch": 1.3274733937721719, "grad_norm": 0.4661828279495239, "learning_rate": 0.0005, "loss": 1.4281110763549805, "step": 16839 }, { "epoch": 1.3275522270398108, "grad_norm": 0.51832515001297, "learning_rate": 0.0005, "loss": 1.7223907709121704, "step": 16840 }, { "epoch": 1.3276310603074497, "grad_norm": 0.5193499326705933, "learning_rate": 0.0005, "loss": 1.5641207695007324, "step": 16841 }, { "epoch": 1.3277098935750886, "grad_norm": 0.5889339447021484, "learning_rate": 0.0005, "loss": 1.6596217155456543, "step": 16842 }, { "epoch": 1.3277887268427278, "grad_norm": 0.46248361468315125, "learning_rate": 0.0005, "loss": 1.625476598739624, "step": 16843 }, { "epoch": 1.3278675601103664, "grad_norm": 0.6134294867515564, "learning_rate": 0.0005, "loss": 1.8218353986740112, "step": 16844 }, { "epoch": 1.3279463933780056, "grad_norm": 0.5303857326507568, "learning_rate": 0.0005, "loss": 1.9261305332183838, "step": 16845 }, { "epoch": 1.3280252266456445, "grad_norm": 0.5838702321052551, "learning_rate": 0.0005, "loss": 1.6577413082122803, "step": 16846 }, { "epoch": 1.3281040599132834, "grad_norm": 0.45579686760902405, "learning_rate": 0.0005, "loss": 1.4036556482315063, "step": 16847 }, { "epoch": 1.3281828931809223, "grad_norm": 0.893968939781189, "learning_rate": 0.0005, "loss": 1.7247114181518555, "step": 16848 }, { "epoch": 1.3282617264485612, "grad_norm": 0.5089982151985168, "learning_rate": 0.0005, "loss": 1.3704856634140015, "step": 16849 }, { "epoch": 1.3283405597162004, "grad_norm": 0.5840005874633789, "learning_rate": 0.0005, "loss": 1.6475273370742798, "step": 16850 }, { "epoch": 1.328419392983839, "grad_norm": 0.601606547832489, "learning_rate": 0.0005, "loss": 1.6376583576202393, "step": 16851 }, { "epoch": 1.3284982262514782, "grad_norm": 0.569098174571991, "learning_rate": 0.0005, "loss": 1.7164902687072754, "step": 16852 }, { "epoch": 1.328577059519117, "grad_norm": 0.6106184720993042, "learning_rate": 0.0005, "loss": 1.6485390663146973, "step": 16853 }, { "epoch": 1.328655892786756, "grad_norm": 0.509954035282135, "learning_rate": 0.0005, "loss": 1.6659119129180908, "step": 16854 }, { "epoch": 1.328734726054395, "grad_norm": 0.5327334403991699, "learning_rate": 0.0005, "loss": 1.6192271709442139, "step": 16855 }, { "epoch": 1.3288135593220338, "grad_norm": 0.4962873160839081, "learning_rate": 0.0005, "loss": 1.6016026735305786, "step": 16856 }, { "epoch": 1.328892392589673, "grad_norm": 0.5432236194610596, "learning_rate": 0.0005, "loss": 1.559098482131958, "step": 16857 }, { "epoch": 1.3289712258573119, "grad_norm": 0.6010438203811646, "learning_rate": 0.0005, "loss": 1.7514228820800781, "step": 16858 }, { "epoch": 1.3290500591249508, "grad_norm": 0.5309979915618896, "learning_rate": 0.0005, "loss": 1.3503390550613403, "step": 16859 }, { "epoch": 1.3291288923925897, "grad_norm": 0.6446905732154846, "learning_rate": 0.0005, "loss": 1.8189964294433594, "step": 16860 }, { "epoch": 1.3292077256602286, "grad_norm": 0.7670047283172607, "learning_rate": 0.0005, "loss": 1.829784870147705, "step": 16861 }, { "epoch": 1.3292865589278675, "grad_norm": 0.5117147564888, "learning_rate": 0.0005, "loss": 1.5890719890594482, "step": 16862 }, { "epoch": 1.3293653921955064, "grad_norm": 0.48775333166122437, "learning_rate": 0.0005, "loss": 1.5159354209899902, "step": 16863 }, { "epoch": 1.3294442254631456, "grad_norm": 0.41660887002944946, "learning_rate": 0.0005, "loss": 1.4293099641799927, "step": 16864 }, { "epoch": 1.3295230587307845, "grad_norm": 0.5522422790527344, "learning_rate": 0.0005, "loss": 1.5009883642196655, "step": 16865 }, { "epoch": 1.3296018919984234, "grad_norm": 0.50323486328125, "learning_rate": 0.0005, "loss": 1.5335023403167725, "step": 16866 }, { "epoch": 1.3296807252660623, "grad_norm": 0.5073800683021545, "learning_rate": 0.0005, "loss": 1.5515562295913696, "step": 16867 }, { "epoch": 1.3297595585337012, "grad_norm": 0.5130758881568909, "learning_rate": 0.0005, "loss": 1.7713326215744019, "step": 16868 }, { "epoch": 1.3298383918013401, "grad_norm": 0.674985945224762, "learning_rate": 0.0005, "loss": 1.9373245239257812, "step": 16869 }, { "epoch": 1.329917225068979, "grad_norm": 0.6094023585319519, "learning_rate": 0.0005, "loss": 1.617706537246704, "step": 16870 }, { "epoch": 1.3299960583366182, "grad_norm": 0.5139305591583252, "learning_rate": 0.0005, "loss": 1.8048208951950073, "step": 16871 }, { "epoch": 1.330074891604257, "grad_norm": 0.6558878421783447, "learning_rate": 0.0005, "loss": 1.9837301969528198, "step": 16872 }, { "epoch": 1.330153724871896, "grad_norm": 0.7642211318016052, "learning_rate": 0.0005, "loss": 1.7004175186157227, "step": 16873 }, { "epoch": 1.330232558139535, "grad_norm": 0.5689683556556702, "learning_rate": 0.0005, "loss": 1.6451294422149658, "step": 16874 }, { "epoch": 1.3303113914071738, "grad_norm": 0.7426556944847107, "learning_rate": 0.0005, "loss": 1.8283424377441406, "step": 16875 }, { "epoch": 1.3303902246748127, "grad_norm": 0.5537647008895874, "learning_rate": 0.0005, "loss": 1.60361647605896, "step": 16876 }, { "epoch": 1.3304690579424516, "grad_norm": 0.5976791381835938, "learning_rate": 0.0005, "loss": 1.6264349222183228, "step": 16877 }, { "epoch": 1.3305478912100908, "grad_norm": 0.4409138560295105, "learning_rate": 0.0005, "loss": 1.547218680381775, "step": 16878 }, { "epoch": 1.3306267244777297, "grad_norm": 0.8629791140556335, "learning_rate": 0.0005, "loss": 1.8128836154937744, "step": 16879 }, { "epoch": 1.3307055577453686, "grad_norm": 0.5680426955223083, "learning_rate": 0.0005, "loss": 1.7132320404052734, "step": 16880 }, { "epoch": 1.3307843910130075, "grad_norm": 0.8244442939758301, "learning_rate": 0.0005, "loss": 2.0156283378601074, "step": 16881 }, { "epoch": 1.3308632242806464, "grad_norm": 0.47517770528793335, "learning_rate": 0.0005, "loss": 1.6255171298980713, "step": 16882 }, { "epoch": 1.3309420575482853, "grad_norm": 0.46853670477867126, "learning_rate": 0.0005, "loss": 1.7952383756637573, "step": 16883 }, { "epoch": 1.3310208908159242, "grad_norm": 0.4693332314491272, "learning_rate": 0.0005, "loss": 1.5713026523590088, "step": 16884 }, { "epoch": 1.3310997240835634, "grad_norm": 0.5575738549232483, "learning_rate": 0.0005, "loss": 1.5700933933258057, "step": 16885 }, { "epoch": 1.3311785573512023, "grad_norm": 0.6556845903396606, "learning_rate": 0.0005, "loss": 1.5718047618865967, "step": 16886 }, { "epoch": 1.3312573906188412, "grad_norm": 0.5132062435150146, "learning_rate": 0.0005, "loss": 1.7150286436080933, "step": 16887 }, { "epoch": 1.33133622388648, "grad_norm": 0.5197620987892151, "learning_rate": 0.0005, "loss": 1.639517068862915, "step": 16888 }, { "epoch": 1.331415057154119, "grad_norm": 0.5464296936988831, "learning_rate": 0.0005, "loss": 1.687882661819458, "step": 16889 }, { "epoch": 1.331493890421758, "grad_norm": 0.5540121793746948, "learning_rate": 0.0005, "loss": 1.70509934425354, "step": 16890 }, { "epoch": 1.3315727236893968, "grad_norm": 0.7697968482971191, "learning_rate": 0.0005, "loss": 1.8186192512512207, "step": 16891 }, { "epoch": 1.331651556957036, "grad_norm": 0.44400161504745483, "learning_rate": 0.0005, "loss": 1.38496732711792, "step": 16892 }, { "epoch": 1.3317303902246749, "grad_norm": 0.5192385315895081, "learning_rate": 0.0005, "loss": 1.651491641998291, "step": 16893 }, { "epoch": 1.3318092234923138, "grad_norm": 0.5936610102653503, "learning_rate": 0.0005, "loss": 1.7065863609313965, "step": 16894 }, { "epoch": 1.3318880567599527, "grad_norm": 0.4194827377796173, "learning_rate": 0.0005, "loss": 1.419227123260498, "step": 16895 }, { "epoch": 1.3319668900275916, "grad_norm": 0.5695642828941345, "learning_rate": 0.0005, "loss": 1.6324567794799805, "step": 16896 }, { "epoch": 1.3320457232952305, "grad_norm": 0.503535807132721, "learning_rate": 0.0005, "loss": 1.5814604759216309, "step": 16897 }, { "epoch": 1.3321245565628694, "grad_norm": 0.9928335547447205, "learning_rate": 0.0005, "loss": 1.9580655097961426, "step": 16898 }, { "epoch": 1.3322033898305086, "grad_norm": 0.6006572246551514, "learning_rate": 0.0005, "loss": 1.6789021492004395, "step": 16899 }, { "epoch": 1.3322822230981475, "grad_norm": 0.6571973562240601, "learning_rate": 0.0005, "loss": 1.6787354946136475, "step": 16900 }, { "epoch": 1.3323610563657864, "grad_norm": 0.6871304512023926, "learning_rate": 0.0005, "loss": 1.4417222738265991, "step": 16901 }, { "epoch": 1.3324398896334253, "grad_norm": 0.5151475667953491, "learning_rate": 0.0005, "loss": 1.6084492206573486, "step": 16902 }, { "epoch": 1.3325187229010642, "grad_norm": 0.6876646280288696, "learning_rate": 0.0005, "loss": 1.7806549072265625, "step": 16903 }, { "epoch": 1.3325975561687031, "grad_norm": 0.7290560603141785, "learning_rate": 0.0005, "loss": 1.609520435333252, "step": 16904 }, { "epoch": 1.332676389436342, "grad_norm": 0.5447189807891846, "learning_rate": 0.0005, "loss": 1.5435338020324707, "step": 16905 }, { "epoch": 1.3327552227039812, "grad_norm": 1.025856375694275, "learning_rate": 0.0005, "loss": 1.9553332328796387, "step": 16906 }, { "epoch": 1.33283405597162, "grad_norm": 0.5619547367095947, "learning_rate": 0.0005, "loss": 1.6681246757507324, "step": 16907 }, { "epoch": 1.332912889239259, "grad_norm": 1.292044997215271, "learning_rate": 0.0005, "loss": 1.7090983390808105, "step": 16908 }, { "epoch": 1.332991722506898, "grad_norm": 0.5507563352584839, "learning_rate": 0.0005, "loss": 1.743485689163208, "step": 16909 }, { "epoch": 1.3330705557745368, "grad_norm": 0.548495352268219, "learning_rate": 0.0005, "loss": 1.525486946105957, "step": 16910 }, { "epoch": 1.3331493890421757, "grad_norm": 0.5247253179550171, "learning_rate": 0.0005, "loss": 1.7512247562408447, "step": 16911 }, { "epoch": 1.3332282223098146, "grad_norm": 0.5171428322792053, "learning_rate": 0.0005, "loss": 1.5358014106750488, "step": 16912 }, { "epoch": 1.3333070555774538, "grad_norm": 0.5454784035682678, "learning_rate": 0.0005, "loss": 1.4416462182998657, "step": 16913 }, { "epoch": 1.3333858888450927, "grad_norm": 0.5537891983985901, "learning_rate": 0.0005, "loss": 1.7064478397369385, "step": 16914 }, { "epoch": 1.3334647221127316, "grad_norm": 0.6269331574440002, "learning_rate": 0.0005, "loss": 1.7312119007110596, "step": 16915 }, { "epoch": 1.3335435553803705, "grad_norm": 0.5874144434928894, "learning_rate": 0.0005, "loss": 1.8993470668792725, "step": 16916 }, { "epoch": 1.3336223886480094, "grad_norm": 0.5490393042564392, "learning_rate": 0.0005, "loss": 1.5315639972686768, "step": 16917 }, { "epoch": 1.3337012219156483, "grad_norm": 0.41242173314094543, "learning_rate": 0.0005, "loss": 1.4168155193328857, "step": 16918 }, { "epoch": 1.3337800551832872, "grad_norm": 0.5957509875297546, "learning_rate": 0.0005, "loss": 1.803608775138855, "step": 16919 }, { "epoch": 1.3338588884509264, "grad_norm": 0.7503387331962585, "learning_rate": 0.0005, "loss": 1.7424396276474, "step": 16920 }, { "epoch": 1.3339377217185653, "grad_norm": 0.7155773043632507, "learning_rate": 0.0005, "loss": 2.1235909461975098, "step": 16921 }, { "epoch": 1.3340165549862042, "grad_norm": 0.4858764708042145, "learning_rate": 0.0005, "loss": 1.641960859298706, "step": 16922 }, { "epoch": 1.334095388253843, "grad_norm": 0.4970431327819824, "learning_rate": 0.0005, "loss": 1.5575897693634033, "step": 16923 }, { "epoch": 1.334174221521482, "grad_norm": 0.5299962759017944, "learning_rate": 0.0005, "loss": 1.7028207778930664, "step": 16924 }, { "epoch": 1.334253054789121, "grad_norm": 0.505479633808136, "learning_rate": 0.0005, "loss": 1.8572123050689697, "step": 16925 }, { "epoch": 1.3343318880567598, "grad_norm": 0.6059858202934265, "learning_rate": 0.0005, "loss": 1.7347261905670166, "step": 16926 }, { "epoch": 1.334410721324399, "grad_norm": 0.6173233985900879, "learning_rate": 0.0005, "loss": 1.7245781421661377, "step": 16927 }, { "epoch": 1.3344895545920379, "grad_norm": 0.5362166166305542, "learning_rate": 0.0005, "loss": 1.6951087713241577, "step": 16928 }, { "epoch": 1.3345683878596768, "grad_norm": 0.43700364232063293, "learning_rate": 0.0005, "loss": 1.4979478120803833, "step": 16929 }, { "epoch": 1.3346472211273157, "grad_norm": 0.58250492811203, "learning_rate": 0.0005, "loss": 1.5925517082214355, "step": 16930 }, { "epoch": 1.3347260543949546, "grad_norm": 0.7128509283065796, "learning_rate": 0.0005, "loss": 1.6983437538146973, "step": 16931 }, { "epoch": 1.3348048876625938, "grad_norm": 0.544287383556366, "learning_rate": 0.0005, "loss": 1.8097116947174072, "step": 16932 }, { "epoch": 1.3348837209302324, "grad_norm": 0.6177117824554443, "learning_rate": 0.0005, "loss": 1.5577425956726074, "step": 16933 }, { "epoch": 1.3349625541978716, "grad_norm": 0.5186113715171814, "learning_rate": 0.0005, "loss": 1.4073991775512695, "step": 16934 }, { "epoch": 1.3350413874655105, "grad_norm": 0.5741175413131714, "learning_rate": 0.0005, "loss": 1.7111492156982422, "step": 16935 }, { "epoch": 1.3351202207331494, "grad_norm": 0.6187341809272766, "learning_rate": 0.0005, "loss": 1.7181061506271362, "step": 16936 }, { "epoch": 1.3351990540007883, "grad_norm": 0.4950772523880005, "learning_rate": 0.0005, "loss": 1.6527501344680786, "step": 16937 }, { "epoch": 1.3352778872684272, "grad_norm": 0.4622464179992676, "learning_rate": 0.0005, "loss": 1.7594656944274902, "step": 16938 }, { "epoch": 1.3353567205360664, "grad_norm": 0.7308312654495239, "learning_rate": 0.0005, "loss": 1.7393159866333008, "step": 16939 }, { "epoch": 1.335435553803705, "grad_norm": 0.5796992182731628, "learning_rate": 0.0005, "loss": 1.5967717170715332, "step": 16940 }, { "epoch": 1.3355143870713442, "grad_norm": 0.5957880616188049, "learning_rate": 0.0005, "loss": 1.7418410778045654, "step": 16941 }, { "epoch": 1.335593220338983, "grad_norm": 0.5202543139457703, "learning_rate": 0.0005, "loss": 1.6510734558105469, "step": 16942 }, { "epoch": 1.335672053606622, "grad_norm": 0.526402473449707, "learning_rate": 0.0005, "loss": 1.8252778053283691, "step": 16943 }, { "epoch": 1.335750886874261, "grad_norm": 0.4951610565185547, "learning_rate": 0.0005, "loss": 1.4505565166473389, "step": 16944 }, { "epoch": 1.3358297201418998, "grad_norm": 0.7212283611297607, "learning_rate": 0.0005, "loss": 1.7298418283462524, "step": 16945 }, { "epoch": 1.335908553409539, "grad_norm": 0.4997425377368927, "learning_rate": 0.0005, "loss": 1.7165595293045044, "step": 16946 }, { "epoch": 1.3359873866771776, "grad_norm": 0.7056652307510376, "learning_rate": 0.0005, "loss": 1.846520185470581, "step": 16947 }, { "epoch": 1.3360662199448168, "grad_norm": 0.5023261308670044, "learning_rate": 0.0005, "loss": 1.393963098526001, "step": 16948 }, { "epoch": 1.3361450532124557, "grad_norm": 0.4681639075279236, "learning_rate": 0.0005, "loss": 1.6763732433319092, "step": 16949 }, { "epoch": 1.3362238864800946, "grad_norm": 0.5076488852500916, "learning_rate": 0.0005, "loss": 1.7168161869049072, "step": 16950 }, { "epoch": 1.3363027197477335, "grad_norm": 0.612993597984314, "learning_rate": 0.0005, "loss": 1.7384026050567627, "step": 16951 }, { "epoch": 1.3363815530153724, "grad_norm": 0.48663151264190674, "learning_rate": 0.0005, "loss": 1.4754984378814697, "step": 16952 }, { "epoch": 1.3364603862830116, "grad_norm": 0.6103612780570984, "learning_rate": 0.0005, "loss": 1.7870266437530518, "step": 16953 }, { "epoch": 1.3365392195506502, "grad_norm": 0.5898058414459229, "learning_rate": 0.0005, "loss": 1.5409669876098633, "step": 16954 }, { "epoch": 1.3366180528182894, "grad_norm": 0.6173139214515686, "learning_rate": 0.0005, "loss": 1.8586721420288086, "step": 16955 }, { "epoch": 1.3366968860859283, "grad_norm": 0.49603408575057983, "learning_rate": 0.0005, "loss": 1.4288065433502197, "step": 16956 }, { "epoch": 1.3367757193535672, "grad_norm": 0.5292368531227112, "learning_rate": 0.0005, "loss": 1.7799513339996338, "step": 16957 }, { "epoch": 1.3368545526212061, "grad_norm": 0.5299243927001953, "learning_rate": 0.0005, "loss": 1.5897530317306519, "step": 16958 }, { "epoch": 1.336933385888845, "grad_norm": 0.5300573110580444, "learning_rate": 0.0005, "loss": 1.7065484523773193, "step": 16959 }, { "epoch": 1.3370122191564842, "grad_norm": 0.7080182433128357, "learning_rate": 0.0005, "loss": 1.7593626976013184, "step": 16960 }, { "epoch": 1.3370910524241229, "grad_norm": 0.5564019680023193, "learning_rate": 0.0005, "loss": 1.71016526222229, "step": 16961 }, { "epoch": 1.337169885691762, "grad_norm": 0.6850488781929016, "learning_rate": 0.0005, "loss": 1.6678187847137451, "step": 16962 }, { "epoch": 1.337248718959401, "grad_norm": 0.4638557732105255, "learning_rate": 0.0005, "loss": 1.7010598182678223, "step": 16963 }, { "epoch": 1.3373275522270398, "grad_norm": 0.5041342377662659, "learning_rate": 0.0005, "loss": 1.6805236339569092, "step": 16964 }, { "epoch": 1.3374063854946787, "grad_norm": 0.6960902810096741, "learning_rate": 0.0005, "loss": 1.9233505725860596, "step": 16965 }, { "epoch": 1.3374852187623176, "grad_norm": 0.5762426257133484, "learning_rate": 0.0005, "loss": 1.844785451889038, "step": 16966 }, { "epoch": 1.3375640520299568, "grad_norm": 0.4871593415737152, "learning_rate": 0.0005, "loss": 1.7171049118041992, "step": 16967 }, { "epoch": 1.3376428852975955, "grad_norm": 0.49222445487976074, "learning_rate": 0.0005, "loss": 1.51504385471344, "step": 16968 }, { "epoch": 1.3377217185652346, "grad_norm": 0.5683438181877136, "learning_rate": 0.0005, "loss": 1.652848482131958, "step": 16969 }, { "epoch": 1.3378005518328735, "grad_norm": 0.7175502777099609, "learning_rate": 0.0005, "loss": 1.6586049795150757, "step": 16970 }, { "epoch": 1.3378793851005124, "grad_norm": 0.45409640669822693, "learning_rate": 0.0005, "loss": 1.4577202796936035, "step": 16971 }, { "epoch": 1.3379582183681513, "grad_norm": 0.4971363842487335, "learning_rate": 0.0005, "loss": 1.7718961238861084, "step": 16972 }, { "epoch": 1.3380370516357902, "grad_norm": 0.6113878488540649, "learning_rate": 0.0005, "loss": 1.6849390268325806, "step": 16973 }, { "epoch": 1.3381158849034294, "grad_norm": 0.6373469829559326, "learning_rate": 0.0005, "loss": 1.623523473739624, "step": 16974 }, { "epoch": 1.338194718171068, "grad_norm": 0.5392663478851318, "learning_rate": 0.0005, "loss": 1.6488440036773682, "step": 16975 }, { "epoch": 1.3382735514387072, "grad_norm": 0.4781433939933777, "learning_rate": 0.0005, "loss": 1.659077525138855, "step": 16976 }, { "epoch": 1.338352384706346, "grad_norm": 0.5587323904037476, "learning_rate": 0.0005, "loss": 1.5809290409088135, "step": 16977 }, { "epoch": 1.338431217973985, "grad_norm": 0.5346900224685669, "learning_rate": 0.0005, "loss": 1.70656418800354, "step": 16978 }, { "epoch": 1.338510051241624, "grad_norm": 0.6432377099990845, "learning_rate": 0.0005, "loss": 1.4612666368484497, "step": 16979 }, { "epoch": 1.3385888845092628, "grad_norm": 0.6280015110969543, "learning_rate": 0.0005, "loss": 1.5303640365600586, "step": 16980 }, { "epoch": 1.338667717776902, "grad_norm": 0.49356600642204285, "learning_rate": 0.0005, "loss": 1.5283433198928833, "step": 16981 }, { "epoch": 1.3387465510445409, "grad_norm": 0.5094454884529114, "learning_rate": 0.0005, "loss": 1.7141809463500977, "step": 16982 }, { "epoch": 1.3388253843121798, "grad_norm": 0.5570679306983948, "learning_rate": 0.0005, "loss": 1.7511003017425537, "step": 16983 }, { "epoch": 1.3389042175798187, "grad_norm": 0.5165258049964905, "learning_rate": 0.0005, "loss": 1.5399703979492188, "step": 16984 }, { "epoch": 1.3389830508474576, "grad_norm": 0.6288835406303406, "learning_rate": 0.0005, "loss": 1.7844598293304443, "step": 16985 }, { "epoch": 1.3390618841150965, "grad_norm": 0.5380828380584717, "learning_rate": 0.0005, "loss": 1.5671800374984741, "step": 16986 }, { "epoch": 1.3391407173827354, "grad_norm": 0.7399821281433105, "learning_rate": 0.0005, "loss": 1.6752773523330688, "step": 16987 }, { "epoch": 1.3392195506503746, "grad_norm": 0.6010465025901794, "learning_rate": 0.0005, "loss": 1.7480437755584717, "step": 16988 }, { "epoch": 1.3392983839180135, "grad_norm": 0.6236857175827026, "learning_rate": 0.0005, "loss": 1.708019495010376, "step": 16989 }, { "epoch": 1.3393772171856524, "grad_norm": 0.5302771329879761, "learning_rate": 0.0005, "loss": 1.707139253616333, "step": 16990 }, { "epoch": 1.3394560504532913, "grad_norm": 0.5874641537666321, "learning_rate": 0.0005, "loss": 1.5410146713256836, "step": 16991 }, { "epoch": 1.3395348837209302, "grad_norm": 0.6338520050048828, "learning_rate": 0.0005, "loss": 1.8006501197814941, "step": 16992 }, { "epoch": 1.3396137169885691, "grad_norm": 0.5555810332298279, "learning_rate": 0.0005, "loss": 1.7060173749923706, "step": 16993 }, { "epoch": 1.339692550256208, "grad_norm": 0.78398597240448, "learning_rate": 0.0005, "loss": 1.8229913711547852, "step": 16994 }, { "epoch": 1.3397713835238472, "grad_norm": 0.5389684438705444, "learning_rate": 0.0005, "loss": 1.4644050598144531, "step": 16995 }, { "epoch": 1.339850216791486, "grad_norm": 0.46051064133644104, "learning_rate": 0.0005, "loss": 1.629917860031128, "step": 16996 }, { "epoch": 1.339929050059125, "grad_norm": 0.531683623790741, "learning_rate": 0.0005, "loss": 1.560346007347107, "step": 16997 }, { "epoch": 1.340007883326764, "grad_norm": 0.47184935212135315, "learning_rate": 0.0005, "loss": 1.5897605419158936, "step": 16998 }, { "epoch": 1.3400867165944028, "grad_norm": 0.7700624465942383, "learning_rate": 0.0005, "loss": 1.7229812145233154, "step": 16999 }, { "epoch": 1.3401655498620417, "grad_norm": 0.4971868097782135, "learning_rate": 0.0005, "loss": 1.7167397737503052, "step": 17000 }, { "epoch": 1.3402443831296806, "grad_norm": 0.4233873784542084, "learning_rate": 0.0005, "loss": 1.5974860191345215, "step": 17001 }, { "epoch": 1.3403232163973198, "grad_norm": 0.5481891632080078, "learning_rate": 0.0005, "loss": 1.7537307739257812, "step": 17002 }, { "epoch": 1.3404020496649587, "grad_norm": 0.6615986227989197, "learning_rate": 0.0005, "loss": 1.5779263973236084, "step": 17003 }, { "epoch": 1.3404808829325976, "grad_norm": 0.47640034556388855, "learning_rate": 0.0005, "loss": 1.5923107862472534, "step": 17004 }, { "epoch": 1.3405597162002365, "grad_norm": 0.5534741878509521, "learning_rate": 0.0005, "loss": 1.7501517534255981, "step": 17005 }, { "epoch": 1.3406385494678754, "grad_norm": 0.6524899005889893, "learning_rate": 0.0005, "loss": 1.894429326057434, "step": 17006 }, { "epoch": 1.3407173827355143, "grad_norm": 0.8490433692932129, "learning_rate": 0.0005, "loss": 1.6078155040740967, "step": 17007 }, { "epoch": 1.3407962160031532, "grad_norm": 0.630115270614624, "learning_rate": 0.0005, "loss": 1.6360828876495361, "step": 17008 }, { "epoch": 1.3408750492707924, "grad_norm": 0.5048837065696716, "learning_rate": 0.0005, "loss": 1.730169653892517, "step": 17009 }, { "epoch": 1.3409538825384313, "grad_norm": 0.48553621768951416, "learning_rate": 0.0005, "loss": 1.6200367212295532, "step": 17010 }, { "epoch": 1.3410327158060702, "grad_norm": 0.5963506102561951, "learning_rate": 0.0005, "loss": 1.863588809967041, "step": 17011 }, { "epoch": 1.341111549073709, "grad_norm": 0.6107916235923767, "learning_rate": 0.0005, "loss": 1.8010754585266113, "step": 17012 }, { "epoch": 1.341190382341348, "grad_norm": 0.5801379680633545, "learning_rate": 0.0005, "loss": 1.3960468769073486, "step": 17013 }, { "epoch": 1.341269215608987, "grad_norm": 0.4944095015525818, "learning_rate": 0.0005, "loss": 1.5803580284118652, "step": 17014 }, { "epoch": 1.3413480488766258, "grad_norm": 0.48520517349243164, "learning_rate": 0.0005, "loss": 1.6403133869171143, "step": 17015 }, { "epoch": 1.341426882144265, "grad_norm": 0.5625271797180176, "learning_rate": 0.0005, "loss": 1.8372949361801147, "step": 17016 }, { "epoch": 1.3415057154119039, "grad_norm": 0.4856607913970947, "learning_rate": 0.0005, "loss": 1.6977821588516235, "step": 17017 }, { "epoch": 1.3415845486795428, "grad_norm": 0.5223902463912964, "learning_rate": 0.0005, "loss": 1.6485676765441895, "step": 17018 }, { "epoch": 1.3416633819471817, "grad_norm": 0.5719022750854492, "learning_rate": 0.0005, "loss": 1.5993974208831787, "step": 17019 }, { "epoch": 1.3417422152148206, "grad_norm": 0.5049694776535034, "learning_rate": 0.0005, "loss": 1.6934373378753662, "step": 17020 }, { "epoch": 1.3418210484824595, "grad_norm": 0.5137405395507812, "learning_rate": 0.0005, "loss": 1.7620875835418701, "step": 17021 }, { "epoch": 1.3418998817500984, "grad_norm": 0.547039806842804, "learning_rate": 0.0005, "loss": 1.4800035953521729, "step": 17022 }, { "epoch": 1.3419787150177376, "grad_norm": 0.5283913016319275, "learning_rate": 0.0005, "loss": 1.4967503547668457, "step": 17023 }, { "epoch": 1.3420575482853765, "grad_norm": 0.5018166303634644, "learning_rate": 0.0005, "loss": 1.717423677444458, "step": 17024 }, { "epoch": 1.3421363815530154, "grad_norm": 0.5190656781196594, "learning_rate": 0.0005, "loss": 1.6795458793640137, "step": 17025 }, { "epoch": 1.3422152148206543, "grad_norm": 0.6441159844398499, "learning_rate": 0.0005, "loss": 1.7446043491363525, "step": 17026 }, { "epoch": 1.3422940480882932, "grad_norm": 0.5652977228164673, "learning_rate": 0.0005, "loss": 1.7817862033843994, "step": 17027 }, { "epoch": 1.3423728813559321, "grad_norm": 0.4619113504886627, "learning_rate": 0.0005, "loss": 1.6316249370574951, "step": 17028 }, { "epoch": 1.342451714623571, "grad_norm": 0.5345745086669922, "learning_rate": 0.0005, "loss": 1.515997290611267, "step": 17029 }, { "epoch": 1.3425305478912102, "grad_norm": 0.5831396579742432, "learning_rate": 0.0005, "loss": 1.6847405433654785, "step": 17030 }, { "epoch": 1.342609381158849, "grad_norm": 0.4145679175853729, "learning_rate": 0.0005, "loss": 1.5810259580612183, "step": 17031 }, { "epoch": 1.342688214426488, "grad_norm": 0.5625083446502686, "learning_rate": 0.0005, "loss": 1.7485748529434204, "step": 17032 }, { "epoch": 1.342767047694127, "grad_norm": 0.585161566734314, "learning_rate": 0.0005, "loss": 1.7141554355621338, "step": 17033 }, { "epoch": 1.3428458809617658, "grad_norm": 0.5542265176773071, "learning_rate": 0.0005, "loss": 1.7106025218963623, "step": 17034 }, { "epoch": 1.3429247142294047, "grad_norm": 0.9021106958389282, "learning_rate": 0.0005, "loss": 1.8119115829467773, "step": 17035 }, { "epoch": 1.3430035474970436, "grad_norm": 0.5373887419700623, "learning_rate": 0.0005, "loss": 1.7378952503204346, "step": 17036 }, { "epoch": 1.3430823807646828, "grad_norm": 0.5519514679908752, "learning_rate": 0.0005, "loss": 1.537126064300537, "step": 17037 }, { "epoch": 1.3431612140323217, "grad_norm": 0.5338985323905945, "learning_rate": 0.0005, "loss": 1.6900038719177246, "step": 17038 }, { "epoch": 1.3432400472999606, "grad_norm": 0.41779783368110657, "learning_rate": 0.0005, "loss": 1.471431016921997, "step": 17039 }, { "epoch": 1.3433188805675995, "grad_norm": 0.5864909887313843, "learning_rate": 0.0005, "loss": 1.755692958831787, "step": 17040 }, { "epoch": 1.3433977138352384, "grad_norm": 0.483761727809906, "learning_rate": 0.0005, "loss": 1.5190379619598389, "step": 17041 }, { "epoch": 1.3434765471028773, "grad_norm": 0.645695149898529, "learning_rate": 0.0005, "loss": 1.7380057573318481, "step": 17042 }, { "epoch": 1.3435553803705163, "grad_norm": 0.5441130995750427, "learning_rate": 0.0005, "loss": 1.658154010772705, "step": 17043 }, { "epoch": 1.3436342136381554, "grad_norm": 0.466880202293396, "learning_rate": 0.0005, "loss": 1.6713414192199707, "step": 17044 }, { "epoch": 1.3437130469057943, "grad_norm": 0.6388596892356873, "learning_rate": 0.0005, "loss": 1.7367303371429443, "step": 17045 }, { "epoch": 1.3437918801734332, "grad_norm": 0.6789270043373108, "learning_rate": 0.0005, "loss": 1.7516765594482422, "step": 17046 }, { "epoch": 1.3438707134410721, "grad_norm": 0.5842152237892151, "learning_rate": 0.0005, "loss": 1.672668218612671, "step": 17047 }, { "epoch": 1.343949546708711, "grad_norm": 0.4480317533016205, "learning_rate": 0.0005, "loss": 1.4897655248641968, "step": 17048 }, { "epoch": 1.34402837997635, "grad_norm": 0.49826592206954956, "learning_rate": 0.0005, "loss": 1.4452495574951172, "step": 17049 }, { "epoch": 1.3441072132439889, "grad_norm": 0.5963983535766602, "learning_rate": 0.0005, "loss": 1.5083253383636475, "step": 17050 }, { "epoch": 1.344186046511628, "grad_norm": 0.7950472235679626, "learning_rate": 0.0005, "loss": 1.765429973602295, "step": 17051 }, { "epoch": 1.344264879779267, "grad_norm": 0.5133347511291504, "learning_rate": 0.0005, "loss": 1.6868290901184082, "step": 17052 }, { "epoch": 1.3443437130469058, "grad_norm": 0.4407353401184082, "learning_rate": 0.0005, "loss": 1.4550836086273193, "step": 17053 }, { "epoch": 1.3444225463145447, "grad_norm": 0.601595401763916, "learning_rate": 0.0005, "loss": 1.717302680015564, "step": 17054 }, { "epoch": 1.3445013795821836, "grad_norm": 0.5537901520729065, "learning_rate": 0.0005, "loss": 1.4151949882507324, "step": 17055 }, { "epoch": 1.3445802128498225, "grad_norm": 0.4051155745983124, "learning_rate": 0.0005, "loss": 1.6515791416168213, "step": 17056 }, { "epoch": 1.3446590461174615, "grad_norm": 0.8217101097106934, "learning_rate": 0.0005, "loss": 1.6991403102874756, "step": 17057 }, { "epoch": 1.3447378793851006, "grad_norm": 0.5353253483772278, "learning_rate": 0.0005, "loss": 1.519499659538269, "step": 17058 }, { "epoch": 1.3448167126527395, "grad_norm": 0.6414149403572083, "learning_rate": 0.0005, "loss": 1.8250749111175537, "step": 17059 }, { "epoch": 1.3448955459203784, "grad_norm": 0.7501948475837708, "learning_rate": 0.0005, "loss": 1.8319177627563477, "step": 17060 }, { "epoch": 1.3449743791880173, "grad_norm": 0.6088826656341553, "learning_rate": 0.0005, "loss": 1.7805736064910889, "step": 17061 }, { "epoch": 1.3450532124556562, "grad_norm": 0.48338520526885986, "learning_rate": 0.0005, "loss": 1.6280204057693481, "step": 17062 }, { "epoch": 1.3451320457232954, "grad_norm": 0.508361279964447, "learning_rate": 0.0005, "loss": 1.7342158555984497, "step": 17063 }, { "epoch": 1.345210878990934, "grad_norm": 0.6847641468048096, "learning_rate": 0.0005, "loss": 1.5838111639022827, "step": 17064 }, { "epoch": 1.3452897122585732, "grad_norm": 0.49824976921081543, "learning_rate": 0.0005, "loss": 1.800977110862732, "step": 17065 }, { "epoch": 1.345368545526212, "grad_norm": 0.550194263458252, "learning_rate": 0.0005, "loss": 1.6551902294158936, "step": 17066 }, { "epoch": 1.345447378793851, "grad_norm": 0.39826086163520813, "learning_rate": 0.0005, "loss": 1.4792275428771973, "step": 17067 }, { "epoch": 1.34552621206149, "grad_norm": 0.45803171396255493, "learning_rate": 0.0005, "loss": 1.4470582008361816, "step": 17068 }, { "epoch": 1.3456050453291288, "grad_norm": 0.5372465252876282, "learning_rate": 0.0005, "loss": 1.7819095849990845, "step": 17069 }, { "epoch": 1.345683878596768, "grad_norm": 0.4642549753189087, "learning_rate": 0.0005, "loss": 1.524604082107544, "step": 17070 }, { "epoch": 1.3457627118644067, "grad_norm": 0.5478715896606445, "learning_rate": 0.0005, "loss": 1.8152968883514404, "step": 17071 }, { "epoch": 1.3458415451320458, "grad_norm": 0.5097022652626038, "learning_rate": 0.0005, "loss": 1.4854466915130615, "step": 17072 }, { "epoch": 1.3459203783996847, "grad_norm": 0.506356954574585, "learning_rate": 0.0005, "loss": 1.6740210056304932, "step": 17073 }, { "epoch": 1.3459992116673236, "grad_norm": 0.615975022315979, "learning_rate": 0.0005, "loss": 1.5882198810577393, "step": 17074 }, { "epoch": 1.3460780449349625, "grad_norm": 0.7683869004249573, "learning_rate": 0.0005, "loss": 1.8043291568756104, "step": 17075 }, { "epoch": 1.3461568782026014, "grad_norm": 0.7712413668632507, "learning_rate": 0.0005, "loss": 1.5797712802886963, "step": 17076 }, { "epoch": 1.3462357114702406, "grad_norm": 0.5193880200386047, "learning_rate": 0.0005, "loss": 1.533888816833496, "step": 17077 }, { "epoch": 1.3463145447378793, "grad_norm": 0.4726329743862152, "learning_rate": 0.0005, "loss": 1.5851855278015137, "step": 17078 }, { "epoch": 1.3463933780055184, "grad_norm": 0.5251489877700806, "learning_rate": 0.0005, "loss": 1.5475599765777588, "step": 17079 }, { "epoch": 1.3464722112731573, "grad_norm": 0.6779044270515442, "learning_rate": 0.0005, "loss": 1.8220195770263672, "step": 17080 }, { "epoch": 1.3465510445407962, "grad_norm": 0.5268751978874207, "learning_rate": 0.0005, "loss": 1.7823302745819092, "step": 17081 }, { "epoch": 1.3466298778084351, "grad_norm": 0.5980405807495117, "learning_rate": 0.0005, "loss": 1.9042986631393433, "step": 17082 }, { "epoch": 1.346708711076074, "grad_norm": 0.6359687447547913, "learning_rate": 0.0005, "loss": 1.772046446800232, "step": 17083 }, { "epoch": 1.3467875443437132, "grad_norm": 0.5152822732925415, "learning_rate": 0.0005, "loss": 1.7348508834838867, "step": 17084 }, { "epoch": 1.3468663776113519, "grad_norm": 0.6680793166160583, "learning_rate": 0.0005, "loss": 1.7397360801696777, "step": 17085 }, { "epoch": 1.346945210878991, "grad_norm": 0.5632125735282898, "learning_rate": 0.0005, "loss": 1.6741634607315063, "step": 17086 }, { "epoch": 1.34702404414663, "grad_norm": 0.49546244740486145, "learning_rate": 0.0005, "loss": 1.7984944581985474, "step": 17087 }, { "epoch": 1.3471028774142688, "grad_norm": 0.5890067219734192, "learning_rate": 0.0005, "loss": 1.7372117042541504, "step": 17088 }, { "epoch": 1.3471817106819077, "grad_norm": 0.5546272993087769, "learning_rate": 0.0005, "loss": 1.7343032360076904, "step": 17089 }, { "epoch": 1.3472605439495466, "grad_norm": 0.541762113571167, "learning_rate": 0.0005, "loss": 1.5399340391159058, "step": 17090 }, { "epoch": 1.3473393772171858, "grad_norm": 0.5874713659286499, "learning_rate": 0.0005, "loss": 1.6467387676239014, "step": 17091 }, { "epoch": 1.3474182104848245, "grad_norm": 0.5704975128173828, "learning_rate": 0.0005, "loss": 1.7038002014160156, "step": 17092 }, { "epoch": 1.3474970437524636, "grad_norm": 1.124488353729248, "learning_rate": 0.0005, "loss": 1.5276261568069458, "step": 17093 }, { "epoch": 1.3475758770201025, "grad_norm": 0.5486860275268555, "learning_rate": 0.0005, "loss": 1.6457841396331787, "step": 17094 }, { "epoch": 1.3476547102877414, "grad_norm": 0.6773713827133179, "learning_rate": 0.0005, "loss": 1.9089505672454834, "step": 17095 }, { "epoch": 1.3477335435553803, "grad_norm": 0.5657913684844971, "learning_rate": 0.0005, "loss": 1.8521475791931152, "step": 17096 }, { "epoch": 1.3478123768230192, "grad_norm": 0.5104454755783081, "learning_rate": 0.0005, "loss": 1.645660638809204, "step": 17097 }, { "epoch": 1.3478912100906584, "grad_norm": 0.587005615234375, "learning_rate": 0.0005, "loss": 1.665873646736145, "step": 17098 }, { "epoch": 1.347970043358297, "grad_norm": 0.6403939723968506, "learning_rate": 0.0005, "loss": 1.4510047435760498, "step": 17099 }, { "epoch": 1.3480488766259362, "grad_norm": 0.6017516255378723, "learning_rate": 0.0005, "loss": 1.72989821434021, "step": 17100 }, { "epoch": 1.348127709893575, "grad_norm": 0.5438207387924194, "learning_rate": 0.0005, "loss": 1.7222671508789062, "step": 17101 }, { "epoch": 1.348206543161214, "grad_norm": 0.5817335844039917, "learning_rate": 0.0005, "loss": 1.7806241512298584, "step": 17102 }, { "epoch": 1.348285376428853, "grad_norm": 0.6607393622398376, "learning_rate": 0.0005, "loss": 1.6497421264648438, "step": 17103 }, { "epoch": 1.3483642096964918, "grad_norm": 0.6827430725097656, "learning_rate": 0.0005, "loss": 1.6488960981369019, "step": 17104 }, { "epoch": 1.348443042964131, "grad_norm": 0.5664358735084534, "learning_rate": 0.0005, "loss": 1.636850118637085, "step": 17105 }, { "epoch": 1.34852187623177, "grad_norm": 0.4973175525665283, "learning_rate": 0.0005, "loss": 1.3384976387023926, "step": 17106 }, { "epoch": 1.3486007094994088, "grad_norm": 0.41458138823509216, "learning_rate": 0.0005, "loss": 1.6611055135726929, "step": 17107 }, { "epoch": 1.3486795427670477, "grad_norm": 0.6224233508110046, "learning_rate": 0.0005, "loss": 1.9072901010513306, "step": 17108 }, { "epoch": 1.3487583760346866, "grad_norm": 0.5369774699211121, "learning_rate": 0.0005, "loss": 1.828887939453125, "step": 17109 }, { "epoch": 1.3488372093023255, "grad_norm": 0.5309738516807556, "learning_rate": 0.0005, "loss": 1.8112417459487915, "step": 17110 }, { "epoch": 1.3489160425699644, "grad_norm": 0.4732830822467804, "learning_rate": 0.0005, "loss": 1.6342535018920898, "step": 17111 }, { "epoch": 1.3489948758376036, "grad_norm": 0.4555629789829254, "learning_rate": 0.0005, "loss": 1.5016251802444458, "step": 17112 }, { "epoch": 1.3490737091052425, "grad_norm": 0.44840729236602783, "learning_rate": 0.0005, "loss": 1.640192985534668, "step": 17113 }, { "epoch": 1.3491525423728814, "grad_norm": 0.42311668395996094, "learning_rate": 0.0005, "loss": 1.4574625492095947, "step": 17114 }, { "epoch": 1.3492313756405203, "grad_norm": 0.4713023602962494, "learning_rate": 0.0005, "loss": 1.5737159252166748, "step": 17115 }, { "epoch": 1.3493102089081592, "grad_norm": 0.48386484384536743, "learning_rate": 0.0005, "loss": 1.5227258205413818, "step": 17116 }, { "epoch": 1.3493890421757981, "grad_norm": 0.5228965282440186, "learning_rate": 0.0005, "loss": 1.728213906288147, "step": 17117 }, { "epoch": 1.349467875443437, "grad_norm": 0.5026319622993469, "learning_rate": 0.0005, "loss": 1.643510341644287, "step": 17118 }, { "epoch": 1.3495467087110762, "grad_norm": 0.5232526659965515, "learning_rate": 0.0005, "loss": 1.6517261266708374, "step": 17119 }, { "epoch": 1.349625541978715, "grad_norm": 0.6554617881774902, "learning_rate": 0.0005, "loss": 1.7363702058792114, "step": 17120 }, { "epoch": 1.349704375246354, "grad_norm": 0.6593866348266602, "learning_rate": 0.0005, "loss": 1.6288890838623047, "step": 17121 }, { "epoch": 1.349783208513993, "grad_norm": 0.5488573908805847, "learning_rate": 0.0005, "loss": 1.5052838325500488, "step": 17122 }, { "epoch": 1.3498620417816318, "grad_norm": 0.5787818431854248, "learning_rate": 0.0005, "loss": 1.474936842918396, "step": 17123 }, { "epoch": 1.3499408750492707, "grad_norm": 0.4924056828022003, "learning_rate": 0.0005, "loss": 1.5210938453674316, "step": 17124 }, { "epoch": 1.3500197083169097, "grad_norm": 0.516377866268158, "learning_rate": 0.0005, "loss": 1.5386385917663574, "step": 17125 }, { "epoch": 1.3500985415845488, "grad_norm": 0.6717383861541748, "learning_rate": 0.0005, "loss": 1.6923868656158447, "step": 17126 }, { "epoch": 1.3501773748521877, "grad_norm": 0.5841708183288574, "learning_rate": 0.0005, "loss": 1.9282565116882324, "step": 17127 }, { "epoch": 1.3502562081198266, "grad_norm": 0.535391628742218, "learning_rate": 0.0005, "loss": 1.4236773252487183, "step": 17128 }, { "epoch": 1.3503350413874655, "grad_norm": 0.5887537598609924, "learning_rate": 0.0005, "loss": 1.8559350967407227, "step": 17129 }, { "epoch": 1.3504138746551044, "grad_norm": 0.6893585324287415, "learning_rate": 0.0005, "loss": 1.9127929210662842, "step": 17130 }, { "epoch": 1.3504927079227433, "grad_norm": 0.5775913596153259, "learning_rate": 0.0005, "loss": 1.802130937576294, "step": 17131 }, { "epoch": 1.3505715411903823, "grad_norm": 0.5377784371376038, "learning_rate": 0.0005, "loss": 1.5831488370895386, "step": 17132 }, { "epoch": 1.3506503744580214, "grad_norm": 0.5315208435058594, "learning_rate": 0.0005, "loss": 1.5722458362579346, "step": 17133 }, { "epoch": 1.3507292077256603, "grad_norm": 0.5234000086784363, "learning_rate": 0.0005, "loss": 1.5578501224517822, "step": 17134 }, { "epoch": 1.3508080409932992, "grad_norm": 0.8958198428153992, "learning_rate": 0.0005, "loss": 1.8092880249023438, "step": 17135 }, { "epoch": 1.3508868742609381, "grad_norm": 0.6511502861976624, "learning_rate": 0.0005, "loss": 1.700268268585205, "step": 17136 }, { "epoch": 1.350965707528577, "grad_norm": 0.7296676635742188, "learning_rate": 0.0005, "loss": 1.8092541694641113, "step": 17137 }, { "epoch": 1.351044540796216, "grad_norm": 0.6335142850875854, "learning_rate": 0.0005, "loss": 1.8993659019470215, "step": 17138 }, { "epoch": 1.3511233740638549, "grad_norm": 0.6569833755493164, "learning_rate": 0.0005, "loss": 1.8339855670928955, "step": 17139 }, { "epoch": 1.351202207331494, "grad_norm": 0.6562516689300537, "learning_rate": 0.0005, "loss": 1.7173426151275635, "step": 17140 }, { "epoch": 1.351281040599133, "grad_norm": 0.5282251238822937, "learning_rate": 0.0005, "loss": 1.5670900344848633, "step": 17141 }, { "epoch": 1.3513598738667718, "grad_norm": 0.6520634889602661, "learning_rate": 0.0005, "loss": 1.8651912212371826, "step": 17142 }, { "epoch": 1.3514387071344107, "grad_norm": 0.6296178698539734, "learning_rate": 0.0005, "loss": 1.644235610961914, "step": 17143 }, { "epoch": 1.3515175404020496, "grad_norm": 0.6594524383544922, "learning_rate": 0.0005, "loss": 1.699150800704956, "step": 17144 }, { "epoch": 1.3515963736696885, "grad_norm": 0.5543392896652222, "learning_rate": 0.0005, "loss": 1.7029423713684082, "step": 17145 }, { "epoch": 1.3516752069373275, "grad_norm": 0.6721811890602112, "learning_rate": 0.0005, "loss": 1.6640892028808594, "step": 17146 }, { "epoch": 1.3517540402049666, "grad_norm": 0.6319493055343628, "learning_rate": 0.0005, "loss": 1.508495807647705, "step": 17147 }, { "epoch": 1.3518328734726055, "grad_norm": 0.5269558429718018, "learning_rate": 0.0005, "loss": 1.57564377784729, "step": 17148 }, { "epoch": 1.3519117067402444, "grad_norm": 0.5751450657844543, "learning_rate": 0.0005, "loss": 1.7831977605819702, "step": 17149 }, { "epoch": 1.3519905400078833, "grad_norm": 0.7881328463554382, "learning_rate": 0.0005, "loss": 1.7067288160324097, "step": 17150 }, { "epoch": 1.3520693732755222, "grad_norm": 0.45325544476509094, "learning_rate": 0.0005, "loss": 1.4965550899505615, "step": 17151 }, { "epoch": 1.3521482065431611, "grad_norm": 0.4223017692565918, "learning_rate": 0.0005, "loss": 1.5304384231567383, "step": 17152 }, { "epoch": 1.3522270398108, "grad_norm": 0.6316917538642883, "learning_rate": 0.0005, "loss": 1.7477412223815918, "step": 17153 }, { "epoch": 1.3523058730784392, "grad_norm": 0.7882511019706726, "learning_rate": 0.0005, "loss": 1.7524311542510986, "step": 17154 }, { "epoch": 1.352384706346078, "grad_norm": 1.3765270709991455, "learning_rate": 0.0005, "loss": 1.5450154542922974, "step": 17155 }, { "epoch": 1.352463539613717, "grad_norm": 0.5003346800804138, "learning_rate": 0.0005, "loss": 1.4752535820007324, "step": 17156 }, { "epoch": 1.352542372881356, "grad_norm": 0.5728618502616882, "learning_rate": 0.0005, "loss": 1.3971855640411377, "step": 17157 }, { "epoch": 1.3526212061489948, "grad_norm": 0.7966087460517883, "learning_rate": 0.0005, "loss": 1.7361726760864258, "step": 17158 }, { "epoch": 1.3527000394166337, "grad_norm": 0.44306474924087524, "learning_rate": 0.0005, "loss": 1.529078483581543, "step": 17159 }, { "epoch": 1.3527788726842727, "grad_norm": 0.5620417594909668, "learning_rate": 0.0005, "loss": 1.3947417736053467, "step": 17160 }, { "epoch": 1.3528577059519118, "grad_norm": 0.45588743686676025, "learning_rate": 0.0005, "loss": 1.5279216766357422, "step": 17161 }, { "epoch": 1.3529365392195507, "grad_norm": 0.559544026851654, "learning_rate": 0.0005, "loss": 1.6822443008422852, "step": 17162 }, { "epoch": 1.3530153724871896, "grad_norm": 0.553739070892334, "learning_rate": 0.0005, "loss": 1.7007851600646973, "step": 17163 }, { "epoch": 1.3530942057548285, "grad_norm": 0.5682520866394043, "learning_rate": 0.0005, "loss": 1.9269413948059082, "step": 17164 }, { "epoch": 1.3531730390224674, "grad_norm": 0.5711919069290161, "learning_rate": 0.0005, "loss": 1.548638105392456, "step": 17165 }, { "epoch": 1.3532518722901064, "grad_norm": 0.5684086084365845, "learning_rate": 0.0005, "loss": 1.6193699836730957, "step": 17166 }, { "epoch": 1.3533307055577453, "grad_norm": 0.5456379652023315, "learning_rate": 0.0005, "loss": 1.8824515342712402, "step": 17167 }, { "epoch": 1.3534095388253844, "grad_norm": 0.7498877048492432, "learning_rate": 0.0005, "loss": 1.9317262172698975, "step": 17168 }, { "epoch": 1.3534883720930233, "grad_norm": 0.5697430372238159, "learning_rate": 0.0005, "loss": 1.8523226976394653, "step": 17169 }, { "epoch": 1.3535672053606622, "grad_norm": 0.47641539573669434, "learning_rate": 0.0005, "loss": 1.6648991107940674, "step": 17170 }, { "epoch": 1.3536460386283011, "grad_norm": 0.4157319962978363, "learning_rate": 0.0005, "loss": 1.5441856384277344, "step": 17171 }, { "epoch": 1.35372487189594, "grad_norm": 0.6079725027084351, "learning_rate": 0.0005, "loss": 1.589877724647522, "step": 17172 }, { "epoch": 1.353803705163579, "grad_norm": 0.5432350039482117, "learning_rate": 0.0005, "loss": 1.396538496017456, "step": 17173 }, { "epoch": 1.3538825384312179, "grad_norm": 0.5010142922401428, "learning_rate": 0.0005, "loss": 1.614587664604187, "step": 17174 }, { "epoch": 1.353961371698857, "grad_norm": 0.6102522611618042, "learning_rate": 0.0005, "loss": 1.5167549848556519, "step": 17175 }, { "epoch": 1.354040204966496, "grad_norm": 0.6310710906982422, "learning_rate": 0.0005, "loss": 2.0243821144104004, "step": 17176 }, { "epoch": 1.3541190382341348, "grad_norm": 0.6059812903404236, "learning_rate": 0.0005, "loss": 1.603079080581665, "step": 17177 }, { "epoch": 1.3541978715017737, "grad_norm": 0.5553594827651978, "learning_rate": 0.0005, "loss": 1.8112993240356445, "step": 17178 }, { "epoch": 1.3542767047694126, "grad_norm": 0.6159341335296631, "learning_rate": 0.0005, "loss": 1.8128857612609863, "step": 17179 }, { "epoch": 1.3543555380370516, "grad_norm": 0.5294838547706604, "learning_rate": 0.0005, "loss": 1.5612292289733887, "step": 17180 }, { "epoch": 1.3544343713046905, "grad_norm": 0.5541191101074219, "learning_rate": 0.0005, "loss": 1.6449618339538574, "step": 17181 }, { "epoch": 1.3545132045723296, "grad_norm": 0.5965999364852905, "learning_rate": 0.0005, "loss": 1.901390552520752, "step": 17182 }, { "epoch": 1.3545920378399685, "grad_norm": 0.6320273876190186, "learning_rate": 0.0005, "loss": 1.6997301578521729, "step": 17183 }, { "epoch": 1.3546708711076074, "grad_norm": 0.5265032649040222, "learning_rate": 0.0005, "loss": 1.87396240234375, "step": 17184 }, { "epoch": 1.3547497043752463, "grad_norm": 0.46828365325927734, "learning_rate": 0.0005, "loss": 1.6405274868011475, "step": 17185 }, { "epoch": 1.3548285376428852, "grad_norm": 0.5557410717010498, "learning_rate": 0.0005, "loss": 1.8917269706726074, "step": 17186 }, { "epoch": 1.3549073709105244, "grad_norm": 0.5215706825256348, "learning_rate": 0.0005, "loss": 1.4273598194122314, "step": 17187 }, { "epoch": 1.354986204178163, "grad_norm": 0.5149388909339905, "learning_rate": 0.0005, "loss": 1.5180574655532837, "step": 17188 }, { "epoch": 1.3550650374458022, "grad_norm": 0.5709458589553833, "learning_rate": 0.0005, "loss": 1.4489469528198242, "step": 17189 }, { "epoch": 1.3551438707134411, "grad_norm": 0.4256974458694458, "learning_rate": 0.0005, "loss": 1.3680119514465332, "step": 17190 }, { "epoch": 1.35522270398108, "grad_norm": 0.5257768034934998, "learning_rate": 0.0005, "loss": 1.871734857559204, "step": 17191 }, { "epoch": 1.355301537248719, "grad_norm": 0.8742582201957703, "learning_rate": 0.0005, "loss": 1.9197802543640137, "step": 17192 }, { "epoch": 1.3553803705163578, "grad_norm": 0.6017221212387085, "learning_rate": 0.0005, "loss": 1.8253034353256226, "step": 17193 }, { "epoch": 1.355459203783997, "grad_norm": 0.46632248163223267, "learning_rate": 0.0005, "loss": 1.728867769241333, "step": 17194 }, { "epoch": 1.3555380370516357, "grad_norm": 0.48666253685951233, "learning_rate": 0.0005, "loss": 1.4234262704849243, "step": 17195 }, { "epoch": 1.3556168703192748, "grad_norm": 0.5124859809875488, "learning_rate": 0.0005, "loss": 1.6774544715881348, "step": 17196 }, { "epoch": 1.3556957035869137, "grad_norm": 0.4525345265865326, "learning_rate": 0.0005, "loss": 1.6357308626174927, "step": 17197 }, { "epoch": 1.3557745368545526, "grad_norm": 0.4986307621002197, "learning_rate": 0.0005, "loss": 1.6204173564910889, "step": 17198 }, { "epoch": 1.3558533701221915, "grad_norm": 0.4670056998729706, "learning_rate": 0.0005, "loss": 1.521482229232788, "step": 17199 }, { "epoch": 1.3559322033898304, "grad_norm": 0.5538585186004639, "learning_rate": 0.0005, "loss": 1.6876883506774902, "step": 17200 }, { "epoch": 1.3560110366574696, "grad_norm": 0.5851719379425049, "learning_rate": 0.0005, "loss": 1.6831257343292236, "step": 17201 }, { "epoch": 1.3560898699251083, "grad_norm": 0.6258948445320129, "learning_rate": 0.0005, "loss": 1.7822717428207397, "step": 17202 }, { "epoch": 1.3561687031927474, "grad_norm": 0.4351820945739746, "learning_rate": 0.0005, "loss": 1.5442229509353638, "step": 17203 }, { "epoch": 1.3562475364603863, "grad_norm": 0.5198051929473877, "learning_rate": 0.0005, "loss": 1.4503676891326904, "step": 17204 }, { "epoch": 1.3563263697280252, "grad_norm": 0.45013877749443054, "learning_rate": 0.0005, "loss": 1.6244392395019531, "step": 17205 }, { "epoch": 1.3564052029956641, "grad_norm": 0.5035511255264282, "learning_rate": 0.0005, "loss": 1.6204787492752075, "step": 17206 }, { "epoch": 1.356484036263303, "grad_norm": 0.55112624168396, "learning_rate": 0.0005, "loss": 1.8795125484466553, "step": 17207 }, { "epoch": 1.3565628695309422, "grad_norm": 0.6282762289047241, "learning_rate": 0.0005, "loss": 1.7118849754333496, "step": 17208 }, { "epoch": 1.3566417027985809, "grad_norm": 0.49336135387420654, "learning_rate": 0.0005, "loss": 1.4802072048187256, "step": 17209 }, { "epoch": 1.35672053606622, "grad_norm": 0.7111196517944336, "learning_rate": 0.0005, "loss": 1.5931544303894043, "step": 17210 }, { "epoch": 1.356799369333859, "grad_norm": 0.5080333948135376, "learning_rate": 0.0005, "loss": 1.9707684516906738, "step": 17211 }, { "epoch": 1.3568782026014978, "grad_norm": 0.5576715469360352, "learning_rate": 0.0005, "loss": 1.5539486408233643, "step": 17212 }, { "epoch": 1.3569570358691367, "grad_norm": 0.6536260843276978, "learning_rate": 0.0005, "loss": 1.6363937854766846, "step": 17213 }, { "epoch": 1.3570358691367757, "grad_norm": 0.849033534526825, "learning_rate": 0.0005, "loss": 1.6807420253753662, "step": 17214 }, { "epoch": 1.3571147024044148, "grad_norm": 2.1993229389190674, "learning_rate": 0.0005, "loss": 1.633439540863037, "step": 17215 }, { "epoch": 1.3571935356720535, "grad_norm": 0.47725769877433777, "learning_rate": 0.0005, "loss": 1.446747064590454, "step": 17216 }, { "epoch": 1.3572723689396926, "grad_norm": 0.4737336039543152, "learning_rate": 0.0005, "loss": 1.5091557502746582, "step": 17217 }, { "epoch": 1.3573512022073315, "grad_norm": 0.5464900135993958, "learning_rate": 0.0005, "loss": 1.4505352973937988, "step": 17218 }, { "epoch": 1.3574300354749704, "grad_norm": 0.6193884611129761, "learning_rate": 0.0005, "loss": 1.5930209159851074, "step": 17219 }, { "epoch": 1.3575088687426093, "grad_norm": 0.4914098381996155, "learning_rate": 0.0005, "loss": 1.4433622360229492, "step": 17220 }, { "epoch": 1.3575877020102483, "grad_norm": 0.5851256847381592, "learning_rate": 0.0005, "loss": 1.6919710636138916, "step": 17221 }, { "epoch": 1.3576665352778874, "grad_norm": 0.5225026607513428, "learning_rate": 0.0005, "loss": 1.729638934135437, "step": 17222 }, { "epoch": 1.357745368545526, "grad_norm": 0.5001034736633301, "learning_rate": 0.0005, "loss": 1.5927401781082153, "step": 17223 }, { "epoch": 1.3578242018131652, "grad_norm": 0.515135645866394, "learning_rate": 0.0005, "loss": 1.6684808731079102, "step": 17224 }, { "epoch": 1.3579030350808041, "grad_norm": 0.5876696109771729, "learning_rate": 0.0005, "loss": 1.6528677940368652, "step": 17225 }, { "epoch": 1.357981868348443, "grad_norm": 0.5402278304100037, "learning_rate": 0.0005, "loss": 1.686070203781128, "step": 17226 }, { "epoch": 1.358060701616082, "grad_norm": 0.524831235408783, "learning_rate": 0.0005, "loss": 1.677875280380249, "step": 17227 }, { "epoch": 1.3581395348837209, "grad_norm": 0.5970550179481506, "learning_rate": 0.0005, "loss": 1.565124273300171, "step": 17228 }, { "epoch": 1.35821836815136, "grad_norm": 0.6735490560531616, "learning_rate": 0.0005, "loss": 1.7841308116912842, "step": 17229 }, { "epoch": 1.358297201418999, "grad_norm": 0.6973879933357239, "learning_rate": 0.0005, "loss": 1.777159333229065, "step": 17230 }, { "epoch": 1.3583760346866378, "grad_norm": 0.5624114871025085, "learning_rate": 0.0005, "loss": 1.7452449798583984, "step": 17231 }, { "epoch": 1.3584548679542767, "grad_norm": 0.541442334651947, "learning_rate": 0.0005, "loss": 1.3859009742736816, "step": 17232 }, { "epoch": 1.3585337012219156, "grad_norm": 0.5040990710258484, "learning_rate": 0.0005, "loss": 1.7308505773544312, "step": 17233 }, { "epoch": 1.3586125344895545, "grad_norm": 0.5754929780960083, "learning_rate": 0.0005, "loss": 1.7917420864105225, "step": 17234 }, { "epoch": 1.3586913677571935, "grad_norm": 0.4726966917514801, "learning_rate": 0.0005, "loss": 1.5124716758728027, "step": 17235 }, { "epoch": 1.3587702010248326, "grad_norm": 0.5373145341873169, "learning_rate": 0.0005, "loss": 1.6752099990844727, "step": 17236 }, { "epoch": 1.3588490342924715, "grad_norm": 0.49021294713020325, "learning_rate": 0.0005, "loss": 1.5788180828094482, "step": 17237 }, { "epoch": 1.3589278675601104, "grad_norm": 0.6465246081352234, "learning_rate": 0.0005, "loss": 1.7592064142227173, "step": 17238 }, { "epoch": 1.3590067008277493, "grad_norm": 0.5337891578674316, "learning_rate": 0.0005, "loss": 1.745560884475708, "step": 17239 }, { "epoch": 1.3590855340953882, "grad_norm": 0.5254685878753662, "learning_rate": 0.0005, "loss": 1.5339128971099854, "step": 17240 }, { "epoch": 1.3591643673630271, "grad_norm": 0.5877822041511536, "learning_rate": 0.0005, "loss": 1.782679796218872, "step": 17241 }, { "epoch": 1.359243200630666, "grad_norm": 0.519334614276886, "learning_rate": 0.0005, "loss": 1.4427814483642578, "step": 17242 }, { "epoch": 1.3593220338983052, "grad_norm": 0.45315414667129517, "learning_rate": 0.0005, "loss": 1.583279013633728, "step": 17243 }, { "epoch": 1.359400867165944, "grad_norm": 0.505949854850769, "learning_rate": 0.0005, "loss": 1.6436285972595215, "step": 17244 }, { "epoch": 1.359479700433583, "grad_norm": 0.4702330231666565, "learning_rate": 0.0005, "loss": 1.5237236022949219, "step": 17245 }, { "epoch": 1.359558533701222, "grad_norm": 0.5035107731819153, "learning_rate": 0.0005, "loss": 1.6709251403808594, "step": 17246 }, { "epoch": 1.3596373669688608, "grad_norm": 0.634102463722229, "learning_rate": 0.0005, "loss": 1.7235009670257568, "step": 17247 }, { "epoch": 1.3597162002364998, "grad_norm": 0.5234236121177673, "learning_rate": 0.0005, "loss": 1.7140711545944214, "step": 17248 }, { "epoch": 1.3597950335041387, "grad_norm": 0.6108877062797546, "learning_rate": 0.0005, "loss": 1.605180025100708, "step": 17249 }, { "epoch": 1.3598738667717778, "grad_norm": 0.41720089316368103, "learning_rate": 0.0005, "loss": 1.6725428104400635, "step": 17250 }, { "epoch": 1.3599527000394167, "grad_norm": 0.608498215675354, "learning_rate": 0.0005, "loss": 1.7064135074615479, "step": 17251 }, { "epoch": 1.3600315333070556, "grad_norm": 0.5034244656562805, "learning_rate": 0.0005, "loss": 1.667833924293518, "step": 17252 }, { "epoch": 1.3601103665746945, "grad_norm": 0.49218887090682983, "learning_rate": 0.0005, "loss": 1.762136697769165, "step": 17253 }, { "epoch": 1.3601891998423334, "grad_norm": 0.37130677700042725, "learning_rate": 0.0005, "loss": 1.3904671669006348, "step": 17254 }, { "epoch": 1.3602680331099724, "grad_norm": 0.5047028660774231, "learning_rate": 0.0005, "loss": 1.6745305061340332, "step": 17255 }, { "epoch": 1.3603468663776113, "grad_norm": 0.5170533657073975, "learning_rate": 0.0005, "loss": 1.8696136474609375, "step": 17256 }, { "epoch": 1.3604256996452504, "grad_norm": 0.48192381858825684, "learning_rate": 0.0005, "loss": 1.544228196144104, "step": 17257 }, { "epoch": 1.3605045329128893, "grad_norm": 0.48426318168640137, "learning_rate": 0.0005, "loss": 1.6078505516052246, "step": 17258 }, { "epoch": 1.3605833661805282, "grad_norm": 0.5573115944862366, "learning_rate": 0.0005, "loss": 1.530025601387024, "step": 17259 }, { "epoch": 1.3606621994481671, "grad_norm": 0.604529082775116, "learning_rate": 0.0005, "loss": 1.7133762836456299, "step": 17260 }, { "epoch": 1.360741032715806, "grad_norm": 0.5122144818305969, "learning_rate": 0.0005, "loss": 1.7121288776397705, "step": 17261 }, { "epoch": 1.360819865983445, "grad_norm": 0.5519498586654663, "learning_rate": 0.0005, "loss": 1.5350476503372192, "step": 17262 }, { "epoch": 1.3608986992510839, "grad_norm": 0.5049575567245483, "learning_rate": 0.0005, "loss": 1.4954521656036377, "step": 17263 }, { "epoch": 1.360977532518723, "grad_norm": 0.5660420656204224, "learning_rate": 0.0005, "loss": 1.8833794593811035, "step": 17264 }, { "epoch": 1.361056365786362, "grad_norm": 0.5217080116271973, "learning_rate": 0.0005, "loss": 1.6111552715301514, "step": 17265 }, { "epoch": 1.3611351990540008, "grad_norm": 0.4407006502151489, "learning_rate": 0.0005, "loss": 1.8013222217559814, "step": 17266 }, { "epoch": 1.3612140323216397, "grad_norm": 0.4684563875198364, "learning_rate": 0.0005, "loss": 1.5282375812530518, "step": 17267 }, { "epoch": 1.3612928655892786, "grad_norm": 0.5573767423629761, "learning_rate": 0.0005, "loss": 1.7059487104415894, "step": 17268 }, { "epoch": 1.3613716988569176, "grad_norm": 0.5241824984550476, "learning_rate": 0.0005, "loss": 1.6754860877990723, "step": 17269 }, { "epoch": 1.3614505321245565, "grad_norm": 0.5105058550834656, "learning_rate": 0.0005, "loss": 1.7957279682159424, "step": 17270 }, { "epoch": 1.3615293653921956, "grad_norm": 0.5474650263786316, "learning_rate": 0.0005, "loss": 1.4699790477752686, "step": 17271 }, { "epoch": 1.3616081986598345, "grad_norm": 0.372599720954895, "learning_rate": 0.0005, "loss": 1.3671274185180664, "step": 17272 }, { "epoch": 1.3616870319274734, "grad_norm": 0.44618064165115356, "learning_rate": 0.0005, "loss": 1.525580644607544, "step": 17273 }, { "epoch": 1.3617658651951123, "grad_norm": 0.7474930882453918, "learning_rate": 0.0005, "loss": 1.725264310836792, "step": 17274 }, { "epoch": 1.3618446984627512, "grad_norm": 0.5127471685409546, "learning_rate": 0.0005, "loss": 1.5933517217636108, "step": 17275 }, { "epoch": 1.3619235317303902, "grad_norm": 0.440309077501297, "learning_rate": 0.0005, "loss": 1.511641502380371, "step": 17276 }, { "epoch": 1.362002364998029, "grad_norm": 0.4252738058567047, "learning_rate": 0.0005, "loss": 1.551337480545044, "step": 17277 }, { "epoch": 1.3620811982656682, "grad_norm": 0.5709101557731628, "learning_rate": 0.0005, "loss": 1.8041672706604004, "step": 17278 }, { "epoch": 1.3621600315333071, "grad_norm": 0.5159763097763062, "learning_rate": 0.0005, "loss": 1.6313354969024658, "step": 17279 }, { "epoch": 1.362238864800946, "grad_norm": 0.6035664081573486, "learning_rate": 0.0005, "loss": 1.7547180652618408, "step": 17280 }, { "epoch": 1.362317698068585, "grad_norm": 0.5131642818450928, "learning_rate": 0.0005, "loss": 1.6928585767745972, "step": 17281 }, { "epoch": 1.3623965313362238, "grad_norm": 0.5681644082069397, "learning_rate": 0.0005, "loss": 1.8462886810302734, "step": 17282 }, { "epoch": 1.3624753646038628, "grad_norm": 0.5998347401618958, "learning_rate": 0.0005, "loss": 1.5588685274124146, "step": 17283 }, { "epoch": 1.3625541978715017, "grad_norm": 0.5431112051010132, "learning_rate": 0.0005, "loss": 1.4357436895370483, "step": 17284 }, { "epoch": 1.3626330311391408, "grad_norm": 0.672191858291626, "learning_rate": 0.0005, "loss": 1.7319390773773193, "step": 17285 }, { "epoch": 1.3627118644067797, "grad_norm": 0.4757251441478729, "learning_rate": 0.0005, "loss": 1.5766925811767578, "step": 17286 }, { "epoch": 1.3627906976744186, "grad_norm": 0.6220654249191284, "learning_rate": 0.0005, "loss": 1.601241111755371, "step": 17287 }, { "epoch": 1.3628695309420575, "grad_norm": 0.43254467844963074, "learning_rate": 0.0005, "loss": 1.749943494796753, "step": 17288 }, { "epoch": 1.3629483642096965, "grad_norm": 0.4647444784641266, "learning_rate": 0.0005, "loss": 1.3993079662322998, "step": 17289 }, { "epoch": 1.3630271974773354, "grad_norm": 0.7147379517555237, "learning_rate": 0.0005, "loss": 1.9275940656661987, "step": 17290 }, { "epoch": 1.3631060307449743, "grad_norm": 0.5686982274055481, "learning_rate": 0.0005, "loss": 1.913888692855835, "step": 17291 }, { "epoch": 1.3631848640126134, "grad_norm": 0.5118114948272705, "learning_rate": 0.0005, "loss": 1.7773696184158325, "step": 17292 }, { "epoch": 1.3632636972802523, "grad_norm": 0.4238831102848053, "learning_rate": 0.0005, "loss": 1.734074592590332, "step": 17293 }, { "epoch": 1.3633425305478912, "grad_norm": 0.48870211839675903, "learning_rate": 0.0005, "loss": 1.6503700017929077, "step": 17294 }, { "epoch": 1.3634213638155301, "grad_norm": 0.40445399284362793, "learning_rate": 0.0005, "loss": 1.4883761405944824, "step": 17295 }, { "epoch": 1.363500197083169, "grad_norm": 0.5391893982887268, "learning_rate": 0.0005, "loss": 1.6567683219909668, "step": 17296 }, { "epoch": 1.363579030350808, "grad_norm": 0.5784791707992554, "learning_rate": 0.0005, "loss": 1.714264154434204, "step": 17297 }, { "epoch": 1.3636578636184469, "grad_norm": 0.5130650997161865, "learning_rate": 0.0005, "loss": 1.5587793588638306, "step": 17298 }, { "epoch": 1.363736696886086, "grad_norm": 0.5016538500785828, "learning_rate": 0.0005, "loss": 1.5291293859481812, "step": 17299 }, { "epoch": 1.363815530153725, "grad_norm": 0.6125237345695496, "learning_rate": 0.0005, "loss": 1.7992541790008545, "step": 17300 }, { "epoch": 1.3638943634213638, "grad_norm": 0.5168865323066711, "learning_rate": 0.0005, "loss": 1.6259465217590332, "step": 17301 }, { "epoch": 1.3639731966890027, "grad_norm": 0.5662439465522766, "learning_rate": 0.0005, "loss": 1.579263687133789, "step": 17302 }, { "epoch": 1.3640520299566417, "grad_norm": 0.532974123954773, "learning_rate": 0.0005, "loss": 1.5589940547943115, "step": 17303 }, { "epoch": 1.3641308632242806, "grad_norm": 0.5125232934951782, "learning_rate": 0.0005, "loss": 1.4944288730621338, "step": 17304 }, { "epoch": 1.3642096964919195, "grad_norm": 0.5091344118118286, "learning_rate": 0.0005, "loss": 1.714896559715271, "step": 17305 }, { "epoch": 1.3642885297595586, "grad_norm": 0.5044904351234436, "learning_rate": 0.0005, "loss": 1.6274690628051758, "step": 17306 }, { "epoch": 1.3643673630271975, "grad_norm": 0.4978024661540985, "learning_rate": 0.0005, "loss": 1.7948176860809326, "step": 17307 }, { "epoch": 1.3644461962948364, "grad_norm": 0.6117041110992432, "learning_rate": 0.0005, "loss": 1.8803783655166626, "step": 17308 }, { "epoch": 1.3645250295624753, "grad_norm": 0.5770624279975891, "learning_rate": 0.0005, "loss": 1.7084026336669922, "step": 17309 }, { "epoch": 1.3646038628301143, "grad_norm": 0.6307606101036072, "learning_rate": 0.0005, "loss": 1.6901530027389526, "step": 17310 }, { "epoch": 1.3646826960977534, "grad_norm": 0.7263793349266052, "learning_rate": 0.0005, "loss": 1.5400909185409546, "step": 17311 }, { "epoch": 1.364761529365392, "grad_norm": 0.4213646948337555, "learning_rate": 0.0005, "loss": 1.5723717212677002, "step": 17312 }, { "epoch": 1.3648403626330312, "grad_norm": 0.5439039468765259, "learning_rate": 0.0005, "loss": 1.7046265602111816, "step": 17313 }, { "epoch": 1.3649191959006701, "grad_norm": 0.5131985545158386, "learning_rate": 0.0005, "loss": 1.7109849452972412, "step": 17314 }, { "epoch": 1.364998029168309, "grad_norm": 0.6213986873626709, "learning_rate": 0.0005, "loss": 1.6872361898422241, "step": 17315 }, { "epoch": 1.365076862435948, "grad_norm": 0.6720889210700989, "learning_rate": 0.0005, "loss": 1.4730141162872314, "step": 17316 }, { "epoch": 1.3651556957035869, "grad_norm": 0.6603830456733704, "learning_rate": 0.0005, "loss": 1.5758509635925293, "step": 17317 }, { "epoch": 1.365234528971226, "grad_norm": 0.49689981341362, "learning_rate": 0.0005, "loss": 1.6130475997924805, "step": 17318 }, { "epoch": 1.3653133622388647, "grad_norm": 0.5675230026245117, "learning_rate": 0.0005, "loss": 1.685211420059204, "step": 17319 }, { "epoch": 1.3653921955065038, "grad_norm": 0.6973572373390198, "learning_rate": 0.0005, "loss": 1.8036136627197266, "step": 17320 }, { "epoch": 1.3654710287741427, "grad_norm": 1.5273290872573853, "learning_rate": 0.0005, "loss": 2.0074715614318848, "step": 17321 }, { "epoch": 1.3655498620417816, "grad_norm": 0.6147873997688293, "learning_rate": 0.0005, "loss": 2.000474691390991, "step": 17322 }, { "epoch": 1.3656286953094205, "grad_norm": 0.5882684588432312, "learning_rate": 0.0005, "loss": 1.6836769580841064, "step": 17323 }, { "epoch": 1.3657075285770595, "grad_norm": 0.4774385094642639, "learning_rate": 0.0005, "loss": 1.6517945528030396, "step": 17324 }, { "epoch": 1.3657863618446986, "grad_norm": 0.9453412294387817, "learning_rate": 0.0005, "loss": 1.8227074146270752, "step": 17325 }, { "epoch": 1.3658651951123373, "grad_norm": 0.5374214053153992, "learning_rate": 0.0005, "loss": 1.8594746589660645, "step": 17326 }, { "epoch": 1.3659440283799764, "grad_norm": 0.5715250968933105, "learning_rate": 0.0005, "loss": 1.7901132106781006, "step": 17327 }, { "epoch": 1.3660228616476153, "grad_norm": 0.5210283398628235, "learning_rate": 0.0005, "loss": 1.404152750968933, "step": 17328 }, { "epoch": 1.3661016949152542, "grad_norm": 0.8351995348930359, "learning_rate": 0.0005, "loss": 1.6350529193878174, "step": 17329 }, { "epoch": 1.3661805281828932, "grad_norm": 0.6588473916053772, "learning_rate": 0.0005, "loss": 1.5944952964782715, "step": 17330 }, { "epoch": 1.366259361450532, "grad_norm": 0.5824810862541199, "learning_rate": 0.0005, "loss": 1.6514275074005127, "step": 17331 }, { "epoch": 1.3663381947181712, "grad_norm": 0.7102769017219543, "learning_rate": 0.0005, "loss": 1.6655336618423462, "step": 17332 }, { "epoch": 1.3664170279858099, "grad_norm": 0.5523498058319092, "learning_rate": 0.0005, "loss": 1.7087749242782593, "step": 17333 }, { "epoch": 1.366495861253449, "grad_norm": 0.9803417325019836, "learning_rate": 0.0005, "loss": 1.9031195640563965, "step": 17334 }, { "epoch": 1.366574694521088, "grad_norm": 0.5749778747558594, "learning_rate": 0.0005, "loss": 1.5555810928344727, "step": 17335 }, { "epoch": 1.3666535277887268, "grad_norm": 0.6947079300880432, "learning_rate": 0.0005, "loss": 1.8782256841659546, "step": 17336 }, { "epoch": 1.3667323610563658, "grad_norm": 0.6590032577514648, "learning_rate": 0.0005, "loss": 1.762688159942627, "step": 17337 }, { "epoch": 1.3668111943240047, "grad_norm": 0.7034677863121033, "learning_rate": 0.0005, "loss": 1.8523489236831665, "step": 17338 }, { "epoch": 1.3668900275916438, "grad_norm": 0.5734550356864929, "learning_rate": 0.0005, "loss": 1.8683483600616455, "step": 17339 }, { "epoch": 1.3669688608592825, "grad_norm": 0.6441373825073242, "learning_rate": 0.0005, "loss": 1.6640046834945679, "step": 17340 }, { "epoch": 1.3670476941269216, "grad_norm": 0.6723586916923523, "learning_rate": 0.0005, "loss": 1.5410969257354736, "step": 17341 }, { "epoch": 1.3671265273945605, "grad_norm": 0.6412404775619507, "learning_rate": 0.0005, "loss": 1.8183552026748657, "step": 17342 }, { "epoch": 1.3672053606621994, "grad_norm": 0.5392540097236633, "learning_rate": 0.0005, "loss": 1.6333192586898804, "step": 17343 }, { "epoch": 1.3672841939298384, "grad_norm": 0.75698322057724, "learning_rate": 0.0005, "loss": 1.7564258575439453, "step": 17344 }, { "epoch": 1.3673630271974773, "grad_norm": 0.5678043365478516, "learning_rate": 0.0005, "loss": 1.6509381532669067, "step": 17345 }, { "epoch": 1.3674418604651164, "grad_norm": 0.7661992311477661, "learning_rate": 0.0005, "loss": 1.78767991065979, "step": 17346 }, { "epoch": 1.367520693732755, "grad_norm": 0.5086861252784729, "learning_rate": 0.0005, "loss": 1.5026326179504395, "step": 17347 }, { "epoch": 1.3675995270003942, "grad_norm": 0.6250180006027222, "learning_rate": 0.0005, "loss": 1.7434208393096924, "step": 17348 }, { "epoch": 1.3676783602680331, "grad_norm": 0.667822003364563, "learning_rate": 0.0005, "loss": 1.6164264678955078, "step": 17349 }, { "epoch": 1.367757193535672, "grad_norm": 0.6013590693473816, "learning_rate": 0.0005, "loss": 1.8153095245361328, "step": 17350 }, { "epoch": 1.367836026803311, "grad_norm": 0.48850733041763306, "learning_rate": 0.0005, "loss": 1.6018435955047607, "step": 17351 }, { "epoch": 1.3679148600709499, "grad_norm": 0.5738555192947388, "learning_rate": 0.0005, "loss": 1.8830795288085938, "step": 17352 }, { "epoch": 1.367993693338589, "grad_norm": 0.4843405783176422, "learning_rate": 0.0005, "loss": 1.8044781684875488, "step": 17353 }, { "epoch": 1.3680725266062277, "grad_norm": 0.6540858149528503, "learning_rate": 0.0005, "loss": 1.6801745891571045, "step": 17354 }, { "epoch": 1.3681513598738668, "grad_norm": 0.6682325601577759, "learning_rate": 0.0005, "loss": 1.7700797319412231, "step": 17355 }, { "epoch": 1.3682301931415057, "grad_norm": 0.5114214420318604, "learning_rate": 0.0005, "loss": 1.708786129951477, "step": 17356 }, { "epoch": 1.3683090264091446, "grad_norm": 0.5849758386611938, "learning_rate": 0.0005, "loss": 1.541905403137207, "step": 17357 }, { "epoch": 1.3683878596767836, "grad_norm": 0.6676995754241943, "learning_rate": 0.0005, "loss": 1.774275541305542, "step": 17358 }, { "epoch": 1.3684666929444225, "grad_norm": 0.6330255270004272, "learning_rate": 0.0005, "loss": 2.0278983116149902, "step": 17359 }, { "epoch": 1.3685455262120616, "grad_norm": 0.5293812155723572, "learning_rate": 0.0005, "loss": 1.7979364395141602, "step": 17360 }, { "epoch": 1.3686243594797005, "grad_norm": 0.5340026617050171, "learning_rate": 0.0005, "loss": 1.4184908866882324, "step": 17361 }, { "epoch": 1.3687031927473394, "grad_norm": 0.656493604183197, "learning_rate": 0.0005, "loss": 1.8071675300598145, "step": 17362 }, { "epoch": 1.3687820260149783, "grad_norm": 0.555426299571991, "learning_rate": 0.0005, "loss": 1.6622260808944702, "step": 17363 }, { "epoch": 1.3688608592826172, "grad_norm": 0.6178216338157654, "learning_rate": 0.0005, "loss": 1.6363248825073242, "step": 17364 }, { "epoch": 1.3689396925502562, "grad_norm": 0.5268725156784058, "learning_rate": 0.0005, "loss": 1.609370231628418, "step": 17365 }, { "epoch": 1.369018525817895, "grad_norm": 0.4800722301006317, "learning_rate": 0.0005, "loss": 1.6780825853347778, "step": 17366 }, { "epoch": 1.3690973590855342, "grad_norm": 0.5969995856285095, "learning_rate": 0.0005, "loss": 1.801511526107788, "step": 17367 }, { "epoch": 1.3691761923531731, "grad_norm": 0.6221404671669006, "learning_rate": 0.0005, "loss": 1.793346881866455, "step": 17368 }, { "epoch": 1.369255025620812, "grad_norm": 0.6203998923301697, "learning_rate": 0.0005, "loss": 1.6235802173614502, "step": 17369 }, { "epoch": 1.369333858888451, "grad_norm": 0.547067403793335, "learning_rate": 0.0005, "loss": 1.6839210987091064, "step": 17370 }, { "epoch": 1.3694126921560899, "grad_norm": 0.5512396693229675, "learning_rate": 0.0005, "loss": 1.6091663837432861, "step": 17371 }, { "epoch": 1.3694915254237288, "grad_norm": 0.5170918107032776, "learning_rate": 0.0005, "loss": 1.6400548219680786, "step": 17372 }, { "epoch": 1.3695703586913677, "grad_norm": 0.43638280034065247, "learning_rate": 0.0005, "loss": 1.8204100131988525, "step": 17373 }, { "epoch": 1.3696491919590068, "grad_norm": 0.5349046587944031, "learning_rate": 0.0005, "loss": 1.6236982345581055, "step": 17374 }, { "epoch": 1.3697280252266457, "grad_norm": 0.5620160698890686, "learning_rate": 0.0005, "loss": 1.6857293844223022, "step": 17375 }, { "epoch": 1.3698068584942846, "grad_norm": 0.5549071431159973, "learning_rate": 0.0005, "loss": 2.0065970420837402, "step": 17376 }, { "epoch": 1.3698856917619235, "grad_norm": 0.6078242063522339, "learning_rate": 0.0005, "loss": 1.596587896347046, "step": 17377 }, { "epoch": 1.3699645250295625, "grad_norm": 0.4656960964202881, "learning_rate": 0.0005, "loss": 1.628504753112793, "step": 17378 }, { "epoch": 1.3700433582972014, "grad_norm": 0.5042251348495483, "learning_rate": 0.0005, "loss": 1.7512447834014893, "step": 17379 }, { "epoch": 1.3701221915648403, "grad_norm": 0.5050324201583862, "learning_rate": 0.0005, "loss": 1.7002544403076172, "step": 17380 }, { "epoch": 1.3702010248324794, "grad_norm": 0.5478106141090393, "learning_rate": 0.0005, "loss": 1.6233524084091187, "step": 17381 }, { "epoch": 1.3702798581001183, "grad_norm": 0.46585607528686523, "learning_rate": 0.0005, "loss": 1.5794672966003418, "step": 17382 }, { "epoch": 1.3703586913677572, "grad_norm": 0.5119292140007019, "learning_rate": 0.0005, "loss": 1.6288821697235107, "step": 17383 }, { "epoch": 1.3704375246353961, "grad_norm": 0.4444394111633301, "learning_rate": 0.0005, "loss": 1.5670127868652344, "step": 17384 }, { "epoch": 1.370516357903035, "grad_norm": 0.47318676114082336, "learning_rate": 0.0005, "loss": 1.637143850326538, "step": 17385 }, { "epoch": 1.370595191170674, "grad_norm": 0.42561104893684387, "learning_rate": 0.0005, "loss": 1.4916999340057373, "step": 17386 }, { "epoch": 1.3706740244383129, "grad_norm": 0.5894281268119812, "learning_rate": 0.0005, "loss": 1.6818010807037354, "step": 17387 }, { "epoch": 1.370752857705952, "grad_norm": 0.48334914445877075, "learning_rate": 0.0005, "loss": 1.8360761404037476, "step": 17388 }, { "epoch": 1.370831690973591, "grad_norm": 0.6802514791488647, "learning_rate": 0.0005, "loss": 1.86427640914917, "step": 17389 }, { "epoch": 1.3709105242412298, "grad_norm": 0.564669668674469, "learning_rate": 0.0005, "loss": 1.7250535488128662, "step": 17390 }, { "epoch": 1.3709893575088687, "grad_norm": 0.654254674911499, "learning_rate": 0.0005, "loss": 1.7090097665786743, "step": 17391 }, { "epoch": 1.3710681907765077, "grad_norm": 0.4508095383644104, "learning_rate": 0.0005, "loss": 1.6844453811645508, "step": 17392 }, { "epoch": 1.3711470240441466, "grad_norm": 0.7150849103927612, "learning_rate": 0.0005, "loss": 2.037921190261841, "step": 17393 }, { "epoch": 1.3712258573117855, "grad_norm": 0.4680277407169342, "learning_rate": 0.0005, "loss": 1.4826633930206299, "step": 17394 }, { "epoch": 1.3713046905794246, "grad_norm": 0.7506968379020691, "learning_rate": 0.0005, "loss": 2.033823251724243, "step": 17395 }, { "epoch": 1.3713835238470635, "grad_norm": 0.6101263165473938, "learning_rate": 0.0005, "loss": 1.7632519006729126, "step": 17396 }, { "epoch": 1.3714623571147024, "grad_norm": 0.45341575145721436, "learning_rate": 0.0005, "loss": 1.5368027687072754, "step": 17397 }, { "epoch": 1.3715411903823413, "grad_norm": 0.5304655432701111, "learning_rate": 0.0005, "loss": 1.7220813035964966, "step": 17398 }, { "epoch": 1.3716200236499803, "grad_norm": 1.1405975818634033, "learning_rate": 0.0005, "loss": 1.5277373790740967, "step": 17399 }, { "epoch": 1.3716988569176192, "grad_norm": 0.4886414408683777, "learning_rate": 0.0005, "loss": 1.4139816761016846, "step": 17400 }, { "epoch": 1.371777690185258, "grad_norm": 0.5370834469795227, "learning_rate": 0.0005, "loss": 1.523166298866272, "step": 17401 }, { "epoch": 1.3718565234528972, "grad_norm": 0.5695014595985413, "learning_rate": 0.0005, "loss": 1.542421579360962, "step": 17402 }, { "epoch": 1.3719353567205361, "grad_norm": 0.5664206743240356, "learning_rate": 0.0005, "loss": 1.5808658599853516, "step": 17403 }, { "epoch": 1.372014189988175, "grad_norm": 0.5306940078735352, "learning_rate": 0.0005, "loss": 1.526432752609253, "step": 17404 }, { "epoch": 1.372093023255814, "grad_norm": 0.5952506065368652, "learning_rate": 0.0005, "loss": 1.6765122413635254, "step": 17405 }, { "epoch": 1.3721718565234529, "grad_norm": 0.5900137424468994, "learning_rate": 0.0005, "loss": 1.8675284385681152, "step": 17406 }, { "epoch": 1.3722506897910918, "grad_norm": 0.46819084882736206, "learning_rate": 0.0005, "loss": 1.4116321802139282, "step": 17407 }, { "epoch": 1.3723295230587307, "grad_norm": 0.563843309879303, "learning_rate": 0.0005, "loss": 1.713700771331787, "step": 17408 }, { "epoch": 1.3724083563263698, "grad_norm": 0.5492296814918518, "learning_rate": 0.0005, "loss": 1.5620839595794678, "step": 17409 }, { "epoch": 1.3724871895940087, "grad_norm": 0.5471511483192444, "learning_rate": 0.0005, "loss": 1.5713390111923218, "step": 17410 }, { "epoch": 1.3725660228616476, "grad_norm": 0.5004238486289978, "learning_rate": 0.0005, "loss": 1.6616432666778564, "step": 17411 }, { "epoch": 1.3726448561292866, "grad_norm": 0.5314290523529053, "learning_rate": 0.0005, "loss": 1.6957767009735107, "step": 17412 }, { "epoch": 1.3727236893969255, "grad_norm": 0.41570955514907837, "learning_rate": 0.0005, "loss": 1.4630811214447021, "step": 17413 }, { "epoch": 1.3728025226645644, "grad_norm": 0.5609035491943359, "learning_rate": 0.0005, "loss": 1.5371589660644531, "step": 17414 }, { "epoch": 1.3728813559322033, "grad_norm": 0.6791101098060608, "learning_rate": 0.0005, "loss": 1.6175475120544434, "step": 17415 }, { "epoch": 1.3729601891998424, "grad_norm": 0.5485063195228577, "learning_rate": 0.0005, "loss": 1.7478266954421997, "step": 17416 }, { "epoch": 1.3730390224674813, "grad_norm": 0.5340632796287537, "learning_rate": 0.0005, "loss": 1.6728966236114502, "step": 17417 }, { "epoch": 1.3731178557351202, "grad_norm": 0.45593225955963135, "learning_rate": 0.0005, "loss": 1.5040969848632812, "step": 17418 }, { "epoch": 1.3731966890027592, "grad_norm": 0.45627695322036743, "learning_rate": 0.0005, "loss": 1.5264232158660889, "step": 17419 }, { "epoch": 1.373275522270398, "grad_norm": 0.5092164874076843, "learning_rate": 0.0005, "loss": 1.714218258857727, "step": 17420 }, { "epoch": 1.373354355538037, "grad_norm": 0.4896455705165863, "learning_rate": 0.0005, "loss": 1.5261133909225464, "step": 17421 }, { "epoch": 1.3734331888056759, "grad_norm": 0.43288102746009827, "learning_rate": 0.0005, "loss": 1.5807969570159912, "step": 17422 }, { "epoch": 1.373512022073315, "grad_norm": 0.5377787351608276, "learning_rate": 0.0005, "loss": 1.564132571220398, "step": 17423 }, { "epoch": 1.373590855340954, "grad_norm": 0.7617811560630798, "learning_rate": 0.0005, "loss": 1.4405944347381592, "step": 17424 }, { "epoch": 1.3736696886085928, "grad_norm": 0.466364324092865, "learning_rate": 0.0005, "loss": 1.7134816646575928, "step": 17425 }, { "epoch": 1.3737485218762318, "grad_norm": 0.4415959119796753, "learning_rate": 0.0005, "loss": 1.6074013710021973, "step": 17426 }, { "epoch": 1.3738273551438707, "grad_norm": 0.4797590374946594, "learning_rate": 0.0005, "loss": 1.5198249816894531, "step": 17427 }, { "epoch": 1.3739061884115096, "grad_norm": 0.4591352939605713, "learning_rate": 0.0005, "loss": 1.5078797340393066, "step": 17428 }, { "epoch": 1.3739850216791485, "grad_norm": 0.45581015944480896, "learning_rate": 0.0005, "loss": 1.539942741394043, "step": 17429 }, { "epoch": 1.3740638549467876, "grad_norm": 0.6401448249816895, "learning_rate": 0.0005, "loss": 1.7622158527374268, "step": 17430 }, { "epoch": 1.3741426882144265, "grad_norm": 0.47363969683647156, "learning_rate": 0.0005, "loss": 1.5517476797103882, "step": 17431 }, { "epoch": 1.3742215214820654, "grad_norm": 0.5015142560005188, "learning_rate": 0.0005, "loss": 1.4750529527664185, "step": 17432 }, { "epoch": 1.3743003547497044, "grad_norm": 0.5423388481140137, "learning_rate": 0.0005, "loss": 1.7436950206756592, "step": 17433 }, { "epoch": 1.3743791880173433, "grad_norm": 0.5036119222640991, "learning_rate": 0.0005, "loss": 1.6544630527496338, "step": 17434 }, { "epoch": 1.3744580212849824, "grad_norm": 0.4661634564399719, "learning_rate": 0.0005, "loss": 1.621344804763794, "step": 17435 }, { "epoch": 1.374536854552621, "grad_norm": 0.6713943481445312, "learning_rate": 0.0005, "loss": 1.6276097297668457, "step": 17436 }, { "epoch": 1.3746156878202602, "grad_norm": 0.43719175457954407, "learning_rate": 0.0005, "loss": 1.6018589735031128, "step": 17437 }, { "epoch": 1.3746945210878991, "grad_norm": 0.7106541991233826, "learning_rate": 0.0005, "loss": 1.9477988481521606, "step": 17438 }, { "epoch": 1.374773354355538, "grad_norm": 0.4266330301761627, "learning_rate": 0.0005, "loss": 1.6713340282440186, "step": 17439 }, { "epoch": 1.374852187623177, "grad_norm": 0.5868669152259827, "learning_rate": 0.0005, "loss": 1.5986278057098389, "step": 17440 }, { "epoch": 1.3749310208908159, "grad_norm": 0.5254281163215637, "learning_rate": 0.0005, "loss": 1.5834472179412842, "step": 17441 }, { "epoch": 1.375009854158455, "grad_norm": 0.5305377244949341, "learning_rate": 0.0005, "loss": 1.7715160846710205, "step": 17442 }, { "epoch": 1.3750886874260937, "grad_norm": 0.7974480390548706, "learning_rate": 0.0005, "loss": 1.8116780519485474, "step": 17443 }, { "epoch": 1.3751675206937328, "grad_norm": 0.6392038464546204, "learning_rate": 0.0005, "loss": 1.7573150396347046, "step": 17444 }, { "epoch": 1.3752463539613717, "grad_norm": 0.58235102891922, "learning_rate": 0.0005, "loss": 1.7507658004760742, "step": 17445 }, { "epoch": 1.3753251872290106, "grad_norm": 0.5255270600318909, "learning_rate": 0.0005, "loss": 1.5091661214828491, "step": 17446 }, { "epoch": 1.3754040204966496, "grad_norm": 0.5578363537788391, "learning_rate": 0.0005, "loss": 1.505528450012207, "step": 17447 }, { "epoch": 1.3754828537642885, "grad_norm": 0.6959245800971985, "learning_rate": 0.0005, "loss": 1.6645557880401611, "step": 17448 }, { "epoch": 1.3755616870319276, "grad_norm": 0.5492010116577148, "learning_rate": 0.0005, "loss": 1.642179012298584, "step": 17449 }, { "epoch": 1.3756405202995663, "grad_norm": 0.5830948352813721, "learning_rate": 0.0005, "loss": 1.6225237846374512, "step": 17450 }, { "epoch": 1.3757193535672054, "grad_norm": 0.42955878376960754, "learning_rate": 0.0005, "loss": 1.6612372398376465, "step": 17451 }, { "epoch": 1.3757981868348443, "grad_norm": 0.6210378408432007, "learning_rate": 0.0005, "loss": 1.8343394994735718, "step": 17452 }, { "epoch": 1.3758770201024833, "grad_norm": 0.543485164642334, "learning_rate": 0.0005, "loss": 1.8357280492782593, "step": 17453 }, { "epoch": 1.3759558533701222, "grad_norm": 0.6030030846595764, "learning_rate": 0.0005, "loss": 1.838940143585205, "step": 17454 }, { "epoch": 1.376034686637761, "grad_norm": 0.44520890712738037, "learning_rate": 0.0005, "loss": 1.4548907279968262, "step": 17455 }, { "epoch": 1.3761135199054002, "grad_norm": 0.4925338923931122, "learning_rate": 0.0005, "loss": 1.3632385730743408, "step": 17456 }, { "epoch": 1.376192353173039, "grad_norm": 0.57874995470047, "learning_rate": 0.0005, "loss": 1.8240725994110107, "step": 17457 }, { "epoch": 1.376271186440678, "grad_norm": 0.5890082716941833, "learning_rate": 0.0005, "loss": 1.7894130945205688, "step": 17458 }, { "epoch": 1.376350019708317, "grad_norm": 0.6313393712043762, "learning_rate": 0.0005, "loss": 1.642976999282837, "step": 17459 }, { "epoch": 1.3764288529759559, "grad_norm": 0.567104697227478, "learning_rate": 0.0005, "loss": 1.6402623653411865, "step": 17460 }, { "epoch": 1.3765076862435948, "grad_norm": 0.5451782941818237, "learning_rate": 0.0005, "loss": 1.647635817527771, "step": 17461 }, { "epoch": 1.3765865195112337, "grad_norm": 0.5061021447181702, "learning_rate": 0.0005, "loss": 1.5431885719299316, "step": 17462 }, { "epoch": 1.3766653527788728, "grad_norm": 0.5840619802474976, "learning_rate": 0.0005, "loss": 1.7519803047180176, "step": 17463 }, { "epoch": 1.3767441860465115, "grad_norm": 0.5572916865348816, "learning_rate": 0.0005, "loss": 1.683495283126831, "step": 17464 }, { "epoch": 1.3768230193141506, "grad_norm": 0.5626078844070435, "learning_rate": 0.0005, "loss": 1.748042106628418, "step": 17465 }, { "epoch": 1.3769018525817895, "grad_norm": 0.5293061137199402, "learning_rate": 0.0005, "loss": 1.6458227634429932, "step": 17466 }, { "epoch": 1.3769806858494285, "grad_norm": 0.5533518195152283, "learning_rate": 0.0005, "loss": 1.6928064823150635, "step": 17467 }, { "epoch": 1.3770595191170674, "grad_norm": 0.45283088088035583, "learning_rate": 0.0005, "loss": 1.7102408409118652, "step": 17468 }, { "epoch": 1.3771383523847063, "grad_norm": 0.5909164547920227, "learning_rate": 0.0005, "loss": 1.6994364261627197, "step": 17469 }, { "epoch": 1.3772171856523454, "grad_norm": 0.49982643127441406, "learning_rate": 0.0005, "loss": 1.6635637283325195, "step": 17470 }, { "epoch": 1.377296018919984, "grad_norm": 0.5439406037330627, "learning_rate": 0.0005, "loss": 1.7898223400115967, "step": 17471 }, { "epoch": 1.3773748521876232, "grad_norm": 0.4533155560493469, "learning_rate": 0.0005, "loss": 1.5209916830062866, "step": 17472 }, { "epoch": 1.3774536854552621, "grad_norm": 0.47990551590919495, "learning_rate": 0.0005, "loss": 1.6488349437713623, "step": 17473 }, { "epoch": 1.377532518722901, "grad_norm": 0.4928818643093109, "learning_rate": 0.0005, "loss": 1.724998950958252, "step": 17474 }, { "epoch": 1.37761135199054, "grad_norm": 0.6022133231163025, "learning_rate": 0.0005, "loss": 1.935835838317871, "step": 17475 }, { "epoch": 1.3776901852581789, "grad_norm": 0.5422400236129761, "learning_rate": 0.0005, "loss": 1.7989453077316284, "step": 17476 }, { "epoch": 1.377769018525818, "grad_norm": 0.5539448857307434, "learning_rate": 0.0005, "loss": 1.6467080116271973, "step": 17477 }, { "epoch": 1.3778478517934567, "grad_norm": 0.545149028301239, "learning_rate": 0.0005, "loss": 1.609246015548706, "step": 17478 }, { "epoch": 1.3779266850610958, "grad_norm": 0.5356553196907043, "learning_rate": 0.0005, "loss": 1.5349516868591309, "step": 17479 }, { "epoch": 1.3780055183287347, "grad_norm": 0.8765607476234436, "learning_rate": 0.0005, "loss": 1.8934202194213867, "step": 17480 }, { "epoch": 1.3780843515963737, "grad_norm": 0.6898635029792786, "learning_rate": 0.0005, "loss": 1.625579833984375, "step": 17481 }, { "epoch": 1.3781631848640126, "grad_norm": 0.49556612968444824, "learning_rate": 0.0005, "loss": 1.6526639461517334, "step": 17482 }, { "epoch": 1.3782420181316515, "grad_norm": 0.4311843514442444, "learning_rate": 0.0005, "loss": 1.3187167644500732, "step": 17483 }, { "epoch": 1.3783208513992906, "grad_norm": 0.5290573835372925, "learning_rate": 0.0005, "loss": 1.6817620992660522, "step": 17484 }, { "epoch": 1.3783996846669295, "grad_norm": 0.5582433342933655, "learning_rate": 0.0005, "loss": 1.5480101108551025, "step": 17485 }, { "epoch": 1.3784785179345684, "grad_norm": 0.4964052140712738, "learning_rate": 0.0005, "loss": 1.4916841983795166, "step": 17486 }, { "epoch": 1.3785573512022073, "grad_norm": 0.5409356355667114, "learning_rate": 0.0005, "loss": 1.787112832069397, "step": 17487 }, { "epoch": 1.3786361844698463, "grad_norm": 0.5913007259368896, "learning_rate": 0.0005, "loss": 1.696838140487671, "step": 17488 }, { "epoch": 1.3787150177374852, "grad_norm": 0.670464277267456, "learning_rate": 0.0005, "loss": 1.568371057510376, "step": 17489 }, { "epoch": 1.378793851005124, "grad_norm": 0.46094271540641785, "learning_rate": 0.0005, "loss": 1.6806217432022095, "step": 17490 }, { "epoch": 1.3788726842727632, "grad_norm": 0.5147321224212646, "learning_rate": 0.0005, "loss": 1.6638246774673462, "step": 17491 }, { "epoch": 1.3789515175404021, "grad_norm": 0.6189643144607544, "learning_rate": 0.0005, "loss": 1.9024646282196045, "step": 17492 }, { "epoch": 1.379030350808041, "grad_norm": 0.48527297377586365, "learning_rate": 0.0005, "loss": 1.3293676376342773, "step": 17493 }, { "epoch": 1.37910918407568, "grad_norm": 0.46936577558517456, "learning_rate": 0.0005, "loss": 1.6518447399139404, "step": 17494 }, { "epoch": 1.3791880173433189, "grad_norm": 0.6325878500938416, "learning_rate": 0.0005, "loss": 1.8025250434875488, "step": 17495 }, { "epoch": 1.3792668506109578, "grad_norm": 0.6499530076980591, "learning_rate": 0.0005, "loss": 1.6754744052886963, "step": 17496 }, { "epoch": 1.3793456838785967, "grad_norm": 0.42145153880119324, "learning_rate": 0.0005, "loss": 1.543175458908081, "step": 17497 }, { "epoch": 1.3794245171462358, "grad_norm": 0.6094610691070557, "learning_rate": 0.0005, "loss": 1.8025519847869873, "step": 17498 }, { "epoch": 1.3795033504138747, "grad_norm": 0.5110717415809631, "learning_rate": 0.0005, "loss": 1.6137096881866455, "step": 17499 }, { "epoch": 1.3795821836815136, "grad_norm": 0.5126603841781616, "learning_rate": 0.0005, "loss": 1.6533372402191162, "step": 17500 }, { "epoch": 1.3796610169491526, "grad_norm": 1.3141403198242188, "learning_rate": 0.0005, "loss": 1.6094187498092651, "step": 17501 }, { "epoch": 1.3797398502167915, "grad_norm": 0.4920274019241333, "learning_rate": 0.0005, "loss": 1.5784201622009277, "step": 17502 }, { "epoch": 1.3798186834844304, "grad_norm": 0.49544385075569153, "learning_rate": 0.0005, "loss": 1.6109505891799927, "step": 17503 }, { "epoch": 1.3798975167520693, "grad_norm": 0.5269982814788818, "learning_rate": 0.0005, "loss": 1.4376715421676636, "step": 17504 }, { "epoch": 1.3799763500197084, "grad_norm": 0.4949776232242584, "learning_rate": 0.0005, "loss": 1.4403382539749146, "step": 17505 }, { "epoch": 1.3800551832873473, "grad_norm": 0.5504803657531738, "learning_rate": 0.0005, "loss": 1.70155668258667, "step": 17506 }, { "epoch": 1.3801340165549862, "grad_norm": 0.6347150802612305, "learning_rate": 0.0005, "loss": 1.728074312210083, "step": 17507 }, { "epoch": 1.3802128498226252, "grad_norm": 0.48366233706474304, "learning_rate": 0.0005, "loss": 1.6046760082244873, "step": 17508 }, { "epoch": 1.380291683090264, "grad_norm": 0.5797331929206848, "learning_rate": 0.0005, "loss": 1.6707191467285156, "step": 17509 }, { "epoch": 1.380370516357903, "grad_norm": 0.43178245425224304, "learning_rate": 0.0005, "loss": 1.6742793321609497, "step": 17510 }, { "epoch": 1.3804493496255419, "grad_norm": 0.6991283297538757, "learning_rate": 0.0005, "loss": 1.9337353706359863, "step": 17511 }, { "epoch": 1.380528182893181, "grad_norm": 0.4487788677215576, "learning_rate": 0.0005, "loss": 1.4489059448242188, "step": 17512 }, { "epoch": 1.38060701616082, "grad_norm": 0.6545231938362122, "learning_rate": 0.0005, "loss": 1.713373064994812, "step": 17513 }, { "epoch": 1.3806858494284588, "grad_norm": 0.8558782935142517, "learning_rate": 0.0005, "loss": 1.917580008506775, "step": 17514 }, { "epoch": 1.3807646826960978, "grad_norm": 0.4297652244567871, "learning_rate": 0.0005, "loss": 1.380263090133667, "step": 17515 }, { "epoch": 1.3808435159637367, "grad_norm": 0.662704586982727, "learning_rate": 0.0005, "loss": 1.8035613298416138, "step": 17516 }, { "epoch": 1.3809223492313756, "grad_norm": 0.5717306137084961, "learning_rate": 0.0005, "loss": 1.5905537605285645, "step": 17517 }, { "epoch": 1.3810011824990145, "grad_norm": 0.6024381518363953, "learning_rate": 0.0005, "loss": 1.5815694332122803, "step": 17518 }, { "epoch": 1.3810800157666536, "grad_norm": 0.6241723895072937, "learning_rate": 0.0005, "loss": 1.5970439910888672, "step": 17519 }, { "epoch": 1.3811588490342925, "grad_norm": 0.5501461029052734, "learning_rate": 0.0005, "loss": 1.7177588939666748, "step": 17520 }, { "epoch": 1.3812376823019314, "grad_norm": 0.5865594744682312, "learning_rate": 0.0005, "loss": 1.3851897716522217, "step": 17521 }, { "epoch": 1.3813165155695704, "grad_norm": 0.5673017501831055, "learning_rate": 0.0005, "loss": 1.8831019401550293, "step": 17522 }, { "epoch": 1.3813953488372093, "grad_norm": 0.5462400317192078, "learning_rate": 0.0005, "loss": 1.656437635421753, "step": 17523 }, { "epoch": 1.3814741821048482, "grad_norm": 0.5582334399223328, "learning_rate": 0.0005, "loss": 1.8299744129180908, "step": 17524 }, { "epoch": 1.381553015372487, "grad_norm": 0.5322210788726807, "learning_rate": 0.0005, "loss": 1.557441234588623, "step": 17525 }, { "epoch": 1.3816318486401262, "grad_norm": 0.5832675695419312, "learning_rate": 0.0005, "loss": 1.499987006187439, "step": 17526 }, { "epoch": 1.3817106819077651, "grad_norm": 0.7325941920280457, "learning_rate": 0.0005, "loss": 1.4890358448028564, "step": 17527 }, { "epoch": 1.381789515175404, "grad_norm": 0.45986825227737427, "learning_rate": 0.0005, "loss": 1.5519206523895264, "step": 17528 }, { "epoch": 1.381868348443043, "grad_norm": 0.5953738689422607, "learning_rate": 0.0005, "loss": 1.711858868598938, "step": 17529 }, { "epoch": 1.3819471817106819, "grad_norm": 0.5768945813179016, "learning_rate": 0.0005, "loss": 1.5479919910430908, "step": 17530 }, { "epoch": 1.3820260149783208, "grad_norm": 0.5486420392990112, "learning_rate": 0.0005, "loss": 1.784715175628662, "step": 17531 }, { "epoch": 1.3821048482459597, "grad_norm": 0.5458763837814331, "learning_rate": 0.0005, "loss": 1.6640193462371826, "step": 17532 }, { "epoch": 1.3821836815135988, "grad_norm": 0.7509956359863281, "learning_rate": 0.0005, "loss": 1.7395381927490234, "step": 17533 }, { "epoch": 1.3822625147812377, "grad_norm": 0.4635433256626129, "learning_rate": 0.0005, "loss": 1.332080364227295, "step": 17534 }, { "epoch": 1.3823413480488767, "grad_norm": 0.5697617530822754, "learning_rate": 0.0005, "loss": 1.7412855625152588, "step": 17535 }, { "epoch": 1.3824201813165156, "grad_norm": 0.5578228235244751, "learning_rate": 0.0005, "loss": 1.754868745803833, "step": 17536 }, { "epoch": 1.3824990145841545, "grad_norm": 0.4834102988243103, "learning_rate": 0.0005, "loss": 1.629166603088379, "step": 17537 }, { "epoch": 1.3825778478517934, "grad_norm": 0.6156257390975952, "learning_rate": 0.0005, "loss": 1.5536398887634277, "step": 17538 }, { "epoch": 1.3826566811194323, "grad_norm": 0.5940042734146118, "learning_rate": 0.0005, "loss": 1.3967862129211426, "step": 17539 }, { "epoch": 1.3827355143870714, "grad_norm": 0.5452595949172974, "learning_rate": 0.0005, "loss": 1.5840343236923218, "step": 17540 }, { "epoch": 1.3828143476547103, "grad_norm": 0.48438066244125366, "learning_rate": 0.0005, "loss": 1.4672974348068237, "step": 17541 }, { "epoch": 1.3828931809223493, "grad_norm": 0.5857391953468323, "learning_rate": 0.0005, "loss": 1.9010565280914307, "step": 17542 }, { "epoch": 1.3829720141899882, "grad_norm": 0.4931679368019104, "learning_rate": 0.0005, "loss": 1.8296244144439697, "step": 17543 }, { "epoch": 1.383050847457627, "grad_norm": 0.4788997769355774, "learning_rate": 0.0005, "loss": 1.580112338066101, "step": 17544 }, { "epoch": 1.383129680725266, "grad_norm": 0.5867066383361816, "learning_rate": 0.0005, "loss": 1.8149666786193848, "step": 17545 }, { "epoch": 1.383208513992905, "grad_norm": 0.6835524439811707, "learning_rate": 0.0005, "loss": 1.934654712677002, "step": 17546 }, { "epoch": 1.383287347260544, "grad_norm": 0.6061577796936035, "learning_rate": 0.0005, "loss": 1.702193021774292, "step": 17547 }, { "epoch": 1.383366180528183, "grad_norm": 0.5562195777893066, "learning_rate": 0.0005, "loss": 1.6134693622589111, "step": 17548 }, { "epoch": 1.3834450137958219, "grad_norm": 0.5722169280052185, "learning_rate": 0.0005, "loss": 1.5428528785705566, "step": 17549 }, { "epoch": 1.3835238470634608, "grad_norm": 0.6132703423500061, "learning_rate": 0.0005, "loss": 1.5919907093048096, "step": 17550 }, { "epoch": 1.3836026803310997, "grad_norm": 0.6123729944229126, "learning_rate": 0.0005, "loss": 1.5646650791168213, "step": 17551 }, { "epoch": 1.3836815135987386, "grad_norm": 0.5516452193260193, "learning_rate": 0.0005, "loss": 1.4990293979644775, "step": 17552 }, { "epoch": 1.3837603468663775, "grad_norm": 0.6910133957862854, "learning_rate": 0.0005, "loss": 1.6377673149108887, "step": 17553 }, { "epoch": 1.3838391801340166, "grad_norm": 0.6813619136810303, "learning_rate": 0.0005, "loss": 1.676628589630127, "step": 17554 }, { "epoch": 1.3839180134016555, "grad_norm": 0.6516686677932739, "learning_rate": 0.0005, "loss": 1.6825861930847168, "step": 17555 }, { "epoch": 1.3839968466692945, "grad_norm": 0.5280671119689941, "learning_rate": 0.0005, "loss": 1.596034288406372, "step": 17556 }, { "epoch": 1.3840756799369334, "grad_norm": 0.6040950417518616, "learning_rate": 0.0005, "loss": 1.485388994216919, "step": 17557 }, { "epoch": 1.3841545132045723, "grad_norm": 0.5409973859786987, "learning_rate": 0.0005, "loss": 1.4546082019805908, "step": 17558 }, { "epoch": 1.3842333464722112, "grad_norm": 0.8314376473426819, "learning_rate": 0.0005, "loss": 1.745091199874878, "step": 17559 }, { "epoch": 1.38431217973985, "grad_norm": 0.5371506810188293, "learning_rate": 0.0005, "loss": 1.5422226190567017, "step": 17560 }, { "epoch": 1.3843910130074892, "grad_norm": 0.5294052362442017, "learning_rate": 0.0005, "loss": 1.6428827047348022, "step": 17561 }, { "epoch": 1.3844698462751281, "grad_norm": 0.5364479422569275, "learning_rate": 0.0005, "loss": 1.5364744663238525, "step": 17562 }, { "epoch": 1.384548679542767, "grad_norm": 0.5216532349586487, "learning_rate": 0.0005, "loss": 1.7426636219024658, "step": 17563 }, { "epoch": 1.384627512810406, "grad_norm": 0.612472653388977, "learning_rate": 0.0005, "loss": 2.0011959075927734, "step": 17564 }, { "epoch": 1.3847063460780449, "grad_norm": 0.5396302938461304, "learning_rate": 0.0005, "loss": 1.8087642192840576, "step": 17565 }, { "epoch": 1.384785179345684, "grad_norm": 0.5807135105133057, "learning_rate": 0.0005, "loss": 1.7834851741790771, "step": 17566 }, { "epoch": 1.3848640126133227, "grad_norm": 0.45643237233161926, "learning_rate": 0.0005, "loss": 1.4037712812423706, "step": 17567 }, { "epoch": 1.3849428458809618, "grad_norm": 0.5255887508392334, "learning_rate": 0.0005, "loss": 1.681849718093872, "step": 17568 }, { "epoch": 1.3850216791486007, "grad_norm": 0.45101505517959595, "learning_rate": 0.0005, "loss": 1.442105770111084, "step": 17569 }, { "epoch": 1.3851005124162397, "grad_norm": 0.5849168300628662, "learning_rate": 0.0005, "loss": 1.9546921253204346, "step": 17570 }, { "epoch": 1.3851793456838786, "grad_norm": 0.6098735928535461, "learning_rate": 0.0005, "loss": 1.6716006994247437, "step": 17571 }, { "epoch": 1.3852581789515175, "grad_norm": 0.5871578454971313, "learning_rate": 0.0005, "loss": 1.8888407945632935, "step": 17572 }, { "epoch": 1.3853370122191566, "grad_norm": 0.47880351543426514, "learning_rate": 0.0005, "loss": 1.6061795949935913, "step": 17573 }, { "epoch": 1.3854158454867953, "grad_norm": 0.673171877861023, "learning_rate": 0.0005, "loss": 1.6973631381988525, "step": 17574 }, { "epoch": 1.3854946787544344, "grad_norm": 0.5835726261138916, "learning_rate": 0.0005, "loss": 1.509143590927124, "step": 17575 }, { "epoch": 1.3855735120220734, "grad_norm": 0.5508349537849426, "learning_rate": 0.0005, "loss": 1.618952989578247, "step": 17576 }, { "epoch": 1.3856523452897123, "grad_norm": 0.6174833178520203, "learning_rate": 0.0005, "loss": 1.8587815761566162, "step": 17577 }, { "epoch": 1.3857311785573512, "grad_norm": 0.6306304931640625, "learning_rate": 0.0005, "loss": 1.8840508460998535, "step": 17578 }, { "epoch": 1.38581001182499, "grad_norm": 0.5047604441642761, "learning_rate": 0.0005, "loss": 1.7227586507797241, "step": 17579 }, { "epoch": 1.3858888450926292, "grad_norm": 0.5233816504478455, "learning_rate": 0.0005, "loss": 1.4949102401733398, "step": 17580 }, { "epoch": 1.385967678360268, "grad_norm": 0.570075511932373, "learning_rate": 0.0005, "loss": 1.6420888900756836, "step": 17581 }, { "epoch": 1.386046511627907, "grad_norm": 0.711867094039917, "learning_rate": 0.0005, "loss": 1.614354133605957, "step": 17582 }, { "epoch": 1.386125344895546, "grad_norm": 0.47903093695640564, "learning_rate": 0.0005, "loss": 1.6826941967010498, "step": 17583 }, { "epoch": 1.3862041781631849, "grad_norm": 0.7894241809844971, "learning_rate": 0.0005, "loss": 1.8604238033294678, "step": 17584 }, { "epoch": 1.3862830114308238, "grad_norm": 0.47284752130508423, "learning_rate": 0.0005, "loss": 1.5838348865509033, "step": 17585 }, { "epoch": 1.3863618446984627, "grad_norm": 0.5517527461051941, "learning_rate": 0.0005, "loss": 1.5608456134796143, "step": 17586 }, { "epoch": 1.3864406779661018, "grad_norm": 0.6235765218734741, "learning_rate": 0.0005, "loss": 1.7902247905731201, "step": 17587 }, { "epoch": 1.3865195112337405, "grad_norm": 0.5354238748550415, "learning_rate": 0.0005, "loss": 1.6967360973358154, "step": 17588 }, { "epoch": 1.3865983445013796, "grad_norm": 0.7126419544219971, "learning_rate": 0.0005, "loss": 1.9403656721115112, "step": 17589 }, { "epoch": 1.3866771777690186, "grad_norm": 0.414749413728714, "learning_rate": 0.0005, "loss": 1.3998186588287354, "step": 17590 }, { "epoch": 1.3867560110366575, "grad_norm": 0.6471443176269531, "learning_rate": 0.0005, "loss": 1.7563263177871704, "step": 17591 }, { "epoch": 1.3868348443042964, "grad_norm": 0.5367490649223328, "learning_rate": 0.0005, "loss": 1.5607333183288574, "step": 17592 }, { "epoch": 1.3869136775719353, "grad_norm": 0.4950553774833679, "learning_rate": 0.0005, "loss": 1.5251351594924927, "step": 17593 }, { "epoch": 1.3869925108395744, "grad_norm": 0.7296187281608582, "learning_rate": 0.0005, "loss": 1.7588729858398438, "step": 17594 }, { "epoch": 1.3870713441072131, "grad_norm": 0.4649164080619812, "learning_rate": 0.0005, "loss": 1.4848498106002808, "step": 17595 }, { "epoch": 1.3871501773748522, "grad_norm": 0.5250757932662964, "learning_rate": 0.0005, "loss": 1.5723490715026855, "step": 17596 }, { "epoch": 1.3872290106424912, "grad_norm": 0.4351179003715515, "learning_rate": 0.0005, "loss": 1.6803616285324097, "step": 17597 }, { "epoch": 1.38730784391013, "grad_norm": 0.47867482900619507, "learning_rate": 0.0005, "loss": 1.4220906496047974, "step": 17598 }, { "epoch": 1.387386677177769, "grad_norm": 0.6017785668373108, "learning_rate": 0.0005, "loss": 1.674107313156128, "step": 17599 }, { "epoch": 1.387465510445408, "grad_norm": 0.7746047377586365, "learning_rate": 0.0005, "loss": 1.6982994079589844, "step": 17600 }, { "epoch": 1.387544343713047, "grad_norm": 0.7630569934844971, "learning_rate": 0.0005, "loss": 1.7204127311706543, "step": 17601 }, { "epoch": 1.3876231769806857, "grad_norm": 0.6804999113082886, "learning_rate": 0.0005, "loss": 1.5729438066482544, "step": 17602 }, { "epoch": 1.3877020102483248, "grad_norm": 0.5803329944610596, "learning_rate": 0.0005, "loss": 1.6693103313446045, "step": 17603 }, { "epoch": 1.3877808435159638, "grad_norm": 0.7445664405822754, "learning_rate": 0.0005, "loss": 1.46351158618927, "step": 17604 }, { "epoch": 1.3878596767836027, "grad_norm": 0.4822577238082886, "learning_rate": 0.0005, "loss": 1.8223896026611328, "step": 17605 }, { "epoch": 1.3879385100512416, "grad_norm": 0.3836173415184021, "learning_rate": 0.0005, "loss": 1.4581761360168457, "step": 17606 }, { "epoch": 1.3880173433188805, "grad_norm": 0.7389049530029297, "learning_rate": 0.0005, "loss": 1.7940446138381958, "step": 17607 }, { "epoch": 1.3880961765865196, "grad_norm": 0.5761511325836182, "learning_rate": 0.0005, "loss": 1.5437138080596924, "step": 17608 }, { "epoch": 1.3881750098541585, "grad_norm": 0.6152516007423401, "learning_rate": 0.0005, "loss": 1.7254561185836792, "step": 17609 }, { "epoch": 1.3882538431217974, "grad_norm": 0.48500320315361023, "learning_rate": 0.0005, "loss": 1.6733001470565796, "step": 17610 }, { "epoch": 1.3883326763894364, "grad_norm": 0.6081732511520386, "learning_rate": 0.0005, "loss": 1.9272091388702393, "step": 17611 }, { "epoch": 1.3884115096570753, "grad_norm": 0.5265952944755554, "learning_rate": 0.0005, "loss": 1.7559109926223755, "step": 17612 }, { "epoch": 1.3884903429247142, "grad_norm": 0.5105114579200745, "learning_rate": 0.0005, "loss": 1.615587592124939, "step": 17613 }, { "epoch": 1.388569176192353, "grad_norm": 0.6452805399894714, "learning_rate": 0.0005, "loss": 1.8013718128204346, "step": 17614 }, { "epoch": 1.3886480094599922, "grad_norm": 0.6117101311683655, "learning_rate": 0.0005, "loss": 1.6605546474456787, "step": 17615 }, { "epoch": 1.3887268427276311, "grad_norm": 0.49737855792045593, "learning_rate": 0.0005, "loss": 1.4900469779968262, "step": 17616 }, { "epoch": 1.38880567599527, "grad_norm": 0.5519174337387085, "learning_rate": 0.0005, "loss": 1.6443896293640137, "step": 17617 }, { "epoch": 1.388884509262909, "grad_norm": 0.4413321018218994, "learning_rate": 0.0005, "loss": 1.7856695652008057, "step": 17618 }, { "epoch": 1.3889633425305479, "grad_norm": 0.4702939987182617, "learning_rate": 0.0005, "loss": 1.8710176944732666, "step": 17619 }, { "epoch": 1.3890421757981868, "grad_norm": 0.5213655233383179, "learning_rate": 0.0005, "loss": 1.4902141094207764, "step": 17620 }, { "epoch": 1.3891210090658257, "grad_norm": 0.6447195410728455, "learning_rate": 0.0005, "loss": 1.6826293468475342, "step": 17621 }, { "epoch": 1.3891998423334648, "grad_norm": 0.5896797776222229, "learning_rate": 0.0005, "loss": 1.5421054363250732, "step": 17622 }, { "epoch": 1.3892786756011037, "grad_norm": 0.5316080451011658, "learning_rate": 0.0005, "loss": 1.5201549530029297, "step": 17623 }, { "epoch": 1.3893575088687427, "grad_norm": 0.5682146549224854, "learning_rate": 0.0005, "loss": 1.7273173332214355, "step": 17624 }, { "epoch": 1.3894363421363816, "grad_norm": 0.5525058507919312, "learning_rate": 0.0005, "loss": 1.6915783882141113, "step": 17625 }, { "epoch": 1.3895151754040205, "grad_norm": 0.6766118407249451, "learning_rate": 0.0005, "loss": 1.705938458442688, "step": 17626 }, { "epoch": 1.3895940086716594, "grad_norm": 0.5387895703315735, "learning_rate": 0.0005, "loss": 1.7320938110351562, "step": 17627 }, { "epoch": 1.3896728419392983, "grad_norm": 0.5642080307006836, "learning_rate": 0.0005, "loss": 1.6782875061035156, "step": 17628 }, { "epoch": 1.3897516752069374, "grad_norm": 1.051867127418518, "learning_rate": 0.0005, "loss": 1.6019054651260376, "step": 17629 }, { "epoch": 1.3898305084745763, "grad_norm": 0.4772418141365051, "learning_rate": 0.0005, "loss": 1.793873906135559, "step": 17630 }, { "epoch": 1.3899093417422153, "grad_norm": 0.8088918328285217, "learning_rate": 0.0005, "loss": 1.7693769931793213, "step": 17631 }, { "epoch": 1.3899881750098542, "grad_norm": 1.4098577499389648, "learning_rate": 0.0005, "loss": 1.610899806022644, "step": 17632 }, { "epoch": 1.390067008277493, "grad_norm": 0.6730016469955444, "learning_rate": 0.0005, "loss": 1.5483072996139526, "step": 17633 }, { "epoch": 1.390145841545132, "grad_norm": 0.6038340330123901, "learning_rate": 0.0005, "loss": 1.6363152265548706, "step": 17634 }, { "epoch": 1.390224674812771, "grad_norm": 2.8545565605163574, "learning_rate": 0.0005, "loss": 1.5833380222320557, "step": 17635 }, { "epoch": 1.39030350808041, "grad_norm": 0.8377736806869507, "learning_rate": 0.0005, "loss": 1.808153748512268, "step": 17636 }, { "epoch": 1.390382341348049, "grad_norm": 0.5570921301841736, "learning_rate": 0.0005, "loss": 1.626481533050537, "step": 17637 }, { "epoch": 1.3904611746156879, "grad_norm": 0.6433542966842651, "learning_rate": 0.0005, "loss": 1.5875235795974731, "step": 17638 }, { "epoch": 1.3905400078833268, "grad_norm": 0.8149234056472778, "learning_rate": 0.0005, "loss": 1.801818609237671, "step": 17639 }, { "epoch": 1.3906188411509657, "grad_norm": 0.6954132318496704, "learning_rate": 0.0005, "loss": 1.5634877681732178, "step": 17640 }, { "epoch": 1.3906976744186046, "grad_norm": 0.6514596939086914, "learning_rate": 0.0005, "loss": 1.6508938074111938, "step": 17641 }, { "epoch": 1.3907765076862435, "grad_norm": 0.5318807363510132, "learning_rate": 0.0005, "loss": 1.6132676601409912, "step": 17642 }, { "epoch": 1.3908553409538826, "grad_norm": 0.5318027138710022, "learning_rate": 0.0005, "loss": 1.6008148193359375, "step": 17643 }, { "epoch": 1.3909341742215215, "grad_norm": 0.5632503628730774, "learning_rate": 0.0005, "loss": 1.6269762516021729, "step": 17644 }, { "epoch": 1.3910130074891605, "grad_norm": 0.6375603675842285, "learning_rate": 0.0005, "loss": 1.6977920532226562, "step": 17645 }, { "epoch": 1.3910918407567994, "grad_norm": 0.5401625633239746, "learning_rate": 0.0005, "loss": 1.4776298999786377, "step": 17646 }, { "epoch": 1.3911706740244383, "grad_norm": 0.7520381808280945, "learning_rate": 0.0005, "loss": 1.782745122909546, "step": 17647 }, { "epoch": 1.3912495072920772, "grad_norm": 0.5327936410903931, "learning_rate": 0.0005, "loss": 1.6885948181152344, "step": 17648 }, { "epoch": 1.391328340559716, "grad_norm": 0.4541899561882019, "learning_rate": 0.0005, "loss": 1.7022286653518677, "step": 17649 }, { "epoch": 1.3914071738273552, "grad_norm": 0.7745869755744934, "learning_rate": 0.0005, "loss": 1.613438606262207, "step": 17650 }, { "epoch": 1.3914860070949941, "grad_norm": 0.6280807852745056, "learning_rate": 0.0005, "loss": 1.9453794956207275, "step": 17651 }, { "epoch": 1.391564840362633, "grad_norm": 0.4706399738788605, "learning_rate": 0.0005, "loss": 1.6141135692596436, "step": 17652 }, { "epoch": 1.391643673630272, "grad_norm": 0.4667402505874634, "learning_rate": 0.0005, "loss": 1.5544803142547607, "step": 17653 }, { "epoch": 1.3917225068979109, "grad_norm": 0.5823675394058228, "learning_rate": 0.0005, "loss": 1.5481035709381104, "step": 17654 }, { "epoch": 1.3918013401655498, "grad_norm": 0.5416653752326965, "learning_rate": 0.0005, "loss": 1.5127671957015991, "step": 17655 }, { "epoch": 1.3918801734331887, "grad_norm": 0.5290887355804443, "learning_rate": 0.0005, "loss": 1.4135717153549194, "step": 17656 }, { "epoch": 1.3919590067008278, "grad_norm": 0.5646299123764038, "learning_rate": 0.0005, "loss": 1.8240430355072021, "step": 17657 }, { "epoch": 1.3920378399684668, "grad_norm": 0.4559284448623657, "learning_rate": 0.0005, "loss": 1.4253718852996826, "step": 17658 }, { "epoch": 1.3921166732361057, "grad_norm": 0.46883657574653625, "learning_rate": 0.0005, "loss": 1.7530066967010498, "step": 17659 }, { "epoch": 1.3921955065037446, "grad_norm": 0.7334628701210022, "learning_rate": 0.0005, "loss": 1.7683653831481934, "step": 17660 }, { "epoch": 1.3922743397713835, "grad_norm": 0.390840619802475, "learning_rate": 0.0005, "loss": 1.4568061828613281, "step": 17661 }, { "epoch": 1.3923531730390224, "grad_norm": 0.4164239168167114, "learning_rate": 0.0005, "loss": 1.5676958560943604, "step": 17662 }, { "epoch": 1.3924320063066613, "grad_norm": 0.5019234418869019, "learning_rate": 0.0005, "loss": 1.5286296606063843, "step": 17663 }, { "epoch": 1.3925108395743004, "grad_norm": 0.6326568126678467, "learning_rate": 0.0005, "loss": 2.028672218322754, "step": 17664 }, { "epoch": 1.3925896728419394, "grad_norm": 0.4820796251296997, "learning_rate": 0.0005, "loss": 1.5845799446105957, "step": 17665 }, { "epoch": 1.3926685061095783, "grad_norm": 0.7072994709014893, "learning_rate": 0.0005, "loss": 1.7033363580703735, "step": 17666 }, { "epoch": 1.3927473393772172, "grad_norm": 0.589426577091217, "learning_rate": 0.0005, "loss": 1.7395179271697998, "step": 17667 }, { "epoch": 1.392826172644856, "grad_norm": 0.508472204208374, "learning_rate": 0.0005, "loss": 1.5604047775268555, "step": 17668 }, { "epoch": 1.392905005912495, "grad_norm": 0.5520613789558411, "learning_rate": 0.0005, "loss": 1.8933489322662354, "step": 17669 }, { "epoch": 1.392983839180134, "grad_norm": 0.6379362344741821, "learning_rate": 0.0005, "loss": 1.9300611019134521, "step": 17670 }, { "epoch": 1.393062672447773, "grad_norm": 0.6676884889602661, "learning_rate": 0.0005, "loss": 1.638296127319336, "step": 17671 }, { "epoch": 1.393141505715412, "grad_norm": 0.542091429233551, "learning_rate": 0.0005, "loss": 1.7607922554016113, "step": 17672 }, { "epoch": 1.3932203389830509, "grad_norm": 0.48863548040390015, "learning_rate": 0.0005, "loss": 1.4757864475250244, "step": 17673 }, { "epoch": 1.3932991722506898, "grad_norm": 0.6353327631950378, "learning_rate": 0.0005, "loss": 1.7706080675125122, "step": 17674 }, { "epoch": 1.3933780055183287, "grad_norm": 0.8621069192886353, "learning_rate": 0.0005, "loss": 1.731710433959961, "step": 17675 }, { "epoch": 1.3934568387859676, "grad_norm": 0.6182321310043335, "learning_rate": 0.0005, "loss": 1.6615835428237915, "step": 17676 }, { "epoch": 1.3935356720536065, "grad_norm": 0.43884655833244324, "learning_rate": 0.0005, "loss": 1.619289517402649, "step": 17677 }, { "epoch": 1.3936145053212456, "grad_norm": 0.5439490079879761, "learning_rate": 0.0005, "loss": 1.5285630226135254, "step": 17678 }, { "epoch": 1.3936933385888846, "grad_norm": 0.6019987463951111, "learning_rate": 0.0005, "loss": 1.7957956790924072, "step": 17679 }, { "epoch": 1.3937721718565235, "grad_norm": 0.7032543420791626, "learning_rate": 0.0005, "loss": 1.7870430946350098, "step": 17680 }, { "epoch": 1.3938510051241624, "grad_norm": 0.47983017563819885, "learning_rate": 0.0005, "loss": 1.6374002695083618, "step": 17681 }, { "epoch": 1.3939298383918013, "grad_norm": 0.5761467814445496, "learning_rate": 0.0005, "loss": 1.910607099533081, "step": 17682 }, { "epoch": 1.3940086716594402, "grad_norm": 0.5091814398765564, "learning_rate": 0.0005, "loss": 1.4794151782989502, "step": 17683 }, { "epoch": 1.3940875049270791, "grad_norm": 0.49301525950431824, "learning_rate": 0.0005, "loss": 1.7339104413986206, "step": 17684 }, { "epoch": 1.3941663381947182, "grad_norm": 0.48710280656814575, "learning_rate": 0.0005, "loss": 1.486040472984314, "step": 17685 }, { "epoch": 1.3942451714623572, "grad_norm": 0.5135279297828674, "learning_rate": 0.0005, "loss": 1.642303705215454, "step": 17686 }, { "epoch": 1.394324004729996, "grad_norm": 0.5790889263153076, "learning_rate": 0.0005, "loss": 1.7972286939620972, "step": 17687 }, { "epoch": 1.394402837997635, "grad_norm": 0.5957538485527039, "learning_rate": 0.0005, "loss": 1.8150876760482788, "step": 17688 }, { "epoch": 1.394481671265274, "grad_norm": 0.6041916608810425, "learning_rate": 0.0005, "loss": 1.6554296016693115, "step": 17689 }, { "epoch": 1.394560504532913, "grad_norm": 0.4641612470149994, "learning_rate": 0.0005, "loss": 1.3306951522827148, "step": 17690 }, { "epoch": 1.3946393378005517, "grad_norm": 0.622562050819397, "learning_rate": 0.0005, "loss": 1.666790246963501, "step": 17691 }, { "epoch": 1.3947181710681908, "grad_norm": 0.4816757142543793, "learning_rate": 0.0005, "loss": 1.6657047271728516, "step": 17692 }, { "epoch": 1.3947970043358298, "grad_norm": 0.6008554100990295, "learning_rate": 0.0005, "loss": 1.674079179763794, "step": 17693 }, { "epoch": 1.3948758376034687, "grad_norm": 0.5676476359367371, "learning_rate": 0.0005, "loss": 1.6062569618225098, "step": 17694 }, { "epoch": 1.3949546708711076, "grad_norm": 0.5322876572608948, "learning_rate": 0.0005, "loss": 1.7559709548950195, "step": 17695 }, { "epoch": 1.3950335041387465, "grad_norm": 0.5179872512817383, "learning_rate": 0.0005, "loss": 1.7364370822906494, "step": 17696 }, { "epoch": 1.3951123374063856, "grad_norm": 0.596644937992096, "learning_rate": 0.0005, "loss": 1.729201316833496, "step": 17697 }, { "epoch": 1.3951911706740243, "grad_norm": 0.5254254341125488, "learning_rate": 0.0005, "loss": 1.6624799966812134, "step": 17698 }, { "epoch": 1.3952700039416635, "grad_norm": 0.5160804986953735, "learning_rate": 0.0005, "loss": 1.6497552394866943, "step": 17699 }, { "epoch": 1.3953488372093024, "grad_norm": 0.5607420206069946, "learning_rate": 0.0005, "loss": 1.6368372440338135, "step": 17700 }, { "epoch": 1.3954276704769413, "grad_norm": 0.6855161786079407, "learning_rate": 0.0005, "loss": 1.84486985206604, "step": 17701 }, { "epoch": 1.3955065037445802, "grad_norm": 0.6333573460578918, "learning_rate": 0.0005, "loss": 1.5997929573059082, "step": 17702 }, { "epoch": 1.395585337012219, "grad_norm": 0.6738271117210388, "learning_rate": 0.0005, "loss": 1.6868094205856323, "step": 17703 }, { "epoch": 1.3956641702798582, "grad_norm": 0.5427346229553223, "learning_rate": 0.0005, "loss": 1.5688419342041016, "step": 17704 }, { "epoch": 1.395743003547497, "grad_norm": 0.5175478458404541, "learning_rate": 0.0005, "loss": 1.6360352039337158, "step": 17705 }, { "epoch": 1.395821836815136, "grad_norm": 0.5462894439697266, "learning_rate": 0.0005, "loss": 1.8464184999465942, "step": 17706 }, { "epoch": 1.395900670082775, "grad_norm": 0.5296373963356018, "learning_rate": 0.0005, "loss": 1.6121402978897095, "step": 17707 }, { "epoch": 1.3959795033504139, "grad_norm": 0.5842381715774536, "learning_rate": 0.0005, "loss": 1.801140308380127, "step": 17708 }, { "epoch": 1.3960583366180528, "grad_norm": 0.6295526623725891, "learning_rate": 0.0005, "loss": 1.6307785511016846, "step": 17709 }, { "epoch": 1.3961371698856917, "grad_norm": 0.5358921885490417, "learning_rate": 0.0005, "loss": 1.6697319746017456, "step": 17710 }, { "epoch": 1.3962160031533308, "grad_norm": 0.5732030868530273, "learning_rate": 0.0005, "loss": 1.5616176128387451, "step": 17711 }, { "epoch": 1.3962948364209695, "grad_norm": 0.6986496448516846, "learning_rate": 0.0005, "loss": 1.9250152111053467, "step": 17712 }, { "epoch": 1.3963736696886087, "grad_norm": 0.5850552916526794, "learning_rate": 0.0005, "loss": 1.5251915454864502, "step": 17713 }, { "epoch": 1.3964525029562476, "grad_norm": 0.4439782202243805, "learning_rate": 0.0005, "loss": 1.40358567237854, "step": 17714 }, { "epoch": 1.3965313362238865, "grad_norm": 0.649588406085968, "learning_rate": 0.0005, "loss": 1.9052541255950928, "step": 17715 }, { "epoch": 1.3966101694915254, "grad_norm": 0.4699243903160095, "learning_rate": 0.0005, "loss": 1.7606300115585327, "step": 17716 }, { "epoch": 1.3966890027591643, "grad_norm": 0.42448481917381287, "learning_rate": 0.0005, "loss": 1.6427860260009766, "step": 17717 }, { "epoch": 1.3967678360268034, "grad_norm": 0.5080121159553528, "learning_rate": 0.0005, "loss": 1.68709397315979, "step": 17718 }, { "epoch": 1.3968466692944421, "grad_norm": 0.7137026190757751, "learning_rate": 0.0005, "loss": 1.8462646007537842, "step": 17719 }, { "epoch": 1.3969255025620813, "grad_norm": 0.6293359994888306, "learning_rate": 0.0005, "loss": 1.9869096279144287, "step": 17720 }, { "epoch": 1.3970043358297202, "grad_norm": 0.43592873215675354, "learning_rate": 0.0005, "loss": 1.3188713788986206, "step": 17721 }, { "epoch": 1.397083169097359, "grad_norm": 0.7281566262245178, "learning_rate": 0.0005, "loss": 1.7266273498535156, "step": 17722 }, { "epoch": 1.397162002364998, "grad_norm": 0.6257559061050415, "learning_rate": 0.0005, "loss": 1.650429368019104, "step": 17723 }, { "epoch": 1.397240835632637, "grad_norm": 0.6312050223350525, "learning_rate": 0.0005, "loss": 1.9590470790863037, "step": 17724 }, { "epoch": 1.397319668900276, "grad_norm": 0.5918472409248352, "learning_rate": 0.0005, "loss": 1.5880463123321533, "step": 17725 }, { "epoch": 1.3973985021679147, "grad_norm": 0.5568513870239258, "learning_rate": 0.0005, "loss": 1.800898551940918, "step": 17726 }, { "epoch": 1.3974773354355539, "grad_norm": 0.7404624819755554, "learning_rate": 0.0005, "loss": 1.5681169033050537, "step": 17727 }, { "epoch": 1.3975561687031928, "grad_norm": 0.4858884811401367, "learning_rate": 0.0005, "loss": 1.6888306140899658, "step": 17728 }, { "epoch": 1.3976350019708317, "grad_norm": 0.4845072329044342, "learning_rate": 0.0005, "loss": 1.6663053035736084, "step": 17729 }, { "epoch": 1.3977138352384706, "grad_norm": 0.48583221435546875, "learning_rate": 0.0005, "loss": 1.8588597774505615, "step": 17730 }, { "epoch": 1.3977926685061095, "grad_norm": 0.559478223323822, "learning_rate": 0.0005, "loss": 1.6721910238265991, "step": 17731 }, { "epoch": 1.3978715017737486, "grad_norm": 0.4485657215118408, "learning_rate": 0.0005, "loss": 1.4280585050582886, "step": 17732 }, { "epoch": 1.3979503350413875, "grad_norm": 0.5997580885887146, "learning_rate": 0.0005, "loss": 1.7528949975967407, "step": 17733 }, { "epoch": 1.3980291683090265, "grad_norm": 0.5892365574836731, "learning_rate": 0.0005, "loss": 1.7981040477752686, "step": 17734 }, { "epoch": 1.3981080015766654, "grad_norm": 0.501697301864624, "learning_rate": 0.0005, "loss": 1.5676910877227783, "step": 17735 }, { "epoch": 1.3981868348443043, "grad_norm": 0.5615264773368835, "learning_rate": 0.0005, "loss": 1.710439920425415, "step": 17736 }, { "epoch": 1.3982656681119432, "grad_norm": 0.42758190631866455, "learning_rate": 0.0005, "loss": 1.4055051803588867, "step": 17737 }, { "epoch": 1.398344501379582, "grad_norm": 0.4974707365036011, "learning_rate": 0.0005, "loss": 1.8646966218948364, "step": 17738 }, { "epoch": 1.3984233346472212, "grad_norm": 0.4259658753871918, "learning_rate": 0.0005, "loss": 1.533799409866333, "step": 17739 }, { "epoch": 1.3985021679148602, "grad_norm": 0.4347567856311798, "learning_rate": 0.0005, "loss": 1.7067639827728271, "step": 17740 }, { "epoch": 1.398581001182499, "grad_norm": 0.5729212164878845, "learning_rate": 0.0005, "loss": 1.4857068061828613, "step": 17741 }, { "epoch": 1.398659834450138, "grad_norm": 0.5787422060966492, "learning_rate": 0.0005, "loss": 1.8022804260253906, "step": 17742 }, { "epoch": 1.3987386677177769, "grad_norm": 0.4977446496486664, "learning_rate": 0.0005, "loss": 1.4348022937774658, "step": 17743 }, { "epoch": 1.3988175009854158, "grad_norm": 0.6409289836883545, "learning_rate": 0.0005, "loss": 1.7247610092163086, "step": 17744 }, { "epoch": 1.3988963342530547, "grad_norm": 0.4817656874656677, "learning_rate": 0.0005, "loss": 1.6258805990219116, "step": 17745 }, { "epoch": 1.3989751675206938, "grad_norm": 0.8498767018318176, "learning_rate": 0.0005, "loss": 1.450278878211975, "step": 17746 }, { "epoch": 1.3990540007883328, "grad_norm": 0.5805449485778809, "learning_rate": 0.0005, "loss": 1.825178861618042, "step": 17747 }, { "epoch": 1.3991328340559717, "grad_norm": 0.6872147917747498, "learning_rate": 0.0005, "loss": 2.206674098968506, "step": 17748 }, { "epoch": 1.3992116673236106, "grad_norm": 0.6223957538604736, "learning_rate": 0.0005, "loss": 1.573038101196289, "step": 17749 }, { "epoch": 1.3992905005912495, "grad_norm": 0.4294319748878479, "learning_rate": 0.0005, "loss": 1.5554661750793457, "step": 17750 }, { "epoch": 1.3993693338588884, "grad_norm": 0.49258455634117126, "learning_rate": 0.0005, "loss": 1.674957513809204, "step": 17751 }, { "epoch": 1.3994481671265273, "grad_norm": 0.4739917516708374, "learning_rate": 0.0005, "loss": 1.6396819353103638, "step": 17752 }, { "epoch": 1.3995270003941664, "grad_norm": 0.60429447889328, "learning_rate": 0.0005, "loss": 1.6024982929229736, "step": 17753 }, { "epoch": 1.3996058336618054, "grad_norm": 0.5391382575035095, "learning_rate": 0.0005, "loss": 1.735326886177063, "step": 17754 }, { "epoch": 1.3996846669294443, "grad_norm": 0.6493931412696838, "learning_rate": 0.0005, "loss": 1.6437872648239136, "step": 17755 }, { "epoch": 1.3997635001970832, "grad_norm": 0.44567206501960754, "learning_rate": 0.0005, "loss": 1.483115553855896, "step": 17756 }, { "epoch": 1.399842333464722, "grad_norm": 0.5882771015167236, "learning_rate": 0.0005, "loss": 1.566483497619629, "step": 17757 }, { "epoch": 1.399921166732361, "grad_norm": 0.7249649167060852, "learning_rate": 0.0005, "loss": 1.5724174976348877, "step": 17758 }, { "epoch": 1.4, "grad_norm": 0.515267014503479, "learning_rate": 0.0005, "loss": 1.533198356628418, "step": 17759 }, { "epoch": 1.400078833267639, "grad_norm": 0.6302537322044373, "learning_rate": 0.0005, "loss": 1.587886095046997, "step": 17760 }, { "epoch": 1.400157666535278, "grad_norm": 0.5824231505393982, "learning_rate": 0.0005, "loss": 1.5439575910568237, "step": 17761 }, { "epoch": 1.4002364998029169, "grad_norm": 0.48517823219299316, "learning_rate": 0.0005, "loss": 1.5874826908111572, "step": 17762 }, { "epoch": 1.4003153330705558, "grad_norm": 0.5643566250801086, "learning_rate": 0.0005, "loss": 1.5974737405776978, "step": 17763 }, { "epoch": 1.4003941663381947, "grad_norm": 0.5905844569206238, "learning_rate": 0.0005, "loss": 1.7909595966339111, "step": 17764 }, { "epoch": 1.4004729996058336, "grad_norm": 0.5202131271362305, "learning_rate": 0.0005, "loss": 1.6221916675567627, "step": 17765 }, { "epoch": 1.4005518328734725, "grad_norm": 0.4876650273799896, "learning_rate": 0.0005, "loss": 1.7209458351135254, "step": 17766 }, { "epoch": 1.4006306661411116, "grad_norm": 0.6496974229812622, "learning_rate": 0.0005, "loss": 1.7249186038970947, "step": 17767 }, { "epoch": 1.4007094994087506, "grad_norm": 0.5484450459480286, "learning_rate": 0.0005, "loss": 1.7911376953125, "step": 17768 }, { "epoch": 1.4007883326763895, "grad_norm": 0.6526243686676025, "learning_rate": 0.0005, "loss": 1.694242238998413, "step": 17769 }, { "epoch": 1.4008671659440284, "grad_norm": 0.6017048358917236, "learning_rate": 0.0005, "loss": 1.4486690759658813, "step": 17770 }, { "epoch": 1.4009459992116673, "grad_norm": 0.8146442770957947, "learning_rate": 0.0005, "loss": 1.652780532836914, "step": 17771 }, { "epoch": 1.4010248324793062, "grad_norm": 0.6137316823005676, "learning_rate": 0.0005, "loss": 1.677142858505249, "step": 17772 }, { "epoch": 1.4011036657469451, "grad_norm": 0.5810580253601074, "learning_rate": 0.0005, "loss": 1.7195501327514648, "step": 17773 }, { "epoch": 1.4011824990145842, "grad_norm": 0.5589307546615601, "learning_rate": 0.0005, "loss": 1.7814013957977295, "step": 17774 }, { "epoch": 1.4012613322822232, "grad_norm": 0.5519812107086182, "learning_rate": 0.0005, "loss": 1.762145757675171, "step": 17775 }, { "epoch": 1.401340165549862, "grad_norm": 0.9110981225967407, "learning_rate": 0.0005, "loss": 1.6758805513381958, "step": 17776 }, { "epoch": 1.401418998817501, "grad_norm": 0.6774035692214966, "learning_rate": 0.0005, "loss": 1.5637187957763672, "step": 17777 }, { "epoch": 1.40149783208514, "grad_norm": 0.5061923861503601, "learning_rate": 0.0005, "loss": 1.5968925952911377, "step": 17778 }, { "epoch": 1.4015766653527788, "grad_norm": 0.5463191270828247, "learning_rate": 0.0005, "loss": 1.6404521465301514, "step": 17779 }, { "epoch": 1.4016554986204177, "grad_norm": 0.8408077955245972, "learning_rate": 0.0005, "loss": 1.762610673904419, "step": 17780 }, { "epoch": 1.4017343318880569, "grad_norm": 0.5292584896087646, "learning_rate": 0.0005, "loss": 1.732090950012207, "step": 17781 }, { "epoch": 1.4018131651556958, "grad_norm": 0.5229979753494263, "learning_rate": 0.0005, "loss": 1.474757432937622, "step": 17782 }, { "epoch": 1.4018919984233347, "grad_norm": 0.621577262878418, "learning_rate": 0.0005, "loss": 1.7691668272018433, "step": 17783 }, { "epoch": 1.4019708316909736, "grad_norm": 0.5498042106628418, "learning_rate": 0.0005, "loss": 1.5377877950668335, "step": 17784 }, { "epoch": 1.4020496649586125, "grad_norm": 0.5158327221870422, "learning_rate": 0.0005, "loss": 1.4411029815673828, "step": 17785 }, { "epoch": 1.4021284982262514, "grad_norm": 0.5755889415740967, "learning_rate": 0.0005, "loss": 1.6838600635528564, "step": 17786 }, { "epoch": 1.4022073314938903, "grad_norm": 0.5595468878746033, "learning_rate": 0.0005, "loss": 1.642289400100708, "step": 17787 }, { "epoch": 1.4022861647615295, "grad_norm": 0.4834989905357361, "learning_rate": 0.0005, "loss": 1.7126036882400513, "step": 17788 }, { "epoch": 1.4023649980291684, "grad_norm": 0.6297243237495422, "learning_rate": 0.0005, "loss": 1.7137625217437744, "step": 17789 }, { "epoch": 1.4024438312968073, "grad_norm": 0.5513780117034912, "learning_rate": 0.0005, "loss": 1.5961982011795044, "step": 17790 }, { "epoch": 1.4025226645644462, "grad_norm": 0.4000791907310486, "learning_rate": 0.0005, "loss": 1.4881101846694946, "step": 17791 }, { "epoch": 1.402601497832085, "grad_norm": 0.5593380928039551, "learning_rate": 0.0005, "loss": 1.639240026473999, "step": 17792 }, { "epoch": 1.402680331099724, "grad_norm": 0.6656678318977356, "learning_rate": 0.0005, "loss": 1.6452887058258057, "step": 17793 }, { "epoch": 1.402759164367363, "grad_norm": 0.5882890224456787, "learning_rate": 0.0005, "loss": 1.7039843797683716, "step": 17794 }, { "epoch": 1.402837997635002, "grad_norm": 0.5195398330688477, "learning_rate": 0.0005, "loss": 1.5399307012557983, "step": 17795 }, { "epoch": 1.402916830902641, "grad_norm": 0.5469403862953186, "learning_rate": 0.0005, "loss": 1.784780740737915, "step": 17796 }, { "epoch": 1.4029956641702799, "grad_norm": 0.6655886173248291, "learning_rate": 0.0005, "loss": 1.5818055868148804, "step": 17797 }, { "epoch": 1.4030744974379188, "grad_norm": 0.576526939868927, "learning_rate": 0.0005, "loss": 1.9001960754394531, "step": 17798 }, { "epoch": 1.4031533307055577, "grad_norm": 1.848745346069336, "learning_rate": 0.0005, "loss": 1.7405471801757812, "step": 17799 }, { "epoch": 1.4032321639731966, "grad_norm": 0.4567711353302002, "learning_rate": 0.0005, "loss": 1.416520357131958, "step": 17800 }, { "epoch": 1.4033109972408355, "grad_norm": 0.4858984053134918, "learning_rate": 0.0005, "loss": 1.5181262493133545, "step": 17801 }, { "epoch": 1.4033898305084747, "grad_norm": 0.5507832169532776, "learning_rate": 0.0005, "loss": 1.7070932388305664, "step": 17802 }, { "epoch": 1.4034686637761136, "grad_norm": 0.4520828127861023, "learning_rate": 0.0005, "loss": 1.6379480361938477, "step": 17803 }, { "epoch": 1.4035474970437525, "grad_norm": 0.5617595911026001, "learning_rate": 0.0005, "loss": 1.8232042789459229, "step": 17804 }, { "epoch": 1.4036263303113914, "grad_norm": 0.5192264914512634, "learning_rate": 0.0005, "loss": 1.7138887643814087, "step": 17805 }, { "epoch": 1.4037051635790303, "grad_norm": 0.565160870552063, "learning_rate": 0.0005, "loss": 1.7805813550949097, "step": 17806 }, { "epoch": 1.4037839968466692, "grad_norm": 0.47024285793304443, "learning_rate": 0.0005, "loss": 1.846991777420044, "step": 17807 }, { "epoch": 1.4038628301143081, "grad_norm": 0.6001161932945251, "learning_rate": 0.0005, "loss": 1.6646313667297363, "step": 17808 }, { "epoch": 1.4039416633819473, "grad_norm": 0.525334358215332, "learning_rate": 0.0005, "loss": 1.701120376586914, "step": 17809 }, { "epoch": 1.4040204966495862, "grad_norm": 0.46620041131973267, "learning_rate": 0.0005, "loss": 1.8051910400390625, "step": 17810 }, { "epoch": 1.404099329917225, "grad_norm": 0.5579496622085571, "learning_rate": 0.0005, "loss": 1.5608552694320679, "step": 17811 }, { "epoch": 1.404178163184864, "grad_norm": 0.5352239012718201, "learning_rate": 0.0005, "loss": 1.6480716466903687, "step": 17812 }, { "epoch": 1.404256996452503, "grad_norm": 0.6893509030342102, "learning_rate": 0.0005, "loss": 1.8897829055786133, "step": 17813 }, { "epoch": 1.404335829720142, "grad_norm": 0.49806201457977295, "learning_rate": 0.0005, "loss": 1.838236689567566, "step": 17814 }, { "epoch": 1.4044146629877807, "grad_norm": 0.5855843424797058, "learning_rate": 0.0005, "loss": 1.7519720792770386, "step": 17815 }, { "epoch": 1.4044934962554199, "grad_norm": 0.5885607600212097, "learning_rate": 0.0005, "loss": 1.4603040218353271, "step": 17816 }, { "epoch": 1.4045723295230588, "grad_norm": 0.47990620136260986, "learning_rate": 0.0005, "loss": 1.5350518226623535, "step": 17817 }, { "epoch": 1.4046511627906977, "grad_norm": 0.5406903028488159, "learning_rate": 0.0005, "loss": 1.8866677284240723, "step": 17818 }, { "epoch": 1.4047299960583366, "grad_norm": 0.6660327911376953, "learning_rate": 0.0005, "loss": 1.447962760925293, "step": 17819 }, { "epoch": 1.4048088293259755, "grad_norm": 0.5182692408561707, "learning_rate": 0.0005, "loss": 1.6364285945892334, "step": 17820 }, { "epoch": 1.4048876625936146, "grad_norm": 0.5677609443664551, "learning_rate": 0.0005, "loss": 1.617231845855713, "step": 17821 }, { "epoch": 1.4049664958612533, "grad_norm": 0.6063398718833923, "learning_rate": 0.0005, "loss": 1.584855318069458, "step": 17822 }, { "epoch": 1.4050453291288925, "grad_norm": 0.6082733869552612, "learning_rate": 0.0005, "loss": 1.761582612991333, "step": 17823 }, { "epoch": 1.4051241623965314, "grad_norm": 0.5812894701957703, "learning_rate": 0.0005, "loss": 1.9536213874816895, "step": 17824 }, { "epoch": 1.4052029956641703, "grad_norm": 0.5268751382827759, "learning_rate": 0.0005, "loss": 1.6302086114883423, "step": 17825 }, { "epoch": 1.4052818289318092, "grad_norm": 0.5306284427642822, "learning_rate": 0.0005, "loss": 1.5607874393463135, "step": 17826 }, { "epoch": 1.405360662199448, "grad_norm": 0.48224493861198425, "learning_rate": 0.0005, "loss": 1.5584383010864258, "step": 17827 }, { "epoch": 1.4054394954670872, "grad_norm": 0.4584861099720001, "learning_rate": 0.0005, "loss": 1.6593554019927979, "step": 17828 }, { "epoch": 1.405518328734726, "grad_norm": 0.6101586818695068, "learning_rate": 0.0005, "loss": 1.6111727952957153, "step": 17829 }, { "epoch": 1.405597162002365, "grad_norm": 0.6137068867683411, "learning_rate": 0.0005, "loss": 1.6690566539764404, "step": 17830 }, { "epoch": 1.405675995270004, "grad_norm": 0.552422285079956, "learning_rate": 0.0005, "loss": 1.58889901638031, "step": 17831 }, { "epoch": 1.4057548285376429, "grad_norm": 0.4982948899269104, "learning_rate": 0.0005, "loss": 1.6209347248077393, "step": 17832 }, { "epoch": 1.4058336618052818, "grad_norm": 0.5279427766799927, "learning_rate": 0.0005, "loss": 1.5397619009017944, "step": 17833 }, { "epoch": 1.4059124950729207, "grad_norm": 0.6773228645324707, "learning_rate": 0.0005, "loss": 1.7954872846603394, "step": 17834 }, { "epoch": 1.4059913283405598, "grad_norm": 0.5219531655311584, "learning_rate": 0.0005, "loss": 1.5160458087921143, "step": 17835 }, { "epoch": 1.4060701616081985, "grad_norm": 0.5472835898399353, "learning_rate": 0.0005, "loss": 1.6677411794662476, "step": 17836 }, { "epoch": 1.4061489948758377, "grad_norm": 0.7159569263458252, "learning_rate": 0.0005, "loss": 1.9177448749542236, "step": 17837 }, { "epoch": 1.4062278281434766, "grad_norm": 0.6089558601379395, "learning_rate": 0.0005, "loss": 1.6561781167984009, "step": 17838 }, { "epoch": 1.4063066614111155, "grad_norm": 0.6101461052894592, "learning_rate": 0.0005, "loss": 1.6473381519317627, "step": 17839 }, { "epoch": 1.4063854946787544, "grad_norm": 0.5520686507225037, "learning_rate": 0.0005, "loss": 1.7550064325332642, "step": 17840 }, { "epoch": 1.4064643279463933, "grad_norm": 0.6543068289756775, "learning_rate": 0.0005, "loss": 1.7639127969741821, "step": 17841 }, { "epoch": 1.4065431612140324, "grad_norm": 1.0217841863632202, "learning_rate": 0.0005, "loss": 1.7036635875701904, "step": 17842 }, { "epoch": 1.4066219944816711, "grad_norm": 0.5401498079299927, "learning_rate": 0.0005, "loss": 1.4851593971252441, "step": 17843 }, { "epoch": 1.4067008277493103, "grad_norm": 0.5925602912902832, "learning_rate": 0.0005, "loss": 1.787709355354309, "step": 17844 }, { "epoch": 1.4067796610169492, "grad_norm": 0.7553813457489014, "learning_rate": 0.0005, "loss": 1.798458218574524, "step": 17845 }, { "epoch": 1.406858494284588, "grad_norm": 0.8667287230491638, "learning_rate": 0.0005, "loss": 1.7769042253494263, "step": 17846 }, { "epoch": 1.406937327552227, "grad_norm": 1.014936089515686, "learning_rate": 0.0005, "loss": 1.7326698303222656, "step": 17847 }, { "epoch": 1.407016160819866, "grad_norm": 0.4792172908782959, "learning_rate": 0.0005, "loss": 1.596195936203003, "step": 17848 }, { "epoch": 1.407094994087505, "grad_norm": 0.5033048391342163, "learning_rate": 0.0005, "loss": 1.5873159170150757, "step": 17849 }, { "epoch": 1.4071738273551437, "grad_norm": 0.5863817930221558, "learning_rate": 0.0005, "loss": 1.6239008903503418, "step": 17850 }, { "epoch": 1.4072526606227829, "grad_norm": 0.476552277803421, "learning_rate": 0.0005, "loss": 1.5212211608886719, "step": 17851 }, { "epoch": 1.4073314938904218, "grad_norm": 0.6132009625434875, "learning_rate": 0.0005, "loss": 1.4455177783966064, "step": 17852 }, { "epoch": 1.4074103271580607, "grad_norm": 0.6948667764663696, "learning_rate": 0.0005, "loss": 1.5322868824005127, "step": 17853 }, { "epoch": 1.4074891604256996, "grad_norm": 0.4866085350513458, "learning_rate": 0.0005, "loss": 1.535003900527954, "step": 17854 }, { "epoch": 1.4075679936933385, "grad_norm": 0.44714394211769104, "learning_rate": 0.0005, "loss": 1.3223657608032227, "step": 17855 }, { "epoch": 1.4076468269609776, "grad_norm": 0.4848935008049011, "learning_rate": 0.0005, "loss": 1.4262611865997314, "step": 17856 }, { "epoch": 1.4077256602286163, "grad_norm": 0.5070715546607971, "learning_rate": 0.0005, "loss": 1.7371368408203125, "step": 17857 }, { "epoch": 1.4078044934962555, "grad_norm": 0.49339044094085693, "learning_rate": 0.0005, "loss": 1.6878620386123657, "step": 17858 }, { "epoch": 1.4078833267638944, "grad_norm": 0.5401666760444641, "learning_rate": 0.0005, "loss": 1.87659752368927, "step": 17859 }, { "epoch": 1.4079621600315333, "grad_norm": 0.5907949209213257, "learning_rate": 0.0005, "loss": 1.5419623851776123, "step": 17860 }, { "epoch": 1.4080409932991722, "grad_norm": 0.59051114320755, "learning_rate": 0.0005, "loss": 1.477123737335205, "step": 17861 }, { "epoch": 1.4081198265668111, "grad_norm": 0.41955581307411194, "learning_rate": 0.0005, "loss": 1.5643916130065918, "step": 17862 }, { "epoch": 1.4081986598344503, "grad_norm": 0.6022123098373413, "learning_rate": 0.0005, "loss": 1.7019919157028198, "step": 17863 }, { "epoch": 1.4082774931020892, "grad_norm": 0.42598894238471985, "learning_rate": 0.0005, "loss": 1.4308109283447266, "step": 17864 }, { "epoch": 1.408356326369728, "grad_norm": 0.45881205797195435, "learning_rate": 0.0005, "loss": 1.467475175857544, "step": 17865 }, { "epoch": 1.408435159637367, "grad_norm": 0.6095852255821228, "learning_rate": 0.0005, "loss": 1.8843994140625, "step": 17866 }, { "epoch": 1.408513992905006, "grad_norm": 0.5284559726715088, "learning_rate": 0.0005, "loss": 1.6449072360992432, "step": 17867 }, { "epoch": 1.4085928261726448, "grad_norm": 0.7318809032440186, "learning_rate": 0.0005, "loss": 1.7640553712844849, "step": 17868 }, { "epoch": 1.4086716594402837, "grad_norm": 0.5961716175079346, "learning_rate": 0.0005, "loss": 1.8233052492141724, "step": 17869 }, { "epoch": 1.4087504927079229, "grad_norm": 0.536765456199646, "learning_rate": 0.0005, "loss": 1.709937572479248, "step": 17870 }, { "epoch": 1.4088293259755618, "grad_norm": 0.5789074897766113, "learning_rate": 0.0005, "loss": 1.7332470417022705, "step": 17871 }, { "epoch": 1.4089081592432007, "grad_norm": 0.6036719679832458, "learning_rate": 0.0005, "loss": 1.7358428239822388, "step": 17872 }, { "epoch": 1.4089869925108396, "grad_norm": 0.43518683314323425, "learning_rate": 0.0005, "loss": 1.4099719524383545, "step": 17873 }, { "epoch": 1.4090658257784785, "grad_norm": 0.48584237694740295, "learning_rate": 0.0005, "loss": 1.7529296875, "step": 17874 }, { "epoch": 1.4091446590461174, "grad_norm": 0.5052088499069214, "learning_rate": 0.0005, "loss": 1.450270652770996, "step": 17875 }, { "epoch": 1.4092234923137563, "grad_norm": 1.181352972984314, "learning_rate": 0.0005, "loss": 1.5459766387939453, "step": 17876 }, { "epoch": 1.4093023255813955, "grad_norm": 0.5834619402885437, "learning_rate": 0.0005, "loss": 1.7822158336639404, "step": 17877 }, { "epoch": 1.4093811588490344, "grad_norm": 0.5109771490097046, "learning_rate": 0.0005, "loss": 1.6983153820037842, "step": 17878 }, { "epoch": 1.4094599921166733, "grad_norm": 0.48370254039764404, "learning_rate": 0.0005, "loss": 1.5888513326644897, "step": 17879 }, { "epoch": 1.4095388253843122, "grad_norm": 0.571031391620636, "learning_rate": 0.0005, "loss": 1.6197564601898193, "step": 17880 }, { "epoch": 1.409617658651951, "grad_norm": 0.4853353500366211, "learning_rate": 0.0005, "loss": 1.5524303913116455, "step": 17881 }, { "epoch": 1.40969649191959, "grad_norm": 0.66045743227005, "learning_rate": 0.0005, "loss": 1.6615145206451416, "step": 17882 }, { "epoch": 1.409775325187229, "grad_norm": 0.6153470873832703, "learning_rate": 0.0005, "loss": 1.7840754985809326, "step": 17883 }, { "epoch": 1.409854158454868, "grad_norm": 0.5585677027702332, "learning_rate": 0.0005, "loss": 1.7335821390151978, "step": 17884 }, { "epoch": 1.409932991722507, "grad_norm": 0.5247589945793152, "learning_rate": 0.0005, "loss": 1.5750956535339355, "step": 17885 }, { "epoch": 1.4100118249901459, "grad_norm": 0.5858162641525269, "learning_rate": 0.0005, "loss": 1.7358627319335938, "step": 17886 }, { "epoch": 1.4100906582577848, "grad_norm": 0.44595080614089966, "learning_rate": 0.0005, "loss": 1.4077205657958984, "step": 17887 }, { "epoch": 1.4101694915254237, "grad_norm": 0.542314887046814, "learning_rate": 0.0005, "loss": 1.4312059879302979, "step": 17888 }, { "epoch": 1.4102483247930626, "grad_norm": 0.6151508092880249, "learning_rate": 0.0005, "loss": 1.5896621942520142, "step": 17889 }, { "epoch": 1.4103271580607015, "grad_norm": 0.5610237121582031, "learning_rate": 0.0005, "loss": 1.7844600677490234, "step": 17890 }, { "epoch": 1.4104059913283407, "grad_norm": 1.9071640968322754, "learning_rate": 0.0005, "loss": 1.6273624897003174, "step": 17891 }, { "epoch": 1.4104848245959796, "grad_norm": 0.5454379916191101, "learning_rate": 0.0005, "loss": 1.7289613485336304, "step": 17892 }, { "epoch": 1.4105636578636185, "grad_norm": 0.590336799621582, "learning_rate": 0.0005, "loss": 1.7341228723526, "step": 17893 }, { "epoch": 1.4106424911312574, "grad_norm": 0.5409152507781982, "learning_rate": 0.0005, "loss": 1.5528079271316528, "step": 17894 }, { "epoch": 1.4107213243988963, "grad_norm": 0.6051382422447205, "learning_rate": 0.0005, "loss": 1.7225043773651123, "step": 17895 }, { "epoch": 1.4108001576665352, "grad_norm": 0.583005428314209, "learning_rate": 0.0005, "loss": 1.517632246017456, "step": 17896 }, { "epoch": 1.4108789909341741, "grad_norm": 0.4800766110420227, "learning_rate": 0.0005, "loss": 1.8867530822753906, "step": 17897 }, { "epoch": 1.4109578242018133, "grad_norm": 0.49898841977119446, "learning_rate": 0.0005, "loss": 1.849722146987915, "step": 17898 }, { "epoch": 1.4110366574694522, "grad_norm": 0.5400261878967285, "learning_rate": 0.0005, "loss": 1.5758607387542725, "step": 17899 }, { "epoch": 1.411115490737091, "grad_norm": 0.690485954284668, "learning_rate": 0.0005, "loss": 1.9901679754257202, "step": 17900 }, { "epoch": 1.41119432400473, "grad_norm": 0.5748558640480042, "learning_rate": 0.0005, "loss": 1.5900856256484985, "step": 17901 }, { "epoch": 1.411273157272369, "grad_norm": 0.6083551645278931, "learning_rate": 0.0005, "loss": 1.9917774200439453, "step": 17902 }, { "epoch": 1.4113519905400078, "grad_norm": 0.5179466009140015, "learning_rate": 0.0005, "loss": 1.6228457689285278, "step": 17903 }, { "epoch": 1.4114308238076467, "grad_norm": 0.4328206777572632, "learning_rate": 0.0005, "loss": 1.3944720029830933, "step": 17904 }, { "epoch": 1.4115096570752859, "grad_norm": 0.5272871851921082, "learning_rate": 0.0005, "loss": 1.6213891506195068, "step": 17905 }, { "epoch": 1.4115884903429248, "grad_norm": 0.5915178656578064, "learning_rate": 0.0005, "loss": 1.8333531618118286, "step": 17906 }, { "epoch": 1.4116673236105637, "grad_norm": 0.5149552226066589, "learning_rate": 0.0005, "loss": 1.7133264541625977, "step": 17907 }, { "epoch": 1.4117461568782026, "grad_norm": 1.5514273643493652, "learning_rate": 0.0005, "loss": 1.6731767654418945, "step": 17908 }, { "epoch": 1.4118249901458415, "grad_norm": 0.6326767802238464, "learning_rate": 0.0005, "loss": 1.680906057357788, "step": 17909 }, { "epoch": 1.4119038234134804, "grad_norm": 0.47571876645088196, "learning_rate": 0.0005, "loss": 1.6771576404571533, "step": 17910 }, { "epoch": 1.4119826566811193, "grad_norm": 0.545567512512207, "learning_rate": 0.0005, "loss": 1.5053668022155762, "step": 17911 }, { "epoch": 1.4120614899487585, "grad_norm": 0.6379565596580505, "learning_rate": 0.0005, "loss": 1.7484514713287354, "step": 17912 }, { "epoch": 1.4121403232163974, "grad_norm": 0.4830426871776581, "learning_rate": 0.0005, "loss": 1.460460901260376, "step": 17913 }, { "epoch": 1.4122191564840363, "grad_norm": 0.6343438625335693, "learning_rate": 0.0005, "loss": 1.962947130203247, "step": 17914 }, { "epoch": 1.4122979897516752, "grad_norm": 0.5531227588653564, "learning_rate": 0.0005, "loss": 1.7984129190444946, "step": 17915 }, { "epoch": 1.412376823019314, "grad_norm": 0.6790801882743835, "learning_rate": 0.0005, "loss": 1.7677414417266846, "step": 17916 }, { "epoch": 1.412455656286953, "grad_norm": 0.4874057173728943, "learning_rate": 0.0005, "loss": 1.5899070501327515, "step": 17917 }, { "epoch": 1.412534489554592, "grad_norm": 0.6549473404884338, "learning_rate": 0.0005, "loss": 1.5511157512664795, "step": 17918 }, { "epoch": 1.412613322822231, "grad_norm": 0.5422781109809875, "learning_rate": 0.0005, "loss": 1.6706618070602417, "step": 17919 }, { "epoch": 1.41269215608987, "grad_norm": 1.221441388130188, "learning_rate": 0.0005, "loss": 1.8617058992385864, "step": 17920 }, { "epoch": 1.4127709893575089, "grad_norm": 0.6855868697166443, "learning_rate": 0.0005, "loss": 1.954875111579895, "step": 17921 }, { "epoch": 1.4128498226251478, "grad_norm": 0.7771157026290894, "learning_rate": 0.0005, "loss": 1.6670396327972412, "step": 17922 }, { "epoch": 1.4129286558927867, "grad_norm": 0.6977402567863464, "learning_rate": 0.0005, "loss": 1.7454664707183838, "step": 17923 }, { "epoch": 1.4130074891604256, "grad_norm": 0.5923154950141907, "learning_rate": 0.0005, "loss": 1.6932318210601807, "step": 17924 }, { "epoch": 1.4130863224280645, "grad_norm": 0.7103945016860962, "learning_rate": 0.0005, "loss": 1.9697306156158447, "step": 17925 }, { "epoch": 1.4131651556957037, "grad_norm": 0.5845494866371155, "learning_rate": 0.0005, "loss": 1.8258981704711914, "step": 17926 }, { "epoch": 1.4132439889633426, "grad_norm": 0.5906262993812561, "learning_rate": 0.0005, "loss": 1.7589507102966309, "step": 17927 }, { "epoch": 1.4133228222309815, "grad_norm": 0.6237615942955017, "learning_rate": 0.0005, "loss": 1.6364684104919434, "step": 17928 }, { "epoch": 1.4134016554986204, "grad_norm": 0.5639280676841736, "learning_rate": 0.0005, "loss": 1.4659968614578247, "step": 17929 }, { "epoch": 1.4134804887662593, "grad_norm": 0.863051176071167, "learning_rate": 0.0005, "loss": 1.8906347751617432, "step": 17930 }, { "epoch": 1.4135593220338982, "grad_norm": 0.5862988829612732, "learning_rate": 0.0005, "loss": 1.6014039516448975, "step": 17931 }, { "epoch": 1.4136381553015371, "grad_norm": 0.5461753606796265, "learning_rate": 0.0005, "loss": 1.6714662313461304, "step": 17932 }, { "epoch": 1.4137169885691763, "grad_norm": 0.5676496624946594, "learning_rate": 0.0005, "loss": 1.6601370573043823, "step": 17933 }, { "epoch": 1.4137958218368152, "grad_norm": 0.5767855644226074, "learning_rate": 0.0005, "loss": 1.9264309406280518, "step": 17934 }, { "epoch": 1.413874655104454, "grad_norm": 0.557385265827179, "learning_rate": 0.0005, "loss": 1.6594626903533936, "step": 17935 }, { "epoch": 1.413953488372093, "grad_norm": 0.40786823630332947, "learning_rate": 0.0005, "loss": 1.4769283533096313, "step": 17936 }, { "epoch": 1.414032321639732, "grad_norm": 0.45682063698768616, "learning_rate": 0.0005, "loss": 1.4708880186080933, "step": 17937 }, { "epoch": 1.414111154907371, "grad_norm": 0.7367778420448303, "learning_rate": 0.0005, "loss": 1.734446406364441, "step": 17938 }, { "epoch": 1.4141899881750097, "grad_norm": 0.45465773344039917, "learning_rate": 0.0005, "loss": 1.4973715543746948, "step": 17939 }, { "epoch": 1.4142688214426489, "grad_norm": 0.5716232061386108, "learning_rate": 0.0005, "loss": 1.5298421382904053, "step": 17940 }, { "epoch": 1.4143476547102878, "grad_norm": 0.5835791826248169, "learning_rate": 0.0005, "loss": 1.914473533630371, "step": 17941 }, { "epoch": 1.4144264879779267, "grad_norm": 0.4925743043422699, "learning_rate": 0.0005, "loss": 1.666237473487854, "step": 17942 }, { "epoch": 1.4145053212455656, "grad_norm": 0.5740711688995361, "learning_rate": 0.0005, "loss": 1.7096620798110962, "step": 17943 }, { "epoch": 1.4145841545132045, "grad_norm": 0.5962057709693909, "learning_rate": 0.0005, "loss": 1.8056302070617676, "step": 17944 }, { "epoch": 1.4146629877808437, "grad_norm": 0.6571897268295288, "learning_rate": 0.0005, "loss": 1.643202304840088, "step": 17945 }, { "epoch": 1.4147418210484823, "grad_norm": 0.5935253500938416, "learning_rate": 0.0005, "loss": 1.6052955389022827, "step": 17946 }, { "epoch": 1.4148206543161215, "grad_norm": 0.4992504417896271, "learning_rate": 0.0005, "loss": 1.5257906913757324, "step": 17947 }, { "epoch": 1.4148994875837604, "grad_norm": 0.5935795903205872, "learning_rate": 0.0005, "loss": 1.7766480445861816, "step": 17948 }, { "epoch": 1.4149783208513993, "grad_norm": 0.4556349813938141, "learning_rate": 0.0005, "loss": 1.5109236240386963, "step": 17949 }, { "epoch": 1.4150571541190382, "grad_norm": 0.48983636498451233, "learning_rate": 0.0005, "loss": 1.597562313079834, "step": 17950 }, { "epoch": 1.4151359873866771, "grad_norm": 0.5471872091293335, "learning_rate": 0.0005, "loss": 1.559545874595642, "step": 17951 }, { "epoch": 1.4152148206543163, "grad_norm": 0.5639530420303345, "learning_rate": 0.0005, "loss": 1.811074137687683, "step": 17952 }, { "epoch": 1.415293653921955, "grad_norm": 1.1427233219146729, "learning_rate": 0.0005, "loss": 1.7631146907806396, "step": 17953 }, { "epoch": 1.415372487189594, "grad_norm": 0.5027994513511658, "learning_rate": 0.0005, "loss": 1.4978488683700562, "step": 17954 }, { "epoch": 1.415451320457233, "grad_norm": 0.6226286292076111, "learning_rate": 0.0005, "loss": 1.720266342163086, "step": 17955 }, { "epoch": 1.415530153724872, "grad_norm": 0.5060862302780151, "learning_rate": 0.0005, "loss": 1.393925666809082, "step": 17956 }, { "epoch": 1.4156089869925108, "grad_norm": 0.5083896517753601, "learning_rate": 0.0005, "loss": 1.3607146739959717, "step": 17957 }, { "epoch": 1.4156878202601497, "grad_norm": 0.7272875308990479, "learning_rate": 0.0005, "loss": 1.6118947267532349, "step": 17958 }, { "epoch": 1.4157666535277889, "grad_norm": 0.5269272327423096, "learning_rate": 0.0005, "loss": 1.5723936557769775, "step": 17959 }, { "epoch": 1.4158454867954275, "grad_norm": 0.5657223463058472, "learning_rate": 0.0005, "loss": 1.6538608074188232, "step": 17960 }, { "epoch": 1.4159243200630667, "grad_norm": 2.299595594406128, "learning_rate": 0.0005, "loss": 1.7248222827911377, "step": 17961 }, { "epoch": 1.4160031533307056, "grad_norm": 0.5835013389587402, "learning_rate": 0.0005, "loss": 1.6890710592269897, "step": 17962 }, { "epoch": 1.4160819865983445, "grad_norm": 0.49586644768714905, "learning_rate": 0.0005, "loss": 1.625611424446106, "step": 17963 }, { "epoch": 1.4161608198659834, "grad_norm": 0.5005297064781189, "learning_rate": 0.0005, "loss": 1.5409343242645264, "step": 17964 }, { "epoch": 1.4162396531336223, "grad_norm": 0.5823498964309692, "learning_rate": 0.0005, "loss": 1.8556967973709106, "step": 17965 }, { "epoch": 1.4163184864012615, "grad_norm": 0.4657386541366577, "learning_rate": 0.0005, "loss": 1.7251685857772827, "step": 17966 }, { "epoch": 1.4163973196689001, "grad_norm": 0.6327925324440002, "learning_rate": 0.0005, "loss": 1.705597162246704, "step": 17967 }, { "epoch": 1.4164761529365393, "grad_norm": 0.7811388373374939, "learning_rate": 0.0005, "loss": 1.7072443962097168, "step": 17968 }, { "epoch": 1.4165549862041782, "grad_norm": 0.5356619358062744, "learning_rate": 0.0005, "loss": 1.505435585975647, "step": 17969 }, { "epoch": 1.416633819471817, "grad_norm": 0.6212309002876282, "learning_rate": 0.0005, "loss": 1.8283905982971191, "step": 17970 }, { "epoch": 1.416712652739456, "grad_norm": 0.8986990451812744, "learning_rate": 0.0005, "loss": 2.000952959060669, "step": 17971 }, { "epoch": 1.416791486007095, "grad_norm": 0.5188690423965454, "learning_rate": 0.0005, "loss": 1.4950584173202515, "step": 17972 }, { "epoch": 1.416870319274734, "grad_norm": 0.5154556035995483, "learning_rate": 0.0005, "loss": 1.5475561618804932, "step": 17973 }, { "epoch": 1.4169491525423727, "grad_norm": 0.5232121348381042, "learning_rate": 0.0005, "loss": 1.5670568943023682, "step": 17974 }, { "epoch": 1.4170279858100119, "grad_norm": 0.7531101703643799, "learning_rate": 0.0005, "loss": 1.5768100023269653, "step": 17975 }, { "epoch": 1.4171068190776508, "grad_norm": 0.5151351690292358, "learning_rate": 0.0005, "loss": 1.7453546524047852, "step": 17976 }, { "epoch": 1.4171856523452897, "grad_norm": 0.528175413608551, "learning_rate": 0.0005, "loss": 1.7348041534423828, "step": 17977 }, { "epoch": 1.4172644856129286, "grad_norm": 0.7317957878112793, "learning_rate": 0.0005, "loss": 1.607041597366333, "step": 17978 }, { "epoch": 1.4173433188805675, "grad_norm": 0.5386849045753479, "learning_rate": 0.0005, "loss": 1.6284607648849487, "step": 17979 }, { "epoch": 1.4174221521482067, "grad_norm": 0.8137750029563904, "learning_rate": 0.0005, "loss": 1.5550627708435059, "step": 17980 }, { "epoch": 1.4175009854158453, "grad_norm": 0.6860857605934143, "learning_rate": 0.0005, "loss": 1.7092686891555786, "step": 17981 }, { "epoch": 1.4175798186834845, "grad_norm": 0.5108044147491455, "learning_rate": 0.0005, "loss": 1.50739324092865, "step": 17982 }, { "epoch": 1.4176586519511234, "grad_norm": 0.5815989971160889, "learning_rate": 0.0005, "loss": 1.656732201576233, "step": 17983 }, { "epoch": 1.4177374852187623, "grad_norm": 0.5162238478660583, "learning_rate": 0.0005, "loss": 1.837972640991211, "step": 17984 }, { "epoch": 1.4178163184864012, "grad_norm": 0.5137572288513184, "learning_rate": 0.0005, "loss": 1.711674451828003, "step": 17985 }, { "epoch": 1.4178951517540401, "grad_norm": 0.7486334443092346, "learning_rate": 0.0005, "loss": 1.6403520107269287, "step": 17986 }, { "epoch": 1.4179739850216793, "grad_norm": 0.47780388593673706, "learning_rate": 0.0005, "loss": 1.625573754310608, "step": 17987 }, { "epoch": 1.4180528182893182, "grad_norm": 0.4993252754211426, "learning_rate": 0.0005, "loss": 1.6403989791870117, "step": 17988 }, { "epoch": 1.418131651556957, "grad_norm": 0.5217691659927368, "learning_rate": 0.0005, "loss": 1.5044444799423218, "step": 17989 }, { "epoch": 1.418210484824596, "grad_norm": 0.6635963320732117, "learning_rate": 0.0005, "loss": 1.6902698278427124, "step": 17990 }, { "epoch": 1.418289318092235, "grad_norm": 0.6450632214546204, "learning_rate": 0.0005, "loss": 1.5960757732391357, "step": 17991 }, { "epoch": 1.4183681513598738, "grad_norm": 0.6145427823066711, "learning_rate": 0.0005, "loss": 1.8246171474456787, "step": 17992 }, { "epoch": 1.4184469846275127, "grad_norm": 0.7174932956695557, "learning_rate": 0.0005, "loss": 1.7115682363510132, "step": 17993 }, { "epoch": 1.4185258178951519, "grad_norm": 0.4634760022163391, "learning_rate": 0.0005, "loss": 1.7248893976211548, "step": 17994 }, { "epoch": 1.4186046511627908, "grad_norm": 0.5004574656486511, "learning_rate": 0.0005, "loss": 1.5690054893493652, "step": 17995 }, { "epoch": 1.4186834844304297, "grad_norm": 0.45991823077201843, "learning_rate": 0.0005, "loss": 1.579089879989624, "step": 17996 }, { "epoch": 1.4187623176980686, "grad_norm": 0.8303950428962708, "learning_rate": 0.0005, "loss": 1.5556392669677734, "step": 17997 }, { "epoch": 1.4188411509657075, "grad_norm": 0.4366444945335388, "learning_rate": 0.0005, "loss": 1.4517998695373535, "step": 17998 }, { "epoch": 1.4189199842333464, "grad_norm": 0.5369849801063538, "learning_rate": 0.0005, "loss": 1.5159533023834229, "step": 17999 }, { "epoch": 1.4189988175009853, "grad_norm": 0.5883844494819641, "learning_rate": 0.0005, "loss": 1.7480294704437256, "step": 18000 }, { "epoch": 1.4190776507686245, "grad_norm": 0.45063266158103943, "learning_rate": 0.0005, "loss": 1.4577971696853638, "step": 18001 }, { "epoch": 1.4191564840362634, "grad_norm": 0.5115717649459839, "learning_rate": 0.0005, "loss": 1.5923762321472168, "step": 18002 }, { "epoch": 1.4192353173039023, "grad_norm": 0.5894367098808289, "learning_rate": 0.0005, "loss": 1.8677321672439575, "step": 18003 }, { "epoch": 1.4193141505715412, "grad_norm": 0.5025847554206848, "learning_rate": 0.0005, "loss": 1.653193473815918, "step": 18004 }, { "epoch": 1.4193929838391801, "grad_norm": 0.5894160866737366, "learning_rate": 0.0005, "loss": 1.6381025314331055, "step": 18005 }, { "epoch": 1.419471817106819, "grad_norm": 0.5747858285903931, "learning_rate": 0.0005, "loss": 1.8133995532989502, "step": 18006 }, { "epoch": 1.419550650374458, "grad_norm": 0.5856685042381287, "learning_rate": 0.0005, "loss": 1.7882375717163086, "step": 18007 }, { "epoch": 1.419629483642097, "grad_norm": 0.6872134208679199, "learning_rate": 0.0005, "loss": 1.8952752351760864, "step": 18008 }, { "epoch": 1.419708316909736, "grad_norm": 0.43271109461784363, "learning_rate": 0.0005, "loss": 1.4533872604370117, "step": 18009 }, { "epoch": 1.419787150177375, "grad_norm": 0.5527859330177307, "learning_rate": 0.0005, "loss": 1.6220932006835938, "step": 18010 }, { "epoch": 1.4198659834450138, "grad_norm": 0.49197888374328613, "learning_rate": 0.0005, "loss": 1.4708106517791748, "step": 18011 }, { "epoch": 1.4199448167126527, "grad_norm": 0.6568602919578552, "learning_rate": 0.0005, "loss": 1.5844122171401978, "step": 18012 }, { "epoch": 1.4200236499802916, "grad_norm": 0.785605251789093, "learning_rate": 0.0005, "loss": 1.9292852878570557, "step": 18013 }, { "epoch": 1.4201024832479305, "grad_norm": 0.6576514840126038, "learning_rate": 0.0005, "loss": 1.953701376914978, "step": 18014 }, { "epoch": 1.4201813165155697, "grad_norm": 0.9495972990989685, "learning_rate": 0.0005, "loss": 1.7679595947265625, "step": 18015 }, { "epoch": 1.4202601497832086, "grad_norm": 0.6094061732292175, "learning_rate": 0.0005, "loss": 1.645149827003479, "step": 18016 }, { "epoch": 1.4203389830508475, "grad_norm": 0.8168232440948486, "learning_rate": 0.0005, "loss": 1.7998549938201904, "step": 18017 }, { "epoch": 1.4204178163184864, "grad_norm": 0.5356431007385254, "learning_rate": 0.0005, "loss": 1.9034357070922852, "step": 18018 }, { "epoch": 1.4204966495861253, "grad_norm": 0.6297908425331116, "learning_rate": 0.0005, "loss": 1.6778242588043213, "step": 18019 }, { "epoch": 1.4205754828537642, "grad_norm": 0.5850971341133118, "learning_rate": 0.0005, "loss": 1.5479028224945068, "step": 18020 }, { "epoch": 1.4206543161214031, "grad_norm": 0.43459632992744446, "learning_rate": 0.0005, "loss": 1.4752492904663086, "step": 18021 }, { "epoch": 1.4207331493890423, "grad_norm": 0.5929307341575623, "learning_rate": 0.0005, "loss": 1.6307017803192139, "step": 18022 }, { "epoch": 1.4208119826566812, "grad_norm": 0.4839049279689789, "learning_rate": 0.0005, "loss": 1.5663373470306396, "step": 18023 }, { "epoch": 1.42089081592432, "grad_norm": 0.5556445717811584, "learning_rate": 0.0005, "loss": 1.697678565979004, "step": 18024 }, { "epoch": 1.420969649191959, "grad_norm": 0.5232328772544861, "learning_rate": 0.0005, "loss": 1.7143592834472656, "step": 18025 }, { "epoch": 1.421048482459598, "grad_norm": 0.6287497282028198, "learning_rate": 0.0005, "loss": 1.8375746011734009, "step": 18026 }, { "epoch": 1.4211273157272368, "grad_norm": 0.5706701874732971, "learning_rate": 0.0005, "loss": 1.524297833442688, "step": 18027 }, { "epoch": 1.4212061489948757, "grad_norm": 0.5095168352127075, "learning_rate": 0.0005, "loss": 1.5399796962738037, "step": 18028 }, { "epoch": 1.4212849822625149, "grad_norm": 0.9282760620117188, "learning_rate": 0.0005, "loss": 1.5276708602905273, "step": 18029 }, { "epoch": 1.4213638155301538, "grad_norm": 0.6318007707595825, "learning_rate": 0.0005, "loss": 1.7719115018844604, "step": 18030 }, { "epoch": 1.4214426487977927, "grad_norm": 0.5010900497436523, "learning_rate": 0.0005, "loss": 1.5931806564331055, "step": 18031 }, { "epoch": 1.4215214820654316, "grad_norm": 0.7367658019065857, "learning_rate": 0.0005, "loss": 1.7218364477157593, "step": 18032 }, { "epoch": 1.4216003153330705, "grad_norm": 0.45310717821121216, "learning_rate": 0.0005, "loss": 1.6209747791290283, "step": 18033 }, { "epoch": 1.4216791486007094, "grad_norm": 0.6067272424697876, "learning_rate": 0.0005, "loss": 1.7523236274719238, "step": 18034 }, { "epoch": 1.4217579818683483, "grad_norm": 0.5290053486824036, "learning_rate": 0.0005, "loss": 1.8674359321594238, "step": 18035 }, { "epoch": 1.4218368151359875, "grad_norm": 0.44821345806121826, "learning_rate": 0.0005, "loss": 1.7143937349319458, "step": 18036 }, { "epoch": 1.4219156484036264, "grad_norm": 0.49633657932281494, "learning_rate": 0.0005, "loss": 1.511022925376892, "step": 18037 }, { "epoch": 1.4219944816712653, "grad_norm": 0.7511879801750183, "learning_rate": 0.0005, "loss": 1.785735011100769, "step": 18038 }, { "epoch": 1.4220733149389042, "grad_norm": 0.5595651268959045, "learning_rate": 0.0005, "loss": 1.5383307933807373, "step": 18039 }, { "epoch": 1.4221521482065431, "grad_norm": 0.8023701310157776, "learning_rate": 0.0005, "loss": 1.6486270427703857, "step": 18040 }, { "epoch": 1.422230981474182, "grad_norm": 0.5348163843154907, "learning_rate": 0.0005, "loss": 1.6313400268554688, "step": 18041 }, { "epoch": 1.422309814741821, "grad_norm": 0.5882527232170105, "learning_rate": 0.0005, "loss": 1.8383979797363281, "step": 18042 }, { "epoch": 1.42238864800946, "grad_norm": 0.6616843342781067, "learning_rate": 0.0005, "loss": 1.5484375953674316, "step": 18043 }, { "epoch": 1.422467481277099, "grad_norm": 0.4756772816181183, "learning_rate": 0.0005, "loss": 1.5533710718154907, "step": 18044 }, { "epoch": 1.422546314544738, "grad_norm": 0.5459569692611694, "learning_rate": 0.0005, "loss": 1.6976890563964844, "step": 18045 }, { "epoch": 1.4226251478123768, "grad_norm": 1.3364108800888062, "learning_rate": 0.0005, "loss": 1.553476333618164, "step": 18046 }, { "epoch": 1.4227039810800157, "grad_norm": 0.5705585479736328, "learning_rate": 0.0005, "loss": 1.4739022254943848, "step": 18047 }, { "epoch": 1.4227828143476546, "grad_norm": 0.5242613554000854, "learning_rate": 0.0005, "loss": 1.4929580688476562, "step": 18048 }, { "epoch": 1.4228616476152935, "grad_norm": 0.5566743016242981, "learning_rate": 0.0005, "loss": 1.5338380336761475, "step": 18049 }, { "epoch": 1.4229404808829327, "grad_norm": 0.5545186996459961, "learning_rate": 0.0005, "loss": 1.6921882629394531, "step": 18050 }, { "epoch": 1.4230193141505716, "grad_norm": 0.6486117839813232, "learning_rate": 0.0005, "loss": 1.6811163425445557, "step": 18051 }, { "epoch": 1.4230981474182105, "grad_norm": 0.47428494691848755, "learning_rate": 0.0005, "loss": 1.4174580574035645, "step": 18052 }, { "epoch": 1.4231769806858494, "grad_norm": 0.5714849233627319, "learning_rate": 0.0005, "loss": 1.4221735000610352, "step": 18053 }, { "epoch": 1.4232558139534883, "grad_norm": 0.4549608826637268, "learning_rate": 0.0005, "loss": 1.7501640319824219, "step": 18054 }, { "epoch": 1.4233346472211272, "grad_norm": 0.6206644177436829, "learning_rate": 0.0005, "loss": 1.739863395690918, "step": 18055 }, { "epoch": 1.4234134804887661, "grad_norm": 0.4632502496242523, "learning_rate": 0.0005, "loss": 1.3493494987487793, "step": 18056 }, { "epoch": 1.4234923137564053, "grad_norm": 2.036098003387451, "learning_rate": 0.0005, "loss": 1.4091134071350098, "step": 18057 }, { "epoch": 1.4235711470240442, "grad_norm": 0.48099756240844727, "learning_rate": 0.0005, "loss": 1.621483564376831, "step": 18058 }, { "epoch": 1.423649980291683, "grad_norm": 0.9768662452697754, "learning_rate": 0.0005, "loss": 1.631115198135376, "step": 18059 }, { "epoch": 1.423728813559322, "grad_norm": 0.5218826532363892, "learning_rate": 0.0005, "loss": 1.366109013557434, "step": 18060 }, { "epoch": 1.423807646826961, "grad_norm": 0.9190846085548401, "learning_rate": 0.0005, "loss": 1.3945176601409912, "step": 18061 }, { "epoch": 1.4238864800946, "grad_norm": 0.7127844095230103, "learning_rate": 0.0005, "loss": 1.6862386465072632, "step": 18062 }, { "epoch": 1.4239653133622387, "grad_norm": 0.7714099884033203, "learning_rate": 0.0005, "loss": 1.570607304573059, "step": 18063 }, { "epoch": 1.4240441466298779, "grad_norm": 0.4365707337856293, "learning_rate": 0.0005, "loss": 1.3564033508300781, "step": 18064 }, { "epoch": 1.4241229798975168, "grad_norm": 0.6067990064620972, "learning_rate": 0.0005, "loss": 1.5934449434280396, "step": 18065 }, { "epoch": 1.4242018131651557, "grad_norm": 0.5429278016090393, "learning_rate": 0.0005, "loss": 1.6629782915115356, "step": 18066 }, { "epoch": 1.4242806464327946, "grad_norm": 0.64335036277771, "learning_rate": 0.0005, "loss": 1.6127262115478516, "step": 18067 }, { "epoch": 1.4243594797004335, "grad_norm": 0.6516739130020142, "learning_rate": 0.0005, "loss": 1.515712857246399, "step": 18068 }, { "epoch": 1.4244383129680727, "grad_norm": 0.5432168245315552, "learning_rate": 0.0005, "loss": 1.312178611755371, "step": 18069 }, { "epoch": 1.4245171462357114, "grad_norm": 0.640876293182373, "learning_rate": 0.0005, "loss": 1.5013256072998047, "step": 18070 }, { "epoch": 1.4245959795033505, "grad_norm": 0.5736388564109802, "learning_rate": 0.0005, "loss": 1.492866039276123, "step": 18071 }, { "epoch": 1.4246748127709894, "grad_norm": 0.6299993395805359, "learning_rate": 0.0005, "loss": 1.4958901405334473, "step": 18072 }, { "epoch": 1.4247536460386283, "grad_norm": 0.649423360824585, "learning_rate": 0.0005, "loss": 1.6104681491851807, "step": 18073 }, { "epoch": 1.4248324793062672, "grad_norm": 0.5327181220054626, "learning_rate": 0.0005, "loss": 1.3051953315734863, "step": 18074 }, { "epoch": 1.4249113125739061, "grad_norm": 0.6036944389343262, "learning_rate": 0.0005, "loss": 1.6149581670761108, "step": 18075 }, { "epoch": 1.4249901458415453, "grad_norm": 0.5102326273918152, "learning_rate": 0.0005, "loss": 1.4720951318740845, "step": 18076 }, { "epoch": 1.425068979109184, "grad_norm": 0.6134922504425049, "learning_rate": 0.0005, "loss": 1.4492816925048828, "step": 18077 }, { "epoch": 1.425147812376823, "grad_norm": 0.4539903700351715, "learning_rate": 0.0005, "loss": 1.463838815689087, "step": 18078 }, { "epoch": 1.425226645644462, "grad_norm": 0.7158799171447754, "learning_rate": 0.0005, "loss": 1.468166470527649, "step": 18079 }, { "epoch": 1.425305478912101, "grad_norm": 0.6498600840568542, "learning_rate": 0.0005, "loss": 1.8313804864883423, "step": 18080 }, { "epoch": 1.4253843121797398, "grad_norm": 0.5836257934570312, "learning_rate": 0.0005, "loss": 1.6028704643249512, "step": 18081 }, { "epoch": 1.4254631454473787, "grad_norm": 0.5757127404212952, "learning_rate": 0.0005, "loss": 1.4542899131774902, "step": 18082 }, { "epoch": 1.4255419787150179, "grad_norm": 0.6999983191490173, "learning_rate": 0.0005, "loss": 1.7469899654388428, "step": 18083 }, { "epoch": 1.4256208119826566, "grad_norm": 0.569021463394165, "learning_rate": 0.0005, "loss": 1.5718941688537598, "step": 18084 }, { "epoch": 1.4256996452502957, "grad_norm": 0.531564474105835, "learning_rate": 0.0005, "loss": 1.7012748718261719, "step": 18085 }, { "epoch": 1.4257784785179346, "grad_norm": 0.6856561899185181, "learning_rate": 0.0005, "loss": 1.511169672012329, "step": 18086 }, { "epoch": 1.4258573117855735, "grad_norm": 0.5291321277618408, "learning_rate": 0.0005, "loss": 1.7552852630615234, "step": 18087 }, { "epoch": 1.4259361450532124, "grad_norm": 0.632685661315918, "learning_rate": 0.0005, "loss": 1.467096209526062, "step": 18088 }, { "epoch": 1.4260149783208513, "grad_norm": 0.535967230796814, "learning_rate": 0.0005, "loss": 1.3949589729309082, "step": 18089 }, { "epoch": 1.4260938115884905, "grad_norm": 0.5378242135047913, "learning_rate": 0.0005, "loss": 1.537881851196289, "step": 18090 }, { "epoch": 1.4261726448561292, "grad_norm": 0.5609685182571411, "learning_rate": 0.0005, "loss": 1.4867833852767944, "step": 18091 }, { "epoch": 1.4262514781237683, "grad_norm": 0.4232846796512604, "learning_rate": 0.0005, "loss": 1.3628818988800049, "step": 18092 }, { "epoch": 1.4263303113914072, "grad_norm": 0.5190722942352295, "learning_rate": 0.0005, "loss": 1.4379093647003174, "step": 18093 }, { "epoch": 1.4264091446590461, "grad_norm": 0.3948274850845337, "learning_rate": 0.0005, "loss": 1.4337258338928223, "step": 18094 }, { "epoch": 1.426487977926685, "grad_norm": 0.5713648796081543, "learning_rate": 0.0005, "loss": 1.6705468893051147, "step": 18095 }, { "epoch": 1.426566811194324, "grad_norm": 0.4919995367527008, "learning_rate": 0.0005, "loss": 1.5495784282684326, "step": 18096 }, { "epoch": 1.426645644461963, "grad_norm": 0.4487287700176239, "learning_rate": 0.0005, "loss": 1.5404826402664185, "step": 18097 }, { "epoch": 1.4267244777296018, "grad_norm": 0.5022193789482117, "learning_rate": 0.0005, "loss": 1.6611804962158203, "step": 18098 }, { "epoch": 1.426803310997241, "grad_norm": 0.5769148468971252, "learning_rate": 0.0005, "loss": 1.4756968021392822, "step": 18099 }, { "epoch": 1.4268821442648798, "grad_norm": 0.43121495842933655, "learning_rate": 0.0005, "loss": 1.305872917175293, "step": 18100 }, { "epoch": 1.4269609775325187, "grad_norm": 0.5686548352241516, "learning_rate": 0.0005, "loss": 1.4683623313903809, "step": 18101 }, { "epoch": 1.4270398108001576, "grad_norm": 0.548097550868988, "learning_rate": 0.0005, "loss": 1.512220859527588, "step": 18102 }, { "epoch": 1.4271186440677965, "grad_norm": 0.4946097135543823, "learning_rate": 0.0005, "loss": 1.5525790452957153, "step": 18103 }, { "epoch": 1.4271974773354357, "grad_norm": 0.7440329790115356, "learning_rate": 0.0005, "loss": 1.5233604907989502, "step": 18104 }, { "epoch": 1.4272763106030744, "grad_norm": 0.5067028403282166, "learning_rate": 0.0005, "loss": 1.7889726161956787, "step": 18105 }, { "epoch": 1.4273551438707135, "grad_norm": 0.8299403786659241, "learning_rate": 0.0005, "loss": 1.3475215435028076, "step": 18106 }, { "epoch": 1.4274339771383524, "grad_norm": 0.5108566880226135, "learning_rate": 0.0005, "loss": 1.6836175918579102, "step": 18107 }, { "epoch": 1.4275128104059913, "grad_norm": 0.5997037291526794, "learning_rate": 0.0005, "loss": 1.6750340461730957, "step": 18108 }, { "epoch": 1.4275916436736302, "grad_norm": 0.4936523735523224, "learning_rate": 0.0005, "loss": 1.5182147026062012, "step": 18109 }, { "epoch": 1.4276704769412691, "grad_norm": 0.44762369990348816, "learning_rate": 0.0005, "loss": 1.5299752950668335, "step": 18110 }, { "epoch": 1.4277493102089083, "grad_norm": 0.5241575241088867, "learning_rate": 0.0005, "loss": 1.7907655239105225, "step": 18111 }, { "epoch": 1.4278281434765472, "grad_norm": 0.496969610452652, "learning_rate": 0.0005, "loss": 1.5255632400512695, "step": 18112 }, { "epoch": 1.427906976744186, "grad_norm": 0.4436023235321045, "learning_rate": 0.0005, "loss": 1.6215498447418213, "step": 18113 }, { "epoch": 1.427985810011825, "grad_norm": 0.5339767336845398, "learning_rate": 0.0005, "loss": 1.4030451774597168, "step": 18114 }, { "epoch": 1.428064643279464, "grad_norm": 0.4794771373271942, "learning_rate": 0.0005, "loss": 1.5968079566955566, "step": 18115 }, { "epoch": 1.4281434765471028, "grad_norm": 0.650251567363739, "learning_rate": 0.0005, "loss": 1.5200426578521729, "step": 18116 }, { "epoch": 1.4282223098147417, "grad_norm": 0.5985834002494812, "learning_rate": 0.0005, "loss": 1.9389439821243286, "step": 18117 }, { "epoch": 1.4283011430823809, "grad_norm": 0.4803158640861511, "learning_rate": 0.0005, "loss": 1.5798406600952148, "step": 18118 }, { "epoch": 1.4283799763500198, "grad_norm": 0.933738648891449, "learning_rate": 0.0005, "loss": 1.6833672523498535, "step": 18119 }, { "epoch": 1.4284588096176587, "grad_norm": 0.5459342002868652, "learning_rate": 0.0005, "loss": 1.3527499437332153, "step": 18120 }, { "epoch": 1.4285376428852976, "grad_norm": 0.6431871652603149, "learning_rate": 0.0005, "loss": 1.5765632390975952, "step": 18121 }, { "epoch": 1.4286164761529365, "grad_norm": 0.80776047706604, "learning_rate": 0.0005, "loss": 1.6616921424865723, "step": 18122 }, { "epoch": 1.4286953094205754, "grad_norm": 0.5980730652809143, "learning_rate": 0.0005, "loss": 1.580777883529663, "step": 18123 }, { "epoch": 1.4287741426882143, "grad_norm": 0.655921220779419, "learning_rate": 0.0005, "loss": 1.5851839780807495, "step": 18124 }, { "epoch": 1.4288529759558535, "grad_norm": 0.6648486852645874, "learning_rate": 0.0005, "loss": 1.631403923034668, "step": 18125 }, { "epoch": 1.4289318092234924, "grad_norm": 0.4692300260066986, "learning_rate": 0.0005, "loss": 1.1513304710388184, "step": 18126 }, { "epoch": 1.4290106424911313, "grad_norm": 0.7237858176231384, "learning_rate": 0.0005, "loss": 1.4404324293136597, "step": 18127 }, { "epoch": 1.4290894757587702, "grad_norm": 0.5340555906295776, "learning_rate": 0.0005, "loss": 1.428213357925415, "step": 18128 }, { "epoch": 1.4291683090264091, "grad_norm": 0.38167035579681396, "learning_rate": 0.0005, "loss": 1.253452181816101, "step": 18129 }, { "epoch": 1.429247142294048, "grad_norm": 0.42538660764694214, "learning_rate": 0.0005, "loss": 1.2813737392425537, "step": 18130 }, { "epoch": 1.429325975561687, "grad_norm": 0.5646782517433167, "learning_rate": 0.0005, "loss": 1.5402326583862305, "step": 18131 }, { "epoch": 1.429404808829326, "grad_norm": 0.5334259271621704, "learning_rate": 0.0005, "loss": 1.6145083904266357, "step": 18132 }, { "epoch": 1.429483642096965, "grad_norm": 0.5219849348068237, "learning_rate": 0.0005, "loss": 1.4986680746078491, "step": 18133 }, { "epoch": 1.429562475364604, "grad_norm": 0.572425127029419, "learning_rate": 0.0005, "loss": 1.6031382083892822, "step": 18134 }, { "epoch": 1.4296413086322428, "grad_norm": 0.6192176342010498, "learning_rate": 0.0005, "loss": 1.6123046875, "step": 18135 }, { "epoch": 1.4297201418998817, "grad_norm": 0.553924560546875, "learning_rate": 0.0005, "loss": 1.6302118301391602, "step": 18136 }, { "epoch": 1.4297989751675206, "grad_norm": 0.5038108825683594, "learning_rate": 0.0005, "loss": 1.5317349433898926, "step": 18137 }, { "epoch": 1.4298778084351595, "grad_norm": 0.665999710559845, "learning_rate": 0.0005, "loss": 1.6936472654342651, "step": 18138 }, { "epoch": 1.4299566417027987, "grad_norm": 0.5791017413139343, "learning_rate": 0.0005, "loss": 1.551087737083435, "step": 18139 }, { "epoch": 1.4300354749704376, "grad_norm": 0.6612746715545654, "learning_rate": 0.0005, "loss": 1.6044682264328003, "step": 18140 }, { "epoch": 1.4301143082380765, "grad_norm": 0.5256674885749817, "learning_rate": 0.0005, "loss": 1.5013978481292725, "step": 18141 }, { "epoch": 1.4301931415057154, "grad_norm": 0.4381096065044403, "learning_rate": 0.0005, "loss": 1.3229646682739258, "step": 18142 }, { "epoch": 1.4302719747733543, "grad_norm": 0.46028536558151245, "learning_rate": 0.0005, "loss": 1.6104834079742432, "step": 18143 }, { "epoch": 1.4303508080409932, "grad_norm": 0.459459513425827, "learning_rate": 0.0005, "loss": 1.4688270092010498, "step": 18144 }, { "epoch": 1.4304296413086321, "grad_norm": 0.5072638392448425, "learning_rate": 0.0005, "loss": 1.5568925142288208, "step": 18145 }, { "epoch": 1.4305084745762713, "grad_norm": 0.5327897667884827, "learning_rate": 0.0005, "loss": 1.527001142501831, "step": 18146 }, { "epoch": 1.4305873078439102, "grad_norm": 0.640428364276886, "learning_rate": 0.0005, "loss": 1.6957796812057495, "step": 18147 }, { "epoch": 1.430666141111549, "grad_norm": 0.703810453414917, "learning_rate": 0.0005, "loss": 1.6719452142715454, "step": 18148 }, { "epoch": 1.430744974379188, "grad_norm": 0.5417083501815796, "learning_rate": 0.0005, "loss": 1.4812443256378174, "step": 18149 }, { "epoch": 1.430823807646827, "grad_norm": 0.5653744339942932, "learning_rate": 0.0005, "loss": 1.4051823616027832, "step": 18150 }, { "epoch": 1.4309026409144658, "grad_norm": 0.5731260180473328, "learning_rate": 0.0005, "loss": 1.7281715869903564, "step": 18151 }, { "epoch": 1.4309814741821048, "grad_norm": 0.5814076662063599, "learning_rate": 0.0005, "loss": 1.5529839992523193, "step": 18152 }, { "epoch": 1.4310603074497439, "grad_norm": 0.49449682235717773, "learning_rate": 0.0005, "loss": 1.448716402053833, "step": 18153 }, { "epoch": 1.4311391407173828, "grad_norm": 0.4919857978820801, "learning_rate": 0.0005, "loss": 1.5063118934631348, "step": 18154 }, { "epoch": 1.4312179739850217, "grad_norm": 0.6239030957221985, "learning_rate": 0.0005, "loss": 1.6055359840393066, "step": 18155 }, { "epoch": 1.4312968072526606, "grad_norm": 0.5261671543121338, "learning_rate": 0.0005, "loss": 1.4696520566940308, "step": 18156 }, { "epoch": 1.4313756405202995, "grad_norm": 2.697176456451416, "learning_rate": 0.0005, "loss": 1.696718454360962, "step": 18157 }, { "epoch": 1.4314544737879384, "grad_norm": 0.6686839461326599, "learning_rate": 0.0005, "loss": 1.683318853378296, "step": 18158 }, { "epoch": 1.4315333070555774, "grad_norm": 0.49512580037117004, "learning_rate": 0.0005, "loss": 1.3538885116577148, "step": 18159 }, { "epoch": 1.4316121403232165, "grad_norm": 0.5833287239074707, "learning_rate": 0.0005, "loss": 1.592195987701416, "step": 18160 }, { "epoch": 1.4316909735908554, "grad_norm": 0.6039069890975952, "learning_rate": 0.0005, "loss": 1.5678948163986206, "step": 18161 }, { "epoch": 1.4317698068584943, "grad_norm": 0.5949841737747192, "learning_rate": 0.0005, "loss": 1.4967241287231445, "step": 18162 }, { "epoch": 1.4318486401261332, "grad_norm": 0.47892242670059204, "learning_rate": 0.0005, "loss": 1.3951845169067383, "step": 18163 }, { "epoch": 1.4319274733937721, "grad_norm": 0.611004114151001, "learning_rate": 0.0005, "loss": 1.3335020542144775, "step": 18164 }, { "epoch": 1.432006306661411, "grad_norm": 0.4872860014438629, "learning_rate": 0.0005, "loss": 1.5174082517623901, "step": 18165 }, { "epoch": 1.43208513992905, "grad_norm": 0.4794843792915344, "learning_rate": 0.0005, "loss": 1.4742172956466675, "step": 18166 }, { "epoch": 1.432163973196689, "grad_norm": 0.4757530391216278, "learning_rate": 0.0005, "loss": 1.5029656887054443, "step": 18167 }, { "epoch": 1.432242806464328, "grad_norm": 0.5227969288825989, "learning_rate": 0.0005, "loss": 1.4908143281936646, "step": 18168 }, { "epoch": 1.432321639731967, "grad_norm": 0.5637831687927246, "learning_rate": 0.0005, "loss": 1.4864925146102905, "step": 18169 }, { "epoch": 1.4324004729996058, "grad_norm": 0.8256455063819885, "learning_rate": 0.0005, "loss": 1.4354643821716309, "step": 18170 }, { "epoch": 1.4324793062672447, "grad_norm": 0.6734108924865723, "learning_rate": 0.0005, "loss": 1.4823167324066162, "step": 18171 }, { "epoch": 1.4325581395348836, "grad_norm": 0.5226425528526306, "learning_rate": 0.0005, "loss": 1.5064113140106201, "step": 18172 }, { "epoch": 1.4326369728025226, "grad_norm": 0.515003502368927, "learning_rate": 0.0005, "loss": 1.4714820384979248, "step": 18173 }, { "epoch": 1.4327158060701617, "grad_norm": 0.5499460101127625, "learning_rate": 0.0005, "loss": 1.6346015930175781, "step": 18174 }, { "epoch": 1.4327946393378006, "grad_norm": 0.5242586731910706, "learning_rate": 0.0005, "loss": 1.4034143686294556, "step": 18175 }, { "epoch": 1.4328734726054395, "grad_norm": 0.45493271946907043, "learning_rate": 0.0005, "loss": 1.5098199844360352, "step": 18176 }, { "epoch": 1.4329523058730784, "grad_norm": 0.7552780508995056, "learning_rate": 0.0005, "loss": 1.648679494857788, "step": 18177 }, { "epoch": 1.4330311391407173, "grad_norm": 0.5058891773223877, "learning_rate": 0.0005, "loss": 1.4178166389465332, "step": 18178 }, { "epoch": 1.4331099724083562, "grad_norm": 0.5048027038574219, "learning_rate": 0.0005, "loss": 1.552688717842102, "step": 18179 }, { "epoch": 1.4331888056759952, "grad_norm": 0.48415616154670715, "learning_rate": 0.0005, "loss": 1.514884114265442, "step": 18180 }, { "epoch": 1.4332676389436343, "grad_norm": 0.6065613627433777, "learning_rate": 0.0005, "loss": 1.5597752332687378, "step": 18181 }, { "epoch": 1.4333464722112732, "grad_norm": 0.8768861889839172, "learning_rate": 0.0005, "loss": 1.6776509284973145, "step": 18182 }, { "epoch": 1.4334253054789121, "grad_norm": 0.535388171672821, "learning_rate": 0.0005, "loss": 1.7047226428985596, "step": 18183 }, { "epoch": 1.433504138746551, "grad_norm": 2.704557418823242, "learning_rate": 0.0005, "loss": 1.659869909286499, "step": 18184 }, { "epoch": 1.43358297201419, "grad_norm": 1.215714931488037, "learning_rate": 0.0005, "loss": 1.7339203357696533, "step": 18185 }, { "epoch": 1.4336618052818288, "grad_norm": 0.5369459986686707, "learning_rate": 0.0005, "loss": 1.6599695682525635, "step": 18186 }, { "epoch": 1.4337406385494678, "grad_norm": 0.6184649467468262, "learning_rate": 0.0005, "loss": 1.4755945205688477, "step": 18187 }, { "epoch": 1.433819471817107, "grad_norm": 2.2297675609588623, "learning_rate": 0.0005, "loss": 1.8614087104797363, "step": 18188 }, { "epoch": 1.4338983050847458, "grad_norm": 0.662015974521637, "learning_rate": 0.0005, "loss": 1.8038747310638428, "step": 18189 }, { "epoch": 1.4339771383523847, "grad_norm": 0.6218988299369812, "learning_rate": 0.0005, "loss": 1.4063279628753662, "step": 18190 }, { "epoch": 1.4340559716200236, "grad_norm": 0.5634599328041077, "learning_rate": 0.0005, "loss": 1.4541059732437134, "step": 18191 }, { "epoch": 1.4341348048876625, "grad_norm": 0.5045937299728394, "learning_rate": 0.0005, "loss": 1.644694447517395, "step": 18192 }, { "epoch": 1.4342136381553017, "grad_norm": 0.8923017978668213, "learning_rate": 0.0005, "loss": 1.7669637203216553, "step": 18193 }, { "epoch": 1.4342924714229404, "grad_norm": 0.577501118183136, "learning_rate": 0.0005, "loss": 1.5071161985397339, "step": 18194 }, { "epoch": 1.4343713046905795, "grad_norm": 0.5387530326843262, "learning_rate": 0.0005, "loss": 1.6832911968231201, "step": 18195 }, { "epoch": 1.4344501379582184, "grad_norm": 0.555419921875, "learning_rate": 0.0005, "loss": 1.473764181137085, "step": 18196 }, { "epoch": 1.4345289712258573, "grad_norm": 0.5631279945373535, "learning_rate": 0.0005, "loss": 1.4831781387329102, "step": 18197 }, { "epoch": 1.4346078044934962, "grad_norm": 0.6111732125282288, "learning_rate": 0.0005, "loss": 1.5417364835739136, "step": 18198 }, { "epoch": 1.4346866377611351, "grad_norm": 0.556697428226471, "learning_rate": 0.0005, "loss": 1.776658058166504, "step": 18199 }, { "epoch": 1.4347654710287743, "grad_norm": 0.6652551293373108, "learning_rate": 0.0005, "loss": 1.503436803817749, "step": 18200 }, { "epoch": 1.434844304296413, "grad_norm": 0.5765525102615356, "learning_rate": 0.0005, "loss": 1.394256830215454, "step": 18201 }, { "epoch": 1.434923137564052, "grad_norm": 0.5644264221191406, "learning_rate": 0.0005, "loss": 1.683868169784546, "step": 18202 }, { "epoch": 1.435001970831691, "grad_norm": 0.6956952214241028, "learning_rate": 0.0005, "loss": 1.4867854118347168, "step": 18203 }, { "epoch": 1.43508080409933, "grad_norm": 0.46442916989326477, "learning_rate": 0.0005, "loss": 1.37192964553833, "step": 18204 }, { "epoch": 1.4351596373669688, "grad_norm": 0.6427491307258606, "learning_rate": 0.0005, "loss": 1.4404315948486328, "step": 18205 }, { "epoch": 1.4352384706346077, "grad_norm": 0.6076834201812744, "learning_rate": 0.0005, "loss": 1.5226513147354126, "step": 18206 }, { "epoch": 1.4353173039022469, "grad_norm": 0.4714129865169525, "learning_rate": 0.0005, "loss": 1.3747849464416504, "step": 18207 }, { "epoch": 1.4353961371698856, "grad_norm": 0.6152233481407166, "learning_rate": 0.0005, "loss": 1.8476874828338623, "step": 18208 }, { "epoch": 1.4354749704375247, "grad_norm": 0.657916784286499, "learning_rate": 0.0005, "loss": 1.7465648651123047, "step": 18209 }, { "epoch": 1.4355538037051636, "grad_norm": 0.6133403778076172, "learning_rate": 0.0005, "loss": 1.6764001846313477, "step": 18210 }, { "epoch": 1.4356326369728025, "grad_norm": 0.8684918880462646, "learning_rate": 0.0005, "loss": 1.6016957759857178, "step": 18211 }, { "epoch": 1.4357114702404414, "grad_norm": 0.5311144590377808, "learning_rate": 0.0005, "loss": 1.4713695049285889, "step": 18212 }, { "epoch": 1.4357903035080803, "grad_norm": 1.028303623199463, "learning_rate": 0.0005, "loss": 1.319137692451477, "step": 18213 }, { "epoch": 1.4358691367757195, "grad_norm": 0.5245403051376343, "learning_rate": 0.0005, "loss": 1.5088897943496704, "step": 18214 }, { "epoch": 1.4359479700433582, "grad_norm": 0.4813705086708069, "learning_rate": 0.0005, "loss": 1.659970760345459, "step": 18215 }, { "epoch": 1.4360268033109973, "grad_norm": 0.6089697480201721, "learning_rate": 0.0005, "loss": 1.8091329336166382, "step": 18216 }, { "epoch": 1.4361056365786362, "grad_norm": 0.6799575686454773, "learning_rate": 0.0005, "loss": 1.6673356294631958, "step": 18217 }, { "epoch": 1.4361844698462751, "grad_norm": 0.44926583766937256, "learning_rate": 0.0005, "loss": 1.401130199432373, "step": 18218 }, { "epoch": 1.436263303113914, "grad_norm": 0.6038839221000671, "learning_rate": 0.0005, "loss": 1.4807422161102295, "step": 18219 }, { "epoch": 1.436342136381553, "grad_norm": 1.0430004596710205, "learning_rate": 0.0005, "loss": 1.7064523696899414, "step": 18220 }, { "epoch": 1.436420969649192, "grad_norm": 0.5205963850021362, "learning_rate": 0.0005, "loss": 1.4584159851074219, "step": 18221 }, { "epoch": 1.4364998029168308, "grad_norm": 0.5610911846160889, "learning_rate": 0.0005, "loss": 1.4928579330444336, "step": 18222 }, { "epoch": 1.43657863618447, "grad_norm": 0.6343045234680176, "learning_rate": 0.0005, "loss": 1.283599615097046, "step": 18223 }, { "epoch": 1.4366574694521088, "grad_norm": 0.5766785740852356, "learning_rate": 0.0005, "loss": 1.424560546875, "step": 18224 }, { "epoch": 1.4367363027197477, "grad_norm": 0.4474974572658539, "learning_rate": 0.0005, "loss": 1.3510024547576904, "step": 18225 }, { "epoch": 1.4368151359873866, "grad_norm": 0.5608589053153992, "learning_rate": 0.0005, "loss": 1.463153600692749, "step": 18226 }, { "epoch": 1.4368939692550255, "grad_norm": 0.6196734309196472, "learning_rate": 0.0005, "loss": 1.6083660125732422, "step": 18227 }, { "epoch": 1.4369728025226647, "grad_norm": 0.7509151697158813, "learning_rate": 0.0005, "loss": 1.6253068447113037, "step": 18228 }, { "epoch": 1.4370516357903034, "grad_norm": 0.6081675887107849, "learning_rate": 0.0005, "loss": 1.8016992807388306, "step": 18229 }, { "epoch": 1.4371304690579425, "grad_norm": 0.5212462544441223, "learning_rate": 0.0005, "loss": 1.6019341945648193, "step": 18230 }, { "epoch": 1.4372093023255814, "grad_norm": 0.5548858046531677, "learning_rate": 0.0005, "loss": 1.5212459564208984, "step": 18231 }, { "epoch": 1.4372881355932203, "grad_norm": 0.6626866459846497, "learning_rate": 0.0005, "loss": 1.5036442279815674, "step": 18232 }, { "epoch": 1.4373669688608592, "grad_norm": 0.5213717222213745, "learning_rate": 0.0005, "loss": 1.5016777515411377, "step": 18233 }, { "epoch": 1.4374458021284982, "grad_norm": 0.6377875208854675, "learning_rate": 0.0005, "loss": 1.56058931350708, "step": 18234 }, { "epoch": 1.4375246353961373, "grad_norm": 0.5878604650497437, "learning_rate": 0.0005, "loss": 1.4663281440734863, "step": 18235 }, { "epoch": 1.4376034686637762, "grad_norm": 0.6428807973861694, "learning_rate": 0.0005, "loss": 1.5297856330871582, "step": 18236 }, { "epoch": 1.437682301931415, "grad_norm": 0.5598263740539551, "learning_rate": 0.0005, "loss": 1.7053470611572266, "step": 18237 }, { "epoch": 1.437761135199054, "grad_norm": 0.5810748338699341, "learning_rate": 0.0005, "loss": 1.672430157661438, "step": 18238 }, { "epoch": 1.437839968466693, "grad_norm": 0.433024525642395, "learning_rate": 0.0005, "loss": 1.2522658109664917, "step": 18239 }, { "epoch": 1.4379188017343318, "grad_norm": 0.4653678834438324, "learning_rate": 0.0005, "loss": 1.5057249069213867, "step": 18240 }, { "epoch": 1.4379976350019708, "grad_norm": 0.43940967321395874, "learning_rate": 0.0005, "loss": 1.5972861051559448, "step": 18241 }, { "epoch": 1.4380764682696099, "grad_norm": 1.562538504600525, "learning_rate": 0.0005, "loss": 1.6385514736175537, "step": 18242 }, { "epoch": 1.4381553015372488, "grad_norm": 0.5196695923805237, "learning_rate": 0.0005, "loss": 1.5442698001861572, "step": 18243 }, { "epoch": 1.4382341348048877, "grad_norm": 0.6473336219787598, "learning_rate": 0.0005, "loss": 1.4240760803222656, "step": 18244 }, { "epoch": 1.4383129680725266, "grad_norm": 0.5471493005752563, "learning_rate": 0.0005, "loss": 1.6122164726257324, "step": 18245 }, { "epoch": 1.4383918013401655, "grad_norm": 0.6011939644813538, "learning_rate": 0.0005, "loss": 1.6120789051055908, "step": 18246 }, { "epoch": 1.4384706346078044, "grad_norm": 0.6042835116386414, "learning_rate": 0.0005, "loss": 1.7405641078948975, "step": 18247 }, { "epoch": 1.4385494678754434, "grad_norm": 0.660534679889679, "learning_rate": 0.0005, "loss": 1.6556096076965332, "step": 18248 }, { "epoch": 1.4386283011430825, "grad_norm": 0.590548038482666, "learning_rate": 0.0005, "loss": 1.5354173183441162, "step": 18249 }, { "epoch": 1.4387071344107214, "grad_norm": 0.5405352115631104, "learning_rate": 0.0005, "loss": 1.5707690715789795, "step": 18250 }, { "epoch": 1.4387859676783603, "grad_norm": 0.5984033346176147, "learning_rate": 0.0005, "loss": 1.4289919137954712, "step": 18251 }, { "epoch": 1.4388648009459992, "grad_norm": 0.6902880072593689, "learning_rate": 0.0005, "loss": 1.7909810543060303, "step": 18252 }, { "epoch": 1.4389436342136381, "grad_norm": 0.505810558795929, "learning_rate": 0.0005, "loss": 1.6151304244995117, "step": 18253 }, { "epoch": 1.439022467481277, "grad_norm": 0.6471545696258545, "learning_rate": 0.0005, "loss": 1.4856781959533691, "step": 18254 }, { "epoch": 1.439101300748916, "grad_norm": 0.5459662675857544, "learning_rate": 0.0005, "loss": 1.463200569152832, "step": 18255 }, { "epoch": 1.439180134016555, "grad_norm": 0.5708228945732117, "learning_rate": 0.0005, "loss": 1.669968843460083, "step": 18256 }, { "epoch": 1.439258967284194, "grad_norm": 0.5496770739555359, "learning_rate": 0.0005, "loss": 1.7310338020324707, "step": 18257 }, { "epoch": 1.439337800551833, "grad_norm": 0.7865806221961975, "learning_rate": 0.0005, "loss": 1.6107300519943237, "step": 18258 }, { "epoch": 1.4394166338194718, "grad_norm": 0.5091579556465149, "learning_rate": 0.0005, "loss": 1.4089221954345703, "step": 18259 }, { "epoch": 1.4394954670871107, "grad_norm": 0.5920461416244507, "learning_rate": 0.0005, "loss": 1.7411473989486694, "step": 18260 }, { "epoch": 1.4395743003547496, "grad_norm": 0.5228467583656311, "learning_rate": 0.0005, "loss": 1.8277994394302368, "step": 18261 }, { "epoch": 1.4396531336223886, "grad_norm": 0.4743756949901581, "learning_rate": 0.0005, "loss": 1.3124439716339111, "step": 18262 }, { "epoch": 1.4397319668900277, "grad_norm": 0.5991233587265015, "learning_rate": 0.0005, "loss": 1.5951323509216309, "step": 18263 }, { "epoch": 1.4398108001576666, "grad_norm": 0.5084676146507263, "learning_rate": 0.0005, "loss": 1.404418706893921, "step": 18264 }, { "epoch": 1.4398896334253055, "grad_norm": 0.5207567811012268, "learning_rate": 0.0005, "loss": 1.648526668548584, "step": 18265 }, { "epoch": 1.4399684666929444, "grad_norm": 0.6270499229431152, "learning_rate": 0.0005, "loss": 1.7091777324676514, "step": 18266 }, { "epoch": 1.4400472999605833, "grad_norm": 0.5802664160728455, "learning_rate": 0.0005, "loss": 1.4389631748199463, "step": 18267 }, { "epoch": 1.4401261332282222, "grad_norm": 0.5729081034660339, "learning_rate": 0.0005, "loss": 1.5975077152252197, "step": 18268 }, { "epoch": 1.4402049664958612, "grad_norm": 0.5284788608551025, "learning_rate": 0.0005, "loss": 1.5696319341659546, "step": 18269 }, { "epoch": 1.4402837997635003, "grad_norm": 0.46680891513824463, "learning_rate": 0.0005, "loss": 1.502188801765442, "step": 18270 }, { "epoch": 1.4403626330311392, "grad_norm": 0.5932631492614746, "learning_rate": 0.0005, "loss": 1.609484314918518, "step": 18271 }, { "epoch": 1.4404414662987781, "grad_norm": 0.4992603361606598, "learning_rate": 0.0005, "loss": 1.4756457805633545, "step": 18272 }, { "epoch": 1.440520299566417, "grad_norm": 0.4964163899421692, "learning_rate": 0.0005, "loss": 1.6804707050323486, "step": 18273 }, { "epoch": 1.440599132834056, "grad_norm": 0.550455629825592, "learning_rate": 0.0005, "loss": 1.5966527462005615, "step": 18274 }, { "epoch": 1.4406779661016949, "grad_norm": 0.5981539487838745, "learning_rate": 0.0005, "loss": 1.5276507139205933, "step": 18275 }, { "epoch": 1.4407567993693338, "grad_norm": 0.6197121739387512, "learning_rate": 0.0005, "loss": 1.803031325340271, "step": 18276 }, { "epoch": 1.440835632636973, "grad_norm": 0.5946598052978516, "learning_rate": 0.0005, "loss": 1.5464320182800293, "step": 18277 }, { "epoch": 1.4409144659046118, "grad_norm": 0.6455270051956177, "learning_rate": 0.0005, "loss": 1.5975420475006104, "step": 18278 }, { "epoch": 1.4409932991722507, "grad_norm": 0.6181178689002991, "learning_rate": 0.0005, "loss": 1.4191043376922607, "step": 18279 }, { "epoch": 1.4410721324398896, "grad_norm": 0.5065026879310608, "learning_rate": 0.0005, "loss": 1.5102050304412842, "step": 18280 }, { "epoch": 1.4411509657075285, "grad_norm": 0.5465022325515747, "learning_rate": 0.0005, "loss": 1.498712420463562, "step": 18281 }, { "epoch": 1.4412297989751675, "grad_norm": 0.5191230177879333, "learning_rate": 0.0005, "loss": 1.559133768081665, "step": 18282 }, { "epoch": 1.4413086322428064, "grad_norm": 0.48866263031959534, "learning_rate": 0.0005, "loss": 1.3849329948425293, "step": 18283 }, { "epoch": 1.4413874655104455, "grad_norm": 0.5160747766494751, "learning_rate": 0.0005, "loss": 1.632049322128296, "step": 18284 }, { "epoch": 1.4414662987780844, "grad_norm": 0.5961524844169617, "learning_rate": 0.0005, "loss": 1.7944568395614624, "step": 18285 }, { "epoch": 1.4415451320457233, "grad_norm": 0.4912027418613434, "learning_rate": 0.0005, "loss": 1.6230734586715698, "step": 18286 }, { "epoch": 1.4416239653133622, "grad_norm": 0.48564115166664124, "learning_rate": 0.0005, "loss": 1.5537819862365723, "step": 18287 }, { "epoch": 1.4417027985810011, "grad_norm": 0.7725891470909119, "learning_rate": 0.0005, "loss": 1.86110258102417, "step": 18288 }, { "epoch": 1.44178163184864, "grad_norm": 0.5994765758514404, "learning_rate": 0.0005, "loss": 1.6222419738769531, "step": 18289 }, { "epoch": 1.441860465116279, "grad_norm": 0.5901739597320557, "learning_rate": 0.0005, "loss": 1.5061602592468262, "step": 18290 }, { "epoch": 1.441939298383918, "grad_norm": 0.5363523364067078, "learning_rate": 0.0005, "loss": 1.716773271560669, "step": 18291 }, { "epoch": 1.442018131651557, "grad_norm": 0.5425997376441956, "learning_rate": 0.0005, "loss": 1.5996087789535522, "step": 18292 }, { "epoch": 1.442096964919196, "grad_norm": 0.4699994623661041, "learning_rate": 0.0005, "loss": 1.5607340335845947, "step": 18293 }, { "epoch": 1.4421757981868348, "grad_norm": 0.5907993316650391, "learning_rate": 0.0005, "loss": 1.4268760681152344, "step": 18294 }, { "epoch": 1.4422546314544737, "grad_norm": 0.5601575970649719, "learning_rate": 0.0005, "loss": 1.514922022819519, "step": 18295 }, { "epoch": 1.4423334647221127, "grad_norm": 0.5627927184104919, "learning_rate": 0.0005, "loss": 1.6260449886322021, "step": 18296 }, { "epoch": 1.4424122979897516, "grad_norm": 0.5551924705505371, "learning_rate": 0.0005, "loss": 1.3532569408416748, "step": 18297 }, { "epoch": 1.4424911312573907, "grad_norm": 0.5676875114440918, "learning_rate": 0.0005, "loss": 1.5765390396118164, "step": 18298 }, { "epoch": 1.4425699645250296, "grad_norm": 0.400832861661911, "learning_rate": 0.0005, "loss": 1.4059476852416992, "step": 18299 }, { "epoch": 1.4426487977926685, "grad_norm": 0.49042728543281555, "learning_rate": 0.0005, "loss": 1.5254011154174805, "step": 18300 }, { "epoch": 1.4427276310603074, "grad_norm": 0.6804899573326111, "learning_rate": 0.0005, "loss": 1.352621078491211, "step": 18301 }, { "epoch": 1.4428064643279463, "grad_norm": 0.4509565532207489, "learning_rate": 0.0005, "loss": 1.4147567749023438, "step": 18302 }, { "epoch": 1.4428852975955853, "grad_norm": 0.5822299122810364, "learning_rate": 0.0005, "loss": 1.7925357818603516, "step": 18303 }, { "epoch": 1.4429641308632242, "grad_norm": 0.5128981471061707, "learning_rate": 0.0005, "loss": 1.6678242683410645, "step": 18304 }, { "epoch": 1.4430429641308633, "grad_norm": 0.5441682934761047, "learning_rate": 0.0005, "loss": 1.4658600091934204, "step": 18305 }, { "epoch": 1.4431217973985022, "grad_norm": 0.4750417172908783, "learning_rate": 0.0005, "loss": 1.3762625455856323, "step": 18306 }, { "epoch": 1.4432006306661411, "grad_norm": 0.5867278575897217, "learning_rate": 0.0005, "loss": 1.497441053390503, "step": 18307 }, { "epoch": 1.44327946393378, "grad_norm": 0.475021630525589, "learning_rate": 0.0005, "loss": 1.6444363594055176, "step": 18308 }, { "epoch": 1.443358297201419, "grad_norm": 0.43657249212265015, "learning_rate": 0.0005, "loss": 1.5077886581420898, "step": 18309 }, { "epoch": 1.4434371304690579, "grad_norm": 0.49824288487434387, "learning_rate": 0.0005, "loss": 1.5280261039733887, "step": 18310 }, { "epoch": 1.4435159637366968, "grad_norm": 0.5091392993927002, "learning_rate": 0.0005, "loss": 1.659201979637146, "step": 18311 }, { "epoch": 1.443594797004336, "grad_norm": 0.4622626006603241, "learning_rate": 0.0005, "loss": 1.5145536661148071, "step": 18312 }, { "epoch": 1.4436736302719748, "grad_norm": 0.4530857801437378, "learning_rate": 0.0005, "loss": 1.444155216217041, "step": 18313 }, { "epoch": 1.4437524635396137, "grad_norm": 0.41380780935287476, "learning_rate": 0.0005, "loss": 1.3541033267974854, "step": 18314 }, { "epoch": 1.4438312968072526, "grad_norm": 0.6250597834587097, "learning_rate": 0.0005, "loss": 1.4784157276153564, "step": 18315 }, { "epoch": 1.4439101300748916, "grad_norm": 0.6626115441322327, "learning_rate": 0.0005, "loss": 1.6017343997955322, "step": 18316 }, { "epoch": 1.4439889633425307, "grad_norm": 0.5226798057556152, "learning_rate": 0.0005, "loss": 1.4397269487380981, "step": 18317 }, { "epoch": 1.4440677966101694, "grad_norm": 0.5041396617889404, "learning_rate": 0.0005, "loss": 1.406071662902832, "step": 18318 }, { "epoch": 1.4441466298778085, "grad_norm": 0.7571945786476135, "learning_rate": 0.0005, "loss": 1.5987436771392822, "step": 18319 }, { "epoch": 1.4442254631454474, "grad_norm": 0.5456740856170654, "learning_rate": 0.0005, "loss": 1.5021450519561768, "step": 18320 }, { "epoch": 1.4443042964130863, "grad_norm": 0.5663691759109497, "learning_rate": 0.0005, "loss": 1.4542720317840576, "step": 18321 }, { "epoch": 1.4443831296807252, "grad_norm": 0.518421471118927, "learning_rate": 0.0005, "loss": 1.5157780647277832, "step": 18322 }, { "epoch": 1.4444619629483642, "grad_norm": 0.4864352345466614, "learning_rate": 0.0005, "loss": 1.494179606437683, "step": 18323 }, { "epoch": 1.4445407962160033, "grad_norm": 0.613680362701416, "learning_rate": 0.0005, "loss": 1.5018397569656372, "step": 18324 }, { "epoch": 1.444619629483642, "grad_norm": 0.6537707448005676, "learning_rate": 0.0005, "loss": 1.833773136138916, "step": 18325 }, { "epoch": 1.444698462751281, "grad_norm": 0.4146498143672943, "learning_rate": 0.0005, "loss": 1.370278000831604, "step": 18326 }, { "epoch": 1.44477729601892, "grad_norm": 0.5269195437431335, "learning_rate": 0.0005, "loss": 1.4444146156311035, "step": 18327 }, { "epoch": 1.444856129286559, "grad_norm": 0.6404036283493042, "learning_rate": 0.0005, "loss": 1.677384853363037, "step": 18328 }, { "epoch": 1.4449349625541978, "grad_norm": 0.39911726117134094, "learning_rate": 0.0005, "loss": 1.3698768615722656, "step": 18329 }, { "epoch": 1.4450137958218368, "grad_norm": 0.5438420176506042, "learning_rate": 0.0005, "loss": 1.6656012535095215, "step": 18330 }, { "epoch": 1.4450926290894759, "grad_norm": 0.59666907787323, "learning_rate": 0.0005, "loss": 1.6360218524932861, "step": 18331 }, { "epoch": 1.4451714623571146, "grad_norm": 0.5523399114608765, "learning_rate": 0.0005, "loss": 1.5600297451019287, "step": 18332 }, { "epoch": 1.4452502956247537, "grad_norm": 0.5666605234146118, "learning_rate": 0.0005, "loss": 1.5934550762176514, "step": 18333 }, { "epoch": 1.4453291288923926, "grad_norm": 0.4940355718135834, "learning_rate": 0.0005, "loss": 1.5315238237380981, "step": 18334 }, { "epoch": 1.4454079621600315, "grad_norm": 0.4655812382698059, "learning_rate": 0.0005, "loss": 1.5971254110336304, "step": 18335 }, { "epoch": 1.4454867954276704, "grad_norm": 0.6504653096199036, "learning_rate": 0.0005, "loss": 1.4683524370193481, "step": 18336 }, { "epoch": 1.4455656286953094, "grad_norm": 0.4966825842857361, "learning_rate": 0.0005, "loss": 1.3554425239562988, "step": 18337 }, { "epoch": 1.4456444619629485, "grad_norm": 0.5329834222793579, "learning_rate": 0.0005, "loss": 1.501102089881897, "step": 18338 }, { "epoch": 1.4457232952305872, "grad_norm": 0.5879033207893372, "learning_rate": 0.0005, "loss": 1.7008652687072754, "step": 18339 }, { "epoch": 1.4458021284982263, "grad_norm": 0.6657845973968506, "learning_rate": 0.0005, "loss": 1.6564414501190186, "step": 18340 }, { "epoch": 1.4458809617658652, "grad_norm": 0.6672524809837341, "learning_rate": 0.0005, "loss": 1.5605740547180176, "step": 18341 }, { "epoch": 1.4459597950335041, "grad_norm": 0.5595958232879639, "learning_rate": 0.0005, "loss": 1.5749704837799072, "step": 18342 }, { "epoch": 1.446038628301143, "grad_norm": 0.5239148139953613, "learning_rate": 0.0005, "loss": 1.4293551445007324, "step": 18343 }, { "epoch": 1.446117461568782, "grad_norm": 0.46732214093208313, "learning_rate": 0.0005, "loss": 1.5215740203857422, "step": 18344 }, { "epoch": 1.446196294836421, "grad_norm": 0.5617603659629822, "learning_rate": 0.0005, "loss": 1.5920411348342896, "step": 18345 }, { "epoch": 1.4462751281040598, "grad_norm": 0.5058388113975525, "learning_rate": 0.0005, "loss": 1.6188898086547852, "step": 18346 }, { "epoch": 1.446353961371699, "grad_norm": 0.436443030834198, "learning_rate": 0.0005, "loss": 1.3667842149734497, "step": 18347 }, { "epoch": 1.4464327946393378, "grad_norm": 0.49228349328041077, "learning_rate": 0.0005, "loss": 1.4381424188613892, "step": 18348 }, { "epoch": 1.4465116279069767, "grad_norm": 0.614795982837677, "learning_rate": 0.0005, "loss": 1.5945020914077759, "step": 18349 }, { "epoch": 1.4465904611746156, "grad_norm": 0.5746588706970215, "learning_rate": 0.0005, "loss": 1.5311859846115112, "step": 18350 }, { "epoch": 1.4466692944422546, "grad_norm": 0.49815309047698975, "learning_rate": 0.0005, "loss": 1.611359715461731, "step": 18351 }, { "epoch": 1.4467481277098937, "grad_norm": 0.5027337670326233, "learning_rate": 0.0005, "loss": 1.7370798587799072, "step": 18352 }, { "epoch": 1.4468269609775324, "grad_norm": 0.5560373067855835, "learning_rate": 0.0005, "loss": 1.3995556831359863, "step": 18353 }, { "epoch": 1.4469057942451715, "grad_norm": 0.47139808535575867, "learning_rate": 0.0005, "loss": 1.3141306638717651, "step": 18354 }, { "epoch": 1.4469846275128104, "grad_norm": 0.520909309387207, "learning_rate": 0.0005, "loss": 1.5708153247833252, "step": 18355 }, { "epoch": 1.4470634607804493, "grad_norm": 0.5495195984840393, "learning_rate": 0.0005, "loss": 1.7011795043945312, "step": 18356 }, { "epoch": 1.4471422940480883, "grad_norm": 0.5233219861984253, "learning_rate": 0.0005, "loss": 1.8262782096862793, "step": 18357 }, { "epoch": 1.4472211273157272, "grad_norm": 0.5808504819869995, "learning_rate": 0.0005, "loss": 1.552396297454834, "step": 18358 }, { "epoch": 1.4472999605833663, "grad_norm": 0.5818819999694824, "learning_rate": 0.0005, "loss": 1.6199944019317627, "step": 18359 }, { "epoch": 1.447378793851005, "grad_norm": 0.5307348966598511, "learning_rate": 0.0005, "loss": 1.4875084161758423, "step": 18360 }, { "epoch": 1.4474576271186441, "grad_norm": 0.5789532661437988, "learning_rate": 0.0005, "loss": 1.5869032144546509, "step": 18361 }, { "epoch": 1.447536460386283, "grad_norm": 0.5317620635032654, "learning_rate": 0.0005, "loss": 1.5473791360855103, "step": 18362 }, { "epoch": 1.447615293653922, "grad_norm": 0.6449772715568542, "learning_rate": 0.0005, "loss": 1.7697442770004272, "step": 18363 }, { "epoch": 1.4476941269215609, "grad_norm": 0.5508953928947449, "learning_rate": 0.0005, "loss": 1.539652705192566, "step": 18364 }, { "epoch": 1.4477729601891998, "grad_norm": 0.5158616304397583, "learning_rate": 0.0005, "loss": 1.7255252599716187, "step": 18365 }, { "epoch": 1.447851793456839, "grad_norm": 0.5431115031242371, "learning_rate": 0.0005, "loss": 1.3619487285614014, "step": 18366 }, { "epoch": 1.4479306267244778, "grad_norm": 0.6904632449150085, "learning_rate": 0.0005, "loss": 1.5562158823013306, "step": 18367 }, { "epoch": 1.4480094599921167, "grad_norm": 0.561911404132843, "learning_rate": 0.0005, "loss": 1.5105780363082886, "step": 18368 }, { "epoch": 1.4480882932597556, "grad_norm": 0.6528193950653076, "learning_rate": 0.0005, "loss": 1.398064374923706, "step": 18369 }, { "epoch": 1.4481671265273945, "grad_norm": 0.5331279635429382, "learning_rate": 0.0005, "loss": 1.457141637802124, "step": 18370 }, { "epoch": 1.4482459597950335, "grad_norm": 0.5158295631408691, "learning_rate": 0.0005, "loss": 1.4111876487731934, "step": 18371 }, { "epoch": 1.4483247930626724, "grad_norm": 0.5250354409217834, "learning_rate": 0.0005, "loss": 1.490342378616333, "step": 18372 }, { "epoch": 1.4484036263303115, "grad_norm": 0.5178812742233276, "learning_rate": 0.0005, "loss": 1.380880355834961, "step": 18373 }, { "epoch": 1.4484824595979504, "grad_norm": 0.5390326380729675, "learning_rate": 0.0005, "loss": 1.5423331260681152, "step": 18374 }, { "epoch": 1.4485612928655893, "grad_norm": 0.4023730456829071, "learning_rate": 0.0005, "loss": 1.4567147493362427, "step": 18375 }, { "epoch": 1.4486401261332282, "grad_norm": 0.45049214363098145, "learning_rate": 0.0005, "loss": 1.3267054557800293, "step": 18376 }, { "epoch": 1.4487189594008671, "grad_norm": 0.4610974192619324, "learning_rate": 0.0005, "loss": 1.4773722887039185, "step": 18377 }, { "epoch": 1.448797792668506, "grad_norm": 0.5939114689826965, "learning_rate": 0.0005, "loss": 1.814326524734497, "step": 18378 }, { "epoch": 1.448876625936145, "grad_norm": 0.5154197812080383, "learning_rate": 0.0005, "loss": 1.6291873455047607, "step": 18379 }, { "epoch": 1.448955459203784, "grad_norm": 0.6585137844085693, "learning_rate": 0.0005, "loss": 1.7125678062438965, "step": 18380 }, { "epoch": 1.449034292471423, "grad_norm": 0.547078549861908, "learning_rate": 0.0005, "loss": 1.7870177030563354, "step": 18381 }, { "epoch": 1.449113125739062, "grad_norm": 0.653736412525177, "learning_rate": 0.0005, "loss": 1.637285590171814, "step": 18382 }, { "epoch": 1.4491919590067008, "grad_norm": 0.5229493975639343, "learning_rate": 0.0005, "loss": 1.5324397087097168, "step": 18383 }, { "epoch": 1.4492707922743397, "grad_norm": 0.6206813454627991, "learning_rate": 0.0005, "loss": 1.7921205759048462, "step": 18384 }, { "epoch": 1.4493496255419787, "grad_norm": 0.5604457855224609, "learning_rate": 0.0005, "loss": 1.5879974365234375, "step": 18385 }, { "epoch": 1.4494284588096176, "grad_norm": 0.5351361036300659, "learning_rate": 0.0005, "loss": 1.4531863927841187, "step": 18386 }, { "epoch": 1.4495072920772567, "grad_norm": 0.5045092105865479, "learning_rate": 0.0005, "loss": 1.364995002746582, "step": 18387 }, { "epoch": 1.4495861253448956, "grad_norm": 0.6559140682220459, "learning_rate": 0.0005, "loss": 1.6549711227416992, "step": 18388 }, { "epoch": 1.4496649586125345, "grad_norm": 0.5536082983016968, "learning_rate": 0.0005, "loss": 1.4736862182617188, "step": 18389 }, { "epoch": 1.4497437918801734, "grad_norm": 0.5556439161300659, "learning_rate": 0.0005, "loss": 1.444307565689087, "step": 18390 }, { "epoch": 1.4498226251478123, "grad_norm": 0.4835375249385834, "learning_rate": 0.0005, "loss": 1.573653221130371, "step": 18391 }, { "epoch": 1.4499014584154513, "grad_norm": 0.40408623218536377, "learning_rate": 0.0005, "loss": 1.322904348373413, "step": 18392 }, { "epoch": 1.4499802916830902, "grad_norm": 0.6345676779747009, "learning_rate": 0.0005, "loss": 1.5325124263763428, "step": 18393 }, { "epoch": 1.4500591249507293, "grad_norm": 0.6839492917060852, "learning_rate": 0.0005, "loss": 1.6518932580947876, "step": 18394 }, { "epoch": 1.4501379582183682, "grad_norm": 0.5542586445808411, "learning_rate": 0.0005, "loss": 1.554962396621704, "step": 18395 }, { "epoch": 1.4502167914860071, "grad_norm": 0.4827882647514343, "learning_rate": 0.0005, "loss": 1.5742318630218506, "step": 18396 }, { "epoch": 1.450295624753646, "grad_norm": 0.4673595428466797, "learning_rate": 0.0005, "loss": 1.6548869609832764, "step": 18397 }, { "epoch": 1.450374458021285, "grad_norm": 0.5535352826118469, "learning_rate": 0.0005, "loss": 1.6446444988250732, "step": 18398 }, { "epoch": 1.4504532912889239, "grad_norm": 0.5715076327323914, "learning_rate": 0.0005, "loss": 1.5219135284423828, "step": 18399 }, { "epoch": 1.4505321245565628, "grad_norm": 0.5780900716781616, "learning_rate": 0.0005, "loss": 1.5880208015441895, "step": 18400 }, { "epoch": 1.450610957824202, "grad_norm": 0.6118278503417969, "learning_rate": 0.0005, "loss": 1.6055092811584473, "step": 18401 }, { "epoch": 1.4506897910918408, "grad_norm": 0.577955961227417, "learning_rate": 0.0005, "loss": 1.4972689151763916, "step": 18402 }, { "epoch": 1.4507686243594797, "grad_norm": 0.4671214818954468, "learning_rate": 0.0005, "loss": 1.515629529953003, "step": 18403 }, { "epoch": 1.4508474576271186, "grad_norm": 0.5217748284339905, "learning_rate": 0.0005, "loss": 1.4928746223449707, "step": 18404 }, { "epoch": 1.4509262908947576, "grad_norm": 0.5598281621932983, "learning_rate": 0.0005, "loss": 1.7625737190246582, "step": 18405 }, { "epoch": 1.4510051241623965, "grad_norm": 0.5045066475868225, "learning_rate": 0.0005, "loss": 1.6017277240753174, "step": 18406 }, { "epoch": 1.4510839574300354, "grad_norm": 0.5706778764724731, "learning_rate": 0.0005, "loss": 1.809065818786621, "step": 18407 }, { "epoch": 1.4511627906976745, "grad_norm": 0.4710198640823364, "learning_rate": 0.0005, "loss": 1.5081183910369873, "step": 18408 }, { "epoch": 1.4512416239653134, "grad_norm": 0.6490044593811035, "learning_rate": 0.0005, "loss": 1.6331183910369873, "step": 18409 }, { "epoch": 1.4513204572329523, "grad_norm": 0.9176210165023804, "learning_rate": 0.0005, "loss": 1.5747458934783936, "step": 18410 }, { "epoch": 1.4513992905005912, "grad_norm": 0.49968016147613525, "learning_rate": 0.0005, "loss": 1.4310650825500488, "step": 18411 }, { "epoch": 1.4514781237682302, "grad_norm": 0.5440177321434021, "learning_rate": 0.0005, "loss": 1.4716089963912964, "step": 18412 }, { "epoch": 1.451556957035869, "grad_norm": 0.525763988494873, "learning_rate": 0.0005, "loss": 1.6160718202590942, "step": 18413 }, { "epoch": 1.451635790303508, "grad_norm": 0.5067189931869507, "learning_rate": 0.0005, "loss": 1.5275111198425293, "step": 18414 }, { "epoch": 1.4517146235711471, "grad_norm": 0.709246039390564, "learning_rate": 0.0005, "loss": 1.560377836227417, "step": 18415 }, { "epoch": 1.451793456838786, "grad_norm": 0.5911233425140381, "learning_rate": 0.0005, "loss": 1.4788132905960083, "step": 18416 }, { "epoch": 1.451872290106425, "grad_norm": 0.6199328303337097, "learning_rate": 0.0005, "loss": 1.6862322092056274, "step": 18417 }, { "epoch": 1.4519511233740638, "grad_norm": 0.6868869066238403, "learning_rate": 0.0005, "loss": 1.8043205738067627, "step": 18418 }, { "epoch": 1.4520299566417028, "grad_norm": 0.7540428042411804, "learning_rate": 0.0005, "loss": 1.5479918718338013, "step": 18419 }, { "epoch": 1.4521087899093417, "grad_norm": 0.472027063369751, "learning_rate": 0.0005, "loss": 1.501822829246521, "step": 18420 }, { "epoch": 1.4521876231769806, "grad_norm": 0.6687750816345215, "learning_rate": 0.0005, "loss": 1.791884183883667, "step": 18421 }, { "epoch": 1.4522664564446197, "grad_norm": 0.5554295182228088, "learning_rate": 0.0005, "loss": 1.6814546585083008, "step": 18422 }, { "epoch": 1.4523452897122586, "grad_norm": 0.5264018774032593, "learning_rate": 0.0005, "loss": 1.5493077039718628, "step": 18423 }, { "epoch": 1.4524241229798975, "grad_norm": 0.5338947176933289, "learning_rate": 0.0005, "loss": 1.7304351329803467, "step": 18424 }, { "epoch": 1.4525029562475364, "grad_norm": 0.5645350813865662, "learning_rate": 0.0005, "loss": 1.462604284286499, "step": 18425 }, { "epoch": 1.4525817895151754, "grad_norm": 0.5914798974990845, "learning_rate": 0.0005, "loss": 1.6774678230285645, "step": 18426 }, { "epoch": 1.4526606227828143, "grad_norm": 0.664017379283905, "learning_rate": 0.0005, "loss": 1.5338808298110962, "step": 18427 }, { "epoch": 1.4527394560504532, "grad_norm": 0.6094858646392822, "learning_rate": 0.0005, "loss": 1.8070805072784424, "step": 18428 }, { "epoch": 1.4528182893180923, "grad_norm": 0.5109812617301941, "learning_rate": 0.0005, "loss": 1.729445219039917, "step": 18429 }, { "epoch": 1.4528971225857312, "grad_norm": 0.6060531139373779, "learning_rate": 0.0005, "loss": 1.7480167150497437, "step": 18430 }, { "epoch": 1.4529759558533701, "grad_norm": 0.6317071318626404, "learning_rate": 0.0005, "loss": 1.6161375045776367, "step": 18431 }, { "epoch": 1.453054789121009, "grad_norm": 0.5565896034240723, "learning_rate": 0.0005, "loss": 1.5577079057693481, "step": 18432 }, { "epoch": 1.453133622388648, "grad_norm": 0.5100029110908508, "learning_rate": 0.0005, "loss": 1.5118017196655273, "step": 18433 }, { "epoch": 1.4532124556562869, "grad_norm": 0.6188757419586182, "learning_rate": 0.0005, "loss": 2.0080199241638184, "step": 18434 }, { "epoch": 1.4532912889239258, "grad_norm": 0.5926173329353333, "learning_rate": 0.0005, "loss": 1.5232740640640259, "step": 18435 }, { "epoch": 1.453370122191565, "grad_norm": 0.6248822212219238, "learning_rate": 0.0005, "loss": 1.693871259689331, "step": 18436 }, { "epoch": 1.4534489554592038, "grad_norm": 0.6684752702713013, "learning_rate": 0.0005, "loss": 1.6100980043411255, "step": 18437 }, { "epoch": 1.4535277887268427, "grad_norm": 0.7060002684593201, "learning_rate": 0.0005, "loss": 1.7560532093048096, "step": 18438 }, { "epoch": 1.4536066219944817, "grad_norm": 0.571383535861969, "learning_rate": 0.0005, "loss": 1.7420753240585327, "step": 18439 }, { "epoch": 1.4536854552621206, "grad_norm": 0.6551961898803711, "learning_rate": 0.0005, "loss": 1.4599664211273193, "step": 18440 }, { "epoch": 1.4537642885297597, "grad_norm": 0.4839245080947876, "learning_rate": 0.0005, "loss": 1.388306975364685, "step": 18441 }, { "epoch": 1.4538431217973984, "grad_norm": 0.5532660484313965, "learning_rate": 0.0005, "loss": 1.6872138977050781, "step": 18442 }, { "epoch": 1.4539219550650375, "grad_norm": 0.6723770499229431, "learning_rate": 0.0005, "loss": 1.6368474960327148, "step": 18443 }, { "epoch": 1.4540007883326764, "grad_norm": 0.5529150366783142, "learning_rate": 0.0005, "loss": 1.5213054418563843, "step": 18444 }, { "epoch": 1.4540796216003153, "grad_norm": 0.5091696381568909, "learning_rate": 0.0005, "loss": 1.7173678874969482, "step": 18445 }, { "epoch": 1.4541584548679543, "grad_norm": 0.614034116268158, "learning_rate": 0.0005, "loss": 1.6442759037017822, "step": 18446 }, { "epoch": 1.4542372881355932, "grad_norm": 0.41878554224967957, "learning_rate": 0.0005, "loss": 1.4081687927246094, "step": 18447 }, { "epoch": 1.4543161214032323, "grad_norm": 0.588874340057373, "learning_rate": 0.0005, "loss": 1.6730639934539795, "step": 18448 }, { "epoch": 1.454394954670871, "grad_norm": 0.5311592817306519, "learning_rate": 0.0005, "loss": 1.4935519695281982, "step": 18449 }, { "epoch": 1.4544737879385101, "grad_norm": 0.5701539516448975, "learning_rate": 0.0005, "loss": 1.2833034992218018, "step": 18450 }, { "epoch": 1.454552621206149, "grad_norm": 0.7691032886505127, "learning_rate": 0.0005, "loss": 1.6039886474609375, "step": 18451 }, { "epoch": 1.454631454473788, "grad_norm": 0.6838710308074951, "learning_rate": 0.0005, "loss": 1.5035862922668457, "step": 18452 }, { "epoch": 1.4547102877414269, "grad_norm": 0.8329035639762878, "learning_rate": 0.0005, "loss": 1.4720348119735718, "step": 18453 }, { "epoch": 1.4547891210090658, "grad_norm": 0.4659077227115631, "learning_rate": 0.0005, "loss": 1.4697911739349365, "step": 18454 }, { "epoch": 1.454867954276705, "grad_norm": 0.6524291038513184, "learning_rate": 0.0005, "loss": 1.4715733528137207, "step": 18455 }, { "epoch": 1.4549467875443436, "grad_norm": 0.5030775666236877, "learning_rate": 0.0005, "loss": 1.3723702430725098, "step": 18456 }, { "epoch": 1.4550256208119827, "grad_norm": 0.5849975347518921, "learning_rate": 0.0005, "loss": 1.7036657333374023, "step": 18457 }, { "epoch": 1.4551044540796216, "grad_norm": 0.5028986930847168, "learning_rate": 0.0005, "loss": 1.4338308572769165, "step": 18458 }, { "epoch": 1.4551832873472605, "grad_norm": 0.6446177363395691, "learning_rate": 0.0005, "loss": 1.5934655666351318, "step": 18459 }, { "epoch": 1.4552621206148995, "grad_norm": 0.5657001733779907, "learning_rate": 0.0005, "loss": 1.5629446506500244, "step": 18460 }, { "epoch": 1.4553409538825384, "grad_norm": 0.5250626802444458, "learning_rate": 0.0005, "loss": 1.6349436044692993, "step": 18461 }, { "epoch": 1.4554197871501775, "grad_norm": 0.5314151644706726, "learning_rate": 0.0005, "loss": 1.6187585592269897, "step": 18462 }, { "epoch": 1.4554986204178162, "grad_norm": 0.5459542274475098, "learning_rate": 0.0005, "loss": 1.4610371589660645, "step": 18463 }, { "epoch": 1.4555774536854553, "grad_norm": 0.661373496055603, "learning_rate": 0.0005, "loss": 1.6624729633331299, "step": 18464 }, { "epoch": 1.4556562869530942, "grad_norm": 0.48577991127967834, "learning_rate": 0.0005, "loss": 1.4862909317016602, "step": 18465 }, { "epoch": 1.4557351202207331, "grad_norm": 0.4961652159690857, "learning_rate": 0.0005, "loss": 1.6940741539001465, "step": 18466 }, { "epoch": 1.455813953488372, "grad_norm": 0.55262690782547, "learning_rate": 0.0005, "loss": 1.661586046218872, "step": 18467 }, { "epoch": 1.455892786756011, "grad_norm": 0.5103346109390259, "learning_rate": 0.0005, "loss": 1.4329752922058105, "step": 18468 }, { "epoch": 1.45597162002365, "grad_norm": 0.5674987435340881, "learning_rate": 0.0005, "loss": 1.5942089557647705, "step": 18469 }, { "epoch": 1.4560504532912888, "grad_norm": 0.6366316080093384, "learning_rate": 0.0005, "loss": 1.5513845682144165, "step": 18470 }, { "epoch": 1.456129286558928, "grad_norm": 0.49384400248527527, "learning_rate": 0.0005, "loss": 1.4982815980911255, "step": 18471 }, { "epoch": 1.4562081198265668, "grad_norm": 0.4544993042945862, "learning_rate": 0.0005, "loss": 1.6449533700942993, "step": 18472 }, { "epoch": 1.4562869530942057, "grad_norm": 0.5082897543907166, "learning_rate": 0.0005, "loss": 1.4075546264648438, "step": 18473 }, { "epoch": 1.4563657863618447, "grad_norm": 0.6556231379508972, "learning_rate": 0.0005, "loss": 1.470893383026123, "step": 18474 }, { "epoch": 1.4564446196294836, "grad_norm": 0.48416683077812195, "learning_rate": 0.0005, "loss": 1.3501524925231934, "step": 18475 }, { "epoch": 1.4565234528971227, "grad_norm": 0.5142368674278259, "learning_rate": 0.0005, "loss": 1.431037425994873, "step": 18476 }, { "epoch": 1.4566022861647614, "grad_norm": 0.5316392779350281, "learning_rate": 0.0005, "loss": 1.5135364532470703, "step": 18477 }, { "epoch": 1.4566811194324005, "grad_norm": 0.6094940304756165, "learning_rate": 0.0005, "loss": 2.003354072570801, "step": 18478 }, { "epoch": 1.4567599527000394, "grad_norm": 0.40933623909950256, "learning_rate": 0.0005, "loss": 1.3424381017684937, "step": 18479 }, { "epoch": 1.4568387859676784, "grad_norm": 0.4941824674606323, "learning_rate": 0.0005, "loss": 1.3017475605010986, "step": 18480 }, { "epoch": 1.4569176192353173, "grad_norm": 0.5608862638473511, "learning_rate": 0.0005, "loss": 1.6533998250961304, "step": 18481 }, { "epoch": 1.4569964525029562, "grad_norm": 0.6326661109924316, "learning_rate": 0.0005, "loss": 1.4413344860076904, "step": 18482 }, { "epoch": 1.4570752857705953, "grad_norm": 0.671050488948822, "learning_rate": 0.0005, "loss": 1.575411081314087, "step": 18483 }, { "epoch": 1.457154119038234, "grad_norm": 0.5450177788734436, "learning_rate": 0.0005, "loss": 1.5283786058425903, "step": 18484 }, { "epoch": 1.4572329523058731, "grad_norm": 0.49051976203918457, "learning_rate": 0.0005, "loss": 1.3280540704727173, "step": 18485 }, { "epoch": 1.457311785573512, "grad_norm": 0.5847792029380798, "learning_rate": 0.0005, "loss": 1.4664052724838257, "step": 18486 }, { "epoch": 1.457390618841151, "grad_norm": 0.7484239339828491, "learning_rate": 0.0005, "loss": 1.2945985794067383, "step": 18487 }, { "epoch": 1.4574694521087899, "grad_norm": 0.4938788115978241, "learning_rate": 0.0005, "loss": 1.4736557006835938, "step": 18488 }, { "epoch": 1.4575482853764288, "grad_norm": 0.6360384821891785, "learning_rate": 0.0005, "loss": 1.5511057376861572, "step": 18489 }, { "epoch": 1.457627118644068, "grad_norm": 0.49718281626701355, "learning_rate": 0.0005, "loss": 1.571552038192749, "step": 18490 }, { "epoch": 1.4577059519117068, "grad_norm": 0.48706814646720886, "learning_rate": 0.0005, "loss": 1.262915849685669, "step": 18491 }, { "epoch": 1.4577847851793457, "grad_norm": 0.49007806181907654, "learning_rate": 0.0005, "loss": 1.5126280784606934, "step": 18492 }, { "epoch": 1.4578636184469846, "grad_norm": 0.589840292930603, "learning_rate": 0.0005, "loss": 1.8025858402252197, "step": 18493 }, { "epoch": 1.4579424517146236, "grad_norm": 0.816976010799408, "learning_rate": 0.0005, "loss": 1.5412962436676025, "step": 18494 }, { "epoch": 1.4580212849822625, "grad_norm": 0.6744569540023804, "learning_rate": 0.0005, "loss": 1.6515167951583862, "step": 18495 }, { "epoch": 1.4581001182499014, "grad_norm": 0.5373943448066711, "learning_rate": 0.0005, "loss": 1.5332646369934082, "step": 18496 }, { "epoch": 1.4581789515175405, "grad_norm": 0.6393885016441345, "learning_rate": 0.0005, "loss": 1.679480791091919, "step": 18497 }, { "epoch": 1.4582577847851794, "grad_norm": 0.5204517841339111, "learning_rate": 0.0005, "loss": 1.4979431629180908, "step": 18498 }, { "epoch": 1.4583366180528183, "grad_norm": 0.6415833830833435, "learning_rate": 0.0005, "loss": 1.4166960716247559, "step": 18499 }, { "epoch": 1.4584154513204572, "grad_norm": 0.5816512107849121, "learning_rate": 0.0005, "loss": 1.6142343282699585, "step": 18500 }, { "epoch": 1.4584942845880962, "grad_norm": 0.47599300742149353, "learning_rate": 0.0005, "loss": 1.5912773609161377, "step": 18501 }, { "epoch": 1.458573117855735, "grad_norm": 0.47126373648643494, "learning_rate": 0.0005, "loss": 1.38517427444458, "step": 18502 }, { "epoch": 1.458651951123374, "grad_norm": 0.4518100917339325, "learning_rate": 0.0005, "loss": 1.4775910377502441, "step": 18503 }, { "epoch": 1.4587307843910131, "grad_norm": 0.658376932144165, "learning_rate": 0.0005, "loss": 1.4781041145324707, "step": 18504 }, { "epoch": 1.458809617658652, "grad_norm": 0.5234808325767517, "learning_rate": 0.0005, "loss": 1.4313995838165283, "step": 18505 }, { "epoch": 1.458888450926291, "grad_norm": 0.6325788497924805, "learning_rate": 0.0005, "loss": 1.738681674003601, "step": 18506 }, { "epoch": 1.4589672841939298, "grad_norm": 0.6246323585510254, "learning_rate": 0.0005, "loss": 1.4441750049591064, "step": 18507 }, { "epoch": 1.4590461174615688, "grad_norm": 0.6367803812026978, "learning_rate": 0.0005, "loss": 1.6306041479110718, "step": 18508 }, { "epoch": 1.4591249507292077, "grad_norm": 0.6055464744567871, "learning_rate": 0.0005, "loss": 1.7267738580703735, "step": 18509 }, { "epoch": 1.4592037839968466, "grad_norm": 0.5832085609436035, "learning_rate": 0.0005, "loss": 1.625730037689209, "step": 18510 }, { "epoch": 1.4592826172644857, "grad_norm": 0.5651509165763855, "learning_rate": 0.0005, "loss": 1.5824928283691406, "step": 18511 }, { "epoch": 1.4593614505321246, "grad_norm": 0.7238428592681885, "learning_rate": 0.0005, "loss": 1.9025299549102783, "step": 18512 }, { "epoch": 1.4594402837997635, "grad_norm": 0.4907800853252411, "learning_rate": 0.0005, "loss": 1.553914189338684, "step": 18513 }, { "epoch": 1.4595191170674024, "grad_norm": 0.5217940211296082, "learning_rate": 0.0005, "loss": 1.6820868253707886, "step": 18514 }, { "epoch": 1.4595979503350414, "grad_norm": 0.4410321116447449, "learning_rate": 0.0005, "loss": 1.5378575325012207, "step": 18515 }, { "epoch": 1.4596767836026803, "grad_norm": 0.47296956181526184, "learning_rate": 0.0005, "loss": 1.3213229179382324, "step": 18516 }, { "epoch": 1.4597556168703192, "grad_norm": 0.5257241725921631, "learning_rate": 0.0005, "loss": 1.5343105792999268, "step": 18517 }, { "epoch": 1.4598344501379583, "grad_norm": 0.5216602087020874, "learning_rate": 0.0005, "loss": 1.5663926601409912, "step": 18518 }, { "epoch": 1.4599132834055972, "grad_norm": 0.5368348956108093, "learning_rate": 0.0005, "loss": 1.456824541091919, "step": 18519 }, { "epoch": 1.4599921166732361, "grad_norm": 0.5133164525032043, "learning_rate": 0.0005, "loss": 1.5012009143829346, "step": 18520 }, { "epoch": 1.460070949940875, "grad_norm": 0.5446235537528992, "learning_rate": 0.0005, "loss": 1.467183232307434, "step": 18521 }, { "epoch": 1.460149783208514, "grad_norm": 0.4742785096168518, "learning_rate": 0.0005, "loss": 1.5545554161071777, "step": 18522 }, { "epoch": 1.4602286164761529, "grad_norm": 0.5000776052474976, "learning_rate": 0.0005, "loss": 1.479906439781189, "step": 18523 }, { "epoch": 1.4603074497437918, "grad_norm": 0.5073516964912415, "learning_rate": 0.0005, "loss": 1.5215435028076172, "step": 18524 }, { "epoch": 1.460386283011431, "grad_norm": 0.5284565091133118, "learning_rate": 0.0005, "loss": 1.6140925884246826, "step": 18525 }, { "epoch": 1.4604651162790698, "grad_norm": 0.698981523513794, "learning_rate": 0.0005, "loss": 1.6057136058807373, "step": 18526 }, { "epoch": 1.4605439495467087, "grad_norm": 0.5012643337249756, "learning_rate": 0.0005, "loss": 1.4063138961791992, "step": 18527 }, { "epoch": 1.4606227828143477, "grad_norm": 0.5450873970985413, "learning_rate": 0.0005, "loss": 1.5800666809082031, "step": 18528 }, { "epoch": 1.4607016160819866, "grad_norm": 0.4719851613044739, "learning_rate": 0.0005, "loss": 1.4253102540969849, "step": 18529 }, { "epoch": 1.4607804493496255, "grad_norm": 0.515674889087677, "learning_rate": 0.0005, "loss": 1.6045395135879517, "step": 18530 }, { "epoch": 1.4608592826172644, "grad_norm": 0.653081476688385, "learning_rate": 0.0005, "loss": 1.5516557693481445, "step": 18531 }, { "epoch": 1.4609381158849035, "grad_norm": 0.6082471013069153, "learning_rate": 0.0005, "loss": 1.8217897415161133, "step": 18532 }, { "epoch": 1.4610169491525424, "grad_norm": 0.6620026230812073, "learning_rate": 0.0005, "loss": 1.6308478116989136, "step": 18533 }, { "epoch": 1.4610957824201813, "grad_norm": 0.6405550241470337, "learning_rate": 0.0005, "loss": 1.6149792671203613, "step": 18534 }, { "epoch": 1.4611746156878203, "grad_norm": 0.605317234992981, "learning_rate": 0.0005, "loss": 1.5331716537475586, "step": 18535 }, { "epoch": 1.4612534489554592, "grad_norm": 0.5420599579811096, "learning_rate": 0.0005, "loss": 1.4863696098327637, "step": 18536 }, { "epoch": 1.461332282223098, "grad_norm": 0.5005265474319458, "learning_rate": 0.0005, "loss": 1.7158868312835693, "step": 18537 }, { "epoch": 1.461411115490737, "grad_norm": 0.48944443464279175, "learning_rate": 0.0005, "loss": 1.5428516864776611, "step": 18538 }, { "epoch": 1.4614899487583761, "grad_norm": 0.5112290382385254, "learning_rate": 0.0005, "loss": 1.657697319984436, "step": 18539 }, { "epoch": 1.461568782026015, "grad_norm": 0.504127025604248, "learning_rate": 0.0005, "loss": 1.4270691871643066, "step": 18540 }, { "epoch": 1.461647615293654, "grad_norm": 0.5039304494857788, "learning_rate": 0.0005, "loss": 1.7444534301757812, "step": 18541 }, { "epoch": 1.4617264485612929, "grad_norm": 0.5883944630622864, "learning_rate": 0.0005, "loss": 1.5783803462982178, "step": 18542 }, { "epoch": 1.4618052818289318, "grad_norm": 0.4963913559913635, "learning_rate": 0.0005, "loss": 1.3774309158325195, "step": 18543 }, { "epoch": 1.4618841150965707, "grad_norm": 0.5570245981216431, "learning_rate": 0.0005, "loss": 1.4282970428466797, "step": 18544 }, { "epoch": 1.4619629483642096, "grad_norm": 0.5860284566879272, "learning_rate": 0.0005, "loss": 1.4570198059082031, "step": 18545 }, { "epoch": 1.4620417816318487, "grad_norm": 0.4913499057292938, "learning_rate": 0.0005, "loss": 1.5583012104034424, "step": 18546 }, { "epoch": 1.4621206148994876, "grad_norm": 0.7831045389175415, "learning_rate": 0.0005, "loss": 1.702361822128296, "step": 18547 }, { "epoch": 1.4621994481671265, "grad_norm": 0.47941359877586365, "learning_rate": 0.0005, "loss": 1.6020090579986572, "step": 18548 }, { "epoch": 1.4622782814347655, "grad_norm": 0.6191660761833191, "learning_rate": 0.0005, "loss": 1.5567363500595093, "step": 18549 }, { "epoch": 1.4623571147024044, "grad_norm": 0.5771525502204895, "learning_rate": 0.0005, "loss": 1.297126293182373, "step": 18550 }, { "epoch": 1.4624359479700433, "grad_norm": 0.5880986452102661, "learning_rate": 0.0005, "loss": 1.444462776184082, "step": 18551 }, { "epoch": 1.4625147812376822, "grad_norm": 0.6059014201164246, "learning_rate": 0.0005, "loss": 1.6398391723632812, "step": 18552 }, { "epoch": 1.4625936145053213, "grad_norm": 0.8194805979728699, "learning_rate": 0.0005, "loss": 1.450001835823059, "step": 18553 }, { "epoch": 1.4626724477729602, "grad_norm": 0.5871614813804626, "learning_rate": 0.0005, "loss": 1.596125841140747, "step": 18554 }, { "epoch": 1.4627512810405991, "grad_norm": 0.5973357558250427, "learning_rate": 0.0005, "loss": 1.5689094066619873, "step": 18555 }, { "epoch": 1.462830114308238, "grad_norm": 0.6446822285652161, "learning_rate": 0.0005, "loss": 1.787751317024231, "step": 18556 }, { "epoch": 1.462908947575877, "grad_norm": 0.6125985980033875, "learning_rate": 0.0005, "loss": 1.47917640209198, "step": 18557 }, { "epoch": 1.4629877808435159, "grad_norm": 0.4273500144481659, "learning_rate": 0.0005, "loss": 1.4182943105697632, "step": 18558 }, { "epoch": 1.4630666141111548, "grad_norm": 0.6039583086967468, "learning_rate": 0.0005, "loss": 1.454128623008728, "step": 18559 }, { "epoch": 1.463145447378794, "grad_norm": 0.6028890013694763, "learning_rate": 0.0005, "loss": 1.4942073822021484, "step": 18560 }, { "epoch": 1.4632242806464328, "grad_norm": 0.5603935718536377, "learning_rate": 0.0005, "loss": 1.7748523950576782, "step": 18561 }, { "epoch": 1.4633031139140718, "grad_norm": 0.617389976978302, "learning_rate": 0.0005, "loss": 1.5787012577056885, "step": 18562 }, { "epoch": 1.4633819471817107, "grad_norm": 0.6048170328140259, "learning_rate": 0.0005, "loss": 1.6412932872772217, "step": 18563 }, { "epoch": 1.4634607804493496, "grad_norm": 0.5039775967597961, "learning_rate": 0.0005, "loss": 1.497228980064392, "step": 18564 }, { "epoch": 1.4635396137169887, "grad_norm": 0.6018979549407959, "learning_rate": 0.0005, "loss": 1.5088285207748413, "step": 18565 }, { "epoch": 1.4636184469846274, "grad_norm": 0.5548940896987915, "learning_rate": 0.0005, "loss": 1.5390591621398926, "step": 18566 }, { "epoch": 1.4636972802522665, "grad_norm": 0.466126024723053, "learning_rate": 0.0005, "loss": 1.2645511627197266, "step": 18567 }, { "epoch": 1.4637761135199054, "grad_norm": 0.6719940304756165, "learning_rate": 0.0005, "loss": 1.570406198501587, "step": 18568 }, { "epoch": 1.4638549467875444, "grad_norm": 1.7740321159362793, "learning_rate": 0.0005, "loss": 1.4058150053024292, "step": 18569 }, { "epoch": 1.4639337800551833, "grad_norm": 0.5987128615379333, "learning_rate": 0.0005, "loss": 1.6060853004455566, "step": 18570 }, { "epoch": 1.4640126133228222, "grad_norm": 0.5511863231658936, "learning_rate": 0.0005, "loss": 1.4837538003921509, "step": 18571 }, { "epoch": 1.4640914465904613, "grad_norm": 1.236272931098938, "learning_rate": 0.0005, "loss": 1.740121603012085, "step": 18572 }, { "epoch": 1.4641702798581, "grad_norm": 0.49613848328590393, "learning_rate": 0.0005, "loss": 1.4665307998657227, "step": 18573 }, { "epoch": 1.4642491131257391, "grad_norm": 0.4552609324455261, "learning_rate": 0.0005, "loss": 1.4148263931274414, "step": 18574 }, { "epoch": 1.464327946393378, "grad_norm": 0.5391438603401184, "learning_rate": 0.0005, "loss": 1.365105152130127, "step": 18575 }, { "epoch": 1.464406779661017, "grad_norm": 0.4897105395793915, "learning_rate": 0.0005, "loss": 1.5491664409637451, "step": 18576 }, { "epoch": 1.4644856129286559, "grad_norm": 0.6891091465950012, "learning_rate": 0.0005, "loss": 1.4828745126724243, "step": 18577 }, { "epoch": 1.4645644461962948, "grad_norm": 0.5640395879745483, "learning_rate": 0.0005, "loss": 1.6730930805206299, "step": 18578 }, { "epoch": 1.464643279463934, "grad_norm": 0.5105741620063782, "learning_rate": 0.0005, "loss": 1.4801619052886963, "step": 18579 }, { "epoch": 1.4647221127315726, "grad_norm": 0.5597403645515442, "learning_rate": 0.0005, "loss": 1.4736394882202148, "step": 18580 }, { "epoch": 1.4648009459992117, "grad_norm": 0.5230032205581665, "learning_rate": 0.0005, "loss": 1.620247483253479, "step": 18581 }, { "epoch": 1.4648797792668506, "grad_norm": 0.49392545223236084, "learning_rate": 0.0005, "loss": 1.4213204383850098, "step": 18582 }, { "epoch": 1.4649586125344896, "grad_norm": 0.5102058053016663, "learning_rate": 0.0005, "loss": 1.5832910537719727, "step": 18583 }, { "epoch": 1.4650374458021285, "grad_norm": 0.5559806227684021, "learning_rate": 0.0005, "loss": 1.6229053735733032, "step": 18584 }, { "epoch": 1.4651162790697674, "grad_norm": 0.5231059789657593, "learning_rate": 0.0005, "loss": 1.614450216293335, "step": 18585 }, { "epoch": 1.4651951123374065, "grad_norm": 0.5784267783164978, "learning_rate": 0.0005, "loss": 1.4696332216262817, "step": 18586 }, { "epoch": 1.4652739456050452, "grad_norm": 0.5395393967628479, "learning_rate": 0.0005, "loss": 1.5380821228027344, "step": 18587 }, { "epoch": 1.4653527788726843, "grad_norm": 0.5622324347496033, "learning_rate": 0.0005, "loss": 1.5512642860412598, "step": 18588 }, { "epoch": 1.4654316121403232, "grad_norm": 0.5045481324195862, "learning_rate": 0.0005, "loss": 1.4598928689956665, "step": 18589 }, { "epoch": 1.4655104454079622, "grad_norm": 0.5478038787841797, "learning_rate": 0.0005, "loss": 1.5107401609420776, "step": 18590 }, { "epoch": 1.465589278675601, "grad_norm": 0.7122552990913391, "learning_rate": 0.0005, "loss": 1.6498637199401855, "step": 18591 }, { "epoch": 1.46566811194324, "grad_norm": 0.9952462911605835, "learning_rate": 0.0005, "loss": 1.704038143157959, "step": 18592 }, { "epoch": 1.4657469452108791, "grad_norm": 0.5479901432991028, "learning_rate": 0.0005, "loss": 1.5266563892364502, "step": 18593 }, { "epoch": 1.4658257784785178, "grad_norm": 0.5155211687088013, "learning_rate": 0.0005, "loss": 1.5216901302337646, "step": 18594 }, { "epoch": 1.465904611746157, "grad_norm": 0.5939970016479492, "learning_rate": 0.0005, "loss": 1.603890299797058, "step": 18595 }, { "epoch": 1.4659834450137958, "grad_norm": 0.7834717631340027, "learning_rate": 0.0005, "loss": 1.6229586601257324, "step": 18596 }, { "epoch": 1.4660622782814348, "grad_norm": 0.43369096517562866, "learning_rate": 0.0005, "loss": 1.2124755382537842, "step": 18597 }, { "epoch": 1.4661411115490737, "grad_norm": 0.5185592174530029, "learning_rate": 0.0005, "loss": 1.5285053253173828, "step": 18598 }, { "epoch": 1.4662199448167126, "grad_norm": 0.5514821410179138, "learning_rate": 0.0005, "loss": 1.3680310249328613, "step": 18599 }, { "epoch": 1.4662987780843517, "grad_norm": 0.6760275959968567, "learning_rate": 0.0005, "loss": 1.646705985069275, "step": 18600 }, { "epoch": 1.4663776113519904, "grad_norm": 0.45016196370124817, "learning_rate": 0.0005, "loss": 1.368591070175171, "step": 18601 }, { "epoch": 1.4664564446196295, "grad_norm": 0.5622938871383667, "learning_rate": 0.0005, "loss": 1.367490291595459, "step": 18602 }, { "epoch": 1.4665352778872685, "grad_norm": 0.5427884459495544, "learning_rate": 0.0005, "loss": 1.6323573589324951, "step": 18603 }, { "epoch": 1.4666141111549074, "grad_norm": 0.585270881652832, "learning_rate": 0.0005, "loss": 1.4850714206695557, "step": 18604 }, { "epoch": 1.4666929444225463, "grad_norm": 0.5069491267204285, "learning_rate": 0.0005, "loss": 1.5938515663146973, "step": 18605 }, { "epoch": 1.4667717776901852, "grad_norm": 0.6610408425331116, "learning_rate": 0.0005, "loss": 1.481856107711792, "step": 18606 }, { "epoch": 1.4668506109578243, "grad_norm": 0.584933876991272, "learning_rate": 0.0005, "loss": 1.5421794652938843, "step": 18607 }, { "epoch": 1.466929444225463, "grad_norm": 0.6093875169754028, "learning_rate": 0.0005, "loss": 1.5249452590942383, "step": 18608 }, { "epoch": 1.4670082774931021, "grad_norm": 0.6124743819236755, "learning_rate": 0.0005, "loss": 1.5582597255706787, "step": 18609 }, { "epoch": 1.467087110760741, "grad_norm": 0.460452675819397, "learning_rate": 0.0005, "loss": 1.3997340202331543, "step": 18610 }, { "epoch": 1.46716594402838, "grad_norm": 0.6129119396209717, "learning_rate": 0.0005, "loss": 1.6337331533432007, "step": 18611 }, { "epoch": 1.4672447772960189, "grad_norm": 0.5661406517028809, "learning_rate": 0.0005, "loss": 1.769364833831787, "step": 18612 }, { "epoch": 1.4673236105636578, "grad_norm": 0.583329439163208, "learning_rate": 0.0005, "loss": 1.3303742408752441, "step": 18613 }, { "epoch": 1.467402443831297, "grad_norm": 0.44605427980422974, "learning_rate": 0.0005, "loss": 1.4598461389541626, "step": 18614 }, { "epoch": 1.4674812770989358, "grad_norm": 0.5496675372123718, "learning_rate": 0.0005, "loss": 1.5013459920883179, "step": 18615 }, { "epoch": 1.4675601103665747, "grad_norm": 0.621345043182373, "learning_rate": 0.0005, "loss": 1.353905439376831, "step": 18616 }, { "epoch": 1.4676389436342137, "grad_norm": 0.5033307671546936, "learning_rate": 0.0005, "loss": 1.3973380327224731, "step": 18617 }, { "epoch": 1.4677177769018526, "grad_norm": 0.5017857551574707, "learning_rate": 0.0005, "loss": 1.4852168560028076, "step": 18618 }, { "epoch": 1.4677966101694915, "grad_norm": 0.5661218762397766, "learning_rate": 0.0005, "loss": 1.2239692211151123, "step": 18619 }, { "epoch": 1.4678754434371304, "grad_norm": 0.6297494769096375, "learning_rate": 0.0005, "loss": 1.8941212892532349, "step": 18620 }, { "epoch": 1.4679542767047695, "grad_norm": 0.4661872982978821, "learning_rate": 0.0005, "loss": 1.4376273155212402, "step": 18621 }, { "epoch": 1.4680331099724084, "grad_norm": 0.4893653094768524, "learning_rate": 0.0005, "loss": 1.4994080066680908, "step": 18622 }, { "epoch": 1.4681119432400473, "grad_norm": 0.4824410676956177, "learning_rate": 0.0005, "loss": 1.3642009496688843, "step": 18623 }, { "epoch": 1.4681907765076863, "grad_norm": 0.5204107165336609, "learning_rate": 0.0005, "loss": 1.6385552883148193, "step": 18624 }, { "epoch": 1.4682696097753252, "grad_norm": 0.574953019618988, "learning_rate": 0.0005, "loss": 1.5015408992767334, "step": 18625 }, { "epoch": 1.468348443042964, "grad_norm": 0.5146443843841553, "learning_rate": 0.0005, "loss": 1.2971824407577515, "step": 18626 }, { "epoch": 1.468427276310603, "grad_norm": 0.5046496987342834, "learning_rate": 0.0005, "loss": 1.6057744026184082, "step": 18627 }, { "epoch": 1.4685061095782421, "grad_norm": 0.41512808203697205, "learning_rate": 0.0005, "loss": 1.5808755159378052, "step": 18628 }, { "epoch": 1.468584942845881, "grad_norm": 0.4611751139163971, "learning_rate": 0.0005, "loss": 1.6168264150619507, "step": 18629 }, { "epoch": 1.46866377611352, "grad_norm": 0.44088852405548096, "learning_rate": 0.0005, "loss": 1.4786921739578247, "step": 18630 }, { "epoch": 1.4687426093811589, "grad_norm": 0.5217136740684509, "learning_rate": 0.0005, "loss": 1.4877774715423584, "step": 18631 }, { "epoch": 1.4688214426487978, "grad_norm": 0.5539605617523193, "learning_rate": 0.0005, "loss": 1.48087739944458, "step": 18632 }, { "epoch": 1.4689002759164367, "grad_norm": 0.5355449914932251, "learning_rate": 0.0005, "loss": 1.5057510137557983, "step": 18633 }, { "epoch": 1.4689791091840756, "grad_norm": 0.45570915937423706, "learning_rate": 0.0005, "loss": 1.5486698150634766, "step": 18634 }, { "epoch": 1.4690579424517147, "grad_norm": 0.5656583905220032, "learning_rate": 0.0005, "loss": 1.705441951751709, "step": 18635 }, { "epoch": 1.4691367757193536, "grad_norm": 0.6829404234886169, "learning_rate": 0.0005, "loss": 1.628021001815796, "step": 18636 }, { "epoch": 1.4692156089869925, "grad_norm": 0.7555238604545593, "learning_rate": 0.0005, "loss": 1.677377700805664, "step": 18637 }, { "epoch": 1.4692944422546315, "grad_norm": 0.5158718228340149, "learning_rate": 0.0005, "loss": 1.2731351852416992, "step": 18638 }, { "epoch": 1.4693732755222704, "grad_norm": 0.5166403651237488, "learning_rate": 0.0005, "loss": 1.5496602058410645, "step": 18639 }, { "epoch": 1.4694521087899093, "grad_norm": 0.6058884263038635, "learning_rate": 0.0005, "loss": 1.519731044769287, "step": 18640 }, { "epoch": 1.4695309420575482, "grad_norm": 0.5907191634178162, "learning_rate": 0.0005, "loss": 1.509252905845642, "step": 18641 }, { "epoch": 1.4696097753251873, "grad_norm": 1.0423080921173096, "learning_rate": 0.0005, "loss": 1.6552293300628662, "step": 18642 }, { "epoch": 1.4696886085928262, "grad_norm": 0.5354748964309692, "learning_rate": 0.0005, "loss": 1.599258542060852, "step": 18643 }, { "epoch": 1.4697674418604652, "grad_norm": 0.6604129076004028, "learning_rate": 0.0005, "loss": 1.815958023071289, "step": 18644 }, { "epoch": 1.469846275128104, "grad_norm": 0.5159467458724976, "learning_rate": 0.0005, "loss": 1.5456619262695312, "step": 18645 }, { "epoch": 1.469925108395743, "grad_norm": 0.6374120712280273, "learning_rate": 0.0005, "loss": 1.5885775089263916, "step": 18646 }, { "epoch": 1.4700039416633819, "grad_norm": 0.5989954471588135, "learning_rate": 0.0005, "loss": 1.6538347005844116, "step": 18647 }, { "epoch": 1.4700827749310208, "grad_norm": 0.5889043807983398, "learning_rate": 0.0005, "loss": 1.540393590927124, "step": 18648 }, { "epoch": 1.47016160819866, "grad_norm": 0.6607267260551453, "learning_rate": 0.0005, "loss": 1.9582207202911377, "step": 18649 }, { "epoch": 1.4702404414662988, "grad_norm": 0.5487546324729919, "learning_rate": 0.0005, "loss": 1.7281177043914795, "step": 18650 }, { "epoch": 1.4703192747339378, "grad_norm": 0.5585628151893616, "learning_rate": 0.0005, "loss": 1.5901696681976318, "step": 18651 }, { "epoch": 1.4703981080015767, "grad_norm": 0.454790323972702, "learning_rate": 0.0005, "loss": 1.4851887226104736, "step": 18652 }, { "epoch": 1.4704769412692156, "grad_norm": 0.6719042658805847, "learning_rate": 0.0005, "loss": 1.6398577690124512, "step": 18653 }, { "epoch": 1.4705557745368545, "grad_norm": 0.6512759923934937, "learning_rate": 0.0005, "loss": 1.563410758972168, "step": 18654 }, { "epoch": 1.4706346078044934, "grad_norm": 0.5266735553741455, "learning_rate": 0.0005, "loss": 1.4775856733322144, "step": 18655 }, { "epoch": 1.4707134410721325, "grad_norm": 0.5641393065452576, "learning_rate": 0.0005, "loss": 1.525881290435791, "step": 18656 }, { "epoch": 1.4707922743397714, "grad_norm": 0.524372935295105, "learning_rate": 0.0005, "loss": 1.5523484945297241, "step": 18657 }, { "epoch": 1.4708711076074104, "grad_norm": 0.4899701774120331, "learning_rate": 0.0005, "loss": 1.3342056274414062, "step": 18658 }, { "epoch": 1.4709499408750493, "grad_norm": 0.5397632718086243, "learning_rate": 0.0005, "loss": 1.5811947584152222, "step": 18659 }, { "epoch": 1.4710287741426882, "grad_norm": 0.5468121767044067, "learning_rate": 0.0005, "loss": 1.592027187347412, "step": 18660 }, { "epoch": 1.471107607410327, "grad_norm": 0.5400420427322388, "learning_rate": 0.0005, "loss": 1.5286340713500977, "step": 18661 }, { "epoch": 1.471186440677966, "grad_norm": 0.6080054640769958, "learning_rate": 0.0005, "loss": 1.7326772212982178, "step": 18662 }, { "epoch": 1.4712652739456051, "grad_norm": 0.5533917546272278, "learning_rate": 0.0005, "loss": 1.6474758386611938, "step": 18663 }, { "epoch": 1.471344107213244, "grad_norm": 0.7332508563995361, "learning_rate": 0.0005, "loss": 1.7995491027832031, "step": 18664 }, { "epoch": 1.471422940480883, "grad_norm": 0.6475421190261841, "learning_rate": 0.0005, "loss": 1.7876555919647217, "step": 18665 }, { "epoch": 1.4715017737485219, "grad_norm": 0.5010652542114258, "learning_rate": 0.0005, "loss": 1.6622488498687744, "step": 18666 }, { "epoch": 1.4715806070161608, "grad_norm": 0.6260507106781006, "learning_rate": 0.0005, "loss": 1.5391521453857422, "step": 18667 }, { "epoch": 1.4716594402837997, "grad_norm": 0.6079713106155396, "learning_rate": 0.0005, "loss": 1.4711194038391113, "step": 18668 }, { "epoch": 1.4717382735514386, "grad_norm": 0.7505218982696533, "learning_rate": 0.0005, "loss": 1.5498347282409668, "step": 18669 }, { "epoch": 1.4718171068190777, "grad_norm": 0.5653551816940308, "learning_rate": 0.0005, "loss": 1.5077837705612183, "step": 18670 }, { "epoch": 1.4718959400867166, "grad_norm": 0.4811045825481415, "learning_rate": 0.0005, "loss": 1.3986656665802002, "step": 18671 }, { "epoch": 1.4719747733543556, "grad_norm": 0.5649226903915405, "learning_rate": 0.0005, "loss": 1.5241880416870117, "step": 18672 }, { "epoch": 1.4720536066219945, "grad_norm": 0.4938950538635254, "learning_rate": 0.0005, "loss": 1.6901823282241821, "step": 18673 }, { "epoch": 1.4721324398896334, "grad_norm": 0.4481825828552246, "learning_rate": 0.0005, "loss": 1.436638593673706, "step": 18674 }, { "epoch": 1.4722112731572723, "grad_norm": 0.5708146095275879, "learning_rate": 0.0005, "loss": 1.6870572566986084, "step": 18675 }, { "epoch": 1.4722901064249112, "grad_norm": 0.6327770352363586, "learning_rate": 0.0005, "loss": 1.6557989120483398, "step": 18676 }, { "epoch": 1.4723689396925503, "grad_norm": 0.6121146082878113, "learning_rate": 0.0005, "loss": 1.6440458297729492, "step": 18677 }, { "epoch": 1.4724477729601892, "grad_norm": 8.077611923217773, "learning_rate": 0.0005, "loss": 1.4905890226364136, "step": 18678 }, { "epoch": 1.4725266062278282, "grad_norm": 0.6220871210098267, "learning_rate": 0.0005, "loss": 1.725819706916809, "step": 18679 }, { "epoch": 1.472605439495467, "grad_norm": 0.6926801800727844, "learning_rate": 0.0005, "loss": 1.5118839740753174, "step": 18680 }, { "epoch": 1.472684272763106, "grad_norm": 0.43873611092567444, "learning_rate": 0.0005, "loss": 1.3795832395553589, "step": 18681 }, { "epoch": 1.472763106030745, "grad_norm": 0.5521286129951477, "learning_rate": 0.0005, "loss": 1.3958957195281982, "step": 18682 }, { "epoch": 1.4728419392983838, "grad_norm": 0.4245142638683319, "learning_rate": 0.0005, "loss": 1.3943746089935303, "step": 18683 }, { "epoch": 1.472920772566023, "grad_norm": 0.5410659313201904, "learning_rate": 0.0005, "loss": 1.4264510869979858, "step": 18684 }, { "epoch": 1.4729996058336619, "grad_norm": 0.5993654131889343, "learning_rate": 0.0005, "loss": 1.8068742752075195, "step": 18685 }, { "epoch": 1.4730784391013008, "grad_norm": 0.5199418663978577, "learning_rate": 0.0005, "loss": 1.419400930404663, "step": 18686 }, { "epoch": 1.4731572723689397, "grad_norm": 0.577435314655304, "learning_rate": 0.0005, "loss": 1.7050480842590332, "step": 18687 }, { "epoch": 1.4732361056365786, "grad_norm": 0.5293827056884766, "learning_rate": 0.0005, "loss": 1.6311957836151123, "step": 18688 }, { "epoch": 1.4733149389042175, "grad_norm": 0.5371300578117371, "learning_rate": 0.0005, "loss": 1.7518589496612549, "step": 18689 }, { "epoch": 1.4733937721718564, "grad_norm": 0.4728884696960449, "learning_rate": 0.0005, "loss": 1.383728265762329, "step": 18690 }, { "epoch": 1.4734726054394955, "grad_norm": 0.5927107334136963, "learning_rate": 0.0005, "loss": 1.4156208038330078, "step": 18691 }, { "epoch": 1.4735514387071345, "grad_norm": 0.6297469735145569, "learning_rate": 0.0005, "loss": 1.727290153503418, "step": 18692 }, { "epoch": 1.4736302719747734, "grad_norm": 0.5758646726608276, "learning_rate": 0.0005, "loss": 1.5123586654663086, "step": 18693 }, { "epoch": 1.4737091052424123, "grad_norm": 0.48419514298439026, "learning_rate": 0.0005, "loss": 1.6632927656173706, "step": 18694 }, { "epoch": 1.4737879385100512, "grad_norm": 0.5585771203041077, "learning_rate": 0.0005, "loss": 1.5085073709487915, "step": 18695 }, { "epoch": 1.4738667717776903, "grad_norm": 0.4897589385509491, "learning_rate": 0.0005, "loss": 1.4749398231506348, "step": 18696 }, { "epoch": 1.473945605045329, "grad_norm": 0.49155691266059875, "learning_rate": 0.0005, "loss": 1.324842095375061, "step": 18697 }, { "epoch": 1.4740244383129681, "grad_norm": 0.47970134019851685, "learning_rate": 0.0005, "loss": 1.4711627960205078, "step": 18698 }, { "epoch": 1.474103271580607, "grad_norm": 0.49508535861968994, "learning_rate": 0.0005, "loss": 1.6372369527816772, "step": 18699 }, { "epoch": 1.474182104848246, "grad_norm": 0.5666181445121765, "learning_rate": 0.0005, "loss": 1.432633399963379, "step": 18700 }, { "epoch": 1.4742609381158849, "grad_norm": 0.5499024987220764, "learning_rate": 0.0005, "loss": 1.6004695892333984, "step": 18701 }, { "epoch": 1.4743397713835238, "grad_norm": 0.7377043962478638, "learning_rate": 0.0005, "loss": 1.714327096939087, "step": 18702 }, { "epoch": 1.474418604651163, "grad_norm": 0.43991097807884216, "learning_rate": 0.0005, "loss": 1.4133278131484985, "step": 18703 }, { "epoch": 1.4744974379188016, "grad_norm": 0.5437347292900085, "learning_rate": 0.0005, "loss": 1.728104829788208, "step": 18704 }, { "epoch": 1.4745762711864407, "grad_norm": 0.6158232092857361, "learning_rate": 0.0005, "loss": 1.7289296388626099, "step": 18705 }, { "epoch": 1.4746551044540797, "grad_norm": 0.6162159442901611, "learning_rate": 0.0005, "loss": 1.7854177951812744, "step": 18706 }, { "epoch": 1.4747339377217186, "grad_norm": 0.47023338079452515, "learning_rate": 0.0005, "loss": 1.3810453414916992, "step": 18707 }, { "epoch": 1.4748127709893575, "grad_norm": 0.5080119967460632, "learning_rate": 0.0005, "loss": 1.4985580444335938, "step": 18708 }, { "epoch": 1.4748916042569964, "grad_norm": 0.5358078479766846, "learning_rate": 0.0005, "loss": 1.4003740549087524, "step": 18709 }, { "epoch": 1.4749704375246355, "grad_norm": 0.674243688583374, "learning_rate": 0.0005, "loss": 1.7816376686096191, "step": 18710 }, { "epoch": 1.4750492707922742, "grad_norm": 0.5975065231323242, "learning_rate": 0.0005, "loss": 1.7327288389205933, "step": 18711 }, { "epoch": 1.4751281040599133, "grad_norm": 0.5788961052894592, "learning_rate": 0.0005, "loss": 1.5632572174072266, "step": 18712 }, { "epoch": 1.4752069373275523, "grad_norm": 0.6146063208580017, "learning_rate": 0.0005, "loss": 1.262216329574585, "step": 18713 }, { "epoch": 1.4752857705951912, "grad_norm": 0.5947038531303406, "learning_rate": 0.0005, "loss": 1.377497911453247, "step": 18714 }, { "epoch": 1.47536460386283, "grad_norm": 0.5238063335418701, "learning_rate": 0.0005, "loss": 1.5164129734039307, "step": 18715 }, { "epoch": 1.475443437130469, "grad_norm": 0.6450586915016174, "learning_rate": 0.0005, "loss": 1.5429866313934326, "step": 18716 }, { "epoch": 1.4755222703981081, "grad_norm": 0.49126145243644714, "learning_rate": 0.0005, "loss": 1.6106568574905396, "step": 18717 }, { "epoch": 1.4756011036657468, "grad_norm": 0.6852032542228699, "learning_rate": 0.0005, "loss": 1.8248543739318848, "step": 18718 }, { "epoch": 1.475679936933386, "grad_norm": 0.4594644010066986, "learning_rate": 0.0005, "loss": 1.348781943321228, "step": 18719 }, { "epoch": 1.4757587702010249, "grad_norm": 0.5824351906776428, "learning_rate": 0.0005, "loss": 1.6551306247711182, "step": 18720 }, { "epoch": 1.4758376034686638, "grad_norm": 0.6347643136978149, "learning_rate": 0.0005, "loss": 1.6900185346603394, "step": 18721 }, { "epoch": 1.4759164367363027, "grad_norm": 0.5698778033256531, "learning_rate": 0.0005, "loss": 1.560767412185669, "step": 18722 }, { "epoch": 1.4759952700039416, "grad_norm": 0.5474114418029785, "learning_rate": 0.0005, "loss": 1.5113950967788696, "step": 18723 }, { "epoch": 1.4760741032715807, "grad_norm": 0.8355982303619385, "learning_rate": 0.0005, "loss": 1.846712350845337, "step": 18724 }, { "epoch": 1.4761529365392194, "grad_norm": 0.49726006388664246, "learning_rate": 0.0005, "loss": 1.3941489458084106, "step": 18725 }, { "epoch": 1.4762317698068586, "grad_norm": 0.5441770553588867, "learning_rate": 0.0005, "loss": 1.5762275457382202, "step": 18726 }, { "epoch": 1.4763106030744975, "grad_norm": 0.6673865914344788, "learning_rate": 0.0005, "loss": 1.5600554943084717, "step": 18727 }, { "epoch": 1.4763894363421364, "grad_norm": 0.4615429639816284, "learning_rate": 0.0005, "loss": 1.6371500492095947, "step": 18728 }, { "epoch": 1.4764682696097753, "grad_norm": 0.49760693311691284, "learning_rate": 0.0005, "loss": 1.4010082483291626, "step": 18729 }, { "epoch": 1.4765471028774142, "grad_norm": 0.48534199595451355, "learning_rate": 0.0005, "loss": 1.3397748470306396, "step": 18730 }, { "epoch": 1.4766259361450533, "grad_norm": 0.5402879118919373, "learning_rate": 0.0005, "loss": 1.2570470571517944, "step": 18731 }, { "epoch": 1.476704769412692, "grad_norm": 0.6174069046974182, "learning_rate": 0.0005, "loss": 1.718618392944336, "step": 18732 }, { "epoch": 1.4767836026803312, "grad_norm": 0.6158902049064636, "learning_rate": 0.0005, "loss": 1.4433934688568115, "step": 18733 }, { "epoch": 1.47686243594797, "grad_norm": 0.6349669694900513, "learning_rate": 0.0005, "loss": 1.6739411354064941, "step": 18734 }, { "epoch": 1.476941269215609, "grad_norm": 0.4794568419456482, "learning_rate": 0.0005, "loss": 1.3687005043029785, "step": 18735 }, { "epoch": 1.4770201024832479, "grad_norm": 0.5050145387649536, "learning_rate": 0.0005, "loss": 1.5559889078140259, "step": 18736 }, { "epoch": 1.4770989357508868, "grad_norm": 0.6701340675354004, "learning_rate": 0.0005, "loss": 1.621382713317871, "step": 18737 }, { "epoch": 1.477177769018526, "grad_norm": 0.4429433345794678, "learning_rate": 0.0005, "loss": 1.4106388092041016, "step": 18738 }, { "epoch": 1.4772566022861648, "grad_norm": 0.8083549737930298, "learning_rate": 0.0005, "loss": 1.699920415878296, "step": 18739 }, { "epoch": 1.4773354355538038, "grad_norm": 0.5009528398513794, "learning_rate": 0.0005, "loss": 1.274500846862793, "step": 18740 }, { "epoch": 1.4774142688214427, "grad_norm": 0.5655731558799744, "learning_rate": 0.0005, "loss": 1.7418384552001953, "step": 18741 }, { "epoch": 1.4774931020890816, "grad_norm": 0.5330174565315247, "learning_rate": 0.0005, "loss": 1.7071102857589722, "step": 18742 }, { "epoch": 1.4775719353567205, "grad_norm": 0.5254819989204407, "learning_rate": 0.0005, "loss": 1.3765389919281006, "step": 18743 }, { "epoch": 1.4776507686243594, "grad_norm": 0.4710848331451416, "learning_rate": 0.0005, "loss": 1.7627959251403809, "step": 18744 }, { "epoch": 1.4777296018919985, "grad_norm": 0.42708972096443176, "learning_rate": 0.0005, "loss": 1.3243508338928223, "step": 18745 }, { "epoch": 1.4778084351596374, "grad_norm": 0.5347493290901184, "learning_rate": 0.0005, "loss": 1.6601536273956299, "step": 18746 }, { "epoch": 1.4778872684272764, "grad_norm": 0.8730654716491699, "learning_rate": 0.0005, "loss": 1.5415129661560059, "step": 18747 }, { "epoch": 1.4779661016949153, "grad_norm": 0.5046872496604919, "learning_rate": 0.0005, "loss": 1.3900682926177979, "step": 18748 }, { "epoch": 1.4780449349625542, "grad_norm": 0.5555546879768372, "learning_rate": 0.0005, "loss": 1.6545472145080566, "step": 18749 }, { "epoch": 1.478123768230193, "grad_norm": 0.5341812968254089, "learning_rate": 0.0005, "loss": 1.4152967929840088, "step": 18750 }, { "epoch": 1.478202601497832, "grad_norm": 0.5254328846931458, "learning_rate": 0.0005, "loss": 1.5963820219039917, "step": 18751 }, { "epoch": 1.4782814347654711, "grad_norm": 0.5266505479812622, "learning_rate": 0.0005, "loss": 1.8198943138122559, "step": 18752 }, { "epoch": 1.47836026803311, "grad_norm": 0.5616018176078796, "learning_rate": 0.0005, "loss": 1.6395127773284912, "step": 18753 }, { "epoch": 1.478439101300749, "grad_norm": 0.5295567512512207, "learning_rate": 0.0005, "loss": 1.4894394874572754, "step": 18754 }, { "epoch": 1.4785179345683879, "grad_norm": 0.4700465500354767, "learning_rate": 0.0005, "loss": 1.4400612115859985, "step": 18755 }, { "epoch": 1.4785967678360268, "grad_norm": 0.6325626373291016, "learning_rate": 0.0005, "loss": 1.7267340421676636, "step": 18756 }, { "epoch": 1.4786756011036657, "grad_norm": 0.5462515950202942, "learning_rate": 0.0005, "loss": 1.5803784132003784, "step": 18757 }, { "epoch": 1.4787544343713046, "grad_norm": 0.4097168445587158, "learning_rate": 0.0005, "loss": 1.415894865989685, "step": 18758 }, { "epoch": 1.4788332676389437, "grad_norm": 0.5026829242706299, "learning_rate": 0.0005, "loss": 1.6804182529449463, "step": 18759 }, { "epoch": 1.4789121009065826, "grad_norm": 0.5941232442855835, "learning_rate": 0.0005, "loss": 1.6219968795776367, "step": 18760 }, { "epoch": 1.4789909341742216, "grad_norm": 0.6206104159355164, "learning_rate": 0.0005, "loss": 1.7390103340148926, "step": 18761 }, { "epoch": 1.4790697674418605, "grad_norm": 0.5426709651947021, "learning_rate": 0.0005, "loss": 1.551713466644287, "step": 18762 }, { "epoch": 1.4791486007094994, "grad_norm": 0.6977813243865967, "learning_rate": 0.0005, "loss": 1.6534152030944824, "step": 18763 }, { "epoch": 1.4792274339771383, "grad_norm": 0.5933881998062134, "learning_rate": 0.0005, "loss": 1.5163465738296509, "step": 18764 }, { "epoch": 1.4793062672447772, "grad_norm": 1.0507673025131226, "learning_rate": 0.0005, "loss": 1.8412539958953857, "step": 18765 }, { "epoch": 1.4793851005124163, "grad_norm": 0.5754284858703613, "learning_rate": 0.0005, "loss": 1.4746991395950317, "step": 18766 }, { "epoch": 1.4794639337800553, "grad_norm": 0.597541332244873, "learning_rate": 0.0005, "loss": 1.3845274448394775, "step": 18767 }, { "epoch": 1.4795427670476942, "grad_norm": 0.5217368006706238, "learning_rate": 0.0005, "loss": 1.3834576606750488, "step": 18768 }, { "epoch": 1.479621600315333, "grad_norm": 0.5588569641113281, "learning_rate": 0.0005, "loss": 1.766450047492981, "step": 18769 }, { "epoch": 1.479700433582972, "grad_norm": 0.5675318241119385, "learning_rate": 0.0005, "loss": 1.5913560390472412, "step": 18770 }, { "epoch": 1.479779266850611, "grad_norm": 0.48032352328300476, "learning_rate": 0.0005, "loss": 1.5341850519180298, "step": 18771 }, { "epoch": 1.4798581001182498, "grad_norm": 0.5014305710792542, "learning_rate": 0.0005, "loss": 1.5651650428771973, "step": 18772 }, { "epoch": 1.479936933385889, "grad_norm": 0.48887452483177185, "learning_rate": 0.0005, "loss": 1.6601625680923462, "step": 18773 }, { "epoch": 1.4800157666535279, "grad_norm": 0.5403523445129395, "learning_rate": 0.0005, "loss": 1.5549793243408203, "step": 18774 }, { "epoch": 1.4800945999211668, "grad_norm": 0.5213795304298401, "learning_rate": 0.0005, "loss": 1.3865461349487305, "step": 18775 }, { "epoch": 1.4801734331888057, "grad_norm": 0.5362594127655029, "learning_rate": 0.0005, "loss": 1.5748473405838013, "step": 18776 }, { "epoch": 1.4802522664564446, "grad_norm": 0.45277684926986694, "learning_rate": 0.0005, "loss": 1.5902609825134277, "step": 18777 }, { "epoch": 1.4803310997240835, "grad_norm": 0.5334819555282593, "learning_rate": 0.0005, "loss": 1.505707025527954, "step": 18778 }, { "epoch": 1.4804099329917224, "grad_norm": 0.5889208912849426, "learning_rate": 0.0005, "loss": 1.63484525680542, "step": 18779 }, { "epoch": 1.4804887662593615, "grad_norm": 0.5083793997764587, "learning_rate": 0.0005, "loss": 1.6153165102005005, "step": 18780 }, { "epoch": 1.4805675995270005, "grad_norm": 0.7108519077301025, "learning_rate": 0.0005, "loss": 1.502827525138855, "step": 18781 }, { "epoch": 1.4806464327946394, "grad_norm": 0.48816099762916565, "learning_rate": 0.0005, "loss": 1.6066457033157349, "step": 18782 }, { "epoch": 1.4807252660622783, "grad_norm": 0.49083203077316284, "learning_rate": 0.0005, "loss": 1.5302321910858154, "step": 18783 }, { "epoch": 1.4808040993299172, "grad_norm": 0.5065883994102478, "learning_rate": 0.0005, "loss": 1.4135477542877197, "step": 18784 }, { "epoch": 1.480882932597556, "grad_norm": 0.6892670392990112, "learning_rate": 0.0005, "loss": 1.364311695098877, "step": 18785 }, { "epoch": 1.480961765865195, "grad_norm": 0.6169010996818542, "learning_rate": 0.0005, "loss": 1.6114537715911865, "step": 18786 }, { "epoch": 1.4810405991328341, "grad_norm": 0.5152228474617004, "learning_rate": 0.0005, "loss": 1.5024123191833496, "step": 18787 }, { "epoch": 1.481119432400473, "grad_norm": 0.7218847274780273, "learning_rate": 0.0005, "loss": 1.456569790840149, "step": 18788 }, { "epoch": 1.481198265668112, "grad_norm": 1.2828353643417358, "learning_rate": 0.0005, "loss": 1.5685341358184814, "step": 18789 }, { "epoch": 1.4812770989357509, "grad_norm": 0.40143531560897827, "learning_rate": 0.0005, "loss": 1.2614401578903198, "step": 18790 }, { "epoch": 1.4813559322033898, "grad_norm": 0.6043359637260437, "learning_rate": 0.0005, "loss": 1.6237175464630127, "step": 18791 }, { "epoch": 1.4814347654710287, "grad_norm": 0.5233984589576721, "learning_rate": 0.0005, "loss": 1.4077755212783813, "step": 18792 }, { "epoch": 1.4815135987386676, "grad_norm": 0.6173467040061951, "learning_rate": 0.0005, "loss": 1.7875456809997559, "step": 18793 }, { "epoch": 1.4815924320063067, "grad_norm": 0.5500341653823853, "learning_rate": 0.0005, "loss": 1.3515675067901611, "step": 18794 }, { "epoch": 1.4816712652739457, "grad_norm": 0.5177299380302429, "learning_rate": 0.0005, "loss": 1.4650611877441406, "step": 18795 }, { "epoch": 1.4817500985415846, "grad_norm": 0.6201850175857544, "learning_rate": 0.0005, "loss": 1.7620129585266113, "step": 18796 }, { "epoch": 1.4818289318092235, "grad_norm": 0.6369944214820862, "learning_rate": 0.0005, "loss": 1.689692497253418, "step": 18797 }, { "epoch": 1.4819077650768624, "grad_norm": 0.6363118290901184, "learning_rate": 0.0005, "loss": 1.6747660636901855, "step": 18798 }, { "epoch": 1.4819865983445013, "grad_norm": 0.7394455075263977, "learning_rate": 0.0005, "loss": 1.562653660774231, "step": 18799 }, { "epoch": 1.4820654316121402, "grad_norm": 0.4883328080177307, "learning_rate": 0.0005, "loss": 1.4429938793182373, "step": 18800 }, { "epoch": 1.4821442648797793, "grad_norm": 0.6064762473106384, "learning_rate": 0.0005, "loss": 1.3874754905700684, "step": 18801 }, { "epoch": 1.4822230981474183, "grad_norm": 0.5342431664466858, "learning_rate": 0.0005, "loss": 1.4058616161346436, "step": 18802 }, { "epoch": 1.4823019314150572, "grad_norm": 0.5336841344833374, "learning_rate": 0.0005, "loss": 1.5691063404083252, "step": 18803 }, { "epoch": 1.482380764682696, "grad_norm": 0.5077459216117859, "learning_rate": 0.0005, "loss": 1.4616329669952393, "step": 18804 }, { "epoch": 1.482459597950335, "grad_norm": 0.6171361207962036, "learning_rate": 0.0005, "loss": 1.7328397035598755, "step": 18805 }, { "epoch": 1.482538431217974, "grad_norm": 0.47130343317985535, "learning_rate": 0.0005, "loss": 1.3543236255645752, "step": 18806 }, { "epoch": 1.4826172644856128, "grad_norm": 0.7924461960792542, "learning_rate": 0.0005, "loss": 1.707050085067749, "step": 18807 }, { "epoch": 1.482696097753252, "grad_norm": 0.7344732880592346, "learning_rate": 0.0005, "loss": 1.453406572341919, "step": 18808 }, { "epoch": 1.4827749310208909, "grad_norm": 0.498478502035141, "learning_rate": 0.0005, "loss": 1.4046469926834106, "step": 18809 }, { "epoch": 1.4828537642885298, "grad_norm": 0.6211154460906982, "learning_rate": 0.0005, "loss": 1.587603211402893, "step": 18810 }, { "epoch": 1.4829325975561687, "grad_norm": 0.5538830757141113, "learning_rate": 0.0005, "loss": 1.6375489234924316, "step": 18811 }, { "epoch": 1.4830114308238076, "grad_norm": 0.5671169757843018, "learning_rate": 0.0005, "loss": 1.2953583002090454, "step": 18812 }, { "epoch": 1.4830902640914465, "grad_norm": 0.5965685248374939, "learning_rate": 0.0005, "loss": 1.6744086742401123, "step": 18813 }, { "epoch": 1.4831690973590854, "grad_norm": 0.544318437576294, "learning_rate": 0.0005, "loss": 1.4843275547027588, "step": 18814 }, { "epoch": 1.4832479306267246, "grad_norm": 0.5974894165992737, "learning_rate": 0.0005, "loss": 1.9286653995513916, "step": 18815 }, { "epoch": 1.4833267638943635, "grad_norm": 0.5931473970413208, "learning_rate": 0.0005, "loss": 1.4454740285873413, "step": 18816 }, { "epoch": 1.4834055971620024, "grad_norm": 0.6247178912162781, "learning_rate": 0.0005, "loss": 1.8156061172485352, "step": 18817 }, { "epoch": 1.4834844304296413, "grad_norm": 0.4656178653240204, "learning_rate": 0.0005, "loss": 1.6961675882339478, "step": 18818 }, { "epoch": 1.4835632636972802, "grad_norm": 0.43854689598083496, "learning_rate": 0.0005, "loss": 1.2211089134216309, "step": 18819 }, { "epoch": 1.4836420969649193, "grad_norm": 0.4171479642391205, "learning_rate": 0.0005, "loss": 1.2068414688110352, "step": 18820 }, { "epoch": 1.483720930232558, "grad_norm": 0.6019686460494995, "learning_rate": 0.0005, "loss": 2.065075397491455, "step": 18821 }, { "epoch": 1.4837997635001972, "grad_norm": 0.5586364269256592, "learning_rate": 0.0005, "loss": 1.6714519262313843, "step": 18822 }, { "epoch": 1.483878596767836, "grad_norm": 0.6591013669967651, "learning_rate": 0.0005, "loss": 1.7726491689682007, "step": 18823 }, { "epoch": 1.483957430035475, "grad_norm": 0.5044469237327576, "learning_rate": 0.0005, "loss": 1.4520976543426514, "step": 18824 }, { "epoch": 1.484036263303114, "grad_norm": 0.6993827223777771, "learning_rate": 0.0005, "loss": 1.5695290565490723, "step": 18825 }, { "epoch": 1.4841150965707528, "grad_norm": 0.5752113461494446, "learning_rate": 0.0005, "loss": 1.5642786026000977, "step": 18826 }, { "epoch": 1.484193929838392, "grad_norm": 0.4606107175350189, "learning_rate": 0.0005, "loss": 1.620469331741333, "step": 18827 }, { "epoch": 1.4842727631060306, "grad_norm": 0.6489894390106201, "learning_rate": 0.0005, "loss": 1.5468264818191528, "step": 18828 }, { "epoch": 1.4843515963736698, "grad_norm": 0.5551856160163879, "learning_rate": 0.0005, "loss": 1.8142036199569702, "step": 18829 }, { "epoch": 1.4844304296413087, "grad_norm": 0.5267294049263, "learning_rate": 0.0005, "loss": 1.5523107051849365, "step": 18830 }, { "epoch": 1.4845092629089476, "grad_norm": 0.6777371764183044, "learning_rate": 0.0005, "loss": 1.4365204572677612, "step": 18831 }, { "epoch": 1.4845880961765865, "grad_norm": 0.6445501446723938, "learning_rate": 0.0005, "loss": 1.6372299194335938, "step": 18832 }, { "epoch": 1.4846669294442254, "grad_norm": 0.5764298439025879, "learning_rate": 0.0005, "loss": 1.6386138200759888, "step": 18833 }, { "epoch": 1.4847457627118645, "grad_norm": 0.517131507396698, "learning_rate": 0.0005, "loss": 1.6047550439834595, "step": 18834 }, { "epoch": 1.4848245959795032, "grad_norm": 0.5486319661140442, "learning_rate": 0.0005, "loss": 1.431877851486206, "step": 18835 }, { "epoch": 1.4849034292471424, "grad_norm": 0.6605930924415588, "learning_rate": 0.0005, "loss": 1.3292080163955688, "step": 18836 }, { "epoch": 1.4849822625147813, "grad_norm": 0.5474550127983093, "learning_rate": 0.0005, "loss": 1.4257482290267944, "step": 18837 }, { "epoch": 1.4850610957824202, "grad_norm": 0.5559670925140381, "learning_rate": 0.0005, "loss": 1.689076542854309, "step": 18838 }, { "epoch": 1.485139929050059, "grad_norm": 0.6061447858810425, "learning_rate": 0.0005, "loss": 1.64093017578125, "step": 18839 }, { "epoch": 1.485218762317698, "grad_norm": 0.5611516237258911, "learning_rate": 0.0005, "loss": 1.5806564092636108, "step": 18840 }, { "epoch": 1.4852975955853371, "grad_norm": 0.6080458164215088, "learning_rate": 0.0005, "loss": 1.7194359302520752, "step": 18841 }, { "epoch": 1.4853764288529758, "grad_norm": 0.5220870971679688, "learning_rate": 0.0005, "loss": 1.4675347805023193, "step": 18842 }, { "epoch": 1.485455262120615, "grad_norm": 0.4984515607357025, "learning_rate": 0.0005, "loss": 1.4430136680603027, "step": 18843 }, { "epoch": 1.4855340953882539, "grad_norm": 0.629084050655365, "learning_rate": 0.0005, "loss": 1.5906238555908203, "step": 18844 }, { "epoch": 1.4856129286558928, "grad_norm": 0.45475107431411743, "learning_rate": 0.0005, "loss": 1.323914647102356, "step": 18845 }, { "epoch": 1.4856917619235317, "grad_norm": 0.47816625237464905, "learning_rate": 0.0005, "loss": 1.4438252449035645, "step": 18846 }, { "epoch": 1.4857705951911706, "grad_norm": 0.6289719939231873, "learning_rate": 0.0005, "loss": 1.461321234703064, "step": 18847 }, { "epoch": 1.4858494284588097, "grad_norm": 0.511318564414978, "learning_rate": 0.0005, "loss": 1.5174237489700317, "step": 18848 }, { "epoch": 1.4859282617264484, "grad_norm": 0.5118265748023987, "learning_rate": 0.0005, "loss": 1.5804569721221924, "step": 18849 }, { "epoch": 1.4860070949940876, "grad_norm": 0.4792841076850891, "learning_rate": 0.0005, "loss": 1.5199347734451294, "step": 18850 }, { "epoch": 1.4860859282617265, "grad_norm": 2.1887784004211426, "learning_rate": 0.0005, "loss": 1.615092158317566, "step": 18851 }, { "epoch": 1.4861647615293654, "grad_norm": 0.6089398264884949, "learning_rate": 0.0005, "loss": 1.5103106498718262, "step": 18852 }, { "epoch": 1.4862435947970043, "grad_norm": 0.4904678463935852, "learning_rate": 0.0005, "loss": 1.488015055656433, "step": 18853 }, { "epoch": 1.4863224280646432, "grad_norm": 0.6788133978843689, "learning_rate": 0.0005, "loss": 1.5802576541900635, "step": 18854 }, { "epoch": 1.4864012613322823, "grad_norm": 0.8103557825088501, "learning_rate": 0.0005, "loss": 1.654723882675171, "step": 18855 }, { "epoch": 1.486480094599921, "grad_norm": 0.8779900670051575, "learning_rate": 0.0005, "loss": 1.7670702934265137, "step": 18856 }, { "epoch": 1.4865589278675602, "grad_norm": 0.71710604429245, "learning_rate": 0.0005, "loss": 1.62647545337677, "step": 18857 }, { "epoch": 1.486637761135199, "grad_norm": 0.677033543586731, "learning_rate": 0.0005, "loss": 1.4572640657424927, "step": 18858 }, { "epoch": 1.486716594402838, "grad_norm": 0.61747807264328, "learning_rate": 0.0005, "loss": 1.6367902755737305, "step": 18859 }, { "epoch": 1.486795427670477, "grad_norm": 0.5597173571586609, "learning_rate": 0.0005, "loss": 1.5892763137817383, "step": 18860 }, { "epoch": 1.4868742609381158, "grad_norm": 1.7539052963256836, "learning_rate": 0.0005, "loss": 1.7508816719055176, "step": 18861 }, { "epoch": 1.486953094205755, "grad_norm": 0.49012690782546997, "learning_rate": 0.0005, "loss": 1.443648338317871, "step": 18862 }, { "epoch": 1.4870319274733939, "grad_norm": 0.3590208888053894, "learning_rate": 0.0005, "loss": 1.2874176502227783, "step": 18863 }, { "epoch": 1.4871107607410328, "grad_norm": 0.6164718270301819, "learning_rate": 0.0005, "loss": 1.4729712009429932, "step": 18864 }, { "epoch": 1.4871895940086717, "grad_norm": 0.583010733127594, "learning_rate": 0.0005, "loss": 1.5404139757156372, "step": 18865 }, { "epoch": 1.4872684272763106, "grad_norm": 0.5150565505027771, "learning_rate": 0.0005, "loss": 1.4693195819854736, "step": 18866 }, { "epoch": 1.4873472605439495, "grad_norm": 0.5625718832015991, "learning_rate": 0.0005, "loss": 1.400557279586792, "step": 18867 }, { "epoch": 1.4874260938115884, "grad_norm": 0.6306213736534119, "learning_rate": 0.0005, "loss": 1.6540534496307373, "step": 18868 }, { "epoch": 1.4875049270792275, "grad_norm": 0.6331478357315063, "learning_rate": 0.0005, "loss": 1.5856192111968994, "step": 18869 }, { "epoch": 1.4875837603468665, "grad_norm": 0.5065707564353943, "learning_rate": 0.0005, "loss": 1.516959547996521, "step": 18870 }, { "epoch": 1.4876625936145054, "grad_norm": 0.6467381715774536, "learning_rate": 0.0005, "loss": 1.6196225881576538, "step": 18871 }, { "epoch": 1.4877414268821443, "grad_norm": 0.737778902053833, "learning_rate": 0.0005, "loss": 1.7074429988861084, "step": 18872 }, { "epoch": 1.4878202601497832, "grad_norm": 0.7867509722709656, "learning_rate": 0.0005, "loss": 1.597981333732605, "step": 18873 }, { "epoch": 1.487899093417422, "grad_norm": 0.6181394457817078, "learning_rate": 0.0005, "loss": 1.583282709121704, "step": 18874 }, { "epoch": 1.487977926685061, "grad_norm": 0.6019860506057739, "learning_rate": 0.0005, "loss": 1.3863356113433838, "step": 18875 }, { "epoch": 1.4880567599527001, "grad_norm": 2.8916220664978027, "learning_rate": 0.0005, "loss": 1.430659532546997, "step": 18876 }, { "epoch": 1.488135593220339, "grad_norm": 0.627650260925293, "learning_rate": 0.0005, "loss": 1.6092617511749268, "step": 18877 }, { "epoch": 1.488214426487978, "grad_norm": 0.4860258996486664, "learning_rate": 0.0005, "loss": 1.4869219064712524, "step": 18878 }, { "epoch": 1.4882932597556169, "grad_norm": 0.45637246966362, "learning_rate": 0.0005, "loss": 1.4507253170013428, "step": 18879 }, { "epoch": 1.4883720930232558, "grad_norm": 0.6174437999725342, "learning_rate": 0.0005, "loss": 1.654125690460205, "step": 18880 }, { "epoch": 1.4884509262908947, "grad_norm": 0.5661627054214478, "learning_rate": 0.0005, "loss": 1.4918431043624878, "step": 18881 }, { "epoch": 1.4885297595585336, "grad_norm": 0.5645570158958435, "learning_rate": 0.0005, "loss": 1.6807610988616943, "step": 18882 }, { "epoch": 1.4886085928261727, "grad_norm": 0.6210455894470215, "learning_rate": 0.0005, "loss": 1.7436509132385254, "step": 18883 }, { "epoch": 1.4886874260938117, "grad_norm": 0.7096271514892578, "learning_rate": 0.0005, "loss": 1.4411523342132568, "step": 18884 }, { "epoch": 1.4887662593614506, "grad_norm": 0.5255767107009888, "learning_rate": 0.0005, "loss": 1.4846556186676025, "step": 18885 }, { "epoch": 1.4888450926290895, "grad_norm": 1.1674028635025024, "learning_rate": 0.0005, "loss": 1.6786777973175049, "step": 18886 }, { "epoch": 1.4889239258967284, "grad_norm": 0.5596100687980652, "learning_rate": 0.0005, "loss": 1.5567654371261597, "step": 18887 }, { "epoch": 1.4890027591643673, "grad_norm": 0.771895706653595, "learning_rate": 0.0005, "loss": 1.3803534507751465, "step": 18888 }, { "epoch": 1.4890815924320062, "grad_norm": 0.5990436673164368, "learning_rate": 0.0005, "loss": 1.6507465839385986, "step": 18889 }, { "epoch": 1.4891604256996454, "grad_norm": 0.5787796378135681, "learning_rate": 0.0005, "loss": 1.651247262954712, "step": 18890 }, { "epoch": 1.4892392589672843, "grad_norm": 0.5184265971183777, "learning_rate": 0.0005, "loss": 1.4156078100204468, "step": 18891 }, { "epoch": 1.4893180922349232, "grad_norm": 0.8353110551834106, "learning_rate": 0.0005, "loss": 1.5413389205932617, "step": 18892 }, { "epoch": 1.489396925502562, "grad_norm": 1.118077278137207, "learning_rate": 0.0005, "loss": 1.3880739212036133, "step": 18893 }, { "epoch": 1.489475758770201, "grad_norm": 0.5461457967758179, "learning_rate": 0.0005, "loss": 1.6246533393859863, "step": 18894 }, { "epoch": 1.48955459203784, "grad_norm": 0.6604285836219788, "learning_rate": 0.0005, "loss": 1.5601425170898438, "step": 18895 }, { "epoch": 1.4896334253054788, "grad_norm": 0.6612204313278198, "learning_rate": 0.0005, "loss": 1.7244762182235718, "step": 18896 }, { "epoch": 1.489712258573118, "grad_norm": 0.4972551465034485, "learning_rate": 0.0005, "loss": 1.4974514245986938, "step": 18897 }, { "epoch": 1.4897910918407569, "grad_norm": 0.5454086661338806, "learning_rate": 0.0005, "loss": 1.6395106315612793, "step": 18898 }, { "epoch": 1.4898699251083958, "grad_norm": 0.5865721702575684, "learning_rate": 0.0005, "loss": 1.611452579498291, "step": 18899 }, { "epoch": 1.4899487583760347, "grad_norm": 0.4195939302444458, "learning_rate": 0.0005, "loss": 1.5445330142974854, "step": 18900 }, { "epoch": 1.4900275916436736, "grad_norm": 0.5126273036003113, "learning_rate": 0.0005, "loss": 1.664971947669983, "step": 18901 }, { "epoch": 1.4901064249113125, "grad_norm": 0.5012211799621582, "learning_rate": 0.0005, "loss": 1.4705156087875366, "step": 18902 }, { "epoch": 1.4901852581789514, "grad_norm": 0.5857654213905334, "learning_rate": 0.0005, "loss": 1.5319404602050781, "step": 18903 }, { "epoch": 1.4902640914465906, "grad_norm": 0.648891031742096, "learning_rate": 0.0005, "loss": 1.5496788024902344, "step": 18904 }, { "epoch": 1.4903429247142295, "grad_norm": 0.5236276984214783, "learning_rate": 0.0005, "loss": 1.793349266052246, "step": 18905 }, { "epoch": 1.4904217579818684, "grad_norm": 0.46766600012779236, "learning_rate": 0.0005, "loss": 1.3248025178909302, "step": 18906 }, { "epoch": 1.4905005912495073, "grad_norm": 0.6670910716056824, "learning_rate": 0.0005, "loss": 1.5130438804626465, "step": 18907 }, { "epoch": 1.4905794245171462, "grad_norm": 0.779711902141571, "learning_rate": 0.0005, "loss": 1.6510276794433594, "step": 18908 }, { "epoch": 1.4906582577847851, "grad_norm": 0.47427839040756226, "learning_rate": 0.0005, "loss": 1.4149179458618164, "step": 18909 }, { "epoch": 1.490737091052424, "grad_norm": 0.675178587436676, "learning_rate": 0.0005, "loss": 1.5925347805023193, "step": 18910 }, { "epoch": 1.4908159243200632, "grad_norm": 0.601211667060852, "learning_rate": 0.0005, "loss": 1.762142300605774, "step": 18911 }, { "epoch": 1.490894757587702, "grad_norm": 0.6307711005210876, "learning_rate": 0.0005, "loss": 1.7199344635009766, "step": 18912 }, { "epoch": 1.490973590855341, "grad_norm": 0.9268468618392944, "learning_rate": 0.0005, "loss": 1.494646430015564, "step": 18913 }, { "epoch": 1.49105242412298, "grad_norm": 0.5201156735420227, "learning_rate": 0.0005, "loss": 1.5858898162841797, "step": 18914 }, { "epoch": 1.4911312573906188, "grad_norm": 0.47462862730026245, "learning_rate": 0.0005, "loss": 1.5012879371643066, "step": 18915 }, { "epoch": 1.4912100906582577, "grad_norm": 0.618753969669342, "learning_rate": 0.0005, "loss": 1.5767097473144531, "step": 18916 }, { "epoch": 1.4912889239258966, "grad_norm": 0.4816386103630066, "learning_rate": 0.0005, "loss": 1.237365961074829, "step": 18917 }, { "epoch": 1.4913677571935358, "grad_norm": 0.6417741775512695, "learning_rate": 0.0005, "loss": 1.6019477844238281, "step": 18918 }, { "epoch": 1.4914465904611747, "grad_norm": 0.6307916045188904, "learning_rate": 0.0005, "loss": 1.7557826042175293, "step": 18919 }, { "epoch": 1.4915254237288136, "grad_norm": 0.6926873922348022, "learning_rate": 0.0005, "loss": 1.6227965354919434, "step": 18920 }, { "epoch": 1.4916042569964525, "grad_norm": 0.5657250881195068, "learning_rate": 0.0005, "loss": 1.5226507186889648, "step": 18921 }, { "epoch": 1.4916830902640914, "grad_norm": 0.683236837387085, "learning_rate": 0.0005, "loss": 1.3839181661605835, "step": 18922 }, { "epoch": 1.4917619235317303, "grad_norm": 0.42497119307518005, "learning_rate": 0.0005, "loss": 1.4141407012939453, "step": 18923 }, { "epoch": 1.4918407567993692, "grad_norm": 0.5662760138511658, "learning_rate": 0.0005, "loss": 1.6644477844238281, "step": 18924 }, { "epoch": 1.4919195900670084, "grad_norm": 0.498881071805954, "learning_rate": 0.0005, "loss": 1.5006120204925537, "step": 18925 }, { "epoch": 1.4919984233346473, "grad_norm": 0.5474951267242432, "learning_rate": 0.0005, "loss": 1.623252511024475, "step": 18926 }, { "epoch": 1.4920772566022862, "grad_norm": 0.46463829278945923, "learning_rate": 0.0005, "loss": 1.375083088874817, "step": 18927 }, { "epoch": 1.492156089869925, "grad_norm": 0.6317634582519531, "learning_rate": 0.0005, "loss": 1.7152917385101318, "step": 18928 }, { "epoch": 1.492234923137564, "grad_norm": 0.6163399815559387, "learning_rate": 0.0005, "loss": 1.5992703437805176, "step": 18929 }, { "epoch": 1.492313756405203, "grad_norm": 0.639386773109436, "learning_rate": 0.0005, "loss": 1.4841432571411133, "step": 18930 }, { "epoch": 1.4923925896728418, "grad_norm": 0.6671057343482971, "learning_rate": 0.0005, "loss": 1.6163029670715332, "step": 18931 }, { "epoch": 1.492471422940481, "grad_norm": 0.4792271852493286, "learning_rate": 0.0005, "loss": 1.4868402481079102, "step": 18932 }, { "epoch": 1.4925502562081199, "grad_norm": 0.5536367893218994, "learning_rate": 0.0005, "loss": 1.3936831951141357, "step": 18933 }, { "epoch": 1.4926290894757588, "grad_norm": 0.4711001515388489, "learning_rate": 0.0005, "loss": 1.438993215560913, "step": 18934 }, { "epoch": 1.4927079227433977, "grad_norm": 0.41733992099761963, "learning_rate": 0.0005, "loss": 1.2754392623901367, "step": 18935 }, { "epoch": 1.4927867560110366, "grad_norm": 0.6740659475326538, "learning_rate": 0.0005, "loss": 1.5683221817016602, "step": 18936 }, { "epoch": 1.4928655892786755, "grad_norm": 0.57024085521698, "learning_rate": 0.0005, "loss": 1.4297199249267578, "step": 18937 }, { "epoch": 1.4929444225463144, "grad_norm": 0.5850717425346375, "learning_rate": 0.0005, "loss": 1.5672138929367065, "step": 18938 }, { "epoch": 1.4930232558139536, "grad_norm": 0.518185555934906, "learning_rate": 0.0005, "loss": 1.703260898590088, "step": 18939 }, { "epoch": 1.4931020890815925, "grad_norm": 0.5578528046607971, "learning_rate": 0.0005, "loss": 1.6513628959655762, "step": 18940 }, { "epoch": 1.4931809223492314, "grad_norm": 1.549047827720642, "learning_rate": 0.0005, "loss": 1.3886451721191406, "step": 18941 }, { "epoch": 1.4932597556168703, "grad_norm": 0.5190911293029785, "learning_rate": 0.0005, "loss": 1.586730718612671, "step": 18942 }, { "epoch": 1.4933385888845092, "grad_norm": 0.6096680760383606, "learning_rate": 0.0005, "loss": 1.6556832790374756, "step": 18943 }, { "epoch": 1.4934174221521483, "grad_norm": 0.656189501285553, "learning_rate": 0.0005, "loss": 1.6010010242462158, "step": 18944 }, { "epoch": 1.493496255419787, "grad_norm": 0.5754037499427795, "learning_rate": 0.0005, "loss": 1.597303867340088, "step": 18945 }, { "epoch": 1.4935750886874262, "grad_norm": 0.47335952520370483, "learning_rate": 0.0005, "loss": 1.4528309106826782, "step": 18946 }, { "epoch": 1.493653921955065, "grad_norm": 0.6350466012954712, "learning_rate": 0.0005, "loss": 1.6172735691070557, "step": 18947 }, { "epoch": 1.493732755222704, "grad_norm": 0.492086261510849, "learning_rate": 0.0005, "loss": 1.64137864112854, "step": 18948 }, { "epoch": 1.493811588490343, "grad_norm": 0.4762728214263916, "learning_rate": 0.0005, "loss": 1.4093345403671265, "step": 18949 }, { "epoch": 1.4938904217579818, "grad_norm": 0.5273800492286682, "learning_rate": 0.0005, "loss": 1.5556327104568481, "step": 18950 }, { "epoch": 1.493969255025621, "grad_norm": 0.6411032676696777, "learning_rate": 0.0005, "loss": 1.5331387519836426, "step": 18951 }, { "epoch": 1.4940480882932596, "grad_norm": 0.5700822472572327, "learning_rate": 0.0005, "loss": 1.7476521730422974, "step": 18952 }, { "epoch": 1.4941269215608988, "grad_norm": 0.5429718494415283, "learning_rate": 0.0005, "loss": 1.384004831314087, "step": 18953 }, { "epoch": 1.4942057548285377, "grad_norm": 0.9104993343353271, "learning_rate": 0.0005, "loss": 1.8245664834976196, "step": 18954 }, { "epoch": 1.4942845880961766, "grad_norm": 0.5279613137245178, "learning_rate": 0.0005, "loss": 1.3894857168197632, "step": 18955 }, { "epoch": 1.4943634213638155, "grad_norm": 0.528190016746521, "learning_rate": 0.0005, "loss": 1.5954402685165405, "step": 18956 }, { "epoch": 1.4944422546314544, "grad_norm": 0.49243417382240295, "learning_rate": 0.0005, "loss": 1.3588918447494507, "step": 18957 }, { "epoch": 1.4945210878990935, "grad_norm": 0.6104783415794373, "learning_rate": 0.0005, "loss": 1.845964789390564, "step": 18958 }, { "epoch": 1.4945999211667322, "grad_norm": 0.540144681930542, "learning_rate": 0.0005, "loss": 1.5808436870574951, "step": 18959 }, { "epoch": 1.4946787544343714, "grad_norm": 0.47859182953834534, "learning_rate": 0.0005, "loss": 1.4697391986846924, "step": 18960 }, { "epoch": 1.4947575877020103, "grad_norm": 0.5454341769218445, "learning_rate": 0.0005, "loss": 1.5735394954681396, "step": 18961 }, { "epoch": 1.4948364209696492, "grad_norm": 0.5761973261833191, "learning_rate": 0.0005, "loss": 1.6170074939727783, "step": 18962 }, { "epoch": 1.494915254237288, "grad_norm": 0.526823878288269, "learning_rate": 0.0005, "loss": 1.5651394128799438, "step": 18963 }, { "epoch": 1.494994087504927, "grad_norm": 0.7284742593765259, "learning_rate": 0.0005, "loss": 1.6455259323120117, "step": 18964 }, { "epoch": 1.4950729207725661, "grad_norm": 0.5637213587760925, "learning_rate": 0.0005, "loss": 1.6667613983154297, "step": 18965 }, { "epoch": 1.4951517540402048, "grad_norm": 0.5572342276573181, "learning_rate": 0.0005, "loss": 1.3475689888000488, "step": 18966 }, { "epoch": 1.495230587307844, "grad_norm": 0.6554843187332153, "learning_rate": 0.0005, "loss": 1.6037366390228271, "step": 18967 }, { "epoch": 1.4953094205754829, "grad_norm": 0.4746939241886139, "learning_rate": 0.0005, "loss": 1.536975622177124, "step": 18968 }, { "epoch": 1.4953882538431218, "grad_norm": 0.5602352619171143, "learning_rate": 0.0005, "loss": 1.452653169631958, "step": 18969 }, { "epoch": 1.4954670871107607, "grad_norm": 0.5034471154212952, "learning_rate": 0.0005, "loss": 1.4456841945648193, "step": 18970 }, { "epoch": 1.4955459203783996, "grad_norm": 0.5418146252632141, "learning_rate": 0.0005, "loss": 1.5074224472045898, "step": 18971 }, { "epoch": 1.4956247536460388, "grad_norm": 0.5025279521942139, "learning_rate": 0.0005, "loss": 1.5060884952545166, "step": 18972 }, { "epoch": 1.4957035869136774, "grad_norm": 0.4797510802745819, "learning_rate": 0.0005, "loss": 1.5545692443847656, "step": 18973 }, { "epoch": 1.4957824201813166, "grad_norm": 0.45314499735832214, "learning_rate": 0.0005, "loss": 1.4410054683685303, "step": 18974 }, { "epoch": 1.4958612534489555, "grad_norm": 0.5379012227058411, "learning_rate": 0.0005, "loss": 1.5107570886611938, "step": 18975 }, { "epoch": 1.4959400867165944, "grad_norm": 0.5107573866844177, "learning_rate": 0.0005, "loss": 1.770992636680603, "step": 18976 }, { "epoch": 1.4960189199842333, "grad_norm": 0.4866693615913391, "learning_rate": 0.0005, "loss": 1.6872098445892334, "step": 18977 }, { "epoch": 1.4960977532518722, "grad_norm": 0.6127998232841492, "learning_rate": 0.0005, "loss": 1.5814974308013916, "step": 18978 }, { "epoch": 1.4961765865195114, "grad_norm": 0.572788417339325, "learning_rate": 0.0005, "loss": 1.5537358522415161, "step": 18979 }, { "epoch": 1.49625541978715, "grad_norm": 0.5647261738777161, "learning_rate": 0.0005, "loss": 1.5500729084014893, "step": 18980 }, { "epoch": 1.4963342530547892, "grad_norm": 0.5386168956756592, "learning_rate": 0.0005, "loss": 1.5871505737304688, "step": 18981 }, { "epoch": 1.496413086322428, "grad_norm": 0.516465961933136, "learning_rate": 0.0005, "loss": 1.748814582824707, "step": 18982 }, { "epoch": 1.496491919590067, "grad_norm": 0.50306636095047, "learning_rate": 0.0005, "loss": 1.693569302558899, "step": 18983 }, { "epoch": 1.496570752857706, "grad_norm": 0.502731442451477, "learning_rate": 0.0005, "loss": 1.4207597970962524, "step": 18984 }, { "epoch": 1.4966495861253448, "grad_norm": 0.6298551559448242, "learning_rate": 0.0005, "loss": 1.4935951232910156, "step": 18985 }, { "epoch": 1.496728419392984, "grad_norm": 0.578412652015686, "learning_rate": 0.0005, "loss": 1.5355441570281982, "step": 18986 }, { "epoch": 1.4968072526606226, "grad_norm": 0.4528798758983612, "learning_rate": 0.0005, "loss": 1.4922337532043457, "step": 18987 }, { "epoch": 1.4968860859282618, "grad_norm": 0.5333957672119141, "learning_rate": 0.0005, "loss": 1.6092008352279663, "step": 18988 }, { "epoch": 1.4969649191959007, "grad_norm": 0.5066562294960022, "learning_rate": 0.0005, "loss": 1.4307008981704712, "step": 18989 }, { "epoch": 1.4970437524635396, "grad_norm": 0.6385390758514404, "learning_rate": 0.0005, "loss": 1.5044108629226685, "step": 18990 }, { "epoch": 1.4971225857311785, "grad_norm": 0.5105125904083252, "learning_rate": 0.0005, "loss": 1.3762481212615967, "step": 18991 }, { "epoch": 1.4972014189988174, "grad_norm": 0.5463109016418457, "learning_rate": 0.0005, "loss": 1.5234525203704834, "step": 18992 }, { "epoch": 1.4972802522664566, "grad_norm": 0.596412718296051, "learning_rate": 0.0005, "loss": 1.6181328296661377, "step": 18993 }, { "epoch": 1.4973590855340955, "grad_norm": 0.4215875267982483, "learning_rate": 0.0005, "loss": 1.522836446762085, "step": 18994 }, { "epoch": 1.4974379188017344, "grad_norm": 0.4732562303543091, "learning_rate": 0.0005, "loss": 1.6559820175170898, "step": 18995 }, { "epoch": 1.4975167520693733, "grad_norm": 0.5831548571586609, "learning_rate": 0.0005, "loss": 1.3289210796356201, "step": 18996 }, { "epoch": 1.4975955853370122, "grad_norm": 0.48010560870170593, "learning_rate": 0.0005, "loss": 1.425536036491394, "step": 18997 }, { "epoch": 1.4976744186046511, "grad_norm": 0.4882485270500183, "learning_rate": 0.0005, "loss": 1.4089215993881226, "step": 18998 }, { "epoch": 1.49775325187229, "grad_norm": 0.5911916494369507, "learning_rate": 0.0005, "loss": 1.602327585220337, "step": 18999 }, { "epoch": 1.4978320851399292, "grad_norm": 0.8198747634887695, "learning_rate": 0.0005, "loss": 1.818402647972107, "step": 19000 }, { "epoch": 1.497910918407568, "grad_norm": 0.607598066329956, "learning_rate": 0.0005, "loss": 1.661077857017517, "step": 19001 }, { "epoch": 1.497989751675207, "grad_norm": 0.5244778990745544, "learning_rate": 0.0005, "loss": 1.5994315147399902, "step": 19002 }, { "epoch": 1.498068584942846, "grad_norm": 0.6409752368927002, "learning_rate": 0.0005, "loss": 1.7281036376953125, "step": 19003 }, { "epoch": 1.4981474182104848, "grad_norm": 0.5488102436065674, "learning_rate": 0.0005, "loss": 1.4812897443771362, "step": 19004 }, { "epoch": 1.4982262514781237, "grad_norm": 0.5041568875312805, "learning_rate": 0.0005, "loss": 1.3731814622879028, "step": 19005 }, { "epoch": 1.4983050847457626, "grad_norm": 0.6286882162094116, "learning_rate": 0.0005, "loss": 1.5998482704162598, "step": 19006 }, { "epoch": 1.4983839180134018, "grad_norm": 0.5061158537864685, "learning_rate": 0.0005, "loss": 1.5093269348144531, "step": 19007 }, { "epoch": 1.4984627512810407, "grad_norm": 0.502465546131134, "learning_rate": 0.0005, "loss": 1.4974911212921143, "step": 19008 }, { "epoch": 1.4985415845486796, "grad_norm": 0.5566292405128479, "learning_rate": 0.0005, "loss": 1.492716908454895, "step": 19009 }, { "epoch": 1.4986204178163185, "grad_norm": 0.5398218631744385, "learning_rate": 0.0005, "loss": 1.5145314931869507, "step": 19010 }, { "epoch": 1.4986992510839574, "grad_norm": 0.5178619623184204, "learning_rate": 0.0005, "loss": 1.570709228515625, "step": 19011 }, { "epoch": 1.4987780843515963, "grad_norm": 0.7490725517272949, "learning_rate": 0.0005, "loss": 1.4274097681045532, "step": 19012 }, { "epoch": 1.4988569176192352, "grad_norm": 0.5482683777809143, "learning_rate": 0.0005, "loss": 1.5134190320968628, "step": 19013 }, { "epoch": 1.4989357508868744, "grad_norm": 0.5146867036819458, "learning_rate": 0.0005, "loss": 1.5488076210021973, "step": 19014 }, { "epoch": 1.4990145841545133, "grad_norm": 0.5653437972068787, "learning_rate": 0.0005, "loss": 1.6386888027191162, "step": 19015 }, { "epoch": 1.4990934174221522, "grad_norm": 0.5008283257484436, "learning_rate": 0.0005, "loss": 1.4833104610443115, "step": 19016 }, { "epoch": 1.499172250689791, "grad_norm": 0.587890088558197, "learning_rate": 0.0005, "loss": 1.448524832725525, "step": 19017 }, { "epoch": 1.49925108395743, "grad_norm": 0.4791563153266907, "learning_rate": 0.0005, "loss": 1.4884158372879028, "step": 19018 }, { "epoch": 1.499329917225069, "grad_norm": 0.6514338850975037, "learning_rate": 0.0005, "loss": 1.4874498844146729, "step": 19019 }, { "epoch": 1.4994087504927078, "grad_norm": 0.5472586154937744, "learning_rate": 0.0005, "loss": 1.4658803939819336, "step": 19020 }, { "epoch": 1.499487583760347, "grad_norm": 0.5530808568000793, "learning_rate": 0.0005, "loss": 1.473398208618164, "step": 19021 }, { "epoch": 1.4995664170279859, "grad_norm": 0.554567277431488, "learning_rate": 0.0005, "loss": 1.6545716524124146, "step": 19022 }, { "epoch": 1.4996452502956248, "grad_norm": 0.5242253541946411, "learning_rate": 0.0005, "loss": 1.5248230695724487, "step": 19023 }, { "epoch": 1.4997240835632637, "grad_norm": 0.5726906061172485, "learning_rate": 0.0005, "loss": 1.7815989255905151, "step": 19024 }, { "epoch": 1.4998029168309026, "grad_norm": 0.5547699928283691, "learning_rate": 0.0005, "loss": 1.4417808055877686, "step": 19025 }, { "epoch": 1.4998817500985415, "grad_norm": 0.6320492625236511, "learning_rate": 0.0005, "loss": 1.6180955171585083, "step": 19026 }, { "epoch": 1.4999605833661804, "grad_norm": 0.6131688356399536, "learning_rate": 0.0005, "loss": 1.7911304235458374, "step": 19027 }, { "epoch": 1.5000394166338196, "grad_norm": 0.5987929105758667, "learning_rate": 0.0005, "loss": 1.3095134496688843, "step": 19028 }, { "epoch": 1.5001182499014583, "grad_norm": 0.8105596899986267, "learning_rate": 0.0005, "loss": 1.475048542022705, "step": 19029 }, { "epoch": 1.5001970831690974, "grad_norm": 0.5779790878295898, "learning_rate": 0.0005, "loss": 1.6497935056686401, "step": 19030 }, { "epoch": 1.5002759164367363, "grad_norm": 0.6300104260444641, "learning_rate": 0.0005, "loss": 1.6354892253875732, "step": 19031 }, { "epoch": 1.5003547497043752, "grad_norm": 0.6009039878845215, "learning_rate": 0.0005, "loss": 1.5928504467010498, "step": 19032 }, { "epoch": 1.5004335829720143, "grad_norm": 0.46531981229782104, "learning_rate": 0.0005, "loss": 1.3439027070999146, "step": 19033 }, { "epoch": 1.500512416239653, "grad_norm": 0.5641887784004211, "learning_rate": 0.0005, "loss": 1.4531768560409546, "step": 19034 }, { "epoch": 1.5005912495072922, "grad_norm": 1.161579966545105, "learning_rate": 0.0005, "loss": 1.4117469787597656, "step": 19035 }, { "epoch": 1.5006700827749309, "grad_norm": 0.5622219443321228, "learning_rate": 0.0005, "loss": 1.6752252578735352, "step": 19036 }, { "epoch": 1.50074891604257, "grad_norm": 0.5964399576187134, "learning_rate": 0.0005, "loss": 1.5981028079986572, "step": 19037 }, { "epoch": 1.500827749310209, "grad_norm": 0.5567134618759155, "learning_rate": 0.0005, "loss": 1.4082047939300537, "step": 19038 }, { "epoch": 1.5009065825778478, "grad_norm": 0.4884454607963562, "learning_rate": 0.0005, "loss": 1.4692325592041016, "step": 19039 }, { "epoch": 1.500985415845487, "grad_norm": 0.6417675018310547, "learning_rate": 0.0005, "loss": 1.5948121547698975, "step": 19040 }, { "epoch": 1.5010642491131256, "grad_norm": 0.6606191396713257, "learning_rate": 0.0005, "loss": 1.6417174339294434, "step": 19041 }, { "epoch": 1.5011430823807648, "grad_norm": 0.45000162720680237, "learning_rate": 0.0005, "loss": 1.485776662826538, "step": 19042 }, { "epoch": 1.5012219156484035, "grad_norm": 0.4651131331920624, "learning_rate": 0.0005, "loss": 1.5431907176971436, "step": 19043 }, { "epoch": 1.5013007489160426, "grad_norm": 0.6473037600517273, "learning_rate": 0.0005, "loss": 1.54447603225708, "step": 19044 }, { "epoch": 1.5013795821836815, "grad_norm": 0.5277532935142517, "learning_rate": 0.0005, "loss": 1.497049331665039, "step": 19045 }, { "epoch": 1.5014584154513204, "grad_norm": 0.7292407155036926, "learning_rate": 0.0005, "loss": 1.3550596237182617, "step": 19046 }, { "epoch": 1.5015372487189595, "grad_norm": 0.6297629475593567, "learning_rate": 0.0005, "loss": 1.6552510261535645, "step": 19047 }, { "epoch": 1.5016160819865982, "grad_norm": 0.5998572707176208, "learning_rate": 0.0005, "loss": 1.7599573135375977, "step": 19048 }, { "epoch": 1.5016949152542374, "grad_norm": 0.5605143308639526, "learning_rate": 0.0005, "loss": 1.783853530883789, "step": 19049 }, { "epoch": 1.5017737485218763, "grad_norm": 0.6151507496833801, "learning_rate": 0.0005, "loss": 1.7972595691680908, "step": 19050 }, { "epoch": 1.5018525817895152, "grad_norm": 0.5668942928314209, "learning_rate": 0.0005, "loss": 1.4222630262374878, "step": 19051 }, { "epoch": 1.501931415057154, "grad_norm": 0.6089854836463928, "learning_rate": 0.0005, "loss": 1.667975902557373, "step": 19052 }, { "epoch": 1.502010248324793, "grad_norm": 0.6287724375724792, "learning_rate": 0.0005, "loss": 1.5193644762039185, "step": 19053 }, { "epoch": 1.5020890815924322, "grad_norm": 0.6863144040107727, "learning_rate": 0.0005, "loss": 1.58742094039917, "step": 19054 }, { "epoch": 1.5021679148600708, "grad_norm": 0.7791173458099365, "learning_rate": 0.0005, "loss": 1.75214684009552, "step": 19055 }, { "epoch": 1.50224674812771, "grad_norm": 0.7201704978942871, "learning_rate": 0.0005, "loss": 1.6176552772521973, "step": 19056 }, { "epoch": 1.5023255813953489, "grad_norm": 0.5282507538795471, "learning_rate": 0.0005, "loss": 1.452593445777893, "step": 19057 }, { "epoch": 1.5024044146629878, "grad_norm": 0.45269691944122314, "learning_rate": 0.0005, "loss": 1.5013957023620605, "step": 19058 }, { "epoch": 1.5024832479306267, "grad_norm": 0.5509517192840576, "learning_rate": 0.0005, "loss": 1.6033276319503784, "step": 19059 }, { "epoch": 1.5025620811982656, "grad_norm": 0.5104091167449951, "learning_rate": 0.0005, "loss": 1.4047929048538208, "step": 19060 }, { "epoch": 1.5026409144659048, "grad_norm": 0.48143041133880615, "learning_rate": 0.0005, "loss": 1.5838968753814697, "step": 19061 }, { "epoch": 1.5027197477335434, "grad_norm": 0.8815457224845886, "learning_rate": 0.0005, "loss": 1.701430082321167, "step": 19062 }, { "epoch": 1.5027985810011826, "grad_norm": 0.4769608974456787, "learning_rate": 0.0005, "loss": 1.3280032873153687, "step": 19063 }, { "epoch": 1.5028774142688215, "grad_norm": 0.5425614714622498, "learning_rate": 0.0005, "loss": 1.737388253211975, "step": 19064 }, { "epoch": 1.5029562475364604, "grad_norm": 0.6384403109550476, "learning_rate": 0.0005, "loss": 1.7078049182891846, "step": 19065 }, { "epoch": 1.5030350808040993, "grad_norm": 0.5229787230491638, "learning_rate": 0.0005, "loss": 1.6881787776947021, "step": 19066 }, { "epoch": 1.5031139140717382, "grad_norm": 0.61234450340271, "learning_rate": 0.0005, "loss": 1.5271944999694824, "step": 19067 }, { "epoch": 1.5031927473393774, "grad_norm": 0.46309059858322144, "learning_rate": 0.0005, "loss": 1.5695672035217285, "step": 19068 }, { "epoch": 1.503271580607016, "grad_norm": 0.5299398899078369, "learning_rate": 0.0005, "loss": 1.676396131515503, "step": 19069 }, { "epoch": 1.5033504138746552, "grad_norm": 0.48716261982917786, "learning_rate": 0.0005, "loss": 1.7486456632614136, "step": 19070 }, { "epoch": 1.503429247142294, "grad_norm": 0.5836460590362549, "learning_rate": 0.0005, "loss": 1.68025541305542, "step": 19071 }, { "epoch": 1.503508080409933, "grad_norm": 0.44978007674217224, "learning_rate": 0.0005, "loss": 1.5986968278884888, "step": 19072 }, { "epoch": 1.503586913677572, "grad_norm": 0.8116584420204163, "learning_rate": 0.0005, "loss": 1.7388026714324951, "step": 19073 }, { "epoch": 1.5036657469452108, "grad_norm": 0.6331611275672913, "learning_rate": 0.0005, "loss": 1.4209522008895874, "step": 19074 }, { "epoch": 1.50374458021285, "grad_norm": 0.5935725569725037, "learning_rate": 0.0005, "loss": 1.5587316751480103, "step": 19075 }, { "epoch": 1.5038234134804886, "grad_norm": 0.6124907732009888, "learning_rate": 0.0005, "loss": 1.459870457649231, "step": 19076 }, { "epoch": 1.5039022467481278, "grad_norm": 0.5132763385772705, "learning_rate": 0.0005, "loss": 1.559282660484314, "step": 19077 }, { "epoch": 1.5039810800157667, "grad_norm": 0.5147002935409546, "learning_rate": 0.0005, "loss": 1.5698587894439697, "step": 19078 }, { "epoch": 1.5040599132834056, "grad_norm": 0.6499040126800537, "learning_rate": 0.0005, "loss": 1.4665899276733398, "step": 19079 }, { "epoch": 1.5041387465510445, "grad_norm": 0.4525987505912781, "learning_rate": 0.0005, "loss": 1.6021857261657715, "step": 19080 }, { "epoch": 1.5042175798186834, "grad_norm": 0.5564363598823547, "learning_rate": 0.0005, "loss": 1.6000945568084717, "step": 19081 }, { "epoch": 1.5042964130863226, "grad_norm": 0.5723351240158081, "learning_rate": 0.0005, "loss": 1.593919038772583, "step": 19082 }, { "epoch": 1.5043752463539612, "grad_norm": 0.5562725067138672, "learning_rate": 0.0005, "loss": 1.6124811172485352, "step": 19083 }, { "epoch": 1.5044540796216004, "grad_norm": 0.5477896928787231, "learning_rate": 0.0005, "loss": 1.7219839096069336, "step": 19084 }, { "epoch": 1.5045329128892393, "grad_norm": 0.8810445666313171, "learning_rate": 0.0005, "loss": 1.673036813735962, "step": 19085 }, { "epoch": 1.5046117461568782, "grad_norm": 0.5906404852867126, "learning_rate": 0.0005, "loss": 1.6724284887313843, "step": 19086 }, { "epoch": 1.5046905794245171, "grad_norm": 0.470736026763916, "learning_rate": 0.0005, "loss": 1.4745365381240845, "step": 19087 }, { "epoch": 1.504769412692156, "grad_norm": 0.46122249960899353, "learning_rate": 0.0005, "loss": 1.3426203727722168, "step": 19088 }, { "epoch": 1.5048482459597952, "grad_norm": 1.1355929374694824, "learning_rate": 0.0005, "loss": 1.4080721139907837, "step": 19089 }, { "epoch": 1.5049270792274339, "grad_norm": 0.5367047190666199, "learning_rate": 0.0005, "loss": 1.5403074026107788, "step": 19090 }, { "epoch": 1.505005912495073, "grad_norm": 0.5107235908508301, "learning_rate": 0.0005, "loss": 1.4398458003997803, "step": 19091 }, { "epoch": 1.505084745762712, "grad_norm": 0.6352507472038269, "learning_rate": 0.0005, "loss": 1.7305331230163574, "step": 19092 }, { "epoch": 1.5051635790303508, "grad_norm": 0.5091521739959717, "learning_rate": 0.0005, "loss": 1.464115023612976, "step": 19093 }, { "epoch": 1.5052424122979897, "grad_norm": 0.47947725653648376, "learning_rate": 0.0005, "loss": 1.67667555809021, "step": 19094 }, { "epoch": 1.5053212455656286, "grad_norm": 0.6031323075294495, "learning_rate": 0.0005, "loss": 1.5237317085266113, "step": 19095 }, { "epoch": 1.5054000788332678, "grad_norm": 0.5228472948074341, "learning_rate": 0.0005, "loss": 1.7492139339447021, "step": 19096 }, { "epoch": 1.5054789121009065, "grad_norm": 0.6045218706130981, "learning_rate": 0.0005, "loss": 1.4237806797027588, "step": 19097 }, { "epoch": 1.5055577453685456, "grad_norm": 0.5512602925300598, "learning_rate": 0.0005, "loss": 1.5083916187286377, "step": 19098 }, { "epoch": 1.5056365786361845, "grad_norm": 0.5528185963630676, "learning_rate": 0.0005, "loss": 1.602211594581604, "step": 19099 }, { "epoch": 1.5057154119038234, "grad_norm": 0.5986568331718445, "learning_rate": 0.0005, "loss": 1.6533684730529785, "step": 19100 }, { "epoch": 1.5057942451714623, "grad_norm": 0.6698511838912964, "learning_rate": 0.0005, "loss": 1.5532522201538086, "step": 19101 }, { "epoch": 1.5058730784391012, "grad_norm": 0.6647225022315979, "learning_rate": 0.0005, "loss": 1.58912193775177, "step": 19102 }, { "epoch": 1.5059519117067404, "grad_norm": 0.4919146001338959, "learning_rate": 0.0005, "loss": 1.624906301498413, "step": 19103 }, { "epoch": 1.506030744974379, "grad_norm": 0.7061482667922974, "learning_rate": 0.0005, "loss": 1.5154216289520264, "step": 19104 }, { "epoch": 1.5061095782420182, "grad_norm": 0.5861024856567383, "learning_rate": 0.0005, "loss": 1.4152661561965942, "step": 19105 }, { "epoch": 1.506188411509657, "grad_norm": 0.6242055892944336, "learning_rate": 0.0005, "loss": 1.493908166885376, "step": 19106 }, { "epoch": 1.506267244777296, "grad_norm": 0.5441517233848572, "learning_rate": 0.0005, "loss": 1.67356276512146, "step": 19107 }, { "epoch": 1.506346078044935, "grad_norm": 0.45822036266326904, "learning_rate": 0.0005, "loss": 1.3523433208465576, "step": 19108 }, { "epoch": 1.5064249113125738, "grad_norm": 0.519274890422821, "learning_rate": 0.0005, "loss": 1.6002804040908813, "step": 19109 }, { "epoch": 1.506503744580213, "grad_norm": 0.5435398817062378, "learning_rate": 0.0005, "loss": 1.5016331672668457, "step": 19110 }, { "epoch": 1.5065825778478517, "grad_norm": 0.60097736120224, "learning_rate": 0.0005, "loss": 1.7555437088012695, "step": 19111 }, { "epoch": 1.5066614111154908, "grad_norm": 0.5903410911560059, "learning_rate": 0.0005, "loss": 1.5397224426269531, "step": 19112 }, { "epoch": 1.5067402443831297, "grad_norm": 0.4137866199016571, "learning_rate": 0.0005, "loss": 1.4336386919021606, "step": 19113 }, { "epoch": 1.5068190776507686, "grad_norm": 0.48267102241516113, "learning_rate": 0.0005, "loss": 1.7152544260025024, "step": 19114 }, { "epoch": 1.5068979109184075, "grad_norm": 0.4884333312511444, "learning_rate": 0.0005, "loss": 1.4533941745758057, "step": 19115 }, { "epoch": 1.5069767441860464, "grad_norm": 0.5967293381690979, "learning_rate": 0.0005, "loss": 1.7066869735717773, "step": 19116 }, { "epoch": 1.5070555774536856, "grad_norm": 0.4571492075920105, "learning_rate": 0.0005, "loss": 1.2938511371612549, "step": 19117 }, { "epoch": 1.5071344107213243, "grad_norm": 0.5123435258865356, "learning_rate": 0.0005, "loss": 1.7312300205230713, "step": 19118 }, { "epoch": 1.5072132439889634, "grad_norm": 0.5643547177314758, "learning_rate": 0.0005, "loss": 1.5979456901550293, "step": 19119 }, { "epoch": 1.5072920772566023, "grad_norm": 0.46622562408447266, "learning_rate": 0.0005, "loss": 1.5713250637054443, "step": 19120 }, { "epoch": 1.5073709105242412, "grad_norm": 0.4955555200576782, "learning_rate": 0.0005, "loss": 1.5093410015106201, "step": 19121 }, { "epoch": 1.5074497437918801, "grad_norm": 0.5226908922195435, "learning_rate": 0.0005, "loss": 1.5600045919418335, "step": 19122 }, { "epoch": 1.507528577059519, "grad_norm": 0.5490460991859436, "learning_rate": 0.0005, "loss": 1.5922870635986328, "step": 19123 }, { "epoch": 1.5076074103271582, "grad_norm": 0.5671875476837158, "learning_rate": 0.0005, "loss": 1.4213895797729492, "step": 19124 }, { "epoch": 1.5076862435947969, "grad_norm": 0.5315271019935608, "learning_rate": 0.0005, "loss": 1.6314486265182495, "step": 19125 }, { "epoch": 1.507765076862436, "grad_norm": 0.7242768406867981, "learning_rate": 0.0005, "loss": 1.4506025314331055, "step": 19126 }, { "epoch": 1.507843910130075, "grad_norm": 0.5342339277267456, "learning_rate": 0.0005, "loss": 1.577649474143982, "step": 19127 }, { "epoch": 1.5079227433977138, "grad_norm": 0.5334575176239014, "learning_rate": 0.0005, "loss": 1.4963346719741821, "step": 19128 }, { "epoch": 1.5080015766653527, "grad_norm": 0.6115246415138245, "learning_rate": 0.0005, "loss": 1.6754720211029053, "step": 19129 }, { "epoch": 1.5080804099329916, "grad_norm": 0.4892459213733673, "learning_rate": 0.0005, "loss": 1.4893577098846436, "step": 19130 }, { "epoch": 1.5081592432006308, "grad_norm": 0.5280182957649231, "learning_rate": 0.0005, "loss": 1.5730643272399902, "step": 19131 }, { "epoch": 1.5082380764682695, "grad_norm": 0.5256316065788269, "learning_rate": 0.0005, "loss": 1.5522902011871338, "step": 19132 }, { "epoch": 1.5083169097359086, "grad_norm": 0.6143466830253601, "learning_rate": 0.0005, "loss": 1.7176234722137451, "step": 19133 }, { "epoch": 1.5083957430035475, "grad_norm": 0.6378703117370605, "learning_rate": 0.0005, "loss": 1.477725863456726, "step": 19134 }, { "epoch": 1.5084745762711864, "grad_norm": 0.6604808568954468, "learning_rate": 0.0005, "loss": 1.549964189529419, "step": 19135 }, { "epoch": 1.5085534095388253, "grad_norm": 0.5430654287338257, "learning_rate": 0.0005, "loss": 1.4992599487304688, "step": 19136 }, { "epoch": 1.5086322428064642, "grad_norm": 0.5117629766464233, "learning_rate": 0.0005, "loss": 1.5195850133895874, "step": 19137 }, { "epoch": 1.5087110760741034, "grad_norm": 0.7102320790290833, "learning_rate": 0.0005, "loss": 1.607269048690796, "step": 19138 }, { "epoch": 1.508789909341742, "grad_norm": 0.5837268233299255, "learning_rate": 0.0005, "loss": 1.4097868204116821, "step": 19139 }, { "epoch": 1.5088687426093812, "grad_norm": 0.44195154309272766, "learning_rate": 0.0005, "loss": 1.6244735717773438, "step": 19140 }, { "epoch": 1.50894757587702, "grad_norm": 0.4433422386646271, "learning_rate": 0.0005, "loss": 1.4016250371932983, "step": 19141 }, { "epoch": 1.509026409144659, "grad_norm": 0.5655063390731812, "learning_rate": 0.0005, "loss": 1.7170934677124023, "step": 19142 }, { "epoch": 1.5091052424122982, "grad_norm": 0.5870783925056458, "learning_rate": 0.0005, "loss": 1.538527488708496, "step": 19143 }, { "epoch": 1.5091840756799368, "grad_norm": 0.564079225063324, "learning_rate": 0.0005, "loss": 1.6524876356124878, "step": 19144 }, { "epoch": 1.509262908947576, "grad_norm": 0.6341103315353394, "learning_rate": 0.0005, "loss": 1.4960181713104248, "step": 19145 }, { "epoch": 1.5093417422152147, "grad_norm": 1.01304292678833, "learning_rate": 0.0005, "loss": 1.5079971551895142, "step": 19146 }, { "epoch": 1.5094205754828538, "grad_norm": 0.502099335193634, "learning_rate": 0.0005, "loss": 1.6336442232131958, "step": 19147 }, { "epoch": 1.5094994087504927, "grad_norm": 0.624480664730072, "learning_rate": 0.0005, "loss": 1.3921468257904053, "step": 19148 }, { "epoch": 1.5095782420181316, "grad_norm": 0.5256893634796143, "learning_rate": 0.0005, "loss": 1.4525831937789917, "step": 19149 }, { "epoch": 1.5096570752857708, "grad_norm": 0.5215163230895996, "learning_rate": 0.0005, "loss": 1.4330110549926758, "step": 19150 }, { "epoch": 1.5097359085534094, "grad_norm": 0.682526171207428, "learning_rate": 0.0005, "loss": 1.5076313018798828, "step": 19151 }, { "epoch": 1.5098147418210486, "grad_norm": 0.5353213548660278, "learning_rate": 0.0005, "loss": 1.6491217613220215, "step": 19152 }, { "epoch": 1.5098935750886873, "grad_norm": 0.48632335662841797, "learning_rate": 0.0005, "loss": 1.532787799835205, "step": 19153 }, { "epoch": 1.5099724083563264, "grad_norm": 0.6219934225082397, "learning_rate": 0.0005, "loss": 1.707835078239441, "step": 19154 }, { "epoch": 1.5100512416239653, "grad_norm": 0.5929855108261108, "learning_rate": 0.0005, "loss": 1.5170869827270508, "step": 19155 }, { "epoch": 1.5101300748916042, "grad_norm": 0.4442032277584076, "learning_rate": 0.0005, "loss": 1.4580059051513672, "step": 19156 }, { "epoch": 1.5102089081592434, "grad_norm": 0.6376791000366211, "learning_rate": 0.0005, "loss": 1.581571340560913, "step": 19157 }, { "epoch": 1.510287741426882, "grad_norm": 0.6924313902854919, "learning_rate": 0.0005, "loss": 1.5876882076263428, "step": 19158 }, { "epoch": 1.5103665746945212, "grad_norm": 0.558370053768158, "learning_rate": 0.0005, "loss": 1.7528799772262573, "step": 19159 }, { "epoch": 1.5104454079621599, "grad_norm": 0.6267319917678833, "learning_rate": 0.0005, "loss": 1.5552537441253662, "step": 19160 }, { "epoch": 1.510524241229799, "grad_norm": 0.5317492485046387, "learning_rate": 0.0005, "loss": 1.5820190906524658, "step": 19161 }, { "epoch": 1.510603074497438, "grad_norm": 0.5958977341651917, "learning_rate": 0.0005, "loss": 1.5933973789215088, "step": 19162 }, { "epoch": 1.5106819077650768, "grad_norm": 4.573745250701904, "learning_rate": 0.0005, "loss": 1.4337241649627686, "step": 19163 }, { "epoch": 1.510760741032716, "grad_norm": 0.4831746220588684, "learning_rate": 0.0005, "loss": 1.4051707983016968, "step": 19164 }, { "epoch": 1.5108395743003546, "grad_norm": 0.4836547374725342, "learning_rate": 0.0005, "loss": 1.4204388856887817, "step": 19165 }, { "epoch": 1.5109184075679938, "grad_norm": 0.48326048254966736, "learning_rate": 0.0005, "loss": 1.505614161491394, "step": 19166 }, { "epoch": 1.5109972408356325, "grad_norm": 0.4387560784816742, "learning_rate": 0.0005, "loss": 1.5445624589920044, "step": 19167 }, { "epoch": 1.5110760741032716, "grad_norm": 0.5448732972145081, "learning_rate": 0.0005, "loss": 1.5803446769714355, "step": 19168 }, { "epoch": 1.5111549073709105, "grad_norm": 0.5862087607383728, "learning_rate": 0.0005, "loss": 1.7205212116241455, "step": 19169 }, { "epoch": 1.5112337406385494, "grad_norm": 0.6235566139221191, "learning_rate": 0.0005, "loss": 1.4778103828430176, "step": 19170 }, { "epoch": 1.5113125739061886, "grad_norm": 0.4540335237979889, "learning_rate": 0.0005, "loss": 1.5100609064102173, "step": 19171 }, { "epoch": 1.5113914071738273, "grad_norm": 0.6890647411346436, "learning_rate": 0.0005, "loss": 1.853156328201294, "step": 19172 }, { "epoch": 1.5114702404414664, "grad_norm": 0.5417882204055786, "learning_rate": 0.0005, "loss": 1.4747133255004883, "step": 19173 }, { "epoch": 1.5115490737091053, "grad_norm": 0.5582586526870728, "learning_rate": 0.0005, "loss": 1.3768316507339478, "step": 19174 }, { "epoch": 1.5116279069767442, "grad_norm": 0.6533219218254089, "learning_rate": 0.0005, "loss": 1.725266456604004, "step": 19175 }, { "epoch": 1.5117067402443831, "grad_norm": 0.7064075469970703, "learning_rate": 0.0005, "loss": 1.6705994606018066, "step": 19176 }, { "epoch": 1.511785573512022, "grad_norm": 0.7208245992660522, "learning_rate": 0.0005, "loss": 1.622588872909546, "step": 19177 }, { "epoch": 1.5118644067796612, "grad_norm": 0.6483103632926941, "learning_rate": 0.0005, "loss": 1.7608988285064697, "step": 19178 }, { "epoch": 1.5119432400472999, "grad_norm": 0.4638268053531647, "learning_rate": 0.0005, "loss": 1.479405403137207, "step": 19179 }, { "epoch": 1.512022073314939, "grad_norm": 0.6691874265670776, "learning_rate": 0.0005, "loss": 1.585249900817871, "step": 19180 }, { "epoch": 1.512100906582578, "grad_norm": 0.6146473288536072, "learning_rate": 0.0005, "loss": 1.5224480628967285, "step": 19181 }, { "epoch": 1.5121797398502168, "grad_norm": 0.5049465298652649, "learning_rate": 0.0005, "loss": 1.7166190147399902, "step": 19182 }, { "epoch": 1.5122585731178557, "grad_norm": 0.5123357176780701, "learning_rate": 0.0005, "loss": 1.436234712600708, "step": 19183 }, { "epoch": 1.5123374063854946, "grad_norm": 0.5605053305625916, "learning_rate": 0.0005, "loss": 1.5055426359176636, "step": 19184 }, { "epoch": 1.5124162396531338, "grad_norm": 0.49053820967674255, "learning_rate": 0.0005, "loss": 1.4000825881958008, "step": 19185 }, { "epoch": 1.5124950729207725, "grad_norm": 0.5151282548904419, "learning_rate": 0.0005, "loss": 1.3627369403839111, "step": 19186 }, { "epoch": 1.5125739061884116, "grad_norm": 0.5624537467956543, "learning_rate": 0.0005, "loss": 1.5969635248184204, "step": 19187 }, { "epoch": 1.5126527394560505, "grad_norm": 0.5985035300254822, "learning_rate": 0.0005, "loss": 1.6079480648040771, "step": 19188 }, { "epoch": 1.5127315727236894, "grad_norm": 0.46971574425697327, "learning_rate": 0.0005, "loss": 1.5664422512054443, "step": 19189 }, { "epoch": 1.5128104059913283, "grad_norm": 0.5206037759780884, "learning_rate": 0.0005, "loss": 1.426285743713379, "step": 19190 }, { "epoch": 1.5128892392589672, "grad_norm": 0.6064851880073547, "learning_rate": 0.0005, "loss": 1.31062912940979, "step": 19191 }, { "epoch": 1.5129680725266064, "grad_norm": 0.4831634759902954, "learning_rate": 0.0005, "loss": 1.6115894317626953, "step": 19192 }, { "epoch": 1.513046905794245, "grad_norm": 0.4843224585056305, "learning_rate": 0.0005, "loss": 1.5017969608306885, "step": 19193 }, { "epoch": 1.5131257390618842, "grad_norm": 0.5324512720108032, "learning_rate": 0.0005, "loss": 1.5882489681243896, "step": 19194 }, { "epoch": 1.513204572329523, "grad_norm": 0.620982825756073, "learning_rate": 0.0005, "loss": 1.3185350894927979, "step": 19195 }, { "epoch": 1.513283405597162, "grad_norm": 0.931898832321167, "learning_rate": 0.0005, "loss": 1.956964373588562, "step": 19196 }, { "epoch": 1.513362238864801, "grad_norm": 0.5161101818084717, "learning_rate": 0.0005, "loss": 1.7280912399291992, "step": 19197 }, { "epoch": 1.5134410721324398, "grad_norm": 0.58689284324646, "learning_rate": 0.0005, "loss": 1.550415277481079, "step": 19198 }, { "epoch": 1.513519905400079, "grad_norm": 0.44149503111839294, "learning_rate": 0.0005, "loss": 1.4579834938049316, "step": 19199 }, { "epoch": 1.5135987386677177, "grad_norm": 0.4544791281223297, "learning_rate": 0.0005, "loss": 1.38119637966156, "step": 19200 }, { "epoch": 1.5136775719353568, "grad_norm": 0.5530012845993042, "learning_rate": 0.0005, "loss": 1.533142328262329, "step": 19201 }, { "epoch": 1.5137564052029957, "grad_norm": 0.599982500076294, "learning_rate": 0.0005, "loss": 1.6149041652679443, "step": 19202 }, { "epoch": 1.5138352384706346, "grad_norm": 0.4710254669189453, "learning_rate": 0.0005, "loss": 1.5465435981750488, "step": 19203 }, { "epoch": 1.5139140717382735, "grad_norm": 0.6245821714401245, "learning_rate": 0.0005, "loss": 1.6518018245697021, "step": 19204 }, { "epoch": 1.5139929050059124, "grad_norm": 0.5442267656326294, "learning_rate": 0.0005, "loss": 1.719082236289978, "step": 19205 }, { "epoch": 1.5140717382735516, "grad_norm": 0.5251057744026184, "learning_rate": 0.0005, "loss": 1.5318810939788818, "step": 19206 }, { "epoch": 1.5141505715411903, "grad_norm": 0.5349637866020203, "learning_rate": 0.0005, "loss": 1.7294771671295166, "step": 19207 }, { "epoch": 1.5142294048088294, "grad_norm": 0.43701767921447754, "learning_rate": 0.0005, "loss": 1.3346749544143677, "step": 19208 }, { "epoch": 1.5143082380764683, "grad_norm": 0.6618719696998596, "learning_rate": 0.0005, "loss": 1.6437277793884277, "step": 19209 }, { "epoch": 1.5143870713441072, "grad_norm": 0.516731858253479, "learning_rate": 0.0005, "loss": 1.5940135717391968, "step": 19210 }, { "epoch": 1.5144659046117461, "grad_norm": 0.5643858909606934, "learning_rate": 0.0005, "loss": 1.5015153884887695, "step": 19211 }, { "epoch": 1.514544737879385, "grad_norm": 0.5509527325630188, "learning_rate": 0.0005, "loss": 1.6066408157348633, "step": 19212 }, { "epoch": 1.5146235711470242, "grad_norm": 0.498063862323761, "learning_rate": 0.0005, "loss": 1.434888482093811, "step": 19213 }, { "epoch": 1.5147024044146629, "grad_norm": 0.5699406862258911, "learning_rate": 0.0005, "loss": 1.6629842519760132, "step": 19214 }, { "epoch": 1.514781237682302, "grad_norm": 0.45955634117126465, "learning_rate": 0.0005, "loss": 1.3523879051208496, "step": 19215 }, { "epoch": 1.514860070949941, "grad_norm": 0.44642263650894165, "learning_rate": 0.0005, "loss": 1.5633548498153687, "step": 19216 }, { "epoch": 1.5149389042175798, "grad_norm": 0.6637911200523376, "learning_rate": 0.0005, "loss": 1.8470134735107422, "step": 19217 }, { "epoch": 1.5150177374852187, "grad_norm": 0.4950198531150818, "learning_rate": 0.0005, "loss": 1.421441912651062, "step": 19218 }, { "epoch": 1.5150965707528576, "grad_norm": 0.4681603014469147, "learning_rate": 0.0005, "loss": 1.546076774597168, "step": 19219 }, { "epoch": 1.5151754040204968, "grad_norm": 0.5937545895576477, "learning_rate": 0.0005, "loss": 1.6945068836212158, "step": 19220 }, { "epoch": 1.5152542372881355, "grad_norm": 0.6125290989875793, "learning_rate": 0.0005, "loss": 1.7708022594451904, "step": 19221 }, { "epoch": 1.5153330705557746, "grad_norm": 0.5288066864013672, "learning_rate": 0.0005, "loss": 1.299316644668579, "step": 19222 }, { "epoch": 1.5154119038234135, "grad_norm": 0.5075029134750366, "learning_rate": 0.0005, "loss": 1.3705755472183228, "step": 19223 }, { "epoch": 1.5154907370910524, "grad_norm": 0.49560973048210144, "learning_rate": 0.0005, "loss": 1.3931307792663574, "step": 19224 }, { "epoch": 1.5155695703586913, "grad_norm": 0.5695331692695618, "learning_rate": 0.0005, "loss": 1.4431953430175781, "step": 19225 }, { "epoch": 1.5156484036263302, "grad_norm": 0.6708330512046814, "learning_rate": 0.0005, "loss": 1.686226487159729, "step": 19226 }, { "epoch": 1.5157272368939694, "grad_norm": 1.235150694847107, "learning_rate": 0.0005, "loss": 1.6125813722610474, "step": 19227 }, { "epoch": 1.515806070161608, "grad_norm": 0.6384680271148682, "learning_rate": 0.0005, "loss": 1.6504430770874023, "step": 19228 }, { "epoch": 1.5158849034292472, "grad_norm": 0.5759577751159668, "learning_rate": 0.0005, "loss": 1.4661858081817627, "step": 19229 }, { "epoch": 1.515963736696886, "grad_norm": 0.7068628668785095, "learning_rate": 0.0005, "loss": 1.8216878175735474, "step": 19230 }, { "epoch": 1.516042569964525, "grad_norm": 0.49470850825309753, "learning_rate": 0.0005, "loss": 1.5562876462936401, "step": 19231 }, { "epoch": 1.516121403232164, "grad_norm": 0.5000607371330261, "learning_rate": 0.0005, "loss": 1.6385650634765625, "step": 19232 }, { "epoch": 1.5162002364998028, "grad_norm": 0.5760279297828674, "learning_rate": 0.0005, "loss": 1.70433509349823, "step": 19233 }, { "epoch": 1.516279069767442, "grad_norm": 0.5327394008636475, "learning_rate": 0.0005, "loss": 1.5768921375274658, "step": 19234 }, { "epoch": 1.5163579030350807, "grad_norm": 0.6323138475418091, "learning_rate": 0.0005, "loss": 1.484383225440979, "step": 19235 }, { "epoch": 1.5164367363027198, "grad_norm": 0.4820611774921417, "learning_rate": 0.0005, "loss": 1.56680428981781, "step": 19236 }, { "epoch": 1.5165155695703587, "grad_norm": 0.5826329588890076, "learning_rate": 0.0005, "loss": 1.6502070426940918, "step": 19237 }, { "epoch": 1.5165944028379976, "grad_norm": 0.6300175189971924, "learning_rate": 0.0005, "loss": 1.5483729839324951, "step": 19238 }, { "epoch": 1.5166732361056365, "grad_norm": 0.6050657033920288, "learning_rate": 0.0005, "loss": 1.4102615118026733, "step": 19239 }, { "epoch": 1.5167520693732754, "grad_norm": 0.6557567119598389, "learning_rate": 0.0005, "loss": 1.6745184659957886, "step": 19240 }, { "epoch": 1.5168309026409146, "grad_norm": 0.7620349526405334, "learning_rate": 0.0005, "loss": 1.5842344760894775, "step": 19241 }, { "epoch": 1.5169097359085533, "grad_norm": 0.5778401494026184, "learning_rate": 0.0005, "loss": 1.510174036026001, "step": 19242 }, { "epoch": 1.5169885691761924, "grad_norm": 0.5128012895584106, "learning_rate": 0.0005, "loss": 1.5950112342834473, "step": 19243 }, { "epoch": 1.5170674024438313, "grad_norm": 0.46980035305023193, "learning_rate": 0.0005, "loss": 1.4514694213867188, "step": 19244 }, { "epoch": 1.5171462357114702, "grad_norm": 0.4970816671848297, "learning_rate": 0.0005, "loss": 1.4739925861358643, "step": 19245 }, { "epoch": 1.5172250689791091, "grad_norm": 0.6182286143302917, "learning_rate": 0.0005, "loss": 1.413862705230713, "step": 19246 }, { "epoch": 1.517303902246748, "grad_norm": 0.5484476089477539, "learning_rate": 0.0005, "loss": 1.5540151596069336, "step": 19247 }, { "epoch": 1.5173827355143872, "grad_norm": 0.4477660655975342, "learning_rate": 0.0005, "loss": 1.4028167724609375, "step": 19248 }, { "epoch": 1.5174615687820259, "grad_norm": 0.5129781365394592, "learning_rate": 0.0005, "loss": 1.586594581604004, "step": 19249 }, { "epoch": 1.517540402049665, "grad_norm": 0.6817841529846191, "learning_rate": 0.0005, "loss": 1.591049313545227, "step": 19250 }, { "epoch": 1.517619235317304, "grad_norm": 0.4817225933074951, "learning_rate": 0.0005, "loss": 1.5236217975616455, "step": 19251 }, { "epoch": 1.5176980685849428, "grad_norm": 0.6488516926765442, "learning_rate": 0.0005, "loss": 1.5374150276184082, "step": 19252 }, { "epoch": 1.5177769018525817, "grad_norm": 0.6825243830680847, "learning_rate": 0.0005, "loss": 1.6791958808898926, "step": 19253 }, { "epoch": 1.5178557351202207, "grad_norm": 0.47545281052589417, "learning_rate": 0.0005, "loss": 1.6717807054519653, "step": 19254 }, { "epoch": 1.5179345683878598, "grad_norm": 0.46508967876434326, "learning_rate": 0.0005, "loss": 1.54142165184021, "step": 19255 }, { "epoch": 1.5180134016554985, "grad_norm": 0.5855541229248047, "learning_rate": 0.0005, "loss": 1.6603431701660156, "step": 19256 }, { "epoch": 1.5180922349231376, "grad_norm": 0.5683813095092773, "learning_rate": 0.0005, "loss": 1.637122631072998, "step": 19257 }, { "epoch": 1.5181710681907765, "grad_norm": 0.5885130763053894, "learning_rate": 0.0005, "loss": 1.6949176788330078, "step": 19258 }, { "epoch": 1.5182499014584154, "grad_norm": 0.5525788068771362, "learning_rate": 0.0005, "loss": 1.4418095350265503, "step": 19259 }, { "epoch": 1.5183287347260543, "grad_norm": 0.5056939125061035, "learning_rate": 0.0005, "loss": 1.768908143043518, "step": 19260 }, { "epoch": 1.5184075679936933, "grad_norm": 0.5607999563217163, "learning_rate": 0.0005, "loss": 1.6405696868896484, "step": 19261 }, { "epoch": 1.5184864012613324, "grad_norm": 0.5012036561965942, "learning_rate": 0.0005, "loss": 1.585172176361084, "step": 19262 }, { "epoch": 1.518565234528971, "grad_norm": 0.7116379737854004, "learning_rate": 0.0005, "loss": 1.4820621013641357, "step": 19263 }, { "epoch": 1.5186440677966102, "grad_norm": 0.4875897169113159, "learning_rate": 0.0005, "loss": 1.5065035820007324, "step": 19264 }, { "epoch": 1.5187229010642491, "grad_norm": 0.49760064482688904, "learning_rate": 0.0005, "loss": 1.5278120040893555, "step": 19265 }, { "epoch": 1.518801734331888, "grad_norm": 0.4773021936416626, "learning_rate": 0.0005, "loss": 1.4744856357574463, "step": 19266 }, { "epoch": 1.5188805675995272, "grad_norm": 0.5073879361152649, "learning_rate": 0.0005, "loss": 1.6177361011505127, "step": 19267 }, { "epoch": 1.5189594008671659, "grad_norm": 0.5701811909675598, "learning_rate": 0.0005, "loss": 1.5429935455322266, "step": 19268 }, { "epoch": 1.519038234134805, "grad_norm": 0.7116031050682068, "learning_rate": 0.0005, "loss": 1.4696201086044312, "step": 19269 }, { "epoch": 1.5191170674024437, "grad_norm": 0.43653345108032227, "learning_rate": 0.0005, "loss": 1.5303951501846313, "step": 19270 }, { "epoch": 1.5191959006700828, "grad_norm": 0.4468163847923279, "learning_rate": 0.0005, "loss": 1.5832550525665283, "step": 19271 }, { "epoch": 1.5192747339377217, "grad_norm": 0.5649757385253906, "learning_rate": 0.0005, "loss": 1.5959560871124268, "step": 19272 }, { "epoch": 1.5193535672053606, "grad_norm": 0.5839231610298157, "learning_rate": 0.0005, "loss": 1.6522239446640015, "step": 19273 }, { "epoch": 1.5194324004729998, "grad_norm": 0.5398616194725037, "learning_rate": 0.0005, "loss": 1.5806984901428223, "step": 19274 }, { "epoch": 1.5195112337406385, "grad_norm": 0.5744966268539429, "learning_rate": 0.0005, "loss": 1.5506517887115479, "step": 19275 }, { "epoch": 1.5195900670082776, "grad_norm": 0.5126648545265198, "learning_rate": 0.0005, "loss": 1.3302257061004639, "step": 19276 }, { "epoch": 1.5196689002759163, "grad_norm": 0.48348164558410645, "learning_rate": 0.0005, "loss": 1.4587315320968628, "step": 19277 }, { "epoch": 1.5197477335435554, "grad_norm": 0.6559180617332458, "learning_rate": 0.0005, "loss": 1.6001745462417603, "step": 19278 }, { "epoch": 1.5198265668111943, "grad_norm": 0.6747543811798096, "learning_rate": 0.0005, "loss": 1.8655887842178345, "step": 19279 }, { "epoch": 1.5199054000788332, "grad_norm": 0.5144509673118591, "learning_rate": 0.0005, "loss": 1.5677180290222168, "step": 19280 }, { "epoch": 1.5199842333464724, "grad_norm": 0.5075835585594177, "learning_rate": 0.0005, "loss": 1.6250725984573364, "step": 19281 }, { "epoch": 1.520063066614111, "grad_norm": 0.901983916759491, "learning_rate": 0.0005, "loss": 1.722257137298584, "step": 19282 }, { "epoch": 1.5201418998817502, "grad_norm": 0.5553669333457947, "learning_rate": 0.0005, "loss": 1.8642765283584595, "step": 19283 }, { "epoch": 1.5202207331493889, "grad_norm": 0.5391161441802979, "learning_rate": 0.0005, "loss": 1.437309980392456, "step": 19284 }, { "epoch": 1.520299566417028, "grad_norm": 0.5037269592285156, "learning_rate": 0.0005, "loss": 1.3551723957061768, "step": 19285 }, { "epoch": 1.520378399684667, "grad_norm": 0.5269114971160889, "learning_rate": 0.0005, "loss": 1.5531203746795654, "step": 19286 }, { "epoch": 1.5204572329523058, "grad_norm": 0.5009928345680237, "learning_rate": 0.0005, "loss": 1.5482261180877686, "step": 19287 }, { "epoch": 1.520536066219945, "grad_norm": 0.6921692490577698, "learning_rate": 0.0005, "loss": 1.6440248489379883, "step": 19288 }, { "epoch": 1.5206148994875837, "grad_norm": 0.5052967071533203, "learning_rate": 0.0005, "loss": 1.3744198083877563, "step": 19289 }, { "epoch": 1.5206937327552228, "grad_norm": 0.6758437156677246, "learning_rate": 0.0005, "loss": 1.7435866594314575, "step": 19290 }, { "epoch": 1.5207725660228615, "grad_norm": 0.6687431931495667, "learning_rate": 0.0005, "loss": 1.6905581951141357, "step": 19291 }, { "epoch": 1.5208513992905006, "grad_norm": 0.5859460234642029, "learning_rate": 0.0005, "loss": 1.670198678970337, "step": 19292 }, { "epoch": 1.5209302325581395, "grad_norm": 0.5190531611442566, "learning_rate": 0.0005, "loss": 1.4792554378509521, "step": 19293 }, { "epoch": 1.5210090658257784, "grad_norm": 0.45922034978866577, "learning_rate": 0.0005, "loss": 1.4546473026275635, "step": 19294 }, { "epoch": 1.5210878990934176, "grad_norm": 0.6780383586883545, "learning_rate": 0.0005, "loss": 1.648223638534546, "step": 19295 }, { "epoch": 1.5211667323610563, "grad_norm": 0.5340293049812317, "learning_rate": 0.0005, "loss": 1.3351823091506958, "step": 19296 }, { "epoch": 1.5212455656286954, "grad_norm": 0.5325226783752441, "learning_rate": 0.0005, "loss": 1.4907550811767578, "step": 19297 }, { "epoch": 1.5213243988963343, "grad_norm": 0.5925288200378418, "learning_rate": 0.0005, "loss": 1.6858808994293213, "step": 19298 }, { "epoch": 1.5214032321639732, "grad_norm": 0.5105945467948914, "learning_rate": 0.0005, "loss": 1.4602885246276855, "step": 19299 }, { "epoch": 1.5214820654316121, "grad_norm": 0.5461451411247253, "learning_rate": 0.0005, "loss": 1.7590663433074951, "step": 19300 }, { "epoch": 1.521560898699251, "grad_norm": 0.592885434627533, "learning_rate": 0.0005, "loss": 1.525626301765442, "step": 19301 }, { "epoch": 1.5216397319668902, "grad_norm": 0.5323163866996765, "learning_rate": 0.0005, "loss": 1.3538607358932495, "step": 19302 }, { "epoch": 1.5217185652345289, "grad_norm": 0.567512035369873, "learning_rate": 0.0005, "loss": 1.4591357707977295, "step": 19303 }, { "epoch": 1.521797398502168, "grad_norm": 0.4947790503501892, "learning_rate": 0.0005, "loss": 1.5932475328445435, "step": 19304 }, { "epoch": 1.521876231769807, "grad_norm": 0.5044007301330566, "learning_rate": 0.0005, "loss": 1.4536490440368652, "step": 19305 }, { "epoch": 1.5219550650374458, "grad_norm": 0.4907529652118683, "learning_rate": 0.0005, "loss": 1.5330122709274292, "step": 19306 }, { "epoch": 1.5220338983050847, "grad_norm": 0.44278666377067566, "learning_rate": 0.0005, "loss": 1.3988237380981445, "step": 19307 }, { "epoch": 1.5221127315727236, "grad_norm": 0.5733844041824341, "learning_rate": 0.0005, "loss": 1.4588420391082764, "step": 19308 }, { "epoch": 1.5221915648403628, "grad_norm": 0.5176694393157959, "learning_rate": 0.0005, "loss": 1.3852990865707397, "step": 19309 }, { "epoch": 1.5222703981080015, "grad_norm": 0.6698898673057556, "learning_rate": 0.0005, "loss": 1.6117947101593018, "step": 19310 }, { "epoch": 1.5223492313756406, "grad_norm": 0.4974994957447052, "learning_rate": 0.0005, "loss": 1.5682644844055176, "step": 19311 }, { "epoch": 1.5224280646432795, "grad_norm": 0.5311765074729919, "learning_rate": 0.0005, "loss": 1.3902677297592163, "step": 19312 }, { "epoch": 1.5225068979109184, "grad_norm": 0.6498476266860962, "learning_rate": 0.0005, "loss": 1.5885791778564453, "step": 19313 }, { "epoch": 1.5225857311785573, "grad_norm": 0.5014387369155884, "learning_rate": 0.0005, "loss": 1.5227699279785156, "step": 19314 }, { "epoch": 1.5226645644461962, "grad_norm": 0.5118228793144226, "learning_rate": 0.0005, "loss": 1.4789724349975586, "step": 19315 }, { "epoch": 1.5227433977138354, "grad_norm": 0.5895670056343079, "learning_rate": 0.0005, "loss": 1.773667335510254, "step": 19316 }, { "epoch": 1.522822230981474, "grad_norm": 0.719028890132904, "learning_rate": 0.0005, "loss": 1.5987823009490967, "step": 19317 }, { "epoch": 1.5229010642491132, "grad_norm": 0.4905689060688019, "learning_rate": 0.0005, "loss": 1.379040241241455, "step": 19318 }, { "epoch": 1.5229798975167521, "grad_norm": 0.5825524926185608, "learning_rate": 0.0005, "loss": 1.7703912258148193, "step": 19319 }, { "epoch": 1.523058730784391, "grad_norm": 0.5097815990447998, "learning_rate": 0.0005, "loss": 1.7215073108673096, "step": 19320 }, { "epoch": 1.52313756405203, "grad_norm": 0.4648544490337372, "learning_rate": 0.0005, "loss": 1.6521785259246826, "step": 19321 }, { "epoch": 1.5232163973196688, "grad_norm": 0.602543830871582, "learning_rate": 0.0005, "loss": 1.8802194595336914, "step": 19322 }, { "epoch": 1.523295230587308, "grad_norm": 0.44335976243019104, "learning_rate": 0.0005, "loss": 1.3450281620025635, "step": 19323 }, { "epoch": 1.5233740638549467, "grad_norm": 0.4423327147960663, "learning_rate": 0.0005, "loss": 1.26300847530365, "step": 19324 }, { "epoch": 1.5234528971225858, "grad_norm": 0.6424245238304138, "learning_rate": 0.0005, "loss": 1.7209420204162598, "step": 19325 }, { "epoch": 1.5235317303902247, "grad_norm": 0.5498539805412292, "learning_rate": 0.0005, "loss": 1.4846373796463013, "step": 19326 }, { "epoch": 1.5236105636578636, "grad_norm": 0.5538972616195679, "learning_rate": 0.0005, "loss": 1.716189980506897, "step": 19327 }, { "epoch": 1.5236893969255025, "grad_norm": 0.5048043131828308, "learning_rate": 0.0005, "loss": 1.512782096862793, "step": 19328 }, { "epoch": 1.5237682301931414, "grad_norm": 0.5064697861671448, "learning_rate": 0.0005, "loss": 1.4265573024749756, "step": 19329 }, { "epoch": 1.5238470634607806, "grad_norm": 0.6106984615325928, "learning_rate": 0.0005, "loss": 1.5302938222885132, "step": 19330 }, { "epoch": 1.5239258967284193, "grad_norm": 0.5212523341178894, "learning_rate": 0.0005, "loss": 1.3730032444000244, "step": 19331 }, { "epoch": 1.5240047299960584, "grad_norm": 0.5707043409347534, "learning_rate": 0.0005, "loss": 1.6224697828292847, "step": 19332 }, { "epoch": 1.5240835632636973, "grad_norm": 0.5428808331489563, "learning_rate": 0.0005, "loss": 1.452035665512085, "step": 19333 }, { "epoch": 1.5241623965313362, "grad_norm": 0.4918259382247925, "learning_rate": 0.0005, "loss": 1.3740559816360474, "step": 19334 }, { "epoch": 1.5242412297989751, "grad_norm": 0.5765498876571655, "learning_rate": 0.0005, "loss": 1.5686354637145996, "step": 19335 }, { "epoch": 1.524320063066614, "grad_norm": 0.5335713624954224, "learning_rate": 0.0005, "loss": 1.4234586954116821, "step": 19336 }, { "epoch": 1.5243988963342532, "grad_norm": 0.640862226486206, "learning_rate": 0.0005, "loss": 1.4772865772247314, "step": 19337 }, { "epoch": 1.5244777296018919, "grad_norm": 0.5014054775238037, "learning_rate": 0.0005, "loss": 1.3707313537597656, "step": 19338 }, { "epoch": 1.524556562869531, "grad_norm": 0.4717402160167694, "learning_rate": 0.0005, "loss": 1.3922163248062134, "step": 19339 }, { "epoch": 1.52463539613717, "grad_norm": 0.4770664870738983, "learning_rate": 0.0005, "loss": 1.4271867275238037, "step": 19340 }, { "epoch": 1.5247142294048088, "grad_norm": 0.8281715512275696, "learning_rate": 0.0005, "loss": 1.7785987854003906, "step": 19341 }, { "epoch": 1.5247930626724477, "grad_norm": 0.5450688004493713, "learning_rate": 0.0005, "loss": 1.5710272789001465, "step": 19342 }, { "epoch": 1.5248718959400867, "grad_norm": 0.6592721343040466, "learning_rate": 0.0005, "loss": 1.4584999084472656, "step": 19343 }, { "epoch": 1.5249507292077258, "grad_norm": 0.5704932808876038, "learning_rate": 0.0005, "loss": 1.4373064041137695, "step": 19344 }, { "epoch": 1.5250295624753645, "grad_norm": 0.45566850900650024, "learning_rate": 0.0005, "loss": 1.319812297821045, "step": 19345 }, { "epoch": 1.5251083957430036, "grad_norm": 0.5568011403083801, "learning_rate": 0.0005, "loss": 1.7148154973983765, "step": 19346 }, { "epoch": 1.5251872290106425, "grad_norm": 0.5682177543640137, "learning_rate": 0.0005, "loss": 1.5678775310516357, "step": 19347 }, { "epoch": 1.5252660622782814, "grad_norm": 0.6464712619781494, "learning_rate": 0.0005, "loss": 1.4908076524734497, "step": 19348 }, { "epoch": 1.5253448955459203, "grad_norm": 0.5520623922348022, "learning_rate": 0.0005, "loss": 1.603163242340088, "step": 19349 }, { "epoch": 1.5254237288135593, "grad_norm": 0.6719757318496704, "learning_rate": 0.0005, "loss": 1.4596550464630127, "step": 19350 }, { "epoch": 1.5255025620811984, "grad_norm": 1.0769610404968262, "learning_rate": 0.0005, "loss": 1.67398202419281, "step": 19351 }, { "epoch": 1.525581395348837, "grad_norm": 0.5486689805984497, "learning_rate": 0.0005, "loss": 1.422568917274475, "step": 19352 }, { "epoch": 1.5256602286164762, "grad_norm": 0.5387164354324341, "learning_rate": 0.0005, "loss": 1.7278118133544922, "step": 19353 }, { "epoch": 1.5257390618841151, "grad_norm": 0.542275071144104, "learning_rate": 0.0005, "loss": 1.5870741605758667, "step": 19354 }, { "epoch": 1.525817895151754, "grad_norm": 0.726323664188385, "learning_rate": 0.0005, "loss": 1.5929548740386963, "step": 19355 }, { "epoch": 1.525896728419393, "grad_norm": 0.7409104704856873, "learning_rate": 0.0005, "loss": 1.5699882507324219, "step": 19356 }, { "epoch": 1.5259755616870319, "grad_norm": 0.6182475090026855, "learning_rate": 0.0005, "loss": 1.5952811241149902, "step": 19357 }, { "epoch": 1.526054394954671, "grad_norm": 0.440968781709671, "learning_rate": 0.0005, "loss": 1.170052170753479, "step": 19358 }, { "epoch": 1.5261332282223097, "grad_norm": 0.6715908050537109, "learning_rate": 0.0005, "loss": 1.550886869430542, "step": 19359 }, { "epoch": 1.5262120614899488, "grad_norm": 0.7371456623077393, "learning_rate": 0.0005, "loss": 1.617647647857666, "step": 19360 }, { "epoch": 1.5262908947575877, "grad_norm": 0.5688115358352661, "learning_rate": 0.0005, "loss": 1.4772536754608154, "step": 19361 }, { "epoch": 1.5263697280252266, "grad_norm": 0.598503053188324, "learning_rate": 0.0005, "loss": 1.6248667240142822, "step": 19362 }, { "epoch": 1.5264485612928655, "grad_norm": 0.48790594935417175, "learning_rate": 0.0005, "loss": 1.231553554534912, "step": 19363 }, { "epoch": 1.5265273945605045, "grad_norm": 0.5022356510162354, "learning_rate": 0.0005, "loss": 1.3833434581756592, "step": 19364 }, { "epoch": 1.5266062278281436, "grad_norm": 0.5287041664123535, "learning_rate": 0.0005, "loss": 1.4700373411178589, "step": 19365 }, { "epoch": 1.5266850610957823, "grad_norm": 0.4438592493534088, "learning_rate": 0.0005, "loss": 1.417677402496338, "step": 19366 }, { "epoch": 1.5267638943634214, "grad_norm": 0.6203730702400208, "learning_rate": 0.0005, "loss": 1.7563867568969727, "step": 19367 }, { "epoch": 1.5268427276310603, "grad_norm": 0.4582865536212921, "learning_rate": 0.0005, "loss": 1.4242630004882812, "step": 19368 }, { "epoch": 1.5269215608986992, "grad_norm": 0.508272647857666, "learning_rate": 0.0005, "loss": 1.4347286224365234, "step": 19369 }, { "epoch": 1.5270003941663381, "grad_norm": 0.6258699297904968, "learning_rate": 0.0005, "loss": 1.5245647430419922, "step": 19370 }, { "epoch": 1.527079227433977, "grad_norm": 0.58566814661026, "learning_rate": 0.0005, "loss": 1.5244762897491455, "step": 19371 }, { "epoch": 1.5271580607016162, "grad_norm": 0.6333565711975098, "learning_rate": 0.0005, "loss": 1.5574238300323486, "step": 19372 }, { "epoch": 1.5272368939692549, "grad_norm": 0.5748895406723022, "learning_rate": 0.0005, "loss": 1.604187250137329, "step": 19373 }, { "epoch": 1.527315727236894, "grad_norm": 0.45083722472190857, "learning_rate": 0.0005, "loss": 1.3944878578186035, "step": 19374 }, { "epoch": 1.527394560504533, "grad_norm": 0.591529130935669, "learning_rate": 0.0005, "loss": 1.740501880645752, "step": 19375 }, { "epoch": 1.5274733937721718, "grad_norm": 0.49848419427871704, "learning_rate": 0.0005, "loss": 1.535544514656067, "step": 19376 }, { "epoch": 1.5275522270398108, "grad_norm": 0.470573365688324, "learning_rate": 0.0005, "loss": 1.6377971172332764, "step": 19377 }, { "epoch": 1.5276310603074497, "grad_norm": 0.45418354868888855, "learning_rate": 0.0005, "loss": 1.5211206674575806, "step": 19378 }, { "epoch": 1.5277098935750888, "grad_norm": 0.5395928621292114, "learning_rate": 0.0005, "loss": 1.6316356658935547, "step": 19379 }, { "epoch": 1.5277887268427275, "grad_norm": 0.4818088114261627, "learning_rate": 0.0005, "loss": 1.5879158973693848, "step": 19380 }, { "epoch": 1.5278675601103666, "grad_norm": 0.4304219186306, "learning_rate": 0.0005, "loss": 1.4138436317443848, "step": 19381 }, { "epoch": 1.5279463933780055, "grad_norm": 0.6183920502662659, "learning_rate": 0.0005, "loss": 1.6063799858093262, "step": 19382 }, { "epoch": 1.5280252266456444, "grad_norm": 0.4652847647666931, "learning_rate": 0.0005, "loss": 1.3859151601791382, "step": 19383 }, { "epoch": 1.5281040599132834, "grad_norm": 0.6666008830070496, "learning_rate": 0.0005, "loss": 1.4911093711853027, "step": 19384 }, { "epoch": 1.5281828931809223, "grad_norm": 0.6058593988418579, "learning_rate": 0.0005, "loss": 1.5617365837097168, "step": 19385 }, { "epoch": 1.5282617264485614, "grad_norm": 0.6309180855751038, "learning_rate": 0.0005, "loss": 1.4730576276779175, "step": 19386 }, { "epoch": 1.5283405597162, "grad_norm": 0.4552272856235504, "learning_rate": 0.0005, "loss": 1.5259766578674316, "step": 19387 }, { "epoch": 1.5284193929838392, "grad_norm": 0.4975752532482147, "learning_rate": 0.0005, "loss": 1.645035982131958, "step": 19388 }, { "epoch": 1.5284982262514781, "grad_norm": 0.7111083269119263, "learning_rate": 0.0005, "loss": 1.5215139389038086, "step": 19389 }, { "epoch": 1.528577059519117, "grad_norm": 0.5652013421058655, "learning_rate": 0.0005, "loss": 1.4986870288848877, "step": 19390 }, { "epoch": 1.5286558927867562, "grad_norm": 0.5369547009468079, "learning_rate": 0.0005, "loss": 1.7171283960342407, "step": 19391 }, { "epoch": 1.5287347260543949, "grad_norm": 0.8371809720993042, "learning_rate": 0.0005, "loss": 1.4176933765411377, "step": 19392 }, { "epoch": 1.528813559322034, "grad_norm": 0.6441593766212463, "learning_rate": 0.0005, "loss": 1.6457099914550781, "step": 19393 }, { "epoch": 1.5288923925896727, "grad_norm": 0.6895162463188171, "learning_rate": 0.0005, "loss": 1.495711088180542, "step": 19394 }, { "epoch": 1.5289712258573118, "grad_norm": 0.5833298563957214, "learning_rate": 0.0005, "loss": 1.7845863103866577, "step": 19395 }, { "epoch": 1.5290500591249507, "grad_norm": 0.42367836833000183, "learning_rate": 0.0005, "loss": 1.3184889554977417, "step": 19396 }, { "epoch": 1.5291288923925896, "grad_norm": 0.5140174031257629, "learning_rate": 0.0005, "loss": 1.6160273551940918, "step": 19397 }, { "epoch": 1.5292077256602288, "grad_norm": 0.4856659770011902, "learning_rate": 0.0005, "loss": 1.6238558292388916, "step": 19398 }, { "epoch": 1.5292865589278675, "grad_norm": 0.4525842070579529, "learning_rate": 0.0005, "loss": 1.5541510581970215, "step": 19399 }, { "epoch": 1.5293653921955066, "grad_norm": 0.5695974826812744, "learning_rate": 0.0005, "loss": 1.6813743114471436, "step": 19400 }, { "epoch": 1.5294442254631453, "grad_norm": 0.5213729739189148, "learning_rate": 0.0005, "loss": 1.3803315162658691, "step": 19401 }, { "epoch": 1.5295230587307844, "grad_norm": 0.505635142326355, "learning_rate": 0.0005, "loss": 1.6291697025299072, "step": 19402 }, { "epoch": 1.5296018919984233, "grad_norm": 0.4427184760570526, "learning_rate": 0.0005, "loss": 1.4106323719024658, "step": 19403 }, { "epoch": 1.5296807252660622, "grad_norm": 0.5284523367881775, "learning_rate": 0.0005, "loss": 1.5373493432998657, "step": 19404 }, { "epoch": 1.5297595585337014, "grad_norm": 0.5738934874534607, "learning_rate": 0.0005, "loss": 1.630420446395874, "step": 19405 }, { "epoch": 1.52983839180134, "grad_norm": 0.48752957582473755, "learning_rate": 0.0005, "loss": 1.7676990032196045, "step": 19406 }, { "epoch": 1.5299172250689792, "grad_norm": 0.42141813039779663, "learning_rate": 0.0005, "loss": 1.491835594177246, "step": 19407 }, { "epoch": 1.529996058336618, "grad_norm": 0.6723464131355286, "learning_rate": 0.0005, "loss": 1.35422682762146, "step": 19408 }, { "epoch": 1.530074891604257, "grad_norm": 0.653671145439148, "learning_rate": 0.0005, "loss": 1.5404072999954224, "step": 19409 }, { "epoch": 1.530153724871896, "grad_norm": 0.563911497592926, "learning_rate": 0.0005, "loss": 1.734060287475586, "step": 19410 }, { "epoch": 1.5302325581395348, "grad_norm": 0.47046688199043274, "learning_rate": 0.0005, "loss": 1.4184602499008179, "step": 19411 }, { "epoch": 1.530311391407174, "grad_norm": 0.47834163904190063, "learning_rate": 0.0005, "loss": 1.514030933380127, "step": 19412 }, { "epoch": 1.5303902246748127, "grad_norm": 0.696757972240448, "learning_rate": 0.0005, "loss": 1.7497904300689697, "step": 19413 }, { "epoch": 1.5304690579424518, "grad_norm": 0.6311876177787781, "learning_rate": 0.0005, "loss": 1.6233779191970825, "step": 19414 }, { "epoch": 1.5305478912100905, "grad_norm": 0.5061299800872803, "learning_rate": 0.0005, "loss": 1.3606688976287842, "step": 19415 }, { "epoch": 1.5306267244777296, "grad_norm": 0.5198113918304443, "learning_rate": 0.0005, "loss": 1.3848744630813599, "step": 19416 }, { "epoch": 1.5307055577453685, "grad_norm": 0.5151234865188599, "learning_rate": 0.0005, "loss": 1.393184781074524, "step": 19417 }, { "epoch": 1.5307843910130075, "grad_norm": 0.5711473226547241, "learning_rate": 0.0005, "loss": 1.4394965171813965, "step": 19418 }, { "epoch": 1.5308632242806466, "grad_norm": 0.49051475524902344, "learning_rate": 0.0005, "loss": 1.4795360565185547, "step": 19419 }, { "epoch": 1.5309420575482853, "grad_norm": 0.5340608954429626, "learning_rate": 0.0005, "loss": 1.7864782810211182, "step": 19420 }, { "epoch": 1.5310208908159244, "grad_norm": 0.7921137809753418, "learning_rate": 0.0005, "loss": 1.585490107536316, "step": 19421 }, { "epoch": 1.5310997240835633, "grad_norm": 0.5838379263877869, "learning_rate": 0.0005, "loss": 1.53501296043396, "step": 19422 }, { "epoch": 1.5311785573512022, "grad_norm": 0.579537034034729, "learning_rate": 0.0005, "loss": 1.4521247148513794, "step": 19423 }, { "epoch": 1.5312573906188411, "grad_norm": 0.5032334923744202, "learning_rate": 0.0005, "loss": 1.6496647596359253, "step": 19424 }, { "epoch": 1.53133622388648, "grad_norm": 0.5378977060317993, "learning_rate": 0.0005, "loss": 1.4760452508926392, "step": 19425 }, { "epoch": 1.5314150571541192, "grad_norm": 0.5965228080749512, "learning_rate": 0.0005, "loss": 1.5241073369979858, "step": 19426 }, { "epoch": 1.5314938904217579, "grad_norm": 0.6666907668113708, "learning_rate": 0.0005, "loss": 1.9133687019348145, "step": 19427 }, { "epoch": 1.531572723689397, "grad_norm": 0.5877796411514282, "learning_rate": 0.0005, "loss": 1.593907356262207, "step": 19428 }, { "epoch": 1.531651556957036, "grad_norm": 0.46532759070396423, "learning_rate": 0.0005, "loss": 1.3258185386657715, "step": 19429 }, { "epoch": 1.5317303902246748, "grad_norm": 0.5696181654930115, "learning_rate": 0.0005, "loss": 1.3855634927749634, "step": 19430 }, { "epoch": 1.5318092234923137, "grad_norm": 0.6014952659606934, "learning_rate": 0.0005, "loss": 1.471851110458374, "step": 19431 }, { "epoch": 1.5318880567599527, "grad_norm": 0.6868013143539429, "learning_rate": 0.0005, "loss": 1.572101354598999, "step": 19432 }, { "epoch": 1.5319668900275918, "grad_norm": 0.5133861899375916, "learning_rate": 0.0005, "loss": 1.6174167394638062, "step": 19433 }, { "epoch": 1.5320457232952305, "grad_norm": 0.844113826751709, "learning_rate": 0.0005, "loss": 1.807054042816162, "step": 19434 }, { "epoch": 1.5321245565628696, "grad_norm": 0.5294204354286194, "learning_rate": 0.0005, "loss": 1.504568099975586, "step": 19435 }, { "epoch": 1.5322033898305085, "grad_norm": 0.5391306281089783, "learning_rate": 0.0005, "loss": 1.3975167274475098, "step": 19436 }, { "epoch": 1.5322822230981474, "grad_norm": 0.6526678204536438, "learning_rate": 0.0005, "loss": 1.4622080326080322, "step": 19437 }, { "epoch": 1.5323610563657863, "grad_norm": 0.5084076523780823, "learning_rate": 0.0005, "loss": 1.7570409774780273, "step": 19438 }, { "epoch": 1.5324398896334253, "grad_norm": 0.6101983189582825, "learning_rate": 0.0005, "loss": 1.6158431768417358, "step": 19439 }, { "epoch": 1.5325187229010644, "grad_norm": 0.5567003488540649, "learning_rate": 0.0005, "loss": 1.6666314601898193, "step": 19440 }, { "epoch": 1.532597556168703, "grad_norm": 0.4917961657047272, "learning_rate": 0.0005, "loss": 1.3440537452697754, "step": 19441 }, { "epoch": 1.5326763894363422, "grad_norm": 0.49139904975891113, "learning_rate": 0.0005, "loss": 1.5964579582214355, "step": 19442 }, { "epoch": 1.5327552227039811, "grad_norm": 0.5795454382896423, "learning_rate": 0.0005, "loss": 1.9663403034210205, "step": 19443 }, { "epoch": 1.53283405597162, "grad_norm": 0.5201746821403503, "learning_rate": 0.0005, "loss": 1.5116897821426392, "step": 19444 }, { "epoch": 1.532912889239259, "grad_norm": 0.5841438174247742, "learning_rate": 0.0005, "loss": 1.5591603517532349, "step": 19445 }, { "epoch": 1.5329917225068979, "grad_norm": 0.6789024472236633, "learning_rate": 0.0005, "loss": 1.7230650186538696, "step": 19446 }, { "epoch": 1.533070555774537, "grad_norm": 0.5461655259132385, "learning_rate": 0.0005, "loss": 1.552613377571106, "step": 19447 }, { "epoch": 1.5331493890421757, "grad_norm": 0.5721172094345093, "learning_rate": 0.0005, "loss": 1.6633336544036865, "step": 19448 }, { "epoch": 1.5332282223098148, "grad_norm": 0.571118950843811, "learning_rate": 0.0005, "loss": 1.4958925247192383, "step": 19449 }, { "epoch": 1.5333070555774537, "grad_norm": 0.5419797301292419, "learning_rate": 0.0005, "loss": 1.7297477722167969, "step": 19450 }, { "epoch": 1.5333858888450926, "grad_norm": 0.5383723378181458, "learning_rate": 0.0005, "loss": 1.6223492622375488, "step": 19451 }, { "epoch": 1.5334647221127315, "grad_norm": 0.5922327637672424, "learning_rate": 0.0005, "loss": 1.5611419677734375, "step": 19452 }, { "epoch": 1.5335435553803705, "grad_norm": 0.5680668950080872, "learning_rate": 0.0005, "loss": 1.6488877534866333, "step": 19453 }, { "epoch": 1.5336223886480096, "grad_norm": 0.4860149919986725, "learning_rate": 0.0005, "loss": 1.6174038648605347, "step": 19454 }, { "epoch": 1.5337012219156483, "grad_norm": 0.6525938510894775, "learning_rate": 0.0005, "loss": 1.6713671684265137, "step": 19455 }, { "epoch": 1.5337800551832874, "grad_norm": 0.7500918507575989, "learning_rate": 0.0005, "loss": 1.6580924987792969, "step": 19456 }, { "epoch": 1.5338588884509263, "grad_norm": 0.5326566100120544, "learning_rate": 0.0005, "loss": 1.4233299493789673, "step": 19457 }, { "epoch": 1.5339377217185652, "grad_norm": 0.5144008994102478, "learning_rate": 0.0005, "loss": 1.3265798091888428, "step": 19458 }, { "epoch": 1.5340165549862042, "grad_norm": 0.6197567582130432, "learning_rate": 0.0005, "loss": 1.6540768146514893, "step": 19459 }, { "epoch": 1.534095388253843, "grad_norm": 0.5268070101737976, "learning_rate": 0.0005, "loss": 1.5352401733398438, "step": 19460 }, { "epoch": 1.5341742215214822, "grad_norm": 0.6147392988204956, "learning_rate": 0.0005, "loss": 1.4902853965759277, "step": 19461 }, { "epoch": 1.5342530547891209, "grad_norm": 0.5198679566383362, "learning_rate": 0.0005, "loss": 1.6121257543563843, "step": 19462 }, { "epoch": 1.53433188805676, "grad_norm": 0.4811047613620758, "learning_rate": 0.0005, "loss": 1.4689691066741943, "step": 19463 }, { "epoch": 1.534410721324399, "grad_norm": 0.6149007678031921, "learning_rate": 0.0005, "loss": 1.5004220008850098, "step": 19464 }, { "epoch": 1.5344895545920378, "grad_norm": 0.5572224855422974, "learning_rate": 0.0005, "loss": 1.6598831415176392, "step": 19465 }, { "epoch": 1.5345683878596768, "grad_norm": 0.6234301924705505, "learning_rate": 0.0005, "loss": 1.684316873550415, "step": 19466 }, { "epoch": 1.5346472211273157, "grad_norm": 0.4961688220500946, "learning_rate": 0.0005, "loss": 1.4841338396072388, "step": 19467 }, { "epoch": 1.5347260543949548, "grad_norm": 0.5416467189788818, "learning_rate": 0.0005, "loss": 1.6022570133209229, "step": 19468 }, { "epoch": 1.5348048876625935, "grad_norm": 0.5532649159431458, "learning_rate": 0.0005, "loss": 1.479647159576416, "step": 19469 }, { "epoch": 1.5348837209302326, "grad_norm": 0.47538644075393677, "learning_rate": 0.0005, "loss": 1.367937684059143, "step": 19470 }, { "epoch": 1.5349625541978715, "grad_norm": 0.45285600423812866, "learning_rate": 0.0005, "loss": 1.3778223991394043, "step": 19471 }, { "epoch": 1.5350413874655104, "grad_norm": 0.5839003920555115, "learning_rate": 0.0005, "loss": 1.5682129859924316, "step": 19472 }, { "epoch": 1.5351202207331494, "grad_norm": 0.5632314085960388, "learning_rate": 0.0005, "loss": 1.7651667594909668, "step": 19473 }, { "epoch": 1.5351990540007883, "grad_norm": 0.5662464499473572, "learning_rate": 0.0005, "loss": 1.564499020576477, "step": 19474 }, { "epoch": 1.5352778872684274, "grad_norm": 0.4812318682670593, "learning_rate": 0.0005, "loss": 1.3342273235321045, "step": 19475 }, { "epoch": 1.535356720536066, "grad_norm": 0.569506049156189, "learning_rate": 0.0005, "loss": 1.554553508758545, "step": 19476 }, { "epoch": 1.5354355538037052, "grad_norm": 0.5755322575569153, "learning_rate": 0.0005, "loss": 1.5436012744903564, "step": 19477 }, { "epoch": 1.5355143870713441, "grad_norm": 0.7930798530578613, "learning_rate": 0.0005, "loss": 1.5017435550689697, "step": 19478 }, { "epoch": 1.535593220338983, "grad_norm": 0.5510436296463013, "learning_rate": 0.0005, "loss": 1.366492509841919, "step": 19479 }, { "epoch": 1.535672053606622, "grad_norm": 0.4452045261859894, "learning_rate": 0.0005, "loss": 1.5772478580474854, "step": 19480 }, { "epoch": 1.5357508868742609, "grad_norm": 0.5708244442939758, "learning_rate": 0.0005, "loss": 1.515848159790039, "step": 19481 }, { "epoch": 1.5358297201419, "grad_norm": 0.46795448660850525, "learning_rate": 0.0005, "loss": 1.5518380403518677, "step": 19482 }, { "epoch": 1.5359085534095387, "grad_norm": 0.5952281355857849, "learning_rate": 0.0005, "loss": 1.4736433029174805, "step": 19483 }, { "epoch": 1.5359873866771778, "grad_norm": 0.4700448513031006, "learning_rate": 0.0005, "loss": 1.4472627639770508, "step": 19484 }, { "epoch": 1.5360662199448167, "grad_norm": 0.5715034008026123, "learning_rate": 0.0005, "loss": 1.7266680002212524, "step": 19485 }, { "epoch": 1.5361450532124556, "grad_norm": 0.4378282427787781, "learning_rate": 0.0005, "loss": 1.376562237739563, "step": 19486 }, { "epoch": 1.5362238864800946, "grad_norm": 0.5286903381347656, "learning_rate": 0.0005, "loss": 1.4291484355926514, "step": 19487 }, { "epoch": 1.5363027197477335, "grad_norm": 0.4633185565471649, "learning_rate": 0.0005, "loss": 1.4216346740722656, "step": 19488 }, { "epoch": 1.5363815530153726, "grad_norm": 0.5866735577583313, "learning_rate": 0.0005, "loss": 1.5276036262512207, "step": 19489 }, { "epoch": 1.5364603862830113, "grad_norm": 0.47343379259109497, "learning_rate": 0.0005, "loss": 1.5343341827392578, "step": 19490 }, { "epoch": 1.5365392195506504, "grad_norm": 0.5301913619041443, "learning_rate": 0.0005, "loss": 1.540318489074707, "step": 19491 }, { "epoch": 1.5366180528182893, "grad_norm": 0.521561324596405, "learning_rate": 0.0005, "loss": 1.4939794540405273, "step": 19492 }, { "epoch": 1.5366968860859282, "grad_norm": 0.5808326601982117, "learning_rate": 0.0005, "loss": 1.5013659000396729, "step": 19493 }, { "epoch": 1.5367757193535672, "grad_norm": 0.5722569227218628, "learning_rate": 0.0005, "loss": 1.4322110414505005, "step": 19494 }, { "epoch": 1.536854552621206, "grad_norm": 0.5399038791656494, "learning_rate": 0.0005, "loss": 1.56930673122406, "step": 19495 }, { "epoch": 1.5369333858888452, "grad_norm": 0.5172888040542603, "learning_rate": 0.0005, "loss": 1.5122349262237549, "step": 19496 }, { "epoch": 1.537012219156484, "grad_norm": 0.6021977066993713, "learning_rate": 0.0005, "loss": 1.6706912517547607, "step": 19497 }, { "epoch": 1.537091052424123, "grad_norm": 0.602043628692627, "learning_rate": 0.0005, "loss": 1.46766996383667, "step": 19498 }, { "epoch": 1.537169885691762, "grad_norm": 0.6133300065994263, "learning_rate": 0.0005, "loss": 1.7362258434295654, "step": 19499 }, { "epoch": 1.5372487189594009, "grad_norm": 0.49415504932403564, "learning_rate": 0.0005, "loss": 1.4978291988372803, "step": 19500 }, { "epoch": 1.5373275522270398, "grad_norm": 0.455116868019104, "learning_rate": 0.0005, "loss": 1.3943562507629395, "step": 19501 }, { "epoch": 1.5374063854946787, "grad_norm": 0.6172548532485962, "learning_rate": 0.0005, "loss": 1.529006004333496, "step": 19502 }, { "epoch": 1.5374852187623178, "grad_norm": 0.6356939077377319, "learning_rate": 0.0005, "loss": 1.6480364799499512, "step": 19503 }, { "epoch": 1.5375640520299565, "grad_norm": 0.6122215986251831, "learning_rate": 0.0005, "loss": 1.7291431427001953, "step": 19504 }, { "epoch": 1.5376428852975956, "grad_norm": 0.5876595377922058, "learning_rate": 0.0005, "loss": 1.629155158996582, "step": 19505 }, { "epoch": 1.5377217185652345, "grad_norm": 0.6147444844245911, "learning_rate": 0.0005, "loss": 1.5043801069259644, "step": 19506 }, { "epoch": 1.5378005518328735, "grad_norm": 0.5523903369903564, "learning_rate": 0.0005, "loss": 1.6755404472351074, "step": 19507 }, { "epoch": 1.5378793851005124, "grad_norm": 0.5704623460769653, "learning_rate": 0.0005, "loss": 1.7262389659881592, "step": 19508 }, { "epoch": 1.5379582183681513, "grad_norm": 0.6171040534973145, "learning_rate": 0.0005, "loss": 1.5692545175552368, "step": 19509 }, { "epoch": 1.5380370516357904, "grad_norm": 0.4577794373035431, "learning_rate": 0.0005, "loss": 1.4530425071716309, "step": 19510 }, { "epoch": 1.538115884903429, "grad_norm": 0.5474368333816528, "learning_rate": 0.0005, "loss": 1.5108413696289062, "step": 19511 }, { "epoch": 1.5381947181710682, "grad_norm": 0.5464686155319214, "learning_rate": 0.0005, "loss": 1.740654468536377, "step": 19512 }, { "epoch": 1.5382735514387071, "grad_norm": 0.46768632531166077, "learning_rate": 0.0005, "loss": 1.5298992395401, "step": 19513 }, { "epoch": 1.538352384706346, "grad_norm": 0.5862149000167847, "learning_rate": 0.0005, "loss": 1.5688090324401855, "step": 19514 }, { "epoch": 1.5384312179739852, "grad_norm": 0.5979347825050354, "learning_rate": 0.0005, "loss": 1.6520253419876099, "step": 19515 }, { "epoch": 1.5385100512416239, "grad_norm": 0.5203544497489929, "learning_rate": 0.0005, "loss": 1.5591753721237183, "step": 19516 }, { "epoch": 1.538588884509263, "grad_norm": 0.49849820137023926, "learning_rate": 0.0005, "loss": 1.4020934104919434, "step": 19517 }, { "epoch": 1.5386677177769017, "grad_norm": 0.6862909197807312, "learning_rate": 0.0005, "loss": 1.6325374841690063, "step": 19518 }, { "epoch": 1.5387465510445408, "grad_norm": 0.5005233883857727, "learning_rate": 0.0005, "loss": 1.4631235599517822, "step": 19519 }, { "epoch": 1.5388253843121797, "grad_norm": 0.5421812534332275, "learning_rate": 0.0005, "loss": 1.6123398542404175, "step": 19520 }, { "epoch": 1.5389042175798187, "grad_norm": 0.8085467219352722, "learning_rate": 0.0005, "loss": 1.750006079673767, "step": 19521 }, { "epoch": 1.5389830508474578, "grad_norm": 0.5330812931060791, "learning_rate": 0.0005, "loss": 1.7022403478622437, "step": 19522 }, { "epoch": 1.5390618841150965, "grad_norm": 0.6152060627937317, "learning_rate": 0.0005, "loss": 1.7177983522415161, "step": 19523 }, { "epoch": 1.5391407173827356, "grad_norm": 0.5403072834014893, "learning_rate": 0.0005, "loss": 1.5395047664642334, "step": 19524 }, { "epoch": 1.5392195506503743, "grad_norm": 0.4786369502544403, "learning_rate": 0.0005, "loss": 1.2382503747940063, "step": 19525 }, { "epoch": 1.5392983839180134, "grad_norm": 0.47108274698257446, "learning_rate": 0.0005, "loss": 1.3870772123336792, "step": 19526 }, { "epoch": 1.5393772171856523, "grad_norm": 0.6351776123046875, "learning_rate": 0.0005, "loss": 1.5372412204742432, "step": 19527 }, { "epoch": 1.5394560504532913, "grad_norm": 0.6446002125740051, "learning_rate": 0.0005, "loss": 1.543208122253418, "step": 19528 }, { "epoch": 1.5395348837209304, "grad_norm": 0.654339075088501, "learning_rate": 0.0005, "loss": 1.5426284074783325, "step": 19529 }, { "epoch": 1.539613716988569, "grad_norm": 0.5921127796173096, "learning_rate": 0.0005, "loss": 1.7043728828430176, "step": 19530 }, { "epoch": 1.5396925502562082, "grad_norm": 0.5325080752372742, "learning_rate": 0.0005, "loss": 1.5033117532730103, "step": 19531 }, { "epoch": 1.539771383523847, "grad_norm": 0.4295633137226105, "learning_rate": 0.0005, "loss": 1.3158788681030273, "step": 19532 }, { "epoch": 1.539850216791486, "grad_norm": 0.5279756784439087, "learning_rate": 0.0005, "loss": 1.394982099533081, "step": 19533 }, { "epoch": 1.539929050059125, "grad_norm": 0.5806331634521484, "learning_rate": 0.0005, "loss": 1.6573359966278076, "step": 19534 }, { "epoch": 1.5400078833267639, "grad_norm": 0.5281479954719543, "learning_rate": 0.0005, "loss": 1.4858498573303223, "step": 19535 }, { "epoch": 1.540086716594403, "grad_norm": 0.5390262007713318, "learning_rate": 0.0005, "loss": 1.5598950386047363, "step": 19536 }, { "epoch": 1.5401655498620417, "grad_norm": 0.6385834813117981, "learning_rate": 0.0005, "loss": 1.5996195077896118, "step": 19537 }, { "epoch": 1.5402443831296808, "grad_norm": 0.5597723722457886, "learning_rate": 0.0005, "loss": 1.5475932359695435, "step": 19538 }, { "epoch": 1.5403232163973195, "grad_norm": 0.6497030854225159, "learning_rate": 0.0005, "loss": 1.6075656414031982, "step": 19539 }, { "epoch": 1.5404020496649586, "grad_norm": 0.5979787707328796, "learning_rate": 0.0005, "loss": 1.5923585891723633, "step": 19540 }, { "epoch": 1.5404808829325976, "grad_norm": 1.1190487146377563, "learning_rate": 0.0005, "loss": 1.5791819095611572, "step": 19541 }, { "epoch": 1.5405597162002365, "grad_norm": 0.5100627541542053, "learning_rate": 0.0005, "loss": 1.4546111822128296, "step": 19542 }, { "epoch": 1.5406385494678756, "grad_norm": 0.4773593544960022, "learning_rate": 0.0005, "loss": 1.3517889976501465, "step": 19543 }, { "epoch": 1.5407173827355143, "grad_norm": 0.5134219527244568, "learning_rate": 0.0005, "loss": 1.610304832458496, "step": 19544 }, { "epoch": 1.5407962160031534, "grad_norm": 0.5806288719177246, "learning_rate": 0.0005, "loss": 1.562691330909729, "step": 19545 }, { "epoch": 1.5408750492707923, "grad_norm": 0.6640546917915344, "learning_rate": 0.0005, "loss": 1.716275930404663, "step": 19546 }, { "epoch": 1.5409538825384312, "grad_norm": 0.5803289413452148, "learning_rate": 0.0005, "loss": 1.7524194717407227, "step": 19547 }, { "epoch": 1.5410327158060702, "grad_norm": 0.4416132867336273, "learning_rate": 0.0005, "loss": 1.3738999366760254, "step": 19548 }, { "epoch": 1.541111549073709, "grad_norm": 0.6507914662361145, "learning_rate": 0.0005, "loss": 1.7455542087554932, "step": 19549 }, { "epoch": 1.5411903823413482, "grad_norm": 0.5002096891403198, "learning_rate": 0.0005, "loss": 1.6543530225753784, "step": 19550 }, { "epoch": 1.5412692156089869, "grad_norm": 0.4252298176288605, "learning_rate": 0.0005, "loss": 1.3788570165634155, "step": 19551 }, { "epoch": 1.541348048876626, "grad_norm": 0.46657541394233704, "learning_rate": 0.0005, "loss": 1.5666606426239014, "step": 19552 }, { "epoch": 1.541426882144265, "grad_norm": 0.4588630497455597, "learning_rate": 0.0005, "loss": 1.36967134475708, "step": 19553 }, { "epoch": 1.5415057154119038, "grad_norm": 0.5270981192588806, "learning_rate": 0.0005, "loss": 1.6469182968139648, "step": 19554 }, { "epoch": 1.5415845486795428, "grad_norm": 0.5456660985946655, "learning_rate": 0.0005, "loss": 1.4946482181549072, "step": 19555 }, { "epoch": 1.5416633819471817, "grad_norm": 0.4438604712486267, "learning_rate": 0.0005, "loss": 1.3508055210113525, "step": 19556 }, { "epoch": 1.5417422152148208, "grad_norm": 0.4605864882469177, "learning_rate": 0.0005, "loss": 1.3612301349639893, "step": 19557 }, { "epoch": 1.5418210484824595, "grad_norm": 0.5532088279724121, "learning_rate": 0.0005, "loss": 1.6999497413635254, "step": 19558 }, { "epoch": 1.5418998817500986, "grad_norm": 0.6067919731140137, "learning_rate": 0.0005, "loss": 1.7744228839874268, "step": 19559 }, { "epoch": 1.5419787150177375, "grad_norm": 0.5843632817268372, "learning_rate": 0.0005, "loss": 1.4628448486328125, "step": 19560 }, { "epoch": 1.5420575482853764, "grad_norm": 0.47923338413238525, "learning_rate": 0.0005, "loss": 1.692469596862793, "step": 19561 }, { "epoch": 1.5421363815530154, "grad_norm": 0.4674394726753235, "learning_rate": 0.0005, "loss": 1.307719349861145, "step": 19562 }, { "epoch": 1.5422152148206543, "grad_norm": 0.5109938383102417, "learning_rate": 0.0005, "loss": 1.5770642757415771, "step": 19563 }, { "epoch": 1.5422940480882934, "grad_norm": 0.5700423717498779, "learning_rate": 0.0005, "loss": 1.784738302230835, "step": 19564 }, { "epoch": 1.542372881355932, "grad_norm": 0.9343637228012085, "learning_rate": 0.0005, "loss": 1.5990935564041138, "step": 19565 }, { "epoch": 1.5424517146235712, "grad_norm": 0.5563976764678955, "learning_rate": 0.0005, "loss": 1.7833123207092285, "step": 19566 }, { "epoch": 1.5425305478912101, "grad_norm": 0.6130082607269287, "learning_rate": 0.0005, "loss": 1.4010043144226074, "step": 19567 }, { "epoch": 1.542609381158849, "grad_norm": 0.500282883644104, "learning_rate": 0.0005, "loss": 1.6472446918487549, "step": 19568 }, { "epoch": 1.542688214426488, "grad_norm": 0.5261356830596924, "learning_rate": 0.0005, "loss": 1.5160717964172363, "step": 19569 }, { "epoch": 1.5427670476941269, "grad_norm": 0.5387008786201477, "learning_rate": 0.0005, "loss": 1.6738909482955933, "step": 19570 }, { "epoch": 1.542845880961766, "grad_norm": 0.6119384765625, "learning_rate": 0.0005, "loss": 1.4148563146591187, "step": 19571 }, { "epoch": 1.5429247142294047, "grad_norm": 0.6397662162780762, "learning_rate": 0.0005, "loss": 1.6534106731414795, "step": 19572 }, { "epoch": 1.5430035474970438, "grad_norm": 0.4709840416908264, "learning_rate": 0.0005, "loss": 1.5140445232391357, "step": 19573 }, { "epoch": 1.5430823807646827, "grad_norm": 0.4879363179206848, "learning_rate": 0.0005, "loss": 1.4378341436386108, "step": 19574 }, { "epoch": 1.5431612140323216, "grad_norm": 0.4545418620109558, "learning_rate": 0.0005, "loss": 1.4523413181304932, "step": 19575 }, { "epoch": 1.5432400472999606, "grad_norm": 0.6762699484825134, "learning_rate": 0.0005, "loss": 1.5918865203857422, "step": 19576 }, { "epoch": 1.5433188805675995, "grad_norm": 0.526179313659668, "learning_rate": 0.0005, "loss": 1.6177234649658203, "step": 19577 }, { "epoch": 1.5433977138352386, "grad_norm": 0.5617743134498596, "learning_rate": 0.0005, "loss": 1.5346266031265259, "step": 19578 }, { "epoch": 1.5434765471028773, "grad_norm": 0.38179656863212585, "learning_rate": 0.0005, "loss": 1.3022143840789795, "step": 19579 }, { "epoch": 1.5435553803705164, "grad_norm": 0.5495828986167908, "learning_rate": 0.0005, "loss": 1.715901255607605, "step": 19580 }, { "epoch": 1.5436342136381553, "grad_norm": 0.5706589221954346, "learning_rate": 0.0005, "loss": 1.482836127281189, "step": 19581 }, { "epoch": 1.5437130469057943, "grad_norm": 0.5275523066520691, "learning_rate": 0.0005, "loss": 1.4805186986923218, "step": 19582 }, { "epoch": 1.5437918801734332, "grad_norm": 0.4647997319698334, "learning_rate": 0.0005, "loss": 1.6786681413650513, "step": 19583 }, { "epoch": 1.543870713441072, "grad_norm": 0.46750348806381226, "learning_rate": 0.0005, "loss": 1.5802063941955566, "step": 19584 }, { "epoch": 1.5439495467087112, "grad_norm": 0.5237088203430176, "learning_rate": 0.0005, "loss": 1.5907623767852783, "step": 19585 }, { "epoch": 1.54402837997635, "grad_norm": 0.7025359869003296, "learning_rate": 0.0005, "loss": 1.5672376155853271, "step": 19586 }, { "epoch": 1.544107213243989, "grad_norm": 0.6465267539024353, "learning_rate": 0.0005, "loss": 1.7529799938201904, "step": 19587 }, { "epoch": 1.544186046511628, "grad_norm": 0.5960948467254639, "learning_rate": 0.0005, "loss": 1.5684998035430908, "step": 19588 }, { "epoch": 1.5442648797792669, "grad_norm": 0.5331440567970276, "learning_rate": 0.0005, "loss": 1.530611515045166, "step": 19589 }, { "epoch": 1.5443437130469058, "grad_norm": 0.5970811247825623, "learning_rate": 0.0005, "loss": 1.458200216293335, "step": 19590 }, { "epoch": 1.5444225463145447, "grad_norm": 0.4358120560646057, "learning_rate": 0.0005, "loss": 1.3579716682434082, "step": 19591 }, { "epoch": 1.5445013795821838, "grad_norm": 0.6511334180831909, "learning_rate": 0.0005, "loss": 1.5114548206329346, "step": 19592 }, { "epoch": 1.5445802128498225, "grad_norm": 0.539556086063385, "learning_rate": 0.0005, "loss": 1.4677815437316895, "step": 19593 }, { "epoch": 1.5446590461174616, "grad_norm": 0.5398144125938416, "learning_rate": 0.0005, "loss": 1.6020679473876953, "step": 19594 }, { "epoch": 1.5447378793851005, "grad_norm": 0.5015532970428467, "learning_rate": 0.0005, "loss": 1.5713393688201904, "step": 19595 }, { "epoch": 1.5448167126527395, "grad_norm": 0.5524657964706421, "learning_rate": 0.0005, "loss": 1.4388046264648438, "step": 19596 }, { "epoch": 1.5448955459203784, "grad_norm": 0.5731117129325867, "learning_rate": 0.0005, "loss": 1.5847136974334717, "step": 19597 }, { "epoch": 1.5449743791880173, "grad_norm": 0.4874648451805115, "learning_rate": 0.0005, "loss": 1.4797680377960205, "step": 19598 }, { "epoch": 1.5450532124556564, "grad_norm": 0.556990385055542, "learning_rate": 0.0005, "loss": 1.7234678268432617, "step": 19599 }, { "epoch": 1.545132045723295, "grad_norm": 0.7499949932098389, "learning_rate": 0.0005, "loss": 1.5299755334854126, "step": 19600 }, { "epoch": 1.5452108789909342, "grad_norm": 0.49247515201568604, "learning_rate": 0.0005, "loss": 1.486886739730835, "step": 19601 }, { "epoch": 1.5452897122585731, "grad_norm": 0.4911351203918457, "learning_rate": 0.0005, "loss": 1.602248191833496, "step": 19602 }, { "epoch": 1.545368545526212, "grad_norm": 0.5256833434104919, "learning_rate": 0.0005, "loss": 1.390932321548462, "step": 19603 }, { "epoch": 1.545447378793851, "grad_norm": 0.5744481682777405, "learning_rate": 0.0005, "loss": 1.6388521194458008, "step": 19604 }, { "epoch": 1.5455262120614899, "grad_norm": 0.5545353889465332, "learning_rate": 0.0005, "loss": 1.517421007156372, "step": 19605 }, { "epoch": 1.545605045329129, "grad_norm": 0.5332380533218384, "learning_rate": 0.0005, "loss": 1.5522093772888184, "step": 19606 }, { "epoch": 1.5456838785967677, "grad_norm": 0.6257928013801575, "learning_rate": 0.0005, "loss": 1.62589430809021, "step": 19607 }, { "epoch": 1.5457627118644068, "grad_norm": 0.4925989806652069, "learning_rate": 0.0005, "loss": 1.5922839641571045, "step": 19608 }, { "epoch": 1.5458415451320457, "grad_norm": 1.1161285638809204, "learning_rate": 0.0005, "loss": 1.951039433479309, "step": 19609 }, { "epoch": 1.5459203783996847, "grad_norm": 0.6730616688728333, "learning_rate": 0.0005, "loss": 1.3711440563201904, "step": 19610 }, { "epoch": 1.5459992116673236, "grad_norm": 0.6180527806282043, "learning_rate": 0.0005, "loss": 1.5706899166107178, "step": 19611 }, { "epoch": 1.5460780449349625, "grad_norm": 0.5142155289649963, "learning_rate": 0.0005, "loss": 1.6015040874481201, "step": 19612 }, { "epoch": 1.5461568782026016, "grad_norm": 0.5685593485832214, "learning_rate": 0.0005, "loss": 1.6320507526397705, "step": 19613 }, { "epoch": 1.5462357114702403, "grad_norm": 0.5599167943000793, "learning_rate": 0.0005, "loss": 1.6700708866119385, "step": 19614 }, { "epoch": 1.5463145447378794, "grad_norm": 0.6168386936187744, "learning_rate": 0.0005, "loss": 1.5622155666351318, "step": 19615 }, { "epoch": 1.5463933780055183, "grad_norm": 0.582665741443634, "learning_rate": 0.0005, "loss": 1.6183395385742188, "step": 19616 }, { "epoch": 1.5464722112731573, "grad_norm": 0.5338843464851379, "learning_rate": 0.0005, "loss": 1.5455546379089355, "step": 19617 }, { "epoch": 1.5465510445407962, "grad_norm": 0.5045832395553589, "learning_rate": 0.0005, "loss": 1.6382009983062744, "step": 19618 }, { "epoch": 1.546629877808435, "grad_norm": 0.566198468208313, "learning_rate": 0.0005, "loss": 1.5696724653244019, "step": 19619 }, { "epoch": 1.5467087110760742, "grad_norm": 0.48350557684898376, "learning_rate": 0.0005, "loss": 1.5088233947753906, "step": 19620 }, { "epoch": 1.546787544343713, "grad_norm": 0.5134986042976379, "learning_rate": 0.0005, "loss": 1.4273691177368164, "step": 19621 }, { "epoch": 1.546866377611352, "grad_norm": 0.5635643005371094, "learning_rate": 0.0005, "loss": 1.6406973600387573, "step": 19622 }, { "epoch": 1.546945210878991, "grad_norm": 0.5840214490890503, "learning_rate": 0.0005, "loss": 1.345428705215454, "step": 19623 }, { "epoch": 1.5470240441466299, "grad_norm": 0.6834420561790466, "learning_rate": 0.0005, "loss": 1.423083782196045, "step": 19624 }, { "epoch": 1.5471028774142688, "grad_norm": 0.7019655108451843, "learning_rate": 0.0005, "loss": 1.3508954048156738, "step": 19625 }, { "epoch": 1.5471817106819077, "grad_norm": 0.6034872531890869, "learning_rate": 0.0005, "loss": 1.574864149093628, "step": 19626 }, { "epoch": 1.5472605439495468, "grad_norm": 0.5639840364456177, "learning_rate": 0.0005, "loss": 1.3751542568206787, "step": 19627 }, { "epoch": 1.5473393772171855, "grad_norm": 0.4814646244049072, "learning_rate": 0.0005, "loss": 1.4664608240127563, "step": 19628 }, { "epoch": 1.5474182104848246, "grad_norm": 0.6487942934036255, "learning_rate": 0.0005, "loss": 1.5839797258377075, "step": 19629 }, { "epoch": 1.5474970437524636, "grad_norm": 0.5464285612106323, "learning_rate": 0.0005, "loss": 1.6544363498687744, "step": 19630 }, { "epoch": 1.5475758770201025, "grad_norm": 0.5177013874053955, "learning_rate": 0.0005, "loss": 1.4789352416992188, "step": 19631 }, { "epoch": 1.5476547102877414, "grad_norm": 0.6199619174003601, "learning_rate": 0.0005, "loss": 1.5130928754806519, "step": 19632 }, { "epoch": 1.5477335435553803, "grad_norm": 0.6901916265487671, "learning_rate": 0.0005, "loss": 1.847154140472412, "step": 19633 }, { "epoch": 1.5478123768230194, "grad_norm": 0.6352301239967346, "learning_rate": 0.0005, "loss": 1.6330952644348145, "step": 19634 }, { "epoch": 1.547891210090658, "grad_norm": 0.46171319484710693, "learning_rate": 0.0005, "loss": 1.44740891456604, "step": 19635 }, { "epoch": 1.5479700433582972, "grad_norm": 0.5334004759788513, "learning_rate": 0.0005, "loss": 1.588336706161499, "step": 19636 }, { "epoch": 1.5480488766259362, "grad_norm": 0.6384318470954895, "learning_rate": 0.0005, "loss": 1.8735547065734863, "step": 19637 }, { "epoch": 1.548127709893575, "grad_norm": 0.6347470879554749, "learning_rate": 0.0005, "loss": 2.0199413299560547, "step": 19638 }, { "epoch": 1.548206543161214, "grad_norm": 0.45957455039024353, "learning_rate": 0.0005, "loss": 1.4879677295684814, "step": 19639 }, { "epoch": 1.5482853764288529, "grad_norm": 0.5593852400779724, "learning_rate": 0.0005, "loss": 1.5576756000518799, "step": 19640 }, { "epoch": 1.548364209696492, "grad_norm": 0.5810107588768005, "learning_rate": 0.0005, "loss": 1.6322367191314697, "step": 19641 }, { "epoch": 1.5484430429641307, "grad_norm": 0.5771303176879883, "learning_rate": 0.0005, "loss": 1.630070686340332, "step": 19642 }, { "epoch": 1.5485218762317698, "grad_norm": 0.6422816514968872, "learning_rate": 0.0005, "loss": 1.681787133216858, "step": 19643 }, { "epoch": 1.5486007094994088, "grad_norm": 0.5717079639434814, "learning_rate": 0.0005, "loss": 1.570319652557373, "step": 19644 }, { "epoch": 1.5486795427670477, "grad_norm": 1.1960867643356323, "learning_rate": 0.0005, "loss": 1.5677266120910645, "step": 19645 }, { "epoch": 1.5487583760346868, "grad_norm": 0.5291934013366699, "learning_rate": 0.0005, "loss": 1.4012401103973389, "step": 19646 }, { "epoch": 1.5488372093023255, "grad_norm": 5.881665229797363, "learning_rate": 0.0005, "loss": 1.4925466775894165, "step": 19647 }, { "epoch": 1.5489160425699646, "grad_norm": 0.6591437458992004, "learning_rate": 0.0005, "loss": 1.3500438928604126, "step": 19648 }, { "epoch": 1.5489948758376033, "grad_norm": 0.6374945044517517, "learning_rate": 0.0005, "loss": 1.6465089321136475, "step": 19649 }, { "epoch": 1.5490737091052424, "grad_norm": 0.6342684626579285, "learning_rate": 0.0005, "loss": 1.3793777227401733, "step": 19650 }, { "epoch": 1.5491525423728814, "grad_norm": 0.679259717464447, "learning_rate": 0.0005, "loss": 1.5893166065216064, "step": 19651 }, { "epoch": 1.5492313756405203, "grad_norm": 0.7155263423919678, "learning_rate": 0.0005, "loss": 1.5560890436172485, "step": 19652 }, { "epoch": 1.5493102089081594, "grad_norm": 0.48285797238349915, "learning_rate": 0.0005, "loss": 1.6793391704559326, "step": 19653 }, { "epoch": 1.549389042175798, "grad_norm": 0.5562827587127686, "learning_rate": 0.0005, "loss": 1.5954774618148804, "step": 19654 }, { "epoch": 1.5494678754434372, "grad_norm": 0.48024117946624756, "learning_rate": 0.0005, "loss": 1.4436333179473877, "step": 19655 }, { "epoch": 1.549546708711076, "grad_norm": 0.5895359516143799, "learning_rate": 0.0005, "loss": 1.5460832118988037, "step": 19656 }, { "epoch": 1.549625541978715, "grad_norm": 0.7056698799133301, "learning_rate": 0.0005, "loss": 1.7874126434326172, "step": 19657 }, { "epoch": 1.549704375246354, "grad_norm": 0.5403852462768555, "learning_rate": 0.0005, "loss": 1.6422384977340698, "step": 19658 }, { "epoch": 1.5497832085139929, "grad_norm": 0.5191279053688049, "learning_rate": 0.0005, "loss": 1.394652247428894, "step": 19659 }, { "epoch": 1.549862041781632, "grad_norm": 0.5435397028923035, "learning_rate": 0.0005, "loss": 1.4961950778961182, "step": 19660 }, { "epoch": 1.5499408750492707, "grad_norm": 0.5134607553482056, "learning_rate": 0.0005, "loss": 1.5135908126831055, "step": 19661 }, { "epoch": 1.5500197083169098, "grad_norm": 0.48368382453918457, "learning_rate": 0.0005, "loss": 1.5384867191314697, "step": 19662 }, { "epoch": 1.5500985415845485, "grad_norm": 0.706109881401062, "learning_rate": 0.0005, "loss": 1.6955604553222656, "step": 19663 }, { "epoch": 1.5501773748521877, "grad_norm": 0.577704131603241, "learning_rate": 0.0005, "loss": 1.577962040901184, "step": 19664 }, { "epoch": 1.5502562081198266, "grad_norm": 0.514494001865387, "learning_rate": 0.0005, "loss": 1.6821147203445435, "step": 19665 }, { "epoch": 1.5503350413874655, "grad_norm": 0.5875115394592285, "learning_rate": 0.0005, "loss": 1.463786005973816, "step": 19666 }, { "epoch": 1.5504138746551046, "grad_norm": 0.5972405672073364, "learning_rate": 0.0005, "loss": 1.3591458797454834, "step": 19667 }, { "epoch": 1.5504927079227433, "grad_norm": 0.5968289375305176, "learning_rate": 0.0005, "loss": 1.4383783340454102, "step": 19668 }, { "epoch": 1.5505715411903824, "grad_norm": 0.5253419876098633, "learning_rate": 0.0005, "loss": 1.435054063796997, "step": 19669 }, { "epoch": 1.5506503744580211, "grad_norm": 0.47498008608818054, "learning_rate": 0.0005, "loss": 1.5799436569213867, "step": 19670 }, { "epoch": 1.5507292077256603, "grad_norm": 0.4863629937171936, "learning_rate": 0.0005, "loss": 1.4464812278747559, "step": 19671 }, { "epoch": 1.5508080409932992, "grad_norm": 0.5811082124710083, "learning_rate": 0.0005, "loss": 1.508321762084961, "step": 19672 }, { "epoch": 1.550886874260938, "grad_norm": 0.5072318315505981, "learning_rate": 0.0005, "loss": 1.4084259271621704, "step": 19673 }, { "epoch": 1.5509657075285772, "grad_norm": 0.5326075553894043, "learning_rate": 0.0005, "loss": 1.491483211517334, "step": 19674 }, { "epoch": 1.551044540796216, "grad_norm": 0.48377978801727295, "learning_rate": 0.0005, "loss": 1.5741758346557617, "step": 19675 }, { "epoch": 1.551123374063855, "grad_norm": 0.5369216203689575, "learning_rate": 0.0005, "loss": 1.498117446899414, "step": 19676 }, { "epoch": 1.551202207331494, "grad_norm": 0.9106840491294861, "learning_rate": 0.0005, "loss": 1.5166256427764893, "step": 19677 }, { "epoch": 1.5512810405991329, "grad_norm": 0.4649035334587097, "learning_rate": 0.0005, "loss": 1.4607532024383545, "step": 19678 }, { "epoch": 1.5513598738667718, "grad_norm": 0.5343866348266602, "learning_rate": 0.0005, "loss": 1.5368497371673584, "step": 19679 }, { "epoch": 1.5514387071344107, "grad_norm": 0.5397884249687195, "learning_rate": 0.0005, "loss": 1.6163630485534668, "step": 19680 }, { "epoch": 1.5515175404020498, "grad_norm": 0.6040364503860474, "learning_rate": 0.0005, "loss": 1.5049972534179688, "step": 19681 }, { "epoch": 1.5515963736696885, "grad_norm": 0.6662084460258484, "learning_rate": 0.0005, "loss": 1.552247405052185, "step": 19682 }, { "epoch": 1.5516752069373276, "grad_norm": 0.6274585723876953, "learning_rate": 0.0005, "loss": 1.5071239471435547, "step": 19683 }, { "epoch": 1.5517540402049665, "grad_norm": 0.4962712824344635, "learning_rate": 0.0005, "loss": 1.5090553760528564, "step": 19684 }, { "epoch": 1.5518328734726055, "grad_norm": 0.4556903541088104, "learning_rate": 0.0005, "loss": 1.2967075109481812, "step": 19685 }, { "epoch": 1.5519117067402444, "grad_norm": 0.67535001039505, "learning_rate": 0.0005, "loss": 1.657528042793274, "step": 19686 }, { "epoch": 1.5519905400078833, "grad_norm": 0.6185239553451538, "learning_rate": 0.0005, "loss": 1.651633620262146, "step": 19687 }, { "epoch": 1.5520693732755224, "grad_norm": 0.4480634033679962, "learning_rate": 0.0005, "loss": 1.4085893630981445, "step": 19688 }, { "epoch": 1.552148206543161, "grad_norm": 0.8138524293899536, "learning_rate": 0.0005, "loss": 1.730079174041748, "step": 19689 }, { "epoch": 1.5522270398108002, "grad_norm": 0.5654172897338867, "learning_rate": 0.0005, "loss": 1.6460285186767578, "step": 19690 }, { "epoch": 1.5523058730784391, "grad_norm": 0.667956531047821, "learning_rate": 0.0005, "loss": 1.5354130268096924, "step": 19691 }, { "epoch": 1.552384706346078, "grad_norm": 0.5038551688194275, "learning_rate": 0.0005, "loss": 1.6059495210647583, "step": 19692 }, { "epoch": 1.552463539613717, "grad_norm": 0.5669296979904175, "learning_rate": 0.0005, "loss": 1.5215750932693481, "step": 19693 }, { "epoch": 1.5525423728813559, "grad_norm": 0.5076632499694824, "learning_rate": 0.0005, "loss": 1.5431113243103027, "step": 19694 }, { "epoch": 1.552621206148995, "grad_norm": 0.5354441404342651, "learning_rate": 0.0005, "loss": 1.5101544857025146, "step": 19695 }, { "epoch": 1.5527000394166337, "grad_norm": 1.076217532157898, "learning_rate": 0.0005, "loss": 1.7309457063674927, "step": 19696 }, { "epoch": 1.5527788726842728, "grad_norm": 1.0128782987594604, "learning_rate": 0.0005, "loss": 1.6304104328155518, "step": 19697 }, { "epoch": 1.5528577059519117, "grad_norm": 0.5250582695007324, "learning_rate": 0.0005, "loss": 1.5135579109191895, "step": 19698 }, { "epoch": 1.5529365392195507, "grad_norm": 0.5511694550514221, "learning_rate": 0.0005, "loss": 1.7434828281402588, "step": 19699 }, { "epoch": 1.5530153724871896, "grad_norm": 0.6636543273925781, "learning_rate": 0.0005, "loss": 1.7594120502471924, "step": 19700 }, { "epoch": 1.5530942057548285, "grad_norm": 0.5139010548591614, "learning_rate": 0.0005, "loss": 1.414884328842163, "step": 19701 }, { "epoch": 1.5531730390224676, "grad_norm": 0.563316822052002, "learning_rate": 0.0005, "loss": 1.5272719860076904, "step": 19702 }, { "epoch": 1.5532518722901063, "grad_norm": 0.6669163107872009, "learning_rate": 0.0005, "loss": 1.6897757053375244, "step": 19703 }, { "epoch": 1.5533307055577454, "grad_norm": 0.6442907452583313, "learning_rate": 0.0005, "loss": 1.6684136390686035, "step": 19704 }, { "epoch": 1.5534095388253844, "grad_norm": 0.6974250674247742, "learning_rate": 0.0005, "loss": 1.5062354803085327, "step": 19705 }, { "epoch": 1.5534883720930233, "grad_norm": 1.1903812885284424, "learning_rate": 0.0005, "loss": 1.687474012374878, "step": 19706 }, { "epoch": 1.5535672053606622, "grad_norm": 0.5893844962120056, "learning_rate": 0.0005, "loss": 1.6174988746643066, "step": 19707 }, { "epoch": 1.553646038628301, "grad_norm": 0.49352917075157166, "learning_rate": 0.0005, "loss": 1.5007020235061646, "step": 19708 }, { "epoch": 1.5537248718959402, "grad_norm": 0.5119666457176208, "learning_rate": 0.0005, "loss": 1.6997791528701782, "step": 19709 }, { "epoch": 1.553803705163579, "grad_norm": 0.5381094217300415, "learning_rate": 0.0005, "loss": 1.6170272827148438, "step": 19710 }, { "epoch": 1.553882538431218, "grad_norm": 0.5269429087638855, "learning_rate": 0.0005, "loss": 1.5054630041122437, "step": 19711 }, { "epoch": 1.553961371698857, "grad_norm": 0.6291996240615845, "learning_rate": 0.0005, "loss": 1.3869595527648926, "step": 19712 }, { "epoch": 1.5540402049664959, "grad_norm": 0.8796890377998352, "learning_rate": 0.0005, "loss": 1.441603183746338, "step": 19713 }, { "epoch": 1.5541190382341348, "grad_norm": 0.5063916444778442, "learning_rate": 0.0005, "loss": 1.5758854150772095, "step": 19714 }, { "epoch": 1.5541978715017737, "grad_norm": 0.5685228705406189, "learning_rate": 0.0005, "loss": 1.5251699686050415, "step": 19715 }, { "epoch": 1.5542767047694128, "grad_norm": 0.47968238592147827, "learning_rate": 0.0005, "loss": 1.5410377979278564, "step": 19716 }, { "epoch": 1.5543555380370515, "grad_norm": 0.5281760096549988, "learning_rate": 0.0005, "loss": 1.5240588188171387, "step": 19717 }, { "epoch": 1.5544343713046906, "grad_norm": 0.4608296751976013, "learning_rate": 0.0005, "loss": 1.3854275941848755, "step": 19718 }, { "epoch": 1.5545132045723296, "grad_norm": 0.5934652090072632, "learning_rate": 0.0005, "loss": 1.5467112064361572, "step": 19719 }, { "epoch": 1.5545920378399685, "grad_norm": 0.6463242769241333, "learning_rate": 0.0005, "loss": 1.8809927701950073, "step": 19720 }, { "epoch": 1.5546708711076074, "grad_norm": 0.5297014713287354, "learning_rate": 0.0005, "loss": 1.617077112197876, "step": 19721 }, { "epoch": 1.5547497043752463, "grad_norm": 0.42379850149154663, "learning_rate": 0.0005, "loss": 1.2709887027740479, "step": 19722 }, { "epoch": 1.5548285376428854, "grad_norm": 0.5420941114425659, "learning_rate": 0.0005, "loss": 1.6907455921173096, "step": 19723 }, { "epoch": 1.5549073709105241, "grad_norm": 0.39023393392562866, "learning_rate": 0.0005, "loss": 1.2146073579788208, "step": 19724 }, { "epoch": 1.5549862041781632, "grad_norm": 0.527066707611084, "learning_rate": 0.0005, "loss": 1.3626686334609985, "step": 19725 }, { "epoch": 1.5550650374458022, "grad_norm": 0.5280495882034302, "learning_rate": 0.0005, "loss": 1.5315814018249512, "step": 19726 }, { "epoch": 1.555143870713441, "grad_norm": 0.5716065168380737, "learning_rate": 0.0005, "loss": 1.4650869369506836, "step": 19727 }, { "epoch": 1.55522270398108, "grad_norm": 1.1176574230194092, "learning_rate": 0.0005, "loss": 1.8877359628677368, "step": 19728 }, { "epoch": 1.555301537248719, "grad_norm": 0.5273757576942444, "learning_rate": 0.0005, "loss": 1.4364604949951172, "step": 19729 }, { "epoch": 1.555380370516358, "grad_norm": 0.5226492881774902, "learning_rate": 0.0005, "loss": 1.5689170360565186, "step": 19730 }, { "epoch": 1.5554592037839967, "grad_norm": 0.46967649459838867, "learning_rate": 0.0005, "loss": 1.3392282724380493, "step": 19731 }, { "epoch": 1.5555380370516358, "grad_norm": 0.5064849853515625, "learning_rate": 0.0005, "loss": 1.4287848472595215, "step": 19732 }, { "epoch": 1.5556168703192748, "grad_norm": 0.49599334597587585, "learning_rate": 0.0005, "loss": 1.5809221267700195, "step": 19733 }, { "epoch": 1.5556957035869137, "grad_norm": 0.46662095189094543, "learning_rate": 0.0005, "loss": 1.2863688468933105, "step": 19734 }, { "epoch": 1.5557745368545526, "grad_norm": 0.7326014041900635, "learning_rate": 0.0005, "loss": 1.4876400232315063, "step": 19735 }, { "epoch": 1.5558533701221915, "grad_norm": 0.5363245010375977, "learning_rate": 0.0005, "loss": 1.6221959590911865, "step": 19736 }, { "epoch": 1.5559322033898306, "grad_norm": 0.6544051170349121, "learning_rate": 0.0005, "loss": 1.6072977781295776, "step": 19737 }, { "epoch": 1.5560110366574693, "grad_norm": 0.47843247652053833, "learning_rate": 0.0005, "loss": 1.5167748928070068, "step": 19738 }, { "epoch": 1.5560898699251084, "grad_norm": 0.6095370650291443, "learning_rate": 0.0005, "loss": 1.5805342197418213, "step": 19739 }, { "epoch": 1.5561687031927474, "grad_norm": 0.6458099484443665, "learning_rate": 0.0005, "loss": 1.7132747173309326, "step": 19740 }, { "epoch": 1.5562475364603863, "grad_norm": 0.5071788430213928, "learning_rate": 0.0005, "loss": 1.556880235671997, "step": 19741 }, { "epoch": 1.5563263697280252, "grad_norm": 1.109416127204895, "learning_rate": 0.0005, "loss": 1.6902086734771729, "step": 19742 }, { "epoch": 1.556405202995664, "grad_norm": 0.7497566342353821, "learning_rate": 0.0005, "loss": 1.827305793762207, "step": 19743 }, { "epoch": 1.5564840362633032, "grad_norm": 0.597360372543335, "learning_rate": 0.0005, "loss": 1.7240142822265625, "step": 19744 }, { "epoch": 1.556562869530942, "grad_norm": 0.4551863372325897, "learning_rate": 0.0005, "loss": 1.4587080478668213, "step": 19745 }, { "epoch": 1.556641702798581, "grad_norm": 0.703524112701416, "learning_rate": 0.0005, "loss": 1.6950104236602783, "step": 19746 }, { "epoch": 1.55672053606622, "grad_norm": 0.6247777342796326, "learning_rate": 0.0005, "loss": 1.6579949855804443, "step": 19747 }, { "epoch": 1.5567993693338589, "grad_norm": 0.4569522738456726, "learning_rate": 0.0005, "loss": 1.4693912267684937, "step": 19748 }, { "epoch": 1.5568782026014978, "grad_norm": 0.5270912051200867, "learning_rate": 0.0005, "loss": 1.5251274108886719, "step": 19749 }, { "epoch": 1.5569570358691367, "grad_norm": 0.6855652332305908, "learning_rate": 0.0005, "loss": 1.6121506690979004, "step": 19750 }, { "epoch": 1.5570358691367758, "grad_norm": 0.46906477212905884, "learning_rate": 0.0005, "loss": 1.628469467163086, "step": 19751 }, { "epoch": 1.5571147024044145, "grad_norm": 0.47418490052223206, "learning_rate": 0.0005, "loss": 1.5117729902267456, "step": 19752 }, { "epoch": 1.5571935356720537, "grad_norm": 0.51566082239151, "learning_rate": 0.0005, "loss": 1.3657371997833252, "step": 19753 }, { "epoch": 1.5572723689396926, "grad_norm": 0.4877828359603882, "learning_rate": 0.0005, "loss": 1.3214807510375977, "step": 19754 }, { "epoch": 1.5573512022073315, "grad_norm": 0.573559045791626, "learning_rate": 0.0005, "loss": 1.3723938465118408, "step": 19755 }, { "epoch": 1.5574300354749704, "grad_norm": 0.4865860044956207, "learning_rate": 0.0005, "loss": 1.5377745628356934, "step": 19756 }, { "epoch": 1.5575088687426093, "grad_norm": 0.5243500471115112, "learning_rate": 0.0005, "loss": 1.7404468059539795, "step": 19757 }, { "epoch": 1.5575877020102484, "grad_norm": 0.4743046164512634, "learning_rate": 0.0005, "loss": 1.5697338581085205, "step": 19758 }, { "epoch": 1.5576665352778871, "grad_norm": 0.561610221862793, "learning_rate": 0.0005, "loss": 1.6086112260818481, "step": 19759 }, { "epoch": 1.5577453685455263, "grad_norm": 0.4433843195438385, "learning_rate": 0.0005, "loss": 1.4703558683395386, "step": 19760 }, { "epoch": 1.5578242018131652, "grad_norm": 0.5215603113174438, "learning_rate": 0.0005, "loss": 1.5319812297821045, "step": 19761 }, { "epoch": 1.557903035080804, "grad_norm": 0.7561676502227783, "learning_rate": 0.0005, "loss": 1.7116196155548096, "step": 19762 }, { "epoch": 1.557981868348443, "grad_norm": 0.4463227391242981, "learning_rate": 0.0005, "loss": 1.4280166625976562, "step": 19763 }, { "epoch": 1.558060701616082, "grad_norm": 0.4763438105583191, "learning_rate": 0.0005, "loss": 1.4408669471740723, "step": 19764 }, { "epoch": 1.558139534883721, "grad_norm": 0.545638382434845, "learning_rate": 0.0005, "loss": 1.6253743171691895, "step": 19765 }, { "epoch": 1.5582183681513597, "grad_norm": 0.6641735434532166, "learning_rate": 0.0005, "loss": 1.7271331548690796, "step": 19766 }, { "epoch": 1.5582972014189989, "grad_norm": 0.5353785157203674, "learning_rate": 0.0005, "loss": 1.7179278135299683, "step": 19767 }, { "epoch": 1.5583760346866378, "grad_norm": 1.0487425327301025, "learning_rate": 0.0005, "loss": 1.5241321325302124, "step": 19768 }, { "epoch": 1.5584548679542767, "grad_norm": 0.6272028684616089, "learning_rate": 0.0005, "loss": 1.4507802724838257, "step": 19769 }, { "epoch": 1.5585337012219158, "grad_norm": 0.5809229016304016, "learning_rate": 0.0005, "loss": 1.6572608947753906, "step": 19770 }, { "epoch": 1.5586125344895545, "grad_norm": 0.6719454526901245, "learning_rate": 0.0005, "loss": 1.5005877017974854, "step": 19771 }, { "epoch": 1.5586913677571936, "grad_norm": 0.5506093502044678, "learning_rate": 0.0005, "loss": 1.4284985065460205, "step": 19772 }, { "epoch": 1.5587702010248323, "grad_norm": 0.6506146192550659, "learning_rate": 0.0005, "loss": 1.3246270418167114, "step": 19773 }, { "epoch": 1.5588490342924715, "grad_norm": 0.5195205211639404, "learning_rate": 0.0005, "loss": 1.5227640867233276, "step": 19774 }, { "epoch": 1.5589278675601104, "grad_norm": 0.4827975928783417, "learning_rate": 0.0005, "loss": 1.3170130252838135, "step": 19775 }, { "epoch": 1.5590067008277493, "grad_norm": 0.5166674852371216, "learning_rate": 0.0005, "loss": 1.7404050827026367, "step": 19776 }, { "epoch": 1.5590855340953884, "grad_norm": 0.6012392044067383, "learning_rate": 0.0005, "loss": 1.4187813997268677, "step": 19777 }, { "epoch": 1.559164367363027, "grad_norm": 0.4498205780982971, "learning_rate": 0.0005, "loss": 1.5111154317855835, "step": 19778 }, { "epoch": 1.5592432006306662, "grad_norm": 0.6638094782829285, "learning_rate": 0.0005, "loss": 1.6496758460998535, "step": 19779 }, { "epoch": 1.559322033898305, "grad_norm": 0.5789825916290283, "learning_rate": 0.0005, "loss": 1.5686616897583008, "step": 19780 }, { "epoch": 1.559400867165944, "grad_norm": 0.707893967628479, "learning_rate": 0.0005, "loss": 1.6616016626358032, "step": 19781 }, { "epoch": 1.559479700433583, "grad_norm": 0.48828235268592834, "learning_rate": 0.0005, "loss": 1.5950568914413452, "step": 19782 }, { "epoch": 1.5595585337012219, "grad_norm": 0.5015894770622253, "learning_rate": 0.0005, "loss": 1.6871025562286377, "step": 19783 }, { "epoch": 1.559637366968861, "grad_norm": 0.7354364395141602, "learning_rate": 0.0005, "loss": 1.6144989728927612, "step": 19784 }, { "epoch": 1.5597162002364997, "grad_norm": 0.46749213337898254, "learning_rate": 0.0005, "loss": 1.6534814834594727, "step": 19785 }, { "epoch": 1.5597950335041388, "grad_norm": 0.6604134440422058, "learning_rate": 0.0005, "loss": 1.4072277545928955, "step": 19786 }, { "epoch": 1.5598738667717775, "grad_norm": 0.6367827653884888, "learning_rate": 0.0005, "loss": 1.7881163358688354, "step": 19787 }, { "epoch": 1.5599527000394167, "grad_norm": 1.8307955265045166, "learning_rate": 0.0005, "loss": 1.6099399328231812, "step": 19788 }, { "epoch": 1.5600315333070556, "grad_norm": 0.5386556386947632, "learning_rate": 0.0005, "loss": 1.5504348278045654, "step": 19789 }, { "epoch": 1.5601103665746945, "grad_norm": 0.5974133610725403, "learning_rate": 0.0005, "loss": 1.7844607830047607, "step": 19790 }, { "epoch": 1.5601891998423336, "grad_norm": 0.5034463405609131, "learning_rate": 0.0005, "loss": 1.5169429779052734, "step": 19791 }, { "epoch": 1.5602680331099723, "grad_norm": 0.5273550748825073, "learning_rate": 0.0005, "loss": 1.6650230884552002, "step": 19792 }, { "epoch": 1.5603468663776114, "grad_norm": 0.48712289333343506, "learning_rate": 0.0005, "loss": 1.389925241470337, "step": 19793 }, { "epoch": 1.5604256996452501, "grad_norm": 0.5565230250358582, "learning_rate": 0.0005, "loss": 1.614568829536438, "step": 19794 }, { "epoch": 1.5605045329128893, "grad_norm": 0.5834309458732605, "learning_rate": 0.0005, "loss": 1.625650405883789, "step": 19795 }, { "epoch": 1.5605833661805282, "grad_norm": 0.4858713746070862, "learning_rate": 0.0005, "loss": 1.4469845294952393, "step": 19796 }, { "epoch": 1.560662199448167, "grad_norm": 0.5865093469619751, "learning_rate": 0.0005, "loss": 1.5857877731323242, "step": 19797 }, { "epoch": 1.5607410327158062, "grad_norm": 0.5261680483818054, "learning_rate": 0.0005, "loss": 1.5339775085449219, "step": 19798 }, { "epoch": 1.560819865983445, "grad_norm": 0.4911908209323883, "learning_rate": 0.0005, "loss": 1.355757236480713, "step": 19799 }, { "epoch": 1.560898699251084, "grad_norm": 0.5200951099395752, "learning_rate": 0.0005, "loss": 1.424095630645752, "step": 19800 }, { "epoch": 1.560977532518723, "grad_norm": 0.5348259806632996, "learning_rate": 0.0005, "loss": 1.3521902561187744, "step": 19801 }, { "epoch": 1.5610563657863619, "grad_norm": 0.4344632625579834, "learning_rate": 0.0005, "loss": 1.4028728008270264, "step": 19802 }, { "epoch": 1.5611351990540008, "grad_norm": 0.7281544208526611, "learning_rate": 0.0005, "loss": 1.4102756977081299, "step": 19803 }, { "epoch": 1.5612140323216397, "grad_norm": 0.5955629348754883, "learning_rate": 0.0005, "loss": 1.659498691558838, "step": 19804 }, { "epoch": 1.5612928655892788, "grad_norm": 0.4713360369205475, "learning_rate": 0.0005, "loss": 1.4325308799743652, "step": 19805 }, { "epoch": 1.5613716988569175, "grad_norm": 0.5228844881057739, "learning_rate": 0.0005, "loss": 1.5277459621429443, "step": 19806 }, { "epoch": 1.5614505321245566, "grad_norm": 0.6432943344116211, "learning_rate": 0.0005, "loss": 1.4689490795135498, "step": 19807 }, { "epoch": 1.5615293653921956, "grad_norm": 0.49797317385673523, "learning_rate": 0.0005, "loss": 1.4767239093780518, "step": 19808 }, { "epoch": 1.5616081986598345, "grad_norm": 0.7328535318374634, "learning_rate": 0.0005, "loss": 1.6056454181671143, "step": 19809 }, { "epoch": 1.5616870319274734, "grad_norm": 0.6343538165092468, "learning_rate": 0.0005, "loss": 1.527167797088623, "step": 19810 }, { "epoch": 1.5617658651951123, "grad_norm": 0.6055540442466736, "learning_rate": 0.0005, "loss": 1.6417882442474365, "step": 19811 }, { "epoch": 1.5618446984627514, "grad_norm": 0.5301403403282166, "learning_rate": 0.0005, "loss": 1.5897533893585205, "step": 19812 }, { "epoch": 1.5619235317303901, "grad_norm": 0.4993075132369995, "learning_rate": 0.0005, "loss": 1.5508112907409668, "step": 19813 }, { "epoch": 1.5620023649980292, "grad_norm": 0.49052074551582336, "learning_rate": 0.0005, "loss": 1.458543300628662, "step": 19814 }, { "epoch": 1.5620811982656682, "grad_norm": 0.5147911906242371, "learning_rate": 0.0005, "loss": 1.6858503818511963, "step": 19815 }, { "epoch": 1.562160031533307, "grad_norm": 0.5218755006790161, "learning_rate": 0.0005, "loss": 1.4481589794158936, "step": 19816 }, { "epoch": 1.562238864800946, "grad_norm": 0.5731242895126343, "learning_rate": 0.0005, "loss": 1.4331560134887695, "step": 19817 }, { "epoch": 1.562317698068585, "grad_norm": 0.5751716494560242, "learning_rate": 0.0005, "loss": 1.4261621236801147, "step": 19818 }, { "epoch": 1.562396531336224, "grad_norm": 0.5704008936882019, "learning_rate": 0.0005, "loss": 1.5356101989746094, "step": 19819 }, { "epoch": 1.5624753646038627, "grad_norm": 0.5968854427337646, "learning_rate": 0.0005, "loss": 1.5229922533035278, "step": 19820 }, { "epoch": 1.5625541978715018, "grad_norm": 0.4398500621318817, "learning_rate": 0.0005, "loss": 1.4174935817718506, "step": 19821 }, { "epoch": 1.5626330311391408, "grad_norm": 0.5428361892700195, "learning_rate": 0.0005, "loss": 1.5846266746520996, "step": 19822 }, { "epoch": 1.5627118644067797, "grad_norm": 0.5455884337425232, "learning_rate": 0.0005, "loss": 1.6147969961166382, "step": 19823 }, { "epoch": 1.5627906976744186, "grad_norm": 0.5607340335845947, "learning_rate": 0.0005, "loss": 1.4695857763290405, "step": 19824 }, { "epoch": 1.5628695309420575, "grad_norm": 0.5237943530082703, "learning_rate": 0.0005, "loss": 1.5964360237121582, "step": 19825 }, { "epoch": 1.5629483642096966, "grad_norm": 0.4422690272331238, "learning_rate": 0.0005, "loss": 1.5347964763641357, "step": 19826 }, { "epoch": 1.5630271974773353, "grad_norm": 0.7082313299179077, "learning_rate": 0.0005, "loss": 1.5956377983093262, "step": 19827 }, { "epoch": 1.5631060307449745, "grad_norm": 0.5005473494529724, "learning_rate": 0.0005, "loss": 1.4625285863876343, "step": 19828 }, { "epoch": 1.5631848640126134, "grad_norm": 0.5564694404602051, "learning_rate": 0.0005, "loss": 1.4142863750457764, "step": 19829 }, { "epoch": 1.5632636972802523, "grad_norm": 0.5010802149772644, "learning_rate": 0.0005, "loss": 1.5024712085723877, "step": 19830 }, { "epoch": 1.5633425305478912, "grad_norm": 0.5675053000450134, "learning_rate": 0.0005, "loss": 1.5374789237976074, "step": 19831 }, { "epoch": 1.56342136381553, "grad_norm": 0.524679958820343, "learning_rate": 0.0005, "loss": 1.4534666538238525, "step": 19832 }, { "epoch": 1.5635001970831692, "grad_norm": 0.726982593536377, "learning_rate": 0.0005, "loss": 1.7355576753616333, "step": 19833 }, { "epoch": 1.563579030350808, "grad_norm": 0.5217223167419434, "learning_rate": 0.0005, "loss": 1.727003574371338, "step": 19834 }, { "epoch": 1.563657863618447, "grad_norm": 0.4698665142059326, "learning_rate": 0.0005, "loss": 1.3907344341278076, "step": 19835 }, { "epoch": 1.563736696886086, "grad_norm": 0.5156121253967285, "learning_rate": 0.0005, "loss": 1.7851836681365967, "step": 19836 }, { "epoch": 1.5638155301537249, "grad_norm": 0.6131583452224731, "learning_rate": 0.0005, "loss": 1.633774757385254, "step": 19837 }, { "epoch": 1.5638943634213638, "grad_norm": 0.4783322811126709, "learning_rate": 0.0005, "loss": 1.5067510604858398, "step": 19838 }, { "epoch": 1.5639731966890027, "grad_norm": 0.5093605518341064, "learning_rate": 0.0005, "loss": 1.5482134819030762, "step": 19839 }, { "epoch": 1.5640520299566418, "grad_norm": 0.8044610023498535, "learning_rate": 0.0005, "loss": 1.5339162349700928, "step": 19840 }, { "epoch": 1.5641308632242805, "grad_norm": 0.4895888566970825, "learning_rate": 0.0005, "loss": 1.6134495735168457, "step": 19841 }, { "epoch": 1.5642096964919197, "grad_norm": 0.47805750370025635, "learning_rate": 0.0005, "loss": 1.5481524467468262, "step": 19842 }, { "epoch": 1.5642885297595586, "grad_norm": 0.48411136865615845, "learning_rate": 0.0005, "loss": 1.323287010192871, "step": 19843 }, { "epoch": 1.5643673630271975, "grad_norm": 0.5503264665603638, "learning_rate": 0.0005, "loss": 1.7134690284729004, "step": 19844 }, { "epoch": 1.5644461962948364, "grad_norm": 0.7070354223251343, "learning_rate": 0.0005, "loss": 1.639793872833252, "step": 19845 }, { "epoch": 1.5645250295624753, "grad_norm": 0.6957533955574036, "learning_rate": 0.0005, "loss": 1.4969165325164795, "step": 19846 }, { "epoch": 1.5646038628301144, "grad_norm": 0.46077388525009155, "learning_rate": 0.0005, "loss": 1.4226219654083252, "step": 19847 }, { "epoch": 1.5646826960977531, "grad_norm": 0.4358273148536682, "learning_rate": 0.0005, "loss": 1.2457284927368164, "step": 19848 }, { "epoch": 1.5647615293653923, "grad_norm": 0.5472033619880676, "learning_rate": 0.0005, "loss": 1.4721636772155762, "step": 19849 }, { "epoch": 1.5648403626330312, "grad_norm": 0.7306118011474609, "learning_rate": 0.0005, "loss": 1.6731669902801514, "step": 19850 }, { "epoch": 1.56491919590067, "grad_norm": 0.4769541025161743, "learning_rate": 0.0005, "loss": 1.5121173858642578, "step": 19851 }, { "epoch": 1.564998029168309, "grad_norm": 0.5639711618423462, "learning_rate": 0.0005, "loss": 1.5014338493347168, "step": 19852 }, { "epoch": 1.565076862435948, "grad_norm": 0.5011350512504578, "learning_rate": 0.0005, "loss": 1.5882024765014648, "step": 19853 }, { "epoch": 1.565155695703587, "grad_norm": 0.6355578899383545, "learning_rate": 0.0005, "loss": 1.571921467781067, "step": 19854 }, { "epoch": 1.5652345289712257, "grad_norm": 0.6126861572265625, "learning_rate": 0.0005, "loss": 1.6408312320709229, "step": 19855 }, { "epoch": 1.5653133622388649, "grad_norm": 0.5174622535705566, "learning_rate": 0.0005, "loss": 1.6840250492095947, "step": 19856 }, { "epoch": 1.5653921955065038, "grad_norm": 0.6423972845077515, "learning_rate": 0.0005, "loss": 1.3605611324310303, "step": 19857 }, { "epoch": 1.5654710287741427, "grad_norm": 0.5348234176635742, "learning_rate": 0.0005, "loss": 1.4680044651031494, "step": 19858 }, { "epoch": 1.5655498620417816, "grad_norm": 0.7380887866020203, "learning_rate": 0.0005, "loss": 1.6796069145202637, "step": 19859 }, { "epoch": 1.5656286953094205, "grad_norm": 0.46628981828689575, "learning_rate": 0.0005, "loss": 1.3930267095565796, "step": 19860 }, { "epoch": 1.5657075285770596, "grad_norm": 0.592948853969574, "learning_rate": 0.0005, "loss": 1.6681252717971802, "step": 19861 }, { "epoch": 1.5657863618446983, "grad_norm": 0.5629725456237793, "learning_rate": 0.0005, "loss": 1.5447722673416138, "step": 19862 }, { "epoch": 1.5658651951123375, "grad_norm": 0.6904111504554749, "learning_rate": 0.0005, "loss": 1.6279916763305664, "step": 19863 }, { "epoch": 1.5659440283799764, "grad_norm": 0.47864073514938354, "learning_rate": 0.0005, "loss": 1.5155470371246338, "step": 19864 }, { "epoch": 1.5660228616476153, "grad_norm": 0.5760975480079651, "learning_rate": 0.0005, "loss": 1.595238447189331, "step": 19865 }, { "epoch": 1.5661016949152542, "grad_norm": 0.5623554587364197, "learning_rate": 0.0005, "loss": 1.8091299533843994, "step": 19866 }, { "epoch": 1.566180528182893, "grad_norm": 0.6224246621131897, "learning_rate": 0.0005, "loss": 1.432585597038269, "step": 19867 }, { "epoch": 1.5662593614505322, "grad_norm": 0.6642881035804749, "learning_rate": 0.0005, "loss": 1.5117967128753662, "step": 19868 }, { "epoch": 1.566338194718171, "grad_norm": 0.5778220295906067, "learning_rate": 0.0005, "loss": 1.6832174062728882, "step": 19869 }, { "epoch": 1.56641702798581, "grad_norm": 0.5550704002380371, "learning_rate": 0.0005, "loss": 1.738490104675293, "step": 19870 }, { "epoch": 1.566495861253449, "grad_norm": 0.6650827527046204, "learning_rate": 0.0005, "loss": 1.8039507865905762, "step": 19871 }, { "epoch": 1.5665746945210879, "grad_norm": 0.5185505151748657, "learning_rate": 0.0005, "loss": 1.5292391777038574, "step": 19872 }, { "epoch": 1.5666535277887268, "grad_norm": 0.47920387983322144, "learning_rate": 0.0005, "loss": 1.5319147109985352, "step": 19873 }, { "epoch": 1.5667323610563657, "grad_norm": 0.6132413744926453, "learning_rate": 0.0005, "loss": 1.646848201751709, "step": 19874 }, { "epoch": 1.5668111943240048, "grad_norm": 0.6305503845214844, "learning_rate": 0.0005, "loss": 1.6917849779129028, "step": 19875 }, { "epoch": 1.5668900275916435, "grad_norm": 0.4796798825263977, "learning_rate": 0.0005, "loss": 1.5637733936309814, "step": 19876 }, { "epoch": 1.5669688608592827, "grad_norm": 0.3992313742637634, "learning_rate": 0.0005, "loss": 1.4351696968078613, "step": 19877 }, { "epoch": 1.5670476941269216, "grad_norm": 0.4508243799209595, "learning_rate": 0.0005, "loss": 1.4713318347930908, "step": 19878 }, { "epoch": 1.5671265273945605, "grad_norm": 0.5040808916091919, "learning_rate": 0.0005, "loss": 1.4883551597595215, "step": 19879 }, { "epoch": 1.5672053606621994, "grad_norm": 0.5019368529319763, "learning_rate": 0.0005, "loss": 1.6413989067077637, "step": 19880 }, { "epoch": 1.5672841939298383, "grad_norm": 0.5641030669212341, "learning_rate": 0.0005, "loss": 1.5365166664123535, "step": 19881 }, { "epoch": 1.5673630271974774, "grad_norm": 0.45554330945014954, "learning_rate": 0.0005, "loss": 1.4449447393417358, "step": 19882 }, { "epoch": 1.5674418604651161, "grad_norm": 0.5509178042411804, "learning_rate": 0.0005, "loss": 1.3833876848220825, "step": 19883 }, { "epoch": 1.5675206937327553, "grad_norm": 0.43660733103752136, "learning_rate": 0.0005, "loss": 1.3530571460723877, "step": 19884 }, { "epoch": 1.5675995270003942, "grad_norm": 0.555699348449707, "learning_rate": 0.0005, "loss": 1.6235411167144775, "step": 19885 }, { "epoch": 1.567678360268033, "grad_norm": 0.6219339966773987, "learning_rate": 0.0005, "loss": 1.4521539211273193, "step": 19886 }, { "epoch": 1.567757193535672, "grad_norm": 0.6796762347221375, "learning_rate": 0.0005, "loss": 1.452269196510315, "step": 19887 }, { "epoch": 1.567836026803311, "grad_norm": 0.5446475148200989, "learning_rate": 0.0005, "loss": 1.5073609352111816, "step": 19888 }, { "epoch": 1.56791486007095, "grad_norm": 0.5794260501861572, "learning_rate": 0.0005, "loss": 1.9033317565917969, "step": 19889 }, { "epoch": 1.5679936933385887, "grad_norm": 0.48477524518966675, "learning_rate": 0.0005, "loss": 1.446602702140808, "step": 19890 }, { "epoch": 1.5680725266062279, "grad_norm": 0.5272138118743896, "learning_rate": 0.0005, "loss": 1.6189284324645996, "step": 19891 }, { "epoch": 1.5681513598738668, "grad_norm": 0.6715236306190491, "learning_rate": 0.0005, "loss": 1.5479099750518799, "step": 19892 }, { "epoch": 1.5682301931415057, "grad_norm": 0.591887354850769, "learning_rate": 0.0005, "loss": 1.6758904457092285, "step": 19893 }, { "epoch": 1.5683090264091448, "grad_norm": 0.546523928642273, "learning_rate": 0.0005, "loss": 1.7486646175384521, "step": 19894 }, { "epoch": 1.5683878596767835, "grad_norm": 0.5755802392959595, "learning_rate": 0.0005, "loss": 1.5082207918167114, "step": 19895 }, { "epoch": 1.5684666929444226, "grad_norm": 0.6370263695716858, "learning_rate": 0.0005, "loss": 1.5226857662200928, "step": 19896 }, { "epoch": 1.5685455262120613, "grad_norm": 0.5717113614082336, "learning_rate": 0.0005, "loss": 1.7979419231414795, "step": 19897 }, { "epoch": 1.5686243594797005, "grad_norm": 0.4870511591434479, "learning_rate": 0.0005, "loss": 1.3197531700134277, "step": 19898 }, { "epoch": 1.5687031927473394, "grad_norm": 0.5504252910614014, "learning_rate": 0.0005, "loss": 1.5360643863677979, "step": 19899 }, { "epoch": 1.5687820260149783, "grad_norm": 0.6891335248947144, "learning_rate": 0.0005, "loss": 1.6599102020263672, "step": 19900 }, { "epoch": 1.5688608592826174, "grad_norm": 0.49537545442581177, "learning_rate": 0.0005, "loss": 1.510077714920044, "step": 19901 }, { "epoch": 1.5689396925502561, "grad_norm": 0.3558230698108673, "learning_rate": 0.0005, "loss": 1.2651216983795166, "step": 19902 }, { "epoch": 1.5690185258178952, "grad_norm": 0.42245110869407654, "learning_rate": 0.0005, "loss": 1.4575767517089844, "step": 19903 }, { "epoch": 1.569097359085534, "grad_norm": 0.6366778016090393, "learning_rate": 0.0005, "loss": 1.4287222623825073, "step": 19904 }, { "epoch": 1.569176192353173, "grad_norm": 0.5734127759933472, "learning_rate": 0.0005, "loss": 1.5238332748413086, "step": 19905 }, { "epoch": 1.569255025620812, "grad_norm": 0.5424825549125671, "learning_rate": 0.0005, "loss": 1.5217304229736328, "step": 19906 }, { "epoch": 1.569333858888451, "grad_norm": 0.45397743582725525, "learning_rate": 0.0005, "loss": 1.645878791809082, "step": 19907 }, { "epoch": 1.56941269215609, "grad_norm": 0.5883371233940125, "learning_rate": 0.0005, "loss": 1.8971924781799316, "step": 19908 }, { "epoch": 1.5694915254237287, "grad_norm": 0.5794674754142761, "learning_rate": 0.0005, "loss": 1.493697166442871, "step": 19909 }, { "epoch": 1.5695703586913679, "grad_norm": 0.554754376411438, "learning_rate": 0.0005, "loss": 1.6348087787628174, "step": 19910 }, { "epoch": 1.5696491919590065, "grad_norm": 0.4983925223350525, "learning_rate": 0.0005, "loss": 1.3361033201217651, "step": 19911 }, { "epoch": 1.5697280252266457, "grad_norm": 0.5475403070449829, "learning_rate": 0.0005, "loss": 1.6277316808700562, "step": 19912 }, { "epoch": 1.5698068584942846, "grad_norm": 0.5930315256118774, "learning_rate": 0.0005, "loss": 1.5037293434143066, "step": 19913 }, { "epoch": 1.5698856917619235, "grad_norm": 0.4960458278656006, "learning_rate": 0.0005, "loss": 1.6106202602386475, "step": 19914 }, { "epoch": 1.5699645250295626, "grad_norm": 0.42017531394958496, "learning_rate": 0.0005, "loss": 1.3861682415008545, "step": 19915 }, { "epoch": 1.5700433582972013, "grad_norm": 0.4209113121032715, "learning_rate": 0.0005, "loss": 1.4648244380950928, "step": 19916 }, { "epoch": 1.5701221915648405, "grad_norm": 0.5202683210372925, "learning_rate": 0.0005, "loss": 1.741981029510498, "step": 19917 }, { "epoch": 1.5702010248324791, "grad_norm": 0.6663088798522949, "learning_rate": 0.0005, "loss": 1.7550139427185059, "step": 19918 }, { "epoch": 1.5702798581001183, "grad_norm": 0.5132095813751221, "learning_rate": 0.0005, "loss": 1.3496921062469482, "step": 19919 }, { "epoch": 1.5703586913677572, "grad_norm": 0.6172436475753784, "learning_rate": 0.0005, "loss": 1.6465177536010742, "step": 19920 }, { "epoch": 1.570437524635396, "grad_norm": 0.5582453608512878, "learning_rate": 0.0005, "loss": 1.6203112602233887, "step": 19921 }, { "epoch": 1.5705163579030352, "grad_norm": 0.4943954050540924, "learning_rate": 0.0005, "loss": 1.3801636695861816, "step": 19922 }, { "epoch": 1.570595191170674, "grad_norm": 0.7752821445465088, "learning_rate": 0.0005, "loss": 2.1007471084594727, "step": 19923 }, { "epoch": 1.570674024438313, "grad_norm": 0.6648535132408142, "learning_rate": 0.0005, "loss": 1.6260623931884766, "step": 19924 }, { "epoch": 1.570752857705952, "grad_norm": 0.6209015846252441, "learning_rate": 0.0005, "loss": 1.5802857875823975, "step": 19925 }, { "epoch": 1.5708316909735909, "grad_norm": 0.5476644039154053, "learning_rate": 0.0005, "loss": 1.7722241878509521, "step": 19926 }, { "epoch": 1.5709105242412298, "grad_norm": 0.5248309373855591, "learning_rate": 0.0005, "loss": 1.3081355094909668, "step": 19927 }, { "epoch": 1.5709893575088687, "grad_norm": 0.43918293714523315, "learning_rate": 0.0005, "loss": 1.459713339805603, "step": 19928 }, { "epoch": 1.5710681907765078, "grad_norm": 0.5094053149223328, "learning_rate": 0.0005, "loss": 1.5034754276275635, "step": 19929 }, { "epoch": 1.5711470240441465, "grad_norm": 0.7006212472915649, "learning_rate": 0.0005, "loss": 1.5021979808807373, "step": 19930 }, { "epoch": 1.5712258573117857, "grad_norm": 0.54270339012146, "learning_rate": 0.0005, "loss": 1.6411086320877075, "step": 19931 }, { "epoch": 1.5713046905794246, "grad_norm": 0.4618132710456848, "learning_rate": 0.0005, "loss": 1.377242088317871, "step": 19932 }, { "epoch": 1.5713835238470635, "grad_norm": 0.517924964427948, "learning_rate": 0.0005, "loss": 1.4246898889541626, "step": 19933 }, { "epoch": 1.5714623571147024, "grad_norm": 0.742157518863678, "learning_rate": 0.0005, "loss": 1.527626633644104, "step": 19934 }, { "epoch": 1.5715411903823413, "grad_norm": 0.6224204301834106, "learning_rate": 0.0005, "loss": 1.524343729019165, "step": 19935 }, { "epoch": 1.5716200236499804, "grad_norm": 0.738197922706604, "learning_rate": 0.0005, "loss": 1.7889573574066162, "step": 19936 }, { "epoch": 1.5716988569176191, "grad_norm": 0.42455917596817017, "learning_rate": 0.0005, "loss": 1.4234397411346436, "step": 19937 }, { "epoch": 1.5717776901852583, "grad_norm": 0.561639130115509, "learning_rate": 0.0005, "loss": 1.7949323654174805, "step": 19938 }, { "epoch": 1.5718565234528972, "grad_norm": 0.6154459118843079, "learning_rate": 0.0005, "loss": 1.658434271812439, "step": 19939 }, { "epoch": 1.571935356720536, "grad_norm": 0.5964336395263672, "learning_rate": 0.0005, "loss": 1.5697988271713257, "step": 19940 }, { "epoch": 1.572014189988175, "grad_norm": 0.47486621141433716, "learning_rate": 0.0005, "loss": 1.6728394031524658, "step": 19941 }, { "epoch": 1.572093023255814, "grad_norm": 0.5357563495635986, "learning_rate": 0.0005, "loss": 1.687808632850647, "step": 19942 }, { "epoch": 1.572171856523453, "grad_norm": 0.47001394629478455, "learning_rate": 0.0005, "loss": 1.3004590272903442, "step": 19943 }, { "epoch": 1.5722506897910917, "grad_norm": 0.5240363478660583, "learning_rate": 0.0005, "loss": 1.3937060832977295, "step": 19944 }, { "epoch": 1.5723295230587309, "grad_norm": 0.49583762884140015, "learning_rate": 0.0005, "loss": 1.5026549100875854, "step": 19945 }, { "epoch": 1.5724083563263698, "grad_norm": 0.5638539791107178, "learning_rate": 0.0005, "loss": 1.3291511535644531, "step": 19946 }, { "epoch": 1.5724871895940087, "grad_norm": 0.5513716340065002, "learning_rate": 0.0005, "loss": 1.5105507373809814, "step": 19947 }, { "epoch": 1.5725660228616476, "grad_norm": 0.4778349697589874, "learning_rate": 0.0005, "loss": 1.368255615234375, "step": 19948 }, { "epoch": 1.5726448561292865, "grad_norm": 0.5903651118278503, "learning_rate": 0.0005, "loss": 1.6394751071929932, "step": 19949 }, { "epoch": 1.5727236893969256, "grad_norm": 0.5695475339889526, "learning_rate": 0.0005, "loss": 1.4837796688079834, "step": 19950 }, { "epoch": 1.5728025226645643, "grad_norm": 0.5335524082183838, "learning_rate": 0.0005, "loss": 1.4166574478149414, "step": 19951 }, { "epoch": 1.5728813559322035, "grad_norm": 0.4807003438472748, "learning_rate": 0.0005, "loss": 1.446114420890808, "step": 19952 }, { "epoch": 1.5729601891998424, "grad_norm": 0.6878008842468262, "learning_rate": 0.0005, "loss": 1.5721330642700195, "step": 19953 }, { "epoch": 1.5730390224674813, "grad_norm": 0.49317413568496704, "learning_rate": 0.0005, "loss": 1.4197115898132324, "step": 19954 }, { "epoch": 1.5731178557351202, "grad_norm": 0.6195862293243408, "learning_rate": 0.0005, "loss": 1.5402729511260986, "step": 19955 }, { "epoch": 1.573196689002759, "grad_norm": 0.6570501923561096, "learning_rate": 0.0005, "loss": 1.456112265586853, "step": 19956 }, { "epoch": 1.5732755222703982, "grad_norm": 0.7044344544410706, "learning_rate": 0.0005, "loss": 1.7770471572875977, "step": 19957 }, { "epoch": 1.573354355538037, "grad_norm": 0.5181312561035156, "learning_rate": 0.0005, "loss": 1.4844954013824463, "step": 19958 }, { "epoch": 1.573433188805676, "grad_norm": 0.4703270494937897, "learning_rate": 0.0005, "loss": 1.5724968910217285, "step": 19959 }, { "epoch": 1.573512022073315, "grad_norm": 0.5555245876312256, "learning_rate": 0.0005, "loss": 1.5172133445739746, "step": 19960 }, { "epoch": 1.5735908553409539, "grad_norm": 0.5227251052856445, "learning_rate": 0.0005, "loss": 1.3776342868804932, "step": 19961 }, { "epoch": 1.5736696886085928, "grad_norm": 0.5101168155670166, "learning_rate": 0.0005, "loss": 1.5947742462158203, "step": 19962 }, { "epoch": 1.5737485218762317, "grad_norm": 0.4916606545448303, "learning_rate": 0.0005, "loss": 1.4442248344421387, "step": 19963 }, { "epoch": 1.5738273551438708, "grad_norm": 0.5163887739181519, "learning_rate": 0.0005, "loss": 1.5167241096496582, "step": 19964 }, { "epoch": 1.5739061884115095, "grad_norm": 0.5376144051551819, "learning_rate": 0.0005, "loss": 1.6151983737945557, "step": 19965 }, { "epoch": 1.5739850216791487, "grad_norm": 0.8086995482444763, "learning_rate": 0.0005, "loss": 1.7528365850448608, "step": 19966 }, { "epoch": 1.5740638549467876, "grad_norm": 0.5903028249740601, "learning_rate": 0.0005, "loss": 1.6475896835327148, "step": 19967 }, { "epoch": 1.5741426882144265, "grad_norm": 0.5666660666465759, "learning_rate": 0.0005, "loss": 1.6561354398727417, "step": 19968 }, { "epoch": 1.5742215214820654, "grad_norm": 0.5178571343421936, "learning_rate": 0.0005, "loss": 1.3861961364746094, "step": 19969 }, { "epoch": 1.5743003547497043, "grad_norm": 0.647386908531189, "learning_rate": 0.0005, "loss": 1.655413031578064, "step": 19970 }, { "epoch": 1.5743791880173434, "grad_norm": 0.8454822301864624, "learning_rate": 0.0005, "loss": 1.669664978981018, "step": 19971 }, { "epoch": 1.5744580212849821, "grad_norm": 0.5761784315109253, "learning_rate": 0.0005, "loss": 1.8064080476760864, "step": 19972 }, { "epoch": 1.5745368545526213, "grad_norm": 0.5621063709259033, "learning_rate": 0.0005, "loss": 1.6024107933044434, "step": 19973 }, { "epoch": 1.5746156878202602, "grad_norm": 0.5572052001953125, "learning_rate": 0.0005, "loss": 1.5870410203933716, "step": 19974 }, { "epoch": 1.574694521087899, "grad_norm": 0.5340757966041565, "learning_rate": 0.0005, "loss": 1.521765947341919, "step": 19975 }, { "epoch": 1.574773354355538, "grad_norm": 0.3847709596157074, "learning_rate": 0.0005, "loss": 1.4243875741958618, "step": 19976 }, { "epoch": 1.574852187623177, "grad_norm": 0.6185348033905029, "learning_rate": 0.0005, "loss": 1.570279836654663, "step": 19977 }, { "epoch": 1.574931020890816, "grad_norm": 0.5556166172027588, "learning_rate": 0.0005, "loss": 1.5541388988494873, "step": 19978 }, { "epoch": 1.5750098541584547, "grad_norm": 0.489196240901947, "learning_rate": 0.0005, "loss": 1.4518424272537231, "step": 19979 }, { "epoch": 1.5750886874260939, "grad_norm": 0.516140878200531, "learning_rate": 0.0005, "loss": 1.3570547103881836, "step": 19980 }, { "epoch": 1.5751675206937328, "grad_norm": 0.4801502525806427, "learning_rate": 0.0005, "loss": 1.7059885263442993, "step": 19981 }, { "epoch": 1.5752463539613717, "grad_norm": 0.5353987812995911, "learning_rate": 0.0005, "loss": 1.513854742050171, "step": 19982 }, { "epoch": 1.5753251872290106, "grad_norm": 0.45753610134124756, "learning_rate": 0.0005, "loss": 1.673701524734497, "step": 19983 }, { "epoch": 1.5754040204966495, "grad_norm": 0.6774518489837646, "learning_rate": 0.0005, "loss": 1.7543550729751587, "step": 19984 }, { "epoch": 1.5754828537642886, "grad_norm": 0.49618008732795715, "learning_rate": 0.0005, "loss": 1.4078552722930908, "step": 19985 }, { "epoch": 1.5755616870319273, "grad_norm": 0.611081063747406, "learning_rate": 0.0005, "loss": 1.5407111644744873, "step": 19986 }, { "epoch": 1.5756405202995665, "grad_norm": 0.4781423509120941, "learning_rate": 0.0005, "loss": 1.5753940343856812, "step": 19987 }, { "epoch": 1.5757193535672054, "grad_norm": 0.48474258184432983, "learning_rate": 0.0005, "loss": 1.674117088317871, "step": 19988 }, { "epoch": 1.5757981868348443, "grad_norm": 0.6362603306770325, "learning_rate": 0.0005, "loss": 1.7123467922210693, "step": 19989 }, { "epoch": 1.5758770201024832, "grad_norm": 0.7909643650054932, "learning_rate": 0.0005, "loss": 1.6252145767211914, "step": 19990 }, { "epoch": 1.5759558533701221, "grad_norm": 0.6751253604888916, "learning_rate": 0.0005, "loss": 1.6013293266296387, "step": 19991 }, { "epoch": 1.5760346866377613, "grad_norm": 0.5870827436447144, "learning_rate": 0.0005, "loss": 1.484285831451416, "step": 19992 }, { "epoch": 1.5761135199054, "grad_norm": 0.5446977615356445, "learning_rate": 0.0005, "loss": 1.5846493244171143, "step": 19993 }, { "epoch": 1.576192353173039, "grad_norm": 0.5849726796150208, "learning_rate": 0.0005, "loss": 1.5650430917739868, "step": 19994 }, { "epoch": 1.576271186440678, "grad_norm": 0.49869877099990845, "learning_rate": 0.0005, "loss": 1.4527223110198975, "step": 19995 }, { "epoch": 1.576350019708317, "grad_norm": 0.7036727666854858, "learning_rate": 0.0005, "loss": 1.5519566535949707, "step": 19996 }, { "epoch": 1.5764288529759558, "grad_norm": 0.556085467338562, "learning_rate": 0.0005, "loss": 1.5050878524780273, "step": 19997 }, { "epoch": 1.5765076862435947, "grad_norm": 0.43622514605522156, "learning_rate": 0.0005, "loss": 1.3003411293029785, "step": 19998 }, { "epoch": 1.5765865195112339, "grad_norm": 0.5488667488098145, "learning_rate": 0.0005, "loss": 1.5718095302581787, "step": 19999 }, { "epoch": 1.5766653527788725, "grad_norm": 0.5121585130691528, "learning_rate": 0.0005, "loss": 1.7926112413406372, "step": 20000 }, { "epoch": 1.5767441860465117, "grad_norm": 0.526033341884613, "learning_rate": 0.0005, "loss": 1.6448101997375488, "step": 20001 }, { "epoch": 1.5768230193141506, "grad_norm": 0.5018616318702698, "learning_rate": 0.0005, "loss": 1.627398133277893, "step": 20002 }, { "epoch": 1.5769018525817895, "grad_norm": 0.5108018517494202, "learning_rate": 0.0005, "loss": 1.667128086090088, "step": 20003 }, { "epoch": 1.5769806858494284, "grad_norm": 0.6271840333938599, "learning_rate": 0.0005, "loss": 1.6391949653625488, "step": 20004 }, { "epoch": 1.5770595191170673, "grad_norm": 0.5104289054870605, "learning_rate": 0.0005, "loss": 1.5184564590454102, "step": 20005 }, { "epoch": 1.5771383523847065, "grad_norm": 0.49527737498283386, "learning_rate": 0.0005, "loss": 1.4929156303405762, "step": 20006 }, { "epoch": 1.5772171856523451, "grad_norm": 0.592999279499054, "learning_rate": 0.0005, "loss": 1.5149227380752563, "step": 20007 }, { "epoch": 1.5772960189199843, "grad_norm": 0.5723611116409302, "learning_rate": 0.0005, "loss": 1.5509830713272095, "step": 20008 }, { "epoch": 1.5773748521876232, "grad_norm": 0.5699309706687927, "learning_rate": 0.0005, "loss": 1.5578222274780273, "step": 20009 }, { "epoch": 1.577453685455262, "grad_norm": 0.45231619477272034, "learning_rate": 0.0005, "loss": 1.3836959600448608, "step": 20010 }, { "epoch": 1.577532518722901, "grad_norm": 0.6258509159088135, "learning_rate": 0.0005, "loss": 1.6311582326889038, "step": 20011 }, { "epoch": 1.57761135199054, "grad_norm": 0.434650719165802, "learning_rate": 0.0005, "loss": 1.4134236574172974, "step": 20012 }, { "epoch": 1.577690185258179, "grad_norm": 0.6738800406455994, "learning_rate": 0.0005, "loss": 1.6509735584259033, "step": 20013 }, { "epoch": 1.5777690185258177, "grad_norm": 0.5405076146125793, "learning_rate": 0.0005, "loss": 1.4109715223312378, "step": 20014 }, { "epoch": 1.5778478517934569, "grad_norm": 0.5526902675628662, "learning_rate": 0.0005, "loss": 1.4995713233947754, "step": 20015 }, { "epoch": 1.5779266850610958, "grad_norm": 0.5613029599189758, "learning_rate": 0.0005, "loss": 1.748056411743164, "step": 20016 }, { "epoch": 1.5780055183287347, "grad_norm": 0.683659017086029, "learning_rate": 0.0005, "loss": 1.5414419174194336, "step": 20017 }, { "epoch": 1.5780843515963738, "grad_norm": 0.527938723564148, "learning_rate": 0.0005, "loss": 1.6401398181915283, "step": 20018 }, { "epoch": 1.5781631848640125, "grad_norm": 0.5195587873458862, "learning_rate": 0.0005, "loss": 1.6384994983673096, "step": 20019 }, { "epoch": 1.5782420181316517, "grad_norm": 0.5358655452728271, "learning_rate": 0.0005, "loss": 1.570749044418335, "step": 20020 }, { "epoch": 1.5783208513992903, "grad_norm": 0.5312113761901855, "learning_rate": 0.0005, "loss": 1.4825539588928223, "step": 20021 }, { "epoch": 1.5783996846669295, "grad_norm": 0.4708325266838074, "learning_rate": 0.0005, "loss": 1.383565068244934, "step": 20022 }, { "epoch": 1.5784785179345684, "grad_norm": 0.6335670351982117, "learning_rate": 0.0005, "loss": 1.406121015548706, "step": 20023 }, { "epoch": 1.5785573512022073, "grad_norm": 0.5769656300544739, "learning_rate": 0.0005, "loss": 1.6093999147415161, "step": 20024 }, { "epoch": 1.5786361844698464, "grad_norm": 0.5121458768844604, "learning_rate": 0.0005, "loss": 1.4507384300231934, "step": 20025 }, { "epoch": 1.5787150177374851, "grad_norm": 0.5655582547187805, "learning_rate": 0.0005, "loss": 1.700329065322876, "step": 20026 }, { "epoch": 1.5787938510051243, "grad_norm": 0.5249013304710388, "learning_rate": 0.0005, "loss": 1.4504806995391846, "step": 20027 }, { "epoch": 1.578872684272763, "grad_norm": 0.47413620352745056, "learning_rate": 0.0005, "loss": 1.421999454498291, "step": 20028 }, { "epoch": 1.578951517540402, "grad_norm": 0.5094723105430603, "learning_rate": 0.0005, "loss": 1.502589225769043, "step": 20029 }, { "epoch": 1.579030350808041, "grad_norm": 0.49245205521583557, "learning_rate": 0.0005, "loss": 1.5600477457046509, "step": 20030 }, { "epoch": 1.57910918407568, "grad_norm": 0.5493650436401367, "learning_rate": 0.0005, "loss": 1.850649356842041, "step": 20031 }, { "epoch": 1.579188017343319, "grad_norm": 0.6004036068916321, "learning_rate": 0.0005, "loss": 1.7736458778381348, "step": 20032 }, { "epoch": 1.5792668506109577, "grad_norm": 0.4613739848136902, "learning_rate": 0.0005, "loss": 1.7314956188201904, "step": 20033 }, { "epoch": 1.5793456838785969, "grad_norm": 0.6506115198135376, "learning_rate": 0.0005, "loss": 1.4410231113433838, "step": 20034 }, { "epoch": 1.5794245171462356, "grad_norm": 0.5763068795204163, "learning_rate": 0.0005, "loss": 1.4836771488189697, "step": 20035 }, { "epoch": 1.5795033504138747, "grad_norm": 0.7279819250106812, "learning_rate": 0.0005, "loss": 1.7182711362838745, "step": 20036 }, { "epoch": 1.5795821836815136, "grad_norm": 0.5660682916641235, "learning_rate": 0.0005, "loss": 1.425559639930725, "step": 20037 }, { "epoch": 1.5796610169491525, "grad_norm": 0.5408155918121338, "learning_rate": 0.0005, "loss": 1.4537994861602783, "step": 20038 }, { "epoch": 1.5797398502167916, "grad_norm": 0.5930622220039368, "learning_rate": 0.0005, "loss": 1.7432256937026978, "step": 20039 }, { "epoch": 1.5798186834844303, "grad_norm": 0.6209388971328735, "learning_rate": 0.0005, "loss": 1.4663128852844238, "step": 20040 }, { "epoch": 1.5798975167520695, "grad_norm": 0.6317963004112244, "learning_rate": 0.0005, "loss": 1.4992185831069946, "step": 20041 }, { "epoch": 1.5799763500197082, "grad_norm": 0.4732333719730377, "learning_rate": 0.0005, "loss": 1.486732006072998, "step": 20042 }, { "epoch": 1.5800551832873473, "grad_norm": 0.5110026001930237, "learning_rate": 0.0005, "loss": 1.404806137084961, "step": 20043 }, { "epoch": 1.5801340165549862, "grad_norm": 0.4285452961921692, "learning_rate": 0.0005, "loss": 1.2640776634216309, "step": 20044 }, { "epoch": 1.580212849822625, "grad_norm": 0.5178750157356262, "learning_rate": 0.0005, "loss": 1.659637689590454, "step": 20045 }, { "epoch": 1.5802916830902642, "grad_norm": 0.5047821402549744, "learning_rate": 0.0005, "loss": 1.6572027206420898, "step": 20046 }, { "epoch": 1.580370516357903, "grad_norm": 0.6025425791740417, "learning_rate": 0.0005, "loss": 1.4664978981018066, "step": 20047 }, { "epoch": 1.580449349625542, "grad_norm": 0.4689987897872925, "learning_rate": 0.0005, "loss": 1.2301232814788818, "step": 20048 }, { "epoch": 1.580528182893181, "grad_norm": 0.593807578086853, "learning_rate": 0.0005, "loss": 1.4341281652450562, "step": 20049 }, { "epoch": 1.5806070161608199, "grad_norm": 0.5454502105712891, "learning_rate": 0.0005, "loss": 1.6898376941680908, "step": 20050 }, { "epoch": 1.5806858494284588, "grad_norm": 0.49229857325553894, "learning_rate": 0.0005, "loss": 1.5752601623535156, "step": 20051 }, { "epoch": 1.5807646826960977, "grad_norm": 0.5181506276130676, "learning_rate": 0.0005, "loss": 1.72629714012146, "step": 20052 }, { "epoch": 1.5808435159637368, "grad_norm": 0.5094754099845886, "learning_rate": 0.0005, "loss": 1.6767395734786987, "step": 20053 }, { "epoch": 1.5809223492313755, "grad_norm": 0.5388217568397522, "learning_rate": 0.0005, "loss": 1.5994020700454712, "step": 20054 }, { "epoch": 1.5810011824990147, "grad_norm": 0.6336783766746521, "learning_rate": 0.0005, "loss": 1.4497936964035034, "step": 20055 }, { "epoch": 1.5810800157666536, "grad_norm": 0.6808990836143494, "learning_rate": 0.0005, "loss": 1.8602670431137085, "step": 20056 }, { "epoch": 1.5811588490342925, "grad_norm": 0.6144062876701355, "learning_rate": 0.0005, "loss": 1.5040066242218018, "step": 20057 }, { "epoch": 1.5812376823019314, "grad_norm": 0.6138811707496643, "learning_rate": 0.0005, "loss": 1.5485426187515259, "step": 20058 }, { "epoch": 1.5813165155695703, "grad_norm": 0.46002718806266785, "learning_rate": 0.0005, "loss": 1.5410665273666382, "step": 20059 }, { "epoch": 1.5813953488372094, "grad_norm": 0.6341962218284607, "learning_rate": 0.0005, "loss": 1.9839646816253662, "step": 20060 }, { "epoch": 1.5814741821048481, "grad_norm": 0.6929411292076111, "learning_rate": 0.0005, "loss": 1.5382592678070068, "step": 20061 }, { "epoch": 1.5815530153724873, "grad_norm": 0.5296130776405334, "learning_rate": 0.0005, "loss": 1.471045970916748, "step": 20062 }, { "epoch": 1.5816318486401262, "grad_norm": 0.4869406223297119, "learning_rate": 0.0005, "loss": 1.6456867456436157, "step": 20063 }, { "epoch": 1.581710681907765, "grad_norm": 0.5549480319023132, "learning_rate": 0.0005, "loss": 1.7005138397216797, "step": 20064 }, { "epoch": 1.581789515175404, "grad_norm": 0.4570363759994507, "learning_rate": 0.0005, "loss": 1.3685212135314941, "step": 20065 }, { "epoch": 1.581868348443043, "grad_norm": 0.7756417989730835, "learning_rate": 0.0005, "loss": 1.6076350212097168, "step": 20066 }, { "epoch": 1.581947181710682, "grad_norm": 0.6168166995048523, "learning_rate": 0.0005, "loss": 1.536350965499878, "step": 20067 }, { "epoch": 1.5820260149783207, "grad_norm": 0.6543287634849548, "learning_rate": 0.0005, "loss": 1.5693379640579224, "step": 20068 }, { "epoch": 1.5821048482459599, "grad_norm": 0.5278130769729614, "learning_rate": 0.0005, "loss": 1.4915326833724976, "step": 20069 }, { "epoch": 1.5821836815135988, "grad_norm": 0.4870705008506775, "learning_rate": 0.0005, "loss": 1.530439019203186, "step": 20070 }, { "epoch": 1.5822625147812377, "grad_norm": 0.4908730983734131, "learning_rate": 0.0005, "loss": 1.6328383684158325, "step": 20071 }, { "epoch": 1.5823413480488766, "grad_norm": 0.7084442377090454, "learning_rate": 0.0005, "loss": 1.5559861660003662, "step": 20072 }, { "epoch": 1.5824201813165155, "grad_norm": 0.4717891216278076, "learning_rate": 0.0005, "loss": 1.5268476009368896, "step": 20073 }, { "epoch": 1.5824990145841547, "grad_norm": 1.3757529258728027, "learning_rate": 0.0005, "loss": 1.7175315618515015, "step": 20074 }, { "epoch": 1.5825778478517933, "grad_norm": 0.4979763925075531, "learning_rate": 0.0005, "loss": 1.4603271484375, "step": 20075 }, { "epoch": 1.5826566811194325, "grad_norm": 0.5542280673980713, "learning_rate": 0.0005, "loss": 1.5217788219451904, "step": 20076 }, { "epoch": 1.5827355143870714, "grad_norm": 0.47064247727394104, "learning_rate": 0.0005, "loss": 1.5786525011062622, "step": 20077 }, { "epoch": 1.5828143476547103, "grad_norm": 0.6701662540435791, "learning_rate": 0.0005, "loss": 1.7067594528198242, "step": 20078 }, { "epoch": 1.5828931809223492, "grad_norm": 0.5396292209625244, "learning_rate": 0.0005, "loss": 1.6921730041503906, "step": 20079 }, { "epoch": 1.5829720141899881, "grad_norm": 0.5033188462257385, "learning_rate": 0.0005, "loss": 1.6283113956451416, "step": 20080 }, { "epoch": 1.5830508474576273, "grad_norm": 0.46077272295951843, "learning_rate": 0.0005, "loss": 1.4852967262268066, "step": 20081 }, { "epoch": 1.583129680725266, "grad_norm": 0.5071188807487488, "learning_rate": 0.0005, "loss": 1.4764988422393799, "step": 20082 }, { "epoch": 1.583208513992905, "grad_norm": 0.4896652102470398, "learning_rate": 0.0005, "loss": 1.5733916759490967, "step": 20083 }, { "epoch": 1.583287347260544, "grad_norm": 0.5307304263114929, "learning_rate": 0.0005, "loss": 1.542913794517517, "step": 20084 }, { "epoch": 1.583366180528183, "grad_norm": 0.5021454691886902, "learning_rate": 0.0005, "loss": 1.4863009452819824, "step": 20085 }, { "epoch": 1.5834450137958218, "grad_norm": 0.5426695942878723, "learning_rate": 0.0005, "loss": 1.9180554151535034, "step": 20086 }, { "epoch": 1.5835238470634607, "grad_norm": 0.590164065361023, "learning_rate": 0.0005, "loss": 1.6125702857971191, "step": 20087 }, { "epoch": 1.5836026803310999, "grad_norm": 0.5978298783302307, "learning_rate": 0.0005, "loss": 1.7044012546539307, "step": 20088 }, { "epoch": 1.5836815135987385, "grad_norm": 0.4779786765575409, "learning_rate": 0.0005, "loss": 1.4881558418273926, "step": 20089 }, { "epoch": 1.5837603468663777, "grad_norm": 0.638698160648346, "learning_rate": 0.0005, "loss": 1.405362606048584, "step": 20090 }, { "epoch": 1.5838391801340166, "grad_norm": 0.45399272441864014, "learning_rate": 0.0005, "loss": 1.4969085454940796, "step": 20091 }, { "epoch": 1.5839180134016555, "grad_norm": 0.5599287152290344, "learning_rate": 0.0005, "loss": 1.4563695192337036, "step": 20092 }, { "epoch": 1.5839968466692944, "grad_norm": 0.7540850043296814, "learning_rate": 0.0005, "loss": 1.7135100364685059, "step": 20093 }, { "epoch": 1.5840756799369333, "grad_norm": 0.480146586894989, "learning_rate": 0.0005, "loss": 1.4575767517089844, "step": 20094 }, { "epoch": 1.5841545132045725, "grad_norm": 0.46585872769355774, "learning_rate": 0.0005, "loss": 1.4771289825439453, "step": 20095 }, { "epoch": 1.5842333464722111, "grad_norm": 0.7886176109313965, "learning_rate": 0.0005, "loss": 1.584325909614563, "step": 20096 }, { "epoch": 1.5843121797398503, "grad_norm": 0.5642716288566589, "learning_rate": 0.0005, "loss": 1.6293121576309204, "step": 20097 }, { "epoch": 1.5843910130074892, "grad_norm": 0.5184932947158813, "learning_rate": 0.0005, "loss": 1.363161563873291, "step": 20098 }, { "epoch": 1.584469846275128, "grad_norm": 0.540571928024292, "learning_rate": 0.0005, "loss": 1.6608092784881592, "step": 20099 }, { "epoch": 1.584548679542767, "grad_norm": 0.5573523640632629, "learning_rate": 0.0005, "loss": 1.4950573444366455, "step": 20100 }, { "epoch": 1.584627512810406, "grad_norm": 0.4526379704475403, "learning_rate": 0.0005, "loss": 1.3574364185333252, "step": 20101 }, { "epoch": 1.584706346078045, "grad_norm": 0.49295416474342346, "learning_rate": 0.0005, "loss": 1.5120081901550293, "step": 20102 }, { "epoch": 1.5847851793456837, "grad_norm": 0.5909630656242371, "learning_rate": 0.0005, "loss": 1.7900121212005615, "step": 20103 }, { "epoch": 1.5848640126133229, "grad_norm": 0.5055525898933411, "learning_rate": 0.0005, "loss": 1.5524420738220215, "step": 20104 }, { "epoch": 1.5849428458809618, "grad_norm": 0.5883702635765076, "learning_rate": 0.0005, "loss": 1.5216296911239624, "step": 20105 }, { "epoch": 1.5850216791486007, "grad_norm": 0.4907141923904419, "learning_rate": 0.0005, "loss": 1.461307168006897, "step": 20106 }, { "epoch": 1.5851005124162396, "grad_norm": 0.5840463042259216, "learning_rate": 0.0005, "loss": 1.6748881340026855, "step": 20107 }, { "epoch": 1.5851793456838785, "grad_norm": 0.728513240814209, "learning_rate": 0.0005, "loss": 1.5321049690246582, "step": 20108 }, { "epoch": 1.5852581789515177, "grad_norm": 0.599566638469696, "learning_rate": 0.0005, "loss": 1.5661170482635498, "step": 20109 }, { "epoch": 1.5853370122191563, "grad_norm": 0.6969887614250183, "learning_rate": 0.0005, "loss": 1.6764377355575562, "step": 20110 }, { "epoch": 1.5854158454867955, "grad_norm": 0.5647141337394714, "learning_rate": 0.0005, "loss": 1.5140310525894165, "step": 20111 }, { "epoch": 1.5854946787544344, "grad_norm": 0.5234126448631287, "learning_rate": 0.0005, "loss": 1.384472131729126, "step": 20112 }, { "epoch": 1.5855735120220733, "grad_norm": 0.4870048761367798, "learning_rate": 0.0005, "loss": 1.4596647024154663, "step": 20113 }, { "epoch": 1.5856523452897122, "grad_norm": 0.5176976919174194, "learning_rate": 0.0005, "loss": 1.5228865146636963, "step": 20114 }, { "epoch": 1.5857311785573511, "grad_norm": 0.5473840832710266, "learning_rate": 0.0005, "loss": 1.6847097873687744, "step": 20115 }, { "epoch": 1.5858100118249903, "grad_norm": 0.5236552953720093, "learning_rate": 0.0005, "loss": 1.6325907707214355, "step": 20116 }, { "epoch": 1.585888845092629, "grad_norm": 0.42853668332099915, "learning_rate": 0.0005, "loss": 1.310107707977295, "step": 20117 }, { "epoch": 1.585967678360268, "grad_norm": 0.5144787430763245, "learning_rate": 0.0005, "loss": 1.6473822593688965, "step": 20118 }, { "epoch": 1.586046511627907, "grad_norm": 0.5471339821815491, "learning_rate": 0.0005, "loss": 1.4553208351135254, "step": 20119 }, { "epoch": 1.586125344895546, "grad_norm": 0.6339956521987915, "learning_rate": 0.0005, "loss": 1.7176607847213745, "step": 20120 }, { "epoch": 1.5862041781631848, "grad_norm": 0.637601912021637, "learning_rate": 0.0005, "loss": 1.5733106136322021, "step": 20121 }, { "epoch": 1.5862830114308237, "grad_norm": 0.6184015274047852, "learning_rate": 0.0005, "loss": 1.6264402866363525, "step": 20122 }, { "epoch": 1.5863618446984629, "grad_norm": 0.5627318620681763, "learning_rate": 0.0005, "loss": 1.3879352807998657, "step": 20123 }, { "epoch": 1.5864406779661016, "grad_norm": 0.6306058764457703, "learning_rate": 0.0005, "loss": 1.721421241760254, "step": 20124 }, { "epoch": 1.5865195112337407, "grad_norm": 0.5626788139343262, "learning_rate": 0.0005, "loss": 1.6936547756195068, "step": 20125 }, { "epoch": 1.5865983445013796, "grad_norm": 0.5493868589401245, "learning_rate": 0.0005, "loss": 1.5439170598983765, "step": 20126 }, { "epoch": 1.5866771777690185, "grad_norm": 0.6021806597709656, "learning_rate": 0.0005, "loss": 1.8131438493728638, "step": 20127 }, { "epoch": 1.5867560110366574, "grad_norm": 0.5447887778282166, "learning_rate": 0.0005, "loss": 1.3701163530349731, "step": 20128 }, { "epoch": 1.5868348443042963, "grad_norm": 0.5836582779884338, "learning_rate": 0.0005, "loss": 1.5506620407104492, "step": 20129 }, { "epoch": 1.5869136775719355, "grad_norm": 0.5381477475166321, "learning_rate": 0.0005, "loss": 1.4796775579452515, "step": 20130 }, { "epoch": 1.5869925108395742, "grad_norm": 0.6878584623336792, "learning_rate": 0.0005, "loss": 1.5922664403915405, "step": 20131 }, { "epoch": 1.5870713441072133, "grad_norm": 0.5772703289985657, "learning_rate": 0.0005, "loss": 1.8164925575256348, "step": 20132 }, { "epoch": 1.5871501773748522, "grad_norm": 0.5611766576766968, "learning_rate": 0.0005, "loss": 1.5979735851287842, "step": 20133 }, { "epoch": 1.5872290106424911, "grad_norm": 0.6182534098625183, "learning_rate": 0.0005, "loss": 1.618438959121704, "step": 20134 }, { "epoch": 1.58730784391013, "grad_norm": 0.4749431908130646, "learning_rate": 0.0005, "loss": 1.2047696113586426, "step": 20135 }, { "epoch": 1.587386677177769, "grad_norm": 0.49873676896095276, "learning_rate": 0.0005, "loss": 1.3716251850128174, "step": 20136 }, { "epoch": 1.587465510445408, "grad_norm": 0.5926699042320251, "learning_rate": 0.0005, "loss": 1.4069483280181885, "step": 20137 }, { "epoch": 1.5875443437130468, "grad_norm": 0.4990085959434509, "learning_rate": 0.0005, "loss": 1.5280588865280151, "step": 20138 }, { "epoch": 1.587623176980686, "grad_norm": 0.6425282955169678, "learning_rate": 0.0005, "loss": 1.7325916290283203, "step": 20139 }, { "epoch": 1.5877020102483248, "grad_norm": 0.5548253655433655, "learning_rate": 0.0005, "loss": 1.5500073432922363, "step": 20140 }, { "epoch": 1.5877808435159637, "grad_norm": 0.49639418721199036, "learning_rate": 0.0005, "loss": 1.6232019662857056, "step": 20141 }, { "epoch": 1.5878596767836026, "grad_norm": 0.6093775629997253, "learning_rate": 0.0005, "loss": 1.6291933059692383, "step": 20142 }, { "epoch": 1.5879385100512415, "grad_norm": 0.6359719038009644, "learning_rate": 0.0005, "loss": 1.7253714799880981, "step": 20143 }, { "epoch": 1.5880173433188807, "grad_norm": 0.5155386328697205, "learning_rate": 0.0005, "loss": 1.3926382064819336, "step": 20144 }, { "epoch": 1.5880961765865194, "grad_norm": 0.6457966566085815, "learning_rate": 0.0005, "loss": 1.544257402420044, "step": 20145 }, { "epoch": 1.5881750098541585, "grad_norm": 0.5314083695411682, "learning_rate": 0.0005, "loss": 1.5593664646148682, "step": 20146 }, { "epoch": 1.5882538431217974, "grad_norm": 0.791240394115448, "learning_rate": 0.0005, "loss": 1.5232957601547241, "step": 20147 }, { "epoch": 1.5883326763894363, "grad_norm": 0.5523686408996582, "learning_rate": 0.0005, "loss": 1.6154066324234009, "step": 20148 }, { "epoch": 1.5884115096570754, "grad_norm": 0.5567497611045837, "learning_rate": 0.0005, "loss": 1.4878642559051514, "step": 20149 }, { "epoch": 1.5884903429247141, "grad_norm": 0.6205223202705383, "learning_rate": 0.0005, "loss": 1.7032233476638794, "step": 20150 }, { "epoch": 1.5885691761923533, "grad_norm": 0.6099977493286133, "learning_rate": 0.0005, "loss": 1.5156958103179932, "step": 20151 }, { "epoch": 1.588648009459992, "grad_norm": 0.5984874963760376, "learning_rate": 0.0005, "loss": 1.6696844100952148, "step": 20152 }, { "epoch": 1.588726842727631, "grad_norm": 0.8032742142677307, "learning_rate": 0.0005, "loss": 1.333247423171997, "step": 20153 }, { "epoch": 1.58880567599527, "grad_norm": 0.4373151659965515, "learning_rate": 0.0005, "loss": 1.469832181930542, "step": 20154 }, { "epoch": 1.588884509262909, "grad_norm": 0.620126485824585, "learning_rate": 0.0005, "loss": 1.5294909477233887, "step": 20155 }, { "epoch": 1.588963342530548, "grad_norm": 0.5635503530502319, "learning_rate": 0.0005, "loss": 1.5962834358215332, "step": 20156 }, { "epoch": 1.5890421757981867, "grad_norm": 0.5083311200141907, "learning_rate": 0.0005, "loss": 1.8069932460784912, "step": 20157 }, { "epoch": 1.5891210090658259, "grad_norm": 0.5086647272109985, "learning_rate": 0.0005, "loss": 1.6490880250930786, "step": 20158 }, { "epoch": 1.5891998423334646, "grad_norm": 0.6587280631065369, "learning_rate": 0.0005, "loss": 1.4666248559951782, "step": 20159 }, { "epoch": 1.5892786756011037, "grad_norm": 0.6103153824806213, "learning_rate": 0.0005, "loss": 1.2970850467681885, "step": 20160 }, { "epoch": 1.5893575088687426, "grad_norm": 0.5495579242706299, "learning_rate": 0.0005, "loss": 1.560276746749878, "step": 20161 }, { "epoch": 1.5894363421363815, "grad_norm": 0.5110227465629578, "learning_rate": 0.0005, "loss": 1.7184351682662964, "step": 20162 }, { "epoch": 1.5895151754040207, "grad_norm": 0.8544759750366211, "learning_rate": 0.0005, "loss": 1.7339930534362793, "step": 20163 }, { "epoch": 1.5895940086716593, "grad_norm": 0.4755246639251709, "learning_rate": 0.0005, "loss": 1.4086605310440063, "step": 20164 }, { "epoch": 1.5896728419392985, "grad_norm": 0.4728875756263733, "learning_rate": 0.0005, "loss": 1.4909882545471191, "step": 20165 }, { "epoch": 1.5897516752069372, "grad_norm": 0.4906085133552551, "learning_rate": 0.0005, "loss": 1.392741084098816, "step": 20166 }, { "epoch": 1.5898305084745763, "grad_norm": 0.4703906774520874, "learning_rate": 0.0005, "loss": 1.3203402757644653, "step": 20167 }, { "epoch": 1.5899093417422152, "grad_norm": 0.5502248406410217, "learning_rate": 0.0005, "loss": 1.6120012998580933, "step": 20168 }, { "epoch": 1.5899881750098541, "grad_norm": 0.5238102078437805, "learning_rate": 0.0005, "loss": 1.6176141500473022, "step": 20169 }, { "epoch": 1.5900670082774933, "grad_norm": 0.6751999855041504, "learning_rate": 0.0005, "loss": 1.7315995693206787, "step": 20170 }, { "epoch": 1.590145841545132, "grad_norm": 0.5354189872741699, "learning_rate": 0.0005, "loss": 1.5424411296844482, "step": 20171 }, { "epoch": 1.590224674812771, "grad_norm": 0.5647180676460266, "learning_rate": 0.0005, "loss": 1.9855833053588867, "step": 20172 }, { "epoch": 1.5903035080804098, "grad_norm": 0.6315957903862, "learning_rate": 0.0005, "loss": 1.4780395030975342, "step": 20173 }, { "epoch": 1.590382341348049, "grad_norm": 0.6321343779563904, "learning_rate": 0.0005, "loss": 1.7309895753860474, "step": 20174 }, { "epoch": 1.5904611746156878, "grad_norm": 0.5034669041633606, "learning_rate": 0.0005, "loss": 1.442095398902893, "step": 20175 }, { "epoch": 1.5905400078833267, "grad_norm": 0.49003270268440247, "learning_rate": 0.0005, "loss": 1.5462045669555664, "step": 20176 }, { "epoch": 1.5906188411509659, "grad_norm": 0.573429524898529, "learning_rate": 0.0005, "loss": 1.448575735092163, "step": 20177 }, { "epoch": 1.5906976744186045, "grad_norm": 0.531572163105011, "learning_rate": 0.0005, "loss": 1.457322597503662, "step": 20178 }, { "epoch": 1.5907765076862437, "grad_norm": 0.5242337584495544, "learning_rate": 0.0005, "loss": 1.6411051750183105, "step": 20179 }, { "epoch": 1.5908553409538826, "grad_norm": 0.5651090741157532, "learning_rate": 0.0005, "loss": 1.7408392429351807, "step": 20180 }, { "epoch": 1.5909341742215215, "grad_norm": 0.6576106548309326, "learning_rate": 0.0005, "loss": 1.7002177238464355, "step": 20181 }, { "epoch": 1.5910130074891604, "grad_norm": 0.5712661743164062, "learning_rate": 0.0005, "loss": 1.494766116142273, "step": 20182 }, { "epoch": 1.5910918407567993, "grad_norm": 0.5627052783966064, "learning_rate": 0.0005, "loss": 1.439476728439331, "step": 20183 }, { "epoch": 1.5911706740244385, "grad_norm": 0.6222819685935974, "learning_rate": 0.0005, "loss": 1.6087074279785156, "step": 20184 }, { "epoch": 1.5912495072920771, "grad_norm": 0.6494059562683105, "learning_rate": 0.0005, "loss": 1.760382890701294, "step": 20185 }, { "epoch": 1.5913283405597163, "grad_norm": 0.45512890815734863, "learning_rate": 0.0005, "loss": 1.468899130821228, "step": 20186 }, { "epoch": 1.5914071738273552, "grad_norm": 0.5777000784873962, "learning_rate": 0.0005, "loss": 1.4249345064163208, "step": 20187 }, { "epoch": 1.591486007094994, "grad_norm": 0.5661172270774841, "learning_rate": 0.0005, "loss": 1.5052745342254639, "step": 20188 }, { "epoch": 1.591564840362633, "grad_norm": 0.5072908997535706, "learning_rate": 0.0005, "loss": 1.7215949296951294, "step": 20189 }, { "epoch": 1.591643673630272, "grad_norm": 0.9648778438568115, "learning_rate": 0.0005, "loss": 1.3494490385055542, "step": 20190 }, { "epoch": 1.591722506897911, "grad_norm": 0.4907442629337311, "learning_rate": 0.0005, "loss": 1.3910373449325562, "step": 20191 }, { "epoch": 1.5918013401655497, "grad_norm": 0.47081977128982544, "learning_rate": 0.0005, "loss": 1.2894468307495117, "step": 20192 }, { "epoch": 1.5918801734331889, "grad_norm": 0.567572295665741, "learning_rate": 0.0005, "loss": 1.5273311138153076, "step": 20193 }, { "epoch": 1.5919590067008278, "grad_norm": 0.4606159031391144, "learning_rate": 0.0005, "loss": 1.432596206665039, "step": 20194 }, { "epoch": 1.5920378399684667, "grad_norm": 0.6367543935775757, "learning_rate": 0.0005, "loss": 1.8062634468078613, "step": 20195 }, { "epoch": 1.5921166732361056, "grad_norm": 0.5172688961029053, "learning_rate": 0.0005, "loss": 1.5164875984191895, "step": 20196 }, { "epoch": 1.5921955065037445, "grad_norm": 0.5840278267860413, "learning_rate": 0.0005, "loss": 1.6336381435394287, "step": 20197 }, { "epoch": 1.5922743397713837, "grad_norm": 0.54144287109375, "learning_rate": 0.0005, "loss": 1.5390311479568481, "step": 20198 }, { "epoch": 1.5923531730390224, "grad_norm": 0.5946009755134583, "learning_rate": 0.0005, "loss": 1.5195860862731934, "step": 20199 }, { "epoch": 1.5924320063066615, "grad_norm": 0.7396077513694763, "learning_rate": 0.0005, "loss": 1.603283166885376, "step": 20200 }, { "epoch": 1.5925108395743004, "grad_norm": 0.5759289264678955, "learning_rate": 0.0005, "loss": 1.673986554145813, "step": 20201 }, { "epoch": 1.5925896728419393, "grad_norm": 0.5617164373397827, "learning_rate": 0.0005, "loss": 1.6213761568069458, "step": 20202 }, { "epoch": 1.5926685061095782, "grad_norm": 0.5163395404815674, "learning_rate": 0.0005, "loss": 1.71282160282135, "step": 20203 }, { "epoch": 1.5927473393772171, "grad_norm": 0.5585283637046814, "learning_rate": 0.0005, "loss": 1.5919272899627686, "step": 20204 }, { "epoch": 1.5928261726448563, "grad_norm": 0.720607578754425, "learning_rate": 0.0005, "loss": 1.6605887413024902, "step": 20205 }, { "epoch": 1.592905005912495, "grad_norm": 0.598469614982605, "learning_rate": 0.0005, "loss": 1.5601760149002075, "step": 20206 }, { "epoch": 1.592983839180134, "grad_norm": 0.5344237685203552, "learning_rate": 0.0005, "loss": 1.5839723348617554, "step": 20207 }, { "epoch": 1.593062672447773, "grad_norm": 0.516196608543396, "learning_rate": 0.0005, "loss": 1.5230135917663574, "step": 20208 }, { "epoch": 1.593141505715412, "grad_norm": 0.5197098851203918, "learning_rate": 0.0005, "loss": 1.345551609992981, "step": 20209 }, { "epoch": 1.5932203389830508, "grad_norm": 0.5539714097976685, "learning_rate": 0.0005, "loss": 1.6487820148468018, "step": 20210 }, { "epoch": 1.5932991722506897, "grad_norm": 0.7599003314971924, "learning_rate": 0.0005, "loss": 1.4045354127883911, "step": 20211 }, { "epoch": 1.5933780055183289, "grad_norm": 0.5959638953208923, "learning_rate": 0.0005, "loss": 1.676145076751709, "step": 20212 }, { "epoch": 1.5934568387859676, "grad_norm": 0.5126998424530029, "learning_rate": 0.0005, "loss": 1.4228551387786865, "step": 20213 }, { "epoch": 1.5935356720536067, "grad_norm": 0.43781498074531555, "learning_rate": 0.0005, "loss": 1.545196771621704, "step": 20214 }, { "epoch": 1.5936145053212456, "grad_norm": 0.6106997132301331, "learning_rate": 0.0005, "loss": 1.5454820394515991, "step": 20215 }, { "epoch": 1.5936933385888845, "grad_norm": 0.5300567746162415, "learning_rate": 0.0005, "loss": 1.6829006671905518, "step": 20216 }, { "epoch": 1.5937721718565234, "grad_norm": 0.5473849177360535, "learning_rate": 0.0005, "loss": 1.5184208154678345, "step": 20217 }, { "epoch": 1.5938510051241623, "grad_norm": 0.5666316747665405, "learning_rate": 0.0005, "loss": 1.568267583847046, "step": 20218 }, { "epoch": 1.5939298383918015, "grad_norm": 0.5432575941085815, "learning_rate": 0.0005, "loss": 1.5584895610809326, "step": 20219 }, { "epoch": 1.5940086716594402, "grad_norm": 0.5144845247268677, "learning_rate": 0.0005, "loss": 1.67486572265625, "step": 20220 }, { "epoch": 1.5940875049270793, "grad_norm": 0.5162945985794067, "learning_rate": 0.0005, "loss": 1.3077545166015625, "step": 20221 }, { "epoch": 1.5941663381947182, "grad_norm": 0.48338285088539124, "learning_rate": 0.0005, "loss": 1.3911993503570557, "step": 20222 }, { "epoch": 1.5942451714623571, "grad_norm": 0.5631784200668335, "learning_rate": 0.0005, "loss": 1.37461519241333, "step": 20223 }, { "epoch": 1.594324004729996, "grad_norm": 0.5772686004638672, "learning_rate": 0.0005, "loss": 1.566323161125183, "step": 20224 }, { "epoch": 1.594402837997635, "grad_norm": 0.8876932859420776, "learning_rate": 0.0005, "loss": 1.6070730686187744, "step": 20225 }, { "epoch": 1.594481671265274, "grad_norm": 0.5081992745399475, "learning_rate": 0.0005, "loss": 1.5864169597625732, "step": 20226 }, { "epoch": 1.5945605045329128, "grad_norm": 0.5253522992134094, "learning_rate": 0.0005, "loss": 1.6117219924926758, "step": 20227 }, { "epoch": 1.594639337800552, "grad_norm": 0.45340970158576965, "learning_rate": 0.0005, "loss": 1.505492925643921, "step": 20228 }, { "epoch": 1.5947181710681908, "grad_norm": 0.7094890475273132, "learning_rate": 0.0005, "loss": 1.6400505304336548, "step": 20229 }, { "epoch": 1.5947970043358297, "grad_norm": 0.466526597738266, "learning_rate": 0.0005, "loss": 1.5448410511016846, "step": 20230 }, { "epoch": 1.5948758376034686, "grad_norm": 0.5082319974899292, "learning_rate": 0.0005, "loss": 1.4997286796569824, "step": 20231 }, { "epoch": 1.5949546708711075, "grad_norm": 0.7607429027557373, "learning_rate": 0.0005, "loss": 1.5324304103851318, "step": 20232 }, { "epoch": 1.5950335041387467, "grad_norm": 0.7132608890533447, "learning_rate": 0.0005, "loss": 1.5166637897491455, "step": 20233 }, { "epoch": 1.5951123374063854, "grad_norm": 0.5014457106590271, "learning_rate": 0.0005, "loss": 1.4745831489562988, "step": 20234 }, { "epoch": 1.5951911706740245, "grad_norm": 0.5155643224716187, "learning_rate": 0.0005, "loss": 1.469435214996338, "step": 20235 }, { "epoch": 1.5952700039416634, "grad_norm": 0.4936375319957733, "learning_rate": 0.0005, "loss": 1.5169713497161865, "step": 20236 }, { "epoch": 1.5953488372093023, "grad_norm": 0.6174567937850952, "learning_rate": 0.0005, "loss": 1.7008788585662842, "step": 20237 }, { "epoch": 1.5954276704769412, "grad_norm": 0.48364296555519104, "learning_rate": 0.0005, "loss": 1.5027220249176025, "step": 20238 }, { "epoch": 1.5955065037445801, "grad_norm": 0.48215222358703613, "learning_rate": 0.0005, "loss": 1.5402085781097412, "step": 20239 }, { "epoch": 1.5955853370122193, "grad_norm": 0.6319337487220764, "learning_rate": 0.0005, "loss": 1.643497347831726, "step": 20240 }, { "epoch": 1.595664170279858, "grad_norm": 0.585131049156189, "learning_rate": 0.0005, "loss": 1.6440293788909912, "step": 20241 }, { "epoch": 1.595743003547497, "grad_norm": 0.6104868054389954, "learning_rate": 0.0005, "loss": 1.5862007141113281, "step": 20242 }, { "epoch": 1.595821836815136, "grad_norm": 0.5917229056358337, "learning_rate": 0.0005, "loss": 1.5303484201431274, "step": 20243 }, { "epoch": 1.595900670082775, "grad_norm": 0.6503164172172546, "learning_rate": 0.0005, "loss": 1.5200650691986084, "step": 20244 }, { "epoch": 1.5959795033504138, "grad_norm": 0.5083632469177246, "learning_rate": 0.0005, "loss": 1.4704976081848145, "step": 20245 }, { "epoch": 1.5960583366180527, "grad_norm": 0.6480758190155029, "learning_rate": 0.0005, "loss": 1.511312484741211, "step": 20246 }, { "epoch": 1.5961371698856919, "grad_norm": 0.775947630405426, "learning_rate": 0.0005, "loss": 1.4932702779769897, "step": 20247 }, { "epoch": 1.5962160031533306, "grad_norm": 0.48989737033843994, "learning_rate": 0.0005, "loss": 1.3130595684051514, "step": 20248 }, { "epoch": 1.5962948364209697, "grad_norm": 0.5983516573905945, "learning_rate": 0.0005, "loss": 1.5538573265075684, "step": 20249 }, { "epoch": 1.5963736696886086, "grad_norm": 0.6026800274848938, "learning_rate": 0.0005, "loss": 1.6064767837524414, "step": 20250 }, { "epoch": 1.5964525029562475, "grad_norm": 0.6747030019760132, "learning_rate": 0.0005, "loss": 1.5737690925598145, "step": 20251 }, { "epoch": 1.5965313362238864, "grad_norm": 0.5605964660644531, "learning_rate": 0.0005, "loss": 1.5994360446929932, "step": 20252 }, { "epoch": 1.5966101694915253, "grad_norm": 0.45241671800613403, "learning_rate": 0.0005, "loss": 1.498701572418213, "step": 20253 }, { "epoch": 1.5966890027591645, "grad_norm": 0.4738037884235382, "learning_rate": 0.0005, "loss": 1.4511505365371704, "step": 20254 }, { "epoch": 1.5967678360268032, "grad_norm": 0.5600130558013916, "learning_rate": 0.0005, "loss": 1.4344828128814697, "step": 20255 }, { "epoch": 1.5968466692944423, "grad_norm": 0.6828835606575012, "learning_rate": 0.0005, "loss": 1.4948137998580933, "step": 20256 }, { "epoch": 1.5969255025620812, "grad_norm": 0.5731354355812073, "learning_rate": 0.0005, "loss": 1.5339455604553223, "step": 20257 }, { "epoch": 1.5970043358297201, "grad_norm": 0.7127403020858765, "learning_rate": 0.0005, "loss": 1.418893575668335, "step": 20258 }, { "epoch": 1.597083169097359, "grad_norm": 0.5709318518638611, "learning_rate": 0.0005, "loss": 1.5932643413543701, "step": 20259 }, { "epoch": 1.597162002364998, "grad_norm": 0.5134925246238708, "learning_rate": 0.0005, "loss": 1.4939377307891846, "step": 20260 }, { "epoch": 1.597240835632637, "grad_norm": 0.6270905137062073, "learning_rate": 0.0005, "loss": 1.7762773036956787, "step": 20261 }, { "epoch": 1.5973196689002758, "grad_norm": 0.8725540041923523, "learning_rate": 0.0005, "loss": 1.6191363334655762, "step": 20262 }, { "epoch": 1.597398502167915, "grad_norm": 0.5293430685997009, "learning_rate": 0.0005, "loss": 1.4756430387496948, "step": 20263 }, { "epoch": 1.5974773354355538, "grad_norm": 0.5411155223846436, "learning_rate": 0.0005, "loss": 1.6276819705963135, "step": 20264 }, { "epoch": 1.5975561687031927, "grad_norm": 0.5373553037643433, "learning_rate": 0.0005, "loss": 1.5985685586929321, "step": 20265 }, { "epoch": 1.5976350019708316, "grad_norm": 0.5993455648422241, "learning_rate": 0.0005, "loss": 1.4091334342956543, "step": 20266 }, { "epoch": 1.5977138352384705, "grad_norm": 0.5630147457122803, "learning_rate": 0.0005, "loss": 1.5603904724121094, "step": 20267 }, { "epoch": 1.5977926685061097, "grad_norm": 0.5160983204841614, "learning_rate": 0.0005, "loss": 1.5980370044708252, "step": 20268 }, { "epoch": 1.5978715017737484, "grad_norm": 0.8652966022491455, "learning_rate": 0.0005, "loss": 1.6588594913482666, "step": 20269 }, { "epoch": 1.5979503350413875, "grad_norm": 0.418407678604126, "learning_rate": 0.0005, "loss": 1.3775277137756348, "step": 20270 }, { "epoch": 1.5980291683090264, "grad_norm": 0.7514693140983582, "learning_rate": 0.0005, "loss": 1.4979817867279053, "step": 20271 }, { "epoch": 1.5981080015766653, "grad_norm": 0.6660211682319641, "learning_rate": 0.0005, "loss": 1.5397956371307373, "step": 20272 }, { "epoch": 1.5981868348443045, "grad_norm": 0.6284776926040649, "learning_rate": 0.0005, "loss": 1.40921151638031, "step": 20273 }, { "epoch": 1.5982656681119431, "grad_norm": 0.8100696206092834, "learning_rate": 0.0005, "loss": 1.6177952289581299, "step": 20274 }, { "epoch": 1.5983445013795823, "grad_norm": 0.6138513088226318, "learning_rate": 0.0005, "loss": 1.7491483688354492, "step": 20275 }, { "epoch": 1.598423334647221, "grad_norm": 0.5569469332695007, "learning_rate": 0.0005, "loss": 1.3153572082519531, "step": 20276 }, { "epoch": 1.59850216791486, "grad_norm": 0.48597925901412964, "learning_rate": 0.0005, "loss": 1.5595465898513794, "step": 20277 }, { "epoch": 1.598581001182499, "grad_norm": 0.6430094242095947, "learning_rate": 0.0005, "loss": 1.6683281660079956, "step": 20278 }, { "epoch": 1.598659834450138, "grad_norm": 0.6728565692901611, "learning_rate": 0.0005, "loss": 1.5618674755096436, "step": 20279 }, { "epoch": 1.598738667717777, "grad_norm": 0.4912533760070801, "learning_rate": 0.0005, "loss": 1.4958305358886719, "step": 20280 }, { "epoch": 1.5988175009854158, "grad_norm": 0.5612702965736389, "learning_rate": 0.0005, "loss": 1.4942502975463867, "step": 20281 }, { "epoch": 1.5988963342530549, "grad_norm": 0.5087068676948547, "learning_rate": 0.0005, "loss": 1.5670132637023926, "step": 20282 }, { "epoch": 1.5989751675206936, "grad_norm": 0.5805888175964355, "learning_rate": 0.0005, "loss": 1.3939852714538574, "step": 20283 }, { "epoch": 1.5990540007883327, "grad_norm": 0.5808336138725281, "learning_rate": 0.0005, "loss": 1.594534158706665, "step": 20284 }, { "epoch": 1.5991328340559716, "grad_norm": 0.5559574961662292, "learning_rate": 0.0005, "loss": 1.5637378692626953, "step": 20285 }, { "epoch": 1.5992116673236105, "grad_norm": 0.5523954033851624, "learning_rate": 0.0005, "loss": 1.611214518547058, "step": 20286 }, { "epoch": 1.5992905005912497, "grad_norm": 0.4836892783641815, "learning_rate": 0.0005, "loss": 1.5156075954437256, "step": 20287 }, { "epoch": 1.5993693338588884, "grad_norm": 0.5562049746513367, "learning_rate": 0.0005, "loss": 1.320557951927185, "step": 20288 }, { "epoch": 1.5994481671265275, "grad_norm": 0.4594704806804657, "learning_rate": 0.0005, "loss": 1.315596103668213, "step": 20289 }, { "epoch": 1.5995270003941662, "grad_norm": 0.5643720030784607, "learning_rate": 0.0005, "loss": 1.429114580154419, "step": 20290 }, { "epoch": 1.5996058336618053, "grad_norm": 0.47643110156059265, "learning_rate": 0.0005, "loss": 1.333624243736267, "step": 20291 }, { "epoch": 1.5996846669294442, "grad_norm": 0.4904854893684387, "learning_rate": 0.0005, "loss": 1.543224573135376, "step": 20292 }, { "epoch": 1.5997635001970831, "grad_norm": 0.5684260129928589, "learning_rate": 0.0005, "loss": 1.5482006072998047, "step": 20293 }, { "epoch": 1.5998423334647223, "grad_norm": 0.6400415301322937, "learning_rate": 0.0005, "loss": 1.3930703401565552, "step": 20294 }, { "epoch": 1.599921166732361, "grad_norm": 0.503038763999939, "learning_rate": 0.0005, "loss": 1.5684176683425903, "step": 20295 }, { "epoch": 1.6, "grad_norm": 0.5464173555374146, "learning_rate": 0.0005, "loss": 1.4360755681991577, "step": 20296 }, { "epoch": 1.6000788332676388, "grad_norm": 0.5471649169921875, "learning_rate": 0.0005, "loss": 1.6043860912322998, "step": 20297 }, { "epoch": 1.600157666535278, "grad_norm": 0.46664658188819885, "learning_rate": 0.0005, "loss": 1.3941893577575684, "step": 20298 }, { "epoch": 1.6002364998029168, "grad_norm": 0.5945401191711426, "learning_rate": 0.0005, "loss": 1.5483349561691284, "step": 20299 }, { "epoch": 1.6003153330705557, "grad_norm": 0.4998900890350342, "learning_rate": 0.0005, "loss": 1.3213849067687988, "step": 20300 }, { "epoch": 1.6003941663381949, "grad_norm": 0.609220027923584, "learning_rate": 0.0005, "loss": 1.606028437614441, "step": 20301 }, { "epoch": 1.6004729996058336, "grad_norm": 0.5205419063568115, "learning_rate": 0.0005, "loss": 1.4185914993286133, "step": 20302 }, { "epoch": 1.6005518328734727, "grad_norm": 0.6448402404785156, "learning_rate": 0.0005, "loss": 1.509916067123413, "step": 20303 }, { "epoch": 1.6006306661411116, "grad_norm": 0.5264423489570618, "learning_rate": 0.0005, "loss": 1.6114948987960815, "step": 20304 }, { "epoch": 1.6007094994087505, "grad_norm": 0.49410101771354675, "learning_rate": 0.0005, "loss": 1.5606989860534668, "step": 20305 }, { "epoch": 1.6007883326763894, "grad_norm": 0.6961990594863892, "learning_rate": 0.0005, "loss": 1.5583899021148682, "step": 20306 }, { "epoch": 1.6008671659440283, "grad_norm": 0.45395416021347046, "learning_rate": 0.0005, "loss": 1.478236436843872, "step": 20307 }, { "epoch": 1.6009459992116675, "grad_norm": 0.959571361541748, "learning_rate": 0.0005, "loss": 1.596665620803833, "step": 20308 }, { "epoch": 1.6010248324793062, "grad_norm": 0.5569546222686768, "learning_rate": 0.0005, "loss": 1.5889875888824463, "step": 20309 }, { "epoch": 1.6011036657469453, "grad_norm": 0.6032266020774841, "learning_rate": 0.0005, "loss": 1.3477973937988281, "step": 20310 }, { "epoch": 1.6011824990145842, "grad_norm": 0.6134401559829712, "learning_rate": 0.0005, "loss": 1.5718096494674683, "step": 20311 }, { "epoch": 1.6012613322822231, "grad_norm": 0.6132933497428894, "learning_rate": 0.0005, "loss": 1.4760854244232178, "step": 20312 }, { "epoch": 1.601340165549862, "grad_norm": 0.5853224992752075, "learning_rate": 0.0005, "loss": 1.5767107009887695, "step": 20313 }, { "epoch": 1.601418998817501, "grad_norm": 0.4783216118812561, "learning_rate": 0.0005, "loss": 1.37588632106781, "step": 20314 }, { "epoch": 1.60149783208514, "grad_norm": 0.6481124758720398, "learning_rate": 0.0005, "loss": 1.54624605178833, "step": 20315 }, { "epoch": 1.6015766653527788, "grad_norm": 0.5353172421455383, "learning_rate": 0.0005, "loss": 1.7306137084960938, "step": 20316 }, { "epoch": 1.601655498620418, "grad_norm": 0.5331284999847412, "learning_rate": 0.0005, "loss": 1.5678030252456665, "step": 20317 }, { "epoch": 1.6017343318880568, "grad_norm": 0.48241275548934937, "learning_rate": 0.0005, "loss": 1.5059068202972412, "step": 20318 }, { "epoch": 1.6018131651556957, "grad_norm": 0.5453187823295593, "learning_rate": 0.0005, "loss": 1.7237387895584106, "step": 20319 }, { "epoch": 1.6018919984233346, "grad_norm": 0.5436475276947021, "learning_rate": 0.0005, "loss": 1.565974473953247, "step": 20320 }, { "epoch": 1.6019708316909735, "grad_norm": 0.5154250860214233, "learning_rate": 0.0005, "loss": 1.4838054180145264, "step": 20321 }, { "epoch": 1.6020496649586127, "grad_norm": 0.49989721179008484, "learning_rate": 0.0005, "loss": 1.4063501358032227, "step": 20322 }, { "epoch": 1.6021284982262514, "grad_norm": 0.5373973846435547, "learning_rate": 0.0005, "loss": 1.6027517318725586, "step": 20323 }, { "epoch": 1.6022073314938905, "grad_norm": 0.5127970576286316, "learning_rate": 0.0005, "loss": 1.5582454204559326, "step": 20324 }, { "epoch": 1.6022861647615294, "grad_norm": 0.5216987729072571, "learning_rate": 0.0005, "loss": 1.632378101348877, "step": 20325 }, { "epoch": 1.6023649980291683, "grad_norm": 0.5955527424812317, "learning_rate": 0.0005, "loss": 1.511495590209961, "step": 20326 }, { "epoch": 1.6024438312968072, "grad_norm": 0.5196991562843323, "learning_rate": 0.0005, "loss": 1.4221811294555664, "step": 20327 }, { "epoch": 1.6025226645644461, "grad_norm": 0.6505822539329529, "learning_rate": 0.0005, "loss": 1.5929114818572998, "step": 20328 }, { "epoch": 1.6026014978320853, "grad_norm": 0.5295389294624329, "learning_rate": 0.0005, "loss": 1.645833969116211, "step": 20329 }, { "epoch": 1.602680331099724, "grad_norm": 0.645939290523529, "learning_rate": 0.0005, "loss": 1.6649606227874756, "step": 20330 }, { "epoch": 1.602759164367363, "grad_norm": 0.5925757884979248, "learning_rate": 0.0005, "loss": 1.545918345451355, "step": 20331 }, { "epoch": 1.602837997635002, "grad_norm": 0.6158999800682068, "learning_rate": 0.0005, "loss": 1.543971300125122, "step": 20332 }, { "epoch": 1.602916830902641, "grad_norm": 0.46045544743537903, "learning_rate": 0.0005, "loss": 1.3072118759155273, "step": 20333 }, { "epoch": 1.6029956641702798, "grad_norm": 0.5239981412887573, "learning_rate": 0.0005, "loss": 1.4672963619232178, "step": 20334 }, { "epoch": 1.6030744974379187, "grad_norm": 0.3850407898426056, "learning_rate": 0.0005, "loss": 1.335668683052063, "step": 20335 }, { "epoch": 1.6031533307055579, "grad_norm": 0.5071186423301697, "learning_rate": 0.0005, "loss": 1.5588598251342773, "step": 20336 }, { "epoch": 1.6032321639731966, "grad_norm": 0.4774695932865143, "learning_rate": 0.0005, "loss": 1.6567697525024414, "step": 20337 }, { "epoch": 1.6033109972408357, "grad_norm": 0.5151464343070984, "learning_rate": 0.0005, "loss": 1.4142651557922363, "step": 20338 }, { "epoch": 1.6033898305084746, "grad_norm": 0.5989657640457153, "learning_rate": 0.0005, "loss": 1.6436057090759277, "step": 20339 }, { "epoch": 1.6034686637761135, "grad_norm": 0.5740789771080017, "learning_rate": 0.0005, "loss": 1.6059753894805908, "step": 20340 }, { "epoch": 1.6035474970437524, "grad_norm": 0.4847888648509979, "learning_rate": 0.0005, "loss": 1.3477158546447754, "step": 20341 }, { "epoch": 1.6036263303113913, "grad_norm": 0.7129935622215271, "learning_rate": 0.0005, "loss": 1.5643264055252075, "step": 20342 }, { "epoch": 1.6037051635790305, "grad_norm": 0.5583203434944153, "learning_rate": 0.0005, "loss": 1.743905782699585, "step": 20343 }, { "epoch": 1.6037839968466692, "grad_norm": 0.5996180772781372, "learning_rate": 0.0005, "loss": 1.6914153099060059, "step": 20344 }, { "epoch": 1.6038628301143083, "grad_norm": 0.5336808562278748, "learning_rate": 0.0005, "loss": 1.5225167274475098, "step": 20345 }, { "epoch": 1.6039416633819472, "grad_norm": 0.4692632853984833, "learning_rate": 0.0005, "loss": 1.3170578479766846, "step": 20346 }, { "epoch": 1.6040204966495861, "grad_norm": 0.7059935331344604, "learning_rate": 0.0005, "loss": 1.556130290031433, "step": 20347 }, { "epoch": 1.604099329917225, "grad_norm": 0.6213382482528687, "learning_rate": 0.0005, "loss": 1.6236296892166138, "step": 20348 }, { "epoch": 1.604178163184864, "grad_norm": 0.46809157729148865, "learning_rate": 0.0005, "loss": 1.4264867305755615, "step": 20349 }, { "epoch": 1.604256996452503, "grad_norm": 0.5109420418739319, "learning_rate": 0.0005, "loss": 1.5998497009277344, "step": 20350 }, { "epoch": 1.6043358297201418, "grad_norm": 0.47187989950180054, "learning_rate": 0.0005, "loss": 1.4192383289337158, "step": 20351 }, { "epoch": 1.604414662987781, "grad_norm": 0.5106498599052429, "learning_rate": 0.0005, "loss": 1.5256435871124268, "step": 20352 }, { "epoch": 1.6044934962554198, "grad_norm": 0.5498698949813843, "learning_rate": 0.0005, "loss": 1.394706130027771, "step": 20353 }, { "epoch": 1.6045723295230587, "grad_norm": 0.5373678207397461, "learning_rate": 0.0005, "loss": 1.5628318786621094, "step": 20354 }, { "epoch": 1.6046511627906976, "grad_norm": 0.5470842719078064, "learning_rate": 0.0005, "loss": 1.488821029663086, "step": 20355 }, { "epoch": 1.6047299960583365, "grad_norm": 0.4468986690044403, "learning_rate": 0.0005, "loss": 1.3286805152893066, "step": 20356 }, { "epoch": 1.6048088293259757, "grad_norm": 0.4648987352848053, "learning_rate": 0.0005, "loss": 1.3099336624145508, "step": 20357 }, { "epoch": 1.6048876625936144, "grad_norm": 0.5209488868713379, "learning_rate": 0.0005, "loss": 1.517721176147461, "step": 20358 }, { "epoch": 1.6049664958612535, "grad_norm": 0.4680064916610718, "learning_rate": 0.0005, "loss": 1.5325393676757812, "step": 20359 }, { "epoch": 1.6050453291288924, "grad_norm": 0.5126939415931702, "learning_rate": 0.0005, "loss": 1.6455473899841309, "step": 20360 }, { "epoch": 1.6051241623965313, "grad_norm": 0.5020177364349365, "learning_rate": 0.0005, "loss": 1.5441548824310303, "step": 20361 }, { "epoch": 1.6052029956641702, "grad_norm": 0.5496110320091248, "learning_rate": 0.0005, "loss": 1.4510389566421509, "step": 20362 }, { "epoch": 1.6052818289318092, "grad_norm": 0.66142338514328, "learning_rate": 0.0005, "loss": 1.5988597869873047, "step": 20363 }, { "epoch": 1.6053606621994483, "grad_norm": 0.6140632629394531, "learning_rate": 0.0005, "loss": 1.584855556488037, "step": 20364 }, { "epoch": 1.605439495467087, "grad_norm": 0.5960325002670288, "learning_rate": 0.0005, "loss": 1.655490756034851, "step": 20365 }, { "epoch": 1.605518328734726, "grad_norm": 0.7050233483314514, "learning_rate": 0.0005, "loss": 1.6240220069885254, "step": 20366 }, { "epoch": 1.605597162002365, "grad_norm": 0.551548421382904, "learning_rate": 0.0005, "loss": 1.3966388702392578, "step": 20367 }, { "epoch": 1.605675995270004, "grad_norm": 0.5841392278671265, "learning_rate": 0.0005, "loss": 1.6430280208587646, "step": 20368 }, { "epoch": 1.6057548285376428, "grad_norm": 0.49314334988594055, "learning_rate": 0.0005, "loss": 1.3585789203643799, "step": 20369 }, { "epoch": 1.6058336618052818, "grad_norm": 0.5421125292778015, "learning_rate": 0.0005, "loss": 1.687301754951477, "step": 20370 }, { "epoch": 1.6059124950729209, "grad_norm": 0.5383904576301575, "learning_rate": 0.0005, "loss": 1.5703750848770142, "step": 20371 }, { "epoch": 1.6059913283405596, "grad_norm": 0.5616369843482971, "learning_rate": 0.0005, "loss": 1.5267592668533325, "step": 20372 }, { "epoch": 1.6060701616081987, "grad_norm": 0.523323655128479, "learning_rate": 0.0005, "loss": 1.4482104778289795, "step": 20373 }, { "epoch": 1.6061489948758376, "grad_norm": 0.476111501455307, "learning_rate": 0.0005, "loss": 1.498616099357605, "step": 20374 }, { "epoch": 1.6062278281434765, "grad_norm": 0.5422998070716858, "learning_rate": 0.0005, "loss": 1.5343215465545654, "step": 20375 }, { "epoch": 1.6063066614111154, "grad_norm": 0.5417318940162659, "learning_rate": 0.0005, "loss": 1.6029891967773438, "step": 20376 }, { "epoch": 1.6063854946787544, "grad_norm": 0.45214343070983887, "learning_rate": 0.0005, "loss": 1.5223753452301025, "step": 20377 }, { "epoch": 1.6064643279463935, "grad_norm": 0.4259418845176697, "learning_rate": 0.0005, "loss": 1.3701459169387817, "step": 20378 }, { "epoch": 1.6065431612140322, "grad_norm": 0.6647807955741882, "learning_rate": 0.0005, "loss": 1.528245449066162, "step": 20379 }, { "epoch": 1.6066219944816713, "grad_norm": 0.5342736840248108, "learning_rate": 0.0005, "loss": 1.5456733703613281, "step": 20380 }, { "epoch": 1.6067008277493102, "grad_norm": 0.49726441502571106, "learning_rate": 0.0005, "loss": 1.5510540008544922, "step": 20381 }, { "epoch": 1.6067796610169491, "grad_norm": 0.5346857309341431, "learning_rate": 0.0005, "loss": 1.6354140043258667, "step": 20382 }, { "epoch": 1.606858494284588, "grad_norm": 0.45000723004341125, "learning_rate": 0.0005, "loss": 1.472417950630188, "step": 20383 }, { "epoch": 1.606937327552227, "grad_norm": 0.5242339968681335, "learning_rate": 0.0005, "loss": 1.4094831943511963, "step": 20384 }, { "epoch": 1.607016160819866, "grad_norm": 0.7111644744873047, "learning_rate": 0.0005, "loss": 1.6165976524353027, "step": 20385 }, { "epoch": 1.6070949940875048, "grad_norm": 0.6002621650695801, "learning_rate": 0.0005, "loss": 1.3468928337097168, "step": 20386 }, { "epoch": 1.607173827355144, "grad_norm": 0.661840558052063, "learning_rate": 0.0005, "loss": 1.480351209640503, "step": 20387 }, { "epoch": 1.6072526606227828, "grad_norm": 0.5965357422828674, "learning_rate": 0.0005, "loss": 1.6256639957427979, "step": 20388 }, { "epoch": 1.6073314938904217, "grad_norm": 0.4427889585494995, "learning_rate": 0.0005, "loss": 1.4707696437835693, "step": 20389 }, { "epoch": 1.6074103271580606, "grad_norm": 0.548439085483551, "learning_rate": 0.0005, "loss": 1.581063985824585, "step": 20390 }, { "epoch": 1.6074891604256996, "grad_norm": 0.6527343988418579, "learning_rate": 0.0005, "loss": 1.5623486042022705, "step": 20391 }, { "epoch": 1.6075679936933387, "grad_norm": 0.49684274196624756, "learning_rate": 0.0005, "loss": 1.3888905048370361, "step": 20392 }, { "epoch": 1.6076468269609774, "grad_norm": 0.6714881658554077, "learning_rate": 0.0005, "loss": 1.4481905698776245, "step": 20393 }, { "epoch": 1.6077256602286165, "grad_norm": 0.6868915557861328, "learning_rate": 0.0005, "loss": 1.6253437995910645, "step": 20394 }, { "epoch": 1.6078044934962554, "grad_norm": 0.5148478746414185, "learning_rate": 0.0005, "loss": 1.4673845767974854, "step": 20395 }, { "epoch": 1.6078833267638943, "grad_norm": 0.8949520587921143, "learning_rate": 0.0005, "loss": 1.5703595876693726, "step": 20396 }, { "epoch": 1.6079621600315335, "grad_norm": 0.6116507053375244, "learning_rate": 0.0005, "loss": 1.5530999898910522, "step": 20397 }, { "epoch": 1.6080409932991722, "grad_norm": 0.43290022015571594, "learning_rate": 0.0005, "loss": 1.4654247760772705, "step": 20398 }, { "epoch": 1.6081198265668113, "grad_norm": 0.5688941478729248, "learning_rate": 0.0005, "loss": 1.50961434841156, "step": 20399 }, { "epoch": 1.60819865983445, "grad_norm": 0.6069545745849609, "learning_rate": 0.0005, "loss": 1.6057208776474, "step": 20400 }, { "epoch": 1.6082774931020891, "grad_norm": 0.4956449270248413, "learning_rate": 0.0005, "loss": 1.339266300201416, "step": 20401 }, { "epoch": 1.608356326369728, "grad_norm": 0.6099702715873718, "learning_rate": 0.0005, "loss": 1.614506721496582, "step": 20402 }, { "epoch": 1.608435159637367, "grad_norm": 0.5434842109680176, "learning_rate": 0.0005, "loss": 1.5735199451446533, "step": 20403 }, { "epoch": 1.608513992905006, "grad_norm": 0.5724125504493713, "learning_rate": 0.0005, "loss": 1.7302687168121338, "step": 20404 }, { "epoch": 1.6085928261726448, "grad_norm": 0.5157455205917358, "learning_rate": 0.0005, "loss": 1.5547850131988525, "step": 20405 }, { "epoch": 1.608671659440284, "grad_norm": 0.622747004032135, "learning_rate": 0.0005, "loss": 1.5162700414657593, "step": 20406 }, { "epoch": 1.6087504927079226, "grad_norm": 0.49866244196891785, "learning_rate": 0.0005, "loss": 1.4822994470596313, "step": 20407 }, { "epoch": 1.6088293259755617, "grad_norm": 0.6191032528877258, "learning_rate": 0.0005, "loss": 1.8223798274993896, "step": 20408 }, { "epoch": 1.6089081592432006, "grad_norm": 0.46869519352912903, "learning_rate": 0.0005, "loss": 1.4144392013549805, "step": 20409 }, { "epoch": 1.6089869925108395, "grad_norm": 0.5437976121902466, "learning_rate": 0.0005, "loss": 1.4935885667800903, "step": 20410 }, { "epoch": 1.6090658257784787, "grad_norm": 0.5275483131408691, "learning_rate": 0.0005, "loss": 1.5446600914001465, "step": 20411 }, { "epoch": 1.6091446590461174, "grad_norm": 0.5847412943840027, "learning_rate": 0.0005, "loss": 1.7285635471343994, "step": 20412 }, { "epoch": 1.6092234923137565, "grad_norm": 0.6801449656486511, "learning_rate": 0.0005, "loss": 1.6474404335021973, "step": 20413 }, { "epoch": 1.6093023255813952, "grad_norm": 0.6031590700149536, "learning_rate": 0.0005, "loss": 1.7189521789550781, "step": 20414 }, { "epoch": 1.6093811588490343, "grad_norm": 0.4694622755050659, "learning_rate": 0.0005, "loss": 1.3919610977172852, "step": 20415 }, { "epoch": 1.6094599921166732, "grad_norm": 0.4850090742111206, "learning_rate": 0.0005, "loss": 1.3821756839752197, "step": 20416 }, { "epoch": 1.6095388253843121, "grad_norm": 0.5247910022735596, "learning_rate": 0.0005, "loss": 1.5152612924575806, "step": 20417 }, { "epoch": 1.6096176586519513, "grad_norm": 0.650809645652771, "learning_rate": 0.0005, "loss": 1.6518864631652832, "step": 20418 }, { "epoch": 1.60969649191959, "grad_norm": 0.6049489378929138, "learning_rate": 0.0005, "loss": 1.597702145576477, "step": 20419 }, { "epoch": 1.609775325187229, "grad_norm": 0.5532341599464417, "learning_rate": 0.0005, "loss": 1.3392367362976074, "step": 20420 }, { "epoch": 1.6098541584548678, "grad_norm": 0.5034478306770325, "learning_rate": 0.0005, "loss": 1.4178324937820435, "step": 20421 }, { "epoch": 1.609932991722507, "grad_norm": 0.6352291107177734, "learning_rate": 0.0005, "loss": 1.8241462707519531, "step": 20422 }, { "epoch": 1.6100118249901458, "grad_norm": 0.6961014270782471, "learning_rate": 0.0005, "loss": 1.5558452606201172, "step": 20423 }, { "epoch": 1.6100906582577847, "grad_norm": 0.48760730028152466, "learning_rate": 0.0005, "loss": 1.4085969924926758, "step": 20424 }, { "epoch": 1.6101694915254239, "grad_norm": 0.48918479681015015, "learning_rate": 0.0005, "loss": 1.590768814086914, "step": 20425 }, { "epoch": 1.6102483247930626, "grad_norm": 0.46471303701400757, "learning_rate": 0.0005, "loss": 1.2948607206344604, "step": 20426 }, { "epoch": 1.6103271580607017, "grad_norm": 0.5294415354728699, "learning_rate": 0.0005, "loss": 1.6195507049560547, "step": 20427 }, { "epoch": 1.6104059913283406, "grad_norm": 0.4549155831336975, "learning_rate": 0.0005, "loss": 1.6596949100494385, "step": 20428 }, { "epoch": 1.6104848245959795, "grad_norm": 0.45535728335380554, "learning_rate": 0.0005, "loss": 1.4036206007003784, "step": 20429 }, { "epoch": 1.6105636578636184, "grad_norm": 0.5251830816268921, "learning_rate": 0.0005, "loss": 1.6497070789337158, "step": 20430 }, { "epoch": 1.6106424911312573, "grad_norm": 0.5702690482139587, "learning_rate": 0.0005, "loss": 1.5762319564819336, "step": 20431 }, { "epoch": 1.6107213243988965, "grad_norm": 0.5279005765914917, "learning_rate": 0.0005, "loss": 1.4515342712402344, "step": 20432 }, { "epoch": 1.6108001576665352, "grad_norm": 0.5542717576026917, "learning_rate": 0.0005, "loss": 1.4965441226959229, "step": 20433 }, { "epoch": 1.6108789909341743, "grad_norm": 0.9242464303970337, "learning_rate": 0.0005, "loss": 1.5710773468017578, "step": 20434 }, { "epoch": 1.6109578242018132, "grad_norm": 0.5159815549850464, "learning_rate": 0.0005, "loss": 1.539984941482544, "step": 20435 }, { "epoch": 1.6110366574694521, "grad_norm": 0.5348140597343445, "learning_rate": 0.0005, "loss": 1.48854660987854, "step": 20436 }, { "epoch": 1.611115490737091, "grad_norm": 0.557345986366272, "learning_rate": 0.0005, "loss": 1.659612774848938, "step": 20437 }, { "epoch": 1.61119432400473, "grad_norm": 0.6187362670898438, "learning_rate": 0.0005, "loss": 1.5455255508422852, "step": 20438 }, { "epoch": 1.611273157272369, "grad_norm": 0.570055365562439, "learning_rate": 0.0005, "loss": 1.6016672849655151, "step": 20439 }, { "epoch": 1.6113519905400078, "grad_norm": 0.530120313167572, "learning_rate": 0.0005, "loss": 1.449720859527588, "step": 20440 }, { "epoch": 1.611430823807647, "grad_norm": 0.4974634647369385, "learning_rate": 0.0005, "loss": 1.5237021446228027, "step": 20441 }, { "epoch": 1.6115096570752858, "grad_norm": 0.5021990537643433, "learning_rate": 0.0005, "loss": 1.4868650436401367, "step": 20442 }, { "epoch": 1.6115884903429247, "grad_norm": 0.47023487091064453, "learning_rate": 0.0005, "loss": 1.534895658493042, "step": 20443 }, { "epoch": 1.6116673236105636, "grad_norm": 0.4224652349948883, "learning_rate": 0.0005, "loss": 1.4163894653320312, "step": 20444 }, { "epoch": 1.6117461568782026, "grad_norm": 0.4593052566051483, "learning_rate": 0.0005, "loss": 1.4819360971450806, "step": 20445 }, { "epoch": 1.6118249901458417, "grad_norm": 0.5751845240592957, "learning_rate": 0.0005, "loss": 1.6899349689483643, "step": 20446 }, { "epoch": 1.6119038234134804, "grad_norm": 0.5944148302078247, "learning_rate": 0.0005, "loss": 1.6296298503875732, "step": 20447 }, { "epoch": 1.6119826566811195, "grad_norm": 0.5662528872489929, "learning_rate": 0.0005, "loss": 1.6747665405273438, "step": 20448 }, { "epoch": 1.6120614899487584, "grad_norm": 0.41773831844329834, "learning_rate": 0.0005, "loss": 1.5113447904586792, "step": 20449 }, { "epoch": 1.6121403232163973, "grad_norm": 0.4627283215522766, "learning_rate": 0.0005, "loss": 1.5921201705932617, "step": 20450 }, { "epoch": 1.6122191564840362, "grad_norm": 0.6411904692649841, "learning_rate": 0.0005, "loss": 1.9194837808609009, "step": 20451 }, { "epoch": 1.6122979897516752, "grad_norm": 0.410435289144516, "learning_rate": 0.0005, "loss": 1.2481679916381836, "step": 20452 }, { "epoch": 1.6123768230193143, "grad_norm": 0.46839749813079834, "learning_rate": 0.0005, "loss": 1.415310025215149, "step": 20453 }, { "epoch": 1.612455656286953, "grad_norm": 0.5112153887748718, "learning_rate": 0.0005, "loss": 1.5398797988891602, "step": 20454 }, { "epoch": 1.612534489554592, "grad_norm": 0.5658907890319824, "learning_rate": 0.0005, "loss": 1.4584081172943115, "step": 20455 }, { "epoch": 1.612613322822231, "grad_norm": 0.5636485815048218, "learning_rate": 0.0005, "loss": 1.5771493911743164, "step": 20456 }, { "epoch": 1.61269215608987, "grad_norm": 0.5912833213806152, "learning_rate": 0.0005, "loss": 1.7602627277374268, "step": 20457 }, { "epoch": 1.6127709893575088, "grad_norm": 0.5275498032569885, "learning_rate": 0.0005, "loss": 1.463514804840088, "step": 20458 }, { "epoch": 1.6128498226251478, "grad_norm": 0.5372166037559509, "learning_rate": 0.0005, "loss": 1.473167896270752, "step": 20459 }, { "epoch": 1.6129286558927869, "grad_norm": 0.6274397969245911, "learning_rate": 0.0005, "loss": 1.5690889358520508, "step": 20460 }, { "epoch": 1.6130074891604256, "grad_norm": 0.6335968971252441, "learning_rate": 0.0005, "loss": 1.8193397521972656, "step": 20461 }, { "epoch": 1.6130863224280647, "grad_norm": 0.457718163728714, "learning_rate": 0.0005, "loss": 1.6088495254516602, "step": 20462 }, { "epoch": 1.6131651556957036, "grad_norm": 0.44666388630867004, "learning_rate": 0.0005, "loss": 1.5546209812164307, "step": 20463 }, { "epoch": 1.6132439889633425, "grad_norm": 0.5340826511383057, "learning_rate": 0.0005, "loss": 1.641500473022461, "step": 20464 }, { "epoch": 1.6133228222309814, "grad_norm": 0.6070077419281006, "learning_rate": 0.0005, "loss": 1.5646761655807495, "step": 20465 }, { "epoch": 1.6134016554986204, "grad_norm": 0.5703017711639404, "learning_rate": 0.0005, "loss": 1.7093454599380493, "step": 20466 }, { "epoch": 1.6134804887662595, "grad_norm": 0.4412250518798828, "learning_rate": 0.0005, "loss": 1.33376145362854, "step": 20467 }, { "epoch": 1.6135593220338982, "grad_norm": 0.5118864178657532, "learning_rate": 0.0005, "loss": 1.5916589498519897, "step": 20468 }, { "epoch": 1.6136381553015373, "grad_norm": 0.5321909785270691, "learning_rate": 0.0005, "loss": 1.4949400424957275, "step": 20469 }, { "epoch": 1.6137169885691762, "grad_norm": 0.49530813097953796, "learning_rate": 0.0005, "loss": 1.5339245796203613, "step": 20470 }, { "epoch": 1.6137958218368151, "grad_norm": 0.5499112010002136, "learning_rate": 0.0005, "loss": 1.6643056869506836, "step": 20471 }, { "epoch": 1.613874655104454, "grad_norm": 0.6575942635536194, "learning_rate": 0.0005, "loss": 1.5954537391662598, "step": 20472 }, { "epoch": 1.613953488372093, "grad_norm": 0.42367708683013916, "learning_rate": 0.0005, "loss": 1.398618221282959, "step": 20473 }, { "epoch": 1.614032321639732, "grad_norm": 0.5914220213890076, "learning_rate": 0.0005, "loss": 1.3883484601974487, "step": 20474 }, { "epoch": 1.6141111549073708, "grad_norm": 0.7846211194992065, "learning_rate": 0.0005, "loss": 1.6617140769958496, "step": 20475 }, { "epoch": 1.61418998817501, "grad_norm": 0.611412525177002, "learning_rate": 0.0005, "loss": 1.5867117643356323, "step": 20476 }, { "epoch": 1.6142688214426488, "grad_norm": 0.5229613780975342, "learning_rate": 0.0005, "loss": 1.543129324913025, "step": 20477 }, { "epoch": 1.6143476547102877, "grad_norm": 0.5212550759315491, "learning_rate": 0.0005, "loss": 1.5505783557891846, "step": 20478 }, { "epoch": 1.6144264879779266, "grad_norm": 0.45528149604797363, "learning_rate": 0.0005, "loss": 1.558267593383789, "step": 20479 }, { "epoch": 1.6145053212455656, "grad_norm": 0.5904635787010193, "learning_rate": 0.0005, "loss": 1.5483487844467163, "step": 20480 }, { "epoch": 1.6145841545132047, "grad_norm": 0.9534227252006531, "learning_rate": 0.0005, "loss": 1.6192675828933716, "step": 20481 }, { "epoch": 1.6146629877808434, "grad_norm": 0.47067490220069885, "learning_rate": 0.0005, "loss": 1.3474135398864746, "step": 20482 }, { "epoch": 1.6147418210484825, "grad_norm": 0.6053767204284668, "learning_rate": 0.0005, "loss": 1.7751739025115967, "step": 20483 }, { "epoch": 1.6148206543161214, "grad_norm": 0.8353320360183716, "learning_rate": 0.0005, "loss": 1.5504999160766602, "step": 20484 }, { "epoch": 1.6148994875837603, "grad_norm": 0.6181759238243103, "learning_rate": 0.0005, "loss": 1.4970834255218506, "step": 20485 }, { "epoch": 1.6149783208513993, "grad_norm": 0.6389626264572144, "learning_rate": 0.0005, "loss": 1.5796802043914795, "step": 20486 }, { "epoch": 1.6150571541190382, "grad_norm": 0.4492841958999634, "learning_rate": 0.0005, "loss": 1.4241325855255127, "step": 20487 }, { "epoch": 1.6151359873866773, "grad_norm": 0.6593549847602844, "learning_rate": 0.0005, "loss": 1.535016417503357, "step": 20488 }, { "epoch": 1.615214820654316, "grad_norm": 0.5184337496757507, "learning_rate": 0.0005, "loss": 1.284294843673706, "step": 20489 }, { "epoch": 1.6152936539219551, "grad_norm": 0.5236125588417053, "learning_rate": 0.0005, "loss": 1.3786869049072266, "step": 20490 }, { "epoch": 1.615372487189594, "grad_norm": 0.4555380046367645, "learning_rate": 0.0005, "loss": 1.4526922702789307, "step": 20491 }, { "epoch": 1.615451320457233, "grad_norm": 0.5213923454284668, "learning_rate": 0.0005, "loss": 1.3936327695846558, "step": 20492 }, { "epoch": 1.6155301537248719, "grad_norm": 0.5232530832290649, "learning_rate": 0.0005, "loss": 1.6262338161468506, "step": 20493 }, { "epoch": 1.6156089869925108, "grad_norm": 0.7067184448242188, "learning_rate": 0.0005, "loss": 1.5216360092163086, "step": 20494 }, { "epoch": 1.61568782026015, "grad_norm": 0.4874688982963562, "learning_rate": 0.0005, "loss": 1.6170214414596558, "step": 20495 }, { "epoch": 1.6157666535277886, "grad_norm": 0.6706907749176025, "learning_rate": 0.0005, "loss": 1.5784903764724731, "step": 20496 }, { "epoch": 1.6158454867954277, "grad_norm": 0.5173236727714539, "learning_rate": 0.0005, "loss": 1.4352082014083862, "step": 20497 }, { "epoch": 1.6159243200630666, "grad_norm": 0.6686833500862122, "learning_rate": 0.0005, "loss": 1.5821278095245361, "step": 20498 }, { "epoch": 1.6160031533307055, "grad_norm": 0.6191912889480591, "learning_rate": 0.0005, "loss": 1.6416585445404053, "step": 20499 }, { "epoch": 1.6160819865983445, "grad_norm": 0.5210091471672058, "learning_rate": 0.0005, "loss": 1.3456451892852783, "step": 20500 }, { "epoch": 1.6161608198659834, "grad_norm": 0.4423484206199646, "learning_rate": 0.0005, "loss": 1.4276623725891113, "step": 20501 }, { "epoch": 1.6162396531336225, "grad_norm": 0.5255235433578491, "learning_rate": 0.0005, "loss": 1.1996803283691406, "step": 20502 }, { "epoch": 1.6163184864012612, "grad_norm": 0.5489200949668884, "learning_rate": 0.0005, "loss": 1.6033536195755005, "step": 20503 }, { "epoch": 1.6163973196689003, "grad_norm": 0.658805251121521, "learning_rate": 0.0005, "loss": 1.5943573713302612, "step": 20504 }, { "epoch": 1.6164761529365392, "grad_norm": 0.5050609707832336, "learning_rate": 0.0005, "loss": 1.2247198820114136, "step": 20505 }, { "epoch": 1.6165549862041781, "grad_norm": 0.46763843297958374, "learning_rate": 0.0005, "loss": 1.567543387413025, "step": 20506 }, { "epoch": 1.616633819471817, "grad_norm": 0.48645246028900146, "learning_rate": 0.0005, "loss": 1.6312426328659058, "step": 20507 }, { "epoch": 1.616712652739456, "grad_norm": 0.47056031227111816, "learning_rate": 0.0005, "loss": 1.580528974533081, "step": 20508 }, { "epoch": 1.616791486007095, "grad_norm": 0.5264910459518433, "learning_rate": 0.0005, "loss": 1.6501531600952148, "step": 20509 }, { "epoch": 1.6168703192747338, "grad_norm": 0.4762555658817291, "learning_rate": 0.0005, "loss": 1.659812331199646, "step": 20510 }, { "epoch": 1.616949152542373, "grad_norm": 0.460283100605011, "learning_rate": 0.0005, "loss": 1.2999951839447021, "step": 20511 }, { "epoch": 1.6170279858100118, "grad_norm": 0.7263084053993225, "learning_rate": 0.0005, "loss": 1.6981494426727295, "step": 20512 }, { "epoch": 1.6171068190776507, "grad_norm": 0.48943760991096497, "learning_rate": 0.0005, "loss": 1.4562921524047852, "step": 20513 }, { "epoch": 1.6171856523452897, "grad_norm": 0.5216946601867676, "learning_rate": 0.0005, "loss": 1.3256148099899292, "step": 20514 }, { "epoch": 1.6172644856129286, "grad_norm": 0.5730533003807068, "learning_rate": 0.0005, "loss": 1.5916929244995117, "step": 20515 }, { "epoch": 1.6173433188805677, "grad_norm": 0.4628243148326874, "learning_rate": 0.0005, "loss": 1.4577943086624146, "step": 20516 }, { "epoch": 1.6174221521482064, "grad_norm": 0.49933093786239624, "learning_rate": 0.0005, "loss": 1.4841738939285278, "step": 20517 }, { "epoch": 1.6175009854158455, "grad_norm": 0.59376460313797, "learning_rate": 0.0005, "loss": 1.6135715246200562, "step": 20518 }, { "epoch": 1.6175798186834844, "grad_norm": 0.8285892605781555, "learning_rate": 0.0005, "loss": 1.5598039627075195, "step": 20519 }, { "epoch": 1.6176586519511233, "grad_norm": 0.5909146070480347, "learning_rate": 0.0005, "loss": 1.4553170204162598, "step": 20520 }, { "epoch": 1.6177374852187625, "grad_norm": 0.46579939126968384, "learning_rate": 0.0005, "loss": 1.370094656944275, "step": 20521 }, { "epoch": 1.6178163184864012, "grad_norm": 0.46114376187324524, "learning_rate": 0.0005, "loss": 1.3614864349365234, "step": 20522 }, { "epoch": 1.6178951517540403, "grad_norm": 0.5201641917228699, "learning_rate": 0.0005, "loss": 1.433738350868225, "step": 20523 }, { "epoch": 1.617973985021679, "grad_norm": 0.7626358866691589, "learning_rate": 0.0005, "loss": 1.5587719678878784, "step": 20524 }, { "epoch": 1.6180528182893181, "grad_norm": 0.6388518214225769, "learning_rate": 0.0005, "loss": 1.5849441289901733, "step": 20525 }, { "epoch": 1.618131651556957, "grad_norm": 0.6755558848381042, "learning_rate": 0.0005, "loss": 1.6422991752624512, "step": 20526 }, { "epoch": 1.618210484824596, "grad_norm": 0.5160205364227295, "learning_rate": 0.0005, "loss": 1.7109969854354858, "step": 20527 }, { "epoch": 1.618289318092235, "grad_norm": 0.5324077010154724, "learning_rate": 0.0005, "loss": 1.5271193981170654, "step": 20528 }, { "epoch": 1.6183681513598738, "grad_norm": 0.5724530220031738, "learning_rate": 0.0005, "loss": 1.583158254623413, "step": 20529 }, { "epoch": 1.618446984627513, "grad_norm": 0.6497920155525208, "learning_rate": 0.0005, "loss": 1.4260241985321045, "step": 20530 }, { "epoch": 1.6185258178951516, "grad_norm": 0.5835663676261902, "learning_rate": 0.0005, "loss": 1.3326706886291504, "step": 20531 }, { "epoch": 1.6186046511627907, "grad_norm": 0.5145649313926697, "learning_rate": 0.0005, "loss": 1.5844337940216064, "step": 20532 }, { "epoch": 1.6186834844304296, "grad_norm": 0.5692671537399292, "learning_rate": 0.0005, "loss": 1.615809440612793, "step": 20533 }, { "epoch": 1.6187623176980686, "grad_norm": 0.5327842235565186, "learning_rate": 0.0005, "loss": 1.4297785758972168, "step": 20534 }, { "epoch": 1.6188411509657077, "grad_norm": 0.741813600063324, "learning_rate": 0.0005, "loss": 1.5304820537567139, "step": 20535 }, { "epoch": 1.6189199842333464, "grad_norm": 0.7416790127754211, "learning_rate": 0.0005, "loss": 1.5785961151123047, "step": 20536 }, { "epoch": 1.6189988175009855, "grad_norm": 0.5292701125144958, "learning_rate": 0.0005, "loss": 1.4490798711776733, "step": 20537 }, { "epoch": 1.6190776507686242, "grad_norm": 0.6171858906745911, "learning_rate": 0.0005, "loss": 1.740692138671875, "step": 20538 }, { "epoch": 1.6191564840362633, "grad_norm": 0.6212210059165955, "learning_rate": 0.0005, "loss": 1.541264533996582, "step": 20539 }, { "epoch": 1.6192353173039022, "grad_norm": 0.5057427287101746, "learning_rate": 0.0005, "loss": 1.5594756603240967, "step": 20540 }, { "epoch": 1.6193141505715412, "grad_norm": 0.5253952741622925, "learning_rate": 0.0005, "loss": 1.4699674844741821, "step": 20541 }, { "epoch": 1.6193929838391803, "grad_norm": 0.6250691413879395, "learning_rate": 0.0005, "loss": 1.513145923614502, "step": 20542 }, { "epoch": 1.619471817106819, "grad_norm": 0.47588101029396057, "learning_rate": 0.0005, "loss": 1.5125746726989746, "step": 20543 }, { "epoch": 1.6195506503744581, "grad_norm": 0.5839968323707581, "learning_rate": 0.0005, "loss": 1.5791916847229004, "step": 20544 }, { "epoch": 1.6196294836420968, "grad_norm": 0.6335669159889221, "learning_rate": 0.0005, "loss": 1.7394386529922485, "step": 20545 }, { "epoch": 1.619708316909736, "grad_norm": 0.7968951463699341, "learning_rate": 0.0005, "loss": 1.3938379287719727, "step": 20546 }, { "epoch": 1.6197871501773748, "grad_norm": 0.5698379874229431, "learning_rate": 0.0005, "loss": 1.543485403060913, "step": 20547 }, { "epoch": 1.6198659834450138, "grad_norm": 0.526298463344574, "learning_rate": 0.0005, "loss": 1.6243237257003784, "step": 20548 }, { "epoch": 1.619944816712653, "grad_norm": 0.5193718075752258, "learning_rate": 0.0005, "loss": 1.401277780532837, "step": 20549 }, { "epoch": 1.6200236499802916, "grad_norm": 0.5550444722175598, "learning_rate": 0.0005, "loss": 1.4035289287567139, "step": 20550 }, { "epoch": 1.6201024832479307, "grad_norm": 0.7371702194213867, "learning_rate": 0.0005, "loss": 1.726191520690918, "step": 20551 }, { "epoch": 1.6201813165155696, "grad_norm": 0.8046377897262573, "learning_rate": 0.0005, "loss": 1.5463694334030151, "step": 20552 }, { "epoch": 1.6202601497832085, "grad_norm": 0.49300098419189453, "learning_rate": 0.0005, "loss": 1.730656385421753, "step": 20553 }, { "epoch": 1.6203389830508474, "grad_norm": 0.5037415623664856, "learning_rate": 0.0005, "loss": 1.4809353351593018, "step": 20554 }, { "epoch": 1.6204178163184864, "grad_norm": 0.5488771796226501, "learning_rate": 0.0005, "loss": 1.6317260265350342, "step": 20555 }, { "epoch": 1.6204966495861255, "grad_norm": 0.7632330060005188, "learning_rate": 0.0005, "loss": 1.4416335821151733, "step": 20556 }, { "epoch": 1.6205754828537642, "grad_norm": 0.6272100806236267, "learning_rate": 0.0005, "loss": 1.3898043632507324, "step": 20557 }, { "epoch": 1.6206543161214033, "grad_norm": 0.5710864663124084, "learning_rate": 0.0005, "loss": 1.4270354509353638, "step": 20558 }, { "epoch": 1.6207331493890422, "grad_norm": 0.6559444665908813, "learning_rate": 0.0005, "loss": 1.704782485961914, "step": 20559 }, { "epoch": 1.6208119826566811, "grad_norm": 0.6549155712127686, "learning_rate": 0.0005, "loss": 1.3897870779037476, "step": 20560 }, { "epoch": 1.62089081592432, "grad_norm": 0.6607330441474915, "learning_rate": 0.0005, "loss": 1.6210358142852783, "step": 20561 }, { "epoch": 1.620969649191959, "grad_norm": 0.599876344203949, "learning_rate": 0.0005, "loss": 1.5549640655517578, "step": 20562 }, { "epoch": 1.621048482459598, "grad_norm": 0.513763427734375, "learning_rate": 0.0005, "loss": 1.4893271923065186, "step": 20563 }, { "epoch": 1.6211273157272368, "grad_norm": 0.5562874674797058, "learning_rate": 0.0005, "loss": 1.5813188552856445, "step": 20564 }, { "epoch": 1.621206148994876, "grad_norm": 0.47718387842178345, "learning_rate": 0.0005, "loss": 1.6054257154464722, "step": 20565 }, { "epoch": 1.6212849822625148, "grad_norm": 0.5751086473464966, "learning_rate": 0.0005, "loss": 1.5621534585952759, "step": 20566 }, { "epoch": 1.6213638155301537, "grad_norm": 0.6050529479980469, "learning_rate": 0.0005, "loss": 1.5950592756271362, "step": 20567 }, { "epoch": 1.6214426487977927, "grad_norm": 0.4522819221019745, "learning_rate": 0.0005, "loss": 1.5155987739562988, "step": 20568 }, { "epoch": 1.6215214820654316, "grad_norm": 0.7769677042961121, "learning_rate": 0.0005, "loss": 1.7012152671813965, "step": 20569 }, { "epoch": 1.6216003153330707, "grad_norm": 0.5588132739067078, "learning_rate": 0.0005, "loss": 1.5692957639694214, "step": 20570 }, { "epoch": 1.6216791486007094, "grad_norm": 0.7069228887557983, "learning_rate": 0.0005, "loss": 1.7136108875274658, "step": 20571 }, { "epoch": 1.6217579818683485, "grad_norm": 0.7725850939750671, "learning_rate": 0.0005, "loss": 1.575476884841919, "step": 20572 }, { "epoch": 1.6218368151359874, "grad_norm": 0.6192856431007385, "learning_rate": 0.0005, "loss": 1.4675722122192383, "step": 20573 }, { "epoch": 1.6219156484036263, "grad_norm": 0.6283297538757324, "learning_rate": 0.0005, "loss": 1.6976948976516724, "step": 20574 }, { "epoch": 1.6219944816712653, "grad_norm": 0.6295080184936523, "learning_rate": 0.0005, "loss": 1.5556516647338867, "step": 20575 }, { "epoch": 1.6220733149389042, "grad_norm": 0.6959401369094849, "learning_rate": 0.0005, "loss": 1.6621595621109009, "step": 20576 }, { "epoch": 1.6221521482065433, "grad_norm": 0.5890717506408691, "learning_rate": 0.0005, "loss": 1.676703691482544, "step": 20577 }, { "epoch": 1.622230981474182, "grad_norm": 0.5552148222923279, "learning_rate": 0.0005, "loss": 1.5267760753631592, "step": 20578 }, { "epoch": 1.6223098147418211, "grad_norm": 0.5089848637580872, "learning_rate": 0.0005, "loss": 1.593213438987732, "step": 20579 }, { "epoch": 1.62238864800946, "grad_norm": 0.5070852041244507, "learning_rate": 0.0005, "loss": 1.4564862251281738, "step": 20580 }, { "epoch": 1.622467481277099, "grad_norm": 0.5937349200248718, "learning_rate": 0.0005, "loss": 1.7741563320159912, "step": 20581 }, { "epoch": 1.6225463145447379, "grad_norm": 0.5019801259040833, "learning_rate": 0.0005, "loss": 1.471091389656067, "step": 20582 }, { "epoch": 1.6226251478123768, "grad_norm": 0.5537994503974915, "learning_rate": 0.0005, "loss": 1.5452299118041992, "step": 20583 }, { "epoch": 1.622703981080016, "grad_norm": 0.5079523921012878, "learning_rate": 0.0005, "loss": 1.5181368589401245, "step": 20584 }, { "epoch": 1.6227828143476546, "grad_norm": 0.6077383160591125, "learning_rate": 0.0005, "loss": 1.6520051956176758, "step": 20585 }, { "epoch": 1.6228616476152937, "grad_norm": 0.5550092458724976, "learning_rate": 0.0005, "loss": 1.6765122413635254, "step": 20586 }, { "epoch": 1.6229404808829326, "grad_norm": 0.49415329098701477, "learning_rate": 0.0005, "loss": 1.4044288396835327, "step": 20587 }, { "epoch": 1.6230193141505715, "grad_norm": 0.5751165747642517, "learning_rate": 0.0005, "loss": 1.5766264200210571, "step": 20588 }, { "epoch": 1.6230981474182105, "grad_norm": 0.6409439444541931, "learning_rate": 0.0005, "loss": 1.6675227880477905, "step": 20589 }, { "epoch": 1.6231769806858494, "grad_norm": 0.44904714822769165, "learning_rate": 0.0005, "loss": 1.3971760272979736, "step": 20590 }, { "epoch": 1.6232558139534885, "grad_norm": 0.6729641556739807, "learning_rate": 0.0005, "loss": 1.535550594329834, "step": 20591 }, { "epoch": 1.6233346472211272, "grad_norm": 0.640238344669342, "learning_rate": 0.0005, "loss": 1.5144374370574951, "step": 20592 }, { "epoch": 1.6234134804887663, "grad_norm": 0.5349183678627014, "learning_rate": 0.0005, "loss": 1.4459037780761719, "step": 20593 }, { "epoch": 1.6234923137564052, "grad_norm": 0.7734835147857666, "learning_rate": 0.0005, "loss": 1.7681503295898438, "step": 20594 }, { "epoch": 1.6235711470240441, "grad_norm": 0.6339074373245239, "learning_rate": 0.0005, "loss": 1.6778779029846191, "step": 20595 }, { "epoch": 1.623649980291683, "grad_norm": 0.5333705544471741, "learning_rate": 0.0005, "loss": 1.472543716430664, "step": 20596 }, { "epoch": 1.623728813559322, "grad_norm": 0.7261560559272766, "learning_rate": 0.0005, "loss": 1.4253062009811401, "step": 20597 }, { "epoch": 1.623807646826961, "grad_norm": 0.46159783005714417, "learning_rate": 0.0005, "loss": 1.5892832279205322, "step": 20598 }, { "epoch": 1.6238864800945998, "grad_norm": 0.5260881781578064, "learning_rate": 0.0005, "loss": 1.6041145324707031, "step": 20599 }, { "epoch": 1.623965313362239, "grad_norm": 0.6581235527992249, "learning_rate": 0.0005, "loss": 1.6115002632141113, "step": 20600 }, { "epoch": 1.6240441466298778, "grad_norm": 0.44916921854019165, "learning_rate": 0.0005, "loss": 1.4949781894683838, "step": 20601 }, { "epoch": 1.6241229798975167, "grad_norm": 0.37320268154144287, "learning_rate": 0.0005, "loss": 1.4279234409332275, "step": 20602 }, { "epoch": 1.6242018131651557, "grad_norm": 0.6046928763389587, "learning_rate": 0.0005, "loss": 1.5029940605163574, "step": 20603 }, { "epoch": 1.6242806464327946, "grad_norm": 0.4936620891094208, "learning_rate": 0.0005, "loss": 1.4375019073486328, "step": 20604 }, { "epoch": 1.6243594797004337, "grad_norm": 0.5440322160720825, "learning_rate": 0.0005, "loss": 1.4632318019866943, "step": 20605 }, { "epoch": 1.6244383129680724, "grad_norm": 0.687942385673523, "learning_rate": 0.0005, "loss": 1.7252216339111328, "step": 20606 }, { "epoch": 1.6245171462357115, "grad_norm": 0.510895848274231, "learning_rate": 0.0005, "loss": 1.4090604782104492, "step": 20607 }, { "epoch": 1.6245959795033504, "grad_norm": 0.4968702495098114, "learning_rate": 0.0005, "loss": 1.308164119720459, "step": 20608 }, { "epoch": 1.6246748127709894, "grad_norm": 0.5886460542678833, "learning_rate": 0.0005, "loss": 1.545089840888977, "step": 20609 }, { "epoch": 1.6247536460386283, "grad_norm": 0.5760421752929688, "learning_rate": 0.0005, "loss": 1.577754020690918, "step": 20610 }, { "epoch": 1.6248324793062672, "grad_norm": 0.5887495875358582, "learning_rate": 0.0005, "loss": 1.6441487073898315, "step": 20611 }, { "epoch": 1.6249113125739063, "grad_norm": 0.5175300240516663, "learning_rate": 0.0005, "loss": 1.3487513065338135, "step": 20612 }, { "epoch": 1.624990145841545, "grad_norm": 0.6068321466445923, "learning_rate": 0.0005, "loss": 1.4463036060333252, "step": 20613 }, { "epoch": 1.6250689791091841, "grad_norm": 0.5815064907073975, "learning_rate": 0.0005, "loss": 1.645200252532959, "step": 20614 }, { "epoch": 1.625147812376823, "grad_norm": 0.5915573835372925, "learning_rate": 0.0005, "loss": 1.5930356979370117, "step": 20615 }, { "epoch": 1.625226645644462, "grad_norm": 0.5567572116851807, "learning_rate": 0.0005, "loss": 1.3970205783843994, "step": 20616 }, { "epoch": 1.6253054789121009, "grad_norm": 0.47930335998535156, "learning_rate": 0.0005, "loss": 1.3769996166229248, "step": 20617 }, { "epoch": 1.6253843121797398, "grad_norm": 0.5522221922874451, "learning_rate": 0.0005, "loss": 1.739001989364624, "step": 20618 }, { "epoch": 1.625463145447379, "grad_norm": 0.4894832670688629, "learning_rate": 0.0005, "loss": 1.4322896003723145, "step": 20619 }, { "epoch": 1.6255419787150176, "grad_norm": 0.49821925163269043, "learning_rate": 0.0005, "loss": 1.3827033042907715, "step": 20620 }, { "epoch": 1.6256208119826567, "grad_norm": 0.6404128670692444, "learning_rate": 0.0005, "loss": 1.8156216144561768, "step": 20621 }, { "epoch": 1.6256996452502956, "grad_norm": 0.4822237491607666, "learning_rate": 0.0005, "loss": 1.654579758644104, "step": 20622 }, { "epoch": 1.6257784785179346, "grad_norm": 0.5476738214492798, "learning_rate": 0.0005, "loss": 1.4641749858856201, "step": 20623 }, { "epoch": 1.6258573117855735, "grad_norm": 0.5333063006401062, "learning_rate": 0.0005, "loss": 1.3492809534072876, "step": 20624 }, { "epoch": 1.6259361450532124, "grad_norm": 0.6833055019378662, "learning_rate": 0.0005, "loss": 1.679746389389038, "step": 20625 }, { "epoch": 1.6260149783208515, "grad_norm": 0.483792781829834, "learning_rate": 0.0005, "loss": 1.510095238685608, "step": 20626 }, { "epoch": 1.6260938115884902, "grad_norm": 0.6846621036529541, "learning_rate": 0.0005, "loss": 1.6893484592437744, "step": 20627 }, { "epoch": 1.6261726448561293, "grad_norm": 0.4568367600440979, "learning_rate": 0.0005, "loss": 1.4449708461761475, "step": 20628 }, { "epoch": 1.6262514781237682, "grad_norm": 0.535333514213562, "learning_rate": 0.0005, "loss": 1.277276873588562, "step": 20629 }, { "epoch": 1.6263303113914072, "grad_norm": 0.6004951000213623, "learning_rate": 0.0005, "loss": 1.512972116470337, "step": 20630 }, { "epoch": 1.626409144659046, "grad_norm": 0.4964338541030884, "learning_rate": 0.0005, "loss": 1.4543004035949707, "step": 20631 }, { "epoch": 1.626487977926685, "grad_norm": 0.581827700138092, "learning_rate": 0.0005, "loss": 1.4699393510818481, "step": 20632 }, { "epoch": 1.6265668111943241, "grad_norm": 0.8096680641174316, "learning_rate": 0.0005, "loss": 1.5033035278320312, "step": 20633 }, { "epoch": 1.6266456444619628, "grad_norm": 0.6529608964920044, "learning_rate": 0.0005, "loss": 1.7094969749450684, "step": 20634 }, { "epoch": 1.626724477729602, "grad_norm": 0.5369023084640503, "learning_rate": 0.0005, "loss": 1.671673059463501, "step": 20635 }, { "epoch": 1.6268033109972408, "grad_norm": 0.5327973961830139, "learning_rate": 0.0005, "loss": 1.3764564990997314, "step": 20636 }, { "epoch": 1.6268821442648798, "grad_norm": 0.5482549667358398, "learning_rate": 0.0005, "loss": 1.5913803577423096, "step": 20637 }, { "epoch": 1.6269609775325187, "grad_norm": 0.6056907176971436, "learning_rate": 0.0005, "loss": 1.6194002628326416, "step": 20638 }, { "epoch": 1.6270398108001576, "grad_norm": 0.503104031085968, "learning_rate": 0.0005, "loss": 1.4612571001052856, "step": 20639 }, { "epoch": 1.6271186440677967, "grad_norm": 0.5502234101295471, "learning_rate": 0.0005, "loss": 1.3926019668579102, "step": 20640 }, { "epoch": 1.6271974773354354, "grad_norm": 0.4778502881526947, "learning_rate": 0.0005, "loss": 1.3487823009490967, "step": 20641 }, { "epoch": 1.6272763106030745, "grad_norm": 0.514681339263916, "learning_rate": 0.0005, "loss": 1.5335849523544312, "step": 20642 }, { "epoch": 1.6273551438707134, "grad_norm": 0.4704325795173645, "learning_rate": 0.0005, "loss": 1.666780710220337, "step": 20643 }, { "epoch": 1.6274339771383524, "grad_norm": 0.5039701461791992, "learning_rate": 0.0005, "loss": 1.7082867622375488, "step": 20644 }, { "epoch": 1.6275128104059915, "grad_norm": 0.6140480637550354, "learning_rate": 0.0005, "loss": 1.538801670074463, "step": 20645 }, { "epoch": 1.6275916436736302, "grad_norm": 0.41113823652267456, "learning_rate": 0.0005, "loss": 1.4958865642547607, "step": 20646 }, { "epoch": 1.6276704769412693, "grad_norm": 0.520996630191803, "learning_rate": 0.0005, "loss": 1.4143110513687134, "step": 20647 }, { "epoch": 1.627749310208908, "grad_norm": 0.5723183155059814, "learning_rate": 0.0005, "loss": 1.640990138053894, "step": 20648 }, { "epoch": 1.6278281434765471, "grad_norm": 0.4897891581058502, "learning_rate": 0.0005, "loss": 1.4021074771881104, "step": 20649 }, { "epoch": 1.627906976744186, "grad_norm": 0.45359840989112854, "learning_rate": 0.0005, "loss": 1.4333335161209106, "step": 20650 }, { "epoch": 1.627985810011825, "grad_norm": 0.5247246623039246, "learning_rate": 0.0005, "loss": 1.5953092575073242, "step": 20651 }, { "epoch": 1.628064643279464, "grad_norm": 0.697189450263977, "learning_rate": 0.0005, "loss": 1.767627239227295, "step": 20652 }, { "epoch": 1.6281434765471028, "grad_norm": 0.4069802463054657, "learning_rate": 0.0005, "loss": 1.4156383275985718, "step": 20653 }, { "epoch": 1.628222309814742, "grad_norm": 0.5506216883659363, "learning_rate": 0.0005, "loss": 1.4831171035766602, "step": 20654 }, { "epoch": 1.6283011430823806, "grad_norm": 0.6283564567565918, "learning_rate": 0.0005, "loss": 1.5966970920562744, "step": 20655 }, { "epoch": 1.6283799763500197, "grad_norm": 0.641135036945343, "learning_rate": 0.0005, "loss": 1.679836630821228, "step": 20656 }, { "epoch": 1.6284588096176587, "grad_norm": 0.48645883798599243, "learning_rate": 0.0005, "loss": 1.436992883682251, "step": 20657 }, { "epoch": 1.6285376428852976, "grad_norm": 0.697907567024231, "learning_rate": 0.0005, "loss": 1.5839887857437134, "step": 20658 }, { "epoch": 1.6286164761529367, "grad_norm": 0.5803326964378357, "learning_rate": 0.0005, "loss": 1.6153424978256226, "step": 20659 }, { "epoch": 1.6286953094205754, "grad_norm": 0.461388498544693, "learning_rate": 0.0005, "loss": 1.4530115127563477, "step": 20660 }, { "epoch": 1.6287741426882145, "grad_norm": 0.471880167722702, "learning_rate": 0.0005, "loss": 1.4491289854049683, "step": 20661 }, { "epoch": 1.6288529759558532, "grad_norm": 0.4798164963722229, "learning_rate": 0.0005, "loss": 1.6154909133911133, "step": 20662 }, { "epoch": 1.6289318092234923, "grad_norm": 0.6179368495941162, "learning_rate": 0.0005, "loss": 1.7496647834777832, "step": 20663 }, { "epoch": 1.6290106424911313, "grad_norm": 0.45262888073921204, "learning_rate": 0.0005, "loss": 1.5314668416976929, "step": 20664 }, { "epoch": 1.6290894757587702, "grad_norm": 1.8342106342315674, "learning_rate": 0.0005, "loss": 1.3208167552947998, "step": 20665 }, { "epoch": 1.6291683090264093, "grad_norm": 0.4645298421382904, "learning_rate": 0.0005, "loss": 1.611379861831665, "step": 20666 }, { "epoch": 1.629247142294048, "grad_norm": 0.446695476770401, "learning_rate": 0.0005, "loss": 1.4829647541046143, "step": 20667 }, { "epoch": 1.6293259755616871, "grad_norm": 0.5063820481300354, "learning_rate": 0.0005, "loss": 1.394683599472046, "step": 20668 }, { "epoch": 1.6294048088293258, "grad_norm": 0.4935797452926636, "learning_rate": 0.0005, "loss": 1.4319677352905273, "step": 20669 }, { "epoch": 1.629483642096965, "grad_norm": 0.4861660897731781, "learning_rate": 0.0005, "loss": 1.518455982208252, "step": 20670 }, { "epoch": 1.6295624753646039, "grad_norm": 0.7791189551353455, "learning_rate": 0.0005, "loss": 1.3401076793670654, "step": 20671 }, { "epoch": 1.6296413086322428, "grad_norm": 0.521312415599823, "learning_rate": 0.0005, "loss": 1.4947023391723633, "step": 20672 }, { "epoch": 1.629720141899882, "grad_norm": 0.5713019371032715, "learning_rate": 0.0005, "loss": 1.5800515413284302, "step": 20673 }, { "epoch": 1.6297989751675206, "grad_norm": 0.658523678779602, "learning_rate": 0.0005, "loss": 1.5510705709457397, "step": 20674 }, { "epoch": 1.6298778084351597, "grad_norm": 0.5554473400115967, "learning_rate": 0.0005, "loss": 1.4408187866210938, "step": 20675 }, { "epoch": 1.6299566417027986, "grad_norm": 0.5660321116447449, "learning_rate": 0.0005, "loss": 1.675065279006958, "step": 20676 }, { "epoch": 1.6300354749704375, "grad_norm": 0.5604825019836426, "learning_rate": 0.0005, "loss": 1.6659351587295532, "step": 20677 }, { "epoch": 1.6301143082380765, "grad_norm": 0.6055976748466492, "learning_rate": 0.0005, "loss": 1.8063396215438843, "step": 20678 }, { "epoch": 1.6301931415057154, "grad_norm": 0.4899306893348694, "learning_rate": 0.0005, "loss": 1.64933180809021, "step": 20679 }, { "epoch": 1.6302719747733545, "grad_norm": 0.6408381462097168, "learning_rate": 0.0005, "loss": 1.6037111282348633, "step": 20680 }, { "epoch": 1.6303508080409932, "grad_norm": 0.5776358246803284, "learning_rate": 0.0005, "loss": 1.8067833185195923, "step": 20681 }, { "epoch": 1.6304296413086323, "grad_norm": 0.47253212332725525, "learning_rate": 0.0005, "loss": 1.4904773235321045, "step": 20682 }, { "epoch": 1.6305084745762712, "grad_norm": 0.5510042905807495, "learning_rate": 0.0005, "loss": 1.6637754440307617, "step": 20683 }, { "epoch": 1.6305873078439101, "grad_norm": 0.5515197515487671, "learning_rate": 0.0005, "loss": 1.3851678371429443, "step": 20684 }, { "epoch": 1.630666141111549, "grad_norm": 0.6631860136985779, "learning_rate": 0.0005, "loss": 1.4753272533416748, "step": 20685 }, { "epoch": 1.630744974379188, "grad_norm": 0.6920223236083984, "learning_rate": 0.0005, "loss": 1.4211442470550537, "step": 20686 }, { "epoch": 1.630823807646827, "grad_norm": 0.5852996110916138, "learning_rate": 0.0005, "loss": 1.4226305484771729, "step": 20687 }, { "epoch": 1.6309026409144658, "grad_norm": 0.5311477184295654, "learning_rate": 0.0005, "loss": 1.4640535116195679, "step": 20688 }, { "epoch": 1.630981474182105, "grad_norm": 0.6843675374984741, "learning_rate": 0.0005, "loss": 1.5546073913574219, "step": 20689 }, { "epoch": 1.6310603074497438, "grad_norm": 0.6695809960365295, "learning_rate": 0.0005, "loss": 1.7355821132659912, "step": 20690 }, { "epoch": 1.6311391407173828, "grad_norm": 0.5176110863685608, "learning_rate": 0.0005, "loss": 1.2550827264785767, "step": 20691 }, { "epoch": 1.6312179739850217, "grad_norm": 0.5080353617668152, "learning_rate": 0.0005, "loss": 1.5811583995819092, "step": 20692 }, { "epoch": 1.6312968072526606, "grad_norm": 0.5637723207473755, "learning_rate": 0.0005, "loss": 1.7585489749908447, "step": 20693 }, { "epoch": 1.6313756405202997, "grad_norm": 0.5279530882835388, "learning_rate": 0.0005, "loss": 1.33778977394104, "step": 20694 }, { "epoch": 1.6314544737879384, "grad_norm": 0.5283194780349731, "learning_rate": 0.0005, "loss": 1.486420750617981, "step": 20695 }, { "epoch": 1.6315333070555775, "grad_norm": 0.5703471302986145, "learning_rate": 0.0005, "loss": 1.3533095121383667, "step": 20696 }, { "epoch": 1.6316121403232164, "grad_norm": 0.5501649379730225, "learning_rate": 0.0005, "loss": 1.6179347038269043, "step": 20697 }, { "epoch": 1.6316909735908554, "grad_norm": 0.48991039395332336, "learning_rate": 0.0005, "loss": 1.463748574256897, "step": 20698 }, { "epoch": 1.6317698068584943, "grad_norm": 0.7235181331634521, "learning_rate": 0.0005, "loss": 1.4282076358795166, "step": 20699 }, { "epoch": 1.6318486401261332, "grad_norm": 0.5554194450378418, "learning_rate": 0.0005, "loss": 1.2597839832305908, "step": 20700 }, { "epoch": 1.6319274733937723, "grad_norm": 0.5436860918998718, "learning_rate": 0.0005, "loss": 1.5493509769439697, "step": 20701 }, { "epoch": 1.632006306661411, "grad_norm": 0.5020498633384705, "learning_rate": 0.0005, "loss": 1.8424975872039795, "step": 20702 }, { "epoch": 1.6320851399290501, "grad_norm": 0.5662791728973389, "learning_rate": 0.0005, "loss": 1.6089434623718262, "step": 20703 }, { "epoch": 1.632163973196689, "grad_norm": 0.4712584614753723, "learning_rate": 0.0005, "loss": 1.489565134048462, "step": 20704 }, { "epoch": 1.632242806464328, "grad_norm": 0.6655226945877075, "learning_rate": 0.0005, "loss": 1.6414028406143188, "step": 20705 }, { "epoch": 1.6323216397319669, "grad_norm": 0.439321368932724, "learning_rate": 0.0005, "loss": 1.4027390480041504, "step": 20706 }, { "epoch": 1.6324004729996058, "grad_norm": 0.5375426411628723, "learning_rate": 0.0005, "loss": 1.451078176498413, "step": 20707 }, { "epoch": 1.632479306267245, "grad_norm": 0.5759960412979126, "learning_rate": 0.0005, "loss": 1.7457160949707031, "step": 20708 }, { "epoch": 1.6325581395348836, "grad_norm": 0.6338203549385071, "learning_rate": 0.0005, "loss": 1.8014953136444092, "step": 20709 }, { "epoch": 1.6326369728025227, "grad_norm": 0.6185089945793152, "learning_rate": 0.0005, "loss": 1.6554770469665527, "step": 20710 }, { "epoch": 1.6327158060701616, "grad_norm": 0.5370264053344727, "learning_rate": 0.0005, "loss": 1.4872281551361084, "step": 20711 }, { "epoch": 1.6327946393378006, "grad_norm": 0.5144832730293274, "learning_rate": 0.0005, "loss": 1.5033873319625854, "step": 20712 }, { "epoch": 1.6328734726054395, "grad_norm": 0.5613031387329102, "learning_rate": 0.0005, "loss": 1.7985548973083496, "step": 20713 }, { "epoch": 1.6329523058730784, "grad_norm": 0.5384852886199951, "learning_rate": 0.0005, "loss": 1.535794973373413, "step": 20714 }, { "epoch": 1.6330311391407175, "grad_norm": 0.45082756876945496, "learning_rate": 0.0005, "loss": 1.3109776973724365, "step": 20715 }, { "epoch": 1.6331099724083562, "grad_norm": 0.4077441394329071, "learning_rate": 0.0005, "loss": 1.4132126569747925, "step": 20716 }, { "epoch": 1.6331888056759953, "grad_norm": 0.6925930380821228, "learning_rate": 0.0005, "loss": 1.5097980499267578, "step": 20717 }, { "epoch": 1.6332676389436342, "grad_norm": 0.7034832239151001, "learning_rate": 0.0005, "loss": 1.6146703958511353, "step": 20718 }, { "epoch": 1.6333464722112732, "grad_norm": 0.48456093668937683, "learning_rate": 0.0005, "loss": 1.5621222257614136, "step": 20719 }, { "epoch": 1.633425305478912, "grad_norm": 0.6260166764259338, "learning_rate": 0.0005, "loss": 1.5914077758789062, "step": 20720 }, { "epoch": 1.633504138746551, "grad_norm": 0.5399007201194763, "learning_rate": 0.0005, "loss": 1.5782692432403564, "step": 20721 }, { "epoch": 1.6335829720141901, "grad_norm": 0.6229583024978638, "learning_rate": 0.0005, "loss": 1.6013219356536865, "step": 20722 }, { "epoch": 1.6336618052818288, "grad_norm": 0.6228519678115845, "learning_rate": 0.0005, "loss": 1.3975721597671509, "step": 20723 }, { "epoch": 1.633740638549468, "grad_norm": 0.5781974196434021, "learning_rate": 0.0005, "loss": 1.5739798545837402, "step": 20724 }, { "epoch": 1.6338194718171068, "grad_norm": 0.6362464427947998, "learning_rate": 0.0005, "loss": 1.5922367572784424, "step": 20725 }, { "epoch": 1.6338983050847458, "grad_norm": 0.4233147203922272, "learning_rate": 0.0005, "loss": 1.4945212602615356, "step": 20726 }, { "epoch": 1.6339771383523847, "grad_norm": 0.42628213763237, "learning_rate": 0.0005, "loss": 1.384608507156372, "step": 20727 }, { "epoch": 1.6340559716200236, "grad_norm": 0.7132258415222168, "learning_rate": 0.0005, "loss": 1.8317701816558838, "step": 20728 }, { "epoch": 1.6341348048876627, "grad_norm": 0.5958670377731323, "learning_rate": 0.0005, "loss": 1.4923768043518066, "step": 20729 }, { "epoch": 1.6342136381553014, "grad_norm": 1.0366919040679932, "learning_rate": 0.0005, "loss": 1.5389988422393799, "step": 20730 }, { "epoch": 1.6342924714229405, "grad_norm": 0.6731536388397217, "learning_rate": 0.0005, "loss": 1.4684704542160034, "step": 20731 }, { "epoch": 1.6343713046905795, "grad_norm": 0.48952245712280273, "learning_rate": 0.0005, "loss": 1.4089797735214233, "step": 20732 }, { "epoch": 1.6344501379582184, "grad_norm": 0.6879194378852844, "learning_rate": 0.0005, "loss": 1.4871711730957031, "step": 20733 }, { "epoch": 1.6345289712258573, "grad_norm": 0.5709495544433594, "learning_rate": 0.0005, "loss": 1.4748985767364502, "step": 20734 }, { "epoch": 1.6346078044934962, "grad_norm": 0.7975462675094604, "learning_rate": 0.0005, "loss": 1.5039935111999512, "step": 20735 }, { "epoch": 1.6346866377611353, "grad_norm": 0.6721401810646057, "learning_rate": 0.0005, "loss": 1.5400173664093018, "step": 20736 }, { "epoch": 1.634765471028774, "grad_norm": 0.5547158718109131, "learning_rate": 0.0005, "loss": 1.6974446773529053, "step": 20737 }, { "epoch": 1.6348443042964131, "grad_norm": 0.5192938446998596, "learning_rate": 0.0005, "loss": 1.8883919715881348, "step": 20738 }, { "epoch": 1.634923137564052, "grad_norm": 0.548061192035675, "learning_rate": 0.0005, "loss": 1.5974746942520142, "step": 20739 }, { "epoch": 1.635001970831691, "grad_norm": 0.6520075798034668, "learning_rate": 0.0005, "loss": 1.418155550956726, "step": 20740 }, { "epoch": 1.6350808040993299, "grad_norm": 0.45307227969169617, "learning_rate": 0.0005, "loss": 1.381088137626648, "step": 20741 }, { "epoch": 1.6351596373669688, "grad_norm": 0.670340359210968, "learning_rate": 0.0005, "loss": 1.6929707527160645, "step": 20742 }, { "epoch": 1.635238470634608, "grad_norm": 0.5356225967407227, "learning_rate": 0.0005, "loss": 1.6820199489593506, "step": 20743 }, { "epoch": 1.6353173039022466, "grad_norm": 0.5213468074798584, "learning_rate": 0.0005, "loss": 1.4729902744293213, "step": 20744 }, { "epoch": 1.6353961371698857, "grad_norm": 0.7095328569412231, "learning_rate": 0.0005, "loss": 1.585932970046997, "step": 20745 }, { "epoch": 1.6354749704375247, "grad_norm": 0.5626591444015503, "learning_rate": 0.0005, "loss": 1.3652796745300293, "step": 20746 }, { "epoch": 1.6355538037051636, "grad_norm": 0.5325178503990173, "learning_rate": 0.0005, "loss": 1.669161081314087, "step": 20747 }, { "epoch": 1.6356326369728025, "grad_norm": 0.5191550850868225, "learning_rate": 0.0005, "loss": 1.6277579069137573, "step": 20748 }, { "epoch": 1.6357114702404414, "grad_norm": 0.46001896262168884, "learning_rate": 0.0005, "loss": 1.7195736169815063, "step": 20749 }, { "epoch": 1.6357903035080805, "grad_norm": 0.4619337320327759, "learning_rate": 0.0005, "loss": 1.5396177768707275, "step": 20750 }, { "epoch": 1.6358691367757192, "grad_norm": 0.49436965584754944, "learning_rate": 0.0005, "loss": 1.5994439125061035, "step": 20751 }, { "epoch": 1.6359479700433583, "grad_norm": 0.5481864213943481, "learning_rate": 0.0005, "loss": 1.5612846612930298, "step": 20752 }, { "epoch": 1.6360268033109973, "grad_norm": 0.5675292611122131, "learning_rate": 0.0005, "loss": 1.683229684829712, "step": 20753 }, { "epoch": 1.6361056365786362, "grad_norm": 0.5478784441947937, "learning_rate": 0.0005, "loss": 1.7545406818389893, "step": 20754 }, { "epoch": 1.636184469846275, "grad_norm": 0.5039761066436768, "learning_rate": 0.0005, "loss": 1.688307285308838, "step": 20755 }, { "epoch": 1.636263303113914, "grad_norm": 0.4224841594696045, "learning_rate": 0.0005, "loss": 1.3930418491363525, "step": 20756 }, { "epoch": 1.6363421363815531, "grad_norm": 0.6043253540992737, "learning_rate": 0.0005, "loss": 1.473418116569519, "step": 20757 }, { "epoch": 1.6364209696491918, "grad_norm": 0.5472943186759949, "learning_rate": 0.0005, "loss": 1.662163257598877, "step": 20758 }, { "epoch": 1.636499802916831, "grad_norm": 0.5142081379890442, "learning_rate": 0.0005, "loss": 1.3453803062438965, "step": 20759 }, { "epoch": 1.6365786361844699, "grad_norm": 0.5220242738723755, "learning_rate": 0.0005, "loss": 1.6270859241485596, "step": 20760 }, { "epoch": 1.6366574694521088, "grad_norm": 0.8197159767150879, "learning_rate": 0.0005, "loss": 1.3686349391937256, "step": 20761 }, { "epoch": 1.6367363027197477, "grad_norm": 0.6291656494140625, "learning_rate": 0.0005, "loss": 1.635004997253418, "step": 20762 }, { "epoch": 1.6368151359873866, "grad_norm": 0.696678102016449, "learning_rate": 0.0005, "loss": 1.4903059005737305, "step": 20763 }, { "epoch": 1.6368939692550257, "grad_norm": 0.641184389591217, "learning_rate": 0.0005, "loss": 1.9264099597930908, "step": 20764 }, { "epoch": 1.6369728025226644, "grad_norm": 0.45234355330467224, "learning_rate": 0.0005, "loss": 1.3502076864242554, "step": 20765 }, { "epoch": 1.6370516357903035, "grad_norm": 0.5368102192878723, "learning_rate": 0.0005, "loss": 1.6194064617156982, "step": 20766 }, { "epoch": 1.6371304690579425, "grad_norm": 0.7280899286270142, "learning_rate": 0.0005, "loss": 1.7226262092590332, "step": 20767 }, { "epoch": 1.6372093023255814, "grad_norm": 0.5072065591812134, "learning_rate": 0.0005, "loss": 1.420772671699524, "step": 20768 }, { "epoch": 1.6372881355932203, "grad_norm": 0.6074749827384949, "learning_rate": 0.0005, "loss": 1.5790454149246216, "step": 20769 }, { "epoch": 1.6373669688608592, "grad_norm": 0.8135126829147339, "learning_rate": 0.0005, "loss": 1.51338791847229, "step": 20770 }, { "epoch": 1.6374458021284983, "grad_norm": 0.5359103083610535, "learning_rate": 0.0005, "loss": 1.611234188079834, "step": 20771 }, { "epoch": 1.637524635396137, "grad_norm": 0.7782632112503052, "learning_rate": 0.0005, "loss": 1.624124526977539, "step": 20772 }, { "epoch": 1.6376034686637762, "grad_norm": 0.5456526279449463, "learning_rate": 0.0005, "loss": 1.5598112344741821, "step": 20773 }, { "epoch": 1.637682301931415, "grad_norm": 0.7016406655311584, "learning_rate": 0.0005, "loss": 1.6262285709381104, "step": 20774 }, { "epoch": 1.637761135199054, "grad_norm": 0.6274771690368652, "learning_rate": 0.0005, "loss": 1.5499601364135742, "step": 20775 }, { "epoch": 1.637839968466693, "grad_norm": 0.4833981394767761, "learning_rate": 0.0005, "loss": 1.4357434511184692, "step": 20776 }, { "epoch": 1.6379188017343318, "grad_norm": 0.511727511882782, "learning_rate": 0.0005, "loss": 1.5043178796768188, "step": 20777 }, { "epoch": 1.637997635001971, "grad_norm": 0.6627973914146423, "learning_rate": 0.0005, "loss": 1.4235053062438965, "step": 20778 }, { "epoch": 1.6380764682696096, "grad_norm": 0.6760739684104919, "learning_rate": 0.0005, "loss": 1.711972951889038, "step": 20779 }, { "epoch": 1.6381553015372488, "grad_norm": 0.4432756006717682, "learning_rate": 0.0005, "loss": 1.5391573905944824, "step": 20780 }, { "epoch": 1.6382341348048877, "grad_norm": 0.5736929178237915, "learning_rate": 0.0005, "loss": 1.6187118291854858, "step": 20781 }, { "epoch": 1.6383129680725266, "grad_norm": 0.6029667854309082, "learning_rate": 0.0005, "loss": 1.6221530437469482, "step": 20782 }, { "epoch": 1.6383918013401657, "grad_norm": 0.696133017539978, "learning_rate": 0.0005, "loss": 1.7157957553863525, "step": 20783 }, { "epoch": 1.6384706346078044, "grad_norm": 0.5492125153541565, "learning_rate": 0.0005, "loss": 1.4676344394683838, "step": 20784 }, { "epoch": 1.6385494678754435, "grad_norm": 0.5407068729400635, "learning_rate": 0.0005, "loss": 1.6312451362609863, "step": 20785 }, { "epoch": 1.6386283011430822, "grad_norm": 0.480885773897171, "learning_rate": 0.0005, "loss": 1.4872832298278809, "step": 20786 }, { "epoch": 1.6387071344107214, "grad_norm": 0.5520555377006531, "learning_rate": 0.0005, "loss": 1.6673076152801514, "step": 20787 }, { "epoch": 1.6387859676783603, "grad_norm": 0.6160959601402283, "learning_rate": 0.0005, "loss": 1.6952664852142334, "step": 20788 }, { "epoch": 1.6388648009459992, "grad_norm": 0.6055582761764526, "learning_rate": 0.0005, "loss": 1.5576152801513672, "step": 20789 }, { "epoch": 1.6389436342136383, "grad_norm": 0.5931528210639954, "learning_rate": 0.0005, "loss": 1.5135654211044312, "step": 20790 }, { "epoch": 1.639022467481277, "grad_norm": 0.47401484847068787, "learning_rate": 0.0005, "loss": 1.124260425567627, "step": 20791 }, { "epoch": 1.6391013007489161, "grad_norm": 0.5306836366653442, "learning_rate": 0.0005, "loss": 1.4603362083435059, "step": 20792 }, { "epoch": 1.6391801340165548, "grad_norm": 0.5818536281585693, "learning_rate": 0.0005, "loss": 1.5013890266418457, "step": 20793 }, { "epoch": 1.639258967284194, "grad_norm": 0.5335177183151245, "learning_rate": 0.0005, "loss": 1.6152085065841675, "step": 20794 }, { "epoch": 1.6393378005518329, "grad_norm": 0.6045417785644531, "learning_rate": 0.0005, "loss": 1.4309182167053223, "step": 20795 }, { "epoch": 1.6394166338194718, "grad_norm": 0.47309669852256775, "learning_rate": 0.0005, "loss": 1.4024267196655273, "step": 20796 }, { "epoch": 1.639495467087111, "grad_norm": 0.6900392770767212, "learning_rate": 0.0005, "loss": 1.7169933319091797, "step": 20797 }, { "epoch": 1.6395743003547496, "grad_norm": 0.6325284242630005, "learning_rate": 0.0005, "loss": 1.4239013195037842, "step": 20798 }, { "epoch": 1.6396531336223887, "grad_norm": 0.47926589846611023, "learning_rate": 0.0005, "loss": 1.6269938945770264, "step": 20799 }, { "epoch": 1.6397319668900274, "grad_norm": 0.5233786106109619, "learning_rate": 0.0005, "loss": 1.6799042224884033, "step": 20800 }, { "epoch": 1.6398108001576666, "grad_norm": 0.5259880423545837, "learning_rate": 0.0005, "loss": 1.5254995822906494, "step": 20801 }, { "epoch": 1.6398896334253055, "grad_norm": 0.6079058051109314, "learning_rate": 0.0005, "loss": 1.4715518951416016, "step": 20802 }, { "epoch": 1.6399684666929444, "grad_norm": 0.5111764073371887, "learning_rate": 0.0005, "loss": 1.5951043367385864, "step": 20803 }, { "epoch": 1.6400472999605835, "grad_norm": 0.6617665886878967, "learning_rate": 0.0005, "loss": 1.6510738134384155, "step": 20804 }, { "epoch": 1.6401261332282222, "grad_norm": 0.6020872592926025, "learning_rate": 0.0005, "loss": 1.5741138458251953, "step": 20805 }, { "epoch": 1.6402049664958613, "grad_norm": 0.5266449451446533, "learning_rate": 0.0005, "loss": 1.4223558902740479, "step": 20806 }, { "epoch": 1.6402837997635002, "grad_norm": 0.48336461186408997, "learning_rate": 0.0005, "loss": 1.5418903827667236, "step": 20807 }, { "epoch": 1.6403626330311392, "grad_norm": 0.482136070728302, "learning_rate": 0.0005, "loss": 1.5152775049209595, "step": 20808 }, { "epoch": 1.640441466298778, "grad_norm": 0.5360079407691956, "learning_rate": 0.0005, "loss": 1.527482509613037, "step": 20809 }, { "epoch": 1.640520299566417, "grad_norm": 0.612061083316803, "learning_rate": 0.0005, "loss": 1.6572948694229126, "step": 20810 }, { "epoch": 1.6405991328340561, "grad_norm": 0.5518805980682373, "learning_rate": 0.0005, "loss": 1.6327636241912842, "step": 20811 }, { "epoch": 1.6406779661016948, "grad_norm": 0.505515992641449, "learning_rate": 0.0005, "loss": 1.6909468173980713, "step": 20812 }, { "epoch": 1.640756799369334, "grad_norm": 0.5323324203491211, "learning_rate": 0.0005, "loss": 1.6613044738769531, "step": 20813 }, { "epoch": 1.6408356326369729, "grad_norm": 0.5417746901512146, "learning_rate": 0.0005, "loss": 1.5005006790161133, "step": 20814 }, { "epoch": 1.6409144659046118, "grad_norm": 0.4171655774116516, "learning_rate": 0.0005, "loss": 1.4022674560546875, "step": 20815 }, { "epoch": 1.6409932991722507, "grad_norm": 0.4583095610141754, "learning_rate": 0.0005, "loss": 1.3023560047149658, "step": 20816 }, { "epoch": 1.6410721324398896, "grad_norm": 0.5278512239456177, "learning_rate": 0.0005, "loss": 1.5022363662719727, "step": 20817 }, { "epoch": 1.6411509657075287, "grad_norm": 0.5409275889396667, "learning_rate": 0.0005, "loss": 1.6380994319915771, "step": 20818 }, { "epoch": 1.6412297989751674, "grad_norm": 0.5361637473106384, "learning_rate": 0.0005, "loss": 1.5790376663208008, "step": 20819 }, { "epoch": 1.6413086322428065, "grad_norm": 0.6666021347045898, "learning_rate": 0.0005, "loss": 1.7122673988342285, "step": 20820 }, { "epoch": 1.6413874655104455, "grad_norm": 1.4296107292175293, "learning_rate": 0.0005, "loss": 1.4428420066833496, "step": 20821 }, { "epoch": 1.6414662987780844, "grad_norm": 0.4948878586292267, "learning_rate": 0.0005, "loss": 1.6431164741516113, "step": 20822 }, { "epoch": 1.6415451320457233, "grad_norm": 0.5131826400756836, "learning_rate": 0.0005, "loss": 1.4693646430969238, "step": 20823 }, { "epoch": 1.6416239653133622, "grad_norm": 0.5990829467773438, "learning_rate": 0.0005, "loss": 1.632338523864746, "step": 20824 }, { "epoch": 1.6417027985810013, "grad_norm": 0.4465945363044739, "learning_rate": 0.0005, "loss": 1.4477299451828003, "step": 20825 }, { "epoch": 1.64178163184864, "grad_norm": 0.5143782496452332, "learning_rate": 0.0005, "loss": 1.5995368957519531, "step": 20826 }, { "epoch": 1.6418604651162791, "grad_norm": 0.46873152256011963, "learning_rate": 0.0005, "loss": 1.6852922439575195, "step": 20827 }, { "epoch": 1.641939298383918, "grad_norm": 0.5935359001159668, "learning_rate": 0.0005, "loss": 1.788779616355896, "step": 20828 }, { "epoch": 1.642018131651557, "grad_norm": 0.6518723368644714, "learning_rate": 0.0005, "loss": 1.6512261629104614, "step": 20829 }, { "epoch": 1.6420969649191959, "grad_norm": 0.5859129428863525, "learning_rate": 0.0005, "loss": 1.5452525615692139, "step": 20830 }, { "epoch": 1.6421757981868348, "grad_norm": 1.0928055047988892, "learning_rate": 0.0005, "loss": 1.6836726665496826, "step": 20831 }, { "epoch": 1.642254631454474, "grad_norm": 0.6272920370101929, "learning_rate": 0.0005, "loss": 1.5262693166732788, "step": 20832 }, { "epoch": 1.6423334647221126, "grad_norm": 0.584421694278717, "learning_rate": 0.0005, "loss": 1.5629194974899292, "step": 20833 }, { "epoch": 1.6424122979897517, "grad_norm": 0.7181170582771301, "learning_rate": 0.0005, "loss": 1.3833271265029907, "step": 20834 }, { "epoch": 1.6424911312573907, "grad_norm": 0.627314031124115, "learning_rate": 0.0005, "loss": 1.532863736152649, "step": 20835 }, { "epoch": 1.6425699645250296, "grad_norm": 0.5092777609825134, "learning_rate": 0.0005, "loss": 1.3715484142303467, "step": 20836 }, { "epoch": 1.6426487977926685, "grad_norm": 0.5527842044830322, "learning_rate": 0.0005, "loss": 1.4079667329788208, "step": 20837 }, { "epoch": 1.6427276310603074, "grad_norm": 0.5781968235969543, "learning_rate": 0.0005, "loss": 1.4627405405044556, "step": 20838 }, { "epoch": 1.6428064643279465, "grad_norm": 0.4597231149673462, "learning_rate": 0.0005, "loss": 1.375290036201477, "step": 20839 }, { "epoch": 1.6428852975955852, "grad_norm": 0.5623408555984497, "learning_rate": 0.0005, "loss": 1.4572043418884277, "step": 20840 }, { "epoch": 1.6429641308632243, "grad_norm": 0.5983046293258667, "learning_rate": 0.0005, "loss": 1.51194429397583, "step": 20841 }, { "epoch": 1.6430429641308633, "grad_norm": 0.48753708600997925, "learning_rate": 0.0005, "loss": 1.6603760719299316, "step": 20842 }, { "epoch": 1.6431217973985022, "grad_norm": 1.964273452758789, "learning_rate": 0.0005, "loss": 1.6696621179580688, "step": 20843 }, { "epoch": 1.643200630666141, "grad_norm": 0.6389749646186829, "learning_rate": 0.0005, "loss": 1.6455031633377075, "step": 20844 }, { "epoch": 1.64327946393378, "grad_norm": 0.473866730928421, "learning_rate": 0.0005, "loss": 1.4379334449768066, "step": 20845 }, { "epoch": 1.6433582972014191, "grad_norm": 0.833111584186554, "learning_rate": 0.0005, "loss": 1.4949709177017212, "step": 20846 }, { "epoch": 1.6434371304690578, "grad_norm": 0.5835444927215576, "learning_rate": 0.0005, "loss": 1.5156712532043457, "step": 20847 }, { "epoch": 1.643515963736697, "grad_norm": 0.5513944029808044, "learning_rate": 0.0005, "loss": 1.5778666734695435, "step": 20848 }, { "epoch": 1.6435947970043359, "grad_norm": 0.5131604671478271, "learning_rate": 0.0005, "loss": 1.430678367614746, "step": 20849 }, { "epoch": 1.6436736302719748, "grad_norm": 0.6649570465087891, "learning_rate": 0.0005, "loss": 1.5829647779464722, "step": 20850 }, { "epoch": 1.6437524635396137, "grad_norm": 0.6706787347793579, "learning_rate": 0.0005, "loss": 1.6130926609039307, "step": 20851 }, { "epoch": 1.6438312968072526, "grad_norm": 0.5476754903793335, "learning_rate": 0.0005, "loss": 1.3706414699554443, "step": 20852 }, { "epoch": 1.6439101300748917, "grad_norm": 0.4391385316848755, "learning_rate": 0.0005, "loss": 1.6083942651748657, "step": 20853 }, { "epoch": 1.6439889633425304, "grad_norm": 0.5295541286468506, "learning_rate": 0.0005, "loss": 1.731278657913208, "step": 20854 }, { "epoch": 1.6440677966101696, "grad_norm": 0.5368552803993225, "learning_rate": 0.0005, "loss": 1.657569169998169, "step": 20855 }, { "epoch": 1.6441466298778085, "grad_norm": 0.5716474652290344, "learning_rate": 0.0005, "loss": 1.598726511001587, "step": 20856 }, { "epoch": 1.6442254631454474, "grad_norm": 3.068086862564087, "learning_rate": 0.0005, "loss": 1.6313002109527588, "step": 20857 }, { "epoch": 1.6443042964130863, "grad_norm": 0.5047104954719543, "learning_rate": 0.0005, "loss": 1.3215439319610596, "step": 20858 }, { "epoch": 1.6443831296807252, "grad_norm": 0.5692322850227356, "learning_rate": 0.0005, "loss": 1.5211565494537354, "step": 20859 }, { "epoch": 1.6444619629483643, "grad_norm": 48.88533020019531, "learning_rate": 0.0005, "loss": 1.6406526565551758, "step": 20860 }, { "epoch": 1.644540796216003, "grad_norm": 0.6216261982917786, "learning_rate": 0.0005, "loss": 1.5796284675598145, "step": 20861 }, { "epoch": 1.6446196294836422, "grad_norm": 0.5821511745452881, "learning_rate": 0.0005, "loss": 1.5697624683380127, "step": 20862 }, { "epoch": 1.644698462751281, "grad_norm": 0.6712638139724731, "learning_rate": 0.0005, "loss": 1.4588432312011719, "step": 20863 }, { "epoch": 1.64477729601892, "grad_norm": 0.7643897533416748, "learning_rate": 0.0005, "loss": 1.40967857837677, "step": 20864 }, { "epoch": 1.6448561292865589, "grad_norm": 0.689814031124115, "learning_rate": 0.0005, "loss": 1.7761093378067017, "step": 20865 }, { "epoch": 1.6449349625541978, "grad_norm": 0.5530067682266235, "learning_rate": 0.0005, "loss": 1.6253199577331543, "step": 20866 }, { "epoch": 1.645013795821837, "grad_norm": 0.5396491885185242, "learning_rate": 0.0005, "loss": 1.4963223934173584, "step": 20867 }, { "epoch": 1.6450926290894756, "grad_norm": 0.49093201756477356, "learning_rate": 0.0005, "loss": 1.5867302417755127, "step": 20868 }, { "epoch": 1.6451714623571148, "grad_norm": 0.550275444984436, "learning_rate": 0.0005, "loss": 1.6068212985992432, "step": 20869 }, { "epoch": 1.6452502956247537, "grad_norm": 0.4446662366390228, "learning_rate": 0.0005, "loss": 1.315872311592102, "step": 20870 }, { "epoch": 1.6453291288923926, "grad_norm": 0.528600811958313, "learning_rate": 0.0005, "loss": 1.5585230588912964, "step": 20871 }, { "epoch": 1.6454079621600315, "grad_norm": 0.5896075367927551, "learning_rate": 0.0005, "loss": 1.4996446371078491, "step": 20872 }, { "epoch": 1.6454867954276704, "grad_norm": 0.5609591007232666, "learning_rate": 0.0005, "loss": 1.4600777626037598, "step": 20873 }, { "epoch": 1.6455656286953095, "grad_norm": 0.636327862739563, "learning_rate": 0.0005, "loss": 1.6161894798278809, "step": 20874 }, { "epoch": 1.6456444619629482, "grad_norm": 0.522129237651825, "learning_rate": 0.0005, "loss": 1.3340390920639038, "step": 20875 }, { "epoch": 1.6457232952305874, "grad_norm": 0.5446076393127441, "learning_rate": 0.0005, "loss": 1.4868706464767456, "step": 20876 }, { "epoch": 1.6458021284982263, "grad_norm": 0.5673097372055054, "learning_rate": 0.0005, "loss": 1.5792210102081299, "step": 20877 }, { "epoch": 1.6458809617658652, "grad_norm": 0.5598545074462891, "learning_rate": 0.0005, "loss": 1.4051541090011597, "step": 20878 }, { "epoch": 1.645959795033504, "grad_norm": 0.4826231598854065, "learning_rate": 0.0005, "loss": 1.6160266399383545, "step": 20879 }, { "epoch": 1.646038628301143, "grad_norm": 0.5595191717147827, "learning_rate": 0.0005, "loss": 1.4601739645004272, "step": 20880 }, { "epoch": 1.6461174615687821, "grad_norm": 0.5255511999130249, "learning_rate": 0.0005, "loss": 1.4836652278900146, "step": 20881 }, { "epoch": 1.6461962948364208, "grad_norm": 0.4871220588684082, "learning_rate": 0.0005, "loss": 1.5394186973571777, "step": 20882 }, { "epoch": 1.64627512810406, "grad_norm": 0.47771063446998596, "learning_rate": 0.0005, "loss": 1.3164567947387695, "step": 20883 }, { "epoch": 1.6463539613716989, "grad_norm": 0.6102815866470337, "learning_rate": 0.0005, "loss": 1.6708238124847412, "step": 20884 }, { "epoch": 1.6464327946393378, "grad_norm": 0.5250336527824402, "learning_rate": 0.0005, "loss": 1.6320126056671143, "step": 20885 }, { "epoch": 1.6465116279069767, "grad_norm": 1.110454797744751, "learning_rate": 0.0005, "loss": 1.294691801071167, "step": 20886 }, { "epoch": 1.6465904611746156, "grad_norm": 0.48465052247047424, "learning_rate": 0.0005, "loss": 1.43733549118042, "step": 20887 }, { "epoch": 1.6466692944422547, "grad_norm": 0.548979640007019, "learning_rate": 0.0005, "loss": 1.5010154247283936, "step": 20888 }, { "epoch": 1.6467481277098934, "grad_norm": 0.5660932064056396, "learning_rate": 0.0005, "loss": 1.7522403001785278, "step": 20889 }, { "epoch": 1.6468269609775326, "grad_norm": 0.5737589001655579, "learning_rate": 0.0005, "loss": 1.5399367809295654, "step": 20890 }, { "epoch": 1.6469057942451715, "grad_norm": 0.7373583912849426, "learning_rate": 0.0005, "loss": 1.6123805046081543, "step": 20891 }, { "epoch": 1.6469846275128104, "grad_norm": 1.2552915811538696, "learning_rate": 0.0005, "loss": 1.605318307876587, "step": 20892 }, { "epoch": 1.6470634607804493, "grad_norm": 0.5184423923492432, "learning_rate": 0.0005, "loss": 1.459920883178711, "step": 20893 }, { "epoch": 1.6471422940480882, "grad_norm": 0.5681564211845398, "learning_rate": 0.0005, "loss": 1.4809414148330688, "step": 20894 }, { "epoch": 1.6472211273157273, "grad_norm": 0.4796416759490967, "learning_rate": 0.0005, "loss": 1.514397144317627, "step": 20895 }, { "epoch": 1.647299960583366, "grad_norm": 0.5781636238098145, "learning_rate": 0.0005, "loss": 1.7713083028793335, "step": 20896 }, { "epoch": 1.6473787938510052, "grad_norm": 0.6205325722694397, "learning_rate": 0.0005, "loss": 1.5427970886230469, "step": 20897 }, { "epoch": 1.647457627118644, "grad_norm": 0.5055297613143921, "learning_rate": 0.0005, "loss": 1.4852454662322998, "step": 20898 }, { "epoch": 1.647536460386283, "grad_norm": 0.5359715223312378, "learning_rate": 0.0005, "loss": 1.4839794635772705, "step": 20899 }, { "epoch": 1.6476152936539221, "grad_norm": 0.5840942859649658, "learning_rate": 0.0005, "loss": 1.7673091888427734, "step": 20900 }, { "epoch": 1.6476941269215608, "grad_norm": 0.5024877190589905, "learning_rate": 0.0005, "loss": 1.566853642463684, "step": 20901 }, { "epoch": 1.6477729601892, "grad_norm": 0.55311119556427, "learning_rate": 0.0005, "loss": 1.5095359086990356, "step": 20902 }, { "epoch": 1.6478517934568386, "grad_norm": 0.5407224893569946, "learning_rate": 0.0005, "loss": 1.3146448135375977, "step": 20903 }, { "epoch": 1.6479306267244778, "grad_norm": 0.6751870512962341, "learning_rate": 0.0005, "loss": 1.5823229551315308, "step": 20904 }, { "epoch": 1.6480094599921167, "grad_norm": 0.6045406460762024, "learning_rate": 0.0005, "loss": 1.708961009979248, "step": 20905 }, { "epoch": 1.6480882932597556, "grad_norm": 0.4800070822238922, "learning_rate": 0.0005, "loss": 1.6661187410354614, "step": 20906 }, { "epoch": 1.6481671265273947, "grad_norm": 0.47396641969680786, "learning_rate": 0.0005, "loss": 1.73224937915802, "step": 20907 }, { "epoch": 1.6482459597950334, "grad_norm": 0.5677468180656433, "learning_rate": 0.0005, "loss": 1.645176887512207, "step": 20908 }, { "epoch": 1.6483247930626725, "grad_norm": 0.6234180927276611, "learning_rate": 0.0005, "loss": 1.733595609664917, "step": 20909 }, { "epoch": 1.6484036263303112, "grad_norm": 0.5952596068382263, "learning_rate": 0.0005, "loss": 1.6440485715866089, "step": 20910 }, { "epoch": 1.6484824595979504, "grad_norm": 0.619974672794342, "learning_rate": 0.0005, "loss": 1.3138335943222046, "step": 20911 }, { "epoch": 1.6485612928655893, "grad_norm": 0.4737203121185303, "learning_rate": 0.0005, "loss": 1.4593806266784668, "step": 20912 }, { "epoch": 1.6486401261332282, "grad_norm": 0.4921294152736664, "learning_rate": 0.0005, "loss": 1.581039547920227, "step": 20913 }, { "epoch": 1.6487189594008673, "grad_norm": 0.6734951138496399, "learning_rate": 0.0005, "loss": 1.656029462814331, "step": 20914 }, { "epoch": 1.648797792668506, "grad_norm": 0.6119799017906189, "learning_rate": 0.0005, "loss": 1.534684658050537, "step": 20915 }, { "epoch": 1.6488766259361451, "grad_norm": 0.6359831690788269, "learning_rate": 0.0005, "loss": 1.481231451034546, "step": 20916 }, { "epoch": 1.6489554592037838, "grad_norm": 0.5431870818138123, "learning_rate": 0.0005, "loss": 1.3909567594528198, "step": 20917 }, { "epoch": 1.649034292471423, "grad_norm": 0.4849521517753601, "learning_rate": 0.0005, "loss": 1.6115036010742188, "step": 20918 }, { "epoch": 1.6491131257390619, "grad_norm": 0.4568207561969757, "learning_rate": 0.0005, "loss": 1.4194440841674805, "step": 20919 }, { "epoch": 1.6491919590067008, "grad_norm": 0.5026752352714539, "learning_rate": 0.0005, "loss": 1.5049457550048828, "step": 20920 }, { "epoch": 1.64927079227434, "grad_norm": 0.5293405055999756, "learning_rate": 0.0005, "loss": 1.62351655960083, "step": 20921 }, { "epoch": 1.6493496255419786, "grad_norm": 0.6048030853271484, "learning_rate": 0.0005, "loss": 1.7125486135482788, "step": 20922 }, { "epoch": 1.6494284588096177, "grad_norm": 0.5910265445709229, "learning_rate": 0.0005, "loss": 1.7357057332992554, "step": 20923 }, { "epoch": 1.6495072920772564, "grad_norm": 0.7108701467514038, "learning_rate": 0.0005, "loss": 1.5388834476470947, "step": 20924 }, { "epoch": 1.6495861253448956, "grad_norm": 0.590270459651947, "learning_rate": 0.0005, "loss": 1.6423321962356567, "step": 20925 }, { "epoch": 1.6496649586125345, "grad_norm": 0.5049388408660889, "learning_rate": 0.0005, "loss": 1.487466812133789, "step": 20926 }, { "epoch": 1.6497437918801734, "grad_norm": 1.0059741735458374, "learning_rate": 0.0005, "loss": 1.5691794157028198, "step": 20927 }, { "epoch": 1.6498226251478125, "grad_norm": 0.551813542842865, "learning_rate": 0.0005, "loss": 1.446375846862793, "step": 20928 }, { "epoch": 1.6499014584154512, "grad_norm": 0.645647406578064, "learning_rate": 0.0005, "loss": 1.5799198150634766, "step": 20929 }, { "epoch": 1.6499802916830903, "grad_norm": 0.5724457502365112, "learning_rate": 0.0005, "loss": 1.7011618614196777, "step": 20930 }, { "epoch": 1.6500591249507293, "grad_norm": 0.5421975255012512, "learning_rate": 0.0005, "loss": 1.5851951837539673, "step": 20931 }, { "epoch": 1.6501379582183682, "grad_norm": 0.5688921213150024, "learning_rate": 0.0005, "loss": 1.7293835878372192, "step": 20932 }, { "epoch": 1.650216791486007, "grad_norm": 0.8777599930763245, "learning_rate": 0.0005, "loss": 1.4835405349731445, "step": 20933 }, { "epoch": 1.650295624753646, "grad_norm": 0.6105015873908997, "learning_rate": 0.0005, "loss": 1.634573221206665, "step": 20934 }, { "epoch": 1.6503744580212851, "grad_norm": 0.5503270030021667, "learning_rate": 0.0005, "loss": 1.4849579334259033, "step": 20935 }, { "epoch": 1.6504532912889238, "grad_norm": 0.544424831867218, "learning_rate": 0.0005, "loss": 1.5645084381103516, "step": 20936 }, { "epoch": 1.650532124556563, "grad_norm": 0.6290438175201416, "learning_rate": 0.0005, "loss": 1.565868854522705, "step": 20937 }, { "epoch": 1.6506109578242019, "grad_norm": 0.5339131951332092, "learning_rate": 0.0005, "loss": 1.3030121326446533, "step": 20938 }, { "epoch": 1.6506897910918408, "grad_norm": 0.5643314719200134, "learning_rate": 0.0005, "loss": 1.3784797191619873, "step": 20939 }, { "epoch": 1.6507686243594797, "grad_norm": 0.6276827454566956, "learning_rate": 0.0005, "loss": 1.5613458156585693, "step": 20940 }, { "epoch": 1.6508474576271186, "grad_norm": 0.5861340761184692, "learning_rate": 0.0005, "loss": 1.3344777822494507, "step": 20941 }, { "epoch": 1.6509262908947577, "grad_norm": 0.5892863869667053, "learning_rate": 0.0005, "loss": 1.5101697444915771, "step": 20942 }, { "epoch": 1.6510051241623964, "grad_norm": 0.6242847442626953, "learning_rate": 0.0005, "loss": 1.47728431224823, "step": 20943 }, { "epoch": 1.6510839574300356, "grad_norm": 0.5280116200447083, "learning_rate": 0.0005, "loss": 1.4044933319091797, "step": 20944 }, { "epoch": 1.6511627906976745, "grad_norm": 0.39980822801589966, "learning_rate": 0.0005, "loss": 1.366534948348999, "step": 20945 }, { "epoch": 1.6512416239653134, "grad_norm": 0.6126689910888672, "learning_rate": 0.0005, "loss": 1.6278774738311768, "step": 20946 }, { "epoch": 1.6513204572329523, "grad_norm": 0.5552807450294495, "learning_rate": 0.0005, "loss": 1.4601393938064575, "step": 20947 }, { "epoch": 1.6513992905005912, "grad_norm": 0.5307433605194092, "learning_rate": 0.0005, "loss": 1.426918625831604, "step": 20948 }, { "epoch": 1.6514781237682303, "grad_norm": 0.5011197328567505, "learning_rate": 0.0005, "loss": 1.5278034210205078, "step": 20949 }, { "epoch": 1.651556957035869, "grad_norm": 0.6163325309753418, "learning_rate": 0.0005, "loss": 1.6504806280136108, "step": 20950 }, { "epoch": 1.6516357903035082, "grad_norm": 0.5644012093544006, "learning_rate": 0.0005, "loss": 1.676493525505066, "step": 20951 }, { "epoch": 1.651714623571147, "grad_norm": 0.5064153671264648, "learning_rate": 0.0005, "loss": 1.4781521558761597, "step": 20952 }, { "epoch": 1.651793456838786, "grad_norm": 0.604226291179657, "learning_rate": 0.0005, "loss": 1.4503616094589233, "step": 20953 }, { "epoch": 1.651872290106425, "grad_norm": 0.6319594383239746, "learning_rate": 0.0005, "loss": 1.7162686586380005, "step": 20954 }, { "epoch": 1.6519511233740638, "grad_norm": 0.6469209790229797, "learning_rate": 0.0005, "loss": 1.580491304397583, "step": 20955 }, { "epoch": 1.652029956641703, "grad_norm": 0.4795662462711334, "learning_rate": 0.0005, "loss": 1.5513596534729004, "step": 20956 }, { "epoch": 1.6521087899093416, "grad_norm": 0.5020750164985657, "learning_rate": 0.0005, "loss": 1.3705475330352783, "step": 20957 }, { "epoch": 1.6521876231769808, "grad_norm": 0.4967027008533478, "learning_rate": 0.0005, "loss": 1.5232644081115723, "step": 20958 }, { "epoch": 1.6522664564446197, "grad_norm": 0.5298828482627869, "learning_rate": 0.0005, "loss": 1.4479806423187256, "step": 20959 }, { "epoch": 1.6523452897122586, "grad_norm": 0.4433194696903229, "learning_rate": 0.0005, "loss": 1.5861682891845703, "step": 20960 }, { "epoch": 1.6524241229798975, "grad_norm": 0.5253267288208008, "learning_rate": 0.0005, "loss": 1.658909797668457, "step": 20961 }, { "epoch": 1.6525029562475364, "grad_norm": 0.5482895374298096, "learning_rate": 0.0005, "loss": 1.4229850769042969, "step": 20962 }, { "epoch": 1.6525817895151755, "grad_norm": 0.5103633403778076, "learning_rate": 0.0005, "loss": 1.597904920578003, "step": 20963 }, { "epoch": 1.6526606227828142, "grad_norm": 0.45444536209106445, "learning_rate": 0.0005, "loss": 1.2812929153442383, "step": 20964 }, { "epoch": 1.6527394560504534, "grad_norm": 0.4115501940250397, "learning_rate": 0.0005, "loss": 1.4332129955291748, "step": 20965 }, { "epoch": 1.6528182893180923, "grad_norm": 0.734294056892395, "learning_rate": 0.0005, "loss": 1.5878067016601562, "step": 20966 }, { "epoch": 1.6528971225857312, "grad_norm": 0.7113077044487, "learning_rate": 0.0005, "loss": 1.6992055177688599, "step": 20967 }, { "epoch": 1.65297595585337, "grad_norm": 0.5345306396484375, "learning_rate": 0.0005, "loss": 1.5135293006896973, "step": 20968 }, { "epoch": 1.653054789121009, "grad_norm": 0.6105307936668396, "learning_rate": 0.0005, "loss": 1.4498536586761475, "step": 20969 }, { "epoch": 1.6531336223886481, "grad_norm": 0.4818789064884186, "learning_rate": 0.0005, "loss": 1.5064826011657715, "step": 20970 }, { "epoch": 1.6532124556562868, "grad_norm": 0.4975793957710266, "learning_rate": 0.0005, "loss": 1.3953654766082764, "step": 20971 }, { "epoch": 1.653291288923926, "grad_norm": 0.5943265557289124, "learning_rate": 0.0005, "loss": 1.6111834049224854, "step": 20972 }, { "epoch": 1.6533701221915649, "grad_norm": 0.4654368758201599, "learning_rate": 0.0005, "loss": 1.2741219997406006, "step": 20973 }, { "epoch": 1.6534489554592038, "grad_norm": 0.4201742708683014, "learning_rate": 0.0005, "loss": 1.3178856372833252, "step": 20974 }, { "epoch": 1.6535277887268427, "grad_norm": 0.7180718183517456, "learning_rate": 0.0005, "loss": 1.414544939994812, "step": 20975 }, { "epoch": 1.6536066219944816, "grad_norm": 0.5071773529052734, "learning_rate": 0.0005, "loss": 1.3684282302856445, "step": 20976 }, { "epoch": 1.6536854552621207, "grad_norm": 0.6413366198539734, "learning_rate": 0.0005, "loss": 1.470966100692749, "step": 20977 }, { "epoch": 1.6537642885297594, "grad_norm": 0.5589358806610107, "learning_rate": 0.0005, "loss": 1.5419429540634155, "step": 20978 }, { "epoch": 1.6538431217973986, "grad_norm": 0.6866450905799866, "learning_rate": 0.0005, "loss": 1.6025251150131226, "step": 20979 }, { "epoch": 1.6539219550650375, "grad_norm": 0.45399296283721924, "learning_rate": 0.0005, "loss": 1.3349339962005615, "step": 20980 }, { "epoch": 1.6540007883326764, "grad_norm": 0.4914950132369995, "learning_rate": 0.0005, "loss": 1.6270954608917236, "step": 20981 }, { "epoch": 1.6540796216003153, "grad_norm": 0.4845762252807617, "learning_rate": 0.0005, "loss": 1.3922693729400635, "step": 20982 }, { "epoch": 1.6541584548679542, "grad_norm": 0.4508427381515503, "learning_rate": 0.0005, "loss": 1.3311328887939453, "step": 20983 }, { "epoch": 1.6542372881355933, "grad_norm": 0.5088620781898499, "learning_rate": 0.0005, "loss": 1.4560015201568604, "step": 20984 }, { "epoch": 1.654316121403232, "grad_norm": 0.5864693522453308, "learning_rate": 0.0005, "loss": 1.7889783382415771, "step": 20985 }, { "epoch": 1.6543949546708712, "grad_norm": 0.591680109500885, "learning_rate": 0.0005, "loss": 1.4193488359451294, "step": 20986 }, { "epoch": 1.65447378793851, "grad_norm": 0.5619969367980957, "learning_rate": 0.0005, "loss": 1.4711909294128418, "step": 20987 }, { "epoch": 1.654552621206149, "grad_norm": 0.465874046087265, "learning_rate": 0.0005, "loss": 1.3999996185302734, "step": 20988 }, { "epoch": 1.654631454473788, "grad_norm": 0.4507821798324585, "learning_rate": 0.0005, "loss": 1.4590578079223633, "step": 20989 }, { "epoch": 1.6547102877414268, "grad_norm": 0.451953262090683, "learning_rate": 0.0005, "loss": 1.4913671016693115, "step": 20990 }, { "epoch": 1.654789121009066, "grad_norm": 0.6731542348861694, "learning_rate": 0.0005, "loss": 1.5458974838256836, "step": 20991 }, { "epoch": 1.6548679542767046, "grad_norm": 0.4566788077354431, "learning_rate": 0.0005, "loss": 1.737136721611023, "step": 20992 }, { "epoch": 1.6549467875443438, "grad_norm": 0.49276331067085266, "learning_rate": 0.0005, "loss": 1.4853698015213013, "step": 20993 }, { "epoch": 1.6550256208119827, "grad_norm": 0.6522365808486938, "learning_rate": 0.0005, "loss": 1.6595187187194824, "step": 20994 }, { "epoch": 1.6551044540796216, "grad_norm": 0.5288601517677307, "learning_rate": 0.0005, "loss": 1.4409589767456055, "step": 20995 }, { "epoch": 1.6551832873472605, "grad_norm": 0.5638149380683899, "learning_rate": 0.0005, "loss": 1.3738564252853394, "step": 20996 }, { "epoch": 1.6552621206148994, "grad_norm": 0.5136705636978149, "learning_rate": 0.0005, "loss": 1.557805061340332, "step": 20997 }, { "epoch": 1.6553409538825385, "grad_norm": 0.5153913497924805, "learning_rate": 0.0005, "loss": 1.5214452743530273, "step": 20998 }, { "epoch": 1.6554197871501772, "grad_norm": 0.5125582814216614, "learning_rate": 0.0005, "loss": 1.4191060066223145, "step": 20999 }, { "epoch": 1.6554986204178164, "grad_norm": 0.4983755052089691, "learning_rate": 0.0005, "loss": 1.4242205619812012, "step": 21000 }, { "epoch": 1.6555774536854553, "grad_norm": 0.4797202944755554, "learning_rate": 0.0005, "loss": 1.3844118118286133, "step": 21001 }, { "epoch": 1.6556562869530942, "grad_norm": 0.448053777217865, "learning_rate": 0.0005, "loss": 1.5437126159667969, "step": 21002 }, { "epoch": 1.655735120220733, "grad_norm": 0.45424526929855347, "learning_rate": 0.0005, "loss": 1.2788722515106201, "step": 21003 }, { "epoch": 1.655813953488372, "grad_norm": 0.5369907021522522, "learning_rate": 0.0005, "loss": 1.3696110248565674, "step": 21004 }, { "epoch": 1.6558927867560111, "grad_norm": 0.4852959215641022, "learning_rate": 0.0005, "loss": 1.56241774559021, "step": 21005 }, { "epoch": 1.6559716200236498, "grad_norm": 0.46816399693489075, "learning_rate": 0.0005, "loss": 1.3107764720916748, "step": 21006 }, { "epoch": 1.656050453291289, "grad_norm": 0.6829035878181458, "learning_rate": 0.0005, "loss": 1.6004263162612915, "step": 21007 }, { "epoch": 1.6561292865589279, "grad_norm": 0.743863046169281, "learning_rate": 0.0005, "loss": 1.6189985275268555, "step": 21008 }, { "epoch": 1.6562081198265668, "grad_norm": 0.5078911185264587, "learning_rate": 0.0005, "loss": 1.466585636138916, "step": 21009 }, { "epoch": 1.6562869530942057, "grad_norm": 1.001049280166626, "learning_rate": 0.0005, "loss": 1.5044448375701904, "step": 21010 }, { "epoch": 1.6563657863618446, "grad_norm": 0.5182727575302124, "learning_rate": 0.0005, "loss": 1.394545316696167, "step": 21011 }, { "epoch": 1.6564446196294837, "grad_norm": 0.5873708724975586, "learning_rate": 0.0005, "loss": 1.243028163909912, "step": 21012 }, { "epoch": 1.6565234528971224, "grad_norm": 0.6003201007843018, "learning_rate": 0.0005, "loss": 1.4364763498306274, "step": 21013 }, { "epoch": 1.6566022861647616, "grad_norm": 0.5287325382232666, "learning_rate": 0.0005, "loss": 1.4062203168869019, "step": 21014 }, { "epoch": 1.6566811194324005, "grad_norm": 0.6400753259658813, "learning_rate": 0.0005, "loss": 1.5785479545593262, "step": 21015 }, { "epoch": 1.6567599527000394, "grad_norm": 0.5102083683013916, "learning_rate": 0.0005, "loss": 1.2503199577331543, "step": 21016 }, { "epoch": 1.6568387859676783, "grad_norm": 0.4986606538295746, "learning_rate": 0.0005, "loss": 1.57731032371521, "step": 21017 }, { "epoch": 1.6569176192353172, "grad_norm": 0.6527727246284485, "learning_rate": 0.0005, "loss": 1.4510091543197632, "step": 21018 }, { "epoch": 1.6569964525029564, "grad_norm": 0.5642832517623901, "learning_rate": 0.0005, "loss": 1.279613733291626, "step": 21019 }, { "epoch": 1.657075285770595, "grad_norm": 0.45486199855804443, "learning_rate": 0.0005, "loss": 1.3220114707946777, "step": 21020 }, { "epoch": 1.6571541190382342, "grad_norm": 0.526772141456604, "learning_rate": 0.0005, "loss": 1.5905640125274658, "step": 21021 }, { "epoch": 1.657232952305873, "grad_norm": 0.6082532405853271, "learning_rate": 0.0005, "loss": 1.4134268760681152, "step": 21022 }, { "epoch": 1.657311785573512, "grad_norm": 0.5425394773483276, "learning_rate": 0.0005, "loss": 1.5455591678619385, "step": 21023 }, { "epoch": 1.6573906188411511, "grad_norm": 0.6720877289772034, "learning_rate": 0.0005, "loss": 1.3773860931396484, "step": 21024 }, { "epoch": 1.6574694521087898, "grad_norm": 0.5081473588943481, "learning_rate": 0.0005, "loss": 1.402604341506958, "step": 21025 }, { "epoch": 1.657548285376429, "grad_norm": 0.5651527047157288, "learning_rate": 0.0005, "loss": 1.78103506565094, "step": 21026 }, { "epoch": 1.6576271186440676, "grad_norm": 0.5140022039413452, "learning_rate": 0.0005, "loss": 1.580637812614441, "step": 21027 }, { "epoch": 1.6577059519117068, "grad_norm": 0.5721357464790344, "learning_rate": 0.0005, "loss": 1.5391316413879395, "step": 21028 }, { "epoch": 1.6577847851793457, "grad_norm": 0.47219955921173096, "learning_rate": 0.0005, "loss": 1.3299472332000732, "step": 21029 }, { "epoch": 1.6578636184469846, "grad_norm": 0.5463356375694275, "learning_rate": 0.0005, "loss": 1.4687080383300781, "step": 21030 }, { "epoch": 1.6579424517146237, "grad_norm": 0.6064199805259705, "learning_rate": 0.0005, "loss": 1.6803053617477417, "step": 21031 }, { "epoch": 1.6580212849822624, "grad_norm": 0.5312573313713074, "learning_rate": 0.0005, "loss": 1.333935260772705, "step": 21032 }, { "epoch": 1.6581001182499016, "grad_norm": 0.638114869594574, "learning_rate": 0.0005, "loss": 1.5960161685943604, "step": 21033 }, { "epoch": 1.6581789515175402, "grad_norm": 0.5470179319381714, "learning_rate": 0.0005, "loss": 1.4906485080718994, "step": 21034 }, { "epoch": 1.6582577847851794, "grad_norm": 0.514458954334259, "learning_rate": 0.0005, "loss": 1.390239953994751, "step": 21035 }, { "epoch": 1.6583366180528183, "grad_norm": 0.4952683448791504, "learning_rate": 0.0005, "loss": 1.5522572994232178, "step": 21036 }, { "epoch": 1.6584154513204572, "grad_norm": 0.5632529258728027, "learning_rate": 0.0005, "loss": 1.5057960748672485, "step": 21037 }, { "epoch": 1.6584942845880963, "grad_norm": 0.6199240684509277, "learning_rate": 0.0005, "loss": 1.6463468074798584, "step": 21038 }, { "epoch": 1.658573117855735, "grad_norm": 0.44624006748199463, "learning_rate": 0.0005, "loss": 1.5314384698867798, "step": 21039 }, { "epoch": 1.6586519511233742, "grad_norm": 0.4802686274051666, "learning_rate": 0.0005, "loss": 1.3728208541870117, "step": 21040 }, { "epoch": 1.6587307843910128, "grad_norm": 0.5909751653671265, "learning_rate": 0.0005, "loss": 1.3870205879211426, "step": 21041 }, { "epoch": 1.658809617658652, "grad_norm": 0.5629855394363403, "learning_rate": 0.0005, "loss": 1.7595373392105103, "step": 21042 }, { "epoch": 1.658888450926291, "grad_norm": 0.612025797367096, "learning_rate": 0.0005, "loss": 1.6065298318862915, "step": 21043 }, { "epoch": 1.6589672841939298, "grad_norm": 0.5591163039207458, "learning_rate": 0.0005, "loss": 1.6432815790176392, "step": 21044 }, { "epoch": 1.659046117461569, "grad_norm": 1.2440885305404663, "learning_rate": 0.0005, "loss": 1.400371789932251, "step": 21045 }, { "epoch": 1.6591249507292076, "grad_norm": 0.6086916923522949, "learning_rate": 0.0005, "loss": 1.680842638015747, "step": 21046 }, { "epoch": 1.6592037839968468, "grad_norm": 0.5709381103515625, "learning_rate": 0.0005, "loss": 1.675325870513916, "step": 21047 }, { "epoch": 1.6592826172644854, "grad_norm": 0.5837227702140808, "learning_rate": 0.0005, "loss": 1.551957130432129, "step": 21048 }, { "epoch": 1.6593614505321246, "grad_norm": 0.5349732637405396, "learning_rate": 0.0005, "loss": 1.6584532260894775, "step": 21049 }, { "epoch": 1.6594402837997635, "grad_norm": 0.484728068113327, "learning_rate": 0.0005, "loss": 1.4397403001785278, "step": 21050 }, { "epoch": 1.6595191170674024, "grad_norm": 0.46050724387168884, "learning_rate": 0.0005, "loss": 1.4708757400512695, "step": 21051 }, { "epoch": 1.6595979503350415, "grad_norm": 0.6777785420417786, "learning_rate": 0.0005, "loss": 1.9402880668640137, "step": 21052 }, { "epoch": 1.6596767836026802, "grad_norm": 0.5612389445304871, "learning_rate": 0.0005, "loss": 1.541993498802185, "step": 21053 }, { "epoch": 1.6597556168703194, "grad_norm": 0.7289721965789795, "learning_rate": 0.0005, "loss": 1.6048731803894043, "step": 21054 }, { "epoch": 1.6598344501379583, "grad_norm": 0.5236691236495972, "learning_rate": 0.0005, "loss": 1.4447126388549805, "step": 21055 }, { "epoch": 1.6599132834055972, "grad_norm": 0.5383650660514832, "learning_rate": 0.0005, "loss": 1.5193333625793457, "step": 21056 }, { "epoch": 1.659992116673236, "grad_norm": 0.5748400688171387, "learning_rate": 0.0005, "loss": 1.3988394737243652, "step": 21057 }, { "epoch": 1.660070949940875, "grad_norm": 0.44421714544296265, "learning_rate": 0.0005, "loss": 1.3249855041503906, "step": 21058 }, { "epoch": 1.6601497832085141, "grad_norm": 0.48334819078445435, "learning_rate": 0.0005, "loss": 1.5544257164001465, "step": 21059 }, { "epoch": 1.6602286164761528, "grad_norm": 0.47919556498527527, "learning_rate": 0.0005, "loss": 1.4329464435577393, "step": 21060 }, { "epoch": 1.660307449743792, "grad_norm": 0.5508739948272705, "learning_rate": 0.0005, "loss": 1.5515297651290894, "step": 21061 }, { "epoch": 1.6603862830114309, "grad_norm": 1.095947504043579, "learning_rate": 0.0005, "loss": 1.800389051437378, "step": 21062 }, { "epoch": 1.6604651162790698, "grad_norm": 0.6613087058067322, "learning_rate": 0.0005, "loss": 1.6821718215942383, "step": 21063 }, { "epoch": 1.6605439495467087, "grad_norm": 0.5461618900299072, "learning_rate": 0.0005, "loss": 1.3730018138885498, "step": 21064 }, { "epoch": 1.6606227828143476, "grad_norm": 0.64578777551651, "learning_rate": 0.0005, "loss": 1.6282460689544678, "step": 21065 }, { "epoch": 1.6607016160819867, "grad_norm": 0.46598756313323975, "learning_rate": 0.0005, "loss": 1.3564176559448242, "step": 21066 }, { "epoch": 1.6607804493496254, "grad_norm": 0.5404032468795776, "learning_rate": 0.0005, "loss": 1.4264802932739258, "step": 21067 }, { "epoch": 1.6608592826172646, "grad_norm": 0.5135000348091125, "learning_rate": 0.0005, "loss": 1.4193365573883057, "step": 21068 }, { "epoch": 1.6609381158849035, "grad_norm": 0.44117996096611023, "learning_rate": 0.0005, "loss": 1.2922163009643555, "step": 21069 }, { "epoch": 1.6610169491525424, "grad_norm": 0.6036137342453003, "learning_rate": 0.0005, "loss": 1.6849958896636963, "step": 21070 }, { "epoch": 1.6610957824201813, "grad_norm": 0.5034080743789673, "learning_rate": 0.0005, "loss": 1.3837354183197021, "step": 21071 }, { "epoch": 1.6611746156878202, "grad_norm": 0.5165024995803833, "learning_rate": 0.0005, "loss": 1.5235947370529175, "step": 21072 }, { "epoch": 1.6612534489554593, "grad_norm": 0.6229649186134338, "learning_rate": 0.0005, "loss": 1.6331896781921387, "step": 21073 }, { "epoch": 1.661332282223098, "grad_norm": 0.52829509973526, "learning_rate": 0.0005, "loss": 1.593611478805542, "step": 21074 }, { "epoch": 1.6614111154907372, "grad_norm": 0.48793503642082214, "learning_rate": 0.0005, "loss": 1.43751859664917, "step": 21075 }, { "epoch": 1.661489948758376, "grad_norm": 0.6003831028938293, "learning_rate": 0.0005, "loss": 1.6514395475387573, "step": 21076 }, { "epoch": 1.661568782026015, "grad_norm": 0.5456095933914185, "learning_rate": 0.0005, "loss": 1.614188551902771, "step": 21077 }, { "epoch": 1.661647615293654, "grad_norm": 0.5228973031044006, "learning_rate": 0.0005, "loss": 1.7073878049850464, "step": 21078 }, { "epoch": 1.6617264485612928, "grad_norm": 0.5669638514518738, "learning_rate": 0.0005, "loss": 1.5030990839004517, "step": 21079 }, { "epoch": 1.661805281828932, "grad_norm": 0.7049926519393921, "learning_rate": 0.0005, "loss": 1.7154595851898193, "step": 21080 }, { "epoch": 1.6618841150965706, "grad_norm": 0.4783031642436981, "learning_rate": 0.0005, "loss": 1.2515895366668701, "step": 21081 }, { "epoch": 1.6619629483642098, "grad_norm": 0.7449193596839905, "learning_rate": 0.0005, "loss": 1.7964550256729126, "step": 21082 }, { "epoch": 1.6620417816318487, "grad_norm": 0.5954854488372803, "learning_rate": 0.0005, "loss": 1.574992060661316, "step": 21083 }, { "epoch": 1.6621206148994876, "grad_norm": 0.5961385369300842, "learning_rate": 0.0005, "loss": 1.5658972263336182, "step": 21084 }, { "epoch": 1.6621994481671265, "grad_norm": 0.49345681071281433, "learning_rate": 0.0005, "loss": 1.5272531509399414, "step": 21085 }, { "epoch": 1.6622782814347654, "grad_norm": 0.5260521173477173, "learning_rate": 0.0005, "loss": 1.4863388538360596, "step": 21086 }, { "epoch": 1.6623571147024045, "grad_norm": 0.6248588562011719, "learning_rate": 0.0005, "loss": 1.5369508266448975, "step": 21087 }, { "epoch": 1.6624359479700432, "grad_norm": 0.46214741468429565, "learning_rate": 0.0005, "loss": 1.4612454175949097, "step": 21088 }, { "epoch": 1.6625147812376824, "grad_norm": 0.6653791069984436, "learning_rate": 0.0005, "loss": 1.3330371379852295, "step": 21089 }, { "epoch": 1.6625936145053213, "grad_norm": 0.6106939911842346, "learning_rate": 0.0005, "loss": 1.864674687385559, "step": 21090 }, { "epoch": 1.6626724477729602, "grad_norm": 0.5461164116859436, "learning_rate": 0.0005, "loss": 1.3252744674682617, "step": 21091 }, { "epoch": 1.662751281040599, "grad_norm": 0.4746450185775757, "learning_rate": 0.0005, "loss": 1.5621956586837769, "step": 21092 }, { "epoch": 1.662830114308238, "grad_norm": 0.5798081159591675, "learning_rate": 0.0005, "loss": 1.6621241569519043, "step": 21093 }, { "epoch": 1.6629089475758771, "grad_norm": 0.41790926456451416, "learning_rate": 0.0005, "loss": 1.4008612632751465, "step": 21094 }, { "epoch": 1.6629877808435158, "grad_norm": 0.46501174569129944, "learning_rate": 0.0005, "loss": 1.4200716018676758, "step": 21095 }, { "epoch": 1.663066614111155, "grad_norm": 0.6396472454071045, "learning_rate": 0.0005, "loss": 1.4486539363861084, "step": 21096 }, { "epoch": 1.6631454473787939, "grad_norm": 0.5347394943237305, "learning_rate": 0.0005, "loss": 1.5507454872131348, "step": 21097 }, { "epoch": 1.6632242806464328, "grad_norm": 0.4275778532028198, "learning_rate": 0.0005, "loss": 1.4042021036148071, "step": 21098 }, { "epoch": 1.6633031139140717, "grad_norm": 0.5594001412391663, "learning_rate": 0.0005, "loss": 1.4800810813903809, "step": 21099 }, { "epoch": 1.6633819471817106, "grad_norm": 0.747820258140564, "learning_rate": 0.0005, "loss": 1.551661491394043, "step": 21100 }, { "epoch": 1.6634607804493498, "grad_norm": 0.5116536617279053, "learning_rate": 0.0005, "loss": 1.5158205032348633, "step": 21101 }, { "epoch": 1.6635396137169884, "grad_norm": 0.4531552493572235, "learning_rate": 0.0005, "loss": 1.3993351459503174, "step": 21102 }, { "epoch": 1.6636184469846276, "grad_norm": 0.6315262913703918, "learning_rate": 0.0005, "loss": 1.6968555450439453, "step": 21103 }, { "epoch": 1.6636972802522665, "grad_norm": 0.6786733269691467, "learning_rate": 0.0005, "loss": 1.6243863105773926, "step": 21104 }, { "epoch": 1.6637761135199054, "grad_norm": 0.45017266273498535, "learning_rate": 0.0005, "loss": 1.4626126289367676, "step": 21105 }, { "epoch": 1.6638549467875443, "grad_norm": 0.6241053342819214, "learning_rate": 0.0005, "loss": 1.5347975492477417, "step": 21106 }, { "epoch": 1.6639337800551832, "grad_norm": 0.4653460681438446, "learning_rate": 0.0005, "loss": 1.5245798826217651, "step": 21107 }, { "epoch": 1.6640126133228224, "grad_norm": 0.5791532397270203, "learning_rate": 0.0005, "loss": 1.6334238052368164, "step": 21108 }, { "epoch": 1.664091446590461, "grad_norm": 0.5632613897323608, "learning_rate": 0.0005, "loss": 1.673478126525879, "step": 21109 }, { "epoch": 1.6641702798581002, "grad_norm": 0.6732578873634338, "learning_rate": 0.0005, "loss": 1.6589868068695068, "step": 21110 }, { "epoch": 1.664249113125739, "grad_norm": 0.6877232193946838, "learning_rate": 0.0005, "loss": 1.6645472049713135, "step": 21111 }, { "epoch": 1.664327946393378, "grad_norm": 0.4200761616230011, "learning_rate": 0.0005, "loss": 1.534942865371704, "step": 21112 }, { "epoch": 1.664406779661017, "grad_norm": 0.5949514508247375, "learning_rate": 0.0005, "loss": 1.4719090461730957, "step": 21113 }, { "epoch": 1.6644856129286558, "grad_norm": 0.43803486227989197, "learning_rate": 0.0005, "loss": 1.508532166481018, "step": 21114 }, { "epoch": 1.664564446196295, "grad_norm": 0.5626775026321411, "learning_rate": 0.0005, "loss": 1.4960672855377197, "step": 21115 }, { "epoch": 1.6646432794639336, "grad_norm": 0.49703359603881836, "learning_rate": 0.0005, "loss": 1.4685494899749756, "step": 21116 }, { "epoch": 1.6647221127315728, "grad_norm": 0.5064552426338196, "learning_rate": 0.0005, "loss": 1.4830243587493896, "step": 21117 }, { "epoch": 1.6648009459992117, "grad_norm": 0.6064276695251465, "learning_rate": 0.0005, "loss": 1.4129421710968018, "step": 21118 }, { "epoch": 1.6648797792668506, "grad_norm": 0.5589470267295837, "learning_rate": 0.0005, "loss": 1.6394299268722534, "step": 21119 }, { "epoch": 1.6649586125344895, "grad_norm": 0.45731163024902344, "learning_rate": 0.0005, "loss": 1.6932042837142944, "step": 21120 }, { "epoch": 1.6650374458021284, "grad_norm": 0.6100581288337708, "learning_rate": 0.0005, "loss": 1.7072784900665283, "step": 21121 }, { "epoch": 1.6651162790697676, "grad_norm": 0.506862998008728, "learning_rate": 0.0005, "loss": 1.552275538444519, "step": 21122 }, { "epoch": 1.6651951123374062, "grad_norm": 0.5051805377006531, "learning_rate": 0.0005, "loss": 1.5852270126342773, "step": 21123 }, { "epoch": 1.6652739456050454, "grad_norm": 0.5587247610092163, "learning_rate": 0.0005, "loss": 1.6407313346862793, "step": 21124 }, { "epoch": 1.6653527788726843, "grad_norm": 0.48363906145095825, "learning_rate": 0.0005, "loss": 1.4764208793640137, "step": 21125 }, { "epoch": 1.6654316121403232, "grad_norm": 0.6549580693244934, "learning_rate": 0.0005, "loss": 1.7618464231491089, "step": 21126 }, { "epoch": 1.6655104454079621, "grad_norm": 0.5448475480079651, "learning_rate": 0.0005, "loss": 1.5775548219680786, "step": 21127 }, { "epoch": 1.665589278675601, "grad_norm": 0.44087228178977966, "learning_rate": 0.0005, "loss": 1.5052196979522705, "step": 21128 }, { "epoch": 1.6656681119432402, "grad_norm": 0.5390545129776001, "learning_rate": 0.0005, "loss": 1.4238371849060059, "step": 21129 }, { "epoch": 1.6657469452108788, "grad_norm": 0.47052785754203796, "learning_rate": 0.0005, "loss": 1.4473036527633667, "step": 21130 }, { "epoch": 1.665825778478518, "grad_norm": 0.44947144389152527, "learning_rate": 0.0005, "loss": 1.410029649734497, "step": 21131 }, { "epoch": 1.665904611746157, "grad_norm": 0.5415597558021545, "learning_rate": 0.0005, "loss": 1.5229597091674805, "step": 21132 }, { "epoch": 1.6659834450137958, "grad_norm": 0.5061737895011902, "learning_rate": 0.0005, "loss": 1.3065829277038574, "step": 21133 }, { "epoch": 1.6660622782814347, "grad_norm": 0.6009261012077332, "learning_rate": 0.0005, "loss": 1.7652488946914673, "step": 21134 }, { "epoch": 1.6661411115490736, "grad_norm": 1.0193018913269043, "learning_rate": 0.0005, "loss": 1.7545032501220703, "step": 21135 }, { "epoch": 1.6662199448167128, "grad_norm": 0.4724634289741516, "learning_rate": 0.0005, "loss": 1.3655178546905518, "step": 21136 }, { "epoch": 1.6662987780843515, "grad_norm": 0.6989250779151917, "learning_rate": 0.0005, "loss": 1.7931957244873047, "step": 21137 }, { "epoch": 1.6663776113519906, "grad_norm": 0.6323803663253784, "learning_rate": 0.0005, "loss": 1.637340784072876, "step": 21138 }, { "epoch": 1.6664564446196295, "grad_norm": 0.5589497685432434, "learning_rate": 0.0005, "loss": 1.4466087818145752, "step": 21139 }, { "epoch": 1.6665352778872684, "grad_norm": 0.47551682591438293, "learning_rate": 0.0005, "loss": 1.5158436298370361, "step": 21140 }, { "epoch": 1.6666141111549073, "grad_norm": 0.5232344269752502, "learning_rate": 0.0005, "loss": 1.435910940170288, "step": 21141 }, { "epoch": 1.6666929444225462, "grad_norm": 0.5481008291244507, "learning_rate": 0.0005, "loss": 1.5104597806930542, "step": 21142 }, { "epoch": 1.6667717776901854, "grad_norm": 0.609245240688324, "learning_rate": 0.0005, "loss": 1.5461989641189575, "step": 21143 }, { "epoch": 1.666850610957824, "grad_norm": 0.48295682668685913, "learning_rate": 0.0005, "loss": 1.6233381032943726, "step": 21144 }, { "epoch": 1.6669294442254632, "grad_norm": 0.4912166893482208, "learning_rate": 0.0005, "loss": 1.4881561994552612, "step": 21145 }, { "epoch": 1.667008277493102, "grad_norm": 0.4559948146343231, "learning_rate": 0.0005, "loss": 1.4789785146713257, "step": 21146 }, { "epoch": 1.667087110760741, "grad_norm": 0.5092375874519348, "learning_rate": 0.0005, "loss": 1.4975467920303345, "step": 21147 }, { "epoch": 1.6671659440283801, "grad_norm": 0.5398054122924805, "learning_rate": 0.0005, "loss": 1.5418343544006348, "step": 21148 }, { "epoch": 1.6672447772960188, "grad_norm": 0.5284707546234131, "learning_rate": 0.0005, "loss": 1.748206615447998, "step": 21149 }, { "epoch": 1.667323610563658, "grad_norm": 0.4196009337902069, "learning_rate": 0.0005, "loss": 1.2730953693389893, "step": 21150 }, { "epoch": 1.6674024438312967, "grad_norm": 0.6926175951957703, "learning_rate": 0.0005, "loss": 1.7419235706329346, "step": 21151 }, { "epoch": 1.6674812770989358, "grad_norm": 0.49747738242149353, "learning_rate": 0.0005, "loss": 1.5091654062271118, "step": 21152 }, { "epoch": 1.6675601103665747, "grad_norm": 0.4580635130405426, "learning_rate": 0.0005, "loss": 1.2293550968170166, "step": 21153 }, { "epoch": 1.6676389436342136, "grad_norm": 0.6414307951927185, "learning_rate": 0.0005, "loss": 1.6578848361968994, "step": 21154 }, { "epoch": 1.6677177769018527, "grad_norm": 0.5039505958557129, "learning_rate": 0.0005, "loss": 1.820021390914917, "step": 21155 }, { "epoch": 1.6677966101694914, "grad_norm": 0.6206626892089844, "learning_rate": 0.0005, "loss": 1.608213186264038, "step": 21156 }, { "epoch": 1.6678754434371306, "grad_norm": 0.5533369779586792, "learning_rate": 0.0005, "loss": 1.4922733306884766, "step": 21157 }, { "epoch": 1.6679542767047693, "grad_norm": 0.4954988360404968, "learning_rate": 0.0005, "loss": 1.3748316764831543, "step": 21158 }, { "epoch": 1.6680331099724084, "grad_norm": 0.5889355540275574, "learning_rate": 0.0005, "loss": 1.4528805017471313, "step": 21159 }, { "epoch": 1.6681119432400473, "grad_norm": 0.4254750609397888, "learning_rate": 0.0005, "loss": 1.2753124237060547, "step": 21160 }, { "epoch": 1.6681907765076862, "grad_norm": 0.5030461549758911, "learning_rate": 0.0005, "loss": 1.5467147827148438, "step": 21161 }, { "epoch": 1.6682696097753253, "grad_norm": 0.4291951358318329, "learning_rate": 0.0005, "loss": 1.2854723930358887, "step": 21162 }, { "epoch": 1.668348443042964, "grad_norm": 0.5326120257377625, "learning_rate": 0.0005, "loss": 1.6298316717147827, "step": 21163 }, { "epoch": 1.6684272763106032, "grad_norm": 0.5477397441864014, "learning_rate": 0.0005, "loss": 1.5085868835449219, "step": 21164 }, { "epoch": 1.6685061095782419, "grad_norm": 0.46156424283981323, "learning_rate": 0.0005, "loss": 1.4000697135925293, "step": 21165 }, { "epoch": 1.668584942845881, "grad_norm": 0.5372523665428162, "learning_rate": 0.0005, "loss": 1.5839271545410156, "step": 21166 }, { "epoch": 1.66866377611352, "grad_norm": 0.6355705261230469, "learning_rate": 0.0005, "loss": 1.476306676864624, "step": 21167 }, { "epoch": 1.6687426093811588, "grad_norm": 0.5082347989082336, "learning_rate": 0.0005, "loss": 1.4292967319488525, "step": 21168 }, { "epoch": 1.668821442648798, "grad_norm": 0.47402405738830566, "learning_rate": 0.0005, "loss": 1.3573689460754395, "step": 21169 }, { "epoch": 1.6689002759164366, "grad_norm": 0.5277413725852966, "learning_rate": 0.0005, "loss": 1.4694101810455322, "step": 21170 }, { "epoch": 1.6689791091840758, "grad_norm": 0.5184693336486816, "learning_rate": 0.0005, "loss": 1.6104001998901367, "step": 21171 }, { "epoch": 1.6690579424517145, "grad_norm": 0.5806632041931152, "learning_rate": 0.0005, "loss": 1.7332000732421875, "step": 21172 }, { "epoch": 1.6691367757193536, "grad_norm": 0.48436102271080017, "learning_rate": 0.0005, "loss": 1.62422776222229, "step": 21173 }, { "epoch": 1.6692156089869925, "grad_norm": 0.5255080461502075, "learning_rate": 0.0005, "loss": 1.6225881576538086, "step": 21174 }, { "epoch": 1.6692944422546314, "grad_norm": 0.4696008861064911, "learning_rate": 0.0005, "loss": 1.503750205039978, "step": 21175 }, { "epoch": 1.6693732755222705, "grad_norm": 0.5288411974906921, "learning_rate": 0.0005, "loss": 1.5268042087554932, "step": 21176 }, { "epoch": 1.6694521087899092, "grad_norm": 0.5393336415290833, "learning_rate": 0.0005, "loss": 1.5777020454406738, "step": 21177 }, { "epoch": 1.6695309420575484, "grad_norm": 0.5859458446502686, "learning_rate": 0.0005, "loss": 1.709320068359375, "step": 21178 }, { "epoch": 1.6696097753251873, "grad_norm": 0.5190134644508362, "learning_rate": 0.0005, "loss": 1.5182360410690308, "step": 21179 }, { "epoch": 1.6696886085928262, "grad_norm": 0.6263392567634583, "learning_rate": 0.0005, "loss": 1.5149248838424683, "step": 21180 }, { "epoch": 1.669767441860465, "grad_norm": 0.513663113117218, "learning_rate": 0.0005, "loss": 1.3378691673278809, "step": 21181 }, { "epoch": 1.669846275128104, "grad_norm": 0.5977299213409424, "learning_rate": 0.0005, "loss": 1.6507683992385864, "step": 21182 }, { "epoch": 1.6699251083957432, "grad_norm": 0.5404443144798279, "learning_rate": 0.0005, "loss": 1.5193145275115967, "step": 21183 }, { "epoch": 1.6700039416633818, "grad_norm": 0.4768325984477997, "learning_rate": 0.0005, "loss": 1.4772207736968994, "step": 21184 }, { "epoch": 1.670082774931021, "grad_norm": 0.6254348754882812, "learning_rate": 0.0005, "loss": 1.6618537902832031, "step": 21185 }, { "epoch": 1.6701616081986599, "grad_norm": 0.5742611289024353, "learning_rate": 0.0005, "loss": 1.5311824083328247, "step": 21186 }, { "epoch": 1.6702404414662988, "grad_norm": 0.691673219203949, "learning_rate": 0.0005, "loss": 1.5744941234588623, "step": 21187 }, { "epoch": 1.6703192747339377, "grad_norm": 0.590212345123291, "learning_rate": 0.0005, "loss": 1.496096134185791, "step": 21188 }, { "epoch": 1.6703981080015766, "grad_norm": 0.5437690615653992, "learning_rate": 0.0005, "loss": 1.8206781148910522, "step": 21189 }, { "epoch": 1.6704769412692158, "grad_norm": 0.6457638144493103, "learning_rate": 0.0005, "loss": 1.5891368389129639, "step": 21190 }, { "epoch": 1.6705557745368544, "grad_norm": 0.5199992060661316, "learning_rate": 0.0005, "loss": 1.6284968852996826, "step": 21191 }, { "epoch": 1.6706346078044936, "grad_norm": 0.5412567853927612, "learning_rate": 0.0005, "loss": 1.7730185985565186, "step": 21192 }, { "epoch": 1.6707134410721325, "grad_norm": 0.5142151117324829, "learning_rate": 0.0005, "loss": 1.504685401916504, "step": 21193 }, { "epoch": 1.6707922743397714, "grad_norm": 0.5848337411880493, "learning_rate": 0.0005, "loss": 1.5891144275665283, "step": 21194 }, { "epoch": 1.6708711076074103, "grad_norm": 0.7597752213478088, "learning_rate": 0.0005, "loss": 1.5176374912261963, "step": 21195 }, { "epoch": 1.6709499408750492, "grad_norm": 0.9014325141906738, "learning_rate": 0.0005, "loss": 1.574663519859314, "step": 21196 }, { "epoch": 1.6710287741426884, "grad_norm": 0.5137558579444885, "learning_rate": 0.0005, "loss": 1.4381897449493408, "step": 21197 }, { "epoch": 1.671107607410327, "grad_norm": 0.6000928282737732, "learning_rate": 0.0005, "loss": 1.5600794553756714, "step": 21198 }, { "epoch": 1.6711864406779662, "grad_norm": 0.8445848822593689, "learning_rate": 0.0005, "loss": 1.549450159072876, "step": 21199 }, { "epoch": 1.671265273945605, "grad_norm": 0.6590767502784729, "learning_rate": 0.0005, "loss": 1.9375308752059937, "step": 21200 }, { "epoch": 1.671344107213244, "grad_norm": 0.5166246891021729, "learning_rate": 0.0005, "loss": 1.2672019004821777, "step": 21201 }, { "epoch": 1.671422940480883, "grad_norm": 0.4965016841888428, "learning_rate": 0.0005, "loss": 1.5659681558609009, "step": 21202 }, { "epoch": 1.6715017737485218, "grad_norm": 1.176421880722046, "learning_rate": 0.0005, "loss": 1.7000880241394043, "step": 21203 }, { "epoch": 1.671580607016161, "grad_norm": 0.6346256732940674, "learning_rate": 0.0005, "loss": 1.6182429790496826, "step": 21204 }, { "epoch": 1.6716594402837996, "grad_norm": 0.49015647172927856, "learning_rate": 0.0005, "loss": 1.4839420318603516, "step": 21205 }, { "epoch": 1.6717382735514388, "grad_norm": 0.6061911582946777, "learning_rate": 0.0005, "loss": 1.5517547130584717, "step": 21206 }, { "epoch": 1.6718171068190777, "grad_norm": 0.5366809368133545, "learning_rate": 0.0005, "loss": 1.7507929801940918, "step": 21207 }, { "epoch": 1.6718959400867166, "grad_norm": 0.5835647583007812, "learning_rate": 0.0005, "loss": 1.4931640625, "step": 21208 }, { "epoch": 1.6719747733543555, "grad_norm": 0.5600824356079102, "learning_rate": 0.0005, "loss": 1.6771867275238037, "step": 21209 }, { "epoch": 1.6720536066219944, "grad_norm": 0.5547491908073425, "learning_rate": 0.0005, "loss": 1.3644859790802002, "step": 21210 }, { "epoch": 1.6721324398896336, "grad_norm": 0.6148784756660461, "learning_rate": 0.0005, "loss": 1.5380818843841553, "step": 21211 }, { "epoch": 1.6722112731572722, "grad_norm": 0.4442363381385803, "learning_rate": 0.0005, "loss": 1.2925293445587158, "step": 21212 }, { "epoch": 1.6722901064249114, "grad_norm": 0.5870134830474854, "learning_rate": 0.0005, "loss": 1.500255823135376, "step": 21213 }, { "epoch": 1.6723689396925503, "grad_norm": 0.8029642105102539, "learning_rate": 0.0005, "loss": 1.7621212005615234, "step": 21214 }, { "epoch": 1.6724477729601892, "grad_norm": 0.460800439119339, "learning_rate": 0.0005, "loss": 1.4889459609985352, "step": 21215 }, { "epoch": 1.6725266062278281, "grad_norm": 0.5348964333534241, "learning_rate": 0.0005, "loss": 1.478711724281311, "step": 21216 }, { "epoch": 1.672605439495467, "grad_norm": 0.611358106136322, "learning_rate": 0.0005, "loss": 1.576977252960205, "step": 21217 }, { "epoch": 1.6726842727631062, "grad_norm": 0.5753032565116882, "learning_rate": 0.0005, "loss": 1.584038496017456, "step": 21218 }, { "epoch": 1.6727631060307449, "grad_norm": 0.4665696322917938, "learning_rate": 0.0005, "loss": 1.5042617321014404, "step": 21219 }, { "epoch": 1.672841939298384, "grad_norm": 0.509811282157898, "learning_rate": 0.0005, "loss": 1.3752844333648682, "step": 21220 }, { "epoch": 1.672920772566023, "grad_norm": 0.53099524974823, "learning_rate": 0.0005, "loss": 1.274451494216919, "step": 21221 }, { "epoch": 1.6729996058336618, "grad_norm": 0.6325508952140808, "learning_rate": 0.0005, "loss": 1.5799357891082764, "step": 21222 }, { "epoch": 1.6730784391013007, "grad_norm": 0.9086959362030029, "learning_rate": 0.0005, "loss": 1.4995546340942383, "step": 21223 }, { "epoch": 1.6731572723689396, "grad_norm": 0.7318281531333923, "learning_rate": 0.0005, "loss": 1.4126813411712646, "step": 21224 }, { "epoch": 1.6732361056365788, "grad_norm": 0.47251957654953003, "learning_rate": 0.0005, "loss": 1.5863043069839478, "step": 21225 }, { "epoch": 1.6733149389042175, "grad_norm": 0.6308983564376831, "learning_rate": 0.0005, "loss": 1.400101661682129, "step": 21226 }, { "epoch": 1.6733937721718566, "grad_norm": 0.5643022656440735, "learning_rate": 0.0005, "loss": 1.4908369779586792, "step": 21227 }, { "epoch": 1.6734726054394955, "grad_norm": 0.4819878339767456, "learning_rate": 0.0005, "loss": 1.434654712677002, "step": 21228 }, { "epoch": 1.6735514387071344, "grad_norm": 0.4937872886657715, "learning_rate": 0.0005, "loss": 1.4650356769561768, "step": 21229 }, { "epoch": 1.6736302719747733, "grad_norm": 0.5787145495414734, "learning_rate": 0.0005, "loss": 1.5769703388214111, "step": 21230 }, { "epoch": 1.6737091052424122, "grad_norm": 0.5072233080863953, "learning_rate": 0.0005, "loss": 1.4936076402664185, "step": 21231 }, { "epoch": 1.6737879385100514, "grad_norm": 0.6122329235076904, "learning_rate": 0.0005, "loss": 1.5694997310638428, "step": 21232 }, { "epoch": 1.67386677177769, "grad_norm": 0.5482499599456787, "learning_rate": 0.0005, "loss": 1.4140247106552124, "step": 21233 }, { "epoch": 1.6739456050453292, "grad_norm": 0.4901953637599945, "learning_rate": 0.0005, "loss": 1.495239019393921, "step": 21234 }, { "epoch": 1.674024438312968, "grad_norm": 0.6410316824913025, "learning_rate": 0.0005, "loss": 1.760533332824707, "step": 21235 }, { "epoch": 1.674103271580607, "grad_norm": 0.555449366569519, "learning_rate": 0.0005, "loss": 1.5816349983215332, "step": 21236 }, { "epoch": 1.674182104848246, "grad_norm": 0.537314772605896, "learning_rate": 0.0005, "loss": 1.4732658863067627, "step": 21237 }, { "epoch": 1.6742609381158848, "grad_norm": 0.5529289841651917, "learning_rate": 0.0005, "loss": 1.4801313877105713, "step": 21238 }, { "epoch": 1.674339771383524, "grad_norm": 0.44961196184158325, "learning_rate": 0.0005, "loss": 1.4142752885818481, "step": 21239 }, { "epoch": 1.6744186046511627, "grad_norm": 0.7616148591041565, "learning_rate": 0.0005, "loss": 1.519324779510498, "step": 21240 }, { "epoch": 1.6744974379188018, "grad_norm": 0.6736546158790588, "learning_rate": 0.0005, "loss": 1.570857048034668, "step": 21241 }, { "epoch": 1.6745762711864407, "grad_norm": 0.46991389989852905, "learning_rate": 0.0005, "loss": 1.6050822734832764, "step": 21242 }, { "epoch": 1.6746551044540796, "grad_norm": 0.5850968360900879, "learning_rate": 0.0005, "loss": 1.6044039726257324, "step": 21243 }, { "epoch": 1.6747339377217185, "grad_norm": 0.5861243605613708, "learning_rate": 0.0005, "loss": 1.5294535160064697, "step": 21244 }, { "epoch": 1.6748127709893574, "grad_norm": 0.595313310623169, "learning_rate": 0.0005, "loss": 1.5351616144180298, "step": 21245 }, { "epoch": 1.6748916042569966, "grad_norm": 0.5380128622055054, "learning_rate": 0.0005, "loss": 1.726198673248291, "step": 21246 }, { "epoch": 1.6749704375246353, "grad_norm": 0.5013042092323303, "learning_rate": 0.0005, "loss": 1.3319423198699951, "step": 21247 }, { "epoch": 1.6750492707922744, "grad_norm": 0.4687642455101013, "learning_rate": 0.0005, "loss": 1.3739845752716064, "step": 21248 }, { "epoch": 1.6751281040599133, "grad_norm": 0.6204565167427063, "learning_rate": 0.0005, "loss": 1.6840498447418213, "step": 21249 }, { "epoch": 1.6752069373275522, "grad_norm": 0.6491953134536743, "learning_rate": 0.0005, "loss": 1.7292988300323486, "step": 21250 }, { "epoch": 1.6752857705951911, "grad_norm": 0.5682840943336487, "learning_rate": 0.0005, "loss": 1.3277556896209717, "step": 21251 }, { "epoch": 1.67536460386283, "grad_norm": 0.6202302575111389, "learning_rate": 0.0005, "loss": 1.4985880851745605, "step": 21252 }, { "epoch": 1.6754434371304692, "grad_norm": 0.4606776237487793, "learning_rate": 0.0005, "loss": 1.4266715049743652, "step": 21253 }, { "epoch": 1.6755222703981079, "grad_norm": 0.5354514718055725, "learning_rate": 0.0005, "loss": 1.6884889602661133, "step": 21254 }, { "epoch": 1.675601103665747, "grad_norm": 0.49274230003356934, "learning_rate": 0.0005, "loss": 1.4777226448059082, "step": 21255 }, { "epoch": 1.675679936933386, "grad_norm": 0.8287373185157776, "learning_rate": 0.0005, "loss": 1.5846965312957764, "step": 21256 }, { "epoch": 1.6757587702010248, "grad_norm": 0.5221323370933533, "learning_rate": 0.0005, "loss": 1.3886709213256836, "step": 21257 }, { "epoch": 1.6758376034686637, "grad_norm": 0.40242770314216614, "learning_rate": 0.0005, "loss": 1.3644936084747314, "step": 21258 }, { "epoch": 1.6759164367363026, "grad_norm": 0.6805283427238464, "learning_rate": 0.0005, "loss": 1.5214247703552246, "step": 21259 }, { "epoch": 1.6759952700039418, "grad_norm": 0.5034233331680298, "learning_rate": 0.0005, "loss": 1.5792875289916992, "step": 21260 }, { "epoch": 1.6760741032715805, "grad_norm": 0.5127204656600952, "learning_rate": 0.0005, "loss": 1.478168249130249, "step": 21261 }, { "epoch": 1.6761529365392196, "grad_norm": 0.4315296411514282, "learning_rate": 0.0005, "loss": 1.3987313508987427, "step": 21262 }, { "epoch": 1.6762317698068585, "grad_norm": 0.6150132417678833, "learning_rate": 0.0005, "loss": 1.6858932971954346, "step": 21263 }, { "epoch": 1.6763106030744974, "grad_norm": 0.4847913682460785, "learning_rate": 0.0005, "loss": 1.3808925151824951, "step": 21264 }, { "epoch": 1.6763894363421363, "grad_norm": 0.6477484703063965, "learning_rate": 0.0005, "loss": 1.5782361030578613, "step": 21265 }, { "epoch": 1.6764682696097752, "grad_norm": 0.4851715564727783, "learning_rate": 0.0005, "loss": 1.5977731943130493, "step": 21266 }, { "epoch": 1.6765471028774144, "grad_norm": 0.4556877613067627, "learning_rate": 0.0005, "loss": 1.5623414516448975, "step": 21267 }, { "epoch": 1.676625936145053, "grad_norm": 0.7109310626983643, "learning_rate": 0.0005, "loss": 1.5674134492874146, "step": 21268 }, { "epoch": 1.6767047694126922, "grad_norm": 0.5042744278907776, "learning_rate": 0.0005, "loss": 1.4871108531951904, "step": 21269 }, { "epoch": 1.676783602680331, "grad_norm": 0.5847841501235962, "learning_rate": 0.0005, "loss": 1.529712200164795, "step": 21270 }, { "epoch": 1.67686243594797, "grad_norm": 0.5727434754371643, "learning_rate": 0.0005, "loss": 1.4691497087478638, "step": 21271 }, { "epoch": 1.676941269215609, "grad_norm": 0.5136074423789978, "learning_rate": 0.0005, "loss": 1.612821340560913, "step": 21272 }, { "epoch": 1.6770201024832478, "grad_norm": 0.5169385075569153, "learning_rate": 0.0005, "loss": 1.5398820638656616, "step": 21273 }, { "epoch": 1.677098935750887, "grad_norm": 0.5027373433113098, "learning_rate": 0.0005, "loss": 1.4785016775131226, "step": 21274 }, { "epoch": 1.6771777690185257, "grad_norm": 0.5392289757728577, "learning_rate": 0.0005, "loss": 1.3633391857147217, "step": 21275 }, { "epoch": 1.6772566022861648, "grad_norm": 0.7371841073036194, "learning_rate": 0.0005, "loss": 1.7332868576049805, "step": 21276 }, { "epoch": 1.6773354355538037, "grad_norm": 0.46456289291381836, "learning_rate": 0.0005, "loss": 1.3995736837387085, "step": 21277 }, { "epoch": 1.6774142688214426, "grad_norm": 0.57439786195755, "learning_rate": 0.0005, "loss": 1.4039244651794434, "step": 21278 }, { "epoch": 1.6774931020890818, "grad_norm": 0.48434972763061523, "learning_rate": 0.0005, "loss": 1.707383155822754, "step": 21279 }, { "epoch": 1.6775719353567204, "grad_norm": 0.41354867815971375, "learning_rate": 0.0005, "loss": 1.5170117616653442, "step": 21280 }, { "epoch": 1.6776507686243596, "grad_norm": 0.5300009846687317, "learning_rate": 0.0005, "loss": 1.4714646339416504, "step": 21281 }, { "epoch": 1.6777296018919983, "grad_norm": 0.46138429641723633, "learning_rate": 0.0005, "loss": 1.309457540512085, "step": 21282 }, { "epoch": 1.6778084351596374, "grad_norm": 0.489078551530838, "learning_rate": 0.0005, "loss": 1.6503081321716309, "step": 21283 }, { "epoch": 1.6778872684272763, "grad_norm": 0.5496011972427368, "learning_rate": 0.0005, "loss": 1.4063540697097778, "step": 21284 }, { "epoch": 1.6779661016949152, "grad_norm": 0.6197680234909058, "learning_rate": 0.0005, "loss": 1.8612582683563232, "step": 21285 }, { "epoch": 1.6780449349625544, "grad_norm": 0.7000594735145569, "learning_rate": 0.0005, "loss": 1.8202691078186035, "step": 21286 }, { "epoch": 1.678123768230193, "grad_norm": 0.5085948705673218, "learning_rate": 0.0005, "loss": 1.5510069131851196, "step": 21287 }, { "epoch": 1.6782026014978322, "grad_norm": 0.6529666185379028, "learning_rate": 0.0005, "loss": 1.5349581241607666, "step": 21288 }, { "epoch": 1.6782814347654709, "grad_norm": 0.4830019176006317, "learning_rate": 0.0005, "loss": 1.5463230609893799, "step": 21289 }, { "epoch": 1.67836026803311, "grad_norm": 0.622183620929718, "learning_rate": 0.0005, "loss": 1.5293285846710205, "step": 21290 }, { "epoch": 1.678439101300749, "grad_norm": 0.4602434039115906, "learning_rate": 0.0005, "loss": 1.5999224185943604, "step": 21291 }, { "epoch": 1.6785179345683878, "grad_norm": 0.4557435214519501, "learning_rate": 0.0005, "loss": 1.5399199724197388, "step": 21292 }, { "epoch": 1.678596767836027, "grad_norm": 0.48750436305999756, "learning_rate": 0.0005, "loss": 1.4298410415649414, "step": 21293 }, { "epoch": 1.6786756011036656, "grad_norm": 0.4179522693157196, "learning_rate": 0.0005, "loss": 1.4231027364730835, "step": 21294 }, { "epoch": 1.6787544343713048, "grad_norm": 0.498384952545166, "learning_rate": 0.0005, "loss": 1.5912537574768066, "step": 21295 }, { "epoch": 1.6788332676389435, "grad_norm": 0.5481845140457153, "learning_rate": 0.0005, "loss": 1.486323356628418, "step": 21296 }, { "epoch": 1.6789121009065826, "grad_norm": 0.4742782413959503, "learning_rate": 0.0005, "loss": 1.4713878631591797, "step": 21297 }, { "epoch": 1.6789909341742215, "grad_norm": 0.3955267667770386, "learning_rate": 0.0005, "loss": 1.2534325122833252, "step": 21298 }, { "epoch": 1.6790697674418604, "grad_norm": 0.3730837404727936, "learning_rate": 0.0005, "loss": 1.4111583232879639, "step": 21299 }, { "epoch": 1.6791486007094996, "grad_norm": 0.41037654876708984, "learning_rate": 0.0005, "loss": 1.4179962873458862, "step": 21300 }, { "epoch": 1.6792274339771383, "grad_norm": 0.44581928849220276, "learning_rate": 0.0005, "loss": 1.310441017150879, "step": 21301 }, { "epoch": 1.6793062672447774, "grad_norm": 0.6189009547233582, "learning_rate": 0.0005, "loss": 1.6130781173706055, "step": 21302 }, { "epoch": 1.679385100512416, "grad_norm": 0.4621475338935852, "learning_rate": 0.0005, "loss": 1.575876235961914, "step": 21303 }, { "epoch": 1.6794639337800552, "grad_norm": 0.47170570492744446, "learning_rate": 0.0005, "loss": 1.4359488487243652, "step": 21304 }, { "epoch": 1.6795427670476941, "grad_norm": 0.7420197129249573, "learning_rate": 0.0005, "loss": 1.605010986328125, "step": 21305 }, { "epoch": 1.679621600315333, "grad_norm": 0.49723801016807556, "learning_rate": 0.0005, "loss": 1.3342807292938232, "step": 21306 }, { "epoch": 1.6797004335829722, "grad_norm": 0.5715844035148621, "learning_rate": 0.0005, "loss": 1.3595142364501953, "step": 21307 }, { "epoch": 1.6797792668506109, "grad_norm": 0.6281555891036987, "learning_rate": 0.0005, "loss": 1.5976581573486328, "step": 21308 }, { "epoch": 1.67985810011825, "grad_norm": 0.5077154636383057, "learning_rate": 0.0005, "loss": 1.499190092086792, "step": 21309 }, { "epoch": 1.679936933385889, "grad_norm": 0.4829657971858978, "learning_rate": 0.0005, "loss": 1.5192975997924805, "step": 21310 }, { "epoch": 1.6800157666535278, "grad_norm": 0.49199429154396057, "learning_rate": 0.0005, "loss": 1.5486071109771729, "step": 21311 }, { "epoch": 1.6800945999211667, "grad_norm": 0.6253074407577515, "learning_rate": 0.0005, "loss": 1.5441606044769287, "step": 21312 }, { "epoch": 1.6801734331888056, "grad_norm": 0.572220504283905, "learning_rate": 0.0005, "loss": 1.4316561222076416, "step": 21313 }, { "epoch": 1.6802522664564448, "grad_norm": 0.4569258987903595, "learning_rate": 0.0005, "loss": 1.3641167879104614, "step": 21314 }, { "epoch": 1.6803310997240835, "grad_norm": 0.7035942077636719, "learning_rate": 0.0005, "loss": 1.5745664834976196, "step": 21315 }, { "epoch": 1.6804099329917226, "grad_norm": 0.513917088508606, "learning_rate": 0.0005, "loss": 1.6294779777526855, "step": 21316 }, { "epoch": 1.6804887662593615, "grad_norm": 0.650382399559021, "learning_rate": 0.0005, "loss": 1.6827571392059326, "step": 21317 }, { "epoch": 1.6805675995270004, "grad_norm": 0.5252799391746521, "learning_rate": 0.0005, "loss": 1.4549827575683594, "step": 21318 }, { "epoch": 1.6806464327946393, "grad_norm": 0.5868139266967773, "learning_rate": 0.0005, "loss": 1.4781185388565063, "step": 21319 }, { "epoch": 1.6807252660622782, "grad_norm": 0.5917463302612305, "learning_rate": 0.0005, "loss": 1.5409436225891113, "step": 21320 }, { "epoch": 1.6808040993299174, "grad_norm": 0.5671718120574951, "learning_rate": 0.0005, "loss": 1.384865403175354, "step": 21321 }, { "epoch": 1.680882932597556, "grad_norm": 0.5836434364318848, "learning_rate": 0.0005, "loss": 1.6100287437438965, "step": 21322 }, { "epoch": 1.6809617658651952, "grad_norm": 0.5596973896026611, "learning_rate": 0.0005, "loss": 1.4141931533813477, "step": 21323 }, { "epoch": 1.681040599132834, "grad_norm": 0.6407793760299683, "learning_rate": 0.0005, "loss": 1.4184720516204834, "step": 21324 }, { "epoch": 1.681119432400473, "grad_norm": 0.47218313813209534, "learning_rate": 0.0005, "loss": 1.4048826694488525, "step": 21325 }, { "epoch": 1.681198265668112, "grad_norm": 0.553570032119751, "learning_rate": 0.0005, "loss": 1.4152801036834717, "step": 21326 }, { "epoch": 1.6812770989357508, "grad_norm": 0.48543664813041687, "learning_rate": 0.0005, "loss": 1.6117076873779297, "step": 21327 }, { "epoch": 1.68135593220339, "grad_norm": 0.7521994709968567, "learning_rate": 0.0005, "loss": 1.4228929281234741, "step": 21328 }, { "epoch": 1.6814347654710287, "grad_norm": 0.6019423604011536, "learning_rate": 0.0005, "loss": 1.5757442712783813, "step": 21329 }, { "epoch": 1.6815135987386678, "grad_norm": 0.709175169467926, "learning_rate": 0.0005, "loss": 1.4802300930023193, "step": 21330 }, { "epoch": 1.6815924320063067, "grad_norm": 0.4759320020675659, "learning_rate": 0.0005, "loss": 1.2608678340911865, "step": 21331 }, { "epoch": 1.6816712652739456, "grad_norm": 0.6939142942428589, "learning_rate": 0.0005, "loss": 1.6021322011947632, "step": 21332 }, { "epoch": 1.6817500985415845, "grad_norm": 0.5419148802757263, "learning_rate": 0.0005, "loss": 1.5334824323654175, "step": 21333 }, { "epoch": 1.6818289318092234, "grad_norm": 0.5727614760398865, "learning_rate": 0.0005, "loss": 1.5260825157165527, "step": 21334 }, { "epoch": 1.6819077650768626, "grad_norm": 0.8462558388710022, "learning_rate": 0.0005, "loss": 1.4292200803756714, "step": 21335 }, { "epoch": 1.6819865983445013, "grad_norm": 0.5843731164932251, "learning_rate": 0.0005, "loss": 1.5262062549591064, "step": 21336 }, { "epoch": 1.6820654316121404, "grad_norm": 0.5110996961593628, "learning_rate": 0.0005, "loss": 1.5881869792938232, "step": 21337 }, { "epoch": 1.6821442648797793, "grad_norm": 0.6692005395889282, "learning_rate": 0.0005, "loss": 1.5580651760101318, "step": 21338 }, { "epoch": 1.6822230981474182, "grad_norm": 0.5132883191108704, "learning_rate": 0.0005, "loss": 1.4887785911560059, "step": 21339 }, { "epoch": 1.6823019314150571, "grad_norm": 0.6648879647254944, "learning_rate": 0.0005, "loss": 1.4056551456451416, "step": 21340 }, { "epoch": 1.682380764682696, "grad_norm": 0.5515730977058411, "learning_rate": 0.0005, "loss": 1.6216274499893188, "step": 21341 }, { "epoch": 1.6824595979503352, "grad_norm": 0.4930439591407776, "learning_rate": 0.0005, "loss": 1.5235662460327148, "step": 21342 }, { "epoch": 1.6825384312179739, "grad_norm": 0.5636270046234131, "learning_rate": 0.0005, "loss": 1.407762050628662, "step": 21343 }, { "epoch": 1.682617264485613, "grad_norm": 0.5610216856002808, "learning_rate": 0.0005, "loss": 1.4738593101501465, "step": 21344 }, { "epoch": 1.682696097753252, "grad_norm": 0.5048080086708069, "learning_rate": 0.0005, "loss": 1.4919252395629883, "step": 21345 }, { "epoch": 1.6827749310208908, "grad_norm": 1.4792906045913696, "learning_rate": 0.0005, "loss": 1.6161582469940186, "step": 21346 }, { "epoch": 1.6828537642885297, "grad_norm": 0.4734300374984741, "learning_rate": 0.0005, "loss": 1.66582453250885, "step": 21347 }, { "epoch": 1.6829325975561686, "grad_norm": 0.4831835925579071, "learning_rate": 0.0005, "loss": 1.2456371784210205, "step": 21348 }, { "epoch": 1.6830114308238078, "grad_norm": 0.9419171214103699, "learning_rate": 0.0005, "loss": 1.70216703414917, "step": 21349 }, { "epoch": 1.6830902640914465, "grad_norm": 0.6643315553665161, "learning_rate": 0.0005, "loss": 1.6112332344055176, "step": 21350 }, { "epoch": 1.6831690973590856, "grad_norm": 0.4823569655418396, "learning_rate": 0.0005, "loss": 1.386343002319336, "step": 21351 }, { "epoch": 1.6832479306267245, "grad_norm": 0.5716826915740967, "learning_rate": 0.0005, "loss": 1.6117794513702393, "step": 21352 }, { "epoch": 1.6833267638943634, "grad_norm": 0.7559685111045837, "learning_rate": 0.0005, "loss": 1.8366122245788574, "step": 21353 }, { "epoch": 1.6834055971620023, "grad_norm": 0.6202883720397949, "learning_rate": 0.0005, "loss": 1.5935653448104858, "step": 21354 }, { "epoch": 1.6834844304296412, "grad_norm": 0.6558135747909546, "learning_rate": 0.0005, "loss": 1.4560465812683105, "step": 21355 }, { "epoch": 1.6835632636972804, "grad_norm": 0.5550829172134399, "learning_rate": 0.0005, "loss": 1.4175786972045898, "step": 21356 }, { "epoch": 1.683642096964919, "grad_norm": 0.5966548919677734, "learning_rate": 0.0005, "loss": 1.3875880241394043, "step": 21357 }, { "epoch": 1.6837209302325582, "grad_norm": 0.5319175720214844, "learning_rate": 0.0005, "loss": 1.4612782001495361, "step": 21358 }, { "epoch": 1.683799763500197, "grad_norm": 1.5611854791641235, "learning_rate": 0.0005, "loss": 1.5584588050842285, "step": 21359 }, { "epoch": 1.683878596767836, "grad_norm": 0.7237808108329773, "learning_rate": 0.0005, "loss": 1.5840904712677002, "step": 21360 }, { "epoch": 1.683957430035475, "grad_norm": 0.5938839316368103, "learning_rate": 0.0005, "loss": 1.5910937786102295, "step": 21361 }, { "epoch": 1.6840362633031138, "grad_norm": 0.5695095062255859, "learning_rate": 0.0005, "loss": 1.5655772686004639, "step": 21362 }, { "epoch": 1.684115096570753, "grad_norm": 0.6204985976219177, "learning_rate": 0.0005, "loss": 1.6768629550933838, "step": 21363 }, { "epoch": 1.6841939298383917, "grad_norm": 0.628274142742157, "learning_rate": 0.0005, "loss": 1.6411411762237549, "step": 21364 }, { "epoch": 1.6842727631060308, "grad_norm": 0.6181867718696594, "learning_rate": 0.0005, "loss": 1.6387052536010742, "step": 21365 }, { "epoch": 1.6843515963736697, "grad_norm": 0.5031808018684387, "learning_rate": 0.0005, "loss": 1.331203818321228, "step": 21366 }, { "epoch": 1.6844304296413086, "grad_norm": 0.4812154769897461, "learning_rate": 0.0005, "loss": 1.305384635925293, "step": 21367 }, { "epoch": 1.6845092629089475, "grad_norm": 0.5729302763938904, "learning_rate": 0.0005, "loss": 1.8436212539672852, "step": 21368 }, { "epoch": 1.6845880961765864, "grad_norm": 0.5805495977401733, "learning_rate": 0.0005, "loss": 1.511937141418457, "step": 21369 }, { "epoch": 1.6846669294442256, "grad_norm": 0.7339022755622864, "learning_rate": 0.0005, "loss": 1.6019282341003418, "step": 21370 }, { "epoch": 1.6847457627118643, "grad_norm": 0.5200239419937134, "learning_rate": 0.0005, "loss": 1.4607940912246704, "step": 21371 }, { "epoch": 1.6848245959795034, "grad_norm": 0.4769006669521332, "learning_rate": 0.0005, "loss": 1.4039386510849, "step": 21372 }, { "epoch": 1.6849034292471423, "grad_norm": 0.6097788214683533, "learning_rate": 0.0005, "loss": 1.5287940502166748, "step": 21373 }, { "epoch": 1.6849822625147812, "grad_norm": 0.6516484022140503, "learning_rate": 0.0005, "loss": 1.7291733026504517, "step": 21374 }, { "epoch": 1.6850610957824201, "grad_norm": 0.5642214417457581, "learning_rate": 0.0005, "loss": 1.5850497484207153, "step": 21375 }, { "epoch": 1.685139929050059, "grad_norm": 0.486103355884552, "learning_rate": 0.0005, "loss": 1.481713891029358, "step": 21376 }, { "epoch": 1.6852187623176982, "grad_norm": 3.0755538940429688, "learning_rate": 0.0005, "loss": 1.6566057205200195, "step": 21377 }, { "epoch": 1.6852975955853369, "grad_norm": 0.49421873688697815, "learning_rate": 0.0005, "loss": 1.3438315391540527, "step": 21378 }, { "epoch": 1.685376428852976, "grad_norm": 0.6059948801994324, "learning_rate": 0.0005, "loss": 1.6828768253326416, "step": 21379 }, { "epoch": 1.685455262120615, "grad_norm": 0.5069456696510315, "learning_rate": 0.0005, "loss": 1.4677983522415161, "step": 21380 }, { "epoch": 1.6855340953882538, "grad_norm": 0.47593072056770325, "learning_rate": 0.0005, "loss": 1.4947786331176758, "step": 21381 }, { "epoch": 1.6856129286558927, "grad_norm": 0.53012615442276, "learning_rate": 0.0005, "loss": 1.3400487899780273, "step": 21382 }, { "epoch": 1.6856917619235317, "grad_norm": 0.4621303379535675, "learning_rate": 0.0005, "loss": 1.4268646240234375, "step": 21383 }, { "epoch": 1.6857705951911708, "grad_norm": 0.6327840089797974, "learning_rate": 0.0005, "loss": 1.484337329864502, "step": 21384 }, { "epoch": 1.6858494284588095, "grad_norm": 0.7647790908813477, "learning_rate": 0.0005, "loss": 1.5029194355010986, "step": 21385 }, { "epoch": 1.6859282617264486, "grad_norm": 0.4265764057636261, "learning_rate": 0.0005, "loss": 1.5299301147460938, "step": 21386 }, { "epoch": 1.6860070949940875, "grad_norm": 0.5361238121986389, "learning_rate": 0.0005, "loss": 1.4942188262939453, "step": 21387 }, { "epoch": 1.6860859282617264, "grad_norm": 0.5589739680290222, "learning_rate": 0.0005, "loss": 1.5134758949279785, "step": 21388 }, { "epoch": 1.6861647615293653, "grad_norm": 0.7337017059326172, "learning_rate": 0.0005, "loss": 1.7494137287139893, "step": 21389 }, { "epoch": 1.6862435947970043, "grad_norm": 0.5610273480415344, "learning_rate": 0.0005, "loss": 1.5416059494018555, "step": 21390 }, { "epoch": 1.6863224280646434, "grad_norm": 0.44779691100120544, "learning_rate": 0.0005, "loss": 1.5295374393463135, "step": 21391 }, { "epoch": 1.686401261332282, "grad_norm": 0.6696111559867859, "learning_rate": 0.0005, "loss": 1.4613025188446045, "step": 21392 }, { "epoch": 1.6864800945999212, "grad_norm": 0.605413556098938, "learning_rate": 0.0005, "loss": 1.6400039196014404, "step": 21393 }, { "epoch": 1.6865589278675601, "grad_norm": 0.48569321632385254, "learning_rate": 0.0005, "loss": 1.4764049053192139, "step": 21394 }, { "epoch": 1.686637761135199, "grad_norm": 0.5420471429824829, "learning_rate": 0.0005, "loss": 1.4175195693969727, "step": 21395 }, { "epoch": 1.686716594402838, "grad_norm": 0.5570815801620483, "learning_rate": 0.0005, "loss": 1.499018907546997, "step": 21396 }, { "epoch": 1.6867954276704769, "grad_norm": 0.5236347913742065, "learning_rate": 0.0005, "loss": 1.460977554321289, "step": 21397 }, { "epoch": 1.686874260938116, "grad_norm": 0.6051521301269531, "learning_rate": 0.0005, "loss": 1.6737945079803467, "step": 21398 }, { "epoch": 1.6869530942057547, "grad_norm": 0.6296345591545105, "learning_rate": 0.0005, "loss": 1.4990839958190918, "step": 21399 }, { "epoch": 1.6870319274733938, "grad_norm": 0.7399601936340332, "learning_rate": 0.0005, "loss": 1.5823062658309937, "step": 21400 }, { "epoch": 1.6871107607410327, "grad_norm": 0.5173717737197876, "learning_rate": 0.0005, "loss": 1.428317904472351, "step": 21401 }, { "epoch": 1.6871895940086716, "grad_norm": 0.7164467573165894, "learning_rate": 0.0005, "loss": 1.3592381477355957, "step": 21402 }, { "epoch": 1.6872684272763108, "grad_norm": 0.7484819889068604, "learning_rate": 0.0005, "loss": 1.4557664394378662, "step": 21403 }, { "epoch": 1.6873472605439495, "grad_norm": 0.6248862147331238, "learning_rate": 0.0005, "loss": 1.8500972986221313, "step": 21404 }, { "epoch": 1.6874260938115886, "grad_norm": 0.6916636824607849, "learning_rate": 0.0005, "loss": 1.6220965385437012, "step": 21405 }, { "epoch": 1.6875049270792273, "grad_norm": 0.6368805170059204, "learning_rate": 0.0005, "loss": 1.5045692920684814, "step": 21406 }, { "epoch": 1.6875837603468664, "grad_norm": 0.6422517895698547, "learning_rate": 0.0005, "loss": 1.5246206521987915, "step": 21407 }, { "epoch": 1.6876625936145053, "grad_norm": 0.6165899634361267, "learning_rate": 0.0005, "loss": 1.5642023086547852, "step": 21408 }, { "epoch": 1.6877414268821442, "grad_norm": 0.6367220282554626, "learning_rate": 0.0005, "loss": 1.6436529159545898, "step": 21409 }, { "epoch": 1.6878202601497834, "grad_norm": 0.6112971305847168, "learning_rate": 0.0005, "loss": 1.630846619606018, "step": 21410 }, { "epoch": 1.687899093417422, "grad_norm": 0.6170445680618286, "learning_rate": 0.0005, "loss": 1.615311861038208, "step": 21411 }, { "epoch": 1.6879779266850612, "grad_norm": 0.5271315574645996, "learning_rate": 0.0005, "loss": 1.5388169288635254, "step": 21412 }, { "epoch": 1.6880567599526999, "grad_norm": 0.6238420009613037, "learning_rate": 0.0005, "loss": 1.6290698051452637, "step": 21413 }, { "epoch": 1.688135593220339, "grad_norm": 0.5198583602905273, "learning_rate": 0.0005, "loss": 1.4355075359344482, "step": 21414 }, { "epoch": 1.688214426487978, "grad_norm": 0.6008421778678894, "learning_rate": 0.0005, "loss": 1.5344223976135254, "step": 21415 }, { "epoch": 1.6882932597556168, "grad_norm": 1.5994768142700195, "learning_rate": 0.0005, "loss": 1.29441499710083, "step": 21416 }, { "epoch": 1.688372093023256, "grad_norm": 0.5582906007766724, "learning_rate": 0.0005, "loss": 1.7279362678527832, "step": 21417 }, { "epoch": 1.6884509262908947, "grad_norm": 0.6846465468406677, "learning_rate": 0.0005, "loss": 1.6139471530914307, "step": 21418 }, { "epoch": 1.6885297595585338, "grad_norm": 0.5137640833854675, "learning_rate": 0.0005, "loss": 1.5240076780319214, "step": 21419 }, { "epoch": 1.6886085928261725, "grad_norm": 0.5015140771865845, "learning_rate": 0.0005, "loss": 1.430729627609253, "step": 21420 }, { "epoch": 1.6886874260938116, "grad_norm": 0.46146759390830994, "learning_rate": 0.0005, "loss": 1.3839945793151855, "step": 21421 }, { "epoch": 1.6887662593614505, "grad_norm": 0.7766097784042358, "learning_rate": 0.0005, "loss": 1.6599558591842651, "step": 21422 }, { "epoch": 1.6888450926290894, "grad_norm": 0.5885025262832642, "learning_rate": 0.0005, "loss": 1.7213815450668335, "step": 21423 }, { "epoch": 1.6889239258967286, "grad_norm": 0.5286197662353516, "learning_rate": 0.0005, "loss": 1.429141640663147, "step": 21424 }, { "epoch": 1.6890027591643673, "grad_norm": 0.5698201656341553, "learning_rate": 0.0005, "loss": 1.5445923805236816, "step": 21425 }, { "epoch": 1.6890815924320064, "grad_norm": 0.6041655540466309, "learning_rate": 0.0005, "loss": 1.7248393297195435, "step": 21426 }, { "epoch": 1.689160425699645, "grad_norm": 0.5205870270729065, "learning_rate": 0.0005, "loss": 1.624462366104126, "step": 21427 }, { "epoch": 1.6892392589672842, "grad_norm": 0.7683230638504028, "learning_rate": 0.0005, "loss": 1.6859850883483887, "step": 21428 }, { "epoch": 1.6893180922349231, "grad_norm": 0.6332648396492004, "learning_rate": 0.0005, "loss": 1.632882833480835, "step": 21429 }, { "epoch": 1.689396925502562, "grad_norm": 0.5820958614349365, "learning_rate": 0.0005, "loss": 1.7817445993423462, "step": 21430 }, { "epoch": 1.6894757587702012, "grad_norm": 0.43958038091659546, "learning_rate": 0.0005, "loss": 1.4365347623825073, "step": 21431 }, { "epoch": 1.6895545920378399, "grad_norm": 0.49702000617980957, "learning_rate": 0.0005, "loss": 1.3051338195800781, "step": 21432 }, { "epoch": 1.689633425305479, "grad_norm": 0.9561672210693359, "learning_rate": 0.0005, "loss": 1.5865428447723389, "step": 21433 }, { "epoch": 1.689712258573118, "grad_norm": 0.5078128576278687, "learning_rate": 0.0005, "loss": 1.6867496967315674, "step": 21434 }, { "epoch": 1.6897910918407568, "grad_norm": 0.8866844177246094, "learning_rate": 0.0005, "loss": 1.5659337043762207, "step": 21435 }, { "epoch": 1.6898699251083957, "grad_norm": 0.6107611060142517, "learning_rate": 0.0005, "loss": 1.316454291343689, "step": 21436 }, { "epoch": 1.6899487583760346, "grad_norm": 0.555533766746521, "learning_rate": 0.0005, "loss": 1.517256498336792, "step": 21437 }, { "epoch": 1.6900275916436738, "grad_norm": 0.5587668418884277, "learning_rate": 0.0005, "loss": 1.4505761861801147, "step": 21438 }, { "epoch": 1.6901064249113125, "grad_norm": 0.5428497791290283, "learning_rate": 0.0005, "loss": 1.4997625350952148, "step": 21439 }, { "epoch": 1.6901852581789516, "grad_norm": 0.6097031235694885, "learning_rate": 0.0005, "loss": 1.5768792629241943, "step": 21440 }, { "epoch": 1.6902640914465905, "grad_norm": 0.502083420753479, "learning_rate": 0.0005, "loss": 1.4940241575241089, "step": 21441 }, { "epoch": 1.6903429247142294, "grad_norm": 0.4375523030757904, "learning_rate": 0.0005, "loss": 1.4576904773712158, "step": 21442 }, { "epoch": 1.6904217579818683, "grad_norm": 0.5412420630455017, "learning_rate": 0.0005, "loss": 1.3737072944641113, "step": 21443 }, { "epoch": 1.6905005912495072, "grad_norm": 0.4351121783256531, "learning_rate": 0.0005, "loss": 1.4434255361557007, "step": 21444 }, { "epoch": 1.6905794245171464, "grad_norm": 0.5545802116394043, "learning_rate": 0.0005, "loss": 1.500145673751831, "step": 21445 }, { "epoch": 1.690658257784785, "grad_norm": 0.4511280655860901, "learning_rate": 0.0005, "loss": 1.5598957538604736, "step": 21446 }, { "epoch": 1.6907370910524242, "grad_norm": 0.4656373858451843, "learning_rate": 0.0005, "loss": 1.4850167036056519, "step": 21447 }, { "epoch": 1.6908159243200631, "grad_norm": 0.6413028836250305, "learning_rate": 0.0005, "loss": 1.4252643585205078, "step": 21448 }, { "epoch": 1.690894757587702, "grad_norm": 0.5742338299751282, "learning_rate": 0.0005, "loss": 1.636603832244873, "step": 21449 }, { "epoch": 1.690973590855341, "grad_norm": 0.48444753885269165, "learning_rate": 0.0005, "loss": 1.2952823638916016, "step": 21450 }, { "epoch": 1.6910524241229798, "grad_norm": 0.39741697907447815, "learning_rate": 0.0005, "loss": 1.3128254413604736, "step": 21451 }, { "epoch": 1.691131257390619, "grad_norm": 0.4660680294036865, "learning_rate": 0.0005, "loss": 1.4716556072235107, "step": 21452 }, { "epoch": 1.6912100906582577, "grad_norm": 0.5088121891021729, "learning_rate": 0.0005, "loss": 1.3526558876037598, "step": 21453 }, { "epoch": 1.6912889239258968, "grad_norm": 0.733681321144104, "learning_rate": 0.0005, "loss": 1.4308868646621704, "step": 21454 }, { "epoch": 1.6913677571935357, "grad_norm": 0.43275946378707886, "learning_rate": 0.0005, "loss": 1.5602972507476807, "step": 21455 }, { "epoch": 1.6914465904611746, "grad_norm": 0.4915398359298706, "learning_rate": 0.0005, "loss": 1.307565450668335, "step": 21456 }, { "epoch": 1.6915254237288135, "grad_norm": 0.6873262524604797, "learning_rate": 0.0005, "loss": 1.7841033935546875, "step": 21457 }, { "epoch": 1.6916042569964524, "grad_norm": 0.5610173344612122, "learning_rate": 0.0005, "loss": 1.6337146759033203, "step": 21458 }, { "epoch": 1.6916830902640916, "grad_norm": 0.7924908399581909, "learning_rate": 0.0005, "loss": 1.7359939813613892, "step": 21459 }, { "epoch": 1.6917619235317303, "grad_norm": 0.5755789279937744, "learning_rate": 0.0005, "loss": 1.5550878047943115, "step": 21460 }, { "epoch": 1.6918407567993694, "grad_norm": 0.5512991547584534, "learning_rate": 0.0005, "loss": 1.5189825296401978, "step": 21461 }, { "epoch": 1.6919195900670083, "grad_norm": 0.8472611904144287, "learning_rate": 0.0005, "loss": 1.5289695262908936, "step": 21462 }, { "epoch": 1.6919984233346472, "grad_norm": 0.4907180666923523, "learning_rate": 0.0005, "loss": 1.5464165210723877, "step": 21463 }, { "epoch": 1.6920772566022861, "grad_norm": 0.5785623788833618, "learning_rate": 0.0005, "loss": 1.7495229244232178, "step": 21464 }, { "epoch": 1.692156089869925, "grad_norm": 0.5855929851531982, "learning_rate": 0.0005, "loss": 1.6750690937042236, "step": 21465 }, { "epoch": 1.6922349231375642, "grad_norm": 0.514405369758606, "learning_rate": 0.0005, "loss": 1.4919419288635254, "step": 21466 }, { "epoch": 1.6923137564052029, "grad_norm": 0.549229621887207, "learning_rate": 0.0005, "loss": 1.449223518371582, "step": 21467 }, { "epoch": 1.692392589672842, "grad_norm": 0.4733528792858124, "learning_rate": 0.0005, "loss": 1.492717981338501, "step": 21468 }, { "epoch": 1.692471422940481, "grad_norm": 0.4996286928653717, "learning_rate": 0.0005, "loss": 1.3217753171920776, "step": 21469 }, { "epoch": 1.6925502562081198, "grad_norm": 0.6761807799339294, "learning_rate": 0.0005, "loss": 1.646465539932251, "step": 21470 }, { "epoch": 1.6926290894757587, "grad_norm": 0.5057198405265808, "learning_rate": 0.0005, "loss": 1.4773396253585815, "step": 21471 }, { "epoch": 1.6927079227433977, "grad_norm": 0.4607239365577698, "learning_rate": 0.0005, "loss": 1.5348896980285645, "step": 21472 }, { "epoch": 1.6927867560110368, "grad_norm": 0.5905177593231201, "learning_rate": 0.0005, "loss": 1.4100360870361328, "step": 21473 }, { "epoch": 1.6928655892786755, "grad_norm": 0.6600927710533142, "learning_rate": 0.0005, "loss": 1.5267256498336792, "step": 21474 }, { "epoch": 1.6929444225463146, "grad_norm": 0.4614400565624237, "learning_rate": 0.0005, "loss": 1.4970775842666626, "step": 21475 }, { "epoch": 1.6930232558139535, "grad_norm": 0.4974558651447296, "learning_rate": 0.0005, "loss": 1.322650671005249, "step": 21476 }, { "epoch": 1.6931020890815924, "grad_norm": 0.5476842522621155, "learning_rate": 0.0005, "loss": 1.7041667699813843, "step": 21477 }, { "epoch": 1.6931809223492313, "grad_norm": 0.5236541628837585, "learning_rate": 0.0005, "loss": 1.4998763799667358, "step": 21478 }, { "epoch": 1.6932597556168703, "grad_norm": 0.43800848722457886, "learning_rate": 0.0005, "loss": 1.3062857389450073, "step": 21479 }, { "epoch": 1.6933385888845094, "grad_norm": 0.5316466689109802, "learning_rate": 0.0005, "loss": 1.3868660926818848, "step": 21480 }, { "epoch": 1.693417422152148, "grad_norm": 0.6109251976013184, "learning_rate": 0.0005, "loss": 1.6552025079727173, "step": 21481 }, { "epoch": 1.6934962554197872, "grad_norm": 0.49343472719192505, "learning_rate": 0.0005, "loss": 1.518776535987854, "step": 21482 }, { "epoch": 1.6935750886874261, "grad_norm": 0.5342288613319397, "learning_rate": 0.0005, "loss": 1.5242351293563843, "step": 21483 }, { "epoch": 1.693653921955065, "grad_norm": 0.43715447187423706, "learning_rate": 0.0005, "loss": 1.2527791261672974, "step": 21484 }, { "epoch": 1.693732755222704, "grad_norm": 0.5250473618507385, "learning_rate": 0.0005, "loss": 1.4045171737670898, "step": 21485 }, { "epoch": 1.6938115884903429, "grad_norm": 0.5518880486488342, "learning_rate": 0.0005, "loss": 1.5731322765350342, "step": 21486 }, { "epoch": 1.693890421757982, "grad_norm": 0.4960857033729553, "learning_rate": 0.0005, "loss": 1.6480931043624878, "step": 21487 }, { "epoch": 1.6939692550256207, "grad_norm": 0.47736120223999023, "learning_rate": 0.0005, "loss": 1.490107536315918, "step": 21488 }, { "epoch": 1.6940480882932598, "grad_norm": 0.4092390239238739, "learning_rate": 0.0005, "loss": 1.258798599243164, "step": 21489 }, { "epoch": 1.6941269215608987, "grad_norm": 0.5073920488357544, "learning_rate": 0.0005, "loss": 1.6053780317306519, "step": 21490 }, { "epoch": 1.6942057548285376, "grad_norm": 1.0801042318344116, "learning_rate": 0.0005, "loss": 1.6753435134887695, "step": 21491 }, { "epoch": 1.6942845880961765, "grad_norm": 0.5494601130485535, "learning_rate": 0.0005, "loss": 1.522440791130066, "step": 21492 }, { "epoch": 1.6943634213638155, "grad_norm": 0.45328521728515625, "learning_rate": 0.0005, "loss": 1.389697790145874, "step": 21493 }, { "epoch": 1.6944422546314546, "grad_norm": 0.6382527947425842, "learning_rate": 0.0005, "loss": 1.4502627849578857, "step": 21494 }, { "epoch": 1.6945210878990933, "grad_norm": 0.4280547499656677, "learning_rate": 0.0005, "loss": 1.352444052696228, "step": 21495 }, { "epoch": 1.6945999211667324, "grad_norm": 0.5667036771774292, "learning_rate": 0.0005, "loss": 1.5959253311157227, "step": 21496 }, { "epoch": 1.6946787544343713, "grad_norm": 0.6276058554649353, "learning_rate": 0.0005, "loss": 1.524512529373169, "step": 21497 }, { "epoch": 1.6947575877020102, "grad_norm": 0.5867217183113098, "learning_rate": 0.0005, "loss": 1.5722463130950928, "step": 21498 }, { "epoch": 1.6948364209696491, "grad_norm": 0.5963463187217712, "learning_rate": 0.0005, "loss": 1.5596740245819092, "step": 21499 }, { "epoch": 1.694915254237288, "grad_norm": 0.4427495002746582, "learning_rate": 0.0005, "loss": 1.5540928840637207, "step": 21500 }, { "epoch": 1.6949940875049272, "grad_norm": 0.49471694231033325, "learning_rate": 0.0005, "loss": 1.4702038764953613, "step": 21501 }, { "epoch": 1.6950729207725659, "grad_norm": 0.5200812220573425, "learning_rate": 0.0005, "loss": 1.6042838096618652, "step": 21502 }, { "epoch": 1.695151754040205, "grad_norm": 0.44095203280448914, "learning_rate": 0.0005, "loss": 1.3652063608169556, "step": 21503 }, { "epoch": 1.695230587307844, "grad_norm": 0.49358025193214417, "learning_rate": 0.0005, "loss": 1.5574923753738403, "step": 21504 }, { "epoch": 1.6953094205754828, "grad_norm": 0.642719566822052, "learning_rate": 0.0005, "loss": 1.5915215015411377, "step": 21505 }, { "epoch": 1.6953882538431218, "grad_norm": 0.5913493037223816, "learning_rate": 0.0005, "loss": 1.5765661001205444, "step": 21506 }, { "epoch": 1.6954670871107607, "grad_norm": 0.5703864097595215, "learning_rate": 0.0005, "loss": 1.6901997327804565, "step": 21507 }, { "epoch": 1.6955459203783998, "grad_norm": 0.6332307457923889, "learning_rate": 0.0005, "loss": 1.7833058834075928, "step": 21508 }, { "epoch": 1.6956247536460385, "grad_norm": 0.6581528782844543, "learning_rate": 0.0005, "loss": 1.7301884889602661, "step": 21509 }, { "epoch": 1.6957035869136776, "grad_norm": 0.4864603281021118, "learning_rate": 0.0005, "loss": 1.5659838914871216, "step": 21510 }, { "epoch": 1.6957824201813165, "grad_norm": 0.45740076899528503, "learning_rate": 0.0005, "loss": 1.4941192865371704, "step": 21511 }, { "epoch": 1.6958612534489554, "grad_norm": 0.6083017587661743, "learning_rate": 0.0005, "loss": 1.6069750785827637, "step": 21512 }, { "epoch": 1.6959400867165944, "grad_norm": 0.4421718120574951, "learning_rate": 0.0005, "loss": 1.4572479724884033, "step": 21513 }, { "epoch": 1.6960189199842333, "grad_norm": 0.5138692259788513, "learning_rate": 0.0005, "loss": 1.3214843273162842, "step": 21514 }, { "epoch": 1.6960977532518724, "grad_norm": 0.543843150138855, "learning_rate": 0.0005, "loss": 1.5457148551940918, "step": 21515 }, { "epoch": 1.696176586519511, "grad_norm": 0.5633861422538757, "learning_rate": 0.0005, "loss": 1.8441286087036133, "step": 21516 }, { "epoch": 1.6962554197871502, "grad_norm": 0.5642068982124329, "learning_rate": 0.0005, "loss": 1.6143126487731934, "step": 21517 }, { "epoch": 1.6963342530547891, "grad_norm": 0.6920640468597412, "learning_rate": 0.0005, "loss": 1.478647232055664, "step": 21518 }, { "epoch": 1.696413086322428, "grad_norm": 0.49596115946769714, "learning_rate": 0.0005, "loss": 1.6380317211151123, "step": 21519 }, { "epoch": 1.696491919590067, "grad_norm": 0.5376101136207581, "learning_rate": 0.0005, "loss": 1.409696340560913, "step": 21520 }, { "epoch": 1.6965707528577059, "grad_norm": 7.343854904174805, "learning_rate": 0.0005, "loss": 1.7464275360107422, "step": 21521 }, { "epoch": 1.696649586125345, "grad_norm": 0.4427522122859955, "learning_rate": 0.0005, "loss": 1.242173433303833, "step": 21522 }, { "epoch": 1.6967284193929837, "grad_norm": 0.5598273277282715, "learning_rate": 0.0005, "loss": 1.3230255842208862, "step": 21523 }, { "epoch": 1.6968072526606228, "grad_norm": 0.6491459608078003, "learning_rate": 0.0005, "loss": 1.635430097579956, "step": 21524 }, { "epoch": 1.6968860859282617, "grad_norm": 0.4876185655593872, "learning_rate": 0.0005, "loss": 1.359444499015808, "step": 21525 }, { "epoch": 1.6969649191959006, "grad_norm": 0.6339570879936218, "learning_rate": 0.0005, "loss": 1.606996774673462, "step": 21526 }, { "epoch": 1.6970437524635398, "grad_norm": 0.5224868059158325, "learning_rate": 0.0005, "loss": 1.5983774662017822, "step": 21527 }, { "epoch": 1.6971225857311785, "grad_norm": 1.0134574174880981, "learning_rate": 0.0005, "loss": 1.832491159439087, "step": 21528 }, { "epoch": 1.6972014189988176, "grad_norm": 0.5606747269630432, "learning_rate": 0.0005, "loss": 1.5298540592193604, "step": 21529 }, { "epoch": 1.6972802522664563, "grad_norm": 0.6950727105140686, "learning_rate": 0.0005, "loss": 1.4737690687179565, "step": 21530 }, { "epoch": 1.6973590855340954, "grad_norm": 0.6601229310035706, "learning_rate": 0.0005, "loss": 1.8504257202148438, "step": 21531 }, { "epoch": 1.6974379188017343, "grad_norm": 0.5480759739875793, "learning_rate": 0.0005, "loss": 1.4698498249053955, "step": 21532 }, { "epoch": 1.6975167520693732, "grad_norm": 0.5614720582962036, "learning_rate": 0.0005, "loss": 1.5667059421539307, "step": 21533 }, { "epoch": 1.6975955853370124, "grad_norm": 0.5018177032470703, "learning_rate": 0.0005, "loss": 1.5403108596801758, "step": 21534 }, { "epoch": 1.697674418604651, "grad_norm": 0.6987289190292358, "learning_rate": 0.0005, "loss": 1.3485798835754395, "step": 21535 }, { "epoch": 1.6977532518722902, "grad_norm": 0.5357650518417358, "learning_rate": 0.0005, "loss": 1.6660971641540527, "step": 21536 }, { "epoch": 1.697832085139929, "grad_norm": 0.5775173902511597, "learning_rate": 0.0005, "loss": 1.6427710056304932, "step": 21537 }, { "epoch": 1.697910918407568, "grad_norm": 0.5896909832954407, "learning_rate": 0.0005, "loss": 1.551230788230896, "step": 21538 }, { "epoch": 1.697989751675207, "grad_norm": 0.49486157298088074, "learning_rate": 0.0005, "loss": 1.327049970626831, "step": 21539 }, { "epoch": 1.6980685849428458, "grad_norm": 0.535625696182251, "learning_rate": 0.0005, "loss": 1.6097445487976074, "step": 21540 }, { "epoch": 1.698147418210485, "grad_norm": 0.465686559677124, "learning_rate": 0.0005, "loss": 1.5027397871017456, "step": 21541 }, { "epoch": 1.6982262514781237, "grad_norm": 0.581446647644043, "learning_rate": 0.0005, "loss": 1.675684928894043, "step": 21542 }, { "epoch": 1.6983050847457628, "grad_norm": 0.621318519115448, "learning_rate": 0.0005, "loss": 1.7264763116836548, "step": 21543 }, { "epoch": 1.6983839180134015, "grad_norm": 0.6523090600967407, "learning_rate": 0.0005, "loss": 1.6368926763534546, "step": 21544 }, { "epoch": 1.6984627512810406, "grad_norm": 0.5298461318016052, "learning_rate": 0.0005, "loss": 1.463230848312378, "step": 21545 }, { "epoch": 1.6985415845486795, "grad_norm": 0.5468012094497681, "learning_rate": 0.0005, "loss": 1.6558892726898193, "step": 21546 }, { "epoch": 1.6986204178163185, "grad_norm": 0.5921990275382996, "learning_rate": 0.0005, "loss": 1.4541667699813843, "step": 21547 }, { "epoch": 1.6986992510839576, "grad_norm": 0.6555826663970947, "learning_rate": 0.0005, "loss": 1.586613655090332, "step": 21548 }, { "epoch": 1.6987780843515963, "grad_norm": 0.4971165359020233, "learning_rate": 0.0005, "loss": 1.688910961151123, "step": 21549 }, { "epoch": 1.6988569176192354, "grad_norm": 0.5232377052307129, "learning_rate": 0.0005, "loss": 1.4748730659484863, "step": 21550 }, { "epoch": 1.698935750886874, "grad_norm": 0.5087164640426636, "learning_rate": 0.0005, "loss": 1.5298120975494385, "step": 21551 }, { "epoch": 1.6990145841545132, "grad_norm": 0.6864384412765503, "learning_rate": 0.0005, "loss": 1.5339581966400146, "step": 21552 }, { "epoch": 1.6990934174221521, "grad_norm": 0.4864540100097656, "learning_rate": 0.0005, "loss": 1.559251308441162, "step": 21553 }, { "epoch": 1.699172250689791, "grad_norm": 0.5670126676559448, "learning_rate": 0.0005, "loss": 1.490214228630066, "step": 21554 }, { "epoch": 1.6992510839574302, "grad_norm": 0.4942685067653656, "learning_rate": 0.0005, "loss": 1.747978925704956, "step": 21555 }, { "epoch": 1.6993299172250689, "grad_norm": 0.4763447940349579, "learning_rate": 0.0005, "loss": 1.4713276624679565, "step": 21556 }, { "epoch": 1.699408750492708, "grad_norm": 0.6878358721733093, "learning_rate": 0.0005, "loss": 1.5565659999847412, "step": 21557 }, { "epoch": 1.699487583760347, "grad_norm": 0.8212553262710571, "learning_rate": 0.0005, "loss": 1.7148948907852173, "step": 21558 }, { "epoch": 1.6995664170279858, "grad_norm": 0.5876061320304871, "learning_rate": 0.0005, "loss": 1.6810507774353027, "step": 21559 }, { "epoch": 1.6996452502956247, "grad_norm": 0.4293077886104584, "learning_rate": 0.0005, "loss": 1.3397375345230103, "step": 21560 }, { "epoch": 1.6997240835632637, "grad_norm": 0.48181602358818054, "learning_rate": 0.0005, "loss": 1.3904681205749512, "step": 21561 }, { "epoch": 1.6998029168309028, "grad_norm": 0.6221205592155457, "learning_rate": 0.0005, "loss": 1.3783708810806274, "step": 21562 }, { "epoch": 1.6998817500985415, "grad_norm": 0.7335625290870667, "learning_rate": 0.0005, "loss": 1.7708661556243896, "step": 21563 }, { "epoch": 1.6999605833661806, "grad_norm": 0.5559492707252502, "learning_rate": 0.0005, "loss": 1.6771209239959717, "step": 21564 }, { "epoch": 1.7000394166338195, "grad_norm": 0.5654436945915222, "learning_rate": 0.0005, "loss": 1.5858632326126099, "step": 21565 }, { "epoch": 1.7001182499014584, "grad_norm": 0.5477433204650879, "learning_rate": 0.0005, "loss": 1.5564298629760742, "step": 21566 }, { "epoch": 1.7001970831690973, "grad_norm": 0.5427210330963135, "learning_rate": 0.0005, "loss": 1.6784796714782715, "step": 21567 }, { "epoch": 1.7002759164367363, "grad_norm": 0.516532301902771, "learning_rate": 0.0005, "loss": 1.598536729812622, "step": 21568 }, { "epoch": 1.7003547497043754, "grad_norm": 0.5828993320465088, "learning_rate": 0.0005, "loss": 1.617271900177002, "step": 21569 }, { "epoch": 1.700433582972014, "grad_norm": 0.5363432765007019, "learning_rate": 0.0005, "loss": 1.6311134099960327, "step": 21570 }, { "epoch": 1.7005124162396532, "grad_norm": 0.480074942111969, "learning_rate": 0.0005, "loss": 1.4816608428955078, "step": 21571 }, { "epoch": 1.7005912495072921, "grad_norm": 0.5152977705001831, "learning_rate": 0.0005, "loss": 1.4678339958190918, "step": 21572 }, { "epoch": 1.700670082774931, "grad_norm": 0.7611609101295471, "learning_rate": 0.0005, "loss": 1.614933967590332, "step": 21573 }, { "epoch": 1.70074891604257, "grad_norm": 0.6527485847473145, "learning_rate": 0.0005, "loss": 1.4029817581176758, "step": 21574 }, { "epoch": 1.7008277493102089, "grad_norm": 0.550880491733551, "learning_rate": 0.0005, "loss": 1.7154600620269775, "step": 21575 }, { "epoch": 1.700906582577848, "grad_norm": 0.7141940593719482, "learning_rate": 0.0005, "loss": 1.406245231628418, "step": 21576 }, { "epoch": 1.7009854158454867, "grad_norm": 0.5184985399246216, "learning_rate": 0.0005, "loss": 1.5075972080230713, "step": 21577 }, { "epoch": 1.7010642491131258, "grad_norm": 0.5376935005187988, "learning_rate": 0.0005, "loss": 1.6036860942840576, "step": 21578 }, { "epoch": 1.7011430823807647, "grad_norm": 0.5462068319320679, "learning_rate": 0.0005, "loss": 1.2595924139022827, "step": 21579 }, { "epoch": 1.7012219156484036, "grad_norm": 0.6883465051651001, "learning_rate": 0.0005, "loss": 1.6415053606033325, "step": 21580 }, { "epoch": 1.7013007489160425, "grad_norm": 0.5173743367195129, "learning_rate": 0.0005, "loss": 1.6331026554107666, "step": 21581 }, { "epoch": 1.7013795821836815, "grad_norm": 1.1698731184005737, "learning_rate": 0.0005, "loss": 1.6316521167755127, "step": 21582 }, { "epoch": 1.7014584154513206, "grad_norm": 0.505793035030365, "learning_rate": 0.0005, "loss": 1.3942763805389404, "step": 21583 }, { "epoch": 1.7015372487189593, "grad_norm": 0.6193028688430786, "learning_rate": 0.0005, "loss": 1.506166696548462, "step": 21584 }, { "epoch": 1.7016160819865984, "grad_norm": 0.516597330570221, "learning_rate": 0.0005, "loss": 1.6947858333587646, "step": 21585 }, { "epoch": 1.7016949152542373, "grad_norm": 0.5152596235275269, "learning_rate": 0.0005, "loss": 1.5004420280456543, "step": 21586 }, { "epoch": 1.7017737485218762, "grad_norm": 0.680996835231781, "learning_rate": 0.0005, "loss": 1.62503981590271, "step": 21587 }, { "epoch": 1.7018525817895152, "grad_norm": 0.5533725619316101, "learning_rate": 0.0005, "loss": 1.410535216331482, "step": 21588 }, { "epoch": 1.701931415057154, "grad_norm": 0.5612267851829529, "learning_rate": 0.0005, "loss": 1.7556051015853882, "step": 21589 }, { "epoch": 1.7020102483247932, "grad_norm": 0.4797227680683136, "learning_rate": 0.0005, "loss": 1.5751841068267822, "step": 21590 }, { "epoch": 1.7020890815924319, "grad_norm": 0.47555482387542725, "learning_rate": 0.0005, "loss": 1.5927150249481201, "step": 21591 }, { "epoch": 1.702167914860071, "grad_norm": 0.5038221478462219, "learning_rate": 0.0005, "loss": 1.448936939239502, "step": 21592 }, { "epoch": 1.70224674812771, "grad_norm": 0.5007674098014832, "learning_rate": 0.0005, "loss": 1.5844215154647827, "step": 21593 }, { "epoch": 1.7023255813953488, "grad_norm": 0.47154104709625244, "learning_rate": 0.0005, "loss": 1.6787019968032837, "step": 21594 }, { "epoch": 1.7024044146629878, "grad_norm": 0.5665490627288818, "learning_rate": 0.0005, "loss": 1.633817434310913, "step": 21595 }, { "epoch": 1.7024832479306267, "grad_norm": 0.5585203170776367, "learning_rate": 0.0005, "loss": 1.5037364959716797, "step": 21596 }, { "epoch": 1.7025620811982658, "grad_norm": 0.5772924423217773, "learning_rate": 0.0005, "loss": 1.6528129577636719, "step": 21597 }, { "epoch": 1.7026409144659045, "grad_norm": 0.4745276868343353, "learning_rate": 0.0005, "loss": 1.6447186470031738, "step": 21598 }, { "epoch": 1.7027197477335436, "grad_norm": 0.5560197234153748, "learning_rate": 0.0005, "loss": 1.5708746910095215, "step": 21599 }, { "epoch": 1.7027985810011825, "grad_norm": 0.43540075421333313, "learning_rate": 0.0005, "loss": 1.320059895515442, "step": 21600 }, { "epoch": 1.7028774142688214, "grad_norm": 0.5377354621887207, "learning_rate": 0.0005, "loss": 1.579559087753296, "step": 21601 }, { "epoch": 1.7029562475364604, "grad_norm": 0.6264450550079346, "learning_rate": 0.0005, "loss": 1.6592596769332886, "step": 21602 }, { "epoch": 1.7030350808040993, "grad_norm": 0.5046733021736145, "learning_rate": 0.0005, "loss": 1.5071933269500732, "step": 21603 }, { "epoch": 1.7031139140717384, "grad_norm": 0.48499324917793274, "learning_rate": 0.0005, "loss": 1.5004265308380127, "step": 21604 }, { "epoch": 1.703192747339377, "grad_norm": 0.5166761875152588, "learning_rate": 0.0005, "loss": 1.8365671634674072, "step": 21605 }, { "epoch": 1.7032715806070162, "grad_norm": 0.69697505235672, "learning_rate": 0.0005, "loss": 1.7520244121551514, "step": 21606 }, { "epoch": 1.7033504138746551, "grad_norm": 0.5162304639816284, "learning_rate": 0.0005, "loss": 1.3782727718353271, "step": 21607 }, { "epoch": 1.703429247142294, "grad_norm": 0.4629330635070801, "learning_rate": 0.0005, "loss": 1.4977152347564697, "step": 21608 }, { "epoch": 1.703508080409933, "grad_norm": 0.4895473122596741, "learning_rate": 0.0005, "loss": 1.6908705234527588, "step": 21609 }, { "epoch": 1.7035869136775719, "grad_norm": 0.5163699388504028, "learning_rate": 0.0005, "loss": 1.4919264316558838, "step": 21610 }, { "epoch": 1.703665746945211, "grad_norm": 0.5840470194816589, "learning_rate": 0.0005, "loss": 1.6612043380737305, "step": 21611 }, { "epoch": 1.7037445802128497, "grad_norm": 0.49740543961524963, "learning_rate": 0.0005, "loss": 1.6492834091186523, "step": 21612 }, { "epoch": 1.7038234134804888, "grad_norm": 0.49013185501098633, "learning_rate": 0.0005, "loss": 1.488684058189392, "step": 21613 }, { "epoch": 1.7039022467481277, "grad_norm": 0.5456778407096863, "learning_rate": 0.0005, "loss": 1.3529751300811768, "step": 21614 }, { "epoch": 1.7039810800157666, "grad_norm": 0.5066627860069275, "learning_rate": 0.0005, "loss": 1.5448659658432007, "step": 21615 }, { "epoch": 1.7040599132834056, "grad_norm": 0.4861051142215729, "learning_rate": 0.0005, "loss": 1.6786911487579346, "step": 21616 }, { "epoch": 1.7041387465510445, "grad_norm": 0.5129250884056091, "learning_rate": 0.0005, "loss": 1.583373785018921, "step": 21617 }, { "epoch": 1.7042175798186836, "grad_norm": 0.5588913559913635, "learning_rate": 0.0005, "loss": 1.6058990955352783, "step": 21618 }, { "epoch": 1.7042964130863223, "grad_norm": 0.6481613516807556, "learning_rate": 0.0005, "loss": 1.807389259338379, "step": 21619 }, { "epoch": 1.7043752463539614, "grad_norm": 0.520734429359436, "learning_rate": 0.0005, "loss": 1.4553462266921997, "step": 21620 }, { "epoch": 1.7044540796216003, "grad_norm": 0.649677574634552, "learning_rate": 0.0005, "loss": 1.512101173400879, "step": 21621 }, { "epoch": 1.7045329128892392, "grad_norm": 0.6860060691833496, "learning_rate": 0.0005, "loss": 1.6954494714736938, "step": 21622 }, { "epoch": 1.7046117461568782, "grad_norm": 0.7269502282142639, "learning_rate": 0.0005, "loss": 1.779829978942871, "step": 21623 }, { "epoch": 1.704690579424517, "grad_norm": 0.6289247870445251, "learning_rate": 0.0005, "loss": 1.5588860511779785, "step": 21624 }, { "epoch": 1.7047694126921562, "grad_norm": 0.4869358241558075, "learning_rate": 0.0005, "loss": 1.3801381587982178, "step": 21625 }, { "epoch": 1.704848245959795, "grad_norm": 0.44421499967575073, "learning_rate": 0.0005, "loss": 1.3820486068725586, "step": 21626 }, { "epoch": 1.704927079227434, "grad_norm": 0.5641135573387146, "learning_rate": 0.0005, "loss": 1.599919080734253, "step": 21627 }, { "epoch": 1.705005912495073, "grad_norm": 0.7306692004203796, "learning_rate": 0.0005, "loss": 1.5747324228286743, "step": 21628 }, { "epoch": 1.7050847457627119, "grad_norm": 0.6278361082077026, "learning_rate": 0.0005, "loss": 1.673255443572998, "step": 21629 }, { "epoch": 1.7051635790303508, "grad_norm": 0.5528503060340881, "learning_rate": 0.0005, "loss": 1.6543292999267578, "step": 21630 }, { "epoch": 1.7052424122979897, "grad_norm": 0.47013595700263977, "learning_rate": 0.0005, "loss": 1.4940677881240845, "step": 21631 }, { "epoch": 1.7053212455656288, "grad_norm": 0.6365728378295898, "learning_rate": 0.0005, "loss": 1.462310791015625, "step": 21632 }, { "epoch": 1.7054000788332675, "grad_norm": 0.5505436658859253, "learning_rate": 0.0005, "loss": 1.6416401863098145, "step": 21633 }, { "epoch": 1.7054789121009066, "grad_norm": 0.5934075117111206, "learning_rate": 0.0005, "loss": 1.6402053833007812, "step": 21634 }, { "epoch": 1.7055577453685455, "grad_norm": 0.4894122779369354, "learning_rate": 0.0005, "loss": 1.377343773841858, "step": 21635 }, { "epoch": 1.7056365786361845, "grad_norm": 0.5625057220458984, "learning_rate": 0.0005, "loss": 1.5477608442306519, "step": 21636 }, { "epoch": 1.7057154119038234, "grad_norm": 0.8034493923187256, "learning_rate": 0.0005, "loss": 1.5504305362701416, "step": 21637 }, { "epoch": 1.7057942451714623, "grad_norm": 0.4803009629249573, "learning_rate": 0.0005, "loss": 1.3956122398376465, "step": 21638 }, { "epoch": 1.7058730784391014, "grad_norm": 0.5530923008918762, "learning_rate": 0.0005, "loss": 1.5583934783935547, "step": 21639 }, { "epoch": 1.70595191170674, "grad_norm": 0.5365877747535706, "learning_rate": 0.0005, "loss": 1.5408508777618408, "step": 21640 }, { "epoch": 1.7060307449743792, "grad_norm": 0.5695661902427673, "learning_rate": 0.0005, "loss": 1.608806848526001, "step": 21641 }, { "epoch": 1.7061095782420181, "grad_norm": 0.43485012650489807, "learning_rate": 0.0005, "loss": 1.388301134109497, "step": 21642 }, { "epoch": 1.706188411509657, "grad_norm": 0.5008630156517029, "learning_rate": 0.0005, "loss": 1.4861873388290405, "step": 21643 }, { "epoch": 1.706267244777296, "grad_norm": 0.6223931312561035, "learning_rate": 0.0005, "loss": 1.5252199172973633, "step": 21644 }, { "epoch": 1.7063460780449349, "grad_norm": 0.6613342761993408, "learning_rate": 0.0005, "loss": 1.5435304641723633, "step": 21645 }, { "epoch": 1.706424911312574, "grad_norm": 0.5535884499549866, "learning_rate": 0.0005, "loss": 1.4562110900878906, "step": 21646 }, { "epoch": 1.7065037445802127, "grad_norm": 0.5365167856216431, "learning_rate": 0.0005, "loss": 1.5957326889038086, "step": 21647 }, { "epoch": 1.7065825778478518, "grad_norm": 0.5405226349830627, "learning_rate": 0.0005, "loss": 1.485774278640747, "step": 21648 }, { "epoch": 1.7066614111154907, "grad_norm": 0.4059893488883972, "learning_rate": 0.0005, "loss": 1.5238310098648071, "step": 21649 }, { "epoch": 1.7067402443831297, "grad_norm": 0.7549487948417664, "learning_rate": 0.0005, "loss": 1.3456038236618042, "step": 21650 }, { "epoch": 1.7068190776507688, "grad_norm": 0.5241590738296509, "learning_rate": 0.0005, "loss": 1.5961650609970093, "step": 21651 }, { "epoch": 1.7068979109184075, "grad_norm": 0.5759649276733398, "learning_rate": 0.0005, "loss": 1.6453543901443481, "step": 21652 }, { "epoch": 1.7069767441860466, "grad_norm": 0.6377111077308655, "learning_rate": 0.0005, "loss": 1.5882067680358887, "step": 21653 }, { "epoch": 1.7070555774536853, "grad_norm": 0.5902553200721741, "learning_rate": 0.0005, "loss": 1.7322795391082764, "step": 21654 }, { "epoch": 1.7071344107213244, "grad_norm": 0.6725587248802185, "learning_rate": 0.0005, "loss": 1.4678077697753906, "step": 21655 }, { "epoch": 1.7072132439889633, "grad_norm": 0.4604891538619995, "learning_rate": 0.0005, "loss": 1.3459076881408691, "step": 21656 }, { "epoch": 1.7072920772566023, "grad_norm": 0.5024973750114441, "learning_rate": 0.0005, "loss": 1.5988750457763672, "step": 21657 }, { "epoch": 1.7073709105242414, "grad_norm": 0.37590548396110535, "learning_rate": 0.0005, "loss": 1.313313364982605, "step": 21658 }, { "epoch": 1.70744974379188, "grad_norm": 0.4905117154121399, "learning_rate": 0.0005, "loss": 1.5954170227050781, "step": 21659 }, { "epoch": 1.7075285770595192, "grad_norm": 0.5192587375640869, "learning_rate": 0.0005, "loss": 1.4334352016448975, "step": 21660 }, { "epoch": 1.707607410327158, "grad_norm": 0.5554630756378174, "learning_rate": 0.0005, "loss": 1.4233975410461426, "step": 21661 }, { "epoch": 1.707686243594797, "grad_norm": 0.6175593733787537, "learning_rate": 0.0005, "loss": 1.7072384357452393, "step": 21662 }, { "epoch": 1.707765076862436, "grad_norm": 0.46914634108543396, "learning_rate": 0.0005, "loss": 1.5218839645385742, "step": 21663 }, { "epoch": 1.7078439101300749, "grad_norm": 0.4834691286087036, "learning_rate": 0.0005, "loss": 1.2947766780853271, "step": 21664 }, { "epoch": 1.707922743397714, "grad_norm": 0.4415954649448395, "learning_rate": 0.0005, "loss": 1.514359951019287, "step": 21665 }, { "epoch": 1.7080015766653527, "grad_norm": 0.5476226806640625, "learning_rate": 0.0005, "loss": 1.3669666051864624, "step": 21666 }, { "epoch": 1.7080804099329918, "grad_norm": 0.5526743531227112, "learning_rate": 0.0005, "loss": 1.6554744243621826, "step": 21667 }, { "epoch": 1.7081592432006305, "grad_norm": 0.5367377400398254, "learning_rate": 0.0005, "loss": 1.3849079608917236, "step": 21668 }, { "epoch": 1.7082380764682696, "grad_norm": 0.5001732707023621, "learning_rate": 0.0005, "loss": 1.4588415622711182, "step": 21669 }, { "epoch": 1.7083169097359086, "grad_norm": 0.5279959440231323, "learning_rate": 0.0005, "loss": 1.7171307802200317, "step": 21670 }, { "epoch": 1.7083957430035475, "grad_norm": 0.4600473940372467, "learning_rate": 0.0005, "loss": 1.4142497777938843, "step": 21671 }, { "epoch": 1.7084745762711866, "grad_norm": 0.5674562454223633, "learning_rate": 0.0005, "loss": 1.7393431663513184, "step": 21672 }, { "epoch": 1.7085534095388253, "grad_norm": 0.6738004088401794, "learning_rate": 0.0005, "loss": 1.6390767097473145, "step": 21673 }, { "epoch": 1.7086322428064644, "grad_norm": 0.42276066541671753, "learning_rate": 0.0005, "loss": 1.4077792167663574, "step": 21674 }, { "epoch": 1.708711076074103, "grad_norm": 0.5067522525787354, "learning_rate": 0.0005, "loss": 1.5448644161224365, "step": 21675 }, { "epoch": 1.7087899093417422, "grad_norm": 0.44191277027130127, "learning_rate": 0.0005, "loss": 1.4379431009292603, "step": 21676 }, { "epoch": 1.7088687426093812, "grad_norm": 0.5373340845108032, "learning_rate": 0.0005, "loss": 1.6712098121643066, "step": 21677 }, { "epoch": 1.70894757587702, "grad_norm": 0.6438045501708984, "learning_rate": 0.0005, "loss": 1.6671545505523682, "step": 21678 }, { "epoch": 1.7090264091446592, "grad_norm": 0.5328651070594788, "learning_rate": 0.0005, "loss": 1.4437241554260254, "step": 21679 }, { "epoch": 1.7091052424122979, "grad_norm": 0.7860925793647766, "learning_rate": 0.0005, "loss": 1.388302206993103, "step": 21680 }, { "epoch": 1.709184075679937, "grad_norm": 0.6460858583450317, "learning_rate": 0.0005, "loss": 1.339792013168335, "step": 21681 }, { "epoch": 1.709262908947576, "grad_norm": 0.5391345024108887, "learning_rate": 0.0005, "loss": 1.2913352251052856, "step": 21682 }, { "epoch": 1.7093417422152148, "grad_norm": 0.5123593807220459, "learning_rate": 0.0005, "loss": 1.4971586465835571, "step": 21683 }, { "epoch": 1.7094205754828538, "grad_norm": 0.5448681712150574, "learning_rate": 0.0005, "loss": 1.734000325202942, "step": 21684 }, { "epoch": 1.7094994087504927, "grad_norm": 0.6226806640625, "learning_rate": 0.0005, "loss": 1.4916783571243286, "step": 21685 }, { "epoch": 1.7095782420181318, "grad_norm": 0.6314006447792053, "learning_rate": 0.0005, "loss": 1.6592419147491455, "step": 21686 }, { "epoch": 1.7096570752857705, "grad_norm": 0.40773439407348633, "learning_rate": 0.0005, "loss": 1.4058904647827148, "step": 21687 }, { "epoch": 1.7097359085534096, "grad_norm": 0.51808100938797, "learning_rate": 0.0005, "loss": 1.5353715419769287, "step": 21688 }, { "epoch": 1.7098147418210485, "grad_norm": 0.6146659851074219, "learning_rate": 0.0005, "loss": 1.5557692050933838, "step": 21689 }, { "epoch": 1.7098935750886874, "grad_norm": 0.5133965015411377, "learning_rate": 0.0005, "loss": 1.543165922164917, "step": 21690 }, { "epoch": 1.7099724083563264, "grad_norm": 0.46175020933151245, "learning_rate": 0.0005, "loss": 1.3336807489395142, "step": 21691 }, { "epoch": 1.7100512416239653, "grad_norm": 0.6243048310279846, "learning_rate": 0.0005, "loss": 1.4302361011505127, "step": 21692 }, { "epoch": 1.7101300748916044, "grad_norm": 0.4545100927352905, "learning_rate": 0.0005, "loss": 1.57607102394104, "step": 21693 }, { "epoch": 1.710208908159243, "grad_norm": 0.5349515676498413, "learning_rate": 0.0005, "loss": 1.6777665615081787, "step": 21694 }, { "epoch": 1.7102877414268822, "grad_norm": 0.4878886938095093, "learning_rate": 0.0005, "loss": 1.5799167156219482, "step": 21695 }, { "epoch": 1.7103665746945211, "grad_norm": 0.535771369934082, "learning_rate": 0.0005, "loss": 1.3821167945861816, "step": 21696 }, { "epoch": 1.71044540796216, "grad_norm": 0.538075864315033, "learning_rate": 0.0005, "loss": 1.5012460947036743, "step": 21697 }, { "epoch": 1.710524241229799, "grad_norm": 0.6371062994003296, "learning_rate": 0.0005, "loss": 1.734190821647644, "step": 21698 }, { "epoch": 1.7106030744974379, "grad_norm": 0.5713402628898621, "learning_rate": 0.0005, "loss": 1.5442938804626465, "step": 21699 }, { "epoch": 1.710681907765077, "grad_norm": 0.45998910069465637, "learning_rate": 0.0005, "loss": 1.488926887512207, "step": 21700 }, { "epoch": 1.7107607410327157, "grad_norm": 0.49158573150634766, "learning_rate": 0.0005, "loss": 1.3702783584594727, "step": 21701 }, { "epoch": 1.7108395743003548, "grad_norm": 0.5183160901069641, "learning_rate": 0.0005, "loss": 1.6792209148406982, "step": 21702 }, { "epoch": 1.7109184075679937, "grad_norm": 0.513177752494812, "learning_rate": 0.0005, "loss": 1.4373549222946167, "step": 21703 }, { "epoch": 1.7109972408356326, "grad_norm": 0.47615525126457214, "learning_rate": 0.0005, "loss": 1.6368930339813232, "step": 21704 }, { "epoch": 1.7110760741032716, "grad_norm": 0.5393474102020264, "learning_rate": 0.0005, "loss": 1.6102763414382935, "step": 21705 }, { "epoch": 1.7111549073709105, "grad_norm": 0.3832797408103943, "learning_rate": 0.0005, "loss": 1.121454119682312, "step": 21706 }, { "epoch": 1.7112337406385496, "grad_norm": 0.5609638094902039, "learning_rate": 0.0005, "loss": 1.447601556777954, "step": 21707 }, { "epoch": 1.7113125739061883, "grad_norm": 0.6470914483070374, "learning_rate": 0.0005, "loss": 1.7417588233947754, "step": 21708 }, { "epoch": 1.7113914071738274, "grad_norm": 0.4229204058647156, "learning_rate": 0.0005, "loss": 1.438661813735962, "step": 21709 }, { "epoch": 1.7114702404414663, "grad_norm": 0.531406581401825, "learning_rate": 0.0005, "loss": 1.615992546081543, "step": 21710 }, { "epoch": 1.7115490737091053, "grad_norm": 0.46372464299201965, "learning_rate": 0.0005, "loss": 1.5506092309951782, "step": 21711 }, { "epoch": 1.7116279069767442, "grad_norm": 0.4933246374130249, "learning_rate": 0.0005, "loss": 1.5515928268432617, "step": 21712 }, { "epoch": 1.711706740244383, "grad_norm": 0.6088716387748718, "learning_rate": 0.0005, "loss": 1.639503002166748, "step": 21713 }, { "epoch": 1.7117855735120222, "grad_norm": 0.5264164805412292, "learning_rate": 0.0005, "loss": 1.43143630027771, "step": 21714 }, { "epoch": 1.711864406779661, "grad_norm": 0.5254316329956055, "learning_rate": 0.0005, "loss": 1.8002703189849854, "step": 21715 }, { "epoch": 1.7119432400473, "grad_norm": 0.8610274791717529, "learning_rate": 0.0005, "loss": 1.7585668563842773, "step": 21716 }, { "epoch": 1.712022073314939, "grad_norm": 0.47902917861938477, "learning_rate": 0.0005, "loss": 1.4557429552078247, "step": 21717 }, { "epoch": 1.7121009065825779, "grad_norm": 0.6731776595115662, "learning_rate": 0.0005, "loss": 1.6660144329071045, "step": 21718 }, { "epoch": 1.7121797398502168, "grad_norm": 0.6187340021133423, "learning_rate": 0.0005, "loss": 1.5080711841583252, "step": 21719 }, { "epoch": 1.7122585731178557, "grad_norm": 0.5215447545051575, "learning_rate": 0.0005, "loss": 1.3793647289276123, "step": 21720 }, { "epoch": 1.7123374063854948, "grad_norm": 0.5173535943031311, "learning_rate": 0.0005, "loss": 1.6476380825042725, "step": 21721 }, { "epoch": 1.7124162396531335, "grad_norm": 0.5606768131256104, "learning_rate": 0.0005, "loss": 1.6231861114501953, "step": 21722 }, { "epoch": 1.7124950729207726, "grad_norm": 0.7000393271446228, "learning_rate": 0.0005, "loss": 1.606370449066162, "step": 21723 }, { "epoch": 1.7125739061884115, "grad_norm": 0.48143863677978516, "learning_rate": 0.0005, "loss": 1.2832615375518799, "step": 21724 }, { "epoch": 1.7126527394560505, "grad_norm": 0.4978977143764496, "learning_rate": 0.0005, "loss": 1.5226714611053467, "step": 21725 }, { "epoch": 1.7127315727236894, "grad_norm": 0.5702511668205261, "learning_rate": 0.0005, "loss": 1.416264295578003, "step": 21726 }, { "epoch": 1.7128104059913283, "grad_norm": 0.46818119287490845, "learning_rate": 0.0005, "loss": 1.3606665134429932, "step": 21727 }, { "epoch": 1.7128892392589674, "grad_norm": 0.8052366375923157, "learning_rate": 0.0005, "loss": 1.4119787216186523, "step": 21728 }, { "epoch": 1.712968072526606, "grad_norm": 0.6575655341148376, "learning_rate": 0.0005, "loss": 1.3739280700683594, "step": 21729 }, { "epoch": 1.7130469057942452, "grad_norm": 0.522261381149292, "learning_rate": 0.0005, "loss": 1.6382911205291748, "step": 21730 }, { "epoch": 1.7131257390618841, "grad_norm": 0.6015633940696716, "learning_rate": 0.0005, "loss": 1.763721227645874, "step": 21731 }, { "epoch": 1.713204572329523, "grad_norm": 0.47465360164642334, "learning_rate": 0.0005, "loss": 1.7338252067565918, "step": 21732 }, { "epoch": 1.713283405597162, "grad_norm": 0.5171680450439453, "learning_rate": 0.0005, "loss": 1.5041301250457764, "step": 21733 }, { "epoch": 1.7133622388648009, "grad_norm": 0.7514985799789429, "learning_rate": 0.0005, "loss": 1.504667043685913, "step": 21734 }, { "epoch": 1.71344107213244, "grad_norm": 0.7515149116516113, "learning_rate": 0.0005, "loss": 1.3889751434326172, "step": 21735 }, { "epoch": 1.7135199054000787, "grad_norm": 0.49120599031448364, "learning_rate": 0.0005, "loss": 1.4151394367218018, "step": 21736 }, { "epoch": 1.7135987386677178, "grad_norm": 0.5485717058181763, "learning_rate": 0.0005, "loss": 1.3138928413391113, "step": 21737 }, { "epoch": 1.7136775719353567, "grad_norm": 0.7466760873794556, "learning_rate": 0.0005, "loss": 1.440076231956482, "step": 21738 }, { "epoch": 1.7137564052029957, "grad_norm": 0.4731091260910034, "learning_rate": 0.0005, "loss": 1.5841667652130127, "step": 21739 }, { "epoch": 1.7138352384706346, "grad_norm": 0.439907044172287, "learning_rate": 0.0005, "loss": 1.5132205486297607, "step": 21740 }, { "epoch": 1.7139140717382735, "grad_norm": 0.46985551714897156, "learning_rate": 0.0005, "loss": 1.6552374362945557, "step": 21741 }, { "epoch": 1.7139929050059126, "grad_norm": 0.6806699633598328, "learning_rate": 0.0005, "loss": 1.294552206993103, "step": 21742 }, { "epoch": 1.7140717382735513, "grad_norm": 0.6235138177871704, "learning_rate": 0.0005, "loss": 1.462644100189209, "step": 21743 }, { "epoch": 1.7141505715411904, "grad_norm": 0.6067761182785034, "learning_rate": 0.0005, "loss": 1.7611500024795532, "step": 21744 }, { "epoch": 1.7142294048088293, "grad_norm": 0.6037691235542297, "learning_rate": 0.0005, "loss": 1.5357595682144165, "step": 21745 }, { "epoch": 1.7143082380764683, "grad_norm": 0.45106497406959534, "learning_rate": 0.0005, "loss": 1.355652928352356, "step": 21746 }, { "epoch": 1.7143870713441072, "grad_norm": 0.5074778199195862, "learning_rate": 0.0005, "loss": 1.4563772678375244, "step": 21747 }, { "epoch": 1.714465904611746, "grad_norm": 0.45978403091430664, "learning_rate": 0.0005, "loss": 1.3399784564971924, "step": 21748 }, { "epoch": 1.7145447378793852, "grad_norm": 0.5338943600654602, "learning_rate": 0.0005, "loss": 1.5306470394134521, "step": 21749 }, { "epoch": 1.714623571147024, "grad_norm": 0.5802909135818481, "learning_rate": 0.0005, "loss": 1.408853530883789, "step": 21750 }, { "epoch": 1.714702404414663, "grad_norm": 0.5188345313072205, "learning_rate": 0.0005, "loss": 1.55070960521698, "step": 21751 }, { "epoch": 1.714781237682302, "grad_norm": 0.6512863636016846, "learning_rate": 0.0005, "loss": 1.6860238313674927, "step": 21752 }, { "epoch": 1.7148600709499409, "grad_norm": 0.6477066874504089, "learning_rate": 0.0005, "loss": 1.534234642982483, "step": 21753 }, { "epoch": 1.7149389042175798, "grad_norm": 0.5013363361358643, "learning_rate": 0.0005, "loss": 1.5259296894073486, "step": 21754 }, { "epoch": 1.7150177374852187, "grad_norm": 0.5200147032737732, "learning_rate": 0.0005, "loss": 1.5105738639831543, "step": 21755 }, { "epoch": 1.7150965707528578, "grad_norm": 0.5869823694229126, "learning_rate": 0.0005, "loss": 1.6707242727279663, "step": 21756 }, { "epoch": 1.7151754040204965, "grad_norm": 0.6662521958351135, "learning_rate": 0.0005, "loss": 1.5199154615402222, "step": 21757 }, { "epoch": 1.7152542372881356, "grad_norm": 0.5471006035804749, "learning_rate": 0.0005, "loss": 1.3289843797683716, "step": 21758 }, { "epoch": 1.7153330705557746, "grad_norm": 0.7034871578216553, "learning_rate": 0.0005, "loss": 1.6003096103668213, "step": 21759 }, { "epoch": 1.7154119038234135, "grad_norm": 0.4714607894420624, "learning_rate": 0.0005, "loss": 1.5491256713867188, "step": 21760 }, { "epoch": 1.7154907370910524, "grad_norm": 0.5430799722671509, "learning_rate": 0.0005, "loss": 1.8089994192123413, "step": 21761 }, { "epoch": 1.7155695703586913, "grad_norm": 0.808127760887146, "learning_rate": 0.0005, "loss": 1.3377459049224854, "step": 21762 }, { "epoch": 1.7156484036263304, "grad_norm": 0.7095861434936523, "learning_rate": 0.0005, "loss": 1.4467798471450806, "step": 21763 }, { "epoch": 1.715727236893969, "grad_norm": 0.4450950622558594, "learning_rate": 0.0005, "loss": 1.5825629234313965, "step": 21764 }, { "epoch": 1.7158060701616082, "grad_norm": 0.5314921736717224, "learning_rate": 0.0005, "loss": 1.3270819187164307, "step": 21765 }, { "epoch": 1.7158849034292472, "grad_norm": 0.6080847382545471, "learning_rate": 0.0005, "loss": 1.5505359172821045, "step": 21766 }, { "epoch": 1.715963736696886, "grad_norm": 0.6166338920593262, "learning_rate": 0.0005, "loss": 1.6271357536315918, "step": 21767 }, { "epoch": 1.716042569964525, "grad_norm": 0.4898524284362793, "learning_rate": 0.0005, "loss": 1.4150114059448242, "step": 21768 }, { "epoch": 1.7161214032321639, "grad_norm": 0.40077531337738037, "learning_rate": 0.0005, "loss": 1.3925833702087402, "step": 21769 }, { "epoch": 1.716200236499803, "grad_norm": 0.7274437546730042, "learning_rate": 0.0005, "loss": 1.591759443283081, "step": 21770 }, { "epoch": 1.7162790697674417, "grad_norm": 0.6295168995857239, "learning_rate": 0.0005, "loss": 1.422332525253296, "step": 21771 }, { "epoch": 1.7163579030350808, "grad_norm": 0.5456303358078003, "learning_rate": 0.0005, "loss": 1.6388871669769287, "step": 21772 }, { "epoch": 1.7164367363027198, "grad_norm": 0.5687652230262756, "learning_rate": 0.0005, "loss": 1.4573423862457275, "step": 21773 }, { "epoch": 1.7165155695703587, "grad_norm": 0.5724497437477112, "learning_rate": 0.0005, "loss": 1.3324341773986816, "step": 21774 }, { "epoch": 1.7165944028379978, "grad_norm": 0.5333130359649658, "learning_rate": 0.0005, "loss": 1.4447120428085327, "step": 21775 }, { "epoch": 1.7166732361056365, "grad_norm": 0.5138967037200928, "learning_rate": 0.0005, "loss": 1.6376397609710693, "step": 21776 }, { "epoch": 1.7167520693732756, "grad_norm": 0.45730429887771606, "learning_rate": 0.0005, "loss": 1.568865418434143, "step": 21777 }, { "epoch": 1.7168309026409143, "grad_norm": 0.6187118291854858, "learning_rate": 0.0005, "loss": 1.3718324899673462, "step": 21778 }, { "epoch": 1.7169097359085534, "grad_norm": 0.5582715272903442, "learning_rate": 0.0005, "loss": 1.5980441570281982, "step": 21779 }, { "epoch": 1.7169885691761924, "grad_norm": 0.5474666953086853, "learning_rate": 0.0005, "loss": 1.7003587484359741, "step": 21780 }, { "epoch": 1.7170674024438313, "grad_norm": 0.6392870545387268, "learning_rate": 0.0005, "loss": 1.650176763534546, "step": 21781 }, { "epoch": 1.7171462357114704, "grad_norm": 0.5465999841690063, "learning_rate": 0.0005, "loss": 1.6863670349121094, "step": 21782 }, { "epoch": 1.717225068979109, "grad_norm": 0.5290398597717285, "learning_rate": 0.0005, "loss": 1.4591755867004395, "step": 21783 }, { "epoch": 1.7173039022467482, "grad_norm": 0.5421339869499207, "learning_rate": 0.0005, "loss": 1.489201307296753, "step": 21784 }, { "epoch": 1.717382735514387, "grad_norm": 0.8524811267852783, "learning_rate": 0.0005, "loss": 1.379781723022461, "step": 21785 }, { "epoch": 1.717461568782026, "grad_norm": 0.5481410622596741, "learning_rate": 0.0005, "loss": 1.5411889553070068, "step": 21786 }, { "epoch": 1.717540402049665, "grad_norm": 0.5135102868080139, "learning_rate": 0.0005, "loss": 1.6088812351226807, "step": 21787 }, { "epoch": 1.7176192353173039, "grad_norm": 0.49722373485565186, "learning_rate": 0.0005, "loss": 1.56985604763031, "step": 21788 }, { "epoch": 1.717698068584943, "grad_norm": 0.4992399215698242, "learning_rate": 0.0005, "loss": 1.6039615869522095, "step": 21789 }, { "epoch": 1.7177769018525817, "grad_norm": 0.4949343502521515, "learning_rate": 0.0005, "loss": 1.4452564716339111, "step": 21790 }, { "epoch": 1.7178557351202208, "grad_norm": 0.47987157106399536, "learning_rate": 0.0005, "loss": 1.6130361557006836, "step": 21791 }, { "epoch": 1.7179345683878595, "grad_norm": 0.5993808507919312, "learning_rate": 0.0005, "loss": 1.8082224130630493, "step": 21792 }, { "epoch": 1.7180134016554987, "grad_norm": 0.487781822681427, "learning_rate": 0.0005, "loss": 1.4303086996078491, "step": 21793 }, { "epoch": 1.7180922349231376, "grad_norm": 0.6644641757011414, "learning_rate": 0.0005, "loss": 1.5848740339279175, "step": 21794 }, { "epoch": 1.7181710681907765, "grad_norm": 0.5173114538192749, "learning_rate": 0.0005, "loss": 1.3504005670547485, "step": 21795 }, { "epoch": 1.7182499014584156, "grad_norm": 0.5148541927337646, "learning_rate": 0.0005, "loss": 1.4148459434509277, "step": 21796 }, { "epoch": 1.7183287347260543, "grad_norm": 0.5896414518356323, "learning_rate": 0.0005, "loss": 1.7902517318725586, "step": 21797 }, { "epoch": 1.7184075679936934, "grad_norm": 0.730786919593811, "learning_rate": 0.0005, "loss": 1.549825668334961, "step": 21798 }, { "epoch": 1.7184864012613321, "grad_norm": 0.540742814540863, "learning_rate": 0.0005, "loss": 1.6714937686920166, "step": 21799 }, { "epoch": 1.7185652345289713, "grad_norm": 0.5185617804527283, "learning_rate": 0.0005, "loss": 1.4274792671203613, "step": 21800 }, { "epoch": 1.7186440677966102, "grad_norm": 0.5148643851280212, "learning_rate": 0.0005, "loss": 1.4551093578338623, "step": 21801 }, { "epoch": 1.718722901064249, "grad_norm": 0.5068654417991638, "learning_rate": 0.0005, "loss": 1.5403720140457153, "step": 21802 }, { "epoch": 1.7188017343318882, "grad_norm": 0.6228445768356323, "learning_rate": 0.0005, "loss": 1.3712424039840698, "step": 21803 }, { "epoch": 1.718880567599527, "grad_norm": 0.4854722321033478, "learning_rate": 0.0005, "loss": 1.4140372276306152, "step": 21804 }, { "epoch": 1.718959400867166, "grad_norm": 0.6638522744178772, "learning_rate": 0.0005, "loss": 1.5213041305541992, "step": 21805 }, { "epoch": 1.7190382341348047, "grad_norm": 0.542906641960144, "learning_rate": 0.0005, "loss": 1.6090974807739258, "step": 21806 }, { "epoch": 1.7191170674024439, "grad_norm": 0.5293405652046204, "learning_rate": 0.0005, "loss": 1.3810724020004272, "step": 21807 }, { "epoch": 1.7191959006700828, "grad_norm": 0.7409855127334595, "learning_rate": 0.0005, "loss": 1.4691441059112549, "step": 21808 }, { "epoch": 1.7192747339377217, "grad_norm": 0.6354610919952393, "learning_rate": 0.0005, "loss": 1.4716846942901611, "step": 21809 }, { "epoch": 1.7193535672053608, "grad_norm": 0.7251202464103699, "learning_rate": 0.0005, "loss": 1.5000876188278198, "step": 21810 }, { "epoch": 1.7194324004729995, "grad_norm": 1.157734990119934, "learning_rate": 0.0005, "loss": 1.7052276134490967, "step": 21811 }, { "epoch": 1.7195112337406386, "grad_norm": 0.5596595406532288, "learning_rate": 0.0005, "loss": 1.5204248428344727, "step": 21812 }, { "epoch": 1.7195900670082775, "grad_norm": 0.4656897783279419, "learning_rate": 0.0005, "loss": 1.1631866693496704, "step": 21813 }, { "epoch": 1.7196689002759165, "grad_norm": 0.7453405261039734, "learning_rate": 0.0005, "loss": 1.8924283981323242, "step": 21814 }, { "epoch": 1.7197477335435554, "grad_norm": 0.5979515314102173, "learning_rate": 0.0005, "loss": 1.4454342126846313, "step": 21815 }, { "epoch": 1.7198265668111943, "grad_norm": 0.5898576378822327, "learning_rate": 0.0005, "loss": 1.5134778022766113, "step": 21816 }, { "epoch": 1.7199054000788334, "grad_norm": 0.4951651096343994, "learning_rate": 0.0005, "loss": 1.6547319889068604, "step": 21817 }, { "epoch": 1.719984233346472, "grad_norm": 0.49259257316589355, "learning_rate": 0.0005, "loss": 1.3707388639450073, "step": 21818 }, { "epoch": 1.7200630666141112, "grad_norm": 0.7408431768417358, "learning_rate": 0.0005, "loss": 1.584813117980957, "step": 21819 }, { "epoch": 1.7201418998817501, "grad_norm": 0.5402087569236755, "learning_rate": 0.0005, "loss": 1.6078472137451172, "step": 21820 }, { "epoch": 1.720220733149389, "grad_norm": 0.5540790557861328, "learning_rate": 0.0005, "loss": 1.489404559135437, "step": 21821 }, { "epoch": 1.720299566417028, "grad_norm": 0.6189216375350952, "learning_rate": 0.0005, "loss": 1.600019931793213, "step": 21822 }, { "epoch": 1.7203783996846669, "grad_norm": 0.5737939476966858, "learning_rate": 0.0005, "loss": 1.530731201171875, "step": 21823 }, { "epoch": 1.720457232952306, "grad_norm": 0.5625503659248352, "learning_rate": 0.0005, "loss": 1.6712961196899414, "step": 21824 }, { "epoch": 1.7205360662199447, "grad_norm": 0.46648916602134705, "learning_rate": 0.0005, "loss": 1.5307927131652832, "step": 21825 }, { "epoch": 1.7206148994875838, "grad_norm": 0.5698933601379395, "learning_rate": 0.0005, "loss": 1.6058517694473267, "step": 21826 }, { "epoch": 1.7206937327552227, "grad_norm": 0.5867814421653748, "learning_rate": 0.0005, "loss": 1.468017339706421, "step": 21827 }, { "epoch": 1.7207725660228617, "grad_norm": 0.4690709412097931, "learning_rate": 0.0005, "loss": 1.4743287563323975, "step": 21828 }, { "epoch": 1.7208513992905006, "grad_norm": 0.5201100707054138, "learning_rate": 0.0005, "loss": 1.5417394638061523, "step": 21829 }, { "epoch": 1.7209302325581395, "grad_norm": 0.610183596611023, "learning_rate": 0.0005, "loss": 1.5441441535949707, "step": 21830 }, { "epoch": 1.7210090658257786, "grad_norm": 0.4973241686820984, "learning_rate": 0.0005, "loss": 1.395263433456421, "step": 21831 }, { "epoch": 1.7210878990934173, "grad_norm": 0.6054885387420654, "learning_rate": 0.0005, "loss": 1.478954792022705, "step": 21832 }, { "epoch": 1.7211667323610564, "grad_norm": 0.4572086036205292, "learning_rate": 0.0005, "loss": 1.2996872663497925, "step": 21833 }, { "epoch": 1.7212455656286954, "grad_norm": 0.6783793568611145, "learning_rate": 0.0005, "loss": 1.8138219118118286, "step": 21834 }, { "epoch": 1.7213243988963343, "grad_norm": 0.726398229598999, "learning_rate": 0.0005, "loss": 1.704862117767334, "step": 21835 }, { "epoch": 1.7214032321639732, "grad_norm": 0.48076462745666504, "learning_rate": 0.0005, "loss": 1.3087496757507324, "step": 21836 }, { "epoch": 1.721482065431612, "grad_norm": 0.5997559428215027, "learning_rate": 0.0005, "loss": 2.0415258407592773, "step": 21837 }, { "epoch": 1.7215608986992512, "grad_norm": 0.4815102517604828, "learning_rate": 0.0005, "loss": 1.5547581911087036, "step": 21838 }, { "epoch": 1.72163973196689, "grad_norm": 0.4344836175441742, "learning_rate": 0.0005, "loss": 1.2670854330062866, "step": 21839 }, { "epoch": 1.721718565234529, "grad_norm": 0.5781961679458618, "learning_rate": 0.0005, "loss": 1.6392230987548828, "step": 21840 }, { "epoch": 1.721797398502168, "grad_norm": 0.6530513167381287, "learning_rate": 0.0005, "loss": 1.3274481296539307, "step": 21841 }, { "epoch": 1.7218762317698069, "grad_norm": 0.6067551970481873, "learning_rate": 0.0005, "loss": 1.7329918146133423, "step": 21842 }, { "epoch": 1.7219550650374458, "grad_norm": 0.5045309662818909, "learning_rate": 0.0005, "loss": 1.4766502380371094, "step": 21843 }, { "epoch": 1.7220338983050847, "grad_norm": 0.5081125497817993, "learning_rate": 0.0005, "loss": 1.3748043775558472, "step": 21844 }, { "epoch": 1.7221127315727238, "grad_norm": 0.6135352253913879, "learning_rate": 0.0005, "loss": 1.7763532400131226, "step": 21845 }, { "epoch": 1.7221915648403625, "grad_norm": 0.5509445667266846, "learning_rate": 0.0005, "loss": 1.5085046291351318, "step": 21846 }, { "epoch": 1.7222703981080016, "grad_norm": 0.4069308340549469, "learning_rate": 0.0005, "loss": 1.3133105039596558, "step": 21847 }, { "epoch": 1.7223492313756406, "grad_norm": 0.4743083417415619, "learning_rate": 0.0005, "loss": 1.3826396465301514, "step": 21848 }, { "epoch": 1.7224280646432795, "grad_norm": 0.5941360592842102, "learning_rate": 0.0005, "loss": 1.3795689344406128, "step": 21849 }, { "epoch": 1.7225068979109184, "grad_norm": 0.44954532384872437, "learning_rate": 0.0005, "loss": 1.4212639331817627, "step": 21850 }, { "epoch": 1.7225857311785573, "grad_norm": 1.026429295539856, "learning_rate": 0.0005, "loss": 1.4753530025482178, "step": 21851 }, { "epoch": 1.7226645644461964, "grad_norm": 0.5854177474975586, "learning_rate": 0.0005, "loss": 1.553795576095581, "step": 21852 }, { "epoch": 1.7227433977138351, "grad_norm": 0.5327762365341187, "learning_rate": 0.0005, "loss": 1.5769033432006836, "step": 21853 }, { "epoch": 1.7228222309814742, "grad_norm": 0.42170315980911255, "learning_rate": 0.0005, "loss": 1.4348094463348389, "step": 21854 }, { "epoch": 1.7229010642491132, "grad_norm": 0.5432132482528687, "learning_rate": 0.0005, "loss": 1.8828529119491577, "step": 21855 }, { "epoch": 1.722979897516752, "grad_norm": 0.5744550824165344, "learning_rate": 0.0005, "loss": 1.3874562978744507, "step": 21856 }, { "epoch": 1.723058730784391, "grad_norm": 0.5202658176422119, "learning_rate": 0.0005, "loss": 1.5027990341186523, "step": 21857 }, { "epoch": 1.72313756405203, "grad_norm": 0.5166282057762146, "learning_rate": 0.0005, "loss": 1.4621362686157227, "step": 21858 }, { "epoch": 1.723216397319669, "grad_norm": 0.4864627718925476, "learning_rate": 0.0005, "loss": 1.3979535102844238, "step": 21859 }, { "epoch": 1.7232952305873077, "grad_norm": 0.5700053572654724, "learning_rate": 0.0005, "loss": 1.6136411428451538, "step": 21860 }, { "epoch": 1.7233740638549468, "grad_norm": 0.4856971800327301, "learning_rate": 0.0005, "loss": 1.5632565021514893, "step": 21861 }, { "epoch": 1.7234528971225858, "grad_norm": 0.5948509573936462, "learning_rate": 0.0005, "loss": 1.7107548713684082, "step": 21862 }, { "epoch": 1.7235317303902247, "grad_norm": 0.5835323929786682, "learning_rate": 0.0005, "loss": 1.5544838905334473, "step": 21863 }, { "epoch": 1.7236105636578636, "grad_norm": 0.49539467692375183, "learning_rate": 0.0005, "loss": 1.545295000076294, "step": 21864 }, { "epoch": 1.7236893969255025, "grad_norm": 0.4659227132797241, "learning_rate": 0.0005, "loss": 1.4187579154968262, "step": 21865 }, { "epoch": 1.7237682301931416, "grad_norm": 0.4666129946708679, "learning_rate": 0.0005, "loss": 1.533090591430664, "step": 21866 }, { "epoch": 1.7238470634607803, "grad_norm": 0.5479272603988647, "learning_rate": 0.0005, "loss": 1.5109095573425293, "step": 21867 }, { "epoch": 1.7239258967284194, "grad_norm": 0.4623171389102936, "learning_rate": 0.0005, "loss": 1.225074291229248, "step": 21868 }, { "epoch": 1.7240047299960584, "grad_norm": 0.6257527470588684, "learning_rate": 0.0005, "loss": 1.3789699077606201, "step": 21869 }, { "epoch": 1.7240835632636973, "grad_norm": 0.5486267805099487, "learning_rate": 0.0005, "loss": 1.5839321613311768, "step": 21870 }, { "epoch": 1.7241623965313362, "grad_norm": 0.5525999665260315, "learning_rate": 0.0005, "loss": 1.8256301879882812, "step": 21871 }, { "epoch": 1.724241229798975, "grad_norm": 0.5752257704734802, "learning_rate": 0.0005, "loss": 1.3861345052719116, "step": 21872 }, { "epoch": 1.7243200630666142, "grad_norm": 0.6960929036140442, "learning_rate": 0.0005, "loss": 1.613675594329834, "step": 21873 }, { "epoch": 1.724398896334253, "grad_norm": 0.49679505825042725, "learning_rate": 0.0005, "loss": 1.3577818870544434, "step": 21874 }, { "epoch": 1.724477729601892, "grad_norm": 0.4795374870300293, "learning_rate": 0.0005, "loss": 1.3133349418640137, "step": 21875 }, { "epoch": 1.724556562869531, "grad_norm": 0.46807271242141724, "learning_rate": 0.0005, "loss": 1.4612622261047363, "step": 21876 }, { "epoch": 1.7246353961371699, "grad_norm": 0.6198059916496277, "learning_rate": 0.0005, "loss": 1.597882866859436, "step": 21877 }, { "epoch": 1.7247142294048088, "grad_norm": 0.5082650780677795, "learning_rate": 0.0005, "loss": 1.6250758171081543, "step": 21878 }, { "epoch": 1.7247930626724477, "grad_norm": 0.523014485836029, "learning_rate": 0.0005, "loss": 1.3477632999420166, "step": 21879 }, { "epoch": 1.7248718959400868, "grad_norm": 0.5232305526733398, "learning_rate": 0.0005, "loss": 1.4068610668182373, "step": 21880 }, { "epoch": 1.7249507292077255, "grad_norm": 0.5367713570594788, "learning_rate": 0.0005, "loss": 1.4114629030227661, "step": 21881 }, { "epoch": 1.7250295624753647, "grad_norm": 0.5624872446060181, "learning_rate": 0.0005, "loss": 1.4941551685333252, "step": 21882 }, { "epoch": 1.7251083957430036, "grad_norm": 0.5758808851242065, "learning_rate": 0.0005, "loss": 1.4278780221939087, "step": 21883 }, { "epoch": 1.7251872290106425, "grad_norm": 0.9917543530464172, "learning_rate": 0.0005, "loss": 1.4120830297470093, "step": 21884 }, { "epoch": 1.7252660622782814, "grad_norm": 0.5288838744163513, "learning_rate": 0.0005, "loss": 1.5136973857879639, "step": 21885 }, { "epoch": 1.7253448955459203, "grad_norm": 0.4798814356327057, "learning_rate": 0.0005, "loss": 1.495638370513916, "step": 21886 }, { "epoch": 1.7254237288135594, "grad_norm": 0.6402269005775452, "learning_rate": 0.0005, "loss": 1.4319612979888916, "step": 21887 }, { "epoch": 1.7255025620811981, "grad_norm": 0.4410199522972107, "learning_rate": 0.0005, "loss": 1.2576682567596436, "step": 21888 }, { "epoch": 1.7255813953488373, "grad_norm": 0.60560142993927, "learning_rate": 0.0005, "loss": 1.6018548011779785, "step": 21889 }, { "epoch": 1.7256602286164762, "grad_norm": 0.49052318930625916, "learning_rate": 0.0005, "loss": 1.509070873260498, "step": 21890 }, { "epoch": 1.725739061884115, "grad_norm": 0.4326682984828949, "learning_rate": 0.0005, "loss": 1.2046928405761719, "step": 21891 }, { "epoch": 1.725817895151754, "grad_norm": 0.4841190576553345, "learning_rate": 0.0005, "loss": 1.3092210292816162, "step": 21892 }, { "epoch": 1.725896728419393, "grad_norm": 0.48810648918151855, "learning_rate": 0.0005, "loss": 1.3441112041473389, "step": 21893 }, { "epoch": 1.725975561687032, "grad_norm": 0.5561065673828125, "learning_rate": 0.0005, "loss": 1.6367621421813965, "step": 21894 }, { "epoch": 1.7260543949546707, "grad_norm": 0.7405359745025635, "learning_rate": 0.0005, "loss": 1.3911166191101074, "step": 21895 }, { "epoch": 1.7261332282223099, "grad_norm": 0.5631016492843628, "learning_rate": 0.0005, "loss": 1.685662031173706, "step": 21896 }, { "epoch": 1.7262120614899488, "grad_norm": 0.5288599133491516, "learning_rate": 0.0005, "loss": 1.6422584056854248, "step": 21897 }, { "epoch": 1.7262908947575877, "grad_norm": 0.531507134437561, "learning_rate": 0.0005, "loss": 1.5743634700775146, "step": 21898 }, { "epoch": 1.7263697280252266, "grad_norm": 0.4991095960140228, "learning_rate": 0.0005, "loss": 1.5992708206176758, "step": 21899 }, { "epoch": 1.7264485612928655, "grad_norm": 0.5392904877662659, "learning_rate": 0.0005, "loss": 1.4212628602981567, "step": 21900 }, { "epoch": 1.7265273945605046, "grad_norm": 0.4943121671676636, "learning_rate": 0.0005, "loss": 1.402176856994629, "step": 21901 }, { "epoch": 1.7266062278281433, "grad_norm": 0.8986660838127136, "learning_rate": 0.0005, "loss": 1.7445859909057617, "step": 21902 }, { "epoch": 1.7266850610957825, "grad_norm": 0.4898858070373535, "learning_rate": 0.0005, "loss": 1.3297734260559082, "step": 21903 }, { "epoch": 1.7267638943634214, "grad_norm": 0.6080693006515503, "learning_rate": 0.0005, "loss": 1.3709752559661865, "step": 21904 }, { "epoch": 1.7268427276310603, "grad_norm": 0.44794732332229614, "learning_rate": 0.0005, "loss": 1.3118317127227783, "step": 21905 }, { "epoch": 1.7269215608986994, "grad_norm": 0.5701594948768616, "learning_rate": 0.0005, "loss": 1.6291695833206177, "step": 21906 }, { "epoch": 1.727000394166338, "grad_norm": 0.5286087989807129, "learning_rate": 0.0005, "loss": 1.7252100706100464, "step": 21907 }, { "epoch": 1.7270792274339772, "grad_norm": 0.5759288668632507, "learning_rate": 0.0005, "loss": 1.2977285385131836, "step": 21908 }, { "epoch": 1.727158060701616, "grad_norm": 0.5217968225479126, "learning_rate": 0.0005, "loss": 1.4186545610427856, "step": 21909 }, { "epoch": 1.727236893969255, "grad_norm": 0.48471927642822266, "learning_rate": 0.0005, "loss": 1.5348083972930908, "step": 21910 }, { "epoch": 1.727315727236894, "grad_norm": 0.5164737105369568, "learning_rate": 0.0005, "loss": 1.695281982421875, "step": 21911 }, { "epoch": 1.7273945605045329, "grad_norm": 0.6130142211914062, "learning_rate": 0.0005, "loss": 1.5930931568145752, "step": 21912 }, { "epoch": 1.727473393772172, "grad_norm": 0.5621614456176758, "learning_rate": 0.0005, "loss": 1.6030910015106201, "step": 21913 }, { "epoch": 1.7275522270398107, "grad_norm": 0.4771735370159149, "learning_rate": 0.0005, "loss": 1.3939077854156494, "step": 21914 }, { "epoch": 1.7276310603074498, "grad_norm": 0.5171753168106079, "learning_rate": 0.0005, "loss": 1.3124415874481201, "step": 21915 }, { "epoch": 1.7277098935750885, "grad_norm": 0.47140711545944214, "learning_rate": 0.0005, "loss": 1.3068039417266846, "step": 21916 }, { "epoch": 1.7277887268427277, "grad_norm": 0.8685002326965332, "learning_rate": 0.0005, "loss": 1.8847447633743286, "step": 21917 }, { "epoch": 1.7278675601103666, "grad_norm": 3.496366500854492, "learning_rate": 0.0005, "loss": 1.6297276020050049, "step": 21918 }, { "epoch": 1.7279463933780055, "grad_norm": 0.4586787521839142, "learning_rate": 0.0005, "loss": 1.3901939392089844, "step": 21919 }, { "epoch": 1.7280252266456446, "grad_norm": 0.6120342016220093, "learning_rate": 0.0005, "loss": 1.6882038116455078, "step": 21920 }, { "epoch": 1.7281040599132833, "grad_norm": 0.6110941767692566, "learning_rate": 0.0005, "loss": 1.571197748184204, "step": 21921 }, { "epoch": 1.7281828931809224, "grad_norm": 0.45454636216163635, "learning_rate": 0.0005, "loss": 1.7829573154449463, "step": 21922 }, { "epoch": 1.7282617264485611, "grad_norm": 0.6253743171691895, "learning_rate": 0.0005, "loss": 1.609267234802246, "step": 21923 }, { "epoch": 1.7283405597162003, "grad_norm": 0.6022738218307495, "learning_rate": 0.0005, "loss": 1.636157512664795, "step": 21924 }, { "epoch": 1.7284193929838392, "grad_norm": 0.49140480160713196, "learning_rate": 0.0005, "loss": 1.398878812789917, "step": 21925 }, { "epoch": 1.728498226251478, "grad_norm": 0.7219703793525696, "learning_rate": 0.0005, "loss": 1.5920395851135254, "step": 21926 }, { "epoch": 1.7285770595191172, "grad_norm": 0.6057623028755188, "learning_rate": 0.0005, "loss": 1.3212974071502686, "step": 21927 }, { "epoch": 1.728655892786756, "grad_norm": 0.48924168944358826, "learning_rate": 0.0005, "loss": 1.3038334846496582, "step": 21928 }, { "epoch": 1.728734726054395, "grad_norm": 0.5384736657142639, "learning_rate": 0.0005, "loss": 1.4858582019805908, "step": 21929 }, { "epoch": 1.7288135593220337, "grad_norm": 0.4919775128364563, "learning_rate": 0.0005, "loss": 1.46958589553833, "step": 21930 }, { "epoch": 1.7288923925896729, "grad_norm": 0.5784628391265869, "learning_rate": 0.0005, "loss": 1.6000852584838867, "step": 21931 }, { "epoch": 1.7289712258573118, "grad_norm": 0.504661500453949, "learning_rate": 0.0005, "loss": 1.438838005065918, "step": 21932 }, { "epoch": 1.7290500591249507, "grad_norm": 0.4786759912967682, "learning_rate": 0.0005, "loss": 1.3583279848098755, "step": 21933 }, { "epoch": 1.7291288923925898, "grad_norm": 0.6572934985160828, "learning_rate": 0.0005, "loss": 1.6276264190673828, "step": 21934 }, { "epoch": 1.7292077256602285, "grad_norm": 0.5650954842567444, "learning_rate": 0.0005, "loss": 1.6508708000183105, "step": 21935 }, { "epoch": 1.7292865589278676, "grad_norm": 0.47262904047966003, "learning_rate": 0.0005, "loss": 1.4795753955841064, "step": 21936 }, { "epoch": 1.7293653921955066, "grad_norm": 0.5561597943305969, "learning_rate": 0.0005, "loss": 1.5851620435714722, "step": 21937 }, { "epoch": 1.7294442254631455, "grad_norm": 0.45863720774650574, "learning_rate": 0.0005, "loss": 1.4399590492248535, "step": 21938 }, { "epoch": 1.7295230587307844, "grad_norm": 0.7237648963928223, "learning_rate": 0.0005, "loss": 1.4938974380493164, "step": 21939 }, { "epoch": 1.7296018919984233, "grad_norm": 0.49468591809272766, "learning_rate": 0.0005, "loss": 1.5274754762649536, "step": 21940 }, { "epoch": 1.7296807252660624, "grad_norm": 0.6073204278945923, "learning_rate": 0.0005, "loss": 1.6414337158203125, "step": 21941 }, { "epoch": 1.7297595585337011, "grad_norm": 0.4765653908252716, "learning_rate": 0.0005, "loss": 1.411259412765503, "step": 21942 }, { "epoch": 1.7298383918013402, "grad_norm": 0.5241172313690186, "learning_rate": 0.0005, "loss": 1.3930858373641968, "step": 21943 }, { "epoch": 1.7299172250689792, "grad_norm": 0.5923259854316711, "learning_rate": 0.0005, "loss": 1.527956247329712, "step": 21944 }, { "epoch": 1.729996058336618, "grad_norm": 0.49095389246940613, "learning_rate": 0.0005, "loss": 1.3020591735839844, "step": 21945 }, { "epoch": 1.730074891604257, "grad_norm": 0.46354037523269653, "learning_rate": 0.0005, "loss": 1.4258314371109009, "step": 21946 }, { "epoch": 1.730153724871896, "grad_norm": 0.504221498966217, "learning_rate": 0.0005, "loss": 1.4357091188430786, "step": 21947 }, { "epoch": 1.730232558139535, "grad_norm": 0.46768951416015625, "learning_rate": 0.0005, "loss": 1.4040699005126953, "step": 21948 }, { "epoch": 1.7303113914071737, "grad_norm": 0.590655505657196, "learning_rate": 0.0005, "loss": 1.3753831386566162, "step": 21949 }, { "epoch": 1.7303902246748128, "grad_norm": 0.6372356414794922, "learning_rate": 0.0005, "loss": 1.6287626028060913, "step": 21950 }, { "epoch": 1.7304690579424518, "grad_norm": 0.5228093266487122, "learning_rate": 0.0005, "loss": 1.6404184103012085, "step": 21951 }, { "epoch": 1.7305478912100907, "grad_norm": 0.4830459952354431, "learning_rate": 0.0005, "loss": 1.2995569705963135, "step": 21952 }, { "epoch": 1.7306267244777296, "grad_norm": 0.4821501076221466, "learning_rate": 0.0005, "loss": 1.469409704208374, "step": 21953 }, { "epoch": 1.7307055577453685, "grad_norm": 0.5266149640083313, "learning_rate": 0.0005, "loss": 1.5845990180969238, "step": 21954 }, { "epoch": 1.7307843910130076, "grad_norm": 0.42998185753822327, "learning_rate": 0.0005, "loss": 1.5860023498535156, "step": 21955 }, { "epoch": 1.7308632242806463, "grad_norm": 0.45454666018486023, "learning_rate": 0.0005, "loss": 1.4719269275665283, "step": 21956 }, { "epoch": 1.7309420575482855, "grad_norm": 0.551798939704895, "learning_rate": 0.0005, "loss": 1.5936548709869385, "step": 21957 }, { "epoch": 1.7310208908159244, "grad_norm": 0.5121216773986816, "learning_rate": 0.0005, "loss": 1.5448318719863892, "step": 21958 }, { "epoch": 1.7310997240835633, "grad_norm": 0.4152575135231018, "learning_rate": 0.0005, "loss": 1.3685953617095947, "step": 21959 }, { "epoch": 1.7311785573512022, "grad_norm": 0.6541596055030823, "learning_rate": 0.0005, "loss": 1.7214646339416504, "step": 21960 }, { "epoch": 1.731257390618841, "grad_norm": 0.5316774249076843, "learning_rate": 0.0005, "loss": 1.652017593383789, "step": 21961 }, { "epoch": 1.7313362238864802, "grad_norm": 0.5790810585021973, "learning_rate": 0.0005, "loss": 1.4465858936309814, "step": 21962 }, { "epoch": 1.731415057154119, "grad_norm": 0.6274158954620361, "learning_rate": 0.0005, "loss": 1.688499927520752, "step": 21963 }, { "epoch": 1.731493890421758, "grad_norm": 0.6406245231628418, "learning_rate": 0.0005, "loss": 1.422989845275879, "step": 21964 }, { "epoch": 1.731572723689397, "grad_norm": 0.49931004643440247, "learning_rate": 0.0005, "loss": 1.3692591190338135, "step": 21965 }, { "epoch": 1.7316515569570359, "grad_norm": 0.5327887535095215, "learning_rate": 0.0005, "loss": 1.725827693939209, "step": 21966 }, { "epoch": 1.7317303902246748, "grad_norm": 0.7097545266151428, "learning_rate": 0.0005, "loss": 1.6797643899917603, "step": 21967 }, { "epoch": 1.7318092234923137, "grad_norm": 0.5064079761505127, "learning_rate": 0.0005, "loss": 1.4024579524993896, "step": 21968 }, { "epoch": 1.7318880567599528, "grad_norm": 0.5014252066612244, "learning_rate": 0.0005, "loss": 1.576056718826294, "step": 21969 }, { "epoch": 1.7319668900275915, "grad_norm": 0.46558237075805664, "learning_rate": 0.0005, "loss": 1.4937841892242432, "step": 21970 }, { "epoch": 1.7320457232952307, "grad_norm": 0.699431836605072, "learning_rate": 0.0005, "loss": 1.689181923866272, "step": 21971 }, { "epoch": 1.7321245565628696, "grad_norm": 0.5269167423248291, "learning_rate": 0.0005, "loss": 1.8102140426635742, "step": 21972 }, { "epoch": 1.7322033898305085, "grad_norm": 0.6427755951881409, "learning_rate": 0.0005, "loss": 1.3204796314239502, "step": 21973 }, { "epoch": 1.7322822230981474, "grad_norm": 0.5566294193267822, "learning_rate": 0.0005, "loss": 1.7500311136245728, "step": 21974 }, { "epoch": 1.7323610563657863, "grad_norm": 0.48007434606552124, "learning_rate": 0.0005, "loss": 1.539689540863037, "step": 21975 }, { "epoch": 1.7324398896334254, "grad_norm": 0.4510003626346588, "learning_rate": 0.0005, "loss": 1.494768500328064, "step": 21976 }, { "epoch": 1.7325187229010641, "grad_norm": 0.5841524004936218, "learning_rate": 0.0005, "loss": 1.654639720916748, "step": 21977 }, { "epoch": 1.7325975561687033, "grad_norm": 0.5339412093162537, "learning_rate": 0.0005, "loss": 1.4227005243301392, "step": 21978 }, { "epoch": 1.7326763894363422, "grad_norm": 0.48645415902137756, "learning_rate": 0.0005, "loss": 1.4754841327667236, "step": 21979 }, { "epoch": 1.732755222703981, "grad_norm": 0.515313446521759, "learning_rate": 0.0005, "loss": 1.516471266746521, "step": 21980 }, { "epoch": 1.73283405597162, "grad_norm": 0.6842474937438965, "learning_rate": 0.0005, "loss": 1.6675448417663574, "step": 21981 }, { "epoch": 1.732912889239259, "grad_norm": 0.49837034940719604, "learning_rate": 0.0005, "loss": 1.5295920372009277, "step": 21982 }, { "epoch": 1.732991722506898, "grad_norm": 0.5462870001792908, "learning_rate": 0.0005, "loss": 1.539536714553833, "step": 21983 }, { "epoch": 1.7330705557745367, "grad_norm": 0.48569199442863464, "learning_rate": 0.0005, "loss": 1.5365712642669678, "step": 21984 }, { "epoch": 1.7331493890421759, "grad_norm": 0.8113732933998108, "learning_rate": 0.0005, "loss": 1.690633773803711, "step": 21985 }, { "epoch": 1.7332282223098148, "grad_norm": 0.578326404094696, "learning_rate": 0.0005, "loss": 1.612921953201294, "step": 21986 }, { "epoch": 1.7333070555774537, "grad_norm": 0.524144172668457, "learning_rate": 0.0005, "loss": 1.4256224632263184, "step": 21987 }, { "epoch": 1.7333858888450926, "grad_norm": 0.5936036705970764, "learning_rate": 0.0005, "loss": 1.5788637399673462, "step": 21988 }, { "epoch": 1.7334647221127315, "grad_norm": 0.5099895000457764, "learning_rate": 0.0005, "loss": 1.402815818786621, "step": 21989 }, { "epoch": 1.7335435553803706, "grad_norm": 0.6770365238189697, "learning_rate": 0.0005, "loss": 1.623105764389038, "step": 21990 }, { "epoch": 1.7336223886480093, "grad_norm": 0.547048807144165, "learning_rate": 0.0005, "loss": 1.543436884880066, "step": 21991 }, { "epoch": 1.7337012219156485, "grad_norm": 0.6498579382896423, "learning_rate": 0.0005, "loss": 1.5848174095153809, "step": 21992 }, { "epoch": 1.7337800551832874, "grad_norm": 0.5504601001739502, "learning_rate": 0.0005, "loss": 1.4026598930358887, "step": 21993 }, { "epoch": 1.7338588884509263, "grad_norm": 0.696463406085968, "learning_rate": 0.0005, "loss": 1.4461236000061035, "step": 21994 }, { "epoch": 1.7339377217185652, "grad_norm": 0.6704970002174377, "learning_rate": 0.0005, "loss": 1.4629970788955688, "step": 21995 }, { "epoch": 1.734016554986204, "grad_norm": 0.45562055706977844, "learning_rate": 0.0005, "loss": 1.5269516706466675, "step": 21996 }, { "epoch": 1.7340953882538432, "grad_norm": 0.5277665257453918, "learning_rate": 0.0005, "loss": 1.5765302181243896, "step": 21997 }, { "epoch": 1.734174221521482, "grad_norm": 0.4466695487499237, "learning_rate": 0.0005, "loss": 1.518965721130371, "step": 21998 }, { "epoch": 1.734253054789121, "grad_norm": 0.46675413846969604, "learning_rate": 0.0005, "loss": 1.3574861288070679, "step": 21999 }, { "epoch": 1.73433188805676, "grad_norm": 0.837855339050293, "learning_rate": 0.0005, "loss": 1.639132022857666, "step": 22000 }, { "epoch": 1.7344107213243989, "grad_norm": 0.5941665172576904, "learning_rate": 0.0005, "loss": 1.4826738834381104, "step": 22001 }, { "epoch": 1.7344895545920378, "grad_norm": 0.5392221808433533, "learning_rate": 0.0005, "loss": 1.5387208461761475, "step": 22002 }, { "epoch": 1.7345683878596767, "grad_norm": 0.6510298848152161, "learning_rate": 0.0005, "loss": 1.660487413406372, "step": 22003 }, { "epoch": 1.7346472211273158, "grad_norm": 0.5151875615119934, "learning_rate": 0.0005, "loss": 1.525575876235962, "step": 22004 }, { "epoch": 1.7347260543949545, "grad_norm": 0.5195518136024475, "learning_rate": 0.0005, "loss": 1.4372023344039917, "step": 22005 }, { "epoch": 1.7348048876625937, "grad_norm": 0.591515302658081, "learning_rate": 0.0005, "loss": 1.8247891664505005, "step": 22006 }, { "epoch": 1.7348837209302326, "grad_norm": 0.4628894031047821, "learning_rate": 0.0005, "loss": 1.487154483795166, "step": 22007 }, { "epoch": 1.7349625541978715, "grad_norm": 0.5030980110168457, "learning_rate": 0.0005, "loss": 1.3529095649719238, "step": 22008 }, { "epoch": 1.7350413874655104, "grad_norm": 0.44927215576171875, "learning_rate": 0.0005, "loss": 1.3340766429901123, "step": 22009 }, { "epoch": 1.7351202207331493, "grad_norm": 0.5456018447875977, "learning_rate": 0.0005, "loss": 1.3577947616577148, "step": 22010 }, { "epoch": 1.7351990540007884, "grad_norm": 0.5693246126174927, "learning_rate": 0.0005, "loss": 1.6003830432891846, "step": 22011 }, { "epoch": 1.7352778872684271, "grad_norm": 0.5440634489059448, "learning_rate": 0.0005, "loss": 1.4770679473876953, "step": 22012 }, { "epoch": 1.7353567205360663, "grad_norm": 0.7100788354873657, "learning_rate": 0.0005, "loss": 1.6687211990356445, "step": 22013 }, { "epoch": 1.7354355538037052, "grad_norm": 0.5779743790626526, "learning_rate": 0.0005, "loss": 1.6812000274658203, "step": 22014 }, { "epoch": 1.735514387071344, "grad_norm": 0.5370977520942688, "learning_rate": 0.0005, "loss": 1.6123788356781006, "step": 22015 }, { "epoch": 1.735593220338983, "grad_norm": 0.7110399603843689, "learning_rate": 0.0005, "loss": 1.4526481628417969, "step": 22016 }, { "epoch": 1.735672053606622, "grad_norm": 0.5580442547798157, "learning_rate": 0.0005, "loss": 1.5403285026550293, "step": 22017 }, { "epoch": 1.735750886874261, "grad_norm": 0.5155510902404785, "learning_rate": 0.0005, "loss": 1.3316096067428589, "step": 22018 }, { "epoch": 1.7358297201418997, "grad_norm": 0.6115117073059082, "learning_rate": 0.0005, "loss": 1.7305471897125244, "step": 22019 }, { "epoch": 1.7359085534095389, "grad_norm": 0.6221814751625061, "learning_rate": 0.0005, "loss": 1.5677680969238281, "step": 22020 }, { "epoch": 1.7359873866771778, "grad_norm": 0.6090749502182007, "learning_rate": 0.0005, "loss": 1.592748761177063, "step": 22021 }, { "epoch": 1.7360662199448167, "grad_norm": 0.5389518737792969, "learning_rate": 0.0005, "loss": 1.6063976287841797, "step": 22022 }, { "epoch": 1.7361450532124556, "grad_norm": 0.44692954421043396, "learning_rate": 0.0005, "loss": 1.539499044418335, "step": 22023 }, { "epoch": 1.7362238864800945, "grad_norm": 0.5430145859718323, "learning_rate": 0.0005, "loss": 1.4410009384155273, "step": 22024 }, { "epoch": 1.7363027197477336, "grad_norm": 0.6035743951797485, "learning_rate": 0.0005, "loss": 1.6601271629333496, "step": 22025 }, { "epoch": 1.7363815530153723, "grad_norm": 0.48837992548942566, "learning_rate": 0.0005, "loss": 1.6668026447296143, "step": 22026 }, { "epoch": 1.7364603862830115, "grad_norm": 0.5416516065597534, "learning_rate": 0.0005, "loss": 1.4979541301727295, "step": 22027 }, { "epoch": 1.7365392195506504, "grad_norm": 0.48273080587387085, "learning_rate": 0.0005, "loss": 1.4364755153656006, "step": 22028 }, { "epoch": 1.7366180528182893, "grad_norm": 0.5520984530448914, "learning_rate": 0.0005, "loss": 1.6140351295471191, "step": 22029 }, { "epoch": 1.7366968860859284, "grad_norm": 0.6259403228759766, "learning_rate": 0.0005, "loss": 1.6441118717193604, "step": 22030 }, { "epoch": 1.7367757193535671, "grad_norm": 0.45243048667907715, "learning_rate": 0.0005, "loss": 1.475382924079895, "step": 22031 }, { "epoch": 1.7368545526212062, "grad_norm": 0.516157329082489, "learning_rate": 0.0005, "loss": 1.5652700662612915, "step": 22032 }, { "epoch": 1.736933385888845, "grad_norm": 0.7263015508651733, "learning_rate": 0.0005, "loss": 1.4891078472137451, "step": 22033 }, { "epoch": 1.737012219156484, "grad_norm": 0.4935564696788788, "learning_rate": 0.0005, "loss": 1.4264452457427979, "step": 22034 }, { "epoch": 1.737091052424123, "grad_norm": 0.43360114097595215, "learning_rate": 0.0005, "loss": 1.5397813320159912, "step": 22035 }, { "epoch": 1.737169885691762, "grad_norm": 0.5945929884910583, "learning_rate": 0.0005, "loss": 1.4997462034225464, "step": 22036 }, { "epoch": 1.737248718959401, "grad_norm": 0.721606969833374, "learning_rate": 0.0005, "loss": 2.0199787616729736, "step": 22037 }, { "epoch": 1.7373275522270397, "grad_norm": 0.49004119634628296, "learning_rate": 0.0005, "loss": 1.4954793453216553, "step": 22038 }, { "epoch": 1.7374063854946789, "grad_norm": 0.6207900047302246, "learning_rate": 0.0005, "loss": 1.6613407135009766, "step": 22039 }, { "epoch": 1.7374852187623175, "grad_norm": 0.5715898871421814, "learning_rate": 0.0005, "loss": 1.4220139980316162, "step": 22040 }, { "epoch": 1.7375640520299567, "grad_norm": 0.5637536644935608, "learning_rate": 0.0005, "loss": 1.5342763662338257, "step": 22041 }, { "epoch": 1.7376428852975956, "grad_norm": 0.6071308851242065, "learning_rate": 0.0005, "loss": 1.5226548910140991, "step": 22042 }, { "epoch": 1.7377217185652345, "grad_norm": 0.7418345212936401, "learning_rate": 0.0005, "loss": 1.5943882465362549, "step": 22043 }, { "epoch": 1.7378005518328736, "grad_norm": 0.5815160870552063, "learning_rate": 0.0005, "loss": 1.6124179363250732, "step": 22044 }, { "epoch": 1.7378793851005123, "grad_norm": 0.5008618831634521, "learning_rate": 0.0005, "loss": 1.3929660320281982, "step": 22045 }, { "epoch": 1.7379582183681515, "grad_norm": 0.4468517005443573, "learning_rate": 0.0005, "loss": 1.362034559249878, "step": 22046 }, { "epoch": 1.7380370516357901, "grad_norm": 0.5371538996696472, "learning_rate": 0.0005, "loss": 1.5055608749389648, "step": 22047 }, { "epoch": 1.7381158849034293, "grad_norm": 0.4680107831954956, "learning_rate": 0.0005, "loss": 1.3298382759094238, "step": 22048 }, { "epoch": 1.7381947181710682, "grad_norm": 0.564751148223877, "learning_rate": 0.0005, "loss": 1.5678248405456543, "step": 22049 }, { "epoch": 1.738273551438707, "grad_norm": 0.41753846406936646, "learning_rate": 0.0005, "loss": 1.2657145261764526, "step": 22050 }, { "epoch": 1.7383523847063462, "grad_norm": 0.5193027853965759, "learning_rate": 0.0005, "loss": 1.67433762550354, "step": 22051 }, { "epoch": 1.738431217973985, "grad_norm": 0.46853652596473694, "learning_rate": 0.0005, "loss": 1.5026203393936157, "step": 22052 }, { "epoch": 1.738510051241624, "grad_norm": 0.43761390447616577, "learning_rate": 0.0005, "loss": 1.4385634660720825, "step": 22053 }, { "epoch": 1.7385888845092627, "grad_norm": 0.5190703868865967, "learning_rate": 0.0005, "loss": 1.6037321090698242, "step": 22054 }, { "epoch": 1.7386677177769019, "grad_norm": 0.520945131778717, "learning_rate": 0.0005, "loss": 1.5061066150665283, "step": 22055 }, { "epoch": 1.7387465510445408, "grad_norm": 0.5662879347801208, "learning_rate": 0.0005, "loss": 1.4612393379211426, "step": 22056 }, { "epoch": 1.7388253843121797, "grad_norm": 0.686489999294281, "learning_rate": 0.0005, "loss": 1.4301122426986694, "step": 22057 }, { "epoch": 1.7389042175798188, "grad_norm": 0.5314396619796753, "learning_rate": 0.0005, "loss": 1.7074308395385742, "step": 22058 }, { "epoch": 1.7389830508474575, "grad_norm": 0.6875690817832947, "learning_rate": 0.0005, "loss": 1.7002482414245605, "step": 22059 }, { "epoch": 1.7390618841150967, "grad_norm": 0.5114969611167908, "learning_rate": 0.0005, "loss": 1.3846948146820068, "step": 22060 }, { "epoch": 1.7391407173827356, "grad_norm": 0.5437233448028564, "learning_rate": 0.0005, "loss": 1.5876524448394775, "step": 22061 }, { "epoch": 1.7392195506503745, "grad_norm": 0.5453805923461914, "learning_rate": 0.0005, "loss": 1.4798870086669922, "step": 22062 }, { "epoch": 1.7392983839180134, "grad_norm": 0.6392782330513, "learning_rate": 0.0005, "loss": 1.9159659147262573, "step": 22063 }, { "epoch": 1.7393772171856523, "grad_norm": 0.6612883806228638, "learning_rate": 0.0005, "loss": 1.589658498764038, "step": 22064 }, { "epoch": 1.7394560504532914, "grad_norm": 0.46068379282951355, "learning_rate": 0.0005, "loss": 1.4026479721069336, "step": 22065 }, { "epoch": 1.7395348837209301, "grad_norm": 0.5868152379989624, "learning_rate": 0.0005, "loss": 1.5551185607910156, "step": 22066 }, { "epoch": 1.7396137169885693, "grad_norm": 0.6643368005752563, "learning_rate": 0.0005, "loss": 1.2888340950012207, "step": 22067 }, { "epoch": 1.7396925502562082, "grad_norm": 1.136112093925476, "learning_rate": 0.0005, "loss": 1.466517686843872, "step": 22068 }, { "epoch": 1.739771383523847, "grad_norm": 0.49912989139556885, "learning_rate": 0.0005, "loss": 1.385568380355835, "step": 22069 }, { "epoch": 1.739850216791486, "grad_norm": 0.4898243248462677, "learning_rate": 0.0005, "loss": 1.335652470588684, "step": 22070 }, { "epoch": 1.739929050059125, "grad_norm": 0.5497227907180786, "learning_rate": 0.0005, "loss": 1.646139144897461, "step": 22071 }, { "epoch": 1.740007883326764, "grad_norm": 0.4815087616443634, "learning_rate": 0.0005, "loss": 1.60465669631958, "step": 22072 }, { "epoch": 1.7400867165944027, "grad_norm": 0.5332428216934204, "learning_rate": 0.0005, "loss": 1.4493863582611084, "step": 22073 }, { "epoch": 1.7401655498620419, "grad_norm": 0.4775089919567108, "learning_rate": 0.0005, "loss": 1.5446279048919678, "step": 22074 }, { "epoch": 1.7402443831296808, "grad_norm": 0.6202564239501953, "learning_rate": 0.0005, "loss": 1.6218161582946777, "step": 22075 }, { "epoch": 1.7403232163973197, "grad_norm": 0.5899133682250977, "learning_rate": 0.0005, "loss": 1.489583969116211, "step": 22076 }, { "epoch": 1.7404020496649586, "grad_norm": 0.46432551741600037, "learning_rate": 0.0005, "loss": 1.5801324844360352, "step": 22077 }, { "epoch": 1.7404808829325975, "grad_norm": 0.4293266236782074, "learning_rate": 0.0005, "loss": 1.6497077941894531, "step": 22078 }, { "epoch": 1.7405597162002366, "grad_norm": 0.7838032841682434, "learning_rate": 0.0005, "loss": 1.685968041419983, "step": 22079 }, { "epoch": 1.7406385494678753, "grad_norm": 0.6214744448661804, "learning_rate": 0.0005, "loss": 1.4697740077972412, "step": 22080 }, { "epoch": 1.7407173827355145, "grad_norm": 0.5242800116539001, "learning_rate": 0.0005, "loss": 1.2703622579574585, "step": 22081 }, { "epoch": 1.7407962160031534, "grad_norm": 0.479557067155838, "learning_rate": 0.0005, "loss": 1.3929507732391357, "step": 22082 }, { "epoch": 1.7408750492707923, "grad_norm": 0.4830333888530731, "learning_rate": 0.0005, "loss": 1.531579852104187, "step": 22083 }, { "epoch": 1.7409538825384312, "grad_norm": 0.6422291994094849, "learning_rate": 0.0005, "loss": 1.7394554615020752, "step": 22084 }, { "epoch": 1.74103271580607, "grad_norm": 0.5324879884719849, "learning_rate": 0.0005, "loss": 1.604265809059143, "step": 22085 }, { "epoch": 1.7411115490737092, "grad_norm": 0.4285329580307007, "learning_rate": 0.0005, "loss": 1.383513331413269, "step": 22086 }, { "epoch": 1.741190382341348, "grad_norm": 0.44928860664367676, "learning_rate": 0.0005, "loss": 1.4825870990753174, "step": 22087 }, { "epoch": 1.741269215608987, "grad_norm": 0.446191668510437, "learning_rate": 0.0005, "loss": 1.6038684844970703, "step": 22088 }, { "epoch": 1.741348048876626, "grad_norm": 0.8203321695327759, "learning_rate": 0.0005, "loss": 1.516869306564331, "step": 22089 }, { "epoch": 1.7414268821442649, "grad_norm": 0.5459567904472351, "learning_rate": 0.0005, "loss": 1.4901618957519531, "step": 22090 }, { "epoch": 1.7415057154119038, "grad_norm": 0.5852111577987671, "learning_rate": 0.0005, "loss": 1.4322247505187988, "step": 22091 }, { "epoch": 1.7415845486795427, "grad_norm": 0.8361773490905762, "learning_rate": 0.0005, "loss": 1.654677152633667, "step": 22092 }, { "epoch": 1.7416633819471818, "grad_norm": 0.5922197103500366, "learning_rate": 0.0005, "loss": 1.547959566116333, "step": 22093 }, { "epoch": 1.7417422152148205, "grad_norm": 0.5150166153907776, "learning_rate": 0.0005, "loss": 1.7244787216186523, "step": 22094 }, { "epoch": 1.7418210484824597, "grad_norm": 0.48574134707450867, "learning_rate": 0.0005, "loss": 1.4763988256454468, "step": 22095 }, { "epoch": 1.7418998817500986, "grad_norm": 0.539340078830719, "learning_rate": 0.0005, "loss": 1.3711729049682617, "step": 22096 }, { "epoch": 1.7419787150177375, "grad_norm": 0.9777900576591492, "learning_rate": 0.0005, "loss": 1.875993251800537, "step": 22097 }, { "epoch": 1.7420575482853764, "grad_norm": 0.6494413614273071, "learning_rate": 0.0005, "loss": 1.5190749168395996, "step": 22098 }, { "epoch": 1.7421363815530153, "grad_norm": 0.4873253405094147, "learning_rate": 0.0005, "loss": 1.4881200790405273, "step": 22099 }, { "epoch": 1.7422152148206544, "grad_norm": 0.6097508072853088, "learning_rate": 0.0005, "loss": 1.5507341623306274, "step": 22100 }, { "epoch": 1.7422940480882931, "grad_norm": 0.7049877643585205, "learning_rate": 0.0005, "loss": 1.6292692422866821, "step": 22101 }, { "epoch": 1.7423728813559323, "grad_norm": 0.5858089327812195, "learning_rate": 0.0005, "loss": 1.5620853900909424, "step": 22102 }, { "epoch": 1.7424517146235712, "grad_norm": 0.5689389109611511, "learning_rate": 0.0005, "loss": 1.5836862325668335, "step": 22103 }, { "epoch": 1.74253054789121, "grad_norm": 0.6036681532859802, "learning_rate": 0.0005, "loss": 1.5730732679367065, "step": 22104 }, { "epoch": 1.742609381158849, "grad_norm": 0.5835628509521484, "learning_rate": 0.0005, "loss": 1.3323276042938232, "step": 22105 }, { "epoch": 1.742688214426488, "grad_norm": 0.519993782043457, "learning_rate": 0.0005, "loss": 1.5063276290893555, "step": 22106 }, { "epoch": 1.742767047694127, "grad_norm": 0.7396320700645447, "learning_rate": 0.0005, "loss": 1.8083951473236084, "step": 22107 }, { "epoch": 1.7428458809617657, "grad_norm": 0.5471712946891785, "learning_rate": 0.0005, "loss": 1.5928783416748047, "step": 22108 }, { "epoch": 1.7429247142294049, "grad_norm": 0.528871476650238, "learning_rate": 0.0005, "loss": 1.6005746126174927, "step": 22109 }, { "epoch": 1.7430035474970438, "grad_norm": 0.5609644651412964, "learning_rate": 0.0005, "loss": 1.5227928161621094, "step": 22110 }, { "epoch": 1.7430823807646827, "grad_norm": 0.7434040904045105, "learning_rate": 0.0005, "loss": 1.7640293836593628, "step": 22111 }, { "epoch": 1.7431612140323216, "grad_norm": 0.5920652747154236, "learning_rate": 0.0005, "loss": 1.5937485694885254, "step": 22112 }, { "epoch": 1.7432400472999605, "grad_norm": 0.5753087997436523, "learning_rate": 0.0005, "loss": 1.6364927291870117, "step": 22113 }, { "epoch": 1.7433188805675996, "grad_norm": 6.052260875701904, "learning_rate": 0.0005, "loss": 1.6016167402267456, "step": 22114 }, { "epoch": 1.7433977138352383, "grad_norm": 0.6484853625297546, "learning_rate": 0.0005, "loss": 1.571093201637268, "step": 22115 }, { "epoch": 1.7434765471028775, "grad_norm": 0.4211050271987915, "learning_rate": 0.0005, "loss": 1.2542004585266113, "step": 22116 }, { "epoch": 1.7435553803705164, "grad_norm": 0.5166206955909729, "learning_rate": 0.0005, "loss": 1.5883578062057495, "step": 22117 }, { "epoch": 1.7436342136381553, "grad_norm": 0.5054952502250671, "learning_rate": 0.0005, "loss": 1.4334399700164795, "step": 22118 }, { "epoch": 1.7437130469057942, "grad_norm": 0.6109302043914795, "learning_rate": 0.0005, "loss": 1.3028264045715332, "step": 22119 }, { "epoch": 1.7437918801734331, "grad_norm": 0.5230967998504639, "learning_rate": 0.0005, "loss": 1.7117869853973389, "step": 22120 }, { "epoch": 1.7438707134410723, "grad_norm": 0.5715888738632202, "learning_rate": 0.0005, "loss": 1.3410427570343018, "step": 22121 }, { "epoch": 1.743949546708711, "grad_norm": 0.5671563744544983, "learning_rate": 0.0005, "loss": 1.657825231552124, "step": 22122 }, { "epoch": 1.74402837997635, "grad_norm": 0.6009573340415955, "learning_rate": 0.0005, "loss": 1.7341864109039307, "step": 22123 }, { "epoch": 1.744107213243989, "grad_norm": 0.500636637210846, "learning_rate": 0.0005, "loss": 1.4700597524642944, "step": 22124 }, { "epoch": 1.744186046511628, "grad_norm": 0.5498971939086914, "learning_rate": 0.0005, "loss": 1.6782485246658325, "step": 22125 }, { "epoch": 1.7442648797792668, "grad_norm": 0.819351077079773, "learning_rate": 0.0005, "loss": 1.5791542530059814, "step": 22126 }, { "epoch": 1.7443437130469057, "grad_norm": 0.516137957572937, "learning_rate": 0.0005, "loss": 1.3826360702514648, "step": 22127 }, { "epoch": 1.7444225463145449, "grad_norm": 0.8698538541793823, "learning_rate": 0.0005, "loss": 1.6698963642120361, "step": 22128 }, { "epoch": 1.7445013795821835, "grad_norm": 0.6077801585197449, "learning_rate": 0.0005, "loss": 1.6835134029388428, "step": 22129 }, { "epoch": 1.7445802128498227, "grad_norm": 0.6115272045135498, "learning_rate": 0.0005, "loss": 1.5759426355361938, "step": 22130 }, { "epoch": 1.7446590461174616, "grad_norm": 0.6744192838668823, "learning_rate": 0.0005, "loss": 1.471484899520874, "step": 22131 }, { "epoch": 1.7447378793851005, "grad_norm": 0.554101288318634, "learning_rate": 0.0005, "loss": 1.479060173034668, "step": 22132 }, { "epoch": 1.7448167126527394, "grad_norm": 0.638917088508606, "learning_rate": 0.0005, "loss": 1.5234060287475586, "step": 22133 }, { "epoch": 1.7448955459203783, "grad_norm": 0.4608238637447357, "learning_rate": 0.0005, "loss": 1.4557596445083618, "step": 22134 }, { "epoch": 1.7449743791880175, "grad_norm": 0.6928451657295227, "learning_rate": 0.0005, "loss": 1.7900421619415283, "step": 22135 }, { "epoch": 1.7450532124556561, "grad_norm": 0.46888455748558044, "learning_rate": 0.0005, "loss": 1.5155389308929443, "step": 22136 }, { "epoch": 1.7451320457232953, "grad_norm": 0.550778329372406, "learning_rate": 0.0005, "loss": 1.6277430057525635, "step": 22137 }, { "epoch": 1.7452108789909342, "grad_norm": 0.47967082262039185, "learning_rate": 0.0005, "loss": 1.2905831336975098, "step": 22138 }, { "epoch": 1.745289712258573, "grad_norm": 0.5468399524688721, "learning_rate": 0.0005, "loss": 1.8392266035079956, "step": 22139 }, { "epoch": 1.745368545526212, "grad_norm": 0.5955362319946289, "learning_rate": 0.0005, "loss": 1.9314260482788086, "step": 22140 }, { "epoch": 1.745447378793851, "grad_norm": 0.745478093624115, "learning_rate": 0.0005, "loss": 1.5888631343841553, "step": 22141 }, { "epoch": 1.74552621206149, "grad_norm": 0.46907636523246765, "learning_rate": 0.0005, "loss": 1.4304516315460205, "step": 22142 }, { "epoch": 1.7456050453291287, "grad_norm": 0.573177695274353, "learning_rate": 0.0005, "loss": 1.6269536018371582, "step": 22143 }, { "epoch": 1.7456838785967679, "grad_norm": 0.5297671556472778, "learning_rate": 0.0005, "loss": 1.4879850149154663, "step": 22144 }, { "epoch": 1.7457627118644068, "grad_norm": 0.4756115972995758, "learning_rate": 0.0005, "loss": 1.2439672946929932, "step": 22145 }, { "epoch": 1.7458415451320457, "grad_norm": 0.5989753007888794, "learning_rate": 0.0005, "loss": 1.5902868509292603, "step": 22146 }, { "epoch": 1.7459203783996846, "grad_norm": 0.5620501041412354, "learning_rate": 0.0005, "loss": 1.451033592224121, "step": 22147 }, { "epoch": 1.7459992116673235, "grad_norm": 0.47354888916015625, "learning_rate": 0.0005, "loss": 1.2668211460113525, "step": 22148 }, { "epoch": 1.7460780449349627, "grad_norm": 0.5247774124145508, "learning_rate": 0.0005, "loss": 1.4277793169021606, "step": 22149 }, { "epoch": 1.7461568782026013, "grad_norm": 0.5228763818740845, "learning_rate": 0.0005, "loss": 1.6304357051849365, "step": 22150 }, { "epoch": 1.7462357114702405, "grad_norm": 0.6336472034454346, "learning_rate": 0.0005, "loss": 1.7900433540344238, "step": 22151 }, { "epoch": 1.7463145447378794, "grad_norm": 0.8710147142410278, "learning_rate": 0.0005, "loss": 1.5451693534851074, "step": 22152 }, { "epoch": 1.7463933780055183, "grad_norm": 0.6346061825752258, "learning_rate": 0.0005, "loss": 1.5375723838806152, "step": 22153 }, { "epoch": 1.7464722112731574, "grad_norm": 0.5708315968513489, "learning_rate": 0.0005, "loss": 1.4211089611053467, "step": 22154 }, { "epoch": 1.7465510445407961, "grad_norm": 0.5020568370819092, "learning_rate": 0.0005, "loss": 1.4621195793151855, "step": 22155 }, { "epoch": 1.7466298778084353, "grad_norm": 0.5084263682365417, "learning_rate": 0.0005, "loss": 1.3204506635665894, "step": 22156 }, { "epoch": 1.746708711076074, "grad_norm": 0.64316725730896, "learning_rate": 0.0005, "loss": 1.5980947017669678, "step": 22157 }, { "epoch": 1.746787544343713, "grad_norm": 0.41355541348457336, "learning_rate": 0.0005, "loss": 1.2091224193572998, "step": 22158 }, { "epoch": 1.746866377611352, "grad_norm": 0.7301132082939148, "learning_rate": 0.0005, "loss": 1.776657223701477, "step": 22159 }, { "epoch": 1.746945210878991, "grad_norm": 0.5793591737747192, "learning_rate": 0.0005, "loss": 1.515331745147705, "step": 22160 }, { "epoch": 1.74702404414663, "grad_norm": 0.5410075783729553, "learning_rate": 0.0005, "loss": 1.5623948574066162, "step": 22161 }, { "epoch": 1.7471028774142687, "grad_norm": 0.6033591032028198, "learning_rate": 0.0005, "loss": 1.6535823345184326, "step": 22162 }, { "epoch": 1.7471817106819079, "grad_norm": 0.5826404094696045, "learning_rate": 0.0005, "loss": 1.286969780921936, "step": 22163 }, { "epoch": 1.7472605439495466, "grad_norm": 0.8239920735359192, "learning_rate": 0.0005, "loss": 1.7792601585388184, "step": 22164 }, { "epoch": 1.7473393772171857, "grad_norm": 0.5492702126502991, "learning_rate": 0.0005, "loss": 1.372276782989502, "step": 22165 }, { "epoch": 1.7474182104848246, "grad_norm": 0.6249783039093018, "learning_rate": 0.0005, "loss": 1.672616720199585, "step": 22166 }, { "epoch": 1.7474970437524635, "grad_norm": 0.5146420001983643, "learning_rate": 0.0005, "loss": 1.4751710891723633, "step": 22167 }, { "epoch": 1.7475758770201026, "grad_norm": 0.5592894554138184, "learning_rate": 0.0005, "loss": 1.559921145439148, "step": 22168 }, { "epoch": 1.7476547102877413, "grad_norm": 0.5407558083534241, "learning_rate": 0.0005, "loss": 1.6722239255905151, "step": 22169 }, { "epoch": 1.7477335435553805, "grad_norm": 0.6109877824783325, "learning_rate": 0.0005, "loss": 1.7227246761322021, "step": 22170 }, { "epoch": 1.7478123768230192, "grad_norm": 0.4969709813594818, "learning_rate": 0.0005, "loss": 1.625145435333252, "step": 22171 }, { "epoch": 1.7478912100906583, "grad_norm": 0.4668976664543152, "learning_rate": 0.0005, "loss": 1.4613542556762695, "step": 22172 }, { "epoch": 1.7479700433582972, "grad_norm": 0.5337780714035034, "learning_rate": 0.0005, "loss": 1.5979084968566895, "step": 22173 }, { "epoch": 1.748048876625936, "grad_norm": 0.49653422832489014, "learning_rate": 0.0005, "loss": 1.5366716384887695, "step": 22174 }, { "epoch": 1.7481277098935752, "grad_norm": 0.5703859925270081, "learning_rate": 0.0005, "loss": 1.58671236038208, "step": 22175 }, { "epoch": 1.748206543161214, "grad_norm": 0.6282569766044617, "learning_rate": 0.0005, "loss": 1.5288519859313965, "step": 22176 }, { "epoch": 1.748285376428853, "grad_norm": 0.7502233386039734, "learning_rate": 0.0005, "loss": 1.5591559410095215, "step": 22177 }, { "epoch": 1.7483642096964918, "grad_norm": 0.5858901739120483, "learning_rate": 0.0005, "loss": 1.6068755388259888, "step": 22178 }, { "epoch": 1.748443042964131, "grad_norm": 0.5630286335945129, "learning_rate": 0.0005, "loss": 1.562070369720459, "step": 22179 }, { "epoch": 1.7485218762317698, "grad_norm": 0.6069396138191223, "learning_rate": 0.0005, "loss": 1.7292060852050781, "step": 22180 }, { "epoch": 1.7486007094994087, "grad_norm": 0.5775426626205444, "learning_rate": 0.0005, "loss": 1.3662936687469482, "step": 22181 }, { "epoch": 1.7486795427670478, "grad_norm": 0.5312990546226501, "learning_rate": 0.0005, "loss": 1.5035091638565063, "step": 22182 }, { "epoch": 1.7487583760346865, "grad_norm": 0.5563662052154541, "learning_rate": 0.0005, "loss": 1.524416208267212, "step": 22183 }, { "epoch": 1.7488372093023257, "grad_norm": 0.4861357808113098, "learning_rate": 0.0005, "loss": 1.302269697189331, "step": 22184 }, { "epoch": 1.7489160425699646, "grad_norm": 0.6509703397750854, "learning_rate": 0.0005, "loss": 1.5239579677581787, "step": 22185 }, { "epoch": 1.7489948758376035, "grad_norm": 0.4238901734352112, "learning_rate": 0.0005, "loss": 1.5865477323532104, "step": 22186 }, { "epoch": 1.7490737091052424, "grad_norm": 0.5496618747711182, "learning_rate": 0.0005, "loss": 1.4293185472488403, "step": 22187 }, { "epoch": 1.7491525423728813, "grad_norm": 0.5024526715278625, "learning_rate": 0.0005, "loss": 1.6701250076293945, "step": 22188 }, { "epoch": 1.7492313756405204, "grad_norm": 0.47577783465385437, "learning_rate": 0.0005, "loss": 1.5649244785308838, "step": 22189 }, { "epoch": 1.7493102089081591, "grad_norm": 0.6266753673553467, "learning_rate": 0.0005, "loss": 1.673080563545227, "step": 22190 }, { "epoch": 1.7493890421757983, "grad_norm": 0.49507325887680054, "learning_rate": 0.0005, "loss": 1.4525065422058105, "step": 22191 }, { "epoch": 1.7494678754434372, "grad_norm": 0.48301663994789124, "learning_rate": 0.0005, "loss": 1.4994287490844727, "step": 22192 }, { "epoch": 1.749546708711076, "grad_norm": 0.5317489504814148, "learning_rate": 0.0005, "loss": 1.5780680179595947, "step": 22193 }, { "epoch": 1.749625541978715, "grad_norm": 0.5279366374015808, "learning_rate": 0.0005, "loss": 1.5744961500167847, "step": 22194 }, { "epoch": 1.749704375246354, "grad_norm": 0.5607012510299683, "learning_rate": 0.0005, "loss": 1.6665301322937012, "step": 22195 }, { "epoch": 1.749783208513993, "grad_norm": 0.48635974526405334, "learning_rate": 0.0005, "loss": 1.7642626762390137, "step": 22196 }, { "epoch": 1.7498620417816317, "grad_norm": 0.5356876850128174, "learning_rate": 0.0005, "loss": 1.6027225255966187, "step": 22197 }, { "epoch": 1.7499408750492709, "grad_norm": 0.5181992053985596, "learning_rate": 0.0005, "loss": 1.4538123607635498, "step": 22198 }, { "epoch": 1.7500197083169098, "grad_norm": 0.7146251797676086, "learning_rate": 0.0005, "loss": 1.5622844696044922, "step": 22199 }, { "epoch": 1.7500985415845487, "grad_norm": 0.45318281650543213, "learning_rate": 0.0005, "loss": 1.498050332069397, "step": 22200 }, { "epoch": 1.7501773748521876, "grad_norm": 0.6466466784477234, "learning_rate": 0.0005, "loss": 1.6530840396881104, "step": 22201 }, { "epoch": 1.7502562081198265, "grad_norm": 0.4979407787322998, "learning_rate": 0.0005, "loss": 1.5234737396240234, "step": 22202 }, { "epoch": 1.7503350413874657, "grad_norm": 0.5405541658401489, "learning_rate": 0.0005, "loss": 1.4986462593078613, "step": 22203 }, { "epoch": 1.7504138746551043, "grad_norm": 0.4742850363254547, "learning_rate": 0.0005, "loss": 1.3433668613433838, "step": 22204 }, { "epoch": 1.7504927079227435, "grad_norm": 0.46509289741516113, "learning_rate": 0.0005, "loss": 1.3566093444824219, "step": 22205 }, { "epoch": 1.7505715411903824, "grad_norm": 0.4967745840549469, "learning_rate": 0.0005, "loss": 1.4064209461212158, "step": 22206 }, { "epoch": 1.7506503744580213, "grad_norm": 0.6382189393043518, "learning_rate": 0.0005, "loss": 1.444244384765625, "step": 22207 }, { "epoch": 1.7507292077256602, "grad_norm": 0.5046707391738892, "learning_rate": 0.0005, "loss": 1.4817380905151367, "step": 22208 }, { "epoch": 1.7508080409932991, "grad_norm": 0.536960780620575, "learning_rate": 0.0005, "loss": 1.4580905437469482, "step": 22209 }, { "epoch": 1.7508868742609383, "grad_norm": 0.5823990106582642, "learning_rate": 0.0005, "loss": 1.5418646335601807, "step": 22210 }, { "epoch": 1.750965707528577, "grad_norm": 0.5029988884925842, "learning_rate": 0.0005, "loss": 1.6356987953186035, "step": 22211 }, { "epoch": 1.751044540796216, "grad_norm": 0.46464914083480835, "learning_rate": 0.0005, "loss": 1.3855524063110352, "step": 22212 }, { "epoch": 1.751123374063855, "grad_norm": 0.6180651783943176, "learning_rate": 0.0005, "loss": 1.2720293998718262, "step": 22213 }, { "epoch": 1.751202207331494, "grad_norm": 0.9796146750450134, "learning_rate": 0.0005, "loss": 1.484910488128662, "step": 22214 }, { "epoch": 1.7512810405991328, "grad_norm": 0.5098244547843933, "learning_rate": 0.0005, "loss": 1.5196515321731567, "step": 22215 }, { "epoch": 1.7513598738667717, "grad_norm": 0.5509717464447021, "learning_rate": 0.0005, "loss": 1.3907527923583984, "step": 22216 }, { "epoch": 1.7514387071344109, "grad_norm": 0.5584455132484436, "learning_rate": 0.0005, "loss": 1.6047008037567139, "step": 22217 }, { "epoch": 1.7515175404020495, "grad_norm": 0.7213476300239563, "learning_rate": 0.0005, "loss": 1.7288110256195068, "step": 22218 }, { "epoch": 1.7515963736696887, "grad_norm": 0.5366341471672058, "learning_rate": 0.0005, "loss": 1.4144115447998047, "step": 22219 }, { "epoch": 1.7516752069373276, "grad_norm": 0.5809858441352844, "learning_rate": 0.0005, "loss": 1.371375560760498, "step": 22220 }, { "epoch": 1.7517540402049665, "grad_norm": 0.5527459979057312, "learning_rate": 0.0005, "loss": 1.4364542961120605, "step": 22221 }, { "epoch": 1.7518328734726054, "grad_norm": 0.5703834891319275, "learning_rate": 0.0005, "loss": 1.6679631471633911, "step": 22222 }, { "epoch": 1.7519117067402443, "grad_norm": 0.6645456552505493, "learning_rate": 0.0005, "loss": 1.6714534759521484, "step": 22223 }, { "epoch": 1.7519905400078835, "grad_norm": 0.5562776327133179, "learning_rate": 0.0005, "loss": 1.4626719951629639, "step": 22224 }, { "epoch": 1.7520693732755221, "grad_norm": 0.5795906186103821, "learning_rate": 0.0005, "loss": 1.5947315692901611, "step": 22225 }, { "epoch": 1.7521482065431613, "grad_norm": 0.5810609459877014, "learning_rate": 0.0005, "loss": 1.4647681713104248, "step": 22226 }, { "epoch": 1.7522270398108002, "grad_norm": 0.4445863664150238, "learning_rate": 0.0005, "loss": 1.3296407461166382, "step": 22227 }, { "epoch": 1.752305873078439, "grad_norm": 0.5281003713607788, "learning_rate": 0.0005, "loss": 1.404952049255371, "step": 22228 }, { "epoch": 1.752384706346078, "grad_norm": 0.520204484462738, "learning_rate": 0.0005, "loss": 1.5050392150878906, "step": 22229 }, { "epoch": 1.752463539613717, "grad_norm": 0.6084606647491455, "learning_rate": 0.0005, "loss": 1.546443223953247, "step": 22230 }, { "epoch": 1.752542372881356, "grad_norm": 0.6538749933242798, "learning_rate": 0.0005, "loss": 1.59940767288208, "step": 22231 }, { "epoch": 1.7526212061489947, "grad_norm": 0.5497674942016602, "learning_rate": 0.0005, "loss": 1.5318456888198853, "step": 22232 }, { "epoch": 1.7527000394166339, "grad_norm": 0.4900047183036804, "learning_rate": 0.0005, "loss": 1.4185948371887207, "step": 22233 }, { "epoch": 1.7527788726842728, "grad_norm": 0.5150946378707886, "learning_rate": 0.0005, "loss": 1.5989325046539307, "step": 22234 }, { "epoch": 1.7528577059519117, "grad_norm": 0.6024685502052307, "learning_rate": 0.0005, "loss": 1.5697859525680542, "step": 22235 }, { "epoch": 1.7529365392195506, "grad_norm": 0.5742936134338379, "learning_rate": 0.0005, "loss": 1.6547709703445435, "step": 22236 }, { "epoch": 1.7530153724871895, "grad_norm": 0.5449380874633789, "learning_rate": 0.0005, "loss": 1.6353442668914795, "step": 22237 }, { "epoch": 1.7530942057548287, "grad_norm": 0.5127573609352112, "learning_rate": 0.0005, "loss": 1.570955514907837, "step": 22238 }, { "epoch": 1.7531730390224673, "grad_norm": 0.6188480854034424, "learning_rate": 0.0005, "loss": 1.5689451694488525, "step": 22239 }, { "epoch": 1.7532518722901065, "grad_norm": 0.5433536767959595, "learning_rate": 0.0005, "loss": 1.3876131772994995, "step": 22240 }, { "epoch": 1.7533307055577454, "grad_norm": 0.6042562127113342, "learning_rate": 0.0005, "loss": 1.7122879028320312, "step": 22241 }, { "epoch": 1.7534095388253843, "grad_norm": 0.5694528818130493, "learning_rate": 0.0005, "loss": 1.4037758111953735, "step": 22242 }, { "epoch": 1.7534883720930232, "grad_norm": 0.6136531829833984, "learning_rate": 0.0005, "loss": 1.4267454147338867, "step": 22243 }, { "epoch": 1.7535672053606621, "grad_norm": 0.6028197407722473, "learning_rate": 0.0005, "loss": 1.4729183912277222, "step": 22244 }, { "epoch": 1.7536460386283013, "grad_norm": 0.593247652053833, "learning_rate": 0.0005, "loss": 1.4764328002929688, "step": 22245 }, { "epoch": 1.75372487189594, "grad_norm": 0.5043308734893799, "learning_rate": 0.0005, "loss": 1.4304157495498657, "step": 22246 }, { "epoch": 1.753803705163579, "grad_norm": 0.6084796190261841, "learning_rate": 0.0005, "loss": 1.81436026096344, "step": 22247 }, { "epoch": 1.753882538431218, "grad_norm": 0.6270420551300049, "learning_rate": 0.0005, "loss": 1.6219135522842407, "step": 22248 }, { "epoch": 1.753961371698857, "grad_norm": 0.6192869544029236, "learning_rate": 0.0005, "loss": 1.5058543682098389, "step": 22249 }, { "epoch": 1.7540402049664958, "grad_norm": 0.5451578497886658, "learning_rate": 0.0005, "loss": 1.6696041822433472, "step": 22250 }, { "epoch": 1.7541190382341347, "grad_norm": 0.5437502264976501, "learning_rate": 0.0005, "loss": 1.7311301231384277, "step": 22251 }, { "epoch": 1.7541978715017739, "grad_norm": 0.43566787242889404, "learning_rate": 0.0005, "loss": 1.506524682044983, "step": 22252 }, { "epoch": 1.7542767047694126, "grad_norm": 0.5333601236343384, "learning_rate": 0.0005, "loss": 1.5287905931472778, "step": 22253 }, { "epoch": 1.7543555380370517, "grad_norm": 0.5110286474227905, "learning_rate": 0.0005, "loss": 1.5209722518920898, "step": 22254 }, { "epoch": 1.7544343713046906, "grad_norm": 0.4983314871788025, "learning_rate": 0.0005, "loss": 1.4910712242126465, "step": 22255 }, { "epoch": 1.7545132045723295, "grad_norm": 0.6046286225318909, "learning_rate": 0.0005, "loss": 1.472686767578125, "step": 22256 }, { "epoch": 1.7545920378399684, "grad_norm": 0.4966401159763336, "learning_rate": 0.0005, "loss": 1.523909330368042, "step": 22257 }, { "epoch": 1.7546708711076073, "grad_norm": 0.5366296768188477, "learning_rate": 0.0005, "loss": 1.5429880619049072, "step": 22258 }, { "epoch": 1.7547497043752465, "grad_norm": 0.5965622067451477, "learning_rate": 0.0005, "loss": 1.5554211139678955, "step": 22259 }, { "epoch": 1.7548285376428852, "grad_norm": 0.5648240447044373, "learning_rate": 0.0005, "loss": 1.4396932125091553, "step": 22260 }, { "epoch": 1.7549073709105243, "grad_norm": 0.5750551223754883, "learning_rate": 0.0005, "loss": 1.7054479122161865, "step": 22261 }, { "epoch": 1.7549862041781632, "grad_norm": 0.596057653427124, "learning_rate": 0.0005, "loss": 1.7582156658172607, "step": 22262 }, { "epoch": 1.7550650374458021, "grad_norm": 0.7636519074440002, "learning_rate": 0.0005, "loss": 1.5476601123809814, "step": 22263 }, { "epoch": 1.755143870713441, "grad_norm": 0.6934292316436768, "learning_rate": 0.0005, "loss": 1.6356016397476196, "step": 22264 }, { "epoch": 1.75522270398108, "grad_norm": 0.5680999755859375, "learning_rate": 0.0005, "loss": 1.3709139823913574, "step": 22265 }, { "epoch": 1.755301537248719, "grad_norm": 0.58539879322052, "learning_rate": 0.0005, "loss": 1.585442304611206, "step": 22266 }, { "epoch": 1.7553803705163578, "grad_norm": 0.7934898734092712, "learning_rate": 0.0005, "loss": 1.2855831384658813, "step": 22267 }, { "epoch": 1.755459203783997, "grad_norm": 0.7797712683677673, "learning_rate": 0.0005, "loss": 1.5240015983581543, "step": 22268 }, { "epoch": 1.7555380370516358, "grad_norm": 0.5509831309318542, "learning_rate": 0.0005, "loss": 1.6566784381866455, "step": 22269 }, { "epoch": 1.7556168703192747, "grad_norm": 0.5154466032981873, "learning_rate": 0.0005, "loss": 1.5359952449798584, "step": 22270 }, { "epoch": 1.7556957035869136, "grad_norm": 0.7171050906181335, "learning_rate": 0.0005, "loss": 1.4702887535095215, "step": 22271 }, { "epoch": 1.7557745368545525, "grad_norm": 0.4223062992095947, "learning_rate": 0.0005, "loss": 1.3364319801330566, "step": 22272 }, { "epoch": 1.7558533701221917, "grad_norm": 0.5468816161155701, "learning_rate": 0.0005, "loss": 1.405218243598938, "step": 22273 }, { "epoch": 1.7559322033898304, "grad_norm": 0.7278499007225037, "learning_rate": 0.0005, "loss": 1.6011748313903809, "step": 22274 }, { "epoch": 1.7560110366574695, "grad_norm": 0.5961043834686279, "learning_rate": 0.0005, "loss": 1.5554101467132568, "step": 22275 }, { "epoch": 1.7560898699251084, "grad_norm": 0.7701905965805054, "learning_rate": 0.0005, "loss": 1.3152393102645874, "step": 22276 }, { "epoch": 1.7561687031927473, "grad_norm": 0.6660118103027344, "learning_rate": 0.0005, "loss": 1.3064141273498535, "step": 22277 }, { "epoch": 1.7562475364603864, "grad_norm": 0.49363386631011963, "learning_rate": 0.0005, "loss": 1.5286043882369995, "step": 22278 }, { "epoch": 1.7563263697280251, "grad_norm": 0.6523692011833191, "learning_rate": 0.0005, "loss": 1.541176199913025, "step": 22279 }, { "epoch": 1.7564052029956643, "grad_norm": 0.913049042224884, "learning_rate": 0.0005, "loss": 1.8601806163787842, "step": 22280 }, { "epoch": 1.756484036263303, "grad_norm": 0.5812204480171204, "learning_rate": 0.0005, "loss": 1.7731010913848877, "step": 22281 }, { "epoch": 1.756562869530942, "grad_norm": 0.5181196928024292, "learning_rate": 0.0005, "loss": 1.8436427116394043, "step": 22282 }, { "epoch": 1.756641702798581, "grad_norm": 0.47986096143722534, "learning_rate": 0.0005, "loss": 1.4854826927185059, "step": 22283 }, { "epoch": 1.75672053606622, "grad_norm": 0.5866219997406006, "learning_rate": 0.0005, "loss": 1.3786327838897705, "step": 22284 }, { "epoch": 1.756799369333859, "grad_norm": 0.6087055206298828, "learning_rate": 0.0005, "loss": 1.3228322267532349, "step": 22285 }, { "epoch": 1.7568782026014977, "grad_norm": 0.5157248377799988, "learning_rate": 0.0005, "loss": 1.3506968021392822, "step": 22286 }, { "epoch": 1.7569570358691369, "grad_norm": 0.6288594007492065, "learning_rate": 0.0005, "loss": 1.4691848754882812, "step": 22287 }, { "epoch": 1.7570358691367756, "grad_norm": 0.5198593735694885, "learning_rate": 0.0005, "loss": 1.5799551010131836, "step": 22288 }, { "epoch": 1.7571147024044147, "grad_norm": 0.5268341302871704, "learning_rate": 0.0005, "loss": 1.5756583213806152, "step": 22289 }, { "epoch": 1.7571935356720536, "grad_norm": 0.6149864792823792, "learning_rate": 0.0005, "loss": 1.6713370084762573, "step": 22290 }, { "epoch": 1.7572723689396925, "grad_norm": 0.5699703097343445, "learning_rate": 0.0005, "loss": 1.5786571502685547, "step": 22291 }, { "epoch": 1.7573512022073317, "grad_norm": 0.6702895760536194, "learning_rate": 0.0005, "loss": 2.0025362968444824, "step": 22292 }, { "epoch": 1.7574300354749703, "grad_norm": 0.46239686012268066, "learning_rate": 0.0005, "loss": 1.3377785682678223, "step": 22293 }, { "epoch": 1.7575088687426095, "grad_norm": 0.4158715307712555, "learning_rate": 0.0005, "loss": 1.3979840278625488, "step": 22294 }, { "epoch": 1.7575877020102482, "grad_norm": 0.5473300218582153, "learning_rate": 0.0005, "loss": 1.4911001920700073, "step": 22295 }, { "epoch": 1.7576665352778873, "grad_norm": 0.5758176445960999, "learning_rate": 0.0005, "loss": 1.3982290029525757, "step": 22296 }, { "epoch": 1.7577453685455262, "grad_norm": 0.5252426862716675, "learning_rate": 0.0005, "loss": 1.4175280332565308, "step": 22297 }, { "epoch": 1.7578242018131651, "grad_norm": 0.4798647165298462, "learning_rate": 0.0005, "loss": 1.4778289794921875, "step": 22298 }, { "epoch": 1.7579030350808043, "grad_norm": 0.5799322128295898, "learning_rate": 0.0005, "loss": 1.475254774093628, "step": 22299 }, { "epoch": 1.757981868348443, "grad_norm": 0.6080901026725769, "learning_rate": 0.0005, "loss": 1.4892265796661377, "step": 22300 }, { "epoch": 1.758060701616082, "grad_norm": 0.43481314182281494, "learning_rate": 0.0005, "loss": 1.4126851558685303, "step": 22301 }, { "epoch": 1.7581395348837208, "grad_norm": 0.8150699138641357, "learning_rate": 0.0005, "loss": 1.569413661956787, "step": 22302 }, { "epoch": 1.75821836815136, "grad_norm": 0.9612080454826355, "learning_rate": 0.0005, "loss": 1.7695567607879639, "step": 22303 }, { "epoch": 1.7582972014189988, "grad_norm": 0.5541564226150513, "learning_rate": 0.0005, "loss": 1.535609483718872, "step": 22304 }, { "epoch": 1.7583760346866377, "grad_norm": 0.6491855978965759, "learning_rate": 0.0005, "loss": 1.666764259338379, "step": 22305 }, { "epoch": 1.7584548679542769, "grad_norm": 0.7410210967063904, "learning_rate": 0.0005, "loss": 1.6506798267364502, "step": 22306 }, { "epoch": 1.7585337012219155, "grad_norm": 0.8306615948677063, "learning_rate": 0.0005, "loss": 1.666919469833374, "step": 22307 }, { "epoch": 1.7586125344895547, "grad_norm": 0.5305435061454773, "learning_rate": 0.0005, "loss": 1.5316479206085205, "step": 22308 }, { "epoch": 1.7586913677571936, "grad_norm": 0.5366313457489014, "learning_rate": 0.0005, "loss": 1.4014174938201904, "step": 22309 }, { "epoch": 1.7587702010248325, "grad_norm": 0.5334072113037109, "learning_rate": 0.0005, "loss": 1.5063519477844238, "step": 22310 }, { "epoch": 1.7588490342924714, "grad_norm": 0.5393713116645813, "learning_rate": 0.0005, "loss": 1.3922621011734009, "step": 22311 }, { "epoch": 1.7589278675601103, "grad_norm": 0.6088690161705017, "learning_rate": 0.0005, "loss": 1.4957630634307861, "step": 22312 }, { "epoch": 1.7590067008277495, "grad_norm": 0.7214776873588562, "learning_rate": 0.0005, "loss": 1.4738435745239258, "step": 22313 }, { "epoch": 1.7590855340953881, "grad_norm": 0.9552748799324036, "learning_rate": 0.0005, "loss": 1.5272564888000488, "step": 22314 }, { "epoch": 1.7591643673630273, "grad_norm": 0.5339271426200867, "learning_rate": 0.0005, "loss": 1.4893765449523926, "step": 22315 }, { "epoch": 1.7592432006306662, "grad_norm": 0.5560135841369629, "learning_rate": 0.0005, "loss": 1.591779112815857, "step": 22316 }, { "epoch": 1.759322033898305, "grad_norm": 0.7084506750106812, "learning_rate": 0.0005, "loss": 1.4889389276504517, "step": 22317 }, { "epoch": 1.759400867165944, "grad_norm": 0.7239204049110413, "learning_rate": 0.0005, "loss": 1.6739776134490967, "step": 22318 }, { "epoch": 1.759479700433583, "grad_norm": 0.5724716782569885, "learning_rate": 0.0005, "loss": 1.5325618982315063, "step": 22319 }, { "epoch": 1.759558533701222, "grad_norm": 0.649634599685669, "learning_rate": 0.0005, "loss": 1.5170351266860962, "step": 22320 }, { "epoch": 1.7596373669688607, "grad_norm": 0.5126960873603821, "learning_rate": 0.0005, "loss": 1.6817210912704468, "step": 22321 }, { "epoch": 1.7597162002364999, "grad_norm": 0.6600993871688843, "learning_rate": 0.0005, "loss": 1.619773268699646, "step": 22322 }, { "epoch": 1.7597950335041388, "grad_norm": 0.8003473281860352, "learning_rate": 0.0005, "loss": 1.712131142616272, "step": 22323 }, { "epoch": 1.7598738667717777, "grad_norm": 0.4716939330101013, "learning_rate": 0.0005, "loss": 1.405795931816101, "step": 22324 }, { "epoch": 1.7599527000394166, "grad_norm": 0.5469992160797119, "learning_rate": 0.0005, "loss": 1.4076347351074219, "step": 22325 }, { "epoch": 1.7600315333070555, "grad_norm": 0.6341031193733215, "learning_rate": 0.0005, "loss": 1.5467026233673096, "step": 22326 }, { "epoch": 1.7601103665746947, "grad_norm": 0.6816801428794861, "learning_rate": 0.0005, "loss": 1.5345778465270996, "step": 22327 }, { "epoch": 1.7601891998423334, "grad_norm": 0.6382531523704529, "learning_rate": 0.0005, "loss": 1.6577863693237305, "step": 22328 }, { "epoch": 1.7602680331099725, "grad_norm": 0.8012211918830872, "learning_rate": 0.0005, "loss": 1.4526760578155518, "step": 22329 }, { "epoch": 1.7603468663776114, "grad_norm": 0.4854387640953064, "learning_rate": 0.0005, "loss": 1.4934444427490234, "step": 22330 }, { "epoch": 1.7604256996452503, "grad_norm": 0.5163257122039795, "learning_rate": 0.0005, "loss": 1.5287961959838867, "step": 22331 }, { "epoch": 1.7605045329128892, "grad_norm": 0.5700352191925049, "learning_rate": 0.0005, "loss": 1.6226251125335693, "step": 22332 }, { "epoch": 1.7605833661805281, "grad_norm": 0.4483304023742676, "learning_rate": 0.0005, "loss": 1.3422825336456299, "step": 22333 }, { "epoch": 1.7606621994481673, "grad_norm": 0.5072913765907288, "learning_rate": 0.0005, "loss": 1.3102540969848633, "step": 22334 }, { "epoch": 1.760741032715806, "grad_norm": 0.5306575894355774, "learning_rate": 0.0005, "loss": 1.697749137878418, "step": 22335 }, { "epoch": 1.760819865983445, "grad_norm": 0.6114729642868042, "learning_rate": 0.0005, "loss": 1.6201276779174805, "step": 22336 }, { "epoch": 1.760898699251084, "grad_norm": 0.5920649766921997, "learning_rate": 0.0005, "loss": 1.6673479080200195, "step": 22337 }, { "epoch": 1.760977532518723, "grad_norm": 0.5252963900566101, "learning_rate": 0.0005, "loss": 1.6143157482147217, "step": 22338 }, { "epoch": 1.7610563657863618, "grad_norm": 0.5746499300003052, "learning_rate": 0.0005, "loss": 1.3716535568237305, "step": 22339 }, { "epoch": 1.7611351990540007, "grad_norm": 0.6073278784751892, "learning_rate": 0.0005, "loss": 1.4508155584335327, "step": 22340 }, { "epoch": 1.7612140323216399, "grad_norm": 0.46034398674964905, "learning_rate": 0.0005, "loss": 1.3074859380722046, "step": 22341 }, { "epoch": 1.7612928655892786, "grad_norm": 0.6117244362831116, "learning_rate": 0.0005, "loss": 1.6277965307235718, "step": 22342 }, { "epoch": 1.7613716988569177, "grad_norm": 0.46729394793510437, "learning_rate": 0.0005, "loss": 1.3999946117401123, "step": 22343 }, { "epoch": 1.7614505321245566, "grad_norm": 0.5451481342315674, "learning_rate": 0.0005, "loss": 1.4255917072296143, "step": 22344 }, { "epoch": 1.7615293653921955, "grad_norm": 1.1945351362228394, "learning_rate": 0.0005, "loss": 1.4266573190689087, "step": 22345 }, { "epoch": 1.7616081986598344, "grad_norm": 0.57242351770401, "learning_rate": 0.0005, "loss": 1.6406140327453613, "step": 22346 }, { "epoch": 1.7616870319274733, "grad_norm": 0.5668070912361145, "learning_rate": 0.0005, "loss": 1.617643117904663, "step": 22347 }, { "epoch": 1.7617658651951125, "grad_norm": 1.259249210357666, "learning_rate": 0.0005, "loss": 1.5465753078460693, "step": 22348 }, { "epoch": 1.7618446984627512, "grad_norm": 0.5727540850639343, "learning_rate": 0.0005, "loss": 1.4773989915847778, "step": 22349 }, { "epoch": 1.7619235317303903, "grad_norm": 0.6158459186553955, "learning_rate": 0.0005, "loss": 1.6550476551055908, "step": 22350 }, { "epoch": 1.7620023649980292, "grad_norm": 0.4776817262172699, "learning_rate": 0.0005, "loss": 1.5327696800231934, "step": 22351 }, { "epoch": 1.7620811982656681, "grad_norm": 0.5130782127380371, "learning_rate": 0.0005, "loss": 1.4848848581314087, "step": 22352 }, { "epoch": 1.762160031533307, "grad_norm": 0.504220724105835, "learning_rate": 0.0005, "loss": 1.4600739479064941, "step": 22353 }, { "epoch": 1.762238864800946, "grad_norm": 0.49806663393974304, "learning_rate": 0.0005, "loss": 1.6501049995422363, "step": 22354 }, { "epoch": 1.762317698068585, "grad_norm": 0.5486788749694824, "learning_rate": 0.0005, "loss": 1.537137746810913, "step": 22355 }, { "epoch": 1.7623965313362238, "grad_norm": 0.6186773180961609, "learning_rate": 0.0005, "loss": 1.5505194664001465, "step": 22356 }, { "epoch": 1.762475364603863, "grad_norm": 0.5554373264312744, "learning_rate": 0.0005, "loss": 1.6376099586486816, "step": 22357 }, { "epoch": 1.7625541978715018, "grad_norm": 0.6250249147415161, "learning_rate": 0.0005, "loss": 1.5237313508987427, "step": 22358 }, { "epoch": 1.7626330311391407, "grad_norm": 0.5248473882675171, "learning_rate": 0.0005, "loss": 1.5295169353485107, "step": 22359 }, { "epoch": 1.7627118644067796, "grad_norm": 0.6358642578125, "learning_rate": 0.0005, "loss": 1.3770999908447266, "step": 22360 }, { "epoch": 1.7627906976744185, "grad_norm": 0.7253949046134949, "learning_rate": 0.0005, "loss": 1.3439422845840454, "step": 22361 }, { "epoch": 1.7628695309420577, "grad_norm": 0.5277262330055237, "learning_rate": 0.0005, "loss": 1.5792195796966553, "step": 22362 }, { "epoch": 1.7629483642096964, "grad_norm": 0.5116611123085022, "learning_rate": 0.0005, "loss": 1.4159828424453735, "step": 22363 }, { "epoch": 1.7630271974773355, "grad_norm": 0.5368548631668091, "learning_rate": 0.0005, "loss": 1.5228538513183594, "step": 22364 }, { "epoch": 1.7631060307449744, "grad_norm": 0.5793263912200928, "learning_rate": 0.0005, "loss": 1.6249078512191772, "step": 22365 }, { "epoch": 1.7631848640126133, "grad_norm": 0.5428863167762756, "learning_rate": 0.0005, "loss": 1.550909399986267, "step": 22366 }, { "epoch": 1.7632636972802522, "grad_norm": 0.6098292469978333, "learning_rate": 0.0005, "loss": 1.4785295724868774, "step": 22367 }, { "epoch": 1.7633425305478911, "grad_norm": 0.5442129373550415, "learning_rate": 0.0005, "loss": 1.4101784229278564, "step": 22368 }, { "epoch": 1.7634213638155303, "grad_norm": 0.8873050808906555, "learning_rate": 0.0005, "loss": 1.530444860458374, "step": 22369 }, { "epoch": 1.763500197083169, "grad_norm": 0.5358284711837769, "learning_rate": 0.0005, "loss": 1.4144642353057861, "step": 22370 }, { "epoch": 1.763579030350808, "grad_norm": 0.47552090883255005, "learning_rate": 0.0005, "loss": 1.6565868854522705, "step": 22371 }, { "epoch": 1.763657863618447, "grad_norm": 0.5218567252159119, "learning_rate": 0.0005, "loss": 1.5973217487335205, "step": 22372 }, { "epoch": 1.763736696886086, "grad_norm": 0.5348984599113464, "learning_rate": 0.0005, "loss": 1.4865477085113525, "step": 22373 }, { "epoch": 1.7638155301537248, "grad_norm": 0.5375361442565918, "learning_rate": 0.0005, "loss": 1.719369649887085, "step": 22374 }, { "epoch": 1.7638943634213637, "grad_norm": 0.5850346088409424, "learning_rate": 0.0005, "loss": 1.6096978187561035, "step": 22375 }, { "epoch": 1.7639731966890029, "grad_norm": 0.5067002177238464, "learning_rate": 0.0005, "loss": 1.4289970397949219, "step": 22376 }, { "epoch": 1.7640520299566416, "grad_norm": 0.6123795509338379, "learning_rate": 0.0005, "loss": 1.4939908981323242, "step": 22377 }, { "epoch": 1.7641308632242807, "grad_norm": 0.47053709626197815, "learning_rate": 0.0005, "loss": 1.6675543785095215, "step": 22378 }, { "epoch": 1.7642096964919196, "grad_norm": 0.675641655921936, "learning_rate": 0.0005, "loss": 1.6342486143112183, "step": 22379 }, { "epoch": 1.7642885297595585, "grad_norm": 0.5595127940177917, "learning_rate": 0.0005, "loss": 1.458911657333374, "step": 22380 }, { "epoch": 1.7643673630271974, "grad_norm": 0.465631365776062, "learning_rate": 0.0005, "loss": 1.434987187385559, "step": 22381 }, { "epoch": 1.7644461962948363, "grad_norm": 0.49898114800453186, "learning_rate": 0.0005, "loss": 1.489607334136963, "step": 22382 }, { "epoch": 1.7645250295624755, "grad_norm": 0.5012046694755554, "learning_rate": 0.0005, "loss": 1.4904123544692993, "step": 22383 }, { "epoch": 1.7646038628301142, "grad_norm": 0.5084783434867859, "learning_rate": 0.0005, "loss": 1.7705204486846924, "step": 22384 }, { "epoch": 1.7646826960977533, "grad_norm": 0.5355629920959473, "learning_rate": 0.0005, "loss": 1.553009271621704, "step": 22385 }, { "epoch": 1.7647615293653922, "grad_norm": 0.5752256512641907, "learning_rate": 0.0005, "loss": 1.6583255529403687, "step": 22386 }, { "epoch": 1.7648403626330311, "grad_norm": 0.5445222854614258, "learning_rate": 0.0005, "loss": 1.5472419261932373, "step": 22387 }, { "epoch": 1.76491919590067, "grad_norm": 0.5841153860092163, "learning_rate": 0.0005, "loss": 1.659433126449585, "step": 22388 }, { "epoch": 1.764998029168309, "grad_norm": 0.4755266606807709, "learning_rate": 0.0005, "loss": 1.409329891204834, "step": 22389 }, { "epoch": 1.765076862435948, "grad_norm": 0.5875340104103088, "learning_rate": 0.0005, "loss": 1.629259467124939, "step": 22390 }, { "epoch": 1.7651556957035868, "grad_norm": 0.7553720474243164, "learning_rate": 0.0005, "loss": 1.4778591394424438, "step": 22391 }, { "epoch": 1.765234528971226, "grad_norm": 0.48392432928085327, "learning_rate": 0.0005, "loss": 1.491405725479126, "step": 22392 }, { "epoch": 1.7653133622388648, "grad_norm": 0.5609515905380249, "learning_rate": 0.0005, "loss": 1.634131908416748, "step": 22393 }, { "epoch": 1.7653921955065037, "grad_norm": 0.54167240858078, "learning_rate": 0.0005, "loss": 1.4053781032562256, "step": 22394 }, { "epoch": 1.7654710287741426, "grad_norm": 0.5639134049415588, "learning_rate": 0.0005, "loss": 1.5644172430038452, "step": 22395 }, { "epoch": 1.7655498620417815, "grad_norm": 0.5848146080970764, "learning_rate": 0.0005, "loss": 1.5880706310272217, "step": 22396 }, { "epoch": 1.7656286953094207, "grad_norm": 0.5651455521583557, "learning_rate": 0.0005, "loss": 1.6830475330352783, "step": 22397 }, { "epoch": 1.7657075285770594, "grad_norm": 0.4502131938934326, "learning_rate": 0.0005, "loss": 1.4665287733078003, "step": 22398 }, { "epoch": 1.7657863618446985, "grad_norm": 1.446658968925476, "learning_rate": 0.0005, "loss": 1.5392587184906006, "step": 22399 }, { "epoch": 1.7658651951123374, "grad_norm": 0.6254510879516602, "learning_rate": 0.0005, "loss": 1.3733539581298828, "step": 22400 }, { "epoch": 1.7659440283799763, "grad_norm": 0.5083229541778564, "learning_rate": 0.0005, "loss": 1.5845232009887695, "step": 22401 }, { "epoch": 1.7660228616476152, "grad_norm": 0.6324080228805542, "learning_rate": 0.0005, "loss": 1.5324512720108032, "step": 22402 }, { "epoch": 1.7661016949152541, "grad_norm": 0.5775660276412964, "learning_rate": 0.0005, "loss": 1.4162490367889404, "step": 22403 }, { "epoch": 1.7661805281828933, "grad_norm": 0.4715989828109741, "learning_rate": 0.0005, "loss": 1.4572006464004517, "step": 22404 }, { "epoch": 1.766259361450532, "grad_norm": 0.49462810158729553, "learning_rate": 0.0005, "loss": 1.3851443529129028, "step": 22405 }, { "epoch": 1.766338194718171, "grad_norm": 0.5498688817024231, "learning_rate": 0.0005, "loss": 1.6076023578643799, "step": 22406 }, { "epoch": 1.76641702798581, "grad_norm": 0.5152403712272644, "learning_rate": 0.0005, "loss": 1.5981793403625488, "step": 22407 }, { "epoch": 1.766495861253449, "grad_norm": 0.5598039627075195, "learning_rate": 0.0005, "loss": 1.451444387435913, "step": 22408 }, { "epoch": 1.766574694521088, "grad_norm": 0.4902159869670868, "learning_rate": 0.0005, "loss": 1.62644624710083, "step": 22409 }, { "epoch": 1.7666535277887268, "grad_norm": 0.5835474729537964, "learning_rate": 0.0005, "loss": 1.603759765625, "step": 22410 }, { "epoch": 1.7667323610563659, "grad_norm": 0.4420822560787201, "learning_rate": 0.0005, "loss": 1.4927845001220703, "step": 22411 }, { "epoch": 1.7668111943240046, "grad_norm": 0.5014759302139282, "learning_rate": 0.0005, "loss": 1.5356884002685547, "step": 22412 }, { "epoch": 1.7668900275916437, "grad_norm": 0.5878500938415527, "learning_rate": 0.0005, "loss": 1.6057301759719849, "step": 22413 }, { "epoch": 1.7669688608592826, "grad_norm": 0.7305504083633423, "learning_rate": 0.0005, "loss": 1.7032544612884521, "step": 22414 }, { "epoch": 1.7670476941269215, "grad_norm": 0.5638360381126404, "learning_rate": 0.0005, "loss": 1.7426105737686157, "step": 22415 }, { "epoch": 1.7671265273945607, "grad_norm": 0.5546634793281555, "learning_rate": 0.0005, "loss": 1.5749742984771729, "step": 22416 }, { "epoch": 1.7672053606621994, "grad_norm": 0.41997581720352173, "learning_rate": 0.0005, "loss": 1.3052396774291992, "step": 22417 }, { "epoch": 1.7672841939298385, "grad_norm": 0.6159180998802185, "learning_rate": 0.0005, "loss": 1.5378895998001099, "step": 22418 }, { "epoch": 1.7673630271974772, "grad_norm": 0.733204185962677, "learning_rate": 0.0005, "loss": 1.4003245830535889, "step": 22419 }, { "epoch": 1.7674418604651163, "grad_norm": 0.4949570894241333, "learning_rate": 0.0005, "loss": 1.5518290996551514, "step": 22420 }, { "epoch": 1.7675206937327552, "grad_norm": 0.5290282368659973, "learning_rate": 0.0005, "loss": 1.4630963802337646, "step": 22421 }, { "epoch": 1.7675995270003941, "grad_norm": 0.4219824969768524, "learning_rate": 0.0005, "loss": 1.3722143173217773, "step": 22422 }, { "epoch": 1.7676783602680333, "grad_norm": 0.6490407586097717, "learning_rate": 0.0005, "loss": 1.4383207559585571, "step": 22423 }, { "epoch": 1.767757193535672, "grad_norm": 0.537717878818512, "learning_rate": 0.0005, "loss": 1.706805944442749, "step": 22424 }, { "epoch": 1.767836026803311, "grad_norm": 0.555168867111206, "learning_rate": 0.0005, "loss": 1.6903542280197144, "step": 22425 }, { "epoch": 1.7679148600709498, "grad_norm": 0.49887093901634216, "learning_rate": 0.0005, "loss": 1.447352647781372, "step": 22426 }, { "epoch": 1.767993693338589, "grad_norm": 0.5744181871414185, "learning_rate": 0.0005, "loss": 1.4548406600952148, "step": 22427 }, { "epoch": 1.7680725266062278, "grad_norm": 0.3987027704715729, "learning_rate": 0.0005, "loss": 1.4391778707504272, "step": 22428 }, { "epoch": 1.7681513598738667, "grad_norm": 0.957760751247406, "learning_rate": 0.0005, "loss": 1.7231365442276, "step": 22429 }, { "epoch": 1.7682301931415059, "grad_norm": 0.3791276216506958, "learning_rate": 0.0005, "loss": 1.3210108280181885, "step": 22430 }, { "epoch": 1.7683090264091446, "grad_norm": 0.5953080654144287, "learning_rate": 0.0005, "loss": 1.5449039936065674, "step": 22431 }, { "epoch": 1.7683878596767837, "grad_norm": 0.46106502413749695, "learning_rate": 0.0005, "loss": 1.329751968383789, "step": 22432 }, { "epoch": 1.7684666929444224, "grad_norm": 0.5788050889968872, "learning_rate": 0.0005, "loss": 1.5970149040222168, "step": 22433 }, { "epoch": 1.7685455262120615, "grad_norm": 0.590456485748291, "learning_rate": 0.0005, "loss": 1.6055750846862793, "step": 22434 }, { "epoch": 1.7686243594797004, "grad_norm": 0.5004450678825378, "learning_rate": 0.0005, "loss": 1.3021130561828613, "step": 22435 }, { "epoch": 1.7687031927473393, "grad_norm": 0.4983404576778412, "learning_rate": 0.0005, "loss": 1.3856744766235352, "step": 22436 }, { "epoch": 1.7687820260149785, "grad_norm": 0.6127499341964722, "learning_rate": 0.0005, "loss": 1.5067082643508911, "step": 22437 }, { "epoch": 1.7688608592826172, "grad_norm": 0.5194012522697449, "learning_rate": 0.0005, "loss": 1.4192991256713867, "step": 22438 }, { "epoch": 1.7689396925502563, "grad_norm": 0.5060303211212158, "learning_rate": 0.0005, "loss": 1.6596616506576538, "step": 22439 }, { "epoch": 1.7690185258178952, "grad_norm": 0.6112472414970398, "learning_rate": 0.0005, "loss": 1.5024504661560059, "step": 22440 }, { "epoch": 1.7690973590855341, "grad_norm": 0.4724609851837158, "learning_rate": 0.0005, "loss": 1.692133903503418, "step": 22441 }, { "epoch": 1.769176192353173, "grad_norm": 0.7028041481971741, "learning_rate": 0.0005, "loss": 1.5876212120056152, "step": 22442 }, { "epoch": 1.769255025620812, "grad_norm": 0.5731878876686096, "learning_rate": 0.0005, "loss": 1.4472498893737793, "step": 22443 }, { "epoch": 1.769333858888451, "grad_norm": 0.5139586925506592, "learning_rate": 0.0005, "loss": 1.5232203006744385, "step": 22444 }, { "epoch": 1.7694126921560898, "grad_norm": 0.5651959776878357, "learning_rate": 0.0005, "loss": 1.5108788013458252, "step": 22445 }, { "epoch": 1.769491525423729, "grad_norm": 0.5058184266090393, "learning_rate": 0.0005, "loss": 1.5630484819412231, "step": 22446 }, { "epoch": 1.7695703586913678, "grad_norm": 0.5844562649726868, "learning_rate": 0.0005, "loss": 1.87043035030365, "step": 22447 }, { "epoch": 1.7696491919590067, "grad_norm": 0.5493311882019043, "learning_rate": 0.0005, "loss": 1.5696687698364258, "step": 22448 }, { "epoch": 1.7697280252266456, "grad_norm": 0.6024640202522278, "learning_rate": 0.0005, "loss": 1.632359504699707, "step": 22449 }, { "epoch": 1.7698068584942845, "grad_norm": 0.5440139770507812, "learning_rate": 0.0005, "loss": 1.6335138082504272, "step": 22450 }, { "epoch": 1.7698856917619237, "grad_norm": 0.8682785034179688, "learning_rate": 0.0005, "loss": 1.26123046875, "step": 22451 }, { "epoch": 1.7699645250295624, "grad_norm": 0.4929570257663727, "learning_rate": 0.0005, "loss": 1.5683040618896484, "step": 22452 }, { "epoch": 1.7700433582972015, "grad_norm": 0.6970149874687195, "learning_rate": 0.0005, "loss": 1.5951805114746094, "step": 22453 }, { "epoch": 1.7701221915648404, "grad_norm": 0.527478814125061, "learning_rate": 0.0005, "loss": 1.4003299474716187, "step": 22454 }, { "epoch": 1.7702010248324793, "grad_norm": 0.5268338918685913, "learning_rate": 0.0005, "loss": 1.5084823369979858, "step": 22455 }, { "epoch": 1.7702798581001182, "grad_norm": 0.5725881457328796, "learning_rate": 0.0005, "loss": 1.485783338546753, "step": 22456 }, { "epoch": 1.7703586913677571, "grad_norm": 0.5202808380126953, "learning_rate": 0.0005, "loss": 1.648612380027771, "step": 22457 }, { "epoch": 1.7704375246353963, "grad_norm": 0.7294802069664001, "learning_rate": 0.0005, "loss": 1.5086848735809326, "step": 22458 }, { "epoch": 1.770516357903035, "grad_norm": 0.49991655349731445, "learning_rate": 0.0005, "loss": 1.4075331687927246, "step": 22459 }, { "epoch": 1.770595191170674, "grad_norm": 0.5370447635650635, "learning_rate": 0.0005, "loss": 1.667099952697754, "step": 22460 }, { "epoch": 1.770674024438313, "grad_norm": 0.502500593662262, "learning_rate": 0.0005, "loss": 1.6063642501831055, "step": 22461 }, { "epoch": 1.770752857705952, "grad_norm": 0.5728196501731873, "learning_rate": 0.0005, "loss": 1.3771086931228638, "step": 22462 }, { "epoch": 1.7708316909735908, "grad_norm": 0.5621744990348816, "learning_rate": 0.0005, "loss": 1.6291673183441162, "step": 22463 }, { "epoch": 1.7709105242412297, "grad_norm": 0.6074678301811218, "learning_rate": 0.0005, "loss": 1.588512659072876, "step": 22464 }, { "epoch": 1.7709893575088689, "grad_norm": 0.7720529437065125, "learning_rate": 0.0005, "loss": 1.6511502265930176, "step": 22465 }, { "epoch": 1.7710681907765076, "grad_norm": 0.5472240447998047, "learning_rate": 0.0005, "loss": 1.5418041944503784, "step": 22466 }, { "epoch": 1.7711470240441467, "grad_norm": 0.4986798167228699, "learning_rate": 0.0005, "loss": 1.6157466173171997, "step": 22467 }, { "epoch": 1.7712258573117856, "grad_norm": 0.5896509289741516, "learning_rate": 0.0005, "loss": 1.4885507822036743, "step": 22468 }, { "epoch": 1.7713046905794245, "grad_norm": 0.4964873492717743, "learning_rate": 0.0005, "loss": 1.8478004932403564, "step": 22469 }, { "epoch": 1.7713835238470634, "grad_norm": 0.6016989946365356, "learning_rate": 0.0005, "loss": 1.6146531105041504, "step": 22470 }, { "epoch": 1.7714623571147023, "grad_norm": 0.4647643268108368, "learning_rate": 0.0005, "loss": 1.509155035018921, "step": 22471 }, { "epoch": 1.7715411903823415, "grad_norm": 0.6466583013534546, "learning_rate": 0.0005, "loss": 1.3677685260772705, "step": 22472 }, { "epoch": 1.7716200236499802, "grad_norm": 0.508104681968689, "learning_rate": 0.0005, "loss": 1.4779479503631592, "step": 22473 }, { "epoch": 1.7716988569176193, "grad_norm": 0.620344877243042, "learning_rate": 0.0005, "loss": 1.7060449123382568, "step": 22474 }, { "epoch": 1.7717776901852582, "grad_norm": 0.49286723136901855, "learning_rate": 0.0005, "loss": 1.431809425354004, "step": 22475 }, { "epoch": 1.7718565234528971, "grad_norm": 0.583734929561615, "learning_rate": 0.0005, "loss": 1.4546523094177246, "step": 22476 }, { "epoch": 1.771935356720536, "grad_norm": 0.4639899730682373, "learning_rate": 0.0005, "loss": 1.378558874130249, "step": 22477 }, { "epoch": 1.772014189988175, "grad_norm": 0.5574430227279663, "learning_rate": 0.0005, "loss": 1.4991061687469482, "step": 22478 }, { "epoch": 1.772093023255814, "grad_norm": 0.4821792542934418, "learning_rate": 0.0005, "loss": 1.4916306734085083, "step": 22479 }, { "epoch": 1.7721718565234528, "grad_norm": 0.41080600023269653, "learning_rate": 0.0005, "loss": 1.3219391107559204, "step": 22480 }, { "epoch": 1.772250689791092, "grad_norm": 1.2066575288772583, "learning_rate": 0.0005, "loss": 1.565443754196167, "step": 22481 }, { "epoch": 1.7723295230587308, "grad_norm": 0.5421777963638306, "learning_rate": 0.0005, "loss": 1.1945903301239014, "step": 22482 }, { "epoch": 1.7724083563263697, "grad_norm": 0.4913107752799988, "learning_rate": 0.0005, "loss": 1.377514362335205, "step": 22483 }, { "epoch": 1.7724871895940086, "grad_norm": 0.4861231744289398, "learning_rate": 0.0005, "loss": 1.3685377836227417, "step": 22484 }, { "epoch": 1.7725660228616476, "grad_norm": 0.5688515305519104, "learning_rate": 0.0005, "loss": 1.329089879989624, "step": 22485 }, { "epoch": 1.7726448561292867, "grad_norm": 0.5175274014472961, "learning_rate": 0.0005, "loss": 1.2918246984481812, "step": 22486 }, { "epoch": 1.7727236893969254, "grad_norm": 0.5719701647758484, "learning_rate": 0.0005, "loss": 1.3116440773010254, "step": 22487 }, { "epoch": 1.7728025226645645, "grad_norm": 0.4178825616836548, "learning_rate": 0.0005, "loss": 1.2580225467681885, "step": 22488 }, { "epoch": 1.7728813559322034, "grad_norm": 0.4510028064250946, "learning_rate": 0.0005, "loss": 1.3943814039230347, "step": 22489 }, { "epoch": 1.7729601891998423, "grad_norm": 0.5174756646156311, "learning_rate": 0.0005, "loss": 1.29452383518219, "step": 22490 }, { "epoch": 1.7730390224674812, "grad_norm": 0.5421199202537537, "learning_rate": 0.0005, "loss": 1.645904541015625, "step": 22491 }, { "epoch": 1.7731178557351202, "grad_norm": 0.6214456558227539, "learning_rate": 0.0005, "loss": 1.3134771585464478, "step": 22492 }, { "epoch": 1.7731966890027593, "grad_norm": 0.586601197719574, "learning_rate": 0.0005, "loss": 1.4282928705215454, "step": 22493 }, { "epoch": 1.773275522270398, "grad_norm": 0.5524228811264038, "learning_rate": 0.0005, "loss": 1.3129122257232666, "step": 22494 }, { "epoch": 1.773354355538037, "grad_norm": 0.5415006279945374, "learning_rate": 0.0005, "loss": 1.6790945529937744, "step": 22495 }, { "epoch": 1.773433188805676, "grad_norm": 0.5325291156768799, "learning_rate": 0.0005, "loss": 1.3542351722717285, "step": 22496 }, { "epoch": 1.773512022073315, "grad_norm": 0.43653205037117004, "learning_rate": 0.0005, "loss": 1.305090308189392, "step": 22497 }, { "epoch": 1.7735908553409538, "grad_norm": 0.4837966859340668, "learning_rate": 0.0005, "loss": 1.283416509628296, "step": 22498 }, { "epoch": 1.7736696886085928, "grad_norm": 0.5160357356071472, "learning_rate": 0.0005, "loss": 1.4932624101638794, "step": 22499 }, { "epoch": 1.7737485218762319, "grad_norm": 0.5255301594734192, "learning_rate": 0.0005, "loss": 1.4218908548355103, "step": 22500 }, { "epoch": 1.7738273551438706, "grad_norm": 0.609417200088501, "learning_rate": 0.0005, "loss": 1.4091650247573853, "step": 22501 }, { "epoch": 1.7739061884115097, "grad_norm": 0.513729453086853, "learning_rate": 0.0005, "loss": 1.5201139450073242, "step": 22502 }, { "epoch": 1.7739850216791486, "grad_norm": 0.46825313568115234, "learning_rate": 0.0005, "loss": 1.3843467235565186, "step": 22503 }, { "epoch": 1.7740638549467875, "grad_norm": 0.501670777797699, "learning_rate": 0.0005, "loss": 1.3437812328338623, "step": 22504 }, { "epoch": 1.7741426882144264, "grad_norm": 0.5071296691894531, "learning_rate": 0.0005, "loss": 1.3766388893127441, "step": 22505 }, { "epoch": 1.7742215214820654, "grad_norm": 0.5749173164367676, "learning_rate": 0.0005, "loss": 1.5245198011398315, "step": 22506 }, { "epoch": 1.7743003547497045, "grad_norm": 0.5311675667762756, "learning_rate": 0.0005, "loss": 1.1837671995162964, "step": 22507 }, { "epoch": 1.7743791880173432, "grad_norm": 0.4429663419723511, "learning_rate": 0.0005, "loss": 1.41860032081604, "step": 22508 }, { "epoch": 1.7744580212849823, "grad_norm": 0.6478672027587891, "learning_rate": 0.0005, "loss": 1.2229605913162231, "step": 22509 }, { "epoch": 1.7745368545526212, "grad_norm": 0.840254008769989, "learning_rate": 0.0005, "loss": 1.5903551578521729, "step": 22510 }, { "epoch": 1.7746156878202601, "grad_norm": 0.44504061341285706, "learning_rate": 0.0005, "loss": 1.23471200466156, "step": 22511 }, { "epoch": 1.774694521087899, "grad_norm": 0.5147503018379211, "learning_rate": 0.0005, "loss": 1.5065019130706787, "step": 22512 }, { "epoch": 1.774773354355538, "grad_norm": 0.5339493751525879, "learning_rate": 0.0005, "loss": 1.3762898445129395, "step": 22513 }, { "epoch": 1.774852187623177, "grad_norm": 0.6997901797294617, "learning_rate": 0.0005, "loss": 1.50685715675354, "step": 22514 }, { "epoch": 1.7749310208908158, "grad_norm": 0.5979337096214294, "learning_rate": 0.0005, "loss": 1.4451249837875366, "step": 22515 }, { "epoch": 1.775009854158455, "grad_norm": 0.5248516798019409, "learning_rate": 0.0005, "loss": 1.557275652885437, "step": 22516 }, { "epoch": 1.7750886874260938, "grad_norm": 0.6152183413505554, "learning_rate": 0.0005, "loss": 1.667576789855957, "step": 22517 }, { "epoch": 1.7751675206937327, "grad_norm": 0.609661877155304, "learning_rate": 0.0005, "loss": 1.3648689985275269, "step": 22518 }, { "epoch": 1.7752463539613716, "grad_norm": 0.5557528734207153, "learning_rate": 0.0005, "loss": 1.6224474906921387, "step": 22519 }, { "epoch": 1.7753251872290106, "grad_norm": 0.5636690855026245, "learning_rate": 0.0005, "loss": 1.4595143795013428, "step": 22520 }, { "epoch": 1.7754040204966497, "grad_norm": 0.5701295733451843, "learning_rate": 0.0005, "loss": 1.554677963256836, "step": 22521 }, { "epoch": 1.7754828537642884, "grad_norm": 0.524443507194519, "learning_rate": 0.0005, "loss": 1.5030148029327393, "step": 22522 }, { "epoch": 1.7755616870319275, "grad_norm": 0.5086710453033447, "learning_rate": 0.0005, "loss": 1.415223240852356, "step": 22523 }, { "epoch": 1.7756405202995664, "grad_norm": 0.5466631054878235, "learning_rate": 0.0005, "loss": 1.2871215343475342, "step": 22524 }, { "epoch": 1.7757193535672053, "grad_norm": 0.5750148296356201, "learning_rate": 0.0005, "loss": 1.7578403949737549, "step": 22525 }, { "epoch": 1.7757981868348443, "grad_norm": 0.5459619164466858, "learning_rate": 0.0005, "loss": 1.6260710954666138, "step": 22526 }, { "epoch": 1.7758770201024832, "grad_norm": 0.4639584720134735, "learning_rate": 0.0005, "loss": 1.4990320205688477, "step": 22527 }, { "epoch": 1.7759558533701223, "grad_norm": 0.5098311901092529, "learning_rate": 0.0005, "loss": 1.3815462589263916, "step": 22528 }, { "epoch": 1.776034686637761, "grad_norm": 0.5297479629516602, "learning_rate": 0.0005, "loss": 1.56084144115448, "step": 22529 }, { "epoch": 1.7761135199054001, "grad_norm": 0.5145705342292786, "learning_rate": 0.0005, "loss": 1.366114616394043, "step": 22530 }, { "epoch": 1.776192353173039, "grad_norm": 0.6877720952033997, "learning_rate": 0.0005, "loss": 1.5517454147338867, "step": 22531 }, { "epoch": 1.776271186440678, "grad_norm": 0.557102620601654, "learning_rate": 0.0005, "loss": 1.5200469493865967, "step": 22532 }, { "epoch": 1.776350019708317, "grad_norm": 0.5485517382621765, "learning_rate": 0.0005, "loss": 1.5417410135269165, "step": 22533 }, { "epoch": 1.7764288529759558, "grad_norm": 0.5513147711753845, "learning_rate": 0.0005, "loss": 1.3642406463623047, "step": 22534 }, { "epoch": 1.776507686243595, "grad_norm": 0.7637077569961548, "learning_rate": 0.0005, "loss": 1.4824509620666504, "step": 22535 }, { "epoch": 1.7765865195112336, "grad_norm": 0.41843006014823914, "learning_rate": 0.0005, "loss": 1.1897046566009521, "step": 22536 }, { "epoch": 1.7766653527788727, "grad_norm": 0.5405715703964233, "learning_rate": 0.0005, "loss": 1.5875905752182007, "step": 22537 }, { "epoch": 1.7767441860465116, "grad_norm": 0.5133263468742371, "learning_rate": 0.0005, "loss": 1.568833589553833, "step": 22538 }, { "epoch": 1.7768230193141505, "grad_norm": 0.4865667521953583, "learning_rate": 0.0005, "loss": 1.4725139141082764, "step": 22539 }, { "epoch": 1.7769018525817897, "grad_norm": 0.42663902044296265, "learning_rate": 0.0005, "loss": 1.3418962955474854, "step": 22540 }, { "epoch": 1.7769806858494284, "grad_norm": 0.777372419834137, "learning_rate": 0.0005, "loss": 1.6200435161590576, "step": 22541 }, { "epoch": 1.7770595191170675, "grad_norm": 0.5501112937927246, "learning_rate": 0.0005, "loss": 1.5209579467773438, "step": 22542 }, { "epoch": 1.7771383523847062, "grad_norm": 0.5402752757072449, "learning_rate": 0.0005, "loss": 1.4021129608154297, "step": 22543 }, { "epoch": 1.7772171856523453, "grad_norm": 0.5594871640205383, "learning_rate": 0.0005, "loss": 1.4524399042129517, "step": 22544 }, { "epoch": 1.7772960189199842, "grad_norm": 0.6118046641349792, "learning_rate": 0.0005, "loss": 1.3936560153961182, "step": 22545 }, { "epoch": 1.7773748521876231, "grad_norm": 0.524897575378418, "learning_rate": 0.0005, "loss": 1.3123146295547485, "step": 22546 }, { "epoch": 1.7774536854552623, "grad_norm": 0.5091708302497864, "learning_rate": 0.0005, "loss": 1.3370652198791504, "step": 22547 }, { "epoch": 1.777532518722901, "grad_norm": 0.5071724653244019, "learning_rate": 0.0005, "loss": 1.51499342918396, "step": 22548 }, { "epoch": 1.77761135199054, "grad_norm": 0.44698214530944824, "learning_rate": 0.0005, "loss": 1.4196217060089111, "step": 22549 }, { "epoch": 1.7776901852581788, "grad_norm": 0.545242965221405, "learning_rate": 0.0005, "loss": 1.3059821128845215, "step": 22550 }, { "epoch": 1.777769018525818, "grad_norm": 0.5285550355911255, "learning_rate": 0.0005, "loss": 1.2120596170425415, "step": 22551 }, { "epoch": 1.7778478517934568, "grad_norm": 0.5121293663978577, "learning_rate": 0.0005, "loss": 1.5249998569488525, "step": 22552 }, { "epoch": 1.7779266850610957, "grad_norm": 0.7666745185852051, "learning_rate": 0.0005, "loss": 1.4193360805511475, "step": 22553 }, { "epoch": 1.7780055183287349, "grad_norm": 0.8311735987663269, "learning_rate": 0.0005, "loss": 1.3086291551589966, "step": 22554 }, { "epoch": 1.7780843515963736, "grad_norm": 0.7082197070121765, "learning_rate": 0.0005, "loss": 1.5832264423370361, "step": 22555 }, { "epoch": 1.7781631848640127, "grad_norm": 0.5890167355537415, "learning_rate": 0.0005, "loss": 1.2997546195983887, "step": 22556 }, { "epoch": 1.7782420181316514, "grad_norm": 0.6376140117645264, "learning_rate": 0.0005, "loss": 1.429706335067749, "step": 22557 }, { "epoch": 1.7783208513992905, "grad_norm": 0.6991009712219238, "learning_rate": 0.0005, "loss": 1.5116618871688843, "step": 22558 }, { "epoch": 1.7783996846669294, "grad_norm": 0.5885045528411865, "learning_rate": 0.0005, "loss": 1.5729670524597168, "step": 22559 }, { "epoch": 1.7784785179345683, "grad_norm": 0.5372837781906128, "learning_rate": 0.0005, "loss": 1.548241376876831, "step": 22560 }, { "epoch": 1.7785573512022075, "grad_norm": 0.46147799491882324, "learning_rate": 0.0005, "loss": 1.3816941976547241, "step": 22561 }, { "epoch": 1.7786361844698462, "grad_norm": 0.5169636607170105, "learning_rate": 0.0005, "loss": 1.429211139678955, "step": 22562 }, { "epoch": 1.7787150177374853, "grad_norm": 0.5473636388778687, "learning_rate": 0.0005, "loss": 1.335101842880249, "step": 22563 }, { "epoch": 1.7787938510051242, "grad_norm": 0.4703125059604645, "learning_rate": 0.0005, "loss": 1.2713589668273926, "step": 22564 }, { "epoch": 1.7788726842727631, "grad_norm": 0.5968855619430542, "learning_rate": 0.0005, "loss": 1.4046564102172852, "step": 22565 }, { "epoch": 1.778951517540402, "grad_norm": 0.7165980935096741, "learning_rate": 0.0005, "loss": 1.3893332481384277, "step": 22566 }, { "epoch": 1.779030350808041, "grad_norm": 0.5636768341064453, "learning_rate": 0.0005, "loss": 1.340556263923645, "step": 22567 }, { "epoch": 1.77910918407568, "grad_norm": 0.5123860239982605, "learning_rate": 0.0005, "loss": 1.2451097965240479, "step": 22568 }, { "epoch": 1.7791880173433188, "grad_norm": 0.4712737500667572, "learning_rate": 0.0005, "loss": 1.4036892652511597, "step": 22569 }, { "epoch": 1.779266850610958, "grad_norm": 0.6069917678833008, "learning_rate": 0.0005, "loss": 1.3108758926391602, "step": 22570 }, { "epoch": 1.7793456838785968, "grad_norm": 0.5412351489067078, "learning_rate": 0.0005, "loss": 1.5227081775665283, "step": 22571 }, { "epoch": 1.7794245171462357, "grad_norm": 0.5224571824073792, "learning_rate": 0.0005, "loss": 1.4841549396514893, "step": 22572 }, { "epoch": 1.7795033504138746, "grad_norm": 0.5066606402397156, "learning_rate": 0.0005, "loss": 1.4441896677017212, "step": 22573 }, { "epoch": 1.7795821836815136, "grad_norm": 0.534496545791626, "learning_rate": 0.0005, "loss": 1.2852952480316162, "step": 22574 }, { "epoch": 1.7796610169491527, "grad_norm": 0.5496341586112976, "learning_rate": 0.0005, "loss": 1.5493409633636475, "step": 22575 }, { "epoch": 1.7797398502167914, "grad_norm": 0.47307050228118896, "learning_rate": 0.0005, "loss": 1.2984859943389893, "step": 22576 }, { "epoch": 1.7798186834844305, "grad_norm": 0.527742326259613, "learning_rate": 0.0005, "loss": 1.2120471000671387, "step": 22577 }, { "epoch": 1.7798975167520694, "grad_norm": 0.6982256174087524, "learning_rate": 0.0005, "loss": 1.1261160373687744, "step": 22578 }, { "epoch": 1.7799763500197083, "grad_norm": 0.6203644275665283, "learning_rate": 0.0005, "loss": 1.4086838960647583, "step": 22579 }, { "epoch": 1.7800551832873472, "grad_norm": 0.4899720847606659, "learning_rate": 0.0005, "loss": 1.377353549003601, "step": 22580 }, { "epoch": 1.7801340165549862, "grad_norm": 0.5147315263748169, "learning_rate": 0.0005, "loss": 1.5417636632919312, "step": 22581 }, { "epoch": 1.7802128498226253, "grad_norm": 0.551360547542572, "learning_rate": 0.0005, "loss": 1.8457257747650146, "step": 22582 }, { "epoch": 1.780291683090264, "grad_norm": 0.50001460313797, "learning_rate": 0.0005, "loss": 1.3034873008728027, "step": 22583 }, { "epoch": 1.780370516357903, "grad_norm": 0.518866240978241, "learning_rate": 0.0005, "loss": 1.5414973497390747, "step": 22584 }, { "epoch": 1.780449349625542, "grad_norm": 0.47673317790031433, "learning_rate": 0.0005, "loss": 1.4100725650787354, "step": 22585 }, { "epoch": 1.780528182893181, "grad_norm": 1.320175051689148, "learning_rate": 0.0005, "loss": 1.561930775642395, "step": 22586 }, { "epoch": 1.7806070161608198, "grad_norm": 0.564521849155426, "learning_rate": 0.0005, "loss": 1.3375141620635986, "step": 22587 }, { "epoch": 1.7806858494284588, "grad_norm": 0.4858275055885315, "learning_rate": 0.0005, "loss": 1.3196375370025635, "step": 22588 }, { "epoch": 1.780764682696098, "grad_norm": 0.6525586247444153, "learning_rate": 0.0005, "loss": 1.5521175861358643, "step": 22589 }, { "epoch": 1.7808435159637366, "grad_norm": 0.7114933729171753, "learning_rate": 0.0005, "loss": 1.324035882949829, "step": 22590 }, { "epoch": 1.7809223492313757, "grad_norm": 0.4803784191608429, "learning_rate": 0.0005, "loss": 1.2585880756378174, "step": 22591 }, { "epoch": 1.7810011824990146, "grad_norm": 0.5559855699539185, "learning_rate": 0.0005, "loss": 1.4457485675811768, "step": 22592 }, { "epoch": 1.7810800157666535, "grad_norm": 0.4954228699207306, "learning_rate": 0.0005, "loss": 1.4599781036376953, "step": 22593 }, { "epoch": 1.7811588490342924, "grad_norm": 0.5464569926261902, "learning_rate": 0.0005, "loss": 1.3029552698135376, "step": 22594 }, { "epoch": 1.7812376823019314, "grad_norm": 0.47880640625953674, "learning_rate": 0.0005, "loss": 1.2407307624816895, "step": 22595 }, { "epoch": 1.7813165155695705, "grad_norm": 0.6154198050498962, "learning_rate": 0.0005, "loss": 1.422410249710083, "step": 22596 }, { "epoch": 1.7813953488372092, "grad_norm": 0.467203289270401, "learning_rate": 0.0005, "loss": 1.3491003513336182, "step": 22597 }, { "epoch": 1.7814741821048483, "grad_norm": 0.46166113018989563, "learning_rate": 0.0005, "loss": 1.2626818418502808, "step": 22598 }, { "epoch": 1.7815530153724872, "grad_norm": 0.5349207520484924, "learning_rate": 0.0005, "loss": 1.5995166301727295, "step": 22599 }, { "epoch": 1.7816318486401261, "grad_norm": 0.6608578562736511, "learning_rate": 0.0005, "loss": 1.5337104797363281, "step": 22600 }, { "epoch": 1.781710681907765, "grad_norm": 0.6008841395378113, "learning_rate": 0.0005, "loss": 1.4024739265441895, "step": 22601 }, { "epoch": 1.781789515175404, "grad_norm": 0.5112203359603882, "learning_rate": 0.0005, "loss": 1.390628457069397, "step": 22602 }, { "epoch": 1.781868348443043, "grad_norm": 0.7468284368515015, "learning_rate": 0.0005, "loss": 1.4332078695297241, "step": 22603 }, { "epoch": 1.7819471817106818, "grad_norm": 0.8223575353622437, "learning_rate": 0.0005, "loss": 1.412862777709961, "step": 22604 }, { "epoch": 1.782026014978321, "grad_norm": 0.591698169708252, "learning_rate": 0.0005, "loss": 1.4674363136291504, "step": 22605 }, { "epoch": 1.7821048482459598, "grad_norm": 0.5472087860107422, "learning_rate": 0.0005, "loss": 1.5071463584899902, "step": 22606 }, { "epoch": 1.7821836815135987, "grad_norm": 0.496693879365921, "learning_rate": 0.0005, "loss": 1.2026188373565674, "step": 22607 }, { "epoch": 1.7822625147812377, "grad_norm": 0.5536317825317383, "learning_rate": 0.0005, "loss": 1.562056064605713, "step": 22608 }, { "epoch": 1.7823413480488766, "grad_norm": 0.5241742134094238, "learning_rate": 0.0005, "loss": 1.5335228443145752, "step": 22609 }, { "epoch": 1.7824201813165157, "grad_norm": 0.6186166405677795, "learning_rate": 0.0005, "loss": 1.5679244995117188, "step": 22610 }, { "epoch": 1.7824990145841544, "grad_norm": 0.5458030700683594, "learning_rate": 0.0005, "loss": 1.582129716873169, "step": 22611 }, { "epoch": 1.7825778478517935, "grad_norm": 0.5929936766624451, "learning_rate": 0.0005, "loss": 1.6531974077224731, "step": 22612 }, { "epoch": 1.7826566811194324, "grad_norm": 0.4663059413433075, "learning_rate": 0.0005, "loss": 1.433281660079956, "step": 22613 }, { "epoch": 1.7827355143870713, "grad_norm": 0.5007315278053284, "learning_rate": 0.0005, "loss": 1.57509183883667, "step": 22614 }, { "epoch": 1.7828143476547103, "grad_norm": 0.3831201493740082, "learning_rate": 0.0005, "loss": 1.2968385219573975, "step": 22615 }, { "epoch": 1.7828931809223492, "grad_norm": 0.5199676752090454, "learning_rate": 0.0005, "loss": 1.4040848016738892, "step": 22616 }, { "epoch": 1.7829720141899883, "grad_norm": 0.6100780367851257, "learning_rate": 0.0005, "loss": 1.3737585544586182, "step": 22617 }, { "epoch": 1.783050847457627, "grad_norm": 0.3808830976486206, "learning_rate": 0.0005, "loss": 1.3558456897735596, "step": 22618 }, { "epoch": 1.7831296807252661, "grad_norm": 0.6873778104782104, "learning_rate": 0.0005, "loss": 1.3483960628509521, "step": 22619 }, { "epoch": 1.783208513992905, "grad_norm": 0.46942630410194397, "learning_rate": 0.0005, "loss": 1.5650768280029297, "step": 22620 }, { "epoch": 1.783287347260544, "grad_norm": 0.5494170188903809, "learning_rate": 0.0005, "loss": 1.3854506015777588, "step": 22621 }, { "epoch": 1.7833661805281829, "grad_norm": 0.5501565337181091, "learning_rate": 0.0005, "loss": 1.5173949003219604, "step": 22622 }, { "epoch": 1.7834450137958218, "grad_norm": 0.5114428997039795, "learning_rate": 0.0005, "loss": 1.3841650485992432, "step": 22623 }, { "epoch": 1.783523847063461, "grad_norm": 0.4736252427101135, "learning_rate": 0.0005, "loss": 1.200829029083252, "step": 22624 }, { "epoch": 1.7836026803310996, "grad_norm": 0.5191583633422852, "learning_rate": 0.0005, "loss": 1.5838507413864136, "step": 22625 }, { "epoch": 1.7836815135987387, "grad_norm": 0.6654964685440063, "learning_rate": 0.0005, "loss": 1.866034746170044, "step": 22626 }, { "epoch": 1.7837603468663776, "grad_norm": 0.665036141872406, "learning_rate": 0.0005, "loss": 1.67592191696167, "step": 22627 }, { "epoch": 1.7838391801340165, "grad_norm": 0.6733751893043518, "learning_rate": 0.0005, "loss": 1.618419885635376, "step": 22628 }, { "epoch": 1.7839180134016555, "grad_norm": 1.6290446519851685, "learning_rate": 0.0005, "loss": 1.363039255142212, "step": 22629 }, { "epoch": 1.7839968466692944, "grad_norm": 0.4954422414302826, "learning_rate": 0.0005, "loss": 1.4160306453704834, "step": 22630 }, { "epoch": 1.7840756799369335, "grad_norm": 0.4657799005508423, "learning_rate": 0.0005, "loss": 1.409473180770874, "step": 22631 }, { "epoch": 1.7841545132045722, "grad_norm": 0.6873657703399658, "learning_rate": 0.0005, "loss": 1.5164926052093506, "step": 22632 }, { "epoch": 1.7842333464722113, "grad_norm": 0.7362475991249084, "learning_rate": 0.0005, "loss": 1.4877926111221313, "step": 22633 }, { "epoch": 1.7843121797398502, "grad_norm": 0.5032035708427429, "learning_rate": 0.0005, "loss": 1.3600668907165527, "step": 22634 }, { "epoch": 1.7843910130074891, "grad_norm": 0.44814613461494446, "learning_rate": 0.0005, "loss": 1.3024427890777588, "step": 22635 }, { "epoch": 1.784469846275128, "grad_norm": 0.5786076784133911, "learning_rate": 0.0005, "loss": 1.5036227703094482, "step": 22636 }, { "epoch": 1.784548679542767, "grad_norm": 0.6058298945426941, "learning_rate": 0.0005, "loss": 1.3000810146331787, "step": 22637 }, { "epoch": 1.784627512810406, "grad_norm": 0.432034432888031, "learning_rate": 0.0005, "loss": 1.1636028289794922, "step": 22638 }, { "epoch": 1.7847063460780448, "grad_norm": 0.4706326425075531, "learning_rate": 0.0005, "loss": 1.3147028684616089, "step": 22639 }, { "epoch": 1.784785179345684, "grad_norm": 0.6265386939048767, "learning_rate": 0.0005, "loss": 1.222797155380249, "step": 22640 }, { "epoch": 1.7848640126133228, "grad_norm": 0.38462182879447937, "learning_rate": 0.0005, "loss": 1.210402488708496, "step": 22641 }, { "epoch": 1.7849428458809617, "grad_norm": 0.6613408327102661, "learning_rate": 0.0005, "loss": 1.1431007385253906, "step": 22642 }, { "epoch": 1.7850216791486007, "grad_norm": 0.5548549294471741, "learning_rate": 0.0005, "loss": 1.4024004936218262, "step": 22643 }, { "epoch": 1.7851005124162396, "grad_norm": 0.5519188642501831, "learning_rate": 0.0005, "loss": 1.3366615772247314, "step": 22644 }, { "epoch": 1.7851793456838787, "grad_norm": 0.3898884057998657, "learning_rate": 0.0005, "loss": 1.2926084995269775, "step": 22645 }, { "epoch": 1.7852581789515174, "grad_norm": 0.5563687086105347, "learning_rate": 0.0005, "loss": 1.6417236328125, "step": 22646 }, { "epoch": 1.7853370122191565, "grad_norm": 0.5727868676185608, "learning_rate": 0.0005, "loss": 1.4775797128677368, "step": 22647 }, { "epoch": 1.7854158454867954, "grad_norm": 0.5156089663505554, "learning_rate": 0.0005, "loss": 1.492578387260437, "step": 22648 }, { "epoch": 1.7854946787544344, "grad_norm": 0.5044112801551819, "learning_rate": 0.0005, "loss": 1.2632321119308472, "step": 22649 }, { "epoch": 1.7855735120220733, "grad_norm": 0.5394083261489868, "learning_rate": 0.0005, "loss": 1.203660249710083, "step": 22650 }, { "epoch": 1.7856523452897122, "grad_norm": 0.4708552956581116, "learning_rate": 0.0005, "loss": 1.5858708620071411, "step": 22651 }, { "epoch": 1.7857311785573513, "grad_norm": 0.5606760382652283, "learning_rate": 0.0005, "loss": 1.497512936592102, "step": 22652 }, { "epoch": 1.78581001182499, "grad_norm": 0.59232497215271, "learning_rate": 0.0005, "loss": 1.4367269277572632, "step": 22653 }, { "epoch": 1.7858888450926291, "grad_norm": 0.5860583186149597, "learning_rate": 0.0005, "loss": 1.471624732017517, "step": 22654 }, { "epoch": 1.785967678360268, "grad_norm": 0.5904543995857239, "learning_rate": 0.0005, "loss": 1.3720808029174805, "step": 22655 }, { "epoch": 1.786046511627907, "grad_norm": 0.6154714226722717, "learning_rate": 0.0005, "loss": 1.7087218761444092, "step": 22656 }, { "epoch": 1.786125344895546, "grad_norm": 0.5216888785362244, "learning_rate": 0.0005, "loss": 1.4333826303482056, "step": 22657 }, { "epoch": 1.7862041781631848, "grad_norm": 0.5119118094444275, "learning_rate": 0.0005, "loss": 1.3155403137207031, "step": 22658 }, { "epoch": 1.786283011430824, "grad_norm": 0.5137838125228882, "learning_rate": 0.0005, "loss": 1.5891728401184082, "step": 22659 }, { "epoch": 1.7863618446984626, "grad_norm": 0.8643196821212769, "learning_rate": 0.0005, "loss": 1.6298694610595703, "step": 22660 }, { "epoch": 1.7864406779661017, "grad_norm": 0.491737961769104, "learning_rate": 0.0005, "loss": 1.5137264728546143, "step": 22661 }, { "epoch": 1.7865195112337406, "grad_norm": 0.6832163333892822, "learning_rate": 0.0005, "loss": 1.5245506763458252, "step": 22662 }, { "epoch": 1.7865983445013796, "grad_norm": 0.6592420339584351, "learning_rate": 0.0005, "loss": 1.4637150764465332, "step": 22663 }, { "epoch": 1.7866771777690187, "grad_norm": 0.6180171370506287, "learning_rate": 0.0005, "loss": 1.358137607574463, "step": 22664 }, { "epoch": 1.7867560110366574, "grad_norm": 0.5453283786773682, "learning_rate": 0.0005, "loss": 1.2778319120407104, "step": 22665 }, { "epoch": 1.7868348443042965, "grad_norm": 0.47009384632110596, "learning_rate": 0.0005, "loss": 1.2617878913879395, "step": 22666 }, { "epoch": 1.7869136775719352, "grad_norm": 1.1083837747573853, "learning_rate": 0.0005, "loss": 1.4379832744598389, "step": 22667 }, { "epoch": 1.7869925108395743, "grad_norm": 0.48036786913871765, "learning_rate": 0.0005, "loss": 1.4805372953414917, "step": 22668 }, { "epoch": 1.7870713441072132, "grad_norm": 0.4883362650871277, "learning_rate": 0.0005, "loss": 1.5064380168914795, "step": 22669 }, { "epoch": 1.7871501773748522, "grad_norm": 0.5586695075035095, "learning_rate": 0.0005, "loss": 1.4481890201568604, "step": 22670 }, { "epoch": 1.7872290106424913, "grad_norm": 0.3990878760814667, "learning_rate": 0.0005, "loss": 1.1189970970153809, "step": 22671 }, { "epoch": 1.78730784391013, "grad_norm": 0.856768012046814, "learning_rate": 0.0005, "loss": 1.5087292194366455, "step": 22672 }, { "epoch": 1.7873866771777691, "grad_norm": 0.42583608627319336, "learning_rate": 0.0005, "loss": 1.3468928337097168, "step": 22673 }, { "epoch": 1.7874655104454078, "grad_norm": 0.5752578973770142, "learning_rate": 0.0005, "loss": 1.3949147462844849, "step": 22674 }, { "epoch": 1.787544343713047, "grad_norm": 0.5437924861907959, "learning_rate": 0.0005, "loss": 1.4353554248809814, "step": 22675 }, { "epoch": 1.7876231769806858, "grad_norm": 0.5813539624214172, "learning_rate": 0.0005, "loss": 1.7087600231170654, "step": 22676 }, { "epoch": 1.7877020102483248, "grad_norm": 0.5577717423439026, "learning_rate": 0.0005, "loss": 1.3834784030914307, "step": 22677 }, { "epoch": 1.787780843515964, "grad_norm": 0.567316472530365, "learning_rate": 0.0005, "loss": 1.5179413557052612, "step": 22678 }, { "epoch": 1.7878596767836026, "grad_norm": 0.5899013876914978, "learning_rate": 0.0005, "loss": 1.448240041732788, "step": 22679 }, { "epoch": 1.7879385100512417, "grad_norm": 0.6519467830657959, "learning_rate": 0.0005, "loss": 1.7001596689224243, "step": 22680 }, { "epoch": 1.7880173433188804, "grad_norm": 0.6173560619354248, "learning_rate": 0.0005, "loss": 1.5518794059753418, "step": 22681 }, { "epoch": 1.7880961765865195, "grad_norm": 0.5860998034477234, "learning_rate": 0.0005, "loss": 1.7477270364761353, "step": 22682 }, { "epoch": 1.7881750098541584, "grad_norm": 0.5184025764465332, "learning_rate": 0.0005, "loss": 1.6280431747436523, "step": 22683 }, { "epoch": 1.7882538431217974, "grad_norm": 0.6136748790740967, "learning_rate": 0.0005, "loss": 1.529400110244751, "step": 22684 }, { "epoch": 1.7883326763894365, "grad_norm": 0.5748792290687561, "learning_rate": 0.0005, "loss": 1.5276299715042114, "step": 22685 }, { "epoch": 1.7884115096570752, "grad_norm": 0.6051006317138672, "learning_rate": 0.0005, "loss": 1.5199280977249146, "step": 22686 }, { "epoch": 1.7884903429247143, "grad_norm": 0.46351107954978943, "learning_rate": 0.0005, "loss": 1.273559808731079, "step": 22687 }, { "epoch": 1.7885691761923532, "grad_norm": 0.45362377166748047, "learning_rate": 0.0005, "loss": 1.243173360824585, "step": 22688 }, { "epoch": 1.7886480094599921, "grad_norm": 0.5816758871078491, "learning_rate": 0.0005, "loss": 1.6855204105377197, "step": 22689 }, { "epoch": 1.788726842727631, "grad_norm": 0.5618335008621216, "learning_rate": 0.0005, "loss": 1.2486536502838135, "step": 22690 }, { "epoch": 1.78880567599527, "grad_norm": 0.44947585463523865, "learning_rate": 0.0005, "loss": 1.554518461227417, "step": 22691 }, { "epoch": 1.788884509262909, "grad_norm": 0.4801051914691925, "learning_rate": 0.0005, "loss": 1.3154504299163818, "step": 22692 }, { "epoch": 1.7889633425305478, "grad_norm": 0.5501791834831238, "learning_rate": 0.0005, "loss": 1.6070919036865234, "step": 22693 }, { "epoch": 1.789042175798187, "grad_norm": 0.4107869565486908, "learning_rate": 0.0005, "loss": 1.4233217239379883, "step": 22694 }, { "epoch": 1.7891210090658258, "grad_norm": 0.6355103850364685, "learning_rate": 0.0005, "loss": 1.4073797464370728, "step": 22695 }, { "epoch": 1.7891998423334647, "grad_norm": 0.5427471399307251, "learning_rate": 0.0005, "loss": 1.2534558773040771, "step": 22696 }, { "epoch": 1.7892786756011037, "grad_norm": 0.4658125936985016, "learning_rate": 0.0005, "loss": 1.363289713859558, "step": 22697 }, { "epoch": 1.7893575088687426, "grad_norm": 0.5894048810005188, "learning_rate": 0.0005, "loss": 1.3804731369018555, "step": 22698 }, { "epoch": 1.7894363421363817, "grad_norm": 0.631162166595459, "learning_rate": 0.0005, "loss": 1.4238908290863037, "step": 22699 }, { "epoch": 1.7895151754040204, "grad_norm": 0.6011555194854736, "learning_rate": 0.0005, "loss": 1.438248872756958, "step": 22700 }, { "epoch": 1.7895940086716595, "grad_norm": 0.42942652106285095, "learning_rate": 0.0005, "loss": 1.1552766561508179, "step": 22701 }, { "epoch": 1.7896728419392984, "grad_norm": 0.4960547089576721, "learning_rate": 0.0005, "loss": 1.3778729438781738, "step": 22702 }, { "epoch": 1.7897516752069373, "grad_norm": 0.4912598133087158, "learning_rate": 0.0005, "loss": 1.2175573110580444, "step": 22703 }, { "epoch": 1.7898305084745763, "grad_norm": 0.4655326008796692, "learning_rate": 0.0005, "loss": 1.3105220794677734, "step": 22704 }, { "epoch": 1.7899093417422152, "grad_norm": 0.6407281756401062, "learning_rate": 0.0005, "loss": 1.3994524478912354, "step": 22705 }, { "epoch": 1.7899881750098543, "grad_norm": 0.48105698823928833, "learning_rate": 0.0005, "loss": 1.2037492990493774, "step": 22706 }, { "epoch": 1.790067008277493, "grad_norm": 0.5040737390518188, "learning_rate": 0.0005, "loss": 1.298828363418579, "step": 22707 }, { "epoch": 1.7901458415451321, "grad_norm": 0.5239306092262268, "learning_rate": 0.0005, "loss": 1.3879914283752441, "step": 22708 }, { "epoch": 1.790224674812771, "grad_norm": 0.4891584813594818, "learning_rate": 0.0005, "loss": 1.269632339477539, "step": 22709 }, { "epoch": 1.79030350808041, "grad_norm": 0.548546552658081, "learning_rate": 0.0005, "loss": 1.2260197401046753, "step": 22710 }, { "epoch": 1.7903823413480489, "grad_norm": 0.5065268278121948, "learning_rate": 0.0005, "loss": 1.391162633895874, "step": 22711 }, { "epoch": 1.7904611746156878, "grad_norm": 0.5714570879936218, "learning_rate": 0.0005, "loss": 1.431518316268921, "step": 22712 }, { "epoch": 1.790540007883327, "grad_norm": 0.4832351505756378, "learning_rate": 0.0005, "loss": 1.3973580598831177, "step": 22713 }, { "epoch": 1.7906188411509656, "grad_norm": 0.5151243805885315, "learning_rate": 0.0005, "loss": 1.4258161783218384, "step": 22714 }, { "epoch": 1.7906976744186047, "grad_norm": 0.5607725977897644, "learning_rate": 0.0005, "loss": 1.5912648439407349, "step": 22715 }, { "epoch": 1.7907765076862436, "grad_norm": 0.7379305362701416, "learning_rate": 0.0005, "loss": 1.831409215927124, "step": 22716 }, { "epoch": 1.7908553409538825, "grad_norm": 0.6340919733047485, "learning_rate": 0.0005, "loss": 1.5873680114746094, "step": 22717 }, { "epoch": 1.7909341742215215, "grad_norm": 0.5038774013519287, "learning_rate": 0.0005, "loss": 1.5099965333938599, "step": 22718 }, { "epoch": 1.7910130074891604, "grad_norm": 0.632929801940918, "learning_rate": 0.0005, "loss": 1.462160587310791, "step": 22719 }, { "epoch": 1.7910918407567995, "grad_norm": 0.5866271257400513, "learning_rate": 0.0005, "loss": 1.315180778503418, "step": 22720 }, { "epoch": 1.7911706740244382, "grad_norm": 0.6883978843688965, "learning_rate": 0.0005, "loss": 1.6460354328155518, "step": 22721 }, { "epoch": 1.7912495072920773, "grad_norm": 0.5228068232536316, "learning_rate": 0.0005, "loss": 1.3895756006240845, "step": 22722 }, { "epoch": 1.7913283405597162, "grad_norm": 0.5831588506698608, "learning_rate": 0.0005, "loss": 1.5022063255310059, "step": 22723 }, { "epoch": 1.7914071738273551, "grad_norm": 0.6857739686965942, "learning_rate": 0.0005, "loss": 1.5151960849761963, "step": 22724 }, { "epoch": 1.791486007094994, "grad_norm": 0.47271212935447693, "learning_rate": 0.0005, "loss": 1.416717767715454, "step": 22725 }, { "epoch": 1.791564840362633, "grad_norm": 0.5038370490074158, "learning_rate": 0.0005, "loss": 1.5014015436172485, "step": 22726 }, { "epoch": 1.791643673630272, "grad_norm": 0.6411710381507874, "learning_rate": 0.0005, "loss": 1.3617390394210815, "step": 22727 }, { "epoch": 1.7917225068979108, "grad_norm": 0.5361503958702087, "learning_rate": 0.0005, "loss": 1.5359233617782593, "step": 22728 }, { "epoch": 1.79180134016555, "grad_norm": 0.6552590131759644, "learning_rate": 0.0005, "loss": 1.548129677772522, "step": 22729 }, { "epoch": 1.7918801734331888, "grad_norm": 0.5910602807998657, "learning_rate": 0.0005, "loss": 1.3779571056365967, "step": 22730 }, { "epoch": 1.7919590067008278, "grad_norm": 0.5409265756607056, "learning_rate": 0.0005, "loss": 1.5304068326950073, "step": 22731 }, { "epoch": 1.7920378399684667, "grad_norm": 0.5097289681434631, "learning_rate": 0.0005, "loss": 1.4608190059661865, "step": 22732 }, { "epoch": 1.7921166732361056, "grad_norm": 0.5358109474182129, "learning_rate": 0.0005, "loss": 1.5470051765441895, "step": 22733 }, { "epoch": 1.7921955065037447, "grad_norm": 0.6418728828430176, "learning_rate": 0.0005, "loss": 1.6001644134521484, "step": 22734 }, { "epoch": 1.7922743397713834, "grad_norm": 0.5672656893730164, "learning_rate": 0.0005, "loss": 1.6198933124542236, "step": 22735 }, { "epoch": 1.7923531730390225, "grad_norm": 0.730124831199646, "learning_rate": 0.0005, "loss": 1.5165331363677979, "step": 22736 }, { "epoch": 1.7924320063066614, "grad_norm": 0.5661762952804565, "learning_rate": 0.0005, "loss": 1.3497483730316162, "step": 22737 }, { "epoch": 1.7925108395743004, "grad_norm": 0.5155206918716431, "learning_rate": 0.0005, "loss": 1.329573392868042, "step": 22738 }, { "epoch": 1.7925896728419393, "grad_norm": 0.46795976161956787, "learning_rate": 0.0005, "loss": 1.586303949356079, "step": 22739 }, { "epoch": 1.7926685061095782, "grad_norm": 0.8151921033859253, "learning_rate": 0.0005, "loss": 1.752071499824524, "step": 22740 }, { "epoch": 1.7927473393772173, "grad_norm": 0.4586901366710663, "learning_rate": 0.0005, "loss": 1.4505659341812134, "step": 22741 }, { "epoch": 1.792826172644856, "grad_norm": 0.5267964005470276, "learning_rate": 0.0005, "loss": 1.4310400485992432, "step": 22742 }, { "epoch": 1.7929050059124951, "grad_norm": 0.5085208415985107, "learning_rate": 0.0005, "loss": 1.3103975057601929, "step": 22743 }, { "epoch": 1.792983839180134, "grad_norm": 0.5078415274620056, "learning_rate": 0.0005, "loss": 1.5712529420852661, "step": 22744 }, { "epoch": 1.793062672447773, "grad_norm": 0.5722546577453613, "learning_rate": 0.0005, "loss": 1.6200342178344727, "step": 22745 }, { "epoch": 1.7931415057154119, "grad_norm": 0.6528353095054626, "learning_rate": 0.0005, "loss": 1.6582589149475098, "step": 22746 }, { "epoch": 1.7932203389830508, "grad_norm": 0.48867887258529663, "learning_rate": 0.0005, "loss": 1.3125157356262207, "step": 22747 }, { "epoch": 1.79329917225069, "grad_norm": 0.7637954950332642, "learning_rate": 0.0005, "loss": 1.5467135906219482, "step": 22748 }, { "epoch": 1.7933780055183286, "grad_norm": 0.6327767372131348, "learning_rate": 0.0005, "loss": 1.5179023742675781, "step": 22749 }, { "epoch": 1.7934568387859677, "grad_norm": 0.48708897829055786, "learning_rate": 0.0005, "loss": 1.281712532043457, "step": 22750 }, { "epoch": 1.7935356720536066, "grad_norm": 0.47288796305656433, "learning_rate": 0.0005, "loss": 1.4175231456756592, "step": 22751 }, { "epoch": 1.7936145053212456, "grad_norm": 0.5897042155265808, "learning_rate": 0.0005, "loss": 1.693718433380127, "step": 22752 }, { "epoch": 1.7936933385888845, "grad_norm": 0.6211864948272705, "learning_rate": 0.0005, "loss": 1.2860558032989502, "step": 22753 }, { "epoch": 1.7937721718565234, "grad_norm": 0.49773770570755005, "learning_rate": 0.0005, "loss": 1.4176596403121948, "step": 22754 }, { "epoch": 1.7938510051241625, "grad_norm": 0.49402880668640137, "learning_rate": 0.0005, "loss": 1.5336580276489258, "step": 22755 }, { "epoch": 1.7939298383918012, "grad_norm": 0.5045387148857117, "learning_rate": 0.0005, "loss": 1.3532710075378418, "step": 22756 }, { "epoch": 1.7940086716594403, "grad_norm": 0.5568884015083313, "learning_rate": 0.0005, "loss": 1.4524626731872559, "step": 22757 }, { "epoch": 1.7940875049270792, "grad_norm": 0.4269793629646301, "learning_rate": 0.0005, "loss": 1.3485267162322998, "step": 22758 }, { "epoch": 1.7941663381947182, "grad_norm": 0.5362957715988159, "learning_rate": 0.0005, "loss": 1.355146884918213, "step": 22759 }, { "epoch": 1.794245171462357, "grad_norm": 0.5264495015144348, "learning_rate": 0.0005, "loss": 1.447123408317566, "step": 22760 }, { "epoch": 1.794324004729996, "grad_norm": 0.5910956859588623, "learning_rate": 0.0005, "loss": 1.4073357582092285, "step": 22761 }, { "epoch": 1.7944028379976351, "grad_norm": 0.5241232514381409, "learning_rate": 0.0005, "loss": 1.312151551246643, "step": 22762 }, { "epoch": 1.7944816712652738, "grad_norm": 0.5598194599151611, "learning_rate": 0.0005, "loss": 1.2639473676681519, "step": 22763 }, { "epoch": 1.794560504532913, "grad_norm": 0.6331833600997925, "learning_rate": 0.0005, "loss": 1.417029857635498, "step": 22764 }, { "epoch": 1.7946393378005518, "grad_norm": 0.5496374368667603, "learning_rate": 0.0005, "loss": 1.545253038406372, "step": 22765 }, { "epoch": 1.7947181710681908, "grad_norm": 0.5862923860549927, "learning_rate": 0.0005, "loss": 1.550729513168335, "step": 22766 }, { "epoch": 1.7947970043358297, "grad_norm": 0.5583064556121826, "learning_rate": 0.0005, "loss": 1.4212017059326172, "step": 22767 }, { "epoch": 1.7948758376034686, "grad_norm": 0.5093920230865479, "learning_rate": 0.0005, "loss": 1.3470656871795654, "step": 22768 }, { "epoch": 1.7949546708711077, "grad_norm": 0.648613691329956, "learning_rate": 0.0005, "loss": 1.3526333570480347, "step": 22769 }, { "epoch": 1.7950335041387464, "grad_norm": 0.4746837615966797, "learning_rate": 0.0005, "loss": 1.3648877143859863, "step": 22770 }, { "epoch": 1.7951123374063855, "grad_norm": 0.6930164098739624, "learning_rate": 0.0005, "loss": 1.5332298278808594, "step": 22771 }, { "epoch": 1.7951911706740245, "grad_norm": 0.6232282519340515, "learning_rate": 0.0005, "loss": 1.3297436237335205, "step": 22772 }, { "epoch": 1.7952700039416634, "grad_norm": 0.4554489552974701, "learning_rate": 0.0005, "loss": 1.3386707305908203, "step": 22773 }, { "epoch": 1.7953488372093023, "grad_norm": 0.4871816039085388, "learning_rate": 0.0005, "loss": 1.5327537059783936, "step": 22774 }, { "epoch": 1.7954276704769412, "grad_norm": 0.5568931698799133, "learning_rate": 0.0005, "loss": 1.3291666507720947, "step": 22775 }, { "epoch": 1.7955065037445803, "grad_norm": 0.4624127745628357, "learning_rate": 0.0005, "loss": 1.2545193433761597, "step": 22776 }, { "epoch": 1.795585337012219, "grad_norm": 0.510151207447052, "learning_rate": 0.0005, "loss": 1.5821090936660767, "step": 22777 }, { "epoch": 1.7956641702798581, "grad_norm": 0.4841904640197754, "learning_rate": 0.0005, "loss": 1.2389812469482422, "step": 22778 }, { "epoch": 1.795743003547497, "grad_norm": 0.6093188524246216, "learning_rate": 0.0005, "loss": 1.4601929187774658, "step": 22779 }, { "epoch": 1.795821836815136, "grad_norm": 0.5976606011390686, "learning_rate": 0.0005, "loss": 1.6915967464447021, "step": 22780 }, { "epoch": 1.795900670082775, "grad_norm": 0.49678343534469604, "learning_rate": 0.0005, "loss": 1.4935026168823242, "step": 22781 }, { "epoch": 1.7959795033504138, "grad_norm": 0.6039320826530457, "learning_rate": 0.0005, "loss": 1.177854299545288, "step": 22782 }, { "epoch": 1.796058336618053, "grad_norm": 0.5128666758537292, "learning_rate": 0.0005, "loss": 1.4067190885543823, "step": 22783 }, { "epoch": 1.7961371698856916, "grad_norm": 0.5233785510063171, "learning_rate": 0.0005, "loss": 1.2073898315429688, "step": 22784 }, { "epoch": 1.7962160031533307, "grad_norm": 0.6055616736412048, "learning_rate": 0.0005, "loss": 1.5195443630218506, "step": 22785 }, { "epoch": 1.7962948364209697, "grad_norm": 0.5592026114463806, "learning_rate": 0.0005, "loss": 1.4035427570343018, "step": 22786 }, { "epoch": 1.7963736696886086, "grad_norm": 0.5659366250038147, "learning_rate": 0.0005, "loss": 1.5356327295303345, "step": 22787 }, { "epoch": 1.7964525029562477, "grad_norm": 0.6059513092041016, "learning_rate": 0.0005, "loss": 1.7646915912628174, "step": 22788 }, { "epoch": 1.7965313362238864, "grad_norm": 1.74540376663208, "learning_rate": 0.0005, "loss": 1.318199872970581, "step": 22789 }, { "epoch": 1.7966101694915255, "grad_norm": 0.5981232523918152, "learning_rate": 0.0005, "loss": 1.338147521018982, "step": 22790 }, { "epoch": 1.7966890027591642, "grad_norm": 0.54915452003479, "learning_rate": 0.0005, "loss": 1.2726521492004395, "step": 22791 }, { "epoch": 1.7967678360268033, "grad_norm": 0.4463846683502197, "learning_rate": 0.0005, "loss": 1.467031478881836, "step": 22792 }, { "epoch": 1.7968466692944423, "grad_norm": 0.47241097688674927, "learning_rate": 0.0005, "loss": 1.349658489227295, "step": 22793 }, { "epoch": 1.7969255025620812, "grad_norm": 0.5553281903266907, "learning_rate": 0.0005, "loss": 1.5537378787994385, "step": 22794 }, { "epoch": 1.7970043358297203, "grad_norm": 0.4667033851146698, "learning_rate": 0.0005, "loss": 1.2680816650390625, "step": 22795 }, { "epoch": 1.797083169097359, "grad_norm": 0.4564851224422455, "learning_rate": 0.0005, "loss": 1.4427192211151123, "step": 22796 }, { "epoch": 1.7971620023649981, "grad_norm": 0.4861820340156555, "learning_rate": 0.0005, "loss": 1.3757785558700562, "step": 22797 }, { "epoch": 1.7972408356326368, "grad_norm": 0.45073994994163513, "learning_rate": 0.0005, "loss": 1.359359860420227, "step": 22798 }, { "epoch": 1.797319668900276, "grad_norm": 0.7536913752555847, "learning_rate": 0.0005, "loss": 1.5330305099487305, "step": 22799 }, { "epoch": 1.7973985021679149, "grad_norm": 0.5570499300956726, "learning_rate": 0.0005, "loss": 1.3493130207061768, "step": 22800 }, { "epoch": 1.7974773354355538, "grad_norm": 0.5557130575180054, "learning_rate": 0.0005, "loss": 1.2515615224838257, "step": 22801 }, { "epoch": 1.797556168703193, "grad_norm": 0.55040043592453, "learning_rate": 0.0005, "loss": 1.4563467502593994, "step": 22802 }, { "epoch": 1.7976350019708316, "grad_norm": 0.6182907223701477, "learning_rate": 0.0005, "loss": 1.5467970371246338, "step": 22803 }, { "epoch": 1.7977138352384707, "grad_norm": 0.5389537811279297, "learning_rate": 0.0005, "loss": 1.5966026782989502, "step": 22804 }, { "epoch": 1.7977926685061094, "grad_norm": 0.4756237268447876, "learning_rate": 0.0005, "loss": 1.293698787689209, "step": 22805 }, { "epoch": 1.7978715017737485, "grad_norm": 0.5456290245056152, "learning_rate": 0.0005, "loss": 1.4152777194976807, "step": 22806 }, { "epoch": 1.7979503350413875, "grad_norm": 0.49464571475982666, "learning_rate": 0.0005, "loss": 1.3589539527893066, "step": 22807 }, { "epoch": 1.7980291683090264, "grad_norm": 0.6444945335388184, "learning_rate": 0.0005, "loss": 1.404484748840332, "step": 22808 }, { "epoch": 1.7981080015766655, "grad_norm": 0.4309540390968323, "learning_rate": 0.0005, "loss": 1.445611834526062, "step": 22809 }, { "epoch": 1.7981868348443042, "grad_norm": 0.6357278227806091, "learning_rate": 0.0005, "loss": 1.489999771118164, "step": 22810 }, { "epoch": 1.7982656681119433, "grad_norm": 0.531817615032196, "learning_rate": 0.0005, "loss": 1.416061282157898, "step": 22811 }, { "epoch": 1.7983445013795822, "grad_norm": 0.5871997475624084, "learning_rate": 0.0005, "loss": 1.4688551425933838, "step": 22812 }, { "epoch": 1.7984233346472212, "grad_norm": 0.6243696212768555, "learning_rate": 0.0005, "loss": 1.6507611274719238, "step": 22813 }, { "epoch": 1.79850216791486, "grad_norm": 0.5541896820068359, "learning_rate": 0.0005, "loss": 1.6299670934677124, "step": 22814 }, { "epoch": 1.798581001182499, "grad_norm": 0.43268030881881714, "learning_rate": 0.0005, "loss": 1.3756461143493652, "step": 22815 }, { "epoch": 1.798659834450138, "grad_norm": 0.48182252049446106, "learning_rate": 0.0005, "loss": 1.4015015363693237, "step": 22816 }, { "epoch": 1.7987386677177768, "grad_norm": 0.5394688844680786, "learning_rate": 0.0005, "loss": 1.3377659320831299, "step": 22817 }, { "epoch": 1.798817500985416, "grad_norm": 0.5735259652137756, "learning_rate": 0.0005, "loss": 1.4187613725662231, "step": 22818 }, { "epoch": 1.7988963342530548, "grad_norm": 0.44359350204467773, "learning_rate": 0.0005, "loss": 1.2251842021942139, "step": 22819 }, { "epoch": 1.7989751675206938, "grad_norm": 0.436896413564682, "learning_rate": 0.0005, "loss": 1.3298232555389404, "step": 22820 }, { "epoch": 1.7990540007883327, "grad_norm": 0.5170955657958984, "learning_rate": 0.0005, "loss": 1.6612632274627686, "step": 22821 }, { "epoch": 1.7991328340559716, "grad_norm": 0.4549706280231476, "learning_rate": 0.0005, "loss": 1.3531997203826904, "step": 22822 }, { "epoch": 1.7992116673236107, "grad_norm": 0.7589332461357117, "learning_rate": 0.0005, "loss": 1.5578277111053467, "step": 22823 }, { "epoch": 1.7992905005912494, "grad_norm": 0.5821592211723328, "learning_rate": 0.0005, "loss": 1.5023221969604492, "step": 22824 }, { "epoch": 1.7993693338588885, "grad_norm": 0.648128867149353, "learning_rate": 0.0005, "loss": 1.5774705410003662, "step": 22825 }, { "epoch": 1.7994481671265274, "grad_norm": 0.5594761371612549, "learning_rate": 0.0005, "loss": 1.5638515949249268, "step": 22826 }, { "epoch": 1.7995270003941664, "grad_norm": 0.4311095178127289, "learning_rate": 0.0005, "loss": 1.195074439048767, "step": 22827 }, { "epoch": 1.7996058336618053, "grad_norm": 0.4400213956832886, "learning_rate": 0.0005, "loss": 1.416641354560852, "step": 22828 }, { "epoch": 1.7996846669294442, "grad_norm": 0.4379582703113556, "learning_rate": 0.0005, "loss": 1.2305176258087158, "step": 22829 }, { "epoch": 1.7997635001970833, "grad_norm": 0.6052280068397522, "learning_rate": 0.0005, "loss": 1.449979543685913, "step": 22830 }, { "epoch": 1.799842333464722, "grad_norm": 0.5444471836090088, "learning_rate": 0.0005, "loss": 1.7057673931121826, "step": 22831 }, { "epoch": 1.7999211667323611, "grad_norm": 0.6101317405700684, "learning_rate": 0.0005, "loss": 1.4813635349273682, "step": 22832 }, { "epoch": 1.8, "grad_norm": 0.5476999878883362, "learning_rate": 0.0005, "loss": 1.4904627799987793, "step": 22833 }, { "epoch": 1.800078833267639, "grad_norm": 0.6079757809638977, "learning_rate": 0.0005, "loss": 1.3492366075515747, "step": 22834 }, { "epoch": 1.8001576665352779, "grad_norm": 0.565659761428833, "learning_rate": 0.0005, "loss": 1.4392321109771729, "step": 22835 }, { "epoch": 1.8002364998029168, "grad_norm": 0.5661056637763977, "learning_rate": 0.0005, "loss": 1.622026801109314, "step": 22836 }, { "epoch": 1.800315333070556, "grad_norm": 0.5525078773498535, "learning_rate": 0.0005, "loss": 1.540705680847168, "step": 22837 }, { "epoch": 1.8003941663381946, "grad_norm": 0.6046969890594482, "learning_rate": 0.0005, "loss": 1.423137903213501, "step": 22838 }, { "epoch": 1.8004729996058337, "grad_norm": 0.8545042276382446, "learning_rate": 0.0005, "loss": 1.3695182800292969, "step": 22839 }, { "epoch": 1.8005518328734726, "grad_norm": 0.6306965351104736, "learning_rate": 0.0005, "loss": 1.5466111898422241, "step": 22840 }, { "epoch": 1.8006306661411116, "grad_norm": 0.5148847699165344, "learning_rate": 0.0005, "loss": 1.4053940773010254, "step": 22841 }, { "epoch": 1.8007094994087505, "grad_norm": 0.5339689254760742, "learning_rate": 0.0005, "loss": 1.3958778381347656, "step": 22842 }, { "epoch": 1.8007883326763894, "grad_norm": 0.6664681434631348, "learning_rate": 0.0005, "loss": 1.5153148174285889, "step": 22843 }, { "epoch": 1.8008671659440285, "grad_norm": 0.5231455564498901, "learning_rate": 0.0005, "loss": 1.3842744827270508, "step": 22844 }, { "epoch": 1.8009459992116672, "grad_norm": 0.6519200801849365, "learning_rate": 0.0005, "loss": 1.4552594423294067, "step": 22845 }, { "epoch": 1.8010248324793063, "grad_norm": 0.4990823268890381, "learning_rate": 0.0005, "loss": 1.4158198833465576, "step": 22846 }, { "epoch": 1.8011036657469452, "grad_norm": 0.46609511971473694, "learning_rate": 0.0005, "loss": 1.3057222366333008, "step": 22847 }, { "epoch": 1.8011824990145842, "grad_norm": 0.4903106987476349, "learning_rate": 0.0005, "loss": 1.2806310653686523, "step": 22848 }, { "epoch": 1.801261332282223, "grad_norm": 0.44850146770477295, "learning_rate": 0.0005, "loss": 1.2918919324874878, "step": 22849 }, { "epoch": 1.801340165549862, "grad_norm": 0.5508014559745789, "learning_rate": 0.0005, "loss": 1.5351898670196533, "step": 22850 }, { "epoch": 1.8014189988175011, "grad_norm": 0.7726701498031616, "learning_rate": 0.0005, "loss": 1.6425859928131104, "step": 22851 }, { "epoch": 1.8014978320851398, "grad_norm": 0.4856741428375244, "learning_rate": 0.0005, "loss": 1.2913942337036133, "step": 22852 }, { "epoch": 1.801576665352779, "grad_norm": 0.5885029435157776, "learning_rate": 0.0005, "loss": 1.4272868633270264, "step": 22853 }, { "epoch": 1.8016554986204179, "grad_norm": 0.546312153339386, "learning_rate": 0.0005, "loss": 1.3724603652954102, "step": 22854 }, { "epoch": 1.8017343318880568, "grad_norm": 0.7676123380661011, "learning_rate": 0.0005, "loss": 1.3270777463912964, "step": 22855 }, { "epoch": 1.8018131651556957, "grad_norm": 0.558930516242981, "learning_rate": 0.0005, "loss": 1.39509916305542, "step": 22856 }, { "epoch": 1.8018919984233346, "grad_norm": 0.483004093170166, "learning_rate": 0.0005, "loss": 1.3093500137329102, "step": 22857 }, { "epoch": 1.8019708316909737, "grad_norm": 0.4913797676563263, "learning_rate": 0.0005, "loss": 1.5429861545562744, "step": 22858 }, { "epoch": 1.8020496649586124, "grad_norm": 0.5336236953735352, "learning_rate": 0.0005, "loss": 1.438943862915039, "step": 22859 }, { "epoch": 1.8021284982262515, "grad_norm": 0.5437742471694946, "learning_rate": 0.0005, "loss": 1.28341805934906, "step": 22860 }, { "epoch": 1.8022073314938905, "grad_norm": 0.48932456970214844, "learning_rate": 0.0005, "loss": 1.2689120769500732, "step": 22861 }, { "epoch": 1.8022861647615294, "grad_norm": 0.4584873914718628, "learning_rate": 0.0005, "loss": 1.447824478149414, "step": 22862 }, { "epoch": 1.8023649980291683, "grad_norm": 0.44538164138793945, "learning_rate": 0.0005, "loss": 1.427325963973999, "step": 22863 }, { "epoch": 1.8024438312968072, "grad_norm": 0.4951231777667999, "learning_rate": 0.0005, "loss": 1.4221184253692627, "step": 22864 }, { "epoch": 1.8025226645644463, "grad_norm": 0.5620147585868835, "learning_rate": 0.0005, "loss": 1.3807735443115234, "step": 22865 }, { "epoch": 1.802601497832085, "grad_norm": 0.5236591100692749, "learning_rate": 0.0005, "loss": 1.4980969429016113, "step": 22866 }, { "epoch": 1.8026803310997241, "grad_norm": 0.670869767665863, "learning_rate": 0.0005, "loss": 1.5481319427490234, "step": 22867 }, { "epoch": 1.802759164367363, "grad_norm": 0.5859401226043701, "learning_rate": 0.0005, "loss": 1.2182496786117554, "step": 22868 }, { "epoch": 1.802837997635002, "grad_norm": 0.49679064750671387, "learning_rate": 0.0005, "loss": 1.3124542236328125, "step": 22869 }, { "epoch": 1.8029168309026409, "grad_norm": 0.48766881227493286, "learning_rate": 0.0005, "loss": 1.2178616523742676, "step": 22870 }, { "epoch": 1.8029956641702798, "grad_norm": 0.5229663252830505, "learning_rate": 0.0005, "loss": 1.4062583446502686, "step": 22871 }, { "epoch": 1.803074497437919, "grad_norm": 1.3023478984832764, "learning_rate": 0.0005, "loss": 1.578294277191162, "step": 22872 }, { "epoch": 1.8031533307055576, "grad_norm": 0.4567146301269531, "learning_rate": 0.0005, "loss": 1.4307198524475098, "step": 22873 }, { "epoch": 1.8032321639731967, "grad_norm": 0.5034380555152893, "learning_rate": 0.0005, "loss": 1.3278207778930664, "step": 22874 }, { "epoch": 1.8033109972408357, "grad_norm": 0.602480947971344, "learning_rate": 0.0005, "loss": 1.5897669792175293, "step": 22875 }, { "epoch": 1.8033898305084746, "grad_norm": 0.8408892750740051, "learning_rate": 0.0005, "loss": 1.4931182861328125, "step": 22876 }, { "epoch": 1.8034686637761135, "grad_norm": 0.5922077298164368, "learning_rate": 0.0005, "loss": 1.4438467025756836, "step": 22877 }, { "epoch": 1.8035474970437524, "grad_norm": 0.4708620607852936, "learning_rate": 0.0005, "loss": 1.3344058990478516, "step": 22878 }, { "epoch": 1.8036263303113915, "grad_norm": 0.567344605922699, "learning_rate": 0.0005, "loss": 1.269315242767334, "step": 22879 }, { "epoch": 1.8037051635790302, "grad_norm": 0.4948488473892212, "learning_rate": 0.0005, "loss": 1.265803337097168, "step": 22880 }, { "epoch": 1.8037839968466693, "grad_norm": 0.4308474659919739, "learning_rate": 0.0005, "loss": 1.4484660625457764, "step": 22881 }, { "epoch": 1.8038628301143083, "grad_norm": 0.4725874364376068, "learning_rate": 0.0005, "loss": 1.4786697626113892, "step": 22882 }, { "epoch": 1.8039416633819472, "grad_norm": 0.49110183119773865, "learning_rate": 0.0005, "loss": 1.1954069137573242, "step": 22883 }, { "epoch": 1.804020496649586, "grad_norm": 0.48072147369384766, "learning_rate": 0.0005, "loss": 1.5414788722991943, "step": 22884 }, { "epoch": 1.804099329917225, "grad_norm": 0.9860129952430725, "learning_rate": 0.0005, "loss": 1.4345958232879639, "step": 22885 }, { "epoch": 1.8041781631848641, "grad_norm": 0.6401737332344055, "learning_rate": 0.0005, "loss": 1.4323883056640625, "step": 22886 }, { "epoch": 1.8042569964525028, "grad_norm": 0.48692458868026733, "learning_rate": 0.0005, "loss": 1.4114983081817627, "step": 22887 }, { "epoch": 1.804335829720142, "grad_norm": 0.4905116558074951, "learning_rate": 0.0005, "loss": 1.2873884439468384, "step": 22888 }, { "epoch": 1.8044146629877809, "grad_norm": 0.7094005942344666, "learning_rate": 0.0005, "loss": 1.5196458101272583, "step": 22889 }, { "epoch": 1.8044934962554198, "grad_norm": 0.541853129863739, "learning_rate": 0.0005, "loss": 1.3212306499481201, "step": 22890 }, { "epoch": 1.8045723295230587, "grad_norm": 0.5328664183616638, "learning_rate": 0.0005, "loss": 1.352843165397644, "step": 22891 }, { "epoch": 1.8046511627906976, "grad_norm": 0.5682697892189026, "learning_rate": 0.0005, "loss": 1.4918463230133057, "step": 22892 }, { "epoch": 1.8047299960583367, "grad_norm": 0.5322390198707581, "learning_rate": 0.0005, "loss": 1.4469926357269287, "step": 22893 }, { "epoch": 1.8048088293259754, "grad_norm": 0.573460578918457, "learning_rate": 0.0005, "loss": 1.654482126235962, "step": 22894 }, { "epoch": 1.8048876625936146, "grad_norm": 0.5748060345649719, "learning_rate": 0.0005, "loss": 1.3427214622497559, "step": 22895 }, { "epoch": 1.8049664958612535, "grad_norm": 0.4858880043029785, "learning_rate": 0.0005, "loss": 1.45064377784729, "step": 22896 }, { "epoch": 1.8050453291288924, "grad_norm": 0.5237690806388855, "learning_rate": 0.0005, "loss": 1.557331919670105, "step": 22897 }, { "epoch": 1.8051241623965313, "grad_norm": 0.560499906539917, "learning_rate": 0.0005, "loss": 1.5188663005828857, "step": 22898 }, { "epoch": 1.8052029956641702, "grad_norm": 0.5052613615989685, "learning_rate": 0.0005, "loss": 1.350538730621338, "step": 22899 }, { "epoch": 1.8052818289318093, "grad_norm": 0.4511988162994385, "learning_rate": 0.0005, "loss": 1.239283561706543, "step": 22900 }, { "epoch": 1.805360662199448, "grad_norm": 0.6202725172042847, "learning_rate": 0.0005, "loss": 1.3154178857803345, "step": 22901 }, { "epoch": 1.8054394954670872, "grad_norm": 0.5119534134864807, "learning_rate": 0.0005, "loss": 1.5002803802490234, "step": 22902 }, { "epoch": 1.805518328734726, "grad_norm": 0.7250404357910156, "learning_rate": 0.0005, "loss": 1.4468505382537842, "step": 22903 }, { "epoch": 1.805597162002365, "grad_norm": 0.49471578001976013, "learning_rate": 0.0005, "loss": 1.3224705457687378, "step": 22904 }, { "epoch": 1.8056759952700039, "grad_norm": 0.5134555101394653, "learning_rate": 0.0005, "loss": 1.3144619464874268, "step": 22905 }, { "epoch": 1.8057548285376428, "grad_norm": 0.5205124020576477, "learning_rate": 0.0005, "loss": 1.4005796909332275, "step": 22906 }, { "epoch": 1.805833661805282, "grad_norm": 0.6091569066047668, "learning_rate": 0.0005, "loss": 1.3441336154937744, "step": 22907 }, { "epoch": 1.8059124950729206, "grad_norm": 0.4966760277748108, "learning_rate": 0.0005, "loss": 1.577440857887268, "step": 22908 }, { "epoch": 1.8059913283405598, "grad_norm": 0.49749109148979187, "learning_rate": 0.0005, "loss": 1.546342372894287, "step": 22909 }, { "epoch": 1.8060701616081987, "grad_norm": 0.5769877433776855, "learning_rate": 0.0005, "loss": 1.3549318313598633, "step": 22910 }, { "epoch": 1.8061489948758376, "grad_norm": 0.542664647102356, "learning_rate": 0.0005, "loss": 1.418137788772583, "step": 22911 }, { "epoch": 1.8062278281434767, "grad_norm": 0.6389974355697632, "learning_rate": 0.0005, "loss": 1.4154841899871826, "step": 22912 }, { "epoch": 1.8063066614111154, "grad_norm": 0.6126923561096191, "learning_rate": 0.0005, "loss": 1.4435834884643555, "step": 22913 }, { "epoch": 1.8063854946787545, "grad_norm": 0.5790419578552246, "learning_rate": 0.0005, "loss": 1.4594407081604004, "step": 22914 }, { "epoch": 1.8064643279463932, "grad_norm": 0.5660926103591919, "learning_rate": 0.0005, "loss": 1.2849663496017456, "step": 22915 }, { "epoch": 1.8065431612140324, "grad_norm": 0.5252892971038818, "learning_rate": 0.0005, "loss": 1.2970435619354248, "step": 22916 }, { "epoch": 1.8066219944816713, "grad_norm": 0.5590305328369141, "learning_rate": 0.0005, "loss": 1.3046679496765137, "step": 22917 }, { "epoch": 1.8067008277493102, "grad_norm": 0.579853355884552, "learning_rate": 0.0005, "loss": 1.47303307056427, "step": 22918 }, { "epoch": 1.8067796610169493, "grad_norm": 0.6891006827354431, "learning_rate": 0.0005, "loss": 1.3474063873291016, "step": 22919 }, { "epoch": 1.806858494284588, "grad_norm": 0.5593375563621521, "learning_rate": 0.0005, "loss": 1.5438780784606934, "step": 22920 }, { "epoch": 1.8069373275522271, "grad_norm": 0.6094783544540405, "learning_rate": 0.0005, "loss": 1.6266868114471436, "step": 22921 }, { "epoch": 1.8070161608198658, "grad_norm": 0.5023624300956726, "learning_rate": 0.0005, "loss": 1.5245497226715088, "step": 22922 }, { "epoch": 1.807094994087505, "grad_norm": 0.47927409410476685, "learning_rate": 0.0005, "loss": 1.6947557926177979, "step": 22923 }, { "epoch": 1.8071738273551439, "grad_norm": 0.5424547791481018, "learning_rate": 0.0005, "loss": 1.282105565071106, "step": 22924 }, { "epoch": 1.8072526606227828, "grad_norm": 0.4644622504711151, "learning_rate": 0.0005, "loss": 1.4629199504852295, "step": 22925 }, { "epoch": 1.807331493890422, "grad_norm": 0.6744329333305359, "learning_rate": 0.0005, "loss": 1.3723745346069336, "step": 22926 }, { "epoch": 1.8074103271580606, "grad_norm": 0.5107508301734924, "learning_rate": 0.0005, "loss": 1.428844928741455, "step": 22927 }, { "epoch": 1.8074891604256997, "grad_norm": 0.4793033301830292, "learning_rate": 0.0005, "loss": 1.3874590396881104, "step": 22928 }, { "epoch": 1.8075679936933384, "grad_norm": 0.5201350450515747, "learning_rate": 0.0005, "loss": 1.3534421920776367, "step": 22929 }, { "epoch": 1.8076468269609776, "grad_norm": 0.6377310156822205, "learning_rate": 0.0005, "loss": 1.5222820043563843, "step": 22930 }, { "epoch": 1.8077256602286165, "grad_norm": 0.5113776922225952, "learning_rate": 0.0005, "loss": 1.3359391689300537, "step": 22931 }, { "epoch": 1.8078044934962554, "grad_norm": 0.8395070433616638, "learning_rate": 0.0005, "loss": 1.5857625007629395, "step": 22932 }, { "epoch": 1.8078833267638945, "grad_norm": 0.5490095019340515, "learning_rate": 0.0005, "loss": 1.3210068941116333, "step": 22933 }, { "epoch": 1.8079621600315332, "grad_norm": 0.512921929359436, "learning_rate": 0.0005, "loss": 1.34214186668396, "step": 22934 }, { "epoch": 1.8080409932991723, "grad_norm": 0.5160278677940369, "learning_rate": 0.0005, "loss": 1.3625397682189941, "step": 22935 }, { "epoch": 1.808119826566811, "grad_norm": 0.5298879742622375, "learning_rate": 0.0005, "loss": 1.3787972927093506, "step": 22936 }, { "epoch": 1.8081986598344502, "grad_norm": 0.47915706038475037, "learning_rate": 0.0005, "loss": 1.2173174619674683, "step": 22937 }, { "epoch": 1.808277493102089, "grad_norm": 0.6011403799057007, "learning_rate": 0.0005, "loss": 1.529166340827942, "step": 22938 }, { "epoch": 1.808356326369728, "grad_norm": 1.1140927076339722, "learning_rate": 0.0005, "loss": 1.18330979347229, "step": 22939 }, { "epoch": 1.8084351596373671, "grad_norm": 0.523809015750885, "learning_rate": 0.0005, "loss": 1.441561222076416, "step": 22940 }, { "epoch": 1.8085139929050058, "grad_norm": 0.5033941864967346, "learning_rate": 0.0005, "loss": 1.505114197731018, "step": 22941 }, { "epoch": 1.808592826172645, "grad_norm": 0.5161979794502258, "learning_rate": 0.0005, "loss": 1.67655611038208, "step": 22942 }, { "epoch": 1.8086716594402839, "grad_norm": 0.6753272414207458, "learning_rate": 0.0005, "loss": 1.6847916841506958, "step": 22943 }, { "epoch": 1.8087504927079228, "grad_norm": 0.5099573731422424, "learning_rate": 0.0005, "loss": 1.5760512351989746, "step": 22944 }, { "epoch": 1.8088293259755617, "grad_norm": 0.5772497057914734, "learning_rate": 0.0005, "loss": 1.39942467212677, "step": 22945 }, { "epoch": 1.8089081592432006, "grad_norm": 0.5908588171005249, "learning_rate": 0.0005, "loss": 1.435213565826416, "step": 22946 }, { "epoch": 1.8089869925108397, "grad_norm": 0.5497503280639648, "learning_rate": 0.0005, "loss": 1.5377388000488281, "step": 22947 }, { "epoch": 1.8090658257784784, "grad_norm": 0.47341182827949524, "learning_rate": 0.0005, "loss": 1.2106249332427979, "step": 22948 }, { "epoch": 1.8091446590461175, "grad_norm": 0.7401187419891357, "learning_rate": 0.0005, "loss": 1.6026611328125, "step": 22949 }, { "epoch": 1.8092234923137565, "grad_norm": 0.7000168561935425, "learning_rate": 0.0005, "loss": 1.7021281719207764, "step": 22950 }, { "epoch": 1.8093023255813954, "grad_norm": 0.4220260679721832, "learning_rate": 0.0005, "loss": 1.3450853824615479, "step": 22951 }, { "epoch": 1.8093811588490343, "grad_norm": 0.4310682713985443, "learning_rate": 0.0005, "loss": 1.469390630722046, "step": 22952 }, { "epoch": 1.8094599921166732, "grad_norm": 0.5731669068336487, "learning_rate": 0.0005, "loss": 1.2487446069717407, "step": 22953 }, { "epoch": 1.8095388253843123, "grad_norm": 0.607976496219635, "learning_rate": 0.0005, "loss": 1.1793900728225708, "step": 22954 }, { "epoch": 1.809617658651951, "grad_norm": 0.5821358561515808, "learning_rate": 0.0005, "loss": 1.2794573307037354, "step": 22955 }, { "epoch": 1.8096964919195901, "grad_norm": 0.7131853103637695, "learning_rate": 0.0005, "loss": 1.4127486944198608, "step": 22956 }, { "epoch": 1.809775325187229, "grad_norm": 0.5323979258537292, "learning_rate": 0.0005, "loss": 1.431509256362915, "step": 22957 }, { "epoch": 1.809854158454868, "grad_norm": 0.6768002510070801, "learning_rate": 0.0005, "loss": 1.6914539337158203, "step": 22958 }, { "epoch": 1.8099329917225069, "grad_norm": 0.6089015007019043, "learning_rate": 0.0005, "loss": 1.4916743040084839, "step": 22959 }, { "epoch": 1.8100118249901458, "grad_norm": 0.5957471132278442, "learning_rate": 0.0005, "loss": 1.4683539867401123, "step": 22960 }, { "epoch": 1.810090658257785, "grad_norm": 0.4336991012096405, "learning_rate": 0.0005, "loss": 1.1915596723556519, "step": 22961 }, { "epoch": 1.8101694915254236, "grad_norm": 0.4972025156021118, "learning_rate": 0.0005, "loss": 1.4096052646636963, "step": 22962 }, { "epoch": 1.8102483247930627, "grad_norm": 0.5374481678009033, "learning_rate": 0.0005, "loss": 1.3445179462432861, "step": 22963 }, { "epoch": 1.8103271580607017, "grad_norm": 0.5823454260826111, "learning_rate": 0.0005, "loss": 1.6055569648742676, "step": 22964 }, { "epoch": 1.8104059913283406, "grad_norm": 0.44720259308815, "learning_rate": 0.0005, "loss": 1.3387730121612549, "step": 22965 }, { "epoch": 1.8104848245959795, "grad_norm": 0.4762636423110962, "learning_rate": 0.0005, "loss": 1.2851359844207764, "step": 22966 }, { "epoch": 1.8105636578636184, "grad_norm": 0.5342941880226135, "learning_rate": 0.0005, "loss": 1.510939121246338, "step": 22967 }, { "epoch": 1.8106424911312575, "grad_norm": 0.48578596115112305, "learning_rate": 0.0005, "loss": 1.384242296218872, "step": 22968 }, { "epoch": 1.8107213243988962, "grad_norm": 0.6064190864562988, "learning_rate": 0.0005, "loss": 1.4096877574920654, "step": 22969 }, { "epoch": 1.8108001576665353, "grad_norm": 0.6229479312896729, "learning_rate": 0.0005, "loss": 1.612969160079956, "step": 22970 }, { "epoch": 1.8108789909341743, "grad_norm": 0.5613697171211243, "learning_rate": 0.0005, "loss": 1.4462875127792358, "step": 22971 }, { "epoch": 1.8109578242018132, "grad_norm": 0.5126926898956299, "learning_rate": 0.0005, "loss": 1.2765066623687744, "step": 22972 }, { "epoch": 1.811036657469452, "grad_norm": 0.717696487903595, "learning_rate": 0.0005, "loss": 1.7031948566436768, "step": 22973 }, { "epoch": 1.811115490737091, "grad_norm": 0.522471010684967, "learning_rate": 0.0005, "loss": 1.3773794174194336, "step": 22974 }, { "epoch": 1.8111943240047301, "grad_norm": 0.5237188935279846, "learning_rate": 0.0005, "loss": 1.4576048851013184, "step": 22975 }, { "epoch": 1.8112731572723688, "grad_norm": 0.5024278163909912, "learning_rate": 0.0005, "loss": 1.1709771156311035, "step": 22976 }, { "epoch": 1.811351990540008, "grad_norm": 0.5062069892883301, "learning_rate": 0.0005, "loss": 1.3995981216430664, "step": 22977 }, { "epoch": 1.8114308238076469, "grad_norm": 0.6508666276931763, "learning_rate": 0.0005, "loss": 1.4287822246551514, "step": 22978 }, { "epoch": 1.8115096570752858, "grad_norm": 0.6436783671379089, "learning_rate": 0.0005, "loss": 1.2736356258392334, "step": 22979 }, { "epoch": 1.8115884903429247, "grad_norm": 0.5130868554115295, "learning_rate": 0.0005, "loss": 1.4226644039154053, "step": 22980 }, { "epoch": 1.8116673236105636, "grad_norm": 0.5135653018951416, "learning_rate": 0.0005, "loss": 1.159797191619873, "step": 22981 }, { "epoch": 1.8117461568782027, "grad_norm": 0.5350592732429504, "learning_rate": 0.0005, "loss": 1.4080235958099365, "step": 22982 }, { "epoch": 1.8118249901458414, "grad_norm": 0.5197455883026123, "learning_rate": 0.0005, "loss": 1.3688668012619019, "step": 22983 }, { "epoch": 1.8119038234134806, "grad_norm": 0.47323960065841675, "learning_rate": 0.0005, "loss": 1.292768120765686, "step": 22984 }, { "epoch": 1.8119826566811195, "grad_norm": 0.7079096436500549, "learning_rate": 0.0005, "loss": 1.3652153015136719, "step": 22985 }, { "epoch": 1.8120614899487584, "grad_norm": 0.4717675745487213, "learning_rate": 0.0005, "loss": 1.2547097206115723, "step": 22986 }, { "epoch": 1.8121403232163973, "grad_norm": 0.5759062170982361, "learning_rate": 0.0005, "loss": 1.4585392475128174, "step": 22987 }, { "epoch": 1.8122191564840362, "grad_norm": 0.5663658976554871, "learning_rate": 0.0005, "loss": 1.5355582237243652, "step": 22988 }, { "epoch": 1.8122979897516753, "grad_norm": 0.5088832378387451, "learning_rate": 0.0005, "loss": 1.523322343826294, "step": 22989 }, { "epoch": 1.812376823019314, "grad_norm": 0.5722275972366333, "learning_rate": 0.0005, "loss": 1.3909302949905396, "step": 22990 }, { "epoch": 1.8124556562869532, "grad_norm": 0.5535815358161926, "learning_rate": 0.0005, "loss": 1.4758896827697754, "step": 22991 }, { "epoch": 1.812534489554592, "grad_norm": 0.6297440528869629, "learning_rate": 0.0005, "loss": 1.4480613470077515, "step": 22992 }, { "epoch": 1.812613322822231, "grad_norm": 0.6506247520446777, "learning_rate": 0.0005, "loss": 1.4502346515655518, "step": 22993 }, { "epoch": 1.8126921560898699, "grad_norm": 0.5373637080192566, "learning_rate": 0.0005, "loss": 1.54630446434021, "step": 22994 }, { "epoch": 1.8127709893575088, "grad_norm": 0.4566446542739868, "learning_rate": 0.0005, "loss": 1.1683216094970703, "step": 22995 }, { "epoch": 1.812849822625148, "grad_norm": 0.5667019486427307, "learning_rate": 0.0005, "loss": 1.6170769929885864, "step": 22996 }, { "epoch": 1.8129286558927866, "grad_norm": 0.5225763916969299, "learning_rate": 0.0005, "loss": 1.4820704460144043, "step": 22997 }, { "epoch": 1.8130074891604258, "grad_norm": 0.6210606694221497, "learning_rate": 0.0005, "loss": 1.213170051574707, "step": 22998 }, { "epoch": 1.8130863224280647, "grad_norm": 0.5913444757461548, "learning_rate": 0.0005, "loss": 1.4864370822906494, "step": 22999 }, { "epoch": 1.8131651556957036, "grad_norm": 0.47681474685668945, "learning_rate": 0.0005, "loss": 1.417222261428833, "step": 23000 }, { "epoch": 1.8132439889633425, "grad_norm": 0.578087329864502, "learning_rate": 0.0005, "loss": 1.4981279373168945, "step": 23001 }, { "epoch": 1.8133228222309814, "grad_norm": 0.6344078779220581, "learning_rate": 0.0005, "loss": 1.675628662109375, "step": 23002 }, { "epoch": 1.8134016554986205, "grad_norm": 0.6614150404930115, "learning_rate": 0.0005, "loss": 1.600584864616394, "step": 23003 }, { "epoch": 1.8134804887662592, "grad_norm": 0.4781332314014435, "learning_rate": 0.0005, "loss": 1.368727445602417, "step": 23004 }, { "epoch": 1.8135593220338984, "grad_norm": 0.4661031663417816, "learning_rate": 0.0005, "loss": 1.3702502250671387, "step": 23005 }, { "epoch": 1.8136381553015373, "grad_norm": 0.54755038022995, "learning_rate": 0.0005, "loss": 1.364823818206787, "step": 23006 }, { "epoch": 1.8137169885691762, "grad_norm": 0.5384311079978943, "learning_rate": 0.0005, "loss": 1.407853603363037, "step": 23007 }, { "epoch": 1.813795821836815, "grad_norm": 0.5072499513626099, "learning_rate": 0.0005, "loss": 1.4255123138427734, "step": 23008 }, { "epoch": 1.813874655104454, "grad_norm": 0.6565051078796387, "learning_rate": 0.0005, "loss": 1.4379329681396484, "step": 23009 }, { "epoch": 1.8139534883720931, "grad_norm": 0.5290448665618896, "learning_rate": 0.0005, "loss": 1.3386070728302002, "step": 23010 }, { "epoch": 1.8140323216397318, "grad_norm": 0.563398003578186, "learning_rate": 0.0005, "loss": 1.735917329788208, "step": 23011 }, { "epoch": 1.814111154907371, "grad_norm": 0.8016701340675354, "learning_rate": 0.0005, "loss": 1.2829571962356567, "step": 23012 }, { "epoch": 1.8141899881750099, "grad_norm": 0.5259914994239807, "learning_rate": 0.0005, "loss": 1.687705636024475, "step": 23013 }, { "epoch": 1.8142688214426488, "grad_norm": 0.47698265314102173, "learning_rate": 0.0005, "loss": 1.3892581462860107, "step": 23014 }, { "epoch": 1.8143476547102877, "grad_norm": 0.5459185838699341, "learning_rate": 0.0005, "loss": 1.4024474620819092, "step": 23015 }, { "epoch": 1.8144264879779266, "grad_norm": 0.6727336645126343, "learning_rate": 0.0005, "loss": 1.6862707138061523, "step": 23016 }, { "epoch": 1.8145053212455657, "grad_norm": 0.863376259803772, "learning_rate": 0.0005, "loss": 1.5644927024841309, "step": 23017 }, { "epoch": 1.8145841545132044, "grad_norm": 0.5629061460494995, "learning_rate": 0.0005, "loss": 1.4200894832611084, "step": 23018 }, { "epoch": 1.8146629877808436, "grad_norm": 0.6459577083587646, "learning_rate": 0.0005, "loss": 1.4773807525634766, "step": 23019 }, { "epoch": 1.8147418210484825, "grad_norm": 0.6738964319229126, "learning_rate": 0.0005, "loss": 1.4004931449890137, "step": 23020 }, { "epoch": 1.8148206543161214, "grad_norm": 0.4535559117794037, "learning_rate": 0.0005, "loss": 1.3498687744140625, "step": 23021 }, { "epoch": 1.8148994875837603, "grad_norm": 0.5596552491188049, "learning_rate": 0.0005, "loss": 1.4048666954040527, "step": 23022 }, { "epoch": 1.8149783208513992, "grad_norm": 0.6333991885185242, "learning_rate": 0.0005, "loss": 1.4553685188293457, "step": 23023 }, { "epoch": 1.8150571541190383, "grad_norm": 0.4775437116622925, "learning_rate": 0.0005, "loss": 1.5172868967056274, "step": 23024 }, { "epoch": 1.815135987386677, "grad_norm": 0.48333555459976196, "learning_rate": 0.0005, "loss": 1.324965000152588, "step": 23025 }, { "epoch": 1.8152148206543162, "grad_norm": 0.5243455171585083, "learning_rate": 0.0005, "loss": 1.4187946319580078, "step": 23026 }, { "epoch": 1.815293653921955, "grad_norm": 0.4851725399494171, "learning_rate": 0.0005, "loss": 1.3455479145050049, "step": 23027 }, { "epoch": 1.815372487189594, "grad_norm": 0.5943806171417236, "learning_rate": 0.0005, "loss": 1.5266444683074951, "step": 23028 }, { "epoch": 1.815451320457233, "grad_norm": 0.5273694396018982, "learning_rate": 0.0005, "loss": 1.5675544738769531, "step": 23029 }, { "epoch": 1.8155301537248718, "grad_norm": 0.48655441403388977, "learning_rate": 0.0005, "loss": 1.416957139968872, "step": 23030 }, { "epoch": 1.815608986992511, "grad_norm": 0.6178381443023682, "learning_rate": 0.0005, "loss": 1.345216155052185, "step": 23031 }, { "epoch": 1.8156878202601496, "grad_norm": 0.5402007699012756, "learning_rate": 0.0005, "loss": 1.6363472938537598, "step": 23032 }, { "epoch": 1.8157666535277888, "grad_norm": 0.4897860884666443, "learning_rate": 0.0005, "loss": 1.367060661315918, "step": 23033 }, { "epoch": 1.8158454867954277, "grad_norm": 0.5423726439476013, "learning_rate": 0.0005, "loss": 1.7150142192840576, "step": 23034 }, { "epoch": 1.8159243200630666, "grad_norm": 0.5312161445617676, "learning_rate": 0.0005, "loss": 1.28290855884552, "step": 23035 }, { "epoch": 1.8160031533307057, "grad_norm": 0.6192254424095154, "learning_rate": 0.0005, "loss": 1.3780887126922607, "step": 23036 }, { "epoch": 1.8160819865983444, "grad_norm": 0.5970808267593384, "learning_rate": 0.0005, "loss": 1.252578616142273, "step": 23037 }, { "epoch": 1.8161608198659835, "grad_norm": 0.5482264757156372, "learning_rate": 0.0005, "loss": 1.3350521326065063, "step": 23038 }, { "epoch": 1.8162396531336222, "grad_norm": 0.5979709625244141, "learning_rate": 0.0005, "loss": 1.5712428092956543, "step": 23039 }, { "epoch": 1.8163184864012614, "grad_norm": 0.6664729714393616, "learning_rate": 0.0005, "loss": 1.5781214237213135, "step": 23040 }, { "epoch": 1.8163973196689003, "grad_norm": 0.5396178960800171, "learning_rate": 0.0005, "loss": 1.4999724626541138, "step": 23041 }, { "epoch": 1.8164761529365392, "grad_norm": 0.5844659209251404, "learning_rate": 0.0005, "loss": 1.5991578102111816, "step": 23042 }, { "epoch": 1.8165549862041783, "grad_norm": 1.802781581878662, "learning_rate": 0.0005, "loss": 1.4698889255523682, "step": 23043 }, { "epoch": 1.816633819471817, "grad_norm": 0.6239618062973022, "learning_rate": 0.0005, "loss": 1.5061211585998535, "step": 23044 }, { "epoch": 1.8167126527394561, "grad_norm": 0.5284643173217773, "learning_rate": 0.0005, "loss": 1.332298994064331, "step": 23045 }, { "epoch": 1.8167914860070948, "grad_norm": 0.5132949352264404, "learning_rate": 0.0005, "loss": 1.5136165618896484, "step": 23046 }, { "epoch": 1.816870319274734, "grad_norm": 0.5251705646514893, "learning_rate": 0.0005, "loss": 1.626673698425293, "step": 23047 }, { "epoch": 1.8169491525423729, "grad_norm": 0.5897322297096252, "learning_rate": 0.0005, "loss": 1.427452802658081, "step": 23048 }, { "epoch": 1.8170279858100118, "grad_norm": 0.5394028425216675, "learning_rate": 0.0005, "loss": 1.6420888900756836, "step": 23049 }, { "epoch": 1.817106819077651, "grad_norm": 0.6510637402534485, "learning_rate": 0.0005, "loss": 1.349069356918335, "step": 23050 }, { "epoch": 1.8171856523452896, "grad_norm": 0.44711047410964966, "learning_rate": 0.0005, "loss": 1.3920451402664185, "step": 23051 }, { "epoch": 1.8172644856129287, "grad_norm": 0.6315614581108093, "learning_rate": 0.0005, "loss": 1.3933675289154053, "step": 23052 }, { "epoch": 1.8173433188805674, "grad_norm": 0.5516133904457092, "learning_rate": 0.0005, "loss": 1.448230504989624, "step": 23053 }, { "epoch": 1.8174221521482066, "grad_norm": 0.5140467882156372, "learning_rate": 0.0005, "loss": 1.2950950860977173, "step": 23054 }, { "epoch": 1.8175009854158455, "grad_norm": 0.519484281539917, "learning_rate": 0.0005, "loss": 1.5470852851867676, "step": 23055 }, { "epoch": 1.8175798186834844, "grad_norm": 0.5193914771080017, "learning_rate": 0.0005, "loss": 1.3653907775878906, "step": 23056 }, { "epoch": 1.8176586519511235, "grad_norm": 0.47152188420295715, "learning_rate": 0.0005, "loss": 1.5021710395812988, "step": 23057 }, { "epoch": 1.8177374852187622, "grad_norm": 0.6768068671226501, "learning_rate": 0.0005, "loss": 1.217478632926941, "step": 23058 }, { "epoch": 1.8178163184864014, "grad_norm": 0.4875575304031372, "learning_rate": 0.0005, "loss": 1.3444428443908691, "step": 23059 }, { "epoch": 1.81789515175404, "grad_norm": 0.5544682741165161, "learning_rate": 0.0005, "loss": 1.4767318964004517, "step": 23060 }, { "epoch": 1.8179739850216792, "grad_norm": 0.4824865758419037, "learning_rate": 0.0005, "loss": 1.283609390258789, "step": 23061 }, { "epoch": 1.818052818289318, "grad_norm": 0.7025818228721619, "learning_rate": 0.0005, "loss": 1.679704189300537, "step": 23062 }, { "epoch": 1.818131651556957, "grad_norm": 0.5896900296211243, "learning_rate": 0.0005, "loss": 1.390567421913147, "step": 23063 }, { "epoch": 1.8182104848245961, "grad_norm": 0.4580833613872528, "learning_rate": 0.0005, "loss": 1.271805763244629, "step": 23064 }, { "epoch": 1.8182893180922348, "grad_norm": 0.5724038481712341, "learning_rate": 0.0005, "loss": 1.2629444599151611, "step": 23065 }, { "epoch": 1.818368151359874, "grad_norm": 0.4820258319377899, "learning_rate": 0.0005, "loss": 1.4387125968933105, "step": 23066 }, { "epoch": 1.8184469846275129, "grad_norm": 0.5069721341133118, "learning_rate": 0.0005, "loss": 1.4961748123168945, "step": 23067 }, { "epoch": 1.8185258178951518, "grad_norm": 0.5557899475097656, "learning_rate": 0.0005, "loss": 1.5765067338943481, "step": 23068 }, { "epoch": 1.8186046511627907, "grad_norm": 0.549435019493103, "learning_rate": 0.0005, "loss": 1.4605553150177002, "step": 23069 }, { "epoch": 1.8186834844304296, "grad_norm": 0.6162946224212646, "learning_rate": 0.0005, "loss": 1.3757789134979248, "step": 23070 }, { "epoch": 1.8187623176980687, "grad_norm": 0.5946944355964661, "learning_rate": 0.0005, "loss": 1.2184474468231201, "step": 23071 }, { "epoch": 1.8188411509657074, "grad_norm": 0.4591265916824341, "learning_rate": 0.0005, "loss": 1.314439058303833, "step": 23072 }, { "epoch": 1.8189199842333466, "grad_norm": 0.5143582820892334, "learning_rate": 0.0005, "loss": 1.4552032947540283, "step": 23073 }, { "epoch": 1.8189988175009855, "grad_norm": 0.5753969550132751, "learning_rate": 0.0005, "loss": 1.5278444290161133, "step": 23074 }, { "epoch": 1.8190776507686244, "grad_norm": 0.6284903883934021, "learning_rate": 0.0005, "loss": 1.6359639167785645, "step": 23075 }, { "epoch": 1.8191564840362633, "grad_norm": 0.45518842339515686, "learning_rate": 0.0005, "loss": 1.26322340965271, "step": 23076 }, { "epoch": 1.8192353173039022, "grad_norm": 0.5978152751922607, "learning_rate": 0.0005, "loss": 1.2219793796539307, "step": 23077 }, { "epoch": 1.8193141505715413, "grad_norm": 0.4587823152542114, "learning_rate": 0.0005, "loss": 1.3610626459121704, "step": 23078 }, { "epoch": 1.81939298383918, "grad_norm": 0.43696653842926025, "learning_rate": 0.0005, "loss": 1.509560227394104, "step": 23079 }, { "epoch": 1.8194718171068192, "grad_norm": 0.527725338935852, "learning_rate": 0.0005, "loss": 1.4535472393035889, "step": 23080 }, { "epoch": 1.819550650374458, "grad_norm": 0.5018261671066284, "learning_rate": 0.0005, "loss": 1.3874306678771973, "step": 23081 }, { "epoch": 1.819629483642097, "grad_norm": 0.48031264543533325, "learning_rate": 0.0005, "loss": 1.4240520000457764, "step": 23082 }, { "epoch": 1.819708316909736, "grad_norm": 0.4578114151954651, "learning_rate": 0.0005, "loss": 1.4884350299835205, "step": 23083 }, { "epoch": 1.8197871501773748, "grad_norm": 0.6114435791969299, "learning_rate": 0.0005, "loss": 1.4998584985733032, "step": 23084 }, { "epoch": 1.819865983445014, "grad_norm": 0.7282840609550476, "learning_rate": 0.0005, "loss": 1.5782595872879028, "step": 23085 }, { "epoch": 1.8199448167126526, "grad_norm": 0.4759194850921631, "learning_rate": 0.0005, "loss": 1.3538448810577393, "step": 23086 }, { "epoch": 1.8200236499802918, "grad_norm": 0.42216646671295166, "learning_rate": 0.0005, "loss": 1.249879240989685, "step": 23087 }, { "epoch": 1.8201024832479307, "grad_norm": 0.5512657761573792, "learning_rate": 0.0005, "loss": 1.1271281242370605, "step": 23088 }, { "epoch": 1.8201813165155696, "grad_norm": 0.5822150707244873, "learning_rate": 0.0005, "loss": 1.4523704051971436, "step": 23089 }, { "epoch": 1.8202601497832085, "grad_norm": 0.6582887172698975, "learning_rate": 0.0005, "loss": 1.3924753665924072, "step": 23090 }, { "epoch": 1.8203389830508474, "grad_norm": 0.675878643989563, "learning_rate": 0.0005, "loss": 1.6106257438659668, "step": 23091 }, { "epoch": 1.8204178163184865, "grad_norm": 0.6277300119400024, "learning_rate": 0.0005, "loss": 1.478168249130249, "step": 23092 }, { "epoch": 1.8204966495861252, "grad_norm": 0.5362387895584106, "learning_rate": 0.0005, "loss": 1.409184455871582, "step": 23093 }, { "epoch": 1.8205754828537644, "grad_norm": 0.4744289815425873, "learning_rate": 0.0005, "loss": 1.328214406967163, "step": 23094 }, { "epoch": 1.8206543161214033, "grad_norm": 0.9198684096336365, "learning_rate": 0.0005, "loss": 1.5943306684494019, "step": 23095 }, { "epoch": 1.8207331493890422, "grad_norm": 0.5615416169166565, "learning_rate": 0.0005, "loss": 1.5718252658843994, "step": 23096 }, { "epoch": 1.820811982656681, "grad_norm": 0.430774450302124, "learning_rate": 0.0005, "loss": 1.286468267440796, "step": 23097 }, { "epoch": 1.82089081592432, "grad_norm": 0.5616686940193176, "learning_rate": 0.0005, "loss": 1.429549217224121, "step": 23098 }, { "epoch": 1.8209696491919591, "grad_norm": 0.8871796131134033, "learning_rate": 0.0005, "loss": 1.5608965158462524, "step": 23099 }, { "epoch": 1.8210484824595978, "grad_norm": 0.4805773198604584, "learning_rate": 0.0005, "loss": 1.2769831418991089, "step": 23100 }, { "epoch": 1.821127315727237, "grad_norm": 0.6348820924758911, "learning_rate": 0.0005, "loss": 1.4917073249816895, "step": 23101 }, { "epoch": 1.8212061489948759, "grad_norm": 0.45978426933288574, "learning_rate": 0.0005, "loss": 1.3265405893325806, "step": 23102 }, { "epoch": 1.8212849822625148, "grad_norm": 0.5907807350158691, "learning_rate": 0.0005, "loss": 1.6387531757354736, "step": 23103 }, { "epoch": 1.8213638155301537, "grad_norm": 0.7386070489883423, "learning_rate": 0.0005, "loss": 1.5752848386764526, "step": 23104 }, { "epoch": 1.8214426487977926, "grad_norm": 0.5225613117218018, "learning_rate": 0.0005, "loss": 1.259178876876831, "step": 23105 }, { "epoch": 1.8215214820654317, "grad_norm": 0.5071259140968323, "learning_rate": 0.0005, "loss": 1.4728559255599976, "step": 23106 }, { "epoch": 1.8216003153330704, "grad_norm": 0.4842360019683838, "learning_rate": 0.0005, "loss": 1.4129751920700073, "step": 23107 }, { "epoch": 1.8216791486007096, "grad_norm": 0.5265721082687378, "learning_rate": 0.0005, "loss": 1.632142424583435, "step": 23108 }, { "epoch": 1.8217579818683485, "grad_norm": 0.8462803959846497, "learning_rate": 0.0005, "loss": 1.3127981424331665, "step": 23109 }, { "epoch": 1.8218368151359874, "grad_norm": 0.6148164868354797, "learning_rate": 0.0005, "loss": 1.620774745941162, "step": 23110 }, { "epoch": 1.8219156484036263, "grad_norm": 0.510895311832428, "learning_rate": 0.0005, "loss": 1.2346442937850952, "step": 23111 }, { "epoch": 1.8219944816712652, "grad_norm": 0.4922958314418793, "learning_rate": 0.0005, "loss": 1.39780855178833, "step": 23112 }, { "epoch": 1.8220733149389043, "grad_norm": 0.642118513584137, "learning_rate": 0.0005, "loss": 1.6508386135101318, "step": 23113 }, { "epoch": 1.822152148206543, "grad_norm": 0.5642767548561096, "learning_rate": 0.0005, "loss": 1.3502922058105469, "step": 23114 }, { "epoch": 1.8222309814741822, "grad_norm": 0.6357119083404541, "learning_rate": 0.0005, "loss": 1.3981279134750366, "step": 23115 }, { "epoch": 1.822309814741821, "grad_norm": 0.5466272830963135, "learning_rate": 0.0005, "loss": 1.5668411254882812, "step": 23116 }, { "epoch": 1.82238864800946, "grad_norm": 0.5967015027999878, "learning_rate": 0.0005, "loss": 1.3640813827514648, "step": 23117 }, { "epoch": 1.822467481277099, "grad_norm": 0.5638597011566162, "learning_rate": 0.0005, "loss": 1.3805865049362183, "step": 23118 }, { "epoch": 1.8225463145447378, "grad_norm": 0.8293335437774658, "learning_rate": 0.0005, "loss": 1.4764068126678467, "step": 23119 }, { "epoch": 1.822625147812377, "grad_norm": 0.7007662653923035, "learning_rate": 0.0005, "loss": 1.4299806356430054, "step": 23120 }, { "epoch": 1.8227039810800156, "grad_norm": 0.5252344012260437, "learning_rate": 0.0005, "loss": 1.4071155786514282, "step": 23121 }, { "epoch": 1.8227828143476548, "grad_norm": 0.5944501757621765, "learning_rate": 0.0005, "loss": 1.604041576385498, "step": 23122 }, { "epoch": 1.8228616476152937, "grad_norm": 0.6310076117515564, "learning_rate": 0.0005, "loss": 1.15987229347229, "step": 23123 }, { "epoch": 1.8229404808829326, "grad_norm": 0.485439270734787, "learning_rate": 0.0005, "loss": 1.5754225254058838, "step": 23124 }, { "epoch": 1.8230193141505715, "grad_norm": 0.49173781275749207, "learning_rate": 0.0005, "loss": 1.4581539630889893, "step": 23125 }, { "epoch": 1.8230981474182104, "grad_norm": 0.6429556608200073, "learning_rate": 0.0005, "loss": 1.3259401321411133, "step": 23126 }, { "epoch": 1.8231769806858495, "grad_norm": 0.672414243221283, "learning_rate": 0.0005, "loss": 1.4633123874664307, "step": 23127 }, { "epoch": 1.8232558139534882, "grad_norm": 0.5974224209785461, "learning_rate": 0.0005, "loss": 1.5060265064239502, "step": 23128 }, { "epoch": 1.8233346472211274, "grad_norm": 0.6597726941108704, "learning_rate": 0.0005, "loss": 1.5036792755126953, "step": 23129 }, { "epoch": 1.8234134804887663, "grad_norm": 0.652574896812439, "learning_rate": 0.0005, "loss": 1.531653642654419, "step": 23130 }, { "epoch": 1.8234923137564052, "grad_norm": 0.48839637637138367, "learning_rate": 0.0005, "loss": 1.4145450592041016, "step": 23131 }, { "epoch": 1.823571147024044, "grad_norm": 0.9088963866233826, "learning_rate": 0.0005, "loss": 1.5116822719573975, "step": 23132 }, { "epoch": 1.823649980291683, "grad_norm": 0.5958254337310791, "learning_rate": 0.0005, "loss": 1.4182097911834717, "step": 23133 }, { "epoch": 1.8237288135593221, "grad_norm": 0.5535228252410889, "learning_rate": 0.0005, "loss": 1.537917971611023, "step": 23134 }, { "epoch": 1.8238076468269608, "grad_norm": 0.45967769622802734, "learning_rate": 0.0005, "loss": 1.3328666687011719, "step": 23135 }, { "epoch": 1.8238864800946, "grad_norm": 0.5080254077911377, "learning_rate": 0.0005, "loss": 1.3221158981323242, "step": 23136 }, { "epoch": 1.8239653133622389, "grad_norm": 0.577930748462677, "learning_rate": 0.0005, "loss": 1.3503282070159912, "step": 23137 }, { "epoch": 1.8240441466298778, "grad_norm": 0.5074753165245056, "learning_rate": 0.0005, "loss": 1.6099666357040405, "step": 23138 }, { "epoch": 1.8241229798975167, "grad_norm": 0.5578420758247375, "learning_rate": 0.0005, "loss": 1.4660699367523193, "step": 23139 }, { "epoch": 1.8242018131651556, "grad_norm": 0.8498037457466125, "learning_rate": 0.0005, "loss": 1.3371942043304443, "step": 23140 }, { "epoch": 1.8242806464327948, "grad_norm": 0.5593240857124329, "learning_rate": 0.0005, "loss": 1.272581934928894, "step": 23141 }, { "epoch": 1.8243594797004334, "grad_norm": 0.5776036381721497, "learning_rate": 0.0005, "loss": 1.4977940320968628, "step": 23142 }, { "epoch": 1.8244383129680726, "grad_norm": 0.5029624104499817, "learning_rate": 0.0005, "loss": 1.3435004949569702, "step": 23143 }, { "epoch": 1.8245171462357115, "grad_norm": 0.44456857442855835, "learning_rate": 0.0005, "loss": 1.3361070156097412, "step": 23144 }, { "epoch": 1.8245959795033504, "grad_norm": 0.6058523654937744, "learning_rate": 0.0005, "loss": 1.2418416738510132, "step": 23145 }, { "epoch": 1.8246748127709893, "grad_norm": 0.49845072627067566, "learning_rate": 0.0005, "loss": 1.3086735010147095, "step": 23146 }, { "epoch": 1.8247536460386282, "grad_norm": 0.5175688862800598, "learning_rate": 0.0005, "loss": 1.4577058553695679, "step": 23147 }, { "epoch": 1.8248324793062674, "grad_norm": 0.6574548482894897, "learning_rate": 0.0005, "loss": 1.1754616498947144, "step": 23148 }, { "epoch": 1.824911312573906, "grad_norm": 0.44016033411026, "learning_rate": 0.0005, "loss": 1.5578745603561401, "step": 23149 }, { "epoch": 1.8249901458415452, "grad_norm": 0.5277263522148132, "learning_rate": 0.0005, "loss": 1.2029304504394531, "step": 23150 }, { "epoch": 1.825068979109184, "grad_norm": 0.5168580412864685, "learning_rate": 0.0005, "loss": 1.3689366579055786, "step": 23151 }, { "epoch": 1.825147812376823, "grad_norm": 0.4633859097957611, "learning_rate": 0.0005, "loss": 1.4601439237594604, "step": 23152 }, { "epoch": 1.825226645644462, "grad_norm": 0.4493776857852936, "learning_rate": 0.0005, "loss": 1.4243535995483398, "step": 23153 }, { "epoch": 1.8253054789121008, "grad_norm": 0.4746842086315155, "learning_rate": 0.0005, "loss": 1.2766175270080566, "step": 23154 }, { "epoch": 1.82538431217974, "grad_norm": 0.8263513445854187, "learning_rate": 0.0005, "loss": 1.5026888847351074, "step": 23155 }, { "epoch": 1.8254631454473786, "grad_norm": 0.43688783049583435, "learning_rate": 0.0005, "loss": 1.2629327774047852, "step": 23156 }, { "epoch": 1.8255419787150178, "grad_norm": 0.6713218688964844, "learning_rate": 0.0005, "loss": 1.6081089973449707, "step": 23157 }, { "epoch": 1.8256208119826567, "grad_norm": 0.5431917905807495, "learning_rate": 0.0005, "loss": 1.4135873317718506, "step": 23158 }, { "epoch": 1.8256996452502956, "grad_norm": 0.4952802062034607, "learning_rate": 0.0005, "loss": 1.206880807876587, "step": 23159 }, { "epoch": 1.8257784785179347, "grad_norm": 0.6006522178649902, "learning_rate": 0.0005, "loss": 1.4338417053222656, "step": 23160 }, { "epoch": 1.8258573117855734, "grad_norm": 0.575497567653656, "learning_rate": 0.0005, "loss": 1.5693515539169312, "step": 23161 }, { "epoch": 1.8259361450532126, "grad_norm": 0.48363980650901794, "learning_rate": 0.0005, "loss": 1.3631553649902344, "step": 23162 }, { "epoch": 1.8260149783208512, "grad_norm": 0.8451667428016663, "learning_rate": 0.0005, "loss": 1.2746225595474243, "step": 23163 }, { "epoch": 1.8260938115884904, "grad_norm": 0.5010931491851807, "learning_rate": 0.0005, "loss": 1.388484001159668, "step": 23164 }, { "epoch": 1.8261726448561293, "grad_norm": 0.610578715801239, "learning_rate": 0.0005, "loss": 1.2467398643493652, "step": 23165 }, { "epoch": 1.8262514781237682, "grad_norm": 0.9483770728111267, "learning_rate": 0.0005, "loss": 1.6115474700927734, "step": 23166 }, { "epoch": 1.8263303113914073, "grad_norm": 0.43901029229164124, "learning_rate": 0.0005, "loss": 1.4078764915466309, "step": 23167 }, { "epoch": 1.826409144659046, "grad_norm": 0.5715233087539673, "learning_rate": 0.0005, "loss": 1.5447943210601807, "step": 23168 }, { "epoch": 1.8264879779266852, "grad_norm": 0.4233435094356537, "learning_rate": 0.0005, "loss": 1.1409456729888916, "step": 23169 }, { "epoch": 1.8265668111943238, "grad_norm": 0.4782620072364807, "learning_rate": 0.0005, "loss": 1.2454625368118286, "step": 23170 }, { "epoch": 1.826645644461963, "grad_norm": 0.6471235752105713, "learning_rate": 0.0005, "loss": 1.5720715522766113, "step": 23171 }, { "epoch": 1.826724477729602, "grad_norm": 0.4837944507598877, "learning_rate": 0.0005, "loss": 1.3342173099517822, "step": 23172 }, { "epoch": 1.8268033109972408, "grad_norm": 0.5372462868690491, "learning_rate": 0.0005, "loss": 1.3400781154632568, "step": 23173 }, { "epoch": 1.82688214426488, "grad_norm": 0.6053290963172913, "learning_rate": 0.0005, "loss": 1.3793330192565918, "step": 23174 }, { "epoch": 1.8269609775325186, "grad_norm": 0.7051578760147095, "learning_rate": 0.0005, "loss": 1.674834966659546, "step": 23175 }, { "epoch": 1.8270398108001578, "grad_norm": 0.5888729691505432, "learning_rate": 0.0005, "loss": 1.2905075550079346, "step": 23176 }, { "epoch": 1.8271186440677964, "grad_norm": 0.5324245095252991, "learning_rate": 0.0005, "loss": 1.481186866760254, "step": 23177 }, { "epoch": 1.8271974773354356, "grad_norm": 0.5416761040687561, "learning_rate": 0.0005, "loss": 1.5321022272109985, "step": 23178 }, { "epoch": 1.8272763106030745, "grad_norm": 0.523622453212738, "learning_rate": 0.0005, "loss": 1.6293455362319946, "step": 23179 }, { "epoch": 1.8273551438707134, "grad_norm": 0.4445190727710724, "learning_rate": 0.0005, "loss": 1.2791866064071655, "step": 23180 }, { "epoch": 1.8274339771383525, "grad_norm": 0.69785076379776, "learning_rate": 0.0005, "loss": 1.5090296268463135, "step": 23181 }, { "epoch": 1.8275128104059912, "grad_norm": 0.48440513014793396, "learning_rate": 0.0005, "loss": 1.4983099699020386, "step": 23182 }, { "epoch": 1.8275916436736304, "grad_norm": 0.5403756499290466, "learning_rate": 0.0005, "loss": 1.3413021564483643, "step": 23183 }, { "epoch": 1.827670476941269, "grad_norm": 0.5802545547485352, "learning_rate": 0.0005, "loss": 1.5633442401885986, "step": 23184 }, { "epoch": 1.8277493102089082, "grad_norm": 0.5275660157203674, "learning_rate": 0.0005, "loss": 1.600274920463562, "step": 23185 }, { "epoch": 1.827828143476547, "grad_norm": 0.44858384132385254, "learning_rate": 0.0005, "loss": 1.3115352392196655, "step": 23186 }, { "epoch": 1.827906976744186, "grad_norm": 0.7031864523887634, "learning_rate": 0.0005, "loss": 1.3603039979934692, "step": 23187 }, { "epoch": 1.8279858100118251, "grad_norm": 0.5241556763648987, "learning_rate": 0.0005, "loss": 1.5676158666610718, "step": 23188 }, { "epoch": 1.8280646432794638, "grad_norm": 0.47758421301841736, "learning_rate": 0.0005, "loss": 1.4337670803070068, "step": 23189 }, { "epoch": 1.828143476547103, "grad_norm": 0.5674795508384705, "learning_rate": 0.0005, "loss": 1.6018036603927612, "step": 23190 }, { "epoch": 1.8282223098147419, "grad_norm": 0.588385283946991, "learning_rate": 0.0005, "loss": 1.5624523162841797, "step": 23191 }, { "epoch": 1.8283011430823808, "grad_norm": 0.5003953576087952, "learning_rate": 0.0005, "loss": 1.579073190689087, "step": 23192 }, { "epoch": 1.8283799763500197, "grad_norm": 0.5269278287887573, "learning_rate": 0.0005, "loss": 1.6634209156036377, "step": 23193 }, { "epoch": 1.8284588096176586, "grad_norm": 0.5773500204086304, "learning_rate": 0.0005, "loss": 1.4397938251495361, "step": 23194 }, { "epoch": 1.8285376428852977, "grad_norm": 0.582288920879364, "learning_rate": 0.0005, "loss": 1.2968847751617432, "step": 23195 }, { "epoch": 1.8286164761529364, "grad_norm": 0.43913567066192627, "learning_rate": 0.0005, "loss": 1.4480760097503662, "step": 23196 }, { "epoch": 1.8286953094205756, "grad_norm": 0.580078661441803, "learning_rate": 0.0005, "loss": 1.4476277828216553, "step": 23197 }, { "epoch": 1.8287741426882145, "grad_norm": 0.5517754554748535, "learning_rate": 0.0005, "loss": 1.3412694931030273, "step": 23198 }, { "epoch": 1.8288529759558534, "grad_norm": 0.553691029548645, "learning_rate": 0.0005, "loss": 1.499643325805664, "step": 23199 }, { "epoch": 1.8289318092234923, "grad_norm": 0.5359959006309509, "learning_rate": 0.0005, "loss": 1.524405598640442, "step": 23200 }, { "epoch": 1.8290106424911312, "grad_norm": 0.5158573389053345, "learning_rate": 0.0005, "loss": 1.4826922416687012, "step": 23201 }, { "epoch": 1.8290894757587703, "grad_norm": 0.5099812746047974, "learning_rate": 0.0005, "loss": 1.3918886184692383, "step": 23202 }, { "epoch": 1.829168309026409, "grad_norm": 0.9525508880615234, "learning_rate": 0.0005, "loss": 1.5672839879989624, "step": 23203 }, { "epoch": 1.8292471422940482, "grad_norm": 0.5490679740905762, "learning_rate": 0.0005, "loss": 1.2889361381530762, "step": 23204 }, { "epoch": 1.829325975561687, "grad_norm": 0.5561361312866211, "learning_rate": 0.0005, "loss": 1.5469154119491577, "step": 23205 }, { "epoch": 1.829404808829326, "grad_norm": 0.5582130551338196, "learning_rate": 0.0005, "loss": 1.5389559268951416, "step": 23206 }, { "epoch": 1.829483642096965, "grad_norm": 0.5031798481941223, "learning_rate": 0.0005, "loss": 1.4568181037902832, "step": 23207 }, { "epoch": 1.8295624753646038, "grad_norm": 0.4460766017436981, "learning_rate": 0.0005, "loss": 1.2883379459381104, "step": 23208 }, { "epoch": 1.829641308632243, "grad_norm": 0.5159488916397095, "learning_rate": 0.0005, "loss": 1.4747041463851929, "step": 23209 }, { "epoch": 1.8297201418998816, "grad_norm": 0.45126768946647644, "learning_rate": 0.0005, "loss": 1.2628366947174072, "step": 23210 }, { "epoch": 1.8297989751675208, "grad_norm": 0.5241240859031677, "learning_rate": 0.0005, "loss": 1.5049461126327515, "step": 23211 }, { "epoch": 1.8298778084351597, "grad_norm": 0.4718007445335388, "learning_rate": 0.0005, "loss": 1.2881035804748535, "step": 23212 }, { "epoch": 1.8299566417027986, "grad_norm": 0.5929357409477234, "learning_rate": 0.0005, "loss": 1.1500298976898193, "step": 23213 }, { "epoch": 1.8300354749704375, "grad_norm": 0.8359423279762268, "learning_rate": 0.0005, "loss": 1.3490766286849976, "step": 23214 }, { "epoch": 1.8301143082380764, "grad_norm": 0.3878961503505707, "learning_rate": 0.0005, "loss": 1.1693018674850464, "step": 23215 }, { "epoch": 1.8301931415057155, "grad_norm": 0.593203067779541, "learning_rate": 0.0005, "loss": 1.713895559310913, "step": 23216 }, { "epoch": 1.8302719747733542, "grad_norm": 0.5295685529708862, "learning_rate": 0.0005, "loss": 1.1958694458007812, "step": 23217 }, { "epoch": 1.8303508080409934, "grad_norm": 7.11527681350708, "learning_rate": 0.0005, "loss": 1.4772006273269653, "step": 23218 }, { "epoch": 1.8304296413086323, "grad_norm": 0.6302016377449036, "learning_rate": 0.0005, "loss": 1.6860955953598022, "step": 23219 }, { "epoch": 1.8305084745762712, "grad_norm": 0.4297496974468231, "learning_rate": 0.0005, "loss": 1.2825946807861328, "step": 23220 }, { "epoch": 1.83058730784391, "grad_norm": 0.482278972864151, "learning_rate": 0.0005, "loss": 1.5968830585479736, "step": 23221 }, { "epoch": 1.830666141111549, "grad_norm": 0.5196806788444519, "learning_rate": 0.0005, "loss": 1.545278787612915, "step": 23222 }, { "epoch": 1.8307449743791882, "grad_norm": 0.49250921607017517, "learning_rate": 0.0005, "loss": 1.5928385257720947, "step": 23223 }, { "epoch": 1.8308238076468268, "grad_norm": 0.5245506763458252, "learning_rate": 0.0005, "loss": 1.5984266996383667, "step": 23224 }, { "epoch": 1.830902640914466, "grad_norm": 0.5533901453018188, "learning_rate": 0.0005, "loss": 1.1775542497634888, "step": 23225 }, { "epoch": 1.8309814741821049, "grad_norm": 0.4838946759700775, "learning_rate": 0.0005, "loss": 1.2601966857910156, "step": 23226 }, { "epoch": 1.8310603074497438, "grad_norm": 0.5927227139472961, "learning_rate": 0.0005, "loss": 1.553652048110962, "step": 23227 }, { "epoch": 1.8311391407173827, "grad_norm": 0.6387679576873779, "learning_rate": 0.0005, "loss": 1.3371638059616089, "step": 23228 }, { "epoch": 1.8312179739850216, "grad_norm": 0.5876486897468567, "learning_rate": 0.0005, "loss": 1.4052391052246094, "step": 23229 }, { "epoch": 1.8312968072526608, "grad_norm": 0.5332551002502441, "learning_rate": 0.0005, "loss": 1.3901649713516235, "step": 23230 }, { "epoch": 1.8313756405202994, "grad_norm": 0.5210973024368286, "learning_rate": 0.0005, "loss": 1.3985153436660767, "step": 23231 }, { "epoch": 1.8314544737879386, "grad_norm": 0.507850170135498, "learning_rate": 0.0005, "loss": 1.3518459796905518, "step": 23232 }, { "epoch": 1.8315333070555775, "grad_norm": 0.5762166380882263, "learning_rate": 0.0005, "loss": 1.4259767532348633, "step": 23233 }, { "epoch": 1.8316121403232164, "grad_norm": 0.545634388923645, "learning_rate": 0.0005, "loss": 1.3964366912841797, "step": 23234 }, { "epoch": 1.8316909735908553, "grad_norm": 0.5670674443244934, "learning_rate": 0.0005, "loss": 1.332843542098999, "step": 23235 }, { "epoch": 1.8317698068584942, "grad_norm": 0.49740374088287354, "learning_rate": 0.0005, "loss": 1.5246310234069824, "step": 23236 }, { "epoch": 1.8318486401261334, "grad_norm": 0.6456851959228516, "learning_rate": 0.0005, "loss": 1.4354307651519775, "step": 23237 }, { "epoch": 1.831927473393772, "grad_norm": 0.5571869015693665, "learning_rate": 0.0005, "loss": 1.3749052286148071, "step": 23238 }, { "epoch": 1.8320063066614112, "grad_norm": 0.6115732192993164, "learning_rate": 0.0005, "loss": 1.5935053825378418, "step": 23239 }, { "epoch": 1.83208513992905, "grad_norm": 0.5109109878540039, "learning_rate": 0.0005, "loss": 1.4907621145248413, "step": 23240 }, { "epoch": 1.832163973196689, "grad_norm": 0.5039693117141724, "learning_rate": 0.0005, "loss": 1.2167489528656006, "step": 23241 }, { "epoch": 1.832242806464328, "grad_norm": 0.6058614253997803, "learning_rate": 0.0005, "loss": 1.419785499572754, "step": 23242 }, { "epoch": 1.8323216397319668, "grad_norm": 0.7069304585456848, "learning_rate": 0.0005, "loss": 1.5531456470489502, "step": 23243 }, { "epoch": 1.832400472999606, "grad_norm": 0.7467698454856873, "learning_rate": 0.0005, "loss": 1.4392989873886108, "step": 23244 }, { "epoch": 1.8324793062672446, "grad_norm": 0.45918673276901245, "learning_rate": 0.0005, "loss": 1.2558047771453857, "step": 23245 }, { "epoch": 1.8325581395348838, "grad_norm": 0.6426324248313904, "learning_rate": 0.0005, "loss": 1.3888180255889893, "step": 23246 }, { "epoch": 1.8326369728025227, "grad_norm": 0.7483536601066589, "learning_rate": 0.0005, "loss": 1.5884044170379639, "step": 23247 }, { "epoch": 1.8327158060701616, "grad_norm": 0.5401531457901001, "learning_rate": 0.0005, "loss": 1.3699685335159302, "step": 23248 }, { "epoch": 1.8327946393378005, "grad_norm": 0.522011935710907, "learning_rate": 0.0005, "loss": 1.345535397529602, "step": 23249 }, { "epoch": 1.8328734726054394, "grad_norm": 0.6356693506240845, "learning_rate": 0.0005, "loss": 1.4201486110687256, "step": 23250 }, { "epoch": 1.8329523058730786, "grad_norm": 0.4700799584388733, "learning_rate": 0.0005, "loss": 1.4665557146072388, "step": 23251 }, { "epoch": 1.8330311391407172, "grad_norm": 0.6430524587631226, "learning_rate": 0.0005, "loss": 1.4417884349822998, "step": 23252 }, { "epoch": 1.8331099724083564, "grad_norm": 0.4885501265525818, "learning_rate": 0.0005, "loss": 1.3979829549789429, "step": 23253 }, { "epoch": 1.8331888056759953, "grad_norm": 0.5546348094940186, "learning_rate": 0.0005, "loss": 1.4280294179916382, "step": 23254 }, { "epoch": 1.8332676389436342, "grad_norm": 0.5955674648284912, "learning_rate": 0.0005, "loss": 1.4332094192504883, "step": 23255 }, { "epoch": 1.8333464722112731, "grad_norm": 0.46145808696746826, "learning_rate": 0.0005, "loss": 1.429762601852417, "step": 23256 }, { "epoch": 1.833425305478912, "grad_norm": 0.463565468788147, "learning_rate": 0.0005, "loss": 1.4861454963684082, "step": 23257 }, { "epoch": 1.8335041387465512, "grad_norm": 0.5853506326675415, "learning_rate": 0.0005, "loss": 1.2963614463806152, "step": 23258 }, { "epoch": 1.8335829720141898, "grad_norm": 0.535658597946167, "learning_rate": 0.0005, "loss": 1.2875139713287354, "step": 23259 }, { "epoch": 1.833661805281829, "grad_norm": 0.6660099625587463, "learning_rate": 0.0005, "loss": 1.4081549644470215, "step": 23260 }, { "epoch": 1.833740638549468, "grad_norm": 0.7519803643226624, "learning_rate": 0.0005, "loss": 1.328193187713623, "step": 23261 }, { "epoch": 1.8338194718171068, "grad_norm": 0.8150760531425476, "learning_rate": 0.0005, "loss": 1.521477460861206, "step": 23262 }, { "epoch": 1.8338983050847457, "grad_norm": 0.5285564064979553, "learning_rate": 0.0005, "loss": 1.3395718336105347, "step": 23263 }, { "epoch": 1.8339771383523846, "grad_norm": 0.5864153504371643, "learning_rate": 0.0005, "loss": 1.571298360824585, "step": 23264 }, { "epoch": 1.8340559716200238, "grad_norm": 0.5924807190895081, "learning_rate": 0.0005, "loss": 1.3482296466827393, "step": 23265 }, { "epoch": 1.8341348048876625, "grad_norm": 0.5044493079185486, "learning_rate": 0.0005, "loss": 1.3638471364974976, "step": 23266 }, { "epoch": 1.8342136381553016, "grad_norm": 0.4914510250091553, "learning_rate": 0.0005, "loss": 1.3132836818695068, "step": 23267 }, { "epoch": 1.8342924714229405, "grad_norm": 0.6490963101387024, "learning_rate": 0.0005, "loss": 1.7253916263580322, "step": 23268 }, { "epoch": 1.8343713046905794, "grad_norm": 0.5109258890151978, "learning_rate": 0.0005, "loss": 1.458082914352417, "step": 23269 }, { "epoch": 1.8344501379582183, "grad_norm": 0.5582271814346313, "learning_rate": 0.0005, "loss": 1.2577288150787354, "step": 23270 }, { "epoch": 1.8345289712258572, "grad_norm": 0.47571322321891785, "learning_rate": 0.0005, "loss": 1.0876855850219727, "step": 23271 }, { "epoch": 1.8346078044934964, "grad_norm": 0.5421921610832214, "learning_rate": 0.0005, "loss": 1.343474268913269, "step": 23272 }, { "epoch": 1.834686637761135, "grad_norm": 0.5772224068641663, "learning_rate": 0.0005, "loss": 1.5387511253356934, "step": 23273 }, { "epoch": 1.8347654710287742, "grad_norm": 0.5537532567977905, "learning_rate": 0.0005, "loss": 1.3843786716461182, "step": 23274 }, { "epoch": 1.834844304296413, "grad_norm": 0.5309494733810425, "learning_rate": 0.0005, "loss": 1.3201994895935059, "step": 23275 }, { "epoch": 1.834923137564052, "grad_norm": 0.5252618193626404, "learning_rate": 0.0005, "loss": 1.3052219152450562, "step": 23276 }, { "epoch": 1.835001970831691, "grad_norm": 0.7390072345733643, "learning_rate": 0.0005, "loss": 1.4923478364944458, "step": 23277 }, { "epoch": 1.8350808040993298, "grad_norm": 0.49570170044898987, "learning_rate": 0.0005, "loss": 1.4002208709716797, "step": 23278 }, { "epoch": 1.835159637366969, "grad_norm": 0.5112387537956238, "learning_rate": 0.0005, "loss": 1.1874873638153076, "step": 23279 }, { "epoch": 1.8352384706346077, "grad_norm": 0.5669663548469543, "learning_rate": 0.0005, "loss": 1.4402620792388916, "step": 23280 }, { "epoch": 1.8353173039022468, "grad_norm": 0.6299240589141846, "learning_rate": 0.0005, "loss": 1.4225194454193115, "step": 23281 }, { "epoch": 1.8353961371698857, "grad_norm": 0.6961755752563477, "learning_rate": 0.0005, "loss": 1.456176996231079, "step": 23282 }, { "epoch": 1.8354749704375246, "grad_norm": 0.670085608959198, "learning_rate": 0.0005, "loss": 1.6724319458007812, "step": 23283 }, { "epoch": 1.8355538037051637, "grad_norm": 0.5086932182312012, "learning_rate": 0.0005, "loss": 1.1788864135742188, "step": 23284 }, { "epoch": 1.8356326369728024, "grad_norm": 0.7157238721847534, "learning_rate": 0.0005, "loss": 1.3950071334838867, "step": 23285 }, { "epoch": 1.8357114702404416, "grad_norm": 0.5958041548728943, "learning_rate": 0.0005, "loss": 1.3753526210784912, "step": 23286 }, { "epoch": 1.8357903035080803, "grad_norm": 0.6959262490272522, "learning_rate": 0.0005, "loss": 1.540252923965454, "step": 23287 }, { "epoch": 1.8358691367757194, "grad_norm": 0.599137544631958, "learning_rate": 0.0005, "loss": 1.3131465911865234, "step": 23288 }, { "epoch": 1.8359479700433583, "grad_norm": 0.5769075751304626, "learning_rate": 0.0005, "loss": 1.57562255859375, "step": 23289 }, { "epoch": 1.8360268033109972, "grad_norm": 0.5462573170661926, "learning_rate": 0.0005, "loss": 1.474766492843628, "step": 23290 }, { "epoch": 1.8361056365786363, "grad_norm": 0.6067498922348022, "learning_rate": 0.0005, "loss": 1.4297555685043335, "step": 23291 }, { "epoch": 1.836184469846275, "grad_norm": 0.5256586074829102, "learning_rate": 0.0005, "loss": 1.3471336364746094, "step": 23292 }, { "epoch": 1.8362633031139142, "grad_norm": 0.40619131922721863, "learning_rate": 0.0005, "loss": 1.0881068706512451, "step": 23293 }, { "epoch": 1.8363421363815529, "grad_norm": 0.5498100519180298, "learning_rate": 0.0005, "loss": 1.3617937564849854, "step": 23294 }, { "epoch": 1.836420969649192, "grad_norm": 0.6266626715660095, "learning_rate": 0.0005, "loss": 1.4829802513122559, "step": 23295 }, { "epoch": 1.836499802916831, "grad_norm": 0.4906150698661804, "learning_rate": 0.0005, "loss": 1.4325149059295654, "step": 23296 }, { "epoch": 1.8365786361844698, "grad_norm": 0.9773784875869751, "learning_rate": 0.0005, "loss": 1.7419977188110352, "step": 23297 }, { "epoch": 1.836657469452109, "grad_norm": 0.5547174215316772, "learning_rate": 0.0005, "loss": 1.1307404041290283, "step": 23298 }, { "epoch": 1.8367363027197476, "grad_norm": 0.5122674703598022, "learning_rate": 0.0005, "loss": 1.3756489753723145, "step": 23299 }, { "epoch": 1.8368151359873868, "grad_norm": 0.5972546935081482, "learning_rate": 0.0005, "loss": 1.5427167415618896, "step": 23300 }, { "epoch": 1.8368939692550255, "grad_norm": 0.5226167440414429, "learning_rate": 0.0005, "loss": 1.577776312828064, "step": 23301 }, { "epoch": 1.8369728025226646, "grad_norm": 0.46961429715156555, "learning_rate": 0.0005, "loss": 1.3253555297851562, "step": 23302 }, { "epoch": 1.8370516357903035, "grad_norm": 0.6291471719741821, "learning_rate": 0.0005, "loss": 1.3752952814102173, "step": 23303 }, { "epoch": 1.8371304690579424, "grad_norm": 2.6016814708709717, "learning_rate": 0.0005, "loss": 1.389634609222412, "step": 23304 }, { "epoch": 1.8372093023255816, "grad_norm": 0.5214268565177917, "learning_rate": 0.0005, "loss": 1.3570406436920166, "step": 23305 }, { "epoch": 1.8372881355932202, "grad_norm": 0.4770331084728241, "learning_rate": 0.0005, "loss": 1.288365364074707, "step": 23306 }, { "epoch": 1.8373669688608594, "grad_norm": 0.6178100109100342, "learning_rate": 0.0005, "loss": 1.7144076824188232, "step": 23307 }, { "epoch": 1.837445802128498, "grad_norm": 0.6416524052619934, "learning_rate": 0.0005, "loss": 1.3584723472595215, "step": 23308 }, { "epoch": 1.8375246353961372, "grad_norm": 0.6383986473083496, "learning_rate": 0.0005, "loss": 1.2341346740722656, "step": 23309 }, { "epoch": 1.837603468663776, "grad_norm": 0.48897314071655273, "learning_rate": 0.0005, "loss": 1.393867015838623, "step": 23310 }, { "epoch": 1.837682301931415, "grad_norm": 0.5422999858856201, "learning_rate": 0.0005, "loss": 1.5761706829071045, "step": 23311 }, { "epoch": 1.8377611351990542, "grad_norm": 0.545076310634613, "learning_rate": 0.0005, "loss": 1.2818808555603027, "step": 23312 }, { "epoch": 1.8378399684666928, "grad_norm": 0.5561673045158386, "learning_rate": 0.0005, "loss": 1.3103101253509521, "step": 23313 }, { "epoch": 1.837918801734332, "grad_norm": 0.5391635894775391, "learning_rate": 0.0005, "loss": 1.467985987663269, "step": 23314 }, { "epoch": 1.8379976350019709, "grad_norm": 0.7719009518623352, "learning_rate": 0.0005, "loss": 1.4639906883239746, "step": 23315 }, { "epoch": 1.8380764682696098, "grad_norm": 0.7085506319999695, "learning_rate": 0.0005, "loss": 1.417564034461975, "step": 23316 }, { "epoch": 1.8381553015372487, "grad_norm": 0.4777352213859558, "learning_rate": 0.0005, "loss": 1.3457221984863281, "step": 23317 }, { "epoch": 1.8382341348048876, "grad_norm": 0.5699125528335571, "learning_rate": 0.0005, "loss": 1.56608247756958, "step": 23318 }, { "epoch": 1.8383129680725268, "grad_norm": 0.5917111039161682, "learning_rate": 0.0005, "loss": 1.4079334735870361, "step": 23319 }, { "epoch": 1.8383918013401654, "grad_norm": 0.7570137977600098, "learning_rate": 0.0005, "loss": 1.5746042728424072, "step": 23320 }, { "epoch": 1.8384706346078046, "grad_norm": 0.6203581094741821, "learning_rate": 0.0005, "loss": 1.35932195186615, "step": 23321 }, { "epoch": 1.8385494678754435, "grad_norm": 0.5483229160308838, "learning_rate": 0.0005, "loss": 1.4521989822387695, "step": 23322 }, { "epoch": 1.8386283011430824, "grad_norm": 0.6185917258262634, "learning_rate": 0.0005, "loss": 1.3907053470611572, "step": 23323 }, { "epoch": 1.8387071344107213, "grad_norm": 0.47575798630714417, "learning_rate": 0.0005, "loss": 1.2453402280807495, "step": 23324 }, { "epoch": 1.8387859676783602, "grad_norm": 0.7017619013786316, "learning_rate": 0.0005, "loss": 1.4265621900558472, "step": 23325 }, { "epoch": 1.8388648009459994, "grad_norm": 0.7800955772399902, "learning_rate": 0.0005, "loss": 1.4963550567626953, "step": 23326 }, { "epoch": 1.838943634213638, "grad_norm": 0.6004244089126587, "learning_rate": 0.0005, "loss": 1.495938777923584, "step": 23327 }, { "epoch": 1.8390224674812772, "grad_norm": 0.5637751221656799, "learning_rate": 0.0005, "loss": 1.3885736465454102, "step": 23328 }, { "epoch": 1.839101300748916, "grad_norm": 0.6102562546730042, "learning_rate": 0.0005, "loss": 1.5201059579849243, "step": 23329 }, { "epoch": 1.839180134016555, "grad_norm": 0.479554682970047, "learning_rate": 0.0005, "loss": 1.541595697402954, "step": 23330 }, { "epoch": 1.839258967284194, "grad_norm": 0.4447583258152008, "learning_rate": 0.0005, "loss": 1.2589302062988281, "step": 23331 }, { "epoch": 1.8393378005518328, "grad_norm": 0.5860775709152222, "learning_rate": 0.0005, "loss": 1.742018461227417, "step": 23332 }, { "epoch": 1.839416633819472, "grad_norm": 0.473245769739151, "learning_rate": 0.0005, "loss": 1.4550707340240479, "step": 23333 }, { "epoch": 1.8394954670871106, "grad_norm": 0.6163071990013123, "learning_rate": 0.0005, "loss": 1.6150035858154297, "step": 23334 }, { "epoch": 1.8395743003547498, "grad_norm": 0.5504190921783447, "learning_rate": 0.0005, "loss": 1.3452894687652588, "step": 23335 }, { "epoch": 1.8396531336223887, "grad_norm": 0.49671661853790283, "learning_rate": 0.0005, "loss": 1.4554390907287598, "step": 23336 }, { "epoch": 1.8397319668900276, "grad_norm": 0.5130431652069092, "learning_rate": 0.0005, "loss": 1.4996888637542725, "step": 23337 }, { "epoch": 1.8398108001576665, "grad_norm": 0.47327423095703125, "learning_rate": 0.0005, "loss": 1.2184648513793945, "step": 23338 }, { "epoch": 1.8398896334253054, "grad_norm": 0.4578743875026703, "learning_rate": 0.0005, "loss": 1.131885051727295, "step": 23339 }, { "epoch": 1.8399684666929446, "grad_norm": 0.5040324330329895, "learning_rate": 0.0005, "loss": 1.642090082168579, "step": 23340 }, { "epoch": 1.8400472999605832, "grad_norm": 0.46691858768463135, "learning_rate": 0.0005, "loss": 1.2916693687438965, "step": 23341 }, { "epoch": 1.8401261332282224, "grad_norm": 0.5856416821479797, "learning_rate": 0.0005, "loss": 1.3772400617599487, "step": 23342 }, { "epoch": 1.8402049664958613, "grad_norm": 0.519212007522583, "learning_rate": 0.0005, "loss": 1.564565896987915, "step": 23343 }, { "epoch": 1.8402837997635002, "grad_norm": 0.6516175270080566, "learning_rate": 0.0005, "loss": 1.5872294902801514, "step": 23344 }, { "epoch": 1.8403626330311391, "grad_norm": 0.526123046875, "learning_rate": 0.0005, "loss": 1.5074293613433838, "step": 23345 }, { "epoch": 1.840441466298778, "grad_norm": 0.6700418591499329, "learning_rate": 0.0005, "loss": 1.42311692237854, "step": 23346 }, { "epoch": 1.8405202995664172, "grad_norm": 0.6610210537910461, "learning_rate": 0.0005, "loss": 1.4445183277130127, "step": 23347 }, { "epoch": 1.8405991328340559, "grad_norm": 0.566655695438385, "learning_rate": 0.0005, "loss": 1.3963422775268555, "step": 23348 }, { "epoch": 1.840677966101695, "grad_norm": 0.47164809703826904, "learning_rate": 0.0005, "loss": 1.3590035438537598, "step": 23349 }, { "epoch": 1.840756799369334, "grad_norm": 0.530436635017395, "learning_rate": 0.0005, "loss": 1.5385830402374268, "step": 23350 }, { "epoch": 1.8408356326369728, "grad_norm": 0.4783095121383667, "learning_rate": 0.0005, "loss": 1.5218315124511719, "step": 23351 }, { "epoch": 1.8409144659046117, "grad_norm": 0.5457934737205505, "learning_rate": 0.0005, "loss": 1.4897480010986328, "step": 23352 }, { "epoch": 1.8409932991722506, "grad_norm": 0.5481833219528198, "learning_rate": 0.0005, "loss": 1.3242912292480469, "step": 23353 }, { "epoch": 1.8410721324398898, "grad_norm": 0.7978190183639526, "learning_rate": 0.0005, "loss": 1.5441241264343262, "step": 23354 }, { "epoch": 1.8411509657075285, "grad_norm": 0.5601710677146912, "learning_rate": 0.0005, "loss": 1.3342444896697998, "step": 23355 }, { "epoch": 1.8412297989751676, "grad_norm": 0.6540018320083618, "learning_rate": 0.0005, "loss": 1.7055883407592773, "step": 23356 }, { "epoch": 1.8413086322428065, "grad_norm": 0.6004454493522644, "learning_rate": 0.0005, "loss": 1.4364793300628662, "step": 23357 }, { "epoch": 1.8413874655104454, "grad_norm": 0.5005871653556824, "learning_rate": 0.0005, "loss": 1.427042007446289, "step": 23358 }, { "epoch": 1.8414662987780843, "grad_norm": 0.5985106229782104, "learning_rate": 0.0005, "loss": 1.536594271659851, "step": 23359 }, { "epoch": 1.8415451320457232, "grad_norm": 0.6285635828971863, "learning_rate": 0.0005, "loss": 1.4138153791427612, "step": 23360 }, { "epoch": 1.8416239653133624, "grad_norm": 0.45997804403305054, "learning_rate": 0.0005, "loss": 1.2038707733154297, "step": 23361 }, { "epoch": 1.841702798581001, "grad_norm": 0.5135199427604675, "learning_rate": 0.0005, "loss": 1.497545838356018, "step": 23362 }, { "epoch": 1.8417816318486402, "grad_norm": 0.5209810733795166, "learning_rate": 0.0005, "loss": 1.3321419954299927, "step": 23363 }, { "epoch": 1.841860465116279, "grad_norm": 0.46754229068756104, "learning_rate": 0.0005, "loss": 1.542872667312622, "step": 23364 }, { "epoch": 1.841939298383918, "grad_norm": 0.4949207901954651, "learning_rate": 0.0005, "loss": 1.1807365417480469, "step": 23365 }, { "epoch": 1.842018131651557, "grad_norm": 0.45596206188201904, "learning_rate": 0.0005, "loss": 1.390358567237854, "step": 23366 }, { "epoch": 1.8420969649191958, "grad_norm": 0.5076777935028076, "learning_rate": 0.0005, "loss": 1.51216721534729, "step": 23367 }, { "epoch": 1.842175798186835, "grad_norm": 0.5724864602088928, "learning_rate": 0.0005, "loss": 1.3742649555206299, "step": 23368 }, { "epoch": 1.8422546314544737, "grad_norm": 0.6228057146072388, "learning_rate": 0.0005, "loss": 1.5512135028839111, "step": 23369 }, { "epoch": 1.8423334647221128, "grad_norm": 2.1368541717529297, "learning_rate": 0.0005, "loss": 1.3230063915252686, "step": 23370 }, { "epoch": 1.8424122979897517, "grad_norm": 0.5018858313560486, "learning_rate": 0.0005, "loss": 1.433545708656311, "step": 23371 }, { "epoch": 1.8424911312573906, "grad_norm": 0.5339753031730652, "learning_rate": 0.0005, "loss": 1.3258488178253174, "step": 23372 }, { "epoch": 1.8425699645250295, "grad_norm": 0.5184809565544128, "learning_rate": 0.0005, "loss": 1.5048450231552124, "step": 23373 }, { "epoch": 1.8426487977926684, "grad_norm": 0.6468438506126404, "learning_rate": 0.0005, "loss": 1.5115408897399902, "step": 23374 }, { "epoch": 1.8427276310603076, "grad_norm": 0.590343713760376, "learning_rate": 0.0005, "loss": 1.541229486465454, "step": 23375 }, { "epoch": 1.8428064643279463, "grad_norm": 0.5977745652198792, "learning_rate": 0.0005, "loss": 1.5572428703308105, "step": 23376 }, { "epoch": 1.8428852975955854, "grad_norm": 0.4957522451877594, "learning_rate": 0.0005, "loss": 1.2291595935821533, "step": 23377 }, { "epoch": 1.8429641308632243, "grad_norm": 0.5964626669883728, "learning_rate": 0.0005, "loss": 1.4333748817443848, "step": 23378 }, { "epoch": 1.8430429641308632, "grad_norm": 0.5920746326446533, "learning_rate": 0.0005, "loss": 1.3548705577850342, "step": 23379 }, { "epoch": 1.8431217973985021, "grad_norm": 0.5535283088684082, "learning_rate": 0.0005, "loss": 1.357829213142395, "step": 23380 }, { "epoch": 1.843200630666141, "grad_norm": 0.689247190952301, "learning_rate": 0.0005, "loss": 1.5618870258331299, "step": 23381 }, { "epoch": 1.8432794639337802, "grad_norm": 0.4557496905326843, "learning_rate": 0.0005, "loss": 1.3087990283966064, "step": 23382 }, { "epoch": 1.8433582972014189, "grad_norm": 0.5174607634544373, "learning_rate": 0.0005, "loss": 1.3142328262329102, "step": 23383 }, { "epoch": 1.843437130469058, "grad_norm": 0.634797215461731, "learning_rate": 0.0005, "loss": 1.5208293199539185, "step": 23384 }, { "epoch": 1.843515963736697, "grad_norm": 0.6258471608161926, "learning_rate": 0.0005, "loss": 1.5757215023040771, "step": 23385 }, { "epoch": 1.8435947970043358, "grad_norm": 0.569120466709137, "learning_rate": 0.0005, "loss": 1.2294408082962036, "step": 23386 }, { "epoch": 1.8436736302719747, "grad_norm": 0.46925869584083557, "learning_rate": 0.0005, "loss": 1.4401568174362183, "step": 23387 }, { "epoch": 1.8437524635396136, "grad_norm": 0.6022915840148926, "learning_rate": 0.0005, "loss": 1.5017859935760498, "step": 23388 }, { "epoch": 1.8438312968072528, "grad_norm": 0.5188673734664917, "learning_rate": 0.0005, "loss": 1.2952477931976318, "step": 23389 }, { "epoch": 1.8439101300748915, "grad_norm": 0.5035306811332703, "learning_rate": 0.0005, "loss": 1.2242612838745117, "step": 23390 }, { "epoch": 1.8439889633425306, "grad_norm": 0.6306962966918945, "learning_rate": 0.0005, "loss": 1.6338531970977783, "step": 23391 }, { "epoch": 1.8440677966101695, "grad_norm": 0.599274754524231, "learning_rate": 0.0005, "loss": 1.3454656600952148, "step": 23392 }, { "epoch": 1.8441466298778084, "grad_norm": 0.643869936466217, "learning_rate": 0.0005, "loss": 1.485562801361084, "step": 23393 }, { "epoch": 1.8442254631454473, "grad_norm": 0.4995450973510742, "learning_rate": 0.0005, "loss": 1.3707685470581055, "step": 23394 }, { "epoch": 1.8443042964130862, "grad_norm": 0.5637321472167969, "learning_rate": 0.0005, "loss": 1.4139636754989624, "step": 23395 }, { "epoch": 1.8443831296807254, "grad_norm": 0.4731181859970093, "learning_rate": 0.0005, "loss": 1.343681812286377, "step": 23396 }, { "epoch": 1.844461962948364, "grad_norm": 0.45625969767570496, "learning_rate": 0.0005, "loss": 1.1317435503005981, "step": 23397 }, { "epoch": 1.8445407962160032, "grad_norm": 0.48440223932266235, "learning_rate": 0.0005, "loss": 1.3569626808166504, "step": 23398 }, { "epoch": 1.844619629483642, "grad_norm": 0.5794059634208679, "learning_rate": 0.0005, "loss": 1.2831188440322876, "step": 23399 }, { "epoch": 1.844698462751281, "grad_norm": 0.4352644383907318, "learning_rate": 0.0005, "loss": 1.266782283782959, "step": 23400 }, { "epoch": 1.84477729601892, "grad_norm": 0.575577974319458, "learning_rate": 0.0005, "loss": 1.3218315839767456, "step": 23401 }, { "epoch": 1.8448561292865588, "grad_norm": 0.565032422542572, "learning_rate": 0.0005, "loss": 1.2820777893066406, "step": 23402 }, { "epoch": 1.844934962554198, "grad_norm": 0.5011157989501953, "learning_rate": 0.0005, "loss": 1.2715160846710205, "step": 23403 }, { "epoch": 1.8450137958218367, "grad_norm": 0.5292083024978638, "learning_rate": 0.0005, "loss": 1.317386507987976, "step": 23404 }, { "epoch": 1.8450926290894758, "grad_norm": 0.6517220735549927, "learning_rate": 0.0005, "loss": 1.4844770431518555, "step": 23405 }, { "epoch": 1.8451714623571147, "grad_norm": 0.47695210576057434, "learning_rate": 0.0005, "loss": 1.3136632442474365, "step": 23406 }, { "epoch": 1.8452502956247536, "grad_norm": 0.5726032257080078, "learning_rate": 0.0005, "loss": 1.4909056425094604, "step": 23407 }, { "epoch": 1.8453291288923928, "grad_norm": 0.5235799551010132, "learning_rate": 0.0005, "loss": 1.2072720527648926, "step": 23408 }, { "epoch": 1.8454079621600314, "grad_norm": 0.5199565291404724, "learning_rate": 0.0005, "loss": 1.4416896104812622, "step": 23409 }, { "epoch": 1.8454867954276706, "grad_norm": 0.5463159680366516, "learning_rate": 0.0005, "loss": 1.3712702989578247, "step": 23410 }, { "epoch": 1.8455656286953093, "grad_norm": 0.4855205714702606, "learning_rate": 0.0005, "loss": 1.2049627304077148, "step": 23411 }, { "epoch": 1.8456444619629484, "grad_norm": 0.5478397011756897, "learning_rate": 0.0005, "loss": 1.4782795906066895, "step": 23412 }, { "epoch": 1.8457232952305873, "grad_norm": 0.549791157245636, "learning_rate": 0.0005, "loss": 1.4628536701202393, "step": 23413 }, { "epoch": 1.8458021284982262, "grad_norm": 0.5058426856994629, "learning_rate": 0.0005, "loss": 1.2616167068481445, "step": 23414 }, { "epoch": 1.8458809617658654, "grad_norm": 0.6511301398277283, "learning_rate": 0.0005, "loss": 1.581319808959961, "step": 23415 }, { "epoch": 1.845959795033504, "grad_norm": 0.4270801246166229, "learning_rate": 0.0005, "loss": 1.184985876083374, "step": 23416 }, { "epoch": 1.8460386283011432, "grad_norm": 0.6218233108520508, "learning_rate": 0.0005, "loss": 1.3508102893829346, "step": 23417 }, { "epoch": 1.8461174615687819, "grad_norm": 0.5194718837738037, "learning_rate": 0.0005, "loss": 1.545576572418213, "step": 23418 }, { "epoch": 1.846196294836421, "grad_norm": 0.7491401433944702, "learning_rate": 0.0005, "loss": 1.572843313217163, "step": 23419 }, { "epoch": 1.84627512810406, "grad_norm": 0.6055231690406799, "learning_rate": 0.0005, "loss": 1.4949384927749634, "step": 23420 }, { "epoch": 1.8463539613716988, "grad_norm": 0.5966270565986633, "learning_rate": 0.0005, "loss": 1.4337835311889648, "step": 23421 }, { "epoch": 1.846432794639338, "grad_norm": 0.6457280516624451, "learning_rate": 0.0005, "loss": 1.74051833152771, "step": 23422 }, { "epoch": 1.8465116279069766, "grad_norm": 0.49880021810531616, "learning_rate": 0.0005, "loss": 1.504023790359497, "step": 23423 }, { "epoch": 1.8465904611746158, "grad_norm": 0.6125290393829346, "learning_rate": 0.0005, "loss": 1.2513923645019531, "step": 23424 }, { "epoch": 1.8466692944422545, "grad_norm": 0.6863263845443726, "learning_rate": 0.0005, "loss": 1.399748682975769, "step": 23425 }, { "epoch": 1.8467481277098936, "grad_norm": 0.4648396968841553, "learning_rate": 0.0005, "loss": 1.1919991970062256, "step": 23426 }, { "epoch": 1.8468269609775325, "grad_norm": 0.6594045758247375, "learning_rate": 0.0005, "loss": 1.6341257095336914, "step": 23427 }, { "epoch": 1.8469057942451714, "grad_norm": 0.5765437483787537, "learning_rate": 0.0005, "loss": 1.7101504802703857, "step": 23428 }, { "epoch": 1.8469846275128106, "grad_norm": 0.5990508198738098, "learning_rate": 0.0005, "loss": 1.5914238691329956, "step": 23429 }, { "epoch": 1.8470634607804493, "grad_norm": 0.5294389128684998, "learning_rate": 0.0005, "loss": 1.3249292373657227, "step": 23430 }, { "epoch": 1.8471422940480884, "grad_norm": 0.4888399839401245, "learning_rate": 0.0005, "loss": 1.2968144416809082, "step": 23431 }, { "epoch": 1.847221127315727, "grad_norm": 0.5136693120002747, "learning_rate": 0.0005, "loss": 1.5745142698287964, "step": 23432 }, { "epoch": 1.8472999605833662, "grad_norm": 0.5117024183273315, "learning_rate": 0.0005, "loss": 1.3164265155792236, "step": 23433 }, { "epoch": 1.8473787938510051, "grad_norm": 0.49625203013420105, "learning_rate": 0.0005, "loss": 1.197678804397583, "step": 23434 }, { "epoch": 1.847457627118644, "grad_norm": 0.46143507957458496, "learning_rate": 0.0005, "loss": 1.4962267875671387, "step": 23435 }, { "epoch": 1.8475364603862832, "grad_norm": 0.5818212032318115, "learning_rate": 0.0005, "loss": 1.3017158508300781, "step": 23436 }, { "epoch": 1.8476152936539219, "grad_norm": 0.5402818322181702, "learning_rate": 0.0005, "loss": 1.4429538249969482, "step": 23437 }, { "epoch": 1.847694126921561, "grad_norm": 0.5831083655357361, "learning_rate": 0.0005, "loss": 1.4075877666473389, "step": 23438 }, { "epoch": 1.8477729601892, "grad_norm": 0.5110570788383484, "learning_rate": 0.0005, "loss": 1.4411836862564087, "step": 23439 }, { "epoch": 1.8478517934568388, "grad_norm": 0.606852114200592, "learning_rate": 0.0005, "loss": 1.3250341415405273, "step": 23440 }, { "epoch": 1.8479306267244777, "grad_norm": 0.4654426872730255, "learning_rate": 0.0005, "loss": 1.3139989376068115, "step": 23441 }, { "epoch": 1.8480094599921166, "grad_norm": 0.6198574900627136, "learning_rate": 0.0005, "loss": 1.4656786918640137, "step": 23442 }, { "epoch": 1.8480882932597558, "grad_norm": 0.48884832859039307, "learning_rate": 0.0005, "loss": 1.4597604274749756, "step": 23443 }, { "epoch": 1.8481671265273945, "grad_norm": 0.5989366173744202, "learning_rate": 0.0005, "loss": 1.5132485628128052, "step": 23444 }, { "epoch": 1.8482459597950336, "grad_norm": 0.5520968437194824, "learning_rate": 0.0005, "loss": 1.6713749170303345, "step": 23445 }, { "epoch": 1.8483247930626725, "grad_norm": 0.500352144241333, "learning_rate": 0.0005, "loss": 1.4205883741378784, "step": 23446 }, { "epoch": 1.8484036263303114, "grad_norm": 0.8239074349403381, "learning_rate": 0.0005, "loss": 1.8152930736541748, "step": 23447 }, { "epoch": 1.8484824595979503, "grad_norm": 0.6471874713897705, "learning_rate": 0.0005, "loss": 1.6759896278381348, "step": 23448 }, { "epoch": 1.8485612928655892, "grad_norm": 0.5568311214447021, "learning_rate": 0.0005, "loss": 1.4575035572052002, "step": 23449 }, { "epoch": 1.8486401261332284, "grad_norm": 0.5190111398696899, "learning_rate": 0.0005, "loss": 1.2855275869369507, "step": 23450 }, { "epoch": 1.848718959400867, "grad_norm": 0.4681684970855713, "learning_rate": 0.0005, "loss": 1.4121956825256348, "step": 23451 }, { "epoch": 1.8487977926685062, "grad_norm": 0.547496497631073, "learning_rate": 0.0005, "loss": 1.4640084505081177, "step": 23452 }, { "epoch": 1.848876625936145, "grad_norm": 0.5860222578048706, "learning_rate": 0.0005, "loss": 1.5397980213165283, "step": 23453 }, { "epoch": 1.848955459203784, "grad_norm": 0.574165940284729, "learning_rate": 0.0005, "loss": 1.3312394618988037, "step": 23454 }, { "epoch": 1.849034292471423, "grad_norm": 0.5124307870864868, "learning_rate": 0.0005, "loss": 1.3900666236877441, "step": 23455 }, { "epoch": 1.8491131257390618, "grad_norm": 0.5157648324966431, "learning_rate": 0.0005, "loss": 1.2222349643707275, "step": 23456 }, { "epoch": 1.849191959006701, "grad_norm": 0.604264497756958, "learning_rate": 0.0005, "loss": 1.580331802368164, "step": 23457 }, { "epoch": 1.8492707922743397, "grad_norm": 0.4889174699783325, "learning_rate": 0.0005, "loss": 1.4570157527923584, "step": 23458 }, { "epoch": 1.8493496255419788, "grad_norm": 0.5728299021720886, "learning_rate": 0.0005, "loss": 1.398688554763794, "step": 23459 }, { "epoch": 1.8494284588096177, "grad_norm": 0.5472698211669922, "learning_rate": 0.0005, "loss": 1.359126329421997, "step": 23460 }, { "epoch": 1.8495072920772566, "grad_norm": 0.5810621380805969, "learning_rate": 0.0005, "loss": 1.2281837463378906, "step": 23461 }, { "epoch": 1.8495861253448955, "grad_norm": 0.46158766746520996, "learning_rate": 0.0005, "loss": 1.470973253250122, "step": 23462 }, { "epoch": 1.8496649586125344, "grad_norm": 0.5768253207206726, "learning_rate": 0.0005, "loss": 1.4105792045593262, "step": 23463 }, { "epoch": 1.8497437918801736, "grad_norm": 0.44262930750846863, "learning_rate": 0.0005, "loss": 1.428929090499878, "step": 23464 }, { "epoch": 1.8498226251478123, "grad_norm": 0.561576247215271, "learning_rate": 0.0005, "loss": 1.540401816368103, "step": 23465 }, { "epoch": 1.8499014584154514, "grad_norm": 0.41058608889579773, "learning_rate": 0.0005, "loss": 1.2217121124267578, "step": 23466 }, { "epoch": 1.8499802916830903, "grad_norm": 0.5361054539680481, "learning_rate": 0.0005, "loss": 1.3398799896240234, "step": 23467 }, { "epoch": 1.8500591249507292, "grad_norm": 0.4996017813682556, "learning_rate": 0.0005, "loss": 1.3702223300933838, "step": 23468 }, { "epoch": 1.8501379582183681, "grad_norm": 0.5700631737709045, "learning_rate": 0.0005, "loss": 1.501374363899231, "step": 23469 }, { "epoch": 1.850216791486007, "grad_norm": 0.47007694840431213, "learning_rate": 0.0005, "loss": 1.4482154846191406, "step": 23470 }, { "epoch": 1.8502956247536462, "grad_norm": 0.5802125334739685, "learning_rate": 0.0005, "loss": 1.159811019897461, "step": 23471 }, { "epoch": 1.8503744580212849, "grad_norm": 0.5273355841636658, "learning_rate": 0.0005, "loss": 1.3871192932128906, "step": 23472 }, { "epoch": 1.850453291288924, "grad_norm": 0.6133183836936951, "learning_rate": 0.0005, "loss": 1.3673392534255981, "step": 23473 }, { "epoch": 1.850532124556563, "grad_norm": 0.46395841240882874, "learning_rate": 0.0005, "loss": 1.5124491453170776, "step": 23474 }, { "epoch": 1.8506109578242018, "grad_norm": 0.5311673283576965, "learning_rate": 0.0005, "loss": 1.224217414855957, "step": 23475 }, { "epoch": 1.8506897910918407, "grad_norm": 0.5086119174957275, "learning_rate": 0.0005, "loss": 1.4233119487762451, "step": 23476 }, { "epoch": 1.8507686243594796, "grad_norm": 0.4416154623031616, "learning_rate": 0.0005, "loss": 1.2260160446166992, "step": 23477 }, { "epoch": 1.8508474576271188, "grad_norm": 0.5889816880226135, "learning_rate": 0.0005, "loss": 1.3527212142944336, "step": 23478 }, { "epoch": 1.8509262908947575, "grad_norm": 0.4987791180610657, "learning_rate": 0.0005, "loss": 1.547392725944519, "step": 23479 }, { "epoch": 1.8510051241623966, "grad_norm": 0.4498094916343689, "learning_rate": 0.0005, "loss": 1.441619873046875, "step": 23480 }, { "epoch": 1.8510839574300355, "grad_norm": 0.6479506492614746, "learning_rate": 0.0005, "loss": 1.2321460247039795, "step": 23481 }, { "epoch": 1.8511627906976744, "grad_norm": 0.47763147950172424, "learning_rate": 0.0005, "loss": 1.4493781328201294, "step": 23482 }, { "epoch": 1.8512416239653133, "grad_norm": 0.5322019457817078, "learning_rate": 0.0005, "loss": 1.675053596496582, "step": 23483 }, { "epoch": 1.8513204572329522, "grad_norm": 0.4630219340324402, "learning_rate": 0.0005, "loss": 1.147693157196045, "step": 23484 }, { "epoch": 1.8513992905005914, "grad_norm": 0.5030446648597717, "learning_rate": 0.0005, "loss": 1.4151880741119385, "step": 23485 }, { "epoch": 1.85147812376823, "grad_norm": 0.5354030132293701, "learning_rate": 0.0005, "loss": 1.3362538814544678, "step": 23486 }, { "epoch": 1.8515569570358692, "grad_norm": 0.7595809102058411, "learning_rate": 0.0005, "loss": 1.4324705600738525, "step": 23487 }, { "epoch": 1.8516357903035081, "grad_norm": 0.5056577324867249, "learning_rate": 0.0005, "loss": 1.5385171175003052, "step": 23488 }, { "epoch": 1.851714623571147, "grad_norm": 0.7105023264884949, "learning_rate": 0.0005, "loss": 1.5720036029815674, "step": 23489 }, { "epoch": 1.851793456838786, "grad_norm": 0.6131541728973389, "learning_rate": 0.0005, "loss": 1.3838770389556885, "step": 23490 }, { "epoch": 1.8518722901064248, "grad_norm": 0.5024108290672302, "learning_rate": 0.0005, "loss": 1.3425607681274414, "step": 23491 }, { "epoch": 1.851951123374064, "grad_norm": 0.5403223037719727, "learning_rate": 0.0005, "loss": 1.570399522781372, "step": 23492 }, { "epoch": 1.8520299566417027, "grad_norm": 0.5862281918525696, "learning_rate": 0.0005, "loss": 1.526357650756836, "step": 23493 }, { "epoch": 1.8521087899093418, "grad_norm": 0.5210744738578796, "learning_rate": 0.0005, "loss": 1.551325798034668, "step": 23494 }, { "epoch": 1.8521876231769807, "grad_norm": 0.4683040380477905, "learning_rate": 0.0005, "loss": 1.4226868152618408, "step": 23495 }, { "epoch": 1.8522664564446196, "grad_norm": 0.474931538105011, "learning_rate": 0.0005, "loss": 1.380878210067749, "step": 23496 }, { "epoch": 1.8523452897122585, "grad_norm": 0.7546175718307495, "learning_rate": 0.0005, "loss": 1.3407986164093018, "step": 23497 }, { "epoch": 1.8524241229798974, "grad_norm": 0.5243952870368958, "learning_rate": 0.0005, "loss": 1.5902342796325684, "step": 23498 }, { "epoch": 1.8525029562475366, "grad_norm": 0.4959186911582947, "learning_rate": 0.0005, "loss": 1.586332082748413, "step": 23499 }, { "epoch": 1.8525817895151753, "grad_norm": 0.5046550631523132, "learning_rate": 0.0005, "loss": 1.5007741451263428, "step": 23500 }, { "epoch": 1.8526606227828144, "grad_norm": 0.4787771999835968, "learning_rate": 0.0005, "loss": 1.312827229499817, "step": 23501 }, { "epoch": 1.8527394560504533, "grad_norm": 0.6479243636131287, "learning_rate": 0.0005, "loss": 1.3866113424301147, "step": 23502 }, { "epoch": 1.8528182893180922, "grad_norm": 0.47035446763038635, "learning_rate": 0.0005, "loss": 1.3932220935821533, "step": 23503 }, { "epoch": 1.8528971225857311, "grad_norm": 0.5994569063186646, "learning_rate": 0.0005, "loss": 1.5141900777816772, "step": 23504 }, { "epoch": 1.85297595585337, "grad_norm": 0.5001334547996521, "learning_rate": 0.0005, "loss": 1.3229517936706543, "step": 23505 }, { "epoch": 1.8530547891210092, "grad_norm": 0.677775502204895, "learning_rate": 0.0005, "loss": 1.5677719116210938, "step": 23506 }, { "epoch": 1.8531336223886479, "grad_norm": 0.4548695385456085, "learning_rate": 0.0005, "loss": 1.4146971702575684, "step": 23507 }, { "epoch": 1.853212455656287, "grad_norm": 0.4729333817958832, "learning_rate": 0.0005, "loss": 1.1829874515533447, "step": 23508 }, { "epoch": 1.853291288923926, "grad_norm": 0.42619866132736206, "learning_rate": 0.0005, "loss": 1.0899271965026855, "step": 23509 }, { "epoch": 1.8533701221915648, "grad_norm": 0.5361050963401794, "learning_rate": 0.0005, "loss": 1.3501582145690918, "step": 23510 }, { "epoch": 1.8534489554592037, "grad_norm": 0.7448636889457703, "learning_rate": 0.0005, "loss": 1.595721960067749, "step": 23511 }, { "epoch": 1.8535277887268427, "grad_norm": 0.5928338766098022, "learning_rate": 0.0005, "loss": 1.7060335874557495, "step": 23512 }, { "epoch": 1.8536066219944818, "grad_norm": 0.5223305821418762, "learning_rate": 0.0005, "loss": 1.3932298421859741, "step": 23513 }, { "epoch": 1.8536854552621205, "grad_norm": 0.5324954390525818, "learning_rate": 0.0005, "loss": 1.4590463638305664, "step": 23514 }, { "epoch": 1.8537642885297596, "grad_norm": 0.6058258414268494, "learning_rate": 0.0005, "loss": 1.6142463684082031, "step": 23515 }, { "epoch": 1.8538431217973985, "grad_norm": 0.5658043026924133, "learning_rate": 0.0005, "loss": 1.4087449312210083, "step": 23516 }, { "epoch": 1.8539219550650374, "grad_norm": 1.1682939529418945, "learning_rate": 0.0005, "loss": 1.5469255447387695, "step": 23517 }, { "epoch": 1.8540007883326763, "grad_norm": 0.6640550494194031, "learning_rate": 0.0005, "loss": 1.582470417022705, "step": 23518 }, { "epoch": 1.8540796216003153, "grad_norm": 0.5348238945007324, "learning_rate": 0.0005, "loss": 1.3888108730316162, "step": 23519 }, { "epoch": 1.8541584548679544, "grad_norm": 0.6033409833908081, "learning_rate": 0.0005, "loss": 1.626779556274414, "step": 23520 }, { "epoch": 1.854237288135593, "grad_norm": 0.6917160153388977, "learning_rate": 0.0005, "loss": 1.6172757148742676, "step": 23521 }, { "epoch": 1.8543161214032322, "grad_norm": 0.8983170986175537, "learning_rate": 0.0005, "loss": 1.4505326747894287, "step": 23522 }, { "epoch": 1.8543949546708711, "grad_norm": 0.5081727504730225, "learning_rate": 0.0005, "loss": 1.420950174331665, "step": 23523 }, { "epoch": 1.85447378793851, "grad_norm": 0.7239608764648438, "learning_rate": 0.0005, "loss": 1.5827237367630005, "step": 23524 }, { "epoch": 1.854552621206149, "grad_norm": 0.6263051629066467, "learning_rate": 0.0005, "loss": 1.4637815952301025, "step": 23525 }, { "epoch": 1.8546314544737879, "grad_norm": 0.8565559387207031, "learning_rate": 0.0005, "loss": 1.5730568170547485, "step": 23526 }, { "epoch": 1.854710287741427, "grad_norm": 0.7046420574188232, "learning_rate": 0.0005, "loss": 1.5802648067474365, "step": 23527 }, { "epoch": 1.8547891210090657, "grad_norm": 0.625119149684906, "learning_rate": 0.0005, "loss": 1.3361937999725342, "step": 23528 }, { "epoch": 1.8548679542767048, "grad_norm": 0.6279028058052063, "learning_rate": 0.0005, "loss": 1.4035015106201172, "step": 23529 }, { "epoch": 1.8549467875443437, "grad_norm": 0.5315653681755066, "learning_rate": 0.0005, "loss": 1.2408745288848877, "step": 23530 }, { "epoch": 1.8550256208119826, "grad_norm": 0.5567100048065186, "learning_rate": 0.0005, "loss": 1.2908046245574951, "step": 23531 }, { "epoch": 1.8551044540796215, "grad_norm": 0.7474879622459412, "learning_rate": 0.0005, "loss": 1.2012171745300293, "step": 23532 }, { "epoch": 1.8551832873472605, "grad_norm": 0.5938000082969666, "learning_rate": 0.0005, "loss": 1.2596170902252197, "step": 23533 }, { "epoch": 1.8552621206148996, "grad_norm": 0.5675271153450012, "learning_rate": 0.0005, "loss": 1.2674661874771118, "step": 23534 }, { "epoch": 1.8553409538825383, "grad_norm": 0.6684500575065613, "learning_rate": 0.0005, "loss": 1.536529541015625, "step": 23535 }, { "epoch": 1.8554197871501774, "grad_norm": 0.4516332447528839, "learning_rate": 0.0005, "loss": 1.2306301593780518, "step": 23536 }, { "epoch": 1.8554986204178163, "grad_norm": 0.5628244280815125, "learning_rate": 0.0005, "loss": 1.5671310424804688, "step": 23537 }, { "epoch": 1.8555774536854552, "grad_norm": 0.4584277868270874, "learning_rate": 0.0005, "loss": 1.2629196643829346, "step": 23538 }, { "epoch": 1.8556562869530944, "grad_norm": 0.7270463705062866, "learning_rate": 0.0005, "loss": 1.7088825702667236, "step": 23539 }, { "epoch": 1.855735120220733, "grad_norm": 0.4644773602485657, "learning_rate": 0.0005, "loss": 1.3085529804229736, "step": 23540 }, { "epoch": 1.8558139534883722, "grad_norm": 0.5517537593841553, "learning_rate": 0.0005, "loss": 1.2515413761138916, "step": 23541 }, { "epoch": 1.8558927867560109, "grad_norm": 0.6114245057106018, "learning_rate": 0.0005, "loss": 1.3738207817077637, "step": 23542 }, { "epoch": 1.85597162002365, "grad_norm": 0.5575141906738281, "learning_rate": 0.0005, "loss": 1.6177349090576172, "step": 23543 }, { "epoch": 1.856050453291289, "grad_norm": 0.6934220790863037, "learning_rate": 0.0005, "loss": 1.3549704551696777, "step": 23544 }, { "epoch": 1.8561292865589278, "grad_norm": 0.3897227346897125, "learning_rate": 0.0005, "loss": 1.2510638236999512, "step": 23545 }, { "epoch": 1.856208119826567, "grad_norm": 0.5524296164512634, "learning_rate": 0.0005, "loss": 1.3280212879180908, "step": 23546 }, { "epoch": 1.8562869530942057, "grad_norm": 0.590591311454773, "learning_rate": 0.0005, "loss": 1.4835293292999268, "step": 23547 }, { "epoch": 1.8563657863618448, "grad_norm": 0.4297630190849304, "learning_rate": 0.0005, "loss": 1.1604408025741577, "step": 23548 }, { "epoch": 1.8564446196294835, "grad_norm": 0.48802170157432556, "learning_rate": 0.0005, "loss": 1.2861328125, "step": 23549 }, { "epoch": 1.8565234528971226, "grad_norm": 0.4540104269981384, "learning_rate": 0.0005, "loss": 1.3881185054779053, "step": 23550 }, { "epoch": 1.8566022861647615, "grad_norm": 0.8774752616882324, "learning_rate": 0.0005, "loss": 1.6199376583099365, "step": 23551 }, { "epoch": 1.8566811194324004, "grad_norm": 0.5634396076202393, "learning_rate": 0.0005, "loss": 1.4443119764328003, "step": 23552 }, { "epoch": 1.8567599527000396, "grad_norm": 0.7053295969963074, "learning_rate": 0.0005, "loss": 1.4518409967422485, "step": 23553 }, { "epoch": 1.8568387859676783, "grad_norm": 0.5017399787902832, "learning_rate": 0.0005, "loss": 1.332909345626831, "step": 23554 }, { "epoch": 1.8569176192353174, "grad_norm": 0.5642102956771851, "learning_rate": 0.0005, "loss": 1.3840382099151611, "step": 23555 }, { "epoch": 1.856996452502956, "grad_norm": 0.8360024094581604, "learning_rate": 0.0005, "loss": 1.6280686855316162, "step": 23556 }, { "epoch": 1.8570752857705952, "grad_norm": 0.5640100240707397, "learning_rate": 0.0005, "loss": 1.376633882522583, "step": 23557 }, { "epoch": 1.8571541190382341, "grad_norm": 0.5971359014511108, "learning_rate": 0.0005, "loss": 1.6134059429168701, "step": 23558 }, { "epoch": 1.857232952305873, "grad_norm": 0.6779577136039734, "learning_rate": 0.0005, "loss": 1.578758716583252, "step": 23559 }, { "epoch": 1.8573117855735122, "grad_norm": 0.5230961441993713, "learning_rate": 0.0005, "loss": 1.3163408041000366, "step": 23560 }, { "epoch": 1.8573906188411509, "grad_norm": 0.45876631140708923, "learning_rate": 0.0005, "loss": 1.343875765800476, "step": 23561 }, { "epoch": 1.85746945210879, "grad_norm": 0.49023517966270447, "learning_rate": 0.0005, "loss": 1.3516179323196411, "step": 23562 }, { "epoch": 1.8575482853764287, "grad_norm": 0.5069427490234375, "learning_rate": 0.0005, "loss": 1.519100308418274, "step": 23563 }, { "epoch": 1.8576271186440678, "grad_norm": 0.7864930033683777, "learning_rate": 0.0005, "loss": 1.6463334560394287, "step": 23564 }, { "epoch": 1.8577059519117067, "grad_norm": 0.6604558229446411, "learning_rate": 0.0005, "loss": 1.4350558519363403, "step": 23565 }, { "epoch": 1.8577847851793456, "grad_norm": 0.502998948097229, "learning_rate": 0.0005, "loss": 1.4844028949737549, "step": 23566 }, { "epoch": 1.8578636184469848, "grad_norm": 0.5693939328193665, "learning_rate": 0.0005, "loss": 1.5673905611038208, "step": 23567 }, { "epoch": 1.8579424517146235, "grad_norm": 0.4520007073879242, "learning_rate": 0.0005, "loss": 1.286769151687622, "step": 23568 }, { "epoch": 1.8580212849822626, "grad_norm": 0.6415697932243347, "learning_rate": 0.0005, "loss": 1.6448736190795898, "step": 23569 }, { "epoch": 1.8581001182499015, "grad_norm": 0.49464666843414307, "learning_rate": 0.0005, "loss": 1.4456498622894287, "step": 23570 }, { "epoch": 1.8581789515175404, "grad_norm": 0.5116291642189026, "learning_rate": 0.0005, "loss": 1.3335262537002563, "step": 23571 }, { "epoch": 1.8582577847851793, "grad_norm": 0.6129238605499268, "learning_rate": 0.0005, "loss": 1.461801290512085, "step": 23572 }, { "epoch": 1.8583366180528182, "grad_norm": 0.6549037098884583, "learning_rate": 0.0005, "loss": 1.3336989879608154, "step": 23573 }, { "epoch": 1.8584154513204574, "grad_norm": 0.4574674367904663, "learning_rate": 0.0005, "loss": 1.2416865825653076, "step": 23574 }, { "epoch": 1.858494284588096, "grad_norm": 0.46626725792884827, "learning_rate": 0.0005, "loss": 1.368304967880249, "step": 23575 }, { "epoch": 1.8585731178557352, "grad_norm": 0.5129485726356506, "learning_rate": 0.0005, "loss": 1.285369873046875, "step": 23576 }, { "epoch": 1.8586519511233741, "grad_norm": 0.589818000793457, "learning_rate": 0.0005, "loss": 1.4748640060424805, "step": 23577 }, { "epoch": 1.858730784391013, "grad_norm": 0.5561649203300476, "learning_rate": 0.0005, "loss": 1.385319709777832, "step": 23578 }, { "epoch": 1.858809617658652, "grad_norm": 0.5019283294677734, "learning_rate": 0.0005, "loss": 1.5240103006362915, "step": 23579 }, { "epoch": 1.8588884509262908, "grad_norm": 0.6473442912101746, "learning_rate": 0.0005, "loss": 1.62517511844635, "step": 23580 }, { "epoch": 1.85896728419393, "grad_norm": 0.6070648431777954, "learning_rate": 0.0005, "loss": 1.6119771003723145, "step": 23581 }, { "epoch": 1.8590461174615687, "grad_norm": 0.6632732152938843, "learning_rate": 0.0005, "loss": 1.421737551689148, "step": 23582 }, { "epoch": 1.8591249507292078, "grad_norm": 0.5766666531562805, "learning_rate": 0.0005, "loss": 1.47239351272583, "step": 23583 }, { "epoch": 1.8592037839968467, "grad_norm": 0.4222869575023651, "learning_rate": 0.0005, "loss": 1.208919882774353, "step": 23584 }, { "epoch": 1.8592826172644856, "grad_norm": 0.6048305034637451, "learning_rate": 0.0005, "loss": 1.5219066143035889, "step": 23585 }, { "epoch": 1.8593614505321245, "grad_norm": 0.6318528056144714, "learning_rate": 0.0005, "loss": 1.5258557796478271, "step": 23586 }, { "epoch": 1.8594402837997634, "grad_norm": 0.5392831563949585, "learning_rate": 0.0005, "loss": 1.3331183195114136, "step": 23587 }, { "epoch": 1.8595191170674026, "grad_norm": 0.5291194915771484, "learning_rate": 0.0005, "loss": 1.4053888320922852, "step": 23588 }, { "epoch": 1.8595979503350413, "grad_norm": 0.5010339617729187, "learning_rate": 0.0005, "loss": 1.3671770095825195, "step": 23589 }, { "epoch": 1.8596767836026804, "grad_norm": 0.518571674823761, "learning_rate": 0.0005, "loss": 1.5326968431472778, "step": 23590 }, { "epoch": 1.8597556168703193, "grad_norm": 0.46577131748199463, "learning_rate": 0.0005, "loss": 1.2321934700012207, "step": 23591 }, { "epoch": 1.8598344501379582, "grad_norm": 0.5093267560005188, "learning_rate": 0.0005, "loss": 1.2054654359817505, "step": 23592 }, { "epoch": 1.8599132834055971, "grad_norm": 0.43440860509872437, "learning_rate": 0.0005, "loss": 1.3167574405670166, "step": 23593 }, { "epoch": 1.859992116673236, "grad_norm": 0.39686816930770874, "learning_rate": 0.0005, "loss": 1.337632417678833, "step": 23594 }, { "epoch": 1.8600709499408752, "grad_norm": 0.5500378012657166, "learning_rate": 0.0005, "loss": 1.3306258916854858, "step": 23595 }, { "epoch": 1.8601497832085139, "grad_norm": 0.7054713368415833, "learning_rate": 0.0005, "loss": 1.4890509843826294, "step": 23596 }, { "epoch": 1.860228616476153, "grad_norm": 0.42585524916648865, "learning_rate": 0.0005, "loss": 1.1895780563354492, "step": 23597 }, { "epoch": 1.860307449743792, "grad_norm": 0.5297362208366394, "learning_rate": 0.0005, "loss": 1.3097715377807617, "step": 23598 }, { "epoch": 1.8603862830114308, "grad_norm": 0.5664333701133728, "learning_rate": 0.0005, "loss": 1.4132344722747803, "step": 23599 }, { "epoch": 1.8604651162790697, "grad_norm": 0.4432826042175293, "learning_rate": 0.0005, "loss": 1.3926270008087158, "step": 23600 }, { "epoch": 1.8605439495467087, "grad_norm": 0.4877128303050995, "learning_rate": 0.0005, "loss": 1.4219152927398682, "step": 23601 }, { "epoch": 1.8606227828143478, "grad_norm": 0.4193555414676666, "learning_rate": 0.0005, "loss": 1.2339768409729004, "step": 23602 }, { "epoch": 1.8607016160819865, "grad_norm": 0.6075363755226135, "learning_rate": 0.0005, "loss": 1.6089707612991333, "step": 23603 }, { "epoch": 1.8607804493496256, "grad_norm": 0.508412778377533, "learning_rate": 0.0005, "loss": 1.415505051612854, "step": 23604 }, { "epoch": 1.8608592826172645, "grad_norm": 0.4380953311920166, "learning_rate": 0.0005, "loss": 1.255100965499878, "step": 23605 }, { "epoch": 1.8609381158849034, "grad_norm": 0.5024629831314087, "learning_rate": 0.0005, "loss": 1.3637099266052246, "step": 23606 }, { "epoch": 1.8610169491525423, "grad_norm": 0.6920555233955383, "learning_rate": 0.0005, "loss": 1.6780481338500977, "step": 23607 }, { "epoch": 1.8610957824201813, "grad_norm": 0.5447340607643127, "learning_rate": 0.0005, "loss": 1.404611349105835, "step": 23608 }, { "epoch": 1.8611746156878204, "grad_norm": 0.41327646374702454, "learning_rate": 0.0005, "loss": 1.2765120267868042, "step": 23609 }, { "epoch": 1.861253448955459, "grad_norm": 0.45467400550842285, "learning_rate": 0.0005, "loss": 1.1473159790039062, "step": 23610 }, { "epoch": 1.8613322822230982, "grad_norm": 0.4886544942855835, "learning_rate": 0.0005, "loss": 1.190057396888733, "step": 23611 }, { "epoch": 1.8614111154907371, "grad_norm": 0.5446645617485046, "learning_rate": 0.0005, "loss": 1.6463508605957031, "step": 23612 }, { "epoch": 1.861489948758376, "grad_norm": 0.584043025970459, "learning_rate": 0.0005, "loss": 1.417604684829712, "step": 23613 }, { "epoch": 1.861568782026015, "grad_norm": 0.4643111228942871, "learning_rate": 0.0005, "loss": 1.252030611038208, "step": 23614 }, { "epoch": 1.8616476152936539, "grad_norm": 0.6163658499717712, "learning_rate": 0.0005, "loss": 1.66792893409729, "step": 23615 }, { "epoch": 1.861726448561293, "grad_norm": 0.6171296834945679, "learning_rate": 0.0005, "loss": 1.4364873170852661, "step": 23616 }, { "epoch": 1.8618052818289317, "grad_norm": 0.5805437564849854, "learning_rate": 0.0005, "loss": 1.629557490348816, "step": 23617 }, { "epoch": 1.8618841150965708, "grad_norm": 0.5798267126083374, "learning_rate": 0.0005, "loss": 1.5213775634765625, "step": 23618 }, { "epoch": 1.8619629483642097, "grad_norm": 0.5160387754440308, "learning_rate": 0.0005, "loss": 1.37752103805542, "step": 23619 }, { "epoch": 1.8620417816318486, "grad_norm": 0.4913024604320526, "learning_rate": 0.0005, "loss": 1.332474708557129, "step": 23620 }, { "epoch": 1.8621206148994875, "grad_norm": 0.4950511157512665, "learning_rate": 0.0005, "loss": 1.5423721075057983, "step": 23621 }, { "epoch": 1.8621994481671265, "grad_norm": 0.49281004071235657, "learning_rate": 0.0005, "loss": 1.5437875986099243, "step": 23622 }, { "epoch": 1.8622782814347656, "grad_norm": 0.5656769871711731, "learning_rate": 0.0005, "loss": 1.4404239654541016, "step": 23623 }, { "epoch": 1.8623571147024043, "grad_norm": 0.5488854646682739, "learning_rate": 0.0005, "loss": 1.5035254955291748, "step": 23624 }, { "epoch": 1.8624359479700434, "grad_norm": 0.5387671589851379, "learning_rate": 0.0005, "loss": 1.3020728826522827, "step": 23625 }, { "epoch": 1.8625147812376823, "grad_norm": 0.6272432804107666, "learning_rate": 0.0005, "loss": 1.5105082988739014, "step": 23626 }, { "epoch": 1.8625936145053212, "grad_norm": 0.5291734337806702, "learning_rate": 0.0005, "loss": 1.320277214050293, "step": 23627 }, { "epoch": 1.8626724477729601, "grad_norm": 0.5343986749649048, "learning_rate": 0.0005, "loss": 1.5017626285552979, "step": 23628 }, { "epoch": 1.862751281040599, "grad_norm": 0.5049145817756653, "learning_rate": 0.0005, "loss": 1.413816213607788, "step": 23629 }, { "epoch": 1.8628301143082382, "grad_norm": 0.5691815614700317, "learning_rate": 0.0005, "loss": 1.220943808555603, "step": 23630 }, { "epoch": 1.8629089475758769, "grad_norm": 0.4915756583213806, "learning_rate": 0.0005, "loss": 1.361169457435608, "step": 23631 }, { "epoch": 1.862987780843516, "grad_norm": 0.5558705925941467, "learning_rate": 0.0005, "loss": 1.5917541980743408, "step": 23632 }, { "epoch": 1.863066614111155, "grad_norm": 0.6075499653816223, "learning_rate": 0.0005, "loss": 1.3985854387283325, "step": 23633 }, { "epoch": 1.8631454473787938, "grad_norm": 0.5223944187164307, "learning_rate": 0.0005, "loss": 1.4386894702911377, "step": 23634 }, { "epoch": 1.8632242806464328, "grad_norm": 0.46690863370895386, "learning_rate": 0.0005, "loss": 1.567162275314331, "step": 23635 }, { "epoch": 1.8633031139140717, "grad_norm": 0.5731343030929565, "learning_rate": 0.0005, "loss": 1.577233076095581, "step": 23636 }, { "epoch": 1.8633819471817108, "grad_norm": 0.5885968208312988, "learning_rate": 0.0005, "loss": 1.4393165111541748, "step": 23637 }, { "epoch": 1.8634607804493495, "grad_norm": 0.49429646134376526, "learning_rate": 0.0005, "loss": 1.4721946716308594, "step": 23638 }, { "epoch": 1.8635396137169886, "grad_norm": 0.8831396698951721, "learning_rate": 0.0005, "loss": 1.510696530342102, "step": 23639 }, { "epoch": 1.8636184469846275, "grad_norm": 0.5833253860473633, "learning_rate": 0.0005, "loss": 1.4138879776000977, "step": 23640 }, { "epoch": 1.8636972802522664, "grad_norm": 0.5191771388053894, "learning_rate": 0.0005, "loss": 1.506822109222412, "step": 23641 }, { "epoch": 1.8637761135199054, "grad_norm": 0.5180182456970215, "learning_rate": 0.0005, "loss": 1.33829927444458, "step": 23642 }, { "epoch": 1.8638549467875443, "grad_norm": 0.5552522540092468, "learning_rate": 0.0005, "loss": 1.2884984016418457, "step": 23643 }, { "epoch": 1.8639337800551834, "grad_norm": 0.7142946124076843, "learning_rate": 0.0005, "loss": 1.3729243278503418, "step": 23644 }, { "epoch": 1.864012613322822, "grad_norm": 0.6567128300666809, "learning_rate": 0.0005, "loss": 1.5426379442214966, "step": 23645 }, { "epoch": 1.8640914465904612, "grad_norm": 0.5486409068107605, "learning_rate": 0.0005, "loss": 1.4363747835159302, "step": 23646 }, { "epoch": 1.8641702798581001, "grad_norm": 0.561511218547821, "learning_rate": 0.0005, "loss": 1.4673495292663574, "step": 23647 }, { "epoch": 1.864249113125739, "grad_norm": 0.5058190822601318, "learning_rate": 0.0005, "loss": 1.2800488471984863, "step": 23648 }, { "epoch": 1.864327946393378, "grad_norm": 0.45868685841560364, "learning_rate": 0.0005, "loss": 1.4371249675750732, "step": 23649 }, { "epoch": 1.8644067796610169, "grad_norm": 0.5335931181907654, "learning_rate": 0.0005, "loss": 1.425839900970459, "step": 23650 }, { "epoch": 1.864485612928656, "grad_norm": 0.6899412274360657, "learning_rate": 0.0005, "loss": 1.4179589748382568, "step": 23651 }, { "epoch": 1.8645644461962947, "grad_norm": 0.4893094599246979, "learning_rate": 0.0005, "loss": 1.5392570495605469, "step": 23652 }, { "epoch": 1.8646432794639338, "grad_norm": 0.37333953380584717, "learning_rate": 0.0005, "loss": 1.1530102491378784, "step": 23653 }, { "epoch": 1.8647221127315727, "grad_norm": 0.5589725971221924, "learning_rate": 0.0005, "loss": 1.4354603290557861, "step": 23654 }, { "epoch": 1.8648009459992116, "grad_norm": 0.5746198296546936, "learning_rate": 0.0005, "loss": 1.4338104724884033, "step": 23655 }, { "epoch": 1.8648797792668506, "grad_norm": 0.6230252981185913, "learning_rate": 0.0005, "loss": 1.2532129287719727, "step": 23656 }, { "epoch": 1.8649586125344895, "grad_norm": 0.6198953986167908, "learning_rate": 0.0005, "loss": 1.4564766883850098, "step": 23657 }, { "epoch": 1.8650374458021286, "grad_norm": 0.497220903635025, "learning_rate": 0.0005, "loss": 1.398306965827942, "step": 23658 }, { "epoch": 1.8651162790697673, "grad_norm": 0.4840158522129059, "learning_rate": 0.0005, "loss": 1.5516918897628784, "step": 23659 }, { "epoch": 1.8651951123374064, "grad_norm": 0.596148669719696, "learning_rate": 0.0005, "loss": 1.478136658668518, "step": 23660 }, { "epoch": 1.8652739456050453, "grad_norm": 0.5315226316452026, "learning_rate": 0.0005, "loss": 1.4830952882766724, "step": 23661 }, { "epoch": 1.8653527788726842, "grad_norm": 0.5169782638549805, "learning_rate": 0.0005, "loss": 1.4809991121292114, "step": 23662 }, { "epoch": 1.8654316121403234, "grad_norm": 0.5484287738800049, "learning_rate": 0.0005, "loss": 1.2382957935333252, "step": 23663 }, { "epoch": 1.865510445407962, "grad_norm": 0.478012353181839, "learning_rate": 0.0005, "loss": 1.4548836946487427, "step": 23664 }, { "epoch": 1.8655892786756012, "grad_norm": 0.5394073128700256, "learning_rate": 0.0005, "loss": 1.3490233421325684, "step": 23665 }, { "epoch": 1.86566811194324, "grad_norm": 0.4809238612651825, "learning_rate": 0.0005, "loss": 1.4939255714416504, "step": 23666 }, { "epoch": 1.865746945210879, "grad_norm": 0.5358151197433472, "learning_rate": 0.0005, "loss": 1.527946949005127, "step": 23667 }, { "epoch": 1.865825778478518, "grad_norm": 0.6704825162887573, "learning_rate": 0.0005, "loss": 1.8789174556732178, "step": 23668 }, { "epoch": 1.8659046117461568, "grad_norm": 1.390505075454712, "learning_rate": 0.0005, "loss": 1.5931346416473389, "step": 23669 }, { "epoch": 1.865983445013796, "grad_norm": 0.7029448747634888, "learning_rate": 0.0005, "loss": 1.639782428741455, "step": 23670 }, { "epoch": 1.8660622782814347, "grad_norm": 0.3822570741176605, "learning_rate": 0.0005, "loss": 1.1947476863861084, "step": 23671 }, { "epoch": 1.8661411115490738, "grad_norm": 0.5508467555046082, "learning_rate": 0.0005, "loss": 1.6663453578948975, "step": 23672 }, { "epoch": 1.8662199448167125, "grad_norm": 0.4481557607650757, "learning_rate": 0.0005, "loss": 1.411515712738037, "step": 23673 }, { "epoch": 1.8662987780843516, "grad_norm": 0.8471829891204834, "learning_rate": 0.0005, "loss": 1.2747397422790527, "step": 23674 }, { "epoch": 1.8663776113519905, "grad_norm": 0.7291288375854492, "learning_rate": 0.0005, "loss": 1.359855055809021, "step": 23675 }, { "epoch": 1.8664564446196295, "grad_norm": 0.5618938207626343, "learning_rate": 0.0005, "loss": 1.5059272050857544, "step": 23676 }, { "epoch": 1.8665352778872686, "grad_norm": 0.5314373970031738, "learning_rate": 0.0005, "loss": 1.4666874408721924, "step": 23677 }, { "epoch": 1.8666141111549073, "grad_norm": 0.5178978443145752, "learning_rate": 0.0005, "loss": 1.3424687385559082, "step": 23678 }, { "epoch": 1.8666929444225464, "grad_norm": 0.502097487449646, "learning_rate": 0.0005, "loss": 1.5755972862243652, "step": 23679 }, { "epoch": 1.866771777690185, "grad_norm": 0.528412401676178, "learning_rate": 0.0005, "loss": 1.2451549768447876, "step": 23680 }, { "epoch": 1.8668506109578242, "grad_norm": 0.5202513933181763, "learning_rate": 0.0005, "loss": 1.2491763830184937, "step": 23681 }, { "epoch": 1.8669294442254631, "grad_norm": 0.5551318526268005, "learning_rate": 0.0005, "loss": 1.388763427734375, "step": 23682 }, { "epoch": 1.867008277493102, "grad_norm": 0.4499835968017578, "learning_rate": 0.0005, "loss": 1.530656099319458, "step": 23683 }, { "epoch": 1.8670871107607412, "grad_norm": 0.47773444652557373, "learning_rate": 0.0005, "loss": 1.426835060119629, "step": 23684 }, { "epoch": 1.8671659440283799, "grad_norm": 0.5153303742408752, "learning_rate": 0.0005, "loss": 1.610838532447815, "step": 23685 }, { "epoch": 1.867244777296019, "grad_norm": 0.8046932816505432, "learning_rate": 0.0005, "loss": 1.3811862468719482, "step": 23686 }, { "epoch": 1.8673236105636577, "grad_norm": 0.5761110782623291, "learning_rate": 0.0005, "loss": 1.5749751329421997, "step": 23687 }, { "epoch": 1.8674024438312968, "grad_norm": 0.49754759669303894, "learning_rate": 0.0005, "loss": 1.3318853378295898, "step": 23688 }, { "epoch": 1.8674812770989357, "grad_norm": 0.7789523005485535, "learning_rate": 0.0005, "loss": 1.5820552110671997, "step": 23689 }, { "epoch": 1.8675601103665747, "grad_norm": 0.6033883094787598, "learning_rate": 0.0005, "loss": 1.5299805402755737, "step": 23690 }, { "epoch": 1.8676389436342138, "grad_norm": 0.6480230689048767, "learning_rate": 0.0005, "loss": 1.4267247915267944, "step": 23691 }, { "epoch": 1.8677177769018525, "grad_norm": 0.4635336101055145, "learning_rate": 0.0005, "loss": 1.2567439079284668, "step": 23692 }, { "epoch": 1.8677966101694916, "grad_norm": 0.43646690249443054, "learning_rate": 0.0005, "loss": 1.173506259918213, "step": 23693 }, { "epoch": 1.8678754434371305, "grad_norm": 0.5385292172431946, "learning_rate": 0.0005, "loss": 1.4247949123382568, "step": 23694 }, { "epoch": 1.8679542767047694, "grad_norm": 0.3981388807296753, "learning_rate": 0.0005, "loss": 1.1554566621780396, "step": 23695 }, { "epoch": 1.8680331099724083, "grad_norm": 0.43006739020347595, "learning_rate": 0.0005, "loss": 1.395981788635254, "step": 23696 }, { "epoch": 1.8681119432400473, "grad_norm": 0.6690655946731567, "learning_rate": 0.0005, "loss": 1.6148786544799805, "step": 23697 }, { "epoch": 1.8681907765076864, "grad_norm": 0.47278714179992676, "learning_rate": 0.0005, "loss": 1.3594639301300049, "step": 23698 }, { "epoch": 1.868269609775325, "grad_norm": 0.6245909929275513, "learning_rate": 0.0005, "loss": 1.4584105014801025, "step": 23699 }, { "epoch": 1.8683484430429642, "grad_norm": 0.4186045229434967, "learning_rate": 0.0005, "loss": 1.1758145093917847, "step": 23700 }, { "epoch": 1.8684272763106031, "grad_norm": 0.4167667329311371, "learning_rate": 0.0005, "loss": 1.4001522064208984, "step": 23701 }, { "epoch": 1.868506109578242, "grad_norm": 0.49237850308418274, "learning_rate": 0.0005, "loss": 1.4950923919677734, "step": 23702 }, { "epoch": 1.868584942845881, "grad_norm": 0.5474368333816528, "learning_rate": 0.0005, "loss": 1.3533885478973389, "step": 23703 }, { "epoch": 1.8686637761135199, "grad_norm": 0.46595022082328796, "learning_rate": 0.0005, "loss": 1.3949012756347656, "step": 23704 }, { "epoch": 1.868742609381159, "grad_norm": 0.5458409786224365, "learning_rate": 0.0005, "loss": 1.3765466213226318, "step": 23705 }, { "epoch": 1.8688214426487977, "grad_norm": 0.43082138895988464, "learning_rate": 0.0005, "loss": 1.1430160999298096, "step": 23706 }, { "epoch": 1.8689002759164368, "grad_norm": 1.36240816116333, "learning_rate": 0.0005, "loss": 1.5977002382278442, "step": 23707 }, { "epoch": 1.8689791091840757, "grad_norm": 0.4785881042480469, "learning_rate": 0.0005, "loss": 1.473578691482544, "step": 23708 }, { "epoch": 1.8690579424517146, "grad_norm": 0.5602726936340332, "learning_rate": 0.0005, "loss": 1.355406403541565, "step": 23709 }, { "epoch": 1.8691367757193535, "grad_norm": 0.5697775483131409, "learning_rate": 0.0005, "loss": 1.6465328931808472, "step": 23710 }, { "epoch": 1.8692156089869925, "grad_norm": 0.6194960474967957, "learning_rate": 0.0005, "loss": 1.5917575359344482, "step": 23711 }, { "epoch": 1.8692944422546316, "grad_norm": 0.47366949915885925, "learning_rate": 0.0005, "loss": 1.4338328838348389, "step": 23712 }, { "epoch": 1.8693732755222703, "grad_norm": 0.511394202709198, "learning_rate": 0.0005, "loss": 1.4630334377288818, "step": 23713 }, { "epoch": 1.8694521087899094, "grad_norm": 0.5305991768836975, "learning_rate": 0.0005, "loss": 1.3706552982330322, "step": 23714 }, { "epoch": 1.8695309420575483, "grad_norm": 0.6244462132453918, "learning_rate": 0.0005, "loss": 1.5423591136932373, "step": 23715 }, { "epoch": 1.8696097753251872, "grad_norm": 0.4850066602230072, "learning_rate": 0.0005, "loss": 1.4266431331634521, "step": 23716 }, { "epoch": 1.8696886085928262, "grad_norm": 0.6055346131324768, "learning_rate": 0.0005, "loss": 1.4944443702697754, "step": 23717 }, { "epoch": 1.869767441860465, "grad_norm": 0.59834223985672, "learning_rate": 0.0005, "loss": 1.578970193862915, "step": 23718 }, { "epoch": 1.8698462751281042, "grad_norm": 0.7466036677360535, "learning_rate": 0.0005, "loss": 1.424818515777588, "step": 23719 }, { "epoch": 1.8699251083957429, "grad_norm": 0.8782433867454529, "learning_rate": 0.0005, "loss": 1.413479208946228, "step": 23720 }, { "epoch": 1.870003941663382, "grad_norm": 0.5389399528503418, "learning_rate": 0.0005, "loss": 1.3517217636108398, "step": 23721 }, { "epoch": 1.870082774931021, "grad_norm": 0.5898737907409668, "learning_rate": 0.0005, "loss": 1.4372591972351074, "step": 23722 }, { "epoch": 1.8701616081986598, "grad_norm": 0.5984390377998352, "learning_rate": 0.0005, "loss": 1.4722763299942017, "step": 23723 }, { "epoch": 1.8702404414662988, "grad_norm": 0.5810006856918335, "learning_rate": 0.0005, "loss": 1.3685678243637085, "step": 23724 }, { "epoch": 1.8703192747339377, "grad_norm": 0.527762770652771, "learning_rate": 0.0005, "loss": 1.28109872341156, "step": 23725 }, { "epoch": 1.8703981080015768, "grad_norm": 0.4263181984424591, "learning_rate": 0.0005, "loss": 1.1493961811065674, "step": 23726 }, { "epoch": 1.8704769412692155, "grad_norm": 0.5621269941329956, "learning_rate": 0.0005, "loss": 1.6064029932022095, "step": 23727 }, { "epoch": 1.8705557745368546, "grad_norm": 0.660320520401001, "learning_rate": 0.0005, "loss": 1.5461089611053467, "step": 23728 }, { "epoch": 1.8706346078044935, "grad_norm": 1.1579952239990234, "learning_rate": 0.0005, "loss": 1.5493793487548828, "step": 23729 }, { "epoch": 1.8707134410721324, "grad_norm": 0.5667781829833984, "learning_rate": 0.0005, "loss": 1.4547972679138184, "step": 23730 }, { "epoch": 1.8707922743397714, "grad_norm": 0.617824375629425, "learning_rate": 0.0005, "loss": 1.3078033924102783, "step": 23731 }, { "epoch": 1.8708711076074103, "grad_norm": 0.6992983222007751, "learning_rate": 0.0005, "loss": 1.3618303537368774, "step": 23732 }, { "epoch": 1.8709499408750494, "grad_norm": 0.5176411867141724, "learning_rate": 0.0005, "loss": 1.3722972869873047, "step": 23733 }, { "epoch": 1.871028774142688, "grad_norm": 0.5197322964668274, "learning_rate": 0.0005, "loss": 1.4116040468215942, "step": 23734 }, { "epoch": 1.8711076074103272, "grad_norm": 0.6462016105651855, "learning_rate": 0.0005, "loss": 1.517533302307129, "step": 23735 }, { "epoch": 1.8711864406779661, "grad_norm": 0.5111731886863708, "learning_rate": 0.0005, "loss": 1.3679380416870117, "step": 23736 }, { "epoch": 1.871265273945605, "grad_norm": 0.8553335070610046, "learning_rate": 0.0005, "loss": 1.463789939880371, "step": 23737 }, { "epoch": 1.871344107213244, "grad_norm": 0.5812970399856567, "learning_rate": 0.0005, "loss": 1.5671638250350952, "step": 23738 }, { "epoch": 1.8714229404808829, "grad_norm": 0.6246856451034546, "learning_rate": 0.0005, "loss": 1.1097214221954346, "step": 23739 }, { "epoch": 1.871501773748522, "grad_norm": 0.6035729050636292, "learning_rate": 0.0005, "loss": 1.543452501296997, "step": 23740 }, { "epoch": 1.8715806070161607, "grad_norm": 0.5672152638435364, "learning_rate": 0.0005, "loss": 1.479752779006958, "step": 23741 }, { "epoch": 1.8716594402837998, "grad_norm": 0.5869781970977783, "learning_rate": 0.0005, "loss": 1.4914600849151611, "step": 23742 }, { "epoch": 1.8717382735514387, "grad_norm": 0.5725894570350647, "learning_rate": 0.0005, "loss": 1.4597892761230469, "step": 23743 }, { "epoch": 1.8718171068190776, "grad_norm": 0.6070759296417236, "learning_rate": 0.0005, "loss": 1.0464696884155273, "step": 23744 }, { "epoch": 1.8718959400867166, "grad_norm": 0.4907833933830261, "learning_rate": 0.0005, "loss": 1.4733946323394775, "step": 23745 }, { "epoch": 1.8719747733543555, "grad_norm": 0.4892760217189789, "learning_rate": 0.0005, "loss": 1.3422669172286987, "step": 23746 }, { "epoch": 1.8720536066219946, "grad_norm": 0.5940787196159363, "learning_rate": 0.0005, "loss": 1.4561278820037842, "step": 23747 }, { "epoch": 1.8721324398896333, "grad_norm": 0.5689781904220581, "learning_rate": 0.0005, "loss": 1.243351697921753, "step": 23748 }, { "epoch": 1.8722112731572724, "grad_norm": 0.6485065221786499, "learning_rate": 0.0005, "loss": 1.581129789352417, "step": 23749 }, { "epoch": 1.8722901064249113, "grad_norm": 0.5572364926338196, "learning_rate": 0.0005, "loss": 1.3127599954605103, "step": 23750 }, { "epoch": 1.8723689396925502, "grad_norm": 0.6406884789466858, "learning_rate": 0.0005, "loss": 1.5032331943511963, "step": 23751 }, { "epoch": 1.8724477729601892, "grad_norm": 0.8234274387359619, "learning_rate": 0.0005, "loss": 1.5720062255859375, "step": 23752 }, { "epoch": 1.872526606227828, "grad_norm": 0.47060608863830566, "learning_rate": 0.0005, "loss": 1.3100444078445435, "step": 23753 }, { "epoch": 1.8726054394954672, "grad_norm": 0.683551549911499, "learning_rate": 0.0005, "loss": 1.8001635074615479, "step": 23754 }, { "epoch": 1.872684272763106, "grad_norm": 0.4768396019935608, "learning_rate": 0.0005, "loss": 1.3913334608078003, "step": 23755 }, { "epoch": 1.872763106030745, "grad_norm": 0.6294510960578918, "learning_rate": 0.0005, "loss": 1.4448366165161133, "step": 23756 }, { "epoch": 1.872841939298384, "grad_norm": 0.603285014629364, "learning_rate": 0.0005, "loss": 1.4916982650756836, "step": 23757 }, { "epoch": 1.8729207725660229, "grad_norm": 0.4895891547203064, "learning_rate": 0.0005, "loss": 1.589585781097412, "step": 23758 }, { "epoch": 1.8729996058336618, "grad_norm": 0.5696115493774414, "learning_rate": 0.0005, "loss": 1.189638614654541, "step": 23759 }, { "epoch": 1.8730784391013007, "grad_norm": 0.6961236000061035, "learning_rate": 0.0005, "loss": 1.404139757156372, "step": 23760 }, { "epoch": 1.8731572723689398, "grad_norm": 0.5338184237480164, "learning_rate": 0.0005, "loss": 1.4714808464050293, "step": 23761 }, { "epoch": 1.8732361056365785, "grad_norm": 0.5075303316116333, "learning_rate": 0.0005, "loss": 1.5881240367889404, "step": 23762 }, { "epoch": 1.8733149389042176, "grad_norm": 0.612348198890686, "learning_rate": 0.0005, "loss": 1.464938998222351, "step": 23763 }, { "epoch": 1.8733937721718565, "grad_norm": 1.1497215032577515, "learning_rate": 0.0005, "loss": 1.4171639680862427, "step": 23764 }, { "epoch": 1.8734726054394955, "grad_norm": 0.5003660917282104, "learning_rate": 0.0005, "loss": 1.1994603872299194, "step": 23765 }, { "epoch": 1.8735514387071344, "grad_norm": 0.5008100867271423, "learning_rate": 0.0005, "loss": 1.2087513208389282, "step": 23766 }, { "epoch": 1.8736302719747733, "grad_norm": 0.5468364357948303, "learning_rate": 0.0005, "loss": 1.60171639919281, "step": 23767 }, { "epoch": 1.8737091052424124, "grad_norm": 0.5173486471176147, "learning_rate": 0.0005, "loss": 1.5820332765579224, "step": 23768 }, { "epoch": 1.873787938510051, "grad_norm": 0.6084025502204895, "learning_rate": 0.0005, "loss": 1.4193581342697144, "step": 23769 }, { "epoch": 1.8738667717776902, "grad_norm": 0.537355899810791, "learning_rate": 0.0005, "loss": 1.3102710247039795, "step": 23770 }, { "epoch": 1.8739456050453291, "grad_norm": 0.5780490040779114, "learning_rate": 0.0005, "loss": 1.414402723312378, "step": 23771 }, { "epoch": 1.874024438312968, "grad_norm": 0.5564703345298767, "learning_rate": 0.0005, "loss": 1.5093128681182861, "step": 23772 }, { "epoch": 1.874103271580607, "grad_norm": 0.4780953824520111, "learning_rate": 0.0005, "loss": 1.383799433708191, "step": 23773 }, { "epoch": 1.8741821048482459, "grad_norm": 0.5568417906761169, "learning_rate": 0.0005, "loss": 1.5397090911865234, "step": 23774 }, { "epoch": 1.874260938115885, "grad_norm": 0.6993827819824219, "learning_rate": 0.0005, "loss": 1.66800057888031, "step": 23775 }, { "epoch": 1.8743397713835237, "grad_norm": 0.5673666596412659, "learning_rate": 0.0005, "loss": 1.5157880783081055, "step": 23776 }, { "epoch": 1.8744186046511628, "grad_norm": 0.5367709994316101, "learning_rate": 0.0005, "loss": 1.3573434352874756, "step": 23777 }, { "epoch": 1.8744974379188017, "grad_norm": 0.6035906076431274, "learning_rate": 0.0005, "loss": 1.4542804956436157, "step": 23778 }, { "epoch": 1.8745762711864407, "grad_norm": 0.44430994987487793, "learning_rate": 0.0005, "loss": 1.2764878273010254, "step": 23779 }, { "epoch": 1.8746551044540796, "grad_norm": 0.5594751834869385, "learning_rate": 0.0005, "loss": 1.4141753911972046, "step": 23780 }, { "epoch": 1.8747339377217185, "grad_norm": 0.6573612689971924, "learning_rate": 0.0005, "loss": 1.5404417514801025, "step": 23781 }, { "epoch": 1.8748127709893576, "grad_norm": 0.5141458511352539, "learning_rate": 0.0005, "loss": 1.4027206897735596, "step": 23782 }, { "epoch": 1.8748916042569963, "grad_norm": 0.5066289305686951, "learning_rate": 0.0005, "loss": 1.2207530736923218, "step": 23783 }, { "epoch": 1.8749704375246354, "grad_norm": 0.48087161779403687, "learning_rate": 0.0005, "loss": 1.3534519672393799, "step": 23784 }, { "epoch": 1.8750492707922743, "grad_norm": 0.668256402015686, "learning_rate": 0.0005, "loss": 1.6440318822860718, "step": 23785 }, { "epoch": 1.8751281040599133, "grad_norm": 0.7321467995643616, "learning_rate": 0.0005, "loss": 1.5305649042129517, "step": 23786 }, { "epoch": 1.8752069373275524, "grad_norm": 1.0346617698669434, "learning_rate": 0.0005, "loss": 1.4514086246490479, "step": 23787 }, { "epoch": 1.875285770595191, "grad_norm": 0.4561253786087036, "learning_rate": 0.0005, "loss": 1.365946650505066, "step": 23788 }, { "epoch": 1.8753646038628302, "grad_norm": 0.5228276252746582, "learning_rate": 0.0005, "loss": 1.3887341022491455, "step": 23789 }, { "epoch": 1.875443437130469, "grad_norm": 0.9840098023414612, "learning_rate": 0.0005, "loss": 1.2818102836608887, "step": 23790 }, { "epoch": 1.875522270398108, "grad_norm": 0.5279324054718018, "learning_rate": 0.0005, "loss": 1.1785616874694824, "step": 23791 }, { "epoch": 1.875601103665747, "grad_norm": 0.5929882526397705, "learning_rate": 0.0005, "loss": 1.5033408403396606, "step": 23792 }, { "epoch": 1.8756799369333859, "grad_norm": 0.5246434211730957, "learning_rate": 0.0005, "loss": 1.3626928329467773, "step": 23793 }, { "epoch": 1.875758770201025, "grad_norm": 0.5647168159484863, "learning_rate": 0.0005, "loss": 1.5653719902038574, "step": 23794 }, { "epoch": 1.8758376034686637, "grad_norm": 0.5546678304672241, "learning_rate": 0.0005, "loss": 1.432466983795166, "step": 23795 }, { "epoch": 1.8759164367363028, "grad_norm": 0.6543492078781128, "learning_rate": 0.0005, "loss": 1.5329320430755615, "step": 23796 }, { "epoch": 1.8759952700039415, "grad_norm": 0.4892437756061554, "learning_rate": 0.0005, "loss": 1.2931607961654663, "step": 23797 }, { "epoch": 1.8760741032715806, "grad_norm": 0.6411371827125549, "learning_rate": 0.0005, "loss": 1.7530176639556885, "step": 23798 }, { "epoch": 1.8761529365392196, "grad_norm": 0.5942831635475159, "learning_rate": 0.0005, "loss": 1.4762144088745117, "step": 23799 }, { "epoch": 1.8762317698068585, "grad_norm": 0.5302202105522156, "learning_rate": 0.0005, "loss": 1.3858869075775146, "step": 23800 }, { "epoch": 1.8763106030744976, "grad_norm": 0.5673287510871887, "learning_rate": 0.0005, "loss": 1.2338253259658813, "step": 23801 }, { "epoch": 1.8763894363421363, "grad_norm": 0.6393483877182007, "learning_rate": 0.0005, "loss": 1.489972472190857, "step": 23802 }, { "epoch": 1.8764682696097754, "grad_norm": 0.5445929765701294, "learning_rate": 0.0005, "loss": 1.3241311311721802, "step": 23803 }, { "epoch": 1.876547102877414, "grad_norm": 0.5357926487922668, "learning_rate": 0.0005, "loss": 1.4005300998687744, "step": 23804 }, { "epoch": 1.8766259361450532, "grad_norm": 0.5476162433624268, "learning_rate": 0.0005, "loss": 1.2105979919433594, "step": 23805 }, { "epoch": 1.8767047694126922, "grad_norm": 0.47866156697273254, "learning_rate": 0.0005, "loss": 1.5738730430603027, "step": 23806 }, { "epoch": 1.876783602680331, "grad_norm": 0.5073629021644592, "learning_rate": 0.0005, "loss": 1.4890364408493042, "step": 23807 }, { "epoch": 1.8768624359479702, "grad_norm": 0.6576064229011536, "learning_rate": 0.0005, "loss": 1.5143263339996338, "step": 23808 }, { "epoch": 1.8769412692156089, "grad_norm": 0.512420654296875, "learning_rate": 0.0005, "loss": 1.3895256519317627, "step": 23809 }, { "epoch": 1.877020102483248, "grad_norm": 0.47115790843963623, "learning_rate": 0.0005, "loss": 1.3544278144836426, "step": 23810 }, { "epoch": 1.8770989357508867, "grad_norm": 0.5222851634025574, "learning_rate": 0.0005, "loss": 1.5439329147338867, "step": 23811 }, { "epoch": 1.8771777690185258, "grad_norm": 0.6329902410507202, "learning_rate": 0.0005, "loss": 1.6099228858947754, "step": 23812 }, { "epoch": 1.8772566022861648, "grad_norm": 0.4649602174758911, "learning_rate": 0.0005, "loss": 1.1753313541412354, "step": 23813 }, { "epoch": 1.8773354355538037, "grad_norm": 0.55898118019104, "learning_rate": 0.0005, "loss": 1.4509837627410889, "step": 23814 }, { "epoch": 1.8774142688214428, "grad_norm": 0.4873017370700836, "learning_rate": 0.0005, "loss": 1.2835294008255005, "step": 23815 }, { "epoch": 1.8774931020890815, "grad_norm": 0.7164594531059265, "learning_rate": 0.0005, "loss": 1.6459953784942627, "step": 23816 }, { "epoch": 1.8775719353567206, "grad_norm": 0.5237869620323181, "learning_rate": 0.0005, "loss": 1.7507429122924805, "step": 23817 }, { "epoch": 1.8776507686243595, "grad_norm": 0.6024656891822815, "learning_rate": 0.0005, "loss": 1.4238133430480957, "step": 23818 }, { "epoch": 1.8777296018919984, "grad_norm": 0.5243712067604065, "learning_rate": 0.0005, "loss": 1.643244981765747, "step": 23819 }, { "epoch": 1.8778084351596374, "grad_norm": 0.4877072274684906, "learning_rate": 0.0005, "loss": 1.4263432025909424, "step": 23820 }, { "epoch": 1.8778872684272763, "grad_norm": 0.5836644172668457, "learning_rate": 0.0005, "loss": 1.4592357873916626, "step": 23821 }, { "epoch": 1.8779661016949154, "grad_norm": 0.5771045684814453, "learning_rate": 0.0005, "loss": 1.5346589088439941, "step": 23822 }, { "epoch": 1.878044934962554, "grad_norm": 0.6653230786323547, "learning_rate": 0.0005, "loss": 1.4272407293319702, "step": 23823 }, { "epoch": 1.8781237682301932, "grad_norm": 0.5900045037269592, "learning_rate": 0.0005, "loss": 1.2324419021606445, "step": 23824 }, { "epoch": 1.8782026014978321, "grad_norm": 0.5146934390068054, "learning_rate": 0.0005, "loss": 1.499751329421997, "step": 23825 }, { "epoch": 1.878281434765471, "grad_norm": 0.5712001323699951, "learning_rate": 0.0005, "loss": 1.4531021118164062, "step": 23826 }, { "epoch": 1.87836026803311, "grad_norm": 0.6427950859069824, "learning_rate": 0.0005, "loss": 1.3142837285995483, "step": 23827 }, { "epoch": 1.8784391013007489, "grad_norm": 0.4510335624217987, "learning_rate": 0.0005, "loss": 1.212944507598877, "step": 23828 }, { "epoch": 1.878517934568388, "grad_norm": 0.5095226168632507, "learning_rate": 0.0005, "loss": 1.2476047277450562, "step": 23829 }, { "epoch": 1.8785967678360267, "grad_norm": 0.593359649181366, "learning_rate": 0.0005, "loss": 1.6804499626159668, "step": 23830 }, { "epoch": 1.8786756011036658, "grad_norm": 0.5113111138343811, "learning_rate": 0.0005, "loss": 1.4272143840789795, "step": 23831 }, { "epoch": 1.8787544343713047, "grad_norm": 0.561557948589325, "learning_rate": 0.0005, "loss": 1.429944634437561, "step": 23832 }, { "epoch": 1.8788332676389436, "grad_norm": 0.5462425947189331, "learning_rate": 0.0005, "loss": 1.3455780744552612, "step": 23833 }, { "epoch": 1.8789121009065826, "grad_norm": 0.46851852536201477, "learning_rate": 0.0005, "loss": 1.4766156673431396, "step": 23834 }, { "epoch": 1.8789909341742215, "grad_norm": 0.4842769503593445, "learning_rate": 0.0005, "loss": 1.4299715757369995, "step": 23835 }, { "epoch": 1.8790697674418606, "grad_norm": 0.5888895988464355, "learning_rate": 0.0005, "loss": 1.3826546669006348, "step": 23836 }, { "epoch": 1.8791486007094993, "grad_norm": 0.5116888880729675, "learning_rate": 0.0005, "loss": 1.4293627738952637, "step": 23837 }, { "epoch": 1.8792274339771384, "grad_norm": 0.5020262002944946, "learning_rate": 0.0005, "loss": 1.34408438205719, "step": 23838 }, { "epoch": 1.8793062672447773, "grad_norm": 0.4364674389362335, "learning_rate": 0.0005, "loss": 1.3274208307266235, "step": 23839 }, { "epoch": 1.8793851005124163, "grad_norm": 0.4761766791343689, "learning_rate": 0.0005, "loss": 1.3834959268569946, "step": 23840 }, { "epoch": 1.8794639337800552, "grad_norm": 0.4954790771007538, "learning_rate": 0.0005, "loss": 1.424452543258667, "step": 23841 }, { "epoch": 1.879542767047694, "grad_norm": 0.6264981627464294, "learning_rate": 0.0005, "loss": 1.297333002090454, "step": 23842 }, { "epoch": 1.8796216003153332, "grad_norm": 0.509253203868866, "learning_rate": 0.0005, "loss": 1.5046427249908447, "step": 23843 }, { "epoch": 1.879700433582972, "grad_norm": 0.5095364451408386, "learning_rate": 0.0005, "loss": 1.6632373332977295, "step": 23844 }, { "epoch": 1.879779266850611, "grad_norm": 0.707360029220581, "learning_rate": 0.0005, "loss": 1.6113758087158203, "step": 23845 }, { "epoch": 1.87985810011825, "grad_norm": 0.696727991104126, "learning_rate": 0.0005, "loss": 1.450363278388977, "step": 23846 }, { "epoch": 1.8799369333858889, "grad_norm": 0.547213613986969, "learning_rate": 0.0005, "loss": 1.4516968727111816, "step": 23847 }, { "epoch": 1.8800157666535278, "grad_norm": 0.609790563583374, "learning_rate": 0.0005, "loss": 1.690569519996643, "step": 23848 }, { "epoch": 1.8800945999211667, "grad_norm": 0.4382527470588684, "learning_rate": 0.0005, "loss": 1.245445966720581, "step": 23849 }, { "epoch": 1.8801734331888058, "grad_norm": 0.48031896352767944, "learning_rate": 0.0005, "loss": 1.5290257930755615, "step": 23850 }, { "epoch": 1.8802522664564445, "grad_norm": 0.6108389496803284, "learning_rate": 0.0005, "loss": 1.1426708698272705, "step": 23851 }, { "epoch": 1.8803310997240836, "grad_norm": 0.5890876650810242, "learning_rate": 0.0005, "loss": 1.5314491987228394, "step": 23852 }, { "epoch": 1.8804099329917225, "grad_norm": 0.5099321007728577, "learning_rate": 0.0005, "loss": 1.4719974994659424, "step": 23853 }, { "epoch": 1.8804887662593615, "grad_norm": 0.6258978843688965, "learning_rate": 0.0005, "loss": 1.4666725397109985, "step": 23854 }, { "epoch": 1.8805675995270004, "grad_norm": 0.5647324323654175, "learning_rate": 0.0005, "loss": 1.3585447072982788, "step": 23855 }, { "epoch": 1.8806464327946393, "grad_norm": 0.4752306342124939, "learning_rate": 0.0005, "loss": 1.5338237285614014, "step": 23856 }, { "epoch": 1.8807252660622784, "grad_norm": 0.5510400533676147, "learning_rate": 0.0005, "loss": 1.3388195037841797, "step": 23857 }, { "epoch": 1.880804099329917, "grad_norm": 0.438960462808609, "learning_rate": 0.0005, "loss": 1.3384127616882324, "step": 23858 }, { "epoch": 1.8808829325975562, "grad_norm": 0.5790637135505676, "learning_rate": 0.0005, "loss": 1.6794474124908447, "step": 23859 }, { "epoch": 1.8809617658651951, "grad_norm": 0.4583468735218048, "learning_rate": 0.0005, "loss": 1.296128511428833, "step": 23860 }, { "epoch": 1.881040599132834, "grad_norm": 0.43541306257247925, "learning_rate": 0.0005, "loss": 1.4789073467254639, "step": 23861 }, { "epoch": 1.881119432400473, "grad_norm": 0.4659639894962311, "learning_rate": 0.0005, "loss": 1.2865890264511108, "step": 23862 }, { "epoch": 1.8811982656681119, "grad_norm": 0.5403299331665039, "learning_rate": 0.0005, "loss": 1.3007667064666748, "step": 23863 }, { "epoch": 1.881277098935751, "grad_norm": 0.4265335202217102, "learning_rate": 0.0005, "loss": 1.2875714302062988, "step": 23864 }, { "epoch": 1.8813559322033897, "grad_norm": 0.6260302662849426, "learning_rate": 0.0005, "loss": 1.3896870613098145, "step": 23865 }, { "epoch": 1.8814347654710288, "grad_norm": 0.46856310963630676, "learning_rate": 0.0005, "loss": 1.3980714082717896, "step": 23866 }, { "epoch": 1.8815135987386677, "grad_norm": 0.6960768699645996, "learning_rate": 0.0005, "loss": 1.426340103149414, "step": 23867 }, { "epoch": 1.8815924320063067, "grad_norm": 0.5286604762077332, "learning_rate": 0.0005, "loss": 1.4404617547988892, "step": 23868 }, { "epoch": 1.8816712652739456, "grad_norm": 0.7136574387550354, "learning_rate": 0.0005, "loss": 1.4720847606658936, "step": 23869 }, { "epoch": 1.8817500985415845, "grad_norm": 0.4922610819339752, "learning_rate": 0.0005, "loss": 1.4835395812988281, "step": 23870 }, { "epoch": 1.8818289318092236, "grad_norm": 0.6467944979667664, "learning_rate": 0.0005, "loss": 1.3717787265777588, "step": 23871 }, { "epoch": 1.8819077650768623, "grad_norm": 0.4952716529369354, "learning_rate": 0.0005, "loss": 1.461955189704895, "step": 23872 }, { "epoch": 1.8819865983445014, "grad_norm": 0.5668825507164001, "learning_rate": 0.0005, "loss": 1.4535436630249023, "step": 23873 }, { "epoch": 1.8820654316121403, "grad_norm": 0.5018689036369324, "learning_rate": 0.0005, "loss": 1.3720529079437256, "step": 23874 }, { "epoch": 1.8821442648797793, "grad_norm": 0.8144994378089905, "learning_rate": 0.0005, "loss": 1.5017119646072388, "step": 23875 }, { "epoch": 1.8822230981474182, "grad_norm": 0.658717691898346, "learning_rate": 0.0005, "loss": 1.6516599655151367, "step": 23876 }, { "epoch": 1.882301931415057, "grad_norm": 0.4419926404953003, "learning_rate": 0.0005, "loss": 1.3293622732162476, "step": 23877 }, { "epoch": 1.8823807646826962, "grad_norm": 0.7642676830291748, "learning_rate": 0.0005, "loss": 1.4087677001953125, "step": 23878 }, { "epoch": 1.882459597950335, "grad_norm": 1.0827176570892334, "learning_rate": 0.0005, "loss": 1.7455772161483765, "step": 23879 }, { "epoch": 1.882538431217974, "grad_norm": 0.4512898623943329, "learning_rate": 0.0005, "loss": 1.5035371780395508, "step": 23880 }, { "epoch": 1.882617264485613, "grad_norm": 0.5546684265136719, "learning_rate": 0.0005, "loss": 1.5162861347198486, "step": 23881 }, { "epoch": 1.8826960977532519, "grad_norm": 0.7604513168334961, "learning_rate": 0.0005, "loss": 1.5441396236419678, "step": 23882 }, { "epoch": 1.8827749310208908, "grad_norm": 0.43607959151268005, "learning_rate": 0.0005, "loss": 1.2045104503631592, "step": 23883 }, { "epoch": 1.8828537642885297, "grad_norm": 0.6361058950424194, "learning_rate": 0.0005, "loss": 1.4626035690307617, "step": 23884 }, { "epoch": 1.8829325975561688, "grad_norm": 0.5472556948661804, "learning_rate": 0.0005, "loss": 1.494134783744812, "step": 23885 }, { "epoch": 1.8830114308238075, "grad_norm": 0.5344252586364746, "learning_rate": 0.0005, "loss": 1.2793028354644775, "step": 23886 }, { "epoch": 1.8830902640914466, "grad_norm": 0.5599009990692139, "learning_rate": 0.0005, "loss": 1.437709093093872, "step": 23887 }, { "epoch": 1.8831690973590856, "grad_norm": 1.2197595834732056, "learning_rate": 0.0005, "loss": 1.4152790307998657, "step": 23888 }, { "epoch": 1.8832479306267245, "grad_norm": 0.566270649433136, "learning_rate": 0.0005, "loss": 1.65595281124115, "step": 23889 }, { "epoch": 1.8833267638943634, "grad_norm": 0.5047798752784729, "learning_rate": 0.0005, "loss": 1.425778865814209, "step": 23890 }, { "epoch": 1.8834055971620023, "grad_norm": 0.5077242255210876, "learning_rate": 0.0005, "loss": 1.4842478036880493, "step": 23891 }, { "epoch": 1.8834844304296414, "grad_norm": 0.5248075127601624, "learning_rate": 0.0005, "loss": 1.2028589248657227, "step": 23892 }, { "epoch": 1.88356326369728, "grad_norm": 0.7324362397193909, "learning_rate": 0.0005, "loss": 1.4992778301239014, "step": 23893 }, { "epoch": 1.8836420969649192, "grad_norm": 0.4940844476222992, "learning_rate": 0.0005, "loss": 1.3846626281738281, "step": 23894 }, { "epoch": 1.8837209302325582, "grad_norm": 0.4996662139892578, "learning_rate": 0.0005, "loss": 1.4534889459609985, "step": 23895 }, { "epoch": 1.883799763500197, "grad_norm": 0.7079864144325256, "learning_rate": 0.0005, "loss": 1.6240313053131104, "step": 23896 }, { "epoch": 1.883878596767836, "grad_norm": 0.5091513395309448, "learning_rate": 0.0005, "loss": 1.410499095916748, "step": 23897 }, { "epoch": 1.883957430035475, "grad_norm": 0.5250381231307983, "learning_rate": 0.0005, "loss": 1.5854815244674683, "step": 23898 }, { "epoch": 1.884036263303114, "grad_norm": 0.6369062066078186, "learning_rate": 0.0005, "loss": 1.4054298400878906, "step": 23899 }, { "epoch": 1.8841150965707527, "grad_norm": 0.7100235819816589, "learning_rate": 0.0005, "loss": 1.4508275985717773, "step": 23900 }, { "epoch": 1.8841939298383918, "grad_norm": 0.5561302900314331, "learning_rate": 0.0005, "loss": 1.3913626670837402, "step": 23901 }, { "epoch": 1.8842727631060308, "grad_norm": 0.534751832485199, "learning_rate": 0.0005, "loss": 1.5616319179534912, "step": 23902 }, { "epoch": 1.8843515963736697, "grad_norm": 0.5330801010131836, "learning_rate": 0.0005, "loss": 1.5066800117492676, "step": 23903 }, { "epoch": 1.8844304296413086, "grad_norm": 0.679252028465271, "learning_rate": 0.0005, "loss": 1.6794955730438232, "step": 23904 }, { "epoch": 1.8845092629089475, "grad_norm": 0.49706748127937317, "learning_rate": 0.0005, "loss": 1.4175136089324951, "step": 23905 }, { "epoch": 1.8845880961765866, "grad_norm": 0.5110360980033875, "learning_rate": 0.0005, "loss": 1.4820489883422852, "step": 23906 }, { "epoch": 1.8846669294442253, "grad_norm": 3.230769157409668, "learning_rate": 0.0005, "loss": 1.3989636898040771, "step": 23907 }, { "epoch": 1.8847457627118644, "grad_norm": 0.5183500647544861, "learning_rate": 0.0005, "loss": 1.2852182388305664, "step": 23908 }, { "epoch": 1.8848245959795034, "grad_norm": 0.9078460335731506, "learning_rate": 0.0005, "loss": 1.5328307151794434, "step": 23909 }, { "epoch": 1.8849034292471423, "grad_norm": 0.4733680784702301, "learning_rate": 0.0005, "loss": 1.6188538074493408, "step": 23910 }, { "epoch": 1.8849822625147814, "grad_norm": 0.48797640204429626, "learning_rate": 0.0005, "loss": 1.3526828289031982, "step": 23911 }, { "epoch": 1.88506109578242, "grad_norm": 0.5075798034667969, "learning_rate": 0.0005, "loss": 1.1734633445739746, "step": 23912 }, { "epoch": 1.8851399290500592, "grad_norm": 0.5454968214035034, "learning_rate": 0.0005, "loss": 1.3475747108459473, "step": 23913 }, { "epoch": 1.885218762317698, "grad_norm": 0.4787675142288208, "learning_rate": 0.0005, "loss": 1.3630297183990479, "step": 23914 }, { "epoch": 1.885297595585337, "grad_norm": 0.6356174349784851, "learning_rate": 0.0005, "loss": 1.4004424810409546, "step": 23915 }, { "epoch": 1.885376428852976, "grad_norm": 0.546868085861206, "learning_rate": 0.0005, "loss": 1.4397305250167847, "step": 23916 }, { "epoch": 1.8854552621206149, "grad_norm": 0.45935046672821045, "learning_rate": 0.0005, "loss": 1.3564741611480713, "step": 23917 }, { "epoch": 1.885534095388254, "grad_norm": 0.509850025177002, "learning_rate": 0.0005, "loss": 1.3257434368133545, "step": 23918 }, { "epoch": 1.8856129286558927, "grad_norm": 0.5761356949806213, "learning_rate": 0.0005, "loss": 1.432732343673706, "step": 23919 }, { "epoch": 1.8856917619235318, "grad_norm": 0.4499971270561218, "learning_rate": 0.0005, "loss": 1.3731952905654907, "step": 23920 }, { "epoch": 1.8857705951911705, "grad_norm": 0.5310025811195374, "learning_rate": 0.0005, "loss": 1.4619473218917847, "step": 23921 }, { "epoch": 1.8858494284588097, "grad_norm": 0.565495491027832, "learning_rate": 0.0005, "loss": 1.6230707168579102, "step": 23922 }, { "epoch": 1.8859282617264486, "grad_norm": 0.5746888518333435, "learning_rate": 0.0005, "loss": 1.3779481649398804, "step": 23923 }, { "epoch": 1.8860070949940875, "grad_norm": 0.4760454297065735, "learning_rate": 0.0005, "loss": 1.4747579097747803, "step": 23924 }, { "epoch": 1.8860859282617266, "grad_norm": 0.6427157521247864, "learning_rate": 0.0005, "loss": 1.3883979320526123, "step": 23925 }, { "epoch": 1.8861647615293653, "grad_norm": 0.527866780757904, "learning_rate": 0.0005, "loss": 1.2772815227508545, "step": 23926 }, { "epoch": 1.8862435947970044, "grad_norm": 0.6394774317741394, "learning_rate": 0.0005, "loss": 1.6663129329681396, "step": 23927 }, { "epoch": 1.8863224280646431, "grad_norm": 0.5013526082038879, "learning_rate": 0.0005, "loss": 1.5596437454223633, "step": 23928 }, { "epoch": 1.8864012613322823, "grad_norm": 0.516772985458374, "learning_rate": 0.0005, "loss": 1.480690360069275, "step": 23929 }, { "epoch": 1.8864800945999212, "grad_norm": 0.993809163570404, "learning_rate": 0.0005, "loss": 1.5327534675598145, "step": 23930 }, { "epoch": 1.88655892786756, "grad_norm": 0.5683853030204773, "learning_rate": 0.0005, "loss": 1.560854196548462, "step": 23931 }, { "epoch": 1.8866377611351992, "grad_norm": 0.4899543523788452, "learning_rate": 0.0005, "loss": 1.2741026878356934, "step": 23932 }, { "epoch": 1.886716594402838, "grad_norm": 0.49599945545196533, "learning_rate": 0.0005, "loss": 1.3233866691589355, "step": 23933 }, { "epoch": 1.886795427670477, "grad_norm": 0.5536879301071167, "learning_rate": 0.0005, "loss": 1.3528791666030884, "step": 23934 }, { "epoch": 1.8868742609381157, "grad_norm": 0.5731295347213745, "learning_rate": 0.0005, "loss": 1.4270260334014893, "step": 23935 }, { "epoch": 1.8869530942057549, "grad_norm": 0.49214887619018555, "learning_rate": 0.0005, "loss": 1.526113748550415, "step": 23936 }, { "epoch": 1.8870319274733938, "grad_norm": 0.5837427377700806, "learning_rate": 0.0005, "loss": 1.620266318321228, "step": 23937 }, { "epoch": 1.8871107607410327, "grad_norm": 0.6561414003372192, "learning_rate": 0.0005, "loss": 1.4372384548187256, "step": 23938 }, { "epoch": 1.8871895940086718, "grad_norm": 0.5580058097839355, "learning_rate": 0.0005, "loss": 1.2838799953460693, "step": 23939 }, { "epoch": 1.8872684272763105, "grad_norm": 0.5044751763343811, "learning_rate": 0.0005, "loss": 1.4465062618255615, "step": 23940 }, { "epoch": 1.8873472605439496, "grad_norm": 0.48982101678848267, "learning_rate": 0.0005, "loss": 1.2488129138946533, "step": 23941 }, { "epoch": 1.8874260938115885, "grad_norm": 0.4546608030796051, "learning_rate": 0.0005, "loss": 1.284224271774292, "step": 23942 }, { "epoch": 1.8875049270792275, "grad_norm": 0.4274555742740631, "learning_rate": 0.0005, "loss": 1.0916948318481445, "step": 23943 }, { "epoch": 1.8875837603468664, "grad_norm": 0.4158613085746765, "learning_rate": 0.0005, "loss": 1.4001879692077637, "step": 23944 }, { "epoch": 1.8876625936145053, "grad_norm": 0.5108771920204163, "learning_rate": 0.0005, "loss": 1.410674810409546, "step": 23945 }, { "epoch": 1.8877414268821444, "grad_norm": 0.5891469120979309, "learning_rate": 0.0005, "loss": 1.5853890180587769, "step": 23946 }, { "epoch": 1.887820260149783, "grad_norm": 0.5049085021018982, "learning_rate": 0.0005, "loss": 1.3569358587265015, "step": 23947 }, { "epoch": 1.8878990934174222, "grad_norm": 0.5647748708724976, "learning_rate": 0.0005, "loss": 1.6589562892913818, "step": 23948 }, { "epoch": 1.8879779266850611, "grad_norm": 0.445140540599823, "learning_rate": 0.0005, "loss": 1.411893367767334, "step": 23949 }, { "epoch": 1.8880567599527, "grad_norm": 0.5110052227973938, "learning_rate": 0.0005, "loss": 1.4007654190063477, "step": 23950 }, { "epoch": 1.888135593220339, "grad_norm": 0.5987735390663147, "learning_rate": 0.0005, "loss": 1.5154597759246826, "step": 23951 }, { "epoch": 1.8882144264879779, "grad_norm": 0.6063928604125977, "learning_rate": 0.0005, "loss": 1.521589994430542, "step": 23952 }, { "epoch": 1.888293259755617, "grad_norm": 0.49288350343704224, "learning_rate": 0.0005, "loss": 1.2658584117889404, "step": 23953 }, { "epoch": 1.8883720930232557, "grad_norm": 0.4746440649032593, "learning_rate": 0.0005, "loss": 1.3456830978393555, "step": 23954 }, { "epoch": 1.8884509262908948, "grad_norm": 0.6836821436882019, "learning_rate": 0.0005, "loss": 1.4045515060424805, "step": 23955 }, { "epoch": 1.8885297595585337, "grad_norm": 0.5685904026031494, "learning_rate": 0.0005, "loss": 1.2702414989471436, "step": 23956 }, { "epoch": 1.8886085928261727, "grad_norm": 0.46558311581611633, "learning_rate": 0.0005, "loss": 1.3889656066894531, "step": 23957 }, { "epoch": 1.8886874260938116, "grad_norm": 0.6061565279960632, "learning_rate": 0.0005, "loss": 1.4011541604995728, "step": 23958 }, { "epoch": 1.8887662593614505, "grad_norm": 0.48149189352989197, "learning_rate": 0.0005, "loss": 1.4561619758605957, "step": 23959 }, { "epoch": 1.8888450926290896, "grad_norm": 0.445147305727005, "learning_rate": 0.0005, "loss": 1.0933904647827148, "step": 23960 }, { "epoch": 1.8889239258967283, "grad_norm": 0.48889151215553284, "learning_rate": 0.0005, "loss": 1.44444739818573, "step": 23961 }, { "epoch": 1.8890027591643674, "grad_norm": 0.4643503427505493, "learning_rate": 0.0005, "loss": 1.3294785022735596, "step": 23962 }, { "epoch": 1.8890815924320064, "grad_norm": 0.44427138566970825, "learning_rate": 0.0005, "loss": 1.4361653327941895, "step": 23963 }, { "epoch": 1.8891604256996453, "grad_norm": 0.5675172805786133, "learning_rate": 0.0005, "loss": 1.4223005771636963, "step": 23964 }, { "epoch": 1.8892392589672842, "grad_norm": 0.6379579901695251, "learning_rate": 0.0005, "loss": 1.7876479625701904, "step": 23965 }, { "epoch": 1.889318092234923, "grad_norm": 0.5305483341217041, "learning_rate": 0.0005, "loss": 1.1991031169891357, "step": 23966 }, { "epoch": 1.8893969255025622, "grad_norm": 0.567683219909668, "learning_rate": 0.0005, "loss": 1.2935662269592285, "step": 23967 }, { "epoch": 1.889475758770201, "grad_norm": 0.55818772315979, "learning_rate": 0.0005, "loss": 1.3312454223632812, "step": 23968 }, { "epoch": 1.88955459203784, "grad_norm": 0.41390764713287354, "learning_rate": 0.0005, "loss": 1.292923092842102, "step": 23969 }, { "epoch": 1.889633425305479, "grad_norm": 0.7644583582878113, "learning_rate": 0.0005, "loss": 1.4836726188659668, "step": 23970 }, { "epoch": 1.8897122585731179, "grad_norm": 0.6563935875892639, "learning_rate": 0.0005, "loss": 1.357433557510376, "step": 23971 }, { "epoch": 1.8897910918407568, "grad_norm": 0.5976136326789856, "learning_rate": 0.0005, "loss": 1.5183627605438232, "step": 23972 }, { "epoch": 1.8898699251083957, "grad_norm": 0.5696526169776917, "learning_rate": 0.0005, "loss": 1.4402387142181396, "step": 23973 }, { "epoch": 1.8899487583760348, "grad_norm": 0.48319151997566223, "learning_rate": 0.0005, "loss": 1.2594873905181885, "step": 23974 }, { "epoch": 1.8900275916436735, "grad_norm": 0.4201761484146118, "learning_rate": 0.0005, "loss": 1.29158353805542, "step": 23975 }, { "epoch": 1.8901064249113126, "grad_norm": 0.5967512130737305, "learning_rate": 0.0005, "loss": 1.8708789348602295, "step": 23976 }, { "epoch": 1.8901852581789516, "grad_norm": 0.5772320628166199, "learning_rate": 0.0005, "loss": 1.4271621704101562, "step": 23977 }, { "epoch": 1.8902640914465905, "grad_norm": 0.44555866718292236, "learning_rate": 0.0005, "loss": 1.372086524963379, "step": 23978 }, { "epoch": 1.8903429247142294, "grad_norm": 0.4969414472579956, "learning_rate": 0.0005, "loss": 1.3540077209472656, "step": 23979 }, { "epoch": 1.8904217579818683, "grad_norm": 0.5096916556358337, "learning_rate": 0.0005, "loss": 1.2374637126922607, "step": 23980 }, { "epoch": 1.8905005912495074, "grad_norm": 0.6714761257171631, "learning_rate": 0.0005, "loss": 1.3201851844787598, "step": 23981 }, { "epoch": 1.8905794245171461, "grad_norm": 0.48966890573501587, "learning_rate": 0.0005, "loss": 1.2967734336853027, "step": 23982 }, { "epoch": 1.8906582577847852, "grad_norm": 0.46102505922317505, "learning_rate": 0.0005, "loss": 1.1661417484283447, "step": 23983 }, { "epoch": 1.8907370910524242, "grad_norm": 0.49543070793151855, "learning_rate": 0.0005, "loss": 1.1421411037445068, "step": 23984 }, { "epoch": 1.890815924320063, "grad_norm": 0.595529317855835, "learning_rate": 0.0005, "loss": 1.4964585304260254, "step": 23985 }, { "epoch": 1.890894757587702, "grad_norm": 0.6050096750259399, "learning_rate": 0.0005, "loss": 1.213409423828125, "step": 23986 }, { "epoch": 1.890973590855341, "grad_norm": 0.5094044208526611, "learning_rate": 0.0005, "loss": 1.5520342588424683, "step": 23987 }, { "epoch": 1.89105242412298, "grad_norm": 2.619056224822998, "learning_rate": 0.0005, "loss": 1.4678168296813965, "step": 23988 }, { "epoch": 1.8911312573906187, "grad_norm": 0.5153116583824158, "learning_rate": 0.0005, "loss": 1.272480845451355, "step": 23989 }, { "epoch": 1.8912100906582578, "grad_norm": 0.656925618648529, "learning_rate": 0.0005, "loss": 1.5016236305236816, "step": 23990 }, { "epoch": 1.8912889239258968, "grad_norm": 0.6173787117004395, "learning_rate": 0.0005, "loss": 1.5269088745117188, "step": 23991 }, { "epoch": 1.8913677571935357, "grad_norm": 0.579452633857727, "learning_rate": 0.0005, "loss": 1.645056962966919, "step": 23992 }, { "epoch": 1.8914465904611746, "grad_norm": 0.464253306388855, "learning_rate": 0.0005, "loss": 1.319634199142456, "step": 23993 }, { "epoch": 1.8915254237288135, "grad_norm": 0.4777824282646179, "learning_rate": 0.0005, "loss": 1.584057331085205, "step": 23994 }, { "epoch": 1.8916042569964526, "grad_norm": 0.5050303936004639, "learning_rate": 0.0005, "loss": 1.3304145336151123, "step": 23995 }, { "epoch": 1.8916830902640913, "grad_norm": 0.6393989324569702, "learning_rate": 0.0005, "loss": 1.443845510482788, "step": 23996 }, { "epoch": 1.8917619235317304, "grad_norm": 0.6326448321342468, "learning_rate": 0.0005, "loss": 1.57344651222229, "step": 23997 }, { "epoch": 1.8918407567993694, "grad_norm": 0.5091986060142517, "learning_rate": 0.0005, "loss": 1.3082191944122314, "step": 23998 }, { "epoch": 1.8919195900670083, "grad_norm": 0.5826073884963989, "learning_rate": 0.0005, "loss": 1.6293690204620361, "step": 23999 }, { "epoch": 1.8919984233346472, "grad_norm": 0.44252684712409973, "learning_rate": 0.0005, "loss": 1.2181047201156616, "step": 24000 }, { "epoch": 1.892077256602286, "grad_norm": 0.4841567277908325, "learning_rate": 0.0005, "loss": 1.3118350505828857, "step": 24001 }, { "epoch": 1.8921560898699252, "grad_norm": 0.5052866339683533, "learning_rate": 0.0005, "loss": 1.2490482330322266, "step": 24002 }, { "epoch": 1.892234923137564, "grad_norm": 0.4749729037284851, "learning_rate": 0.0005, "loss": 1.3303229808807373, "step": 24003 }, { "epoch": 1.892313756405203, "grad_norm": 0.4889245629310608, "learning_rate": 0.0005, "loss": 1.5022902488708496, "step": 24004 }, { "epoch": 1.892392589672842, "grad_norm": 0.5052606463432312, "learning_rate": 0.0005, "loss": 1.545386791229248, "step": 24005 }, { "epoch": 1.8924714229404809, "grad_norm": 0.504146158695221, "learning_rate": 0.0005, "loss": 1.6056565046310425, "step": 24006 }, { "epoch": 1.8925502562081198, "grad_norm": 0.4986785650253296, "learning_rate": 0.0005, "loss": 1.5071765184402466, "step": 24007 }, { "epoch": 1.8926290894757587, "grad_norm": 0.491793155670166, "learning_rate": 0.0005, "loss": 1.2589843273162842, "step": 24008 }, { "epoch": 1.8927079227433978, "grad_norm": 0.4338926672935486, "learning_rate": 0.0005, "loss": 1.2589328289031982, "step": 24009 }, { "epoch": 1.8927867560110365, "grad_norm": 0.4582003355026245, "learning_rate": 0.0005, "loss": 1.327627420425415, "step": 24010 }, { "epoch": 1.8928655892786757, "grad_norm": 0.5403716564178467, "learning_rate": 0.0005, "loss": 1.3441355228424072, "step": 24011 }, { "epoch": 1.8929444225463146, "grad_norm": 0.5353524088859558, "learning_rate": 0.0005, "loss": 1.5753041505813599, "step": 24012 }, { "epoch": 1.8930232558139535, "grad_norm": 0.614112138748169, "learning_rate": 0.0005, "loss": 1.4969158172607422, "step": 24013 }, { "epoch": 1.8931020890815924, "grad_norm": 0.6451431512832642, "learning_rate": 0.0005, "loss": 1.6581792831420898, "step": 24014 }, { "epoch": 1.8931809223492313, "grad_norm": 0.4720894992351532, "learning_rate": 0.0005, "loss": 1.350774884223938, "step": 24015 }, { "epoch": 1.8932597556168704, "grad_norm": 0.3800542950630188, "learning_rate": 0.0005, "loss": 1.0882292985916138, "step": 24016 }, { "epoch": 1.8933385888845091, "grad_norm": 0.4673779308795929, "learning_rate": 0.0005, "loss": 1.3429677486419678, "step": 24017 }, { "epoch": 1.8934174221521483, "grad_norm": 0.46770697832107544, "learning_rate": 0.0005, "loss": 1.2996580600738525, "step": 24018 }, { "epoch": 1.8934962554197872, "grad_norm": 0.4853592813014984, "learning_rate": 0.0005, "loss": 1.5059173107147217, "step": 24019 }, { "epoch": 1.893575088687426, "grad_norm": 0.44740042090415955, "learning_rate": 0.0005, "loss": 1.2954790592193604, "step": 24020 }, { "epoch": 1.893653921955065, "grad_norm": 0.47294294834136963, "learning_rate": 0.0005, "loss": 1.2118752002716064, "step": 24021 }, { "epoch": 1.893732755222704, "grad_norm": 0.5832002758979797, "learning_rate": 0.0005, "loss": 1.5467805862426758, "step": 24022 }, { "epoch": 1.893811588490343, "grad_norm": 0.4235585033893585, "learning_rate": 0.0005, "loss": 1.3696446418762207, "step": 24023 }, { "epoch": 1.8938904217579817, "grad_norm": 0.48565536737442017, "learning_rate": 0.0005, "loss": 1.333214282989502, "step": 24024 }, { "epoch": 1.8939692550256209, "grad_norm": 0.6132749915122986, "learning_rate": 0.0005, "loss": 1.6319198608398438, "step": 24025 }, { "epoch": 1.8940480882932598, "grad_norm": 0.4516279399394989, "learning_rate": 0.0005, "loss": 1.2383291721343994, "step": 24026 }, { "epoch": 1.8941269215608987, "grad_norm": 0.7760177254676819, "learning_rate": 0.0005, "loss": 1.442167043685913, "step": 24027 }, { "epoch": 1.8942057548285376, "grad_norm": 0.5560088753700256, "learning_rate": 0.0005, "loss": 1.4579596519470215, "step": 24028 }, { "epoch": 1.8942845880961765, "grad_norm": 0.48117342591285706, "learning_rate": 0.0005, "loss": 1.2388204336166382, "step": 24029 }, { "epoch": 1.8943634213638156, "grad_norm": 0.5294064283370972, "learning_rate": 0.0005, "loss": 1.5099014043807983, "step": 24030 }, { "epoch": 1.8944422546314543, "grad_norm": 0.4744924306869507, "learning_rate": 0.0005, "loss": 1.306433081626892, "step": 24031 }, { "epoch": 1.8945210878990935, "grad_norm": 0.5009076595306396, "learning_rate": 0.0005, "loss": 1.4777882099151611, "step": 24032 }, { "epoch": 1.8945999211667324, "grad_norm": 0.4137043058872223, "learning_rate": 0.0005, "loss": 1.32151198387146, "step": 24033 }, { "epoch": 1.8946787544343713, "grad_norm": 0.6463325619697571, "learning_rate": 0.0005, "loss": 1.65596604347229, "step": 24034 }, { "epoch": 1.8947575877020102, "grad_norm": 0.6101812720298767, "learning_rate": 0.0005, "loss": 1.6203014850616455, "step": 24035 }, { "epoch": 1.894836420969649, "grad_norm": 0.3964526951313019, "learning_rate": 0.0005, "loss": 1.293778896331787, "step": 24036 }, { "epoch": 1.8949152542372882, "grad_norm": 0.47864827513694763, "learning_rate": 0.0005, "loss": 1.2378817796707153, "step": 24037 }, { "epoch": 1.894994087504927, "grad_norm": 0.530979335308075, "learning_rate": 0.0005, "loss": 1.5723339319229126, "step": 24038 }, { "epoch": 1.895072920772566, "grad_norm": 0.6677145957946777, "learning_rate": 0.0005, "loss": 1.440863847732544, "step": 24039 }, { "epoch": 1.895151754040205, "grad_norm": 0.4326111078262329, "learning_rate": 0.0005, "loss": 1.2220325469970703, "step": 24040 }, { "epoch": 1.8952305873078439, "grad_norm": 0.6031753420829773, "learning_rate": 0.0005, "loss": 1.520111322402954, "step": 24041 }, { "epoch": 1.895309420575483, "grad_norm": 0.6359008550643921, "learning_rate": 0.0005, "loss": 1.429610013961792, "step": 24042 }, { "epoch": 1.8953882538431217, "grad_norm": 0.6133980751037598, "learning_rate": 0.0005, "loss": 1.5741629600524902, "step": 24043 }, { "epoch": 1.8954670871107608, "grad_norm": 0.7191737294197083, "learning_rate": 0.0005, "loss": 1.3933653831481934, "step": 24044 }, { "epoch": 1.8955459203783995, "grad_norm": 0.5073503255844116, "learning_rate": 0.0005, "loss": 1.2970821857452393, "step": 24045 }, { "epoch": 1.8956247536460387, "grad_norm": 0.5835318565368652, "learning_rate": 0.0005, "loss": 1.3675215244293213, "step": 24046 }, { "epoch": 1.8957035869136776, "grad_norm": 0.5002620220184326, "learning_rate": 0.0005, "loss": 1.4452018737792969, "step": 24047 }, { "epoch": 1.8957824201813165, "grad_norm": 0.601768434047699, "learning_rate": 0.0005, "loss": 1.4753098487854004, "step": 24048 }, { "epoch": 1.8958612534489556, "grad_norm": 0.5419475436210632, "learning_rate": 0.0005, "loss": 1.5282976627349854, "step": 24049 }, { "epoch": 1.8959400867165943, "grad_norm": 0.49869975447654724, "learning_rate": 0.0005, "loss": 1.2633638381958008, "step": 24050 }, { "epoch": 1.8960189199842334, "grad_norm": 0.4728391170501709, "learning_rate": 0.0005, "loss": 1.174152135848999, "step": 24051 }, { "epoch": 1.8960977532518721, "grad_norm": 0.6055663228034973, "learning_rate": 0.0005, "loss": 1.4019821882247925, "step": 24052 }, { "epoch": 1.8961765865195113, "grad_norm": 0.6739127039909363, "learning_rate": 0.0005, "loss": 1.5538111925125122, "step": 24053 }, { "epoch": 1.8962554197871502, "grad_norm": 0.6413898468017578, "learning_rate": 0.0005, "loss": 1.428091049194336, "step": 24054 }, { "epoch": 1.896334253054789, "grad_norm": 0.6087861657142639, "learning_rate": 0.0005, "loss": 1.553602933883667, "step": 24055 }, { "epoch": 1.8964130863224282, "grad_norm": 0.5091407299041748, "learning_rate": 0.0005, "loss": 1.2595850229263306, "step": 24056 }, { "epoch": 1.896491919590067, "grad_norm": 0.44786298274993896, "learning_rate": 0.0005, "loss": 1.2559890747070312, "step": 24057 }, { "epoch": 1.896570752857706, "grad_norm": 0.4649122357368469, "learning_rate": 0.0005, "loss": 1.4502228498458862, "step": 24058 }, { "epoch": 1.8966495861253447, "grad_norm": 0.5528388619422913, "learning_rate": 0.0005, "loss": 1.388331651687622, "step": 24059 }, { "epoch": 1.8967284193929839, "grad_norm": 0.5337737202644348, "learning_rate": 0.0005, "loss": 1.788862705230713, "step": 24060 }, { "epoch": 1.8968072526606228, "grad_norm": 0.5356422066688538, "learning_rate": 0.0005, "loss": 1.386509656906128, "step": 24061 }, { "epoch": 1.8968860859282617, "grad_norm": 0.5460609793663025, "learning_rate": 0.0005, "loss": 1.5426661968231201, "step": 24062 }, { "epoch": 1.8969649191959008, "grad_norm": 0.46871647238731384, "learning_rate": 0.0005, "loss": 1.4868322610855103, "step": 24063 }, { "epoch": 1.8970437524635395, "grad_norm": 0.4382016658782959, "learning_rate": 0.0005, "loss": 1.1223926544189453, "step": 24064 }, { "epoch": 1.8971225857311786, "grad_norm": 0.4272528290748596, "learning_rate": 0.0005, "loss": 1.2568345069885254, "step": 24065 }, { "epoch": 1.8972014189988173, "grad_norm": 0.5166786909103394, "learning_rate": 0.0005, "loss": 1.4373246431350708, "step": 24066 }, { "epoch": 1.8972802522664565, "grad_norm": 0.513899564743042, "learning_rate": 0.0005, "loss": 1.4543968439102173, "step": 24067 }, { "epoch": 1.8973590855340954, "grad_norm": 0.5106244087219238, "learning_rate": 0.0005, "loss": 1.5622929334640503, "step": 24068 }, { "epoch": 1.8974379188017343, "grad_norm": 0.44603049755096436, "learning_rate": 0.0005, "loss": 1.3441808223724365, "step": 24069 }, { "epoch": 1.8975167520693734, "grad_norm": 0.5024532079696655, "learning_rate": 0.0005, "loss": 1.399125099182129, "step": 24070 }, { "epoch": 1.8975955853370121, "grad_norm": 0.4533578157424927, "learning_rate": 0.0005, "loss": 1.3391469717025757, "step": 24071 }, { "epoch": 1.8976744186046512, "grad_norm": 0.6266641020774841, "learning_rate": 0.0005, "loss": 1.3563793897628784, "step": 24072 }, { "epoch": 1.8977532518722902, "grad_norm": 0.5345409512519836, "learning_rate": 0.0005, "loss": 1.2507704496383667, "step": 24073 }, { "epoch": 1.897832085139929, "grad_norm": 0.47614586353302, "learning_rate": 0.0005, "loss": 1.3198418617248535, "step": 24074 }, { "epoch": 1.897910918407568, "grad_norm": 0.5096871852874756, "learning_rate": 0.0005, "loss": 1.6696184873580933, "step": 24075 }, { "epoch": 1.897989751675207, "grad_norm": 0.5700991153717041, "learning_rate": 0.0005, "loss": 1.2909209728240967, "step": 24076 }, { "epoch": 1.898068584942846, "grad_norm": 0.5908649563789368, "learning_rate": 0.0005, "loss": 1.5334177017211914, "step": 24077 }, { "epoch": 1.8981474182104847, "grad_norm": 0.6427224278450012, "learning_rate": 0.0005, "loss": 1.441955327987671, "step": 24078 }, { "epoch": 1.8982262514781238, "grad_norm": 0.6405501961708069, "learning_rate": 0.0005, "loss": 1.6769890785217285, "step": 24079 }, { "epoch": 1.8983050847457628, "grad_norm": 0.6032131910324097, "learning_rate": 0.0005, "loss": 1.6398518085479736, "step": 24080 }, { "epoch": 1.8983839180134017, "grad_norm": 0.6471253037452698, "learning_rate": 0.0005, "loss": 1.5031278133392334, "step": 24081 }, { "epoch": 1.8984627512810406, "grad_norm": 0.4829252362251282, "learning_rate": 0.0005, "loss": 1.1702384948730469, "step": 24082 }, { "epoch": 1.8985415845486795, "grad_norm": 0.5201511979103088, "learning_rate": 0.0005, "loss": 1.5672407150268555, "step": 24083 }, { "epoch": 1.8986204178163186, "grad_norm": 0.5376798510551453, "learning_rate": 0.0005, "loss": 1.3787827491760254, "step": 24084 }, { "epoch": 1.8986992510839573, "grad_norm": 0.5148823261260986, "learning_rate": 0.0005, "loss": 1.5585604906082153, "step": 24085 }, { "epoch": 1.8987780843515965, "grad_norm": 0.5600709915161133, "learning_rate": 0.0005, "loss": 1.410808801651001, "step": 24086 }, { "epoch": 1.8988569176192354, "grad_norm": 0.4951234757900238, "learning_rate": 0.0005, "loss": 1.4385125637054443, "step": 24087 }, { "epoch": 1.8989357508868743, "grad_norm": 0.4071592688560486, "learning_rate": 0.0005, "loss": 1.17533278465271, "step": 24088 }, { "epoch": 1.8990145841545132, "grad_norm": 0.49421048164367676, "learning_rate": 0.0005, "loss": 1.4888054132461548, "step": 24089 }, { "epoch": 1.899093417422152, "grad_norm": 0.6942282915115356, "learning_rate": 0.0005, "loss": 1.5608733892440796, "step": 24090 }, { "epoch": 1.8991722506897912, "grad_norm": 0.5275148749351501, "learning_rate": 0.0005, "loss": 1.3057353496551514, "step": 24091 }, { "epoch": 1.89925108395743, "grad_norm": 0.5199688076972961, "learning_rate": 0.0005, "loss": 1.6727290153503418, "step": 24092 }, { "epoch": 1.899329917225069, "grad_norm": 0.4226379692554474, "learning_rate": 0.0005, "loss": 1.2739626169204712, "step": 24093 }, { "epoch": 1.899408750492708, "grad_norm": 0.46918347477912903, "learning_rate": 0.0005, "loss": 1.4132654666900635, "step": 24094 }, { "epoch": 1.8994875837603469, "grad_norm": 0.5215540528297424, "learning_rate": 0.0005, "loss": 1.4640462398529053, "step": 24095 }, { "epoch": 1.8995664170279858, "grad_norm": 0.5165569186210632, "learning_rate": 0.0005, "loss": 1.4227477312088013, "step": 24096 }, { "epoch": 1.8996452502956247, "grad_norm": 0.8527712225914001, "learning_rate": 0.0005, "loss": 1.286385178565979, "step": 24097 }, { "epoch": 1.8997240835632638, "grad_norm": 0.4895016551017761, "learning_rate": 0.0005, "loss": 1.507005214691162, "step": 24098 }, { "epoch": 1.8998029168309025, "grad_norm": 0.789887011051178, "learning_rate": 0.0005, "loss": 1.8076437711715698, "step": 24099 }, { "epoch": 1.8998817500985417, "grad_norm": 0.6190564036369324, "learning_rate": 0.0005, "loss": 1.391403317451477, "step": 24100 }, { "epoch": 1.8999605833661806, "grad_norm": 0.6526584625244141, "learning_rate": 0.0005, "loss": 1.3182475566864014, "step": 24101 }, { "epoch": 1.9000394166338195, "grad_norm": 0.48861461877822876, "learning_rate": 0.0005, "loss": 1.4078389406204224, "step": 24102 }, { "epoch": 1.9001182499014584, "grad_norm": 0.48237207531929016, "learning_rate": 0.0005, "loss": 1.2399249076843262, "step": 24103 }, { "epoch": 1.9001970831690973, "grad_norm": 0.533789336681366, "learning_rate": 0.0005, "loss": 1.2541723251342773, "step": 24104 }, { "epoch": 1.9002759164367364, "grad_norm": 0.462827205657959, "learning_rate": 0.0005, "loss": 1.2027575969696045, "step": 24105 }, { "epoch": 1.9003547497043751, "grad_norm": 0.5811426639556885, "learning_rate": 0.0005, "loss": 1.4656846523284912, "step": 24106 }, { "epoch": 1.9004335829720143, "grad_norm": 0.6428518295288086, "learning_rate": 0.0005, "loss": 1.5715432167053223, "step": 24107 }, { "epoch": 1.9005124162396532, "grad_norm": 0.534451425075531, "learning_rate": 0.0005, "loss": 1.4946694374084473, "step": 24108 }, { "epoch": 1.900591249507292, "grad_norm": 0.5913828015327454, "learning_rate": 0.0005, "loss": 1.2600841522216797, "step": 24109 }, { "epoch": 1.900670082774931, "grad_norm": 0.4475043714046478, "learning_rate": 0.0005, "loss": 1.3500306606292725, "step": 24110 }, { "epoch": 1.90074891604257, "grad_norm": 0.48440366983413696, "learning_rate": 0.0005, "loss": 1.3041492700576782, "step": 24111 }, { "epoch": 1.900827749310209, "grad_norm": 0.5762292146682739, "learning_rate": 0.0005, "loss": 1.4703236818313599, "step": 24112 }, { "epoch": 1.9009065825778477, "grad_norm": 0.46618184447288513, "learning_rate": 0.0005, "loss": 1.4290225505828857, "step": 24113 }, { "epoch": 1.9009854158454869, "grad_norm": 0.42200878262519836, "learning_rate": 0.0005, "loss": 1.2759664058685303, "step": 24114 }, { "epoch": 1.9010642491131258, "grad_norm": 0.6212743520736694, "learning_rate": 0.0005, "loss": 1.554560661315918, "step": 24115 }, { "epoch": 1.9011430823807647, "grad_norm": 0.4852451682090759, "learning_rate": 0.0005, "loss": 1.3101685047149658, "step": 24116 }, { "epoch": 1.9012219156484036, "grad_norm": 0.5037696361541748, "learning_rate": 0.0005, "loss": 1.633332371711731, "step": 24117 }, { "epoch": 1.9013007489160425, "grad_norm": 0.6424763798713684, "learning_rate": 0.0005, "loss": 1.4634597301483154, "step": 24118 }, { "epoch": 1.9013795821836816, "grad_norm": 0.4674868583679199, "learning_rate": 0.0005, "loss": 1.2824140787124634, "step": 24119 }, { "epoch": 1.9014584154513203, "grad_norm": 0.605269193649292, "learning_rate": 0.0005, "loss": 1.434227705001831, "step": 24120 }, { "epoch": 1.9015372487189595, "grad_norm": 0.5243204832077026, "learning_rate": 0.0005, "loss": 1.330427646636963, "step": 24121 }, { "epoch": 1.9016160819865984, "grad_norm": 0.5158457159996033, "learning_rate": 0.0005, "loss": 1.47892165184021, "step": 24122 }, { "epoch": 1.9016949152542373, "grad_norm": 0.6120995283126831, "learning_rate": 0.0005, "loss": 1.531685471534729, "step": 24123 }, { "epoch": 1.9017737485218762, "grad_norm": 0.6543110013008118, "learning_rate": 0.0005, "loss": 1.4205936193466187, "step": 24124 }, { "epoch": 1.901852581789515, "grad_norm": 0.684709906578064, "learning_rate": 0.0005, "loss": 1.4714689254760742, "step": 24125 }, { "epoch": 1.9019314150571542, "grad_norm": 0.5626693964004517, "learning_rate": 0.0005, "loss": 1.330406904220581, "step": 24126 }, { "epoch": 1.902010248324793, "grad_norm": 0.4969647526741028, "learning_rate": 0.0005, "loss": 1.4444855451583862, "step": 24127 }, { "epoch": 1.902089081592432, "grad_norm": 0.5637059211730957, "learning_rate": 0.0005, "loss": 1.5109634399414062, "step": 24128 }, { "epoch": 1.902167914860071, "grad_norm": 0.8876002430915833, "learning_rate": 0.0005, "loss": 1.3124141693115234, "step": 24129 }, { "epoch": 1.9022467481277099, "grad_norm": 0.7259595990180969, "learning_rate": 0.0005, "loss": 1.4022468328475952, "step": 24130 }, { "epoch": 1.9023255813953488, "grad_norm": 0.6541531085968018, "learning_rate": 0.0005, "loss": 1.3455638885498047, "step": 24131 }, { "epoch": 1.9024044146629877, "grad_norm": 0.5199626684188843, "learning_rate": 0.0005, "loss": 1.4418964385986328, "step": 24132 }, { "epoch": 1.9024832479306268, "grad_norm": 0.5764614343643188, "learning_rate": 0.0005, "loss": 1.473543643951416, "step": 24133 }, { "epoch": 1.9025620811982655, "grad_norm": 0.5705220103263855, "learning_rate": 0.0005, "loss": 1.4576292037963867, "step": 24134 }, { "epoch": 1.9026409144659047, "grad_norm": 0.5298311710357666, "learning_rate": 0.0005, "loss": 1.5199151039123535, "step": 24135 }, { "epoch": 1.9027197477335436, "grad_norm": 0.5402519702911377, "learning_rate": 0.0005, "loss": 1.263251543045044, "step": 24136 }, { "epoch": 1.9027985810011825, "grad_norm": 0.5225520133972168, "learning_rate": 0.0005, "loss": 1.2926729917526245, "step": 24137 }, { "epoch": 1.9028774142688214, "grad_norm": 0.7538648843765259, "learning_rate": 0.0005, "loss": 1.6475118398666382, "step": 24138 }, { "epoch": 1.9029562475364603, "grad_norm": 0.6153720021247864, "learning_rate": 0.0005, "loss": 1.5985782146453857, "step": 24139 }, { "epoch": 1.9030350808040994, "grad_norm": 0.7853623032569885, "learning_rate": 0.0005, "loss": 1.6429650783538818, "step": 24140 }, { "epoch": 1.9031139140717381, "grad_norm": 0.6974040865898132, "learning_rate": 0.0005, "loss": 1.4926449060440063, "step": 24141 }, { "epoch": 1.9031927473393773, "grad_norm": 0.5627703070640564, "learning_rate": 0.0005, "loss": 1.4329590797424316, "step": 24142 }, { "epoch": 1.9032715806070162, "grad_norm": 0.5220901370048523, "learning_rate": 0.0005, "loss": 1.3961058855056763, "step": 24143 }, { "epoch": 1.903350413874655, "grad_norm": 0.5322819352149963, "learning_rate": 0.0005, "loss": 1.4222553968429565, "step": 24144 }, { "epoch": 1.903429247142294, "grad_norm": 0.6216225028038025, "learning_rate": 0.0005, "loss": 1.3930972814559937, "step": 24145 }, { "epoch": 1.903508080409933, "grad_norm": 0.5679385662078857, "learning_rate": 0.0005, "loss": 1.3887345790863037, "step": 24146 }, { "epoch": 1.903586913677572, "grad_norm": 0.5207493901252747, "learning_rate": 0.0005, "loss": 1.3381245136260986, "step": 24147 }, { "epoch": 1.9036657469452107, "grad_norm": 0.47389864921569824, "learning_rate": 0.0005, "loss": 1.2872109413146973, "step": 24148 }, { "epoch": 1.9037445802128499, "grad_norm": 0.6591587066650391, "learning_rate": 0.0005, "loss": 1.441314935684204, "step": 24149 }, { "epoch": 1.9038234134804888, "grad_norm": 0.6241520643234253, "learning_rate": 0.0005, "loss": 1.5400371551513672, "step": 24150 }, { "epoch": 1.9039022467481277, "grad_norm": 0.6244001984596252, "learning_rate": 0.0005, "loss": 1.287581443786621, "step": 24151 }, { "epoch": 1.9039810800157666, "grad_norm": 0.6020779013633728, "learning_rate": 0.0005, "loss": 1.4456464052200317, "step": 24152 }, { "epoch": 1.9040599132834055, "grad_norm": 0.45437437295913696, "learning_rate": 0.0005, "loss": 1.3730484247207642, "step": 24153 }, { "epoch": 1.9041387465510446, "grad_norm": 0.6018922924995422, "learning_rate": 0.0005, "loss": 1.606943964958191, "step": 24154 }, { "epoch": 1.9042175798186833, "grad_norm": 0.43583300709724426, "learning_rate": 0.0005, "loss": 1.184483289718628, "step": 24155 }, { "epoch": 1.9042964130863225, "grad_norm": 0.6054439544677734, "learning_rate": 0.0005, "loss": 1.5494260787963867, "step": 24156 }, { "epoch": 1.9043752463539614, "grad_norm": 0.5326022505760193, "learning_rate": 0.0005, "loss": 1.3357959985733032, "step": 24157 }, { "epoch": 1.9044540796216003, "grad_norm": 0.43804508447647095, "learning_rate": 0.0005, "loss": 1.5165770053863525, "step": 24158 }, { "epoch": 1.9045329128892392, "grad_norm": 0.43655890226364136, "learning_rate": 0.0005, "loss": 1.092180848121643, "step": 24159 }, { "epoch": 1.9046117461568781, "grad_norm": 0.5910447239875793, "learning_rate": 0.0005, "loss": 1.4248037338256836, "step": 24160 }, { "epoch": 1.9046905794245172, "grad_norm": 0.5182461142539978, "learning_rate": 0.0005, "loss": 1.4067178964614868, "step": 24161 }, { "epoch": 1.904769412692156, "grad_norm": 0.6828846335411072, "learning_rate": 0.0005, "loss": 1.4813811779022217, "step": 24162 }, { "epoch": 1.904848245959795, "grad_norm": 0.6536094546318054, "learning_rate": 0.0005, "loss": 1.547763705253601, "step": 24163 }, { "epoch": 1.904927079227434, "grad_norm": 0.4699629843235016, "learning_rate": 0.0005, "loss": 1.4332268238067627, "step": 24164 }, { "epoch": 1.905005912495073, "grad_norm": 0.6409323215484619, "learning_rate": 0.0005, "loss": 1.7844910621643066, "step": 24165 }, { "epoch": 1.905084745762712, "grad_norm": 0.6014025211334229, "learning_rate": 0.0005, "loss": 1.636444091796875, "step": 24166 }, { "epoch": 1.9051635790303507, "grad_norm": 0.5244306325912476, "learning_rate": 0.0005, "loss": 1.5483202934265137, "step": 24167 }, { "epoch": 1.9052424122979899, "grad_norm": 0.550290584564209, "learning_rate": 0.0005, "loss": 1.3120241165161133, "step": 24168 }, { "epoch": 1.9053212455656285, "grad_norm": 0.5201104283332825, "learning_rate": 0.0005, "loss": 1.40083646774292, "step": 24169 }, { "epoch": 1.9054000788332677, "grad_norm": 0.5356069207191467, "learning_rate": 0.0005, "loss": 1.3655650615692139, "step": 24170 }, { "epoch": 1.9054789121009066, "grad_norm": 0.6118682026863098, "learning_rate": 0.0005, "loss": 1.4561891555786133, "step": 24171 }, { "epoch": 1.9055577453685455, "grad_norm": 0.6479302048683167, "learning_rate": 0.0005, "loss": 1.68260657787323, "step": 24172 }, { "epoch": 1.9056365786361846, "grad_norm": 0.4908037483692169, "learning_rate": 0.0005, "loss": 1.3331165313720703, "step": 24173 }, { "epoch": 1.9057154119038233, "grad_norm": 0.5000961422920227, "learning_rate": 0.0005, "loss": 1.4965505599975586, "step": 24174 }, { "epoch": 1.9057942451714625, "grad_norm": 0.5823988914489746, "learning_rate": 0.0005, "loss": 1.600165605545044, "step": 24175 }, { "epoch": 1.9058730784391011, "grad_norm": 0.5274562239646912, "learning_rate": 0.0005, "loss": 1.3749022483825684, "step": 24176 }, { "epoch": 1.9059519117067403, "grad_norm": 0.48119598627090454, "learning_rate": 0.0005, "loss": 1.241206407546997, "step": 24177 }, { "epoch": 1.9060307449743792, "grad_norm": 0.6030813455581665, "learning_rate": 0.0005, "loss": 1.7433722019195557, "step": 24178 }, { "epoch": 1.906109578242018, "grad_norm": 0.5302752256393433, "learning_rate": 0.0005, "loss": 1.3136940002441406, "step": 24179 }, { "epoch": 1.9061884115096572, "grad_norm": 0.609930157661438, "learning_rate": 0.0005, "loss": 1.512592077255249, "step": 24180 }, { "epoch": 1.906267244777296, "grad_norm": 0.6421342492103577, "learning_rate": 0.0005, "loss": 1.480590581893921, "step": 24181 }, { "epoch": 1.906346078044935, "grad_norm": 0.6270186901092529, "learning_rate": 0.0005, "loss": 1.2561702728271484, "step": 24182 }, { "epoch": 1.9064249113125737, "grad_norm": 0.5486481189727783, "learning_rate": 0.0005, "loss": 1.4159976243972778, "step": 24183 }, { "epoch": 1.9065037445802129, "grad_norm": 0.5187151432037354, "learning_rate": 0.0005, "loss": 1.4539273977279663, "step": 24184 }, { "epoch": 1.9065825778478518, "grad_norm": 0.49551695585250854, "learning_rate": 0.0005, "loss": 1.3583829402923584, "step": 24185 }, { "epoch": 1.9066614111154907, "grad_norm": 0.645971953868866, "learning_rate": 0.0005, "loss": 1.5115439891815186, "step": 24186 }, { "epoch": 1.9067402443831298, "grad_norm": 0.698934018611908, "learning_rate": 0.0005, "loss": 1.5218877792358398, "step": 24187 }, { "epoch": 1.9068190776507685, "grad_norm": 0.4264519512653351, "learning_rate": 0.0005, "loss": 1.311238169670105, "step": 24188 }, { "epoch": 1.9068979109184077, "grad_norm": 0.6789985299110413, "learning_rate": 0.0005, "loss": 1.397465467453003, "step": 24189 }, { "epoch": 1.9069767441860463, "grad_norm": 0.8462187647819519, "learning_rate": 0.0005, "loss": 1.6839642524719238, "step": 24190 }, { "epoch": 1.9070555774536855, "grad_norm": 0.4542976915836334, "learning_rate": 0.0005, "loss": 1.2550243139266968, "step": 24191 }, { "epoch": 1.9071344107213244, "grad_norm": 0.6405741572380066, "learning_rate": 0.0005, "loss": 1.4631563425064087, "step": 24192 }, { "epoch": 1.9072132439889633, "grad_norm": 0.4409458041191101, "learning_rate": 0.0005, "loss": 1.2447295188903809, "step": 24193 }, { "epoch": 1.9072920772566024, "grad_norm": 0.6132073998451233, "learning_rate": 0.0005, "loss": 1.3927371501922607, "step": 24194 }, { "epoch": 1.9073709105242411, "grad_norm": 0.583335816860199, "learning_rate": 0.0005, "loss": 1.288894534111023, "step": 24195 }, { "epoch": 1.9074497437918803, "grad_norm": 0.5162605047225952, "learning_rate": 0.0005, "loss": 1.4415991306304932, "step": 24196 }, { "epoch": 1.9075285770595192, "grad_norm": 0.5658568739891052, "learning_rate": 0.0005, "loss": 1.3631706237792969, "step": 24197 }, { "epoch": 1.907607410327158, "grad_norm": 0.4163399040699005, "learning_rate": 0.0005, "loss": 1.2193320989608765, "step": 24198 }, { "epoch": 1.907686243594797, "grad_norm": 0.41645586490631104, "learning_rate": 0.0005, "loss": 1.2036633491516113, "step": 24199 }, { "epoch": 1.907765076862436, "grad_norm": 0.488432377576828, "learning_rate": 0.0005, "loss": 1.4110674858093262, "step": 24200 }, { "epoch": 1.907843910130075, "grad_norm": 0.5677105188369751, "learning_rate": 0.0005, "loss": 1.389165997505188, "step": 24201 }, { "epoch": 1.9079227433977137, "grad_norm": 0.6850674152374268, "learning_rate": 0.0005, "loss": 1.4893889427185059, "step": 24202 }, { "epoch": 1.9080015766653529, "grad_norm": 0.44888201355934143, "learning_rate": 0.0005, "loss": 1.5047035217285156, "step": 24203 }, { "epoch": 1.9080804099329918, "grad_norm": 0.5368985533714294, "learning_rate": 0.0005, "loss": 1.419827938079834, "step": 24204 }, { "epoch": 1.9081592432006307, "grad_norm": 0.5607704520225525, "learning_rate": 0.0005, "loss": 1.3150222301483154, "step": 24205 }, { "epoch": 1.9082380764682696, "grad_norm": 0.4774921238422394, "learning_rate": 0.0005, "loss": 1.2372992038726807, "step": 24206 }, { "epoch": 1.9083169097359085, "grad_norm": 0.5607859492301941, "learning_rate": 0.0005, "loss": 1.1896194219589233, "step": 24207 }, { "epoch": 1.9083957430035476, "grad_norm": 0.4510273337364197, "learning_rate": 0.0005, "loss": 1.5077276229858398, "step": 24208 }, { "epoch": 1.9084745762711863, "grad_norm": 1.0908573865890503, "learning_rate": 0.0005, "loss": 1.4494845867156982, "step": 24209 }, { "epoch": 1.9085534095388255, "grad_norm": 0.5803927779197693, "learning_rate": 0.0005, "loss": 1.5096890926361084, "step": 24210 }, { "epoch": 1.9086322428064644, "grad_norm": 0.6287021636962891, "learning_rate": 0.0005, "loss": 1.3884174823760986, "step": 24211 }, { "epoch": 1.9087110760741033, "grad_norm": 0.6923620700836182, "learning_rate": 0.0005, "loss": 1.6851966381072998, "step": 24212 }, { "epoch": 1.9087899093417422, "grad_norm": 0.6427436470985413, "learning_rate": 0.0005, "loss": 1.7263166904449463, "step": 24213 }, { "epoch": 1.908868742609381, "grad_norm": 0.6790623664855957, "learning_rate": 0.0005, "loss": 1.5332964658737183, "step": 24214 }, { "epoch": 1.9089475758770202, "grad_norm": 0.5385189652442932, "learning_rate": 0.0005, "loss": 1.4478003978729248, "step": 24215 }, { "epoch": 1.909026409144659, "grad_norm": 0.5905685424804688, "learning_rate": 0.0005, "loss": 1.2962026596069336, "step": 24216 }, { "epoch": 1.909105242412298, "grad_norm": 0.5529810190200806, "learning_rate": 0.0005, "loss": 1.413771629333496, "step": 24217 }, { "epoch": 1.909184075679937, "grad_norm": 0.6378756761550903, "learning_rate": 0.0005, "loss": 1.4431012868881226, "step": 24218 }, { "epoch": 1.9092629089475759, "grad_norm": 0.5572518110275269, "learning_rate": 0.0005, "loss": 1.3550187349319458, "step": 24219 }, { "epoch": 1.9093417422152148, "grad_norm": 0.5137205123901367, "learning_rate": 0.0005, "loss": 1.4710556268692017, "step": 24220 }, { "epoch": 1.9094205754828537, "grad_norm": 0.5184030532836914, "learning_rate": 0.0005, "loss": 1.3400044441223145, "step": 24221 }, { "epoch": 1.9094994087504928, "grad_norm": 0.4120320677757263, "learning_rate": 0.0005, "loss": 1.318865180015564, "step": 24222 }, { "epoch": 1.9095782420181315, "grad_norm": 0.5786510109901428, "learning_rate": 0.0005, "loss": 1.28070068359375, "step": 24223 }, { "epoch": 1.9096570752857707, "grad_norm": 0.5272017121315002, "learning_rate": 0.0005, "loss": 1.4412521123886108, "step": 24224 }, { "epoch": 1.9097359085534096, "grad_norm": 0.4787392020225525, "learning_rate": 0.0005, "loss": 1.2263890504837036, "step": 24225 }, { "epoch": 1.9098147418210485, "grad_norm": 0.6633605360984802, "learning_rate": 0.0005, "loss": 1.3244541883468628, "step": 24226 }, { "epoch": 1.9098935750886874, "grad_norm": 0.5069360733032227, "learning_rate": 0.0005, "loss": 1.5924105644226074, "step": 24227 }, { "epoch": 1.9099724083563263, "grad_norm": 0.4830150902271271, "learning_rate": 0.0005, "loss": 1.405248999595642, "step": 24228 }, { "epoch": 1.9100512416239654, "grad_norm": 0.6147918701171875, "learning_rate": 0.0005, "loss": 1.5905125141143799, "step": 24229 }, { "epoch": 1.9101300748916041, "grad_norm": 0.56755530834198, "learning_rate": 0.0005, "loss": 1.7062170505523682, "step": 24230 }, { "epoch": 1.9102089081592433, "grad_norm": 0.5100781321525574, "learning_rate": 0.0005, "loss": 1.3071792125701904, "step": 24231 }, { "epoch": 1.9102877414268822, "grad_norm": 0.4929361045360565, "learning_rate": 0.0005, "loss": 1.4068846702575684, "step": 24232 }, { "epoch": 1.910366574694521, "grad_norm": 0.5791650414466858, "learning_rate": 0.0005, "loss": 1.507542371749878, "step": 24233 }, { "epoch": 1.91044540796216, "grad_norm": 0.496008962392807, "learning_rate": 0.0005, "loss": 1.3431731462478638, "step": 24234 }, { "epoch": 1.910524241229799, "grad_norm": 0.5647189617156982, "learning_rate": 0.0005, "loss": 1.3196516036987305, "step": 24235 }, { "epoch": 1.910603074497438, "grad_norm": 0.521544873714447, "learning_rate": 0.0005, "loss": 1.3806430101394653, "step": 24236 }, { "epoch": 1.9106819077650767, "grad_norm": 0.5211564898490906, "learning_rate": 0.0005, "loss": 1.287386417388916, "step": 24237 }, { "epoch": 1.9107607410327159, "grad_norm": 0.6232188940048218, "learning_rate": 0.0005, "loss": 1.454309344291687, "step": 24238 }, { "epoch": 1.9108395743003548, "grad_norm": 0.47062185406684875, "learning_rate": 0.0005, "loss": 1.3202176094055176, "step": 24239 }, { "epoch": 1.9109184075679937, "grad_norm": 0.4927828907966614, "learning_rate": 0.0005, "loss": 1.2517712116241455, "step": 24240 }, { "epoch": 1.9109972408356326, "grad_norm": 1.9967751502990723, "learning_rate": 0.0005, "loss": 1.2520434856414795, "step": 24241 }, { "epoch": 1.9110760741032715, "grad_norm": 0.48660144209861755, "learning_rate": 0.0005, "loss": 1.4481697082519531, "step": 24242 }, { "epoch": 1.9111549073709106, "grad_norm": 0.49448105692863464, "learning_rate": 0.0005, "loss": 1.2196967601776123, "step": 24243 }, { "epoch": 1.9112337406385493, "grad_norm": 0.490906685590744, "learning_rate": 0.0005, "loss": 1.557420253753662, "step": 24244 }, { "epoch": 1.9113125739061885, "grad_norm": 11.974090576171875, "learning_rate": 0.0005, "loss": 1.3118596076965332, "step": 24245 }, { "epoch": 1.9113914071738274, "grad_norm": 0.6383022665977478, "learning_rate": 0.0005, "loss": 1.6145236492156982, "step": 24246 }, { "epoch": 1.9114702404414663, "grad_norm": 0.46001219749450684, "learning_rate": 0.0005, "loss": 1.3980016708374023, "step": 24247 }, { "epoch": 1.9115490737091052, "grad_norm": 0.5607569217681885, "learning_rate": 0.0005, "loss": 1.4502798318862915, "step": 24248 }, { "epoch": 1.9116279069767441, "grad_norm": 0.4820522665977478, "learning_rate": 0.0005, "loss": 1.4503146409988403, "step": 24249 }, { "epoch": 1.9117067402443833, "grad_norm": 0.5961846709251404, "learning_rate": 0.0005, "loss": 1.629238247871399, "step": 24250 }, { "epoch": 1.911785573512022, "grad_norm": 0.5669131875038147, "learning_rate": 0.0005, "loss": 1.5700982809066772, "step": 24251 }, { "epoch": 1.911864406779661, "grad_norm": 0.6659059524536133, "learning_rate": 0.0005, "loss": 1.3431029319763184, "step": 24252 }, { "epoch": 1.9119432400473, "grad_norm": 0.6026536822319031, "learning_rate": 0.0005, "loss": 1.4330976009368896, "step": 24253 }, { "epoch": 1.912022073314939, "grad_norm": 0.6763783693313599, "learning_rate": 0.0005, "loss": 1.490582823753357, "step": 24254 }, { "epoch": 1.9121009065825778, "grad_norm": 0.46082672476768494, "learning_rate": 0.0005, "loss": 1.398747205734253, "step": 24255 }, { "epoch": 1.9121797398502167, "grad_norm": 0.48526155948638916, "learning_rate": 0.0005, "loss": 1.36368989944458, "step": 24256 }, { "epoch": 1.9122585731178559, "grad_norm": 0.4589354693889618, "learning_rate": 0.0005, "loss": 1.315187931060791, "step": 24257 }, { "epoch": 1.9123374063854945, "grad_norm": 0.724831223487854, "learning_rate": 0.0005, "loss": 1.4892328977584839, "step": 24258 }, { "epoch": 1.9124162396531337, "grad_norm": 0.6440284252166748, "learning_rate": 0.0005, "loss": 1.4936306476593018, "step": 24259 }, { "epoch": 1.9124950729207726, "grad_norm": 0.46814414858818054, "learning_rate": 0.0005, "loss": 1.5817002058029175, "step": 24260 }, { "epoch": 1.9125739061884115, "grad_norm": 0.5196642875671387, "learning_rate": 0.0005, "loss": 1.2621088027954102, "step": 24261 }, { "epoch": 1.9126527394560504, "grad_norm": 0.6362979412078857, "learning_rate": 0.0005, "loss": 1.4354212284088135, "step": 24262 }, { "epoch": 1.9127315727236893, "grad_norm": 0.5413639545440674, "learning_rate": 0.0005, "loss": 1.3169760704040527, "step": 24263 }, { "epoch": 1.9128104059913285, "grad_norm": 0.5504941940307617, "learning_rate": 0.0005, "loss": 1.389962077140808, "step": 24264 }, { "epoch": 1.9128892392589671, "grad_norm": 0.4516962468624115, "learning_rate": 0.0005, "loss": 1.1662750244140625, "step": 24265 }, { "epoch": 1.9129680725266063, "grad_norm": 0.5182923674583435, "learning_rate": 0.0005, "loss": 1.4792766571044922, "step": 24266 }, { "epoch": 1.9130469057942452, "grad_norm": 0.5581590533256531, "learning_rate": 0.0005, "loss": 1.4065985679626465, "step": 24267 }, { "epoch": 1.913125739061884, "grad_norm": 0.9370785355567932, "learning_rate": 0.0005, "loss": 1.4308741092681885, "step": 24268 }, { "epoch": 1.913204572329523, "grad_norm": 0.4379395842552185, "learning_rate": 0.0005, "loss": 1.015881061553955, "step": 24269 }, { "epoch": 1.913283405597162, "grad_norm": 0.5388845801353455, "learning_rate": 0.0005, "loss": 1.5425148010253906, "step": 24270 }, { "epoch": 1.913362238864801, "grad_norm": 0.5817891359329224, "learning_rate": 0.0005, "loss": 1.471840739250183, "step": 24271 }, { "epoch": 1.9134410721324397, "grad_norm": 0.9626772403717041, "learning_rate": 0.0005, "loss": 1.4194408655166626, "step": 24272 }, { "epoch": 1.9135199054000789, "grad_norm": 0.6273197531700134, "learning_rate": 0.0005, "loss": 1.350590705871582, "step": 24273 }, { "epoch": 1.9135987386677178, "grad_norm": 0.6048389077186584, "learning_rate": 0.0005, "loss": 1.454946756362915, "step": 24274 }, { "epoch": 1.9136775719353567, "grad_norm": 0.482596755027771, "learning_rate": 0.0005, "loss": 1.3187884092330933, "step": 24275 }, { "epoch": 1.9137564052029956, "grad_norm": 0.46305033564567566, "learning_rate": 0.0005, "loss": 1.3720781803131104, "step": 24276 }, { "epoch": 1.9138352384706345, "grad_norm": 0.49948644638061523, "learning_rate": 0.0005, "loss": 1.638354778289795, "step": 24277 }, { "epoch": 1.9139140717382737, "grad_norm": 0.4334929883480072, "learning_rate": 0.0005, "loss": 1.3754234313964844, "step": 24278 }, { "epoch": 1.9139929050059123, "grad_norm": 0.5368766188621521, "learning_rate": 0.0005, "loss": 1.3821871280670166, "step": 24279 }, { "epoch": 1.9140717382735515, "grad_norm": 0.573400616645813, "learning_rate": 0.0005, "loss": 1.2487294673919678, "step": 24280 }, { "epoch": 1.9141505715411904, "grad_norm": 0.5794155597686768, "learning_rate": 0.0005, "loss": 1.4769920110702515, "step": 24281 }, { "epoch": 1.9142294048088293, "grad_norm": 0.6262540817260742, "learning_rate": 0.0005, "loss": 1.2707923650741577, "step": 24282 }, { "epoch": 1.9143082380764682, "grad_norm": 0.5473018288612366, "learning_rate": 0.0005, "loss": 1.2703185081481934, "step": 24283 }, { "epoch": 1.9143870713441071, "grad_norm": 0.4974385201931, "learning_rate": 0.0005, "loss": 1.4248214960098267, "step": 24284 }, { "epoch": 1.9144659046117463, "grad_norm": 0.4442727267742157, "learning_rate": 0.0005, "loss": 1.1990060806274414, "step": 24285 }, { "epoch": 1.914544737879385, "grad_norm": 0.47467461228370667, "learning_rate": 0.0005, "loss": 1.453545331954956, "step": 24286 }, { "epoch": 1.914623571147024, "grad_norm": 0.47268185019493103, "learning_rate": 0.0005, "loss": 1.2086621522903442, "step": 24287 }, { "epoch": 1.914702404414663, "grad_norm": 0.5828182101249695, "learning_rate": 0.0005, "loss": 1.3127003908157349, "step": 24288 }, { "epoch": 1.914781237682302, "grad_norm": 0.44447118043899536, "learning_rate": 0.0005, "loss": 1.2311067581176758, "step": 24289 }, { "epoch": 1.914860070949941, "grad_norm": 0.5475199818611145, "learning_rate": 0.0005, "loss": 1.3035229444503784, "step": 24290 }, { "epoch": 1.9149389042175797, "grad_norm": 0.5839252471923828, "learning_rate": 0.0005, "loss": 1.455306887626648, "step": 24291 }, { "epoch": 1.9150177374852189, "grad_norm": 0.4670107662677765, "learning_rate": 0.0005, "loss": 1.1167149543762207, "step": 24292 }, { "epoch": 1.9150965707528576, "grad_norm": 0.5262584686279297, "learning_rate": 0.0005, "loss": 1.4182058572769165, "step": 24293 }, { "epoch": 1.9151754040204967, "grad_norm": 0.5565694570541382, "learning_rate": 0.0005, "loss": 1.5364329814910889, "step": 24294 }, { "epoch": 1.9152542372881356, "grad_norm": 0.5516197085380554, "learning_rate": 0.0005, "loss": 1.590052843093872, "step": 24295 }, { "epoch": 1.9153330705557745, "grad_norm": 0.5969058871269226, "learning_rate": 0.0005, "loss": 1.3498904705047607, "step": 24296 }, { "epoch": 1.9154119038234136, "grad_norm": 0.5697839260101318, "learning_rate": 0.0005, "loss": 1.3217509984970093, "step": 24297 }, { "epoch": 1.9154907370910523, "grad_norm": 0.47323253750801086, "learning_rate": 0.0005, "loss": 1.479886770248413, "step": 24298 }, { "epoch": 1.9155695703586915, "grad_norm": 0.47699546813964844, "learning_rate": 0.0005, "loss": 1.3483355045318604, "step": 24299 }, { "epoch": 1.9156484036263302, "grad_norm": 0.4695797860622406, "learning_rate": 0.0005, "loss": 1.4067777395248413, "step": 24300 }, { "epoch": 1.9157272368939693, "grad_norm": 0.5637030601501465, "learning_rate": 0.0005, "loss": 1.613060474395752, "step": 24301 }, { "epoch": 1.9158060701616082, "grad_norm": 0.7269173264503479, "learning_rate": 0.0005, "loss": 1.476715087890625, "step": 24302 }, { "epoch": 1.915884903429247, "grad_norm": 0.5081171989440918, "learning_rate": 0.0005, "loss": 1.396267294883728, "step": 24303 }, { "epoch": 1.9159637366968862, "grad_norm": 0.49414464831352234, "learning_rate": 0.0005, "loss": 1.5439647436141968, "step": 24304 }, { "epoch": 1.916042569964525, "grad_norm": 0.6864214539527893, "learning_rate": 0.0005, "loss": 1.4550058841705322, "step": 24305 }, { "epoch": 1.916121403232164, "grad_norm": 0.4549950361251831, "learning_rate": 0.0005, "loss": 1.411124348640442, "step": 24306 }, { "epoch": 1.9162002364998028, "grad_norm": 0.5361577868461609, "learning_rate": 0.0005, "loss": 1.4601631164550781, "step": 24307 }, { "epoch": 1.916279069767442, "grad_norm": 0.5236725211143494, "learning_rate": 0.0005, "loss": 1.528144121170044, "step": 24308 }, { "epoch": 1.9163579030350808, "grad_norm": 0.5547469854354858, "learning_rate": 0.0005, "loss": 1.5211975574493408, "step": 24309 }, { "epoch": 1.9164367363027197, "grad_norm": 0.6623060703277588, "learning_rate": 0.0005, "loss": 1.3767613172531128, "step": 24310 }, { "epoch": 1.9165155695703588, "grad_norm": 0.4852456748485565, "learning_rate": 0.0005, "loss": 1.3414534330368042, "step": 24311 }, { "epoch": 1.9165944028379975, "grad_norm": 0.5068003535270691, "learning_rate": 0.0005, "loss": 1.3505351543426514, "step": 24312 }, { "epoch": 1.9166732361056367, "grad_norm": 0.4638619124889374, "learning_rate": 0.0005, "loss": 1.2114181518554688, "step": 24313 }, { "epoch": 1.9167520693732754, "grad_norm": 0.47558358311653137, "learning_rate": 0.0005, "loss": 1.3624941110610962, "step": 24314 }, { "epoch": 1.9168309026409145, "grad_norm": 0.5615140199661255, "learning_rate": 0.0005, "loss": 1.3411389589309692, "step": 24315 }, { "epoch": 1.9169097359085534, "grad_norm": 0.8030446767807007, "learning_rate": 0.0005, "loss": 1.3023722171783447, "step": 24316 }, { "epoch": 1.9169885691761923, "grad_norm": 0.6073014736175537, "learning_rate": 0.0005, "loss": 1.5690040588378906, "step": 24317 }, { "epoch": 1.9170674024438314, "grad_norm": 0.4943205714225769, "learning_rate": 0.0005, "loss": 1.292672872543335, "step": 24318 }, { "epoch": 1.9171462357114701, "grad_norm": 0.641524612903595, "learning_rate": 0.0005, "loss": 1.4287723302841187, "step": 24319 }, { "epoch": 1.9172250689791093, "grad_norm": 0.5573528409004211, "learning_rate": 0.0005, "loss": 1.2599871158599854, "step": 24320 }, { "epoch": 1.9173039022467482, "grad_norm": 0.4643790125846863, "learning_rate": 0.0005, "loss": 1.3693604469299316, "step": 24321 }, { "epoch": 1.917382735514387, "grad_norm": 0.6123452186584473, "learning_rate": 0.0005, "loss": 1.4764769077301025, "step": 24322 }, { "epoch": 1.917461568782026, "grad_norm": 0.5930224657058716, "learning_rate": 0.0005, "loss": 1.3659791946411133, "step": 24323 }, { "epoch": 1.917540402049665, "grad_norm": 0.6154563426971436, "learning_rate": 0.0005, "loss": 1.597325325012207, "step": 24324 }, { "epoch": 1.917619235317304, "grad_norm": 0.5510326623916626, "learning_rate": 0.0005, "loss": 1.3510396480560303, "step": 24325 }, { "epoch": 1.9176980685849427, "grad_norm": 0.6259074211120605, "learning_rate": 0.0005, "loss": 1.4465147256851196, "step": 24326 }, { "epoch": 1.9177769018525819, "grad_norm": 0.5363280177116394, "learning_rate": 0.0005, "loss": 1.237114429473877, "step": 24327 }, { "epoch": 1.9178557351202208, "grad_norm": 0.5634337663650513, "learning_rate": 0.0005, "loss": 1.5310224294662476, "step": 24328 }, { "epoch": 1.9179345683878597, "grad_norm": 0.6556835770606995, "learning_rate": 0.0005, "loss": 1.5577025413513184, "step": 24329 }, { "epoch": 1.9180134016554986, "grad_norm": 0.4643647372722626, "learning_rate": 0.0005, "loss": 1.0731306076049805, "step": 24330 }, { "epoch": 1.9180922349231375, "grad_norm": 0.6177454590797424, "learning_rate": 0.0005, "loss": 1.5698210000991821, "step": 24331 }, { "epoch": 1.9181710681907767, "grad_norm": 0.5236209034919739, "learning_rate": 0.0005, "loss": 1.595345139503479, "step": 24332 }, { "epoch": 1.9182499014584153, "grad_norm": 0.5308405160903931, "learning_rate": 0.0005, "loss": 1.2766271829605103, "step": 24333 }, { "epoch": 1.9183287347260545, "grad_norm": 0.6442750096321106, "learning_rate": 0.0005, "loss": 1.3948838710784912, "step": 24334 }, { "epoch": 1.9184075679936934, "grad_norm": 0.5044070482254028, "learning_rate": 0.0005, "loss": 1.425093650817871, "step": 24335 }, { "epoch": 1.9184864012613323, "grad_norm": 0.45092490315437317, "learning_rate": 0.0005, "loss": 1.218209981918335, "step": 24336 }, { "epoch": 1.9185652345289712, "grad_norm": 0.5673171877861023, "learning_rate": 0.0005, "loss": 1.3807883262634277, "step": 24337 }, { "epoch": 1.9186440677966101, "grad_norm": 0.5030297040939331, "learning_rate": 0.0005, "loss": 1.2592599391937256, "step": 24338 }, { "epoch": 1.9187229010642493, "grad_norm": 0.4340202510356903, "learning_rate": 0.0005, "loss": 1.3776659965515137, "step": 24339 }, { "epoch": 1.918801734331888, "grad_norm": 0.6242750883102417, "learning_rate": 0.0005, "loss": 1.6070332527160645, "step": 24340 }, { "epoch": 1.918880567599527, "grad_norm": 0.4496705234050751, "learning_rate": 0.0005, "loss": 1.3455322980880737, "step": 24341 }, { "epoch": 1.918959400867166, "grad_norm": 0.5476945042610168, "learning_rate": 0.0005, "loss": 1.4075242280960083, "step": 24342 }, { "epoch": 1.919038234134805, "grad_norm": 0.6048492789268494, "learning_rate": 0.0005, "loss": 1.5724931955337524, "step": 24343 }, { "epoch": 1.9191170674024438, "grad_norm": 0.5401883125305176, "learning_rate": 0.0005, "loss": 1.3773725032806396, "step": 24344 }, { "epoch": 1.9191959006700827, "grad_norm": 0.6176822781562805, "learning_rate": 0.0005, "loss": 1.4027466773986816, "step": 24345 }, { "epoch": 1.9192747339377219, "grad_norm": 0.5980076193809509, "learning_rate": 0.0005, "loss": 1.5782701969146729, "step": 24346 }, { "epoch": 1.9193535672053605, "grad_norm": 0.40624573826789856, "learning_rate": 0.0005, "loss": 1.2366868257522583, "step": 24347 }, { "epoch": 1.9194324004729997, "grad_norm": 0.5929281711578369, "learning_rate": 0.0005, "loss": 1.5299599170684814, "step": 24348 }, { "epoch": 1.9195112337406386, "grad_norm": 0.6124657392501831, "learning_rate": 0.0005, "loss": 1.668517827987671, "step": 24349 }, { "epoch": 1.9195900670082775, "grad_norm": 0.9084593653678894, "learning_rate": 0.0005, "loss": 1.7575223445892334, "step": 24350 }, { "epoch": 1.9196689002759164, "grad_norm": 0.6769436597824097, "learning_rate": 0.0005, "loss": 1.3853285312652588, "step": 24351 }, { "epoch": 1.9197477335435553, "grad_norm": 0.5866071581840515, "learning_rate": 0.0005, "loss": 1.5591601133346558, "step": 24352 }, { "epoch": 1.9198265668111945, "grad_norm": 0.48993465304374695, "learning_rate": 0.0005, "loss": 1.4470000267028809, "step": 24353 }, { "epoch": 1.9199054000788331, "grad_norm": 0.45241355895996094, "learning_rate": 0.0005, "loss": 1.1535327434539795, "step": 24354 }, { "epoch": 1.9199842333464723, "grad_norm": 0.5720622539520264, "learning_rate": 0.0005, "loss": 1.3896236419677734, "step": 24355 }, { "epoch": 1.9200630666141112, "grad_norm": 0.5498647689819336, "learning_rate": 0.0005, "loss": 1.3543413877487183, "step": 24356 }, { "epoch": 1.92014189988175, "grad_norm": 0.4416719079017639, "learning_rate": 0.0005, "loss": 1.2922720909118652, "step": 24357 }, { "epoch": 1.920220733149389, "grad_norm": 0.4120197296142578, "learning_rate": 0.0005, "loss": 1.290554404258728, "step": 24358 }, { "epoch": 1.920299566417028, "grad_norm": 1.0734699964523315, "learning_rate": 0.0005, "loss": 1.4385788440704346, "step": 24359 }, { "epoch": 1.920378399684667, "grad_norm": 0.7433494925498962, "learning_rate": 0.0005, "loss": 1.5637598037719727, "step": 24360 }, { "epoch": 1.9204572329523057, "grad_norm": 0.5414905548095703, "learning_rate": 0.0005, "loss": 1.5883095264434814, "step": 24361 }, { "epoch": 1.9205360662199449, "grad_norm": 0.5276691317558289, "learning_rate": 0.0005, "loss": 1.2838371992111206, "step": 24362 }, { "epoch": 1.9206148994875838, "grad_norm": 0.543913722038269, "learning_rate": 0.0005, "loss": 1.439793348312378, "step": 24363 }, { "epoch": 1.9206937327552227, "grad_norm": 0.5413733720779419, "learning_rate": 0.0005, "loss": 1.2994868755340576, "step": 24364 }, { "epoch": 1.9207725660228616, "grad_norm": 0.621295690536499, "learning_rate": 0.0005, "loss": 1.2915443181991577, "step": 24365 }, { "epoch": 1.9208513992905005, "grad_norm": 1.421960711479187, "learning_rate": 0.0005, "loss": 1.6374765634536743, "step": 24366 }, { "epoch": 1.9209302325581397, "grad_norm": 0.5453924536705017, "learning_rate": 0.0005, "loss": 1.3032608032226562, "step": 24367 }, { "epoch": 1.9210090658257784, "grad_norm": 0.7197274565696716, "learning_rate": 0.0005, "loss": 1.3075551986694336, "step": 24368 }, { "epoch": 1.9210878990934175, "grad_norm": 1.086258053779602, "learning_rate": 0.0005, "loss": 1.44706130027771, "step": 24369 }, { "epoch": 1.9211667323610564, "grad_norm": 0.5150155425071716, "learning_rate": 0.0005, "loss": 1.2983732223510742, "step": 24370 }, { "epoch": 1.9212455656286953, "grad_norm": 1.1117891073226929, "learning_rate": 0.0005, "loss": 1.4808744192123413, "step": 24371 }, { "epoch": 1.9213243988963342, "grad_norm": 0.6003172397613525, "learning_rate": 0.0005, "loss": 1.3384658098220825, "step": 24372 }, { "epoch": 1.9214032321639731, "grad_norm": 0.8244845271110535, "learning_rate": 0.0005, "loss": 1.366694450378418, "step": 24373 }, { "epoch": 1.9214820654316123, "grad_norm": 0.5563856959342957, "learning_rate": 0.0005, "loss": 1.4973680973052979, "step": 24374 }, { "epoch": 1.921560898699251, "grad_norm": 0.542155385017395, "learning_rate": 0.0005, "loss": 1.4099005460739136, "step": 24375 }, { "epoch": 1.92163973196689, "grad_norm": 0.5368578433990479, "learning_rate": 0.0005, "loss": 1.55499267578125, "step": 24376 }, { "epoch": 1.921718565234529, "grad_norm": 0.5014187693595886, "learning_rate": 0.0005, "loss": 1.404533863067627, "step": 24377 }, { "epoch": 1.921797398502168, "grad_norm": 0.47815826535224915, "learning_rate": 0.0005, "loss": 1.3644306659698486, "step": 24378 }, { "epoch": 1.9218762317698068, "grad_norm": 0.47934091091156006, "learning_rate": 0.0005, "loss": 1.3619904518127441, "step": 24379 }, { "epoch": 1.9219550650374457, "grad_norm": 0.5664995312690735, "learning_rate": 0.0005, "loss": 1.2935611009597778, "step": 24380 }, { "epoch": 1.9220338983050849, "grad_norm": 0.5095530152320862, "learning_rate": 0.0005, "loss": 1.3337174654006958, "step": 24381 }, { "epoch": 1.9221127315727236, "grad_norm": 0.5065775513648987, "learning_rate": 0.0005, "loss": 1.490999460220337, "step": 24382 }, { "epoch": 1.9221915648403627, "grad_norm": 0.5794327259063721, "learning_rate": 0.0005, "loss": 1.554779052734375, "step": 24383 }, { "epoch": 1.9222703981080016, "grad_norm": 0.623996376991272, "learning_rate": 0.0005, "loss": 1.4476442337036133, "step": 24384 }, { "epoch": 1.9223492313756405, "grad_norm": 0.5625943541526794, "learning_rate": 0.0005, "loss": 1.557742953300476, "step": 24385 }, { "epoch": 1.9224280646432794, "grad_norm": 0.6618022322654724, "learning_rate": 0.0005, "loss": 1.5719980001449585, "step": 24386 }, { "epoch": 1.9225068979109183, "grad_norm": 0.594416081905365, "learning_rate": 0.0005, "loss": 1.2833292484283447, "step": 24387 }, { "epoch": 1.9225857311785575, "grad_norm": 0.6880358457565308, "learning_rate": 0.0005, "loss": 1.4713550806045532, "step": 24388 }, { "epoch": 1.9226645644461962, "grad_norm": 0.5657851099967957, "learning_rate": 0.0005, "loss": 1.30720853805542, "step": 24389 }, { "epoch": 1.9227433977138353, "grad_norm": 0.6579865217208862, "learning_rate": 0.0005, "loss": 1.546804428100586, "step": 24390 }, { "epoch": 1.9228222309814742, "grad_norm": 0.5971181392669678, "learning_rate": 0.0005, "loss": 1.5033246278762817, "step": 24391 }, { "epoch": 1.9229010642491131, "grad_norm": 0.49790310859680176, "learning_rate": 0.0005, "loss": 1.5851879119873047, "step": 24392 }, { "epoch": 1.922979897516752, "grad_norm": 0.5406858921051025, "learning_rate": 0.0005, "loss": 1.39369535446167, "step": 24393 }, { "epoch": 1.923058730784391, "grad_norm": 0.7330262660980225, "learning_rate": 0.0005, "loss": 1.3166773319244385, "step": 24394 }, { "epoch": 1.92313756405203, "grad_norm": 0.54268479347229, "learning_rate": 0.0005, "loss": 1.2142040729522705, "step": 24395 }, { "epoch": 1.9232163973196688, "grad_norm": 0.6029713749885559, "learning_rate": 0.0005, "loss": 1.3811256885528564, "step": 24396 }, { "epoch": 1.923295230587308, "grad_norm": 0.5764508247375488, "learning_rate": 0.0005, "loss": 1.479633092880249, "step": 24397 }, { "epoch": 1.9233740638549468, "grad_norm": 0.5550993084907532, "learning_rate": 0.0005, "loss": 1.4571068286895752, "step": 24398 }, { "epoch": 1.9234528971225857, "grad_norm": 0.5869020223617554, "learning_rate": 0.0005, "loss": 1.4437384605407715, "step": 24399 }, { "epoch": 1.9235317303902246, "grad_norm": 0.5708479881286621, "learning_rate": 0.0005, "loss": 1.4739294052124023, "step": 24400 }, { "epoch": 1.9236105636578635, "grad_norm": 0.4666683077812195, "learning_rate": 0.0005, "loss": 1.285144329071045, "step": 24401 }, { "epoch": 1.9236893969255027, "grad_norm": 0.5053818821907043, "learning_rate": 0.0005, "loss": 1.5089179277420044, "step": 24402 }, { "epoch": 1.9237682301931414, "grad_norm": 0.5910482406616211, "learning_rate": 0.0005, "loss": 1.2803466320037842, "step": 24403 }, { "epoch": 1.9238470634607805, "grad_norm": 0.48355719447135925, "learning_rate": 0.0005, "loss": 1.358601450920105, "step": 24404 }, { "epoch": 1.9239258967284194, "grad_norm": 0.5108476281166077, "learning_rate": 0.0005, "loss": 1.446128010749817, "step": 24405 }, { "epoch": 1.9240047299960583, "grad_norm": 0.5373287796974182, "learning_rate": 0.0005, "loss": 1.4190404415130615, "step": 24406 }, { "epoch": 1.9240835632636972, "grad_norm": 0.5067636370658875, "learning_rate": 0.0005, "loss": 1.4678378105163574, "step": 24407 }, { "epoch": 1.9241623965313361, "grad_norm": 0.476167768239975, "learning_rate": 0.0005, "loss": 1.479886770248413, "step": 24408 }, { "epoch": 1.9242412297989753, "grad_norm": 0.5039871335029602, "learning_rate": 0.0005, "loss": 1.3152105808258057, "step": 24409 }, { "epoch": 1.924320063066614, "grad_norm": 0.4439409375190735, "learning_rate": 0.0005, "loss": 1.3651702404022217, "step": 24410 }, { "epoch": 1.924398896334253, "grad_norm": 0.43194323778152466, "learning_rate": 0.0005, "loss": 1.2883390188217163, "step": 24411 }, { "epoch": 1.924477729601892, "grad_norm": 0.7653734087944031, "learning_rate": 0.0005, "loss": 1.5756103992462158, "step": 24412 }, { "epoch": 1.924556562869531, "grad_norm": 0.6241344213485718, "learning_rate": 0.0005, "loss": 1.572580337524414, "step": 24413 }, { "epoch": 1.92463539613717, "grad_norm": 0.6275749206542969, "learning_rate": 0.0005, "loss": 1.6702936887741089, "step": 24414 }, { "epoch": 1.9247142294048087, "grad_norm": 0.5057171583175659, "learning_rate": 0.0005, "loss": 1.4280650615692139, "step": 24415 }, { "epoch": 1.9247930626724479, "grad_norm": 0.5260318517684937, "learning_rate": 0.0005, "loss": 1.3069453239440918, "step": 24416 }, { "epoch": 1.9248718959400866, "grad_norm": 0.6147386431694031, "learning_rate": 0.0005, "loss": 1.5685417652130127, "step": 24417 }, { "epoch": 1.9249507292077257, "grad_norm": 0.4221687912940979, "learning_rate": 0.0005, "loss": 1.5566885471343994, "step": 24418 }, { "epoch": 1.9250295624753646, "grad_norm": 0.4642684757709503, "learning_rate": 0.0005, "loss": 1.4540992975234985, "step": 24419 }, { "epoch": 1.9251083957430035, "grad_norm": 0.6554713249206543, "learning_rate": 0.0005, "loss": 1.4068349599838257, "step": 24420 }, { "epoch": 1.9251872290106427, "grad_norm": 0.4172988831996918, "learning_rate": 0.0005, "loss": 1.478283166885376, "step": 24421 }, { "epoch": 1.9252660622782813, "grad_norm": 0.4841265082359314, "learning_rate": 0.0005, "loss": 1.4182593822479248, "step": 24422 }, { "epoch": 1.9253448955459205, "grad_norm": 0.6117002964019775, "learning_rate": 0.0005, "loss": 1.5059078931808472, "step": 24423 }, { "epoch": 1.9254237288135592, "grad_norm": 0.5046015381813049, "learning_rate": 0.0005, "loss": 1.3472386598587036, "step": 24424 }, { "epoch": 1.9255025620811983, "grad_norm": 0.5659657120704651, "learning_rate": 0.0005, "loss": 1.3974876403808594, "step": 24425 }, { "epoch": 1.9255813953488372, "grad_norm": 0.44306910037994385, "learning_rate": 0.0005, "loss": 1.4297072887420654, "step": 24426 }, { "epoch": 1.9256602286164761, "grad_norm": 0.4626452624797821, "learning_rate": 0.0005, "loss": 1.2477673292160034, "step": 24427 }, { "epoch": 1.9257390618841153, "grad_norm": 0.6675531268119812, "learning_rate": 0.0005, "loss": 1.42181396484375, "step": 24428 }, { "epoch": 1.925817895151754, "grad_norm": 0.5019369721412659, "learning_rate": 0.0005, "loss": 1.3965930938720703, "step": 24429 }, { "epoch": 1.925896728419393, "grad_norm": 0.47716039419174194, "learning_rate": 0.0005, "loss": 1.5559048652648926, "step": 24430 }, { "epoch": 1.9259755616870318, "grad_norm": 0.48380154371261597, "learning_rate": 0.0005, "loss": 1.4005666971206665, "step": 24431 }, { "epoch": 1.926054394954671, "grad_norm": 0.690140962600708, "learning_rate": 0.0005, "loss": 1.4628084897994995, "step": 24432 }, { "epoch": 1.9261332282223098, "grad_norm": 0.5744273066520691, "learning_rate": 0.0005, "loss": 1.2651870250701904, "step": 24433 }, { "epoch": 1.9262120614899487, "grad_norm": 0.5062691569328308, "learning_rate": 0.0005, "loss": 1.368613600730896, "step": 24434 }, { "epoch": 1.9262908947575879, "grad_norm": 0.692597508430481, "learning_rate": 0.0005, "loss": 1.5048795938491821, "step": 24435 }, { "epoch": 1.9263697280252265, "grad_norm": 0.5536736249923706, "learning_rate": 0.0005, "loss": 1.4332082271575928, "step": 24436 }, { "epoch": 1.9264485612928657, "grad_norm": 0.6871480345726013, "learning_rate": 0.0005, "loss": 1.3424795866012573, "step": 24437 }, { "epoch": 1.9265273945605044, "grad_norm": 0.5072031617164612, "learning_rate": 0.0005, "loss": 1.504044771194458, "step": 24438 }, { "epoch": 1.9266062278281435, "grad_norm": 0.451981782913208, "learning_rate": 0.0005, "loss": 1.4442288875579834, "step": 24439 }, { "epoch": 1.9266850610957824, "grad_norm": 0.5329338908195496, "learning_rate": 0.0005, "loss": 1.4621069431304932, "step": 24440 }, { "epoch": 1.9267638943634213, "grad_norm": 0.4948655366897583, "learning_rate": 0.0005, "loss": 1.399381399154663, "step": 24441 }, { "epoch": 1.9268427276310605, "grad_norm": 0.5205783843994141, "learning_rate": 0.0005, "loss": 1.4579908847808838, "step": 24442 }, { "epoch": 1.9269215608986991, "grad_norm": 0.48815155029296875, "learning_rate": 0.0005, "loss": 1.5062923431396484, "step": 24443 }, { "epoch": 1.9270003941663383, "grad_norm": 0.5851317048072815, "learning_rate": 0.0005, "loss": 1.3934460878372192, "step": 24444 }, { "epoch": 1.9270792274339772, "grad_norm": 0.6053221225738525, "learning_rate": 0.0005, "loss": 1.4833805561065674, "step": 24445 }, { "epoch": 1.927158060701616, "grad_norm": 0.5151376724243164, "learning_rate": 0.0005, "loss": 1.440891981124878, "step": 24446 }, { "epoch": 1.927236893969255, "grad_norm": 0.3922993540763855, "learning_rate": 0.0005, "loss": 1.2134745121002197, "step": 24447 }, { "epoch": 1.927315727236894, "grad_norm": 0.46538591384887695, "learning_rate": 0.0005, "loss": 1.430792212486267, "step": 24448 }, { "epoch": 1.927394560504533, "grad_norm": 0.5888000130653381, "learning_rate": 0.0005, "loss": 1.49434494972229, "step": 24449 }, { "epoch": 1.9274733937721718, "grad_norm": 0.5228363871574402, "learning_rate": 0.0005, "loss": 1.4496948719024658, "step": 24450 }, { "epoch": 1.9275522270398109, "grad_norm": 0.5776951313018799, "learning_rate": 0.0005, "loss": 1.351206660270691, "step": 24451 }, { "epoch": 1.9276310603074498, "grad_norm": 0.5481374263763428, "learning_rate": 0.0005, "loss": 1.3191213607788086, "step": 24452 }, { "epoch": 1.9277098935750887, "grad_norm": 0.5303412079811096, "learning_rate": 0.0005, "loss": 1.5037903785705566, "step": 24453 }, { "epoch": 1.9277887268427276, "grad_norm": 0.5584839582443237, "learning_rate": 0.0005, "loss": 1.8133193254470825, "step": 24454 }, { "epoch": 1.9278675601103665, "grad_norm": 0.4813138246536255, "learning_rate": 0.0005, "loss": 1.4840781688690186, "step": 24455 }, { "epoch": 1.9279463933780057, "grad_norm": 0.5918380618095398, "learning_rate": 0.0005, "loss": 1.506775140762329, "step": 24456 }, { "epoch": 1.9280252266456444, "grad_norm": 1.1105397939682007, "learning_rate": 0.0005, "loss": 1.2324168682098389, "step": 24457 }, { "epoch": 1.9281040599132835, "grad_norm": 0.5765606760978699, "learning_rate": 0.0005, "loss": 1.2914986610412598, "step": 24458 }, { "epoch": 1.9281828931809224, "grad_norm": 0.5032544136047363, "learning_rate": 0.0005, "loss": 1.5727767944335938, "step": 24459 }, { "epoch": 1.9282617264485613, "grad_norm": 0.6876063942909241, "learning_rate": 0.0005, "loss": 1.4213614463806152, "step": 24460 }, { "epoch": 1.9283405597162002, "grad_norm": 0.8009763956069946, "learning_rate": 0.0005, "loss": 1.5592231750488281, "step": 24461 }, { "epoch": 1.9284193929838391, "grad_norm": 0.5631070137023926, "learning_rate": 0.0005, "loss": 1.5817327499389648, "step": 24462 }, { "epoch": 1.9284982262514783, "grad_norm": 0.5296075344085693, "learning_rate": 0.0005, "loss": 1.3183209896087646, "step": 24463 }, { "epoch": 1.928577059519117, "grad_norm": 0.4569118618965149, "learning_rate": 0.0005, "loss": 1.279296875, "step": 24464 }, { "epoch": 1.928655892786756, "grad_norm": 0.524080753326416, "learning_rate": 0.0005, "loss": 1.2034170627593994, "step": 24465 }, { "epoch": 1.928734726054395, "grad_norm": 0.556717038154602, "learning_rate": 0.0005, "loss": 1.356947660446167, "step": 24466 }, { "epoch": 1.928813559322034, "grad_norm": 0.5259890556335449, "learning_rate": 0.0005, "loss": 1.3336100578308105, "step": 24467 }, { "epoch": 1.9288923925896728, "grad_norm": 0.5583544373512268, "learning_rate": 0.0005, "loss": 1.3884469270706177, "step": 24468 }, { "epoch": 1.9289712258573117, "grad_norm": 0.6291914582252502, "learning_rate": 0.0005, "loss": 1.4012035131454468, "step": 24469 }, { "epoch": 1.9290500591249509, "grad_norm": 0.5386402010917664, "learning_rate": 0.0005, "loss": 1.1638920307159424, "step": 24470 }, { "epoch": 1.9291288923925896, "grad_norm": 0.5670875906944275, "learning_rate": 0.0005, "loss": 1.2516896724700928, "step": 24471 }, { "epoch": 1.9292077256602287, "grad_norm": 0.5508254170417786, "learning_rate": 0.0005, "loss": 1.5076241493225098, "step": 24472 }, { "epoch": 1.9292865589278676, "grad_norm": 0.5835577249526978, "learning_rate": 0.0005, "loss": 1.4213616847991943, "step": 24473 }, { "epoch": 1.9293653921955065, "grad_norm": 0.4360629618167877, "learning_rate": 0.0005, "loss": 1.4697017669677734, "step": 24474 }, { "epoch": 1.9294442254631454, "grad_norm": 0.4396136999130249, "learning_rate": 0.0005, "loss": 1.2602192163467407, "step": 24475 }, { "epoch": 1.9295230587307843, "grad_norm": 0.6172070503234863, "learning_rate": 0.0005, "loss": 1.4943877458572388, "step": 24476 }, { "epoch": 1.9296018919984235, "grad_norm": 0.6812646389007568, "learning_rate": 0.0005, "loss": 1.2556111812591553, "step": 24477 }, { "epoch": 1.9296807252660622, "grad_norm": 0.5650076866149902, "learning_rate": 0.0005, "loss": 1.362327218055725, "step": 24478 }, { "epoch": 1.9297595585337013, "grad_norm": 0.4926915764808655, "learning_rate": 0.0005, "loss": 1.393711805343628, "step": 24479 }, { "epoch": 1.9298383918013402, "grad_norm": 0.5191940665245056, "learning_rate": 0.0005, "loss": 1.3855693340301514, "step": 24480 }, { "epoch": 1.9299172250689791, "grad_norm": 0.5068911910057068, "learning_rate": 0.0005, "loss": 1.386624813079834, "step": 24481 }, { "epoch": 1.929996058336618, "grad_norm": 0.5585964322090149, "learning_rate": 0.0005, "loss": 1.411760926246643, "step": 24482 }, { "epoch": 1.930074891604257, "grad_norm": 0.39505624771118164, "learning_rate": 0.0005, "loss": 1.3014225959777832, "step": 24483 }, { "epoch": 1.930153724871896, "grad_norm": 0.6826627254486084, "learning_rate": 0.0005, "loss": 1.4804120063781738, "step": 24484 }, { "epoch": 1.9302325581395348, "grad_norm": 0.4690701961517334, "learning_rate": 0.0005, "loss": 1.2920241355895996, "step": 24485 }, { "epoch": 1.930311391407174, "grad_norm": 0.635047435760498, "learning_rate": 0.0005, "loss": 1.3669497966766357, "step": 24486 }, { "epoch": 1.9303902246748128, "grad_norm": 0.5437239408493042, "learning_rate": 0.0005, "loss": 1.438560962677002, "step": 24487 }, { "epoch": 1.9304690579424517, "grad_norm": 0.5586326122283936, "learning_rate": 0.0005, "loss": 1.6329575777053833, "step": 24488 }, { "epoch": 1.9305478912100906, "grad_norm": 0.5426324605941772, "learning_rate": 0.0005, "loss": 1.5008177757263184, "step": 24489 }, { "epoch": 1.9306267244777295, "grad_norm": 0.44427725672721863, "learning_rate": 0.0005, "loss": 1.285771131515503, "step": 24490 }, { "epoch": 1.9307055577453687, "grad_norm": 0.556120753288269, "learning_rate": 0.0005, "loss": 1.2742633819580078, "step": 24491 }, { "epoch": 1.9307843910130074, "grad_norm": 0.5200291275978088, "learning_rate": 0.0005, "loss": 1.4700649976730347, "step": 24492 }, { "epoch": 1.9308632242806465, "grad_norm": 0.48457852005958557, "learning_rate": 0.0005, "loss": 1.3117280006408691, "step": 24493 }, { "epoch": 1.9309420575482854, "grad_norm": 0.4948635399341583, "learning_rate": 0.0005, "loss": 1.3723995685577393, "step": 24494 }, { "epoch": 1.9310208908159243, "grad_norm": 0.45660582184791565, "learning_rate": 0.0005, "loss": 1.3546624183654785, "step": 24495 }, { "epoch": 1.9310997240835632, "grad_norm": 1.5183701515197754, "learning_rate": 0.0005, "loss": 1.3651750087738037, "step": 24496 }, { "epoch": 1.9311785573512021, "grad_norm": 0.4654697775840759, "learning_rate": 0.0005, "loss": 1.3217352628707886, "step": 24497 }, { "epoch": 1.9312573906188413, "grad_norm": 0.5928992629051208, "learning_rate": 0.0005, "loss": 1.4171838760375977, "step": 24498 }, { "epoch": 1.93133622388648, "grad_norm": 0.49899178743362427, "learning_rate": 0.0005, "loss": 1.2794418334960938, "step": 24499 }, { "epoch": 1.931415057154119, "grad_norm": 0.45750492811203003, "learning_rate": 0.0005, "loss": 1.2481498718261719, "step": 24500 }, { "epoch": 1.931493890421758, "grad_norm": 0.5243788361549377, "learning_rate": 0.0005, "loss": 1.3984708786010742, "step": 24501 }, { "epoch": 1.931572723689397, "grad_norm": 0.5591636300086975, "learning_rate": 0.0005, "loss": 1.4514408111572266, "step": 24502 }, { "epoch": 1.9316515569570358, "grad_norm": 0.5552404522895813, "learning_rate": 0.0005, "loss": 1.320812463760376, "step": 24503 }, { "epoch": 1.9317303902246747, "grad_norm": 0.5352044701576233, "learning_rate": 0.0005, "loss": 1.636597990989685, "step": 24504 }, { "epoch": 1.9318092234923139, "grad_norm": 0.5352669358253479, "learning_rate": 0.0005, "loss": 1.488500952720642, "step": 24505 }, { "epoch": 1.9318880567599526, "grad_norm": 0.5217730402946472, "learning_rate": 0.0005, "loss": 1.2761049270629883, "step": 24506 }, { "epoch": 1.9319668900275917, "grad_norm": 0.9388657808303833, "learning_rate": 0.0005, "loss": 1.3530628681182861, "step": 24507 }, { "epoch": 1.9320457232952306, "grad_norm": 0.567829966545105, "learning_rate": 0.0005, "loss": 1.3767127990722656, "step": 24508 }, { "epoch": 1.9321245565628695, "grad_norm": 0.5654349327087402, "learning_rate": 0.0005, "loss": 1.537414312362671, "step": 24509 }, { "epoch": 1.9322033898305084, "grad_norm": 0.49568164348602295, "learning_rate": 0.0005, "loss": 1.5029513835906982, "step": 24510 }, { "epoch": 1.9322822230981473, "grad_norm": 0.6604014039039612, "learning_rate": 0.0005, "loss": 1.4832862615585327, "step": 24511 }, { "epoch": 1.9323610563657865, "grad_norm": 0.5120615363121033, "learning_rate": 0.0005, "loss": 1.393507957458496, "step": 24512 }, { "epoch": 1.9324398896334252, "grad_norm": 0.5543691515922546, "learning_rate": 0.0005, "loss": 1.4647798538208008, "step": 24513 }, { "epoch": 1.9325187229010643, "grad_norm": 0.5980587601661682, "learning_rate": 0.0005, "loss": 1.475411295890808, "step": 24514 }, { "epoch": 1.9325975561687032, "grad_norm": 0.5965562462806702, "learning_rate": 0.0005, "loss": 1.2701632976531982, "step": 24515 }, { "epoch": 1.9326763894363421, "grad_norm": 0.4640447795391083, "learning_rate": 0.0005, "loss": 1.4922935962677002, "step": 24516 }, { "epoch": 1.932755222703981, "grad_norm": 0.5722428560256958, "learning_rate": 0.0005, "loss": 1.3999723196029663, "step": 24517 }, { "epoch": 1.93283405597162, "grad_norm": 0.51554936170578, "learning_rate": 0.0005, "loss": 1.4607863426208496, "step": 24518 }, { "epoch": 1.932912889239259, "grad_norm": 0.5188196897506714, "learning_rate": 0.0005, "loss": 1.3761038780212402, "step": 24519 }, { "epoch": 1.9329917225068978, "grad_norm": 0.5111024975776672, "learning_rate": 0.0005, "loss": 1.6340224742889404, "step": 24520 }, { "epoch": 1.933070555774537, "grad_norm": 0.5173315405845642, "learning_rate": 0.0005, "loss": 1.4580726623535156, "step": 24521 }, { "epoch": 1.9331493890421758, "grad_norm": 0.4475342035293579, "learning_rate": 0.0005, "loss": 1.386723518371582, "step": 24522 }, { "epoch": 1.9332282223098147, "grad_norm": 0.5016500949859619, "learning_rate": 0.0005, "loss": 1.265272855758667, "step": 24523 }, { "epoch": 1.9333070555774536, "grad_norm": 0.4800044000148773, "learning_rate": 0.0005, "loss": 1.5669995546340942, "step": 24524 }, { "epoch": 1.9333858888450925, "grad_norm": 0.4502902626991272, "learning_rate": 0.0005, "loss": 1.3217720985412598, "step": 24525 }, { "epoch": 1.9334647221127317, "grad_norm": 0.5244324207305908, "learning_rate": 0.0005, "loss": 1.1566365957260132, "step": 24526 }, { "epoch": 1.9335435553803704, "grad_norm": 0.5141406059265137, "learning_rate": 0.0005, "loss": 1.6303070783615112, "step": 24527 }, { "epoch": 1.9336223886480095, "grad_norm": 0.4038507640361786, "learning_rate": 0.0005, "loss": 1.1001918315887451, "step": 24528 }, { "epoch": 1.9337012219156484, "grad_norm": 0.5213164687156677, "learning_rate": 0.0005, "loss": 1.128091812133789, "step": 24529 }, { "epoch": 1.9337800551832873, "grad_norm": 0.42426183819770813, "learning_rate": 0.0005, "loss": 1.3770278692245483, "step": 24530 }, { "epoch": 1.9338588884509262, "grad_norm": 0.5705059170722961, "learning_rate": 0.0005, "loss": 1.4727967977523804, "step": 24531 }, { "epoch": 1.9339377217185652, "grad_norm": 0.4404623806476593, "learning_rate": 0.0005, "loss": 1.1223602294921875, "step": 24532 }, { "epoch": 1.9340165549862043, "grad_norm": 0.5423168540000916, "learning_rate": 0.0005, "loss": 1.538102149963379, "step": 24533 }, { "epoch": 1.934095388253843, "grad_norm": 0.7550796866416931, "learning_rate": 0.0005, "loss": 1.5128271579742432, "step": 24534 }, { "epoch": 1.934174221521482, "grad_norm": 0.7022950649261475, "learning_rate": 0.0005, "loss": 1.3568439483642578, "step": 24535 }, { "epoch": 1.934253054789121, "grad_norm": 0.7773563861846924, "learning_rate": 0.0005, "loss": 1.532670021057129, "step": 24536 }, { "epoch": 1.93433188805676, "grad_norm": 0.7052679657936096, "learning_rate": 0.0005, "loss": 1.3337794542312622, "step": 24537 }, { "epoch": 1.934410721324399, "grad_norm": 0.5822761654853821, "learning_rate": 0.0005, "loss": 1.3569295406341553, "step": 24538 }, { "epoch": 1.9344895545920378, "grad_norm": 0.6687442064285278, "learning_rate": 0.0005, "loss": 1.6672369241714478, "step": 24539 }, { "epoch": 1.9345683878596769, "grad_norm": 1.046474814414978, "learning_rate": 0.0005, "loss": 1.5965160131454468, "step": 24540 }, { "epoch": 1.9346472211273156, "grad_norm": 0.5394074320793152, "learning_rate": 0.0005, "loss": 1.3567311763763428, "step": 24541 }, { "epoch": 1.9347260543949547, "grad_norm": 0.590846598148346, "learning_rate": 0.0005, "loss": 1.359426498413086, "step": 24542 }, { "epoch": 1.9348048876625936, "grad_norm": 0.6243076920509338, "learning_rate": 0.0005, "loss": 1.3001179695129395, "step": 24543 }, { "epoch": 1.9348837209302325, "grad_norm": 0.5271902084350586, "learning_rate": 0.0005, "loss": 1.434558629989624, "step": 24544 }, { "epoch": 1.9349625541978717, "grad_norm": 0.595016360282898, "learning_rate": 0.0005, "loss": 1.3471436500549316, "step": 24545 }, { "epoch": 1.9350413874655104, "grad_norm": 0.6006484031677246, "learning_rate": 0.0005, "loss": 1.4639794826507568, "step": 24546 }, { "epoch": 1.9351202207331495, "grad_norm": 0.605642557144165, "learning_rate": 0.0005, "loss": 1.5412702560424805, "step": 24547 }, { "epoch": 1.9351990540007882, "grad_norm": 0.5468659400939941, "learning_rate": 0.0005, "loss": 1.4154341220855713, "step": 24548 }, { "epoch": 1.9352778872684273, "grad_norm": 0.6147962212562561, "learning_rate": 0.0005, "loss": 1.691087007522583, "step": 24549 }, { "epoch": 1.9353567205360662, "grad_norm": 0.49767735600471497, "learning_rate": 0.0005, "loss": 1.4493941068649292, "step": 24550 }, { "epoch": 1.9354355538037051, "grad_norm": 0.6371117830276489, "learning_rate": 0.0005, "loss": 1.3800251483917236, "step": 24551 }, { "epoch": 1.9355143870713443, "grad_norm": 0.6988334059715271, "learning_rate": 0.0005, "loss": 1.4226628541946411, "step": 24552 }, { "epoch": 1.935593220338983, "grad_norm": 0.5561619400978088, "learning_rate": 0.0005, "loss": 1.509983777999878, "step": 24553 }, { "epoch": 1.935672053606622, "grad_norm": 0.8251403570175171, "learning_rate": 0.0005, "loss": 1.7818307876586914, "step": 24554 }, { "epoch": 1.9357508868742608, "grad_norm": 0.6187715530395508, "learning_rate": 0.0005, "loss": 1.4901609420776367, "step": 24555 }, { "epoch": 1.9358297201419, "grad_norm": 0.6316965222358704, "learning_rate": 0.0005, "loss": 1.4482178688049316, "step": 24556 }, { "epoch": 1.9359085534095388, "grad_norm": 0.5974162817001343, "learning_rate": 0.0005, "loss": 1.3620034456253052, "step": 24557 }, { "epoch": 1.9359873866771777, "grad_norm": 0.5030356049537659, "learning_rate": 0.0005, "loss": 1.1798592805862427, "step": 24558 }, { "epoch": 1.9360662199448169, "grad_norm": 0.6988682746887207, "learning_rate": 0.0005, "loss": 1.3001632690429688, "step": 24559 }, { "epoch": 1.9361450532124556, "grad_norm": 0.5165719389915466, "learning_rate": 0.0005, "loss": 1.354055404663086, "step": 24560 }, { "epoch": 1.9362238864800947, "grad_norm": 0.6271710395812988, "learning_rate": 0.0005, "loss": 1.4872703552246094, "step": 24561 }, { "epoch": 1.9363027197477334, "grad_norm": 0.4338729679584503, "learning_rate": 0.0005, "loss": 1.3222320079803467, "step": 24562 }, { "epoch": 1.9363815530153725, "grad_norm": 0.5343237519264221, "learning_rate": 0.0005, "loss": 1.4459998607635498, "step": 24563 }, { "epoch": 1.9364603862830114, "grad_norm": 0.4936565160751343, "learning_rate": 0.0005, "loss": 1.4183403253555298, "step": 24564 }, { "epoch": 1.9365392195506503, "grad_norm": 0.61846524477005, "learning_rate": 0.0005, "loss": 1.533992052078247, "step": 24565 }, { "epoch": 1.9366180528182895, "grad_norm": 0.6069936156272888, "learning_rate": 0.0005, "loss": 1.2019596099853516, "step": 24566 }, { "epoch": 1.9366968860859282, "grad_norm": 0.5426865220069885, "learning_rate": 0.0005, "loss": 1.2508559226989746, "step": 24567 }, { "epoch": 1.9367757193535673, "grad_norm": 0.5207188129425049, "learning_rate": 0.0005, "loss": 1.1993744373321533, "step": 24568 }, { "epoch": 1.936854552621206, "grad_norm": 0.5761152505874634, "learning_rate": 0.0005, "loss": 1.3621132373809814, "step": 24569 }, { "epoch": 1.9369333858888451, "grad_norm": 0.6476828455924988, "learning_rate": 0.0005, "loss": 1.4666728973388672, "step": 24570 }, { "epoch": 1.937012219156484, "grad_norm": 0.5630218386650085, "learning_rate": 0.0005, "loss": 1.4961810111999512, "step": 24571 }, { "epoch": 1.937091052424123, "grad_norm": 0.4724741280078888, "learning_rate": 0.0005, "loss": 1.455665111541748, "step": 24572 }, { "epoch": 1.937169885691762, "grad_norm": 0.6562091708183289, "learning_rate": 0.0005, "loss": 1.3605867624282837, "step": 24573 }, { "epoch": 1.9372487189594008, "grad_norm": 0.44682520627975464, "learning_rate": 0.0005, "loss": 1.3398425579071045, "step": 24574 }, { "epoch": 1.93732755222704, "grad_norm": 0.5097748041152954, "learning_rate": 0.0005, "loss": 1.3939757347106934, "step": 24575 }, { "epoch": 1.9374063854946788, "grad_norm": 0.5425564050674438, "learning_rate": 0.0005, "loss": 1.5602763891220093, "step": 24576 }, { "epoch": 1.9374852187623177, "grad_norm": 0.5446891784667969, "learning_rate": 0.0005, "loss": 1.5073661804199219, "step": 24577 }, { "epoch": 1.9375640520299566, "grad_norm": 0.5288560390472412, "learning_rate": 0.0005, "loss": 1.5810065269470215, "step": 24578 }, { "epoch": 1.9376428852975955, "grad_norm": 0.5370967388153076, "learning_rate": 0.0005, "loss": 1.4048690795898438, "step": 24579 }, { "epoch": 1.9377217185652347, "grad_norm": 0.480472594499588, "learning_rate": 0.0005, "loss": 1.3031433820724487, "step": 24580 }, { "epoch": 1.9378005518328734, "grad_norm": 0.4516882598400116, "learning_rate": 0.0005, "loss": 1.2375681400299072, "step": 24581 }, { "epoch": 1.9378793851005125, "grad_norm": 0.4666668176651001, "learning_rate": 0.0005, "loss": 1.3438694477081299, "step": 24582 }, { "epoch": 1.9379582183681514, "grad_norm": 0.5417614579200745, "learning_rate": 0.0005, "loss": 1.3277256488800049, "step": 24583 }, { "epoch": 1.9380370516357903, "grad_norm": 0.6602696776390076, "learning_rate": 0.0005, "loss": 1.6119494438171387, "step": 24584 }, { "epoch": 1.9381158849034292, "grad_norm": 0.5500161647796631, "learning_rate": 0.0005, "loss": 1.469539761543274, "step": 24585 }, { "epoch": 1.9381947181710681, "grad_norm": 0.6128736734390259, "learning_rate": 0.0005, "loss": 1.7281761169433594, "step": 24586 }, { "epoch": 1.9382735514387073, "grad_norm": 0.4840189814567566, "learning_rate": 0.0005, "loss": 1.3391444683074951, "step": 24587 }, { "epoch": 1.938352384706346, "grad_norm": 0.5447763800621033, "learning_rate": 0.0005, "loss": 1.5003540515899658, "step": 24588 }, { "epoch": 1.938431217973985, "grad_norm": 0.43253055214881897, "learning_rate": 0.0005, "loss": 1.1504504680633545, "step": 24589 }, { "epoch": 1.938510051241624, "grad_norm": 0.47074732184410095, "learning_rate": 0.0005, "loss": 1.468596339225769, "step": 24590 }, { "epoch": 1.938588884509263, "grad_norm": 0.6327366828918457, "learning_rate": 0.0005, "loss": 1.4274086952209473, "step": 24591 }, { "epoch": 1.9386677177769018, "grad_norm": 0.5450735092163086, "learning_rate": 0.0005, "loss": 1.5762460231781006, "step": 24592 }, { "epoch": 1.9387465510445407, "grad_norm": 0.632172703742981, "learning_rate": 0.0005, "loss": 1.595252513885498, "step": 24593 }, { "epoch": 1.9388253843121799, "grad_norm": 0.569487988948822, "learning_rate": 0.0005, "loss": 1.4262962341308594, "step": 24594 }, { "epoch": 1.9389042175798186, "grad_norm": 0.607310950756073, "learning_rate": 0.0005, "loss": 1.4739413261413574, "step": 24595 }, { "epoch": 1.9389830508474577, "grad_norm": 0.5195026993751526, "learning_rate": 0.0005, "loss": 1.2936826944351196, "step": 24596 }, { "epoch": 1.9390618841150966, "grad_norm": 0.5879037380218506, "learning_rate": 0.0005, "loss": 1.3264026641845703, "step": 24597 }, { "epoch": 1.9391407173827355, "grad_norm": 0.6486097574234009, "learning_rate": 0.0005, "loss": 1.3585104942321777, "step": 24598 }, { "epoch": 1.9392195506503744, "grad_norm": 0.5330471396446228, "learning_rate": 0.0005, "loss": 1.2220760583877563, "step": 24599 }, { "epoch": 1.9392983839180133, "grad_norm": 0.6249052882194519, "learning_rate": 0.0005, "loss": 1.4562938213348389, "step": 24600 }, { "epoch": 1.9393772171856525, "grad_norm": 0.8706095814704895, "learning_rate": 0.0005, "loss": 1.5056136846542358, "step": 24601 }, { "epoch": 1.9394560504532912, "grad_norm": 0.5434760451316833, "learning_rate": 0.0005, "loss": 1.3528516292572021, "step": 24602 }, { "epoch": 1.9395348837209303, "grad_norm": 0.48103639483451843, "learning_rate": 0.0005, "loss": 1.5971124172210693, "step": 24603 }, { "epoch": 1.9396137169885692, "grad_norm": 0.4087833762168884, "learning_rate": 0.0005, "loss": 1.167368769645691, "step": 24604 }, { "epoch": 1.9396925502562081, "grad_norm": 0.5832739472389221, "learning_rate": 0.0005, "loss": 1.2411389350891113, "step": 24605 }, { "epoch": 1.939771383523847, "grad_norm": 0.525641679763794, "learning_rate": 0.0005, "loss": 1.420147180557251, "step": 24606 }, { "epoch": 1.939850216791486, "grad_norm": 0.9723790884017944, "learning_rate": 0.0005, "loss": 1.4908946752548218, "step": 24607 }, { "epoch": 1.939929050059125, "grad_norm": 0.5913490056991577, "learning_rate": 0.0005, "loss": 1.6316194534301758, "step": 24608 }, { "epoch": 1.9400078833267638, "grad_norm": 0.5835849046707153, "learning_rate": 0.0005, "loss": 1.3464415073394775, "step": 24609 }, { "epoch": 1.940086716594403, "grad_norm": 0.5610374808311462, "learning_rate": 0.0005, "loss": 1.4397999048233032, "step": 24610 }, { "epoch": 1.9401655498620418, "grad_norm": 0.5385233759880066, "learning_rate": 0.0005, "loss": 1.4762537479400635, "step": 24611 }, { "epoch": 1.9402443831296807, "grad_norm": 0.5573030114173889, "learning_rate": 0.0005, "loss": 1.6436705589294434, "step": 24612 }, { "epoch": 1.9403232163973196, "grad_norm": 0.5182440280914307, "learning_rate": 0.0005, "loss": 1.460935354232788, "step": 24613 }, { "epoch": 1.9404020496649586, "grad_norm": 0.531086802482605, "learning_rate": 0.0005, "loss": 1.2321248054504395, "step": 24614 }, { "epoch": 1.9404808829325977, "grad_norm": 0.6878035664558411, "learning_rate": 0.0005, "loss": 1.455424427986145, "step": 24615 }, { "epoch": 1.9405597162002364, "grad_norm": 0.5150464773178101, "learning_rate": 0.0005, "loss": 1.2275519371032715, "step": 24616 }, { "epoch": 1.9406385494678755, "grad_norm": 0.5914342999458313, "learning_rate": 0.0005, "loss": 1.3498618602752686, "step": 24617 }, { "epoch": 1.9407173827355144, "grad_norm": 0.4744066596031189, "learning_rate": 0.0005, "loss": 1.3255826234817505, "step": 24618 }, { "epoch": 1.9407962160031533, "grad_norm": 0.5533033609390259, "learning_rate": 0.0005, "loss": 1.4139764308929443, "step": 24619 }, { "epoch": 1.9408750492707922, "grad_norm": 0.5414235591888428, "learning_rate": 0.0005, "loss": 1.4229962825775146, "step": 24620 }, { "epoch": 1.9409538825384312, "grad_norm": 0.5344661474227905, "learning_rate": 0.0005, "loss": 1.2342686653137207, "step": 24621 }, { "epoch": 1.9410327158060703, "grad_norm": 0.43975532054901123, "learning_rate": 0.0005, "loss": 1.349703073501587, "step": 24622 }, { "epoch": 1.941111549073709, "grad_norm": 0.5214027762413025, "learning_rate": 0.0005, "loss": 1.4038163423538208, "step": 24623 }, { "epoch": 1.941190382341348, "grad_norm": 0.394950270652771, "learning_rate": 0.0005, "loss": 1.1203854084014893, "step": 24624 }, { "epoch": 1.941269215608987, "grad_norm": 0.49904173612594604, "learning_rate": 0.0005, "loss": 1.3281277418136597, "step": 24625 }, { "epoch": 1.941348048876626, "grad_norm": 0.4308326244354248, "learning_rate": 0.0005, "loss": 1.2171565294265747, "step": 24626 }, { "epoch": 1.9414268821442648, "grad_norm": 0.5528531074523926, "learning_rate": 0.0005, "loss": 1.426939845085144, "step": 24627 }, { "epoch": 1.9415057154119038, "grad_norm": 0.5310920476913452, "learning_rate": 0.0005, "loss": 1.6300597190856934, "step": 24628 }, { "epoch": 1.9415845486795429, "grad_norm": 0.49523645639419556, "learning_rate": 0.0005, "loss": 1.4891408681869507, "step": 24629 }, { "epoch": 1.9416633819471816, "grad_norm": 0.5257865786552429, "learning_rate": 0.0005, "loss": 1.3470852375030518, "step": 24630 }, { "epoch": 1.9417422152148207, "grad_norm": 0.487213134765625, "learning_rate": 0.0005, "loss": 1.5568325519561768, "step": 24631 }, { "epoch": 1.9418210484824596, "grad_norm": 0.4451267123222351, "learning_rate": 0.0005, "loss": 1.4143388271331787, "step": 24632 }, { "epoch": 1.9418998817500985, "grad_norm": 0.5364731550216675, "learning_rate": 0.0005, "loss": 1.5115966796875, "step": 24633 }, { "epoch": 1.9419787150177374, "grad_norm": 0.49155232310295105, "learning_rate": 0.0005, "loss": 1.3708668947219849, "step": 24634 }, { "epoch": 1.9420575482853764, "grad_norm": 0.5716988444328308, "learning_rate": 0.0005, "loss": 1.3537664413452148, "step": 24635 }, { "epoch": 1.9421363815530155, "grad_norm": 0.4712021052837372, "learning_rate": 0.0005, "loss": 1.1928510665893555, "step": 24636 }, { "epoch": 1.9422152148206542, "grad_norm": 0.5075589418411255, "learning_rate": 0.0005, "loss": 1.3602594137191772, "step": 24637 }, { "epoch": 1.9422940480882933, "grad_norm": 0.539126455783844, "learning_rate": 0.0005, "loss": 1.4724891185760498, "step": 24638 }, { "epoch": 1.9423728813559322, "grad_norm": 0.6541890501976013, "learning_rate": 0.0005, "loss": 1.589978814125061, "step": 24639 }, { "epoch": 1.9424517146235711, "grad_norm": 0.47189638018608093, "learning_rate": 0.0005, "loss": 1.4492919445037842, "step": 24640 }, { "epoch": 1.94253054789121, "grad_norm": 0.5442137122154236, "learning_rate": 0.0005, "loss": 1.6594812870025635, "step": 24641 }, { "epoch": 1.942609381158849, "grad_norm": 0.5168443918228149, "learning_rate": 0.0005, "loss": 1.3879014253616333, "step": 24642 }, { "epoch": 1.942688214426488, "grad_norm": 0.44380098581314087, "learning_rate": 0.0005, "loss": 1.570703148841858, "step": 24643 }, { "epoch": 1.9427670476941268, "grad_norm": 0.5398082733154297, "learning_rate": 0.0005, "loss": 1.2616033554077148, "step": 24644 }, { "epoch": 1.942845880961766, "grad_norm": 0.5590677261352539, "learning_rate": 0.0005, "loss": 1.6131362915039062, "step": 24645 }, { "epoch": 1.9429247142294048, "grad_norm": 0.5228447914123535, "learning_rate": 0.0005, "loss": 1.4445374011993408, "step": 24646 }, { "epoch": 1.9430035474970437, "grad_norm": 0.6246071457862854, "learning_rate": 0.0005, "loss": 1.410557508468628, "step": 24647 }, { "epoch": 1.9430823807646826, "grad_norm": 0.4233081638813019, "learning_rate": 0.0005, "loss": 1.5443720817565918, "step": 24648 }, { "epoch": 1.9431612140323216, "grad_norm": 0.4896569848060608, "learning_rate": 0.0005, "loss": 1.3374507427215576, "step": 24649 }, { "epoch": 1.9432400472999607, "grad_norm": 0.42514076828956604, "learning_rate": 0.0005, "loss": 1.1585878133773804, "step": 24650 }, { "epoch": 1.9433188805675994, "grad_norm": 0.42154616117477417, "learning_rate": 0.0005, "loss": 1.2252628803253174, "step": 24651 }, { "epoch": 1.9433977138352385, "grad_norm": 0.48318591713905334, "learning_rate": 0.0005, "loss": 1.3843486309051514, "step": 24652 }, { "epoch": 1.9434765471028774, "grad_norm": 0.43606022000312805, "learning_rate": 0.0005, "loss": 1.1436207294464111, "step": 24653 }, { "epoch": 1.9435553803705163, "grad_norm": 0.4950844347476959, "learning_rate": 0.0005, "loss": 1.2943134307861328, "step": 24654 }, { "epoch": 1.9436342136381553, "grad_norm": 0.6481460332870483, "learning_rate": 0.0005, "loss": 1.5334233045578003, "step": 24655 }, { "epoch": 1.9437130469057942, "grad_norm": 0.5159989595413208, "learning_rate": 0.0005, "loss": 1.6528825759887695, "step": 24656 }, { "epoch": 1.9437918801734333, "grad_norm": 0.5106387734413147, "learning_rate": 0.0005, "loss": 1.4449529647827148, "step": 24657 }, { "epoch": 1.943870713441072, "grad_norm": 0.5899324417114258, "learning_rate": 0.0005, "loss": 1.3026361465454102, "step": 24658 }, { "epoch": 1.9439495467087111, "grad_norm": 0.4928252696990967, "learning_rate": 0.0005, "loss": 1.418407917022705, "step": 24659 }, { "epoch": 1.94402837997635, "grad_norm": 0.534391462802887, "learning_rate": 0.0005, "loss": 1.3168680667877197, "step": 24660 }, { "epoch": 1.944107213243989, "grad_norm": 0.5580299496650696, "learning_rate": 0.0005, "loss": 1.5310612916946411, "step": 24661 }, { "epoch": 1.9441860465116279, "grad_norm": 0.5288225412368774, "learning_rate": 0.0005, "loss": 1.2685105800628662, "step": 24662 }, { "epoch": 1.9442648797792668, "grad_norm": 0.42815709114074707, "learning_rate": 0.0005, "loss": 1.179560661315918, "step": 24663 }, { "epoch": 1.944343713046906, "grad_norm": 0.4668693542480469, "learning_rate": 0.0005, "loss": 1.5399079322814941, "step": 24664 }, { "epoch": 1.9444225463145446, "grad_norm": 0.4657135307788849, "learning_rate": 0.0005, "loss": 1.38594388961792, "step": 24665 }, { "epoch": 1.9445013795821837, "grad_norm": 0.4906948506832123, "learning_rate": 0.0005, "loss": 1.4705731868743896, "step": 24666 }, { "epoch": 1.9445802128498226, "grad_norm": 2.1462533473968506, "learning_rate": 0.0005, "loss": 1.2979416847229004, "step": 24667 }, { "epoch": 1.9446590461174615, "grad_norm": 0.4996684789657593, "learning_rate": 0.0005, "loss": 1.3532435894012451, "step": 24668 }, { "epoch": 1.9447378793851007, "grad_norm": 0.4930557310581207, "learning_rate": 0.0005, "loss": 1.582947015762329, "step": 24669 }, { "epoch": 1.9448167126527394, "grad_norm": 0.576318621635437, "learning_rate": 0.0005, "loss": 1.4261443614959717, "step": 24670 }, { "epoch": 1.9448955459203785, "grad_norm": 0.4884152114391327, "learning_rate": 0.0005, "loss": 1.4676631689071655, "step": 24671 }, { "epoch": 1.9449743791880172, "grad_norm": 0.5752428770065308, "learning_rate": 0.0005, "loss": 1.5056848526000977, "step": 24672 }, { "epoch": 1.9450532124556563, "grad_norm": 0.4077000319957733, "learning_rate": 0.0005, "loss": 1.4917078018188477, "step": 24673 }, { "epoch": 1.9451320457232952, "grad_norm": 0.6120359301567078, "learning_rate": 0.0005, "loss": 1.4590847492218018, "step": 24674 }, { "epoch": 1.9452108789909341, "grad_norm": 0.7030614614486694, "learning_rate": 0.0005, "loss": 1.6276044845581055, "step": 24675 }, { "epoch": 1.9452897122585733, "grad_norm": 0.5504812002182007, "learning_rate": 0.0005, "loss": 1.4378762245178223, "step": 24676 }, { "epoch": 1.945368545526212, "grad_norm": 0.5651683807373047, "learning_rate": 0.0005, "loss": 1.2144460678100586, "step": 24677 }, { "epoch": 1.945447378793851, "grad_norm": 0.5327439904212952, "learning_rate": 0.0005, "loss": 1.2847275733947754, "step": 24678 }, { "epoch": 1.9455262120614898, "grad_norm": 0.5812879800796509, "learning_rate": 0.0005, "loss": 1.4635140895843506, "step": 24679 }, { "epoch": 1.945605045329129, "grad_norm": 0.8452461361885071, "learning_rate": 0.0005, "loss": 1.4424374103546143, "step": 24680 }, { "epoch": 1.9456838785967678, "grad_norm": 0.4738445580005646, "learning_rate": 0.0005, "loss": 1.4369196891784668, "step": 24681 }, { "epoch": 1.9457627118644067, "grad_norm": 0.5658569931983948, "learning_rate": 0.0005, "loss": 1.415168046951294, "step": 24682 }, { "epoch": 1.9458415451320459, "grad_norm": 0.6148220300674438, "learning_rate": 0.0005, "loss": 1.4370014667510986, "step": 24683 }, { "epoch": 1.9459203783996846, "grad_norm": 0.5588802695274353, "learning_rate": 0.0005, "loss": 1.4759987592697144, "step": 24684 }, { "epoch": 1.9459992116673237, "grad_norm": 0.44618096947669983, "learning_rate": 0.0005, "loss": 1.1631150245666504, "step": 24685 }, { "epoch": 1.9460780449349624, "grad_norm": 0.5564996600151062, "learning_rate": 0.0005, "loss": 1.422325611114502, "step": 24686 }, { "epoch": 1.9461568782026015, "grad_norm": 0.5683233141899109, "learning_rate": 0.0005, "loss": 1.4102168083190918, "step": 24687 }, { "epoch": 1.9462357114702404, "grad_norm": 0.5143250823020935, "learning_rate": 0.0005, "loss": 1.3666764497756958, "step": 24688 }, { "epoch": 1.9463145447378793, "grad_norm": 0.623100996017456, "learning_rate": 0.0005, "loss": 1.4860832691192627, "step": 24689 }, { "epoch": 1.9463933780055185, "grad_norm": 0.6069087386131287, "learning_rate": 0.0005, "loss": 1.510918378829956, "step": 24690 }, { "epoch": 1.9464722112731572, "grad_norm": 0.8283535242080688, "learning_rate": 0.0005, "loss": 1.360150933265686, "step": 24691 }, { "epoch": 1.9465510445407963, "grad_norm": 0.6187388300895691, "learning_rate": 0.0005, "loss": 1.5316863059997559, "step": 24692 }, { "epoch": 1.946629877808435, "grad_norm": 0.5859988927841187, "learning_rate": 0.0005, "loss": 1.4358699321746826, "step": 24693 }, { "epoch": 1.9467087110760741, "grad_norm": 0.5357921719551086, "learning_rate": 0.0005, "loss": 1.5653014183044434, "step": 24694 }, { "epoch": 1.946787544343713, "grad_norm": 0.6506510376930237, "learning_rate": 0.0005, "loss": 1.5226541757583618, "step": 24695 }, { "epoch": 1.946866377611352, "grad_norm": 0.5637574791908264, "learning_rate": 0.0005, "loss": 1.43064284324646, "step": 24696 }, { "epoch": 1.946945210878991, "grad_norm": 0.46826139092445374, "learning_rate": 0.0005, "loss": 1.464106798171997, "step": 24697 }, { "epoch": 1.9470240441466298, "grad_norm": 1.9665263891220093, "learning_rate": 0.0005, "loss": 1.3351800441741943, "step": 24698 }, { "epoch": 1.947102877414269, "grad_norm": 0.4840831458568573, "learning_rate": 0.0005, "loss": 1.559955358505249, "step": 24699 }, { "epoch": 1.9471817106819078, "grad_norm": 0.511868417263031, "learning_rate": 0.0005, "loss": 1.530245065689087, "step": 24700 }, { "epoch": 1.9472605439495467, "grad_norm": 0.523720383644104, "learning_rate": 0.0005, "loss": 1.6315090656280518, "step": 24701 }, { "epoch": 1.9473393772171856, "grad_norm": 0.5327965021133423, "learning_rate": 0.0005, "loss": 1.5888867378234863, "step": 24702 }, { "epoch": 1.9474182104848246, "grad_norm": 0.39963245391845703, "learning_rate": 0.0005, "loss": 1.28794527053833, "step": 24703 }, { "epoch": 1.9474970437524637, "grad_norm": 0.4526200294494629, "learning_rate": 0.0005, "loss": 1.459590196609497, "step": 24704 }, { "epoch": 1.9475758770201024, "grad_norm": 0.5604483485221863, "learning_rate": 0.0005, "loss": 1.3305622339248657, "step": 24705 }, { "epoch": 1.9476547102877415, "grad_norm": 0.42825615406036377, "learning_rate": 0.0005, "loss": 1.3979644775390625, "step": 24706 }, { "epoch": 1.9477335435553804, "grad_norm": 0.5985821485519409, "learning_rate": 0.0005, "loss": 1.499695062637329, "step": 24707 }, { "epoch": 1.9478123768230193, "grad_norm": 0.5913828015327454, "learning_rate": 0.0005, "loss": 1.3613719940185547, "step": 24708 }, { "epoch": 1.9478912100906582, "grad_norm": 0.618781328201294, "learning_rate": 0.0005, "loss": 1.4102272987365723, "step": 24709 }, { "epoch": 1.9479700433582972, "grad_norm": 0.6250736713409424, "learning_rate": 0.0005, "loss": 1.3730425834655762, "step": 24710 }, { "epoch": 1.9480488766259363, "grad_norm": 0.45868563652038574, "learning_rate": 0.0005, "loss": 1.3634209632873535, "step": 24711 }, { "epoch": 1.948127709893575, "grad_norm": 0.4997306168079376, "learning_rate": 0.0005, "loss": 1.4134838581085205, "step": 24712 }, { "epoch": 1.948206543161214, "grad_norm": 0.5546479821205139, "learning_rate": 0.0005, "loss": 1.324844479560852, "step": 24713 }, { "epoch": 1.948285376428853, "grad_norm": 0.5742992162704468, "learning_rate": 0.0005, "loss": 1.6148220300674438, "step": 24714 }, { "epoch": 1.948364209696492, "grad_norm": 0.3933088481426239, "learning_rate": 0.0005, "loss": 1.269321084022522, "step": 24715 }, { "epoch": 1.9484430429641308, "grad_norm": 0.43865087628364563, "learning_rate": 0.0005, "loss": 1.2589716911315918, "step": 24716 }, { "epoch": 1.9485218762317698, "grad_norm": 0.4685736298561096, "learning_rate": 0.0005, "loss": 1.3288354873657227, "step": 24717 }, { "epoch": 1.948600709499409, "grad_norm": 0.4982452690601349, "learning_rate": 0.0005, "loss": 1.3117765188217163, "step": 24718 }, { "epoch": 1.9486795427670476, "grad_norm": 0.5615741014480591, "learning_rate": 0.0005, "loss": 1.5656747817993164, "step": 24719 }, { "epoch": 1.9487583760346867, "grad_norm": 0.43446335196495056, "learning_rate": 0.0005, "loss": 1.5175577402114868, "step": 24720 }, { "epoch": 1.9488372093023256, "grad_norm": 0.5142613053321838, "learning_rate": 0.0005, "loss": 1.3571456670761108, "step": 24721 }, { "epoch": 1.9489160425699645, "grad_norm": 0.5919089913368225, "learning_rate": 0.0005, "loss": 1.273384690284729, "step": 24722 }, { "epoch": 1.9489948758376034, "grad_norm": 0.6429592370986938, "learning_rate": 0.0005, "loss": 1.5878041982650757, "step": 24723 }, { "epoch": 1.9490737091052424, "grad_norm": 0.46019741892814636, "learning_rate": 0.0005, "loss": 1.5612657070159912, "step": 24724 }, { "epoch": 1.9491525423728815, "grad_norm": 0.5825235843658447, "learning_rate": 0.0005, "loss": 1.307660698890686, "step": 24725 }, { "epoch": 1.9492313756405202, "grad_norm": 0.5274300575256348, "learning_rate": 0.0005, "loss": 1.5172069072723389, "step": 24726 }, { "epoch": 1.9493102089081593, "grad_norm": 0.5538339018821716, "learning_rate": 0.0005, "loss": 1.5672657489776611, "step": 24727 }, { "epoch": 1.9493890421757982, "grad_norm": 0.5661598443984985, "learning_rate": 0.0005, "loss": 1.3208940029144287, "step": 24728 }, { "epoch": 1.9494678754434371, "grad_norm": 0.47241777181625366, "learning_rate": 0.0005, "loss": 1.382323980331421, "step": 24729 }, { "epoch": 1.949546708711076, "grad_norm": 0.5456783771514893, "learning_rate": 0.0005, "loss": 1.3768507242202759, "step": 24730 }, { "epoch": 1.949625541978715, "grad_norm": 0.4445515275001526, "learning_rate": 0.0005, "loss": 1.1353859901428223, "step": 24731 }, { "epoch": 1.949704375246354, "grad_norm": 0.5033071637153625, "learning_rate": 0.0005, "loss": 1.3996589183807373, "step": 24732 }, { "epoch": 1.9497832085139928, "grad_norm": 0.5411146879196167, "learning_rate": 0.0005, "loss": 1.7286721467971802, "step": 24733 }, { "epoch": 1.949862041781632, "grad_norm": 0.5407171249389648, "learning_rate": 0.0005, "loss": 1.2654719352722168, "step": 24734 }, { "epoch": 1.9499408750492708, "grad_norm": 0.47341498732566833, "learning_rate": 0.0005, "loss": 1.5084729194641113, "step": 24735 }, { "epoch": 1.9500197083169097, "grad_norm": 0.4878650903701782, "learning_rate": 0.0005, "loss": 1.3805551528930664, "step": 24736 }, { "epoch": 1.9500985415845487, "grad_norm": 0.5624491572380066, "learning_rate": 0.0005, "loss": 1.5512504577636719, "step": 24737 }, { "epoch": 1.9501773748521876, "grad_norm": 0.6046503782272339, "learning_rate": 0.0005, "loss": 1.5731823444366455, "step": 24738 }, { "epoch": 1.9502562081198267, "grad_norm": 0.4522798955440521, "learning_rate": 0.0005, "loss": 1.3703747987747192, "step": 24739 }, { "epoch": 1.9503350413874654, "grad_norm": 0.6169461011886597, "learning_rate": 0.0005, "loss": 1.613677740097046, "step": 24740 }, { "epoch": 1.9504138746551045, "grad_norm": 0.5178878307342529, "learning_rate": 0.0005, "loss": 1.3542009592056274, "step": 24741 }, { "epoch": 1.9504927079227434, "grad_norm": 0.5977669954299927, "learning_rate": 0.0005, "loss": 1.3776495456695557, "step": 24742 }, { "epoch": 1.9505715411903823, "grad_norm": 0.534570574760437, "learning_rate": 0.0005, "loss": 1.3658583164215088, "step": 24743 }, { "epoch": 1.9506503744580213, "grad_norm": 0.5356298685073853, "learning_rate": 0.0005, "loss": 1.6639983654022217, "step": 24744 }, { "epoch": 1.9507292077256602, "grad_norm": 0.7068485021591187, "learning_rate": 0.0005, "loss": 1.4122023582458496, "step": 24745 }, { "epoch": 1.9508080409932993, "grad_norm": 0.4446273744106293, "learning_rate": 0.0005, "loss": 1.305208444595337, "step": 24746 }, { "epoch": 1.950886874260938, "grad_norm": 0.6868321895599365, "learning_rate": 0.0005, "loss": 1.5371949672698975, "step": 24747 }, { "epoch": 1.9509657075285771, "grad_norm": 0.6715918779373169, "learning_rate": 0.0005, "loss": 1.3629851341247559, "step": 24748 }, { "epoch": 1.951044540796216, "grad_norm": 0.5113963484764099, "learning_rate": 0.0005, "loss": 1.5160728693008423, "step": 24749 }, { "epoch": 1.951123374063855, "grad_norm": 0.49596133828163147, "learning_rate": 0.0005, "loss": 1.3498915433883667, "step": 24750 }, { "epoch": 1.9512022073314939, "grad_norm": 0.5110723376274109, "learning_rate": 0.0005, "loss": 1.216398000717163, "step": 24751 }, { "epoch": 1.9512810405991328, "grad_norm": 0.3927602469921112, "learning_rate": 0.0005, "loss": 1.37586510181427, "step": 24752 }, { "epoch": 1.951359873866772, "grad_norm": 0.4259149432182312, "learning_rate": 0.0005, "loss": 1.3135164976119995, "step": 24753 }, { "epoch": 1.9514387071344106, "grad_norm": 0.563340961933136, "learning_rate": 0.0005, "loss": 1.546308159828186, "step": 24754 }, { "epoch": 1.9515175404020497, "grad_norm": 0.3852337598800659, "learning_rate": 0.0005, "loss": 1.3698537349700928, "step": 24755 }, { "epoch": 1.9515963736696886, "grad_norm": 0.5599778294563293, "learning_rate": 0.0005, "loss": 1.6090964078903198, "step": 24756 }, { "epoch": 1.9516752069373275, "grad_norm": 0.41108202934265137, "learning_rate": 0.0005, "loss": 1.364292860031128, "step": 24757 }, { "epoch": 1.9517540402049665, "grad_norm": 0.5524052381515503, "learning_rate": 0.0005, "loss": 1.5427817106246948, "step": 24758 }, { "epoch": 1.9518328734726054, "grad_norm": 0.578970193862915, "learning_rate": 0.0005, "loss": 1.3691973686218262, "step": 24759 }, { "epoch": 1.9519117067402445, "grad_norm": 0.4700206518173218, "learning_rate": 0.0005, "loss": 1.4147310256958008, "step": 24760 }, { "epoch": 1.9519905400078832, "grad_norm": 0.4212303161621094, "learning_rate": 0.0005, "loss": 1.2592663764953613, "step": 24761 }, { "epoch": 1.9520693732755223, "grad_norm": 0.5351167917251587, "learning_rate": 0.0005, "loss": 1.3647292852401733, "step": 24762 }, { "epoch": 1.9521482065431612, "grad_norm": 0.5877925753593445, "learning_rate": 0.0005, "loss": 1.6142258644104004, "step": 24763 }, { "epoch": 1.9522270398108001, "grad_norm": 0.5654385685920715, "learning_rate": 0.0005, "loss": 1.3919628858566284, "step": 24764 }, { "epoch": 1.952305873078439, "grad_norm": 0.4914868175983429, "learning_rate": 0.0005, "loss": 1.279747486114502, "step": 24765 }, { "epoch": 1.952384706346078, "grad_norm": 0.5047804713249207, "learning_rate": 0.0005, "loss": 1.1266555786132812, "step": 24766 }, { "epoch": 1.952463539613717, "grad_norm": 0.6010770201683044, "learning_rate": 0.0005, "loss": 1.2308859825134277, "step": 24767 }, { "epoch": 1.9525423728813558, "grad_norm": 0.5838909149169922, "learning_rate": 0.0005, "loss": 1.42218017578125, "step": 24768 }, { "epoch": 1.952621206148995, "grad_norm": 0.7148910164833069, "learning_rate": 0.0005, "loss": 1.4439728260040283, "step": 24769 }, { "epoch": 1.9527000394166338, "grad_norm": 0.4841240644454956, "learning_rate": 0.0005, "loss": 1.2638381719589233, "step": 24770 }, { "epoch": 1.9527788726842727, "grad_norm": 0.5207822322845459, "learning_rate": 0.0005, "loss": 1.694691777229309, "step": 24771 }, { "epoch": 1.9528577059519117, "grad_norm": 0.47691208124160767, "learning_rate": 0.0005, "loss": 1.325652837753296, "step": 24772 }, { "epoch": 1.9529365392195506, "grad_norm": 0.5399647355079651, "learning_rate": 0.0005, "loss": 1.673903465270996, "step": 24773 }, { "epoch": 1.9530153724871897, "grad_norm": 0.5198584794998169, "learning_rate": 0.0005, "loss": 1.1970051527023315, "step": 24774 }, { "epoch": 1.9530942057548284, "grad_norm": 0.5631554126739502, "learning_rate": 0.0005, "loss": 1.4808852672576904, "step": 24775 }, { "epoch": 1.9531730390224675, "grad_norm": 0.5929896235466003, "learning_rate": 0.0005, "loss": 1.3894588947296143, "step": 24776 }, { "epoch": 1.9532518722901064, "grad_norm": 0.527570366859436, "learning_rate": 0.0005, "loss": 1.441200613975525, "step": 24777 }, { "epoch": 1.9533307055577454, "grad_norm": 0.5172514915466309, "learning_rate": 0.0005, "loss": 1.5876373052597046, "step": 24778 }, { "epoch": 1.9534095388253843, "grad_norm": 0.502514660358429, "learning_rate": 0.0005, "loss": 1.558253526687622, "step": 24779 }, { "epoch": 1.9534883720930232, "grad_norm": 0.6365373134613037, "learning_rate": 0.0005, "loss": 1.6959795951843262, "step": 24780 }, { "epoch": 1.9535672053606623, "grad_norm": 0.6713334918022156, "learning_rate": 0.0005, "loss": 1.686417818069458, "step": 24781 }, { "epoch": 1.953646038628301, "grad_norm": 0.657658040523529, "learning_rate": 0.0005, "loss": 1.631813645362854, "step": 24782 }, { "epoch": 1.9537248718959401, "grad_norm": 0.5011796951293945, "learning_rate": 0.0005, "loss": 1.4313167333602905, "step": 24783 }, { "epoch": 1.953803705163579, "grad_norm": 0.3996656835079193, "learning_rate": 0.0005, "loss": 1.1533342599868774, "step": 24784 }, { "epoch": 1.953882538431218, "grad_norm": 0.9018830060958862, "learning_rate": 0.0005, "loss": 1.3585091829299927, "step": 24785 }, { "epoch": 1.9539613716988569, "grad_norm": 0.5389336943626404, "learning_rate": 0.0005, "loss": 1.4555467367172241, "step": 24786 }, { "epoch": 1.9540402049664958, "grad_norm": 0.4244304299354553, "learning_rate": 0.0005, "loss": 1.2640377283096313, "step": 24787 }, { "epoch": 1.954119038234135, "grad_norm": 0.48730340600013733, "learning_rate": 0.0005, "loss": 1.401637077331543, "step": 24788 }, { "epoch": 1.9541978715017736, "grad_norm": 0.7369993925094604, "learning_rate": 0.0005, "loss": 1.6277941465377808, "step": 24789 }, { "epoch": 1.9542767047694127, "grad_norm": 0.4628885090351105, "learning_rate": 0.0005, "loss": 1.1488635540008545, "step": 24790 }, { "epoch": 1.9543555380370516, "grad_norm": 0.5900282859802246, "learning_rate": 0.0005, "loss": 1.5862374305725098, "step": 24791 }, { "epoch": 1.9544343713046906, "grad_norm": 0.4862293601036072, "learning_rate": 0.0005, "loss": 1.4750027656555176, "step": 24792 }, { "epoch": 1.9545132045723297, "grad_norm": 0.5426033735275269, "learning_rate": 0.0005, "loss": 1.444379210472107, "step": 24793 }, { "epoch": 1.9545920378399684, "grad_norm": 0.6543498039245605, "learning_rate": 0.0005, "loss": 1.4659122228622437, "step": 24794 }, { "epoch": 1.9546708711076075, "grad_norm": 0.536250650882721, "learning_rate": 0.0005, "loss": 1.1949183940887451, "step": 24795 }, { "epoch": 1.9547497043752462, "grad_norm": 0.5815622210502625, "learning_rate": 0.0005, "loss": 1.6003950834274292, "step": 24796 }, { "epoch": 1.9548285376428853, "grad_norm": 0.5340210795402527, "learning_rate": 0.0005, "loss": 1.4868243932724, "step": 24797 }, { "epoch": 1.9549073709105242, "grad_norm": 0.4253895580768585, "learning_rate": 0.0005, "loss": 1.3431459665298462, "step": 24798 }, { "epoch": 1.9549862041781632, "grad_norm": 0.5170643925666809, "learning_rate": 0.0005, "loss": 1.5638015270233154, "step": 24799 }, { "epoch": 1.9550650374458023, "grad_norm": 0.45326554775238037, "learning_rate": 0.0005, "loss": 1.1762571334838867, "step": 24800 }, { "epoch": 1.955143870713441, "grad_norm": 0.48176851868629456, "learning_rate": 0.0005, "loss": 1.3400275707244873, "step": 24801 }, { "epoch": 1.9552227039810801, "grad_norm": 0.4908596873283386, "learning_rate": 0.0005, "loss": 1.3084546327590942, "step": 24802 }, { "epoch": 1.9553015372487188, "grad_norm": 0.44655150175094604, "learning_rate": 0.0005, "loss": 1.3202943801879883, "step": 24803 }, { "epoch": 1.955380370516358, "grad_norm": 0.6951836943626404, "learning_rate": 0.0005, "loss": 1.494018316268921, "step": 24804 }, { "epoch": 1.9554592037839968, "grad_norm": 0.489952951669693, "learning_rate": 0.0005, "loss": 1.094618320465088, "step": 24805 }, { "epoch": 1.9555380370516358, "grad_norm": 0.5373262166976929, "learning_rate": 0.0005, "loss": 1.5884408950805664, "step": 24806 }, { "epoch": 1.955616870319275, "grad_norm": 0.6189429759979248, "learning_rate": 0.0005, "loss": 1.5190545320510864, "step": 24807 }, { "epoch": 1.9556957035869136, "grad_norm": 0.4870590269565582, "learning_rate": 0.0005, "loss": 1.3155608177185059, "step": 24808 }, { "epoch": 1.9557745368545527, "grad_norm": 0.478567898273468, "learning_rate": 0.0005, "loss": 1.2779593467712402, "step": 24809 }, { "epoch": 1.9558533701221914, "grad_norm": 0.49879294633865356, "learning_rate": 0.0005, "loss": 1.4539159536361694, "step": 24810 }, { "epoch": 1.9559322033898305, "grad_norm": 0.49690547585487366, "learning_rate": 0.0005, "loss": 1.504091739654541, "step": 24811 }, { "epoch": 1.9560110366574694, "grad_norm": 0.5179885625839233, "learning_rate": 0.0005, "loss": 1.450289249420166, "step": 24812 }, { "epoch": 1.9560898699251084, "grad_norm": 0.6273043155670166, "learning_rate": 0.0005, "loss": 1.4243500232696533, "step": 24813 }, { "epoch": 1.9561687031927475, "grad_norm": 0.5596001744270325, "learning_rate": 0.0005, "loss": 1.676457166671753, "step": 24814 }, { "epoch": 1.9562475364603862, "grad_norm": 0.6216135025024414, "learning_rate": 0.0005, "loss": 1.3886828422546387, "step": 24815 }, { "epoch": 1.9563263697280253, "grad_norm": 0.6185089945793152, "learning_rate": 0.0005, "loss": 1.5446734428405762, "step": 24816 }, { "epoch": 1.956405202995664, "grad_norm": 0.5460031032562256, "learning_rate": 0.0005, "loss": 1.3816883563995361, "step": 24817 }, { "epoch": 1.9564840362633031, "grad_norm": 0.5305683612823486, "learning_rate": 0.0005, "loss": 1.3519489765167236, "step": 24818 }, { "epoch": 1.956562869530942, "grad_norm": 0.728142499923706, "learning_rate": 0.0005, "loss": 1.4839236736297607, "step": 24819 }, { "epoch": 1.956641702798581, "grad_norm": 0.5639535784721375, "learning_rate": 0.0005, "loss": 1.4307646751403809, "step": 24820 }, { "epoch": 1.95672053606622, "grad_norm": 0.4857827126979828, "learning_rate": 0.0005, "loss": 1.4582736492156982, "step": 24821 }, { "epoch": 1.9567993693338588, "grad_norm": 0.5229424238204956, "learning_rate": 0.0005, "loss": 1.2879891395568848, "step": 24822 }, { "epoch": 1.956878202601498, "grad_norm": 0.6664652228355408, "learning_rate": 0.0005, "loss": 1.4283933639526367, "step": 24823 }, { "epoch": 1.9569570358691368, "grad_norm": 0.7384924292564392, "learning_rate": 0.0005, "loss": 1.399507761001587, "step": 24824 }, { "epoch": 1.9570358691367757, "grad_norm": 0.4986245334148407, "learning_rate": 0.0005, "loss": 1.3643193244934082, "step": 24825 }, { "epoch": 1.9571147024044147, "grad_norm": 0.42464199662208557, "learning_rate": 0.0005, "loss": 1.3956239223480225, "step": 24826 }, { "epoch": 1.9571935356720536, "grad_norm": 0.6689065098762512, "learning_rate": 0.0005, "loss": 1.5233300924301147, "step": 24827 }, { "epoch": 1.9572723689396927, "grad_norm": 0.47171446681022644, "learning_rate": 0.0005, "loss": 1.468146562576294, "step": 24828 }, { "epoch": 1.9573512022073314, "grad_norm": 0.4566923677921295, "learning_rate": 0.0005, "loss": 1.2415159940719604, "step": 24829 }, { "epoch": 1.9574300354749705, "grad_norm": 0.6807218194007874, "learning_rate": 0.0005, "loss": 1.4958062171936035, "step": 24830 }, { "epoch": 1.9575088687426094, "grad_norm": 0.6791133284568787, "learning_rate": 0.0005, "loss": 1.356287956237793, "step": 24831 }, { "epoch": 1.9575877020102483, "grad_norm": 0.4940199553966522, "learning_rate": 0.0005, "loss": 1.5608575344085693, "step": 24832 }, { "epoch": 1.9576665352778873, "grad_norm": 0.6264495253562927, "learning_rate": 0.0005, "loss": 1.4740824699401855, "step": 24833 }, { "epoch": 1.9577453685455262, "grad_norm": 0.6342796683311462, "learning_rate": 0.0005, "loss": 1.4898600578308105, "step": 24834 }, { "epoch": 1.9578242018131653, "grad_norm": 0.4582189917564392, "learning_rate": 0.0005, "loss": 1.3229498863220215, "step": 24835 }, { "epoch": 1.957903035080804, "grad_norm": 0.5224904417991638, "learning_rate": 0.0005, "loss": 1.6223313808441162, "step": 24836 }, { "epoch": 1.9579818683484431, "grad_norm": 0.6621999144554138, "learning_rate": 0.0005, "loss": 1.5995248556137085, "step": 24837 }, { "epoch": 1.958060701616082, "grad_norm": 0.4784214198589325, "learning_rate": 0.0005, "loss": 1.4250011444091797, "step": 24838 }, { "epoch": 1.958139534883721, "grad_norm": 0.5401505827903748, "learning_rate": 0.0005, "loss": 1.3401789665222168, "step": 24839 }, { "epoch": 1.9582183681513599, "grad_norm": 0.43996796011924744, "learning_rate": 0.0005, "loss": 1.3824191093444824, "step": 24840 }, { "epoch": 1.9582972014189988, "grad_norm": 0.5066400170326233, "learning_rate": 0.0005, "loss": 1.2753351926803589, "step": 24841 }, { "epoch": 1.958376034686638, "grad_norm": 0.7876383066177368, "learning_rate": 0.0005, "loss": 1.5160365104675293, "step": 24842 }, { "epoch": 1.9584548679542766, "grad_norm": 0.5040366649627686, "learning_rate": 0.0005, "loss": 1.4112212657928467, "step": 24843 }, { "epoch": 1.9585337012219157, "grad_norm": 0.4589550495147705, "learning_rate": 0.0005, "loss": 1.135077714920044, "step": 24844 }, { "epoch": 1.9586125344895546, "grad_norm": 0.6616760492324829, "learning_rate": 0.0005, "loss": 1.6302363872528076, "step": 24845 }, { "epoch": 1.9586913677571935, "grad_norm": 0.4986949563026428, "learning_rate": 0.0005, "loss": 1.376328706741333, "step": 24846 }, { "epoch": 1.9587702010248325, "grad_norm": 0.6227468252182007, "learning_rate": 0.0005, "loss": 1.6913286447525024, "step": 24847 }, { "epoch": 1.9588490342924714, "grad_norm": 0.4915201663970947, "learning_rate": 0.0005, "loss": 1.5967997312545776, "step": 24848 }, { "epoch": 1.9589278675601105, "grad_norm": 0.46899452805519104, "learning_rate": 0.0005, "loss": 1.3211747407913208, "step": 24849 }, { "epoch": 1.9590067008277492, "grad_norm": 0.4842683970928192, "learning_rate": 0.0005, "loss": 1.3806840181350708, "step": 24850 }, { "epoch": 1.9590855340953883, "grad_norm": 0.511908769607544, "learning_rate": 0.0005, "loss": 1.2391667366027832, "step": 24851 }, { "epoch": 1.9591643673630272, "grad_norm": 0.494303822517395, "learning_rate": 0.0005, "loss": 1.23435378074646, "step": 24852 }, { "epoch": 1.9592432006306661, "grad_norm": 0.5940409302711487, "learning_rate": 0.0005, "loss": 1.4910646677017212, "step": 24853 }, { "epoch": 1.959322033898305, "grad_norm": 0.472465842962265, "learning_rate": 0.0005, "loss": 1.480460524559021, "step": 24854 }, { "epoch": 1.959400867165944, "grad_norm": 0.5256670117378235, "learning_rate": 0.0005, "loss": 1.2174594402313232, "step": 24855 }, { "epoch": 1.959479700433583, "grad_norm": 0.6912091970443726, "learning_rate": 0.0005, "loss": 1.5858643054962158, "step": 24856 }, { "epoch": 1.9595585337012218, "grad_norm": 0.5332239866256714, "learning_rate": 0.0005, "loss": 1.3800323009490967, "step": 24857 }, { "epoch": 1.959637366968861, "grad_norm": 0.5172095894813538, "learning_rate": 0.0005, "loss": 1.4042590856552124, "step": 24858 }, { "epoch": 1.9597162002364998, "grad_norm": 0.4781522750854492, "learning_rate": 0.0005, "loss": 1.3690173625946045, "step": 24859 }, { "epoch": 1.9597950335041388, "grad_norm": 0.5885156393051147, "learning_rate": 0.0005, "loss": 1.6291375160217285, "step": 24860 }, { "epoch": 1.9598738667717777, "grad_norm": 0.7137308716773987, "learning_rate": 0.0005, "loss": 1.368044137954712, "step": 24861 }, { "epoch": 1.9599527000394166, "grad_norm": 0.576308012008667, "learning_rate": 0.0005, "loss": 1.6610368490219116, "step": 24862 }, { "epoch": 1.9600315333070557, "grad_norm": 0.5041666030883789, "learning_rate": 0.0005, "loss": 1.3750224113464355, "step": 24863 }, { "epoch": 1.9601103665746944, "grad_norm": 0.49703249335289, "learning_rate": 0.0005, "loss": 1.377565860748291, "step": 24864 }, { "epoch": 1.9601891998423335, "grad_norm": 0.4292219877243042, "learning_rate": 0.0005, "loss": 1.2810755968093872, "step": 24865 }, { "epoch": 1.9602680331099724, "grad_norm": 0.5599587559700012, "learning_rate": 0.0005, "loss": 1.3634998798370361, "step": 24866 }, { "epoch": 1.9603468663776114, "grad_norm": 0.6924136877059937, "learning_rate": 0.0005, "loss": 1.6027278900146484, "step": 24867 }, { "epoch": 1.9604256996452503, "grad_norm": 0.654858410358429, "learning_rate": 0.0005, "loss": 1.647662878036499, "step": 24868 }, { "epoch": 1.9605045329128892, "grad_norm": 0.43090179562568665, "learning_rate": 0.0005, "loss": 1.140434741973877, "step": 24869 }, { "epoch": 1.9605833661805283, "grad_norm": 0.5216646194458008, "learning_rate": 0.0005, "loss": 1.4586939811706543, "step": 24870 }, { "epoch": 1.960662199448167, "grad_norm": 0.596085250377655, "learning_rate": 0.0005, "loss": 1.6078277826309204, "step": 24871 }, { "epoch": 1.9607410327158061, "grad_norm": 0.488230437040329, "learning_rate": 0.0005, "loss": 1.5626648664474487, "step": 24872 }, { "epoch": 1.960819865983445, "grad_norm": 0.508643627166748, "learning_rate": 0.0005, "loss": 1.2963531017303467, "step": 24873 }, { "epoch": 1.960898699251084, "grad_norm": 0.5217302441596985, "learning_rate": 0.0005, "loss": 1.2180806398391724, "step": 24874 }, { "epoch": 1.9609775325187229, "grad_norm": 0.5666150450706482, "learning_rate": 0.0005, "loss": 1.3464257717132568, "step": 24875 }, { "epoch": 1.9610563657863618, "grad_norm": 0.5376191735267639, "learning_rate": 0.0005, "loss": 1.3124741315841675, "step": 24876 }, { "epoch": 1.961135199054001, "grad_norm": 0.5850287079811096, "learning_rate": 0.0005, "loss": 1.5591628551483154, "step": 24877 }, { "epoch": 1.9612140323216396, "grad_norm": 0.4560088813304901, "learning_rate": 0.0005, "loss": 1.3566615581512451, "step": 24878 }, { "epoch": 1.9612928655892787, "grad_norm": 0.554508626461029, "learning_rate": 0.0005, "loss": 1.522613525390625, "step": 24879 }, { "epoch": 1.9613716988569176, "grad_norm": 0.6798605918884277, "learning_rate": 0.0005, "loss": 1.544169306755066, "step": 24880 }, { "epoch": 1.9614505321245566, "grad_norm": 0.49994391202926636, "learning_rate": 0.0005, "loss": 1.2767415046691895, "step": 24881 }, { "epoch": 1.9615293653921955, "grad_norm": 1.0527381896972656, "learning_rate": 0.0005, "loss": 1.498246192932129, "step": 24882 }, { "epoch": 1.9616081986598344, "grad_norm": 0.6287873387336731, "learning_rate": 0.0005, "loss": 1.6995997428894043, "step": 24883 }, { "epoch": 1.9616870319274735, "grad_norm": 0.6121591925621033, "learning_rate": 0.0005, "loss": 1.36293625831604, "step": 24884 }, { "epoch": 1.9617658651951122, "grad_norm": 0.7128411531448364, "learning_rate": 0.0005, "loss": 1.5326528549194336, "step": 24885 }, { "epoch": 1.9618446984627513, "grad_norm": 0.46233808994293213, "learning_rate": 0.0005, "loss": 1.311061143875122, "step": 24886 }, { "epoch": 1.9619235317303902, "grad_norm": 0.62225741147995, "learning_rate": 0.0005, "loss": 1.3924524784088135, "step": 24887 }, { "epoch": 1.9620023649980292, "grad_norm": 0.509358286857605, "learning_rate": 0.0005, "loss": 1.291762113571167, "step": 24888 }, { "epoch": 1.962081198265668, "grad_norm": 0.5248419642448425, "learning_rate": 0.0005, "loss": 1.6190439462661743, "step": 24889 }, { "epoch": 1.962160031533307, "grad_norm": 0.5363579392433167, "learning_rate": 0.0005, "loss": 1.326037049293518, "step": 24890 }, { "epoch": 1.9622388648009461, "grad_norm": 0.6443800330162048, "learning_rate": 0.0005, "loss": 1.3748873472213745, "step": 24891 }, { "epoch": 1.9623176980685848, "grad_norm": 0.5857399106025696, "learning_rate": 0.0005, "loss": 1.4763820171356201, "step": 24892 }, { "epoch": 1.962396531336224, "grad_norm": 0.5582281947135925, "learning_rate": 0.0005, "loss": 1.376295566558838, "step": 24893 }, { "epoch": 1.9624753646038628, "grad_norm": 0.6528254151344299, "learning_rate": 0.0005, "loss": 1.6464498043060303, "step": 24894 }, { "epoch": 1.9625541978715018, "grad_norm": 0.6148533821105957, "learning_rate": 0.0005, "loss": 1.518383264541626, "step": 24895 }, { "epoch": 1.9626330311391407, "grad_norm": 0.5652647614479065, "learning_rate": 0.0005, "loss": 1.303762674331665, "step": 24896 }, { "epoch": 1.9627118644067796, "grad_norm": 0.5635592341423035, "learning_rate": 0.0005, "loss": 1.5072211027145386, "step": 24897 }, { "epoch": 1.9627906976744187, "grad_norm": 0.5566683411598206, "learning_rate": 0.0005, "loss": 1.6429386138916016, "step": 24898 }, { "epoch": 1.9628695309420574, "grad_norm": 0.6104251146316528, "learning_rate": 0.0005, "loss": 1.4794609546661377, "step": 24899 }, { "epoch": 1.9629483642096965, "grad_norm": 0.6309837698936462, "learning_rate": 0.0005, "loss": 1.6536774635314941, "step": 24900 }, { "epoch": 1.9630271974773355, "grad_norm": 0.48580798506736755, "learning_rate": 0.0005, "loss": 1.3392627239227295, "step": 24901 }, { "epoch": 1.9631060307449744, "grad_norm": 0.5800302028656006, "learning_rate": 0.0005, "loss": 1.5254459381103516, "step": 24902 }, { "epoch": 1.9631848640126133, "grad_norm": 0.5469232797622681, "learning_rate": 0.0005, "loss": 1.5419061183929443, "step": 24903 }, { "epoch": 1.9632636972802522, "grad_norm": 0.5098510980606079, "learning_rate": 0.0005, "loss": 1.461613655090332, "step": 24904 }, { "epoch": 1.9633425305478913, "grad_norm": 0.4633042514324188, "learning_rate": 0.0005, "loss": 1.3695992231369019, "step": 24905 }, { "epoch": 1.96342136381553, "grad_norm": 0.5949276089668274, "learning_rate": 0.0005, "loss": 1.371596336364746, "step": 24906 }, { "epoch": 1.9635001970831691, "grad_norm": 0.5003650188446045, "learning_rate": 0.0005, "loss": 1.3755037784576416, "step": 24907 }, { "epoch": 1.963579030350808, "grad_norm": 0.513681173324585, "learning_rate": 0.0005, "loss": 1.1943141222000122, "step": 24908 }, { "epoch": 1.963657863618447, "grad_norm": 0.46457746624946594, "learning_rate": 0.0005, "loss": 1.3769276142120361, "step": 24909 }, { "epoch": 1.9637366968860859, "grad_norm": 0.6129611134529114, "learning_rate": 0.0005, "loss": 1.6497154235839844, "step": 24910 }, { "epoch": 1.9638155301537248, "grad_norm": 0.6123008728027344, "learning_rate": 0.0005, "loss": 1.313324213027954, "step": 24911 }, { "epoch": 1.963894363421364, "grad_norm": 0.6553804278373718, "learning_rate": 0.0005, "loss": 1.4076895713806152, "step": 24912 }, { "epoch": 1.9639731966890026, "grad_norm": 0.4651118516921997, "learning_rate": 0.0005, "loss": 1.4761521816253662, "step": 24913 }, { "epoch": 1.9640520299566417, "grad_norm": 0.5240194201469421, "learning_rate": 0.0005, "loss": 1.3741037845611572, "step": 24914 }, { "epoch": 1.9641308632242807, "grad_norm": 2.5338664054870605, "learning_rate": 0.0005, "loss": 1.3481130599975586, "step": 24915 }, { "epoch": 1.9642096964919196, "grad_norm": 0.381040096282959, "learning_rate": 0.0005, "loss": 1.1902227401733398, "step": 24916 }, { "epoch": 1.9642885297595587, "grad_norm": 0.4910030663013458, "learning_rate": 0.0005, "loss": 1.5482265949249268, "step": 24917 }, { "epoch": 1.9643673630271974, "grad_norm": 0.4848385751247406, "learning_rate": 0.0005, "loss": 1.4664767980575562, "step": 24918 }, { "epoch": 1.9644461962948365, "grad_norm": 0.4844103157520294, "learning_rate": 0.0005, "loss": 1.5608389377593994, "step": 24919 }, { "epoch": 1.9645250295624752, "grad_norm": 0.532008707523346, "learning_rate": 0.0005, "loss": 1.336605191230774, "step": 24920 }, { "epoch": 1.9646038628301143, "grad_norm": 0.5812627673149109, "learning_rate": 0.0005, "loss": 1.3670144081115723, "step": 24921 }, { "epoch": 1.9646826960977533, "grad_norm": 0.4530871510505676, "learning_rate": 0.0005, "loss": 1.5001466274261475, "step": 24922 }, { "epoch": 1.9647615293653922, "grad_norm": 0.48917755484580994, "learning_rate": 0.0005, "loss": 1.2595101594924927, "step": 24923 }, { "epoch": 1.9648403626330313, "grad_norm": 0.42247238755226135, "learning_rate": 0.0005, "loss": 1.2421290874481201, "step": 24924 }, { "epoch": 1.96491919590067, "grad_norm": 0.5699629187583923, "learning_rate": 0.0005, "loss": 1.480217456817627, "step": 24925 }, { "epoch": 1.9649980291683091, "grad_norm": 0.4891074299812317, "learning_rate": 0.0005, "loss": 1.5255026817321777, "step": 24926 }, { "epoch": 1.9650768624359478, "grad_norm": 0.508868396282196, "learning_rate": 0.0005, "loss": 1.3453779220581055, "step": 24927 }, { "epoch": 1.965155695703587, "grad_norm": 0.5543299913406372, "learning_rate": 0.0005, "loss": 1.5026791095733643, "step": 24928 }, { "epoch": 1.9652345289712259, "grad_norm": 0.5126544237136841, "learning_rate": 0.0005, "loss": 1.416669487953186, "step": 24929 }, { "epoch": 1.9653133622388648, "grad_norm": 0.4516282081604004, "learning_rate": 0.0005, "loss": 1.465691089630127, "step": 24930 }, { "epoch": 1.965392195506504, "grad_norm": 0.5230121612548828, "learning_rate": 0.0005, "loss": 1.229182481765747, "step": 24931 }, { "epoch": 1.9654710287741426, "grad_norm": 0.5210802555084229, "learning_rate": 0.0005, "loss": 1.343186855316162, "step": 24932 }, { "epoch": 1.9655498620417817, "grad_norm": 0.390745609998703, "learning_rate": 0.0005, "loss": 1.312514305114746, "step": 24933 }, { "epoch": 1.9656286953094204, "grad_norm": 0.6453042030334473, "learning_rate": 0.0005, "loss": 1.4882946014404297, "step": 24934 }, { "epoch": 1.9657075285770595, "grad_norm": 0.47575169801712036, "learning_rate": 0.0005, "loss": 1.3140299320220947, "step": 24935 }, { "epoch": 1.9657863618446985, "grad_norm": 0.6093855500221252, "learning_rate": 0.0005, "loss": 1.3169327974319458, "step": 24936 }, { "epoch": 1.9658651951123374, "grad_norm": 0.5957979559898376, "learning_rate": 0.0005, "loss": 1.360913872718811, "step": 24937 }, { "epoch": 1.9659440283799765, "grad_norm": 0.6268312335014343, "learning_rate": 0.0005, "loss": 1.5997495651245117, "step": 24938 }, { "epoch": 1.9660228616476152, "grad_norm": 0.5135242938995361, "learning_rate": 0.0005, "loss": 1.1861224174499512, "step": 24939 }, { "epoch": 1.9661016949152543, "grad_norm": 0.4608425199985504, "learning_rate": 0.0005, "loss": 1.4929932355880737, "step": 24940 }, { "epoch": 1.966180528182893, "grad_norm": 0.5624517798423767, "learning_rate": 0.0005, "loss": 1.6333088874816895, "step": 24941 }, { "epoch": 1.9662593614505322, "grad_norm": 0.4400518834590912, "learning_rate": 0.0005, "loss": 1.367597222328186, "step": 24942 }, { "epoch": 1.966338194718171, "grad_norm": 0.6648906469345093, "learning_rate": 0.0005, "loss": 1.6535049676895142, "step": 24943 }, { "epoch": 1.96641702798581, "grad_norm": 0.3949342370033264, "learning_rate": 0.0005, "loss": 1.3460147380828857, "step": 24944 }, { "epoch": 1.966495861253449, "grad_norm": 0.6071423292160034, "learning_rate": 0.0005, "loss": 1.1975090503692627, "step": 24945 }, { "epoch": 1.9665746945210878, "grad_norm": 0.4827975928783417, "learning_rate": 0.0005, "loss": 1.548621416091919, "step": 24946 }, { "epoch": 1.966653527788727, "grad_norm": 0.5100916028022766, "learning_rate": 0.0005, "loss": 1.3927221298217773, "step": 24947 }, { "epoch": 1.9667323610563658, "grad_norm": 0.6082321405410767, "learning_rate": 0.0005, "loss": 1.424451231956482, "step": 24948 }, { "epoch": 1.9668111943240048, "grad_norm": 0.4613998234272003, "learning_rate": 0.0005, "loss": 1.2428009510040283, "step": 24949 }, { "epoch": 1.9668900275916437, "grad_norm": 0.4311657249927521, "learning_rate": 0.0005, "loss": 1.319776177406311, "step": 24950 }, { "epoch": 1.9669688608592826, "grad_norm": 0.5102778673171997, "learning_rate": 0.0005, "loss": 1.589203119277954, "step": 24951 }, { "epoch": 1.9670476941269217, "grad_norm": 0.4999220669269562, "learning_rate": 0.0005, "loss": 1.2176878452301025, "step": 24952 }, { "epoch": 1.9671265273945604, "grad_norm": 0.5182673931121826, "learning_rate": 0.0005, "loss": 1.31759774684906, "step": 24953 }, { "epoch": 1.9672053606621995, "grad_norm": 0.5945820212364197, "learning_rate": 0.0005, "loss": 1.559579849243164, "step": 24954 }, { "epoch": 1.9672841939298384, "grad_norm": 0.5436625480651855, "learning_rate": 0.0005, "loss": 1.5118906497955322, "step": 24955 }, { "epoch": 1.9673630271974774, "grad_norm": 0.496426522731781, "learning_rate": 0.0005, "loss": 1.3009891510009766, "step": 24956 }, { "epoch": 1.9674418604651163, "grad_norm": 0.4571567475795746, "learning_rate": 0.0005, "loss": 1.2846068143844604, "step": 24957 }, { "epoch": 1.9675206937327552, "grad_norm": 0.6578035354614258, "learning_rate": 0.0005, "loss": 1.7227673530578613, "step": 24958 }, { "epoch": 1.9675995270003943, "grad_norm": 0.46803706884384155, "learning_rate": 0.0005, "loss": 1.2416285276412964, "step": 24959 }, { "epoch": 1.967678360268033, "grad_norm": 0.42203617095947266, "learning_rate": 0.0005, "loss": 1.340395450592041, "step": 24960 }, { "epoch": 1.9677571935356721, "grad_norm": 0.49425503611564636, "learning_rate": 0.0005, "loss": 1.374201774597168, "step": 24961 }, { "epoch": 1.967836026803311, "grad_norm": 0.46327248215675354, "learning_rate": 0.0005, "loss": 1.376952052116394, "step": 24962 }, { "epoch": 1.96791486007095, "grad_norm": 0.4226859509944916, "learning_rate": 0.0005, "loss": 1.2818784713745117, "step": 24963 }, { "epoch": 1.9679936933385889, "grad_norm": 0.5440379977226257, "learning_rate": 0.0005, "loss": 1.3715662956237793, "step": 24964 }, { "epoch": 1.9680725266062278, "grad_norm": 0.47967785596847534, "learning_rate": 0.0005, "loss": 1.4807530641555786, "step": 24965 }, { "epoch": 1.968151359873867, "grad_norm": 0.41088593006134033, "learning_rate": 0.0005, "loss": 1.2194069623947144, "step": 24966 }, { "epoch": 1.9682301931415056, "grad_norm": 0.489582896232605, "learning_rate": 0.0005, "loss": 1.3255126476287842, "step": 24967 }, { "epoch": 1.9683090264091447, "grad_norm": 2.546905994415283, "learning_rate": 0.0005, "loss": 1.5931921005249023, "step": 24968 }, { "epoch": 1.9683878596767836, "grad_norm": 0.5148877501487732, "learning_rate": 0.0005, "loss": 1.5232446193695068, "step": 24969 }, { "epoch": 1.9684666929444226, "grad_norm": 0.5370032787322998, "learning_rate": 0.0005, "loss": 1.3644442558288574, "step": 24970 }, { "epoch": 1.9685455262120615, "grad_norm": 0.5801675319671631, "learning_rate": 0.0005, "loss": 1.401313304901123, "step": 24971 }, { "epoch": 1.9686243594797004, "grad_norm": 0.4957856833934784, "learning_rate": 0.0005, "loss": 1.4970860481262207, "step": 24972 }, { "epoch": 1.9687031927473395, "grad_norm": 0.4268580377101898, "learning_rate": 0.0005, "loss": 1.3150817155838013, "step": 24973 }, { "epoch": 1.9687820260149782, "grad_norm": 0.5086383819580078, "learning_rate": 0.0005, "loss": 1.6802699565887451, "step": 24974 }, { "epoch": 1.9688608592826173, "grad_norm": 0.5388337969779968, "learning_rate": 0.0005, "loss": 1.4872983694076538, "step": 24975 }, { "epoch": 1.9689396925502562, "grad_norm": 0.5400332808494568, "learning_rate": 0.0005, "loss": 1.3692522048950195, "step": 24976 }, { "epoch": 1.9690185258178952, "grad_norm": 0.6121078133583069, "learning_rate": 0.0005, "loss": 1.337350845336914, "step": 24977 }, { "epoch": 1.969097359085534, "grad_norm": 0.5272785425186157, "learning_rate": 0.0005, "loss": 1.3251783847808838, "step": 24978 }, { "epoch": 1.969176192353173, "grad_norm": 0.5375644564628601, "learning_rate": 0.0005, "loss": 1.3777532577514648, "step": 24979 }, { "epoch": 1.9692550256208121, "grad_norm": 0.5040822625160217, "learning_rate": 0.0005, "loss": 1.4012525081634521, "step": 24980 }, { "epoch": 1.9693338588884508, "grad_norm": 0.5253827571868896, "learning_rate": 0.0005, "loss": 1.3791475296020508, "step": 24981 }, { "epoch": 1.96941269215609, "grad_norm": 0.6504855155944824, "learning_rate": 0.0005, "loss": 1.4015823602676392, "step": 24982 }, { "epoch": 1.9694915254237289, "grad_norm": 0.7738586068153381, "learning_rate": 0.0005, "loss": 1.5129022598266602, "step": 24983 }, { "epoch": 1.9695703586913678, "grad_norm": 0.6981523633003235, "learning_rate": 0.0005, "loss": 1.380893349647522, "step": 24984 }, { "epoch": 1.9696491919590067, "grad_norm": 0.5646354556083679, "learning_rate": 0.0005, "loss": 1.290388584136963, "step": 24985 }, { "epoch": 1.9697280252266456, "grad_norm": 0.6227567195892334, "learning_rate": 0.0005, "loss": 1.5490565299987793, "step": 24986 }, { "epoch": 1.9698068584942847, "grad_norm": 0.590634286403656, "learning_rate": 0.0005, "loss": 1.4823291301727295, "step": 24987 }, { "epoch": 1.9698856917619234, "grad_norm": 0.5019006133079529, "learning_rate": 0.0005, "loss": 1.135931134223938, "step": 24988 }, { "epoch": 1.9699645250295625, "grad_norm": 0.5157438516616821, "learning_rate": 0.0005, "loss": 1.5200303792953491, "step": 24989 }, { "epoch": 1.9700433582972015, "grad_norm": 0.5695444941520691, "learning_rate": 0.0005, "loss": 1.4509763717651367, "step": 24990 }, { "epoch": 1.9701221915648404, "grad_norm": 0.5953799486160278, "learning_rate": 0.0005, "loss": 1.3473405838012695, "step": 24991 }, { "epoch": 1.9702010248324793, "grad_norm": 0.567672610282898, "learning_rate": 0.0005, "loss": 1.6114857196807861, "step": 24992 }, { "epoch": 1.9702798581001182, "grad_norm": 0.4443112015724182, "learning_rate": 0.0005, "loss": 1.4129940271377563, "step": 24993 }, { "epoch": 1.9703586913677573, "grad_norm": 0.7407101392745972, "learning_rate": 0.0005, "loss": 1.4196231365203857, "step": 24994 }, { "epoch": 1.970437524635396, "grad_norm": 0.6391596794128418, "learning_rate": 0.0005, "loss": 1.4930078983306885, "step": 24995 }, { "epoch": 1.9705163579030351, "grad_norm": 0.5073044300079346, "learning_rate": 0.0005, "loss": 1.3294658660888672, "step": 24996 }, { "epoch": 1.970595191170674, "grad_norm": 0.5845140218734741, "learning_rate": 0.0005, "loss": 1.3203904628753662, "step": 24997 }, { "epoch": 1.970674024438313, "grad_norm": 0.4564439356327057, "learning_rate": 0.0005, "loss": 1.0624241828918457, "step": 24998 }, { "epoch": 1.9707528577059519, "grad_norm": 0.44973763823509216, "learning_rate": 0.0005, "loss": 1.2272756099700928, "step": 24999 }, { "epoch": 1.9708316909735908, "grad_norm": 0.4972911775112152, "learning_rate": 0.0005, "loss": 1.269951581954956, "step": 25000 }, { "epoch": 1.97091052424123, "grad_norm": 0.4812910556793213, "learning_rate": 0.0005, "loss": 1.6668553352355957, "step": 25001 }, { "epoch": 1.9709893575088686, "grad_norm": 0.46884992718696594, "learning_rate": 0.0005, "loss": 1.3961215019226074, "step": 25002 }, { "epoch": 1.9710681907765077, "grad_norm": 0.5123382210731506, "learning_rate": 0.0005, "loss": 1.2841072082519531, "step": 25003 }, { "epoch": 1.9711470240441467, "grad_norm": 0.5696836709976196, "learning_rate": 0.0005, "loss": 1.4397192001342773, "step": 25004 }, { "epoch": 1.9712258573117856, "grad_norm": 0.36226093769073486, "learning_rate": 0.0005, "loss": 1.095371127128601, "step": 25005 }, { "epoch": 1.9713046905794245, "grad_norm": 0.737094521522522, "learning_rate": 0.0005, "loss": 1.5674428939819336, "step": 25006 }, { "epoch": 1.9713835238470634, "grad_norm": 0.47246259450912476, "learning_rate": 0.0005, "loss": 1.4708747863769531, "step": 25007 }, { "epoch": 1.9714623571147025, "grad_norm": 0.5879527926445007, "learning_rate": 0.0005, "loss": 1.5158156156539917, "step": 25008 }, { "epoch": 1.9715411903823412, "grad_norm": 0.5174111127853394, "learning_rate": 0.0005, "loss": 1.4554290771484375, "step": 25009 }, { "epoch": 1.9716200236499803, "grad_norm": 0.5688305497169495, "learning_rate": 0.0005, "loss": 1.4419710636138916, "step": 25010 }, { "epoch": 1.9716988569176193, "grad_norm": 0.43417516350746155, "learning_rate": 0.0005, "loss": 1.31459379196167, "step": 25011 }, { "epoch": 1.9717776901852582, "grad_norm": 0.5013535618782043, "learning_rate": 0.0005, "loss": 1.4513391256332397, "step": 25012 }, { "epoch": 1.971856523452897, "grad_norm": 0.4666633903980255, "learning_rate": 0.0005, "loss": 1.3183026313781738, "step": 25013 }, { "epoch": 1.971935356720536, "grad_norm": 0.5167515873908997, "learning_rate": 0.0005, "loss": 1.41348135471344, "step": 25014 }, { "epoch": 1.9720141899881751, "grad_norm": 0.5031669735908508, "learning_rate": 0.0005, "loss": 1.296677589416504, "step": 25015 }, { "epoch": 1.9720930232558138, "grad_norm": 0.4940018057823181, "learning_rate": 0.0005, "loss": 1.1849207878112793, "step": 25016 }, { "epoch": 1.972171856523453, "grad_norm": 0.5588218569755554, "learning_rate": 0.0005, "loss": 1.474614143371582, "step": 25017 }, { "epoch": 1.9722506897910919, "grad_norm": 0.49903908371925354, "learning_rate": 0.0005, "loss": 1.2768728733062744, "step": 25018 }, { "epoch": 1.9723295230587308, "grad_norm": 0.5585418343544006, "learning_rate": 0.0005, "loss": 1.5338319540023804, "step": 25019 }, { "epoch": 1.9724083563263697, "grad_norm": 0.5094550848007202, "learning_rate": 0.0005, "loss": 1.4016997814178467, "step": 25020 }, { "epoch": 1.9724871895940086, "grad_norm": 5.783502578735352, "learning_rate": 0.0005, "loss": 1.51921808719635, "step": 25021 }, { "epoch": 1.9725660228616477, "grad_norm": 0.49401697516441345, "learning_rate": 0.0005, "loss": 1.3424015045166016, "step": 25022 }, { "epoch": 1.9726448561292864, "grad_norm": 0.5026566982269287, "learning_rate": 0.0005, "loss": 1.3912084102630615, "step": 25023 }, { "epoch": 1.9727236893969256, "grad_norm": 0.5581346750259399, "learning_rate": 0.0005, "loss": 1.4086092710494995, "step": 25024 }, { "epoch": 1.9728025226645645, "grad_norm": 0.5473636388778687, "learning_rate": 0.0005, "loss": 1.5890793800354004, "step": 25025 }, { "epoch": 1.9728813559322034, "grad_norm": 0.4654653072357178, "learning_rate": 0.0005, "loss": 1.3634824752807617, "step": 25026 }, { "epoch": 1.9729601891998423, "grad_norm": 0.5440317392349243, "learning_rate": 0.0005, "loss": 1.4129369258880615, "step": 25027 }, { "epoch": 1.9730390224674812, "grad_norm": 0.49488428235054016, "learning_rate": 0.0005, "loss": 1.4045214653015137, "step": 25028 }, { "epoch": 1.9731178557351203, "grad_norm": 0.5416846871376038, "learning_rate": 0.0005, "loss": 1.3645551204681396, "step": 25029 }, { "epoch": 1.973196689002759, "grad_norm": 0.46134451031684875, "learning_rate": 0.0005, "loss": 1.2033360004425049, "step": 25030 }, { "epoch": 1.9732755222703982, "grad_norm": 0.5070168375968933, "learning_rate": 0.0005, "loss": 1.3958526849746704, "step": 25031 }, { "epoch": 1.973354355538037, "grad_norm": 0.4147501587867737, "learning_rate": 0.0005, "loss": 1.3938578367233276, "step": 25032 }, { "epoch": 1.973433188805676, "grad_norm": 0.500283420085907, "learning_rate": 0.0005, "loss": 1.284645438194275, "step": 25033 }, { "epoch": 1.9735120220733149, "grad_norm": 0.5355873703956604, "learning_rate": 0.0005, "loss": 1.4539680480957031, "step": 25034 }, { "epoch": 1.9735908553409538, "grad_norm": 0.5233542323112488, "learning_rate": 0.0005, "loss": 1.672470211982727, "step": 25035 }, { "epoch": 1.973669688608593, "grad_norm": 0.497526615858078, "learning_rate": 0.0005, "loss": 1.221092700958252, "step": 25036 }, { "epoch": 1.9737485218762316, "grad_norm": 0.4844081699848175, "learning_rate": 0.0005, "loss": 1.4348297119140625, "step": 25037 }, { "epoch": 1.9738273551438708, "grad_norm": 0.5050142407417297, "learning_rate": 0.0005, "loss": 1.3663921356201172, "step": 25038 }, { "epoch": 1.9739061884115097, "grad_norm": 76.78945922851562, "learning_rate": 0.0005, "loss": 1.3762260675430298, "step": 25039 }, { "epoch": 1.9739850216791486, "grad_norm": 0.4747130274772644, "learning_rate": 0.0005, "loss": 1.5395877361297607, "step": 25040 }, { "epoch": 1.9740638549467877, "grad_norm": 0.5980457067489624, "learning_rate": 0.0005, "loss": 1.3725874423980713, "step": 25041 }, { "epoch": 1.9741426882144264, "grad_norm": 0.7224206328392029, "learning_rate": 0.0005, "loss": 1.6089496612548828, "step": 25042 }, { "epoch": 1.9742215214820655, "grad_norm": 0.5692657232284546, "learning_rate": 0.0005, "loss": 1.374925136566162, "step": 25043 }, { "epoch": 1.9743003547497042, "grad_norm": 0.5330213308334351, "learning_rate": 0.0005, "loss": 1.4389984607696533, "step": 25044 }, { "epoch": 1.9743791880173434, "grad_norm": 0.5279732942581177, "learning_rate": 0.0005, "loss": 1.5342210531234741, "step": 25045 }, { "epoch": 1.9744580212849823, "grad_norm": 0.5304780602455139, "learning_rate": 0.0005, "loss": 1.5219554901123047, "step": 25046 }, { "epoch": 1.9745368545526212, "grad_norm": 0.6289305090904236, "learning_rate": 0.0005, "loss": 1.3334436416625977, "step": 25047 }, { "epoch": 1.9746156878202603, "grad_norm": 0.4210643768310547, "learning_rate": 0.0005, "loss": 1.3190813064575195, "step": 25048 }, { "epoch": 1.974694521087899, "grad_norm": 0.5072302222251892, "learning_rate": 0.0005, "loss": 1.3969087600708008, "step": 25049 }, { "epoch": 1.9747733543555381, "grad_norm": 0.659637451171875, "learning_rate": 0.0005, "loss": 1.2977417707443237, "step": 25050 }, { "epoch": 1.9748521876231768, "grad_norm": 0.4712854325771332, "learning_rate": 0.0005, "loss": 1.3221832513809204, "step": 25051 }, { "epoch": 1.974931020890816, "grad_norm": 0.5597484707832336, "learning_rate": 0.0005, "loss": 1.3612279891967773, "step": 25052 }, { "epoch": 1.9750098541584549, "grad_norm": 0.604058563709259, "learning_rate": 0.0005, "loss": 1.4085638523101807, "step": 25053 }, { "epoch": 1.9750886874260938, "grad_norm": 0.6139672994613647, "learning_rate": 0.0005, "loss": 1.418470859527588, "step": 25054 }, { "epoch": 1.975167520693733, "grad_norm": 0.5358469486236572, "learning_rate": 0.0005, "loss": 1.316131830215454, "step": 25055 }, { "epoch": 1.9752463539613716, "grad_norm": 0.503260612487793, "learning_rate": 0.0005, "loss": 1.2793734073638916, "step": 25056 }, { "epoch": 1.9753251872290107, "grad_norm": 0.590822160243988, "learning_rate": 0.0005, "loss": 1.354462742805481, "step": 25057 }, { "epoch": 1.9754040204966494, "grad_norm": 0.4804537892341614, "learning_rate": 0.0005, "loss": 1.5024561882019043, "step": 25058 }, { "epoch": 1.9754828537642886, "grad_norm": 0.4175472557544708, "learning_rate": 0.0005, "loss": 1.326850175857544, "step": 25059 }, { "epoch": 1.9755616870319275, "grad_norm": 0.5696437358856201, "learning_rate": 0.0005, "loss": 1.44034743309021, "step": 25060 }, { "epoch": 1.9756405202995664, "grad_norm": 0.6501047015190125, "learning_rate": 0.0005, "loss": 1.3917956352233887, "step": 25061 }, { "epoch": 1.9757193535672055, "grad_norm": 0.5832938551902771, "learning_rate": 0.0005, "loss": 1.495079755783081, "step": 25062 }, { "epoch": 1.9757981868348442, "grad_norm": 0.5421666502952576, "learning_rate": 0.0005, "loss": 1.3963184356689453, "step": 25063 }, { "epoch": 1.9758770201024833, "grad_norm": 0.5059025883674622, "learning_rate": 0.0005, "loss": 1.3858815431594849, "step": 25064 }, { "epoch": 1.975955853370122, "grad_norm": 0.4805103540420532, "learning_rate": 0.0005, "loss": 1.1740642786026, "step": 25065 }, { "epoch": 1.9760346866377612, "grad_norm": 0.4404635429382324, "learning_rate": 0.0005, "loss": 1.257978081703186, "step": 25066 }, { "epoch": 1.9761135199054, "grad_norm": 0.6083424091339111, "learning_rate": 0.0005, "loss": 1.4355964660644531, "step": 25067 }, { "epoch": 1.976192353173039, "grad_norm": 0.7525056600570679, "learning_rate": 0.0005, "loss": 1.5564513206481934, "step": 25068 }, { "epoch": 1.9762711864406781, "grad_norm": 0.6426652073860168, "learning_rate": 0.0005, "loss": 1.504638671875, "step": 25069 }, { "epoch": 1.9763500197083168, "grad_norm": 0.5278668403625488, "learning_rate": 0.0005, "loss": 1.4476470947265625, "step": 25070 }, { "epoch": 1.976428852975956, "grad_norm": 0.59187251329422, "learning_rate": 0.0005, "loss": 1.4791882038116455, "step": 25071 }, { "epoch": 1.9765076862435949, "grad_norm": 0.4237889349460602, "learning_rate": 0.0005, "loss": 1.5180103778839111, "step": 25072 }, { "epoch": 1.9765865195112338, "grad_norm": 0.4555828273296356, "learning_rate": 0.0005, "loss": 1.3415372371673584, "step": 25073 }, { "epoch": 1.9766653527788727, "grad_norm": 0.5198084712028503, "learning_rate": 0.0005, "loss": 1.2852753400802612, "step": 25074 }, { "epoch": 1.9767441860465116, "grad_norm": 0.4710637331008911, "learning_rate": 0.0005, "loss": 1.2640430927276611, "step": 25075 }, { "epoch": 1.9768230193141507, "grad_norm": 0.541823148727417, "learning_rate": 0.0005, "loss": 1.4292988777160645, "step": 25076 }, { "epoch": 1.9769018525817894, "grad_norm": 0.5918586254119873, "learning_rate": 0.0005, "loss": 1.518563985824585, "step": 25077 }, { "epoch": 1.9769806858494285, "grad_norm": 0.5271667838096619, "learning_rate": 0.0005, "loss": 1.3507659435272217, "step": 25078 }, { "epoch": 1.9770595191170675, "grad_norm": 0.5119953155517578, "learning_rate": 0.0005, "loss": 1.3709535598754883, "step": 25079 }, { "epoch": 1.9771383523847064, "grad_norm": 0.5728475451469421, "learning_rate": 0.0005, "loss": 1.4814397096633911, "step": 25080 }, { "epoch": 1.9772171856523453, "grad_norm": 0.6335472464561462, "learning_rate": 0.0005, "loss": 1.3580849170684814, "step": 25081 }, { "epoch": 1.9772960189199842, "grad_norm": 0.5843734741210938, "learning_rate": 0.0005, "loss": 1.372987151145935, "step": 25082 }, { "epoch": 1.9773748521876233, "grad_norm": 0.537794828414917, "learning_rate": 0.0005, "loss": 1.3230222463607788, "step": 25083 }, { "epoch": 1.977453685455262, "grad_norm": 0.5329272150993347, "learning_rate": 0.0005, "loss": 1.3254282474517822, "step": 25084 }, { "epoch": 1.9775325187229011, "grad_norm": 0.4632936716079712, "learning_rate": 0.0005, "loss": 1.110353946685791, "step": 25085 }, { "epoch": 1.97761135199054, "grad_norm": 0.8726377487182617, "learning_rate": 0.0005, "loss": 1.6564550399780273, "step": 25086 }, { "epoch": 1.977690185258179, "grad_norm": 0.6446006894111633, "learning_rate": 0.0005, "loss": 1.4819204807281494, "step": 25087 }, { "epoch": 1.9777690185258179, "grad_norm": 0.4936222732067108, "learning_rate": 0.0005, "loss": 1.3724639415740967, "step": 25088 }, { "epoch": 1.9778478517934568, "grad_norm": 0.5488932132720947, "learning_rate": 0.0005, "loss": 1.570337176322937, "step": 25089 }, { "epoch": 1.977926685061096, "grad_norm": 0.5553224682807922, "learning_rate": 0.0005, "loss": 1.3173242807388306, "step": 25090 }, { "epoch": 1.9780055183287346, "grad_norm": 0.4790200889110565, "learning_rate": 0.0005, "loss": 1.280405044555664, "step": 25091 }, { "epoch": 1.9780843515963737, "grad_norm": 0.5389537811279297, "learning_rate": 0.0005, "loss": 1.644714593887329, "step": 25092 }, { "epoch": 1.9781631848640127, "grad_norm": 0.7300888895988464, "learning_rate": 0.0005, "loss": 1.4932026863098145, "step": 25093 }, { "epoch": 1.9782420181316516, "grad_norm": 0.5089248418807983, "learning_rate": 0.0005, "loss": 1.3748643398284912, "step": 25094 }, { "epoch": 1.9783208513992905, "grad_norm": 0.5294334888458252, "learning_rate": 0.0005, "loss": 1.3988200426101685, "step": 25095 }, { "epoch": 1.9783996846669294, "grad_norm": 0.5362831950187683, "learning_rate": 0.0005, "loss": 1.3898468017578125, "step": 25096 }, { "epoch": 1.9784785179345685, "grad_norm": 0.4548598527908325, "learning_rate": 0.0005, "loss": 1.3978173732757568, "step": 25097 }, { "epoch": 1.9785573512022072, "grad_norm": 0.5755941867828369, "learning_rate": 0.0005, "loss": 1.346916675567627, "step": 25098 }, { "epoch": 1.9786361844698463, "grad_norm": 0.5216706991195679, "learning_rate": 0.0005, "loss": 1.2194817066192627, "step": 25099 }, { "epoch": 1.9787150177374853, "grad_norm": 0.46256881952285767, "learning_rate": 0.0005, "loss": 1.3148328065872192, "step": 25100 }, { "epoch": 1.9787938510051242, "grad_norm": 0.5910412669181824, "learning_rate": 0.0005, "loss": 1.4525426626205444, "step": 25101 }, { "epoch": 1.978872684272763, "grad_norm": 1.2181392908096313, "learning_rate": 0.0005, "loss": 1.3811497688293457, "step": 25102 }, { "epoch": 1.978951517540402, "grad_norm": 0.5015256404876709, "learning_rate": 0.0005, "loss": 1.2201734781265259, "step": 25103 }, { "epoch": 1.9790303508080411, "grad_norm": 0.5103805065155029, "learning_rate": 0.0005, "loss": 1.3997712135314941, "step": 25104 }, { "epoch": 1.9791091840756798, "grad_norm": 0.5101880431175232, "learning_rate": 0.0005, "loss": 1.1228666305541992, "step": 25105 }, { "epoch": 1.979188017343319, "grad_norm": 0.5484932065010071, "learning_rate": 0.0005, "loss": 1.5191900730133057, "step": 25106 }, { "epoch": 1.9792668506109579, "grad_norm": 0.6071258783340454, "learning_rate": 0.0005, "loss": 1.498398780822754, "step": 25107 }, { "epoch": 1.9793456838785968, "grad_norm": 0.4581443965435028, "learning_rate": 0.0005, "loss": 1.2829760313034058, "step": 25108 }, { "epoch": 1.9794245171462357, "grad_norm": 0.5111004710197449, "learning_rate": 0.0005, "loss": 1.6040983200073242, "step": 25109 }, { "epoch": 1.9795033504138746, "grad_norm": 0.5006632804870605, "learning_rate": 0.0005, "loss": 1.3870700597763062, "step": 25110 }, { "epoch": 1.9795821836815137, "grad_norm": 0.4926561117172241, "learning_rate": 0.0005, "loss": 1.2964332103729248, "step": 25111 }, { "epoch": 1.9796610169491524, "grad_norm": 0.5280366539955139, "learning_rate": 0.0005, "loss": 1.462505578994751, "step": 25112 }, { "epoch": 1.9797398502167916, "grad_norm": 0.7461651563644409, "learning_rate": 0.0005, "loss": 1.257493495941162, "step": 25113 }, { "epoch": 1.9798186834844305, "grad_norm": 0.6987009644508362, "learning_rate": 0.0005, "loss": 1.5575997829437256, "step": 25114 }, { "epoch": 1.9798975167520694, "grad_norm": 0.49457278847694397, "learning_rate": 0.0005, "loss": 1.1797651052474976, "step": 25115 }, { "epoch": 1.9799763500197083, "grad_norm": 0.4833633303642273, "learning_rate": 0.0005, "loss": 1.2987416982650757, "step": 25116 }, { "epoch": 1.9800551832873472, "grad_norm": 0.979095995426178, "learning_rate": 0.0005, "loss": 1.328970193862915, "step": 25117 }, { "epoch": 1.9801340165549863, "grad_norm": 0.5135771036148071, "learning_rate": 0.0005, "loss": 1.4917012453079224, "step": 25118 }, { "epoch": 1.980212849822625, "grad_norm": 0.5743287205696106, "learning_rate": 0.0005, "loss": 1.562122106552124, "step": 25119 }, { "epoch": 1.9802916830902642, "grad_norm": 0.5209737420082092, "learning_rate": 0.0005, "loss": 1.5631214380264282, "step": 25120 }, { "epoch": 1.980370516357903, "grad_norm": 0.4504718780517578, "learning_rate": 0.0005, "loss": 1.1462085247039795, "step": 25121 }, { "epoch": 1.980449349625542, "grad_norm": 0.548972487449646, "learning_rate": 0.0005, "loss": 1.5730973482131958, "step": 25122 }, { "epoch": 1.9805281828931809, "grad_norm": 0.6325787901878357, "learning_rate": 0.0005, "loss": 1.4636788368225098, "step": 25123 }, { "epoch": 1.9806070161608198, "grad_norm": 0.6202803254127502, "learning_rate": 0.0005, "loss": 1.4248179197311401, "step": 25124 }, { "epoch": 1.980685849428459, "grad_norm": 0.641784131526947, "learning_rate": 0.0005, "loss": 1.408477544784546, "step": 25125 }, { "epoch": 1.9807646826960976, "grad_norm": 0.568298876285553, "learning_rate": 0.0005, "loss": 1.430272102355957, "step": 25126 }, { "epoch": 1.9808435159637368, "grad_norm": 0.4342455267906189, "learning_rate": 0.0005, "loss": 1.250288486480713, "step": 25127 }, { "epoch": 1.9809223492313757, "grad_norm": 0.5354779958724976, "learning_rate": 0.0005, "loss": 1.6329529285430908, "step": 25128 }, { "epoch": 1.9810011824990146, "grad_norm": 0.45544615387916565, "learning_rate": 0.0005, "loss": 1.2686753273010254, "step": 25129 }, { "epoch": 1.9810800157666535, "grad_norm": 1.3118590116500854, "learning_rate": 0.0005, "loss": 1.4501017332077026, "step": 25130 }, { "epoch": 1.9811588490342924, "grad_norm": 0.5850420594215393, "learning_rate": 0.0005, "loss": 1.5322363376617432, "step": 25131 }, { "epoch": 1.9812376823019315, "grad_norm": 0.5844403505325317, "learning_rate": 0.0005, "loss": 1.3830928802490234, "step": 25132 }, { "epoch": 1.9813165155695702, "grad_norm": 0.5730516910552979, "learning_rate": 0.0005, "loss": 1.5414454936981201, "step": 25133 }, { "epoch": 1.9813953488372094, "grad_norm": 0.5849870443344116, "learning_rate": 0.0005, "loss": 1.3356419801712036, "step": 25134 }, { "epoch": 1.9814741821048483, "grad_norm": 0.46888089179992676, "learning_rate": 0.0005, "loss": 1.2706680297851562, "step": 25135 }, { "epoch": 1.9815530153724872, "grad_norm": 0.5741980671882629, "learning_rate": 0.0005, "loss": 1.4337137937545776, "step": 25136 }, { "epoch": 1.981631848640126, "grad_norm": 0.5205303430557251, "learning_rate": 0.0005, "loss": 1.3250024318695068, "step": 25137 }, { "epoch": 1.981710681907765, "grad_norm": 0.5601536631584167, "learning_rate": 0.0005, "loss": 1.3314836025238037, "step": 25138 }, { "epoch": 1.9817895151754041, "grad_norm": 0.6659562587738037, "learning_rate": 0.0005, "loss": 1.4064316749572754, "step": 25139 }, { "epoch": 1.9818683484430428, "grad_norm": 0.603331983089447, "learning_rate": 0.0005, "loss": 1.482686996459961, "step": 25140 }, { "epoch": 1.981947181710682, "grad_norm": 0.457487016916275, "learning_rate": 0.0005, "loss": 1.215787410736084, "step": 25141 }, { "epoch": 1.9820260149783209, "grad_norm": 0.545551061630249, "learning_rate": 0.0005, "loss": 1.4816367626190186, "step": 25142 }, { "epoch": 1.9821048482459598, "grad_norm": 0.5043964982032776, "learning_rate": 0.0005, "loss": 1.3896679878234863, "step": 25143 }, { "epoch": 1.9821836815135987, "grad_norm": 0.497626394033432, "learning_rate": 0.0005, "loss": 1.463201642036438, "step": 25144 }, { "epoch": 1.9822625147812376, "grad_norm": 0.5181422829627991, "learning_rate": 0.0005, "loss": 1.4371986389160156, "step": 25145 }, { "epoch": 1.9823413480488767, "grad_norm": 0.5035252571105957, "learning_rate": 0.0005, "loss": 1.574533462524414, "step": 25146 }, { "epoch": 1.9824201813165154, "grad_norm": 0.7371193170547485, "learning_rate": 0.0005, "loss": 1.3590703010559082, "step": 25147 }, { "epoch": 1.9824990145841546, "grad_norm": 0.426607608795166, "learning_rate": 0.0005, "loss": 1.2892628908157349, "step": 25148 }, { "epoch": 1.9825778478517935, "grad_norm": 0.539919912815094, "learning_rate": 0.0005, "loss": 1.420403242111206, "step": 25149 }, { "epoch": 1.9826566811194324, "grad_norm": 0.4854939579963684, "learning_rate": 0.0005, "loss": 1.3527905941009521, "step": 25150 }, { "epoch": 1.9827355143870713, "grad_norm": 0.6329705119132996, "learning_rate": 0.0005, "loss": 1.4054867029190063, "step": 25151 }, { "epoch": 1.9828143476547102, "grad_norm": 0.4366145431995392, "learning_rate": 0.0005, "loss": 1.1188626289367676, "step": 25152 }, { "epoch": 1.9828931809223493, "grad_norm": 0.5389226078987122, "learning_rate": 0.0005, "loss": 1.489077091217041, "step": 25153 }, { "epoch": 1.982972014189988, "grad_norm": 0.4405030608177185, "learning_rate": 0.0005, "loss": 1.4089174270629883, "step": 25154 }, { "epoch": 1.9830508474576272, "grad_norm": 0.5309379696846008, "learning_rate": 0.0005, "loss": 1.3872227668762207, "step": 25155 }, { "epoch": 1.983129680725266, "grad_norm": 0.585411787033081, "learning_rate": 0.0005, "loss": 1.5262410640716553, "step": 25156 }, { "epoch": 1.983208513992905, "grad_norm": 0.5866402387619019, "learning_rate": 0.0005, "loss": 1.5157045125961304, "step": 25157 }, { "epoch": 1.983287347260544, "grad_norm": 0.5443923473358154, "learning_rate": 0.0005, "loss": 1.5258015394210815, "step": 25158 }, { "epoch": 1.9833661805281828, "grad_norm": 0.5974991917610168, "learning_rate": 0.0005, "loss": 1.3751940727233887, "step": 25159 }, { "epoch": 1.983445013795822, "grad_norm": 0.789190411567688, "learning_rate": 0.0005, "loss": 1.4028127193450928, "step": 25160 }, { "epoch": 1.9835238470634606, "grad_norm": 0.4642137587070465, "learning_rate": 0.0005, "loss": 1.3379241228103638, "step": 25161 }, { "epoch": 1.9836026803310998, "grad_norm": 0.5758548378944397, "learning_rate": 0.0005, "loss": 1.4156568050384521, "step": 25162 }, { "epoch": 1.9836815135987387, "grad_norm": 0.5174841284751892, "learning_rate": 0.0005, "loss": 1.5572829246520996, "step": 25163 }, { "epoch": 1.9837603468663776, "grad_norm": 0.6210100054740906, "learning_rate": 0.0005, "loss": 1.233607292175293, "step": 25164 }, { "epoch": 1.9838391801340165, "grad_norm": 0.5239400267601013, "learning_rate": 0.0005, "loss": 1.3417246341705322, "step": 25165 }, { "epoch": 1.9839180134016554, "grad_norm": 1.2455872297286987, "learning_rate": 0.0005, "loss": 1.7580177783966064, "step": 25166 }, { "epoch": 1.9839968466692945, "grad_norm": 0.709125816822052, "learning_rate": 0.0005, "loss": 1.3855812549591064, "step": 25167 }, { "epoch": 1.9840756799369332, "grad_norm": 0.5277063250541687, "learning_rate": 0.0005, "loss": 1.6015746593475342, "step": 25168 }, { "epoch": 1.9841545132045724, "grad_norm": 0.5434011816978455, "learning_rate": 0.0005, "loss": 1.5692542791366577, "step": 25169 }, { "epoch": 1.9842333464722113, "grad_norm": 0.8823921084403992, "learning_rate": 0.0005, "loss": 1.3839521408081055, "step": 25170 }, { "epoch": 1.9843121797398502, "grad_norm": 0.6521664261817932, "learning_rate": 0.0005, "loss": 1.699528694152832, "step": 25171 }, { "epoch": 1.9843910130074893, "grad_norm": 0.5295312404632568, "learning_rate": 0.0005, "loss": 1.4831688404083252, "step": 25172 }, { "epoch": 1.984469846275128, "grad_norm": 0.544187068939209, "learning_rate": 0.0005, "loss": 1.333571434020996, "step": 25173 }, { "epoch": 1.9845486795427671, "grad_norm": 0.5889992117881775, "learning_rate": 0.0005, "loss": 1.6279265880584717, "step": 25174 }, { "epoch": 1.9846275128104058, "grad_norm": 0.6353528499603271, "learning_rate": 0.0005, "loss": 1.6331918239593506, "step": 25175 }, { "epoch": 1.984706346078045, "grad_norm": 0.610388994216919, "learning_rate": 0.0005, "loss": 1.6026301383972168, "step": 25176 }, { "epoch": 1.9847851793456839, "grad_norm": 0.5124680995941162, "learning_rate": 0.0005, "loss": 1.4267369508743286, "step": 25177 }, { "epoch": 1.9848640126133228, "grad_norm": 0.46584033966064453, "learning_rate": 0.0005, "loss": 1.4086893796920776, "step": 25178 }, { "epoch": 1.984942845880962, "grad_norm": 0.4660071134567261, "learning_rate": 0.0005, "loss": 1.3520134687423706, "step": 25179 }, { "epoch": 1.9850216791486006, "grad_norm": 0.5281440615653992, "learning_rate": 0.0005, "loss": 1.5104418992996216, "step": 25180 }, { "epoch": 1.9851005124162397, "grad_norm": 0.6409110426902771, "learning_rate": 0.0005, "loss": 1.5431089401245117, "step": 25181 }, { "epoch": 1.9851793456838784, "grad_norm": 0.5323613882064819, "learning_rate": 0.0005, "loss": 1.3759151697158813, "step": 25182 }, { "epoch": 1.9852581789515176, "grad_norm": 0.5379147529602051, "learning_rate": 0.0005, "loss": 1.4728450775146484, "step": 25183 }, { "epoch": 1.9853370122191565, "grad_norm": 0.5153061747550964, "learning_rate": 0.0005, "loss": 1.3360437154769897, "step": 25184 }, { "epoch": 1.9854158454867954, "grad_norm": 0.4698616564273834, "learning_rate": 0.0005, "loss": 1.387531042098999, "step": 25185 }, { "epoch": 1.9854946787544345, "grad_norm": 0.47004422545433044, "learning_rate": 0.0005, "loss": 1.3073482513427734, "step": 25186 }, { "epoch": 1.9855735120220732, "grad_norm": 0.7255147695541382, "learning_rate": 0.0005, "loss": 1.439171314239502, "step": 25187 }, { "epoch": 1.9856523452897124, "grad_norm": 0.5003434419631958, "learning_rate": 0.0005, "loss": 1.3833723068237305, "step": 25188 }, { "epoch": 1.985731178557351, "grad_norm": 0.5058868527412415, "learning_rate": 0.0005, "loss": 1.4425727128982544, "step": 25189 }, { "epoch": 1.9858100118249902, "grad_norm": 0.5600583553314209, "learning_rate": 0.0005, "loss": 1.624056100845337, "step": 25190 }, { "epoch": 1.985888845092629, "grad_norm": 0.5745914578437805, "learning_rate": 0.0005, "loss": 1.4450833797454834, "step": 25191 }, { "epoch": 1.985967678360268, "grad_norm": 0.42177900671958923, "learning_rate": 0.0005, "loss": 1.3364006280899048, "step": 25192 }, { "epoch": 1.9860465116279071, "grad_norm": 0.5019512176513672, "learning_rate": 0.0005, "loss": 1.3753395080566406, "step": 25193 }, { "epoch": 1.9861253448955458, "grad_norm": 0.6245512962341309, "learning_rate": 0.0005, "loss": 1.3462731838226318, "step": 25194 }, { "epoch": 1.986204178163185, "grad_norm": 0.46290305256843567, "learning_rate": 0.0005, "loss": 1.3787312507629395, "step": 25195 }, { "epoch": 1.9862830114308236, "grad_norm": 0.5502561330795288, "learning_rate": 0.0005, "loss": 1.5061924457550049, "step": 25196 }, { "epoch": 1.9863618446984628, "grad_norm": 0.6110765337944031, "learning_rate": 0.0005, "loss": 1.4807648658752441, "step": 25197 }, { "epoch": 1.9864406779661017, "grad_norm": 0.5214844942092896, "learning_rate": 0.0005, "loss": 1.4325298070907593, "step": 25198 }, { "epoch": 1.9865195112337406, "grad_norm": 0.5077952742576599, "learning_rate": 0.0005, "loss": 1.3600289821624756, "step": 25199 }, { "epoch": 1.9865983445013797, "grad_norm": 0.6351954340934753, "learning_rate": 0.0005, "loss": 1.426103115081787, "step": 25200 }, { "epoch": 1.9866771777690184, "grad_norm": 0.483843058347702, "learning_rate": 0.0005, "loss": 1.3695368766784668, "step": 25201 }, { "epoch": 1.9867560110366576, "grad_norm": 0.7182548642158508, "learning_rate": 0.0005, "loss": 1.3117282390594482, "step": 25202 }, { "epoch": 1.9868348443042965, "grad_norm": 0.48674747347831726, "learning_rate": 0.0005, "loss": 1.3237717151641846, "step": 25203 }, { "epoch": 1.9869136775719354, "grad_norm": 0.6737329363822937, "learning_rate": 0.0005, "loss": 1.4473577737808228, "step": 25204 }, { "epoch": 1.9869925108395743, "grad_norm": 0.6981221437454224, "learning_rate": 0.0005, "loss": 1.5094690322875977, "step": 25205 }, { "epoch": 1.9870713441072132, "grad_norm": 0.5375104546546936, "learning_rate": 0.0005, "loss": 1.4151500463485718, "step": 25206 }, { "epoch": 1.9871501773748523, "grad_norm": 0.5472936630249023, "learning_rate": 0.0005, "loss": 1.3018016815185547, "step": 25207 }, { "epoch": 1.987229010642491, "grad_norm": 0.4697345793247223, "learning_rate": 0.0005, "loss": 1.3065845966339111, "step": 25208 }, { "epoch": 1.9873078439101302, "grad_norm": 0.6254900693893433, "learning_rate": 0.0005, "loss": 1.6312925815582275, "step": 25209 }, { "epoch": 1.987386677177769, "grad_norm": 0.4480826258659363, "learning_rate": 0.0005, "loss": 1.2487502098083496, "step": 25210 }, { "epoch": 1.987465510445408, "grad_norm": 0.5397255420684814, "learning_rate": 0.0005, "loss": 1.4356675148010254, "step": 25211 }, { "epoch": 1.987544343713047, "grad_norm": 0.5738361477851868, "learning_rate": 0.0005, "loss": 1.3847911357879639, "step": 25212 }, { "epoch": 1.9876231769806858, "grad_norm": 0.5403699278831482, "learning_rate": 0.0005, "loss": 1.4395785331726074, "step": 25213 }, { "epoch": 1.987702010248325, "grad_norm": 0.5855913162231445, "learning_rate": 0.0005, "loss": 1.2919032573699951, "step": 25214 }, { "epoch": 1.9877808435159636, "grad_norm": 0.52583247423172, "learning_rate": 0.0005, "loss": 1.2045836448669434, "step": 25215 }, { "epoch": 1.9878596767836028, "grad_norm": 0.5407021641731262, "learning_rate": 0.0005, "loss": 1.3134108781814575, "step": 25216 }, { "epoch": 1.9879385100512417, "grad_norm": 0.49484655261039734, "learning_rate": 0.0005, "loss": 1.6692633628845215, "step": 25217 }, { "epoch": 1.9880173433188806, "grad_norm": 0.514312207698822, "learning_rate": 0.0005, "loss": 1.4384301900863647, "step": 25218 }, { "epoch": 1.9880961765865195, "grad_norm": 0.5825915336608887, "learning_rate": 0.0005, "loss": 1.2950377464294434, "step": 25219 }, { "epoch": 1.9881750098541584, "grad_norm": 0.42053893208503723, "learning_rate": 0.0005, "loss": 1.4026023149490356, "step": 25220 }, { "epoch": 1.9882538431217975, "grad_norm": 0.7540911436080933, "learning_rate": 0.0005, "loss": 1.5119394063949585, "step": 25221 }, { "epoch": 1.9883326763894362, "grad_norm": 0.4581170976161957, "learning_rate": 0.0005, "loss": 1.4953901767730713, "step": 25222 }, { "epoch": 1.9884115096570754, "grad_norm": 0.516459047794342, "learning_rate": 0.0005, "loss": 1.3339282274246216, "step": 25223 }, { "epoch": 1.9884903429247143, "grad_norm": 0.7017471194267273, "learning_rate": 0.0005, "loss": 1.3574556112289429, "step": 25224 }, { "epoch": 1.9885691761923532, "grad_norm": 0.5905671715736389, "learning_rate": 0.0005, "loss": 1.2606174945831299, "step": 25225 }, { "epoch": 1.988648009459992, "grad_norm": 0.5616070032119751, "learning_rate": 0.0005, "loss": 1.3755033016204834, "step": 25226 }, { "epoch": 1.988726842727631, "grad_norm": 0.46376651525497437, "learning_rate": 0.0005, "loss": 1.2207322120666504, "step": 25227 }, { "epoch": 1.9888056759952701, "grad_norm": 0.5448832511901855, "learning_rate": 0.0005, "loss": 1.512906551361084, "step": 25228 }, { "epoch": 1.9888845092629088, "grad_norm": 0.5051071047782898, "learning_rate": 0.0005, "loss": 1.2776553630828857, "step": 25229 }, { "epoch": 1.988963342530548, "grad_norm": 0.500160276889801, "learning_rate": 0.0005, "loss": 1.4573016166687012, "step": 25230 }, { "epoch": 1.9890421757981869, "grad_norm": 0.48284804821014404, "learning_rate": 0.0005, "loss": 1.2389178276062012, "step": 25231 }, { "epoch": 1.9891210090658258, "grad_norm": 0.46598055958747864, "learning_rate": 0.0005, "loss": 1.4241706132888794, "step": 25232 }, { "epoch": 1.9891998423334647, "grad_norm": 0.45565715432167053, "learning_rate": 0.0005, "loss": 1.1063857078552246, "step": 25233 }, { "epoch": 1.9892786756011036, "grad_norm": 0.47343501448631287, "learning_rate": 0.0005, "loss": 1.2535682916641235, "step": 25234 }, { "epoch": 1.9893575088687427, "grad_norm": 0.6492318511009216, "learning_rate": 0.0005, "loss": 1.3772846460342407, "step": 25235 }, { "epoch": 1.9894363421363814, "grad_norm": 0.49835440516471863, "learning_rate": 0.0005, "loss": 1.3463512659072876, "step": 25236 }, { "epoch": 1.9895151754040206, "grad_norm": 0.5997062921524048, "learning_rate": 0.0005, "loss": 1.5276737213134766, "step": 25237 }, { "epoch": 1.9895940086716595, "grad_norm": 0.5895761847496033, "learning_rate": 0.0005, "loss": 1.5416004657745361, "step": 25238 }, { "epoch": 1.9896728419392984, "grad_norm": 0.5115691423416138, "learning_rate": 0.0005, "loss": 1.5647616386413574, "step": 25239 }, { "epoch": 1.9897516752069373, "grad_norm": 0.638894259929657, "learning_rate": 0.0005, "loss": 1.2937836647033691, "step": 25240 }, { "epoch": 1.9898305084745762, "grad_norm": 0.580352246761322, "learning_rate": 0.0005, "loss": 1.6723270416259766, "step": 25241 }, { "epoch": 1.9899093417422153, "grad_norm": 0.3977384865283966, "learning_rate": 0.0005, "loss": 1.1515109539031982, "step": 25242 }, { "epoch": 1.989988175009854, "grad_norm": 0.4940672516822815, "learning_rate": 0.0005, "loss": 1.187917709350586, "step": 25243 }, { "epoch": 1.9900670082774932, "grad_norm": 0.5045962333679199, "learning_rate": 0.0005, "loss": 1.439579725265503, "step": 25244 }, { "epoch": 1.990145841545132, "grad_norm": 0.5898249745368958, "learning_rate": 0.0005, "loss": 1.4144035577774048, "step": 25245 }, { "epoch": 1.990224674812771, "grad_norm": 2.6566505432128906, "learning_rate": 0.0005, "loss": 1.3503949642181396, "step": 25246 }, { "epoch": 1.99030350808041, "grad_norm": 0.6861431002616882, "learning_rate": 0.0005, "loss": 1.2237918376922607, "step": 25247 }, { "epoch": 1.9903823413480488, "grad_norm": 3.660459041595459, "learning_rate": 0.0005, "loss": 1.8744783401489258, "step": 25248 }, { "epoch": 1.990461174615688, "grad_norm": 0.43808168172836304, "learning_rate": 0.0005, "loss": 1.4594533443450928, "step": 25249 }, { "epoch": 1.9905400078833266, "grad_norm": 0.5618487596511841, "learning_rate": 0.0005, "loss": 1.3335461616516113, "step": 25250 }, { "epoch": 1.9906188411509658, "grad_norm": 0.5441924333572388, "learning_rate": 0.0005, "loss": 1.2718398571014404, "step": 25251 }, { "epoch": 1.9906976744186047, "grad_norm": 0.596676230430603, "learning_rate": 0.0005, "loss": 1.5022051334381104, "step": 25252 }, { "epoch": 1.9907765076862436, "grad_norm": 0.5029641389846802, "learning_rate": 0.0005, "loss": 1.4793977737426758, "step": 25253 }, { "epoch": 1.9908553409538825, "grad_norm": 0.5046746134757996, "learning_rate": 0.0005, "loss": 1.2943224906921387, "step": 25254 }, { "epoch": 1.9909341742215214, "grad_norm": 0.5201146006584167, "learning_rate": 0.0005, "loss": 1.3533413410186768, "step": 25255 }, { "epoch": 1.9910130074891605, "grad_norm": 1.2093440294265747, "learning_rate": 0.0005, "loss": 1.6341172456741333, "step": 25256 }, { "epoch": 1.9910918407567992, "grad_norm": 0.6113665103912354, "learning_rate": 0.0005, "loss": 1.4601151943206787, "step": 25257 }, { "epoch": 1.9911706740244384, "grad_norm": 0.508308470249176, "learning_rate": 0.0005, "loss": 1.4894795417785645, "step": 25258 }, { "epoch": 1.9912495072920773, "grad_norm": 0.5744897127151489, "learning_rate": 0.0005, "loss": 1.411222219467163, "step": 25259 }, { "epoch": 1.9913283405597162, "grad_norm": 0.5273433327674866, "learning_rate": 0.0005, "loss": 1.1691207885742188, "step": 25260 }, { "epoch": 1.991407173827355, "grad_norm": 0.4334932267665863, "learning_rate": 0.0005, "loss": 1.2532535791397095, "step": 25261 }, { "epoch": 1.991486007094994, "grad_norm": 0.5047194361686707, "learning_rate": 0.0005, "loss": 1.4960887432098389, "step": 25262 }, { "epoch": 1.9915648403626331, "grad_norm": 0.5429132580757141, "learning_rate": 0.0005, "loss": 1.5573561191558838, "step": 25263 }, { "epoch": 1.9916436736302718, "grad_norm": 0.5383802652359009, "learning_rate": 0.0005, "loss": 1.5399259328842163, "step": 25264 }, { "epoch": 1.991722506897911, "grad_norm": 0.4866751730442047, "learning_rate": 0.0005, "loss": 1.1971638202667236, "step": 25265 }, { "epoch": 1.9918013401655499, "grad_norm": 0.4893697202205658, "learning_rate": 0.0005, "loss": 1.168955683708191, "step": 25266 }, { "epoch": 1.9918801734331888, "grad_norm": 0.5565026998519897, "learning_rate": 0.0005, "loss": 1.4330863952636719, "step": 25267 }, { "epoch": 1.9919590067008277, "grad_norm": 0.6643816828727722, "learning_rate": 0.0005, "loss": 1.3913203477859497, "step": 25268 }, { "epoch": 1.9920378399684666, "grad_norm": 0.47413936257362366, "learning_rate": 0.0005, "loss": 1.4522517919540405, "step": 25269 }, { "epoch": 1.9921166732361058, "grad_norm": 0.49781593680381775, "learning_rate": 0.0005, "loss": 1.3070385456085205, "step": 25270 }, { "epoch": 1.9921955065037444, "grad_norm": 0.5969171524047852, "learning_rate": 0.0005, "loss": 1.5142583847045898, "step": 25271 }, { "epoch": 1.9922743397713836, "grad_norm": 0.5342310070991516, "learning_rate": 0.0005, "loss": 1.3152790069580078, "step": 25272 }, { "epoch": 1.9923531730390225, "grad_norm": 0.7308405041694641, "learning_rate": 0.0005, "loss": 1.4041390419006348, "step": 25273 }, { "epoch": 1.9924320063066614, "grad_norm": 0.6210900545120239, "learning_rate": 0.0005, "loss": 1.5298776626586914, "step": 25274 }, { "epoch": 1.9925108395743003, "grad_norm": 0.6455862522125244, "learning_rate": 0.0005, "loss": 1.535036325454712, "step": 25275 }, { "epoch": 1.9925896728419392, "grad_norm": 0.5841387510299683, "learning_rate": 0.0005, "loss": 1.2762507200241089, "step": 25276 }, { "epoch": 1.9926685061095784, "grad_norm": 0.6908183693885803, "learning_rate": 0.0005, "loss": 1.784210205078125, "step": 25277 }, { "epoch": 1.992747339377217, "grad_norm": 0.4618799090385437, "learning_rate": 0.0005, "loss": 1.2231063842773438, "step": 25278 }, { "epoch": 1.9928261726448562, "grad_norm": 0.6107218861579895, "learning_rate": 0.0005, "loss": 1.5809926986694336, "step": 25279 }, { "epoch": 1.992905005912495, "grad_norm": 0.5478295683860779, "learning_rate": 0.0005, "loss": 1.2578070163726807, "step": 25280 }, { "epoch": 1.992983839180134, "grad_norm": 0.5813312530517578, "learning_rate": 0.0005, "loss": 1.5125617980957031, "step": 25281 }, { "epoch": 1.993062672447773, "grad_norm": 0.543853223323822, "learning_rate": 0.0005, "loss": 1.411787986755371, "step": 25282 }, { "epoch": 1.9931415057154118, "grad_norm": 0.5666842460632324, "learning_rate": 0.0005, "loss": 1.3768882751464844, "step": 25283 }, { "epoch": 1.993220338983051, "grad_norm": 0.5024076700210571, "learning_rate": 0.0005, "loss": 1.168628454208374, "step": 25284 }, { "epoch": 1.9932991722506896, "grad_norm": 0.520977258682251, "learning_rate": 0.0005, "loss": 1.1400468349456787, "step": 25285 }, { "epoch": 1.9933780055183288, "grad_norm": 0.5179328918457031, "learning_rate": 0.0005, "loss": 1.2702174186706543, "step": 25286 }, { "epoch": 1.9934568387859677, "grad_norm": 0.7674295902252197, "learning_rate": 0.0005, "loss": 1.4005019664764404, "step": 25287 }, { "epoch": 1.9935356720536066, "grad_norm": 0.5346574187278748, "learning_rate": 0.0005, "loss": 1.3095676898956299, "step": 25288 }, { "epoch": 1.9936145053212455, "grad_norm": 0.5671604871749878, "learning_rate": 0.0005, "loss": 1.561652660369873, "step": 25289 }, { "epoch": 1.9936933385888844, "grad_norm": 0.4847627878189087, "learning_rate": 0.0005, "loss": 1.1932342052459717, "step": 25290 }, { "epoch": 1.9937721718565236, "grad_norm": 0.5049234628677368, "learning_rate": 0.0005, "loss": 1.3438193798065186, "step": 25291 }, { "epoch": 1.9938510051241622, "grad_norm": 0.5666925311088562, "learning_rate": 0.0005, "loss": 1.3183186054229736, "step": 25292 }, { "epoch": 1.9939298383918014, "grad_norm": 0.4814063608646393, "learning_rate": 0.0005, "loss": 1.3639780282974243, "step": 25293 }, { "epoch": 1.9940086716594403, "grad_norm": 1.406113862991333, "learning_rate": 0.0005, "loss": 1.4706125259399414, "step": 25294 }, { "epoch": 1.9940875049270792, "grad_norm": 0.5845130085945129, "learning_rate": 0.0005, "loss": 1.3881001472473145, "step": 25295 }, { "epoch": 1.9941663381947183, "grad_norm": 0.5785846710205078, "learning_rate": 0.0005, "loss": 1.211900234222412, "step": 25296 }, { "epoch": 1.994245171462357, "grad_norm": 0.49947065114974976, "learning_rate": 0.0005, "loss": 1.351431131362915, "step": 25297 }, { "epoch": 1.9943240047299962, "grad_norm": 0.40672147274017334, "learning_rate": 0.0005, "loss": 1.1444063186645508, "step": 25298 }, { "epoch": 1.9944028379976348, "grad_norm": 0.6287363767623901, "learning_rate": 0.0005, "loss": 1.3299822807312012, "step": 25299 }, { "epoch": 1.994481671265274, "grad_norm": 0.4239160120487213, "learning_rate": 0.0005, "loss": 1.2145264148712158, "step": 25300 }, { "epoch": 1.994560504532913, "grad_norm": 0.5497996211051941, "learning_rate": 0.0005, "loss": 1.4423081874847412, "step": 25301 }, { "epoch": 1.9946393378005518, "grad_norm": 0.4994964301586151, "learning_rate": 0.0005, "loss": 1.475836157798767, "step": 25302 }, { "epoch": 1.994718171068191, "grad_norm": 0.49658823013305664, "learning_rate": 0.0005, "loss": 1.3187031745910645, "step": 25303 }, { "epoch": 1.9947970043358296, "grad_norm": 0.494527667760849, "learning_rate": 0.0005, "loss": 1.2057603597640991, "step": 25304 }, { "epoch": 1.9948758376034688, "grad_norm": 0.5504069328308105, "learning_rate": 0.0005, "loss": 1.316467523574829, "step": 25305 }, { "epoch": 1.9949546708711074, "grad_norm": 0.4471483528614044, "learning_rate": 0.0005, "loss": 1.3145880699157715, "step": 25306 }, { "epoch": 1.9950335041387466, "grad_norm": 0.43233397603034973, "learning_rate": 0.0005, "loss": 1.387308955192566, "step": 25307 }, { "epoch": 1.9951123374063855, "grad_norm": 0.5056670904159546, "learning_rate": 0.0005, "loss": 1.4157209396362305, "step": 25308 }, { "epoch": 1.9951911706740244, "grad_norm": 0.6467129588127136, "learning_rate": 0.0005, "loss": 1.7594587802886963, "step": 25309 }, { "epoch": 1.9952700039416635, "grad_norm": 0.5362975597381592, "learning_rate": 0.0005, "loss": 1.5900368690490723, "step": 25310 }, { "epoch": 1.9953488372093022, "grad_norm": 0.7148752808570862, "learning_rate": 0.0005, "loss": 1.663668155670166, "step": 25311 }, { "epoch": 1.9954276704769414, "grad_norm": 0.5884960889816284, "learning_rate": 0.0005, "loss": 1.5303106307983398, "step": 25312 }, { "epoch": 1.99550650374458, "grad_norm": 0.7689217329025269, "learning_rate": 0.0005, "loss": 1.5837275981903076, "step": 25313 }, { "epoch": 1.9955853370122192, "grad_norm": 0.4613844156265259, "learning_rate": 0.0005, "loss": 1.2928316593170166, "step": 25314 }, { "epoch": 1.995664170279858, "grad_norm": 0.458141952753067, "learning_rate": 0.0005, "loss": 1.2630056142807007, "step": 25315 }, { "epoch": 1.995743003547497, "grad_norm": 0.5375730991363525, "learning_rate": 0.0005, "loss": 1.3118319511413574, "step": 25316 }, { "epoch": 1.9958218368151361, "grad_norm": 0.6879655718803406, "learning_rate": 0.0005, "loss": 1.5155012607574463, "step": 25317 }, { "epoch": 1.9959006700827748, "grad_norm": 0.5273895263671875, "learning_rate": 0.0005, "loss": 1.4038842916488647, "step": 25318 }, { "epoch": 1.995979503350414, "grad_norm": 0.499158650636673, "learning_rate": 0.0005, "loss": 1.4905844926834106, "step": 25319 }, { "epoch": 1.9960583366180527, "grad_norm": 0.4971262514591217, "learning_rate": 0.0005, "loss": 1.4397337436676025, "step": 25320 }, { "epoch": 1.9961371698856918, "grad_norm": 0.6314005255699158, "learning_rate": 0.0005, "loss": 1.4091708660125732, "step": 25321 }, { "epoch": 1.9962160031533307, "grad_norm": 0.502153217792511, "learning_rate": 0.0005, "loss": 1.3684523105621338, "step": 25322 }, { "epoch": 1.9962948364209696, "grad_norm": 0.49439480900764465, "learning_rate": 0.0005, "loss": 1.3546202182769775, "step": 25323 }, { "epoch": 1.9963736696886087, "grad_norm": 0.5416577458381653, "learning_rate": 0.0005, "loss": 1.4259529113769531, "step": 25324 }, { "epoch": 1.9964525029562474, "grad_norm": 0.5412954092025757, "learning_rate": 0.0005, "loss": 1.5770971775054932, "step": 25325 }, { "epoch": 1.9965313362238866, "grad_norm": 0.9008951187133789, "learning_rate": 0.0005, "loss": 1.3117637634277344, "step": 25326 }, { "epoch": 1.9966101694915255, "grad_norm": 0.5557647347450256, "learning_rate": 0.0005, "loss": 1.423335313796997, "step": 25327 }, { "epoch": 1.9966890027591644, "grad_norm": 0.48188236355781555, "learning_rate": 0.0005, "loss": 1.3224692344665527, "step": 25328 }, { "epoch": 1.9967678360268033, "grad_norm": 0.6460180282592773, "learning_rate": 0.0005, "loss": 1.4753332138061523, "step": 25329 }, { "epoch": 1.9968466692944422, "grad_norm": 0.43567612767219543, "learning_rate": 0.0005, "loss": 1.4378561973571777, "step": 25330 }, { "epoch": 1.9969255025620813, "grad_norm": 0.5610160231590271, "learning_rate": 0.0005, "loss": 1.3426423072814941, "step": 25331 }, { "epoch": 1.99700433582972, "grad_norm": 0.5857498645782471, "learning_rate": 0.0005, "loss": 1.397874355316162, "step": 25332 }, { "epoch": 1.9970831690973592, "grad_norm": 0.5381183624267578, "learning_rate": 0.0005, "loss": 1.432317852973938, "step": 25333 }, { "epoch": 1.997162002364998, "grad_norm": 0.5571179986000061, "learning_rate": 0.0005, "loss": 1.3745100498199463, "step": 25334 }, { "epoch": 1.997240835632637, "grad_norm": 0.47819605469703674, "learning_rate": 0.0005, "loss": 1.3440332412719727, "step": 25335 }, { "epoch": 1.997319668900276, "grad_norm": 0.69869464635849, "learning_rate": 0.0005, "loss": 1.3052319288253784, "step": 25336 }, { "epoch": 1.9973985021679148, "grad_norm": 0.5044248700141907, "learning_rate": 0.0005, "loss": 1.4296523332595825, "step": 25337 }, { "epoch": 1.997477335435554, "grad_norm": 0.4761711657047272, "learning_rate": 0.0005, "loss": 1.3579821586608887, "step": 25338 }, { "epoch": 1.9975561687031926, "grad_norm": 0.4909198582172394, "learning_rate": 0.0005, "loss": 1.3232920169830322, "step": 25339 }, { "epoch": 1.9976350019708318, "grad_norm": 0.4825349748134613, "learning_rate": 0.0005, "loss": 1.3694640398025513, "step": 25340 }, { "epoch": 1.9977138352384707, "grad_norm": 0.4595666229724884, "learning_rate": 0.0005, "loss": 1.2524449825286865, "step": 25341 }, { "epoch": 1.9977926685061096, "grad_norm": 0.4704570174217224, "learning_rate": 0.0005, "loss": 1.2984308004379272, "step": 25342 }, { "epoch": 1.9978715017737485, "grad_norm": 0.6014832854270935, "learning_rate": 0.0005, "loss": 1.8274905681610107, "step": 25343 }, { "epoch": 1.9979503350413874, "grad_norm": 0.4757893681526184, "learning_rate": 0.0005, "loss": 1.3605694770812988, "step": 25344 }, { "epoch": 1.9980291683090265, "grad_norm": 0.487394779920578, "learning_rate": 0.0005, "loss": 1.443488597869873, "step": 25345 }, { "epoch": 1.9981080015766652, "grad_norm": 0.5097968578338623, "learning_rate": 0.0005, "loss": 1.3535382747650146, "step": 25346 }, { "epoch": 1.9981868348443044, "grad_norm": 0.7971606254577637, "learning_rate": 0.0005, "loss": 1.3156540393829346, "step": 25347 }, { "epoch": 1.9982656681119433, "grad_norm": 0.5645010471343994, "learning_rate": 0.0005, "loss": 1.5232435464859009, "step": 25348 }, { "epoch": 1.9983445013795822, "grad_norm": 0.4695464074611664, "learning_rate": 0.0005, "loss": 1.5214475393295288, "step": 25349 }, { "epoch": 1.998423334647221, "grad_norm": 0.5036425590515137, "learning_rate": 0.0005, "loss": 1.3599472045898438, "step": 25350 }, { "epoch": 1.99850216791486, "grad_norm": 0.5453616976737976, "learning_rate": 0.0005, "loss": 1.4996635913848877, "step": 25351 }, { "epoch": 1.9985810011824992, "grad_norm": 0.4820610582828522, "learning_rate": 0.0005, "loss": 1.4571399688720703, "step": 25352 }, { "epoch": 1.9986598344501378, "grad_norm": 0.44884994626045227, "learning_rate": 0.0005, "loss": 1.3835482597351074, "step": 25353 }, { "epoch": 1.998738667717777, "grad_norm": 0.5154063105583191, "learning_rate": 0.0005, "loss": 1.203787922859192, "step": 25354 }, { "epoch": 1.9988175009854159, "grad_norm": 0.45287129282951355, "learning_rate": 0.0005, "loss": 1.3969167470932007, "step": 25355 }, { "epoch": 1.9988963342530548, "grad_norm": 0.43558764457702637, "learning_rate": 0.0005, "loss": 1.2654049396514893, "step": 25356 }, { "epoch": 1.9989751675206937, "grad_norm": 0.7166481018066406, "learning_rate": 0.0005, "loss": 1.3247942924499512, "step": 25357 }, { "epoch": 1.9990540007883326, "grad_norm": 0.6128932237625122, "learning_rate": 0.0005, "loss": 1.5097525119781494, "step": 25358 }, { "epoch": 1.9991328340559718, "grad_norm": 0.46733686327934265, "learning_rate": 0.0005, "loss": 1.3807382583618164, "step": 25359 }, { "epoch": 1.9992116673236104, "grad_norm": 0.49091199040412903, "learning_rate": 0.0005, "loss": 1.5517324209213257, "step": 25360 }, { "epoch": 1.9992905005912496, "grad_norm": 0.6090822219848633, "learning_rate": 0.0005, "loss": 1.5844299793243408, "step": 25361 }, { "epoch": 1.9993693338588885, "grad_norm": 0.6873476505279541, "learning_rate": 0.0005, "loss": 1.2963311672210693, "step": 25362 }, { "epoch": 1.9994481671265274, "grad_norm": 0.7117175459861755, "learning_rate": 0.0005, "loss": 1.4582147598266602, "step": 25363 }, { "epoch": 1.9995270003941663, "grad_norm": 0.5071216225624084, "learning_rate": 0.0005, "loss": 1.3106682300567627, "step": 25364 }, { "epoch": 1.9996058336618052, "grad_norm": 0.6262487173080444, "learning_rate": 0.0005, "loss": 1.4439027309417725, "step": 25365 }, { "epoch": 1.9996846669294444, "grad_norm": 0.47458264231681824, "learning_rate": 0.0005, "loss": 1.3750312328338623, "step": 25366 }, { "epoch": 1.999763500197083, "grad_norm": 0.74904465675354, "learning_rate": 0.0005, "loss": 1.5205800533294678, "step": 25367 }, { "epoch": 1.9998423334647222, "grad_norm": 0.5013642907142639, "learning_rate": 0.0005, "loss": 1.259000301361084, "step": 25368 }, { "epoch": 1.999921166732361, "grad_norm": 0.4805360436439514, "learning_rate": 0.0005, "loss": 1.3346930742263794, "step": 25369 }, { "epoch": 2.0, "grad_norm": 0.6130425333976746, "learning_rate": 0.0005, "loss": 1.5335655212402344, "step": 25370 }, { "epoch": 2.000078833267639, "grad_norm": 0.4844476282596588, "learning_rate": 0.0005, "loss": 1.375461220741272, "step": 25371 }, { "epoch": 2.000157666535278, "grad_norm": 0.43867602944374084, "learning_rate": 0.0005, "loss": 1.3876852989196777, "step": 25372 }, { "epoch": 2.000236499802917, "grad_norm": 0.46498817205429077, "learning_rate": 0.0005, "loss": 1.2597548961639404, "step": 25373 }, { "epoch": 2.0003153330705556, "grad_norm": 0.44054630398750305, "learning_rate": 0.0005, "loss": 1.2905182838439941, "step": 25374 }, { "epoch": 2.000394166338195, "grad_norm": 0.5131959915161133, "learning_rate": 0.0005, "loss": 1.4223661422729492, "step": 25375 }, { "epoch": 2.0004729996058335, "grad_norm": 0.598905622959137, "learning_rate": 0.0005, "loss": 1.377525806427002, "step": 25376 }, { "epoch": 2.0005518328734726, "grad_norm": 0.5742391347885132, "learning_rate": 0.0005, "loss": 1.413971185684204, "step": 25377 }, { "epoch": 2.0006306661411117, "grad_norm": 0.4286418855190277, "learning_rate": 0.0005, "loss": 1.266080617904663, "step": 25378 }, { "epoch": 2.0007094994087504, "grad_norm": 0.45555368065834045, "learning_rate": 0.0005, "loss": 1.1955821514129639, "step": 25379 }, { "epoch": 2.0007883326763896, "grad_norm": 0.4470624029636383, "learning_rate": 0.0005, "loss": 1.181827187538147, "step": 25380 }, { "epoch": 2.0008671659440282, "grad_norm": 0.4352395832538605, "learning_rate": 0.0005, "loss": 1.2600064277648926, "step": 25381 }, { "epoch": 2.0009459992116674, "grad_norm": 0.5247476696968079, "learning_rate": 0.0005, "loss": 1.428823471069336, "step": 25382 }, { "epoch": 2.001024832479306, "grad_norm": 0.5813039541244507, "learning_rate": 0.0005, "loss": 1.2580926418304443, "step": 25383 }, { "epoch": 2.001103665746945, "grad_norm": 0.4848006069660187, "learning_rate": 0.0005, "loss": 1.2119460105895996, "step": 25384 }, { "epoch": 2.0011824990145843, "grad_norm": 0.6344539523124695, "learning_rate": 0.0005, "loss": 1.249845027923584, "step": 25385 }, { "epoch": 2.001261332282223, "grad_norm": 0.8027376532554626, "learning_rate": 0.0005, "loss": 1.2106051445007324, "step": 25386 }, { "epoch": 2.001340165549862, "grad_norm": 0.764394223690033, "learning_rate": 0.0005, "loss": 1.5260225534439087, "step": 25387 }, { "epoch": 2.001418998817501, "grad_norm": 0.587064266204834, "learning_rate": 0.0005, "loss": 1.2738280296325684, "step": 25388 }, { "epoch": 2.00149783208514, "grad_norm": 0.4803970456123352, "learning_rate": 0.0005, "loss": 1.2576799392700195, "step": 25389 }, { "epoch": 2.0015766653527787, "grad_norm": 0.9015636444091797, "learning_rate": 0.0005, "loss": 1.346237301826477, "step": 25390 }, { "epoch": 2.001655498620418, "grad_norm": 0.4310043156147003, "learning_rate": 0.0005, "loss": 1.275752067565918, "step": 25391 }, { "epoch": 2.001734331888057, "grad_norm": 0.4300380349159241, "learning_rate": 0.0005, "loss": 1.2761034965515137, "step": 25392 }, { "epoch": 2.0018131651556956, "grad_norm": 0.6402103304862976, "learning_rate": 0.0005, "loss": 1.4627461433410645, "step": 25393 }, { "epoch": 2.0018919984233348, "grad_norm": 0.5015472769737244, "learning_rate": 0.0005, "loss": 1.509500503540039, "step": 25394 }, { "epoch": 2.0019708316909735, "grad_norm": 0.5646257400512695, "learning_rate": 0.0005, "loss": 1.4458236694335938, "step": 25395 }, { "epoch": 2.0020496649586126, "grad_norm": 0.5070555806159973, "learning_rate": 0.0005, "loss": 1.597170114517212, "step": 25396 }, { "epoch": 2.0021284982262513, "grad_norm": 0.4721750319004059, "learning_rate": 0.0005, "loss": 1.304668664932251, "step": 25397 }, { "epoch": 2.0022073314938904, "grad_norm": 0.4696309566497803, "learning_rate": 0.0005, "loss": 1.2357374429702759, "step": 25398 }, { "epoch": 2.0022861647615295, "grad_norm": 0.5781183242797852, "learning_rate": 0.0005, "loss": 1.1725385189056396, "step": 25399 }, { "epoch": 2.0023649980291682, "grad_norm": 0.57637619972229, "learning_rate": 0.0005, "loss": 1.3733835220336914, "step": 25400 }, { "epoch": 2.0024438312968074, "grad_norm": 0.4089905917644501, "learning_rate": 0.0005, "loss": 1.0265905857086182, "step": 25401 }, { "epoch": 2.002522664564446, "grad_norm": 0.6082478165626526, "learning_rate": 0.0005, "loss": 1.2227888107299805, "step": 25402 }, { "epoch": 2.002601497832085, "grad_norm": 0.5416755080223083, "learning_rate": 0.0005, "loss": 1.42763352394104, "step": 25403 }, { "epoch": 2.002680331099724, "grad_norm": 0.4749482274055481, "learning_rate": 0.0005, "loss": 1.1968666315078735, "step": 25404 }, { "epoch": 2.002759164367363, "grad_norm": 0.515005350112915, "learning_rate": 0.0005, "loss": 1.3500325679779053, "step": 25405 }, { "epoch": 2.002837997635002, "grad_norm": 0.5088449120521545, "learning_rate": 0.0005, "loss": 1.555507779121399, "step": 25406 }, { "epoch": 2.002916830902641, "grad_norm": 0.43033722043037415, "learning_rate": 0.0005, "loss": 1.109618902206421, "step": 25407 }, { "epoch": 2.00299566417028, "grad_norm": 0.5137749314308167, "learning_rate": 0.0005, "loss": 1.3042595386505127, "step": 25408 }, { "epoch": 2.0030744974379187, "grad_norm": 0.5421142578125, "learning_rate": 0.0005, "loss": 1.5197232961654663, "step": 25409 }, { "epoch": 2.003153330705558, "grad_norm": 0.5163419246673584, "learning_rate": 0.0005, "loss": 1.3435472249984741, "step": 25410 }, { "epoch": 2.0032321639731965, "grad_norm": 0.4653502106666565, "learning_rate": 0.0005, "loss": 1.3997843265533447, "step": 25411 }, { "epoch": 2.0033109972408356, "grad_norm": 2.672546148300171, "learning_rate": 0.0005, "loss": 1.4029603004455566, "step": 25412 }, { "epoch": 2.0033898305084747, "grad_norm": 0.5766341686248779, "learning_rate": 0.0005, "loss": 1.4545540809631348, "step": 25413 }, { "epoch": 2.0034686637761134, "grad_norm": 0.588307797908783, "learning_rate": 0.0005, "loss": 1.6660892963409424, "step": 25414 }, { "epoch": 2.0035474970437526, "grad_norm": 0.4816195070743561, "learning_rate": 0.0005, "loss": 1.4098527431488037, "step": 25415 }, { "epoch": 2.0036263303113913, "grad_norm": 0.5443080067634583, "learning_rate": 0.0005, "loss": 1.3438050746917725, "step": 25416 }, { "epoch": 2.0037051635790304, "grad_norm": 0.8250000476837158, "learning_rate": 0.0005, "loss": 1.4603729248046875, "step": 25417 }, { "epoch": 2.003783996846669, "grad_norm": 0.524767279624939, "learning_rate": 0.0005, "loss": 1.398042917251587, "step": 25418 }, { "epoch": 2.003862830114308, "grad_norm": 0.5616950392723083, "learning_rate": 0.0005, "loss": 1.4839541912078857, "step": 25419 }, { "epoch": 2.0039416633819473, "grad_norm": 0.4602590799331665, "learning_rate": 0.0005, "loss": 1.2321275472640991, "step": 25420 }, { "epoch": 2.004020496649586, "grad_norm": 0.490573525428772, "learning_rate": 0.0005, "loss": 1.4212117195129395, "step": 25421 }, { "epoch": 2.004099329917225, "grad_norm": 0.5087755918502808, "learning_rate": 0.0005, "loss": 1.488010287284851, "step": 25422 }, { "epoch": 2.004178163184864, "grad_norm": 0.41953569650650024, "learning_rate": 0.0005, "loss": 1.3554562330245972, "step": 25423 }, { "epoch": 2.004256996452503, "grad_norm": 0.4838646948337555, "learning_rate": 0.0005, "loss": 1.291334867477417, "step": 25424 }, { "epoch": 2.0043358297201417, "grad_norm": 0.4726409316062927, "learning_rate": 0.0005, "loss": 1.260172963142395, "step": 25425 }, { "epoch": 2.004414662987781, "grad_norm": 0.47952184081077576, "learning_rate": 0.0005, "loss": 1.2601304054260254, "step": 25426 }, { "epoch": 2.00449349625542, "grad_norm": 0.5434567928314209, "learning_rate": 0.0005, "loss": 1.452139139175415, "step": 25427 }, { "epoch": 2.0045723295230586, "grad_norm": 0.6623910665512085, "learning_rate": 0.0005, "loss": 1.6306099891662598, "step": 25428 }, { "epoch": 2.0046511627906978, "grad_norm": 0.4650191068649292, "learning_rate": 0.0005, "loss": 1.2757251262664795, "step": 25429 }, { "epoch": 2.0047299960583365, "grad_norm": 0.46563127636909485, "learning_rate": 0.0005, "loss": 1.2851639986038208, "step": 25430 }, { "epoch": 2.0048088293259756, "grad_norm": 0.42642128467559814, "learning_rate": 0.0005, "loss": 1.3169766664505005, "step": 25431 }, { "epoch": 2.0048876625936147, "grad_norm": 0.6218727827072144, "learning_rate": 0.0005, "loss": 1.498189091682434, "step": 25432 }, { "epoch": 2.0049664958612534, "grad_norm": 0.5062916278839111, "learning_rate": 0.0005, "loss": 1.4890153408050537, "step": 25433 }, { "epoch": 2.0050453291288926, "grad_norm": 0.4972400963306427, "learning_rate": 0.0005, "loss": 1.3318195343017578, "step": 25434 }, { "epoch": 2.0051241623965312, "grad_norm": 0.5835239291191101, "learning_rate": 0.0005, "loss": 1.209925651550293, "step": 25435 }, { "epoch": 2.0052029956641704, "grad_norm": 0.41708904504776, "learning_rate": 0.0005, "loss": 1.2621169090270996, "step": 25436 }, { "epoch": 2.005281828931809, "grad_norm": 0.46543389558792114, "learning_rate": 0.0005, "loss": 1.3597073554992676, "step": 25437 }, { "epoch": 2.005360662199448, "grad_norm": 0.36465680599212646, "learning_rate": 0.0005, "loss": 1.0608340501785278, "step": 25438 }, { "epoch": 2.0054394954670873, "grad_norm": 0.5841076970100403, "learning_rate": 0.0005, "loss": 1.3806310892105103, "step": 25439 }, { "epoch": 2.005518328734726, "grad_norm": 0.503609836101532, "learning_rate": 0.0005, "loss": 1.323622226715088, "step": 25440 }, { "epoch": 2.005597162002365, "grad_norm": 0.5791953802108765, "learning_rate": 0.0005, "loss": 1.4395747184753418, "step": 25441 }, { "epoch": 2.005675995270004, "grad_norm": 0.48905718326568604, "learning_rate": 0.0005, "loss": 1.3635263442993164, "step": 25442 }, { "epoch": 2.005754828537643, "grad_norm": 0.4839935600757599, "learning_rate": 0.0005, "loss": 1.5210509300231934, "step": 25443 }, { "epoch": 2.0058336618052817, "grad_norm": 0.4537733197212219, "learning_rate": 0.0005, "loss": 1.4106544256210327, "step": 25444 }, { "epoch": 2.005912495072921, "grad_norm": 0.47243115305900574, "learning_rate": 0.0005, "loss": 1.341402530670166, "step": 25445 }, { "epoch": 2.00599132834056, "grad_norm": 0.4487893879413605, "learning_rate": 0.0005, "loss": 1.4072049856185913, "step": 25446 }, { "epoch": 2.0060701616081986, "grad_norm": 0.5342990159988403, "learning_rate": 0.0005, "loss": 1.2562040090560913, "step": 25447 }, { "epoch": 2.0061489948758378, "grad_norm": 0.45543941855430603, "learning_rate": 0.0005, "loss": 1.2916810512542725, "step": 25448 }, { "epoch": 2.0062278281434764, "grad_norm": 0.5773393511772156, "learning_rate": 0.0005, "loss": 1.334897756576538, "step": 25449 }, { "epoch": 2.0063066614111156, "grad_norm": 0.5145148038864136, "learning_rate": 0.0005, "loss": 1.3798929452896118, "step": 25450 }, { "epoch": 2.0063854946787543, "grad_norm": 0.6435299515724182, "learning_rate": 0.0005, "loss": 1.3421844244003296, "step": 25451 }, { "epoch": 2.0064643279463934, "grad_norm": 0.5692755579948425, "learning_rate": 0.0005, "loss": 1.3954136371612549, "step": 25452 }, { "epoch": 2.0065431612140325, "grad_norm": 0.44034063816070557, "learning_rate": 0.0005, "loss": 1.1998955011367798, "step": 25453 }, { "epoch": 2.0066219944816712, "grad_norm": 0.5989120602607727, "learning_rate": 0.0005, "loss": 1.5472028255462646, "step": 25454 }, { "epoch": 2.0067008277493104, "grad_norm": 0.4457125663757324, "learning_rate": 0.0005, "loss": 1.2978962659835815, "step": 25455 }, { "epoch": 2.006779661016949, "grad_norm": 0.49546030163764954, "learning_rate": 0.0005, "loss": 1.3003143072128296, "step": 25456 }, { "epoch": 2.006858494284588, "grad_norm": 0.5529619455337524, "learning_rate": 0.0005, "loss": 1.4007679224014282, "step": 25457 }, { "epoch": 2.006937327552227, "grad_norm": 0.4853510856628418, "learning_rate": 0.0005, "loss": 1.3992544412612915, "step": 25458 }, { "epoch": 2.007016160819866, "grad_norm": 0.43212538957595825, "learning_rate": 0.0005, "loss": 1.1527636051177979, "step": 25459 }, { "epoch": 2.007094994087505, "grad_norm": 0.5432956218719482, "learning_rate": 0.0005, "loss": 1.3995989561080933, "step": 25460 }, { "epoch": 2.007173827355144, "grad_norm": 0.5370071530342102, "learning_rate": 0.0005, "loss": 1.5219595432281494, "step": 25461 }, { "epoch": 2.007252660622783, "grad_norm": 0.47514304518699646, "learning_rate": 0.0005, "loss": 1.1653592586517334, "step": 25462 }, { "epoch": 2.0073314938904216, "grad_norm": 0.49271371960639954, "learning_rate": 0.0005, "loss": 1.2276506423950195, "step": 25463 }, { "epoch": 2.007410327158061, "grad_norm": 0.41672250628471375, "learning_rate": 0.0005, "loss": 1.1517879962921143, "step": 25464 }, { "epoch": 2.0074891604256995, "grad_norm": 0.5004397630691528, "learning_rate": 0.0005, "loss": 1.2683063745498657, "step": 25465 }, { "epoch": 2.0075679936933386, "grad_norm": 0.3457619547843933, "learning_rate": 0.0005, "loss": 1.1331543922424316, "step": 25466 }, { "epoch": 2.0076468269609777, "grad_norm": 0.4666423499584198, "learning_rate": 0.0005, "loss": 1.413119912147522, "step": 25467 }, { "epoch": 2.0077256602286164, "grad_norm": 0.5314549803733826, "learning_rate": 0.0005, "loss": 1.3330166339874268, "step": 25468 }, { "epoch": 2.0078044934962556, "grad_norm": 0.5261837840080261, "learning_rate": 0.0005, "loss": 1.600661277770996, "step": 25469 }, { "epoch": 2.0078833267638942, "grad_norm": 0.5177896022796631, "learning_rate": 0.0005, "loss": 1.4704152345657349, "step": 25470 }, { "epoch": 2.0079621600315334, "grad_norm": 0.48356741666793823, "learning_rate": 0.0005, "loss": 1.0975641012191772, "step": 25471 }, { "epoch": 2.008040993299172, "grad_norm": 0.6868694424629211, "learning_rate": 0.0005, "loss": 1.3667829036712646, "step": 25472 }, { "epoch": 2.008119826566811, "grad_norm": 0.44305387139320374, "learning_rate": 0.0005, "loss": 1.4085092544555664, "step": 25473 }, { "epoch": 2.0081986598344503, "grad_norm": 0.6382785439491272, "learning_rate": 0.0005, "loss": 1.5021107196807861, "step": 25474 }, { "epoch": 2.008277493102089, "grad_norm": 0.6003319621086121, "learning_rate": 0.0005, "loss": 1.5757267475128174, "step": 25475 }, { "epoch": 2.008356326369728, "grad_norm": 0.4983018934726715, "learning_rate": 0.0005, "loss": 1.3401193618774414, "step": 25476 }, { "epoch": 2.008435159637367, "grad_norm": 0.5308544039726257, "learning_rate": 0.0005, "loss": 1.3524699211120605, "step": 25477 }, { "epoch": 2.008513992905006, "grad_norm": 0.56147700548172, "learning_rate": 0.0005, "loss": 1.3813574314117432, "step": 25478 }, { "epoch": 2.0085928261726447, "grad_norm": 0.6334036588668823, "learning_rate": 0.0005, "loss": 1.5029643774032593, "step": 25479 }, { "epoch": 2.008671659440284, "grad_norm": 0.4877364933490753, "learning_rate": 0.0005, "loss": 1.239137887954712, "step": 25480 }, { "epoch": 2.008750492707923, "grad_norm": 1.184895634651184, "learning_rate": 0.0005, "loss": 1.284862756729126, "step": 25481 }, { "epoch": 2.0088293259755616, "grad_norm": 0.511462390422821, "learning_rate": 0.0005, "loss": 1.3351476192474365, "step": 25482 }, { "epoch": 2.0089081592432008, "grad_norm": 0.37895849347114563, "learning_rate": 0.0005, "loss": 1.1459834575653076, "step": 25483 }, { "epoch": 2.0089869925108395, "grad_norm": 0.5166416764259338, "learning_rate": 0.0005, "loss": 1.3633836507797241, "step": 25484 }, { "epoch": 2.0090658257784786, "grad_norm": 0.46985867619514465, "learning_rate": 0.0005, "loss": 1.3613977432250977, "step": 25485 }, { "epoch": 2.0091446590461173, "grad_norm": 0.44571453332901, "learning_rate": 0.0005, "loss": 1.2137833833694458, "step": 25486 }, { "epoch": 2.0092234923137564, "grad_norm": 0.5105843544006348, "learning_rate": 0.0005, "loss": 1.3968849182128906, "step": 25487 }, { "epoch": 2.0093023255813955, "grad_norm": 0.46019840240478516, "learning_rate": 0.0005, "loss": 1.5328681468963623, "step": 25488 }, { "epoch": 2.0093811588490342, "grad_norm": 0.5308184623718262, "learning_rate": 0.0005, "loss": 1.31915283203125, "step": 25489 }, { "epoch": 2.0094599921166734, "grad_norm": 0.4481327533721924, "learning_rate": 0.0005, "loss": 1.0412920713424683, "step": 25490 }, { "epoch": 2.009538825384312, "grad_norm": 0.49038323760032654, "learning_rate": 0.0005, "loss": 1.3150345087051392, "step": 25491 }, { "epoch": 2.009617658651951, "grad_norm": 0.49161550402641296, "learning_rate": 0.0005, "loss": 1.2936291694641113, "step": 25492 }, { "epoch": 2.00969649191959, "grad_norm": 0.4636573791503906, "learning_rate": 0.0005, "loss": 1.2377543449401855, "step": 25493 }, { "epoch": 2.009775325187229, "grad_norm": 0.4320600628852844, "learning_rate": 0.0005, "loss": 1.197031021118164, "step": 25494 }, { "epoch": 2.009854158454868, "grad_norm": 0.5236601829528809, "learning_rate": 0.0005, "loss": 1.439544677734375, "step": 25495 }, { "epoch": 2.009932991722507, "grad_norm": 0.4916367828845978, "learning_rate": 0.0005, "loss": 1.310528039932251, "step": 25496 }, { "epoch": 2.010011824990146, "grad_norm": 0.5646289587020874, "learning_rate": 0.0005, "loss": 1.4772734642028809, "step": 25497 }, { "epoch": 2.0100906582577847, "grad_norm": 0.5480055212974548, "learning_rate": 0.0005, "loss": 1.402064323425293, "step": 25498 }, { "epoch": 2.010169491525424, "grad_norm": 0.45783984661102295, "learning_rate": 0.0005, "loss": 1.4072743654251099, "step": 25499 }, { "epoch": 2.0102483247930625, "grad_norm": 0.6328992247581482, "learning_rate": 0.0005, "loss": 1.2506279945373535, "step": 25500 }, { "epoch": 2.0103271580607016, "grad_norm": 0.42806127667427063, "learning_rate": 0.0005, "loss": 1.3310701847076416, "step": 25501 }, { "epoch": 2.0104059913283407, "grad_norm": 0.4474397897720337, "learning_rate": 0.0005, "loss": 1.3241279125213623, "step": 25502 }, { "epoch": 2.0104848245959794, "grad_norm": 0.4508497714996338, "learning_rate": 0.0005, "loss": 1.3329271078109741, "step": 25503 }, { "epoch": 2.0105636578636186, "grad_norm": 0.44219154119491577, "learning_rate": 0.0005, "loss": 1.4035205841064453, "step": 25504 }, { "epoch": 2.0106424911312573, "grad_norm": 0.5095399618148804, "learning_rate": 0.0005, "loss": 1.431404948234558, "step": 25505 }, { "epoch": 2.0107213243988964, "grad_norm": 0.5269772410392761, "learning_rate": 0.0005, "loss": 1.2031526565551758, "step": 25506 }, { "epoch": 2.010800157666535, "grad_norm": 0.4010747969150543, "learning_rate": 0.0005, "loss": 1.4174913167953491, "step": 25507 }, { "epoch": 2.010878990934174, "grad_norm": 0.44916826486587524, "learning_rate": 0.0005, "loss": 1.2310924530029297, "step": 25508 }, { "epoch": 2.0109578242018133, "grad_norm": 0.40948548913002014, "learning_rate": 0.0005, "loss": 1.215998649597168, "step": 25509 }, { "epoch": 2.011036657469452, "grad_norm": 0.43039047718048096, "learning_rate": 0.0005, "loss": 1.343451976776123, "step": 25510 }, { "epoch": 2.011115490737091, "grad_norm": 0.420984148979187, "learning_rate": 0.0005, "loss": 1.213666558265686, "step": 25511 }, { "epoch": 2.01119432400473, "grad_norm": 0.45632311701774597, "learning_rate": 0.0005, "loss": 1.1614208221435547, "step": 25512 }, { "epoch": 2.011273157272369, "grad_norm": 0.5607126355171204, "learning_rate": 0.0005, "loss": 1.4346364736557007, "step": 25513 }, { "epoch": 2.0113519905400077, "grad_norm": 3.6429386138916016, "learning_rate": 0.0005, "loss": 1.2243046760559082, "step": 25514 }, { "epoch": 2.011430823807647, "grad_norm": 0.5219910144805908, "learning_rate": 0.0005, "loss": 1.4383764266967773, "step": 25515 }, { "epoch": 2.011509657075286, "grad_norm": 0.4306250512599945, "learning_rate": 0.0005, "loss": 1.2037065029144287, "step": 25516 }, { "epoch": 2.0115884903429246, "grad_norm": 0.5726975202560425, "learning_rate": 0.0005, "loss": 1.4660406112670898, "step": 25517 }, { "epoch": 2.0116673236105638, "grad_norm": 0.44023922085762024, "learning_rate": 0.0005, "loss": 1.1453418731689453, "step": 25518 }, { "epoch": 2.0117461568782025, "grad_norm": 0.49748262763023376, "learning_rate": 0.0005, "loss": 1.1518667936325073, "step": 25519 }, { "epoch": 2.0118249901458416, "grad_norm": 0.4656429886817932, "learning_rate": 0.0005, "loss": 1.4784436225891113, "step": 25520 }, { "epoch": 2.0119038234134803, "grad_norm": 0.5438622236251831, "learning_rate": 0.0005, "loss": 1.3469806909561157, "step": 25521 }, { "epoch": 2.0119826566811194, "grad_norm": 0.6243153810501099, "learning_rate": 0.0005, "loss": 1.4737272262573242, "step": 25522 }, { "epoch": 2.0120614899487586, "grad_norm": 0.41985422372817993, "learning_rate": 0.0005, "loss": 1.2358949184417725, "step": 25523 }, { "epoch": 2.0121403232163972, "grad_norm": 0.4566446840763092, "learning_rate": 0.0005, "loss": 1.2680182456970215, "step": 25524 }, { "epoch": 2.0122191564840364, "grad_norm": 0.5203542113304138, "learning_rate": 0.0005, "loss": 1.3502378463745117, "step": 25525 }, { "epoch": 2.012297989751675, "grad_norm": 0.4565547704696655, "learning_rate": 0.0005, "loss": 1.394946813583374, "step": 25526 }, { "epoch": 2.012376823019314, "grad_norm": 0.5016083121299744, "learning_rate": 0.0005, "loss": 1.1276410818099976, "step": 25527 }, { "epoch": 2.012455656286953, "grad_norm": 0.48240405321121216, "learning_rate": 0.0005, "loss": 1.2072207927703857, "step": 25528 }, { "epoch": 2.012534489554592, "grad_norm": 0.538972020149231, "learning_rate": 0.0005, "loss": 1.2292660474777222, "step": 25529 }, { "epoch": 2.012613322822231, "grad_norm": 0.49811869859695435, "learning_rate": 0.0005, "loss": 1.563690423965454, "step": 25530 }, { "epoch": 2.01269215608987, "grad_norm": 0.4904085695743561, "learning_rate": 0.0005, "loss": 1.3615241050720215, "step": 25531 }, { "epoch": 2.012770989357509, "grad_norm": 0.5293667316436768, "learning_rate": 0.0005, "loss": 1.1203120946884155, "step": 25532 }, { "epoch": 2.0128498226251477, "grad_norm": 0.6459764242172241, "learning_rate": 0.0005, "loss": 1.3767626285552979, "step": 25533 }, { "epoch": 2.012928655892787, "grad_norm": 0.5947456359863281, "learning_rate": 0.0005, "loss": 1.3521170616149902, "step": 25534 }, { "epoch": 2.0130074891604255, "grad_norm": 0.4543931186199188, "learning_rate": 0.0005, "loss": 1.3047919273376465, "step": 25535 }, { "epoch": 2.0130863224280646, "grad_norm": 0.569966197013855, "learning_rate": 0.0005, "loss": 1.3110196590423584, "step": 25536 }, { "epoch": 2.0131651556957038, "grad_norm": 0.6172413229942322, "learning_rate": 0.0005, "loss": 1.3118999004364014, "step": 25537 }, { "epoch": 2.0132439889633424, "grad_norm": 0.5582471489906311, "learning_rate": 0.0005, "loss": 1.289313554763794, "step": 25538 }, { "epoch": 2.0133228222309816, "grad_norm": 0.5578001737594604, "learning_rate": 0.0005, "loss": 1.5957285165786743, "step": 25539 }, { "epoch": 2.0134016554986203, "grad_norm": 0.4884175658226013, "learning_rate": 0.0005, "loss": 1.3328396081924438, "step": 25540 }, { "epoch": 2.0134804887662594, "grad_norm": 0.4791537821292877, "learning_rate": 0.0005, "loss": 1.1804916858673096, "step": 25541 }, { "epoch": 2.013559322033898, "grad_norm": 0.4645650386810303, "learning_rate": 0.0005, "loss": 1.2461466789245605, "step": 25542 }, { "epoch": 2.0136381553015372, "grad_norm": 0.430136501789093, "learning_rate": 0.0005, "loss": 1.2181122303009033, "step": 25543 }, { "epoch": 2.0137169885691764, "grad_norm": 0.44457492232322693, "learning_rate": 0.0005, "loss": 1.2906230688095093, "step": 25544 }, { "epoch": 2.013795821836815, "grad_norm": 0.4759080410003662, "learning_rate": 0.0005, "loss": 1.399078607559204, "step": 25545 }, { "epoch": 2.013874655104454, "grad_norm": 0.48601576685905457, "learning_rate": 0.0005, "loss": 1.4377343654632568, "step": 25546 }, { "epoch": 2.013953488372093, "grad_norm": 0.5108949542045593, "learning_rate": 0.0005, "loss": 1.2589443922042847, "step": 25547 }, { "epoch": 2.014032321639732, "grad_norm": 0.5167447328567505, "learning_rate": 0.0005, "loss": 1.4057612419128418, "step": 25548 }, { "epoch": 2.0141111549073707, "grad_norm": 0.5149713754653931, "learning_rate": 0.0005, "loss": 1.3785622119903564, "step": 25549 }, { "epoch": 2.01418998817501, "grad_norm": 0.8121715784072876, "learning_rate": 0.0005, "loss": 1.4475488662719727, "step": 25550 }, { "epoch": 2.014268821442649, "grad_norm": 0.4988751709461212, "learning_rate": 0.0005, "loss": 1.1797025203704834, "step": 25551 }, { "epoch": 2.0143476547102876, "grad_norm": 0.6493135094642639, "learning_rate": 0.0005, "loss": 1.451093316078186, "step": 25552 }, { "epoch": 2.014426487977927, "grad_norm": 0.46310955286026, "learning_rate": 0.0005, "loss": 1.316598892211914, "step": 25553 }, { "epoch": 2.0145053212455655, "grad_norm": 0.611181914806366, "learning_rate": 0.0005, "loss": 1.5139672756195068, "step": 25554 }, { "epoch": 2.0145841545132046, "grad_norm": 0.4939940571784973, "learning_rate": 0.0005, "loss": 1.2087750434875488, "step": 25555 }, { "epoch": 2.0146629877808433, "grad_norm": 0.5994412302970886, "learning_rate": 0.0005, "loss": 1.3317381143569946, "step": 25556 }, { "epoch": 2.0147418210484824, "grad_norm": 0.4550897479057312, "learning_rate": 0.0005, "loss": 1.2929024696350098, "step": 25557 }, { "epoch": 2.0148206543161216, "grad_norm": 0.46430420875549316, "learning_rate": 0.0005, "loss": 1.4430456161499023, "step": 25558 }, { "epoch": 2.0148994875837603, "grad_norm": 0.577686607837677, "learning_rate": 0.0005, "loss": 1.294283390045166, "step": 25559 }, { "epoch": 2.0149783208513994, "grad_norm": 0.4939334988594055, "learning_rate": 0.0005, "loss": 1.279783010482788, "step": 25560 }, { "epoch": 2.015057154119038, "grad_norm": 0.5491834878921509, "learning_rate": 0.0005, "loss": 1.7228769063949585, "step": 25561 }, { "epoch": 2.015135987386677, "grad_norm": 0.4102042317390442, "learning_rate": 0.0005, "loss": 1.2451207637786865, "step": 25562 }, { "epoch": 2.0152148206543163, "grad_norm": 0.465961217880249, "learning_rate": 0.0005, "loss": 1.332501769065857, "step": 25563 }, { "epoch": 2.015293653921955, "grad_norm": 0.6038301587104797, "learning_rate": 0.0005, "loss": 1.281380534172058, "step": 25564 }, { "epoch": 2.015372487189594, "grad_norm": 0.4285869002342224, "learning_rate": 0.0005, "loss": 1.189847707748413, "step": 25565 }, { "epoch": 2.015451320457233, "grad_norm": 0.4876983165740967, "learning_rate": 0.0005, "loss": 1.4119579792022705, "step": 25566 }, { "epoch": 2.015530153724872, "grad_norm": 0.8954164981842041, "learning_rate": 0.0005, "loss": 1.4316940307617188, "step": 25567 }, { "epoch": 2.0156089869925107, "grad_norm": 0.5896000862121582, "learning_rate": 0.0005, "loss": 1.2610456943511963, "step": 25568 }, { "epoch": 2.01568782026015, "grad_norm": 0.51893550157547, "learning_rate": 0.0005, "loss": 1.4062871932983398, "step": 25569 }, { "epoch": 2.015766653527789, "grad_norm": 0.7308017015457153, "learning_rate": 0.0005, "loss": 1.426328182220459, "step": 25570 }, { "epoch": 2.0158454867954276, "grad_norm": 0.5125136375427246, "learning_rate": 0.0005, "loss": 1.5154809951782227, "step": 25571 }, { "epoch": 2.0159243200630668, "grad_norm": 0.4530503451824188, "learning_rate": 0.0005, "loss": 1.4017348289489746, "step": 25572 }, { "epoch": 2.0160031533307055, "grad_norm": 0.45690807700157166, "learning_rate": 0.0005, "loss": 1.219698429107666, "step": 25573 }, { "epoch": 2.0160819865983446, "grad_norm": 0.42692235112190247, "learning_rate": 0.0005, "loss": 1.155718207359314, "step": 25574 }, { "epoch": 2.0161608198659833, "grad_norm": 0.6755460500717163, "learning_rate": 0.0005, "loss": 1.3610608577728271, "step": 25575 }, { "epoch": 2.0162396531336224, "grad_norm": 0.4764673113822937, "learning_rate": 0.0005, "loss": 1.2698440551757812, "step": 25576 }, { "epoch": 2.0163184864012615, "grad_norm": 0.5668029189109802, "learning_rate": 0.0005, "loss": 1.2269713878631592, "step": 25577 }, { "epoch": 2.0163973196689002, "grad_norm": 0.524672269821167, "learning_rate": 0.0005, "loss": 1.3978612422943115, "step": 25578 }, { "epoch": 2.0164761529365394, "grad_norm": 0.4619525671005249, "learning_rate": 0.0005, "loss": 1.247739315032959, "step": 25579 }, { "epoch": 2.016554986204178, "grad_norm": 0.5262101292610168, "learning_rate": 0.0005, "loss": 1.3142049312591553, "step": 25580 }, { "epoch": 2.016633819471817, "grad_norm": 0.5495548248291016, "learning_rate": 0.0005, "loss": 1.2145054340362549, "step": 25581 }, { "epoch": 2.016712652739456, "grad_norm": 0.4217558801174164, "learning_rate": 0.0005, "loss": 1.2604713439941406, "step": 25582 }, { "epoch": 2.016791486007095, "grad_norm": 0.45171695947647095, "learning_rate": 0.0005, "loss": 1.259453535079956, "step": 25583 }, { "epoch": 2.016870319274734, "grad_norm": 0.46279749274253845, "learning_rate": 0.0005, "loss": 1.3973150253295898, "step": 25584 }, { "epoch": 2.016949152542373, "grad_norm": 0.5491155385971069, "learning_rate": 0.0005, "loss": 1.296222448348999, "step": 25585 }, { "epoch": 2.017027985810012, "grad_norm": 0.5594919323921204, "learning_rate": 0.0005, "loss": 1.4464515447616577, "step": 25586 }, { "epoch": 2.0171068190776507, "grad_norm": 0.5663610100746155, "learning_rate": 0.0005, "loss": 1.6049904823303223, "step": 25587 }, { "epoch": 2.01718565234529, "grad_norm": 0.46146130561828613, "learning_rate": 0.0005, "loss": 1.2852506637573242, "step": 25588 }, { "epoch": 2.0172644856129285, "grad_norm": 0.4207161068916321, "learning_rate": 0.0005, "loss": 1.3154457807540894, "step": 25589 }, { "epoch": 2.0173433188805676, "grad_norm": 0.5112995505332947, "learning_rate": 0.0005, "loss": 1.397399663925171, "step": 25590 }, { "epoch": 2.0174221521482067, "grad_norm": 0.48055678606033325, "learning_rate": 0.0005, "loss": 1.4665127992630005, "step": 25591 }, { "epoch": 2.0175009854158454, "grad_norm": 0.4791247844696045, "learning_rate": 0.0005, "loss": 1.226323127746582, "step": 25592 }, { "epoch": 2.0175798186834846, "grad_norm": 0.4440983831882477, "learning_rate": 0.0005, "loss": 1.3374321460723877, "step": 25593 }, { "epoch": 2.0176586519511233, "grad_norm": 0.6187724471092224, "learning_rate": 0.0005, "loss": 1.3741586208343506, "step": 25594 }, { "epoch": 2.0177374852187624, "grad_norm": 0.4963103234767914, "learning_rate": 0.0005, "loss": 1.4424912929534912, "step": 25595 }, { "epoch": 2.017816318486401, "grad_norm": 0.5345882773399353, "learning_rate": 0.0005, "loss": 1.445630431175232, "step": 25596 }, { "epoch": 2.01789515175404, "grad_norm": 0.5125211477279663, "learning_rate": 0.0005, "loss": 1.2899720668792725, "step": 25597 }, { "epoch": 2.0179739850216794, "grad_norm": 0.5254883766174316, "learning_rate": 0.0005, "loss": 1.320017695426941, "step": 25598 }, { "epoch": 2.018052818289318, "grad_norm": 0.4672008752822876, "learning_rate": 0.0005, "loss": 1.1920275688171387, "step": 25599 }, { "epoch": 2.018131651556957, "grad_norm": 0.5063022375106812, "learning_rate": 0.0005, "loss": 1.366944670677185, "step": 25600 }, { "epoch": 2.018210484824596, "grad_norm": 0.470180869102478, "learning_rate": 0.0005, "loss": 1.4580755233764648, "step": 25601 }, { "epoch": 2.018289318092235, "grad_norm": 0.626242995262146, "learning_rate": 0.0005, "loss": 1.3533697128295898, "step": 25602 }, { "epoch": 2.0183681513598737, "grad_norm": 0.6353836059570312, "learning_rate": 0.0005, "loss": 1.6795861721038818, "step": 25603 }, { "epoch": 2.018446984627513, "grad_norm": 0.510725736618042, "learning_rate": 0.0005, "loss": 1.3024466037750244, "step": 25604 }, { "epoch": 2.018525817895152, "grad_norm": 0.4798612594604492, "learning_rate": 0.0005, "loss": 1.4561436176300049, "step": 25605 }, { "epoch": 2.0186046511627906, "grad_norm": 0.42005518078804016, "learning_rate": 0.0005, "loss": 1.3677698373794556, "step": 25606 }, { "epoch": 2.0186834844304298, "grad_norm": 0.5678736567497253, "learning_rate": 0.0005, "loss": 1.4166840314865112, "step": 25607 }, { "epoch": 2.0187623176980685, "grad_norm": 0.47210627794265747, "learning_rate": 0.0005, "loss": 1.3398840427398682, "step": 25608 }, { "epoch": 2.0188411509657076, "grad_norm": 0.532051146030426, "learning_rate": 0.0005, "loss": 1.5298136472702026, "step": 25609 }, { "epoch": 2.0189199842333463, "grad_norm": 0.4007030427455902, "learning_rate": 0.0005, "loss": 1.1968169212341309, "step": 25610 }, { "epoch": 2.0189988175009854, "grad_norm": 0.5195993781089783, "learning_rate": 0.0005, "loss": 1.5737147331237793, "step": 25611 }, { "epoch": 2.0190776507686246, "grad_norm": 0.5189354419708252, "learning_rate": 0.0005, "loss": 1.4687012434005737, "step": 25612 }, { "epoch": 2.0191564840362632, "grad_norm": 0.5460285544395447, "learning_rate": 0.0005, "loss": 1.296036720275879, "step": 25613 }, { "epoch": 2.0192353173039024, "grad_norm": 0.5327387452125549, "learning_rate": 0.0005, "loss": 1.666279911994934, "step": 25614 }, { "epoch": 2.019314150571541, "grad_norm": 0.4538417160511017, "learning_rate": 0.0005, "loss": 1.305694818496704, "step": 25615 }, { "epoch": 2.01939298383918, "grad_norm": 0.4545949697494507, "learning_rate": 0.0005, "loss": 1.3869445323944092, "step": 25616 }, { "epoch": 2.019471817106819, "grad_norm": 0.41746553778648376, "learning_rate": 0.0005, "loss": 1.4393101930618286, "step": 25617 }, { "epoch": 2.019550650374458, "grad_norm": 0.5339491367340088, "learning_rate": 0.0005, "loss": 1.51729416847229, "step": 25618 }, { "epoch": 2.019629483642097, "grad_norm": 0.5175211429595947, "learning_rate": 0.0005, "loss": 1.551635980606079, "step": 25619 }, { "epoch": 2.019708316909736, "grad_norm": 0.46469539403915405, "learning_rate": 0.0005, "loss": 1.2329676151275635, "step": 25620 }, { "epoch": 2.019787150177375, "grad_norm": 0.6120604872703552, "learning_rate": 0.0005, "loss": 1.2820438146591187, "step": 25621 }, { "epoch": 2.0198659834450137, "grad_norm": 0.39691880345344543, "learning_rate": 0.0005, "loss": 1.170812964439392, "step": 25622 }, { "epoch": 2.019944816712653, "grad_norm": 0.5146130323410034, "learning_rate": 0.0005, "loss": 1.2871702909469604, "step": 25623 }, { "epoch": 2.0200236499802915, "grad_norm": 0.4637707769870758, "learning_rate": 0.0005, "loss": 1.22311270236969, "step": 25624 }, { "epoch": 2.0201024832479306, "grad_norm": 0.5375871062278748, "learning_rate": 0.0005, "loss": 1.401620864868164, "step": 25625 }, { "epoch": 2.0201813165155698, "grad_norm": 0.5170924663543701, "learning_rate": 0.0005, "loss": 1.1355550289154053, "step": 25626 }, { "epoch": 2.0202601497832084, "grad_norm": 0.4678173363208771, "learning_rate": 0.0005, "loss": 1.302869439125061, "step": 25627 }, { "epoch": 2.0203389830508476, "grad_norm": 0.5576592683792114, "learning_rate": 0.0005, "loss": 1.3226888179779053, "step": 25628 }, { "epoch": 2.0204178163184863, "grad_norm": 0.5434167385101318, "learning_rate": 0.0005, "loss": 1.2765016555786133, "step": 25629 }, { "epoch": 2.0204966495861254, "grad_norm": 0.6575555205345154, "learning_rate": 0.0005, "loss": 1.389719009399414, "step": 25630 }, { "epoch": 2.020575482853764, "grad_norm": 0.434038907289505, "learning_rate": 0.0005, "loss": 1.0881798267364502, "step": 25631 }, { "epoch": 2.0206543161214032, "grad_norm": 0.752082109451294, "learning_rate": 0.0005, "loss": 1.4327144622802734, "step": 25632 }, { "epoch": 2.0207331493890424, "grad_norm": 0.49704593420028687, "learning_rate": 0.0005, "loss": 1.462121844291687, "step": 25633 }, { "epoch": 2.020811982656681, "grad_norm": 0.38998499512672424, "learning_rate": 0.0005, "loss": 1.1762539148330688, "step": 25634 }, { "epoch": 2.02089081592432, "grad_norm": 0.5000308752059937, "learning_rate": 0.0005, "loss": 1.3331154584884644, "step": 25635 }, { "epoch": 2.020969649191959, "grad_norm": 0.4479173719882965, "learning_rate": 0.0005, "loss": 1.4101307392120361, "step": 25636 }, { "epoch": 2.021048482459598, "grad_norm": 0.5704500675201416, "learning_rate": 0.0005, "loss": 1.226992130279541, "step": 25637 }, { "epoch": 2.0211273157272367, "grad_norm": 0.49466294050216675, "learning_rate": 0.0005, "loss": 1.125589370727539, "step": 25638 }, { "epoch": 2.021206148994876, "grad_norm": 0.44770383834838867, "learning_rate": 0.0005, "loss": 1.213244915008545, "step": 25639 }, { "epoch": 2.021284982262515, "grad_norm": 0.6558617949485779, "learning_rate": 0.0005, "loss": 1.0877506732940674, "step": 25640 }, { "epoch": 2.0213638155301537, "grad_norm": 0.5503889918327332, "learning_rate": 0.0005, "loss": 1.5148866176605225, "step": 25641 }, { "epoch": 2.021442648797793, "grad_norm": 0.5070462226867676, "learning_rate": 0.0005, "loss": 1.3547353744506836, "step": 25642 }, { "epoch": 2.0215214820654315, "grad_norm": 0.5468460917472839, "learning_rate": 0.0005, "loss": 1.3388861417770386, "step": 25643 }, { "epoch": 2.0216003153330706, "grad_norm": 0.5460335612297058, "learning_rate": 0.0005, "loss": 1.2217408418655396, "step": 25644 }, { "epoch": 2.0216791486007093, "grad_norm": 0.5352476239204407, "learning_rate": 0.0005, "loss": 1.5861951112747192, "step": 25645 }, { "epoch": 2.0217579818683484, "grad_norm": 0.4856778681278229, "learning_rate": 0.0005, "loss": 1.3027355670928955, "step": 25646 }, { "epoch": 2.0218368151359876, "grad_norm": 0.49385178089141846, "learning_rate": 0.0005, "loss": 1.451851487159729, "step": 25647 }, { "epoch": 2.0219156484036263, "grad_norm": 0.4716639816761017, "learning_rate": 0.0005, "loss": 1.3809645175933838, "step": 25648 }, { "epoch": 2.0219944816712654, "grad_norm": 0.4594634175300598, "learning_rate": 0.0005, "loss": 1.2347493171691895, "step": 25649 }, { "epoch": 2.022073314938904, "grad_norm": 0.4003063440322876, "learning_rate": 0.0005, "loss": 1.0836527347564697, "step": 25650 }, { "epoch": 2.022152148206543, "grad_norm": 0.5320907831192017, "learning_rate": 0.0005, "loss": 1.3814866542816162, "step": 25651 }, { "epoch": 2.022230981474182, "grad_norm": 0.5291727185249329, "learning_rate": 0.0005, "loss": 1.25081467628479, "step": 25652 }, { "epoch": 2.022309814741821, "grad_norm": 0.6115028858184814, "learning_rate": 0.0005, "loss": 1.2298212051391602, "step": 25653 }, { "epoch": 2.02238864800946, "grad_norm": 0.5307687520980835, "learning_rate": 0.0005, "loss": 1.1952745914459229, "step": 25654 }, { "epoch": 2.022467481277099, "grad_norm": 0.5668959021568298, "learning_rate": 0.0005, "loss": 1.4844201803207397, "step": 25655 }, { "epoch": 2.022546314544738, "grad_norm": 0.5277495384216309, "learning_rate": 0.0005, "loss": 1.5620145797729492, "step": 25656 }, { "epoch": 2.0226251478123767, "grad_norm": 0.5313717126846313, "learning_rate": 0.0005, "loss": 1.4169845581054688, "step": 25657 }, { "epoch": 2.022703981080016, "grad_norm": 0.4844748079776764, "learning_rate": 0.0005, "loss": 1.423574447631836, "step": 25658 }, { "epoch": 2.0227828143476545, "grad_norm": 0.4355904459953308, "learning_rate": 0.0005, "loss": 1.3937585353851318, "step": 25659 }, { "epoch": 2.0228616476152936, "grad_norm": 0.5007655620574951, "learning_rate": 0.0005, "loss": 1.3405615091323853, "step": 25660 }, { "epoch": 2.0229404808829328, "grad_norm": 0.5041772127151489, "learning_rate": 0.0005, "loss": 1.1815471649169922, "step": 25661 }, { "epoch": 2.0230193141505715, "grad_norm": 0.5030718445777893, "learning_rate": 0.0005, "loss": 1.1576286554336548, "step": 25662 }, { "epoch": 2.0230981474182106, "grad_norm": 0.5336852669715881, "learning_rate": 0.0005, "loss": 1.301905632019043, "step": 25663 }, { "epoch": 2.0231769806858493, "grad_norm": 0.5363289713859558, "learning_rate": 0.0005, "loss": 1.2351667881011963, "step": 25664 }, { "epoch": 2.0232558139534884, "grad_norm": 0.5071254372596741, "learning_rate": 0.0005, "loss": 1.4801979064941406, "step": 25665 }, { "epoch": 2.023334647221127, "grad_norm": 0.5683687329292297, "learning_rate": 0.0005, "loss": 1.2938251495361328, "step": 25666 }, { "epoch": 2.0234134804887662, "grad_norm": 0.5313791632652283, "learning_rate": 0.0005, "loss": 1.1970162391662598, "step": 25667 }, { "epoch": 2.0234923137564054, "grad_norm": 0.45279741287231445, "learning_rate": 0.0005, "loss": 1.322184681892395, "step": 25668 }, { "epoch": 2.023571147024044, "grad_norm": 0.48947930335998535, "learning_rate": 0.0005, "loss": 1.2407363653182983, "step": 25669 }, { "epoch": 2.023649980291683, "grad_norm": 0.4148239493370056, "learning_rate": 0.0005, "loss": 1.18296217918396, "step": 25670 }, { "epoch": 2.023728813559322, "grad_norm": 0.4976901710033417, "learning_rate": 0.0005, "loss": 1.5103859901428223, "step": 25671 }, { "epoch": 2.023807646826961, "grad_norm": 0.6374571323394775, "learning_rate": 0.0005, "loss": 1.4334046840667725, "step": 25672 }, { "epoch": 2.0238864800945997, "grad_norm": 0.4715052545070648, "learning_rate": 0.0005, "loss": 1.1592499017715454, "step": 25673 }, { "epoch": 2.023965313362239, "grad_norm": 0.4894990921020508, "learning_rate": 0.0005, "loss": 1.2179803848266602, "step": 25674 }, { "epoch": 2.024044146629878, "grad_norm": 0.5445259809494019, "learning_rate": 0.0005, "loss": 1.3263260126113892, "step": 25675 }, { "epoch": 2.0241229798975167, "grad_norm": 0.5195654630661011, "learning_rate": 0.0005, "loss": 1.5097990036010742, "step": 25676 }, { "epoch": 2.024201813165156, "grad_norm": 0.552475094795227, "learning_rate": 0.0005, "loss": 1.6754788160324097, "step": 25677 }, { "epoch": 2.0242806464327945, "grad_norm": 0.4638843834400177, "learning_rate": 0.0005, "loss": 1.413649082183838, "step": 25678 }, { "epoch": 2.0243594797004336, "grad_norm": 0.4922431707382202, "learning_rate": 0.0005, "loss": 1.3691314458847046, "step": 25679 }, { "epoch": 2.0244383129680723, "grad_norm": 0.45916616916656494, "learning_rate": 0.0005, "loss": 1.2388973236083984, "step": 25680 }, { "epoch": 2.0245171462357114, "grad_norm": 0.5941755771636963, "learning_rate": 0.0005, "loss": 1.5923686027526855, "step": 25681 }, { "epoch": 2.0245959795033506, "grad_norm": 0.4335319995880127, "learning_rate": 0.0005, "loss": 1.316328525543213, "step": 25682 }, { "epoch": 2.0246748127709893, "grad_norm": 0.4962831437587738, "learning_rate": 0.0005, "loss": 1.5536723136901855, "step": 25683 }, { "epoch": 2.0247536460386284, "grad_norm": 0.49245116114616394, "learning_rate": 0.0005, "loss": 1.1694793701171875, "step": 25684 }, { "epoch": 2.024832479306267, "grad_norm": 0.5904091596603394, "learning_rate": 0.0005, "loss": 1.5702338218688965, "step": 25685 }, { "epoch": 2.024911312573906, "grad_norm": 0.372908353805542, "learning_rate": 0.0005, "loss": 1.1163501739501953, "step": 25686 }, { "epoch": 2.024990145841545, "grad_norm": 0.43279561400413513, "learning_rate": 0.0005, "loss": 1.4279696941375732, "step": 25687 }, { "epoch": 2.025068979109184, "grad_norm": 0.4137355089187622, "learning_rate": 0.0005, "loss": 1.2137333154678345, "step": 25688 }, { "epoch": 2.025147812376823, "grad_norm": 0.5287660360336304, "learning_rate": 0.0005, "loss": 1.4082159996032715, "step": 25689 }, { "epoch": 2.025226645644462, "grad_norm": 0.5693302750587463, "learning_rate": 0.0005, "loss": 1.6705431938171387, "step": 25690 }, { "epoch": 2.025305478912101, "grad_norm": 0.5268979668617249, "learning_rate": 0.0005, "loss": 1.255332350730896, "step": 25691 }, { "epoch": 2.0253843121797397, "grad_norm": 0.46048903465270996, "learning_rate": 0.0005, "loss": 1.4130690097808838, "step": 25692 }, { "epoch": 2.025463145447379, "grad_norm": 0.44037824869155884, "learning_rate": 0.0005, "loss": 1.4045979976654053, "step": 25693 }, { "epoch": 2.025541978715018, "grad_norm": 0.5067592263221741, "learning_rate": 0.0005, "loss": 1.3018128871917725, "step": 25694 }, { "epoch": 2.0256208119826566, "grad_norm": 0.44473132491111755, "learning_rate": 0.0005, "loss": 1.3704155683517456, "step": 25695 }, { "epoch": 2.0256996452502958, "grad_norm": 0.4651006758213043, "learning_rate": 0.0005, "loss": 1.1164679527282715, "step": 25696 }, { "epoch": 2.0257784785179345, "grad_norm": 0.5034859776496887, "learning_rate": 0.0005, "loss": 1.157757043838501, "step": 25697 }, { "epoch": 2.0258573117855736, "grad_norm": 0.4588681757450104, "learning_rate": 0.0005, "loss": 1.4450081586837769, "step": 25698 }, { "epoch": 2.0259361450532123, "grad_norm": 0.43745967745780945, "learning_rate": 0.0005, "loss": 1.3127527236938477, "step": 25699 }, { "epoch": 2.0260149783208514, "grad_norm": 0.4796521067619324, "learning_rate": 0.0005, "loss": 1.233642816543579, "step": 25700 }, { "epoch": 2.0260938115884906, "grad_norm": 0.5058556199073792, "learning_rate": 0.0005, "loss": 1.2484047412872314, "step": 25701 }, { "epoch": 2.0261726448561292, "grad_norm": 0.4717019200325012, "learning_rate": 0.0005, "loss": 1.2272855043411255, "step": 25702 }, { "epoch": 2.0262514781237684, "grad_norm": 0.5540373921394348, "learning_rate": 0.0005, "loss": 1.3991062641143799, "step": 25703 }, { "epoch": 2.026330311391407, "grad_norm": 0.4802062511444092, "learning_rate": 0.0005, "loss": 1.3443408012390137, "step": 25704 }, { "epoch": 2.026409144659046, "grad_norm": 0.4381880760192871, "learning_rate": 0.0005, "loss": 1.4050757884979248, "step": 25705 }, { "epoch": 2.026487977926685, "grad_norm": 0.6591938734054565, "learning_rate": 0.0005, "loss": 1.4550182819366455, "step": 25706 }, { "epoch": 2.026566811194324, "grad_norm": 0.4314993917942047, "learning_rate": 0.0005, "loss": 1.1708579063415527, "step": 25707 }, { "epoch": 2.026645644461963, "grad_norm": 0.4185716509819031, "learning_rate": 0.0005, "loss": 1.2868006229400635, "step": 25708 }, { "epoch": 2.026724477729602, "grad_norm": 0.4128323495388031, "learning_rate": 0.0005, "loss": 1.1732637882232666, "step": 25709 }, { "epoch": 2.026803310997241, "grad_norm": 0.4040917754173279, "learning_rate": 0.0005, "loss": 1.2474727630615234, "step": 25710 }, { "epoch": 2.0268821442648797, "grad_norm": 0.40388721227645874, "learning_rate": 0.0005, "loss": 1.3669596910476685, "step": 25711 }, { "epoch": 2.026960977532519, "grad_norm": 0.4726903736591339, "learning_rate": 0.0005, "loss": 1.236743450164795, "step": 25712 }, { "epoch": 2.0270398108001575, "grad_norm": 0.4519961178302765, "learning_rate": 0.0005, "loss": 1.3006999492645264, "step": 25713 }, { "epoch": 2.0271186440677966, "grad_norm": 0.6448515057563782, "learning_rate": 0.0005, "loss": 1.2498739957809448, "step": 25714 }, { "epoch": 2.0271974773354358, "grad_norm": 0.501335084438324, "learning_rate": 0.0005, "loss": 1.3653111457824707, "step": 25715 }, { "epoch": 2.0272763106030744, "grad_norm": 0.4965323507785797, "learning_rate": 0.0005, "loss": 1.3359698057174683, "step": 25716 }, { "epoch": 2.0273551438707136, "grad_norm": 0.4707380533218384, "learning_rate": 0.0005, "loss": 1.3243303298950195, "step": 25717 }, { "epoch": 2.0274339771383523, "grad_norm": 0.5571938157081604, "learning_rate": 0.0005, "loss": 1.358659029006958, "step": 25718 }, { "epoch": 2.0275128104059914, "grad_norm": 0.44774311780929565, "learning_rate": 0.0005, "loss": 1.443221092224121, "step": 25719 }, { "epoch": 2.02759164367363, "grad_norm": 0.5301305651664734, "learning_rate": 0.0005, "loss": 1.3955228328704834, "step": 25720 }, { "epoch": 2.0276704769412692, "grad_norm": 0.44859591126441956, "learning_rate": 0.0005, "loss": 1.3259655237197876, "step": 25721 }, { "epoch": 2.0277493102089084, "grad_norm": 0.44205814599990845, "learning_rate": 0.0005, "loss": 1.2943345308303833, "step": 25722 }, { "epoch": 2.027828143476547, "grad_norm": 0.42209336161613464, "learning_rate": 0.0005, "loss": 1.4307175874710083, "step": 25723 }, { "epoch": 2.027906976744186, "grad_norm": 0.4996713101863861, "learning_rate": 0.0005, "loss": 1.5194969177246094, "step": 25724 }, { "epoch": 2.027985810011825, "grad_norm": 0.5187596678733826, "learning_rate": 0.0005, "loss": 1.4339983463287354, "step": 25725 }, { "epoch": 2.028064643279464, "grad_norm": 0.4646996855735779, "learning_rate": 0.0005, "loss": 1.3864479064941406, "step": 25726 }, { "epoch": 2.0281434765471027, "grad_norm": 0.44976067543029785, "learning_rate": 0.0005, "loss": 1.378490924835205, "step": 25727 }, { "epoch": 2.028222309814742, "grad_norm": 0.48659569025039673, "learning_rate": 0.0005, "loss": 1.2997726202011108, "step": 25728 }, { "epoch": 2.028301143082381, "grad_norm": 0.5010913014411926, "learning_rate": 0.0005, "loss": 1.359352469444275, "step": 25729 }, { "epoch": 2.0283799763500197, "grad_norm": 0.4750998914241791, "learning_rate": 0.0005, "loss": 1.4062905311584473, "step": 25730 }, { "epoch": 2.028458809617659, "grad_norm": 0.4259273409843445, "learning_rate": 0.0005, "loss": 1.2781989574432373, "step": 25731 }, { "epoch": 2.0285376428852975, "grad_norm": 0.45818522572517395, "learning_rate": 0.0005, "loss": 1.243394374847412, "step": 25732 }, { "epoch": 2.0286164761529366, "grad_norm": 0.43337878584861755, "learning_rate": 0.0005, "loss": 1.2112269401550293, "step": 25733 }, { "epoch": 2.0286953094205753, "grad_norm": 0.6298087239265442, "learning_rate": 0.0005, "loss": 1.4283345937728882, "step": 25734 }, { "epoch": 2.0287741426882144, "grad_norm": 0.41173845529556274, "learning_rate": 0.0005, "loss": 1.228991150856018, "step": 25735 }, { "epoch": 2.0288529759558536, "grad_norm": 0.4482218325138092, "learning_rate": 0.0005, "loss": 1.171658992767334, "step": 25736 }, { "epoch": 2.0289318092234923, "grad_norm": 0.6933199167251587, "learning_rate": 0.0005, "loss": 1.5098628997802734, "step": 25737 }, { "epoch": 2.0290106424911314, "grad_norm": 0.3500322997570038, "learning_rate": 0.0005, "loss": 1.2245064973831177, "step": 25738 }, { "epoch": 2.02908947575877, "grad_norm": 0.7457129955291748, "learning_rate": 0.0005, "loss": 1.4452089071273804, "step": 25739 }, { "epoch": 2.029168309026409, "grad_norm": 0.46156051754951477, "learning_rate": 0.0005, "loss": 1.3123167753219604, "step": 25740 }, { "epoch": 2.029247142294048, "grad_norm": 0.4177224934101105, "learning_rate": 0.0005, "loss": 1.2603219747543335, "step": 25741 }, { "epoch": 2.029325975561687, "grad_norm": 0.5173746347427368, "learning_rate": 0.0005, "loss": 1.2413201332092285, "step": 25742 }, { "epoch": 2.029404808829326, "grad_norm": 0.44859203696250916, "learning_rate": 0.0005, "loss": 1.2200621366500854, "step": 25743 }, { "epoch": 2.029483642096965, "grad_norm": 0.5234670639038086, "learning_rate": 0.0005, "loss": 1.136059284210205, "step": 25744 }, { "epoch": 2.029562475364604, "grad_norm": 0.522351086139679, "learning_rate": 0.0005, "loss": 1.3978837728500366, "step": 25745 }, { "epoch": 2.0296413086322427, "grad_norm": 0.5045843124389648, "learning_rate": 0.0005, "loss": 1.2644338607788086, "step": 25746 }, { "epoch": 2.029720141899882, "grad_norm": 0.4909164309501648, "learning_rate": 0.0005, "loss": 1.3613084554672241, "step": 25747 }, { "epoch": 2.0297989751675205, "grad_norm": 0.6382652521133423, "learning_rate": 0.0005, "loss": 1.3851174116134644, "step": 25748 }, { "epoch": 2.0298778084351596, "grad_norm": 0.4846304655075073, "learning_rate": 0.0005, "loss": 1.3998652696609497, "step": 25749 }, { "epoch": 2.0299566417027988, "grad_norm": 0.4556191563606262, "learning_rate": 0.0005, "loss": 1.117775321006775, "step": 25750 }, { "epoch": 2.0300354749704375, "grad_norm": 0.44516506791114807, "learning_rate": 0.0005, "loss": 1.1880656480789185, "step": 25751 }, { "epoch": 2.0301143082380766, "grad_norm": 0.5330653786659241, "learning_rate": 0.0005, "loss": 1.4656684398651123, "step": 25752 }, { "epoch": 2.0301931415057153, "grad_norm": 0.4866721034049988, "learning_rate": 0.0005, "loss": 1.489072322845459, "step": 25753 }, { "epoch": 2.0302719747733544, "grad_norm": 0.47885453701019287, "learning_rate": 0.0005, "loss": 1.3168158531188965, "step": 25754 }, { "epoch": 2.030350808040993, "grad_norm": 0.5387957692146301, "learning_rate": 0.0005, "loss": 1.3557195663452148, "step": 25755 }, { "epoch": 2.0304296413086322, "grad_norm": 0.6189814805984497, "learning_rate": 0.0005, "loss": 1.4242172241210938, "step": 25756 }, { "epoch": 2.0305084745762714, "grad_norm": 0.4574461579322815, "learning_rate": 0.0005, "loss": 1.3173742294311523, "step": 25757 }, { "epoch": 2.03058730784391, "grad_norm": 0.5728470683097839, "learning_rate": 0.0005, "loss": 1.43380868434906, "step": 25758 }, { "epoch": 2.030666141111549, "grad_norm": 0.40895262360572815, "learning_rate": 0.0005, "loss": 1.2080732583999634, "step": 25759 }, { "epoch": 2.030744974379188, "grad_norm": 0.4607085585594177, "learning_rate": 0.0005, "loss": 1.2729074954986572, "step": 25760 }, { "epoch": 2.030823807646827, "grad_norm": 0.48186302185058594, "learning_rate": 0.0005, "loss": 1.316016435623169, "step": 25761 }, { "epoch": 2.0309026409144657, "grad_norm": 0.4377124607563019, "learning_rate": 0.0005, "loss": 1.3673818111419678, "step": 25762 }, { "epoch": 2.030981474182105, "grad_norm": 0.4452407956123352, "learning_rate": 0.0005, "loss": 1.3563082218170166, "step": 25763 }, { "epoch": 2.031060307449744, "grad_norm": 0.44913265109062195, "learning_rate": 0.0005, "loss": 1.3371615409851074, "step": 25764 }, { "epoch": 2.0311391407173827, "grad_norm": 0.5445312261581421, "learning_rate": 0.0005, "loss": 1.3187370300292969, "step": 25765 }, { "epoch": 2.031217973985022, "grad_norm": 0.547603964805603, "learning_rate": 0.0005, "loss": 1.4294122457504272, "step": 25766 }, { "epoch": 2.0312968072526605, "grad_norm": 0.4199143946170807, "learning_rate": 0.0005, "loss": 1.2664754390716553, "step": 25767 }, { "epoch": 2.0313756405202996, "grad_norm": 0.4714868664741516, "learning_rate": 0.0005, "loss": 1.3142404556274414, "step": 25768 }, { "epoch": 2.0314544737879383, "grad_norm": 0.371897429227829, "learning_rate": 0.0005, "loss": 1.2385504245758057, "step": 25769 }, { "epoch": 2.0315333070555774, "grad_norm": 0.4005303680896759, "learning_rate": 0.0005, "loss": 1.2335796356201172, "step": 25770 }, { "epoch": 2.0316121403232166, "grad_norm": 0.5157444477081299, "learning_rate": 0.0005, "loss": 1.402143955230713, "step": 25771 }, { "epoch": 2.0316909735908553, "grad_norm": 0.5277809500694275, "learning_rate": 0.0005, "loss": 1.383528232574463, "step": 25772 }, { "epoch": 2.0317698068584944, "grad_norm": 0.4860006272792816, "learning_rate": 0.0005, "loss": 1.2398061752319336, "step": 25773 }, { "epoch": 2.031848640126133, "grad_norm": 0.44446563720703125, "learning_rate": 0.0005, "loss": 1.2585158348083496, "step": 25774 }, { "epoch": 2.031927473393772, "grad_norm": 0.46703022718429565, "learning_rate": 0.0005, "loss": 1.5294075012207031, "step": 25775 }, { "epoch": 2.032006306661411, "grad_norm": 0.47410351037979126, "learning_rate": 0.0005, "loss": 1.185046911239624, "step": 25776 }, { "epoch": 2.03208513992905, "grad_norm": 0.48913636803627014, "learning_rate": 0.0005, "loss": 1.2483527660369873, "step": 25777 }, { "epoch": 2.032163973196689, "grad_norm": 0.3812761902809143, "learning_rate": 0.0005, "loss": 1.277618169784546, "step": 25778 }, { "epoch": 2.032242806464328, "grad_norm": 0.5424320101737976, "learning_rate": 0.0005, "loss": 1.515049934387207, "step": 25779 }, { "epoch": 2.032321639731967, "grad_norm": 0.5232023596763611, "learning_rate": 0.0005, "loss": 1.3035829067230225, "step": 25780 }, { "epoch": 2.0324004729996057, "grad_norm": 0.4820781648159027, "learning_rate": 0.0005, "loss": 1.4409990310668945, "step": 25781 }, { "epoch": 2.032479306267245, "grad_norm": 0.4591449499130249, "learning_rate": 0.0005, "loss": 1.1904480457305908, "step": 25782 }, { "epoch": 2.0325581395348835, "grad_norm": 0.5437905192375183, "learning_rate": 0.0005, "loss": 1.198387861251831, "step": 25783 }, { "epoch": 2.0326369728025226, "grad_norm": 0.43797457218170166, "learning_rate": 0.0005, "loss": 1.417259693145752, "step": 25784 }, { "epoch": 2.032715806070162, "grad_norm": 0.6194391846656799, "learning_rate": 0.0005, "loss": 1.3536567687988281, "step": 25785 }, { "epoch": 2.0327946393378005, "grad_norm": 0.50995934009552, "learning_rate": 0.0005, "loss": 1.3914320468902588, "step": 25786 }, { "epoch": 2.0328734726054396, "grad_norm": 0.4191380441188812, "learning_rate": 0.0005, "loss": 1.2326734066009521, "step": 25787 }, { "epoch": 2.0329523058730783, "grad_norm": 0.4220889210700989, "learning_rate": 0.0005, "loss": 1.2651698589324951, "step": 25788 }, { "epoch": 2.0330311391407174, "grad_norm": 0.5217933058738708, "learning_rate": 0.0005, "loss": 1.2854087352752686, "step": 25789 }, { "epoch": 2.033109972408356, "grad_norm": 0.44369685649871826, "learning_rate": 0.0005, "loss": 1.3735677003860474, "step": 25790 }, { "epoch": 2.0331888056759952, "grad_norm": 0.4848063290119171, "learning_rate": 0.0005, "loss": 1.386103868484497, "step": 25791 }, { "epoch": 2.0332676389436344, "grad_norm": 0.5479732751846313, "learning_rate": 0.0005, "loss": 1.5455467700958252, "step": 25792 }, { "epoch": 2.033346472211273, "grad_norm": 0.7722265124320984, "learning_rate": 0.0005, "loss": 1.4154939651489258, "step": 25793 }, { "epoch": 2.033425305478912, "grad_norm": 0.41794419288635254, "learning_rate": 0.0005, "loss": 1.3217928409576416, "step": 25794 }, { "epoch": 2.033504138746551, "grad_norm": 0.5307248830795288, "learning_rate": 0.0005, "loss": 1.3379113674163818, "step": 25795 }, { "epoch": 2.03358297201419, "grad_norm": 0.4382171928882599, "learning_rate": 0.0005, "loss": 1.224556565284729, "step": 25796 }, { "epoch": 2.0336618052818287, "grad_norm": 0.45520341396331787, "learning_rate": 0.0005, "loss": 1.3021748065948486, "step": 25797 }, { "epoch": 2.033740638549468, "grad_norm": 0.5068667531013489, "learning_rate": 0.0005, "loss": 1.261427879333496, "step": 25798 }, { "epoch": 2.033819471817107, "grad_norm": 0.46275290846824646, "learning_rate": 0.0005, "loss": 1.4761502742767334, "step": 25799 }, { "epoch": 2.0338983050847457, "grad_norm": 0.6249605417251587, "learning_rate": 0.0005, "loss": 1.577368974685669, "step": 25800 }, { "epoch": 2.033977138352385, "grad_norm": 0.45277678966522217, "learning_rate": 0.0005, "loss": 1.2953896522521973, "step": 25801 }, { "epoch": 2.0340559716200235, "grad_norm": 0.446200966835022, "learning_rate": 0.0005, "loss": 1.4074985980987549, "step": 25802 }, { "epoch": 2.0341348048876626, "grad_norm": 0.5586902499198914, "learning_rate": 0.0005, "loss": 1.3762948513031006, "step": 25803 }, { "epoch": 2.0342136381553013, "grad_norm": 0.42211073637008667, "learning_rate": 0.0005, "loss": 1.3383394479751587, "step": 25804 }, { "epoch": 2.0342924714229405, "grad_norm": 0.5201801061630249, "learning_rate": 0.0005, "loss": 1.5352697372436523, "step": 25805 }, { "epoch": 2.0343713046905796, "grad_norm": 0.5538772344589233, "learning_rate": 0.0005, "loss": 1.5714802742004395, "step": 25806 }, { "epoch": 2.0344501379582183, "grad_norm": 0.5257241129875183, "learning_rate": 0.0005, "loss": 1.3541680574417114, "step": 25807 }, { "epoch": 2.0345289712258574, "grad_norm": 0.514863908290863, "learning_rate": 0.0005, "loss": 1.2901626825332642, "step": 25808 }, { "epoch": 2.034607804493496, "grad_norm": 0.572181224822998, "learning_rate": 0.0005, "loss": 1.371313452720642, "step": 25809 }, { "epoch": 2.0346866377611352, "grad_norm": 0.6498968601226807, "learning_rate": 0.0005, "loss": 1.4214646816253662, "step": 25810 }, { "epoch": 2.0347654710287744, "grad_norm": 0.5701812505722046, "learning_rate": 0.0005, "loss": 1.3057849407196045, "step": 25811 }, { "epoch": 2.034844304296413, "grad_norm": 0.538731038570404, "learning_rate": 0.0005, "loss": 1.4356393814086914, "step": 25812 }, { "epoch": 2.034923137564052, "grad_norm": 0.4600314795970917, "learning_rate": 0.0005, "loss": 1.2327338457107544, "step": 25813 }, { "epoch": 2.035001970831691, "grad_norm": 0.5983346104621887, "learning_rate": 0.0005, "loss": 1.3738479614257812, "step": 25814 }, { "epoch": 2.03508080409933, "grad_norm": 0.5731672048568726, "learning_rate": 0.0005, "loss": 1.3992934226989746, "step": 25815 }, { "epoch": 2.0351596373669687, "grad_norm": 0.38320809602737427, "learning_rate": 0.0005, "loss": 1.1328471899032593, "step": 25816 }, { "epoch": 2.035238470634608, "grad_norm": 0.45921891927719116, "learning_rate": 0.0005, "loss": 1.2885465621948242, "step": 25817 }, { "epoch": 2.035317303902247, "grad_norm": 0.7549616694450378, "learning_rate": 0.0005, "loss": 1.4162471294403076, "step": 25818 }, { "epoch": 2.0353961371698857, "grad_norm": 0.4781205952167511, "learning_rate": 0.0005, "loss": 1.148973822593689, "step": 25819 }, { "epoch": 2.035474970437525, "grad_norm": 0.45982906222343445, "learning_rate": 0.0005, "loss": 1.317587971687317, "step": 25820 }, { "epoch": 2.0355538037051635, "grad_norm": 0.42882809042930603, "learning_rate": 0.0005, "loss": 1.2960759401321411, "step": 25821 }, { "epoch": 2.0356326369728026, "grad_norm": 0.3821488618850708, "learning_rate": 0.0005, "loss": 1.0792508125305176, "step": 25822 }, { "epoch": 2.0357114702404413, "grad_norm": 0.4664309322834015, "learning_rate": 0.0005, "loss": 1.4821979999542236, "step": 25823 }, { "epoch": 2.0357903035080804, "grad_norm": 0.44816887378692627, "learning_rate": 0.0005, "loss": 1.3651758432388306, "step": 25824 }, { "epoch": 2.0358691367757196, "grad_norm": 0.5054049491882324, "learning_rate": 0.0005, "loss": 1.3531312942504883, "step": 25825 }, { "epoch": 2.0359479700433583, "grad_norm": 0.5448969602584839, "learning_rate": 0.0005, "loss": 1.562027931213379, "step": 25826 }, { "epoch": 2.0360268033109974, "grad_norm": 0.47697970271110535, "learning_rate": 0.0005, "loss": 1.2627599239349365, "step": 25827 }, { "epoch": 2.036105636578636, "grad_norm": 0.452659547328949, "learning_rate": 0.0005, "loss": 1.3492196798324585, "step": 25828 }, { "epoch": 2.036184469846275, "grad_norm": 0.5497245192527771, "learning_rate": 0.0005, "loss": 1.306304931640625, "step": 25829 }, { "epoch": 2.036263303113914, "grad_norm": 0.4237005114555359, "learning_rate": 0.0005, "loss": 1.3656939268112183, "step": 25830 }, { "epoch": 2.036342136381553, "grad_norm": 0.465305358171463, "learning_rate": 0.0005, "loss": 1.2387149333953857, "step": 25831 }, { "epoch": 2.036420969649192, "grad_norm": 0.5208359360694885, "learning_rate": 0.0005, "loss": 1.462969422340393, "step": 25832 }, { "epoch": 2.036499802916831, "grad_norm": 0.5746637582778931, "learning_rate": 0.0005, "loss": 1.3557426929473877, "step": 25833 }, { "epoch": 2.03657863618447, "grad_norm": 1.511511206626892, "learning_rate": 0.0005, "loss": 1.3558201789855957, "step": 25834 }, { "epoch": 2.0366574694521087, "grad_norm": 0.4789819121360779, "learning_rate": 0.0005, "loss": 1.3235273361206055, "step": 25835 }, { "epoch": 2.036736302719748, "grad_norm": 0.5247492790222168, "learning_rate": 0.0005, "loss": 1.3138320446014404, "step": 25836 }, { "epoch": 2.0368151359873865, "grad_norm": 0.45246997475624084, "learning_rate": 0.0005, "loss": 1.2062983512878418, "step": 25837 }, { "epoch": 2.0368939692550256, "grad_norm": 3.073770761489868, "learning_rate": 0.0005, "loss": 1.2836086750030518, "step": 25838 }, { "epoch": 2.0369728025226648, "grad_norm": 0.5360348224639893, "learning_rate": 0.0005, "loss": 1.3496239185333252, "step": 25839 }, { "epoch": 2.0370516357903035, "grad_norm": 0.4808471202850342, "learning_rate": 0.0005, "loss": 1.1810534000396729, "step": 25840 }, { "epoch": 2.0371304690579426, "grad_norm": 0.5509829521179199, "learning_rate": 0.0005, "loss": 1.2054057121276855, "step": 25841 }, { "epoch": 2.0372093023255813, "grad_norm": 0.5122857093811035, "learning_rate": 0.0005, "loss": 1.2907171249389648, "step": 25842 }, { "epoch": 2.0372881355932204, "grad_norm": 0.4709721505641937, "learning_rate": 0.0005, "loss": 1.2898155450820923, "step": 25843 }, { "epoch": 2.037366968860859, "grad_norm": 0.5045527219772339, "learning_rate": 0.0005, "loss": 1.392820119857788, "step": 25844 }, { "epoch": 2.0374458021284982, "grad_norm": 0.435611754655838, "learning_rate": 0.0005, "loss": 1.2102200984954834, "step": 25845 }, { "epoch": 2.0375246353961374, "grad_norm": 0.5449081063270569, "learning_rate": 0.0005, "loss": 1.4946346282958984, "step": 25846 }, { "epoch": 2.037603468663776, "grad_norm": 0.46490442752838135, "learning_rate": 0.0005, "loss": 1.2263498306274414, "step": 25847 }, { "epoch": 2.037682301931415, "grad_norm": 0.3509756624698639, "learning_rate": 0.0005, "loss": 1.1104435920715332, "step": 25848 }, { "epoch": 2.037761135199054, "grad_norm": 0.6430375576019287, "learning_rate": 0.0005, "loss": 1.3367109298706055, "step": 25849 }, { "epoch": 2.037839968466693, "grad_norm": 0.4449286162853241, "learning_rate": 0.0005, "loss": 1.3152053356170654, "step": 25850 }, { "epoch": 2.0379188017343317, "grad_norm": 0.46236684918403625, "learning_rate": 0.0005, "loss": 1.2985289096832275, "step": 25851 }, { "epoch": 2.037997635001971, "grad_norm": 0.5265916585922241, "learning_rate": 0.0005, "loss": 1.3540974855422974, "step": 25852 }, { "epoch": 2.03807646826961, "grad_norm": 0.6534435749053955, "learning_rate": 0.0005, "loss": 1.442159652709961, "step": 25853 }, { "epoch": 2.0381553015372487, "grad_norm": 0.5196419358253479, "learning_rate": 0.0005, "loss": 1.4390641450881958, "step": 25854 }, { "epoch": 2.038234134804888, "grad_norm": 0.47509050369262695, "learning_rate": 0.0005, "loss": 1.2315864562988281, "step": 25855 }, { "epoch": 2.0383129680725265, "grad_norm": 0.4653134346008301, "learning_rate": 0.0005, "loss": 1.2751697301864624, "step": 25856 }, { "epoch": 2.0383918013401656, "grad_norm": 0.4334017336368561, "learning_rate": 0.0005, "loss": 1.2176328897476196, "step": 25857 }, { "epoch": 2.0384706346078043, "grad_norm": 0.507304310798645, "learning_rate": 0.0005, "loss": 1.3539224863052368, "step": 25858 }, { "epoch": 2.0385494678754434, "grad_norm": 0.35646262764930725, "learning_rate": 0.0005, "loss": 1.1666958332061768, "step": 25859 }, { "epoch": 2.0386283011430826, "grad_norm": 0.517910897731781, "learning_rate": 0.0005, "loss": 1.3134859800338745, "step": 25860 }, { "epoch": 2.0387071344107213, "grad_norm": 0.6228623390197754, "learning_rate": 0.0005, "loss": 1.2629708051681519, "step": 25861 }, { "epoch": 2.0387859676783604, "grad_norm": 0.5290647149085999, "learning_rate": 0.0005, "loss": 1.3815007209777832, "step": 25862 }, { "epoch": 2.038864800945999, "grad_norm": 0.40665891766548157, "learning_rate": 0.0005, "loss": 1.1650655269622803, "step": 25863 }, { "epoch": 2.0389436342136382, "grad_norm": 0.4165058135986328, "learning_rate": 0.0005, "loss": 1.308924674987793, "step": 25864 }, { "epoch": 2.039022467481277, "grad_norm": 0.5147824883460999, "learning_rate": 0.0005, "loss": 1.2336459159851074, "step": 25865 }, { "epoch": 2.039101300748916, "grad_norm": 0.555951714515686, "learning_rate": 0.0005, "loss": 1.40524160861969, "step": 25866 }, { "epoch": 2.039180134016555, "grad_norm": 0.5517331957817078, "learning_rate": 0.0005, "loss": 1.4090948104858398, "step": 25867 }, { "epoch": 2.039258967284194, "grad_norm": 0.4561215937137604, "learning_rate": 0.0005, "loss": 1.389420747756958, "step": 25868 }, { "epoch": 2.039337800551833, "grad_norm": 0.46855875849723816, "learning_rate": 0.0005, "loss": 1.3429501056671143, "step": 25869 }, { "epoch": 2.0394166338194717, "grad_norm": 0.515363335609436, "learning_rate": 0.0005, "loss": 1.2461585998535156, "step": 25870 }, { "epoch": 2.039495467087111, "grad_norm": 0.50583815574646, "learning_rate": 0.0005, "loss": 1.3040274381637573, "step": 25871 }, { "epoch": 2.0395743003547495, "grad_norm": 0.5096978545188904, "learning_rate": 0.0005, "loss": 1.3615280389785767, "step": 25872 }, { "epoch": 2.0396531336223886, "grad_norm": 0.5794070959091187, "learning_rate": 0.0005, "loss": 1.3610093593597412, "step": 25873 }, { "epoch": 2.039731966890028, "grad_norm": 0.4441487789154053, "learning_rate": 0.0005, "loss": 1.1858278512954712, "step": 25874 }, { "epoch": 2.0398108001576665, "grad_norm": 0.5291289687156677, "learning_rate": 0.0005, "loss": 1.5402607917785645, "step": 25875 }, { "epoch": 2.0398896334253056, "grad_norm": 0.665604829788208, "learning_rate": 0.0005, "loss": 1.4551827907562256, "step": 25876 }, { "epoch": 2.0399684666929443, "grad_norm": 0.5100714564323425, "learning_rate": 0.0005, "loss": 1.3095762729644775, "step": 25877 }, { "epoch": 2.0400472999605834, "grad_norm": 0.5459918975830078, "learning_rate": 0.0005, "loss": 1.2755733728408813, "step": 25878 }, { "epoch": 2.040126133228222, "grad_norm": 0.3988088369369507, "learning_rate": 0.0005, "loss": 1.289839744567871, "step": 25879 }, { "epoch": 2.0402049664958612, "grad_norm": 0.4645090699195862, "learning_rate": 0.0005, "loss": 1.323857307434082, "step": 25880 }, { "epoch": 2.0402837997635004, "grad_norm": 1.1831871271133423, "learning_rate": 0.0005, "loss": 1.6056007146835327, "step": 25881 }, { "epoch": 2.040362633031139, "grad_norm": 0.4496667981147766, "learning_rate": 0.0005, "loss": 1.1531164646148682, "step": 25882 }, { "epoch": 2.040441466298778, "grad_norm": 0.48245683312416077, "learning_rate": 0.0005, "loss": 1.300846815109253, "step": 25883 }, { "epoch": 2.040520299566417, "grad_norm": 0.5869547128677368, "learning_rate": 0.0005, "loss": 1.3660414218902588, "step": 25884 }, { "epoch": 2.040599132834056, "grad_norm": 0.45702099800109863, "learning_rate": 0.0005, "loss": 1.2580024003982544, "step": 25885 }, { "epoch": 2.0406779661016947, "grad_norm": 0.4776441752910614, "learning_rate": 0.0005, "loss": 1.3528566360473633, "step": 25886 }, { "epoch": 2.040756799369334, "grad_norm": 0.41057121753692627, "learning_rate": 0.0005, "loss": 1.2316539287567139, "step": 25887 }, { "epoch": 2.040835632636973, "grad_norm": 0.5256665945053101, "learning_rate": 0.0005, "loss": 1.3216958045959473, "step": 25888 }, { "epoch": 2.0409144659046117, "grad_norm": 0.5916429758071899, "learning_rate": 0.0005, "loss": 1.4133155345916748, "step": 25889 }, { "epoch": 2.040993299172251, "grad_norm": 0.45906031131744385, "learning_rate": 0.0005, "loss": 1.2031363248825073, "step": 25890 }, { "epoch": 2.0410721324398895, "grad_norm": 0.5658990144729614, "learning_rate": 0.0005, "loss": 1.4385572671890259, "step": 25891 }, { "epoch": 2.0411509657075286, "grad_norm": 0.46224692463874817, "learning_rate": 0.0005, "loss": 1.208073616027832, "step": 25892 }, { "epoch": 2.0412297989751673, "grad_norm": 0.47055572271347046, "learning_rate": 0.0005, "loss": 1.386534333229065, "step": 25893 }, { "epoch": 2.0413086322428065, "grad_norm": 0.46191710233688354, "learning_rate": 0.0005, "loss": 1.3820738792419434, "step": 25894 }, { "epoch": 2.0413874655104456, "grad_norm": 0.5390366911888123, "learning_rate": 0.0005, "loss": 1.2794831991195679, "step": 25895 }, { "epoch": 2.0414662987780843, "grad_norm": 0.5871425867080688, "learning_rate": 0.0005, "loss": 1.447662353515625, "step": 25896 }, { "epoch": 2.0415451320457234, "grad_norm": 0.47545376420021057, "learning_rate": 0.0005, "loss": 1.1697163581848145, "step": 25897 }, { "epoch": 2.041623965313362, "grad_norm": 0.47761043906211853, "learning_rate": 0.0005, "loss": 1.2741779088974, "step": 25898 }, { "epoch": 2.0417027985810012, "grad_norm": 0.4707794487476349, "learning_rate": 0.0005, "loss": 1.1861385107040405, "step": 25899 }, { "epoch": 2.04178163184864, "grad_norm": 0.404356986284256, "learning_rate": 0.0005, "loss": 1.1992124319076538, "step": 25900 }, { "epoch": 2.041860465116279, "grad_norm": 0.5031038522720337, "learning_rate": 0.0005, "loss": 1.466793179512024, "step": 25901 }, { "epoch": 2.041939298383918, "grad_norm": 0.41071996092796326, "learning_rate": 0.0005, "loss": 1.2978264093399048, "step": 25902 }, { "epoch": 2.042018131651557, "grad_norm": 0.4756166934967041, "learning_rate": 0.0005, "loss": 1.3460561037063599, "step": 25903 }, { "epoch": 2.042096964919196, "grad_norm": 0.5015752911567688, "learning_rate": 0.0005, "loss": 1.606749176979065, "step": 25904 }, { "epoch": 2.0421757981868347, "grad_norm": 0.4365467131137848, "learning_rate": 0.0005, "loss": 1.2315632104873657, "step": 25905 }, { "epoch": 2.042254631454474, "grad_norm": 0.4923493564128876, "learning_rate": 0.0005, "loss": 1.4233920574188232, "step": 25906 }, { "epoch": 2.0423334647221125, "grad_norm": 0.4787697494029999, "learning_rate": 0.0005, "loss": 1.284686803817749, "step": 25907 }, { "epoch": 2.0424122979897517, "grad_norm": 0.4112808108329773, "learning_rate": 0.0005, "loss": 1.2996660470962524, "step": 25908 }, { "epoch": 2.042491131257391, "grad_norm": 0.437529057264328, "learning_rate": 0.0005, "loss": 1.3131804466247559, "step": 25909 }, { "epoch": 2.0425699645250295, "grad_norm": 0.4595186114311218, "learning_rate": 0.0005, "loss": 1.199190616607666, "step": 25910 }, { "epoch": 2.0426487977926686, "grad_norm": 0.4905855059623718, "learning_rate": 0.0005, "loss": 1.287261962890625, "step": 25911 }, { "epoch": 2.0427276310603073, "grad_norm": 0.5581175684928894, "learning_rate": 0.0005, "loss": 1.4527904987335205, "step": 25912 }, { "epoch": 2.0428064643279464, "grad_norm": 0.5289123058319092, "learning_rate": 0.0005, "loss": 1.4502187967300415, "step": 25913 }, { "epoch": 2.042885297595585, "grad_norm": 0.5578375458717346, "learning_rate": 0.0005, "loss": 1.1905789375305176, "step": 25914 }, { "epoch": 2.0429641308632243, "grad_norm": 0.44866952300071716, "learning_rate": 0.0005, "loss": 1.2149345874786377, "step": 25915 }, { "epoch": 2.0430429641308634, "grad_norm": 0.5030577182769775, "learning_rate": 0.0005, "loss": 1.221590280532837, "step": 25916 }, { "epoch": 2.043121797398502, "grad_norm": 0.46245449781417847, "learning_rate": 0.0005, "loss": 1.19920015335083, "step": 25917 }, { "epoch": 2.043200630666141, "grad_norm": 0.5158913731575012, "learning_rate": 0.0005, "loss": 1.3632481098175049, "step": 25918 }, { "epoch": 2.04327946393378, "grad_norm": 0.647331178188324, "learning_rate": 0.0005, "loss": 1.5059735774993896, "step": 25919 }, { "epoch": 2.043358297201419, "grad_norm": 0.4681124985218048, "learning_rate": 0.0005, "loss": 1.2346159219741821, "step": 25920 }, { "epoch": 2.0434371304690577, "grad_norm": 0.5064473152160645, "learning_rate": 0.0005, "loss": 1.3079479932785034, "step": 25921 }, { "epoch": 2.043515963736697, "grad_norm": 0.41897329688072205, "learning_rate": 0.0005, "loss": 1.1236554384231567, "step": 25922 }, { "epoch": 2.043594797004336, "grad_norm": 0.5459412932395935, "learning_rate": 0.0005, "loss": 1.3840069770812988, "step": 25923 }, { "epoch": 2.0436736302719747, "grad_norm": 0.571243166923523, "learning_rate": 0.0005, "loss": 1.221712589263916, "step": 25924 }, { "epoch": 2.043752463539614, "grad_norm": 0.4314865171909332, "learning_rate": 0.0005, "loss": 1.308150291442871, "step": 25925 }, { "epoch": 2.0438312968072525, "grad_norm": 0.5474942326545715, "learning_rate": 0.0005, "loss": 1.3871889114379883, "step": 25926 }, { "epoch": 2.0439101300748916, "grad_norm": 0.5197153687477112, "learning_rate": 0.0005, "loss": 1.414132833480835, "step": 25927 }, { "epoch": 2.0439889633425303, "grad_norm": 0.543913722038269, "learning_rate": 0.0005, "loss": 1.3671762943267822, "step": 25928 }, { "epoch": 2.0440677966101695, "grad_norm": 0.6077635288238525, "learning_rate": 0.0005, "loss": 1.543853521347046, "step": 25929 }, { "epoch": 2.0441466298778086, "grad_norm": 0.576226532459259, "learning_rate": 0.0005, "loss": 1.1273465156555176, "step": 25930 }, { "epoch": 2.0442254631454473, "grad_norm": 0.5064948797225952, "learning_rate": 0.0005, "loss": 1.3234130144119263, "step": 25931 }, { "epoch": 2.0443042964130864, "grad_norm": 0.4515829384326935, "learning_rate": 0.0005, "loss": 1.3052902221679688, "step": 25932 }, { "epoch": 2.044383129680725, "grad_norm": 0.6098263263702393, "learning_rate": 0.0005, "loss": 1.5205204486846924, "step": 25933 }, { "epoch": 2.0444619629483642, "grad_norm": 0.6964028477668762, "learning_rate": 0.0005, "loss": 1.5278195142745972, "step": 25934 }, { "epoch": 2.044540796216003, "grad_norm": 0.4603413939476013, "learning_rate": 0.0005, "loss": 1.5019168853759766, "step": 25935 }, { "epoch": 2.044619629483642, "grad_norm": 0.4598941206932068, "learning_rate": 0.0005, "loss": 1.2531487941741943, "step": 25936 }, { "epoch": 2.044698462751281, "grad_norm": 0.5297213792800903, "learning_rate": 0.0005, "loss": 1.151389479637146, "step": 25937 }, { "epoch": 2.04477729601892, "grad_norm": 0.49874353408813477, "learning_rate": 0.0005, "loss": 1.403632402420044, "step": 25938 }, { "epoch": 2.044856129286559, "grad_norm": 0.7100439071655273, "learning_rate": 0.0005, "loss": 1.3674168586730957, "step": 25939 }, { "epoch": 2.0449349625541977, "grad_norm": 0.5120784044265747, "learning_rate": 0.0005, "loss": 1.432999849319458, "step": 25940 }, { "epoch": 2.045013795821837, "grad_norm": 0.5083380937576294, "learning_rate": 0.0005, "loss": 1.2073500156402588, "step": 25941 }, { "epoch": 2.045092629089476, "grad_norm": 0.39979979395866394, "learning_rate": 0.0005, "loss": 1.1482867002487183, "step": 25942 }, { "epoch": 2.0451714623571147, "grad_norm": 0.7462778091430664, "learning_rate": 0.0005, "loss": 1.2952430248260498, "step": 25943 }, { "epoch": 2.045250295624754, "grad_norm": 0.4308054745197296, "learning_rate": 0.0005, "loss": 1.3447154760360718, "step": 25944 }, { "epoch": 2.0453291288923925, "grad_norm": 0.5292385220527649, "learning_rate": 0.0005, "loss": 1.2515599727630615, "step": 25945 }, { "epoch": 2.0454079621600316, "grad_norm": 0.5581149458885193, "learning_rate": 0.0005, "loss": 1.401707649230957, "step": 25946 }, { "epoch": 2.0454867954276703, "grad_norm": 0.4668840169906616, "learning_rate": 0.0005, "loss": 1.4202598333358765, "step": 25947 }, { "epoch": 2.0455656286953094, "grad_norm": 0.5736372470855713, "learning_rate": 0.0005, "loss": 1.4392409324645996, "step": 25948 }, { "epoch": 2.0456444619629486, "grad_norm": 0.5498201847076416, "learning_rate": 0.0005, "loss": 1.3352608680725098, "step": 25949 }, { "epoch": 2.0457232952305873, "grad_norm": 0.5703774690628052, "learning_rate": 0.0005, "loss": 1.5548429489135742, "step": 25950 }, { "epoch": 2.0458021284982264, "grad_norm": 0.4191127121448517, "learning_rate": 0.0005, "loss": 1.2498667240142822, "step": 25951 }, { "epoch": 2.045880961765865, "grad_norm": 0.6222320795059204, "learning_rate": 0.0005, "loss": 1.5342967510223389, "step": 25952 }, { "epoch": 2.0459597950335042, "grad_norm": 0.4869459867477417, "learning_rate": 0.0005, "loss": 1.3468945026397705, "step": 25953 }, { "epoch": 2.046038628301143, "grad_norm": 1.0536131858825684, "learning_rate": 0.0005, "loss": 1.5477769374847412, "step": 25954 }, { "epoch": 2.046117461568782, "grad_norm": 0.3915019631385803, "learning_rate": 0.0005, "loss": 1.1857571601867676, "step": 25955 }, { "epoch": 2.046196294836421, "grad_norm": 0.5280746221542358, "learning_rate": 0.0005, "loss": 1.3114570379257202, "step": 25956 }, { "epoch": 2.04627512810406, "grad_norm": 0.4204423129558563, "learning_rate": 0.0005, "loss": 1.2302619218826294, "step": 25957 }, { "epoch": 2.046353961371699, "grad_norm": 0.5202769041061401, "learning_rate": 0.0005, "loss": 1.404017448425293, "step": 25958 }, { "epoch": 2.0464327946393377, "grad_norm": 0.46404749155044556, "learning_rate": 0.0005, "loss": 1.353058099746704, "step": 25959 }, { "epoch": 2.046511627906977, "grad_norm": 0.4196571409702301, "learning_rate": 0.0005, "loss": 1.3333780765533447, "step": 25960 }, { "epoch": 2.0465904611746155, "grad_norm": 0.5043893456459045, "learning_rate": 0.0005, "loss": 1.422480821609497, "step": 25961 }, { "epoch": 2.0466692944422546, "grad_norm": 0.47215819358825684, "learning_rate": 0.0005, "loss": 1.3753836154937744, "step": 25962 }, { "epoch": 2.046748127709894, "grad_norm": 0.7479338645935059, "learning_rate": 0.0005, "loss": 1.525530219078064, "step": 25963 }, { "epoch": 2.0468269609775325, "grad_norm": 0.4649406373500824, "learning_rate": 0.0005, "loss": 1.2305498123168945, "step": 25964 }, { "epoch": 2.0469057942451716, "grad_norm": 0.5306258201599121, "learning_rate": 0.0005, "loss": 1.3221172094345093, "step": 25965 }, { "epoch": 2.0469846275128103, "grad_norm": 0.4682806134223938, "learning_rate": 0.0005, "loss": 1.0806324481964111, "step": 25966 }, { "epoch": 2.0470634607804494, "grad_norm": 0.5250673294067383, "learning_rate": 0.0005, "loss": 1.343286156654358, "step": 25967 }, { "epoch": 2.047142294048088, "grad_norm": 0.5506716966629028, "learning_rate": 0.0005, "loss": 1.379350185394287, "step": 25968 }, { "epoch": 2.0472211273157273, "grad_norm": 0.5383493900299072, "learning_rate": 0.0005, "loss": 1.3142857551574707, "step": 25969 }, { "epoch": 2.0472999605833664, "grad_norm": 0.45245206356048584, "learning_rate": 0.0005, "loss": 1.2067570686340332, "step": 25970 }, { "epoch": 2.047378793851005, "grad_norm": 0.8878797888755798, "learning_rate": 0.0005, "loss": 1.1604111194610596, "step": 25971 }, { "epoch": 2.047457627118644, "grad_norm": 0.5710999369621277, "learning_rate": 0.0005, "loss": 1.5108617544174194, "step": 25972 }, { "epoch": 2.047536460386283, "grad_norm": 0.5731435418128967, "learning_rate": 0.0005, "loss": 1.4016075134277344, "step": 25973 }, { "epoch": 2.047615293653922, "grad_norm": 0.4407535195350647, "learning_rate": 0.0005, "loss": 1.2991772890090942, "step": 25974 }, { "epoch": 2.0476941269215607, "grad_norm": 0.49786701798439026, "learning_rate": 0.0005, "loss": 1.3976199626922607, "step": 25975 }, { "epoch": 2.0477729601892, "grad_norm": 0.4382496476173401, "learning_rate": 0.0005, "loss": 1.328330159187317, "step": 25976 }, { "epoch": 2.047851793456839, "grad_norm": 0.4920887053012848, "learning_rate": 0.0005, "loss": 1.4191033840179443, "step": 25977 }, { "epoch": 2.0479306267244777, "grad_norm": 0.4229976534843445, "learning_rate": 0.0005, "loss": 1.1952873468399048, "step": 25978 }, { "epoch": 2.048009459992117, "grad_norm": 0.4728405773639679, "learning_rate": 0.0005, "loss": 1.068189263343811, "step": 25979 }, { "epoch": 2.0480882932597555, "grad_norm": 0.42026814818382263, "learning_rate": 0.0005, "loss": 1.291037678718567, "step": 25980 }, { "epoch": 2.0481671265273946, "grad_norm": 0.5612468123435974, "learning_rate": 0.0005, "loss": 1.4818928241729736, "step": 25981 }, { "epoch": 2.0482459597950333, "grad_norm": 0.4672275483608246, "learning_rate": 0.0005, "loss": 1.263731837272644, "step": 25982 }, { "epoch": 2.0483247930626725, "grad_norm": 0.44281983375549316, "learning_rate": 0.0005, "loss": 1.2821874618530273, "step": 25983 }, { "epoch": 2.0484036263303116, "grad_norm": 0.5071999430656433, "learning_rate": 0.0005, "loss": 1.3415954113006592, "step": 25984 }, { "epoch": 2.0484824595979503, "grad_norm": 0.4778009057044983, "learning_rate": 0.0005, "loss": 1.2944352626800537, "step": 25985 }, { "epoch": 2.0485612928655894, "grad_norm": 0.546836793422699, "learning_rate": 0.0005, "loss": 1.3022494316101074, "step": 25986 }, { "epoch": 2.048640126133228, "grad_norm": 0.606317400932312, "learning_rate": 0.0005, "loss": 1.5044126510620117, "step": 25987 }, { "epoch": 2.0487189594008672, "grad_norm": 0.41674768924713135, "learning_rate": 0.0005, "loss": 1.3098076581954956, "step": 25988 }, { "epoch": 2.048797792668506, "grad_norm": 0.6456282734870911, "learning_rate": 0.0005, "loss": 1.210364580154419, "step": 25989 }, { "epoch": 2.048876625936145, "grad_norm": 0.43467816710472107, "learning_rate": 0.0005, "loss": 1.3037762641906738, "step": 25990 }, { "epoch": 2.048955459203784, "grad_norm": 0.5796558856964111, "learning_rate": 0.0005, "loss": 1.3945484161376953, "step": 25991 }, { "epoch": 2.049034292471423, "grad_norm": 0.5241223573684692, "learning_rate": 0.0005, "loss": 1.3895514011383057, "step": 25992 }, { "epoch": 2.049113125739062, "grad_norm": 0.4807490408420563, "learning_rate": 0.0005, "loss": 1.3323206901550293, "step": 25993 }, { "epoch": 2.0491919590067007, "grad_norm": 0.5277844071388245, "learning_rate": 0.0005, "loss": 1.2123658657073975, "step": 25994 }, { "epoch": 2.04927079227434, "grad_norm": 0.454829603433609, "learning_rate": 0.0005, "loss": 1.2493098974227905, "step": 25995 }, { "epoch": 2.0493496255419785, "grad_norm": 0.5344671607017517, "learning_rate": 0.0005, "loss": 1.3457744121551514, "step": 25996 }, { "epoch": 2.0494284588096177, "grad_norm": 0.5047208070755005, "learning_rate": 0.0005, "loss": 1.4393858909606934, "step": 25997 }, { "epoch": 2.049507292077257, "grad_norm": 0.4584576189517975, "learning_rate": 0.0005, "loss": 1.3167507648468018, "step": 25998 }, { "epoch": 2.0495861253448955, "grad_norm": 0.4747964143753052, "learning_rate": 0.0005, "loss": 1.2233164310455322, "step": 25999 }, { "epoch": 2.0496649586125346, "grad_norm": 0.5681489706039429, "learning_rate": 0.0005, "loss": 1.3512516021728516, "step": 26000 }, { "epoch": 2.0497437918801733, "grad_norm": 0.506511926651001, "learning_rate": 0.0005, "loss": 1.4200959205627441, "step": 26001 }, { "epoch": 2.0498226251478124, "grad_norm": 0.5472273826599121, "learning_rate": 0.0005, "loss": 1.5578193664550781, "step": 26002 }, { "epoch": 2.049901458415451, "grad_norm": 0.49793294072151184, "learning_rate": 0.0005, "loss": 1.4470385313034058, "step": 26003 }, { "epoch": 2.0499802916830903, "grad_norm": 0.6539878249168396, "learning_rate": 0.0005, "loss": 1.391902208328247, "step": 26004 }, { "epoch": 2.0500591249507294, "grad_norm": 0.5039684176445007, "learning_rate": 0.0005, "loss": 1.1323373317718506, "step": 26005 }, { "epoch": 2.050137958218368, "grad_norm": 0.47458887100219727, "learning_rate": 0.0005, "loss": 1.3087029457092285, "step": 26006 }, { "epoch": 2.050216791486007, "grad_norm": 0.504326343536377, "learning_rate": 0.0005, "loss": 1.3232548236846924, "step": 26007 }, { "epoch": 2.050295624753646, "grad_norm": 0.5371454954147339, "learning_rate": 0.0005, "loss": 1.2742867469787598, "step": 26008 }, { "epoch": 2.050374458021285, "grad_norm": 0.5604483485221863, "learning_rate": 0.0005, "loss": 1.3706746101379395, "step": 26009 }, { "epoch": 2.0504532912889237, "grad_norm": 0.5587525367736816, "learning_rate": 0.0005, "loss": 1.3009823560714722, "step": 26010 }, { "epoch": 2.050532124556563, "grad_norm": 0.4784590005874634, "learning_rate": 0.0005, "loss": 1.3461824655532837, "step": 26011 }, { "epoch": 2.050610957824202, "grad_norm": 0.503360390663147, "learning_rate": 0.0005, "loss": 1.3306314945220947, "step": 26012 }, { "epoch": 2.0506897910918407, "grad_norm": 0.627159595489502, "learning_rate": 0.0005, "loss": 1.3403444290161133, "step": 26013 }, { "epoch": 2.05076862435948, "grad_norm": 0.38417714834213257, "learning_rate": 0.0005, "loss": 1.145771861076355, "step": 26014 }, { "epoch": 2.0508474576271185, "grad_norm": 0.5056053400039673, "learning_rate": 0.0005, "loss": 1.3599517345428467, "step": 26015 }, { "epoch": 2.0509262908947576, "grad_norm": 0.5349212288856506, "learning_rate": 0.0005, "loss": 1.3879218101501465, "step": 26016 }, { "epoch": 2.0510051241623963, "grad_norm": 0.520756185054779, "learning_rate": 0.0005, "loss": 1.4119635820388794, "step": 26017 }, { "epoch": 2.0510839574300355, "grad_norm": 0.58586186170578, "learning_rate": 0.0005, "loss": 1.4175653457641602, "step": 26018 }, { "epoch": 2.0511627906976746, "grad_norm": 0.4202165901660919, "learning_rate": 0.0005, "loss": 1.334930181503296, "step": 26019 }, { "epoch": 2.0512416239653133, "grad_norm": 0.47426697611808777, "learning_rate": 0.0005, "loss": 1.2241554260253906, "step": 26020 }, { "epoch": 2.0513204572329524, "grad_norm": 0.45038002729415894, "learning_rate": 0.0005, "loss": 1.3401178121566772, "step": 26021 }, { "epoch": 2.051399290500591, "grad_norm": 0.4914502501487732, "learning_rate": 0.0005, "loss": 1.366774082183838, "step": 26022 }, { "epoch": 2.0514781237682302, "grad_norm": 0.461548775434494, "learning_rate": 0.0005, "loss": 1.4790688753128052, "step": 26023 }, { "epoch": 2.051556957035869, "grad_norm": 0.4989475905895233, "learning_rate": 0.0005, "loss": 1.440550446510315, "step": 26024 }, { "epoch": 2.051635790303508, "grad_norm": 0.5314591526985168, "learning_rate": 0.0005, "loss": 1.2614035606384277, "step": 26025 }, { "epoch": 2.051714623571147, "grad_norm": 0.6231496334075928, "learning_rate": 0.0005, "loss": 1.2121734619140625, "step": 26026 }, { "epoch": 2.051793456838786, "grad_norm": 0.48908084630966187, "learning_rate": 0.0005, "loss": 1.3192412853240967, "step": 26027 }, { "epoch": 2.051872290106425, "grad_norm": 0.6027082204818726, "learning_rate": 0.0005, "loss": 1.5537816286087036, "step": 26028 }, { "epoch": 2.0519511233740637, "grad_norm": 0.4944860339164734, "learning_rate": 0.0005, "loss": 1.1343872547149658, "step": 26029 }, { "epoch": 2.052029956641703, "grad_norm": 0.5656280517578125, "learning_rate": 0.0005, "loss": 1.4311761856079102, "step": 26030 }, { "epoch": 2.0521087899093415, "grad_norm": 0.40245532989501953, "learning_rate": 0.0005, "loss": 1.3001396656036377, "step": 26031 }, { "epoch": 2.0521876231769807, "grad_norm": 0.41766366362571716, "learning_rate": 0.0005, "loss": 1.2039988040924072, "step": 26032 }, { "epoch": 2.05226645644462, "grad_norm": 0.42674288153648376, "learning_rate": 0.0005, "loss": 1.1287574768066406, "step": 26033 }, { "epoch": 2.0523452897122585, "grad_norm": 0.42391273379325867, "learning_rate": 0.0005, "loss": 1.257225751876831, "step": 26034 }, { "epoch": 2.0524241229798976, "grad_norm": 0.431138813495636, "learning_rate": 0.0005, "loss": 1.3713734149932861, "step": 26035 }, { "epoch": 2.0525029562475363, "grad_norm": 0.506570041179657, "learning_rate": 0.0005, "loss": 1.305964708328247, "step": 26036 }, { "epoch": 2.0525817895151754, "grad_norm": 0.42371538281440735, "learning_rate": 0.0005, "loss": 1.342293381690979, "step": 26037 }, { "epoch": 2.052660622782814, "grad_norm": 0.5594090819358826, "learning_rate": 0.0005, "loss": 1.5281188488006592, "step": 26038 }, { "epoch": 2.0527394560504533, "grad_norm": 0.5625066161155701, "learning_rate": 0.0005, "loss": 1.4247159957885742, "step": 26039 }, { "epoch": 2.0528182893180924, "grad_norm": 0.5424532890319824, "learning_rate": 0.0005, "loss": 1.4009859561920166, "step": 26040 }, { "epoch": 2.052897122585731, "grad_norm": 0.39211827516555786, "learning_rate": 0.0005, "loss": 1.304377794265747, "step": 26041 }, { "epoch": 2.0529759558533702, "grad_norm": 0.6763414144515991, "learning_rate": 0.0005, "loss": 1.3358135223388672, "step": 26042 }, { "epoch": 2.053054789121009, "grad_norm": 0.5556341409683228, "learning_rate": 0.0005, "loss": 1.4453814029693604, "step": 26043 }, { "epoch": 2.053133622388648, "grad_norm": 0.6293908953666687, "learning_rate": 0.0005, "loss": 1.5662565231323242, "step": 26044 }, { "epoch": 2.0532124556562867, "grad_norm": 0.5730984210968018, "learning_rate": 0.0005, "loss": 1.2433536052703857, "step": 26045 }, { "epoch": 2.053291288923926, "grad_norm": 0.469746470451355, "learning_rate": 0.0005, "loss": 1.239975929260254, "step": 26046 }, { "epoch": 2.053370122191565, "grad_norm": 0.48505881428718567, "learning_rate": 0.0005, "loss": 1.3102622032165527, "step": 26047 }, { "epoch": 2.0534489554592037, "grad_norm": 0.41437965631484985, "learning_rate": 0.0005, "loss": 1.345567226409912, "step": 26048 }, { "epoch": 2.053527788726843, "grad_norm": 0.5623729228973389, "learning_rate": 0.0005, "loss": 1.2369619607925415, "step": 26049 }, { "epoch": 2.0536066219944815, "grad_norm": 0.48564818501472473, "learning_rate": 0.0005, "loss": 1.4053372144699097, "step": 26050 }, { "epoch": 2.0536854552621207, "grad_norm": 0.5006498098373413, "learning_rate": 0.0005, "loss": 1.3070412874221802, "step": 26051 }, { "epoch": 2.0537642885297593, "grad_norm": 0.563949704170227, "learning_rate": 0.0005, "loss": 1.417680025100708, "step": 26052 }, { "epoch": 2.0538431217973985, "grad_norm": 0.5111717581748962, "learning_rate": 0.0005, "loss": 1.541813611984253, "step": 26053 }, { "epoch": 2.0539219550650376, "grad_norm": 0.45781415700912476, "learning_rate": 0.0005, "loss": 1.2979516983032227, "step": 26054 }, { "epoch": 2.0540007883326763, "grad_norm": 0.5119307637214661, "learning_rate": 0.0005, "loss": 1.3380321264266968, "step": 26055 }, { "epoch": 2.0540796216003154, "grad_norm": 0.4543687701225281, "learning_rate": 0.0005, "loss": 1.451972484588623, "step": 26056 }, { "epoch": 2.054158454867954, "grad_norm": 0.4437936544418335, "learning_rate": 0.0005, "loss": 1.303943157196045, "step": 26057 }, { "epoch": 2.0542372881355933, "grad_norm": 0.5903225541114807, "learning_rate": 0.0005, "loss": 1.4127800464630127, "step": 26058 }, { "epoch": 2.0543161214032324, "grad_norm": 0.5491580963134766, "learning_rate": 0.0005, "loss": 1.3871026039123535, "step": 26059 }, { "epoch": 2.054394954670871, "grad_norm": 0.8071620464324951, "learning_rate": 0.0005, "loss": 1.3720210790634155, "step": 26060 }, { "epoch": 2.05447378793851, "grad_norm": 0.475902795791626, "learning_rate": 0.0005, "loss": 1.2421443462371826, "step": 26061 }, { "epoch": 2.054552621206149, "grad_norm": 0.5846650004386902, "learning_rate": 0.0005, "loss": 1.1714290380477905, "step": 26062 }, { "epoch": 2.054631454473788, "grad_norm": 0.5927297472953796, "learning_rate": 0.0005, "loss": 1.2332534790039062, "step": 26063 }, { "epoch": 2.0547102877414267, "grad_norm": 0.45592284202575684, "learning_rate": 0.0005, "loss": 1.1452618837356567, "step": 26064 }, { "epoch": 2.054789121009066, "grad_norm": 0.587213397026062, "learning_rate": 0.0005, "loss": 1.3288356065750122, "step": 26065 }, { "epoch": 2.0548679542767045, "grad_norm": 0.5449798703193665, "learning_rate": 0.0005, "loss": 1.298187017440796, "step": 26066 }, { "epoch": 2.0549467875443437, "grad_norm": 0.539412260055542, "learning_rate": 0.0005, "loss": 1.4726409912109375, "step": 26067 }, { "epoch": 2.055025620811983, "grad_norm": 0.5523949265480042, "learning_rate": 0.0005, "loss": 1.5726017951965332, "step": 26068 }, { "epoch": 2.0551044540796215, "grad_norm": 1.754319190979004, "learning_rate": 0.0005, "loss": 1.3078532218933105, "step": 26069 }, { "epoch": 2.0551832873472606, "grad_norm": 0.46531572937965393, "learning_rate": 0.0005, "loss": 1.3187493085861206, "step": 26070 }, { "epoch": 2.0552621206148993, "grad_norm": 0.4503249228000641, "learning_rate": 0.0005, "loss": 1.2475333213806152, "step": 26071 }, { "epoch": 2.0553409538825385, "grad_norm": 0.4064640700817108, "learning_rate": 0.0005, "loss": 1.1679120063781738, "step": 26072 }, { "epoch": 2.0554197871501776, "grad_norm": 0.6974788904190063, "learning_rate": 0.0005, "loss": 1.2573621273040771, "step": 26073 }, { "epoch": 2.0554986204178163, "grad_norm": 0.4905189275741577, "learning_rate": 0.0005, "loss": 1.325180172920227, "step": 26074 }, { "epoch": 2.0555774536854554, "grad_norm": 0.5646219253540039, "learning_rate": 0.0005, "loss": 1.2026419639587402, "step": 26075 }, { "epoch": 2.055656286953094, "grad_norm": 0.490592896938324, "learning_rate": 0.0005, "loss": 1.2899816036224365, "step": 26076 }, { "epoch": 2.0557351202207332, "grad_norm": 0.6363793015480042, "learning_rate": 0.0005, "loss": 1.5571080446243286, "step": 26077 }, { "epoch": 2.055813953488372, "grad_norm": 0.5064179301261902, "learning_rate": 0.0005, "loss": 1.486234188079834, "step": 26078 }, { "epoch": 2.055892786756011, "grad_norm": 0.47060734033584595, "learning_rate": 0.0005, "loss": 1.2881888151168823, "step": 26079 }, { "epoch": 2.05597162002365, "grad_norm": 0.5274221301078796, "learning_rate": 0.0005, "loss": 1.6219630241394043, "step": 26080 }, { "epoch": 2.056050453291289, "grad_norm": 0.5341755747795105, "learning_rate": 0.0005, "loss": 1.3136624097824097, "step": 26081 }, { "epoch": 2.056129286558928, "grad_norm": 0.4575740396976471, "learning_rate": 0.0005, "loss": 1.0844788551330566, "step": 26082 }, { "epoch": 2.0562081198265667, "grad_norm": 0.5568159222602844, "learning_rate": 0.0005, "loss": 1.4755520820617676, "step": 26083 }, { "epoch": 2.056286953094206, "grad_norm": 0.5292547941207886, "learning_rate": 0.0005, "loss": 1.2695740461349487, "step": 26084 }, { "epoch": 2.0563657863618445, "grad_norm": 0.6274425387382507, "learning_rate": 0.0005, "loss": 1.4515904188156128, "step": 26085 }, { "epoch": 2.0564446196294837, "grad_norm": 0.4028599262237549, "learning_rate": 0.0005, "loss": 1.1777567863464355, "step": 26086 }, { "epoch": 2.056523452897123, "grad_norm": 0.42265185713768005, "learning_rate": 0.0005, "loss": 1.3456206321716309, "step": 26087 }, { "epoch": 2.0566022861647615, "grad_norm": 0.5733193159103394, "learning_rate": 0.0005, "loss": 1.5263638496398926, "step": 26088 }, { "epoch": 2.0566811194324006, "grad_norm": 0.46479806303977966, "learning_rate": 0.0005, "loss": 1.262009859085083, "step": 26089 }, { "epoch": 2.0567599527000393, "grad_norm": 0.4101850986480713, "learning_rate": 0.0005, "loss": 1.2869325876235962, "step": 26090 }, { "epoch": 2.0568387859676784, "grad_norm": 0.4943801760673523, "learning_rate": 0.0005, "loss": 1.4120111465454102, "step": 26091 }, { "epoch": 2.056917619235317, "grad_norm": 0.5870614051818848, "learning_rate": 0.0005, "loss": 1.5056970119476318, "step": 26092 }, { "epoch": 2.0569964525029563, "grad_norm": 0.45471277832984924, "learning_rate": 0.0005, "loss": 1.238511323928833, "step": 26093 }, { "epoch": 2.0570752857705954, "grad_norm": 0.5035858750343323, "learning_rate": 0.0005, "loss": 1.3648781776428223, "step": 26094 }, { "epoch": 2.057154119038234, "grad_norm": 0.5325511693954468, "learning_rate": 0.0005, "loss": 1.3655023574829102, "step": 26095 }, { "epoch": 2.057232952305873, "grad_norm": 0.4794139564037323, "learning_rate": 0.0005, "loss": 1.2277307510375977, "step": 26096 }, { "epoch": 2.057311785573512, "grad_norm": 0.5772789716720581, "learning_rate": 0.0005, "loss": 1.490315556526184, "step": 26097 }, { "epoch": 2.057390618841151, "grad_norm": 0.489040732383728, "learning_rate": 0.0005, "loss": 1.3040776252746582, "step": 26098 }, { "epoch": 2.0574694521087897, "grad_norm": 0.5307244658470154, "learning_rate": 0.0005, "loss": 1.3768761157989502, "step": 26099 }, { "epoch": 2.057548285376429, "grad_norm": 0.5415270328521729, "learning_rate": 0.0005, "loss": 1.269577980041504, "step": 26100 }, { "epoch": 2.057627118644068, "grad_norm": 0.4653851389884949, "learning_rate": 0.0005, "loss": 1.367538571357727, "step": 26101 }, { "epoch": 2.0577059519117067, "grad_norm": 0.5233780145645142, "learning_rate": 0.0005, "loss": 1.3080568313598633, "step": 26102 }, { "epoch": 2.057784785179346, "grad_norm": 0.4650595486164093, "learning_rate": 0.0005, "loss": 1.3032171726226807, "step": 26103 }, { "epoch": 2.0578636184469845, "grad_norm": 0.5179500579833984, "learning_rate": 0.0005, "loss": 1.3352720737457275, "step": 26104 }, { "epoch": 2.0579424517146236, "grad_norm": 0.5632611513137817, "learning_rate": 0.0005, "loss": 1.5998713970184326, "step": 26105 }, { "epoch": 2.0580212849822623, "grad_norm": 0.5725659728050232, "learning_rate": 0.0005, "loss": 1.2856817245483398, "step": 26106 }, { "epoch": 2.0581001182499015, "grad_norm": 0.5952774286270142, "learning_rate": 0.0005, "loss": 1.3225619792938232, "step": 26107 }, { "epoch": 2.0581789515175406, "grad_norm": 0.47732964158058167, "learning_rate": 0.0005, "loss": 1.3587751388549805, "step": 26108 }, { "epoch": 2.0582577847851793, "grad_norm": 0.5095391869544983, "learning_rate": 0.0005, "loss": 1.2455742359161377, "step": 26109 }, { "epoch": 2.0583366180528184, "grad_norm": 0.398939311504364, "learning_rate": 0.0005, "loss": 1.2185876369476318, "step": 26110 }, { "epoch": 2.058415451320457, "grad_norm": 0.5753827095031738, "learning_rate": 0.0005, "loss": 1.2737908363342285, "step": 26111 }, { "epoch": 2.0584942845880962, "grad_norm": 0.5128607153892517, "learning_rate": 0.0005, "loss": 1.3578349351882935, "step": 26112 }, { "epoch": 2.058573117855735, "grad_norm": 0.4522571265697479, "learning_rate": 0.0005, "loss": 1.3217823505401611, "step": 26113 }, { "epoch": 2.058651951123374, "grad_norm": 0.5891952514648438, "learning_rate": 0.0005, "loss": 1.467913269996643, "step": 26114 }, { "epoch": 2.058730784391013, "grad_norm": 0.4758884906768799, "learning_rate": 0.0005, "loss": 1.376963496208191, "step": 26115 }, { "epoch": 2.058809617658652, "grad_norm": 0.5983033776283264, "learning_rate": 0.0005, "loss": 1.34568190574646, "step": 26116 }, { "epoch": 2.058888450926291, "grad_norm": 0.4530392289161682, "learning_rate": 0.0005, "loss": 1.3850443363189697, "step": 26117 }, { "epoch": 2.0589672841939297, "grad_norm": 0.4581982493400574, "learning_rate": 0.0005, "loss": 1.2300224304199219, "step": 26118 }, { "epoch": 2.059046117461569, "grad_norm": 0.624767541885376, "learning_rate": 0.0005, "loss": 1.496584415435791, "step": 26119 }, { "epoch": 2.0591249507292075, "grad_norm": 0.47105684876441956, "learning_rate": 0.0005, "loss": 1.3094731569290161, "step": 26120 }, { "epoch": 2.0592037839968467, "grad_norm": 0.5039126873016357, "learning_rate": 0.0005, "loss": 1.2505736351013184, "step": 26121 }, { "epoch": 2.059282617264486, "grad_norm": 0.4324607849121094, "learning_rate": 0.0005, "loss": 1.2916126251220703, "step": 26122 }, { "epoch": 2.0593614505321245, "grad_norm": 0.4899342656135559, "learning_rate": 0.0005, "loss": 1.4796727895736694, "step": 26123 }, { "epoch": 2.0594402837997636, "grad_norm": 0.4601598381996155, "learning_rate": 0.0005, "loss": 1.2673275470733643, "step": 26124 }, { "epoch": 2.0595191170674023, "grad_norm": 0.7624434232711792, "learning_rate": 0.0005, "loss": 1.3332208395004272, "step": 26125 }, { "epoch": 2.0595979503350414, "grad_norm": 0.47046250104904175, "learning_rate": 0.0005, "loss": 1.247234582901001, "step": 26126 }, { "epoch": 2.05967678360268, "grad_norm": 0.5423744916915894, "learning_rate": 0.0005, "loss": 1.3464919328689575, "step": 26127 }, { "epoch": 2.0597556168703193, "grad_norm": 0.451653391122818, "learning_rate": 0.0005, "loss": 1.2313518524169922, "step": 26128 }, { "epoch": 2.0598344501379584, "grad_norm": 0.4745149612426758, "learning_rate": 0.0005, "loss": 1.4283463954925537, "step": 26129 }, { "epoch": 2.059913283405597, "grad_norm": 0.6426177620887756, "learning_rate": 0.0005, "loss": 1.3910300731658936, "step": 26130 }, { "epoch": 2.0599921166732362, "grad_norm": 0.5362503528594971, "learning_rate": 0.0005, "loss": 1.3062779903411865, "step": 26131 }, { "epoch": 2.060070949940875, "grad_norm": 0.5045172572135925, "learning_rate": 0.0005, "loss": 1.3472027778625488, "step": 26132 }, { "epoch": 2.060149783208514, "grad_norm": 0.5059715509414673, "learning_rate": 0.0005, "loss": 1.359498381614685, "step": 26133 }, { "epoch": 2.0602286164761527, "grad_norm": 0.46807968616485596, "learning_rate": 0.0005, "loss": 1.2116832733154297, "step": 26134 }, { "epoch": 2.060307449743792, "grad_norm": 0.5301356315612793, "learning_rate": 0.0005, "loss": 1.4039640426635742, "step": 26135 }, { "epoch": 2.060386283011431, "grad_norm": 0.7080212235450745, "learning_rate": 0.0005, "loss": 1.4368679523468018, "step": 26136 }, { "epoch": 2.0604651162790697, "grad_norm": 0.5569285750389099, "learning_rate": 0.0005, "loss": 1.221907615661621, "step": 26137 }, { "epoch": 2.060543949546709, "grad_norm": 0.4491094648838043, "learning_rate": 0.0005, "loss": 1.4298678636550903, "step": 26138 }, { "epoch": 2.0606227828143475, "grad_norm": 0.5313845872879028, "learning_rate": 0.0005, "loss": 1.4003193378448486, "step": 26139 }, { "epoch": 2.0607016160819867, "grad_norm": 0.48234298825263977, "learning_rate": 0.0005, "loss": 1.365091323852539, "step": 26140 }, { "epoch": 2.0607804493496253, "grad_norm": 0.4502596855163574, "learning_rate": 0.0005, "loss": 1.1905769109725952, "step": 26141 }, { "epoch": 2.0608592826172645, "grad_norm": 0.8298578262329102, "learning_rate": 0.0005, "loss": 1.5398426055908203, "step": 26142 }, { "epoch": 2.0609381158849036, "grad_norm": 0.49538537859916687, "learning_rate": 0.0005, "loss": 1.470757246017456, "step": 26143 }, { "epoch": 2.0610169491525423, "grad_norm": 0.4570119380950928, "learning_rate": 0.0005, "loss": 1.3866679668426514, "step": 26144 }, { "epoch": 2.0610957824201814, "grad_norm": 0.4270229637622833, "learning_rate": 0.0005, "loss": 1.2782902717590332, "step": 26145 }, { "epoch": 2.06117461568782, "grad_norm": 0.4299197494983673, "learning_rate": 0.0005, "loss": 1.3285731077194214, "step": 26146 }, { "epoch": 2.0612534489554593, "grad_norm": 0.42375144362449646, "learning_rate": 0.0005, "loss": 1.2923424243927002, "step": 26147 }, { "epoch": 2.061332282223098, "grad_norm": 2.493250608444214, "learning_rate": 0.0005, "loss": 1.7473993301391602, "step": 26148 }, { "epoch": 2.061411115490737, "grad_norm": 0.5301604866981506, "learning_rate": 0.0005, "loss": 1.218468427658081, "step": 26149 }, { "epoch": 2.061489948758376, "grad_norm": 0.3794141411781311, "learning_rate": 0.0005, "loss": 1.19722580909729, "step": 26150 }, { "epoch": 2.061568782026015, "grad_norm": 0.45819520950317383, "learning_rate": 0.0005, "loss": 1.1609017848968506, "step": 26151 }, { "epoch": 2.061647615293654, "grad_norm": 0.40520623326301575, "learning_rate": 0.0005, "loss": 1.2123444080352783, "step": 26152 }, { "epoch": 2.0617264485612927, "grad_norm": 0.5624277591705322, "learning_rate": 0.0005, "loss": 1.4696695804595947, "step": 26153 }, { "epoch": 2.061805281828932, "grad_norm": 0.5211732983589172, "learning_rate": 0.0005, "loss": 1.427342414855957, "step": 26154 }, { "epoch": 2.0618841150965705, "grad_norm": 0.43879199028015137, "learning_rate": 0.0005, "loss": 1.3566312789916992, "step": 26155 }, { "epoch": 2.0619629483642097, "grad_norm": 0.41555193066596985, "learning_rate": 0.0005, "loss": 1.3233883380889893, "step": 26156 }, { "epoch": 2.062041781631849, "grad_norm": 0.537180483341217, "learning_rate": 0.0005, "loss": 1.4116483926773071, "step": 26157 }, { "epoch": 2.0621206148994875, "grad_norm": 0.4327099323272705, "learning_rate": 0.0005, "loss": 1.1129668951034546, "step": 26158 }, { "epoch": 2.0621994481671266, "grad_norm": 0.454500287771225, "learning_rate": 0.0005, "loss": 1.3327293395996094, "step": 26159 }, { "epoch": 2.0622782814347653, "grad_norm": 0.45552772283554077, "learning_rate": 0.0005, "loss": 1.1758997440338135, "step": 26160 }, { "epoch": 2.0623571147024045, "grad_norm": 0.4946140646934509, "learning_rate": 0.0005, "loss": 1.3738595247268677, "step": 26161 }, { "epoch": 2.062435947970043, "grad_norm": 0.4865277111530304, "learning_rate": 0.0005, "loss": 1.5538923740386963, "step": 26162 }, { "epoch": 2.0625147812376823, "grad_norm": 0.5108970999717712, "learning_rate": 0.0005, "loss": 1.3375895023345947, "step": 26163 }, { "epoch": 2.0625936145053214, "grad_norm": 0.5134490728378296, "learning_rate": 0.0005, "loss": 1.3889154195785522, "step": 26164 }, { "epoch": 2.06267244777296, "grad_norm": 0.46356117725372314, "learning_rate": 0.0005, "loss": 1.418088436126709, "step": 26165 }, { "epoch": 2.0627512810405992, "grad_norm": 0.45072707533836365, "learning_rate": 0.0005, "loss": 1.2962034940719604, "step": 26166 }, { "epoch": 2.062830114308238, "grad_norm": 0.5364370346069336, "learning_rate": 0.0005, "loss": 1.3975880146026611, "step": 26167 }, { "epoch": 2.062908947575877, "grad_norm": 0.4317660331726074, "learning_rate": 0.0005, "loss": 1.2425143718719482, "step": 26168 }, { "epoch": 2.0629877808435158, "grad_norm": 0.43725940585136414, "learning_rate": 0.0005, "loss": 1.3467236757278442, "step": 26169 }, { "epoch": 2.063066614111155, "grad_norm": 0.46387600898742676, "learning_rate": 0.0005, "loss": 1.3348348140716553, "step": 26170 }, { "epoch": 2.063145447378794, "grad_norm": 0.3986544907093048, "learning_rate": 0.0005, "loss": 1.052905797958374, "step": 26171 }, { "epoch": 2.0632242806464327, "grad_norm": 0.4140316843986511, "learning_rate": 0.0005, "loss": 1.3092883825302124, "step": 26172 }, { "epoch": 2.063303113914072, "grad_norm": 0.42044979333877563, "learning_rate": 0.0005, "loss": 1.339584469795227, "step": 26173 }, { "epoch": 2.0633819471817105, "grad_norm": 0.4491294324398041, "learning_rate": 0.0005, "loss": 1.5198967456817627, "step": 26174 }, { "epoch": 2.0634607804493497, "grad_norm": 0.48887190222740173, "learning_rate": 0.0005, "loss": 1.228736400604248, "step": 26175 }, { "epoch": 2.063539613716989, "grad_norm": 0.5733411908149719, "learning_rate": 0.0005, "loss": 1.2869467735290527, "step": 26176 }, { "epoch": 2.0636184469846275, "grad_norm": 0.41920405626296997, "learning_rate": 0.0005, "loss": 1.3338674306869507, "step": 26177 }, { "epoch": 2.0636972802522666, "grad_norm": 0.597021222114563, "learning_rate": 0.0005, "loss": 1.1854743957519531, "step": 26178 }, { "epoch": 2.0637761135199053, "grad_norm": 0.6811203360557556, "learning_rate": 0.0005, "loss": 1.3439218997955322, "step": 26179 }, { "epoch": 2.0638549467875444, "grad_norm": 0.3846299350261688, "learning_rate": 0.0005, "loss": 1.2323882579803467, "step": 26180 }, { "epoch": 2.063933780055183, "grad_norm": 0.538248598575592, "learning_rate": 0.0005, "loss": 1.5235769748687744, "step": 26181 }, { "epoch": 2.0640126133228223, "grad_norm": 0.5472362637519836, "learning_rate": 0.0005, "loss": 1.334388256072998, "step": 26182 }, { "epoch": 2.064091446590461, "grad_norm": 0.47440141439437866, "learning_rate": 0.0005, "loss": 1.3539377450942993, "step": 26183 }, { "epoch": 2.0641702798581, "grad_norm": 0.4586237967014313, "learning_rate": 0.0005, "loss": 1.2454380989074707, "step": 26184 }, { "epoch": 2.064249113125739, "grad_norm": 0.5235264897346497, "learning_rate": 0.0005, "loss": 1.152263879776001, "step": 26185 }, { "epoch": 2.064327946393378, "grad_norm": 0.5057611465454102, "learning_rate": 0.0005, "loss": 1.2691045999526978, "step": 26186 }, { "epoch": 2.064406779661017, "grad_norm": 0.3589950203895569, "learning_rate": 0.0005, "loss": 1.0768084526062012, "step": 26187 }, { "epoch": 2.0644856129286557, "grad_norm": 0.5104169249534607, "learning_rate": 0.0005, "loss": 1.437357783317566, "step": 26188 }, { "epoch": 2.064564446196295, "grad_norm": 0.5190882682800293, "learning_rate": 0.0005, "loss": 1.1874256134033203, "step": 26189 }, { "epoch": 2.064643279463934, "grad_norm": 0.7031969428062439, "learning_rate": 0.0005, "loss": 1.6062849760055542, "step": 26190 }, { "epoch": 2.0647221127315727, "grad_norm": 0.48382800817489624, "learning_rate": 0.0005, "loss": 1.3279739618301392, "step": 26191 }, { "epoch": 2.064800945999212, "grad_norm": 0.4557611048221588, "learning_rate": 0.0005, "loss": 1.3339898586273193, "step": 26192 }, { "epoch": 2.0648797792668505, "grad_norm": 0.5550474524497986, "learning_rate": 0.0005, "loss": 1.4717440605163574, "step": 26193 }, { "epoch": 2.0649586125344896, "grad_norm": 0.6089634895324707, "learning_rate": 0.0005, "loss": 1.3441108465194702, "step": 26194 }, { "epoch": 2.0650374458021283, "grad_norm": 0.46007290482521057, "learning_rate": 0.0005, "loss": 1.358776569366455, "step": 26195 }, { "epoch": 2.0651162790697675, "grad_norm": 0.4793678820133209, "learning_rate": 0.0005, "loss": 1.392226219177246, "step": 26196 }, { "epoch": 2.065195112337406, "grad_norm": 0.446194052696228, "learning_rate": 0.0005, "loss": 1.2410287857055664, "step": 26197 }, { "epoch": 2.0652739456050453, "grad_norm": 0.5172684788703918, "learning_rate": 0.0005, "loss": 1.4661061763763428, "step": 26198 }, { "epoch": 2.0653527788726844, "grad_norm": 0.5599110126495361, "learning_rate": 0.0005, "loss": 1.490797758102417, "step": 26199 }, { "epoch": 2.065431612140323, "grad_norm": 0.6492161154747009, "learning_rate": 0.0005, "loss": 1.2999132871627808, "step": 26200 }, { "epoch": 2.0655104454079622, "grad_norm": 0.7709826231002808, "learning_rate": 0.0005, "loss": 1.4820376634597778, "step": 26201 }, { "epoch": 2.065589278675601, "grad_norm": 0.38739079236984253, "learning_rate": 0.0005, "loss": 1.2318708896636963, "step": 26202 }, { "epoch": 2.06566811194324, "grad_norm": 0.4718497097492218, "learning_rate": 0.0005, "loss": 1.2549676895141602, "step": 26203 }, { "epoch": 2.065746945210879, "grad_norm": 0.5476653575897217, "learning_rate": 0.0005, "loss": 1.3518742322921753, "step": 26204 }, { "epoch": 2.065825778478518, "grad_norm": 0.4445473253726959, "learning_rate": 0.0005, "loss": 1.139662742614746, "step": 26205 }, { "epoch": 2.065904611746157, "grad_norm": 0.8053118586540222, "learning_rate": 0.0005, "loss": 1.3136637210845947, "step": 26206 }, { "epoch": 2.0659834450137957, "grad_norm": 0.4608003795146942, "learning_rate": 0.0005, "loss": 1.3297617435455322, "step": 26207 }, { "epoch": 2.066062278281435, "grad_norm": 0.40733107924461365, "learning_rate": 0.0005, "loss": 1.240935206413269, "step": 26208 }, { "epoch": 2.0661411115490735, "grad_norm": 0.5162707567214966, "learning_rate": 0.0005, "loss": 1.3585593700408936, "step": 26209 }, { "epoch": 2.0662199448167127, "grad_norm": 0.4832861125469208, "learning_rate": 0.0005, "loss": 1.4153797626495361, "step": 26210 }, { "epoch": 2.066298778084352, "grad_norm": 0.575348973274231, "learning_rate": 0.0005, "loss": 1.3248250484466553, "step": 26211 }, { "epoch": 2.0663776113519905, "grad_norm": 0.5661430954933167, "learning_rate": 0.0005, "loss": 1.4809119701385498, "step": 26212 }, { "epoch": 2.0664564446196296, "grad_norm": 0.4598996937274933, "learning_rate": 0.0005, "loss": 1.2580559253692627, "step": 26213 }, { "epoch": 2.0665352778872683, "grad_norm": 0.39104315638542175, "learning_rate": 0.0005, "loss": 1.310994029045105, "step": 26214 }, { "epoch": 2.0666141111549075, "grad_norm": 0.5356432199478149, "learning_rate": 0.0005, "loss": 1.6165417432785034, "step": 26215 }, { "epoch": 2.066692944422546, "grad_norm": 0.6289679408073425, "learning_rate": 0.0005, "loss": 1.409163475036621, "step": 26216 }, { "epoch": 2.0667717776901853, "grad_norm": 0.5514901280403137, "learning_rate": 0.0005, "loss": 1.4876179695129395, "step": 26217 }, { "epoch": 2.0668506109578244, "grad_norm": 0.7932664155960083, "learning_rate": 0.0005, "loss": 1.256666660308838, "step": 26218 }, { "epoch": 2.066929444225463, "grad_norm": 0.5000296235084534, "learning_rate": 0.0005, "loss": 1.3619263172149658, "step": 26219 }, { "epoch": 2.0670082774931022, "grad_norm": 0.4312294125556946, "learning_rate": 0.0005, "loss": 1.226897120475769, "step": 26220 }, { "epoch": 2.067087110760741, "grad_norm": 0.44897520542144775, "learning_rate": 0.0005, "loss": 1.2252099514007568, "step": 26221 }, { "epoch": 2.06716594402838, "grad_norm": 0.4880331754684448, "learning_rate": 0.0005, "loss": 1.4097113609313965, "step": 26222 }, { "epoch": 2.0672447772960187, "grad_norm": 0.5877812504768372, "learning_rate": 0.0005, "loss": 1.508880376815796, "step": 26223 }, { "epoch": 2.067323610563658, "grad_norm": 0.5249083042144775, "learning_rate": 0.0005, "loss": 1.3613948822021484, "step": 26224 }, { "epoch": 2.067402443831297, "grad_norm": 0.5553280115127563, "learning_rate": 0.0005, "loss": 1.1816471815109253, "step": 26225 }, { "epoch": 2.0674812770989357, "grad_norm": 0.5129079222679138, "learning_rate": 0.0005, "loss": 1.251250982284546, "step": 26226 }, { "epoch": 2.067560110366575, "grad_norm": 0.4699951112270355, "learning_rate": 0.0005, "loss": 1.368505597114563, "step": 26227 }, { "epoch": 2.0676389436342135, "grad_norm": 0.7096647024154663, "learning_rate": 0.0005, "loss": 1.4347493648529053, "step": 26228 }, { "epoch": 2.0677177769018527, "grad_norm": 0.4715301990509033, "learning_rate": 0.0005, "loss": 1.2407677173614502, "step": 26229 }, { "epoch": 2.0677966101694913, "grad_norm": 0.6202036738395691, "learning_rate": 0.0005, "loss": 1.3831779956817627, "step": 26230 }, { "epoch": 2.0678754434371305, "grad_norm": 0.38797664642333984, "learning_rate": 0.0005, "loss": 1.1708711385726929, "step": 26231 }, { "epoch": 2.0679542767047696, "grad_norm": 0.5696514844894409, "learning_rate": 0.0005, "loss": 1.383823037147522, "step": 26232 }, { "epoch": 2.0680331099724083, "grad_norm": 0.6499230265617371, "learning_rate": 0.0005, "loss": 1.4600844383239746, "step": 26233 }, { "epoch": 2.0681119432400474, "grad_norm": 0.516302227973938, "learning_rate": 0.0005, "loss": 1.0191045999526978, "step": 26234 }, { "epoch": 2.068190776507686, "grad_norm": 0.6446506977081299, "learning_rate": 0.0005, "loss": 1.2642326354980469, "step": 26235 }, { "epoch": 2.0682696097753253, "grad_norm": 0.5676352977752686, "learning_rate": 0.0005, "loss": 1.470373272895813, "step": 26236 }, { "epoch": 2.068348443042964, "grad_norm": 0.5237923860549927, "learning_rate": 0.0005, "loss": 1.2775076627731323, "step": 26237 }, { "epoch": 2.068427276310603, "grad_norm": 0.4682997763156891, "learning_rate": 0.0005, "loss": 1.2994096279144287, "step": 26238 }, { "epoch": 2.068506109578242, "grad_norm": 0.4586489498615265, "learning_rate": 0.0005, "loss": 1.4046237468719482, "step": 26239 }, { "epoch": 2.068584942845881, "grad_norm": 0.5559287071228027, "learning_rate": 0.0005, "loss": 1.6093745231628418, "step": 26240 }, { "epoch": 2.06866377611352, "grad_norm": 0.5095992088317871, "learning_rate": 0.0005, "loss": 1.3288787603378296, "step": 26241 }, { "epoch": 2.0687426093811587, "grad_norm": 0.42358726263046265, "learning_rate": 0.0005, "loss": 1.238767147064209, "step": 26242 }, { "epoch": 2.068821442648798, "grad_norm": 0.4248358905315399, "learning_rate": 0.0005, "loss": 1.3808822631835938, "step": 26243 }, { "epoch": 2.0689002759164365, "grad_norm": 0.45812833309173584, "learning_rate": 0.0005, "loss": 1.4686801433563232, "step": 26244 }, { "epoch": 2.0689791091840757, "grad_norm": 0.574765682220459, "learning_rate": 0.0005, "loss": 1.33446204662323, "step": 26245 }, { "epoch": 2.069057942451715, "grad_norm": 0.4834950864315033, "learning_rate": 0.0005, "loss": 1.2896814346313477, "step": 26246 }, { "epoch": 2.0691367757193535, "grad_norm": 0.5853464603424072, "learning_rate": 0.0005, "loss": 1.2950642108917236, "step": 26247 }, { "epoch": 2.0692156089869926, "grad_norm": 0.49528828263282776, "learning_rate": 0.0005, "loss": 1.1902562379837036, "step": 26248 }, { "epoch": 2.0692944422546313, "grad_norm": 0.45268678665161133, "learning_rate": 0.0005, "loss": 1.2491109371185303, "step": 26249 }, { "epoch": 2.0693732755222705, "grad_norm": 0.43458685278892517, "learning_rate": 0.0005, "loss": 1.1315736770629883, "step": 26250 }, { "epoch": 2.069452108789909, "grad_norm": 0.46882110834121704, "learning_rate": 0.0005, "loss": 1.4226231575012207, "step": 26251 }, { "epoch": 2.0695309420575483, "grad_norm": 0.4203227460384369, "learning_rate": 0.0005, "loss": 1.1779518127441406, "step": 26252 }, { "epoch": 2.0696097753251874, "grad_norm": 0.4690994322299957, "learning_rate": 0.0005, "loss": 1.255221962928772, "step": 26253 }, { "epoch": 2.069688608592826, "grad_norm": 0.5845348834991455, "learning_rate": 0.0005, "loss": 1.372707724571228, "step": 26254 }, { "epoch": 2.0697674418604652, "grad_norm": 0.5714594125747681, "learning_rate": 0.0005, "loss": 1.1761763095855713, "step": 26255 }, { "epoch": 2.069846275128104, "grad_norm": 0.5300986766815186, "learning_rate": 0.0005, "loss": 1.5376759767532349, "step": 26256 }, { "epoch": 2.069925108395743, "grad_norm": 0.49966058135032654, "learning_rate": 0.0005, "loss": 1.2941640615463257, "step": 26257 }, { "epoch": 2.0700039416633818, "grad_norm": 0.4937705099582672, "learning_rate": 0.0005, "loss": 1.4441876411437988, "step": 26258 }, { "epoch": 2.070082774931021, "grad_norm": 0.5416471362113953, "learning_rate": 0.0005, "loss": 1.5326204299926758, "step": 26259 }, { "epoch": 2.07016160819866, "grad_norm": 0.45961934328079224, "learning_rate": 0.0005, "loss": 1.6483099460601807, "step": 26260 }, { "epoch": 2.0702404414662987, "grad_norm": 0.4371648132801056, "learning_rate": 0.0005, "loss": 1.32437002658844, "step": 26261 }, { "epoch": 2.070319274733938, "grad_norm": 0.5007100701332092, "learning_rate": 0.0005, "loss": 1.3273701667785645, "step": 26262 }, { "epoch": 2.0703981080015765, "grad_norm": 0.4950787127017975, "learning_rate": 0.0005, "loss": 1.3175770044326782, "step": 26263 }, { "epoch": 2.0704769412692157, "grad_norm": 0.48894205689430237, "learning_rate": 0.0005, "loss": 1.410872220993042, "step": 26264 }, { "epoch": 2.0705557745368544, "grad_norm": 0.5363978147506714, "learning_rate": 0.0005, "loss": 1.4259998798370361, "step": 26265 }, { "epoch": 2.0706346078044935, "grad_norm": 0.44246721267700195, "learning_rate": 0.0005, "loss": 1.235090970993042, "step": 26266 }, { "epoch": 2.0707134410721326, "grad_norm": 0.6122936606407166, "learning_rate": 0.0005, "loss": 1.5301300287246704, "step": 26267 }, { "epoch": 2.0707922743397713, "grad_norm": 0.48861557245254517, "learning_rate": 0.0005, "loss": 1.4697597026824951, "step": 26268 }, { "epoch": 2.0708711076074104, "grad_norm": 0.5628800988197327, "learning_rate": 0.0005, "loss": 1.566460371017456, "step": 26269 }, { "epoch": 2.070949940875049, "grad_norm": 0.5259683728218079, "learning_rate": 0.0005, "loss": 1.300368309020996, "step": 26270 }, { "epoch": 2.0710287741426883, "grad_norm": 0.43880313634872437, "learning_rate": 0.0005, "loss": 1.2369779348373413, "step": 26271 }, { "epoch": 2.071107607410327, "grad_norm": 0.44366759061813354, "learning_rate": 0.0005, "loss": 1.2046396732330322, "step": 26272 }, { "epoch": 2.071186440677966, "grad_norm": 0.5343159437179565, "learning_rate": 0.0005, "loss": 1.3910834789276123, "step": 26273 }, { "epoch": 2.0712652739456052, "grad_norm": 0.5416838526725769, "learning_rate": 0.0005, "loss": 1.3756046295166016, "step": 26274 }, { "epoch": 2.071344107213244, "grad_norm": 0.4452388286590576, "learning_rate": 0.0005, "loss": 1.206963062286377, "step": 26275 }, { "epoch": 2.071422940480883, "grad_norm": 0.475495845079422, "learning_rate": 0.0005, "loss": 1.315056562423706, "step": 26276 }, { "epoch": 2.0715017737485217, "grad_norm": 0.7518742084503174, "learning_rate": 0.0005, "loss": 1.2847566604614258, "step": 26277 }, { "epoch": 2.071580607016161, "grad_norm": 0.6563513278961182, "learning_rate": 0.0005, "loss": 1.3193659782409668, "step": 26278 }, { "epoch": 2.0716594402837996, "grad_norm": 0.5411086082458496, "learning_rate": 0.0005, "loss": 1.3769891262054443, "step": 26279 }, { "epoch": 2.0717382735514387, "grad_norm": 0.5720568895339966, "learning_rate": 0.0005, "loss": 1.500504493713379, "step": 26280 }, { "epoch": 2.071817106819078, "grad_norm": 0.6808109283447266, "learning_rate": 0.0005, "loss": 1.4487320184707642, "step": 26281 }, { "epoch": 2.0718959400867165, "grad_norm": 7.5164313316345215, "learning_rate": 0.0005, "loss": 1.3475608825683594, "step": 26282 }, { "epoch": 2.0719747733543556, "grad_norm": 0.4538402855396271, "learning_rate": 0.0005, "loss": 1.387838363647461, "step": 26283 }, { "epoch": 2.0720536066219943, "grad_norm": 0.5209347009658813, "learning_rate": 0.0005, "loss": 1.5310158729553223, "step": 26284 }, { "epoch": 2.0721324398896335, "grad_norm": 0.509210467338562, "learning_rate": 0.0005, "loss": 1.4463493824005127, "step": 26285 }, { "epoch": 2.072211273157272, "grad_norm": 0.5205390453338623, "learning_rate": 0.0005, "loss": 1.4669723510742188, "step": 26286 }, { "epoch": 2.0722901064249113, "grad_norm": 0.5430582761764526, "learning_rate": 0.0005, "loss": 1.319760799407959, "step": 26287 }, { "epoch": 2.0723689396925504, "grad_norm": 0.47845983505249023, "learning_rate": 0.0005, "loss": 1.2843036651611328, "step": 26288 }, { "epoch": 2.072447772960189, "grad_norm": 0.44209110736846924, "learning_rate": 0.0005, "loss": 1.3503377437591553, "step": 26289 }, { "epoch": 2.0725266062278282, "grad_norm": 0.5840727090835571, "learning_rate": 0.0005, "loss": 1.3474740982055664, "step": 26290 }, { "epoch": 2.072605439495467, "grad_norm": 0.5212722420692444, "learning_rate": 0.0005, "loss": 1.1809464693069458, "step": 26291 }, { "epoch": 2.072684272763106, "grad_norm": 0.566265881061554, "learning_rate": 0.0005, "loss": 1.2617467641830444, "step": 26292 }, { "epoch": 2.0727631060307448, "grad_norm": 0.5432077050209045, "learning_rate": 0.0005, "loss": 1.3982887268066406, "step": 26293 }, { "epoch": 2.072841939298384, "grad_norm": 0.5626224875450134, "learning_rate": 0.0005, "loss": 1.273923397064209, "step": 26294 }, { "epoch": 2.072920772566023, "grad_norm": 0.4884198009967804, "learning_rate": 0.0005, "loss": 1.3993537425994873, "step": 26295 }, { "epoch": 2.0729996058336617, "grad_norm": 0.4866470694541931, "learning_rate": 0.0005, "loss": 1.1817121505737305, "step": 26296 }, { "epoch": 2.073078439101301, "grad_norm": 0.522750735282898, "learning_rate": 0.0005, "loss": 1.3855867385864258, "step": 26297 }, { "epoch": 2.0731572723689395, "grad_norm": 0.5064902305603027, "learning_rate": 0.0005, "loss": 1.2606570720672607, "step": 26298 }, { "epoch": 2.0732361056365787, "grad_norm": 0.5677387714385986, "learning_rate": 0.0005, "loss": 1.293777346611023, "step": 26299 }, { "epoch": 2.0733149389042174, "grad_norm": 0.4216587543487549, "learning_rate": 0.0005, "loss": 1.461191177368164, "step": 26300 }, { "epoch": 2.0733937721718565, "grad_norm": 0.5638648271560669, "learning_rate": 0.0005, "loss": 1.3926582336425781, "step": 26301 }, { "epoch": 2.0734726054394956, "grad_norm": 0.5241127014160156, "learning_rate": 0.0005, "loss": 1.6101006269454956, "step": 26302 }, { "epoch": 2.0735514387071343, "grad_norm": 0.592376708984375, "learning_rate": 0.0005, "loss": 1.3839486837387085, "step": 26303 }, { "epoch": 2.0736302719747735, "grad_norm": 0.5269502997398376, "learning_rate": 0.0005, "loss": 1.3370327949523926, "step": 26304 }, { "epoch": 2.073709105242412, "grad_norm": 0.49900996685028076, "learning_rate": 0.0005, "loss": 1.368514895439148, "step": 26305 }, { "epoch": 2.0737879385100513, "grad_norm": 0.39259347319602966, "learning_rate": 0.0005, "loss": 1.2191102504730225, "step": 26306 }, { "epoch": 2.0738667717776904, "grad_norm": 0.4903949499130249, "learning_rate": 0.0005, "loss": 1.141759991645813, "step": 26307 }, { "epoch": 2.073945605045329, "grad_norm": 0.44607895612716675, "learning_rate": 0.0005, "loss": 1.3440828323364258, "step": 26308 }, { "epoch": 2.0740244383129682, "grad_norm": 0.5303677320480347, "learning_rate": 0.0005, "loss": 1.3527216911315918, "step": 26309 }, { "epoch": 2.074103271580607, "grad_norm": 0.5171297192573547, "learning_rate": 0.0005, "loss": 1.3716974258422852, "step": 26310 }, { "epoch": 2.074182104848246, "grad_norm": 0.47818389534950256, "learning_rate": 0.0005, "loss": 1.2328016757965088, "step": 26311 }, { "epoch": 2.0742609381158847, "grad_norm": 0.6120519042015076, "learning_rate": 0.0005, "loss": 1.4406001567840576, "step": 26312 }, { "epoch": 2.074339771383524, "grad_norm": 0.4993940591812134, "learning_rate": 0.0005, "loss": 1.3121061325073242, "step": 26313 }, { "epoch": 2.0744186046511626, "grad_norm": 0.5405462980270386, "learning_rate": 0.0005, "loss": 1.435687780380249, "step": 26314 }, { "epoch": 2.0744974379188017, "grad_norm": 0.5672962665557861, "learning_rate": 0.0005, "loss": 1.363737940788269, "step": 26315 }, { "epoch": 2.074576271186441, "grad_norm": 0.4729853570461273, "learning_rate": 0.0005, "loss": 1.3118741512298584, "step": 26316 }, { "epoch": 2.0746551044540795, "grad_norm": 0.7419790029525757, "learning_rate": 0.0005, "loss": 1.3744926452636719, "step": 26317 }, { "epoch": 2.0747339377217187, "grad_norm": 0.4789588451385498, "learning_rate": 0.0005, "loss": 1.2016335725784302, "step": 26318 }, { "epoch": 2.0748127709893573, "grad_norm": 0.7856791019439697, "learning_rate": 0.0005, "loss": 1.2368683815002441, "step": 26319 }, { "epoch": 2.0748916042569965, "grad_norm": 0.5493692755699158, "learning_rate": 0.0005, "loss": 1.526232361793518, "step": 26320 }, { "epoch": 2.0749704375246356, "grad_norm": 0.5763341784477234, "learning_rate": 0.0005, "loss": 1.2870373725891113, "step": 26321 }, { "epoch": 2.0750492707922743, "grad_norm": 0.4233342409133911, "learning_rate": 0.0005, "loss": 1.232743501663208, "step": 26322 }, { "epoch": 2.0751281040599134, "grad_norm": 0.394156813621521, "learning_rate": 0.0005, "loss": 1.243631362915039, "step": 26323 }, { "epoch": 2.075206937327552, "grad_norm": 0.4739505350589752, "learning_rate": 0.0005, "loss": 1.2796850204467773, "step": 26324 }, { "epoch": 2.0752857705951913, "grad_norm": 0.39814913272857666, "learning_rate": 0.0005, "loss": 1.208800196647644, "step": 26325 }, { "epoch": 2.07536460386283, "grad_norm": 0.5081117153167725, "learning_rate": 0.0005, "loss": 1.4676883220672607, "step": 26326 }, { "epoch": 2.075443437130469, "grad_norm": 0.47042641043663025, "learning_rate": 0.0005, "loss": 1.2494395971298218, "step": 26327 }, { "epoch": 2.075522270398108, "grad_norm": 0.5848693251609802, "learning_rate": 0.0005, "loss": 1.1830724477767944, "step": 26328 }, { "epoch": 2.075601103665747, "grad_norm": 0.557465672492981, "learning_rate": 0.0005, "loss": 1.3435535430908203, "step": 26329 }, { "epoch": 2.075679936933386, "grad_norm": 0.47777995467185974, "learning_rate": 0.0005, "loss": 1.4630283117294312, "step": 26330 }, { "epoch": 2.0757587702010247, "grad_norm": 0.5046231746673584, "learning_rate": 0.0005, "loss": 1.3812170028686523, "step": 26331 }, { "epoch": 2.075837603468664, "grad_norm": 0.5482017993927002, "learning_rate": 0.0005, "loss": 1.3299052715301514, "step": 26332 }, { "epoch": 2.0759164367363026, "grad_norm": 0.749986469745636, "learning_rate": 0.0005, "loss": 1.30045485496521, "step": 26333 }, { "epoch": 2.0759952700039417, "grad_norm": 0.47622713446617126, "learning_rate": 0.0005, "loss": 1.4174621105194092, "step": 26334 }, { "epoch": 2.076074103271581, "grad_norm": 0.4864453375339508, "learning_rate": 0.0005, "loss": 1.2728230953216553, "step": 26335 }, { "epoch": 2.0761529365392195, "grad_norm": 0.5165749788284302, "learning_rate": 0.0005, "loss": 1.4037179946899414, "step": 26336 }, { "epoch": 2.0762317698068586, "grad_norm": 0.4215461015701294, "learning_rate": 0.0005, "loss": 1.1194894313812256, "step": 26337 }, { "epoch": 2.0763106030744973, "grad_norm": 0.4647725820541382, "learning_rate": 0.0005, "loss": 1.3030898571014404, "step": 26338 }, { "epoch": 2.0763894363421365, "grad_norm": 0.5302726030349731, "learning_rate": 0.0005, "loss": 1.3409960269927979, "step": 26339 }, { "epoch": 2.076468269609775, "grad_norm": 0.6210141777992249, "learning_rate": 0.0005, "loss": 1.3581386804580688, "step": 26340 }, { "epoch": 2.0765471028774143, "grad_norm": 0.5378154516220093, "learning_rate": 0.0005, "loss": 1.2960357666015625, "step": 26341 }, { "epoch": 2.0766259361450534, "grad_norm": 0.50899738073349, "learning_rate": 0.0005, "loss": 1.3533183336257935, "step": 26342 }, { "epoch": 2.076704769412692, "grad_norm": 0.5279836654663086, "learning_rate": 0.0005, "loss": 1.4361200332641602, "step": 26343 }, { "epoch": 2.0767836026803312, "grad_norm": 0.4309505522251129, "learning_rate": 0.0005, "loss": 1.1758582592010498, "step": 26344 }, { "epoch": 2.07686243594797, "grad_norm": 0.617037296295166, "learning_rate": 0.0005, "loss": 1.3562649488449097, "step": 26345 }, { "epoch": 2.076941269215609, "grad_norm": 0.41204410791397095, "learning_rate": 0.0005, "loss": 1.1574023962020874, "step": 26346 }, { "epoch": 2.0770201024832478, "grad_norm": 0.4591894745826721, "learning_rate": 0.0005, "loss": 1.2667845487594604, "step": 26347 }, { "epoch": 2.077098935750887, "grad_norm": 0.509913444519043, "learning_rate": 0.0005, "loss": 1.4426606893539429, "step": 26348 }, { "epoch": 2.077177769018526, "grad_norm": 0.5012990832328796, "learning_rate": 0.0005, "loss": 1.494668960571289, "step": 26349 }, { "epoch": 2.0772566022861647, "grad_norm": 0.4097670316696167, "learning_rate": 0.0005, "loss": 1.3632630109786987, "step": 26350 }, { "epoch": 2.077335435553804, "grad_norm": 0.5079538822174072, "learning_rate": 0.0005, "loss": 1.4367578029632568, "step": 26351 }, { "epoch": 2.0774142688214425, "grad_norm": 0.4691762626171112, "learning_rate": 0.0005, "loss": 1.3812370300292969, "step": 26352 }, { "epoch": 2.0774931020890817, "grad_norm": 0.42674484848976135, "learning_rate": 0.0005, "loss": 1.2276737689971924, "step": 26353 }, { "epoch": 2.0775719353567204, "grad_norm": 0.5765203237533569, "learning_rate": 0.0005, "loss": 1.36905837059021, "step": 26354 }, { "epoch": 2.0776507686243595, "grad_norm": 0.5413273572921753, "learning_rate": 0.0005, "loss": 1.4204492568969727, "step": 26355 }, { "epoch": 2.0777296018919986, "grad_norm": 0.5857071280479431, "learning_rate": 0.0005, "loss": 1.4172186851501465, "step": 26356 }, { "epoch": 2.0778084351596373, "grad_norm": 0.4956175982952118, "learning_rate": 0.0005, "loss": 1.3437237739562988, "step": 26357 }, { "epoch": 2.0778872684272764, "grad_norm": 0.630507230758667, "learning_rate": 0.0005, "loss": 1.2432188987731934, "step": 26358 }, { "epoch": 2.077966101694915, "grad_norm": 0.5617859959602356, "learning_rate": 0.0005, "loss": 1.297194242477417, "step": 26359 }, { "epoch": 2.0780449349625543, "grad_norm": 0.49090075492858887, "learning_rate": 0.0005, "loss": 1.2860486507415771, "step": 26360 }, { "epoch": 2.078123768230193, "grad_norm": 0.4443911612033844, "learning_rate": 0.0005, "loss": 1.1853094100952148, "step": 26361 }, { "epoch": 2.078202601497832, "grad_norm": 0.44999703764915466, "learning_rate": 0.0005, "loss": 1.2545289993286133, "step": 26362 }, { "epoch": 2.0782814347654712, "grad_norm": 0.5022109746932983, "learning_rate": 0.0005, "loss": 1.3852331638336182, "step": 26363 }, { "epoch": 2.07836026803311, "grad_norm": 0.4973868727684021, "learning_rate": 0.0005, "loss": 1.3875139951705933, "step": 26364 }, { "epoch": 2.078439101300749, "grad_norm": 0.4493730962276459, "learning_rate": 0.0005, "loss": 1.3646084070205688, "step": 26365 }, { "epoch": 2.0785179345683877, "grad_norm": 0.5662777423858643, "learning_rate": 0.0005, "loss": 1.405762791633606, "step": 26366 }, { "epoch": 2.078596767836027, "grad_norm": 0.4571525752544403, "learning_rate": 0.0005, "loss": 1.1531068086624146, "step": 26367 }, { "epoch": 2.0786756011036656, "grad_norm": 0.5099523067474365, "learning_rate": 0.0005, "loss": 1.4068453311920166, "step": 26368 }, { "epoch": 2.0787544343713047, "grad_norm": 0.46535536646842957, "learning_rate": 0.0005, "loss": 1.3291774988174438, "step": 26369 }, { "epoch": 2.078833267638944, "grad_norm": 0.6135430932044983, "learning_rate": 0.0005, "loss": 1.5488498210906982, "step": 26370 }, { "epoch": 2.0789121009065825, "grad_norm": 0.6723419427871704, "learning_rate": 0.0005, "loss": 1.3381659984588623, "step": 26371 }, { "epoch": 2.0789909341742216, "grad_norm": 0.5481259226799011, "learning_rate": 0.0005, "loss": 1.4807381629943848, "step": 26372 }, { "epoch": 2.0790697674418603, "grad_norm": 0.4756018817424774, "learning_rate": 0.0005, "loss": 1.470879077911377, "step": 26373 }, { "epoch": 2.0791486007094995, "grad_norm": 0.47188568115234375, "learning_rate": 0.0005, "loss": 1.2217137813568115, "step": 26374 }, { "epoch": 2.079227433977138, "grad_norm": 1.1201797723770142, "learning_rate": 0.0005, "loss": 1.3479857444763184, "step": 26375 }, { "epoch": 2.0793062672447773, "grad_norm": 0.4782116711139679, "learning_rate": 0.0005, "loss": 1.2696747779846191, "step": 26376 }, { "epoch": 2.0793851005124164, "grad_norm": 0.4170061945915222, "learning_rate": 0.0005, "loss": 1.2860100269317627, "step": 26377 }, { "epoch": 2.079463933780055, "grad_norm": 0.4296755790710449, "learning_rate": 0.0005, "loss": 1.202928066253662, "step": 26378 }, { "epoch": 2.0795427670476943, "grad_norm": 0.5167216658592224, "learning_rate": 0.0005, "loss": 1.4431257247924805, "step": 26379 }, { "epoch": 2.079621600315333, "grad_norm": 0.547199547290802, "learning_rate": 0.0005, "loss": 1.5190935134887695, "step": 26380 }, { "epoch": 2.079700433582972, "grad_norm": 0.48808717727661133, "learning_rate": 0.0005, "loss": 1.4518224000930786, "step": 26381 }, { "epoch": 2.0797792668506108, "grad_norm": 0.5243397951126099, "learning_rate": 0.0005, "loss": 1.5275639295578003, "step": 26382 }, { "epoch": 2.07985810011825, "grad_norm": 0.41078871488571167, "learning_rate": 0.0005, "loss": 1.3328759670257568, "step": 26383 }, { "epoch": 2.079936933385889, "grad_norm": 0.5241022109985352, "learning_rate": 0.0005, "loss": 1.3913002014160156, "step": 26384 }, { "epoch": 2.0800157666535277, "grad_norm": 0.7244629263877869, "learning_rate": 0.0005, "loss": 1.2898736000061035, "step": 26385 }, { "epoch": 2.080094599921167, "grad_norm": 0.5100358128547668, "learning_rate": 0.0005, "loss": 1.396945595741272, "step": 26386 }, { "epoch": 2.0801734331888055, "grad_norm": 0.47384896874427795, "learning_rate": 0.0005, "loss": 1.1636269092559814, "step": 26387 }, { "epoch": 2.0802522664564447, "grad_norm": 0.4137584865093231, "learning_rate": 0.0005, "loss": 1.3582520484924316, "step": 26388 }, { "epoch": 2.0803310997240834, "grad_norm": 0.41413140296936035, "learning_rate": 0.0005, "loss": 1.4142422676086426, "step": 26389 }, { "epoch": 2.0804099329917225, "grad_norm": 0.47414496541023254, "learning_rate": 0.0005, "loss": 1.3211395740509033, "step": 26390 }, { "epoch": 2.0804887662593616, "grad_norm": 0.6072850227355957, "learning_rate": 0.0005, "loss": 1.463818073272705, "step": 26391 }, { "epoch": 2.0805675995270003, "grad_norm": 0.49882593750953674, "learning_rate": 0.0005, "loss": 1.2331304550170898, "step": 26392 }, { "epoch": 2.0806464327946395, "grad_norm": 0.5965675711631775, "learning_rate": 0.0005, "loss": 1.2494102716445923, "step": 26393 }, { "epoch": 2.080725266062278, "grad_norm": 0.4446408748626709, "learning_rate": 0.0005, "loss": 1.320008397102356, "step": 26394 }, { "epoch": 2.0808040993299173, "grad_norm": 0.4724127948284149, "learning_rate": 0.0005, "loss": 1.3345727920532227, "step": 26395 }, { "epoch": 2.080882932597556, "grad_norm": 0.5044773817062378, "learning_rate": 0.0005, "loss": 1.4764564037322998, "step": 26396 }, { "epoch": 2.080961765865195, "grad_norm": 0.41431963443756104, "learning_rate": 0.0005, "loss": 1.308261513710022, "step": 26397 }, { "epoch": 2.0810405991328342, "grad_norm": 0.48109790682792664, "learning_rate": 0.0005, "loss": 1.4167463779449463, "step": 26398 }, { "epoch": 2.081119432400473, "grad_norm": 0.539066731929779, "learning_rate": 0.0005, "loss": 1.3568732738494873, "step": 26399 }, { "epoch": 2.081198265668112, "grad_norm": 0.4382838010787964, "learning_rate": 0.0005, "loss": 1.319368600845337, "step": 26400 }, { "epoch": 2.0812770989357507, "grad_norm": 0.4106091558933258, "learning_rate": 0.0005, "loss": 1.2076596021652222, "step": 26401 }, { "epoch": 2.08135593220339, "grad_norm": 0.46504178643226624, "learning_rate": 0.0005, "loss": 1.1517693996429443, "step": 26402 }, { "epoch": 2.0814347654710286, "grad_norm": 0.44512084126472473, "learning_rate": 0.0005, "loss": 1.1694351434707642, "step": 26403 }, { "epoch": 2.0815135987386677, "grad_norm": 0.4378456175327301, "learning_rate": 0.0005, "loss": 1.2862462997436523, "step": 26404 }, { "epoch": 2.081592432006307, "grad_norm": 0.4992227852344513, "learning_rate": 0.0005, "loss": 1.426210641860962, "step": 26405 }, { "epoch": 2.0816712652739455, "grad_norm": 0.5274015665054321, "learning_rate": 0.0005, "loss": 1.3115111589431763, "step": 26406 }, { "epoch": 2.0817500985415847, "grad_norm": 0.4705880284309387, "learning_rate": 0.0005, "loss": 1.542388677597046, "step": 26407 }, { "epoch": 2.0818289318092233, "grad_norm": 0.5516986846923828, "learning_rate": 0.0005, "loss": 1.2539761066436768, "step": 26408 }, { "epoch": 2.0819077650768625, "grad_norm": 0.49736589193344116, "learning_rate": 0.0005, "loss": 1.4504687786102295, "step": 26409 }, { "epoch": 2.081986598344501, "grad_norm": 0.48722895979881287, "learning_rate": 0.0005, "loss": 1.2253084182739258, "step": 26410 }, { "epoch": 2.0820654316121403, "grad_norm": 0.5906611680984497, "learning_rate": 0.0005, "loss": 1.4432101249694824, "step": 26411 }, { "epoch": 2.0821442648797794, "grad_norm": 0.49909085035324097, "learning_rate": 0.0005, "loss": 1.1886379718780518, "step": 26412 }, { "epoch": 2.082223098147418, "grad_norm": 0.5057275295257568, "learning_rate": 0.0005, "loss": 1.4233646392822266, "step": 26413 }, { "epoch": 2.0823019314150573, "grad_norm": 0.5841972231864929, "learning_rate": 0.0005, "loss": 1.4822614192962646, "step": 26414 }, { "epoch": 2.082380764682696, "grad_norm": 0.6065002679824829, "learning_rate": 0.0005, "loss": 1.3011119365692139, "step": 26415 }, { "epoch": 2.082459597950335, "grad_norm": 1.3094778060913086, "learning_rate": 0.0005, "loss": 1.3801159858703613, "step": 26416 }, { "epoch": 2.0825384312179738, "grad_norm": 0.5331070423126221, "learning_rate": 0.0005, "loss": 1.5111134052276611, "step": 26417 }, { "epoch": 2.082617264485613, "grad_norm": 0.5379615426063538, "learning_rate": 0.0005, "loss": 1.3627071380615234, "step": 26418 }, { "epoch": 2.082696097753252, "grad_norm": 0.6474922895431519, "learning_rate": 0.0005, "loss": 1.6545339822769165, "step": 26419 }, { "epoch": 2.0827749310208907, "grad_norm": 0.4936586618423462, "learning_rate": 0.0005, "loss": 1.3927216529846191, "step": 26420 }, { "epoch": 2.08285376428853, "grad_norm": 0.43466508388519287, "learning_rate": 0.0005, "loss": 1.253913402557373, "step": 26421 }, { "epoch": 2.0829325975561686, "grad_norm": 0.4780969023704529, "learning_rate": 0.0005, "loss": 1.325810194015503, "step": 26422 }, { "epoch": 2.0830114308238077, "grad_norm": 0.5364126563072205, "learning_rate": 0.0005, "loss": 1.2726585865020752, "step": 26423 }, { "epoch": 2.083090264091447, "grad_norm": 0.4838408827781677, "learning_rate": 0.0005, "loss": 1.1445860862731934, "step": 26424 }, { "epoch": 2.0831690973590855, "grad_norm": 0.4977594316005707, "learning_rate": 0.0005, "loss": 1.2441182136535645, "step": 26425 }, { "epoch": 2.0832479306267246, "grad_norm": 0.5357880592346191, "learning_rate": 0.0005, "loss": 1.2165539264678955, "step": 26426 }, { "epoch": 2.0833267638943633, "grad_norm": 0.5480741858482361, "learning_rate": 0.0005, "loss": 1.4226768016815186, "step": 26427 }, { "epoch": 2.0834055971620025, "grad_norm": 0.52931809425354, "learning_rate": 0.0005, "loss": 1.3531923294067383, "step": 26428 }, { "epoch": 2.083484430429641, "grad_norm": 0.575537383556366, "learning_rate": 0.0005, "loss": 1.4929628372192383, "step": 26429 }, { "epoch": 2.0835632636972803, "grad_norm": 0.5417758226394653, "learning_rate": 0.0005, "loss": 1.207369327545166, "step": 26430 }, { "epoch": 2.083642096964919, "grad_norm": 0.4912359118461609, "learning_rate": 0.0005, "loss": 1.397172212600708, "step": 26431 }, { "epoch": 2.083720930232558, "grad_norm": 0.45645377039909363, "learning_rate": 0.0005, "loss": 1.205959439277649, "step": 26432 }, { "epoch": 2.0837997635001972, "grad_norm": 0.543433427810669, "learning_rate": 0.0005, "loss": 1.334630012512207, "step": 26433 }, { "epoch": 2.083878596767836, "grad_norm": 0.4909295439720154, "learning_rate": 0.0005, "loss": 1.4505095481872559, "step": 26434 }, { "epoch": 2.083957430035475, "grad_norm": 0.5358245968818665, "learning_rate": 0.0005, "loss": 1.262691855430603, "step": 26435 }, { "epoch": 2.0840362633031138, "grad_norm": 0.6155993342399597, "learning_rate": 0.0005, "loss": 1.5801007747650146, "step": 26436 }, { "epoch": 2.084115096570753, "grad_norm": 0.4013620615005493, "learning_rate": 0.0005, "loss": 1.192949891090393, "step": 26437 }, { "epoch": 2.084193929838392, "grad_norm": 0.4663076400756836, "learning_rate": 0.0005, "loss": 1.3194866180419922, "step": 26438 }, { "epoch": 2.0842727631060307, "grad_norm": 0.47972622513771057, "learning_rate": 0.0005, "loss": 1.3422226905822754, "step": 26439 }, { "epoch": 2.08435159637367, "grad_norm": 0.6583388447761536, "learning_rate": 0.0005, "loss": 1.4608657360076904, "step": 26440 }, { "epoch": 2.0844304296413085, "grad_norm": 0.4615953862667084, "learning_rate": 0.0005, "loss": 1.204973578453064, "step": 26441 }, { "epoch": 2.0845092629089477, "grad_norm": 0.435368150472641, "learning_rate": 0.0005, "loss": 1.3135380744934082, "step": 26442 }, { "epoch": 2.0845880961765864, "grad_norm": 2.2219762802124023, "learning_rate": 0.0005, "loss": 1.5753659009933472, "step": 26443 }, { "epoch": 2.0846669294442255, "grad_norm": 0.6050194501876831, "learning_rate": 0.0005, "loss": 1.526914119720459, "step": 26444 }, { "epoch": 2.084745762711864, "grad_norm": 0.5792437791824341, "learning_rate": 0.0005, "loss": 1.3600690364837646, "step": 26445 }, { "epoch": 2.0848245959795033, "grad_norm": 0.48773714900016785, "learning_rate": 0.0005, "loss": 1.3499054908752441, "step": 26446 }, { "epoch": 2.0849034292471424, "grad_norm": 0.4455837607383728, "learning_rate": 0.0005, "loss": 1.2810001373291016, "step": 26447 }, { "epoch": 2.084982262514781, "grad_norm": 0.5640823841094971, "learning_rate": 0.0005, "loss": 1.3729113340377808, "step": 26448 }, { "epoch": 2.0850610957824203, "grad_norm": 0.46651923656463623, "learning_rate": 0.0005, "loss": 1.201513648033142, "step": 26449 }, { "epoch": 2.085139929050059, "grad_norm": 0.5661782622337341, "learning_rate": 0.0005, "loss": 1.1673641204833984, "step": 26450 }, { "epoch": 2.085218762317698, "grad_norm": 0.42615920305252075, "learning_rate": 0.0005, "loss": 1.2851736545562744, "step": 26451 }, { "epoch": 2.0852975955853372, "grad_norm": 0.4695848524570465, "learning_rate": 0.0005, "loss": 1.2359604835510254, "step": 26452 }, { "epoch": 2.085376428852976, "grad_norm": 0.5817915797233582, "learning_rate": 0.0005, "loss": 1.342572569847107, "step": 26453 }, { "epoch": 2.085455262120615, "grad_norm": 0.5821157097816467, "learning_rate": 0.0005, "loss": 1.4132726192474365, "step": 26454 }, { "epoch": 2.0855340953882537, "grad_norm": 0.5816364884376526, "learning_rate": 0.0005, "loss": 1.480480670928955, "step": 26455 }, { "epoch": 2.085612928655893, "grad_norm": 0.47160014510154724, "learning_rate": 0.0005, "loss": 1.338395357131958, "step": 26456 }, { "epoch": 2.0856917619235316, "grad_norm": 1.5050791501998901, "learning_rate": 0.0005, "loss": 1.554542064666748, "step": 26457 }, { "epoch": 2.0857705951911707, "grad_norm": 0.4564628601074219, "learning_rate": 0.0005, "loss": 1.2694838047027588, "step": 26458 }, { "epoch": 2.08584942845881, "grad_norm": 0.48986169695854187, "learning_rate": 0.0005, "loss": 1.3173466920852661, "step": 26459 }, { "epoch": 2.0859282617264485, "grad_norm": 0.5512104034423828, "learning_rate": 0.0005, "loss": 1.4146862030029297, "step": 26460 }, { "epoch": 2.0860070949940877, "grad_norm": 0.4919430911540985, "learning_rate": 0.0005, "loss": 1.3602917194366455, "step": 26461 }, { "epoch": 2.0860859282617263, "grad_norm": 0.5030795931816101, "learning_rate": 0.0005, "loss": 1.3222928047180176, "step": 26462 }, { "epoch": 2.0861647615293655, "grad_norm": 0.4744514524936676, "learning_rate": 0.0005, "loss": 1.2755389213562012, "step": 26463 }, { "epoch": 2.086243594797004, "grad_norm": 0.5518189072608948, "learning_rate": 0.0005, "loss": 1.2889583110809326, "step": 26464 }, { "epoch": 2.0863224280646433, "grad_norm": 0.4709981679916382, "learning_rate": 0.0005, "loss": 1.293360710144043, "step": 26465 }, { "epoch": 2.0864012613322824, "grad_norm": 0.4472816586494446, "learning_rate": 0.0005, "loss": 1.1948895454406738, "step": 26466 }, { "epoch": 2.086480094599921, "grad_norm": 0.5539504289627075, "learning_rate": 0.0005, "loss": 1.5025955438613892, "step": 26467 }, { "epoch": 2.0865589278675603, "grad_norm": 0.5609390139579773, "learning_rate": 0.0005, "loss": 1.3721375465393066, "step": 26468 }, { "epoch": 2.086637761135199, "grad_norm": 0.44404858350753784, "learning_rate": 0.0005, "loss": 1.3667070865631104, "step": 26469 }, { "epoch": 2.086716594402838, "grad_norm": 0.6090096235275269, "learning_rate": 0.0005, "loss": 1.55195951461792, "step": 26470 }, { "epoch": 2.0867954276704768, "grad_norm": 0.3982713222503662, "learning_rate": 0.0005, "loss": 1.29142427444458, "step": 26471 }, { "epoch": 2.086874260938116, "grad_norm": 0.47274136543273926, "learning_rate": 0.0005, "loss": 1.3815323114395142, "step": 26472 }, { "epoch": 2.086953094205755, "grad_norm": 0.5143447518348694, "learning_rate": 0.0005, "loss": 1.1679043769836426, "step": 26473 }, { "epoch": 2.0870319274733937, "grad_norm": 0.578531801700592, "learning_rate": 0.0005, "loss": 1.3810962438583374, "step": 26474 }, { "epoch": 2.087110760741033, "grad_norm": 0.4734732508659363, "learning_rate": 0.0005, "loss": 1.2249943017959595, "step": 26475 }, { "epoch": 2.0871895940086715, "grad_norm": 0.44278571009635925, "learning_rate": 0.0005, "loss": 1.3787026405334473, "step": 26476 }, { "epoch": 2.0872684272763107, "grad_norm": 0.5010221004486084, "learning_rate": 0.0005, "loss": 1.534867763519287, "step": 26477 }, { "epoch": 2.0873472605439494, "grad_norm": 0.44251349568367004, "learning_rate": 0.0005, "loss": 1.275276780128479, "step": 26478 }, { "epoch": 2.0874260938115885, "grad_norm": 0.46499788761138916, "learning_rate": 0.0005, "loss": 1.2642228603363037, "step": 26479 }, { "epoch": 2.0875049270792276, "grad_norm": 0.6292316913604736, "learning_rate": 0.0005, "loss": 1.4245893955230713, "step": 26480 }, { "epoch": 2.0875837603468663, "grad_norm": 0.5108434557914734, "learning_rate": 0.0005, "loss": 1.416887640953064, "step": 26481 }, { "epoch": 2.0876625936145055, "grad_norm": 0.4472116529941559, "learning_rate": 0.0005, "loss": 1.5502147674560547, "step": 26482 }, { "epoch": 2.087741426882144, "grad_norm": 0.538687527179718, "learning_rate": 0.0005, "loss": 1.2612791061401367, "step": 26483 }, { "epoch": 2.0878202601497833, "grad_norm": 0.41957157850265503, "learning_rate": 0.0005, "loss": 1.160696268081665, "step": 26484 }, { "epoch": 2.087899093417422, "grad_norm": 0.5035828948020935, "learning_rate": 0.0005, "loss": 1.3119361400604248, "step": 26485 }, { "epoch": 2.087977926685061, "grad_norm": 0.5504425764083862, "learning_rate": 0.0005, "loss": 1.361236572265625, "step": 26486 }, { "epoch": 2.0880567599527002, "grad_norm": 0.4038675129413605, "learning_rate": 0.0005, "loss": 1.1384234428405762, "step": 26487 }, { "epoch": 2.088135593220339, "grad_norm": 0.5634648203849792, "learning_rate": 0.0005, "loss": 1.529903531074524, "step": 26488 }, { "epoch": 2.088214426487978, "grad_norm": 0.4663954973220825, "learning_rate": 0.0005, "loss": 1.1162397861480713, "step": 26489 }, { "epoch": 2.0882932597556167, "grad_norm": 0.5696865320205688, "learning_rate": 0.0005, "loss": 1.4792357683181763, "step": 26490 }, { "epoch": 2.088372093023256, "grad_norm": 0.46810734272003174, "learning_rate": 0.0005, "loss": 1.4012787342071533, "step": 26491 }, { "epoch": 2.0884509262908946, "grad_norm": 0.48859062790870667, "learning_rate": 0.0005, "loss": 1.319450855255127, "step": 26492 }, { "epoch": 2.0885297595585337, "grad_norm": 0.5500312447547913, "learning_rate": 0.0005, "loss": 1.4801149368286133, "step": 26493 }, { "epoch": 2.088608592826173, "grad_norm": 0.5519118309020996, "learning_rate": 0.0005, "loss": 1.3249558210372925, "step": 26494 }, { "epoch": 2.0886874260938115, "grad_norm": 0.47162747383117676, "learning_rate": 0.0005, "loss": 1.3441853523254395, "step": 26495 }, { "epoch": 2.0887662593614507, "grad_norm": 0.48513102531433105, "learning_rate": 0.0005, "loss": 1.360318899154663, "step": 26496 }, { "epoch": 2.0888450926290894, "grad_norm": 0.48052528500556946, "learning_rate": 0.0005, "loss": 1.2106707096099854, "step": 26497 }, { "epoch": 2.0889239258967285, "grad_norm": 0.5268300175666809, "learning_rate": 0.0005, "loss": 1.375649333000183, "step": 26498 }, { "epoch": 2.089002759164367, "grad_norm": 0.4570111334323883, "learning_rate": 0.0005, "loss": 1.2836973667144775, "step": 26499 }, { "epoch": 2.0890815924320063, "grad_norm": 0.5218257308006287, "learning_rate": 0.0005, "loss": 1.324657678604126, "step": 26500 }, { "epoch": 2.0891604256996454, "grad_norm": 0.40047547221183777, "learning_rate": 0.0005, "loss": 1.2498302459716797, "step": 26501 }, { "epoch": 2.089239258967284, "grad_norm": 0.5122251510620117, "learning_rate": 0.0005, "loss": 1.303177833557129, "step": 26502 }, { "epoch": 2.0893180922349233, "grad_norm": 0.5476664900779724, "learning_rate": 0.0005, "loss": 1.3564361333847046, "step": 26503 }, { "epoch": 2.089396925502562, "grad_norm": 0.4802100658416748, "learning_rate": 0.0005, "loss": 1.4799106121063232, "step": 26504 }, { "epoch": 2.089475758770201, "grad_norm": 0.5322856903076172, "learning_rate": 0.0005, "loss": 1.3122060298919678, "step": 26505 }, { "epoch": 2.0895545920378398, "grad_norm": 0.5132149457931519, "learning_rate": 0.0005, "loss": 1.3056714534759521, "step": 26506 }, { "epoch": 2.089633425305479, "grad_norm": 0.5446670055389404, "learning_rate": 0.0005, "loss": 1.4299777746200562, "step": 26507 }, { "epoch": 2.089712258573118, "grad_norm": 0.5145368576049805, "learning_rate": 0.0005, "loss": 1.3428677320480347, "step": 26508 }, { "epoch": 2.0897910918407567, "grad_norm": 0.5574523210525513, "learning_rate": 0.0005, "loss": 1.3854358196258545, "step": 26509 }, { "epoch": 2.089869925108396, "grad_norm": 0.5761517286300659, "learning_rate": 0.0005, "loss": 1.2511029243469238, "step": 26510 }, { "epoch": 2.0899487583760346, "grad_norm": 0.3983974754810333, "learning_rate": 0.0005, "loss": 1.1544586420059204, "step": 26511 }, { "epoch": 2.0900275916436737, "grad_norm": 0.4172777831554413, "learning_rate": 0.0005, "loss": 1.2871909141540527, "step": 26512 }, { "epoch": 2.0901064249113124, "grad_norm": 0.5807667970657349, "learning_rate": 0.0005, "loss": 1.463935136795044, "step": 26513 }, { "epoch": 2.0901852581789515, "grad_norm": 0.5916624069213867, "learning_rate": 0.0005, "loss": 1.374410629272461, "step": 26514 }, { "epoch": 2.0902640914465906, "grad_norm": 0.48256897926330566, "learning_rate": 0.0005, "loss": 1.2018083333969116, "step": 26515 }, { "epoch": 2.0903429247142293, "grad_norm": 0.5482420325279236, "learning_rate": 0.0005, "loss": 1.4165596961975098, "step": 26516 }, { "epoch": 2.0904217579818685, "grad_norm": 0.5041403770446777, "learning_rate": 0.0005, "loss": 1.3348156213760376, "step": 26517 }, { "epoch": 2.090500591249507, "grad_norm": 0.4941957890987396, "learning_rate": 0.0005, "loss": 1.3811150789260864, "step": 26518 }, { "epoch": 2.0905794245171463, "grad_norm": 0.4654499888420105, "learning_rate": 0.0005, "loss": 1.300545334815979, "step": 26519 }, { "epoch": 2.090658257784785, "grad_norm": 0.43829113245010376, "learning_rate": 0.0005, "loss": 1.2805235385894775, "step": 26520 }, { "epoch": 2.090737091052424, "grad_norm": 0.5469900965690613, "learning_rate": 0.0005, "loss": 1.3865330219268799, "step": 26521 }, { "epoch": 2.0908159243200632, "grad_norm": 0.47850269079208374, "learning_rate": 0.0005, "loss": 1.2993323802947998, "step": 26522 }, { "epoch": 2.090894757587702, "grad_norm": 0.43453943729400635, "learning_rate": 0.0005, "loss": 1.3826274871826172, "step": 26523 }, { "epoch": 2.090973590855341, "grad_norm": 0.5829694867134094, "learning_rate": 0.0005, "loss": 1.327226161956787, "step": 26524 }, { "epoch": 2.0910524241229798, "grad_norm": 0.575933575630188, "learning_rate": 0.0005, "loss": 1.4485299587249756, "step": 26525 }, { "epoch": 2.091131257390619, "grad_norm": 0.45094332098960876, "learning_rate": 0.0005, "loss": 1.4175441265106201, "step": 26526 }, { "epoch": 2.0912100906582576, "grad_norm": 0.6321477890014648, "learning_rate": 0.0005, "loss": 1.330049753189087, "step": 26527 }, { "epoch": 2.0912889239258967, "grad_norm": 0.47167345881462097, "learning_rate": 0.0005, "loss": 1.2454900741577148, "step": 26528 }, { "epoch": 2.091367757193536, "grad_norm": 0.46740952134132385, "learning_rate": 0.0005, "loss": 1.425235629081726, "step": 26529 }, { "epoch": 2.0914465904611745, "grad_norm": 0.5413058400154114, "learning_rate": 0.0005, "loss": 1.4891811609268188, "step": 26530 }, { "epoch": 2.0915254237288137, "grad_norm": 0.49016115069389343, "learning_rate": 0.0005, "loss": 1.4338700771331787, "step": 26531 }, { "epoch": 2.0916042569964524, "grad_norm": 0.40601223707199097, "learning_rate": 0.0005, "loss": 1.204411268234253, "step": 26532 }, { "epoch": 2.0916830902640915, "grad_norm": 0.4204823672771454, "learning_rate": 0.0005, "loss": 1.1920106410980225, "step": 26533 }, { "epoch": 2.09176192353173, "grad_norm": 0.4803312420845032, "learning_rate": 0.0005, "loss": 1.2793973684310913, "step": 26534 }, { "epoch": 2.0918407567993693, "grad_norm": 0.47403889894485474, "learning_rate": 0.0005, "loss": 1.314465045928955, "step": 26535 }, { "epoch": 2.0919195900670084, "grad_norm": 0.44896695017814636, "learning_rate": 0.0005, "loss": 1.2753417491912842, "step": 26536 }, { "epoch": 2.091998423334647, "grad_norm": 0.5065422654151917, "learning_rate": 0.0005, "loss": 1.2132939100265503, "step": 26537 }, { "epoch": 2.0920772566022863, "grad_norm": 0.6260666847229004, "learning_rate": 0.0005, "loss": 1.3520829677581787, "step": 26538 }, { "epoch": 2.092156089869925, "grad_norm": 0.38654565811157227, "learning_rate": 0.0005, "loss": 1.2388241291046143, "step": 26539 }, { "epoch": 2.092234923137564, "grad_norm": 0.45916473865509033, "learning_rate": 0.0005, "loss": 1.1903237104415894, "step": 26540 }, { "epoch": 2.092313756405203, "grad_norm": 0.4479231834411621, "learning_rate": 0.0005, "loss": 1.2519999742507935, "step": 26541 }, { "epoch": 2.092392589672842, "grad_norm": 0.43171995878219604, "learning_rate": 0.0005, "loss": 1.370948314666748, "step": 26542 }, { "epoch": 2.092471422940481, "grad_norm": 0.47534531354904175, "learning_rate": 0.0005, "loss": 1.4416009187698364, "step": 26543 }, { "epoch": 2.0925502562081197, "grad_norm": 0.5156137943267822, "learning_rate": 0.0005, "loss": 1.4418295621871948, "step": 26544 }, { "epoch": 2.092629089475759, "grad_norm": 0.427735835313797, "learning_rate": 0.0005, "loss": 1.4148403406143188, "step": 26545 }, { "epoch": 2.0927079227433976, "grad_norm": 0.542026937007904, "learning_rate": 0.0005, "loss": 1.2656499147415161, "step": 26546 }, { "epoch": 2.0927867560110367, "grad_norm": 0.7000836133956909, "learning_rate": 0.0005, "loss": 1.6627026796340942, "step": 26547 }, { "epoch": 2.0928655892786754, "grad_norm": 0.5453939437866211, "learning_rate": 0.0005, "loss": 1.4827706813812256, "step": 26548 }, { "epoch": 2.0929444225463145, "grad_norm": 0.5165998935699463, "learning_rate": 0.0005, "loss": 1.5829956531524658, "step": 26549 }, { "epoch": 2.0930232558139537, "grad_norm": 0.472006231546402, "learning_rate": 0.0005, "loss": 1.2846653461456299, "step": 26550 }, { "epoch": 2.0931020890815923, "grad_norm": 0.5310214161872864, "learning_rate": 0.0005, "loss": 1.328263521194458, "step": 26551 }, { "epoch": 2.0931809223492315, "grad_norm": 0.4457871913909912, "learning_rate": 0.0005, "loss": 1.234137773513794, "step": 26552 }, { "epoch": 2.09325975561687, "grad_norm": 0.5284172892570496, "learning_rate": 0.0005, "loss": 1.2505861520767212, "step": 26553 }, { "epoch": 2.0933385888845093, "grad_norm": 0.5736480355262756, "learning_rate": 0.0005, "loss": 1.3041932582855225, "step": 26554 }, { "epoch": 2.0934174221521484, "grad_norm": 0.43800467252731323, "learning_rate": 0.0005, "loss": 1.2580747604370117, "step": 26555 }, { "epoch": 2.093496255419787, "grad_norm": 0.5527966022491455, "learning_rate": 0.0005, "loss": 1.3395719528198242, "step": 26556 }, { "epoch": 2.0935750886874263, "grad_norm": 0.5112220644950867, "learning_rate": 0.0005, "loss": 1.2711269855499268, "step": 26557 }, { "epoch": 2.093653921955065, "grad_norm": 0.5015066266059875, "learning_rate": 0.0005, "loss": 1.3406150341033936, "step": 26558 }, { "epoch": 2.093732755222704, "grad_norm": 0.4273538887500763, "learning_rate": 0.0005, "loss": 1.332496166229248, "step": 26559 }, { "epoch": 2.0938115884903428, "grad_norm": 0.5686925649642944, "learning_rate": 0.0005, "loss": 1.4534878730773926, "step": 26560 }, { "epoch": 2.093890421757982, "grad_norm": 0.5580359101295471, "learning_rate": 0.0005, "loss": 1.3708603382110596, "step": 26561 }, { "epoch": 2.0939692550256206, "grad_norm": 0.42044422030448914, "learning_rate": 0.0005, "loss": 1.348956823348999, "step": 26562 }, { "epoch": 2.0940480882932597, "grad_norm": 0.49537530541419983, "learning_rate": 0.0005, "loss": 1.3105931282043457, "step": 26563 }, { "epoch": 2.094126921560899, "grad_norm": 0.46530231833457947, "learning_rate": 0.0005, "loss": 1.2564821243286133, "step": 26564 }, { "epoch": 2.0942057548285375, "grad_norm": 0.5039637684822083, "learning_rate": 0.0005, "loss": 1.2449520826339722, "step": 26565 }, { "epoch": 2.0942845880961767, "grad_norm": 0.41392141580581665, "learning_rate": 0.0005, "loss": 1.3272719383239746, "step": 26566 }, { "epoch": 2.0943634213638154, "grad_norm": 0.5117407441139221, "learning_rate": 0.0005, "loss": 1.1916425228118896, "step": 26567 }, { "epoch": 2.0944422546314545, "grad_norm": 0.45357590913772583, "learning_rate": 0.0005, "loss": 1.3471202850341797, "step": 26568 }, { "epoch": 2.0945210878990936, "grad_norm": 0.49245139956474304, "learning_rate": 0.0005, "loss": 1.2969768047332764, "step": 26569 }, { "epoch": 2.0945999211667323, "grad_norm": 0.49801263213157654, "learning_rate": 0.0005, "loss": 1.1944952011108398, "step": 26570 }, { "epoch": 2.0946787544343715, "grad_norm": 0.4745931327342987, "learning_rate": 0.0005, "loss": 1.1495137214660645, "step": 26571 }, { "epoch": 2.09475758770201, "grad_norm": 0.42984235286712646, "learning_rate": 0.0005, "loss": 1.3693161010742188, "step": 26572 }, { "epoch": 2.0948364209696493, "grad_norm": 0.5027828812599182, "learning_rate": 0.0005, "loss": 1.233719825744629, "step": 26573 }, { "epoch": 2.094915254237288, "grad_norm": 0.47850361466407776, "learning_rate": 0.0005, "loss": 1.2767040729522705, "step": 26574 }, { "epoch": 2.094994087504927, "grad_norm": 0.3916518986225128, "learning_rate": 0.0005, "loss": 1.3770726919174194, "step": 26575 }, { "epoch": 2.095072920772566, "grad_norm": 0.4546310603618622, "learning_rate": 0.0005, "loss": 1.2231253385543823, "step": 26576 }, { "epoch": 2.095151754040205, "grad_norm": 0.4076124429702759, "learning_rate": 0.0005, "loss": 1.183056354522705, "step": 26577 }, { "epoch": 2.095230587307844, "grad_norm": 0.4524003267288208, "learning_rate": 0.0005, "loss": 1.3998100757598877, "step": 26578 }, { "epoch": 2.0953094205754828, "grad_norm": 0.39041075110435486, "learning_rate": 0.0005, "loss": 1.1094574928283691, "step": 26579 }, { "epoch": 2.095388253843122, "grad_norm": 0.43976014852523804, "learning_rate": 0.0005, "loss": 1.2460410594940186, "step": 26580 }, { "epoch": 2.0954670871107606, "grad_norm": 0.6310276389122009, "learning_rate": 0.0005, "loss": 1.1855461597442627, "step": 26581 }, { "epoch": 2.0955459203783997, "grad_norm": 0.4850495159626007, "learning_rate": 0.0005, "loss": 1.3696818351745605, "step": 26582 }, { "epoch": 2.095624753646039, "grad_norm": 0.7026390433311462, "learning_rate": 0.0005, "loss": 1.4586446285247803, "step": 26583 }, { "epoch": 2.0957035869136775, "grad_norm": 0.47819215059280396, "learning_rate": 0.0005, "loss": 1.3711748123168945, "step": 26584 }, { "epoch": 2.0957824201813167, "grad_norm": 0.5661620497703552, "learning_rate": 0.0005, "loss": 1.2948753833770752, "step": 26585 }, { "epoch": 2.0958612534489554, "grad_norm": 0.48586028814315796, "learning_rate": 0.0005, "loss": 1.1804912090301514, "step": 26586 }, { "epoch": 2.0959400867165945, "grad_norm": 0.5581140518188477, "learning_rate": 0.0005, "loss": 1.2528502941131592, "step": 26587 }, { "epoch": 2.096018919984233, "grad_norm": 0.49968841671943665, "learning_rate": 0.0005, "loss": 1.1808366775512695, "step": 26588 }, { "epoch": 2.0960977532518723, "grad_norm": 0.48062416911125183, "learning_rate": 0.0005, "loss": 1.2149598598480225, "step": 26589 }, { "epoch": 2.0961765865195114, "grad_norm": 0.44507071375846863, "learning_rate": 0.0005, "loss": 1.2553766965866089, "step": 26590 }, { "epoch": 2.09625541978715, "grad_norm": 0.48387226462364197, "learning_rate": 0.0005, "loss": 1.185063362121582, "step": 26591 }, { "epoch": 2.0963342530547893, "grad_norm": 0.4935299754142761, "learning_rate": 0.0005, "loss": 1.1714375019073486, "step": 26592 }, { "epoch": 2.096413086322428, "grad_norm": 0.4888610541820526, "learning_rate": 0.0005, "loss": 1.4021656513214111, "step": 26593 }, { "epoch": 2.096491919590067, "grad_norm": 0.46229684352874756, "learning_rate": 0.0005, "loss": 1.2310841083526611, "step": 26594 }, { "epoch": 2.0965707528577058, "grad_norm": 0.4655475914478302, "learning_rate": 0.0005, "loss": 1.3973764181137085, "step": 26595 }, { "epoch": 2.096649586125345, "grad_norm": 0.5252377986907959, "learning_rate": 0.0005, "loss": 1.4407329559326172, "step": 26596 }, { "epoch": 2.096728419392984, "grad_norm": 0.5101561546325684, "learning_rate": 0.0005, "loss": 1.2097783088684082, "step": 26597 }, { "epoch": 2.0968072526606227, "grad_norm": 0.5257771015167236, "learning_rate": 0.0005, "loss": 1.3643739223480225, "step": 26598 }, { "epoch": 2.096886085928262, "grad_norm": 0.5254238247871399, "learning_rate": 0.0005, "loss": 1.4968984127044678, "step": 26599 }, { "epoch": 2.0969649191959006, "grad_norm": 0.4308755695819855, "learning_rate": 0.0005, "loss": 1.192000150680542, "step": 26600 }, { "epoch": 2.0970437524635397, "grad_norm": 0.4078505039215088, "learning_rate": 0.0005, "loss": 1.1851763725280762, "step": 26601 }, { "epoch": 2.0971225857311784, "grad_norm": 0.5141676664352417, "learning_rate": 0.0005, "loss": 1.5737677812576294, "step": 26602 }, { "epoch": 2.0972014189988175, "grad_norm": 0.5575531125068665, "learning_rate": 0.0005, "loss": 1.5216014385223389, "step": 26603 }, { "epoch": 2.0972802522664566, "grad_norm": 0.489803284406662, "learning_rate": 0.0005, "loss": 1.5596739053726196, "step": 26604 }, { "epoch": 2.0973590855340953, "grad_norm": 0.5865331292152405, "learning_rate": 0.0005, "loss": 1.3443204164505005, "step": 26605 }, { "epoch": 2.0974379188017345, "grad_norm": 0.43874746561050415, "learning_rate": 0.0005, "loss": 1.1930466890335083, "step": 26606 }, { "epoch": 2.097516752069373, "grad_norm": 0.6314362287521362, "learning_rate": 0.0005, "loss": 1.62313711643219, "step": 26607 }, { "epoch": 2.0975955853370123, "grad_norm": 0.42300212383270264, "learning_rate": 0.0005, "loss": 1.3604960441589355, "step": 26608 }, { "epoch": 2.097674418604651, "grad_norm": 0.4898924231529236, "learning_rate": 0.0005, "loss": 1.244034767150879, "step": 26609 }, { "epoch": 2.09775325187229, "grad_norm": 0.46586674451828003, "learning_rate": 0.0005, "loss": 1.3463780879974365, "step": 26610 }, { "epoch": 2.0978320851399292, "grad_norm": 0.6226338148117065, "learning_rate": 0.0005, "loss": 1.3869824409484863, "step": 26611 }, { "epoch": 2.097910918407568, "grad_norm": 0.4511406421661377, "learning_rate": 0.0005, "loss": 1.54093337059021, "step": 26612 }, { "epoch": 2.097989751675207, "grad_norm": 0.44498229026794434, "learning_rate": 0.0005, "loss": 1.3339793682098389, "step": 26613 }, { "epoch": 2.0980685849428458, "grad_norm": 0.4497690200805664, "learning_rate": 0.0005, "loss": 1.4005314111709595, "step": 26614 }, { "epoch": 2.098147418210485, "grad_norm": 0.4465106427669525, "learning_rate": 0.0005, "loss": 1.282731056213379, "step": 26615 }, { "epoch": 2.0982262514781236, "grad_norm": 0.4700862765312195, "learning_rate": 0.0005, "loss": 1.21079683303833, "step": 26616 }, { "epoch": 2.0983050847457627, "grad_norm": 0.44045716524124146, "learning_rate": 0.0005, "loss": 1.3236937522888184, "step": 26617 }, { "epoch": 2.098383918013402, "grad_norm": 0.47824838757514954, "learning_rate": 0.0005, "loss": 1.2293550968170166, "step": 26618 }, { "epoch": 2.0984627512810405, "grad_norm": 0.4434813857078552, "learning_rate": 0.0005, "loss": 1.2100927829742432, "step": 26619 }, { "epoch": 2.0985415845486797, "grad_norm": 0.5115941166877747, "learning_rate": 0.0005, "loss": 1.253886103630066, "step": 26620 }, { "epoch": 2.0986204178163184, "grad_norm": 0.4301083981990814, "learning_rate": 0.0005, "loss": 1.2517632246017456, "step": 26621 }, { "epoch": 2.0986992510839575, "grad_norm": 0.4912784993648529, "learning_rate": 0.0005, "loss": 1.4422452449798584, "step": 26622 }, { "epoch": 2.098778084351596, "grad_norm": 0.5689113140106201, "learning_rate": 0.0005, "loss": 1.585265874862671, "step": 26623 }, { "epoch": 2.0988569176192353, "grad_norm": 0.5594332814216614, "learning_rate": 0.0005, "loss": 1.3842406272888184, "step": 26624 }, { "epoch": 2.0989357508868745, "grad_norm": 0.5359715819358826, "learning_rate": 0.0005, "loss": 1.5175158977508545, "step": 26625 }, { "epoch": 2.099014584154513, "grad_norm": 0.5009867548942566, "learning_rate": 0.0005, "loss": 1.375704050064087, "step": 26626 }, { "epoch": 2.0990934174221523, "grad_norm": 0.4340704381465912, "learning_rate": 0.0005, "loss": 1.3194875717163086, "step": 26627 }, { "epoch": 2.099172250689791, "grad_norm": 0.46528634428977966, "learning_rate": 0.0005, "loss": 1.199991226196289, "step": 26628 }, { "epoch": 2.09925108395743, "grad_norm": 0.459877610206604, "learning_rate": 0.0005, "loss": 1.241070032119751, "step": 26629 }, { "epoch": 2.099329917225069, "grad_norm": 0.5208485722541809, "learning_rate": 0.0005, "loss": 1.276430606842041, "step": 26630 }, { "epoch": 2.099408750492708, "grad_norm": 0.4470897614955902, "learning_rate": 0.0005, "loss": 1.3426244258880615, "step": 26631 }, { "epoch": 2.099487583760347, "grad_norm": 0.5274063944816589, "learning_rate": 0.0005, "loss": 1.335557460784912, "step": 26632 }, { "epoch": 2.0995664170279857, "grad_norm": 0.5588079690933228, "learning_rate": 0.0005, "loss": 1.3048239946365356, "step": 26633 }, { "epoch": 2.099645250295625, "grad_norm": 0.4202389121055603, "learning_rate": 0.0005, "loss": 1.2738230228424072, "step": 26634 }, { "epoch": 2.0997240835632636, "grad_norm": 0.5578058362007141, "learning_rate": 0.0005, "loss": 1.3119304180145264, "step": 26635 }, { "epoch": 2.0998029168309027, "grad_norm": 0.42033275961875916, "learning_rate": 0.0005, "loss": 0.9940165281295776, "step": 26636 }, { "epoch": 2.0998817500985414, "grad_norm": 0.4667305052280426, "learning_rate": 0.0005, "loss": 1.3885438442230225, "step": 26637 }, { "epoch": 2.0999605833661805, "grad_norm": 0.559457004070282, "learning_rate": 0.0005, "loss": 1.3081451654434204, "step": 26638 }, { "epoch": 2.1000394166338197, "grad_norm": 0.424810528755188, "learning_rate": 0.0005, "loss": 1.0894579887390137, "step": 26639 }, { "epoch": 2.1001182499014583, "grad_norm": 0.46675747632980347, "learning_rate": 0.0005, "loss": 1.2995351552963257, "step": 26640 }, { "epoch": 2.1001970831690975, "grad_norm": 0.45602625608444214, "learning_rate": 0.0005, "loss": 1.2985692024230957, "step": 26641 }, { "epoch": 2.100275916436736, "grad_norm": 0.5042321681976318, "learning_rate": 0.0005, "loss": 1.4795889854431152, "step": 26642 }, { "epoch": 2.1003547497043753, "grad_norm": 0.5356577038764954, "learning_rate": 0.0005, "loss": 1.4865641593933105, "step": 26643 }, { "epoch": 2.100433582972014, "grad_norm": 0.470095694065094, "learning_rate": 0.0005, "loss": 1.2228407859802246, "step": 26644 }, { "epoch": 2.100512416239653, "grad_norm": 0.5255061984062195, "learning_rate": 0.0005, "loss": 1.2277858257293701, "step": 26645 }, { "epoch": 2.1005912495072923, "grad_norm": 0.8386034369468689, "learning_rate": 0.0005, "loss": 1.4041781425476074, "step": 26646 }, { "epoch": 2.100670082774931, "grad_norm": 0.4523332118988037, "learning_rate": 0.0005, "loss": 1.2398512363433838, "step": 26647 }, { "epoch": 2.10074891604257, "grad_norm": 0.4645203948020935, "learning_rate": 0.0005, "loss": 1.3534176349639893, "step": 26648 }, { "epoch": 2.1008277493102088, "grad_norm": 0.4931237995624542, "learning_rate": 0.0005, "loss": 1.2162868976593018, "step": 26649 }, { "epoch": 2.100906582577848, "grad_norm": 0.4869278371334076, "learning_rate": 0.0005, "loss": 1.2879033088684082, "step": 26650 }, { "epoch": 2.1009854158454866, "grad_norm": 0.5333353281021118, "learning_rate": 0.0005, "loss": 1.227020025253296, "step": 26651 }, { "epoch": 2.1010642491131257, "grad_norm": 0.48297882080078125, "learning_rate": 0.0005, "loss": 1.274574875831604, "step": 26652 }, { "epoch": 2.101143082380765, "grad_norm": 0.5194970369338989, "learning_rate": 0.0005, "loss": 1.400665521621704, "step": 26653 }, { "epoch": 2.1012219156484035, "grad_norm": 0.5751033425331116, "learning_rate": 0.0005, "loss": 1.2881404161453247, "step": 26654 }, { "epoch": 2.1013007489160427, "grad_norm": 0.4604642689228058, "learning_rate": 0.0005, "loss": 1.333867073059082, "step": 26655 }, { "epoch": 2.1013795821836814, "grad_norm": 0.5391600131988525, "learning_rate": 0.0005, "loss": 1.4501550197601318, "step": 26656 }, { "epoch": 2.1014584154513205, "grad_norm": 0.5226851105690002, "learning_rate": 0.0005, "loss": 1.2126247882843018, "step": 26657 }, { "epoch": 2.101537248718959, "grad_norm": 0.49384021759033203, "learning_rate": 0.0005, "loss": 1.4043142795562744, "step": 26658 }, { "epoch": 2.1016160819865983, "grad_norm": 0.4875600039958954, "learning_rate": 0.0005, "loss": 1.387028455734253, "step": 26659 }, { "epoch": 2.1016949152542375, "grad_norm": 0.4471265971660614, "learning_rate": 0.0005, "loss": 1.178729772567749, "step": 26660 }, { "epoch": 2.101773748521876, "grad_norm": 0.5769782662391663, "learning_rate": 0.0005, "loss": 1.4219996929168701, "step": 26661 }, { "epoch": 2.1018525817895153, "grad_norm": 0.4560075104236603, "learning_rate": 0.0005, "loss": 1.4717111587524414, "step": 26662 }, { "epoch": 2.101931415057154, "grad_norm": 0.5724084377288818, "learning_rate": 0.0005, "loss": 1.3527443408966064, "step": 26663 }, { "epoch": 2.102010248324793, "grad_norm": 0.45214658975601196, "learning_rate": 0.0005, "loss": 1.1508619785308838, "step": 26664 }, { "epoch": 2.102089081592432, "grad_norm": 0.5611408352851868, "learning_rate": 0.0005, "loss": 1.259108543395996, "step": 26665 }, { "epoch": 2.102167914860071, "grad_norm": 0.4492090344429016, "learning_rate": 0.0005, "loss": 1.4110383987426758, "step": 26666 }, { "epoch": 2.10224674812771, "grad_norm": 0.43987056612968445, "learning_rate": 0.0005, "loss": 1.2683230638504028, "step": 26667 }, { "epoch": 2.1023255813953488, "grad_norm": 0.4687987267971039, "learning_rate": 0.0005, "loss": 1.3021810054779053, "step": 26668 }, { "epoch": 2.102404414662988, "grad_norm": 0.5017737150192261, "learning_rate": 0.0005, "loss": 1.2334046363830566, "step": 26669 }, { "epoch": 2.1024832479306266, "grad_norm": 0.5160804390907288, "learning_rate": 0.0005, "loss": 1.3039453029632568, "step": 26670 }, { "epoch": 2.1025620811982657, "grad_norm": 0.4517328143119812, "learning_rate": 0.0005, "loss": 1.382965087890625, "step": 26671 }, { "epoch": 2.1026409144659044, "grad_norm": 0.4526328444480896, "learning_rate": 0.0005, "loss": 1.4418026208877563, "step": 26672 }, { "epoch": 2.1027197477335435, "grad_norm": 0.5344780087471008, "learning_rate": 0.0005, "loss": 1.2991050481796265, "step": 26673 }, { "epoch": 2.1027985810011827, "grad_norm": 0.47583869099617004, "learning_rate": 0.0005, "loss": 1.2191811800003052, "step": 26674 }, { "epoch": 2.1028774142688214, "grad_norm": 0.5333475470542908, "learning_rate": 0.0005, "loss": 1.3511693477630615, "step": 26675 }, { "epoch": 2.1029562475364605, "grad_norm": 0.49352407455444336, "learning_rate": 0.0005, "loss": 1.315260887145996, "step": 26676 }, { "epoch": 2.103035080804099, "grad_norm": 0.46644338965415955, "learning_rate": 0.0005, "loss": 1.2584668397903442, "step": 26677 }, { "epoch": 2.1031139140717383, "grad_norm": 0.4627395272254944, "learning_rate": 0.0005, "loss": 1.269075632095337, "step": 26678 }, { "epoch": 2.103192747339377, "grad_norm": 0.4011194705963135, "learning_rate": 0.0005, "loss": 1.1424301862716675, "step": 26679 }, { "epoch": 2.103271580607016, "grad_norm": 0.5376883149147034, "learning_rate": 0.0005, "loss": 1.3048001527786255, "step": 26680 }, { "epoch": 2.1033504138746553, "grad_norm": 0.51097571849823, "learning_rate": 0.0005, "loss": 1.4042105674743652, "step": 26681 }, { "epoch": 2.103429247142294, "grad_norm": 0.5398691892623901, "learning_rate": 0.0005, "loss": 1.3534514904022217, "step": 26682 }, { "epoch": 2.103508080409933, "grad_norm": 0.6048057079315186, "learning_rate": 0.0005, "loss": 1.389974594116211, "step": 26683 }, { "epoch": 2.103586913677572, "grad_norm": 0.512852132320404, "learning_rate": 0.0005, "loss": 1.4307767152786255, "step": 26684 }, { "epoch": 2.103665746945211, "grad_norm": 0.4507489800453186, "learning_rate": 0.0005, "loss": 1.1828153133392334, "step": 26685 }, { "epoch": 2.10374458021285, "grad_norm": 0.5519083142280579, "learning_rate": 0.0005, "loss": 1.5260857343673706, "step": 26686 }, { "epoch": 2.1038234134804887, "grad_norm": 0.5399293303489685, "learning_rate": 0.0005, "loss": 1.193921446800232, "step": 26687 }, { "epoch": 2.103902246748128, "grad_norm": 0.4898064136505127, "learning_rate": 0.0005, "loss": 1.2025657892227173, "step": 26688 }, { "epoch": 2.1039810800157666, "grad_norm": 0.6050623655319214, "learning_rate": 0.0005, "loss": 1.12577486038208, "step": 26689 }, { "epoch": 2.1040599132834057, "grad_norm": 0.6068325042724609, "learning_rate": 0.0005, "loss": 1.5359747409820557, "step": 26690 }, { "epoch": 2.1041387465510444, "grad_norm": 0.5222295522689819, "learning_rate": 0.0005, "loss": 1.3870829343795776, "step": 26691 }, { "epoch": 2.1042175798186835, "grad_norm": 0.3949841260910034, "learning_rate": 0.0005, "loss": 1.0629020929336548, "step": 26692 }, { "epoch": 2.104296413086322, "grad_norm": 0.4270123541355133, "learning_rate": 0.0005, "loss": 1.2326486110687256, "step": 26693 }, { "epoch": 2.1043752463539613, "grad_norm": 0.4926638901233673, "learning_rate": 0.0005, "loss": 1.3478578329086304, "step": 26694 }, { "epoch": 2.1044540796216005, "grad_norm": 0.5027843117713928, "learning_rate": 0.0005, "loss": 1.4875602722167969, "step": 26695 }, { "epoch": 2.104532912889239, "grad_norm": 0.413832426071167, "learning_rate": 0.0005, "loss": 1.108567237854004, "step": 26696 }, { "epoch": 2.1046117461568783, "grad_norm": 0.41370004415512085, "learning_rate": 0.0005, "loss": 1.177777647972107, "step": 26697 }, { "epoch": 2.104690579424517, "grad_norm": 0.45750999450683594, "learning_rate": 0.0005, "loss": 1.2928450107574463, "step": 26698 }, { "epoch": 2.104769412692156, "grad_norm": 0.4796393811702728, "learning_rate": 0.0005, "loss": 1.298269510269165, "step": 26699 }, { "epoch": 2.1048482459597952, "grad_norm": 0.4444155693054199, "learning_rate": 0.0005, "loss": 1.3927435874938965, "step": 26700 }, { "epoch": 2.104927079227434, "grad_norm": 0.5002509355545044, "learning_rate": 0.0005, "loss": 1.236060619354248, "step": 26701 }, { "epoch": 2.105005912495073, "grad_norm": 0.5191283822059631, "learning_rate": 0.0005, "loss": 1.2791781425476074, "step": 26702 }, { "epoch": 2.1050847457627118, "grad_norm": 0.6829298138618469, "learning_rate": 0.0005, "loss": 1.2588508129119873, "step": 26703 }, { "epoch": 2.105163579030351, "grad_norm": 0.5508917570114136, "learning_rate": 0.0005, "loss": 1.2544784545898438, "step": 26704 }, { "epoch": 2.1052424122979896, "grad_norm": 0.4511146545410156, "learning_rate": 0.0005, "loss": 1.3598806858062744, "step": 26705 }, { "epoch": 2.1053212455656287, "grad_norm": 0.6953930258750916, "learning_rate": 0.0005, "loss": 1.39387845993042, "step": 26706 }, { "epoch": 2.105400078833268, "grad_norm": 0.5470321178436279, "learning_rate": 0.0005, "loss": 1.4137060642242432, "step": 26707 }, { "epoch": 2.1054789121009065, "grad_norm": 0.5698243975639343, "learning_rate": 0.0005, "loss": 1.2613731622695923, "step": 26708 }, { "epoch": 2.1055577453685457, "grad_norm": 0.44187793135643005, "learning_rate": 0.0005, "loss": 1.1508291959762573, "step": 26709 }, { "epoch": 2.1056365786361844, "grad_norm": 0.45922771096229553, "learning_rate": 0.0005, "loss": 1.1228585243225098, "step": 26710 }, { "epoch": 2.1057154119038235, "grad_norm": 0.560935378074646, "learning_rate": 0.0005, "loss": 1.3475767374038696, "step": 26711 }, { "epoch": 2.105794245171462, "grad_norm": 0.5268117785453796, "learning_rate": 0.0005, "loss": 1.337080955505371, "step": 26712 }, { "epoch": 2.1058730784391013, "grad_norm": 0.5301322340965271, "learning_rate": 0.0005, "loss": 1.2533454895019531, "step": 26713 }, { "epoch": 2.1059519117067405, "grad_norm": 0.533323347568512, "learning_rate": 0.0005, "loss": 1.52571439743042, "step": 26714 }, { "epoch": 2.106030744974379, "grad_norm": 0.5107543468475342, "learning_rate": 0.0005, "loss": 1.5538150072097778, "step": 26715 }, { "epoch": 2.1061095782420183, "grad_norm": 0.46392273902893066, "learning_rate": 0.0005, "loss": 1.3092873096466064, "step": 26716 }, { "epoch": 2.106188411509657, "grad_norm": 0.6330254077911377, "learning_rate": 0.0005, "loss": 1.5389679670333862, "step": 26717 }, { "epoch": 2.106267244777296, "grad_norm": 0.48007264733314514, "learning_rate": 0.0005, "loss": 1.2956337928771973, "step": 26718 }, { "epoch": 2.106346078044935, "grad_norm": 0.7270219922065735, "learning_rate": 0.0005, "loss": 1.5948576927185059, "step": 26719 }, { "epoch": 2.106424911312574, "grad_norm": 0.517892062664032, "learning_rate": 0.0005, "loss": 1.357945203781128, "step": 26720 }, { "epoch": 2.106503744580213, "grad_norm": 0.6024863719940186, "learning_rate": 0.0005, "loss": 1.345337152481079, "step": 26721 }, { "epoch": 2.1065825778478517, "grad_norm": 0.46232903003692627, "learning_rate": 0.0005, "loss": 1.286480188369751, "step": 26722 }, { "epoch": 2.106661411115491, "grad_norm": 0.5358757972717285, "learning_rate": 0.0005, "loss": 1.2966779470443726, "step": 26723 }, { "epoch": 2.1067402443831296, "grad_norm": 0.48959460854530334, "learning_rate": 0.0005, "loss": 1.297929286956787, "step": 26724 }, { "epoch": 2.1068190776507687, "grad_norm": 0.568043053150177, "learning_rate": 0.0005, "loss": 1.3518853187561035, "step": 26725 }, { "epoch": 2.1068979109184074, "grad_norm": 0.4999444782733917, "learning_rate": 0.0005, "loss": 1.5752067565917969, "step": 26726 }, { "epoch": 2.1069767441860465, "grad_norm": 0.48276418447494507, "learning_rate": 0.0005, "loss": 1.269975185394287, "step": 26727 }, { "epoch": 2.1070555774536857, "grad_norm": 0.5334150195121765, "learning_rate": 0.0005, "loss": 1.3620271682739258, "step": 26728 }, { "epoch": 2.1071344107213243, "grad_norm": 0.4413844347000122, "learning_rate": 0.0005, "loss": 1.3096089363098145, "step": 26729 }, { "epoch": 2.1072132439889635, "grad_norm": 0.6193228960037231, "learning_rate": 0.0005, "loss": 1.5619893074035645, "step": 26730 }, { "epoch": 2.107292077256602, "grad_norm": 0.4494479298591614, "learning_rate": 0.0005, "loss": 1.3650951385498047, "step": 26731 }, { "epoch": 2.1073709105242413, "grad_norm": 0.5101513862609863, "learning_rate": 0.0005, "loss": 1.3819317817687988, "step": 26732 }, { "epoch": 2.10744974379188, "grad_norm": 0.5452712178230286, "learning_rate": 0.0005, "loss": 1.2609010934829712, "step": 26733 }, { "epoch": 2.107528577059519, "grad_norm": 0.4525890052318573, "learning_rate": 0.0005, "loss": 1.2870163917541504, "step": 26734 }, { "epoch": 2.1076074103271583, "grad_norm": 0.4489509165287018, "learning_rate": 0.0005, "loss": 1.104935884475708, "step": 26735 }, { "epoch": 2.107686243594797, "grad_norm": 0.4770914614200592, "learning_rate": 0.0005, "loss": 1.1805239915847778, "step": 26736 }, { "epoch": 2.107765076862436, "grad_norm": 0.5449503660202026, "learning_rate": 0.0005, "loss": 1.444449782371521, "step": 26737 }, { "epoch": 2.1078439101300748, "grad_norm": 0.5024905204772949, "learning_rate": 0.0005, "loss": 1.3487446308135986, "step": 26738 }, { "epoch": 2.107922743397714, "grad_norm": 0.5285104513168335, "learning_rate": 0.0005, "loss": 1.1886746883392334, "step": 26739 }, { "epoch": 2.1080015766653526, "grad_norm": 0.39738738536834717, "learning_rate": 0.0005, "loss": 1.0709409713745117, "step": 26740 }, { "epoch": 2.1080804099329917, "grad_norm": 0.5247101187705994, "learning_rate": 0.0005, "loss": 1.1740487813949585, "step": 26741 }, { "epoch": 2.108159243200631, "grad_norm": 0.6083625555038452, "learning_rate": 0.0005, "loss": 1.3785343170166016, "step": 26742 }, { "epoch": 2.1082380764682696, "grad_norm": 0.4623282849788666, "learning_rate": 0.0005, "loss": 1.1655991077423096, "step": 26743 }, { "epoch": 2.1083169097359087, "grad_norm": 0.4548473656177521, "learning_rate": 0.0005, "loss": 1.3986992835998535, "step": 26744 }, { "epoch": 2.1083957430035474, "grad_norm": 0.48333263397216797, "learning_rate": 0.0005, "loss": 1.331477403640747, "step": 26745 }, { "epoch": 2.1084745762711865, "grad_norm": 0.432716965675354, "learning_rate": 0.0005, "loss": 1.2538384199142456, "step": 26746 }, { "epoch": 2.108553409538825, "grad_norm": 0.5618542432785034, "learning_rate": 0.0005, "loss": 1.2643897533416748, "step": 26747 }, { "epoch": 2.1086322428064643, "grad_norm": 0.5318111777305603, "learning_rate": 0.0005, "loss": 1.2046043872833252, "step": 26748 }, { "epoch": 2.1087110760741035, "grad_norm": 0.5036038160324097, "learning_rate": 0.0005, "loss": 1.1053603887557983, "step": 26749 }, { "epoch": 2.108789909341742, "grad_norm": 0.5919277667999268, "learning_rate": 0.0005, "loss": 1.569443941116333, "step": 26750 }, { "epoch": 2.1088687426093813, "grad_norm": 0.5208078026771545, "learning_rate": 0.0005, "loss": 1.2702417373657227, "step": 26751 }, { "epoch": 2.10894757587702, "grad_norm": 0.4925954043865204, "learning_rate": 0.0005, "loss": 1.4519387483596802, "step": 26752 }, { "epoch": 2.109026409144659, "grad_norm": 0.4534876048564911, "learning_rate": 0.0005, "loss": 1.2838032245635986, "step": 26753 }, { "epoch": 2.109105242412298, "grad_norm": 0.5099811553955078, "learning_rate": 0.0005, "loss": 1.2700557708740234, "step": 26754 }, { "epoch": 2.109184075679937, "grad_norm": 0.509060800075531, "learning_rate": 0.0005, "loss": 1.430159568786621, "step": 26755 }, { "epoch": 2.109262908947576, "grad_norm": 0.42561665177345276, "learning_rate": 0.0005, "loss": 1.232008695602417, "step": 26756 }, { "epoch": 2.1093417422152148, "grad_norm": 0.4590260684490204, "learning_rate": 0.0005, "loss": 1.280921220779419, "step": 26757 }, { "epoch": 2.109420575482854, "grad_norm": 0.5233948826789856, "learning_rate": 0.0005, "loss": 1.2185829877853394, "step": 26758 }, { "epoch": 2.1094994087504926, "grad_norm": 0.5428136587142944, "learning_rate": 0.0005, "loss": 1.427823781967163, "step": 26759 }, { "epoch": 2.1095782420181317, "grad_norm": 0.5770111083984375, "learning_rate": 0.0005, "loss": 1.4217530488967896, "step": 26760 }, { "epoch": 2.1096570752857704, "grad_norm": 0.525330662727356, "learning_rate": 0.0005, "loss": 1.4227869510650635, "step": 26761 }, { "epoch": 2.1097359085534095, "grad_norm": 0.5218520760536194, "learning_rate": 0.0005, "loss": 1.3029725551605225, "step": 26762 }, { "epoch": 2.1098147418210487, "grad_norm": 0.4829341769218445, "learning_rate": 0.0005, "loss": 1.3702654838562012, "step": 26763 }, { "epoch": 2.1098935750886874, "grad_norm": 0.5882985591888428, "learning_rate": 0.0005, "loss": 1.4662684202194214, "step": 26764 }, { "epoch": 2.1099724083563265, "grad_norm": 0.468966007232666, "learning_rate": 0.0005, "loss": 1.4199590682983398, "step": 26765 }, { "epoch": 2.110051241623965, "grad_norm": 0.533839225769043, "learning_rate": 0.0005, "loss": 1.2055118083953857, "step": 26766 }, { "epoch": 2.1101300748916043, "grad_norm": 0.5746224522590637, "learning_rate": 0.0005, "loss": 1.4924839735031128, "step": 26767 }, { "epoch": 2.110208908159243, "grad_norm": 0.5070051550865173, "learning_rate": 0.0005, "loss": 1.4373456239700317, "step": 26768 }, { "epoch": 2.110287741426882, "grad_norm": 0.5442166924476624, "learning_rate": 0.0005, "loss": 1.2348449230194092, "step": 26769 }, { "epoch": 2.1103665746945213, "grad_norm": 0.47298160195350647, "learning_rate": 0.0005, "loss": 1.3362529277801514, "step": 26770 }, { "epoch": 2.11044540796216, "grad_norm": 5.094367504119873, "learning_rate": 0.0005, "loss": 1.35184907913208, "step": 26771 }, { "epoch": 2.110524241229799, "grad_norm": 0.5649314522743225, "learning_rate": 0.0005, "loss": 1.4738118648529053, "step": 26772 }, { "epoch": 2.110603074497438, "grad_norm": 0.4502819776535034, "learning_rate": 0.0005, "loss": 1.418975830078125, "step": 26773 }, { "epoch": 2.110681907765077, "grad_norm": 0.47065088152885437, "learning_rate": 0.0005, "loss": 1.318141222000122, "step": 26774 }, { "epoch": 2.1107607410327156, "grad_norm": 0.5279043912887573, "learning_rate": 0.0005, "loss": 1.535410761833191, "step": 26775 }, { "epoch": 2.1108395743003547, "grad_norm": 0.5448740720748901, "learning_rate": 0.0005, "loss": 1.4905827045440674, "step": 26776 }, { "epoch": 2.110918407567994, "grad_norm": 0.5764236450195312, "learning_rate": 0.0005, "loss": 1.6858494281768799, "step": 26777 }, { "epoch": 2.1109972408356326, "grad_norm": 0.4711470305919647, "learning_rate": 0.0005, "loss": 1.345945954322815, "step": 26778 }, { "epoch": 2.1110760741032717, "grad_norm": 0.3945142924785614, "learning_rate": 0.0005, "loss": 1.1404063701629639, "step": 26779 }, { "epoch": 2.1111549073709104, "grad_norm": 0.5591504573822021, "learning_rate": 0.0005, "loss": 1.360634446144104, "step": 26780 }, { "epoch": 2.1112337406385495, "grad_norm": 0.48588940501213074, "learning_rate": 0.0005, "loss": 1.3543404340744019, "step": 26781 }, { "epoch": 2.111312573906188, "grad_norm": 0.5083717107772827, "learning_rate": 0.0005, "loss": 1.4381740093231201, "step": 26782 }, { "epoch": 2.1113914071738273, "grad_norm": 0.445243239402771, "learning_rate": 0.0005, "loss": 1.4285712242126465, "step": 26783 }, { "epoch": 2.1114702404414665, "grad_norm": 0.4171890616416931, "learning_rate": 0.0005, "loss": 1.235175371170044, "step": 26784 }, { "epoch": 2.111549073709105, "grad_norm": 0.4959377646446228, "learning_rate": 0.0005, "loss": 1.2062981128692627, "step": 26785 }, { "epoch": 2.1116279069767443, "grad_norm": 0.4596100449562073, "learning_rate": 0.0005, "loss": 1.327298879623413, "step": 26786 }, { "epoch": 2.111706740244383, "grad_norm": 0.5557661056518555, "learning_rate": 0.0005, "loss": 1.386292576789856, "step": 26787 }, { "epoch": 2.111785573512022, "grad_norm": 0.4592265188694, "learning_rate": 0.0005, "loss": 1.2620952129364014, "step": 26788 }, { "epoch": 2.111864406779661, "grad_norm": 0.5129113793373108, "learning_rate": 0.0005, "loss": 1.4388710260391235, "step": 26789 }, { "epoch": 2.1119432400473, "grad_norm": 0.4657931327819824, "learning_rate": 0.0005, "loss": 1.3323817253112793, "step": 26790 }, { "epoch": 2.112022073314939, "grad_norm": 0.49116894602775574, "learning_rate": 0.0005, "loss": 1.3147304058074951, "step": 26791 }, { "epoch": 2.1121009065825778, "grad_norm": 0.39522144198417664, "learning_rate": 0.0005, "loss": 1.1470434665679932, "step": 26792 }, { "epoch": 2.112179739850217, "grad_norm": 0.44380491971969604, "learning_rate": 0.0005, "loss": 1.3684357404708862, "step": 26793 }, { "epoch": 2.1122585731178556, "grad_norm": 0.4432888627052307, "learning_rate": 0.0005, "loss": 1.2753443717956543, "step": 26794 }, { "epoch": 2.1123374063854947, "grad_norm": 0.4967398941516876, "learning_rate": 0.0005, "loss": 1.2807626724243164, "step": 26795 }, { "epoch": 2.1124162396531334, "grad_norm": 0.7028184533119202, "learning_rate": 0.0005, "loss": 1.3803730010986328, "step": 26796 }, { "epoch": 2.1124950729207725, "grad_norm": 0.5372222661972046, "learning_rate": 0.0005, "loss": 1.1803255081176758, "step": 26797 }, { "epoch": 2.1125739061884117, "grad_norm": 0.44148027896881104, "learning_rate": 0.0005, "loss": 1.2712681293487549, "step": 26798 }, { "epoch": 2.1126527394560504, "grad_norm": 0.41212883591651917, "learning_rate": 0.0005, "loss": 1.2969622611999512, "step": 26799 }, { "epoch": 2.1127315727236895, "grad_norm": 0.5244424343109131, "learning_rate": 0.0005, "loss": 1.4337025880813599, "step": 26800 }, { "epoch": 2.112810405991328, "grad_norm": 0.492530882358551, "learning_rate": 0.0005, "loss": 1.3988386392593384, "step": 26801 }, { "epoch": 2.1128892392589673, "grad_norm": 0.4565468728542328, "learning_rate": 0.0005, "loss": 1.425658941268921, "step": 26802 }, { "epoch": 2.1129680725266065, "grad_norm": 0.4449803829193115, "learning_rate": 0.0005, "loss": 1.371911883354187, "step": 26803 }, { "epoch": 2.113046905794245, "grad_norm": 0.7242526412010193, "learning_rate": 0.0005, "loss": 1.323730707168579, "step": 26804 }, { "epoch": 2.1131257390618843, "grad_norm": 0.4889277219772339, "learning_rate": 0.0005, "loss": 1.4414563179016113, "step": 26805 }, { "epoch": 2.113204572329523, "grad_norm": 0.4567081034183502, "learning_rate": 0.0005, "loss": 1.3892903327941895, "step": 26806 }, { "epoch": 2.113283405597162, "grad_norm": 0.5948565602302551, "learning_rate": 0.0005, "loss": 1.3770549297332764, "step": 26807 }, { "epoch": 2.113362238864801, "grad_norm": 0.6256197094917297, "learning_rate": 0.0005, "loss": 1.3272268772125244, "step": 26808 }, { "epoch": 2.11344107213244, "grad_norm": 0.4427895247936249, "learning_rate": 0.0005, "loss": 1.337061882019043, "step": 26809 }, { "epoch": 2.1135199054000786, "grad_norm": 0.7135319709777832, "learning_rate": 0.0005, "loss": 1.409035086631775, "step": 26810 }, { "epoch": 2.1135987386677177, "grad_norm": 0.44318801164627075, "learning_rate": 0.0005, "loss": 1.2728846073150635, "step": 26811 }, { "epoch": 2.113677571935357, "grad_norm": 0.603938102722168, "learning_rate": 0.0005, "loss": 1.3312604427337646, "step": 26812 }, { "epoch": 2.1137564052029956, "grad_norm": 0.6614887118339539, "learning_rate": 0.0005, "loss": 1.1609697341918945, "step": 26813 }, { "epoch": 2.1138352384706347, "grad_norm": 0.5545361638069153, "learning_rate": 0.0005, "loss": 1.6032088994979858, "step": 26814 }, { "epoch": 2.1139140717382734, "grad_norm": 0.4722113013267517, "learning_rate": 0.0005, "loss": 1.28596830368042, "step": 26815 }, { "epoch": 2.1139929050059125, "grad_norm": 0.4831857979297638, "learning_rate": 0.0005, "loss": 1.3309271335601807, "step": 26816 }, { "epoch": 2.1140717382735517, "grad_norm": 0.6472719311714172, "learning_rate": 0.0005, "loss": 1.338250756263733, "step": 26817 }, { "epoch": 2.1141505715411903, "grad_norm": 0.47838330268859863, "learning_rate": 0.0005, "loss": 1.3494477272033691, "step": 26818 }, { "epoch": 2.1142294048088295, "grad_norm": 0.6077718138694763, "learning_rate": 0.0005, "loss": 1.4012579917907715, "step": 26819 }, { "epoch": 2.114308238076468, "grad_norm": 0.49413102865219116, "learning_rate": 0.0005, "loss": 1.4465222358703613, "step": 26820 }, { "epoch": 2.1143870713441073, "grad_norm": 0.5476879477500916, "learning_rate": 0.0005, "loss": 1.3563483953475952, "step": 26821 }, { "epoch": 2.114465904611746, "grad_norm": 0.5708193182945251, "learning_rate": 0.0005, "loss": 1.3745614290237427, "step": 26822 }, { "epoch": 2.114544737879385, "grad_norm": 0.5432684421539307, "learning_rate": 0.0005, "loss": 1.3203380107879639, "step": 26823 }, { "epoch": 2.114623571147024, "grad_norm": 0.5049065351486206, "learning_rate": 0.0005, "loss": 1.2121107578277588, "step": 26824 }, { "epoch": 2.114702404414663, "grad_norm": 0.5510658621788025, "learning_rate": 0.0005, "loss": 1.450998067855835, "step": 26825 }, { "epoch": 2.114781237682302, "grad_norm": 0.5777895450592041, "learning_rate": 0.0005, "loss": 1.4022185802459717, "step": 26826 }, { "epoch": 2.1148600709499408, "grad_norm": 0.48968297243118286, "learning_rate": 0.0005, "loss": 1.2818759679794312, "step": 26827 }, { "epoch": 2.11493890421758, "grad_norm": 0.4669308662414551, "learning_rate": 0.0005, "loss": 1.3741058111190796, "step": 26828 }, { "epoch": 2.1150177374852186, "grad_norm": 0.48598483204841614, "learning_rate": 0.0005, "loss": 1.3323050737380981, "step": 26829 }, { "epoch": 2.1150965707528577, "grad_norm": 0.4070380926132202, "learning_rate": 0.0005, "loss": 1.3568115234375, "step": 26830 }, { "epoch": 2.115175404020497, "grad_norm": 0.5702078938484192, "learning_rate": 0.0005, "loss": 1.3906246423721313, "step": 26831 }, { "epoch": 2.1152542372881356, "grad_norm": 0.583177924156189, "learning_rate": 0.0005, "loss": 1.4464144706726074, "step": 26832 }, { "epoch": 2.1153330705557747, "grad_norm": 0.46884042024612427, "learning_rate": 0.0005, "loss": 1.3065650463104248, "step": 26833 }, { "epoch": 2.1154119038234134, "grad_norm": 0.43143370747566223, "learning_rate": 0.0005, "loss": 1.183756709098816, "step": 26834 }, { "epoch": 2.1154907370910525, "grad_norm": 0.5318250060081482, "learning_rate": 0.0005, "loss": 1.4595108032226562, "step": 26835 }, { "epoch": 2.115569570358691, "grad_norm": 0.44135087728500366, "learning_rate": 0.0005, "loss": 1.3035483360290527, "step": 26836 }, { "epoch": 2.1156484036263303, "grad_norm": 0.4138813316822052, "learning_rate": 0.0005, "loss": 1.3939014673233032, "step": 26837 }, { "epoch": 2.1157272368939695, "grad_norm": 0.48974499106407166, "learning_rate": 0.0005, "loss": 1.3355616331100464, "step": 26838 }, { "epoch": 2.115806070161608, "grad_norm": 0.48463213443756104, "learning_rate": 0.0005, "loss": 1.3755767345428467, "step": 26839 }, { "epoch": 2.1158849034292473, "grad_norm": 0.5118165016174316, "learning_rate": 0.0005, "loss": 1.4339655637741089, "step": 26840 }, { "epoch": 2.115963736696886, "grad_norm": 0.4951791763305664, "learning_rate": 0.0005, "loss": 1.4871571063995361, "step": 26841 }, { "epoch": 2.116042569964525, "grad_norm": 0.6983973383903503, "learning_rate": 0.0005, "loss": 1.6261038780212402, "step": 26842 }, { "epoch": 2.116121403232164, "grad_norm": 0.5233349204063416, "learning_rate": 0.0005, "loss": 1.2783377170562744, "step": 26843 }, { "epoch": 2.116200236499803, "grad_norm": 0.42744192481040955, "learning_rate": 0.0005, "loss": 1.3351860046386719, "step": 26844 }, { "epoch": 2.116279069767442, "grad_norm": 0.4375164210796356, "learning_rate": 0.0005, "loss": 1.1492037773132324, "step": 26845 }, { "epoch": 2.1163579030350808, "grad_norm": 0.5093523859977722, "learning_rate": 0.0005, "loss": 1.2983990907669067, "step": 26846 }, { "epoch": 2.11643673630272, "grad_norm": 0.47669246792793274, "learning_rate": 0.0005, "loss": 1.2999367713928223, "step": 26847 }, { "epoch": 2.1165155695703586, "grad_norm": 0.6425495147705078, "learning_rate": 0.0005, "loss": 1.2232146263122559, "step": 26848 }, { "epoch": 2.1165944028379977, "grad_norm": 0.4834910035133362, "learning_rate": 0.0005, "loss": 1.3942375183105469, "step": 26849 }, { "epoch": 2.1166732361056364, "grad_norm": 0.6689178943634033, "learning_rate": 0.0005, "loss": 1.6675121784210205, "step": 26850 }, { "epoch": 2.1167520693732755, "grad_norm": 0.41394561529159546, "learning_rate": 0.0005, "loss": 1.1070194244384766, "step": 26851 }, { "epoch": 2.1168309026409147, "grad_norm": 0.5801023244857788, "learning_rate": 0.0005, "loss": 1.3724477291107178, "step": 26852 }, { "epoch": 2.1169097359085534, "grad_norm": 0.5133398771286011, "learning_rate": 0.0005, "loss": 1.4785267114639282, "step": 26853 }, { "epoch": 2.1169885691761925, "grad_norm": 0.4633357524871826, "learning_rate": 0.0005, "loss": 1.3043787479400635, "step": 26854 }, { "epoch": 2.117067402443831, "grad_norm": 0.6300947666168213, "learning_rate": 0.0005, "loss": 1.3747174739837646, "step": 26855 }, { "epoch": 2.1171462357114703, "grad_norm": 0.5099165439605713, "learning_rate": 0.0005, "loss": 1.3963241577148438, "step": 26856 }, { "epoch": 2.117225068979109, "grad_norm": 0.5997077822685242, "learning_rate": 0.0005, "loss": 1.4649949073791504, "step": 26857 }, { "epoch": 2.117303902246748, "grad_norm": 0.48921674489974976, "learning_rate": 0.0005, "loss": 1.2657504081726074, "step": 26858 }, { "epoch": 2.1173827355143873, "grad_norm": 0.45762762427330017, "learning_rate": 0.0005, "loss": 1.2659425735473633, "step": 26859 }, { "epoch": 2.117461568782026, "grad_norm": 0.5780647397041321, "learning_rate": 0.0005, "loss": 1.3822678327560425, "step": 26860 }, { "epoch": 2.117540402049665, "grad_norm": 0.4806910753250122, "learning_rate": 0.0005, "loss": 1.1128277778625488, "step": 26861 }, { "epoch": 2.117619235317304, "grad_norm": 0.6057684421539307, "learning_rate": 0.0005, "loss": 1.346189260482788, "step": 26862 }, { "epoch": 2.117698068584943, "grad_norm": 0.8940443396568298, "learning_rate": 0.0005, "loss": 1.351213812828064, "step": 26863 }, { "epoch": 2.1177769018525816, "grad_norm": 0.5659908056259155, "learning_rate": 0.0005, "loss": 1.29114830493927, "step": 26864 }, { "epoch": 2.1178557351202207, "grad_norm": 0.5902377963066101, "learning_rate": 0.0005, "loss": 1.1635254621505737, "step": 26865 }, { "epoch": 2.11793456838786, "grad_norm": 0.5107098817825317, "learning_rate": 0.0005, "loss": 1.3810245990753174, "step": 26866 }, { "epoch": 2.1180134016554986, "grad_norm": 0.4611211121082306, "learning_rate": 0.0005, "loss": 1.2720757722854614, "step": 26867 }, { "epoch": 2.1180922349231377, "grad_norm": 0.5669696927070618, "learning_rate": 0.0005, "loss": 1.4177693128585815, "step": 26868 }, { "epoch": 2.1181710681907764, "grad_norm": 0.5893335938453674, "learning_rate": 0.0005, "loss": 1.213749885559082, "step": 26869 }, { "epoch": 2.1182499014584155, "grad_norm": 0.4930095970630646, "learning_rate": 0.0005, "loss": 1.4182829856872559, "step": 26870 }, { "epoch": 2.118328734726054, "grad_norm": 0.5726637244224548, "learning_rate": 0.0005, "loss": 1.2948040962219238, "step": 26871 }, { "epoch": 2.1184075679936933, "grad_norm": 0.45133641362190247, "learning_rate": 0.0005, "loss": 1.21319580078125, "step": 26872 }, { "epoch": 2.1184864012613325, "grad_norm": 0.4815495014190674, "learning_rate": 0.0005, "loss": 1.3987300395965576, "step": 26873 }, { "epoch": 2.118565234528971, "grad_norm": 0.901519775390625, "learning_rate": 0.0005, "loss": 1.4607746601104736, "step": 26874 }, { "epoch": 2.1186440677966103, "grad_norm": 0.5598378777503967, "learning_rate": 0.0005, "loss": 1.4647364616394043, "step": 26875 }, { "epoch": 2.118722901064249, "grad_norm": 0.49437791109085083, "learning_rate": 0.0005, "loss": 1.2919789552688599, "step": 26876 }, { "epoch": 2.118801734331888, "grad_norm": 0.4544653594493866, "learning_rate": 0.0005, "loss": 1.3804763555526733, "step": 26877 }, { "epoch": 2.118880567599527, "grad_norm": 0.5866126418113708, "learning_rate": 0.0005, "loss": 1.2537429332733154, "step": 26878 }, { "epoch": 2.118959400867166, "grad_norm": 0.4298707842826843, "learning_rate": 0.0005, "loss": 1.3071531057357788, "step": 26879 }, { "epoch": 2.119038234134805, "grad_norm": 0.4706357717514038, "learning_rate": 0.0005, "loss": 1.2894564867019653, "step": 26880 }, { "epoch": 2.1191170674024438, "grad_norm": 0.4855218231678009, "learning_rate": 0.0005, "loss": 1.2763309478759766, "step": 26881 }, { "epoch": 2.119195900670083, "grad_norm": 0.4507063329219818, "learning_rate": 0.0005, "loss": 1.421474575996399, "step": 26882 }, { "epoch": 2.1192747339377216, "grad_norm": 0.4616182744503021, "learning_rate": 0.0005, "loss": 1.2614355087280273, "step": 26883 }, { "epoch": 2.1193535672053607, "grad_norm": 0.4229793846607208, "learning_rate": 0.0005, "loss": 1.171598196029663, "step": 26884 }, { "epoch": 2.1194324004729994, "grad_norm": 0.6838713884353638, "learning_rate": 0.0005, "loss": 1.6174994707107544, "step": 26885 }, { "epoch": 2.1195112337406385, "grad_norm": 0.504052460193634, "learning_rate": 0.0005, "loss": 1.513229489326477, "step": 26886 }, { "epoch": 2.1195900670082777, "grad_norm": 0.5996045470237732, "learning_rate": 0.0005, "loss": 1.1617058515548706, "step": 26887 }, { "epoch": 2.1196689002759164, "grad_norm": 0.5633692145347595, "learning_rate": 0.0005, "loss": 1.373208999633789, "step": 26888 }, { "epoch": 2.1197477335435555, "grad_norm": 0.5322270393371582, "learning_rate": 0.0005, "loss": 1.4363675117492676, "step": 26889 }, { "epoch": 2.119826566811194, "grad_norm": 0.5547828674316406, "learning_rate": 0.0005, "loss": 1.3894057273864746, "step": 26890 }, { "epoch": 2.1199054000788333, "grad_norm": 0.4949618875980377, "learning_rate": 0.0005, "loss": 1.289000391960144, "step": 26891 }, { "epoch": 2.119984233346472, "grad_norm": 0.43998658657073975, "learning_rate": 0.0005, "loss": 1.1605603694915771, "step": 26892 }, { "epoch": 2.120063066614111, "grad_norm": 0.46416574716567993, "learning_rate": 0.0005, "loss": 1.5829116106033325, "step": 26893 }, { "epoch": 2.1201418998817503, "grad_norm": 0.4601830542087555, "learning_rate": 0.0005, "loss": 1.1942778825759888, "step": 26894 }, { "epoch": 2.120220733149389, "grad_norm": 0.5135525465011597, "learning_rate": 0.0005, "loss": 1.3756641149520874, "step": 26895 }, { "epoch": 2.120299566417028, "grad_norm": 0.48523175716400146, "learning_rate": 0.0005, "loss": 1.223623514175415, "step": 26896 }, { "epoch": 2.120378399684667, "grad_norm": 0.4757380187511444, "learning_rate": 0.0005, "loss": 1.3971402645111084, "step": 26897 }, { "epoch": 2.120457232952306, "grad_norm": 0.5124884247779846, "learning_rate": 0.0005, "loss": 1.2797517776489258, "step": 26898 }, { "epoch": 2.1205360662199446, "grad_norm": 0.46263474225997925, "learning_rate": 0.0005, "loss": 1.3163509368896484, "step": 26899 }, { "epoch": 2.1206148994875837, "grad_norm": 0.7649278044700623, "learning_rate": 0.0005, "loss": 1.4210140705108643, "step": 26900 }, { "epoch": 2.120693732755223, "grad_norm": 0.5032075047492981, "learning_rate": 0.0005, "loss": 1.4166839122772217, "step": 26901 }, { "epoch": 2.1207725660228616, "grad_norm": 1.6103140115737915, "learning_rate": 0.0005, "loss": 1.2635825872421265, "step": 26902 }, { "epoch": 2.1208513992905007, "grad_norm": 0.4624016284942627, "learning_rate": 0.0005, "loss": 1.3030173778533936, "step": 26903 }, { "epoch": 2.1209302325581394, "grad_norm": 0.42118191719055176, "learning_rate": 0.0005, "loss": 1.2165952920913696, "step": 26904 }, { "epoch": 2.1210090658257785, "grad_norm": 0.619827926158905, "learning_rate": 0.0005, "loss": 1.3722882270812988, "step": 26905 }, { "epoch": 2.121087899093417, "grad_norm": 0.613913893699646, "learning_rate": 0.0005, "loss": 1.3438494205474854, "step": 26906 }, { "epoch": 2.1211667323610564, "grad_norm": 0.41880476474761963, "learning_rate": 0.0005, "loss": 1.2003655433654785, "step": 26907 }, { "epoch": 2.1212455656286955, "grad_norm": 0.46743667125701904, "learning_rate": 0.0005, "loss": 1.366593599319458, "step": 26908 }, { "epoch": 2.121324398896334, "grad_norm": 0.7058080434799194, "learning_rate": 0.0005, "loss": 1.3736228942871094, "step": 26909 }, { "epoch": 2.1214032321639733, "grad_norm": 0.5478537678718567, "learning_rate": 0.0005, "loss": 1.580685019493103, "step": 26910 }, { "epoch": 2.121482065431612, "grad_norm": 0.4644332528114319, "learning_rate": 0.0005, "loss": 1.2689744234085083, "step": 26911 }, { "epoch": 2.121560898699251, "grad_norm": 0.5020297765731812, "learning_rate": 0.0005, "loss": 1.3467800617218018, "step": 26912 }, { "epoch": 2.12163973196689, "grad_norm": 0.5759838819503784, "learning_rate": 0.0005, "loss": 1.5477689504623413, "step": 26913 }, { "epoch": 2.121718565234529, "grad_norm": 0.5400076508522034, "learning_rate": 0.0005, "loss": 1.2633049488067627, "step": 26914 }, { "epoch": 2.121797398502168, "grad_norm": 0.5891651511192322, "learning_rate": 0.0005, "loss": 1.3761757612228394, "step": 26915 }, { "epoch": 2.1218762317698068, "grad_norm": 0.5550164580345154, "learning_rate": 0.0005, "loss": 1.468531847000122, "step": 26916 }, { "epoch": 2.121955065037446, "grad_norm": 0.684187650680542, "learning_rate": 0.0005, "loss": 1.358460545539856, "step": 26917 }, { "epoch": 2.1220338983050846, "grad_norm": 0.5008035898208618, "learning_rate": 0.0005, "loss": 1.3392590284347534, "step": 26918 }, { "epoch": 2.1221127315727237, "grad_norm": 0.4998774528503418, "learning_rate": 0.0005, "loss": 1.4389289617538452, "step": 26919 }, { "epoch": 2.1221915648403624, "grad_norm": 0.5011453628540039, "learning_rate": 0.0005, "loss": 1.4682073593139648, "step": 26920 }, { "epoch": 2.1222703981080016, "grad_norm": 0.5977146029472351, "learning_rate": 0.0005, "loss": 1.3701670169830322, "step": 26921 }, { "epoch": 2.1223492313756407, "grad_norm": 0.46815237402915955, "learning_rate": 0.0005, "loss": 1.2576333284378052, "step": 26922 }, { "epoch": 2.1224280646432794, "grad_norm": 0.4117608368396759, "learning_rate": 0.0005, "loss": 1.3484035730361938, "step": 26923 }, { "epoch": 2.1225068979109185, "grad_norm": 0.5893922448158264, "learning_rate": 0.0005, "loss": 1.461017370223999, "step": 26924 }, { "epoch": 2.122585731178557, "grad_norm": 0.4706491529941559, "learning_rate": 0.0005, "loss": 1.244741439819336, "step": 26925 }, { "epoch": 2.1226645644461963, "grad_norm": 0.5460847616195679, "learning_rate": 0.0005, "loss": 1.558423638343811, "step": 26926 }, { "epoch": 2.122743397713835, "grad_norm": 0.47515761852264404, "learning_rate": 0.0005, "loss": 1.2800253629684448, "step": 26927 }, { "epoch": 2.122822230981474, "grad_norm": 0.48355039954185486, "learning_rate": 0.0005, "loss": 1.4857147932052612, "step": 26928 }, { "epoch": 2.1229010642491133, "grad_norm": 0.5196611285209656, "learning_rate": 0.0005, "loss": 1.3484156131744385, "step": 26929 }, { "epoch": 2.122979897516752, "grad_norm": 0.5261896252632141, "learning_rate": 0.0005, "loss": 1.3645095825195312, "step": 26930 }, { "epoch": 2.123058730784391, "grad_norm": 0.5289736390113831, "learning_rate": 0.0005, "loss": 1.5295073986053467, "step": 26931 }, { "epoch": 2.12313756405203, "grad_norm": 0.4577232301235199, "learning_rate": 0.0005, "loss": 1.1928677558898926, "step": 26932 }, { "epoch": 2.123216397319669, "grad_norm": 0.5572698712348938, "learning_rate": 0.0005, "loss": 1.2877137660980225, "step": 26933 }, { "epoch": 2.123295230587308, "grad_norm": 0.5956705808639526, "learning_rate": 0.0005, "loss": 1.586169719696045, "step": 26934 }, { "epoch": 2.1233740638549468, "grad_norm": 0.5416970252990723, "learning_rate": 0.0005, "loss": 1.3141251802444458, "step": 26935 }, { "epoch": 2.123452897122586, "grad_norm": 0.4328560531139374, "learning_rate": 0.0005, "loss": 1.1891969442367554, "step": 26936 }, { "epoch": 2.1235317303902246, "grad_norm": 0.5565972924232483, "learning_rate": 0.0005, "loss": 1.168947458267212, "step": 26937 }, { "epoch": 2.1236105636578637, "grad_norm": 0.3828931748867035, "learning_rate": 0.0005, "loss": 1.1608448028564453, "step": 26938 }, { "epoch": 2.1236893969255024, "grad_norm": 0.4984820485115051, "learning_rate": 0.0005, "loss": 1.2639834880828857, "step": 26939 }, { "epoch": 2.1237682301931415, "grad_norm": 0.42394134402275085, "learning_rate": 0.0005, "loss": 1.3215759992599487, "step": 26940 }, { "epoch": 2.1238470634607802, "grad_norm": 0.524249792098999, "learning_rate": 0.0005, "loss": 1.489059567451477, "step": 26941 }, { "epoch": 2.1239258967284194, "grad_norm": 0.4632517695426941, "learning_rate": 0.0005, "loss": 1.2624136209487915, "step": 26942 }, { "epoch": 2.1240047299960585, "grad_norm": 0.449825644493103, "learning_rate": 0.0005, "loss": 1.285505771636963, "step": 26943 }, { "epoch": 2.124083563263697, "grad_norm": 0.43691638112068176, "learning_rate": 0.0005, "loss": 1.2959870100021362, "step": 26944 }, { "epoch": 2.1241623965313363, "grad_norm": 0.4846224784851074, "learning_rate": 0.0005, "loss": 1.2098090648651123, "step": 26945 }, { "epoch": 2.124241229798975, "grad_norm": 0.43792539834976196, "learning_rate": 0.0005, "loss": 1.234505295753479, "step": 26946 }, { "epoch": 2.124320063066614, "grad_norm": 0.5866377949714661, "learning_rate": 0.0005, "loss": 1.4071686267852783, "step": 26947 }, { "epoch": 2.1243988963342533, "grad_norm": 0.6891833543777466, "learning_rate": 0.0005, "loss": 1.3874824047088623, "step": 26948 }, { "epoch": 2.124477729601892, "grad_norm": 0.5200789570808411, "learning_rate": 0.0005, "loss": 1.4339447021484375, "step": 26949 }, { "epoch": 2.124556562869531, "grad_norm": 0.5662601590156555, "learning_rate": 0.0005, "loss": 1.3141899108886719, "step": 26950 }, { "epoch": 2.12463539613717, "grad_norm": 0.4786653220653534, "learning_rate": 0.0005, "loss": 1.3176145553588867, "step": 26951 }, { "epoch": 2.124714229404809, "grad_norm": 0.5475524067878723, "learning_rate": 0.0005, "loss": 1.489964246749878, "step": 26952 }, { "epoch": 2.1247930626724476, "grad_norm": 0.5016205310821533, "learning_rate": 0.0005, "loss": 1.352194905281067, "step": 26953 }, { "epoch": 2.1248718959400867, "grad_norm": 0.5384412407875061, "learning_rate": 0.0005, "loss": 1.2800025939941406, "step": 26954 }, { "epoch": 2.124950729207726, "grad_norm": 0.47275593876838684, "learning_rate": 0.0005, "loss": 1.2841639518737793, "step": 26955 }, { "epoch": 2.1250295624753646, "grad_norm": 0.5753605961799622, "learning_rate": 0.0005, "loss": 1.461608648300171, "step": 26956 }, { "epoch": 2.1251083957430037, "grad_norm": 0.4626355767250061, "learning_rate": 0.0005, "loss": 1.3674767017364502, "step": 26957 }, { "epoch": 2.1251872290106424, "grad_norm": 0.5655148029327393, "learning_rate": 0.0005, "loss": 1.4812896251678467, "step": 26958 }, { "epoch": 2.1252660622782815, "grad_norm": 0.5034832954406738, "learning_rate": 0.0005, "loss": 1.3237247467041016, "step": 26959 }, { "epoch": 2.12534489554592, "grad_norm": 0.5430567860603333, "learning_rate": 0.0005, "loss": 1.3940571546554565, "step": 26960 }, { "epoch": 2.1254237288135593, "grad_norm": 0.49134036898612976, "learning_rate": 0.0005, "loss": 1.1752411127090454, "step": 26961 }, { "epoch": 2.1255025620811985, "grad_norm": 0.4927274286746979, "learning_rate": 0.0005, "loss": 1.2989234924316406, "step": 26962 }, { "epoch": 2.125581395348837, "grad_norm": 0.4606649577617645, "learning_rate": 0.0005, "loss": 1.2509562969207764, "step": 26963 }, { "epoch": 2.1256602286164763, "grad_norm": 0.4988281726837158, "learning_rate": 0.0005, "loss": 1.417634129524231, "step": 26964 }, { "epoch": 2.125739061884115, "grad_norm": 0.551068127155304, "learning_rate": 0.0005, "loss": 1.2636783123016357, "step": 26965 }, { "epoch": 2.125817895151754, "grad_norm": 0.4753150939941406, "learning_rate": 0.0005, "loss": 1.376847505569458, "step": 26966 }, { "epoch": 2.125896728419393, "grad_norm": 0.5105608701705933, "learning_rate": 0.0005, "loss": 1.2884838581085205, "step": 26967 }, { "epoch": 2.125975561687032, "grad_norm": 0.4338875412940979, "learning_rate": 0.0005, "loss": 1.210939645767212, "step": 26968 }, { "epoch": 2.126054394954671, "grad_norm": 0.5443215370178223, "learning_rate": 0.0005, "loss": 1.3931374549865723, "step": 26969 }, { "epoch": 2.1261332282223098, "grad_norm": 0.5981549024581909, "learning_rate": 0.0005, "loss": 1.3084867000579834, "step": 26970 }, { "epoch": 2.126212061489949, "grad_norm": 0.565924882888794, "learning_rate": 0.0005, "loss": 1.3581137657165527, "step": 26971 }, { "epoch": 2.1262908947575876, "grad_norm": 0.5460693836212158, "learning_rate": 0.0005, "loss": 1.3541842699050903, "step": 26972 }, { "epoch": 2.1263697280252267, "grad_norm": 0.5484476685523987, "learning_rate": 0.0005, "loss": 1.359999418258667, "step": 26973 }, { "epoch": 2.1264485612928654, "grad_norm": 0.5296201109886169, "learning_rate": 0.0005, "loss": 1.230364441871643, "step": 26974 }, { "epoch": 2.1265273945605045, "grad_norm": 0.49952709674835205, "learning_rate": 0.0005, "loss": 1.4783798456192017, "step": 26975 }, { "epoch": 2.1266062278281437, "grad_norm": 0.47603070735931396, "learning_rate": 0.0005, "loss": 1.3625233173370361, "step": 26976 }, { "epoch": 2.1266850610957824, "grad_norm": 0.598407506942749, "learning_rate": 0.0005, "loss": 1.3951406478881836, "step": 26977 }, { "epoch": 2.1267638943634215, "grad_norm": 0.5317953824996948, "learning_rate": 0.0005, "loss": 1.516484022140503, "step": 26978 }, { "epoch": 2.12684272763106, "grad_norm": 0.4357448220252991, "learning_rate": 0.0005, "loss": 1.3035844564437866, "step": 26979 }, { "epoch": 2.1269215608986993, "grad_norm": 0.4844651520252228, "learning_rate": 0.0005, "loss": 1.2912253141403198, "step": 26980 }, { "epoch": 2.127000394166338, "grad_norm": 0.4648030996322632, "learning_rate": 0.0005, "loss": 1.3071684837341309, "step": 26981 }, { "epoch": 2.127079227433977, "grad_norm": 0.505386471748352, "learning_rate": 0.0005, "loss": 1.3308100700378418, "step": 26982 }, { "epoch": 2.1271580607016163, "grad_norm": 0.46145349740982056, "learning_rate": 0.0005, "loss": 1.3351866006851196, "step": 26983 }, { "epoch": 2.127236893969255, "grad_norm": 0.48207786679267883, "learning_rate": 0.0005, "loss": 1.4370613098144531, "step": 26984 }, { "epoch": 2.127315727236894, "grad_norm": 0.4576849639415741, "learning_rate": 0.0005, "loss": 1.4213955402374268, "step": 26985 }, { "epoch": 2.127394560504533, "grad_norm": 1.0556507110595703, "learning_rate": 0.0005, "loss": 1.3945741653442383, "step": 26986 }, { "epoch": 2.127473393772172, "grad_norm": 5.530577182769775, "learning_rate": 0.0005, "loss": 1.4754990339279175, "step": 26987 }, { "epoch": 2.1275522270398106, "grad_norm": 0.4354069232940674, "learning_rate": 0.0005, "loss": 1.2464940547943115, "step": 26988 }, { "epoch": 2.1276310603074498, "grad_norm": 0.737655520439148, "learning_rate": 0.0005, "loss": 1.2347602844238281, "step": 26989 }, { "epoch": 2.127709893575089, "grad_norm": 0.44601818919181824, "learning_rate": 0.0005, "loss": 1.2870956659317017, "step": 26990 }, { "epoch": 2.1277887268427276, "grad_norm": 0.4022752344608307, "learning_rate": 0.0005, "loss": 1.3694465160369873, "step": 26991 }, { "epoch": 2.1278675601103667, "grad_norm": 0.6110041737556458, "learning_rate": 0.0005, "loss": 1.3337613344192505, "step": 26992 }, { "epoch": 2.1279463933780054, "grad_norm": 0.5472148060798645, "learning_rate": 0.0005, "loss": 1.308249831199646, "step": 26993 }, { "epoch": 2.1280252266456445, "grad_norm": 0.4934740662574768, "learning_rate": 0.0005, "loss": 1.594772458076477, "step": 26994 }, { "epoch": 2.128104059913283, "grad_norm": 0.39603790640830994, "learning_rate": 0.0005, "loss": 1.3439826965332031, "step": 26995 }, { "epoch": 2.1281828931809224, "grad_norm": 0.5792034864425659, "learning_rate": 0.0005, "loss": 1.4118293523788452, "step": 26996 }, { "epoch": 2.1282617264485615, "grad_norm": 0.673384964466095, "learning_rate": 0.0005, "loss": 1.3508942127227783, "step": 26997 }, { "epoch": 2.1283405597162, "grad_norm": 0.5437629222869873, "learning_rate": 0.0005, "loss": 1.270000696182251, "step": 26998 }, { "epoch": 2.1284193929838393, "grad_norm": 0.5632542967796326, "learning_rate": 0.0005, "loss": 1.3887221813201904, "step": 26999 }, { "epoch": 2.128498226251478, "grad_norm": 1.089429259300232, "learning_rate": 0.0005, "loss": 1.5657281875610352, "step": 27000 }, { "epoch": 2.128577059519117, "grad_norm": 0.4617481827735901, "learning_rate": 0.0005, "loss": 1.293982744216919, "step": 27001 }, { "epoch": 2.128655892786756, "grad_norm": 0.49639609456062317, "learning_rate": 0.0005, "loss": 1.1864080429077148, "step": 27002 }, { "epoch": 2.128734726054395, "grad_norm": 0.5471315979957581, "learning_rate": 0.0005, "loss": 1.429617166519165, "step": 27003 }, { "epoch": 2.128813559322034, "grad_norm": 0.862116277217865, "learning_rate": 0.0005, "loss": 1.5719764232635498, "step": 27004 }, { "epoch": 2.1288923925896728, "grad_norm": 0.41654884815216064, "learning_rate": 0.0005, "loss": 1.1221438646316528, "step": 27005 }, { "epoch": 2.128971225857312, "grad_norm": 0.5327066779136658, "learning_rate": 0.0005, "loss": 1.2316876649856567, "step": 27006 }, { "epoch": 2.1290500591249506, "grad_norm": 0.5242629647254944, "learning_rate": 0.0005, "loss": 1.3479485511779785, "step": 27007 }, { "epoch": 2.1291288923925897, "grad_norm": 0.43429914116859436, "learning_rate": 0.0005, "loss": 1.1293973922729492, "step": 27008 }, { "epoch": 2.1292077256602284, "grad_norm": 0.6019238233566284, "learning_rate": 0.0005, "loss": 1.2818410396575928, "step": 27009 }, { "epoch": 2.1292865589278676, "grad_norm": 0.5953699350357056, "learning_rate": 0.0005, "loss": 1.1336700916290283, "step": 27010 }, { "epoch": 2.1293653921955067, "grad_norm": 0.443296879529953, "learning_rate": 0.0005, "loss": 1.2101523876190186, "step": 27011 }, { "epoch": 2.1294442254631454, "grad_norm": 0.566448450088501, "learning_rate": 0.0005, "loss": 1.2330379486083984, "step": 27012 }, { "epoch": 2.1295230587307845, "grad_norm": 0.461503267288208, "learning_rate": 0.0005, "loss": 1.181344985961914, "step": 27013 }, { "epoch": 2.129601891998423, "grad_norm": 0.725669801235199, "learning_rate": 0.0005, "loss": 1.412185788154602, "step": 27014 }, { "epoch": 2.1296807252660623, "grad_norm": 0.4856896996498108, "learning_rate": 0.0005, "loss": 1.3086763620376587, "step": 27015 }, { "epoch": 2.129759558533701, "grad_norm": 0.552272379398346, "learning_rate": 0.0005, "loss": 1.4907821416854858, "step": 27016 }, { "epoch": 2.12983839180134, "grad_norm": 0.416608601808548, "learning_rate": 0.0005, "loss": 1.466355562210083, "step": 27017 }, { "epoch": 2.1299172250689793, "grad_norm": 0.612754225730896, "learning_rate": 0.0005, "loss": 1.3553330898284912, "step": 27018 }, { "epoch": 2.129996058336618, "grad_norm": 0.4012983739376068, "learning_rate": 0.0005, "loss": 1.2874550819396973, "step": 27019 }, { "epoch": 2.130074891604257, "grad_norm": 0.4816524088382721, "learning_rate": 0.0005, "loss": 1.3475018739700317, "step": 27020 }, { "epoch": 2.130153724871896, "grad_norm": 0.5637679696083069, "learning_rate": 0.0005, "loss": 1.2615113258361816, "step": 27021 }, { "epoch": 2.130232558139535, "grad_norm": 0.4343198537826538, "learning_rate": 0.0005, "loss": 1.3279072046279907, "step": 27022 }, { "epoch": 2.1303113914071736, "grad_norm": 0.6616766452789307, "learning_rate": 0.0005, "loss": 1.2455098628997803, "step": 27023 }, { "epoch": 2.1303902246748128, "grad_norm": 0.4436037242412567, "learning_rate": 0.0005, "loss": 1.327303409576416, "step": 27024 }, { "epoch": 2.130469057942452, "grad_norm": 0.4900820851325989, "learning_rate": 0.0005, "loss": 1.0398389101028442, "step": 27025 }, { "epoch": 2.1305478912100906, "grad_norm": 0.5417866706848145, "learning_rate": 0.0005, "loss": 1.3514422178268433, "step": 27026 }, { "epoch": 2.1306267244777297, "grad_norm": 0.4616592824459076, "learning_rate": 0.0005, "loss": 1.251145601272583, "step": 27027 }, { "epoch": 2.1307055577453684, "grad_norm": 0.46202757954597473, "learning_rate": 0.0005, "loss": 1.2096951007843018, "step": 27028 }, { "epoch": 2.1307843910130075, "grad_norm": 0.4627648591995239, "learning_rate": 0.0005, "loss": 1.100036382675171, "step": 27029 }, { "epoch": 2.1308632242806462, "grad_norm": 0.5606397986412048, "learning_rate": 0.0005, "loss": 1.3126778602600098, "step": 27030 }, { "epoch": 2.1309420575482854, "grad_norm": 0.4844619035720825, "learning_rate": 0.0005, "loss": 1.3371400833129883, "step": 27031 }, { "epoch": 2.1310208908159245, "grad_norm": 0.4644457697868347, "learning_rate": 0.0005, "loss": 1.3666965961456299, "step": 27032 }, { "epoch": 2.131099724083563, "grad_norm": 0.4696897268295288, "learning_rate": 0.0005, "loss": 1.418176531791687, "step": 27033 }, { "epoch": 2.1311785573512023, "grad_norm": 0.6010242700576782, "learning_rate": 0.0005, "loss": 1.490004301071167, "step": 27034 }, { "epoch": 2.131257390618841, "grad_norm": 0.44139885902404785, "learning_rate": 0.0005, "loss": 1.1076805591583252, "step": 27035 }, { "epoch": 2.13133622388648, "grad_norm": 0.4644111096858978, "learning_rate": 0.0005, "loss": 1.3096399307250977, "step": 27036 }, { "epoch": 2.131415057154119, "grad_norm": 0.4182441830635071, "learning_rate": 0.0005, "loss": 1.265188217163086, "step": 27037 }, { "epoch": 2.131493890421758, "grad_norm": 0.5551152229309082, "learning_rate": 0.0005, "loss": 0.9832320213317871, "step": 27038 }, { "epoch": 2.131572723689397, "grad_norm": 0.5455562472343445, "learning_rate": 0.0005, "loss": 1.3040374517440796, "step": 27039 }, { "epoch": 2.131651556957036, "grad_norm": 0.5400180816650391, "learning_rate": 0.0005, "loss": 1.4299819469451904, "step": 27040 }, { "epoch": 2.131730390224675, "grad_norm": 0.557996392250061, "learning_rate": 0.0005, "loss": 1.4147472381591797, "step": 27041 }, { "epoch": 2.1318092234923136, "grad_norm": 0.6126329898834229, "learning_rate": 0.0005, "loss": 1.5585174560546875, "step": 27042 }, { "epoch": 2.1318880567599527, "grad_norm": 0.4942045211791992, "learning_rate": 0.0005, "loss": 1.472877025604248, "step": 27043 }, { "epoch": 2.1319668900275914, "grad_norm": 0.4894773066043854, "learning_rate": 0.0005, "loss": 1.4545786380767822, "step": 27044 }, { "epoch": 2.1320457232952306, "grad_norm": 0.4751163721084595, "learning_rate": 0.0005, "loss": 1.2364301681518555, "step": 27045 }, { "epoch": 2.1321245565628697, "grad_norm": 0.5364949107170105, "learning_rate": 0.0005, "loss": 1.3293730020523071, "step": 27046 }, { "epoch": 2.1322033898305084, "grad_norm": 0.5983586311340332, "learning_rate": 0.0005, "loss": 1.3426637649536133, "step": 27047 }, { "epoch": 2.1322822230981475, "grad_norm": 0.47397327423095703, "learning_rate": 0.0005, "loss": 1.4031862020492554, "step": 27048 }, { "epoch": 2.132361056365786, "grad_norm": 0.565634548664093, "learning_rate": 0.0005, "loss": 1.3312304019927979, "step": 27049 }, { "epoch": 2.1324398896334253, "grad_norm": 0.5121806859970093, "learning_rate": 0.0005, "loss": 1.201064109802246, "step": 27050 }, { "epoch": 2.1325187229010645, "grad_norm": 0.5406442880630493, "learning_rate": 0.0005, "loss": 1.3153040409088135, "step": 27051 }, { "epoch": 2.132597556168703, "grad_norm": 0.44489750266075134, "learning_rate": 0.0005, "loss": 1.2117061614990234, "step": 27052 }, { "epoch": 2.1326763894363423, "grad_norm": 0.47096917033195496, "learning_rate": 0.0005, "loss": 1.310539960861206, "step": 27053 }, { "epoch": 2.132755222703981, "grad_norm": 0.5449689030647278, "learning_rate": 0.0005, "loss": 1.3390063047409058, "step": 27054 }, { "epoch": 2.13283405597162, "grad_norm": 0.49742430448532104, "learning_rate": 0.0005, "loss": 1.3071417808532715, "step": 27055 }, { "epoch": 2.132912889239259, "grad_norm": 0.48463183641433716, "learning_rate": 0.0005, "loss": 1.3044092655181885, "step": 27056 }, { "epoch": 2.132991722506898, "grad_norm": 0.4717086851596832, "learning_rate": 0.0005, "loss": 1.1914408206939697, "step": 27057 }, { "epoch": 2.1330705557745366, "grad_norm": 0.49107682704925537, "learning_rate": 0.0005, "loss": 1.2213863134384155, "step": 27058 }, { "epoch": 2.1331493890421758, "grad_norm": 0.4217834770679474, "learning_rate": 0.0005, "loss": 1.4307682514190674, "step": 27059 }, { "epoch": 2.133228222309815, "grad_norm": 0.4821282923221588, "learning_rate": 0.0005, "loss": 1.4845658540725708, "step": 27060 }, { "epoch": 2.1333070555774536, "grad_norm": 0.46238580346107483, "learning_rate": 0.0005, "loss": 1.1832597255706787, "step": 27061 }, { "epoch": 2.1333858888450927, "grad_norm": 0.5055215358734131, "learning_rate": 0.0005, "loss": 1.479938268661499, "step": 27062 }, { "epoch": 2.1334647221127314, "grad_norm": 0.5425200462341309, "learning_rate": 0.0005, "loss": 1.1169097423553467, "step": 27063 }, { "epoch": 2.1335435553803705, "grad_norm": 0.5016424655914307, "learning_rate": 0.0005, "loss": 1.2224411964416504, "step": 27064 }, { "epoch": 2.1336223886480097, "grad_norm": 0.4380292296409607, "learning_rate": 0.0005, "loss": 1.287719488143921, "step": 27065 }, { "epoch": 2.1337012219156484, "grad_norm": 0.49015456438064575, "learning_rate": 0.0005, "loss": 1.2490956783294678, "step": 27066 }, { "epoch": 2.1337800551832875, "grad_norm": 0.5561738610267639, "learning_rate": 0.0005, "loss": 1.4148027896881104, "step": 27067 }, { "epoch": 2.133858888450926, "grad_norm": 0.4694425165653229, "learning_rate": 0.0005, "loss": 1.245600938796997, "step": 27068 }, { "epoch": 2.1339377217185653, "grad_norm": 0.6288057565689087, "learning_rate": 0.0005, "loss": 1.4301402568817139, "step": 27069 }, { "epoch": 2.134016554986204, "grad_norm": 0.5481507778167725, "learning_rate": 0.0005, "loss": 1.4452250003814697, "step": 27070 }, { "epoch": 2.134095388253843, "grad_norm": 0.6769814491271973, "learning_rate": 0.0005, "loss": 1.5578548908233643, "step": 27071 }, { "epoch": 2.134174221521482, "grad_norm": 0.48888033628463745, "learning_rate": 0.0005, "loss": 1.285015344619751, "step": 27072 }, { "epoch": 2.134253054789121, "grad_norm": 0.6068102717399597, "learning_rate": 0.0005, "loss": 1.2647705078125, "step": 27073 }, { "epoch": 2.13433188805676, "grad_norm": 0.5293742418289185, "learning_rate": 0.0005, "loss": 1.316286563873291, "step": 27074 }, { "epoch": 2.134410721324399, "grad_norm": 0.5213347673416138, "learning_rate": 0.0005, "loss": 1.440390706062317, "step": 27075 }, { "epoch": 2.134489554592038, "grad_norm": 0.4582940936088562, "learning_rate": 0.0005, "loss": 1.4732365608215332, "step": 27076 }, { "epoch": 2.1345683878596766, "grad_norm": 0.5863044261932373, "learning_rate": 0.0005, "loss": 1.4710837602615356, "step": 27077 }, { "epoch": 2.1346472211273158, "grad_norm": 0.46854376792907715, "learning_rate": 0.0005, "loss": 1.218604326248169, "step": 27078 }, { "epoch": 2.134726054394955, "grad_norm": 0.5488191246986389, "learning_rate": 0.0005, "loss": 1.415503978729248, "step": 27079 }, { "epoch": 2.1348048876625936, "grad_norm": 0.6744683384895325, "learning_rate": 0.0005, "loss": 1.5219550132751465, "step": 27080 }, { "epoch": 2.1348837209302327, "grad_norm": 0.44983625411987305, "learning_rate": 0.0005, "loss": 1.3507390022277832, "step": 27081 }, { "epoch": 2.1349625541978714, "grad_norm": 0.5367063283920288, "learning_rate": 0.0005, "loss": 1.373473882675171, "step": 27082 }, { "epoch": 2.1350413874655105, "grad_norm": 0.6915757060050964, "learning_rate": 0.0005, "loss": 1.341470718383789, "step": 27083 }, { "epoch": 2.135120220733149, "grad_norm": 0.5916113257408142, "learning_rate": 0.0005, "loss": 1.5482949018478394, "step": 27084 }, { "epoch": 2.1351990540007884, "grad_norm": 0.469958633184433, "learning_rate": 0.0005, "loss": 1.2099733352661133, "step": 27085 }, { "epoch": 2.135277887268427, "grad_norm": 0.9529886841773987, "learning_rate": 0.0005, "loss": 1.526787519454956, "step": 27086 }, { "epoch": 2.135356720536066, "grad_norm": 0.4403456747531891, "learning_rate": 0.0005, "loss": 1.2129018306732178, "step": 27087 }, { "epoch": 2.1354355538037053, "grad_norm": 1.2060317993164062, "learning_rate": 0.0005, "loss": 1.3653010129928589, "step": 27088 }, { "epoch": 2.135514387071344, "grad_norm": 0.5236196517944336, "learning_rate": 0.0005, "loss": 1.4031890630722046, "step": 27089 }, { "epoch": 2.135593220338983, "grad_norm": 0.5466326475143433, "learning_rate": 0.0005, "loss": 1.311643123626709, "step": 27090 }, { "epoch": 2.135672053606622, "grad_norm": 0.45260390639305115, "learning_rate": 0.0005, "loss": 1.1564913988113403, "step": 27091 }, { "epoch": 2.135750886874261, "grad_norm": 0.5846644639968872, "learning_rate": 0.0005, "loss": 1.469173550605774, "step": 27092 }, { "epoch": 2.1358297201419, "grad_norm": 0.5698908567428589, "learning_rate": 0.0005, "loss": 1.3946222066879272, "step": 27093 }, { "epoch": 2.135908553409539, "grad_norm": 0.623528778553009, "learning_rate": 0.0005, "loss": 1.231994867324829, "step": 27094 }, { "epoch": 2.135987386677178, "grad_norm": 0.6672744750976562, "learning_rate": 0.0005, "loss": 1.5561046600341797, "step": 27095 }, { "epoch": 2.1360662199448166, "grad_norm": 0.34443143010139465, "learning_rate": 0.0005, "loss": 1.0760639905929565, "step": 27096 }, { "epoch": 2.1361450532124557, "grad_norm": 0.5476701259613037, "learning_rate": 0.0005, "loss": 1.339896559715271, "step": 27097 }, { "epoch": 2.1362238864800944, "grad_norm": 0.5183145403862, "learning_rate": 0.0005, "loss": 1.2075285911560059, "step": 27098 }, { "epoch": 2.1363027197477336, "grad_norm": 0.49402859807014465, "learning_rate": 0.0005, "loss": 1.1700053215026855, "step": 27099 }, { "epoch": 2.1363815530153727, "grad_norm": 0.5412951111793518, "learning_rate": 0.0005, "loss": 1.3990669250488281, "step": 27100 }, { "epoch": 2.1364603862830114, "grad_norm": 0.38927793502807617, "learning_rate": 0.0005, "loss": 1.006485939025879, "step": 27101 }, { "epoch": 2.1365392195506505, "grad_norm": 0.5951307415962219, "learning_rate": 0.0005, "loss": 1.2375092506408691, "step": 27102 }, { "epoch": 2.136618052818289, "grad_norm": 0.4968469738960266, "learning_rate": 0.0005, "loss": 1.3349915742874146, "step": 27103 }, { "epoch": 2.1366968860859283, "grad_norm": 0.49027401208877563, "learning_rate": 0.0005, "loss": 1.2251505851745605, "step": 27104 }, { "epoch": 2.136775719353567, "grad_norm": 0.5367497205734253, "learning_rate": 0.0005, "loss": 1.525101900100708, "step": 27105 }, { "epoch": 2.136854552621206, "grad_norm": 0.35696589946746826, "learning_rate": 0.0005, "loss": 1.2399507761001587, "step": 27106 }, { "epoch": 2.1369333858888453, "grad_norm": 0.4419367015361786, "learning_rate": 0.0005, "loss": 1.2074501514434814, "step": 27107 }, { "epoch": 2.137012219156484, "grad_norm": 0.4672611355781555, "learning_rate": 0.0005, "loss": 1.3889245986938477, "step": 27108 }, { "epoch": 2.137091052424123, "grad_norm": 0.48583921790122986, "learning_rate": 0.0005, "loss": 1.3298404216766357, "step": 27109 }, { "epoch": 2.137169885691762, "grad_norm": 0.5512411594390869, "learning_rate": 0.0005, "loss": 1.256309986114502, "step": 27110 }, { "epoch": 2.137248718959401, "grad_norm": 0.4847561717033386, "learning_rate": 0.0005, "loss": 1.4447368383407593, "step": 27111 }, { "epoch": 2.1373275522270396, "grad_norm": 0.4886783957481384, "learning_rate": 0.0005, "loss": 1.4051836729049683, "step": 27112 }, { "epoch": 2.1374063854946788, "grad_norm": 0.4577326476573944, "learning_rate": 0.0005, "loss": 1.2689759731292725, "step": 27113 }, { "epoch": 2.137485218762318, "grad_norm": 0.46764764189720154, "learning_rate": 0.0005, "loss": 1.1652989387512207, "step": 27114 }, { "epoch": 2.1375640520299566, "grad_norm": 0.5087417364120483, "learning_rate": 0.0005, "loss": 1.4867630004882812, "step": 27115 }, { "epoch": 2.1376428852975957, "grad_norm": 0.5707322359085083, "learning_rate": 0.0005, "loss": 1.4821957349777222, "step": 27116 }, { "epoch": 2.1377217185652344, "grad_norm": 0.4915529787540436, "learning_rate": 0.0005, "loss": 1.2952589988708496, "step": 27117 }, { "epoch": 2.1378005518328735, "grad_norm": 0.41171950101852417, "learning_rate": 0.0005, "loss": 1.4775111675262451, "step": 27118 }, { "epoch": 2.1378793851005122, "grad_norm": 0.4581865072250366, "learning_rate": 0.0005, "loss": 1.1615535020828247, "step": 27119 }, { "epoch": 2.1379582183681514, "grad_norm": 0.9273145794868469, "learning_rate": 0.0005, "loss": 1.3795692920684814, "step": 27120 }, { "epoch": 2.1380370516357905, "grad_norm": 0.7219826579093933, "learning_rate": 0.0005, "loss": 1.2763937711715698, "step": 27121 }, { "epoch": 2.138115884903429, "grad_norm": 0.4509213864803314, "learning_rate": 0.0005, "loss": 1.381944179534912, "step": 27122 }, { "epoch": 2.1381947181710683, "grad_norm": 0.454377681016922, "learning_rate": 0.0005, "loss": 1.293838381767273, "step": 27123 }, { "epoch": 2.138273551438707, "grad_norm": 0.4428834319114685, "learning_rate": 0.0005, "loss": 1.1710190773010254, "step": 27124 }, { "epoch": 2.138352384706346, "grad_norm": 0.5020681023597717, "learning_rate": 0.0005, "loss": 1.3047834634780884, "step": 27125 }, { "epoch": 2.138431217973985, "grad_norm": 0.4913986325263977, "learning_rate": 0.0005, "loss": 1.2694294452667236, "step": 27126 }, { "epoch": 2.138510051241624, "grad_norm": 0.5032405853271484, "learning_rate": 0.0005, "loss": 1.3310794830322266, "step": 27127 }, { "epoch": 2.138588884509263, "grad_norm": 0.5145487189292908, "learning_rate": 0.0005, "loss": 1.3406883478164673, "step": 27128 }, { "epoch": 2.138667717776902, "grad_norm": 0.5814300775527954, "learning_rate": 0.0005, "loss": 1.269892930984497, "step": 27129 }, { "epoch": 2.138746551044541, "grad_norm": 0.5466656684875488, "learning_rate": 0.0005, "loss": 1.5755465030670166, "step": 27130 }, { "epoch": 2.1388253843121796, "grad_norm": 0.5792414546012878, "learning_rate": 0.0005, "loss": 1.4072203636169434, "step": 27131 }, { "epoch": 2.1389042175798187, "grad_norm": 0.5087300539016724, "learning_rate": 0.0005, "loss": 1.5055427551269531, "step": 27132 }, { "epoch": 2.1389830508474574, "grad_norm": 0.40998923778533936, "learning_rate": 0.0005, "loss": 1.1443673372268677, "step": 27133 }, { "epoch": 2.1390618841150966, "grad_norm": 0.5620078444480896, "learning_rate": 0.0005, "loss": 1.4684169292449951, "step": 27134 }, { "epoch": 2.1391407173827357, "grad_norm": 0.519406795501709, "learning_rate": 0.0005, "loss": 1.2450894117355347, "step": 27135 }, { "epoch": 2.1392195506503744, "grad_norm": 0.4896411597728729, "learning_rate": 0.0005, "loss": 1.2347707748413086, "step": 27136 }, { "epoch": 2.1392983839180135, "grad_norm": 0.4115515947341919, "learning_rate": 0.0005, "loss": 1.1946943998336792, "step": 27137 }, { "epoch": 2.139377217185652, "grad_norm": 0.4097763001918793, "learning_rate": 0.0005, "loss": 1.2937266826629639, "step": 27138 }, { "epoch": 2.1394560504532913, "grad_norm": 1.0539695024490356, "learning_rate": 0.0005, "loss": 1.3997621536254883, "step": 27139 }, { "epoch": 2.13953488372093, "grad_norm": 0.5296152234077454, "learning_rate": 0.0005, "loss": 1.3856873512268066, "step": 27140 }, { "epoch": 2.139613716988569, "grad_norm": 0.4987337291240692, "learning_rate": 0.0005, "loss": 1.470468282699585, "step": 27141 }, { "epoch": 2.1396925502562083, "grad_norm": 0.5745412111282349, "learning_rate": 0.0005, "loss": 1.3385334014892578, "step": 27142 }, { "epoch": 2.139771383523847, "grad_norm": 0.5113898515701294, "learning_rate": 0.0005, "loss": 1.3759686946868896, "step": 27143 }, { "epoch": 2.139850216791486, "grad_norm": 0.4892486035823822, "learning_rate": 0.0005, "loss": 1.2770404815673828, "step": 27144 }, { "epoch": 2.139929050059125, "grad_norm": 0.42706361413002014, "learning_rate": 0.0005, "loss": 1.4186694622039795, "step": 27145 }, { "epoch": 2.140007883326764, "grad_norm": 0.6058576107025146, "learning_rate": 0.0005, "loss": 1.4478025436401367, "step": 27146 }, { "epoch": 2.1400867165944026, "grad_norm": 0.5283636450767517, "learning_rate": 0.0005, "loss": 1.3664963245391846, "step": 27147 }, { "epoch": 2.1401655498620418, "grad_norm": 0.4033110439777374, "learning_rate": 0.0005, "loss": 1.2257342338562012, "step": 27148 }, { "epoch": 2.140244383129681, "grad_norm": 0.44847893714904785, "learning_rate": 0.0005, "loss": 1.2037475109100342, "step": 27149 }, { "epoch": 2.1403232163973196, "grad_norm": 0.42618417739868164, "learning_rate": 0.0005, "loss": 1.1235570907592773, "step": 27150 }, { "epoch": 2.1404020496649587, "grad_norm": 0.43730852007865906, "learning_rate": 0.0005, "loss": 1.332943320274353, "step": 27151 }, { "epoch": 2.1404808829325974, "grad_norm": 0.45944100618362427, "learning_rate": 0.0005, "loss": 1.1926020383834839, "step": 27152 }, { "epoch": 2.1405597162002366, "grad_norm": 0.5869740843772888, "learning_rate": 0.0005, "loss": 1.237208604812622, "step": 27153 }, { "epoch": 2.1406385494678752, "grad_norm": 0.653548002243042, "learning_rate": 0.0005, "loss": 1.4547317028045654, "step": 27154 }, { "epoch": 2.1407173827355144, "grad_norm": 0.49579328298568726, "learning_rate": 0.0005, "loss": 1.4066027402877808, "step": 27155 }, { "epoch": 2.1407962160031535, "grad_norm": 0.4479902684688568, "learning_rate": 0.0005, "loss": 1.287057876586914, "step": 27156 }, { "epoch": 2.140875049270792, "grad_norm": 0.38254493474960327, "learning_rate": 0.0005, "loss": 1.300372838973999, "step": 27157 }, { "epoch": 2.1409538825384313, "grad_norm": 0.5261876583099365, "learning_rate": 0.0005, "loss": 1.4925856590270996, "step": 27158 }, { "epoch": 2.14103271580607, "grad_norm": 0.562926173210144, "learning_rate": 0.0005, "loss": 1.2886619567871094, "step": 27159 }, { "epoch": 2.141111549073709, "grad_norm": 0.4376583695411682, "learning_rate": 0.0005, "loss": 1.1569509506225586, "step": 27160 }, { "epoch": 2.141190382341348, "grad_norm": 0.530860424041748, "learning_rate": 0.0005, "loss": 1.3898532390594482, "step": 27161 }, { "epoch": 2.141269215608987, "grad_norm": 0.52207350730896, "learning_rate": 0.0005, "loss": 1.4984610080718994, "step": 27162 }, { "epoch": 2.141348048876626, "grad_norm": 0.6566421389579773, "learning_rate": 0.0005, "loss": 1.2247143983840942, "step": 27163 }, { "epoch": 2.141426882144265, "grad_norm": 0.47160443663597107, "learning_rate": 0.0005, "loss": 1.5061018466949463, "step": 27164 }, { "epoch": 2.141505715411904, "grad_norm": 0.5185300707817078, "learning_rate": 0.0005, "loss": 1.487655520439148, "step": 27165 }, { "epoch": 2.1415845486795426, "grad_norm": 0.5117151737213135, "learning_rate": 0.0005, "loss": 1.4105916023254395, "step": 27166 }, { "epoch": 2.1416633819471818, "grad_norm": 0.6783576011657715, "learning_rate": 0.0005, "loss": 1.38258957862854, "step": 27167 }, { "epoch": 2.141742215214821, "grad_norm": 0.4659268260002136, "learning_rate": 0.0005, "loss": 1.2786146402359009, "step": 27168 }, { "epoch": 2.1418210484824596, "grad_norm": 0.4982964098453522, "learning_rate": 0.0005, "loss": 1.3712867498397827, "step": 27169 }, { "epoch": 2.1418998817500987, "grad_norm": 0.572752058506012, "learning_rate": 0.0005, "loss": 1.441123127937317, "step": 27170 }, { "epoch": 2.1419787150177374, "grad_norm": 0.5250593423843384, "learning_rate": 0.0005, "loss": 1.3919764757156372, "step": 27171 }, { "epoch": 2.1420575482853765, "grad_norm": 0.43675893545150757, "learning_rate": 0.0005, "loss": 1.3236504793167114, "step": 27172 }, { "epoch": 2.1421363815530152, "grad_norm": 0.4669401943683624, "learning_rate": 0.0005, "loss": 1.3302714824676514, "step": 27173 }, { "epoch": 2.1422152148206544, "grad_norm": 0.5749316811561584, "learning_rate": 0.0005, "loss": 1.3788068294525146, "step": 27174 }, { "epoch": 2.142294048088293, "grad_norm": 1.9429774284362793, "learning_rate": 0.0005, "loss": 1.4274399280548096, "step": 27175 }, { "epoch": 2.142372881355932, "grad_norm": 0.7269758582115173, "learning_rate": 0.0005, "loss": 1.4323923587799072, "step": 27176 }, { "epoch": 2.1424517146235713, "grad_norm": 0.568446934223175, "learning_rate": 0.0005, "loss": 1.3033665418624878, "step": 27177 }, { "epoch": 2.14253054789121, "grad_norm": 0.5572582483291626, "learning_rate": 0.0005, "loss": 1.313852071762085, "step": 27178 }, { "epoch": 2.142609381158849, "grad_norm": 0.5373883247375488, "learning_rate": 0.0005, "loss": 1.296433687210083, "step": 27179 }, { "epoch": 2.142688214426488, "grad_norm": 0.574622631072998, "learning_rate": 0.0005, "loss": 1.5646262168884277, "step": 27180 }, { "epoch": 2.142767047694127, "grad_norm": 0.45949867367744446, "learning_rate": 0.0005, "loss": 1.1966402530670166, "step": 27181 }, { "epoch": 2.142845880961766, "grad_norm": 0.5099313855171204, "learning_rate": 0.0005, "loss": 1.330974817276001, "step": 27182 }, { "epoch": 2.142924714229405, "grad_norm": 0.5005158185958862, "learning_rate": 0.0005, "loss": 1.2582905292510986, "step": 27183 }, { "epoch": 2.143003547497044, "grad_norm": 0.5459979772567749, "learning_rate": 0.0005, "loss": 1.5336034297943115, "step": 27184 }, { "epoch": 2.1430823807646826, "grad_norm": 0.5161260962486267, "learning_rate": 0.0005, "loss": 1.1982336044311523, "step": 27185 }, { "epoch": 2.1431612140323217, "grad_norm": 0.509185254573822, "learning_rate": 0.0005, "loss": 1.2848716974258423, "step": 27186 }, { "epoch": 2.1432400472999604, "grad_norm": 0.5853108167648315, "learning_rate": 0.0005, "loss": 1.2804449796676636, "step": 27187 }, { "epoch": 2.1433188805675996, "grad_norm": 0.7551941871643066, "learning_rate": 0.0005, "loss": 1.4252350330352783, "step": 27188 }, { "epoch": 2.1433977138352382, "grad_norm": 0.5087641477584839, "learning_rate": 0.0005, "loss": 1.496720790863037, "step": 27189 }, { "epoch": 2.1434765471028774, "grad_norm": 0.5528695583343506, "learning_rate": 0.0005, "loss": 1.2473175525665283, "step": 27190 }, { "epoch": 2.1435553803705165, "grad_norm": 0.561103880405426, "learning_rate": 0.0005, "loss": 1.4234994649887085, "step": 27191 }, { "epoch": 2.143634213638155, "grad_norm": 0.5432384610176086, "learning_rate": 0.0005, "loss": 1.3503851890563965, "step": 27192 }, { "epoch": 2.1437130469057943, "grad_norm": 0.4379121959209442, "learning_rate": 0.0005, "loss": 1.3475027084350586, "step": 27193 }, { "epoch": 2.143791880173433, "grad_norm": 0.505054235458374, "learning_rate": 0.0005, "loss": 1.3702373504638672, "step": 27194 }, { "epoch": 2.143870713441072, "grad_norm": 0.5432612895965576, "learning_rate": 0.0005, "loss": 1.2814326286315918, "step": 27195 }, { "epoch": 2.1439495467087113, "grad_norm": 0.5011560916900635, "learning_rate": 0.0005, "loss": 1.6009114980697632, "step": 27196 }, { "epoch": 2.14402837997635, "grad_norm": 0.6003344655036926, "learning_rate": 0.0005, "loss": 1.26027250289917, "step": 27197 }, { "epoch": 2.144107213243989, "grad_norm": 0.48553982377052307, "learning_rate": 0.0005, "loss": 1.4610145092010498, "step": 27198 }, { "epoch": 2.144186046511628, "grad_norm": 0.5620834231376648, "learning_rate": 0.0005, "loss": 1.3035780191421509, "step": 27199 }, { "epoch": 2.144264879779267, "grad_norm": 0.5279732346534729, "learning_rate": 0.0005, "loss": 1.3345427513122559, "step": 27200 }, { "epoch": 2.1443437130469056, "grad_norm": 0.795960545539856, "learning_rate": 0.0005, "loss": 1.547668695449829, "step": 27201 }, { "epoch": 2.1444225463145448, "grad_norm": 0.6582344770431519, "learning_rate": 0.0005, "loss": 1.3770651817321777, "step": 27202 }, { "epoch": 2.1445013795821835, "grad_norm": 0.4409809708595276, "learning_rate": 0.0005, "loss": 1.3290164470672607, "step": 27203 }, { "epoch": 2.1445802128498226, "grad_norm": 0.5213303565979004, "learning_rate": 0.0005, "loss": 1.3364888429641724, "step": 27204 }, { "epoch": 2.1446590461174617, "grad_norm": 0.5539698004722595, "learning_rate": 0.0005, "loss": 1.236280918121338, "step": 27205 }, { "epoch": 2.1447378793851004, "grad_norm": 0.5229923129081726, "learning_rate": 0.0005, "loss": 1.4161489009857178, "step": 27206 }, { "epoch": 2.1448167126527395, "grad_norm": 0.5290010571479797, "learning_rate": 0.0005, "loss": 1.3834705352783203, "step": 27207 }, { "epoch": 2.1448955459203782, "grad_norm": 0.6437997817993164, "learning_rate": 0.0005, "loss": 1.247004508972168, "step": 27208 }, { "epoch": 2.1449743791880174, "grad_norm": 0.4546716511249542, "learning_rate": 0.0005, "loss": 1.3674852848052979, "step": 27209 }, { "epoch": 2.1450532124556565, "grad_norm": 0.9626653790473938, "learning_rate": 0.0005, "loss": 1.533980131149292, "step": 27210 }, { "epoch": 2.145132045723295, "grad_norm": 0.5899215936660767, "learning_rate": 0.0005, "loss": 1.355034589767456, "step": 27211 }, { "epoch": 2.1452108789909343, "grad_norm": 0.5015677213668823, "learning_rate": 0.0005, "loss": 1.2732884883880615, "step": 27212 }, { "epoch": 2.145289712258573, "grad_norm": 0.45877382159233093, "learning_rate": 0.0005, "loss": 1.3668285608291626, "step": 27213 }, { "epoch": 2.145368545526212, "grad_norm": 0.6109424829483032, "learning_rate": 0.0005, "loss": 1.385956883430481, "step": 27214 }, { "epoch": 2.145447378793851, "grad_norm": 0.45786330103874207, "learning_rate": 0.0005, "loss": 1.3011581897735596, "step": 27215 }, { "epoch": 2.14552621206149, "grad_norm": 0.514470636844635, "learning_rate": 0.0005, "loss": 1.312092900276184, "step": 27216 }, { "epoch": 2.1456050453291287, "grad_norm": 0.46047574281692505, "learning_rate": 0.0005, "loss": 1.1165432929992676, "step": 27217 }, { "epoch": 2.145683878596768, "grad_norm": 0.4392514228820801, "learning_rate": 0.0005, "loss": 1.356337070465088, "step": 27218 }, { "epoch": 2.145762711864407, "grad_norm": 0.4414447546005249, "learning_rate": 0.0005, "loss": 1.0968353748321533, "step": 27219 }, { "epoch": 2.1458415451320456, "grad_norm": 0.4955121576786041, "learning_rate": 0.0005, "loss": 1.256683588027954, "step": 27220 }, { "epoch": 2.1459203783996847, "grad_norm": 0.5465173721313477, "learning_rate": 0.0005, "loss": 1.4434043169021606, "step": 27221 }, { "epoch": 2.1459992116673234, "grad_norm": 0.5355545878410339, "learning_rate": 0.0005, "loss": 1.4082268476486206, "step": 27222 }, { "epoch": 2.1460780449349626, "grad_norm": 0.4693126976490021, "learning_rate": 0.0005, "loss": 1.148498773574829, "step": 27223 }, { "epoch": 2.1461568782026017, "grad_norm": 0.46821653842926025, "learning_rate": 0.0005, "loss": 1.4796087741851807, "step": 27224 }, { "epoch": 2.1462357114702404, "grad_norm": 0.4490322172641754, "learning_rate": 0.0005, "loss": 1.2308671474456787, "step": 27225 }, { "epoch": 2.1463145447378795, "grad_norm": 0.5879162549972534, "learning_rate": 0.0005, "loss": 1.4521443843841553, "step": 27226 }, { "epoch": 2.146393378005518, "grad_norm": 0.5032158493995667, "learning_rate": 0.0005, "loss": 1.3877949714660645, "step": 27227 }, { "epoch": 2.1464722112731573, "grad_norm": 0.41957083344459534, "learning_rate": 0.0005, "loss": 1.2897298336029053, "step": 27228 }, { "epoch": 2.146551044540796, "grad_norm": 0.3951965868473053, "learning_rate": 0.0005, "loss": 1.11492919921875, "step": 27229 }, { "epoch": 2.146629877808435, "grad_norm": 0.4808464050292969, "learning_rate": 0.0005, "loss": 1.3181339502334595, "step": 27230 }, { "epoch": 2.1467087110760743, "grad_norm": 0.4757348597049713, "learning_rate": 0.0005, "loss": 1.359499216079712, "step": 27231 }, { "epoch": 2.146787544343713, "grad_norm": 0.4733354151248932, "learning_rate": 0.0005, "loss": 1.1990294456481934, "step": 27232 }, { "epoch": 2.146866377611352, "grad_norm": 0.5414417386054993, "learning_rate": 0.0005, "loss": 1.5217702388763428, "step": 27233 }, { "epoch": 2.146945210878991, "grad_norm": 0.42424026131629944, "learning_rate": 0.0005, "loss": 1.269776463508606, "step": 27234 }, { "epoch": 2.14702404414663, "grad_norm": 0.5220842361450195, "learning_rate": 0.0005, "loss": 1.4925295114517212, "step": 27235 }, { "epoch": 2.1471028774142686, "grad_norm": 0.5109006762504578, "learning_rate": 0.0005, "loss": 1.3347638845443726, "step": 27236 }, { "epoch": 2.1471817106819078, "grad_norm": 0.49304306507110596, "learning_rate": 0.0005, "loss": 1.3457087278366089, "step": 27237 }, { "epoch": 2.147260543949547, "grad_norm": 0.4338401257991791, "learning_rate": 0.0005, "loss": 1.317120909690857, "step": 27238 }, { "epoch": 2.1473393772171856, "grad_norm": 0.5675132870674133, "learning_rate": 0.0005, "loss": 1.4711365699768066, "step": 27239 }, { "epoch": 2.1474182104848247, "grad_norm": 0.4310219883918762, "learning_rate": 0.0005, "loss": 1.2354025840759277, "step": 27240 }, { "epoch": 2.1474970437524634, "grad_norm": 0.4246249794960022, "learning_rate": 0.0005, "loss": 1.3494783639907837, "step": 27241 }, { "epoch": 2.1475758770201026, "grad_norm": 0.408225953578949, "learning_rate": 0.0005, "loss": 1.37996244430542, "step": 27242 }, { "epoch": 2.1476547102877412, "grad_norm": 0.47519341111183167, "learning_rate": 0.0005, "loss": 1.2743316888809204, "step": 27243 }, { "epoch": 2.1477335435553804, "grad_norm": 0.4798603951931, "learning_rate": 0.0005, "loss": 1.3143904209136963, "step": 27244 }, { "epoch": 2.1478123768230195, "grad_norm": 0.49170035123825073, "learning_rate": 0.0005, "loss": 1.2429389953613281, "step": 27245 }, { "epoch": 2.147891210090658, "grad_norm": 0.4949691891670227, "learning_rate": 0.0005, "loss": 1.4211571216583252, "step": 27246 }, { "epoch": 2.1479700433582973, "grad_norm": 0.5860175490379333, "learning_rate": 0.0005, "loss": 1.4882678985595703, "step": 27247 }, { "epoch": 2.148048876625936, "grad_norm": 0.45503950119018555, "learning_rate": 0.0005, "loss": 1.373907446861267, "step": 27248 }, { "epoch": 2.148127709893575, "grad_norm": 0.506507933139801, "learning_rate": 0.0005, "loss": 1.4695709943771362, "step": 27249 }, { "epoch": 2.148206543161214, "grad_norm": 0.5503748655319214, "learning_rate": 0.0005, "loss": 1.4353615045547485, "step": 27250 }, { "epoch": 2.148285376428853, "grad_norm": 0.8266297578811646, "learning_rate": 0.0005, "loss": 1.463804006576538, "step": 27251 }, { "epoch": 2.148364209696492, "grad_norm": 0.46975359320640564, "learning_rate": 0.0005, "loss": 1.4296494722366333, "step": 27252 }, { "epoch": 2.148443042964131, "grad_norm": 0.5065155625343323, "learning_rate": 0.0005, "loss": 1.367065668106079, "step": 27253 }, { "epoch": 2.14852187623177, "grad_norm": 0.5132631659507751, "learning_rate": 0.0005, "loss": 1.269370436668396, "step": 27254 }, { "epoch": 2.1486007094994086, "grad_norm": 0.5254696011543274, "learning_rate": 0.0005, "loss": 1.3928160667419434, "step": 27255 }, { "epoch": 2.1486795427670478, "grad_norm": 0.4186227023601532, "learning_rate": 0.0005, "loss": 1.1180579662322998, "step": 27256 }, { "epoch": 2.1487583760346864, "grad_norm": 0.4663425087928772, "learning_rate": 0.0005, "loss": 1.381063461303711, "step": 27257 }, { "epoch": 2.1488372093023256, "grad_norm": 0.4664452373981476, "learning_rate": 0.0005, "loss": 1.161353588104248, "step": 27258 }, { "epoch": 2.1489160425699647, "grad_norm": 0.44816598296165466, "learning_rate": 0.0005, "loss": 1.2955985069274902, "step": 27259 }, { "epoch": 2.1489948758376034, "grad_norm": 0.49553585052490234, "learning_rate": 0.0005, "loss": 1.3443632125854492, "step": 27260 }, { "epoch": 2.1490737091052425, "grad_norm": 0.4674740135669708, "learning_rate": 0.0005, "loss": 1.331254482269287, "step": 27261 }, { "epoch": 2.1491525423728812, "grad_norm": 0.48473748564720154, "learning_rate": 0.0005, "loss": 1.4459656476974487, "step": 27262 }, { "epoch": 2.1492313756405204, "grad_norm": 0.4483121931552887, "learning_rate": 0.0005, "loss": 1.2489840984344482, "step": 27263 }, { "epoch": 2.149310208908159, "grad_norm": 0.5224089026451111, "learning_rate": 0.0005, "loss": 1.4954644441604614, "step": 27264 }, { "epoch": 2.149389042175798, "grad_norm": 0.5581604242324829, "learning_rate": 0.0005, "loss": 1.2392995357513428, "step": 27265 }, { "epoch": 2.1494678754434373, "grad_norm": 0.45965808629989624, "learning_rate": 0.0005, "loss": 1.2726547718048096, "step": 27266 }, { "epoch": 2.149546708711076, "grad_norm": 0.5497046113014221, "learning_rate": 0.0005, "loss": 1.6885956525802612, "step": 27267 }, { "epoch": 2.149625541978715, "grad_norm": 0.7334465980529785, "learning_rate": 0.0005, "loss": 1.330723524093628, "step": 27268 }, { "epoch": 2.149704375246354, "grad_norm": 0.4219808578491211, "learning_rate": 0.0005, "loss": 1.3305304050445557, "step": 27269 }, { "epoch": 2.149783208513993, "grad_norm": 0.5567870140075684, "learning_rate": 0.0005, "loss": 1.354392170906067, "step": 27270 }, { "epoch": 2.1498620417816316, "grad_norm": 0.4277614653110504, "learning_rate": 0.0005, "loss": 1.4141902923583984, "step": 27271 }, { "epoch": 2.149940875049271, "grad_norm": 0.4783684313297272, "learning_rate": 0.0005, "loss": 1.2748953104019165, "step": 27272 }, { "epoch": 2.15001970831691, "grad_norm": 0.5606405735015869, "learning_rate": 0.0005, "loss": 1.527223825454712, "step": 27273 }, { "epoch": 2.1500985415845486, "grad_norm": 0.4635342061519623, "learning_rate": 0.0005, "loss": 1.3530328273773193, "step": 27274 }, { "epoch": 2.1501773748521877, "grad_norm": 0.4070497155189514, "learning_rate": 0.0005, "loss": 1.2531027793884277, "step": 27275 }, { "epoch": 2.1502562081198264, "grad_norm": 0.5344916582107544, "learning_rate": 0.0005, "loss": 1.542257308959961, "step": 27276 }, { "epoch": 2.1503350413874656, "grad_norm": 0.4955732226371765, "learning_rate": 0.0005, "loss": 1.3355774879455566, "step": 27277 }, { "epoch": 2.1504138746551043, "grad_norm": 0.455303817987442, "learning_rate": 0.0005, "loss": 1.1551876068115234, "step": 27278 }, { "epoch": 2.1504927079227434, "grad_norm": 0.5358603000640869, "learning_rate": 0.0005, "loss": 1.1644880771636963, "step": 27279 }, { "epoch": 2.1505715411903825, "grad_norm": 0.4294539988040924, "learning_rate": 0.0005, "loss": 1.1826586723327637, "step": 27280 }, { "epoch": 2.150650374458021, "grad_norm": 0.48424091935157776, "learning_rate": 0.0005, "loss": 1.401713252067566, "step": 27281 }, { "epoch": 2.1507292077256603, "grad_norm": 0.5342462658882141, "learning_rate": 0.0005, "loss": 1.2474982738494873, "step": 27282 }, { "epoch": 2.150808040993299, "grad_norm": 0.5268414616584778, "learning_rate": 0.0005, "loss": 1.409125566482544, "step": 27283 }, { "epoch": 2.150886874260938, "grad_norm": 0.47878411412239075, "learning_rate": 0.0005, "loss": 1.2697033882141113, "step": 27284 }, { "epoch": 2.150965707528577, "grad_norm": 0.42013272643089294, "learning_rate": 0.0005, "loss": 1.2951428890228271, "step": 27285 }, { "epoch": 2.151044540796216, "grad_norm": 0.5597307682037354, "learning_rate": 0.0005, "loss": 1.432518482208252, "step": 27286 }, { "epoch": 2.151123374063855, "grad_norm": 0.5408802032470703, "learning_rate": 0.0005, "loss": 1.55144464969635, "step": 27287 }, { "epoch": 2.151202207331494, "grad_norm": 0.4804445803165436, "learning_rate": 0.0005, "loss": 1.3231170177459717, "step": 27288 }, { "epoch": 2.151281040599133, "grad_norm": 0.5110616087913513, "learning_rate": 0.0005, "loss": 1.4390186071395874, "step": 27289 }, { "epoch": 2.1513598738667716, "grad_norm": 0.617900013923645, "learning_rate": 0.0005, "loss": 1.512008547782898, "step": 27290 }, { "epoch": 2.1514387071344108, "grad_norm": 0.5303274989128113, "learning_rate": 0.0005, "loss": 1.467862606048584, "step": 27291 }, { "epoch": 2.1515175404020495, "grad_norm": 0.5379756093025208, "learning_rate": 0.0005, "loss": 1.4776275157928467, "step": 27292 }, { "epoch": 2.1515963736696886, "grad_norm": 0.5342470407485962, "learning_rate": 0.0005, "loss": 1.3088370561599731, "step": 27293 }, { "epoch": 2.1516752069373277, "grad_norm": 0.47021663188934326, "learning_rate": 0.0005, "loss": 1.3471834659576416, "step": 27294 }, { "epoch": 2.1517540402049664, "grad_norm": 0.523395299911499, "learning_rate": 0.0005, "loss": 1.3184746503829956, "step": 27295 }, { "epoch": 2.1518328734726055, "grad_norm": 0.4478170871734619, "learning_rate": 0.0005, "loss": 1.375126600265503, "step": 27296 }, { "epoch": 2.1519117067402442, "grad_norm": 0.49847134947776794, "learning_rate": 0.0005, "loss": 1.3063830137252808, "step": 27297 }, { "epoch": 2.1519905400078834, "grad_norm": 0.6101563572883606, "learning_rate": 0.0005, "loss": 1.647322416305542, "step": 27298 }, { "epoch": 2.1520693732755225, "grad_norm": 0.3994496762752533, "learning_rate": 0.0005, "loss": 1.207880973815918, "step": 27299 }, { "epoch": 2.152148206543161, "grad_norm": 0.46823447942733765, "learning_rate": 0.0005, "loss": 1.2996282577514648, "step": 27300 }, { "epoch": 2.1522270398108003, "grad_norm": 0.4126242995262146, "learning_rate": 0.0005, "loss": 1.2315080165863037, "step": 27301 }, { "epoch": 2.152305873078439, "grad_norm": 0.49988460540771484, "learning_rate": 0.0005, "loss": 1.4156177043914795, "step": 27302 }, { "epoch": 2.152384706346078, "grad_norm": 0.5503394603729248, "learning_rate": 0.0005, "loss": 1.557363510131836, "step": 27303 }, { "epoch": 2.152463539613717, "grad_norm": 0.4611770510673523, "learning_rate": 0.0005, "loss": 1.2287688255310059, "step": 27304 }, { "epoch": 2.152542372881356, "grad_norm": 0.6231275796890259, "learning_rate": 0.0005, "loss": 1.2798497676849365, "step": 27305 }, { "epoch": 2.1526212061489947, "grad_norm": 0.5950547456741333, "learning_rate": 0.0005, "loss": 1.458927035331726, "step": 27306 }, { "epoch": 2.152700039416634, "grad_norm": 0.4593525528907776, "learning_rate": 0.0005, "loss": 1.3286330699920654, "step": 27307 }, { "epoch": 2.152778872684273, "grad_norm": 0.5261961221694946, "learning_rate": 0.0005, "loss": 1.326803207397461, "step": 27308 }, { "epoch": 2.1528577059519116, "grad_norm": 0.4583260416984558, "learning_rate": 0.0005, "loss": 1.3187886476516724, "step": 27309 }, { "epoch": 2.1529365392195507, "grad_norm": 0.5825979113578796, "learning_rate": 0.0005, "loss": 1.4724931716918945, "step": 27310 }, { "epoch": 2.1530153724871894, "grad_norm": 0.573304295539856, "learning_rate": 0.0005, "loss": 1.429426670074463, "step": 27311 }, { "epoch": 2.1530942057548286, "grad_norm": 0.6238038539886475, "learning_rate": 0.0005, "loss": 1.6626276969909668, "step": 27312 }, { "epoch": 2.1531730390224677, "grad_norm": 0.5169690847396851, "learning_rate": 0.0005, "loss": 1.4129480123519897, "step": 27313 }, { "epoch": 2.1532518722901064, "grad_norm": 0.6138249039649963, "learning_rate": 0.0005, "loss": 1.3304228782653809, "step": 27314 }, { "epoch": 2.1533307055577455, "grad_norm": 0.5167621970176697, "learning_rate": 0.0005, "loss": 1.2675400972366333, "step": 27315 }, { "epoch": 2.153409538825384, "grad_norm": 0.5411226749420166, "learning_rate": 0.0005, "loss": 1.392890214920044, "step": 27316 }, { "epoch": 2.1534883720930234, "grad_norm": 0.44147989153862, "learning_rate": 0.0005, "loss": 1.3600419759750366, "step": 27317 }, { "epoch": 2.153567205360662, "grad_norm": 0.485969603061676, "learning_rate": 0.0005, "loss": 1.3056493997573853, "step": 27318 }, { "epoch": 2.153646038628301, "grad_norm": 0.5554858446121216, "learning_rate": 0.0005, "loss": 1.6389610767364502, "step": 27319 }, { "epoch": 2.15372487189594, "grad_norm": 0.5492916107177734, "learning_rate": 0.0005, "loss": 1.11387038230896, "step": 27320 }, { "epoch": 2.153803705163579, "grad_norm": 0.527731716632843, "learning_rate": 0.0005, "loss": 1.379407525062561, "step": 27321 }, { "epoch": 2.153882538431218, "grad_norm": 0.8037453293800354, "learning_rate": 0.0005, "loss": 1.6384663581848145, "step": 27322 }, { "epoch": 2.153961371698857, "grad_norm": 0.41033735871315, "learning_rate": 0.0005, "loss": 1.1995470523834229, "step": 27323 }, { "epoch": 2.154040204966496, "grad_norm": 0.578694224357605, "learning_rate": 0.0005, "loss": 1.4791297912597656, "step": 27324 }, { "epoch": 2.1541190382341346, "grad_norm": 0.4486454129219055, "learning_rate": 0.0005, "loss": 1.239915132522583, "step": 27325 }, { "epoch": 2.1541978715017738, "grad_norm": 0.4676307737827301, "learning_rate": 0.0005, "loss": 1.4495818614959717, "step": 27326 }, { "epoch": 2.154276704769413, "grad_norm": 0.4912464916706085, "learning_rate": 0.0005, "loss": 1.351727843284607, "step": 27327 }, { "epoch": 2.1543555380370516, "grad_norm": 0.46351251006126404, "learning_rate": 0.0005, "loss": 1.4274336099624634, "step": 27328 }, { "epoch": 2.1544343713046907, "grad_norm": 0.4950752556324005, "learning_rate": 0.0005, "loss": 1.3731670379638672, "step": 27329 }, { "epoch": 2.1545132045723294, "grad_norm": 0.4970203936100006, "learning_rate": 0.0005, "loss": 1.2619271278381348, "step": 27330 }, { "epoch": 2.1545920378399686, "grad_norm": 0.6972354054450989, "learning_rate": 0.0005, "loss": 1.3418471813201904, "step": 27331 }, { "epoch": 2.1546708711076072, "grad_norm": 0.47531986236572266, "learning_rate": 0.0005, "loss": 1.3964800834655762, "step": 27332 }, { "epoch": 2.1547497043752464, "grad_norm": 0.5159022808074951, "learning_rate": 0.0005, "loss": 1.323026180267334, "step": 27333 }, { "epoch": 2.154828537642885, "grad_norm": 0.6321675777435303, "learning_rate": 0.0005, "loss": 1.4695799350738525, "step": 27334 }, { "epoch": 2.154907370910524, "grad_norm": 0.532347559928894, "learning_rate": 0.0005, "loss": 1.2574646472930908, "step": 27335 }, { "epoch": 2.1549862041781633, "grad_norm": 0.5329158306121826, "learning_rate": 0.0005, "loss": 1.4918339252471924, "step": 27336 }, { "epoch": 2.155065037445802, "grad_norm": 0.4969373047351837, "learning_rate": 0.0005, "loss": 1.2577757835388184, "step": 27337 }, { "epoch": 2.155143870713441, "grad_norm": 0.3753022849559784, "learning_rate": 0.0005, "loss": 1.0503456592559814, "step": 27338 }, { "epoch": 2.15522270398108, "grad_norm": 0.5297386050224304, "learning_rate": 0.0005, "loss": 1.427445411682129, "step": 27339 }, { "epoch": 2.155301537248719, "grad_norm": 0.47152066230773926, "learning_rate": 0.0005, "loss": 1.120119333267212, "step": 27340 }, { "epoch": 2.155380370516358, "grad_norm": 0.43627631664276123, "learning_rate": 0.0005, "loss": 1.2509527206420898, "step": 27341 }, { "epoch": 2.155459203783997, "grad_norm": 0.6182729005813599, "learning_rate": 0.0005, "loss": 1.2640454769134521, "step": 27342 }, { "epoch": 2.155538037051636, "grad_norm": 0.875244677066803, "learning_rate": 0.0005, "loss": 1.3010563850402832, "step": 27343 }, { "epoch": 2.1556168703192746, "grad_norm": 0.44165265560150146, "learning_rate": 0.0005, "loss": 1.1378397941589355, "step": 27344 }, { "epoch": 2.1556957035869138, "grad_norm": 0.38971808552742004, "learning_rate": 0.0005, "loss": 1.1598179340362549, "step": 27345 }, { "epoch": 2.1557745368545524, "grad_norm": 0.5830932259559631, "learning_rate": 0.0005, "loss": 1.2961407899856567, "step": 27346 }, { "epoch": 2.1558533701221916, "grad_norm": 0.5220316052436829, "learning_rate": 0.0005, "loss": 1.4444935321807861, "step": 27347 }, { "epoch": 2.1559322033898307, "grad_norm": 0.5410134792327881, "learning_rate": 0.0005, "loss": 1.235090732574463, "step": 27348 }, { "epoch": 2.1560110366574694, "grad_norm": 0.3693549931049347, "learning_rate": 0.0005, "loss": 1.1282813549041748, "step": 27349 }, { "epoch": 2.1560898699251085, "grad_norm": 0.5815514326095581, "learning_rate": 0.0005, "loss": 1.3367788791656494, "step": 27350 }, { "epoch": 2.1561687031927472, "grad_norm": 0.5471286177635193, "learning_rate": 0.0005, "loss": 1.312549114227295, "step": 27351 }, { "epoch": 2.1562475364603864, "grad_norm": 0.4878244698047638, "learning_rate": 0.0005, "loss": 1.1420091390609741, "step": 27352 }, { "epoch": 2.156326369728025, "grad_norm": 0.6458975076675415, "learning_rate": 0.0005, "loss": 1.3110419511795044, "step": 27353 }, { "epoch": 2.156405202995664, "grad_norm": 0.5640323162078857, "learning_rate": 0.0005, "loss": 1.4076809883117676, "step": 27354 }, { "epoch": 2.1564840362633033, "grad_norm": 0.5586226582527161, "learning_rate": 0.0005, "loss": 1.5106016397476196, "step": 27355 }, { "epoch": 2.156562869530942, "grad_norm": 0.5104104280471802, "learning_rate": 0.0005, "loss": 1.198176622390747, "step": 27356 }, { "epoch": 2.156641702798581, "grad_norm": 0.553212583065033, "learning_rate": 0.0005, "loss": 1.3921144008636475, "step": 27357 }, { "epoch": 2.15672053606622, "grad_norm": 0.520225465297699, "learning_rate": 0.0005, "loss": 1.4435617923736572, "step": 27358 }, { "epoch": 2.156799369333859, "grad_norm": 0.43031221628189087, "learning_rate": 0.0005, "loss": 1.1902480125427246, "step": 27359 }, { "epoch": 2.1568782026014977, "grad_norm": 0.4884396493434906, "learning_rate": 0.0005, "loss": 1.3338860273361206, "step": 27360 }, { "epoch": 2.156957035869137, "grad_norm": 0.5402728319168091, "learning_rate": 0.0005, "loss": 1.3681060075759888, "step": 27361 }, { "epoch": 2.157035869136776, "grad_norm": 0.48711660504341125, "learning_rate": 0.0005, "loss": 1.2900149822235107, "step": 27362 }, { "epoch": 2.1571147024044146, "grad_norm": 0.5656374096870422, "learning_rate": 0.0005, "loss": 1.3977880477905273, "step": 27363 }, { "epoch": 2.1571935356720537, "grad_norm": 0.5376161336898804, "learning_rate": 0.0005, "loss": 1.2083576917648315, "step": 27364 }, { "epoch": 2.1572723689396924, "grad_norm": 0.4929778277873993, "learning_rate": 0.0005, "loss": 1.5096924304962158, "step": 27365 }, { "epoch": 2.1573512022073316, "grad_norm": 0.43697360157966614, "learning_rate": 0.0005, "loss": 1.2859797477722168, "step": 27366 }, { "epoch": 2.1574300354749703, "grad_norm": 0.6454047560691833, "learning_rate": 0.0005, "loss": 1.322743535041809, "step": 27367 }, { "epoch": 2.1575088687426094, "grad_norm": 0.46523797512054443, "learning_rate": 0.0005, "loss": 1.1042766571044922, "step": 27368 }, { "epoch": 2.1575877020102485, "grad_norm": 0.5144999027252197, "learning_rate": 0.0005, "loss": 1.3264951705932617, "step": 27369 }, { "epoch": 2.157666535277887, "grad_norm": 0.45614972710609436, "learning_rate": 0.0005, "loss": 1.005190134048462, "step": 27370 }, { "epoch": 2.1577453685455263, "grad_norm": 0.5452471375465393, "learning_rate": 0.0005, "loss": 1.440441370010376, "step": 27371 }, { "epoch": 2.157824201813165, "grad_norm": 0.49926891922950745, "learning_rate": 0.0005, "loss": 1.1473116874694824, "step": 27372 }, { "epoch": 2.157903035080804, "grad_norm": 0.5788209438323975, "learning_rate": 0.0005, "loss": 1.4457416534423828, "step": 27373 }, { "epoch": 2.157981868348443, "grad_norm": 0.5664433836936951, "learning_rate": 0.0005, "loss": 1.3407286405563354, "step": 27374 }, { "epoch": 2.158060701616082, "grad_norm": 0.466473788022995, "learning_rate": 0.0005, "loss": 1.3146171569824219, "step": 27375 }, { "epoch": 2.158139534883721, "grad_norm": 0.5006668567657471, "learning_rate": 0.0005, "loss": 1.3137062788009644, "step": 27376 }, { "epoch": 2.15821836815136, "grad_norm": 0.4394311308860779, "learning_rate": 0.0005, "loss": 1.214484691619873, "step": 27377 }, { "epoch": 2.158297201418999, "grad_norm": 0.505588948726654, "learning_rate": 0.0005, "loss": 1.385405421257019, "step": 27378 }, { "epoch": 2.1583760346866376, "grad_norm": 0.6609221696853638, "learning_rate": 0.0005, "loss": 1.2231374979019165, "step": 27379 }, { "epoch": 2.1584548679542768, "grad_norm": 0.41071662306785583, "learning_rate": 0.0005, "loss": 1.2072598934173584, "step": 27380 }, { "epoch": 2.1585337012219155, "grad_norm": 0.4451150596141815, "learning_rate": 0.0005, "loss": 1.3419936895370483, "step": 27381 }, { "epoch": 2.1586125344895546, "grad_norm": 0.5032547116279602, "learning_rate": 0.0005, "loss": 1.4857364892959595, "step": 27382 }, { "epoch": 2.1586913677571937, "grad_norm": 0.6600362062454224, "learning_rate": 0.0005, "loss": 1.407715082168579, "step": 27383 }, { "epoch": 2.1587702010248324, "grad_norm": 0.45872220396995544, "learning_rate": 0.0005, "loss": 1.3164201974868774, "step": 27384 }, { "epoch": 2.1588490342924715, "grad_norm": 0.49621957540512085, "learning_rate": 0.0005, "loss": 1.2405258417129517, "step": 27385 }, { "epoch": 2.1589278675601102, "grad_norm": 0.5353471040725708, "learning_rate": 0.0005, "loss": 1.2710812091827393, "step": 27386 }, { "epoch": 2.1590067008277494, "grad_norm": 0.4926385283470154, "learning_rate": 0.0005, "loss": 1.5052192211151123, "step": 27387 }, { "epoch": 2.159085534095388, "grad_norm": 0.9656508564949036, "learning_rate": 0.0005, "loss": 1.6212480068206787, "step": 27388 }, { "epoch": 2.159164367363027, "grad_norm": 0.5232201218605042, "learning_rate": 0.0005, "loss": 1.3254706859588623, "step": 27389 }, { "epoch": 2.1592432006306663, "grad_norm": 0.5573775172233582, "learning_rate": 0.0005, "loss": 1.4467650651931763, "step": 27390 }, { "epoch": 2.159322033898305, "grad_norm": 0.5092108845710754, "learning_rate": 0.0005, "loss": 1.3612788915634155, "step": 27391 }, { "epoch": 2.159400867165944, "grad_norm": 0.7133508324623108, "learning_rate": 0.0005, "loss": 1.4778923988342285, "step": 27392 }, { "epoch": 2.159479700433583, "grad_norm": 0.41116851568222046, "learning_rate": 0.0005, "loss": 1.3611468076705933, "step": 27393 }, { "epoch": 2.159558533701222, "grad_norm": 0.4562957286834717, "learning_rate": 0.0005, "loss": 1.311288833618164, "step": 27394 }, { "epoch": 2.1596373669688607, "grad_norm": 0.47586700320243835, "learning_rate": 0.0005, "loss": 1.3687924146652222, "step": 27395 }, { "epoch": 2.1597162002365, "grad_norm": 1.3073155879974365, "learning_rate": 0.0005, "loss": 1.4488704204559326, "step": 27396 }, { "epoch": 2.159795033504139, "grad_norm": 0.5076315999031067, "learning_rate": 0.0005, "loss": 1.5352030992507935, "step": 27397 }, { "epoch": 2.1598738667717776, "grad_norm": 0.610550582408905, "learning_rate": 0.0005, "loss": 1.4669303894042969, "step": 27398 }, { "epoch": 2.1599527000394168, "grad_norm": 0.6011984944343567, "learning_rate": 0.0005, "loss": 1.1845251321792603, "step": 27399 }, { "epoch": 2.1600315333070554, "grad_norm": 0.503394365310669, "learning_rate": 0.0005, "loss": 1.313166618347168, "step": 27400 }, { "epoch": 2.1601103665746946, "grad_norm": 0.4098977744579315, "learning_rate": 0.0005, "loss": 1.3780368566513062, "step": 27401 }, { "epoch": 2.1601891998423333, "grad_norm": 0.6956062316894531, "learning_rate": 0.0005, "loss": 1.4086612462997437, "step": 27402 }, { "epoch": 2.1602680331099724, "grad_norm": 0.5385603308677673, "learning_rate": 0.0005, "loss": 1.425194263458252, "step": 27403 }, { "epoch": 2.1603468663776115, "grad_norm": 0.4520590603351593, "learning_rate": 0.0005, "loss": 1.2558250427246094, "step": 27404 }, { "epoch": 2.16042569964525, "grad_norm": 0.4304790198802948, "learning_rate": 0.0005, "loss": 1.2951667308807373, "step": 27405 }, { "epoch": 2.1605045329128894, "grad_norm": 0.40696218609809875, "learning_rate": 0.0005, "loss": 1.1251230239868164, "step": 27406 }, { "epoch": 2.160583366180528, "grad_norm": 0.4769846498966217, "learning_rate": 0.0005, "loss": 1.2107822895050049, "step": 27407 }, { "epoch": 2.160662199448167, "grad_norm": 0.4637022912502289, "learning_rate": 0.0005, "loss": 1.3035619258880615, "step": 27408 }, { "epoch": 2.160741032715806, "grad_norm": 0.5983501672744751, "learning_rate": 0.0005, "loss": 1.4724653959274292, "step": 27409 }, { "epoch": 2.160819865983445, "grad_norm": 0.535204291343689, "learning_rate": 0.0005, "loss": 1.5798240900039673, "step": 27410 }, { "epoch": 2.160898699251084, "grad_norm": 0.5403060913085938, "learning_rate": 0.0005, "loss": 1.3291183710098267, "step": 27411 }, { "epoch": 2.160977532518723, "grad_norm": 0.46190130710601807, "learning_rate": 0.0005, "loss": 1.4131405353546143, "step": 27412 }, { "epoch": 2.161056365786362, "grad_norm": 0.4510885179042816, "learning_rate": 0.0005, "loss": 1.1708362102508545, "step": 27413 }, { "epoch": 2.1611351990540006, "grad_norm": 0.49150753021240234, "learning_rate": 0.0005, "loss": 1.2118544578552246, "step": 27414 }, { "epoch": 2.1612140323216398, "grad_norm": 0.6019675731658936, "learning_rate": 0.0005, "loss": 1.2269442081451416, "step": 27415 }, { "epoch": 2.161292865589279, "grad_norm": 0.41259241104125977, "learning_rate": 0.0005, "loss": 1.2428514957427979, "step": 27416 }, { "epoch": 2.1613716988569176, "grad_norm": 0.48512348532676697, "learning_rate": 0.0005, "loss": 1.441457748413086, "step": 27417 }, { "epoch": 2.1614505321245567, "grad_norm": 0.5200076699256897, "learning_rate": 0.0005, "loss": 1.1982483863830566, "step": 27418 }, { "epoch": 2.1615293653921954, "grad_norm": 0.5226909518241882, "learning_rate": 0.0005, "loss": 1.461573839187622, "step": 27419 }, { "epoch": 2.1616081986598346, "grad_norm": 0.4811708331108093, "learning_rate": 0.0005, "loss": 1.40803062915802, "step": 27420 }, { "epoch": 2.1616870319274732, "grad_norm": 0.6406228542327881, "learning_rate": 0.0005, "loss": 1.2832236289978027, "step": 27421 }, { "epoch": 2.1617658651951124, "grad_norm": 0.4456280469894409, "learning_rate": 0.0005, "loss": 1.3461065292358398, "step": 27422 }, { "epoch": 2.161844698462751, "grad_norm": 0.48251691460609436, "learning_rate": 0.0005, "loss": 1.1853344440460205, "step": 27423 }, { "epoch": 2.16192353173039, "grad_norm": 0.49220430850982666, "learning_rate": 0.0005, "loss": 1.300126075744629, "step": 27424 }, { "epoch": 2.1620023649980293, "grad_norm": 0.5256655216217041, "learning_rate": 0.0005, "loss": 1.2859644889831543, "step": 27425 }, { "epoch": 2.162081198265668, "grad_norm": 0.4990346133708954, "learning_rate": 0.0005, "loss": 1.3118470907211304, "step": 27426 }, { "epoch": 2.162160031533307, "grad_norm": 0.5875138640403748, "learning_rate": 0.0005, "loss": 1.4389287233352661, "step": 27427 }, { "epoch": 2.162238864800946, "grad_norm": 0.4390171468257904, "learning_rate": 0.0005, "loss": 1.3723440170288086, "step": 27428 }, { "epoch": 2.162317698068585, "grad_norm": 0.573722243309021, "learning_rate": 0.0005, "loss": 1.2182583808898926, "step": 27429 }, { "epoch": 2.162396531336224, "grad_norm": 0.4866783022880554, "learning_rate": 0.0005, "loss": 1.22709321975708, "step": 27430 }, { "epoch": 2.162475364603863, "grad_norm": 0.524867594242096, "learning_rate": 0.0005, "loss": 1.2606165409088135, "step": 27431 }, { "epoch": 2.162554197871502, "grad_norm": 0.41707223653793335, "learning_rate": 0.0005, "loss": 1.2292490005493164, "step": 27432 }, { "epoch": 2.1626330311391406, "grad_norm": 0.5799057483673096, "learning_rate": 0.0005, "loss": 1.423406720161438, "step": 27433 }, { "epoch": 2.1627118644067798, "grad_norm": 0.41132673621177673, "learning_rate": 0.0005, "loss": 1.1701093912124634, "step": 27434 }, { "epoch": 2.1627906976744184, "grad_norm": 0.5431414246559143, "learning_rate": 0.0005, "loss": 1.4239511489868164, "step": 27435 }, { "epoch": 2.1628695309420576, "grad_norm": 0.7630829811096191, "learning_rate": 0.0005, "loss": 1.3280742168426514, "step": 27436 }, { "epoch": 2.1629483642096963, "grad_norm": 0.5344016551971436, "learning_rate": 0.0005, "loss": 1.40238618850708, "step": 27437 }, { "epoch": 2.1630271974773354, "grad_norm": 0.7415120601654053, "learning_rate": 0.0005, "loss": 1.6627445220947266, "step": 27438 }, { "epoch": 2.1631060307449745, "grad_norm": 0.4944869577884674, "learning_rate": 0.0005, "loss": 1.1508585214614868, "step": 27439 }, { "epoch": 2.1631848640126132, "grad_norm": 0.5220844745635986, "learning_rate": 0.0005, "loss": 1.287415862083435, "step": 27440 }, { "epoch": 2.1632636972802524, "grad_norm": 0.5758074522018433, "learning_rate": 0.0005, "loss": 1.1496084928512573, "step": 27441 }, { "epoch": 2.163342530547891, "grad_norm": 0.48461198806762695, "learning_rate": 0.0005, "loss": 1.196155071258545, "step": 27442 }, { "epoch": 2.16342136381553, "grad_norm": 0.5026702880859375, "learning_rate": 0.0005, "loss": 1.223049283027649, "step": 27443 }, { "epoch": 2.1635001970831693, "grad_norm": 0.6754467487335205, "learning_rate": 0.0005, "loss": 1.431331992149353, "step": 27444 }, { "epoch": 2.163579030350808, "grad_norm": 0.5331754088401794, "learning_rate": 0.0005, "loss": 1.3101906776428223, "step": 27445 }, { "epoch": 2.163657863618447, "grad_norm": 0.8430993556976318, "learning_rate": 0.0005, "loss": 1.3992836475372314, "step": 27446 }, { "epoch": 2.163736696886086, "grad_norm": 0.4646667242050171, "learning_rate": 0.0005, "loss": 1.3269643783569336, "step": 27447 }, { "epoch": 2.163815530153725, "grad_norm": 0.4779578745365143, "learning_rate": 0.0005, "loss": 1.2156157493591309, "step": 27448 }, { "epoch": 2.1638943634213637, "grad_norm": 0.6853093504905701, "learning_rate": 0.0005, "loss": 1.3998258113861084, "step": 27449 }, { "epoch": 2.163973196689003, "grad_norm": 0.5433935523033142, "learning_rate": 0.0005, "loss": 1.5001882314682007, "step": 27450 }, { "epoch": 2.1640520299566415, "grad_norm": 0.5348283648490906, "learning_rate": 0.0005, "loss": 1.1563599109649658, "step": 27451 }, { "epoch": 2.1641308632242806, "grad_norm": 0.5955511927604675, "learning_rate": 0.0005, "loss": 1.4494285583496094, "step": 27452 }, { "epoch": 2.1642096964919197, "grad_norm": 0.4391326308250427, "learning_rate": 0.0005, "loss": 1.346765160560608, "step": 27453 }, { "epoch": 2.1642885297595584, "grad_norm": 0.5735595226287842, "learning_rate": 0.0005, "loss": 1.5284714698791504, "step": 27454 }, { "epoch": 2.1643673630271976, "grad_norm": 0.4838627576828003, "learning_rate": 0.0005, "loss": 1.3929426670074463, "step": 27455 }, { "epoch": 2.1644461962948363, "grad_norm": 0.5982485413551331, "learning_rate": 0.0005, "loss": 1.1808173656463623, "step": 27456 }, { "epoch": 2.1645250295624754, "grad_norm": 0.559810221195221, "learning_rate": 0.0005, "loss": 1.279038906097412, "step": 27457 }, { "epoch": 2.1646038628301145, "grad_norm": 0.4999902546405792, "learning_rate": 0.0005, "loss": 1.251842737197876, "step": 27458 }, { "epoch": 2.164682696097753, "grad_norm": 0.615106999874115, "learning_rate": 0.0005, "loss": 1.580924153327942, "step": 27459 }, { "epoch": 2.1647615293653923, "grad_norm": 0.5337401628494263, "learning_rate": 0.0005, "loss": 1.2941954135894775, "step": 27460 }, { "epoch": 2.164840362633031, "grad_norm": 0.46448007225990295, "learning_rate": 0.0005, "loss": 1.1881439685821533, "step": 27461 }, { "epoch": 2.16491919590067, "grad_norm": 0.4550086259841919, "learning_rate": 0.0005, "loss": 1.2583616971969604, "step": 27462 }, { "epoch": 2.164998029168309, "grad_norm": 0.38978061079978943, "learning_rate": 0.0005, "loss": 1.2281239032745361, "step": 27463 }, { "epoch": 2.165076862435948, "grad_norm": 0.4305703639984131, "learning_rate": 0.0005, "loss": 1.2724804878234863, "step": 27464 }, { "epoch": 2.1651556957035867, "grad_norm": 0.4740018844604492, "learning_rate": 0.0005, "loss": 1.4317495822906494, "step": 27465 }, { "epoch": 2.165234528971226, "grad_norm": 0.5217959880828857, "learning_rate": 0.0005, "loss": 1.2750141620635986, "step": 27466 }, { "epoch": 2.165313362238865, "grad_norm": 0.45628413558006287, "learning_rate": 0.0005, "loss": 1.282094955444336, "step": 27467 }, { "epoch": 2.1653921955065036, "grad_norm": 0.4505848288536072, "learning_rate": 0.0005, "loss": 1.5113470554351807, "step": 27468 }, { "epoch": 2.1654710287741428, "grad_norm": 0.5010233521461487, "learning_rate": 0.0005, "loss": 1.2359172105789185, "step": 27469 }, { "epoch": 2.1655498620417815, "grad_norm": 0.39682716131210327, "learning_rate": 0.0005, "loss": 1.0997419357299805, "step": 27470 }, { "epoch": 2.1656286953094206, "grad_norm": 0.4448334872722626, "learning_rate": 0.0005, "loss": 1.295107364654541, "step": 27471 }, { "epoch": 2.1657075285770597, "grad_norm": 0.44630926847457886, "learning_rate": 0.0005, "loss": 1.4205762147903442, "step": 27472 }, { "epoch": 2.1657863618446984, "grad_norm": 0.43045610189437866, "learning_rate": 0.0005, "loss": 1.3476665019989014, "step": 27473 }, { "epoch": 2.1658651951123375, "grad_norm": 0.7121990323066711, "learning_rate": 0.0005, "loss": 1.2586288452148438, "step": 27474 }, { "epoch": 2.1659440283799762, "grad_norm": 0.48267829418182373, "learning_rate": 0.0005, "loss": 1.5059905052185059, "step": 27475 }, { "epoch": 2.1660228616476154, "grad_norm": 0.48496145009994507, "learning_rate": 0.0005, "loss": 1.3488962650299072, "step": 27476 }, { "epoch": 2.166101694915254, "grad_norm": 0.4966132640838623, "learning_rate": 0.0005, "loss": 1.1896274089813232, "step": 27477 }, { "epoch": 2.166180528182893, "grad_norm": 0.5711697936058044, "learning_rate": 0.0005, "loss": 1.3830816745758057, "step": 27478 }, { "epoch": 2.1662593614505323, "grad_norm": 0.5045295357704163, "learning_rate": 0.0005, "loss": 1.352189540863037, "step": 27479 }, { "epoch": 2.166338194718171, "grad_norm": 0.5248717665672302, "learning_rate": 0.0005, "loss": 1.4680516719818115, "step": 27480 }, { "epoch": 2.16641702798581, "grad_norm": 0.5816226601600647, "learning_rate": 0.0005, "loss": 1.410457968711853, "step": 27481 }, { "epoch": 2.166495861253449, "grad_norm": 0.5352383852005005, "learning_rate": 0.0005, "loss": 1.4567720890045166, "step": 27482 }, { "epoch": 2.166574694521088, "grad_norm": 0.5559389591217041, "learning_rate": 0.0005, "loss": 1.5246608257293701, "step": 27483 }, { "epoch": 2.1666535277887267, "grad_norm": 0.5199693441390991, "learning_rate": 0.0005, "loss": 1.5168499946594238, "step": 27484 }, { "epoch": 2.166732361056366, "grad_norm": 0.5204653739929199, "learning_rate": 0.0005, "loss": 1.3564796447753906, "step": 27485 }, { "epoch": 2.166811194324005, "grad_norm": 0.5846946239471436, "learning_rate": 0.0005, "loss": 1.3861937522888184, "step": 27486 }, { "epoch": 2.1668900275916436, "grad_norm": 0.5389056205749512, "learning_rate": 0.0005, "loss": 1.212834358215332, "step": 27487 }, { "epoch": 2.1669688608592828, "grad_norm": 0.47275835275650024, "learning_rate": 0.0005, "loss": 1.1158714294433594, "step": 27488 }, { "epoch": 2.1670476941269214, "grad_norm": 0.4762061834335327, "learning_rate": 0.0005, "loss": 1.395725965499878, "step": 27489 }, { "epoch": 2.1671265273945606, "grad_norm": 0.5952422022819519, "learning_rate": 0.0005, "loss": 1.5306321382522583, "step": 27490 }, { "epoch": 2.1672053606621993, "grad_norm": 0.4765929579734802, "learning_rate": 0.0005, "loss": 1.4011101722717285, "step": 27491 }, { "epoch": 2.1672841939298384, "grad_norm": 0.5409395694732666, "learning_rate": 0.0005, "loss": 1.4733643531799316, "step": 27492 }, { "epoch": 2.1673630271974775, "grad_norm": 0.7283906936645508, "learning_rate": 0.0005, "loss": 1.4030550718307495, "step": 27493 }, { "epoch": 2.167441860465116, "grad_norm": 0.4220145642757416, "learning_rate": 0.0005, "loss": 1.0700643062591553, "step": 27494 }, { "epoch": 2.1675206937327554, "grad_norm": 0.5124160051345825, "learning_rate": 0.0005, "loss": 1.2444485425949097, "step": 27495 }, { "epoch": 2.167599527000394, "grad_norm": 0.628696858882904, "learning_rate": 0.0005, "loss": 1.4928772449493408, "step": 27496 }, { "epoch": 2.167678360268033, "grad_norm": 0.5305566191673279, "learning_rate": 0.0005, "loss": 1.3970417976379395, "step": 27497 }, { "epoch": 2.167757193535672, "grad_norm": 0.5780356526374817, "learning_rate": 0.0005, "loss": 1.2554000616073608, "step": 27498 }, { "epoch": 2.167836026803311, "grad_norm": 0.5396630167961121, "learning_rate": 0.0005, "loss": 1.4465036392211914, "step": 27499 }, { "epoch": 2.16791486007095, "grad_norm": 0.6107105612754822, "learning_rate": 0.0005, "loss": 1.4364196062088013, "step": 27500 }, { "epoch": 2.167993693338589, "grad_norm": 0.5880661010742188, "learning_rate": 0.0005, "loss": 1.2160673141479492, "step": 27501 }, { "epoch": 2.168072526606228, "grad_norm": 0.5789574980735779, "learning_rate": 0.0005, "loss": 1.4076007604599, "step": 27502 }, { "epoch": 2.1681513598738666, "grad_norm": 0.6293506622314453, "learning_rate": 0.0005, "loss": 1.4424800872802734, "step": 27503 }, { "epoch": 2.168230193141506, "grad_norm": 0.4708743691444397, "learning_rate": 0.0005, "loss": 1.4319756031036377, "step": 27504 }, { "epoch": 2.1683090264091445, "grad_norm": 0.5107477903366089, "learning_rate": 0.0005, "loss": 1.3326072692871094, "step": 27505 }, { "epoch": 2.1683878596767836, "grad_norm": 0.5358905792236328, "learning_rate": 0.0005, "loss": 1.3837628364562988, "step": 27506 }, { "epoch": 2.1684666929444227, "grad_norm": 0.5452705025672913, "learning_rate": 0.0005, "loss": 1.3836148977279663, "step": 27507 }, { "epoch": 2.1685455262120614, "grad_norm": 0.6324033141136169, "learning_rate": 0.0005, "loss": 1.1745178699493408, "step": 27508 }, { "epoch": 2.1686243594797006, "grad_norm": 0.5590446591377258, "learning_rate": 0.0005, "loss": 1.3729228973388672, "step": 27509 }, { "epoch": 2.1687031927473392, "grad_norm": 0.48543408513069153, "learning_rate": 0.0005, "loss": 1.29756498336792, "step": 27510 }, { "epoch": 2.1687820260149784, "grad_norm": 0.6485567688941956, "learning_rate": 0.0005, "loss": 1.5489521026611328, "step": 27511 }, { "epoch": 2.168860859282617, "grad_norm": 0.6638679504394531, "learning_rate": 0.0005, "loss": 1.3459478616714478, "step": 27512 }, { "epoch": 2.168939692550256, "grad_norm": 0.41218480467796326, "learning_rate": 0.0005, "loss": 1.1951099634170532, "step": 27513 }, { "epoch": 2.1690185258178953, "grad_norm": 0.5667170286178589, "learning_rate": 0.0005, "loss": 1.4495484828948975, "step": 27514 }, { "epoch": 2.169097359085534, "grad_norm": 0.429950088262558, "learning_rate": 0.0005, "loss": 1.4694409370422363, "step": 27515 }, { "epoch": 2.169176192353173, "grad_norm": 0.4699883460998535, "learning_rate": 0.0005, "loss": 1.338432788848877, "step": 27516 }, { "epoch": 2.169255025620812, "grad_norm": 0.7093356847763062, "learning_rate": 0.0005, "loss": 1.302236557006836, "step": 27517 }, { "epoch": 2.169333858888451, "grad_norm": 0.4425326883792877, "learning_rate": 0.0005, "loss": 1.372826099395752, "step": 27518 }, { "epoch": 2.1694126921560897, "grad_norm": 0.60284423828125, "learning_rate": 0.0005, "loss": 1.4824074506759644, "step": 27519 }, { "epoch": 2.169491525423729, "grad_norm": 0.5774567127227783, "learning_rate": 0.0005, "loss": 1.412614107131958, "step": 27520 }, { "epoch": 2.169570358691368, "grad_norm": 0.4388711154460907, "learning_rate": 0.0005, "loss": 1.1684534549713135, "step": 27521 }, { "epoch": 2.1696491919590066, "grad_norm": 0.5107621550559998, "learning_rate": 0.0005, "loss": 1.3162480592727661, "step": 27522 }, { "epoch": 2.1697280252266458, "grad_norm": 0.5461861491203308, "learning_rate": 0.0005, "loss": 1.4128295183181763, "step": 27523 }, { "epoch": 2.1698068584942845, "grad_norm": 0.586693525314331, "learning_rate": 0.0005, "loss": 1.3809468746185303, "step": 27524 }, { "epoch": 2.1698856917619236, "grad_norm": 0.5384500622749329, "learning_rate": 0.0005, "loss": 1.455622911453247, "step": 27525 }, { "epoch": 2.1699645250295623, "grad_norm": 0.4898207187652588, "learning_rate": 0.0005, "loss": 1.5062220096588135, "step": 27526 }, { "epoch": 2.1700433582972014, "grad_norm": 0.4617673456668854, "learning_rate": 0.0005, "loss": 1.465308666229248, "step": 27527 }, { "epoch": 2.1701221915648405, "grad_norm": 0.48299095034599304, "learning_rate": 0.0005, "loss": 1.5456924438476562, "step": 27528 }, { "epoch": 2.1702010248324792, "grad_norm": 0.40425747632980347, "learning_rate": 0.0005, "loss": 1.2626807689666748, "step": 27529 }, { "epoch": 2.1702798581001184, "grad_norm": 0.9797006845474243, "learning_rate": 0.0005, "loss": 1.5991251468658447, "step": 27530 }, { "epoch": 2.170358691367757, "grad_norm": 0.5156222581863403, "learning_rate": 0.0005, "loss": 1.271704077720642, "step": 27531 }, { "epoch": 2.170437524635396, "grad_norm": 0.5077350735664368, "learning_rate": 0.0005, "loss": 1.5268434286117554, "step": 27532 }, { "epoch": 2.170516357903035, "grad_norm": 0.415330171585083, "learning_rate": 0.0005, "loss": 1.2999153137207031, "step": 27533 }, { "epoch": 2.170595191170674, "grad_norm": 0.46642524003982544, "learning_rate": 0.0005, "loss": 1.445737361907959, "step": 27534 }, { "epoch": 2.170674024438313, "grad_norm": 0.46366339921951294, "learning_rate": 0.0005, "loss": 1.4314711093902588, "step": 27535 }, { "epoch": 2.170752857705952, "grad_norm": 0.5374650955200195, "learning_rate": 0.0005, "loss": 1.528376579284668, "step": 27536 }, { "epoch": 2.170831690973591, "grad_norm": 0.6063327789306641, "learning_rate": 0.0005, "loss": 1.4012620449066162, "step": 27537 }, { "epoch": 2.1709105242412297, "grad_norm": 0.4610099792480469, "learning_rate": 0.0005, "loss": 1.3310035467147827, "step": 27538 }, { "epoch": 2.170989357508869, "grad_norm": 0.4702998995780945, "learning_rate": 0.0005, "loss": 1.280925989151001, "step": 27539 }, { "epoch": 2.1710681907765075, "grad_norm": 0.49642589688301086, "learning_rate": 0.0005, "loss": 1.2983016967773438, "step": 27540 }, { "epoch": 2.1711470240441466, "grad_norm": 0.3856564462184906, "learning_rate": 0.0005, "loss": 1.1301240921020508, "step": 27541 }, { "epoch": 2.1712258573117857, "grad_norm": 0.5518009662628174, "learning_rate": 0.0005, "loss": 1.3676834106445312, "step": 27542 }, { "epoch": 2.1713046905794244, "grad_norm": 0.5571284294128418, "learning_rate": 0.0005, "loss": 1.3923007249832153, "step": 27543 }, { "epoch": 2.1713835238470636, "grad_norm": 0.585163414478302, "learning_rate": 0.0005, "loss": 1.345544457435608, "step": 27544 }, { "epoch": 2.1714623571147023, "grad_norm": 0.47822844982147217, "learning_rate": 0.0005, "loss": 1.386531114578247, "step": 27545 }, { "epoch": 2.1715411903823414, "grad_norm": 0.5706571936607361, "learning_rate": 0.0005, "loss": 1.4435505867004395, "step": 27546 }, { "epoch": 2.1716200236499805, "grad_norm": 0.4665642976760864, "learning_rate": 0.0005, "loss": 1.294475793838501, "step": 27547 }, { "epoch": 2.171698856917619, "grad_norm": 0.6053951978683472, "learning_rate": 0.0005, "loss": 1.6923658847808838, "step": 27548 }, { "epoch": 2.1717776901852583, "grad_norm": 0.4493299126625061, "learning_rate": 0.0005, "loss": 1.281708836555481, "step": 27549 }, { "epoch": 2.171856523452897, "grad_norm": 0.48973676562309265, "learning_rate": 0.0005, "loss": 1.4618117809295654, "step": 27550 }, { "epoch": 2.171935356720536, "grad_norm": 0.4267178773880005, "learning_rate": 0.0005, "loss": 1.2741620540618896, "step": 27551 }, { "epoch": 2.172014189988175, "grad_norm": 0.49680736660957336, "learning_rate": 0.0005, "loss": 1.1910252571105957, "step": 27552 }, { "epoch": 2.172093023255814, "grad_norm": 0.49873077869415283, "learning_rate": 0.0005, "loss": 1.4317631721496582, "step": 27553 }, { "epoch": 2.1721718565234527, "grad_norm": 0.35162612795829773, "learning_rate": 0.0005, "loss": 1.0929182767868042, "step": 27554 }, { "epoch": 2.172250689791092, "grad_norm": 0.4630352854728699, "learning_rate": 0.0005, "loss": 1.1911669969558716, "step": 27555 }, { "epoch": 2.172329523058731, "grad_norm": 0.6102254986763, "learning_rate": 0.0005, "loss": 1.5006558895111084, "step": 27556 }, { "epoch": 2.1724083563263696, "grad_norm": 0.660662055015564, "learning_rate": 0.0005, "loss": 1.4764301776885986, "step": 27557 }, { "epoch": 2.1724871895940088, "grad_norm": 0.4834437370300293, "learning_rate": 0.0005, "loss": 1.2005906105041504, "step": 27558 }, { "epoch": 2.1725660228616475, "grad_norm": 0.5818244218826294, "learning_rate": 0.0005, "loss": 1.4441840648651123, "step": 27559 }, { "epoch": 2.1726448561292866, "grad_norm": 0.478035569190979, "learning_rate": 0.0005, "loss": 1.3900333642959595, "step": 27560 }, { "epoch": 2.1727236893969257, "grad_norm": 0.540344774723053, "learning_rate": 0.0005, "loss": 1.3306775093078613, "step": 27561 }, { "epoch": 2.1728025226645644, "grad_norm": 0.6645601987838745, "learning_rate": 0.0005, "loss": 1.5515834093093872, "step": 27562 }, { "epoch": 2.1728813559322036, "grad_norm": 0.8927807807922363, "learning_rate": 0.0005, "loss": 1.2612457275390625, "step": 27563 }, { "epoch": 2.1729601891998422, "grad_norm": 0.4536847174167633, "learning_rate": 0.0005, "loss": 1.2214946746826172, "step": 27564 }, { "epoch": 2.1730390224674814, "grad_norm": 0.5903763771057129, "learning_rate": 0.0005, "loss": 1.3638743162155151, "step": 27565 }, { "epoch": 2.17311785573512, "grad_norm": 0.5433675646781921, "learning_rate": 0.0005, "loss": 1.5186235904693604, "step": 27566 }, { "epoch": 2.173196689002759, "grad_norm": 0.4515667259693146, "learning_rate": 0.0005, "loss": 1.3378907442092896, "step": 27567 }, { "epoch": 2.173275522270398, "grad_norm": 0.5241944789886475, "learning_rate": 0.0005, "loss": 1.22069251537323, "step": 27568 }, { "epoch": 2.173354355538037, "grad_norm": 0.444309800863266, "learning_rate": 0.0005, "loss": 1.3301793336868286, "step": 27569 }, { "epoch": 2.173433188805676, "grad_norm": 0.42961233854293823, "learning_rate": 0.0005, "loss": 1.3166182041168213, "step": 27570 }, { "epoch": 2.173512022073315, "grad_norm": 0.5171361565589905, "learning_rate": 0.0005, "loss": 1.4762818813323975, "step": 27571 }, { "epoch": 2.173590855340954, "grad_norm": 0.5138381123542786, "learning_rate": 0.0005, "loss": 1.271815538406372, "step": 27572 }, { "epoch": 2.1736696886085927, "grad_norm": 0.5725290775299072, "learning_rate": 0.0005, "loss": 1.3050994873046875, "step": 27573 }, { "epoch": 2.173748521876232, "grad_norm": 0.4647907614707947, "learning_rate": 0.0005, "loss": 1.266909122467041, "step": 27574 }, { "epoch": 2.173827355143871, "grad_norm": 0.5883275866508484, "learning_rate": 0.0005, "loss": 1.3714646100997925, "step": 27575 }, { "epoch": 2.1739061884115096, "grad_norm": 0.4932706952095032, "learning_rate": 0.0005, "loss": 1.3678081035614014, "step": 27576 }, { "epoch": 2.1739850216791488, "grad_norm": 0.44902151823043823, "learning_rate": 0.0005, "loss": 1.289536714553833, "step": 27577 }, { "epoch": 2.1740638549467874, "grad_norm": 0.4902109205722809, "learning_rate": 0.0005, "loss": 1.2494415044784546, "step": 27578 }, { "epoch": 2.1741426882144266, "grad_norm": 0.47250404953956604, "learning_rate": 0.0005, "loss": 1.450727105140686, "step": 27579 }, { "epoch": 2.1742215214820653, "grad_norm": 0.4935055077075958, "learning_rate": 0.0005, "loss": 1.4250589609146118, "step": 27580 }, { "epoch": 2.1743003547497044, "grad_norm": 0.5474644899368286, "learning_rate": 0.0005, "loss": 1.288017749786377, "step": 27581 }, { "epoch": 2.174379188017343, "grad_norm": 0.5487800240516663, "learning_rate": 0.0005, "loss": 1.4291024208068848, "step": 27582 }, { "epoch": 2.1744580212849822, "grad_norm": 0.6379727721214294, "learning_rate": 0.0005, "loss": 1.5408598184585571, "step": 27583 }, { "epoch": 2.1745368545526214, "grad_norm": 0.48860323429107666, "learning_rate": 0.0005, "loss": 1.2370400428771973, "step": 27584 }, { "epoch": 2.17461568782026, "grad_norm": 0.48919805884361267, "learning_rate": 0.0005, "loss": 1.3926665782928467, "step": 27585 }, { "epoch": 2.174694521087899, "grad_norm": 0.592754065990448, "learning_rate": 0.0005, "loss": 1.4296027421951294, "step": 27586 }, { "epoch": 2.174773354355538, "grad_norm": 0.4773910939693451, "learning_rate": 0.0005, "loss": 1.3363618850708008, "step": 27587 }, { "epoch": 2.174852187623177, "grad_norm": 0.5020500421524048, "learning_rate": 0.0005, "loss": 1.3471078872680664, "step": 27588 }, { "epoch": 2.174931020890816, "grad_norm": 0.495501309633255, "learning_rate": 0.0005, "loss": 1.3447015285491943, "step": 27589 }, { "epoch": 2.175009854158455, "grad_norm": 0.44590049982070923, "learning_rate": 0.0005, "loss": 1.366783857345581, "step": 27590 }, { "epoch": 2.175088687426094, "grad_norm": 0.4708639681339264, "learning_rate": 0.0005, "loss": 1.1536586284637451, "step": 27591 }, { "epoch": 2.1751675206937326, "grad_norm": 0.6354984045028687, "learning_rate": 0.0005, "loss": 1.2283836603164673, "step": 27592 }, { "epoch": 2.175246353961372, "grad_norm": 0.45278143882751465, "learning_rate": 0.0005, "loss": 1.2775393724441528, "step": 27593 }, { "epoch": 2.1753251872290105, "grad_norm": 0.45724573731422424, "learning_rate": 0.0005, "loss": 1.279259443283081, "step": 27594 }, { "epoch": 2.1754040204966496, "grad_norm": 0.5508413910865784, "learning_rate": 0.0005, "loss": 1.5379363298416138, "step": 27595 }, { "epoch": 2.1754828537642887, "grad_norm": 0.4033348858356476, "learning_rate": 0.0005, "loss": 1.3253198862075806, "step": 27596 }, { "epoch": 2.1755616870319274, "grad_norm": 0.5461651682853699, "learning_rate": 0.0005, "loss": 1.3504350185394287, "step": 27597 }, { "epoch": 2.1756405202995666, "grad_norm": 0.8761882781982422, "learning_rate": 0.0005, "loss": 1.2466838359832764, "step": 27598 }, { "epoch": 2.1757193535672052, "grad_norm": 0.5071178078651428, "learning_rate": 0.0005, "loss": 1.331453800201416, "step": 27599 }, { "epoch": 2.1757981868348444, "grad_norm": 0.5325493812561035, "learning_rate": 0.0005, "loss": 1.261171579360962, "step": 27600 }, { "epoch": 2.175877020102483, "grad_norm": 0.4493563771247864, "learning_rate": 0.0005, "loss": 1.1730892658233643, "step": 27601 }, { "epoch": 2.175955853370122, "grad_norm": 0.5815888047218323, "learning_rate": 0.0005, "loss": 1.392591118812561, "step": 27602 }, { "epoch": 2.1760346866377613, "grad_norm": 0.5177597403526306, "learning_rate": 0.0005, "loss": 1.3805838823318481, "step": 27603 }, { "epoch": 2.1761135199054, "grad_norm": 0.556453287601471, "learning_rate": 0.0005, "loss": 1.5164711475372314, "step": 27604 }, { "epoch": 2.176192353173039, "grad_norm": 0.6750112175941467, "learning_rate": 0.0005, "loss": 1.4850624799728394, "step": 27605 }, { "epoch": 2.176271186440678, "grad_norm": 0.46068090200424194, "learning_rate": 0.0005, "loss": 1.249948263168335, "step": 27606 }, { "epoch": 2.176350019708317, "grad_norm": 0.44659173488616943, "learning_rate": 0.0005, "loss": 1.369436264038086, "step": 27607 }, { "epoch": 2.1764288529759557, "grad_norm": 0.5710673332214355, "learning_rate": 0.0005, "loss": 1.3180794715881348, "step": 27608 }, { "epoch": 2.176507686243595, "grad_norm": 0.44866397976875305, "learning_rate": 0.0005, "loss": 1.4536375999450684, "step": 27609 }, { "epoch": 2.176586519511234, "grad_norm": 0.46902772784233093, "learning_rate": 0.0005, "loss": 1.2094831466674805, "step": 27610 }, { "epoch": 2.1766653527788726, "grad_norm": 0.49361497163772583, "learning_rate": 0.0005, "loss": 1.4575138092041016, "step": 27611 }, { "epoch": 2.1767441860465118, "grad_norm": 0.5155346989631653, "learning_rate": 0.0005, "loss": 1.5970345735549927, "step": 27612 }, { "epoch": 2.1768230193141505, "grad_norm": 0.4414619505405426, "learning_rate": 0.0005, "loss": 1.3596667051315308, "step": 27613 }, { "epoch": 2.1769018525817896, "grad_norm": 0.5183904767036438, "learning_rate": 0.0005, "loss": 1.1844736337661743, "step": 27614 }, { "epoch": 2.1769806858494283, "grad_norm": 0.5917145609855652, "learning_rate": 0.0005, "loss": 1.5281623601913452, "step": 27615 }, { "epoch": 2.1770595191170674, "grad_norm": 0.494838684797287, "learning_rate": 0.0005, "loss": 1.274242877960205, "step": 27616 }, { "epoch": 2.1771383523847065, "grad_norm": 0.46972280740737915, "learning_rate": 0.0005, "loss": 1.241025447845459, "step": 27617 }, { "epoch": 2.1772171856523452, "grad_norm": 0.49321869015693665, "learning_rate": 0.0005, "loss": 1.1704301834106445, "step": 27618 }, { "epoch": 2.1772960189199844, "grad_norm": 0.4246821701526642, "learning_rate": 0.0005, "loss": 1.3185691833496094, "step": 27619 }, { "epoch": 2.177374852187623, "grad_norm": 0.5493200421333313, "learning_rate": 0.0005, "loss": 1.3165397644042969, "step": 27620 }, { "epoch": 2.177453685455262, "grad_norm": 0.5985090136528015, "learning_rate": 0.0005, "loss": 1.4462189674377441, "step": 27621 }, { "epoch": 2.177532518722901, "grad_norm": 0.4625054895877838, "learning_rate": 0.0005, "loss": 1.3981496095657349, "step": 27622 }, { "epoch": 2.17761135199054, "grad_norm": 0.5549979209899902, "learning_rate": 0.0005, "loss": 1.3929076194763184, "step": 27623 }, { "epoch": 2.177690185258179, "grad_norm": 0.5136436820030212, "learning_rate": 0.0005, "loss": 1.3519010543823242, "step": 27624 }, { "epoch": 2.177769018525818, "grad_norm": 0.5368277430534363, "learning_rate": 0.0005, "loss": 1.1255407333374023, "step": 27625 }, { "epoch": 2.177847851793457, "grad_norm": 0.5366101264953613, "learning_rate": 0.0005, "loss": 1.5252673625946045, "step": 27626 }, { "epoch": 2.1779266850610957, "grad_norm": 0.4249308705329895, "learning_rate": 0.0005, "loss": 1.2418856620788574, "step": 27627 }, { "epoch": 2.178005518328735, "grad_norm": 0.6976162195205688, "learning_rate": 0.0005, "loss": 1.4082496166229248, "step": 27628 }, { "epoch": 2.1780843515963735, "grad_norm": 0.5312796831130981, "learning_rate": 0.0005, "loss": 1.1329810619354248, "step": 27629 }, { "epoch": 2.1781631848640126, "grad_norm": 0.45619991421699524, "learning_rate": 0.0005, "loss": 1.2151200771331787, "step": 27630 }, { "epoch": 2.1782420181316517, "grad_norm": 0.44485655426979065, "learning_rate": 0.0005, "loss": 1.3562127351760864, "step": 27631 }, { "epoch": 2.1783208513992904, "grad_norm": 0.4563349187374115, "learning_rate": 0.0005, "loss": 1.412502646446228, "step": 27632 }, { "epoch": 2.1783996846669296, "grad_norm": 0.5149260759353638, "learning_rate": 0.0005, "loss": 1.4460983276367188, "step": 27633 }, { "epoch": 2.1784785179345683, "grad_norm": 0.5163233280181885, "learning_rate": 0.0005, "loss": 1.2747533321380615, "step": 27634 }, { "epoch": 2.1785573512022074, "grad_norm": 0.5192527770996094, "learning_rate": 0.0005, "loss": 1.255420207977295, "step": 27635 }, { "epoch": 2.178636184469846, "grad_norm": 0.4596096873283386, "learning_rate": 0.0005, "loss": 1.3003183603286743, "step": 27636 }, { "epoch": 2.178715017737485, "grad_norm": 0.6593555212020874, "learning_rate": 0.0005, "loss": 1.5684133768081665, "step": 27637 }, { "epoch": 2.1787938510051243, "grad_norm": 3.0441524982452393, "learning_rate": 0.0005, "loss": 1.4521965980529785, "step": 27638 }, { "epoch": 2.178872684272763, "grad_norm": 0.5590920448303223, "learning_rate": 0.0005, "loss": 1.3530832529067993, "step": 27639 }, { "epoch": 2.178951517540402, "grad_norm": 0.4735488295555115, "learning_rate": 0.0005, "loss": 1.2720997333526611, "step": 27640 }, { "epoch": 2.179030350808041, "grad_norm": 0.5170111060142517, "learning_rate": 0.0005, "loss": 1.4199914932250977, "step": 27641 }, { "epoch": 2.17910918407568, "grad_norm": 0.6276448369026184, "learning_rate": 0.0005, "loss": 1.2999762296676636, "step": 27642 }, { "epoch": 2.1791880173433187, "grad_norm": 0.5769432783126831, "learning_rate": 0.0005, "loss": 1.4733651876449585, "step": 27643 }, { "epoch": 2.179266850610958, "grad_norm": 0.502501904964447, "learning_rate": 0.0005, "loss": 1.2786762714385986, "step": 27644 }, { "epoch": 2.179345683878597, "grad_norm": 0.41500526666641235, "learning_rate": 0.0005, "loss": 1.2306265830993652, "step": 27645 }, { "epoch": 2.1794245171462356, "grad_norm": 0.4267703890800476, "learning_rate": 0.0005, "loss": 1.103479027748108, "step": 27646 }, { "epoch": 2.1795033504138748, "grad_norm": 0.5669747591018677, "learning_rate": 0.0005, "loss": 1.322683334350586, "step": 27647 }, { "epoch": 2.1795821836815135, "grad_norm": 0.4611818790435791, "learning_rate": 0.0005, "loss": 1.4514983892440796, "step": 27648 }, { "epoch": 2.1796610169491526, "grad_norm": 0.45423421263694763, "learning_rate": 0.0005, "loss": 1.1857774257659912, "step": 27649 }, { "epoch": 2.1797398502167913, "grad_norm": 0.45535868406295776, "learning_rate": 0.0005, "loss": 1.3071507215499878, "step": 27650 }, { "epoch": 2.1798186834844304, "grad_norm": 0.5080641508102417, "learning_rate": 0.0005, "loss": 1.2339204549789429, "step": 27651 }, { "epoch": 2.1798975167520696, "grad_norm": 0.5249460339546204, "learning_rate": 0.0005, "loss": 1.495373249053955, "step": 27652 }, { "epoch": 2.1799763500197082, "grad_norm": 0.5041049122810364, "learning_rate": 0.0005, "loss": 1.2777587175369263, "step": 27653 }, { "epoch": 2.1800551832873474, "grad_norm": 0.5891790390014648, "learning_rate": 0.0005, "loss": 1.6040997505187988, "step": 27654 }, { "epoch": 2.180134016554986, "grad_norm": 0.4701104164123535, "learning_rate": 0.0005, "loss": 1.354749083518982, "step": 27655 }, { "epoch": 2.180212849822625, "grad_norm": 0.4358569383621216, "learning_rate": 0.0005, "loss": 1.243260145187378, "step": 27656 }, { "epoch": 2.180291683090264, "grad_norm": 0.4510538876056671, "learning_rate": 0.0005, "loss": 1.1543437242507935, "step": 27657 }, { "epoch": 2.180370516357903, "grad_norm": 0.46469616889953613, "learning_rate": 0.0005, "loss": 1.2994678020477295, "step": 27658 }, { "epoch": 2.180449349625542, "grad_norm": 0.4904063940048218, "learning_rate": 0.0005, "loss": 1.5047954320907593, "step": 27659 }, { "epoch": 2.180528182893181, "grad_norm": 0.47130975127220154, "learning_rate": 0.0005, "loss": 1.4935688972473145, "step": 27660 }, { "epoch": 2.18060701616082, "grad_norm": 0.5166906118392944, "learning_rate": 0.0005, "loss": 1.3979263305664062, "step": 27661 }, { "epoch": 2.1806858494284587, "grad_norm": 0.5441630482673645, "learning_rate": 0.0005, "loss": 1.2357468605041504, "step": 27662 }, { "epoch": 2.180764682696098, "grad_norm": 0.5389992594718933, "learning_rate": 0.0005, "loss": 1.516142725944519, "step": 27663 }, { "epoch": 2.1808435159637365, "grad_norm": 0.4335763156414032, "learning_rate": 0.0005, "loss": 1.2646028995513916, "step": 27664 }, { "epoch": 2.1809223492313756, "grad_norm": 0.6444676518440247, "learning_rate": 0.0005, "loss": 1.3825103044509888, "step": 27665 }, { "epoch": 2.1810011824990148, "grad_norm": 0.6920354962348938, "learning_rate": 0.0005, "loss": 1.5000321865081787, "step": 27666 }, { "epoch": 2.1810800157666534, "grad_norm": 0.3735381066799164, "learning_rate": 0.0005, "loss": 1.08680260181427, "step": 27667 }, { "epoch": 2.1811588490342926, "grad_norm": 0.5200391411781311, "learning_rate": 0.0005, "loss": 1.4083906412124634, "step": 27668 }, { "epoch": 2.1812376823019313, "grad_norm": 0.5711115002632141, "learning_rate": 0.0005, "loss": 1.5322073698043823, "step": 27669 }, { "epoch": 2.1813165155695704, "grad_norm": 0.47897347807884216, "learning_rate": 0.0005, "loss": 1.2976932525634766, "step": 27670 }, { "epoch": 2.181395348837209, "grad_norm": 0.38750821352005005, "learning_rate": 0.0005, "loss": 1.2387890815734863, "step": 27671 }, { "epoch": 2.1814741821048482, "grad_norm": 0.5675501823425293, "learning_rate": 0.0005, "loss": 1.3850998878479004, "step": 27672 }, { "epoch": 2.1815530153724874, "grad_norm": 0.5166848301887512, "learning_rate": 0.0005, "loss": 1.5364718437194824, "step": 27673 }, { "epoch": 2.181631848640126, "grad_norm": 0.42702269554138184, "learning_rate": 0.0005, "loss": 1.283437967300415, "step": 27674 }, { "epoch": 2.181710681907765, "grad_norm": 0.5550701022148132, "learning_rate": 0.0005, "loss": 1.568805456161499, "step": 27675 }, { "epoch": 2.181789515175404, "grad_norm": 0.5374664068222046, "learning_rate": 0.0005, "loss": 1.341552495956421, "step": 27676 }, { "epoch": 2.181868348443043, "grad_norm": 0.48991432785987854, "learning_rate": 0.0005, "loss": 1.5392396450042725, "step": 27677 }, { "epoch": 2.181947181710682, "grad_norm": 2.5054690837860107, "learning_rate": 0.0005, "loss": 1.4839539527893066, "step": 27678 }, { "epoch": 2.182026014978321, "grad_norm": 0.5036318898200989, "learning_rate": 0.0005, "loss": 1.2680221796035767, "step": 27679 }, { "epoch": 2.18210484824596, "grad_norm": 0.4600680470466614, "learning_rate": 0.0005, "loss": 1.2146286964416504, "step": 27680 }, { "epoch": 2.1821836815135986, "grad_norm": 0.4764317274093628, "learning_rate": 0.0005, "loss": 1.501007080078125, "step": 27681 }, { "epoch": 2.182262514781238, "grad_norm": 0.6786947250366211, "learning_rate": 0.0005, "loss": 1.4325916767120361, "step": 27682 }, { "epoch": 2.1823413480488765, "grad_norm": 0.3761128783226013, "learning_rate": 0.0005, "loss": 1.0941821336746216, "step": 27683 }, { "epoch": 2.1824201813165156, "grad_norm": 0.524448037147522, "learning_rate": 0.0005, "loss": 1.3483102321624756, "step": 27684 }, { "epoch": 2.1824990145841543, "grad_norm": 0.4767543375492096, "learning_rate": 0.0005, "loss": 1.3113863468170166, "step": 27685 }, { "epoch": 2.1825778478517934, "grad_norm": 0.5446077585220337, "learning_rate": 0.0005, "loss": 1.339738368988037, "step": 27686 }, { "epoch": 2.1826566811194326, "grad_norm": 0.572858452796936, "learning_rate": 0.0005, "loss": 1.3647276163101196, "step": 27687 }, { "epoch": 2.1827355143870713, "grad_norm": 0.5312337279319763, "learning_rate": 0.0005, "loss": 1.4095325469970703, "step": 27688 }, { "epoch": 2.1828143476547104, "grad_norm": 0.7269866466522217, "learning_rate": 0.0005, "loss": 1.2298945188522339, "step": 27689 }, { "epoch": 2.182893180922349, "grad_norm": 0.4780610203742981, "learning_rate": 0.0005, "loss": 1.2754275798797607, "step": 27690 }, { "epoch": 2.182972014189988, "grad_norm": 0.509300172328949, "learning_rate": 0.0005, "loss": 1.255280613899231, "step": 27691 }, { "epoch": 2.1830508474576273, "grad_norm": 0.5419343113899231, "learning_rate": 0.0005, "loss": 1.2533175945281982, "step": 27692 }, { "epoch": 2.183129680725266, "grad_norm": 0.5093937516212463, "learning_rate": 0.0005, "loss": 1.337254285812378, "step": 27693 }, { "epoch": 2.183208513992905, "grad_norm": 0.46612393856048584, "learning_rate": 0.0005, "loss": 1.271531581878662, "step": 27694 }, { "epoch": 2.183287347260544, "grad_norm": 0.4068555235862732, "learning_rate": 0.0005, "loss": 1.0949132442474365, "step": 27695 }, { "epoch": 2.183366180528183, "grad_norm": 0.5252244472503662, "learning_rate": 0.0005, "loss": 1.2819585800170898, "step": 27696 }, { "epoch": 2.1834450137958217, "grad_norm": 0.3704591989517212, "learning_rate": 0.0005, "loss": 1.074916124343872, "step": 27697 }, { "epoch": 2.183523847063461, "grad_norm": 0.40643662214279175, "learning_rate": 0.0005, "loss": 1.2607924938201904, "step": 27698 }, { "epoch": 2.1836026803310995, "grad_norm": 0.7642269730567932, "learning_rate": 0.0005, "loss": 1.2855381965637207, "step": 27699 }, { "epoch": 2.1836815135987386, "grad_norm": 0.4493327736854553, "learning_rate": 0.0005, "loss": 1.163886308670044, "step": 27700 }, { "epoch": 2.1837603468663778, "grad_norm": 0.4192567467689514, "learning_rate": 0.0005, "loss": 1.3088490962982178, "step": 27701 }, { "epoch": 2.1838391801340165, "grad_norm": 0.41711023449897766, "learning_rate": 0.0005, "loss": 1.1446614265441895, "step": 27702 }, { "epoch": 2.1839180134016556, "grad_norm": 0.42565780878067017, "learning_rate": 0.0005, "loss": 1.3711488246917725, "step": 27703 }, { "epoch": 2.1839968466692943, "grad_norm": 0.42619577050209045, "learning_rate": 0.0005, "loss": 1.2836185693740845, "step": 27704 }, { "epoch": 2.1840756799369334, "grad_norm": 0.4920169413089752, "learning_rate": 0.0005, "loss": 1.2760883569717407, "step": 27705 }, { "epoch": 2.1841545132045725, "grad_norm": 0.4234945774078369, "learning_rate": 0.0005, "loss": 1.2630122900009155, "step": 27706 }, { "epoch": 2.1842333464722112, "grad_norm": 0.6795536279678345, "learning_rate": 0.0005, "loss": 1.394885540008545, "step": 27707 }, { "epoch": 2.1843121797398504, "grad_norm": 0.4940604567527771, "learning_rate": 0.0005, "loss": 1.380946397781372, "step": 27708 }, { "epoch": 2.184391013007489, "grad_norm": 0.5195534229278564, "learning_rate": 0.0005, "loss": 1.3816691637039185, "step": 27709 }, { "epoch": 2.184469846275128, "grad_norm": 0.5331981778144836, "learning_rate": 0.0005, "loss": 1.480778694152832, "step": 27710 }, { "epoch": 2.184548679542767, "grad_norm": 0.4758094251155853, "learning_rate": 0.0005, "loss": 1.4604027271270752, "step": 27711 }, { "epoch": 2.184627512810406, "grad_norm": 0.4909934997558594, "learning_rate": 0.0005, "loss": 1.2851133346557617, "step": 27712 }, { "epoch": 2.1847063460780447, "grad_norm": 0.46651187539100647, "learning_rate": 0.0005, "loss": 1.232893943786621, "step": 27713 }, { "epoch": 2.184785179345684, "grad_norm": 0.46942567825317383, "learning_rate": 0.0005, "loss": 1.2691830396652222, "step": 27714 }, { "epoch": 2.184864012613323, "grad_norm": 0.6054465770721436, "learning_rate": 0.0005, "loss": 1.394193172454834, "step": 27715 }, { "epoch": 2.1849428458809617, "grad_norm": 0.5066328644752502, "learning_rate": 0.0005, "loss": 1.20823335647583, "step": 27716 }, { "epoch": 2.185021679148601, "grad_norm": 0.5449123382568359, "learning_rate": 0.0005, "loss": 1.2806026935577393, "step": 27717 }, { "epoch": 2.1851005124162395, "grad_norm": 0.5993679761886597, "learning_rate": 0.0005, "loss": 1.3063628673553467, "step": 27718 }, { "epoch": 2.1851793456838786, "grad_norm": 0.5265068411827087, "learning_rate": 0.0005, "loss": 1.2717852592468262, "step": 27719 }, { "epoch": 2.1852581789515177, "grad_norm": 0.5441412329673767, "learning_rate": 0.0005, "loss": 1.398644208908081, "step": 27720 }, { "epoch": 2.1853370122191564, "grad_norm": 0.4457579255104065, "learning_rate": 0.0005, "loss": 1.1251857280731201, "step": 27721 }, { "epoch": 2.1854158454867956, "grad_norm": 0.458389550447464, "learning_rate": 0.0005, "loss": 1.3302277326583862, "step": 27722 }, { "epoch": 2.1854946787544343, "grad_norm": 0.803477942943573, "learning_rate": 0.0005, "loss": 1.4678056240081787, "step": 27723 }, { "epoch": 2.1855735120220734, "grad_norm": 0.6065056324005127, "learning_rate": 0.0005, "loss": 1.2989871501922607, "step": 27724 }, { "epoch": 2.185652345289712, "grad_norm": 0.5019353628158569, "learning_rate": 0.0005, "loss": 1.3296856880187988, "step": 27725 }, { "epoch": 2.185731178557351, "grad_norm": 0.7894207835197449, "learning_rate": 0.0005, "loss": 1.5745421648025513, "step": 27726 }, { "epoch": 2.1858100118249904, "grad_norm": 0.4566679000854492, "learning_rate": 0.0005, "loss": 1.2711156606674194, "step": 27727 }, { "epoch": 2.185888845092629, "grad_norm": 0.5708795785903931, "learning_rate": 0.0005, "loss": 1.2684450149536133, "step": 27728 }, { "epoch": 2.185967678360268, "grad_norm": 0.6953080892562866, "learning_rate": 0.0005, "loss": 1.4225122928619385, "step": 27729 }, { "epoch": 2.186046511627907, "grad_norm": 0.4775546193122864, "learning_rate": 0.0005, "loss": 1.129538655281067, "step": 27730 }, { "epoch": 2.186125344895546, "grad_norm": 0.44954246282577515, "learning_rate": 0.0005, "loss": 1.2709577083587646, "step": 27731 }, { "epoch": 2.1862041781631847, "grad_norm": 0.6675820350646973, "learning_rate": 0.0005, "loss": 1.3878173828125, "step": 27732 }, { "epoch": 2.186283011430824, "grad_norm": 0.5447203516960144, "learning_rate": 0.0005, "loss": 1.2424778938293457, "step": 27733 }, { "epoch": 2.186361844698463, "grad_norm": 0.5563379526138306, "learning_rate": 0.0005, "loss": 1.2602083683013916, "step": 27734 }, { "epoch": 2.1864406779661016, "grad_norm": 0.42287933826446533, "learning_rate": 0.0005, "loss": 1.2191188335418701, "step": 27735 }, { "epoch": 2.1865195112337408, "grad_norm": 0.9027332663536072, "learning_rate": 0.0005, "loss": 1.348097324371338, "step": 27736 }, { "epoch": 2.1865983445013795, "grad_norm": 0.5264265537261963, "learning_rate": 0.0005, "loss": 1.2499160766601562, "step": 27737 }, { "epoch": 2.1866771777690186, "grad_norm": 0.6319389939308167, "learning_rate": 0.0005, "loss": 1.5991322994232178, "step": 27738 }, { "epoch": 2.1867560110366573, "grad_norm": 0.7383378148078918, "learning_rate": 0.0005, "loss": 1.4906294345855713, "step": 27739 }, { "epoch": 2.1868348443042964, "grad_norm": 0.44784799218177795, "learning_rate": 0.0005, "loss": 1.19917893409729, "step": 27740 }, { "epoch": 2.1869136775719356, "grad_norm": 0.5673123598098755, "learning_rate": 0.0005, "loss": 1.3631813526153564, "step": 27741 }, { "epoch": 2.1869925108395742, "grad_norm": 0.542415201663971, "learning_rate": 0.0005, "loss": 1.3756437301635742, "step": 27742 }, { "epoch": 2.1870713441072134, "grad_norm": 7.745838165283203, "learning_rate": 0.0005, "loss": 1.7548190355300903, "step": 27743 }, { "epoch": 2.187150177374852, "grad_norm": 0.5640819072723389, "learning_rate": 0.0005, "loss": 1.2831377983093262, "step": 27744 }, { "epoch": 2.187229010642491, "grad_norm": 0.5193212032318115, "learning_rate": 0.0005, "loss": 1.5080392360687256, "step": 27745 }, { "epoch": 2.18730784391013, "grad_norm": 0.6116464734077454, "learning_rate": 0.0005, "loss": 1.418459177017212, "step": 27746 }, { "epoch": 2.187386677177769, "grad_norm": 0.6327936053276062, "learning_rate": 0.0005, "loss": 1.4977573156356812, "step": 27747 }, { "epoch": 2.187465510445408, "grad_norm": 0.5322327017784119, "learning_rate": 0.0005, "loss": 1.480426549911499, "step": 27748 }, { "epoch": 2.187544343713047, "grad_norm": 0.5469965934753418, "learning_rate": 0.0005, "loss": 1.434739112854004, "step": 27749 }, { "epoch": 2.187623176980686, "grad_norm": 0.5407512784004211, "learning_rate": 0.0005, "loss": 1.356715440750122, "step": 27750 }, { "epoch": 2.1877020102483247, "grad_norm": 0.4603252410888672, "learning_rate": 0.0005, "loss": 1.3839421272277832, "step": 27751 }, { "epoch": 2.187780843515964, "grad_norm": 1.0692445039749146, "learning_rate": 0.0005, "loss": 1.3059179782867432, "step": 27752 }, { "epoch": 2.1878596767836025, "grad_norm": 0.6250012516975403, "learning_rate": 0.0005, "loss": 1.511838436126709, "step": 27753 }, { "epoch": 2.1879385100512416, "grad_norm": 0.45684266090393066, "learning_rate": 0.0005, "loss": 1.4880704879760742, "step": 27754 }, { "epoch": 2.1880173433188808, "grad_norm": 0.5728824734687805, "learning_rate": 0.0005, "loss": 1.2465887069702148, "step": 27755 }, { "epoch": 2.1880961765865194, "grad_norm": 0.5025216341018677, "learning_rate": 0.0005, "loss": 1.5192410945892334, "step": 27756 }, { "epoch": 2.1881750098541586, "grad_norm": 0.567696750164032, "learning_rate": 0.0005, "loss": 1.4759483337402344, "step": 27757 }, { "epoch": 2.1882538431217973, "grad_norm": 0.5759769082069397, "learning_rate": 0.0005, "loss": 1.3225669860839844, "step": 27758 }, { "epoch": 2.1883326763894364, "grad_norm": 0.4783301055431366, "learning_rate": 0.0005, "loss": 1.316308856010437, "step": 27759 }, { "epoch": 2.188411509657075, "grad_norm": 0.6725620627403259, "learning_rate": 0.0005, "loss": 1.5425996780395508, "step": 27760 }, { "epoch": 2.1884903429247142, "grad_norm": 0.43339115381240845, "learning_rate": 0.0005, "loss": 1.3676786422729492, "step": 27761 }, { "epoch": 2.1885691761923534, "grad_norm": 0.5214246511459351, "learning_rate": 0.0005, "loss": 1.2629644870758057, "step": 27762 }, { "epoch": 2.188648009459992, "grad_norm": 0.6745139360427856, "learning_rate": 0.0005, "loss": 1.4266982078552246, "step": 27763 }, { "epoch": 2.188726842727631, "grad_norm": 0.5112407803535461, "learning_rate": 0.0005, "loss": 1.3032844066619873, "step": 27764 }, { "epoch": 2.18880567599527, "grad_norm": 0.4930895268917084, "learning_rate": 0.0005, "loss": 1.2430802583694458, "step": 27765 }, { "epoch": 2.188884509262909, "grad_norm": 0.5410493016242981, "learning_rate": 0.0005, "loss": 1.345458745956421, "step": 27766 }, { "epoch": 2.1889633425305477, "grad_norm": 0.5137597322463989, "learning_rate": 0.0005, "loss": 1.388688325881958, "step": 27767 }, { "epoch": 2.189042175798187, "grad_norm": 0.36626315116882324, "learning_rate": 0.0005, "loss": 1.0307035446166992, "step": 27768 }, { "epoch": 2.189121009065826, "grad_norm": 0.7316939830780029, "learning_rate": 0.0005, "loss": 1.4711179733276367, "step": 27769 }, { "epoch": 2.1891998423334647, "grad_norm": 0.5830821394920349, "learning_rate": 0.0005, "loss": 1.4417293071746826, "step": 27770 }, { "epoch": 2.189278675601104, "grad_norm": 0.5594969987869263, "learning_rate": 0.0005, "loss": 1.4572117328643799, "step": 27771 }, { "epoch": 2.1893575088687425, "grad_norm": 0.5155237317085266, "learning_rate": 0.0005, "loss": 1.3399802446365356, "step": 27772 }, { "epoch": 2.1894363421363816, "grad_norm": 0.5062478184700012, "learning_rate": 0.0005, "loss": 1.2338764667510986, "step": 27773 }, { "epoch": 2.1895151754040203, "grad_norm": 0.44250231981277466, "learning_rate": 0.0005, "loss": 1.1257375478744507, "step": 27774 }, { "epoch": 2.1895940086716594, "grad_norm": 0.44209834933280945, "learning_rate": 0.0005, "loss": 1.3019788265228271, "step": 27775 }, { "epoch": 2.1896728419392986, "grad_norm": 1.131515622138977, "learning_rate": 0.0005, "loss": 1.435795545578003, "step": 27776 }, { "epoch": 2.1897516752069373, "grad_norm": 0.47751787304878235, "learning_rate": 0.0005, "loss": 1.325796127319336, "step": 27777 }, { "epoch": 2.1898305084745764, "grad_norm": 0.5255373120307922, "learning_rate": 0.0005, "loss": 1.2323622703552246, "step": 27778 }, { "epoch": 2.189909341742215, "grad_norm": 0.4297317564487457, "learning_rate": 0.0005, "loss": 1.1307127475738525, "step": 27779 }, { "epoch": 2.189988175009854, "grad_norm": 0.5358938574790955, "learning_rate": 0.0005, "loss": 1.473414421081543, "step": 27780 }, { "epoch": 2.190067008277493, "grad_norm": 0.5185040235519409, "learning_rate": 0.0005, "loss": 1.2580742835998535, "step": 27781 }, { "epoch": 2.190145841545132, "grad_norm": 0.4454087018966675, "learning_rate": 0.0005, "loss": 1.4056241512298584, "step": 27782 }, { "epoch": 2.190224674812771, "grad_norm": 1.0049656629562378, "learning_rate": 0.0005, "loss": 1.271371603012085, "step": 27783 }, { "epoch": 2.19030350808041, "grad_norm": 0.4429449141025543, "learning_rate": 0.0005, "loss": 1.27479887008667, "step": 27784 }, { "epoch": 2.190382341348049, "grad_norm": 0.5950801968574524, "learning_rate": 0.0005, "loss": 1.3200147151947021, "step": 27785 }, { "epoch": 2.1904611746156877, "grad_norm": 0.5794751644134521, "learning_rate": 0.0005, "loss": 1.4722471237182617, "step": 27786 }, { "epoch": 2.190540007883327, "grad_norm": 0.6374685168266296, "learning_rate": 0.0005, "loss": 1.6517953872680664, "step": 27787 }, { "epoch": 2.1906188411509655, "grad_norm": 0.48910099267959595, "learning_rate": 0.0005, "loss": 1.284630537033081, "step": 27788 }, { "epoch": 2.1906976744186046, "grad_norm": 0.5433534979820251, "learning_rate": 0.0005, "loss": 1.3223607540130615, "step": 27789 }, { "epoch": 2.1907765076862438, "grad_norm": 0.5917786955833435, "learning_rate": 0.0005, "loss": 1.3559645414352417, "step": 27790 }, { "epoch": 2.1908553409538825, "grad_norm": 0.5698956847190857, "learning_rate": 0.0005, "loss": 1.2869715690612793, "step": 27791 }, { "epoch": 2.1909341742215216, "grad_norm": 0.47180792689323425, "learning_rate": 0.0005, "loss": 1.4927091598510742, "step": 27792 }, { "epoch": 2.1910130074891603, "grad_norm": 0.6114998459815979, "learning_rate": 0.0005, "loss": 1.4327150583267212, "step": 27793 }, { "epoch": 2.1910918407567994, "grad_norm": 0.44503524899482727, "learning_rate": 0.0005, "loss": 1.2468047142028809, "step": 27794 }, { "epoch": 2.1911706740244385, "grad_norm": 0.4936981797218323, "learning_rate": 0.0005, "loss": 1.325667142868042, "step": 27795 }, { "epoch": 2.1912495072920772, "grad_norm": 0.5287659764289856, "learning_rate": 0.0005, "loss": 1.3126200437545776, "step": 27796 }, { "epoch": 2.1913283405597164, "grad_norm": 0.5476386547088623, "learning_rate": 0.0005, "loss": 1.4491124153137207, "step": 27797 }, { "epoch": 2.191407173827355, "grad_norm": 0.49607110023498535, "learning_rate": 0.0005, "loss": 1.2501037120819092, "step": 27798 }, { "epoch": 2.191486007094994, "grad_norm": 0.4707038402557373, "learning_rate": 0.0005, "loss": 1.3699522018432617, "step": 27799 }, { "epoch": 2.191564840362633, "grad_norm": 0.4899224638938904, "learning_rate": 0.0005, "loss": 1.3743363618850708, "step": 27800 }, { "epoch": 2.191643673630272, "grad_norm": 0.6282480955123901, "learning_rate": 0.0005, "loss": 1.2408411502838135, "step": 27801 }, { "epoch": 2.1917225068979107, "grad_norm": 0.8222194314002991, "learning_rate": 0.0005, "loss": 1.6779954433441162, "step": 27802 }, { "epoch": 2.19180134016555, "grad_norm": 0.3601199984550476, "learning_rate": 0.0005, "loss": 1.1345767974853516, "step": 27803 }, { "epoch": 2.191880173433189, "grad_norm": 0.478230744600296, "learning_rate": 0.0005, "loss": 1.2592988014221191, "step": 27804 }, { "epoch": 2.1919590067008277, "grad_norm": 0.4928836226463318, "learning_rate": 0.0005, "loss": 1.3915033340454102, "step": 27805 }, { "epoch": 2.192037839968467, "grad_norm": 0.4408017694950104, "learning_rate": 0.0005, "loss": 1.305987000465393, "step": 27806 }, { "epoch": 2.1921166732361055, "grad_norm": 0.491658478975296, "learning_rate": 0.0005, "loss": 1.592307209968567, "step": 27807 }, { "epoch": 2.1921955065037446, "grad_norm": 0.6413040161132812, "learning_rate": 0.0005, "loss": 1.6362318992614746, "step": 27808 }, { "epoch": 2.1922743397713838, "grad_norm": 0.5647026300430298, "learning_rate": 0.0005, "loss": 1.2283484935760498, "step": 27809 }, { "epoch": 2.1923531730390224, "grad_norm": 0.6236225366592407, "learning_rate": 0.0005, "loss": 1.2524120807647705, "step": 27810 }, { "epoch": 2.1924320063066616, "grad_norm": 0.4788973033428192, "learning_rate": 0.0005, "loss": 1.144381046295166, "step": 27811 }, { "epoch": 2.1925108395743003, "grad_norm": 0.5034769177436829, "learning_rate": 0.0005, "loss": 1.3596726655960083, "step": 27812 }, { "epoch": 2.1925896728419394, "grad_norm": 0.49705931544303894, "learning_rate": 0.0005, "loss": 1.4693809747695923, "step": 27813 }, { "epoch": 2.192668506109578, "grad_norm": 0.5223347544670105, "learning_rate": 0.0005, "loss": 1.4503681659698486, "step": 27814 }, { "epoch": 2.192747339377217, "grad_norm": 0.4743332862854004, "learning_rate": 0.0005, "loss": 1.1646113395690918, "step": 27815 }, { "epoch": 2.192826172644856, "grad_norm": 1.3140103816986084, "learning_rate": 0.0005, "loss": 1.7365353107452393, "step": 27816 }, { "epoch": 2.192905005912495, "grad_norm": 0.43793797492980957, "learning_rate": 0.0005, "loss": 1.2720794677734375, "step": 27817 }, { "epoch": 2.192983839180134, "grad_norm": 0.37919679284095764, "learning_rate": 0.0005, "loss": 1.1524677276611328, "step": 27818 }, { "epoch": 2.193062672447773, "grad_norm": 0.44660019874572754, "learning_rate": 0.0005, "loss": 1.3214728832244873, "step": 27819 }, { "epoch": 2.193141505715412, "grad_norm": 0.5110461115837097, "learning_rate": 0.0005, "loss": 1.3000774383544922, "step": 27820 }, { "epoch": 2.1932203389830507, "grad_norm": 0.5338295698165894, "learning_rate": 0.0005, "loss": 1.4550660848617554, "step": 27821 }, { "epoch": 2.19329917225069, "grad_norm": 0.41133442521095276, "learning_rate": 0.0005, "loss": 1.3235058784484863, "step": 27822 }, { "epoch": 2.193378005518329, "grad_norm": 0.5288609266281128, "learning_rate": 0.0005, "loss": 1.2981685400009155, "step": 27823 }, { "epoch": 2.1934568387859676, "grad_norm": 0.4138929843902588, "learning_rate": 0.0005, "loss": 1.216118335723877, "step": 27824 }, { "epoch": 2.1935356720536068, "grad_norm": 0.5476473569869995, "learning_rate": 0.0005, "loss": 1.4834017753601074, "step": 27825 }, { "epoch": 2.1936145053212455, "grad_norm": 0.5490391850471497, "learning_rate": 0.0005, "loss": 1.190150499343872, "step": 27826 }, { "epoch": 2.1936933385888846, "grad_norm": 0.5210945010185242, "learning_rate": 0.0005, "loss": 1.2592507600784302, "step": 27827 }, { "epoch": 2.1937721718565233, "grad_norm": 0.5080657601356506, "learning_rate": 0.0005, "loss": 1.292677640914917, "step": 27828 }, { "epoch": 2.1938510051241624, "grad_norm": 0.445519357919693, "learning_rate": 0.0005, "loss": 1.3817509412765503, "step": 27829 }, { "epoch": 2.193929838391801, "grad_norm": 0.46419423818588257, "learning_rate": 0.0005, "loss": 1.3586746454238892, "step": 27830 }, { "epoch": 2.1940086716594402, "grad_norm": 0.5315458178520203, "learning_rate": 0.0005, "loss": 1.1092090606689453, "step": 27831 }, { "epoch": 2.1940875049270794, "grad_norm": 0.4773588180541992, "learning_rate": 0.0005, "loss": 1.331829309463501, "step": 27832 }, { "epoch": 2.194166338194718, "grad_norm": 0.5996008515357971, "learning_rate": 0.0005, "loss": 1.5074864625930786, "step": 27833 }, { "epoch": 2.194245171462357, "grad_norm": 0.49655798077583313, "learning_rate": 0.0005, "loss": 1.4293363094329834, "step": 27834 }, { "epoch": 2.194324004729996, "grad_norm": 0.5204288363456726, "learning_rate": 0.0005, "loss": 1.4169412851333618, "step": 27835 }, { "epoch": 2.194402837997635, "grad_norm": 0.4569747745990753, "learning_rate": 0.0005, "loss": 1.2108293771743774, "step": 27836 }, { "epoch": 2.194481671265274, "grad_norm": 0.39021891355514526, "learning_rate": 0.0005, "loss": 1.0132306814193726, "step": 27837 }, { "epoch": 2.194560504532913, "grad_norm": 0.5794723033905029, "learning_rate": 0.0005, "loss": 1.3778666257858276, "step": 27838 }, { "epoch": 2.194639337800552, "grad_norm": 0.5563450455665588, "learning_rate": 0.0005, "loss": 1.4099339246749878, "step": 27839 }, { "epoch": 2.1947181710681907, "grad_norm": 0.4138978123664856, "learning_rate": 0.0005, "loss": 1.4068458080291748, "step": 27840 }, { "epoch": 2.19479700433583, "grad_norm": 0.5578877329826355, "learning_rate": 0.0005, "loss": 1.478403091430664, "step": 27841 }, { "epoch": 2.1948758376034685, "grad_norm": 0.49691399931907654, "learning_rate": 0.0005, "loss": 1.296120047569275, "step": 27842 }, { "epoch": 2.1949546708711076, "grad_norm": 0.4705130457878113, "learning_rate": 0.0005, "loss": 1.370605230331421, "step": 27843 }, { "epoch": 2.1950335041387463, "grad_norm": 0.5934165120124817, "learning_rate": 0.0005, "loss": 1.3161338567733765, "step": 27844 }, { "epoch": 2.1951123374063854, "grad_norm": 0.5189238786697388, "learning_rate": 0.0005, "loss": 1.3906548023223877, "step": 27845 }, { "epoch": 2.1951911706740246, "grad_norm": 0.42332276701927185, "learning_rate": 0.0005, "loss": 1.3015575408935547, "step": 27846 }, { "epoch": 2.1952700039416633, "grad_norm": 0.6186092495918274, "learning_rate": 0.0005, "loss": 1.4960711002349854, "step": 27847 }, { "epoch": 2.1953488372093024, "grad_norm": 0.45271480083465576, "learning_rate": 0.0005, "loss": 1.1098026037216187, "step": 27848 }, { "epoch": 2.195427670476941, "grad_norm": 0.6334086060523987, "learning_rate": 0.0005, "loss": 1.3407896757125854, "step": 27849 }, { "epoch": 2.1955065037445802, "grad_norm": 0.4570189416408539, "learning_rate": 0.0005, "loss": 1.1727328300476074, "step": 27850 }, { "epoch": 2.1955853370122194, "grad_norm": 0.4419551491737366, "learning_rate": 0.0005, "loss": 1.2676388025283813, "step": 27851 }, { "epoch": 2.195664170279858, "grad_norm": 0.633565366268158, "learning_rate": 0.0005, "loss": 1.3328675031661987, "step": 27852 }, { "epoch": 2.195743003547497, "grad_norm": 0.5359633564949036, "learning_rate": 0.0005, "loss": 1.4696424007415771, "step": 27853 }, { "epoch": 2.195821836815136, "grad_norm": 0.47989410161972046, "learning_rate": 0.0005, "loss": 1.2342946529388428, "step": 27854 }, { "epoch": 2.195900670082775, "grad_norm": 0.49607381224632263, "learning_rate": 0.0005, "loss": 1.4369629621505737, "step": 27855 }, { "epoch": 2.1959795033504137, "grad_norm": 0.5072454214096069, "learning_rate": 0.0005, "loss": 1.1530345678329468, "step": 27856 }, { "epoch": 2.196058336618053, "grad_norm": 0.561635434627533, "learning_rate": 0.0005, "loss": 1.2299081087112427, "step": 27857 }, { "epoch": 2.196137169885692, "grad_norm": 0.4598703384399414, "learning_rate": 0.0005, "loss": 1.359130620956421, "step": 27858 }, { "epoch": 2.1962160031533307, "grad_norm": 0.5413108468055725, "learning_rate": 0.0005, "loss": 1.4448230266571045, "step": 27859 }, { "epoch": 2.19629483642097, "grad_norm": 0.5562036633491516, "learning_rate": 0.0005, "loss": 1.2914361953735352, "step": 27860 }, { "epoch": 2.1963736696886085, "grad_norm": 0.5403205752372742, "learning_rate": 0.0005, "loss": 1.3750169277191162, "step": 27861 }, { "epoch": 2.1964525029562476, "grad_norm": 0.42434924840927124, "learning_rate": 0.0005, "loss": 1.40468430519104, "step": 27862 }, { "epoch": 2.1965313362238863, "grad_norm": 0.4843977093696594, "learning_rate": 0.0005, "loss": 1.1436368227005005, "step": 27863 }, { "epoch": 2.1966101694915254, "grad_norm": 0.6149324178695679, "learning_rate": 0.0005, "loss": 1.5714011192321777, "step": 27864 }, { "epoch": 2.1966890027591646, "grad_norm": 0.5008436441421509, "learning_rate": 0.0005, "loss": 1.2375198602676392, "step": 27865 }, { "epoch": 2.1967678360268033, "grad_norm": 0.8136425018310547, "learning_rate": 0.0005, "loss": 1.466689109802246, "step": 27866 }, { "epoch": 2.1968466692944424, "grad_norm": 0.5504489541053772, "learning_rate": 0.0005, "loss": 1.4808542728424072, "step": 27867 }, { "epoch": 2.196925502562081, "grad_norm": 0.5073716044425964, "learning_rate": 0.0005, "loss": 1.1805098056793213, "step": 27868 }, { "epoch": 2.19700433582972, "grad_norm": 0.39206093549728394, "learning_rate": 0.0005, "loss": 1.1030974388122559, "step": 27869 }, { "epoch": 2.197083169097359, "grad_norm": 0.6166918873786926, "learning_rate": 0.0005, "loss": 1.3237240314483643, "step": 27870 }, { "epoch": 2.197162002364998, "grad_norm": 0.41232630610466003, "learning_rate": 0.0005, "loss": 1.1181750297546387, "step": 27871 }, { "epoch": 2.197240835632637, "grad_norm": 0.45969876646995544, "learning_rate": 0.0005, "loss": 1.2695735692977905, "step": 27872 }, { "epoch": 2.197319668900276, "grad_norm": 0.5083996057510376, "learning_rate": 0.0005, "loss": 1.4279650449752808, "step": 27873 }, { "epoch": 2.197398502167915, "grad_norm": 0.406352162361145, "learning_rate": 0.0005, "loss": 1.2588129043579102, "step": 27874 }, { "epoch": 2.1974773354355537, "grad_norm": 0.4551678001880646, "learning_rate": 0.0005, "loss": 1.372770071029663, "step": 27875 }, { "epoch": 2.197556168703193, "grad_norm": 0.4556359648704529, "learning_rate": 0.0005, "loss": 1.3726125955581665, "step": 27876 }, { "epoch": 2.1976350019708315, "grad_norm": 0.4956900477409363, "learning_rate": 0.0005, "loss": 1.3108341693878174, "step": 27877 }, { "epoch": 2.1977138352384706, "grad_norm": 0.5356830954551697, "learning_rate": 0.0005, "loss": 1.4167072772979736, "step": 27878 }, { "epoch": 2.1977926685061098, "grad_norm": 0.5013542175292969, "learning_rate": 0.0005, "loss": 1.2842164039611816, "step": 27879 }, { "epoch": 2.1978715017737485, "grad_norm": 0.4207148253917694, "learning_rate": 0.0005, "loss": 1.3471002578735352, "step": 27880 }, { "epoch": 2.1979503350413876, "grad_norm": 0.5412474274635315, "learning_rate": 0.0005, "loss": 1.4844999313354492, "step": 27881 }, { "epoch": 2.1980291683090263, "grad_norm": 0.4476868510246277, "learning_rate": 0.0005, "loss": 1.4738998413085938, "step": 27882 }, { "epoch": 2.1981080015766654, "grad_norm": 0.629207968711853, "learning_rate": 0.0005, "loss": 1.5242938995361328, "step": 27883 }, { "epoch": 2.198186834844304, "grad_norm": 0.4392184019088745, "learning_rate": 0.0005, "loss": 1.2814953327178955, "step": 27884 }, { "epoch": 2.1982656681119432, "grad_norm": 0.4584081172943115, "learning_rate": 0.0005, "loss": 1.3891340494155884, "step": 27885 }, { "epoch": 2.1983445013795824, "grad_norm": 0.5080301761627197, "learning_rate": 0.0005, "loss": 1.3674256801605225, "step": 27886 }, { "epoch": 2.198423334647221, "grad_norm": 0.5673618316650391, "learning_rate": 0.0005, "loss": 1.3824388980865479, "step": 27887 }, { "epoch": 2.19850216791486, "grad_norm": 0.595508337020874, "learning_rate": 0.0005, "loss": 1.3369214534759521, "step": 27888 }, { "epoch": 2.198581001182499, "grad_norm": 0.4721008241176605, "learning_rate": 0.0005, "loss": 1.474686622619629, "step": 27889 }, { "epoch": 2.198659834450138, "grad_norm": 0.5458170771598816, "learning_rate": 0.0005, "loss": 1.4463481903076172, "step": 27890 }, { "epoch": 2.1987386677177767, "grad_norm": 0.45045116543769836, "learning_rate": 0.0005, "loss": 1.381020426750183, "step": 27891 }, { "epoch": 2.198817500985416, "grad_norm": 0.4522446095943451, "learning_rate": 0.0005, "loss": 1.3416600227355957, "step": 27892 }, { "epoch": 2.198896334253055, "grad_norm": 0.5563418865203857, "learning_rate": 0.0005, "loss": 1.3220547437667847, "step": 27893 }, { "epoch": 2.1989751675206937, "grad_norm": 0.47135835886001587, "learning_rate": 0.0005, "loss": 1.3414733409881592, "step": 27894 }, { "epoch": 2.199054000788333, "grad_norm": 0.5069609880447388, "learning_rate": 0.0005, "loss": 1.532949686050415, "step": 27895 }, { "epoch": 2.1991328340559715, "grad_norm": 0.4538988173007965, "learning_rate": 0.0005, "loss": 1.3581438064575195, "step": 27896 }, { "epoch": 2.1992116673236106, "grad_norm": 25.63484764099121, "learning_rate": 0.0005, "loss": 1.3800382614135742, "step": 27897 }, { "epoch": 2.1992905005912493, "grad_norm": 0.4155508279800415, "learning_rate": 0.0005, "loss": 1.137019395828247, "step": 27898 }, { "epoch": 2.1993693338588884, "grad_norm": 0.48936930298805237, "learning_rate": 0.0005, "loss": 1.4391981363296509, "step": 27899 }, { "epoch": 2.1994481671265276, "grad_norm": 0.5655322670936584, "learning_rate": 0.0005, "loss": 1.3450230360031128, "step": 27900 }, { "epoch": 2.1995270003941663, "grad_norm": 0.5710260272026062, "learning_rate": 0.0005, "loss": 1.4492791891098022, "step": 27901 }, { "epoch": 2.1996058336618054, "grad_norm": 0.6034753918647766, "learning_rate": 0.0005, "loss": 1.342924952507019, "step": 27902 }, { "epoch": 2.199684666929444, "grad_norm": 0.4820983111858368, "learning_rate": 0.0005, "loss": 1.4285314083099365, "step": 27903 }, { "epoch": 2.199763500197083, "grad_norm": 0.4744948744773865, "learning_rate": 0.0005, "loss": 1.1849052906036377, "step": 27904 }, { "epoch": 2.199842333464722, "grad_norm": 0.511855959892273, "learning_rate": 0.0005, "loss": 1.2725286483764648, "step": 27905 }, { "epoch": 2.199921166732361, "grad_norm": 0.5460681319236755, "learning_rate": 0.0005, "loss": 1.2454581260681152, "step": 27906 }, { "epoch": 2.2, "grad_norm": 0.5077573657035828, "learning_rate": 0.0005, "loss": 1.5344822406768799, "step": 27907 }, { "epoch": 2.200078833267639, "grad_norm": 0.5024734735488892, "learning_rate": 0.0005, "loss": 1.3837299346923828, "step": 27908 }, { "epoch": 2.200157666535278, "grad_norm": 0.5500574111938477, "learning_rate": 0.0005, "loss": 1.4019920825958252, "step": 27909 }, { "epoch": 2.2002364998029167, "grad_norm": 0.5230045318603516, "learning_rate": 0.0005, "loss": 1.2189364433288574, "step": 27910 }, { "epoch": 2.200315333070556, "grad_norm": 0.6307011246681213, "learning_rate": 0.0005, "loss": 1.474265694618225, "step": 27911 }, { "epoch": 2.2003941663381945, "grad_norm": 0.6690076589584351, "learning_rate": 0.0005, "loss": 1.566910982131958, "step": 27912 }, { "epoch": 2.2004729996058336, "grad_norm": 0.49873480200767517, "learning_rate": 0.0005, "loss": 1.2016183137893677, "step": 27913 }, { "epoch": 2.200551832873473, "grad_norm": 0.3968283534049988, "learning_rate": 0.0005, "loss": 1.1612341403961182, "step": 27914 }, { "epoch": 2.2006306661411115, "grad_norm": 0.5358133316040039, "learning_rate": 0.0005, "loss": 1.5486812591552734, "step": 27915 }, { "epoch": 2.2007094994087506, "grad_norm": 0.6065979599952698, "learning_rate": 0.0005, "loss": 1.5135891437530518, "step": 27916 }, { "epoch": 2.2007883326763893, "grad_norm": 0.48316118121147156, "learning_rate": 0.0005, "loss": 1.2349755764007568, "step": 27917 }, { "epoch": 2.2008671659440284, "grad_norm": 0.6204690933227539, "learning_rate": 0.0005, "loss": 1.564160943031311, "step": 27918 }, { "epoch": 2.200945999211667, "grad_norm": 0.531414806842804, "learning_rate": 0.0005, "loss": 1.5220924615859985, "step": 27919 }, { "epoch": 2.2010248324793062, "grad_norm": 0.4732803404331207, "learning_rate": 0.0005, "loss": 1.2413685321807861, "step": 27920 }, { "epoch": 2.2011036657469454, "grad_norm": 0.5202165246009827, "learning_rate": 0.0005, "loss": 1.2506434917449951, "step": 27921 }, { "epoch": 2.201182499014584, "grad_norm": 0.43099498748779297, "learning_rate": 0.0005, "loss": 1.2127492427825928, "step": 27922 }, { "epoch": 2.201261332282223, "grad_norm": 0.5542266368865967, "learning_rate": 0.0005, "loss": 1.1885403394699097, "step": 27923 }, { "epoch": 2.201340165549862, "grad_norm": 0.49249404668807983, "learning_rate": 0.0005, "loss": 1.2907211780548096, "step": 27924 }, { "epoch": 2.201418998817501, "grad_norm": 0.44364607334136963, "learning_rate": 0.0005, "loss": 1.3865842819213867, "step": 27925 }, { "epoch": 2.20149783208514, "grad_norm": 0.5388743877410889, "learning_rate": 0.0005, "loss": 1.3632603883743286, "step": 27926 }, { "epoch": 2.201576665352779, "grad_norm": 0.4214671850204468, "learning_rate": 0.0005, "loss": 1.4523632526397705, "step": 27927 }, { "epoch": 2.201655498620418, "grad_norm": 0.5730457901954651, "learning_rate": 0.0005, "loss": 1.353194236755371, "step": 27928 }, { "epoch": 2.2017343318880567, "grad_norm": 0.7648257613182068, "learning_rate": 0.0005, "loss": 1.5529224872589111, "step": 27929 }, { "epoch": 2.201813165155696, "grad_norm": 0.47907811403274536, "learning_rate": 0.0005, "loss": 1.4823110103607178, "step": 27930 }, { "epoch": 2.2018919984233345, "grad_norm": 0.5662774443626404, "learning_rate": 0.0005, "loss": 1.2972595691680908, "step": 27931 }, { "epoch": 2.2019708316909736, "grad_norm": 0.40501904487609863, "learning_rate": 0.0005, "loss": 1.3335200548171997, "step": 27932 }, { "epoch": 2.2020496649586123, "grad_norm": 0.4439844489097595, "learning_rate": 0.0005, "loss": 1.2854640483856201, "step": 27933 }, { "epoch": 2.2021284982262515, "grad_norm": 0.5320517420768738, "learning_rate": 0.0005, "loss": 1.369127631187439, "step": 27934 }, { "epoch": 2.2022073314938906, "grad_norm": 0.45257139205932617, "learning_rate": 0.0005, "loss": 1.082801342010498, "step": 27935 }, { "epoch": 2.2022861647615293, "grad_norm": 0.48150861263275146, "learning_rate": 0.0005, "loss": 1.2765254974365234, "step": 27936 }, { "epoch": 2.2023649980291684, "grad_norm": 0.5392519235610962, "learning_rate": 0.0005, "loss": 1.4111303091049194, "step": 27937 }, { "epoch": 2.202443831296807, "grad_norm": 0.52596116065979, "learning_rate": 0.0005, "loss": 1.1786730289459229, "step": 27938 }, { "epoch": 2.2025226645644462, "grad_norm": 0.453215628862381, "learning_rate": 0.0005, "loss": 1.247328519821167, "step": 27939 }, { "epoch": 2.2026014978320854, "grad_norm": 0.48248091340065, "learning_rate": 0.0005, "loss": 1.4002366065979004, "step": 27940 }, { "epoch": 2.202680331099724, "grad_norm": 0.5782668590545654, "learning_rate": 0.0005, "loss": 1.2750482559204102, "step": 27941 }, { "epoch": 2.202759164367363, "grad_norm": 0.41530686616897583, "learning_rate": 0.0005, "loss": 1.5102392435073853, "step": 27942 }, { "epoch": 2.202837997635002, "grad_norm": 0.5310244560241699, "learning_rate": 0.0005, "loss": 1.2519490718841553, "step": 27943 }, { "epoch": 2.202916830902641, "grad_norm": 0.5071327090263367, "learning_rate": 0.0005, "loss": 1.2948200702667236, "step": 27944 }, { "epoch": 2.2029956641702797, "grad_norm": 0.5181206464767456, "learning_rate": 0.0005, "loss": 1.282888650894165, "step": 27945 }, { "epoch": 2.203074497437919, "grad_norm": 0.5581212043762207, "learning_rate": 0.0005, "loss": 1.2756259441375732, "step": 27946 }, { "epoch": 2.2031533307055575, "grad_norm": 0.5511798858642578, "learning_rate": 0.0005, "loss": 1.38687264919281, "step": 27947 }, { "epoch": 2.2032321639731967, "grad_norm": 0.5220913887023926, "learning_rate": 0.0005, "loss": 1.3144744634628296, "step": 27948 }, { "epoch": 2.203310997240836, "grad_norm": 0.40891462564468384, "learning_rate": 0.0005, "loss": 1.218044638633728, "step": 27949 }, { "epoch": 2.2033898305084745, "grad_norm": 0.435507208108902, "learning_rate": 0.0005, "loss": 1.1646286249160767, "step": 27950 }, { "epoch": 2.2034686637761136, "grad_norm": 0.6300994753837585, "learning_rate": 0.0005, "loss": 1.2831320762634277, "step": 27951 }, { "epoch": 2.2035474970437523, "grad_norm": 0.4652246832847595, "learning_rate": 0.0005, "loss": 1.0464963912963867, "step": 27952 }, { "epoch": 2.2036263303113914, "grad_norm": 0.5755884051322937, "learning_rate": 0.0005, "loss": 1.3757524490356445, "step": 27953 }, { "epoch": 2.2037051635790306, "grad_norm": 0.548721194267273, "learning_rate": 0.0005, "loss": 1.2266013622283936, "step": 27954 }, { "epoch": 2.2037839968466693, "grad_norm": 0.5526302456855774, "learning_rate": 0.0005, "loss": 1.2881479263305664, "step": 27955 }, { "epoch": 2.2038628301143084, "grad_norm": 0.4472816586494446, "learning_rate": 0.0005, "loss": 1.462912917137146, "step": 27956 }, { "epoch": 2.203941663381947, "grad_norm": 0.4647897183895111, "learning_rate": 0.0005, "loss": 1.1530084609985352, "step": 27957 }, { "epoch": 2.204020496649586, "grad_norm": 0.45029720664024353, "learning_rate": 0.0005, "loss": 1.2376128435134888, "step": 27958 }, { "epoch": 2.204099329917225, "grad_norm": 0.4242222309112549, "learning_rate": 0.0005, "loss": 1.0556235313415527, "step": 27959 }, { "epoch": 2.204178163184864, "grad_norm": 0.6419163346290588, "learning_rate": 0.0005, "loss": 1.4132533073425293, "step": 27960 }, { "epoch": 2.2042569964525027, "grad_norm": 0.49611973762512207, "learning_rate": 0.0005, "loss": 1.3072911500930786, "step": 27961 }, { "epoch": 2.204335829720142, "grad_norm": 0.7645719647407532, "learning_rate": 0.0005, "loss": 1.5364333391189575, "step": 27962 }, { "epoch": 2.204414662987781, "grad_norm": 0.5286379456520081, "learning_rate": 0.0005, "loss": 1.317049503326416, "step": 27963 }, { "epoch": 2.2044934962554197, "grad_norm": 0.520850419998169, "learning_rate": 0.0005, "loss": 1.2069611549377441, "step": 27964 }, { "epoch": 2.204572329523059, "grad_norm": 0.5703724026679993, "learning_rate": 0.0005, "loss": 1.5907814502716064, "step": 27965 }, { "epoch": 2.2046511627906975, "grad_norm": 0.48705044388771057, "learning_rate": 0.0005, "loss": 1.266958475112915, "step": 27966 }, { "epoch": 2.2047299960583366, "grad_norm": 0.4983966052532196, "learning_rate": 0.0005, "loss": 1.208754062652588, "step": 27967 }, { "epoch": 2.2048088293259758, "grad_norm": 0.3760683536529541, "learning_rate": 0.0005, "loss": 1.1046453714370728, "step": 27968 }, { "epoch": 2.2048876625936145, "grad_norm": 0.5783560872077942, "learning_rate": 0.0005, "loss": 1.3443684577941895, "step": 27969 }, { "epoch": 2.2049664958612536, "grad_norm": 0.46287012100219727, "learning_rate": 0.0005, "loss": 1.4475656747817993, "step": 27970 }, { "epoch": 2.2050453291288923, "grad_norm": 0.5259210467338562, "learning_rate": 0.0005, "loss": 1.4079341888427734, "step": 27971 }, { "epoch": 2.2051241623965314, "grad_norm": 0.47268739342689514, "learning_rate": 0.0005, "loss": 1.2532222270965576, "step": 27972 }, { "epoch": 2.20520299566417, "grad_norm": 0.511793315410614, "learning_rate": 0.0005, "loss": 1.2503832578659058, "step": 27973 }, { "epoch": 2.2052818289318092, "grad_norm": 0.4951827824115753, "learning_rate": 0.0005, "loss": 1.37379789352417, "step": 27974 }, { "epoch": 2.2053606621994484, "grad_norm": 0.5887207984924316, "learning_rate": 0.0005, "loss": 1.532292366027832, "step": 27975 }, { "epoch": 2.205439495467087, "grad_norm": 0.4255489110946655, "learning_rate": 0.0005, "loss": 1.3057111501693726, "step": 27976 }, { "epoch": 2.205518328734726, "grad_norm": 0.5061526894569397, "learning_rate": 0.0005, "loss": 1.243502140045166, "step": 27977 }, { "epoch": 2.205597162002365, "grad_norm": 0.5652040839195251, "learning_rate": 0.0005, "loss": 1.5549230575561523, "step": 27978 }, { "epoch": 2.205675995270004, "grad_norm": 0.5237714648246765, "learning_rate": 0.0005, "loss": 1.2540475130081177, "step": 27979 }, { "epoch": 2.2057548285376427, "grad_norm": 0.43018239736557007, "learning_rate": 0.0005, "loss": 1.1495288610458374, "step": 27980 }, { "epoch": 2.205833661805282, "grad_norm": 2.218775510787964, "learning_rate": 0.0005, "loss": 1.618375539779663, "step": 27981 }, { "epoch": 2.205912495072921, "grad_norm": 0.5966813564300537, "learning_rate": 0.0005, "loss": 1.4298126697540283, "step": 27982 }, { "epoch": 2.2059913283405597, "grad_norm": 0.4319118857383728, "learning_rate": 0.0005, "loss": 1.3678388595581055, "step": 27983 }, { "epoch": 2.206070161608199, "grad_norm": 0.4704047739505768, "learning_rate": 0.0005, "loss": 1.2279331684112549, "step": 27984 }, { "epoch": 2.2061489948758375, "grad_norm": 0.7986894845962524, "learning_rate": 0.0005, "loss": 1.6484133005142212, "step": 27985 }, { "epoch": 2.2062278281434766, "grad_norm": 0.3979606330394745, "learning_rate": 0.0005, "loss": 1.070385456085205, "step": 27986 }, { "epoch": 2.2063066614111153, "grad_norm": 0.5162236094474792, "learning_rate": 0.0005, "loss": 1.0944013595581055, "step": 27987 }, { "epoch": 2.2063854946787544, "grad_norm": 0.536496639251709, "learning_rate": 0.0005, "loss": 1.3742318153381348, "step": 27988 }, { "epoch": 2.2064643279463936, "grad_norm": 0.4746531844139099, "learning_rate": 0.0005, "loss": 1.0902125835418701, "step": 27989 }, { "epoch": 2.2065431612140323, "grad_norm": 0.5610944032669067, "learning_rate": 0.0005, "loss": 1.1811714172363281, "step": 27990 }, { "epoch": 2.2066219944816714, "grad_norm": 0.49661463499069214, "learning_rate": 0.0005, "loss": 1.359764575958252, "step": 27991 }, { "epoch": 2.20670082774931, "grad_norm": 0.3853755593299866, "learning_rate": 0.0005, "loss": 1.143378734588623, "step": 27992 }, { "epoch": 2.2067796610169492, "grad_norm": 0.41014522314071655, "learning_rate": 0.0005, "loss": 1.1645015478134155, "step": 27993 }, { "epoch": 2.206858494284588, "grad_norm": 0.4540558457374573, "learning_rate": 0.0005, "loss": 1.4251859188079834, "step": 27994 }, { "epoch": 2.206937327552227, "grad_norm": 0.3665669560432434, "learning_rate": 0.0005, "loss": 1.0708119869232178, "step": 27995 }, { "epoch": 2.207016160819866, "grad_norm": 0.6369413137435913, "learning_rate": 0.0005, "loss": 1.0994585752487183, "step": 27996 }, { "epoch": 2.207094994087505, "grad_norm": 0.49055004119873047, "learning_rate": 0.0005, "loss": 1.178572654724121, "step": 27997 }, { "epoch": 2.207173827355144, "grad_norm": 0.5361788272857666, "learning_rate": 0.0005, "loss": 1.377532720565796, "step": 27998 }, { "epoch": 2.2072526606227827, "grad_norm": 0.5988501906394958, "learning_rate": 0.0005, "loss": 1.5263221263885498, "step": 27999 }, { "epoch": 2.207331493890422, "grad_norm": 0.6485698223114014, "learning_rate": 0.0005, "loss": 1.4180439710617065, "step": 28000 }, { "epoch": 2.2074103271580605, "grad_norm": 0.4824581444263458, "learning_rate": 0.0005, "loss": 1.3363704681396484, "step": 28001 }, { "epoch": 2.2074891604256996, "grad_norm": 0.4905555844306946, "learning_rate": 0.0005, "loss": 1.3184460401535034, "step": 28002 }, { "epoch": 2.207567993693339, "grad_norm": 0.4873829782009125, "learning_rate": 0.0005, "loss": 1.3552650213241577, "step": 28003 }, { "epoch": 2.2076468269609775, "grad_norm": 0.539918839931488, "learning_rate": 0.0005, "loss": 1.2325267791748047, "step": 28004 }, { "epoch": 2.2077256602286166, "grad_norm": 0.7169219255447388, "learning_rate": 0.0005, "loss": 1.430525541305542, "step": 28005 }, { "epoch": 2.2078044934962553, "grad_norm": 0.5983539819717407, "learning_rate": 0.0005, "loss": 1.5015358924865723, "step": 28006 }, { "epoch": 2.2078833267638944, "grad_norm": 0.493236780166626, "learning_rate": 0.0005, "loss": 1.4496347904205322, "step": 28007 }, { "epoch": 2.207962160031533, "grad_norm": 0.49347424507141113, "learning_rate": 0.0005, "loss": 1.3848580121994019, "step": 28008 }, { "epoch": 2.2080409932991722, "grad_norm": 0.6001780033111572, "learning_rate": 0.0005, "loss": 1.5404351949691772, "step": 28009 }, { "epoch": 2.2081198265668114, "grad_norm": 0.5167549848556519, "learning_rate": 0.0005, "loss": 1.4046458005905151, "step": 28010 }, { "epoch": 2.20819865983445, "grad_norm": 0.6308948397636414, "learning_rate": 0.0005, "loss": 1.2406442165374756, "step": 28011 }, { "epoch": 2.208277493102089, "grad_norm": 0.657170295715332, "learning_rate": 0.0005, "loss": 1.5343873500823975, "step": 28012 }, { "epoch": 2.208356326369728, "grad_norm": 0.538204550743103, "learning_rate": 0.0005, "loss": 1.4031250476837158, "step": 28013 }, { "epoch": 2.208435159637367, "grad_norm": 0.5138254165649414, "learning_rate": 0.0005, "loss": 1.3299063444137573, "step": 28014 }, { "epoch": 2.2085139929050057, "grad_norm": 0.5647298097610474, "learning_rate": 0.0005, "loss": 1.4045844078063965, "step": 28015 }, { "epoch": 2.208592826172645, "grad_norm": 0.5539336800575256, "learning_rate": 0.0005, "loss": 1.289568305015564, "step": 28016 }, { "epoch": 2.208671659440284, "grad_norm": 0.461311936378479, "learning_rate": 0.0005, "loss": 1.3925731182098389, "step": 28017 }, { "epoch": 2.2087504927079227, "grad_norm": 0.5228728652000427, "learning_rate": 0.0005, "loss": 1.3139212131500244, "step": 28018 }, { "epoch": 2.208829325975562, "grad_norm": 0.4998694658279419, "learning_rate": 0.0005, "loss": 1.2731215953826904, "step": 28019 }, { "epoch": 2.2089081592432005, "grad_norm": 0.4142064154148102, "learning_rate": 0.0005, "loss": 1.3926488161087036, "step": 28020 }, { "epoch": 2.2089869925108396, "grad_norm": 0.5278007388114929, "learning_rate": 0.0005, "loss": 1.4625635147094727, "step": 28021 }, { "epoch": 2.2090658257784783, "grad_norm": 0.5304992198944092, "learning_rate": 0.0005, "loss": 1.4073717594146729, "step": 28022 }, { "epoch": 2.2091446590461175, "grad_norm": 0.4569474756717682, "learning_rate": 0.0005, "loss": 1.4447705745697021, "step": 28023 }, { "epoch": 2.2092234923137566, "grad_norm": 0.5582717657089233, "learning_rate": 0.0005, "loss": 1.2524281740188599, "step": 28024 }, { "epoch": 2.2093023255813953, "grad_norm": 0.500990629196167, "learning_rate": 0.0005, "loss": 1.5127344131469727, "step": 28025 }, { "epoch": 2.2093811588490344, "grad_norm": 0.42168304324150085, "learning_rate": 0.0005, "loss": 1.2021474838256836, "step": 28026 }, { "epoch": 2.209459992116673, "grad_norm": 0.49987152218818665, "learning_rate": 0.0005, "loss": 1.4336270093917847, "step": 28027 }, { "epoch": 2.2095388253843122, "grad_norm": 0.4426795542240143, "learning_rate": 0.0005, "loss": 1.0962363481521606, "step": 28028 }, { "epoch": 2.209617658651951, "grad_norm": 0.45463988184928894, "learning_rate": 0.0005, "loss": 1.161238670349121, "step": 28029 }, { "epoch": 2.20969649191959, "grad_norm": 0.575139045715332, "learning_rate": 0.0005, "loss": 1.4847842454910278, "step": 28030 }, { "epoch": 2.209775325187229, "grad_norm": 0.46156883239746094, "learning_rate": 0.0005, "loss": 1.4824540615081787, "step": 28031 }, { "epoch": 2.209854158454868, "grad_norm": 0.5072803497314453, "learning_rate": 0.0005, "loss": 1.5318596363067627, "step": 28032 }, { "epoch": 2.209932991722507, "grad_norm": 1.4695745706558228, "learning_rate": 0.0005, "loss": 1.3433804512023926, "step": 28033 }, { "epoch": 2.2100118249901457, "grad_norm": 0.6229614019393921, "learning_rate": 0.0005, "loss": 1.5349102020263672, "step": 28034 }, { "epoch": 2.210090658257785, "grad_norm": 0.6344738602638245, "learning_rate": 0.0005, "loss": 1.2445603609085083, "step": 28035 }, { "epoch": 2.2101694915254235, "grad_norm": 0.4200132489204407, "learning_rate": 0.0005, "loss": 1.1776931285858154, "step": 28036 }, { "epoch": 2.2102483247930627, "grad_norm": 0.5297606587409973, "learning_rate": 0.0005, "loss": 1.428652048110962, "step": 28037 }, { "epoch": 2.210327158060702, "grad_norm": 0.5403637290000916, "learning_rate": 0.0005, "loss": 1.3777177333831787, "step": 28038 }, { "epoch": 2.2104059913283405, "grad_norm": 0.43586277961730957, "learning_rate": 0.0005, "loss": 1.3578076362609863, "step": 28039 }, { "epoch": 2.2104848245959796, "grad_norm": 1.0485551357269287, "learning_rate": 0.0005, "loss": 1.3840810060501099, "step": 28040 }, { "epoch": 2.2105636578636183, "grad_norm": 0.6019833087921143, "learning_rate": 0.0005, "loss": 1.3308100700378418, "step": 28041 }, { "epoch": 2.2106424911312574, "grad_norm": 0.4653358459472656, "learning_rate": 0.0005, "loss": 1.290482759475708, "step": 28042 }, { "epoch": 2.2107213243988966, "grad_norm": 0.5399667024612427, "learning_rate": 0.0005, "loss": 1.3420196771621704, "step": 28043 }, { "epoch": 2.2108001576665353, "grad_norm": 0.5353893637657166, "learning_rate": 0.0005, "loss": 1.266454815864563, "step": 28044 }, { "epoch": 2.2108789909341744, "grad_norm": 0.7604720592498779, "learning_rate": 0.0005, "loss": 1.207632303237915, "step": 28045 }, { "epoch": 2.210957824201813, "grad_norm": 0.5446599721908569, "learning_rate": 0.0005, "loss": 1.4303232431411743, "step": 28046 }, { "epoch": 2.211036657469452, "grad_norm": 0.674304187297821, "learning_rate": 0.0005, "loss": 1.4712111949920654, "step": 28047 }, { "epoch": 2.211115490737091, "grad_norm": 0.45469391345977783, "learning_rate": 0.0005, "loss": 1.3300009965896606, "step": 28048 }, { "epoch": 2.21119432400473, "grad_norm": 0.5080896019935608, "learning_rate": 0.0005, "loss": 1.3818384408950806, "step": 28049 }, { "epoch": 2.2112731572723687, "grad_norm": 0.5675098896026611, "learning_rate": 0.0005, "loss": 1.4139926433563232, "step": 28050 }, { "epoch": 2.211351990540008, "grad_norm": 0.529258668422699, "learning_rate": 0.0005, "loss": 1.371788740158081, "step": 28051 }, { "epoch": 2.211430823807647, "grad_norm": 0.5105621814727783, "learning_rate": 0.0005, "loss": 1.4415857791900635, "step": 28052 }, { "epoch": 2.2115096570752857, "grad_norm": 0.44461745023727417, "learning_rate": 0.0005, "loss": 1.2106163501739502, "step": 28053 }, { "epoch": 2.211588490342925, "grad_norm": 0.5219492316246033, "learning_rate": 0.0005, "loss": 1.4576524496078491, "step": 28054 }, { "epoch": 2.2116673236105635, "grad_norm": 0.6073534488677979, "learning_rate": 0.0005, "loss": 1.4472317695617676, "step": 28055 }, { "epoch": 2.2117461568782026, "grad_norm": 0.49920305609703064, "learning_rate": 0.0005, "loss": 1.4188088178634644, "step": 28056 }, { "epoch": 2.2118249901458418, "grad_norm": 12.985356330871582, "learning_rate": 0.0005, "loss": 1.5881452560424805, "step": 28057 }, { "epoch": 2.2119038234134805, "grad_norm": 0.5526951551437378, "learning_rate": 0.0005, "loss": 1.5435676574707031, "step": 28058 }, { "epoch": 2.2119826566811196, "grad_norm": 0.48161840438842773, "learning_rate": 0.0005, "loss": 1.3088462352752686, "step": 28059 }, { "epoch": 2.2120614899487583, "grad_norm": 0.40235626697540283, "learning_rate": 0.0005, "loss": 1.0839132070541382, "step": 28060 }, { "epoch": 2.2121403232163974, "grad_norm": 0.5586727261543274, "learning_rate": 0.0005, "loss": 1.4092905521392822, "step": 28061 }, { "epoch": 2.212219156484036, "grad_norm": 1.1850736141204834, "learning_rate": 0.0005, "loss": 1.437485933303833, "step": 28062 }, { "epoch": 2.2122979897516752, "grad_norm": 0.44489213824272156, "learning_rate": 0.0005, "loss": 1.2288309335708618, "step": 28063 }, { "epoch": 2.212376823019314, "grad_norm": 0.4109313189983368, "learning_rate": 0.0005, "loss": 1.0122376680374146, "step": 28064 }, { "epoch": 2.212455656286953, "grad_norm": 0.5917171239852905, "learning_rate": 0.0005, "loss": 1.402101993560791, "step": 28065 }, { "epoch": 2.212534489554592, "grad_norm": 0.3843154311180115, "learning_rate": 0.0005, "loss": 1.2661393880844116, "step": 28066 }, { "epoch": 2.212613322822231, "grad_norm": 0.38426873087882996, "learning_rate": 0.0005, "loss": 1.2229588031768799, "step": 28067 }, { "epoch": 2.21269215608987, "grad_norm": 0.527280330657959, "learning_rate": 0.0005, "loss": 1.3614712953567505, "step": 28068 }, { "epoch": 2.2127709893575087, "grad_norm": 0.384133517742157, "learning_rate": 0.0005, "loss": 1.2874548435211182, "step": 28069 }, { "epoch": 2.212849822625148, "grad_norm": 0.47788432240486145, "learning_rate": 0.0005, "loss": 1.1873348951339722, "step": 28070 }, { "epoch": 2.212928655892787, "grad_norm": 0.42300379276275635, "learning_rate": 0.0005, "loss": 1.3779428005218506, "step": 28071 }, { "epoch": 2.2130074891604257, "grad_norm": 0.38691842555999756, "learning_rate": 0.0005, "loss": 1.2290998697280884, "step": 28072 }, { "epoch": 2.213086322428065, "grad_norm": 0.46984755992889404, "learning_rate": 0.0005, "loss": 1.3638395071029663, "step": 28073 }, { "epoch": 2.2131651556957035, "grad_norm": 0.40839117765426636, "learning_rate": 0.0005, "loss": 1.2510731220245361, "step": 28074 }, { "epoch": 2.2132439889633426, "grad_norm": 0.5153862833976746, "learning_rate": 0.0005, "loss": 1.3431611061096191, "step": 28075 }, { "epoch": 2.2133228222309813, "grad_norm": 0.665321946144104, "learning_rate": 0.0005, "loss": 1.3979268074035645, "step": 28076 }, { "epoch": 2.2134016554986204, "grad_norm": 0.4817754328250885, "learning_rate": 0.0005, "loss": 1.310808539390564, "step": 28077 }, { "epoch": 2.213480488766259, "grad_norm": 0.7394447922706604, "learning_rate": 0.0005, "loss": 1.4349579811096191, "step": 28078 }, { "epoch": 2.2135593220338983, "grad_norm": 0.5321196913719177, "learning_rate": 0.0005, "loss": 1.280158519744873, "step": 28079 }, { "epoch": 2.2136381553015374, "grad_norm": 0.564336359500885, "learning_rate": 0.0005, "loss": 1.350539207458496, "step": 28080 }, { "epoch": 2.213716988569176, "grad_norm": 0.5981886982917786, "learning_rate": 0.0005, "loss": 1.236685037612915, "step": 28081 }, { "epoch": 2.2137958218368152, "grad_norm": 0.5868616700172424, "learning_rate": 0.0005, "loss": 1.513134241104126, "step": 28082 }, { "epoch": 2.213874655104454, "grad_norm": 0.5524038672447205, "learning_rate": 0.0005, "loss": 1.203886866569519, "step": 28083 }, { "epoch": 2.213953488372093, "grad_norm": 0.8516825437545776, "learning_rate": 0.0005, "loss": 1.2745492458343506, "step": 28084 }, { "epoch": 2.214032321639732, "grad_norm": 0.6143586039543152, "learning_rate": 0.0005, "loss": 1.4543328285217285, "step": 28085 }, { "epoch": 2.214111154907371, "grad_norm": 0.49751728773117065, "learning_rate": 0.0005, "loss": 1.3510386943817139, "step": 28086 }, { "epoch": 2.21418998817501, "grad_norm": 1.8000199794769287, "learning_rate": 0.0005, "loss": 1.1120846271514893, "step": 28087 }, { "epoch": 2.2142688214426487, "grad_norm": 0.486001193523407, "learning_rate": 0.0005, "loss": 1.547659158706665, "step": 28088 }, { "epoch": 2.214347654710288, "grad_norm": 0.5405229926109314, "learning_rate": 0.0005, "loss": 1.3785946369171143, "step": 28089 }, { "epoch": 2.2144264879779265, "grad_norm": 0.4477788209915161, "learning_rate": 0.0005, "loss": 1.62911057472229, "step": 28090 }, { "epoch": 2.2145053212455656, "grad_norm": 0.6208744645118713, "learning_rate": 0.0005, "loss": 1.2090966701507568, "step": 28091 }, { "epoch": 2.2145841545132043, "grad_norm": 0.45143330097198486, "learning_rate": 0.0005, "loss": 1.2573449611663818, "step": 28092 }, { "epoch": 2.2146629877808435, "grad_norm": 0.5282835364341736, "learning_rate": 0.0005, "loss": 1.4924376010894775, "step": 28093 }, { "epoch": 2.2147418210484826, "grad_norm": 0.4907190799713135, "learning_rate": 0.0005, "loss": 1.39773428440094, "step": 28094 }, { "epoch": 2.2148206543161213, "grad_norm": 0.49030452966690063, "learning_rate": 0.0005, "loss": 1.1910934448242188, "step": 28095 }, { "epoch": 2.2148994875837604, "grad_norm": 0.4959755539894104, "learning_rate": 0.0005, "loss": 1.3933666944503784, "step": 28096 }, { "epoch": 2.214978320851399, "grad_norm": 0.5047264099121094, "learning_rate": 0.0005, "loss": 1.3455413579940796, "step": 28097 }, { "epoch": 2.2150571541190383, "grad_norm": 0.7529411911964417, "learning_rate": 0.0005, "loss": 1.44376540184021, "step": 28098 }, { "epoch": 2.2151359873866774, "grad_norm": 0.5354129076004028, "learning_rate": 0.0005, "loss": 1.4541966915130615, "step": 28099 }, { "epoch": 2.215214820654316, "grad_norm": 0.618251383304596, "learning_rate": 0.0005, "loss": 1.3652982711791992, "step": 28100 }, { "epoch": 2.215293653921955, "grad_norm": 0.5532675981521606, "learning_rate": 0.0005, "loss": 1.25990891456604, "step": 28101 }, { "epoch": 2.215372487189594, "grad_norm": 0.5485885739326477, "learning_rate": 0.0005, "loss": 1.5330921411514282, "step": 28102 }, { "epoch": 2.215451320457233, "grad_norm": 0.5220564603805542, "learning_rate": 0.0005, "loss": 1.4930769205093384, "step": 28103 }, { "epoch": 2.2155301537248717, "grad_norm": 0.5069320201873779, "learning_rate": 0.0005, "loss": 1.301710844039917, "step": 28104 }, { "epoch": 2.215608986992511, "grad_norm": 0.46293169260025024, "learning_rate": 0.0005, "loss": 1.154249906539917, "step": 28105 }, { "epoch": 2.21568782026015, "grad_norm": 0.4661119878292084, "learning_rate": 0.0005, "loss": 1.2502645254135132, "step": 28106 }, { "epoch": 2.2157666535277887, "grad_norm": 0.4850916266441345, "learning_rate": 0.0005, "loss": 1.3760360479354858, "step": 28107 }, { "epoch": 2.215845486795428, "grad_norm": 0.5432465672492981, "learning_rate": 0.0005, "loss": 1.4553828239440918, "step": 28108 }, { "epoch": 2.2159243200630665, "grad_norm": 0.4546773433685303, "learning_rate": 0.0005, "loss": 1.233812928199768, "step": 28109 }, { "epoch": 2.2160031533307056, "grad_norm": 0.7259103655815125, "learning_rate": 0.0005, "loss": 1.4829655885696411, "step": 28110 }, { "epoch": 2.2160819865983443, "grad_norm": 0.5644806623458862, "learning_rate": 0.0005, "loss": 1.3526406288146973, "step": 28111 }, { "epoch": 2.2161608198659835, "grad_norm": 0.46524396538734436, "learning_rate": 0.0005, "loss": 1.304742455482483, "step": 28112 }, { "epoch": 2.2162396531336226, "grad_norm": 0.6269053816795349, "learning_rate": 0.0005, "loss": 1.2538272142410278, "step": 28113 }, { "epoch": 2.2163184864012613, "grad_norm": 0.4732273817062378, "learning_rate": 0.0005, "loss": 1.52284574508667, "step": 28114 }, { "epoch": 2.2163973196689004, "grad_norm": 0.7474794387817383, "learning_rate": 0.0005, "loss": 1.2704282999038696, "step": 28115 }, { "epoch": 2.216476152936539, "grad_norm": 0.5075207948684692, "learning_rate": 0.0005, "loss": 1.1117916107177734, "step": 28116 }, { "epoch": 2.2165549862041782, "grad_norm": 0.4679621458053589, "learning_rate": 0.0005, "loss": 1.2030097246170044, "step": 28117 }, { "epoch": 2.216633819471817, "grad_norm": 0.6169427037239075, "learning_rate": 0.0005, "loss": 1.3684508800506592, "step": 28118 }, { "epoch": 2.216712652739456, "grad_norm": 0.46635204553604126, "learning_rate": 0.0005, "loss": 1.5075790882110596, "step": 28119 }, { "epoch": 2.216791486007095, "grad_norm": 0.4695439040660858, "learning_rate": 0.0005, "loss": 1.4869831800460815, "step": 28120 }, { "epoch": 2.216870319274734, "grad_norm": 1.4147498607635498, "learning_rate": 0.0005, "loss": 1.4104816913604736, "step": 28121 }, { "epoch": 2.216949152542373, "grad_norm": 0.5292565822601318, "learning_rate": 0.0005, "loss": 1.3401508331298828, "step": 28122 }, { "epoch": 2.2170279858100117, "grad_norm": 0.5934411287307739, "learning_rate": 0.0005, "loss": 1.276687741279602, "step": 28123 }, { "epoch": 2.217106819077651, "grad_norm": 0.46997854113578796, "learning_rate": 0.0005, "loss": 1.3598456382751465, "step": 28124 }, { "epoch": 2.2171856523452895, "grad_norm": 0.526493489742279, "learning_rate": 0.0005, "loss": 1.5257024765014648, "step": 28125 }, { "epoch": 2.2172644856129287, "grad_norm": 0.5485840439796448, "learning_rate": 0.0005, "loss": 1.6485029458999634, "step": 28126 }, { "epoch": 2.217343318880568, "grad_norm": 0.4954574704170227, "learning_rate": 0.0005, "loss": 1.356999397277832, "step": 28127 }, { "epoch": 2.2174221521482065, "grad_norm": 0.471579909324646, "learning_rate": 0.0005, "loss": 1.4658643007278442, "step": 28128 }, { "epoch": 2.2175009854158456, "grad_norm": 0.44382575154304504, "learning_rate": 0.0005, "loss": 1.2718422412872314, "step": 28129 }, { "epoch": 2.2175798186834843, "grad_norm": 0.6475181579589844, "learning_rate": 0.0005, "loss": 1.2943742275238037, "step": 28130 }, { "epoch": 2.2176586519511234, "grad_norm": 0.6278185248374939, "learning_rate": 0.0005, "loss": 1.3446811437606812, "step": 28131 }, { "epoch": 2.217737485218762, "grad_norm": 0.8883947134017944, "learning_rate": 0.0005, "loss": 1.457927942276001, "step": 28132 }, { "epoch": 2.2178163184864013, "grad_norm": 0.6602464914321899, "learning_rate": 0.0005, "loss": 1.7482712268829346, "step": 28133 }, { "epoch": 2.2178951517540404, "grad_norm": 0.5333126187324524, "learning_rate": 0.0005, "loss": 1.3907525539398193, "step": 28134 }, { "epoch": 2.217973985021679, "grad_norm": 0.49170830845832825, "learning_rate": 0.0005, "loss": 1.296612024307251, "step": 28135 }, { "epoch": 2.218052818289318, "grad_norm": 0.5894104242324829, "learning_rate": 0.0005, "loss": 1.458662986755371, "step": 28136 }, { "epoch": 2.218131651556957, "grad_norm": 0.396893173456192, "learning_rate": 0.0005, "loss": 1.127937912940979, "step": 28137 }, { "epoch": 2.218210484824596, "grad_norm": 0.5916754007339478, "learning_rate": 0.0005, "loss": 1.352137565612793, "step": 28138 }, { "epoch": 2.2182893180922347, "grad_norm": 0.6038634181022644, "learning_rate": 0.0005, "loss": 1.4423249959945679, "step": 28139 }, { "epoch": 2.218368151359874, "grad_norm": 0.5208482146263123, "learning_rate": 0.0005, "loss": 1.2961286306381226, "step": 28140 }, { "epoch": 2.218446984627513, "grad_norm": 0.47418028116226196, "learning_rate": 0.0005, "loss": 1.2838451862335205, "step": 28141 }, { "epoch": 2.2185258178951517, "grad_norm": 0.42058125138282776, "learning_rate": 0.0005, "loss": 1.1942323446273804, "step": 28142 }, { "epoch": 2.218604651162791, "grad_norm": 0.7091530561447144, "learning_rate": 0.0005, "loss": 1.28257155418396, "step": 28143 }, { "epoch": 2.2186834844304295, "grad_norm": 0.5336547493934631, "learning_rate": 0.0005, "loss": 1.4118967056274414, "step": 28144 }, { "epoch": 2.2187623176980686, "grad_norm": 0.5204633474349976, "learning_rate": 0.0005, "loss": 1.2882006168365479, "step": 28145 }, { "epoch": 2.2188411509657073, "grad_norm": 0.5979517102241516, "learning_rate": 0.0005, "loss": 1.4242146015167236, "step": 28146 }, { "epoch": 2.2189199842333465, "grad_norm": 0.4670138657093048, "learning_rate": 0.0005, "loss": 1.3833692073822021, "step": 28147 }, { "epoch": 2.2189988175009856, "grad_norm": 0.8631899952888489, "learning_rate": 0.0005, "loss": 1.41217041015625, "step": 28148 }, { "epoch": 2.2190776507686243, "grad_norm": 0.504803478717804, "learning_rate": 0.0005, "loss": 1.2835404872894287, "step": 28149 }, { "epoch": 2.2191564840362634, "grad_norm": 0.6138834953308105, "learning_rate": 0.0005, "loss": 1.308610439300537, "step": 28150 }, { "epoch": 2.219235317303902, "grad_norm": 0.5118200778961182, "learning_rate": 0.0005, "loss": 1.2142682075500488, "step": 28151 }, { "epoch": 2.2193141505715412, "grad_norm": 0.5319210290908813, "learning_rate": 0.0005, "loss": 1.2096772193908691, "step": 28152 }, { "epoch": 2.21939298383918, "grad_norm": 0.5492151379585266, "learning_rate": 0.0005, "loss": 1.2173171043395996, "step": 28153 }, { "epoch": 2.219471817106819, "grad_norm": 0.47068917751312256, "learning_rate": 0.0005, "loss": 1.2471325397491455, "step": 28154 }, { "epoch": 2.219550650374458, "grad_norm": 0.5701571702957153, "learning_rate": 0.0005, "loss": 1.3589977025985718, "step": 28155 }, { "epoch": 2.219629483642097, "grad_norm": 0.44548991322517395, "learning_rate": 0.0005, "loss": 1.3023900985717773, "step": 28156 }, { "epoch": 2.219708316909736, "grad_norm": 0.5885774493217468, "learning_rate": 0.0005, "loss": 1.2720341682434082, "step": 28157 }, { "epoch": 2.2197871501773747, "grad_norm": 0.4237268269062042, "learning_rate": 0.0005, "loss": 1.1937141418457031, "step": 28158 }, { "epoch": 2.219865983445014, "grad_norm": 0.47499987483024597, "learning_rate": 0.0005, "loss": 1.2938203811645508, "step": 28159 }, { "epoch": 2.2199448167126525, "grad_norm": 0.5320918560028076, "learning_rate": 0.0005, "loss": 1.3760184049606323, "step": 28160 }, { "epoch": 2.2200236499802917, "grad_norm": 0.41404831409454346, "learning_rate": 0.0005, "loss": 1.4019091129302979, "step": 28161 }, { "epoch": 2.220102483247931, "grad_norm": 0.47037532925605774, "learning_rate": 0.0005, "loss": 1.2563773393630981, "step": 28162 }, { "epoch": 2.2201813165155695, "grad_norm": 0.466570645570755, "learning_rate": 0.0005, "loss": 1.2579448223114014, "step": 28163 }, { "epoch": 2.2202601497832086, "grad_norm": 0.5747981071472168, "learning_rate": 0.0005, "loss": 1.5786316394805908, "step": 28164 }, { "epoch": 2.2203389830508473, "grad_norm": 0.4834991991519928, "learning_rate": 0.0005, "loss": 1.2019810676574707, "step": 28165 }, { "epoch": 2.2204178163184864, "grad_norm": 0.5266520380973816, "learning_rate": 0.0005, "loss": 1.4031192064285278, "step": 28166 }, { "epoch": 2.220496649586125, "grad_norm": 0.45673274993896484, "learning_rate": 0.0005, "loss": 1.2711577415466309, "step": 28167 }, { "epoch": 2.2205754828537643, "grad_norm": 0.5015288591384888, "learning_rate": 0.0005, "loss": 1.5616204738616943, "step": 28168 }, { "epoch": 2.2206543161214034, "grad_norm": 0.6400312781333923, "learning_rate": 0.0005, "loss": 1.5672788619995117, "step": 28169 }, { "epoch": 2.220733149389042, "grad_norm": 0.5035474896430969, "learning_rate": 0.0005, "loss": 1.2721012830734253, "step": 28170 }, { "epoch": 2.2208119826566812, "grad_norm": 0.541779637336731, "learning_rate": 0.0005, "loss": 1.3571349382400513, "step": 28171 }, { "epoch": 2.22089081592432, "grad_norm": 0.8403359055519104, "learning_rate": 0.0005, "loss": 1.3854143619537354, "step": 28172 }, { "epoch": 2.220969649191959, "grad_norm": 0.5335131287574768, "learning_rate": 0.0005, "loss": 1.3230624198913574, "step": 28173 }, { "epoch": 2.221048482459598, "grad_norm": 0.46434444189071655, "learning_rate": 0.0005, "loss": 1.392005205154419, "step": 28174 }, { "epoch": 2.221127315727237, "grad_norm": 0.5662758350372314, "learning_rate": 0.0005, "loss": 1.4198079109191895, "step": 28175 }, { "epoch": 2.221206148994876, "grad_norm": 0.5227227807044983, "learning_rate": 0.0005, "loss": 1.3226711750030518, "step": 28176 }, { "epoch": 2.2212849822625147, "grad_norm": 0.49041205644607544, "learning_rate": 0.0005, "loss": 1.3492305278778076, "step": 28177 }, { "epoch": 2.221363815530154, "grad_norm": 0.470822811126709, "learning_rate": 0.0005, "loss": 1.4498542547225952, "step": 28178 }, { "epoch": 2.2214426487977925, "grad_norm": 0.5844265222549438, "learning_rate": 0.0005, "loss": 1.416053295135498, "step": 28179 }, { "epoch": 2.2215214820654317, "grad_norm": 0.5356816649436951, "learning_rate": 0.0005, "loss": 1.3318109512329102, "step": 28180 }, { "epoch": 2.2216003153330703, "grad_norm": 0.540799617767334, "learning_rate": 0.0005, "loss": 1.4614174365997314, "step": 28181 }, { "epoch": 2.2216791486007095, "grad_norm": 0.46169042587280273, "learning_rate": 0.0005, "loss": 1.3277201652526855, "step": 28182 }, { "epoch": 2.2217579818683486, "grad_norm": 0.5908328294754028, "learning_rate": 0.0005, "loss": 1.4217023849487305, "step": 28183 }, { "epoch": 2.2218368151359873, "grad_norm": 0.8367754220962524, "learning_rate": 0.0005, "loss": 1.4082396030426025, "step": 28184 }, { "epoch": 2.2219156484036264, "grad_norm": 0.5410225987434387, "learning_rate": 0.0005, "loss": 1.363398790359497, "step": 28185 }, { "epoch": 2.221994481671265, "grad_norm": 0.5548796653747559, "learning_rate": 0.0005, "loss": 1.531591773033142, "step": 28186 }, { "epoch": 2.2220733149389043, "grad_norm": 0.5119682550430298, "learning_rate": 0.0005, "loss": 1.3042234182357788, "step": 28187 }, { "epoch": 2.2221521482065434, "grad_norm": 0.6520854234695435, "learning_rate": 0.0005, "loss": 1.2277212142944336, "step": 28188 }, { "epoch": 2.222230981474182, "grad_norm": 0.6383541822433472, "learning_rate": 0.0005, "loss": 1.4900413751602173, "step": 28189 }, { "epoch": 2.222309814741821, "grad_norm": 0.41763240098953247, "learning_rate": 0.0005, "loss": 1.2300087213516235, "step": 28190 }, { "epoch": 2.22238864800946, "grad_norm": 0.4540647864341736, "learning_rate": 0.0005, "loss": 1.2228796482086182, "step": 28191 }, { "epoch": 2.222467481277099, "grad_norm": 0.507465124130249, "learning_rate": 0.0005, "loss": 1.1349124908447266, "step": 28192 }, { "epoch": 2.2225463145447377, "grad_norm": 0.38847726583480835, "learning_rate": 0.0005, "loss": 1.1798081398010254, "step": 28193 }, { "epoch": 2.222625147812377, "grad_norm": 0.5153325796127319, "learning_rate": 0.0005, "loss": 1.1173031330108643, "step": 28194 }, { "epoch": 2.2227039810800155, "grad_norm": 0.4788224399089813, "learning_rate": 0.0005, "loss": 1.5445729494094849, "step": 28195 }, { "epoch": 2.2227828143476547, "grad_norm": 0.4977625012397766, "learning_rate": 0.0005, "loss": 1.3034924268722534, "step": 28196 }, { "epoch": 2.222861647615294, "grad_norm": 0.5942710041999817, "learning_rate": 0.0005, "loss": 1.2546679973602295, "step": 28197 }, { "epoch": 2.2229404808829325, "grad_norm": 0.4123685359954834, "learning_rate": 0.0005, "loss": 1.2403056621551514, "step": 28198 }, { "epoch": 2.2230193141505716, "grad_norm": 0.4256189465522766, "learning_rate": 0.0005, "loss": 1.2768008708953857, "step": 28199 }, { "epoch": 2.2230981474182103, "grad_norm": 0.601397693157196, "learning_rate": 0.0005, "loss": 1.282243251800537, "step": 28200 }, { "epoch": 2.2231769806858495, "grad_norm": 0.5214434266090393, "learning_rate": 0.0005, "loss": 1.396662712097168, "step": 28201 }, { "epoch": 2.2232558139534886, "grad_norm": 0.5084714889526367, "learning_rate": 0.0005, "loss": 1.367687702178955, "step": 28202 }, { "epoch": 2.2233346472211273, "grad_norm": 0.5941954851150513, "learning_rate": 0.0005, "loss": 1.418924331665039, "step": 28203 }, { "epoch": 2.2234134804887664, "grad_norm": 0.4913848638534546, "learning_rate": 0.0005, "loss": 1.305260419845581, "step": 28204 }, { "epoch": 2.223492313756405, "grad_norm": 0.6644192337989807, "learning_rate": 0.0005, "loss": 1.5284085273742676, "step": 28205 }, { "epoch": 2.2235711470240442, "grad_norm": 0.5984609127044678, "learning_rate": 0.0005, "loss": 1.4483516216278076, "step": 28206 }, { "epoch": 2.223649980291683, "grad_norm": 0.5315937995910645, "learning_rate": 0.0005, "loss": 1.2130683660507202, "step": 28207 }, { "epoch": 2.223728813559322, "grad_norm": 0.5935890674591064, "learning_rate": 0.0005, "loss": 1.4894979000091553, "step": 28208 }, { "epoch": 2.2238076468269607, "grad_norm": 0.5965295433998108, "learning_rate": 0.0005, "loss": 1.4862635135650635, "step": 28209 }, { "epoch": 2.2238864800946, "grad_norm": 0.6013268828392029, "learning_rate": 0.0005, "loss": 1.565476655960083, "step": 28210 }, { "epoch": 2.223965313362239, "grad_norm": 0.4662242829799652, "learning_rate": 0.0005, "loss": 1.2661032676696777, "step": 28211 }, { "epoch": 2.2240441466298777, "grad_norm": 0.48014986515045166, "learning_rate": 0.0005, "loss": 1.4041411876678467, "step": 28212 }, { "epoch": 2.224122979897517, "grad_norm": 0.5316572189331055, "learning_rate": 0.0005, "loss": 1.327656626701355, "step": 28213 }, { "epoch": 2.2242018131651555, "grad_norm": 0.6000217199325562, "learning_rate": 0.0005, "loss": 1.5273069143295288, "step": 28214 }, { "epoch": 2.2242806464327947, "grad_norm": 0.5999860167503357, "learning_rate": 0.0005, "loss": 1.4322108030319214, "step": 28215 }, { "epoch": 2.224359479700434, "grad_norm": 4.821021556854248, "learning_rate": 0.0005, "loss": 1.5717363357543945, "step": 28216 }, { "epoch": 2.2244383129680725, "grad_norm": 0.4739372432231903, "learning_rate": 0.0005, "loss": 1.260483980178833, "step": 28217 }, { "epoch": 2.2245171462357116, "grad_norm": 0.5098862051963806, "learning_rate": 0.0005, "loss": 1.409429669380188, "step": 28218 }, { "epoch": 2.2245959795033503, "grad_norm": 0.5054876804351807, "learning_rate": 0.0005, "loss": 1.6051883697509766, "step": 28219 }, { "epoch": 2.2246748127709894, "grad_norm": 0.8098068833351135, "learning_rate": 0.0005, "loss": 1.378183364868164, "step": 28220 }, { "epoch": 2.224753646038628, "grad_norm": 0.5589635968208313, "learning_rate": 0.0005, "loss": 1.3851147890090942, "step": 28221 }, { "epoch": 2.2248324793062673, "grad_norm": 0.5878076553344727, "learning_rate": 0.0005, "loss": 1.3161544799804688, "step": 28222 }, { "epoch": 2.224911312573906, "grad_norm": 0.45599445700645447, "learning_rate": 0.0005, "loss": 1.224602222442627, "step": 28223 }, { "epoch": 2.224990145841545, "grad_norm": 0.5629924535751343, "learning_rate": 0.0005, "loss": 1.4576449394226074, "step": 28224 }, { "epoch": 2.225068979109184, "grad_norm": 0.5662726759910583, "learning_rate": 0.0005, "loss": 1.4521100521087646, "step": 28225 }, { "epoch": 2.225147812376823, "grad_norm": 0.7009643316268921, "learning_rate": 0.0005, "loss": 1.4131039381027222, "step": 28226 }, { "epoch": 2.225226645644462, "grad_norm": 0.391719788312912, "learning_rate": 0.0005, "loss": 1.2172620296478271, "step": 28227 }, { "epoch": 2.2253054789121007, "grad_norm": 0.5929821133613586, "learning_rate": 0.0005, "loss": 1.6102478504180908, "step": 28228 }, { "epoch": 2.22538431217974, "grad_norm": 0.5962135791778564, "learning_rate": 0.0005, "loss": 1.496955394744873, "step": 28229 }, { "epoch": 2.225463145447379, "grad_norm": 0.5167925357818604, "learning_rate": 0.0005, "loss": 1.468814730644226, "step": 28230 }, { "epoch": 2.2255419787150177, "grad_norm": 0.4921940863132477, "learning_rate": 0.0005, "loss": 1.3665499687194824, "step": 28231 }, { "epoch": 2.225620811982657, "grad_norm": 0.47416624426841736, "learning_rate": 0.0005, "loss": 1.2608864307403564, "step": 28232 }, { "epoch": 2.2256996452502955, "grad_norm": 0.4454050362110138, "learning_rate": 0.0005, "loss": 1.246201992034912, "step": 28233 }, { "epoch": 2.2257784785179346, "grad_norm": 0.7999304533004761, "learning_rate": 0.0005, "loss": 1.2366151809692383, "step": 28234 }, { "epoch": 2.2258573117855733, "grad_norm": 0.43021395802497864, "learning_rate": 0.0005, "loss": 1.0645344257354736, "step": 28235 }, { "epoch": 2.2259361450532125, "grad_norm": 0.44071102142333984, "learning_rate": 0.0005, "loss": 1.1233670711517334, "step": 28236 }, { "epoch": 2.2260149783208516, "grad_norm": 0.4166194796562195, "learning_rate": 0.0005, "loss": 1.3415379524230957, "step": 28237 }, { "epoch": 2.2260938115884903, "grad_norm": 0.508009672164917, "learning_rate": 0.0005, "loss": 1.32627534866333, "step": 28238 }, { "epoch": 2.2261726448561294, "grad_norm": 0.457265168428421, "learning_rate": 0.0005, "loss": 1.243280291557312, "step": 28239 }, { "epoch": 2.226251478123768, "grad_norm": 0.5949178338050842, "learning_rate": 0.0005, "loss": 1.2630360126495361, "step": 28240 }, { "epoch": 2.2263303113914072, "grad_norm": 0.41789060831069946, "learning_rate": 0.0005, "loss": 1.1344091892242432, "step": 28241 }, { "epoch": 2.226409144659046, "grad_norm": 0.45605817437171936, "learning_rate": 0.0005, "loss": 1.4114993810653687, "step": 28242 }, { "epoch": 2.226487977926685, "grad_norm": 0.4177136719226837, "learning_rate": 0.0005, "loss": 1.132951021194458, "step": 28243 }, { "epoch": 2.226566811194324, "grad_norm": 0.3992886245250702, "learning_rate": 0.0005, "loss": 1.1988589763641357, "step": 28244 }, { "epoch": 2.226645644461963, "grad_norm": 0.5009450316429138, "learning_rate": 0.0005, "loss": 1.4758412837982178, "step": 28245 }, { "epoch": 2.226724477729602, "grad_norm": 0.48783695697784424, "learning_rate": 0.0005, "loss": 1.4583213329315186, "step": 28246 }, { "epoch": 2.2268033109972407, "grad_norm": 0.6137518286705017, "learning_rate": 0.0005, "loss": 1.3939704895019531, "step": 28247 }, { "epoch": 2.22688214426488, "grad_norm": 0.7966024875640869, "learning_rate": 0.0005, "loss": 1.396247148513794, "step": 28248 }, { "epoch": 2.2269609775325185, "grad_norm": 0.46422597765922546, "learning_rate": 0.0005, "loss": 1.1822296380996704, "step": 28249 }, { "epoch": 2.2270398108001577, "grad_norm": 0.41611552238464355, "learning_rate": 0.0005, "loss": 1.2669541835784912, "step": 28250 }, { "epoch": 2.227118644067797, "grad_norm": 0.5642392635345459, "learning_rate": 0.0005, "loss": 1.311094880104065, "step": 28251 }, { "epoch": 2.2271974773354355, "grad_norm": 0.4697209298610687, "learning_rate": 0.0005, "loss": 1.362326741218567, "step": 28252 }, { "epoch": 2.2272763106030746, "grad_norm": 0.4317683279514313, "learning_rate": 0.0005, "loss": 1.2768685817718506, "step": 28253 }, { "epoch": 2.2273551438707133, "grad_norm": 0.5391834378242493, "learning_rate": 0.0005, "loss": 1.3777238130569458, "step": 28254 }, { "epoch": 2.2274339771383524, "grad_norm": 0.5049025416374207, "learning_rate": 0.0005, "loss": 1.3126637935638428, "step": 28255 }, { "epoch": 2.227512810405991, "grad_norm": 0.5462983846664429, "learning_rate": 0.0005, "loss": 1.2309212684631348, "step": 28256 }, { "epoch": 2.2275916436736303, "grad_norm": 0.48735880851745605, "learning_rate": 0.0005, "loss": 1.3676725625991821, "step": 28257 }, { "epoch": 2.2276704769412694, "grad_norm": 0.5360366106033325, "learning_rate": 0.0005, "loss": 1.5137593746185303, "step": 28258 }, { "epoch": 2.227749310208908, "grad_norm": 0.49543261528015137, "learning_rate": 0.0005, "loss": 1.4323115348815918, "step": 28259 }, { "epoch": 2.2278281434765472, "grad_norm": 0.5159831047058105, "learning_rate": 0.0005, "loss": 1.3750250339508057, "step": 28260 }, { "epoch": 2.227906976744186, "grad_norm": 0.46255654096603394, "learning_rate": 0.0005, "loss": 1.2775720357894897, "step": 28261 }, { "epoch": 2.227985810011825, "grad_norm": 0.46719393134117126, "learning_rate": 0.0005, "loss": 1.323882818222046, "step": 28262 }, { "epoch": 2.2280646432794637, "grad_norm": 0.4752274453639984, "learning_rate": 0.0005, "loss": 1.296593189239502, "step": 28263 }, { "epoch": 2.228143476547103, "grad_norm": 0.5308703780174255, "learning_rate": 0.0005, "loss": 1.347449779510498, "step": 28264 }, { "epoch": 2.228222309814742, "grad_norm": 0.48951104283332825, "learning_rate": 0.0005, "loss": 1.3687975406646729, "step": 28265 }, { "epoch": 2.2283011430823807, "grad_norm": 0.6376722455024719, "learning_rate": 0.0005, "loss": 1.4428948163986206, "step": 28266 }, { "epoch": 2.22837997635002, "grad_norm": 0.4224732220172882, "learning_rate": 0.0005, "loss": 1.185880184173584, "step": 28267 }, { "epoch": 2.2284588096176585, "grad_norm": 0.5144929885864258, "learning_rate": 0.0005, "loss": 1.4229938983917236, "step": 28268 }, { "epoch": 2.2285376428852977, "grad_norm": 0.42759597301483154, "learning_rate": 0.0005, "loss": 1.1727612018585205, "step": 28269 }, { "epoch": 2.2286164761529363, "grad_norm": 0.43267711997032166, "learning_rate": 0.0005, "loss": 1.2208752632141113, "step": 28270 }, { "epoch": 2.2286953094205755, "grad_norm": 0.5198342800140381, "learning_rate": 0.0005, "loss": 1.304787039756775, "step": 28271 }, { "epoch": 2.2287741426882146, "grad_norm": 0.47126585245132446, "learning_rate": 0.0005, "loss": 1.4309909343719482, "step": 28272 }, { "epoch": 2.2288529759558533, "grad_norm": 0.4335620403289795, "learning_rate": 0.0005, "loss": 1.1008021831512451, "step": 28273 }, { "epoch": 2.2289318092234924, "grad_norm": 0.5264834761619568, "learning_rate": 0.0005, "loss": 1.2263833284378052, "step": 28274 }, { "epoch": 2.229010642491131, "grad_norm": 0.4277198910713196, "learning_rate": 0.0005, "loss": 1.0925233364105225, "step": 28275 }, { "epoch": 2.2290894757587703, "grad_norm": 0.5283845067024231, "learning_rate": 0.0005, "loss": 1.3418478965759277, "step": 28276 }, { "epoch": 2.229168309026409, "grad_norm": 0.4911375641822815, "learning_rate": 0.0005, "loss": 1.316511631011963, "step": 28277 }, { "epoch": 2.229247142294048, "grad_norm": 0.5067079663276672, "learning_rate": 0.0005, "loss": 1.2693312168121338, "step": 28278 }, { "epoch": 2.229325975561687, "grad_norm": 0.542371928691864, "learning_rate": 0.0005, "loss": 1.2722718715667725, "step": 28279 }, { "epoch": 2.229404808829326, "grad_norm": 0.5223414301872253, "learning_rate": 0.0005, "loss": 1.4343128204345703, "step": 28280 }, { "epoch": 2.229483642096965, "grad_norm": 0.4489825367927551, "learning_rate": 0.0005, "loss": 1.220667839050293, "step": 28281 }, { "epoch": 2.2295624753646037, "grad_norm": 0.4571412205696106, "learning_rate": 0.0005, "loss": 1.2984240055084229, "step": 28282 }, { "epoch": 2.229641308632243, "grad_norm": 0.5125874280929565, "learning_rate": 0.0005, "loss": 1.2618333101272583, "step": 28283 }, { "epoch": 2.2297201418998815, "grad_norm": 0.5444291830062866, "learning_rate": 0.0005, "loss": 1.521125078201294, "step": 28284 }, { "epoch": 2.2297989751675207, "grad_norm": 0.4905124604701996, "learning_rate": 0.0005, "loss": 1.501265525817871, "step": 28285 }, { "epoch": 2.22987780843516, "grad_norm": 0.6032671928405762, "learning_rate": 0.0005, "loss": 1.3913402557373047, "step": 28286 }, { "epoch": 2.2299566417027985, "grad_norm": 0.5523597002029419, "learning_rate": 0.0005, "loss": 1.4067420959472656, "step": 28287 }, { "epoch": 2.2300354749704376, "grad_norm": 0.5269906520843506, "learning_rate": 0.0005, "loss": 1.559096097946167, "step": 28288 }, { "epoch": 2.2301143082380763, "grad_norm": 0.519748866558075, "learning_rate": 0.0005, "loss": 1.3509902954101562, "step": 28289 }, { "epoch": 2.2301931415057155, "grad_norm": 0.5563883185386658, "learning_rate": 0.0005, "loss": 1.4940741062164307, "step": 28290 }, { "epoch": 2.230271974773354, "grad_norm": 0.6995545625686646, "learning_rate": 0.0005, "loss": 1.296851396560669, "step": 28291 }, { "epoch": 2.2303508080409933, "grad_norm": 0.5323556065559387, "learning_rate": 0.0005, "loss": 1.217052936553955, "step": 28292 }, { "epoch": 2.2304296413086324, "grad_norm": 0.47651419043540955, "learning_rate": 0.0005, "loss": 1.4323604106903076, "step": 28293 }, { "epoch": 2.230508474576271, "grad_norm": 0.4763250946998596, "learning_rate": 0.0005, "loss": 1.330334186553955, "step": 28294 }, { "epoch": 2.2305873078439102, "grad_norm": 0.6083263158798218, "learning_rate": 0.0005, "loss": 1.5371222496032715, "step": 28295 }, { "epoch": 2.230666141111549, "grad_norm": 0.5299860835075378, "learning_rate": 0.0005, "loss": 1.4024302959442139, "step": 28296 }, { "epoch": 2.230744974379188, "grad_norm": 0.5638259053230286, "learning_rate": 0.0005, "loss": 1.4244225025177002, "step": 28297 }, { "epoch": 2.2308238076468268, "grad_norm": 0.5550103187561035, "learning_rate": 0.0005, "loss": 1.3061132431030273, "step": 28298 }, { "epoch": 2.230902640914466, "grad_norm": 0.5125486254692078, "learning_rate": 0.0005, "loss": 1.43929123878479, "step": 28299 }, { "epoch": 2.230981474182105, "grad_norm": 0.547882080078125, "learning_rate": 0.0005, "loss": 1.30816650390625, "step": 28300 }, { "epoch": 2.2310603074497437, "grad_norm": 0.49638402462005615, "learning_rate": 0.0005, "loss": 1.202868938446045, "step": 28301 }, { "epoch": 2.231139140717383, "grad_norm": 0.5125197768211365, "learning_rate": 0.0005, "loss": 1.3193581104278564, "step": 28302 }, { "epoch": 2.2312179739850215, "grad_norm": 0.5280899405479431, "learning_rate": 0.0005, "loss": 1.2763159275054932, "step": 28303 }, { "epoch": 2.2312968072526607, "grad_norm": 0.4716741442680359, "learning_rate": 0.0005, "loss": 1.2462282180786133, "step": 28304 }, { "epoch": 2.2313756405203, "grad_norm": 0.4699528217315674, "learning_rate": 0.0005, "loss": 1.4089586734771729, "step": 28305 }, { "epoch": 2.2314544737879385, "grad_norm": 0.4811128079891205, "learning_rate": 0.0005, "loss": 1.5121757984161377, "step": 28306 }, { "epoch": 2.2315333070555776, "grad_norm": 0.4150410592556, "learning_rate": 0.0005, "loss": 1.3427718877792358, "step": 28307 }, { "epoch": 2.2316121403232163, "grad_norm": 0.49756062030792236, "learning_rate": 0.0005, "loss": 1.213001012802124, "step": 28308 }, { "epoch": 2.2316909735908554, "grad_norm": 0.5094100832939148, "learning_rate": 0.0005, "loss": 1.4733242988586426, "step": 28309 }, { "epoch": 2.231769806858494, "grad_norm": 0.5623432397842407, "learning_rate": 0.0005, "loss": 1.552504062652588, "step": 28310 }, { "epoch": 2.2318486401261333, "grad_norm": 0.8086846470832825, "learning_rate": 0.0005, "loss": 1.3687739372253418, "step": 28311 }, { "epoch": 2.231927473393772, "grad_norm": 0.5281391143798828, "learning_rate": 0.0005, "loss": 1.1459877490997314, "step": 28312 }, { "epoch": 2.232006306661411, "grad_norm": 0.5876305103302002, "learning_rate": 0.0005, "loss": 1.4160478115081787, "step": 28313 }, { "epoch": 2.23208513992905, "grad_norm": 0.5496618151664734, "learning_rate": 0.0005, "loss": 1.359523057937622, "step": 28314 }, { "epoch": 2.232163973196689, "grad_norm": 0.4438712000846863, "learning_rate": 0.0005, "loss": 1.237743616104126, "step": 28315 }, { "epoch": 2.232242806464328, "grad_norm": 0.5380332469940186, "learning_rate": 0.0005, "loss": 1.3690487146377563, "step": 28316 }, { "epoch": 2.2323216397319667, "grad_norm": 0.5351756811141968, "learning_rate": 0.0005, "loss": 1.3879570960998535, "step": 28317 }, { "epoch": 2.232400472999606, "grad_norm": 0.48867282271385193, "learning_rate": 0.0005, "loss": 1.217395305633545, "step": 28318 }, { "epoch": 2.232479306267245, "grad_norm": 0.43734312057495117, "learning_rate": 0.0005, "loss": 1.2969038486480713, "step": 28319 }, { "epoch": 2.2325581395348837, "grad_norm": 0.5328992009162903, "learning_rate": 0.0005, "loss": 1.2778558731079102, "step": 28320 }, { "epoch": 2.232636972802523, "grad_norm": 0.49548134207725525, "learning_rate": 0.0005, "loss": 1.4627113342285156, "step": 28321 }, { "epoch": 2.2327158060701615, "grad_norm": 0.6662355065345764, "learning_rate": 0.0005, "loss": 1.471863865852356, "step": 28322 }, { "epoch": 2.2327946393378006, "grad_norm": 0.39586570858955383, "learning_rate": 0.0005, "loss": 1.1750822067260742, "step": 28323 }, { "epoch": 2.2328734726054393, "grad_norm": 0.49369359016418457, "learning_rate": 0.0005, "loss": 1.2935587167739868, "step": 28324 }, { "epoch": 2.2329523058730785, "grad_norm": 0.6898428201675415, "learning_rate": 0.0005, "loss": 1.5742745399475098, "step": 28325 }, { "epoch": 2.233031139140717, "grad_norm": 0.59518963098526, "learning_rate": 0.0005, "loss": 1.368821620941162, "step": 28326 }, { "epoch": 2.2331099724083563, "grad_norm": 0.5032368898391724, "learning_rate": 0.0005, "loss": 1.2941372394561768, "step": 28327 }, { "epoch": 2.2331888056759954, "grad_norm": 0.5163507461547852, "learning_rate": 0.0005, "loss": 1.24222731590271, "step": 28328 }, { "epoch": 2.233267638943634, "grad_norm": 0.4987413287162781, "learning_rate": 0.0005, "loss": 1.5107555389404297, "step": 28329 }, { "epoch": 2.2333464722112732, "grad_norm": 0.46460071206092834, "learning_rate": 0.0005, "loss": 1.3387900590896606, "step": 28330 }, { "epoch": 2.233425305478912, "grad_norm": 0.5368793606758118, "learning_rate": 0.0005, "loss": 1.359290599822998, "step": 28331 }, { "epoch": 2.233504138746551, "grad_norm": 0.4977240562438965, "learning_rate": 0.0005, "loss": 1.4373054504394531, "step": 28332 }, { "epoch": 2.23358297201419, "grad_norm": 0.6982146501541138, "learning_rate": 0.0005, "loss": 1.1873652935028076, "step": 28333 }, { "epoch": 2.233661805281829, "grad_norm": 0.4234570264816284, "learning_rate": 0.0005, "loss": 1.2317533493041992, "step": 28334 }, { "epoch": 2.233740638549468, "grad_norm": 0.4409540891647339, "learning_rate": 0.0005, "loss": 1.2804285287857056, "step": 28335 }, { "epoch": 2.2338194718171067, "grad_norm": 0.6481072902679443, "learning_rate": 0.0005, "loss": 1.5065727233886719, "step": 28336 }, { "epoch": 2.233898305084746, "grad_norm": 0.3883019983768463, "learning_rate": 0.0005, "loss": 1.2576574087142944, "step": 28337 }, { "epoch": 2.2339771383523845, "grad_norm": 0.4484882354736328, "learning_rate": 0.0005, "loss": 1.290391445159912, "step": 28338 }, { "epoch": 2.2340559716200237, "grad_norm": 0.4911743104457855, "learning_rate": 0.0005, "loss": 1.2393630743026733, "step": 28339 }, { "epoch": 2.2341348048876624, "grad_norm": 0.4770371913909912, "learning_rate": 0.0005, "loss": 1.5375313758850098, "step": 28340 }, { "epoch": 2.2342136381553015, "grad_norm": 0.44699111580848694, "learning_rate": 0.0005, "loss": 1.3592926263809204, "step": 28341 }, { "epoch": 2.2342924714229406, "grad_norm": 0.5119693279266357, "learning_rate": 0.0005, "loss": 1.3119409084320068, "step": 28342 }, { "epoch": 2.2343713046905793, "grad_norm": 0.5904557108879089, "learning_rate": 0.0005, "loss": 1.5335086584091187, "step": 28343 }, { "epoch": 2.2344501379582185, "grad_norm": 0.5961230397224426, "learning_rate": 0.0005, "loss": 1.4152917861938477, "step": 28344 }, { "epoch": 2.234528971225857, "grad_norm": 0.5381563305854797, "learning_rate": 0.0005, "loss": 1.2500399351119995, "step": 28345 }, { "epoch": 2.2346078044934963, "grad_norm": 0.5139386653900146, "learning_rate": 0.0005, "loss": 1.212216854095459, "step": 28346 }, { "epoch": 2.2346866377611354, "grad_norm": 0.6341782808303833, "learning_rate": 0.0005, "loss": 1.2688472270965576, "step": 28347 }, { "epoch": 2.234765471028774, "grad_norm": 0.4588647484779358, "learning_rate": 0.0005, "loss": 1.1522936820983887, "step": 28348 }, { "epoch": 2.2348443042964132, "grad_norm": 0.47718480229377747, "learning_rate": 0.0005, "loss": 1.226550817489624, "step": 28349 }, { "epoch": 2.234923137564052, "grad_norm": 0.5412312150001526, "learning_rate": 0.0005, "loss": 1.457826852798462, "step": 28350 }, { "epoch": 2.235001970831691, "grad_norm": 0.44143128395080566, "learning_rate": 0.0005, "loss": 1.2216209173202515, "step": 28351 }, { "epoch": 2.2350808040993297, "grad_norm": 0.5144141912460327, "learning_rate": 0.0005, "loss": 1.4149727821350098, "step": 28352 }, { "epoch": 2.235159637366969, "grad_norm": 0.466494083404541, "learning_rate": 0.0005, "loss": 1.3401377201080322, "step": 28353 }, { "epoch": 2.235238470634608, "grad_norm": 0.45979180932044983, "learning_rate": 0.0005, "loss": 1.3563512563705444, "step": 28354 }, { "epoch": 2.2353173039022467, "grad_norm": 0.5786073803901672, "learning_rate": 0.0005, "loss": 1.3689839839935303, "step": 28355 }, { "epoch": 2.235396137169886, "grad_norm": 0.49227893352508545, "learning_rate": 0.0005, "loss": 1.2502741813659668, "step": 28356 }, { "epoch": 2.2354749704375245, "grad_norm": 0.5763033032417297, "learning_rate": 0.0005, "loss": 1.364515781402588, "step": 28357 }, { "epoch": 2.2355538037051637, "grad_norm": 0.5169336199760437, "learning_rate": 0.0005, "loss": 1.3076586723327637, "step": 28358 }, { "epoch": 2.2356326369728023, "grad_norm": 0.5720324516296387, "learning_rate": 0.0005, "loss": 1.3706350326538086, "step": 28359 }, { "epoch": 2.2357114702404415, "grad_norm": 0.4289506673812866, "learning_rate": 0.0005, "loss": 1.2303900718688965, "step": 28360 }, { "epoch": 2.2357903035080806, "grad_norm": 0.5899069905281067, "learning_rate": 0.0005, "loss": 1.3485007286071777, "step": 28361 }, { "epoch": 2.2358691367757193, "grad_norm": 0.4414888620376587, "learning_rate": 0.0005, "loss": 1.327021598815918, "step": 28362 }, { "epoch": 2.2359479700433584, "grad_norm": 0.609804093837738, "learning_rate": 0.0005, "loss": 1.533104658126831, "step": 28363 }, { "epoch": 2.236026803310997, "grad_norm": 0.482669472694397, "learning_rate": 0.0005, "loss": 1.3207401037216187, "step": 28364 }, { "epoch": 2.2361056365786363, "grad_norm": 0.48686784505844116, "learning_rate": 0.0005, "loss": 1.3763009309768677, "step": 28365 }, { "epoch": 2.236184469846275, "grad_norm": 0.6702589988708496, "learning_rate": 0.0005, "loss": 1.2917460203170776, "step": 28366 }, { "epoch": 2.236263303113914, "grad_norm": 0.5389851331710815, "learning_rate": 0.0005, "loss": 1.2365130186080933, "step": 28367 }, { "epoch": 2.236342136381553, "grad_norm": 0.629477322101593, "learning_rate": 0.0005, "loss": 1.4437835216522217, "step": 28368 }, { "epoch": 2.236420969649192, "grad_norm": 0.48788365721702576, "learning_rate": 0.0005, "loss": 1.3035224676132202, "step": 28369 }, { "epoch": 2.236499802916831, "grad_norm": 0.435146689414978, "learning_rate": 0.0005, "loss": 1.184600591659546, "step": 28370 }, { "epoch": 2.2365786361844697, "grad_norm": 0.529528021812439, "learning_rate": 0.0005, "loss": 1.3167977333068848, "step": 28371 }, { "epoch": 2.236657469452109, "grad_norm": 0.639398992061615, "learning_rate": 0.0005, "loss": 1.3970062732696533, "step": 28372 }, { "epoch": 2.2367363027197475, "grad_norm": 0.4538811147212982, "learning_rate": 0.0005, "loss": 1.1913673877716064, "step": 28373 }, { "epoch": 2.2368151359873867, "grad_norm": 0.6050177216529846, "learning_rate": 0.0005, "loss": 1.226893424987793, "step": 28374 }, { "epoch": 2.236893969255026, "grad_norm": 0.5940211415290833, "learning_rate": 0.0005, "loss": 1.3790466785430908, "step": 28375 }, { "epoch": 2.2369728025226645, "grad_norm": 0.47973233461380005, "learning_rate": 0.0005, "loss": 1.3182933330535889, "step": 28376 }, { "epoch": 2.2370516357903036, "grad_norm": 0.41227710247039795, "learning_rate": 0.0005, "loss": 1.282382845878601, "step": 28377 }, { "epoch": 2.2371304690579423, "grad_norm": 0.4592263996601105, "learning_rate": 0.0005, "loss": 1.255638599395752, "step": 28378 }, { "epoch": 2.2372093023255815, "grad_norm": 0.4631735682487488, "learning_rate": 0.0005, "loss": 1.4013277292251587, "step": 28379 }, { "epoch": 2.23728813559322, "grad_norm": 0.730207085609436, "learning_rate": 0.0005, "loss": 1.5423531532287598, "step": 28380 }, { "epoch": 2.2373669688608593, "grad_norm": 0.6225348711013794, "learning_rate": 0.0005, "loss": 1.373375415802002, "step": 28381 }, { "epoch": 2.2374458021284984, "grad_norm": 0.4428099989891052, "learning_rate": 0.0005, "loss": 1.1076823472976685, "step": 28382 }, { "epoch": 2.237524635396137, "grad_norm": 0.5027745962142944, "learning_rate": 0.0005, "loss": 1.205167531967163, "step": 28383 }, { "epoch": 2.2376034686637762, "grad_norm": 0.3962562382221222, "learning_rate": 0.0005, "loss": 1.3036644458770752, "step": 28384 }, { "epoch": 2.237682301931415, "grad_norm": 0.5511302351951599, "learning_rate": 0.0005, "loss": 1.3590672016143799, "step": 28385 }, { "epoch": 2.237761135199054, "grad_norm": 0.6630741953849792, "learning_rate": 0.0005, "loss": 1.5458712577819824, "step": 28386 }, { "epoch": 2.2378399684666928, "grad_norm": 0.42270469665527344, "learning_rate": 0.0005, "loss": 1.346629023551941, "step": 28387 }, { "epoch": 2.237918801734332, "grad_norm": 0.5497311353683472, "learning_rate": 0.0005, "loss": 1.3645567893981934, "step": 28388 }, { "epoch": 2.237997635001971, "grad_norm": 0.9106725454330444, "learning_rate": 0.0005, "loss": 1.3820433616638184, "step": 28389 }, { "epoch": 2.2380764682696097, "grad_norm": 0.3594663441181183, "learning_rate": 0.0005, "loss": 0.9498317837715149, "step": 28390 }, { "epoch": 2.238155301537249, "grad_norm": 0.504538357257843, "learning_rate": 0.0005, "loss": 1.3264365196228027, "step": 28391 }, { "epoch": 2.2382341348048875, "grad_norm": 0.5188793540000916, "learning_rate": 0.0005, "loss": 1.2391198873519897, "step": 28392 }, { "epoch": 2.2383129680725267, "grad_norm": 0.49276435375213623, "learning_rate": 0.0005, "loss": 1.3476094007492065, "step": 28393 }, { "epoch": 2.2383918013401654, "grad_norm": 0.5314370393753052, "learning_rate": 0.0005, "loss": 1.4405262470245361, "step": 28394 }, { "epoch": 2.2384706346078045, "grad_norm": 0.49331963062286377, "learning_rate": 0.0005, "loss": 1.4720408916473389, "step": 28395 }, { "epoch": 2.2385494678754436, "grad_norm": 0.5415603518486023, "learning_rate": 0.0005, "loss": 1.4092936515808105, "step": 28396 }, { "epoch": 2.2386283011430823, "grad_norm": 0.5668002963066101, "learning_rate": 0.0005, "loss": 1.6774792671203613, "step": 28397 }, { "epoch": 2.2387071344107214, "grad_norm": 0.6497027277946472, "learning_rate": 0.0005, "loss": 1.3749496936798096, "step": 28398 }, { "epoch": 2.23878596767836, "grad_norm": 0.7441266179084778, "learning_rate": 0.0005, "loss": 1.6634995937347412, "step": 28399 }, { "epoch": 2.2388648009459993, "grad_norm": 0.502310574054718, "learning_rate": 0.0005, "loss": 1.159023642539978, "step": 28400 }, { "epoch": 2.238943634213638, "grad_norm": 0.46823886036872864, "learning_rate": 0.0005, "loss": 1.3636115789413452, "step": 28401 }, { "epoch": 2.239022467481277, "grad_norm": 0.489016592502594, "learning_rate": 0.0005, "loss": 1.2682477235794067, "step": 28402 }, { "epoch": 2.2391013007489162, "grad_norm": 0.502773106098175, "learning_rate": 0.0005, "loss": 1.3606348037719727, "step": 28403 }, { "epoch": 2.239180134016555, "grad_norm": 0.47774654626846313, "learning_rate": 0.0005, "loss": 1.3269758224487305, "step": 28404 }, { "epoch": 2.239258967284194, "grad_norm": 0.8385021090507507, "learning_rate": 0.0005, "loss": 1.6869032382965088, "step": 28405 }, { "epoch": 2.2393378005518327, "grad_norm": 0.6071922183036804, "learning_rate": 0.0005, "loss": 1.4006364345550537, "step": 28406 }, { "epoch": 2.239416633819472, "grad_norm": 0.6648318767547607, "learning_rate": 0.0005, "loss": 1.3374054431915283, "step": 28407 }, { "epoch": 2.2394954670871106, "grad_norm": 0.48305797576904297, "learning_rate": 0.0005, "loss": 1.3526068925857544, "step": 28408 }, { "epoch": 2.2395743003547497, "grad_norm": 0.6373107433319092, "learning_rate": 0.0005, "loss": 1.5126726627349854, "step": 28409 }, { "epoch": 2.239653133622389, "grad_norm": 0.42272624373435974, "learning_rate": 0.0005, "loss": 1.1389379501342773, "step": 28410 }, { "epoch": 2.2397319668900275, "grad_norm": 0.5023343563079834, "learning_rate": 0.0005, "loss": 1.0965346097946167, "step": 28411 }, { "epoch": 2.2398108001576666, "grad_norm": 0.6451847553253174, "learning_rate": 0.0005, "loss": 1.4927183389663696, "step": 28412 }, { "epoch": 2.2398896334253053, "grad_norm": 0.5319225192070007, "learning_rate": 0.0005, "loss": 1.4618232250213623, "step": 28413 }, { "epoch": 2.2399684666929445, "grad_norm": 0.49161168932914734, "learning_rate": 0.0005, "loss": 1.3221060037612915, "step": 28414 }, { "epoch": 2.240047299960583, "grad_norm": 0.5417340993881226, "learning_rate": 0.0005, "loss": 1.4669475555419922, "step": 28415 }, { "epoch": 2.2401261332282223, "grad_norm": 0.6341031193733215, "learning_rate": 0.0005, "loss": 1.528857946395874, "step": 28416 }, { "epoch": 2.2402049664958614, "grad_norm": 0.5408728122711182, "learning_rate": 0.0005, "loss": 1.4812068939208984, "step": 28417 }, { "epoch": 2.2402837997635, "grad_norm": 0.5388551354408264, "learning_rate": 0.0005, "loss": 1.3786089420318604, "step": 28418 }, { "epoch": 2.2403626330311393, "grad_norm": 0.42148274183273315, "learning_rate": 0.0005, "loss": 1.3684653043746948, "step": 28419 }, { "epoch": 2.240441466298778, "grad_norm": 0.5472095608711243, "learning_rate": 0.0005, "loss": 1.3172168731689453, "step": 28420 }, { "epoch": 2.240520299566417, "grad_norm": 0.44222137331962585, "learning_rate": 0.0005, "loss": 1.4725449085235596, "step": 28421 }, { "epoch": 2.240599132834056, "grad_norm": 0.4642624855041504, "learning_rate": 0.0005, "loss": 1.368072509765625, "step": 28422 }, { "epoch": 2.240677966101695, "grad_norm": 0.4365944266319275, "learning_rate": 0.0005, "loss": 1.4070451259613037, "step": 28423 }, { "epoch": 2.240756799369334, "grad_norm": 0.644487202167511, "learning_rate": 0.0005, "loss": 1.416745901107788, "step": 28424 }, { "epoch": 2.2408356326369727, "grad_norm": 0.44969913363456726, "learning_rate": 0.0005, "loss": 1.2227444648742676, "step": 28425 }, { "epoch": 2.240914465904612, "grad_norm": 0.6050341129302979, "learning_rate": 0.0005, "loss": 1.4559073448181152, "step": 28426 }, { "epoch": 2.2409932991722505, "grad_norm": 0.37655848264694214, "learning_rate": 0.0005, "loss": 1.0511226654052734, "step": 28427 }, { "epoch": 2.2410721324398897, "grad_norm": 0.480129212141037, "learning_rate": 0.0005, "loss": 1.3433384895324707, "step": 28428 }, { "epoch": 2.2411509657075284, "grad_norm": 0.564453125, "learning_rate": 0.0005, "loss": 1.5441453456878662, "step": 28429 }, { "epoch": 2.2412297989751675, "grad_norm": 0.5952548980712891, "learning_rate": 0.0005, "loss": 1.266481876373291, "step": 28430 }, { "epoch": 2.2413086322428066, "grad_norm": 0.4241182506084442, "learning_rate": 0.0005, "loss": 1.2822465896606445, "step": 28431 }, { "epoch": 2.2413874655104453, "grad_norm": 0.4971015155315399, "learning_rate": 0.0005, "loss": 1.280429482460022, "step": 28432 }, { "epoch": 2.2414662987780845, "grad_norm": 0.5348160862922668, "learning_rate": 0.0005, "loss": 1.5373106002807617, "step": 28433 }, { "epoch": 2.241545132045723, "grad_norm": 0.5571949481964111, "learning_rate": 0.0005, "loss": 1.4324359893798828, "step": 28434 }, { "epoch": 2.2416239653133623, "grad_norm": 0.481395423412323, "learning_rate": 0.0005, "loss": 1.36289644241333, "step": 28435 }, { "epoch": 2.2417027985810014, "grad_norm": 0.565912663936615, "learning_rate": 0.0005, "loss": 1.600555419921875, "step": 28436 }, { "epoch": 2.24178163184864, "grad_norm": 0.5143362283706665, "learning_rate": 0.0005, "loss": 1.5878263711929321, "step": 28437 }, { "epoch": 2.2418604651162792, "grad_norm": 0.47159793972969055, "learning_rate": 0.0005, "loss": 1.2649167776107788, "step": 28438 }, { "epoch": 2.241939298383918, "grad_norm": 0.4591664969921112, "learning_rate": 0.0005, "loss": 1.2786098718643188, "step": 28439 }, { "epoch": 2.242018131651557, "grad_norm": 0.7051511406898499, "learning_rate": 0.0005, "loss": 1.3868796825408936, "step": 28440 }, { "epoch": 2.2420969649191957, "grad_norm": 0.5024992823600769, "learning_rate": 0.0005, "loss": 1.3751788139343262, "step": 28441 }, { "epoch": 2.242175798186835, "grad_norm": 0.5198324918746948, "learning_rate": 0.0005, "loss": 1.2971917390823364, "step": 28442 }, { "epoch": 2.2422546314544736, "grad_norm": 0.5418713092803955, "learning_rate": 0.0005, "loss": 1.310049295425415, "step": 28443 }, { "epoch": 2.2423334647221127, "grad_norm": 0.6255244016647339, "learning_rate": 0.0005, "loss": 1.1538772583007812, "step": 28444 }, { "epoch": 2.242412297989752, "grad_norm": 0.39601993560791016, "learning_rate": 0.0005, "loss": 1.1309095621109009, "step": 28445 }, { "epoch": 2.2424911312573905, "grad_norm": 0.41473841667175293, "learning_rate": 0.0005, "loss": 1.3077284097671509, "step": 28446 }, { "epoch": 2.2425699645250297, "grad_norm": 0.460053950548172, "learning_rate": 0.0005, "loss": 1.266249656677246, "step": 28447 }, { "epoch": 2.2426487977926683, "grad_norm": 0.5130355358123779, "learning_rate": 0.0005, "loss": 1.4912464618682861, "step": 28448 }, { "epoch": 2.2427276310603075, "grad_norm": 0.42376890778541565, "learning_rate": 0.0005, "loss": 1.2779161930084229, "step": 28449 }, { "epoch": 2.2428064643279466, "grad_norm": 0.5917222499847412, "learning_rate": 0.0005, "loss": 1.5974746942520142, "step": 28450 }, { "epoch": 2.2428852975955853, "grad_norm": 0.4103774428367615, "learning_rate": 0.0005, "loss": 1.3412809371948242, "step": 28451 }, { "epoch": 2.2429641308632244, "grad_norm": 0.45856013894081116, "learning_rate": 0.0005, "loss": 1.2076914310455322, "step": 28452 }, { "epoch": 2.243042964130863, "grad_norm": 0.45346224308013916, "learning_rate": 0.0005, "loss": 1.3514689207077026, "step": 28453 }, { "epoch": 2.2431217973985023, "grad_norm": 0.48722729086875916, "learning_rate": 0.0005, "loss": 1.3236128091812134, "step": 28454 }, { "epoch": 2.243200630666141, "grad_norm": 0.44927752017974854, "learning_rate": 0.0005, "loss": 1.2867567539215088, "step": 28455 }, { "epoch": 2.24327946393378, "grad_norm": 0.5446193814277649, "learning_rate": 0.0005, "loss": 1.2888693809509277, "step": 28456 }, { "epoch": 2.2433582972014188, "grad_norm": 0.5213134288787842, "learning_rate": 0.0005, "loss": 1.332294225692749, "step": 28457 }, { "epoch": 2.243437130469058, "grad_norm": 0.5426693558692932, "learning_rate": 0.0005, "loss": 1.3760108947753906, "step": 28458 }, { "epoch": 2.243515963736697, "grad_norm": 0.5270718932151794, "learning_rate": 0.0005, "loss": 1.3933823108673096, "step": 28459 }, { "epoch": 2.2435947970043357, "grad_norm": 0.44669392704963684, "learning_rate": 0.0005, "loss": 1.338042140007019, "step": 28460 }, { "epoch": 2.243673630271975, "grad_norm": 0.46809446811676025, "learning_rate": 0.0005, "loss": 1.3430283069610596, "step": 28461 }, { "epoch": 2.2437524635396136, "grad_norm": 0.7341825366020203, "learning_rate": 0.0005, "loss": 1.2614343166351318, "step": 28462 }, { "epoch": 2.2438312968072527, "grad_norm": 0.4491893947124481, "learning_rate": 0.0005, "loss": 1.4280903339385986, "step": 28463 }, { "epoch": 2.243910130074892, "grad_norm": 0.6250624656677246, "learning_rate": 0.0005, "loss": 1.5002474784851074, "step": 28464 }, { "epoch": 2.2439889633425305, "grad_norm": 0.5783436894416809, "learning_rate": 0.0005, "loss": 1.4098632335662842, "step": 28465 }, { "epoch": 2.2440677966101696, "grad_norm": 0.5574479103088379, "learning_rate": 0.0005, "loss": 1.2771928310394287, "step": 28466 }, { "epoch": 2.2441466298778083, "grad_norm": 0.5234494209289551, "learning_rate": 0.0005, "loss": 1.4338059425354004, "step": 28467 }, { "epoch": 2.2442254631454475, "grad_norm": 0.595983624458313, "learning_rate": 0.0005, "loss": 1.3838156461715698, "step": 28468 }, { "epoch": 2.244304296413086, "grad_norm": 0.5434097051620483, "learning_rate": 0.0005, "loss": 1.500624656677246, "step": 28469 }, { "epoch": 2.2443831296807253, "grad_norm": 0.5444976687431335, "learning_rate": 0.0005, "loss": 1.292299747467041, "step": 28470 }, { "epoch": 2.244461962948364, "grad_norm": 0.5186905264854431, "learning_rate": 0.0005, "loss": 1.260101318359375, "step": 28471 }, { "epoch": 2.244540796216003, "grad_norm": 0.44244152307510376, "learning_rate": 0.0005, "loss": 1.447317123413086, "step": 28472 }, { "epoch": 2.2446196294836422, "grad_norm": 0.48735079169273376, "learning_rate": 0.0005, "loss": 1.3508121967315674, "step": 28473 }, { "epoch": 2.244698462751281, "grad_norm": 0.5281165838241577, "learning_rate": 0.0005, "loss": 1.4678990840911865, "step": 28474 }, { "epoch": 2.24477729601892, "grad_norm": 0.5409751534461975, "learning_rate": 0.0005, "loss": 1.3473832607269287, "step": 28475 }, { "epoch": 2.2448561292865588, "grad_norm": 0.44460928440093994, "learning_rate": 0.0005, "loss": 1.2363938093185425, "step": 28476 }, { "epoch": 2.244934962554198, "grad_norm": 0.5509315729141235, "learning_rate": 0.0005, "loss": 1.5542523860931396, "step": 28477 }, { "epoch": 2.245013795821837, "grad_norm": 0.5221925973892212, "learning_rate": 0.0005, "loss": 1.5483078956604004, "step": 28478 }, { "epoch": 2.2450926290894757, "grad_norm": 0.508836030960083, "learning_rate": 0.0005, "loss": 1.3390183448791504, "step": 28479 }, { "epoch": 2.245171462357115, "grad_norm": 0.501251757144928, "learning_rate": 0.0005, "loss": 1.2703335285186768, "step": 28480 }, { "epoch": 2.2452502956247535, "grad_norm": 0.5041272640228271, "learning_rate": 0.0005, "loss": 1.4196299314498901, "step": 28481 }, { "epoch": 2.2453291288923927, "grad_norm": 0.46308428049087524, "learning_rate": 0.0005, "loss": 1.1585841178894043, "step": 28482 }, { "epoch": 2.2454079621600314, "grad_norm": 0.4393145740032196, "learning_rate": 0.0005, "loss": 1.1454393863677979, "step": 28483 }, { "epoch": 2.2454867954276705, "grad_norm": 0.45329341292381287, "learning_rate": 0.0005, "loss": 1.1907448768615723, "step": 28484 }, { "epoch": 2.2455656286953096, "grad_norm": 0.4363984167575836, "learning_rate": 0.0005, "loss": 1.2412889003753662, "step": 28485 }, { "epoch": 2.2456444619629483, "grad_norm": 0.5642125606536865, "learning_rate": 0.0005, "loss": 1.3990893363952637, "step": 28486 }, { "epoch": 2.2457232952305874, "grad_norm": 0.4326314926147461, "learning_rate": 0.0005, "loss": 1.1084333658218384, "step": 28487 }, { "epoch": 2.245802128498226, "grad_norm": 0.41715025901794434, "learning_rate": 0.0005, "loss": 1.2793376445770264, "step": 28488 }, { "epoch": 2.2458809617658653, "grad_norm": 0.4445454776287079, "learning_rate": 0.0005, "loss": 1.2448219060897827, "step": 28489 }, { "epoch": 2.245959795033504, "grad_norm": 0.49211937189102173, "learning_rate": 0.0005, "loss": 1.14261794090271, "step": 28490 }, { "epoch": 2.246038628301143, "grad_norm": 0.5074578523635864, "learning_rate": 0.0005, "loss": 1.4261950254440308, "step": 28491 }, { "epoch": 2.2461174615687822, "grad_norm": 0.5091302990913391, "learning_rate": 0.0005, "loss": 1.4444701671600342, "step": 28492 }, { "epoch": 2.246196294836421, "grad_norm": 0.5148383378982544, "learning_rate": 0.0005, "loss": 1.6014552116394043, "step": 28493 }, { "epoch": 2.24627512810406, "grad_norm": 0.513265073299408, "learning_rate": 0.0005, "loss": 1.3750839233398438, "step": 28494 }, { "epoch": 2.2463539613716987, "grad_norm": 0.5174127221107483, "learning_rate": 0.0005, "loss": 1.4097282886505127, "step": 28495 }, { "epoch": 2.246432794639338, "grad_norm": 0.4637565016746521, "learning_rate": 0.0005, "loss": 1.376439094543457, "step": 28496 }, { "epoch": 2.2465116279069766, "grad_norm": 0.6196467280387878, "learning_rate": 0.0005, "loss": 1.3751027584075928, "step": 28497 }, { "epoch": 2.2465904611746157, "grad_norm": 0.6468899846076965, "learning_rate": 0.0005, "loss": 1.4814658164978027, "step": 28498 }, { "epoch": 2.246669294442255, "grad_norm": 0.45948582887649536, "learning_rate": 0.0005, "loss": 1.5103914737701416, "step": 28499 }, { "epoch": 2.2467481277098935, "grad_norm": 0.4647112786769867, "learning_rate": 0.0005, "loss": 1.1599888801574707, "step": 28500 }, { "epoch": 2.2468269609775327, "grad_norm": 0.5645521879196167, "learning_rate": 0.0005, "loss": 1.28670334815979, "step": 28501 }, { "epoch": 2.2469057942451713, "grad_norm": 0.5346479415893555, "learning_rate": 0.0005, "loss": 1.3312318325042725, "step": 28502 }, { "epoch": 2.2469846275128105, "grad_norm": 0.41177114844322205, "learning_rate": 0.0005, "loss": 1.231684923171997, "step": 28503 }, { "epoch": 2.247063460780449, "grad_norm": 0.5979390740394592, "learning_rate": 0.0005, "loss": 1.2990278005599976, "step": 28504 }, { "epoch": 2.2471422940480883, "grad_norm": 0.4657324552536011, "learning_rate": 0.0005, "loss": 1.1692402362823486, "step": 28505 }, { "epoch": 2.2472211273157274, "grad_norm": 0.4415534436702728, "learning_rate": 0.0005, "loss": 1.2619636058807373, "step": 28506 }, { "epoch": 2.247299960583366, "grad_norm": 0.46042734384536743, "learning_rate": 0.0005, "loss": 1.2452425956726074, "step": 28507 }, { "epoch": 2.2473787938510053, "grad_norm": 0.6193164587020874, "learning_rate": 0.0005, "loss": 1.2454779148101807, "step": 28508 }, { "epoch": 2.247457627118644, "grad_norm": 0.5388053059577942, "learning_rate": 0.0005, "loss": 1.1258587837219238, "step": 28509 }, { "epoch": 2.247536460386283, "grad_norm": 0.639972448348999, "learning_rate": 0.0005, "loss": 1.3947594165802002, "step": 28510 }, { "epoch": 2.2476152936539218, "grad_norm": 0.5073971152305603, "learning_rate": 0.0005, "loss": 1.3274192810058594, "step": 28511 }, { "epoch": 2.247694126921561, "grad_norm": 0.449145644903183, "learning_rate": 0.0005, "loss": 1.2793238162994385, "step": 28512 }, { "epoch": 2.2477729601892, "grad_norm": 0.5904021263122559, "learning_rate": 0.0005, "loss": 1.5587012767791748, "step": 28513 }, { "epoch": 2.2478517934568387, "grad_norm": 0.6251996159553528, "learning_rate": 0.0005, "loss": 1.3785815238952637, "step": 28514 }, { "epoch": 2.247930626724478, "grad_norm": 0.5100457668304443, "learning_rate": 0.0005, "loss": 1.4757981300354004, "step": 28515 }, { "epoch": 2.2480094599921165, "grad_norm": 0.5402491688728333, "learning_rate": 0.0005, "loss": 1.368995189666748, "step": 28516 }, { "epoch": 2.2480882932597557, "grad_norm": 0.5219271779060364, "learning_rate": 0.0005, "loss": 1.3415186405181885, "step": 28517 }, { "epoch": 2.2481671265273944, "grad_norm": 0.49212855100631714, "learning_rate": 0.0005, "loss": 1.2614893913269043, "step": 28518 }, { "epoch": 2.2482459597950335, "grad_norm": 0.5302221775054932, "learning_rate": 0.0005, "loss": 1.3059687614440918, "step": 28519 }, { "epoch": 2.2483247930626726, "grad_norm": 0.4731104373931885, "learning_rate": 0.0005, "loss": 1.3829377889633179, "step": 28520 }, { "epoch": 2.2484036263303113, "grad_norm": 0.6247063875198364, "learning_rate": 0.0005, "loss": 1.4338791370391846, "step": 28521 }, { "epoch": 2.2484824595979505, "grad_norm": 0.49258577823638916, "learning_rate": 0.0005, "loss": 1.1415668725967407, "step": 28522 }, { "epoch": 2.248561292865589, "grad_norm": 0.4221245050430298, "learning_rate": 0.0005, "loss": 1.2892072200775146, "step": 28523 }, { "epoch": 2.2486401261332283, "grad_norm": 0.46489134430885315, "learning_rate": 0.0005, "loss": 1.2483187913894653, "step": 28524 }, { "epoch": 2.248718959400867, "grad_norm": 0.5142554640769958, "learning_rate": 0.0005, "loss": 1.2788453102111816, "step": 28525 }, { "epoch": 2.248797792668506, "grad_norm": 0.46835359930992126, "learning_rate": 0.0005, "loss": 1.1347079277038574, "step": 28526 }, { "epoch": 2.2488766259361452, "grad_norm": 0.562149167060852, "learning_rate": 0.0005, "loss": 1.3549432754516602, "step": 28527 }, { "epoch": 2.248955459203784, "grad_norm": 0.42771515250205994, "learning_rate": 0.0005, "loss": 1.2001264095306396, "step": 28528 }, { "epoch": 2.249034292471423, "grad_norm": 0.48193129897117615, "learning_rate": 0.0005, "loss": 1.448634147644043, "step": 28529 }, { "epoch": 2.2491131257390617, "grad_norm": 0.5477007031440735, "learning_rate": 0.0005, "loss": 1.3992607593536377, "step": 28530 }, { "epoch": 2.249191959006701, "grad_norm": 0.5430341958999634, "learning_rate": 0.0005, "loss": 1.5126348733901978, "step": 28531 }, { "epoch": 2.2492707922743396, "grad_norm": 0.7403337359428406, "learning_rate": 0.0005, "loss": 1.3677715063095093, "step": 28532 }, { "epoch": 2.2493496255419787, "grad_norm": 0.5472297668457031, "learning_rate": 0.0005, "loss": 1.3093788623809814, "step": 28533 }, { "epoch": 2.249428458809618, "grad_norm": 0.5213742852210999, "learning_rate": 0.0005, "loss": 1.3612136840820312, "step": 28534 }, { "epoch": 2.2495072920772565, "grad_norm": 0.44936487078666687, "learning_rate": 0.0005, "loss": 1.3696274757385254, "step": 28535 }, { "epoch": 2.2495861253448957, "grad_norm": 0.477001428604126, "learning_rate": 0.0005, "loss": 1.3826216459274292, "step": 28536 }, { "epoch": 2.2496649586125343, "grad_norm": 0.6418971419334412, "learning_rate": 0.0005, "loss": 1.4994392395019531, "step": 28537 }, { "epoch": 2.2497437918801735, "grad_norm": 0.48980578780174255, "learning_rate": 0.0005, "loss": 1.380096435546875, "step": 28538 }, { "epoch": 2.249822625147812, "grad_norm": 0.4826269745826721, "learning_rate": 0.0005, "loss": 1.3315351009368896, "step": 28539 }, { "epoch": 2.2499014584154513, "grad_norm": 0.4837029278278351, "learning_rate": 0.0005, "loss": 1.220395565032959, "step": 28540 }, { "epoch": 2.2499802916830904, "grad_norm": 2.2203996181488037, "learning_rate": 0.0005, "loss": 1.246446132659912, "step": 28541 }, { "epoch": 2.250059124950729, "grad_norm": 0.42804622650146484, "learning_rate": 0.0005, "loss": 1.4036507606506348, "step": 28542 }, { "epoch": 2.2501379582183683, "grad_norm": 0.46552494168281555, "learning_rate": 0.0005, "loss": 1.4070907831192017, "step": 28543 }, { "epoch": 2.250216791486007, "grad_norm": 0.6436789631843567, "learning_rate": 0.0005, "loss": 1.3044230937957764, "step": 28544 }, { "epoch": 2.250295624753646, "grad_norm": 1.5567327737808228, "learning_rate": 0.0005, "loss": 1.3173401355743408, "step": 28545 }, { "epoch": 2.2503744580212848, "grad_norm": 0.589477002620697, "learning_rate": 0.0005, "loss": 1.584924578666687, "step": 28546 }, { "epoch": 2.250453291288924, "grad_norm": 0.4893999695777893, "learning_rate": 0.0005, "loss": 1.3199288845062256, "step": 28547 }, { "epoch": 2.250532124556563, "grad_norm": 0.6080403327941895, "learning_rate": 0.0005, "loss": 1.4420108795166016, "step": 28548 }, { "epoch": 2.2506109578242017, "grad_norm": 0.5898521542549133, "learning_rate": 0.0005, "loss": 1.2023669481277466, "step": 28549 }, { "epoch": 2.250689791091841, "grad_norm": 0.44773828983306885, "learning_rate": 0.0005, "loss": 1.386101245880127, "step": 28550 }, { "epoch": 2.2507686243594796, "grad_norm": 0.4740775525569916, "learning_rate": 0.0005, "loss": 1.1893072128295898, "step": 28551 }, { "epoch": 2.2508474576271187, "grad_norm": 0.49847543239593506, "learning_rate": 0.0005, "loss": 1.133293867111206, "step": 28552 }, { "epoch": 2.250926290894758, "grad_norm": 0.550651490688324, "learning_rate": 0.0005, "loss": 1.2976913452148438, "step": 28553 }, { "epoch": 2.2510051241623965, "grad_norm": 0.37264662981033325, "learning_rate": 0.0005, "loss": 1.0733095407485962, "step": 28554 }, { "epoch": 2.2510839574300356, "grad_norm": 0.431944876909256, "learning_rate": 0.0005, "loss": 1.441061019897461, "step": 28555 }, { "epoch": 2.2511627906976743, "grad_norm": 0.5484790802001953, "learning_rate": 0.0005, "loss": 1.472949504852295, "step": 28556 }, { "epoch": 2.2512416239653135, "grad_norm": 0.3929477632045746, "learning_rate": 0.0005, "loss": 1.1998164653778076, "step": 28557 }, { "epoch": 2.251320457232952, "grad_norm": 0.773348331451416, "learning_rate": 0.0005, "loss": 1.5239591598510742, "step": 28558 }, { "epoch": 2.2513992905005913, "grad_norm": 0.5065878033638, "learning_rate": 0.0005, "loss": 1.2662631273269653, "step": 28559 }, { "epoch": 2.25147812376823, "grad_norm": 0.4745733439922333, "learning_rate": 0.0005, "loss": 1.4212356805801392, "step": 28560 }, { "epoch": 2.251556957035869, "grad_norm": 0.4608389437198639, "learning_rate": 0.0005, "loss": 1.2765185832977295, "step": 28561 }, { "epoch": 2.2516357903035082, "grad_norm": 0.5438791513442993, "learning_rate": 0.0005, "loss": 1.4696707725524902, "step": 28562 }, { "epoch": 2.251714623571147, "grad_norm": 0.5213825106620789, "learning_rate": 0.0005, "loss": 1.2005329132080078, "step": 28563 }, { "epoch": 2.251793456838786, "grad_norm": 0.4452472925186157, "learning_rate": 0.0005, "loss": 1.233370065689087, "step": 28564 }, { "epoch": 2.2518722901064248, "grad_norm": 0.5574288368225098, "learning_rate": 0.0005, "loss": 1.3186616897583008, "step": 28565 }, { "epoch": 2.251951123374064, "grad_norm": 0.5407736897468567, "learning_rate": 0.0005, "loss": 1.4604239463806152, "step": 28566 }, { "epoch": 2.252029956641703, "grad_norm": 0.580973744392395, "learning_rate": 0.0005, "loss": 1.3630931377410889, "step": 28567 }, { "epoch": 2.2521087899093417, "grad_norm": 0.5597187876701355, "learning_rate": 0.0005, "loss": 1.2393534183502197, "step": 28568 }, { "epoch": 2.252187623176981, "grad_norm": 0.4546279013156891, "learning_rate": 0.0005, "loss": 1.182790994644165, "step": 28569 }, { "epoch": 2.2522664564446195, "grad_norm": 0.5066560506820679, "learning_rate": 0.0005, "loss": 1.3189946413040161, "step": 28570 }, { "epoch": 2.2523452897122587, "grad_norm": 0.4829961657524109, "learning_rate": 0.0005, "loss": 1.2864106893539429, "step": 28571 }, { "epoch": 2.2524241229798974, "grad_norm": 0.5804902911186218, "learning_rate": 0.0005, "loss": 1.2772471904754639, "step": 28572 }, { "epoch": 2.2525029562475365, "grad_norm": 0.48657551407814026, "learning_rate": 0.0005, "loss": 1.5341826677322388, "step": 28573 }, { "epoch": 2.252581789515175, "grad_norm": 0.44515734910964966, "learning_rate": 0.0005, "loss": 1.2546215057373047, "step": 28574 }, { "epoch": 2.2526606227828143, "grad_norm": 0.5103757381439209, "learning_rate": 0.0005, "loss": 1.2355735301971436, "step": 28575 }, { "epoch": 2.2527394560504534, "grad_norm": 0.5077601075172424, "learning_rate": 0.0005, "loss": 1.2301993370056152, "step": 28576 }, { "epoch": 2.252818289318092, "grad_norm": 0.4849399924278259, "learning_rate": 0.0005, "loss": 1.3130205869674683, "step": 28577 }, { "epoch": 2.2528971225857313, "grad_norm": 0.49401652812957764, "learning_rate": 0.0005, "loss": 1.3243567943572998, "step": 28578 }, { "epoch": 2.25297595585337, "grad_norm": 0.4828014671802521, "learning_rate": 0.0005, "loss": 1.1881383657455444, "step": 28579 }, { "epoch": 2.253054789121009, "grad_norm": 0.5219692587852478, "learning_rate": 0.0005, "loss": 1.5980855226516724, "step": 28580 }, { "epoch": 2.2531336223886482, "grad_norm": 0.5123537182807922, "learning_rate": 0.0005, "loss": 1.471876621246338, "step": 28581 }, { "epoch": 2.253212455656287, "grad_norm": 0.5775736570358276, "learning_rate": 0.0005, "loss": 1.3849740028381348, "step": 28582 }, { "epoch": 2.253291288923926, "grad_norm": 0.51310795545578, "learning_rate": 0.0005, "loss": 1.3731698989868164, "step": 28583 }, { "epoch": 2.2533701221915647, "grad_norm": 0.5545991063117981, "learning_rate": 0.0005, "loss": 1.2720506191253662, "step": 28584 }, { "epoch": 2.253448955459204, "grad_norm": 0.5689169764518738, "learning_rate": 0.0005, "loss": 1.2338833808898926, "step": 28585 }, { "epoch": 2.2535277887268426, "grad_norm": 0.6531279683113098, "learning_rate": 0.0005, "loss": 1.4121909141540527, "step": 28586 }, { "epoch": 2.2536066219944817, "grad_norm": 0.532010555267334, "learning_rate": 0.0005, "loss": 1.4485687017440796, "step": 28587 }, { "epoch": 2.2536854552621204, "grad_norm": 0.4901706576347351, "learning_rate": 0.0005, "loss": 1.219452142715454, "step": 28588 }, { "epoch": 2.2537642885297595, "grad_norm": 0.6466668844223022, "learning_rate": 0.0005, "loss": 1.6433556079864502, "step": 28589 }, { "epoch": 2.2538431217973987, "grad_norm": 0.4590657353401184, "learning_rate": 0.0005, "loss": 1.1750582456588745, "step": 28590 }, { "epoch": 2.2539219550650373, "grad_norm": 0.5106403231620789, "learning_rate": 0.0005, "loss": 1.3996511697769165, "step": 28591 }, { "epoch": 2.2540007883326765, "grad_norm": 0.7709904313087463, "learning_rate": 0.0005, "loss": 1.5000817775726318, "step": 28592 }, { "epoch": 2.254079621600315, "grad_norm": 0.5366540551185608, "learning_rate": 0.0005, "loss": 1.2439734935760498, "step": 28593 }, { "epoch": 2.2541584548679543, "grad_norm": 0.4900868535041809, "learning_rate": 0.0005, "loss": 1.273329257965088, "step": 28594 }, { "epoch": 2.2542372881355934, "grad_norm": 0.6384656429290771, "learning_rate": 0.0005, "loss": 1.4981873035430908, "step": 28595 }, { "epoch": 2.254316121403232, "grad_norm": 0.47889310121536255, "learning_rate": 0.0005, "loss": 1.4753401279449463, "step": 28596 }, { "epoch": 2.2543949546708713, "grad_norm": 0.457071989774704, "learning_rate": 0.0005, "loss": 1.199586272239685, "step": 28597 }, { "epoch": 2.25447378793851, "grad_norm": 0.5733769536018372, "learning_rate": 0.0005, "loss": 1.2979305982589722, "step": 28598 }, { "epoch": 2.254552621206149, "grad_norm": 0.5362685918807983, "learning_rate": 0.0005, "loss": 1.4369754791259766, "step": 28599 }, { "epoch": 2.2546314544737878, "grad_norm": 0.6232174634933472, "learning_rate": 0.0005, "loss": 1.575753092765808, "step": 28600 }, { "epoch": 2.254710287741427, "grad_norm": 0.9332584142684937, "learning_rate": 0.0005, "loss": 1.3541710376739502, "step": 28601 }, { "epoch": 2.2547891210090656, "grad_norm": 0.4624936282634735, "learning_rate": 0.0005, "loss": 1.5001301765441895, "step": 28602 }, { "epoch": 2.2548679542767047, "grad_norm": 0.5089897513389587, "learning_rate": 0.0005, "loss": 1.4447216987609863, "step": 28603 }, { "epoch": 2.254946787544344, "grad_norm": 0.5854209065437317, "learning_rate": 0.0005, "loss": 1.3956248760223389, "step": 28604 }, { "epoch": 2.2550256208119825, "grad_norm": 0.7249283194541931, "learning_rate": 0.0005, "loss": 1.351374864578247, "step": 28605 }, { "epoch": 2.2551044540796217, "grad_norm": 0.5070708394050598, "learning_rate": 0.0005, "loss": 1.3951300382614136, "step": 28606 }, { "epoch": 2.2551832873472604, "grad_norm": 0.8906657695770264, "learning_rate": 0.0005, "loss": 1.3196685314178467, "step": 28607 }, { "epoch": 2.2552621206148995, "grad_norm": 0.6193034052848816, "learning_rate": 0.0005, "loss": 1.4614591598510742, "step": 28608 }, { "epoch": 2.2553409538825386, "grad_norm": 0.48871245980262756, "learning_rate": 0.0005, "loss": 1.2879347801208496, "step": 28609 }, { "epoch": 2.2554197871501773, "grad_norm": 0.4336366653442383, "learning_rate": 0.0005, "loss": 1.2201141119003296, "step": 28610 }, { "epoch": 2.2554986204178165, "grad_norm": 0.6017689108848572, "learning_rate": 0.0005, "loss": 1.4485149383544922, "step": 28611 }, { "epoch": 2.255577453685455, "grad_norm": 0.45139312744140625, "learning_rate": 0.0005, "loss": 1.2018202543258667, "step": 28612 }, { "epoch": 2.2556562869530943, "grad_norm": 0.46974891424179077, "learning_rate": 0.0005, "loss": 1.1354421377182007, "step": 28613 }, { "epoch": 2.255735120220733, "grad_norm": 0.6507522463798523, "learning_rate": 0.0005, "loss": 1.4843612909317017, "step": 28614 }, { "epoch": 2.255813953488372, "grad_norm": 0.44764629006385803, "learning_rate": 0.0005, "loss": 1.5093458890914917, "step": 28615 }, { "epoch": 2.255892786756011, "grad_norm": 0.460978627204895, "learning_rate": 0.0005, "loss": 1.0298047065734863, "step": 28616 }, { "epoch": 2.25597162002365, "grad_norm": 0.39590710401535034, "learning_rate": 0.0005, "loss": 1.2342960834503174, "step": 28617 }, { "epoch": 2.256050453291289, "grad_norm": 0.52634596824646, "learning_rate": 0.0005, "loss": 1.4356528520584106, "step": 28618 }, { "epoch": 2.2561292865589277, "grad_norm": 0.5258265137672424, "learning_rate": 0.0005, "loss": 1.3077069520950317, "step": 28619 }, { "epoch": 2.256208119826567, "grad_norm": 0.5743854641914368, "learning_rate": 0.0005, "loss": 1.3350118398666382, "step": 28620 }, { "epoch": 2.2562869530942056, "grad_norm": 0.5227643847465515, "learning_rate": 0.0005, "loss": 1.5313820838928223, "step": 28621 }, { "epoch": 2.2563657863618447, "grad_norm": 0.5758839249610901, "learning_rate": 0.0005, "loss": 1.5026087760925293, "step": 28622 }, { "epoch": 2.256444619629484, "grad_norm": 0.5030068755149841, "learning_rate": 0.0005, "loss": 1.4569538831710815, "step": 28623 }, { "epoch": 2.2565234528971225, "grad_norm": 0.5914273262023926, "learning_rate": 0.0005, "loss": 1.2916624546051025, "step": 28624 }, { "epoch": 2.2566022861647617, "grad_norm": 0.6758742928504944, "learning_rate": 0.0005, "loss": 1.4406706094741821, "step": 28625 }, { "epoch": 2.2566811194324004, "grad_norm": 0.5288490056991577, "learning_rate": 0.0005, "loss": 1.4847979545593262, "step": 28626 }, { "epoch": 2.2567599527000395, "grad_norm": 0.560650110244751, "learning_rate": 0.0005, "loss": 1.3546395301818848, "step": 28627 }, { "epoch": 2.256838785967678, "grad_norm": 0.6209045648574829, "learning_rate": 0.0005, "loss": 1.4416831731796265, "step": 28628 }, { "epoch": 2.2569176192353173, "grad_norm": 0.44749605655670166, "learning_rate": 0.0005, "loss": 1.1760269403457642, "step": 28629 }, { "epoch": 2.2569964525029564, "grad_norm": 0.5836713314056396, "learning_rate": 0.0005, "loss": 1.3187593221664429, "step": 28630 }, { "epoch": 2.257075285770595, "grad_norm": 0.5433988571166992, "learning_rate": 0.0005, "loss": 1.479934811592102, "step": 28631 }, { "epoch": 2.2571541190382343, "grad_norm": 0.5791031718254089, "learning_rate": 0.0005, "loss": 1.252198338508606, "step": 28632 }, { "epoch": 2.257232952305873, "grad_norm": 0.4629761278629303, "learning_rate": 0.0005, "loss": 1.2765321731567383, "step": 28633 }, { "epoch": 2.257311785573512, "grad_norm": 0.4307008683681488, "learning_rate": 0.0005, "loss": 1.326423168182373, "step": 28634 }, { "epoch": 2.2573906188411508, "grad_norm": 0.4871238172054291, "learning_rate": 0.0005, "loss": 1.417229413986206, "step": 28635 }, { "epoch": 2.25746945210879, "grad_norm": 0.48069679737091064, "learning_rate": 0.0005, "loss": 1.4327491521835327, "step": 28636 }, { "epoch": 2.257548285376429, "grad_norm": 0.5574925541877747, "learning_rate": 0.0005, "loss": 1.64048433303833, "step": 28637 }, { "epoch": 2.2576271186440677, "grad_norm": 0.6192947626113892, "learning_rate": 0.0005, "loss": 1.4204246997833252, "step": 28638 }, { "epoch": 2.257705951911707, "grad_norm": 0.6204711198806763, "learning_rate": 0.0005, "loss": 1.5206825733184814, "step": 28639 }, { "epoch": 2.2577847851793456, "grad_norm": 0.4693007469177246, "learning_rate": 0.0005, "loss": 1.3237000703811646, "step": 28640 }, { "epoch": 2.2578636184469847, "grad_norm": 0.5367225408554077, "learning_rate": 0.0005, "loss": 1.2751590013504028, "step": 28641 }, { "epoch": 2.2579424517146234, "grad_norm": 0.5450319051742554, "learning_rate": 0.0005, "loss": 1.3826912641525269, "step": 28642 }, { "epoch": 2.2580212849822625, "grad_norm": 0.5969168543815613, "learning_rate": 0.0005, "loss": 1.3359088897705078, "step": 28643 }, { "epoch": 2.2581001182499016, "grad_norm": 0.49261462688446045, "learning_rate": 0.0005, "loss": 1.323568344116211, "step": 28644 }, { "epoch": 2.2581789515175403, "grad_norm": 0.5540012717247009, "learning_rate": 0.0005, "loss": 1.5231084823608398, "step": 28645 }, { "epoch": 2.2582577847851795, "grad_norm": 0.5283226370811462, "learning_rate": 0.0005, "loss": 1.3859877586364746, "step": 28646 }, { "epoch": 2.258336618052818, "grad_norm": 0.5053200721740723, "learning_rate": 0.0005, "loss": 1.2701942920684814, "step": 28647 }, { "epoch": 2.2584154513204573, "grad_norm": 0.5083266496658325, "learning_rate": 0.0005, "loss": 1.1656941175460815, "step": 28648 }, { "epoch": 2.258494284588096, "grad_norm": 0.5301598310470581, "learning_rate": 0.0005, "loss": 1.3107655048370361, "step": 28649 }, { "epoch": 2.258573117855735, "grad_norm": 0.49310600757598877, "learning_rate": 0.0005, "loss": 1.3944332599639893, "step": 28650 }, { "epoch": 2.2586519511233742, "grad_norm": 0.4497842490673065, "learning_rate": 0.0005, "loss": 1.2125728130340576, "step": 28651 }, { "epoch": 2.258730784391013, "grad_norm": 0.7317894697189331, "learning_rate": 0.0005, "loss": 1.5746291875839233, "step": 28652 }, { "epoch": 2.258809617658652, "grad_norm": 0.6368351578712463, "learning_rate": 0.0005, "loss": 1.3735394477844238, "step": 28653 }, { "epoch": 2.2588884509262908, "grad_norm": 3.2226603031158447, "learning_rate": 0.0005, "loss": 1.6255207061767578, "step": 28654 }, { "epoch": 2.25896728419393, "grad_norm": 0.6410723924636841, "learning_rate": 0.0005, "loss": 1.4340105056762695, "step": 28655 }, { "epoch": 2.259046117461569, "grad_norm": 0.5753961801528931, "learning_rate": 0.0005, "loss": 1.548767328262329, "step": 28656 }, { "epoch": 2.2591249507292077, "grad_norm": 0.5169352293014526, "learning_rate": 0.0005, "loss": 1.264958143234253, "step": 28657 }, { "epoch": 2.259203783996847, "grad_norm": 0.489719420671463, "learning_rate": 0.0005, "loss": 1.176412582397461, "step": 28658 }, { "epoch": 2.2592826172644855, "grad_norm": 0.5451294779777527, "learning_rate": 0.0005, "loss": 1.1531411409378052, "step": 28659 }, { "epoch": 2.2593614505321247, "grad_norm": 0.38071098923683167, "learning_rate": 0.0005, "loss": 1.269216775894165, "step": 28660 }, { "epoch": 2.2594402837997634, "grad_norm": 0.48220711946487427, "learning_rate": 0.0005, "loss": 1.22428560256958, "step": 28661 }, { "epoch": 2.2595191170674025, "grad_norm": 0.669424295425415, "learning_rate": 0.0005, "loss": 1.4851053953170776, "step": 28662 }, { "epoch": 2.259597950335041, "grad_norm": 0.41062629222869873, "learning_rate": 0.0005, "loss": 1.1754837036132812, "step": 28663 }, { "epoch": 2.2596767836026803, "grad_norm": 0.4608090817928314, "learning_rate": 0.0005, "loss": 1.1799695491790771, "step": 28664 }, { "epoch": 2.2597556168703195, "grad_norm": 0.49205532670021057, "learning_rate": 0.0005, "loss": 1.1861023902893066, "step": 28665 }, { "epoch": 2.259834450137958, "grad_norm": 0.4740235209465027, "learning_rate": 0.0005, "loss": 1.3016090393066406, "step": 28666 }, { "epoch": 2.2599132834055973, "grad_norm": 0.42701515555381775, "learning_rate": 0.0005, "loss": 1.1839873790740967, "step": 28667 }, { "epoch": 2.259992116673236, "grad_norm": 0.5239979028701782, "learning_rate": 0.0005, "loss": 1.4440865516662598, "step": 28668 }, { "epoch": 2.260070949940875, "grad_norm": 0.4599105715751648, "learning_rate": 0.0005, "loss": 1.2971360683441162, "step": 28669 }, { "epoch": 2.2601497832085142, "grad_norm": 0.47862425446510315, "learning_rate": 0.0005, "loss": 1.3063504695892334, "step": 28670 }, { "epoch": 2.260228616476153, "grad_norm": 0.4244419038295746, "learning_rate": 0.0005, "loss": 1.2636297941207886, "step": 28671 }, { "epoch": 2.260307449743792, "grad_norm": 0.4740692377090454, "learning_rate": 0.0005, "loss": 1.3635722398757935, "step": 28672 }, { "epoch": 2.2603862830114307, "grad_norm": 0.5598007440567017, "learning_rate": 0.0005, "loss": 1.2857388257980347, "step": 28673 }, { "epoch": 2.26046511627907, "grad_norm": 0.5830134153366089, "learning_rate": 0.0005, "loss": 1.3842735290527344, "step": 28674 }, { "epoch": 2.2605439495467086, "grad_norm": 0.6141247153282166, "learning_rate": 0.0005, "loss": 1.2013238668441772, "step": 28675 }, { "epoch": 2.2606227828143477, "grad_norm": 0.4558883011341095, "learning_rate": 0.0005, "loss": 1.387843132019043, "step": 28676 }, { "epoch": 2.2607016160819864, "grad_norm": 0.5830237865447998, "learning_rate": 0.0005, "loss": 1.454613208770752, "step": 28677 }, { "epoch": 2.2607804493496255, "grad_norm": 0.4507771134376526, "learning_rate": 0.0005, "loss": 1.2601890563964844, "step": 28678 }, { "epoch": 2.2608592826172647, "grad_norm": 0.48284801840782166, "learning_rate": 0.0005, "loss": 1.3194470405578613, "step": 28679 }, { "epoch": 2.2609381158849033, "grad_norm": 0.4812576472759247, "learning_rate": 0.0005, "loss": 1.2668471336364746, "step": 28680 }, { "epoch": 2.2610169491525425, "grad_norm": 0.5040473937988281, "learning_rate": 0.0005, "loss": 1.3800030946731567, "step": 28681 }, { "epoch": 2.261095782420181, "grad_norm": 0.4795081615447998, "learning_rate": 0.0005, "loss": 1.3930151462554932, "step": 28682 }, { "epoch": 2.2611746156878203, "grad_norm": 0.49580106139183044, "learning_rate": 0.0005, "loss": 1.4160542488098145, "step": 28683 }, { "epoch": 2.2612534489554594, "grad_norm": 0.4383034110069275, "learning_rate": 0.0005, "loss": 1.295581579208374, "step": 28684 }, { "epoch": 2.261332282223098, "grad_norm": 0.6781310439109802, "learning_rate": 0.0005, "loss": 1.391574501991272, "step": 28685 }, { "epoch": 2.2614111154907373, "grad_norm": 0.5553976893424988, "learning_rate": 0.0005, "loss": 1.2887985706329346, "step": 28686 }, { "epoch": 2.261489948758376, "grad_norm": 0.5687925815582275, "learning_rate": 0.0005, "loss": 1.4284058809280396, "step": 28687 }, { "epoch": 2.261568782026015, "grad_norm": 0.39459648728370667, "learning_rate": 0.0005, "loss": 1.1737068891525269, "step": 28688 }, { "epoch": 2.2616476152936538, "grad_norm": 0.43846097588539124, "learning_rate": 0.0005, "loss": 1.1730014085769653, "step": 28689 }, { "epoch": 2.261726448561293, "grad_norm": 0.514498233795166, "learning_rate": 0.0005, "loss": 1.3545403480529785, "step": 28690 }, { "epoch": 2.2618052818289316, "grad_norm": 0.5020714998245239, "learning_rate": 0.0005, "loss": 1.4408092498779297, "step": 28691 }, { "epoch": 2.2618841150965707, "grad_norm": 0.4806060791015625, "learning_rate": 0.0005, "loss": 1.4032588005065918, "step": 28692 }, { "epoch": 2.26196294836421, "grad_norm": 0.652370810508728, "learning_rate": 0.0005, "loss": 1.269108772277832, "step": 28693 }, { "epoch": 2.2620417816318485, "grad_norm": 0.6706218123435974, "learning_rate": 0.0005, "loss": 1.4250575304031372, "step": 28694 }, { "epoch": 2.2621206148994877, "grad_norm": 0.5403236746788025, "learning_rate": 0.0005, "loss": 1.480243444442749, "step": 28695 }, { "epoch": 2.2621994481671264, "grad_norm": 0.43111929297447205, "learning_rate": 0.0005, "loss": 1.2038593292236328, "step": 28696 }, { "epoch": 2.2622782814347655, "grad_norm": 0.5164404511451721, "learning_rate": 0.0005, "loss": 1.182520866394043, "step": 28697 }, { "epoch": 2.2623571147024046, "grad_norm": 0.47229278087615967, "learning_rate": 0.0005, "loss": 1.2752552032470703, "step": 28698 }, { "epoch": 2.2624359479700433, "grad_norm": 0.6235831379890442, "learning_rate": 0.0005, "loss": 1.2548308372497559, "step": 28699 }, { "epoch": 2.2625147812376825, "grad_norm": 0.6393117904663086, "learning_rate": 0.0005, "loss": 1.3368842601776123, "step": 28700 }, { "epoch": 2.262593614505321, "grad_norm": 0.46121838688850403, "learning_rate": 0.0005, "loss": 1.363006830215454, "step": 28701 }, { "epoch": 2.2626724477729603, "grad_norm": 0.4489399790763855, "learning_rate": 0.0005, "loss": 1.1906778812408447, "step": 28702 }, { "epoch": 2.262751281040599, "grad_norm": 0.5478276014328003, "learning_rate": 0.0005, "loss": 1.159285068511963, "step": 28703 }, { "epoch": 2.262830114308238, "grad_norm": 0.6960415244102478, "learning_rate": 0.0005, "loss": 1.5913538932800293, "step": 28704 }, { "epoch": 2.262908947575877, "grad_norm": 0.6570788025856018, "learning_rate": 0.0005, "loss": 1.3991496562957764, "step": 28705 }, { "epoch": 2.262987780843516, "grad_norm": 0.5116711854934692, "learning_rate": 0.0005, "loss": 1.5002706050872803, "step": 28706 }, { "epoch": 2.263066614111155, "grad_norm": 0.570675790309906, "learning_rate": 0.0005, "loss": 1.576697826385498, "step": 28707 }, { "epoch": 2.2631454473787938, "grad_norm": 0.5803750157356262, "learning_rate": 0.0005, "loss": 1.2791972160339355, "step": 28708 }, { "epoch": 2.263224280646433, "grad_norm": 0.564689576625824, "learning_rate": 0.0005, "loss": 1.3741823434829712, "step": 28709 }, { "epoch": 2.2633031139140716, "grad_norm": 0.5586678981781006, "learning_rate": 0.0005, "loss": 1.3767905235290527, "step": 28710 }, { "epoch": 2.2633819471817107, "grad_norm": 0.4759526252746582, "learning_rate": 0.0005, "loss": 1.360079288482666, "step": 28711 }, { "epoch": 2.26346078044935, "grad_norm": 0.41562172770500183, "learning_rate": 0.0005, "loss": 1.2439899444580078, "step": 28712 }, { "epoch": 2.2635396137169885, "grad_norm": 0.527123212814331, "learning_rate": 0.0005, "loss": 1.2734010219573975, "step": 28713 }, { "epoch": 2.2636184469846277, "grad_norm": 0.5241063237190247, "learning_rate": 0.0005, "loss": 1.1639622449874878, "step": 28714 }, { "epoch": 2.2636972802522664, "grad_norm": 0.4890584647655487, "learning_rate": 0.0005, "loss": 1.263767957687378, "step": 28715 }, { "epoch": 2.2637761135199055, "grad_norm": 0.5611205697059631, "learning_rate": 0.0005, "loss": 1.4006166458129883, "step": 28716 }, { "epoch": 2.263854946787544, "grad_norm": 0.4077916145324707, "learning_rate": 0.0005, "loss": 1.2095619440078735, "step": 28717 }, { "epoch": 2.2639337800551833, "grad_norm": 0.46552401781082153, "learning_rate": 0.0005, "loss": 1.4418734312057495, "step": 28718 }, { "epoch": 2.264012613322822, "grad_norm": 0.5389455556869507, "learning_rate": 0.0005, "loss": 1.393714189529419, "step": 28719 }, { "epoch": 2.264091446590461, "grad_norm": 0.42999640107154846, "learning_rate": 0.0005, "loss": 1.1946725845336914, "step": 28720 }, { "epoch": 2.2641702798581003, "grad_norm": 0.4088696837425232, "learning_rate": 0.0005, "loss": 1.2268097400665283, "step": 28721 }, { "epoch": 2.264249113125739, "grad_norm": 0.5336601734161377, "learning_rate": 0.0005, "loss": 1.2235186100006104, "step": 28722 }, { "epoch": 2.264327946393378, "grad_norm": 0.4760437309741974, "learning_rate": 0.0005, "loss": 1.3893615007400513, "step": 28723 }, { "epoch": 2.2644067796610168, "grad_norm": 0.5204164981842041, "learning_rate": 0.0005, "loss": 1.4292628765106201, "step": 28724 }, { "epoch": 2.264485612928656, "grad_norm": 0.45131897926330566, "learning_rate": 0.0005, "loss": 1.13694429397583, "step": 28725 }, { "epoch": 2.264564446196295, "grad_norm": 0.5236952304840088, "learning_rate": 0.0005, "loss": 1.203801155090332, "step": 28726 }, { "epoch": 2.2646432794639337, "grad_norm": 0.5115203261375427, "learning_rate": 0.0005, "loss": 1.2953336238861084, "step": 28727 }, { "epoch": 2.264722112731573, "grad_norm": 0.50447016954422, "learning_rate": 0.0005, "loss": 1.3055024147033691, "step": 28728 }, { "epoch": 2.2648009459992116, "grad_norm": 0.7220543622970581, "learning_rate": 0.0005, "loss": 1.2273836135864258, "step": 28729 }, { "epoch": 2.2648797792668507, "grad_norm": 0.5353177189826965, "learning_rate": 0.0005, "loss": 1.4123568534851074, "step": 28730 }, { "epoch": 2.2649586125344894, "grad_norm": 0.4994518756866455, "learning_rate": 0.0005, "loss": 1.4557111263275146, "step": 28731 }, { "epoch": 2.2650374458021285, "grad_norm": 0.5881084203720093, "learning_rate": 0.0005, "loss": 1.4830498695373535, "step": 28732 }, { "epoch": 2.265116279069767, "grad_norm": 0.505046546459198, "learning_rate": 0.0005, "loss": 1.2132360935211182, "step": 28733 }, { "epoch": 2.2651951123374063, "grad_norm": 0.47939708828926086, "learning_rate": 0.0005, "loss": 1.3930459022521973, "step": 28734 }, { "epoch": 2.2652739456050455, "grad_norm": 0.5340130925178528, "learning_rate": 0.0005, "loss": 1.4142155647277832, "step": 28735 }, { "epoch": 2.265352778872684, "grad_norm": 0.4709519147872925, "learning_rate": 0.0005, "loss": 1.2595775127410889, "step": 28736 }, { "epoch": 2.2654316121403233, "grad_norm": 0.6432875394821167, "learning_rate": 0.0005, "loss": 1.601877212524414, "step": 28737 }, { "epoch": 2.265510445407962, "grad_norm": 0.5021792054176331, "learning_rate": 0.0005, "loss": 1.3646538257598877, "step": 28738 }, { "epoch": 2.265589278675601, "grad_norm": 0.4712774455547333, "learning_rate": 0.0005, "loss": 1.2970983982086182, "step": 28739 }, { "epoch": 2.2656681119432402, "grad_norm": 0.45644626021385193, "learning_rate": 0.0005, "loss": 1.0968384742736816, "step": 28740 }, { "epoch": 2.265746945210879, "grad_norm": 0.4031358063220978, "learning_rate": 0.0005, "loss": 1.1106395721435547, "step": 28741 }, { "epoch": 2.265825778478518, "grad_norm": 0.543571412563324, "learning_rate": 0.0005, "loss": 1.3254443407058716, "step": 28742 }, { "epoch": 2.2659046117461568, "grad_norm": 0.4793737530708313, "learning_rate": 0.0005, "loss": 1.3396804332733154, "step": 28743 }, { "epoch": 2.265983445013796, "grad_norm": 0.5610702037811279, "learning_rate": 0.0005, "loss": 1.2686353921890259, "step": 28744 }, { "epoch": 2.2660622782814346, "grad_norm": 0.49262556433677673, "learning_rate": 0.0005, "loss": 1.277053713798523, "step": 28745 }, { "epoch": 2.2661411115490737, "grad_norm": 0.7361495494842529, "learning_rate": 0.0005, "loss": 1.4164971113204956, "step": 28746 }, { "epoch": 2.2662199448167124, "grad_norm": 0.48239609599113464, "learning_rate": 0.0005, "loss": 1.2189713716506958, "step": 28747 }, { "epoch": 2.2662987780843515, "grad_norm": 0.4458026587963104, "learning_rate": 0.0005, "loss": 1.392974615097046, "step": 28748 }, { "epoch": 2.2663776113519907, "grad_norm": 0.4428337812423706, "learning_rate": 0.0005, "loss": 1.3235840797424316, "step": 28749 }, { "epoch": 2.2664564446196294, "grad_norm": 0.4059251546859741, "learning_rate": 0.0005, "loss": 1.2086803913116455, "step": 28750 }, { "epoch": 2.2665352778872685, "grad_norm": 0.492516428232193, "learning_rate": 0.0005, "loss": 1.2353472709655762, "step": 28751 }, { "epoch": 2.266614111154907, "grad_norm": 0.5309558510780334, "learning_rate": 0.0005, "loss": 1.6023588180541992, "step": 28752 }, { "epoch": 2.2666929444225463, "grad_norm": 0.5013905763626099, "learning_rate": 0.0005, "loss": 1.4082918167114258, "step": 28753 }, { "epoch": 2.2667717776901855, "grad_norm": 0.5307531356811523, "learning_rate": 0.0005, "loss": 1.2755851745605469, "step": 28754 }, { "epoch": 2.266850610957824, "grad_norm": 0.48447731137275696, "learning_rate": 0.0005, "loss": 1.1547021865844727, "step": 28755 }, { "epoch": 2.2669294442254633, "grad_norm": 0.6270573735237122, "learning_rate": 0.0005, "loss": 1.4864370822906494, "step": 28756 }, { "epoch": 2.267008277493102, "grad_norm": 0.5123396515846252, "learning_rate": 0.0005, "loss": 1.310857892036438, "step": 28757 }, { "epoch": 2.267087110760741, "grad_norm": 0.5316138863563538, "learning_rate": 0.0005, "loss": 1.3312480449676514, "step": 28758 }, { "epoch": 2.26716594402838, "grad_norm": 0.5192509889602661, "learning_rate": 0.0005, "loss": 1.2742359638214111, "step": 28759 }, { "epoch": 2.267244777296019, "grad_norm": 0.6670733690261841, "learning_rate": 0.0005, "loss": 1.2683939933776855, "step": 28760 }, { "epoch": 2.267323610563658, "grad_norm": 0.6514979600906372, "learning_rate": 0.0005, "loss": 1.3163120746612549, "step": 28761 }, { "epoch": 2.2674024438312967, "grad_norm": 0.4315471351146698, "learning_rate": 0.0005, "loss": 1.2290453910827637, "step": 28762 }, { "epoch": 2.267481277098936, "grad_norm": 0.46979400515556335, "learning_rate": 0.0005, "loss": 1.3135883808135986, "step": 28763 }, { "epoch": 2.2675601103665746, "grad_norm": 0.4096827805042267, "learning_rate": 0.0005, "loss": 1.2364261150360107, "step": 28764 }, { "epoch": 2.2676389436342137, "grad_norm": 0.8463770747184753, "learning_rate": 0.0005, "loss": 1.3518683910369873, "step": 28765 }, { "epoch": 2.2677177769018524, "grad_norm": 0.4325273036956787, "learning_rate": 0.0005, "loss": 1.1271560192108154, "step": 28766 }, { "epoch": 2.2677966101694915, "grad_norm": 0.481896311044693, "learning_rate": 0.0005, "loss": 1.3385708332061768, "step": 28767 }, { "epoch": 2.2678754434371307, "grad_norm": 0.5071457028388977, "learning_rate": 0.0005, "loss": 1.3817973136901855, "step": 28768 }, { "epoch": 2.2679542767047693, "grad_norm": 0.5773692727088928, "learning_rate": 0.0005, "loss": 1.2560632228851318, "step": 28769 }, { "epoch": 2.2680331099724085, "grad_norm": 0.45132115483283997, "learning_rate": 0.0005, "loss": 1.2264095544815063, "step": 28770 }, { "epoch": 2.268111943240047, "grad_norm": 0.4201437532901764, "learning_rate": 0.0005, "loss": 1.4269137382507324, "step": 28771 }, { "epoch": 2.2681907765076863, "grad_norm": 0.5378060936927795, "learning_rate": 0.0005, "loss": 1.494061827659607, "step": 28772 }, { "epoch": 2.2682696097753254, "grad_norm": 0.43205249309539795, "learning_rate": 0.0005, "loss": 1.2214252948760986, "step": 28773 }, { "epoch": 2.268348443042964, "grad_norm": 0.5019948482513428, "learning_rate": 0.0005, "loss": 1.4207597970962524, "step": 28774 }, { "epoch": 2.2684272763106033, "grad_norm": 0.5277582406997681, "learning_rate": 0.0005, "loss": 1.2076700925827026, "step": 28775 }, { "epoch": 2.268506109578242, "grad_norm": 0.5389999747276306, "learning_rate": 0.0005, "loss": 1.3235676288604736, "step": 28776 }, { "epoch": 2.268584942845881, "grad_norm": 0.5367347002029419, "learning_rate": 0.0005, "loss": 1.3192858695983887, "step": 28777 }, { "epoch": 2.2686637761135198, "grad_norm": 0.5410000085830688, "learning_rate": 0.0005, "loss": 1.3934729099273682, "step": 28778 }, { "epoch": 2.268742609381159, "grad_norm": 0.43384894728660583, "learning_rate": 0.0005, "loss": 1.3311572074890137, "step": 28779 }, { "epoch": 2.2688214426487976, "grad_norm": 0.5378380417823792, "learning_rate": 0.0005, "loss": 1.2939372062683105, "step": 28780 }, { "epoch": 2.2689002759164367, "grad_norm": 0.8561527132987976, "learning_rate": 0.0005, "loss": 1.6113122701644897, "step": 28781 }, { "epoch": 2.268979109184076, "grad_norm": 0.4938151240348816, "learning_rate": 0.0005, "loss": 1.3416550159454346, "step": 28782 }, { "epoch": 2.2690579424517145, "grad_norm": 0.5362287163734436, "learning_rate": 0.0005, "loss": 1.4850181341171265, "step": 28783 }, { "epoch": 2.2691367757193537, "grad_norm": 0.5445340871810913, "learning_rate": 0.0005, "loss": 1.346372365951538, "step": 28784 }, { "epoch": 2.2692156089869924, "grad_norm": 0.4875338077545166, "learning_rate": 0.0005, "loss": 1.4322799444198608, "step": 28785 }, { "epoch": 2.2692944422546315, "grad_norm": 0.557942807674408, "learning_rate": 0.0005, "loss": 1.1209882497787476, "step": 28786 }, { "epoch": 2.2693732755222706, "grad_norm": 0.5376218557357788, "learning_rate": 0.0005, "loss": 1.4283418655395508, "step": 28787 }, { "epoch": 2.2694521087899093, "grad_norm": 0.46929407119750977, "learning_rate": 0.0005, "loss": 1.3462274074554443, "step": 28788 }, { "epoch": 2.2695309420575485, "grad_norm": 0.5353450775146484, "learning_rate": 0.0005, "loss": 1.6046667098999023, "step": 28789 }, { "epoch": 2.269609775325187, "grad_norm": 0.501680314540863, "learning_rate": 0.0005, "loss": 1.158231496810913, "step": 28790 }, { "epoch": 2.2696886085928263, "grad_norm": 0.4877545237541199, "learning_rate": 0.0005, "loss": 1.2038612365722656, "step": 28791 }, { "epoch": 2.269767441860465, "grad_norm": 0.4750353991985321, "learning_rate": 0.0005, "loss": 1.2280148267745972, "step": 28792 }, { "epoch": 2.269846275128104, "grad_norm": 0.46791404485702515, "learning_rate": 0.0005, "loss": 1.2125284671783447, "step": 28793 }, { "epoch": 2.269925108395743, "grad_norm": 0.4785260856151581, "learning_rate": 0.0005, "loss": 1.389780044555664, "step": 28794 }, { "epoch": 2.270003941663382, "grad_norm": 0.4734067916870117, "learning_rate": 0.0005, "loss": 1.2146186828613281, "step": 28795 }, { "epoch": 2.270082774931021, "grad_norm": 0.4887966215610504, "learning_rate": 0.0005, "loss": 1.4081064462661743, "step": 28796 }, { "epoch": 2.2701616081986598, "grad_norm": 0.544192373752594, "learning_rate": 0.0005, "loss": 1.1972970962524414, "step": 28797 }, { "epoch": 2.270240441466299, "grad_norm": 0.6477658152580261, "learning_rate": 0.0005, "loss": 1.3953951597213745, "step": 28798 }, { "epoch": 2.2703192747339376, "grad_norm": 0.4699813425540924, "learning_rate": 0.0005, "loss": 1.3717188835144043, "step": 28799 }, { "epoch": 2.2703981080015767, "grad_norm": 0.5209044218063354, "learning_rate": 0.0005, "loss": 1.273726463317871, "step": 28800 }, { "epoch": 2.270476941269216, "grad_norm": 0.7051225900650024, "learning_rate": 0.0005, "loss": 1.3476545810699463, "step": 28801 }, { "epoch": 2.2705557745368545, "grad_norm": 0.3557417392730713, "learning_rate": 0.0005, "loss": 1.154286503791809, "step": 28802 }, { "epoch": 2.2706346078044937, "grad_norm": 0.544378936290741, "learning_rate": 0.0005, "loss": 1.3234714269638062, "step": 28803 }, { "epoch": 2.2707134410721324, "grad_norm": 0.4843049943447113, "learning_rate": 0.0005, "loss": 1.1684317588806152, "step": 28804 }, { "epoch": 2.2707922743397715, "grad_norm": 0.4913036525249481, "learning_rate": 0.0005, "loss": 1.3234035968780518, "step": 28805 }, { "epoch": 2.27087110760741, "grad_norm": 0.6198538541793823, "learning_rate": 0.0005, "loss": 1.5628716945648193, "step": 28806 }, { "epoch": 2.2709499408750493, "grad_norm": 0.5377777218818665, "learning_rate": 0.0005, "loss": 1.1692352294921875, "step": 28807 }, { "epoch": 2.271028774142688, "grad_norm": 0.632902204990387, "learning_rate": 0.0005, "loss": 1.4905688762664795, "step": 28808 }, { "epoch": 2.271107607410327, "grad_norm": 0.47781118750572205, "learning_rate": 0.0005, "loss": 1.3351075649261475, "step": 28809 }, { "epoch": 2.2711864406779663, "grad_norm": 0.5152208209037781, "learning_rate": 0.0005, "loss": 1.4554052352905273, "step": 28810 }, { "epoch": 2.271265273945605, "grad_norm": 0.42385706305503845, "learning_rate": 0.0005, "loss": 1.1814744472503662, "step": 28811 }, { "epoch": 2.271344107213244, "grad_norm": 0.5478805303573608, "learning_rate": 0.0005, "loss": 1.466605305671692, "step": 28812 }, { "epoch": 2.271422940480883, "grad_norm": 0.5442754626274109, "learning_rate": 0.0005, "loss": 1.6355924606323242, "step": 28813 }, { "epoch": 2.271501773748522, "grad_norm": 0.4899595081806183, "learning_rate": 0.0005, "loss": 1.378572702407837, "step": 28814 }, { "epoch": 2.271580607016161, "grad_norm": 0.5275533199310303, "learning_rate": 0.0005, "loss": 1.3384524583816528, "step": 28815 }, { "epoch": 2.2716594402837997, "grad_norm": 0.6180388927459717, "learning_rate": 0.0005, "loss": 1.5955700874328613, "step": 28816 }, { "epoch": 2.271738273551439, "grad_norm": 0.481843501329422, "learning_rate": 0.0005, "loss": 1.404659390449524, "step": 28817 }, { "epoch": 2.2718171068190776, "grad_norm": 0.5893514752388, "learning_rate": 0.0005, "loss": 1.5815688371658325, "step": 28818 }, { "epoch": 2.2718959400867167, "grad_norm": 0.3858608603477478, "learning_rate": 0.0005, "loss": 1.2499809265136719, "step": 28819 }, { "epoch": 2.2719747733543554, "grad_norm": 0.5468712449073792, "learning_rate": 0.0005, "loss": 1.3539940118789673, "step": 28820 }, { "epoch": 2.2720536066219945, "grad_norm": 0.5281586647033691, "learning_rate": 0.0005, "loss": 1.5415072441101074, "step": 28821 }, { "epoch": 2.272132439889633, "grad_norm": 0.448169082403183, "learning_rate": 0.0005, "loss": 1.2637277841567993, "step": 28822 }, { "epoch": 2.2722112731572723, "grad_norm": 0.43425434827804565, "learning_rate": 0.0005, "loss": 1.3092193603515625, "step": 28823 }, { "epoch": 2.2722901064249115, "grad_norm": 0.6949057579040527, "learning_rate": 0.0005, "loss": 1.1254756450653076, "step": 28824 }, { "epoch": 2.27236893969255, "grad_norm": 0.49661022424697876, "learning_rate": 0.0005, "loss": 1.3267465829849243, "step": 28825 }, { "epoch": 2.2724477729601893, "grad_norm": 0.6948183178901672, "learning_rate": 0.0005, "loss": 1.369750738143921, "step": 28826 }, { "epoch": 2.272526606227828, "grad_norm": 0.4785037338733673, "learning_rate": 0.0005, "loss": 1.3760770559310913, "step": 28827 }, { "epoch": 2.272605439495467, "grad_norm": 0.5395675897598267, "learning_rate": 0.0005, "loss": 1.410213828086853, "step": 28828 }, { "epoch": 2.2726842727631063, "grad_norm": 0.4778164029121399, "learning_rate": 0.0005, "loss": 1.2401304244995117, "step": 28829 }, { "epoch": 2.272763106030745, "grad_norm": 0.42769280076026917, "learning_rate": 0.0005, "loss": 1.2183349132537842, "step": 28830 }, { "epoch": 2.272841939298384, "grad_norm": 0.5127547979354858, "learning_rate": 0.0005, "loss": 1.2996811866760254, "step": 28831 }, { "epoch": 2.2729207725660228, "grad_norm": 0.4021841287612915, "learning_rate": 0.0005, "loss": 1.164615511894226, "step": 28832 }, { "epoch": 2.272999605833662, "grad_norm": 0.6223256587982178, "learning_rate": 0.0005, "loss": 1.4194685220718384, "step": 28833 }, { "epoch": 2.2730784391013006, "grad_norm": 0.4472966194152832, "learning_rate": 0.0005, "loss": 1.1657480001449585, "step": 28834 }, { "epoch": 2.2731572723689397, "grad_norm": 0.40930676460266113, "learning_rate": 0.0005, "loss": 1.2861788272857666, "step": 28835 }, { "epoch": 2.2732361056365784, "grad_norm": 0.5021288394927979, "learning_rate": 0.0005, "loss": 1.2564826011657715, "step": 28836 }, { "epoch": 2.2733149389042175, "grad_norm": 0.4681864380836487, "learning_rate": 0.0005, "loss": 1.36588454246521, "step": 28837 }, { "epoch": 2.2733937721718567, "grad_norm": 0.5304824709892273, "learning_rate": 0.0005, "loss": 1.3773918151855469, "step": 28838 }, { "epoch": 2.2734726054394954, "grad_norm": 0.507178544998169, "learning_rate": 0.0005, "loss": 1.2488813400268555, "step": 28839 }, { "epoch": 2.2735514387071345, "grad_norm": 0.537821352481842, "learning_rate": 0.0005, "loss": 1.4266693592071533, "step": 28840 }, { "epoch": 2.273630271974773, "grad_norm": 0.48898085951805115, "learning_rate": 0.0005, "loss": 1.3599013090133667, "step": 28841 }, { "epoch": 2.2737091052424123, "grad_norm": 0.41889438033103943, "learning_rate": 0.0005, "loss": 1.252770185470581, "step": 28842 }, { "epoch": 2.2737879385100515, "grad_norm": 0.6777895092964172, "learning_rate": 0.0005, "loss": 1.4775676727294922, "step": 28843 }, { "epoch": 2.27386677177769, "grad_norm": 0.4639146029949188, "learning_rate": 0.0005, "loss": 1.463588833808899, "step": 28844 }, { "epoch": 2.2739456050453293, "grad_norm": 0.557970404624939, "learning_rate": 0.0005, "loss": 1.3970444202423096, "step": 28845 }, { "epoch": 2.274024438312968, "grad_norm": 0.38126352429389954, "learning_rate": 0.0005, "loss": 1.1444237232208252, "step": 28846 }, { "epoch": 2.274103271580607, "grad_norm": 0.4865472614765167, "learning_rate": 0.0005, "loss": 1.552847981452942, "step": 28847 }, { "epoch": 2.274182104848246, "grad_norm": 0.4555506408214569, "learning_rate": 0.0005, "loss": 1.2910454273223877, "step": 28848 }, { "epoch": 2.274260938115885, "grad_norm": 0.5290615558624268, "learning_rate": 0.0005, "loss": 1.373844027519226, "step": 28849 }, { "epoch": 2.2743397713835236, "grad_norm": 0.563605546951294, "learning_rate": 0.0005, "loss": 1.2094199657440186, "step": 28850 }, { "epoch": 2.2744186046511627, "grad_norm": 0.48221516609191895, "learning_rate": 0.0005, "loss": 1.260610818862915, "step": 28851 }, { "epoch": 2.274497437918802, "grad_norm": 0.47556474804878235, "learning_rate": 0.0005, "loss": 1.3924750089645386, "step": 28852 }, { "epoch": 2.2745762711864406, "grad_norm": 0.6252780556678772, "learning_rate": 0.0005, "loss": 1.359950065612793, "step": 28853 }, { "epoch": 2.2746551044540797, "grad_norm": 0.5407022833824158, "learning_rate": 0.0005, "loss": 1.3925741910934448, "step": 28854 }, { "epoch": 2.2747339377217184, "grad_norm": 0.5285715460777283, "learning_rate": 0.0005, "loss": 1.3465291261672974, "step": 28855 }, { "epoch": 2.2748127709893575, "grad_norm": 0.620711088180542, "learning_rate": 0.0005, "loss": 1.2612112760543823, "step": 28856 }, { "epoch": 2.2748916042569967, "grad_norm": 0.583773136138916, "learning_rate": 0.0005, "loss": 1.5929369926452637, "step": 28857 }, { "epoch": 2.2749704375246353, "grad_norm": 0.4690468907356262, "learning_rate": 0.0005, "loss": 1.2977149486541748, "step": 28858 }, { "epoch": 2.2750492707922745, "grad_norm": 0.5316932201385498, "learning_rate": 0.0005, "loss": 1.456287145614624, "step": 28859 }, { "epoch": 2.275128104059913, "grad_norm": 0.515688419342041, "learning_rate": 0.0005, "loss": 1.4563817977905273, "step": 28860 }, { "epoch": 2.2752069373275523, "grad_norm": 0.5177460312843323, "learning_rate": 0.0005, "loss": 1.2350434064865112, "step": 28861 }, { "epoch": 2.275285770595191, "grad_norm": 0.42966362833976746, "learning_rate": 0.0005, "loss": 1.2275928258895874, "step": 28862 }, { "epoch": 2.27536460386283, "grad_norm": 0.8067923188209534, "learning_rate": 0.0005, "loss": 1.492238998413086, "step": 28863 }, { "epoch": 2.275443437130469, "grad_norm": 0.624171793460846, "learning_rate": 0.0005, "loss": 1.4303257465362549, "step": 28864 }, { "epoch": 2.275522270398108, "grad_norm": 0.4656570255756378, "learning_rate": 0.0005, "loss": 1.2759459018707275, "step": 28865 }, { "epoch": 2.275601103665747, "grad_norm": 0.6353452205657959, "learning_rate": 0.0005, "loss": 1.5167031288146973, "step": 28866 }, { "epoch": 2.2756799369333858, "grad_norm": 0.435785710811615, "learning_rate": 0.0005, "loss": 1.2651597261428833, "step": 28867 }, { "epoch": 2.275758770201025, "grad_norm": 0.4607624113559723, "learning_rate": 0.0005, "loss": 1.2592748403549194, "step": 28868 }, { "epoch": 2.2758376034686636, "grad_norm": 0.4930793344974518, "learning_rate": 0.0005, "loss": 1.4066593647003174, "step": 28869 }, { "epoch": 2.2759164367363027, "grad_norm": 0.5191872119903564, "learning_rate": 0.0005, "loss": 1.5706866979599, "step": 28870 }, { "epoch": 2.275995270003942, "grad_norm": 0.4701330065727234, "learning_rate": 0.0005, "loss": 1.2940752506256104, "step": 28871 }, { "epoch": 2.2760741032715806, "grad_norm": 0.3951129913330078, "learning_rate": 0.0005, "loss": 1.1282718181610107, "step": 28872 }, { "epoch": 2.2761529365392197, "grad_norm": 0.575200617313385, "learning_rate": 0.0005, "loss": 1.325427770614624, "step": 28873 }, { "epoch": 2.2762317698068584, "grad_norm": 0.5881286859512329, "learning_rate": 0.0005, "loss": 1.4966659545898438, "step": 28874 }, { "epoch": 2.2763106030744975, "grad_norm": 0.5207692384719849, "learning_rate": 0.0005, "loss": 1.4154260158538818, "step": 28875 }, { "epoch": 2.276389436342136, "grad_norm": 0.44563430547714233, "learning_rate": 0.0005, "loss": 1.183640956878662, "step": 28876 }, { "epoch": 2.2764682696097753, "grad_norm": 0.48216819763183594, "learning_rate": 0.0005, "loss": 1.1883273124694824, "step": 28877 }, { "epoch": 2.276547102877414, "grad_norm": 0.5350058674812317, "learning_rate": 0.0005, "loss": 1.3486121892929077, "step": 28878 }, { "epoch": 2.276625936145053, "grad_norm": 0.6233708262443542, "learning_rate": 0.0005, "loss": 1.192752718925476, "step": 28879 }, { "epoch": 2.2767047694126923, "grad_norm": 0.4946453869342804, "learning_rate": 0.0005, "loss": 1.4509103298187256, "step": 28880 }, { "epoch": 2.276783602680331, "grad_norm": 0.5229690670967102, "learning_rate": 0.0005, "loss": 1.4164023399353027, "step": 28881 }, { "epoch": 2.27686243594797, "grad_norm": 0.590629518032074, "learning_rate": 0.0005, "loss": 1.4988770484924316, "step": 28882 }, { "epoch": 2.276941269215609, "grad_norm": 0.5044748783111572, "learning_rate": 0.0005, "loss": 1.2128697633743286, "step": 28883 }, { "epoch": 2.277020102483248, "grad_norm": 0.6225656867027283, "learning_rate": 0.0005, "loss": 1.3379298448562622, "step": 28884 }, { "epoch": 2.277098935750887, "grad_norm": 0.4183368682861328, "learning_rate": 0.0005, "loss": 1.2288215160369873, "step": 28885 }, { "epoch": 2.2771777690185258, "grad_norm": 0.6082575917243958, "learning_rate": 0.0005, "loss": 1.440094232559204, "step": 28886 }, { "epoch": 2.277256602286165, "grad_norm": 0.5082867741584778, "learning_rate": 0.0005, "loss": 1.3625946044921875, "step": 28887 }, { "epoch": 2.2773354355538036, "grad_norm": 0.43178457021713257, "learning_rate": 0.0005, "loss": 1.1957921981811523, "step": 28888 }, { "epoch": 2.2774142688214427, "grad_norm": 0.5130634307861328, "learning_rate": 0.0005, "loss": 1.3703508377075195, "step": 28889 }, { "epoch": 2.2774931020890814, "grad_norm": 0.5833248496055603, "learning_rate": 0.0005, "loss": 1.5214035511016846, "step": 28890 }, { "epoch": 2.2775719353567205, "grad_norm": 0.45847272872924805, "learning_rate": 0.0005, "loss": 1.3980176448822021, "step": 28891 }, { "epoch": 2.2776507686243597, "grad_norm": 0.5289212465286255, "learning_rate": 0.0005, "loss": 1.1943796873092651, "step": 28892 }, { "epoch": 2.2777296018919984, "grad_norm": 0.4836083948612213, "learning_rate": 0.0005, "loss": 1.223038911819458, "step": 28893 }, { "epoch": 2.2778084351596375, "grad_norm": 0.46608513593673706, "learning_rate": 0.0005, "loss": 1.185274362564087, "step": 28894 }, { "epoch": 2.277887268427276, "grad_norm": 0.586902379989624, "learning_rate": 0.0005, "loss": 1.5074772834777832, "step": 28895 }, { "epoch": 2.2779661016949153, "grad_norm": 0.4560072124004364, "learning_rate": 0.0005, "loss": 1.3642983436584473, "step": 28896 }, { "epoch": 2.278044934962554, "grad_norm": 0.40629616379737854, "learning_rate": 0.0005, "loss": 1.2842553853988647, "step": 28897 }, { "epoch": 2.278123768230193, "grad_norm": 0.562389075756073, "learning_rate": 0.0005, "loss": 1.3100537061691284, "step": 28898 }, { "epoch": 2.2782026014978323, "grad_norm": 0.5597556829452515, "learning_rate": 0.0005, "loss": 1.3995780944824219, "step": 28899 }, { "epoch": 2.278281434765471, "grad_norm": 0.5543556809425354, "learning_rate": 0.0005, "loss": 1.4597312211990356, "step": 28900 }, { "epoch": 2.27836026803311, "grad_norm": 0.4212431013584137, "learning_rate": 0.0005, "loss": 1.2454673051834106, "step": 28901 }, { "epoch": 2.278439101300749, "grad_norm": 0.4159720540046692, "learning_rate": 0.0005, "loss": 1.1702320575714111, "step": 28902 }, { "epoch": 2.278517934568388, "grad_norm": 0.6436994075775146, "learning_rate": 0.0005, "loss": 1.5103111267089844, "step": 28903 }, { "epoch": 2.278596767836027, "grad_norm": 0.5747270584106445, "learning_rate": 0.0005, "loss": 1.5190162658691406, "step": 28904 }, { "epoch": 2.2786756011036657, "grad_norm": 0.5065715312957764, "learning_rate": 0.0005, "loss": 1.2715466022491455, "step": 28905 }, { "epoch": 2.278754434371305, "grad_norm": 0.4289155602455139, "learning_rate": 0.0005, "loss": 1.3333591222763062, "step": 28906 }, { "epoch": 2.2788332676389436, "grad_norm": 0.5330187082290649, "learning_rate": 0.0005, "loss": 1.3980236053466797, "step": 28907 }, { "epoch": 2.2789121009065827, "grad_norm": 0.4876154959201813, "learning_rate": 0.0005, "loss": 1.315696358680725, "step": 28908 }, { "epoch": 2.2789909341742214, "grad_norm": 0.37179237604141235, "learning_rate": 0.0005, "loss": 1.1410378217697144, "step": 28909 }, { "epoch": 2.2790697674418605, "grad_norm": 0.5552800297737122, "learning_rate": 0.0005, "loss": 1.446557879447937, "step": 28910 }, { "epoch": 2.279148600709499, "grad_norm": 0.6509450674057007, "learning_rate": 0.0005, "loss": 1.3544530868530273, "step": 28911 }, { "epoch": 2.2792274339771383, "grad_norm": 0.5134510397911072, "learning_rate": 0.0005, "loss": 1.2907625436782837, "step": 28912 }, { "epoch": 2.2793062672447775, "grad_norm": 0.5843950510025024, "learning_rate": 0.0005, "loss": 1.535287618637085, "step": 28913 }, { "epoch": 2.279385100512416, "grad_norm": 0.6351397037506104, "learning_rate": 0.0005, "loss": 1.733862280845642, "step": 28914 }, { "epoch": 2.2794639337800553, "grad_norm": 0.41935619711875916, "learning_rate": 0.0005, "loss": 1.3505584001541138, "step": 28915 }, { "epoch": 2.279542767047694, "grad_norm": 0.5180190205574036, "learning_rate": 0.0005, "loss": 1.3705379962921143, "step": 28916 }, { "epoch": 2.279621600315333, "grad_norm": 0.5025182366371155, "learning_rate": 0.0005, "loss": 1.2459625005722046, "step": 28917 }, { "epoch": 2.2797004335829723, "grad_norm": 0.5950396656990051, "learning_rate": 0.0005, "loss": 1.4684159755706787, "step": 28918 }, { "epoch": 2.279779266850611, "grad_norm": 0.513068437576294, "learning_rate": 0.0005, "loss": 1.405266284942627, "step": 28919 }, { "epoch": 2.27985810011825, "grad_norm": 0.36954644322395325, "learning_rate": 0.0005, "loss": 1.051224946975708, "step": 28920 }, { "epoch": 2.2799369333858888, "grad_norm": 0.5664017796516418, "learning_rate": 0.0005, "loss": 1.4693732261657715, "step": 28921 }, { "epoch": 2.280015766653528, "grad_norm": 0.35569262504577637, "learning_rate": 0.0005, "loss": 1.000015139579773, "step": 28922 }, { "epoch": 2.2800945999211666, "grad_norm": 0.49027568101882935, "learning_rate": 0.0005, "loss": 1.273602843284607, "step": 28923 }, { "epoch": 2.2801734331888057, "grad_norm": 0.6512258052825928, "learning_rate": 0.0005, "loss": 1.4611037969589233, "step": 28924 }, { "epoch": 2.2802522664564444, "grad_norm": 0.6181666254997253, "learning_rate": 0.0005, "loss": 1.3554993867874146, "step": 28925 }, { "epoch": 2.2803310997240835, "grad_norm": 0.4926924407482147, "learning_rate": 0.0005, "loss": 1.3885414600372314, "step": 28926 }, { "epoch": 2.2804099329917227, "grad_norm": 0.5878501534461975, "learning_rate": 0.0005, "loss": 1.4155666828155518, "step": 28927 }, { "epoch": 2.2804887662593614, "grad_norm": 0.4216955602169037, "learning_rate": 0.0005, "loss": 1.2671139240264893, "step": 28928 }, { "epoch": 2.2805675995270005, "grad_norm": 0.671625554561615, "learning_rate": 0.0005, "loss": 1.6871938705444336, "step": 28929 }, { "epoch": 2.280646432794639, "grad_norm": 0.6402845978736877, "learning_rate": 0.0005, "loss": 1.3278591632843018, "step": 28930 }, { "epoch": 2.2807252660622783, "grad_norm": 0.6084861159324646, "learning_rate": 0.0005, "loss": 1.4894609451293945, "step": 28931 }, { "epoch": 2.2808040993299175, "grad_norm": 0.5129626989364624, "learning_rate": 0.0005, "loss": 1.3679628372192383, "step": 28932 }, { "epoch": 2.280882932597556, "grad_norm": 0.536119818687439, "learning_rate": 0.0005, "loss": 1.3280935287475586, "step": 28933 }, { "epoch": 2.2809617658651953, "grad_norm": 0.44859305024147034, "learning_rate": 0.0005, "loss": 1.2001934051513672, "step": 28934 }, { "epoch": 2.281040599132834, "grad_norm": 0.4442327618598938, "learning_rate": 0.0005, "loss": 1.193136215209961, "step": 28935 }, { "epoch": 2.281119432400473, "grad_norm": 0.5250164866447449, "learning_rate": 0.0005, "loss": 1.3367925882339478, "step": 28936 }, { "epoch": 2.281198265668112, "grad_norm": 0.45913591980934143, "learning_rate": 0.0005, "loss": 1.2889533042907715, "step": 28937 }, { "epoch": 2.281277098935751, "grad_norm": 0.5640679597854614, "learning_rate": 0.0005, "loss": 1.3262579441070557, "step": 28938 }, { "epoch": 2.2813559322033896, "grad_norm": 0.4726530611515045, "learning_rate": 0.0005, "loss": 1.4245483875274658, "step": 28939 }, { "epoch": 2.2814347654710287, "grad_norm": 0.49005162715911865, "learning_rate": 0.0005, "loss": 1.4161604642868042, "step": 28940 }, { "epoch": 2.281513598738668, "grad_norm": 0.5723552107810974, "learning_rate": 0.0005, "loss": 1.3497116565704346, "step": 28941 }, { "epoch": 2.2815924320063066, "grad_norm": 0.5368781685829163, "learning_rate": 0.0005, "loss": 1.3515723943710327, "step": 28942 }, { "epoch": 2.2816712652739457, "grad_norm": 0.47064051032066345, "learning_rate": 0.0005, "loss": 1.3090602159500122, "step": 28943 }, { "epoch": 2.2817500985415844, "grad_norm": 0.602982223033905, "learning_rate": 0.0005, "loss": 1.2206119298934937, "step": 28944 }, { "epoch": 2.2818289318092235, "grad_norm": 0.4929329454898834, "learning_rate": 0.0005, "loss": 1.367737054824829, "step": 28945 }, { "epoch": 2.2819077650768627, "grad_norm": 0.4315864145755768, "learning_rate": 0.0005, "loss": 1.2875981330871582, "step": 28946 }, { "epoch": 2.2819865983445013, "grad_norm": 0.5157550573348999, "learning_rate": 0.0005, "loss": 1.2510812282562256, "step": 28947 }, { "epoch": 2.2820654316121405, "grad_norm": 0.5882952213287354, "learning_rate": 0.0005, "loss": 1.5332090854644775, "step": 28948 }, { "epoch": 2.282144264879779, "grad_norm": 0.5854296088218689, "learning_rate": 0.0005, "loss": 1.504854679107666, "step": 28949 }, { "epoch": 2.2822230981474183, "grad_norm": 0.5184423923492432, "learning_rate": 0.0005, "loss": 1.444281816482544, "step": 28950 }, { "epoch": 2.282301931415057, "grad_norm": 0.5254591107368469, "learning_rate": 0.0005, "loss": 1.0409601926803589, "step": 28951 }, { "epoch": 2.282380764682696, "grad_norm": 0.7678333520889282, "learning_rate": 0.0005, "loss": 1.5006747245788574, "step": 28952 }, { "epoch": 2.282459597950335, "grad_norm": 0.49301496148109436, "learning_rate": 0.0005, "loss": 1.2809617519378662, "step": 28953 }, { "epoch": 2.282538431217974, "grad_norm": 0.7719206213951111, "learning_rate": 0.0005, "loss": 1.3672165870666504, "step": 28954 }, { "epoch": 2.282617264485613, "grad_norm": 0.4962874948978424, "learning_rate": 0.0005, "loss": 1.3380613327026367, "step": 28955 }, { "epoch": 2.2826960977532518, "grad_norm": 0.5950779318809509, "learning_rate": 0.0005, "loss": 1.312945008277893, "step": 28956 }, { "epoch": 2.282774931020891, "grad_norm": 0.5320306420326233, "learning_rate": 0.0005, "loss": 1.247112512588501, "step": 28957 }, { "epoch": 2.2828537642885296, "grad_norm": 0.7153400182723999, "learning_rate": 0.0005, "loss": 1.3712666034698486, "step": 28958 }, { "epoch": 2.2829325975561687, "grad_norm": 0.5024011731147766, "learning_rate": 0.0005, "loss": 1.4701833724975586, "step": 28959 }, { "epoch": 2.283011430823808, "grad_norm": 0.6635242700576782, "learning_rate": 0.0005, "loss": 1.4407918453216553, "step": 28960 }, { "epoch": 2.2830902640914466, "grad_norm": 0.48243430256843567, "learning_rate": 0.0005, "loss": 1.4822301864624023, "step": 28961 }, { "epoch": 2.2831690973590857, "grad_norm": 0.4691857099533081, "learning_rate": 0.0005, "loss": 1.299107313156128, "step": 28962 }, { "epoch": 2.2832479306267244, "grad_norm": 0.4203658699989319, "learning_rate": 0.0005, "loss": 1.2616100311279297, "step": 28963 }, { "epoch": 2.2833267638943635, "grad_norm": 0.551429271697998, "learning_rate": 0.0005, "loss": 1.4554970264434814, "step": 28964 }, { "epoch": 2.283405597162002, "grad_norm": 0.42174291610717773, "learning_rate": 0.0005, "loss": 1.0545908212661743, "step": 28965 }, { "epoch": 2.2834844304296413, "grad_norm": 0.5706891417503357, "learning_rate": 0.0005, "loss": 1.4383795261383057, "step": 28966 }, { "epoch": 2.28356326369728, "grad_norm": 0.4403604567050934, "learning_rate": 0.0005, "loss": 1.2945888042449951, "step": 28967 }, { "epoch": 2.283642096964919, "grad_norm": 0.6871048212051392, "learning_rate": 0.0005, "loss": 1.4243505001068115, "step": 28968 }, { "epoch": 2.2837209302325583, "grad_norm": 0.6017324328422546, "learning_rate": 0.0005, "loss": 1.470763087272644, "step": 28969 }, { "epoch": 2.283799763500197, "grad_norm": 0.6983972787857056, "learning_rate": 0.0005, "loss": 1.5359399318695068, "step": 28970 }, { "epoch": 2.283878596767836, "grad_norm": 0.7198784351348877, "learning_rate": 0.0005, "loss": 1.5403416156768799, "step": 28971 }, { "epoch": 2.283957430035475, "grad_norm": 0.47894516587257385, "learning_rate": 0.0005, "loss": 1.1268924474716187, "step": 28972 }, { "epoch": 2.284036263303114, "grad_norm": 0.4750271141529083, "learning_rate": 0.0005, "loss": 1.3034210205078125, "step": 28973 }, { "epoch": 2.284115096570753, "grad_norm": 0.49388837814331055, "learning_rate": 0.0005, "loss": 1.4515626430511475, "step": 28974 }, { "epoch": 2.2841939298383918, "grad_norm": 0.5757851600646973, "learning_rate": 0.0005, "loss": 1.4076391458511353, "step": 28975 }, { "epoch": 2.284272763106031, "grad_norm": 0.5688809156417847, "learning_rate": 0.0005, "loss": 1.4439880847930908, "step": 28976 }, { "epoch": 2.2843515963736696, "grad_norm": 0.5734180212020874, "learning_rate": 0.0005, "loss": 1.4091739654541016, "step": 28977 }, { "epoch": 2.2844304296413087, "grad_norm": 0.49589547514915466, "learning_rate": 0.0005, "loss": 1.274763584136963, "step": 28978 }, { "epoch": 2.2845092629089474, "grad_norm": 0.6071733236312866, "learning_rate": 0.0005, "loss": 1.5370577573776245, "step": 28979 }, { "epoch": 2.2845880961765865, "grad_norm": 0.6616787910461426, "learning_rate": 0.0005, "loss": 1.7073158025741577, "step": 28980 }, { "epoch": 2.2846669294442252, "grad_norm": 0.6052953004837036, "learning_rate": 0.0005, "loss": 1.5626847743988037, "step": 28981 }, { "epoch": 2.2847457627118644, "grad_norm": 0.4623996317386627, "learning_rate": 0.0005, "loss": 1.2568449974060059, "step": 28982 }, { "epoch": 2.2848245959795035, "grad_norm": 0.4671591520309448, "learning_rate": 0.0005, "loss": 1.4814509153366089, "step": 28983 }, { "epoch": 2.284903429247142, "grad_norm": 0.5525689721107483, "learning_rate": 0.0005, "loss": 1.503726840019226, "step": 28984 }, { "epoch": 2.2849822625147813, "grad_norm": 0.4611649811267853, "learning_rate": 0.0005, "loss": 1.469797134399414, "step": 28985 }, { "epoch": 2.28506109578242, "grad_norm": 0.44693446159362793, "learning_rate": 0.0005, "loss": 1.3303794860839844, "step": 28986 }, { "epoch": 2.285139929050059, "grad_norm": 0.41345691680908203, "learning_rate": 0.0005, "loss": 1.3490700721740723, "step": 28987 }, { "epoch": 2.2852187623176983, "grad_norm": 0.5032830834388733, "learning_rate": 0.0005, "loss": 1.3632380962371826, "step": 28988 }, { "epoch": 2.285297595585337, "grad_norm": 0.5454185009002686, "learning_rate": 0.0005, "loss": 1.4355947971343994, "step": 28989 }, { "epoch": 2.285376428852976, "grad_norm": 0.5303391218185425, "learning_rate": 0.0005, "loss": 1.2651185989379883, "step": 28990 }, { "epoch": 2.285455262120615, "grad_norm": 0.5140413641929626, "learning_rate": 0.0005, "loss": 1.237222671508789, "step": 28991 }, { "epoch": 2.285534095388254, "grad_norm": 0.478669136762619, "learning_rate": 0.0005, "loss": 1.342797875404358, "step": 28992 }, { "epoch": 2.2856129286558926, "grad_norm": 0.5332680344581604, "learning_rate": 0.0005, "loss": 1.3492395877838135, "step": 28993 }, { "epoch": 2.2856917619235317, "grad_norm": 0.5256980061531067, "learning_rate": 0.0005, "loss": 1.3532688617706299, "step": 28994 }, { "epoch": 2.2857705951911704, "grad_norm": 0.467244029045105, "learning_rate": 0.0005, "loss": 1.2659653425216675, "step": 28995 }, { "epoch": 2.2858494284588096, "grad_norm": 0.5291227698326111, "learning_rate": 0.0005, "loss": 1.3229762315750122, "step": 28996 }, { "epoch": 2.2859282617264487, "grad_norm": 0.6224977970123291, "learning_rate": 0.0005, "loss": 1.490654706954956, "step": 28997 }, { "epoch": 2.2860070949940874, "grad_norm": 0.5657118558883667, "learning_rate": 0.0005, "loss": 1.4822531938552856, "step": 28998 }, { "epoch": 2.2860859282617265, "grad_norm": 0.40433579683303833, "learning_rate": 0.0005, "loss": 1.2896279096603394, "step": 28999 }, { "epoch": 2.286164761529365, "grad_norm": 0.5325729846954346, "learning_rate": 0.0005, "loss": 1.204144835472107, "step": 29000 }, { "epoch": 2.2862435947970043, "grad_norm": 0.535036027431488, "learning_rate": 0.0005, "loss": 1.2654964923858643, "step": 29001 }, { "epoch": 2.2863224280646435, "grad_norm": 0.470975786447525, "learning_rate": 0.0005, "loss": 1.3961933851242065, "step": 29002 }, { "epoch": 2.286401261332282, "grad_norm": 0.42571237683296204, "learning_rate": 0.0005, "loss": 1.2018057107925415, "step": 29003 }, { "epoch": 2.2864800945999213, "grad_norm": 0.40841013193130493, "learning_rate": 0.0005, "loss": 1.2834306955337524, "step": 29004 }, { "epoch": 2.28655892786756, "grad_norm": 0.4853125214576721, "learning_rate": 0.0005, "loss": 1.314662218093872, "step": 29005 }, { "epoch": 2.286637761135199, "grad_norm": 0.5824649930000305, "learning_rate": 0.0005, "loss": 1.4448422193527222, "step": 29006 }, { "epoch": 2.286716594402838, "grad_norm": 0.49429789185523987, "learning_rate": 0.0005, "loss": 1.4430032968521118, "step": 29007 }, { "epoch": 2.286795427670477, "grad_norm": 0.5263201594352722, "learning_rate": 0.0005, "loss": 1.1440531015396118, "step": 29008 }, { "epoch": 2.286874260938116, "grad_norm": 1.1453770399093628, "learning_rate": 0.0005, "loss": 1.24808931350708, "step": 29009 }, { "epoch": 2.2869530942057548, "grad_norm": 0.6494777202606201, "learning_rate": 0.0005, "loss": 1.5708154439926147, "step": 29010 }, { "epoch": 2.287031927473394, "grad_norm": 0.6224031448364258, "learning_rate": 0.0005, "loss": 1.545536994934082, "step": 29011 }, { "epoch": 2.2871107607410326, "grad_norm": 0.4080049693584442, "learning_rate": 0.0005, "loss": 1.298836588859558, "step": 29012 }, { "epoch": 2.2871895940086717, "grad_norm": 0.528359591960907, "learning_rate": 0.0005, "loss": 1.3801679611206055, "step": 29013 }, { "epoch": 2.2872684272763104, "grad_norm": 0.5761806964874268, "learning_rate": 0.0005, "loss": 1.2862653732299805, "step": 29014 }, { "epoch": 2.2873472605439495, "grad_norm": 0.5054315328598022, "learning_rate": 0.0005, "loss": 1.142972707748413, "step": 29015 }, { "epoch": 2.2874260938115887, "grad_norm": 0.36531704664230347, "learning_rate": 0.0005, "loss": 1.1851903200149536, "step": 29016 }, { "epoch": 2.2875049270792274, "grad_norm": 0.5392433404922485, "learning_rate": 0.0005, "loss": 1.4159488677978516, "step": 29017 }, { "epoch": 2.2875837603468665, "grad_norm": 0.5451984405517578, "learning_rate": 0.0005, "loss": 1.4989657402038574, "step": 29018 }, { "epoch": 2.287662593614505, "grad_norm": 0.48756641149520874, "learning_rate": 0.0005, "loss": 1.282254695892334, "step": 29019 }, { "epoch": 2.2877414268821443, "grad_norm": 0.592441976070404, "learning_rate": 0.0005, "loss": 1.2489534616470337, "step": 29020 }, { "epoch": 2.2878202601497835, "grad_norm": 0.5468217134475708, "learning_rate": 0.0005, "loss": 1.4609100818634033, "step": 29021 }, { "epoch": 2.287899093417422, "grad_norm": 0.4921221435070038, "learning_rate": 0.0005, "loss": 1.2082433700561523, "step": 29022 }, { "epoch": 2.2879779266850613, "grad_norm": 0.4417054057121277, "learning_rate": 0.0005, "loss": 1.2886509895324707, "step": 29023 }, { "epoch": 2.2880567599527, "grad_norm": 0.49785295128822327, "learning_rate": 0.0005, "loss": 1.3954252004623413, "step": 29024 }, { "epoch": 2.288135593220339, "grad_norm": 0.5102649927139282, "learning_rate": 0.0005, "loss": 1.3859820365905762, "step": 29025 }, { "epoch": 2.288214426487978, "grad_norm": 0.520269513130188, "learning_rate": 0.0005, "loss": 1.293760895729065, "step": 29026 }, { "epoch": 2.288293259755617, "grad_norm": 0.49025091528892517, "learning_rate": 0.0005, "loss": 1.5053668022155762, "step": 29027 }, { "epoch": 2.2883720930232556, "grad_norm": 0.522539496421814, "learning_rate": 0.0005, "loss": 1.3656706809997559, "step": 29028 }, { "epoch": 2.2884509262908947, "grad_norm": 0.5044251084327698, "learning_rate": 0.0005, "loss": 1.31689453125, "step": 29029 }, { "epoch": 2.288529759558534, "grad_norm": 0.6468967795372009, "learning_rate": 0.0005, "loss": 1.480623722076416, "step": 29030 }, { "epoch": 2.2886085928261726, "grad_norm": 0.4834815561771393, "learning_rate": 0.0005, "loss": 1.3721246719360352, "step": 29031 }, { "epoch": 2.2886874260938117, "grad_norm": 0.5693092942237854, "learning_rate": 0.0005, "loss": 1.3099689483642578, "step": 29032 }, { "epoch": 2.2887662593614504, "grad_norm": 0.49079644680023193, "learning_rate": 0.0005, "loss": 1.3555066585540771, "step": 29033 }, { "epoch": 2.2888450926290895, "grad_norm": 0.44291365146636963, "learning_rate": 0.0005, "loss": 1.2695107460021973, "step": 29034 }, { "epoch": 2.2889239258967287, "grad_norm": 0.5794968605041504, "learning_rate": 0.0005, "loss": 1.4151601791381836, "step": 29035 }, { "epoch": 2.2890027591643674, "grad_norm": 0.5057776570320129, "learning_rate": 0.0005, "loss": 1.3678045272827148, "step": 29036 }, { "epoch": 2.2890815924320065, "grad_norm": 0.48909541964530945, "learning_rate": 0.0005, "loss": 1.3151532411575317, "step": 29037 }, { "epoch": 2.289160425699645, "grad_norm": 0.49301058053970337, "learning_rate": 0.0005, "loss": 1.3045907020568848, "step": 29038 }, { "epoch": 2.2892392589672843, "grad_norm": 0.4511166512966156, "learning_rate": 0.0005, "loss": 1.302473545074463, "step": 29039 }, { "epoch": 2.289318092234923, "grad_norm": 0.46100395917892456, "learning_rate": 0.0005, "loss": 1.3133143186569214, "step": 29040 }, { "epoch": 2.289396925502562, "grad_norm": 0.4958931505680084, "learning_rate": 0.0005, "loss": 1.4139000177383423, "step": 29041 }, { "epoch": 2.289475758770201, "grad_norm": 0.5378164052963257, "learning_rate": 0.0005, "loss": 1.362199068069458, "step": 29042 }, { "epoch": 2.28955459203784, "grad_norm": 0.43771103024482727, "learning_rate": 0.0005, "loss": 1.406479835510254, "step": 29043 }, { "epoch": 2.289633425305479, "grad_norm": 0.545386016368866, "learning_rate": 0.0005, "loss": 1.458448052406311, "step": 29044 }, { "epoch": 2.2897122585731178, "grad_norm": 0.4482380747795105, "learning_rate": 0.0005, "loss": 1.171295404434204, "step": 29045 }, { "epoch": 2.289791091840757, "grad_norm": 0.5132929086685181, "learning_rate": 0.0005, "loss": 1.3908576965332031, "step": 29046 }, { "epoch": 2.2898699251083956, "grad_norm": 0.4516247808933258, "learning_rate": 0.0005, "loss": 1.3507133722305298, "step": 29047 }, { "epoch": 2.2899487583760347, "grad_norm": 0.48992663621902466, "learning_rate": 0.0005, "loss": 1.3624781370162964, "step": 29048 }, { "epoch": 2.290027591643674, "grad_norm": 0.5118864178657532, "learning_rate": 0.0005, "loss": 1.3013842105865479, "step": 29049 }, { "epoch": 2.2901064249113126, "grad_norm": 0.5060271620750427, "learning_rate": 0.0005, "loss": 1.2890288829803467, "step": 29050 }, { "epoch": 2.2901852581789517, "grad_norm": 0.5349798798561096, "learning_rate": 0.0005, "loss": 1.3668127059936523, "step": 29051 }, { "epoch": 2.2902640914465904, "grad_norm": 0.5164602398872375, "learning_rate": 0.0005, "loss": 1.461834192276001, "step": 29052 }, { "epoch": 2.2903429247142295, "grad_norm": 0.563750147819519, "learning_rate": 0.0005, "loss": 1.5468097925186157, "step": 29053 }, { "epoch": 2.290421757981868, "grad_norm": 0.5351457595825195, "learning_rate": 0.0005, "loss": 1.2568069696426392, "step": 29054 }, { "epoch": 2.2905005912495073, "grad_norm": 0.4799802303314209, "learning_rate": 0.0005, "loss": 1.3072949647903442, "step": 29055 }, { "epoch": 2.290579424517146, "grad_norm": 0.5398627519607544, "learning_rate": 0.0005, "loss": 1.197171926498413, "step": 29056 }, { "epoch": 2.290658257784785, "grad_norm": 0.523725688457489, "learning_rate": 0.0005, "loss": 1.1927353143692017, "step": 29057 }, { "epoch": 2.2907370910524243, "grad_norm": 0.4570219814777374, "learning_rate": 0.0005, "loss": 1.3333418369293213, "step": 29058 }, { "epoch": 2.290815924320063, "grad_norm": 0.43586254119873047, "learning_rate": 0.0005, "loss": 1.268032431602478, "step": 29059 }, { "epoch": 2.290894757587702, "grad_norm": 0.5368947386741638, "learning_rate": 0.0005, "loss": 1.4132541418075562, "step": 29060 }, { "epoch": 2.290973590855341, "grad_norm": 0.5144193172454834, "learning_rate": 0.0005, "loss": 1.5438811779022217, "step": 29061 }, { "epoch": 2.29105242412298, "grad_norm": 0.5511051416397095, "learning_rate": 0.0005, "loss": 1.2964863777160645, "step": 29062 }, { "epoch": 2.291131257390619, "grad_norm": 0.46951133012771606, "learning_rate": 0.0005, "loss": 1.4145135879516602, "step": 29063 }, { "epoch": 2.2912100906582578, "grad_norm": 0.5019378066062927, "learning_rate": 0.0005, "loss": 1.4848945140838623, "step": 29064 }, { "epoch": 2.291288923925897, "grad_norm": 0.5621135830879211, "learning_rate": 0.0005, "loss": 1.39613676071167, "step": 29065 }, { "epoch": 2.2913677571935356, "grad_norm": 0.5919157862663269, "learning_rate": 0.0005, "loss": 1.5194642543792725, "step": 29066 }, { "epoch": 2.2914465904611747, "grad_norm": 0.4549678862094879, "learning_rate": 0.0005, "loss": 1.4331426620483398, "step": 29067 }, { "epoch": 2.2915254237288134, "grad_norm": 0.40976274013519287, "learning_rate": 0.0005, "loss": 1.149647831916809, "step": 29068 }, { "epoch": 2.2916042569964525, "grad_norm": 0.4621034264564514, "learning_rate": 0.0005, "loss": 1.1612548828125, "step": 29069 }, { "epoch": 2.2916830902640912, "grad_norm": 0.47881361842155457, "learning_rate": 0.0005, "loss": 1.2114346027374268, "step": 29070 }, { "epoch": 2.2917619235317304, "grad_norm": 0.5479875206947327, "learning_rate": 0.0005, "loss": 1.3679265975952148, "step": 29071 }, { "epoch": 2.2918407567993695, "grad_norm": 0.44265979528427124, "learning_rate": 0.0005, "loss": 1.3183058500289917, "step": 29072 }, { "epoch": 2.291919590067008, "grad_norm": 0.5378918051719666, "learning_rate": 0.0005, "loss": 1.559144377708435, "step": 29073 }, { "epoch": 2.2919984233346473, "grad_norm": 0.5661687254905701, "learning_rate": 0.0005, "loss": 1.518662929534912, "step": 29074 }, { "epoch": 2.292077256602286, "grad_norm": 0.5820776224136353, "learning_rate": 0.0005, "loss": 1.4326939582824707, "step": 29075 }, { "epoch": 2.292156089869925, "grad_norm": 0.4665055274963379, "learning_rate": 0.0005, "loss": 1.3953287601470947, "step": 29076 }, { "epoch": 2.2922349231375643, "grad_norm": 0.46766284108161926, "learning_rate": 0.0005, "loss": 1.3578664064407349, "step": 29077 }, { "epoch": 2.292313756405203, "grad_norm": 0.46160003542900085, "learning_rate": 0.0005, "loss": 1.2592560052871704, "step": 29078 }, { "epoch": 2.292392589672842, "grad_norm": 0.46271681785583496, "learning_rate": 0.0005, "loss": 1.3441866636276245, "step": 29079 }, { "epoch": 2.292471422940481, "grad_norm": 0.4776032269001007, "learning_rate": 0.0005, "loss": 1.5969257354736328, "step": 29080 }, { "epoch": 2.29255025620812, "grad_norm": 0.5482228994369507, "learning_rate": 0.0005, "loss": 1.158461570739746, "step": 29081 }, { "epoch": 2.2926290894757586, "grad_norm": 0.4777494966983795, "learning_rate": 0.0005, "loss": 1.2664034366607666, "step": 29082 }, { "epoch": 2.2927079227433977, "grad_norm": 0.5119122266769409, "learning_rate": 0.0005, "loss": 1.4290721416473389, "step": 29083 }, { "epoch": 2.2927867560110364, "grad_norm": 0.5380182862281799, "learning_rate": 0.0005, "loss": 1.3527984619140625, "step": 29084 }, { "epoch": 2.2928655892786756, "grad_norm": 0.9545491933822632, "learning_rate": 0.0005, "loss": 1.4154279232025146, "step": 29085 }, { "epoch": 2.2929444225463147, "grad_norm": 0.4120566248893738, "learning_rate": 0.0005, "loss": 1.1915799379348755, "step": 29086 }, { "epoch": 2.2930232558139534, "grad_norm": 0.6546730399131775, "learning_rate": 0.0005, "loss": 1.4673051834106445, "step": 29087 }, { "epoch": 2.2931020890815925, "grad_norm": 0.49661579728126526, "learning_rate": 0.0005, "loss": 1.27524995803833, "step": 29088 }, { "epoch": 2.293180922349231, "grad_norm": 0.5113427639007568, "learning_rate": 0.0005, "loss": 1.5127744674682617, "step": 29089 }, { "epoch": 2.2932597556168703, "grad_norm": 0.4201370179653168, "learning_rate": 0.0005, "loss": 1.1610159873962402, "step": 29090 }, { "epoch": 2.2933385888845095, "grad_norm": 0.4924260377883911, "learning_rate": 0.0005, "loss": 1.3104684352874756, "step": 29091 }, { "epoch": 2.293417422152148, "grad_norm": 0.4778805673122406, "learning_rate": 0.0005, "loss": 1.3802850246429443, "step": 29092 }, { "epoch": 2.2934962554197873, "grad_norm": 0.5175332427024841, "learning_rate": 0.0005, "loss": 1.2481842041015625, "step": 29093 }, { "epoch": 2.293575088687426, "grad_norm": 0.46059882640838623, "learning_rate": 0.0005, "loss": 1.3400373458862305, "step": 29094 }, { "epoch": 2.293653921955065, "grad_norm": 0.43510910868644714, "learning_rate": 0.0005, "loss": 1.2479339838027954, "step": 29095 }, { "epoch": 2.293732755222704, "grad_norm": 0.419080525636673, "learning_rate": 0.0005, "loss": 1.140971302986145, "step": 29096 }, { "epoch": 2.293811588490343, "grad_norm": 0.5990159511566162, "learning_rate": 0.0005, "loss": 1.3699870109558105, "step": 29097 }, { "epoch": 2.2938904217579816, "grad_norm": 0.4153096377849579, "learning_rate": 0.0005, "loss": 1.1616382598876953, "step": 29098 }, { "epoch": 2.2939692550256208, "grad_norm": 0.4706675708293915, "learning_rate": 0.0005, "loss": 1.3473117351531982, "step": 29099 }, { "epoch": 2.29404808829326, "grad_norm": 1.289912223815918, "learning_rate": 0.0005, "loss": 1.1599745750427246, "step": 29100 }, { "epoch": 2.2941269215608986, "grad_norm": 0.5392539501190186, "learning_rate": 0.0005, "loss": 1.609201192855835, "step": 29101 }, { "epoch": 2.2942057548285377, "grad_norm": 0.43966513872146606, "learning_rate": 0.0005, "loss": 1.304646372795105, "step": 29102 }, { "epoch": 2.2942845880961764, "grad_norm": 0.4533075988292694, "learning_rate": 0.0005, "loss": 1.1939290761947632, "step": 29103 }, { "epoch": 2.2943634213638155, "grad_norm": 0.516863226890564, "learning_rate": 0.0005, "loss": 1.2875640392303467, "step": 29104 }, { "epoch": 2.2944422546314547, "grad_norm": 0.6503157019615173, "learning_rate": 0.0005, "loss": 1.382497787475586, "step": 29105 }, { "epoch": 2.2945210878990934, "grad_norm": 0.41035395860671997, "learning_rate": 0.0005, "loss": 1.1362472772598267, "step": 29106 }, { "epoch": 2.2945999211667325, "grad_norm": 0.4769211411476135, "learning_rate": 0.0005, "loss": 1.3046962022781372, "step": 29107 }, { "epoch": 2.294678754434371, "grad_norm": 0.5300397276878357, "learning_rate": 0.0005, "loss": 1.4093009233474731, "step": 29108 }, { "epoch": 2.2947575877020103, "grad_norm": 0.4790233075618744, "learning_rate": 0.0005, "loss": 1.3253393173217773, "step": 29109 }, { "epoch": 2.294836420969649, "grad_norm": 0.47139427065849304, "learning_rate": 0.0005, "loss": 1.2176718711853027, "step": 29110 }, { "epoch": 2.294915254237288, "grad_norm": 0.4071701765060425, "learning_rate": 0.0005, "loss": 1.2436203956604004, "step": 29111 }, { "epoch": 2.294994087504927, "grad_norm": 0.46685850620269775, "learning_rate": 0.0005, "loss": 1.421795129776001, "step": 29112 }, { "epoch": 2.295072920772566, "grad_norm": 0.6146024465560913, "learning_rate": 0.0005, "loss": 1.340409278869629, "step": 29113 }, { "epoch": 2.295151754040205, "grad_norm": 0.5280202627182007, "learning_rate": 0.0005, "loss": 1.5517523288726807, "step": 29114 }, { "epoch": 2.295230587307844, "grad_norm": 0.5432345271110535, "learning_rate": 0.0005, "loss": 1.3361384868621826, "step": 29115 }, { "epoch": 2.295309420575483, "grad_norm": 0.4871332347393036, "learning_rate": 0.0005, "loss": 1.2530055046081543, "step": 29116 }, { "epoch": 2.2953882538431216, "grad_norm": 0.4338105320930481, "learning_rate": 0.0005, "loss": 1.2643578052520752, "step": 29117 }, { "epoch": 2.2954670871107608, "grad_norm": 0.4430939257144928, "learning_rate": 0.0005, "loss": 1.2417898178100586, "step": 29118 }, { "epoch": 2.2955459203784, "grad_norm": 0.6012816429138184, "learning_rate": 0.0005, "loss": 1.5006041526794434, "step": 29119 }, { "epoch": 2.2956247536460386, "grad_norm": 0.5436656475067139, "learning_rate": 0.0005, "loss": 1.4998939037322998, "step": 29120 }, { "epoch": 2.2957035869136777, "grad_norm": 0.5017439126968384, "learning_rate": 0.0005, "loss": 1.428248643875122, "step": 29121 }, { "epoch": 2.2957824201813164, "grad_norm": 0.45672178268432617, "learning_rate": 0.0005, "loss": 1.3469630479812622, "step": 29122 }, { "epoch": 2.2958612534489555, "grad_norm": 0.3595009744167328, "learning_rate": 0.0005, "loss": 1.1924365758895874, "step": 29123 }, { "epoch": 2.295940086716594, "grad_norm": 0.48042726516723633, "learning_rate": 0.0005, "loss": 1.423924446105957, "step": 29124 }, { "epoch": 2.2960189199842334, "grad_norm": 0.7553935647010803, "learning_rate": 0.0005, "loss": 1.3647503852844238, "step": 29125 }, { "epoch": 2.296097753251872, "grad_norm": 0.4548274278640747, "learning_rate": 0.0005, "loss": 1.4019649028778076, "step": 29126 }, { "epoch": 2.296176586519511, "grad_norm": 0.4888233244419098, "learning_rate": 0.0005, "loss": 1.206188440322876, "step": 29127 }, { "epoch": 2.2962554197871503, "grad_norm": 0.49247506260871887, "learning_rate": 0.0005, "loss": 1.3840934038162231, "step": 29128 }, { "epoch": 2.296334253054789, "grad_norm": 0.6910765767097473, "learning_rate": 0.0005, "loss": 1.5362430810928345, "step": 29129 }, { "epoch": 2.296413086322428, "grad_norm": 0.47917184233665466, "learning_rate": 0.0005, "loss": 1.1834211349487305, "step": 29130 }, { "epoch": 2.296491919590067, "grad_norm": 0.670360267162323, "learning_rate": 0.0005, "loss": 1.3303587436676025, "step": 29131 }, { "epoch": 2.296570752857706, "grad_norm": 0.5058040618896484, "learning_rate": 0.0005, "loss": 1.3044047355651855, "step": 29132 }, { "epoch": 2.296649586125345, "grad_norm": 0.4593468904495239, "learning_rate": 0.0005, "loss": 1.3600459098815918, "step": 29133 }, { "epoch": 2.2967284193929838, "grad_norm": 0.4739610552787781, "learning_rate": 0.0005, "loss": 1.1792097091674805, "step": 29134 }, { "epoch": 2.296807252660623, "grad_norm": 0.45193618535995483, "learning_rate": 0.0005, "loss": 1.1324936151504517, "step": 29135 }, { "epoch": 2.2968860859282616, "grad_norm": 0.48701751232147217, "learning_rate": 0.0005, "loss": 1.1626850366592407, "step": 29136 }, { "epoch": 2.2969649191959007, "grad_norm": 0.48099416494369507, "learning_rate": 0.0005, "loss": 1.480790138244629, "step": 29137 }, { "epoch": 2.2970437524635394, "grad_norm": 0.4057827293872833, "learning_rate": 0.0005, "loss": 1.1755844354629517, "step": 29138 }, { "epoch": 2.2971225857311786, "grad_norm": 0.5029832124710083, "learning_rate": 0.0005, "loss": 1.5190298557281494, "step": 29139 }, { "epoch": 2.2972014189988177, "grad_norm": 0.5425437688827515, "learning_rate": 0.0005, "loss": 1.249701976776123, "step": 29140 }, { "epoch": 2.2972802522664564, "grad_norm": 0.5426517724990845, "learning_rate": 0.0005, "loss": 1.3920152187347412, "step": 29141 }, { "epoch": 2.2973590855340955, "grad_norm": 0.4840370714664459, "learning_rate": 0.0005, "loss": 1.3078961372375488, "step": 29142 }, { "epoch": 2.297437918801734, "grad_norm": 0.5174046754837036, "learning_rate": 0.0005, "loss": 1.5377113819122314, "step": 29143 }, { "epoch": 2.2975167520693733, "grad_norm": 0.585950493812561, "learning_rate": 0.0005, "loss": 1.401261568069458, "step": 29144 }, { "epoch": 2.297595585337012, "grad_norm": 0.560701847076416, "learning_rate": 0.0005, "loss": 1.4339736700057983, "step": 29145 }, { "epoch": 2.297674418604651, "grad_norm": 0.43908676505088806, "learning_rate": 0.0005, "loss": 1.232649326324463, "step": 29146 }, { "epoch": 2.2977532518722903, "grad_norm": 0.5162631273269653, "learning_rate": 0.0005, "loss": 1.2057092189788818, "step": 29147 }, { "epoch": 2.297832085139929, "grad_norm": 0.6617437601089478, "learning_rate": 0.0005, "loss": 1.280212640762329, "step": 29148 }, { "epoch": 2.297910918407568, "grad_norm": 0.42062050104141235, "learning_rate": 0.0005, "loss": 1.2127649784088135, "step": 29149 }, { "epoch": 2.297989751675207, "grad_norm": 0.7822335362434387, "learning_rate": 0.0005, "loss": 1.3538944721221924, "step": 29150 }, { "epoch": 2.298068584942846, "grad_norm": 0.6650148034095764, "learning_rate": 0.0005, "loss": 1.4120720624923706, "step": 29151 }, { "epoch": 2.298147418210485, "grad_norm": 0.5490576028823853, "learning_rate": 0.0005, "loss": 1.4736462831497192, "step": 29152 }, { "epoch": 2.2982262514781238, "grad_norm": 0.5485876202583313, "learning_rate": 0.0005, "loss": 1.2135324478149414, "step": 29153 }, { "epoch": 2.298305084745763, "grad_norm": 0.4426538646221161, "learning_rate": 0.0005, "loss": 1.2010760307312012, "step": 29154 }, { "epoch": 2.2983839180134016, "grad_norm": 0.45742642879486084, "learning_rate": 0.0005, "loss": 1.3530547618865967, "step": 29155 }, { "epoch": 2.2984627512810407, "grad_norm": 0.4158134162425995, "learning_rate": 0.0005, "loss": 1.3775081634521484, "step": 29156 }, { "epoch": 2.2985415845486794, "grad_norm": 0.4725950360298157, "learning_rate": 0.0005, "loss": 1.3266751766204834, "step": 29157 }, { "epoch": 2.2986204178163185, "grad_norm": 0.4846983551979065, "learning_rate": 0.0005, "loss": 1.4340577125549316, "step": 29158 }, { "epoch": 2.2986992510839572, "grad_norm": 0.7794314026832581, "learning_rate": 0.0005, "loss": 1.407256007194519, "step": 29159 }, { "epoch": 2.2987780843515964, "grad_norm": 0.47940659523010254, "learning_rate": 0.0005, "loss": 1.4631006717681885, "step": 29160 }, { "epoch": 2.2988569176192355, "grad_norm": 0.6355035901069641, "learning_rate": 0.0005, "loss": 1.4493330717086792, "step": 29161 }, { "epoch": 2.298935750886874, "grad_norm": 0.48995333909988403, "learning_rate": 0.0005, "loss": 1.475733995437622, "step": 29162 }, { "epoch": 2.2990145841545133, "grad_norm": 0.47130274772644043, "learning_rate": 0.0005, "loss": 1.283747911453247, "step": 29163 }, { "epoch": 2.299093417422152, "grad_norm": 0.5019466876983643, "learning_rate": 0.0005, "loss": 1.3548405170440674, "step": 29164 }, { "epoch": 2.299172250689791, "grad_norm": 0.555072546005249, "learning_rate": 0.0005, "loss": 1.1704561710357666, "step": 29165 }, { "epoch": 2.2992510839574303, "grad_norm": 0.5388231873512268, "learning_rate": 0.0005, "loss": 1.3551933765411377, "step": 29166 }, { "epoch": 2.299329917225069, "grad_norm": 0.5450826287269592, "learning_rate": 0.0005, "loss": 1.3827617168426514, "step": 29167 }, { "epoch": 2.299408750492708, "grad_norm": 0.816519558429718, "learning_rate": 0.0005, "loss": 1.6260206699371338, "step": 29168 }, { "epoch": 2.299487583760347, "grad_norm": 0.5406827926635742, "learning_rate": 0.0005, "loss": 1.4332728385925293, "step": 29169 }, { "epoch": 2.299566417027986, "grad_norm": 0.44467028975486755, "learning_rate": 0.0005, "loss": 1.271571397781372, "step": 29170 }, { "epoch": 2.2996452502956246, "grad_norm": 0.4592059552669525, "learning_rate": 0.0005, "loss": 1.3114746809005737, "step": 29171 }, { "epoch": 2.2997240835632637, "grad_norm": 0.4702364206314087, "learning_rate": 0.0005, "loss": 1.324587106704712, "step": 29172 }, { "epoch": 2.2998029168309024, "grad_norm": 0.5304548740386963, "learning_rate": 0.0005, "loss": 1.3314249515533447, "step": 29173 }, { "epoch": 2.2998817500985416, "grad_norm": 0.45415520668029785, "learning_rate": 0.0005, "loss": 1.4459710121154785, "step": 29174 }, { "epoch": 2.2999605833661807, "grad_norm": 0.4487791359424591, "learning_rate": 0.0005, "loss": 1.2821615934371948, "step": 29175 }, { "epoch": 2.3000394166338194, "grad_norm": 0.5029311180114746, "learning_rate": 0.0005, "loss": 1.2928004264831543, "step": 29176 }, { "epoch": 2.3001182499014585, "grad_norm": 0.5149456858634949, "learning_rate": 0.0005, "loss": 1.2644717693328857, "step": 29177 }, { "epoch": 2.300197083169097, "grad_norm": 0.4940154552459717, "learning_rate": 0.0005, "loss": 1.2778887748718262, "step": 29178 }, { "epoch": 2.3002759164367363, "grad_norm": 0.5088689923286438, "learning_rate": 0.0005, "loss": 1.2749272584915161, "step": 29179 }, { "epoch": 2.3003547497043755, "grad_norm": 0.47395795583724976, "learning_rate": 0.0005, "loss": 1.33345627784729, "step": 29180 }, { "epoch": 2.300433582972014, "grad_norm": 0.4617423415184021, "learning_rate": 0.0005, "loss": 1.2607784271240234, "step": 29181 }, { "epoch": 2.3005124162396533, "grad_norm": 0.4582507312297821, "learning_rate": 0.0005, "loss": 1.221321702003479, "step": 29182 }, { "epoch": 2.300591249507292, "grad_norm": 0.41994497179985046, "learning_rate": 0.0005, "loss": 1.2069251537322998, "step": 29183 }, { "epoch": 2.300670082774931, "grad_norm": 0.5605723261833191, "learning_rate": 0.0005, "loss": 1.408940315246582, "step": 29184 }, { "epoch": 2.30074891604257, "grad_norm": 0.49071526527404785, "learning_rate": 0.0005, "loss": 1.2239713668823242, "step": 29185 }, { "epoch": 2.300827749310209, "grad_norm": 0.36678555607795715, "learning_rate": 0.0005, "loss": 1.0409035682678223, "step": 29186 }, { "epoch": 2.3009065825778476, "grad_norm": 0.44771501421928406, "learning_rate": 0.0005, "loss": 1.2189998626708984, "step": 29187 }, { "epoch": 2.3009854158454868, "grad_norm": 0.4330102205276489, "learning_rate": 0.0005, "loss": 1.2571914196014404, "step": 29188 }, { "epoch": 2.301064249113126, "grad_norm": 0.42676082253456116, "learning_rate": 0.0005, "loss": 1.2913669347763062, "step": 29189 }, { "epoch": 2.3011430823807646, "grad_norm": 0.5186669826507568, "learning_rate": 0.0005, "loss": 1.56605863571167, "step": 29190 }, { "epoch": 2.3012219156484037, "grad_norm": 0.4835547208786011, "learning_rate": 0.0005, "loss": 1.4109278917312622, "step": 29191 }, { "epoch": 2.3013007489160424, "grad_norm": 0.47551479935646057, "learning_rate": 0.0005, "loss": 1.3993589878082275, "step": 29192 }, { "epoch": 2.3013795821836815, "grad_norm": 0.5225621461868286, "learning_rate": 0.0005, "loss": 1.4128793478012085, "step": 29193 }, { "epoch": 2.3014584154513207, "grad_norm": 0.5115175247192383, "learning_rate": 0.0005, "loss": 1.6532127857208252, "step": 29194 }, { "epoch": 2.3015372487189594, "grad_norm": 0.42983555793762207, "learning_rate": 0.0005, "loss": 1.381091833114624, "step": 29195 }, { "epoch": 2.3016160819865985, "grad_norm": 0.5800891518592834, "learning_rate": 0.0005, "loss": 1.5278420448303223, "step": 29196 }, { "epoch": 2.301694915254237, "grad_norm": 0.39645227789878845, "learning_rate": 0.0005, "loss": 1.282818078994751, "step": 29197 }, { "epoch": 2.3017737485218763, "grad_norm": 0.5066376328468323, "learning_rate": 0.0005, "loss": 1.3215904235839844, "step": 29198 }, { "epoch": 2.301852581789515, "grad_norm": 0.5418393015861511, "learning_rate": 0.0005, "loss": 1.4419200420379639, "step": 29199 }, { "epoch": 2.301931415057154, "grad_norm": 0.5085105299949646, "learning_rate": 0.0005, "loss": 1.1754049062728882, "step": 29200 }, { "epoch": 2.302010248324793, "grad_norm": 0.39287373423576355, "learning_rate": 0.0005, "loss": 1.3095942735671997, "step": 29201 }, { "epoch": 2.302089081592432, "grad_norm": 0.49622485041618347, "learning_rate": 0.0005, "loss": 1.2978099584579468, "step": 29202 }, { "epoch": 2.302167914860071, "grad_norm": 0.4633843004703522, "learning_rate": 0.0005, "loss": 1.229011058807373, "step": 29203 }, { "epoch": 2.30224674812771, "grad_norm": 0.4828718304634094, "learning_rate": 0.0005, "loss": 1.5298473834991455, "step": 29204 }, { "epoch": 2.302325581395349, "grad_norm": 0.4989362359046936, "learning_rate": 0.0005, "loss": 1.2445975542068481, "step": 29205 }, { "epoch": 2.3024044146629876, "grad_norm": 0.40040770173072815, "learning_rate": 0.0005, "loss": 1.33236563205719, "step": 29206 }, { "epoch": 2.3024832479306268, "grad_norm": 0.4616028666496277, "learning_rate": 0.0005, "loss": 1.391531229019165, "step": 29207 }, { "epoch": 2.302562081198266, "grad_norm": 0.5290864706039429, "learning_rate": 0.0005, "loss": 1.3646259307861328, "step": 29208 }, { "epoch": 2.3026409144659046, "grad_norm": 0.41246727108955383, "learning_rate": 0.0005, "loss": 1.3559644222259521, "step": 29209 }, { "epoch": 2.3027197477335437, "grad_norm": 0.44204723834991455, "learning_rate": 0.0005, "loss": 1.2473139762878418, "step": 29210 }, { "epoch": 2.3027985810011824, "grad_norm": 0.6211934089660645, "learning_rate": 0.0005, "loss": 1.5365722179412842, "step": 29211 }, { "epoch": 2.3028774142688215, "grad_norm": 0.432576447725296, "learning_rate": 0.0005, "loss": 1.2270547151565552, "step": 29212 }, { "epoch": 2.30295624753646, "grad_norm": 0.6529998779296875, "learning_rate": 0.0005, "loss": 1.2522763013839722, "step": 29213 }, { "epoch": 2.3030350808040994, "grad_norm": 0.4566775858402252, "learning_rate": 0.0005, "loss": 1.3051097393035889, "step": 29214 }, { "epoch": 2.303113914071738, "grad_norm": 0.5338020324707031, "learning_rate": 0.0005, "loss": 1.182873010635376, "step": 29215 }, { "epoch": 2.303192747339377, "grad_norm": 0.9437354803085327, "learning_rate": 0.0005, "loss": 1.3885430097579956, "step": 29216 }, { "epoch": 2.3032715806070163, "grad_norm": 0.5440269708633423, "learning_rate": 0.0005, "loss": 1.3564460277557373, "step": 29217 }, { "epoch": 2.303350413874655, "grad_norm": 0.7309747338294983, "learning_rate": 0.0005, "loss": 1.533369541168213, "step": 29218 }, { "epoch": 2.303429247142294, "grad_norm": 0.5029125809669495, "learning_rate": 0.0005, "loss": 1.4721198081970215, "step": 29219 }, { "epoch": 2.303508080409933, "grad_norm": 0.4797074794769287, "learning_rate": 0.0005, "loss": 1.2418451309204102, "step": 29220 }, { "epoch": 2.303586913677572, "grad_norm": 0.45833805203437805, "learning_rate": 0.0005, "loss": 1.0813541412353516, "step": 29221 }, { "epoch": 2.303665746945211, "grad_norm": 0.5311841368675232, "learning_rate": 0.0005, "loss": 1.343146562576294, "step": 29222 }, { "epoch": 2.30374458021285, "grad_norm": 0.6335597038269043, "learning_rate": 0.0005, "loss": 1.6148159503936768, "step": 29223 }, { "epoch": 2.303823413480489, "grad_norm": 0.4880114793777466, "learning_rate": 0.0005, "loss": 1.3504787683486938, "step": 29224 }, { "epoch": 2.3039022467481276, "grad_norm": 0.47335386276245117, "learning_rate": 0.0005, "loss": 1.2143290042877197, "step": 29225 }, { "epoch": 2.3039810800157667, "grad_norm": 0.5051199197769165, "learning_rate": 0.0005, "loss": 1.3318068981170654, "step": 29226 }, { "epoch": 2.3040599132834054, "grad_norm": 0.4258058965206146, "learning_rate": 0.0005, "loss": 1.2064772844314575, "step": 29227 }, { "epoch": 2.3041387465510446, "grad_norm": 0.530337393283844, "learning_rate": 0.0005, "loss": 1.3619956970214844, "step": 29228 }, { "epoch": 2.3042175798186832, "grad_norm": 0.5927718281745911, "learning_rate": 0.0005, "loss": 1.4024806022644043, "step": 29229 }, { "epoch": 2.3042964130863224, "grad_norm": 0.5483943223953247, "learning_rate": 0.0005, "loss": 1.4856632947921753, "step": 29230 }, { "epoch": 2.3043752463539615, "grad_norm": 0.47250792384147644, "learning_rate": 0.0005, "loss": 1.5823575258255005, "step": 29231 }, { "epoch": 2.3044540796216, "grad_norm": 0.47916051745414734, "learning_rate": 0.0005, "loss": 1.2044652700424194, "step": 29232 }, { "epoch": 2.3045329128892393, "grad_norm": 0.485562801361084, "learning_rate": 0.0005, "loss": 1.194715142250061, "step": 29233 }, { "epoch": 2.304611746156878, "grad_norm": 0.5247547626495361, "learning_rate": 0.0005, "loss": 1.136519193649292, "step": 29234 }, { "epoch": 2.304690579424517, "grad_norm": 0.49408408999443054, "learning_rate": 0.0005, "loss": 1.3147666454315186, "step": 29235 }, { "epoch": 2.3047694126921563, "grad_norm": 0.5104248523712158, "learning_rate": 0.0005, "loss": 1.3901169300079346, "step": 29236 }, { "epoch": 2.304848245959795, "grad_norm": 0.44367507100105286, "learning_rate": 0.0005, "loss": 1.263371229171753, "step": 29237 }, { "epoch": 2.304927079227434, "grad_norm": 0.5055084228515625, "learning_rate": 0.0005, "loss": 1.1936532258987427, "step": 29238 }, { "epoch": 2.305005912495073, "grad_norm": 0.4555608928203583, "learning_rate": 0.0005, "loss": 1.1527628898620605, "step": 29239 }, { "epoch": 2.305084745762712, "grad_norm": 0.6082990765571594, "learning_rate": 0.0005, "loss": 1.4655916690826416, "step": 29240 }, { "epoch": 2.3051635790303506, "grad_norm": 0.6135585904121399, "learning_rate": 0.0005, "loss": 1.3654112815856934, "step": 29241 }, { "epoch": 2.3052424122979898, "grad_norm": 0.5057182312011719, "learning_rate": 0.0005, "loss": 1.5095094442367554, "step": 29242 }, { "epoch": 2.3053212455656285, "grad_norm": 0.527814507484436, "learning_rate": 0.0005, "loss": 1.4818212985992432, "step": 29243 }, { "epoch": 2.3054000788332676, "grad_norm": 0.72654128074646, "learning_rate": 0.0005, "loss": 1.3780412673950195, "step": 29244 }, { "epoch": 2.3054789121009067, "grad_norm": 0.529618501663208, "learning_rate": 0.0005, "loss": 1.2774577140808105, "step": 29245 }, { "epoch": 2.3055577453685454, "grad_norm": 0.5769365429878235, "learning_rate": 0.0005, "loss": 1.4383232593536377, "step": 29246 }, { "epoch": 2.3056365786361845, "grad_norm": 0.7275834083557129, "learning_rate": 0.0005, "loss": 1.3621485233306885, "step": 29247 }, { "epoch": 2.3057154119038232, "grad_norm": 0.5078981518745422, "learning_rate": 0.0005, "loss": 1.197710633277893, "step": 29248 }, { "epoch": 2.3057942451714624, "grad_norm": 0.5635799169540405, "learning_rate": 0.0005, "loss": 1.3754501342773438, "step": 29249 }, { "epoch": 2.3058730784391015, "grad_norm": 0.5266578197479248, "learning_rate": 0.0005, "loss": 1.5289666652679443, "step": 29250 }, { "epoch": 2.30595191170674, "grad_norm": 0.5036000609397888, "learning_rate": 0.0005, "loss": 1.6390879154205322, "step": 29251 }, { "epoch": 2.3060307449743793, "grad_norm": 0.47327736020088196, "learning_rate": 0.0005, "loss": 1.3343729972839355, "step": 29252 }, { "epoch": 2.306109578242018, "grad_norm": 0.5126079320907593, "learning_rate": 0.0005, "loss": 1.3339731693267822, "step": 29253 }, { "epoch": 2.306188411509657, "grad_norm": 0.45196178555488586, "learning_rate": 0.0005, "loss": 1.2102453708648682, "step": 29254 }, { "epoch": 2.306267244777296, "grad_norm": 0.4970180094242096, "learning_rate": 0.0005, "loss": 1.3213419914245605, "step": 29255 }, { "epoch": 2.306346078044935, "grad_norm": 0.6090889573097229, "learning_rate": 0.0005, "loss": 1.6361393928527832, "step": 29256 }, { "epoch": 2.306424911312574, "grad_norm": 0.5417943596839905, "learning_rate": 0.0005, "loss": 1.359641671180725, "step": 29257 }, { "epoch": 2.306503744580213, "grad_norm": 0.47679829597473145, "learning_rate": 0.0005, "loss": 1.2865092754364014, "step": 29258 }, { "epoch": 2.306582577847852, "grad_norm": 0.46185949444770813, "learning_rate": 0.0005, "loss": 1.2382179498672485, "step": 29259 }, { "epoch": 2.3066614111154906, "grad_norm": 0.4569644629955292, "learning_rate": 0.0005, "loss": 1.3191072940826416, "step": 29260 }, { "epoch": 2.3067402443831297, "grad_norm": 0.4990721344947815, "learning_rate": 0.0005, "loss": 1.155250072479248, "step": 29261 }, { "epoch": 2.3068190776507684, "grad_norm": 0.5305508971214294, "learning_rate": 0.0005, "loss": 1.4419852495193481, "step": 29262 }, { "epoch": 2.3068979109184076, "grad_norm": 0.47220373153686523, "learning_rate": 0.0005, "loss": 1.2702090740203857, "step": 29263 }, { "epoch": 2.3069767441860467, "grad_norm": 0.48881179094314575, "learning_rate": 0.0005, "loss": 1.2299281358718872, "step": 29264 }, { "epoch": 2.3070555774536854, "grad_norm": 0.5786565542221069, "learning_rate": 0.0005, "loss": 1.3700565099716187, "step": 29265 }, { "epoch": 2.3071344107213245, "grad_norm": 0.44400250911712646, "learning_rate": 0.0005, "loss": 1.3453540802001953, "step": 29266 }, { "epoch": 2.307213243988963, "grad_norm": 0.5334617495536804, "learning_rate": 0.0005, "loss": 1.3722419738769531, "step": 29267 }, { "epoch": 2.3072920772566023, "grad_norm": 0.4611859619617462, "learning_rate": 0.0005, "loss": 1.429640293121338, "step": 29268 }, { "epoch": 2.307370910524241, "grad_norm": 0.49211108684539795, "learning_rate": 0.0005, "loss": 1.4080522060394287, "step": 29269 }, { "epoch": 2.30744974379188, "grad_norm": 0.6585845351219177, "learning_rate": 0.0005, "loss": 1.561617374420166, "step": 29270 }, { "epoch": 2.3075285770595193, "grad_norm": 0.4810577630996704, "learning_rate": 0.0005, "loss": 1.271639347076416, "step": 29271 }, { "epoch": 2.307607410327158, "grad_norm": 0.5982868671417236, "learning_rate": 0.0005, "loss": 1.4505345821380615, "step": 29272 }, { "epoch": 2.307686243594797, "grad_norm": 0.4303814172744751, "learning_rate": 0.0005, "loss": 1.3891716003417969, "step": 29273 }, { "epoch": 2.307765076862436, "grad_norm": 0.5097802877426147, "learning_rate": 0.0005, "loss": 1.3990273475646973, "step": 29274 }, { "epoch": 2.307843910130075, "grad_norm": 0.5547155737876892, "learning_rate": 0.0005, "loss": 1.2138912677764893, "step": 29275 }, { "epoch": 2.3079227433977136, "grad_norm": 0.42373818159103394, "learning_rate": 0.0005, "loss": 1.2123386859893799, "step": 29276 }, { "epoch": 2.3080015766653528, "grad_norm": 0.969386637210846, "learning_rate": 0.0005, "loss": 1.385505199432373, "step": 29277 }, { "epoch": 2.308080409932992, "grad_norm": 0.5216411352157593, "learning_rate": 0.0005, "loss": 1.3447513580322266, "step": 29278 }, { "epoch": 2.3081592432006306, "grad_norm": 0.5614843964576721, "learning_rate": 0.0005, "loss": 1.4004393815994263, "step": 29279 }, { "epoch": 2.3082380764682697, "grad_norm": 0.5636408925056458, "learning_rate": 0.0005, "loss": 1.2199732065200806, "step": 29280 }, { "epoch": 2.3083169097359084, "grad_norm": 0.4802688658237457, "learning_rate": 0.0005, "loss": 1.4491246938705444, "step": 29281 }, { "epoch": 2.3083957430035476, "grad_norm": 0.48926830291748047, "learning_rate": 0.0005, "loss": 1.117281436920166, "step": 29282 }, { "epoch": 2.3084745762711867, "grad_norm": 0.48904091119766235, "learning_rate": 0.0005, "loss": 1.368489384651184, "step": 29283 }, { "epoch": 2.3085534095388254, "grad_norm": 0.5093595385551453, "learning_rate": 0.0005, "loss": 1.2455923557281494, "step": 29284 }, { "epoch": 2.3086322428064645, "grad_norm": 0.5304158926010132, "learning_rate": 0.0005, "loss": 1.3228787183761597, "step": 29285 }, { "epoch": 2.308711076074103, "grad_norm": 0.5473637580871582, "learning_rate": 0.0005, "loss": 1.2714104652404785, "step": 29286 }, { "epoch": 2.3087899093417423, "grad_norm": 0.4758114516735077, "learning_rate": 0.0005, "loss": 1.2836203575134277, "step": 29287 }, { "epoch": 2.308868742609381, "grad_norm": 0.6460287570953369, "learning_rate": 0.0005, "loss": 1.4842767715454102, "step": 29288 }, { "epoch": 2.30894757587702, "grad_norm": 0.45813828706741333, "learning_rate": 0.0005, "loss": 1.381105661392212, "step": 29289 }, { "epoch": 2.309026409144659, "grad_norm": 0.6621305346488953, "learning_rate": 0.0005, "loss": 1.2609257698059082, "step": 29290 }, { "epoch": 2.309105242412298, "grad_norm": 0.4310280382633209, "learning_rate": 0.0005, "loss": 1.1593906879425049, "step": 29291 }, { "epoch": 2.309184075679937, "grad_norm": 0.49177202582359314, "learning_rate": 0.0005, "loss": 1.4545913934707642, "step": 29292 }, { "epoch": 2.309262908947576, "grad_norm": 0.5085462331771851, "learning_rate": 0.0005, "loss": 1.457416296005249, "step": 29293 }, { "epoch": 2.309341742215215, "grad_norm": 0.5636540055274963, "learning_rate": 0.0005, "loss": 1.387047290802002, "step": 29294 }, { "epoch": 2.3094205754828536, "grad_norm": 0.41606467962265015, "learning_rate": 0.0005, "loss": 1.2947943210601807, "step": 29295 }, { "epoch": 2.3094994087504928, "grad_norm": 0.5453097820281982, "learning_rate": 0.0005, "loss": 1.1966863870620728, "step": 29296 }, { "epoch": 2.309578242018132, "grad_norm": 0.5261029601097107, "learning_rate": 0.0005, "loss": 1.3922269344329834, "step": 29297 }, { "epoch": 2.3096570752857706, "grad_norm": 0.6389018297195435, "learning_rate": 0.0005, "loss": 1.6629984378814697, "step": 29298 }, { "epoch": 2.3097359085534097, "grad_norm": 0.4604496657848358, "learning_rate": 0.0005, "loss": 1.4825196266174316, "step": 29299 }, { "epoch": 2.3098147418210484, "grad_norm": 0.4928005337715149, "learning_rate": 0.0005, "loss": 1.4175596237182617, "step": 29300 }, { "epoch": 2.3098935750886875, "grad_norm": 0.6148955225944519, "learning_rate": 0.0005, "loss": 1.4394460916519165, "step": 29301 }, { "epoch": 2.3099724083563262, "grad_norm": 0.5321065187454224, "learning_rate": 0.0005, "loss": 1.4783380031585693, "step": 29302 }, { "epoch": 2.3100512416239654, "grad_norm": 0.4112943112850189, "learning_rate": 0.0005, "loss": 1.2180843353271484, "step": 29303 }, { "epoch": 2.310130074891604, "grad_norm": 0.4594585597515106, "learning_rate": 0.0005, "loss": 1.5024564266204834, "step": 29304 }, { "epoch": 2.310208908159243, "grad_norm": 0.566455602645874, "learning_rate": 0.0005, "loss": 1.2333452701568604, "step": 29305 }, { "epoch": 2.3102877414268823, "grad_norm": 0.5731611847877502, "learning_rate": 0.0005, "loss": 1.4040769338607788, "step": 29306 }, { "epoch": 2.310366574694521, "grad_norm": 0.5145308375358582, "learning_rate": 0.0005, "loss": 1.4227137565612793, "step": 29307 }, { "epoch": 2.31044540796216, "grad_norm": 0.5011743903160095, "learning_rate": 0.0005, "loss": 1.3241956233978271, "step": 29308 }, { "epoch": 2.310524241229799, "grad_norm": 0.5512197017669678, "learning_rate": 0.0005, "loss": 1.4055449962615967, "step": 29309 }, { "epoch": 2.310603074497438, "grad_norm": 0.5732846856117249, "learning_rate": 0.0005, "loss": 1.4948325157165527, "step": 29310 }, { "epoch": 2.310681907765077, "grad_norm": 0.47321459650993347, "learning_rate": 0.0005, "loss": 1.4014294147491455, "step": 29311 }, { "epoch": 2.310760741032716, "grad_norm": 0.47695791721343994, "learning_rate": 0.0005, "loss": 1.3300483226776123, "step": 29312 }, { "epoch": 2.310839574300355, "grad_norm": 0.5175589323043823, "learning_rate": 0.0005, "loss": 1.3150475025177002, "step": 29313 }, { "epoch": 2.3109184075679936, "grad_norm": 0.5884442925453186, "learning_rate": 0.0005, "loss": 1.1746406555175781, "step": 29314 }, { "epoch": 2.3109972408356327, "grad_norm": 0.5196976065635681, "learning_rate": 0.0005, "loss": 1.5394744873046875, "step": 29315 }, { "epoch": 2.3110760741032714, "grad_norm": 0.533464789390564, "learning_rate": 0.0005, "loss": 1.3480439186096191, "step": 29316 }, { "epoch": 2.3111549073709106, "grad_norm": 0.5606815218925476, "learning_rate": 0.0005, "loss": 1.1836590766906738, "step": 29317 }, { "epoch": 2.3112337406385492, "grad_norm": 0.5214725732803345, "learning_rate": 0.0005, "loss": 1.3399767875671387, "step": 29318 }, { "epoch": 2.3113125739061884, "grad_norm": 0.5765326619148254, "learning_rate": 0.0005, "loss": 1.4210264682769775, "step": 29319 }, { "epoch": 2.3113914071738275, "grad_norm": 0.4829457700252533, "learning_rate": 0.0005, "loss": 1.3321828842163086, "step": 29320 }, { "epoch": 2.311470240441466, "grad_norm": 0.6846831440925598, "learning_rate": 0.0005, "loss": 1.3700226545333862, "step": 29321 }, { "epoch": 2.3115490737091053, "grad_norm": 0.6258596777915955, "learning_rate": 0.0005, "loss": 1.4610610008239746, "step": 29322 }, { "epoch": 2.311627906976744, "grad_norm": 0.5177095532417297, "learning_rate": 0.0005, "loss": 1.4066238403320312, "step": 29323 }, { "epoch": 2.311706740244383, "grad_norm": 0.47185632586479187, "learning_rate": 0.0005, "loss": 1.2420907020568848, "step": 29324 }, { "epoch": 2.3117855735120223, "grad_norm": 0.5727642774581909, "learning_rate": 0.0005, "loss": 1.4432458877563477, "step": 29325 }, { "epoch": 2.311864406779661, "grad_norm": 0.557120144367218, "learning_rate": 0.0005, "loss": 1.34927237033844, "step": 29326 }, { "epoch": 2.3119432400473, "grad_norm": 0.48233336210250854, "learning_rate": 0.0005, "loss": 1.4898688793182373, "step": 29327 }, { "epoch": 2.312022073314939, "grad_norm": 0.4617082476615906, "learning_rate": 0.0005, "loss": 1.3051905632019043, "step": 29328 }, { "epoch": 2.312100906582578, "grad_norm": 0.6206318736076355, "learning_rate": 0.0005, "loss": 1.0674114227294922, "step": 29329 }, { "epoch": 2.3121797398502166, "grad_norm": 0.5334299206733704, "learning_rate": 0.0005, "loss": 1.5808565616607666, "step": 29330 }, { "epoch": 2.3122585731178558, "grad_norm": 0.4970583915710449, "learning_rate": 0.0005, "loss": 1.3156464099884033, "step": 29331 }, { "epoch": 2.3123374063854945, "grad_norm": 0.6994494795799255, "learning_rate": 0.0005, "loss": 1.4014091491699219, "step": 29332 }, { "epoch": 2.3124162396531336, "grad_norm": 0.4967040717601776, "learning_rate": 0.0005, "loss": 1.4014718532562256, "step": 29333 }, { "epoch": 2.3124950729207727, "grad_norm": 0.477893203496933, "learning_rate": 0.0005, "loss": 1.344271183013916, "step": 29334 }, { "epoch": 2.3125739061884114, "grad_norm": 0.4878353178501129, "learning_rate": 0.0005, "loss": 1.1750847101211548, "step": 29335 }, { "epoch": 2.3126527394560505, "grad_norm": 0.6586218476295471, "learning_rate": 0.0005, "loss": 1.4561152458190918, "step": 29336 }, { "epoch": 2.3127315727236892, "grad_norm": 0.528066098690033, "learning_rate": 0.0005, "loss": 1.186248779296875, "step": 29337 }, { "epoch": 2.3128104059913284, "grad_norm": 0.4572283923625946, "learning_rate": 0.0005, "loss": 1.3750603199005127, "step": 29338 }, { "epoch": 2.3128892392589675, "grad_norm": 0.4865357577800751, "learning_rate": 0.0005, "loss": 1.2571625709533691, "step": 29339 }, { "epoch": 2.312968072526606, "grad_norm": 0.5226272940635681, "learning_rate": 0.0005, "loss": 1.3271739482879639, "step": 29340 }, { "epoch": 2.3130469057942453, "grad_norm": 0.539326012134552, "learning_rate": 0.0005, "loss": 1.4485995769500732, "step": 29341 }, { "epoch": 2.313125739061884, "grad_norm": 0.6106353998184204, "learning_rate": 0.0005, "loss": 1.572745680809021, "step": 29342 }, { "epoch": 2.313204572329523, "grad_norm": 0.5648543238639832, "learning_rate": 0.0005, "loss": 1.3165265321731567, "step": 29343 }, { "epoch": 2.313283405597162, "grad_norm": 0.5719794034957886, "learning_rate": 0.0005, "loss": 1.4094219207763672, "step": 29344 }, { "epoch": 2.313362238864801, "grad_norm": 0.5435794591903687, "learning_rate": 0.0005, "loss": 1.4035595655441284, "step": 29345 }, { "epoch": 2.3134410721324397, "grad_norm": 0.5248809456825256, "learning_rate": 0.0005, "loss": 1.3496081829071045, "step": 29346 }, { "epoch": 2.313519905400079, "grad_norm": 0.46600568294525146, "learning_rate": 0.0005, "loss": 1.2490181922912598, "step": 29347 }, { "epoch": 2.313598738667718, "grad_norm": 0.46279609203338623, "learning_rate": 0.0005, "loss": 1.217458963394165, "step": 29348 }, { "epoch": 2.3136775719353566, "grad_norm": 0.5105347037315369, "learning_rate": 0.0005, "loss": 1.3509926795959473, "step": 29349 }, { "epoch": 2.3137564052029957, "grad_norm": 0.46831172704696655, "learning_rate": 0.0005, "loss": 1.5153769254684448, "step": 29350 }, { "epoch": 2.3138352384706344, "grad_norm": 0.5317655801773071, "learning_rate": 0.0005, "loss": 1.4279588460922241, "step": 29351 }, { "epoch": 2.3139140717382736, "grad_norm": 0.5667725801467896, "learning_rate": 0.0005, "loss": 1.2132776975631714, "step": 29352 }, { "epoch": 2.3139929050059127, "grad_norm": 0.49366307258605957, "learning_rate": 0.0005, "loss": 1.184870719909668, "step": 29353 }, { "epoch": 2.3140717382735514, "grad_norm": 0.5248345732688904, "learning_rate": 0.0005, "loss": 1.4655112028121948, "step": 29354 }, { "epoch": 2.3141505715411905, "grad_norm": 0.5486785769462585, "learning_rate": 0.0005, "loss": 1.1946266889572144, "step": 29355 }, { "epoch": 2.314229404808829, "grad_norm": 0.5757296085357666, "learning_rate": 0.0005, "loss": 1.4222004413604736, "step": 29356 }, { "epoch": 2.3143082380764683, "grad_norm": 0.4909772276878357, "learning_rate": 0.0005, "loss": 1.4278239011764526, "step": 29357 }, { "epoch": 2.314387071344107, "grad_norm": 0.5433705449104309, "learning_rate": 0.0005, "loss": 1.585273265838623, "step": 29358 }, { "epoch": 2.314465904611746, "grad_norm": 0.703474760055542, "learning_rate": 0.0005, "loss": 1.4177988767623901, "step": 29359 }, { "epoch": 2.314544737879385, "grad_norm": 0.48760029673576355, "learning_rate": 0.0005, "loss": 1.1813693046569824, "step": 29360 }, { "epoch": 2.314623571147024, "grad_norm": 0.4424382448196411, "learning_rate": 0.0005, "loss": 1.18929922580719, "step": 29361 }, { "epoch": 2.314702404414663, "grad_norm": 0.4978351294994354, "learning_rate": 0.0005, "loss": 1.2208876609802246, "step": 29362 }, { "epoch": 2.314781237682302, "grad_norm": 0.5097944140434265, "learning_rate": 0.0005, "loss": 1.2852425575256348, "step": 29363 }, { "epoch": 2.314860070949941, "grad_norm": 0.4255083501338959, "learning_rate": 0.0005, "loss": 1.2482025623321533, "step": 29364 }, { "epoch": 2.3149389042175796, "grad_norm": 0.3929468095302582, "learning_rate": 0.0005, "loss": 1.2121460437774658, "step": 29365 }, { "epoch": 2.3150177374852188, "grad_norm": 0.4808891713619232, "learning_rate": 0.0005, "loss": 1.3074440956115723, "step": 29366 }, { "epoch": 2.315096570752858, "grad_norm": 0.4980674386024475, "learning_rate": 0.0005, "loss": 1.7004843950271606, "step": 29367 }, { "epoch": 2.3151754040204966, "grad_norm": 0.5327815413475037, "learning_rate": 0.0005, "loss": 1.4893059730529785, "step": 29368 }, { "epoch": 2.3152542372881357, "grad_norm": 0.5634644031524658, "learning_rate": 0.0005, "loss": 1.1337480545043945, "step": 29369 }, { "epoch": 2.3153330705557744, "grad_norm": 0.5233767628669739, "learning_rate": 0.0005, "loss": 1.4317067861557007, "step": 29370 }, { "epoch": 2.3154119038234136, "grad_norm": 0.5492593050003052, "learning_rate": 0.0005, "loss": 1.4135046005249023, "step": 29371 }, { "epoch": 2.3154907370910522, "grad_norm": 0.43932658433914185, "learning_rate": 0.0005, "loss": 1.353095293045044, "step": 29372 }, { "epoch": 2.3155695703586914, "grad_norm": 0.597683846950531, "learning_rate": 0.0005, "loss": 1.4291610717773438, "step": 29373 }, { "epoch": 2.31564840362633, "grad_norm": 0.5133095979690552, "learning_rate": 0.0005, "loss": 1.3368477821350098, "step": 29374 }, { "epoch": 2.315727236893969, "grad_norm": 0.718845009803772, "learning_rate": 0.0005, "loss": 1.4371423721313477, "step": 29375 }, { "epoch": 2.3158060701616083, "grad_norm": 0.49449172616004944, "learning_rate": 0.0005, "loss": 1.4159291982650757, "step": 29376 }, { "epoch": 2.315884903429247, "grad_norm": 0.47104376554489136, "learning_rate": 0.0005, "loss": 1.3318008184432983, "step": 29377 }, { "epoch": 2.315963736696886, "grad_norm": 0.4155963063240051, "learning_rate": 0.0005, "loss": 1.2611567974090576, "step": 29378 }, { "epoch": 2.316042569964525, "grad_norm": 0.4841531217098236, "learning_rate": 0.0005, "loss": 1.2807697057724, "step": 29379 }, { "epoch": 2.316121403232164, "grad_norm": 0.46969637274742126, "learning_rate": 0.0005, "loss": 1.3341937065124512, "step": 29380 }, { "epoch": 2.316200236499803, "grad_norm": 0.5032976865768433, "learning_rate": 0.0005, "loss": 1.5138452053070068, "step": 29381 }, { "epoch": 2.316279069767442, "grad_norm": 0.4606962203979492, "learning_rate": 0.0005, "loss": 1.4774315357208252, "step": 29382 }, { "epoch": 2.316357903035081, "grad_norm": 0.45712921023368835, "learning_rate": 0.0005, "loss": 1.160987377166748, "step": 29383 }, { "epoch": 2.3164367363027196, "grad_norm": 0.5583056807518005, "learning_rate": 0.0005, "loss": 1.2385530471801758, "step": 29384 }, { "epoch": 2.3165155695703588, "grad_norm": 0.4840314984321594, "learning_rate": 0.0005, "loss": 1.0959550142288208, "step": 29385 }, { "epoch": 2.3165944028379974, "grad_norm": 0.709324300289154, "learning_rate": 0.0005, "loss": 1.4345052242279053, "step": 29386 }, { "epoch": 2.3166732361056366, "grad_norm": 0.5863023400306702, "learning_rate": 0.0005, "loss": 1.2953639030456543, "step": 29387 }, { "epoch": 2.3167520693732757, "grad_norm": 0.5978643298149109, "learning_rate": 0.0005, "loss": 1.4138946533203125, "step": 29388 }, { "epoch": 2.3168309026409144, "grad_norm": 0.7696920037269592, "learning_rate": 0.0005, "loss": 1.5081448554992676, "step": 29389 }, { "epoch": 2.3169097359085535, "grad_norm": 0.5356228351593018, "learning_rate": 0.0005, "loss": 1.3309710025787354, "step": 29390 }, { "epoch": 2.3169885691761922, "grad_norm": 0.5804892778396606, "learning_rate": 0.0005, "loss": 1.5165128707885742, "step": 29391 }, { "epoch": 2.3170674024438314, "grad_norm": 0.6539053916931152, "learning_rate": 0.0005, "loss": 1.308242917060852, "step": 29392 }, { "epoch": 2.31714623571147, "grad_norm": 0.7319834232330322, "learning_rate": 0.0005, "loss": 1.5304696559906006, "step": 29393 }, { "epoch": 2.317225068979109, "grad_norm": 0.47691068053245544, "learning_rate": 0.0005, "loss": 1.2719008922576904, "step": 29394 }, { "epoch": 2.3173039022467483, "grad_norm": 0.5802240371704102, "learning_rate": 0.0005, "loss": 1.2408719062805176, "step": 29395 }, { "epoch": 2.317382735514387, "grad_norm": 0.6031660437583923, "learning_rate": 0.0005, "loss": 1.4960360527038574, "step": 29396 }, { "epoch": 2.317461568782026, "grad_norm": 0.5860222578048706, "learning_rate": 0.0005, "loss": 1.4849703311920166, "step": 29397 }, { "epoch": 2.317540402049665, "grad_norm": 0.49042248725891113, "learning_rate": 0.0005, "loss": 1.5088756084442139, "step": 29398 }, { "epoch": 2.317619235317304, "grad_norm": 0.5506858229637146, "learning_rate": 0.0005, "loss": 1.419100046157837, "step": 29399 }, { "epoch": 2.317698068584943, "grad_norm": 0.5147431492805481, "learning_rate": 0.0005, "loss": 1.2734429836273193, "step": 29400 }, { "epoch": 2.317776901852582, "grad_norm": 0.5308499932289124, "learning_rate": 0.0005, "loss": 1.350034236907959, "step": 29401 }, { "epoch": 2.317855735120221, "grad_norm": 0.42171356081962585, "learning_rate": 0.0005, "loss": 1.3195350170135498, "step": 29402 }, { "epoch": 2.3179345683878596, "grad_norm": 0.5032863020896912, "learning_rate": 0.0005, "loss": 1.495275855064392, "step": 29403 }, { "epoch": 2.3180134016554987, "grad_norm": 0.46898582577705383, "learning_rate": 0.0005, "loss": 1.2248324155807495, "step": 29404 }, { "epoch": 2.3180922349231374, "grad_norm": 0.43124809861183167, "learning_rate": 0.0005, "loss": 1.2360514402389526, "step": 29405 }, { "epoch": 2.3181710681907766, "grad_norm": 0.8553649187088013, "learning_rate": 0.0005, "loss": 1.1979999542236328, "step": 29406 }, { "epoch": 2.3182499014584153, "grad_norm": 0.4758189916610718, "learning_rate": 0.0005, "loss": 1.391028642654419, "step": 29407 }, { "epoch": 2.3183287347260544, "grad_norm": 0.44416508078575134, "learning_rate": 0.0005, "loss": 1.284488320350647, "step": 29408 }, { "epoch": 2.3184075679936935, "grad_norm": 0.5913455486297607, "learning_rate": 0.0005, "loss": 1.4696900844573975, "step": 29409 }, { "epoch": 2.318486401261332, "grad_norm": 0.48340556025505066, "learning_rate": 0.0005, "loss": 1.3019163608551025, "step": 29410 }, { "epoch": 2.3185652345289713, "grad_norm": 0.5932342410087585, "learning_rate": 0.0005, "loss": 1.1060864925384521, "step": 29411 }, { "epoch": 2.31864406779661, "grad_norm": 0.4832332730293274, "learning_rate": 0.0005, "loss": 1.4704707860946655, "step": 29412 }, { "epoch": 2.318722901064249, "grad_norm": 0.4389096200466156, "learning_rate": 0.0005, "loss": 1.2371547222137451, "step": 29413 }, { "epoch": 2.3188017343318883, "grad_norm": 0.4742664694786072, "learning_rate": 0.0005, "loss": 1.237579107284546, "step": 29414 }, { "epoch": 2.318880567599527, "grad_norm": 0.4914182424545288, "learning_rate": 0.0005, "loss": 1.3627395629882812, "step": 29415 }, { "epoch": 2.318959400867166, "grad_norm": 0.959120512008667, "learning_rate": 0.0005, "loss": 1.3649073839187622, "step": 29416 }, { "epoch": 2.319038234134805, "grad_norm": 0.49548742175102234, "learning_rate": 0.0005, "loss": 1.2296528816223145, "step": 29417 }, { "epoch": 2.319117067402444, "grad_norm": 0.44737520813941956, "learning_rate": 0.0005, "loss": 1.3736364841461182, "step": 29418 }, { "epoch": 2.3191959006700826, "grad_norm": 0.46013355255126953, "learning_rate": 0.0005, "loss": 1.4600086212158203, "step": 29419 }, { "epoch": 2.3192747339377218, "grad_norm": 0.5619417428970337, "learning_rate": 0.0005, "loss": 1.2725224494934082, "step": 29420 }, { "epoch": 2.3193535672053605, "grad_norm": 0.5207204222679138, "learning_rate": 0.0005, "loss": 1.4031755924224854, "step": 29421 }, { "epoch": 2.3194324004729996, "grad_norm": 0.4613555073738098, "learning_rate": 0.0005, "loss": 1.434736728668213, "step": 29422 }, { "epoch": 2.3195112337406387, "grad_norm": 0.46432313323020935, "learning_rate": 0.0005, "loss": 1.3229477405548096, "step": 29423 }, { "epoch": 2.3195900670082774, "grad_norm": 0.4856320321559906, "learning_rate": 0.0005, "loss": 1.327134609222412, "step": 29424 }, { "epoch": 2.3196689002759165, "grad_norm": 0.5170731544494629, "learning_rate": 0.0005, "loss": 1.2863454818725586, "step": 29425 }, { "epoch": 2.3197477335435552, "grad_norm": 0.5125287175178528, "learning_rate": 0.0005, "loss": 1.5044589042663574, "step": 29426 }, { "epoch": 2.3198265668111944, "grad_norm": 0.40533939003944397, "learning_rate": 0.0005, "loss": 1.1343984603881836, "step": 29427 }, { "epoch": 2.3199054000788335, "grad_norm": 0.4619789123535156, "learning_rate": 0.0005, "loss": 1.2446038722991943, "step": 29428 }, { "epoch": 2.319984233346472, "grad_norm": 0.5044095516204834, "learning_rate": 0.0005, "loss": 1.1787868738174438, "step": 29429 }, { "epoch": 2.3200630666141113, "grad_norm": 0.591339111328125, "learning_rate": 0.0005, "loss": 1.5201034545898438, "step": 29430 }, { "epoch": 2.32014189988175, "grad_norm": 0.4980287253856659, "learning_rate": 0.0005, "loss": 1.269036889076233, "step": 29431 }, { "epoch": 2.320220733149389, "grad_norm": 0.49115175008773804, "learning_rate": 0.0005, "loss": 1.3570688962936401, "step": 29432 }, { "epoch": 2.320299566417028, "grad_norm": 0.5401461124420166, "learning_rate": 0.0005, "loss": 1.363229751586914, "step": 29433 }, { "epoch": 2.320378399684667, "grad_norm": 0.5082563161849976, "learning_rate": 0.0005, "loss": 1.6940653324127197, "step": 29434 }, { "epoch": 2.3204572329523057, "grad_norm": 0.5653221011161804, "learning_rate": 0.0005, "loss": 1.4386627674102783, "step": 29435 }, { "epoch": 2.320536066219945, "grad_norm": 0.5165693759918213, "learning_rate": 0.0005, "loss": 1.4035542011260986, "step": 29436 }, { "epoch": 2.320614899487584, "grad_norm": 0.5892328023910522, "learning_rate": 0.0005, "loss": 1.24039888381958, "step": 29437 }, { "epoch": 2.3206937327552226, "grad_norm": 0.4091211259365082, "learning_rate": 0.0005, "loss": 1.3659534454345703, "step": 29438 }, { "epoch": 2.3207725660228617, "grad_norm": 0.5287430286407471, "learning_rate": 0.0005, "loss": 1.4724431037902832, "step": 29439 }, { "epoch": 2.3208513992905004, "grad_norm": 1.7793617248535156, "learning_rate": 0.0005, "loss": 1.454497218132019, "step": 29440 }, { "epoch": 2.3209302325581396, "grad_norm": 0.5564354658126831, "learning_rate": 0.0005, "loss": 1.3057854175567627, "step": 29441 }, { "epoch": 2.3210090658257787, "grad_norm": 0.6150012612342834, "learning_rate": 0.0005, "loss": 1.483602523803711, "step": 29442 }, { "epoch": 2.3210878990934174, "grad_norm": 0.4627707898616791, "learning_rate": 0.0005, "loss": 1.4183753728866577, "step": 29443 }, { "epoch": 2.3211667323610565, "grad_norm": 0.49284249544143677, "learning_rate": 0.0005, "loss": 1.2613956928253174, "step": 29444 }, { "epoch": 2.321245565628695, "grad_norm": 0.533073365688324, "learning_rate": 0.0005, "loss": 1.3445930480957031, "step": 29445 }, { "epoch": 2.3213243988963344, "grad_norm": 0.5009053945541382, "learning_rate": 0.0005, "loss": 1.338615894317627, "step": 29446 }, { "epoch": 2.321403232163973, "grad_norm": 0.47287485003471375, "learning_rate": 0.0005, "loss": 1.1865345239639282, "step": 29447 }, { "epoch": 2.321482065431612, "grad_norm": 0.4810163378715515, "learning_rate": 0.0005, "loss": 1.3042415380477905, "step": 29448 }, { "epoch": 2.321560898699251, "grad_norm": 0.5499052405357361, "learning_rate": 0.0005, "loss": 1.464986801147461, "step": 29449 }, { "epoch": 2.32163973196689, "grad_norm": 0.5157127380371094, "learning_rate": 0.0005, "loss": 1.441908597946167, "step": 29450 }, { "epoch": 2.321718565234529, "grad_norm": 0.6903409957885742, "learning_rate": 0.0005, "loss": 1.4385440349578857, "step": 29451 }, { "epoch": 2.321797398502168, "grad_norm": 0.5969696044921875, "learning_rate": 0.0005, "loss": 1.251081943511963, "step": 29452 }, { "epoch": 2.321876231769807, "grad_norm": 0.4454423487186432, "learning_rate": 0.0005, "loss": 1.348360538482666, "step": 29453 }, { "epoch": 2.3219550650374456, "grad_norm": 0.5005387663841248, "learning_rate": 0.0005, "loss": 1.2793740034103394, "step": 29454 }, { "epoch": 2.3220338983050848, "grad_norm": 0.5251312255859375, "learning_rate": 0.0005, "loss": 1.4764258861541748, "step": 29455 }, { "epoch": 2.322112731572724, "grad_norm": 0.43240347504615784, "learning_rate": 0.0005, "loss": 1.2316415309906006, "step": 29456 }, { "epoch": 2.3221915648403626, "grad_norm": 0.5562621355056763, "learning_rate": 0.0005, "loss": 1.3149465322494507, "step": 29457 }, { "epoch": 2.3222703981080017, "grad_norm": 0.47268936038017273, "learning_rate": 0.0005, "loss": 1.2519135475158691, "step": 29458 }, { "epoch": 2.3223492313756404, "grad_norm": 0.5142735242843628, "learning_rate": 0.0005, "loss": 1.3159983158111572, "step": 29459 }, { "epoch": 2.3224280646432796, "grad_norm": 0.5716788172721863, "learning_rate": 0.0005, "loss": 1.318398356437683, "step": 29460 }, { "epoch": 2.3225068979109182, "grad_norm": 0.711582601070404, "learning_rate": 0.0005, "loss": 1.2799091339111328, "step": 29461 }, { "epoch": 2.3225857311785574, "grad_norm": 0.6098706126213074, "learning_rate": 0.0005, "loss": 1.4342117309570312, "step": 29462 }, { "epoch": 2.322664564446196, "grad_norm": 0.6735028624534607, "learning_rate": 0.0005, "loss": 1.6001594066619873, "step": 29463 }, { "epoch": 2.322743397713835, "grad_norm": 0.5481131672859192, "learning_rate": 0.0005, "loss": 1.3092527389526367, "step": 29464 }, { "epoch": 2.3228222309814743, "grad_norm": 0.4864257574081421, "learning_rate": 0.0005, "loss": 1.4920530319213867, "step": 29465 }, { "epoch": 2.322901064249113, "grad_norm": 0.4424471855163574, "learning_rate": 0.0005, "loss": 1.1533949375152588, "step": 29466 }, { "epoch": 2.322979897516752, "grad_norm": 0.42737916111946106, "learning_rate": 0.0005, "loss": 1.14589262008667, "step": 29467 }, { "epoch": 2.323058730784391, "grad_norm": 0.5038342475891113, "learning_rate": 0.0005, "loss": 1.2094813585281372, "step": 29468 }, { "epoch": 2.32313756405203, "grad_norm": 0.5973790287971497, "learning_rate": 0.0005, "loss": 1.1379725933074951, "step": 29469 }, { "epoch": 2.323216397319669, "grad_norm": 0.4899692237377167, "learning_rate": 0.0005, "loss": 1.4472095966339111, "step": 29470 }, { "epoch": 2.323295230587308, "grad_norm": 0.38620439171791077, "learning_rate": 0.0005, "loss": 1.1889532804489136, "step": 29471 }, { "epoch": 2.323374063854947, "grad_norm": 0.4710465967655182, "learning_rate": 0.0005, "loss": 1.3432602882385254, "step": 29472 }, { "epoch": 2.3234528971225856, "grad_norm": 0.4507255256175995, "learning_rate": 0.0005, "loss": 1.2151315212249756, "step": 29473 }, { "epoch": 2.3235317303902248, "grad_norm": 0.49103161692619324, "learning_rate": 0.0005, "loss": 1.2720043659210205, "step": 29474 }, { "epoch": 2.3236105636578634, "grad_norm": 0.5042454600334167, "learning_rate": 0.0005, "loss": 1.3187925815582275, "step": 29475 }, { "epoch": 2.3236893969255026, "grad_norm": 0.5454970598220825, "learning_rate": 0.0005, "loss": 1.3911055326461792, "step": 29476 }, { "epoch": 2.3237682301931413, "grad_norm": 0.5032107830047607, "learning_rate": 0.0005, "loss": 1.2289646863937378, "step": 29477 }, { "epoch": 2.3238470634607804, "grad_norm": 0.4276891052722931, "learning_rate": 0.0005, "loss": 1.480756402015686, "step": 29478 }, { "epoch": 2.3239258967284195, "grad_norm": 0.4698525369167328, "learning_rate": 0.0005, "loss": 1.2929896116256714, "step": 29479 }, { "epoch": 2.3240047299960582, "grad_norm": 0.5599037408828735, "learning_rate": 0.0005, "loss": 1.3530747890472412, "step": 29480 }, { "epoch": 2.3240835632636974, "grad_norm": 0.44546109437942505, "learning_rate": 0.0005, "loss": 1.34027099609375, "step": 29481 }, { "epoch": 2.324162396531336, "grad_norm": 0.5500847101211548, "learning_rate": 0.0005, "loss": 1.331454873085022, "step": 29482 }, { "epoch": 2.324241229798975, "grad_norm": 0.49636417627334595, "learning_rate": 0.0005, "loss": 1.1841678619384766, "step": 29483 }, { "epoch": 2.3243200630666143, "grad_norm": 0.4505471885204315, "learning_rate": 0.0005, "loss": 1.41017484664917, "step": 29484 }, { "epoch": 2.324398896334253, "grad_norm": 0.528117835521698, "learning_rate": 0.0005, "loss": 1.5917460918426514, "step": 29485 }, { "epoch": 2.324477729601892, "grad_norm": 0.4950646460056305, "learning_rate": 0.0005, "loss": 1.1113665103912354, "step": 29486 }, { "epoch": 2.324556562869531, "grad_norm": 0.5087864398956299, "learning_rate": 0.0005, "loss": 1.4584484100341797, "step": 29487 }, { "epoch": 2.32463539613717, "grad_norm": 0.433132141828537, "learning_rate": 0.0005, "loss": 1.2838045358657837, "step": 29488 }, { "epoch": 2.3247142294048087, "grad_norm": 0.4942021071910858, "learning_rate": 0.0005, "loss": 1.3688914775848389, "step": 29489 }, { "epoch": 2.324793062672448, "grad_norm": 0.49007099866867065, "learning_rate": 0.0005, "loss": 1.3722257614135742, "step": 29490 }, { "epoch": 2.3248718959400865, "grad_norm": 0.5169275403022766, "learning_rate": 0.0005, "loss": 1.2892988920211792, "step": 29491 }, { "epoch": 2.3249507292077256, "grad_norm": 0.46421918272972107, "learning_rate": 0.0005, "loss": 1.3662217855453491, "step": 29492 }, { "epoch": 2.3250295624753647, "grad_norm": 0.46989235281944275, "learning_rate": 0.0005, "loss": 1.325634241104126, "step": 29493 }, { "epoch": 2.3251083957430034, "grad_norm": 0.5203493237495422, "learning_rate": 0.0005, "loss": 1.3771061897277832, "step": 29494 }, { "epoch": 2.3251872290106426, "grad_norm": 0.4745006561279297, "learning_rate": 0.0005, "loss": 1.267183780670166, "step": 29495 }, { "epoch": 2.3252660622782813, "grad_norm": 0.5231673121452332, "learning_rate": 0.0005, "loss": 1.2334470748901367, "step": 29496 }, { "epoch": 2.3253448955459204, "grad_norm": 0.5033909678459167, "learning_rate": 0.0005, "loss": 1.3923912048339844, "step": 29497 }, { "epoch": 2.3254237288135595, "grad_norm": 0.4907428026199341, "learning_rate": 0.0005, "loss": 1.2588388919830322, "step": 29498 }, { "epoch": 2.325502562081198, "grad_norm": 0.3755699098110199, "learning_rate": 0.0005, "loss": 1.0709491968154907, "step": 29499 }, { "epoch": 2.3255813953488373, "grad_norm": 0.46097680926322937, "learning_rate": 0.0005, "loss": 1.2378088235855103, "step": 29500 }, { "epoch": 2.325660228616476, "grad_norm": 0.6249547600746155, "learning_rate": 0.0005, "loss": 1.4744670391082764, "step": 29501 }, { "epoch": 2.325739061884115, "grad_norm": 0.5333526730537415, "learning_rate": 0.0005, "loss": 1.2828097343444824, "step": 29502 }, { "epoch": 2.325817895151754, "grad_norm": 0.6920005679130554, "learning_rate": 0.0005, "loss": 1.4760043621063232, "step": 29503 }, { "epoch": 2.325896728419393, "grad_norm": 0.5524982810020447, "learning_rate": 0.0005, "loss": 1.427239179611206, "step": 29504 }, { "epoch": 2.3259755616870317, "grad_norm": 0.39460283517837524, "learning_rate": 0.0005, "loss": 1.250813603401184, "step": 29505 }, { "epoch": 2.326054394954671, "grad_norm": 0.5656500458717346, "learning_rate": 0.0005, "loss": 1.3828036785125732, "step": 29506 }, { "epoch": 2.32613322822231, "grad_norm": 0.4281432330608368, "learning_rate": 0.0005, "loss": 1.1799522638320923, "step": 29507 }, { "epoch": 2.3262120614899486, "grad_norm": 0.45090848207473755, "learning_rate": 0.0005, "loss": 1.2224321365356445, "step": 29508 }, { "epoch": 2.3262908947575878, "grad_norm": 1.0726735591888428, "learning_rate": 0.0005, "loss": 1.2735732793807983, "step": 29509 }, { "epoch": 2.3263697280252265, "grad_norm": 0.4653061628341675, "learning_rate": 0.0005, "loss": 1.3624135255813599, "step": 29510 }, { "epoch": 2.3264485612928656, "grad_norm": 0.46535927057266235, "learning_rate": 0.0005, "loss": 1.39949631690979, "step": 29511 }, { "epoch": 2.3265273945605047, "grad_norm": 0.5177687406539917, "learning_rate": 0.0005, "loss": 1.40506911277771, "step": 29512 }, { "epoch": 2.3266062278281434, "grad_norm": 0.4827820360660553, "learning_rate": 0.0005, "loss": 1.3447086811065674, "step": 29513 }, { "epoch": 2.3266850610957825, "grad_norm": 0.6944893002510071, "learning_rate": 0.0005, "loss": 1.5024960041046143, "step": 29514 }, { "epoch": 2.3267638943634212, "grad_norm": 0.4782346785068512, "learning_rate": 0.0005, "loss": 1.195715308189392, "step": 29515 }, { "epoch": 2.3268427276310604, "grad_norm": 0.45510631799697876, "learning_rate": 0.0005, "loss": 1.3581985235214233, "step": 29516 }, { "epoch": 2.326921560898699, "grad_norm": 0.4164368212223053, "learning_rate": 0.0005, "loss": 1.1755321025848389, "step": 29517 }, { "epoch": 2.327000394166338, "grad_norm": 0.5250107049942017, "learning_rate": 0.0005, "loss": 1.2422890663146973, "step": 29518 }, { "epoch": 2.3270792274339773, "grad_norm": 0.5641327500343323, "learning_rate": 0.0005, "loss": 1.5845458507537842, "step": 29519 }, { "epoch": 2.327158060701616, "grad_norm": 0.5170787572860718, "learning_rate": 0.0005, "loss": 1.2908142805099487, "step": 29520 }, { "epoch": 2.327236893969255, "grad_norm": 0.611452579498291, "learning_rate": 0.0005, "loss": 1.4897451400756836, "step": 29521 }, { "epoch": 2.327315727236894, "grad_norm": 0.49725955724716187, "learning_rate": 0.0005, "loss": 1.3069555759429932, "step": 29522 }, { "epoch": 2.327394560504533, "grad_norm": 0.5873163938522339, "learning_rate": 0.0005, "loss": 1.3331389427185059, "step": 29523 }, { "epoch": 2.3274733937721717, "grad_norm": 0.6440572738647461, "learning_rate": 0.0005, "loss": 1.5571058988571167, "step": 29524 }, { "epoch": 2.327552227039811, "grad_norm": 0.4459313154220581, "learning_rate": 0.0005, "loss": 1.52634596824646, "step": 29525 }, { "epoch": 2.32763106030745, "grad_norm": 0.48287561535835266, "learning_rate": 0.0005, "loss": 1.3431665897369385, "step": 29526 }, { "epoch": 2.3277098935750886, "grad_norm": 0.4422958493232727, "learning_rate": 0.0005, "loss": 1.3475325107574463, "step": 29527 }, { "epoch": 2.3277887268427278, "grad_norm": 0.5207160115242004, "learning_rate": 0.0005, "loss": 1.4526879787445068, "step": 29528 }, { "epoch": 2.3278675601103664, "grad_norm": 0.5147568583488464, "learning_rate": 0.0005, "loss": 1.3774209022521973, "step": 29529 }, { "epoch": 2.3279463933780056, "grad_norm": 0.5794849395751953, "learning_rate": 0.0005, "loss": 1.3523380756378174, "step": 29530 }, { "epoch": 2.3280252266456447, "grad_norm": 0.5447437763214111, "learning_rate": 0.0005, "loss": 1.3343865871429443, "step": 29531 }, { "epoch": 2.3281040599132834, "grad_norm": 0.6825800538063049, "learning_rate": 0.0005, "loss": 1.3758341073989868, "step": 29532 }, { "epoch": 2.3281828931809225, "grad_norm": 0.6115847826004028, "learning_rate": 0.0005, "loss": 1.1291897296905518, "step": 29533 }, { "epoch": 2.328261726448561, "grad_norm": 0.5228520631790161, "learning_rate": 0.0005, "loss": 1.263817310333252, "step": 29534 }, { "epoch": 2.3283405597162004, "grad_norm": 0.4674103260040283, "learning_rate": 0.0005, "loss": 1.2140123844146729, "step": 29535 }, { "epoch": 2.328419392983839, "grad_norm": 0.4394422471523285, "learning_rate": 0.0005, "loss": 1.3604120016098022, "step": 29536 }, { "epoch": 2.328498226251478, "grad_norm": 0.5508050322532654, "learning_rate": 0.0005, "loss": 1.5349767208099365, "step": 29537 }, { "epoch": 2.328577059519117, "grad_norm": 0.45867639780044556, "learning_rate": 0.0005, "loss": 1.3630177974700928, "step": 29538 }, { "epoch": 2.328655892786756, "grad_norm": 0.47389543056488037, "learning_rate": 0.0005, "loss": 1.303797960281372, "step": 29539 }, { "epoch": 2.328734726054395, "grad_norm": 0.47089913487434387, "learning_rate": 0.0005, "loss": 1.2751221656799316, "step": 29540 }, { "epoch": 2.328813559322034, "grad_norm": 0.4343458116054535, "learning_rate": 0.0005, "loss": 1.257032036781311, "step": 29541 }, { "epoch": 2.328892392589673, "grad_norm": 0.5818983912467957, "learning_rate": 0.0005, "loss": 1.3972814083099365, "step": 29542 }, { "epoch": 2.3289712258573116, "grad_norm": 0.4131072461605072, "learning_rate": 0.0005, "loss": 1.2401527166366577, "step": 29543 }, { "epoch": 2.3290500591249508, "grad_norm": 0.4728400409221649, "learning_rate": 0.0005, "loss": 1.0908583402633667, "step": 29544 }, { "epoch": 2.32912889239259, "grad_norm": 0.4698943495750427, "learning_rate": 0.0005, "loss": 1.4152264595031738, "step": 29545 }, { "epoch": 2.3292077256602286, "grad_norm": 0.51828932762146, "learning_rate": 0.0005, "loss": 1.2870184183120728, "step": 29546 }, { "epoch": 2.3292865589278677, "grad_norm": 0.5426226854324341, "learning_rate": 0.0005, "loss": 1.3232314586639404, "step": 29547 }, { "epoch": 2.3293653921955064, "grad_norm": 0.5800778269767761, "learning_rate": 0.0005, "loss": 1.2883723974227905, "step": 29548 }, { "epoch": 2.3294442254631456, "grad_norm": 0.46195298433303833, "learning_rate": 0.0005, "loss": 1.2401893138885498, "step": 29549 }, { "epoch": 2.3295230587307842, "grad_norm": 0.5545164942741394, "learning_rate": 0.0005, "loss": 1.3709938526153564, "step": 29550 }, { "epoch": 2.3296018919984234, "grad_norm": 0.5137676000595093, "learning_rate": 0.0005, "loss": 1.1953353881835938, "step": 29551 }, { "epoch": 2.329680725266062, "grad_norm": 0.5311254858970642, "learning_rate": 0.0005, "loss": 1.4271197319030762, "step": 29552 }, { "epoch": 2.329759558533701, "grad_norm": 0.4712845981121063, "learning_rate": 0.0005, "loss": 1.4160616397857666, "step": 29553 }, { "epoch": 2.3298383918013403, "grad_norm": 0.4391297698020935, "learning_rate": 0.0005, "loss": 1.2346012592315674, "step": 29554 }, { "epoch": 2.329917225068979, "grad_norm": 0.5073040127754211, "learning_rate": 0.0005, "loss": 1.352420449256897, "step": 29555 }, { "epoch": 2.329996058336618, "grad_norm": 0.8256681561470032, "learning_rate": 0.0005, "loss": 1.6026874780654907, "step": 29556 }, { "epoch": 2.330074891604257, "grad_norm": 0.5107319951057434, "learning_rate": 0.0005, "loss": 1.2118749618530273, "step": 29557 }, { "epoch": 2.330153724871896, "grad_norm": 0.47453534603118896, "learning_rate": 0.0005, "loss": 1.3377383947372437, "step": 29558 }, { "epoch": 2.330232558139535, "grad_norm": 0.44168412685394287, "learning_rate": 0.0005, "loss": 1.2877764701843262, "step": 29559 }, { "epoch": 2.330311391407174, "grad_norm": 0.5814324617385864, "learning_rate": 0.0005, "loss": 1.4824562072753906, "step": 29560 }, { "epoch": 2.330390224674813, "grad_norm": 0.6185000538825989, "learning_rate": 0.0005, "loss": 1.4720427989959717, "step": 29561 }, { "epoch": 2.3304690579424516, "grad_norm": 0.4424804449081421, "learning_rate": 0.0005, "loss": 1.3434584140777588, "step": 29562 }, { "epoch": 2.3305478912100908, "grad_norm": 0.41427871584892273, "learning_rate": 0.0005, "loss": 1.3138508796691895, "step": 29563 }, { "epoch": 2.3306267244777295, "grad_norm": 0.5430837273597717, "learning_rate": 0.0005, "loss": 1.3822579383850098, "step": 29564 }, { "epoch": 2.3307055577453686, "grad_norm": 0.6435716152191162, "learning_rate": 0.0005, "loss": 1.613608956336975, "step": 29565 }, { "epoch": 2.3307843910130073, "grad_norm": 0.5633654594421387, "learning_rate": 0.0005, "loss": 1.2795605659484863, "step": 29566 }, { "epoch": 2.3308632242806464, "grad_norm": 0.48120805621147156, "learning_rate": 0.0005, "loss": 1.46905517578125, "step": 29567 }, { "epoch": 2.3309420575482855, "grad_norm": 0.5518143177032471, "learning_rate": 0.0005, "loss": 1.312685251235962, "step": 29568 }, { "epoch": 2.3310208908159242, "grad_norm": 0.41971439123153687, "learning_rate": 0.0005, "loss": 1.2364866733551025, "step": 29569 }, { "epoch": 2.3310997240835634, "grad_norm": 0.6726107597351074, "learning_rate": 0.0005, "loss": 1.3174097537994385, "step": 29570 }, { "epoch": 2.331178557351202, "grad_norm": 0.44287073612213135, "learning_rate": 0.0005, "loss": 1.1624447107315063, "step": 29571 }, { "epoch": 2.331257390618841, "grad_norm": 0.6292403936386108, "learning_rate": 0.0005, "loss": 1.5684938430786133, "step": 29572 }, { "epoch": 2.3313362238864803, "grad_norm": 0.43219614028930664, "learning_rate": 0.0005, "loss": 1.1883301734924316, "step": 29573 }, { "epoch": 2.331415057154119, "grad_norm": 0.6379328370094299, "learning_rate": 0.0005, "loss": 1.3624086380004883, "step": 29574 }, { "epoch": 2.331493890421758, "grad_norm": 0.4841654300689697, "learning_rate": 0.0005, "loss": 1.4661149978637695, "step": 29575 }, { "epoch": 2.331572723689397, "grad_norm": 0.6088178157806396, "learning_rate": 0.0005, "loss": 1.4913800954818726, "step": 29576 }, { "epoch": 2.331651556957036, "grad_norm": 0.6050405502319336, "learning_rate": 0.0005, "loss": 1.637343168258667, "step": 29577 }, { "epoch": 2.3317303902246747, "grad_norm": 0.534100353717804, "learning_rate": 0.0005, "loss": 1.3984382152557373, "step": 29578 }, { "epoch": 2.331809223492314, "grad_norm": 0.4440942406654358, "learning_rate": 0.0005, "loss": 1.396169900894165, "step": 29579 }, { "epoch": 2.3318880567599525, "grad_norm": 0.4737273156642914, "learning_rate": 0.0005, "loss": 1.1734358072280884, "step": 29580 }, { "epoch": 2.3319668900275916, "grad_norm": 0.541495680809021, "learning_rate": 0.0005, "loss": 1.4312025308609009, "step": 29581 }, { "epoch": 2.3320457232952307, "grad_norm": 0.5070326924324036, "learning_rate": 0.0005, "loss": 1.459580659866333, "step": 29582 }, { "epoch": 2.3321245565628694, "grad_norm": 0.6573865413665771, "learning_rate": 0.0005, "loss": 1.3696677684783936, "step": 29583 }, { "epoch": 2.3322033898305086, "grad_norm": 0.45373547077178955, "learning_rate": 0.0005, "loss": 1.2755467891693115, "step": 29584 }, { "epoch": 2.3322822230981473, "grad_norm": 0.5577782988548279, "learning_rate": 0.0005, "loss": 1.411484718322754, "step": 29585 }, { "epoch": 2.3323610563657864, "grad_norm": 0.4997299313545227, "learning_rate": 0.0005, "loss": 1.3520503044128418, "step": 29586 }, { "epoch": 2.3324398896334255, "grad_norm": 0.5019572973251343, "learning_rate": 0.0005, "loss": 1.5169589519500732, "step": 29587 }, { "epoch": 2.332518722901064, "grad_norm": 0.5207830667495728, "learning_rate": 0.0005, "loss": 1.4246631860733032, "step": 29588 }, { "epoch": 2.3325975561687033, "grad_norm": 0.5268940329551697, "learning_rate": 0.0005, "loss": 1.3120405673980713, "step": 29589 }, { "epoch": 2.332676389436342, "grad_norm": 0.4819983243942261, "learning_rate": 0.0005, "loss": 1.2509125471115112, "step": 29590 }, { "epoch": 2.332755222703981, "grad_norm": 0.7505690455436707, "learning_rate": 0.0005, "loss": 1.426804542541504, "step": 29591 }, { "epoch": 2.33283405597162, "grad_norm": 0.5602685213088989, "learning_rate": 0.0005, "loss": 1.6753960847854614, "step": 29592 }, { "epoch": 2.332912889239259, "grad_norm": 0.589451014995575, "learning_rate": 0.0005, "loss": 1.4125466346740723, "step": 29593 }, { "epoch": 2.3329917225068977, "grad_norm": 0.5880053043365479, "learning_rate": 0.0005, "loss": 1.4890422821044922, "step": 29594 }, { "epoch": 2.333070555774537, "grad_norm": 0.509287416934967, "learning_rate": 0.0005, "loss": 1.233768105506897, "step": 29595 }, { "epoch": 2.333149389042176, "grad_norm": 0.5170086026191711, "learning_rate": 0.0005, "loss": 1.3098106384277344, "step": 29596 }, { "epoch": 2.3332282223098146, "grad_norm": 0.4966113269329071, "learning_rate": 0.0005, "loss": 1.4113152027130127, "step": 29597 }, { "epoch": 2.3333070555774538, "grad_norm": 0.5368326902389526, "learning_rate": 0.0005, "loss": 1.2742270231246948, "step": 29598 }, { "epoch": 2.3333858888450925, "grad_norm": 0.49962469935417175, "learning_rate": 0.0005, "loss": 1.2774732112884521, "step": 29599 }, { "epoch": 2.3334647221127316, "grad_norm": 0.5156096816062927, "learning_rate": 0.0005, "loss": 1.328262448310852, "step": 29600 }, { "epoch": 2.3335435553803707, "grad_norm": 0.5285976529121399, "learning_rate": 0.0005, "loss": 1.3911693096160889, "step": 29601 }, { "epoch": 2.3336223886480094, "grad_norm": 0.5282471179962158, "learning_rate": 0.0005, "loss": 1.3474528789520264, "step": 29602 }, { "epoch": 2.3337012219156485, "grad_norm": 0.48772963881492615, "learning_rate": 0.0005, "loss": 1.213475227355957, "step": 29603 }, { "epoch": 2.3337800551832872, "grad_norm": 0.47989168763160706, "learning_rate": 0.0005, "loss": 1.115077018737793, "step": 29604 }, { "epoch": 2.3338588884509264, "grad_norm": 0.5290720462799072, "learning_rate": 0.0005, "loss": 1.402511715888977, "step": 29605 }, { "epoch": 2.333937721718565, "grad_norm": 0.5546876192092896, "learning_rate": 0.0005, "loss": 1.434524655342102, "step": 29606 }, { "epoch": 2.334016554986204, "grad_norm": 0.4461762309074402, "learning_rate": 0.0005, "loss": 1.1463382244110107, "step": 29607 }, { "epoch": 2.334095388253843, "grad_norm": 0.5067219138145447, "learning_rate": 0.0005, "loss": 1.2733842134475708, "step": 29608 }, { "epoch": 2.334174221521482, "grad_norm": 0.4441699683666229, "learning_rate": 0.0005, "loss": 1.2629859447479248, "step": 29609 }, { "epoch": 2.334253054789121, "grad_norm": 0.5340917706489563, "learning_rate": 0.0005, "loss": 1.3910274505615234, "step": 29610 }, { "epoch": 2.33433188805676, "grad_norm": 0.6279860734939575, "learning_rate": 0.0005, "loss": 1.6006724834442139, "step": 29611 }, { "epoch": 2.334410721324399, "grad_norm": 0.4554206430912018, "learning_rate": 0.0005, "loss": 1.43483567237854, "step": 29612 }, { "epoch": 2.3344895545920377, "grad_norm": 0.5220403671264648, "learning_rate": 0.0005, "loss": 1.206761360168457, "step": 29613 }, { "epoch": 2.334568387859677, "grad_norm": 0.578116238117218, "learning_rate": 0.0005, "loss": 1.3708124160766602, "step": 29614 }, { "epoch": 2.334647221127316, "grad_norm": 0.5004519820213318, "learning_rate": 0.0005, "loss": 1.428912878036499, "step": 29615 }, { "epoch": 2.3347260543949546, "grad_norm": 0.6140292882919312, "learning_rate": 0.0005, "loss": 1.5059483051300049, "step": 29616 }, { "epoch": 2.3348048876625938, "grad_norm": 0.5052311420440674, "learning_rate": 0.0005, "loss": 1.2249053716659546, "step": 29617 }, { "epoch": 2.3348837209302324, "grad_norm": 0.44366133213043213, "learning_rate": 0.0005, "loss": 1.3033537864685059, "step": 29618 }, { "epoch": 2.3349625541978716, "grad_norm": 0.6158618330955505, "learning_rate": 0.0005, "loss": 1.3456687927246094, "step": 29619 }, { "epoch": 2.3350413874655103, "grad_norm": 0.5672368407249451, "learning_rate": 0.0005, "loss": 1.2802764177322388, "step": 29620 }, { "epoch": 2.3351202207331494, "grad_norm": 0.556364893913269, "learning_rate": 0.0005, "loss": 1.343886375427246, "step": 29621 }, { "epoch": 2.335199054000788, "grad_norm": 0.6286629438400269, "learning_rate": 0.0005, "loss": 1.4837119579315186, "step": 29622 }, { "epoch": 2.335277887268427, "grad_norm": 0.5069075226783752, "learning_rate": 0.0005, "loss": 1.2271615266799927, "step": 29623 }, { "epoch": 2.3353567205360664, "grad_norm": 0.5140282511711121, "learning_rate": 0.0005, "loss": 1.4507951736450195, "step": 29624 }, { "epoch": 2.335435553803705, "grad_norm": 0.4448780119419098, "learning_rate": 0.0005, "loss": 1.216355800628662, "step": 29625 }, { "epoch": 2.335514387071344, "grad_norm": 0.60933917760849, "learning_rate": 0.0005, "loss": 1.4297600984573364, "step": 29626 }, { "epoch": 2.335593220338983, "grad_norm": 0.47323691844940186, "learning_rate": 0.0005, "loss": 1.486675500869751, "step": 29627 }, { "epoch": 2.335672053606622, "grad_norm": 0.4604658782482147, "learning_rate": 0.0005, "loss": 1.510624885559082, "step": 29628 }, { "epoch": 2.335750886874261, "grad_norm": 0.467332661151886, "learning_rate": 0.0005, "loss": 1.2316336631774902, "step": 29629 }, { "epoch": 2.3358297201419, "grad_norm": 0.4945216476917267, "learning_rate": 0.0005, "loss": 1.4377944469451904, "step": 29630 }, { "epoch": 2.335908553409539, "grad_norm": 0.4899480640888214, "learning_rate": 0.0005, "loss": 1.364368200302124, "step": 29631 }, { "epoch": 2.3359873866771776, "grad_norm": 0.5327165126800537, "learning_rate": 0.0005, "loss": 1.37844717502594, "step": 29632 }, { "epoch": 2.336066219944817, "grad_norm": 0.5494226217269897, "learning_rate": 0.0005, "loss": 1.5726187229156494, "step": 29633 }, { "epoch": 2.3361450532124555, "grad_norm": 0.6253211498260498, "learning_rate": 0.0005, "loss": 1.4647059440612793, "step": 29634 }, { "epoch": 2.3362238864800946, "grad_norm": 0.4245481789112091, "learning_rate": 0.0005, "loss": 1.3871712684631348, "step": 29635 }, { "epoch": 2.3363027197477337, "grad_norm": 0.5379452705383301, "learning_rate": 0.0005, "loss": 1.3500597476959229, "step": 29636 }, { "epoch": 2.3363815530153724, "grad_norm": 0.4577617049217224, "learning_rate": 0.0005, "loss": 1.2950040102005005, "step": 29637 }, { "epoch": 2.3364603862830116, "grad_norm": 0.5052088499069214, "learning_rate": 0.0005, "loss": 1.3257722854614258, "step": 29638 }, { "epoch": 2.3365392195506502, "grad_norm": 0.5041306018829346, "learning_rate": 0.0005, "loss": 1.3914985656738281, "step": 29639 }, { "epoch": 2.3366180528182894, "grad_norm": 0.4999498724937439, "learning_rate": 0.0005, "loss": 1.3799943923950195, "step": 29640 }, { "epoch": 2.336696886085928, "grad_norm": 0.48458632826805115, "learning_rate": 0.0005, "loss": 1.276695728302002, "step": 29641 }, { "epoch": 2.336775719353567, "grad_norm": 0.6448889374732971, "learning_rate": 0.0005, "loss": 1.2278655767440796, "step": 29642 }, { "epoch": 2.3368545526212063, "grad_norm": 0.4306105375289917, "learning_rate": 0.0005, "loss": 1.1734652519226074, "step": 29643 }, { "epoch": 2.336933385888845, "grad_norm": 0.5720920562744141, "learning_rate": 0.0005, "loss": 1.4004502296447754, "step": 29644 }, { "epoch": 2.337012219156484, "grad_norm": 0.77936851978302, "learning_rate": 0.0005, "loss": 1.181879997253418, "step": 29645 }, { "epoch": 2.337091052424123, "grad_norm": 0.4859161972999573, "learning_rate": 0.0005, "loss": 1.0256423950195312, "step": 29646 }, { "epoch": 2.337169885691762, "grad_norm": 0.4940865933895111, "learning_rate": 0.0005, "loss": 1.5009669065475464, "step": 29647 }, { "epoch": 2.3372487189594007, "grad_norm": 0.4931454658508301, "learning_rate": 0.0005, "loss": 1.376664161682129, "step": 29648 }, { "epoch": 2.33732755222704, "grad_norm": 0.4693700969219208, "learning_rate": 0.0005, "loss": 1.0600268840789795, "step": 29649 }, { "epoch": 2.337406385494679, "grad_norm": 0.5059502720832825, "learning_rate": 0.0005, "loss": 1.5343196392059326, "step": 29650 }, { "epoch": 2.3374852187623176, "grad_norm": 0.5693846344947815, "learning_rate": 0.0005, "loss": 1.5316319465637207, "step": 29651 }, { "epoch": 2.3375640520299568, "grad_norm": 0.46482548117637634, "learning_rate": 0.0005, "loss": 1.3727209568023682, "step": 29652 }, { "epoch": 2.3376428852975955, "grad_norm": 0.4764374792575836, "learning_rate": 0.0005, "loss": 1.1286630630493164, "step": 29653 }, { "epoch": 2.3377217185652346, "grad_norm": 0.4374273717403412, "learning_rate": 0.0005, "loss": 1.2362045049667358, "step": 29654 }, { "epoch": 2.3378005518328733, "grad_norm": 0.6914564371109009, "learning_rate": 0.0005, "loss": 1.1990516185760498, "step": 29655 }, { "epoch": 2.3378793851005124, "grad_norm": 0.5125323534011841, "learning_rate": 0.0005, "loss": 1.2864649295806885, "step": 29656 }, { "epoch": 2.3379582183681515, "grad_norm": 0.4632607102394104, "learning_rate": 0.0005, "loss": 1.4376682043075562, "step": 29657 }, { "epoch": 2.3380370516357902, "grad_norm": 0.6477321982383728, "learning_rate": 0.0005, "loss": 1.2321550846099854, "step": 29658 }, { "epoch": 2.3381158849034294, "grad_norm": 0.4409422278404236, "learning_rate": 0.0005, "loss": 1.345360517501831, "step": 29659 }, { "epoch": 2.338194718171068, "grad_norm": 0.6845542192459106, "learning_rate": 0.0005, "loss": 1.5529053211212158, "step": 29660 }, { "epoch": 2.338273551438707, "grad_norm": 0.5469959378242493, "learning_rate": 0.0005, "loss": 1.5103247165679932, "step": 29661 }, { "epoch": 2.3383523847063463, "grad_norm": 0.5211078524589539, "learning_rate": 0.0005, "loss": 1.3123962879180908, "step": 29662 }, { "epoch": 2.338431217973985, "grad_norm": 0.40139153599739075, "learning_rate": 0.0005, "loss": 1.1940566301345825, "step": 29663 }, { "epoch": 2.338510051241624, "grad_norm": 0.49467962980270386, "learning_rate": 0.0005, "loss": 1.4333564043045044, "step": 29664 }, { "epoch": 2.338588884509263, "grad_norm": 0.46789661049842834, "learning_rate": 0.0005, "loss": 1.253535509109497, "step": 29665 }, { "epoch": 2.338667717776902, "grad_norm": 0.6969893574714661, "learning_rate": 0.0005, "loss": 1.467712163925171, "step": 29666 }, { "epoch": 2.3387465510445407, "grad_norm": 0.9617526531219482, "learning_rate": 0.0005, "loss": 1.4172132015228271, "step": 29667 }, { "epoch": 2.33882538431218, "grad_norm": 0.5249393582344055, "learning_rate": 0.0005, "loss": 1.330103874206543, "step": 29668 }, { "epoch": 2.3389042175798185, "grad_norm": 0.532810628414154, "learning_rate": 0.0005, "loss": 1.3234816789627075, "step": 29669 }, { "epoch": 2.3389830508474576, "grad_norm": 0.4019145369529724, "learning_rate": 0.0005, "loss": 1.1011183261871338, "step": 29670 }, { "epoch": 2.3390618841150967, "grad_norm": 0.6057009100914001, "learning_rate": 0.0005, "loss": 1.2894970178604126, "step": 29671 }, { "epoch": 2.3391407173827354, "grad_norm": 0.42514532804489136, "learning_rate": 0.0005, "loss": 1.280982494354248, "step": 29672 }, { "epoch": 2.3392195506503746, "grad_norm": 0.4729122519493103, "learning_rate": 0.0005, "loss": 1.2965795993804932, "step": 29673 }, { "epoch": 2.3392983839180133, "grad_norm": 0.5204010009765625, "learning_rate": 0.0005, "loss": 1.5115153789520264, "step": 29674 }, { "epoch": 2.3393772171856524, "grad_norm": 0.46968716382980347, "learning_rate": 0.0005, "loss": 1.2936691045761108, "step": 29675 }, { "epoch": 2.3394560504532915, "grad_norm": 0.532665491104126, "learning_rate": 0.0005, "loss": 1.2264987230300903, "step": 29676 }, { "epoch": 2.33953488372093, "grad_norm": 0.4828891158103943, "learning_rate": 0.0005, "loss": 1.212092638015747, "step": 29677 }, { "epoch": 2.3396137169885693, "grad_norm": 0.5192029476165771, "learning_rate": 0.0005, "loss": 1.3268611431121826, "step": 29678 }, { "epoch": 2.339692550256208, "grad_norm": 0.7764449715614319, "learning_rate": 0.0005, "loss": 1.214795470237732, "step": 29679 }, { "epoch": 2.339771383523847, "grad_norm": 0.5459150075912476, "learning_rate": 0.0005, "loss": 1.3277077674865723, "step": 29680 }, { "epoch": 2.339850216791486, "grad_norm": 0.4100345969200134, "learning_rate": 0.0005, "loss": 1.2316555976867676, "step": 29681 }, { "epoch": 2.339929050059125, "grad_norm": 0.49626246094703674, "learning_rate": 0.0005, "loss": 1.4694080352783203, "step": 29682 }, { "epoch": 2.3400078833267637, "grad_norm": 0.519753098487854, "learning_rate": 0.0005, "loss": 1.3671501874923706, "step": 29683 }, { "epoch": 2.340086716594403, "grad_norm": 0.45536836981773376, "learning_rate": 0.0005, "loss": 1.324605941772461, "step": 29684 }, { "epoch": 2.340165549862042, "grad_norm": 0.4795760214328766, "learning_rate": 0.0005, "loss": 1.3727686405181885, "step": 29685 }, { "epoch": 2.3402443831296806, "grad_norm": 0.5428752899169922, "learning_rate": 0.0005, "loss": 1.3402512073516846, "step": 29686 }, { "epoch": 2.3403232163973198, "grad_norm": 0.6257197260856628, "learning_rate": 0.0005, "loss": 1.3625056743621826, "step": 29687 }, { "epoch": 2.3404020496649585, "grad_norm": 0.40950170159339905, "learning_rate": 0.0005, "loss": 1.0883078575134277, "step": 29688 }, { "epoch": 2.3404808829325976, "grad_norm": 0.47835573554039, "learning_rate": 0.0005, "loss": 1.2833409309387207, "step": 29689 }, { "epoch": 2.3405597162002367, "grad_norm": 0.43539121747016907, "learning_rate": 0.0005, "loss": 1.0325621366500854, "step": 29690 }, { "epoch": 2.3406385494678754, "grad_norm": 0.45867207646369934, "learning_rate": 0.0005, "loss": 1.3112061023712158, "step": 29691 }, { "epoch": 2.3407173827355146, "grad_norm": 0.5972163677215576, "learning_rate": 0.0005, "loss": 1.449168086051941, "step": 29692 }, { "epoch": 2.3407962160031532, "grad_norm": 0.4338204562664032, "learning_rate": 0.0005, "loss": 1.2834051847457886, "step": 29693 }, { "epoch": 2.3408750492707924, "grad_norm": 0.47263747453689575, "learning_rate": 0.0005, "loss": 1.465532660484314, "step": 29694 }, { "epoch": 2.340953882538431, "grad_norm": 0.5667664408683777, "learning_rate": 0.0005, "loss": 1.3193774223327637, "step": 29695 }, { "epoch": 2.34103271580607, "grad_norm": 0.5330780148506165, "learning_rate": 0.0005, "loss": 1.6262261867523193, "step": 29696 }, { "epoch": 2.341111549073709, "grad_norm": 0.47423848509788513, "learning_rate": 0.0005, "loss": 1.294267177581787, "step": 29697 }, { "epoch": 2.341190382341348, "grad_norm": 0.543823778629303, "learning_rate": 0.0005, "loss": 1.4425033330917358, "step": 29698 }, { "epoch": 2.341269215608987, "grad_norm": 0.467242568731308, "learning_rate": 0.0005, "loss": 1.498171091079712, "step": 29699 }, { "epoch": 2.341348048876626, "grad_norm": 0.48684486746788025, "learning_rate": 0.0005, "loss": 1.4292855262756348, "step": 29700 }, { "epoch": 2.341426882144265, "grad_norm": 0.5955221056938171, "learning_rate": 0.0005, "loss": 1.4762738943099976, "step": 29701 }, { "epoch": 2.3415057154119037, "grad_norm": 0.5100277662277222, "learning_rate": 0.0005, "loss": 1.3901150226593018, "step": 29702 }, { "epoch": 2.341584548679543, "grad_norm": 0.5602821111679077, "learning_rate": 0.0005, "loss": 1.4396018981933594, "step": 29703 }, { "epoch": 2.341663381947182, "grad_norm": 0.3959665596485138, "learning_rate": 0.0005, "loss": 1.2599722146987915, "step": 29704 }, { "epoch": 2.3417422152148206, "grad_norm": 0.5286287665367126, "learning_rate": 0.0005, "loss": 1.3957185745239258, "step": 29705 }, { "epoch": 2.3418210484824598, "grad_norm": 0.49393847584724426, "learning_rate": 0.0005, "loss": 1.386176347732544, "step": 29706 }, { "epoch": 2.3418998817500984, "grad_norm": 0.4371560215950012, "learning_rate": 0.0005, "loss": 1.1886396408081055, "step": 29707 }, { "epoch": 2.3419787150177376, "grad_norm": 0.5427003502845764, "learning_rate": 0.0005, "loss": 1.292320728302002, "step": 29708 }, { "epoch": 2.3420575482853763, "grad_norm": 0.4342179298400879, "learning_rate": 0.0005, "loss": 1.202883005142212, "step": 29709 }, { "epoch": 2.3421363815530154, "grad_norm": 0.5141597986221313, "learning_rate": 0.0005, "loss": 1.280987024307251, "step": 29710 }, { "epoch": 2.342215214820654, "grad_norm": 0.6167502999305725, "learning_rate": 0.0005, "loss": 1.3507966995239258, "step": 29711 }, { "epoch": 2.3422940480882932, "grad_norm": 0.4402848780155182, "learning_rate": 0.0005, "loss": 1.2653181552886963, "step": 29712 }, { "epoch": 2.3423728813559324, "grad_norm": 0.4350348114967346, "learning_rate": 0.0005, "loss": 1.2936341762542725, "step": 29713 }, { "epoch": 2.342451714623571, "grad_norm": 0.48087117075920105, "learning_rate": 0.0005, "loss": 1.4521994590759277, "step": 29714 }, { "epoch": 2.34253054789121, "grad_norm": 0.449282169342041, "learning_rate": 0.0005, "loss": 1.2199143171310425, "step": 29715 }, { "epoch": 2.342609381158849, "grad_norm": 0.6254090666770935, "learning_rate": 0.0005, "loss": 1.502145767211914, "step": 29716 }, { "epoch": 2.342688214426488, "grad_norm": 0.47621139883995056, "learning_rate": 0.0005, "loss": 1.6003408432006836, "step": 29717 }, { "epoch": 2.342767047694127, "grad_norm": 0.49681994318962097, "learning_rate": 0.0005, "loss": 1.4699809551239014, "step": 29718 }, { "epoch": 2.342845880961766, "grad_norm": 0.5929498076438904, "learning_rate": 0.0005, "loss": 1.5096657276153564, "step": 29719 }, { "epoch": 2.342924714229405, "grad_norm": 0.4800785183906555, "learning_rate": 0.0005, "loss": 1.3174562454223633, "step": 29720 }, { "epoch": 2.3430035474970436, "grad_norm": 0.4458729922771454, "learning_rate": 0.0005, "loss": 1.1536085605621338, "step": 29721 }, { "epoch": 2.343082380764683, "grad_norm": 0.5115292072296143, "learning_rate": 0.0005, "loss": 1.2695183753967285, "step": 29722 }, { "epoch": 2.3431612140323215, "grad_norm": 0.5260081887245178, "learning_rate": 0.0005, "loss": 1.4497511386871338, "step": 29723 }, { "epoch": 2.3432400472999606, "grad_norm": 0.6380015015602112, "learning_rate": 0.0005, "loss": 1.3843929767608643, "step": 29724 }, { "epoch": 2.3433188805675993, "grad_norm": 0.6023973822593689, "learning_rate": 0.0005, "loss": 1.4487316608428955, "step": 29725 }, { "epoch": 2.3433977138352384, "grad_norm": 0.5994687080383301, "learning_rate": 0.0005, "loss": 1.325862169265747, "step": 29726 }, { "epoch": 2.3434765471028776, "grad_norm": 0.4927842319011688, "learning_rate": 0.0005, "loss": 1.3452856540679932, "step": 29727 }, { "epoch": 2.3435553803705163, "grad_norm": 0.4679536521434784, "learning_rate": 0.0005, "loss": 1.3054448366165161, "step": 29728 }, { "epoch": 2.3436342136381554, "grad_norm": 0.4617908298969269, "learning_rate": 0.0005, "loss": 1.3997918367385864, "step": 29729 }, { "epoch": 2.343713046905794, "grad_norm": 0.4660905599594116, "learning_rate": 0.0005, "loss": 1.2607128620147705, "step": 29730 }, { "epoch": 2.343791880173433, "grad_norm": 0.41090676188468933, "learning_rate": 0.0005, "loss": 1.1623897552490234, "step": 29731 }, { "epoch": 2.3438707134410723, "grad_norm": 0.5881739258766174, "learning_rate": 0.0005, "loss": 1.2754347324371338, "step": 29732 }, { "epoch": 2.343949546708711, "grad_norm": 0.49911707639694214, "learning_rate": 0.0005, "loss": 1.31876540184021, "step": 29733 }, { "epoch": 2.34402837997635, "grad_norm": 0.45402029156684875, "learning_rate": 0.0005, "loss": 1.322155475616455, "step": 29734 }, { "epoch": 2.344107213243989, "grad_norm": 0.4345840513706207, "learning_rate": 0.0005, "loss": 1.2291057109832764, "step": 29735 }, { "epoch": 2.344186046511628, "grad_norm": 0.45449891686439514, "learning_rate": 0.0005, "loss": 1.2227020263671875, "step": 29736 }, { "epoch": 2.3442648797792667, "grad_norm": 0.5084713697433472, "learning_rate": 0.0005, "loss": 1.406026005744934, "step": 29737 }, { "epoch": 2.344343713046906, "grad_norm": 0.5276031494140625, "learning_rate": 0.0005, "loss": 1.373060703277588, "step": 29738 }, { "epoch": 2.3444225463145445, "grad_norm": 0.4903833270072937, "learning_rate": 0.0005, "loss": 1.3113126754760742, "step": 29739 }, { "epoch": 2.3445013795821836, "grad_norm": 0.5164191126823425, "learning_rate": 0.0005, "loss": 1.3803671598434448, "step": 29740 }, { "epoch": 2.3445802128498228, "grad_norm": 0.583541214466095, "learning_rate": 0.0005, "loss": 1.4733967781066895, "step": 29741 }, { "epoch": 2.3446590461174615, "grad_norm": 0.5890262722969055, "learning_rate": 0.0005, "loss": 1.4925117492675781, "step": 29742 }, { "epoch": 2.3447378793851006, "grad_norm": 0.45510080456733704, "learning_rate": 0.0005, "loss": 1.281174898147583, "step": 29743 }, { "epoch": 2.3448167126527393, "grad_norm": 0.5465981364250183, "learning_rate": 0.0005, "loss": 1.354162335395813, "step": 29744 }, { "epoch": 2.3448955459203784, "grad_norm": 0.4664878845214844, "learning_rate": 0.0005, "loss": 1.4574840068817139, "step": 29745 }, { "epoch": 2.3449743791880175, "grad_norm": 0.7389584183692932, "learning_rate": 0.0005, "loss": 1.3333854675292969, "step": 29746 }, { "epoch": 2.3450532124556562, "grad_norm": 0.4306030869483948, "learning_rate": 0.0005, "loss": 1.2465434074401855, "step": 29747 }, { "epoch": 2.3451320457232954, "grad_norm": 0.4220992624759674, "learning_rate": 0.0005, "loss": 1.1996722221374512, "step": 29748 }, { "epoch": 2.345210878990934, "grad_norm": 0.4679851233959198, "learning_rate": 0.0005, "loss": 1.2464573383331299, "step": 29749 }, { "epoch": 2.345289712258573, "grad_norm": 0.5658190250396729, "learning_rate": 0.0005, "loss": 1.3379040956497192, "step": 29750 }, { "epoch": 2.345368545526212, "grad_norm": 0.5947656631469727, "learning_rate": 0.0005, "loss": 1.4646046161651611, "step": 29751 }, { "epoch": 2.345447378793851, "grad_norm": 0.5835243463516235, "learning_rate": 0.0005, "loss": 1.628892183303833, "step": 29752 }, { "epoch": 2.3455262120614897, "grad_norm": 0.45976635813713074, "learning_rate": 0.0005, "loss": 1.2677987813949585, "step": 29753 }, { "epoch": 2.345605045329129, "grad_norm": 0.4291386604309082, "learning_rate": 0.0005, "loss": 1.409647822380066, "step": 29754 }, { "epoch": 2.345683878596768, "grad_norm": 0.590758740901947, "learning_rate": 0.0005, "loss": 1.3395493030548096, "step": 29755 }, { "epoch": 2.3457627118644067, "grad_norm": 0.848929762840271, "learning_rate": 0.0005, "loss": 1.5009057521820068, "step": 29756 }, { "epoch": 2.345841545132046, "grad_norm": 0.67289799451828, "learning_rate": 0.0005, "loss": 1.468066930770874, "step": 29757 }, { "epoch": 2.3459203783996845, "grad_norm": 0.5895066261291504, "learning_rate": 0.0005, "loss": 1.2933573722839355, "step": 29758 }, { "epoch": 2.3459992116673236, "grad_norm": 0.45610755681991577, "learning_rate": 0.0005, "loss": 1.2476694583892822, "step": 29759 }, { "epoch": 2.3460780449349627, "grad_norm": 0.49399229884147644, "learning_rate": 0.0005, "loss": 1.0377542972564697, "step": 29760 }, { "epoch": 2.3461568782026014, "grad_norm": 0.5211862325668335, "learning_rate": 0.0005, "loss": 1.3198796510696411, "step": 29761 }, { "epoch": 2.3462357114702406, "grad_norm": 0.5314942002296448, "learning_rate": 0.0005, "loss": 1.308964729309082, "step": 29762 }, { "epoch": 2.3463145447378793, "grad_norm": 0.5350594520568848, "learning_rate": 0.0005, "loss": 1.2931454181671143, "step": 29763 }, { "epoch": 2.3463933780055184, "grad_norm": 0.4574263095855713, "learning_rate": 0.0005, "loss": 1.2254774570465088, "step": 29764 }, { "epoch": 2.346472211273157, "grad_norm": 0.5550715923309326, "learning_rate": 0.0005, "loss": 1.2013623714447021, "step": 29765 }, { "epoch": 2.346551044540796, "grad_norm": 0.5205934643745422, "learning_rate": 0.0005, "loss": 1.339331865310669, "step": 29766 }, { "epoch": 2.3466298778084353, "grad_norm": 0.5964100956916809, "learning_rate": 0.0005, "loss": 1.3885756731033325, "step": 29767 }, { "epoch": 2.346708711076074, "grad_norm": 0.44855841994285583, "learning_rate": 0.0005, "loss": 1.330251693725586, "step": 29768 }, { "epoch": 2.346787544343713, "grad_norm": 0.43275701999664307, "learning_rate": 0.0005, "loss": 1.1179132461547852, "step": 29769 }, { "epoch": 2.346866377611352, "grad_norm": 0.5229983925819397, "learning_rate": 0.0005, "loss": 1.3603847026824951, "step": 29770 }, { "epoch": 2.346945210878991, "grad_norm": 0.5030547380447388, "learning_rate": 0.0005, "loss": 1.3507161140441895, "step": 29771 }, { "epoch": 2.3470240441466297, "grad_norm": 0.629706859588623, "learning_rate": 0.0005, "loss": 1.2859481573104858, "step": 29772 }, { "epoch": 2.347102877414269, "grad_norm": 0.46386268734931946, "learning_rate": 0.0005, "loss": 1.3428940773010254, "step": 29773 }, { "epoch": 2.347181710681908, "grad_norm": 0.4892864525318146, "learning_rate": 0.0005, "loss": 1.3423547744750977, "step": 29774 }, { "epoch": 2.3472605439495466, "grad_norm": 0.854488730430603, "learning_rate": 0.0005, "loss": 1.589752197265625, "step": 29775 }, { "epoch": 2.3473393772171858, "grad_norm": 0.4962095320224762, "learning_rate": 0.0005, "loss": 1.2296664714813232, "step": 29776 }, { "epoch": 2.3474182104848245, "grad_norm": 0.5389366149902344, "learning_rate": 0.0005, "loss": 1.4002680778503418, "step": 29777 }, { "epoch": 2.3474970437524636, "grad_norm": 0.5493225455284119, "learning_rate": 0.0005, "loss": 1.5762457847595215, "step": 29778 }, { "epoch": 2.3475758770201027, "grad_norm": 0.5463783740997314, "learning_rate": 0.0005, "loss": 1.4409312009811401, "step": 29779 }, { "epoch": 2.3476547102877414, "grad_norm": 0.5042445659637451, "learning_rate": 0.0005, "loss": 1.2941381931304932, "step": 29780 }, { "epoch": 2.3477335435553806, "grad_norm": 0.4815971553325653, "learning_rate": 0.0005, "loss": 1.3393617868423462, "step": 29781 }, { "epoch": 2.3478123768230192, "grad_norm": 0.6033822298049927, "learning_rate": 0.0005, "loss": 1.1745647192001343, "step": 29782 }, { "epoch": 2.3478912100906584, "grad_norm": 0.48537856340408325, "learning_rate": 0.0005, "loss": 1.2034595012664795, "step": 29783 }, { "epoch": 2.347970043358297, "grad_norm": 0.41726699471473694, "learning_rate": 0.0005, "loss": 1.2699624300003052, "step": 29784 }, { "epoch": 2.348048876625936, "grad_norm": 0.554895281791687, "learning_rate": 0.0005, "loss": 1.3312439918518066, "step": 29785 }, { "epoch": 2.348127709893575, "grad_norm": 0.6464790105819702, "learning_rate": 0.0005, "loss": 1.6772509813308716, "step": 29786 }, { "epoch": 2.348206543161214, "grad_norm": 0.5756717324256897, "learning_rate": 0.0005, "loss": 1.3084871768951416, "step": 29787 }, { "epoch": 2.348285376428853, "grad_norm": 0.49585333466529846, "learning_rate": 0.0005, "loss": 1.23455810546875, "step": 29788 }, { "epoch": 2.348364209696492, "grad_norm": 0.5995557308197021, "learning_rate": 0.0005, "loss": 1.4644120931625366, "step": 29789 }, { "epoch": 2.348443042964131, "grad_norm": 0.5201660394668579, "learning_rate": 0.0005, "loss": 1.3213891983032227, "step": 29790 }, { "epoch": 2.3485218762317697, "grad_norm": 0.5745948553085327, "learning_rate": 0.0005, "loss": 1.3019262552261353, "step": 29791 }, { "epoch": 2.348600709499409, "grad_norm": 0.5486956238746643, "learning_rate": 0.0005, "loss": 1.3312931060791016, "step": 29792 }, { "epoch": 2.348679542767048, "grad_norm": 0.5394384860992432, "learning_rate": 0.0005, "loss": 1.244167447090149, "step": 29793 }, { "epoch": 2.3487583760346866, "grad_norm": 0.6172693371772766, "learning_rate": 0.0005, "loss": 1.2236112356185913, "step": 29794 }, { "epoch": 2.3488372093023258, "grad_norm": 0.8526232242584229, "learning_rate": 0.0005, "loss": 1.4479360580444336, "step": 29795 }, { "epoch": 2.3489160425699644, "grad_norm": 0.602843165397644, "learning_rate": 0.0005, "loss": 1.5389524698257446, "step": 29796 }, { "epoch": 2.3489948758376036, "grad_norm": 0.4974783658981323, "learning_rate": 0.0005, "loss": 1.24717378616333, "step": 29797 }, { "epoch": 2.3490737091052423, "grad_norm": 0.5187832713127136, "learning_rate": 0.0005, "loss": 1.298536777496338, "step": 29798 }, { "epoch": 2.3491525423728814, "grad_norm": 0.562798798084259, "learning_rate": 0.0005, "loss": 1.4143173694610596, "step": 29799 }, { "epoch": 2.34923137564052, "grad_norm": 0.4934246242046356, "learning_rate": 0.0005, "loss": 1.4681086540222168, "step": 29800 }, { "epoch": 2.3493102089081592, "grad_norm": 0.447225421667099, "learning_rate": 0.0005, "loss": 1.3457939624786377, "step": 29801 }, { "epoch": 2.3493890421757984, "grad_norm": 0.6486877202987671, "learning_rate": 0.0005, "loss": 1.399961233139038, "step": 29802 }, { "epoch": 2.349467875443437, "grad_norm": 0.5038895010948181, "learning_rate": 0.0005, "loss": 1.419695496559143, "step": 29803 }, { "epoch": 2.349546708711076, "grad_norm": 0.5508754253387451, "learning_rate": 0.0005, "loss": 1.6021473407745361, "step": 29804 }, { "epoch": 2.349625541978715, "grad_norm": 0.43484118580818176, "learning_rate": 0.0005, "loss": 1.3676137924194336, "step": 29805 }, { "epoch": 2.349704375246354, "grad_norm": 0.7565494775772095, "learning_rate": 0.0005, "loss": 1.552258849143982, "step": 29806 }, { "epoch": 2.349783208513993, "grad_norm": 0.5194353461265564, "learning_rate": 0.0005, "loss": 1.4606791734695435, "step": 29807 }, { "epoch": 2.349862041781632, "grad_norm": 0.5364132523536682, "learning_rate": 0.0005, "loss": 1.581895112991333, "step": 29808 }, { "epoch": 2.349940875049271, "grad_norm": 0.6638505458831787, "learning_rate": 0.0005, "loss": 1.2257665395736694, "step": 29809 }, { "epoch": 2.3500197083169097, "grad_norm": 0.41883015632629395, "learning_rate": 0.0005, "loss": 1.2862628698349, "step": 29810 }, { "epoch": 2.350098541584549, "grad_norm": 0.5894836187362671, "learning_rate": 0.0005, "loss": 1.4473545551300049, "step": 29811 }, { "epoch": 2.3501773748521875, "grad_norm": 0.4731278121471405, "learning_rate": 0.0005, "loss": 1.3653656244277954, "step": 29812 }, { "epoch": 2.3502562081198266, "grad_norm": 0.586151123046875, "learning_rate": 0.0005, "loss": 1.3808417320251465, "step": 29813 }, { "epoch": 2.3503350413874653, "grad_norm": 0.5631766319274902, "learning_rate": 0.0005, "loss": 1.452590823173523, "step": 29814 }, { "epoch": 2.3504138746551044, "grad_norm": 0.5188016891479492, "learning_rate": 0.0005, "loss": 1.2675740718841553, "step": 29815 }, { "epoch": 2.3504927079227436, "grad_norm": 0.4278404712677002, "learning_rate": 0.0005, "loss": 1.2551813125610352, "step": 29816 }, { "epoch": 2.3505715411903823, "grad_norm": 0.5567995309829712, "learning_rate": 0.0005, "loss": 1.473742961883545, "step": 29817 }, { "epoch": 2.3506503744580214, "grad_norm": 0.6140612363815308, "learning_rate": 0.0005, "loss": 1.2052123546600342, "step": 29818 }, { "epoch": 2.35072920772566, "grad_norm": 0.6379767060279846, "learning_rate": 0.0005, "loss": 1.3161542415618896, "step": 29819 }, { "epoch": 2.350808040993299, "grad_norm": 0.5795385837554932, "learning_rate": 0.0005, "loss": 1.3902734518051147, "step": 29820 }, { "epoch": 2.3508868742609383, "grad_norm": 0.43913713097572327, "learning_rate": 0.0005, "loss": 1.1400772333145142, "step": 29821 }, { "epoch": 2.350965707528577, "grad_norm": 0.4819330871105194, "learning_rate": 0.0005, "loss": 1.4118688106536865, "step": 29822 }, { "epoch": 2.351044540796216, "grad_norm": 0.49559518694877625, "learning_rate": 0.0005, "loss": 1.2361881732940674, "step": 29823 }, { "epoch": 2.351123374063855, "grad_norm": 0.544879674911499, "learning_rate": 0.0005, "loss": 1.570585012435913, "step": 29824 }, { "epoch": 2.351202207331494, "grad_norm": 0.9083262085914612, "learning_rate": 0.0005, "loss": 1.1695582866668701, "step": 29825 }, { "epoch": 2.3512810405991327, "grad_norm": 0.5157922506332397, "learning_rate": 0.0005, "loss": 1.2358708381652832, "step": 29826 }, { "epoch": 2.351359873866772, "grad_norm": 0.47001373767852783, "learning_rate": 0.0005, "loss": 1.370786428451538, "step": 29827 }, { "epoch": 2.3514387071344105, "grad_norm": 0.4833025336265564, "learning_rate": 0.0005, "loss": 1.3890902996063232, "step": 29828 }, { "epoch": 2.3515175404020496, "grad_norm": 0.5609699487686157, "learning_rate": 0.0005, "loss": 1.5688809156417847, "step": 29829 }, { "epoch": 2.3515963736696888, "grad_norm": 0.5570526719093323, "learning_rate": 0.0005, "loss": 1.2064547538757324, "step": 29830 }, { "epoch": 2.3516752069373275, "grad_norm": 1.0167076587677002, "learning_rate": 0.0005, "loss": 1.4829970598220825, "step": 29831 }, { "epoch": 2.3517540402049666, "grad_norm": 0.41844648122787476, "learning_rate": 0.0005, "loss": 1.405210256576538, "step": 29832 }, { "epoch": 2.3518328734726053, "grad_norm": 0.47143855690956116, "learning_rate": 0.0005, "loss": 1.3501613140106201, "step": 29833 }, { "epoch": 2.3519117067402444, "grad_norm": 0.5268496870994568, "learning_rate": 0.0005, "loss": 1.3035179376602173, "step": 29834 }, { "epoch": 2.3519905400078835, "grad_norm": 0.4971902668476105, "learning_rate": 0.0005, "loss": 1.1646974086761475, "step": 29835 }, { "epoch": 2.3520693732755222, "grad_norm": 0.6627982258796692, "learning_rate": 0.0005, "loss": 1.3116717338562012, "step": 29836 }, { "epoch": 2.3521482065431614, "grad_norm": 0.65299391746521, "learning_rate": 0.0005, "loss": 1.56675124168396, "step": 29837 }, { "epoch": 2.3522270398108, "grad_norm": 0.4248954653739929, "learning_rate": 0.0005, "loss": 1.1604795455932617, "step": 29838 }, { "epoch": 2.352305873078439, "grad_norm": 0.4798109233379364, "learning_rate": 0.0005, "loss": 1.1726986169815063, "step": 29839 }, { "epoch": 2.352384706346078, "grad_norm": 0.5242350101470947, "learning_rate": 0.0005, "loss": 1.205500602722168, "step": 29840 }, { "epoch": 2.352463539613717, "grad_norm": 0.5376061201095581, "learning_rate": 0.0005, "loss": 1.3206958770751953, "step": 29841 }, { "epoch": 2.3525423728813557, "grad_norm": 0.6301591396331787, "learning_rate": 0.0005, "loss": 1.453125, "step": 29842 }, { "epoch": 2.352621206148995, "grad_norm": 0.4340759515762329, "learning_rate": 0.0005, "loss": 1.1526826620101929, "step": 29843 }, { "epoch": 2.352700039416634, "grad_norm": 0.5482279658317566, "learning_rate": 0.0005, "loss": 1.3644938468933105, "step": 29844 }, { "epoch": 2.3527788726842727, "grad_norm": 0.8431817293167114, "learning_rate": 0.0005, "loss": 1.365450382232666, "step": 29845 }, { "epoch": 2.352857705951912, "grad_norm": 0.5081895589828491, "learning_rate": 0.0005, "loss": 1.282212734222412, "step": 29846 }, { "epoch": 2.3529365392195505, "grad_norm": 0.49616190791130066, "learning_rate": 0.0005, "loss": 1.1868888139724731, "step": 29847 }, { "epoch": 2.3530153724871896, "grad_norm": 0.4421338737010956, "learning_rate": 0.0005, "loss": 1.0616519451141357, "step": 29848 }, { "epoch": 2.3530942057548287, "grad_norm": 0.4650859534740448, "learning_rate": 0.0005, "loss": 1.194366693496704, "step": 29849 }, { "epoch": 2.3531730390224674, "grad_norm": 0.5605385899543762, "learning_rate": 0.0005, "loss": 1.1474733352661133, "step": 29850 }, { "epoch": 2.3532518722901066, "grad_norm": 0.5045424103736877, "learning_rate": 0.0005, "loss": 1.2644380331039429, "step": 29851 }, { "epoch": 2.3533307055577453, "grad_norm": 0.46529045701026917, "learning_rate": 0.0005, "loss": 1.3340669870376587, "step": 29852 }, { "epoch": 2.3534095388253844, "grad_norm": 0.5936150550842285, "learning_rate": 0.0005, "loss": 1.4268754720687866, "step": 29853 }, { "epoch": 2.353488372093023, "grad_norm": 0.5313565135002136, "learning_rate": 0.0005, "loss": 1.1492900848388672, "step": 29854 }, { "epoch": 2.353567205360662, "grad_norm": 0.46332892775535583, "learning_rate": 0.0005, "loss": 1.282096028327942, "step": 29855 }, { "epoch": 2.353646038628301, "grad_norm": 0.5017772316932678, "learning_rate": 0.0005, "loss": 1.2821483612060547, "step": 29856 }, { "epoch": 2.35372487189594, "grad_norm": 0.4822135269641876, "learning_rate": 0.0005, "loss": 1.2034270763397217, "step": 29857 }, { "epoch": 2.353803705163579, "grad_norm": 0.6434010863304138, "learning_rate": 0.0005, "loss": 1.429667353630066, "step": 29858 }, { "epoch": 2.353882538431218, "grad_norm": 0.6090368032455444, "learning_rate": 0.0005, "loss": 1.6983025074005127, "step": 29859 }, { "epoch": 2.353961371698857, "grad_norm": 0.4897746443748474, "learning_rate": 0.0005, "loss": 1.0997459888458252, "step": 29860 }, { "epoch": 2.3540402049664957, "grad_norm": 0.6743404865264893, "learning_rate": 0.0005, "loss": 1.5048177242279053, "step": 29861 }, { "epoch": 2.354119038234135, "grad_norm": 0.42197659611701965, "learning_rate": 0.0005, "loss": 1.1055253744125366, "step": 29862 }, { "epoch": 2.354197871501774, "grad_norm": 0.4771088659763336, "learning_rate": 0.0005, "loss": 1.2548534870147705, "step": 29863 }, { "epoch": 2.3542767047694126, "grad_norm": 0.4753413200378418, "learning_rate": 0.0005, "loss": 1.2740179300308228, "step": 29864 }, { "epoch": 2.3543555380370518, "grad_norm": 0.5555104613304138, "learning_rate": 0.0005, "loss": 1.3337171077728271, "step": 29865 }, { "epoch": 2.3544343713046905, "grad_norm": 0.5847665667533875, "learning_rate": 0.0005, "loss": 1.5780980587005615, "step": 29866 }, { "epoch": 2.3545132045723296, "grad_norm": 0.48071014881134033, "learning_rate": 0.0005, "loss": 1.3114397525787354, "step": 29867 }, { "epoch": 2.3545920378399683, "grad_norm": 0.5630660653114319, "learning_rate": 0.0005, "loss": 1.262891173362732, "step": 29868 }, { "epoch": 2.3546708711076074, "grad_norm": 0.5310851335525513, "learning_rate": 0.0005, "loss": 1.336460828781128, "step": 29869 }, { "epoch": 2.354749704375246, "grad_norm": 0.4826541244983673, "learning_rate": 0.0005, "loss": 1.2607030868530273, "step": 29870 }, { "epoch": 2.3548285376428852, "grad_norm": 0.5007094740867615, "learning_rate": 0.0005, "loss": 1.3022308349609375, "step": 29871 }, { "epoch": 2.3549073709105244, "grad_norm": 0.48155829310417175, "learning_rate": 0.0005, "loss": 1.2086113691329956, "step": 29872 }, { "epoch": 2.354986204178163, "grad_norm": 0.5194251537322998, "learning_rate": 0.0005, "loss": 1.3198559284210205, "step": 29873 }, { "epoch": 2.355065037445802, "grad_norm": 0.6017544269561768, "learning_rate": 0.0005, "loss": 1.4343417882919312, "step": 29874 }, { "epoch": 2.355143870713441, "grad_norm": 0.5224237442016602, "learning_rate": 0.0005, "loss": 1.225899577140808, "step": 29875 }, { "epoch": 2.35522270398108, "grad_norm": 0.6899163722991943, "learning_rate": 0.0005, "loss": 1.3634400367736816, "step": 29876 }, { "epoch": 2.355301537248719, "grad_norm": 0.5031598806381226, "learning_rate": 0.0005, "loss": 1.420494556427002, "step": 29877 }, { "epoch": 2.355380370516358, "grad_norm": 0.52874755859375, "learning_rate": 0.0005, "loss": 1.3195948600769043, "step": 29878 }, { "epoch": 2.355459203783997, "grad_norm": 0.5233909487724304, "learning_rate": 0.0005, "loss": 1.3289588689804077, "step": 29879 }, { "epoch": 2.3555380370516357, "grad_norm": 0.42870649695396423, "learning_rate": 0.0005, "loss": 1.254380702972412, "step": 29880 }, { "epoch": 2.355616870319275, "grad_norm": 0.43956321477890015, "learning_rate": 0.0005, "loss": 1.2803540229797363, "step": 29881 }, { "epoch": 2.3556957035869135, "grad_norm": 0.49758630990982056, "learning_rate": 0.0005, "loss": 1.3795613050460815, "step": 29882 }, { "epoch": 2.3557745368545526, "grad_norm": 0.4430769681930542, "learning_rate": 0.0005, "loss": 1.3073408603668213, "step": 29883 }, { "epoch": 2.3558533701221918, "grad_norm": 0.499165803194046, "learning_rate": 0.0005, "loss": 1.492053747177124, "step": 29884 }, { "epoch": 2.3559322033898304, "grad_norm": 0.45367559790611267, "learning_rate": 0.0005, "loss": 1.2607083320617676, "step": 29885 }, { "epoch": 2.3560110366574696, "grad_norm": 0.4375550448894501, "learning_rate": 0.0005, "loss": 1.2858456373214722, "step": 29886 }, { "epoch": 2.3560898699251083, "grad_norm": 0.585554838180542, "learning_rate": 0.0005, "loss": 1.387352466583252, "step": 29887 }, { "epoch": 2.3561687031927474, "grad_norm": 0.46629559993743896, "learning_rate": 0.0005, "loss": 1.3296964168548584, "step": 29888 }, { "epoch": 2.356247536460386, "grad_norm": 0.5616792440414429, "learning_rate": 0.0005, "loss": 1.3652023077011108, "step": 29889 }, { "epoch": 2.3563263697280252, "grad_norm": 0.5068426132202148, "learning_rate": 0.0005, "loss": 1.3282524347305298, "step": 29890 }, { "epoch": 2.3564052029956644, "grad_norm": 0.5389084815979004, "learning_rate": 0.0005, "loss": 1.358194351196289, "step": 29891 }, { "epoch": 2.356484036263303, "grad_norm": 0.5432423949241638, "learning_rate": 0.0005, "loss": 1.4367327690124512, "step": 29892 }, { "epoch": 2.356562869530942, "grad_norm": 0.5373194813728333, "learning_rate": 0.0005, "loss": 1.228715181350708, "step": 29893 }, { "epoch": 2.356641702798581, "grad_norm": 0.43364062905311584, "learning_rate": 0.0005, "loss": 1.1485729217529297, "step": 29894 }, { "epoch": 2.35672053606622, "grad_norm": 0.4649903178215027, "learning_rate": 0.0005, "loss": 1.4629485607147217, "step": 29895 }, { "epoch": 2.3567993693338587, "grad_norm": 0.6131722927093506, "learning_rate": 0.0005, "loss": 1.2556242942810059, "step": 29896 }, { "epoch": 2.356878202601498, "grad_norm": 0.5740959644317627, "learning_rate": 0.0005, "loss": 1.3444101810455322, "step": 29897 }, { "epoch": 2.356957035869137, "grad_norm": 0.4812450706958771, "learning_rate": 0.0005, "loss": 1.2205727100372314, "step": 29898 }, { "epoch": 2.3570358691367757, "grad_norm": 0.6883073449134827, "learning_rate": 0.0005, "loss": 1.5359046459197998, "step": 29899 }, { "epoch": 2.357114702404415, "grad_norm": 0.5131322145462036, "learning_rate": 0.0005, "loss": 1.261400580406189, "step": 29900 }, { "epoch": 2.3571935356720535, "grad_norm": 0.4786875247955322, "learning_rate": 0.0005, "loss": 1.2187222242355347, "step": 29901 }, { "epoch": 2.3572723689396926, "grad_norm": 0.4807109236717224, "learning_rate": 0.0005, "loss": 1.3178564310073853, "step": 29902 }, { "epoch": 2.3573512022073313, "grad_norm": 0.4691309630870819, "learning_rate": 0.0005, "loss": 1.1933424472808838, "step": 29903 }, { "epoch": 2.3574300354749704, "grad_norm": 0.5229849815368652, "learning_rate": 0.0005, "loss": 1.3249733448028564, "step": 29904 }, { "epoch": 2.3575088687426096, "grad_norm": 0.5296149253845215, "learning_rate": 0.0005, "loss": 1.2238500118255615, "step": 29905 }, { "epoch": 2.3575877020102483, "grad_norm": 0.5132721066474915, "learning_rate": 0.0005, "loss": 1.4572277069091797, "step": 29906 }, { "epoch": 2.3576665352778874, "grad_norm": 0.4386815130710602, "learning_rate": 0.0005, "loss": 1.5179202556610107, "step": 29907 }, { "epoch": 2.357745368545526, "grad_norm": 0.4488986134529114, "learning_rate": 0.0005, "loss": 1.184701919555664, "step": 29908 }, { "epoch": 2.357824201813165, "grad_norm": 0.5165168642997742, "learning_rate": 0.0005, "loss": 1.459242820739746, "step": 29909 }, { "epoch": 2.3579030350808043, "grad_norm": 0.48362454771995544, "learning_rate": 0.0005, "loss": 1.3553942441940308, "step": 29910 }, { "epoch": 2.357981868348443, "grad_norm": 0.4303778409957886, "learning_rate": 0.0005, "loss": 1.2730450630187988, "step": 29911 }, { "epoch": 2.358060701616082, "grad_norm": 0.46712929010391235, "learning_rate": 0.0005, "loss": 1.3671252727508545, "step": 29912 }, { "epoch": 2.358139534883721, "grad_norm": 0.5872012376785278, "learning_rate": 0.0005, "loss": 1.4809129238128662, "step": 29913 }, { "epoch": 2.35821836815136, "grad_norm": 0.4204086363315582, "learning_rate": 0.0005, "loss": 1.2825955152511597, "step": 29914 }, { "epoch": 2.3582972014189987, "grad_norm": 0.4179697036743164, "learning_rate": 0.0005, "loss": 1.1655924320220947, "step": 29915 }, { "epoch": 2.358376034686638, "grad_norm": 0.3921860456466675, "learning_rate": 0.0005, "loss": 1.248918056488037, "step": 29916 }, { "epoch": 2.3584548679542765, "grad_norm": 0.5311183929443359, "learning_rate": 0.0005, "loss": 1.4382801055908203, "step": 29917 }, { "epoch": 2.3585337012219156, "grad_norm": 0.5183716416358948, "learning_rate": 0.0005, "loss": 1.613570213317871, "step": 29918 }, { "epoch": 2.3586125344895548, "grad_norm": 0.4273574948310852, "learning_rate": 0.0005, "loss": 1.4494761228561401, "step": 29919 }, { "epoch": 2.3586913677571935, "grad_norm": 0.4963262379169464, "learning_rate": 0.0005, "loss": 1.3602129220962524, "step": 29920 }, { "epoch": 2.3587702010248326, "grad_norm": 0.47829678654670715, "learning_rate": 0.0005, "loss": 1.2268917560577393, "step": 29921 }, { "epoch": 2.3588490342924713, "grad_norm": 0.623970627784729, "learning_rate": 0.0005, "loss": 1.3337407112121582, "step": 29922 }, { "epoch": 2.3589278675601104, "grad_norm": 0.43453580141067505, "learning_rate": 0.0005, "loss": 1.375584363937378, "step": 29923 }, { "epoch": 2.3590067008277495, "grad_norm": 0.5547788143157959, "learning_rate": 0.0005, "loss": 1.3037229776382446, "step": 29924 }, { "epoch": 2.3590855340953882, "grad_norm": 0.44755786657333374, "learning_rate": 0.0005, "loss": 1.1262303590774536, "step": 29925 }, { "epoch": 2.3591643673630274, "grad_norm": 0.5202330350875854, "learning_rate": 0.0005, "loss": 1.327998399734497, "step": 29926 }, { "epoch": 2.359243200630666, "grad_norm": 0.5256668329238892, "learning_rate": 0.0005, "loss": 1.3322362899780273, "step": 29927 }, { "epoch": 2.359322033898305, "grad_norm": 0.39427998661994934, "learning_rate": 0.0005, "loss": 1.27833890914917, "step": 29928 }, { "epoch": 2.359400867165944, "grad_norm": 0.5345752835273743, "learning_rate": 0.0005, "loss": 1.2964625358581543, "step": 29929 }, { "epoch": 2.359479700433583, "grad_norm": 0.4915303587913513, "learning_rate": 0.0005, "loss": 1.2849602699279785, "step": 29930 }, { "epoch": 2.3595585337012217, "grad_norm": 0.5098348259925842, "learning_rate": 0.0005, "loss": 1.3879209756851196, "step": 29931 }, { "epoch": 2.359637366968861, "grad_norm": 0.4803912937641144, "learning_rate": 0.0005, "loss": 1.4337036609649658, "step": 29932 }, { "epoch": 2.3597162002365, "grad_norm": 0.4714362323284149, "learning_rate": 0.0005, "loss": 1.2516206502914429, "step": 29933 }, { "epoch": 2.3597950335041387, "grad_norm": 0.8196424841880798, "learning_rate": 0.0005, "loss": 1.3290908336639404, "step": 29934 }, { "epoch": 2.359873866771778, "grad_norm": 0.6239068508148193, "learning_rate": 0.0005, "loss": 1.2903478145599365, "step": 29935 }, { "epoch": 2.3599527000394165, "grad_norm": 0.6251016855239868, "learning_rate": 0.0005, "loss": 1.366180419921875, "step": 29936 }, { "epoch": 2.3600315333070556, "grad_norm": 0.5311625003814697, "learning_rate": 0.0005, "loss": 1.47716224193573, "step": 29937 }, { "epoch": 2.3601103665746948, "grad_norm": 0.4916900098323822, "learning_rate": 0.0005, "loss": 1.3386956453323364, "step": 29938 }, { "epoch": 2.3601891998423334, "grad_norm": 0.5154948830604553, "learning_rate": 0.0005, "loss": 1.3215432167053223, "step": 29939 }, { "epoch": 2.3602680331099726, "grad_norm": 0.45503464341163635, "learning_rate": 0.0005, "loss": 1.2543482780456543, "step": 29940 }, { "epoch": 2.3603468663776113, "grad_norm": 0.6515610814094543, "learning_rate": 0.0005, "loss": 1.641869306564331, "step": 29941 }, { "epoch": 2.3604256996452504, "grad_norm": 0.536123514175415, "learning_rate": 0.0005, "loss": 1.3362747430801392, "step": 29942 }, { "epoch": 2.360504532912889, "grad_norm": 0.46552711725234985, "learning_rate": 0.0005, "loss": 1.2765291929244995, "step": 29943 }, { "epoch": 2.360583366180528, "grad_norm": 1.557983636856079, "learning_rate": 0.0005, "loss": 1.373271107673645, "step": 29944 }, { "epoch": 2.360662199448167, "grad_norm": 0.4421885013580322, "learning_rate": 0.0005, "loss": 1.3067659139633179, "step": 29945 }, { "epoch": 2.360741032715806, "grad_norm": 0.4420689642429352, "learning_rate": 0.0005, "loss": 1.2687644958496094, "step": 29946 }, { "epoch": 2.360819865983445, "grad_norm": 0.6158350110054016, "learning_rate": 0.0005, "loss": 1.3874590396881104, "step": 29947 }, { "epoch": 2.360898699251084, "grad_norm": 0.6674576997756958, "learning_rate": 0.0005, "loss": 1.1589469909667969, "step": 29948 }, { "epoch": 2.360977532518723, "grad_norm": 0.4377785325050354, "learning_rate": 0.0005, "loss": 1.3086357116699219, "step": 29949 }, { "epoch": 2.3610563657863617, "grad_norm": 0.7119031548500061, "learning_rate": 0.0005, "loss": 1.3978400230407715, "step": 29950 }, { "epoch": 2.361135199054001, "grad_norm": 0.5398812890052795, "learning_rate": 0.0005, "loss": 1.2859554290771484, "step": 29951 }, { "epoch": 2.36121403232164, "grad_norm": 0.5860132575035095, "learning_rate": 0.0005, "loss": 1.543198823928833, "step": 29952 }, { "epoch": 2.3612928655892786, "grad_norm": 0.6272806525230408, "learning_rate": 0.0005, "loss": 1.5637450218200684, "step": 29953 }, { "epoch": 2.3613716988569178, "grad_norm": 0.578426718711853, "learning_rate": 0.0005, "loss": 1.3249083757400513, "step": 29954 }, { "epoch": 2.3614505321245565, "grad_norm": 0.504083514213562, "learning_rate": 0.0005, "loss": 1.53275465965271, "step": 29955 }, { "epoch": 2.3615293653921956, "grad_norm": 0.5188315510749817, "learning_rate": 0.0005, "loss": 1.3980913162231445, "step": 29956 }, { "epoch": 2.3616081986598343, "grad_norm": 0.5124708414077759, "learning_rate": 0.0005, "loss": 1.4013054370880127, "step": 29957 }, { "epoch": 2.3616870319274734, "grad_norm": 0.46597978472709656, "learning_rate": 0.0005, "loss": 1.3262581825256348, "step": 29958 }, { "epoch": 2.361765865195112, "grad_norm": 0.5078048706054688, "learning_rate": 0.0005, "loss": 1.4119741916656494, "step": 29959 }, { "epoch": 2.3618446984627512, "grad_norm": 0.5313335061073303, "learning_rate": 0.0005, "loss": 1.2192041873931885, "step": 29960 }, { "epoch": 2.3619235317303904, "grad_norm": 0.4951128661632538, "learning_rate": 0.0005, "loss": 1.570597529411316, "step": 29961 }, { "epoch": 2.362002364998029, "grad_norm": 0.7915657758712769, "learning_rate": 0.0005, "loss": 1.329973816871643, "step": 29962 }, { "epoch": 2.362081198265668, "grad_norm": 0.5604677796363831, "learning_rate": 0.0005, "loss": 1.449352741241455, "step": 29963 }, { "epoch": 2.362160031533307, "grad_norm": 0.6647801399230957, "learning_rate": 0.0005, "loss": 1.3299099206924438, "step": 29964 }, { "epoch": 2.362238864800946, "grad_norm": 0.5249141454696655, "learning_rate": 0.0005, "loss": 1.2682933807373047, "step": 29965 }, { "epoch": 2.362317698068585, "grad_norm": 0.5139307975769043, "learning_rate": 0.0005, "loss": 1.5369261503219604, "step": 29966 }, { "epoch": 2.362396531336224, "grad_norm": 0.45602670311927795, "learning_rate": 0.0005, "loss": 1.1968605518341064, "step": 29967 }, { "epoch": 2.362475364603863, "grad_norm": 0.6139630079269409, "learning_rate": 0.0005, "loss": 1.4286826848983765, "step": 29968 }, { "epoch": 2.3625541978715017, "grad_norm": 0.4866258502006531, "learning_rate": 0.0005, "loss": 1.1764906644821167, "step": 29969 }, { "epoch": 2.362633031139141, "grad_norm": 0.4530017375946045, "learning_rate": 0.0005, "loss": 1.1415886878967285, "step": 29970 }, { "epoch": 2.3627118644067795, "grad_norm": 0.4896107316017151, "learning_rate": 0.0005, "loss": 1.2185372114181519, "step": 29971 }, { "epoch": 2.3627906976744186, "grad_norm": 0.4553069770336151, "learning_rate": 0.0005, "loss": 1.4482606649398804, "step": 29972 }, { "epoch": 2.3628695309420573, "grad_norm": 0.4454474449157715, "learning_rate": 0.0005, "loss": 1.197018027305603, "step": 29973 }, { "epoch": 2.3629483642096965, "grad_norm": 0.6800053119659424, "learning_rate": 0.0005, "loss": 1.4708077907562256, "step": 29974 }, { "epoch": 2.3630271974773356, "grad_norm": 0.424959272146225, "learning_rate": 0.0005, "loss": 1.1332645416259766, "step": 29975 }, { "epoch": 2.3631060307449743, "grad_norm": 0.462401419878006, "learning_rate": 0.0005, "loss": 1.3826245069503784, "step": 29976 }, { "epoch": 2.3631848640126134, "grad_norm": 0.444868266582489, "learning_rate": 0.0005, "loss": 1.1411446332931519, "step": 29977 }, { "epoch": 2.363263697280252, "grad_norm": 0.5062633752822876, "learning_rate": 0.0005, "loss": 1.1553722620010376, "step": 29978 }, { "epoch": 2.3633425305478912, "grad_norm": 0.6249628067016602, "learning_rate": 0.0005, "loss": 1.516717791557312, "step": 29979 }, { "epoch": 2.3634213638155304, "grad_norm": 0.44154423475265503, "learning_rate": 0.0005, "loss": 1.2983708381652832, "step": 29980 }, { "epoch": 2.363500197083169, "grad_norm": 0.5822848081588745, "learning_rate": 0.0005, "loss": 1.3606245517730713, "step": 29981 }, { "epoch": 2.363579030350808, "grad_norm": 0.814593493938446, "learning_rate": 0.0005, "loss": 1.6079744100570679, "step": 29982 }, { "epoch": 2.363657863618447, "grad_norm": 0.5397751331329346, "learning_rate": 0.0005, "loss": 1.4836218357086182, "step": 29983 }, { "epoch": 2.363736696886086, "grad_norm": 0.5005542635917664, "learning_rate": 0.0005, "loss": 1.286078691482544, "step": 29984 }, { "epoch": 2.3638155301537247, "grad_norm": 0.5944780707359314, "learning_rate": 0.0005, "loss": 1.4519468545913696, "step": 29985 }, { "epoch": 2.363894363421364, "grad_norm": 0.5822734832763672, "learning_rate": 0.0005, "loss": 1.4887441396713257, "step": 29986 }, { "epoch": 2.3639731966890025, "grad_norm": 0.4172214865684509, "learning_rate": 0.0005, "loss": 1.1794251203536987, "step": 29987 }, { "epoch": 2.3640520299566417, "grad_norm": 0.41786131262779236, "learning_rate": 0.0005, "loss": 1.3869545459747314, "step": 29988 }, { "epoch": 2.364130863224281, "grad_norm": 0.5235735177993774, "learning_rate": 0.0005, "loss": 1.5216524600982666, "step": 29989 }, { "epoch": 2.3642096964919195, "grad_norm": 0.4932887554168701, "learning_rate": 0.0005, "loss": 1.3707466125488281, "step": 29990 }, { "epoch": 2.3642885297595586, "grad_norm": 0.5548262596130371, "learning_rate": 0.0005, "loss": 1.259469747543335, "step": 29991 }, { "epoch": 2.3643673630271973, "grad_norm": 0.482637494802475, "learning_rate": 0.0005, "loss": 1.1948281526565552, "step": 29992 }, { "epoch": 2.3644461962948364, "grad_norm": 0.5019259452819824, "learning_rate": 0.0005, "loss": 1.2427693605422974, "step": 29993 }, { "epoch": 2.3645250295624756, "grad_norm": 0.4115406572818756, "learning_rate": 0.0005, "loss": 1.3163872957229614, "step": 29994 }, { "epoch": 2.3646038628301143, "grad_norm": 0.4768030345439911, "learning_rate": 0.0005, "loss": 1.2166938781738281, "step": 29995 }, { "epoch": 2.3646826960977534, "grad_norm": 0.44638073444366455, "learning_rate": 0.0005, "loss": 1.1009962558746338, "step": 29996 }, { "epoch": 2.364761529365392, "grad_norm": 0.41028517484664917, "learning_rate": 0.0005, "loss": 1.2891968488693237, "step": 29997 }, { "epoch": 2.364840362633031, "grad_norm": 0.5259130597114563, "learning_rate": 0.0005, "loss": 1.298140287399292, "step": 29998 }, { "epoch": 2.36491919590067, "grad_norm": 0.4367321729660034, "learning_rate": 0.0005, "loss": 1.2674890756607056, "step": 29999 }, { "epoch": 2.364998029168309, "grad_norm": 0.3657588064670563, "learning_rate": 0.0005, "loss": 1.197331190109253, "step": 30000 }, { "epoch": 2.3650768624359477, "grad_norm": 0.5477893352508545, "learning_rate": 0.0005, "loss": 1.3763580322265625, "step": 30001 }, { "epoch": 2.365155695703587, "grad_norm": 0.4548892080783844, "learning_rate": 0.0005, "loss": 1.2553372383117676, "step": 30002 }, { "epoch": 2.365234528971226, "grad_norm": 0.5779412388801575, "learning_rate": 0.0005, "loss": 1.3359332084655762, "step": 30003 }, { "epoch": 2.3653133622388647, "grad_norm": 0.5327569246292114, "learning_rate": 0.0005, "loss": 1.414262056350708, "step": 30004 }, { "epoch": 2.365392195506504, "grad_norm": 0.3950054347515106, "learning_rate": 0.0005, "loss": 1.3119817972183228, "step": 30005 }, { "epoch": 2.3654710287741425, "grad_norm": 0.5430753827095032, "learning_rate": 0.0005, "loss": 1.458243727684021, "step": 30006 }, { "epoch": 2.3655498620417816, "grad_norm": 0.5170108675956726, "learning_rate": 0.0005, "loss": 1.274674415588379, "step": 30007 }, { "epoch": 2.3656286953094208, "grad_norm": 0.531731903553009, "learning_rate": 0.0005, "loss": 1.4591610431671143, "step": 30008 }, { "epoch": 2.3657075285770595, "grad_norm": 0.551671028137207, "learning_rate": 0.0005, "loss": 1.366450309753418, "step": 30009 }, { "epoch": 2.3657863618446986, "grad_norm": 0.7006297707557678, "learning_rate": 0.0005, "loss": 1.462985634803772, "step": 30010 }, { "epoch": 2.3658651951123373, "grad_norm": 0.4855845272541046, "learning_rate": 0.0005, "loss": 1.3200838565826416, "step": 30011 }, { "epoch": 2.3659440283799764, "grad_norm": 0.5153099894523621, "learning_rate": 0.0005, "loss": 1.3784778118133545, "step": 30012 }, { "epoch": 2.366022861647615, "grad_norm": 0.43396279215812683, "learning_rate": 0.0005, "loss": 1.2015788555145264, "step": 30013 }, { "epoch": 2.3661016949152542, "grad_norm": 0.4460105895996094, "learning_rate": 0.0005, "loss": 1.194399356842041, "step": 30014 }, { "epoch": 2.3661805281828934, "grad_norm": 0.4896378219127655, "learning_rate": 0.0005, "loss": 1.245424747467041, "step": 30015 }, { "epoch": 2.366259361450532, "grad_norm": 0.494023859500885, "learning_rate": 0.0005, "loss": 1.1731886863708496, "step": 30016 }, { "epoch": 2.366338194718171, "grad_norm": 0.6923179626464844, "learning_rate": 0.0005, "loss": 1.1823501586914062, "step": 30017 }, { "epoch": 2.36641702798581, "grad_norm": 0.5072858333587646, "learning_rate": 0.0005, "loss": 1.421388030052185, "step": 30018 }, { "epoch": 2.366495861253449, "grad_norm": 0.46174129843711853, "learning_rate": 0.0005, "loss": 1.1811946630477905, "step": 30019 }, { "epoch": 2.3665746945210877, "grad_norm": 0.47709691524505615, "learning_rate": 0.0005, "loss": 1.45963454246521, "step": 30020 }, { "epoch": 2.366653527788727, "grad_norm": 0.9499860405921936, "learning_rate": 0.0005, "loss": 1.3608123064041138, "step": 30021 }, { "epoch": 2.366732361056366, "grad_norm": 0.4240455627441406, "learning_rate": 0.0005, "loss": 1.277770757675171, "step": 30022 }, { "epoch": 2.3668111943240047, "grad_norm": 0.5114375948905945, "learning_rate": 0.0005, "loss": 1.4085817337036133, "step": 30023 }, { "epoch": 2.366890027591644, "grad_norm": 0.5606784224510193, "learning_rate": 0.0005, "loss": 1.2143006324768066, "step": 30024 }, { "epoch": 2.3669688608592825, "grad_norm": 0.5478686690330505, "learning_rate": 0.0005, "loss": 1.1790274381637573, "step": 30025 }, { "epoch": 2.3670476941269216, "grad_norm": 0.6187275052070618, "learning_rate": 0.0005, "loss": 1.3291308879852295, "step": 30026 }, { "epoch": 2.3671265273945608, "grad_norm": 0.40594762563705444, "learning_rate": 0.0005, "loss": 1.3247573375701904, "step": 30027 }, { "epoch": 2.3672053606621994, "grad_norm": 0.5128663182258606, "learning_rate": 0.0005, "loss": 1.1473400592803955, "step": 30028 }, { "epoch": 2.3672841939298386, "grad_norm": 0.4718889892101288, "learning_rate": 0.0005, "loss": 1.2167441844940186, "step": 30029 }, { "epoch": 2.3673630271974773, "grad_norm": 0.5668595433235168, "learning_rate": 0.0005, "loss": 1.4548096656799316, "step": 30030 }, { "epoch": 2.3674418604651164, "grad_norm": 0.5514861941337585, "learning_rate": 0.0005, "loss": 1.2611370086669922, "step": 30031 }, { "epoch": 2.367520693732755, "grad_norm": 0.5327038168907166, "learning_rate": 0.0005, "loss": 1.4934786558151245, "step": 30032 }, { "epoch": 2.367599527000394, "grad_norm": 0.8250502943992615, "learning_rate": 0.0005, "loss": 1.2391244173049927, "step": 30033 }, { "epoch": 2.367678360268033, "grad_norm": 0.47305822372436523, "learning_rate": 0.0005, "loss": 1.4085944890975952, "step": 30034 }, { "epoch": 2.367757193535672, "grad_norm": 0.4213545322418213, "learning_rate": 0.0005, "loss": 1.1977511644363403, "step": 30035 }, { "epoch": 2.367836026803311, "grad_norm": 0.4552976191043854, "learning_rate": 0.0005, "loss": 1.2374300956726074, "step": 30036 }, { "epoch": 2.36791486007095, "grad_norm": 0.4670136570930481, "learning_rate": 0.0005, "loss": 1.2117431163787842, "step": 30037 }, { "epoch": 2.367993693338589, "grad_norm": 0.4383305013179779, "learning_rate": 0.0005, "loss": 1.290583848953247, "step": 30038 }, { "epoch": 2.3680725266062277, "grad_norm": 0.6682792901992798, "learning_rate": 0.0005, "loss": 1.706153392791748, "step": 30039 }, { "epoch": 2.368151359873867, "grad_norm": 0.49636632204055786, "learning_rate": 0.0005, "loss": 1.4058195352554321, "step": 30040 }, { "epoch": 2.368230193141506, "grad_norm": 0.5099305510520935, "learning_rate": 0.0005, "loss": 1.6306829452514648, "step": 30041 }, { "epoch": 2.3683090264091446, "grad_norm": 0.5023711919784546, "learning_rate": 0.0005, "loss": 1.2696104049682617, "step": 30042 }, { "epoch": 2.368387859676784, "grad_norm": 0.5041861534118652, "learning_rate": 0.0005, "loss": 1.409595251083374, "step": 30043 }, { "epoch": 2.3684666929444225, "grad_norm": 0.6428308486938477, "learning_rate": 0.0005, "loss": 1.1702420711517334, "step": 30044 }, { "epoch": 2.3685455262120616, "grad_norm": 0.5098201632499695, "learning_rate": 0.0005, "loss": 1.137412428855896, "step": 30045 }, { "epoch": 2.3686243594797003, "grad_norm": 0.44393521547317505, "learning_rate": 0.0005, "loss": 1.3804222345352173, "step": 30046 }, { "epoch": 2.3687031927473394, "grad_norm": 0.6177316308021545, "learning_rate": 0.0005, "loss": 1.5806536674499512, "step": 30047 }, { "epoch": 2.368782026014978, "grad_norm": 0.4462162256240845, "learning_rate": 0.0005, "loss": 1.2496337890625, "step": 30048 }, { "epoch": 2.3688608592826172, "grad_norm": 0.4841357171535492, "learning_rate": 0.0005, "loss": 1.2938324213027954, "step": 30049 }, { "epoch": 2.3689396925502564, "grad_norm": 0.48967066407203674, "learning_rate": 0.0005, "loss": 1.3177672624588013, "step": 30050 }, { "epoch": 2.369018525817895, "grad_norm": 0.5396003127098083, "learning_rate": 0.0005, "loss": 1.4204387664794922, "step": 30051 }, { "epoch": 2.369097359085534, "grad_norm": 0.5288501977920532, "learning_rate": 0.0005, "loss": 1.2373483180999756, "step": 30052 }, { "epoch": 2.369176192353173, "grad_norm": 0.637921929359436, "learning_rate": 0.0005, "loss": 1.3425054550170898, "step": 30053 }, { "epoch": 2.369255025620812, "grad_norm": 0.39823973178863525, "learning_rate": 0.0005, "loss": 1.2888462543487549, "step": 30054 }, { "epoch": 2.369333858888451, "grad_norm": 0.4971200227737427, "learning_rate": 0.0005, "loss": 1.4615510702133179, "step": 30055 }, { "epoch": 2.36941269215609, "grad_norm": 0.5605071783065796, "learning_rate": 0.0005, "loss": 1.3879655599594116, "step": 30056 }, { "epoch": 2.369491525423729, "grad_norm": 0.4939003586769104, "learning_rate": 0.0005, "loss": 1.288770318031311, "step": 30057 }, { "epoch": 2.3695703586913677, "grad_norm": 0.4651833772659302, "learning_rate": 0.0005, "loss": 1.2312414646148682, "step": 30058 }, { "epoch": 2.369649191959007, "grad_norm": 0.45238903164863586, "learning_rate": 0.0005, "loss": 1.3700664043426514, "step": 30059 }, { "epoch": 2.3697280252266455, "grad_norm": 0.4860215187072754, "learning_rate": 0.0005, "loss": 1.4753379821777344, "step": 30060 }, { "epoch": 2.3698068584942846, "grad_norm": 0.5657784342765808, "learning_rate": 0.0005, "loss": 1.403398036956787, "step": 30061 }, { "epoch": 2.3698856917619233, "grad_norm": 0.46784549951553345, "learning_rate": 0.0005, "loss": 1.1756377220153809, "step": 30062 }, { "epoch": 2.3699645250295625, "grad_norm": 0.5207504034042358, "learning_rate": 0.0005, "loss": 1.1912845373153687, "step": 30063 }, { "epoch": 2.3700433582972016, "grad_norm": 0.5127072930335999, "learning_rate": 0.0005, "loss": 1.3192956447601318, "step": 30064 }, { "epoch": 2.3701221915648403, "grad_norm": 0.4624457061290741, "learning_rate": 0.0005, "loss": 1.3583805561065674, "step": 30065 }, { "epoch": 2.3702010248324794, "grad_norm": 0.5696218609809875, "learning_rate": 0.0005, "loss": 1.6203581094741821, "step": 30066 }, { "epoch": 2.370279858100118, "grad_norm": 0.5203738212585449, "learning_rate": 0.0005, "loss": 1.52559494972229, "step": 30067 }, { "epoch": 2.3703586913677572, "grad_norm": 0.837454080581665, "learning_rate": 0.0005, "loss": 1.324170470237732, "step": 30068 }, { "epoch": 2.3704375246353964, "grad_norm": 0.5059599280357361, "learning_rate": 0.0005, "loss": 1.3630688190460205, "step": 30069 }, { "epoch": 2.370516357903035, "grad_norm": 0.5411688089370728, "learning_rate": 0.0005, "loss": 1.306419014930725, "step": 30070 }, { "epoch": 2.370595191170674, "grad_norm": 0.6708377003669739, "learning_rate": 0.0005, "loss": 1.4204609394073486, "step": 30071 }, { "epoch": 2.370674024438313, "grad_norm": 0.47135916352272034, "learning_rate": 0.0005, "loss": 1.239159345626831, "step": 30072 }, { "epoch": 2.370752857705952, "grad_norm": 0.5491563677787781, "learning_rate": 0.0005, "loss": 1.227463722229004, "step": 30073 }, { "epoch": 2.3708316909735907, "grad_norm": 0.5396187901496887, "learning_rate": 0.0005, "loss": 1.4274914264678955, "step": 30074 }, { "epoch": 2.37091052424123, "grad_norm": 0.4674709439277649, "learning_rate": 0.0005, "loss": 1.3106472492218018, "step": 30075 }, { "epoch": 2.3709893575088685, "grad_norm": 0.5451788902282715, "learning_rate": 0.0005, "loss": 1.143510103225708, "step": 30076 }, { "epoch": 2.3710681907765077, "grad_norm": 0.5042402148246765, "learning_rate": 0.0005, "loss": 1.4844955205917358, "step": 30077 }, { "epoch": 2.371147024044147, "grad_norm": 0.5195039510726929, "learning_rate": 0.0005, "loss": 1.3984520435333252, "step": 30078 }, { "epoch": 2.3712258573117855, "grad_norm": 0.42944657802581787, "learning_rate": 0.0005, "loss": 1.5208642482757568, "step": 30079 }, { "epoch": 2.3713046905794246, "grad_norm": 0.5781610608100891, "learning_rate": 0.0005, "loss": 1.493553876876831, "step": 30080 }, { "epoch": 2.3713835238470633, "grad_norm": 0.6429010033607483, "learning_rate": 0.0005, "loss": 1.4504445791244507, "step": 30081 }, { "epoch": 2.3714623571147024, "grad_norm": 0.4935840666294098, "learning_rate": 0.0005, "loss": 1.4882423877716064, "step": 30082 }, { "epoch": 2.3715411903823416, "grad_norm": 0.4746541678905487, "learning_rate": 0.0005, "loss": 1.307631254196167, "step": 30083 }, { "epoch": 2.3716200236499803, "grad_norm": 0.4995463192462921, "learning_rate": 0.0005, "loss": 1.28371262550354, "step": 30084 }, { "epoch": 2.3716988569176194, "grad_norm": 0.4857633411884308, "learning_rate": 0.0005, "loss": 1.4469022750854492, "step": 30085 }, { "epoch": 2.371777690185258, "grad_norm": 0.5031493306159973, "learning_rate": 0.0005, "loss": 1.3904613256454468, "step": 30086 }, { "epoch": 2.371856523452897, "grad_norm": 0.4536799490451813, "learning_rate": 0.0005, "loss": 1.4454448223114014, "step": 30087 }, { "epoch": 2.371935356720536, "grad_norm": 0.5309269428253174, "learning_rate": 0.0005, "loss": 1.2669944763183594, "step": 30088 }, { "epoch": 2.372014189988175, "grad_norm": 0.4157131016254425, "learning_rate": 0.0005, "loss": 1.201611876487732, "step": 30089 }, { "epoch": 2.3720930232558137, "grad_norm": 0.4497331380844116, "learning_rate": 0.0005, "loss": 1.2892249822616577, "step": 30090 }, { "epoch": 2.372171856523453, "grad_norm": 0.467038094997406, "learning_rate": 0.0005, "loss": 1.457031011581421, "step": 30091 }, { "epoch": 2.372250689791092, "grad_norm": 0.47995179891586304, "learning_rate": 0.0005, "loss": 1.4565739631652832, "step": 30092 }, { "epoch": 2.3723295230587307, "grad_norm": 0.43792057037353516, "learning_rate": 0.0005, "loss": 1.3221482038497925, "step": 30093 }, { "epoch": 2.37240835632637, "grad_norm": 0.45676663517951965, "learning_rate": 0.0005, "loss": 1.40578031539917, "step": 30094 }, { "epoch": 2.3724871895940085, "grad_norm": 0.6933512687683105, "learning_rate": 0.0005, "loss": 1.433410406112671, "step": 30095 }, { "epoch": 2.3725660228616476, "grad_norm": 0.5001704096794128, "learning_rate": 0.0005, "loss": 1.2487035989761353, "step": 30096 }, { "epoch": 2.3726448561292868, "grad_norm": 0.4875575602054596, "learning_rate": 0.0005, "loss": 1.3255904912948608, "step": 30097 }, { "epoch": 2.3727236893969255, "grad_norm": 0.6441652774810791, "learning_rate": 0.0005, "loss": 1.2727208137512207, "step": 30098 }, { "epoch": 2.3728025226645646, "grad_norm": 3.0620851516723633, "learning_rate": 0.0005, "loss": 1.457458257675171, "step": 30099 }, { "epoch": 2.3728813559322033, "grad_norm": 0.5587818622589111, "learning_rate": 0.0005, "loss": 1.5628801584243774, "step": 30100 }, { "epoch": 2.3729601891998424, "grad_norm": 0.672290563583374, "learning_rate": 0.0005, "loss": 1.436650276184082, "step": 30101 }, { "epoch": 2.373039022467481, "grad_norm": 0.7361665368080139, "learning_rate": 0.0005, "loss": 1.6076841354370117, "step": 30102 }, { "epoch": 2.3731178557351202, "grad_norm": 0.5287460088729858, "learning_rate": 0.0005, "loss": 1.3075940608978271, "step": 30103 }, { "epoch": 2.373196689002759, "grad_norm": 0.47450757026672363, "learning_rate": 0.0005, "loss": 1.4662209749221802, "step": 30104 }, { "epoch": 2.373275522270398, "grad_norm": 0.5054431557655334, "learning_rate": 0.0005, "loss": 1.5029687881469727, "step": 30105 }, { "epoch": 2.373354355538037, "grad_norm": 0.47164320945739746, "learning_rate": 0.0005, "loss": 1.2386198043823242, "step": 30106 }, { "epoch": 2.373433188805676, "grad_norm": 0.44254645705223083, "learning_rate": 0.0005, "loss": 1.1935157775878906, "step": 30107 }, { "epoch": 2.373512022073315, "grad_norm": 0.4529789388179779, "learning_rate": 0.0005, "loss": 1.2149369716644287, "step": 30108 }, { "epoch": 2.3735908553409537, "grad_norm": 0.8672301769256592, "learning_rate": 0.0005, "loss": 1.6090586185455322, "step": 30109 }, { "epoch": 2.373669688608593, "grad_norm": 0.5378122329711914, "learning_rate": 0.0005, "loss": 1.2201387882232666, "step": 30110 }, { "epoch": 2.373748521876232, "grad_norm": 0.5482671856880188, "learning_rate": 0.0005, "loss": 1.3958531618118286, "step": 30111 }, { "epoch": 2.3738273551438707, "grad_norm": 0.4757297933101654, "learning_rate": 0.0005, "loss": 1.229067325592041, "step": 30112 }, { "epoch": 2.37390618841151, "grad_norm": 0.6289435029029846, "learning_rate": 0.0005, "loss": 1.2544509172439575, "step": 30113 }, { "epoch": 2.3739850216791485, "grad_norm": 0.6165823936462402, "learning_rate": 0.0005, "loss": 1.4592037200927734, "step": 30114 }, { "epoch": 2.3740638549467876, "grad_norm": 0.46911516785621643, "learning_rate": 0.0005, "loss": 1.3645575046539307, "step": 30115 }, { "epoch": 2.3741426882144263, "grad_norm": 0.5932166576385498, "learning_rate": 0.0005, "loss": 1.3734478950500488, "step": 30116 }, { "epoch": 2.3742215214820654, "grad_norm": 0.39914971590042114, "learning_rate": 0.0005, "loss": 1.086985468864441, "step": 30117 }, { "epoch": 2.374300354749704, "grad_norm": 0.4591579735279083, "learning_rate": 0.0005, "loss": 1.2167646884918213, "step": 30118 }, { "epoch": 2.3743791880173433, "grad_norm": 0.3735092580318451, "learning_rate": 0.0005, "loss": 1.2321876287460327, "step": 30119 }, { "epoch": 2.3744580212849824, "grad_norm": 0.570983350276947, "learning_rate": 0.0005, "loss": 1.2885477542877197, "step": 30120 }, { "epoch": 2.374536854552621, "grad_norm": 0.4966679513454437, "learning_rate": 0.0005, "loss": 1.4323501586914062, "step": 30121 }, { "epoch": 2.3746156878202602, "grad_norm": 0.4358145296573639, "learning_rate": 0.0005, "loss": 1.2732652425765991, "step": 30122 }, { "epoch": 2.374694521087899, "grad_norm": 0.4769895076751709, "learning_rate": 0.0005, "loss": 1.3219964504241943, "step": 30123 }, { "epoch": 2.374773354355538, "grad_norm": 0.5872752070426941, "learning_rate": 0.0005, "loss": 1.3430187702178955, "step": 30124 }, { "epoch": 2.374852187623177, "grad_norm": 0.5585855841636658, "learning_rate": 0.0005, "loss": 1.631681203842163, "step": 30125 }, { "epoch": 2.374931020890816, "grad_norm": 0.49197667837142944, "learning_rate": 0.0005, "loss": 1.1806745529174805, "step": 30126 }, { "epoch": 2.375009854158455, "grad_norm": 0.5296902060508728, "learning_rate": 0.0005, "loss": 1.1829370260238647, "step": 30127 }, { "epoch": 2.3750886874260937, "grad_norm": 0.47298797965049744, "learning_rate": 0.0005, "loss": 1.2485405206680298, "step": 30128 }, { "epoch": 2.375167520693733, "grad_norm": 0.6482140421867371, "learning_rate": 0.0005, "loss": 1.3441591262817383, "step": 30129 }, { "epoch": 2.3752463539613715, "grad_norm": 0.5139638781547546, "learning_rate": 0.0005, "loss": 1.378369927406311, "step": 30130 }, { "epoch": 2.3753251872290106, "grad_norm": 0.5226482152938843, "learning_rate": 0.0005, "loss": 1.3056434392929077, "step": 30131 }, { "epoch": 2.3754040204966493, "grad_norm": 0.9949754476547241, "learning_rate": 0.0005, "loss": 1.3786439895629883, "step": 30132 }, { "epoch": 2.3754828537642885, "grad_norm": 0.4893180727958679, "learning_rate": 0.0005, "loss": 1.3137400150299072, "step": 30133 }, { "epoch": 2.3755616870319276, "grad_norm": 0.4830407202243805, "learning_rate": 0.0005, "loss": 1.30782151222229, "step": 30134 }, { "epoch": 2.3756405202995663, "grad_norm": 0.7905213832855225, "learning_rate": 0.0005, "loss": 1.4041810035705566, "step": 30135 }, { "epoch": 2.3757193535672054, "grad_norm": 0.4847046136856079, "learning_rate": 0.0005, "loss": 1.3313769102096558, "step": 30136 }, { "epoch": 2.375798186834844, "grad_norm": 0.4009522795677185, "learning_rate": 0.0005, "loss": 1.1980183124542236, "step": 30137 }, { "epoch": 2.3758770201024833, "grad_norm": 0.49673619866371155, "learning_rate": 0.0005, "loss": 1.1109529733657837, "step": 30138 }, { "epoch": 2.3759558533701224, "grad_norm": 0.5287347435951233, "learning_rate": 0.0005, "loss": 1.3465780019760132, "step": 30139 }, { "epoch": 2.376034686637761, "grad_norm": 0.6220276951789856, "learning_rate": 0.0005, "loss": 1.4257547855377197, "step": 30140 }, { "epoch": 2.3761135199054, "grad_norm": 0.4068455100059509, "learning_rate": 0.0005, "loss": 1.167715311050415, "step": 30141 }, { "epoch": 2.376192353173039, "grad_norm": 0.45214831829071045, "learning_rate": 0.0005, "loss": 1.3694944381713867, "step": 30142 }, { "epoch": 2.376271186440678, "grad_norm": 0.5772183537483215, "learning_rate": 0.0005, "loss": 1.386090636253357, "step": 30143 }, { "epoch": 2.3763500197083167, "grad_norm": 0.5050861835479736, "learning_rate": 0.0005, "loss": 1.2156574726104736, "step": 30144 }, { "epoch": 2.376428852975956, "grad_norm": 0.5303711891174316, "learning_rate": 0.0005, "loss": 1.2668242454528809, "step": 30145 }, { "epoch": 2.376507686243595, "grad_norm": 0.4232729375362396, "learning_rate": 0.0005, "loss": 1.1076840162277222, "step": 30146 }, { "epoch": 2.3765865195112337, "grad_norm": 0.397628515958786, "learning_rate": 0.0005, "loss": 1.161267876625061, "step": 30147 }, { "epoch": 2.376665352778873, "grad_norm": 0.7181816697120667, "learning_rate": 0.0005, "loss": 1.2057738304138184, "step": 30148 }, { "epoch": 2.3767441860465115, "grad_norm": 0.5229585766792297, "learning_rate": 0.0005, "loss": 1.574697494506836, "step": 30149 }, { "epoch": 2.3768230193141506, "grad_norm": 0.5511987805366516, "learning_rate": 0.0005, "loss": 1.5227470397949219, "step": 30150 }, { "epoch": 2.3769018525817893, "grad_norm": 0.560058057308197, "learning_rate": 0.0005, "loss": 1.5582488775253296, "step": 30151 }, { "epoch": 2.3769806858494285, "grad_norm": 0.5271748304367065, "learning_rate": 0.0005, "loss": 1.602939486503601, "step": 30152 }, { "epoch": 2.3770595191170676, "grad_norm": 0.6608428359031677, "learning_rate": 0.0005, "loss": 1.2380093336105347, "step": 30153 }, { "epoch": 2.3771383523847063, "grad_norm": 0.5213642120361328, "learning_rate": 0.0005, "loss": 1.4282877445220947, "step": 30154 }, { "epoch": 2.3772171856523454, "grad_norm": 0.4387850761413574, "learning_rate": 0.0005, "loss": 1.239823818206787, "step": 30155 }, { "epoch": 2.377296018919984, "grad_norm": 0.4436415731906891, "learning_rate": 0.0005, "loss": 1.375930905342102, "step": 30156 }, { "epoch": 2.3773748521876232, "grad_norm": 0.5134716033935547, "learning_rate": 0.0005, "loss": 1.312257170677185, "step": 30157 }, { "epoch": 2.3774536854552624, "grad_norm": 0.5445421934127808, "learning_rate": 0.0005, "loss": 1.4398744106292725, "step": 30158 }, { "epoch": 2.377532518722901, "grad_norm": 0.48500674962997437, "learning_rate": 0.0005, "loss": 1.2609920501708984, "step": 30159 }, { "epoch": 2.37761135199054, "grad_norm": 0.528134286403656, "learning_rate": 0.0005, "loss": 1.365997552871704, "step": 30160 }, { "epoch": 2.377690185258179, "grad_norm": 0.5388025641441345, "learning_rate": 0.0005, "loss": 1.4847981929779053, "step": 30161 }, { "epoch": 2.377769018525818, "grad_norm": 0.5135522484779358, "learning_rate": 0.0005, "loss": 1.3640722036361694, "step": 30162 }, { "epoch": 2.3778478517934567, "grad_norm": 0.4956933557987213, "learning_rate": 0.0005, "loss": 1.239133596420288, "step": 30163 }, { "epoch": 2.377926685061096, "grad_norm": 0.6056497097015381, "learning_rate": 0.0005, "loss": 1.4052340984344482, "step": 30164 }, { "epoch": 2.3780055183287345, "grad_norm": 0.4900031089782715, "learning_rate": 0.0005, "loss": 1.4154633283615112, "step": 30165 }, { "epoch": 2.3780843515963737, "grad_norm": 0.5468872785568237, "learning_rate": 0.0005, "loss": 1.4402134418487549, "step": 30166 }, { "epoch": 2.378163184864013, "grad_norm": 1.3147859573364258, "learning_rate": 0.0005, "loss": 1.2699470520019531, "step": 30167 }, { "epoch": 2.3782420181316515, "grad_norm": 0.49649059772491455, "learning_rate": 0.0005, "loss": 1.367445468902588, "step": 30168 }, { "epoch": 2.3783208513992906, "grad_norm": 0.47643810510635376, "learning_rate": 0.0005, "loss": 1.4869654178619385, "step": 30169 }, { "epoch": 2.3783996846669293, "grad_norm": 0.6002255082130432, "learning_rate": 0.0005, "loss": 1.245107650756836, "step": 30170 }, { "epoch": 2.3784785179345684, "grad_norm": 0.47893279790878296, "learning_rate": 0.0005, "loss": 1.1482570171356201, "step": 30171 }, { "epoch": 2.3785573512022076, "grad_norm": 0.496488481760025, "learning_rate": 0.0005, "loss": 1.26063871383667, "step": 30172 }, { "epoch": 2.3786361844698463, "grad_norm": 0.4423706829547882, "learning_rate": 0.0005, "loss": 1.2625503540039062, "step": 30173 }, { "epoch": 2.3787150177374854, "grad_norm": 0.4906306564807892, "learning_rate": 0.0005, "loss": 1.359193205833435, "step": 30174 }, { "epoch": 2.378793851005124, "grad_norm": 0.6089495420455933, "learning_rate": 0.0005, "loss": 1.6559205055236816, "step": 30175 }, { "epoch": 2.378872684272763, "grad_norm": 0.432824045419693, "learning_rate": 0.0005, "loss": 1.26806640625, "step": 30176 }, { "epoch": 2.378951517540402, "grad_norm": 0.4321584105491638, "learning_rate": 0.0005, "loss": 1.3195527791976929, "step": 30177 }, { "epoch": 2.379030350808041, "grad_norm": 0.7134606838226318, "learning_rate": 0.0005, "loss": 1.2950856685638428, "step": 30178 }, { "epoch": 2.3791091840756797, "grad_norm": 0.4414205849170685, "learning_rate": 0.0005, "loss": 1.107662320137024, "step": 30179 }, { "epoch": 2.379188017343319, "grad_norm": 0.6400152444839478, "learning_rate": 0.0005, "loss": 1.4024467468261719, "step": 30180 }, { "epoch": 2.379266850610958, "grad_norm": 0.4829462766647339, "learning_rate": 0.0005, "loss": 1.3110820055007935, "step": 30181 }, { "epoch": 2.3793456838785967, "grad_norm": 0.5315182209014893, "learning_rate": 0.0005, "loss": 1.4391210079193115, "step": 30182 }, { "epoch": 2.379424517146236, "grad_norm": 0.5225861668586731, "learning_rate": 0.0005, "loss": 1.654647707939148, "step": 30183 }, { "epoch": 2.3795033504138745, "grad_norm": 0.49818703532218933, "learning_rate": 0.0005, "loss": 1.1434816122055054, "step": 30184 }, { "epoch": 2.3795821836815136, "grad_norm": 0.5058518648147583, "learning_rate": 0.0005, "loss": 1.3343050479888916, "step": 30185 }, { "epoch": 2.3796610169491528, "grad_norm": 0.5608970522880554, "learning_rate": 0.0005, "loss": 1.4236316680908203, "step": 30186 }, { "epoch": 2.3797398502167915, "grad_norm": 0.6043518781661987, "learning_rate": 0.0005, "loss": 1.254577398300171, "step": 30187 }, { "epoch": 2.3798186834844306, "grad_norm": 0.5579339265823364, "learning_rate": 0.0005, "loss": 1.1635007858276367, "step": 30188 }, { "epoch": 2.3798975167520693, "grad_norm": 0.4936698079109192, "learning_rate": 0.0005, "loss": 1.2660791873931885, "step": 30189 }, { "epoch": 2.3799763500197084, "grad_norm": 0.8462754487991333, "learning_rate": 0.0005, "loss": 1.4142694473266602, "step": 30190 }, { "epoch": 2.380055183287347, "grad_norm": 0.5353327989578247, "learning_rate": 0.0005, "loss": 1.4378662109375, "step": 30191 }, { "epoch": 2.3801340165549862, "grad_norm": 0.6556992530822754, "learning_rate": 0.0005, "loss": 1.4291892051696777, "step": 30192 }, { "epoch": 2.380212849822625, "grad_norm": 0.6284310221672058, "learning_rate": 0.0005, "loss": 1.3972508907318115, "step": 30193 }, { "epoch": 2.380291683090264, "grad_norm": 0.48276832699775696, "learning_rate": 0.0005, "loss": 1.2027299404144287, "step": 30194 }, { "epoch": 2.380370516357903, "grad_norm": 0.5372943878173828, "learning_rate": 0.0005, "loss": 1.3862903118133545, "step": 30195 }, { "epoch": 2.380449349625542, "grad_norm": 0.5802838206291199, "learning_rate": 0.0005, "loss": 1.4630788564682007, "step": 30196 }, { "epoch": 2.380528182893181, "grad_norm": 0.5814650654792786, "learning_rate": 0.0005, "loss": 1.4189157485961914, "step": 30197 }, { "epoch": 2.3806070161608197, "grad_norm": 0.49540799856185913, "learning_rate": 0.0005, "loss": 1.4020146131515503, "step": 30198 }, { "epoch": 2.380685849428459, "grad_norm": 0.4561633765697479, "learning_rate": 0.0005, "loss": 1.2902413606643677, "step": 30199 }, { "epoch": 2.380764682696098, "grad_norm": 0.44129180908203125, "learning_rate": 0.0005, "loss": 1.3330543041229248, "step": 30200 }, { "epoch": 2.3808435159637367, "grad_norm": 0.46777817606925964, "learning_rate": 0.0005, "loss": 1.2167065143585205, "step": 30201 }, { "epoch": 2.380922349231376, "grad_norm": 0.6366841793060303, "learning_rate": 0.0005, "loss": 1.4125967025756836, "step": 30202 }, { "epoch": 2.3810011824990145, "grad_norm": 0.43422386050224304, "learning_rate": 0.0005, "loss": 1.3831101655960083, "step": 30203 }, { "epoch": 2.3810800157666536, "grad_norm": 0.5160349607467651, "learning_rate": 0.0005, "loss": 1.2541038990020752, "step": 30204 }, { "epoch": 2.3811588490342923, "grad_norm": 0.5054811835289001, "learning_rate": 0.0005, "loss": 1.3616201877593994, "step": 30205 }, { "epoch": 2.3812376823019314, "grad_norm": 0.45785507559776306, "learning_rate": 0.0005, "loss": 1.2375035285949707, "step": 30206 }, { "epoch": 2.38131651556957, "grad_norm": 0.4734083414077759, "learning_rate": 0.0005, "loss": 1.1627339124679565, "step": 30207 }, { "epoch": 2.3813953488372093, "grad_norm": 0.5110750198364258, "learning_rate": 0.0005, "loss": 1.3345398902893066, "step": 30208 }, { "epoch": 2.3814741821048484, "grad_norm": 0.469236820936203, "learning_rate": 0.0005, "loss": 1.4356465339660645, "step": 30209 }, { "epoch": 2.381553015372487, "grad_norm": 0.44673681259155273, "learning_rate": 0.0005, "loss": 1.1100417375564575, "step": 30210 }, { "epoch": 2.3816318486401262, "grad_norm": 0.5214987993240356, "learning_rate": 0.0005, "loss": 1.49531888961792, "step": 30211 }, { "epoch": 2.381710681907765, "grad_norm": 0.3872247636318207, "learning_rate": 0.0005, "loss": 1.2123557329177856, "step": 30212 }, { "epoch": 2.381789515175404, "grad_norm": 0.44126418232917786, "learning_rate": 0.0005, "loss": 1.3883012533187866, "step": 30213 }, { "epoch": 2.381868348443043, "grad_norm": 0.39977723360061646, "learning_rate": 0.0005, "loss": 1.1734249591827393, "step": 30214 }, { "epoch": 2.381947181710682, "grad_norm": 0.47004950046539307, "learning_rate": 0.0005, "loss": 1.1941134929656982, "step": 30215 }, { "epoch": 2.382026014978321, "grad_norm": 0.5141206979751587, "learning_rate": 0.0005, "loss": 1.269256591796875, "step": 30216 }, { "epoch": 2.3821048482459597, "grad_norm": 0.4767838716506958, "learning_rate": 0.0005, "loss": 1.1927812099456787, "step": 30217 }, { "epoch": 2.382183681513599, "grad_norm": 0.6788123250007629, "learning_rate": 0.0005, "loss": 1.342758059501648, "step": 30218 }, { "epoch": 2.3822625147812375, "grad_norm": 0.5947864651679993, "learning_rate": 0.0005, "loss": 1.2770380973815918, "step": 30219 }, { "epoch": 2.3823413480488767, "grad_norm": 0.4559033215045929, "learning_rate": 0.0005, "loss": 1.4127331972122192, "step": 30220 }, { "epoch": 2.3824201813165153, "grad_norm": 0.48439279198646545, "learning_rate": 0.0005, "loss": 1.3958330154418945, "step": 30221 }, { "epoch": 2.3824990145841545, "grad_norm": 0.7188881039619446, "learning_rate": 0.0005, "loss": 1.4126808643341064, "step": 30222 }, { "epoch": 2.3825778478517936, "grad_norm": 0.4935908019542694, "learning_rate": 0.0005, "loss": 1.2737290859222412, "step": 30223 }, { "epoch": 2.3826566811194323, "grad_norm": 0.5905735492706299, "learning_rate": 0.0005, "loss": 1.6316865682601929, "step": 30224 }, { "epoch": 2.3827355143870714, "grad_norm": 0.47043219208717346, "learning_rate": 0.0005, "loss": 1.5399575233459473, "step": 30225 }, { "epoch": 2.38281434765471, "grad_norm": 0.5655587315559387, "learning_rate": 0.0005, "loss": 1.1855661869049072, "step": 30226 }, { "epoch": 2.3828931809223493, "grad_norm": 0.4501706659793854, "learning_rate": 0.0005, "loss": 1.4834039211273193, "step": 30227 }, { "epoch": 2.3829720141899884, "grad_norm": 0.5652788281440735, "learning_rate": 0.0005, "loss": 1.314281940460205, "step": 30228 }, { "epoch": 2.383050847457627, "grad_norm": 0.5203577876091003, "learning_rate": 0.0005, "loss": 1.262765645980835, "step": 30229 }, { "epoch": 2.383129680725266, "grad_norm": 0.5265558362007141, "learning_rate": 0.0005, "loss": 1.3616269826889038, "step": 30230 }, { "epoch": 2.383208513992905, "grad_norm": 0.5428612232208252, "learning_rate": 0.0005, "loss": 1.2841349840164185, "step": 30231 }, { "epoch": 2.383287347260544, "grad_norm": 0.46069133281707764, "learning_rate": 0.0005, "loss": 1.3244761228561401, "step": 30232 }, { "epoch": 2.3833661805281827, "grad_norm": 0.6123635768890381, "learning_rate": 0.0005, "loss": 1.4274921417236328, "step": 30233 }, { "epoch": 2.383445013795822, "grad_norm": 0.3228634297847748, "learning_rate": 0.0005, "loss": 0.9651352167129517, "step": 30234 }, { "epoch": 2.3835238470634605, "grad_norm": 0.4619273543357849, "learning_rate": 0.0005, "loss": 1.3752367496490479, "step": 30235 }, { "epoch": 2.3836026803310997, "grad_norm": 0.4412459433078766, "learning_rate": 0.0005, "loss": 1.2692358493804932, "step": 30236 }, { "epoch": 2.383681513598739, "grad_norm": 0.46616753935813904, "learning_rate": 0.0005, "loss": 1.3378629684448242, "step": 30237 }, { "epoch": 2.3837603468663775, "grad_norm": 0.6541306972503662, "learning_rate": 0.0005, "loss": 1.4342527389526367, "step": 30238 }, { "epoch": 2.3838391801340166, "grad_norm": 0.4514096975326538, "learning_rate": 0.0005, "loss": 1.1794270277023315, "step": 30239 }, { "epoch": 2.3839180134016553, "grad_norm": 0.5196794867515564, "learning_rate": 0.0005, "loss": 1.4990146160125732, "step": 30240 }, { "epoch": 2.3839968466692945, "grad_norm": 0.5859379172325134, "learning_rate": 0.0005, "loss": 1.3768041133880615, "step": 30241 }, { "epoch": 2.3840756799369336, "grad_norm": 0.4535490572452545, "learning_rate": 0.0005, "loss": 1.3508392572402954, "step": 30242 }, { "epoch": 2.3841545132045723, "grad_norm": 0.9550881385803223, "learning_rate": 0.0005, "loss": 1.4561182260513306, "step": 30243 }, { "epoch": 2.3842333464722114, "grad_norm": 0.4161125719547272, "learning_rate": 0.0005, "loss": 1.0588164329528809, "step": 30244 }, { "epoch": 2.38431217973985, "grad_norm": 0.5409739017486572, "learning_rate": 0.0005, "loss": 1.5872836112976074, "step": 30245 }, { "epoch": 2.3843910130074892, "grad_norm": 0.43159249424934387, "learning_rate": 0.0005, "loss": 1.1651060581207275, "step": 30246 }, { "epoch": 2.384469846275128, "grad_norm": 0.5697835087776184, "learning_rate": 0.0005, "loss": 1.5069109201431274, "step": 30247 }, { "epoch": 2.384548679542767, "grad_norm": 0.4712488055229187, "learning_rate": 0.0005, "loss": 1.2758334875106812, "step": 30248 }, { "epoch": 2.3846275128104057, "grad_norm": 0.44954830408096313, "learning_rate": 0.0005, "loss": 1.2068026065826416, "step": 30249 }, { "epoch": 2.384706346078045, "grad_norm": 0.47422635555267334, "learning_rate": 0.0005, "loss": 1.2499117851257324, "step": 30250 }, { "epoch": 2.384785179345684, "grad_norm": 0.5291735529899597, "learning_rate": 0.0005, "loss": 1.3761444091796875, "step": 30251 }, { "epoch": 2.3848640126133227, "grad_norm": 0.5375609993934631, "learning_rate": 0.0005, "loss": 1.3086435794830322, "step": 30252 }, { "epoch": 2.384942845880962, "grad_norm": 0.4746270179748535, "learning_rate": 0.0005, "loss": 1.4073172807693481, "step": 30253 }, { "epoch": 2.3850216791486005, "grad_norm": 0.43513864278793335, "learning_rate": 0.0005, "loss": 1.376129388809204, "step": 30254 }, { "epoch": 2.3851005124162397, "grad_norm": 0.625543475151062, "learning_rate": 0.0005, "loss": 1.6315546035766602, "step": 30255 }, { "epoch": 2.385179345683879, "grad_norm": 0.5078287124633789, "learning_rate": 0.0005, "loss": 1.2931612730026245, "step": 30256 }, { "epoch": 2.3852581789515175, "grad_norm": 0.47695595026016235, "learning_rate": 0.0005, "loss": 1.3941835165023804, "step": 30257 }, { "epoch": 2.3853370122191566, "grad_norm": 0.5491047501564026, "learning_rate": 0.0005, "loss": 1.4378387928009033, "step": 30258 }, { "epoch": 2.3854158454867953, "grad_norm": 0.36371880769729614, "learning_rate": 0.0005, "loss": 1.3271374702453613, "step": 30259 }, { "epoch": 2.3854946787544344, "grad_norm": 0.7355305552482605, "learning_rate": 0.0005, "loss": 1.2786048650741577, "step": 30260 }, { "epoch": 2.385573512022073, "grad_norm": 0.5041505694389343, "learning_rate": 0.0005, "loss": 1.3954908847808838, "step": 30261 }, { "epoch": 2.3856523452897123, "grad_norm": 0.5156200528144836, "learning_rate": 0.0005, "loss": 1.3931617736816406, "step": 30262 }, { "epoch": 2.3857311785573514, "grad_norm": 0.473274290561676, "learning_rate": 0.0005, "loss": 1.1599018573760986, "step": 30263 }, { "epoch": 2.38581001182499, "grad_norm": 0.5572308301925659, "learning_rate": 0.0005, "loss": 1.4972243309020996, "step": 30264 }, { "epoch": 2.385888845092629, "grad_norm": 0.5089655518531799, "learning_rate": 0.0005, "loss": 1.285302996635437, "step": 30265 }, { "epoch": 2.385967678360268, "grad_norm": 0.5296815037727356, "learning_rate": 0.0005, "loss": 1.3659393787384033, "step": 30266 }, { "epoch": 2.386046511627907, "grad_norm": 0.5040332078933716, "learning_rate": 0.0005, "loss": 1.6710174083709717, "step": 30267 }, { "epoch": 2.3861253448955457, "grad_norm": 0.6568320393562317, "learning_rate": 0.0005, "loss": 1.419640064239502, "step": 30268 }, { "epoch": 2.386204178163185, "grad_norm": 0.590212345123291, "learning_rate": 0.0005, "loss": 1.267667293548584, "step": 30269 }, { "epoch": 2.386283011430824, "grad_norm": 0.45033425092697144, "learning_rate": 0.0005, "loss": 1.2951768636703491, "step": 30270 }, { "epoch": 2.3863618446984627, "grad_norm": 0.5117624998092651, "learning_rate": 0.0005, "loss": 1.4113436937332153, "step": 30271 }, { "epoch": 2.386440677966102, "grad_norm": 0.4611232876777649, "learning_rate": 0.0005, "loss": 1.134425401687622, "step": 30272 }, { "epoch": 2.3865195112337405, "grad_norm": 0.5296425223350525, "learning_rate": 0.0005, "loss": 1.3103746175765991, "step": 30273 }, { "epoch": 2.3865983445013796, "grad_norm": 0.6755410432815552, "learning_rate": 0.0005, "loss": 1.4233665466308594, "step": 30274 }, { "epoch": 2.3866771777690183, "grad_norm": 0.4360792934894562, "learning_rate": 0.0005, "loss": 1.0338460206985474, "step": 30275 }, { "epoch": 2.3867560110366575, "grad_norm": 0.4643706977367401, "learning_rate": 0.0005, "loss": 1.3429534435272217, "step": 30276 }, { "epoch": 2.3868348443042966, "grad_norm": 0.5268436670303345, "learning_rate": 0.0005, "loss": 1.3129867315292358, "step": 30277 }, { "epoch": 2.3869136775719353, "grad_norm": 0.46728286147117615, "learning_rate": 0.0005, "loss": 1.4429655075073242, "step": 30278 }, { "epoch": 2.3869925108395744, "grad_norm": 0.37464600801467896, "learning_rate": 0.0005, "loss": 1.1929283142089844, "step": 30279 }, { "epoch": 2.387071344107213, "grad_norm": 0.636817455291748, "learning_rate": 0.0005, "loss": 1.4475336074829102, "step": 30280 }, { "epoch": 2.3871501773748522, "grad_norm": 0.4517017602920532, "learning_rate": 0.0005, "loss": 1.2641277313232422, "step": 30281 }, { "epoch": 2.387229010642491, "grad_norm": 0.6403820514678955, "learning_rate": 0.0005, "loss": 1.367430567741394, "step": 30282 }, { "epoch": 2.38730784391013, "grad_norm": 0.44993528723716736, "learning_rate": 0.0005, "loss": 1.3080954551696777, "step": 30283 }, { "epoch": 2.387386677177769, "grad_norm": 0.6737614870071411, "learning_rate": 0.0005, "loss": 1.4926929473876953, "step": 30284 }, { "epoch": 2.387465510445408, "grad_norm": 0.5127688646316528, "learning_rate": 0.0005, "loss": 1.3940373659133911, "step": 30285 }, { "epoch": 2.387544343713047, "grad_norm": 0.4277883470058441, "learning_rate": 0.0005, "loss": 1.0497915744781494, "step": 30286 }, { "epoch": 2.3876231769806857, "grad_norm": 0.4518277049064636, "learning_rate": 0.0005, "loss": 1.3028104305267334, "step": 30287 }, { "epoch": 2.387702010248325, "grad_norm": 0.44155794382095337, "learning_rate": 0.0005, "loss": 1.2224867343902588, "step": 30288 }, { "epoch": 2.387780843515964, "grad_norm": 0.5966147780418396, "learning_rate": 0.0005, "loss": 1.3660459518432617, "step": 30289 }, { "epoch": 2.3878596767836027, "grad_norm": 0.6910567879676819, "learning_rate": 0.0005, "loss": 1.5154756307601929, "step": 30290 }, { "epoch": 2.387938510051242, "grad_norm": 0.536179780960083, "learning_rate": 0.0005, "loss": 1.1943168640136719, "step": 30291 }, { "epoch": 2.3880173433188805, "grad_norm": 0.4574786424636841, "learning_rate": 0.0005, "loss": 1.1882703304290771, "step": 30292 }, { "epoch": 2.3880961765865196, "grad_norm": 0.4594443142414093, "learning_rate": 0.0005, "loss": 1.4007612466812134, "step": 30293 }, { "epoch": 2.3881750098541583, "grad_norm": 0.6016282439231873, "learning_rate": 0.0005, "loss": 1.3591071367263794, "step": 30294 }, { "epoch": 2.3882538431217974, "grad_norm": 0.5329635143280029, "learning_rate": 0.0005, "loss": 1.3344244956970215, "step": 30295 }, { "epoch": 2.388332676389436, "grad_norm": 0.44477254152297974, "learning_rate": 0.0005, "loss": 1.1405292749404907, "step": 30296 }, { "epoch": 2.3884115096570753, "grad_norm": 0.47102808952331543, "learning_rate": 0.0005, "loss": 1.170687198638916, "step": 30297 }, { "epoch": 2.3884903429247144, "grad_norm": 0.5740942358970642, "learning_rate": 0.0005, "loss": 1.2148642539978027, "step": 30298 }, { "epoch": 2.388569176192353, "grad_norm": 0.485402911901474, "learning_rate": 0.0005, "loss": 1.197378158569336, "step": 30299 }, { "epoch": 2.3886480094599922, "grad_norm": 0.8802101016044617, "learning_rate": 0.0005, "loss": 1.4146404266357422, "step": 30300 }, { "epoch": 2.388726842727631, "grad_norm": 0.5635148882865906, "learning_rate": 0.0005, "loss": 1.314103126525879, "step": 30301 }, { "epoch": 2.38880567599527, "grad_norm": 0.488321989774704, "learning_rate": 0.0005, "loss": 1.3105583190917969, "step": 30302 }, { "epoch": 2.388884509262909, "grad_norm": 0.46342933177948, "learning_rate": 0.0005, "loss": 1.4445514678955078, "step": 30303 }, { "epoch": 2.388963342530548, "grad_norm": 0.5607169270515442, "learning_rate": 0.0005, "loss": 1.1070735454559326, "step": 30304 }, { "epoch": 2.389042175798187, "grad_norm": 0.4308702051639557, "learning_rate": 0.0005, "loss": 1.1454136371612549, "step": 30305 }, { "epoch": 2.3891210090658257, "grad_norm": 0.5399348139762878, "learning_rate": 0.0005, "loss": 1.367856502532959, "step": 30306 }, { "epoch": 2.389199842333465, "grad_norm": 0.5009578466415405, "learning_rate": 0.0005, "loss": 1.2455625534057617, "step": 30307 }, { "epoch": 2.3892786756011035, "grad_norm": 0.511857271194458, "learning_rate": 0.0005, "loss": 1.307671308517456, "step": 30308 }, { "epoch": 2.3893575088687427, "grad_norm": 0.4294429123401642, "learning_rate": 0.0005, "loss": 1.3988152742385864, "step": 30309 }, { "epoch": 2.3894363421363813, "grad_norm": 0.5647504329681396, "learning_rate": 0.0005, "loss": 1.2499123811721802, "step": 30310 }, { "epoch": 2.3895151754040205, "grad_norm": 0.5373870134353638, "learning_rate": 0.0005, "loss": 1.2123547792434692, "step": 30311 }, { "epoch": 2.3895940086716596, "grad_norm": 0.45145875215530396, "learning_rate": 0.0005, "loss": 1.321375846862793, "step": 30312 }, { "epoch": 2.3896728419392983, "grad_norm": 0.563499927520752, "learning_rate": 0.0005, "loss": 1.3302652835845947, "step": 30313 }, { "epoch": 2.3897516752069374, "grad_norm": 0.4926310181617737, "learning_rate": 0.0005, "loss": 1.443946361541748, "step": 30314 }, { "epoch": 2.389830508474576, "grad_norm": 0.7077040672302246, "learning_rate": 0.0005, "loss": 1.5227069854736328, "step": 30315 }, { "epoch": 2.3899093417422153, "grad_norm": 0.5002076029777527, "learning_rate": 0.0005, "loss": 1.307321310043335, "step": 30316 }, { "epoch": 2.3899881750098544, "grad_norm": 0.47859373688697815, "learning_rate": 0.0005, "loss": 1.3413622379302979, "step": 30317 }, { "epoch": 2.390067008277493, "grad_norm": 0.5261622667312622, "learning_rate": 0.0005, "loss": 1.2649949789047241, "step": 30318 }, { "epoch": 2.390145841545132, "grad_norm": 0.5135011076927185, "learning_rate": 0.0005, "loss": 1.3286619186401367, "step": 30319 }, { "epoch": 2.390224674812771, "grad_norm": 0.49665647745132446, "learning_rate": 0.0005, "loss": 1.3045527935028076, "step": 30320 }, { "epoch": 2.39030350808041, "grad_norm": 0.4836537539958954, "learning_rate": 0.0005, "loss": 1.2700518369674683, "step": 30321 }, { "epoch": 2.3903823413480487, "grad_norm": 0.48979803919792175, "learning_rate": 0.0005, "loss": 1.2295210361480713, "step": 30322 }, { "epoch": 2.390461174615688, "grad_norm": 0.45719608664512634, "learning_rate": 0.0005, "loss": 1.4660370349884033, "step": 30323 }, { "epoch": 2.3905400078833265, "grad_norm": 0.4187474846839905, "learning_rate": 0.0005, "loss": 1.2296931743621826, "step": 30324 }, { "epoch": 2.3906188411509657, "grad_norm": 0.769026517868042, "learning_rate": 0.0005, "loss": 1.3133822679519653, "step": 30325 }, { "epoch": 2.390697674418605, "grad_norm": 0.45827168226242065, "learning_rate": 0.0005, "loss": 1.2541331052780151, "step": 30326 }, { "epoch": 2.3907765076862435, "grad_norm": 0.4176001250743866, "learning_rate": 0.0005, "loss": 1.2421305179595947, "step": 30327 }, { "epoch": 2.3908553409538826, "grad_norm": 0.5745885968208313, "learning_rate": 0.0005, "loss": 1.3192111253738403, "step": 30328 }, { "epoch": 2.3909341742215213, "grad_norm": 0.5079789757728577, "learning_rate": 0.0005, "loss": 1.100801706314087, "step": 30329 }, { "epoch": 2.3910130074891605, "grad_norm": 0.49012765288352966, "learning_rate": 0.0005, "loss": 1.2879862785339355, "step": 30330 }, { "epoch": 2.3910918407567996, "grad_norm": 0.576930820941925, "learning_rate": 0.0005, "loss": 1.277148723602295, "step": 30331 }, { "epoch": 2.3911706740244383, "grad_norm": 0.6090522408485413, "learning_rate": 0.0005, "loss": 1.3816521167755127, "step": 30332 }, { "epoch": 2.3912495072920774, "grad_norm": 0.5703935027122498, "learning_rate": 0.0005, "loss": 1.3643887042999268, "step": 30333 }, { "epoch": 2.391328340559716, "grad_norm": 0.6218575835227966, "learning_rate": 0.0005, "loss": 1.2155427932739258, "step": 30334 }, { "epoch": 2.3914071738273552, "grad_norm": 0.4690311551094055, "learning_rate": 0.0005, "loss": 1.3001375198364258, "step": 30335 }, { "epoch": 2.391486007094994, "grad_norm": 0.38810572028160095, "learning_rate": 0.0005, "loss": 1.2593190670013428, "step": 30336 }, { "epoch": 2.391564840362633, "grad_norm": 0.5980260968208313, "learning_rate": 0.0005, "loss": 1.623907208442688, "step": 30337 }, { "epoch": 2.3916436736302717, "grad_norm": 0.5147457718849182, "learning_rate": 0.0005, "loss": 1.4259731769561768, "step": 30338 }, { "epoch": 2.391722506897911, "grad_norm": 0.40394076704978943, "learning_rate": 0.0005, "loss": 1.3408312797546387, "step": 30339 }, { "epoch": 2.39180134016555, "grad_norm": 0.6720328330993652, "learning_rate": 0.0005, "loss": 1.3515255451202393, "step": 30340 }, { "epoch": 2.3918801734331887, "grad_norm": 0.465149462223053, "learning_rate": 0.0005, "loss": 1.4142895936965942, "step": 30341 }, { "epoch": 2.391959006700828, "grad_norm": 0.5579989552497864, "learning_rate": 0.0005, "loss": 1.5063385963439941, "step": 30342 }, { "epoch": 2.3920378399684665, "grad_norm": 0.3933086097240448, "learning_rate": 0.0005, "loss": 1.157827377319336, "step": 30343 }, { "epoch": 2.3921166732361057, "grad_norm": 0.4248965084552765, "learning_rate": 0.0005, "loss": 1.2488837242126465, "step": 30344 }, { "epoch": 2.392195506503745, "grad_norm": 0.7556132078170776, "learning_rate": 0.0005, "loss": 1.3824446201324463, "step": 30345 }, { "epoch": 2.3922743397713835, "grad_norm": 0.5252332091331482, "learning_rate": 0.0005, "loss": 1.2432039976119995, "step": 30346 }, { "epoch": 2.3923531730390226, "grad_norm": 0.657310962677002, "learning_rate": 0.0005, "loss": 1.419972538948059, "step": 30347 }, { "epoch": 2.3924320063066613, "grad_norm": 0.49944016337394714, "learning_rate": 0.0005, "loss": 1.2754395008087158, "step": 30348 }, { "epoch": 2.3925108395743004, "grad_norm": 0.4612756371498108, "learning_rate": 0.0005, "loss": 1.281625509262085, "step": 30349 }, { "epoch": 2.392589672841939, "grad_norm": 0.6163269877433777, "learning_rate": 0.0005, "loss": 1.371434211730957, "step": 30350 }, { "epoch": 2.3926685061095783, "grad_norm": 0.9966849684715271, "learning_rate": 0.0005, "loss": 1.343493938446045, "step": 30351 }, { "epoch": 2.392747339377217, "grad_norm": 0.5127993822097778, "learning_rate": 0.0005, "loss": 1.3669514656066895, "step": 30352 }, { "epoch": 2.392826172644856, "grad_norm": 0.4839853346347809, "learning_rate": 0.0005, "loss": 1.3224167823791504, "step": 30353 }, { "epoch": 2.392905005912495, "grad_norm": 0.507139265537262, "learning_rate": 0.0005, "loss": 1.5621837377548218, "step": 30354 }, { "epoch": 2.392983839180134, "grad_norm": 0.5191482901573181, "learning_rate": 0.0005, "loss": 1.2294535636901855, "step": 30355 }, { "epoch": 2.393062672447773, "grad_norm": 0.4793921709060669, "learning_rate": 0.0005, "loss": 1.2169111967086792, "step": 30356 }, { "epoch": 2.3931415057154117, "grad_norm": 0.4213634133338928, "learning_rate": 0.0005, "loss": 1.348297357559204, "step": 30357 }, { "epoch": 2.393220338983051, "grad_norm": 0.4775433838367462, "learning_rate": 0.0005, "loss": 1.2760308980941772, "step": 30358 }, { "epoch": 2.39329917225069, "grad_norm": 0.4491887092590332, "learning_rate": 0.0005, "loss": 1.3751220703125, "step": 30359 }, { "epoch": 2.3933780055183287, "grad_norm": 0.5009178519248962, "learning_rate": 0.0005, "loss": 1.6252223253250122, "step": 30360 }, { "epoch": 2.393456838785968, "grad_norm": 0.5736829042434692, "learning_rate": 0.0005, "loss": 1.3394269943237305, "step": 30361 }, { "epoch": 2.3935356720536065, "grad_norm": 0.4812249541282654, "learning_rate": 0.0005, "loss": 1.2926716804504395, "step": 30362 }, { "epoch": 2.3936145053212456, "grad_norm": 0.5825659036636353, "learning_rate": 0.0005, "loss": 1.3679022789001465, "step": 30363 }, { "epoch": 2.3936933385888843, "grad_norm": 0.450888454914093, "learning_rate": 0.0005, "loss": 1.1155602931976318, "step": 30364 }, { "epoch": 2.3937721718565235, "grad_norm": 0.6599665284156799, "learning_rate": 0.0005, "loss": 1.1958177089691162, "step": 30365 }, { "epoch": 2.393851005124162, "grad_norm": 0.5602315664291382, "learning_rate": 0.0005, "loss": 1.341648817062378, "step": 30366 }, { "epoch": 2.3939298383918013, "grad_norm": 0.49539732933044434, "learning_rate": 0.0005, "loss": 1.1969462633132935, "step": 30367 }, { "epoch": 2.3940086716594404, "grad_norm": 0.4558833837509155, "learning_rate": 0.0005, "loss": 1.271085262298584, "step": 30368 }, { "epoch": 2.394087504927079, "grad_norm": 0.4449945092201233, "learning_rate": 0.0005, "loss": 1.3119112253189087, "step": 30369 }, { "epoch": 2.3941663381947182, "grad_norm": 0.626749575138092, "learning_rate": 0.0005, "loss": 1.439324140548706, "step": 30370 }, { "epoch": 2.394245171462357, "grad_norm": 0.43126219511032104, "learning_rate": 0.0005, "loss": 1.1711950302124023, "step": 30371 }, { "epoch": 2.394324004729996, "grad_norm": 0.5061604380607605, "learning_rate": 0.0005, "loss": 1.2412033081054688, "step": 30372 }, { "epoch": 2.394402837997635, "grad_norm": 0.5014946460723877, "learning_rate": 0.0005, "loss": 1.3039437532424927, "step": 30373 }, { "epoch": 2.394481671265274, "grad_norm": 0.4700201451778412, "learning_rate": 0.0005, "loss": 1.3120774030685425, "step": 30374 }, { "epoch": 2.394560504532913, "grad_norm": 0.5136470198631287, "learning_rate": 0.0005, "loss": 1.2545056343078613, "step": 30375 }, { "epoch": 2.3946393378005517, "grad_norm": 0.48388370871543884, "learning_rate": 0.0005, "loss": 1.1720653772354126, "step": 30376 }, { "epoch": 2.394718171068191, "grad_norm": 0.424188494682312, "learning_rate": 0.0005, "loss": 1.2409367561340332, "step": 30377 }, { "epoch": 2.3947970043358295, "grad_norm": 0.4641985297203064, "learning_rate": 0.0005, "loss": 1.258826732635498, "step": 30378 }, { "epoch": 2.3948758376034687, "grad_norm": 0.502633273601532, "learning_rate": 0.0005, "loss": 1.2684404850006104, "step": 30379 }, { "epoch": 2.3949546708711074, "grad_norm": 0.5564766526222229, "learning_rate": 0.0005, "loss": 1.2369647026062012, "step": 30380 }, { "epoch": 2.3950335041387465, "grad_norm": 0.7702914476394653, "learning_rate": 0.0005, "loss": 1.3960121870040894, "step": 30381 }, { "epoch": 2.3951123374063856, "grad_norm": 0.5055590271949768, "learning_rate": 0.0005, "loss": 1.3600738048553467, "step": 30382 }, { "epoch": 2.3951911706740243, "grad_norm": 0.44732004404067993, "learning_rate": 0.0005, "loss": 1.355454683303833, "step": 30383 }, { "epoch": 2.3952700039416635, "grad_norm": 0.3843744993209839, "learning_rate": 0.0005, "loss": 1.2485435009002686, "step": 30384 }, { "epoch": 2.395348837209302, "grad_norm": 0.7992888689041138, "learning_rate": 0.0005, "loss": 1.4604558944702148, "step": 30385 }, { "epoch": 2.3954276704769413, "grad_norm": 0.5874817371368408, "learning_rate": 0.0005, "loss": 1.3016319274902344, "step": 30386 }, { "epoch": 2.3955065037445804, "grad_norm": 0.5306949615478516, "learning_rate": 0.0005, "loss": 1.4852323532104492, "step": 30387 }, { "epoch": 2.395585337012219, "grad_norm": 0.5337672829627991, "learning_rate": 0.0005, "loss": 1.3445698022842407, "step": 30388 }, { "epoch": 2.3956641702798582, "grad_norm": 0.6551920771598816, "learning_rate": 0.0005, "loss": 1.357160210609436, "step": 30389 }, { "epoch": 2.395743003547497, "grad_norm": 0.6040166020393372, "learning_rate": 0.0005, "loss": 1.4120948314666748, "step": 30390 }, { "epoch": 2.395821836815136, "grad_norm": 0.4479673206806183, "learning_rate": 0.0005, "loss": 1.3588849306106567, "step": 30391 }, { "epoch": 2.3959006700827747, "grad_norm": 0.49974891543388367, "learning_rate": 0.0005, "loss": 1.377187728881836, "step": 30392 }, { "epoch": 2.395979503350414, "grad_norm": 0.5393363237380981, "learning_rate": 0.0005, "loss": 1.347524642944336, "step": 30393 }, { "epoch": 2.396058336618053, "grad_norm": 0.5415778160095215, "learning_rate": 0.0005, "loss": 1.374428391456604, "step": 30394 }, { "epoch": 2.3961371698856917, "grad_norm": 0.4656955897808075, "learning_rate": 0.0005, "loss": 1.4300291538238525, "step": 30395 }, { "epoch": 2.396216003153331, "grad_norm": 0.5300238132476807, "learning_rate": 0.0005, "loss": 1.1706228256225586, "step": 30396 }, { "epoch": 2.3962948364209695, "grad_norm": 0.567595899105072, "learning_rate": 0.0005, "loss": 1.2376526594161987, "step": 30397 }, { "epoch": 2.3963736696886087, "grad_norm": 0.5621939897537231, "learning_rate": 0.0005, "loss": 1.3471131324768066, "step": 30398 }, { "epoch": 2.3964525029562473, "grad_norm": 0.47089684009552, "learning_rate": 0.0005, "loss": 1.4730637073516846, "step": 30399 }, { "epoch": 2.3965313362238865, "grad_norm": 0.43157562613487244, "learning_rate": 0.0005, "loss": 1.1919333934783936, "step": 30400 }, { "epoch": 2.3966101694915256, "grad_norm": 0.4881153106689453, "learning_rate": 0.0005, "loss": 1.2750849723815918, "step": 30401 }, { "epoch": 2.3966890027591643, "grad_norm": 0.4956544041633606, "learning_rate": 0.0005, "loss": 1.3259637355804443, "step": 30402 }, { "epoch": 2.3967678360268034, "grad_norm": 0.4963454604148865, "learning_rate": 0.0005, "loss": 1.3233485221862793, "step": 30403 }, { "epoch": 2.396846669294442, "grad_norm": 0.5161387324333191, "learning_rate": 0.0005, "loss": 1.327066421508789, "step": 30404 }, { "epoch": 2.3969255025620813, "grad_norm": 0.6436188220977783, "learning_rate": 0.0005, "loss": 1.3795855045318604, "step": 30405 }, { "epoch": 2.3970043358297204, "grad_norm": 0.5746564865112305, "learning_rate": 0.0005, "loss": 1.531567096710205, "step": 30406 }, { "epoch": 2.397083169097359, "grad_norm": 0.5275946259498596, "learning_rate": 0.0005, "loss": 1.3925092220306396, "step": 30407 }, { "epoch": 2.397162002364998, "grad_norm": 0.43863216042518616, "learning_rate": 0.0005, "loss": 1.1911404132843018, "step": 30408 }, { "epoch": 2.397240835632637, "grad_norm": 0.969587504863739, "learning_rate": 0.0005, "loss": 1.3755714893341064, "step": 30409 }, { "epoch": 2.397319668900276, "grad_norm": 0.516957700252533, "learning_rate": 0.0005, "loss": 1.2764184474945068, "step": 30410 }, { "epoch": 2.3973985021679147, "grad_norm": 0.5085188150405884, "learning_rate": 0.0005, "loss": 1.319035530090332, "step": 30411 }, { "epoch": 2.397477335435554, "grad_norm": 0.5049147009849548, "learning_rate": 0.0005, "loss": 1.3631007671356201, "step": 30412 }, { "epoch": 2.3975561687031925, "grad_norm": 0.49983301758766174, "learning_rate": 0.0005, "loss": 1.3318787813186646, "step": 30413 }, { "epoch": 2.3976350019708317, "grad_norm": 0.5007420182228088, "learning_rate": 0.0005, "loss": 1.4093818664550781, "step": 30414 }, { "epoch": 2.397713835238471, "grad_norm": 0.5592321753501892, "learning_rate": 0.0005, "loss": 1.2774896621704102, "step": 30415 }, { "epoch": 2.3977926685061095, "grad_norm": 0.450955867767334, "learning_rate": 0.0005, "loss": 1.1586874723434448, "step": 30416 }, { "epoch": 2.3978715017737486, "grad_norm": 0.8543906211853027, "learning_rate": 0.0005, "loss": 1.229564905166626, "step": 30417 }, { "epoch": 2.3979503350413873, "grad_norm": 0.8545133471488953, "learning_rate": 0.0005, "loss": 1.680030107498169, "step": 30418 }, { "epoch": 2.3980291683090265, "grad_norm": 0.441496878862381, "learning_rate": 0.0005, "loss": 1.2430421113967896, "step": 30419 }, { "epoch": 2.3981080015766656, "grad_norm": 0.6613968014717102, "learning_rate": 0.0005, "loss": 1.6025886535644531, "step": 30420 }, { "epoch": 2.3981868348443043, "grad_norm": 0.49773356318473816, "learning_rate": 0.0005, "loss": 1.321803092956543, "step": 30421 }, { "epoch": 2.3982656681119434, "grad_norm": 0.5547326803207397, "learning_rate": 0.0005, "loss": 1.5160813331604004, "step": 30422 }, { "epoch": 2.398344501379582, "grad_norm": 0.7254179120063782, "learning_rate": 0.0005, "loss": 1.4869141578674316, "step": 30423 }, { "epoch": 2.3984233346472212, "grad_norm": 0.599705696105957, "learning_rate": 0.0005, "loss": 1.1315209865570068, "step": 30424 }, { "epoch": 2.39850216791486, "grad_norm": 0.4746115803718567, "learning_rate": 0.0005, "loss": 1.2445961236953735, "step": 30425 }, { "epoch": 2.398581001182499, "grad_norm": 0.5172304511070251, "learning_rate": 0.0005, "loss": 1.4853180646896362, "step": 30426 }, { "epoch": 2.3986598344501378, "grad_norm": 0.46384763717651367, "learning_rate": 0.0005, "loss": 1.3821629285812378, "step": 30427 }, { "epoch": 2.398738667717777, "grad_norm": 0.5024218559265137, "learning_rate": 0.0005, "loss": 1.2833352088928223, "step": 30428 }, { "epoch": 2.398817500985416, "grad_norm": 0.4743287265300751, "learning_rate": 0.0005, "loss": 1.3498518466949463, "step": 30429 }, { "epoch": 2.3988963342530547, "grad_norm": 0.42746245861053467, "learning_rate": 0.0005, "loss": 1.122964859008789, "step": 30430 }, { "epoch": 2.398975167520694, "grad_norm": 0.5626628398895264, "learning_rate": 0.0005, "loss": 1.389521598815918, "step": 30431 }, { "epoch": 2.3990540007883325, "grad_norm": 0.5734891891479492, "learning_rate": 0.0005, "loss": 1.3278441429138184, "step": 30432 }, { "epoch": 2.3991328340559717, "grad_norm": 0.5166229009628296, "learning_rate": 0.0005, "loss": 1.2644315958023071, "step": 30433 }, { "epoch": 2.399211667323611, "grad_norm": 0.5376887321472168, "learning_rate": 0.0005, "loss": 1.377132773399353, "step": 30434 }, { "epoch": 2.3992905005912495, "grad_norm": 0.4381266236305237, "learning_rate": 0.0005, "loss": 1.216172695159912, "step": 30435 }, { "epoch": 2.3993693338588886, "grad_norm": 0.44672632217407227, "learning_rate": 0.0005, "loss": 1.2525180578231812, "step": 30436 }, { "epoch": 2.3994481671265273, "grad_norm": 0.470217227935791, "learning_rate": 0.0005, "loss": 1.2866613864898682, "step": 30437 }, { "epoch": 2.3995270003941664, "grad_norm": 0.4778665006160736, "learning_rate": 0.0005, "loss": 1.321770191192627, "step": 30438 }, { "epoch": 2.399605833661805, "grad_norm": 0.47276487946510315, "learning_rate": 0.0005, "loss": 1.3460183143615723, "step": 30439 }, { "epoch": 2.3996846669294443, "grad_norm": 0.44721361994743347, "learning_rate": 0.0005, "loss": 1.2930207252502441, "step": 30440 }, { "epoch": 2.399763500197083, "grad_norm": 0.502497673034668, "learning_rate": 0.0005, "loss": 1.2435153722763062, "step": 30441 }, { "epoch": 2.399842333464722, "grad_norm": 0.4974271357059479, "learning_rate": 0.0005, "loss": 1.3201580047607422, "step": 30442 }, { "epoch": 2.399921166732361, "grad_norm": 0.48825183510780334, "learning_rate": 0.0005, "loss": 1.3740031719207764, "step": 30443 }, { "epoch": 2.4, "grad_norm": 0.5323765873908997, "learning_rate": 0.0005, "loss": 1.4265806674957275, "step": 30444 }, { "epoch": 2.400078833267639, "grad_norm": 0.5623005032539368, "learning_rate": 0.0005, "loss": 1.362779140472412, "step": 30445 }, { "epoch": 2.4001576665352777, "grad_norm": 0.4368518888950348, "learning_rate": 0.0005, "loss": 1.2719318866729736, "step": 30446 }, { "epoch": 2.400236499802917, "grad_norm": 0.5236346125602722, "learning_rate": 0.0005, "loss": 1.3142364025115967, "step": 30447 }, { "epoch": 2.400315333070556, "grad_norm": 0.41351643204689026, "learning_rate": 0.0005, "loss": 1.1009387969970703, "step": 30448 }, { "epoch": 2.4003941663381947, "grad_norm": 0.5286643505096436, "learning_rate": 0.0005, "loss": 1.2967193126678467, "step": 30449 }, { "epoch": 2.400472999605834, "grad_norm": 0.455782949924469, "learning_rate": 0.0005, "loss": 1.172743320465088, "step": 30450 }, { "epoch": 2.4005518328734725, "grad_norm": 0.4697350263595581, "learning_rate": 0.0005, "loss": 1.3987691402435303, "step": 30451 }, { "epoch": 2.4006306661411116, "grad_norm": 0.552911639213562, "learning_rate": 0.0005, "loss": 1.5007731914520264, "step": 30452 }, { "epoch": 2.4007094994087503, "grad_norm": 0.4552183747291565, "learning_rate": 0.0005, "loss": 1.322939395904541, "step": 30453 }, { "epoch": 2.4007883326763895, "grad_norm": 0.40181177854537964, "learning_rate": 0.0005, "loss": 1.2292112112045288, "step": 30454 }, { "epoch": 2.400867165944028, "grad_norm": 0.6107909679412842, "learning_rate": 0.0005, "loss": 1.3212144374847412, "step": 30455 }, { "epoch": 2.4009459992116673, "grad_norm": 0.37641221284866333, "learning_rate": 0.0005, "loss": 1.241032361984253, "step": 30456 }, { "epoch": 2.4010248324793064, "grad_norm": 0.539331316947937, "learning_rate": 0.0005, "loss": 1.102243423461914, "step": 30457 }, { "epoch": 2.401103665746945, "grad_norm": 0.48375871777534485, "learning_rate": 0.0005, "loss": 1.3968656063079834, "step": 30458 }, { "epoch": 2.4011824990145842, "grad_norm": 0.5259376168251038, "learning_rate": 0.0005, "loss": 1.2845587730407715, "step": 30459 }, { "epoch": 2.401261332282223, "grad_norm": 0.5070331692695618, "learning_rate": 0.0005, "loss": 1.6289260387420654, "step": 30460 }, { "epoch": 2.401340165549862, "grad_norm": 0.6682738661766052, "learning_rate": 0.0005, "loss": 1.356916904449463, "step": 30461 }, { "epoch": 2.401418998817501, "grad_norm": 0.7036702036857605, "learning_rate": 0.0005, "loss": 1.4545228481292725, "step": 30462 }, { "epoch": 2.40149783208514, "grad_norm": 0.5546646118164062, "learning_rate": 0.0005, "loss": 1.454718828201294, "step": 30463 }, { "epoch": 2.401576665352779, "grad_norm": 0.5318651795387268, "learning_rate": 0.0005, "loss": 1.427734613418579, "step": 30464 }, { "epoch": 2.4016554986204177, "grad_norm": 1.3797756433486938, "learning_rate": 0.0005, "loss": 1.3911025524139404, "step": 30465 }, { "epoch": 2.401734331888057, "grad_norm": 0.45674416422843933, "learning_rate": 0.0005, "loss": 1.2954769134521484, "step": 30466 }, { "epoch": 2.4018131651556955, "grad_norm": 0.44507551193237305, "learning_rate": 0.0005, "loss": 1.3502774238586426, "step": 30467 }, { "epoch": 2.4018919984233347, "grad_norm": 0.4997273087501526, "learning_rate": 0.0005, "loss": 1.3235496282577515, "step": 30468 }, { "epoch": 2.4019708316909734, "grad_norm": 0.639823317527771, "learning_rate": 0.0005, "loss": 1.3841407299041748, "step": 30469 }, { "epoch": 2.4020496649586125, "grad_norm": 0.5998167395591736, "learning_rate": 0.0005, "loss": 1.3511409759521484, "step": 30470 }, { "epoch": 2.4021284982262516, "grad_norm": 0.5412846803665161, "learning_rate": 0.0005, "loss": 1.3563960790634155, "step": 30471 }, { "epoch": 2.4022073314938903, "grad_norm": 0.4908992350101471, "learning_rate": 0.0005, "loss": 1.4400646686553955, "step": 30472 }, { "epoch": 2.4022861647615295, "grad_norm": 0.43199098110198975, "learning_rate": 0.0005, "loss": 1.2895941734313965, "step": 30473 }, { "epoch": 2.402364998029168, "grad_norm": 0.5708037614822388, "learning_rate": 0.0005, "loss": 1.474419116973877, "step": 30474 }, { "epoch": 2.4024438312968073, "grad_norm": 0.5026431083679199, "learning_rate": 0.0005, "loss": 1.1351053714752197, "step": 30475 }, { "epoch": 2.4025226645644464, "grad_norm": 0.583845317363739, "learning_rate": 0.0005, "loss": 1.4493539333343506, "step": 30476 }, { "epoch": 2.402601497832085, "grad_norm": 0.5217270255088806, "learning_rate": 0.0005, "loss": 1.1669657230377197, "step": 30477 }, { "epoch": 2.4026803310997242, "grad_norm": 0.5896086692810059, "learning_rate": 0.0005, "loss": 1.3896353244781494, "step": 30478 }, { "epoch": 2.402759164367363, "grad_norm": 0.6192582249641418, "learning_rate": 0.0005, "loss": 1.4010975360870361, "step": 30479 }, { "epoch": 2.402837997635002, "grad_norm": 0.4933750629425049, "learning_rate": 0.0005, "loss": 1.2676587104797363, "step": 30480 }, { "epoch": 2.4029168309026407, "grad_norm": 0.3941836357116699, "learning_rate": 0.0005, "loss": 1.0395466089248657, "step": 30481 }, { "epoch": 2.40299566417028, "grad_norm": 0.44828253984451294, "learning_rate": 0.0005, "loss": 1.2234572172164917, "step": 30482 }, { "epoch": 2.4030744974379186, "grad_norm": 0.6763484477996826, "learning_rate": 0.0005, "loss": 1.429613709449768, "step": 30483 }, { "epoch": 2.4031533307055577, "grad_norm": 0.5391541123390198, "learning_rate": 0.0005, "loss": 1.3094310760498047, "step": 30484 }, { "epoch": 2.403232163973197, "grad_norm": 0.44869333505630493, "learning_rate": 0.0005, "loss": 1.36287522315979, "step": 30485 }, { "epoch": 2.4033109972408355, "grad_norm": 0.4895266592502594, "learning_rate": 0.0005, "loss": 1.419177532196045, "step": 30486 }, { "epoch": 2.4033898305084747, "grad_norm": 0.44044914841651917, "learning_rate": 0.0005, "loss": 1.1709716320037842, "step": 30487 }, { "epoch": 2.4034686637761133, "grad_norm": 0.5635640025138855, "learning_rate": 0.0005, "loss": 1.346233606338501, "step": 30488 }, { "epoch": 2.4035474970437525, "grad_norm": 0.608455240726471, "learning_rate": 0.0005, "loss": 1.3469038009643555, "step": 30489 }, { "epoch": 2.4036263303113916, "grad_norm": 0.5576043128967285, "learning_rate": 0.0005, "loss": 1.162510633468628, "step": 30490 }, { "epoch": 2.4037051635790303, "grad_norm": 0.5679395198822021, "learning_rate": 0.0005, "loss": 1.5415544509887695, "step": 30491 }, { "epoch": 2.4037839968466694, "grad_norm": 0.48910683393478394, "learning_rate": 0.0005, "loss": 1.3118170499801636, "step": 30492 }, { "epoch": 2.403862830114308, "grad_norm": 0.4982037842273712, "learning_rate": 0.0005, "loss": 1.3736493587493896, "step": 30493 }, { "epoch": 2.4039416633819473, "grad_norm": 0.43815016746520996, "learning_rate": 0.0005, "loss": 1.2104679346084595, "step": 30494 }, { "epoch": 2.404020496649586, "grad_norm": 0.746533989906311, "learning_rate": 0.0005, "loss": 1.2933154106140137, "step": 30495 }, { "epoch": 2.404099329917225, "grad_norm": 0.6929632425308228, "learning_rate": 0.0005, "loss": 1.4177696704864502, "step": 30496 }, { "epoch": 2.4041781631848638, "grad_norm": 0.4835120737552643, "learning_rate": 0.0005, "loss": 1.2786915302276611, "step": 30497 }, { "epoch": 2.404256996452503, "grad_norm": 0.6680625081062317, "learning_rate": 0.0005, "loss": 1.2503018379211426, "step": 30498 }, { "epoch": 2.404335829720142, "grad_norm": 0.5483022928237915, "learning_rate": 0.0005, "loss": 1.3920187950134277, "step": 30499 }, { "epoch": 2.4044146629877807, "grad_norm": 0.7250402569770813, "learning_rate": 0.0005, "loss": 1.4452803134918213, "step": 30500 }, { "epoch": 2.40449349625542, "grad_norm": 0.49431121349334717, "learning_rate": 0.0005, "loss": 1.549615740776062, "step": 30501 }, { "epoch": 2.4045723295230585, "grad_norm": 0.5062157511711121, "learning_rate": 0.0005, "loss": 1.4196341037750244, "step": 30502 }, { "epoch": 2.4046511627906977, "grad_norm": 0.5216296315193176, "learning_rate": 0.0005, "loss": 1.3070875406265259, "step": 30503 }, { "epoch": 2.404729996058337, "grad_norm": 0.5829538702964783, "learning_rate": 0.0005, "loss": 1.4407861232757568, "step": 30504 }, { "epoch": 2.4048088293259755, "grad_norm": 0.4288618862628937, "learning_rate": 0.0005, "loss": 1.3110811710357666, "step": 30505 }, { "epoch": 2.4048876625936146, "grad_norm": 0.5768395662307739, "learning_rate": 0.0005, "loss": 1.314454197883606, "step": 30506 }, { "epoch": 2.4049664958612533, "grad_norm": 0.5027259588241577, "learning_rate": 0.0005, "loss": 1.2919946908950806, "step": 30507 }, { "epoch": 2.4050453291288925, "grad_norm": 0.5519617199897766, "learning_rate": 0.0005, "loss": 1.101656198501587, "step": 30508 }, { "epoch": 2.405124162396531, "grad_norm": 0.47430741786956787, "learning_rate": 0.0005, "loss": 1.3772199153900146, "step": 30509 }, { "epoch": 2.4052029956641703, "grad_norm": 0.5551522970199585, "learning_rate": 0.0005, "loss": 1.4249773025512695, "step": 30510 }, { "epoch": 2.405281828931809, "grad_norm": 0.41046497225761414, "learning_rate": 0.0005, "loss": 1.2806729078292847, "step": 30511 }, { "epoch": 2.405360662199448, "grad_norm": 0.4889619052410126, "learning_rate": 0.0005, "loss": 1.2134785652160645, "step": 30512 }, { "epoch": 2.4054394954670872, "grad_norm": 0.49074333906173706, "learning_rate": 0.0005, "loss": 1.2109068632125854, "step": 30513 }, { "epoch": 2.405518328734726, "grad_norm": 0.8271178603172302, "learning_rate": 0.0005, "loss": 1.447887897491455, "step": 30514 }, { "epoch": 2.405597162002365, "grad_norm": 0.6333986520767212, "learning_rate": 0.0005, "loss": 1.4240357875823975, "step": 30515 }, { "epoch": 2.4056759952700038, "grad_norm": 0.47614210844039917, "learning_rate": 0.0005, "loss": 1.304401159286499, "step": 30516 }, { "epoch": 2.405754828537643, "grad_norm": 0.47082462906837463, "learning_rate": 0.0005, "loss": 1.3726606369018555, "step": 30517 }, { "epoch": 2.405833661805282, "grad_norm": 0.500876784324646, "learning_rate": 0.0005, "loss": 1.3550505638122559, "step": 30518 }, { "epoch": 2.4059124950729207, "grad_norm": 0.5815492272377014, "learning_rate": 0.0005, "loss": 1.5343177318572998, "step": 30519 }, { "epoch": 2.40599132834056, "grad_norm": 0.5755940079689026, "learning_rate": 0.0005, "loss": 1.2371983528137207, "step": 30520 }, { "epoch": 2.4060701616081985, "grad_norm": 0.602081298828125, "learning_rate": 0.0005, "loss": 1.2402973175048828, "step": 30521 }, { "epoch": 2.4061489948758377, "grad_norm": 0.47092166543006897, "learning_rate": 0.0005, "loss": 1.3760628700256348, "step": 30522 }, { "epoch": 2.4062278281434764, "grad_norm": 0.4434009790420532, "learning_rate": 0.0005, "loss": 1.2546815872192383, "step": 30523 }, { "epoch": 2.4063066614111155, "grad_norm": 0.4777073562145233, "learning_rate": 0.0005, "loss": 1.1925654411315918, "step": 30524 }, { "epoch": 2.4063854946787546, "grad_norm": 0.4946102499961853, "learning_rate": 0.0005, "loss": 1.3150372505187988, "step": 30525 }, { "epoch": 2.4064643279463933, "grad_norm": 0.6303891539573669, "learning_rate": 0.0005, "loss": 1.5370255708694458, "step": 30526 }, { "epoch": 2.4065431612140324, "grad_norm": 0.47418972849845886, "learning_rate": 0.0005, "loss": 1.2332926988601685, "step": 30527 }, { "epoch": 2.406621994481671, "grad_norm": 0.48251277208328247, "learning_rate": 0.0005, "loss": 1.233788013458252, "step": 30528 }, { "epoch": 2.4067008277493103, "grad_norm": 0.5913625359535217, "learning_rate": 0.0005, "loss": 1.4765043258666992, "step": 30529 }, { "epoch": 2.406779661016949, "grad_norm": 0.4890443980693817, "learning_rate": 0.0005, "loss": 1.2114083766937256, "step": 30530 }, { "epoch": 2.406858494284588, "grad_norm": 0.4929152727127075, "learning_rate": 0.0005, "loss": 1.2470875978469849, "step": 30531 }, { "epoch": 2.4069373275522272, "grad_norm": 0.4520731568336487, "learning_rate": 0.0005, "loss": 1.2874752283096313, "step": 30532 }, { "epoch": 2.407016160819866, "grad_norm": 0.49768054485321045, "learning_rate": 0.0005, "loss": 1.511085033416748, "step": 30533 }, { "epoch": 2.407094994087505, "grad_norm": 0.49754658341407776, "learning_rate": 0.0005, "loss": 1.205278992652893, "step": 30534 }, { "epoch": 2.4071738273551437, "grad_norm": 0.5001226663589478, "learning_rate": 0.0005, "loss": 1.4245188236236572, "step": 30535 }, { "epoch": 2.407252660622783, "grad_norm": 0.5378178954124451, "learning_rate": 0.0005, "loss": 1.4295392036437988, "step": 30536 }, { "epoch": 2.407331493890422, "grad_norm": 0.5880661606788635, "learning_rate": 0.0005, "loss": 1.3722795248031616, "step": 30537 }, { "epoch": 2.4074103271580607, "grad_norm": 0.4769641160964966, "learning_rate": 0.0005, "loss": 1.4733835458755493, "step": 30538 }, { "epoch": 2.4074891604257, "grad_norm": 0.6446800827980042, "learning_rate": 0.0005, "loss": 1.4472602605819702, "step": 30539 }, { "epoch": 2.4075679936933385, "grad_norm": 0.5289832949638367, "learning_rate": 0.0005, "loss": 1.395003080368042, "step": 30540 }, { "epoch": 2.4076468269609776, "grad_norm": 0.4110031723976135, "learning_rate": 0.0005, "loss": 1.2803152799606323, "step": 30541 }, { "epoch": 2.4077256602286163, "grad_norm": 0.45711392164230347, "learning_rate": 0.0005, "loss": 1.320892572402954, "step": 30542 }, { "epoch": 2.4078044934962555, "grad_norm": 0.5098633170127869, "learning_rate": 0.0005, "loss": 1.2170754671096802, "step": 30543 }, { "epoch": 2.407883326763894, "grad_norm": 0.4541298449039459, "learning_rate": 0.0005, "loss": 1.3489844799041748, "step": 30544 }, { "epoch": 2.4079621600315333, "grad_norm": 0.4992615878582001, "learning_rate": 0.0005, "loss": 1.533024787902832, "step": 30545 }, { "epoch": 2.4080409932991724, "grad_norm": 0.8969396352767944, "learning_rate": 0.0005, "loss": 1.394411325454712, "step": 30546 }, { "epoch": 2.408119826566811, "grad_norm": 0.4727603793144226, "learning_rate": 0.0005, "loss": 1.2069675922393799, "step": 30547 }, { "epoch": 2.4081986598344503, "grad_norm": 0.6556857228279114, "learning_rate": 0.0005, "loss": 1.3629916906356812, "step": 30548 }, { "epoch": 2.408277493102089, "grad_norm": 0.4655889868736267, "learning_rate": 0.0005, "loss": 1.2669570446014404, "step": 30549 }, { "epoch": 2.408356326369728, "grad_norm": 0.5764383673667908, "learning_rate": 0.0005, "loss": 1.4212418794631958, "step": 30550 }, { "epoch": 2.408435159637367, "grad_norm": 0.6081236600875854, "learning_rate": 0.0005, "loss": 1.2518376111984253, "step": 30551 }, { "epoch": 2.408513992905006, "grad_norm": 0.51885986328125, "learning_rate": 0.0005, "loss": 1.2113776206970215, "step": 30552 }, { "epoch": 2.408592826172645, "grad_norm": 1.2984600067138672, "learning_rate": 0.0005, "loss": 1.292693853378296, "step": 30553 }, { "epoch": 2.4086716594402837, "grad_norm": 0.6890522837638855, "learning_rate": 0.0005, "loss": 1.4901549816131592, "step": 30554 }, { "epoch": 2.408750492707923, "grad_norm": 0.5124025940895081, "learning_rate": 0.0005, "loss": 1.234236240386963, "step": 30555 }, { "epoch": 2.4088293259755615, "grad_norm": 0.522608757019043, "learning_rate": 0.0005, "loss": 1.3095601797103882, "step": 30556 }, { "epoch": 2.4089081592432007, "grad_norm": 0.45196443796157837, "learning_rate": 0.0005, "loss": 1.3078687191009521, "step": 30557 }, { "epoch": 2.4089869925108394, "grad_norm": 0.4933314025402069, "learning_rate": 0.0005, "loss": 1.2713124752044678, "step": 30558 }, { "epoch": 2.4090658257784785, "grad_norm": 0.9212467074394226, "learning_rate": 0.0005, "loss": 1.431563377380371, "step": 30559 }, { "epoch": 2.4091446590461176, "grad_norm": 0.646995484828949, "learning_rate": 0.0005, "loss": 1.331279993057251, "step": 30560 }, { "epoch": 2.4092234923137563, "grad_norm": 0.48385727405548096, "learning_rate": 0.0005, "loss": 1.3236924409866333, "step": 30561 }, { "epoch": 2.4093023255813955, "grad_norm": 0.5172442197799683, "learning_rate": 0.0005, "loss": 1.420578122138977, "step": 30562 }, { "epoch": 2.409381158849034, "grad_norm": 0.6116666793823242, "learning_rate": 0.0005, "loss": 1.4265294075012207, "step": 30563 }, { "epoch": 2.4094599921166733, "grad_norm": 0.5431395769119263, "learning_rate": 0.0005, "loss": 1.3913729190826416, "step": 30564 }, { "epoch": 2.4095388253843124, "grad_norm": 0.5594216585159302, "learning_rate": 0.0005, "loss": 1.4701480865478516, "step": 30565 }, { "epoch": 2.409617658651951, "grad_norm": 0.5209285020828247, "learning_rate": 0.0005, "loss": 1.4254212379455566, "step": 30566 }, { "epoch": 2.4096964919195902, "grad_norm": 0.45479702949523926, "learning_rate": 0.0005, "loss": 1.3364630937576294, "step": 30567 }, { "epoch": 2.409775325187229, "grad_norm": 0.513300895690918, "learning_rate": 0.0005, "loss": 1.0914933681488037, "step": 30568 }, { "epoch": 2.409854158454868, "grad_norm": 0.6302177309989929, "learning_rate": 0.0005, "loss": 1.4708008766174316, "step": 30569 }, { "epoch": 2.4099329917225067, "grad_norm": 0.4504254162311554, "learning_rate": 0.0005, "loss": 1.2132899761199951, "step": 30570 }, { "epoch": 2.410011824990146, "grad_norm": 0.5193306803703308, "learning_rate": 0.0005, "loss": 1.2172508239746094, "step": 30571 }, { "epoch": 2.4100906582577846, "grad_norm": 0.5766510367393494, "learning_rate": 0.0005, "loss": 1.2678637504577637, "step": 30572 }, { "epoch": 2.4101694915254237, "grad_norm": 1.080938458442688, "learning_rate": 0.0005, "loss": 1.326364278793335, "step": 30573 }, { "epoch": 2.410248324793063, "grad_norm": 0.4788402020931244, "learning_rate": 0.0005, "loss": 1.3597114086151123, "step": 30574 }, { "epoch": 2.4103271580607015, "grad_norm": 0.6683976650238037, "learning_rate": 0.0005, "loss": 1.3065283298492432, "step": 30575 }, { "epoch": 2.4104059913283407, "grad_norm": 0.45508840680122375, "learning_rate": 0.0005, "loss": 1.3040461540222168, "step": 30576 }, { "epoch": 2.4104848245959793, "grad_norm": 0.5224183797836304, "learning_rate": 0.0005, "loss": 1.3315744400024414, "step": 30577 }, { "epoch": 2.4105636578636185, "grad_norm": 0.462990939617157, "learning_rate": 0.0005, "loss": 1.2377793788909912, "step": 30578 }, { "epoch": 2.4106424911312576, "grad_norm": 0.4396880567073822, "learning_rate": 0.0005, "loss": 1.2722495794296265, "step": 30579 }, { "epoch": 2.4107213243988963, "grad_norm": 0.5213751792907715, "learning_rate": 0.0005, "loss": 1.37370765209198, "step": 30580 }, { "epoch": 2.4108001576665354, "grad_norm": 0.4287598729133606, "learning_rate": 0.0005, "loss": 1.198173999786377, "step": 30581 }, { "epoch": 2.410878990934174, "grad_norm": 0.538296639919281, "learning_rate": 0.0005, "loss": 1.370062232017517, "step": 30582 }, { "epoch": 2.4109578242018133, "grad_norm": 0.4506284296512604, "learning_rate": 0.0005, "loss": 1.3831413984298706, "step": 30583 }, { "epoch": 2.411036657469452, "grad_norm": 0.5241426229476929, "learning_rate": 0.0005, "loss": 1.622454285621643, "step": 30584 }, { "epoch": 2.411115490737091, "grad_norm": 0.4868313670158386, "learning_rate": 0.0005, "loss": 1.2602953910827637, "step": 30585 }, { "epoch": 2.4111943240047298, "grad_norm": 0.4831930696964264, "learning_rate": 0.0005, "loss": 1.111852765083313, "step": 30586 }, { "epoch": 2.411273157272369, "grad_norm": 0.6731185913085938, "learning_rate": 0.0005, "loss": 1.3610198497772217, "step": 30587 }, { "epoch": 2.411351990540008, "grad_norm": 0.5919389128684998, "learning_rate": 0.0005, "loss": 1.3131306171417236, "step": 30588 }, { "epoch": 2.4114308238076467, "grad_norm": 0.44931522011756897, "learning_rate": 0.0005, "loss": 1.2468045949935913, "step": 30589 }, { "epoch": 2.411509657075286, "grad_norm": 0.4938110113143921, "learning_rate": 0.0005, "loss": 1.4144374132156372, "step": 30590 }, { "epoch": 2.4115884903429246, "grad_norm": 0.5509161949157715, "learning_rate": 0.0005, "loss": 1.424183964729309, "step": 30591 }, { "epoch": 2.4116673236105637, "grad_norm": 0.5808293223381042, "learning_rate": 0.0005, "loss": 1.422729730606079, "step": 30592 }, { "epoch": 2.411746156878203, "grad_norm": 0.5903156399726868, "learning_rate": 0.0005, "loss": 1.3276481628417969, "step": 30593 }, { "epoch": 2.4118249901458415, "grad_norm": 0.568996012210846, "learning_rate": 0.0005, "loss": 1.3048365116119385, "step": 30594 }, { "epoch": 2.4119038234134806, "grad_norm": 0.46337175369262695, "learning_rate": 0.0005, "loss": 1.3121860027313232, "step": 30595 }, { "epoch": 2.4119826566811193, "grad_norm": 0.5359238386154175, "learning_rate": 0.0005, "loss": 1.4415972232818604, "step": 30596 }, { "epoch": 2.4120614899487585, "grad_norm": 0.5521347522735596, "learning_rate": 0.0005, "loss": 1.4820923805236816, "step": 30597 }, { "epoch": 2.412140323216397, "grad_norm": 0.7418045997619629, "learning_rate": 0.0005, "loss": 1.3108839988708496, "step": 30598 }, { "epoch": 2.4122191564840363, "grad_norm": 0.5179901719093323, "learning_rate": 0.0005, "loss": 1.3724324703216553, "step": 30599 }, { "epoch": 2.412297989751675, "grad_norm": 0.4222748279571533, "learning_rate": 0.0005, "loss": 1.0425699949264526, "step": 30600 }, { "epoch": 2.412376823019314, "grad_norm": 0.4691894054412842, "learning_rate": 0.0005, "loss": 1.2094367742538452, "step": 30601 }, { "epoch": 2.4124556562869532, "grad_norm": 0.4785673916339874, "learning_rate": 0.0005, "loss": 1.4109981060028076, "step": 30602 }, { "epoch": 2.412534489554592, "grad_norm": 0.46460917592048645, "learning_rate": 0.0005, "loss": 1.341155767440796, "step": 30603 }, { "epoch": 2.412613322822231, "grad_norm": 0.4870997667312622, "learning_rate": 0.0005, "loss": 1.3567662239074707, "step": 30604 }, { "epoch": 2.4126921560898698, "grad_norm": 0.5731624960899353, "learning_rate": 0.0005, "loss": 1.3565073013305664, "step": 30605 }, { "epoch": 2.412770989357509, "grad_norm": 0.45278728008270264, "learning_rate": 0.0005, "loss": 1.2598882913589478, "step": 30606 }, { "epoch": 2.412849822625148, "grad_norm": 0.5603128671646118, "learning_rate": 0.0005, "loss": 1.4484221935272217, "step": 30607 }, { "epoch": 2.4129286558927867, "grad_norm": 1.5295919179916382, "learning_rate": 0.0005, "loss": 1.5568950176239014, "step": 30608 }, { "epoch": 2.413007489160426, "grad_norm": 0.4586344063282013, "learning_rate": 0.0005, "loss": 1.1518049240112305, "step": 30609 }, { "epoch": 2.4130863224280645, "grad_norm": 0.4774126708507538, "learning_rate": 0.0005, "loss": 1.4094089269638062, "step": 30610 }, { "epoch": 2.4131651556957037, "grad_norm": 0.5278739929199219, "learning_rate": 0.0005, "loss": 1.3561609983444214, "step": 30611 }, { "epoch": 2.4132439889633424, "grad_norm": 0.5709217190742493, "learning_rate": 0.0005, "loss": 1.4950690269470215, "step": 30612 }, { "epoch": 2.4133228222309815, "grad_norm": 0.4847857356071472, "learning_rate": 0.0005, "loss": 1.3371899127960205, "step": 30613 }, { "epoch": 2.41340165549862, "grad_norm": 0.4559144675731659, "learning_rate": 0.0005, "loss": 1.2946910858154297, "step": 30614 }, { "epoch": 2.4134804887662593, "grad_norm": 0.5684418082237244, "learning_rate": 0.0005, "loss": 1.0579994916915894, "step": 30615 }, { "epoch": 2.4135593220338984, "grad_norm": 0.6157748103141785, "learning_rate": 0.0005, "loss": 1.3619205951690674, "step": 30616 }, { "epoch": 2.413638155301537, "grad_norm": 0.5392101407051086, "learning_rate": 0.0005, "loss": 1.3072583675384521, "step": 30617 }, { "epoch": 2.4137169885691763, "grad_norm": 0.6057757139205933, "learning_rate": 0.0005, "loss": 1.430548906326294, "step": 30618 }, { "epoch": 2.413795821836815, "grad_norm": 0.40195226669311523, "learning_rate": 0.0005, "loss": 1.1601239442825317, "step": 30619 }, { "epoch": 2.413874655104454, "grad_norm": 0.5348336696624756, "learning_rate": 0.0005, "loss": 1.5066537857055664, "step": 30620 }, { "epoch": 2.4139534883720932, "grad_norm": 0.5601163506507874, "learning_rate": 0.0005, "loss": 1.2997939586639404, "step": 30621 }, { "epoch": 2.414032321639732, "grad_norm": 0.4549500346183777, "learning_rate": 0.0005, "loss": 1.3611265420913696, "step": 30622 }, { "epoch": 2.414111154907371, "grad_norm": 0.5962584614753723, "learning_rate": 0.0005, "loss": 1.2837764024734497, "step": 30623 }, { "epoch": 2.4141899881750097, "grad_norm": 0.5828949213027954, "learning_rate": 0.0005, "loss": 1.4140195846557617, "step": 30624 }, { "epoch": 2.414268821442649, "grad_norm": 0.5182405710220337, "learning_rate": 0.0005, "loss": 1.3405530452728271, "step": 30625 }, { "epoch": 2.4143476547102876, "grad_norm": 0.5433297157287598, "learning_rate": 0.0005, "loss": 1.5817170143127441, "step": 30626 }, { "epoch": 2.4144264879779267, "grad_norm": 0.5175686478614807, "learning_rate": 0.0005, "loss": 1.2523703575134277, "step": 30627 }, { "epoch": 2.4145053212455654, "grad_norm": 0.5493719577789307, "learning_rate": 0.0005, "loss": 1.166046380996704, "step": 30628 }, { "epoch": 2.4145841545132045, "grad_norm": 0.49195435643196106, "learning_rate": 0.0005, "loss": 1.2993178367614746, "step": 30629 }, { "epoch": 2.4146629877808437, "grad_norm": 0.610405445098877, "learning_rate": 0.0005, "loss": 1.4790135622024536, "step": 30630 }, { "epoch": 2.4147418210484823, "grad_norm": 0.4325409233570099, "learning_rate": 0.0005, "loss": 1.3123152256011963, "step": 30631 }, { "epoch": 2.4148206543161215, "grad_norm": 0.5121285915374756, "learning_rate": 0.0005, "loss": 1.3425630331039429, "step": 30632 }, { "epoch": 2.41489948758376, "grad_norm": 0.4910906255245209, "learning_rate": 0.0005, "loss": 1.2828325033187866, "step": 30633 }, { "epoch": 2.4149783208513993, "grad_norm": 0.44898679852485657, "learning_rate": 0.0005, "loss": 1.1227657794952393, "step": 30634 }, { "epoch": 2.4150571541190384, "grad_norm": 0.455095112323761, "learning_rate": 0.0005, "loss": 1.3106971979141235, "step": 30635 }, { "epoch": 2.415135987386677, "grad_norm": 0.5768331289291382, "learning_rate": 0.0005, "loss": 1.3806891441345215, "step": 30636 }, { "epoch": 2.4152148206543163, "grad_norm": 0.48838701844215393, "learning_rate": 0.0005, "loss": 1.2130632400512695, "step": 30637 }, { "epoch": 2.415293653921955, "grad_norm": 0.4592244029045105, "learning_rate": 0.0005, "loss": 1.3179519176483154, "step": 30638 }, { "epoch": 2.415372487189594, "grad_norm": 0.49806854128837585, "learning_rate": 0.0005, "loss": 1.3059885501861572, "step": 30639 }, { "epoch": 2.4154513204572328, "grad_norm": 1.2195475101470947, "learning_rate": 0.0005, "loss": 1.1037428379058838, "step": 30640 }, { "epoch": 2.415530153724872, "grad_norm": 0.7803581953048706, "learning_rate": 0.0005, "loss": 1.2208852767944336, "step": 30641 }, { "epoch": 2.415608986992511, "grad_norm": 0.3898411691188812, "learning_rate": 0.0005, "loss": 1.104768991470337, "step": 30642 }, { "epoch": 2.4156878202601497, "grad_norm": 0.5891496539115906, "learning_rate": 0.0005, "loss": 1.2898890972137451, "step": 30643 }, { "epoch": 2.415766653527789, "grad_norm": 0.4938703775405884, "learning_rate": 0.0005, "loss": 1.5854544639587402, "step": 30644 }, { "epoch": 2.4158454867954275, "grad_norm": 0.496986985206604, "learning_rate": 0.0005, "loss": 1.2595252990722656, "step": 30645 }, { "epoch": 2.4159243200630667, "grad_norm": 0.4477407932281494, "learning_rate": 0.0005, "loss": 1.3052858114242554, "step": 30646 }, { "epoch": 2.4160031533307054, "grad_norm": 0.5998209714889526, "learning_rate": 0.0005, "loss": 1.5795317888259888, "step": 30647 }, { "epoch": 2.4160819865983445, "grad_norm": 0.6661169528961182, "learning_rate": 0.0005, "loss": 1.4787359237670898, "step": 30648 }, { "epoch": 2.4161608198659836, "grad_norm": 0.4609835147857666, "learning_rate": 0.0005, "loss": 1.3446701765060425, "step": 30649 }, { "epoch": 2.4162396531336223, "grad_norm": 0.46861833333969116, "learning_rate": 0.0005, "loss": 1.2527384757995605, "step": 30650 }, { "epoch": 2.4163184864012615, "grad_norm": 0.4925483763217926, "learning_rate": 0.0005, "loss": 1.2322232723236084, "step": 30651 }, { "epoch": 2.4163973196689, "grad_norm": 0.5488778352737427, "learning_rate": 0.0005, "loss": 1.3305771350860596, "step": 30652 }, { "epoch": 2.4164761529365393, "grad_norm": 0.5945791602134705, "learning_rate": 0.0005, "loss": 1.3106331825256348, "step": 30653 }, { "epoch": 2.4165549862041784, "grad_norm": 0.5071091651916504, "learning_rate": 0.0005, "loss": 1.3893128633499146, "step": 30654 }, { "epoch": 2.416633819471817, "grad_norm": 0.5201031565666199, "learning_rate": 0.0005, "loss": 1.1247777938842773, "step": 30655 }, { "epoch": 2.4167126527394562, "grad_norm": 0.5926125645637512, "learning_rate": 0.0005, "loss": 1.359976053237915, "step": 30656 }, { "epoch": 2.416791486007095, "grad_norm": 0.4981571137905121, "learning_rate": 0.0005, "loss": 1.4163413047790527, "step": 30657 }, { "epoch": 2.416870319274734, "grad_norm": 0.4330402910709381, "learning_rate": 0.0005, "loss": 1.2749264240264893, "step": 30658 }, { "epoch": 2.4169491525423727, "grad_norm": 0.47200289368629456, "learning_rate": 0.0005, "loss": 1.4037312269210815, "step": 30659 }, { "epoch": 2.417027985810012, "grad_norm": 0.5627111196517944, "learning_rate": 0.0005, "loss": 1.378817081451416, "step": 30660 }, { "epoch": 2.4171068190776506, "grad_norm": 0.4699656069278717, "learning_rate": 0.0005, "loss": 1.223679542541504, "step": 30661 }, { "epoch": 2.4171856523452897, "grad_norm": 1.3224619626998901, "learning_rate": 0.0005, "loss": 1.2166721820831299, "step": 30662 }, { "epoch": 2.417264485612929, "grad_norm": 0.5716208219528198, "learning_rate": 0.0005, "loss": 1.4162561893463135, "step": 30663 }, { "epoch": 2.4173433188805675, "grad_norm": 0.530595064163208, "learning_rate": 0.0005, "loss": 1.41069495677948, "step": 30664 }, { "epoch": 2.4174221521482067, "grad_norm": 0.47567009925842285, "learning_rate": 0.0005, "loss": 1.160891056060791, "step": 30665 }, { "epoch": 2.4175009854158453, "grad_norm": 0.6162993907928467, "learning_rate": 0.0005, "loss": 1.4513540267944336, "step": 30666 }, { "epoch": 2.4175798186834845, "grad_norm": 0.4978019595146179, "learning_rate": 0.0005, "loss": 1.3056349754333496, "step": 30667 }, { "epoch": 2.4176586519511236, "grad_norm": 0.5511399507522583, "learning_rate": 0.0005, "loss": 1.3467040061950684, "step": 30668 }, { "epoch": 2.4177374852187623, "grad_norm": 0.5955223441123962, "learning_rate": 0.0005, "loss": 1.3885068893432617, "step": 30669 }, { "epoch": 2.4178163184864014, "grad_norm": 0.5969151258468628, "learning_rate": 0.0005, "loss": 1.3363133668899536, "step": 30670 }, { "epoch": 2.41789515175404, "grad_norm": 0.504571795463562, "learning_rate": 0.0005, "loss": 1.3453834056854248, "step": 30671 }, { "epoch": 2.4179739850216793, "grad_norm": 0.5147382616996765, "learning_rate": 0.0005, "loss": 1.5107249021530151, "step": 30672 }, { "epoch": 2.418052818289318, "grad_norm": 0.6447324752807617, "learning_rate": 0.0005, "loss": 1.389269471168518, "step": 30673 }, { "epoch": 2.418131651556957, "grad_norm": 0.5733769536018372, "learning_rate": 0.0005, "loss": 1.495435118675232, "step": 30674 }, { "epoch": 2.4182104848245958, "grad_norm": 0.5330589413642883, "learning_rate": 0.0005, "loss": 1.4192478656768799, "step": 30675 }, { "epoch": 2.418289318092235, "grad_norm": 0.5088978409767151, "learning_rate": 0.0005, "loss": 1.1943528652191162, "step": 30676 }, { "epoch": 2.418368151359874, "grad_norm": 0.6003183126449585, "learning_rate": 0.0005, "loss": 1.2924811840057373, "step": 30677 }, { "epoch": 2.4184469846275127, "grad_norm": 0.46052733063697815, "learning_rate": 0.0005, "loss": 1.3941494226455688, "step": 30678 }, { "epoch": 2.418525817895152, "grad_norm": 0.4899691641330719, "learning_rate": 0.0005, "loss": 1.3010740280151367, "step": 30679 }, { "epoch": 2.4186046511627906, "grad_norm": 0.49206992983818054, "learning_rate": 0.0005, "loss": 1.3340131044387817, "step": 30680 }, { "epoch": 2.4186834844304297, "grad_norm": 0.5546557307243347, "learning_rate": 0.0005, "loss": 1.5218164920806885, "step": 30681 }, { "epoch": 2.418762317698069, "grad_norm": 0.6189365386962891, "learning_rate": 0.0005, "loss": 1.4386811256408691, "step": 30682 }, { "epoch": 2.4188411509657075, "grad_norm": 0.5633648037910461, "learning_rate": 0.0005, "loss": 1.2959527969360352, "step": 30683 }, { "epoch": 2.4189199842333466, "grad_norm": 0.5169934034347534, "learning_rate": 0.0005, "loss": 1.100334644317627, "step": 30684 }, { "epoch": 2.4189988175009853, "grad_norm": 0.5164675712585449, "learning_rate": 0.0005, "loss": 1.5335001945495605, "step": 30685 }, { "epoch": 2.4190776507686245, "grad_norm": 0.4149818420410156, "learning_rate": 0.0005, "loss": 1.3925518989562988, "step": 30686 }, { "epoch": 2.419156484036263, "grad_norm": 0.5070524215698242, "learning_rate": 0.0005, "loss": 1.256146788597107, "step": 30687 }, { "epoch": 2.4192353173039023, "grad_norm": 0.5492910742759705, "learning_rate": 0.0005, "loss": 1.3812196254730225, "step": 30688 }, { "epoch": 2.419314150571541, "grad_norm": 0.4844551682472229, "learning_rate": 0.0005, "loss": 1.3787453174591064, "step": 30689 }, { "epoch": 2.41939298383918, "grad_norm": 0.47697868943214417, "learning_rate": 0.0005, "loss": 1.374975562095642, "step": 30690 }, { "epoch": 2.4194718171068192, "grad_norm": 0.462016224861145, "learning_rate": 0.0005, "loss": 1.1839708089828491, "step": 30691 }, { "epoch": 2.419550650374458, "grad_norm": 0.4833526909351349, "learning_rate": 0.0005, "loss": 1.4747188091278076, "step": 30692 }, { "epoch": 2.419629483642097, "grad_norm": 0.4537002742290497, "learning_rate": 0.0005, "loss": 1.1511938571929932, "step": 30693 }, { "epoch": 2.4197083169097358, "grad_norm": 0.67006516456604, "learning_rate": 0.0005, "loss": 1.6263508796691895, "step": 30694 }, { "epoch": 2.419787150177375, "grad_norm": 0.5136757493019104, "learning_rate": 0.0005, "loss": 1.3185484409332275, "step": 30695 }, { "epoch": 2.419865983445014, "grad_norm": 0.680746853351593, "learning_rate": 0.0005, "loss": 1.5716552734375, "step": 30696 }, { "epoch": 2.4199448167126527, "grad_norm": 0.4881090223789215, "learning_rate": 0.0005, "loss": 1.3155391216278076, "step": 30697 }, { "epoch": 2.420023649980292, "grad_norm": 0.43732813000679016, "learning_rate": 0.0005, "loss": 1.24639892578125, "step": 30698 }, { "epoch": 2.4201024832479305, "grad_norm": 0.46682530641555786, "learning_rate": 0.0005, "loss": 1.317072868347168, "step": 30699 }, { "epoch": 2.4201813165155697, "grad_norm": 0.5442099571228027, "learning_rate": 0.0005, "loss": 1.5433769226074219, "step": 30700 }, { "epoch": 2.4202601497832084, "grad_norm": 0.5064399242401123, "learning_rate": 0.0005, "loss": 1.1702213287353516, "step": 30701 }, { "epoch": 2.4203389830508475, "grad_norm": 0.5798147320747375, "learning_rate": 0.0005, "loss": 1.3225146532058716, "step": 30702 }, { "epoch": 2.420417816318486, "grad_norm": 0.5154101848602295, "learning_rate": 0.0005, "loss": 1.5678863525390625, "step": 30703 }, { "epoch": 2.4204966495861253, "grad_norm": 0.4113785922527313, "learning_rate": 0.0005, "loss": 1.1693294048309326, "step": 30704 }, { "epoch": 2.4205754828537644, "grad_norm": 0.42965495586395264, "learning_rate": 0.0005, "loss": 1.3182613849639893, "step": 30705 }, { "epoch": 2.420654316121403, "grad_norm": 0.45730188488960266, "learning_rate": 0.0005, "loss": 1.3516674041748047, "step": 30706 }, { "epoch": 2.4207331493890423, "grad_norm": 0.41763803362846375, "learning_rate": 0.0005, "loss": 1.1365361213684082, "step": 30707 }, { "epoch": 2.420811982656681, "grad_norm": 0.5743700861930847, "learning_rate": 0.0005, "loss": 1.524993896484375, "step": 30708 }, { "epoch": 2.42089081592432, "grad_norm": 1.046927809715271, "learning_rate": 0.0005, "loss": 1.349706768989563, "step": 30709 }, { "epoch": 2.4209696491919592, "grad_norm": 0.5110320448875427, "learning_rate": 0.0005, "loss": 1.2363314628601074, "step": 30710 }, { "epoch": 2.421048482459598, "grad_norm": 0.5289781093597412, "learning_rate": 0.0005, "loss": 1.392580509185791, "step": 30711 }, { "epoch": 2.421127315727237, "grad_norm": 0.5547598600387573, "learning_rate": 0.0005, "loss": 1.350931167602539, "step": 30712 }, { "epoch": 2.4212061489948757, "grad_norm": 0.6290407776832581, "learning_rate": 0.0005, "loss": 1.5037891864776611, "step": 30713 }, { "epoch": 2.421284982262515, "grad_norm": 0.5547909140586853, "learning_rate": 0.0005, "loss": 1.2161083221435547, "step": 30714 }, { "epoch": 2.4213638155301536, "grad_norm": 0.5949156880378723, "learning_rate": 0.0005, "loss": 1.4053608179092407, "step": 30715 }, { "epoch": 2.4214426487977927, "grad_norm": 0.6616426706314087, "learning_rate": 0.0005, "loss": 1.4111003875732422, "step": 30716 }, { "epoch": 2.4215214820654314, "grad_norm": 0.5014055371284485, "learning_rate": 0.0005, "loss": 1.3590660095214844, "step": 30717 }, { "epoch": 2.4216003153330705, "grad_norm": 0.4184182286262512, "learning_rate": 0.0005, "loss": 1.1472580432891846, "step": 30718 }, { "epoch": 2.4216791486007097, "grad_norm": 0.5758016705513, "learning_rate": 0.0005, "loss": 1.2916972637176514, "step": 30719 }, { "epoch": 2.4217579818683483, "grad_norm": 0.6011320948600769, "learning_rate": 0.0005, "loss": 1.3674601316452026, "step": 30720 }, { "epoch": 2.4218368151359875, "grad_norm": 0.5295847058296204, "learning_rate": 0.0005, "loss": 1.170957088470459, "step": 30721 }, { "epoch": 2.421915648403626, "grad_norm": 0.5323660373687744, "learning_rate": 0.0005, "loss": 1.6005034446716309, "step": 30722 }, { "epoch": 2.4219944816712653, "grad_norm": 0.5770719051361084, "learning_rate": 0.0005, "loss": 1.373273491859436, "step": 30723 }, { "epoch": 2.4220733149389044, "grad_norm": 0.6306474804878235, "learning_rate": 0.0005, "loss": 1.4991388320922852, "step": 30724 }, { "epoch": 2.422152148206543, "grad_norm": 0.5813411474227905, "learning_rate": 0.0005, "loss": 1.4565942287445068, "step": 30725 }, { "epoch": 2.4222309814741823, "grad_norm": 0.5480951070785522, "learning_rate": 0.0005, "loss": 1.3807120323181152, "step": 30726 }, { "epoch": 2.422309814741821, "grad_norm": 0.6176353693008423, "learning_rate": 0.0005, "loss": 1.4214870929718018, "step": 30727 }, { "epoch": 2.42238864800946, "grad_norm": 0.5143252611160278, "learning_rate": 0.0005, "loss": 1.3279138803482056, "step": 30728 }, { "epoch": 2.4224674812770988, "grad_norm": 0.48681777715682983, "learning_rate": 0.0005, "loss": 1.364480972290039, "step": 30729 }, { "epoch": 2.422546314544738, "grad_norm": 0.5052858591079712, "learning_rate": 0.0005, "loss": 1.1407647132873535, "step": 30730 }, { "epoch": 2.4226251478123766, "grad_norm": 0.5422652363777161, "learning_rate": 0.0005, "loss": 1.387688398361206, "step": 30731 }, { "epoch": 2.4227039810800157, "grad_norm": 0.5432325601577759, "learning_rate": 0.0005, "loss": 1.742666482925415, "step": 30732 }, { "epoch": 2.422782814347655, "grad_norm": 0.4575015604496002, "learning_rate": 0.0005, "loss": 1.1904571056365967, "step": 30733 }, { "epoch": 2.4228616476152935, "grad_norm": 0.5334998965263367, "learning_rate": 0.0005, "loss": 1.4167407751083374, "step": 30734 }, { "epoch": 2.4229404808829327, "grad_norm": 0.4160500168800354, "learning_rate": 0.0005, "loss": 1.4469828605651855, "step": 30735 }, { "epoch": 2.4230193141505714, "grad_norm": 0.6554198265075684, "learning_rate": 0.0005, "loss": 1.4897536039352417, "step": 30736 }, { "epoch": 2.4230981474182105, "grad_norm": 0.6958966255187988, "learning_rate": 0.0005, "loss": 1.3145427703857422, "step": 30737 }, { "epoch": 2.4231769806858496, "grad_norm": 0.5289972424507141, "learning_rate": 0.0005, "loss": 1.304790735244751, "step": 30738 }, { "epoch": 2.4232558139534883, "grad_norm": 0.5449002981185913, "learning_rate": 0.0005, "loss": 1.2680690288543701, "step": 30739 }, { "epoch": 2.4233346472211275, "grad_norm": 0.515397310256958, "learning_rate": 0.0005, "loss": 1.3596746921539307, "step": 30740 }, { "epoch": 2.423413480488766, "grad_norm": 0.8241817355155945, "learning_rate": 0.0005, "loss": 1.3294672966003418, "step": 30741 }, { "epoch": 2.4234923137564053, "grad_norm": 0.566170334815979, "learning_rate": 0.0005, "loss": 1.49712073802948, "step": 30742 }, { "epoch": 2.423571147024044, "grad_norm": 0.49271976947784424, "learning_rate": 0.0005, "loss": 1.3676804304122925, "step": 30743 }, { "epoch": 2.423649980291683, "grad_norm": 0.6253694891929626, "learning_rate": 0.0005, "loss": 1.5727314949035645, "step": 30744 }, { "epoch": 2.423728813559322, "grad_norm": 0.6021379828453064, "learning_rate": 0.0005, "loss": 1.3071835041046143, "step": 30745 }, { "epoch": 2.423807646826961, "grad_norm": 0.5173640251159668, "learning_rate": 0.0005, "loss": 1.5372025966644287, "step": 30746 }, { "epoch": 2.4238864800946, "grad_norm": 0.5423717498779297, "learning_rate": 0.0005, "loss": 1.3741331100463867, "step": 30747 }, { "epoch": 2.4239653133622387, "grad_norm": 0.4365413784980774, "learning_rate": 0.0005, "loss": 1.2350631952285767, "step": 30748 }, { "epoch": 2.424044146629878, "grad_norm": 0.5321832299232483, "learning_rate": 0.0005, "loss": 1.3952583074569702, "step": 30749 }, { "epoch": 2.4241229798975166, "grad_norm": 0.8235259056091309, "learning_rate": 0.0005, "loss": 1.4582945108413696, "step": 30750 }, { "epoch": 2.4242018131651557, "grad_norm": 0.535898745059967, "learning_rate": 0.0005, "loss": 1.4478093385696411, "step": 30751 }, { "epoch": 2.424280646432795, "grad_norm": 0.4941297173500061, "learning_rate": 0.0005, "loss": 1.2243633270263672, "step": 30752 }, { "epoch": 2.4243594797004335, "grad_norm": 0.5022234916687012, "learning_rate": 0.0005, "loss": 1.4169858694076538, "step": 30753 }, { "epoch": 2.4244383129680727, "grad_norm": 0.45607662200927734, "learning_rate": 0.0005, "loss": 1.3734869956970215, "step": 30754 }, { "epoch": 2.4245171462357114, "grad_norm": 0.47585728764533997, "learning_rate": 0.0005, "loss": 1.3682849407196045, "step": 30755 }, { "epoch": 2.4245959795033505, "grad_norm": 0.4882916510105133, "learning_rate": 0.0005, "loss": 1.3524315357208252, "step": 30756 }, { "epoch": 2.424674812770989, "grad_norm": 0.45717406272888184, "learning_rate": 0.0005, "loss": 1.2806124687194824, "step": 30757 }, { "epoch": 2.4247536460386283, "grad_norm": 0.4708823263645172, "learning_rate": 0.0005, "loss": 1.5354818105697632, "step": 30758 }, { "epoch": 2.424832479306267, "grad_norm": 0.48817381262779236, "learning_rate": 0.0005, "loss": 1.3906598091125488, "step": 30759 }, { "epoch": 2.424911312573906, "grad_norm": 0.5252518653869629, "learning_rate": 0.0005, "loss": 1.2447941303253174, "step": 30760 }, { "epoch": 2.4249901458415453, "grad_norm": 0.5647013783454895, "learning_rate": 0.0005, "loss": 1.3585355281829834, "step": 30761 }, { "epoch": 2.425068979109184, "grad_norm": 0.4827274680137634, "learning_rate": 0.0005, "loss": 1.3483253717422485, "step": 30762 }, { "epoch": 2.425147812376823, "grad_norm": 0.6184762120246887, "learning_rate": 0.0005, "loss": 1.2506544589996338, "step": 30763 }, { "epoch": 2.4252266456444618, "grad_norm": 0.4431993067264557, "learning_rate": 0.0005, "loss": 1.3965706825256348, "step": 30764 }, { "epoch": 2.425305478912101, "grad_norm": 0.6010844111442566, "learning_rate": 0.0005, "loss": 1.7373549938201904, "step": 30765 }, { "epoch": 2.42538431217974, "grad_norm": 0.5112842321395874, "learning_rate": 0.0005, "loss": 1.3509025573730469, "step": 30766 }, { "epoch": 2.4254631454473787, "grad_norm": 0.531225323677063, "learning_rate": 0.0005, "loss": 1.2140040397644043, "step": 30767 }, { "epoch": 2.425541978715018, "grad_norm": 0.8340159058570862, "learning_rate": 0.0005, "loss": 1.4611577987670898, "step": 30768 }, { "epoch": 2.4256208119826566, "grad_norm": 0.4568634033203125, "learning_rate": 0.0005, "loss": 1.4320588111877441, "step": 30769 }, { "epoch": 2.4256996452502957, "grad_norm": 0.4484972357749939, "learning_rate": 0.0005, "loss": 1.2318601608276367, "step": 30770 }, { "epoch": 2.4257784785179344, "grad_norm": 0.32732680439949036, "learning_rate": 0.0005, "loss": 0.9595367908477783, "step": 30771 }, { "epoch": 2.4258573117855735, "grad_norm": 0.47984129190444946, "learning_rate": 0.0005, "loss": 1.2034790515899658, "step": 30772 }, { "epoch": 2.4259361450532126, "grad_norm": 0.48422861099243164, "learning_rate": 0.0005, "loss": 1.3739769458770752, "step": 30773 }, { "epoch": 2.4260149783208513, "grad_norm": 0.4714299440383911, "learning_rate": 0.0005, "loss": 1.3412792682647705, "step": 30774 }, { "epoch": 2.4260938115884905, "grad_norm": 0.6210565567016602, "learning_rate": 0.0005, "loss": 1.459236979484558, "step": 30775 }, { "epoch": 2.426172644856129, "grad_norm": 0.5007604360580444, "learning_rate": 0.0005, "loss": 1.2656421661376953, "step": 30776 }, { "epoch": 2.4262514781237683, "grad_norm": 0.3732263445854187, "learning_rate": 0.0005, "loss": 1.1269268989562988, "step": 30777 }, { "epoch": 2.426330311391407, "grad_norm": 0.46112680435180664, "learning_rate": 0.0005, "loss": 1.4107036590576172, "step": 30778 }, { "epoch": 2.426409144659046, "grad_norm": 0.6643993854522705, "learning_rate": 0.0005, "loss": 1.296168327331543, "step": 30779 }, { "epoch": 2.4264879779266852, "grad_norm": 0.513283908367157, "learning_rate": 0.0005, "loss": 1.3392863273620605, "step": 30780 }, { "epoch": 2.426566811194324, "grad_norm": 0.48412707448005676, "learning_rate": 0.0005, "loss": 1.151773452758789, "step": 30781 }, { "epoch": 2.426645644461963, "grad_norm": 0.5860255360603333, "learning_rate": 0.0005, "loss": 1.4703831672668457, "step": 30782 }, { "epoch": 2.4267244777296018, "grad_norm": 0.5540887713432312, "learning_rate": 0.0005, "loss": 1.5506374835968018, "step": 30783 }, { "epoch": 2.426803310997241, "grad_norm": 0.5466326475143433, "learning_rate": 0.0005, "loss": 1.3672983646392822, "step": 30784 }, { "epoch": 2.42688214426488, "grad_norm": 0.587797224521637, "learning_rate": 0.0005, "loss": 1.4125093221664429, "step": 30785 }, { "epoch": 2.4269609775325187, "grad_norm": 0.5282521843910217, "learning_rate": 0.0005, "loss": 1.2464075088500977, "step": 30786 }, { "epoch": 2.427039810800158, "grad_norm": 0.6247408390045166, "learning_rate": 0.0005, "loss": 1.3407225608825684, "step": 30787 }, { "epoch": 2.4271186440677965, "grad_norm": 0.5147194862365723, "learning_rate": 0.0005, "loss": 1.3923876285552979, "step": 30788 }, { "epoch": 2.4271974773354357, "grad_norm": 0.5271977186203003, "learning_rate": 0.0005, "loss": 1.5243343114852905, "step": 30789 }, { "epoch": 2.4272763106030744, "grad_norm": 0.5644568204879761, "learning_rate": 0.0005, "loss": 1.1258846521377563, "step": 30790 }, { "epoch": 2.4273551438707135, "grad_norm": 0.5538649559020996, "learning_rate": 0.0005, "loss": 1.1798770427703857, "step": 30791 }, { "epoch": 2.427433977138352, "grad_norm": 0.5339332818984985, "learning_rate": 0.0005, "loss": 1.2490836381912231, "step": 30792 }, { "epoch": 2.4275128104059913, "grad_norm": 0.5561832785606384, "learning_rate": 0.0005, "loss": 1.4370534420013428, "step": 30793 }, { "epoch": 2.4275916436736305, "grad_norm": 0.6830036044120789, "learning_rate": 0.0005, "loss": 1.5191190242767334, "step": 30794 }, { "epoch": 2.427670476941269, "grad_norm": 0.4150056838989258, "learning_rate": 0.0005, "loss": 1.3024592399597168, "step": 30795 }, { "epoch": 2.4277493102089083, "grad_norm": 0.482028067111969, "learning_rate": 0.0005, "loss": 1.3136837482452393, "step": 30796 }, { "epoch": 2.427828143476547, "grad_norm": 0.3841225206851959, "learning_rate": 0.0005, "loss": 1.0999921560287476, "step": 30797 }, { "epoch": 2.427906976744186, "grad_norm": 0.4897582232952118, "learning_rate": 0.0005, "loss": 1.4719280004501343, "step": 30798 }, { "epoch": 2.4279858100118252, "grad_norm": 0.4960445463657379, "learning_rate": 0.0005, "loss": 1.1253305673599243, "step": 30799 }, { "epoch": 2.428064643279464, "grad_norm": 0.5509291887283325, "learning_rate": 0.0005, "loss": 1.3515164852142334, "step": 30800 }, { "epoch": 2.428143476547103, "grad_norm": 0.5737284421920776, "learning_rate": 0.0005, "loss": 1.5176419019699097, "step": 30801 }, { "epoch": 2.4282223098147417, "grad_norm": 0.4966351091861725, "learning_rate": 0.0005, "loss": 1.4204308986663818, "step": 30802 }, { "epoch": 2.428301143082381, "grad_norm": 0.4872857630252838, "learning_rate": 0.0005, "loss": 1.2480511665344238, "step": 30803 }, { "epoch": 2.4283799763500196, "grad_norm": 0.49078431725502014, "learning_rate": 0.0005, "loss": 1.2855713367462158, "step": 30804 }, { "epoch": 2.4284588096176587, "grad_norm": 0.6020264029502869, "learning_rate": 0.0005, "loss": 1.42283296585083, "step": 30805 }, { "epoch": 2.4285376428852974, "grad_norm": 1.5955145359039307, "learning_rate": 0.0005, "loss": 1.4404863119125366, "step": 30806 }, { "epoch": 2.4286164761529365, "grad_norm": 0.5582624673843384, "learning_rate": 0.0005, "loss": 1.3925933837890625, "step": 30807 }, { "epoch": 2.4286953094205757, "grad_norm": 0.49774283170700073, "learning_rate": 0.0005, "loss": 1.3521833419799805, "step": 30808 }, { "epoch": 2.4287741426882143, "grad_norm": 0.5614212155342102, "learning_rate": 0.0005, "loss": 1.4497175216674805, "step": 30809 }, { "epoch": 2.4288529759558535, "grad_norm": 0.56398606300354, "learning_rate": 0.0005, "loss": 1.305046796798706, "step": 30810 }, { "epoch": 2.428931809223492, "grad_norm": 0.42362070083618164, "learning_rate": 0.0005, "loss": 1.2929636240005493, "step": 30811 }, { "epoch": 2.4290106424911313, "grad_norm": 0.6080392003059387, "learning_rate": 0.0005, "loss": 1.4883779287338257, "step": 30812 }, { "epoch": 2.4290894757587704, "grad_norm": 0.4736385643482208, "learning_rate": 0.0005, "loss": 1.3450191020965576, "step": 30813 }, { "epoch": 2.429168309026409, "grad_norm": 0.4625619351863861, "learning_rate": 0.0005, "loss": 1.3435289859771729, "step": 30814 }, { "epoch": 2.4292471422940483, "grad_norm": 0.5252005457878113, "learning_rate": 0.0005, "loss": 1.593808889389038, "step": 30815 }, { "epoch": 2.429325975561687, "grad_norm": 0.465986430644989, "learning_rate": 0.0005, "loss": 1.1370606422424316, "step": 30816 }, { "epoch": 2.429404808829326, "grad_norm": 0.42880886793136597, "learning_rate": 0.0005, "loss": 1.3236911296844482, "step": 30817 }, { "epoch": 2.4294836420969648, "grad_norm": 0.5631487369537354, "learning_rate": 0.0005, "loss": 1.3047925233840942, "step": 30818 }, { "epoch": 2.429562475364604, "grad_norm": 0.45931920409202576, "learning_rate": 0.0005, "loss": 1.1923065185546875, "step": 30819 }, { "epoch": 2.4296413086322426, "grad_norm": 0.6695030331611633, "learning_rate": 0.0005, "loss": 1.4957062005996704, "step": 30820 }, { "epoch": 2.4297201418998817, "grad_norm": 0.5849955081939697, "learning_rate": 0.0005, "loss": 1.3498138189315796, "step": 30821 }, { "epoch": 2.429798975167521, "grad_norm": 0.6237931847572327, "learning_rate": 0.0005, "loss": 1.4689488410949707, "step": 30822 }, { "epoch": 2.4298778084351595, "grad_norm": 0.6647284626960754, "learning_rate": 0.0005, "loss": 1.5588934421539307, "step": 30823 }, { "epoch": 2.4299566417027987, "grad_norm": 0.6816727519035339, "learning_rate": 0.0005, "loss": 1.2050604820251465, "step": 30824 }, { "epoch": 2.4300354749704374, "grad_norm": 0.6261335015296936, "learning_rate": 0.0005, "loss": 1.2222859859466553, "step": 30825 }, { "epoch": 2.4301143082380765, "grad_norm": 0.4505392014980316, "learning_rate": 0.0005, "loss": 1.2547340393066406, "step": 30826 }, { "epoch": 2.4301931415057156, "grad_norm": 0.618064284324646, "learning_rate": 0.0005, "loss": 1.6334736347198486, "step": 30827 }, { "epoch": 2.4302719747733543, "grad_norm": 0.45994070172309875, "learning_rate": 0.0005, "loss": 1.3035755157470703, "step": 30828 }, { "epoch": 2.4303508080409935, "grad_norm": 0.5690991282463074, "learning_rate": 0.0005, "loss": 1.324486255645752, "step": 30829 }, { "epoch": 2.430429641308632, "grad_norm": 2.0245018005371094, "learning_rate": 0.0005, "loss": 1.2255992889404297, "step": 30830 }, { "epoch": 2.4305084745762713, "grad_norm": 0.5519663095474243, "learning_rate": 0.0005, "loss": 1.468335747718811, "step": 30831 }, { "epoch": 2.43058730784391, "grad_norm": 0.5721049904823303, "learning_rate": 0.0005, "loss": 1.353319525718689, "step": 30832 }, { "epoch": 2.430666141111549, "grad_norm": 0.6166370511054993, "learning_rate": 0.0005, "loss": 1.4164142608642578, "step": 30833 }, { "epoch": 2.430744974379188, "grad_norm": 0.5780723690986633, "learning_rate": 0.0005, "loss": 1.2437632083892822, "step": 30834 }, { "epoch": 2.430823807646827, "grad_norm": 0.5589006543159485, "learning_rate": 0.0005, "loss": 1.4502413272857666, "step": 30835 }, { "epoch": 2.430902640914466, "grad_norm": 0.5411574244499207, "learning_rate": 0.0005, "loss": 1.3359392881393433, "step": 30836 }, { "epoch": 2.4309814741821048, "grad_norm": 0.5250681638717651, "learning_rate": 0.0005, "loss": 1.3233611583709717, "step": 30837 }, { "epoch": 2.431060307449744, "grad_norm": 0.543354868888855, "learning_rate": 0.0005, "loss": 1.2614150047302246, "step": 30838 }, { "epoch": 2.4311391407173826, "grad_norm": 0.7104278802871704, "learning_rate": 0.0005, "loss": 1.512800931930542, "step": 30839 }, { "epoch": 2.4312179739850217, "grad_norm": 0.41277989745140076, "learning_rate": 0.0005, "loss": 1.4181379079818726, "step": 30840 }, { "epoch": 2.431296807252661, "grad_norm": 0.5353971123695374, "learning_rate": 0.0005, "loss": 1.2656432390213013, "step": 30841 }, { "epoch": 2.4313756405202995, "grad_norm": 0.7761315107345581, "learning_rate": 0.0005, "loss": 1.3314480781555176, "step": 30842 }, { "epoch": 2.4314544737879387, "grad_norm": 0.48204296827316284, "learning_rate": 0.0005, "loss": 1.2777183055877686, "step": 30843 }, { "epoch": 2.4315333070555774, "grad_norm": 0.5053282380104065, "learning_rate": 0.0005, "loss": 1.2343255281448364, "step": 30844 }, { "epoch": 2.4316121403232165, "grad_norm": 0.4917707145214081, "learning_rate": 0.0005, "loss": 1.631880521774292, "step": 30845 }, { "epoch": 2.431690973590855, "grad_norm": 0.415274977684021, "learning_rate": 0.0005, "loss": 1.3581280708312988, "step": 30846 }, { "epoch": 2.4317698068584943, "grad_norm": 0.6044759154319763, "learning_rate": 0.0005, "loss": 1.4452842473983765, "step": 30847 }, { "epoch": 2.431848640126133, "grad_norm": 0.5251237750053406, "learning_rate": 0.0005, "loss": 1.5135784149169922, "step": 30848 }, { "epoch": 2.431927473393772, "grad_norm": 0.5439318418502808, "learning_rate": 0.0005, "loss": 1.2344751358032227, "step": 30849 }, { "epoch": 2.4320063066614113, "grad_norm": 0.4633479714393616, "learning_rate": 0.0005, "loss": 1.3142739534378052, "step": 30850 }, { "epoch": 2.43208513992905, "grad_norm": 0.5820819735527039, "learning_rate": 0.0005, "loss": 1.277217149734497, "step": 30851 }, { "epoch": 2.432163973196689, "grad_norm": 0.5368884205818176, "learning_rate": 0.0005, "loss": 1.423484206199646, "step": 30852 }, { "epoch": 2.4322428064643278, "grad_norm": 0.5221967101097107, "learning_rate": 0.0005, "loss": 1.3812241554260254, "step": 30853 }, { "epoch": 2.432321639731967, "grad_norm": 0.6524625420570374, "learning_rate": 0.0005, "loss": 1.3192267417907715, "step": 30854 }, { "epoch": 2.432400472999606, "grad_norm": 0.5215424299240112, "learning_rate": 0.0005, "loss": 1.4409334659576416, "step": 30855 }, { "epoch": 2.4324793062672447, "grad_norm": 0.4581533670425415, "learning_rate": 0.0005, "loss": 1.2174948453903198, "step": 30856 }, { "epoch": 2.432558139534884, "grad_norm": 0.5173478722572327, "learning_rate": 0.0005, "loss": 1.3590441942214966, "step": 30857 }, { "epoch": 2.4326369728025226, "grad_norm": 0.6815463304519653, "learning_rate": 0.0005, "loss": 1.2854174375534058, "step": 30858 }, { "epoch": 2.4327158060701617, "grad_norm": 0.6703916788101196, "learning_rate": 0.0005, "loss": 1.3338468074798584, "step": 30859 }, { "epoch": 2.4327946393378004, "grad_norm": 0.5353094935417175, "learning_rate": 0.0005, "loss": 1.2099738121032715, "step": 30860 }, { "epoch": 2.4328734726054395, "grad_norm": 0.47005781531333923, "learning_rate": 0.0005, "loss": 1.2723482847213745, "step": 30861 }, { "epoch": 2.432952305873078, "grad_norm": 0.5213046073913574, "learning_rate": 0.0005, "loss": 1.533341646194458, "step": 30862 }, { "epoch": 2.4330311391407173, "grad_norm": 0.49806568026542664, "learning_rate": 0.0005, "loss": 1.5161247253417969, "step": 30863 }, { "epoch": 2.4331099724083565, "grad_norm": 0.5867021679878235, "learning_rate": 0.0005, "loss": 1.2630107402801514, "step": 30864 }, { "epoch": 2.433188805675995, "grad_norm": 0.46492403745651245, "learning_rate": 0.0005, "loss": 1.250267744064331, "step": 30865 }, { "epoch": 2.4332676389436343, "grad_norm": 0.5922428965568542, "learning_rate": 0.0005, "loss": 1.3235636949539185, "step": 30866 }, { "epoch": 2.433346472211273, "grad_norm": 0.5232328772544861, "learning_rate": 0.0005, "loss": 1.4736816883087158, "step": 30867 }, { "epoch": 2.433425305478912, "grad_norm": 0.5750695466995239, "learning_rate": 0.0005, "loss": 1.3165231943130493, "step": 30868 }, { "epoch": 2.4335041387465512, "grad_norm": 0.43595027923583984, "learning_rate": 0.0005, "loss": 1.2410271167755127, "step": 30869 }, { "epoch": 2.43358297201419, "grad_norm": 0.44008898735046387, "learning_rate": 0.0005, "loss": 1.3537012338638306, "step": 30870 }, { "epoch": 2.433661805281829, "grad_norm": 0.5422633290290833, "learning_rate": 0.0005, "loss": 1.3506178855895996, "step": 30871 }, { "epoch": 2.4337406385494678, "grad_norm": 0.5679396986961365, "learning_rate": 0.0005, "loss": 1.498504638671875, "step": 30872 }, { "epoch": 2.433819471817107, "grad_norm": 0.43055787682533264, "learning_rate": 0.0005, "loss": 1.0503294467926025, "step": 30873 }, { "epoch": 2.4338983050847456, "grad_norm": 0.46355026960372925, "learning_rate": 0.0005, "loss": 1.2405816316604614, "step": 30874 }, { "epoch": 2.4339771383523847, "grad_norm": 0.590961754322052, "learning_rate": 0.0005, "loss": 1.3658626079559326, "step": 30875 }, { "epoch": 2.4340559716200234, "grad_norm": 0.4864336848258972, "learning_rate": 0.0005, "loss": 1.3826621770858765, "step": 30876 }, { "epoch": 2.4341348048876625, "grad_norm": 0.5095157027244568, "learning_rate": 0.0005, "loss": 1.6166027784347534, "step": 30877 }, { "epoch": 2.4342136381553017, "grad_norm": 0.6027824878692627, "learning_rate": 0.0005, "loss": 1.4557006359100342, "step": 30878 }, { "epoch": 2.4342924714229404, "grad_norm": 0.5258594751358032, "learning_rate": 0.0005, "loss": 1.4256020784378052, "step": 30879 }, { "epoch": 2.4343713046905795, "grad_norm": 0.5684458613395691, "learning_rate": 0.0005, "loss": 1.3647665977478027, "step": 30880 }, { "epoch": 2.434450137958218, "grad_norm": 0.5634217262268066, "learning_rate": 0.0005, "loss": 1.2724599838256836, "step": 30881 }, { "epoch": 2.4345289712258573, "grad_norm": 0.45076414942741394, "learning_rate": 0.0005, "loss": 1.1899523735046387, "step": 30882 }, { "epoch": 2.4346078044934965, "grad_norm": 0.4646139144897461, "learning_rate": 0.0005, "loss": 1.174534559249878, "step": 30883 }, { "epoch": 2.434686637761135, "grad_norm": 0.4881853759288788, "learning_rate": 0.0005, "loss": 1.3396230936050415, "step": 30884 }, { "epoch": 2.4347654710287743, "grad_norm": 0.45735055208206177, "learning_rate": 0.0005, "loss": 1.1613664627075195, "step": 30885 }, { "epoch": 2.434844304296413, "grad_norm": 0.6640549898147583, "learning_rate": 0.0005, "loss": 1.6085970401763916, "step": 30886 }, { "epoch": 2.434923137564052, "grad_norm": 0.4914020001888275, "learning_rate": 0.0005, "loss": 1.310380220413208, "step": 30887 }, { "epoch": 2.435001970831691, "grad_norm": 0.5402894020080566, "learning_rate": 0.0005, "loss": 1.4937796592712402, "step": 30888 }, { "epoch": 2.43508080409933, "grad_norm": 0.5623892545700073, "learning_rate": 0.0005, "loss": 1.2673871517181396, "step": 30889 }, { "epoch": 2.435159637366969, "grad_norm": 0.631091833114624, "learning_rate": 0.0005, "loss": 1.398457407951355, "step": 30890 }, { "epoch": 2.4352384706346077, "grad_norm": 0.47302404046058655, "learning_rate": 0.0005, "loss": 1.200500726699829, "step": 30891 }, { "epoch": 2.435317303902247, "grad_norm": 0.4829309284687042, "learning_rate": 0.0005, "loss": 1.266160011291504, "step": 30892 }, { "epoch": 2.4353961371698856, "grad_norm": 0.3960700035095215, "learning_rate": 0.0005, "loss": 1.08786940574646, "step": 30893 }, { "epoch": 2.4354749704375247, "grad_norm": 0.35326525568962097, "learning_rate": 0.0005, "loss": 1.046473741531372, "step": 30894 }, { "epoch": 2.4355538037051634, "grad_norm": 0.47941428422927856, "learning_rate": 0.0005, "loss": 1.4652806520462036, "step": 30895 }, { "epoch": 2.4356326369728025, "grad_norm": 0.4434000849723816, "learning_rate": 0.0005, "loss": 1.240380883216858, "step": 30896 }, { "epoch": 2.4357114702404417, "grad_norm": 0.5711491703987122, "learning_rate": 0.0005, "loss": 1.3056445121765137, "step": 30897 }, { "epoch": 2.4357903035080803, "grad_norm": 0.7353262901306152, "learning_rate": 0.0005, "loss": 1.2193224430084229, "step": 30898 }, { "epoch": 2.4358691367757195, "grad_norm": 0.45825910568237305, "learning_rate": 0.0005, "loss": 1.3464891910552979, "step": 30899 }, { "epoch": 2.435947970043358, "grad_norm": 0.5518167614936829, "learning_rate": 0.0005, "loss": 1.1360372304916382, "step": 30900 }, { "epoch": 2.4360268033109973, "grad_norm": 0.5580439567565918, "learning_rate": 0.0005, "loss": 1.3030829429626465, "step": 30901 }, { "epoch": 2.436105636578636, "grad_norm": 0.5711685419082642, "learning_rate": 0.0005, "loss": 1.1991969347000122, "step": 30902 }, { "epoch": 2.436184469846275, "grad_norm": 0.5095979571342468, "learning_rate": 0.0005, "loss": 1.3161708116531372, "step": 30903 }, { "epoch": 2.4362633031139143, "grad_norm": 0.5640423893928528, "learning_rate": 0.0005, "loss": 1.1916383504867554, "step": 30904 }, { "epoch": 2.436342136381553, "grad_norm": 0.5120277404785156, "learning_rate": 0.0005, "loss": 1.3145345449447632, "step": 30905 }, { "epoch": 2.436420969649192, "grad_norm": 0.47233888506889343, "learning_rate": 0.0005, "loss": 1.3075834512710571, "step": 30906 }, { "epoch": 2.4364998029168308, "grad_norm": 0.4738573431968689, "learning_rate": 0.0005, "loss": 1.2997885942459106, "step": 30907 }, { "epoch": 2.43657863618447, "grad_norm": 0.5576995611190796, "learning_rate": 0.0005, "loss": 1.5386056900024414, "step": 30908 }, { "epoch": 2.4366574694521086, "grad_norm": 0.5198659300804138, "learning_rate": 0.0005, "loss": 1.312951683998108, "step": 30909 }, { "epoch": 2.4367363027197477, "grad_norm": 0.6807589530944824, "learning_rate": 0.0005, "loss": 1.472782850265503, "step": 30910 }, { "epoch": 2.436815135987387, "grad_norm": 0.5388994216918945, "learning_rate": 0.0005, "loss": 1.4155040979385376, "step": 30911 }, { "epoch": 2.4368939692550255, "grad_norm": 0.4262712001800537, "learning_rate": 0.0005, "loss": 1.1855366230010986, "step": 30912 }, { "epoch": 2.4369728025226647, "grad_norm": 0.4662224054336548, "learning_rate": 0.0005, "loss": 1.1004308462142944, "step": 30913 }, { "epoch": 2.4370516357903034, "grad_norm": 0.9219315052032471, "learning_rate": 0.0005, "loss": 1.3884546756744385, "step": 30914 }, { "epoch": 2.4371304690579425, "grad_norm": 0.48848044872283936, "learning_rate": 0.0005, "loss": 1.5019240379333496, "step": 30915 }, { "epoch": 2.4372093023255816, "grad_norm": 0.5751808881759644, "learning_rate": 0.0005, "loss": 1.460233211517334, "step": 30916 }, { "epoch": 2.4372881355932203, "grad_norm": 0.5867371559143066, "learning_rate": 0.0005, "loss": 1.2538392543792725, "step": 30917 }, { "epoch": 2.4373669688608595, "grad_norm": 0.5598576664924622, "learning_rate": 0.0005, "loss": 1.4302465915679932, "step": 30918 }, { "epoch": 2.437445802128498, "grad_norm": 0.5062417387962341, "learning_rate": 0.0005, "loss": 1.5063992738723755, "step": 30919 }, { "epoch": 2.4375246353961373, "grad_norm": 0.4908728003501892, "learning_rate": 0.0005, "loss": 1.5334644317626953, "step": 30920 }, { "epoch": 2.437603468663776, "grad_norm": 0.5316117405891418, "learning_rate": 0.0005, "loss": 1.3504635095596313, "step": 30921 }, { "epoch": 2.437682301931415, "grad_norm": 0.7309665083885193, "learning_rate": 0.0005, "loss": 1.4293445348739624, "step": 30922 }, { "epoch": 2.437761135199054, "grad_norm": 0.3966025710105896, "learning_rate": 0.0005, "loss": 1.265750527381897, "step": 30923 }, { "epoch": 2.437839968466693, "grad_norm": 0.4368189573287964, "learning_rate": 0.0005, "loss": 1.2816733121871948, "step": 30924 }, { "epoch": 2.437918801734332, "grad_norm": 0.5578015446662903, "learning_rate": 0.0005, "loss": 1.2126070261001587, "step": 30925 }, { "epoch": 2.4379976350019708, "grad_norm": 0.6610507965087891, "learning_rate": 0.0005, "loss": 1.3658227920532227, "step": 30926 }, { "epoch": 2.43807646826961, "grad_norm": 0.4784904718399048, "learning_rate": 0.0005, "loss": 1.351570725440979, "step": 30927 }, { "epoch": 2.4381553015372486, "grad_norm": 0.5320109724998474, "learning_rate": 0.0005, "loss": 1.21047842502594, "step": 30928 }, { "epoch": 2.4382341348048877, "grad_norm": 0.5577908158302307, "learning_rate": 0.0005, "loss": 1.2529269456863403, "step": 30929 }, { "epoch": 2.438312968072527, "grad_norm": 0.42063575983047485, "learning_rate": 0.0005, "loss": 1.3065602779388428, "step": 30930 }, { "epoch": 2.4383918013401655, "grad_norm": 0.4567702114582062, "learning_rate": 0.0005, "loss": 1.0940914154052734, "step": 30931 }, { "epoch": 2.4384706346078047, "grad_norm": 0.5842582583427429, "learning_rate": 0.0005, "loss": 1.3476605415344238, "step": 30932 }, { "epoch": 2.4385494678754434, "grad_norm": 0.46362820267677307, "learning_rate": 0.0005, "loss": 1.4307870864868164, "step": 30933 }, { "epoch": 2.4386283011430825, "grad_norm": 0.4916169345378876, "learning_rate": 0.0005, "loss": 1.3502802848815918, "step": 30934 }, { "epoch": 2.438707134410721, "grad_norm": 0.47287869453430176, "learning_rate": 0.0005, "loss": 1.2201589345932007, "step": 30935 }, { "epoch": 2.4387859676783603, "grad_norm": 0.4499821960926056, "learning_rate": 0.0005, "loss": 1.4376180171966553, "step": 30936 }, { "epoch": 2.438864800945999, "grad_norm": 0.4477383494377136, "learning_rate": 0.0005, "loss": 1.2043315172195435, "step": 30937 }, { "epoch": 2.438943634213638, "grad_norm": 0.4287828207015991, "learning_rate": 0.0005, "loss": 1.2336972951889038, "step": 30938 }, { "epoch": 2.4390224674812773, "grad_norm": 0.5067172646522522, "learning_rate": 0.0005, "loss": 1.2307072877883911, "step": 30939 }, { "epoch": 2.439101300748916, "grad_norm": 0.3994911015033722, "learning_rate": 0.0005, "loss": 1.3947864770889282, "step": 30940 }, { "epoch": 2.439180134016555, "grad_norm": 0.5190410614013672, "learning_rate": 0.0005, "loss": 1.3196156024932861, "step": 30941 }, { "epoch": 2.439258967284194, "grad_norm": 0.45637112855911255, "learning_rate": 0.0005, "loss": 1.3538013696670532, "step": 30942 }, { "epoch": 2.439337800551833, "grad_norm": 0.4353616535663605, "learning_rate": 0.0005, "loss": 1.3123997449874878, "step": 30943 }, { "epoch": 2.439416633819472, "grad_norm": 0.4316159784793854, "learning_rate": 0.0005, "loss": 1.2796094417572021, "step": 30944 }, { "epoch": 2.4394954670871107, "grad_norm": 0.5313881635665894, "learning_rate": 0.0005, "loss": 1.3432347774505615, "step": 30945 }, { "epoch": 2.43957430035475, "grad_norm": 0.5243086814880371, "learning_rate": 0.0005, "loss": 1.3298611640930176, "step": 30946 }, { "epoch": 2.4396531336223886, "grad_norm": 0.638634204864502, "learning_rate": 0.0005, "loss": 1.4258997440338135, "step": 30947 }, { "epoch": 2.4397319668900277, "grad_norm": 0.5009571313858032, "learning_rate": 0.0005, "loss": 1.4726920127868652, "step": 30948 }, { "epoch": 2.4398108001576664, "grad_norm": 0.46456900238990784, "learning_rate": 0.0005, "loss": 1.4493980407714844, "step": 30949 }, { "epoch": 2.4398896334253055, "grad_norm": 0.6882403492927551, "learning_rate": 0.0005, "loss": 1.5490689277648926, "step": 30950 }, { "epoch": 2.439968466692944, "grad_norm": 0.5233314633369446, "learning_rate": 0.0005, "loss": 1.461263656616211, "step": 30951 }, { "epoch": 2.4400472999605833, "grad_norm": 0.6433753371238708, "learning_rate": 0.0005, "loss": 1.2564313411712646, "step": 30952 }, { "epoch": 2.4401261332282225, "grad_norm": 0.47194233536720276, "learning_rate": 0.0005, "loss": 1.51338529586792, "step": 30953 }, { "epoch": 2.440204966495861, "grad_norm": 0.5139421820640564, "learning_rate": 0.0005, "loss": 1.230543613433838, "step": 30954 }, { "epoch": 2.4402837997635003, "grad_norm": 0.5118805766105652, "learning_rate": 0.0005, "loss": 1.2969497442245483, "step": 30955 }, { "epoch": 2.440362633031139, "grad_norm": 0.5510157346725464, "learning_rate": 0.0005, "loss": 1.3703417778015137, "step": 30956 }, { "epoch": 2.440441466298778, "grad_norm": 0.6925054788589478, "learning_rate": 0.0005, "loss": 1.4131901264190674, "step": 30957 }, { "epoch": 2.4405202995664173, "grad_norm": 0.48664385080337524, "learning_rate": 0.0005, "loss": 1.221395492553711, "step": 30958 }, { "epoch": 2.440599132834056, "grad_norm": 0.6040728688240051, "learning_rate": 0.0005, "loss": 1.355257511138916, "step": 30959 }, { "epoch": 2.440677966101695, "grad_norm": 0.49471092224121094, "learning_rate": 0.0005, "loss": 1.4462530612945557, "step": 30960 }, { "epoch": 2.4407567993693338, "grad_norm": 0.605068027973175, "learning_rate": 0.0005, "loss": 1.393834114074707, "step": 30961 }, { "epoch": 2.440835632636973, "grad_norm": 0.6382551193237305, "learning_rate": 0.0005, "loss": 1.450860857963562, "step": 30962 }, { "epoch": 2.4409144659046116, "grad_norm": 0.518101692199707, "learning_rate": 0.0005, "loss": 1.2392864227294922, "step": 30963 }, { "epoch": 2.4409932991722507, "grad_norm": 0.6825693249702454, "learning_rate": 0.0005, "loss": 1.3301496505737305, "step": 30964 }, { "epoch": 2.4410721324398894, "grad_norm": 0.529977560043335, "learning_rate": 0.0005, "loss": 1.1803200244903564, "step": 30965 }, { "epoch": 2.4411509657075285, "grad_norm": 0.525493860244751, "learning_rate": 0.0005, "loss": 1.1773293018341064, "step": 30966 }, { "epoch": 2.4412297989751677, "grad_norm": 0.49660953879356384, "learning_rate": 0.0005, "loss": 1.2492903470993042, "step": 30967 }, { "epoch": 2.4413086322428064, "grad_norm": 0.5002860426902771, "learning_rate": 0.0005, "loss": 1.4060887098312378, "step": 30968 }, { "epoch": 2.4413874655104455, "grad_norm": 0.5153120756149292, "learning_rate": 0.0005, "loss": 1.2190680503845215, "step": 30969 }, { "epoch": 2.441466298778084, "grad_norm": 0.4609396755695343, "learning_rate": 0.0005, "loss": 1.2384142875671387, "step": 30970 }, { "epoch": 2.4415451320457233, "grad_norm": 0.4189763069152832, "learning_rate": 0.0005, "loss": 1.3378846645355225, "step": 30971 }, { "epoch": 2.4416239653133625, "grad_norm": 0.4467446506023407, "learning_rate": 0.0005, "loss": 1.2586452960968018, "step": 30972 }, { "epoch": 2.441702798581001, "grad_norm": 0.5081932544708252, "learning_rate": 0.0005, "loss": 1.244596242904663, "step": 30973 }, { "epoch": 2.4417816318486403, "grad_norm": 0.7041778564453125, "learning_rate": 0.0005, "loss": 1.5315147638320923, "step": 30974 }, { "epoch": 2.441860465116279, "grad_norm": 0.5500949025154114, "learning_rate": 0.0005, "loss": 1.3957643508911133, "step": 30975 }, { "epoch": 2.441939298383918, "grad_norm": 0.5560375452041626, "learning_rate": 0.0005, "loss": 1.3886412382125854, "step": 30976 }, { "epoch": 2.442018131651557, "grad_norm": 0.45991629362106323, "learning_rate": 0.0005, "loss": 1.2930023670196533, "step": 30977 }, { "epoch": 2.442096964919196, "grad_norm": 0.5298013687133789, "learning_rate": 0.0005, "loss": 1.2864378690719604, "step": 30978 }, { "epoch": 2.4421757981868346, "grad_norm": 0.4418972432613373, "learning_rate": 0.0005, "loss": 1.0154893398284912, "step": 30979 }, { "epoch": 2.4422546314544737, "grad_norm": 0.7426415085792542, "learning_rate": 0.0005, "loss": 1.033188819885254, "step": 30980 }, { "epoch": 2.442333464722113, "grad_norm": 0.5982429385185242, "learning_rate": 0.0005, "loss": 0.9964737296104431, "step": 30981 }, { "epoch": 2.4424122979897516, "grad_norm": 0.48180830478668213, "learning_rate": 0.0005, "loss": 1.3884450197219849, "step": 30982 }, { "epoch": 2.4424911312573907, "grad_norm": 0.5336899757385254, "learning_rate": 0.0005, "loss": 1.531320571899414, "step": 30983 }, { "epoch": 2.4425699645250294, "grad_norm": 0.498050719499588, "learning_rate": 0.0005, "loss": 1.2340425252914429, "step": 30984 }, { "epoch": 2.4426487977926685, "grad_norm": 0.6348779201507568, "learning_rate": 0.0005, "loss": 1.5062896013259888, "step": 30985 }, { "epoch": 2.4427276310603077, "grad_norm": 0.4756895899772644, "learning_rate": 0.0005, "loss": 1.20142662525177, "step": 30986 }, { "epoch": 2.4428064643279463, "grad_norm": 0.5441890954971313, "learning_rate": 0.0005, "loss": 1.3466501235961914, "step": 30987 }, { "epoch": 2.4428852975955855, "grad_norm": 0.5128921866416931, "learning_rate": 0.0005, "loss": 1.5149528980255127, "step": 30988 }, { "epoch": 2.442964130863224, "grad_norm": 0.45815128087997437, "learning_rate": 0.0005, "loss": 1.3883349895477295, "step": 30989 }, { "epoch": 2.4430429641308633, "grad_norm": 0.4108372926712036, "learning_rate": 0.0005, "loss": 1.2339503765106201, "step": 30990 }, { "epoch": 2.443121797398502, "grad_norm": 0.4988607168197632, "learning_rate": 0.0005, "loss": 1.361844778060913, "step": 30991 }, { "epoch": 2.443200630666141, "grad_norm": 0.5952207446098328, "learning_rate": 0.0005, "loss": 1.6143274307250977, "step": 30992 }, { "epoch": 2.44327946393378, "grad_norm": 0.5170751810073853, "learning_rate": 0.0005, "loss": 1.4298548698425293, "step": 30993 }, { "epoch": 2.443358297201419, "grad_norm": 1.0224915742874146, "learning_rate": 0.0005, "loss": 1.5971698760986328, "step": 30994 }, { "epoch": 2.443437130469058, "grad_norm": 0.549325168132782, "learning_rate": 0.0005, "loss": 1.417197346687317, "step": 30995 }, { "epoch": 2.4435159637366968, "grad_norm": 0.57677161693573, "learning_rate": 0.0005, "loss": 1.3919552564620972, "step": 30996 }, { "epoch": 2.443594797004336, "grad_norm": 0.45067062973976135, "learning_rate": 0.0005, "loss": 1.5048725605010986, "step": 30997 }, { "epoch": 2.4436736302719746, "grad_norm": 0.48651251196861267, "learning_rate": 0.0005, "loss": 1.3803685903549194, "step": 30998 }, { "epoch": 2.4437524635396137, "grad_norm": 0.6818010210990906, "learning_rate": 0.0005, "loss": 1.5998101234436035, "step": 30999 }, { "epoch": 2.443831296807253, "grad_norm": 0.4715867340564728, "learning_rate": 0.0005, "loss": 1.3902498483657837, "step": 31000 }, { "epoch": 2.4439101300748916, "grad_norm": 0.48861849308013916, "learning_rate": 0.0005, "loss": 1.340574026107788, "step": 31001 }, { "epoch": 2.4439889633425307, "grad_norm": 0.4769779443740845, "learning_rate": 0.0005, "loss": 1.3474318981170654, "step": 31002 }, { "epoch": 2.4440677966101694, "grad_norm": 0.5810655951499939, "learning_rate": 0.0005, "loss": 1.484231948852539, "step": 31003 }, { "epoch": 2.4441466298778085, "grad_norm": 0.4732959568500519, "learning_rate": 0.0005, "loss": 1.1638134717941284, "step": 31004 }, { "epoch": 2.444225463145447, "grad_norm": 0.5279163718223572, "learning_rate": 0.0005, "loss": 1.47964346408844, "step": 31005 }, { "epoch": 2.4443042964130863, "grad_norm": 0.4630034267902374, "learning_rate": 0.0005, "loss": 1.2373406887054443, "step": 31006 }, { "epoch": 2.444383129680725, "grad_norm": 0.3939669728279114, "learning_rate": 0.0005, "loss": 1.0247927904129028, "step": 31007 }, { "epoch": 2.444461962948364, "grad_norm": 0.4526630938053131, "learning_rate": 0.0005, "loss": 1.3113950490951538, "step": 31008 }, { "epoch": 2.4445407962160033, "grad_norm": 0.5667822957038879, "learning_rate": 0.0005, "loss": 1.1798721551895142, "step": 31009 }, { "epoch": 2.444619629483642, "grad_norm": 0.41610389947891235, "learning_rate": 0.0005, "loss": 1.1741857528686523, "step": 31010 }, { "epoch": 2.444698462751281, "grad_norm": 0.5718691349029541, "learning_rate": 0.0005, "loss": 1.4344264268875122, "step": 31011 }, { "epoch": 2.44477729601892, "grad_norm": 0.7088457942008972, "learning_rate": 0.0005, "loss": 1.679709553718567, "step": 31012 }, { "epoch": 2.444856129286559, "grad_norm": 0.6794201135635376, "learning_rate": 0.0005, "loss": 1.3875244855880737, "step": 31013 }, { "epoch": 2.444934962554198, "grad_norm": 0.4812764525413513, "learning_rate": 0.0005, "loss": 1.1996784210205078, "step": 31014 }, { "epoch": 2.4450137958218368, "grad_norm": 0.7488291263580322, "learning_rate": 0.0005, "loss": 1.2907065153121948, "step": 31015 }, { "epoch": 2.445092629089476, "grad_norm": 0.44438979029655457, "learning_rate": 0.0005, "loss": 1.3097388744354248, "step": 31016 }, { "epoch": 2.4451714623571146, "grad_norm": 0.5324508547782898, "learning_rate": 0.0005, "loss": 1.4209778308868408, "step": 31017 }, { "epoch": 2.4452502956247537, "grad_norm": 0.4425917863845825, "learning_rate": 0.0005, "loss": 1.3166090250015259, "step": 31018 }, { "epoch": 2.4453291288923924, "grad_norm": 0.5010898113250732, "learning_rate": 0.0005, "loss": 1.4425833225250244, "step": 31019 }, { "epoch": 2.4454079621600315, "grad_norm": 0.496791273355484, "learning_rate": 0.0005, "loss": 1.4721728563308716, "step": 31020 }, { "epoch": 2.4454867954276707, "grad_norm": 0.40877047181129456, "learning_rate": 0.0005, "loss": 1.3320014476776123, "step": 31021 }, { "epoch": 2.4455656286953094, "grad_norm": 0.530039370059967, "learning_rate": 0.0005, "loss": 1.3457295894622803, "step": 31022 }, { "epoch": 2.4456444619629485, "grad_norm": 0.6749452352523804, "learning_rate": 0.0005, "loss": 1.4147279262542725, "step": 31023 }, { "epoch": 2.445723295230587, "grad_norm": 0.4761923551559448, "learning_rate": 0.0005, "loss": 1.4373552799224854, "step": 31024 }, { "epoch": 2.4458021284982263, "grad_norm": 0.6224496960639954, "learning_rate": 0.0005, "loss": 1.539440631866455, "step": 31025 }, { "epoch": 2.445880961765865, "grad_norm": 0.49092620611190796, "learning_rate": 0.0005, "loss": 1.308374285697937, "step": 31026 }, { "epoch": 2.445959795033504, "grad_norm": 0.4344802796840668, "learning_rate": 0.0005, "loss": 1.2167320251464844, "step": 31027 }, { "epoch": 2.4460386283011433, "grad_norm": 0.551533043384552, "learning_rate": 0.0005, "loss": 1.4924838542938232, "step": 31028 }, { "epoch": 2.446117461568782, "grad_norm": 0.4842567443847656, "learning_rate": 0.0005, "loss": 1.37593674659729, "step": 31029 }, { "epoch": 2.446196294836421, "grad_norm": 0.47397229075431824, "learning_rate": 0.0005, "loss": 1.2870681285858154, "step": 31030 }, { "epoch": 2.44627512810406, "grad_norm": 0.46836233139038086, "learning_rate": 0.0005, "loss": 1.4301631450653076, "step": 31031 }, { "epoch": 2.446353961371699, "grad_norm": 0.6050163507461548, "learning_rate": 0.0005, "loss": 1.3128982782363892, "step": 31032 }, { "epoch": 2.446432794639338, "grad_norm": 0.5702216029167175, "learning_rate": 0.0005, "loss": 1.616925597190857, "step": 31033 }, { "epoch": 2.4465116279069767, "grad_norm": 0.6212747097015381, "learning_rate": 0.0005, "loss": 1.5647271871566772, "step": 31034 }, { "epoch": 2.446590461174616, "grad_norm": 0.4850890636444092, "learning_rate": 0.0005, "loss": 1.2897578477859497, "step": 31035 }, { "epoch": 2.4466692944422546, "grad_norm": 0.5010289549827576, "learning_rate": 0.0005, "loss": 1.266512393951416, "step": 31036 }, { "epoch": 2.4467481277098937, "grad_norm": 0.5545573830604553, "learning_rate": 0.0005, "loss": 1.438737154006958, "step": 31037 }, { "epoch": 2.4468269609775324, "grad_norm": 0.5534400939941406, "learning_rate": 0.0005, "loss": 1.354934573173523, "step": 31038 }, { "epoch": 2.4469057942451715, "grad_norm": 0.5159761905670166, "learning_rate": 0.0005, "loss": 1.332374095916748, "step": 31039 }, { "epoch": 2.44698462751281, "grad_norm": 0.5297015905380249, "learning_rate": 0.0005, "loss": 1.3789026737213135, "step": 31040 }, { "epoch": 2.4470634607804493, "grad_norm": 0.5167973637580872, "learning_rate": 0.0005, "loss": 1.3720762729644775, "step": 31041 }, { "epoch": 2.4471422940480885, "grad_norm": 0.48303043842315674, "learning_rate": 0.0005, "loss": 1.3955249786376953, "step": 31042 }, { "epoch": 2.447221127315727, "grad_norm": 0.49351391196250916, "learning_rate": 0.0005, "loss": 1.5096778869628906, "step": 31043 }, { "epoch": 2.4472999605833663, "grad_norm": 0.4949978291988373, "learning_rate": 0.0005, "loss": 1.3093557357788086, "step": 31044 }, { "epoch": 2.447378793851005, "grad_norm": 0.8607109785079956, "learning_rate": 0.0005, "loss": 1.4124910831451416, "step": 31045 }, { "epoch": 2.447457627118644, "grad_norm": 0.4106244146823883, "learning_rate": 0.0005, "loss": 1.3837354183197021, "step": 31046 }, { "epoch": 2.4475364603862833, "grad_norm": 0.48346132040023804, "learning_rate": 0.0005, "loss": 1.4043570756912231, "step": 31047 }, { "epoch": 2.447615293653922, "grad_norm": 0.5654935240745544, "learning_rate": 0.0005, "loss": 1.107255220413208, "step": 31048 }, { "epoch": 2.447694126921561, "grad_norm": 0.5348862409591675, "learning_rate": 0.0005, "loss": 1.4418412446975708, "step": 31049 }, { "epoch": 2.4477729601891998, "grad_norm": 0.517836332321167, "learning_rate": 0.0005, "loss": 1.3952064514160156, "step": 31050 }, { "epoch": 2.447851793456839, "grad_norm": 0.4992433190345764, "learning_rate": 0.0005, "loss": 1.2763481140136719, "step": 31051 }, { "epoch": 2.4479306267244776, "grad_norm": 0.47004181146621704, "learning_rate": 0.0005, "loss": 1.2180001735687256, "step": 31052 }, { "epoch": 2.4480094599921167, "grad_norm": 0.4515753388404846, "learning_rate": 0.0005, "loss": 1.1610124111175537, "step": 31053 }, { "epoch": 2.4480882932597554, "grad_norm": 0.5239903330802917, "learning_rate": 0.0005, "loss": 1.3574596643447876, "step": 31054 }, { "epoch": 2.4481671265273945, "grad_norm": 0.5905420184135437, "learning_rate": 0.0005, "loss": 1.3404324054718018, "step": 31055 }, { "epoch": 2.4482459597950337, "grad_norm": 0.5076128244400024, "learning_rate": 0.0005, "loss": 1.2703754901885986, "step": 31056 }, { "epoch": 2.4483247930626724, "grad_norm": 0.47181427478790283, "learning_rate": 0.0005, "loss": 1.3072268962860107, "step": 31057 }, { "epoch": 2.4484036263303115, "grad_norm": 0.4812767207622528, "learning_rate": 0.0005, "loss": 1.2294938564300537, "step": 31058 }, { "epoch": 2.44848245959795, "grad_norm": 0.5557214617729187, "learning_rate": 0.0005, "loss": 1.4098960161209106, "step": 31059 }, { "epoch": 2.4485612928655893, "grad_norm": 0.4856604039669037, "learning_rate": 0.0005, "loss": 1.2553387880325317, "step": 31060 }, { "epoch": 2.4486401261332285, "grad_norm": 0.40796250104904175, "learning_rate": 0.0005, "loss": 1.1366157531738281, "step": 31061 }, { "epoch": 2.448718959400867, "grad_norm": 0.5335592031478882, "learning_rate": 0.0005, "loss": 1.1950581073760986, "step": 31062 }, { "epoch": 2.4487977926685063, "grad_norm": 0.45850297808647156, "learning_rate": 0.0005, "loss": 1.3604644536972046, "step": 31063 }, { "epoch": 2.448876625936145, "grad_norm": 0.5593708753585815, "learning_rate": 0.0005, "loss": 1.3481577634811401, "step": 31064 }, { "epoch": 2.448955459203784, "grad_norm": 0.48668697476387024, "learning_rate": 0.0005, "loss": 1.313283920288086, "step": 31065 }, { "epoch": 2.449034292471423, "grad_norm": 0.5001747608184814, "learning_rate": 0.0005, "loss": 1.3104660511016846, "step": 31066 }, { "epoch": 2.449113125739062, "grad_norm": 0.6354945302009583, "learning_rate": 0.0005, "loss": 1.2001065015792847, "step": 31067 }, { "epoch": 2.4491919590067006, "grad_norm": 0.4089248478412628, "learning_rate": 0.0005, "loss": 0.9915483593940735, "step": 31068 }, { "epoch": 2.4492707922743397, "grad_norm": 0.5557985901832581, "learning_rate": 0.0005, "loss": 1.3773161172866821, "step": 31069 }, { "epoch": 2.449349625541979, "grad_norm": 1.0247862339019775, "learning_rate": 0.0005, "loss": 1.2809127569198608, "step": 31070 }, { "epoch": 2.4494284588096176, "grad_norm": 0.5949125289916992, "learning_rate": 0.0005, "loss": 1.2651751041412354, "step": 31071 }, { "epoch": 2.4495072920772567, "grad_norm": 0.6022316813468933, "learning_rate": 0.0005, "loss": 1.3886804580688477, "step": 31072 }, { "epoch": 2.4495861253448954, "grad_norm": 0.5755407214164734, "learning_rate": 0.0005, "loss": 1.479844093322754, "step": 31073 }, { "epoch": 2.4496649586125345, "grad_norm": 0.5438489317893982, "learning_rate": 0.0005, "loss": 1.5280795097351074, "step": 31074 }, { "epoch": 2.4497437918801737, "grad_norm": 0.405261367559433, "learning_rate": 0.0005, "loss": 1.1798632144927979, "step": 31075 }, { "epoch": 2.4498226251478123, "grad_norm": 0.619079053401947, "learning_rate": 0.0005, "loss": 1.3786840438842773, "step": 31076 }, { "epoch": 2.4499014584154515, "grad_norm": 0.5965307950973511, "learning_rate": 0.0005, "loss": 1.342075228691101, "step": 31077 }, { "epoch": 2.44998029168309, "grad_norm": 0.4215759336948395, "learning_rate": 0.0005, "loss": 1.2663874626159668, "step": 31078 }, { "epoch": 2.4500591249507293, "grad_norm": 0.5602754950523376, "learning_rate": 0.0005, "loss": 1.5195446014404297, "step": 31079 }, { "epoch": 2.450137958218368, "grad_norm": 0.47662195563316345, "learning_rate": 0.0005, "loss": 1.2998420000076294, "step": 31080 }, { "epoch": 2.450216791486007, "grad_norm": 0.5100082159042358, "learning_rate": 0.0005, "loss": 1.1839545965194702, "step": 31081 }, { "epoch": 2.450295624753646, "grad_norm": 0.4622942805290222, "learning_rate": 0.0005, "loss": 1.375532865524292, "step": 31082 }, { "epoch": 2.450374458021285, "grad_norm": 0.5336621999740601, "learning_rate": 0.0005, "loss": 1.2265644073486328, "step": 31083 }, { "epoch": 2.450453291288924, "grad_norm": 0.7805410623550415, "learning_rate": 0.0005, "loss": 1.3572899103164673, "step": 31084 }, { "epoch": 2.4505321245565628, "grad_norm": 0.4951448142528534, "learning_rate": 0.0005, "loss": 1.3344285488128662, "step": 31085 }, { "epoch": 2.450610957824202, "grad_norm": 0.5618443489074707, "learning_rate": 0.0005, "loss": 1.4492088556289673, "step": 31086 }, { "epoch": 2.4506897910918406, "grad_norm": 0.526404857635498, "learning_rate": 0.0005, "loss": 1.3690035343170166, "step": 31087 }, { "epoch": 2.4507686243594797, "grad_norm": 0.7427000999450684, "learning_rate": 0.0005, "loss": 1.6345454454421997, "step": 31088 }, { "epoch": 2.450847457627119, "grad_norm": 0.4622483253479004, "learning_rate": 0.0005, "loss": 1.318878412246704, "step": 31089 }, { "epoch": 2.4509262908947576, "grad_norm": 0.6127062439918518, "learning_rate": 0.0005, "loss": 1.2131106853485107, "step": 31090 }, { "epoch": 2.4510051241623967, "grad_norm": 0.5118359923362732, "learning_rate": 0.0005, "loss": 1.4573712348937988, "step": 31091 }, { "epoch": 2.4510839574300354, "grad_norm": 0.6467957496643066, "learning_rate": 0.0005, "loss": 1.4703285694122314, "step": 31092 }, { "epoch": 2.4511627906976745, "grad_norm": 0.5321919918060303, "learning_rate": 0.0005, "loss": 1.2834917306900024, "step": 31093 }, { "epoch": 2.451241623965313, "grad_norm": 0.6156998872756958, "learning_rate": 0.0005, "loss": 1.4109539985656738, "step": 31094 }, { "epoch": 2.4513204572329523, "grad_norm": 0.5376033186912537, "learning_rate": 0.0005, "loss": 1.3532781600952148, "step": 31095 }, { "epoch": 2.451399290500591, "grad_norm": 0.4901160001754761, "learning_rate": 0.0005, "loss": 1.507258415222168, "step": 31096 }, { "epoch": 2.45147812376823, "grad_norm": 0.3628107011318207, "learning_rate": 0.0005, "loss": 1.2457678318023682, "step": 31097 }, { "epoch": 2.4515569570358693, "grad_norm": 0.45267918705940247, "learning_rate": 0.0005, "loss": 1.2584420442581177, "step": 31098 }, { "epoch": 2.451635790303508, "grad_norm": 0.5138174891471863, "learning_rate": 0.0005, "loss": 1.35421621799469, "step": 31099 }, { "epoch": 2.451714623571147, "grad_norm": 0.521251916885376, "learning_rate": 0.0005, "loss": 1.3469527959823608, "step": 31100 }, { "epoch": 2.451793456838786, "grad_norm": 0.424134761095047, "learning_rate": 0.0005, "loss": 1.3073103427886963, "step": 31101 }, { "epoch": 2.451872290106425, "grad_norm": 0.4857032597064972, "learning_rate": 0.0005, "loss": 1.1442196369171143, "step": 31102 }, { "epoch": 2.451951123374064, "grad_norm": 0.5300744771957397, "learning_rate": 0.0005, "loss": 1.2829203605651855, "step": 31103 }, { "epoch": 2.4520299566417028, "grad_norm": 0.5613262057304382, "learning_rate": 0.0005, "loss": 1.4857721328735352, "step": 31104 }, { "epoch": 2.452108789909342, "grad_norm": 0.39954671263694763, "learning_rate": 0.0005, "loss": 1.119994878768921, "step": 31105 }, { "epoch": 2.4521876231769806, "grad_norm": 0.4948987066745758, "learning_rate": 0.0005, "loss": 1.3615267276763916, "step": 31106 }, { "epoch": 2.4522664564446197, "grad_norm": 0.49375954270362854, "learning_rate": 0.0005, "loss": 1.3951294422149658, "step": 31107 }, { "epoch": 2.4523452897122584, "grad_norm": 0.5314804315567017, "learning_rate": 0.0005, "loss": 1.2739388942718506, "step": 31108 }, { "epoch": 2.4524241229798975, "grad_norm": 0.5992746949195862, "learning_rate": 0.0005, "loss": 1.7474055290222168, "step": 31109 }, { "epoch": 2.4525029562475362, "grad_norm": 0.451109379529953, "learning_rate": 0.0005, "loss": 1.279026985168457, "step": 31110 }, { "epoch": 2.4525817895151754, "grad_norm": 0.43777889013290405, "learning_rate": 0.0005, "loss": 1.2333433628082275, "step": 31111 }, { "epoch": 2.4526606227828145, "grad_norm": 0.46571972966194153, "learning_rate": 0.0005, "loss": 1.274482011795044, "step": 31112 }, { "epoch": 2.452739456050453, "grad_norm": 0.5190407037734985, "learning_rate": 0.0005, "loss": 1.4469959735870361, "step": 31113 }, { "epoch": 2.4528182893180923, "grad_norm": 1.0579761266708374, "learning_rate": 0.0005, "loss": 1.6508125066757202, "step": 31114 }, { "epoch": 2.452897122585731, "grad_norm": 0.45423606038093567, "learning_rate": 0.0005, "loss": 1.3592872619628906, "step": 31115 }, { "epoch": 2.45297595585337, "grad_norm": 0.6402040123939514, "learning_rate": 0.0005, "loss": 1.2866215705871582, "step": 31116 }, { "epoch": 2.4530547891210093, "grad_norm": 0.4849900007247925, "learning_rate": 0.0005, "loss": 1.288297414779663, "step": 31117 }, { "epoch": 2.453133622388648, "grad_norm": 0.45233386754989624, "learning_rate": 0.0005, "loss": 1.3796494007110596, "step": 31118 }, { "epoch": 2.453212455656287, "grad_norm": 0.4405977725982666, "learning_rate": 0.0005, "loss": 1.0879260301589966, "step": 31119 }, { "epoch": 2.453291288923926, "grad_norm": 0.47355300188064575, "learning_rate": 0.0005, "loss": 1.415921688079834, "step": 31120 }, { "epoch": 2.453370122191565, "grad_norm": 0.8477486968040466, "learning_rate": 0.0005, "loss": 1.2477322816848755, "step": 31121 }, { "epoch": 2.4534489554592036, "grad_norm": 0.8038686513900757, "learning_rate": 0.0005, "loss": 1.372488021850586, "step": 31122 }, { "epoch": 2.4535277887268427, "grad_norm": 0.4671168923377991, "learning_rate": 0.0005, "loss": 1.350212574005127, "step": 31123 }, { "epoch": 2.4536066219944814, "grad_norm": 0.6248153448104858, "learning_rate": 0.0005, "loss": 1.3986562490463257, "step": 31124 }, { "epoch": 2.4536854552621206, "grad_norm": 0.5854637026786804, "learning_rate": 0.0005, "loss": 1.3054721355438232, "step": 31125 }, { "epoch": 2.4537642885297597, "grad_norm": 0.8926743268966675, "learning_rate": 0.0005, "loss": 1.334594964981079, "step": 31126 }, { "epoch": 2.4538431217973984, "grad_norm": 0.4804736375808716, "learning_rate": 0.0005, "loss": 1.3057928085327148, "step": 31127 }, { "epoch": 2.4539219550650375, "grad_norm": 0.6077125072479248, "learning_rate": 0.0005, "loss": 1.2163610458374023, "step": 31128 }, { "epoch": 2.454000788332676, "grad_norm": 0.5212759375572205, "learning_rate": 0.0005, "loss": 1.2414112091064453, "step": 31129 }, { "epoch": 2.4540796216003153, "grad_norm": 0.43877172470092773, "learning_rate": 0.0005, "loss": 1.3280696868896484, "step": 31130 }, { "epoch": 2.4541584548679545, "grad_norm": 0.4901784062385559, "learning_rate": 0.0005, "loss": 1.1809492111206055, "step": 31131 }, { "epoch": 2.454237288135593, "grad_norm": 0.447502464056015, "learning_rate": 0.0005, "loss": 1.2124477624893188, "step": 31132 }, { "epoch": 2.4543161214032323, "grad_norm": 0.6401559114456177, "learning_rate": 0.0005, "loss": 1.5208728313446045, "step": 31133 }, { "epoch": 2.454394954670871, "grad_norm": 0.5704196095466614, "learning_rate": 0.0005, "loss": 1.3838657140731812, "step": 31134 }, { "epoch": 2.45447378793851, "grad_norm": 0.5930216312408447, "learning_rate": 0.0005, "loss": 1.7053242921829224, "step": 31135 }, { "epoch": 2.454552621206149, "grad_norm": 0.5042359828948975, "learning_rate": 0.0005, "loss": 1.1931347846984863, "step": 31136 }, { "epoch": 2.454631454473788, "grad_norm": 0.5110070705413818, "learning_rate": 0.0005, "loss": 1.2855767011642456, "step": 31137 }, { "epoch": 2.4547102877414266, "grad_norm": 0.4704417884349823, "learning_rate": 0.0005, "loss": 1.464359998703003, "step": 31138 }, { "epoch": 2.4547891210090658, "grad_norm": 0.47348129749298096, "learning_rate": 0.0005, "loss": 1.259556531906128, "step": 31139 }, { "epoch": 2.454867954276705, "grad_norm": 0.6979036927223206, "learning_rate": 0.0005, "loss": 1.4052361249923706, "step": 31140 }, { "epoch": 2.4549467875443436, "grad_norm": 0.5545876622200012, "learning_rate": 0.0005, "loss": 1.4723310470581055, "step": 31141 }, { "epoch": 2.4550256208119827, "grad_norm": 0.4878053665161133, "learning_rate": 0.0005, "loss": 1.1643788814544678, "step": 31142 }, { "epoch": 2.4551044540796214, "grad_norm": 0.5621257424354553, "learning_rate": 0.0005, "loss": 1.2620166540145874, "step": 31143 }, { "epoch": 2.4551832873472605, "grad_norm": 0.5242526531219482, "learning_rate": 0.0005, "loss": 1.421746015548706, "step": 31144 }, { "epoch": 2.4552621206148997, "grad_norm": 0.5665863156318665, "learning_rate": 0.0005, "loss": 1.3568754196166992, "step": 31145 }, { "epoch": 2.4553409538825384, "grad_norm": 0.45588380098342896, "learning_rate": 0.0005, "loss": 1.2324862480163574, "step": 31146 }, { "epoch": 2.4554197871501775, "grad_norm": 0.5462378263473511, "learning_rate": 0.0005, "loss": 1.1723850965499878, "step": 31147 }, { "epoch": 2.455498620417816, "grad_norm": 0.5152032971382141, "learning_rate": 0.0005, "loss": 1.3838562965393066, "step": 31148 }, { "epoch": 2.4555774536854553, "grad_norm": 0.4589954614639282, "learning_rate": 0.0005, "loss": 1.193835735321045, "step": 31149 }, { "epoch": 2.455656286953094, "grad_norm": 0.4353176951408386, "learning_rate": 0.0005, "loss": 1.416767954826355, "step": 31150 }, { "epoch": 2.455735120220733, "grad_norm": 0.549479067325592, "learning_rate": 0.0005, "loss": 1.3147681951522827, "step": 31151 }, { "epoch": 2.4558139534883723, "grad_norm": 0.4998272955417633, "learning_rate": 0.0005, "loss": 1.2420690059661865, "step": 31152 }, { "epoch": 2.455892786756011, "grad_norm": 0.47392159700393677, "learning_rate": 0.0005, "loss": 1.1775243282318115, "step": 31153 }, { "epoch": 2.45597162002365, "grad_norm": 0.5408713221549988, "learning_rate": 0.0005, "loss": 1.3215057849884033, "step": 31154 }, { "epoch": 2.456050453291289, "grad_norm": 0.562247633934021, "learning_rate": 0.0005, "loss": 1.3594701290130615, "step": 31155 }, { "epoch": 2.456129286558928, "grad_norm": 0.46790069341659546, "learning_rate": 0.0005, "loss": 1.2679245471954346, "step": 31156 }, { "epoch": 2.4562081198265666, "grad_norm": 0.4759604036808014, "learning_rate": 0.0005, "loss": 1.1692575216293335, "step": 31157 }, { "epoch": 2.4562869530942057, "grad_norm": 0.46087804436683655, "learning_rate": 0.0005, "loss": 1.22994065284729, "step": 31158 }, { "epoch": 2.456365786361845, "grad_norm": 0.47502318024635315, "learning_rate": 0.0005, "loss": 1.1697988510131836, "step": 31159 }, { "epoch": 2.4564446196294836, "grad_norm": 0.5398086905479431, "learning_rate": 0.0005, "loss": 1.488337755203247, "step": 31160 }, { "epoch": 2.4565234528971227, "grad_norm": 0.5614643692970276, "learning_rate": 0.0005, "loss": 1.4671461582183838, "step": 31161 }, { "epoch": 2.4566022861647614, "grad_norm": 0.5284438133239746, "learning_rate": 0.0005, "loss": 1.4778084754943848, "step": 31162 }, { "epoch": 2.4566811194324005, "grad_norm": 0.5374048948287964, "learning_rate": 0.0005, "loss": 1.4690881967544556, "step": 31163 }, { "epoch": 2.4567599527000397, "grad_norm": 0.5711380243301392, "learning_rate": 0.0005, "loss": 1.2558602094650269, "step": 31164 }, { "epoch": 2.4568387859676784, "grad_norm": 0.4609944522380829, "learning_rate": 0.0005, "loss": 1.3247171640396118, "step": 31165 }, { "epoch": 2.4569176192353175, "grad_norm": 0.4884938895702362, "learning_rate": 0.0005, "loss": 1.3922052383422852, "step": 31166 }, { "epoch": 2.456996452502956, "grad_norm": 0.4408731162548065, "learning_rate": 0.0005, "loss": 1.2669296264648438, "step": 31167 }, { "epoch": 2.4570752857705953, "grad_norm": 0.5919069647789001, "learning_rate": 0.0005, "loss": 1.4558309316635132, "step": 31168 }, { "epoch": 2.457154119038234, "grad_norm": 0.5468357801437378, "learning_rate": 0.0005, "loss": 1.3233426809310913, "step": 31169 }, { "epoch": 2.457232952305873, "grad_norm": 0.5095190405845642, "learning_rate": 0.0005, "loss": 1.343495488166809, "step": 31170 }, { "epoch": 2.457311785573512, "grad_norm": 0.5058028101921082, "learning_rate": 0.0005, "loss": 1.3634188175201416, "step": 31171 }, { "epoch": 2.457390618841151, "grad_norm": 0.5961276292800903, "learning_rate": 0.0005, "loss": 1.3464045524597168, "step": 31172 }, { "epoch": 2.45746945210879, "grad_norm": 0.5505213141441345, "learning_rate": 0.0005, "loss": 1.389420509338379, "step": 31173 }, { "epoch": 2.4575482853764288, "grad_norm": 0.4735351502895355, "learning_rate": 0.0005, "loss": 1.352800965309143, "step": 31174 }, { "epoch": 2.457627118644068, "grad_norm": 0.4362315833568573, "learning_rate": 0.0005, "loss": 1.2377917766571045, "step": 31175 }, { "epoch": 2.4577059519117066, "grad_norm": 0.53475421667099, "learning_rate": 0.0005, "loss": 1.331169605255127, "step": 31176 }, { "epoch": 2.4577847851793457, "grad_norm": 0.5086625814437866, "learning_rate": 0.0005, "loss": 1.2962400913238525, "step": 31177 }, { "epoch": 2.457863618446985, "grad_norm": 0.4749123156070709, "learning_rate": 0.0005, "loss": 1.435610055923462, "step": 31178 }, { "epoch": 2.4579424517146236, "grad_norm": 0.4940777122974396, "learning_rate": 0.0005, "loss": 1.4392850399017334, "step": 31179 }, { "epoch": 2.4580212849822627, "grad_norm": 0.487348735332489, "learning_rate": 0.0005, "loss": 1.4143431186676025, "step": 31180 }, { "epoch": 2.4581001182499014, "grad_norm": 0.5162346959114075, "learning_rate": 0.0005, "loss": 1.3736536502838135, "step": 31181 }, { "epoch": 2.4581789515175405, "grad_norm": 0.6258837580680847, "learning_rate": 0.0005, "loss": 1.6086355447769165, "step": 31182 }, { "epoch": 2.458257784785179, "grad_norm": 0.46015092730522156, "learning_rate": 0.0005, "loss": 1.3347903490066528, "step": 31183 }, { "epoch": 2.4583366180528183, "grad_norm": 0.5208771824836731, "learning_rate": 0.0005, "loss": 1.4290580749511719, "step": 31184 }, { "epoch": 2.458415451320457, "grad_norm": 0.40481141209602356, "learning_rate": 0.0005, "loss": 1.2535138130187988, "step": 31185 }, { "epoch": 2.458494284588096, "grad_norm": 0.4753303527832031, "learning_rate": 0.0005, "loss": 1.3471702337265015, "step": 31186 }, { "epoch": 2.4585731178557353, "grad_norm": 0.4111517071723938, "learning_rate": 0.0005, "loss": 1.297794222831726, "step": 31187 }, { "epoch": 2.458651951123374, "grad_norm": 0.46836182475090027, "learning_rate": 0.0005, "loss": 1.429689645767212, "step": 31188 }, { "epoch": 2.458730784391013, "grad_norm": 0.5897638201713562, "learning_rate": 0.0005, "loss": 1.7531394958496094, "step": 31189 }, { "epoch": 2.458809617658652, "grad_norm": 0.48518893122673035, "learning_rate": 0.0005, "loss": 1.1965177059173584, "step": 31190 }, { "epoch": 2.458888450926291, "grad_norm": 0.5661410093307495, "learning_rate": 0.0005, "loss": 1.4782969951629639, "step": 31191 }, { "epoch": 2.45896728419393, "grad_norm": 0.6455100774765015, "learning_rate": 0.0005, "loss": 1.1686375141143799, "step": 31192 }, { "epoch": 2.4590461174615688, "grad_norm": 0.47273215651512146, "learning_rate": 0.0005, "loss": 1.2675023078918457, "step": 31193 }, { "epoch": 2.459124950729208, "grad_norm": 0.5281051993370056, "learning_rate": 0.0005, "loss": 1.5472986698150635, "step": 31194 }, { "epoch": 2.4592037839968466, "grad_norm": 0.5893737077713013, "learning_rate": 0.0005, "loss": 1.4646332263946533, "step": 31195 }, { "epoch": 2.4592826172644857, "grad_norm": 0.6701403260231018, "learning_rate": 0.0005, "loss": 1.242173194885254, "step": 31196 }, { "epoch": 2.4593614505321244, "grad_norm": 0.6065593361854553, "learning_rate": 0.0005, "loss": 1.3290563821792603, "step": 31197 }, { "epoch": 2.4594402837997635, "grad_norm": 0.41503801941871643, "learning_rate": 0.0005, "loss": 1.237750768661499, "step": 31198 }, { "epoch": 2.4595191170674022, "grad_norm": 0.46170246601104736, "learning_rate": 0.0005, "loss": 1.3995692729949951, "step": 31199 }, { "epoch": 2.4595979503350414, "grad_norm": 0.4469467103481293, "learning_rate": 0.0005, "loss": 1.2079994678497314, "step": 31200 }, { "epoch": 2.4596767836026805, "grad_norm": 0.5291711091995239, "learning_rate": 0.0005, "loss": 1.3002285957336426, "step": 31201 }, { "epoch": 2.459755616870319, "grad_norm": 0.9347235560417175, "learning_rate": 0.0005, "loss": 1.4102668762207031, "step": 31202 }, { "epoch": 2.4598344501379583, "grad_norm": 0.49314212799072266, "learning_rate": 0.0005, "loss": 1.3610527515411377, "step": 31203 }, { "epoch": 2.459913283405597, "grad_norm": 0.5483952760696411, "learning_rate": 0.0005, "loss": 1.423863172531128, "step": 31204 }, { "epoch": 2.459992116673236, "grad_norm": 0.4567396342754364, "learning_rate": 0.0005, "loss": 1.2255134582519531, "step": 31205 }, { "epoch": 2.4600709499408753, "grad_norm": 0.48338019847869873, "learning_rate": 0.0005, "loss": 1.2911081314086914, "step": 31206 }, { "epoch": 2.460149783208514, "grad_norm": 0.45323508977890015, "learning_rate": 0.0005, "loss": 1.447061538696289, "step": 31207 }, { "epoch": 2.460228616476153, "grad_norm": 0.47824665904045105, "learning_rate": 0.0005, "loss": 1.2574394941329956, "step": 31208 }, { "epoch": 2.460307449743792, "grad_norm": 0.6986267566680908, "learning_rate": 0.0005, "loss": 1.7052860260009766, "step": 31209 }, { "epoch": 2.460386283011431, "grad_norm": 0.5999929904937744, "learning_rate": 0.0005, "loss": 1.3511995077133179, "step": 31210 }, { "epoch": 2.4604651162790696, "grad_norm": 0.6192191243171692, "learning_rate": 0.0005, "loss": 1.4876972436904907, "step": 31211 }, { "epoch": 2.4605439495467087, "grad_norm": 0.5763411521911621, "learning_rate": 0.0005, "loss": 1.3477845191955566, "step": 31212 }, { "epoch": 2.4606227828143474, "grad_norm": 0.4234728515148163, "learning_rate": 0.0005, "loss": 1.23726487159729, "step": 31213 }, { "epoch": 2.4607016160819866, "grad_norm": 0.361299991607666, "learning_rate": 0.0005, "loss": 1.0495957136154175, "step": 31214 }, { "epoch": 2.4607804493496257, "grad_norm": 0.6620896458625793, "learning_rate": 0.0005, "loss": 1.4745025634765625, "step": 31215 }, { "epoch": 2.4608592826172644, "grad_norm": 0.3984895944595337, "learning_rate": 0.0005, "loss": 1.1972451210021973, "step": 31216 }, { "epoch": 2.4609381158849035, "grad_norm": 0.38569435477256775, "learning_rate": 0.0005, "loss": 1.360473871231079, "step": 31217 }, { "epoch": 2.461016949152542, "grad_norm": 0.42910727858543396, "learning_rate": 0.0005, "loss": 1.3616712093353271, "step": 31218 }, { "epoch": 2.4610957824201813, "grad_norm": 0.49782052636146545, "learning_rate": 0.0005, "loss": 1.289839267730713, "step": 31219 }, { "epoch": 2.4611746156878205, "grad_norm": 0.4225383996963501, "learning_rate": 0.0005, "loss": 1.2088088989257812, "step": 31220 }, { "epoch": 2.461253448955459, "grad_norm": 0.4768049418926239, "learning_rate": 0.0005, "loss": 1.3681278228759766, "step": 31221 }, { "epoch": 2.4613322822230983, "grad_norm": 0.47060930728912354, "learning_rate": 0.0005, "loss": 1.2781178951263428, "step": 31222 }, { "epoch": 2.461411115490737, "grad_norm": 0.42954298853874207, "learning_rate": 0.0005, "loss": 1.139355182647705, "step": 31223 }, { "epoch": 2.461489948758376, "grad_norm": 0.4970552623271942, "learning_rate": 0.0005, "loss": 1.4027730226516724, "step": 31224 }, { "epoch": 2.461568782026015, "grad_norm": 0.44956499338150024, "learning_rate": 0.0005, "loss": 1.37091064453125, "step": 31225 }, { "epoch": 2.461647615293654, "grad_norm": 0.4840954840183258, "learning_rate": 0.0005, "loss": 1.2840865850448608, "step": 31226 }, { "epoch": 2.4617264485612926, "grad_norm": 0.4362448453903198, "learning_rate": 0.0005, "loss": 1.2147774696350098, "step": 31227 }, { "epoch": 2.4618052818289318, "grad_norm": 0.45516785979270935, "learning_rate": 0.0005, "loss": 1.318514108657837, "step": 31228 }, { "epoch": 2.461884115096571, "grad_norm": 0.605704128742218, "learning_rate": 0.0005, "loss": 1.3933289051055908, "step": 31229 }, { "epoch": 2.4619629483642096, "grad_norm": 0.4527350068092346, "learning_rate": 0.0005, "loss": 1.058240532875061, "step": 31230 }, { "epoch": 2.4620417816318487, "grad_norm": 0.6153746247291565, "learning_rate": 0.0005, "loss": 1.398686170578003, "step": 31231 }, { "epoch": 2.4621206148994874, "grad_norm": 0.4430326223373413, "learning_rate": 0.0005, "loss": 1.3021990060806274, "step": 31232 }, { "epoch": 2.4621994481671265, "grad_norm": 0.5173069834709167, "learning_rate": 0.0005, "loss": 1.4655125141143799, "step": 31233 }, { "epoch": 2.4622782814347657, "grad_norm": 0.5918560028076172, "learning_rate": 0.0005, "loss": 1.3336939811706543, "step": 31234 }, { "epoch": 2.4623571147024044, "grad_norm": 0.5151349306106567, "learning_rate": 0.0005, "loss": 1.532987356185913, "step": 31235 }, { "epoch": 2.4624359479700435, "grad_norm": 0.4992712140083313, "learning_rate": 0.0005, "loss": 1.1531388759613037, "step": 31236 }, { "epoch": 2.462514781237682, "grad_norm": 0.5374547243118286, "learning_rate": 0.0005, "loss": 1.3223072290420532, "step": 31237 }, { "epoch": 2.4625936145053213, "grad_norm": 0.6296115517616272, "learning_rate": 0.0005, "loss": 1.4087233543395996, "step": 31238 }, { "epoch": 2.46267244777296, "grad_norm": 0.46506646275520325, "learning_rate": 0.0005, "loss": 1.2063281536102295, "step": 31239 }, { "epoch": 2.462751281040599, "grad_norm": 0.5913813710212708, "learning_rate": 0.0005, "loss": 1.524850845336914, "step": 31240 }, { "epoch": 2.462830114308238, "grad_norm": 0.5458057522773743, "learning_rate": 0.0005, "loss": 1.3273506164550781, "step": 31241 }, { "epoch": 2.462908947575877, "grad_norm": 0.5137434005737305, "learning_rate": 0.0005, "loss": 1.3774336576461792, "step": 31242 }, { "epoch": 2.462987780843516, "grad_norm": 0.4848197400569916, "learning_rate": 0.0005, "loss": 1.2879369258880615, "step": 31243 }, { "epoch": 2.463066614111155, "grad_norm": 0.49428024888038635, "learning_rate": 0.0005, "loss": 1.1585371494293213, "step": 31244 }, { "epoch": 2.463145447378794, "grad_norm": 0.4491903483867645, "learning_rate": 0.0005, "loss": 1.3719110488891602, "step": 31245 }, { "epoch": 2.4632242806464326, "grad_norm": 0.5489470362663269, "learning_rate": 0.0005, "loss": 1.2051621675491333, "step": 31246 }, { "epoch": 2.4633031139140718, "grad_norm": 0.4921831786632538, "learning_rate": 0.0005, "loss": 1.2287755012512207, "step": 31247 }, { "epoch": 2.463381947181711, "grad_norm": 0.5534164309501648, "learning_rate": 0.0005, "loss": 1.3983440399169922, "step": 31248 }, { "epoch": 2.4634607804493496, "grad_norm": 0.429897278547287, "learning_rate": 0.0005, "loss": 1.078765869140625, "step": 31249 }, { "epoch": 2.4635396137169887, "grad_norm": 0.5077419877052307, "learning_rate": 0.0005, "loss": 1.2880258560180664, "step": 31250 }, { "epoch": 2.4636184469846274, "grad_norm": 0.48188117146492004, "learning_rate": 0.0005, "loss": 1.341726303100586, "step": 31251 }, { "epoch": 2.4636972802522665, "grad_norm": 0.4709281921386719, "learning_rate": 0.0005, "loss": 1.1857788562774658, "step": 31252 }, { "epoch": 2.463776113519905, "grad_norm": 0.4897935092449188, "learning_rate": 0.0005, "loss": 1.3449397087097168, "step": 31253 }, { "epoch": 2.4638549467875444, "grad_norm": 0.540160596370697, "learning_rate": 0.0005, "loss": 1.5393469333648682, "step": 31254 }, { "epoch": 2.463933780055183, "grad_norm": 0.5373311638832092, "learning_rate": 0.0005, "loss": 1.3515653610229492, "step": 31255 }, { "epoch": 2.464012613322822, "grad_norm": 0.5486889481544495, "learning_rate": 0.0005, "loss": 1.1417349576950073, "step": 31256 }, { "epoch": 2.4640914465904613, "grad_norm": 0.43492069840431213, "learning_rate": 0.0005, "loss": 1.1676620244979858, "step": 31257 }, { "epoch": 2.4641702798581, "grad_norm": 0.5226885080337524, "learning_rate": 0.0005, "loss": 1.208543062210083, "step": 31258 }, { "epoch": 2.464249113125739, "grad_norm": 0.51901775598526, "learning_rate": 0.0005, "loss": 1.3824453353881836, "step": 31259 }, { "epoch": 2.464327946393378, "grad_norm": 0.6714298129081726, "learning_rate": 0.0005, "loss": 1.5058889389038086, "step": 31260 }, { "epoch": 2.464406779661017, "grad_norm": 0.44645607471466064, "learning_rate": 0.0005, "loss": 1.17880117893219, "step": 31261 }, { "epoch": 2.464485612928656, "grad_norm": 0.5729249119758606, "learning_rate": 0.0005, "loss": 1.2261545658111572, "step": 31262 }, { "epoch": 2.464564446196295, "grad_norm": 0.6264291405677795, "learning_rate": 0.0005, "loss": 1.3511319160461426, "step": 31263 }, { "epoch": 2.464643279463934, "grad_norm": 0.5130385160446167, "learning_rate": 0.0005, "loss": 1.2774419784545898, "step": 31264 }, { "epoch": 2.4647221127315726, "grad_norm": 0.5138610601425171, "learning_rate": 0.0005, "loss": 1.1717159748077393, "step": 31265 }, { "epoch": 2.4648009459992117, "grad_norm": 0.468068391084671, "learning_rate": 0.0005, "loss": 1.3313117027282715, "step": 31266 }, { "epoch": 2.4648797792668504, "grad_norm": 0.4909861981868744, "learning_rate": 0.0005, "loss": 1.317183256149292, "step": 31267 }, { "epoch": 2.4649586125344896, "grad_norm": 0.5570735335350037, "learning_rate": 0.0005, "loss": 1.263283133506775, "step": 31268 }, { "epoch": 2.4650374458021287, "grad_norm": 0.46356385946273804, "learning_rate": 0.0005, "loss": 1.3032339811325073, "step": 31269 }, { "epoch": 2.4651162790697674, "grad_norm": 0.6618177890777588, "learning_rate": 0.0005, "loss": 1.2185273170471191, "step": 31270 }, { "epoch": 2.4651951123374065, "grad_norm": 0.5852181911468506, "learning_rate": 0.0005, "loss": 1.2052509784698486, "step": 31271 }, { "epoch": 2.465273945605045, "grad_norm": 0.4741609990596771, "learning_rate": 0.0005, "loss": 1.281721591949463, "step": 31272 }, { "epoch": 2.4653527788726843, "grad_norm": 0.5576865077018738, "learning_rate": 0.0005, "loss": 1.392397165298462, "step": 31273 }, { "epoch": 2.465431612140323, "grad_norm": 0.5427865982055664, "learning_rate": 0.0005, "loss": 1.5380101203918457, "step": 31274 }, { "epoch": 2.465510445407962, "grad_norm": 0.4508167803287506, "learning_rate": 0.0005, "loss": 1.2902584075927734, "step": 31275 }, { "epoch": 2.4655892786756013, "grad_norm": 0.5543376207351685, "learning_rate": 0.0005, "loss": 1.3909928798675537, "step": 31276 }, { "epoch": 2.46566811194324, "grad_norm": 0.5501148104667664, "learning_rate": 0.0005, "loss": 1.4866446256637573, "step": 31277 }, { "epoch": 2.465746945210879, "grad_norm": 0.4376738369464874, "learning_rate": 0.0005, "loss": 1.4065426588058472, "step": 31278 }, { "epoch": 2.465825778478518, "grad_norm": 0.44664233922958374, "learning_rate": 0.0005, "loss": 1.192968726158142, "step": 31279 }, { "epoch": 2.465904611746157, "grad_norm": 0.6731731295585632, "learning_rate": 0.0005, "loss": 1.3380906581878662, "step": 31280 }, { "epoch": 2.465983445013796, "grad_norm": 0.5589506030082703, "learning_rate": 0.0005, "loss": 1.2508310079574585, "step": 31281 }, { "epoch": 2.4660622782814348, "grad_norm": 0.6424784064292908, "learning_rate": 0.0005, "loss": 1.4719762802124023, "step": 31282 }, { "epoch": 2.466141111549074, "grad_norm": 0.5465372204780579, "learning_rate": 0.0005, "loss": 1.4177886247634888, "step": 31283 }, { "epoch": 2.4662199448167126, "grad_norm": 0.5522673726081848, "learning_rate": 0.0005, "loss": 1.4661755561828613, "step": 31284 }, { "epoch": 2.4662987780843517, "grad_norm": 0.5926810503005981, "learning_rate": 0.0005, "loss": 1.415608286857605, "step": 31285 }, { "epoch": 2.4663776113519904, "grad_norm": 0.5557354688644409, "learning_rate": 0.0005, "loss": 1.4798192977905273, "step": 31286 }, { "epoch": 2.4664564446196295, "grad_norm": 0.49689334630966187, "learning_rate": 0.0005, "loss": 1.2002277374267578, "step": 31287 }, { "epoch": 2.4665352778872682, "grad_norm": 0.46007028222084045, "learning_rate": 0.0005, "loss": 1.2327690124511719, "step": 31288 }, { "epoch": 2.4666141111549074, "grad_norm": 0.5421808362007141, "learning_rate": 0.0005, "loss": 1.526469111442566, "step": 31289 }, { "epoch": 2.4666929444225465, "grad_norm": 0.6253250241279602, "learning_rate": 0.0005, "loss": 1.2632269859313965, "step": 31290 }, { "epoch": 2.466771777690185, "grad_norm": 0.5728278756141663, "learning_rate": 0.0005, "loss": 1.3036198616027832, "step": 31291 }, { "epoch": 2.4668506109578243, "grad_norm": 0.5666942596435547, "learning_rate": 0.0005, "loss": 1.300620436668396, "step": 31292 }, { "epoch": 2.466929444225463, "grad_norm": 0.4915117621421814, "learning_rate": 0.0005, "loss": 1.4522457122802734, "step": 31293 }, { "epoch": 2.467008277493102, "grad_norm": 0.5554187297821045, "learning_rate": 0.0005, "loss": 1.2358468770980835, "step": 31294 }, { "epoch": 2.4670871107607413, "grad_norm": 0.41175153851509094, "learning_rate": 0.0005, "loss": 1.1678850650787354, "step": 31295 }, { "epoch": 2.46716594402838, "grad_norm": 0.5143278241157532, "learning_rate": 0.0005, "loss": 1.5400108098983765, "step": 31296 }, { "epoch": 2.467244777296019, "grad_norm": 0.44828125834465027, "learning_rate": 0.0005, "loss": 1.2997117042541504, "step": 31297 }, { "epoch": 2.467323610563658, "grad_norm": 0.4819943904876709, "learning_rate": 0.0005, "loss": 1.2380733489990234, "step": 31298 }, { "epoch": 2.467402443831297, "grad_norm": 0.4767935276031494, "learning_rate": 0.0005, "loss": 1.4965308904647827, "step": 31299 }, { "epoch": 2.4674812770989356, "grad_norm": 0.5462777614593506, "learning_rate": 0.0005, "loss": 1.4435713291168213, "step": 31300 }, { "epoch": 2.4675601103665747, "grad_norm": 1.1885606050491333, "learning_rate": 0.0005, "loss": 1.347824215888977, "step": 31301 }, { "epoch": 2.4676389436342134, "grad_norm": 0.5408488512039185, "learning_rate": 0.0005, "loss": 1.323135256767273, "step": 31302 }, { "epoch": 2.4677177769018526, "grad_norm": 0.4792741537094116, "learning_rate": 0.0005, "loss": 1.3178133964538574, "step": 31303 }, { "epoch": 2.4677966101694917, "grad_norm": 0.4570510983467102, "learning_rate": 0.0005, "loss": 1.1889744997024536, "step": 31304 }, { "epoch": 2.4678754434371304, "grad_norm": 0.4570460617542267, "learning_rate": 0.0005, "loss": 1.220353126525879, "step": 31305 }, { "epoch": 2.4679542767047695, "grad_norm": 0.7030482292175293, "learning_rate": 0.0005, "loss": 1.2795889377593994, "step": 31306 }, { "epoch": 2.468033109972408, "grad_norm": 0.6596150994300842, "learning_rate": 0.0005, "loss": 1.1937437057495117, "step": 31307 }, { "epoch": 2.4681119432400473, "grad_norm": 0.6190639734268188, "learning_rate": 0.0005, "loss": 1.2767963409423828, "step": 31308 }, { "epoch": 2.4681907765076865, "grad_norm": 0.43174251914024353, "learning_rate": 0.0005, "loss": 1.15114164352417, "step": 31309 }, { "epoch": 2.468269609775325, "grad_norm": 0.4886023700237274, "learning_rate": 0.0005, "loss": 1.2616972923278809, "step": 31310 }, { "epoch": 2.4683484430429643, "grad_norm": 0.533765971660614, "learning_rate": 0.0005, "loss": 1.4835556745529175, "step": 31311 }, { "epoch": 2.468427276310603, "grad_norm": 0.5208571553230286, "learning_rate": 0.0005, "loss": 1.258192777633667, "step": 31312 }, { "epoch": 2.468506109578242, "grad_norm": 0.6648029088973999, "learning_rate": 0.0005, "loss": 1.5123398303985596, "step": 31313 }, { "epoch": 2.468584942845881, "grad_norm": 0.46766984462738037, "learning_rate": 0.0005, "loss": 1.3132601976394653, "step": 31314 }, { "epoch": 2.46866377611352, "grad_norm": 0.501915693283081, "learning_rate": 0.0005, "loss": 1.5158650875091553, "step": 31315 }, { "epoch": 2.4687426093811586, "grad_norm": 0.455033540725708, "learning_rate": 0.0005, "loss": 1.1166716814041138, "step": 31316 }, { "epoch": 2.4688214426487978, "grad_norm": 0.5366023182868958, "learning_rate": 0.0005, "loss": 1.3616621494293213, "step": 31317 }, { "epoch": 2.468900275916437, "grad_norm": 0.45581507682800293, "learning_rate": 0.0005, "loss": 1.240263819694519, "step": 31318 }, { "epoch": 2.4689791091840756, "grad_norm": 0.5721766948699951, "learning_rate": 0.0005, "loss": 1.316819190979004, "step": 31319 }, { "epoch": 2.4690579424517147, "grad_norm": 0.5674936771392822, "learning_rate": 0.0005, "loss": 1.2571640014648438, "step": 31320 }, { "epoch": 2.4691367757193534, "grad_norm": 0.4960081875324249, "learning_rate": 0.0005, "loss": 1.4261415004730225, "step": 31321 }, { "epoch": 2.4692156089869925, "grad_norm": 0.49278581142425537, "learning_rate": 0.0005, "loss": 1.5062485933303833, "step": 31322 }, { "epoch": 2.4692944422546317, "grad_norm": 0.4197938144207001, "learning_rate": 0.0005, "loss": 1.2543474435806274, "step": 31323 }, { "epoch": 2.4693732755222704, "grad_norm": 0.45713678002357483, "learning_rate": 0.0005, "loss": 1.341724157333374, "step": 31324 }, { "epoch": 2.4694521087899095, "grad_norm": 0.5428403615951538, "learning_rate": 0.0005, "loss": 1.464605450630188, "step": 31325 }, { "epoch": 2.469530942057548, "grad_norm": 0.45742809772491455, "learning_rate": 0.0005, "loss": 1.3381097316741943, "step": 31326 }, { "epoch": 2.4696097753251873, "grad_norm": 0.4442489445209503, "learning_rate": 0.0005, "loss": 1.394314169883728, "step": 31327 }, { "epoch": 2.469688608592826, "grad_norm": 0.569998562335968, "learning_rate": 0.0005, "loss": 1.4667387008666992, "step": 31328 }, { "epoch": 2.469767441860465, "grad_norm": 0.4463815987110138, "learning_rate": 0.0005, "loss": 1.2490272521972656, "step": 31329 }, { "epoch": 2.469846275128104, "grad_norm": 0.3723249137401581, "learning_rate": 0.0005, "loss": 1.0952767133712769, "step": 31330 }, { "epoch": 2.469925108395743, "grad_norm": 0.39337101578712463, "learning_rate": 0.0005, "loss": 1.4290200471878052, "step": 31331 }, { "epoch": 2.470003941663382, "grad_norm": 0.6159021854400635, "learning_rate": 0.0005, "loss": 1.2274608612060547, "step": 31332 }, { "epoch": 2.470082774931021, "grad_norm": 0.5468648076057434, "learning_rate": 0.0005, "loss": 1.4549736976623535, "step": 31333 }, { "epoch": 2.47016160819866, "grad_norm": 0.4644947648048401, "learning_rate": 0.0005, "loss": 1.3261607885360718, "step": 31334 }, { "epoch": 2.4702404414662986, "grad_norm": 0.469129353761673, "learning_rate": 0.0005, "loss": 1.2920360565185547, "step": 31335 }, { "epoch": 2.4703192747339378, "grad_norm": 0.7750182747840881, "learning_rate": 0.0005, "loss": 1.163599967956543, "step": 31336 }, { "epoch": 2.470398108001577, "grad_norm": 0.6451838612556458, "learning_rate": 0.0005, "loss": 1.4531711339950562, "step": 31337 }, { "epoch": 2.4704769412692156, "grad_norm": 0.6110767126083374, "learning_rate": 0.0005, "loss": 1.5413151979446411, "step": 31338 }, { "epoch": 2.4705557745368547, "grad_norm": 0.5492854714393616, "learning_rate": 0.0005, "loss": 1.2659833431243896, "step": 31339 }, { "epoch": 2.4706346078044934, "grad_norm": 0.5361050367355347, "learning_rate": 0.0005, "loss": 1.4578068256378174, "step": 31340 }, { "epoch": 2.4707134410721325, "grad_norm": 0.513798177242279, "learning_rate": 0.0005, "loss": 1.201242446899414, "step": 31341 }, { "epoch": 2.470792274339771, "grad_norm": 0.49815279245376587, "learning_rate": 0.0005, "loss": 1.1889219284057617, "step": 31342 }, { "epoch": 2.4708711076074104, "grad_norm": 0.5532183051109314, "learning_rate": 0.0005, "loss": 1.306114912033081, "step": 31343 }, { "epoch": 2.470949940875049, "grad_norm": 0.5950362682342529, "learning_rate": 0.0005, "loss": 1.4827978610992432, "step": 31344 }, { "epoch": 2.471028774142688, "grad_norm": 0.6427779197692871, "learning_rate": 0.0005, "loss": 1.4412773847579956, "step": 31345 }, { "epoch": 2.4711076074103273, "grad_norm": 0.5024493932723999, "learning_rate": 0.0005, "loss": 1.388623833656311, "step": 31346 }, { "epoch": 2.471186440677966, "grad_norm": 0.491793692111969, "learning_rate": 0.0005, "loss": 1.3661437034606934, "step": 31347 }, { "epoch": 2.471265273945605, "grad_norm": 0.5295433402061462, "learning_rate": 0.0005, "loss": 1.4184019565582275, "step": 31348 }, { "epoch": 2.471344107213244, "grad_norm": 0.4706728160381317, "learning_rate": 0.0005, "loss": 1.2256615161895752, "step": 31349 }, { "epoch": 2.471422940480883, "grad_norm": 0.5618335008621216, "learning_rate": 0.0005, "loss": 1.4062434434890747, "step": 31350 }, { "epoch": 2.471501773748522, "grad_norm": 0.5012410879135132, "learning_rate": 0.0005, "loss": 1.3251551389694214, "step": 31351 }, { "epoch": 2.471580607016161, "grad_norm": 0.5313887000083923, "learning_rate": 0.0005, "loss": 1.4726431369781494, "step": 31352 }, { "epoch": 2.4716594402838, "grad_norm": 0.5038907527923584, "learning_rate": 0.0005, "loss": 1.3560528755187988, "step": 31353 }, { "epoch": 2.4717382735514386, "grad_norm": 0.44436317682266235, "learning_rate": 0.0005, "loss": 1.262498140335083, "step": 31354 }, { "epoch": 2.4718171068190777, "grad_norm": 0.47526243329048157, "learning_rate": 0.0005, "loss": 1.44800865650177, "step": 31355 }, { "epoch": 2.4718959400867164, "grad_norm": 0.5219289064407349, "learning_rate": 0.0005, "loss": 1.1840726137161255, "step": 31356 }, { "epoch": 2.4719747733543556, "grad_norm": 0.541516900062561, "learning_rate": 0.0005, "loss": 1.4209630489349365, "step": 31357 }, { "epoch": 2.4720536066219942, "grad_norm": 0.5209288597106934, "learning_rate": 0.0005, "loss": 1.5928386449813843, "step": 31358 }, { "epoch": 2.4721324398896334, "grad_norm": 0.3828742206096649, "learning_rate": 0.0005, "loss": 1.123866081237793, "step": 31359 }, { "epoch": 2.4722112731572725, "grad_norm": 0.47745147347450256, "learning_rate": 0.0005, "loss": 1.4544991254806519, "step": 31360 }, { "epoch": 2.472290106424911, "grad_norm": 0.44593101739883423, "learning_rate": 0.0005, "loss": 1.2295243740081787, "step": 31361 }, { "epoch": 2.4723689396925503, "grad_norm": 0.614612877368927, "learning_rate": 0.0005, "loss": 1.3201439380645752, "step": 31362 }, { "epoch": 2.472447772960189, "grad_norm": 0.5924707055091858, "learning_rate": 0.0005, "loss": 1.1911147832870483, "step": 31363 }, { "epoch": 2.472526606227828, "grad_norm": 0.5244059562683105, "learning_rate": 0.0005, "loss": 1.580916166305542, "step": 31364 }, { "epoch": 2.4726054394954673, "grad_norm": 0.4653118848800659, "learning_rate": 0.0005, "loss": 1.2947561740875244, "step": 31365 }, { "epoch": 2.472684272763106, "grad_norm": 0.4936383366584778, "learning_rate": 0.0005, "loss": 1.2661176919937134, "step": 31366 }, { "epoch": 2.472763106030745, "grad_norm": 0.522914707660675, "learning_rate": 0.0005, "loss": 1.3394124507904053, "step": 31367 }, { "epoch": 2.472841939298384, "grad_norm": 0.4744962751865387, "learning_rate": 0.0005, "loss": 1.2245194911956787, "step": 31368 }, { "epoch": 2.472920772566023, "grad_norm": 0.6511791348457336, "learning_rate": 0.0005, "loss": 1.4101183414459229, "step": 31369 }, { "epoch": 2.4729996058336616, "grad_norm": 0.5208705067634583, "learning_rate": 0.0005, "loss": 1.4091566801071167, "step": 31370 }, { "epoch": 2.4730784391013008, "grad_norm": 0.519270122051239, "learning_rate": 0.0005, "loss": 1.356434941291809, "step": 31371 }, { "epoch": 2.4731572723689395, "grad_norm": 0.5536623001098633, "learning_rate": 0.0005, "loss": 1.488904356956482, "step": 31372 }, { "epoch": 2.4732361056365786, "grad_norm": 0.5856107473373413, "learning_rate": 0.0005, "loss": 1.3069703578948975, "step": 31373 }, { "epoch": 2.4733149389042177, "grad_norm": 0.689009428024292, "learning_rate": 0.0005, "loss": 1.3558404445648193, "step": 31374 }, { "epoch": 2.4733937721718564, "grad_norm": 0.6475136876106262, "learning_rate": 0.0005, "loss": 1.4514553546905518, "step": 31375 }, { "epoch": 2.4734726054394955, "grad_norm": 0.41122904419898987, "learning_rate": 0.0005, "loss": 1.3245612382888794, "step": 31376 }, { "epoch": 2.4735514387071342, "grad_norm": 0.5658074617385864, "learning_rate": 0.0005, "loss": 1.4287453889846802, "step": 31377 }, { "epoch": 2.4736302719747734, "grad_norm": 0.6145495176315308, "learning_rate": 0.0005, "loss": 1.637911319732666, "step": 31378 }, { "epoch": 2.4737091052424125, "grad_norm": 0.43192118406295776, "learning_rate": 0.0005, "loss": 1.2154343128204346, "step": 31379 }, { "epoch": 2.473787938510051, "grad_norm": 0.5028480291366577, "learning_rate": 0.0005, "loss": 1.2264378070831299, "step": 31380 }, { "epoch": 2.4738667717776903, "grad_norm": 0.5153944492340088, "learning_rate": 0.0005, "loss": 1.4278922080993652, "step": 31381 }, { "epoch": 2.473945605045329, "grad_norm": 0.878295361995697, "learning_rate": 0.0005, "loss": 1.3058210611343384, "step": 31382 }, { "epoch": 2.474024438312968, "grad_norm": 0.5406653881072998, "learning_rate": 0.0005, "loss": 1.4340033531188965, "step": 31383 }, { "epoch": 2.474103271580607, "grad_norm": 0.6033235192298889, "learning_rate": 0.0005, "loss": 1.2714403867721558, "step": 31384 }, { "epoch": 2.474182104848246, "grad_norm": 0.5447589755058289, "learning_rate": 0.0005, "loss": 1.286008596420288, "step": 31385 }, { "epoch": 2.4742609381158847, "grad_norm": 0.4648478031158447, "learning_rate": 0.0005, "loss": 1.227418065071106, "step": 31386 }, { "epoch": 2.474339771383524, "grad_norm": 0.518970251083374, "learning_rate": 0.0005, "loss": 1.2466938495635986, "step": 31387 }, { "epoch": 2.474418604651163, "grad_norm": 0.4934003949165344, "learning_rate": 0.0005, "loss": 1.1730471849441528, "step": 31388 }, { "epoch": 2.4744974379188016, "grad_norm": 0.517989993095398, "learning_rate": 0.0005, "loss": 1.3401870727539062, "step": 31389 }, { "epoch": 2.4745762711864407, "grad_norm": 0.5154250264167786, "learning_rate": 0.0005, "loss": 1.2641522884368896, "step": 31390 }, { "epoch": 2.4746551044540794, "grad_norm": 0.5703257918357849, "learning_rate": 0.0005, "loss": 1.3553760051727295, "step": 31391 }, { "epoch": 2.4747339377217186, "grad_norm": 0.43969833850860596, "learning_rate": 0.0005, "loss": 1.186655879020691, "step": 31392 }, { "epoch": 2.4748127709893577, "grad_norm": 0.591849148273468, "learning_rate": 0.0005, "loss": 1.7294180393218994, "step": 31393 }, { "epoch": 2.4748916042569964, "grad_norm": 0.42149806022644043, "learning_rate": 0.0005, "loss": 1.3623000383377075, "step": 31394 }, { "epoch": 2.4749704375246355, "grad_norm": 0.45585766434669495, "learning_rate": 0.0005, "loss": 1.2662227153778076, "step": 31395 }, { "epoch": 2.475049270792274, "grad_norm": 0.6509242057800293, "learning_rate": 0.0005, "loss": 1.1815848350524902, "step": 31396 }, { "epoch": 2.4751281040599133, "grad_norm": 0.49892377853393555, "learning_rate": 0.0005, "loss": 1.3226404190063477, "step": 31397 }, { "epoch": 2.475206937327552, "grad_norm": 0.5203495621681213, "learning_rate": 0.0005, "loss": 1.3986319303512573, "step": 31398 }, { "epoch": 2.475285770595191, "grad_norm": 0.4368940591812134, "learning_rate": 0.0005, "loss": 1.3366018533706665, "step": 31399 }, { "epoch": 2.4753646038628303, "grad_norm": 0.670466959476471, "learning_rate": 0.0005, "loss": 1.3288788795471191, "step": 31400 }, { "epoch": 2.475443437130469, "grad_norm": 0.4729747176170349, "learning_rate": 0.0005, "loss": 1.2652781009674072, "step": 31401 }, { "epoch": 2.475522270398108, "grad_norm": 0.406814306974411, "learning_rate": 0.0005, "loss": 1.2827948331832886, "step": 31402 }, { "epoch": 2.475601103665747, "grad_norm": 0.5198704600334167, "learning_rate": 0.0005, "loss": 1.5064057111740112, "step": 31403 }, { "epoch": 2.475679936933386, "grad_norm": 0.5092067718505859, "learning_rate": 0.0005, "loss": 1.3417510986328125, "step": 31404 }, { "epoch": 2.4757587702010246, "grad_norm": 0.44703033566474915, "learning_rate": 0.0005, "loss": 1.2557358741760254, "step": 31405 }, { "epoch": 2.4758376034686638, "grad_norm": 0.6810908913612366, "learning_rate": 0.0005, "loss": 1.2767630815505981, "step": 31406 }, { "epoch": 2.475916436736303, "grad_norm": 0.41537755727767944, "learning_rate": 0.0005, "loss": 1.081099510192871, "step": 31407 }, { "epoch": 2.4759952700039416, "grad_norm": 0.5104990601539612, "learning_rate": 0.0005, "loss": 1.422696828842163, "step": 31408 }, { "epoch": 2.4760741032715807, "grad_norm": 0.5180765986442566, "learning_rate": 0.0005, "loss": 1.3792874813079834, "step": 31409 }, { "epoch": 2.4761529365392194, "grad_norm": 0.44583290815353394, "learning_rate": 0.0005, "loss": 1.1236463785171509, "step": 31410 }, { "epoch": 2.4762317698068586, "grad_norm": 0.6098383665084839, "learning_rate": 0.0005, "loss": 1.485175609588623, "step": 31411 }, { "epoch": 2.4763106030744977, "grad_norm": 0.5118913650512695, "learning_rate": 0.0005, "loss": 1.4252054691314697, "step": 31412 }, { "epoch": 2.4763894363421364, "grad_norm": 0.42898452281951904, "learning_rate": 0.0005, "loss": 1.2299232482910156, "step": 31413 }, { "epoch": 2.4764682696097755, "grad_norm": 0.5410791635513306, "learning_rate": 0.0005, "loss": 1.3279945850372314, "step": 31414 }, { "epoch": 2.476547102877414, "grad_norm": 0.4918934404850006, "learning_rate": 0.0005, "loss": 1.4414455890655518, "step": 31415 }, { "epoch": 2.4766259361450533, "grad_norm": 0.46445557475090027, "learning_rate": 0.0005, "loss": 1.2518821954727173, "step": 31416 }, { "epoch": 2.476704769412692, "grad_norm": 0.5369288325309753, "learning_rate": 0.0005, "loss": 1.3820431232452393, "step": 31417 }, { "epoch": 2.476783602680331, "grad_norm": 0.551941454410553, "learning_rate": 0.0005, "loss": 1.442175030708313, "step": 31418 }, { "epoch": 2.47686243594797, "grad_norm": 0.5734333992004395, "learning_rate": 0.0005, "loss": 1.5070292949676514, "step": 31419 }, { "epoch": 2.476941269215609, "grad_norm": 0.5571244955062866, "learning_rate": 0.0005, "loss": 1.333364725112915, "step": 31420 }, { "epoch": 2.477020102483248, "grad_norm": 0.4735443890094757, "learning_rate": 0.0005, "loss": 1.3474713563919067, "step": 31421 }, { "epoch": 2.477098935750887, "grad_norm": 0.4237407445907593, "learning_rate": 0.0005, "loss": 1.276800513267517, "step": 31422 }, { "epoch": 2.477177769018526, "grad_norm": 0.560003936290741, "learning_rate": 0.0005, "loss": 1.3697102069854736, "step": 31423 }, { "epoch": 2.4772566022861646, "grad_norm": 0.5891830325126648, "learning_rate": 0.0005, "loss": 1.3702375888824463, "step": 31424 }, { "epoch": 2.4773354355538038, "grad_norm": 0.38367152214050293, "learning_rate": 0.0005, "loss": 1.178128719329834, "step": 31425 }, { "epoch": 2.477414268821443, "grad_norm": 0.4460093080997467, "learning_rate": 0.0005, "loss": 1.334695816040039, "step": 31426 }, { "epoch": 2.4774931020890816, "grad_norm": 0.6624167561531067, "learning_rate": 0.0005, "loss": 1.2920868396759033, "step": 31427 }, { "epoch": 2.4775719353567207, "grad_norm": 0.5363987684249878, "learning_rate": 0.0005, "loss": 1.2655885219573975, "step": 31428 }, { "epoch": 2.4776507686243594, "grad_norm": 0.4694497883319855, "learning_rate": 0.0005, "loss": 1.2511922121047974, "step": 31429 }, { "epoch": 2.4777296018919985, "grad_norm": 0.474735826253891, "learning_rate": 0.0005, "loss": 1.3497676849365234, "step": 31430 }, { "epoch": 2.4778084351596372, "grad_norm": 0.7459419965744019, "learning_rate": 0.0005, "loss": 1.3183037042617798, "step": 31431 }, { "epoch": 2.4778872684272764, "grad_norm": 0.43873363733291626, "learning_rate": 0.0005, "loss": 1.237762689590454, "step": 31432 }, { "epoch": 2.477966101694915, "grad_norm": 0.5524219870567322, "learning_rate": 0.0005, "loss": 1.4418708086013794, "step": 31433 }, { "epoch": 2.478044934962554, "grad_norm": 0.4542723000049591, "learning_rate": 0.0005, "loss": 1.3615245819091797, "step": 31434 }, { "epoch": 2.4781237682301933, "grad_norm": 0.4400070309638977, "learning_rate": 0.0005, "loss": 1.2643835544586182, "step": 31435 }, { "epoch": 2.478202601497832, "grad_norm": 0.4978831112384796, "learning_rate": 0.0005, "loss": 1.2505087852478027, "step": 31436 }, { "epoch": 2.478281434765471, "grad_norm": 0.41989031434059143, "learning_rate": 0.0005, "loss": 1.309800624847412, "step": 31437 }, { "epoch": 2.47836026803311, "grad_norm": 0.45762506127357483, "learning_rate": 0.0005, "loss": 1.359163761138916, "step": 31438 }, { "epoch": 2.478439101300749, "grad_norm": 0.4854797422885895, "learning_rate": 0.0005, "loss": 1.4626309871673584, "step": 31439 }, { "epoch": 2.478517934568388, "grad_norm": 0.5124834775924683, "learning_rate": 0.0005, "loss": 1.4556326866149902, "step": 31440 }, { "epoch": 2.478596767836027, "grad_norm": 0.4763081669807434, "learning_rate": 0.0005, "loss": 1.3905744552612305, "step": 31441 }, { "epoch": 2.478675601103666, "grad_norm": 0.4664660096168518, "learning_rate": 0.0005, "loss": 1.4799238443374634, "step": 31442 }, { "epoch": 2.4787544343713046, "grad_norm": 0.5242665410041809, "learning_rate": 0.0005, "loss": 1.3674654960632324, "step": 31443 }, { "epoch": 2.4788332676389437, "grad_norm": 0.4678032100200653, "learning_rate": 0.0005, "loss": 1.4292173385620117, "step": 31444 }, { "epoch": 2.4789121009065824, "grad_norm": 0.5365962386131287, "learning_rate": 0.0005, "loss": 1.3424372673034668, "step": 31445 }, { "epoch": 2.4789909341742216, "grad_norm": 0.4718508720397949, "learning_rate": 0.0005, "loss": 1.3309532403945923, "step": 31446 }, { "epoch": 2.4790697674418603, "grad_norm": 0.4661381244659424, "learning_rate": 0.0005, "loss": 1.1745250225067139, "step": 31447 }, { "epoch": 2.4791486007094994, "grad_norm": 0.4517875909805298, "learning_rate": 0.0005, "loss": 1.262944221496582, "step": 31448 }, { "epoch": 2.4792274339771385, "grad_norm": 0.5106662511825562, "learning_rate": 0.0005, "loss": 1.3693630695343018, "step": 31449 }, { "epoch": 2.479306267244777, "grad_norm": 0.4581713378429413, "learning_rate": 0.0005, "loss": 1.443250298500061, "step": 31450 }, { "epoch": 2.4793851005124163, "grad_norm": 0.5014910697937012, "learning_rate": 0.0005, "loss": 1.3099901676177979, "step": 31451 }, { "epoch": 2.479463933780055, "grad_norm": 0.5190889835357666, "learning_rate": 0.0005, "loss": 1.263413906097412, "step": 31452 }, { "epoch": 2.479542767047694, "grad_norm": 0.4287477135658264, "learning_rate": 0.0005, "loss": 1.252655267715454, "step": 31453 }, { "epoch": 2.4796216003153333, "grad_norm": 0.6046026945114136, "learning_rate": 0.0005, "loss": 1.3414201736450195, "step": 31454 }, { "epoch": 2.479700433582972, "grad_norm": 0.41853412985801697, "learning_rate": 0.0005, "loss": 1.2115428447723389, "step": 31455 }, { "epoch": 2.479779266850611, "grad_norm": 0.5034335851669312, "learning_rate": 0.0005, "loss": 1.4759420156478882, "step": 31456 }, { "epoch": 2.47985810011825, "grad_norm": 0.4802711009979248, "learning_rate": 0.0005, "loss": 1.1656724214553833, "step": 31457 }, { "epoch": 2.479936933385889, "grad_norm": 0.4738692343235016, "learning_rate": 0.0005, "loss": 1.3736457824707031, "step": 31458 }, { "epoch": 2.4800157666535276, "grad_norm": 0.5272923111915588, "learning_rate": 0.0005, "loss": 1.4627292156219482, "step": 31459 }, { "epoch": 2.4800945999211668, "grad_norm": 0.719923198223114, "learning_rate": 0.0005, "loss": 1.485215425491333, "step": 31460 }, { "epoch": 2.4801734331888055, "grad_norm": 0.5194142460823059, "learning_rate": 0.0005, "loss": 1.3931505680084229, "step": 31461 }, { "epoch": 2.4802522664564446, "grad_norm": 0.6017105579376221, "learning_rate": 0.0005, "loss": 1.371741533279419, "step": 31462 }, { "epoch": 2.4803310997240837, "grad_norm": 0.7190672159194946, "learning_rate": 0.0005, "loss": 1.525275468826294, "step": 31463 }, { "epoch": 2.4804099329917224, "grad_norm": 0.5620349049568176, "learning_rate": 0.0005, "loss": 1.307424783706665, "step": 31464 }, { "epoch": 2.4804887662593615, "grad_norm": 0.49172669649124146, "learning_rate": 0.0005, "loss": 1.4414178133010864, "step": 31465 }, { "epoch": 2.4805675995270002, "grad_norm": 0.4696592390537262, "learning_rate": 0.0005, "loss": 1.3020389080047607, "step": 31466 }, { "epoch": 2.4806464327946394, "grad_norm": 0.46746066212654114, "learning_rate": 0.0005, "loss": 1.3231251239776611, "step": 31467 }, { "epoch": 2.4807252660622785, "grad_norm": 0.6537434458732605, "learning_rate": 0.0005, "loss": 1.2492921352386475, "step": 31468 }, { "epoch": 2.480804099329917, "grad_norm": 0.4303223788738251, "learning_rate": 0.0005, "loss": 1.2016196250915527, "step": 31469 }, { "epoch": 2.4808829325975563, "grad_norm": 0.5039765238761902, "learning_rate": 0.0005, "loss": 1.396738886833191, "step": 31470 }, { "epoch": 2.480961765865195, "grad_norm": 0.6321219801902771, "learning_rate": 0.0005, "loss": 1.3737938404083252, "step": 31471 }, { "epoch": 2.481040599132834, "grad_norm": 0.4038560092449188, "learning_rate": 0.0005, "loss": 1.1796423196792603, "step": 31472 }, { "epoch": 2.481119432400473, "grad_norm": 0.5981714725494385, "learning_rate": 0.0005, "loss": 1.6413168907165527, "step": 31473 }, { "epoch": 2.481198265668112, "grad_norm": 0.5060940980911255, "learning_rate": 0.0005, "loss": 1.3517866134643555, "step": 31474 }, { "epoch": 2.4812770989357507, "grad_norm": 0.3695501983165741, "learning_rate": 0.0005, "loss": 1.1956899166107178, "step": 31475 }, { "epoch": 2.48135593220339, "grad_norm": 0.43565833568573, "learning_rate": 0.0005, "loss": 1.2775084972381592, "step": 31476 }, { "epoch": 2.481434765471029, "grad_norm": 0.5088937282562256, "learning_rate": 0.0005, "loss": 1.270320177078247, "step": 31477 }, { "epoch": 2.4815135987386676, "grad_norm": 0.5189148783683777, "learning_rate": 0.0005, "loss": 1.2026329040527344, "step": 31478 }, { "epoch": 2.4815924320063067, "grad_norm": 0.5219128131866455, "learning_rate": 0.0005, "loss": 1.3972500562667847, "step": 31479 }, { "epoch": 2.4816712652739454, "grad_norm": 0.4350242018699646, "learning_rate": 0.0005, "loss": 1.2149221897125244, "step": 31480 }, { "epoch": 2.4817500985415846, "grad_norm": 0.5678172707557678, "learning_rate": 0.0005, "loss": 1.2883754968643188, "step": 31481 }, { "epoch": 2.4818289318092237, "grad_norm": 0.5848571062088013, "learning_rate": 0.0005, "loss": 1.211958885192871, "step": 31482 }, { "epoch": 2.4819077650768624, "grad_norm": 0.4801257252693176, "learning_rate": 0.0005, "loss": 1.374436378479004, "step": 31483 }, { "epoch": 2.4819865983445015, "grad_norm": 0.5876219272613525, "learning_rate": 0.0005, "loss": 1.3254907131195068, "step": 31484 }, { "epoch": 2.48206543161214, "grad_norm": 0.5679991841316223, "learning_rate": 0.0005, "loss": 1.4391651153564453, "step": 31485 }, { "epoch": 2.4821442648797793, "grad_norm": 0.41159456968307495, "learning_rate": 0.0005, "loss": 1.152111291885376, "step": 31486 }, { "epoch": 2.482223098147418, "grad_norm": 0.5157570242881775, "learning_rate": 0.0005, "loss": 1.2928335666656494, "step": 31487 }, { "epoch": 2.482301931415057, "grad_norm": 0.44301414489746094, "learning_rate": 0.0005, "loss": 1.1067116260528564, "step": 31488 }, { "epoch": 2.482380764682696, "grad_norm": 0.46765756607055664, "learning_rate": 0.0005, "loss": 1.380023717880249, "step": 31489 }, { "epoch": 2.482459597950335, "grad_norm": 0.4425901472568512, "learning_rate": 0.0005, "loss": 1.16141676902771, "step": 31490 }, { "epoch": 2.482538431217974, "grad_norm": 0.4294007420539856, "learning_rate": 0.0005, "loss": 1.2606459856033325, "step": 31491 }, { "epoch": 2.482617264485613, "grad_norm": 0.47140049934387207, "learning_rate": 0.0005, "loss": 1.4641873836517334, "step": 31492 }, { "epoch": 2.482696097753252, "grad_norm": 0.5322067141532898, "learning_rate": 0.0005, "loss": 1.3700047731399536, "step": 31493 }, { "epoch": 2.4827749310208906, "grad_norm": 0.43198904395103455, "learning_rate": 0.0005, "loss": 1.148277759552002, "step": 31494 }, { "epoch": 2.4828537642885298, "grad_norm": 0.4951433539390564, "learning_rate": 0.0005, "loss": 1.1966865062713623, "step": 31495 }, { "epoch": 2.482932597556169, "grad_norm": 0.4712073802947998, "learning_rate": 0.0005, "loss": 1.206404685974121, "step": 31496 }, { "epoch": 2.4830114308238076, "grad_norm": 0.663231611251831, "learning_rate": 0.0005, "loss": 1.2637627124786377, "step": 31497 }, { "epoch": 2.4830902640914467, "grad_norm": 0.47348201274871826, "learning_rate": 0.0005, "loss": 1.2288224697113037, "step": 31498 }, { "epoch": 2.4831690973590854, "grad_norm": 0.43472346663475037, "learning_rate": 0.0005, "loss": 1.223466396331787, "step": 31499 }, { "epoch": 2.4832479306267246, "grad_norm": 0.4904049038887024, "learning_rate": 0.0005, "loss": 1.3101025819778442, "step": 31500 }, { "epoch": 2.4833267638943632, "grad_norm": 0.5067062973976135, "learning_rate": 0.0005, "loss": 1.4248442649841309, "step": 31501 }, { "epoch": 2.4834055971620024, "grad_norm": 0.5616899132728577, "learning_rate": 0.0005, "loss": 1.4456074237823486, "step": 31502 }, { "epoch": 2.483484430429641, "grad_norm": 0.6296722292900085, "learning_rate": 0.0005, "loss": 1.4188401699066162, "step": 31503 }, { "epoch": 2.48356326369728, "grad_norm": 0.4868697226047516, "learning_rate": 0.0005, "loss": 1.3194259405136108, "step": 31504 }, { "epoch": 2.4836420969649193, "grad_norm": 0.5730339884757996, "learning_rate": 0.0005, "loss": 1.327782392501831, "step": 31505 }, { "epoch": 2.483720930232558, "grad_norm": 0.7359772324562073, "learning_rate": 0.0005, "loss": 1.4457125663757324, "step": 31506 }, { "epoch": 2.483799763500197, "grad_norm": 0.5172923803329468, "learning_rate": 0.0005, "loss": 1.2786624431610107, "step": 31507 }, { "epoch": 2.483878596767836, "grad_norm": 0.4858214259147644, "learning_rate": 0.0005, "loss": 1.2812836170196533, "step": 31508 }, { "epoch": 2.483957430035475, "grad_norm": 0.6053157448768616, "learning_rate": 0.0005, "loss": 1.3000479936599731, "step": 31509 }, { "epoch": 2.484036263303114, "grad_norm": 0.506136953830719, "learning_rate": 0.0005, "loss": 1.3135173320770264, "step": 31510 }, { "epoch": 2.484115096570753, "grad_norm": 0.5586593747138977, "learning_rate": 0.0005, "loss": 1.5177156925201416, "step": 31511 }, { "epoch": 2.484193929838392, "grad_norm": 0.5669669508934021, "learning_rate": 0.0005, "loss": 1.4331891536712646, "step": 31512 }, { "epoch": 2.4842727631060306, "grad_norm": 0.46806374192237854, "learning_rate": 0.0005, "loss": 1.3174400329589844, "step": 31513 }, { "epoch": 2.4843515963736698, "grad_norm": 0.4732135832309723, "learning_rate": 0.0005, "loss": 1.4290410280227661, "step": 31514 }, { "epoch": 2.4844304296413084, "grad_norm": 0.5560804009437561, "learning_rate": 0.0005, "loss": 1.406538724899292, "step": 31515 }, { "epoch": 2.4845092629089476, "grad_norm": 0.47098129987716675, "learning_rate": 0.0005, "loss": 1.4190292358398438, "step": 31516 }, { "epoch": 2.4845880961765867, "grad_norm": 0.5618056654930115, "learning_rate": 0.0005, "loss": 1.3997743129730225, "step": 31517 }, { "epoch": 2.4846669294442254, "grad_norm": 0.4710392951965332, "learning_rate": 0.0005, "loss": 1.378965139389038, "step": 31518 }, { "epoch": 2.4847457627118645, "grad_norm": 0.6995977163314819, "learning_rate": 0.0005, "loss": 1.3437466621398926, "step": 31519 }, { "epoch": 2.4848245959795032, "grad_norm": 0.6634481549263, "learning_rate": 0.0005, "loss": 1.4473105669021606, "step": 31520 }, { "epoch": 2.4849034292471424, "grad_norm": 0.42459022998809814, "learning_rate": 0.0005, "loss": 1.4173541069030762, "step": 31521 }, { "epoch": 2.484982262514781, "grad_norm": 0.7273522019386292, "learning_rate": 0.0005, "loss": 1.2643862962722778, "step": 31522 }, { "epoch": 2.48506109578242, "grad_norm": 0.5161416530609131, "learning_rate": 0.0005, "loss": 1.4110124111175537, "step": 31523 }, { "epoch": 2.4851399290500593, "grad_norm": 0.597743570804596, "learning_rate": 0.0005, "loss": 1.469230055809021, "step": 31524 }, { "epoch": 2.485218762317698, "grad_norm": 0.49631932377815247, "learning_rate": 0.0005, "loss": 1.438084602355957, "step": 31525 }, { "epoch": 2.485297595585337, "grad_norm": 1.7307007312774658, "learning_rate": 0.0005, "loss": 1.148097038269043, "step": 31526 }, { "epoch": 2.485376428852976, "grad_norm": 0.4683987498283386, "learning_rate": 0.0005, "loss": 1.3572156429290771, "step": 31527 }, { "epoch": 2.485455262120615, "grad_norm": 0.5590559840202332, "learning_rate": 0.0005, "loss": 1.529333233833313, "step": 31528 }, { "epoch": 2.4855340953882537, "grad_norm": 0.5378466844558716, "learning_rate": 0.0005, "loss": 1.2053334712982178, "step": 31529 }, { "epoch": 2.485612928655893, "grad_norm": 0.4487302005290985, "learning_rate": 0.0005, "loss": 1.3195161819458008, "step": 31530 }, { "epoch": 2.485691761923532, "grad_norm": 0.7741996645927429, "learning_rate": 0.0005, "loss": 1.7225797176361084, "step": 31531 }, { "epoch": 2.4857705951911706, "grad_norm": 0.3904399573802948, "learning_rate": 0.0005, "loss": 1.210666537284851, "step": 31532 }, { "epoch": 2.4858494284588097, "grad_norm": 0.4431934654712677, "learning_rate": 0.0005, "loss": 1.2175090312957764, "step": 31533 }, { "epoch": 2.4859282617264484, "grad_norm": 0.5553749799728394, "learning_rate": 0.0005, "loss": 1.2598247528076172, "step": 31534 }, { "epoch": 2.4860070949940876, "grad_norm": 0.6233090758323669, "learning_rate": 0.0005, "loss": 1.4243600368499756, "step": 31535 }, { "epoch": 2.4860859282617263, "grad_norm": 0.6065593957901001, "learning_rate": 0.0005, "loss": 1.4767212867736816, "step": 31536 }, { "epoch": 2.4861647615293654, "grad_norm": 0.5509007573127747, "learning_rate": 0.0005, "loss": 1.2631157636642456, "step": 31537 }, { "epoch": 2.4862435947970045, "grad_norm": 0.4730837047100067, "learning_rate": 0.0005, "loss": 1.2877819538116455, "step": 31538 }, { "epoch": 2.486322428064643, "grad_norm": 0.5659739971160889, "learning_rate": 0.0005, "loss": 1.3505213260650635, "step": 31539 }, { "epoch": 2.4864012613322823, "grad_norm": 0.5094271898269653, "learning_rate": 0.0005, "loss": 1.2306627035140991, "step": 31540 }, { "epoch": 2.486480094599921, "grad_norm": 0.4440736174583435, "learning_rate": 0.0005, "loss": 1.258796215057373, "step": 31541 }, { "epoch": 2.48655892786756, "grad_norm": 0.5441377758979797, "learning_rate": 0.0005, "loss": 1.4095823764801025, "step": 31542 }, { "epoch": 2.4866377611351993, "grad_norm": 0.4508873522281647, "learning_rate": 0.0005, "loss": 1.3177683353424072, "step": 31543 }, { "epoch": 2.486716594402838, "grad_norm": 0.5954551100730896, "learning_rate": 0.0005, "loss": 1.4218227863311768, "step": 31544 }, { "epoch": 2.486795427670477, "grad_norm": 0.5292702317237854, "learning_rate": 0.0005, "loss": 1.4251558780670166, "step": 31545 }, { "epoch": 2.486874260938116, "grad_norm": 0.41342559456825256, "learning_rate": 0.0005, "loss": 1.2872891426086426, "step": 31546 }, { "epoch": 2.486953094205755, "grad_norm": 0.4162583649158478, "learning_rate": 0.0005, "loss": 1.2266268730163574, "step": 31547 }, { "epoch": 2.4870319274733936, "grad_norm": 0.4236064553260803, "learning_rate": 0.0005, "loss": 1.1968283653259277, "step": 31548 }, { "epoch": 2.4871107607410328, "grad_norm": 0.4835060238838196, "learning_rate": 0.0005, "loss": 1.317819356918335, "step": 31549 }, { "epoch": 2.4871895940086715, "grad_norm": 0.46483883261680603, "learning_rate": 0.0005, "loss": 1.3342046737670898, "step": 31550 }, { "epoch": 2.4872684272763106, "grad_norm": 0.5726062655448914, "learning_rate": 0.0005, "loss": 1.3569682836532593, "step": 31551 }, { "epoch": 2.4873472605439497, "grad_norm": 0.6062948107719421, "learning_rate": 0.0005, "loss": 1.4307665824890137, "step": 31552 }, { "epoch": 2.4874260938115884, "grad_norm": 0.4857022166252136, "learning_rate": 0.0005, "loss": 1.2719316482543945, "step": 31553 }, { "epoch": 2.4875049270792275, "grad_norm": 0.5739246010780334, "learning_rate": 0.0005, "loss": 1.3107273578643799, "step": 31554 }, { "epoch": 2.4875837603468662, "grad_norm": 0.5375766158103943, "learning_rate": 0.0005, "loss": 1.484178066253662, "step": 31555 }, { "epoch": 2.4876625936145054, "grad_norm": 0.47803646326065063, "learning_rate": 0.0005, "loss": 1.230139136314392, "step": 31556 }, { "epoch": 2.4877414268821445, "grad_norm": 0.6036074161529541, "learning_rate": 0.0005, "loss": 1.3456099033355713, "step": 31557 }, { "epoch": 2.487820260149783, "grad_norm": 0.5048182606697083, "learning_rate": 0.0005, "loss": 1.3940602540969849, "step": 31558 }, { "epoch": 2.4878990934174223, "grad_norm": 0.7682981491088867, "learning_rate": 0.0005, "loss": 1.4691736698150635, "step": 31559 }, { "epoch": 2.487977926685061, "grad_norm": 0.5100100040435791, "learning_rate": 0.0005, "loss": 1.6059985160827637, "step": 31560 }, { "epoch": 2.4880567599527, "grad_norm": 0.46517008543014526, "learning_rate": 0.0005, "loss": 1.3536845445632935, "step": 31561 }, { "epoch": 2.488135593220339, "grad_norm": 0.4866262972354889, "learning_rate": 0.0005, "loss": 1.3435308933258057, "step": 31562 }, { "epoch": 2.488214426487978, "grad_norm": 0.5359547138214111, "learning_rate": 0.0005, "loss": 1.3371455669403076, "step": 31563 }, { "epoch": 2.4882932597556167, "grad_norm": 0.6067219972610474, "learning_rate": 0.0005, "loss": 1.3025505542755127, "step": 31564 }, { "epoch": 2.488372093023256, "grad_norm": 0.4966129660606384, "learning_rate": 0.0005, "loss": 1.468397855758667, "step": 31565 }, { "epoch": 2.488450926290895, "grad_norm": 0.6080204844474792, "learning_rate": 0.0005, "loss": 1.4697483777999878, "step": 31566 }, { "epoch": 2.4885297595585336, "grad_norm": 0.5930185317993164, "learning_rate": 0.0005, "loss": 1.488455891609192, "step": 31567 }, { "epoch": 2.4886085928261727, "grad_norm": 0.3602691888809204, "learning_rate": 0.0005, "loss": 1.0892763137817383, "step": 31568 }, { "epoch": 2.4886874260938114, "grad_norm": 0.43574872612953186, "learning_rate": 0.0005, "loss": 1.2823474407196045, "step": 31569 }, { "epoch": 2.4887662593614506, "grad_norm": 0.39164629578590393, "learning_rate": 0.0005, "loss": 1.1713032722473145, "step": 31570 }, { "epoch": 2.4888450926290897, "grad_norm": 0.5142799019813538, "learning_rate": 0.0005, "loss": 1.2795004844665527, "step": 31571 }, { "epoch": 2.4889239258967284, "grad_norm": 0.36336004734039307, "learning_rate": 0.0005, "loss": 1.239389419555664, "step": 31572 }, { "epoch": 2.4890027591643675, "grad_norm": 0.5112895965576172, "learning_rate": 0.0005, "loss": 1.2699187994003296, "step": 31573 }, { "epoch": 2.489081592432006, "grad_norm": 0.469439834356308, "learning_rate": 0.0005, "loss": 1.1542800664901733, "step": 31574 }, { "epoch": 2.4891604256996454, "grad_norm": 0.4436454474925995, "learning_rate": 0.0005, "loss": 1.3458359241485596, "step": 31575 }, { "epoch": 2.489239258967284, "grad_norm": 0.46886903047561646, "learning_rate": 0.0005, "loss": 1.248157024383545, "step": 31576 }, { "epoch": 2.489318092234923, "grad_norm": 0.5890640616416931, "learning_rate": 0.0005, "loss": 1.5292247533798218, "step": 31577 }, { "epoch": 2.489396925502562, "grad_norm": 0.5259605050086975, "learning_rate": 0.0005, "loss": 1.2835794687271118, "step": 31578 }, { "epoch": 2.489475758770201, "grad_norm": 0.46392470598220825, "learning_rate": 0.0005, "loss": 1.2518184185028076, "step": 31579 }, { "epoch": 2.48955459203784, "grad_norm": 0.53367680311203, "learning_rate": 0.0005, "loss": 1.2786685228347778, "step": 31580 }, { "epoch": 2.489633425305479, "grad_norm": 0.566243052482605, "learning_rate": 0.0005, "loss": 1.6688497066497803, "step": 31581 }, { "epoch": 2.489712258573118, "grad_norm": 0.5176858901977539, "learning_rate": 0.0005, "loss": 1.2822120189666748, "step": 31582 }, { "epoch": 2.4897910918407566, "grad_norm": 0.5290963053703308, "learning_rate": 0.0005, "loss": 1.3660800457000732, "step": 31583 }, { "epoch": 2.4898699251083958, "grad_norm": 0.44723719358444214, "learning_rate": 0.0005, "loss": 1.223854422569275, "step": 31584 }, { "epoch": 2.489948758376035, "grad_norm": 0.6126503348350525, "learning_rate": 0.0005, "loss": 1.3144581317901611, "step": 31585 }, { "epoch": 2.4900275916436736, "grad_norm": 0.5101999044418335, "learning_rate": 0.0005, "loss": 1.2151894569396973, "step": 31586 }, { "epoch": 2.4901064249113127, "grad_norm": 0.5210403800010681, "learning_rate": 0.0005, "loss": 1.2469241619110107, "step": 31587 }, { "epoch": 2.4901852581789514, "grad_norm": 0.6602094769477844, "learning_rate": 0.0005, "loss": 1.5187379121780396, "step": 31588 }, { "epoch": 2.4902640914465906, "grad_norm": 0.5228450894355774, "learning_rate": 0.0005, "loss": 1.1871205568313599, "step": 31589 }, { "epoch": 2.4903429247142292, "grad_norm": 0.41410794854164124, "learning_rate": 0.0005, "loss": 1.1970388889312744, "step": 31590 }, { "epoch": 2.4904217579818684, "grad_norm": 0.5535680055618286, "learning_rate": 0.0005, "loss": 1.3411004543304443, "step": 31591 }, { "epoch": 2.490500591249507, "grad_norm": 0.5809449553489685, "learning_rate": 0.0005, "loss": 1.3366830348968506, "step": 31592 }, { "epoch": 2.490579424517146, "grad_norm": 0.4176512360572815, "learning_rate": 0.0005, "loss": 1.3914923667907715, "step": 31593 }, { "epoch": 2.4906582577847853, "grad_norm": 0.5419444441795349, "learning_rate": 0.0005, "loss": 1.4214060306549072, "step": 31594 }, { "epoch": 2.490737091052424, "grad_norm": 0.6176844239234924, "learning_rate": 0.0005, "loss": 1.215635895729065, "step": 31595 }, { "epoch": 2.490815924320063, "grad_norm": 0.4920656979084015, "learning_rate": 0.0005, "loss": 1.2024643421173096, "step": 31596 }, { "epoch": 2.490894757587702, "grad_norm": 0.6991656422615051, "learning_rate": 0.0005, "loss": 1.6180012226104736, "step": 31597 }, { "epoch": 2.490973590855341, "grad_norm": 0.5926855802536011, "learning_rate": 0.0005, "loss": 1.0981721878051758, "step": 31598 }, { "epoch": 2.49105242412298, "grad_norm": 0.39514097571372986, "learning_rate": 0.0005, "loss": 1.1143162250518799, "step": 31599 }, { "epoch": 2.491131257390619, "grad_norm": 0.4803113639354706, "learning_rate": 0.0005, "loss": 1.332023024559021, "step": 31600 }, { "epoch": 2.491210090658258, "grad_norm": 0.5029434561729431, "learning_rate": 0.0005, "loss": 1.2449328899383545, "step": 31601 }, { "epoch": 2.4912889239258966, "grad_norm": 0.5110099911689758, "learning_rate": 0.0005, "loss": 1.2028272151947021, "step": 31602 }, { "epoch": 2.4913677571935358, "grad_norm": 0.453792005777359, "learning_rate": 0.0005, "loss": 1.1628541946411133, "step": 31603 }, { "epoch": 2.4914465904611744, "grad_norm": 0.6507580280303955, "learning_rate": 0.0005, "loss": 1.5377916097640991, "step": 31604 }, { "epoch": 2.4915254237288136, "grad_norm": 0.4581383466720581, "learning_rate": 0.0005, "loss": 1.2680237293243408, "step": 31605 }, { "epoch": 2.4916042569964523, "grad_norm": 0.4453122317790985, "learning_rate": 0.0005, "loss": 1.219933032989502, "step": 31606 }, { "epoch": 2.4916830902640914, "grad_norm": 0.44532355666160583, "learning_rate": 0.0005, "loss": 1.1893794536590576, "step": 31607 }, { "epoch": 2.4917619235317305, "grad_norm": 0.5048041343688965, "learning_rate": 0.0005, "loss": 1.1713744401931763, "step": 31608 }, { "epoch": 2.4918407567993692, "grad_norm": 0.48741716146469116, "learning_rate": 0.0005, "loss": 1.231550693511963, "step": 31609 }, { "epoch": 2.4919195900670084, "grad_norm": 0.43842700123786926, "learning_rate": 0.0005, "loss": 1.2683145999908447, "step": 31610 }, { "epoch": 2.491998423334647, "grad_norm": 0.6575219631195068, "learning_rate": 0.0005, "loss": 1.3864281177520752, "step": 31611 }, { "epoch": 2.492077256602286, "grad_norm": 0.5415660738945007, "learning_rate": 0.0005, "loss": 1.440861701965332, "step": 31612 }, { "epoch": 2.4921560898699253, "grad_norm": 0.510513186454773, "learning_rate": 0.0005, "loss": 1.273414134979248, "step": 31613 }, { "epoch": 2.492234923137564, "grad_norm": 0.4066790044307709, "learning_rate": 0.0005, "loss": 1.231701135635376, "step": 31614 }, { "epoch": 2.492313756405203, "grad_norm": 0.4471708834171295, "learning_rate": 0.0005, "loss": 1.2127997875213623, "step": 31615 }, { "epoch": 2.492392589672842, "grad_norm": 0.573276162147522, "learning_rate": 0.0005, "loss": 1.227584958076477, "step": 31616 }, { "epoch": 2.492471422940481, "grad_norm": 0.435668408870697, "learning_rate": 0.0005, "loss": 1.2799381017684937, "step": 31617 }, { "epoch": 2.4925502562081197, "grad_norm": 0.6822594404220581, "learning_rate": 0.0005, "loss": 1.6564826965332031, "step": 31618 }, { "epoch": 2.492629089475759, "grad_norm": 0.5038661956787109, "learning_rate": 0.0005, "loss": 1.4200044870376587, "step": 31619 }, { "epoch": 2.4927079227433975, "grad_norm": 0.43740808963775635, "learning_rate": 0.0005, "loss": 1.1850204467773438, "step": 31620 }, { "epoch": 2.4927867560110366, "grad_norm": 0.4497431218624115, "learning_rate": 0.0005, "loss": 1.1775583028793335, "step": 31621 }, { "epoch": 2.4928655892786757, "grad_norm": 0.6996749639511108, "learning_rate": 0.0005, "loss": 1.485459804534912, "step": 31622 }, { "epoch": 2.4929444225463144, "grad_norm": 0.4506642520427704, "learning_rate": 0.0005, "loss": 1.2674341201782227, "step": 31623 }, { "epoch": 2.4930232558139536, "grad_norm": 0.5577198266983032, "learning_rate": 0.0005, "loss": 1.325955867767334, "step": 31624 }, { "epoch": 2.4931020890815923, "grad_norm": 0.5338730216026306, "learning_rate": 0.0005, "loss": 1.590732455253601, "step": 31625 }, { "epoch": 2.4931809223492314, "grad_norm": 0.8138726949691772, "learning_rate": 0.0005, "loss": 1.4445726871490479, "step": 31626 }, { "epoch": 2.4932597556168705, "grad_norm": 0.6157236695289612, "learning_rate": 0.0005, "loss": 1.366242527961731, "step": 31627 }, { "epoch": 2.493338588884509, "grad_norm": 0.5010959506034851, "learning_rate": 0.0005, "loss": 1.2613399028778076, "step": 31628 }, { "epoch": 2.4934174221521483, "grad_norm": 0.6053504943847656, "learning_rate": 0.0005, "loss": 1.5030183792114258, "step": 31629 }, { "epoch": 2.493496255419787, "grad_norm": 0.4485117197036743, "learning_rate": 0.0005, "loss": 1.3440017700195312, "step": 31630 }, { "epoch": 2.493575088687426, "grad_norm": 1.4998406171798706, "learning_rate": 0.0005, "loss": 1.3093693256378174, "step": 31631 }, { "epoch": 2.493653921955065, "grad_norm": 0.5558107495307922, "learning_rate": 0.0005, "loss": 1.4042489528656006, "step": 31632 }, { "epoch": 2.493732755222704, "grad_norm": 0.5173988342285156, "learning_rate": 0.0005, "loss": 1.1274902820587158, "step": 31633 }, { "epoch": 2.4938115884903427, "grad_norm": 0.4704267680644989, "learning_rate": 0.0005, "loss": 1.1756627559661865, "step": 31634 }, { "epoch": 2.493890421757982, "grad_norm": 0.526669442653656, "learning_rate": 0.0005, "loss": 1.1041642427444458, "step": 31635 }, { "epoch": 2.493969255025621, "grad_norm": 0.49065113067626953, "learning_rate": 0.0005, "loss": 1.2997604608535767, "step": 31636 }, { "epoch": 2.4940480882932596, "grad_norm": 0.5639541149139404, "learning_rate": 0.0005, "loss": 1.3479092121124268, "step": 31637 }, { "epoch": 2.4941269215608988, "grad_norm": 0.41448816657066345, "learning_rate": 0.0005, "loss": 1.226820707321167, "step": 31638 }, { "epoch": 2.4942057548285375, "grad_norm": 0.5586339831352234, "learning_rate": 0.0005, "loss": 1.4073351621627808, "step": 31639 }, { "epoch": 2.4942845880961766, "grad_norm": 0.4729987680912018, "learning_rate": 0.0005, "loss": 1.2413932085037231, "step": 31640 }, { "epoch": 2.4943634213638157, "grad_norm": 0.42799633741378784, "learning_rate": 0.0005, "loss": 1.2919642925262451, "step": 31641 }, { "epoch": 2.4944422546314544, "grad_norm": 0.5028735399246216, "learning_rate": 0.0005, "loss": 1.2064476013183594, "step": 31642 }, { "epoch": 2.4945210878990935, "grad_norm": 0.46922868490219116, "learning_rate": 0.0005, "loss": 1.210785150527954, "step": 31643 }, { "epoch": 2.4945999211667322, "grad_norm": 0.49553561210632324, "learning_rate": 0.0005, "loss": 1.3528029918670654, "step": 31644 }, { "epoch": 2.4946787544343714, "grad_norm": 0.632127583026886, "learning_rate": 0.0005, "loss": 1.1696429252624512, "step": 31645 }, { "epoch": 2.49475758770201, "grad_norm": 0.4728676676750183, "learning_rate": 0.0005, "loss": 1.3638478517532349, "step": 31646 }, { "epoch": 2.494836420969649, "grad_norm": 0.47278401255607605, "learning_rate": 0.0005, "loss": 1.2007064819335938, "step": 31647 }, { "epoch": 2.4949152542372883, "grad_norm": 0.6020183563232422, "learning_rate": 0.0005, "loss": 1.3288514614105225, "step": 31648 }, { "epoch": 2.494994087504927, "grad_norm": 0.4146266281604767, "learning_rate": 0.0005, "loss": 1.1576591730117798, "step": 31649 }, { "epoch": 2.495072920772566, "grad_norm": 0.5208911299705505, "learning_rate": 0.0005, "loss": 1.332526445388794, "step": 31650 }, { "epoch": 2.495151754040205, "grad_norm": 0.5188377499580383, "learning_rate": 0.0005, "loss": 1.403959035873413, "step": 31651 }, { "epoch": 2.495230587307844, "grad_norm": 0.6648988127708435, "learning_rate": 0.0005, "loss": 1.510067343711853, "step": 31652 }, { "epoch": 2.4953094205754827, "grad_norm": 0.45793795585632324, "learning_rate": 0.0005, "loss": 1.3143274784088135, "step": 31653 }, { "epoch": 2.495388253843122, "grad_norm": 0.48992693424224854, "learning_rate": 0.0005, "loss": 1.3243937492370605, "step": 31654 }, { "epoch": 2.495467087110761, "grad_norm": 0.542600154876709, "learning_rate": 0.0005, "loss": 1.2292159795761108, "step": 31655 }, { "epoch": 2.4955459203783996, "grad_norm": 0.43620264530181885, "learning_rate": 0.0005, "loss": 1.324826955795288, "step": 31656 }, { "epoch": 2.4956247536460388, "grad_norm": 0.44079259037971497, "learning_rate": 0.0005, "loss": 1.4963780641555786, "step": 31657 }, { "epoch": 2.4957035869136774, "grad_norm": 0.5196875929832458, "learning_rate": 0.0005, "loss": 1.301250696182251, "step": 31658 }, { "epoch": 2.4957824201813166, "grad_norm": 0.47120925784111023, "learning_rate": 0.0005, "loss": 1.2461276054382324, "step": 31659 }, { "epoch": 2.4958612534489557, "grad_norm": 1.1271272897720337, "learning_rate": 0.0005, "loss": 1.3779761791229248, "step": 31660 }, { "epoch": 2.4959400867165944, "grad_norm": 0.5361678600311279, "learning_rate": 0.0005, "loss": 1.3421046733856201, "step": 31661 }, { "epoch": 2.4960189199842335, "grad_norm": 0.49529728293418884, "learning_rate": 0.0005, "loss": 1.3122813701629639, "step": 31662 }, { "epoch": 2.496097753251872, "grad_norm": 0.4346206784248352, "learning_rate": 0.0005, "loss": 1.1719876527786255, "step": 31663 }, { "epoch": 2.4961765865195114, "grad_norm": 0.516867995262146, "learning_rate": 0.0005, "loss": 1.4056285619735718, "step": 31664 }, { "epoch": 2.49625541978715, "grad_norm": 0.5397174954414368, "learning_rate": 0.0005, "loss": 1.3812496662139893, "step": 31665 }, { "epoch": 2.496334253054789, "grad_norm": 0.5453313589096069, "learning_rate": 0.0005, "loss": 1.2468798160552979, "step": 31666 }, { "epoch": 2.496413086322428, "grad_norm": 0.5658764839172363, "learning_rate": 0.0005, "loss": 1.4398248195648193, "step": 31667 }, { "epoch": 2.496491919590067, "grad_norm": 0.5343732237815857, "learning_rate": 0.0005, "loss": 1.2707595825195312, "step": 31668 }, { "epoch": 2.496570752857706, "grad_norm": 0.5613932013511658, "learning_rate": 0.0005, "loss": 1.3046131134033203, "step": 31669 }, { "epoch": 2.496649586125345, "grad_norm": 0.6231865286827087, "learning_rate": 0.0005, "loss": 1.4864847660064697, "step": 31670 }, { "epoch": 2.496728419392984, "grad_norm": 0.37840551137924194, "learning_rate": 0.0005, "loss": 1.1052712202072144, "step": 31671 }, { "epoch": 2.4968072526606226, "grad_norm": 0.4856078028678894, "learning_rate": 0.0005, "loss": 1.0910084247589111, "step": 31672 }, { "epoch": 2.496886085928262, "grad_norm": 0.5342099666595459, "learning_rate": 0.0005, "loss": 1.1712324619293213, "step": 31673 }, { "epoch": 2.496964919195901, "grad_norm": 0.5624063014984131, "learning_rate": 0.0005, "loss": 1.3658839464187622, "step": 31674 }, { "epoch": 2.4970437524635396, "grad_norm": 0.5862855911254883, "learning_rate": 0.0005, "loss": 1.4491705894470215, "step": 31675 }, { "epoch": 2.4971225857311787, "grad_norm": 0.5455082654953003, "learning_rate": 0.0005, "loss": 1.2670923471450806, "step": 31676 }, { "epoch": 2.4972014189988174, "grad_norm": 0.38195616006851196, "learning_rate": 0.0005, "loss": 1.138654351234436, "step": 31677 }, { "epoch": 2.4972802522664566, "grad_norm": 0.5399524569511414, "learning_rate": 0.0005, "loss": 1.4192785024642944, "step": 31678 }, { "epoch": 2.4973590855340952, "grad_norm": 0.5236343741416931, "learning_rate": 0.0005, "loss": 1.3278589248657227, "step": 31679 }, { "epoch": 2.4974379188017344, "grad_norm": 0.4855891466140747, "learning_rate": 0.0005, "loss": 1.4244968891143799, "step": 31680 }, { "epoch": 2.497516752069373, "grad_norm": 0.5371622443199158, "learning_rate": 0.0005, "loss": 1.4180155992507935, "step": 31681 }, { "epoch": 2.497595585337012, "grad_norm": 0.4848040044307709, "learning_rate": 0.0005, "loss": 1.2105159759521484, "step": 31682 }, { "epoch": 2.4976744186046513, "grad_norm": 0.9793622493743896, "learning_rate": 0.0005, "loss": 1.4296059608459473, "step": 31683 }, { "epoch": 2.49775325187229, "grad_norm": 0.5198027491569519, "learning_rate": 0.0005, "loss": 1.3460452556610107, "step": 31684 }, { "epoch": 2.497832085139929, "grad_norm": 0.5303373336791992, "learning_rate": 0.0005, "loss": 1.3054773807525635, "step": 31685 }, { "epoch": 2.497910918407568, "grad_norm": 0.5690463185310364, "learning_rate": 0.0005, "loss": 1.482013463973999, "step": 31686 }, { "epoch": 2.497989751675207, "grad_norm": 0.4793810248374939, "learning_rate": 0.0005, "loss": 1.2753467559814453, "step": 31687 }, { "epoch": 2.498068584942846, "grad_norm": 0.5287753343582153, "learning_rate": 0.0005, "loss": 1.4490970373153687, "step": 31688 }, { "epoch": 2.498147418210485, "grad_norm": 0.48822253942489624, "learning_rate": 0.0005, "loss": 1.164886236190796, "step": 31689 }, { "epoch": 2.498226251478124, "grad_norm": 0.49522045254707336, "learning_rate": 0.0005, "loss": 1.3382363319396973, "step": 31690 }, { "epoch": 2.4983050847457626, "grad_norm": 0.5115475654602051, "learning_rate": 0.0005, "loss": 1.2484097480773926, "step": 31691 }, { "epoch": 2.4983839180134018, "grad_norm": 0.5589709281921387, "learning_rate": 0.0005, "loss": 1.348520040512085, "step": 31692 }, { "epoch": 2.4984627512810405, "grad_norm": 0.5533628463745117, "learning_rate": 0.0005, "loss": 1.3972206115722656, "step": 31693 }, { "epoch": 2.4985415845486796, "grad_norm": 0.5514285564422607, "learning_rate": 0.0005, "loss": 1.5749404430389404, "step": 31694 }, { "epoch": 2.4986204178163183, "grad_norm": 0.5109338164329529, "learning_rate": 0.0005, "loss": 1.4471936225891113, "step": 31695 }, { "epoch": 2.4986992510839574, "grad_norm": 0.5385640859603882, "learning_rate": 0.0005, "loss": 1.3287869691848755, "step": 31696 }, { "epoch": 2.4987780843515965, "grad_norm": 0.47645747661590576, "learning_rate": 0.0005, "loss": 1.3334519863128662, "step": 31697 }, { "epoch": 2.4988569176192352, "grad_norm": 0.4733853340148926, "learning_rate": 0.0005, "loss": 1.3083974123001099, "step": 31698 }, { "epoch": 2.4989357508868744, "grad_norm": 0.5402556657791138, "learning_rate": 0.0005, "loss": 1.2590115070343018, "step": 31699 }, { "epoch": 2.499014584154513, "grad_norm": 0.5063702464103699, "learning_rate": 0.0005, "loss": 1.3869433403015137, "step": 31700 }, { "epoch": 2.499093417422152, "grad_norm": 0.5280547738075256, "learning_rate": 0.0005, "loss": 1.4453051090240479, "step": 31701 }, { "epoch": 2.4991722506897913, "grad_norm": 0.5500586628913879, "learning_rate": 0.0005, "loss": 1.3517191410064697, "step": 31702 }, { "epoch": 2.49925108395743, "grad_norm": 0.505498468875885, "learning_rate": 0.0005, "loss": 1.4455571174621582, "step": 31703 }, { "epoch": 2.499329917225069, "grad_norm": 0.5047531723976135, "learning_rate": 0.0005, "loss": 1.4011857509613037, "step": 31704 }, { "epoch": 2.499408750492708, "grad_norm": 0.46128207445144653, "learning_rate": 0.0005, "loss": 1.3427228927612305, "step": 31705 }, { "epoch": 2.499487583760347, "grad_norm": 0.5127113461494446, "learning_rate": 0.0005, "loss": 1.2893786430358887, "step": 31706 }, { "epoch": 2.4995664170279857, "grad_norm": 0.4772054851055145, "learning_rate": 0.0005, "loss": 1.5603082180023193, "step": 31707 }, { "epoch": 2.499645250295625, "grad_norm": 0.5022121071815491, "learning_rate": 0.0005, "loss": 1.4553241729736328, "step": 31708 }, { "epoch": 2.4997240835632635, "grad_norm": 0.47568726539611816, "learning_rate": 0.0005, "loss": 1.305498719215393, "step": 31709 }, { "epoch": 2.4998029168309026, "grad_norm": 0.5169969201087952, "learning_rate": 0.0005, "loss": 1.5920051336288452, "step": 31710 }, { "epoch": 2.4998817500985417, "grad_norm": 0.426800400018692, "learning_rate": 0.0005, "loss": 1.2068252563476562, "step": 31711 }, { "epoch": 2.4999605833661804, "grad_norm": 0.47843074798583984, "learning_rate": 0.0005, "loss": 1.119335412979126, "step": 31712 }, { "epoch": 2.5000394166338196, "grad_norm": 0.8102190494537354, "learning_rate": 0.0005, "loss": 1.413417100906372, "step": 31713 }, { "epoch": 2.5001182499014583, "grad_norm": 0.46402111649513245, "learning_rate": 0.0005, "loss": 1.276092290878296, "step": 31714 }, { "epoch": 2.5001970831690974, "grad_norm": 0.4564312696456909, "learning_rate": 0.0005, "loss": 1.3864483833312988, "step": 31715 }, { "epoch": 2.5002759164367365, "grad_norm": 0.49634626507759094, "learning_rate": 0.0005, "loss": 1.3071767091751099, "step": 31716 }, { "epoch": 2.500354749704375, "grad_norm": 0.5215936899185181, "learning_rate": 0.0005, "loss": 1.3598203659057617, "step": 31717 }, { "epoch": 2.5004335829720143, "grad_norm": 0.5381278991699219, "learning_rate": 0.0005, "loss": 1.3544062376022339, "step": 31718 }, { "epoch": 2.500512416239653, "grad_norm": 0.4848356246948242, "learning_rate": 0.0005, "loss": 1.5722837448120117, "step": 31719 }, { "epoch": 2.500591249507292, "grad_norm": 0.5089934468269348, "learning_rate": 0.0005, "loss": 1.3160284757614136, "step": 31720 }, { "epoch": 2.500670082774931, "grad_norm": 0.5239740610122681, "learning_rate": 0.0005, "loss": 1.322840929031372, "step": 31721 }, { "epoch": 2.50074891604257, "grad_norm": 0.4804224967956543, "learning_rate": 0.0005, "loss": 1.4582579135894775, "step": 31722 }, { "epoch": 2.5008277493102087, "grad_norm": 0.4421888291835785, "learning_rate": 0.0005, "loss": 1.3739821910858154, "step": 31723 }, { "epoch": 2.500906582577848, "grad_norm": 0.5589663982391357, "learning_rate": 0.0005, "loss": 1.3966472148895264, "step": 31724 }, { "epoch": 2.500985415845487, "grad_norm": 0.6919220685958862, "learning_rate": 0.0005, "loss": 1.3125861883163452, "step": 31725 }, { "epoch": 2.5010642491131256, "grad_norm": 0.5348147749900818, "learning_rate": 0.0005, "loss": 1.3725001811981201, "step": 31726 }, { "epoch": 2.5011430823807648, "grad_norm": 0.6135135293006897, "learning_rate": 0.0005, "loss": 1.5246789455413818, "step": 31727 }, { "epoch": 2.5012219156484035, "grad_norm": 0.6065400838851929, "learning_rate": 0.0005, "loss": 1.389746904373169, "step": 31728 }, { "epoch": 2.5013007489160426, "grad_norm": 0.4870759844779968, "learning_rate": 0.0005, "loss": 1.3174093961715698, "step": 31729 }, { "epoch": 2.5013795821836817, "grad_norm": 0.45566412806510925, "learning_rate": 0.0005, "loss": 1.2945423126220703, "step": 31730 }, { "epoch": 2.5014584154513204, "grad_norm": 0.4252101182937622, "learning_rate": 0.0005, "loss": 1.3440364599227905, "step": 31731 }, { "epoch": 2.5015372487189595, "grad_norm": 0.5442041754722595, "learning_rate": 0.0005, "loss": 1.6875827312469482, "step": 31732 }, { "epoch": 2.5016160819865982, "grad_norm": 0.5361164212226868, "learning_rate": 0.0005, "loss": 1.428688406944275, "step": 31733 }, { "epoch": 2.5016949152542374, "grad_norm": 0.5825225114822388, "learning_rate": 0.0005, "loss": 1.436211347579956, "step": 31734 }, { "epoch": 2.5017737485218765, "grad_norm": 0.6354440450668335, "learning_rate": 0.0005, "loss": 1.2726143598556519, "step": 31735 }, { "epoch": 2.501852581789515, "grad_norm": 0.5332158803939819, "learning_rate": 0.0005, "loss": 1.278162956237793, "step": 31736 }, { "epoch": 2.501931415057154, "grad_norm": 0.5487943291664124, "learning_rate": 0.0005, "loss": 1.3862802982330322, "step": 31737 }, { "epoch": 2.502010248324793, "grad_norm": 0.5139238238334656, "learning_rate": 0.0005, "loss": 1.442972183227539, "step": 31738 }, { "epoch": 2.502089081592432, "grad_norm": 0.5221526026725769, "learning_rate": 0.0005, "loss": 1.3722655773162842, "step": 31739 }, { "epoch": 2.502167914860071, "grad_norm": 0.4680907726287842, "learning_rate": 0.0005, "loss": 1.393603801727295, "step": 31740 }, { "epoch": 2.50224674812771, "grad_norm": 0.5685150027275085, "learning_rate": 0.0005, "loss": 1.254058599472046, "step": 31741 }, { "epoch": 2.5023255813953487, "grad_norm": 0.5464416146278381, "learning_rate": 0.0005, "loss": 1.3419955968856812, "step": 31742 }, { "epoch": 2.502404414662988, "grad_norm": 0.49742913246154785, "learning_rate": 0.0005, "loss": 1.2423679828643799, "step": 31743 }, { "epoch": 2.502483247930627, "grad_norm": 0.5190958976745605, "learning_rate": 0.0005, "loss": 1.3520019054412842, "step": 31744 }, { "epoch": 2.5025620811982656, "grad_norm": 0.6712267398834229, "learning_rate": 0.0005, "loss": 1.623936414718628, "step": 31745 }, { "epoch": 2.5026409144659048, "grad_norm": 0.44776153564453125, "learning_rate": 0.0005, "loss": 1.1627715826034546, "step": 31746 }, { "epoch": 2.5027197477335434, "grad_norm": 0.41925573348999023, "learning_rate": 0.0005, "loss": 1.2374404668807983, "step": 31747 }, { "epoch": 2.5027985810011826, "grad_norm": 0.39284607768058777, "learning_rate": 0.0005, "loss": 1.2792553901672363, "step": 31748 }, { "epoch": 2.5028774142688217, "grad_norm": 0.44381457567214966, "learning_rate": 0.0005, "loss": 1.0703070163726807, "step": 31749 }, { "epoch": 2.5029562475364604, "grad_norm": 0.5070129036903381, "learning_rate": 0.0005, "loss": 1.350529432296753, "step": 31750 }, { "epoch": 2.503035080804099, "grad_norm": 0.5466955304145813, "learning_rate": 0.0005, "loss": 1.3984417915344238, "step": 31751 }, { "epoch": 2.503113914071738, "grad_norm": 0.5076603293418884, "learning_rate": 0.0005, "loss": 1.1604708433151245, "step": 31752 }, { "epoch": 2.5031927473393774, "grad_norm": 0.479042112827301, "learning_rate": 0.0005, "loss": 1.3705697059631348, "step": 31753 }, { "epoch": 2.503271580607016, "grad_norm": 0.5240731835365295, "learning_rate": 0.0005, "loss": 1.3246519565582275, "step": 31754 }, { "epoch": 2.503350413874655, "grad_norm": 0.4769611656665802, "learning_rate": 0.0005, "loss": 1.2648746967315674, "step": 31755 }, { "epoch": 2.503429247142294, "grad_norm": 0.5088683366775513, "learning_rate": 0.0005, "loss": 1.4371603727340698, "step": 31756 }, { "epoch": 2.503508080409933, "grad_norm": 0.4941759705543518, "learning_rate": 0.0005, "loss": 1.5286386013031006, "step": 31757 }, { "epoch": 2.503586913677572, "grad_norm": 0.5637709498405457, "learning_rate": 0.0005, "loss": 1.4285794496536255, "step": 31758 }, { "epoch": 2.503665746945211, "grad_norm": 0.4914212226867676, "learning_rate": 0.0005, "loss": 1.3909809589385986, "step": 31759 }, { "epoch": 2.50374458021285, "grad_norm": 0.47828736901283264, "learning_rate": 0.0005, "loss": 1.3047386407852173, "step": 31760 }, { "epoch": 2.5038234134804886, "grad_norm": 0.4591734707355499, "learning_rate": 0.0005, "loss": 1.2861284017562866, "step": 31761 }, { "epoch": 2.503902246748128, "grad_norm": 0.47224879264831543, "learning_rate": 0.0005, "loss": 1.232446312904358, "step": 31762 }, { "epoch": 2.503981080015767, "grad_norm": 0.47081464529037476, "learning_rate": 0.0005, "loss": 1.1999340057373047, "step": 31763 }, { "epoch": 2.5040599132834056, "grad_norm": 0.5332292318344116, "learning_rate": 0.0005, "loss": 1.3808677196502686, "step": 31764 }, { "epoch": 2.5041387465510443, "grad_norm": 0.4389898180961609, "learning_rate": 0.0005, "loss": 1.1700811386108398, "step": 31765 }, { "epoch": 2.5042175798186834, "grad_norm": 0.5272949934005737, "learning_rate": 0.0005, "loss": 1.4483811855316162, "step": 31766 }, { "epoch": 2.5042964130863226, "grad_norm": 0.5424938797950745, "learning_rate": 0.0005, "loss": 1.3976290225982666, "step": 31767 }, { "epoch": 2.5043752463539612, "grad_norm": 0.6364608407020569, "learning_rate": 0.0005, "loss": 1.5974452495574951, "step": 31768 }, { "epoch": 2.5044540796216004, "grad_norm": 0.5343613624572754, "learning_rate": 0.0005, "loss": 1.3004417419433594, "step": 31769 }, { "epoch": 2.504532912889239, "grad_norm": 0.5926498770713806, "learning_rate": 0.0005, "loss": 1.3420989513397217, "step": 31770 }, { "epoch": 2.504611746156878, "grad_norm": 0.44294044375419617, "learning_rate": 0.0005, "loss": 1.3108413219451904, "step": 31771 }, { "epoch": 2.5046905794245173, "grad_norm": 0.5534746646881104, "learning_rate": 0.0005, "loss": 1.3495285511016846, "step": 31772 }, { "epoch": 2.504769412692156, "grad_norm": 0.5163394212722778, "learning_rate": 0.0005, "loss": 1.406278371810913, "step": 31773 }, { "epoch": 2.504848245959795, "grad_norm": 0.5731506943702698, "learning_rate": 0.0005, "loss": 1.4449342489242554, "step": 31774 }, { "epoch": 2.504927079227434, "grad_norm": 0.5989260673522949, "learning_rate": 0.0005, "loss": 1.402461290359497, "step": 31775 }, { "epoch": 2.505005912495073, "grad_norm": 0.5682874917984009, "learning_rate": 0.0005, "loss": 1.4059123992919922, "step": 31776 }, { "epoch": 2.505084745762712, "grad_norm": 0.5285612940788269, "learning_rate": 0.0005, "loss": 1.3273186683654785, "step": 31777 }, { "epoch": 2.505163579030351, "grad_norm": 0.4801953136920929, "learning_rate": 0.0005, "loss": 1.298431158065796, "step": 31778 }, { "epoch": 2.5052424122979895, "grad_norm": 0.5605295896530151, "learning_rate": 0.0005, "loss": 1.4668384790420532, "step": 31779 }, { "epoch": 2.5053212455656286, "grad_norm": 0.4267539978027344, "learning_rate": 0.0005, "loss": 1.2409451007843018, "step": 31780 }, { "epoch": 2.5054000788332678, "grad_norm": 0.631769597530365, "learning_rate": 0.0005, "loss": 1.7081414461135864, "step": 31781 }, { "epoch": 2.5054789121009065, "grad_norm": 0.6290945410728455, "learning_rate": 0.0005, "loss": 1.336489200592041, "step": 31782 }, { "epoch": 2.5055577453685456, "grad_norm": 0.5900118350982666, "learning_rate": 0.0005, "loss": 1.305479884147644, "step": 31783 }, { "epoch": 2.5056365786361843, "grad_norm": 0.7029054760932922, "learning_rate": 0.0005, "loss": 1.369390845298767, "step": 31784 }, { "epoch": 2.5057154119038234, "grad_norm": 0.6839621067047119, "learning_rate": 0.0005, "loss": 1.4073336124420166, "step": 31785 }, { "epoch": 2.5057942451714625, "grad_norm": 0.44350671768188477, "learning_rate": 0.0005, "loss": 1.263342022895813, "step": 31786 }, { "epoch": 2.5058730784391012, "grad_norm": 0.4687761068344116, "learning_rate": 0.0005, "loss": 1.2995643615722656, "step": 31787 }, { "epoch": 2.5059519117067404, "grad_norm": 0.4458305537700653, "learning_rate": 0.0005, "loss": 1.400984525680542, "step": 31788 }, { "epoch": 2.506030744974379, "grad_norm": 0.5178697109222412, "learning_rate": 0.0005, "loss": 1.3148815631866455, "step": 31789 }, { "epoch": 2.506109578242018, "grad_norm": 0.40929022431373596, "learning_rate": 0.0005, "loss": 1.1699326038360596, "step": 31790 }, { "epoch": 2.5061884115096573, "grad_norm": 0.615155041217804, "learning_rate": 0.0005, "loss": 1.5397433042526245, "step": 31791 }, { "epoch": 2.506267244777296, "grad_norm": 0.45499467849731445, "learning_rate": 0.0005, "loss": 1.2853126525878906, "step": 31792 }, { "epoch": 2.5063460780449347, "grad_norm": 0.543269157409668, "learning_rate": 0.0005, "loss": 1.340034008026123, "step": 31793 }, { "epoch": 2.506424911312574, "grad_norm": 0.4309289753437042, "learning_rate": 0.0005, "loss": 1.1391677856445312, "step": 31794 }, { "epoch": 2.506503744580213, "grad_norm": 0.5778321027755737, "learning_rate": 0.0005, "loss": 1.452937126159668, "step": 31795 }, { "epoch": 2.5065825778478517, "grad_norm": 0.46861979365348816, "learning_rate": 0.0005, "loss": 1.155630350112915, "step": 31796 }, { "epoch": 2.506661411115491, "grad_norm": 0.5636255145072937, "learning_rate": 0.0005, "loss": 1.1370799541473389, "step": 31797 }, { "epoch": 2.5067402443831295, "grad_norm": 0.5195988416671753, "learning_rate": 0.0005, "loss": 1.2153502702713013, "step": 31798 }, { "epoch": 2.5068190776507686, "grad_norm": 0.6837415099143982, "learning_rate": 0.0005, "loss": 1.2856770753860474, "step": 31799 }, { "epoch": 2.5068979109184077, "grad_norm": 0.5467540621757507, "learning_rate": 0.0005, "loss": 1.3546512126922607, "step": 31800 }, { "epoch": 2.5069767441860464, "grad_norm": 0.42504945397377014, "learning_rate": 0.0005, "loss": 1.158250331878662, "step": 31801 }, { "epoch": 2.5070555774536856, "grad_norm": 0.5051652193069458, "learning_rate": 0.0005, "loss": 1.364120364189148, "step": 31802 }, { "epoch": 2.5071344107213243, "grad_norm": 0.6139587759971619, "learning_rate": 0.0005, "loss": 1.264509916305542, "step": 31803 }, { "epoch": 2.5072132439889634, "grad_norm": 0.4793805480003357, "learning_rate": 0.0005, "loss": 1.382157802581787, "step": 31804 }, { "epoch": 2.5072920772566025, "grad_norm": 0.5508502721786499, "learning_rate": 0.0005, "loss": 1.3554943799972534, "step": 31805 }, { "epoch": 2.507370910524241, "grad_norm": 0.41450172662734985, "learning_rate": 0.0005, "loss": 1.154892921447754, "step": 31806 }, { "epoch": 2.50744974379188, "grad_norm": 0.4948153495788574, "learning_rate": 0.0005, "loss": 1.1792747974395752, "step": 31807 }, { "epoch": 2.507528577059519, "grad_norm": 0.48761290311813354, "learning_rate": 0.0005, "loss": 1.336571455001831, "step": 31808 }, { "epoch": 2.507607410327158, "grad_norm": 0.4427132308483124, "learning_rate": 0.0005, "loss": 1.2432076930999756, "step": 31809 }, { "epoch": 2.507686243594797, "grad_norm": 0.5554542541503906, "learning_rate": 0.0005, "loss": 1.2824325561523438, "step": 31810 }, { "epoch": 2.507765076862436, "grad_norm": 0.44058167934417725, "learning_rate": 0.0005, "loss": 1.2358657121658325, "step": 31811 }, { "epoch": 2.5078439101300747, "grad_norm": 0.5461037158966064, "learning_rate": 0.0005, "loss": 1.1453287601470947, "step": 31812 }, { "epoch": 2.507922743397714, "grad_norm": 0.4678230285644531, "learning_rate": 0.0005, "loss": 1.3071222305297852, "step": 31813 }, { "epoch": 2.508001576665353, "grad_norm": 0.5069259405136108, "learning_rate": 0.0005, "loss": 1.2256518602371216, "step": 31814 }, { "epoch": 2.5080804099329916, "grad_norm": 0.4662998616695404, "learning_rate": 0.0005, "loss": 1.2433197498321533, "step": 31815 }, { "epoch": 2.5081592432006308, "grad_norm": 0.5826437473297119, "learning_rate": 0.0005, "loss": 1.492755651473999, "step": 31816 }, { "epoch": 2.5082380764682695, "grad_norm": 0.5010484457015991, "learning_rate": 0.0005, "loss": 1.2742912769317627, "step": 31817 }, { "epoch": 2.5083169097359086, "grad_norm": 0.449333131313324, "learning_rate": 0.0005, "loss": 1.3382881879806519, "step": 31818 }, { "epoch": 2.5083957430035477, "grad_norm": 0.5085610151290894, "learning_rate": 0.0005, "loss": 1.124898910522461, "step": 31819 }, { "epoch": 2.5084745762711864, "grad_norm": 0.4418710470199585, "learning_rate": 0.0005, "loss": 1.0791501998901367, "step": 31820 }, { "epoch": 2.508553409538825, "grad_norm": 0.47581613063812256, "learning_rate": 0.0005, "loss": 1.4121659994125366, "step": 31821 }, { "epoch": 2.5086322428064642, "grad_norm": 0.5910877585411072, "learning_rate": 0.0005, "loss": 1.144978642463684, "step": 31822 }, { "epoch": 2.5087110760741034, "grad_norm": 0.41829803586006165, "learning_rate": 0.0005, "loss": 1.2514127492904663, "step": 31823 }, { "epoch": 2.508789909341742, "grad_norm": 0.4585312008857727, "learning_rate": 0.0005, "loss": 1.2434885501861572, "step": 31824 }, { "epoch": 2.508868742609381, "grad_norm": 0.47258979082107544, "learning_rate": 0.0005, "loss": 1.2807483673095703, "step": 31825 }, { "epoch": 2.50894757587702, "grad_norm": 0.42528295516967773, "learning_rate": 0.0005, "loss": 1.3558731079101562, "step": 31826 }, { "epoch": 2.509026409144659, "grad_norm": 0.5092818140983582, "learning_rate": 0.0005, "loss": 1.4983906745910645, "step": 31827 }, { "epoch": 2.509105242412298, "grad_norm": 0.4685712456703186, "learning_rate": 0.0005, "loss": 1.1497827768325806, "step": 31828 }, { "epoch": 2.509184075679937, "grad_norm": 0.49332788586616516, "learning_rate": 0.0005, "loss": 1.2977981567382812, "step": 31829 }, { "epoch": 2.509262908947576, "grad_norm": 0.4043170213699341, "learning_rate": 0.0005, "loss": 1.2238361835479736, "step": 31830 }, { "epoch": 2.5093417422152147, "grad_norm": 0.5081291198730469, "learning_rate": 0.0005, "loss": 1.2959924936294556, "step": 31831 }, { "epoch": 2.509420575482854, "grad_norm": 0.5385888814926147, "learning_rate": 0.0005, "loss": 1.2863819599151611, "step": 31832 }, { "epoch": 2.509499408750493, "grad_norm": 0.5734867453575134, "learning_rate": 0.0005, "loss": 1.483015775680542, "step": 31833 }, { "epoch": 2.5095782420181316, "grad_norm": 0.46343794465065, "learning_rate": 0.0005, "loss": 1.2915050983428955, "step": 31834 }, { "epoch": 2.5096570752857708, "grad_norm": 0.4536112844944, "learning_rate": 0.0005, "loss": 1.273097038269043, "step": 31835 }, { "epoch": 2.5097359085534094, "grad_norm": 0.528326153755188, "learning_rate": 0.0005, "loss": 1.4502711296081543, "step": 31836 }, { "epoch": 2.5098147418210486, "grad_norm": 0.7068074941635132, "learning_rate": 0.0005, "loss": 1.4769654273986816, "step": 31837 }, { "epoch": 2.5098935750886873, "grad_norm": 0.5533843636512756, "learning_rate": 0.0005, "loss": 1.3022713661193848, "step": 31838 }, { "epoch": 2.5099724083563264, "grad_norm": 0.4564233720302582, "learning_rate": 0.0005, "loss": 1.2840286493301392, "step": 31839 }, { "epoch": 2.510051241623965, "grad_norm": 0.5003173351287842, "learning_rate": 0.0005, "loss": 1.4538469314575195, "step": 31840 }, { "epoch": 2.5101300748916042, "grad_norm": 0.5504772067070007, "learning_rate": 0.0005, "loss": 1.3150641918182373, "step": 31841 }, { "epoch": 2.5102089081592434, "grad_norm": 0.5953679084777832, "learning_rate": 0.0005, "loss": 1.358593463897705, "step": 31842 }, { "epoch": 2.510287741426882, "grad_norm": 0.44950202107429504, "learning_rate": 0.0005, "loss": 1.3606758117675781, "step": 31843 }, { "epoch": 2.510366574694521, "grad_norm": 0.3709837794303894, "learning_rate": 0.0005, "loss": 1.1956368684768677, "step": 31844 }, { "epoch": 2.51044540796216, "grad_norm": 0.46470126509666443, "learning_rate": 0.0005, "loss": 1.3591477870941162, "step": 31845 }, { "epoch": 2.510524241229799, "grad_norm": 0.4600068926811218, "learning_rate": 0.0005, "loss": 1.298570156097412, "step": 31846 }, { "epoch": 2.510603074497438, "grad_norm": 0.48119473457336426, "learning_rate": 0.0005, "loss": 1.4893194437026978, "step": 31847 }, { "epoch": 2.510681907765077, "grad_norm": 0.5141507387161255, "learning_rate": 0.0005, "loss": 1.566195011138916, "step": 31848 }, { "epoch": 2.510760741032716, "grad_norm": 0.4869093596935272, "learning_rate": 0.0005, "loss": 1.3464823961257935, "step": 31849 }, { "epoch": 2.5108395743003546, "grad_norm": 0.5723012089729309, "learning_rate": 0.0005, "loss": 1.567692518234253, "step": 31850 }, { "epoch": 2.510918407567994, "grad_norm": 0.46126294136047363, "learning_rate": 0.0005, "loss": 1.1948654651641846, "step": 31851 }, { "epoch": 2.5109972408356325, "grad_norm": 0.49016091227531433, "learning_rate": 0.0005, "loss": 1.468190312385559, "step": 31852 }, { "epoch": 2.5110760741032716, "grad_norm": 0.5409206748008728, "learning_rate": 0.0005, "loss": 1.455737590789795, "step": 31853 }, { "epoch": 2.5111549073709103, "grad_norm": 0.6167699694633484, "learning_rate": 0.0005, "loss": 1.3520578145980835, "step": 31854 }, { "epoch": 2.5112337406385494, "grad_norm": 0.7364978790283203, "learning_rate": 0.0005, "loss": 1.3805832862854004, "step": 31855 }, { "epoch": 2.5113125739061886, "grad_norm": 1.1502374410629272, "learning_rate": 0.0005, "loss": 1.460676908493042, "step": 31856 }, { "epoch": 2.5113914071738273, "grad_norm": 0.6217806935310364, "learning_rate": 0.0005, "loss": 1.5078654289245605, "step": 31857 }, { "epoch": 2.5114702404414664, "grad_norm": 0.5574552416801453, "learning_rate": 0.0005, "loss": 1.300079345703125, "step": 31858 }, { "epoch": 2.511549073709105, "grad_norm": 0.49690669775009155, "learning_rate": 0.0005, "loss": 1.3038055896759033, "step": 31859 }, { "epoch": 2.511627906976744, "grad_norm": 0.4583120346069336, "learning_rate": 0.0005, "loss": 1.1100654602050781, "step": 31860 }, { "epoch": 2.5117067402443833, "grad_norm": 0.5933244824409485, "learning_rate": 0.0005, "loss": 1.5064629316329956, "step": 31861 }, { "epoch": 2.511785573512022, "grad_norm": 0.46965524554252625, "learning_rate": 0.0005, "loss": 1.2976598739624023, "step": 31862 }, { "epoch": 2.511864406779661, "grad_norm": 0.5966199040412903, "learning_rate": 0.0005, "loss": 1.3530781269073486, "step": 31863 }, { "epoch": 2.5119432400473, "grad_norm": 0.5334296226501465, "learning_rate": 0.0005, "loss": 1.2152904272079468, "step": 31864 }, { "epoch": 2.512022073314939, "grad_norm": 0.4830121397972107, "learning_rate": 0.0005, "loss": 1.2267358303070068, "step": 31865 }, { "epoch": 2.512100906582578, "grad_norm": 0.5711424350738525, "learning_rate": 0.0005, "loss": 1.3181730508804321, "step": 31866 }, { "epoch": 2.512179739850217, "grad_norm": 0.4775398075580597, "learning_rate": 0.0005, "loss": 1.2552956342697144, "step": 31867 }, { "epoch": 2.5122585731178555, "grad_norm": 0.5473701357841492, "learning_rate": 0.0005, "loss": 1.323310375213623, "step": 31868 }, { "epoch": 2.5123374063854946, "grad_norm": 0.47209903597831726, "learning_rate": 0.0005, "loss": 1.343681812286377, "step": 31869 }, { "epoch": 2.5124162396531338, "grad_norm": 0.6196010112762451, "learning_rate": 0.0005, "loss": 1.293186902999878, "step": 31870 }, { "epoch": 2.5124950729207725, "grad_norm": 0.42704033851623535, "learning_rate": 0.0005, "loss": 1.3028349876403809, "step": 31871 }, { "epoch": 2.5125739061884116, "grad_norm": 0.40719911456108093, "learning_rate": 0.0005, "loss": 1.0387585163116455, "step": 31872 }, { "epoch": 2.5126527394560503, "grad_norm": 0.5095211267471313, "learning_rate": 0.0005, "loss": 1.4856672286987305, "step": 31873 }, { "epoch": 2.5127315727236894, "grad_norm": 0.4035288989543915, "learning_rate": 0.0005, "loss": 1.1878935098648071, "step": 31874 }, { "epoch": 2.5128104059913285, "grad_norm": 0.6002760529518127, "learning_rate": 0.0005, "loss": 1.4601449966430664, "step": 31875 }, { "epoch": 2.5128892392589672, "grad_norm": 0.8489753603935242, "learning_rate": 0.0005, "loss": 1.461686611175537, "step": 31876 }, { "epoch": 2.5129680725266064, "grad_norm": 0.4310103952884674, "learning_rate": 0.0005, "loss": 1.2730650901794434, "step": 31877 }, { "epoch": 2.513046905794245, "grad_norm": 0.6927597522735596, "learning_rate": 0.0005, "loss": 1.3770262002944946, "step": 31878 }, { "epoch": 2.513125739061884, "grad_norm": 0.49830177426338196, "learning_rate": 0.0005, "loss": 1.3027572631835938, "step": 31879 }, { "epoch": 2.5132045723295233, "grad_norm": 0.5801132917404175, "learning_rate": 0.0005, "loss": 1.2626748085021973, "step": 31880 }, { "epoch": 2.513283405597162, "grad_norm": 0.5403900146484375, "learning_rate": 0.0005, "loss": 1.5056262016296387, "step": 31881 }, { "epoch": 2.5133622388648007, "grad_norm": 0.569191575050354, "learning_rate": 0.0005, "loss": 1.2943639755249023, "step": 31882 }, { "epoch": 2.51344107213244, "grad_norm": 0.6653793454170227, "learning_rate": 0.0005, "loss": 1.2709022760391235, "step": 31883 }, { "epoch": 2.513519905400079, "grad_norm": 0.4940057098865509, "learning_rate": 0.0005, "loss": 1.5083096027374268, "step": 31884 }, { "epoch": 2.5135987386677177, "grad_norm": 0.6159873008728027, "learning_rate": 0.0005, "loss": 1.288045883178711, "step": 31885 }, { "epoch": 2.513677571935357, "grad_norm": 0.37297213077545166, "learning_rate": 0.0005, "loss": 1.0347168445587158, "step": 31886 }, { "epoch": 2.5137564052029955, "grad_norm": 0.4787515103816986, "learning_rate": 0.0005, "loss": 1.1829982995986938, "step": 31887 }, { "epoch": 2.5138352384706346, "grad_norm": 0.526118278503418, "learning_rate": 0.0005, "loss": 1.2201972007751465, "step": 31888 }, { "epoch": 2.5139140717382737, "grad_norm": 0.8212949633598328, "learning_rate": 0.0005, "loss": 1.331407070159912, "step": 31889 }, { "epoch": 2.5139929050059124, "grad_norm": 0.6149592995643616, "learning_rate": 0.0005, "loss": 1.4359484910964966, "step": 31890 }, { "epoch": 2.5140717382735516, "grad_norm": 0.48624593019485474, "learning_rate": 0.0005, "loss": 1.2492926120758057, "step": 31891 }, { "epoch": 2.5141505715411903, "grad_norm": 0.50346839427948, "learning_rate": 0.0005, "loss": 1.4171252250671387, "step": 31892 }, { "epoch": 2.5142294048088294, "grad_norm": 0.5505930781364441, "learning_rate": 0.0005, "loss": 1.4969146251678467, "step": 31893 }, { "epoch": 2.5143082380764685, "grad_norm": 0.49496904015541077, "learning_rate": 0.0005, "loss": 1.193111538887024, "step": 31894 }, { "epoch": 2.514387071344107, "grad_norm": 0.545435905456543, "learning_rate": 0.0005, "loss": 1.3120408058166504, "step": 31895 }, { "epoch": 2.514465904611746, "grad_norm": 0.5572736263275146, "learning_rate": 0.0005, "loss": 1.4146006107330322, "step": 31896 }, { "epoch": 2.514544737879385, "grad_norm": 0.6097012758255005, "learning_rate": 0.0005, "loss": 1.3015660047531128, "step": 31897 }, { "epoch": 2.514623571147024, "grad_norm": 0.5869887471199036, "learning_rate": 0.0005, "loss": 1.3229176998138428, "step": 31898 }, { "epoch": 2.514702404414663, "grad_norm": 0.4425044357776642, "learning_rate": 0.0005, "loss": 1.3099360466003418, "step": 31899 }, { "epoch": 2.514781237682302, "grad_norm": 0.4712465703487396, "learning_rate": 0.0005, "loss": 1.3243777751922607, "step": 31900 }, { "epoch": 2.5148600709499407, "grad_norm": 0.5598543882369995, "learning_rate": 0.0005, "loss": 1.400916576385498, "step": 31901 }, { "epoch": 2.51493890421758, "grad_norm": 0.4899856746196747, "learning_rate": 0.0005, "loss": 1.308101773262024, "step": 31902 }, { "epoch": 2.515017737485219, "grad_norm": 0.44490906596183777, "learning_rate": 0.0005, "loss": 1.1769793033599854, "step": 31903 }, { "epoch": 2.5150965707528576, "grad_norm": 0.4818161129951477, "learning_rate": 0.0005, "loss": 1.3030107021331787, "step": 31904 }, { "epoch": 2.5151754040204968, "grad_norm": 0.4925244152545929, "learning_rate": 0.0005, "loss": 1.2121042013168335, "step": 31905 }, { "epoch": 2.5152542372881355, "grad_norm": 0.4296555519104004, "learning_rate": 0.0005, "loss": 1.1476068496704102, "step": 31906 }, { "epoch": 2.5153330705557746, "grad_norm": 0.44502827525138855, "learning_rate": 0.0005, "loss": 1.175580620765686, "step": 31907 }, { "epoch": 2.5154119038234137, "grad_norm": 0.6112645268440247, "learning_rate": 0.0005, "loss": 1.424264669418335, "step": 31908 }, { "epoch": 2.5154907370910524, "grad_norm": 0.5582069158554077, "learning_rate": 0.0005, "loss": 1.4969780445098877, "step": 31909 }, { "epoch": 2.515569570358691, "grad_norm": 0.41058680415153503, "learning_rate": 0.0005, "loss": 1.2417242527008057, "step": 31910 }, { "epoch": 2.5156484036263302, "grad_norm": 0.4839356541633606, "learning_rate": 0.0005, "loss": 1.2868345975875854, "step": 31911 }, { "epoch": 2.5157272368939694, "grad_norm": 0.4994647800922394, "learning_rate": 0.0005, "loss": 1.286043405532837, "step": 31912 }, { "epoch": 2.515806070161608, "grad_norm": 0.42707785964012146, "learning_rate": 0.0005, "loss": 1.1994643211364746, "step": 31913 }, { "epoch": 2.515884903429247, "grad_norm": 0.5765823721885681, "learning_rate": 0.0005, "loss": 1.4037097692489624, "step": 31914 }, { "epoch": 2.515963736696886, "grad_norm": 0.534541666507721, "learning_rate": 0.0005, "loss": 1.4892919063568115, "step": 31915 }, { "epoch": 2.516042569964525, "grad_norm": 0.6548802256584167, "learning_rate": 0.0005, "loss": 1.3805071115493774, "step": 31916 }, { "epoch": 2.516121403232164, "grad_norm": 0.4324514865875244, "learning_rate": 0.0005, "loss": 1.3236219882965088, "step": 31917 }, { "epoch": 2.516200236499803, "grad_norm": 0.3996571898460388, "learning_rate": 0.0005, "loss": 1.2931662797927856, "step": 31918 }, { "epoch": 2.516279069767442, "grad_norm": 0.5384787321090698, "learning_rate": 0.0005, "loss": 1.4816617965698242, "step": 31919 }, { "epoch": 2.5163579030350807, "grad_norm": 0.42014840245246887, "learning_rate": 0.0005, "loss": 1.2974790334701538, "step": 31920 }, { "epoch": 2.51643673630272, "grad_norm": 0.45131370425224304, "learning_rate": 0.0005, "loss": 1.102126121520996, "step": 31921 }, { "epoch": 2.516515569570359, "grad_norm": 0.5095865726470947, "learning_rate": 0.0005, "loss": 1.2128480672836304, "step": 31922 }, { "epoch": 2.5165944028379976, "grad_norm": 0.43147388100624084, "learning_rate": 0.0005, "loss": 1.3883715867996216, "step": 31923 }, { "epoch": 2.5166732361056363, "grad_norm": 0.40823134779930115, "learning_rate": 0.0005, "loss": 1.2084009647369385, "step": 31924 }, { "epoch": 2.5167520693732754, "grad_norm": 0.4309360086917877, "learning_rate": 0.0005, "loss": 1.32999849319458, "step": 31925 }, { "epoch": 2.5168309026409146, "grad_norm": 0.5654857754707336, "learning_rate": 0.0005, "loss": 1.4422600269317627, "step": 31926 }, { "epoch": 2.5169097359085533, "grad_norm": 0.47314581274986267, "learning_rate": 0.0005, "loss": 1.2226406335830688, "step": 31927 }, { "epoch": 2.5169885691761924, "grad_norm": 0.49597808718681335, "learning_rate": 0.0005, "loss": 1.2004282474517822, "step": 31928 }, { "epoch": 2.517067402443831, "grad_norm": 0.4819900691509247, "learning_rate": 0.0005, "loss": 1.4785182476043701, "step": 31929 }, { "epoch": 2.5171462357114702, "grad_norm": 1.052259087562561, "learning_rate": 0.0005, "loss": 1.2731043100357056, "step": 31930 }, { "epoch": 2.5172250689791094, "grad_norm": 0.5280608534812927, "learning_rate": 0.0005, "loss": 1.2941901683807373, "step": 31931 }, { "epoch": 2.517303902246748, "grad_norm": 0.49798399209976196, "learning_rate": 0.0005, "loss": 1.3001958131790161, "step": 31932 }, { "epoch": 2.517382735514387, "grad_norm": 0.42471346259117126, "learning_rate": 0.0005, "loss": 1.1090424060821533, "step": 31933 }, { "epoch": 2.517461568782026, "grad_norm": 0.5330187082290649, "learning_rate": 0.0005, "loss": 1.5688235759735107, "step": 31934 }, { "epoch": 2.517540402049665, "grad_norm": 0.6109955310821533, "learning_rate": 0.0005, "loss": 1.6853244304656982, "step": 31935 }, { "epoch": 2.517619235317304, "grad_norm": 0.5994796752929688, "learning_rate": 0.0005, "loss": 1.4775224924087524, "step": 31936 }, { "epoch": 2.517698068584943, "grad_norm": 0.4900706112384796, "learning_rate": 0.0005, "loss": 1.2376097440719604, "step": 31937 }, { "epoch": 2.5177769018525815, "grad_norm": 0.46536439657211304, "learning_rate": 0.0005, "loss": 1.2739315032958984, "step": 31938 }, { "epoch": 2.5178557351202207, "grad_norm": 0.5518657565116882, "learning_rate": 0.0005, "loss": 1.3472858667373657, "step": 31939 }, { "epoch": 2.51793456838786, "grad_norm": 0.41142386198043823, "learning_rate": 0.0005, "loss": 1.1658470630645752, "step": 31940 }, { "epoch": 2.5180134016554985, "grad_norm": 0.4901652932167053, "learning_rate": 0.0005, "loss": 1.4063551425933838, "step": 31941 }, { "epoch": 2.5180922349231376, "grad_norm": 0.5234354734420776, "learning_rate": 0.0005, "loss": 1.339174747467041, "step": 31942 }, { "epoch": 2.5181710681907763, "grad_norm": 0.5012384057044983, "learning_rate": 0.0005, "loss": 1.2875423431396484, "step": 31943 }, { "epoch": 2.5182499014584154, "grad_norm": 0.5938750505447388, "learning_rate": 0.0005, "loss": 1.399282693862915, "step": 31944 }, { "epoch": 2.5183287347260546, "grad_norm": 0.48763489723205566, "learning_rate": 0.0005, "loss": 1.3488876819610596, "step": 31945 }, { "epoch": 2.5184075679936933, "grad_norm": 0.5479346513748169, "learning_rate": 0.0005, "loss": 1.358186960220337, "step": 31946 }, { "epoch": 2.5184864012613324, "grad_norm": 0.4758586585521698, "learning_rate": 0.0005, "loss": 1.414601445198059, "step": 31947 }, { "epoch": 2.518565234528971, "grad_norm": 0.4769579768180847, "learning_rate": 0.0005, "loss": 1.2551615238189697, "step": 31948 }, { "epoch": 2.51864406779661, "grad_norm": 0.5522769093513489, "learning_rate": 0.0005, "loss": 1.5454059839248657, "step": 31949 }, { "epoch": 2.5187229010642493, "grad_norm": 0.43126481771469116, "learning_rate": 0.0005, "loss": 1.0430371761322021, "step": 31950 }, { "epoch": 2.518801734331888, "grad_norm": 0.39963358640670776, "learning_rate": 0.0005, "loss": 1.117156744003296, "step": 31951 }, { "epoch": 2.518880567599527, "grad_norm": 0.7567199468612671, "learning_rate": 0.0005, "loss": 1.6138118505477905, "step": 31952 }, { "epoch": 2.518959400867166, "grad_norm": 0.5087522864341736, "learning_rate": 0.0005, "loss": 1.3572359085083008, "step": 31953 }, { "epoch": 2.519038234134805, "grad_norm": 0.4476778209209442, "learning_rate": 0.0005, "loss": 1.3250141143798828, "step": 31954 }, { "epoch": 2.5191170674024437, "grad_norm": 0.4185933470726013, "learning_rate": 0.0005, "loss": 1.333097219467163, "step": 31955 }, { "epoch": 2.519195900670083, "grad_norm": 0.559988260269165, "learning_rate": 0.0005, "loss": 1.311127781867981, "step": 31956 }, { "epoch": 2.5192747339377215, "grad_norm": 0.6053204536437988, "learning_rate": 0.0005, "loss": 1.3825633525848389, "step": 31957 }, { "epoch": 2.5193535672053606, "grad_norm": 0.5487314462661743, "learning_rate": 0.0005, "loss": 1.2998777627944946, "step": 31958 }, { "epoch": 2.5194324004729998, "grad_norm": 0.6480600237846375, "learning_rate": 0.0005, "loss": 1.6363818645477295, "step": 31959 }, { "epoch": 2.5195112337406385, "grad_norm": 0.5407010912895203, "learning_rate": 0.0005, "loss": 1.4291139841079712, "step": 31960 }, { "epoch": 2.5195900670082776, "grad_norm": 0.42823290824890137, "learning_rate": 0.0005, "loss": 1.162423849105835, "step": 31961 }, { "epoch": 2.5196689002759163, "grad_norm": 0.45537617802619934, "learning_rate": 0.0005, "loss": 1.341426968574524, "step": 31962 }, { "epoch": 2.5197477335435554, "grad_norm": 0.4974406361579895, "learning_rate": 0.0005, "loss": 1.4538297653198242, "step": 31963 }, { "epoch": 2.5198265668111945, "grad_norm": 0.5229921936988831, "learning_rate": 0.0005, "loss": 1.226770281791687, "step": 31964 }, { "epoch": 2.5199054000788332, "grad_norm": 0.5184757709503174, "learning_rate": 0.0005, "loss": 1.3384087085723877, "step": 31965 }, { "epoch": 2.5199842333464724, "grad_norm": 0.550174355506897, "learning_rate": 0.0005, "loss": 1.2278358936309814, "step": 31966 }, { "epoch": 2.520063066614111, "grad_norm": 0.5141935348510742, "learning_rate": 0.0005, "loss": 1.3252118825912476, "step": 31967 }, { "epoch": 2.52014189988175, "grad_norm": 0.6933255195617676, "learning_rate": 0.0005, "loss": 1.4380507469177246, "step": 31968 }, { "epoch": 2.520220733149389, "grad_norm": 0.6483427882194519, "learning_rate": 0.0005, "loss": 1.5944710969924927, "step": 31969 }, { "epoch": 2.520299566417028, "grad_norm": 0.6753068566322327, "learning_rate": 0.0005, "loss": 1.3216357231140137, "step": 31970 }, { "epoch": 2.5203783996846667, "grad_norm": 0.7433057427406311, "learning_rate": 0.0005, "loss": 1.3164515495300293, "step": 31971 }, { "epoch": 2.520457232952306, "grad_norm": 0.5422965288162231, "learning_rate": 0.0005, "loss": 1.5406641960144043, "step": 31972 }, { "epoch": 2.520536066219945, "grad_norm": 0.47153785824775696, "learning_rate": 0.0005, "loss": 1.245269536972046, "step": 31973 }, { "epoch": 2.5206148994875837, "grad_norm": 0.5439067482948303, "learning_rate": 0.0005, "loss": 1.4138352870941162, "step": 31974 }, { "epoch": 2.520693732755223, "grad_norm": 0.5136392712593079, "learning_rate": 0.0005, "loss": 1.3709988594055176, "step": 31975 }, { "epoch": 2.5207725660228615, "grad_norm": 0.6165544986724854, "learning_rate": 0.0005, "loss": 1.319596529006958, "step": 31976 }, { "epoch": 2.5208513992905006, "grad_norm": 0.4044744074344635, "learning_rate": 0.0005, "loss": 1.0464975833892822, "step": 31977 }, { "epoch": 2.5209302325581397, "grad_norm": 0.5010319352149963, "learning_rate": 0.0005, "loss": 1.2973352670669556, "step": 31978 }, { "epoch": 2.5210090658257784, "grad_norm": 0.6941857933998108, "learning_rate": 0.0005, "loss": 1.3061516284942627, "step": 31979 }, { "epoch": 2.5210878990934176, "grad_norm": 0.5636542439460754, "learning_rate": 0.0005, "loss": 1.285331130027771, "step": 31980 }, { "epoch": 2.5211667323610563, "grad_norm": 0.5627965331077576, "learning_rate": 0.0005, "loss": 1.1623420715332031, "step": 31981 }, { "epoch": 2.5212455656286954, "grad_norm": 1.381604790687561, "learning_rate": 0.0005, "loss": 1.3102456331253052, "step": 31982 }, { "epoch": 2.5213243988963345, "grad_norm": 0.6286538243293762, "learning_rate": 0.0005, "loss": 1.398179054260254, "step": 31983 }, { "epoch": 2.521403232163973, "grad_norm": 0.6229990124702454, "learning_rate": 0.0005, "loss": 1.351379632949829, "step": 31984 }, { "epoch": 2.521482065431612, "grad_norm": 0.4377659261226654, "learning_rate": 0.0005, "loss": 1.4877816438674927, "step": 31985 }, { "epoch": 2.521560898699251, "grad_norm": 0.5998595952987671, "learning_rate": 0.0005, "loss": 1.4346587657928467, "step": 31986 }, { "epoch": 2.52163973196689, "grad_norm": 0.4874817728996277, "learning_rate": 0.0005, "loss": 1.2956860065460205, "step": 31987 }, { "epoch": 2.521718565234529, "grad_norm": 0.5204861164093018, "learning_rate": 0.0005, "loss": 1.074436068534851, "step": 31988 }, { "epoch": 2.521797398502168, "grad_norm": 0.7766536474227905, "learning_rate": 0.0005, "loss": 1.5173221826553345, "step": 31989 }, { "epoch": 2.5218762317698067, "grad_norm": 0.5253868103027344, "learning_rate": 0.0005, "loss": 1.4720669984817505, "step": 31990 }, { "epoch": 2.521955065037446, "grad_norm": 0.4015316665172577, "learning_rate": 0.0005, "loss": 1.1709232330322266, "step": 31991 }, { "epoch": 2.522033898305085, "grad_norm": 0.6234300136566162, "learning_rate": 0.0005, "loss": 1.262994647026062, "step": 31992 }, { "epoch": 2.5221127315727236, "grad_norm": 0.4739050269126892, "learning_rate": 0.0005, "loss": 1.1489816904067993, "step": 31993 }, { "epoch": 2.5221915648403628, "grad_norm": 0.5284963846206665, "learning_rate": 0.0005, "loss": 1.264898419380188, "step": 31994 }, { "epoch": 2.5222703981080015, "grad_norm": 0.4764401614665985, "learning_rate": 0.0005, "loss": 1.3087040185928345, "step": 31995 }, { "epoch": 2.5223492313756406, "grad_norm": 0.5168656706809998, "learning_rate": 0.0005, "loss": 1.42476224899292, "step": 31996 }, { "epoch": 2.5224280646432797, "grad_norm": 0.49817129969596863, "learning_rate": 0.0005, "loss": 1.2519776821136475, "step": 31997 }, { "epoch": 2.5225068979109184, "grad_norm": 0.5384326577186584, "learning_rate": 0.0005, "loss": 1.400308609008789, "step": 31998 }, { "epoch": 2.522585731178557, "grad_norm": 0.5593602061271667, "learning_rate": 0.0005, "loss": 1.4593250751495361, "step": 31999 }, { "epoch": 2.5226645644461962, "grad_norm": 0.7084818482398987, "learning_rate": 0.0005, "loss": 1.5102078914642334, "step": 32000 }, { "epoch": 2.5227433977138354, "grad_norm": 0.46101173758506775, "learning_rate": 0.0005, "loss": 1.3144220113754272, "step": 32001 }, { "epoch": 2.522822230981474, "grad_norm": 0.5387885570526123, "learning_rate": 0.0005, "loss": 1.2469639778137207, "step": 32002 }, { "epoch": 2.522901064249113, "grad_norm": 0.5732427835464478, "learning_rate": 0.0005, "loss": 1.4353704452514648, "step": 32003 }, { "epoch": 2.522979897516752, "grad_norm": 0.5466023087501526, "learning_rate": 0.0005, "loss": 1.3499419689178467, "step": 32004 }, { "epoch": 2.523058730784391, "grad_norm": 0.42817723751068115, "learning_rate": 0.0005, "loss": 1.1419155597686768, "step": 32005 }, { "epoch": 2.52313756405203, "grad_norm": 0.5005800724029541, "learning_rate": 0.0005, "loss": 1.2429691553115845, "step": 32006 }, { "epoch": 2.523216397319669, "grad_norm": 0.40339428186416626, "learning_rate": 0.0005, "loss": 1.1718820333480835, "step": 32007 }, { "epoch": 2.523295230587308, "grad_norm": 0.5314835906028748, "learning_rate": 0.0005, "loss": 1.3429204225540161, "step": 32008 }, { "epoch": 2.5233740638549467, "grad_norm": 0.5241599082946777, "learning_rate": 0.0005, "loss": 1.2791571617126465, "step": 32009 }, { "epoch": 2.523452897122586, "grad_norm": 0.4751538336277008, "learning_rate": 0.0005, "loss": 1.2468531131744385, "step": 32010 }, { "epoch": 2.523531730390225, "grad_norm": 0.5177423357963562, "learning_rate": 0.0005, "loss": 1.2949856519699097, "step": 32011 }, { "epoch": 2.5236105636578636, "grad_norm": 0.41146230697631836, "learning_rate": 0.0005, "loss": 1.1465407609939575, "step": 32012 }, { "epoch": 2.5236893969255023, "grad_norm": 0.5930464863777161, "learning_rate": 0.0005, "loss": 1.5124073028564453, "step": 32013 }, { "epoch": 2.5237682301931414, "grad_norm": 0.4579027593135834, "learning_rate": 0.0005, "loss": 1.2931675910949707, "step": 32014 }, { "epoch": 2.5238470634607806, "grad_norm": 0.5225745439529419, "learning_rate": 0.0005, "loss": 1.3695592880249023, "step": 32015 }, { "epoch": 2.5239258967284193, "grad_norm": 0.5149315595626831, "learning_rate": 0.0005, "loss": 1.3335180282592773, "step": 32016 }, { "epoch": 2.5240047299960584, "grad_norm": 0.5105911493301392, "learning_rate": 0.0005, "loss": 1.313488483428955, "step": 32017 }, { "epoch": 2.524083563263697, "grad_norm": 0.600527822971344, "learning_rate": 0.0005, "loss": 1.4238642454147339, "step": 32018 }, { "epoch": 2.5241623965313362, "grad_norm": 0.42176952958106995, "learning_rate": 0.0005, "loss": 1.2766814231872559, "step": 32019 }, { "epoch": 2.5242412297989754, "grad_norm": 0.39801454544067383, "learning_rate": 0.0005, "loss": 1.3393418788909912, "step": 32020 }, { "epoch": 2.524320063066614, "grad_norm": 0.6608341932296753, "learning_rate": 0.0005, "loss": 1.4506012201309204, "step": 32021 }, { "epoch": 2.524398896334253, "grad_norm": 0.5603823065757751, "learning_rate": 0.0005, "loss": 1.41732919216156, "step": 32022 }, { "epoch": 2.524477729601892, "grad_norm": 0.5497627854347229, "learning_rate": 0.0005, "loss": 1.1886413097381592, "step": 32023 }, { "epoch": 2.524556562869531, "grad_norm": 0.48983561992645264, "learning_rate": 0.0005, "loss": 1.4795094728469849, "step": 32024 }, { "epoch": 2.52463539613717, "grad_norm": 0.4675392806529999, "learning_rate": 0.0005, "loss": 1.2462522983551025, "step": 32025 }, { "epoch": 2.524714229404809, "grad_norm": 0.4715273082256317, "learning_rate": 0.0005, "loss": 1.24049973487854, "step": 32026 }, { "epoch": 2.5247930626724475, "grad_norm": 0.5421560406684875, "learning_rate": 0.0005, "loss": 1.3265812397003174, "step": 32027 }, { "epoch": 2.5248718959400867, "grad_norm": 0.5325769186019897, "learning_rate": 0.0005, "loss": 1.3567287921905518, "step": 32028 }, { "epoch": 2.524950729207726, "grad_norm": 0.557667076587677, "learning_rate": 0.0005, "loss": 1.517961025238037, "step": 32029 }, { "epoch": 2.5250295624753645, "grad_norm": 0.5558415055274963, "learning_rate": 0.0005, "loss": 1.4457951784133911, "step": 32030 }, { "epoch": 2.5251083957430036, "grad_norm": 0.4987509250640869, "learning_rate": 0.0005, "loss": 1.2064303159713745, "step": 32031 }, { "epoch": 2.5251872290106423, "grad_norm": 0.536689281463623, "learning_rate": 0.0005, "loss": 1.5213581323623657, "step": 32032 }, { "epoch": 2.5252660622782814, "grad_norm": 0.4594377875328064, "learning_rate": 0.0005, "loss": 1.204484462738037, "step": 32033 }, { "epoch": 2.5253448955459206, "grad_norm": 0.7349574565887451, "learning_rate": 0.0005, "loss": 1.3258960247039795, "step": 32034 }, { "epoch": 2.5254237288135593, "grad_norm": 0.5983296632766724, "learning_rate": 0.0005, "loss": 1.4049440622329712, "step": 32035 }, { "epoch": 2.5255025620811984, "grad_norm": 0.60123211145401, "learning_rate": 0.0005, "loss": 1.391012191772461, "step": 32036 }, { "epoch": 2.525581395348837, "grad_norm": 0.45255008339881897, "learning_rate": 0.0005, "loss": 1.2081823348999023, "step": 32037 }, { "epoch": 2.525660228616476, "grad_norm": 0.5066790580749512, "learning_rate": 0.0005, "loss": 1.325695276260376, "step": 32038 }, { "epoch": 2.5257390618841153, "grad_norm": 0.7138577699661255, "learning_rate": 0.0005, "loss": 1.550649881362915, "step": 32039 }, { "epoch": 2.525817895151754, "grad_norm": 0.5433018803596497, "learning_rate": 0.0005, "loss": 1.2983365058898926, "step": 32040 }, { "epoch": 2.5258967284193927, "grad_norm": 0.5593668222427368, "learning_rate": 0.0005, "loss": 1.2344703674316406, "step": 32041 }, { "epoch": 2.525975561687032, "grad_norm": 0.5086607336997986, "learning_rate": 0.0005, "loss": 1.2455434799194336, "step": 32042 }, { "epoch": 2.526054394954671, "grad_norm": 0.474703848361969, "learning_rate": 0.0005, "loss": 1.4189958572387695, "step": 32043 }, { "epoch": 2.5261332282223097, "grad_norm": 0.48500749468803406, "learning_rate": 0.0005, "loss": 1.2426083087921143, "step": 32044 }, { "epoch": 2.526212061489949, "grad_norm": 0.4562496244907379, "learning_rate": 0.0005, "loss": 1.2551722526550293, "step": 32045 }, { "epoch": 2.5262908947575875, "grad_norm": 0.4978227913379669, "learning_rate": 0.0005, "loss": 1.3371753692626953, "step": 32046 }, { "epoch": 2.5263697280252266, "grad_norm": 0.5308449268341064, "learning_rate": 0.0005, "loss": 1.3580784797668457, "step": 32047 }, { "epoch": 2.5264485612928658, "grad_norm": 0.644612729549408, "learning_rate": 0.0005, "loss": 1.4872283935546875, "step": 32048 }, { "epoch": 2.5265273945605045, "grad_norm": 0.5478942394256592, "learning_rate": 0.0005, "loss": 1.2001986503601074, "step": 32049 }, { "epoch": 2.5266062278281436, "grad_norm": 0.4321357309818268, "learning_rate": 0.0005, "loss": 1.3658019304275513, "step": 32050 }, { "epoch": 2.5266850610957823, "grad_norm": 0.5445257425308228, "learning_rate": 0.0005, "loss": 1.1923677921295166, "step": 32051 }, { "epoch": 2.5267638943634214, "grad_norm": 0.42238733172416687, "learning_rate": 0.0005, "loss": 1.3210630416870117, "step": 32052 }, { "epoch": 2.5268427276310605, "grad_norm": 0.5601441264152527, "learning_rate": 0.0005, "loss": 1.596853494644165, "step": 32053 }, { "epoch": 2.5269215608986992, "grad_norm": 0.4163070023059845, "learning_rate": 0.0005, "loss": 1.2956907749176025, "step": 32054 }, { "epoch": 2.527000394166338, "grad_norm": 0.5086731314659119, "learning_rate": 0.0005, "loss": 1.2498574256896973, "step": 32055 }, { "epoch": 2.527079227433977, "grad_norm": 0.5312511324882507, "learning_rate": 0.0005, "loss": 1.2023833990097046, "step": 32056 }, { "epoch": 2.527158060701616, "grad_norm": 0.4908161461353302, "learning_rate": 0.0005, "loss": 1.1828947067260742, "step": 32057 }, { "epoch": 2.527236893969255, "grad_norm": 0.412117063999176, "learning_rate": 0.0005, "loss": 1.205425500869751, "step": 32058 }, { "epoch": 2.527315727236894, "grad_norm": 0.4423067271709442, "learning_rate": 0.0005, "loss": 1.255071997642517, "step": 32059 }, { "epoch": 2.5273945605045327, "grad_norm": 0.5188002586364746, "learning_rate": 0.0005, "loss": 1.2895275354385376, "step": 32060 }, { "epoch": 2.527473393772172, "grad_norm": 0.4202776253223419, "learning_rate": 0.0005, "loss": 1.3544437885284424, "step": 32061 }, { "epoch": 2.527552227039811, "grad_norm": 0.4770662784576416, "learning_rate": 0.0005, "loss": 1.182056188583374, "step": 32062 }, { "epoch": 2.5276310603074497, "grad_norm": 0.5400170087814331, "learning_rate": 0.0005, "loss": 1.4297617673873901, "step": 32063 }, { "epoch": 2.527709893575089, "grad_norm": 0.4643709063529968, "learning_rate": 0.0005, "loss": 1.3587334156036377, "step": 32064 }, { "epoch": 2.5277887268427275, "grad_norm": 0.511825680732727, "learning_rate": 0.0005, "loss": 1.581068992614746, "step": 32065 }, { "epoch": 2.5278675601103666, "grad_norm": 0.7426891922950745, "learning_rate": 0.0005, "loss": 1.4973139762878418, "step": 32066 }, { "epoch": 2.5279463933780058, "grad_norm": 0.5435991287231445, "learning_rate": 0.0005, "loss": 1.4418604373931885, "step": 32067 }, { "epoch": 2.5280252266456444, "grad_norm": 0.4349218010902405, "learning_rate": 0.0005, "loss": 1.3544713258743286, "step": 32068 }, { "epoch": 2.528104059913283, "grad_norm": 0.4391182065010071, "learning_rate": 0.0005, "loss": 1.3001537322998047, "step": 32069 }, { "epoch": 2.5281828931809223, "grad_norm": 0.7459487318992615, "learning_rate": 0.0005, "loss": 1.1955053806304932, "step": 32070 }, { "epoch": 2.5282617264485614, "grad_norm": 0.41858717799186707, "learning_rate": 0.0005, "loss": 1.202671766281128, "step": 32071 }, { "epoch": 2.5283405597162, "grad_norm": 0.4848272502422333, "learning_rate": 0.0005, "loss": 1.3684380054473877, "step": 32072 }, { "epoch": 2.528419392983839, "grad_norm": 0.8390281200408936, "learning_rate": 0.0005, "loss": 1.3851568698883057, "step": 32073 }, { "epoch": 2.528498226251478, "grad_norm": 0.5671613216400146, "learning_rate": 0.0005, "loss": 1.4457306861877441, "step": 32074 }, { "epoch": 2.528577059519117, "grad_norm": 0.5453099608421326, "learning_rate": 0.0005, "loss": 1.4255833625793457, "step": 32075 }, { "epoch": 2.528655892786756, "grad_norm": 0.6040134429931641, "learning_rate": 0.0005, "loss": 1.4998767375946045, "step": 32076 }, { "epoch": 2.528734726054395, "grad_norm": 0.5731556415557861, "learning_rate": 0.0005, "loss": 1.203352451324463, "step": 32077 }, { "epoch": 2.528813559322034, "grad_norm": 0.6591706275939941, "learning_rate": 0.0005, "loss": 1.3192795515060425, "step": 32078 }, { "epoch": 2.5288923925896727, "grad_norm": 0.5054331421852112, "learning_rate": 0.0005, "loss": 1.2052048444747925, "step": 32079 }, { "epoch": 2.528971225857312, "grad_norm": 0.782189667224884, "learning_rate": 0.0005, "loss": 1.475109338760376, "step": 32080 }, { "epoch": 2.529050059124951, "grad_norm": 0.8787041306495667, "learning_rate": 0.0005, "loss": 1.3844165802001953, "step": 32081 }, { "epoch": 2.5291288923925896, "grad_norm": 0.6680632829666138, "learning_rate": 0.0005, "loss": 1.2344377040863037, "step": 32082 }, { "epoch": 2.529207725660229, "grad_norm": 0.5101026296615601, "learning_rate": 0.0005, "loss": 1.5773112773895264, "step": 32083 }, { "epoch": 2.5292865589278675, "grad_norm": 0.4494248330593109, "learning_rate": 0.0005, "loss": 1.2254074811935425, "step": 32084 }, { "epoch": 2.5293653921955066, "grad_norm": 0.5518826246261597, "learning_rate": 0.0005, "loss": 1.2732534408569336, "step": 32085 }, { "epoch": 2.5294442254631453, "grad_norm": 0.6107111573219299, "learning_rate": 0.0005, "loss": 1.3676313161849976, "step": 32086 }, { "epoch": 2.5295230587307844, "grad_norm": 0.6148934364318848, "learning_rate": 0.0005, "loss": 1.4001898765563965, "step": 32087 }, { "epoch": 2.529601891998423, "grad_norm": 0.7946754097938538, "learning_rate": 0.0005, "loss": 1.44631028175354, "step": 32088 }, { "epoch": 2.5296807252660622, "grad_norm": 0.4615824818611145, "learning_rate": 0.0005, "loss": 1.2655569314956665, "step": 32089 }, { "epoch": 2.5297595585337014, "grad_norm": 0.6110896468162537, "learning_rate": 0.0005, "loss": 1.219518780708313, "step": 32090 }, { "epoch": 2.52983839180134, "grad_norm": 0.5065288543701172, "learning_rate": 0.0005, "loss": 1.2122522592544556, "step": 32091 }, { "epoch": 2.529917225068979, "grad_norm": 0.6831658482551575, "learning_rate": 0.0005, "loss": 1.5721805095672607, "step": 32092 }, { "epoch": 2.529996058336618, "grad_norm": 0.5124505162239075, "learning_rate": 0.0005, "loss": 1.3654868602752686, "step": 32093 }, { "epoch": 2.530074891604257, "grad_norm": 0.5564518570899963, "learning_rate": 0.0005, "loss": 1.3936331272125244, "step": 32094 }, { "epoch": 2.530153724871896, "grad_norm": 0.5786979794502258, "learning_rate": 0.0005, "loss": 1.303653359413147, "step": 32095 }, { "epoch": 2.530232558139535, "grad_norm": 0.5518765449523926, "learning_rate": 0.0005, "loss": 1.46025812625885, "step": 32096 }, { "epoch": 2.530311391407174, "grad_norm": 0.5117413997650146, "learning_rate": 0.0005, "loss": 1.4584181308746338, "step": 32097 }, { "epoch": 2.5303902246748127, "grad_norm": 0.5376760363578796, "learning_rate": 0.0005, "loss": 1.4352483749389648, "step": 32098 }, { "epoch": 2.530469057942452, "grad_norm": 0.46075454354286194, "learning_rate": 0.0005, "loss": 1.2361371517181396, "step": 32099 }, { "epoch": 2.5305478912100905, "grad_norm": 0.5541406273841858, "learning_rate": 0.0005, "loss": 1.411924123764038, "step": 32100 }, { "epoch": 2.5306267244777296, "grad_norm": 0.47083553671836853, "learning_rate": 0.0005, "loss": 1.3139073848724365, "step": 32101 }, { "epoch": 2.5307055577453683, "grad_norm": 0.43787482380867004, "learning_rate": 0.0005, "loss": 1.3463985919952393, "step": 32102 }, { "epoch": 2.5307843910130075, "grad_norm": 0.44730672240257263, "learning_rate": 0.0005, "loss": 1.4149723052978516, "step": 32103 }, { "epoch": 2.5308632242806466, "grad_norm": 0.49723342061042786, "learning_rate": 0.0005, "loss": 1.4167057275772095, "step": 32104 }, { "epoch": 2.5309420575482853, "grad_norm": 0.6446092128753662, "learning_rate": 0.0005, "loss": 1.4896894693374634, "step": 32105 }, { "epoch": 2.5310208908159244, "grad_norm": 0.49684804677963257, "learning_rate": 0.0005, "loss": 1.3862581253051758, "step": 32106 }, { "epoch": 2.531099724083563, "grad_norm": 0.3839109539985657, "learning_rate": 0.0005, "loss": 1.2585370540618896, "step": 32107 }, { "epoch": 2.5311785573512022, "grad_norm": 0.5554133057594299, "learning_rate": 0.0005, "loss": 1.3995206356048584, "step": 32108 }, { "epoch": 2.5312573906188414, "grad_norm": 0.5187278985977173, "learning_rate": 0.0005, "loss": 1.34891676902771, "step": 32109 }, { "epoch": 2.53133622388648, "grad_norm": 0.39530327916145325, "learning_rate": 0.0005, "loss": 1.230499505996704, "step": 32110 }, { "epoch": 2.531415057154119, "grad_norm": 0.45035606622695923, "learning_rate": 0.0005, "loss": 1.2386376857757568, "step": 32111 }, { "epoch": 2.531493890421758, "grad_norm": 0.531329333782196, "learning_rate": 0.0005, "loss": 1.283882737159729, "step": 32112 }, { "epoch": 2.531572723689397, "grad_norm": 0.36394003033638, "learning_rate": 0.0005, "loss": 1.0161226987838745, "step": 32113 }, { "epoch": 2.531651556957036, "grad_norm": 0.46117231249809265, "learning_rate": 0.0005, "loss": 1.1718794107437134, "step": 32114 }, { "epoch": 2.531730390224675, "grad_norm": 0.6039759516716003, "learning_rate": 0.0005, "loss": 1.3024930953979492, "step": 32115 }, { "epoch": 2.5318092234923135, "grad_norm": 0.47104862332344055, "learning_rate": 0.0005, "loss": 1.3089717626571655, "step": 32116 }, { "epoch": 2.5318880567599527, "grad_norm": 0.4999769628047943, "learning_rate": 0.0005, "loss": 1.6319921016693115, "step": 32117 }, { "epoch": 2.531966890027592, "grad_norm": 0.5464063882827759, "learning_rate": 0.0005, "loss": 1.5740993022918701, "step": 32118 }, { "epoch": 2.5320457232952305, "grad_norm": 0.45367780327796936, "learning_rate": 0.0005, "loss": 1.4137725830078125, "step": 32119 }, { "epoch": 2.5321245565628696, "grad_norm": 0.5458285212516785, "learning_rate": 0.0005, "loss": 1.2936434745788574, "step": 32120 }, { "epoch": 2.5322033898305083, "grad_norm": 0.5023491978645325, "learning_rate": 0.0005, "loss": 1.3981003761291504, "step": 32121 }, { "epoch": 2.5322822230981474, "grad_norm": 0.4817255735397339, "learning_rate": 0.0005, "loss": 1.1526880264282227, "step": 32122 }, { "epoch": 2.5323610563657866, "grad_norm": 0.5631568431854248, "learning_rate": 0.0005, "loss": 1.5461843013763428, "step": 32123 }, { "epoch": 2.5324398896334253, "grad_norm": 0.5816189050674438, "learning_rate": 0.0005, "loss": 1.230910301208496, "step": 32124 }, { "epoch": 2.5325187229010644, "grad_norm": 0.5003822445869446, "learning_rate": 0.0005, "loss": 1.3586276769638062, "step": 32125 }, { "epoch": 2.532597556168703, "grad_norm": 0.5239286422729492, "learning_rate": 0.0005, "loss": 1.4082897901535034, "step": 32126 }, { "epoch": 2.532676389436342, "grad_norm": 0.4782586395740509, "learning_rate": 0.0005, "loss": 1.3459351062774658, "step": 32127 }, { "epoch": 2.5327552227039813, "grad_norm": 0.7735648155212402, "learning_rate": 0.0005, "loss": 1.4778172969818115, "step": 32128 }, { "epoch": 2.53283405597162, "grad_norm": 0.5469108819961548, "learning_rate": 0.0005, "loss": 1.5173790454864502, "step": 32129 }, { "epoch": 2.5329128892392587, "grad_norm": 0.4457463324069977, "learning_rate": 0.0005, "loss": 1.3488460779190063, "step": 32130 }, { "epoch": 2.532991722506898, "grad_norm": 0.49944159388542175, "learning_rate": 0.0005, "loss": 1.3040859699249268, "step": 32131 }, { "epoch": 2.533070555774537, "grad_norm": 0.4427109658718109, "learning_rate": 0.0005, "loss": 1.3737657070159912, "step": 32132 }, { "epoch": 2.5331493890421757, "grad_norm": 0.47279268503189087, "learning_rate": 0.0005, "loss": 1.3122752904891968, "step": 32133 }, { "epoch": 2.533228222309815, "grad_norm": 0.4843633472919464, "learning_rate": 0.0005, "loss": 1.173889398574829, "step": 32134 }, { "epoch": 2.5333070555774535, "grad_norm": 0.5040220022201538, "learning_rate": 0.0005, "loss": 1.3707334995269775, "step": 32135 }, { "epoch": 2.5333858888450926, "grad_norm": 0.4459812343120575, "learning_rate": 0.0005, "loss": 1.3086953163146973, "step": 32136 }, { "epoch": 2.5334647221127318, "grad_norm": 0.47516101598739624, "learning_rate": 0.0005, "loss": 1.303921103477478, "step": 32137 }, { "epoch": 2.5335435553803705, "grad_norm": 0.6002056002616882, "learning_rate": 0.0005, "loss": 1.3650994300842285, "step": 32138 }, { "epoch": 2.5336223886480096, "grad_norm": 0.45833078026771545, "learning_rate": 0.0005, "loss": 1.2670742273330688, "step": 32139 }, { "epoch": 2.5337012219156483, "grad_norm": 0.45786362886428833, "learning_rate": 0.0005, "loss": 1.3296477794647217, "step": 32140 }, { "epoch": 2.5337800551832874, "grad_norm": 0.45966440439224243, "learning_rate": 0.0005, "loss": 1.4681084156036377, "step": 32141 }, { "epoch": 2.5338588884509265, "grad_norm": 0.5147836804389954, "learning_rate": 0.0005, "loss": 1.4578609466552734, "step": 32142 }, { "epoch": 2.5339377217185652, "grad_norm": 0.4661845564842224, "learning_rate": 0.0005, "loss": 1.5629253387451172, "step": 32143 }, { "epoch": 2.534016554986204, "grad_norm": 0.40962332487106323, "learning_rate": 0.0005, "loss": 1.2200883626937866, "step": 32144 }, { "epoch": 2.534095388253843, "grad_norm": 0.5273574590682983, "learning_rate": 0.0005, "loss": 1.4494097232818604, "step": 32145 }, { "epoch": 2.534174221521482, "grad_norm": 0.49743834137916565, "learning_rate": 0.0005, "loss": 1.4288365840911865, "step": 32146 }, { "epoch": 2.534253054789121, "grad_norm": 0.44736480712890625, "learning_rate": 0.0005, "loss": 1.2667088508605957, "step": 32147 }, { "epoch": 2.53433188805676, "grad_norm": 0.665290117263794, "learning_rate": 0.0005, "loss": 1.6268551349639893, "step": 32148 }, { "epoch": 2.5344107213243987, "grad_norm": 0.5670962333679199, "learning_rate": 0.0005, "loss": 1.3490681648254395, "step": 32149 }, { "epoch": 2.534489554592038, "grad_norm": 0.49861589074134827, "learning_rate": 0.0005, "loss": 1.1765053272247314, "step": 32150 }, { "epoch": 2.534568387859677, "grad_norm": 0.4545673131942749, "learning_rate": 0.0005, "loss": 1.510474681854248, "step": 32151 }, { "epoch": 2.5346472211273157, "grad_norm": 0.46556225419044495, "learning_rate": 0.0005, "loss": 1.4044623374938965, "step": 32152 }, { "epoch": 2.534726054394955, "grad_norm": 0.7293512225151062, "learning_rate": 0.0005, "loss": 1.4901483058929443, "step": 32153 }, { "epoch": 2.5348048876625935, "grad_norm": 0.5466985106468201, "learning_rate": 0.0005, "loss": 1.4289616346359253, "step": 32154 }, { "epoch": 2.5348837209302326, "grad_norm": 0.4135802090167999, "learning_rate": 0.0005, "loss": 1.1185745000839233, "step": 32155 }, { "epoch": 2.5349625541978718, "grad_norm": 0.49379557371139526, "learning_rate": 0.0005, "loss": 1.401802659034729, "step": 32156 }, { "epoch": 2.5350413874655104, "grad_norm": 0.4917003810405731, "learning_rate": 0.0005, "loss": 1.3246071338653564, "step": 32157 }, { "epoch": 2.535120220733149, "grad_norm": 0.4582184851169586, "learning_rate": 0.0005, "loss": 1.2605538368225098, "step": 32158 }, { "epoch": 2.5351990540007883, "grad_norm": 0.4384053945541382, "learning_rate": 0.0005, "loss": 1.1583603620529175, "step": 32159 }, { "epoch": 2.5352778872684274, "grad_norm": 0.4413393437862396, "learning_rate": 0.0005, "loss": 1.204969882965088, "step": 32160 }, { "epoch": 2.535356720536066, "grad_norm": 0.5521956086158752, "learning_rate": 0.0005, "loss": 1.4446511268615723, "step": 32161 }, { "epoch": 2.535435553803705, "grad_norm": 0.4735279679298401, "learning_rate": 0.0005, "loss": 1.2913954257965088, "step": 32162 }, { "epoch": 2.535514387071344, "grad_norm": 0.47190359234809875, "learning_rate": 0.0005, "loss": 1.345497727394104, "step": 32163 }, { "epoch": 2.535593220338983, "grad_norm": 0.5349177122116089, "learning_rate": 0.0005, "loss": 1.2837746143341064, "step": 32164 }, { "epoch": 2.535672053606622, "grad_norm": 0.42746850848197937, "learning_rate": 0.0005, "loss": 1.325368881225586, "step": 32165 }, { "epoch": 2.535750886874261, "grad_norm": 0.5252687335014343, "learning_rate": 0.0005, "loss": 1.294215202331543, "step": 32166 }, { "epoch": 2.5358297201419, "grad_norm": 0.48905929923057556, "learning_rate": 0.0005, "loss": 1.1068575382232666, "step": 32167 }, { "epoch": 2.5359085534095387, "grad_norm": 0.524485170841217, "learning_rate": 0.0005, "loss": 1.4519281387329102, "step": 32168 }, { "epoch": 2.535987386677178, "grad_norm": 0.47059792280197144, "learning_rate": 0.0005, "loss": 1.0979423522949219, "step": 32169 }, { "epoch": 2.536066219944817, "grad_norm": 0.5796728730201721, "learning_rate": 0.0005, "loss": 1.371891975402832, "step": 32170 }, { "epoch": 2.5361450532124556, "grad_norm": 0.4934035837650299, "learning_rate": 0.0005, "loss": 1.2845759391784668, "step": 32171 }, { "epoch": 2.5362238864800943, "grad_norm": 0.5341967344284058, "learning_rate": 0.0005, "loss": 1.4255424737930298, "step": 32172 }, { "epoch": 2.5363027197477335, "grad_norm": 0.6145732402801514, "learning_rate": 0.0005, "loss": 1.1935063600540161, "step": 32173 }, { "epoch": 2.5363815530153726, "grad_norm": 0.4933328330516815, "learning_rate": 0.0005, "loss": 1.368635892868042, "step": 32174 }, { "epoch": 2.5364603862830113, "grad_norm": 0.5217344760894775, "learning_rate": 0.0005, "loss": 1.198704481124878, "step": 32175 }, { "epoch": 2.5365392195506504, "grad_norm": 0.5547736883163452, "learning_rate": 0.0005, "loss": 1.3952124118804932, "step": 32176 }, { "epoch": 2.536618052818289, "grad_norm": 0.46881869435310364, "learning_rate": 0.0005, "loss": 1.2086024284362793, "step": 32177 }, { "epoch": 2.5366968860859282, "grad_norm": 0.536138653755188, "learning_rate": 0.0005, "loss": 1.2098037004470825, "step": 32178 }, { "epoch": 2.5367757193535674, "grad_norm": 0.5355392694473267, "learning_rate": 0.0005, "loss": 1.435225248336792, "step": 32179 }, { "epoch": 2.536854552621206, "grad_norm": 0.5307652354240417, "learning_rate": 0.0005, "loss": 1.276422142982483, "step": 32180 }, { "epoch": 2.536933385888845, "grad_norm": 0.44164392352104187, "learning_rate": 0.0005, "loss": 1.3341392278671265, "step": 32181 }, { "epoch": 2.537012219156484, "grad_norm": 0.5784506797790527, "learning_rate": 0.0005, "loss": 1.397531270980835, "step": 32182 }, { "epoch": 2.537091052424123, "grad_norm": 0.6619012951850891, "learning_rate": 0.0005, "loss": 1.442527413368225, "step": 32183 }, { "epoch": 2.537169885691762, "grad_norm": 0.5042853951454163, "learning_rate": 0.0005, "loss": 1.3581101894378662, "step": 32184 }, { "epoch": 2.537248718959401, "grad_norm": 0.45963409543037415, "learning_rate": 0.0005, "loss": 1.2639148235321045, "step": 32185 }, { "epoch": 2.5373275522270395, "grad_norm": 0.542940080165863, "learning_rate": 0.0005, "loss": 1.4047596454620361, "step": 32186 }, { "epoch": 2.5374063854946787, "grad_norm": 0.6394768953323364, "learning_rate": 0.0005, "loss": 1.5984930992126465, "step": 32187 }, { "epoch": 2.537485218762318, "grad_norm": 0.5214694142341614, "learning_rate": 0.0005, "loss": 1.2774295806884766, "step": 32188 }, { "epoch": 2.5375640520299565, "grad_norm": 0.4973539113998413, "learning_rate": 0.0005, "loss": 1.4980792999267578, "step": 32189 }, { "epoch": 2.5376428852975956, "grad_norm": 0.5615149140357971, "learning_rate": 0.0005, "loss": 1.3307437896728516, "step": 32190 }, { "epoch": 2.5377217185652343, "grad_norm": 0.49720850586891174, "learning_rate": 0.0005, "loss": 1.2694392204284668, "step": 32191 }, { "epoch": 2.5378005518328735, "grad_norm": 0.49829670786857605, "learning_rate": 0.0005, "loss": 1.393158197402954, "step": 32192 }, { "epoch": 2.5378793851005126, "grad_norm": 0.5455119609832764, "learning_rate": 0.0005, "loss": 1.2975635528564453, "step": 32193 }, { "epoch": 2.5379582183681513, "grad_norm": 0.4921611249446869, "learning_rate": 0.0005, "loss": 1.4942694902420044, "step": 32194 }, { "epoch": 2.5380370516357904, "grad_norm": 0.4959831237792969, "learning_rate": 0.0005, "loss": 1.3320281505584717, "step": 32195 }, { "epoch": 2.538115884903429, "grad_norm": 0.49517789483070374, "learning_rate": 0.0005, "loss": 1.456118106842041, "step": 32196 }, { "epoch": 2.5381947181710682, "grad_norm": 0.5694095492362976, "learning_rate": 0.0005, "loss": 1.4109996557235718, "step": 32197 }, { "epoch": 2.5382735514387074, "grad_norm": 0.4544921815395355, "learning_rate": 0.0005, "loss": 1.3942160606384277, "step": 32198 }, { "epoch": 2.538352384706346, "grad_norm": 0.3953856825828552, "learning_rate": 0.0005, "loss": 1.18768310546875, "step": 32199 }, { "epoch": 2.538431217973985, "grad_norm": 0.6352625489234924, "learning_rate": 0.0005, "loss": 1.2396903038024902, "step": 32200 }, { "epoch": 2.538510051241624, "grad_norm": 0.4651831388473511, "learning_rate": 0.0005, "loss": 1.1904386281967163, "step": 32201 }, { "epoch": 2.538588884509263, "grad_norm": 0.5055025219917297, "learning_rate": 0.0005, "loss": 1.4007068872451782, "step": 32202 }, { "epoch": 2.5386677177769017, "grad_norm": 0.5014169216156006, "learning_rate": 0.0005, "loss": 1.2352436780929565, "step": 32203 }, { "epoch": 2.538746551044541, "grad_norm": 0.5137884616851807, "learning_rate": 0.0005, "loss": 1.3715184926986694, "step": 32204 }, { "epoch": 2.5388253843121795, "grad_norm": 0.5594609379768372, "learning_rate": 0.0005, "loss": 1.3384416103363037, "step": 32205 }, { "epoch": 2.5389042175798187, "grad_norm": 0.4453800618648529, "learning_rate": 0.0005, "loss": 1.210463523864746, "step": 32206 }, { "epoch": 2.538983050847458, "grad_norm": 0.4415884017944336, "learning_rate": 0.0005, "loss": 1.2944929599761963, "step": 32207 }, { "epoch": 2.5390618841150965, "grad_norm": 0.4354671537876129, "learning_rate": 0.0005, "loss": 1.2351734638214111, "step": 32208 }, { "epoch": 2.5391407173827356, "grad_norm": 0.6109245419502258, "learning_rate": 0.0005, "loss": 1.3618502616882324, "step": 32209 }, { "epoch": 2.5392195506503743, "grad_norm": 0.4913851022720337, "learning_rate": 0.0005, "loss": 1.3385144472122192, "step": 32210 }, { "epoch": 2.5392983839180134, "grad_norm": 0.946568489074707, "learning_rate": 0.0005, "loss": 1.387714147567749, "step": 32211 }, { "epoch": 2.5393772171856526, "grad_norm": 0.47969040274620056, "learning_rate": 0.0005, "loss": 1.1838219165802002, "step": 32212 }, { "epoch": 2.5394560504532913, "grad_norm": 0.4659575819969177, "learning_rate": 0.0005, "loss": 1.294853925704956, "step": 32213 }, { "epoch": 2.5395348837209304, "grad_norm": 0.5473634600639343, "learning_rate": 0.0005, "loss": 1.4505019187927246, "step": 32214 }, { "epoch": 2.539613716988569, "grad_norm": 0.47750553488731384, "learning_rate": 0.0005, "loss": 1.4275710582733154, "step": 32215 }, { "epoch": 2.539692550256208, "grad_norm": 0.513110876083374, "learning_rate": 0.0005, "loss": 1.2167984247207642, "step": 32216 }, { "epoch": 2.539771383523847, "grad_norm": 0.5104542970657349, "learning_rate": 0.0005, "loss": 1.4401204586029053, "step": 32217 }, { "epoch": 2.539850216791486, "grad_norm": 0.4991890490055084, "learning_rate": 0.0005, "loss": 1.3385058641433716, "step": 32218 }, { "epoch": 2.5399290500591247, "grad_norm": 0.6925104856491089, "learning_rate": 0.0005, "loss": 1.4003655910491943, "step": 32219 }, { "epoch": 2.540007883326764, "grad_norm": 0.6512454748153687, "learning_rate": 0.0005, "loss": 1.2746250629425049, "step": 32220 }, { "epoch": 2.540086716594403, "grad_norm": 0.4952813684940338, "learning_rate": 0.0005, "loss": 1.1243857145309448, "step": 32221 }, { "epoch": 2.5401655498620417, "grad_norm": 0.5448160171508789, "learning_rate": 0.0005, "loss": 1.2743840217590332, "step": 32222 }, { "epoch": 2.540244383129681, "grad_norm": 0.5118390917778015, "learning_rate": 0.0005, "loss": 1.2714886665344238, "step": 32223 }, { "epoch": 2.5403232163973195, "grad_norm": 0.4809342622756958, "learning_rate": 0.0005, "loss": 1.39912748336792, "step": 32224 }, { "epoch": 2.5404020496649586, "grad_norm": 0.585709810256958, "learning_rate": 0.0005, "loss": 1.2799758911132812, "step": 32225 }, { "epoch": 2.5404808829325978, "grad_norm": 1.0325970649719238, "learning_rate": 0.0005, "loss": 1.3200316429138184, "step": 32226 }, { "epoch": 2.5405597162002365, "grad_norm": 0.5898028016090393, "learning_rate": 0.0005, "loss": 1.408843994140625, "step": 32227 }, { "epoch": 2.5406385494678756, "grad_norm": 0.4761647880077362, "learning_rate": 0.0005, "loss": 1.2986555099487305, "step": 32228 }, { "epoch": 2.5407173827355143, "grad_norm": 0.5146788358688354, "learning_rate": 0.0005, "loss": 1.365931749343872, "step": 32229 }, { "epoch": 2.5407962160031534, "grad_norm": 0.4492119550704956, "learning_rate": 0.0005, "loss": 1.255744218826294, "step": 32230 }, { "epoch": 2.5408750492707926, "grad_norm": 0.42218929529190063, "learning_rate": 0.0005, "loss": 1.3352124691009521, "step": 32231 }, { "epoch": 2.5409538825384312, "grad_norm": 0.4457130432128906, "learning_rate": 0.0005, "loss": 1.2704085111618042, "step": 32232 }, { "epoch": 2.54103271580607, "grad_norm": 0.502176821231842, "learning_rate": 0.0005, "loss": 1.2695844173431396, "step": 32233 }, { "epoch": 2.541111549073709, "grad_norm": 0.5515351295471191, "learning_rate": 0.0005, "loss": 1.2391730546951294, "step": 32234 }, { "epoch": 2.541190382341348, "grad_norm": 0.4499991834163666, "learning_rate": 0.0005, "loss": 1.5291836261749268, "step": 32235 }, { "epoch": 2.541269215608987, "grad_norm": 0.4638064205646515, "learning_rate": 0.0005, "loss": 1.3715050220489502, "step": 32236 }, { "epoch": 2.541348048876626, "grad_norm": 0.5493003726005554, "learning_rate": 0.0005, "loss": 1.337918996810913, "step": 32237 }, { "epoch": 2.5414268821442647, "grad_norm": 0.4591034948825836, "learning_rate": 0.0005, "loss": 1.1524343490600586, "step": 32238 }, { "epoch": 2.541505715411904, "grad_norm": 0.4456808269023895, "learning_rate": 0.0005, "loss": 1.3124957084655762, "step": 32239 }, { "epoch": 2.541584548679543, "grad_norm": 0.4278355538845062, "learning_rate": 0.0005, "loss": 1.2975337505340576, "step": 32240 }, { "epoch": 2.5416633819471817, "grad_norm": 0.4093661606311798, "learning_rate": 0.0005, "loss": 1.2696936130523682, "step": 32241 }, { "epoch": 2.541742215214821, "grad_norm": 0.4748794734477997, "learning_rate": 0.0005, "loss": 1.3741896152496338, "step": 32242 }, { "epoch": 2.5418210484824595, "grad_norm": 0.4434979557991028, "learning_rate": 0.0005, "loss": 1.4543540477752686, "step": 32243 }, { "epoch": 2.5418998817500986, "grad_norm": 0.4820716977119446, "learning_rate": 0.0005, "loss": 1.2054593563079834, "step": 32244 }, { "epoch": 2.5419787150177378, "grad_norm": 0.43159881234169006, "learning_rate": 0.0005, "loss": 1.217663049697876, "step": 32245 }, { "epoch": 2.5420575482853764, "grad_norm": 0.46170493960380554, "learning_rate": 0.0005, "loss": 1.3532735109329224, "step": 32246 }, { "epoch": 2.542136381553015, "grad_norm": 0.5707929730415344, "learning_rate": 0.0005, "loss": 1.5188133716583252, "step": 32247 }, { "epoch": 2.5422152148206543, "grad_norm": 0.5083668231964111, "learning_rate": 0.0005, "loss": 1.3741240501403809, "step": 32248 }, { "epoch": 2.5422940480882934, "grad_norm": 0.44447991251945496, "learning_rate": 0.0005, "loss": 1.5208579301834106, "step": 32249 }, { "epoch": 2.542372881355932, "grad_norm": 0.43076151609420776, "learning_rate": 0.0005, "loss": 1.3248058557510376, "step": 32250 }, { "epoch": 2.5424517146235712, "grad_norm": 1.782904028892517, "learning_rate": 0.0005, "loss": 1.2483248710632324, "step": 32251 }, { "epoch": 2.54253054789121, "grad_norm": 0.49194058775901794, "learning_rate": 0.0005, "loss": 1.3101186752319336, "step": 32252 }, { "epoch": 2.542609381158849, "grad_norm": 0.43714630603790283, "learning_rate": 0.0005, "loss": 1.2862377166748047, "step": 32253 }, { "epoch": 2.542688214426488, "grad_norm": 0.470528244972229, "learning_rate": 0.0005, "loss": 1.3550844192504883, "step": 32254 }, { "epoch": 2.542767047694127, "grad_norm": 0.7570782899856567, "learning_rate": 0.0005, "loss": 1.3227035999298096, "step": 32255 }, { "epoch": 2.542845880961766, "grad_norm": 0.4029942750930786, "learning_rate": 0.0005, "loss": 1.2769885063171387, "step": 32256 }, { "epoch": 2.5429247142294047, "grad_norm": 0.6180689930915833, "learning_rate": 0.0005, "loss": 1.4225282669067383, "step": 32257 }, { "epoch": 2.543003547497044, "grad_norm": 0.3866922855377197, "learning_rate": 0.0005, "loss": 1.1215643882751465, "step": 32258 }, { "epoch": 2.543082380764683, "grad_norm": 0.46460190415382385, "learning_rate": 0.0005, "loss": 1.155963659286499, "step": 32259 }, { "epoch": 2.5431612140323216, "grad_norm": 0.604115903377533, "learning_rate": 0.0005, "loss": 1.4315757751464844, "step": 32260 }, { "epoch": 2.5432400472999603, "grad_norm": 0.4645061492919922, "learning_rate": 0.0005, "loss": 1.2924844026565552, "step": 32261 }, { "epoch": 2.5433188805675995, "grad_norm": 0.5809798240661621, "learning_rate": 0.0005, "loss": 1.4039125442504883, "step": 32262 }, { "epoch": 2.5433977138352386, "grad_norm": 0.4482874572277069, "learning_rate": 0.0005, "loss": 1.2528777122497559, "step": 32263 }, { "epoch": 2.5434765471028773, "grad_norm": 0.4943104088306427, "learning_rate": 0.0005, "loss": 1.2131165266036987, "step": 32264 }, { "epoch": 2.5435553803705164, "grad_norm": 0.4994133412837982, "learning_rate": 0.0005, "loss": 1.389717698097229, "step": 32265 }, { "epoch": 2.543634213638155, "grad_norm": 0.442740797996521, "learning_rate": 0.0005, "loss": 1.1398000717163086, "step": 32266 }, { "epoch": 2.5437130469057943, "grad_norm": 0.5107588768005371, "learning_rate": 0.0005, "loss": 1.3304266929626465, "step": 32267 }, { "epoch": 2.5437918801734334, "grad_norm": 0.3797009289264679, "learning_rate": 0.0005, "loss": 1.2768311500549316, "step": 32268 }, { "epoch": 2.543870713441072, "grad_norm": 0.5587447881698608, "learning_rate": 0.0005, "loss": 1.2970592975616455, "step": 32269 }, { "epoch": 2.543949546708711, "grad_norm": 0.554263710975647, "learning_rate": 0.0005, "loss": 1.2718889713287354, "step": 32270 }, { "epoch": 2.54402837997635, "grad_norm": 0.4337683320045471, "learning_rate": 0.0005, "loss": 1.1871973276138306, "step": 32271 }, { "epoch": 2.544107213243989, "grad_norm": 0.4349570870399475, "learning_rate": 0.0005, "loss": 1.3817646503448486, "step": 32272 }, { "epoch": 2.544186046511628, "grad_norm": 0.6095805168151855, "learning_rate": 0.0005, "loss": 1.2631064653396606, "step": 32273 }, { "epoch": 2.544264879779267, "grad_norm": 0.4266885221004486, "learning_rate": 0.0005, "loss": 1.3769514560699463, "step": 32274 }, { "epoch": 2.5443437130469055, "grad_norm": 0.4745832085609436, "learning_rate": 0.0005, "loss": 1.2755844593048096, "step": 32275 }, { "epoch": 2.5444225463145447, "grad_norm": 0.7891713976860046, "learning_rate": 0.0005, "loss": 1.277892827987671, "step": 32276 }, { "epoch": 2.544501379582184, "grad_norm": 0.4581741392612457, "learning_rate": 0.0005, "loss": 1.306004524230957, "step": 32277 }, { "epoch": 2.5445802128498225, "grad_norm": 0.6075775027275085, "learning_rate": 0.0005, "loss": 1.4505085945129395, "step": 32278 }, { "epoch": 2.5446590461174616, "grad_norm": 0.6271390318870544, "learning_rate": 0.0005, "loss": 1.2153044939041138, "step": 32279 }, { "epoch": 2.5447378793851003, "grad_norm": 0.6026678681373596, "learning_rate": 0.0005, "loss": 1.2826383113861084, "step": 32280 }, { "epoch": 2.5448167126527395, "grad_norm": 0.5095512270927429, "learning_rate": 0.0005, "loss": 1.2931232452392578, "step": 32281 }, { "epoch": 2.5448955459203786, "grad_norm": 0.6204997897148132, "learning_rate": 0.0005, "loss": 1.3090661764144897, "step": 32282 }, { "epoch": 2.5449743791880173, "grad_norm": 0.5194874405860901, "learning_rate": 0.0005, "loss": 1.356231927871704, "step": 32283 }, { "epoch": 2.5450532124556564, "grad_norm": 0.5321183204650879, "learning_rate": 0.0005, "loss": 1.51966392993927, "step": 32284 }, { "epoch": 2.545132045723295, "grad_norm": 0.5396994948387146, "learning_rate": 0.0005, "loss": 1.4035370349884033, "step": 32285 }, { "epoch": 2.5452108789909342, "grad_norm": 0.5092082023620605, "learning_rate": 0.0005, "loss": 1.316500186920166, "step": 32286 }, { "epoch": 2.5452897122585734, "grad_norm": 0.4237542748451233, "learning_rate": 0.0005, "loss": 1.3889997005462646, "step": 32287 }, { "epoch": 2.545368545526212, "grad_norm": 0.48676884174346924, "learning_rate": 0.0005, "loss": 1.4770488739013672, "step": 32288 }, { "epoch": 2.5454473787938507, "grad_norm": 0.5126323103904724, "learning_rate": 0.0005, "loss": 1.293628215789795, "step": 32289 }, { "epoch": 2.54552621206149, "grad_norm": 0.501608669757843, "learning_rate": 0.0005, "loss": 1.1671199798583984, "step": 32290 }, { "epoch": 2.545605045329129, "grad_norm": 0.738455057144165, "learning_rate": 0.0005, "loss": 1.241267442703247, "step": 32291 }, { "epoch": 2.5456838785967677, "grad_norm": 0.463887482881546, "learning_rate": 0.0005, "loss": 1.4185333251953125, "step": 32292 }, { "epoch": 2.545762711864407, "grad_norm": 0.5494053959846497, "learning_rate": 0.0005, "loss": 1.302247166633606, "step": 32293 }, { "epoch": 2.5458415451320455, "grad_norm": 0.5082263946533203, "learning_rate": 0.0005, "loss": 1.4756970405578613, "step": 32294 }, { "epoch": 2.5459203783996847, "grad_norm": 0.4705933928489685, "learning_rate": 0.0005, "loss": 1.2110867500305176, "step": 32295 }, { "epoch": 2.545999211667324, "grad_norm": 0.4796079695224762, "learning_rate": 0.0005, "loss": 1.5135760307312012, "step": 32296 }, { "epoch": 2.5460780449349625, "grad_norm": 0.5432699918746948, "learning_rate": 0.0005, "loss": 1.3256785869598389, "step": 32297 }, { "epoch": 2.5461568782026016, "grad_norm": 0.4276389479637146, "learning_rate": 0.0005, "loss": 1.3748791217803955, "step": 32298 }, { "epoch": 2.5462357114702403, "grad_norm": 0.4419785439968109, "learning_rate": 0.0005, "loss": 1.353095293045044, "step": 32299 }, { "epoch": 2.5463145447378794, "grad_norm": 0.5097216367721558, "learning_rate": 0.0005, "loss": 1.4755096435546875, "step": 32300 }, { "epoch": 2.5463933780055186, "grad_norm": 0.49629855155944824, "learning_rate": 0.0005, "loss": 1.2796838283538818, "step": 32301 }, { "epoch": 2.5464722112731573, "grad_norm": 0.6370145678520203, "learning_rate": 0.0005, "loss": 1.4160685539245605, "step": 32302 }, { "epoch": 2.546551044540796, "grad_norm": 0.4979534447193146, "learning_rate": 0.0005, "loss": 1.3742783069610596, "step": 32303 }, { "epoch": 2.546629877808435, "grad_norm": 0.4446072280406952, "learning_rate": 0.0005, "loss": 1.3642499446868896, "step": 32304 }, { "epoch": 2.546708711076074, "grad_norm": 0.4403206706047058, "learning_rate": 0.0005, "loss": 1.1746251583099365, "step": 32305 }, { "epoch": 2.546787544343713, "grad_norm": 0.4675465524196625, "learning_rate": 0.0005, "loss": 1.1470286846160889, "step": 32306 }, { "epoch": 2.546866377611352, "grad_norm": 0.45413321256637573, "learning_rate": 0.0005, "loss": 1.2408356666564941, "step": 32307 }, { "epoch": 2.5469452108789907, "grad_norm": 0.5531846880912781, "learning_rate": 0.0005, "loss": 1.4295328855514526, "step": 32308 }, { "epoch": 2.54702404414663, "grad_norm": 0.4066711962223053, "learning_rate": 0.0005, "loss": 1.3403887748718262, "step": 32309 }, { "epoch": 2.547102877414269, "grad_norm": 0.4745272099971771, "learning_rate": 0.0005, "loss": 1.2381243705749512, "step": 32310 }, { "epoch": 2.5471817106819077, "grad_norm": 0.49294930696487427, "learning_rate": 0.0005, "loss": 1.4052057266235352, "step": 32311 }, { "epoch": 2.547260543949547, "grad_norm": 0.491546094417572, "learning_rate": 0.0005, "loss": 1.2888984680175781, "step": 32312 }, { "epoch": 2.5473393772171855, "grad_norm": 0.5177667737007141, "learning_rate": 0.0005, "loss": 1.4742611646652222, "step": 32313 }, { "epoch": 2.5474182104848246, "grad_norm": 0.3748553991317749, "learning_rate": 0.0005, "loss": 1.1823192834854126, "step": 32314 }, { "epoch": 2.5474970437524638, "grad_norm": 0.5867276787757874, "learning_rate": 0.0005, "loss": 1.1700180768966675, "step": 32315 }, { "epoch": 2.5475758770201025, "grad_norm": 0.4814147353172302, "learning_rate": 0.0005, "loss": 1.701122522354126, "step": 32316 }, { "epoch": 2.547654710287741, "grad_norm": 0.5398731827735901, "learning_rate": 0.0005, "loss": 1.4923752546310425, "step": 32317 }, { "epoch": 2.5477335435553803, "grad_norm": 0.5297912359237671, "learning_rate": 0.0005, "loss": 1.470055341720581, "step": 32318 }, { "epoch": 2.5478123768230194, "grad_norm": 0.46966925263404846, "learning_rate": 0.0005, "loss": 1.2766185998916626, "step": 32319 }, { "epoch": 2.547891210090658, "grad_norm": 0.4943172037601471, "learning_rate": 0.0005, "loss": 1.3613289594650269, "step": 32320 }, { "epoch": 2.5479700433582972, "grad_norm": 0.5519980788230896, "learning_rate": 0.0005, "loss": 1.4632291793823242, "step": 32321 }, { "epoch": 2.548048876625936, "grad_norm": 0.7056514620780945, "learning_rate": 0.0005, "loss": 1.2959378957748413, "step": 32322 }, { "epoch": 2.548127709893575, "grad_norm": 0.47727739810943604, "learning_rate": 0.0005, "loss": 1.2788043022155762, "step": 32323 }, { "epoch": 2.548206543161214, "grad_norm": 0.4777047336101532, "learning_rate": 0.0005, "loss": 1.504295825958252, "step": 32324 }, { "epoch": 2.548285376428853, "grad_norm": 6.7856316566467285, "learning_rate": 0.0005, "loss": 1.3654154539108276, "step": 32325 }, { "epoch": 2.548364209696492, "grad_norm": 0.4411064088344574, "learning_rate": 0.0005, "loss": 1.3444573879241943, "step": 32326 }, { "epoch": 2.5484430429641307, "grad_norm": 0.47263914346694946, "learning_rate": 0.0005, "loss": 1.5034619569778442, "step": 32327 }, { "epoch": 2.54852187623177, "grad_norm": 0.4356009364128113, "learning_rate": 0.0005, "loss": 1.2868454456329346, "step": 32328 }, { "epoch": 2.548600709499409, "grad_norm": 0.5196070075035095, "learning_rate": 0.0005, "loss": 1.3631813526153564, "step": 32329 }, { "epoch": 2.5486795427670477, "grad_norm": 0.5559109449386597, "learning_rate": 0.0005, "loss": 1.2386162281036377, "step": 32330 }, { "epoch": 2.548758376034687, "grad_norm": 0.47031059861183167, "learning_rate": 0.0005, "loss": 1.3906419277191162, "step": 32331 }, { "epoch": 2.5488372093023255, "grad_norm": 0.5097913146018982, "learning_rate": 0.0005, "loss": 1.192816972732544, "step": 32332 }, { "epoch": 2.5489160425699646, "grad_norm": 0.5593259334564209, "learning_rate": 0.0005, "loss": 1.3429659605026245, "step": 32333 }, { "epoch": 2.5489948758376033, "grad_norm": 0.5629101395606995, "learning_rate": 0.0005, "loss": 1.4614696502685547, "step": 32334 }, { "epoch": 2.5490737091052424, "grad_norm": 0.5211344361305237, "learning_rate": 0.0005, "loss": 1.1312761306762695, "step": 32335 }, { "epoch": 2.549152542372881, "grad_norm": 0.520366370677948, "learning_rate": 0.0005, "loss": 1.4809242486953735, "step": 32336 }, { "epoch": 2.5492313756405203, "grad_norm": 0.4364528954029083, "learning_rate": 0.0005, "loss": 1.3436975479125977, "step": 32337 }, { "epoch": 2.5493102089081594, "grad_norm": 0.48548945784568787, "learning_rate": 0.0005, "loss": 1.5129225254058838, "step": 32338 }, { "epoch": 2.549389042175798, "grad_norm": 0.43170908093452454, "learning_rate": 0.0005, "loss": 1.223903775215149, "step": 32339 }, { "epoch": 2.5494678754434372, "grad_norm": 0.38540658354759216, "learning_rate": 0.0005, "loss": 1.1042139530181885, "step": 32340 }, { "epoch": 2.549546708711076, "grad_norm": 0.518444299697876, "learning_rate": 0.0005, "loss": 1.4642009735107422, "step": 32341 }, { "epoch": 2.549625541978715, "grad_norm": 0.5128554701805115, "learning_rate": 0.0005, "loss": 1.171647310256958, "step": 32342 }, { "epoch": 2.549704375246354, "grad_norm": 0.3781772255897522, "learning_rate": 0.0005, "loss": 1.1383273601531982, "step": 32343 }, { "epoch": 2.549783208513993, "grad_norm": 0.4679391384124756, "learning_rate": 0.0005, "loss": 1.364978313446045, "step": 32344 }, { "epoch": 2.549862041781632, "grad_norm": 0.6100770831108093, "learning_rate": 0.0005, "loss": 1.5790072679519653, "step": 32345 }, { "epoch": 2.5499408750492707, "grad_norm": 0.4433286786079407, "learning_rate": 0.0005, "loss": 1.1518323421478271, "step": 32346 }, { "epoch": 2.55001970831691, "grad_norm": 0.473690003156662, "learning_rate": 0.0005, "loss": 1.5061489343643188, "step": 32347 }, { "epoch": 2.5500985415845485, "grad_norm": 0.5000922679901123, "learning_rate": 0.0005, "loss": 1.4311854839324951, "step": 32348 }, { "epoch": 2.5501773748521877, "grad_norm": 0.5964311957359314, "learning_rate": 0.0005, "loss": 1.4562104940414429, "step": 32349 }, { "epoch": 2.5502562081198263, "grad_norm": 0.5228803753852844, "learning_rate": 0.0005, "loss": 1.2683894634246826, "step": 32350 }, { "epoch": 2.5503350413874655, "grad_norm": 0.5365608930587769, "learning_rate": 0.0005, "loss": 1.3979274034500122, "step": 32351 }, { "epoch": 2.5504138746551046, "grad_norm": 0.4347136616706848, "learning_rate": 0.0005, "loss": 1.1302454471588135, "step": 32352 }, { "epoch": 2.5504927079227433, "grad_norm": 0.4858206808567047, "learning_rate": 0.0005, "loss": 1.1982581615447998, "step": 32353 }, { "epoch": 2.5505715411903824, "grad_norm": 0.47022897005081177, "learning_rate": 0.0005, "loss": 1.226151466369629, "step": 32354 }, { "epoch": 2.550650374458021, "grad_norm": 0.4293822646141052, "learning_rate": 0.0005, "loss": 1.337437629699707, "step": 32355 }, { "epoch": 2.5507292077256603, "grad_norm": 0.5089747309684753, "learning_rate": 0.0005, "loss": 1.3578380346298218, "step": 32356 }, { "epoch": 2.5508080409932994, "grad_norm": 0.5140353441238403, "learning_rate": 0.0005, "loss": 1.3435022830963135, "step": 32357 }, { "epoch": 2.550886874260938, "grad_norm": 0.8221498727798462, "learning_rate": 0.0005, "loss": 1.167093276977539, "step": 32358 }, { "epoch": 2.550965707528577, "grad_norm": 0.5538358688354492, "learning_rate": 0.0005, "loss": 1.3521406650543213, "step": 32359 }, { "epoch": 2.551044540796216, "grad_norm": 0.5925791263580322, "learning_rate": 0.0005, "loss": 1.4772086143493652, "step": 32360 }, { "epoch": 2.551123374063855, "grad_norm": 0.5053110122680664, "learning_rate": 0.0005, "loss": 1.1116423606872559, "step": 32361 }, { "epoch": 2.551202207331494, "grad_norm": 0.5909560322761536, "learning_rate": 0.0005, "loss": 1.4028923511505127, "step": 32362 }, { "epoch": 2.551281040599133, "grad_norm": 0.5599812269210815, "learning_rate": 0.0005, "loss": 1.3701198101043701, "step": 32363 }, { "epoch": 2.5513598738667715, "grad_norm": 0.5184825658798218, "learning_rate": 0.0005, "loss": 1.4247512817382812, "step": 32364 }, { "epoch": 2.5514387071344107, "grad_norm": 0.5209051966667175, "learning_rate": 0.0005, "loss": 1.5492414236068726, "step": 32365 }, { "epoch": 2.55151754040205, "grad_norm": 0.48375535011291504, "learning_rate": 0.0005, "loss": 1.2786815166473389, "step": 32366 }, { "epoch": 2.5515963736696885, "grad_norm": 0.4649631679058075, "learning_rate": 0.0005, "loss": 1.3444864749908447, "step": 32367 }, { "epoch": 2.5516752069373276, "grad_norm": 0.5098756551742554, "learning_rate": 0.0005, "loss": 1.1803430318832397, "step": 32368 }, { "epoch": 2.5517540402049663, "grad_norm": 0.4945155084133148, "learning_rate": 0.0005, "loss": 1.25588858127594, "step": 32369 }, { "epoch": 2.5518328734726055, "grad_norm": 0.5161387920379639, "learning_rate": 0.0005, "loss": 1.2896004915237427, "step": 32370 }, { "epoch": 2.5519117067402446, "grad_norm": 0.6568662524223328, "learning_rate": 0.0005, "loss": 1.3193867206573486, "step": 32371 }, { "epoch": 2.5519905400078833, "grad_norm": 0.449405699968338, "learning_rate": 0.0005, "loss": 1.3131945133209229, "step": 32372 }, { "epoch": 2.5520693732755224, "grad_norm": 0.5339359045028687, "learning_rate": 0.0005, "loss": 1.3449136018753052, "step": 32373 }, { "epoch": 2.552148206543161, "grad_norm": 0.4476732611656189, "learning_rate": 0.0005, "loss": 1.246968150138855, "step": 32374 }, { "epoch": 2.5522270398108002, "grad_norm": 0.45997872948646545, "learning_rate": 0.0005, "loss": 1.3183341026306152, "step": 32375 }, { "epoch": 2.5523058730784394, "grad_norm": 0.5590380430221558, "learning_rate": 0.0005, "loss": 1.4559235572814941, "step": 32376 }, { "epoch": 2.552384706346078, "grad_norm": 0.4817538559436798, "learning_rate": 0.0005, "loss": 1.2897225618362427, "step": 32377 }, { "epoch": 2.5524635396137167, "grad_norm": 0.5636617541313171, "learning_rate": 0.0005, "loss": 1.3922643661499023, "step": 32378 }, { "epoch": 2.552542372881356, "grad_norm": 0.4062926471233368, "learning_rate": 0.0005, "loss": 1.2602107524871826, "step": 32379 }, { "epoch": 2.552621206148995, "grad_norm": 0.507070779800415, "learning_rate": 0.0005, "loss": 1.5102591514587402, "step": 32380 }, { "epoch": 2.5527000394166337, "grad_norm": 1.0674117803573608, "learning_rate": 0.0005, "loss": 1.4615309238433838, "step": 32381 }, { "epoch": 2.552778872684273, "grad_norm": 0.49346792697906494, "learning_rate": 0.0005, "loss": 1.2171640396118164, "step": 32382 }, { "epoch": 2.5528577059519115, "grad_norm": 0.46739938855171204, "learning_rate": 0.0005, "loss": 1.1554417610168457, "step": 32383 }, { "epoch": 2.5529365392195507, "grad_norm": 0.6856618523597717, "learning_rate": 0.0005, "loss": 1.3271037340164185, "step": 32384 }, { "epoch": 2.55301537248719, "grad_norm": 0.4614388942718506, "learning_rate": 0.0005, "loss": 1.2901554107666016, "step": 32385 }, { "epoch": 2.5530942057548285, "grad_norm": 0.4814038872718811, "learning_rate": 0.0005, "loss": 1.3368868827819824, "step": 32386 }, { "epoch": 2.5531730390224676, "grad_norm": 0.5850225687026978, "learning_rate": 0.0005, "loss": 1.6578876972198486, "step": 32387 }, { "epoch": 2.5532518722901063, "grad_norm": 0.5226104259490967, "learning_rate": 0.0005, "loss": 1.1845835447311401, "step": 32388 }, { "epoch": 2.5533307055577454, "grad_norm": 0.47019636631011963, "learning_rate": 0.0005, "loss": 1.2638623714447021, "step": 32389 }, { "epoch": 2.5534095388253846, "grad_norm": 0.49735015630722046, "learning_rate": 0.0005, "loss": 1.2509506940841675, "step": 32390 }, { "epoch": 2.5534883720930233, "grad_norm": 0.4286350607872009, "learning_rate": 0.0005, "loss": 1.282470941543579, "step": 32391 }, { "epoch": 2.553567205360662, "grad_norm": 0.6220224499702454, "learning_rate": 0.0005, "loss": 1.5719614028930664, "step": 32392 }, { "epoch": 2.553646038628301, "grad_norm": 0.516957700252533, "learning_rate": 0.0005, "loss": 1.3455229997634888, "step": 32393 }, { "epoch": 2.55372487189594, "grad_norm": 0.5132774114608765, "learning_rate": 0.0005, "loss": 1.5096757411956787, "step": 32394 }, { "epoch": 2.553803705163579, "grad_norm": 0.4979429244995117, "learning_rate": 0.0005, "loss": 1.2989106178283691, "step": 32395 }, { "epoch": 2.553882538431218, "grad_norm": 0.549305260181427, "learning_rate": 0.0005, "loss": 1.218817949295044, "step": 32396 }, { "epoch": 2.5539613716988567, "grad_norm": 0.6117478609085083, "learning_rate": 0.0005, "loss": 1.427655816078186, "step": 32397 }, { "epoch": 2.554040204966496, "grad_norm": 0.6308988928794861, "learning_rate": 0.0005, "loss": 1.423842430114746, "step": 32398 }, { "epoch": 2.554119038234135, "grad_norm": 0.5399843454360962, "learning_rate": 0.0005, "loss": 1.1295714378356934, "step": 32399 }, { "epoch": 2.5541978715017737, "grad_norm": 0.5586810111999512, "learning_rate": 0.0005, "loss": 1.407036542892456, "step": 32400 }, { "epoch": 2.554276704769413, "grad_norm": 0.4840601980686188, "learning_rate": 0.0005, "loss": 1.294745683670044, "step": 32401 }, { "epoch": 2.5543555380370515, "grad_norm": 0.573959231376648, "learning_rate": 0.0005, "loss": 1.3628976345062256, "step": 32402 }, { "epoch": 2.5544343713046906, "grad_norm": 0.5455620288848877, "learning_rate": 0.0005, "loss": 1.5283479690551758, "step": 32403 }, { "epoch": 2.5545132045723298, "grad_norm": 0.5063451528549194, "learning_rate": 0.0005, "loss": 1.34525728225708, "step": 32404 }, { "epoch": 2.5545920378399685, "grad_norm": 0.542902946472168, "learning_rate": 0.0005, "loss": 1.3497117757797241, "step": 32405 }, { "epoch": 2.554670871107607, "grad_norm": 0.48419189453125, "learning_rate": 0.0005, "loss": 1.1511030197143555, "step": 32406 }, { "epoch": 2.5547497043752463, "grad_norm": 0.4511726200580597, "learning_rate": 0.0005, "loss": 1.2576987743377686, "step": 32407 }, { "epoch": 2.5548285376428854, "grad_norm": 0.5401514172554016, "learning_rate": 0.0005, "loss": 1.4365339279174805, "step": 32408 }, { "epoch": 2.554907370910524, "grad_norm": 0.42364659905433655, "learning_rate": 0.0005, "loss": 1.136551856994629, "step": 32409 }, { "epoch": 2.5549862041781632, "grad_norm": 0.4243012070655823, "learning_rate": 0.0005, "loss": 1.3317577838897705, "step": 32410 }, { "epoch": 2.555065037445802, "grad_norm": 0.552916944026947, "learning_rate": 0.0005, "loss": 1.3545030355453491, "step": 32411 }, { "epoch": 2.555143870713441, "grad_norm": 0.6230836510658264, "learning_rate": 0.0005, "loss": 1.630906105041504, "step": 32412 }, { "epoch": 2.55522270398108, "grad_norm": 0.5033361911773682, "learning_rate": 0.0005, "loss": 1.2431293725967407, "step": 32413 }, { "epoch": 2.555301537248719, "grad_norm": 0.4834977686405182, "learning_rate": 0.0005, "loss": 1.1963715553283691, "step": 32414 }, { "epoch": 2.555380370516358, "grad_norm": 0.4654691219329834, "learning_rate": 0.0005, "loss": 1.1580302715301514, "step": 32415 }, { "epoch": 2.5554592037839967, "grad_norm": 0.47103530168533325, "learning_rate": 0.0005, "loss": 1.2643338441848755, "step": 32416 }, { "epoch": 2.555538037051636, "grad_norm": 0.44452181458473206, "learning_rate": 0.0005, "loss": 1.3762948513031006, "step": 32417 }, { "epoch": 2.555616870319275, "grad_norm": 0.46411794424057007, "learning_rate": 0.0005, "loss": 1.2705873250961304, "step": 32418 }, { "epoch": 2.5556957035869137, "grad_norm": 0.4373281002044678, "learning_rate": 0.0005, "loss": 1.1906814575195312, "step": 32419 }, { "epoch": 2.5557745368545524, "grad_norm": 0.6617822647094727, "learning_rate": 0.0005, "loss": 1.3261065483093262, "step": 32420 }, { "epoch": 2.5558533701221915, "grad_norm": 0.5062958598136902, "learning_rate": 0.0005, "loss": 1.471867322921753, "step": 32421 }, { "epoch": 2.5559322033898306, "grad_norm": 0.440489262342453, "learning_rate": 0.0005, "loss": 1.296342372894287, "step": 32422 }, { "epoch": 2.5560110366574693, "grad_norm": 0.73605877161026, "learning_rate": 0.0005, "loss": 1.0947377681732178, "step": 32423 }, { "epoch": 2.5560898699251084, "grad_norm": 0.5425975322723389, "learning_rate": 0.0005, "loss": 1.3659491539001465, "step": 32424 }, { "epoch": 2.556168703192747, "grad_norm": 0.5118356347084045, "learning_rate": 0.0005, "loss": 1.2686824798583984, "step": 32425 }, { "epoch": 2.5562475364603863, "grad_norm": 0.5538749098777771, "learning_rate": 0.0005, "loss": 1.3179208040237427, "step": 32426 }, { "epoch": 2.5563263697280254, "grad_norm": 0.4753071665763855, "learning_rate": 0.0005, "loss": 1.1620278358459473, "step": 32427 }, { "epoch": 2.556405202995664, "grad_norm": 0.5007449388504028, "learning_rate": 0.0005, "loss": 1.2189075946807861, "step": 32428 }, { "epoch": 2.5564840362633032, "grad_norm": 0.48427674174308777, "learning_rate": 0.0005, "loss": 1.432796835899353, "step": 32429 }, { "epoch": 2.556562869530942, "grad_norm": 0.5066666007041931, "learning_rate": 0.0005, "loss": 1.1975935697555542, "step": 32430 }, { "epoch": 2.556641702798581, "grad_norm": 0.6043192148208618, "learning_rate": 0.0005, "loss": 1.2166703939437866, "step": 32431 }, { "epoch": 2.55672053606622, "grad_norm": 0.5794361233711243, "learning_rate": 0.0005, "loss": 1.4032481908798218, "step": 32432 }, { "epoch": 2.556799369333859, "grad_norm": 0.5904355645179749, "learning_rate": 0.0005, "loss": 1.3932820558547974, "step": 32433 }, { "epoch": 2.5568782026014976, "grad_norm": 0.4879661798477173, "learning_rate": 0.0005, "loss": 1.1951494216918945, "step": 32434 }, { "epoch": 2.5569570358691367, "grad_norm": 0.5303611755371094, "learning_rate": 0.0005, "loss": 1.5847111940383911, "step": 32435 }, { "epoch": 2.557035869136776, "grad_norm": 0.6491779685020447, "learning_rate": 0.0005, "loss": 1.4168201684951782, "step": 32436 }, { "epoch": 2.5571147024044145, "grad_norm": 0.5786977410316467, "learning_rate": 0.0005, "loss": 1.3711926937103271, "step": 32437 }, { "epoch": 2.5571935356720537, "grad_norm": 0.4762571156024933, "learning_rate": 0.0005, "loss": 1.412589430809021, "step": 32438 }, { "epoch": 2.5572723689396923, "grad_norm": 0.5023295879364014, "learning_rate": 0.0005, "loss": 1.4214966297149658, "step": 32439 }, { "epoch": 2.5573512022073315, "grad_norm": 0.4110919237136841, "learning_rate": 0.0005, "loss": 1.2098984718322754, "step": 32440 }, { "epoch": 2.5574300354749706, "grad_norm": 0.441127210855484, "learning_rate": 0.0005, "loss": 1.283158302307129, "step": 32441 }, { "epoch": 2.5575088687426093, "grad_norm": 1.1784625053405762, "learning_rate": 0.0005, "loss": 1.5353796482086182, "step": 32442 }, { "epoch": 2.5575877020102484, "grad_norm": 0.5519596338272095, "learning_rate": 0.0005, "loss": 1.4750289916992188, "step": 32443 }, { "epoch": 2.557666535277887, "grad_norm": 0.5284465551376343, "learning_rate": 0.0005, "loss": 1.1986886262893677, "step": 32444 }, { "epoch": 2.5577453685455263, "grad_norm": 0.4656965434551239, "learning_rate": 0.0005, "loss": 1.2941628694534302, "step": 32445 }, { "epoch": 2.5578242018131654, "grad_norm": 0.6212873458862305, "learning_rate": 0.0005, "loss": 1.3776271343231201, "step": 32446 }, { "epoch": 2.557903035080804, "grad_norm": 0.4474361836910248, "learning_rate": 0.0005, "loss": 1.2648974657058716, "step": 32447 }, { "epoch": 2.5579818683484428, "grad_norm": 0.47306379675865173, "learning_rate": 0.0005, "loss": 1.5537877082824707, "step": 32448 }, { "epoch": 2.558060701616082, "grad_norm": 0.46390390396118164, "learning_rate": 0.0005, "loss": 1.320211410522461, "step": 32449 }, { "epoch": 2.558139534883721, "grad_norm": 0.4899280369281769, "learning_rate": 0.0005, "loss": 1.4235718250274658, "step": 32450 }, { "epoch": 2.5582183681513597, "grad_norm": 0.46100926399230957, "learning_rate": 0.0005, "loss": 1.3351681232452393, "step": 32451 }, { "epoch": 2.558297201418999, "grad_norm": 0.5044904351234436, "learning_rate": 0.0005, "loss": 1.4681377410888672, "step": 32452 }, { "epoch": 2.5583760346866375, "grad_norm": 0.5316295027732849, "learning_rate": 0.0005, "loss": 1.3984992504119873, "step": 32453 }, { "epoch": 2.5584548679542767, "grad_norm": 0.6310490965843201, "learning_rate": 0.0005, "loss": 1.3667657375335693, "step": 32454 }, { "epoch": 2.558533701221916, "grad_norm": 0.48534366488456726, "learning_rate": 0.0005, "loss": 1.4026470184326172, "step": 32455 }, { "epoch": 2.5586125344895545, "grad_norm": 0.5800508856773376, "learning_rate": 0.0005, "loss": 1.3289417028427124, "step": 32456 }, { "epoch": 2.5586913677571936, "grad_norm": 0.5317406058311462, "learning_rate": 0.0005, "loss": 1.367261528968811, "step": 32457 }, { "epoch": 2.5587702010248323, "grad_norm": 0.5175880789756775, "learning_rate": 0.0005, "loss": 1.3371095657348633, "step": 32458 }, { "epoch": 2.5588490342924715, "grad_norm": 0.4962869882583618, "learning_rate": 0.0005, "loss": 1.2681427001953125, "step": 32459 }, { "epoch": 2.5589278675601106, "grad_norm": 0.5441069602966309, "learning_rate": 0.0005, "loss": 1.3229402303695679, "step": 32460 }, { "epoch": 2.5590067008277493, "grad_norm": 0.4905514419078827, "learning_rate": 0.0005, "loss": 1.3575934171676636, "step": 32461 }, { "epoch": 2.5590855340953884, "grad_norm": 0.553943932056427, "learning_rate": 0.0005, "loss": 1.4013853073120117, "step": 32462 }, { "epoch": 2.559164367363027, "grad_norm": 0.5081813931465149, "learning_rate": 0.0005, "loss": 1.4251010417938232, "step": 32463 }, { "epoch": 2.5592432006306662, "grad_norm": 0.631459653377533, "learning_rate": 0.0005, "loss": 1.4999134540557861, "step": 32464 }, { "epoch": 2.559322033898305, "grad_norm": 0.48507004976272583, "learning_rate": 0.0005, "loss": 1.3055102825164795, "step": 32465 }, { "epoch": 2.559400867165944, "grad_norm": 0.4445476531982422, "learning_rate": 0.0005, "loss": 1.3253288269042969, "step": 32466 }, { "epoch": 2.5594797004335827, "grad_norm": 0.4635987877845764, "learning_rate": 0.0005, "loss": 1.2968199253082275, "step": 32467 }, { "epoch": 2.559558533701222, "grad_norm": 0.5334172248840332, "learning_rate": 0.0005, "loss": 1.46779465675354, "step": 32468 }, { "epoch": 2.559637366968861, "grad_norm": 0.42041465640068054, "learning_rate": 0.0005, "loss": 1.1814873218536377, "step": 32469 }, { "epoch": 2.5597162002364997, "grad_norm": 0.4904279112815857, "learning_rate": 0.0005, "loss": 1.2972387075424194, "step": 32470 }, { "epoch": 2.559795033504139, "grad_norm": 0.5729037523269653, "learning_rate": 0.0005, "loss": 1.4330198764801025, "step": 32471 }, { "epoch": 2.5598738667717775, "grad_norm": 0.6327759027481079, "learning_rate": 0.0005, "loss": 1.5570766925811768, "step": 32472 }, { "epoch": 2.5599527000394167, "grad_norm": 0.41920343041419983, "learning_rate": 0.0005, "loss": 1.3552132844924927, "step": 32473 }, { "epoch": 2.560031533307056, "grad_norm": 0.40485140681266785, "learning_rate": 0.0005, "loss": 1.2531757354736328, "step": 32474 }, { "epoch": 2.5601103665746945, "grad_norm": 0.5738627910614014, "learning_rate": 0.0005, "loss": 1.3178646564483643, "step": 32475 }, { "epoch": 2.5601891998423336, "grad_norm": 0.5646225214004517, "learning_rate": 0.0005, "loss": 1.4690358638763428, "step": 32476 }, { "epoch": 2.5602680331099723, "grad_norm": 0.512671947479248, "learning_rate": 0.0005, "loss": 1.3152472972869873, "step": 32477 }, { "epoch": 2.5603468663776114, "grad_norm": 0.5050573945045471, "learning_rate": 0.0005, "loss": 1.1983493566513062, "step": 32478 }, { "epoch": 2.56042569964525, "grad_norm": 0.5629593133926392, "learning_rate": 0.0005, "loss": 1.2679985761642456, "step": 32479 }, { "epoch": 2.5605045329128893, "grad_norm": 0.505953848361969, "learning_rate": 0.0005, "loss": 1.3361669778823853, "step": 32480 }, { "epoch": 2.560583366180528, "grad_norm": 0.5325884222984314, "learning_rate": 0.0005, "loss": 1.2816270589828491, "step": 32481 }, { "epoch": 2.560662199448167, "grad_norm": 0.4246916174888611, "learning_rate": 0.0005, "loss": 1.2953635454177856, "step": 32482 }, { "epoch": 2.560741032715806, "grad_norm": 0.4531642496585846, "learning_rate": 0.0005, "loss": 1.3703608512878418, "step": 32483 }, { "epoch": 2.560819865983445, "grad_norm": 0.5863823890686035, "learning_rate": 0.0005, "loss": 1.3730274438858032, "step": 32484 }, { "epoch": 2.560898699251084, "grad_norm": 0.47702667117118835, "learning_rate": 0.0005, "loss": 1.332963228225708, "step": 32485 }, { "epoch": 2.5609775325187227, "grad_norm": 0.38934504985809326, "learning_rate": 0.0005, "loss": 1.2475860118865967, "step": 32486 }, { "epoch": 2.561056365786362, "grad_norm": 0.36050403118133545, "learning_rate": 0.0005, "loss": 1.1193079948425293, "step": 32487 }, { "epoch": 2.561135199054001, "grad_norm": 0.4988134205341339, "learning_rate": 0.0005, "loss": 1.2839457988739014, "step": 32488 }, { "epoch": 2.5612140323216397, "grad_norm": 0.46224066615104675, "learning_rate": 0.0005, "loss": 1.3348939418792725, "step": 32489 }, { "epoch": 2.561292865589279, "grad_norm": 0.4985315203666687, "learning_rate": 0.0005, "loss": 1.345533847808838, "step": 32490 }, { "epoch": 2.5613716988569175, "grad_norm": 0.44583919644355774, "learning_rate": 0.0005, "loss": 1.3739680051803589, "step": 32491 }, { "epoch": 2.5614505321245566, "grad_norm": 0.3518903851509094, "learning_rate": 0.0005, "loss": 1.1653389930725098, "step": 32492 }, { "epoch": 2.561529365392196, "grad_norm": 0.5547561645507812, "learning_rate": 0.0005, "loss": 1.2718360424041748, "step": 32493 }, { "epoch": 2.5616081986598345, "grad_norm": 0.5140753984451294, "learning_rate": 0.0005, "loss": 1.410666584968567, "step": 32494 }, { "epoch": 2.561687031927473, "grad_norm": 0.5391907691955566, "learning_rate": 0.0005, "loss": 1.4589500427246094, "step": 32495 }, { "epoch": 2.5617658651951123, "grad_norm": 0.638150691986084, "learning_rate": 0.0005, "loss": 1.4212913513183594, "step": 32496 }, { "epoch": 2.5618446984627514, "grad_norm": 0.4280523657798767, "learning_rate": 0.0005, "loss": 1.321812629699707, "step": 32497 }, { "epoch": 2.56192353173039, "grad_norm": 0.5022722482681274, "learning_rate": 0.0005, "loss": 1.3224599361419678, "step": 32498 }, { "epoch": 2.5620023649980292, "grad_norm": 0.5181060433387756, "learning_rate": 0.0005, "loss": 1.3011929988861084, "step": 32499 }, { "epoch": 2.562081198265668, "grad_norm": 0.41581958532333374, "learning_rate": 0.0005, "loss": 1.1926562786102295, "step": 32500 }, { "epoch": 2.562160031533307, "grad_norm": 0.5820244550704956, "learning_rate": 0.0005, "loss": 1.3646752834320068, "step": 32501 }, { "epoch": 2.562238864800946, "grad_norm": 0.5074565410614014, "learning_rate": 0.0005, "loss": 1.337892770767212, "step": 32502 }, { "epoch": 2.562317698068585, "grad_norm": 0.42317381501197815, "learning_rate": 0.0005, "loss": 1.1233460903167725, "step": 32503 }, { "epoch": 2.562396531336224, "grad_norm": 0.5825227499008179, "learning_rate": 0.0005, "loss": 1.3275657892227173, "step": 32504 }, { "epoch": 2.5624753646038627, "grad_norm": 0.5349224209785461, "learning_rate": 0.0005, "loss": 1.2524056434631348, "step": 32505 }, { "epoch": 2.562554197871502, "grad_norm": 0.4504890441894531, "learning_rate": 0.0005, "loss": 1.3161871433258057, "step": 32506 }, { "epoch": 2.562633031139141, "grad_norm": 0.4166088402271271, "learning_rate": 0.0005, "loss": 1.3160490989685059, "step": 32507 }, { "epoch": 2.5627118644067797, "grad_norm": 0.4630478322505951, "learning_rate": 0.0005, "loss": 1.4232499599456787, "step": 32508 }, { "epoch": 2.5627906976744184, "grad_norm": 0.5507692098617554, "learning_rate": 0.0005, "loss": 1.36110520362854, "step": 32509 }, { "epoch": 2.5628695309420575, "grad_norm": 0.5089375376701355, "learning_rate": 0.0005, "loss": 1.4178690910339355, "step": 32510 }, { "epoch": 2.5629483642096966, "grad_norm": 0.518768846988678, "learning_rate": 0.0005, "loss": 1.3701856136322021, "step": 32511 }, { "epoch": 2.5630271974773353, "grad_norm": 0.48649004101753235, "learning_rate": 0.0005, "loss": 1.2815204858779907, "step": 32512 }, { "epoch": 2.5631060307449745, "grad_norm": 0.5611616373062134, "learning_rate": 0.0005, "loss": 1.2254796028137207, "step": 32513 }, { "epoch": 2.563184864012613, "grad_norm": 0.5176839828491211, "learning_rate": 0.0005, "loss": 1.4402836561203003, "step": 32514 }, { "epoch": 2.5632636972802523, "grad_norm": 0.5422201156616211, "learning_rate": 0.0005, "loss": 1.398719072341919, "step": 32515 }, { "epoch": 2.5633425305478914, "grad_norm": 0.3828018605709076, "learning_rate": 0.0005, "loss": 1.1968584060668945, "step": 32516 }, { "epoch": 2.56342136381553, "grad_norm": 0.530170738697052, "learning_rate": 0.0005, "loss": 1.3623945713043213, "step": 32517 }, { "epoch": 2.5635001970831692, "grad_norm": 0.4549310505390167, "learning_rate": 0.0005, "loss": 1.207369327545166, "step": 32518 }, { "epoch": 2.563579030350808, "grad_norm": 0.6394408941268921, "learning_rate": 0.0005, "loss": 1.4191852807998657, "step": 32519 }, { "epoch": 2.563657863618447, "grad_norm": 0.43876418471336365, "learning_rate": 0.0005, "loss": 1.1605514287948608, "step": 32520 }, { "epoch": 2.563736696886086, "grad_norm": 0.4386845827102661, "learning_rate": 0.0005, "loss": 1.2274839878082275, "step": 32521 }, { "epoch": 2.563815530153725, "grad_norm": 0.512717604637146, "learning_rate": 0.0005, "loss": 1.2878237962722778, "step": 32522 }, { "epoch": 2.5638943634213636, "grad_norm": 0.48033615946769714, "learning_rate": 0.0005, "loss": 1.27629554271698, "step": 32523 }, { "epoch": 2.5639731966890027, "grad_norm": 0.48025885224342346, "learning_rate": 0.0005, "loss": 1.242447018623352, "step": 32524 }, { "epoch": 2.564052029956642, "grad_norm": 0.6044671535491943, "learning_rate": 0.0005, "loss": 1.2892756462097168, "step": 32525 }, { "epoch": 2.5641308632242805, "grad_norm": 0.47842517495155334, "learning_rate": 0.0005, "loss": 1.2391066551208496, "step": 32526 }, { "epoch": 2.5642096964919197, "grad_norm": 0.5541859865188599, "learning_rate": 0.0005, "loss": 1.5463911294937134, "step": 32527 }, { "epoch": 2.5642885297595583, "grad_norm": 1.6018140316009521, "learning_rate": 0.0005, "loss": 1.2723811864852905, "step": 32528 }, { "epoch": 2.5643673630271975, "grad_norm": 0.5331453680992126, "learning_rate": 0.0005, "loss": 1.3913768529891968, "step": 32529 }, { "epoch": 2.5644461962948366, "grad_norm": 0.6400218605995178, "learning_rate": 0.0005, "loss": 1.3271856307983398, "step": 32530 }, { "epoch": 2.5645250295624753, "grad_norm": 0.6707751154899597, "learning_rate": 0.0005, "loss": 1.289570689201355, "step": 32531 }, { "epoch": 2.5646038628301144, "grad_norm": 0.5096116662025452, "learning_rate": 0.0005, "loss": 1.3028619289398193, "step": 32532 }, { "epoch": 2.564682696097753, "grad_norm": 0.5975431203842163, "learning_rate": 0.0005, "loss": 1.3585286140441895, "step": 32533 }, { "epoch": 2.5647615293653923, "grad_norm": 0.507169246673584, "learning_rate": 0.0005, "loss": 1.4451195001602173, "step": 32534 }, { "epoch": 2.5648403626330314, "grad_norm": 0.48108813166618347, "learning_rate": 0.0005, "loss": 1.429686188697815, "step": 32535 }, { "epoch": 2.56491919590067, "grad_norm": 0.5840972661972046, "learning_rate": 0.0005, "loss": 1.42801833152771, "step": 32536 }, { "epoch": 2.5649980291683088, "grad_norm": 0.48424032330513, "learning_rate": 0.0005, "loss": 1.4773979187011719, "step": 32537 }, { "epoch": 2.565076862435948, "grad_norm": 0.4752267897129059, "learning_rate": 0.0005, "loss": 1.262638807296753, "step": 32538 }, { "epoch": 2.565155695703587, "grad_norm": 0.4706689417362213, "learning_rate": 0.0005, "loss": 1.2122989892959595, "step": 32539 }, { "epoch": 2.5652345289712257, "grad_norm": 0.4950271248817444, "learning_rate": 0.0005, "loss": 1.6367924213409424, "step": 32540 }, { "epoch": 2.565313362238865, "grad_norm": 0.5763810873031616, "learning_rate": 0.0005, "loss": 1.4402661323547363, "step": 32541 }, { "epoch": 2.5653921955065035, "grad_norm": 0.5140224099159241, "learning_rate": 0.0005, "loss": 1.4079420566558838, "step": 32542 }, { "epoch": 2.5654710287741427, "grad_norm": 0.4953095316886902, "learning_rate": 0.0005, "loss": 1.2875816822052002, "step": 32543 }, { "epoch": 2.565549862041782, "grad_norm": 0.512090265750885, "learning_rate": 0.0005, "loss": 1.336827039718628, "step": 32544 }, { "epoch": 2.5656286953094205, "grad_norm": 0.49884551763534546, "learning_rate": 0.0005, "loss": 1.4328254461288452, "step": 32545 }, { "epoch": 2.5657075285770596, "grad_norm": 0.4752134680747986, "learning_rate": 0.0005, "loss": 1.380673885345459, "step": 32546 }, { "epoch": 2.5657863618446983, "grad_norm": 0.5474594831466675, "learning_rate": 0.0005, "loss": 1.1058964729309082, "step": 32547 }, { "epoch": 2.5658651951123375, "grad_norm": 0.6009147763252258, "learning_rate": 0.0005, "loss": 1.4631893634796143, "step": 32548 }, { "epoch": 2.5659440283799766, "grad_norm": 0.609139621257782, "learning_rate": 0.0005, "loss": 1.2769412994384766, "step": 32549 }, { "epoch": 2.5660228616476153, "grad_norm": 0.4338550567626953, "learning_rate": 0.0005, "loss": 1.2237532138824463, "step": 32550 }, { "epoch": 2.566101694915254, "grad_norm": 0.5469593405723572, "learning_rate": 0.0005, "loss": 1.3364121913909912, "step": 32551 }, { "epoch": 2.566180528182893, "grad_norm": 0.5360134840011597, "learning_rate": 0.0005, "loss": 1.344095230102539, "step": 32552 }, { "epoch": 2.5662593614505322, "grad_norm": 0.4940560758113861, "learning_rate": 0.0005, "loss": 1.28004789352417, "step": 32553 }, { "epoch": 2.566338194718171, "grad_norm": 0.4575687646865845, "learning_rate": 0.0005, "loss": 1.4000840187072754, "step": 32554 }, { "epoch": 2.56641702798581, "grad_norm": 0.7088473439216614, "learning_rate": 0.0005, "loss": 1.4176452159881592, "step": 32555 }, { "epoch": 2.5664958612534488, "grad_norm": 0.48354825377464294, "learning_rate": 0.0005, "loss": 1.4274711608886719, "step": 32556 }, { "epoch": 2.566574694521088, "grad_norm": 0.6323664784431458, "learning_rate": 0.0005, "loss": 1.3897805213928223, "step": 32557 }, { "epoch": 2.566653527788727, "grad_norm": 0.5371655225753784, "learning_rate": 0.0005, "loss": 1.5202481746673584, "step": 32558 }, { "epoch": 2.5667323610563657, "grad_norm": 0.4620799720287323, "learning_rate": 0.0005, "loss": 1.2404279708862305, "step": 32559 }, { "epoch": 2.566811194324005, "grad_norm": 0.41665077209472656, "learning_rate": 0.0005, "loss": 1.275937795639038, "step": 32560 }, { "epoch": 2.5668900275916435, "grad_norm": 0.556748628616333, "learning_rate": 0.0005, "loss": 1.3141701221466064, "step": 32561 }, { "epoch": 2.5669688608592827, "grad_norm": 0.4678581953048706, "learning_rate": 0.0005, "loss": 1.4001855850219727, "step": 32562 }, { "epoch": 2.567047694126922, "grad_norm": 0.5815408825874329, "learning_rate": 0.0005, "loss": 1.4657692909240723, "step": 32563 }, { "epoch": 2.5671265273945605, "grad_norm": 0.5887289047241211, "learning_rate": 0.0005, "loss": 1.3469699621200562, "step": 32564 }, { "epoch": 2.567205360662199, "grad_norm": 0.5105680227279663, "learning_rate": 0.0005, "loss": 1.3237392902374268, "step": 32565 }, { "epoch": 2.5672841939298383, "grad_norm": 0.4395751357078552, "learning_rate": 0.0005, "loss": 1.246009111404419, "step": 32566 }, { "epoch": 2.5673630271974774, "grad_norm": 0.45419245958328247, "learning_rate": 0.0005, "loss": 1.290248990058899, "step": 32567 }, { "epoch": 2.567441860465116, "grad_norm": 0.4663046896457672, "learning_rate": 0.0005, "loss": 1.1976619958877563, "step": 32568 }, { "epoch": 2.5675206937327553, "grad_norm": 0.44454947113990784, "learning_rate": 0.0005, "loss": 1.2122946977615356, "step": 32569 }, { "epoch": 2.567599527000394, "grad_norm": 0.4819777309894562, "learning_rate": 0.0005, "loss": 1.4817547798156738, "step": 32570 }, { "epoch": 2.567678360268033, "grad_norm": 0.5145958662033081, "learning_rate": 0.0005, "loss": 1.6879647970199585, "step": 32571 }, { "epoch": 2.5677571935356722, "grad_norm": 0.42589694261550903, "learning_rate": 0.0005, "loss": 1.3224587440490723, "step": 32572 }, { "epoch": 2.567836026803311, "grad_norm": 0.5026485323905945, "learning_rate": 0.0005, "loss": 1.3061935901641846, "step": 32573 }, { "epoch": 2.56791486007095, "grad_norm": 0.5498108863830566, "learning_rate": 0.0005, "loss": 1.423830509185791, "step": 32574 }, { "epoch": 2.5679936933385887, "grad_norm": 0.49758821725845337, "learning_rate": 0.0005, "loss": 1.3374587297439575, "step": 32575 }, { "epoch": 2.568072526606228, "grad_norm": 0.4825494587421417, "learning_rate": 0.0005, "loss": 1.3851478099822998, "step": 32576 }, { "epoch": 2.568151359873867, "grad_norm": 0.7408034801483154, "learning_rate": 0.0005, "loss": 1.2927100658416748, "step": 32577 }, { "epoch": 2.5682301931415057, "grad_norm": 0.5138142108917236, "learning_rate": 0.0005, "loss": 1.2045331001281738, "step": 32578 }, { "epoch": 2.568309026409145, "grad_norm": 0.5651276111602783, "learning_rate": 0.0005, "loss": 1.2317042350769043, "step": 32579 }, { "epoch": 2.5683878596767835, "grad_norm": 0.3812873959541321, "learning_rate": 0.0005, "loss": 1.207916021347046, "step": 32580 }, { "epoch": 2.5684666929444226, "grad_norm": 0.4551045596599579, "learning_rate": 0.0005, "loss": 1.4529021978378296, "step": 32581 }, { "epoch": 2.5685455262120613, "grad_norm": 0.48055335879325867, "learning_rate": 0.0005, "loss": 1.3939480781555176, "step": 32582 }, { "epoch": 2.5686243594797005, "grad_norm": 0.5241953730583191, "learning_rate": 0.0005, "loss": 1.2539255619049072, "step": 32583 }, { "epoch": 2.568703192747339, "grad_norm": 0.45320945978164673, "learning_rate": 0.0005, "loss": 1.2638652324676514, "step": 32584 }, { "epoch": 2.5687820260149783, "grad_norm": 0.4918120801448822, "learning_rate": 0.0005, "loss": 1.1581294536590576, "step": 32585 }, { "epoch": 2.5688608592826174, "grad_norm": 0.5115662813186646, "learning_rate": 0.0005, "loss": 1.292039394378662, "step": 32586 }, { "epoch": 2.568939692550256, "grad_norm": 0.5170490741729736, "learning_rate": 0.0005, "loss": 1.4378890991210938, "step": 32587 }, { "epoch": 2.5690185258178952, "grad_norm": 0.4860381782054901, "learning_rate": 0.0005, "loss": 1.2558786869049072, "step": 32588 }, { "epoch": 2.569097359085534, "grad_norm": 0.49638256430625916, "learning_rate": 0.0005, "loss": 1.2669377326965332, "step": 32589 }, { "epoch": 2.569176192353173, "grad_norm": 0.4659890830516815, "learning_rate": 0.0005, "loss": 1.3291221857070923, "step": 32590 }, { "epoch": 2.569255025620812, "grad_norm": 0.46206483244895935, "learning_rate": 0.0005, "loss": 1.4115898609161377, "step": 32591 }, { "epoch": 2.569333858888451, "grad_norm": 0.5345780253410339, "learning_rate": 0.0005, "loss": 1.421597957611084, "step": 32592 }, { "epoch": 2.56941269215609, "grad_norm": 0.6148871183395386, "learning_rate": 0.0005, "loss": 1.2887567281723022, "step": 32593 }, { "epoch": 2.5694915254237287, "grad_norm": 0.5977407097816467, "learning_rate": 0.0005, "loss": 1.443635106086731, "step": 32594 }, { "epoch": 2.569570358691368, "grad_norm": 0.4601444900035858, "learning_rate": 0.0005, "loss": 1.1477184295654297, "step": 32595 }, { "epoch": 2.5696491919590065, "grad_norm": 1.1460943222045898, "learning_rate": 0.0005, "loss": 1.5581952333450317, "step": 32596 }, { "epoch": 2.5697280252266457, "grad_norm": 0.5540674328804016, "learning_rate": 0.0005, "loss": 1.3392419815063477, "step": 32597 }, { "epoch": 2.5698068584942844, "grad_norm": 0.5109248757362366, "learning_rate": 0.0005, "loss": 1.2941412925720215, "step": 32598 }, { "epoch": 2.5698856917619235, "grad_norm": 0.5211352705955505, "learning_rate": 0.0005, "loss": 1.3979249000549316, "step": 32599 }, { "epoch": 2.5699645250295626, "grad_norm": 0.5224384665489197, "learning_rate": 0.0005, "loss": 1.2009276151657104, "step": 32600 }, { "epoch": 2.5700433582972013, "grad_norm": 0.4869164824485779, "learning_rate": 0.0005, "loss": 1.2269437313079834, "step": 32601 }, { "epoch": 2.5701221915648405, "grad_norm": 0.5707406997680664, "learning_rate": 0.0005, "loss": 1.287182092666626, "step": 32602 }, { "epoch": 2.570201024832479, "grad_norm": 0.5240017175674438, "learning_rate": 0.0005, "loss": 1.2992491722106934, "step": 32603 }, { "epoch": 2.5702798581001183, "grad_norm": 0.4507859945297241, "learning_rate": 0.0005, "loss": 1.3500688076019287, "step": 32604 }, { "epoch": 2.5703586913677574, "grad_norm": 0.5207774043083191, "learning_rate": 0.0005, "loss": 1.238175630569458, "step": 32605 }, { "epoch": 2.570437524635396, "grad_norm": 0.4515500068664551, "learning_rate": 0.0005, "loss": 1.37904953956604, "step": 32606 }, { "epoch": 2.5705163579030352, "grad_norm": 0.4260014593601227, "learning_rate": 0.0005, "loss": 1.0160514116287231, "step": 32607 }, { "epoch": 2.570595191170674, "grad_norm": 0.4657723307609558, "learning_rate": 0.0005, "loss": 1.5665638446807861, "step": 32608 }, { "epoch": 2.570674024438313, "grad_norm": 0.49952274560928345, "learning_rate": 0.0005, "loss": 1.4030585289001465, "step": 32609 }, { "epoch": 2.570752857705952, "grad_norm": 0.5532729029655457, "learning_rate": 0.0005, "loss": 1.4094297885894775, "step": 32610 }, { "epoch": 2.570831690973591, "grad_norm": 0.5835930109024048, "learning_rate": 0.0005, "loss": 1.392470359802246, "step": 32611 }, { "epoch": 2.5709105242412296, "grad_norm": 0.4112442433834076, "learning_rate": 0.0005, "loss": 1.2625973224639893, "step": 32612 }, { "epoch": 2.5709893575088687, "grad_norm": 0.6498827934265137, "learning_rate": 0.0005, "loss": 1.388929009437561, "step": 32613 }, { "epoch": 2.571068190776508, "grad_norm": 0.7495924234390259, "learning_rate": 0.0005, "loss": 1.4810023307800293, "step": 32614 }, { "epoch": 2.5711470240441465, "grad_norm": 0.5839231610298157, "learning_rate": 0.0005, "loss": 1.509251594543457, "step": 32615 }, { "epoch": 2.5712258573117857, "grad_norm": 0.43675366044044495, "learning_rate": 0.0005, "loss": 1.3277852535247803, "step": 32616 }, { "epoch": 2.5713046905794243, "grad_norm": 0.48636695742607117, "learning_rate": 0.0005, "loss": 1.3455145359039307, "step": 32617 }, { "epoch": 2.5713835238470635, "grad_norm": 0.3935757875442505, "learning_rate": 0.0005, "loss": 1.0954937934875488, "step": 32618 }, { "epoch": 2.5714623571147026, "grad_norm": 0.5308600068092346, "learning_rate": 0.0005, "loss": 1.3463385105133057, "step": 32619 }, { "epoch": 2.5715411903823413, "grad_norm": 0.5495949983596802, "learning_rate": 0.0005, "loss": 1.394871473312378, "step": 32620 }, { "epoch": 2.5716200236499804, "grad_norm": 0.5397502183914185, "learning_rate": 0.0005, "loss": 1.2569411993026733, "step": 32621 }, { "epoch": 2.571698856917619, "grad_norm": 0.4560234844684601, "learning_rate": 0.0005, "loss": 1.2991396188735962, "step": 32622 }, { "epoch": 2.5717776901852583, "grad_norm": 0.443589985370636, "learning_rate": 0.0005, "loss": 1.2098593711853027, "step": 32623 }, { "epoch": 2.5718565234528974, "grad_norm": 0.6201456785202026, "learning_rate": 0.0005, "loss": 1.3545207977294922, "step": 32624 }, { "epoch": 2.571935356720536, "grad_norm": 0.523514986038208, "learning_rate": 0.0005, "loss": 1.2074697017669678, "step": 32625 }, { "epoch": 2.5720141899881748, "grad_norm": 0.5068216323852539, "learning_rate": 0.0005, "loss": 1.3263566493988037, "step": 32626 }, { "epoch": 2.572093023255814, "grad_norm": 0.4768129587173462, "learning_rate": 0.0005, "loss": 1.2354772090911865, "step": 32627 }, { "epoch": 2.572171856523453, "grad_norm": 0.5059556365013123, "learning_rate": 0.0005, "loss": 1.3908801078796387, "step": 32628 }, { "epoch": 2.5722506897910917, "grad_norm": 0.5616469383239746, "learning_rate": 0.0005, "loss": 1.5787992477416992, "step": 32629 }, { "epoch": 2.572329523058731, "grad_norm": 0.4887404441833496, "learning_rate": 0.0005, "loss": 1.3720042705535889, "step": 32630 }, { "epoch": 2.5724083563263695, "grad_norm": 0.5798875093460083, "learning_rate": 0.0005, "loss": 1.4537079334259033, "step": 32631 }, { "epoch": 2.5724871895940087, "grad_norm": 0.44974544644355774, "learning_rate": 0.0005, "loss": 1.3298065662384033, "step": 32632 }, { "epoch": 2.572566022861648, "grad_norm": 0.48197102546691895, "learning_rate": 0.0005, "loss": 1.372492790222168, "step": 32633 }, { "epoch": 2.5726448561292865, "grad_norm": 0.5456447005271912, "learning_rate": 0.0005, "loss": 1.428696632385254, "step": 32634 }, { "epoch": 2.5727236893969256, "grad_norm": 0.5304083228111267, "learning_rate": 0.0005, "loss": 1.4540295600891113, "step": 32635 }, { "epoch": 2.5728025226645643, "grad_norm": 0.5239014625549316, "learning_rate": 0.0005, "loss": 1.3330247402191162, "step": 32636 }, { "epoch": 2.5728813559322035, "grad_norm": 0.5102386474609375, "learning_rate": 0.0005, "loss": 1.3557868003845215, "step": 32637 }, { "epoch": 2.5729601891998426, "grad_norm": 0.6588415503501892, "learning_rate": 0.0005, "loss": 1.4477171897888184, "step": 32638 }, { "epoch": 2.5730390224674813, "grad_norm": 0.5072263479232788, "learning_rate": 0.0005, "loss": 1.4845606088638306, "step": 32639 }, { "epoch": 2.57311785573512, "grad_norm": 0.5784098505973816, "learning_rate": 0.0005, "loss": 1.515010118484497, "step": 32640 }, { "epoch": 2.573196689002759, "grad_norm": 0.614920437335968, "learning_rate": 0.0005, "loss": 1.2342705726623535, "step": 32641 }, { "epoch": 2.5732755222703982, "grad_norm": 0.4316616952419281, "learning_rate": 0.0005, "loss": 1.0736453533172607, "step": 32642 }, { "epoch": 2.573354355538037, "grad_norm": 0.6014229655265808, "learning_rate": 0.0005, "loss": 1.3302791118621826, "step": 32643 }, { "epoch": 2.573433188805676, "grad_norm": 0.4482574760913849, "learning_rate": 0.0005, "loss": 1.1455049514770508, "step": 32644 }, { "epoch": 2.5735120220733148, "grad_norm": 0.5812934637069702, "learning_rate": 0.0005, "loss": 1.5709619522094727, "step": 32645 }, { "epoch": 2.573590855340954, "grad_norm": 0.45689186453819275, "learning_rate": 0.0005, "loss": 1.1575086116790771, "step": 32646 }, { "epoch": 2.573669688608593, "grad_norm": 0.4847009778022766, "learning_rate": 0.0005, "loss": 1.2573118209838867, "step": 32647 }, { "epoch": 2.5737485218762317, "grad_norm": 0.454814612865448, "learning_rate": 0.0005, "loss": 1.3940621614456177, "step": 32648 }, { "epoch": 2.573827355143871, "grad_norm": 0.4364425837993622, "learning_rate": 0.0005, "loss": 1.2448043823242188, "step": 32649 }, { "epoch": 2.5739061884115095, "grad_norm": 0.5034843683242798, "learning_rate": 0.0005, "loss": 1.260613203048706, "step": 32650 }, { "epoch": 2.5739850216791487, "grad_norm": 0.40174803137779236, "learning_rate": 0.0005, "loss": 1.2310560941696167, "step": 32651 }, { "epoch": 2.574063854946788, "grad_norm": 0.6823201775550842, "learning_rate": 0.0005, "loss": 1.608642339706421, "step": 32652 }, { "epoch": 2.5741426882144265, "grad_norm": 0.5288650393486023, "learning_rate": 0.0005, "loss": 1.3318099975585938, "step": 32653 }, { "epoch": 2.574221521482065, "grad_norm": 0.457306832075119, "learning_rate": 0.0005, "loss": 1.1568255424499512, "step": 32654 }, { "epoch": 2.5743003547497043, "grad_norm": 0.5123719573020935, "learning_rate": 0.0005, "loss": 1.3938182592391968, "step": 32655 }, { "epoch": 2.5743791880173434, "grad_norm": 0.4752579629421234, "learning_rate": 0.0005, "loss": 1.3711755275726318, "step": 32656 }, { "epoch": 2.574458021284982, "grad_norm": 1.5916814804077148, "learning_rate": 0.0005, "loss": 1.3307316303253174, "step": 32657 }, { "epoch": 2.5745368545526213, "grad_norm": 0.588639497756958, "learning_rate": 0.0005, "loss": 1.3269256353378296, "step": 32658 }, { "epoch": 2.57461568782026, "grad_norm": 0.5698415040969849, "learning_rate": 0.0005, "loss": 1.574099063873291, "step": 32659 }, { "epoch": 2.574694521087899, "grad_norm": 0.4677761197090149, "learning_rate": 0.0005, "loss": 1.3191273212432861, "step": 32660 }, { "epoch": 2.5747733543555382, "grad_norm": 0.4166119694709778, "learning_rate": 0.0005, "loss": 1.1495134830474854, "step": 32661 }, { "epoch": 2.574852187623177, "grad_norm": 0.5175060629844666, "learning_rate": 0.0005, "loss": 1.4095022678375244, "step": 32662 }, { "epoch": 2.574931020890816, "grad_norm": 0.5671879053115845, "learning_rate": 0.0005, "loss": 1.4355485439300537, "step": 32663 }, { "epoch": 2.5750098541584547, "grad_norm": 0.5374677181243896, "learning_rate": 0.0005, "loss": 1.4949800968170166, "step": 32664 }, { "epoch": 2.575088687426094, "grad_norm": 0.4699932336807251, "learning_rate": 0.0005, "loss": 1.3937621116638184, "step": 32665 }, { "epoch": 2.575167520693733, "grad_norm": 0.4749791920185089, "learning_rate": 0.0005, "loss": 1.2148255109786987, "step": 32666 }, { "epoch": 2.5752463539613717, "grad_norm": 0.5424721240997314, "learning_rate": 0.0005, "loss": 1.4939407110214233, "step": 32667 }, { "epoch": 2.5753251872290104, "grad_norm": 0.526707649230957, "learning_rate": 0.0005, "loss": 1.2903984785079956, "step": 32668 }, { "epoch": 2.5754040204966495, "grad_norm": 0.4997789263725281, "learning_rate": 0.0005, "loss": 1.215683102607727, "step": 32669 }, { "epoch": 2.5754828537642886, "grad_norm": 0.5052099823951721, "learning_rate": 0.0005, "loss": 1.1958075761795044, "step": 32670 }, { "epoch": 2.5755616870319273, "grad_norm": 0.5429546236991882, "learning_rate": 0.0005, "loss": 1.4539048671722412, "step": 32671 }, { "epoch": 2.5756405202995665, "grad_norm": 0.6362968683242798, "learning_rate": 0.0005, "loss": 1.4012233018875122, "step": 32672 }, { "epoch": 2.575719353567205, "grad_norm": 0.43972769379615784, "learning_rate": 0.0005, "loss": 1.2203469276428223, "step": 32673 }, { "epoch": 2.5757981868348443, "grad_norm": 0.47209909558296204, "learning_rate": 0.0005, "loss": 1.2113232612609863, "step": 32674 }, { "epoch": 2.5758770201024834, "grad_norm": 0.7528533935546875, "learning_rate": 0.0005, "loss": 1.5239970684051514, "step": 32675 }, { "epoch": 2.575955853370122, "grad_norm": 0.6590392589569092, "learning_rate": 0.0005, "loss": 1.5458872318267822, "step": 32676 }, { "epoch": 2.5760346866377613, "grad_norm": 0.7421782612800598, "learning_rate": 0.0005, "loss": 1.5330474376678467, "step": 32677 }, { "epoch": 2.5761135199054, "grad_norm": 0.5891018509864807, "learning_rate": 0.0005, "loss": 1.5500785112380981, "step": 32678 }, { "epoch": 2.576192353173039, "grad_norm": 0.529991090297699, "learning_rate": 0.0005, "loss": 1.3561692237854004, "step": 32679 }, { "epoch": 2.576271186440678, "grad_norm": 0.5643376708030701, "learning_rate": 0.0005, "loss": 1.4235680103302002, "step": 32680 }, { "epoch": 2.576350019708317, "grad_norm": 0.5687635540962219, "learning_rate": 0.0005, "loss": 1.1690951585769653, "step": 32681 }, { "epoch": 2.5764288529759556, "grad_norm": 0.5672497749328613, "learning_rate": 0.0005, "loss": 1.4448665380477905, "step": 32682 }, { "epoch": 2.5765076862435947, "grad_norm": 0.496803879737854, "learning_rate": 0.0005, "loss": 1.2459189891815186, "step": 32683 }, { "epoch": 2.576586519511234, "grad_norm": 0.4450322687625885, "learning_rate": 0.0005, "loss": 1.2904744148254395, "step": 32684 }, { "epoch": 2.5766653527788725, "grad_norm": 0.5583546161651611, "learning_rate": 0.0005, "loss": 1.4330941438674927, "step": 32685 }, { "epoch": 2.5767441860465117, "grad_norm": 0.4307667016983032, "learning_rate": 0.0005, "loss": 1.2829203605651855, "step": 32686 }, { "epoch": 2.5768230193141504, "grad_norm": 0.6142381429672241, "learning_rate": 0.0005, "loss": 1.5080205202102661, "step": 32687 }, { "epoch": 2.5769018525817895, "grad_norm": 0.49763137102127075, "learning_rate": 0.0005, "loss": 1.4031174182891846, "step": 32688 }, { "epoch": 2.5769806858494286, "grad_norm": 0.4729728698730469, "learning_rate": 0.0005, "loss": 1.284257173538208, "step": 32689 }, { "epoch": 2.5770595191170673, "grad_norm": 0.542937695980072, "learning_rate": 0.0005, "loss": 1.4716017246246338, "step": 32690 }, { "epoch": 2.5771383523847065, "grad_norm": 0.4355758726596832, "learning_rate": 0.0005, "loss": 1.2543519735336304, "step": 32691 }, { "epoch": 2.577217185652345, "grad_norm": 0.5249320864677429, "learning_rate": 0.0005, "loss": 1.5726394653320312, "step": 32692 }, { "epoch": 2.5772960189199843, "grad_norm": 0.4509197175502777, "learning_rate": 0.0005, "loss": 1.308472752571106, "step": 32693 }, { "epoch": 2.5773748521876234, "grad_norm": 0.6148849129676819, "learning_rate": 0.0005, "loss": 1.1738053560256958, "step": 32694 }, { "epoch": 2.577453685455262, "grad_norm": 0.45815610885620117, "learning_rate": 0.0005, "loss": 1.2671838998794556, "step": 32695 }, { "epoch": 2.577532518722901, "grad_norm": 0.6252416372299194, "learning_rate": 0.0005, "loss": 1.578665018081665, "step": 32696 }, { "epoch": 2.57761135199054, "grad_norm": 0.4721139073371887, "learning_rate": 0.0005, "loss": 1.3693315982818604, "step": 32697 }, { "epoch": 2.577690185258179, "grad_norm": 0.4494442641735077, "learning_rate": 0.0005, "loss": 1.311471700668335, "step": 32698 }, { "epoch": 2.5777690185258177, "grad_norm": 0.5173017978668213, "learning_rate": 0.0005, "loss": 1.2805397510528564, "step": 32699 }, { "epoch": 2.577847851793457, "grad_norm": 0.5063640475273132, "learning_rate": 0.0005, "loss": 1.622002124786377, "step": 32700 }, { "epoch": 2.5779266850610956, "grad_norm": 0.44504427909851074, "learning_rate": 0.0005, "loss": 1.3425605297088623, "step": 32701 }, { "epoch": 2.5780055183287347, "grad_norm": 0.45325610041618347, "learning_rate": 0.0005, "loss": 1.261533260345459, "step": 32702 }, { "epoch": 2.578084351596374, "grad_norm": 0.5362719297409058, "learning_rate": 0.0005, "loss": 1.606184959411621, "step": 32703 }, { "epoch": 2.5781631848640125, "grad_norm": 0.5285587310791016, "learning_rate": 0.0005, "loss": 1.2734386920928955, "step": 32704 }, { "epoch": 2.5782420181316517, "grad_norm": 0.4591533839702606, "learning_rate": 0.0005, "loss": 1.228601336479187, "step": 32705 }, { "epoch": 2.5783208513992903, "grad_norm": 0.5145815014839172, "learning_rate": 0.0005, "loss": 1.4492225646972656, "step": 32706 }, { "epoch": 2.5783996846669295, "grad_norm": 0.5342621803283691, "learning_rate": 0.0005, "loss": 1.5059833526611328, "step": 32707 }, { "epoch": 2.5784785179345686, "grad_norm": 0.5083213448524475, "learning_rate": 0.0005, "loss": 1.3866198062896729, "step": 32708 }, { "epoch": 2.5785573512022073, "grad_norm": 0.39530259370803833, "learning_rate": 0.0005, "loss": 1.2889446020126343, "step": 32709 }, { "epoch": 2.5786361844698464, "grad_norm": 0.4331224262714386, "learning_rate": 0.0005, "loss": 1.0836282968521118, "step": 32710 }, { "epoch": 2.578715017737485, "grad_norm": 0.4855146110057831, "learning_rate": 0.0005, "loss": 1.250232458114624, "step": 32711 }, { "epoch": 2.5787938510051243, "grad_norm": 0.5949252247810364, "learning_rate": 0.0005, "loss": 1.3331139087677002, "step": 32712 }, { "epoch": 2.578872684272763, "grad_norm": 0.49759113788604736, "learning_rate": 0.0005, "loss": 1.3412950038909912, "step": 32713 }, { "epoch": 2.578951517540402, "grad_norm": 0.5762301683425903, "learning_rate": 0.0005, "loss": 1.299238920211792, "step": 32714 }, { "epoch": 2.5790303508080408, "grad_norm": 0.4770454466342926, "learning_rate": 0.0005, "loss": 1.2200486660003662, "step": 32715 }, { "epoch": 2.57910918407568, "grad_norm": 0.5131155848503113, "learning_rate": 0.0005, "loss": 1.547481894493103, "step": 32716 }, { "epoch": 2.579188017343319, "grad_norm": 0.6126191020011902, "learning_rate": 0.0005, "loss": 1.5821261405944824, "step": 32717 }, { "epoch": 2.5792668506109577, "grad_norm": 0.6123259663581848, "learning_rate": 0.0005, "loss": 1.4714629650115967, "step": 32718 }, { "epoch": 2.579345683878597, "grad_norm": 0.5426167249679565, "learning_rate": 0.0005, "loss": 1.231918454170227, "step": 32719 }, { "epoch": 2.5794245171462356, "grad_norm": 0.5540497303009033, "learning_rate": 0.0005, "loss": 1.5478847026824951, "step": 32720 }, { "epoch": 2.5795033504138747, "grad_norm": 0.37243813276290894, "learning_rate": 0.0005, "loss": 1.1144187450408936, "step": 32721 }, { "epoch": 2.579582183681514, "grad_norm": 0.4534117877483368, "learning_rate": 0.0005, "loss": 1.2581260204315186, "step": 32722 }, { "epoch": 2.5796610169491525, "grad_norm": 0.5330349206924438, "learning_rate": 0.0005, "loss": 1.1667600870132446, "step": 32723 }, { "epoch": 2.5797398502167916, "grad_norm": 0.42924338579177856, "learning_rate": 0.0005, "loss": 1.1638402938842773, "step": 32724 }, { "epoch": 2.5798186834844303, "grad_norm": 0.5140810012817383, "learning_rate": 0.0005, "loss": 1.197540521621704, "step": 32725 }, { "epoch": 2.5798975167520695, "grad_norm": 0.8269230723381042, "learning_rate": 0.0005, "loss": 1.4099739789962769, "step": 32726 }, { "epoch": 2.579976350019708, "grad_norm": 0.46207237243652344, "learning_rate": 0.0005, "loss": 1.1391470432281494, "step": 32727 }, { "epoch": 2.5800551832873473, "grad_norm": 0.6100811958312988, "learning_rate": 0.0005, "loss": 1.403167963027954, "step": 32728 }, { "epoch": 2.580134016554986, "grad_norm": 0.4274415969848633, "learning_rate": 0.0005, "loss": 1.4432883262634277, "step": 32729 }, { "epoch": 2.580212849822625, "grad_norm": 0.5437668561935425, "learning_rate": 0.0005, "loss": 1.3868035078048706, "step": 32730 }, { "epoch": 2.5802916830902642, "grad_norm": 0.7883633375167847, "learning_rate": 0.0005, "loss": 1.4824259281158447, "step": 32731 }, { "epoch": 2.580370516357903, "grad_norm": 0.5506380796432495, "learning_rate": 0.0005, "loss": 1.4638569355010986, "step": 32732 }, { "epoch": 2.580449349625542, "grad_norm": 0.5064609050750732, "learning_rate": 0.0005, "loss": 1.2785985469818115, "step": 32733 }, { "epoch": 2.5805281828931808, "grad_norm": 0.5192337036132812, "learning_rate": 0.0005, "loss": 1.3821892738342285, "step": 32734 }, { "epoch": 2.58060701616082, "grad_norm": 0.5555121302604675, "learning_rate": 0.0005, "loss": 1.3257040977478027, "step": 32735 }, { "epoch": 2.580685849428459, "grad_norm": 0.5954516530036926, "learning_rate": 0.0005, "loss": 1.3666760921478271, "step": 32736 }, { "epoch": 2.5807646826960977, "grad_norm": 0.6015856266021729, "learning_rate": 0.0005, "loss": 1.407482624053955, "step": 32737 }, { "epoch": 2.580843515963737, "grad_norm": 0.6702842116355896, "learning_rate": 0.0005, "loss": 1.3514423370361328, "step": 32738 }, { "epoch": 2.5809223492313755, "grad_norm": 0.5575474500656128, "learning_rate": 0.0005, "loss": 1.4628849029541016, "step": 32739 }, { "epoch": 2.5810011824990147, "grad_norm": 0.613351047039032, "learning_rate": 0.0005, "loss": 1.5722510814666748, "step": 32740 }, { "epoch": 2.581080015766654, "grad_norm": 0.6310606002807617, "learning_rate": 0.0005, "loss": 1.4343459606170654, "step": 32741 }, { "epoch": 2.5811588490342925, "grad_norm": 0.5313856601715088, "learning_rate": 0.0005, "loss": 1.349639654159546, "step": 32742 }, { "epoch": 2.581237682301931, "grad_norm": 0.477687269449234, "learning_rate": 0.0005, "loss": 1.2605408430099487, "step": 32743 }, { "epoch": 2.5813165155695703, "grad_norm": 0.4907461404800415, "learning_rate": 0.0005, "loss": 1.263617992401123, "step": 32744 }, { "epoch": 2.5813953488372094, "grad_norm": 0.5666082501411438, "learning_rate": 0.0005, "loss": 1.236255168914795, "step": 32745 }, { "epoch": 2.581474182104848, "grad_norm": 0.6803351640701294, "learning_rate": 0.0005, "loss": 1.525269627571106, "step": 32746 }, { "epoch": 2.5815530153724873, "grad_norm": 0.5779699683189392, "learning_rate": 0.0005, "loss": 1.3994078636169434, "step": 32747 }, { "epoch": 2.581631848640126, "grad_norm": 1.0358589887619019, "learning_rate": 0.0005, "loss": 1.6216720342636108, "step": 32748 }, { "epoch": 2.581710681907765, "grad_norm": 0.5849839448928833, "learning_rate": 0.0005, "loss": 1.5144243240356445, "step": 32749 }, { "epoch": 2.5817895151754042, "grad_norm": 0.4899088144302368, "learning_rate": 0.0005, "loss": 1.199912667274475, "step": 32750 }, { "epoch": 2.581868348443043, "grad_norm": 0.5554359555244446, "learning_rate": 0.0005, "loss": 1.2921727895736694, "step": 32751 }, { "epoch": 2.581947181710682, "grad_norm": 0.5947186946868896, "learning_rate": 0.0005, "loss": 1.4013181924819946, "step": 32752 }, { "epoch": 2.5820260149783207, "grad_norm": 0.6232037544250488, "learning_rate": 0.0005, "loss": 1.43997061252594, "step": 32753 }, { "epoch": 2.58210484824596, "grad_norm": 0.485194593667984, "learning_rate": 0.0005, "loss": 1.2638441324234009, "step": 32754 }, { "epoch": 2.582183681513599, "grad_norm": 0.5583275556564331, "learning_rate": 0.0005, "loss": 1.5292279720306396, "step": 32755 }, { "epoch": 2.5822625147812377, "grad_norm": 0.6266511082649231, "learning_rate": 0.0005, "loss": 1.415849208831787, "step": 32756 }, { "epoch": 2.5823413480488764, "grad_norm": 0.4859989583492279, "learning_rate": 0.0005, "loss": 1.3181556463241577, "step": 32757 }, { "epoch": 2.5824201813165155, "grad_norm": 0.5746468305587769, "learning_rate": 0.0005, "loss": 1.5887491703033447, "step": 32758 }, { "epoch": 2.5824990145841547, "grad_norm": 0.6486518979072571, "learning_rate": 0.0005, "loss": 1.4431567192077637, "step": 32759 }, { "epoch": 2.5825778478517933, "grad_norm": 0.5181805491447449, "learning_rate": 0.0005, "loss": 1.3499038219451904, "step": 32760 }, { "epoch": 2.5826566811194325, "grad_norm": 0.5118341445922852, "learning_rate": 0.0005, "loss": 1.3962385654449463, "step": 32761 }, { "epoch": 2.582735514387071, "grad_norm": 0.5981252193450928, "learning_rate": 0.0005, "loss": 1.285296082496643, "step": 32762 }, { "epoch": 2.5828143476547103, "grad_norm": 0.5326070785522461, "learning_rate": 0.0005, "loss": 1.3690767288208008, "step": 32763 }, { "epoch": 2.5828931809223494, "grad_norm": 0.45983418822288513, "learning_rate": 0.0005, "loss": 1.6042976379394531, "step": 32764 }, { "epoch": 2.582972014189988, "grad_norm": 0.5644960999488831, "learning_rate": 0.0005, "loss": 1.3357503414154053, "step": 32765 }, { "epoch": 2.5830508474576273, "grad_norm": 0.50779128074646, "learning_rate": 0.0005, "loss": 1.2356233596801758, "step": 32766 }, { "epoch": 2.583129680725266, "grad_norm": 0.45627671480178833, "learning_rate": 0.0005, "loss": 1.2087559700012207, "step": 32767 }, { "epoch": 2.583208513992905, "grad_norm": 0.62236088514328, "learning_rate": 0.0005, "loss": 1.5155680179595947, "step": 32768 }, { "epoch": 2.583287347260544, "grad_norm": 0.45101067423820496, "learning_rate": 0.0005, "loss": 1.2613967657089233, "step": 32769 }, { "epoch": 2.583366180528183, "grad_norm": 0.5714508891105652, "learning_rate": 0.0005, "loss": 1.2373099327087402, "step": 32770 }, { "epoch": 2.5834450137958216, "grad_norm": 4.192469596862793, "learning_rate": 0.0005, "loss": 1.3035011291503906, "step": 32771 }, { "epoch": 2.5835238470634607, "grad_norm": 0.4602147042751312, "learning_rate": 0.0005, "loss": 1.0207432508468628, "step": 32772 }, { "epoch": 2.5836026803311, "grad_norm": 0.5353338122367859, "learning_rate": 0.0005, "loss": 1.2451951503753662, "step": 32773 }, { "epoch": 2.5836815135987385, "grad_norm": 0.5993489027023315, "learning_rate": 0.0005, "loss": 1.4289276599884033, "step": 32774 }, { "epoch": 2.5837603468663777, "grad_norm": 0.468404620885849, "learning_rate": 0.0005, "loss": 1.2923859357833862, "step": 32775 }, { "epoch": 2.5838391801340164, "grad_norm": 0.47966763377189636, "learning_rate": 0.0005, "loss": 1.3685355186462402, "step": 32776 }, { "epoch": 2.5839180134016555, "grad_norm": 0.7418712377548218, "learning_rate": 0.0005, "loss": 1.574949026107788, "step": 32777 }, { "epoch": 2.5839968466692946, "grad_norm": 0.4514469802379608, "learning_rate": 0.0005, "loss": 1.2146254777908325, "step": 32778 }, { "epoch": 2.5840756799369333, "grad_norm": 0.5188712477684021, "learning_rate": 0.0005, "loss": 1.2032772302627563, "step": 32779 }, { "epoch": 2.5841545132045725, "grad_norm": 0.521721363067627, "learning_rate": 0.0005, "loss": 1.449298620223999, "step": 32780 }, { "epoch": 2.584233346472211, "grad_norm": 0.4592546224594116, "learning_rate": 0.0005, "loss": 1.417129397392273, "step": 32781 }, { "epoch": 2.5843121797398503, "grad_norm": 0.6015331149101257, "learning_rate": 0.0005, "loss": 1.3781551122665405, "step": 32782 }, { "epoch": 2.5843910130074894, "grad_norm": 0.48380517959594727, "learning_rate": 0.0005, "loss": 1.4268519878387451, "step": 32783 }, { "epoch": 2.584469846275128, "grad_norm": 0.4180658459663391, "learning_rate": 0.0005, "loss": 1.1280648708343506, "step": 32784 }, { "epoch": 2.584548679542767, "grad_norm": 0.5080854892730713, "learning_rate": 0.0005, "loss": 1.5401579141616821, "step": 32785 }, { "epoch": 2.584627512810406, "grad_norm": 0.4955986440181732, "learning_rate": 0.0005, "loss": 1.2187731266021729, "step": 32786 }, { "epoch": 2.584706346078045, "grad_norm": 0.5767961144447327, "learning_rate": 0.0005, "loss": 1.337209701538086, "step": 32787 }, { "epoch": 2.5847851793456837, "grad_norm": 0.4723765552043915, "learning_rate": 0.0005, "loss": 1.404573917388916, "step": 32788 }, { "epoch": 2.584864012613323, "grad_norm": 0.440700888633728, "learning_rate": 0.0005, "loss": 1.2149577140808105, "step": 32789 }, { "epoch": 2.5849428458809616, "grad_norm": 0.5145580768585205, "learning_rate": 0.0005, "loss": 1.5631444454193115, "step": 32790 }, { "epoch": 2.5850216791486007, "grad_norm": 0.6349011063575745, "learning_rate": 0.0005, "loss": 1.4912437200546265, "step": 32791 }, { "epoch": 2.58510051241624, "grad_norm": 0.4738619327545166, "learning_rate": 0.0005, "loss": 1.4622063636779785, "step": 32792 }, { "epoch": 2.5851793456838785, "grad_norm": 0.7190619707107544, "learning_rate": 0.0005, "loss": 1.2395446300506592, "step": 32793 }, { "epoch": 2.5852581789515177, "grad_norm": 0.5287585854530334, "learning_rate": 0.0005, "loss": 1.3702099323272705, "step": 32794 }, { "epoch": 2.5853370122191563, "grad_norm": 0.4848068356513977, "learning_rate": 0.0005, "loss": 1.424039363861084, "step": 32795 }, { "epoch": 2.5854158454867955, "grad_norm": 0.5179009437561035, "learning_rate": 0.0005, "loss": 1.3139585256576538, "step": 32796 }, { "epoch": 2.5854946787544346, "grad_norm": 0.4466175138950348, "learning_rate": 0.0005, "loss": 1.546673059463501, "step": 32797 }, { "epoch": 2.5855735120220733, "grad_norm": 0.715465784072876, "learning_rate": 0.0005, "loss": 1.4743316173553467, "step": 32798 }, { "epoch": 2.585652345289712, "grad_norm": 0.5248321890830994, "learning_rate": 0.0005, "loss": 1.3451355695724487, "step": 32799 }, { "epoch": 2.585731178557351, "grad_norm": 0.6094547510147095, "learning_rate": 0.0005, "loss": 1.4137489795684814, "step": 32800 }, { "epoch": 2.5858100118249903, "grad_norm": 0.4376065731048584, "learning_rate": 0.0005, "loss": 1.0780603885650635, "step": 32801 }, { "epoch": 2.585888845092629, "grad_norm": 0.5373980402946472, "learning_rate": 0.0005, "loss": 1.2231642007827759, "step": 32802 }, { "epoch": 2.585967678360268, "grad_norm": 0.48264172673225403, "learning_rate": 0.0005, "loss": 1.2660722732543945, "step": 32803 }, { "epoch": 2.5860465116279068, "grad_norm": 0.42713868618011475, "learning_rate": 0.0005, "loss": 1.2848414182662964, "step": 32804 }, { "epoch": 2.586125344895546, "grad_norm": 0.4768590033054352, "learning_rate": 0.0005, "loss": 1.2023779153823853, "step": 32805 }, { "epoch": 2.586204178163185, "grad_norm": 0.5148842930793762, "learning_rate": 0.0005, "loss": 1.3517073392868042, "step": 32806 }, { "epoch": 2.5862830114308237, "grad_norm": 0.6298648118972778, "learning_rate": 0.0005, "loss": 1.3998271226882935, "step": 32807 }, { "epoch": 2.586361844698463, "grad_norm": 0.45454102754592896, "learning_rate": 0.0005, "loss": 1.260740876197815, "step": 32808 }, { "epoch": 2.5864406779661016, "grad_norm": 0.5839904546737671, "learning_rate": 0.0005, "loss": 1.3847087621688843, "step": 32809 }, { "epoch": 2.5865195112337407, "grad_norm": 0.47088804841041565, "learning_rate": 0.0005, "loss": 1.1992239952087402, "step": 32810 }, { "epoch": 2.58659834450138, "grad_norm": 0.4831404387950897, "learning_rate": 0.0005, "loss": 1.193171501159668, "step": 32811 }, { "epoch": 2.5866771777690185, "grad_norm": 0.5413971543312073, "learning_rate": 0.0005, "loss": 1.4365193843841553, "step": 32812 }, { "epoch": 2.586756011036657, "grad_norm": 0.5643041729927063, "learning_rate": 0.0005, "loss": 1.4027864933013916, "step": 32813 }, { "epoch": 2.5868348443042963, "grad_norm": 0.5277060270309448, "learning_rate": 0.0005, "loss": 1.4538860321044922, "step": 32814 }, { "epoch": 2.5869136775719355, "grad_norm": 0.6107758283615112, "learning_rate": 0.0005, "loss": 1.438960075378418, "step": 32815 }, { "epoch": 2.586992510839574, "grad_norm": 0.47768089175224304, "learning_rate": 0.0005, "loss": 1.2982141971588135, "step": 32816 }, { "epoch": 2.5870713441072133, "grad_norm": 0.4569316804409027, "learning_rate": 0.0005, "loss": 1.3524103164672852, "step": 32817 }, { "epoch": 2.587150177374852, "grad_norm": 0.4957115352153778, "learning_rate": 0.0005, "loss": 1.4236717224121094, "step": 32818 }, { "epoch": 2.587229010642491, "grad_norm": 0.520584225654602, "learning_rate": 0.0005, "loss": 1.3613855838775635, "step": 32819 }, { "epoch": 2.5873078439101302, "grad_norm": 0.3985917866230011, "learning_rate": 0.0005, "loss": 1.1667425632476807, "step": 32820 }, { "epoch": 2.587386677177769, "grad_norm": 0.5067381262779236, "learning_rate": 0.0005, "loss": 1.3614802360534668, "step": 32821 }, { "epoch": 2.587465510445408, "grad_norm": 0.4534027576446533, "learning_rate": 0.0005, "loss": 1.1717064380645752, "step": 32822 }, { "epoch": 2.5875443437130468, "grad_norm": 0.44254326820373535, "learning_rate": 0.0005, "loss": 1.29837167263031, "step": 32823 }, { "epoch": 2.587623176980686, "grad_norm": 0.4619345963001251, "learning_rate": 0.0005, "loss": 1.4597561359405518, "step": 32824 }, { "epoch": 2.587702010248325, "grad_norm": 0.5003630518913269, "learning_rate": 0.0005, "loss": 1.537691354751587, "step": 32825 }, { "epoch": 2.5877808435159637, "grad_norm": 0.5017669796943665, "learning_rate": 0.0005, "loss": 1.2838279008865356, "step": 32826 }, { "epoch": 2.5878596767836024, "grad_norm": 0.4270358085632324, "learning_rate": 0.0005, "loss": 1.3156168460845947, "step": 32827 }, { "epoch": 2.5879385100512415, "grad_norm": 0.49452969431877136, "learning_rate": 0.0005, "loss": 1.35624098777771, "step": 32828 }, { "epoch": 2.5880173433188807, "grad_norm": 0.4811599552631378, "learning_rate": 0.0005, "loss": 1.4154095649719238, "step": 32829 }, { "epoch": 2.5880961765865194, "grad_norm": 0.4230044186115265, "learning_rate": 0.0005, "loss": 1.168774962425232, "step": 32830 }, { "epoch": 2.5881750098541585, "grad_norm": 0.3945581018924713, "learning_rate": 0.0005, "loss": 1.2699002027511597, "step": 32831 }, { "epoch": 2.588253843121797, "grad_norm": 0.6830456256866455, "learning_rate": 0.0005, "loss": 1.6286749839782715, "step": 32832 }, { "epoch": 2.5883326763894363, "grad_norm": 0.44865837693214417, "learning_rate": 0.0005, "loss": 1.0565054416656494, "step": 32833 }, { "epoch": 2.5884115096570754, "grad_norm": 0.42990872263908386, "learning_rate": 0.0005, "loss": 1.3293241262435913, "step": 32834 }, { "epoch": 2.588490342924714, "grad_norm": 0.574418306350708, "learning_rate": 0.0005, "loss": 1.4483115673065186, "step": 32835 }, { "epoch": 2.5885691761923533, "grad_norm": 0.5348930954933167, "learning_rate": 0.0005, "loss": 1.3589428663253784, "step": 32836 }, { "epoch": 2.588648009459992, "grad_norm": 0.5367786288261414, "learning_rate": 0.0005, "loss": 1.17903733253479, "step": 32837 }, { "epoch": 2.588726842727631, "grad_norm": 0.5937806963920593, "learning_rate": 0.0005, "loss": 1.3959403038024902, "step": 32838 }, { "epoch": 2.5888056759952702, "grad_norm": 0.7023877501487732, "learning_rate": 0.0005, "loss": 1.2237828969955444, "step": 32839 }, { "epoch": 2.588884509262909, "grad_norm": 0.509854257106781, "learning_rate": 0.0005, "loss": 1.4059126377105713, "step": 32840 }, { "epoch": 2.588963342530548, "grad_norm": 0.48880162835121155, "learning_rate": 0.0005, "loss": 1.2786813974380493, "step": 32841 }, { "epoch": 2.5890421757981867, "grad_norm": 0.4867701232433319, "learning_rate": 0.0005, "loss": 1.2662241458892822, "step": 32842 }, { "epoch": 2.589121009065826, "grad_norm": 0.5259591341018677, "learning_rate": 0.0005, "loss": 1.235687017440796, "step": 32843 }, { "epoch": 2.5891998423334646, "grad_norm": 0.46542102098464966, "learning_rate": 0.0005, "loss": 1.3926743268966675, "step": 32844 }, { "epoch": 2.5892786756011037, "grad_norm": 0.5198084712028503, "learning_rate": 0.0005, "loss": 1.4901264905929565, "step": 32845 }, { "epoch": 2.5893575088687424, "grad_norm": 0.43317100405693054, "learning_rate": 0.0005, "loss": 1.342536449432373, "step": 32846 }, { "epoch": 2.5894363421363815, "grad_norm": 0.48078399896621704, "learning_rate": 0.0005, "loss": 1.4588844776153564, "step": 32847 }, { "epoch": 2.5895151754040207, "grad_norm": 0.5018238425254822, "learning_rate": 0.0005, "loss": 1.1799347400665283, "step": 32848 }, { "epoch": 2.5895940086716593, "grad_norm": 1.1958949565887451, "learning_rate": 0.0005, "loss": 1.3116264343261719, "step": 32849 }, { "epoch": 2.5896728419392985, "grad_norm": 0.5515947937965393, "learning_rate": 0.0005, "loss": 1.1949784755706787, "step": 32850 }, { "epoch": 2.589751675206937, "grad_norm": 0.5171798467636108, "learning_rate": 0.0005, "loss": 1.3869495391845703, "step": 32851 }, { "epoch": 2.5898305084745763, "grad_norm": 0.5807592868804932, "learning_rate": 0.0005, "loss": 1.315859079360962, "step": 32852 }, { "epoch": 2.5899093417422154, "grad_norm": 0.4725738763809204, "learning_rate": 0.0005, "loss": 1.377854824066162, "step": 32853 }, { "epoch": 2.589988175009854, "grad_norm": 0.5018622875213623, "learning_rate": 0.0005, "loss": 1.2766519784927368, "step": 32854 }, { "epoch": 2.5900670082774933, "grad_norm": 0.6358791589736938, "learning_rate": 0.0005, "loss": 1.3280963897705078, "step": 32855 }, { "epoch": 2.590145841545132, "grad_norm": 0.42274343967437744, "learning_rate": 0.0005, "loss": 1.2754511833190918, "step": 32856 }, { "epoch": 2.590224674812771, "grad_norm": 0.44979149103164673, "learning_rate": 0.0005, "loss": 1.2629047632217407, "step": 32857 }, { "epoch": 2.5903035080804098, "grad_norm": 0.4893382489681244, "learning_rate": 0.0005, "loss": 1.2293232679367065, "step": 32858 }, { "epoch": 2.590382341348049, "grad_norm": 0.4631587564945221, "learning_rate": 0.0005, "loss": 1.2172014713287354, "step": 32859 }, { "epoch": 2.5904611746156876, "grad_norm": 0.47576481103897095, "learning_rate": 0.0005, "loss": 1.286666989326477, "step": 32860 }, { "epoch": 2.5905400078833267, "grad_norm": 0.49641847610473633, "learning_rate": 0.0005, "loss": 1.378537893295288, "step": 32861 }, { "epoch": 2.590618841150966, "grad_norm": 0.477779358625412, "learning_rate": 0.0005, "loss": 1.4019352197647095, "step": 32862 }, { "epoch": 2.5906976744186045, "grad_norm": 0.5685275793075562, "learning_rate": 0.0005, "loss": 1.555645227432251, "step": 32863 }, { "epoch": 2.5907765076862437, "grad_norm": 0.5152483582496643, "learning_rate": 0.0005, "loss": 1.5399787425994873, "step": 32864 }, { "epoch": 2.5908553409538824, "grad_norm": 0.3972708582878113, "learning_rate": 0.0005, "loss": 1.0802057981491089, "step": 32865 }, { "epoch": 2.5909341742215215, "grad_norm": 0.5285032391548157, "learning_rate": 0.0005, "loss": 1.2957900762557983, "step": 32866 }, { "epoch": 2.5910130074891606, "grad_norm": 0.39701586961746216, "learning_rate": 0.0005, "loss": 1.0709125995635986, "step": 32867 }, { "epoch": 2.5910918407567993, "grad_norm": 0.7336728572845459, "learning_rate": 0.0005, "loss": 1.6328502893447876, "step": 32868 }, { "epoch": 2.5911706740244385, "grad_norm": 0.5194147825241089, "learning_rate": 0.0005, "loss": 1.3821079730987549, "step": 32869 }, { "epoch": 2.591249507292077, "grad_norm": 0.5840884447097778, "learning_rate": 0.0005, "loss": 1.3301770687103271, "step": 32870 }, { "epoch": 2.5913283405597163, "grad_norm": 0.4620538353919983, "learning_rate": 0.0005, "loss": 1.3775458335876465, "step": 32871 }, { "epoch": 2.5914071738273554, "grad_norm": 0.5310845375061035, "learning_rate": 0.0005, "loss": 1.3471567630767822, "step": 32872 }, { "epoch": 2.591486007094994, "grad_norm": 0.4994390606880188, "learning_rate": 0.0005, "loss": 1.2782189846038818, "step": 32873 }, { "epoch": 2.591564840362633, "grad_norm": 0.6243299841880798, "learning_rate": 0.0005, "loss": 1.5769352912902832, "step": 32874 }, { "epoch": 2.591643673630272, "grad_norm": 0.5735763907432556, "learning_rate": 0.0005, "loss": 1.441634178161621, "step": 32875 }, { "epoch": 2.591722506897911, "grad_norm": 0.5518173575401306, "learning_rate": 0.0005, "loss": 1.372131109237671, "step": 32876 }, { "epoch": 2.5918013401655497, "grad_norm": 0.40812692046165466, "learning_rate": 0.0005, "loss": 1.0783274173736572, "step": 32877 }, { "epoch": 2.591880173433189, "grad_norm": 0.6334328055381775, "learning_rate": 0.0005, "loss": 1.4327163696289062, "step": 32878 }, { "epoch": 2.5919590067008276, "grad_norm": 0.4642881155014038, "learning_rate": 0.0005, "loss": 1.3537964820861816, "step": 32879 }, { "epoch": 2.5920378399684667, "grad_norm": 0.4855653643608093, "learning_rate": 0.0005, "loss": 1.2995574474334717, "step": 32880 }, { "epoch": 2.592116673236106, "grad_norm": 0.42903491854667664, "learning_rate": 0.0005, "loss": 1.2020273208618164, "step": 32881 }, { "epoch": 2.5921955065037445, "grad_norm": 0.4481883943080902, "learning_rate": 0.0005, "loss": 1.2574622631072998, "step": 32882 }, { "epoch": 2.5922743397713837, "grad_norm": 0.5080265998840332, "learning_rate": 0.0005, "loss": 1.450066328048706, "step": 32883 }, { "epoch": 2.5923531730390224, "grad_norm": 0.4045892357826233, "learning_rate": 0.0005, "loss": 1.170837640762329, "step": 32884 }, { "epoch": 2.5924320063066615, "grad_norm": 0.48627474904060364, "learning_rate": 0.0005, "loss": 1.4429688453674316, "step": 32885 }, { "epoch": 2.5925108395743006, "grad_norm": 0.5472208857536316, "learning_rate": 0.0005, "loss": 1.1582337617874146, "step": 32886 }, { "epoch": 2.5925896728419393, "grad_norm": 0.4736669063568115, "learning_rate": 0.0005, "loss": 1.2534149885177612, "step": 32887 }, { "epoch": 2.592668506109578, "grad_norm": 0.5273642539978027, "learning_rate": 0.0005, "loss": 1.3029237985610962, "step": 32888 }, { "epoch": 2.592747339377217, "grad_norm": 0.39830178022384644, "learning_rate": 0.0005, "loss": 1.2291481494903564, "step": 32889 }, { "epoch": 2.5928261726448563, "grad_norm": 0.46495380997657776, "learning_rate": 0.0005, "loss": 1.3127999305725098, "step": 32890 }, { "epoch": 2.592905005912495, "grad_norm": 0.41404399275779724, "learning_rate": 0.0005, "loss": 1.2613399028778076, "step": 32891 }, { "epoch": 2.592983839180134, "grad_norm": 0.57384192943573, "learning_rate": 0.0005, "loss": 1.5888206958770752, "step": 32892 }, { "epoch": 2.5930626724477728, "grad_norm": 0.47905588150024414, "learning_rate": 0.0005, "loss": 1.2471191883087158, "step": 32893 }, { "epoch": 2.593141505715412, "grad_norm": 0.5198813676834106, "learning_rate": 0.0005, "loss": 1.3692851066589355, "step": 32894 }, { "epoch": 2.593220338983051, "grad_norm": 0.4220902621746063, "learning_rate": 0.0005, "loss": 1.2800172567367554, "step": 32895 }, { "epoch": 2.5932991722506897, "grad_norm": 0.5793113112449646, "learning_rate": 0.0005, "loss": 1.2015855312347412, "step": 32896 }, { "epoch": 2.593378005518329, "grad_norm": 0.5248187780380249, "learning_rate": 0.0005, "loss": 1.2856138944625854, "step": 32897 }, { "epoch": 2.5934568387859676, "grad_norm": 0.43006280064582825, "learning_rate": 0.0005, "loss": 1.2495793104171753, "step": 32898 }, { "epoch": 2.5935356720536067, "grad_norm": 0.7027475237846375, "learning_rate": 0.0005, "loss": 1.4399117231369019, "step": 32899 }, { "epoch": 2.593614505321246, "grad_norm": 0.47052183747291565, "learning_rate": 0.0005, "loss": 1.456749439239502, "step": 32900 }, { "epoch": 2.5936933385888845, "grad_norm": 0.6154478788375854, "learning_rate": 0.0005, "loss": 1.498530387878418, "step": 32901 }, { "epoch": 2.593772171856523, "grad_norm": 0.4343186318874359, "learning_rate": 0.0005, "loss": 1.2354868650436401, "step": 32902 }, { "epoch": 2.5938510051241623, "grad_norm": 0.7853288054466248, "learning_rate": 0.0005, "loss": 1.4820506572723389, "step": 32903 }, { "epoch": 2.5939298383918015, "grad_norm": 0.5981270670890808, "learning_rate": 0.0005, "loss": 1.4737067222595215, "step": 32904 }, { "epoch": 2.59400867165944, "grad_norm": 0.4904497265815735, "learning_rate": 0.0005, "loss": 1.598230242729187, "step": 32905 }, { "epoch": 2.5940875049270793, "grad_norm": 0.5099135637283325, "learning_rate": 0.0005, "loss": 1.5827279090881348, "step": 32906 }, { "epoch": 2.594166338194718, "grad_norm": 0.45939135551452637, "learning_rate": 0.0005, "loss": 1.4439164400100708, "step": 32907 }, { "epoch": 2.594245171462357, "grad_norm": 0.5917332172393799, "learning_rate": 0.0005, "loss": 1.3354499340057373, "step": 32908 }, { "epoch": 2.5943240047299962, "grad_norm": 0.5288698673248291, "learning_rate": 0.0005, "loss": 1.0697592496871948, "step": 32909 }, { "epoch": 2.594402837997635, "grad_norm": 0.539688766002655, "learning_rate": 0.0005, "loss": 1.2498395442962646, "step": 32910 }, { "epoch": 2.594481671265274, "grad_norm": 0.5728907585144043, "learning_rate": 0.0005, "loss": 1.5024492740631104, "step": 32911 }, { "epoch": 2.5945605045329128, "grad_norm": 0.5610902905464172, "learning_rate": 0.0005, "loss": 1.4618208408355713, "step": 32912 }, { "epoch": 2.594639337800552, "grad_norm": 0.5894826650619507, "learning_rate": 0.0005, "loss": 1.4615797996520996, "step": 32913 }, { "epoch": 2.594718171068191, "grad_norm": 0.48391327261924744, "learning_rate": 0.0005, "loss": 1.1741137504577637, "step": 32914 }, { "epoch": 2.5947970043358297, "grad_norm": 0.5338665246963501, "learning_rate": 0.0005, "loss": 1.1570874452590942, "step": 32915 }, { "epoch": 2.5948758376034684, "grad_norm": 0.5919241905212402, "learning_rate": 0.0005, "loss": 1.530336618423462, "step": 32916 }, { "epoch": 2.5949546708711075, "grad_norm": 0.5245370268821716, "learning_rate": 0.0005, "loss": 1.1482105255126953, "step": 32917 }, { "epoch": 2.5950335041387467, "grad_norm": 0.4568667411804199, "learning_rate": 0.0005, "loss": 1.2841637134552002, "step": 32918 }, { "epoch": 2.5951123374063854, "grad_norm": 0.48792073130607605, "learning_rate": 0.0005, "loss": 1.3523187637329102, "step": 32919 }, { "epoch": 2.5951911706740245, "grad_norm": 0.4769304096698761, "learning_rate": 0.0005, "loss": 1.2229893207550049, "step": 32920 }, { "epoch": 2.595270003941663, "grad_norm": 0.526735246181488, "learning_rate": 0.0005, "loss": 1.530611276626587, "step": 32921 }, { "epoch": 2.5953488372093023, "grad_norm": 0.6068681478500366, "learning_rate": 0.0005, "loss": 1.4447338581085205, "step": 32922 }, { "epoch": 2.5954276704769415, "grad_norm": 0.5285713076591492, "learning_rate": 0.0005, "loss": 1.2145589590072632, "step": 32923 }, { "epoch": 2.59550650374458, "grad_norm": 0.536134660243988, "learning_rate": 0.0005, "loss": 1.4241664409637451, "step": 32924 }, { "epoch": 2.5955853370122193, "grad_norm": 0.4772467315196991, "learning_rate": 0.0005, "loss": 1.3522405624389648, "step": 32925 }, { "epoch": 2.595664170279858, "grad_norm": 0.6448649168014526, "learning_rate": 0.0005, "loss": 1.2244654893875122, "step": 32926 }, { "epoch": 2.595743003547497, "grad_norm": 0.685684859752655, "learning_rate": 0.0005, "loss": 1.441443681716919, "step": 32927 }, { "epoch": 2.5958218368151362, "grad_norm": 0.45075324177742004, "learning_rate": 0.0005, "loss": 1.2116005420684814, "step": 32928 }, { "epoch": 2.595900670082775, "grad_norm": 0.49423903226852417, "learning_rate": 0.0005, "loss": 1.121978998184204, "step": 32929 }, { "epoch": 2.5959795033504136, "grad_norm": 0.47361132502555847, "learning_rate": 0.0005, "loss": 1.2865456342697144, "step": 32930 }, { "epoch": 2.5960583366180527, "grad_norm": 0.5875464677810669, "learning_rate": 0.0005, "loss": 1.390298843383789, "step": 32931 }, { "epoch": 2.596137169885692, "grad_norm": 0.4691731035709381, "learning_rate": 0.0005, "loss": 1.2741727828979492, "step": 32932 }, { "epoch": 2.5962160031533306, "grad_norm": 0.5915358066558838, "learning_rate": 0.0005, "loss": 1.5219743251800537, "step": 32933 }, { "epoch": 2.5962948364209697, "grad_norm": 0.5663368701934814, "learning_rate": 0.0005, "loss": 1.4042695760726929, "step": 32934 }, { "epoch": 2.5963736696886084, "grad_norm": 0.5184509754180908, "learning_rate": 0.0005, "loss": 1.3257513046264648, "step": 32935 }, { "epoch": 2.5964525029562475, "grad_norm": 0.5607261061668396, "learning_rate": 0.0005, "loss": 1.4165868759155273, "step": 32936 }, { "epoch": 2.5965313362238867, "grad_norm": 0.4572397470474243, "learning_rate": 0.0005, "loss": 1.4057410955429077, "step": 32937 }, { "epoch": 2.5966101694915253, "grad_norm": 0.5207986831665039, "learning_rate": 0.0005, "loss": 1.2408673763275146, "step": 32938 }, { "epoch": 2.5966890027591645, "grad_norm": 0.5581598877906799, "learning_rate": 0.0005, "loss": 1.4437437057495117, "step": 32939 }, { "epoch": 2.596767836026803, "grad_norm": 0.5755200386047363, "learning_rate": 0.0005, "loss": 1.285306453704834, "step": 32940 }, { "epoch": 2.5968466692944423, "grad_norm": 0.5277917981147766, "learning_rate": 0.0005, "loss": 1.172431468963623, "step": 32941 }, { "epoch": 2.5969255025620814, "grad_norm": 0.5055504441261292, "learning_rate": 0.0005, "loss": 1.3227275609970093, "step": 32942 }, { "epoch": 2.59700433582972, "grad_norm": 0.6222394108772278, "learning_rate": 0.0005, "loss": 1.3494253158569336, "step": 32943 }, { "epoch": 2.597083169097359, "grad_norm": 0.8740878105163574, "learning_rate": 0.0005, "loss": 1.3299970626831055, "step": 32944 }, { "epoch": 2.597162002364998, "grad_norm": 0.5426328182220459, "learning_rate": 0.0005, "loss": 1.4512213468551636, "step": 32945 }, { "epoch": 2.597240835632637, "grad_norm": 0.5420626997947693, "learning_rate": 0.0005, "loss": 1.2902597188949585, "step": 32946 }, { "epoch": 2.5973196689002758, "grad_norm": 0.5009362101554871, "learning_rate": 0.0005, "loss": 1.404449701309204, "step": 32947 }, { "epoch": 2.597398502167915, "grad_norm": 0.38759204745292664, "learning_rate": 0.0005, "loss": 1.3978755474090576, "step": 32948 }, { "epoch": 2.5974773354355536, "grad_norm": 0.4881783127784729, "learning_rate": 0.0005, "loss": 1.4664286375045776, "step": 32949 }, { "epoch": 2.5975561687031927, "grad_norm": 0.6110832691192627, "learning_rate": 0.0005, "loss": 1.604058027267456, "step": 32950 }, { "epoch": 2.597635001970832, "grad_norm": 0.5552085638046265, "learning_rate": 0.0005, "loss": 1.6079167127609253, "step": 32951 }, { "epoch": 2.5977138352384705, "grad_norm": 0.45090439915657043, "learning_rate": 0.0005, "loss": 1.413374662399292, "step": 32952 }, { "epoch": 2.5977926685061097, "grad_norm": 0.7192748188972473, "learning_rate": 0.0005, "loss": 1.1585460901260376, "step": 32953 }, { "epoch": 2.5978715017737484, "grad_norm": 0.5188924670219421, "learning_rate": 0.0005, "loss": 1.4945019483566284, "step": 32954 }, { "epoch": 2.5979503350413875, "grad_norm": 0.4248398244380951, "learning_rate": 0.0005, "loss": 1.3146777153015137, "step": 32955 }, { "epoch": 2.5980291683090266, "grad_norm": 0.6240667700767517, "learning_rate": 0.0005, "loss": 1.2632488012313843, "step": 32956 }, { "epoch": 2.5981080015766653, "grad_norm": 0.4448879361152649, "learning_rate": 0.0005, "loss": 1.2566527128219604, "step": 32957 }, { "epoch": 2.5981868348443045, "grad_norm": 0.4566618800163269, "learning_rate": 0.0005, "loss": 1.500891923904419, "step": 32958 }, { "epoch": 2.598265668111943, "grad_norm": 0.6040999293327332, "learning_rate": 0.0005, "loss": 1.414815902709961, "step": 32959 }, { "epoch": 2.5983445013795823, "grad_norm": 0.5066860914230347, "learning_rate": 0.0005, "loss": 1.3866955041885376, "step": 32960 }, { "epoch": 2.598423334647221, "grad_norm": 0.46838581562042236, "learning_rate": 0.0005, "loss": 1.3132603168487549, "step": 32961 }, { "epoch": 2.59850216791486, "grad_norm": 0.43296968936920166, "learning_rate": 0.0005, "loss": 1.4245686531066895, "step": 32962 }, { "epoch": 2.598581001182499, "grad_norm": 0.5646836757659912, "learning_rate": 0.0005, "loss": 1.4440670013427734, "step": 32963 }, { "epoch": 2.598659834450138, "grad_norm": 0.651284396648407, "learning_rate": 0.0005, "loss": 1.4440114498138428, "step": 32964 }, { "epoch": 2.598738667717777, "grad_norm": 0.5070505142211914, "learning_rate": 0.0005, "loss": 1.4081528186798096, "step": 32965 }, { "epoch": 2.5988175009854158, "grad_norm": 0.5453545451164246, "learning_rate": 0.0005, "loss": 1.4557440280914307, "step": 32966 }, { "epoch": 2.598896334253055, "grad_norm": 0.7202898859977722, "learning_rate": 0.0005, "loss": 1.3872716426849365, "step": 32967 }, { "epoch": 2.5989751675206936, "grad_norm": 0.6510143280029297, "learning_rate": 0.0005, "loss": 1.5265891551971436, "step": 32968 }, { "epoch": 2.5990540007883327, "grad_norm": 0.49990808963775635, "learning_rate": 0.0005, "loss": 1.2992169857025146, "step": 32969 }, { "epoch": 2.599132834055972, "grad_norm": 0.46893447637557983, "learning_rate": 0.0005, "loss": 1.2221906185150146, "step": 32970 }, { "epoch": 2.5992116673236105, "grad_norm": 0.43101733922958374, "learning_rate": 0.0005, "loss": 1.1750578880310059, "step": 32971 }, { "epoch": 2.5992905005912497, "grad_norm": 0.561326801776886, "learning_rate": 0.0005, "loss": 1.428242802619934, "step": 32972 }, { "epoch": 2.5993693338588884, "grad_norm": 0.507733941078186, "learning_rate": 0.0005, "loss": 1.1891248226165771, "step": 32973 }, { "epoch": 2.5994481671265275, "grad_norm": 0.46978554129600525, "learning_rate": 0.0005, "loss": 1.3565587997436523, "step": 32974 }, { "epoch": 2.599527000394166, "grad_norm": 0.5277109742164612, "learning_rate": 0.0005, "loss": 1.431435227394104, "step": 32975 }, { "epoch": 2.5996058336618053, "grad_norm": 0.5986407995223999, "learning_rate": 0.0005, "loss": 1.3412381410598755, "step": 32976 }, { "epoch": 2.599684666929444, "grad_norm": 0.4764895737171173, "learning_rate": 0.0005, "loss": 1.3114924430847168, "step": 32977 }, { "epoch": 2.599763500197083, "grad_norm": 0.4348020851612091, "learning_rate": 0.0005, "loss": 1.3526594638824463, "step": 32978 }, { "epoch": 2.5998423334647223, "grad_norm": 0.5287104249000549, "learning_rate": 0.0005, "loss": 1.186742901802063, "step": 32979 }, { "epoch": 2.599921166732361, "grad_norm": 0.4857420027256012, "learning_rate": 0.0005, "loss": 1.119002103805542, "step": 32980 }, { "epoch": 2.6, "grad_norm": 0.4896876811981201, "learning_rate": 0.0005, "loss": 1.2406164407730103, "step": 32981 }, { "epoch": 2.600078833267639, "grad_norm": 0.7121270895004272, "learning_rate": 0.0005, "loss": 1.4490809440612793, "step": 32982 }, { "epoch": 2.600157666535278, "grad_norm": 0.5147850513458252, "learning_rate": 0.0005, "loss": 1.3345884084701538, "step": 32983 }, { "epoch": 2.600236499802917, "grad_norm": 0.5294824838638306, "learning_rate": 0.0005, "loss": 1.250586748123169, "step": 32984 }, { "epoch": 2.6003153330705557, "grad_norm": 0.6830931901931763, "learning_rate": 0.0005, "loss": 1.2490992546081543, "step": 32985 }, { "epoch": 2.600394166338195, "grad_norm": 0.5621172189712524, "learning_rate": 0.0005, "loss": 1.3385804891586304, "step": 32986 }, { "epoch": 2.6004729996058336, "grad_norm": 0.659325122833252, "learning_rate": 0.0005, "loss": 1.3198432922363281, "step": 32987 }, { "epoch": 2.6005518328734727, "grad_norm": 0.4494779706001282, "learning_rate": 0.0005, "loss": 1.2568036317825317, "step": 32988 }, { "epoch": 2.600630666141112, "grad_norm": 0.5206201076507568, "learning_rate": 0.0005, "loss": 1.2960522174835205, "step": 32989 }, { "epoch": 2.6007094994087505, "grad_norm": 0.5356907248497009, "learning_rate": 0.0005, "loss": 1.340620994567871, "step": 32990 }, { "epoch": 2.600788332676389, "grad_norm": 0.5216336846351624, "learning_rate": 0.0005, "loss": 1.422181248664856, "step": 32991 }, { "epoch": 2.6008671659440283, "grad_norm": 0.4260856509208679, "learning_rate": 0.0005, "loss": 1.2034482955932617, "step": 32992 }, { "epoch": 2.6009459992116675, "grad_norm": 0.4453059732913971, "learning_rate": 0.0005, "loss": 1.1736979484558105, "step": 32993 }, { "epoch": 2.601024832479306, "grad_norm": 0.4818801283836365, "learning_rate": 0.0005, "loss": 1.231128215789795, "step": 32994 }, { "epoch": 2.6011036657469453, "grad_norm": 0.49270299077033997, "learning_rate": 0.0005, "loss": 1.401120662689209, "step": 32995 }, { "epoch": 2.601182499014584, "grad_norm": 0.47066181898117065, "learning_rate": 0.0005, "loss": 1.3102097511291504, "step": 32996 }, { "epoch": 2.601261332282223, "grad_norm": 0.4830562472343445, "learning_rate": 0.0005, "loss": 1.2516758441925049, "step": 32997 }, { "epoch": 2.6013401655498622, "grad_norm": 0.6381423473358154, "learning_rate": 0.0005, "loss": 1.464031457901001, "step": 32998 }, { "epoch": 2.601418998817501, "grad_norm": 0.5015918612480164, "learning_rate": 0.0005, "loss": 1.2043625116348267, "step": 32999 }, { "epoch": 2.60149783208514, "grad_norm": 0.49142777919769287, "learning_rate": 0.0005, "loss": 1.1808918714523315, "step": 33000 }, { "epoch": 2.6015766653527788, "grad_norm": 0.46424564719200134, "learning_rate": 0.0005, "loss": 1.4775314331054688, "step": 33001 }, { "epoch": 2.601655498620418, "grad_norm": 0.5835293531417847, "learning_rate": 0.0005, "loss": 1.4543678760528564, "step": 33002 }, { "epoch": 2.601734331888057, "grad_norm": 0.4089450240135193, "learning_rate": 0.0005, "loss": 1.0895862579345703, "step": 33003 }, { "epoch": 2.6018131651556957, "grad_norm": 0.5329614877700806, "learning_rate": 0.0005, "loss": 1.381882905960083, "step": 33004 }, { "epoch": 2.6018919984233344, "grad_norm": 0.5834226012229919, "learning_rate": 0.0005, "loss": 1.487191915512085, "step": 33005 }, { "epoch": 2.6019708316909735, "grad_norm": 0.5298959612846375, "learning_rate": 0.0005, "loss": 1.3888095617294312, "step": 33006 }, { "epoch": 2.6020496649586127, "grad_norm": 0.64737468957901, "learning_rate": 0.0005, "loss": 1.5373947620391846, "step": 33007 }, { "epoch": 2.6021284982262514, "grad_norm": 0.49996218085289, "learning_rate": 0.0005, "loss": 1.367553949356079, "step": 33008 }, { "epoch": 2.6022073314938905, "grad_norm": 0.4416816234588623, "learning_rate": 0.0005, "loss": 1.137847661972046, "step": 33009 }, { "epoch": 2.602286164761529, "grad_norm": 0.5129213929176331, "learning_rate": 0.0005, "loss": 1.3884202241897583, "step": 33010 }, { "epoch": 2.6023649980291683, "grad_norm": 0.5458571910858154, "learning_rate": 0.0005, "loss": 1.3405512571334839, "step": 33011 }, { "epoch": 2.6024438312968075, "grad_norm": 0.4520473778247833, "learning_rate": 0.0005, "loss": 1.3278629779815674, "step": 33012 }, { "epoch": 2.602522664564446, "grad_norm": 0.47399234771728516, "learning_rate": 0.0005, "loss": 1.1577717065811157, "step": 33013 }, { "epoch": 2.6026014978320853, "grad_norm": 0.5370376706123352, "learning_rate": 0.0005, "loss": 1.3034169673919678, "step": 33014 }, { "epoch": 2.602680331099724, "grad_norm": 0.5462002158164978, "learning_rate": 0.0005, "loss": 1.2771015167236328, "step": 33015 }, { "epoch": 2.602759164367363, "grad_norm": 0.5035198926925659, "learning_rate": 0.0005, "loss": 1.3544671535491943, "step": 33016 }, { "epoch": 2.6028379976350022, "grad_norm": 0.5032668113708496, "learning_rate": 0.0005, "loss": 1.1570072174072266, "step": 33017 }, { "epoch": 2.602916830902641, "grad_norm": 0.6029572486877441, "learning_rate": 0.0005, "loss": 1.2546665668487549, "step": 33018 }, { "epoch": 2.6029956641702796, "grad_norm": 0.5385835766792297, "learning_rate": 0.0005, "loss": 1.3752586841583252, "step": 33019 }, { "epoch": 2.6030744974379187, "grad_norm": 0.5286155939102173, "learning_rate": 0.0005, "loss": 1.417609691619873, "step": 33020 }, { "epoch": 2.603153330705558, "grad_norm": 0.5174835324287415, "learning_rate": 0.0005, "loss": 1.288840889930725, "step": 33021 }, { "epoch": 2.6032321639731966, "grad_norm": 0.4802687466144562, "learning_rate": 0.0005, "loss": 1.3294636011123657, "step": 33022 }, { "epoch": 2.6033109972408357, "grad_norm": 0.43891990184783936, "learning_rate": 0.0005, "loss": 1.2423137426376343, "step": 33023 }, { "epoch": 2.6033898305084744, "grad_norm": 0.8110542893409729, "learning_rate": 0.0005, "loss": 1.3955676555633545, "step": 33024 }, { "epoch": 2.6034686637761135, "grad_norm": 0.7417337894439697, "learning_rate": 0.0005, "loss": 1.1707783937454224, "step": 33025 }, { "epoch": 2.6035474970437527, "grad_norm": 0.514332115650177, "learning_rate": 0.0005, "loss": 1.3238822221755981, "step": 33026 }, { "epoch": 2.6036263303113913, "grad_norm": 0.48779308795928955, "learning_rate": 0.0005, "loss": 1.3908374309539795, "step": 33027 }, { "epoch": 2.6037051635790305, "grad_norm": 0.4745892286300659, "learning_rate": 0.0005, "loss": 1.3828125, "step": 33028 }, { "epoch": 2.603783996846669, "grad_norm": 0.539544403553009, "learning_rate": 0.0005, "loss": 1.2092125415802002, "step": 33029 }, { "epoch": 2.6038628301143083, "grad_norm": 0.510694146156311, "learning_rate": 0.0005, "loss": 1.474637508392334, "step": 33030 }, { "epoch": 2.6039416633819474, "grad_norm": 0.6043664813041687, "learning_rate": 0.0005, "loss": 1.3761327266693115, "step": 33031 }, { "epoch": 2.604020496649586, "grad_norm": 0.5955190062522888, "learning_rate": 0.0005, "loss": 1.3572180271148682, "step": 33032 }, { "epoch": 2.604099329917225, "grad_norm": 0.5079712271690369, "learning_rate": 0.0005, "loss": 1.4884846210479736, "step": 33033 }, { "epoch": 2.604178163184864, "grad_norm": 0.578249454498291, "learning_rate": 0.0005, "loss": 1.4436283111572266, "step": 33034 }, { "epoch": 2.604256996452503, "grad_norm": 0.5685158967971802, "learning_rate": 0.0005, "loss": 1.296297311782837, "step": 33035 }, { "epoch": 2.6043358297201418, "grad_norm": 0.4769502580165863, "learning_rate": 0.0005, "loss": 1.3899445533752441, "step": 33036 }, { "epoch": 2.604414662987781, "grad_norm": 0.4742392897605896, "learning_rate": 0.0005, "loss": 1.305938959121704, "step": 33037 }, { "epoch": 2.6044934962554196, "grad_norm": 0.5448983311653137, "learning_rate": 0.0005, "loss": 1.519632339477539, "step": 33038 }, { "epoch": 2.6045723295230587, "grad_norm": 0.45651406049728394, "learning_rate": 0.0005, "loss": 1.0723869800567627, "step": 33039 }, { "epoch": 2.604651162790698, "grad_norm": 0.5783329606056213, "learning_rate": 0.0005, "loss": 1.5182476043701172, "step": 33040 }, { "epoch": 2.6047299960583365, "grad_norm": 0.6573101282119751, "learning_rate": 0.0005, "loss": 1.659970998764038, "step": 33041 }, { "epoch": 2.6048088293259757, "grad_norm": 0.48707786202430725, "learning_rate": 0.0005, "loss": 1.3684465885162354, "step": 33042 }, { "epoch": 2.6048876625936144, "grad_norm": 0.5666033029556274, "learning_rate": 0.0005, "loss": 1.6547253131866455, "step": 33043 }, { "epoch": 2.6049664958612535, "grad_norm": 0.46461451053619385, "learning_rate": 0.0005, "loss": 1.4432222843170166, "step": 33044 }, { "epoch": 2.6050453291288926, "grad_norm": 0.5228970646858215, "learning_rate": 0.0005, "loss": 1.4242382049560547, "step": 33045 }, { "epoch": 2.6051241623965313, "grad_norm": 0.4573476016521454, "learning_rate": 0.0005, "loss": 1.1347284317016602, "step": 33046 }, { "epoch": 2.60520299566417, "grad_norm": 0.6113069653511047, "learning_rate": 0.0005, "loss": 1.181327223777771, "step": 33047 }, { "epoch": 2.605281828931809, "grad_norm": 0.45340847969055176, "learning_rate": 0.0005, "loss": 1.168207049369812, "step": 33048 }, { "epoch": 2.6053606621994483, "grad_norm": 0.480045884847641, "learning_rate": 0.0005, "loss": 1.520430326461792, "step": 33049 }, { "epoch": 2.605439495467087, "grad_norm": 0.4949481785297394, "learning_rate": 0.0005, "loss": 1.3574495315551758, "step": 33050 }, { "epoch": 2.605518328734726, "grad_norm": 0.7017534971237183, "learning_rate": 0.0005, "loss": 1.3854461908340454, "step": 33051 }, { "epoch": 2.605597162002365, "grad_norm": 0.5226337313652039, "learning_rate": 0.0005, "loss": 1.2451812028884888, "step": 33052 }, { "epoch": 2.605675995270004, "grad_norm": 0.4581794738769531, "learning_rate": 0.0005, "loss": 1.1243808269500732, "step": 33053 }, { "epoch": 2.605754828537643, "grad_norm": 0.44526365399360657, "learning_rate": 0.0005, "loss": 1.1930749416351318, "step": 33054 }, { "epoch": 2.6058336618052818, "grad_norm": 0.5400126576423645, "learning_rate": 0.0005, "loss": 1.4162602424621582, "step": 33055 }, { "epoch": 2.605912495072921, "grad_norm": 0.537682831287384, "learning_rate": 0.0005, "loss": 1.2817270755767822, "step": 33056 }, { "epoch": 2.6059913283405596, "grad_norm": 0.540677547454834, "learning_rate": 0.0005, "loss": 1.295871615409851, "step": 33057 }, { "epoch": 2.6060701616081987, "grad_norm": 0.5299403667449951, "learning_rate": 0.0005, "loss": 1.7070534229278564, "step": 33058 }, { "epoch": 2.606148994875838, "grad_norm": 0.5839768648147583, "learning_rate": 0.0005, "loss": 1.6447157859802246, "step": 33059 }, { "epoch": 2.6062278281434765, "grad_norm": 0.47973763942718506, "learning_rate": 0.0005, "loss": 1.1927295923233032, "step": 33060 }, { "epoch": 2.606306661411115, "grad_norm": 0.5650421380996704, "learning_rate": 0.0005, "loss": 1.309861421585083, "step": 33061 }, { "epoch": 2.6063854946787544, "grad_norm": 0.4833071231842041, "learning_rate": 0.0005, "loss": 1.338225245475769, "step": 33062 }, { "epoch": 2.6064643279463935, "grad_norm": 0.6041005849838257, "learning_rate": 0.0005, "loss": 1.2194814682006836, "step": 33063 }, { "epoch": 2.606543161214032, "grad_norm": 0.6166638731956482, "learning_rate": 0.0005, "loss": 1.6692678928375244, "step": 33064 }, { "epoch": 2.6066219944816713, "grad_norm": 0.5082882046699524, "learning_rate": 0.0005, "loss": 1.3913984298706055, "step": 33065 }, { "epoch": 2.60670082774931, "grad_norm": 21.01670265197754, "learning_rate": 0.0005, "loss": 1.2994029521942139, "step": 33066 }, { "epoch": 2.606779661016949, "grad_norm": 0.6752678155899048, "learning_rate": 0.0005, "loss": 1.442111611366272, "step": 33067 }, { "epoch": 2.6068584942845883, "grad_norm": 0.57827228307724, "learning_rate": 0.0005, "loss": 1.3103350400924683, "step": 33068 }, { "epoch": 2.606937327552227, "grad_norm": 0.5228431820869446, "learning_rate": 0.0005, "loss": 1.2986998558044434, "step": 33069 }, { "epoch": 2.607016160819866, "grad_norm": 0.5018703937530518, "learning_rate": 0.0005, "loss": 1.398921251296997, "step": 33070 }, { "epoch": 2.607094994087505, "grad_norm": 0.5243483185768127, "learning_rate": 0.0005, "loss": 1.25922429561615, "step": 33071 }, { "epoch": 2.607173827355144, "grad_norm": 0.47102949023246765, "learning_rate": 0.0005, "loss": 1.2618556022644043, "step": 33072 }, { "epoch": 2.607252660622783, "grad_norm": 0.665993332862854, "learning_rate": 0.0005, "loss": 1.3422871828079224, "step": 33073 }, { "epoch": 2.6073314938904217, "grad_norm": 0.4851827919483185, "learning_rate": 0.0005, "loss": 1.358067274093628, "step": 33074 }, { "epoch": 2.6074103271580604, "grad_norm": 0.44165751338005066, "learning_rate": 0.0005, "loss": 1.21592116355896, "step": 33075 }, { "epoch": 2.6074891604256996, "grad_norm": 0.47644752264022827, "learning_rate": 0.0005, "loss": 1.3685789108276367, "step": 33076 }, { "epoch": 2.6075679936933387, "grad_norm": 0.48150119185447693, "learning_rate": 0.0005, "loss": 1.324134349822998, "step": 33077 }, { "epoch": 2.6076468269609774, "grad_norm": 0.5562043786048889, "learning_rate": 0.0005, "loss": 1.558082938194275, "step": 33078 }, { "epoch": 2.6077256602286165, "grad_norm": 0.5709496736526489, "learning_rate": 0.0005, "loss": 1.3882567882537842, "step": 33079 }, { "epoch": 2.607804493496255, "grad_norm": 0.45654767751693726, "learning_rate": 0.0005, "loss": 1.2286734580993652, "step": 33080 }, { "epoch": 2.6078833267638943, "grad_norm": 0.47821226716041565, "learning_rate": 0.0005, "loss": 1.5256547927856445, "step": 33081 }, { "epoch": 2.6079621600315335, "grad_norm": 0.45334237813949585, "learning_rate": 0.0005, "loss": 1.3989063501358032, "step": 33082 }, { "epoch": 2.608040993299172, "grad_norm": 0.399745911359787, "learning_rate": 0.0005, "loss": 1.139870285987854, "step": 33083 }, { "epoch": 2.6081198265668113, "grad_norm": 0.48161813616752625, "learning_rate": 0.0005, "loss": 1.2983875274658203, "step": 33084 }, { "epoch": 2.60819865983445, "grad_norm": 0.5125957131385803, "learning_rate": 0.0005, "loss": 1.3515771627426147, "step": 33085 }, { "epoch": 2.608277493102089, "grad_norm": 0.43515339493751526, "learning_rate": 0.0005, "loss": 1.2507067918777466, "step": 33086 }, { "epoch": 2.6083563263697283, "grad_norm": 0.6687605381011963, "learning_rate": 0.0005, "loss": 1.4071085453033447, "step": 33087 }, { "epoch": 2.608435159637367, "grad_norm": 0.4711235761642456, "learning_rate": 0.0005, "loss": 1.4243855476379395, "step": 33088 }, { "epoch": 2.608513992905006, "grad_norm": 0.628331184387207, "learning_rate": 0.0005, "loss": 1.2636414766311646, "step": 33089 }, { "epoch": 2.6085928261726448, "grad_norm": 0.530157744884491, "learning_rate": 0.0005, "loss": 1.4559838771820068, "step": 33090 }, { "epoch": 2.608671659440284, "grad_norm": 0.6665238738059998, "learning_rate": 0.0005, "loss": 1.6067931652069092, "step": 33091 }, { "epoch": 2.6087504927079226, "grad_norm": 0.5060410499572754, "learning_rate": 0.0005, "loss": 1.3732666969299316, "step": 33092 }, { "epoch": 2.6088293259755617, "grad_norm": 0.57161545753479, "learning_rate": 0.0005, "loss": 1.33282470703125, "step": 33093 }, { "epoch": 2.6089081592432004, "grad_norm": 0.4891521632671356, "learning_rate": 0.0005, "loss": 1.5000972747802734, "step": 33094 }, { "epoch": 2.6089869925108395, "grad_norm": 0.43621861934661865, "learning_rate": 0.0005, "loss": 1.3685274124145508, "step": 33095 }, { "epoch": 2.6090658257784787, "grad_norm": 0.559556245803833, "learning_rate": 0.0005, "loss": 1.392701268196106, "step": 33096 }, { "epoch": 2.6091446590461174, "grad_norm": 0.4999887943267822, "learning_rate": 0.0005, "loss": 1.3485853672027588, "step": 33097 }, { "epoch": 2.6092234923137565, "grad_norm": 0.6054799556732178, "learning_rate": 0.0005, "loss": 1.4270857572555542, "step": 33098 }, { "epoch": 2.609302325581395, "grad_norm": 0.47081080079078674, "learning_rate": 0.0005, "loss": 1.2808983325958252, "step": 33099 }, { "epoch": 2.6093811588490343, "grad_norm": 0.5134462714195251, "learning_rate": 0.0005, "loss": 1.270695447921753, "step": 33100 }, { "epoch": 2.6094599921166735, "grad_norm": 0.47046521306037903, "learning_rate": 0.0005, "loss": 1.2988204956054688, "step": 33101 }, { "epoch": 2.609538825384312, "grad_norm": 0.48025646805763245, "learning_rate": 0.0005, "loss": 1.2961684465408325, "step": 33102 }, { "epoch": 2.6096176586519513, "grad_norm": 0.6131063103675842, "learning_rate": 0.0005, "loss": 1.4296555519104004, "step": 33103 }, { "epoch": 2.60969649191959, "grad_norm": 0.49125614762306213, "learning_rate": 0.0005, "loss": 1.3735944032669067, "step": 33104 }, { "epoch": 2.609775325187229, "grad_norm": 0.5435960292816162, "learning_rate": 0.0005, "loss": 1.5127630233764648, "step": 33105 }, { "epoch": 2.609854158454868, "grad_norm": 1.063035011291504, "learning_rate": 0.0005, "loss": 1.3393888473510742, "step": 33106 }, { "epoch": 2.609932991722507, "grad_norm": 0.5234800577163696, "learning_rate": 0.0005, "loss": 1.5211305618286133, "step": 33107 }, { "epoch": 2.6100118249901456, "grad_norm": 0.5984623432159424, "learning_rate": 0.0005, "loss": 1.2825713157653809, "step": 33108 }, { "epoch": 2.6100906582577847, "grad_norm": 0.5797029733657837, "learning_rate": 0.0005, "loss": 1.333644151687622, "step": 33109 }, { "epoch": 2.610169491525424, "grad_norm": 0.4804941415786743, "learning_rate": 0.0005, "loss": 1.3459491729736328, "step": 33110 }, { "epoch": 2.6102483247930626, "grad_norm": 0.440540075302124, "learning_rate": 0.0005, "loss": 1.3728535175323486, "step": 33111 }, { "epoch": 2.6103271580607017, "grad_norm": 0.5110069513320923, "learning_rate": 0.0005, "loss": 1.2842504978179932, "step": 33112 }, { "epoch": 2.6104059913283404, "grad_norm": 0.4645403027534485, "learning_rate": 0.0005, "loss": 1.3949226140975952, "step": 33113 }, { "epoch": 2.6104848245959795, "grad_norm": 0.4995853006839752, "learning_rate": 0.0005, "loss": 1.479567050933838, "step": 33114 }, { "epoch": 2.6105636578636187, "grad_norm": 0.5454042553901672, "learning_rate": 0.0005, "loss": 1.3987147808074951, "step": 33115 }, { "epoch": 2.6106424911312573, "grad_norm": 0.5072004199028015, "learning_rate": 0.0005, "loss": 1.2381268739700317, "step": 33116 }, { "epoch": 2.6107213243988965, "grad_norm": 0.5498201847076416, "learning_rate": 0.0005, "loss": 1.527045488357544, "step": 33117 }, { "epoch": 2.610800157666535, "grad_norm": 0.5665250420570374, "learning_rate": 0.0005, "loss": 1.3331595659255981, "step": 33118 }, { "epoch": 2.6108789909341743, "grad_norm": 0.625549852848053, "learning_rate": 0.0005, "loss": 1.5061408281326294, "step": 33119 }, { "epoch": 2.6109578242018134, "grad_norm": 0.5408117771148682, "learning_rate": 0.0005, "loss": 1.373936414718628, "step": 33120 }, { "epoch": 2.611036657469452, "grad_norm": 0.5211285948753357, "learning_rate": 0.0005, "loss": 1.3261449337005615, "step": 33121 }, { "epoch": 2.611115490737091, "grad_norm": 0.5844464898109436, "learning_rate": 0.0005, "loss": 1.5238652229309082, "step": 33122 }, { "epoch": 2.61119432400473, "grad_norm": 0.43551939725875854, "learning_rate": 0.0005, "loss": 1.231722116470337, "step": 33123 }, { "epoch": 2.611273157272369, "grad_norm": 0.5729880928993225, "learning_rate": 0.0005, "loss": 1.3007593154907227, "step": 33124 }, { "epoch": 2.6113519905400078, "grad_norm": 0.5094028115272522, "learning_rate": 0.0005, "loss": 1.4666564464569092, "step": 33125 }, { "epoch": 2.611430823807647, "grad_norm": 0.6433104276657104, "learning_rate": 0.0005, "loss": 1.4613759517669678, "step": 33126 }, { "epoch": 2.6115096570752856, "grad_norm": 0.533022403717041, "learning_rate": 0.0005, "loss": 1.5835857391357422, "step": 33127 }, { "epoch": 2.6115884903429247, "grad_norm": 0.5749051570892334, "learning_rate": 0.0005, "loss": 1.5084242820739746, "step": 33128 }, { "epoch": 2.611667323610564, "grad_norm": 0.4874747693538666, "learning_rate": 0.0005, "loss": 1.4987130165100098, "step": 33129 }, { "epoch": 2.6117461568782026, "grad_norm": 0.4564751386642456, "learning_rate": 0.0005, "loss": 1.2545636892318726, "step": 33130 }, { "epoch": 2.6118249901458417, "grad_norm": 0.5518471002578735, "learning_rate": 0.0005, "loss": 1.1219816207885742, "step": 33131 }, { "epoch": 2.6119038234134804, "grad_norm": 0.5197895765304565, "learning_rate": 0.0005, "loss": 1.2248717546463013, "step": 33132 }, { "epoch": 2.6119826566811195, "grad_norm": 0.4252409040927887, "learning_rate": 0.0005, "loss": 1.2834168672561646, "step": 33133 }, { "epoch": 2.6120614899487586, "grad_norm": 0.6135594248771667, "learning_rate": 0.0005, "loss": 1.4264836311340332, "step": 33134 }, { "epoch": 2.6121403232163973, "grad_norm": 0.5665282011032104, "learning_rate": 0.0005, "loss": 1.3680071830749512, "step": 33135 }, { "epoch": 2.612219156484036, "grad_norm": 0.5396360158920288, "learning_rate": 0.0005, "loss": 1.2215297222137451, "step": 33136 }, { "epoch": 2.612297989751675, "grad_norm": 0.5504568219184875, "learning_rate": 0.0005, "loss": 1.3080322742462158, "step": 33137 }, { "epoch": 2.6123768230193143, "grad_norm": 0.5542226433753967, "learning_rate": 0.0005, "loss": 1.4400346279144287, "step": 33138 }, { "epoch": 2.612455656286953, "grad_norm": 0.568565845489502, "learning_rate": 0.0005, "loss": 1.4538456201553345, "step": 33139 }, { "epoch": 2.612534489554592, "grad_norm": 0.5467010736465454, "learning_rate": 0.0005, "loss": 1.4827251434326172, "step": 33140 }, { "epoch": 2.612613322822231, "grad_norm": 0.540549099445343, "learning_rate": 0.0005, "loss": 1.2130457162857056, "step": 33141 }, { "epoch": 2.61269215608987, "grad_norm": 0.8309035897254944, "learning_rate": 0.0005, "loss": 1.388201117515564, "step": 33142 }, { "epoch": 2.612770989357509, "grad_norm": 0.5435320138931274, "learning_rate": 0.0005, "loss": 1.5540897846221924, "step": 33143 }, { "epoch": 2.6128498226251478, "grad_norm": 0.4072953760623932, "learning_rate": 0.0005, "loss": 1.0392825603485107, "step": 33144 }, { "epoch": 2.612928655892787, "grad_norm": 0.581782341003418, "learning_rate": 0.0005, "loss": 1.616323471069336, "step": 33145 }, { "epoch": 2.6130074891604256, "grad_norm": 0.521102249622345, "learning_rate": 0.0005, "loss": 1.1846749782562256, "step": 33146 }, { "epoch": 2.6130863224280647, "grad_norm": 0.5276725888252258, "learning_rate": 0.0005, "loss": 1.4593725204467773, "step": 33147 }, { "epoch": 2.613165155695704, "grad_norm": 3.6576919555664062, "learning_rate": 0.0005, "loss": 1.1660242080688477, "step": 33148 }, { "epoch": 2.6132439889633425, "grad_norm": 0.5135966539382935, "learning_rate": 0.0005, "loss": 1.2366400957107544, "step": 33149 }, { "epoch": 2.6133228222309812, "grad_norm": 0.6146278977394104, "learning_rate": 0.0005, "loss": 1.570873737335205, "step": 33150 }, { "epoch": 2.6134016554986204, "grad_norm": 0.5005632638931274, "learning_rate": 0.0005, "loss": 1.298348069190979, "step": 33151 }, { "epoch": 2.6134804887662595, "grad_norm": 0.4738679528236389, "learning_rate": 0.0005, "loss": 1.088201642036438, "step": 33152 }, { "epoch": 2.613559322033898, "grad_norm": 0.507039487361908, "learning_rate": 0.0005, "loss": 1.3823504447937012, "step": 33153 }, { "epoch": 2.6136381553015373, "grad_norm": 0.46305328607559204, "learning_rate": 0.0005, "loss": 1.3565781116485596, "step": 33154 }, { "epoch": 2.613716988569176, "grad_norm": 0.4870498478412628, "learning_rate": 0.0005, "loss": 1.2161617279052734, "step": 33155 }, { "epoch": 2.613795821836815, "grad_norm": 0.6603094935417175, "learning_rate": 0.0005, "loss": 1.4456701278686523, "step": 33156 }, { "epoch": 2.6138746551044543, "grad_norm": 0.46725010871887207, "learning_rate": 0.0005, "loss": 1.374194860458374, "step": 33157 }, { "epoch": 2.613953488372093, "grad_norm": 0.4239669144153595, "learning_rate": 0.0005, "loss": 1.34749174118042, "step": 33158 }, { "epoch": 2.614032321639732, "grad_norm": 0.6232500076293945, "learning_rate": 0.0005, "loss": 1.2431094646453857, "step": 33159 }, { "epoch": 2.614111154907371, "grad_norm": 0.5220555067062378, "learning_rate": 0.0005, "loss": 1.2660400867462158, "step": 33160 }, { "epoch": 2.61418998817501, "grad_norm": 0.42575129866600037, "learning_rate": 0.0005, "loss": 1.330287218093872, "step": 33161 }, { "epoch": 2.614268821442649, "grad_norm": 0.5258126258850098, "learning_rate": 0.0005, "loss": 1.503782033920288, "step": 33162 }, { "epoch": 2.6143476547102877, "grad_norm": 0.5136097073554993, "learning_rate": 0.0005, "loss": 1.41843581199646, "step": 33163 }, { "epoch": 2.6144264879779264, "grad_norm": 0.5174190998077393, "learning_rate": 0.0005, "loss": 1.3389630317687988, "step": 33164 }, { "epoch": 2.6145053212455656, "grad_norm": 0.4686427116394043, "learning_rate": 0.0005, "loss": 1.2249276638031006, "step": 33165 }, { "epoch": 2.6145841545132047, "grad_norm": 0.44607990980148315, "learning_rate": 0.0005, "loss": 1.6142611503601074, "step": 33166 }, { "epoch": 2.6146629877808434, "grad_norm": 0.42079752683639526, "learning_rate": 0.0005, "loss": 1.1663814783096313, "step": 33167 }, { "epoch": 2.6147418210484825, "grad_norm": 0.4387598931789398, "learning_rate": 0.0005, "loss": 1.2503888607025146, "step": 33168 }, { "epoch": 2.614820654316121, "grad_norm": 0.8282942175865173, "learning_rate": 0.0005, "loss": 1.392911434173584, "step": 33169 }, { "epoch": 2.6148994875837603, "grad_norm": 0.37685060501098633, "learning_rate": 0.0005, "loss": 1.1520072221755981, "step": 33170 }, { "epoch": 2.6149783208513995, "grad_norm": 0.4950661063194275, "learning_rate": 0.0005, "loss": 1.2891731262207031, "step": 33171 }, { "epoch": 2.615057154119038, "grad_norm": 0.5250833034515381, "learning_rate": 0.0005, "loss": 1.3997358083724976, "step": 33172 }, { "epoch": 2.6151359873866773, "grad_norm": 0.5327101349830627, "learning_rate": 0.0005, "loss": 1.398468255996704, "step": 33173 }, { "epoch": 2.615214820654316, "grad_norm": 0.5122814774513245, "learning_rate": 0.0005, "loss": 1.2572342157363892, "step": 33174 }, { "epoch": 2.615293653921955, "grad_norm": 0.4396073520183563, "learning_rate": 0.0005, "loss": 1.262373924255371, "step": 33175 }, { "epoch": 2.6153724871895943, "grad_norm": 0.4392319619655609, "learning_rate": 0.0005, "loss": 1.0834749937057495, "step": 33176 }, { "epoch": 2.615451320457233, "grad_norm": 0.5222888588905334, "learning_rate": 0.0005, "loss": 1.4229897260665894, "step": 33177 }, { "epoch": 2.6155301537248716, "grad_norm": 0.5193818807601929, "learning_rate": 0.0005, "loss": 1.338463306427002, "step": 33178 }, { "epoch": 2.6156089869925108, "grad_norm": 0.6187266111373901, "learning_rate": 0.0005, "loss": 1.3286139965057373, "step": 33179 }, { "epoch": 2.61568782026015, "grad_norm": 0.5879215002059937, "learning_rate": 0.0005, "loss": 1.3018436431884766, "step": 33180 }, { "epoch": 2.6157666535277886, "grad_norm": 0.4820138216018677, "learning_rate": 0.0005, "loss": 1.3483314514160156, "step": 33181 }, { "epoch": 2.6158454867954277, "grad_norm": 0.6304450631141663, "learning_rate": 0.0005, "loss": 1.3342703580856323, "step": 33182 }, { "epoch": 2.6159243200630664, "grad_norm": 0.4302215278148651, "learning_rate": 0.0005, "loss": 1.1582298278808594, "step": 33183 }, { "epoch": 2.6160031533307055, "grad_norm": 0.5952346920967102, "learning_rate": 0.0005, "loss": 1.44622802734375, "step": 33184 }, { "epoch": 2.6160819865983447, "grad_norm": 0.4299621880054474, "learning_rate": 0.0005, "loss": 1.1892025470733643, "step": 33185 }, { "epoch": 2.6161608198659834, "grad_norm": 0.515381932258606, "learning_rate": 0.0005, "loss": 1.2496705055236816, "step": 33186 }, { "epoch": 2.6162396531336225, "grad_norm": 0.6271296739578247, "learning_rate": 0.0005, "loss": 1.3656264543533325, "step": 33187 }, { "epoch": 2.616318486401261, "grad_norm": 0.5093989968299866, "learning_rate": 0.0005, "loss": 1.27229642868042, "step": 33188 }, { "epoch": 2.6163973196689003, "grad_norm": 0.4410238265991211, "learning_rate": 0.0005, "loss": 1.2504545450210571, "step": 33189 }, { "epoch": 2.6164761529365395, "grad_norm": 0.4986982047557831, "learning_rate": 0.0005, "loss": 1.1491888761520386, "step": 33190 }, { "epoch": 2.616554986204178, "grad_norm": 0.6566145420074463, "learning_rate": 0.0005, "loss": 1.4213095903396606, "step": 33191 }, { "epoch": 2.616633819471817, "grad_norm": 0.47671109437942505, "learning_rate": 0.0005, "loss": 1.2659831047058105, "step": 33192 }, { "epoch": 2.616712652739456, "grad_norm": 0.4471876323223114, "learning_rate": 0.0005, "loss": 1.495803952217102, "step": 33193 }, { "epoch": 2.616791486007095, "grad_norm": 0.6483893990516663, "learning_rate": 0.0005, "loss": 1.2509765625, "step": 33194 }, { "epoch": 2.616870319274734, "grad_norm": 0.47618958353996277, "learning_rate": 0.0005, "loss": 1.4198687076568604, "step": 33195 }, { "epoch": 2.616949152542373, "grad_norm": 0.6776336431503296, "learning_rate": 0.0005, "loss": 1.451201319694519, "step": 33196 }, { "epoch": 2.6170279858100116, "grad_norm": 0.9143702387809753, "learning_rate": 0.0005, "loss": 1.2471318244934082, "step": 33197 }, { "epoch": 2.6171068190776507, "grad_norm": 0.4970910847187042, "learning_rate": 0.0005, "loss": 1.1678707599639893, "step": 33198 }, { "epoch": 2.61718565234529, "grad_norm": 0.4080824851989746, "learning_rate": 0.0005, "loss": 1.310100793838501, "step": 33199 }, { "epoch": 2.6172644856129286, "grad_norm": 0.5453002452850342, "learning_rate": 0.0005, "loss": 1.4973185062408447, "step": 33200 }, { "epoch": 2.6173433188805677, "grad_norm": 0.6699541807174683, "learning_rate": 0.0005, "loss": 1.562092661857605, "step": 33201 }, { "epoch": 2.6174221521482064, "grad_norm": 0.5689789056777954, "learning_rate": 0.0005, "loss": 1.5019636154174805, "step": 33202 }, { "epoch": 2.6175009854158455, "grad_norm": 0.5452170968055725, "learning_rate": 0.0005, "loss": 1.5042552947998047, "step": 33203 }, { "epoch": 2.6175798186834847, "grad_norm": 0.4893738627433777, "learning_rate": 0.0005, "loss": 1.3600385189056396, "step": 33204 }, { "epoch": 2.6176586519511233, "grad_norm": 0.47708994150161743, "learning_rate": 0.0005, "loss": 1.3308048248291016, "step": 33205 }, { "epoch": 2.6177374852187625, "grad_norm": 0.4608895182609558, "learning_rate": 0.0005, "loss": 1.1196186542510986, "step": 33206 }, { "epoch": 2.617816318486401, "grad_norm": 0.526748776435852, "learning_rate": 0.0005, "loss": 1.005804181098938, "step": 33207 }, { "epoch": 2.6178951517540403, "grad_norm": 0.41284507513046265, "learning_rate": 0.0005, "loss": 1.1162679195404053, "step": 33208 }, { "epoch": 2.617973985021679, "grad_norm": 0.5626603364944458, "learning_rate": 0.0005, "loss": 1.5743417739868164, "step": 33209 }, { "epoch": 2.618052818289318, "grad_norm": 0.4969955086708069, "learning_rate": 0.0005, "loss": 1.32199227809906, "step": 33210 }, { "epoch": 2.618131651556957, "grad_norm": 0.4504487216472626, "learning_rate": 0.0005, "loss": 1.4600249528884888, "step": 33211 }, { "epoch": 2.618210484824596, "grad_norm": 0.4357307553291321, "learning_rate": 0.0005, "loss": 1.2518351078033447, "step": 33212 }, { "epoch": 2.618289318092235, "grad_norm": 0.558799684047699, "learning_rate": 0.0005, "loss": 1.2767119407653809, "step": 33213 }, { "epoch": 2.6183681513598738, "grad_norm": 0.6248811483383179, "learning_rate": 0.0005, "loss": 1.3596996068954468, "step": 33214 }, { "epoch": 2.618446984627513, "grad_norm": 0.6177336573600769, "learning_rate": 0.0005, "loss": 1.522209882736206, "step": 33215 }, { "epoch": 2.6185258178951516, "grad_norm": 0.5157703757286072, "learning_rate": 0.0005, "loss": 1.4443656206130981, "step": 33216 }, { "epoch": 2.6186046511627907, "grad_norm": 0.4256075322628021, "learning_rate": 0.0005, "loss": 1.1768379211425781, "step": 33217 }, { "epoch": 2.61868348443043, "grad_norm": 0.48101773858070374, "learning_rate": 0.0005, "loss": 1.4179142713546753, "step": 33218 }, { "epoch": 2.6187623176980686, "grad_norm": 0.49215805530548096, "learning_rate": 0.0005, "loss": 1.4129501581192017, "step": 33219 }, { "epoch": 2.6188411509657077, "grad_norm": 1.2111483812332153, "learning_rate": 0.0005, "loss": 1.1955840587615967, "step": 33220 }, { "epoch": 2.6189199842333464, "grad_norm": 0.5561094880104065, "learning_rate": 0.0005, "loss": 1.6986615657806396, "step": 33221 }, { "epoch": 2.6189988175009855, "grad_norm": 0.46886834502220154, "learning_rate": 0.0005, "loss": 1.1760979890823364, "step": 33222 }, { "epoch": 2.619077650768624, "grad_norm": 0.5861800909042358, "learning_rate": 0.0005, "loss": 1.3117408752441406, "step": 33223 }, { "epoch": 2.6191564840362633, "grad_norm": 0.5862710475921631, "learning_rate": 0.0005, "loss": 1.2067911624908447, "step": 33224 }, { "epoch": 2.619235317303902, "grad_norm": 0.8094763159751892, "learning_rate": 0.0005, "loss": 1.2501920461654663, "step": 33225 }, { "epoch": 2.619314150571541, "grad_norm": 0.5467047095298767, "learning_rate": 0.0005, "loss": 1.285435676574707, "step": 33226 }, { "epoch": 2.6193929838391803, "grad_norm": 0.5042670965194702, "learning_rate": 0.0005, "loss": 1.3512558937072754, "step": 33227 }, { "epoch": 2.619471817106819, "grad_norm": 0.4374619722366333, "learning_rate": 0.0005, "loss": 1.0925418138504028, "step": 33228 }, { "epoch": 2.619550650374458, "grad_norm": 0.5411364436149597, "learning_rate": 0.0005, "loss": 1.367732286453247, "step": 33229 }, { "epoch": 2.619629483642097, "grad_norm": 0.5273290872573853, "learning_rate": 0.0005, "loss": 1.2982269525527954, "step": 33230 }, { "epoch": 2.619708316909736, "grad_norm": 0.5489985942840576, "learning_rate": 0.0005, "loss": 1.4090890884399414, "step": 33231 }, { "epoch": 2.619787150177375, "grad_norm": 0.4938558340072632, "learning_rate": 0.0005, "loss": 1.2902400493621826, "step": 33232 }, { "epoch": 2.6198659834450138, "grad_norm": 0.576500654220581, "learning_rate": 0.0005, "loss": 1.3715708255767822, "step": 33233 }, { "epoch": 2.619944816712653, "grad_norm": 0.5434507727622986, "learning_rate": 0.0005, "loss": 1.233542799949646, "step": 33234 }, { "epoch": 2.6200236499802916, "grad_norm": 0.5575529336929321, "learning_rate": 0.0005, "loss": 1.5360972881317139, "step": 33235 }, { "epoch": 2.6201024832479307, "grad_norm": 0.43582209944725037, "learning_rate": 0.0005, "loss": 1.345378041267395, "step": 33236 }, { "epoch": 2.62018131651557, "grad_norm": 0.7329895496368408, "learning_rate": 0.0005, "loss": 1.2792377471923828, "step": 33237 }, { "epoch": 2.6202601497832085, "grad_norm": 0.5750635862350464, "learning_rate": 0.0005, "loss": 1.1787291765213013, "step": 33238 }, { "epoch": 2.6203389830508472, "grad_norm": 0.5803742408752441, "learning_rate": 0.0005, "loss": 1.3497250080108643, "step": 33239 }, { "epoch": 2.6204178163184864, "grad_norm": 0.5105057954788208, "learning_rate": 0.0005, "loss": 1.2023043632507324, "step": 33240 }, { "epoch": 2.6204966495861255, "grad_norm": 0.5853933691978455, "learning_rate": 0.0005, "loss": 1.4469830989837646, "step": 33241 }, { "epoch": 2.620575482853764, "grad_norm": 0.5184380412101746, "learning_rate": 0.0005, "loss": 1.4410946369171143, "step": 33242 }, { "epoch": 2.6206543161214033, "grad_norm": 0.6439075469970703, "learning_rate": 0.0005, "loss": 1.575116753578186, "step": 33243 }, { "epoch": 2.620733149389042, "grad_norm": 0.4846891164779663, "learning_rate": 0.0005, "loss": 1.1309213638305664, "step": 33244 }, { "epoch": 2.620811982656681, "grad_norm": 0.458767294883728, "learning_rate": 0.0005, "loss": 1.3116950988769531, "step": 33245 }, { "epoch": 2.6208908159243203, "grad_norm": 0.525752604007721, "learning_rate": 0.0005, "loss": 1.2743752002716064, "step": 33246 }, { "epoch": 2.620969649191959, "grad_norm": 0.41010206937789917, "learning_rate": 0.0005, "loss": 1.224822998046875, "step": 33247 }, { "epoch": 2.621048482459598, "grad_norm": 0.5217915177345276, "learning_rate": 0.0005, "loss": 1.2438448667526245, "step": 33248 }, { "epoch": 2.621127315727237, "grad_norm": 0.474724680185318, "learning_rate": 0.0005, "loss": 1.4910439252853394, "step": 33249 }, { "epoch": 2.621206148994876, "grad_norm": 0.5057793259620667, "learning_rate": 0.0005, "loss": 1.158026933670044, "step": 33250 }, { "epoch": 2.621284982262515, "grad_norm": 0.47329190373420715, "learning_rate": 0.0005, "loss": 1.201080560684204, "step": 33251 }, { "epoch": 2.6213638155301537, "grad_norm": 0.4113278090953827, "learning_rate": 0.0005, "loss": 1.1655335426330566, "step": 33252 }, { "epoch": 2.6214426487977924, "grad_norm": 0.529943585395813, "learning_rate": 0.0005, "loss": 1.3215479850769043, "step": 33253 }, { "epoch": 2.6215214820654316, "grad_norm": 0.4522135257720947, "learning_rate": 0.0005, "loss": 1.120420217514038, "step": 33254 }, { "epoch": 2.6216003153330707, "grad_norm": 0.4959421753883362, "learning_rate": 0.0005, "loss": 1.3654803037643433, "step": 33255 }, { "epoch": 2.6216791486007094, "grad_norm": 0.45798027515411377, "learning_rate": 0.0005, "loss": 1.3323734998703003, "step": 33256 }, { "epoch": 2.6217579818683485, "grad_norm": 0.4921828508377075, "learning_rate": 0.0005, "loss": 1.318833827972412, "step": 33257 }, { "epoch": 2.621836815135987, "grad_norm": 0.6126610040664673, "learning_rate": 0.0005, "loss": 1.4034250974655151, "step": 33258 }, { "epoch": 2.6219156484036263, "grad_norm": 0.5059401988983154, "learning_rate": 0.0005, "loss": 1.4639652967453003, "step": 33259 }, { "epoch": 2.6219944816712655, "grad_norm": 0.3974260985851288, "learning_rate": 0.0005, "loss": 1.1975419521331787, "step": 33260 }, { "epoch": 2.622073314938904, "grad_norm": 0.6899964809417725, "learning_rate": 0.0005, "loss": 1.5367202758789062, "step": 33261 }, { "epoch": 2.6221521482065433, "grad_norm": 0.5416306853294373, "learning_rate": 0.0005, "loss": 1.4072282314300537, "step": 33262 }, { "epoch": 2.622230981474182, "grad_norm": 0.6252557039260864, "learning_rate": 0.0005, "loss": 1.6452951431274414, "step": 33263 }, { "epoch": 2.622309814741821, "grad_norm": 0.5892677307128906, "learning_rate": 0.0005, "loss": 1.3639053106307983, "step": 33264 }, { "epoch": 2.6223886480094603, "grad_norm": 0.502219557762146, "learning_rate": 0.0005, "loss": 1.2931629419326782, "step": 33265 }, { "epoch": 2.622467481277099, "grad_norm": 0.5572089552879333, "learning_rate": 0.0005, "loss": 1.3232104778289795, "step": 33266 }, { "epoch": 2.6225463145447376, "grad_norm": 0.5276060104370117, "learning_rate": 0.0005, "loss": 1.3550156354904175, "step": 33267 }, { "epoch": 2.6226251478123768, "grad_norm": 0.4730774760246277, "learning_rate": 0.0005, "loss": 1.3454842567443848, "step": 33268 }, { "epoch": 2.622703981080016, "grad_norm": 0.4623801112174988, "learning_rate": 0.0005, "loss": 1.2123050689697266, "step": 33269 }, { "epoch": 2.6227828143476546, "grad_norm": 0.5357342958450317, "learning_rate": 0.0005, "loss": 1.29185152053833, "step": 33270 }, { "epoch": 2.6228616476152937, "grad_norm": 0.6479941606521606, "learning_rate": 0.0005, "loss": 1.6653320789337158, "step": 33271 }, { "epoch": 2.6229404808829324, "grad_norm": 0.5649018883705139, "learning_rate": 0.0005, "loss": 1.6002066135406494, "step": 33272 }, { "epoch": 2.6230193141505715, "grad_norm": 0.5414933562278748, "learning_rate": 0.0005, "loss": 1.3447868824005127, "step": 33273 }, { "epoch": 2.6230981474182107, "grad_norm": 0.5373132824897766, "learning_rate": 0.0005, "loss": 1.5093951225280762, "step": 33274 }, { "epoch": 2.6231769806858494, "grad_norm": 0.5340108275413513, "learning_rate": 0.0005, "loss": 1.372609257698059, "step": 33275 }, { "epoch": 2.6232558139534885, "grad_norm": 0.5207670331001282, "learning_rate": 0.0005, "loss": 1.3813519477844238, "step": 33276 }, { "epoch": 2.623334647221127, "grad_norm": 0.5432732701301575, "learning_rate": 0.0005, "loss": 1.5401973724365234, "step": 33277 }, { "epoch": 2.6234134804887663, "grad_norm": 0.48110514879226685, "learning_rate": 0.0005, "loss": 1.2771589756011963, "step": 33278 }, { "epoch": 2.6234923137564055, "grad_norm": 0.5404742360115051, "learning_rate": 0.0005, "loss": 1.4072904586791992, "step": 33279 }, { "epoch": 2.623571147024044, "grad_norm": 0.583537757396698, "learning_rate": 0.0005, "loss": 1.3685150146484375, "step": 33280 }, { "epoch": 2.623649980291683, "grad_norm": 0.5888038873672485, "learning_rate": 0.0005, "loss": 1.2622302770614624, "step": 33281 }, { "epoch": 2.623728813559322, "grad_norm": 0.5121458172798157, "learning_rate": 0.0005, "loss": 1.4269905090332031, "step": 33282 }, { "epoch": 2.623807646826961, "grad_norm": 0.49196910858154297, "learning_rate": 0.0005, "loss": 1.354042887687683, "step": 33283 }, { "epoch": 2.6238864800946, "grad_norm": 0.4731396734714508, "learning_rate": 0.0005, "loss": 1.2567462921142578, "step": 33284 }, { "epoch": 2.623965313362239, "grad_norm": 0.5771927833557129, "learning_rate": 0.0005, "loss": 1.3898262977600098, "step": 33285 }, { "epoch": 2.6240441466298776, "grad_norm": 0.4784039556980133, "learning_rate": 0.0005, "loss": 1.3626149892807007, "step": 33286 }, { "epoch": 2.6241229798975167, "grad_norm": 0.4062695801258087, "learning_rate": 0.0005, "loss": 1.248070240020752, "step": 33287 }, { "epoch": 2.624201813165156, "grad_norm": 0.4136970639228821, "learning_rate": 0.0005, "loss": 1.278850793838501, "step": 33288 }, { "epoch": 2.6242806464327946, "grad_norm": 0.5338807702064514, "learning_rate": 0.0005, "loss": 1.2910261154174805, "step": 33289 }, { "epoch": 2.6243594797004337, "grad_norm": 0.3788924515247345, "learning_rate": 0.0005, "loss": 1.076124668121338, "step": 33290 }, { "epoch": 2.6244383129680724, "grad_norm": 0.5314033627510071, "learning_rate": 0.0005, "loss": 1.403780460357666, "step": 33291 }, { "epoch": 2.6245171462357115, "grad_norm": 0.5829641819000244, "learning_rate": 0.0005, "loss": 1.3809986114501953, "step": 33292 }, { "epoch": 2.6245959795033507, "grad_norm": 0.5025653839111328, "learning_rate": 0.0005, "loss": 1.3122397661209106, "step": 33293 }, { "epoch": 2.6246748127709894, "grad_norm": 0.46294039487838745, "learning_rate": 0.0005, "loss": 1.3724561929702759, "step": 33294 }, { "epoch": 2.624753646038628, "grad_norm": 0.6536985635757446, "learning_rate": 0.0005, "loss": 1.3105977773666382, "step": 33295 }, { "epoch": 2.624832479306267, "grad_norm": 0.5197908282279968, "learning_rate": 0.0005, "loss": 1.3370389938354492, "step": 33296 }, { "epoch": 2.6249113125739063, "grad_norm": 0.44701462984085083, "learning_rate": 0.0005, "loss": 1.351096749305725, "step": 33297 }, { "epoch": 2.624990145841545, "grad_norm": 0.5053117871284485, "learning_rate": 0.0005, "loss": 1.3705960512161255, "step": 33298 }, { "epoch": 2.625068979109184, "grad_norm": 0.4896952509880066, "learning_rate": 0.0005, "loss": 1.2126452922821045, "step": 33299 }, { "epoch": 2.625147812376823, "grad_norm": 0.45307445526123047, "learning_rate": 0.0005, "loss": 1.2259386777877808, "step": 33300 }, { "epoch": 2.625226645644462, "grad_norm": 0.49399736523628235, "learning_rate": 0.0005, "loss": 1.2510912418365479, "step": 33301 }, { "epoch": 2.625305478912101, "grad_norm": 0.5730206370353699, "learning_rate": 0.0005, "loss": 1.3672322034835815, "step": 33302 }, { "epoch": 2.6253843121797398, "grad_norm": 0.48706352710723877, "learning_rate": 0.0005, "loss": 1.2095352411270142, "step": 33303 }, { "epoch": 2.625463145447379, "grad_norm": 0.4839398264884949, "learning_rate": 0.0005, "loss": 1.224400281906128, "step": 33304 }, { "epoch": 2.6255419787150176, "grad_norm": 0.42054054141044617, "learning_rate": 0.0005, "loss": 1.0848873853683472, "step": 33305 }, { "epoch": 2.6256208119826567, "grad_norm": 0.6269382238388062, "learning_rate": 0.0005, "loss": 1.429046392440796, "step": 33306 }, { "epoch": 2.625699645250296, "grad_norm": 0.48276472091674805, "learning_rate": 0.0005, "loss": 1.2791473865509033, "step": 33307 }, { "epoch": 2.6257784785179346, "grad_norm": 0.4758417308330536, "learning_rate": 0.0005, "loss": 1.5355817079544067, "step": 33308 }, { "epoch": 2.6258573117855732, "grad_norm": 0.6632909774780273, "learning_rate": 0.0005, "loss": 1.4320762157440186, "step": 33309 }, { "epoch": 2.6259361450532124, "grad_norm": 0.4329789876937866, "learning_rate": 0.0005, "loss": 1.358527660369873, "step": 33310 }, { "epoch": 2.6260149783208515, "grad_norm": 0.5125138759613037, "learning_rate": 0.0005, "loss": 1.503157615661621, "step": 33311 }, { "epoch": 2.62609381158849, "grad_norm": 0.5676660537719727, "learning_rate": 0.0005, "loss": 1.3875526189804077, "step": 33312 }, { "epoch": 2.6261726448561293, "grad_norm": 0.5720034241676331, "learning_rate": 0.0005, "loss": 1.2626441717147827, "step": 33313 }, { "epoch": 2.626251478123768, "grad_norm": 0.4878682494163513, "learning_rate": 0.0005, "loss": 1.4376423358917236, "step": 33314 }, { "epoch": 2.626330311391407, "grad_norm": 0.514887809753418, "learning_rate": 0.0005, "loss": 1.4616782665252686, "step": 33315 }, { "epoch": 2.6264091446590463, "grad_norm": 0.4594153165817261, "learning_rate": 0.0005, "loss": 1.2951695919036865, "step": 33316 }, { "epoch": 2.626487977926685, "grad_norm": 0.47709885239601135, "learning_rate": 0.0005, "loss": 1.3009856939315796, "step": 33317 }, { "epoch": 2.626566811194324, "grad_norm": 0.6041240692138672, "learning_rate": 0.0005, "loss": 1.4425184726715088, "step": 33318 }, { "epoch": 2.626645644461963, "grad_norm": 0.5819308757781982, "learning_rate": 0.0005, "loss": 1.5467352867126465, "step": 33319 }, { "epoch": 2.626724477729602, "grad_norm": 0.4573229253292084, "learning_rate": 0.0005, "loss": 1.322519063949585, "step": 33320 }, { "epoch": 2.626803310997241, "grad_norm": 0.48402121663093567, "learning_rate": 0.0005, "loss": 1.6390128135681152, "step": 33321 }, { "epoch": 2.6268821442648798, "grad_norm": 0.530808687210083, "learning_rate": 0.0005, "loss": 1.2013936042785645, "step": 33322 }, { "epoch": 2.6269609775325184, "grad_norm": 0.4378470182418823, "learning_rate": 0.0005, "loss": 1.228128433227539, "step": 33323 }, { "epoch": 2.6270398108001576, "grad_norm": 0.6039791107177734, "learning_rate": 0.0005, "loss": 1.3036530017852783, "step": 33324 }, { "epoch": 2.6271186440677967, "grad_norm": 0.5684245824813843, "learning_rate": 0.0005, "loss": 1.4417924880981445, "step": 33325 }, { "epoch": 2.6271974773354354, "grad_norm": 0.49801185727119446, "learning_rate": 0.0005, "loss": 1.4467846155166626, "step": 33326 }, { "epoch": 2.6272763106030745, "grad_norm": 0.7216101884841919, "learning_rate": 0.0005, "loss": 1.426048994064331, "step": 33327 }, { "epoch": 2.6273551438707132, "grad_norm": 0.5867412686347961, "learning_rate": 0.0005, "loss": 1.4797724485397339, "step": 33328 }, { "epoch": 2.6274339771383524, "grad_norm": 0.5126895904541016, "learning_rate": 0.0005, "loss": 1.2668530941009521, "step": 33329 }, { "epoch": 2.6275128104059915, "grad_norm": 0.4850185811519623, "learning_rate": 0.0005, "loss": 1.3378121852874756, "step": 33330 }, { "epoch": 2.62759164367363, "grad_norm": 0.5122259855270386, "learning_rate": 0.0005, "loss": 1.3979415893554688, "step": 33331 }, { "epoch": 2.6276704769412693, "grad_norm": 0.6023169159889221, "learning_rate": 0.0005, "loss": 1.5380492210388184, "step": 33332 }, { "epoch": 2.627749310208908, "grad_norm": 0.6160013675689697, "learning_rate": 0.0005, "loss": 1.293165922164917, "step": 33333 }, { "epoch": 2.627828143476547, "grad_norm": 0.43328776955604553, "learning_rate": 0.0005, "loss": 1.1870371103286743, "step": 33334 }, { "epoch": 2.6279069767441863, "grad_norm": 0.45557162165641785, "learning_rate": 0.0005, "loss": 1.1859049797058105, "step": 33335 }, { "epoch": 2.627985810011825, "grad_norm": 0.4343671202659607, "learning_rate": 0.0005, "loss": 1.5133841037750244, "step": 33336 }, { "epoch": 2.628064643279464, "grad_norm": 0.369002103805542, "learning_rate": 0.0005, "loss": 1.2147479057312012, "step": 33337 }, { "epoch": 2.628143476547103, "grad_norm": 0.47106873989105225, "learning_rate": 0.0005, "loss": 1.3309911489486694, "step": 33338 }, { "epoch": 2.628222309814742, "grad_norm": 0.5870872139930725, "learning_rate": 0.0005, "loss": 1.3446451425552368, "step": 33339 }, { "epoch": 2.6283011430823806, "grad_norm": 0.5552120804786682, "learning_rate": 0.0005, "loss": 1.5569597482681274, "step": 33340 }, { "epoch": 2.6283799763500197, "grad_norm": 0.5315641164779663, "learning_rate": 0.0005, "loss": 1.4285047054290771, "step": 33341 }, { "epoch": 2.6284588096176584, "grad_norm": 0.5497264862060547, "learning_rate": 0.0005, "loss": 1.2961621284484863, "step": 33342 }, { "epoch": 2.6285376428852976, "grad_norm": 0.5289729833602905, "learning_rate": 0.0005, "loss": 1.4821956157684326, "step": 33343 }, { "epoch": 2.6286164761529367, "grad_norm": 0.5337557196617126, "learning_rate": 0.0005, "loss": 1.3600730895996094, "step": 33344 }, { "epoch": 2.6286953094205754, "grad_norm": 0.49541664123535156, "learning_rate": 0.0005, "loss": 1.1908636093139648, "step": 33345 }, { "epoch": 2.6287741426882145, "grad_norm": 0.4667038321495056, "learning_rate": 0.0005, "loss": 1.2590808868408203, "step": 33346 }, { "epoch": 2.628852975955853, "grad_norm": 0.4382026493549347, "learning_rate": 0.0005, "loss": 1.1125338077545166, "step": 33347 }, { "epoch": 2.6289318092234923, "grad_norm": 0.4683046340942383, "learning_rate": 0.0005, "loss": 1.388979196548462, "step": 33348 }, { "epoch": 2.6290106424911315, "grad_norm": 0.5159946084022522, "learning_rate": 0.0005, "loss": 1.6165730953216553, "step": 33349 }, { "epoch": 2.62908947575877, "grad_norm": 0.5419397354125977, "learning_rate": 0.0005, "loss": 1.363633632659912, "step": 33350 }, { "epoch": 2.6291683090264093, "grad_norm": 0.5915570259094238, "learning_rate": 0.0005, "loss": 1.3422123193740845, "step": 33351 }, { "epoch": 2.629247142294048, "grad_norm": 0.409839004278183, "learning_rate": 0.0005, "loss": 1.1766178607940674, "step": 33352 }, { "epoch": 2.629325975561687, "grad_norm": 0.583651065826416, "learning_rate": 0.0005, "loss": 1.4731724262237549, "step": 33353 }, { "epoch": 2.629404808829326, "grad_norm": 0.5040656328201294, "learning_rate": 0.0005, "loss": 1.1801098585128784, "step": 33354 }, { "epoch": 2.629483642096965, "grad_norm": 0.4912893772125244, "learning_rate": 0.0005, "loss": 1.4060070514678955, "step": 33355 }, { "epoch": 2.6295624753646036, "grad_norm": 0.5285663604736328, "learning_rate": 0.0005, "loss": 1.529061198234558, "step": 33356 }, { "epoch": 2.6296413086322428, "grad_norm": 0.5557522773742676, "learning_rate": 0.0005, "loss": 1.3673481941223145, "step": 33357 }, { "epoch": 2.629720141899882, "grad_norm": 0.44694557785987854, "learning_rate": 0.0005, "loss": 1.2930997610092163, "step": 33358 }, { "epoch": 2.6297989751675206, "grad_norm": 0.4163639545440674, "learning_rate": 0.0005, "loss": 1.2911577224731445, "step": 33359 }, { "epoch": 2.6298778084351597, "grad_norm": 0.5920175313949585, "learning_rate": 0.0005, "loss": 1.3861169815063477, "step": 33360 }, { "epoch": 2.6299566417027984, "grad_norm": 0.4561840891838074, "learning_rate": 0.0005, "loss": 1.2770897150039673, "step": 33361 }, { "epoch": 2.6300354749704375, "grad_norm": 0.44499051570892334, "learning_rate": 0.0005, "loss": 1.2614246606826782, "step": 33362 }, { "epoch": 2.6301143082380767, "grad_norm": 0.43749144673347473, "learning_rate": 0.0005, "loss": 1.3145439624786377, "step": 33363 }, { "epoch": 2.6301931415057154, "grad_norm": 0.747196614742279, "learning_rate": 0.0005, "loss": 1.3296079635620117, "step": 33364 }, { "epoch": 2.6302719747733545, "grad_norm": 0.45512571930885315, "learning_rate": 0.0005, "loss": 1.169703483581543, "step": 33365 }, { "epoch": 2.630350808040993, "grad_norm": 0.5749698281288147, "learning_rate": 0.0005, "loss": 1.4407100677490234, "step": 33366 }, { "epoch": 2.6304296413086323, "grad_norm": 0.4524020552635193, "learning_rate": 0.0005, "loss": 1.3116364479064941, "step": 33367 }, { "epoch": 2.6305084745762715, "grad_norm": 0.5886411666870117, "learning_rate": 0.0005, "loss": 1.2841733694076538, "step": 33368 }, { "epoch": 2.63058730784391, "grad_norm": 0.5797860622406006, "learning_rate": 0.0005, "loss": 1.3869974613189697, "step": 33369 }, { "epoch": 2.630666141111549, "grad_norm": 0.5344419479370117, "learning_rate": 0.0005, "loss": 1.4397624731063843, "step": 33370 }, { "epoch": 2.630744974379188, "grad_norm": 0.48437008261680603, "learning_rate": 0.0005, "loss": 1.233588457107544, "step": 33371 }, { "epoch": 2.630823807646827, "grad_norm": 0.5720645189285278, "learning_rate": 0.0005, "loss": 1.395477533340454, "step": 33372 }, { "epoch": 2.630902640914466, "grad_norm": 0.4968642294406891, "learning_rate": 0.0005, "loss": 1.3452725410461426, "step": 33373 }, { "epoch": 2.630981474182105, "grad_norm": 0.4803387224674225, "learning_rate": 0.0005, "loss": 1.3648858070373535, "step": 33374 }, { "epoch": 2.6310603074497436, "grad_norm": 0.565524160861969, "learning_rate": 0.0005, "loss": 1.4275476932525635, "step": 33375 }, { "epoch": 2.6311391407173828, "grad_norm": 0.4914422929286957, "learning_rate": 0.0005, "loss": 1.4727482795715332, "step": 33376 }, { "epoch": 2.631217973985022, "grad_norm": 0.5052222609519958, "learning_rate": 0.0005, "loss": 1.3007146120071411, "step": 33377 }, { "epoch": 2.6312968072526606, "grad_norm": 0.49092671275138855, "learning_rate": 0.0005, "loss": 1.257106065750122, "step": 33378 }, { "epoch": 2.6313756405202997, "grad_norm": 0.5009632706642151, "learning_rate": 0.0005, "loss": 1.2354578971862793, "step": 33379 }, { "epoch": 2.6314544737879384, "grad_norm": 0.533110499382019, "learning_rate": 0.0005, "loss": 1.4115447998046875, "step": 33380 }, { "epoch": 2.6315333070555775, "grad_norm": 0.49011191725730896, "learning_rate": 0.0005, "loss": 1.2226827144622803, "step": 33381 }, { "epoch": 2.6316121403232167, "grad_norm": 0.545779287815094, "learning_rate": 0.0005, "loss": 1.5579909086227417, "step": 33382 }, { "epoch": 2.6316909735908554, "grad_norm": 0.4724397361278534, "learning_rate": 0.0005, "loss": 1.3125450611114502, "step": 33383 }, { "epoch": 2.631769806858494, "grad_norm": 0.4773079454898834, "learning_rate": 0.0005, "loss": 1.4330761432647705, "step": 33384 }, { "epoch": 2.631848640126133, "grad_norm": 0.5146748423576355, "learning_rate": 0.0005, "loss": 1.3351616859436035, "step": 33385 }, { "epoch": 2.6319274733937723, "grad_norm": 0.4405498802661896, "learning_rate": 0.0005, "loss": 1.357243537902832, "step": 33386 }, { "epoch": 2.632006306661411, "grad_norm": 0.44625917077064514, "learning_rate": 0.0005, "loss": 1.298612117767334, "step": 33387 }, { "epoch": 2.63208513992905, "grad_norm": 0.3751109540462494, "learning_rate": 0.0005, "loss": 1.1172332763671875, "step": 33388 }, { "epoch": 2.632163973196689, "grad_norm": 0.7051821947097778, "learning_rate": 0.0005, "loss": 1.2291918992996216, "step": 33389 }, { "epoch": 2.632242806464328, "grad_norm": 0.4505590498447418, "learning_rate": 0.0005, "loss": 1.1805148124694824, "step": 33390 }, { "epoch": 2.632321639731967, "grad_norm": 0.5179821848869324, "learning_rate": 0.0005, "loss": 1.3850809335708618, "step": 33391 }, { "epoch": 2.632400472999606, "grad_norm": 0.4909958243370056, "learning_rate": 0.0005, "loss": 1.2571461200714111, "step": 33392 }, { "epoch": 2.632479306267245, "grad_norm": 0.5209231972694397, "learning_rate": 0.0005, "loss": 1.1282899379730225, "step": 33393 }, { "epoch": 2.6325581395348836, "grad_norm": 0.6667652726173401, "learning_rate": 0.0005, "loss": 1.44766366481781, "step": 33394 }, { "epoch": 2.6326369728025227, "grad_norm": 0.5859026312828064, "learning_rate": 0.0005, "loss": 1.6197867393493652, "step": 33395 }, { "epoch": 2.632715806070162, "grad_norm": 0.5191139578819275, "learning_rate": 0.0005, "loss": 1.296653151512146, "step": 33396 }, { "epoch": 2.6327946393378006, "grad_norm": 0.4733642637729645, "learning_rate": 0.0005, "loss": 1.252115249633789, "step": 33397 }, { "epoch": 2.6328734726054392, "grad_norm": 0.6494421362876892, "learning_rate": 0.0005, "loss": 1.6361280679702759, "step": 33398 }, { "epoch": 2.6329523058730784, "grad_norm": 0.7330641746520996, "learning_rate": 0.0005, "loss": 1.3544038534164429, "step": 33399 }, { "epoch": 2.6330311391407175, "grad_norm": 0.5183196067810059, "learning_rate": 0.0005, "loss": 1.2576881647109985, "step": 33400 }, { "epoch": 2.633109972408356, "grad_norm": 0.4992934763431549, "learning_rate": 0.0005, "loss": 1.181452989578247, "step": 33401 }, { "epoch": 2.6331888056759953, "grad_norm": 0.5373656749725342, "learning_rate": 0.0005, "loss": 1.5296258926391602, "step": 33402 }, { "epoch": 2.633267638943634, "grad_norm": 0.48576825857162476, "learning_rate": 0.0005, "loss": 1.4348480701446533, "step": 33403 }, { "epoch": 2.633346472211273, "grad_norm": 0.5006096959114075, "learning_rate": 0.0005, "loss": 1.3584022521972656, "step": 33404 }, { "epoch": 2.6334253054789123, "grad_norm": 0.4642260670661926, "learning_rate": 0.0005, "loss": 1.1362334489822388, "step": 33405 }, { "epoch": 2.633504138746551, "grad_norm": 0.5943835973739624, "learning_rate": 0.0005, "loss": 1.292524814605713, "step": 33406 }, { "epoch": 2.63358297201419, "grad_norm": 0.5167719721794128, "learning_rate": 0.0005, "loss": 1.09218168258667, "step": 33407 }, { "epoch": 2.633661805281829, "grad_norm": 0.44646570086479187, "learning_rate": 0.0005, "loss": 1.3840570449829102, "step": 33408 }, { "epoch": 2.633740638549468, "grad_norm": 0.4561626613140106, "learning_rate": 0.0005, "loss": 1.378770351409912, "step": 33409 }, { "epoch": 2.633819471817107, "grad_norm": 0.48545023798942566, "learning_rate": 0.0005, "loss": 1.284363865852356, "step": 33410 }, { "epoch": 2.6338983050847458, "grad_norm": 0.52882981300354, "learning_rate": 0.0005, "loss": 1.294824242591858, "step": 33411 }, { "epoch": 2.6339771383523845, "grad_norm": 0.5107122659683228, "learning_rate": 0.0005, "loss": 1.3814749717712402, "step": 33412 }, { "epoch": 2.6340559716200236, "grad_norm": 0.5747369527816772, "learning_rate": 0.0005, "loss": 1.3444397449493408, "step": 33413 }, { "epoch": 2.6341348048876627, "grad_norm": 0.4613291025161743, "learning_rate": 0.0005, "loss": 1.2496328353881836, "step": 33414 }, { "epoch": 2.6342136381553014, "grad_norm": 0.42934471368789673, "learning_rate": 0.0005, "loss": 1.156360149383545, "step": 33415 }, { "epoch": 2.6342924714229405, "grad_norm": 0.519546389579773, "learning_rate": 0.0005, "loss": 1.416607141494751, "step": 33416 }, { "epoch": 2.6343713046905792, "grad_norm": 0.41838937997817993, "learning_rate": 0.0005, "loss": 1.2242400646209717, "step": 33417 }, { "epoch": 2.6344501379582184, "grad_norm": 0.47964903712272644, "learning_rate": 0.0005, "loss": 1.1917788982391357, "step": 33418 }, { "epoch": 2.6345289712258575, "grad_norm": 0.7446689009666443, "learning_rate": 0.0005, "loss": 1.143420934677124, "step": 33419 }, { "epoch": 2.634607804493496, "grad_norm": 0.465981662273407, "learning_rate": 0.0005, "loss": 1.3623785972595215, "step": 33420 }, { "epoch": 2.6346866377611353, "grad_norm": 0.5789652466773987, "learning_rate": 0.0005, "loss": 1.4453046321868896, "step": 33421 }, { "epoch": 2.634765471028774, "grad_norm": 0.4745368957519531, "learning_rate": 0.0005, "loss": 1.3751393556594849, "step": 33422 }, { "epoch": 2.634844304296413, "grad_norm": 0.40311694145202637, "learning_rate": 0.0005, "loss": 1.15338933467865, "step": 33423 }, { "epoch": 2.6349231375640523, "grad_norm": 0.5440452098846436, "learning_rate": 0.0005, "loss": 1.2169115543365479, "step": 33424 }, { "epoch": 2.635001970831691, "grad_norm": 0.5340166687965393, "learning_rate": 0.0005, "loss": 1.3415488004684448, "step": 33425 }, { "epoch": 2.6350808040993297, "grad_norm": 0.4770698845386505, "learning_rate": 0.0005, "loss": 1.3891161680221558, "step": 33426 }, { "epoch": 2.635159637366969, "grad_norm": 0.41477248072624207, "learning_rate": 0.0005, "loss": 1.1983323097229004, "step": 33427 }, { "epoch": 2.635238470634608, "grad_norm": 0.5882228016853333, "learning_rate": 0.0005, "loss": 1.2827812433242798, "step": 33428 }, { "epoch": 2.6353173039022466, "grad_norm": 0.5752454400062561, "learning_rate": 0.0005, "loss": 1.4560805559158325, "step": 33429 }, { "epoch": 2.6353961371698857, "grad_norm": 0.4943598210811615, "learning_rate": 0.0005, "loss": 1.5207808017730713, "step": 33430 }, { "epoch": 2.6354749704375244, "grad_norm": 0.455020010471344, "learning_rate": 0.0005, "loss": 1.1699126958847046, "step": 33431 }, { "epoch": 2.6355538037051636, "grad_norm": 0.5123836398124695, "learning_rate": 0.0005, "loss": 1.1320443153381348, "step": 33432 }, { "epoch": 2.6356326369728027, "grad_norm": 0.5001936554908752, "learning_rate": 0.0005, "loss": 1.2846035957336426, "step": 33433 }, { "epoch": 2.6357114702404414, "grad_norm": 0.4399558901786804, "learning_rate": 0.0005, "loss": 1.3653333187103271, "step": 33434 }, { "epoch": 2.6357903035080805, "grad_norm": 0.5007594227790833, "learning_rate": 0.0005, "loss": 1.198104739189148, "step": 33435 }, { "epoch": 2.635869136775719, "grad_norm": 0.46748366951942444, "learning_rate": 0.0005, "loss": 1.2576937675476074, "step": 33436 }, { "epoch": 2.6359479700433583, "grad_norm": 0.5341650247573853, "learning_rate": 0.0005, "loss": 1.5934770107269287, "step": 33437 }, { "epoch": 2.6360268033109975, "grad_norm": 0.4794515073299408, "learning_rate": 0.0005, "loss": 1.3256645202636719, "step": 33438 }, { "epoch": 2.636105636578636, "grad_norm": 0.4309625029563904, "learning_rate": 0.0005, "loss": 1.3948001861572266, "step": 33439 }, { "epoch": 2.636184469846275, "grad_norm": 0.5648868680000305, "learning_rate": 0.0005, "loss": 1.2549080848693848, "step": 33440 }, { "epoch": 2.636263303113914, "grad_norm": 0.494889497756958, "learning_rate": 0.0005, "loss": 1.3832676410675049, "step": 33441 }, { "epoch": 2.636342136381553, "grad_norm": 0.5780321955680847, "learning_rate": 0.0005, "loss": 1.4118694067001343, "step": 33442 }, { "epoch": 2.636420969649192, "grad_norm": 0.48267167806625366, "learning_rate": 0.0005, "loss": 1.371254801750183, "step": 33443 }, { "epoch": 2.636499802916831, "grad_norm": 0.5545873641967773, "learning_rate": 0.0005, "loss": 1.4385337829589844, "step": 33444 }, { "epoch": 2.6365786361844696, "grad_norm": 0.4412277936935425, "learning_rate": 0.0005, "loss": 1.345057725906372, "step": 33445 }, { "epoch": 2.6366574694521088, "grad_norm": 0.7843329906463623, "learning_rate": 0.0005, "loss": 1.382253885269165, "step": 33446 }, { "epoch": 2.636736302719748, "grad_norm": 0.6417945623397827, "learning_rate": 0.0005, "loss": 1.376409649848938, "step": 33447 }, { "epoch": 2.6368151359873866, "grad_norm": 0.5374383926391602, "learning_rate": 0.0005, "loss": 1.487494945526123, "step": 33448 }, { "epoch": 2.6368939692550257, "grad_norm": 0.48447713255882263, "learning_rate": 0.0005, "loss": 1.2889779806137085, "step": 33449 }, { "epoch": 2.6369728025226644, "grad_norm": 0.4707080125808716, "learning_rate": 0.0005, "loss": 1.1957919597625732, "step": 33450 }, { "epoch": 2.6370516357903035, "grad_norm": 0.49077385663986206, "learning_rate": 0.0005, "loss": 1.425029993057251, "step": 33451 }, { "epoch": 2.6371304690579427, "grad_norm": 0.4306030571460724, "learning_rate": 0.0005, "loss": 1.2508114576339722, "step": 33452 }, { "epoch": 2.6372093023255814, "grad_norm": 0.5432855486869812, "learning_rate": 0.0005, "loss": 1.2855956554412842, "step": 33453 }, { "epoch": 2.63728813559322, "grad_norm": 0.5599205493927002, "learning_rate": 0.0005, "loss": 1.538411259651184, "step": 33454 }, { "epoch": 2.637366968860859, "grad_norm": 0.4741128087043762, "learning_rate": 0.0005, "loss": 1.3641371726989746, "step": 33455 }, { "epoch": 2.6374458021284983, "grad_norm": 0.5168046355247498, "learning_rate": 0.0005, "loss": 1.3322727680206299, "step": 33456 }, { "epoch": 2.637524635396137, "grad_norm": 1.1683238744735718, "learning_rate": 0.0005, "loss": 1.9621787071228027, "step": 33457 }, { "epoch": 2.637603468663776, "grad_norm": 0.5716238617897034, "learning_rate": 0.0005, "loss": 1.322698712348938, "step": 33458 }, { "epoch": 2.637682301931415, "grad_norm": 0.42120060324668884, "learning_rate": 0.0005, "loss": 1.138066053390503, "step": 33459 }, { "epoch": 2.637761135199054, "grad_norm": 0.568166971206665, "learning_rate": 0.0005, "loss": 1.4969062805175781, "step": 33460 }, { "epoch": 2.637839968466693, "grad_norm": 0.5784436464309692, "learning_rate": 0.0005, "loss": 1.4406917095184326, "step": 33461 }, { "epoch": 2.637918801734332, "grad_norm": 0.47279128432273865, "learning_rate": 0.0005, "loss": 1.2402805089950562, "step": 33462 }, { "epoch": 2.637997635001971, "grad_norm": 0.6430979371070862, "learning_rate": 0.0005, "loss": 1.5209406614303589, "step": 33463 }, { "epoch": 2.6380764682696096, "grad_norm": 0.5176985859870911, "learning_rate": 0.0005, "loss": 1.4133687019348145, "step": 33464 }, { "epoch": 2.6381553015372488, "grad_norm": 0.5006131529808044, "learning_rate": 0.0005, "loss": 1.3218519687652588, "step": 33465 }, { "epoch": 2.638234134804888, "grad_norm": 0.5116352438926697, "learning_rate": 0.0005, "loss": 1.2182669639587402, "step": 33466 }, { "epoch": 2.6383129680725266, "grad_norm": 0.6056544780731201, "learning_rate": 0.0005, "loss": 1.408677101135254, "step": 33467 }, { "epoch": 2.6383918013401657, "grad_norm": 0.5060456395149231, "learning_rate": 0.0005, "loss": 1.3646457195281982, "step": 33468 }, { "epoch": 2.6384706346078044, "grad_norm": 0.5484968423843384, "learning_rate": 0.0005, "loss": 1.4236247539520264, "step": 33469 }, { "epoch": 2.6385494678754435, "grad_norm": 0.5921052694320679, "learning_rate": 0.0005, "loss": 1.3333919048309326, "step": 33470 }, { "epoch": 2.6386283011430822, "grad_norm": 0.47155675292015076, "learning_rate": 0.0005, "loss": 1.2861833572387695, "step": 33471 }, { "epoch": 2.6387071344107214, "grad_norm": 0.5731092691421509, "learning_rate": 0.0005, "loss": 1.2731645107269287, "step": 33472 }, { "epoch": 2.63878596767836, "grad_norm": 0.5780636072158813, "learning_rate": 0.0005, "loss": 1.4765678644180298, "step": 33473 }, { "epoch": 2.638864800945999, "grad_norm": 0.5268680453300476, "learning_rate": 0.0005, "loss": 1.495502233505249, "step": 33474 }, { "epoch": 2.6389436342136383, "grad_norm": 0.5396550297737122, "learning_rate": 0.0005, "loss": 1.3203654289245605, "step": 33475 }, { "epoch": 2.639022467481277, "grad_norm": 0.5413773059844971, "learning_rate": 0.0005, "loss": 1.35007905960083, "step": 33476 }, { "epoch": 2.639101300748916, "grad_norm": 0.7391840815544128, "learning_rate": 0.0005, "loss": 1.4370911121368408, "step": 33477 }, { "epoch": 2.639180134016555, "grad_norm": 0.44885462522506714, "learning_rate": 0.0005, "loss": 1.2934311628341675, "step": 33478 }, { "epoch": 2.639258967284194, "grad_norm": 0.6150253415107727, "learning_rate": 0.0005, "loss": 1.2853801250457764, "step": 33479 }, { "epoch": 2.639337800551833, "grad_norm": 0.47125956416130066, "learning_rate": 0.0005, "loss": 1.2240793704986572, "step": 33480 }, { "epoch": 2.639416633819472, "grad_norm": 0.5051057934761047, "learning_rate": 0.0005, "loss": 1.3944668769836426, "step": 33481 }, { "epoch": 2.639495467087111, "grad_norm": 0.40606558322906494, "learning_rate": 0.0005, "loss": 1.2392094135284424, "step": 33482 }, { "epoch": 2.6395743003547496, "grad_norm": 0.6184936165809631, "learning_rate": 0.0005, "loss": 1.4825918674468994, "step": 33483 }, { "epoch": 2.6396531336223887, "grad_norm": 0.49773848056793213, "learning_rate": 0.0005, "loss": 1.4576196670532227, "step": 33484 }, { "epoch": 2.6397319668900274, "grad_norm": 1.1382259130477905, "learning_rate": 0.0005, "loss": 1.500133991241455, "step": 33485 }, { "epoch": 2.6398108001576666, "grad_norm": 0.48861491680145264, "learning_rate": 0.0005, "loss": 1.427031397819519, "step": 33486 }, { "epoch": 2.6398896334253052, "grad_norm": 0.4648994207382202, "learning_rate": 0.0005, "loss": 1.2922643423080444, "step": 33487 }, { "epoch": 2.6399684666929444, "grad_norm": 0.49846696853637695, "learning_rate": 0.0005, "loss": 1.3427889347076416, "step": 33488 }, { "epoch": 2.6400472999605835, "grad_norm": 1.253899335861206, "learning_rate": 0.0005, "loss": 1.4970672130584717, "step": 33489 }, { "epoch": 2.640126133228222, "grad_norm": 0.5523549318313599, "learning_rate": 0.0005, "loss": 1.5238502025604248, "step": 33490 }, { "epoch": 2.6402049664958613, "grad_norm": 0.4641844928264618, "learning_rate": 0.0005, "loss": 1.3638837337493896, "step": 33491 }, { "epoch": 2.6402837997635, "grad_norm": 0.5567678213119507, "learning_rate": 0.0005, "loss": 1.196589708328247, "step": 33492 }, { "epoch": 2.640362633031139, "grad_norm": 0.47751808166503906, "learning_rate": 0.0005, "loss": 1.3333783149719238, "step": 33493 }, { "epoch": 2.6404414662987783, "grad_norm": 0.4693811535835266, "learning_rate": 0.0005, "loss": 1.3449366092681885, "step": 33494 }, { "epoch": 2.640520299566417, "grad_norm": 0.5080652832984924, "learning_rate": 0.0005, "loss": 1.2904655933380127, "step": 33495 }, { "epoch": 2.640599132834056, "grad_norm": 0.47568607330322266, "learning_rate": 0.0005, "loss": 1.3134427070617676, "step": 33496 }, { "epoch": 2.640677966101695, "grad_norm": 0.4941195547580719, "learning_rate": 0.0005, "loss": 1.2158801555633545, "step": 33497 }, { "epoch": 2.640756799369334, "grad_norm": 0.529546320438385, "learning_rate": 0.0005, "loss": 1.3220059871673584, "step": 33498 }, { "epoch": 2.640835632636973, "grad_norm": 0.6094770431518555, "learning_rate": 0.0005, "loss": 1.4480350017547607, "step": 33499 }, { "epoch": 2.6409144659046118, "grad_norm": 0.5096790194511414, "learning_rate": 0.0005, "loss": 1.400038242340088, "step": 33500 }, { "epoch": 2.6409932991722505, "grad_norm": 0.5861477255821228, "learning_rate": 0.0005, "loss": 1.3842742443084717, "step": 33501 }, { "epoch": 2.6410721324398896, "grad_norm": 0.5606666207313538, "learning_rate": 0.0005, "loss": 1.40500009059906, "step": 33502 }, { "epoch": 2.6411509657075287, "grad_norm": 0.4684256911277771, "learning_rate": 0.0005, "loss": 1.4512931108474731, "step": 33503 }, { "epoch": 2.6412297989751674, "grad_norm": 0.44804486632347107, "learning_rate": 0.0005, "loss": 1.3858052492141724, "step": 33504 }, { "epoch": 2.6413086322428065, "grad_norm": 0.6051062345504761, "learning_rate": 0.0005, "loss": 1.3339437246322632, "step": 33505 }, { "epoch": 2.6413874655104452, "grad_norm": 0.554682195186615, "learning_rate": 0.0005, "loss": 1.3641453981399536, "step": 33506 }, { "epoch": 2.6414662987780844, "grad_norm": 0.5875897407531738, "learning_rate": 0.0005, "loss": 1.402282476425171, "step": 33507 }, { "epoch": 2.6415451320457235, "grad_norm": 0.6166945099830627, "learning_rate": 0.0005, "loss": 1.4024126529693604, "step": 33508 }, { "epoch": 2.641623965313362, "grad_norm": 0.5123350620269775, "learning_rate": 0.0005, "loss": 1.4182898998260498, "step": 33509 }, { "epoch": 2.6417027985810013, "grad_norm": 0.6019790768623352, "learning_rate": 0.0005, "loss": 1.4252097606658936, "step": 33510 }, { "epoch": 2.64178163184864, "grad_norm": 0.46488156914711, "learning_rate": 0.0005, "loss": 1.2050085067749023, "step": 33511 }, { "epoch": 2.641860465116279, "grad_norm": 0.5400447249412537, "learning_rate": 0.0005, "loss": 1.4205604791641235, "step": 33512 }, { "epoch": 2.6419392983839183, "grad_norm": 0.5303869247436523, "learning_rate": 0.0005, "loss": 1.250161051750183, "step": 33513 }, { "epoch": 2.642018131651557, "grad_norm": 0.4465261399745941, "learning_rate": 0.0005, "loss": 1.3087868690490723, "step": 33514 }, { "epoch": 2.6420969649191957, "grad_norm": 0.4456517696380615, "learning_rate": 0.0005, "loss": 1.1909489631652832, "step": 33515 }, { "epoch": 2.642175798186835, "grad_norm": 0.6763278245925903, "learning_rate": 0.0005, "loss": 1.6312958002090454, "step": 33516 }, { "epoch": 2.642254631454474, "grad_norm": 0.5129315853118896, "learning_rate": 0.0005, "loss": 1.3734538555145264, "step": 33517 }, { "epoch": 2.6423334647221126, "grad_norm": 0.5073295831680298, "learning_rate": 0.0005, "loss": 1.2866718769073486, "step": 33518 }, { "epoch": 2.6424122979897517, "grad_norm": 0.47627508640289307, "learning_rate": 0.0005, "loss": 1.3926639556884766, "step": 33519 }, { "epoch": 2.6424911312573904, "grad_norm": 0.4669264853000641, "learning_rate": 0.0005, "loss": 1.423891305923462, "step": 33520 }, { "epoch": 2.6425699645250296, "grad_norm": 0.5088369846343994, "learning_rate": 0.0005, "loss": 1.5365521907806396, "step": 33521 }, { "epoch": 2.6426487977926687, "grad_norm": 0.3799305856227875, "learning_rate": 0.0005, "loss": 1.0725445747375488, "step": 33522 }, { "epoch": 2.6427276310603074, "grad_norm": 0.4114963710308075, "learning_rate": 0.0005, "loss": 1.2797642946243286, "step": 33523 }, { "epoch": 2.6428064643279465, "grad_norm": 0.4585926830768585, "learning_rate": 0.0005, "loss": 1.3701081275939941, "step": 33524 }, { "epoch": 2.642885297595585, "grad_norm": 0.45728519558906555, "learning_rate": 0.0005, "loss": 1.3136311769485474, "step": 33525 }, { "epoch": 2.6429641308632243, "grad_norm": 0.4514461159706116, "learning_rate": 0.0005, "loss": 1.3151581287384033, "step": 33526 }, { "epoch": 2.6430429641308635, "grad_norm": 0.7148190140724182, "learning_rate": 0.0005, "loss": 1.532740592956543, "step": 33527 }, { "epoch": 2.643121797398502, "grad_norm": 0.5966140627861023, "learning_rate": 0.0005, "loss": 1.603083848953247, "step": 33528 }, { "epoch": 2.643200630666141, "grad_norm": 0.9803163409233093, "learning_rate": 0.0005, "loss": 1.2128145694732666, "step": 33529 }, { "epoch": 2.64327946393378, "grad_norm": 0.46966850757598877, "learning_rate": 0.0005, "loss": 1.2279362678527832, "step": 33530 }, { "epoch": 2.643358297201419, "grad_norm": 0.45650580525398254, "learning_rate": 0.0005, "loss": 1.174450159072876, "step": 33531 }, { "epoch": 2.643437130469058, "grad_norm": 0.50312739610672, "learning_rate": 0.0005, "loss": 1.3135275840759277, "step": 33532 }, { "epoch": 2.643515963736697, "grad_norm": 0.43920719623565674, "learning_rate": 0.0005, "loss": 1.2643628120422363, "step": 33533 }, { "epoch": 2.6435947970043356, "grad_norm": 0.4240169823169708, "learning_rate": 0.0005, "loss": 1.2493975162506104, "step": 33534 }, { "epoch": 2.6436736302719748, "grad_norm": 0.4541798233985901, "learning_rate": 0.0005, "loss": 1.4353017807006836, "step": 33535 }, { "epoch": 2.643752463539614, "grad_norm": 0.5867295265197754, "learning_rate": 0.0005, "loss": 1.4127192497253418, "step": 33536 }, { "epoch": 2.6438312968072526, "grad_norm": 0.5957010388374329, "learning_rate": 0.0005, "loss": 1.3118963241577148, "step": 33537 }, { "epoch": 2.6439101300748917, "grad_norm": 0.5045033097267151, "learning_rate": 0.0005, "loss": 1.2865591049194336, "step": 33538 }, { "epoch": 2.6439889633425304, "grad_norm": 0.5949347615242004, "learning_rate": 0.0005, "loss": 1.2785660028457642, "step": 33539 }, { "epoch": 2.6440677966101696, "grad_norm": 0.5328078269958496, "learning_rate": 0.0005, "loss": 1.2607868909835815, "step": 33540 }, { "epoch": 2.6441466298778087, "grad_norm": 0.5553485155105591, "learning_rate": 0.0005, "loss": 1.3137388229370117, "step": 33541 }, { "epoch": 2.6442254631454474, "grad_norm": 0.5766091346740723, "learning_rate": 0.0005, "loss": 1.3587706089019775, "step": 33542 }, { "epoch": 2.644304296413086, "grad_norm": 0.4434005916118622, "learning_rate": 0.0005, "loss": 1.2276020050048828, "step": 33543 }, { "epoch": 2.644383129680725, "grad_norm": 0.46715831756591797, "learning_rate": 0.0005, "loss": 1.1666455268859863, "step": 33544 }, { "epoch": 2.6444619629483643, "grad_norm": 0.5625002384185791, "learning_rate": 0.0005, "loss": 1.5512712001800537, "step": 33545 }, { "epoch": 2.644540796216003, "grad_norm": 0.6344094276428223, "learning_rate": 0.0005, "loss": 1.5586847066879272, "step": 33546 }, { "epoch": 2.644619629483642, "grad_norm": 0.5991416573524475, "learning_rate": 0.0005, "loss": 1.3712518215179443, "step": 33547 }, { "epoch": 2.644698462751281, "grad_norm": 0.47781842947006226, "learning_rate": 0.0005, "loss": 1.2695878744125366, "step": 33548 }, { "epoch": 2.64477729601892, "grad_norm": 0.49024754762649536, "learning_rate": 0.0005, "loss": 1.2629672288894653, "step": 33549 }, { "epoch": 2.644856129286559, "grad_norm": 0.39942798018455505, "learning_rate": 0.0005, "loss": 1.0639503002166748, "step": 33550 }, { "epoch": 2.644934962554198, "grad_norm": 0.6172943115234375, "learning_rate": 0.0005, "loss": 1.4124102592468262, "step": 33551 }, { "epoch": 2.645013795821837, "grad_norm": 0.6450144648551941, "learning_rate": 0.0005, "loss": 1.4483239650726318, "step": 33552 }, { "epoch": 2.6450926290894756, "grad_norm": 0.5386847257614136, "learning_rate": 0.0005, "loss": 1.4771627187728882, "step": 33553 }, { "epoch": 2.6451714623571148, "grad_norm": 0.6693621873855591, "learning_rate": 0.0005, "loss": 1.4548544883728027, "step": 33554 }, { "epoch": 2.645250295624754, "grad_norm": 0.5586084127426147, "learning_rate": 0.0005, "loss": 1.2743765115737915, "step": 33555 }, { "epoch": 2.6453291288923926, "grad_norm": 0.4651370942592621, "learning_rate": 0.0005, "loss": 1.2192732095718384, "step": 33556 }, { "epoch": 2.6454079621600313, "grad_norm": 0.47790369391441345, "learning_rate": 0.0005, "loss": 1.2181127071380615, "step": 33557 }, { "epoch": 2.6454867954276704, "grad_norm": 0.977759838104248, "learning_rate": 0.0005, "loss": 1.5584865808486938, "step": 33558 }, { "epoch": 2.6455656286953095, "grad_norm": 0.5538826584815979, "learning_rate": 0.0005, "loss": 1.5901219844818115, "step": 33559 }, { "epoch": 2.6456444619629482, "grad_norm": 0.5275359749794006, "learning_rate": 0.0005, "loss": 1.2326536178588867, "step": 33560 }, { "epoch": 2.6457232952305874, "grad_norm": 0.47140899300575256, "learning_rate": 0.0005, "loss": 1.1616257429122925, "step": 33561 }, { "epoch": 2.645802128498226, "grad_norm": 0.4528597295284271, "learning_rate": 0.0005, "loss": 1.1557157039642334, "step": 33562 }, { "epoch": 2.645880961765865, "grad_norm": 0.4189021587371826, "learning_rate": 0.0005, "loss": 1.3136472702026367, "step": 33563 }, { "epoch": 2.6459597950335043, "grad_norm": 0.502751350402832, "learning_rate": 0.0005, "loss": 1.3885537385940552, "step": 33564 }, { "epoch": 2.646038628301143, "grad_norm": 0.5289939045906067, "learning_rate": 0.0005, "loss": 1.202903389930725, "step": 33565 }, { "epoch": 2.646117461568782, "grad_norm": 0.4426627457141876, "learning_rate": 0.0005, "loss": 1.324570894241333, "step": 33566 }, { "epoch": 2.646196294836421, "grad_norm": 0.6073133945465088, "learning_rate": 0.0005, "loss": 1.554371953010559, "step": 33567 }, { "epoch": 2.64627512810406, "grad_norm": 0.5155795216560364, "learning_rate": 0.0005, "loss": 1.5802390575408936, "step": 33568 }, { "epoch": 2.646353961371699, "grad_norm": 0.44159260392189026, "learning_rate": 0.0005, "loss": 1.2687854766845703, "step": 33569 }, { "epoch": 2.646432794639338, "grad_norm": 0.4360869824886322, "learning_rate": 0.0005, "loss": 1.3898769617080688, "step": 33570 }, { "epoch": 2.6465116279069765, "grad_norm": 0.5138629078865051, "learning_rate": 0.0005, "loss": 1.3853904008865356, "step": 33571 }, { "epoch": 2.6465904611746156, "grad_norm": 0.528864324092865, "learning_rate": 0.0005, "loss": 1.2496455907821655, "step": 33572 }, { "epoch": 2.6466692944422547, "grad_norm": 0.5485350489616394, "learning_rate": 0.0005, "loss": 1.4459362030029297, "step": 33573 }, { "epoch": 2.6467481277098934, "grad_norm": 0.5017878413200378, "learning_rate": 0.0005, "loss": 1.4172780513763428, "step": 33574 }, { "epoch": 2.6468269609775326, "grad_norm": 0.9671072363853455, "learning_rate": 0.0005, "loss": 1.4800950288772583, "step": 33575 }, { "epoch": 2.6469057942451713, "grad_norm": 0.44375547766685486, "learning_rate": 0.0005, "loss": 1.3745818138122559, "step": 33576 }, { "epoch": 2.6469846275128104, "grad_norm": 0.527914822101593, "learning_rate": 0.0005, "loss": 1.3419687747955322, "step": 33577 }, { "epoch": 2.6470634607804495, "grad_norm": 0.46404653787612915, "learning_rate": 0.0005, "loss": 1.3520054817199707, "step": 33578 }, { "epoch": 2.647142294048088, "grad_norm": 0.4776843786239624, "learning_rate": 0.0005, "loss": 1.3247077465057373, "step": 33579 }, { "epoch": 2.6472211273157273, "grad_norm": 0.5066940784454346, "learning_rate": 0.0005, "loss": 1.2296264171600342, "step": 33580 }, { "epoch": 2.647299960583366, "grad_norm": 0.4838160574436188, "learning_rate": 0.0005, "loss": 1.2554346323013306, "step": 33581 }, { "epoch": 2.647378793851005, "grad_norm": 0.42023766040802, "learning_rate": 0.0005, "loss": 1.1992511749267578, "step": 33582 }, { "epoch": 2.6474576271186443, "grad_norm": 0.40688616037368774, "learning_rate": 0.0005, "loss": 1.2031276226043701, "step": 33583 }, { "epoch": 2.647536460386283, "grad_norm": 0.4922659993171692, "learning_rate": 0.0005, "loss": 1.524533748626709, "step": 33584 }, { "epoch": 2.647615293653922, "grad_norm": 0.5280145406723022, "learning_rate": 0.0005, "loss": 1.4512656927108765, "step": 33585 }, { "epoch": 2.647694126921561, "grad_norm": 0.4752250611782074, "learning_rate": 0.0005, "loss": 1.2481043338775635, "step": 33586 }, { "epoch": 2.6477729601892, "grad_norm": 0.4457467496395111, "learning_rate": 0.0005, "loss": 1.2790569067001343, "step": 33587 }, { "epoch": 2.6478517934568386, "grad_norm": 0.6912898421287537, "learning_rate": 0.0005, "loss": 1.2861073017120361, "step": 33588 }, { "epoch": 2.6479306267244778, "grad_norm": 0.4327273368835449, "learning_rate": 0.0005, "loss": 1.3021228313446045, "step": 33589 }, { "epoch": 2.6480094599921165, "grad_norm": 0.5260512232780457, "learning_rate": 0.0005, "loss": 1.3744364976882935, "step": 33590 }, { "epoch": 2.6480882932597556, "grad_norm": 0.4567514657974243, "learning_rate": 0.0005, "loss": 1.3462796211242676, "step": 33591 }, { "epoch": 2.6481671265273947, "grad_norm": 0.6700188517570496, "learning_rate": 0.0005, "loss": 1.3125860691070557, "step": 33592 }, { "epoch": 2.6482459597950334, "grad_norm": 0.4771108627319336, "learning_rate": 0.0005, "loss": 1.3599931001663208, "step": 33593 }, { "epoch": 2.6483247930626725, "grad_norm": 0.555113673210144, "learning_rate": 0.0005, "loss": 1.378566026687622, "step": 33594 }, { "epoch": 2.6484036263303112, "grad_norm": 0.4588455259799957, "learning_rate": 0.0005, "loss": 1.3546063899993896, "step": 33595 }, { "epoch": 2.6484824595979504, "grad_norm": 0.564012348651886, "learning_rate": 0.0005, "loss": 1.5151419639587402, "step": 33596 }, { "epoch": 2.6485612928655895, "grad_norm": 0.6011912226676941, "learning_rate": 0.0005, "loss": 1.4289920330047607, "step": 33597 }, { "epoch": 2.648640126133228, "grad_norm": 0.5237701535224915, "learning_rate": 0.0005, "loss": 1.4041165113449097, "step": 33598 }, { "epoch": 2.6487189594008673, "grad_norm": 0.4450945258140564, "learning_rate": 0.0005, "loss": 1.2563035488128662, "step": 33599 }, { "epoch": 2.648797792668506, "grad_norm": 0.4683753252029419, "learning_rate": 0.0005, "loss": 1.2661652565002441, "step": 33600 }, { "epoch": 2.648876625936145, "grad_norm": 0.4753384292125702, "learning_rate": 0.0005, "loss": 1.2197299003601074, "step": 33601 }, { "epoch": 2.648955459203784, "grad_norm": 0.5444121956825256, "learning_rate": 0.0005, "loss": 1.408469319343567, "step": 33602 }, { "epoch": 2.649034292471423, "grad_norm": 0.7447605133056641, "learning_rate": 0.0005, "loss": 1.399470567703247, "step": 33603 }, { "epoch": 2.6491131257390617, "grad_norm": 0.5279456377029419, "learning_rate": 0.0005, "loss": 1.4110051393508911, "step": 33604 }, { "epoch": 2.649191959006701, "grad_norm": 0.4786430895328522, "learning_rate": 0.0005, "loss": 1.393542766571045, "step": 33605 }, { "epoch": 2.64927079227434, "grad_norm": 0.4961661994457245, "learning_rate": 0.0005, "loss": 1.3736696243286133, "step": 33606 }, { "epoch": 2.6493496255419786, "grad_norm": 0.4871704578399658, "learning_rate": 0.0005, "loss": 1.4101181030273438, "step": 33607 }, { "epoch": 2.6494284588096177, "grad_norm": 0.41324689984321594, "learning_rate": 0.0005, "loss": 1.2571452856063843, "step": 33608 }, { "epoch": 2.6495072920772564, "grad_norm": 0.590699315071106, "learning_rate": 0.0005, "loss": 1.4992008209228516, "step": 33609 }, { "epoch": 2.6495861253448956, "grad_norm": 0.5027836561203003, "learning_rate": 0.0005, "loss": 1.3784619569778442, "step": 33610 }, { "epoch": 2.6496649586125347, "grad_norm": 0.482022762298584, "learning_rate": 0.0005, "loss": 1.2747513055801392, "step": 33611 }, { "epoch": 2.6497437918801734, "grad_norm": 0.40827488899230957, "learning_rate": 0.0005, "loss": 1.3348407745361328, "step": 33612 }, { "epoch": 2.6498226251478125, "grad_norm": 0.4030497074127197, "learning_rate": 0.0005, "loss": 1.0289857387542725, "step": 33613 }, { "epoch": 2.649901458415451, "grad_norm": 0.5017326474189758, "learning_rate": 0.0005, "loss": 1.2954990863800049, "step": 33614 }, { "epoch": 2.6499802916830903, "grad_norm": 0.4681086242198944, "learning_rate": 0.0005, "loss": 1.3779009580612183, "step": 33615 }, { "epoch": 2.6500591249507295, "grad_norm": 0.5450279712677002, "learning_rate": 0.0005, "loss": 1.3706430196762085, "step": 33616 }, { "epoch": 2.650137958218368, "grad_norm": 0.47481510043144226, "learning_rate": 0.0005, "loss": 1.4161701202392578, "step": 33617 }, { "epoch": 2.650216791486007, "grad_norm": 0.5277593731880188, "learning_rate": 0.0005, "loss": 1.2203550338745117, "step": 33618 }, { "epoch": 2.650295624753646, "grad_norm": 0.537293553352356, "learning_rate": 0.0005, "loss": 1.1276719570159912, "step": 33619 }, { "epoch": 2.650374458021285, "grad_norm": 0.5977296829223633, "learning_rate": 0.0005, "loss": 1.342604637145996, "step": 33620 }, { "epoch": 2.650453291288924, "grad_norm": 0.5068870782852173, "learning_rate": 0.0005, "loss": 1.3652372360229492, "step": 33621 }, { "epoch": 2.650532124556563, "grad_norm": 0.5008274912834167, "learning_rate": 0.0005, "loss": 1.284395694732666, "step": 33622 }, { "epoch": 2.6506109578242016, "grad_norm": 0.5786631107330322, "learning_rate": 0.0005, "loss": 1.4034452438354492, "step": 33623 }, { "epoch": 2.6506897910918408, "grad_norm": 0.5399787425994873, "learning_rate": 0.0005, "loss": 1.4806742668151855, "step": 33624 }, { "epoch": 2.65076862435948, "grad_norm": 0.5298895239830017, "learning_rate": 0.0005, "loss": 1.4695112705230713, "step": 33625 }, { "epoch": 2.6508474576271186, "grad_norm": 0.41174474358558655, "learning_rate": 0.0005, "loss": 1.1330808401107788, "step": 33626 }, { "epoch": 2.6509262908947577, "grad_norm": 0.5256377458572388, "learning_rate": 0.0005, "loss": 1.4756200313568115, "step": 33627 }, { "epoch": 2.6510051241623964, "grad_norm": 0.398447722196579, "learning_rate": 0.0005, "loss": 1.1296947002410889, "step": 33628 }, { "epoch": 2.6510839574300356, "grad_norm": 0.5427692532539368, "learning_rate": 0.0005, "loss": 1.4767543077468872, "step": 33629 }, { "epoch": 2.6511627906976747, "grad_norm": 0.46019676327705383, "learning_rate": 0.0005, "loss": 1.05708646774292, "step": 33630 }, { "epoch": 2.6512416239653134, "grad_norm": 0.46325695514678955, "learning_rate": 0.0005, "loss": 1.3246889114379883, "step": 33631 }, { "epoch": 2.651320457232952, "grad_norm": 0.45286545157432556, "learning_rate": 0.0005, "loss": 1.4763003587722778, "step": 33632 }, { "epoch": 2.651399290500591, "grad_norm": 0.5068168044090271, "learning_rate": 0.0005, "loss": 1.307686686515808, "step": 33633 }, { "epoch": 2.6514781237682303, "grad_norm": 0.5196764469146729, "learning_rate": 0.0005, "loss": 1.2792487144470215, "step": 33634 }, { "epoch": 2.651556957035869, "grad_norm": 0.42915382981300354, "learning_rate": 0.0005, "loss": 1.2144924402236938, "step": 33635 }, { "epoch": 2.651635790303508, "grad_norm": 0.48312902450561523, "learning_rate": 0.0005, "loss": 1.4322199821472168, "step": 33636 }, { "epoch": 2.651714623571147, "grad_norm": 0.4355242848396301, "learning_rate": 0.0005, "loss": 1.1551077365875244, "step": 33637 }, { "epoch": 2.651793456838786, "grad_norm": 0.49813276529312134, "learning_rate": 0.0005, "loss": 1.2046823501586914, "step": 33638 }, { "epoch": 2.651872290106425, "grad_norm": 0.4718290865421295, "learning_rate": 0.0005, "loss": 1.3160218000411987, "step": 33639 }, { "epoch": 2.651951123374064, "grad_norm": 0.4223007559776306, "learning_rate": 0.0005, "loss": 1.3117276430130005, "step": 33640 }, { "epoch": 2.652029956641703, "grad_norm": 0.5599148273468018, "learning_rate": 0.0005, "loss": 1.5190794467926025, "step": 33641 }, { "epoch": 2.6521087899093416, "grad_norm": 0.5012010931968689, "learning_rate": 0.0005, "loss": 1.3024277687072754, "step": 33642 }, { "epoch": 2.6521876231769808, "grad_norm": 0.4686780869960785, "learning_rate": 0.0005, "loss": 1.4661577939987183, "step": 33643 }, { "epoch": 2.65226645644462, "grad_norm": 0.4840196371078491, "learning_rate": 0.0005, "loss": 1.2447009086608887, "step": 33644 }, { "epoch": 2.6523452897122586, "grad_norm": 0.8975825905799866, "learning_rate": 0.0005, "loss": 1.2639515399932861, "step": 33645 }, { "epoch": 2.6524241229798973, "grad_norm": 0.5354563593864441, "learning_rate": 0.0005, "loss": 1.3201271295547485, "step": 33646 }, { "epoch": 2.6525029562475364, "grad_norm": 0.5410889387130737, "learning_rate": 0.0005, "loss": 1.561202049255371, "step": 33647 }, { "epoch": 2.6525817895151755, "grad_norm": 0.600364089012146, "learning_rate": 0.0005, "loss": 1.4312224388122559, "step": 33648 }, { "epoch": 2.6526606227828142, "grad_norm": 0.6126238703727722, "learning_rate": 0.0005, "loss": 1.313089370727539, "step": 33649 }, { "epoch": 2.6527394560504534, "grad_norm": 0.5180489420890808, "learning_rate": 0.0005, "loss": 1.4095230102539062, "step": 33650 }, { "epoch": 2.652818289318092, "grad_norm": 0.45926713943481445, "learning_rate": 0.0005, "loss": 1.4097263813018799, "step": 33651 }, { "epoch": 2.652897122585731, "grad_norm": 0.5645545721054077, "learning_rate": 0.0005, "loss": 1.38292396068573, "step": 33652 }, { "epoch": 2.6529759558533703, "grad_norm": 0.4324902892112732, "learning_rate": 0.0005, "loss": 1.1720654964447021, "step": 33653 }, { "epoch": 2.653054789121009, "grad_norm": 0.4636397361755371, "learning_rate": 0.0005, "loss": 1.2598358392715454, "step": 33654 }, { "epoch": 2.653133622388648, "grad_norm": 0.47811928391456604, "learning_rate": 0.0005, "loss": 1.5154969692230225, "step": 33655 }, { "epoch": 2.653212455656287, "grad_norm": 0.5247020125389099, "learning_rate": 0.0005, "loss": 1.3748972415924072, "step": 33656 }, { "epoch": 2.653291288923926, "grad_norm": 0.4640606939792633, "learning_rate": 0.0005, "loss": 1.3888518810272217, "step": 33657 }, { "epoch": 2.653370122191565, "grad_norm": 0.5040827393531799, "learning_rate": 0.0005, "loss": 1.206333041191101, "step": 33658 }, { "epoch": 2.653448955459204, "grad_norm": 0.4823290705680847, "learning_rate": 0.0005, "loss": 1.1149040460586548, "step": 33659 }, { "epoch": 2.6535277887268425, "grad_norm": 0.4773423671722412, "learning_rate": 0.0005, "loss": 1.4705860614776611, "step": 33660 }, { "epoch": 2.6536066219944816, "grad_norm": 0.5654106736183167, "learning_rate": 0.0005, "loss": 1.2688984870910645, "step": 33661 }, { "epoch": 2.6536854552621207, "grad_norm": 0.4873439371585846, "learning_rate": 0.0005, "loss": 1.2803388833999634, "step": 33662 }, { "epoch": 2.6537642885297594, "grad_norm": 0.4426348805427551, "learning_rate": 0.0005, "loss": 1.1230038404464722, "step": 33663 }, { "epoch": 2.6538431217973986, "grad_norm": 0.4737931191921234, "learning_rate": 0.0005, "loss": 1.2725656032562256, "step": 33664 }, { "epoch": 2.6539219550650373, "grad_norm": 0.5046881437301636, "learning_rate": 0.0005, "loss": 1.4746499061584473, "step": 33665 }, { "epoch": 2.6540007883326764, "grad_norm": 0.4726548194885254, "learning_rate": 0.0005, "loss": 1.2852321863174438, "step": 33666 }, { "epoch": 2.6540796216003155, "grad_norm": 0.5016148686408997, "learning_rate": 0.0005, "loss": 1.52241051197052, "step": 33667 }, { "epoch": 2.654158454867954, "grad_norm": 0.5615840554237366, "learning_rate": 0.0005, "loss": 1.1948188543319702, "step": 33668 }, { "epoch": 2.6542372881355933, "grad_norm": 0.4166989028453827, "learning_rate": 0.0005, "loss": 1.2330105304718018, "step": 33669 }, { "epoch": 2.654316121403232, "grad_norm": 0.5734806656837463, "learning_rate": 0.0005, "loss": 1.445359706878662, "step": 33670 }, { "epoch": 2.654394954670871, "grad_norm": 0.44761720299720764, "learning_rate": 0.0005, "loss": 1.1852715015411377, "step": 33671 }, { "epoch": 2.6544737879385103, "grad_norm": 0.578900158405304, "learning_rate": 0.0005, "loss": 1.358958125114441, "step": 33672 }, { "epoch": 2.654552621206149, "grad_norm": 0.4962948262691498, "learning_rate": 0.0005, "loss": 1.4228864908218384, "step": 33673 }, { "epoch": 2.6546314544737877, "grad_norm": 0.7945778965950012, "learning_rate": 0.0005, "loss": 1.3683931827545166, "step": 33674 }, { "epoch": 2.654710287741427, "grad_norm": 0.41983655095100403, "learning_rate": 0.0005, "loss": 1.247298240661621, "step": 33675 }, { "epoch": 2.654789121009066, "grad_norm": 0.465510755777359, "learning_rate": 0.0005, "loss": 1.2768547534942627, "step": 33676 }, { "epoch": 2.6548679542767046, "grad_norm": 0.5081749558448792, "learning_rate": 0.0005, "loss": 1.4293160438537598, "step": 33677 }, { "epoch": 2.6549467875443438, "grad_norm": 0.4451064169406891, "learning_rate": 0.0005, "loss": 1.2281261682510376, "step": 33678 }, { "epoch": 2.6550256208119825, "grad_norm": 0.4803715646266937, "learning_rate": 0.0005, "loss": 1.3849678039550781, "step": 33679 }, { "epoch": 2.6551044540796216, "grad_norm": 0.4749743640422821, "learning_rate": 0.0005, "loss": 1.2474571466445923, "step": 33680 }, { "epoch": 2.6551832873472607, "grad_norm": 0.9841918349266052, "learning_rate": 0.0005, "loss": 1.521359920501709, "step": 33681 }, { "epoch": 2.6552621206148994, "grad_norm": 0.5275358557701111, "learning_rate": 0.0005, "loss": 1.26236093044281, "step": 33682 }, { "epoch": 2.6553409538825385, "grad_norm": 0.39670008420944214, "learning_rate": 0.0005, "loss": 1.1350902318954468, "step": 33683 }, { "epoch": 2.6554197871501772, "grad_norm": 0.6691266298294067, "learning_rate": 0.0005, "loss": 1.3325269222259521, "step": 33684 }, { "epoch": 2.6554986204178164, "grad_norm": 0.5680819153785706, "learning_rate": 0.0005, "loss": 1.603912353515625, "step": 33685 }, { "epoch": 2.6555774536854555, "grad_norm": 0.5996812582015991, "learning_rate": 0.0005, "loss": 1.330810546875, "step": 33686 }, { "epoch": 2.655656286953094, "grad_norm": 0.5151841640472412, "learning_rate": 0.0005, "loss": 1.5096197128295898, "step": 33687 }, { "epoch": 2.655735120220733, "grad_norm": 0.5356025695800781, "learning_rate": 0.0005, "loss": 1.377109169960022, "step": 33688 }, { "epoch": 2.655813953488372, "grad_norm": 0.5102872848510742, "learning_rate": 0.0005, "loss": 1.3111319541931152, "step": 33689 }, { "epoch": 2.655892786756011, "grad_norm": 0.4649737775325775, "learning_rate": 0.0005, "loss": 1.2420903444290161, "step": 33690 }, { "epoch": 2.65597162002365, "grad_norm": 0.5651558041572571, "learning_rate": 0.0005, "loss": 1.328413963317871, "step": 33691 }, { "epoch": 2.656050453291289, "grad_norm": 0.4905729293823242, "learning_rate": 0.0005, "loss": 1.2074958086013794, "step": 33692 }, { "epoch": 2.6561292865589277, "grad_norm": 0.46336835622787476, "learning_rate": 0.0005, "loss": 1.2216413021087646, "step": 33693 }, { "epoch": 2.656208119826567, "grad_norm": 0.45129427313804626, "learning_rate": 0.0005, "loss": 1.2743130922317505, "step": 33694 }, { "epoch": 2.656286953094206, "grad_norm": 1.4413117170333862, "learning_rate": 0.0005, "loss": 1.352672815322876, "step": 33695 }, { "epoch": 2.6563657863618446, "grad_norm": 0.551480233669281, "learning_rate": 0.0005, "loss": 1.2911081314086914, "step": 33696 }, { "epoch": 2.6564446196294837, "grad_norm": 0.5643500685691833, "learning_rate": 0.0005, "loss": 1.454586386680603, "step": 33697 }, { "epoch": 2.6565234528971224, "grad_norm": 0.4691450595855713, "learning_rate": 0.0005, "loss": 1.4817579984664917, "step": 33698 }, { "epoch": 2.6566022861647616, "grad_norm": 0.5080392956733704, "learning_rate": 0.0005, "loss": 1.308770775794983, "step": 33699 }, { "epoch": 2.6566811194324007, "grad_norm": 0.4819696247577667, "learning_rate": 0.0005, "loss": 1.3322169780731201, "step": 33700 }, { "epoch": 2.6567599527000394, "grad_norm": 0.48760315775871277, "learning_rate": 0.0005, "loss": 1.2062010765075684, "step": 33701 }, { "epoch": 2.656838785967678, "grad_norm": 0.42091381549835205, "learning_rate": 0.0005, "loss": 1.153127908706665, "step": 33702 }, { "epoch": 2.656917619235317, "grad_norm": 0.43892210721969604, "learning_rate": 0.0005, "loss": 1.2597568035125732, "step": 33703 }, { "epoch": 2.6569964525029564, "grad_norm": 0.5564805269241333, "learning_rate": 0.0005, "loss": 1.3843897581100464, "step": 33704 }, { "epoch": 2.657075285770595, "grad_norm": 0.4358648359775543, "learning_rate": 0.0005, "loss": 1.3949569463729858, "step": 33705 }, { "epoch": 2.657154119038234, "grad_norm": 0.5925253033638, "learning_rate": 0.0005, "loss": 1.5578219890594482, "step": 33706 }, { "epoch": 2.657232952305873, "grad_norm": 0.5457097291946411, "learning_rate": 0.0005, "loss": 1.3189892768859863, "step": 33707 }, { "epoch": 2.657311785573512, "grad_norm": 0.44161584973335266, "learning_rate": 0.0005, "loss": 1.273432731628418, "step": 33708 }, { "epoch": 2.657390618841151, "grad_norm": 0.4466342628002167, "learning_rate": 0.0005, "loss": 1.3277359008789062, "step": 33709 }, { "epoch": 2.65746945210879, "grad_norm": 0.4264914095401764, "learning_rate": 0.0005, "loss": 1.249718189239502, "step": 33710 }, { "epoch": 2.657548285376429, "grad_norm": 0.43509483337402344, "learning_rate": 0.0005, "loss": 1.310449242591858, "step": 33711 }, { "epoch": 2.6576271186440676, "grad_norm": 0.456246942281723, "learning_rate": 0.0005, "loss": 1.3377959728240967, "step": 33712 }, { "epoch": 2.6577059519117068, "grad_norm": 0.5837259888648987, "learning_rate": 0.0005, "loss": 1.4983373880386353, "step": 33713 }, { "epoch": 2.657784785179346, "grad_norm": 0.40323206782341003, "learning_rate": 0.0005, "loss": 1.3930752277374268, "step": 33714 }, { "epoch": 2.6578636184469846, "grad_norm": 0.47487765550613403, "learning_rate": 0.0005, "loss": 1.3630516529083252, "step": 33715 }, { "epoch": 2.6579424517146237, "grad_norm": 0.40774455666542053, "learning_rate": 0.0005, "loss": 1.257164716720581, "step": 33716 }, { "epoch": 2.6580212849822624, "grad_norm": 0.46238815784454346, "learning_rate": 0.0005, "loss": 1.2517235279083252, "step": 33717 }, { "epoch": 2.6581001182499016, "grad_norm": 0.510570764541626, "learning_rate": 0.0005, "loss": 1.4472613334655762, "step": 33718 }, { "epoch": 2.6581789515175402, "grad_norm": 0.4069150388240814, "learning_rate": 0.0005, "loss": 1.2290709018707275, "step": 33719 }, { "epoch": 2.6582577847851794, "grad_norm": 0.5701431035995483, "learning_rate": 0.0005, "loss": 1.3248376846313477, "step": 33720 }, { "epoch": 2.658336618052818, "grad_norm": 0.7813193202018738, "learning_rate": 0.0005, "loss": 1.3064115047454834, "step": 33721 }, { "epoch": 2.658415451320457, "grad_norm": 0.5114237666130066, "learning_rate": 0.0005, "loss": 1.364758014678955, "step": 33722 }, { "epoch": 2.6584942845880963, "grad_norm": 0.689318835735321, "learning_rate": 0.0005, "loss": 1.5231353044509888, "step": 33723 }, { "epoch": 2.658573117855735, "grad_norm": 0.5807201266288757, "learning_rate": 0.0005, "loss": 1.3794034719467163, "step": 33724 }, { "epoch": 2.658651951123374, "grad_norm": 0.4954405128955841, "learning_rate": 0.0005, "loss": 1.3721187114715576, "step": 33725 }, { "epoch": 2.658730784391013, "grad_norm": 0.5644300580024719, "learning_rate": 0.0005, "loss": 1.4056397676467896, "step": 33726 }, { "epoch": 2.658809617658652, "grad_norm": 0.4758417308330536, "learning_rate": 0.0005, "loss": 1.335590124130249, "step": 33727 }, { "epoch": 2.658888450926291, "grad_norm": 0.6150546669960022, "learning_rate": 0.0005, "loss": 1.2879161834716797, "step": 33728 }, { "epoch": 2.65896728419393, "grad_norm": 0.4449315667152405, "learning_rate": 0.0005, "loss": 1.2277908325195312, "step": 33729 }, { "epoch": 2.659046117461569, "grad_norm": 0.4694287180900574, "learning_rate": 0.0005, "loss": 1.267748236656189, "step": 33730 }, { "epoch": 2.6591249507292076, "grad_norm": 0.5227731466293335, "learning_rate": 0.0005, "loss": 1.243405818939209, "step": 33731 }, { "epoch": 2.6592037839968468, "grad_norm": 0.49865320324897766, "learning_rate": 0.0005, "loss": 1.2549018859863281, "step": 33732 }, { "epoch": 2.6592826172644854, "grad_norm": 0.5183569192886353, "learning_rate": 0.0005, "loss": 1.4880660772323608, "step": 33733 }, { "epoch": 2.6593614505321246, "grad_norm": 0.5220175981521606, "learning_rate": 0.0005, "loss": 1.4693822860717773, "step": 33734 }, { "epoch": 2.6594402837997633, "grad_norm": 0.48676109313964844, "learning_rate": 0.0005, "loss": 1.3469157218933105, "step": 33735 }, { "epoch": 2.6595191170674024, "grad_norm": 0.4769632816314697, "learning_rate": 0.0005, "loss": 1.3142707347869873, "step": 33736 }, { "epoch": 2.6595979503350415, "grad_norm": 0.5957263112068176, "learning_rate": 0.0005, "loss": 1.4302374124526978, "step": 33737 }, { "epoch": 2.6596767836026802, "grad_norm": 0.3945508301258087, "learning_rate": 0.0005, "loss": 0.9976146817207336, "step": 33738 }, { "epoch": 2.6597556168703194, "grad_norm": 0.49310624599456787, "learning_rate": 0.0005, "loss": 1.440782070159912, "step": 33739 }, { "epoch": 2.659834450137958, "grad_norm": 0.4272288978099823, "learning_rate": 0.0005, "loss": 1.42258882522583, "step": 33740 }, { "epoch": 2.659913283405597, "grad_norm": 0.5292540192604065, "learning_rate": 0.0005, "loss": 1.4275195598602295, "step": 33741 }, { "epoch": 2.6599921166732363, "grad_norm": 0.4431876838207245, "learning_rate": 0.0005, "loss": 1.3970537185668945, "step": 33742 }, { "epoch": 2.660070949940875, "grad_norm": 0.5323556661605835, "learning_rate": 0.0005, "loss": 1.3561404943466187, "step": 33743 }, { "epoch": 2.660149783208514, "grad_norm": 0.49202728271484375, "learning_rate": 0.0005, "loss": 1.2803194522857666, "step": 33744 }, { "epoch": 2.660228616476153, "grad_norm": 0.48267266154289246, "learning_rate": 0.0005, "loss": 1.3575114011764526, "step": 33745 }, { "epoch": 2.660307449743792, "grad_norm": 0.5358701348304749, "learning_rate": 0.0005, "loss": 1.2765185832977295, "step": 33746 }, { "epoch": 2.660386283011431, "grad_norm": 0.4913252294063568, "learning_rate": 0.0005, "loss": 1.4307267665863037, "step": 33747 }, { "epoch": 2.66046511627907, "grad_norm": 0.46101564168930054, "learning_rate": 0.0005, "loss": 1.1812883615493774, "step": 33748 }, { "epoch": 2.6605439495467085, "grad_norm": 0.4947565495967865, "learning_rate": 0.0005, "loss": 1.2820556163787842, "step": 33749 }, { "epoch": 2.6606227828143476, "grad_norm": 0.4774516820907593, "learning_rate": 0.0005, "loss": 1.3548243045806885, "step": 33750 }, { "epoch": 2.6607016160819867, "grad_norm": 0.5244342088699341, "learning_rate": 0.0005, "loss": 1.4156545400619507, "step": 33751 }, { "epoch": 2.6607804493496254, "grad_norm": 0.6025017499923706, "learning_rate": 0.0005, "loss": 1.556653380393982, "step": 33752 }, { "epoch": 2.6608592826172646, "grad_norm": 0.613896369934082, "learning_rate": 0.0005, "loss": 1.501002311706543, "step": 33753 }, { "epoch": 2.6609381158849033, "grad_norm": 0.4589191675186157, "learning_rate": 0.0005, "loss": 1.2292449474334717, "step": 33754 }, { "epoch": 2.6610169491525424, "grad_norm": 0.42944422364234924, "learning_rate": 0.0005, "loss": 1.2243154048919678, "step": 33755 }, { "epoch": 2.6610957824201815, "grad_norm": 0.44062015414237976, "learning_rate": 0.0005, "loss": 0.9883240461349487, "step": 33756 }, { "epoch": 2.66117461568782, "grad_norm": 0.4757032096385956, "learning_rate": 0.0005, "loss": 1.36079740524292, "step": 33757 }, { "epoch": 2.6612534489554593, "grad_norm": 0.5726158022880554, "learning_rate": 0.0005, "loss": 1.4741795063018799, "step": 33758 }, { "epoch": 2.661332282223098, "grad_norm": 0.5200657248497009, "learning_rate": 0.0005, "loss": 1.202927589416504, "step": 33759 }, { "epoch": 2.661411115490737, "grad_norm": 0.5789062976837158, "learning_rate": 0.0005, "loss": 1.5503957271575928, "step": 33760 }, { "epoch": 2.6614899487583763, "grad_norm": 0.4715130925178528, "learning_rate": 0.0005, "loss": 1.1401832103729248, "step": 33761 }, { "epoch": 2.661568782026015, "grad_norm": 0.5341659784317017, "learning_rate": 0.0005, "loss": 1.4284619092941284, "step": 33762 }, { "epoch": 2.6616476152936537, "grad_norm": 0.5663230419158936, "learning_rate": 0.0005, "loss": 1.2616453170776367, "step": 33763 }, { "epoch": 2.661726448561293, "grad_norm": 0.5116846561431885, "learning_rate": 0.0005, "loss": 1.3550150394439697, "step": 33764 }, { "epoch": 2.661805281828932, "grad_norm": 0.4677538275718689, "learning_rate": 0.0005, "loss": 1.2201824188232422, "step": 33765 }, { "epoch": 2.6618841150965706, "grad_norm": 0.5958003997802734, "learning_rate": 0.0005, "loss": 1.4017715454101562, "step": 33766 }, { "epoch": 2.6619629483642098, "grad_norm": 0.4576050043106079, "learning_rate": 0.0005, "loss": 1.2208449840545654, "step": 33767 }, { "epoch": 2.6620417816318485, "grad_norm": 0.46862152218818665, "learning_rate": 0.0005, "loss": 1.3958518505096436, "step": 33768 }, { "epoch": 2.6621206148994876, "grad_norm": 0.4837351441383362, "learning_rate": 0.0005, "loss": 1.3382446765899658, "step": 33769 }, { "epoch": 2.6621994481671267, "grad_norm": 0.4710775315761566, "learning_rate": 0.0005, "loss": 1.3777129650115967, "step": 33770 }, { "epoch": 2.6622782814347654, "grad_norm": 0.6031640768051147, "learning_rate": 0.0005, "loss": 1.29667067527771, "step": 33771 }, { "epoch": 2.6623571147024045, "grad_norm": 0.5348234176635742, "learning_rate": 0.0005, "loss": 1.3491380214691162, "step": 33772 }, { "epoch": 2.6624359479700432, "grad_norm": 0.49260133504867554, "learning_rate": 0.0005, "loss": 1.457031011581421, "step": 33773 }, { "epoch": 2.6625147812376824, "grad_norm": 0.5406946539878845, "learning_rate": 0.0005, "loss": 1.2778315544128418, "step": 33774 }, { "epoch": 2.6625936145053215, "grad_norm": 0.49835386872291565, "learning_rate": 0.0005, "loss": 1.4253069162368774, "step": 33775 }, { "epoch": 2.66267244777296, "grad_norm": 0.6604825854301453, "learning_rate": 0.0005, "loss": 1.3969089984893799, "step": 33776 }, { "epoch": 2.662751281040599, "grad_norm": 0.4494474530220032, "learning_rate": 0.0005, "loss": 1.2694848775863647, "step": 33777 }, { "epoch": 2.662830114308238, "grad_norm": 0.4789874255657196, "learning_rate": 0.0005, "loss": 1.2723667621612549, "step": 33778 }, { "epoch": 2.662908947575877, "grad_norm": 0.5780317783355713, "learning_rate": 0.0005, "loss": 1.2752379179000854, "step": 33779 }, { "epoch": 2.662987780843516, "grad_norm": 0.5244918465614319, "learning_rate": 0.0005, "loss": 1.2920973300933838, "step": 33780 }, { "epoch": 2.663066614111155, "grad_norm": 0.6159688234329224, "learning_rate": 0.0005, "loss": 1.1839747428894043, "step": 33781 }, { "epoch": 2.6631454473787937, "grad_norm": 0.5682672262191772, "learning_rate": 0.0005, "loss": 1.3762822151184082, "step": 33782 }, { "epoch": 2.663224280646433, "grad_norm": 0.5912015438079834, "learning_rate": 0.0005, "loss": 1.3259114027023315, "step": 33783 }, { "epoch": 2.663303113914072, "grad_norm": 0.7054080963134766, "learning_rate": 0.0005, "loss": 1.422951579093933, "step": 33784 }, { "epoch": 2.6633819471817106, "grad_norm": 0.5057249069213867, "learning_rate": 0.0005, "loss": 1.4024699926376343, "step": 33785 }, { "epoch": 2.6634607804493498, "grad_norm": 0.5119203925132751, "learning_rate": 0.0005, "loss": 1.5764180421829224, "step": 33786 }, { "epoch": 2.6635396137169884, "grad_norm": 0.41626617312431335, "learning_rate": 0.0005, "loss": 1.0577932596206665, "step": 33787 }, { "epoch": 2.6636184469846276, "grad_norm": 0.5147650241851807, "learning_rate": 0.0005, "loss": 1.2695666551589966, "step": 33788 }, { "epoch": 2.6636972802522667, "grad_norm": 0.4718996584415436, "learning_rate": 0.0005, "loss": 1.406927227973938, "step": 33789 }, { "epoch": 2.6637761135199054, "grad_norm": 0.5362313389778137, "learning_rate": 0.0005, "loss": 1.3714255094528198, "step": 33790 }, { "epoch": 2.663854946787544, "grad_norm": 0.5519334673881531, "learning_rate": 0.0005, "loss": 1.4209206104278564, "step": 33791 }, { "epoch": 2.663933780055183, "grad_norm": 0.5894575119018555, "learning_rate": 0.0005, "loss": 1.470672607421875, "step": 33792 }, { "epoch": 2.6640126133228224, "grad_norm": 0.6461178660392761, "learning_rate": 0.0005, "loss": 1.4141392707824707, "step": 33793 }, { "epoch": 2.664091446590461, "grad_norm": 0.5918242335319519, "learning_rate": 0.0005, "loss": 1.426147222518921, "step": 33794 }, { "epoch": 2.6641702798581, "grad_norm": 0.43391162157058716, "learning_rate": 0.0005, "loss": 1.241197109222412, "step": 33795 }, { "epoch": 2.664249113125739, "grad_norm": 0.5304022431373596, "learning_rate": 0.0005, "loss": 1.3114420175552368, "step": 33796 }, { "epoch": 2.664327946393378, "grad_norm": 0.7301998734474182, "learning_rate": 0.0005, "loss": 1.4889031648635864, "step": 33797 }, { "epoch": 2.664406779661017, "grad_norm": 0.5961583256721497, "learning_rate": 0.0005, "loss": 1.467886209487915, "step": 33798 }, { "epoch": 2.664485612928656, "grad_norm": 0.4242018461227417, "learning_rate": 0.0005, "loss": 1.186713695526123, "step": 33799 }, { "epoch": 2.664564446196295, "grad_norm": 0.46894559264183044, "learning_rate": 0.0005, "loss": 1.3560256958007812, "step": 33800 }, { "epoch": 2.6646432794639336, "grad_norm": 0.8182317018508911, "learning_rate": 0.0005, "loss": 1.3924927711486816, "step": 33801 }, { "epoch": 2.664722112731573, "grad_norm": 0.4243202209472656, "learning_rate": 0.0005, "loss": 1.1978793144226074, "step": 33802 }, { "epoch": 2.664800945999212, "grad_norm": 0.49079832434654236, "learning_rate": 0.0005, "loss": 1.408611536026001, "step": 33803 }, { "epoch": 2.6648797792668506, "grad_norm": 0.5421884655952454, "learning_rate": 0.0005, "loss": 1.3313677310943604, "step": 33804 }, { "epoch": 2.6649586125344893, "grad_norm": 0.6103893518447876, "learning_rate": 0.0005, "loss": 1.396601676940918, "step": 33805 }, { "epoch": 2.6650374458021284, "grad_norm": 0.4893818497657776, "learning_rate": 0.0005, "loss": 1.205944538116455, "step": 33806 }, { "epoch": 2.6651162790697676, "grad_norm": 0.45153918862342834, "learning_rate": 0.0005, "loss": 1.2946444749832153, "step": 33807 }, { "epoch": 2.6651951123374062, "grad_norm": 0.5349622368812561, "learning_rate": 0.0005, "loss": 1.5846428871154785, "step": 33808 }, { "epoch": 2.6652739456050454, "grad_norm": 0.6183018088340759, "learning_rate": 0.0005, "loss": 1.3669617176055908, "step": 33809 }, { "epoch": 2.665352778872684, "grad_norm": 0.47421643137931824, "learning_rate": 0.0005, "loss": 1.3435002565383911, "step": 33810 }, { "epoch": 2.665431612140323, "grad_norm": 0.5092594027519226, "learning_rate": 0.0005, "loss": 1.241154670715332, "step": 33811 }, { "epoch": 2.6655104454079623, "grad_norm": 0.44212254881858826, "learning_rate": 0.0005, "loss": 1.2843036651611328, "step": 33812 }, { "epoch": 2.665589278675601, "grad_norm": 0.44874435663223267, "learning_rate": 0.0005, "loss": 1.2170312404632568, "step": 33813 }, { "epoch": 2.66566811194324, "grad_norm": 0.6063355803489685, "learning_rate": 0.0005, "loss": 1.6313846111297607, "step": 33814 }, { "epoch": 2.665746945210879, "grad_norm": 0.5179767608642578, "learning_rate": 0.0005, "loss": 1.314192295074463, "step": 33815 }, { "epoch": 2.665825778478518, "grad_norm": 0.46852800250053406, "learning_rate": 0.0005, "loss": 1.1546424627304077, "step": 33816 }, { "epoch": 2.665904611746157, "grad_norm": 0.6132033467292786, "learning_rate": 0.0005, "loss": 1.4999970197677612, "step": 33817 }, { "epoch": 2.665983445013796, "grad_norm": 0.4474392235279083, "learning_rate": 0.0005, "loss": 1.2795741558074951, "step": 33818 }, { "epoch": 2.6660622782814345, "grad_norm": 0.500007688999176, "learning_rate": 0.0005, "loss": 1.3062646389007568, "step": 33819 }, { "epoch": 2.6661411115490736, "grad_norm": 0.5160308480262756, "learning_rate": 0.0005, "loss": 1.425297737121582, "step": 33820 }, { "epoch": 2.6662199448167128, "grad_norm": 0.515270471572876, "learning_rate": 0.0005, "loss": 1.3667742013931274, "step": 33821 }, { "epoch": 2.6662987780843515, "grad_norm": 0.4450322687625885, "learning_rate": 0.0005, "loss": 1.228259563446045, "step": 33822 }, { "epoch": 2.6663776113519906, "grad_norm": 0.4848383665084839, "learning_rate": 0.0005, "loss": 1.3142478466033936, "step": 33823 }, { "epoch": 2.6664564446196293, "grad_norm": 0.8318251371383667, "learning_rate": 0.0005, "loss": 1.076017141342163, "step": 33824 }, { "epoch": 2.6665352778872684, "grad_norm": 0.5000246167182922, "learning_rate": 0.0005, "loss": 1.3752392530441284, "step": 33825 }, { "epoch": 2.6666141111549075, "grad_norm": 0.547576367855072, "learning_rate": 0.0005, "loss": 1.4139814376831055, "step": 33826 }, { "epoch": 2.6666929444225462, "grad_norm": 0.5781720280647278, "learning_rate": 0.0005, "loss": 1.2427849769592285, "step": 33827 }, { "epoch": 2.6667717776901854, "grad_norm": 0.5503439903259277, "learning_rate": 0.0005, "loss": 1.2478663921356201, "step": 33828 }, { "epoch": 2.666850610957824, "grad_norm": 0.7458693385124207, "learning_rate": 0.0005, "loss": 1.2283287048339844, "step": 33829 }, { "epoch": 2.666929444225463, "grad_norm": 0.5311889052391052, "learning_rate": 0.0005, "loss": 1.4191994667053223, "step": 33830 }, { "epoch": 2.6670082774931023, "grad_norm": 0.5426300168037415, "learning_rate": 0.0005, "loss": 1.2617117166519165, "step": 33831 }, { "epoch": 2.667087110760741, "grad_norm": 0.6445571780204773, "learning_rate": 0.0005, "loss": 1.21909499168396, "step": 33832 }, { "epoch": 2.66716594402838, "grad_norm": 0.4919879734516144, "learning_rate": 0.0005, "loss": 1.163832426071167, "step": 33833 }, { "epoch": 2.667244777296019, "grad_norm": 0.4608358144760132, "learning_rate": 0.0005, "loss": 1.2746515274047852, "step": 33834 }, { "epoch": 2.667323610563658, "grad_norm": 0.4151705503463745, "learning_rate": 0.0005, "loss": 1.1894688606262207, "step": 33835 }, { "epoch": 2.6674024438312967, "grad_norm": 0.4722023606300354, "learning_rate": 0.0005, "loss": 1.2538707256317139, "step": 33836 }, { "epoch": 2.667481277098936, "grad_norm": 0.7168601155281067, "learning_rate": 0.0005, "loss": 1.312393307685852, "step": 33837 }, { "epoch": 2.6675601103665745, "grad_norm": 0.5212687253952026, "learning_rate": 0.0005, "loss": 1.342139482498169, "step": 33838 }, { "epoch": 2.6676389436342136, "grad_norm": 0.4717557430267334, "learning_rate": 0.0005, "loss": 1.3368873596191406, "step": 33839 }, { "epoch": 2.6677177769018527, "grad_norm": 0.4516185522079468, "learning_rate": 0.0005, "loss": 1.2133572101593018, "step": 33840 }, { "epoch": 2.6677966101694914, "grad_norm": 0.4813956022262573, "learning_rate": 0.0005, "loss": 1.328193187713623, "step": 33841 }, { "epoch": 2.6678754434371306, "grad_norm": 0.4993388056755066, "learning_rate": 0.0005, "loss": 1.3070156574249268, "step": 33842 }, { "epoch": 2.6679542767047693, "grad_norm": 0.5149385333061218, "learning_rate": 0.0005, "loss": 1.3466899394989014, "step": 33843 }, { "epoch": 2.6680331099724084, "grad_norm": 0.3995438516139984, "learning_rate": 0.0005, "loss": 1.1705045700073242, "step": 33844 }, { "epoch": 2.6681119432400475, "grad_norm": 0.5742319822311401, "learning_rate": 0.0005, "loss": 1.4808381795883179, "step": 33845 }, { "epoch": 2.668190776507686, "grad_norm": 0.501905083656311, "learning_rate": 0.0005, "loss": 1.3031293153762817, "step": 33846 }, { "epoch": 2.6682696097753253, "grad_norm": 0.5252368450164795, "learning_rate": 0.0005, "loss": 1.2934836149215698, "step": 33847 }, { "epoch": 2.668348443042964, "grad_norm": 0.4672827124595642, "learning_rate": 0.0005, "loss": 1.1974858045578003, "step": 33848 }, { "epoch": 2.668427276310603, "grad_norm": 0.5601332187652588, "learning_rate": 0.0005, "loss": 1.2939271926879883, "step": 33849 }, { "epoch": 2.668506109578242, "grad_norm": 0.548196017742157, "learning_rate": 0.0005, "loss": 1.1951768398284912, "step": 33850 }, { "epoch": 2.668584942845881, "grad_norm": 0.5749002695083618, "learning_rate": 0.0005, "loss": 1.395620346069336, "step": 33851 }, { "epoch": 2.6686637761135197, "grad_norm": 0.4222019612789154, "learning_rate": 0.0005, "loss": 1.2006019353866577, "step": 33852 }, { "epoch": 2.668742609381159, "grad_norm": 0.5158131718635559, "learning_rate": 0.0005, "loss": 1.362812876701355, "step": 33853 }, { "epoch": 2.668821442648798, "grad_norm": 0.48095595836639404, "learning_rate": 0.0005, "loss": 1.2794756889343262, "step": 33854 }, { "epoch": 2.6689002759164366, "grad_norm": 0.42278385162353516, "learning_rate": 0.0005, "loss": 1.2810344696044922, "step": 33855 }, { "epoch": 2.6689791091840758, "grad_norm": 0.5513980984687805, "learning_rate": 0.0005, "loss": 1.4314298629760742, "step": 33856 }, { "epoch": 2.6690579424517145, "grad_norm": 0.5809026956558228, "learning_rate": 0.0005, "loss": 1.2210203409194946, "step": 33857 }, { "epoch": 2.6691367757193536, "grad_norm": 0.9523822665214539, "learning_rate": 0.0005, "loss": 1.5345220565795898, "step": 33858 }, { "epoch": 2.6692156089869927, "grad_norm": 0.4853259027004242, "learning_rate": 0.0005, "loss": 1.4400347471237183, "step": 33859 }, { "epoch": 2.6692944422546314, "grad_norm": 5.733424186706543, "learning_rate": 0.0005, "loss": 1.4353666305541992, "step": 33860 }, { "epoch": 2.6693732755222705, "grad_norm": 0.5180862545967102, "learning_rate": 0.0005, "loss": 1.476618766784668, "step": 33861 }, { "epoch": 2.6694521087899092, "grad_norm": 0.5659561157226562, "learning_rate": 0.0005, "loss": 1.4468082189559937, "step": 33862 }, { "epoch": 2.6695309420575484, "grad_norm": 0.5105794668197632, "learning_rate": 0.0005, "loss": 1.3825397491455078, "step": 33863 }, { "epoch": 2.6696097753251875, "grad_norm": 0.5196848511695862, "learning_rate": 0.0005, "loss": 1.3533294200897217, "step": 33864 }, { "epoch": 2.669688608592826, "grad_norm": 0.5727667808532715, "learning_rate": 0.0005, "loss": 1.4051086902618408, "step": 33865 }, { "epoch": 2.669767441860465, "grad_norm": 0.5104538798332214, "learning_rate": 0.0005, "loss": 1.3243681192398071, "step": 33866 }, { "epoch": 2.669846275128104, "grad_norm": 0.7064172625541687, "learning_rate": 0.0005, "loss": 1.368894100189209, "step": 33867 }, { "epoch": 2.669925108395743, "grad_norm": 0.6068339347839355, "learning_rate": 0.0005, "loss": 1.3997619152069092, "step": 33868 }, { "epoch": 2.670003941663382, "grad_norm": 0.4861721694469452, "learning_rate": 0.0005, "loss": 1.2777612209320068, "step": 33869 }, { "epoch": 2.670082774931021, "grad_norm": 0.5325243473052979, "learning_rate": 0.0005, "loss": 1.1644980907440186, "step": 33870 }, { "epoch": 2.6701616081986597, "grad_norm": 0.5700812935829163, "learning_rate": 0.0005, "loss": 1.0496735572814941, "step": 33871 }, { "epoch": 2.670240441466299, "grad_norm": 0.5465543270111084, "learning_rate": 0.0005, "loss": 1.3915743827819824, "step": 33872 }, { "epoch": 2.670319274733938, "grad_norm": 0.5301595330238342, "learning_rate": 0.0005, "loss": 1.2811415195465088, "step": 33873 }, { "epoch": 2.6703981080015766, "grad_norm": 0.4737597107887268, "learning_rate": 0.0005, "loss": 1.3572691679000854, "step": 33874 }, { "epoch": 2.6704769412692158, "grad_norm": 0.5853643417358398, "learning_rate": 0.0005, "loss": 1.165187120437622, "step": 33875 }, { "epoch": 2.6705557745368544, "grad_norm": 0.46680089831352234, "learning_rate": 0.0005, "loss": 1.301595687866211, "step": 33876 }, { "epoch": 2.6706346078044936, "grad_norm": 0.4705699682235718, "learning_rate": 0.0005, "loss": 1.138878583908081, "step": 33877 }, { "epoch": 2.6707134410721327, "grad_norm": 0.48267191648483276, "learning_rate": 0.0005, "loss": 1.228604793548584, "step": 33878 }, { "epoch": 2.6707922743397714, "grad_norm": 0.5382208228111267, "learning_rate": 0.0005, "loss": 1.438867449760437, "step": 33879 }, { "epoch": 2.67087110760741, "grad_norm": 0.4588310420513153, "learning_rate": 0.0005, "loss": 1.3753867149353027, "step": 33880 }, { "epoch": 2.6709499408750492, "grad_norm": 0.5364039540290833, "learning_rate": 0.0005, "loss": 1.4659104347229004, "step": 33881 }, { "epoch": 2.6710287741426884, "grad_norm": 0.5131874084472656, "learning_rate": 0.0005, "loss": 1.314933180809021, "step": 33882 }, { "epoch": 2.671107607410327, "grad_norm": 0.6386160850524902, "learning_rate": 0.0005, "loss": 1.264268398284912, "step": 33883 }, { "epoch": 2.671186440677966, "grad_norm": 0.5490173101425171, "learning_rate": 0.0005, "loss": 1.4335678815841675, "step": 33884 }, { "epoch": 2.671265273945605, "grad_norm": 0.5631450414657593, "learning_rate": 0.0005, "loss": 1.4168224334716797, "step": 33885 }, { "epoch": 2.671344107213244, "grad_norm": 0.5833771824836731, "learning_rate": 0.0005, "loss": 1.3442010879516602, "step": 33886 }, { "epoch": 2.671422940480883, "grad_norm": 0.5567858815193176, "learning_rate": 0.0005, "loss": 1.2903215885162354, "step": 33887 }, { "epoch": 2.671501773748522, "grad_norm": 0.3749728202819824, "learning_rate": 0.0005, "loss": 1.1705362796783447, "step": 33888 }, { "epoch": 2.671580607016161, "grad_norm": 0.6293832063674927, "learning_rate": 0.0005, "loss": 1.5816221237182617, "step": 33889 }, { "epoch": 2.6716594402837996, "grad_norm": 0.8032751083374023, "learning_rate": 0.0005, "loss": 1.2917070388793945, "step": 33890 }, { "epoch": 2.671738273551439, "grad_norm": 0.5908827185630798, "learning_rate": 0.0005, "loss": 1.586836814880371, "step": 33891 }, { "epoch": 2.671817106819078, "grad_norm": 0.5062404274940491, "learning_rate": 0.0005, "loss": 1.3719185590744019, "step": 33892 }, { "epoch": 2.6718959400867166, "grad_norm": 0.5478272438049316, "learning_rate": 0.0005, "loss": 1.4855923652648926, "step": 33893 }, { "epoch": 2.6719747733543553, "grad_norm": 0.5167933702468872, "learning_rate": 0.0005, "loss": 1.4376423358917236, "step": 33894 }, { "epoch": 2.6720536066219944, "grad_norm": 0.45574280619621277, "learning_rate": 0.0005, "loss": 1.3128795623779297, "step": 33895 }, { "epoch": 2.6721324398896336, "grad_norm": 0.5012115836143494, "learning_rate": 0.0005, "loss": 1.4363763332366943, "step": 33896 }, { "epoch": 2.6722112731572722, "grad_norm": 0.4748725891113281, "learning_rate": 0.0005, "loss": 1.4243347644805908, "step": 33897 }, { "epoch": 2.6722901064249114, "grad_norm": 0.5687037706375122, "learning_rate": 0.0005, "loss": 1.5521485805511475, "step": 33898 }, { "epoch": 2.67236893969255, "grad_norm": 0.4194730818271637, "learning_rate": 0.0005, "loss": 1.14882493019104, "step": 33899 }, { "epoch": 2.672447772960189, "grad_norm": 0.45567643642425537, "learning_rate": 0.0005, "loss": 1.2126619815826416, "step": 33900 }, { "epoch": 2.6725266062278283, "grad_norm": 0.46516066789627075, "learning_rate": 0.0005, "loss": 1.362624168395996, "step": 33901 }, { "epoch": 2.672605439495467, "grad_norm": 0.4462604820728302, "learning_rate": 0.0005, "loss": 1.2281415462493896, "step": 33902 }, { "epoch": 2.672684272763106, "grad_norm": 0.5553525686264038, "learning_rate": 0.0005, "loss": 1.21028470993042, "step": 33903 }, { "epoch": 2.672763106030745, "grad_norm": 0.6395543813705444, "learning_rate": 0.0005, "loss": 1.5755319595336914, "step": 33904 }, { "epoch": 2.672841939298384, "grad_norm": 0.7475113272666931, "learning_rate": 0.0005, "loss": 1.5453994274139404, "step": 33905 }, { "epoch": 2.672920772566023, "grad_norm": 0.5623601078987122, "learning_rate": 0.0005, "loss": 1.3580589294433594, "step": 33906 }, { "epoch": 2.672999605833662, "grad_norm": 0.5542020797729492, "learning_rate": 0.0005, "loss": 1.0655953884124756, "step": 33907 }, { "epoch": 2.6730784391013005, "grad_norm": 0.386536568403244, "learning_rate": 0.0005, "loss": 1.2088366746902466, "step": 33908 }, { "epoch": 2.6731572723689396, "grad_norm": 0.48466306924819946, "learning_rate": 0.0005, "loss": 1.312685251235962, "step": 33909 }, { "epoch": 2.6732361056365788, "grad_norm": 0.46312764286994934, "learning_rate": 0.0005, "loss": 1.3968925476074219, "step": 33910 }, { "epoch": 2.6733149389042175, "grad_norm": 0.551841139793396, "learning_rate": 0.0005, "loss": 1.4296777248382568, "step": 33911 }, { "epoch": 2.6733937721718566, "grad_norm": 0.4785425066947937, "learning_rate": 0.0005, "loss": 1.1207733154296875, "step": 33912 }, { "epoch": 2.6734726054394953, "grad_norm": 0.4847301244735718, "learning_rate": 0.0005, "loss": 1.3955926895141602, "step": 33913 }, { "epoch": 2.6735514387071344, "grad_norm": 0.48918417096138, "learning_rate": 0.0005, "loss": 1.1687650680541992, "step": 33914 }, { "epoch": 2.6736302719747735, "grad_norm": 0.5373988747596741, "learning_rate": 0.0005, "loss": 1.3406813144683838, "step": 33915 }, { "epoch": 2.6737091052424122, "grad_norm": 0.4715743362903595, "learning_rate": 0.0005, "loss": 1.2581615447998047, "step": 33916 }, { "epoch": 2.6737879385100514, "grad_norm": 0.5251074433326721, "learning_rate": 0.0005, "loss": 1.371614933013916, "step": 33917 }, { "epoch": 2.67386677177769, "grad_norm": 0.4505022466182709, "learning_rate": 0.0005, "loss": 1.18610680103302, "step": 33918 }, { "epoch": 2.673945605045329, "grad_norm": 1.0159276723861694, "learning_rate": 0.0005, "loss": 1.3652786016464233, "step": 33919 }, { "epoch": 2.6740244383129683, "grad_norm": 0.5703728199005127, "learning_rate": 0.0005, "loss": 1.446597695350647, "step": 33920 }, { "epoch": 2.674103271580607, "grad_norm": 0.43812260031700134, "learning_rate": 0.0005, "loss": 1.2699707746505737, "step": 33921 }, { "epoch": 2.6741821048482457, "grad_norm": 0.6076892614364624, "learning_rate": 0.0005, "loss": 1.1335222721099854, "step": 33922 }, { "epoch": 2.674260938115885, "grad_norm": 0.5010131001472473, "learning_rate": 0.0005, "loss": 1.2736457586288452, "step": 33923 }, { "epoch": 2.674339771383524, "grad_norm": 0.6034531593322754, "learning_rate": 0.0005, "loss": 1.2445119619369507, "step": 33924 }, { "epoch": 2.6744186046511627, "grad_norm": 0.49044477939605713, "learning_rate": 0.0005, "loss": 1.211247205734253, "step": 33925 }, { "epoch": 2.674497437918802, "grad_norm": 0.5555495619773865, "learning_rate": 0.0005, "loss": 1.2494014501571655, "step": 33926 }, { "epoch": 2.6745762711864405, "grad_norm": 0.45413464307785034, "learning_rate": 0.0005, "loss": 1.3399016857147217, "step": 33927 }, { "epoch": 2.6746551044540796, "grad_norm": 0.6187168955802917, "learning_rate": 0.0005, "loss": 1.4368560314178467, "step": 33928 }, { "epoch": 2.6747339377217187, "grad_norm": 0.82787024974823, "learning_rate": 0.0005, "loss": 1.3994420766830444, "step": 33929 }, { "epoch": 2.6748127709893574, "grad_norm": 0.5371134877204895, "learning_rate": 0.0005, "loss": 1.3240900039672852, "step": 33930 }, { "epoch": 2.6748916042569966, "grad_norm": 0.6181508302688599, "learning_rate": 0.0005, "loss": 1.3065001964569092, "step": 33931 }, { "epoch": 2.6749704375246353, "grad_norm": 0.6040024757385254, "learning_rate": 0.0005, "loss": 1.2902215719223022, "step": 33932 }, { "epoch": 2.6750492707922744, "grad_norm": 0.5273095369338989, "learning_rate": 0.0005, "loss": 1.23165762424469, "step": 33933 }, { "epoch": 2.6751281040599135, "grad_norm": 0.4226905107498169, "learning_rate": 0.0005, "loss": 1.2252329587936401, "step": 33934 }, { "epoch": 2.675206937327552, "grad_norm": 0.46802785992622375, "learning_rate": 0.0005, "loss": 1.2805227041244507, "step": 33935 }, { "epoch": 2.675285770595191, "grad_norm": 0.5329186916351318, "learning_rate": 0.0005, "loss": 1.434942603111267, "step": 33936 }, { "epoch": 2.67536460386283, "grad_norm": 0.42380625009536743, "learning_rate": 0.0005, "loss": 1.2984890937805176, "step": 33937 }, { "epoch": 2.675443437130469, "grad_norm": 0.6089310050010681, "learning_rate": 0.0005, "loss": 1.5269742012023926, "step": 33938 }, { "epoch": 2.675522270398108, "grad_norm": 0.4672035276889801, "learning_rate": 0.0005, "loss": 1.3189048767089844, "step": 33939 }, { "epoch": 2.675601103665747, "grad_norm": 0.3846612572669983, "learning_rate": 0.0005, "loss": 1.2662057876586914, "step": 33940 }, { "epoch": 2.6756799369333857, "grad_norm": 0.480481892824173, "learning_rate": 0.0005, "loss": 1.3038139343261719, "step": 33941 }, { "epoch": 2.675758770201025, "grad_norm": 0.4330766499042511, "learning_rate": 0.0005, "loss": 1.1694836616516113, "step": 33942 }, { "epoch": 2.675837603468664, "grad_norm": 0.5210555195808411, "learning_rate": 0.0005, "loss": 1.3560515642166138, "step": 33943 }, { "epoch": 2.6759164367363026, "grad_norm": 0.4845539927482605, "learning_rate": 0.0005, "loss": 1.36167573928833, "step": 33944 }, { "epoch": 2.6759952700039418, "grad_norm": 0.6145159602165222, "learning_rate": 0.0005, "loss": 1.4767711162567139, "step": 33945 }, { "epoch": 2.6760741032715805, "grad_norm": 0.5049101710319519, "learning_rate": 0.0005, "loss": 1.5406062602996826, "step": 33946 }, { "epoch": 2.6761529365392196, "grad_norm": 0.4825979173183441, "learning_rate": 0.0005, "loss": 1.323183298110962, "step": 33947 }, { "epoch": 2.6762317698068587, "grad_norm": 0.5598328709602356, "learning_rate": 0.0005, "loss": 1.316603660583496, "step": 33948 }, { "epoch": 2.6763106030744974, "grad_norm": 0.5456136465072632, "learning_rate": 0.0005, "loss": 1.5022366046905518, "step": 33949 }, { "epoch": 2.676389436342136, "grad_norm": 0.4529401957988739, "learning_rate": 0.0005, "loss": 1.2541491985321045, "step": 33950 }, { "epoch": 2.6764682696097752, "grad_norm": 0.552245557308197, "learning_rate": 0.0005, "loss": 1.3252071142196655, "step": 33951 }, { "epoch": 2.6765471028774144, "grad_norm": 0.42600709199905396, "learning_rate": 0.0005, "loss": 1.3417456150054932, "step": 33952 }, { "epoch": 2.676625936145053, "grad_norm": 0.43786120414733887, "learning_rate": 0.0005, "loss": 1.3403066396713257, "step": 33953 }, { "epoch": 2.676704769412692, "grad_norm": 0.5255801677703857, "learning_rate": 0.0005, "loss": 1.4360640048980713, "step": 33954 }, { "epoch": 2.676783602680331, "grad_norm": 0.5247758030891418, "learning_rate": 0.0005, "loss": 1.4668772220611572, "step": 33955 }, { "epoch": 2.67686243594797, "grad_norm": 0.5665109753608704, "learning_rate": 0.0005, "loss": 1.3769917488098145, "step": 33956 }, { "epoch": 2.676941269215609, "grad_norm": 0.46971261501312256, "learning_rate": 0.0005, "loss": 1.3565893173217773, "step": 33957 }, { "epoch": 2.677020102483248, "grad_norm": 0.5413627624511719, "learning_rate": 0.0005, "loss": 1.5134693384170532, "step": 33958 }, { "epoch": 2.677098935750887, "grad_norm": 0.47475379705429077, "learning_rate": 0.0005, "loss": 1.2787964344024658, "step": 33959 }, { "epoch": 2.6771777690185257, "grad_norm": 0.45025959610939026, "learning_rate": 0.0005, "loss": 1.213184118270874, "step": 33960 }, { "epoch": 2.677256602286165, "grad_norm": 0.4858839511871338, "learning_rate": 0.0005, "loss": 1.3284151554107666, "step": 33961 }, { "epoch": 2.677335435553804, "grad_norm": 0.6371397376060486, "learning_rate": 0.0005, "loss": 1.3183249235153198, "step": 33962 }, { "epoch": 2.6774142688214426, "grad_norm": 0.5492887496948242, "learning_rate": 0.0005, "loss": 1.3124566078186035, "step": 33963 }, { "epoch": 2.6774931020890818, "grad_norm": 0.47802212834358215, "learning_rate": 0.0005, "loss": 1.2347162961959839, "step": 33964 }, { "epoch": 2.6775719353567204, "grad_norm": 0.6213148236274719, "learning_rate": 0.0005, "loss": 1.297290563583374, "step": 33965 }, { "epoch": 2.6776507686243596, "grad_norm": 0.5404993295669556, "learning_rate": 0.0005, "loss": 1.229185700416565, "step": 33966 }, { "epoch": 2.6777296018919983, "grad_norm": 0.47508516907691956, "learning_rate": 0.0005, "loss": 1.337601661682129, "step": 33967 }, { "epoch": 2.6778084351596374, "grad_norm": 0.4606059193611145, "learning_rate": 0.0005, "loss": 1.3212833404541016, "step": 33968 }, { "epoch": 2.677887268427276, "grad_norm": 0.48039183020591736, "learning_rate": 0.0005, "loss": 1.2984997034072876, "step": 33969 }, { "epoch": 2.6779661016949152, "grad_norm": 0.6862522959709167, "learning_rate": 0.0005, "loss": 1.3312569856643677, "step": 33970 }, { "epoch": 2.6780449349625544, "grad_norm": 0.5424454808235168, "learning_rate": 0.0005, "loss": 1.3248190879821777, "step": 33971 }, { "epoch": 2.678123768230193, "grad_norm": 0.5632786154747009, "learning_rate": 0.0005, "loss": 1.4575965404510498, "step": 33972 }, { "epoch": 2.678202601497832, "grad_norm": 0.5140519142150879, "learning_rate": 0.0005, "loss": 1.331932544708252, "step": 33973 }, { "epoch": 2.678281434765471, "grad_norm": 0.5089076161384583, "learning_rate": 0.0005, "loss": 1.3931267261505127, "step": 33974 }, { "epoch": 2.67836026803311, "grad_norm": 0.4836132526397705, "learning_rate": 0.0005, "loss": 1.4717001914978027, "step": 33975 }, { "epoch": 2.678439101300749, "grad_norm": 0.5692492723464966, "learning_rate": 0.0005, "loss": 1.473764181137085, "step": 33976 }, { "epoch": 2.678517934568388, "grad_norm": 0.5056750774383545, "learning_rate": 0.0005, "loss": 1.2760779857635498, "step": 33977 }, { "epoch": 2.678596767836027, "grad_norm": 0.37895822525024414, "learning_rate": 0.0005, "loss": 1.2134521007537842, "step": 33978 }, { "epoch": 2.6786756011036656, "grad_norm": 0.5795139670372009, "learning_rate": 0.0005, "loss": 1.636075735092163, "step": 33979 }, { "epoch": 2.678754434371305, "grad_norm": 0.5086313486099243, "learning_rate": 0.0005, "loss": 1.4910476207733154, "step": 33980 }, { "epoch": 2.6788332676389435, "grad_norm": 0.46843400597572327, "learning_rate": 0.0005, "loss": 1.217973232269287, "step": 33981 }, { "epoch": 2.6789121009065826, "grad_norm": 0.59428870677948, "learning_rate": 0.0005, "loss": 1.1822926998138428, "step": 33982 }, { "epoch": 2.6789909341742213, "grad_norm": 0.47003039717674255, "learning_rate": 0.0005, "loss": 1.2220299243927002, "step": 33983 }, { "epoch": 2.6790697674418604, "grad_norm": 0.7325011491775513, "learning_rate": 0.0005, "loss": 1.2218904495239258, "step": 33984 }, { "epoch": 2.6791486007094996, "grad_norm": 0.519600510597229, "learning_rate": 0.0005, "loss": 1.4921867847442627, "step": 33985 }, { "epoch": 2.6792274339771383, "grad_norm": 0.5488156080245972, "learning_rate": 0.0005, "loss": 1.316964030265808, "step": 33986 }, { "epoch": 2.6793062672447774, "grad_norm": 0.5052134990692139, "learning_rate": 0.0005, "loss": 1.2740867137908936, "step": 33987 }, { "epoch": 2.679385100512416, "grad_norm": 0.5613943338394165, "learning_rate": 0.0005, "loss": 1.1951031684875488, "step": 33988 }, { "epoch": 2.679463933780055, "grad_norm": 0.5193274021148682, "learning_rate": 0.0005, "loss": 1.4209188222885132, "step": 33989 }, { "epoch": 2.6795427670476943, "grad_norm": 0.7555820345878601, "learning_rate": 0.0005, "loss": 1.363755464553833, "step": 33990 }, { "epoch": 2.679621600315333, "grad_norm": 0.5258013606071472, "learning_rate": 0.0005, "loss": 1.3626832962036133, "step": 33991 }, { "epoch": 2.679700433582972, "grad_norm": 0.5386700630187988, "learning_rate": 0.0005, "loss": 1.5354671478271484, "step": 33992 }, { "epoch": 2.679779266850611, "grad_norm": 0.628376841545105, "learning_rate": 0.0005, "loss": 1.543999195098877, "step": 33993 }, { "epoch": 2.67985810011825, "grad_norm": 0.463573157787323, "learning_rate": 0.0005, "loss": 1.2858139276504517, "step": 33994 }, { "epoch": 2.679936933385889, "grad_norm": 0.42785587906837463, "learning_rate": 0.0005, "loss": 1.2615816593170166, "step": 33995 }, { "epoch": 2.680015766653528, "grad_norm": 0.47738680243492126, "learning_rate": 0.0005, "loss": 1.4573712348937988, "step": 33996 }, { "epoch": 2.6800945999211665, "grad_norm": 0.6962947845458984, "learning_rate": 0.0005, "loss": 1.2430082559585571, "step": 33997 }, { "epoch": 2.6801734331888056, "grad_norm": 0.5620681643486023, "learning_rate": 0.0005, "loss": 1.3721916675567627, "step": 33998 }, { "epoch": 2.6802522664564448, "grad_norm": 0.5474938750267029, "learning_rate": 0.0005, "loss": 1.2972595691680908, "step": 33999 }, { "epoch": 2.6803310997240835, "grad_norm": 0.513755738735199, "learning_rate": 0.0005, "loss": 1.37492036819458, "step": 34000 }, { "epoch": 2.6804099329917226, "grad_norm": 0.5443038940429688, "learning_rate": 0.0005, "loss": 1.4183430671691895, "step": 34001 }, { "epoch": 2.6804887662593613, "grad_norm": 0.49301809072494507, "learning_rate": 0.0005, "loss": 1.0509493350982666, "step": 34002 }, { "epoch": 2.6805675995270004, "grad_norm": 0.5580002069473267, "learning_rate": 0.0005, "loss": 1.2611706256866455, "step": 34003 }, { "epoch": 2.6806464327946395, "grad_norm": 0.5224257111549377, "learning_rate": 0.0005, "loss": 1.3816808462142944, "step": 34004 }, { "epoch": 2.6807252660622782, "grad_norm": 0.5235961675643921, "learning_rate": 0.0005, "loss": 1.4407782554626465, "step": 34005 }, { "epoch": 2.6808040993299174, "grad_norm": 0.5164549946784973, "learning_rate": 0.0005, "loss": 1.2256393432617188, "step": 34006 }, { "epoch": 2.680882932597556, "grad_norm": 0.6517978310585022, "learning_rate": 0.0005, "loss": 1.339606523513794, "step": 34007 }, { "epoch": 2.680961765865195, "grad_norm": 0.604289174079895, "learning_rate": 0.0005, "loss": 1.2738409042358398, "step": 34008 }, { "epoch": 2.6810405991328343, "grad_norm": 0.5548025369644165, "learning_rate": 0.0005, "loss": 1.2664203643798828, "step": 34009 }, { "epoch": 2.681119432400473, "grad_norm": 0.48421627283096313, "learning_rate": 0.0005, "loss": 1.3909764289855957, "step": 34010 }, { "epoch": 2.6811982656681117, "grad_norm": 0.42602765560150146, "learning_rate": 0.0005, "loss": 1.2108213901519775, "step": 34011 }, { "epoch": 2.681277098935751, "grad_norm": 0.48735588788986206, "learning_rate": 0.0005, "loss": 1.2437069416046143, "step": 34012 }, { "epoch": 2.68135593220339, "grad_norm": 0.4727727174758911, "learning_rate": 0.0005, "loss": 1.228255271911621, "step": 34013 }, { "epoch": 2.6814347654710287, "grad_norm": 0.48477283120155334, "learning_rate": 0.0005, "loss": 1.292226791381836, "step": 34014 }, { "epoch": 2.681513598738668, "grad_norm": 0.48807135224342346, "learning_rate": 0.0005, "loss": 1.2686281204223633, "step": 34015 }, { "epoch": 2.6815924320063065, "grad_norm": 0.5334140062332153, "learning_rate": 0.0005, "loss": 1.4674656391143799, "step": 34016 }, { "epoch": 2.6816712652739456, "grad_norm": 0.5173021554946899, "learning_rate": 0.0005, "loss": 1.4385885000228882, "step": 34017 }, { "epoch": 2.6817500985415847, "grad_norm": 0.47132474184036255, "learning_rate": 0.0005, "loss": 1.4737961292266846, "step": 34018 }, { "epoch": 2.6818289318092234, "grad_norm": 0.41258594393730164, "learning_rate": 0.0005, "loss": 1.0701854228973389, "step": 34019 }, { "epoch": 2.6819077650768626, "grad_norm": 0.36558952927589417, "learning_rate": 0.0005, "loss": 1.1331939697265625, "step": 34020 }, { "epoch": 2.6819865983445013, "grad_norm": 0.4680032432079315, "learning_rate": 0.0005, "loss": 1.2217581272125244, "step": 34021 }, { "epoch": 2.6820654316121404, "grad_norm": 0.41532102227211, "learning_rate": 0.0005, "loss": 1.1297175884246826, "step": 34022 }, { "epoch": 2.6821442648797795, "grad_norm": 0.6393837332725525, "learning_rate": 0.0005, "loss": 1.5711930990219116, "step": 34023 }, { "epoch": 2.682223098147418, "grad_norm": 0.4702723026275635, "learning_rate": 0.0005, "loss": 1.4087090492248535, "step": 34024 }, { "epoch": 2.682301931415057, "grad_norm": 0.4999738037586212, "learning_rate": 0.0005, "loss": 1.3974717855453491, "step": 34025 }, { "epoch": 2.682380764682696, "grad_norm": 0.5322011709213257, "learning_rate": 0.0005, "loss": 1.4059444665908813, "step": 34026 }, { "epoch": 2.682459597950335, "grad_norm": 0.577922523021698, "learning_rate": 0.0005, "loss": 1.2725099325180054, "step": 34027 }, { "epoch": 2.682538431217974, "grad_norm": 0.6910742521286011, "learning_rate": 0.0005, "loss": 1.3603602647781372, "step": 34028 }, { "epoch": 2.682617264485613, "grad_norm": 0.45823150873184204, "learning_rate": 0.0005, "loss": 1.1228678226470947, "step": 34029 }, { "epoch": 2.6826960977532517, "grad_norm": 0.49493515491485596, "learning_rate": 0.0005, "loss": 1.344543695449829, "step": 34030 }, { "epoch": 2.682774931020891, "grad_norm": 0.422702819108963, "learning_rate": 0.0005, "loss": 1.2001030445098877, "step": 34031 }, { "epoch": 2.68285376428853, "grad_norm": 0.5670332908630371, "learning_rate": 0.0005, "loss": 1.2407336235046387, "step": 34032 }, { "epoch": 2.6829325975561686, "grad_norm": 0.5340278148651123, "learning_rate": 0.0005, "loss": 1.2713305950164795, "step": 34033 }, { "epoch": 2.6830114308238078, "grad_norm": 0.5713286399841309, "learning_rate": 0.0005, "loss": 1.4641026258468628, "step": 34034 }, { "epoch": 2.6830902640914465, "grad_norm": 0.43654119968414307, "learning_rate": 0.0005, "loss": 1.2749950885772705, "step": 34035 }, { "epoch": 2.6831690973590856, "grad_norm": 0.6169542670249939, "learning_rate": 0.0005, "loss": 1.4550807476043701, "step": 34036 }, { "epoch": 2.6832479306267247, "grad_norm": 0.47455108165740967, "learning_rate": 0.0005, "loss": 1.4879064559936523, "step": 34037 }, { "epoch": 2.6833267638943634, "grad_norm": 0.5850338935852051, "learning_rate": 0.0005, "loss": 1.4028279781341553, "step": 34038 }, { "epoch": 2.683405597162002, "grad_norm": 0.7835137248039246, "learning_rate": 0.0005, "loss": 1.3245153427124023, "step": 34039 }, { "epoch": 2.6834844304296412, "grad_norm": 0.4130430221557617, "learning_rate": 0.0005, "loss": 1.2050780057907104, "step": 34040 }, { "epoch": 2.6835632636972804, "grad_norm": 0.47876405715942383, "learning_rate": 0.0005, "loss": 1.2411532402038574, "step": 34041 }, { "epoch": 2.683642096964919, "grad_norm": 0.5594386458396912, "learning_rate": 0.0005, "loss": 1.2572565078735352, "step": 34042 }, { "epoch": 2.683720930232558, "grad_norm": 0.5469566583633423, "learning_rate": 0.0005, "loss": 1.3144292831420898, "step": 34043 }, { "epoch": 2.683799763500197, "grad_norm": 0.5697798132896423, "learning_rate": 0.0005, "loss": 1.4155257940292358, "step": 34044 }, { "epoch": 2.683878596767836, "grad_norm": 0.45714202523231506, "learning_rate": 0.0005, "loss": 1.2877651453018188, "step": 34045 }, { "epoch": 2.683957430035475, "grad_norm": 0.47960910201072693, "learning_rate": 0.0005, "loss": 1.342479944229126, "step": 34046 }, { "epoch": 2.684036263303114, "grad_norm": 0.37978437542915344, "learning_rate": 0.0005, "loss": 1.2233517169952393, "step": 34047 }, { "epoch": 2.684115096570753, "grad_norm": 0.49223020672798157, "learning_rate": 0.0005, "loss": 1.2403323650360107, "step": 34048 }, { "epoch": 2.6841939298383917, "grad_norm": 0.3605341613292694, "learning_rate": 0.0005, "loss": 1.0615112781524658, "step": 34049 }, { "epoch": 2.684272763106031, "grad_norm": 0.4978126287460327, "learning_rate": 0.0005, "loss": 1.3132336139678955, "step": 34050 }, { "epoch": 2.68435159637367, "grad_norm": 0.6274998188018799, "learning_rate": 0.0005, "loss": 1.4793299436569214, "step": 34051 }, { "epoch": 2.6844304296413086, "grad_norm": 0.5956411957740784, "learning_rate": 0.0005, "loss": 1.4929088354110718, "step": 34052 }, { "epoch": 2.6845092629089473, "grad_norm": 0.5172160863876343, "learning_rate": 0.0005, "loss": 1.438848853111267, "step": 34053 }, { "epoch": 2.6845880961765864, "grad_norm": 0.4572182297706604, "learning_rate": 0.0005, "loss": 1.0915372371673584, "step": 34054 }, { "epoch": 2.6846669294442256, "grad_norm": 0.421068012714386, "learning_rate": 0.0005, "loss": 1.0428494215011597, "step": 34055 }, { "epoch": 2.6847457627118643, "grad_norm": 0.556450605392456, "learning_rate": 0.0005, "loss": 1.4963340759277344, "step": 34056 }, { "epoch": 2.6848245959795034, "grad_norm": 0.5468692183494568, "learning_rate": 0.0005, "loss": 1.2531659603118896, "step": 34057 }, { "epoch": 2.684903429247142, "grad_norm": 0.5096701979637146, "learning_rate": 0.0005, "loss": 1.2882254123687744, "step": 34058 }, { "epoch": 2.6849822625147812, "grad_norm": 0.5115979909896851, "learning_rate": 0.0005, "loss": 1.3582266569137573, "step": 34059 }, { "epoch": 2.6850610957824204, "grad_norm": 0.51907879114151, "learning_rate": 0.0005, "loss": 1.070847988128662, "step": 34060 }, { "epoch": 2.685139929050059, "grad_norm": 0.45394858717918396, "learning_rate": 0.0005, "loss": 1.3770849704742432, "step": 34061 }, { "epoch": 2.685218762317698, "grad_norm": 0.6043967008590698, "learning_rate": 0.0005, "loss": 1.466567039489746, "step": 34062 }, { "epoch": 2.685297595585337, "grad_norm": 0.5101833939552307, "learning_rate": 0.0005, "loss": 1.5174798965454102, "step": 34063 }, { "epoch": 2.685376428852976, "grad_norm": 0.5972816944122314, "learning_rate": 0.0005, "loss": 1.576728105545044, "step": 34064 }, { "epoch": 2.685455262120615, "grad_norm": 0.5579954981803894, "learning_rate": 0.0005, "loss": 1.314244031906128, "step": 34065 }, { "epoch": 2.685534095388254, "grad_norm": 0.4655309319496155, "learning_rate": 0.0005, "loss": 1.374776840209961, "step": 34066 }, { "epoch": 2.6856129286558925, "grad_norm": 0.6483479142189026, "learning_rate": 0.0005, "loss": 1.3421831130981445, "step": 34067 }, { "epoch": 2.6856917619235317, "grad_norm": 0.48787251114845276, "learning_rate": 0.0005, "loss": 1.263231873512268, "step": 34068 }, { "epoch": 2.685770595191171, "grad_norm": 0.48377755284309387, "learning_rate": 0.0005, "loss": 1.0941206216812134, "step": 34069 }, { "epoch": 2.6858494284588095, "grad_norm": 0.6091570854187012, "learning_rate": 0.0005, "loss": 1.5220298767089844, "step": 34070 }, { "epoch": 2.6859282617264486, "grad_norm": 0.6883618235588074, "learning_rate": 0.0005, "loss": 1.342885971069336, "step": 34071 }, { "epoch": 2.6860070949940873, "grad_norm": 0.6043362021446228, "learning_rate": 0.0005, "loss": 1.393620252609253, "step": 34072 }, { "epoch": 2.6860859282617264, "grad_norm": 0.4236912727355957, "learning_rate": 0.0005, "loss": 1.244715690612793, "step": 34073 }, { "epoch": 2.6861647615293656, "grad_norm": 0.45168787240982056, "learning_rate": 0.0005, "loss": 1.3051567077636719, "step": 34074 }, { "epoch": 2.6862435947970043, "grad_norm": 0.5605636239051819, "learning_rate": 0.0005, "loss": 1.3622010946273804, "step": 34075 }, { "epoch": 2.6863224280646434, "grad_norm": 0.5107719898223877, "learning_rate": 0.0005, "loss": 1.3418159484863281, "step": 34076 }, { "epoch": 2.686401261332282, "grad_norm": 0.44036540389060974, "learning_rate": 0.0005, "loss": 1.2414389848709106, "step": 34077 }, { "epoch": 2.686480094599921, "grad_norm": 0.42660075426101685, "learning_rate": 0.0005, "loss": 1.1092808246612549, "step": 34078 }, { "epoch": 2.6865589278675603, "grad_norm": 0.46609801054000854, "learning_rate": 0.0005, "loss": 1.3412666320800781, "step": 34079 }, { "epoch": 2.686637761135199, "grad_norm": 0.5139880776405334, "learning_rate": 0.0005, "loss": 1.2435848712921143, "step": 34080 }, { "epoch": 2.6867165944028377, "grad_norm": 0.4114653766155243, "learning_rate": 0.0005, "loss": 1.1799741983413696, "step": 34081 }, { "epoch": 2.686795427670477, "grad_norm": 0.6431373357772827, "learning_rate": 0.0005, "loss": 1.4906420707702637, "step": 34082 }, { "epoch": 2.686874260938116, "grad_norm": 0.44702479243278503, "learning_rate": 0.0005, "loss": 1.1272194385528564, "step": 34083 }, { "epoch": 2.6869530942057547, "grad_norm": 0.6626941561698914, "learning_rate": 0.0005, "loss": 1.4477155208587646, "step": 34084 }, { "epoch": 2.687031927473394, "grad_norm": 0.5024009346961975, "learning_rate": 0.0005, "loss": 1.1599925756454468, "step": 34085 }, { "epoch": 2.6871107607410325, "grad_norm": 0.5909313559532166, "learning_rate": 0.0005, "loss": 1.3370115756988525, "step": 34086 }, { "epoch": 2.6871895940086716, "grad_norm": 0.512932300567627, "learning_rate": 0.0005, "loss": 1.4183088541030884, "step": 34087 }, { "epoch": 2.6872684272763108, "grad_norm": 0.4647538661956787, "learning_rate": 0.0005, "loss": 1.302398443222046, "step": 34088 }, { "epoch": 2.6873472605439495, "grad_norm": 0.4024282991886139, "learning_rate": 0.0005, "loss": 1.1637481451034546, "step": 34089 }, { "epoch": 2.6874260938115886, "grad_norm": 0.447689950466156, "learning_rate": 0.0005, "loss": 1.181105136871338, "step": 34090 }, { "epoch": 2.6875049270792273, "grad_norm": 0.6015523672103882, "learning_rate": 0.0005, "loss": 1.447562575340271, "step": 34091 }, { "epoch": 2.6875837603468664, "grad_norm": 0.5062685012817383, "learning_rate": 0.0005, "loss": 1.3428304195404053, "step": 34092 }, { "epoch": 2.6876625936145055, "grad_norm": 0.5389657616615295, "learning_rate": 0.0005, "loss": 1.2773375511169434, "step": 34093 }, { "epoch": 2.6877414268821442, "grad_norm": 0.6497871279716492, "learning_rate": 0.0005, "loss": 1.5328234434127808, "step": 34094 }, { "epoch": 2.6878202601497834, "grad_norm": 0.5581339001655579, "learning_rate": 0.0005, "loss": 1.3089308738708496, "step": 34095 }, { "epoch": 2.687899093417422, "grad_norm": 0.44673511385917664, "learning_rate": 0.0005, "loss": 1.2924199104309082, "step": 34096 }, { "epoch": 2.687977926685061, "grad_norm": 0.6095879077911377, "learning_rate": 0.0005, "loss": 1.2495731115341187, "step": 34097 }, { "epoch": 2.6880567599527, "grad_norm": 0.5425824522972107, "learning_rate": 0.0005, "loss": 1.2538968324661255, "step": 34098 }, { "epoch": 2.688135593220339, "grad_norm": 0.5014564990997314, "learning_rate": 0.0005, "loss": 1.384544849395752, "step": 34099 }, { "epoch": 2.6882144264879777, "grad_norm": 0.9678954482078552, "learning_rate": 0.0005, "loss": 1.1660221815109253, "step": 34100 }, { "epoch": 2.688293259755617, "grad_norm": 0.45246371626853943, "learning_rate": 0.0005, "loss": 1.159041404724121, "step": 34101 }, { "epoch": 2.688372093023256, "grad_norm": 0.5779354572296143, "learning_rate": 0.0005, "loss": 1.2970956563949585, "step": 34102 }, { "epoch": 2.6884509262908947, "grad_norm": 0.47021040320396423, "learning_rate": 0.0005, "loss": 1.2813735008239746, "step": 34103 }, { "epoch": 2.688529759558534, "grad_norm": 0.49183452129364014, "learning_rate": 0.0005, "loss": 1.1768544912338257, "step": 34104 }, { "epoch": 2.6886085928261725, "grad_norm": 0.600796639919281, "learning_rate": 0.0005, "loss": 1.4584895372390747, "step": 34105 }, { "epoch": 2.6886874260938116, "grad_norm": 0.4780533015727997, "learning_rate": 0.0005, "loss": 1.3692487478256226, "step": 34106 }, { "epoch": 2.6887662593614508, "grad_norm": 0.4793086349964142, "learning_rate": 0.0005, "loss": 1.3483116626739502, "step": 34107 }, { "epoch": 2.6888450926290894, "grad_norm": 0.4906138777732849, "learning_rate": 0.0005, "loss": 1.2397397756576538, "step": 34108 }, { "epoch": 2.6889239258967286, "grad_norm": 0.5975778102874756, "learning_rate": 0.0005, "loss": 1.4567447900772095, "step": 34109 }, { "epoch": 2.6890027591643673, "grad_norm": 0.4887239933013916, "learning_rate": 0.0005, "loss": 1.4215493202209473, "step": 34110 }, { "epoch": 2.6890815924320064, "grad_norm": 1.0303148031234741, "learning_rate": 0.0005, "loss": 1.2565829753875732, "step": 34111 }, { "epoch": 2.689160425699645, "grad_norm": 0.6328168511390686, "learning_rate": 0.0005, "loss": 1.3414733409881592, "step": 34112 }, { "epoch": 2.689239258967284, "grad_norm": 0.5040991306304932, "learning_rate": 0.0005, "loss": 1.2889206409454346, "step": 34113 }, { "epoch": 2.689318092234923, "grad_norm": 0.538214921951294, "learning_rate": 0.0005, "loss": 1.370912790298462, "step": 34114 }, { "epoch": 2.689396925502562, "grad_norm": 0.45533403754234314, "learning_rate": 0.0005, "loss": 1.4308016300201416, "step": 34115 }, { "epoch": 2.689475758770201, "grad_norm": 0.7183805704116821, "learning_rate": 0.0005, "loss": 1.1858887672424316, "step": 34116 }, { "epoch": 2.68955459203784, "grad_norm": 0.6083846688270569, "learning_rate": 0.0005, "loss": 1.4601978063583374, "step": 34117 }, { "epoch": 2.689633425305479, "grad_norm": 0.5641534924507141, "learning_rate": 0.0005, "loss": 1.288865089416504, "step": 34118 }, { "epoch": 2.6897122585731177, "grad_norm": 0.5633936524391174, "learning_rate": 0.0005, "loss": 1.4224944114685059, "step": 34119 }, { "epoch": 2.689791091840757, "grad_norm": 0.5409612655639648, "learning_rate": 0.0005, "loss": 1.5541942119598389, "step": 34120 }, { "epoch": 2.689869925108396, "grad_norm": 0.5491654276847839, "learning_rate": 0.0005, "loss": 1.4206105470657349, "step": 34121 }, { "epoch": 2.6899487583760346, "grad_norm": 0.4480315148830414, "learning_rate": 0.0005, "loss": 1.2742831707000732, "step": 34122 }, { "epoch": 2.6900275916436738, "grad_norm": 0.5925441384315491, "learning_rate": 0.0005, "loss": 1.5490113496780396, "step": 34123 }, { "epoch": 2.6901064249113125, "grad_norm": 0.5959253907203674, "learning_rate": 0.0005, "loss": 1.2979958057403564, "step": 34124 }, { "epoch": 2.6901852581789516, "grad_norm": 0.5703783631324768, "learning_rate": 0.0005, "loss": 1.361464500427246, "step": 34125 }, { "epoch": 2.6902640914465907, "grad_norm": 0.6369980573654175, "learning_rate": 0.0005, "loss": 1.3405637741088867, "step": 34126 }, { "epoch": 2.6903429247142294, "grad_norm": 0.5217834711074829, "learning_rate": 0.0005, "loss": 1.4066064357757568, "step": 34127 }, { "epoch": 2.690421757981868, "grad_norm": 0.5050923228263855, "learning_rate": 0.0005, "loss": 1.34269118309021, "step": 34128 }, { "epoch": 2.6905005912495072, "grad_norm": 0.6320822238922119, "learning_rate": 0.0005, "loss": 1.4152076244354248, "step": 34129 }, { "epoch": 2.6905794245171464, "grad_norm": 0.5986621379852295, "learning_rate": 0.0005, "loss": 1.202661395072937, "step": 34130 }, { "epoch": 2.690658257784785, "grad_norm": 0.5500098466873169, "learning_rate": 0.0005, "loss": 1.2700475454330444, "step": 34131 }, { "epoch": 2.690737091052424, "grad_norm": 0.4529517889022827, "learning_rate": 0.0005, "loss": 1.3008406162261963, "step": 34132 }, { "epoch": 2.690815924320063, "grad_norm": 0.6588666439056396, "learning_rate": 0.0005, "loss": 1.3703283071517944, "step": 34133 }, { "epoch": 2.690894757587702, "grad_norm": 0.4958450198173523, "learning_rate": 0.0005, "loss": 1.2574303150177002, "step": 34134 }, { "epoch": 2.690973590855341, "grad_norm": 0.5391598343849182, "learning_rate": 0.0005, "loss": 1.4014856815338135, "step": 34135 }, { "epoch": 2.69105242412298, "grad_norm": 0.5214970707893372, "learning_rate": 0.0005, "loss": 1.426864504814148, "step": 34136 }, { "epoch": 2.691131257390619, "grad_norm": 0.5893440246582031, "learning_rate": 0.0005, "loss": 1.496498942375183, "step": 34137 }, { "epoch": 2.6912100906582577, "grad_norm": 0.735465407371521, "learning_rate": 0.0005, "loss": 1.2655949592590332, "step": 34138 }, { "epoch": 2.691288923925897, "grad_norm": 0.4898640513420105, "learning_rate": 0.0005, "loss": 1.3445650339126587, "step": 34139 }, { "epoch": 2.691367757193536, "grad_norm": 0.48438766598701477, "learning_rate": 0.0005, "loss": 1.4202876091003418, "step": 34140 }, { "epoch": 2.6914465904611746, "grad_norm": 0.5674211978912354, "learning_rate": 0.0005, "loss": 1.3003478050231934, "step": 34141 }, { "epoch": 2.6915254237288133, "grad_norm": 0.824489951133728, "learning_rate": 0.0005, "loss": 1.5260001420974731, "step": 34142 }, { "epoch": 2.6916042569964524, "grad_norm": 0.6941741108894348, "learning_rate": 0.0005, "loss": 1.4139833450317383, "step": 34143 }, { "epoch": 2.6916830902640916, "grad_norm": 0.538652777671814, "learning_rate": 0.0005, "loss": 1.1348135471343994, "step": 34144 }, { "epoch": 2.6917619235317303, "grad_norm": 0.6943999528884888, "learning_rate": 0.0005, "loss": 1.4416420459747314, "step": 34145 }, { "epoch": 2.6918407567993694, "grad_norm": 0.48366639018058777, "learning_rate": 0.0005, "loss": 1.2636525630950928, "step": 34146 }, { "epoch": 2.691919590067008, "grad_norm": 0.5231360793113708, "learning_rate": 0.0005, "loss": 1.367215871810913, "step": 34147 }, { "epoch": 2.6919984233346472, "grad_norm": 0.5981862545013428, "learning_rate": 0.0005, "loss": 1.3225820064544678, "step": 34148 }, { "epoch": 2.6920772566022864, "grad_norm": 0.6652439832687378, "learning_rate": 0.0005, "loss": 1.3910695314407349, "step": 34149 }, { "epoch": 2.692156089869925, "grad_norm": 0.563065767288208, "learning_rate": 0.0005, "loss": 1.3467047214508057, "step": 34150 }, { "epoch": 2.692234923137564, "grad_norm": 0.49443483352661133, "learning_rate": 0.0005, "loss": 1.3695399761199951, "step": 34151 }, { "epoch": 2.692313756405203, "grad_norm": 0.5411814451217651, "learning_rate": 0.0005, "loss": 1.4385817050933838, "step": 34152 }, { "epoch": 2.692392589672842, "grad_norm": 0.7904382348060608, "learning_rate": 0.0005, "loss": 1.2765753269195557, "step": 34153 }, { "epoch": 2.692471422940481, "grad_norm": 0.5398232340812683, "learning_rate": 0.0005, "loss": 1.3378876447677612, "step": 34154 }, { "epoch": 2.69255025620812, "grad_norm": 0.4992283284664154, "learning_rate": 0.0005, "loss": 1.2655725479125977, "step": 34155 }, { "epoch": 2.6926290894757585, "grad_norm": 0.47876229882240295, "learning_rate": 0.0005, "loss": 1.2490586042404175, "step": 34156 }, { "epoch": 2.6927079227433977, "grad_norm": 0.4539179801940918, "learning_rate": 0.0005, "loss": 1.2700448036193848, "step": 34157 }, { "epoch": 2.692786756011037, "grad_norm": 0.3856622874736786, "learning_rate": 0.0005, "loss": 1.214568853378296, "step": 34158 }, { "epoch": 2.6928655892786755, "grad_norm": 0.520025372505188, "learning_rate": 0.0005, "loss": 1.3454784154891968, "step": 34159 }, { "epoch": 2.6929444225463146, "grad_norm": 0.556052029132843, "learning_rate": 0.0005, "loss": 1.3757139444351196, "step": 34160 }, { "epoch": 2.6930232558139533, "grad_norm": 0.5780694484710693, "learning_rate": 0.0005, "loss": 1.2942713499069214, "step": 34161 }, { "epoch": 2.6931020890815924, "grad_norm": 0.41720589995384216, "learning_rate": 0.0005, "loss": 1.3164644241333008, "step": 34162 }, { "epoch": 2.6931809223492316, "grad_norm": 0.594257116317749, "learning_rate": 0.0005, "loss": 1.3828728199005127, "step": 34163 }, { "epoch": 2.6932597556168703, "grad_norm": 0.5735880136489868, "learning_rate": 0.0005, "loss": 1.3034695386886597, "step": 34164 }, { "epoch": 2.6933385888845094, "grad_norm": 0.4395493268966675, "learning_rate": 0.0005, "loss": 1.3438339233398438, "step": 34165 }, { "epoch": 2.693417422152148, "grad_norm": 0.5056896805763245, "learning_rate": 0.0005, "loss": 1.2693294286727905, "step": 34166 }, { "epoch": 2.693496255419787, "grad_norm": 0.5521557927131653, "learning_rate": 0.0005, "loss": 1.5134716033935547, "step": 34167 }, { "epoch": 2.6935750886874263, "grad_norm": 0.3777920603752136, "learning_rate": 0.0005, "loss": 1.184358835220337, "step": 34168 }, { "epoch": 2.693653921955065, "grad_norm": 0.40499019622802734, "learning_rate": 0.0005, "loss": 1.267059326171875, "step": 34169 }, { "epoch": 2.6937327552227037, "grad_norm": 0.5253272652626038, "learning_rate": 0.0005, "loss": 1.4606235027313232, "step": 34170 }, { "epoch": 2.693811588490343, "grad_norm": 0.5099503397941589, "learning_rate": 0.0005, "loss": 1.2812631130218506, "step": 34171 }, { "epoch": 2.693890421757982, "grad_norm": 0.44193872809410095, "learning_rate": 0.0005, "loss": 1.285967230796814, "step": 34172 }, { "epoch": 2.6939692550256207, "grad_norm": 0.4751059412956238, "learning_rate": 0.0005, "loss": 1.4006199836730957, "step": 34173 }, { "epoch": 2.69404808829326, "grad_norm": 0.48842018842697144, "learning_rate": 0.0005, "loss": 1.4010769128799438, "step": 34174 }, { "epoch": 2.6941269215608985, "grad_norm": 0.4634193778038025, "learning_rate": 0.0005, "loss": 1.349496603012085, "step": 34175 }, { "epoch": 2.6942057548285376, "grad_norm": 0.45856165885925293, "learning_rate": 0.0005, "loss": 1.3083701133728027, "step": 34176 }, { "epoch": 2.6942845880961768, "grad_norm": 0.5501959323883057, "learning_rate": 0.0005, "loss": 1.365375280380249, "step": 34177 }, { "epoch": 2.6943634213638155, "grad_norm": 0.563369631767273, "learning_rate": 0.0005, "loss": 1.2586308717727661, "step": 34178 }, { "epoch": 2.6944422546314546, "grad_norm": 0.47647371888160706, "learning_rate": 0.0005, "loss": 1.2021965980529785, "step": 34179 }, { "epoch": 2.6945210878990933, "grad_norm": 0.44382303953170776, "learning_rate": 0.0005, "loss": 1.1553806066513062, "step": 34180 }, { "epoch": 2.6945999211667324, "grad_norm": 0.4935358166694641, "learning_rate": 0.0005, "loss": 1.2097750902175903, "step": 34181 }, { "epoch": 2.6946787544343715, "grad_norm": 0.6846004128456116, "learning_rate": 0.0005, "loss": 1.43617844581604, "step": 34182 }, { "epoch": 2.6947575877020102, "grad_norm": 0.44961991906166077, "learning_rate": 0.0005, "loss": 1.1888285875320435, "step": 34183 }, { "epoch": 2.694836420969649, "grad_norm": 0.40807202458381653, "learning_rate": 0.0005, "loss": 1.2607086896896362, "step": 34184 }, { "epoch": 2.694915254237288, "grad_norm": 0.4311344623565674, "learning_rate": 0.0005, "loss": 1.3370203971862793, "step": 34185 }, { "epoch": 2.694994087504927, "grad_norm": 0.3923790156841278, "learning_rate": 0.0005, "loss": 1.3029837608337402, "step": 34186 }, { "epoch": 2.695072920772566, "grad_norm": 0.525323748588562, "learning_rate": 0.0005, "loss": 1.3810725212097168, "step": 34187 }, { "epoch": 2.695151754040205, "grad_norm": 0.5973218083381653, "learning_rate": 0.0005, "loss": 1.5049657821655273, "step": 34188 }, { "epoch": 2.6952305873078437, "grad_norm": 0.43021267652511597, "learning_rate": 0.0005, "loss": 1.2063394784927368, "step": 34189 }, { "epoch": 2.695309420575483, "grad_norm": 0.44739148020744324, "learning_rate": 0.0005, "loss": 1.1988258361816406, "step": 34190 }, { "epoch": 2.695388253843122, "grad_norm": 0.43312346935272217, "learning_rate": 0.0005, "loss": 1.182896614074707, "step": 34191 }, { "epoch": 2.6954670871107607, "grad_norm": 0.415640652179718, "learning_rate": 0.0005, "loss": 1.4924718141555786, "step": 34192 }, { "epoch": 2.6955459203784, "grad_norm": 0.5325759649276733, "learning_rate": 0.0005, "loss": 1.3717200756072998, "step": 34193 }, { "epoch": 2.6956247536460385, "grad_norm": 0.3988426625728607, "learning_rate": 0.0005, "loss": 1.3520548343658447, "step": 34194 }, { "epoch": 2.6957035869136776, "grad_norm": 0.6052148342132568, "learning_rate": 0.0005, "loss": 1.3428728580474854, "step": 34195 }, { "epoch": 2.6957824201813168, "grad_norm": 0.5228135585784912, "learning_rate": 0.0005, "loss": 1.3014438152313232, "step": 34196 }, { "epoch": 2.6958612534489554, "grad_norm": 0.5115324258804321, "learning_rate": 0.0005, "loss": 1.3273617029190063, "step": 34197 }, { "epoch": 2.695940086716594, "grad_norm": 0.5199338793754578, "learning_rate": 0.0005, "loss": 1.3472161293029785, "step": 34198 }, { "epoch": 2.6960189199842333, "grad_norm": 0.4413760304450989, "learning_rate": 0.0005, "loss": 1.3635170459747314, "step": 34199 }, { "epoch": 2.6960977532518724, "grad_norm": 0.5718503594398499, "learning_rate": 0.0005, "loss": 1.2889283895492554, "step": 34200 }, { "epoch": 2.696176586519511, "grad_norm": 0.48457464575767517, "learning_rate": 0.0005, "loss": 1.4600613117218018, "step": 34201 }, { "epoch": 2.69625541978715, "grad_norm": 0.5961516499519348, "learning_rate": 0.0005, "loss": 1.3099493980407715, "step": 34202 }, { "epoch": 2.696334253054789, "grad_norm": 0.5001735687255859, "learning_rate": 0.0005, "loss": 1.579942226409912, "step": 34203 }, { "epoch": 2.696413086322428, "grad_norm": 0.44804948568344116, "learning_rate": 0.0005, "loss": 1.361612319946289, "step": 34204 }, { "epoch": 2.696491919590067, "grad_norm": 0.4747711420059204, "learning_rate": 0.0005, "loss": 1.4246799945831299, "step": 34205 }, { "epoch": 2.696570752857706, "grad_norm": 0.4618068039417267, "learning_rate": 0.0005, "loss": 1.2534260749816895, "step": 34206 }, { "epoch": 2.696649586125345, "grad_norm": 0.5343491435050964, "learning_rate": 0.0005, "loss": 1.2998954057693481, "step": 34207 }, { "epoch": 2.6967284193929837, "grad_norm": 0.44374606013298035, "learning_rate": 0.0005, "loss": 1.4616990089416504, "step": 34208 }, { "epoch": 2.696807252660623, "grad_norm": 0.442790687084198, "learning_rate": 0.0005, "loss": 1.2503992319107056, "step": 34209 }, { "epoch": 2.696886085928262, "grad_norm": 0.5340598225593567, "learning_rate": 0.0005, "loss": 1.380998134613037, "step": 34210 }, { "epoch": 2.6969649191959006, "grad_norm": 0.46583443880081177, "learning_rate": 0.0005, "loss": 1.2494745254516602, "step": 34211 }, { "epoch": 2.69704375246354, "grad_norm": 0.47346869111061096, "learning_rate": 0.0005, "loss": 1.2449915409088135, "step": 34212 }, { "epoch": 2.6971225857311785, "grad_norm": 0.4415161609649658, "learning_rate": 0.0005, "loss": 1.3626543283462524, "step": 34213 }, { "epoch": 2.6972014189988176, "grad_norm": 0.5039005279541016, "learning_rate": 0.0005, "loss": 1.392111897468567, "step": 34214 }, { "epoch": 2.6972802522664563, "grad_norm": 0.4748696982860565, "learning_rate": 0.0005, "loss": 1.170515775680542, "step": 34215 }, { "epoch": 2.6973590855340954, "grad_norm": 0.48270514607429504, "learning_rate": 0.0005, "loss": 1.168766736984253, "step": 34216 }, { "epoch": 2.697437918801734, "grad_norm": 0.562001645565033, "learning_rate": 0.0005, "loss": 1.4458715915679932, "step": 34217 }, { "epoch": 2.6975167520693732, "grad_norm": 0.4657210111618042, "learning_rate": 0.0005, "loss": 1.3038069009780884, "step": 34218 }, { "epoch": 2.6975955853370124, "grad_norm": 0.4778549373149872, "learning_rate": 0.0005, "loss": 1.218085527420044, "step": 34219 }, { "epoch": 2.697674418604651, "grad_norm": 0.5857632160186768, "learning_rate": 0.0005, "loss": 1.4286749362945557, "step": 34220 }, { "epoch": 2.69775325187229, "grad_norm": 0.551235020160675, "learning_rate": 0.0005, "loss": 1.349369764328003, "step": 34221 }, { "epoch": 2.697832085139929, "grad_norm": 0.5736430883407593, "learning_rate": 0.0005, "loss": 1.4805248975753784, "step": 34222 }, { "epoch": 2.697910918407568, "grad_norm": 0.5120357871055603, "learning_rate": 0.0005, "loss": 1.3971490859985352, "step": 34223 }, { "epoch": 2.697989751675207, "grad_norm": 0.5547856688499451, "learning_rate": 0.0005, "loss": 1.3482110500335693, "step": 34224 }, { "epoch": 2.698068584942846, "grad_norm": 0.5093784928321838, "learning_rate": 0.0005, "loss": 1.2357103824615479, "step": 34225 }, { "epoch": 2.698147418210485, "grad_norm": 0.505294919013977, "learning_rate": 0.0005, "loss": 1.342173457145691, "step": 34226 }, { "epoch": 2.6982262514781237, "grad_norm": 0.525395393371582, "learning_rate": 0.0005, "loss": 1.2068160772323608, "step": 34227 }, { "epoch": 2.698305084745763, "grad_norm": 0.5299283862113953, "learning_rate": 0.0005, "loss": 1.319854497909546, "step": 34228 }, { "epoch": 2.6983839180134015, "grad_norm": 0.5209398865699768, "learning_rate": 0.0005, "loss": 1.4122681617736816, "step": 34229 }, { "epoch": 2.6984627512810406, "grad_norm": 0.4613468050956726, "learning_rate": 0.0005, "loss": 1.226306438446045, "step": 34230 }, { "epoch": 2.6985415845486793, "grad_norm": 0.5552906394004822, "learning_rate": 0.0005, "loss": 1.1570600271224976, "step": 34231 }, { "epoch": 2.6986204178163185, "grad_norm": 0.5138357281684875, "learning_rate": 0.0005, "loss": 1.3660328388214111, "step": 34232 }, { "epoch": 2.6986992510839576, "grad_norm": 0.5030091404914856, "learning_rate": 0.0005, "loss": 1.366205096244812, "step": 34233 }, { "epoch": 2.6987780843515963, "grad_norm": 0.5010241866111755, "learning_rate": 0.0005, "loss": 1.226392149925232, "step": 34234 }, { "epoch": 2.6988569176192354, "grad_norm": 0.6238754391670227, "learning_rate": 0.0005, "loss": 1.3574209213256836, "step": 34235 }, { "epoch": 2.698935750886874, "grad_norm": 0.4269697368144989, "learning_rate": 0.0005, "loss": 1.0926640033721924, "step": 34236 }, { "epoch": 2.6990145841545132, "grad_norm": 0.4652954638004303, "learning_rate": 0.0005, "loss": 1.436267375946045, "step": 34237 }, { "epoch": 2.6990934174221524, "grad_norm": 0.4549056589603424, "learning_rate": 0.0005, "loss": 1.2985159158706665, "step": 34238 }, { "epoch": 2.699172250689791, "grad_norm": 0.5300198197364807, "learning_rate": 0.0005, "loss": 1.3268141746520996, "step": 34239 }, { "epoch": 2.69925108395743, "grad_norm": 0.5280650854110718, "learning_rate": 0.0005, "loss": 1.4410145282745361, "step": 34240 }, { "epoch": 2.699329917225069, "grad_norm": 0.5461567640304565, "learning_rate": 0.0005, "loss": 1.3870539665222168, "step": 34241 }, { "epoch": 2.699408750492708, "grad_norm": 0.5246272683143616, "learning_rate": 0.0005, "loss": 1.311160683631897, "step": 34242 }, { "epoch": 2.699487583760347, "grad_norm": 0.4275292158126831, "learning_rate": 0.0005, "loss": 1.1125833988189697, "step": 34243 }, { "epoch": 2.699566417027986, "grad_norm": 0.44458794593811035, "learning_rate": 0.0005, "loss": 1.4093647003173828, "step": 34244 }, { "epoch": 2.6996452502956245, "grad_norm": 0.45639336109161377, "learning_rate": 0.0005, "loss": 1.3408725261688232, "step": 34245 }, { "epoch": 2.6997240835632637, "grad_norm": 0.7979722619056702, "learning_rate": 0.0005, "loss": 1.3453599214553833, "step": 34246 }, { "epoch": 2.699802916830903, "grad_norm": 0.641259491443634, "learning_rate": 0.0005, "loss": 1.4836969375610352, "step": 34247 }, { "epoch": 2.6998817500985415, "grad_norm": 0.5281617045402527, "learning_rate": 0.0005, "loss": 1.498962640762329, "step": 34248 }, { "epoch": 2.6999605833661806, "grad_norm": 0.438931405544281, "learning_rate": 0.0005, "loss": 1.1511375904083252, "step": 34249 }, { "epoch": 2.7000394166338193, "grad_norm": 0.5263501405715942, "learning_rate": 0.0005, "loss": 1.3094379901885986, "step": 34250 }, { "epoch": 2.7001182499014584, "grad_norm": 0.542273223400116, "learning_rate": 0.0005, "loss": 1.485626220703125, "step": 34251 }, { "epoch": 2.7001970831690976, "grad_norm": 0.48738789558410645, "learning_rate": 0.0005, "loss": 1.4860990047454834, "step": 34252 }, { "epoch": 2.7002759164367363, "grad_norm": 0.47611039876937866, "learning_rate": 0.0005, "loss": 1.2954628467559814, "step": 34253 }, { "epoch": 2.7003547497043754, "grad_norm": 0.5729116797447205, "learning_rate": 0.0005, "loss": 1.4678741693496704, "step": 34254 }, { "epoch": 2.700433582972014, "grad_norm": 0.4202398955821991, "learning_rate": 0.0005, "loss": 1.233153223991394, "step": 34255 }, { "epoch": 2.700512416239653, "grad_norm": 0.5203476548194885, "learning_rate": 0.0005, "loss": 1.416849136352539, "step": 34256 }, { "epoch": 2.7005912495072923, "grad_norm": 0.5170702338218689, "learning_rate": 0.0005, "loss": 1.4564874172210693, "step": 34257 }, { "epoch": 2.700670082774931, "grad_norm": 0.41476869583129883, "learning_rate": 0.0005, "loss": 1.1785858869552612, "step": 34258 }, { "epoch": 2.7007489160425697, "grad_norm": 0.7298309206962585, "learning_rate": 0.0005, "loss": 1.301478385925293, "step": 34259 }, { "epoch": 2.700827749310209, "grad_norm": 0.6130403280258179, "learning_rate": 0.0005, "loss": 1.497558355331421, "step": 34260 }, { "epoch": 2.700906582577848, "grad_norm": 0.43737366795539856, "learning_rate": 0.0005, "loss": 1.2075928449630737, "step": 34261 }, { "epoch": 2.7009854158454867, "grad_norm": 0.537415623664856, "learning_rate": 0.0005, "loss": 1.4816150665283203, "step": 34262 }, { "epoch": 2.701064249113126, "grad_norm": 0.49445778131484985, "learning_rate": 0.0005, "loss": 1.375793695449829, "step": 34263 }, { "epoch": 2.7011430823807645, "grad_norm": 0.5743076205253601, "learning_rate": 0.0005, "loss": 1.3050117492675781, "step": 34264 }, { "epoch": 2.7012219156484036, "grad_norm": 0.5075460076332092, "learning_rate": 0.0005, "loss": 1.2879347801208496, "step": 34265 }, { "epoch": 2.7013007489160428, "grad_norm": 0.5372564196586609, "learning_rate": 0.0005, "loss": 1.3301994800567627, "step": 34266 }, { "epoch": 2.7013795821836815, "grad_norm": 0.7220638990402222, "learning_rate": 0.0005, "loss": 1.374069333076477, "step": 34267 }, { "epoch": 2.7014584154513206, "grad_norm": 0.5550183057785034, "learning_rate": 0.0005, "loss": 1.4443931579589844, "step": 34268 }, { "epoch": 2.7015372487189593, "grad_norm": 0.5752488970756531, "learning_rate": 0.0005, "loss": 1.1541621685028076, "step": 34269 }, { "epoch": 2.7016160819865984, "grad_norm": 0.6485399603843689, "learning_rate": 0.0005, "loss": 1.302178144454956, "step": 34270 }, { "epoch": 2.7016949152542376, "grad_norm": 1.1026712656021118, "learning_rate": 0.0005, "loss": 1.335089921951294, "step": 34271 }, { "epoch": 2.7017737485218762, "grad_norm": 0.4720945954322815, "learning_rate": 0.0005, "loss": 1.2915538549423218, "step": 34272 }, { "epoch": 2.701852581789515, "grad_norm": 0.505683183670044, "learning_rate": 0.0005, "loss": 1.2383272647857666, "step": 34273 }, { "epoch": 2.701931415057154, "grad_norm": 0.4395255744457245, "learning_rate": 0.0005, "loss": 1.3407127857208252, "step": 34274 }, { "epoch": 2.702010248324793, "grad_norm": 0.5139374732971191, "learning_rate": 0.0005, "loss": 1.2500966787338257, "step": 34275 }, { "epoch": 2.702089081592432, "grad_norm": 0.5694953203201294, "learning_rate": 0.0005, "loss": 1.4321017265319824, "step": 34276 }, { "epoch": 2.702167914860071, "grad_norm": 0.5355960726737976, "learning_rate": 0.0005, "loss": 1.351982831954956, "step": 34277 }, { "epoch": 2.7022467481277097, "grad_norm": 0.5915052890777588, "learning_rate": 0.0005, "loss": 1.5943603515625, "step": 34278 }, { "epoch": 2.702325581395349, "grad_norm": 0.49189502000808716, "learning_rate": 0.0005, "loss": 1.4696229696273804, "step": 34279 }, { "epoch": 2.702404414662988, "grad_norm": 0.404717355966568, "learning_rate": 0.0005, "loss": 1.133300542831421, "step": 34280 }, { "epoch": 2.7024832479306267, "grad_norm": 0.7651868462562561, "learning_rate": 0.0005, "loss": 1.4942545890808105, "step": 34281 }, { "epoch": 2.702562081198266, "grad_norm": 0.5069852471351624, "learning_rate": 0.0005, "loss": 1.3110642433166504, "step": 34282 }, { "epoch": 2.7026409144659045, "grad_norm": 0.45923876762390137, "learning_rate": 0.0005, "loss": 1.3529200553894043, "step": 34283 }, { "epoch": 2.7027197477335436, "grad_norm": 0.5025811195373535, "learning_rate": 0.0005, "loss": 1.3591285943984985, "step": 34284 }, { "epoch": 2.7027985810011828, "grad_norm": 0.4531281292438507, "learning_rate": 0.0005, "loss": 1.3057053089141846, "step": 34285 }, { "epoch": 2.7028774142688214, "grad_norm": 0.4872249662876129, "learning_rate": 0.0005, "loss": 1.3344926834106445, "step": 34286 }, { "epoch": 2.70295624753646, "grad_norm": 0.44234493374824524, "learning_rate": 0.0005, "loss": 1.342449426651001, "step": 34287 }, { "epoch": 2.7030350808040993, "grad_norm": 0.4345413148403168, "learning_rate": 0.0005, "loss": 1.5073001384735107, "step": 34288 }, { "epoch": 2.7031139140717384, "grad_norm": 0.5779989361763, "learning_rate": 0.0005, "loss": 1.5344065427780151, "step": 34289 }, { "epoch": 2.703192747339377, "grad_norm": 0.5877482295036316, "learning_rate": 0.0005, "loss": 1.5516234636306763, "step": 34290 }, { "epoch": 2.7032715806070162, "grad_norm": 0.5297864675521851, "learning_rate": 0.0005, "loss": 1.346402883529663, "step": 34291 }, { "epoch": 2.703350413874655, "grad_norm": 0.5938069820404053, "learning_rate": 0.0005, "loss": 1.3439007997512817, "step": 34292 }, { "epoch": 2.703429247142294, "grad_norm": 0.6073879599571228, "learning_rate": 0.0005, "loss": 1.3196840286254883, "step": 34293 }, { "epoch": 2.703508080409933, "grad_norm": 0.5117557048797607, "learning_rate": 0.0005, "loss": 1.2829923629760742, "step": 34294 }, { "epoch": 2.703586913677572, "grad_norm": 0.5749456882476807, "learning_rate": 0.0005, "loss": 1.4888336658477783, "step": 34295 }, { "epoch": 2.703665746945211, "grad_norm": 0.5409713983535767, "learning_rate": 0.0005, "loss": 1.4799888134002686, "step": 34296 }, { "epoch": 2.7037445802128497, "grad_norm": 0.5509085655212402, "learning_rate": 0.0005, "loss": 1.4467122554779053, "step": 34297 }, { "epoch": 2.703823413480489, "grad_norm": 0.4332573413848877, "learning_rate": 0.0005, "loss": 1.224259853363037, "step": 34298 }, { "epoch": 2.703902246748128, "grad_norm": 0.5332505702972412, "learning_rate": 0.0005, "loss": 1.3183410167694092, "step": 34299 }, { "epoch": 2.7039810800157666, "grad_norm": 0.49313753843307495, "learning_rate": 0.0005, "loss": 1.1456739902496338, "step": 34300 }, { "epoch": 2.7040599132834053, "grad_norm": 0.4578081965446472, "learning_rate": 0.0005, "loss": 1.1349269151687622, "step": 34301 }, { "epoch": 2.7041387465510445, "grad_norm": 0.4580026865005493, "learning_rate": 0.0005, "loss": 1.2992690801620483, "step": 34302 }, { "epoch": 2.7042175798186836, "grad_norm": 0.47371962666511536, "learning_rate": 0.0005, "loss": 1.2899765968322754, "step": 34303 }, { "epoch": 2.7042964130863223, "grad_norm": 0.6102655529975891, "learning_rate": 0.0005, "loss": 1.2531754970550537, "step": 34304 }, { "epoch": 2.7043752463539614, "grad_norm": 0.48399195075035095, "learning_rate": 0.0005, "loss": 1.3303171396255493, "step": 34305 }, { "epoch": 2.7044540796216, "grad_norm": 0.5079290866851807, "learning_rate": 0.0005, "loss": 1.2858179807662964, "step": 34306 }, { "epoch": 2.7045329128892392, "grad_norm": 0.5043085813522339, "learning_rate": 0.0005, "loss": 1.2361516952514648, "step": 34307 }, { "epoch": 2.7046117461568784, "grad_norm": 0.4535142779350281, "learning_rate": 0.0005, "loss": 1.3442130088806152, "step": 34308 }, { "epoch": 2.704690579424517, "grad_norm": 0.4158138036727905, "learning_rate": 0.0005, "loss": 1.3861427307128906, "step": 34309 }, { "epoch": 2.704769412692156, "grad_norm": 0.6055302023887634, "learning_rate": 0.0005, "loss": 1.5779505968093872, "step": 34310 }, { "epoch": 2.704848245959795, "grad_norm": 0.532481849193573, "learning_rate": 0.0005, "loss": 1.443247675895691, "step": 34311 }, { "epoch": 2.704927079227434, "grad_norm": 0.5242669582366943, "learning_rate": 0.0005, "loss": 1.2639563083648682, "step": 34312 }, { "epoch": 2.705005912495073, "grad_norm": 0.47967422008514404, "learning_rate": 0.0005, "loss": 1.342625379562378, "step": 34313 }, { "epoch": 2.705084745762712, "grad_norm": 0.4890711009502411, "learning_rate": 0.0005, "loss": 1.5542855262756348, "step": 34314 }, { "epoch": 2.7051635790303505, "grad_norm": 0.4252607822418213, "learning_rate": 0.0005, "loss": 1.3260486125946045, "step": 34315 }, { "epoch": 2.7052424122979897, "grad_norm": 0.459984689950943, "learning_rate": 0.0005, "loss": 1.0284347534179688, "step": 34316 }, { "epoch": 2.705321245565629, "grad_norm": 0.516120195388794, "learning_rate": 0.0005, "loss": 1.3960158824920654, "step": 34317 }, { "epoch": 2.7054000788332675, "grad_norm": 0.46135014295578003, "learning_rate": 0.0005, "loss": 1.44343900680542, "step": 34318 }, { "epoch": 2.7054789121009066, "grad_norm": 0.5068410038948059, "learning_rate": 0.0005, "loss": 1.28256094455719, "step": 34319 }, { "epoch": 2.7055577453685453, "grad_norm": 0.40585362911224365, "learning_rate": 0.0005, "loss": 1.1654157638549805, "step": 34320 }, { "epoch": 2.7056365786361845, "grad_norm": 0.4325585663318634, "learning_rate": 0.0005, "loss": 1.2593499422073364, "step": 34321 }, { "epoch": 2.7057154119038236, "grad_norm": 0.535926103591919, "learning_rate": 0.0005, "loss": 1.3719452619552612, "step": 34322 }, { "epoch": 2.7057942451714623, "grad_norm": 0.4926355481147766, "learning_rate": 0.0005, "loss": 1.358567714691162, "step": 34323 }, { "epoch": 2.7058730784391014, "grad_norm": 0.4891452491283417, "learning_rate": 0.0005, "loss": 1.398715615272522, "step": 34324 }, { "epoch": 2.70595191170674, "grad_norm": 0.46007975935935974, "learning_rate": 0.0005, "loss": 1.3158496618270874, "step": 34325 }, { "epoch": 2.7060307449743792, "grad_norm": 0.4637880027294159, "learning_rate": 0.0005, "loss": 1.3024945259094238, "step": 34326 }, { "epoch": 2.7061095782420184, "grad_norm": 0.45651665329933167, "learning_rate": 0.0005, "loss": 1.3926713466644287, "step": 34327 }, { "epoch": 2.706188411509657, "grad_norm": 0.4507651925086975, "learning_rate": 0.0005, "loss": 1.2206110954284668, "step": 34328 }, { "epoch": 2.7062672447772957, "grad_norm": 0.7174240350723267, "learning_rate": 0.0005, "loss": 1.5085864067077637, "step": 34329 }, { "epoch": 2.706346078044935, "grad_norm": 0.5022456049919128, "learning_rate": 0.0005, "loss": 1.4722251892089844, "step": 34330 }, { "epoch": 2.706424911312574, "grad_norm": 0.6288166046142578, "learning_rate": 0.0005, "loss": 1.3522355556488037, "step": 34331 }, { "epoch": 2.7065037445802127, "grad_norm": 0.6827459335327148, "learning_rate": 0.0005, "loss": 1.7805039882659912, "step": 34332 }, { "epoch": 2.706582577847852, "grad_norm": 0.5125550031661987, "learning_rate": 0.0005, "loss": 1.2163945436477661, "step": 34333 }, { "epoch": 2.7066614111154905, "grad_norm": 0.5119127631187439, "learning_rate": 0.0005, "loss": 1.2542352676391602, "step": 34334 }, { "epoch": 2.7067402443831297, "grad_norm": 0.5906087756156921, "learning_rate": 0.0005, "loss": 1.3241891860961914, "step": 34335 }, { "epoch": 2.706819077650769, "grad_norm": 0.45604854822158813, "learning_rate": 0.0005, "loss": 1.2558835744857788, "step": 34336 }, { "epoch": 2.7068979109184075, "grad_norm": 0.9140849709510803, "learning_rate": 0.0005, "loss": 1.2862428426742554, "step": 34337 }, { "epoch": 2.7069767441860466, "grad_norm": 0.45194488763809204, "learning_rate": 0.0005, "loss": 1.1976745128631592, "step": 34338 }, { "epoch": 2.7070555774536853, "grad_norm": 0.5913553237915039, "learning_rate": 0.0005, "loss": 1.4875173568725586, "step": 34339 }, { "epoch": 2.7071344107213244, "grad_norm": 0.5335967540740967, "learning_rate": 0.0005, "loss": 1.4118905067443848, "step": 34340 }, { "epoch": 2.7072132439889636, "grad_norm": 0.5064572691917419, "learning_rate": 0.0005, "loss": 1.4608702659606934, "step": 34341 }, { "epoch": 2.7072920772566023, "grad_norm": 0.45945680141448975, "learning_rate": 0.0005, "loss": 1.3256802558898926, "step": 34342 }, { "epoch": 2.7073709105242414, "grad_norm": 0.5113971829414368, "learning_rate": 0.0005, "loss": 1.267032504081726, "step": 34343 }, { "epoch": 2.70744974379188, "grad_norm": 0.46043118834495544, "learning_rate": 0.0005, "loss": 1.2484004497528076, "step": 34344 }, { "epoch": 2.707528577059519, "grad_norm": 1.0642998218536377, "learning_rate": 0.0005, "loss": 1.1988089084625244, "step": 34345 }, { "epoch": 2.707607410327158, "grad_norm": 0.5768689513206482, "learning_rate": 0.0005, "loss": 1.6431964635849, "step": 34346 }, { "epoch": 2.707686243594797, "grad_norm": 0.49068567156791687, "learning_rate": 0.0005, "loss": 1.3318818807601929, "step": 34347 }, { "epoch": 2.7077650768624357, "grad_norm": 0.6323900818824768, "learning_rate": 0.0005, "loss": 1.250502109527588, "step": 34348 }, { "epoch": 2.707843910130075, "grad_norm": 0.495082825422287, "learning_rate": 0.0005, "loss": 1.5676896572113037, "step": 34349 }, { "epoch": 2.707922743397714, "grad_norm": 0.44642627239227295, "learning_rate": 0.0005, "loss": 1.1636360883712769, "step": 34350 }, { "epoch": 2.7080015766653527, "grad_norm": 0.5660710334777832, "learning_rate": 0.0005, "loss": 1.3252553939819336, "step": 34351 }, { "epoch": 2.708080409932992, "grad_norm": 0.4972708523273468, "learning_rate": 0.0005, "loss": 1.2639727592468262, "step": 34352 }, { "epoch": 2.7081592432006305, "grad_norm": 0.5919227600097656, "learning_rate": 0.0005, "loss": 1.3727376461029053, "step": 34353 }, { "epoch": 2.7082380764682696, "grad_norm": 0.46513810753822327, "learning_rate": 0.0005, "loss": 1.362696886062622, "step": 34354 }, { "epoch": 2.7083169097359088, "grad_norm": 0.44168853759765625, "learning_rate": 0.0005, "loss": 1.2144256830215454, "step": 34355 }, { "epoch": 2.7083957430035475, "grad_norm": 0.4381781816482544, "learning_rate": 0.0005, "loss": 1.186439037322998, "step": 34356 }, { "epoch": 2.7084745762711866, "grad_norm": 0.6017748117446899, "learning_rate": 0.0005, "loss": 1.4248127937316895, "step": 34357 }, { "epoch": 2.7085534095388253, "grad_norm": 0.4297812879085541, "learning_rate": 0.0005, "loss": 1.1825664043426514, "step": 34358 }, { "epoch": 2.7086322428064644, "grad_norm": 0.5193182229995728, "learning_rate": 0.0005, "loss": 1.3505580425262451, "step": 34359 }, { "epoch": 2.708711076074103, "grad_norm": 0.47094839811325073, "learning_rate": 0.0005, "loss": 1.1917037963867188, "step": 34360 }, { "epoch": 2.7087899093417422, "grad_norm": 0.8296535611152649, "learning_rate": 0.0005, "loss": 1.2625658512115479, "step": 34361 }, { "epoch": 2.708868742609381, "grad_norm": 0.48834577202796936, "learning_rate": 0.0005, "loss": 1.4244121313095093, "step": 34362 }, { "epoch": 2.70894757587702, "grad_norm": 0.514889657497406, "learning_rate": 0.0005, "loss": 1.4923105239868164, "step": 34363 }, { "epoch": 2.709026409144659, "grad_norm": 0.49562326073646545, "learning_rate": 0.0005, "loss": 1.456966519355774, "step": 34364 }, { "epoch": 2.709105242412298, "grad_norm": 0.5656524300575256, "learning_rate": 0.0005, "loss": 1.6677725315093994, "step": 34365 }, { "epoch": 2.709184075679937, "grad_norm": 0.46748828887939453, "learning_rate": 0.0005, "loss": 1.2490010261535645, "step": 34366 }, { "epoch": 2.7092629089475757, "grad_norm": 0.4983735680580139, "learning_rate": 0.0005, "loss": 1.2950656414031982, "step": 34367 }, { "epoch": 2.709341742215215, "grad_norm": 0.39837008714675903, "learning_rate": 0.0005, "loss": 1.1164906024932861, "step": 34368 }, { "epoch": 2.709420575482854, "grad_norm": 0.5089168548583984, "learning_rate": 0.0005, "loss": 1.4452180862426758, "step": 34369 }, { "epoch": 2.7094994087504927, "grad_norm": 0.3903900682926178, "learning_rate": 0.0005, "loss": 1.2636367082595825, "step": 34370 }, { "epoch": 2.709578242018132, "grad_norm": 0.5558512806892395, "learning_rate": 0.0005, "loss": 1.463389277458191, "step": 34371 }, { "epoch": 2.7096570752857705, "grad_norm": 0.642298698425293, "learning_rate": 0.0005, "loss": 1.2580437660217285, "step": 34372 }, { "epoch": 2.7097359085534096, "grad_norm": 0.5790566802024841, "learning_rate": 0.0005, "loss": 1.4104782342910767, "step": 34373 }, { "epoch": 2.7098147418210488, "grad_norm": 0.5163925886154175, "learning_rate": 0.0005, "loss": 1.1534312963485718, "step": 34374 }, { "epoch": 2.7098935750886874, "grad_norm": 0.4335675537586212, "learning_rate": 0.0005, "loss": 1.2881135940551758, "step": 34375 }, { "epoch": 2.709972408356326, "grad_norm": 0.5446009039878845, "learning_rate": 0.0005, "loss": 1.3390021324157715, "step": 34376 }, { "epoch": 2.7100512416239653, "grad_norm": 0.43745583295822144, "learning_rate": 0.0005, "loss": 1.3341929912567139, "step": 34377 }, { "epoch": 2.7101300748916044, "grad_norm": 0.7530010342597961, "learning_rate": 0.0005, "loss": 1.3661746978759766, "step": 34378 }, { "epoch": 2.710208908159243, "grad_norm": 0.47476667165756226, "learning_rate": 0.0005, "loss": 1.303658127784729, "step": 34379 }, { "epoch": 2.7102877414268822, "grad_norm": 0.4501314163208008, "learning_rate": 0.0005, "loss": 1.265195608139038, "step": 34380 }, { "epoch": 2.710366574694521, "grad_norm": 0.48211535811424255, "learning_rate": 0.0005, "loss": 1.3159226179122925, "step": 34381 }, { "epoch": 2.71044540796216, "grad_norm": 0.5686248540878296, "learning_rate": 0.0005, "loss": 1.406638264656067, "step": 34382 }, { "epoch": 2.710524241229799, "grad_norm": 0.4947720170021057, "learning_rate": 0.0005, "loss": 1.2616913318634033, "step": 34383 }, { "epoch": 2.710603074497438, "grad_norm": 0.6133501529693604, "learning_rate": 0.0005, "loss": 1.672907829284668, "step": 34384 }, { "epoch": 2.710681907765077, "grad_norm": 0.4976541996002197, "learning_rate": 0.0005, "loss": 1.269314169883728, "step": 34385 }, { "epoch": 2.7107607410327157, "grad_norm": 0.4245009124279022, "learning_rate": 0.0005, "loss": 1.3802217245101929, "step": 34386 }, { "epoch": 2.710839574300355, "grad_norm": 0.5368014574050903, "learning_rate": 0.0005, "loss": 1.3864073753356934, "step": 34387 }, { "epoch": 2.710918407567994, "grad_norm": 0.5074777007102966, "learning_rate": 0.0005, "loss": 1.2617394924163818, "step": 34388 }, { "epoch": 2.7109972408356326, "grad_norm": 0.4545929729938507, "learning_rate": 0.0005, "loss": 1.3778645992279053, "step": 34389 }, { "epoch": 2.7110760741032713, "grad_norm": 0.5514860153198242, "learning_rate": 0.0005, "loss": 1.4928771257400513, "step": 34390 }, { "epoch": 2.7111549073709105, "grad_norm": 0.41025853157043457, "learning_rate": 0.0005, "loss": 1.339051604270935, "step": 34391 }, { "epoch": 2.7112337406385496, "grad_norm": 0.5143442749977112, "learning_rate": 0.0005, "loss": 1.3052936792373657, "step": 34392 }, { "epoch": 2.7113125739061883, "grad_norm": 0.4932185113430023, "learning_rate": 0.0005, "loss": 1.1878407001495361, "step": 34393 }, { "epoch": 2.7113914071738274, "grad_norm": 0.4372447729110718, "learning_rate": 0.0005, "loss": 1.2168514728546143, "step": 34394 }, { "epoch": 2.711470240441466, "grad_norm": 0.4435056447982788, "learning_rate": 0.0005, "loss": 1.2507567405700684, "step": 34395 }, { "epoch": 2.7115490737091053, "grad_norm": 0.4177643358707428, "learning_rate": 0.0005, "loss": 1.2981925010681152, "step": 34396 }, { "epoch": 2.7116279069767444, "grad_norm": 0.43986958265304565, "learning_rate": 0.0005, "loss": 1.3280408382415771, "step": 34397 }, { "epoch": 2.711706740244383, "grad_norm": 0.5365113019943237, "learning_rate": 0.0005, "loss": 1.3589646816253662, "step": 34398 }, { "epoch": 2.711785573512022, "grad_norm": 0.37012407183647156, "learning_rate": 0.0005, "loss": 1.087405800819397, "step": 34399 }, { "epoch": 2.711864406779661, "grad_norm": 0.4986303746700287, "learning_rate": 0.0005, "loss": 1.2338573932647705, "step": 34400 }, { "epoch": 2.7119432400473, "grad_norm": 0.5022056698799133, "learning_rate": 0.0005, "loss": 1.4988081455230713, "step": 34401 }, { "epoch": 2.712022073314939, "grad_norm": 0.5910390019416809, "learning_rate": 0.0005, "loss": 1.3884447813034058, "step": 34402 }, { "epoch": 2.712100906582578, "grad_norm": 0.4404480457305908, "learning_rate": 0.0005, "loss": 1.1383838653564453, "step": 34403 }, { "epoch": 2.7121797398502165, "grad_norm": 0.508614718914032, "learning_rate": 0.0005, "loss": 1.5696090459823608, "step": 34404 }, { "epoch": 2.7122585731178557, "grad_norm": 0.5761644244194031, "learning_rate": 0.0005, "loss": 1.454714059829712, "step": 34405 }, { "epoch": 2.712337406385495, "grad_norm": 0.4993603229522705, "learning_rate": 0.0005, "loss": 1.3233145475387573, "step": 34406 }, { "epoch": 2.7124162396531335, "grad_norm": 0.5048706531524658, "learning_rate": 0.0005, "loss": 1.4541692733764648, "step": 34407 }, { "epoch": 2.7124950729207726, "grad_norm": 0.5044744610786438, "learning_rate": 0.0005, "loss": 1.35781991481781, "step": 34408 }, { "epoch": 2.7125739061884113, "grad_norm": 0.5194818377494812, "learning_rate": 0.0005, "loss": 1.3640310764312744, "step": 34409 }, { "epoch": 2.7126527394560505, "grad_norm": 0.5577256679534912, "learning_rate": 0.0005, "loss": 1.3952603340148926, "step": 34410 }, { "epoch": 2.7127315727236896, "grad_norm": 0.4877791106700897, "learning_rate": 0.0005, "loss": 1.315967321395874, "step": 34411 }, { "epoch": 2.7128104059913283, "grad_norm": 0.5247555375099182, "learning_rate": 0.0005, "loss": 1.3706858158111572, "step": 34412 }, { "epoch": 2.7128892392589674, "grad_norm": 0.45197924971580505, "learning_rate": 0.0005, "loss": 1.1159471273422241, "step": 34413 }, { "epoch": 2.712968072526606, "grad_norm": 0.6237931847572327, "learning_rate": 0.0005, "loss": 1.3756047487258911, "step": 34414 }, { "epoch": 2.7130469057942452, "grad_norm": 0.4967743158340454, "learning_rate": 0.0005, "loss": 1.3572580814361572, "step": 34415 }, { "epoch": 2.7131257390618844, "grad_norm": 0.5485376715660095, "learning_rate": 0.0005, "loss": 1.3366503715515137, "step": 34416 }, { "epoch": 2.713204572329523, "grad_norm": 0.467570036649704, "learning_rate": 0.0005, "loss": 1.2451204061508179, "step": 34417 }, { "epoch": 2.7132834055971617, "grad_norm": 0.7499520182609558, "learning_rate": 0.0005, "loss": 1.1950223445892334, "step": 34418 }, { "epoch": 2.713362238864801, "grad_norm": 0.49260106682777405, "learning_rate": 0.0005, "loss": 1.2253113985061646, "step": 34419 }, { "epoch": 2.71344107213244, "grad_norm": 0.49077194929122925, "learning_rate": 0.0005, "loss": 1.3757343292236328, "step": 34420 }, { "epoch": 2.7135199054000787, "grad_norm": 0.3937557637691498, "learning_rate": 0.0005, "loss": 1.1367545127868652, "step": 34421 }, { "epoch": 2.713598738667718, "grad_norm": 0.47390201687812805, "learning_rate": 0.0005, "loss": 1.3445513248443604, "step": 34422 }, { "epoch": 2.7136775719353565, "grad_norm": 0.6058135628700256, "learning_rate": 0.0005, "loss": 1.3910269737243652, "step": 34423 }, { "epoch": 2.7137564052029957, "grad_norm": 0.4190906286239624, "learning_rate": 0.0005, "loss": 1.229418158531189, "step": 34424 }, { "epoch": 2.713835238470635, "grad_norm": 0.46513015031814575, "learning_rate": 0.0005, "loss": 1.4173791408538818, "step": 34425 }, { "epoch": 2.7139140717382735, "grad_norm": 0.5398637652397156, "learning_rate": 0.0005, "loss": 1.2955036163330078, "step": 34426 }, { "epoch": 2.7139929050059126, "grad_norm": 0.5563393831253052, "learning_rate": 0.0005, "loss": 1.391981601715088, "step": 34427 }, { "epoch": 2.7140717382735513, "grad_norm": 0.4962557852268219, "learning_rate": 0.0005, "loss": 1.1824849843978882, "step": 34428 }, { "epoch": 2.7141505715411904, "grad_norm": 0.48300251364707947, "learning_rate": 0.0005, "loss": 1.305403470993042, "step": 34429 }, { "epoch": 2.7142294048088296, "grad_norm": 0.4115666151046753, "learning_rate": 0.0005, "loss": 1.2349666357040405, "step": 34430 }, { "epoch": 2.7143082380764683, "grad_norm": 0.5902106761932373, "learning_rate": 0.0005, "loss": 1.3720132112503052, "step": 34431 }, { "epoch": 2.714387071344107, "grad_norm": 0.5042228102684021, "learning_rate": 0.0005, "loss": 1.375920295715332, "step": 34432 }, { "epoch": 2.714465904611746, "grad_norm": 0.4226807653903961, "learning_rate": 0.0005, "loss": 1.072523832321167, "step": 34433 }, { "epoch": 2.714544737879385, "grad_norm": 0.4391770660877228, "learning_rate": 0.0005, "loss": 1.217565655708313, "step": 34434 }, { "epoch": 2.714623571147024, "grad_norm": 0.5162904858589172, "learning_rate": 0.0005, "loss": 1.300013780593872, "step": 34435 }, { "epoch": 2.714702404414663, "grad_norm": 0.38875722885131836, "learning_rate": 0.0005, "loss": 1.1913870573043823, "step": 34436 }, { "epoch": 2.7147812376823017, "grad_norm": 0.6098119616508484, "learning_rate": 0.0005, "loss": 1.350438117980957, "step": 34437 }, { "epoch": 2.714860070949941, "grad_norm": 0.5980523228645325, "learning_rate": 0.0005, "loss": 1.3443201780319214, "step": 34438 }, { "epoch": 2.71493890421758, "grad_norm": 0.36822202801704407, "learning_rate": 0.0005, "loss": 1.2385375499725342, "step": 34439 }, { "epoch": 2.7150177374852187, "grad_norm": 0.4685279130935669, "learning_rate": 0.0005, "loss": 1.3774135112762451, "step": 34440 }, { "epoch": 2.715096570752858, "grad_norm": 0.5270456671714783, "learning_rate": 0.0005, "loss": 1.1840925216674805, "step": 34441 }, { "epoch": 2.7151754040204965, "grad_norm": 0.4755566120147705, "learning_rate": 0.0005, "loss": 1.4448833465576172, "step": 34442 }, { "epoch": 2.7152542372881356, "grad_norm": 0.4852381944656372, "learning_rate": 0.0005, "loss": 1.4026987552642822, "step": 34443 }, { "epoch": 2.7153330705557748, "grad_norm": 0.5227851271629333, "learning_rate": 0.0005, "loss": 1.2187533378601074, "step": 34444 }, { "epoch": 2.7154119038234135, "grad_norm": 0.7387179136276245, "learning_rate": 0.0005, "loss": 1.4515290260314941, "step": 34445 }, { "epoch": 2.715490737091052, "grad_norm": 0.614113450050354, "learning_rate": 0.0005, "loss": 1.4632099866867065, "step": 34446 }, { "epoch": 2.7155695703586913, "grad_norm": 0.5398443937301636, "learning_rate": 0.0005, "loss": 1.309295654296875, "step": 34447 }, { "epoch": 2.7156484036263304, "grad_norm": 0.4587075114250183, "learning_rate": 0.0005, "loss": 1.2274540662765503, "step": 34448 }, { "epoch": 2.715727236893969, "grad_norm": 0.582019567489624, "learning_rate": 0.0005, "loss": 1.191056251525879, "step": 34449 }, { "epoch": 2.7158060701616082, "grad_norm": 0.563996434211731, "learning_rate": 0.0005, "loss": 1.3194429874420166, "step": 34450 }, { "epoch": 2.715884903429247, "grad_norm": 0.46707072854042053, "learning_rate": 0.0005, "loss": 1.1440786123275757, "step": 34451 }, { "epoch": 2.715963736696886, "grad_norm": 0.6069936752319336, "learning_rate": 0.0005, "loss": 1.3212177753448486, "step": 34452 }, { "epoch": 2.716042569964525, "grad_norm": 0.5275227427482605, "learning_rate": 0.0005, "loss": 1.4463006258010864, "step": 34453 }, { "epoch": 2.716121403232164, "grad_norm": 0.4838030934333801, "learning_rate": 0.0005, "loss": 1.4699034690856934, "step": 34454 }, { "epoch": 2.716200236499803, "grad_norm": 0.508743166923523, "learning_rate": 0.0005, "loss": 1.4362560510635376, "step": 34455 }, { "epoch": 2.7162790697674417, "grad_norm": 0.4708414673805237, "learning_rate": 0.0005, "loss": 1.360889196395874, "step": 34456 }, { "epoch": 2.716357903035081, "grad_norm": 0.6855814456939697, "learning_rate": 0.0005, "loss": 1.363241195678711, "step": 34457 }, { "epoch": 2.71643673630272, "grad_norm": 0.40793830156326294, "learning_rate": 0.0005, "loss": 1.178438425064087, "step": 34458 }, { "epoch": 2.7165155695703587, "grad_norm": 0.4276219308376312, "learning_rate": 0.0005, "loss": 1.384469747543335, "step": 34459 }, { "epoch": 2.716594402837998, "grad_norm": 0.4812203645706177, "learning_rate": 0.0005, "loss": 1.1919958591461182, "step": 34460 }, { "epoch": 2.7166732361056365, "grad_norm": 0.6388936042785645, "learning_rate": 0.0005, "loss": 1.575253963470459, "step": 34461 }, { "epoch": 2.7167520693732756, "grad_norm": 0.5262898802757263, "learning_rate": 0.0005, "loss": 1.394822120666504, "step": 34462 }, { "epoch": 2.7168309026409143, "grad_norm": 0.489226371049881, "learning_rate": 0.0005, "loss": 1.2845110893249512, "step": 34463 }, { "epoch": 2.7169097359085534, "grad_norm": 0.47668418288230896, "learning_rate": 0.0005, "loss": 1.4214177131652832, "step": 34464 }, { "epoch": 2.716988569176192, "grad_norm": 0.46671178936958313, "learning_rate": 0.0005, "loss": 1.2933151721954346, "step": 34465 }, { "epoch": 2.7170674024438313, "grad_norm": 0.502203106880188, "learning_rate": 0.0005, "loss": 1.37656569480896, "step": 34466 }, { "epoch": 2.7171462357114704, "grad_norm": 0.5004785060882568, "learning_rate": 0.0005, "loss": 1.1476032733917236, "step": 34467 }, { "epoch": 2.717225068979109, "grad_norm": 0.6927358508110046, "learning_rate": 0.0005, "loss": 1.417687177658081, "step": 34468 }, { "epoch": 2.7173039022467482, "grad_norm": 0.3983639180660248, "learning_rate": 0.0005, "loss": 1.203053593635559, "step": 34469 }, { "epoch": 2.717382735514387, "grad_norm": 0.5272011756896973, "learning_rate": 0.0005, "loss": 1.3559467792510986, "step": 34470 }, { "epoch": 2.717461568782026, "grad_norm": 0.4644896984100342, "learning_rate": 0.0005, "loss": 1.336721658706665, "step": 34471 }, { "epoch": 2.717540402049665, "grad_norm": 0.6155636310577393, "learning_rate": 0.0005, "loss": 1.3733206987380981, "step": 34472 }, { "epoch": 2.717619235317304, "grad_norm": 0.5999853610992432, "learning_rate": 0.0005, "loss": 1.3265793323516846, "step": 34473 }, { "epoch": 2.717698068584943, "grad_norm": 0.4797183573246002, "learning_rate": 0.0005, "loss": 1.1823506355285645, "step": 34474 }, { "epoch": 2.7177769018525817, "grad_norm": 0.4729335308074951, "learning_rate": 0.0005, "loss": 1.3875283002853394, "step": 34475 }, { "epoch": 2.717855735120221, "grad_norm": 0.545574963092804, "learning_rate": 0.0005, "loss": 1.35357666015625, "step": 34476 }, { "epoch": 2.7179345683878595, "grad_norm": 0.5478201508522034, "learning_rate": 0.0005, "loss": 1.3335117101669312, "step": 34477 }, { "epoch": 2.7180134016554987, "grad_norm": 0.5438385605812073, "learning_rate": 0.0005, "loss": 1.3878459930419922, "step": 34478 }, { "epoch": 2.7180922349231373, "grad_norm": 0.5385991930961609, "learning_rate": 0.0005, "loss": 1.316450595855713, "step": 34479 }, { "epoch": 2.7181710681907765, "grad_norm": 0.5100459456443787, "learning_rate": 0.0005, "loss": 1.2487046718597412, "step": 34480 }, { "epoch": 2.7182499014584156, "grad_norm": 0.43466314673423767, "learning_rate": 0.0005, "loss": 1.2511366605758667, "step": 34481 }, { "epoch": 2.7183287347260543, "grad_norm": 0.5244939923286438, "learning_rate": 0.0005, "loss": 1.31376051902771, "step": 34482 }, { "epoch": 2.7184075679936934, "grad_norm": 0.5405738949775696, "learning_rate": 0.0005, "loss": 1.2630897760391235, "step": 34483 }, { "epoch": 2.718486401261332, "grad_norm": 0.39223551750183105, "learning_rate": 0.0005, "loss": 1.2482162714004517, "step": 34484 }, { "epoch": 2.7185652345289713, "grad_norm": 0.4640350639820099, "learning_rate": 0.0005, "loss": 1.164923906326294, "step": 34485 }, { "epoch": 2.7186440677966104, "grad_norm": 0.6280139684677124, "learning_rate": 0.0005, "loss": 1.5457276105880737, "step": 34486 }, { "epoch": 2.718722901064249, "grad_norm": 0.41708189249038696, "learning_rate": 0.0005, "loss": 1.31986665725708, "step": 34487 }, { "epoch": 2.718801734331888, "grad_norm": 0.5098757743835449, "learning_rate": 0.0005, "loss": 1.2759697437286377, "step": 34488 }, { "epoch": 2.718880567599527, "grad_norm": 0.5534725189208984, "learning_rate": 0.0005, "loss": 1.6769840717315674, "step": 34489 }, { "epoch": 2.718959400867166, "grad_norm": 0.48769059777259827, "learning_rate": 0.0005, "loss": 1.3384063243865967, "step": 34490 }, { "epoch": 2.7190382341348047, "grad_norm": 0.49945268034935, "learning_rate": 0.0005, "loss": 1.3963102102279663, "step": 34491 }, { "epoch": 2.719117067402444, "grad_norm": 0.5357636213302612, "learning_rate": 0.0005, "loss": 1.2409706115722656, "step": 34492 }, { "epoch": 2.7191959006700825, "grad_norm": 0.6638696193695068, "learning_rate": 0.0005, "loss": 1.48370361328125, "step": 34493 }, { "epoch": 2.7192747339377217, "grad_norm": 0.4353465139865875, "learning_rate": 0.0005, "loss": 1.1081513166427612, "step": 34494 }, { "epoch": 2.719353567205361, "grad_norm": 0.4177990257740021, "learning_rate": 0.0005, "loss": 1.2116913795471191, "step": 34495 }, { "epoch": 2.7194324004729995, "grad_norm": 0.4738903045654297, "learning_rate": 0.0005, "loss": 1.1348426342010498, "step": 34496 }, { "epoch": 2.7195112337406386, "grad_norm": 0.49094295501708984, "learning_rate": 0.0005, "loss": 1.3112601041793823, "step": 34497 }, { "epoch": 2.7195900670082773, "grad_norm": 0.46865561604499817, "learning_rate": 0.0005, "loss": 1.4118248224258423, "step": 34498 }, { "epoch": 2.7196689002759165, "grad_norm": 0.7232500910758972, "learning_rate": 0.0005, "loss": 1.4190462827682495, "step": 34499 }, { "epoch": 2.7197477335435556, "grad_norm": 0.5022234320640564, "learning_rate": 0.0005, "loss": 1.333601474761963, "step": 34500 }, { "epoch": 2.7198265668111943, "grad_norm": 0.4227276146411896, "learning_rate": 0.0005, "loss": 1.3028948307037354, "step": 34501 }, { "epoch": 2.7199054000788334, "grad_norm": 0.45754188299179077, "learning_rate": 0.0005, "loss": 1.1653509140014648, "step": 34502 }, { "epoch": 2.719984233346472, "grad_norm": 0.4251774549484253, "learning_rate": 0.0005, "loss": 1.0564212799072266, "step": 34503 }, { "epoch": 2.7200630666141112, "grad_norm": 0.5699372291564941, "learning_rate": 0.0005, "loss": 1.606750726699829, "step": 34504 }, { "epoch": 2.7201418998817504, "grad_norm": 0.5728313326835632, "learning_rate": 0.0005, "loss": 1.4732816219329834, "step": 34505 }, { "epoch": 2.720220733149389, "grad_norm": 0.5114802718162537, "learning_rate": 0.0005, "loss": 1.2829176187515259, "step": 34506 }, { "epoch": 2.7202995664170277, "grad_norm": 0.5317367315292358, "learning_rate": 0.0005, "loss": 1.5203852653503418, "step": 34507 }, { "epoch": 2.720378399684667, "grad_norm": 0.6534552574157715, "learning_rate": 0.0005, "loss": 1.5185753107070923, "step": 34508 }, { "epoch": 2.720457232952306, "grad_norm": 0.4856049418449402, "learning_rate": 0.0005, "loss": 1.3938857316970825, "step": 34509 }, { "epoch": 2.7205360662199447, "grad_norm": 0.45039525628089905, "learning_rate": 0.0005, "loss": 1.3627585172653198, "step": 34510 }, { "epoch": 2.720614899487584, "grad_norm": 0.47762370109558105, "learning_rate": 0.0005, "loss": 1.4769091606140137, "step": 34511 }, { "epoch": 2.7206937327552225, "grad_norm": 0.394269734621048, "learning_rate": 0.0005, "loss": 1.2627030611038208, "step": 34512 }, { "epoch": 2.7207725660228617, "grad_norm": 0.6318206787109375, "learning_rate": 0.0005, "loss": 1.490736484527588, "step": 34513 }, { "epoch": 2.720851399290501, "grad_norm": 0.4506362974643707, "learning_rate": 0.0005, "loss": 1.4286465644836426, "step": 34514 }, { "epoch": 2.7209302325581395, "grad_norm": 0.5853074789047241, "learning_rate": 0.0005, "loss": 1.4475716352462769, "step": 34515 }, { "epoch": 2.7210090658257786, "grad_norm": 0.4638093411922455, "learning_rate": 0.0005, "loss": 1.281639575958252, "step": 34516 }, { "epoch": 2.7210878990934173, "grad_norm": 0.640788197517395, "learning_rate": 0.0005, "loss": 1.3280980587005615, "step": 34517 }, { "epoch": 2.7211667323610564, "grad_norm": 0.6199787259101868, "learning_rate": 0.0005, "loss": 1.3126767873764038, "step": 34518 }, { "epoch": 2.7212455656286956, "grad_norm": 0.5716682076454163, "learning_rate": 0.0005, "loss": 1.4117319583892822, "step": 34519 }, { "epoch": 2.7213243988963343, "grad_norm": 0.6454521417617798, "learning_rate": 0.0005, "loss": 1.480217695236206, "step": 34520 }, { "epoch": 2.721403232163973, "grad_norm": 0.4426419138908386, "learning_rate": 0.0005, "loss": 1.404619574546814, "step": 34521 }, { "epoch": 2.721482065431612, "grad_norm": 0.43585339188575745, "learning_rate": 0.0005, "loss": 1.1832735538482666, "step": 34522 }, { "epoch": 2.721560898699251, "grad_norm": 0.43522578477859497, "learning_rate": 0.0005, "loss": 1.2561500072479248, "step": 34523 }, { "epoch": 2.72163973196689, "grad_norm": 0.5493221282958984, "learning_rate": 0.0005, "loss": 1.5743104219436646, "step": 34524 }, { "epoch": 2.721718565234529, "grad_norm": 0.4926927387714386, "learning_rate": 0.0005, "loss": 1.3268799781799316, "step": 34525 }, { "epoch": 2.7217973985021677, "grad_norm": 0.5229392051696777, "learning_rate": 0.0005, "loss": 1.1151745319366455, "step": 34526 }, { "epoch": 2.721876231769807, "grad_norm": 0.5970519781112671, "learning_rate": 0.0005, "loss": 1.3080209493637085, "step": 34527 }, { "epoch": 2.721955065037446, "grad_norm": 0.44358521699905396, "learning_rate": 0.0005, "loss": 1.3999872207641602, "step": 34528 }, { "epoch": 2.7220338983050847, "grad_norm": 0.6948875188827515, "learning_rate": 0.0005, "loss": 1.2610138654708862, "step": 34529 }, { "epoch": 2.722112731572724, "grad_norm": 0.5152396559715271, "learning_rate": 0.0005, "loss": 1.3665838241577148, "step": 34530 }, { "epoch": 2.7221915648403625, "grad_norm": 0.521584153175354, "learning_rate": 0.0005, "loss": 1.447760820388794, "step": 34531 }, { "epoch": 2.7222703981080016, "grad_norm": 0.6212850213050842, "learning_rate": 0.0005, "loss": 1.3868591785430908, "step": 34532 }, { "epoch": 2.7223492313756408, "grad_norm": 0.48943111300468445, "learning_rate": 0.0005, "loss": 1.3104356527328491, "step": 34533 }, { "epoch": 2.7224280646432795, "grad_norm": 0.5397793650627136, "learning_rate": 0.0005, "loss": 1.4794647693634033, "step": 34534 }, { "epoch": 2.722506897910918, "grad_norm": 0.49822190403938293, "learning_rate": 0.0005, "loss": 1.2734150886535645, "step": 34535 }, { "epoch": 2.7225857311785573, "grad_norm": 0.5489893555641174, "learning_rate": 0.0005, "loss": 1.3409655094146729, "step": 34536 }, { "epoch": 2.7226645644461964, "grad_norm": 0.47154656052589417, "learning_rate": 0.0005, "loss": 1.347141981124878, "step": 34537 }, { "epoch": 2.722743397713835, "grad_norm": 0.4183318316936493, "learning_rate": 0.0005, "loss": 1.1868255138397217, "step": 34538 }, { "epoch": 2.7228222309814742, "grad_norm": 0.530474841594696, "learning_rate": 0.0005, "loss": 1.6393629312515259, "step": 34539 }, { "epoch": 2.722901064249113, "grad_norm": 0.6008987426757812, "learning_rate": 0.0005, "loss": 1.5707310438156128, "step": 34540 }, { "epoch": 2.722979897516752, "grad_norm": 0.49100276827812195, "learning_rate": 0.0005, "loss": 1.2434152364730835, "step": 34541 }, { "epoch": 2.723058730784391, "grad_norm": 0.49041998386383057, "learning_rate": 0.0005, "loss": 1.2604845762252808, "step": 34542 }, { "epoch": 2.72313756405203, "grad_norm": 0.4922831952571869, "learning_rate": 0.0005, "loss": 1.188921332359314, "step": 34543 }, { "epoch": 2.723216397319669, "grad_norm": 0.39958640933036804, "learning_rate": 0.0005, "loss": 1.1323356628417969, "step": 34544 }, { "epoch": 2.7232952305873077, "grad_norm": 0.7444862723350525, "learning_rate": 0.0005, "loss": 1.453636646270752, "step": 34545 }, { "epoch": 2.723374063854947, "grad_norm": 0.5109712481498718, "learning_rate": 0.0005, "loss": 1.435793399810791, "step": 34546 }, { "epoch": 2.723452897122586, "grad_norm": 0.49103298783302307, "learning_rate": 0.0005, "loss": 1.4862406253814697, "step": 34547 }, { "epoch": 2.7235317303902247, "grad_norm": 0.63120037317276, "learning_rate": 0.0005, "loss": 1.483951449394226, "step": 34548 }, { "epoch": 2.7236105636578634, "grad_norm": 0.541255533695221, "learning_rate": 0.0005, "loss": 1.4143178462982178, "step": 34549 }, { "epoch": 2.7236893969255025, "grad_norm": 0.5555362105369568, "learning_rate": 0.0005, "loss": 1.2795634269714355, "step": 34550 }, { "epoch": 2.7237682301931416, "grad_norm": 0.43105635046958923, "learning_rate": 0.0005, "loss": 1.1893138885498047, "step": 34551 }, { "epoch": 2.7238470634607803, "grad_norm": 0.5301271677017212, "learning_rate": 0.0005, "loss": 1.2994476556777954, "step": 34552 }, { "epoch": 2.7239258967284194, "grad_norm": 0.43364858627319336, "learning_rate": 0.0005, "loss": 1.341444492340088, "step": 34553 }, { "epoch": 2.724004729996058, "grad_norm": 0.5648687481880188, "learning_rate": 0.0005, "loss": 1.5601434707641602, "step": 34554 }, { "epoch": 2.7240835632636973, "grad_norm": 0.38025790452957153, "learning_rate": 0.0005, "loss": 1.3343758583068848, "step": 34555 }, { "epoch": 2.7241623965313364, "grad_norm": 0.47002723813056946, "learning_rate": 0.0005, "loss": 1.4976017475128174, "step": 34556 }, { "epoch": 2.724241229798975, "grad_norm": 0.46433281898498535, "learning_rate": 0.0005, "loss": 1.3066840171813965, "step": 34557 }, { "epoch": 2.7243200630666142, "grad_norm": 0.47189009189605713, "learning_rate": 0.0005, "loss": 1.4405794143676758, "step": 34558 }, { "epoch": 2.724398896334253, "grad_norm": 0.5360025763511658, "learning_rate": 0.0005, "loss": 1.4113508462905884, "step": 34559 }, { "epoch": 2.724477729601892, "grad_norm": 12.92783260345459, "learning_rate": 0.0005, "loss": 1.3518173694610596, "step": 34560 }, { "epoch": 2.724556562869531, "grad_norm": 0.532729983329773, "learning_rate": 0.0005, "loss": 1.607815146446228, "step": 34561 }, { "epoch": 2.72463539613717, "grad_norm": 0.5926269292831421, "learning_rate": 0.0005, "loss": 1.5243850946426392, "step": 34562 }, { "epoch": 2.7247142294048086, "grad_norm": 0.48273494839668274, "learning_rate": 0.0005, "loss": 1.2300134897232056, "step": 34563 }, { "epoch": 2.7247930626724477, "grad_norm": 0.48877236247062683, "learning_rate": 0.0005, "loss": 1.225543737411499, "step": 34564 }, { "epoch": 2.724871895940087, "grad_norm": 0.598260223865509, "learning_rate": 0.0005, "loss": 1.4184378385543823, "step": 34565 }, { "epoch": 2.7249507292077255, "grad_norm": 0.5114186406135559, "learning_rate": 0.0005, "loss": 1.4286353588104248, "step": 34566 }, { "epoch": 2.7250295624753647, "grad_norm": 0.4929168224334717, "learning_rate": 0.0005, "loss": 1.319930911064148, "step": 34567 }, { "epoch": 2.7251083957430033, "grad_norm": 0.5464309453964233, "learning_rate": 0.0005, "loss": 1.565746545791626, "step": 34568 }, { "epoch": 2.7251872290106425, "grad_norm": 0.5192335247993469, "learning_rate": 0.0005, "loss": 1.3696856498718262, "step": 34569 }, { "epoch": 2.7252660622782816, "grad_norm": 0.47441136837005615, "learning_rate": 0.0005, "loss": 1.3674427270889282, "step": 34570 }, { "epoch": 2.7253448955459203, "grad_norm": 0.6936952471733093, "learning_rate": 0.0005, "loss": 1.5754345655441284, "step": 34571 }, { "epoch": 2.7254237288135594, "grad_norm": 0.39392685890197754, "learning_rate": 0.0005, "loss": 1.1965396404266357, "step": 34572 }, { "epoch": 2.725502562081198, "grad_norm": 0.5605627298355103, "learning_rate": 0.0005, "loss": 1.396132469177246, "step": 34573 }, { "epoch": 2.7255813953488373, "grad_norm": 0.49425947666168213, "learning_rate": 0.0005, "loss": 1.2633227109909058, "step": 34574 }, { "epoch": 2.7256602286164764, "grad_norm": 0.48861581087112427, "learning_rate": 0.0005, "loss": 1.2351093292236328, "step": 34575 }, { "epoch": 2.725739061884115, "grad_norm": 0.5581210851669312, "learning_rate": 0.0005, "loss": 1.2784675359725952, "step": 34576 }, { "epoch": 2.7258178951517538, "grad_norm": 0.6908870339393616, "learning_rate": 0.0005, "loss": 1.6224499940872192, "step": 34577 }, { "epoch": 2.725896728419393, "grad_norm": 0.46638408303260803, "learning_rate": 0.0005, "loss": 1.270986795425415, "step": 34578 }, { "epoch": 2.725975561687032, "grad_norm": 0.4545826017856598, "learning_rate": 0.0005, "loss": 1.3506834506988525, "step": 34579 }, { "epoch": 2.7260543949546707, "grad_norm": 0.41275468468666077, "learning_rate": 0.0005, "loss": 1.2078555822372437, "step": 34580 }, { "epoch": 2.72613322822231, "grad_norm": 0.44995203614234924, "learning_rate": 0.0005, "loss": 1.427748441696167, "step": 34581 }, { "epoch": 2.7262120614899485, "grad_norm": 0.5924654603004456, "learning_rate": 0.0005, "loss": 1.350082516670227, "step": 34582 }, { "epoch": 2.7262908947575877, "grad_norm": 0.4602229595184326, "learning_rate": 0.0005, "loss": 1.310196876525879, "step": 34583 }, { "epoch": 2.726369728025227, "grad_norm": 0.5247877240180969, "learning_rate": 0.0005, "loss": 0.9564071893692017, "step": 34584 }, { "epoch": 2.7264485612928655, "grad_norm": 0.8015108704566956, "learning_rate": 0.0005, "loss": 1.236412763595581, "step": 34585 }, { "epoch": 2.7265273945605046, "grad_norm": 0.5025275349617004, "learning_rate": 0.0005, "loss": 1.3577581644058228, "step": 34586 }, { "epoch": 2.7266062278281433, "grad_norm": 0.5696719288825989, "learning_rate": 0.0005, "loss": 1.4415241479873657, "step": 34587 }, { "epoch": 2.7266850610957825, "grad_norm": 0.5131023526191711, "learning_rate": 0.0005, "loss": 1.3498042821884155, "step": 34588 }, { "epoch": 2.7267638943634216, "grad_norm": 0.3941359519958496, "learning_rate": 0.0005, "loss": 1.2937010526657104, "step": 34589 }, { "epoch": 2.7268427276310603, "grad_norm": 0.5517815351486206, "learning_rate": 0.0005, "loss": 1.2269494533538818, "step": 34590 }, { "epoch": 2.7269215608986994, "grad_norm": 0.6758432388305664, "learning_rate": 0.0005, "loss": 1.331673264503479, "step": 34591 }, { "epoch": 2.727000394166338, "grad_norm": 0.4667389392852783, "learning_rate": 0.0005, "loss": 1.2601515054702759, "step": 34592 }, { "epoch": 2.7270792274339772, "grad_norm": 0.4874924123287201, "learning_rate": 0.0005, "loss": 1.282086968421936, "step": 34593 }, { "epoch": 2.727158060701616, "grad_norm": 0.5764530897140503, "learning_rate": 0.0005, "loss": 1.3850980997085571, "step": 34594 }, { "epoch": 2.727236893969255, "grad_norm": 0.44419577717781067, "learning_rate": 0.0005, "loss": 1.08711576461792, "step": 34595 }, { "epoch": 2.7273157272368937, "grad_norm": 0.5341799259185791, "learning_rate": 0.0005, "loss": 1.3430135250091553, "step": 34596 }, { "epoch": 2.727394560504533, "grad_norm": 0.39431485533714294, "learning_rate": 0.0005, "loss": 1.143636703491211, "step": 34597 }, { "epoch": 2.727473393772172, "grad_norm": 0.5213261842727661, "learning_rate": 0.0005, "loss": 1.4436991214752197, "step": 34598 }, { "epoch": 2.7275522270398107, "grad_norm": 0.47864753007888794, "learning_rate": 0.0005, "loss": 1.131453037261963, "step": 34599 }, { "epoch": 2.72763106030745, "grad_norm": 0.5777791738510132, "learning_rate": 0.0005, "loss": 1.5404255390167236, "step": 34600 }, { "epoch": 2.7277098935750885, "grad_norm": 0.5668500661849976, "learning_rate": 0.0005, "loss": 1.5040500164031982, "step": 34601 }, { "epoch": 2.7277887268427277, "grad_norm": 0.4716678261756897, "learning_rate": 0.0005, "loss": 1.3926434516906738, "step": 34602 }, { "epoch": 2.727867560110367, "grad_norm": 0.41620030999183655, "learning_rate": 0.0005, "loss": 1.422330379486084, "step": 34603 }, { "epoch": 2.7279463933780055, "grad_norm": 0.550383448600769, "learning_rate": 0.0005, "loss": 1.4840035438537598, "step": 34604 }, { "epoch": 2.7280252266456446, "grad_norm": 0.5056126117706299, "learning_rate": 0.0005, "loss": 1.2258775234222412, "step": 34605 }, { "epoch": 2.7281040599132833, "grad_norm": 0.49030208587646484, "learning_rate": 0.0005, "loss": 1.3449287414550781, "step": 34606 }, { "epoch": 2.7281828931809224, "grad_norm": 0.5866938829421997, "learning_rate": 0.0005, "loss": 1.3235082626342773, "step": 34607 }, { "epoch": 2.728261726448561, "grad_norm": 0.48396050930023193, "learning_rate": 0.0005, "loss": 1.3685824871063232, "step": 34608 }, { "epoch": 2.7283405597162003, "grad_norm": 0.4838273227214813, "learning_rate": 0.0005, "loss": 1.1472852230072021, "step": 34609 }, { "epoch": 2.728419392983839, "grad_norm": 0.499569296836853, "learning_rate": 0.0005, "loss": 1.389686107635498, "step": 34610 }, { "epoch": 2.728498226251478, "grad_norm": 0.6133782267570496, "learning_rate": 0.0005, "loss": 1.2828099727630615, "step": 34611 }, { "epoch": 2.728577059519117, "grad_norm": 0.5432180166244507, "learning_rate": 0.0005, "loss": 1.4145903587341309, "step": 34612 }, { "epoch": 2.728655892786756, "grad_norm": 0.6862429976463318, "learning_rate": 0.0005, "loss": 1.2700409889221191, "step": 34613 }, { "epoch": 2.728734726054395, "grad_norm": 0.5704070925712585, "learning_rate": 0.0005, "loss": 1.5037142038345337, "step": 34614 }, { "epoch": 2.7288135593220337, "grad_norm": 0.5230950117111206, "learning_rate": 0.0005, "loss": 1.2474329471588135, "step": 34615 }, { "epoch": 2.728892392589673, "grad_norm": 0.4942558705806732, "learning_rate": 0.0005, "loss": 1.2841943502426147, "step": 34616 }, { "epoch": 2.728971225857312, "grad_norm": 0.4576759934425354, "learning_rate": 0.0005, "loss": 1.4601645469665527, "step": 34617 }, { "epoch": 2.7290500591249507, "grad_norm": 0.568053126335144, "learning_rate": 0.0005, "loss": 1.4675281047821045, "step": 34618 }, { "epoch": 2.72912889239259, "grad_norm": 0.46500515937805176, "learning_rate": 0.0005, "loss": 1.3435701131820679, "step": 34619 }, { "epoch": 2.7292077256602285, "grad_norm": 0.6543867588043213, "learning_rate": 0.0005, "loss": 1.286767601966858, "step": 34620 }, { "epoch": 2.7292865589278676, "grad_norm": 0.44562530517578125, "learning_rate": 0.0005, "loss": 1.1223068237304688, "step": 34621 }, { "epoch": 2.729365392195507, "grad_norm": 0.4413242042064667, "learning_rate": 0.0005, "loss": 1.1857272386550903, "step": 34622 }, { "epoch": 2.7294442254631455, "grad_norm": 0.7062446475028992, "learning_rate": 0.0005, "loss": 1.3078460693359375, "step": 34623 }, { "epoch": 2.729523058730784, "grad_norm": 0.5077981948852539, "learning_rate": 0.0005, "loss": 1.406598687171936, "step": 34624 }, { "epoch": 2.7296018919984233, "grad_norm": 0.5019273161888123, "learning_rate": 0.0005, "loss": 1.2530934810638428, "step": 34625 }, { "epoch": 2.7296807252660624, "grad_norm": 0.5044022798538208, "learning_rate": 0.0005, "loss": 1.3815720081329346, "step": 34626 }, { "epoch": 2.729759558533701, "grad_norm": 0.5680587291717529, "learning_rate": 0.0005, "loss": 1.4715360403060913, "step": 34627 }, { "epoch": 2.7298383918013402, "grad_norm": 0.5685674548149109, "learning_rate": 0.0005, "loss": 1.3379182815551758, "step": 34628 }, { "epoch": 2.729917225068979, "grad_norm": 0.49654483795166016, "learning_rate": 0.0005, "loss": 1.2645759582519531, "step": 34629 }, { "epoch": 2.729996058336618, "grad_norm": 0.4417024552822113, "learning_rate": 0.0005, "loss": 1.284379482269287, "step": 34630 }, { "epoch": 2.730074891604257, "grad_norm": 0.5720497965812683, "learning_rate": 0.0005, "loss": 1.3635262250900269, "step": 34631 }, { "epoch": 2.730153724871896, "grad_norm": 0.47761473059654236, "learning_rate": 0.0005, "loss": 1.2348757982254028, "step": 34632 }, { "epoch": 2.730232558139535, "grad_norm": 0.5969586968421936, "learning_rate": 0.0005, "loss": 1.3418631553649902, "step": 34633 }, { "epoch": 2.7303113914071737, "grad_norm": 0.4789227247238159, "learning_rate": 0.0005, "loss": 1.4363946914672852, "step": 34634 }, { "epoch": 2.730390224674813, "grad_norm": 0.5935055613517761, "learning_rate": 0.0005, "loss": 1.2148139476776123, "step": 34635 }, { "epoch": 2.730469057942452, "grad_norm": 0.5257623195648193, "learning_rate": 0.0005, "loss": 1.3768047094345093, "step": 34636 }, { "epoch": 2.7305478912100907, "grad_norm": 0.3996754288673401, "learning_rate": 0.0005, "loss": 1.1891521215438843, "step": 34637 }, { "epoch": 2.7306267244777294, "grad_norm": 0.8805487155914307, "learning_rate": 0.0005, "loss": 1.333930492401123, "step": 34638 }, { "epoch": 2.7307055577453685, "grad_norm": 0.49756285548210144, "learning_rate": 0.0005, "loss": 1.3805980682373047, "step": 34639 }, { "epoch": 2.7307843910130076, "grad_norm": 0.4894419014453888, "learning_rate": 0.0005, "loss": 1.1819109916687012, "step": 34640 }, { "epoch": 2.7308632242806463, "grad_norm": 0.563351571559906, "learning_rate": 0.0005, "loss": 1.304219365119934, "step": 34641 }, { "epoch": 2.7309420575482855, "grad_norm": 0.39694827795028687, "learning_rate": 0.0005, "loss": 1.3235290050506592, "step": 34642 }, { "epoch": 2.731020890815924, "grad_norm": 0.5135737061500549, "learning_rate": 0.0005, "loss": 1.5061523914337158, "step": 34643 }, { "epoch": 2.7310997240835633, "grad_norm": 0.5196470618247986, "learning_rate": 0.0005, "loss": 1.417353630065918, "step": 34644 }, { "epoch": 2.7311785573512024, "grad_norm": 0.9045632481575012, "learning_rate": 0.0005, "loss": 1.578389286994934, "step": 34645 }, { "epoch": 2.731257390618841, "grad_norm": 0.4952940344810486, "learning_rate": 0.0005, "loss": 1.5551154613494873, "step": 34646 }, { "epoch": 2.7313362238864802, "grad_norm": 0.48581811785697937, "learning_rate": 0.0005, "loss": 1.226654052734375, "step": 34647 }, { "epoch": 2.731415057154119, "grad_norm": 0.5583526492118835, "learning_rate": 0.0005, "loss": 1.0698020458221436, "step": 34648 }, { "epoch": 2.731493890421758, "grad_norm": 0.5236606597900391, "learning_rate": 0.0005, "loss": 1.3921701908111572, "step": 34649 }, { "epoch": 2.731572723689397, "grad_norm": 0.4431813359260559, "learning_rate": 0.0005, "loss": 1.3207039833068848, "step": 34650 }, { "epoch": 2.731651556957036, "grad_norm": 0.45107799768447876, "learning_rate": 0.0005, "loss": 1.3474745750427246, "step": 34651 }, { "epoch": 2.7317303902246746, "grad_norm": 0.436102956533432, "learning_rate": 0.0005, "loss": 1.2777377367019653, "step": 34652 }, { "epoch": 2.7318092234923137, "grad_norm": 0.44511905312538147, "learning_rate": 0.0005, "loss": 1.4490351676940918, "step": 34653 }, { "epoch": 2.731888056759953, "grad_norm": 0.6205989122390747, "learning_rate": 0.0005, "loss": 1.623331069946289, "step": 34654 }, { "epoch": 2.7319668900275915, "grad_norm": 0.5175864100456238, "learning_rate": 0.0005, "loss": 1.3361761569976807, "step": 34655 }, { "epoch": 2.7320457232952307, "grad_norm": 0.4432729184627533, "learning_rate": 0.0005, "loss": 1.3986361026763916, "step": 34656 }, { "epoch": 2.7321245565628693, "grad_norm": 0.43246740102767944, "learning_rate": 0.0005, "loss": 1.3252482414245605, "step": 34657 }, { "epoch": 2.7322033898305085, "grad_norm": 0.6308184266090393, "learning_rate": 0.0005, "loss": 1.4923033714294434, "step": 34658 }, { "epoch": 2.7322822230981476, "grad_norm": 0.4356130361557007, "learning_rate": 0.0005, "loss": 1.2521289587020874, "step": 34659 }, { "epoch": 2.7323610563657863, "grad_norm": 0.3619987368583679, "learning_rate": 0.0005, "loss": 1.2149333953857422, "step": 34660 }, { "epoch": 2.7324398896334254, "grad_norm": 0.4094334840774536, "learning_rate": 0.0005, "loss": 1.176759958267212, "step": 34661 }, { "epoch": 2.732518722901064, "grad_norm": 0.4496447741985321, "learning_rate": 0.0005, "loss": 1.364707589149475, "step": 34662 }, { "epoch": 2.7325975561687033, "grad_norm": 0.605646550655365, "learning_rate": 0.0005, "loss": 1.5448799133300781, "step": 34663 }, { "epoch": 2.7326763894363424, "grad_norm": 0.48750120401382446, "learning_rate": 0.0005, "loss": 1.2474050521850586, "step": 34664 }, { "epoch": 2.732755222703981, "grad_norm": 0.5747161507606506, "learning_rate": 0.0005, "loss": 1.500691294670105, "step": 34665 }, { "epoch": 2.7328340559716198, "grad_norm": 0.5207899808883667, "learning_rate": 0.0005, "loss": 1.3671867847442627, "step": 34666 }, { "epoch": 2.732912889239259, "grad_norm": 0.5830774307250977, "learning_rate": 0.0005, "loss": 1.321664571762085, "step": 34667 }, { "epoch": 2.732991722506898, "grad_norm": 0.6869755983352661, "learning_rate": 0.0005, "loss": 1.5112885236740112, "step": 34668 }, { "epoch": 2.7330705557745367, "grad_norm": 0.48444491624832153, "learning_rate": 0.0005, "loss": 1.3915128707885742, "step": 34669 }, { "epoch": 2.733149389042176, "grad_norm": 0.5766134858131409, "learning_rate": 0.0005, "loss": 1.3323678970336914, "step": 34670 }, { "epoch": 2.7332282223098145, "grad_norm": 0.4290318787097931, "learning_rate": 0.0005, "loss": 1.1482734680175781, "step": 34671 }, { "epoch": 2.7333070555774537, "grad_norm": 0.44797828793525696, "learning_rate": 0.0005, "loss": 1.2812745571136475, "step": 34672 }, { "epoch": 2.733385888845093, "grad_norm": 0.45874929428100586, "learning_rate": 0.0005, "loss": 1.469680905342102, "step": 34673 }, { "epoch": 2.7334647221127315, "grad_norm": 0.46702978014945984, "learning_rate": 0.0005, "loss": 1.2602590322494507, "step": 34674 }, { "epoch": 2.7335435553803706, "grad_norm": 1.7148209810256958, "learning_rate": 0.0005, "loss": 1.3141398429870605, "step": 34675 }, { "epoch": 2.7336223886480093, "grad_norm": 0.4004421830177307, "learning_rate": 0.0005, "loss": 1.1726311445236206, "step": 34676 }, { "epoch": 2.7337012219156485, "grad_norm": 0.5038810968399048, "learning_rate": 0.0005, "loss": 1.2913141250610352, "step": 34677 }, { "epoch": 2.7337800551832876, "grad_norm": 0.5583763718605042, "learning_rate": 0.0005, "loss": 1.3473849296569824, "step": 34678 }, { "epoch": 2.7338588884509263, "grad_norm": 0.48114845156669617, "learning_rate": 0.0005, "loss": 1.341178297996521, "step": 34679 }, { "epoch": 2.733937721718565, "grad_norm": 0.40841832756996155, "learning_rate": 0.0005, "loss": 1.1686395406723022, "step": 34680 }, { "epoch": 2.734016554986204, "grad_norm": 0.5455604195594788, "learning_rate": 0.0005, "loss": 1.4042730331420898, "step": 34681 }, { "epoch": 2.7340953882538432, "grad_norm": 0.5566330552101135, "learning_rate": 0.0005, "loss": 1.278278112411499, "step": 34682 }, { "epoch": 2.734174221521482, "grad_norm": 0.5138536095619202, "learning_rate": 0.0005, "loss": 1.252249002456665, "step": 34683 }, { "epoch": 2.734253054789121, "grad_norm": 0.6608846783638, "learning_rate": 0.0005, "loss": 1.247624158859253, "step": 34684 }, { "epoch": 2.7343318880567598, "grad_norm": 0.5734183192253113, "learning_rate": 0.0005, "loss": 1.296412467956543, "step": 34685 }, { "epoch": 2.734410721324399, "grad_norm": 0.5699942111968994, "learning_rate": 0.0005, "loss": 1.291527271270752, "step": 34686 }, { "epoch": 2.734489554592038, "grad_norm": 0.444538950920105, "learning_rate": 0.0005, "loss": 1.1121728420257568, "step": 34687 }, { "epoch": 2.7345683878596767, "grad_norm": 0.5062187910079956, "learning_rate": 0.0005, "loss": 1.3998990058898926, "step": 34688 }, { "epoch": 2.734647221127316, "grad_norm": 0.5252258777618408, "learning_rate": 0.0005, "loss": 1.4239239692687988, "step": 34689 }, { "epoch": 2.7347260543949545, "grad_norm": 0.47707369923591614, "learning_rate": 0.0005, "loss": 1.4028820991516113, "step": 34690 }, { "epoch": 2.7348048876625937, "grad_norm": 0.4821889400482178, "learning_rate": 0.0005, "loss": 1.4391648769378662, "step": 34691 }, { "epoch": 2.734883720930233, "grad_norm": 0.4075077474117279, "learning_rate": 0.0005, "loss": 1.1682606935501099, "step": 34692 }, { "epoch": 2.7349625541978715, "grad_norm": 0.4434376358985901, "learning_rate": 0.0005, "loss": 1.3487411737442017, "step": 34693 }, { "epoch": 2.73504138746551, "grad_norm": 0.7575255036354065, "learning_rate": 0.0005, "loss": 1.4511795043945312, "step": 34694 }, { "epoch": 2.7351202207331493, "grad_norm": 0.43724265694618225, "learning_rate": 0.0005, "loss": 1.1240242719650269, "step": 34695 }, { "epoch": 2.7351990540007884, "grad_norm": 0.4194769263267517, "learning_rate": 0.0005, "loss": 1.1632097959518433, "step": 34696 }, { "epoch": 2.735277887268427, "grad_norm": 0.6663065552711487, "learning_rate": 0.0005, "loss": 1.6323868036270142, "step": 34697 }, { "epoch": 2.7353567205360663, "grad_norm": 0.4553559422492981, "learning_rate": 0.0005, "loss": 1.418558120727539, "step": 34698 }, { "epoch": 2.735435553803705, "grad_norm": 0.4471580982208252, "learning_rate": 0.0005, "loss": 1.1923730373382568, "step": 34699 }, { "epoch": 2.735514387071344, "grad_norm": 0.49134770035743713, "learning_rate": 0.0005, "loss": 1.364731788635254, "step": 34700 }, { "epoch": 2.7355932203389832, "grad_norm": 0.5321424007415771, "learning_rate": 0.0005, "loss": 1.2867918014526367, "step": 34701 }, { "epoch": 2.735672053606622, "grad_norm": 0.5330973863601685, "learning_rate": 0.0005, "loss": 1.3372166156768799, "step": 34702 }, { "epoch": 2.735750886874261, "grad_norm": 0.49329015612602234, "learning_rate": 0.0005, "loss": 1.5235357284545898, "step": 34703 }, { "epoch": 2.7358297201418997, "grad_norm": 0.44210150837898254, "learning_rate": 0.0005, "loss": 1.188318133354187, "step": 34704 }, { "epoch": 2.735908553409539, "grad_norm": 0.638064444065094, "learning_rate": 0.0005, "loss": 1.3948109149932861, "step": 34705 }, { "epoch": 2.735987386677178, "grad_norm": 0.5398856997489929, "learning_rate": 0.0005, "loss": 1.5373542308807373, "step": 34706 }, { "epoch": 2.7360662199448167, "grad_norm": 0.5115448236465454, "learning_rate": 0.0005, "loss": 1.271331548690796, "step": 34707 }, { "epoch": 2.7361450532124554, "grad_norm": 0.46641021966934204, "learning_rate": 0.0005, "loss": 1.4149152040481567, "step": 34708 }, { "epoch": 2.7362238864800945, "grad_norm": 0.4927118122577667, "learning_rate": 0.0005, "loss": 1.4150755405426025, "step": 34709 }, { "epoch": 2.7363027197477336, "grad_norm": 0.44537779688835144, "learning_rate": 0.0005, "loss": 1.2691099643707275, "step": 34710 }, { "epoch": 2.7363815530153723, "grad_norm": 0.6705199480056763, "learning_rate": 0.0005, "loss": 1.4177318811416626, "step": 34711 }, { "epoch": 2.7364603862830115, "grad_norm": 0.5863696932792664, "learning_rate": 0.0005, "loss": 1.5134646892547607, "step": 34712 }, { "epoch": 2.73653921955065, "grad_norm": 0.4009060859680176, "learning_rate": 0.0005, "loss": 1.2754182815551758, "step": 34713 }, { "epoch": 2.7366180528182893, "grad_norm": 0.5625416040420532, "learning_rate": 0.0005, "loss": 1.4605261087417603, "step": 34714 }, { "epoch": 2.7366968860859284, "grad_norm": 0.46339479088783264, "learning_rate": 0.0005, "loss": 1.1615753173828125, "step": 34715 }, { "epoch": 2.736775719353567, "grad_norm": 0.5119382739067078, "learning_rate": 0.0005, "loss": 1.3902850151062012, "step": 34716 }, { "epoch": 2.7368545526212062, "grad_norm": 0.4795725345611572, "learning_rate": 0.0005, "loss": 1.3588624000549316, "step": 34717 }, { "epoch": 2.736933385888845, "grad_norm": 0.5135520696640015, "learning_rate": 0.0005, "loss": 1.2069621086120605, "step": 34718 }, { "epoch": 2.737012219156484, "grad_norm": 0.5000767707824707, "learning_rate": 0.0005, "loss": 1.2755794525146484, "step": 34719 }, { "epoch": 2.737091052424123, "grad_norm": 0.5233899354934692, "learning_rate": 0.0005, "loss": 1.2174280881881714, "step": 34720 }, { "epoch": 2.737169885691762, "grad_norm": 0.582381010055542, "learning_rate": 0.0005, "loss": 1.3953258991241455, "step": 34721 }, { "epoch": 2.737248718959401, "grad_norm": 0.7674686908721924, "learning_rate": 0.0005, "loss": 1.3320380449295044, "step": 34722 }, { "epoch": 2.7373275522270397, "grad_norm": 0.5087751746177673, "learning_rate": 0.0005, "loss": 1.4303960800170898, "step": 34723 }, { "epoch": 2.737406385494679, "grad_norm": 0.5695871710777283, "learning_rate": 0.0005, "loss": 1.409632921218872, "step": 34724 }, { "epoch": 2.7374852187623175, "grad_norm": 0.47948843240737915, "learning_rate": 0.0005, "loss": 1.1664035320281982, "step": 34725 }, { "epoch": 2.7375640520299567, "grad_norm": 0.5093125104904175, "learning_rate": 0.0005, "loss": 1.4157640933990479, "step": 34726 }, { "epoch": 2.7376428852975954, "grad_norm": 0.5551788210868835, "learning_rate": 0.0005, "loss": 1.1297519207000732, "step": 34727 }, { "epoch": 2.7377217185652345, "grad_norm": 0.4684425890445709, "learning_rate": 0.0005, "loss": 1.3817625045776367, "step": 34728 }, { "epoch": 2.7378005518328736, "grad_norm": 0.5653771758079529, "learning_rate": 0.0005, "loss": 1.2957706451416016, "step": 34729 }, { "epoch": 2.7378793851005123, "grad_norm": 0.4942554831504822, "learning_rate": 0.0005, "loss": 1.244481086730957, "step": 34730 }, { "epoch": 2.7379582183681515, "grad_norm": 0.583638608455658, "learning_rate": 0.0005, "loss": 1.1919782161712646, "step": 34731 }, { "epoch": 2.73803705163579, "grad_norm": 0.41056182980537415, "learning_rate": 0.0005, "loss": 1.1291968822479248, "step": 34732 }, { "epoch": 2.7381158849034293, "grad_norm": 0.4924364387989044, "learning_rate": 0.0005, "loss": 1.3943500518798828, "step": 34733 }, { "epoch": 2.7381947181710684, "grad_norm": 0.5153864026069641, "learning_rate": 0.0005, "loss": 1.3576383590698242, "step": 34734 }, { "epoch": 2.738273551438707, "grad_norm": 0.48035600781440735, "learning_rate": 0.0005, "loss": 1.3440343141555786, "step": 34735 }, { "epoch": 2.7383523847063462, "grad_norm": 0.4199031591415405, "learning_rate": 0.0005, "loss": 1.2232375144958496, "step": 34736 }, { "epoch": 2.738431217973985, "grad_norm": 0.5800814628601074, "learning_rate": 0.0005, "loss": 1.537898302078247, "step": 34737 }, { "epoch": 2.738510051241624, "grad_norm": 0.44000133872032166, "learning_rate": 0.0005, "loss": 1.2150371074676514, "step": 34738 }, { "epoch": 2.7385888845092627, "grad_norm": 0.500119686126709, "learning_rate": 0.0005, "loss": 1.28327214717865, "step": 34739 }, { "epoch": 2.738667717776902, "grad_norm": 0.45914924144744873, "learning_rate": 0.0005, "loss": 1.3876779079437256, "step": 34740 }, { "epoch": 2.7387465510445406, "grad_norm": 0.6719428896903992, "learning_rate": 0.0005, "loss": 1.3339650630950928, "step": 34741 }, { "epoch": 2.7388253843121797, "grad_norm": 0.7300020456314087, "learning_rate": 0.0005, "loss": 1.3080642223358154, "step": 34742 }, { "epoch": 2.738904217579819, "grad_norm": 0.5117443799972534, "learning_rate": 0.0005, "loss": 1.3342289924621582, "step": 34743 }, { "epoch": 2.7389830508474575, "grad_norm": 0.5934880375862122, "learning_rate": 0.0005, "loss": 1.4600657224655151, "step": 34744 }, { "epoch": 2.7390618841150967, "grad_norm": 0.47933530807495117, "learning_rate": 0.0005, "loss": 1.2958576679229736, "step": 34745 }, { "epoch": 2.7391407173827353, "grad_norm": 0.5055359601974487, "learning_rate": 0.0005, "loss": 1.3553457260131836, "step": 34746 }, { "epoch": 2.7392195506503745, "grad_norm": 0.471734881401062, "learning_rate": 0.0005, "loss": 1.3117059469223022, "step": 34747 }, { "epoch": 2.7392983839180136, "grad_norm": 0.7461965680122375, "learning_rate": 0.0005, "loss": 1.4833412170410156, "step": 34748 }, { "epoch": 2.7393772171856523, "grad_norm": 0.5285799503326416, "learning_rate": 0.0005, "loss": 1.4426708221435547, "step": 34749 }, { "epoch": 2.7394560504532914, "grad_norm": 0.4586803615093231, "learning_rate": 0.0005, "loss": 1.1603997945785522, "step": 34750 }, { "epoch": 2.73953488372093, "grad_norm": 0.5876525044441223, "learning_rate": 0.0005, "loss": 1.4608933925628662, "step": 34751 }, { "epoch": 2.7396137169885693, "grad_norm": 0.9025155305862427, "learning_rate": 0.0005, "loss": 1.513357400894165, "step": 34752 }, { "epoch": 2.7396925502562084, "grad_norm": 0.5045576691627502, "learning_rate": 0.0005, "loss": 1.5027744770050049, "step": 34753 }, { "epoch": 2.739771383523847, "grad_norm": 0.5020776391029358, "learning_rate": 0.0005, "loss": 1.4077500104904175, "step": 34754 }, { "epoch": 2.7398502167914858, "grad_norm": 0.7016099691390991, "learning_rate": 0.0005, "loss": 1.2910946607589722, "step": 34755 }, { "epoch": 2.739929050059125, "grad_norm": 0.6379935145378113, "learning_rate": 0.0005, "loss": 1.4438767433166504, "step": 34756 }, { "epoch": 2.740007883326764, "grad_norm": 0.46577680110931396, "learning_rate": 0.0005, "loss": 1.1189600229263306, "step": 34757 }, { "epoch": 2.7400867165944027, "grad_norm": 0.48773226141929626, "learning_rate": 0.0005, "loss": 1.3539780378341675, "step": 34758 }, { "epoch": 2.740165549862042, "grad_norm": 0.5478967428207397, "learning_rate": 0.0005, "loss": 1.3393158912658691, "step": 34759 }, { "epoch": 2.7402443831296805, "grad_norm": 0.53985995054245, "learning_rate": 0.0005, "loss": 1.2393622398376465, "step": 34760 }, { "epoch": 2.7403232163973197, "grad_norm": 0.5323547124862671, "learning_rate": 0.0005, "loss": 1.3999528884887695, "step": 34761 }, { "epoch": 2.740402049664959, "grad_norm": 0.5403212904930115, "learning_rate": 0.0005, "loss": 1.5277811288833618, "step": 34762 }, { "epoch": 2.7404808829325975, "grad_norm": 0.4995313286781311, "learning_rate": 0.0005, "loss": 1.3003571033477783, "step": 34763 }, { "epoch": 2.7405597162002366, "grad_norm": 0.5082274079322815, "learning_rate": 0.0005, "loss": 1.4658482074737549, "step": 34764 }, { "epoch": 2.7406385494678753, "grad_norm": 0.4981358051300049, "learning_rate": 0.0005, "loss": 1.3131369352340698, "step": 34765 }, { "epoch": 2.7407173827355145, "grad_norm": 0.6755357384681702, "learning_rate": 0.0005, "loss": 1.574889898300171, "step": 34766 }, { "epoch": 2.7407962160031536, "grad_norm": 0.44797444343566895, "learning_rate": 0.0005, "loss": 1.1374735832214355, "step": 34767 }, { "epoch": 2.7408750492707923, "grad_norm": 0.5044108033180237, "learning_rate": 0.0005, "loss": 1.4062180519104004, "step": 34768 }, { "epoch": 2.740953882538431, "grad_norm": 0.509319007396698, "learning_rate": 0.0005, "loss": 1.555558681488037, "step": 34769 }, { "epoch": 2.74103271580607, "grad_norm": 0.6318219900131226, "learning_rate": 0.0005, "loss": 1.4867990016937256, "step": 34770 }, { "epoch": 2.7411115490737092, "grad_norm": 0.43734869360923767, "learning_rate": 0.0005, "loss": 1.3581700325012207, "step": 34771 }, { "epoch": 2.741190382341348, "grad_norm": 0.5194981098175049, "learning_rate": 0.0005, "loss": 1.368574857711792, "step": 34772 }, { "epoch": 2.741269215608987, "grad_norm": 0.5165910124778748, "learning_rate": 0.0005, "loss": 1.2568299770355225, "step": 34773 }, { "epoch": 2.7413480488766258, "grad_norm": 0.494204044342041, "learning_rate": 0.0005, "loss": 1.1989178657531738, "step": 34774 }, { "epoch": 2.741426882144265, "grad_norm": 0.4027180075645447, "learning_rate": 0.0005, "loss": 1.262819766998291, "step": 34775 }, { "epoch": 2.741505715411904, "grad_norm": 0.5051065683364868, "learning_rate": 0.0005, "loss": 1.3008742332458496, "step": 34776 }, { "epoch": 2.7415845486795427, "grad_norm": 0.49351003766059875, "learning_rate": 0.0005, "loss": 1.3265349864959717, "step": 34777 }, { "epoch": 2.741663381947182, "grad_norm": 0.5481879711151123, "learning_rate": 0.0005, "loss": 1.3945908546447754, "step": 34778 }, { "epoch": 2.7417422152148205, "grad_norm": 0.5713438391685486, "learning_rate": 0.0005, "loss": 1.3369150161743164, "step": 34779 }, { "epoch": 2.7418210484824597, "grad_norm": 0.5812419056892395, "learning_rate": 0.0005, "loss": 1.4158347845077515, "step": 34780 }, { "epoch": 2.741899881750099, "grad_norm": 0.5248993635177612, "learning_rate": 0.0005, "loss": 1.2421798706054688, "step": 34781 }, { "epoch": 2.7419787150177375, "grad_norm": 0.5767098069190979, "learning_rate": 0.0005, "loss": 1.490126132965088, "step": 34782 }, { "epoch": 2.742057548285376, "grad_norm": 0.5529601573944092, "learning_rate": 0.0005, "loss": 1.4415339231491089, "step": 34783 }, { "epoch": 2.7421363815530153, "grad_norm": 0.6464492082595825, "learning_rate": 0.0005, "loss": 1.4177594184875488, "step": 34784 }, { "epoch": 2.7422152148206544, "grad_norm": 0.5053641200065613, "learning_rate": 0.0005, "loss": 1.5251429080963135, "step": 34785 }, { "epoch": 2.742294048088293, "grad_norm": 0.412898451089859, "learning_rate": 0.0005, "loss": 1.1718047857284546, "step": 34786 }, { "epoch": 2.7423728813559323, "grad_norm": 0.6012570261955261, "learning_rate": 0.0005, "loss": 1.501517415046692, "step": 34787 }, { "epoch": 2.742451714623571, "grad_norm": 0.7073889970779419, "learning_rate": 0.0005, "loss": 1.4106640815734863, "step": 34788 }, { "epoch": 2.74253054789121, "grad_norm": 0.5808502435684204, "learning_rate": 0.0005, "loss": 1.2840280532836914, "step": 34789 }, { "epoch": 2.7426093811588492, "grad_norm": 0.47842806577682495, "learning_rate": 0.0005, "loss": 1.6039518117904663, "step": 34790 }, { "epoch": 2.742688214426488, "grad_norm": 0.4906676113605499, "learning_rate": 0.0005, "loss": 1.3160967826843262, "step": 34791 }, { "epoch": 2.742767047694127, "grad_norm": 0.5012238025665283, "learning_rate": 0.0005, "loss": 1.2972240447998047, "step": 34792 }, { "epoch": 2.7428458809617657, "grad_norm": 0.5441476702690125, "learning_rate": 0.0005, "loss": 1.4228408336639404, "step": 34793 }, { "epoch": 2.742924714229405, "grad_norm": 0.5046610832214355, "learning_rate": 0.0005, "loss": 1.3936302661895752, "step": 34794 }, { "epoch": 2.743003547497044, "grad_norm": 0.44990813732147217, "learning_rate": 0.0005, "loss": 1.2411412000656128, "step": 34795 }, { "epoch": 2.7430823807646827, "grad_norm": 0.5014692544937134, "learning_rate": 0.0005, "loss": 1.4289320707321167, "step": 34796 }, { "epoch": 2.7431612140323214, "grad_norm": 0.5662676095962524, "learning_rate": 0.0005, "loss": 1.3269648551940918, "step": 34797 }, { "epoch": 2.7432400472999605, "grad_norm": 0.48370203375816345, "learning_rate": 0.0005, "loss": 1.267270803451538, "step": 34798 }, { "epoch": 2.7433188805675996, "grad_norm": 0.39594563841819763, "learning_rate": 0.0005, "loss": 1.3707172870635986, "step": 34799 }, { "epoch": 2.7433977138352383, "grad_norm": 0.44204527139663696, "learning_rate": 0.0005, "loss": 1.2533998489379883, "step": 34800 }, { "epoch": 2.7434765471028775, "grad_norm": 0.6861317753791809, "learning_rate": 0.0005, "loss": 1.4570621252059937, "step": 34801 }, { "epoch": 2.743555380370516, "grad_norm": 0.4754068851470947, "learning_rate": 0.0005, "loss": 1.2742537260055542, "step": 34802 }, { "epoch": 2.7436342136381553, "grad_norm": 0.5088376998901367, "learning_rate": 0.0005, "loss": 1.362402319908142, "step": 34803 }, { "epoch": 2.7437130469057944, "grad_norm": 0.48540249466896057, "learning_rate": 0.0005, "loss": 1.441791296005249, "step": 34804 }, { "epoch": 2.743791880173433, "grad_norm": 0.5978894233703613, "learning_rate": 0.0005, "loss": 1.5248929262161255, "step": 34805 }, { "epoch": 2.7438707134410723, "grad_norm": 0.569291889667511, "learning_rate": 0.0005, "loss": 1.339136004447937, "step": 34806 }, { "epoch": 2.743949546708711, "grad_norm": 0.558453381061554, "learning_rate": 0.0005, "loss": 1.5407257080078125, "step": 34807 }, { "epoch": 2.74402837997635, "grad_norm": 0.45183849334716797, "learning_rate": 0.0005, "loss": 1.2685296535491943, "step": 34808 }, { "epoch": 2.744107213243989, "grad_norm": 0.6465663909912109, "learning_rate": 0.0005, "loss": 1.3338027000427246, "step": 34809 }, { "epoch": 2.744186046511628, "grad_norm": 0.510302722454071, "learning_rate": 0.0005, "loss": 1.1522743701934814, "step": 34810 }, { "epoch": 2.7442648797792666, "grad_norm": 0.4981374144554138, "learning_rate": 0.0005, "loss": 1.413477897644043, "step": 34811 }, { "epoch": 2.7443437130469057, "grad_norm": 0.4543170630931854, "learning_rate": 0.0005, "loss": 1.3149809837341309, "step": 34812 }, { "epoch": 2.744422546314545, "grad_norm": 0.4401988983154297, "learning_rate": 0.0005, "loss": 1.1704540252685547, "step": 34813 }, { "epoch": 2.7445013795821835, "grad_norm": 0.5065075159072876, "learning_rate": 0.0005, "loss": 1.4379026889801025, "step": 34814 }, { "epoch": 2.7445802128498227, "grad_norm": 0.3685356378555298, "learning_rate": 0.0005, "loss": 1.1442575454711914, "step": 34815 }, { "epoch": 2.7446590461174614, "grad_norm": 0.47510701417922974, "learning_rate": 0.0005, "loss": 1.2574888467788696, "step": 34816 }, { "epoch": 2.7447378793851005, "grad_norm": 0.41588738560676575, "learning_rate": 0.0005, "loss": 1.4223172664642334, "step": 34817 }, { "epoch": 2.7448167126527396, "grad_norm": 0.49553149938583374, "learning_rate": 0.0005, "loss": 1.2598446607589722, "step": 34818 }, { "epoch": 2.7448955459203783, "grad_norm": 0.5666859745979309, "learning_rate": 0.0005, "loss": 1.4159886837005615, "step": 34819 }, { "epoch": 2.7449743791880175, "grad_norm": 0.45618996024131775, "learning_rate": 0.0005, "loss": 1.3100945949554443, "step": 34820 }, { "epoch": 2.745053212455656, "grad_norm": 0.4900345206260681, "learning_rate": 0.0005, "loss": 1.227940320968628, "step": 34821 }, { "epoch": 2.7451320457232953, "grad_norm": 0.5001844763755798, "learning_rate": 0.0005, "loss": 1.3642146587371826, "step": 34822 }, { "epoch": 2.7452108789909344, "grad_norm": 0.5313074588775635, "learning_rate": 0.0005, "loss": 1.3393549919128418, "step": 34823 }, { "epoch": 2.745289712258573, "grad_norm": 0.4464254379272461, "learning_rate": 0.0005, "loss": 1.1927404403686523, "step": 34824 }, { "epoch": 2.745368545526212, "grad_norm": 0.5544670224189758, "learning_rate": 0.0005, "loss": 1.3085668087005615, "step": 34825 }, { "epoch": 2.745447378793851, "grad_norm": 0.4940705895423889, "learning_rate": 0.0005, "loss": 1.2635107040405273, "step": 34826 }, { "epoch": 2.74552621206149, "grad_norm": 0.5022749900817871, "learning_rate": 0.0005, "loss": 1.4567248821258545, "step": 34827 }, { "epoch": 2.7456050453291287, "grad_norm": 0.5535376667976379, "learning_rate": 0.0005, "loss": 1.473111629486084, "step": 34828 }, { "epoch": 2.745683878596768, "grad_norm": 0.4191487729549408, "learning_rate": 0.0005, "loss": 1.266085147857666, "step": 34829 }, { "epoch": 2.7457627118644066, "grad_norm": 0.5717436075210571, "learning_rate": 0.0005, "loss": 1.3692784309387207, "step": 34830 }, { "epoch": 2.7458415451320457, "grad_norm": 0.4471195340156555, "learning_rate": 0.0005, "loss": 1.3075395822525024, "step": 34831 }, { "epoch": 2.745920378399685, "grad_norm": 0.5179681777954102, "learning_rate": 0.0005, "loss": 1.373551368713379, "step": 34832 }, { "epoch": 2.7459992116673235, "grad_norm": 0.5700235366821289, "learning_rate": 0.0005, "loss": 1.4333600997924805, "step": 34833 }, { "epoch": 2.7460780449349627, "grad_norm": 0.47501954436302185, "learning_rate": 0.0005, "loss": 1.319563865661621, "step": 34834 }, { "epoch": 2.7461568782026013, "grad_norm": 0.5446731448173523, "learning_rate": 0.0005, "loss": 1.1851558685302734, "step": 34835 }, { "epoch": 2.7462357114702405, "grad_norm": 0.5940981507301331, "learning_rate": 0.0005, "loss": 1.3569774627685547, "step": 34836 }, { "epoch": 2.7463145447378796, "grad_norm": 0.483003705739975, "learning_rate": 0.0005, "loss": 1.3794056177139282, "step": 34837 }, { "epoch": 2.7463933780055183, "grad_norm": 0.5238203406333923, "learning_rate": 0.0005, "loss": 1.2388782501220703, "step": 34838 }, { "epoch": 2.7464722112731574, "grad_norm": 0.4569060206413269, "learning_rate": 0.0005, "loss": 1.2832839488983154, "step": 34839 }, { "epoch": 2.746551044540796, "grad_norm": 0.517372190952301, "learning_rate": 0.0005, "loss": 1.1355466842651367, "step": 34840 }, { "epoch": 2.7466298778084353, "grad_norm": 0.5779933333396912, "learning_rate": 0.0005, "loss": 1.2833912372589111, "step": 34841 }, { "epoch": 2.746708711076074, "grad_norm": 0.5194288492202759, "learning_rate": 0.0005, "loss": 1.2863364219665527, "step": 34842 }, { "epoch": 2.746787544343713, "grad_norm": 0.4327256679534912, "learning_rate": 0.0005, "loss": 1.2515838146209717, "step": 34843 }, { "epoch": 2.7468663776113518, "grad_norm": 0.4951322078704834, "learning_rate": 0.0005, "loss": 1.2744688987731934, "step": 34844 }, { "epoch": 2.746945210878991, "grad_norm": 0.5521532297134399, "learning_rate": 0.0005, "loss": 1.3211512565612793, "step": 34845 }, { "epoch": 2.74702404414663, "grad_norm": 0.6002913117408752, "learning_rate": 0.0005, "loss": 1.3152844905853271, "step": 34846 }, { "epoch": 2.7471028774142687, "grad_norm": 0.5844975709915161, "learning_rate": 0.0005, "loss": 1.2407922744750977, "step": 34847 }, { "epoch": 2.747181710681908, "grad_norm": 0.650205671787262, "learning_rate": 0.0005, "loss": 1.4847177267074585, "step": 34848 }, { "epoch": 2.7472605439495466, "grad_norm": 0.5160161852836609, "learning_rate": 0.0005, "loss": 1.4477951526641846, "step": 34849 }, { "epoch": 2.7473393772171857, "grad_norm": 0.4936637878417969, "learning_rate": 0.0005, "loss": 1.6995797157287598, "step": 34850 }, { "epoch": 2.747418210484825, "grad_norm": 0.5189223289489746, "learning_rate": 0.0005, "loss": 1.338134765625, "step": 34851 }, { "epoch": 2.7474970437524635, "grad_norm": 0.5475614070892334, "learning_rate": 0.0005, "loss": 1.313559651374817, "step": 34852 }, { "epoch": 2.7475758770201026, "grad_norm": 0.5271425247192383, "learning_rate": 0.0005, "loss": 1.4328498840332031, "step": 34853 }, { "epoch": 2.7476547102877413, "grad_norm": 0.6822482347488403, "learning_rate": 0.0005, "loss": 1.364823579788208, "step": 34854 }, { "epoch": 2.7477335435553805, "grad_norm": 0.5655303597450256, "learning_rate": 0.0005, "loss": 1.3026740550994873, "step": 34855 }, { "epoch": 2.747812376823019, "grad_norm": 0.5808221697807312, "learning_rate": 0.0005, "loss": 1.5279208421707153, "step": 34856 }, { "epoch": 2.7478912100906583, "grad_norm": 0.4234835207462311, "learning_rate": 0.0005, "loss": 1.2820478677749634, "step": 34857 }, { "epoch": 2.747970043358297, "grad_norm": 0.5091322660446167, "learning_rate": 0.0005, "loss": 1.4363303184509277, "step": 34858 }, { "epoch": 2.748048876625936, "grad_norm": 0.5031497478485107, "learning_rate": 0.0005, "loss": 1.1346930265426636, "step": 34859 }, { "epoch": 2.7481277098935752, "grad_norm": 0.5157983899116516, "learning_rate": 0.0005, "loss": 1.2756332159042358, "step": 34860 }, { "epoch": 2.748206543161214, "grad_norm": 0.6425036191940308, "learning_rate": 0.0005, "loss": 1.1598029136657715, "step": 34861 }, { "epoch": 2.748285376428853, "grad_norm": 0.4807990491390228, "learning_rate": 0.0005, "loss": 1.2355308532714844, "step": 34862 }, { "epoch": 2.7483642096964918, "grad_norm": 0.46066048741340637, "learning_rate": 0.0005, "loss": 1.3648868799209595, "step": 34863 }, { "epoch": 2.748443042964131, "grad_norm": 0.5126534104347229, "learning_rate": 0.0005, "loss": 1.3966898918151855, "step": 34864 }, { "epoch": 2.74852187623177, "grad_norm": 0.4834882915019989, "learning_rate": 0.0005, "loss": 1.3657554388046265, "step": 34865 }, { "epoch": 2.7486007094994087, "grad_norm": 0.49955618381500244, "learning_rate": 0.0005, "loss": 1.2557175159454346, "step": 34866 }, { "epoch": 2.748679542767048, "grad_norm": 0.6220505833625793, "learning_rate": 0.0005, "loss": 1.3177911043167114, "step": 34867 }, { "epoch": 2.7487583760346865, "grad_norm": 0.5186170935630798, "learning_rate": 0.0005, "loss": 1.2782474756240845, "step": 34868 }, { "epoch": 2.7488372093023257, "grad_norm": 0.4428377151489258, "learning_rate": 0.0005, "loss": 1.282519817352295, "step": 34869 }, { "epoch": 2.748916042569965, "grad_norm": 0.4627699553966522, "learning_rate": 0.0005, "loss": 1.35170316696167, "step": 34870 }, { "epoch": 2.7489948758376035, "grad_norm": 0.5435940027236938, "learning_rate": 0.0005, "loss": 1.2310153245925903, "step": 34871 }, { "epoch": 2.749073709105242, "grad_norm": 0.5418998003005981, "learning_rate": 0.0005, "loss": 1.238172173500061, "step": 34872 }, { "epoch": 2.7491525423728813, "grad_norm": 0.966023325920105, "learning_rate": 0.0005, "loss": 1.1524772644042969, "step": 34873 }, { "epoch": 2.7492313756405204, "grad_norm": 0.4786128103733063, "learning_rate": 0.0005, "loss": 1.3569986820220947, "step": 34874 }, { "epoch": 2.749310208908159, "grad_norm": 0.4764971137046814, "learning_rate": 0.0005, "loss": 1.3372706174850464, "step": 34875 }, { "epoch": 2.7493890421757983, "grad_norm": 0.4681706726551056, "learning_rate": 0.0005, "loss": 1.2572715282440186, "step": 34876 }, { "epoch": 2.749467875443437, "grad_norm": 0.3867741525173187, "learning_rate": 0.0005, "loss": 1.2791752815246582, "step": 34877 }, { "epoch": 2.749546708711076, "grad_norm": 0.5043726563453674, "learning_rate": 0.0005, "loss": 1.4222453832626343, "step": 34878 }, { "epoch": 2.7496255419787152, "grad_norm": 0.45339030027389526, "learning_rate": 0.0005, "loss": 1.3044918775558472, "step": 34879 }, { "epoch": 2.749704375246354, "grad_norm": 0.4658605456352234, "learning_rate": 0.0005, "loss": 1.2083368301391602, "step": 34880 }, { "epoch": 2.749783208513993, "grad_norm": 0.42672860622406006, "learning_rate": 0.0005, "loss": 1.3306771516799927, "step": 34881 }, { "epoch": 2.7498620417816317, "grad_norm": 0.4873332381248474, "learning_rate": 0.0005, "loss": 1.3127505779266357, "step": 34882 }, { "epoch": 2.749940875049271, "grad_norm": 0.49202418327331543, "learning_rate": 0.0005, "loss": 1.331534743309021, "step": 34883 }, { "epoch": 2.75001970831691, "grad_norm": 0.46652480959892273, "learning_rate": 0.0005, "loss": 1.2660698890686035, "step": 34884 }, { "epoch": 2.7500985415845487, "grad_norm": 0.5417236089706421, "learning_rate": 0.0005, "loss": 1.3118467330932617, "step": 34885 }, { "epoch": 2.7501773748521874, "grad_norm": 0.43189987540245056, "learning_rate": 0.0005, "loss": 1.2060357332229614, "step": 34886 }, { "epoch": 2.7502562081198265, "grad_norm": 0.5774287581443787, "learning_rate": 0.0005, "loss": 1.4486290216445923, "step": 34887 }, { "epoch": 2.7503350413874657, "grad_norm": 0.436198353767395, "learning_rate": 0.0005, "loss": 1.316540241241455, "step": 34888 }, { "epoch": 2.7504138746551043, "grad_norm": 0.4901888370513916, "learning_rate": 0.0005, "loss": 1.4327808618545532, "step": 34889 }, { "epoch": 2.7504927079227435, "grad_norm": 0.4752271771430969, "learning_rate": 0.0005, "loss": 1.2235431671142578, "step": 34890 }, { "epoch": 2.750571541190382, "grad_norm": 0.5183418989181519, "learning_rate": 0.0005, "loss": 1.2308344841003418, "step": 34891 }, { "epoch": 2.7506503744580213, "grad_norm": 0.6075941324234009, "learning_rate": 0.0005, "loss": 1.4666582345962524, "step": 34892 }, { "epoch": 2.7507292077256604, "grad_norm": 0.6930124759674072, "learning_rate": 0.0005, "loss": 1.4396283626556396, "step": 34893 }, { "epoch": 2.750808040993299, "grad_norm": 0.5417836904525757, "learning_rate": 0.0005, "loss": 1.4139994382858276, "step": 34894 }, { "epoch": 2.7508868742609383, "grad_norm": 0.45604920387268066, "learning_rate": 0.0005, "loss": 1.357208251953125, "step": 34895 }, { "epoch": 2.750965707528577, "grad_norm": 0.48304522037506104, "learning_rate": 0.0005, "loss": 1.2829337120056152, "step": 34896 }, { "epoch": 2.751044540796216, "grad_norm": 0.4451044201850891, "learning_rate": 0.0005, "loss": 1.1849830150604248, "step": 34897 }, { "epoch": 2.751123374063855, "grad_norm": 0.6000191569328308, "learning_rate": 0.0005, "loss": 1.6691646575927734, "step": 34898 }, { "epoch": 2.751202207331494, "grad_norm": 0.5570487976074219, "learning_rate": 0.0005, "loss": 1.2145500183105469, "step": 34899 }, { "epoch": 2.7512810405991326, "grad_norm": 0.4457373321056366, "learning_rate": 0.0005, "loss": 1.3772494792938232, "step": 34900 }, { "epoch": 2.7513598738667717, "grad_norm": 0.46152251958847046, "learning_rate": 0.0005, "loss": 1.3986459970474243, "step": 34901 }, { "epoch": 2.751438707134411, "grad_norm": 0.4372180700302124, "learning_rate": 0.0005, "loss": 1.1478271484375, "step": 34902 }, { "epoch": 2.7515175404020495, "grad_norm": 0.6041548848152161, "learning_rate": 0.0005, "loss": 1.2309441566467285, "step": 34903 }, { "epoch": 2.7515963736696887, "grad_norm": 0.5737385749816895, "learning_rate": 0.0005, "loss": 1.2656195163726807, "step": 34904 }, { "epoch": 2.7516752069373274, "grad_norm": 0.5305520296096802, "learning_rate": 0.0005, "loss": 1.43794584274292, "step": 34905 }, { "epoch": 2.7517540402049665, "grad_norm": 0.42892658710479736, "learning_rate": 0.0005, "loss": 1.288461685180664, "step": 34906 }, { "epoch": 2.7518328734726056, "grad_norm": 0.5113341212272644, "learning_rate": 0.0005, "loss": 1.2345904111862183, "step": 34907 }, { "epoch": 2.7519117067402443, "grad_norm": 0.4869813919067383, "learning_rate": 0.0005, "loss": 1.262490153312683, "step": 34908 }, { "epoch": 2.7519905400078835, "grad_norm": 0.5161533951759338, "learning_rate": 0.0005, "loss": 1.178373098373413, "step": 34909 }, { "epoch": 2.752069373275522, "grad_norm": 0.43595558404922485, "learning_rate": 0.0005, "loss": 1.2051925659179688, "step": 34910 }, { "epoch": 2.7521482065431613, "grad_norm": 0.4653729200363159, "learning_rate": 0.0005, "loss": 1.2186895608901978, "step": 34911 }, { "epoch": 2.7522270398108004, "grad_norm": 0.4929564595222473, "learning_rate": 0.0005, "loss": 1.3093923330307007, "step": 34912 }, { "epoch": 2.752305873078439, "grad_norm": 0.42140790820121765, "learning_rate": 0.0005, "loss": 1.1671745777130127, "step": 34913 }, { "epoch": 2.752384706346078, "grad_norm": 0.446024626493454, "learning_rate": 0.0005, "loss": 1.2370309829711914, "step": 34914 }, { "epoch": 2.752463539613717, "grad_norm": 0.5199552178382874, "learning_rate": 0.0005, "loss": 1.4961683750152588, "step": 34915 }, { "epoch": 2.752542372881356, "grad_norm": 0.39524829387664795, "learning_rate": 0.0005, "loss": 1.4120029211044312, "step": 34916 }, { "epoch": 2.7526212061489947, "grad_norm": 0.6146150231361389, "learning_rate": 0.0005, "loss": 1.3124998807907104, "step": 34917 }, { "epoch": 2.752700039416634, "grad_norm": 0.488840252161026, "learning_rate": 0.0005, "loss": 1.3728365898132324, "step": 34918 }, { "epoch": 2.7527788726842726, "grad_norm": 0.4477713406085968, "learning_rate": 0.0005, "loss": 1.3099826574325562, "step": 34919 }, { "epoch": 2.7528577059519117, "grad_norm": 0.5317416191101074, "learning_rate": 0.0005, "loss": 1.3749290704727173, "step": 34920 }, { "epoch": 2.752936539219551, "grad_norm": 0.7436664700508118, "learning_rate": 0.0005, "loss": 1.3736522197723389, "step": 34921 }, { "epoch": 2.7530153724871895, "grad_norm": 0.5202651619911194, "learning_rate": 0.0005, "loss": 1.3780136108398438, "step": 34922 }, { "epoch": 2.7530942057548287, "grad_norm": 0.4530338943004608, "learning_rate": 0.0005, "loss": 1.304619312286377, "step": 34923 }, { "epoch": 2.7531730390224673, "grad_norm": 0.6306197047233582, "learning_rate": 0.0005, "loss": 1.4386961460113525, "step": 34924 }, { "epoch": 2.7532518722901065, "grad_norm": 0.481230229139328, "learning_rate": 0.0005, "loss": 1.3505104780197144, "step": 34925 }, { "epoch": 2.7533307055577456, "grad_norm": 0.47181636095046997, "learning_rate": 0.0005, "loss": 1.5385282039642334, "step": 34926 }, { "epoch": 2.7534095388253843, "grad_norm": 0.432940810918808, "learning_rate": 0.0005, "loss": 1.3528575897216797, "step": 34927 }, { "epoch": 2.753488372093023, "grad_norm": 0.4115554690361023, "learning_rate": 0.0005, "loss": 1.2473318576812744, "step": 34928 }, { "epoch": 2.753567205360662, "grad_norm": 0.5674760341644287, "learning_rate": 0.0005, "loss": 1.314422607421875, "step": 34929 }, { "epoch": 2.7536460386283013, "grad_norm": 0.5226834416389465, "learning_rate": 0.0005, "loss": 1.3901066780090332, "step": 34930 }, { "epoch": 2.75372487189594, "grad_norm": 0.6853922605514526, "learning_rate": 0.0005, "loss": 1.512359619140625, "step": 34931 }, { "epoch": 2.753803705163579, "grad_norm": 0.5861023664474487, "learning_rate": 0.0005, "loss": 1.4495149850845337, "step": 34932 }, { "epoch": 2.7538825384312178, "grad_norm": 0.5756235718727112, "learning_rate": 0.0005, "loss": 1.3832433223724365, "step": 34933 }, { "epoch": 2.753961371698857, "grad_norm": 0.49410977959632874, "learning_rate": 0.0005, "loss": 1.1702297925949097, "step": 34934 }, { "epoch": 2.754040204966496, "grad_norm": 0.47260230779647827, "learning_rate": 0.0005, "loss": 1.2468541860580444, "step": 34935 }, { "epoch": 2.7541190382341347, "grad_norm": 0.4100378453731537, "learning_rate": 0.0005, "loss": 1.265344262123108, "step": 34936 }, { "epoch": 2.754197871501774, "grad_norm": 0.6504765152931213, "learning_rate": 0.0005, "loss": 1.2170336246490479, "step": 34937 }, { "epoch": 2.7542767047694126, "grad_norm": 0.45146387815475464, "learning_rate": 0.0005, "loss": 1.296539545059204, "step": 34938 }, { "epoch": 2.7543555380370517, "grad_norm": 0.5175355076789856, "learning_rate": 0.0005, "loss": 1.3876402378082275, "step": 34939 }, { "epoch": 2.754434371304691, "grad_norm": 0.5439774990081787, "learning_rate": 0.0005, "loss": 1.4792587757110596, "step": 34940 }, { "epoch": 2.7545132045723295, "grad_norm": 0.511220395565033, "learning_rate": 0.0005, "loss": 1.315763235092163, "step": 34941 }, { "epoch": 2.754592037839968, "grad_norm": 0.6637471914291382, "learning_rate": 0.0005, "loss": 1.4948726892471313, "step": 34942 }, { "epoch": 2.7546708711076073, "grad_norm": 0.46674996614456177, "learning_rate": 0.0005, "loss": 1.376097321510315, "step": 34943 }, { "epoch": 2.7547497043752465, "grad_norm": 0.5671642422676086, "learning_rate": 0.0005, "loss": 1.3600184917449951, "step": 34944 }, { "epoch": 2.754828537642885, "grad_norm": 0.6170461773872375, "learning_rate": 0.0005, "loss": 1.3993257284164429, "step": 34945 }, { "epoch": 2.7549073709105243, "grad_norm": 0.8131244778633118, "learning_rate": 0.0005, "loss": 1.6339542865753174, "step": 34946 }, { "epoch": 2.754986204178163, "grad_norm": 0.544120192527771, "learning_rate": 0.0005, "loss": 1.3815555572509766, "step": 34947 }, { "epoch": 2.755065037445802, "grad_norm": 0.49288609623908997, "learning_rate": 0.0005, "loss": 1.2364261150360107, "step": 34948 }, { "epoch": 2.7551438707134412, "grad_norm": 0.5123692154884338, "learning_rate": 0.0005, "loss": 1.5295052528381348, "step": 34949 }, { "epoch": 2.75522270398108, "grad_norm": 0.5686463117599487, "learning_rate": 0.0005, "loss": 1.4414589405059814, "step": 34950 }, { "epoch": 2.755301537248719, "grad_norm": 0.4720242917537689, "learning_rate": 0.0005, "loss": 1.3061070442199707, "step": 34951 }, { "epoch": 2.7553803705163578, "grad_norm": 0.5535873174667358, "learning_rate": 0.0005, "loss": 1.3780066967010498, "step": 34952 }, { "epoch": 2.755459203783997, "grad_norm": 0.49396181106567383, "learning_rate": 0.0005, "loss": 1.3201780319213867, "step": 34953 }, { "epoch": 2.755538037051636, "grad_norm": 0.485050231218338, "learning_rate": 0.0005, "loss": 1.2468312978744507, "step": 34954 }, { "epoch": 2.7556168703192747, "grad_norm": 0.4832231104373932, "learning_rate": 0.0005, "loss": 1.1228790283203125, "step": 34955 }, { "epoch": 2.7556957035869134, "grad_norm": 0.5863429307937622, "learning_rate": 0.0005, "loss": 1.437694787979126, "step": 34956 }, { "epoch": 2.7557745368545525, "grad_norm": 0.6043557524681091, "learning_rate": 0.0005, "loss": 1.4314086437225342, "step": 34957 }, { "epoch": 2.7558533701221917, "grad_norm": 0.4974438548088074, "learning_rate": 0.0005, "loss": 1.4320502281188965, "step": 34958 }, { "epoch": 2.7559322033898304, "grad_norm": 0.4432007670402527, "learning_rate": 0.0005, "loss": 1.3877487182617188, "step": 34959 }, { "epoch": 2.7560110366574695, "grad_norm": 0.5326863527297974, "learning_rate": 0.0005, "loss": 1.4312810897827148, "step": 34960 }, { "epoch": 2.756089869925108, "grad_norm": 0.5053398609161377, "learning_rate": 0.0005, "loss": 1.4751842021942139, "step": 34961 }, { "epoch": 2.7561687031927473, "grad_norm": 0.4038163125514984, "learning_rate": 0.0005, "loss": 1.1545417308807373, "step": 34962 }, { "epoch": 2.7562475364603864, "grad_norm": 0.7040082812309265, "learning_rate": 0.0005, "loss": 1.3699315786361694, "step": 34963 }, { "epoch": 2.756326369728025, "grad_norm": 0.5635662078857422, "learning_rate": 0.0005, "loss": 1.4610607624053955, "step": 34964 }, { "epoch": 2.7564052029956643, "grad_norm": 0.49703657627105713, "learning_rate": 0.0005, "loss": 1.3850090503692627, "step": 34965 }, { "epoch": 2.756484036263303, "grad_norm": 0.49407318234443665, "learning_rate": 0.0005, "loss": 1.2497210502624512, "step": 34966 }, { "epoch": 2.756562869530942, "grad_norm": 0.524013102054596, "learning_rate": 0.0005, "loss": 1.3369054794311523, "step": 34967 }, { "epoch": 2.7566417027985812, "grad_norm": 0.4086448550224304, "learning_rate": 0.0005, "loss": 1.2672762870788574, "step": 34968 }, { "epoch": 2.75672053606622, "grad_norm": 0.45080599188804626, "learning_rate": 0.0005, "loss": 1.5212208032608032, "step": 34969 }, { "epoch": 2.756799369333859, "grad_norm": 0.5645542740821838, "learning_rate": 0.0005, "loss": 1.242132306098938, "step": 34970 }, { "epoch": 2.7568782026014977, "grad_norm": 0.5178118348121643, "learning_rate": 0.0005, "loss": 1.2921110391616821, "step": 34971 }, { "epoch": 2.756957035869137, "grad_norm": 0.45045745372772217, "learning_rate": 0.0005, "loss": 1.4939287900924683, "step": 34972 }, { "epoch": 2.7570358691367756, "grad_norm": 0.45385995507240295, "learning_rate": 0.0005, "loss": 1.4538744688034058, "step": 34973 }, { "epoch": 2.7571147024044147, "grad_norm": 0.5249384641647339, "learning_rate": 0.0005, "loss": 1.1551399230957031, "step": 34974 }, { "epoch": 2.7571935356720534, "grad_norm": 0.46206483244895935, "learning_rate": 0.0005, "loss": 1.2452714443206787, "step": 34975 }, { "epoch": 2.7572723689396925, "grad_norm": 0.47329431772232056, "learning_rate": 0.0005, "loss": 1.3064286708831787, "step": 34976 }, { "epoch": 2.7573512022073317, "grad_norm": 0.424333781003952, "learning_rate": 0.0005, "loss": 1.2146382331848145, "step": 34977 }, { "epoch": 2.7574300354749703, "grad_norm": 0.5245180130004883, "learning_rate": 0.0005, "loss": 1.3184633255004883, "step": 34978 }, { "epoch": 2.7575088687426095, "grad_norm": 0.44706791639328003, "learning_rate": 0.0005, "loss": 1.2204372882843018, "step": 34979 }, { "epoch": 2.757587702010248, "grad_norm": 0.5076934099197388, "learning_rate": 0.0005, "loss": 1.5092921257019043, "step": 34980 }, { "epoch": 2.7576665352778873, "grad_norm": 0.4442497193813324, "learning_rate": 0.0005, "loss": 1.1109659671783447, "step": 34981 }, { "epoch": 2.7577453685455264, "grad_norm": 0.4654630422592163, "learning_rate": 0.0005, "loss": 1.3742889165878296, "step": 34982 }, { "epoch": 2.757824201813165, "grad_norm": 0.4361066222190857, "learning_rate": 0.0005, "loss": 1.298462152481079, "step": 34983 }, { "epoch": 2.7579030350808043, "grad_norm": 0.5754554271697998, "learning_rate": 0.0005, "loss": 1.306315302848816, "step": 34984 }, { "epoch": 2.757981868348443, "grad_norm": 0.3982127010822296, "learning_rate": 0.0005, "loss": 1.1996417045593262, "step": 34985 }, { "epoch": 2.758060701616082, "grad_norm": 0.5606020092964172, "learning_rate": 0.0005, "loss": 1.421832799911499, "step": 34986 }, { "epoch": 2.7581395348837208, "grad_norm": 0.4459489583969116, "learning_rate": 0.0005, "loss": 1.4288910627365112, "step": 34987 }, { "epoch": 2.75821836815136, "grad_norm": 0.5687179565429688, "learning_rate": 0.0005, "loss": 1.3625704050064087, "step": 34988 }, { "epoch": 2.7582972014189986, "grad_norm": 0.42459574341773987, "learning_rate": 0.0005, "loss": 1.223486065864563, "step": 34989 }, { "epoch": 2.7583760346866377, "grad_norm": 0.4901483952999115, "learning_rate": 0.0005, "loss": 1.3785333633422852, "step": 34990 }, { "epoch": 2.758454867954277, "grad_norm": 0.4337049126625061, "learning_rate": 0.0005, "loss": 1.2137131690979004, "step": 34991 }, { "epoch": 2.7585337012219155, "grad_norm": 0.5846812725067139, "learning_rate": 0.0005, "loss": 1.433389663696289, "step": 34992 }, { "epoch": 2.7586125344895547, "grad_norm": 0.48058387637138367, "learning_rate": 0.0005, "loss": 1.2908153533935547, "step": 34993 }, { "epoch": 2.7586913677571934, "grad_norm": 0.41985222697257996, "learning_rate": 0.0005, "loss": 1.1004904508590698, "step": 34994 }, { "epoch": 2.7587702010248325, "grad_norm": 0.5522285103797913, "learning_rate": 0.0005, "loss": 1.4505363702774048, "step": 34995 }, { "epoch": 2.7588490342924716, "grad_norm": 0.5321156978607178, "learning_rate": 0.0005, "loss": 1.31793212890625, "step": 34996 }, { "epoch": 2.7589278675601103, "grad_norm": 0.4713059663772583, "learning_rate": 0.0005, "loss": 1.2745342254638672, "step": 34997 }, { "epoch": 2.7590067008277495, "grad_norm": 0.6735148429870605, "learning_rate": 0.0005, "loss": 1.305952787399292, "step": 34998 }, { "epoch": 2.759085534095388, "grad_norm": 0.45880332589149475, "learning_rate": 0.0005, "loss": 1.1938011646270752, "step": 34999 }, { "epoch": 2.7591643673630273, "grad_norm": 0.5392799973487854, "learning_rate": 0.0005, "loss": 1.2827434539794922, "step": 35000 }, { "epoch": 2.7592432006306664, "grad_norm": 0.46944284439086914, "learning_rate": 0.0005, "loss": 1.4867660999298096, "step": 35001 }, { "epoch": 2.759322033898305, "grad_norm": 0.5043957233428955, "learning_rate": 0.0005, "loss": 1.3049328327178955, "step": 35002 }, { "epoch": 2.759400867165944, "grad_norm": 0.657446026802063, "learning_rate": 0.0005, "loss": 1.5750651359558105, "step": 35003 }, { "epoch": 2.759479700433583, "grad_norm": 0.6176305413246155, "learning_rate": 0.0005, "loss": 1.5241711139678955, "step": 35004 }, { "epoch": 2.759558533701222, "grad_norm": 0.45471087098121643, "learning_rate": 0.0005, "loss": 1.1654467582702637, "step": 35005 }, { "epoch": 2.7596373669688607, "grad_norm": 0.8326579332351685, "learning_rate": 0.0005, "loss": 1.3244984149932861, "step": 35006 }, { "epoch": 2.7597162002365, "grad_norm": 0.6361290216445923, "learning_rate": 0.0005, "loss": 1.4171773195266724, "step": 35007 }, { "epoch": 2.7597950335041386, "grad_norm": 0.5125183463096619, "learning_rate": 0.0005, "loss": 1.3063511848449707, "step": 35008 }, { "epoch": 2.7598738667717777, "grad_norm": 0.5003364682197571, "learning_rate": 0.0005, "loss": 1.1800832748413086, "step": 35009 }, { "epoch": 2.759952700039417, "grad_norm": 0.6321406364440918, "learning_rate": 0.0005, "loss": 1.37758469581604, "step": 35010 }, { "epoch": 2.7600315333070555, "grad_norm": 0.5619624257087708, "learning_rate": 0.0005, "loss": 1.4903311729431152, "step": 35011 }, { "epoch": 2.7601103665746947, "grad_norm": 0.8572874665260315, "learning_rate": 0.0005, "loss": 1.34205162525177, "step": 35012 }, { "epoch": 2.7601891998423334, "grad_norm": 0.44411808252334595, "learning_rate": 0.0005, "loss": 1.2365765571594238, "step": 35013 }, { "epoch": 2.7602680331099725, "grad_norm": 0.4959048330783844, "learning_rate": 0.0005, "loss": 1.4737060070037842, "step": 35014 }, { "epoch": 2.7603468663776116, "grad_norm": 0.6686298251152039, "learning_rate": 0.0005, "loss": 1.285244345664978, "step": 35015 }, { "epoch": 2.7604256996452503, "grad_norm": 0.5015840530395508, "learning_rate": 0.0005, "loss": 1.276381015777588, "step": 35016 }, { "epoch": 2.760504532912889, "grad_norm": 0.557874858379364, "learning_rate": 0.0005, "loss": 1.3120954036712646, "step": 35017 }, { "epoch": 2.760583366180528, "grad_norm": 0.5272992253303528, "learning_rate": 0.0005, "loss": 1.1388188600540161, "step": 35018 }, { "epoch": 2.7606621994481673, "grad_norm": 0.47677865624427795, "learning_rate": 0.0005, "loss": 1.256748914718628, "step": 35019 }, { "epoch": 2.760741032715806, "grad_norm": 0.48763230443000793, "learning_rate": 0.0005, "loss": 1.1441705226898193, "step": 35020 }, { "epoch": 2.760819865983445, "grad_norm": 0.5139769315719604, "learning_rate": 0.0005, "loss": 1.4170726537704468, "step": 35021 }, { "epoch": 2.7608986992510838, "grad_norm": 0.5637743473052979, "learning_rate": 0.0005, "loss": 1.4728760719299316, "step": 35022 }, { "epoch": 2.760977532518723, "grad_norm": 0.4848909080028534, "learning_rate": 0.0005, "loss": 1.1743271350860596, "step": 35023 }, { "epoch": 2.761056365786362, "grad_norm": 0.48080989718437195, "learning_rate": 0.0005, "loss": 1.46140456199646, "step": 35024 }, { "epoch": 2.7611351990540007, "grad_norm": 0.5697471499443054, "learning_rate": 0.0005, "loss": 1.3033396005630493, "step": 35025 }, { "epoch": 2.76121403232164, "grad_norm": 0.5544777512550354, "learning_rate": 0.0005, "loss": 1.5251376628875732, "step": 35026 }, { "epoch": 2.7612928655892786, "grad_norm": 0.4374813735485077, "learning_rate": 0.0005, "loss": 1.2375966310501099, "step": 35027 }, { "epoch": 2.7613716988569177, "grad_norm": 0.49325892329216003, "learning_rate": 0.0005, "loss": 1.1854090690612793, "step": 35028 }, { "epoch": 2.761450532124557, "grad_norm": 0.4361896812915802, "learning_rate": 0.0005, "loss": 1.2267142534255981, "step": 35029 }, { "epoch": 2.7615293653921955, "grad_norm": 0.38313812017440796, "learning_rate": 0.0005, "loss": 1.2345666885375977, "step": 35030 }, { "epoch": 2.761608198659834, "grad_norm": 0.5615782737731934, "learning_rate": 0.0005, "loss": 1.2844420671463013, "step": 35031 }, { "epoch": 2.7616870319274733, "grad_norm": 0.4609934091567993, "learning_rate": 0.0005, "loss": 1.4861005544662476, "step": 35032 }, { "epoch": 2.7617658651951125, "grad_norm": 0.60404372215271, "learning_rate": 0.0005, "loss": 1.5010926723480225, "step": 35033 }, { "epoch": 2.761844698462751, "grad_norm": 0.4767675995826721, "learning_rate": 0.0005, "loss": 1.2450060844421387, "step": 35034 }, { "epoch": 2.7619235317303903, "grad_norm": 0.7059378623962402, "learning_rate": 0.0005, "loss": 1.2593334913253784, "step": 35035 }, { "epoch": 2.762002364998029, "grad_norm": 0.5830900073051453, "learning_rate": 0.0005, "loss": 1.4360296726226807, "step": 35036 }, { "epoch": 2.762081198265668, "grad_norm": 0.47690337896347046, "learning_rate": 0.0005, "loss": 1.2674494981765747, "step": 35037 }, { "epoch": 2.7621600315333072, "grad_norm": 0.5145075917243958, "learning_rate": 0.0005, "loss": 1.4948161840438843, "step": 35038 }, { "epoch": 2.762238864800946, "grad_norm": 0.42166534066200256, "learning_rate": 0.0005, "loss": 1.2119091749191284, "step": 35039 }, { "epoch": 2.762317698068585, "grad_norm": 0.4396718144416809, "learning_rate": 0.0005, "loss": 1.2327148914337158, "step": 35040 }, { "epoch": 2.7623965313362238, "grad_norm": 0.5062065124511719, "learning_rate": 0.0005, "loss": 1.4527270793914795, "step": 35041 }, { "epoch": 2.762475364603863, "grad_norm": 0.550004243850708, "learning_rate": 0.0005, "loss": 1.4464919567108154, "step": 35042 }, { "epoch": 2.762554197871502, "grad_norm": 0.45518580079078674, "learning_rate": 0.0005, "loss": 1.1966358423233032, "step": 35043 }, { "epoch": 2.7626330311391407, "grad_norm": 0.4054470658302307, "learning_rate": 0.0005, "loss": 1.2284560203552246, "step": 35044 }, { "epoch": 2.7627118644067794, "grad_norm": 0.5410326719284058, "learning_rate": 0.0005, "loss": 1.486536979675293, "step": 35045 }, { "epoch": 2.7627906976744185, "grad_norm": 0.4645240604877472, "learning_rate": 0.0005, "loss": 1.2576720714569092, "step": 35046 }, { "epoch": 2.7628695309420577, "grad_norm": 0.45833656191825867, "learning_rate": 0.0005, "loss": 1.402832269668579, "step": 35047 }, { "epoch": 2.7629483642096964, "grad_norm": 0.5006750822067261, "learning_rate": 0.0005, "loss": 1.3101060390472412, "step": 35048 }, { "epoch": 2.7630271974773355, "grad_norm": 0.52263343334198, "learning_rate": 0.0005, "loss": 1.4334136247634888, "step": 35049 }, { "epoch": 2.763106030744974, "grad_norm": 0.47021302580833435, "learning_rate": 0.0005, "loss": 1.3775333166122437, "step": 35050 }, { "epoch": 2.7631848640126133, "grad_norm": 0.48528486490249634, "learning_rate": 0.0005, "loss": 1.3234796524047852, "step": 35051 }, { "epoch": 2.7632636972802525, "grad_norm": 0.41168108582496643, "learning_rate": 0.0005, "loss": 1.2327711582183838, "step": 35052 }, { "epoch": 2.763342530547891, "grad_norm": 0.505822479724884, "learning_rate": 0.0005, "loss": 1.413781762123108, "step": 35053 }, { "epoch": 2.7634213638155303, "grad_norm": 0.5808597803115845, "learning_rate": 0.0005, "loss": 1.31052565574646, "step": 35054 }, { "epoch": 2.763500197083169, "grad_norm": 0.45604410767555237, "learning_rate": 0.0005, "loss": 1.270566701889038, "step": 35055 }, { "epoch": 2.763579030350808, "grad_norm": 0.5159265398979187, "learning_rate": 0.0005, "loss": 1.330162763595581, "step": 35056 }, { "epoch": 2.7636578636184472, "grad_norm": 0.4681130349636078, "learning_rate": 0.0005, "loss": 1.1288608312606812, "step": 35057 }, { "epoch": 2.763736696886086, "grad_norm": 0.6952773928642273, "learning_rate": 0.0005, "loss": 1.381561040878296, "step": 35058 }, { "epoch": 2.7638155301537246, "grad_norm": 0.5351331233978271, "learning_rate": 0.0005, "loss": 1.3236937522888184, "step": 35059 }, { "epoch": 2.7638943634213637, "grad_norm": 0.4219849705696106, "learning_rate": 0.0005, "loss": 1.2828266620635986, "step": 35060 }, { "epoch": 2.763973196689003, "grad_norm": 0.5455524921417236, "learning_rate": 0.0005, "loss": 1.3481359481811523, "step": 35061 }, { "epoch": 2.7640520299566416, "grad_norm": 0.6176311373710632, "learning_rate": 0.0005, "loss": 1.2847661972045898, "step": 35062 }, { "epoch": 2.7641308632242807, "grad_norm": 0.4730428159236908, "learning_rate": 0.0005, "loss": 1.3755073547363281, "step": 35063 }, { "epoch": 2.7642096964919194, "grad_norm": 0.6845508217811584, "learning_rate": 0.0005, "loss": 1.2636921405792236, "step": 35064 }, { "epoch": 2.7642885297595585, "grad_norm": 0.5428585410118103, "learning_rate": 0.0005, "loss": 1.4191005229949951, "step": 35065 }, { "epoch": 2.7643673630271977, "grad_norm": 0.5352934002876282, "learning_rate": 0.0005, "loss": 1.569016695022583, "step": 35066 }, { "epoch": 2.7644461962948363, "grad_norm": 0.48632463812828064, "learning_rate": 0.0005, "loss": 1.2962656021118164, "step": 35067 }, { "epoch": 2.7645250295624755, "grad_norm": 0.5156857371330261, "learning_rate": 0.0005, "loss": 1.4199142456054688, "step": 35068 }, { "epoch": 2.764603862830114, "grad_norm": 0.5703853964805603, "learning_rate": 0.0005, "loss": 1.3793764114379883, "step": 35069 }, { "epoch": 2.7646826960977533, "grad_norm": 0.5091921091079712, "learning_rate": 0.0005, "loss": 1.124887228012085, "step": 35070 }, { "epoch": 2.7647615293653924, "grad_norm": 1.102402925491333, "learning_rate": 0.0005, "loss": 1.4171645641326904, "step": 35071 }, { "epoch": 2.764840362633031, "grad_norm": 0.5407748818397522, "learning_rate": 0.0005, "loss": 1.2657347917556763, "step": 35072 }, { "epoch": 2.76491919590067, "grad_norm": 0.5514525175094604, "learning_rate": 0.0005, "loss": 1.5849506855010986, "step": 35073 }, { "epoch": 2.764998029168309, "grad_norm": 0.5341234803199768, "learning_rate": 0.0005, "loss": 1.3406866788864136, "step": 35074 }, { "epoch": 2.765076862435948, "grad_norm": 0.48595017194747925, "learning_rate": 0.0005, "loss": 1.2695767879486084, "step": 35075 }, { "epoch": 2.7651556957035868, "grad_norm": 0.49382254481315613, "learning_rate": 0.0005, "loss": 1.2776820659637451, "step": 35076 }, { "epoch": 2.765234528971226, "grad_norm": 0.42396217584609985, "learning_rate": 0.0005, "loss": 1.1287555694580078, "step": 35077 }, { "epoch": 2.7653133622388646, "grad_norm": 0.444036066532135, "learning_rate": 0.0005, "loss": 1.283952236175537, "step": 35078 }, { "epoch": 2.7653921955065037, "grad_norm": 0.5693608522415161, "learning_rate": 0.0005, "loss": 1.4592649936676025, "step": 35079 }, { "epoch": 2.765471028774143, "grad_norm": 0.40219423174858093, "learning_rate": 0.0005, "loss": 1.2730298042297363, "step": 35080 }, { "epoch": 2.7655498620417815, "grad_norm": 0.48690664768218994, "learning_rate": 0.0005, "loss": 1.2143394947052002, "step": 35081 }, { "epoch": 2.7656286953094207, "grad_norm": 0.4605317711830139, "learning_rate": 0.0005, "loss": 1.2229373455047607, "step": 35082 }, { "epoch": 2.7657075285770594, "grad_norm": 0.41482725739479065, "learning_rate": 0.0005, "loss": 1.433519959449768, "step": 35083 }, { "epoch": 2.7657863618446985, "grad_norm": 0.5444282293319702, "learning_rate": 0.0005, "loss": 1.4131346940994263, "step": 35084 }, { "epoch": 2.7658651951123376, "grad_norm": 0.4862494468688965, "learning_rate": 0.0005, "loss": 1.3146367073059082, "step": 35085 }, { "epoch": 2.7659440283799763, "grad_norm": 0.4690781831741333, "learning_rate": 0.0005, "loss": 1.3320488929748535, "step": 35086 }, { "epoch": 2.766022861647615, "grad_norm": 0.6115877032279968, "learning_rate": 0.0005, "loss": 1.4180949926376343, "step": 35087 }, { "epoch": 2.766101694915254, "grad_norm": 0.5405425429344177, "learning_rate": 0.0005, "loss": 1.4937617778778076, "step": 35088 }, { "epoch": 2.7661805281828933, "grad_norm": 0.5387265086174011, "learning_rate": 0.0005, "loss": 1.4386975765228271, "step": 35089 }, { "epoch": 2.766259361450532, "grad_norm": 0.6098085045814514, "learning_rate": 0.0005, "loss": 1.3024687767028809, "step": 35090 }, { "epoch": 2.766338194718171, "grad_norm": 0.4072711169719696, "learning_rate": 0.0005, "loss": 1.221750020980835, "step": 35091 }, { "epoch": 2.76641702798581, "grad_norm": 0.487861692905426, "learning_rate": 0.0005, "loss": 1.4898433685302734, "step": 35092 }, { "epoch": 2.766495861253449, "grad_norm": 0.39705803990364075, "learning_rate": 0.0005, "loss": 1.1681411266326904, "step": 35093 }, { "epoch": 2.766574694521088, "grad_norm": 0.450748085975647, "learning_rate": 0.0005, "loss": 1.362160086631775, "step": 35094 }, { "epoch": 2.7666535277887268, "grad_norm": 0.4426794648170471, "learning_rate": 0.0005, "loss": 1.33367121219635, "step": 35095 }, { "epoch": 2.766732361056366, "grad_norm": 0.42696258425712585, "learning_rate": 0.0005, "loss": 1.275627851486206, "step": 35096 }, { "epoch": 2.7668111943240046, "grad_norm": 0.46052807569503784, "learning_rate": 0.0005, "loss": 1.410545825958252, "step": 35097 }, { "epoch": 2.7668900275916437, "grad_norm": 0.5466827154159546, "learning_rate": 0.0005, "loss": 1.356834888458252, "step": 35098 }, { "epoch": 2.766968860859283, "grad_norm": 0.519511878490448, "learning_rate": 0.0005, "loss": 1.4163825511932373, "step": 35099 }, { "epoch": 2.7670476941269215, "grad_norm": 0.5089267492294312, "learning_rate": 0.0005, "loss": 1.3557732105255127, "step": 35100 }, { "epoch": 2.7671265273945607, "grad_norm": 0.5907735824584961, "learning_rate": 0.0005, "loss": 1.2278330326080322, "step": 35101 }, { "epoch": 2.7672053606621994, "grad_norm": 0.4787764549255371, "learning_rate": 0.0005, "loss": 1.3184082508087158, "step": 35102 }, { "epoch": 2.7672841939298385, "grad_norm": 0.4658931791782379, "learning_rate": 0.0005, "loss": 1.3508580923080444, "step": 35103 }, { "epoch": 2.767363027197477, "grad_norm": 0.565757691860199, "learning_rate": 0.0005, "loss": 1.4306108951568604, "step": 35104 }, { "epoch": 2.7674418604651163, "grad_norm": 0.45381075143814087, "learning_rate": 0.0005, "loss": 1.452641487121582, "step": 35105 }, { "epoch": 2.767520693732755, "grad_norm": 0.4613853394985199, "learning_rate": 0.0005, "loss": 1.2949073314666748, "step": 35106 }, { "epoch": 2.767599527000394, "grad_norm": 0.4226963520050049, "learning_rate": 0.0005, "loss": 1.261305809020996, "step": 35107 }, { "epoch": 2.7676783602680333, "grad_norm": 0.5440008044242859, "learning_rate": 0.0005, "loss": 1.2508409023284912, "step": 35108 }, { "epoch": 2.767757193535672, "grad_norm": 0.6452470421791077, "learning_rate": 0.0005, "loss": 1.4121067523956299, "step": 35109 }, { "epoch": 2.767836026803311, "grad_norm": 0.469504177570343, "learning_rate": 0.0005, "loss": 1.191468358039856, "step": 35110 }, { "epoch": 2.76791486007095, "grad_norm": 0.5031636953353882, "learning_rate": 0.0005, "loss": 1.23740816116333, "step": 35111 }, { "epoch": 2.767993693338589, "grad_norm": 0.6183382272720337, "learning_rate": 0.0005, "loss": 1.4761693477630615, "step": 35112 }, { "epoch": 2.768072526606228, "grad_norm": 0.6411771774291992, "learning_rate": 0.0005, "loss": 1.4528467655181885, "step": 35113 }, { "epoch": 2.7681513598738667, "grad_norm": 0.47061219811439514, "learning_rate": 0.0005, "loss": 1.0917043685913086, "step": 35114 }, { "epoch": 2.768230193141506, "grad_norm": 0.5616072416305542, "learning_rate": 0.0005, "loss": 1.2586568593978882, "step": 35115 }, { "epoch": 2.7683090264091446, "grad_norm": 0.5150337219238281, "learning_rate": 0.0005, "loss": 1.3250871896743774, "step": 35116 }, { "epoch": 2.7683878596767837, "grad_norm": 0.5344514846801758, "learning_rate": 0.0005, "loss": 1.2872260808944702, "step": 35117 }, { "epoch": 2.7684666929444224, "grad_norm": 0.444212943315506, "learning_rate": 0.0005, "loss": 1.188401460647583, "step": 35118 }, { "epoch": 2.7685455262120615, "grad_norm": 0.4372423589229584, "learning_rate": 0.0005, "loss": 1.2871301174163818, "step": 35119 }, { "epoch": 2.7686243594797, "grad_norm": 0.4496679902076721, "learning_rate": 0.0005, "loss": 1.2147244215011597, "step": 35120 }, { "epoch": 2.7687031927473393, "grad_norm": 0.5712377429008484, "learning_rate": 0.0005, "loss": 1.3143558502197266, "step": 35121 }, { "epoch": 2.7687820260149785, "grad_norm": 0.4934203624725342, "learning_rate": 0.0005, "loss": 1.3101212978363037, "step": 35122 }, { "epoch": 2.768860859282617, "grad_norm": 0.5710044503211975, "learning_rate": 0.0005, "loss": 1.3994029760360718, "step": 35123 }, { "epoch": 2.7689396925502563, "grad_norm": 0.4857397675514221, "learning_rate": 0.0005, "loss": 1.289191484451294, "step": 35124 }, { "epoch": 2.769018525817895, "grad_norm": 0.36778244376182556, "learning_rate": 0.0005, "loss": 1.0482945442199707, "step": 35125 }, { "epoch": 2.769097359085534, "grad_norm": 0.5983121991157532, "learning_rate": 0.0005, "loss": 1.2380194664001465, "step": 35126 }, { "epoch": 2.7691761923531732, "grad_norm": 0.5457414388656616, "learning_rate": 0.0005, "loss": 1.4469196796417236, "step": 35127 }, { "epoch": 2.769255025620812, "grad_norm": 0.5188705921173096, "learning_rate": 0.0005, "loss": 1.2721197605133057, "step": 35128 }, { "epoch": 2.769333858888451, "grad_norm": 0.4645552933216095, "learning_rate": 0.0005, "loss": 1.1262147426605225, "step": 35129 }, { "epoch": 2.7694126921560898, "grad_norm": 0.5830397009849548, "learning_rate": 0.0005, "loss": 1.4410045146942139, "step": 35130 }, { "epoch": 2.769491525423729, "grad_norm": 0.48457983136177063, "learning_rate": 0.0005, "loss": 1.1817278861999512, "step": 35131 }, { "epoch": 2.769570358691368, "grad_norm": 0.41293784976005554, "learning_rate": 0.0005, "loss": 1.4548702239990234, "step": 35132 }, { "epoch": 2.7696491919590067, "grad_norm": 0.5188993811607361, "learning_rate": 0.0005, "loss": 1.3916821479797363, "step": 35133 }, { "epoch": 2.7697280252266454, "grad_norm": 0.4975719451904297, "learning_rate": 0.0005, "loss": 1.1990940570831299, "step": 35134 }, { "epoch": 2.7698068584942845, "grad_norm": 0.5041093826293945, "learning_rate": 0.0005, "loss": 1.249525547027588, "step": 35135 }, { "epoch": 2.7698856917619237, "grad_norm": 0.5460905432701111, "learning_rate": 0.0005, "loss": 1.2619037628173828, "step": 35136 }, { "epoch": 2.7699645250295624, "grad_norm": 0.5912001132965088, "learning_rate": 0.0005, "loss": 1.576074481010437, "step": 35137 }, { "epoch": 2.7700433582972015, "grad_norm": 0.6842900514602661, "learning_rate": 0.0005, "loss": 1.258848786354065, "step": 35138 }, { "epoch": 2.77012219156484, "grad_norm": 0.7721131443977356, "learning_rate": 0.0005, "loss": 1.3720966577529907, "step": 35139 }, { "epoch": 2.7702010248324793, "grad_norm": 0.4919585585594177, "learning_rate": 0.0005, "loss": 1.480298399925232, "step": 35140 }, { "epoch": 2.7702798581001185, "grad_norm": 0.6910105347633362, "learning_rate": 0.0005, "loss": 1.2493329048156738, "step": 35141 }, { "epoch": 2.770358691367757, "grad_norm": 0.6108620166778564, "learning_rate": 0.0005, "loss": 1.3996834754943848, "step": 35142 }, { "epoch": 2.7704375246353963, "grad_norm": 0.7070860266685486, "learning_rate": 0.0005, "loss": 1.5305626392364502, "step": 35143 }, { "epoch": 2.770516357903035, "grad_norm": 0.5262183547019958, "learning_rate": 0.0005, "loss": 1.2848591804504395, "step": 35144 }, { "epoch": 2.770595191170674, "grad_norm": 0.5312085747718811, "learning_rate": 0.0005, "loss": 1.2796508073806763, "step": 35145 }, { "epoch": 2.7706740244383132, "grad_norm": 0.45963573455810547, "learning_rate": 0.0005, "loss": 1.3869938850402832, "step": 35146 }, { "epoch": 2.770752857705952, "grad_norm": 0.7286555767059326, "learning_rate": 0.0005, "loss": 1.5051696300506592, "step": 35147 }, { "epoch": 2.7708316909735906, "grad_norm": 0.49262377619743347, "learning_rate": 0.0005, "loss": 1.1819349527359009, "step": 35148 }, { "epoch": 2.7709105242412297, "grad_norm": 0.44196733832359314, "learning_rate": 0.0005, "loss": 1.2908989191055298, "step": 35149 }, { "epoch": 2.770989357508869, "grad_norm": 0.5189231634140015, "learning_rate": 0.0005, "loss": 1.6956264972686768, "step": 35150 }, { "epoch": 2.7710681907765076, "grad_norm": 0.6504504680633545, "learning_rate": 0.0005, "loss": 1.344695806503296, "step": 35151 }, { "epoch": 2.7711470240441467, "grad_norm": 0.4633671045303345, "learning_rate": 0.0005, "loss": 1.285200595855713, "step": 35152 }, { "epoch": 2.7712258573117854, "grad_norm": 0.5375608205795288, "learning_rate": 0.0005, "loss": 1.4314948320388794, "step": 35153 }, { "epoch": 2.7713046905794245, "grad_norm": 0.5884613394737244, "learning_rate": 0.0005, "loss": 1.4503982067108154, "step": 35154 }, { "epoch": 2.7713835238470637, "grad_norm": 0.49741414189338684, "learning_rate": 0.0005, "loss": 1.3555986881256104, "step": 35155 }, { "epoch": 2.7714623571147023, "grad_norm": 0.469963937997818, "learning_rate": 0.0005, "loss": 1.3911950588226318, "step": 35156 }, { "epoch": 2.7715411903823415, "grad_norm": 0.5680777430534363, "learning_rate": 0.0005, "loss": 1.4585630893707275, "step": 35157 }, { "epoch": 2.77162002364998, "grad_norm": 0.4486050307750702, "learning_rate": 0.0005, "loss": 1.316826581954956, "step": 35158 }, { "epoch": 2.7716988569176193, "grad_norm": 0.5643481016159058, "learning_rate": 0.0005, "loss": 1.186463713645935, "step": 35159 }, { "epoch": 2.7717776901852584, "grad_norm": 0.5765794515609741, "learning_rate": 0.0005, "loss": 1.2677559852600098, "step": 35160 }, { "epoch": 2.771856523452897, "grad_norm": 0.5216208100318909, "learning_rate": 0.0005, "loss": 1.3418774604797363, "step": 35161 }, { "epoch": 2.771935356720536, "grad_norm": 0.5336684584617615, "learning_rate": 0.0005, "loss": 1.3946523666381836, "step": 35162 }, { "epoch": 2.772014189988175, "grad_norm": 0.5008546113967896, "learning_rate": 0.0005, "loss": 1.2099533081054688, "step": 35163 }, { "epoch": 2.772093023255814, "grad_norm": 0.58526211977005, "learning_rate": 0.0005, "loss": 1.5988523960113525, "step": 35164 }, { "epoch": 2.7721718565234528, "grad_norm": 0.42901819944381714, "learning_rate": 0.0005, "loss": 1.1991785764694214, "step": 35165 }, { "epoch": 2.772250689791092, "grad_norm": 0.3818058967590332, "learning_rate": 0.0005, "loss": 1.1555043458938599, "step": 35166 }, { "epoch": 2.7723295230587306, "grad_norm": 0.416777104139328, "learning_rate": 0.0005, "loss": 1.3278067111968994, "step": 35167 }, { "epoch": 2.7724083563263697, "grad_norm": 0.45372191071510315, "learning_rate": 0.0005, "loss": 1.2969586849212646, "step": 35168 }, { "epoch": 2.772487189594009, "grad_norm": 0.6223942041397095, "learning_rate": 0.0005, "loss": 1.4042062759399414, "step": 35169 }, { "epoch": 2.7725660228616476, "grad_norm": 0.5692116022109985, "learning_rate": 0.0005, "loss": 1.4644204378128052, "step": 35170 }, { "epoch": 2.7726448561292867, "grad_norm": 0.48071423172950745, "learning_rate": 0.0005, "loss": 1.535355567932129, "step": 35171 }, { "epoch": 2.7727236893969254, "grad_norm": 0.5361442565917969, "learning_rate": 0.0005, "loss": 1.3069415092468262, "step": 35172 }, { "epoch": 2.7728025226645645, "grad_norm": 0.5591797828674316, "learning_rate": 0.0005, "loss": 1.3350346088409424, "step": 35173 }, { "epoch": 2.7728813559322036, "grad_norm": 0.44940102100372314, "learning_rate": 0.0005, "loss": 1.400923490524292, "step": 35174 }, { "epoch": 2.7729601891998423, "grad_norm": 0.5414597988128662, "learning_rate": 0.0005, "loss": 1.3564839363098145, "step": 35175 }, { "epoch": 2.773039022467481, "grad_norm": 0.5784767866134644, "learning_rate": 0.0005, "loss": 1.2363098859786987, "step": 35176 }, { "epoch": 2.77311785573512, "grad_norm": 0.4905666708946228, "learning_rate": 0.0005, "loss": 1.2067781686782837, "step": 35177 }, { "epoch": 2.7731966890027593, "grad_norm": 0.5485530495643616, "learning_rate": 0.0005, "loss": 1.4298491477966309, "step": 35178 }, { "epoch": 2.773275522270398, "grad_norm": 0.49835556745529175, "learning_rate": 0.0005, "loss": 1.1516451835632324, "step": 35179 }, { "epoch": 2.773354355538037, "grad_norm": 0.43931248784065247, "learning_rate": 0.0005, "loss": 1.3616554737091064, "step": 35180 }, { "epoch": 2.773433188805676, "grad_norm": 0.47849041223526, "learning_rate": 0.0005, "loss": 1.2910606861114502, "step": 35181 }, { "epoch": 2.773512022073315, "grad_norm": 0.4567050039768219, "learning_rate": 0.0005, "loss": 1.4158450365066528, "step": 35182 }, { "epoch": 2.773590855340954, "grad_norm": 0.47603124380111694, "learning_rate": 0.0005, "loss": 1.387096881866455, "step": 35183 }, { "epoch": 2.7736696886085928, "grad_norm": 0.481159508228302, "learning_rate": 0.0005, "loss": 1.211350679397583, "step": 35184 }, { "epoch": 2.773748521876232, "grad_norm": 0.674504816532135, "learning_rate": 0.0005, "loss": 1.599351167678833, "step": 35185 }, { "epoch": 2.7738273551438706, "grad_norm": 0.4482910931110382, "learning_rate": 0.0005, "loss": 1.4347937107086182, "step": 35186 }, { "epoch": 2.7739061884115097, "grad_norm": 0.5586877465248108, "learning_rate": 0.0005, "loss": 1.2785325050354004, "step": 35187 }, { "epoch": 2.773985021679149, "grad_norm": 0.45329174399375916, "learning_rate": 0.0005, "loss": 1.3212339878082275, "step": 35188 }, { "epoch": 2.7740638549467875, "grad_norm": 0.6331826448440552, "learning_rate": 0.0005, "loss": 1.499731183052063, "step": 35189 }, { "epoch": 2.7741426882144262, "grad_norm": 0.582955539226532, "learning_rate": 0.0005, "loss": 1.1575833559036255, "step": 35190 }, { "epoch": 2.7742215214820654, "grad_norm": 0.48472097516059875, "learning_rate": 0.0005, "loss": 1.2401355504989624, "step": 35191 }, { "epoch": 2.7743003547497045, "grad_norm": 0.5012452602386475, "learning_rate": 0.0005, "loss": 1.2521495819091797, "step": 35192 }, { "epoch": 2.774379188017343, "grad_norm": 0.5120944380760193, "learning_rate": 0.0005, "loss": 1.3102147579193115, "step": 35193 }, { "epoch": 2.7744580212849823, "grad_norm": 0.607650101184845, "learning_rate": 0.0005, "loss": 1.3257551193237305, "step": 35194 }, { "epoch": 2.774536854552621, "grad_norm": 0.5677324533462524, "learning_rate": 0.0005, "loss": 1.5003228187561035, "step": 35195 }, { "epoch": 2.77461568782026, "grad_norm": 0.5337347388267517, "learning_rate": 0.0005, "loss": 1.4026422500610352, "step": 35196 }, { "epoch": 2.7746945210878993, "grad_norm": 0.4947909414768219, "learning_rate": 0.0005, "loss": 1.304818034172058, "step": 35197 }, { "epoch": 2.774773354355538, "grad_norm": 0.47669845819473267, "learning_rate": 0.0005, "loss": 1.1997807025909424, "step": 35198 }, { "epoch": 2.774852187623177, "grad_norm": 0.5578025579452515, "learning_rate": 0.0005, "loss": 1.2920324802398682, "step": 35199 }, { "epoch": 2.774931020890816, "grad_norm": 0.5727450847625732, "learning_rate": 0.0005, "loss": 1.3643748760223389, "step": 35200 }, { "epoch": 2.775009854158455, "grad_norm": 0.49261415004730225, "learning_rate": 0.0005, "loss": 1.3704073429107666, "step": 35201 }, { "epoch": 2.775088687426094, "grad_norm": 0.5461393594741821, "learning_rate": 0.0005, "loss": 1.2222304344177246, "step": 35202 }, { "epoch": 2.7751675206937327, "grad_norm": 0.4473661780357361, "learning_rate": 0.0005, "loss": 1.2400258779525757, "step": 35203 }, { "epoch": 2.7752463539613714, "grad_norm": 0.46619752049446106, "learning_rate": 0.0005, "loss": 1.3606632947921753, "step": 35204 }, { "epoch": 2.7753251872290106, "grad_norm": 0.4975586533546448, "learning_rate": 0.0005, "loss": 1.2995764017105103, "step": 35205 }, { "epoch": 2.7754040204966497, "grad_norm": 0.4775300920009613, "learning_rate": 0.0005, "loss": 1.333056092262268, "step": 35206 }, { "epoch": 2.7754828537642884, "grad_norm": 0.5057500600814819, "learning_rate": 0.0005, "loss": 1.2413666248321533, "step": 35207 }, { "epoch": 2.7755616870319275, "grad_norm": 0.4449741244316101, "learning_rate": 0.0005, "loss": 1.2306241989135742, "step": 35208 }, { "epoch": 2.775640520299566, "grad_norm": 0.5446974635124207, "learning_rate": 0.0005, "loss": 1.2907742261886597, "step": 35209 }, { "epoch": 2.7757193535672053, "grad_norm": 0.5101200938224792, "learning_rate": 0.0005, "loss": 1.522239327430725, "step": 35210 }, { "epoch": 2.7757981868348445, "grad_norm": 0.5106992721557617, "learning_rate": 0.0005, "loss": 1.3783292770385742, "step": 35211 }, { "epoch": 2.775877020102483, "grad_norm": 0.7909073233604431, "learning_rate": 0.0005, "loss": 1.5738970041275024, "step": 35212 }, { "epoch": 2.7759558533701223, "grad_norm": 0.5114784240722656, "learning_rate": 0.0005, "loss": 1.3719439506530762, "step": 35213 }, { "epoch": 2.776034686637761, "grad_norm": 0.49652016162872314, "learning_rate": 0.0005, "loss": 1.3712594509124756, "step": 35214 }, { "epoch": 2.7761135199054, "grad_norm": 0.4591622054576874, "learning_rate": 0.0005, "loss": 1.280439853668213, "step": 35215 }, { "epoch": 2.7761923531730393, "grad_norm": 0.46116694808006287, "learning_rate": 0.0005, "loss": 1.1926805973052979, "step": 35216 }, { "epoch": 2.776271186440678, "grad_norm": 0.5736504197120667, "learning_rate": 0.0005, "loss": 1.2755074501037598, "step": 35217 }, { "epoch": 2.776350019708317, "grad_norm": 0.5557781457901001, "learning_rate": 0.0005, "loss": 1.6150712966918945, "step": 35218 }, { "epoch": 2.7764288529759558, "grad_norm": 0.4679880142211914, "learning_rate": 0.0005, "loss": 1.3619270324707031, "step": 35219 }, { "epoch": 2.776507686243595, "grad_norm": 0.5015659928321838, "learning_rate": 0.0005, "loss": 1.328482985496521, "step": 35220 }, { "epoch": 2.7765865195112336, "grad_norm": 0.4625122845172882, "learning_rate": 0.0005, "loss": 1.443694829940796, "step": 35221 }, { "epoch": 2.7766653527788727, "grad_norm": 0.41245874762535095, "learning_rate": 0.0005, "loss": 1.3449852466583252, "step": 35222 }, { "epoch": 2.7767441860465114, "grad_norm": 0.5713614225387573, "learning_rate": 0.0005, "loss": 1.5138065814971924, "step": 35223 }, { "epoch": 2.7768230193141505, "grad_norm": 0.42741456627845764, "learning_rate": 0.0005, "loss": 1.119429349899292, "step": 35224 }, { "epoch": 2.7769018525817897, "grad_norm": 0.43935707211494446, "learning_rate": 0.0005, "loss": 1.3119044303894043, "step": 35225 }, { "epoch": 2.7769806858494284, "grad_norm": 0.41575321555137634, "learning_rate": 0.0005, "loss": 1.1446130275726318, "step": 35226 }, { "epoch": 2.7770595191170675, "grad_norm": 0.4609648883342743, "learning_rate": 0.0005, "loss": 1.3804649114608765, "step": 35227 }, { "epoch": 2.777138352384706, "grad_norm": 0.46786239743232727, "learning_rate": 0.0005, "loss": 1.3418385982513428, "step": 35228 }, { "epoch": 2.7772171856523453, "grad_norm": 0.5590161085128784, "learning_rate": 0.0005, "loss": 1.2841041088104248, "step": 35229 }, { "epoch": 2.7772960189199845, "grad_norm": 0.6208571791648865, "learning_rate": 0.0005, "loss": 1.4270634651184082, "step": 35230 }, { "epoch": 2.777374852187623, "grad_norm": 0.4518578350543976, "learning_rate": 0.0005, "loss": 1.1673188209533691, "step": 35231 }, { "epoch": 2.7774536854552623, "grad_norm": 0.45162758231163025, "learning_rate": 0.0005, "loss": 1.4450912475585938, "step": 35232 }, { "epoch": 2.777532518722901, "grad_norm": 0.5464154481887817, "learning_rate": 0.0005, "loss": 1.3411893844604492, "step": 35233 }, { "epoch": 2.77761135199054, "grad_norm": 0.48656293749809265, "learning_rate": 0.0005, "loss": 1.3007166385650635, "step": 35234 }, { "epoch": 2.777690185258179, "grad_norm": 0.6134454011917114, "learning_rate": 0.0005, "loss": 1.3225362300872803, "step": 35235 }, { "epoch": 2.777769018525818, "grad_norm": 0.512703537940979, "learning_rate": 0.0005, "loss": 1.2285236120224, "step": 35236 }, { "epoch": 2.7778478517934566, "grad_norm": 0.5000819563865662, "learning_rate": 0.0005, "loss": 1.4771270751953125, "step": 35237 }, { "epoch": 2.7779266850610957, "grad_norm": 0.3718899190425873, "learning_rate": 0.0005, "loss": 1.1006505489349365, "step": 35238 }, { "epoch": 2.778005518328735, "grad_norm": 0.5825523138046265, "learning_rate": 0.0005, "loss": 1.2504230737686157, "step": 35239 }, { "epoch": 2.7780843515963736, "grad_norm": 0.4324297308921814, "learning_rate": 0.0005, "loss": 1.0982413291931152, "step": 35240 }, { "epoch": 2.7781631848640127, "grad_norm": 0.4726937711238861, "learning_rate": 0.0005, "loss": 1.2728807926177979, "step": 35241 }, { "epoch": 2.7782420181316514, "grad_norm": 0.4402315616607666, "learning_rate": 0.0005, "loss": 1.176191806793213, "step": 35242 }, { "epoch": 2.7783208513992905, "grad_norm": 0.4847760796546936, "learning_rate": 0.0005, "loss": 1.291356086730957, "step": 35243 }, { "epoch": 2.7783996846669297, "grad_norm": 0.6779201626777649, "learning_rate": 0.0005, "loss": 1.2188518047332764, "step": 35244 }, { "epoch": 2.7784785179345683, "grad_norm": 0.5447788238525391, "learning_rate": 0.0005, "loss": 1.4408578872680664, "step": 35245 }, { "epoch": 2.7785573512022075, "grad_norm": 0.44674330949783325, "learning_rate": 0.0005, "loss": 1.359452486038208, "step": 35246 }, { "epoch": 2.778636184469846, "grad_norm": 0.5116195678710938, "learning_rate": 0.0005, "loss": 1.343049168586731, "step": 35247 }, { "epoch": 2.7787150177374853, "grad_norm": 0.41971060633659363, "learning_rate": 0.0005, "loss": 1.285649061203003, "step": 35248 }, { "epoch": 2.7787938510051244, "grad_norm": 0.4915664494037628, "learning_rate": 0.0005, "loss": 1.3501752614974976, "step": 35249 }, { "epoch": 2.778872684272763, "grad_norm": 0.4569299817085266, "learning_rate": 0.0005, "loss": 1.1414179801940918, "step": 35250 }, { "epoch": 2.778951517540402, "grad_norm": 0.4843895435333252, "learning_rate": 0.0005, "loss": 1.2114382982254028, "step": 35251 }, { "epoch": 2.779030350808041, "grad_norm": 0.48325636982917786, "learning_rate": 0.0005, "loss": 1.544126033782959, "step": 35252 }, { "epoch": 2.77910918407568, "grad_norm": 0.40552738308906555, "learning_rate": 0.0005, "loss": 1.099283218383789, "step": 35253 }, { "epoch": 2.7791880173433188, "grad_norm": 0.4809303879737854, "learning_rate": 0.0005, "loss": 1.3303194046020508, "step": 35254 }, { "epoch": 2.779266850610958, "grad_norm": 0.5276346802711487, "learning_rate": 0.0005, "loss": 1.3767600059509277, "step": 35255 }, { "epoch": 2.7793456838785966, "grad_norm": 0.5715621709823608, "learning_rate": 0.0005, "loss": 1.1017348766326904, "step": 35256 }, { "epoch": 2.7794245171462357, "grad_norm": 0.640129566192627, "learning_rate": 0.0005, "loss": 1.4598617553710938, "step": 35257 }, { "epoch": 2.779503350413875, "grad_norm": 0.4369833767414093, "learning_rate": 0.0005, "loss": 1.2285585403442383, "step": 35258 }, { "epoch": 2.7795821836815136, "grad_norm": 0.771152913570404, "learning_rate": 0.0005, "loss": 1.4054505825042725, "step": 35259 }, { "epoch": 2.7796610169491527, "grad_norm": 0.4926565885543823, "learning_rate": 0.0005, "loss": 1.1758782863616943, "step": 35260 }, { "epoch": 2.7797398502167914, "grad_norm": 0.5232011079788208, "learning_rate": 0.0005, "loss": 1.4293327331542969, "step": 35261 }, { "epoch": 2.7798186834844305, "grad_norm": 0.44930827617645264, "learning_rate": 0.0005, "loss": 1.20750093460083, "step": 35262 }, { "epoch": 2.7798975167520696, "grad_norm": 0.5671574473381042, "learning_rate": 0.0005, "loss": 1.4557979106903076, "step": 35263 }, { "epoch": 2.7799763500197083, "grad_norm": 0.4984022080898285, "learning_rate": 0.0005, "loss": 1.2983168363571167, "step": 35264 }, { "epoch": 2.780055183287347, "grad_norm": 0.5354489684104919, "learning_rate": 0.0005, "loss": 1.419834852218628, "step": 35265 }, { "epoch": 2.780134016554986, "grad_norm": 0.4075525104999542, "learning_rate": 0.0005, "loss": 1.1809718608856201, "step": 35266 }, { "epoch": 2.7802128498226253, "grad_norm": 0.6233635544776917, "learning_rate": 0.0005, "loss": 1.5011043548583984, "step": 35267 }, { "epoch": 2.780291683090264, "grad_norm": 0.45893412828445435, "learning_rate": 0.0005, "loss": 1.1163023710250854, "step": 35268 }, { "epoch": 2.780370516357903, "grad_norm": 0.5561407804489136, "learning_rate": 0.0005, "loss": 1.3811194896697998, "step": 35269 }, { "epoch": 2.780449349625542, "grad_norm": 0.45073267817497253, "learning_rate": 0.0005, "loss": 1.1209512948989868, "step": 35270 }, { "epoch": 2.780528182893181, "grad_norm": 0.6213690638542175, "learning_rate": 0.0005, "loss": 1.3337552547454834, "step": 35271 }, { "epoch": 2.78060701616082, "grad_norm": 0.47132691740989685, "learning_rate": 0.0005, "loss": 1.3625545501708984, "step": 35272 }, { "epoch": 2.7806858494284588, "grad_norm": 0.5062557458877563, "learning_rate": 0.0005, "loss": 1.3009570837020874, "step": 35273 }, { "epoch": 2.780764682696098, "grad_norm": 0.46844685077667236, "learning_rate": 0.0005, "loss": 1.3104546070098877, "step": 35274 }, { "epoch": 2.7808435159637366, "grad_norm": 0.5330522656440735, "learning_rate": 0.0005, "loss": 1.4833276271820068, "step": 35275 }, { "epoch": 2.7809223492313757, "grad_norm": 0.5497084856033325, "learning_rate": 0.0005, "loss": 1.1613695621490479, "step": 35276 }, { "epoch": 2.781001182499015, "grad_norm": 0.5273580551147461, "learning_rate": 0.0005, "loss": 1.2242324352264404, "step": 35277 }, { "epoch": 2.7810800157666535, "grad_norm": 0.5404471158981323, "learning_rate": 0.0005, "loss": 1.3988656997680664, "step": 35278 }, { "epoch": 2.7811588490342922, "grad_norm": 0.4144365191459656, "learning_rate": 0.0005, "loss": 1.2614887952804565, "step": 35279 }, { "epoch": 2.7812376823019314, "grad_norm": 0.461905837059021, "learning_rate": 0.0005, "loss": 1.3222298622131348, "step": 35280 }, { "epoch": 2.7813165155695705, "grad_norm": 0.4448525011539459, "learning_rate": 0.0005, "loss": 1.281691074371338, "step": 35281 }, { "epoch": 2.781395348837209, "grad_norm": 0.5648767352104187, "learning_rate": 0.0005, "loss": 1.2114951610565186, "step": 35282 }, { "epoch": 2.7814741821048483, "grad_norm": 0.515774130821228, "learning_rate": 0.0005, "loss": 1.2909047603607178, "step": 35283 }, { "epoch": 2.781553015372487, "grad_norm": 0.6170539259910583, "learning_rate": 0.0005, "loss": 1.179885983467102, "step": 35284 }, { "epoch": 2.781631848640126, "grad_norm": 0.4360034763813019, "learning_rate": 0.0005, "loss": 1.2623587846755981, "step": 35285 }, { "epoch": 2.7817106819077653, "grad_norm": 0.6480051279067993, "learning_rate": 0.0005, "loss": 1.3727052211761475, "step": 35286 }, { "epoch": 2.781789515175404, "grad_norm": 0.46969521045684814, "learning_rate": 0.0005, "loss": 1.2079497575759888, "step": 35287 }, { "epoch": 2.781868348443043, "grad_norm": 0.5709038972854614, "learning_rate": 0.0005, "loss": 1.5456035137176514, "step": 35288 }, { "epoch": 2.781947181710682, "grad_norm": 0.5063547492027283, "learning_rate": 0.0005, "loss": 1.2284613847732544, "step": 35289 }, { "epoch": 2.782026014978321, "grad_norm": 0.49435412883758545, "learning_rate": 0.0005, "loss": 1.2462520599365234, "step": 35290 }, { "epoch": 2.78210484824596, "grad_norm": 0.5162838697433472, "learning_rate": 0.0005, "loss": 1.409012794494629, "step": 35291 }, { "epoch": 2.7821836815135987, "grad_norm": 0.5202926993370056, "learning_rate": 0.0005, "loss": 1.2332844734191895, "step": 35292 }, { "epoch": 2.7822625147812374, "grad_norm": 0.6459175944328308, "learning_rate": 0.0005, "loss": 1.3597389459609985, "step": 35293 }, { "epoch": 2.7823413480488766, "grad_norm": 0.5989434123039246, "learning_rate": 0.0005, "loss": 1.4588310718536377, "step": 35294 }, { "epoch": 2.7824201813165157, "grad_norm": 0.5240166783332825, "learning_rate": 0.0005, "loss": 1.185306429862976, "step": 35295 }, { "epoch": 2.7824990145841544, "grad_norm": 0.48237061500549316, "learning_rate": 0.0005, "loss": 1.2149637937545776, "step": 35296 }, { "epoch": 2.7825778478517935, "grad_norm": 0.4367121756076813, "learning_rate": 0.0005, "loss": 1.1645259857177734, "step": 35297 }, { "epoch": 2.782656681119432, "grad_norm": 0.514471709728241, "learning_rate": 0.0005, "loss": 1.3650598526000977, "step": 35298 }, { "epoch": 2.7827355143870713, "grad_norm": 0.5517674088478088, "learning_rate": 0.0005, "loss": 1.4666587114334106, "step": 35299 }, { "epoch": 2.7828143476547105, "grad_norm": 0.5400115251541138, "learning_rate": 0.0005, "loss": 1.2533292770385742, "step": 35300 }, { "epoch": 2.782893180922349, "grad_norm": 0.47886744141578674, "learning_rate": 0.0005, "loss": 1.3720005750656128, "step": 35301 }, { "epoch": 2.7829720141899883, "grad_norm": 0.42047011852264404, "learning_rate": 0.0005, "loss": 1.3630344867706299, "step": 35302 }, { "epoch": 2.783050847457627, "grad_norm": 0.5348663926124573, "learning_rate": 0.0005, "loss": 1.5779738426208496, "step": 35303 }, { "epoch": 2.783129680725266, "grad_norm": 0.447198748588562, "learning_rate": 0.0005, "loss": 1.3882408142089844, "step": 35304 }, { "epoch": 2.7832085139929053, "grad_norm": 0.6287176609039307, "learning_rate": 0.0005, "loss": 1.6118288040161133, "step": 35305 }, { "epoch": 2.783287347260544, "grad_norm": 0.4210548996925354, "learning_rate": 0.0005, "loss": 1.2152442932128906, "step": 35306 }, { "epoch": 2.7833661805281826, "grad_norm": 0.48279574513435364, "learning_rate": 0.0005, "loss": 1.1652371883392334, "step": 35307 }, { "epoch": 2.7834450137958218, "grad_norm": 0.9538668394088745, "learning_rate": 0.0005, "loss": 1.4866065979003906, "step": 35308 }, { "epoch": 2.783523847063461, "grad_norm": 0.5461913347244263, "learning_rate": 0.0005, "loss": 1.2588063478469849, "step": 35309 }, { "epoch": 2.7836026803310996, "grad_norm": 0.5937591195106506, "learning_rate": 0.0005, "loss": 1.412839651107788, "step": 35310 }, { "epoch": 2.7836815135987387, "grad_norm": 0.5593531131744385, "learning_rate": 0.0005, "loss": 1.349914312362671, "step": 35311 }, { "epoch": 2.7837603468663774, "grad_norm": 0.7717745900154114, "learning_rate": 0.0005, "loss": 1.2358746528625488, "step": 35312 }, { "epoch": 2.7838391801340165, "grad_norm": 0.49736592173576355, "learning_rate": 0.0005, "loss": 1.3227434158325195, "step": 35313 }, { "epoch": 2.7839180134016557, "grad_norm": 0.7382441163063049, "learning_rate": 0.0005, "loss": 1.3899701833724976, "step": 35314 }, { "epoch": 2.7839968466692944, "grad_norm": 0.5545008182525635, "learning_rate": 0.0005, "loss": 1.3657478094100952, "step": 35315 }, { "epoch": 2.7840756799369335, "grad_norm": 0.4507749378681183, "learning_rate": 0.0005, "loss": 1.3774609565734863, "step": 35316 }, { "epoch": 2.784154513204572, "grad_norm": 0.6403915286064148, "learning_rate": 0.0005, "loss": 1.3255105018615723, "step": 35317 }, { "epoch": 2.7842333464722113, "grad_norm": 0.4605664312839508, "learning_rate": 0.0005, "loss": 1.38138747215271, "step": 35318 }, { "epoch": 2.7843121797398505, "grad_norm": 0.47178345918655396, "learning_rate": 0.0005, "loss": 1.3508925437927246, "step": 35319 }, { "epoch": 2.784391013007489, "grad_norm": 0.744678795337677, "learning_rate": 0.0005, "loss": 1.276748538017273, "step": 35320 }, { "epoch": 2.784469846275128, "grad_norm": 0.554047167301178, "learning_rate": 0.0005, "loss": 1.376812219619751, "step": 35321 }, { "epoch": 2.784548679542767, "grad_norm": 0.45421090722084045, "learning_rate": 0.0005, "loss": 1.2154293060302734, "step": 35322 }, { "epoch": 2.784627512810406, "grad_norm": 0.7461654543876648, "learning_rate": 0.0005, "loss": 1.460400104522705, "step": 35323 }, { "epoch": 2.784706346078045, "grad_norm": 0.6529826521873474, "learning_rate": 0.0005, "loss": 1.537516713142395, "step": 35324 }, { "epoch": 2.784785179345684, "grad_norm": 0.48492318391799927, "learning_rate": 0.0005, "loss": 1.1794705390930176, "step": 35325 }, { "epoch": 2.7848640126133226, "grad_norm": 0.5008412003517151, "learning_rate": 0.0005, "loss": 1.2756141424179077, "step": 35326 }, { "epoch": 2.7849428458809617, "grad_norm": 0.5941195487976074, "learning_rate": 0.0005, "loss": 1.4626786708831787, "step": 35327 }, { "epoch": 2.785021679148601, "grad_norm": 0.5521820783615112, "learning_rate": 0.0005, "loss": 1.3642668724060059, "step": 35328 }, { "epoch": 2.7851005124162396, "grad_norm": 0.46914514899253845, "learning_rate": 0.0005, "loss": 1.1476433277130127, "step": 35329 }, { "epoch": 2.7851793456838787, "grad_norm": 0.5072062015533447, "learning_rate": 0.0005, "loss": 1.2166013717651367, "step": 35330 }, { "epoch": 2.7852581789515174, "grad_norm": 0.4799390733242035, "learning_rate": 0.0005, "loss": 1.196735143661499, "step": 35331 }, { "epoch": 2.7853370122191565, "grad_norm": 0.9407156109809875, "learning_rate": 0.0005, "loss": 1.3159480094909668, "step": 35332 }, { "epoch": 2.7854158454867957, "grad_norm": 0.5428552031517029, "learning_rate": 0.0005, "loss": 1.2648952007293701, "step": 35333 }, { "epoch": 2.7854946787544344, "grad_norm": 0.8456439971923828, "learning_rate": 0.0005, "loss": 1.410865306854248, "step": 35334 }, { "epoch": 2.785573512022073, "grad_norm": 0.6884878277778625, "learning_rate": 0.0005, "loss": 1.4888975620269775, "step": 35335 }, { "epoch": 2.785652345289712, "grad_norm": 0.4568055272102356, "learning_rate": 0.0005, "loss": 1.3935184478759766, "step": 35336 }, { "epoch": 2.7857311785573513, "grad_norm": 0.5226662755012512, "learning_rate": 0.0005, "loss": 1.4429597854614258, "step": 35337 }, { "epoch": 2.78581001182499, "grad_norm": 0.4831516444683075, "learning_rate": 0.0005, "loss": 1.2735882997512817, "step": 35338 }, { "epoch": 2.785888845092629, "grad_norm": 0.7170106172561646, "learning_rate": 0.0005, "loss": 1.1742924451828003, "step": 35339 }, { "epoch": 2.785967678360268, "grad_norm": 0.6104686856269836, "learning_rate": 0.0005, "loss": 1.3682129383087158, "step": 35340 }, { "epoch": 2.786046511627907, "grad_norm": 0.6470192074775696, "learning_rate": 0.0005, "loss": 1.4633653163909912, "step": 35341 }, { "epoch": 2.786125344895546, "grad_norm": 0.5696902871131897, "learning_rate": 0.0005, "loss": 1.3270525932312012, "step": 35342 }, { "epoch": 2.7862041781631848, "grad_norm": 0.4434334635734558, "learning_rate": 0.0005, "loss": 1.2283241748809814, "step": 35343 }, { "epoch": 2.786283011430824, "grad_norm": 0.4577658474445343, "learning_rate": 0.0005, "loss": 1.2620782852172852, "step": 35344 }, { "epoch": 2.7863618446984626, "grad_norm": 0.49993056058883667, "learning_rate": 0.0005, "loss": 1.3112680912017822, "step": 35345 }, { "epoch": 2.7864406779661017, "grad_norm": 0.4357265532016754, "learning_rate": 0.0005, "loss": 1.3892273902893066, "step": 35346 }, { "epoch": 2.786519511233741, "grad_norm": 0.5375564098358154, "learning_rate": 0.0005, "loss": 1.2883026599884033, "step": 35347 }, { "epoch": 2.7865983445013796, "grad_norm": 0.5572046041488647, "learning_rate": 0.0005, "loss": 1.4871392250061035, "step": 35348 }, { "epoch": 2.7866771777690187, "grad_norm": 0.5206170678138733, "learning_rate": 0.0005, "loss": 1.3463828563690186, "step": 35349 }, { "epoch": 2.7867560110366574, "grad_norm": 0.4902259111404419, "learning_rate": 0.0005, "loss": 1.1556860208511353, "step": 35350 }, { "epoch": 2.7868348443042965, "grad_norm": 0.45940113067626953, "learning_rate": 0.0005, "loss": 1.423339605331421, "step": 35351 }, { "epoch": 2.786913677571935, "grad_norm": 0.47839707136154175, "learning_rate": 0.0005, "loss": 1.3034685850143433, "step": 35352 }, { "epoch": 2.7869925108395743, "grad_norm": 0.5642824172973633, "learning_rate": 0.0005, "loss": 1.3611433506011963, "step": 35353 }, { "epoch": 2.787071344107213, "grad_norm": 0.6189478635787964, "learning_rate": 0.0005, "loss": 1.4088176488876343, "step": 35354 }, { "epoch": 2.787150177374852, "grad_norm": 0.5075517296791077, "learning_rate": 0.0005, "loss": 1.5130928754806519, "step": 35355 }, { "epoch": 2.7872290106424913, "grad_norm": 0.4533331096172333, "learning_rate": 0.0005, "loss": 1.3336126804351807, "step": 35356 }, { "epoch": 2.78730784391013, "grad_norm": 0.47656604647636414, "learning_rate": 0.0005, "loss": 1.2354298830032349, "step": 35357 }, { "epoch": 2.787386677177769, "grad_norm": 0.500564694404602, "learning_rate": 0.0005, "loss": 1.2986801862716675, "step": 35358 }, { "epoch": 2.787465510445408, "grad_norm": 0.594068169593811, "learning_rate": 0.0005, "loss": 1.6658990383148193, "step": 35359 }, { "epoch": 2.787544343713047, "grad_norm": 0.5000338554382324, "learning_rate": 0.0005, "loss": 1.3636350631713867, "step": 35360 }, { "epoch": 2.787623176980686, "grad_norm": 0.5127966403961182, "learning_rate": 0.0005, "loss": 1.4044537544250488, "step": 35361 }, { "epoch": 2.7877020102483248, "grad_norm": 0.5022414922714233, "learning_rate": 0.0005, "loss": 1.3809216022491455, "step": 35362 }, { "epoch": 2.787780843515964, "grad_norm": 0.508131206035614, "learning_rate": 0.0005, "loss": 1.294800043106079, "step": 35363 }, { "epoch": 2.7878596767836026, "grad_norm": 0.5385568141937256, "learning_rate": 0.0005, "loss": 1.3429614305496216, "step": 35364 }, { "epoch": 2.7879385100512417, "grad_norm": 0.5529385805130005, "learning_rate": 0.0005, "loss": 1.1000748872756958, "step": 35365 }, { "epoch": 2.7880173433188804, "grad_norm": 0.6117897629737854, "learning_rate": 0.0005, "loss": 1.3955382108688354, "step": 35366 }, { "epoch": 2.7880961765865195, "grad_norm": 0.5549578070640564, "learning_rate": 0.0005, "loss": 1.483581304550171, "step": 35367 }, { "epoch": 2.7881750098541582, "grad_norm": 0.46683040261268616, "learning_rate": 0.0005, "loss": 1.363358736038208, "step": 35368 }, { "epoch": 2.7882538431217974, "grad_norm": 0.5114413499832153, "learning_rate": 0.0005, "loss": 1.2298866510391235, "step": 35369 }, { "epoch": 2.7883326763894365, "grad_norm": 0.40503808856010437, "learning_rate": 0.0005, "loss": 1.2507009506225586, "step": 35370 }, { "epoch": 2.788411509657075, "grad_norm": 0.5863260626792908, "learning_rate": 0.0005, "loss": 1.3503280878067017, "step": 35371 }, { "epoch": 2.7884903429247143, "grad_norm": 0.4963086247444153, "learning_rate": 0.0005, "loss": 1.2490119934082031, "step": 35372 }, { "epoch": 2.788569176192353, "grad_norm": 0.5729318857192993, "learning_rate": 0.0005, "loss": 1.3181126117706299, "step": 35373 }, { "epoch": 2.788648009459992, "grad_norm": 0.7526267170906067, "learning_rate": 0.0005, "loss": 1.4005249738693237, "step": 35374 }, { "epoch": 2.7887268427276313, "grad_norm": 0.5035160779953003, "learning_rate": 0.0005, "loss": 1.34525465965271, "step": 35375 }, { "epoch": 2.78880567599527, "grad_norm": 0.5170647501945496, "learning_rate": 0.0005, "loss": 1.2483818531036377, "step": 35376 }, { "epoch": 2.788884509262909, "grad_norm": 0.4928140640258789, "learning_rate": 0.0005, "loss": 1.2674673795700073, "step": 35377 }, { "epoch": 2.788963342530548, "grad_norm": 0.4961784780025482, "learning_rate": 0.0005, "loss": 1.4156103134155273, "step": 35378 }, { "epoch": 2.789042175798187, "grad_norm": 0.45513027906417847, "learning_rate": 0.0005, "loss": 1.2096115350723267, "step": 35379 }, { "epoch": 2.789121009065826, "grad_norm": 0.92259681224823, "learning_rate": 0.0005, "loss": 1.1683920621871948, "step": 35380 }, { "epoch": 2.7891998423334647, "grad_norm": 0.5009371638298035, "learning_rate": 0.0005, "loss": 1.2666451930999756, "step": 35381 }, { "epoch": 2.7892786756011034, "grad_norm": 0.4387284219264984, "learning_rate": 0.0005, "loss": 1.144205093383789, "step": 35382 }, { "epoch": 2.7893575088687426, "grad_norm": 0.5213704705238342, "learning_rate": 0.0005, "loss": 1.2520620822906494, "step": 35383 }, { "epoch": 2.7894363421363817, "grad_norm": 0.5695057511329651, "learning_rate": 0.0005, "loss": 1.4311389923095703, "step": 35384 }, { "epoch": 2.7895151754040204, "grad_norm": 0.5722025036811829, "learning_rate": 0.0005, "loss": 1.435600757598877, "step": 35385 }, { "epoch": 2.7895940086716595, "grad_norm": 0.48790889978408813, "learning_rate": 0.0005, "loss": 1.2181282043457031, "step": 35386 }, { "epoch": 2.789672841939298, "grad_norm": 0.6534649133682251, "learning_rate": 0.0005, "loss": 1.2867562770843506, "step": 35387 }, { "epoch": 2.7897516752069373, "grad_norm": 0.5616055130958557, "learning_rate": 0.0005, "loss": 1.4117956161499023, "step": 35388 }, { "epoch": 2.7898305084745765, "grad_norm": 0.5306032299995422, "learning_rate": 0.0005, "loss": 1.4770007133483887, "step": 35389 }, { "epoch": 2.789909341742215, "grad_norm": 0.49331119656562805, "learning_rate": 0.0005, "loss": 1.3949458599090576, "step": 35390 }, { "epoch": 2.7899881750098543, "grad_norm": 0.5206998586654663, "learning_rate": 0.0005, "loss": 1.0513207912445068, "step": 35391 }, { "epoch": 2.790067008277493, "grad_norm": 0.5706011056900024, "learning_rate": 0.0005, "loss": 1.275368332862854, "step": 35392 }, { "epoch": 2.790145841545132, "grad_norm": 0.5276108980178833, "learning_rate": 0.0005, "loss": 1.1801315546035767, "step": 35393 }, { "epoch": 2.7902246748127713, "grad_norm": 0.43846267461776733, "learning_rate": 0.0005, "loss": 1.275244116783142, "step": 35394 }, { "epoch": 2.79030350808041, "grad_norm": 0.43341726064682007, "learning_rate": 0.0005, "loss": 1.313319444656372, "step": 35395 }, { "epoch": 2.7903823413480486, "grad_norm": 0.5024873614311218, "learning_rate": 0.0005, "loss": 1.3321349620819092, "step": 35396 }, { "epoch": 2.7904611746156878, "grad_norm": 0.45027393102645874, "learning_rate": 0.0005, "loss": 1.088753342628479, "step": 35397 }, { "epoch": 2.790540007883327, "grad_norm": 0.5718066096305847, "learning_rate": 0.0005, "loss": 1.3533034324645996, "step": 35398 }, { "epoch": 2.7906188411509656, "grad_norm": 0.5010195374488831, "learning_rate": 0.0005, "loss": 1.293535828590393, "step": 35399 }, { "epoch": 2.7906976744186047, "grad_norm": 0.48105019330978394, "learning_rate": 0.0005, "loss": 1.232562780380249, "step": 35400 }, { "epoch": 2.7907765076862434, "grad_norm": 0.39264070987701416, "learning_rate": 0.0005, "loss": 1.1717435121536255, "step": 35401 }, { "epoch": 2.7908553409538825, "grad_norm": 0.5599244236946106, "learning_rate": 0.0005, "loss": 1.3809516429901123, "step": 35402 }, { "epoch": 2.7909341742215217, "grad_norm": 0.4709784984588623, "learning_rate": 0.0005, "loss": 1.3210721015930176, "step": 35403 }, { "epoch": 2.7910130074891604, "grad_norm": 0.5019218921661377, "learning_rate": 0.0005, "loss": 1.2541319131851196, "step": 35404 }, { "epoch": 2.7910918407567995, "grad_norm": 0.6459190249443054, "learning_rate": 0.0005, "loss": 1.5064746141433716, "step": 35405 }, { "epoch": 2.791170674024438, "grad_norm": 0.5438527464866638, "learning_rate": 0.0005, "loss": 1.5081852674484253, "step": 35406 }, { "epoch": 2.7912495072920773, "grad_norm": 0.5700694918632507, "learning_rate": 0.0005, "loss": 1.313455581665039, "step": 35407 }, { "epoch": 2.7913283405597165, "grad_norm": 0.5505872368812561, "learning_rate": 0.0005, "loss": 1.382063865661621, "step": 35408 }, { "epoch": 2.791407173827355, "grad_norm": 0.5136450529098511, "learning_rate": 0.0005, "loss": 1.3592162132263184, "step": 35409 }, { "epoch": 2.791486007094994, "grad_norm": 0.5419811606407166, "learning_rate": 0.0005, "loss": 1.2935913801193237, "step": 35410 }, { "epoch": 2.791564840362633, "grad_norm": 0.60152667760849, "learning_rate": 0.0005, "loss": 1.6476044654846191, "step": 35411 }, { "epoch": 2.791643673630272, "grad_norm": 0.4995541572570801, "learning_rate": 0.0005, "loss": 1.3622816801071167, "step": 35412 }, { "epoch": 2.791722506897911, "grad_norm": 0.4588533937931061, "learning_rate": 0.0005, "loss": 1.2675889730453491, "step": 35413 }, { "epoch": 2.79180134016555, "grad_norm": 0.450164794921875, "learning_rate": 0.0005, "loss": 1.4247560501098633, "step": 35414 }, { "epoch": 2.7918801734331886, "grad_norm": 0.49735385179519653, "learning_rate": 0.0005, "loss": 1.432377815246582, "step": 35415 }, { "epoch": 2.7919590067008278, "grad_norm": 0.6134056448936462, "learning_rate": 0.0005, "loss": 1.4015555381774902, "step": 35416 }, { "epoch": 2.792037839968467, "grad_norm": 0.4687129557132721, "learning_rate": 0.0005, "loss": 1.2767136096954346, "step": 35417 }, { "epoch": 2.7921166732361056, "grad_norm": 0.5523917078971863, "learning_rate": 0.0005, "loss": 1.241296410560608, "step": 35418 }, { "epoch": 2.7921955065037447, "grad_norm": 0.49407684803009033, "learning_rate": 0.0005, "loss": 1.2822511196136475, "step": 35419 }, { "epoch": 2.7922743397713834, "grad_norm": 0.4702829122543335, "learning_rate": 0.0005, "loss": 1.283189296722412, "step": 35420 }, { "epoch": 2.7923531730390225, "grad_norm": 0.4849196970462799, "learning_rate": 0.0005, "loss": 1.3452156782150269, "step": 35421 }, { "epoch": 2.7924320063066617, "grad_norm": 0.5018777251243591, "learning_rate": 0.0005, "loss": 1.2069892883300781, "step": 35422 }, { "epoch": 2.7925108395743004, "grad_norm": 0.503998875617981, "learning_rate": 0.0005, "loss": 1.3969604969024658, "step": 35423 }, { "epoch": 2.792589672841939, "grad_norm": 0.47258907556533813, "learning_rate": 0.0005, "loss": 1.332491159439087, "step": 35424 }, { "epoch": 2.792668506109578, "grad_norm": 0.47676023840904236, "learning_rate": 0.0005, "loss": 1.3177685737609863, "step": 35425 }, { "epoch": 2.7927473393772173, "grad_norm": 0.43681788444519043, "learning_rate": 0.0005, "loss": 1.1255451440811157, "step": 35426 }, { "epoch": 2.792826172644856, "grad_norm": 0.40964221954345703, "learning_rate": 0.0005, "loss": 1.2041959762573242, "step": 35427 }, { "epoch": 2.792905005912495, "grad_norm": 0.48563575744628906, "learning_rate": 0.0005, "loss": 1.4775028228759766, "step": 35428 }, { "epoch": 2.792983839180134, "grad_norm": 0.3997786045074463, "learning_rate": 0.0005, "loss": 1.140170931816101, "step": 35429 }, { "epoch": 2.793062672447773, "grad_norm": 0.5036562085151672, "learning_rate": 0.0005, "loss": 1.3761169910430908, "step": 35430 }, { "epoch": 2.793141505715412, "grad_norm": 0.42859041690826416, "learning_rate": 0.0005, "loss": 1.3220269680023193, "step": 35431 }, { "epoch": 2.7932203389830508, "grad_norm": 0.4767189621925354, "learning_rate": 0.0005, "loss": 1.1758668422698975, "step": 35432 }, { "epoch": 2.79329917225069, "grad_norm": 0.48391056060791016, "learning_rate": 0.0005, "loss": 1.3447626829147339, "step": 35433 }, { "epoch": 2.7933780055183286, "grad_norm": 0.5525660514831543, "learning_rate": 0.0005, "loss": 1.366556167602539, "step": 35434 }, { "epoch": 2.7934568387859677, "grad_norm": 0.5423786044120789, "learning_rate": 0.0005, "loss": 1.4476335048675537, "step": 35435 }, { "epoch": 2.793535672053607, "grad_norm": 0.47099652886390686, "learning_rate": 0.0005, "loss": 1.3528815507888794, "step": 35436 }, { "epoch": 2.7936145053212456, "grad_norm": 0.532637357711792, "learning_rate": 0.0005, "loss": 1.1212666034698486, "step": 35437 }, { "epoch": 2.7936933385888842, "grad_norm": 0.46810051798820496, "learning_rate": 0.0005, "loss": 1.3453152179718018, "step": 35438 }, { "epoch": 2.7937721718565234, "grad_norm": 0.4251693785190582, "learning_rate": 0.0005, "loss": 1.2823083400726318, "step": 35439 }, { "epoch": 2.7938510051241625, "grad_norm": 0.5039617419242859, "learning_rate": 0.0005, "loss": 1.4573856592178345, "step": 35440 }, { "epoch": 2.793929838391801, "grad_norm": 0.4520880877971649, "learning_rate": 0.0005, "loss": 1.1934144496917725, "step": 35441 }, { "epoch": 2.7940086716594403, "grad_norm": 0.6995242238044739, "learning_rate": 0.0005, "loss": 1.4064817428588867, "step": 35442 }, { "epoch": 2.794087504927079, "grad_norm": 0.8927081823348999, "learning_rate": 0.0005, "loss": 1.6156840324401855, "step": 35443 }, { "epoch": 2.794166338194718, "grad_norm": 0.5814045667648315, "learning_rate": 0.0005, "loss": 1.364278793334961, "step": 35444 }, { "epoch": 2.7942451714623573, "grad_norm": 0.5120818614959717, "learning_rate": 0.0005, "loss": 1.365235447883606, "step": 35445 }, { "epoch": 2.794324004729996, "grad_norm": 0.7997216582298279, "learning_rate": 0.0005, "loss": 1.1507574319839478, "step": 35446 }, { "epoch": 2.794402837997635, "grad_norm": 0.6512022018432617, "learning_rate": 0.0005, "loss": 1.2646076679229736, "step": 35447 }, { "epoch": 2.794481671265274, "grad_norm": 0.6306304931640625, "learning_rate": 0.0005, "loss": 1.5103302001953125, "step": 35448 }, { "epoch": 2.794560504532913, "grad_norm": 0.5908418297767639, "learning_rate": 0.0005, "loss": 1.40375816822052, "step": 35449 }, { "epoch": 2.794639337800552, "grad_norm": 0.5207655429840088, "learning_rate": 0.0005, "loss": 1.4513194561004639, "step": 35450 }, { "epoch": 2.7947181710681908, "grad_norm": 0.5828113555908203, "learning_rate": 0.0005, "loss": 1.451857566833496, "step": 35451 }, { "epoch": 2.7947970043358294, "grad_norm": 0.5666357278823853, "learning_rate": 0.0005, "loss": 1.3600163459777832, "step": 35452 }, { "epoch": 2.7948758376034686, "grad_norm": 0.4737306237220764, "learning_rate": 0.0005, "loss": 1.3102306127548218, "step": 35453 }, { "epoch": 2.7949546708711077, "grad_norm": 0.5868045091629028, "learning_rate": 0.0005, "loss": 1.482133150100708, "step": 35454 }, { "epoch": 2.7950335041387464, "grad_norm": 0.3946956992149353, "learning_rate": 0.0005, "loss": 1.1080667972564697, "step": 35455 }, { "epoch": 2.7951123374063855, "grad_norm": 0.49790310859680176, "learning_rate": 0.0005, "loss": 1.3538222312927246, "step": 35456 }, { "epoch": 2.7951911706740242, "grad_norm": 0.5488779544830322, "learning_rate": 0.0005, "loss": 1.356139898300171, "step": 35457 }, { "epoch": 2.7952700039416634, "grad_norm": 0.5733841061592102, "learning_rate": 0.0005, "loss": 1.4663832187652588, "step": 35458 }, { "epoch": 2.7953488372093025, "grad_norm": 0.46004974842071533, "learning_rate": 0.0005, "loss": 1.243914246559143, "step": 35459 }, { "epoch": 2.795427670476941, "grad_norm": 0.5230311751365662, "learning_rate": 0.0005, "loss": 1.4042303562164307, "step": 35460 }, { "epoch": 2.7955065037445803, "grad_norm": 0.49551624059677124, "learning_rate": 0.0005, "loss": 1.3418545722961426, "step": 35461 }, { "epoch": 2.795585337012219, "grad_norm": 0.7010484337806702, "learning_rate": 0.0005, "loss": 1.1910929679870605, "step": 35462 }, { "epoch": 2.795664170279858, "grad_norm": 0.5198555588722229, "learning_rate": 0.0005, "loss": 1.414612054824829, "step": 35463 }, { "epoch": 2.7957430035474973, "grad_norm": 0.5941894054412842, "learning_rate": 0.0005, "loss": 1.2752940654754639, "step": 35464 }, { "epoch": 2.795821836815136, "grad_norm": 0.526512086391449, "learning_rate": 0.0005, "loss": 1.2341123819351196, "step": 35465 }, { "epoch": 2.795900670082775, "grad_norm": 0.559019923210144, "learning_rate": 0.0005, "loss": 1.422258973121643, "step": 35466 }, { "epoch": 2.795979503350414, "grad_norm": 0.4768661856651306, "learning_rate": 0.0005, "loss": 1.2367998361587524, "step": 35467 }, { "epoch": 2.796058336618053, "grad_norm": 0.5348789095878601, "learning_rate": 0.0005, "loss": 1.4648593664169312, "step": 35468 }, { "epoch": 2.7961371698856916, "grad_norm": 0.5137662291526794, "learning_rate": 0.0005, "loss": 1.4292094707489014, "step": 35469 }, { "epoch": 2.7962160031533307, "grad_norm": 0.47228580713272095, "learning_rate": 0.0005, "loss": 1.2783615589141846, "step": 35470 }, { "epoch": 2.7962948364209694, "grad_norm": 0.6267925500869751, "learning_rate": 0.0005, "loss": 1.3611254692077637, "step": 35471 }, { "epoch": 2.7963736696886086, "grad_norm": 0.4786852300167084, "learning_rate": 0.0005, "loss": 1.2672367095947266, "step": 35472 }, { "epoch": 2.7964525029562477, "grad_norm": 0.6326040625572205, "learning_rate": 0.0005, "loss": 1.1685028076171875, "step": 35473 }, { "epoch": 2.7965313362238864, "grad_norm": 0.42046669125556946, "learning_rate": 0.0005, "loss": 1.165174126625061, "step": 35474 }, { "epoch": 2.7966101694915255, "grad_norm": 0.6649244427680969, "learning_rate": 0.0005, "loss": 1.4987128973007202, "step": 35475 }, { "epoch": 2.796689002759164, "grad_norm": 0.5194183588027954, "learning_rate": 0.0005, "loss": 1.291632890701294, "step": 35476 }, { "epoch": 2.7967678360268033, "grad_norm": 0.5281552672386169, "learning_rate": 0.0005, "loss": 1.3480503559112549, "step": 35477 }, { "epoch": 2.7968466692944425, "grad_norm": 0.7173498272895813, "learning_rate": 0.0005, "loss": 1.5094671249389648, "step": 35478 }, { "epoch": 2.796925502562081, "grad_norm": 0.6504492163658142, "learning_rate": 0.0005, "loss": 1.354894757270813, "step": 35479 }, { "epoch": 2.7970043358297203, "grad_norm": 0.5382473468780518, "learning_rate": 0.0005, "loss": 1.4896950721740723, "step": 35480 }, { "epoch": 2.797083169097359, "grad_norm": 0.4820158779621124, "learning_rate": 0.0005, "loss": 1.354196548461914, "step": 35481 }, { "epoch": 2.797162002364998, "grad_norm": 0.4968825876712799, "learning_rate": 0.0005, "loss": 1.201711893081665, "step": 35482 }, { "epoch": 2.797240835632637, "grad_norm": 0.4425687789916992, "learning_rate": 0.0005, "loss": 1.2391563653945923, "step": 35483 }, { "epoch": 2.797319668900276, "grad_norm": 0.5267871022224426, "learning_rate": 0.0005, "loss": 1.4987764358520508, "step": 35484 }, { "epoch": 2.7973985021679146, "grad_norm": 0.39414992928504944, "learning_rate": 0.0005, "loss": 1.2757781744003296, "step": 35485 }, { "epoch": 2.7974773354355538, "grad_norm": 0.5830772519111633, "learning_rate": 0.0005, "loss": 1.5410557985305786, "step": 35486 }, { "epoch": 2.797556168703193, "grad_norm": 0.5246084928512573, "learning_rate": 0.0005, "loss": 1.169908046722412, "step": 35487 }, { "epoch": 2.7976350019708316, "grad_norm": 0.5111180543899536, "learning_rate": 0.0005, "loss": 1.2632700204849243, "step": 35488 }, { "epoch": 2.7977138352384707, "grad_norm": 0.38146141171455383, "learning_rate": 0.0005, "loss": 1.131029725074768, "step": 35489 }, { "epoch": 2.7977926685061094, "grad_norm": 0.551670491695404, "learning_rate": 0.0005, "loss": 1.566969871520996, "step": 35490 }, { "epoch": 2.7978715017737485, "grad_norm": 0.4843394458293915, "learning_rate": 0.0005, "loss": 1.1559202671051025, "step": 35491 }, { "epoch": 2.7979503350413877, "grad_norm": 0.6012076735496521, "learning_rate": 0.0005, "loss": 1.4343974590301514, "step": 35492 }, { "epoch": 2.7980291683090264, "grad_norm": 0.4889603853225708, "learning_rate": 0.0005, "loss": 1.4148821830749512, "step": 35493 }, { "epoch": 2.7981080015766655, "grad_norm": 0.4936448335647583, "learning_rate": 0.0005, "loss": 1.2167844772338867, "step": 35494 }, { "epoch": 2.798186834844304, "grad_norm": 0.47736355662345886, "learning_rate": 0.0005, "loss": 1.337611198425293, "step": 35495 }, { "epoch": 2.7982656681119433, "grad_norm": 0.47755980491638184, "learning_rate": 0.0005, "loss": 1.3015459775924683, "step": 35496 }, { "epoch": 2.7983445013795825, "grad_norm": 0.44870612025260925, "learning_rate": 0.0005, "loss": 1.1773947477340698, "step": 35497 }, { "epoch": 2.798423334647221, "grad_norm": 0.48465031385421753, "learning_rate": 0.0005, "loss": 1.264983057975769, "step": 35498 }, { "epoch": 2.79850216791486, "grad_norm": 0.6047321557998657, "learning_rate": 0.0005, "loss": 1.1396214962005615, "step": 35499 }, { "epoch": 2.798581001182499, "grad_norm": 0.48656219244003296, "learning_rate": 0.0005, "loss": 1.3876256942749023, "step": 35500 }, { "epoch": 2.798659834450138, "grad_norm": 0.4325038194656372, "learning_rate": 0.0005, "loss": 1.2127137184143066, "step": 35501 }, { "epoch": 2.798738667717777, "grad_norm": 0.372109979391098, "learning_rate": 0.0005, "loss": 1.1899895668029785, "step": 35502 }, { "epoch": 2.798817500985416, "grad_norm": 0.5136397480964661, "learning_rate": 0.0005, "loss": 1.502984642982483, "step": 35503 }, { "epoch": 2.7988963342530546, "grad_norm": 0.5072241425514221, "learning_rate": 0.0005, "loss": 1.2321691513061523, "step": 35504 }, { "epoch": 2.7989751675206938, "grad_norm": 0.8178738355636597, "learning_rate": 0.0005, "loss": 1.4812567234039307, "step": 35505 }, { "epoch": 2.799054000788333, "grad_norm": 0.6064123511314392, "learning_rate": 0.0005, "loss": 1.5396884679794312, "step": 35506 }, { "epoch": 2.7991328340559716, "grad_norm": 0.5112556219100952, "learning_rate": 0.0005, "loss": 1.2484583854675293, "step": 35507 }, { "epoch": 2.7992116673236107, "grad_norm": 0.48583269119262695, "learning_rate": 0.0005, "loss": 1.1890473365783691, "step": 35508 }, { "epoch": 2.7992905005912494, "grad_norm": 0.6465245485305786, "learning_rate": 0.0005, "loss": 1.5055311918258667, "step": 35509 }, { "epoch": 2.7993693338588885, "grad_norm": 0.47642838954925537, "learning_rate": 0.0005, "loss": 1.2836073637008667, "step": 35510 }, { "epoch": 2.7994481671265277, "grad_norm": 0.48595231771469116, "learning_rate": 0.0005, "loss": 1.4406769275665283, "step": 35511 }, { "epoch": 2.7995270003941664, "grad_norm": 0.7599474191665649, "learning_rate": 0.0005, "loss": 1.4477319717407227, "step": 35512 }, { "epoch": 2.799605833661805, "grad_norm": 0.5657469630241394, "learning_rate": 0.0005, "loss": 1.311506748199463, "step": 35513 }, { "epoch": 2.799684666929444, "grad_norm": 0.5002889633178711, "learning_rate": 0.0005, "loss": 1.1868667602539062, "step": 35514 }, { "epoch": 2.7997635001970833, "grad_norm": 0.5842069387435913, "learning_rate": 0.0005, "loss": 1.4658889770507812, "step": 35515 }, { "epoch": 2.799842333464722, "grad_norm": 0.46769392490386963, "learning_rate": 0.0005, "loss": 1.0997464656829834, "step": 35516 }, { "epoch": 2.799921166732361, "grad_norm": 0.570174515247345, "learning_rate": 0.0005, "loss": 1.4457777738571167, "step": 35517 }, { "epoch": 2.8, "grad_norm": 0.6428463459014893, "learning_rate": 0.0005, "loss": 1.2742013931274414, "step": 35518 }, { "epoch": 2.800078833267639, "grad_norm": 0.576225996017456, "learning_rate": 0.0005, "loss": 1.5283184051513672, "step": 35519 }, { "epoch": 2.800157666535278, "grad_norm": 0.7213240265846252, "learning_rate": 0.0005, "loss": 1.5727158784866333, "step": 35520 }, { "epoch": 2.800236499802917, "grad_norm": 0.5922701358795166, "learning_rate": 0.0005, "loss": 1.3935717344284058, "step": 35521 }, { "epoch": 2.800315333070556, "grad_norm": 0.5768064856529236, "learning_rate": 0.0005, "loss": 1.4375343322753906, "step": 35522 }, { "epoch": 2.8003941663381946, "grad_norm": 0.4313143491744995, "learning_rate": 0.0005, "loss": 1.3885523080825806, "step": 35523 }, { "epoch": 2.8004729996058337, "grad_norm": 0.5342037677764893, "learning_rate": 0.0005, "loss": 1.3717396259307861, "step": 35524 }, { "epoch": 2.800551832873473, "grad_norm": 0.45809024572372437, "learning_rate": 0.0005, "loss": 1.4381462335586548, "step": 35525 }, { "epoch": 2.8006306661411116, "grad_norm": 0.6001433730125427, "learning_rate": 0.0005, "loss": 1.3536312580108643, "step": 35526 }, { "epoch": 2.8007094994087502, "grad_norm": 0.47690272331237793, "learning_rate": 0.0005, "loss": 1.2032227516174316, "step": 35527 }, { "epoch": 2.8007883326763894, "grad_norm": 0.5710651874542236, "learning_rate": 0.0005, "loss": 1.4005273580551147, "step": 35528 }, { "epoch": 2.8008671659440285, "grad_norm": 0.4660879671573639, "learning_rate": 0.0005, "loss": 1.1551213264465332, "step": 35529 }, { "epoch": 2.800945999211667, "grad_norm": 0.543633759021759, "learning_rate": 0.0005, "loss": 1.4306640625, "step": 35530 }, { "epoch": 2.8010248324793063, "grad_norm": 0.4336557686328888, "learning_rate": 0.0005, "loss": 1.3084805011749268, "step": 35531 }, { "epoch": 2.801103665746945, "grad_norm": 0.570694625377655, "learning_rate": 0.0005, "loss": 1.3465068340301514, "step": 35532 }, { "epoch": 2.801182499014584, "grad_norm": 0.48609140515327454, "learning_rate": 0.0005, "loss": 1.2707829475402832, "step": 35533 }, { "epoch": 2.8012613322822233, "grad_norm": 0.471985787153244, "learning_rate": 0.0005, "loss": 1.325432300567627, "step": 35534 }, { "epoch": 2.801340165549862, "grad_norm": 0.4708002507686615, "learning_rate": 0.0005, "loss": 1.552912712097168, "step": 35535 }, { "epoch": 2.801418998817501, "grad_norm": 0.5189149975776672, "learning_rate": 0.0005, "loss": 1.454750418663025, "step": 35536 }, { "epoch": 2.80149783208514, "grad_norm": 0.49936071038246155, "learning_rate": 0.0005, "loss": 1.3798503875732422, "step": 35537 }, { "epoch": 2.801576665352779, "grad_norm": 0.7507963180541992, "learning_rate": 0.0005, "loss": 1.369882345199585, "step": 35538 }, { "epoch": 2.801655498620418, "grad_norm": 0.4595860540866852, "learning_rate": 0.0005, "loss": 1.4007800817489624, "step": 35539 }, { "epoch": 2.8017343318880568, "grad_norm": 0.5924350619316101, "learning_rate": 0.0005, "loss": 1.3372323513031006, "step": 35540 }, { "epoch": 2.8018131651556955, "grad_norm": 0.4738388657569885, "learning_rate": 0.0005, "loss": 1.1360008716583252, "step": 35541 }, { "epoch": 2.8018919984233346, "grad_norm": 0.4853561818599701, "learning_rate": 0.0005, "loss": 1.3705781698226929, "step": 35542 }, { "epoch": 2.8019708316909737, "grad_norm": 0.5628430843353271, "learning_rate": 0.0005, "loss": 1.3299460411071777, "step": 35543 }, { "epoch": 2.8020496649586124, "grad_norm": 0.5631573796272278, "learning_rate": 0.0005, "loss": 1.4336448907852173, "step": 35544 }, { "epoch": 2.8021284982262515, "grad_norm": 0.5888688564300537, "learning_rate": 0.0005, "loss": 1.4794609546661377, "step": 35545 }, { "epoch": 2.8022073314938902, "grad_norm": 0.5081086754798889, "learning_rate": 0.0005, "loss": 1.3180410861968994, "step": 35546 }, { "epoch": 2.8022861647615294, "grad_norm": 0.5174721479415894, "learning_rate": 0.0005, "loss": 1.3729890584945679, "step": 35547 }, { "epoch": 2.8023649980291685, "grad_norm": 0.7009722590446472, "learning_rate": 0.0005, "loss": 1.3821017742156982, "step": 35548 }, { "epoch": 2.802443831296807, "grad_norm": 0.5470597147941589, "learning_rate": 0.0005, "loss": 1.329360008239746, "step": 35549 }, { "epoch": 2.8025226645644463, "grad_norm": 0.5032703280448914, "learning_rate": 0.0005, "loss": 1.2602834701538086, "step": 35550 }, { "epoch": 2.802601497832085, "grad_norm": 0.4179422855377197, "learning_rate": 0.0005, "loss": 1.192057728767395, "step": 35551 }, { "epoch": 2.802680331099724, "grad_norm": 0.4336373805999756, "learning_rate": 0.0005, "loss": 1.2037503719329834, "step": 35552 }, { "epoch": 2.8027591643673633, "grad_norm": 0.45719146728515625, "learning_rate": 0.0005, "loss": 1.3340150117874146, "step": 35553 }, { "epoch": 2.802837997635002, "grad_norm": 0.9023056626319885, "learning_rate": 0.0005, "loss": 1.318225622177124, "step": 35554 }, { "epoch": 2.8029168309026407, "grad_norm": 0.36562052369117737, "learning_rate": 0.0005, "loss": 1.0121142864227295, "step": 35555 }, { "epoch": 2.80299566417028, "grad_norm": 0.47637128829956055, "learning_rate": 0.0005, "loss": 1.401021122932434, "step": 35556 }, { "epoch": 2.803074497437919, "grad_norm": 1.277380347251892, "learning_rate": 0.0005, "loss": 1.5750733613967896, "step": 35557 }, { "epoch": 2.8031533307055576, "grad_norm": 0.49693912267684937, "learning_rate": 0.0005, "loss": 1.3542821407318115, "step": 35558 }, { "epoch": 2.8032321639731967, "grad_norm": 0.47353824973106384, "learning_rate": 0.0005, "loss": 1.2704989910125732, "step": 35559 }, { "epoch": 2.8033109972408354, "grad_norm": 0.5329681038856506, "learning_rate": 0.0005, "loss": 1.5820021629333496, "step": 35560 }, { "epoch": 2.8033898305084746, "grad_norm": 0.5184637904167175, "learning_rate": 0.0005, "loss": 1.2694586515426636, "step": 35561 }, { "epoch": 2.8034686637761137, "grad_norm": 0.41246840357780457, "learning_rate": 0.0005, "loss": 1.1325740814208984, "step": 35562 }, { "epoch": 2.8035474970437524, "grad_norm": 0.4722237288951874, "learning_rate": 0.0005, "loss": 1.4762389659881592, "step": 35563 }, { "epoch": 2.8036263303113915, "grad_norm": 0.6486467123031616, "learning_rate": 0.0005, "loss": 1.4024620056152344, "step": 35564 }, { "epoch": 2.80370516357903, "grad_norm": 0.6411939859390259, "learning_rate": 0.0005, "loss": 1.500077724456787, "step": 35565 }, { "epoch": 2.8037839968466693, "grad_norm": 0.5187541246414185, "learning_rate": 0.0005, "loss": 1.2350902557373047, "step": 35566 }, { "epoch": 2.8038628301143085, "grad_norm": 0.4998965263366699, "learning_rate": 0.0005, "loss": 1.323710322380066, "step": 35567 }, { "epoch": 2.803941663381947, "grad_norm": 0.46055081486701965, "learning_rate": 0.0005, "loss": 1.2186894416809082, "step": 35568 }, { "epoch": 2.804020496649586, "grad_norm": 0.48082104325294495, "learning_rate": 0.0005, "loss": 1.3400375843048096, "step": 35569 }, { "epoch": 2.804099329917225, "grad_norm": 0.41938871145248413, "learning_rate": 0.0005, "loss": 1.1959030628204346, "step": 35570 }, { "epoch": 2.804178163184864, "grad_norm": 0.6302804350852966, "learning_rate": 0.0005, "loss": 1.398306131362915, "step": 35571 }, { "epoch": 2.804256996452503, "grad_norm": 0.4715457260608673, "learning_rate": 0.0005, "loss": 1.231543779373169, "step": 35572 }, { "epoch": 2.804335829720142, "grad_norm": 1.0406444072723389, "learning_rate": 0.0005, "loss": 1.6136995553970337, "step": 35573 }, { "epoch": 2.8044146629877806, "grad_norm": 0.40511080622673035, "learning_rate": 0.0005, "loss": 1.247391700744629, "step": 35574 }, { "epoch": 2.8044934962554198, "grad_norm": 0.4944361448287964, "learning_rate": 0.0005, "loss": 1.1717314720153809, "step": 35575 }, { "epoch": 2.804572329523059, "grad_norm": 0.49432268738746643, "learning_rate": 0.0005, "loss": 1.2599496841430664, "step": 35576 }, { "epoch": 2.8046511627906976, "grad_norm": 0.434413343667984, "learning_rate": 0.0005, "loss": 1.2420604228973389, "step": 35577 }, { "epoch": 2.8047299960583367, "grad_norm": 0.6052020788192749, "learning_rate": 0.0005, "loss": 1.3664400577545166, "step": 35578 }, { "epoch": 2.8048088293259754, "grad_norm": 0.520359992980957, "learning_rate": 0.0005, "loss": 1.4050850868225098, "step": 35579 }, { "epoch": 2.8048876625936146, "grad_norm": 0.45094457268714905, "learning_rate": 0.0005, "loss": 1.3572988510131836, "step": 35580 }, { "epoch": 2.8049664958612537, "grad_norm": 0.47412246465682983, "learning_rate": 0.0005, "loss": 1.1958887577056885, "step": 35581 }, { "epoch": 2.8050453291288924, "grad_norm": 0.5595793724060059, "learning_rate": 0.0005, "loss": 1.3523553609848022, "step": 35582 }, { "epoch": 2.805124162396531, "grad_norm": 0.44064760208129883, "learning_rate": 0.0005, "loss": 1.1715624332427979, "step": 35583 }, { "epoch": 2.80520299566417, "grad_norm": 0.7808666229248047, "learning_rate": 0.0005, "loss": 1.4851083755493164, "step": 35584 }, { "epoch": 2.8052818289318093, "grad_norm": 0.5398996472358704, "learning_rate": 0.0005, "loss": 1.2472357749938965, "step": 35585 }, { "epoch": 2.805360662199448, "grad_norm": 0.49520599842071533, "learning_rate": 0.0005, "loss": 1.29744553565979, "step": 35586 }, { "epoch": 2.805439495467087, "grad_norm": 0.5636061429977417, "learning_rate": 0.0005, "loss": 1.5084106922149658, "step": 35587 }, { "epoch": 2.805518328734726, "grad_norm": 0.5855740308761597, "learning_rate": 0.0005, "loss": 1.3417986631393433, "step": 35588 }, { "epoch": 2.805597162002365, "grad_norm": 0.45160844922065735, "learning_rate": 0.0005, "loss": 1.196746826171875, "step": 35589 }, { "epoch": 2.805675995270004, "grad_norm": 0.5379756689071655, "learning_rate": 0.0005, "loss": 1.4103751182556152, "step": 35590 }, { "epoch": 2.805754828537643, "grad_norm": 0.4832322299480438, "learning_rate": 0.0005, "loss": 1.2668792009353638, "step": 35591 }, { "epoch": 2.805833661805282, "grad_norm": 0.5124419927597046, "learning_rate": 0.0005, "loss": 1.380704402923584, "step": 35592 }, { "epoch": 2.8059124950729206, "grad_norm": 0.5717802047729492, "learning_rate": 0.0005, "loss": 1.420027494430542, "step": 35593 }, { "epoch": 2.8059913283405598, "grad_norm": 0.6641525030136108, "learning_rate": 0.0005, "loss": 1.4565958976745605, "step": 35594 }, { "epoch": 2.806070161608199, "grad_norm": 0.4793568551540375, "learning_rate": 0.0005, "loss": 1.401719331741333, "step": 35595 }, { "epoch": 2.8061489948758376, "grad_norm": 0.5438988208770752, "learning_rate": 0.0005, "loss": 1.3570632934570312, "step": 35596 }, { "epoch": 2.8062278281434767, "grad_norm": 0.6359502077102661, "learning_rate": 0.0005, "loss": 1.4648786783218384, "step": 35597 }, { "epoch": 2.8063066614111154, "grad_norm": 0.4319024085998535, "learning_rate": 0.0005, "loss": 1.3238632678985596, "step": 35598 }, { "epoch": 2.8063854946787545, "grad_norm": 0.367497056722641, "learning_rate": 0.0005, "loss": 1.1866977214813232, "step": 35599 }, { "epoch": 2.8064643279463932, "grad_norm": 0.5362018942832947, "learning_rate": 0.0005, "loss": 1.2335875034332275, "step": 35600 }, { "epoch": 2.8065431612140324, "grad_norm": 0.483996719121933, "learning_rate": 0.0005, "loss": 1.1830041408538818, "step": 35601 }, { "epoch": 2.806621994481671, "grad_norm": 0.5331189632415771, "learning_rate": 0.0005, "loss": 1.323622226715088, "step": 35602 }, { "epoch": 2.80670082774931, "grad_norm": 0.4478907585144043, "learning_rate": 0.0005, "loss": 1.1707942485809326, "step": 35603 }, { "epoch": 2.8067796610169493, "grad_norm": 0.4648447632789612, "learning_rate": 0.0005, "loss": 1.3108558654785156, "step": 35604 }, { "epoch": 2.806858494284588, "grad_norm": 0.5117353200912476, "learning_rate": 0.0005, "loss": 1.3710113763809204, "step": 35605 }, { "epoch": 2.806937327552227, "grad_norm": 0.5630084872245789, "learning_rate": 0.0005, "loss": 1.510342001914978, "step": 35606 }, { "epoch": 2.807016160819866, "grad_norm": 0.5287096500396729, "learning_rate": 0.0005, "loss": 1.3975248336791992, "step": 35607 }, { "epoch": 2.807094994087505, "grad_norm": 0.46344029903411865, "learning_rate": 0.0005, "loss": 1.3024117946624756, "step": 35608 }, { "epoch": 2.807173827355144, "grad_norm": 0.48891282081604004, "learning_rate": 0.0005, "loss": 1.3226147890090942, "step": 35609 }, { "epoch": 2.807252660622783, "grad_norm": 0.7630643248558044, "learning_rate": 0.0005, "loss": 1.4055376052856445, "step": 35610 }, { "epoch": 2.807331493890422, "grad_norm": 0.47651001811027527, "learning_rate": 0.0005, "loss": 1.35726797580719, "step": 35611 }, { "epoch": 2.8074103271580606, "grad_norm": 0.45390695333480835, "learning_rate": 0.0005, "loss": 1.2338447570800781, "step": 35612 }, { "epoch": 2.8074891604256997, "grad_norm": 0.555648148059845, "learning_rate": 0.0005, "loss": 1.4205459356307983, "step": 35613 }, { "epoch": 2.8075679936933384, "grad_norm": 0.4733373820781708, "learning_rate": 0.0005, "loss": 1.2184548377990723, "step": 35614 }, { "epoch": 2.8076468269609776, "grad_norm": 0.6637349128723145, "learning_rate": 0.0005, "loss": 1.5688788890838623, "step": 35615 }, { "epoch": 2.8077256602286162, "grad_norm": 0.6692094206809998, "learning_rate": 0.0005, "loss": 1.4090335369110107, "step": 35616 }, { "epoch": 2.8078044934962554, "grad_norm": 0.4942079782485962, "learning_rate": 0.0005, "loss": 1.3283196687698364, "step": 35617 }, { "epoch": 2.8078833267638945, "grad_norm": 0.4493621587753296, "learning_rate": 0.0005, "loss": 1.1650140285491943, "step": 35618 }, { "epoch": 2.807962160031533, "grad_norm": 0.568126916885376, "learning_rate": 0.0005, "loss": 1.2388092279434204, "step": 35619 }, { "epoch": 2.8080409932991723, "grad_norm": 0.7223996520042419, "learning_rate": 0.0005, "loss": 1.2279853820800781, "step": 35620 }, { "epoch": 2.808119826566811, "grad_norm": 0.5393514037132263, "learning_rate": 0.0005, "loss": 1.0971421003341675, "step": 35621 }, { "epoch": 2.80819865983445, "grad_norm": 0.527447521686554, "learning_rate": 0.0005, "loss": 1.3032866716384888, "step": 35622 }, { "epoch": 2.8082774931020893, "grad_norm": 0.5070998668670654, "learning_rate": 0.0005, "loss": 1.2724533081054688, "step": 35623 }, { "epoch": 2.808356326369728, "grad_norm": 0.505079448223114, "learning_rate": 0.0005, "loss": 1.251340389251709, "step": 35624 }, { "epoch": 2.808435159637367, "grad_norm": 0.45977967977523804, "learning_rate": 0.0005, "loss": 1.2828404903411865, "step": 35625 }, { "epoch": 2.808513992905006, "grad_norm": 0.507413387298584, "learning_rate": 0.0005, "loss": 1.4743343591690063, "step": 35626 }, { "epoch": 2.808592826172645, "grad_norm": 0.5202187895774841, "learning_rate": 0.0005, "loss": 1.3484631776809692, "step": 35627 }, { "epoch": 2.808671659440284, "grad_norm": 0.5705924034118652, "learning_rate": 0.0005, "loss": 1.2854433059692383, "step": 35628 }, { "epoch": 2.8087504927079228, "grad_norm": 0.6301109790802002, "learning_rate": 0.0005, "loss": 1.4826576709747314, "step": 35629 }, { "epoch": 2.8088293259755615, "grad_norm": 0.4642392694950104, "learning_rate": 0.0005, "loss": 1.2038073539733887, "step": 35630 }, { "epoch": 2.8089081592432006, "grad_norm": 0.7087087035179138, "learning_rate": 0.0005, "loss": 1.4486886262893677, "step": 35631 }, { "epoch": 2.8089869925108397, "grad_norm": 0.4463273584842682, "learning_rate": 0.0005, "loss": 1.1731815338134766, "step": 35632 }, { "epoch": 2.8090658257784784, "grad_norm": 0.5692229270935059, "learning_rate": 0.0005, "loss": 1.232500433921814, "step": 35633 }, { "epoch": 2.8091446590461175, "grad_norm": 0.5488917231559753, "learning_rate": 0.0005, "loss": 1.437847375869751, "step": 35634 }, { "epoch": 2.8092234923137562, "grad_norm": 0.471609503030777, "learning_rate": 0.0005, "loss": 1.2415457963943481, "step": 35635 }, { "epoch": 2.8093023255813954, "grad_norm": 0.5367149114608765, "learning_rate": 0.0005, "loss": 1.4336071014404297, "step": 35636 }, { "epoch": 2.8093811588490345, "grad_norm": 0.48828989267349243, "learning_rate": 0.0005, "loss": 1.3118023872375488, "step": 35637 }, { "epoch": 2.809459992116673, "grad_norm": 0.4919053614139557, "learning_rate": 0.0005, "loss": 1.211876392364502, "step": 35638 }, { "epoch": 2.8095388253843123, "grad_norm": 0.41506046056747437, "learning_rate": 0.0005, "loss": 1.2495203018188477, "step": 35639 }, { "epoch": 2.809617658651951, "grad_norm": 0.5763742923736572, "learning_rate": 0.0005, "loss": 1.6439263820648193, "step": 35640 }, { "epoch": 2.80969649191959, "grad_norm": 0.4765118956565857, "learning_rate": 0.0005, "loss": 1.36765718460083, "step": 35641 }, { "epoch": 2.8097753251872293, "grad_norm": 0.5590476989746094, "learning_rate": 0.0005, "loss": 1.3634406328201294, "step": 35642 }, { "epoch": 2.809854158454868, "grad_norm": 0.4232810437679291, "learning_rate": 0.0005, "loss": 1.2584565877914429, "step": 35643 }, { "epoch": 2.8099329917225067, "grad_norm": 0.4442887306213379, "learning_rate": 0.0005, "loss": 1.1880979537963867, "step": 35644 }, { "epoch": 2.810011824990146, "grad_norm": 0.48957180976867676, "learning_rate": 0.0005, "loss": 1.2466354370117188, "step": 35645 }, { "epoch": 2.810090658257785, "grad_norm": 0.5009669065475464, "learning_rate": 0.0005, "loss": 1.4118986129760742, "step": 35646 }, { "epoch": 2.8101694915254236, "grad_norm": 0.4374235272407532, "learning_rate": 0.0005, "loss": 1.146602749824524, "step": 35647 }, { "epoch": 2.8102483247930627, "grad_norm": 0.6445745825767517, "learning_rate": 0.0005, "loss": 1.625239610671997, "step": 35648 }, { "epoch": 2.8103271580607014, "grad_norm": 0.5204863548278809, "learning_rate": 0.0005, "loss": 1.4634249210357666, "step": 35649 }, { "epoch": 2.8104059913283406, "grad_norm": 0.5569863319396973, "learning_rate": 0.0005, "loss": 1.2179906368255615, "step": 35650 }, { "epoch": 2.8104848245959797, "grad_norm": 0.4569098949432373, "learning_rate": 0.0005, "loss": 1.177010178565979, "step": 35651 }, { "epoch": 2.8105636578636184, "grad_norm": 0.6002146601676941, "learning_rate": 0.0005, "loss": 1.5207247734069824, "step": 35652 }, { "epoch": 2.8106424911312575, "grad_norm": 0.4964180886745453, "learning_rate": 0.0005, "loss": 1.4055112600326538, "step": 35653 }, { "epoch": 2.810721324398896, "grad_norm": 0.534888744354248, "learning_rate": 0.0005, "loss": 1.257049798965454, "step": 35654 }, { "epoch": 2.8108001576665353, "grad_norm": 0.4562375247478485, "learning_rate": 0.0005, "loss": 1.1952786445617676, "step": 35655 }, { "epoch": 2.8108789909341745, "grad_norm": 0.5557423830032349, "learning_rate": 0.0005, "loss": 1.3219475746154785, "step": 35656 }, { "epoch": 2.810957824201813, "grad_norm": 0.47282344102859497, "learning_rate": 0.0005, "loss": 1.2570091485977173, "step": 35657 }, { "epoch": 2.811036657469452, "grad_norm": 0.45092159509658813, "learning_rate": 0.0005, "loss": 1.1289573907852173, "step": 35658 }, { "epoch": 2.811115490737091, "grad_norm": 0.7692050337791443, "learning_rate": 0.0005, "loss": 1.5143712759017944, "step": 35659 }, { "epoch": 2.81119432400473, "grad_norm": 0.5393772721290588, "learning_rate": 0.0005, "loss": 1.249556541442871, "step": 35660 }, { "epoch": 2.811273157272369, "grad_norm": 0.4378763735294342, "learning_rate": 0.0005, "loss": 1.3628532886505127, "step": 35661 }, { "epoch": 2.811351990540008, "grad_norm": 0.5939468741416931, "learning_rate": 0.0005, "loss": 1.366322636604309, "step": 35662 }, { "epoch": 2.8114308238076466, "grad_norm": 0.4499160945415497, "learning_rate": 0.0005, "loss": 1.2501142024993896, "step": 35663 }, { "epoch": 2.8115096570752858, "grad_norm": 0.515944242477417, "learning_rate": 0.0005, "loss": 1.3282382488250732, "step": 35664 }, { "epoch": 2.811588490342925, "grad_norm": 0.4309483766555786, "learning_rate": 0.0005, "loss": 1.238533854484558, "step": 35665 }, { "epoch": 2.8116673236105636, "grad_norm": 0.47321248054504395, "learning_rate": 0.0005, "loss": 1.4402378797531128, "step": 35666 }, { "epoch": 2.8117461568782027, "grad_norm": 0.5959744453430176, "learning_rate": 0.0005, "loss": 1.4929263591766357, "step": 35667 }, { "epoch": 2.8118249901458414, "grad_norm": 0.8443343043327332, "learning_rate": 0.0005, "loss": 1.509918212890625, "step": 35668 }, { "epoch": 2.8119038234134806, "grad_norm": 0.44269225001335144, "learning_rate": 0.0005, "loss": 1.3206005096435547, "step": 35669 }, { "epoch": 2.8119826566811197, "grad_norm": 0.5284915566444397, "learning_rate": 0.0005, "loss": 1.3718743324279785, "step": 35670 }, { "epoch": 2.8120614899487584, "grad_norm": 0.6195961833000183, "learning_rate": 0.0005, "loss": 1.3485136032104492, "step": 35671 }, { "epoch": 2.812140323216397, "grad_norm": 0.5371698141098022, "learning_rate": 0.0005, "loss": 1.278795838356018, "step": 35672 }, { "epoch": 2.812219156484036, "grad_norm": 0.5266580581665039, "learning_rate": 0.0005, "loss": 1.5474402904510498, "step": 35673 }, { "epoch": 2.8122979897516753, "grad_norm": 0.4955452084541321, "learning_rate": 0.0005, "loss": 1.4548934698104858, "step": 35674 }, { "epoch": 2.812376823019314, "grad_norm": 0.5557435154914856, "learning_rate": 0.0005, "loss": 1.2100722789764404, "step": 35675 }, { "epoch": 2.812455656286953, "grad_norm": 0.5978226661682129, "learning_rate": 0.0005, "loss": 1.5171592235565186, "step": 35676 }, { "epoch": 2.812534489554592, "grad_norm": 0.5033902525901794, "learning_rate": 0.0005, "loss": 1.1635631322860718, "step": 35677 }, { "epoch": 2.812613322822231, "grad_norm": 0.4876658618450165, "learning_rate": 0.0005, "loss": 1.4548059701919556, "step": 35678 }, { "epoch": 2.81269215608987, "grad_norm": 0.6065530776977539, "learning_rate": 0.0005, "loss": 1.627458095550537, "step": 35679 }, { "epoch": 2.812770989357509, "grad_norm": 0.44695356488227844, "learning_rate": 0.0005, "loss": 1.1076971292495728, "step": 35680 }, { "epoch": 2.812849822625148, "grad_norm": 0.4219265878200531, "learning_rate": 0.0005, "loss": 1.3552730083465576, "step": 35681 }, { "epoch": 2.8129286558927866, "grad_norm": 0.51895672082901, "learning_rate": 0.0005, "loss": 1.4051318168640137, "step": 35682 }, { "epoch": 2.8130074891604258, "grad_norm": 0.6372947692871094, "learning_rate": 0.0005, "loss": 1.5019025802612305, "step": 35683 }, { "epoch": 2.813086322428065, "grad_norm": 0.4733608067035675, "learning_rate": 0.0005, "loss": 1.2000813484191895, "step": 35684 }, { "epoch": 2.8131651556957036, "grad_norm": 0.4300583302974701, "learning_rate": 0.0005, "loss": 1.1608667373657227, "step": 35685 }, { "epoch": 2.8132439889633423, "grad_norm": 0.5480557680130005, "learning_rate": 0.0005, "loss": 1.301431655883789, "step": 35686 }, { "epoch": 2.8133228222309814, "grad_norm": 0.4794800281524658, "learning_rate": 0.0005, "loss": 1.3432683944702148, "step": 35687 }, { "epoch": 2.8134016554986205, "grad_norm": 0.4368342459201813, "learning_rate": 0.0005, "loss": 1.1696983575820923, "step": 35688 }, { "epoch": 2.8134804887662592, "grad_norm": 0.5611957311630249, "learning_rate": 0.0005, "loss": 1.4087486267089844, "step": 35689 }, { "epoch": 2.8135593220338984, "grad_norm": 0.5612648129463196, "learning_rate": 0.0005, "loss": 1.2487716674804688, "step": 35690 }, { "epoch": 2.813638155301537, "grad_norm": 0.5976298451423645, "learning_rate": 0.0005, "loss": 1.3976233005523682, "step": 35691 }, { "epoch": 2.813716988569176, "grad_norm": 0.508766233921051, "learning_rate": 0.0005, "loss": 1.3083046674728394, "step": 35692 }, { "epoch": 2.8137958218368153, "grad_norm": 0.7096136808395386, "learning_rate": 0.0005, "loss": 1.3242881298065186, "step": 35693 }, { "epoch": 2.813874655104454, "grad_norm": 0.4450855255126953, "learning_rate": 0.0005, "loss": 1.2728875875473022, "step": 35694 }, { "epoch": 2.813953488372093, "grad_norm": 0.4477703273296356, "learning_rate": 0.0005, "loss": 1.3234343528747559, "step": 35695 }, { "epoch": 2.814032321639732, "grad_norm": 0.4561460018157959, "learning_rate": 0.0005, "loss": 1.2597200870513916, "step": 35696 }, { "epoch": 2.814111154907371, "grad_norm": 0.44892585277557373, "learning_rate": 0.0005, "loss": 1.251467227935791, "step": 35697 }, { "epoch": 2.81418998817501, "grad_norm": 0.396186500787735, "learning_rate": 0.0005, "loss": 1.1389411687850952, "step": 35698 }, { "epoch": 2.814268821442649, "grad_norm": 0.6057844758033752, "learning_rate": 0.0005, "loss": 1.500044584274292, "step": 35699 }, { "epoch": 2.8143476547102875, "grad_norm": 0.49263498187065125, "learning_rate": 0.0005, "loss": 1.2851684093475342, "step": 35700 }, { "epoch": 2.8144264879779266, "grad_norm": 0.43723657727241516, "learning_rate": 0.0005, "loss": 1.2199883460998535, "step": 35701 }, { "epoch": 2.8145053212455657, "grad_norm": 0.4903055429458618, "learning_rate": 0.0005, "loss": 1.330260992050171, "step": 35702 }, { "epoch": 2.8145841545132044, "grad_norm": 0.4418811798095703, "learning_rate": 0.0005, "loss": 1.269400715827942, "step": 35703 }, { "epoch": 2.8146629877808436, "grad_norm": 0.5086963176727295, "learning_rate": 0.0005, "loss": 1.4826403856277466, "step": 35704 }, { "epoch": 2.8147418210484823, "grad_norm": 0.4172731041908264, "learning_rate": 0.0005, "loss": 1.0951145887374878, "step": 35705 }, { "epoch": 2.8148206543161214, "grad_norm": 0.4242112934589386, "learning_rate": 0.0005, "loss": 1.1391069889068604, "step": 35706 }, { "epoch": 2.8148994875837605, "grad_norm": 0.7738272547721863, "learning_rate": 0.0005, "loss": 1.326783537864685, "step": 35707 }, { "epoch": 2.814978320851399, "grad_norm": 0.4221823215484619, "learning_rate": 0.0005, "loss": 1.1273341178894043, "step": 35708 }, { "epoch": 2.8150571541190383, "grad_norm": 0.4252983033657074, "learning_rate": 0.0005, "loss": 1.264691710472107, "step": 35709 }, { "epoch": 2.815135987386677, "grad_norm": 0.5434086918830872, "learning_rate": 0.0005, "loss": 1.3324980735778809, "step": 35710 }, { "epoch": 2.815214820654316, "grad_norm": 1.5545934438705444, "learning_rate": 0.0005, "loss": 1.2919888496398926, "step": 35711 }, { "epoch": 2.8152936539219553, "grad_norm": 0.5282546877861023, "learning_rate": 0.0005, "loss": 1.330623745918274, "step": 35712 }, { "epoch": 2.815372487189594, "grad_norm": 0.5366412997245789, "learning_rate": 0.0005, "loss": 1.3599865436553955, "step": 35713 }, { "epoch": 2.8154513204572327, "grad_norm": 0.49739977717399597, "learning_rate": 0.0005, "loss": 1.471904993057251, "step": 35714 }, { "epoch": 2.815530153724872, "grad_norm": 0.43432965874671936, "learning_rate": 0.0005, "loss": 1.1714763641357422, "step": 35715 }, { "epoch": 2.815608986992511, "grad_norm": 0.5205497145652771, "learning_rate": 0.0005, "loss": 1.3405613899230957, "step": 35716 }, { "epoch": 2.8156878202601496, "grad_norm": 0.5225538611412048, "learning_rate": 0.0005, "loss": 1.2355995178222656, "step": 35717 }, { "epoch": 2.8157666535277888, "grad_norm": 0.6274452209472656, "learning_rate": 0.0005, "loss": 1.5175540447235107, "step": 35718 }, { "epoch": 2.8158454867954275, "grad_norm": 1.1226476430892944, "learning_rate": 0.0005, "loss": 1.3689110279083252, "step": 35719 }, { "epoch": 2.8159243200630666, "grad_norm": 0.5264347195625305, "learning_rate": 0.0005, "loss": 1.2474706172943115, "step": 35720 }, { "epoch": 2.8160031533307057, "grad_norm": 0.6411653757095337, "learning_rate": 0.0005, "loss": 1.517211675643921, "step": 35721 }, { "epoch": 2.8160819865983444, "grad_norm": 0.5517951846122742, "learning_rate": 0.0005, "loss": 1.393297553062439, "step": 35722 }, { "epoch": 2.8161608198659835, "grad_norm": 0.49258679151535034, "learning_rate": 0.0005, "loss": 1.360913634300232, "step": 35723 }, { "epoch": 2.8162396531336222, "grad_norm": 0.44979578256607056, "learning_rate": 0.0005, "loss": 1.2860565185546875, "step": 35724 }, { "epoch": 2.8163184864012614, "grad_norm": 0.550732433795929, "learning_rate": 0.0005, "loss": 1.1950342655181885, "step": 35725 }, { "epoch": 2.8163973196689005, "grad_norm": 0.479994535446167, "learning_rate": 0.0005, "loss": 1.3139243125915527, "step": 35726 }, { "epoch": 2.816476152936539, "grad_norm": 0.5296258330345154, "learning_rate": 0.0005, "loss": 1.4566789865493774, "step": 35727 }, { "epoch": 2.8165549862041783, "grad_norm": 0.575725257396698, "learning_rate": 0.0005, "loss": 1.4417576789855957, "step": 35728 }, { "epoch": 2.816633819471817, "grad_norm": 0.495172917842865, "learning_rate": 0.0005, "loss": 1.5552420616149902, "step": 35729 }, { "epoch": 2.816712652739456, "grad_norm": 0.43924030661582947, "learning_rate": 0.0005, "loss": 1.225578784942627, "step": 35730 }, { "epoch": 2.816791486007095, "grad_norm": 0.4567946791648865, "learning_rate": 0.0005, "loss": 1.3663268089294434, "step": 35731 }, { "epoch": 2.816870319274734, "grad_norm": 0.49591153860092163, "learning_rate": 0.0005, "loss": 1.3355923891067505, "step": 35732 }, { "epoch": 2.8169491525423727, "grad_norm": 0.4762071669101715, "learning_rate": 0.0005, "loss": 1.2583781480789185, "step": 35733 }, { "epoch": 2.817027985810012, "grad_norm": 0.6423025131225586, "learning_rate": 0.0005, "loss": 1.3270132541656494, "step": 35734 }, { "epoch": 2.817106819077651, "grad_norm": 0.6474536061286926, "learning_rate": 0.0005, "loss": 1.4747874736785889, "step": 35735 }, { "epoch": 2.8171856523452896, "grad_norm": 0.4624923765659332, "learning_rate": 0.0005, "loss": 1.1700596809387207, "step": 35736 }, { "epoch": 2.8172644856129287, "grad_norm": 0.42055779695510864, "learning_rate": 0.0005, "loss": 1.432592511177063, "step": 35737 }, { "epoch": 2.8173433188805674, "grad_norm": 0.48154503107070923, "learning_rate": 0.0005, "loss": 1.2622380256652832, "step": 35738 }, { "epoch": 2.8174221521482066, "grad_norm": 0.4777733087539673, "learning_rate": 0.0005, "loss": 1.286649465560913, "step": 35739 }, { "epoch": 2.8175009854158457, "grad_norm": 0.45618578791618347, "learning_rate": 0.0005, "loss": 1.250473976135254, "step": 35740 }, { "epoch": 2.8175798186834844, "grad_norm": 0.45627400279045105, "learning_rate": 0.0005, "loss": 1.1871542930603027, "step": 35741 }, { "epoch": 2.8176586519511235, "grad_norm": 0.48586609959602356, "learning_rate": 0.0005, "loss": 1.4787893295288086, "step": 35742 }, { "epoch": 2.817737485218762, "grad_norm": 0.5122833847999573, "learning_rate": 0.0005, "loss": 1.2743949890136719, "step": 35743 }, { "epoch": 2.8178163184864014, "grad_norm": 0.49526718258857727, "learning_rate": 0.0005, "loss": 1.3055651187896729, "step": 35744 }, { "epoch": 2.81789515175404, "grad_norm": 0.4037719964981079, "learning_rate": 0.0005, "loss": 1.1619422435760498, "step": 35745 }, { "epoch": 2.817973985021679, "grad_norm": 0.49601712822914124, "learning_rate": 0.0005, "loss": 1.2853333950042725, "step": 35746 }, { "epoch": 2.818052818289318, "grad_norm": 0.502354085445404, "learning_rate": 0.0005, "loss": 1.389017939567566, "step": 35747 }, { "epoch": 2.818131651556957, "grad_norm": 0.6504165530204773, "learning_rate": 0.0005, "loss": 1.4677281379699707, "step": 35748 }, { "epoch": 2.818210484824596, "grad_norm": 0.5379419326782227, "learning_rate": 0.0005, "loss": 1.3070204257965088, "step": 35749 }, { "epoch": 2.818289318092235, "grad_norm": 0.60782790184021, "learning_rate": 0.0005, "loss": 1.341320514678955, "step": 35750 }, { "epoch": 2.818368151359874, "grad_norm": 0.5048473477363586, "learning_rate": 0.0005, "loss": 1.3557357788085938, "step": 35751 }, { "epoch": 2.8184469846275126, "grad_norm": 0.5015600323677063, "learning_rate": 0.0005, "loss": 1.3677101135253906, "step": 35752 }, { "epoch": 2.8185258178951518, "grad_norm": 0.49294835329055786, "learning_rate": 0.0005, "loss": 1.2222459316253662, "step": 35753 }, { "epoch": 2.818604651162791, "grad_norm": 0.5273074507713318, "learning_rate": 0.0005, "loss": 1.324500322341919, "step": 35754 }, { "epoch": 2.8186834844304296, "grad_norm": 0.5576359033584595, "learning_rate": 0.0005, "loss": 1.4232494831085205, "step": 35755 }, { "epoch": 2.8187623176980687, "grad_norm": 0.44737327098846436, "learning_rate": 0.0005, "loss": 1.4305598735809326, "step": 35756 }, { "epoch": 2.8188411509657074, "grad_norm": 0.5773645639419556, "learning_rate": 0.0005, "loss": 1.3242409229278564, "step": 35757 }, { "epoch": 2.8189199842333466, "grad_norm": 0.43272310495376587, "learning_rate": 0.0005, "loss": 1.398499608039856, "step": 35758 }, { "epoch": 2.8189988175009857, "grad_norm": 0.5362844467163086, "learning_rate": 0.0005, "loss": 1.4804519414901733, "step": 35759 }, { "epoch": 2.8190776507686244, "grad_norm": 0.44969168305397034, "learning_rate": 0.0005, "loss": 1.1880557537078857, "step": 35760 }, { "epoch": 2.819156484036263, "grad_norm": 0.49160051345825195, "learning_rate": 0.0005, "loss": 1.337464690208435, "step": 35761 }, { "epoch": 2.819235317303902, "grad_norm": 0.4477297365665436, "learning_rate": 0.0005, "loss": 1.268028736114502, "step": 35762 }, { "epoch": 2.8193141505715413, "grad_norm": 0.5534898638725281, "learning_rate": 0.0005, "loss": 1.4777305126190186, "step": 35763 }, { "epoch": 2.81939298383918, "grad_norm": 0.6530993580818176, "learning_rate": 0.0005, "loss": 1.4752471446990967, "step": 35764 }, { "epoch": 2.819471817106819, "grad_norm": 0.7157095670700073, "learning_rate": 0.0005, "loss": 1.2989152669906616, "step": 35765 }, { "epoch": 2.819550650374458, "grad_norm": 0.5683404803276062, "learning_rate": 0.0005, "loss": 1.4198434352874756, "step": 35766 }, { "epoch": 2.819629483642097, "grad_norm": 0.5250231623649597, "learning_rate": 0.0005, "loss": 1.4752819538116455, "step": 35767 }, { "epoch": 2.819708316909736, "grad_norm": 0.4802071452140808, "learning_rate": 0.0005, "loss": 1.2888400554656982, "step": 35768 }, { "epoch": 2.819787150177375, "grad_norm": 0.550271213054657, "learning_rate": 0.0005, "loss": 1.2158230543136597, "step": 35769 }, { "epoch": 2.819865983445014, "grad_norm": 0.41977351903915405, "learning_rate": 0.0005, "loss": 1.1103814840316772, "step": 35770 }, { "epoch": 2.8199448167126526, "grad_norm": 0.44140639901161194, "learning_rate": 0.0005, "loss": 1.2711384296417236, "step": 35771 }, { "epoch": 2.8200236499802918, "grad_norm": 0.5695753693580627, "learning_rate": 0.0005, "loss": 1.5095808506011963, "step": 35772 }, { "epoch": 2.820102483247931, "grad_norm": 0.5353624224662781, "learning_rate": 0.0005, "loss": 1.2939443588256836, "step": 35773 }, { "epoch": 2.8201813165155696, "grad_norm": 0.4457278251647949, "learning_rate": 0.0005, "loss": 1.2705552577972412, "step": 35774 }, { "epoch": 2.8202601497832083, "grad_norm": 0.4701946973800659, "learning_rate": 0.0005, "loss": 1.4053564071655273, "step": 35775 }, { "epoch": 2.8203389830508474, "grad_norm": 0.4540344476699829, "learning_rate": 0.0005, "loss": 1.1998292207717896, "step": 35776 }, { "epoch": 2.8204178163184865, "grad_norm": 0.4919133484363556, "learning_rate": 0.0005, "loss": 1.33720862865448, "step": 35777 }, { "epoch": 2.8204966495861252, "grad_norm": 0.5313456654548645, "learning_rate": 0.0005, "loss": 1.4777779579162598, "step": 35778 }, { "epoch": 2.8205754828537644, "grad_norm": 0.44647422432899475, "learning_rate": 0.0005, "loss": 1.238926887512207, "step": 35779 }, { "epoch": 2.820654316121403, "grad_norm": 0.5187576413154602, "learning_rate": 0.0005, "loss": 1.3512372970581055, "step": 35780 }, { "epoch": 2.820733149389042, "grad_norm": 0.39514461159706116, "learning_rate": 0.0005, "loss": 1.223587989807129, "step": 35781 }, { "epoch": 2.8208119826566813, "grad_norm": 0.4163147211074829, "learning_rate": 0.0005, "loss": 1.2677326202392578, "step": 35782 }, { "epoch": 2.82089081592432, "grad_norm": 0.4519069790840149, "learning_rate": 0.0005, "loss": 1.3855960369110107, "step": 35783 }, { "epoch": 2.820969649191959, "grad_norm": 0.6048614978790283, "learning_rate": 0.0005, "loss": 1.3965413570404053, "step": 35784 }, { "epoch": 2.821048482459598, "grad_norm": 0.6736149191856384, "learning_rate": 0.0005, "loss": 1.4282140731811523, "step": 35785 }, { "epoch": 2.821127315727237, "grad_norm": 0.3892532289028168, "learning_rate": 0.0005, "loss": 1.0515165328979492, "step": 35786 }, { "epoch": 2.821206148994876, "grad_norm": 0.49771755933761597, "learning_rate": 0.0005, "loss": 1.1828522682189941, "step": 35787 }, { "epoch": 2.821284982262515, "grad_norm": 0.5405041575431824, "learning_rate": 0.0005, "loss": 1.2540037631988525, "step": 35788 }, { "epoch": 2.8213638155301535, "grad_norm": 0.6692376136779785, "learning_rate": 0.0005, "loss": 1.266603946685791, "step": 35789 }, { "epoch": 2.8214426487977926, "grad_norm": 0.6435408592224121, "learning_rate": 0.0005, "loss": 1.5519564151763916, "step": 35790 }, { "epoch": 2.8215214820654317, "grad_norm": 0.4982186257839203, "learning_rate": 0.0005, "loss": 1.3719698190689087, "step": 35791 }, { "epoch": 2.8216003153330704, "grad_norm": 0.6325652599334717, "learning_rate": 0.0005, "loss": 1.2045624256134033, "step": 35792 }, { "epoch": 2.8216791486007096, "grad_norm": 0.35822248458862305, "learning_rate": 0.0005, "loss": 1.15723717212677, "step": 35793 }, { "epoch": 2.8217579818683483, "grad_norm": 0.4924563765525818, "learning_rate": 0.0005, "loss": 1.232743501663208, "step": 35794 }, { "epoch": 2.8218368151359874, "grad_norm": 0.46534159779548645, "learning_rate": 0.0005, "loss": 1.3013139963150024, "step": 35795 }, { "epoch": 2.8219156484036265, "grad_norm": 0.4614122211933136, "learning_rate": 0.0005, "loss": 1.2070560455322266, "step": 35796 }, { "epoch": 2.821994481671265, "grad_norm": 0.5502068996429443, "learning_rate": 0.0005, "loss": 1.3655592203140259, "step": 35797 }, { "epoch": 2.8220733149389043, "grad_norm": 0.5329136848449707, "learning_rate": 0.0005, "loss": 1.3492215871810913, "step": 35798 }, { "epoch": 2.822152148206543, "grad_norm": 0.42011430859565735, "learning_rate": 0.0005, "loss": 1.248342514038086, "step": 35799 }, { "epoch": 2.822230981474182, "grad_norm": 0.5822644829750061, "learning_rate": 0.0005, "loss": 1.6130681037902832, "step": 35800 }, { "epoch": 2.8223098147418213, "grad_norm": 0.5714613199234009, "learning_rate": 0.0005, "loss": 1.3841729164123535, "step": 35801 }, { "epoch": 2.82238864800946, "grad_norm": 0.5545135736465454, "learning_rate": 0.0005, "loss": 1.400921106338501, "step": 35802 }, { "epoch": 2.8224674812770987, "grad_norm": 0.5180578827857971, "learning_rate": 0.0005, "loss": 1.421698808670044, "step": 35803 }, { "epoch": 2.822546314544738, "grad_norm": 0.5256057977676392, "learning_rate": 0.0005, "loss": 1.386359691619873, "step": 35804 }, { "epoch": 2.822625147812377, "grad_norm": 0.6165196299552917, "learning_rate": 0.0005, "loss": 1.5624966621398926, "step": 35805 }, { "epoch": 2.8227039810800156, "grad_norm": 0.4906075894832611, "learning_rate": 0.0005, "loss": 1.2774072885513306, "step": 35806 }, { "epoch": 2.8227828143476548, "grad_norm": 0.46398645639419556, "learning_rate": 0.0005, "loss": 1.2534695863723755, "step": 35807 }, { "epoch": 2.8228616476152935, "grad_norm": 0.4130566120147705, "learning_rate": 0.0005, "loss": 1.2110555171966553, "step": 35808 }, { "epoch": 2.8229404808829326, "grad_norm": 0.573638379573822, "learning_rate": 0.0005, "loss": 1.6638767719268799, "step": 35809 }, { "epoch": 2.8230193141505717, "grad_norm": 0.4585062861442566, "learning_rate": 0.0005, "loss": 1.1140098571777344, "step": 35810 }, { "epoch": 2.8230981474182104, "grad_norm": 0.40998220443725586, "learning_rate": 0.0005, "loss": 1.217627763748169, "step": 35811 }, { "epoch": 2.8231769806858495, "grad_norm": 0.5289225578308105, "learning_rate": 0.0005, "loss": 1.3578438758850098, "step": 35812 }, { "epoch": 2.8232558139534882, "grad_norm": 0.5343631505966187, "learning_rate": 0.0005, "loss": 1.4950332641601562, "step": 35813 }, { "epoch": 2.8233346472211274, "grad_norm": 0.3898032605648041, "learning_rate": 0.0005, "loss": 1.2013461589813232, "step": 35814 }, { "epoch": 2.8234134804887665, "grad_norm": 0.49867576360702515, "learning_rate": 0.0005, "loss": 1.3417277336120605, "step": 35815 }, { "epoch": 2.823492313756405, "grad_norm": 0.5624525547027588, "learning_rate": 0.0005, "loss": 1.4418772459030151, "step": 35816 }, { "epoch": 2.823571147024044, "grad_norm": 0.6070742011070251, "learning_rate": 0.0005, "loss": 1.4891371726989746, "step": 35817 }, { "epoch": 2.823649980291683, "grad_norm": 0.5023791193962097, "learning_rate": 0.0005, "loss": 1.2788341045379639, "step": 35818 }, { "epoch": 2.823728813559322, "grad_norm": 0.6515675783157349, "learning_rate": 0.0005, "loss": 1.5984511375427246, "step": 35819 }, { "epoch": 2.823807646826961, "grad_norm": 0.5708863735198975, "learning_rate": 0.0005, "loss": 1.495672583580017, "step": 35820 }, { "epoch": 2.8238864800946, "grad_norm": 0.49415987730026245, "learning_rate": 0.0005, "loss": 1.3287806510925293, "step": 35821 }, { "epoch": 2.8239653133622387, "grad_norm": 0.4585254490375519, "learning_rate": 0.0005, "loss": 1.2027604579925537, "step": 35822 }, { "epoch": 2.824044146629878, "grad_norm": 0.41193750500679016, "learning_rate": 0.0005, "loss": 1.1002228260040283, "step": 35823 }, { "epoch": 2.824122979897517, "grad_norm": 0.5599190592765808, "learning_rate": 0.0005, "loss": 1.283827781677246, "step": 35824 }, { "epoch": 2.8242018131651556, "grad_norm": 0.4776364266872406, "learning_rate": 0.0005, "loss": 1.5108566284179688, "step": 35825 }, { "epoch": 2.8242806464327948, "grad_norm": 0.5305831432342529, "learning_rate": 0.0005, "loss": 1.2914257049560547, "step": 35826 }, { "epoch": 2.8243594797004334, "grad_norm": 0.4344753324985504, "learning_rate": 0.0005, "loss": 1.166401743888855, "step": 35827 }, { "epoch": 2.8244383129680726, "grad_norm": 0.43287813663482666, "learning_rate": 0.0005, "loss": 1.3248260021209717, "step": 35828 }, { "epoch": 2.8245171462357117, "grad_norm": 0.5464059114456177, "learning_rate": 0.0005, "loss": 1.6800007820129395, "step": 35829 }, { "epoch": 2.8245959795033504, "grad_norm": 0.4532462954521179, "learning_rate": 0.0005, "loss": 1.274848461151123, "step": 35830 }, { "epoch": 2.824674812770989, "grad_norm": 0.4929879903793335, "learning_rate": 0.0005, "loss": 1.2738895416259766, "step": 35831 }, { "epoch": 2.824753646038628, "grad_norm": 0.4952239990234375, "learning_rate": 0.0005, "loss": 1.3243801593780518, "step": 35832 }, { "epoch": 2.8248324793062674, "grad_norm": 0.47966495156288147, "learning_rate": 0.0005, "loss": 1.291719913482666, "step": 35833 }, { "epoch": 2.824911312573906, "grad_norm": 0.6024839878082275, "learning_rate": 0.0005, "loss": 1.4362924098968506, "step": 35834 }, { "epoch": 2.824990145841545, "grad_norm": 1.3067147731781006, "learning_rate": 0.0005, "loss": 1.3109030723571777, "step": 35835 }, { "epoch": 2.825068979109184, "grad_norm": 1.0368140935897827, "learning_rate": 0.0005, "loss": 1.397423267364502, "step": 35836 }, { "epoch": 2.825147812376823, "grad_norm": 0.5231382846832275, "learning_rate": 0.0005, "loss": 1.3436927795410156, "step": 35837 }, { "epoch": 2.825226645644462, "grad_norm": 0.4886922836303711, "learning_rate": 0.0005, "loss": 1.3019318580627441, "step": 35838 }, { "epoch": 2.825305478912101, "grad_norm": 0.5262094140052795, "learning_rate": 0.0005, "loss": 1.2944953441619873, "step": 35839 }, { "epoch": 2.82538431217974, "grad_norm": 0.5183604955673218, "learning_rate": 0.0005, "loss": 1.1634855270385742, "step": 35840 }, { "epoch": 2.8254631454473786, "grad_norm": 0.39219510555267334, "learning_rate": 0.0005, "loss": 1.1123777627944946, "step": 35841 }, { "epoch": 2.8255419787150178, "grad_norm": 0.4982168674468994, "learning_rate": 0.0005, "loss": 1.4146411418914795, "step": 35842 }, { "epoch": 2.825620811982657, "grad_norm": 0.42963457107543945, "learning_rate": 0.0005, "loss": 1.1820287704467773, "step": 35843 }, { "epoch": 2.8256996452502956, "grad_norm": 0.42941638827323914, "learning_rate": 0.0005, "loss": 1.2493306398391724, "step": 35844 }, { "epoch": 2.8257784785179347, "grad_norm": 0.5124224424362183, "learning_rate": 0.0005, "loss": 1.429282546043396, "step": 35845 }, { "epoch": 2.8258573117855734, "grad_norm": 0.5268060564994812, "learning_rate": 0.0005, "loss": 1.3866147994995117, "step": 35846 }, { "epoch": 2.8259361450532126, "grad_norm": 0.5345860123634338, "learning_rate": 0.0005, "loss": 1.390498161315918, "step": 35847 }, { "epoch": 2.8260149783208512, "grad_norm": 0.572797954082489, "learning_rate": 0.0005, "loss": 1.2798705101013184, "step": 35848 }, { "epoch": 2.8260938115884904, "grad_norm": 0.44871851801872253, "learning_rate": 0.0005, "loss": 1.261012077331543, "step": 35849 }, { "epoch": 2.826172644856129, "grad_norm": 0.5196883082389832, "learning_rate": 0.0005, "loss": 1.206886649131775, "step": 35850 }, { "epoch": 2.826251478123768, "grad_norm": 0.7343745231628418, "learning_rate": 0.0005, "loss": 1.4477946758270264, "step": 35851 }, { "epoch": 2.8263303113914073, "grad_norm": 0.5350512266159058, "learning_rate": 0.0005, "loss": 1.465003490447998, "step": 35852 }, { "epoch": 2.826409144659046, "grad_norm": 0.47623488306999207, "learning_rate": 0.0005, "loss": 1.1583082675933838, "step": 35853 }, { "epoch": 2.826487977926685, "grad_norm": 0.4392838478088379, "learning_rate": 0.0005, "loss": 1.1891547441482544, "step": 35854 }, { "epoch": 2.826566811194324, "grad_norm": 0.5067265033721924, "learning_rate": 0.0005, "loss": 1.2912721633911133, "step": 35855 }, { "epoch": 2.826645644461963, "grad_norm": 0.5697924494743347, "learning_rate": 0.0005, "loss": 1.6761860847473145, "step": 35856 }, { "epoch": 2.826724477729602, "grad_norm": 0.4960743188858032, "learning_rate": 0.0005, "loss": 1.2751920223236084, "step": 35857 }, { "epoch": 2.826803310997241, "grad_norm": 0.562401533126831, "learning_rate": 0.0005, "loss": 1.3326001167297363, "step": 35858 }, { "epoch": 2.82688214426488, "grad_norm": 0.4006728529930115, "learning_rate": 0.0005, "loss": 1.2814786434173584, "step": 35859 }, { "epoch": 2.8269609775325186, "grad_norm": 0.54234379529953, "learning_rate": 0.0005, "loss": 1.3099517822265625, "step": 35860 }, { "epoch": 2.8270398108001578, "grad_norm": 0.44445887207984924, "learning_rate": 0.0005, "loss": 1.2556155920028687, "step": 35861 }, { "epoch": 2.8271186440677964, "grad_norm": 0.5501300096511841, "learning_rate": 0.0005, "loss": 1.484682559967041, "step": 35862 }, { "epoch": 2.8271974773354356, "grad_norm": 0.41443321108818054, "learning_rate": 0.0005, "loss": 1.365220069885254, "step": 35863 }, { "epoch": 2.8272763106030743, "grad_norm": 0.6052572727203369, "learning_rate": 0.0005, "loss": 1.4261841773986816, "step": 35864 }, { "epoch": 2.8273551438707134, "grad_norm": 0.5329552888870239, "learning_rate": 0.0005, "loss": 1.2656912803649902, "step": 35865 }, { "epoch": 2.8274339771383525, "grad_norm": 0.5201075673103333, "learning_rate": 0.0005, "loss": 1.5898833274841309, "step": 35866 }, { "epoch": 2.8275128104059912, "grad_norm": 0.49059200286865234, "learning_rate": 0.0005, "loss": 1.6247594356536865, "step": 35867 }, { "epoch": 2.8275916436736304, "grad_norm": 0.5110724568367004, "learning_rate": 0.0005, "loss": 1.345461130142212, "step": 35868 }, { "epoch": 2.827670476941269, "grad_norm": 0.4781055748462677, "learning_rate": 0.0005, "loss": 1.2208657264709473, "step": 35869 }, { "epoch": 2.827749310208908, "grad_norm": 0.4752064645290375, "learning_rate": 0.0005, "loss": 1.236767053604126, "step": 35870 }, { "epoch": 2.8278281434765473, "grad_norm": 0.43699514865875244, "learning_rate": 0.0005, "loss": 1.311865210533142, "step": 35871 }, { "epoch": 2.827906976744186, "grad_norm": 0.44895437359809875, "learning_rate": 0.0005, "loss": 1.2702343463897705, "step": 35872 }, { "epoch": 2.827985810011825, "grad_norm": 0.493808388710022, "learning_rate": 0.0005, "loss": 1.3617299795150757, "step": 35873 }, { "epoch": 2.828064643279464, "grad_norm": 0.515774130821228, "learning_rate": 0.0005, "loss": 1.431828260421753, "step": 35874 }, { "epoch": 2.828143476547103, "grad_norm": 0.5214526653289795, "learning_rate": 0.0005, "loss": 1.2576576471328735, "step": 35875 }, { "epoch": 2.828222309814742, "grad_norm": 0.5013112425804138, "learning_rate": 0.0005, "loss": 1.2642035484313965, "step": 35876 }, { "epoch": 2.828301143082381, "grad_norm": 0.5119330286979675, "learning_rate": 0.0005, "loss": 1.4537274837493896, "step": 35877 }, { "epoch": 2.8283799763500195, "grad_norm": 0.5846436023712158, "learning_rate": 0.0005, "loss": 1.3996658325195312, "step": 35878 }, { "epoch": 2.8284588096176586, "grad_norm": 0.46797895431518555, "learning_rate": 0.0005, "loss": 1.1213220357894897, "step": 35879 }, { "epoch": 2.8285376428852977, "grad_norm": 0.44247525930404663, "learning_rate": 0.0005, "loss": 1.2757396697998047, "step": 35880 }, { "epoch": 2.8286164761529364, "grad_norm": 0.5165911912918091, "learning_rate": 0.0005, "loss": 1.2294354438781738, "step": 35881 }, { "epoch": 2.8286953094205756, "grad_norm": 0.4883374273777008, "learning_rate": 0.0005, "loss": 1.2255029678344727, "step": 35882 }, { "epoch": 2.8287741426882143, "grad_norm": 0.6169780492782593, "learning_rate": 0.0005, "loss": 1.3713691234588623, "step": 35883 }, { "epoch": 2.8288529759558534, "grad_norm": 0.5165230631828308, "learning_rate": 0.0005, "loss": 1.2684826850891113, "step": 35884 }, { "epoch": 2.8289318092234925, "grad_norm": 0.5322999954223633, "learning_rate": 0.0005, "loss": 1.2869147062301636, "step": 35885 }, { "epoch": 2.829010642491131, "grad_norm": 0.5465141534805298, "learning_rate": 0.0005, "loss": 1.4553786516189575, "step": 35886 }, { "epoch": 2.8290894757587703, "grad_norm": 0.5730521082878113, "learning_rate": 0.0005, "loss": 1.4129700660705566, "step": 35887 }, { "epoch": 2.829168309026409, "grad_norm": 0.513474702835083, "learning_rate": 0.0005, "loss": 1.5256876945495605, "step": 35888 }, { "epoch": 2.829247142294048, "grad_norm": 0.46530523896217346, "learning_rate": 0.0005, "loss": 1.358452558517456, "step": 35889 }, { "epoch": 2.8293259755616873, "grad_norm": 0.5663270354270935, "learning_rate": 0.0005, "loss": 1.283909559249878, "step": 35890 }, { "epoch": 2.829404808829326, "grad_norm": 0.5295841693878174, "learning_rate": 0.0005, "loss": 1.2063766717910767, "step": 35891 }, { "epoch": 2.8294836420969647, "grad_norm": 0.4135369658470154, "learning_rate": 0.0005, "loss": 1.1695492267608643, "step": 35892 }, { "epoch": 2.829562475364604, "grad_norm": 0.5104460120201111, "learning_rate": 0.0005, "loss": 1.1406828165054321, "step": 35893 }, { "epoch": 2.829641308632243, "grad_norm": 0.4784514605998993, "learning_rate": 0.0005, "loss": 1.3251032829284668, "step": 35894 }, { "epoch": 2.8297201418998816, "grad_norm": 2.005599021911621, "learning_rate": 0.0005, "loss": 1.583680272102356, "step": 35895 }, { "epoch": 2.8297989751675208, "grad_norm": 0.6163385510444641, "learning_rate": 0.0005, "loss": 1.3934252262115479, "step": 35896 }, { "epoch": 2.8298778084351595, "grad_norm": 0.5252891778945923, "learning_rate": 0.0005, "loss": 1.423999309539795, "step": 35897 }, { "epoch": 2.8299566417027986, "grad_norm": 0.47469562292099, "learning_rate": 0.0005, "loss": 1.287034273147583, "step": 35898 }, { "epoch": 2.8300354749704377, "grad_norm": 0.4703845977783203, "learning_rate": 0.0005, "loss": 1.2418699264526367, "step": 35899 }, { "epoch": 2.8301143082380764, "grad_norm": 0.5410225987434387, "learning_rate": 0.0005, "loss": 1.1983741521835327, "step": 35900 }, { "epoch": 2.8301931415057155, "grad_norm": 0.5513836741447449, "learning_rate": 0.0005, "loss": 1.333040714263916, "step": 35901 }, { "epoch": 2.8302719747733542, "grad_norm": 0.6212021112442017, "learning_rate": 0.0005, "loss": 1.2663509845733643, "step": 35902 }, { "epoch": 2.8303508080409934, "grad_norm": 0.6121837496757507, "learning_rate": 0.0005, "loss": 1.5179991722106934, "step": 35903 }, { "epoch": 2.8304296413086325, "grad_norm": 0.41596561670303345, "learning_rate": 0.0005, "loss": 1.2418068647384644, "step": 35904 }, { "epoch": 2.830508474576271, "grad_norm": 0.5966029167175293, "learning_rate": 0.0005, "loss": 1.2949597835540771, "step": 35905 }, { "epoch": 2.83058730784391, "grad_norm": 0.46667566895484924, "learning_rate": 0.0005, "loss": 1.311536192893982, "step": 35906 }, { "epoch": 2.830666141111549, "grad_norm": 0.40420910716056824, "learning_rate": 0.0005, "loss": 1.2903574705123901, "step": 35907 }, { "epoch": 2.830744974379188, "grad_norm": 0.6231656074523926, "learning_rate": 0.0005, "loss": 1.3683922290802002, "step": 35908 }, { "epoch": 2.830823807646827, "grad_norm": 0.5149691104888916, "learning_rate": 0.0005, "loss": 1.377389907836914, "step": 35909 }, { "epoch": 2.830902640914466, "grad_norm": 0.7344058156013489, "learning_rate": 0.0005, "loss": 1.1802489757537842, "step": 35910 }, { "epoch": 2.8309814741821047, "grad_norm": 0.5749437212944031, "learning_rate": 0.0005, "loss": 1.360217809677124, "step": 35911 }, { "epoch": 2.831060307449744, "grad_norm": 0.6261962652206421, "learning_rate": 0.0005, "loss": 1.4253745079040527, "step": 35912 }, { "epoch": 2.831139140717383, "grad_norm": 0.6173784732818604, "learning_rate": 0.0005, "loss": 1.253833532333374, "step": 35913 }, { "epoch": 2.8312179739850216, "grad_norm": 0.4878354072570801, "learning_rate": 0.0005, "loss": 1.14054536819458, "step": 35914 }, { "epoch": 2.8312968072526608, "grad_norm": 0.46029454469680786, "learning_rate": 0.0005, "loss": 1.2027761936187744, "step": 35915 }, { "epoch": 2.8313756405202994, "grad_norm": 0.5876262784004211, "learning_rate": 0.0005, "loss": 1.0615384578704834, "step": 35916 }, { "epoch": 2.8314544737879386, "grad_norm": 0.5520135164260864, "learning_rate": 0.0005, "loss": 1.2895504236221313, "step": 35917 }, { "epoch": 2.8315333070555777, "grad_norm": 0.4574648141860962, "learning_rate": 0.0005, "loss": 1.1978508234024048, "step": 35918 }, { "epoch": 2.8316121403232164, "grad_norm": 0.45049139857292175, "learning_rate": 0.0005, "loss": 1.1330033540725708, "step": 35919 }, { "epoch": 2.831690973590855, "grad_norm": 0.6459941267967224, "learning_rate": 0.0005, "loss": 1.3442708253860474, "step": 35920 }, { "epoch": 2.831769806858494, "grad_norm": 0.47691935300827026, "learning_rate": 0.0005, "loss": 1.1772961616516113, "step": 35921 }, { "epoch": 2.8318486401261334, "grad_norm": 0.5157651901245117, "learning_rate": 0.0005, "loss": 1.634352684020996, "step": 35922 }, { "epoch": 2.831927473393772, "grad_norm": 0.6930768489837646, "learning_rate": 0.0005, "loss": 1.2643375396728516, "step": 35923 }, { "epoch": 2.832006306661411, "grad_norm": 0.4208790361881256, "learning_rate": 0.0005, "loss": 1.1749545335769653, "step": 35924 }, { "epoch": 2.83208513992905, "grad_norm": 0.5217435359954834, "learning_rate": 0.0005, "loss": 1.3452880382537842, "step": 35925 }, { "epoch": 2.832163973196689, "grad_norm": 0.4833647310733795, "learning_rate": 0.0005, "loss": 1.2862573862075806, "step": 35926 }, { "epoch": 2.832242806464328, "grad_norm": 0.8919006586074829, "learning_rate": 0.0005, "loss": 1.8380905389785767, "step": 35927 }, { "epoch": 2.832321639731967, "grad_norm": 0.5299553871154785, "learning_rate": 0.0005, "loss": 1.5274593830108643, "step": 35928 }, { "epoch": 2.832400472999606, "grad_norm": 0.6821218729019165, "learning_rate": 0.0005, "loss": 1.3422125577926636, "step": 35929 }, { "epoch": 2.8324793062672446, "grad_norm": 0.5546900033950806, "learning_rate": 0.0005, "loss": 1.2165130376815796, "step": 35930 }, { "epoch": 2.832558139534884, "grad_norm": 0.5234493017196655, "learning_rate": 0.0005, "loss": 1.3998661041259766, "step": 35931 }, { "epoch": 2.832636972802523, "grad_norm": 0.6322288513183594, "learning_rate": 0.0005, "loss": 1.2490910291671753, "step": 35932 }, { "epoch": 2.8327158060701616, "grad_norm": 0.5042245388031006, "learning_rate": 0.0005, "loss": 1.305911660194397, "step": 35933 }, { "epoch": 2.8327946393378003, "grad_norm": 0.5950256586074829, "learning_rate": 0.0005, "loss": 1.4199410676956177, "step": 35934 }, { "epoch": 2.8328734726054394, "grad_norm": 0.6512607336044312, "learning_rate": 0.0005, "loss": 1.3223124742507935, "step": 35935 }, { "epoch": 2.8329523058730786, "grad_norm": 0.5433279275894165, "learning_rate": 0.0005, "loss": 1.3280351161956787, "step": 35936 }, { "epoch": 2.8330311391407172, "grad_norm": 0.5684239864349365, "learning_rate": 0.0005, "loss": 1.4418812990188599, "step": 35937 }, { "epoch": 2.8331099724083564, "grad_norm": 0.508030354976654, "learning_rate": 0.0005, "loss": 1.4436410665512085, "step": 35938 }, { "epoch": 2.833188805675995, "grad_norm": 0.4560818076133728, "learning_rate": 0.0005, "loss": 1.2467570304870605, "step": 35939 }, { "epoch": 2.833267638943634, "grad_norm": 0.5350269675254822, "learning_rate": 0.0005, "loss": 1.460713267326355, "step": 35940 }, { "epoch": 2.8333464722112733, "grad_norm": 0.4732646942138672, "learning_rate": 0.0005, "loss": 1.3477784395217896, "step": 35941 }, { "epoch": 2.833425305478912, "grad_norm": 0.4764868915081024, "learning_rate": 0.0005, "loss": 1.3742750883102417, "step": 35942 }, { "epoch": 2.833504138746551, "grad_norm": 0.453554630279541, "learning_rate": 0.0005, "loss": 1.408450961112976, "step": 35943 }, { "epoch": 2.83358297201419, "grad_norm": 0.461052268743515, "learning_rate": 0.0005, "loss": 1.4304418563842773, "step": 35944 }, { "epoch": 2.833661805281829, "grad_norm": 0.542158305644989, "learning_rate": 0.0005, "loss": 1.4388575553894043, "step": 35945 }, { "epoch": 2.833740638549468, "grad_norm": 0.5309229493141174, "learning_rate": 0.0005, "loss": 1.345249891281128, "step": 35946 }, { "epoch": 2.833819471817107, "grad_norm": 0.4710659086704254, "learning_rate": 0.0005, "loss": 1.2049740552902222, "step": 35947 }, { "epoch": 2.8338983050847455, "grad_norm": 0.5408002138137817, "learning_rate": 0.0005, "loss": 1.5873732566833496, "step": 35948 }, { "epoch": 2.8339771383523846, "grad_norm": 0.5738587975502014, "learning_rate": 0.0005, "loss": 1.5502291917800903, "step": 35949 }, { "epoch": 2.8340559716200238, "grad_norm": 0.4674985110759735, "learning_rate": 0.0005, "loss": 1.3863674402236938, "step": 35950 }, { "epoch": 2.8341348048876625, "grad_norm": 0.5341168642044067, "learning_rate": 0.0005, "loss": 1.4720944166183472, "step": 35951 }, { "epoch": 2.8342136381553016, "grad_norm": 0.45054104924201965, "learning_rate": 0.0005, "loss": 1.4202960729599, "step": 35952 }, { "epoch": 2.8342924714229403, "grad_norm": 0.5103870630264282, "learning_rate": 0.0005, "loss": 1.5580312013626099, "step": 35953 }, { "epoch": 2.8343713046905794, "grad_norm": 0.638188362121582, "learning_rate": 0.0005, "loss": 1.4369423389434814, "step": 35954 }, { "epoch": 2.8344501379582185, "grad_norm": 0.48764362931251526, "learning_rate": 0.0005, "loss": 1.3509516716003418, "step": 35955 }, { "epoch": 2.8345289712258572, "grad_norm": 0.8060693740844727, "learning_rate": 0.0005, "loss": 1.1868844032287598, "step": 35956 }, { "epoch": 2.8346078044934964, "grad_norm": 0.3917962312698364, "learning_rate": 0.0005, "loss": 1.0709748268127441, "step": 35957 }, { "epoch": 2.834686637761135, "grad_norm": 0.5994484424591064, "learning_rate": 0.0005, "loss": 1.5700833797454834, "step": 35958 }, { "epoch": 2.834765471028774, "grad_norm": 0.5291974544525146, "learning_rate": 0.0005, "loss": 1.3961843252182007, "step": 35959 }, { "epoch": 2.8348443042964133, "grad_norm": 0.49756908416748047, "learning_rate": 0.0005, "loss": 1.3315457105636597, "step": 35960 }, { "epoch": 2.834923137564052, "grad_norm": 0.4606102705001831, "learning_rate": 0.0005, "loss": 1.4932020902633667, "step": 35961 }, { "epoch": 2.8350019708316907, "grad_norm": 0.5408952236175537, "learning_rate": 0.0005, "loss": 1.5486359596252441, "step": 35962 }, { "epoch": 2.83508080409933, "grad_norm": 0.4422507584095001, "learning_rate": 0.0005, "loss": 1.2358866930007935, "step": 35963 }, { "epoch": 2.835159637366969, "grad_norm": 0.45521387457847595, "learning_rate": 0.0005, "loss": 1.3131358623504639, "step": 35964 }, { "epoch": 2.8352384706346077, "grad_norm": 0.4741263687610626, "learning_rate": 0.0005, "loss": 1.4227046966552734, "step": 35965 }, { "epoch": 2.835317303902247, "grad_norm": 0.6017748117446899, "learning_rate": 0.0005, "loss": 1.2723290920257568, "step": 35966 }, { "epoch": 2.8353961371698855, "grad_norm": 0.5354647636413574, "learning_rate": 0.0005, "loss": 1.3699674606323242, "step": 35967 }, { "epoch": 2.8354749704375246, "grad_norm": 0.5708469748497009, "learning_rate": 0.0005, "loss": 1.3104174137115479, "step": 35968 }, { "epoch": 2.8355538037051637, "grad_norm": 0.5183053612709045, "learning_rate": 0.0005, "loss": 1.4351178407669067, "step": 35969 }, { "epoch": 2.8356326369728024, "grad_norm": 0.4396687150001526, "learning_rate": 0.0005, "loss": 1.1738052368164062, "step": 35970 }, { "epoch": 2.8357114702404416, "grad_norm": 0.4702242910861969, "learning_rate": 0.0005, "loss": 1.3790833950042725, "step": 35971 }, { "epoch": 2.8357903035080803, "grad_norm": 0.5137245655059814, "learning_rate": 0.0005, "loss": 1.2787396907806396, "step": 35972 }, { "epoch": 2.8358691367757194, "grad_norm": 0.5599477887153625, "learning_rate": 0.0005, "loss": 1.3544459342956543, "step": 35973 }, { "epoch": 2.8359479700433585, "grad_norm": 0.41885992884635925, "learning_rate": 0.0005, "loss": 1.253818154335022, "step": 35974 }, { "epoch": 2.836026803310997, "grad_norm": 0.48467719554901123, "learning_rate": 0.0005, "loss": 1.2548749446868896, "step": 35975 }, { "epoch": 2.8361056365786363, "grad_norm": 0.5659629702568054, "learning_rate": 0.0005, "loss": 1.5448533296585083, "step": 35976 }, { "epoch": 2.836184469846275, "grad_norm": 0.4592118263244629, "learning_rate": 0.0005, "loss": 1.241767168045044, "step": 35977 }, { "epoch": 2.836263303113914, "grad_norm": 0.6186793446540833, "learning_rate": 0.0005, "loss": 1.6568489074707031, "step": 35978 }, { "epoch": 2.836342136381553, "grad_norm": 0.39196065068244934, "learning_rate": 0.0005, "loss": 1.1997848749160767, "step": 35979 }, { "epoch": 2.836420969649192, "grad_norm": 0.47024786472320557, "learning_rate": 0.0005, "loss": 1.2444138526916504, "step": 35980 }, { "epoch": 2.8364998029168307, "grad_norm": 0.4695277214050293, "learning_rate": 0.0005, "loss": 1.1732203960418701, "step": 35981 }, { "epoch": 2.83657863618447, "grad_norm": 0.4634009301662445, "learning_rate": 0.0005, "loss": 1.2553821802139282, "step": 35982 }, { "epoch": 2.836657469452109, "grad_norm": 0.5172184705734253, "learning_rate": 0.0005, "loss": 1.4051152467727661, "step": 35983 }, { "epoch": 2.8367363027197476, "grad_norm": 0.4886857867240906, "learning_rate": 0.0005, "loss": 1.377942681312561, "step": 35984 }, { "epoch": 2.8368151359873868, "grad_norm": 0.4201924502849579, "learning_rate": 0.0005, "loss": 1.1404966115951538, "step": 35985 }, { "epoch": 2.8368939692550255, "grad_norm": 0.5298252701759338, "learning_rate": 0.0005, "loss": 1.5616064071655273, "step": 35986 }, { "epoch": 2.8369728025226646, "grad_norm": 0.49639540910720825, "learning_rate": 0.0005, "loss": 1.1454969644546509, "step": 35987 }, { "epoch": 2.8370516357903037, "grad_norm": 0.45846402645111084, "learning_rate": 0.0005, "loss": 1.287069320678711, "step": 35988 }, { "epoch": 2.8371304690579424, "grad_norm": 0.4403228759765625, "learning_rate": 0.0005, "loss": 1.2651119232177734, "step": 35989 }, { "epoch": 2.8372093023255816, "grad_norm": 0.5763353705406189, "learning_rate": 0.0005, "loss": 1.339924931526184, "step": 35990 }, { "epoch": 2.8372881355932202, "grad_norm": 0.4044578969478607, "learning_rate": 0.0005, "loss": 1.2551825046539307, "step": 35991 }, { "epoch": 2.8373669688608594, "grad_norm": 0.4548051655292511, "learning_rate": 0.0005, "loss": 1.3466136455535889, "step": 35992 }, { "epoch": 2.837445802128498, "grad_norm": 0.529818594455719, "learning_rate": 0.0005, "loss": 1.25655198097229, "step": 35993 }, { "epoch": 2.837524635396137, "grad_norm": 0.4076502323150635, "learning_rate": 0.0005, "loss": 1.166243314743042, "step": 35994 }, { "epoch": 2.837603468663776, "grad_norm": 0.566624641418457, "learning_rate": 0.0005, "loss": 1.5172741413116455, "step": 35995 }, { "epoch": 2.837682301931415, "grad_norm": 0.439637690782547, "learning_rate": 0.0005, "loss": 1.3937287330627441, "step": 35996 }, { "epoch": 2.837761135199054, "grad_norm": 0.46708744764328003, "learning_rate": 0.0005, "loss": 1.3891041278839111, "step": 35997 }, { "epoch": 2.837839968466693, "grad_norm": 0.4378148913383484, "learning_rate": 0.0005, "loss": 1.3309625387191772, "step": 35998 }, { "epoch": 2.837918801734332, "grad_norm": 0.4446154236793518, "learning_rate": 0.0005, "loss": 1.2771615982055664, "step": 35999 }, { "epoch": 2.8379976350019707, "grad_norm": 3.521660327911377, "learning_rate": 0.0005, "loss": 1.4601085186004639, "step": 36000 }, { "epoch": 2.83807646826961, "grad_norm": 0.7164475917816162, "learning_rate": 0.0005, "loss": 1.4622173309326172, "step": 36001 }, { "epoch": 2.838155301537249, "grad_norm": 0.6148515939712524, "learning_rate": 0.0005, "loss": 1.5326544046401978, "step": 36002 }, { "epoch": 2.8382341348048876, "grad_norm": 0.46418026089668274, "learning_rate": 0.0005, "loss": 1.3089656829833984, "step": 36003 }, { "epoch": 2.8383129680725268, "grad_norm": 0.5261593461036682, "learning_rate": 0.0005, "loss": 1.2466139793395996, "step": 36004 }, { "epoch": 2.8383918013401654, "grad_norm": 0.5643236041069031, "learning_rate": 0.0005, "loss": 1.4368743896484375, "step": 36005 }, { "epoch": 2.8384706346078046, "grad_norm": 0.4956235885620117, "learning_rate": 0.0005, "loss": 1.2969361543655396, "step": 36006 }, { "epoch": 2.8385494678754437, "grad_norm": 0.48850318789482117, "learning_rate": 0.0005, "loss": 1.4005346298217773, "step": 36007 }, { "epoch": 2.8386283011430824, "grad_norm": 0.4170575737953186, "learning_rate": 0.0005, "loss": 1.2080739736557007, "step": 36008 }, { "epoch": 2.838707134410721, "grad_norm": 0.8164767622947693, "learning_rate": 0.0005, "loss": 1.4144420623779297, "step": 36009 }, { "epoch": 2.8387859676783602, "grad_norm": 0.535352885723114, "learning_rate": 0.0005, "loss": 1.3170216083526611, "step": 36010 }, { "epoch": 2.8388648009459994, "grad_norm": 0.5120034217834473, "learning_rate": 0.0005, "loss": 1.3397966623306274, "step": 36011 }, { "epoch": 2.838943634213638, "grad_norm": 0.421056866645813, "learning_rate": 0.0005, "loss": 1.1174484491348267, "step": 36012 }, { "epoch": 2.839022467481277, "grad_norm": 0.531882107257843, "learning_rate": 0.0005, "loss": 1.4125128984451294, "step": 36013 }, { "epoch": 2.839101300748916, "grad_norm": 0.4955808222293854, "learning_rate": 0.0005, "loss": 1.3637065887451172, "step": 36014 }, { "epoch": 2.839180134016555, "grad_norm": 0.48900142312049866, "learning_rate": 0.0005, "loss": 1.4732847213745117, "step": 36015 }, { "epoch": 2.839258967284194, "grad_norm": 0.46672606468200684, "learning_rate": 0.0005, "loss": 1.3532817363739014, "step": 36016 }, { "epoch": 2.839337800551833, "grad_norm": 0.46012648940086365, "learning_rate": 0.0005, "loss": 1.247389316558838, "step": 36017 }, { "epoch": 2.839416633819472, "grad_norm": 0.48177915811538696, "learning_rate": 0.0005, "loss": 1.2821934223175049, "step": 36018 }, { "epoch": 2.8394954670871106, "grad_norm": 0.6180341839790344, "learning_rate": 0.0005, "loss": 1.5774823427200317, "step": 36019 }, { "epoch": 2.83957430035475, "grad_norm": 0.5539329648017883, "learning_rate": 0.0005, "loss": 1.4658123254776, "step": 36020 }, { "epoch": 2.839653133622389, "grad_norm": 0.47739019989967346, "learning_rate": 0.0005, "loss": 1.2901785373687744, "step": 36021 }, { "epoch": 2.8397319668900276, "grad_norm": 0.47414588928222656, "learning_rate": 0.0005, "loss": 1.2700432538986206, "step": 36022 }, { "epoch": 2.8398108001576663, "grad_norm": 0.731360912322998, "learning_rate": 0.0005, "loss": 1.3833407163619995, "step": 36023 }, { "epoch": 2.8398896334253054, "grad_norm": 0.5135458707809448, "learning_rate": 0.0005, "loss": 1.2992300987243652, "step": 36024 }, { "epoch": 2.8399684666929446, "grad_norm": 0.5217403173446655, "learning_rate": 0.0005, "loss": 1.2807576656341553, "step": 36025 }, { "epoch": 2.8400472999605832, "grad_norm": 0.5373261570930481, "learning_rate": 0.0005, "loss": 1.4079113006591797, "step": 36026 }, { "epoch": 2.8401261332282224, "grad_norm": 0.4972243010997772, "learning_rate": 0.0005, "loss": 1.209398627281189, "step": 36027 }, { "epoch": 2.840204966495861, "grad_norm": 0.5656737089157104, "learning_rate": 0.0005, "loss": 1.3547139167785645, "step": 36028 }, { "epoch": 2.8402837997635, "grad_norm": 0.3971233069896698, "learning_rate": 0.0005, "loss": 1.2824993133544922, "step": 36029 }, { "epoch": 2.8403626330311393, "grad_norm": 0.5633384585380554, "learning_rate": 0.0005, "loss": 1.326672911643982, "step": 36030 }, { "epoch": 2.840441466298778, "grad_norm": 0.4122627079486847, "learning_rate": 0.0005, "loss": 1.2575488090515137, "step": 36031 }, { "epoch": 2.840520299566417, "grad_norm": 0.4744301736354828, "learning_rate": 0.0005, "loss": 1.3646748065948486, "step": 36032 }, { "epoch": 2.840599132834056, "grad_norm": 0.4309132397174835, "learning_rate": 0.0005, "loss": 1.2736061811447144, "step": 36033 }, { "epoch": 2.840677966101695, "grad_norm": 0.4231986105442047, "learning_rate": 0.0005, "loss": 1.18625009059906, "step": 36034 }, { "epoch": 2.840756799369334, "grad_norm": 0.4389064610004425, "learning_rate": 0.0005, "loss": 1.2097980976104736, "step": 36035 }, { "epoch": 2.840835632636973, "grad_norm": 0.39507246017456055, "learning_rate": 0.0005, "loss": 1.2769660949707031, "step": 36036 }, { "epoch": 2.8409144659046115, "grad_norm": 0.4987063705921173, "learning_rate": 0.0005, "loss": 1.3751758337020874, "step": 36037 }, { "epoch": 2.8409932991722506, "grad_norm": 0.42613160610198975, "learning_rate": 0.0005, "loss": 1.3905409574508667, "step": 36038 }, { "epoch": 2.8410721324398898, "grad_norm": 0.47142294049263, "learning_rate": 0.0005, "loss": 1.2026293277740479, "step": 36039 }, { "epoch": 2.8411509657075285, "grad_norm": 0.483830064535141, "learning_rate": 0.0005, "loss": 1.2031383514404297, "step": 36040 }, { "epoch": 2.8412297989751676, "grad_norm": 0.5145517587661743, "learning_rate": 0.0005, "loss": 1.1772050857543945, "step": 36041 }, { "epoch": 2.8413086322428063, "grad_norm": 0.49471598863601685, "learning_rate": 0.0005, "loss": 1.248831868171692, "step": 36042 }, { "epoch": 2.8413874655104454, "grad_norm": 0.46926626563072205, "learning_rate": 0.0005, "loss": 1.3720380067825317, "step": 36043 }, { "epoch": 2.8414662987780845, "grad_norm": 0.520767867565155, "learning_rate": 0.0005, "loss": 1.4141185283660889, "step": 36044 }, { "epoch": 2.8415451320457232, "grad_norm": 0.5098776817321777, "learning_rate": 0.0005, "loss": 1.3875033855438232, "step": 36045 }, { "epoch": 2.8416239653133624, "grad_norm": 0.4515739679336548, "learning_rate": 0.0005, "loss": 1.2371983528137207, "step": 36046 }, { "epoch": 2.841702798581001, "grad_norm": 0.5297159552574158, "learning_rate": 0.0005, "loss": 1.2256889343261719, "step": 36047 }, { "epoch": 2.84178163184864, "grad_norm": 0.5257198214530945, "learning_rate": 0.0005, "loss": 1.5164241790771484, "step": 36048 }, { "epoch": 2.8418604651162793, "grad_norm": 0.5199211239814758, "learning_rate": 0.0005, "loss": 1.2717795372009277, "step": 36049 }, { "epoch": 2.841939298383918, "grad_norm": 0.45458778738975525, "learning_rate": 0.0005, "loss": 1.3800344467163086, "step": 36050 }, { "epoch": 2.8420181316515567, "grad_norm": 0.549628734588623, "learning_rate": 0.0005, "loss": 1.2514090538024902, "step": 36051 }, { "epoch": 2.842096964919196, "grad_norm": 0.516248345375061, "learning_rate": 0.0005, "loss": 1.2817418575286865, "step": 36052 }, { "epoch": 2.842175798186835, "grad_norm": 0.4594642221927643, "learning_rate": 0.0005, "loss": 1.1309700012207031, "step": 36053 }, { "epoch": 2.8422546314544737, "grad_norm": 0.4142065644264221, "learning_rate": 0.0005, "loss": 1.1762769222259521, "step": 36054 }, { "epoch": 2.842333464722113, "grad_norm": 0.4839516282081604, "learning_rate": 0.0005, "loss": 1.2501757144927979, "step": 36055 }, { "epoch": 2.8424122979897515, "grad_norm": 0.4394376575946808, "learning_rate": 0.0005, "loss": 1.1517667770385742, "step": 36056 }, { "epoch": 2.8424911312573906, "grad_norm": 0.43389764428138733, "learning_rate": 0.0005, "loss": 1.282761812210083, "step": 36057 }, { "epoch": 2.8425699645250297, "grad_norm": 0.4937703609466553, "learning_rate": 0.0005, "loss": 1.3853464126586914, "step": 36058 }, { "epoch": 2.8426487977926684, "grad_norm": 0.4125724136829376, "learning_rate": 0.0005, "loss": 1.1537706851959229, "step": 36059 }, { "epoch": 2.8427276310603076, "grad_norm": 0.41794735193252563, "learning_rate": 0.0005, "loss": 1.2189048528671265, "step": 36060 }, { "epoch": 2.8428064643279463, "grad_norm": 0.5422064661979675, "learning_rate": 0.0005, "loss": 1.358877182006836, "step": 36061 }, { "epoch": 2.8428852975955854, "grad_norm": 0.7092297077178955, "learning_rate": 0.0005, "loss": 1.3810343742370605, "step": 36062 }, { "epoch": 2.8429641308632245, "grad_norm": 0.3901391327381134, "learning_rate": 0.0005, "loss": 1.1756373643875122, "step": 36063 }, { "epoch": 2.843042964130863, "grad_norm": 0.41576087474823, "learning_rate": 0.0005, "loss": 1.2439630031585693, "step": 36064 }, { "epoch": 2.843121797398502, "grad_norm": 0.5445731282234192, "learning_rate": 0.0005, "loss": 1.3805572986602783, "step": 36065 }, { "epoch": 2.843200630666141, "grad_norm": 0.6342938542366028, "learning_rate": 0.0005, "loss": 1.289658784866333, "step": 36066 }, { "epoch": 2.84327946393378, "grad_norm": 0.5535919070243835, "learning_rate": 0.0005, "loss": 1.4740933179855347, "step": 36067 }, { "epoch": 2.843358297201419, "grad_norm": 0.6771757006645203, "learning_rate": 0.0005, "loss": 1.330573320388794, "step": 36068 }, { "epoch": 2.843437130469058, "grad_norm": 0.5209189653396606, "learning_rate": 0.0005, "loss": 1.2795900106430054, "step": 36069 }, { "epoch": 2.8435159637366967, "grad_norm": 0.6048352122306824, "learning_rate": 0.0005, "loss": 1.4088165760040283, "step": 36070 }, { "epoch": 2.843594797004336, "grad_norm": 0.4863072633743286, "learning_rate": 0.0005, "loss": 1.1312134265899658, "step": 36071 }, { "epoch": 2.843673630271975, "grad_norm": 0.5734896063804626, "learning_rate": 0.0005, "loss": 1.513992190361023, "step": 36072 }, { "epoch": 2.8437524635396136, "grad_norm": 0.36964982748031616, "learning_rate": 0.0005, "loss": 1.2108434438705444, "step": 36073 }, { "epoch": 2.8438312968072528, "grad_norm": 0.5399786829948425, "learning_rate": 0.0005, "loss": 1.4474499225616455, "step": 36074 }, { "epoch": 2.8439101300748915, "grad_norm": 0.4025704264640808, "learning_rate": 0.0005, "loss": 1.2587668895721436, "step": 36075 }, { "epoch": 2.8439889633425306, "grad_norm": 0.6037622094154358, "learning_rate": 0.0005, "loss": 1.6105691194534302, "step": 36076 }, { "epoch": 2.8440677966101697, "grad_norm": 0.4354184865951538, "learning_rate": 0.0005, "loss": 1.2992877960205078, "step": 36077 }, { "epoch": 2.8441466298778084, "grad_norm": 0.42567846179008484, "learning_rate": 0.0005, "loss": 1.1286163330078125, "step": 36078 }, { "epoch": 2.844225463145447, "grad_norm": 0.4631412923336029, "learning_rate": 0.0005, "loss": 1.3444807529449463, "step": 36079 }, { "epoch": 2.8443042964130862, "grad_norm": 0.5664646625518799, "learning_rate": 0.0005, "loss": 1.4606574773788452, "step": 36080 }, { "epoch": 2.8443831296807254, "grad_norm": 0.4680943489074707, "learning_rate": 0.0005, "loss": 1.3208270072937012, "step": 36081 }, { "epoch": 2.844461962948364, "grad_norm": 0.3721962571144104, "learning_rate": 0.0005, "loss": 1.0750582218170166, "step": 36082 }, { "epoch": 2.844540796216003, "grad_norm": 0.6801832914352417, "learning_rate": 0.0005, "loss": 1.347581148147583, "step": 36083 }, { "epoch": 2.844619629483642, "grad_norm": 0.4950854480266571, "learning_rate": 0.0005, "loss": 1.4634414911270142, "step": 36084 }, { "epoch": 2.844698462751281, "grad_norm": 0.6051675081253052, "learning_rate": 0.0005, "loss": 1.426392674446106, "step": 36085 }, { "epoch": 2.84477729601892, "grad_norm": 0.5403684377670288, "learning_rate": 0.0005, "loss": 1.487285852432251, "step": 36086 }, { "epoch": 2.844856129286559, "grad_norm": 0.502718448638916, "learning_rate": 0.0005, "loss": 1.4534716606140137, "step": 36087 }, { "epoch": 2.844934962554198, "grad_norm": 0.40347135066986084, "learning_rate": 0.0005, "loss": 1.1167068481445312, "step": 36088 }, { "epoch": 2.8450137958218367, "grad_norm": 0.5136009454727173, "learning_rate": 0.0005, "loss": 1.4416444301605225, "step": 36089 }, { "epoch": 2.845092629089476, "grad_norm": 0.5572847127914429, "learning_rate": 0.0005, "loss": 1.3245806694030762, "step": 36090 }, { "epoch": 2.845171462357115, "grad_norm": 0.3753911256790161, "learning_rate": 0.0005, "loss": 1.1328692436218262, "step": 36091 }, { "epoch": 2.8452502956247536, "grad_norm": 0.5821122527122498, "learning_rate": 0.0005, "loss": 1.3072704076766968, "step": 36092 }, { "epoch": 2.8453291288923928, "grad_norm": 0.4680050015449524, "learning_rate": 0.0005, "loss": 1.2511975765228271, "step": 36093 }, { "epoch": 2.8454079621600314, "grad_norm": 0.4151647686958313, "learning_rate": 0.0005, "loss": 1.111930251121521, "step": 36094 }, { "epoch": 2.8454867954276706, "grad_norm": 0.6952399015426636, "learning_rate": 0.0005, "loss": 1.430039882659912, "step": 36095 }, { "epoch": 2.8455656286953093, "grad_norm": 0.5586088299751282, "learning_rate": 0.0005, "loss": 1.297037124633789, "step": 36096 }, { "epoch": 2.8456444619629484, "grad_norm": 0.5317789316177368, "learning_rate": 0.0005, "loss": 1.496872901916504, "step": 36097 }, { "epoch": 2.845723295230587, "grad_norm": 0.5022475123405457, "learning_rate": 0.0005, "loss": 1.4402860403060913, "step": 36098 }, { "epoch": 2.8458021284982262, "grad_norm": 0.5127690434455872, "learning_rate": 0.0005, "loss": 1.3357596397399902, "step": 36099 }, { "epoch": 2.8458809617658654, "grad_norm": 0.4985940456390381, "learning_rate": 0.0005, "loss": 1.4441707134246826, "step": 36100 }, { "epoch": 2.845959795033504, "grad_norm": 0.47849780321121216, "learning_rate": 0.0005, "loss": 1.285894513130188, "step": 36101 }, { "epoch": 2.846038628301143, "grad_norm": 0.5001161098480225, "learning_rate": 0.0005, "loss": 1.3330683708190918, "step": 36102 }, { "epoch": 2.846117461568782, "grad_norm": 0.4045601189136505, "learning_rate": 0.0005, "loss": 1.0880837440490723, "step": 36103 }, { "epoch": 2.846196294836421, "grad_norm": 0.4884262979030609, "learning_rate": 0.0005, "loss": 1.3110666275024414, "step": 36104 }, { "epoch": 2.84627512810406, "grad_norm": 0.6488404273986816, "learning_rate": 0.0005, "loss": 1.5615284442901611, "step": 36105 }, { "epoch": 2.846353961371699, "grad_norm": 0.48656633496284485, "learning_rate": 0.0005, "loss": 1.3999085426330566, "step": 36106 }, { "epoch": 2.846432794639338, "grad_norm": 0.4443126320838928, "learning_rate": 0.0005, "loss": 1.2515733242034912, "step": 36107 }, { "epoch": 2.8465116279069766, "grad_norm": 0.6225273013114929, "learning_rate": 0.0005, "loss": 1.3339424133300781, "step": 36108 }, { "epoch": 2.846590461174616, "grad_norm": 0.5549671053886414, "learning_rate": 0.0005, "loss": 1.2030205726623535, "step": 36109 }, { "epoch": 2.8466692944422545, "grad_norm": 0.44832897186279297, "learning_rate": 0.0005, "loss": 1.4049439430236816, "step": 36110 }, { "epoch": 2.8467481277098936, "grad_norm": 0.4877069294452667, "learning_rate": 0.0005, "loss": 1.21645987033844, "step": 36111 }, { "epoch": 2.8468269609775323, "grad_norm": 0.451670378446579, "learning_rate": 0.0005, "loss": 1.5758416652679443, "step": 36112 }, { "epoch": 2.8469057942451714, "grad_norm": 0.462530255317688, "learning_rate": 0.0005, "loss": 1.4111518859863281, "step": 36113 }, { "epoch": 2.8469846275128106, "grad_norm": 0.4691285490989685, "learning_rate": 0.0005, "loss": 1.1891672611236572, "step": 36114 }, { "epoch": 2.8470634607804493, "grad_norm": 0.5414841771125793, "learning_rate": 0.0005, "loss": 1.466681957244873, "step": 36115 }, { "epoch": 2.8471422940480884, "grad_norm": 0.39255988597869873, "learning_rate": 0.0005, "loss": 1.054567813873291, "step": 36116 }, { "epoch": 2.847221127315727, "grad_norm": 0.4342471957206726, "learning_rate": 0.0005, "loss": 1.131041407585144, "step": 36117 }, { "epoch": 2.847299960583366, "grad_norm": 0.5618985295295715, "learning_rate": 0.0005, "loss": 1.6791181564331055, "step": 36118 }, { "epoch": 2.8473787938510053, "grad_norm": 0.5351208448410034, "learning_rate": 0.0005, "loss": 1.2846126556396484, "step": 36119 }, { "epoch": 2.847457627118644, "grad_norm": 0.507532000541687, "learning_rate": 0.0005, "loss": 1.5328723192214966, "step": 36120 }, { "epoch": 2.847536460386283, "grad_norm": 0.44133949279785156, "learning_rate": 0.0005, "loss": 1.3194749355316162, "step": 36121 }, { "epoch": 2.847615293653922, "grad_norm": 0.439343124628067, "learning_rate": 0.0005, "loss": 1.2455723285675049, "step": 36122 }, { "epoch": 2.847694126921561, "grad_norm": 0.41284942626953125, "learning_rate": 0.0005, "loss": 1.1299550533294678, "step": 36123 }, { "epoch": 2.8477729601892, "grad_norm": 0.5812575221061707, "learning_rate": 0.0005, "loss": 1.2824856042861938, "step": 36124 }, { "epoch": 2.847851793456839, "grad_norm": 0.444710910320282, "learning_rate": 0.0005, "loss": 1.3094286918640137, "step": 36125 }, { "epoch": 2.8479306267244775, "grad_norm": 0.43835002183914185, "learning_rate": 0.0005, "loss": 1.2314882278442383, "step": 36126 }, { "epoch": 2.8480094599921166, "grad_norm": 0.5013070702552795, "learning_rate": 0.0005, "loss": 1.3038249015808105, "step": 36127 }, { "epoch": 2.8480882932597558, "grad_norm": 0.44194087386131287, "learning_rate": 0.0005, "loss": 1.2319955825805664, "step": 36128 }, { "epoch": 2.8481671265273945, "grad_norm": 0.42463499307632446, "learning_rate": 0.0005, "loss": 1.1587872505187988, "step": 36129 }, { "epoch": 2.8482459597950336, "grad_norm": 0.49974700808525085, "learning_rate": 0.0005, "loss": 1.3816944360733032, "step": 36130 }, { "epoch": 2.8483247930626723, "grad_norm": 0.4293040931224823, "learning_rate": 0.0005, "loss": 1.2343904972076416, "step": 36131 }, { "epoch": 2.8484036263303114, "grad_norm": 0.3969058394432068, "learning_rate": 0.0005, "loss": 1.1700984239578247, "step": 36132 }, { "epoch": 2.8484824595979505, "grad_norm": 0.5311360955238342, "learning_rate": 0.0005, "loss": 1.2372256517410278, "step": 36133 }, { "epoch": 2.8485612928655892, "grad_norm": 0.43749967217445374, "learning_rate": 0.0005, "loss": 1.341932773590088, "step": 36134 }, { "epoch": 2.8486401261332284, "grad_norm": 0.47193628549575806, "learning_rate": 0.0005, "loss": 1.3010661602020264, "step": 36135 }, { "epoch": 2.848718959400867, "grad_norm": 0.5756685733795166, "learning_rate": 0.0005, "loss": 1.4603582620620728, "step": 36136 }, { "epoch": 2.848797792668506, "grad_norm": 0.4106435775756836, "learning_rate": 0.0005, "loss": 1.2354137897491455, "step": 36137 }, { "epoch": 2.8488766259361453, "grad_norm": 0.4968852698802948, "learning_rate": 0.0005, "loss": 1.3313342332839966, "step": 36138 }, { "epoch": 2.848955459203784, "grad_norm": 0.46589145064353943, "learning_rate": 0.0005, "loss": 1.4820997714996338, "step": 36139 }, { "epoch": 2.8490342924714227, "grad_norm": 0.7258062362670898, "learning_rate": 0.0005, "loss": 1.5482206344604492, "step": 36140 }, { "epoch": 2.849113125739062, "grad_norm": 0.6084006428718567, "learning_rate": 0.0005, "loss": 1.3966058492660522, "step": 36141 }, { "epoch": 2.849191959006701, "grad_norm": 0.5413083434104919, "learning_rate": 0.0005, "loss": 1.344724178314209, "step": 36142 }, { "epoch": 2.8492707922743397, "grad_norm": 0.42722952365875244, "learning_rate": 0.0005, "loss": 1.2694180011749268, "step": 36143 }, { "epoch": 2.849349625541979, "grad_norm": 0.555461585521698, "learning_rate": 0.0005, "loss": 1.4876775741577148, "step": 36144 }, { "epoch": 2.8494284588096175, "grad_norm": 0.5791542530059814, "learning_rate": 0.0005, "loss": 1.2752740383148193, "step": 36145 }, { "epoch": 2.8495072920772566, "grad_norm": 0.4676072299480438, "learning_rate": 0.0005, "loss": 1.3786464929580688, "step": 36146 }, { "epoch": 2.8495861253448957, "grad_norm": 0.4503166079521179, "learning_rate": 0.0005, "loss": 1.2888402938842773, "step": 36147 }, { "epoch": 2.8496649586125344, "grad_norm": 0.6187726259231567, "learning_rate": 0.0005, "loss": 1.5285347700119019, "step": 36148 }, { "epoch": 2.8497437918801736, "grad_norm": 0.5012899041175842, "learning_rate": 0.0005, "loss": 1.3270542621612549, "step": 36149 }, { "epoch": 2.8498226251478123, "grad_norm": 0.7077423930168152, "learning_rate": 0.0005, "loss": 1.2404284477233887, "step": 36150 }, { "epoch": 2.8499014584154514, "grad_norm": 0.5546442866325378, "learning_rate": 0.0005, "loss": 1.5987889766693115, "step": 36151 }, { "epoch": 2.8499802916830905, "grad_norm": 0.4782775938510895, "learning_rate": 0.0005, "loss": 1.0485260486602783, "step": 36152 }, { "epoch": 2.850059124950729, "grad_norm": 0.5464553833007812, "learning_rate": 0.0005, "loss": 1.414482831954956, "step": 36153 }, { "epoch": 2.850137958218368, "grad_norm": 0.46985408663749695, "learning_rate": 0.0005, "loss": 1.3049736022949219, "step": 36154 }, { "epoch": 2.850216791486007, "grad_norm": 0.41478797793388367, "learning_rate": 0.0005, "loss": 1.2550075054168701, "step": 36155 }, { "epoch": 2.850295624753646, "grad_norm": 0.5772413015365601, "learning_rate": 0.0005, "loss": 1.216970682144165, "step": 36156 }, { "epoch": 2.850374458021285, "grad_norm": 0.49150311946868896, "learning_rate": 0.0005, "loss": 1.2388850450515747, "step": 36157 }, { "epoch": 2.850453291288924, "grad_norm": 0.4437761604785919, "learning_rate": 0.0005, "loss": 1.2625441551208496, "step": 36158 }, { "epoch": 2.8505321245565627, "grad_norm": 0.6438060998916626, "learning_rate": 0.0005, "loss": 1.3988569974899292, "step": 36159 }, { "epoch": 2.850610957824202, "grad_norm": 0.5477882623672485, "learning_rate": 0.0005, "loss": 1.3509716987609863, "step": 36160 }, { "epoch": 2.850689791091841, "grad_norm": 0.44537582993507385, "learning_rate": 0.0005, "loss": 1.2312579154968262, "step": 36161 }, { "epoch": 2.8507686243594796, "grad_norm": 0.434452086687088, "learning_rate": 0.0005, "loss": 1.3986201286315918, "step": 36162 }, { "epoch": 2.8508474576271188, "grad_norm": 0.4408824145793915, "learning_rate": 0.0005, "loss": 1.259204626083374, "step": 36163 }, { "epoch": 2.8509262908947575, "grad_norm": 0.49237504601478577, "learning_rate": 0.0005, "loss": 1.3630857467651367, "step": 36164 }, { "epoch": 2.8510051241623966, "grad_norm": 0.509893000125885, "learning_rate": 0.0005, "loss": 1.3457367420196533, "step": 36165 }, { "epoch": 2.8510839574300357, "grad_norm": 0.6368942260742188, "learning_rate": 0.0005, "loss": 1.2530497312545776, "step": 36166 }, { "epoch": 2.8511627906976744, "grad_norm": 0.6296312808990479, "learning_rate": 0.0005, "loss": 1.3815991878509521, "step": 36167 }, { "epoch": 2.851241623965313, "grad_norm": 0.5185847282409668, "learning_rate": 0.0005, "loss": 1.5641676187515259, "step": 36168 }, { "epoch": 2.8513204572329522, "grad_norm": 0.5655089616775513, "learning_rate": 0.0005, "loss": 1.4797313213348389, "step": 36169 }, { "epoch": 2.8513992905005914, "grad_norm": 0.5444799065589905, "learning_rate": 0.0005, "loss": 1.4917868375778198, "step": 36170 }, { "epoch": 2.85147812376823, "grad_norm": 0.8011316657066345, "learning_rate": 0.0005, "loss": 1.322770595550537, "step": 36171 }, { "epoch": 2.851556957035869, "grad_norm": 0.4243159592151642, "learning_rate": 0.0005, "loss": 1.1357989311218262, "step": 36172 }, { "epoch": 2.851635790303508, "grad_norm": 0.5704067945480347, "learning_rate": 0.0005, "loss": 1.4482405185699463, "step": 36173 }, { "epoch": 2.851714623571147, "grad_norm": 0.4394647479057312, "learning_rate": 0.0005, "loss": 1.2000951766967773, "step": 36174 }, { "epoch": 2.851793456838786, "grad_norm": 0.4740605056285858, "learning_rate": 0.0005, "loss": 1.5407195091247559, "step": 36175 }, { "epoch": 2.851872290106425, "grad_norm": 0.4600943624973297, "learning_rate": 0.0005, "loss": 1.2443548440933228, "step": 36176 }, { "epoch": 2.851951123374064, "grad_norm": 0.46236541867256165, "learning_rate": 0.0005, "loss": 1.3336998224258423, "step": 36177 }, { "epoch": 2.8520299566417027, "grad_norm": 0.4487667381763458, "learning_rate": 0.0005, "loss": 1.3458900451660156, "step": 36178 }, { "epoch": 2.852108789909342, "grad_norm": 0.4629138112068176, "learning_rate": 0.0005, "loss": 1.375462293624878, "step": 36179 }, { "epoch": 2.852187623176981, "grad_norm": 0.42161932587623596, "learning_rate": 0.0005, "loss": 1.2063519954681396, "step": 36180 }, { "epoch": 2.8522664564446196, "grad_norm": 0.4375789761543274, "learning_rate": 0.0005, "loss": 1.1973811388015747, "step": 36181 }, { "epoch": 2.8523452897122583, "grad_norm": 0.718357503414154, "learning_rate": 0.0005, "loss": 1.240753412246704, "step": 36182 }, { "epoch": 2.8524241229798974, "grad_norm": 0.44494685530662537, "learning_rate": 0.0005, "loss": 1.1547430753707886, "step": 36183 }, { "epoch": 2.8525029562475366, "grad_norm": 0.4216667413711548, "learning_rate": 0.0005, "loss": 1.202597975730896, "step": 36184 }, { "epoch": 2.8525817895151753, "grad_norm": 0.4512348771095276, "learning_rate": 0.0005, "loss": 1.2970354557037354, "step": 36185 }, { "epoch": 2.8526606227828144, "grad_norm": 0.4916309714317322, "learning_rate": 0.0005, "loss": 1.1122335195541382, "step": 36186 }, { "epoch": 2.852739456050453, "grad_norm": 0.4760510325431824, "learning_rate": 0.0005, "loss": 1.2498760223388672, "step": 36187 }, { "epoch": 2.8528182893180922, "grad_norm": 0.5266228318214417, "learning_rate": 0.0005, "loss": 1.3817977905273438, "step": 36188 }, { "epoch": 2.8528971225857314, "grad_norm": 0.4980674088001251, "learning_rate": 0.0005, "loss": 1.359699010848999, "step": 36189 }, { "epoch": 2.85297595585337, "grad_norm": 0.45476478338241577, "learning_rate": 0.0005, "loss": 1.1685020923614502, "step": 36190 }, { "epoch": 2.853054789121009, "grad_norm": 0.6213122010231018, "learning_rate": 0.0005, "loss": 1.1394802331924438, "step": 36191 }, { "epoch": 2.853133622388648, "grad_norm": 0.5767745971679688, "learning_rate": 0.0005, "loss": 1.4459561109542847, "step": 36192 }, { "epoch": 2.853212455656287, "grad_norm": 0.46516889333724976, "learning_rate": 0.0005, "loss": 1.3109521865844727, "step": 36193 }, { "epoch": 2.853291288923926, "grad_norm": 0.47598257660865784, "learning_rate": 0.0005, "loss": 1.3519320487976074, "step": 36194 }, { "epoch": 2.853370122191565, "grad_norm": 0.4601023197174072, "learning_rate": 0.0005, "loss": 1.4389913082122803, "step": 36195 }, { "epoch": 2.8534489554592035, "grad_norm": 0.5044496655464172, "learning_rate": 0.0005, "loss": 1.3995413780212402, "step": 36196 }, { "epoch": 2.8535277887268427, "grad_norm": 1.5120534896850586, "learning_rate": 0.0005, "loss": 1.3014461994171143, "step": 36197 }, { "epoch": 2.853606621994482, "grad_norm": 0.48469361662864685, "learning_rate": 0.0005, "loss": 1.3454272747039795, "step": 36198 }, { "epoch": 2.8536854552621205, "grad_norm": 0.45363250374794006, "learning_rate": 0.0005, "loss": 1.2264766693115234, "step": 36199 }, { "epoch": 2.8537642885297596, "grad_norm": 0.5985310673713684, "learning_rate": 0.0005, "loss": 1.307572364807129, "step": 36200 }, { "epoch": 2.8538431217973983, "grad_norm": 0.6318244338035583, "learning_rate": 0.0005, "loss": 1.1988089084625244, "step": 36201 }, { "epoch": 2.8539219550650374, "grad_norm": 0.5018049478530884, "learning_rate": 0.0005, "loss": 1.267516851425171, "step": 36202 }, { "epoch": 2.8540007883326766, "grad_norm": 0.7705140113830566, "learning_rate": 0.0005, "loss": 1.3410519361495972, "step": 36203 }, { "epoch": 2.8540796216003153, "grad_norm": 0.4933417737483978, "learning_rate": 0.0005, "loss": 1.1830461025238037, "step": 36204 }, { "epoch": 2.8541584548679544, "grad_norm": 0.5210332870483398, "learning_rate": 0.0005, "loss": 1.395038366317749, "step": 36205 }, { "epoch": 2.854237288135593, "grad_norm": 0.8327938318252563, "learning_rate": 0.0005, "loss": 1.3605232238769531, "step": 36206 }, { "epoch": 2.854316121403232, "grad_norm": 0.4363642632961273, "learning_rate": 0.0005, "loss": 1.2077995538711548, "step": 36207 }, { "epoch": 2.8543949546708713, "grad_norm": 0.5616475343704224, "learning_rate": 0.0005, "loss": 1.3844292163848877, "step": 36208 }, { "epoch": 2.85447378793851, "grad_norm": 0.47028958797454834, "learning_rate": 0.0005, "loss": 1.501277208328247, "step": 36209 }, { "epoch": 2.8545526212061487, "grad_norm": 0.737722635269165, "learning_rate": 0.0005, "loss": 1.3556156158447266, "step": 36210 }, { "epoch": 2.854631454473788, "grad_norm": 0.5317836403846741, "learning_rate": 0.0005, "loss": 1.316382884979248, "step": 36211 }, { "epoch": 2.854710287741427, "grad_norm": 0.48615458607673645, "learning_rate": 0.0005, "loss": 1.3259336948394775, "step": 36212 }, { "epoch": 2.8547891210090657, "grad_norm": 0.658552885055542, "learning_rate": 0.0005, "loss": 1.5371994972229004, "step": 36213 }, { "epoch": 2.854867954276705, "grad_norm": 0.5073145031929016, "learning_rate": 0.0005, "loss": 1.3450504541397095, "step": 36214 }, { "epoch": 2.8549467875443435, "grad_norm": 0.47291457653045654, "learning_rate": 0.0005, "loss": 1.0742855072021484, "step": 36215 }, { "epoch": 2.8550256208119826, "grad_norm": 0.42226606607437134, "learning_rate": 0.0005, "loss": 1.189579725265503, "step": 36216 }, { "epoch": 2.8551044540796218, "grad_norm": 0.44759562611579895, "learning_rate": 0.0005, "loss": 1.5015254020690918, "step": 36217 }, { "epoch": 2.8551832873472605, "grad_norm": 0.7705485224723816, "learning_rate": 0.0005, "loss": 1.3180575370788574, "step": 36218 }, { "epoch": 2.8552621206148996, "grad_norm": 0.4586394429206848, "learning_rate": 0.0005, "loss": 1.2692121267318726, "step": 36219 }, { "epoch": 2.8553409538825383, "grad_norm": 0.6622408628463745, "learning_rate": 0.0005, "loss": 1.4564440250396729, "step": 36220 }, { "epoch": 2.8554197871501774, "grad_norm": 0.4073755443096161, "learning_rate": 0.0005, "loss": 1.1713621616363525, "step": 36221 }, { "epoch": 2.8554986204178165, "grad_norm": 0.6420058012008667, "learning_rate": 0.0005, "loss": 1.3761483430862427, "step": 36222 }, { "epoch": 2.8555774536854552, "grad_norm": 1.0842626094818115, "learning_rate": 0.0005, "loss": 1.466780424118042, "step": 36223 }, { "epoch": 2.8556562869530944, "grad_norm": 0.5011598467826843, "learning_rate": 0.0005, "loss": 1.372583270072937, "step": 36224 }, { "epoch": 2.855735120220733, "grad_norm": 0.6222467422485352, "learning_rate": 0.0005, "loss": 1.2939376831054688, "step": 36225 }, { "epoch": 2.855813953488372, "grad_norm": 0.3925701677799225, "learning_rate": 0.0005, "loss": 1.1175754070281982, "step": 36226 }, { "epoch": 2.855892786756011, "grad_norm": 0.4183143675327301, "learning_rate": 0.0005, "loss": 1.1569197177886963, "step": 36227 }, { "epoch": 2.85597162002365, "grad_norm": 0.5510343909263611, "learning_rate": 0.0005, "loss": 1.3451054096221924, "step": 36228 }, { "epoch": 2.8560504532912887, "grad_norm": 0.5235258340835571, "learning_rate": 0.0005, "loss": 1.14293372631073, "step": 36229 }, { "epoch": 2.856129286558928, "grad_norm": 0.5787665247917175, "learning_rate": 0.0005, "loss": 1.3426395654678345, "step": 36230 }, { "epoch": 2.856208119826567, "grad_norm": 0.37441912293434143, "learning_rate": 0.0005, "loss": 1.1303658485412598, "step": 36231 }, { "epoch": 2.8562869530942057, "grad_norm": 0.48039543628692627, "learning_rate": 0.0005, "loss": 1.2894313335418701, "step": 36232 }, { "epoch": 2.856365786361845, "grad_norm": 0.5084933638572693, "learning_rate": 0.0005, "loss": 1.3697614669799805, "step": 36233 }, { "epoch": 2.8564446196294835, "grad_norm": 0.4773463308811188, "learning_rate": 0.0005, "loss": 1.4209022521972656, "step": 36234 }, { "epoch": 2.8565234528971226, "grad_norm": 0.4005391001701355, "learning_rate": 0.0005, "loss": 1.3400487899780273, "step": 36235 }, { "epoch": 2.8566022861647618, "grad_norm": 0.4833262264728546, "learning_rate": 0.0005, "loss": 1.1760646104812622, "step": 36236 }, { "epoch": 2.8566811194324004, "grad_norm": 0.38210973143577576, "learning_rate": 0.0005, "loss": 1.0769178867340088, "step": 36237 }, { "epoch": 2.8567599527000396, "grad_norm": 0.6363012194633484, "learning_rate": 0.0005, "loss": 1.5940357446670532, "step": 36238 }, { "epoch": 2.8568387859676783, "grad_norm": 0.4396240711212158, "learning_rate": 0.0005, "loss": 1.3212617635726929, "step": 36239 }, { "epoch": 2.8569176192353174, "grad_norm": 0.4479005038738251, "learning_rate": 0.0005, "loss": 1.278071403503418, "step": 36240 }, { "epoch": 2.856996452502956, "grad_norm": 0.45367321372032166, "learning_rate": 0.0005, "loss": 1.2084606885910034, "step": 36241 }, { "epoch": 2.857075285770595, "grad_norm": 0.42109745740890503, "learning_rate": 0.0005, "loss": 1.46474289894104, "step": 36242 }, { "epoch": 2.857154119038234, "grad_norm": 0.5824345350265503, "learning_rate": 0.0005, "loss": 1.451143741607666, "step": 36243 }, { "epoch": 2.857232952305873, "grad_norm": 0.5899002552032471, "learning_rate": 0.0005, "loss": 1.600670337677002, "step": 36244 }, { "epoch": 2.857311785573512, "grad_norm": 0.5822539925575256, "learning_rate": 0.0005, "loss": 1.4226250648498535, "step": 36245 }, { "epoch": 2.857390618841151, "grad_norm": 0.5208282470703125, "learning_rate": 0.0005, "loss": 1.304181456565857, "step": 36246 }, { "epoch": 2.85746945210879, "grad_norm": 0.5145001411437988, "learning_rate": 0.0005, "loss": 1.4262967109680176, "step": 36247 }, { "epoch": 2.8575482853764287, "grad_norm": 0.4859194755554199, "learning_rate": 0.0005, "loss": 1.1868586540222168, "step": 36248 }, { "epoch": 2.857627118644068, "grad_norm": 0.44722798466682434, "learning_rate": 0.0005, "loss": 1.176195502281189, "step": 36249 }, { "epoch": 2.857705951911707, "grad_norm": 0.6091744899749756, "learning_rate": 0.0005, "loss": 1.2208938598632812, "step": 36250 }, { "epoch": 2.8577847851793456, "grad_norm": 0.3971964120864868, "learning_rate": 0.0005, "loss": 1.2037123441696167, "step": 36251 }, { "epoch": 2.8578636184469848, "grad_norm": 0.47066882252693176, "learning_rate": 0.0005, "loss": 1.376763939857483, "step": 36252 }, { "epoch": 2.8579424517146235, "grad_norm": 0.5862494111061096, "learning_rate": 0.0005, "loss": 1.422849178314209, "step": 36253 }, { "epoch": 2.8580212849822626, "grad_norm": 0.5791057348251343, "learning_rate": 0.0005, "loss": 1.1999828815460205, "step": 36254 }, { "epoch": 2.8581001182499017, "grad_norm": 0.6096162796020508, "learning_rate": 0.0005, "loss": 1.3265199661254883, "step": 36255 }, { "epoch": 2.8581789515175404, "grad_norm": 0.5169605016708374, "learning_rate": 0.0005, "loss": 1.3984308242797852, "step": 36256 }, { "epoch": 2.858257784785179, "grad_norm": 0.5011579990386963, "learning_rate": 0.0005, "loss": 1.4581773281097412, "step": 36257 }, { "epoch": 2.8583366180528182, "grad_norm": 0.42029041051864624, "learning_rate": 0.0005, "loss": 1.074968934059143, "step": 36258 }, { "epoch": 2.8584154513204574, "grad_norm": 0.4499454200267792, "learning_rate": 0.0005, "loss": 1.3265352249145508, "step": 36259 }, { "epoch": 2.858494284588096, "grad_norm": 0.468204140663147, "learning_rate": 0.0005, "loss": 1.3999370336532593, "step": 36260 }, { "epoch": 2.858573117855735, "grad_norm": 0.5029659867286682, "learning_rate": 0.0005, "loss": 1.246977686882019, "step": 36261 }, { "epoch": 2.858651951123374, "grad_norm": 0.4372154772281647, "learning_rate": 0.0005, "loss": 1.1699186563491821, "step": 36262 }, { "epoch": 2.858730784391013, "grad_norm": 0.7246295809745789, "learning_rate": 0.0005, "loss": 1.4266579151153564, "step": 36263 }, { "epoch": 2.858809617658652, "grad_norm": 0.6458813548088074, "learning_rate": 0.0005, "loss": 1.1501774787902832, "step": 36264 }, { "epoch": 2.858888450926291, "grad_norm": 0.41274940967559814, "learning_rate": 0.0005, "loss": 1.243507742881775, "step": 36265 }, { "epoch": 2.85896728419393, "grad_norm": 0.5515924096107483, "learning_rate": 0.0005, "loss": 1.3810675144195557, "step": 36266 }, { "epoch": 2.8590461174615687, "grad_norm": 0.4419196844100952, "learning_rate": 0.0005, "loss": 1.2436639070510864, "step": 36267 }, { "epoch": 2.859124950729208, "grad_norm": 0.44155648350715637, "learning_rate": 0.0005, "loss": 1.170969843864441, "step": 36268 }, { "epoch": 2.859203783996847, "grad_norm": 0.5617873072624207, "learning_rate": 0.0005, "loss": 1.3375928401947021, "step": 36269 }, { "epoch": 2.8592826172644856, "grad_norm": 0.5820003747940063, "learning_rate": 0.0005, "loss": 1.2613961696624756, "step": 36270 }, { "epoch": 2.8593614505321243, "grad_norm": 0.5062298774719238, "learning_rate": 0.0005, "loss": 1.5002892017364502, "step": 36271 }, { "epoch": 2.8594402837997634, "grad_norm": 0.43785643577575684, "learning_rate": 0.0005, "loss": 1.193045973777771, "step": 36272 }, { "epoch": 2.8595191170674026, "grad_norm": 0.375458687543869, "learning_rate": 0.0005, "loss": 1.21464204788208, "step": 36273 }, { "epoch": 2.8595979503350413, "grad_norm": 0.6181128621101379, "learning_rate": 0.0005, "loss": 1.5633035898208618, "step": 36274 }, { "epoch": 2.8596767836026804, "grad_norm": 0.516606330871582, "learning_rate": 0.0005, "loss": 1.3235840797424316, "step": 36275 }, { "epoch": 2.859755616870319, "grad_norm": 0.5843068957328796, "learning_rate": 0.0005, "loss": 1.403546690940857, "step": 36276 }, { "epoch": 2.8598344501379582, "grad_norm": 0.47811999917030334, "learning_rate": 0.0005, "loss": 1.2975153923034668, "step": 36277 }, { "epoch": 2.8599132834055974, "grad_norm": 0.4846169650554657, "learning_rate": 0.0005, "loss": 1.2177799940109253, "step": 36278 }, { "epoch": 2.859992116673236, "grad_norm": 0.4759291112422943, "learning_rate": 0.0005, "loss": 1.2958961725234985, "step": 36279 }, { "epoch": 2.860070949940875, "grad_norm": 0.5521359443664551, "learning_rate": 0.0005, "loss": 1.3512067794799805, "step": 36280 }, { "epoch": 2.860149783208514, "grad_norm": 0.5097845792770386, "learning_rate": 0.0005, "loss": 1.3977116346359253, "step": 36281 }, { "epoch": 2.860228616476153, "grad_norm": 0.4959436058998108, "learning_rate": 0.0005, "loss": 1.2764263153076172, "step": 36282 }, { "epoch": 2.860307449743792, "grad_norm": 0.5542135834693909, "learning_rate": 0.0005, "loss": 1.4166929721832275, "step": 36283 }, { "epoch": 2.860386283011431, "grad_norm": 0.5593481659889221, "learning_rate": 0.0005, "loss": 1.3822704553604126, "step": 36284 }, { "epoch": 2.8604651162790695, "grad_norm": 0.5423532128334045, "learning_rate": 0.0005, "loss": 1.297888994216919, "step": 36285 }, { "epoch": 2.8605439495467087, "grad_norm": 0.5135094523429871, "learning_rate": 0.0005, "loss": 1.1709635257720947, "step": 36286 }, { "epoch": 2.860622782814348, "grad_norm": 0.4416372776031494, "learning_rate": 0.0005, "loss": 1.2182998657226562, "step": 36287 }, { "epoch": 2.8607016160819865, "grad_norm": 0.4323945641517639, "learning_rate": 0.0005, "loss": 1.2802568674087524, "step": 36288 }, { "epoch": 2.8607804493496256, "grad_norm": 0.5938848853111267, "learning_rate": 0.0005, "loss": 1.3710020780563354, "step": 36289 }, { "epoch": 2.8608592826172643, "grad_norm": 0.35989993810653687, "learning_rate": 0.0005, "loss": 1.0200378894805908, "step": 36290 }, { "epoch": 2.8609381158849034, "grad_norm": 0.5660712718963623, "learning_rate": 0.0005, "loss": 1.4872443675994873, "step": 36291 }, { "epoch": 2.8610169491525426, "grad_norm": 0.4997495412826538, "learning_rate": 0.0005, "loss": 1.336219072341919, "step": 36292 }, { "epoch": 2.8610957824201813, "grad_norm": 0.6037298440933228, "learning_rate": 0.0005, "loss": 1.3967385292053223, "step": 36293 }, { "epoch": 2.8611746156878204, "grad_norm": 0.4796392619609833, "learning_rate": 0.0005, "loss": 1.40553617477417, "step": 36294 }, { "epoch": 2.861253448955459, "grad_norm": 0.4794475734233856, "learning_rate": 0.0005, "loss": 1.2466659545898438, "step": 36295 }, { "epoch": 2.861332282223098, "grad_norm": 0.40425753593444824, "learning_rate": 0.0005, "loss": 1.3285568952560425, "step": 36296 }, { "epoch": 2.8614111154907373, "grad_norm": 0.6300418376922607, "learning_rate": 0.0005, "loss": 1.3849470615386963, "step": 36297 }, { "epoch": 2.861489948758376, "grad_norm": 0.4595282971858978, "learning_rate": 0.0005, "loss": 1.2133395671844482, "step": 36298 }, { "epoch": 2.8615687820260147, "grad_norm": 0.4633307456970215, "learning_rate": 0.0005, "loss": 1.335188865661621, "step": 36299 }, { "epoch": 2.861647615293654, "grad_norm": 0.46054452657699585, "learning_rate": 0.0005, "loss": 1.4706568717956543, "step": 36300 }, { "epoch": 2.861726448561293, "grad_norm": 0.5331417918205261, "learning_rate": 0.0005, "loss": 1.1517491340637207, "step": 36301 }, { "epoch": 2.8618052818289317, "grad_norm": 0.6136207580566406, "learning_rate": 0.0005, "loss": 1.417161464691162, "step": 36302 }, { "epoch": 2.861884115096571, "grad_norm": 0.48545435070991516, "learning_rate": 0.0005, "loss": 1.200089693069458, "step": 36303 }, { "epoch": 2.8619629483642095, "grad_norm": 0.4443819522857666, "learning_rate": 0.0005, "loss": 1.370524525642395, "step": 36304 }, { "epoch": 2.8620417816318486, "grad_norm": 0.47872161865234375, "learning_rate": 0.0005, "loss": 1.1805611848831177, "step": 36305 }, { "epoch": 2.8621206148994878, "grad_norm": 0.4759101867675781, "learning_rate": 0.0005, "loss": 1.3091365098953247, "step": 36306 }, { "epoch": 2.8621994481671265, "grad_norm": 0.5602080821990967, "learning_rate": 0.0005, "loss": 1.4901803731918335, "step": 36307 }, { "epoch": 2.8622782814347656, "grad_norm": 0.6163511872291565, "learning_rate": 0.0005, "loss": 1.4087284803390503, "step": 36308 }, { "epoch": 2.8623571147024043, "grad_norm": 0.5099495649337769, "learning_rate": 0.0005, "loss": 1.1858935356140137, "step": 36309 }, { "epoch": 2.8624359479700434, "grad_norm": 0.42346805334091187, "learning_rate": 0.0005, "loss": 1.3157132863998413, "step": 36310 }, { "epoch": 2.8625147812376825, "grad_norm": 0.6678423881530762, "learning_rate": 0.0005, "loss": 1.3206415176391602, "step": 36311 }, { "epoch": 2.8625936145053212, "grad_norm": 0.5121201276779175, "learning_rate": 0.0005, "loss": 1.3203704357147217, "step": 36312 }, { "epoch": 2.86267244777296, "grad_norm": 0.5623218417167664, "learning_rate": 0.0005, "loss": 1.3192627429962158, "step": 36313 }, { "epoch": 2.862751281040599, "grad_norm": 0.466569721698761, "learning_rate": 0.0005, "loss": 1.2961323261260986, "step": 36314 }, { "epoch": 2.862830114308238, "grad_norm": 0.396480917930603, "learning_rate": 0.0005, "loss": 1.1344743967056274, "step": 36315 }, { "epoch": 2.862908947575877, "grad_norm": 0.5802823901176453, "learning_rate": 0.0005, "loss": 1.331528663635254, "step": 36316 }, { "epoch": 2.862987780843516, "grad_norm": 0.5557237863540649, "learning_rate": 0.0005, "loss": 1.227156162261963, "step": 36317 }, { "epoch": 2.8630666141111547, "grad_norm": 0.6509055495262146, "learning_rate": 0.0005, "loss": 1.4733413457870483, "step": 36318 }, { "epoch": 2.863145447378794, "grad_norm": 0.4182220697402954, "learning_rate": 0.0005, "loss": 1.1853969097137451, "step": 36319 }, { "epoch": 2.863224280646433, "grad_norm": 0.5049907565116882, "learning_rate": 0.0005, "loss": 1.3218586444854736, "step": 36320 }, { "epoch": 2.8633031139140717, "grad_norm": 0.4815155863761902, "learning_rate": 0.0005, "loss": 1.3536393642425537, "step": 36321 }, { "epoch": 2.863381947181711, "grad_norm": 0.6447001695632935, "learning_rate": 0.0005, "loss": 1.3929002285003662, "step": 36322 }, { "epoch": 2.8634607804493495, "grad_norm": 0.49784451723098755, "learning_rate": 0.0005, "loss": 1.2987186908721924, "step": 36323 }, { "epoch": 2.8635396137169886, "grad_norm": 0.618386447429657, "learning_rate": 0.0005, "loss": 1.3341152667999268, "step": 36324 }, { "epoch": 2.8636184469846278, "grad_norm": 0.4221804141998291, "learning_rate": 0.0005, "loss": 1.1637340784072876, "step": 36325 }, { "epoch": 2.8636972802522664, "grad_norm": 0.6182150840759277, "learning_rate": 0.0005, "loss": 1.358397364616394, "step": 36326 }, { "epoch": 2.863776113519905, "grad_norm": 0.4412781894207001, "learning_rate": 0.0005, "loss": 1.1808732748031616, "step": 36327 }, { "epoch": 2.8638549467875443, "grad_norm": 0.4349184036254883, "learning_rate": 0.0005, "loss": 1.2280447483062744, "step": 36328 }, { "epoch": 2.8639337800551834, "grad_norm": 0.4643597900867462, "learning_rate": 0.0005, "loss": 1.216880202293396, "step": 36329 }, { "epoch": 2.864012613322822, "grad_norm": 0.5504450798034668, "learning_rate": 0.0005, "loss": 1.431397557258606, "step": 36330 }, { "epoch": 2.864091446590461, "grad_norm": 0.5886935591697693, "learning_rate": 0.0005, "loss": 1.278443694114685, "step": 36331 }, { "epoch": 2.8641702798581, "grad_norm": 0.47130000591278076, "learning_rate": 0.0005, "loss": 1.224276065826416, "step": 36332 }, { "epoch": 2.864249113125739, "grad_norm": 0.6790347695350647, "learning_rate": 0.0005, "loss": 1.448099136352539, "step": 36333 }, { "epoch": 2.864327946393378, "grad_norm": 0.7800230979919434, "learning_rate": 0.0005, "loss": 1.5004976987838745, "step": 36334 }, { "epoch": 2.864406779661017, "grad_norm": 0.509070873260498, "learning_rate": 0.0005, "loss": 1.1896898746490479, "step": 36335 }, { "epoch": 2.864485612928656, "grad_norm": 0.7530781030654907, "learning_rate": 0.0005, "loss": 1.421973705291748, "step": 36336 }, { "epoch": 2.8645644461962947, "grad_norm": 0.5053229331970215, "learning_rate": 0.0005, "loss": 1.2711375951766968, "step": 36337 }, { "epoch": 2.864643279463934, "grad_norm": 0.6176142692565918, "learning_rate": 0.0005, "loss": 1.3490252494812012, "step": 36338 }, { "epoch": 2.864722112731573, "grad_norm": 0.5594462752342224, "learning_rate": 0.0005, "loss": 1.378429889678955, "step": 36339 }, { "epoch": 2.8648009459992116, "grad_norm": 0.6441121101379395, "learning_rate": 0.0005, "loss": 1.3794262409210205, "step": 36340 }, { "epoch": 2.8648797792668503, "grad_norm": 0.5217394828796387, "learning_rate": 0.0005, "loss": 1.291242003440857, "step": 36341 }, { "epoch": 2.8649586125344895, "grad_norm": 0.5392839312553406, "learning_rate": 0.0005, "loss": 1.3928364515304565, "step": 36342 }, { "epoch": 2.8650374458021286, "grad_norm": 0.5225619673728943, "learning_rate": 0.0005, "loss": 1.3471910953521729, "step": 36343 }, { "epoch": 2.8651162790697673, "grad_norm": 0.6830936670303345, "learning_rate": 0.0005, "loss": 1.3281821012496948, "step": 36344 }, { "epoch": 2.8651951123374064, "grad_norm": 0.4916857182979584, "learning_rate": 0.0005, "loss": 1.3194257020950317, "step": 36345 }, { "epoch": 2.865273945605045, "grad_norm": 0.7838240265846252, "learning_rate": 0.0005, "loss": 1.2286288738250732, "step": 36346 }, { "epoch": 2.8653527788726842, "grad_norm": 0.5100435614585876, "learning_rate": 0.0005, "loss": 1.326472520828247, "step": 36347 }, { "epoch": 2.8654316121403234, "grad_norm": 0.5245509743690491, "learning_rate": 0.0005, "loss": 1.2924576997756958, "step": 36348 }, { "epoch": 2.865510445407962, "grad_norm": 0.43852099776268005, "learning_rate": 0.0005, "loss": 1.145445704460144, "step": 36349 }, { "epoch": 2.865589278675601, "grad_norm": 0.5632242560386658, "learning_rate": 0.0005, "loss": 1.3052842617034912, "step": 36350 }, { "epoch": 2.86566811194324, "grad_norm": 1.0751465559005737, "learning_rate": 0.0005, "loss": 1.3335846662521362, "step": 36351 }, { "epoch": 2.865746945210879, "grad_norm": 0.7335390448570251, "learning_rate": 0.0005, "loss": 1.3994452953338623, "step": 36352 }, { "epoch": 2.865825778478518, "grad_norm": 0.5373451113700867, "learning_rate": 0.0005, "loss": 1.515984296798706, "step": 36353 }, { "epoch": 2.865904611746157, "grad_norm": 0.5518673658370972, "learning_rate": 0.0005, "loss": 1.4751472473144531, "step": 36354 }, { "epoch": 2.865983445013796, "grad_norm": 0.5031486749649048, "learning_rate": 0.0005, "loss": 1.2973558902740479, "step": 36355 }, { "epoch": 2.8660622782814347, "grad_norm": 0.5119655132293701, "learning_rate": 0.0005, "loss": 1.256422519683838, "step": 36356 }, { "epoch": 2.866141111549074, "grad_norm": 0.4559604525566101, "learning_rate": 0.0005, "loss": 1.150449275970459, "step": 36357 }, { "epoch": 2.8662199448167125, "grad_norm": 0.4861823618412018, "learning_rate": 0.0005, "loss": 1.4644508361816406, "step": 36358 }, { "epoch": 2.8662987780843516, "grad_norm": 0.45528939366340637, "learning_rate": 0.0005, "loss": 1.2880604267120361, "step": 36359 }, { "epoch": 2.8663776113519903, "grad_norm": 0.4721026122570038, "learning_rate": 0.0005, "loss": 1.1188278198242188, "step": 36360 }, { "epoch": 2.8664564446196295, "grad_norm": 0.4689480662345886, "learning_rate": 0.0005, "loss": 1.3662145137786865, "step": 36361 }, { "epoch": 2.8665352778872686, "grad_norm": 0.431325227022171, "learning_rate": 0.0005, "loss": 1.36849045753479, "step": 36362 }, { "epoch": 2.8666141111549073, "grad_norm": 0.4849725365638733, "learning_rate": 0.0005, "loss": 1.1385340690612793, "step": 36363 }, { "epoch": 2.8666929444225464, "grad_norm": 0.526570200920105, "learning_rate": 0.0005, "loss": 1.2010029554367065, "step": 36364 }, { "epoch": 2.866771777690185, "grad_norm": 0.5302569270133972, "learning_rate": 0.0005, "loss": 1.0987359285354614, "step": 36365 }, { "epoch": 2.8668506109578242, "grad_norm": 0.6489048004150391, "learning_rate": 0.0005, "loss": 1.4195998907089233, "step": 36366 }, { "epoch": 2.8669294442254634, "grad_norm": 0.5431743264198303, "learning_rate": 0.0005, "loss": 1.4929485321044922, "step": 36367 }, { "epoch": 2.867008277493102, "grad_norm": 0.4442767798900604, "learning_rate": 0.0005, "loss": 1.0609562397003174, "step": 36368 }, { "epoch": 2.867087110760741, "grad_norm": 0.5132629871368408, "learning_rate": 0.0005, "loss": 1.358802318572998, "step": 36369 }, { "epoch": 2.86716594402838, "grad_norm": 1.0173227787017822, "learning_rate": 0.0005, "loss": 1.457484245300293, "step": 36370 }, { "epoch": 2.867244777296019, "grad_norm": 0.5536442399024963, "learning_rate": 0.0005, "loss": 1.4842491149902344, "step": 36371 }, { "epoch": 2.8673236105636577, "grad_norm": 0.4769110381603241, "learning_rate": 0.0005, "loss": 1.3190367221832275, "step": 36372 }, { "epoch": 2.867402443831297, "grad_norm": 0.5404021739959717, "learning_rate": 0.0005, "loss": 1.3122572898864746, "step": 36373 }, { "epoch": 2.8674812770989355, "grad_norm": 0.44199180603027344, "learning_rate": 0.0005, "loss": 1.3749611377716064, "step": 36374 }, { "epoch": 2.8675601103665747, "grad_norm": 0.41218480467796326, "learning_rate": 0.0005, "loss": 1.2171776294708252, "step": 36375 }, { "epoch": 2.867638943634214, "grad_norm": 0.4659420847892761, "learning_rate": 0.0005, "loss": 1.1351358890533447, "step": 36376 }, { "epoch": 2.8677177769018525, "grad_norm": 0.5018198490142822, "learning_rate": 0.0005, "loss": 1.360595464706421, "step": 36377 }, { "epoch": 2.8677966101694916, "grad_norm": 0.4969657063484192, "learning_rate": 0.0005, "loss": 1.278914213180542, "step": 36378 }, { "epoch": 2.8678754434371303, "grad_norm": 0.5837820172309875, "learning_rate": 0.0005, "loss": 1.3822026252746582, "step": 36379 }, { "epoch": 2.8679542767047694, "grad_norm": 0.4999212324619293, "learning_rate": 0.0005, "loss": 1.3621292114257812, "step": 36380 }, { "epoch": 2.8680331099724086, "grad_norm": 0.5312225222587585, "learning_rate": 0.0005, "loss": 1.337100863456726, "step": 36381 }, { "epoch": 2.8681119432400473, "grad_norm": 0.6102645397186279, "learning_rate": 0.0005, "loss": 1.4410375356674194, "step": 36382 }, { "epoch": 2.8681907765076864, "grad_norm": 0.5107668042182922, "learning_rate": 0.0005, "loss": 1.258094072341919, "step": 36383 }, { "epoch": 2.868269609775325, "grad_norm": 0.6784655451774597, "learning_rate": 0.0005, "loss": 1.4849916696548462, "step": 36384 }, { "epoch": 2.868348443042964, "grad_norm": 0.5184281468391418, "learning_rate": 0.0005, "loss": 1.2383877038955688, "step": 36385 }, { "epoch": 2.8684272763106033, "grad_norm": 0.5103944540023804, "learning_rate": 0.0005, "loss": 1.2680144309997559, "step": 36386 }, { "epoch": 2.868506109578242, "grad_norm": 0.539049506187439, "learning_rate": 0.0005, "loss": 1.3841995000839233, "step": 36387 }, { "epoch": 2.8685849428458807, "grad_norm": 0.46083855628967285, "learning_rate": 0.0005, "loss": 1.3254742622375488, "step": 36388 }, { "epoch": 2.86866377611352, "grad_norm": 0.49320846796035767, "learning_rate": 0.0005, "loss": 1.5498816967010498, "step": 36389 }, { "epoch": 2.868742609381159, "grad_norm": 0.4861029386520386, "learning_rate": 0.0005, "loss": 1.2288026809692383, "step": 36390 }, { "epoch": 2.8688214426487977, "grad_norm": 0.45021650195121765, "learning_rate": 0.0005, "loss": 1.246031641960144, "step": 36391 }, { "epoch": 2.868900275916437, "grad_norm": 0.5017424821853638, "learning_rate": 0.0005, "loss": 1.3353298902511597, "step": 36392 }, { "epoch": 2.8689791091840755, "grad_norm": 0.5067558288574219, "learning_rate": 0.0005, "loss": 1.2730062007904053, "step": 36393 }, { "epoch": 2.8690579424517146, "grad_norm": 0.4906540513038635, "learning_rate": 0.0005, "loss": 1.3297085762023926, "step": 36394 }, { "epoch": 2.8691367757193538, "grad_norm": 0.48767775297164917, "learning_rate": 0.0005, "loss": 1.210322618484497, "step": 36395 }, { "epoch": 2.8692156089869925, "grad_norm": 0.5203908681869507, "learning_rate": 0.0005, "loss": 1.423248291015625, "step": 36396 }, { "epoch": 2.8692944422546316, "grad_norm": 0.608898401260376, "learning_rate": 0.0005, "loss": 1.191123366355896, "step": 36397 }, { "epoch": 2.8693732755222703, "grad_norm": 0.47165852785110474, "learning_rate": 0.0005, "loss": 1.2480038404464722, "step": 36398 }, { "epoch": 2.8694521087899094, "grad_norm": 0.577812910079956, "learning_rate": 0.0005, "loss": 1.3736292123794556, "step": 36399 }, { "epoch": 2.8695309420575486, "grad_norm": 0.554250955581665, "learning_rate": 0.0005, "loss": 1.3499116897583008, "step": 36400 }, { "epoch": 2.8696097753251872, "grad_norm": 0.7287899255752563, "learning_rate": 0.0005, "loss": 1.203416347503662, "step": 36401 }, { "epoch": 2.869688608592826, "grad_norm": 0.5976173281669617, "learning_rate": 0.0005, "loss": 1.2940895557403564, "step": 36402 }, { "epoch": 2.869767441860465, "grad_norm": 0.49950000643730164, "learning_rate": 0.0005, "loss": 1.5660265684127808, "step": 36403 }, { "epoch": 2.869846275128104, "grad_norm": 0.5293172001838684, "learning_rate": 0.0005, "loss": 1.3166106939315796, "step": 36404 }, { "epoch": 2.869925108395743, "grad_norm": 0.682438313961029, "learning_rate": 0.0005, "loss": 1.474092721939087, "step": 36405 }, { "epoch": 2.870003941663382, "grad_norm": 0.607006847858429, "learning_rate": 0.0005, "loss": 1.513258695602417, "step": 36406 }, { "epoch": 2.8700827749310207, "grad_norm": 0.5030569434165955, "learning_rate": 0.0005, "loss": 1.2541546821594238, "step": 36407 }, { "epoch": 2.87016160819866, "grad_norm": 0.5776571035385132, "learning_rate": 0.0005, "loss": 1.4797301292419434, "step": 36408 }, { "epoch": 2.870240441466299, "grad_norm": 0.5455103516578674, "learning_rate": 0.0005, "loss": 1.2821401357650757, "step": 36409 }, { "epoch": 2.8703192747339377, "grad_norm": 0.8657872080802917, "learning_rate": 0.0005, "loss": 1.2330580949783325, "step": 36410 }, { "epoch": 2.870398108001577, "grad_norm": 0.4416126310825348, "learning_rate": 0.0005, "loss": 1.3003556728363037, "step": 36411 }, { "epoch": 2.8704769412692155, "grad_norm": 0.5437811613082886, "learning_rate": 0.0005, "loss": 1.3532922267913818, "step": 36412 }, { "epoch": 2.8705557745368546, "grad_norm": 0.5032335519790649, "learning_rate": 0.0005, "loss": 1.262155294418335, "step": 36413 }, { "epoch": 2.8706346078044938, "grad_norm": 0.6472664475440979, "learning_rate": 0.0005, "loss": 1.2475495338439941, "step": 36414 }, { "epoch": 2.8707134410721324, "grad_norm": 0.7840546369552612, "learning_rate": 0.0005, "loss": 1.4273314476013184, "step": 36415 }, { "epoch": 2.870792274339771, "grad_norm": 0.6058002710342407, "learning_rate": 0.0005, "loss": 1.4861323833465576, "step": 36416 }, { "epoch": 2.8708711076074103, "grad_norm": 0.4757229685783386, "learning_rate": 0.0005, "loss": 1.363844871520996, "step": 36417 }, { "epoch": 2.8709499408750494, "grad_norm": 0.566101610660553, "learning_rate": 0.0005, "loss": 1.385593295097351, "step": 36418 }, { "epoch": 2.871028774142688, "grad_norm": 0.49374327063560486, "learning_rate": 0.0005, "loss": 1.1514040231704712, "step": 36419 }, { "epoch": 2.8711076074103272, "grad_norm": 0.6174482703208923, "learning_rate": 0.0005, "loss": 1.74800443649292, "step": 36420 }, { "epoch": 2.871186440677966, "grad_norm": 0.4772014915943146, "learning_rate": 0.0005, "loss": 1.157059907913208, "step": 36421 }, { "epoch": 2.871265273945605, "grad_norm": 0.4988636374473572, "learning_rate": 0.0005, "loss": 1.4115042686462402, "step": 36422 }, { "epoch": 2.871344107213244, "grad_norm": 0.45444029569625854, "learning_rate": 0.0005, "loss": 1.401146650314331, "step": 36423 }, { "epoch": 2.871422940480883, "grad_norm": 1.706268548965454, "learning_rate": 0.0005, "loss": 1.5717415809631348, "step": 36424 }, { "epoch": 2.871501773748522, "grad_norm": 0.5127494931221008, "learning_rate": 0.0005, "loss": 1.4867854118347168, "step": 36425 }, { "epoch": 2.8715806070161607, "grad_norm": 0.5006505250930786, "learning_rate": 0.0005, "loss": 1.2332863807678223, "step": 36426 }, { "epoch": 2.8716594402838, "grad_norm": 0.5039277672767639, "learning_rate": 0.0005, "loss": 1.0726795196533203, "step": 36427 }, { "epoch": 2.871738273551439, "grad_norm": 0.5767279863357544, "learning_rate": 0.0005, "loss": 1.2823176383972168, "step": 36428 }, { "epoch": 2.8718171068190776, "grad_norm": 0.6162203550338745, "learning_rate": 0.0005, "loss": 1.3606407642364502, "step": 36429 }, { "epoch": 2.8718959400867163, "grad_norm": 0.5408912897109985, "learning_rate": 0.0005, "loss": 1.2213218212127686, "step": 36430 }, { "epoch": 2.8719747733543555, "grad_norm": 0.4781284034252167, "learning_rate": 0.0005, "loss": 1.221275806427002, "step": 36431 }, { "epoch": 2.8720536066219946, "grad_norm": 0.6252883672714233, "learning_rate": 0.0005, "loss": 1.3476035594940186, "step": 36432 }, { "epoch": 2.8721324398896333, "grad_norm": 0.5212606191635132, "learning_rate": 0.0005, "loss": 1.3860183954238892, "step": 36433 }, { "epoch": 2.8722112731572724, "grad_norm": 0.584945797920227, "learning_rate": 0.0005, "loss": 1.5509583950042725, "step": 36434 }, { "epoch": 2.872290106424911, "grad_norm": 0.6418894529342651, "learning_rate": 0.0005, "loss": 1.3559415340423584, "step": 36435 }, { "epoch": 2.8723689396925502, "grad_norm": 0.4410529136657715, "learning_rate": 0.0005, "loss": 1.2026541233062744, "step": 36436 }, { "epoch": 2.8724477729601894, "grad_norm": 0.5473162531852722, "learning_rate": 0.0005, "loss": 1.428722858428955, "step": 36437 }, { "epoch": 2.872526606227828, "grad_norm": 1.3426241874694824, "learning_rate": 0.0005, "loss": 1.2520498037338257, "step": 36438 }, { "epoch": 2.872605439495467, "grad_norm": 0.674670934677124, "learning_rate": 0.0005, "loss": 1.3870956897735596, "step": 36439 }, { "epoch": 2.872684272763106, "grad_norm": 0.4754599928855896, "learning_rate": 0.0005, "loss": 1.2336115837097168, "step": 36440 }, { "epoch": 2.872763106030745, "grad_norm": 0.5915988087654114, "learning_rate": 0.0005, "loss": 1.3151867389678955, "step": 36441 }, { "epoch": 2.872841939298384, "grad_norm": 0.5847609043121338, "learning_rate": 0.0005, "loss": 1.5430705547332764, "step": 36442 }, { "epoch": 2.872920772566023, "grad_norm": 0.4826420545578003, "learning_rate": 0.0005, "loss": 1.2214064598083496, "step": 36443 }, { "epoch": 2.8729996058336615, "grad_norm": 0.5926339626312256, "learning_rate": 0.0005, "loss": 1.2437533140182495, "step": 36444 }, { "epoch": 2.8730784391013007, "grad_norm": 0.6535995006561279, "learning_rate": 0.0005, "loss": 1.3248416185379028, "step": 36445 }, { "epoch": 2.87315727236894, "grad_norm": 0.49551069736480713, "learning_rate": 0.0005, "loss": 1.3866198062896729, "step": 36446 }, { "epoch": 2.8732361056365785, "grad_norm": 0.4658653736114502, "learning_rate": 0.0005, "loss": 1.3366094827651978, "step": 36447 }, { "epoch": 2.8733149389042176, "grad_norm": 0.5940082669258118, "learning_rate": 0.0005, "loss": 1.435829520225525, "step": 36448 }, { "epoch": 2.8733937721718563, "grad_norm": 0.6314881443977356, "learning_rate": 0.0005, "loss": 1.3957905769348145, "step": 36449 }, { "epoch": 2.8734726054394955, "grad_norm": 0.4403114318847656, "learning_rate": 0.0005, "loss": 1.3650102615356445, "step": 36450 }, { "epoch": 2.8735514387071346, "grad_norm": 0.5226274132728577, "learning_rate": 0.0005, "loss": 1.3958321809768677, "step": 36451 }, { "epoch": 2.8736302719747733, "grad_norm": 0.4956425726413727, "learning_rate": 0.0005, "loss": 1.3362326622009277, "step": 36452 }, { "epoch": 2.8737091052424124, "grad_norm": 0.48728057742118835, "learning_rate": 0.0005, "loss": 1.3616461753845215, "step": 36453 }, { "epoch": 2.873787938510051, "grad_norm": 0.5541855692863464, "learning_rate": 0.0005, "loss": 1.4590516090393066, "step": 36454 }, { "epoch": 2.8738667717776902, "grad_norm": 0.6361741423606873, "learning_rate": 0.0005, "loss": 1.3413939476013184, "step": 36455 }, { "epoch": 2.8739456050453294, "grad_norm": 0.5942999124526978, "learning_rate": 0.0005, "loss": 1.3062617778778076, "step": 36456 }, { "epoch": 2.874024438312968, "grad_norm": 0.47501567006111145, "learning_rate": 0.0005, "loss": 1.2418270111083984, "step": 36457 }, { "epoch": 2.8741032715806067, "grad_norm": 0.5448896884918213, "learning_rate": 0.0005, "loss": 1.4455376863479614, "step": 36458 }, { "epoch": 2.874182104848246, "grad_norm": 0.44544824957847595, "learning_rate": 0.0005, "loss": 1.2687664031982422, "step": 36459 }, { "epoch": 2.874260938115885, "grad_norm": 0.6139150857925415, "learning_rate": 0.0005, "loss": 1.307905912399292, "step": 36460 }, { "epoch": 2.8743397713835237, "grad_norm": 0.4934673607349396, "learning_rate": 0.0005, "loss": 1.1584972143173218, "step": 36461 }, { "epoch": 2.874418604651163, "grad_norm": 0.41268807649612427, "learning_rate": 0.0005, "loss": 1.1992933750152588, "step": 36462 }, { "epoch": 2.8744974379188015, "grad_norm": 0.46687594056129456, "learning_rate": 0.0005, "loss": 1.170684576034546, "step": 36463 }, { "epoch": 2.8745762711864407, "grad_norm": 0.4246561825275421, "learning_rate": 0.0005, "loss": 1.2031755447387695, "step": 36464 }, { "epoch": 2.87465510445408, "grad_norm": 0.4579918086528778, "learning_rate": 0.0005, "loss": 1.2654833793640137, "step": 36465 }, { "epoch": 2.8747339377217185, "grad_norm": 0.5000869631767273, "learning_rate": 0.0005, "loss": 1.2679959535598755, "step": 36466 }, { "epoch": 2.8748127709893576, "grad_norm": 0.45521780848503113, "learning_rate": 0.0005, "loss": 1.320356845855713, "step": 36467 }, { "epoch": 2.8748916042569963, "grad_norm": 0.5409853458404541, "learning_rate": 0.0005, "loss": 1.2944061756134033, "step": 36468 }, { "epoch": 2.8749704375246354, "grad_norm": 0.44796863198280334, "learning_rate": 0.0005, "loss": 1.5040076971054077, "step": 36469 }, { "epoch": 2.8750492707922746, "grad_norm": 0.4096658527851105, "learning_rate": 0.0005, "loss": 1.1933269500732422, "step": 36470 }, { "epoch": 2.8751281040599133, "grad_norm": 0.5406467914581299, "learning_rate": 0.0005, "loss": 1.3836318254470825, "step": 36471 }, { "epoch": 2.8752069373275524, "grad_norm": 0.4669812023639679, "learning_rate": 0.0005, "loss": 1.3925971984863281, "step": 36472 }, { "epoch": 2.875285770595191, "grad_norm": 0.41971471905708313, "learning_rate": 0.0005, "loss": 1.1726231575012207, "step": 36473 }, { "epoch": 2.87536460386283, "grad_norm": 0.5653309226036072, "learning_rate": 0.0005, "loss": 1.307144284248352, "step": 36474 }, { "epoch": 2.875443437130469, "grad_norm": 0.3941824734210968, "learning_rate": 0.0005, "loss": 1.2800586223602295, "step": 36475 }, { "epoch": 2.875522270398108, "grad_norm": 0.41450783610343933, "learning_rate": 0.0005, "loss": 1.321289300918579, "step": 36476 }, { "epoch": 2.8756011036657467, "grad_norm": 0.43008849024772644, "learning_rate": 0.0005, "loss": 1.1944661140441895, "step": 36477 }, { "epoch": 2.875679936933386, "grad_norm": 0.5116217732429504, "learning_rate": 0.0005, "loss": 1.3784360885620117, "step": 36478 }, { "epoch": 2.875758770201025, "grad_norm": 0.48642823100090027, "learning_rate": 0.0005, "loss": 1.5005052089691162, "step": 36479 }, { "epoch": 2.8758376034686637, "grad_norm": 0.5637512803077698, "learning_rate": 0.0005, "loss": 1.397763967514038, "step": 36480 }, { "epoch": 2.875916436736303, "grad_norm": 0.4806482195854187, "learning_rate": 0.0005, "loss": 1.2513277530670166, "step": 36481 }, { "epoch": 2.8759952700039415, "grad_norm": 0.6559802889823914, "learning_rate": 0.0005, "loss": 1.385132074356079, "step": 36482 }, { "epoch": 2.8760741032715806, "grad_norm": 0.41846275329589844, "learning_rate": 0.0005, "loss": 1.209188461303711, "step": 36483 }, { "epoch": 2.8761529365392198, "grad_norm": 0.4825528562068939, "learning_rate": 0.0005, "loss": 1.4077049493789673, "step": 36484 }, { "epoch": 2.8762317698068585, "grad_norm": 0.6210333108901978, "learning_rate": 0.0005, "loss": 1.6537244319915771, "step": 36485 }, { "epoch": 2.8763106030744976, "grad_norm": 0.5394681692123413, "learning_rate": 0.0005, "loss": 1.3473321199417114, "step": 36486 }, { "epoch": 2.8763894363421363, "grad_norm": 0.5353785157203674, "learning_rate": 0.0005, "loss": 1.593119740486145, "step": 36487 }, { "epoch": 2.8764682696097754, "grad_norm": 0.5216358304023743, "learning_rate": 0.0005, "loss": 1.403244972229004, "step": 36488 }, { "epoch": 2.876547102877414, "grad_norm": 0.43013185262680054, "learning_rate": 0.0005, "loss": 1.37226402759552, "step": 36489 }, { "epoch": 2.8766259361450532, "grad_norm": 0.49271294474601746, "learning_rate": 0.0005, "loss": 1.0085633993148804, "step": 36490 }, { "epoch": 2.876704769412692, "grad_norm": 0.4824048578739166, "learning_rate": 0.0005, "loss": 1.1717593669891357, "step": 36491 }, { "epoch": 2.876783602680331, "grad_norm": 0.4477463960647583, "learning_rate": 0.0005, "loss": 1.1259623765945435, "step": 36492 }, { "epoch": 2.87686243594797, "grad_norm": 0.49126487970352173, "learning_rate": 0.0005, "loss": 1.438476324081421, "step": 36493 }, { "epoch": 2.876941269215609, "grad_norm": 0.4812384247779846, "learning_rate": 0.0005, "loss": 1.2324113845825195, "step": 36494 }, { "epoch": 2.877020102483248, "grad_norm": 0.49517714977264404, "learning_rate": 0.0005, "loss": 1.2823518514633179, "step": 36495 }, { "epoch": 2.8770989357508867, "grad_norm": 0.5214880704879761, "learning_rate": 0.0005, "loss": 1.3994348049163818, "step": 36496 }, { "epoch": 2.877177769018526, "grad_norm": 0.4755615293979645, "learning_rate": 0.0005, "loss": 1.2833244800567627, "step": 36497 }, { "epoch": 2.877256602286165, "grad_norm": 0.42173290252685547, "learning_rate": 0.0005, "loss": 1.1149656772613525, "step": 36498 }, { "epoch": 2.8773354355538037, "grad_norm": 0.46697232127189636, "learning_rate": 0.0005, "loss": 1.283451795578003, "step": 36499 }, { "epoch": 2.877414268821443, "grad_norm": 0.47220873832702637, "learning_rate": 0.0005, "loss": 1.5344812870025635, "step": 36500 }, { "epoch": 2.8774931020890815, "grad_norm": 0.5496098399162292, "learning_rate": 0.0005, "loss": 1.295677661895752, "step": 36501 }, { "epoch": 2.8775719353567206, "grad_norm": 0.5731765031814575, "learning_rate": 0.0005, "loss": 1.4506466388702393, "step": 36502 }, { "epoch": 2.8776507686243598, "grad_norm": 0.7918035387992859, "learning_rate": 0.0005, "loss": 1.3460232019424438, "step": 36503 }, { "epoch": 2.8777296018919984, "grad_norm": 0.5715234875679016, "learning_rate": 0.0005, "loss": 1.3142470121383667, "step": 36504 }, { "epoch": 2.877808435159637, "grad_norm": 0.49332717061042786, "learning_rate": 0.0005, "loss": 1.20035982131958, "step": 36505 }, { "epoch": 2.8778872684272763, "grad_norm": 0.8982821702957153, "learning_rate": 0.0005, "loss": 1.4307892322540283, "step": 36506 }, { "epoch": 2.8779661016949154, "grad_norm": 0.55624920129776, "learning_rate": 0.0005, "loss": 1.478136658668518, "step": 36507 }, { "epoch": 2.878044934962554, "grad_norm": 0.4595266282558441, "learning_rate": 0.0005, "loss": 1.2232012748718262, "step": 36508 }, { "epoch": 2.8781237682301932, "grad_norm": 0.5992075800895691, "learning_rate": 0.0005, "loss": 1.407752513885498, "step": 36509 }, { "epoch": 2.878202601497832, "grad_norm": 0.4947925806045532, "learning_rate": 0.0005, "loss": 1.2658838033676147, "step": 36510 }, { "epoch": 2.878281434765471, "grad_norm": 0.47399067878723145, "learning_rate": 0.0005, "loss": 1.4824923276901245, "step": 36511 }, { "epoch": 2.87836026803311, "grad_norm": 0.39821237325668335, "learning_rate": 0.0005, "loss": 1.3236138820648193, "step": 36512 }, { "epoch": 2.878439101300749, "grad_norm": 0.4539537727832794, "learning_rate": 0.0005, "loss": 1.3490793704986572, "step": 36513 }, { "epoch": 2.878517934568388, "grad_norm": 0.38521334528923035, "learning_rate": 0.0005, "loss": 1.1875463724136353, "step": 36514 }, { "epoch": 2.8785967678360267, "grad_norm": 0.49308013916015625, "learning_rate": 0.0005, "loss": 1.28165864944458, "step": 36515 }, { "epoch": 2.878675601103666, "grad_norm": 0.7898513078689575, "learning_rate": 0.0005, "loss": 1.2366387844085693, "step": 36516 }, { "epoch": 2.878754434371305, "grad_norm": 0.4429032802581787, "learning_rate": 0.0005, "loss": 1.118362307548523, "step": 36517 }, { "epoch": 2.8788332676389436, "grad_norm": 0.47166961431503296, "learning_rate": 0.0005, "loss": 1.1575363874435425, "step": 36518 }, { "epoch": 2.8789121009065823, "grad_norm": 0.40892836451530457, "learning_rate": 0.0005, "loss": 1.1762402057647705, "step": 36519 }, { "epoch": 2.8789909341742215, "grad_norm": 0.45544272661209106, "learning_rate": 0.0005, "loss": 1.275032877922058, "step": 36520 }, { "epoch": 2.8790697674418606, "grad_norm": 0.6566174030303955, "learning_rate": 0.0005, "loss": 1.3092013597488403, "step": 36521 }, { "epoch": 2.8791486007094993, "grad_norm": 0.5773069858551025, "learning_rate": 0.0005, "loss": 1.3398154973983765, "step": 36522 }, { "epoch": 2.8792274339771384, "grad_norm": 0.45576781034469604, "learning_rate": 0.0005, "loss": 1.3243043422698975, "step": 36523 }, { "epoch": 2.879306267244777, "grad_norm": 0.7728612422943115, "learning_rate": 0.0005, "loss": 1.4137356281280518, "step": 36524 }, { "epoch": 2.8793851005124163, "grad_norm": 0.44013240933418274, "learning_rate": 0.0005, "loss": 1.1524208784103394, "step": 36525 }, { "epoch": 2.8794639337800554, "grad_norm": 0.4502837061882019, "learning_rate": 0.0005, "loss": 1.2408709526062012, "step": 36526 }, { "epoch": 2.879542767047694, "grad_norm": 0.511460542678833, "learning_rate": 0.0005, "loss": 1.2260314226150513, "step": 36527 }, { "epoch": 2.879621600315333, "grad_norm": 0.579742431640625, "learning_rate": 0.0005, "loss": 1.5394642353057861, "step": 36528 }, { "epoch": 2.879700433582972, "grad_norm": 0.5067777037620544, "learning_rate": 0.0005, "loss": 1.3873882293701172, "step": 36529 }, { "epoch": 2.879779266850611, "grad_norm": 0.5702305436134338, "learning_rate": 0.0005, "loss": 1.1850152015686035, "step": 36530 }, { "epoch": 2.87985810011825, "grad_norm": 0.5762619972229004, "learning_rate": 0.0005, "loss": 1.2658295631408691, "step": 36531 }, { "epoch": 2.879936933385889, "grad_norm": 0.5282636880874634, "learning_rate": 0.0005, "loss": 1.4587409496307373, "step": 36532 }, { "epoch": 2.8800157666535275, "grad_norm": 0.5091283321380615, "learning_rate": 0.0005, "loss": 1.4628055095672607, "step": 36533 }, { "epoch": 2.8800945999211667, "grad_norm": 0.4576140344142914, "learning_rate": 0.0005, "loss": 1.3189994096755981, "step": 36534 }, { "epoch": 2.880173433188806, "grad_norm": 0.5879773497581482, "learning_rate": 0.0005, "loss": 1.3657861948013306, "step": 36535 }, { "epoch": 2.8802522664564445, "grad_norm": 0.46028372645378113, "learning_rate": 0.0005, "loss": 1.2227859497070312, "step": 36536 }, { "epoch": 2.8803310997240836, "grad_norm": 0.543895423412323, "learning_rate": 0.0005, "loss": 1.3194923400878906, "step": 36537 }, { "epoch": 2.8804099329917223, "grad_norm": 0.44156959652900696, "learning_rate": 0.0005, "loss": 1.132401704788208, "step": 36538 }, { "epoch": 2.8804887662593615, "grad_norm": 0.5525791645050049, "learning_rate": 0.0005, "loss": 1.398169755935669, "step": 36539 }, { "epoch": 2.8805675995270006, "grad_norm": 0.42689818143844604, "learning_rate": 0.0005, "loss": 1.2630552053451538, "step": 36540 }, { "epoch": 2.8806464327946393, "grad_norm": 0.6045463681221008, "learning_rate": 0.0005, "loss": 1.4759129285812378, "step": 36541 }, { "epoch": 2.8807252660622784, "grad_norm": 3.035430908203125, "learning_rate": 0.0005, "loss": 1.2534135580062866, "step": 36542 }, { "epoch": 2.880804099329917, "grad_norm": 0.4951414167881012, "learning_rate": 0.0005, "loss": 1.4919894933700562, "step": 36543 }, { "epoch": 2.8808829325975562, "grad_norm": 0.5235878229141235, "learning_rate": 0.0005, "loss": 1.320765733718872, "step": 36544 }, { "epoch": 2.8809617658651954, "grad_norm": 0.5848876237869263, "learning_rate": 0.0005, "loss": 1.2418770790100098, "step": 36545 }, { "epoch": 2.881040599132834, "grad_norm": 0.49303409457206726, "learning_rate": 0.0005, "loss": 1.2723557949066162, "step": 36546 }, { "epoch": 2.8811194324004727, "grad_norm": 0.4239048957824707, "learning_rate": 0.0005, "loss": 1.3780986070632935, "step": 36547 }, { "epoch": 2.881198265668112, "grad_norm": 0.49008432030677795, "learning_rate": 0.0005, "loss": 1.331770658493042, "step": 36548 }, { "epoch": 2.881277098935751, "grad_norm": 0.45058968663215637, "learning_rate": 0.0005, "loss": 1.1879088878631592, "step": 36549 }, { "epoch": 2.8813559322033897, "grad_norm": 0.41614705324172974, "learning_rate": 0.0005, "loss": 1.226743221282959, "step": 36550 }, { "epoch": 2.881434765471029, "grad_norm": 0.5377364158630371, "learning_rate": 0.0005, "loss": 1.312401294708252, "step": 36551 }, { "epoch": 2.8815135987386675, "grad_norm": 0.5108594298362732, "learning_rate": 0.0005, "loss": 1.3034828901290894, "step": 36552 }, { "epoch": 2.8815924320063067, "grad_norm": 0.5614683628082275, "learning_rate": 0.0005, "loss": 1.393176555633545, "step": 36553 }, { "epoch": 2.881671265273946, "grad_norm": 0.5196362733840942, "learning_rate": 0.0005, "loss": 1.2811272144317627, "step": 36554 }, { "epoch": 2.8817500985415845, "grad_norm": 0.6044563055038452, "learning_rate": 0.0005, "loss": 1.2322371006011963, "step": 36555 }, { "epoch": 2.8818289318092236, "grad_norm": 0.5909515619277954, "learning_rate": 0.0005, "loss": 1.2539019584655762, "step": 36556 }, { "epoch": 2.8819077650768623, "grad_norm": 0.4927769601345062, "learning_rate": 0.0005, "loss": 1.3609912395477295, "step": 36557 }, { "epoch": 2.8819865983445014, "grad_norm": 0.5916493535041809, "learning_rate": 0.0005, "loss": 1.4071614742279053, "step": 36558 }, { "epoch": 2.8820654316121406, "grad_norm": 0.5332538485527039, "learning_rate": 0.0005, "loss": 1.2661224603652954, "step": 36559 }, { "epoch": 2.8821442648797793, "grad_norm": 0.4344722032546997, "learning_rate": 0.0005, "loss": 1.3528274297714233, "step": 36560 }, { "epoch": 2.882223098147418, "grad_norm": 0.5678786635398865, "learning_rate": 0.0005, "loss": 1.234170913696289, "step": 36561 }, { "epoch": 2.882301931415057, "grad_norm": 0.5429990291595459, "learning_rate": 0.0005, "loss": 1.2726762294769287, "step": 36562 }, { "epoch": 2.882380764682696, "grad_norm": 0.5740481019020081, "learning_rate": 0.0005, "loss": 1.3382914066314697, "step": 36563 }, { "epoch": 2.882459597950335, "grad_norm": 0.4771062433719635, "learning_rate": 0.0005, "loss": 1.5915091037750244, "step": 36564 }, { "epoch": 2.882538431217974, "grad_norm": 0.522742748260498, "learning_rate": 0.0005, "loss": 1.4099340438842773, "step": 36565 }, { "epoch": 2.8826172644856127, "grad_norm": 0.5147879123687744, "learning_rate": 0.0005, "loss": 1.4269065856933594, "step": 36566 }, { "epoch": 2.882696097753252, "grad_norm": 0.4245317280292511, "learning_rate": 0.0005, "loss": 1.3503425121307373, "step": 36567 }, { "epoch": 2.882774931020891, "grad_norm": 0.4129003882408142, "learning_rate": 0.0005, "loss": 1.1745376586914062, "step": 36568 }, { "epoch": 2.8828537642885297, "grad_norm": 0.43738967180252075, "learning_rate": 0.0005, "loss": 1.1904172897338867, "step": 36569 }, { "epoch": 2.882932597556169, "grad_norm": 0.4789882302284241, "learning_rate": 0.0005, "loss": 1.4683058261871338, "step": 36570 }, { "epoch": 2.8830114308238075, "grad_norm": 0.47681760787963867, "learning_rate": 0.0005, "loss": 1.490608811378479, "step": 36571 }, { "epoch": 2.8830902640914466, "grad_norm": 0.6677345037460327, "learning_rate": 0.0005, "loss": 1.3712375164031982, "step": 36572 }, { "epoch": 2.8831690973590858, "grad_norm": 0.43169450759887695, "learning_rate": 0.0005, "loss": 1.2450742721557617, "step": 36573 }, { "epoch": 2.8832479306267245, "grad_norm": 0.5343952775001526, "learning_rate": 0.0005, "loss": 1.3354893922805786, "step": 36574 }, { "epoch": 2.883326763894363, "grad_norm": 0.4634735584259033, "learning_rate": 0.0005, "loss": 1.2075424194335938, "step": 36575 }, { "epoch": 2.8834055971620023, "grad_norm": 0.7725159525871277, "learning_rate": 0.0005, "loss": 1.5516597032546997, "step": 36576 }, { "epoch": 2.8834844304296414, "grad_norm": 0.614054799079895, "learning_rate": 0.0005, "loss": 1.340723991394043, "step": 36577 }, { "epoch": 2.88356326369728, "grad_norm": 0.5417001843452454, "learning_rate": 0.0005, "loss": 1.303684115409851, "step": 36578 }, { "epoch": 2.8836420969649192, "grad_norm": 0.5085877776145935, "learning_rate": 0.0005, "loss": 1.3923159837722778, "step": 36579 }, { "epoch": 2.883720930232558, "grad_norm": 0.49469664692878723, "learning_rate": 0.0005, "loss": 1.2560639381408691, "step": 36580 }, { "epoch": 2.883799763500197, "grad_norm": 0.49097201228141785, "learning_rate": 0.0005, "loss": 1.33750581741333, "step": 36581 }, { "epoch": 2.883878596767836, "grad_norm": 0.501337468624115, "learning_rate": 0.0005, "loss": 1.4418957233428955, "step": 36582 }, { "epoch": 2.883957430035475, "grad_norm": 0.5092079043388367, "learning_rate": 0.0005, "loss": 1.5127615928649902, "step": 36583 }, { "epoch": 2.884036263303114, "grad_norm": 0.5137975215911865, "learning_rate": 0.0005, "loss": 1.2078055143356323, "step": 36584 }, { "epoch": 2.8841150965707527, "grad_norm": 0.4744396507740021, "learning_rate": 0.0005, "loss": 1.4154245853424072, "step": 36585 }, { "epoch": 2.884193929838392, "grad_norm": 0.5584964752197266, "learning_rate": 0.0005, "loss": 1.492332100868225, "step": 36586 }, { "epoch": 2.884272763106031, "grad_norm": 0.49521151185035706, "learning_rate": 0.0005, "loss": 1.2437254190444946, "step": 36587 }, { "epoch": 2.8843515963736697, "grad_norm": 0.4997524321079254, "learning_rate": 0.0005, "loss": 1.3384008407592773, "step": 36588 }, { "epoch": 2.8844304296413084, "grad_norm": 0.5645609498023987, "learning_rate": 0.0005, "loss": 1.3406996726989746, "step": 36589 }, { "epoch": 2.8845092629089475, "grad_norm": 0.46894124150276184, "learning_rate": 0.0005, "loss": 1.2241098880767822, "step": 36590 }, { "epoch": 2.8845880961765866, "grad_norm": 0.45653659105300903, "learning_rate": 0.0005, "loss": 1.2558298110961914, "step": 36591 }, { "epoch": 2.8846669294442253, "grad_norm": 0.5383002161979675, "learning_rate": 0.0005, "loss": 1.0979397296905518, "step": 36592 }, { "epoch": 2.8847457627118644, "grad_norm": 0.455277681350708, "learning_rate": 0.0005, "loss": 1.2801520824432373, "step": 36593 }, { "epoch": 2.884824595979503, "grad_norm": 0.5858082175254822, "learning_rate": 0.0005, "loss": 1.1139312982559204, "step": 36594 }, { "epoch": 2.8849034292471423, "grad_norm": 0.48146671056747437, "learning_rate": 0.0005, "loss": 1.3272417783737183, "step": 36595 }, { "epoch": 2.8849822625147814, "grad_norm": 0.51017826795578, "learning_rate": 0.0005, "loss": 1.2429662942886353, "step": 36596 }, { "epoch": 2.88506109578242, "grad_norm": 0.5731446743011475, "learning_rate": 0.0005, "loss": 1.4230101108551025, "step": 36597 }, { "epoch": 2.8851399290500592, "grad_norm": 0.49743446707725525, "learning_rate": 0.0005, "loss": 1.1959891319274902, "step": 36598 }, { "epoch": 2.885218762317698, "grad_norm": 0.4850253462791443, "learning_rate": 0.0005, "loss": 1.3815579414367676, "step": 36599 }, { "epoch": 2.885297595585337, "grad_norm": 0.4670719802379608, "learning_rate": 0.0005, "loss": 1.3307828903198242, "step": 36600 }, { "epoch": 2.885376428852976, "grad_norm": 0.5878838896751404, "learning_rate": 0.0005, "loss": 1.1317827701568604, "step": 36601 }, { "epoch": 2.885455262120615, "grad_norm": 0.5636633038520813, "learning_rate": 0.0005, "loss": 1.3447240591049194, "step": 36602 }, { "epoch": 2.885534095388254, "grad_norm": 0.4998149573802948, "learning_rate": 0.0005, "loss": 1.2676516771316528, "step": 36603 }, { "epoch": 2.8856129286558927, "grad_norm": 0.5509403347969055, "learning_rate": 0.0005, "loss": 1.270064115524292, "step": 36604 }, { "epoch": 2.885691761923532, "grad_norm": 0.48462438583374023, "learning_rate": 0.0005, "loss": 1.3980445861816406, "step": 36605 }, { "epoch": 2.8857705951911705, "grad_norm": 0.48173078894615173, "learning_rate": 0.0005, "loss": 1.3181543350219727, "step": 36606 }, { "epoch": 2.8858494284588097, "grad_norm": 0.5179102420806885, "learning_rate": 0.0005, "loss": 1.4876116514205933, "step": 36607 }, { "epoch": 2.8859282617264483, "grad_norm": 0.5486032962799072, "learning_rate": 0.0005, "loss": 1.3715121746063232, "step": 36608 }, { "epoch": 2.8860070949940875, "grad_norm": 0.49985283613204956, "learning_rate": 0.0005, "loss": 1.2438287734985352, "step": 36609 }, { "epoch": 2.8860859282617266, "grad_norm": 0.46828755736351013, "learning_rate": 0.0005, "loss": 1.160498857498169, "step": 36610 }, { "epoch": 2.8861647615293653, "grad_norm": 0.4528556168079376, "learning_rate": 0.0005, "loss": 1.3449161052703857, "step": 36611 }, { "epoch": 2.8862435947970044, "grad_norm": 0.4906470775604248, "learning_rate": 0.0005, "loss": 1.13226318359375, "step": 36612 }, { "epoch": 2.886322428064643, "grad_norm": 0.4771851599216461, "learning_rate": 0.0005, "loss": 1.2952516078948975, "step": 36613 }, { "epoch": 2.8864012613322823, "grad_norm": 0.4908040761947632, "learning_rate": 0.0005, "loss": 1.3596446514129639, "step": 36614 }, { "epoch": 2.8864800945999214, "grad_norm": 0.5323699116706848, "learning_rate": 0.0005, "loss": 1.340969443321228, "step": 36615 }, { "epoch": 2.88655892786756, "grad_norm": 0.411833792924881, "learning_rate": 0.0005, "loss": 1.241812825202942, "step": 36616 }, { "epoch": 2.886637761135199, "grad_norm": 0.4855908453464508, "learning_rate": 0.0005, "loss": 1.379179835319519, "step": 36617 }, { "epoch": 2.886716594402838, "grad_norm": 0.8210268020629883, "learning_rate": 0.0005, "loss": 1.3056234121322632, "step": 36618 }, { "epoch": 2.886795427670477, "grad_norm": 0.4772496819496155, "learning_rate": 0.0005, "loss": 1.2544655799865723, "step": 36619 }, { "epoch": 2.8868742609381157, "grad_norm": 0.4532761871814728, "learning_rate": 0.0005, "loss": 1.3918198347091675, "step": 36620 }, { "epoch": 2.886953094205755, "grad_norm": 0.6336896419525146, "learning_rate": 0.0005, "loss": 1.4846184253692627, "step": 36621 }, { "epoch": 2.8870319274733935, "grad_norm": 0.5040792226791382, "learning_rate": 0.0005, "loss": 1.1783925294876099, "step": 36622 }, { "epoch": 2.8871107607410327, "grad_norm": 0.49328091740608215, "learning_rate": 0.0005, "loss": 1.4191315174102783, "step": 36623 }, { "epoch": 2.887189594008672, "grad_norm": 0.45740067958831787, "learning_rate": 0.0005, "loss": 1.1747010946273804, "step": 36624 }, { "epoch": 2.8872684272763105, "grad_norm": 0.5377042293548584, "learning_rate": 0.0005, "loss": 1.246640682220459, "step": 36625 }, { "epoch": 2.8873472605439496, "grad_norm": 0.5591970086097717, "learning_rate": 0.0005, "loss": 1.3120824098587036, "step": 36626 }, { "epoch": 2.8874260938115883, "grad_norm": 0.46619731187820435, "learning_rate": 0.0005, "loss": 1.2639378309249878, "step": 36627 }, { "epoch": 2.8875049270792275, "grad_norm": 0.6204800605773926, "learning_rate": 0.0005, "loss": 1.3976459503173828, "step": 36628 }, { "epoch": 2.8875837603468666, "grad_norm": 0.502713680267334, "learning_rate": 0.0005, "loss": 1.157782793045044, "step": 36629 }, { "epoch": 2.8876625936145053, "grad_norm": 0.5726150870323181, "learning_rate": 0.0005, "loss": 1.4176301956176758, "step": 36630 }, { "epoch": 2.8877414268821444, "grad_norm": 0.4266541600227356, "learning_rate": 0.0005, "loss": 1.2421588897705078, "step": 36631 }, { "epoch": 2.887820260149783, "grad_norm": 0.4284127950668335, "learning_rate": 0.0005, "loss": 1.2162952423095703, "step": 36632 }, { "epoch": 2.8878990934174222, "grad_norm": 0.6360061764717102, "learning_rate": 0.0005, "loss": 1.3387787342071533, "step": 36633 }, { "epoch": 2.8879779266850614, "grad_norm": 0.556770384311676, "learning_rate": 0.0005, "loss": 1.5590592622756958, "step": 36634 }, { "epoch": 2.8880567599527, "grad_norm": 0.5618075728416443, "learning_rate": 0.0005, "loss": 1.2406015396118164, "step": 36635 }, { "epoch": 2.8881355932203387, "grad_norm": 0.5452706217765808, "learning_rate": 0.0005, "loss": 1.4733701944351196, "step": 36636 }, { "epoch": 2.888214426487978, "grad_norm": 0.7496389746665955, "learning_rate": 0.0005, "loss": 1.2727054357528687, "step": 36637 }, { "epoch": 2.888293259755617, "grad_norm": 0.536681056022644, "learning_rate": 0.0005, "loss": 1.3333282470703125, "step": 36638 }, { "epoch": 2.8883720930232557, "grad_norm": 0.5537366271018982, "learning_rate": 0.0005, "loss": 1.1702923774719238, "step": 36639 }, { "epoch": 2.888450926290895, "grad_norm": 0.5071996450424194, "learning_rate": 0.0005, "loss": 1.3774912357330322, "step": 36640 }, { "epoch": 2.8885297595585335, "grad_norm": 0.5115845203399658, "learning_rate": 0.0005, "loss": 1.2797291278839111, "step": 36641 }, { "epoch": 2.8886085928261727, "grad_norm": 0.550294816493988, "learning_rate": 0.0005, "loss": 1.3693984746932983, "step": 36642 }, { "epoch": 2.888687426093812, "grad_norm": 0.40867528319358826, "learning_rate": 0.0005, "loss": 1.2633917331695557, "step": 36643 }, { "epoch": 2.8887662593614505, "grad_norm": 0.4942854344844818, "learning_rate": 0.0005, "loss": 1.2863264083862305, "step": 36644 }, { "epoch": 2.8888450926290896, "grad_norm": 0.5891305208206177, "learning_rate": 0.0005, "loss": 1.7332415580749512, "step": 36645 }, { "epoch": 2.8889239258967283, "grad_norm": 0.652588963508606, "learning_rate": 0.0005, "loss": 1.4553210735321045, "step": 36646 }, { "epoch": 2.8890027591643674, "grad_norm": 0.49839988350868225, "learning_rate": 0.0005, "loss": 1.1935161352157593, "step": 36647 }, { "epoch": 2.8890815924320066, "grad_norm": 0.384211927652359, "learning_rate": 0.0005, "loss": 1.210332989692688, "step": 36648 }, { "epoch": 2.8891604256996453, "grad_norm": 0.47335031628608704, "learning_rate": 0.0005, "loss": 1.330106496810913, "step": 36649 }, { "epoch": 2.889239258967284, "grad_norm": 0.4754545986652374, "learning_rate": 0.0005, "loss": 1.358268141746521, "step": 36650 }, { "epoch": 2.889318092234923, "grad_norm": 0.4620341956615448, "learning_rate": 0.0005, "loss": 1.2679961919784546, "step": 36651 }, { "epoch": 2.889396925502562, "grad_norm": 0.4861159026622772, "learning_rate": 0.0005, "loss": 1.2152135372161865, "step": 36652 }, { "epoch": 2.889475758770201, "grad_norm": 0.4852167069911957, "learning_rate": 0.0005, "loss": 1.4649736881256104, "step": 36653 }, { "epoch": 2.88955459203784, "grad_norm": 0.7000812292098999, "learning_rate": 0.0005, "loss": 1.3027732372283936, "step": 36654 }, { "epoch": 2.8896334253054787, "grad_norm": 0.47545909881591797, "learning_rate": 0.0005, "loss": 1.4213765859603882, "step": 36655 }, { "epoch": 2.889712258573118, "grad_norm": 0.43679147958755493, "learning_rate": 0.0005, "loss": 1.2552562952041626, "step": 36656 }, { "epoch": 2.889791091840757, "grad_norm": 0.45219114422798157, "learning_rate": 0.0005, "loss": 1.2788759469985962, "step": 36657 }, { "epoch": 2.8898699251083957, "grad_norm": 0.5443932414054871, "learning_rate": 0.0005, "loss": 1.1852856874465942, "step": 36658 }, { "epoch": 2.889948758376035, "grad_norm": 0.549875020980835, "learning_rate": 0.0005, "loss": 1.2832789421081543, "step": 36659 }, { "epoch": 2.8900275916436735, "grad_norm": 0.49042338132858276, "learning_rate": 0.0005, "loss": 1.320341944694519, "step": 36660 }, { "epoch": 2.8901064249113126, "grad_norm": 0.6341888308525085, "learning_rate": 0.0005, "loss": 1.4798357486724854, "step": 36661 }, { "epoch": 2.8901852581789518, "grad_norm": 0.5399968028068542, "learning_rate": 0.0005, "loss": 1.406477928161621, "step": 36662 }, { "epoch": 2.8902640914465905, "grad_norm": 0.43819689750671387, "learning_rate": 0.0005, "loss": 1.4089276790618896, "step": 36663 }, { "epoch": 2.890342924714229, "grad_norm": 0.5068790912628174, "learning_rate": 0.0005, "loss": 1.1990835666656494, "step": 36664 }, { "epoch": 2.8904217579818683, "grad_norm": 0.5095681548118591, "learning_rate": 0.0005, "loss": 1.3473317623138428, "step": 36665 }, { "epoch": 2.8905005912495074, "grad_norm": 0.45995938777923584, "learning_rate": 0.0005, "loss": 1.1234157085418701, "step": 36666 }, { "epoch": 2.890579424517146, "grad_norm": 0.5199801325798035, "learning_rate": 0.0005, "loss": 1.4473507404327393, "step": 36667 }, { "epoch": 2.8906582577847852, "grad_norm": 0.6173021793365479, "learning_rate": 0.0005, "loss": 1.2894232273101807, "step": 36668 }, { "epoch": 2.890737091052424, "grad_norm": 0.5569318532943726, "learning_rate": 0.0005, "loss": 1.1856560707092285, "step": 36669 }, { "epoch": 2.890815924320063, "grad_norm": 0.4939688742160797, "learning_rate": 0.0005, "loss": 1.3460266590118408, "step": 36670 }, { "epoch": 2.890894757587702, "grad_norm": 0.635283350944519, "learning_rate": 0.0005, "loss": 1.399092674255371, "step": 36671 }, { "epoch": 2.890973590855341, "grad_norm": 0.46570560336112976, "learning_rate": 0.0005, "loss": 1.0990822315216064, "step": 36672 }, { "epoch": 2.89105242412298, "grad_norm": 0.4969158172607422, "learning_rate": 0.0005, "loss": 1.2675414085388184, "step": 36673 }, { "epoch": 2.8911312573906187, "grad_norm": 0.7520781755447388, "learning_rate": 0.0005, "loss": 1.3088469505310059, "step": 36674 }, { "epoch": 2.891210090658258, "grad_norm": 0.5690545439720154, "learning_rate": 0.0005, "loss": 1.2452881336212158, "step": 36675 }, { "epoch": 2.891288923925897, "grad_norm": 0.5207648873329163, "learning_rate": 0.0005, "loss": 1.362121343612671, "step": 36676 }, { "epoch": 2.8913677571935357, "grad_norm": 0.6014278531074524, "learning_rate": 0.0005, "loss": 1.4600608348846436, "step": 36677 }, { "epoch": 2.8914465904611744, "grad_norm": 0.5676657557487488, "learning_rate": 0.0005, "loss": 1.3546137809753418, "step": 36678 }, { "epoch": 2.8915254237288135, "grad_norm": 0.5947532653808594, "learning_rate": 0.0005, "loss": 1.2963954210281372, "step": 36679 }, { "epoch": 2.8916042569964526, "grad_norm": 0.4260953664779663, "learning_rate": 0.0005, "loss": 1.1589678525924683, "step": 36680 }, { "epoch": 2.8916830902640913, "grad_norm": 0.5594482421875, "learning_rate": 0.0005, "loss": 1.3562583923339844, "step": 36681 }, { "epoch": 2.8917619235317304, "grad_norm": 0.5110114812850952, "learning_rate": 0.0005, "loss": 1.370832920074463, "step": 36682 }, { "epoch": 2.891840756799369, "grad_norm": 0.49684104323387146, "learning_rate": 0.0005, "loss": 1.2606079578399658, "step": 36683 }, { "epoch": 2.8919195900670083, "grad_norm": 0.5273814797401428, "learning_rate": 0.0005, "loss": 1.2303540706634521, "step": 36684 }, { "epoch": 2.8919984233346474, "grad_norm": 0.4776911437511444, "learning_rate": 0.0005, "loss": 1.2918360233306885, "step": 36685 }, { "epoch": 2.892077256602286, "grad_norm": 0.45900487899780273, "learning_rate": 0.0005, "loss": 1.3325746059417725, "step": 36686 }, { "epoch": 2.8921560898699252, "grad_norm": 0.5004479289054871, "learning_rate": 0.0005, "loss": 1.3582656383514404, "step": 36687 }, { "epoch": 2.892234923137564, "grad_norm": 0.5675373673439026, "learning_rate": 0.0005, "loss": 1.2713134288787842, "step": 36688 }, { "epoch": 2.892313756405203, "grad_norm": 0.4524845778942108, "learning_rate": 0.0005, "loss": 1.6005196571350098, "step": 36689 }, { "epoch": 2.892392589672842, "grad_norm": 0.5329134464263916, "learning_rate": 0.0005, "loss": 1.3825722932815552, "step": 36690 }, { "epoch": 2.892471422940481, "grad_norm": 0.4899224638938904, "learning_rate": 0.0005, "loss": 1.4530985355377197, "step": 36691 }, { "epoch": 2.8925502562081196, "grad_norm": 0.47879907488822937, "learning_rate": 0.0005, "loss": 1.2853468656539917, "step": 36692 }, { "epoch": 2.8926290894757587, "grad_norm": 0.49917903542518616, "learning_rate": 0.0005, "loss": 1.259829044342041, "step": 36693 }, { "epoch": 2.892707922743398, "grad_norm": 0.4551365375518799, "learning_rate": 0.0005, "loss": 1.425331473350525, "step": 36694 }, { "epoch": 2.8927867560110365, "grad_norm": 0.6150899529457092, "learning_rate": 0.0005, "loss": 1.3480827808380127, "step": 36695 }, { "epoch": 2.8928655892786757, "grad_norm": 0.820390522480011, "learning_rate": 0.0005, "loss": 1.3410499095916748, "step": 36696 }, { "epoch": 2.8929444225463143, "grad_norm": 0.4912688136100769, "learning_rate": 0.0005, "loss": 1.2151750326156616, "step": 36697 }, { "epoch": 2.8930232558139535, "grad_norm": 0.5761464834213257, "learning_rate": 0.0005, "loss": 1.5012969970703125, "step": 36698 }, { "epoch": 2.8931020890815926, "grad_norm": 0.5400691032409668, "learning_rate": 0.0005, "loss": 1.2553062438964844, "step": 36699 }, { "epoch": 2.8931809223492313, "grad_norm": 0.45898696780204773, "learning_rate": 0.0005, "loss": 1.4189748764038086, "step": 36700 }, { "epoch": 2.8932597556168704, "grad_norm": 0.6329644322395325, "learning_rate": 0.0005, "loss": 1.5335184335708618, "step": 36701 }, { "epoch": 2.893338588884509, "grad_norm": 0.6248409748077393, "learning_rate": 0.0005, "loss": 1.187149167060852, "step": 36702 }, { "epoch": 2.8934174221521483, "grad_norm": 0.4898947775363922, "learning_rate": 0.0005, "loss": 1.2573318481445312, "step": 36703 }, { "epoch": 2.8934962554197874, "grad_norm": 0.49758338928222656, "learning_rate": 0.0005, "loss": 1.4485217332839966, "step": 36704 }, { "epoch": 2.893575088687426, "grad_norm": 0.4529496431350708, "learning_rate": 0.0005, "loss": 1.1734235286712646, "step": 36705 }, { "epoch": 2.8936539219550648, "grad_norm": 0.5621219277381897, "learning_rate": 0.0005, "loss": 1.5433650016784668, "step": 36706 }, { "epoch": 2.893732755222704, "grad_norm": 0.5525111556053162, "learning_rate": 0.0005, "loss": 1.380750298500061, "step": 36707 }, { "epoch": 2.893811588490343, "grad_norm": 0.4579261243343353, "learning_rate": 0.0005, "loss": 1.2293230295181274, "step": 36708 }, { "epoch": 2.8938904217579817, "grad_norm": 0.452323317527771, "learning_rate": 0.0005, "loss": 1.1878979206085205, "step": 36709 }, { "epoch": 2.893969255025621, "grad_norm": 0.48650625348091125, "learning_rate": 0.0005, "loss": 1.3574036359786987, "step": 36710 }, { "epoch": 2.8940480882932595, "grad_norm": 0.5015636086463928, "learning_rate": 0.0005, "loss": 1.3138461112976074, "step": 36711 }, { "epoch": 2.8941269215608987, "grad_norm": 0.44910597801208496, "learning_rate": 0.0005, "loss": 1.4505512714385986, "step": 36712 }, { "epoch": 2.894205754828538, "grad_norm": 0.5515457391738892, "learning_rate": 0.0005, "loss": 1.3395183086395264, "step": 36713 }, { "epoch": 2.8942845880961765, "grad_norm": 0.40720364451408386, "learning_rate": 0.0005, "loss": 1.20424222946167, "step": 36714 }, { "epoch": 2.8943634213638156, "grad_norm": 0.5581760406494141, "learning_rate": 0.0005, "loss": 1.4795511960983276, "step": 36715 }, { "epoch": 2.8944422546314543, "grad_norm": 0.4635666608810425, "learning_rate": 0.0005, "loss": 1.21199369430542, "step": 36716 }, { "epoch": 2.8945210878990935, "grad_norm": 0.4066263735294342, "learning_rate": 0.0005, "loss": 1.1961109638214111, "step": 36717 }, { "epoch": 2.8945999211667326, "grad_norm": 0.6529095768928528, "learning_rate": 0.0005, "loss": 1.2652692794799805, "step": 36718 }, { "epoch": 2.8946787544343713, "grad_norm": 0.4602857232093811, "learning_rate": 0.0005, "loss": 1.1292378902435303, "step": 36719 }, { "epoch": 2.89475758770201, "grad_norm": 0.38684892654418945, "learning_rate": 0.0005, "loss": 1.1925948858261108, "step": 36720 }, { "epoch": 2.894836420969649, "grad_norm": 0.5517552495002747, "learning_rate": 0.0005, "loss": 1.3558080196380615, "step": 36721 }, { "epoch": 2.8949152542372882, "grad_norm": 0.41504526138305664, "learning_rate": 0.0005, "loss": 1.336673378944397, "step": 36722 }, { "epoch": 2.894994087504927, "grad_norm": 0.4466477930545807, "learning_rate": 0.0005, "loss": 1.2620518207550049, "step": 36723 }, { "epoch": 2.895072920772566, "grad_norm": 0.49059340357780457, "learning_rate": 0.0005, "loss": 1.3033088445663452, "step": 36724 }, { "epoch": 2.8951517540402048, "grad_norm": 0.46703845262527466, "learning_rate": 0.0005, "loss": 1.1717168092727661, "step": 36725 }, { "epoch": 2.895230587307844, "grad_norm": 0.4252447485923767, "learning_rate": 0.0005, "loss": 1.1568431854248047, "step": 36726 }, { "epoch": 2.895309420575483, "grad_norm": 0.46573346853256226, "learning_rate": 0.0005, "loss": 1.3144564628601074, "step": 36727 }, { "epoch": 2.8953882538431217, "grad_norm": 0.527675211429596, "learning_rate": 0.0005, "loss": 1.2957504987716675, "step": 36728 }, { "epoch": 2.895467087110761, "grad_norm": 0.6018602252006531, "learning_rate": 0.0005, "loss": 1.451040267944336, "step": 36729 }, { "epoch": 2.8955459203783995, "grad_norm": 0.40227603912353516, "learning_rate": 0.0005, "loss": 1.1167562007904053, "step": 36730 }, { "epoch": 2.8956247536460387, "grad_norm": 0.44111397862434387, "learning_rate": 0.0005, "loss": 1.2232518196105957, "step": 36731 }, { "epoch": 2.895703586913678, "grad_norm": 0.5086554884910583, "learning_rate": 0.0005, "loss": 1.411014437675476, "step": 36732 }, { "epoch": 2.8957824201813165, "grad_norm": 0.5509381294250488, "learning_rate": 0.0005, "loss": 1.3404784202575684, "step": 36733 }, { "epoch": 2.8958612534489556, "grad_norm": 0.41325268149375916, "learning_rate": 0.0005, "loss": 1.3191702365875244, "step": 36734 }, { "epoch": 2.8959400867165943, "grad_norm": 0.43384307622909546, "learning_rate": 0.0005, "loss": 1.2003467082977295, "step": 36735 }, { "epoch": 2.8960189199842334, "grad_norm": 0.585942804813385, "learning_rate": 0.0005, "loss": 1.5461812019348145, "step": 36736 }, { "epoch": 2.896097753251872, "grad_norm": 0.4901870787143707, "learning_rate": 0.0005, "loss": 1.1481542587280273, "step": 36737 }, { "epoch": 2.8961765865195113, "grad_norm": 0.5214243531227112, "learning_rate": 0.0005, "loss": 1.4013404846191406, "step": 36738 }, { "epoch": 2.89625541978715, "grad_norm": 0.5703760981559753, "learning_rate": 0.0005, "loss": 1.3199185132980347, "step": 36739 }, { "epoch": 2.896334253054789, "grad_norm": 0.4895687401294708, "learning_rate": 0.0005, "loss": 1.3940768241882324, "step": 36740 }, { "epoch": 2.896413086322428, "grad_norm": 0.46350759267807007, "learning_rate": 0.0005, "loss": 1.293965220451355, "step": 36741 }, { "epoch": 2.896491919590067, "grad_norm": 0.4743063747882843, "learning_rate": 0.0005, "loss": 1.1780261993408203, "step": 36742 }, { "epoch": 2.896570752857706, "grad_norm": 0.6975225210189819, "learning_rate": 0.0005, "loss": 1.3273730278015137, "step": 36743 }, { "epoch": 2.8966495861253447, "grad_norm": 0.47583600878715515, "learning_rate": 0.0005, "loss": 1.5434019565582275, "step": 36744 }, { "epoch": 2.896728419392984, "grad_norm": 0.4670291543006897, "learning_rate": 0.0005, "loss": 1.2407971620559692, "step": 36745 }, { "epoch": 2.896807252660623, "grad_norm": 0.4789283275604248, "learning_rate": 0.0005, "loss": 1.2851336002349854, "step": 36746 }, { "epoch": 2.8968860859282617, "grad_norm": 0.5325680375099182, "learning_rate": 0.0005, "loss": 1.4048314094543457, "step": 36747 }, { "epoch": 2.896964919195901, "grad_norm": 0.48849475383758545, "learning_rate": 0.0005, "loss": 1.3833730220794678, "step": 36748 }, { "epoch": 2.8970437524635395, "grad_norm": 0.4892805516719818, "learning_rate": 0.0005, "loss": 1.5803611278533936, "step": 36749 }, { "epoch": 2.8971225857311786, "grad_norm": 0.5399143695831299, "learning_rate": 0.0005, "loss": 1.3188152313232422, "step": 36750 }, { "epoch": 2.8972014189988173, "grad_norm": 0.5093458294868469, "learning_rate": 0.0005, "loss": 1.2219609022140503, "step": 36751 }, { "epoch": 2.8972802522664565, "grad_norm": 0.54213947057724, "learning_rate": 0.0005, "loss": 1.3184120655059814, "step": 36752 }, { "epoch": 2.897359085534095, "grad_norm": 0.6416807770729065, "learning_rate": 0.0005, "loss": 1.3956072330474854, "step": 36753 }, { "epoch": 2.8974379188017343, "grad_norm": 0.616252601146698, "learning_rate": 0.0005, "loss": 1.374658465385437, "step": 36754 }, { "epoch": 2.8975167520693734, "grad_norm": 0.4230390787124634, "learning_rate": 0.0005, "loss": 1.2576699256896973, "step": 36755 }, { "epoch": 2.897595585337012, "grad_norm": 0.4093489646911621, "learning_rate": 0.0005, "loss": 1.3429871797561646, "step": 36756 }, { "epoch": 2.8976744186046512, "grad_norm": 0.5828458070755005, "learning_rate": 0.0005, "loss": 1.4076625108718872, "step": 36757 }, { "epoch": 2.89775325187229, "grad_norm": 0.48881909251213074, "learning_rate": 0.0005, "loss": 1.28758704662323, "step": 36758 }, { "epoch": 2.897832085139929, "grad_norm": 0.5428057909011841, "learning_rate": 0.0005, "loss": 1.1639113426208496, "step": 36759 }, { "epoch": 2.897910918407568, "grad_norm": 0.5046159029006958, "learning_rate": 0.0005, "loss": 1.4768900871276855, "step": 36760 }, { "epoch": 2.897989751675207, "grad_norm": 0.5460267066955566, "learning_rate": 0.0005, "loss": 1.3170827627182007, "step": 36761 }, { "epoch": 2.898068584942846, "grad_norm": 0.46296221017837524, "learning_rate": 0.0005, "loss": 1.3767943382263184, "step": 36762 }, { "epoch": 2.8981474182104847, "grad_norm": 0.6035960912704468, "learning_rate": 0.0005, "loss": 1.6327623128890991, "step": 36763 }, { "epoch": 2.898226251478124, "grad_norm": 0.5053081512451172, "learning_rate": 0.0005, "loss": 1.286207675933838, "step": 36764 }, { "epoch": 2.898305084745763, "grad_norm": 0.43283766508102417, "learning_rate": 0.0005, "loss": 1.2825462818145752, "step": 36765 }, { "epoch": 2.8983839180134017, "grad_norm": 0.49128592014312744, "learning_rate": 0.0005, "loss": 1.2733232975006104, "step": 36766 }, { "epoch": 2.8984627512810404, "grad_norm": 0.5010495185852051, "learning_rate": 0.0005, "loss": 1.2415095567703247, "step": 36767 }, { "epoch": 2.8985415845486795, "grad_norm": 0.42941930890083313, "learning_rate": 0.0005, "loss": 1.3755369186401367, "step": 36768 }, { "epoch": 2.8986204178163186, "grad_norm": 0.5234240889549255, "learning_rate": 0.0005, "loss": 1.2094676494598389, "step": 36769 }, { "epoch": 2.8986992510839573, "grad_norm": 0.5597453713417053, "learning_rate": 0.0005, "loss": 1.4264278411865234, "step": 36770 }, { "epoch": 2.8987780843515965, "grad_norm": 0.505950927734375, "learning_rate": 0.0005, "loss": 1.2135320901870728, "step": 36771 }, { "epoch": 2.898856917619235, "grad_norm": 0.40839654207229614, "learning_rate": 0.0005, "loss": 1.0666818618774414, "step": 36772 }, { "epoch": 2.8989357508868743, "grad_norm": 0.4085356295108795, "learning_rate": 0.0005, "loss": 1.3278868198394775, "step": 36773 }, { "epoch": 2.8990145841545134, "grad_norm": 0.7036204934120178, "learning_rate": 0.0005, "loss": 1.322811484336853, "step": 36774 }, { "epoch": 2.899093417422152, "grad_norm": 0.4808635115623474, "learning_rate": 0.0005, "loss": 1.3902359008789062, "step": 36775 }, { "epoch": 2.8991722506897912, "grad_norm": 0.5694419145584106, "learning_rate": 0.0005, "loss": 1.3664884567260742, "step": 36776 }, { "epoch": 2.89925108395743, "grad_norm": 0.426053911447525, "learning_rate": 0.0005, "loss": 1.388908863067627, "step": 36777 }, { "epoch": 2.899329917225069, "grad_norm": 0.46942853927612305, "learning_rate": 0.0005, "loss": 1.4461216926574707, "step": 36778 }, { "epoch": 2.899408750492708, "grad_norm": 11.694939613342285, "learning_rate": 0.0005, "loss": 1.4291036128997803, "step": 36779 }, { "epoch": 2.899487583760347, "grad_norm": 0.5051568150520325, "learning_rate": 0.0005, "loss": 1.2712657451629639, "step": 36780 }, { "epoch": 2.8995664170279856, "grad_norm": 0.460305392742157, "learning_rate": 0.0005, "loss": 1.3159327507019043, "step": 36781 }, { "epoch": 2.8996452502956247, "grad_norm": 0.6185666918754578, "learning_rate": 0.0005, "loss": 1.3986331224441528, "step": 36782 }, { "epoch": 2.899724083563264, "grad_norm": 0.4167322814464569, "learning_rate": 0.0005, "loss": 1.0609657764434814, "step": 36783 }, { "epoch": 2.8998029168309025, "grad_norm": 0.6071489453315735, "learning_rate": 0.0005, "loss": 1.2950794696807861, "step": 36784 }, { "epoch": 2.8998817500985417, "grad_norm": 0.4451110363006592, "learning_rate": 0.0005, "loss": 1.3441550731658936, "step": 36785 }, { "epoch": 2.8999605833661803, "grad_norm": 0.45432472229003906, "learning_rate": 0.0005, "loss": 1.2947885990142822, "step": 36786 }, { "epoch": 2.9000394166338195, "grad_norm": 0.47366970777511597, "learning_rate": 0.0005, "loss": 1.2349305152893066, "step": 36787 }, { "epoch": 2.9001182499014586, "grad_norm": 0.6231240034103394, "learning_rate": 0.0005, "loss": 1.4912769794464111, "step": 36788 }, { "epoch": 2.9001970831690973, "grad_norm": 0.5509688854217529, "learning_rate": 0.0005, "loss": 1.3574941158294678, "step": 36789 }, { "epoch": 2.9002759164367364, "grad_norm": 0.5053091049194336, "learning_rate": 0.0005, "loss": 1.1222865581512451, "step": 36790 }, { "epoch": 2.900354749704375, "grad_norm": 0.4735844135284424, "learning_rate": 0.0005, "loss": 1.2547708749771118, "step": 36791 }, { "epoch": 2.9004335829720143, "grad_norm": 0.5082516074180603, "learning_rate": 0.0005, "loss": 1.4528582096099854, "step": 36792 }, { "epoch": 2.9005124162396534, "grad_norm": 0.45336100459098816, "learning_rate": 0.0005, "loss": 1.2139558792114258, "step": 36793 }, { "epoch": 2.900591249507292, "grad_norm": 0.46317705512046814, "learning_rate": 0.0005, "loss": 1.1317209005355835, "step": 36794 }, { "epoch": 2.9006700827749308, "grad_norm": 0.42370346188545227, "learning_rate": 0.0005, "loss": 1.4371975660324097, "step": 36795 }, { "epoch": 2.90074891604257, "grad_norm": 0.49981528520584106, "learning_rate": 0.0005, "loss": 1.474593162536621, "step": 36796 }, { "epoch": 2.900827749310209, "grad_norm": 0.553472101688385, "learning_rate": 0.0005, "loss": 1.37880539894104, "step": 36797 }, { "epoch": 2.9009065825778477, "grad_norm": 0.5027175545692444, "learning_rate": 0.0005, "loss": 1.351588487625122, "step": 36798 }, { "epoch": 2.900985415845487, "grad_norm": 0.5253069996833801, "learning_rate": 0.0005, "loss": 1.1284818649291992, "step": 36799 }, { "epoch": 2.9010642491131255, "grad_norm": 0.7788543105125427, "learning_rate": 0.0005, "loss": 1.5494321584701538, "step": 36800 }, { "epoch": 2.9011430823807647, "grad_norm": 0.5787361264228821, "learning_rate": 0.0005, "loss": 1.3140846490859985, "step": 36801 }, { "epoch": 2.901221915648404, "grad_norm": 0.4987128674983978, "learning_rate": 0.0005, "loss": 1.3075947761535645, "step": 36802 }, { "epoch": 2.9013007489160425, "grad_norm": 0.49151429533958435, "learning_rate": 0.0005, "loss": 1.3489675521850586, "step": 36803 }, { "epoch": 2.9013795821836816, "grad_norm": 0.4540206789970398, "learning_rate": 0.0005, "loss": 1.172823429107666, "step": 36804 }, { "epoch": 2.9014584154513203, "grad_norm": 0.6013261079788208, "learning_rate": 0.0005, "loss": 1.2793231010437012, "step": 36805 }, { "epoch": 2.9015372487189595, "grad_norm": 0.4671464264392853, "learning_rate": 0.0005, "loss": 1.2946029901504517, "step": 36806 }, { "epoch": 2.9016160819865986, "grad_norm": 0.5369325876235962, "learning_rate": 0.0005, "loss": 1.264878749847412, "step": 36807 }, { "epoch": 2.9016949152542373, "grad_norm": 0.43141305446624756, "learning_rate": 0.0005, "loss": 1.1342965364456177, "step": 36808 }, { "epoch": 2.901773748521876, "grad_norm": 0.405282199382782, "learning_rate": 0.0005, "loss": 1.1282179355621338, "step": 36809 }, { "epoch": 2.901852581789515, "grad_norm": 0.5218422412872314, "learning_rate": 0.0005, "loss": 1.3298006057739258, "step": 36810 }, { "epoch": 2.9019314150571542, "grad_norm": 0.5451775193214417, "learning_rate": 0.0005, "loss": 1.473661184310913, "step": 36811 }, { "epoch": 2.902010248324793, "grad_norm": 0.6710933446884155, "learning_rate": 0.0005, "loss": 1.4373286962509155, "step": 36812 }, { "epoch": 2.902089081592432, "grad_norm": 0.4224764406681061, "learning_rate": 0.0005, "loss": 1.1883533000946045, "step": 36813 }, { "epoch": 2.9021679148600708, "grad_norm": 0.5976907014846802, "learning_rate": 0.0005, "loss": 1.4707016944885254, "step": 36814 }, { "epoch": 2.90224674812771, "grad_norm": 0.47029197216033936, "learning_rate": 0.0005, "loss": 1.2969329357147217, "step": 36815 }, { "epoch": 2.902325581395349, "grad_norm": 0.5608729124069214, "learning_rate": 0.0005, "loss": 1.3762075901031494, "step": 36816 }, { "epoch": 2.9024044146629877, "grad_norm": 0.4488527774810791, "learning_rate": 0.0005, "loss": 1.216983675956726, "step": 36817 }, { "epoch": 2.902483247930627, "grad_norm": 0.4964308738708496, "learning_rate": 0.0005, "loss": 1.3745923042297363, "step": 36818 }, { "epoch": 2.9025620811982655, "grad_norm": 0.5704028010368347, "learning_rate": 0.0005, "loss": 1.2779288291931152, "step": 36819 }, { "epoch": 2.9026409144659047, "grad_norm": 2.6035006046295166, "learning_rate": 0.0005, "loss": 1.2261178493499756, "step": 36820 }, { "epoch": 2.902719747733544, "grad_norm": 0.5083122849464417, "learning_rate": 0.0005, "loss": 1.313826322555542, "step": 36821 }, { "epoch": 2.9027985810011825, "grad_norm": 0.5184966325759888, "learning_rate": 0.0005, "loss": 1.4673044681549072, "step": 36822 }, { "epoch": 2.902877414268821, "grad_norm": 0.4577902555465698, "learning_rate": 0.0005, "loss": 1.2577379941940308, "step": 36823 }, { "epoch": 2.9029562475364603, "grad_norm": 0.45798394083976746, "learning_rate": 0.0005, "loss": 1.2239569425582886, "step": 36824 }, { "epoch": 2.9030350808040994, "grad_norm": 0.46796518564224243, "learning_rate": 0.0005, "loss": 1.2840497493743896, "step": 36825 }, { "epoch": 2.903113914071738, "grad_norm": 0.5849010944366455, "learning_rate": 0.0005, "loss": 1.3421505689620972, "step": 36826 }, { "epoch": 2.9031927473393773, "grad_norm": 0.500816285610199, "learning_rate": 0.0005, "loss": 1.373242735862732, "step": 36827 }, { "epoch": 2.903271580607016, "grad_norm": 0.5756186246871948, "learning_rate": 0.0005, "loss": 1.3892292976379395, "step": 36828 }, { "epoch": 2.903350413874655, "grad_norm": 0.5162991285324097, "learning_rate": 0.0005, "loss": 1.283419132232666, "step": 36829 }, { "epoch": 2.9034292471422942, "grad_norm": 0.5454946756362915, "learning_rate": 0.0005, "loss": 1.3662002086639404, "step": 36830 }, { "epoch": 2.903508080409933, "grad_norm": 0.6178691387176514, "learning_rate": 0.0005, "loss": 1.2519268989562988, "step": 36831 }, { "epoch": 2.903586913677572, "grad_norm": 0.4194605350494385, "learning_rate": 0.0005, "loss": 1.11703622341156, "step": 36832 }, { "epoch": 2.9036657469452107, "grad_norm": 0.4867772161960602, "learning_rate": 0.0005, "loss": 1.3065712451934814, "step": 36833 }, { "epoch": 2.90374458021285, "grad_norm": 0.5890476703643799, "learning_rate": 0.0005, "loss": 1.3907567262649536, "step": 36834 }, { "epoch": 2.903823413480489, "grad_norm": 0.5057650804519653, "learning_rate": 0.0005, "loss": 1.2899993658065796, "step": 36835 }, { "epoch": 2.9039022467481277, "grad_norm": 0.7228745222091675, "learning_rate": 0.0005, "loss": 1.5463531017303467, "step": 36836 }, { "epoch": 2.9039810800157664, "grad_norm": 0.3645826280117035, "learning_rate": 0.0005, "loss": 1.0987279415130615, "step": 36837 }, { "epoch": 2.9040599132834055, "grad_norm": 0.4705654978752136, "learning_rate": 0.0005, "loss": 1.3329952955245972, "step": 36838 }, { "epoch": 2.9041387465510446, "grad_norm": 0.4826219081878662, "learning_rate": 0.0005, "loss": 1.2416729927062988, "step": 36839 }, { "epoch": 2.9042175798186833, "grad_norm": 0.4483441412448883, "learning_rate": 0.0005, "loss": 1.2880961894989014, "step": 36840 }, { "epoch": 2.9042964130863225, "grad_norm": 0.4791613221168518, "learning_rate": 0.0005, "loss": 1.3440179824829102, "step": 36841 }, { "epoch": 2.904375246353961, "grad_norm": 0.45240455865859985, "learning_rate": 0.0005, "loss": 1.2224986553192139, "step": 36842 }, { "epoch": 2.9044540796216003, "grad_norm": 0.568873405456543, "learning_rate": 0.0005, "loss": 1.299743413925171, "step": 36843 }, { "epoch": 2.9045329128892394, "grad_norm": 0.4110872447490692, "learning_rate": 0.0005, "loss": 1.0199389457702637, "step": 36844 }, { "epoch": 2.904611746156878, "grad_norm": 0.539637565612793, "learning_rate": 0.0005, "loss": 1.350076675415039, "step": 36845 }, { "epoch": 2.9046905794245172, "grad_norm": 0.5436388254165649, "learning_rate": 0.0005, "loss": 1.3257602453231812, "step": 36846 }, { "epoch": 2.904769412692156, "grad_norm": 0.7335741519927979, "learning_rate": 0.0005, "loss": 1.4307295083999634, "step": 36847 }, { "epoch": 2.904848245959795, "grad_norm": 0.45329606533050537, "learning_rate": 0.0005, "loss": 1.1404972076416016, "step": 36848 }, { "epoch": 2.904927079227434, "grad_norm": 0.6032395958900452, "learning_rate": 0.0005, "loss": 1.3807969093322754, "step": 36849 }, { "epoch": 2.905005912495073, "grad_norm": 0.42986297607421875, "learning_rate": 0.0005, "loss": 1.1330084800720215, "step": 36850 }, { "epoch": 2.905084745762712, "grad_norm": 0.5610033869743347, "learning_rate": 0.0005, "loss": 1.5124292373657227, "step": 36851 }, { "epoch": 2.9051635790303507, "grad_norm": 0.4787522256374359, "learning_rate": 0.0005, "loss": 1.2319486141204834, "step": 36852 }, { "epoch": 2.90524241229799, "grad_norm": 0.5805981755256653, "learning_rate": 0.0005, "loss": 1.3270442485809326, "step": 36853 }, { "epoch": 2.9053212455656285, "grad_norm": 0.5196069478988647, "learning_rate": 0.0005, "loss": 1.4614890813827515, "step": 36854 }, { "epoch": 2.9054000788332677, "grad_norm": 0.5169482827186584, "learning_rate": 0.0005, "loss": 1.1924915313720703, "step": 36855 }, { "epoch": 2.9054789121009064, "grad_norm": 0.5696468949317932, "learning_rate": 0.0005, "loss": 1.187664270401001, "step": 36856 }, { "epoch": 2.9055577453685455, "grad_norm": 0.5007427334785461, "learning_rate": 0.0005, "loss": 1.3520402908325195, "step": 36857 }, { "epoch": 2.9056365786361846, "grad_norm": 0.5516456961631775, "learning_rate": 0.0005, "loss": 1.4433860778808594, "step": 36858 }, { "epoch": 2.9057154119038233, "grad_norm": 0.5067703127861023, "learning_rate": 0.0005, "loss": 1.2382221221923828, "step": 36859 }, { "epoch": 2.9057942451714625, "grad_norm": 0.536629319190979, "learning_rate": 0.0005, "loss": 1.513944149017334, "step": 36860 }, { "epoch": 2.905873078439101, "grad_norm": 0.5401824712753296, "learning_rate": 0.0005, "loss": 1.246087908744812, "step": 36861 }, { "epoch": 2.9059519117067403, "grad_norm": 0.5476600527763367, "learning_rate": 0.0005, "loss": 1.4476914405822754, "step": 36862 }, { "epoch": 2.9060307449743794, "grad_norm": 0.4756682217121124, "learning_rate": 0.0005, "loss": 1.3331724405288696, "step": 36863 }, { "epoch": 2.906109578242018, "grad_norm": 0.548876941204071, "learning_rate": 0.0005, "loss": 1.3349661827087402, "step": 36864 }, { "epoch": 2.9061884115096572, "grad_norm": 0.5598335862159729, "learning_rate": 0.0005, "loss": 1.382224202156067, "step": 36865 }, { "epoch": 2.906267244777296, "grad_norm": 0.5152824521064758, "learning_rate": 0.0005, "loss": 1.3706045150756836, "step": 36866 }, { "epoch": 2.906346078044935, "grad_norm": 0.5364512205123901, "learning_rate": 0.0005, "loss": 1.404198408126831, "step": 36867 }, { "epoch": 2.9064249113125737, "grad_norm": 0.5211039185523987, "learning_rate": 0.0005, "loss": 1.4866876602172852, "step": 36868 }, { "epoch": 2.906503744580213, "grad_norm": 0.47718870639801025, "learning_rate": 0.0005, "loss": 1.1785593032836914, "step": 36869 }, { "epoch": 2.9065825778478516, "grad_norm": 0.4344758987426758, "learning_rate": 0.0005, "loss": 1.2092057466506958, "step": 36870 }, { "epoch": 2.9066614111154907, "grad_norm": 0.6199631094932556, "learning_rate": 0.0005, "loss": 1.2585867643356323, "step": 36871 }, { "epoch": 2.90674024438313, "grad_norm": 0.490000456571579, "learning_rate": 0.0005, "loss": 1.3021049499511719, "step": 36872 }, { "epoch": 2.9068190776507685, "grad_norm": 0.4341009855270386, "learning_rate": 0.0005, "loss": 1.240456223487854, "step": 36873 }, { "epoch": 2.9068979109184077, "grad_norm": 0.6012352705001831, "learning_rate": 0.0005, "loss": 1.4791661500930786, "step": 36874 }, { "epoch": 2.9069767441860463, "grad_norm": 0.643657922744751, "learning_rate": 0.0005, "loss": 1.4461277723312378, "step": 36875 }, { "epoch": 2.9070555774536855, "grad_norm": 0.5390598773956299, "learning_rate": 0.0005, "loss": 1.2582714557647705, "step": 36876 }, { "epoch": 2.9071344107213246, "grad_norm": 0.5304669141769409, "learning_rate": 0.0005, "loss": 1.2162365913391113, "step": 36877 }, { "epoch": 2.9072132439889633, "grad_norm": 0.5404486060142517, "learning_rate": 0.0005, "loss": 1.2946406602859497, "step": 36878 }, { "epoch": 2.9072920772566024, "grad_norm": 0.44909578561782837, "learning_rate": 0.0005, "loss": 1.140648603439331, "step": 36879 }, { "epoch": 2.907370910524241, "grad_norm": 0.5113206505775452, "learning_rate": 0.0005, "loss": 1.2612228393554688, "step": 36880 }, { "epoch": 2.9074497437918803, "grad_norm": 0.5619951486587524, "learning_rate": 0.0005, "loss": 1.3662995100021362, "step": 36881 }, { "epoch": 2.9075285770595194, "grad_norm": 0.47597911953926086, "learning_rate": 0.0005, "loss": 1.3707349300384521, "step": 36882 }, { "epoch": 2.907607410327158, "grad_norm": 0.5642213821411133, "learning_rate": 0.0005, "loss": 1.4718259572982788, "step": 36883 }, { "epoch": 2.9076862435947968, "grad_norm": 0.5193398594856262, "learning_rate": 0.0005, "loss": 1.2728500366210938, "step": 36884 }, { "epoch": 2.907765076862436, "grad_norm": 0.6732064485549927, "learning_rate": 0.0005, "loss": 1.6674267053604126, "step": 36885 }, { "epoch": 2.907843910130075, "grad_norm": 0.37093421816825867, "learning_rate": 0.0005, "loss": 1.0789014101028442, "step": 36886 }, { "epoch": 2.9079227433977137, "grad_norm": 0.5113570094108582, "learning_rate": 0.0005, "loss": 1.2429125308990479, "step": 36887 }, { "epoch": 2.908001576665353, "grad_norm": 0.511590301990509, "learning_rate": 0.0005, "loss": 1.1941978931427002, "step": 36888 }, { "epoch": 2.9080804099329916, "grad_norm": 0.578389585018158, "learning_rate": 0.0005, "loss": 1.2327163219451904, "step": 36889 }, { "epoch": 2.9081592432006307, "grad_norm": 0.47967520356178284, "learning_rate": 0.0005, "loss": 1.383657455444336, "step": 36890 }, { "epoch": 2.90823807646827, "grad_norm": 0.5332105755805969, "learning_rate": 0.0005, "loss": 1.5180580615997314, "step": 36891 }, { "epoch": 2.9083169097359085, "grad_norm": 0.5633525848388672, "learning_rate": 0.0005, "loss": 1.4314525127410889, "step": 36892 }, { "epoch": 2.9083957430035476, "grad_norm": 0.472918838262558, "learning_rate": 0.0005, "loss": 1.4547393321990967, "step": 36893 }, { "epoch": 2.9084745762711863, "grad_norm": 0.5378448367118835, "learning_rate": 0.0005, "loss": 1.3973791599273682, "step": 36894 }, { "epoch": 2.9085534095388255, "grad_norm": 0.5074716806411743, "learning_rate": 0.0005, "loss": 1.1499539613723755, "step": 36895 }, { "epoch": 2.9086322428064646, "grad_norm": 0.48822909593582153, "learning_rate": 0.0005, "loss": 1.4418070316314697, "step": 36896 }, { "epoch": 2.9087110760741033, "grad_norm": 0.6103960871696472, "learning_rate": 0.0005, "loss": 1.3535436391830444, "step": 36897 }, { "epoch": 2.908789909341742, "grad_norm": 0.582953691482544, "learning_rate": 0.0005, "loss": 1.3718751668930054, "step": 36898 }, { "epoch": 2.908868742609381, "grad_norm": 0.42615842819213867, "learning_rate": 0.0005, "loss": 1.2501366138458252, "step": 36899 }, { "epoch": 2.9089475758770202, "grad_norm": 0.5205328464508057, "learning_rate": 0.0005, "loss": 1.4050803184509277, "step": 36900 }, { "epoch": 2.909026409144659, "grad_norm": 0.4602876603603363, "learning_rate": 0.0005, "loss": 1.0243679285049438, "step": 36901 }, { "epoch": 2.909105242412298, "grad_norm": 0.5279074907302856, "learning_rate": 0.0005, "loss": 1.4532389640808105, "step": 36902 }, { "epoch": 2.9091840756799368, "grad_norm": 0.5173662900924683, "learning_rate": 0.0005, "loss": 1.1823463439941406, "step": 36903 }, { "epoch": 2.909262908947576, "grad_norm": 0.43749842047691345, "learning_rate": 0.0005, "loss": 1.409293532371521, "step": 36904 }, { "epoch": 2.909341742215215, "grad_norm": 0.7937314510345459, "learning_rate": 0.0005, "loss": 1.3623327016830444, "step": 36905 }, { "epoch": 2.9094205754828537, "grad_norm": 0.38382789492607117, "learning_rate": 0.0005, "loss": 1.0575048923492432, "step": 36906 }, { "epoch": 2.909499408750493, "grad_norm": 0.4824194312095642, "learning_rate": 0.0005, "loss": 1.279688835144043, "step": 36907 }, { "epoch": 2.9095782420181315, "grad_norm": 0.46138229966163635, "learning_rate": 0.0005, "loss": 1.3783776760101318, "step": 36908 }, { "epoch": 2.9096570752857707, "grad_norm": 0.5707902908325195, "learning_rate": 0.0005, "loss": 1.3269686698913574, "step": 36909 }, { "epoch": 2.90973590855341, "grad_norm": 0.5032293200492859, "learning_rate": 0.0005, "loss": 1.6175484657287598, "step": 36910 }, { "epoch": 2.9098147418210485, "grad_norm": 0.4156840145587921, "learning_rate": 0.0005, "loss": 1.1850621700286865, "step": 36911 }, { "epoch": 2.909893575088687, "grad_norm": 0.5749214291572571, "learning_rate": 0.0005, "loss": 1.419792890548706, "step": 36912 }, { "epoch": 2.9099724083563263, "grad_norm": 0.4361429810523987, "learning_rate": 0.0005, "loss": 1.0972586870193481, "step": 36913 }, { "epoch": 2.9100512416239654, "grad_norm": 0.4678072929382324, "learning_rate": 0.0005, "loss": 1.2295434474945068, "step": 36914 }, { "epoch": 2.910130074891604, "grad_norm": 0.5872445106506348, "learning_rate": 0.0005, "loss": 1.207399845123291, "step": 36915 }, { "epoch": 2.9102089081592433, "grad_norm": 0.43951353430747986, "learning_rate": 0.0005, "loss": 1.3484883308410645, "step": 36916 }, { "epoch": 2.910287741426882, "grad_norm": 0.6739217638969421, "learning_rate": 0.0005, "loss": 1.4294335842132568, "step": 36917 }, { "epoch": 2.910366574694521, "grad_norm": 0.4090757369995117, "learning_rate": 0.0005, "loss": 1.16990065574646, "step": 36918 }, { "epoch": 2.9104454079621602, "grad_norm": 0.546088695526123, "learning_rate": 0.0005, "loss": 1.2970070838928223, "step": 36919 }, { "epoch": 2.910524241229799, "grad_norm": 0.6768372654914856, "learning_rate": 0.0005, "loss": 1.3515491485595703, "step": 36920 }, { "epoch": 2.910603074497438, "grad_norm": 0.4636662006378174, "learning_rate": 0.0005, "loss": 1.363161563873291, "step": 36921 }, { "epoch": 2.9106819077650767, "grad_norm": 0.5207840204238892, "learning_rate": 0.0005, "loss": 1.271277666091919, "step": 36922 }, { "epoch": 2.910760741032716, "grad_norm": 0.5020939707756042, "learning_rate": 0.0005, "loss": 1.2824287414550781, "step": 36923 }, { "epoch": 2.910839574300355, "grad_norm": 0.42676281929016113, "learning_rate": 0.0005, "loss": 1.18186616897583, "step": 36924 }, { "epoch": 2.9109184075679937, "grad_norm": 0.4876631200313568, "learning_rate": 0.0005, "loss": 1.156577229499817, "step": 36925 }, { "epoch": 2.9109972408356324, "grad_norm": 0.5830894112586975, "learning_rate": 0.0005, "loss": 1.3275277614593506, "step": 36926 }, { "epoch": 2.9110760741032715, "grad_norm": 0.532579243183136, "learning_rate": 0.0005, "loss": 1.4868354797363281, "step": 36927 }, { "epoch": 2.9111549073709106, "grad_norm": 0.43963301181793213, "learning_rate": 0.0005, "loss": 1.2184507846832275, "step": 36928 }, { "epoch": 2.9112337406385493, "grad_norm": 0.5391221642494202, "learning_rate": 0.0005, "loss": 1.4156187772750854, "step": 36929 }, { "epoch": 2.9113125739061885, "grad_norm": 0.48113682866096497, "learning_rate": 0.0005, "loss": 1.3168420791625977, "step": 36930 }, { "epoch": 2.911391407173827, "grad_norm": 0.4864186942577362, "learning_rate": 0.0005, "loss": 1.2234313488006592, "step": 36931 }, { "epoch": 2.9114702404414663, "grad_norm": 0.678974449634552, "learning_rate": 0.0005, "loss": 1.2460447549819946, "step": 36932 }, { "epoch": 2.9115490737091054, "grad_norm": 0.502634584903717, "learning_rate": 0.0005, "loss": 1.4399915933609009, "step": 36933 }, { "epoch": 2.911627906976744, "grad_norm": 0.520383358001709, "learning_rate": 0.0005, "loss": 1.4785938262939453, "step": 36934 }, { "epoch": 2.9117067402443833, "grad_norm": 0.6387502551078796, "learning_rate": 0.0005, "loss": 1.3899025917053223, "step": 36935 }, { "epoch": 2.911785573512022, "grad_norm": 0.4711955189704895, "learning_rate": 0.0005, "loss": 1.5156952142715454, "step": 36936 }, { "epoch": 2.911864406779661, "grad_norm": 0.5116642713546753, "learning_rate": 0.0005, "loss": 1.2529258728027344, "step": 36937 }, { "epoch": 2.9119432400473, "grad_norm": 0.45047616958618164, "learning_rate": 0.0005, "loss": 1.1942095756530762, "step": 36938 }, { "epoch": 2.912022073314939, "grad_norm": 0.5382534265518188, "learning_rate": 0.0005, "loss": 1.3015532493591309, "step": 36939 }, { "epoch": 2.9121009065825776, "grad_norm": 0.5375515818595886, "learning_rate": 0.0005, "loss": 1.5662827491760254, "step": 36940 }, { "epoch": 2.9121797398502167, "grad_norm": 0.45447051525115967, "learning_rate": 0.0005, "loss": 1.336778998374939, "step": 36941 }, { "epoch": 2.912258573117856, "grad_norm": 0.4354180693626404, "learning_rate": 0.0005, "loss": 1.3220653533935547, "step": 36942 }, { "epoch": 2.9123374063854945, "grad_norm": 0.5533021092414856, "learning_rate": 0.0005, "loss": 1.3991013765335083, "step": 36943 }, { "epoch": 2.9124162396531337, "grad_norm": 0.4949522018432617, "learning_rate": 0.0005, "loss": 1.3283640146255493, "step": 36944 }, { "epoch": 2.9124950729207724, "grad_norm": 0.5447515249252319, "learning_rate": 0.0005, "loss": 1.2642217874526978, "step": 36945 }, { "epoch": 2.9125739061884115, "grad_norm": 0.44102150201797485, "learning_rate": 0.0005, "loss": 1.3977701663970947, "step": 36946 }, { "epoch": 2.9126527394560506, "grad_norm": 0.6519960761070251, "learning_rate": 0.0005, "loss": 1.260233759880066, "step": 36947 }, { "epoch": 2.9127315727236893, "grad_norm": 0.4673287868499756, "learning_rate": 0.0005, "loss": 1.1789220571517944, "step": 36948 }, { "epoch": 2.9128104059913285, "grad_norm": 0.46367380023002625, "learning_rate": 0.0005, "loss": 1.33587646484375, "step": 36949 }, { "epoch": 2.912889239258967, "grad_norm": 0.4520171582698822, "learning_rate": 0.0005, "loss": 1.1901905536651611, "step": 36950 }, { "epoch": 2.9129680725266063, "grad_norm": 0.65495765209198, "learning_rate": 0.0005, "loss": 1.6608834266662598, "step": 36951 }, { "epoch": 2.9130469057942454, "grad_norm": 0.38962775468826294, "learning_rate": 0.0005, "loss": 1.2395354509353638, "step": 36952 }, { "epoch": 2.913125739061884, "grad_norm": 0.40841421484947205, "learning_rate": 0.0005, "loss": 1.1027288436889648, "step": 36953 }, { "epoch": 2.913204572329523, "grad_norm": 0.6060893535614014, "learning_rate": 0.0005, "loss": 1.278112769126892, "step": 36954 }, { "epoch": 2.913283405597162, "grad_norm": 0.4811036288738251, "learning_rate": 0.0005, "loss": 1.2699594497680664, "step": 36955 }, { "epoch": 2.913362238864801, "grad_norm": 0.5302385687828064, "learning_rate": 0.0005, "loss": 1.2229913473129272, "step": 36956 }, { "epoch": 2.9134410721324397, "grad_norm": 0.5291746854782104, "learning_rate": 0.0005, "loss": 1.3309895992279053, "step": 36957 }, { "epoch": 2.913519905400079, "grad_norm": 0.4955120384693146, "learning_rate": 0.0005, "loss": 1.2204376459121704, "step": 36958 }, { "epoch": 2.9135987386677176, "grad_norm": 0.48224616050720215, "learning_rate": 0.0005, "loss": 1.3375946283340454, "step": 36959 }, { "epoch": 2.9136775719353567, "grad_norm": 0.546953022480011, "learning_rate": 0.0005, "loss": 1.3697541952133179, "step": 36960 }, { "epoch": 2.913756405202996, "grad_norm": 0.638590931892395, "learning_rate": 0.0005, "loss": 1.3744851350784302, "step": 36961 }, { "epoch": 2.9138352384706345, "grad_norm": 0.508464515209198, "learning_rate": 0.0005, "loss": 1.3519694805145264, "step": 36962 }, { "epoch": 2.9139140717382737, "grad_norm": 0.4419463276863098, "learning_rate": 0.0005, "loss": 1.3030686378479004, "step": 36963 }, { "epoch": 2.9139929050059123, "grad_norm": 0.47826558351516724, "learning_rate": 0.0005, "loss": 1.3329964876174927, "step": 36964 }, { "epoch": 2.9140717382735515, "grad_norm": 0.47328639030456543, "learning_rate": 0.0005, "loss": 1.3267372846603394, "step": 36965 }, { "epoch": 2.9141505715411906, "grad_norm": 0.45916062593460083, "learning_rate": 0.0005, "loss": 1.451324462890625, "step": 36966 }, { "epoch": 2.9142294048088293, "grad_norm": 0.46933698654174805, "learning_rate": 0.0005, "loss": 1.4535340070724487, "step": 36967 }, { "epoch": 2.914308238076468, "grad_norm": 0.4600445032119751, "learning_rate": 0.0005, "loss": 1.437723159790039, "step": 36968 }, { "epoch": 2.914387071344107, "grad_norm": 0.47585296630859375, "learning_rate": 0.0005, "loss": 1.1030261516571045, "step": 36969 }, { "epoch": 2.9144659046117463, "grad_norm": 0.6632093191146851, "learning_rate": 0.0005, "loss": 1.4722957611083984, "step": 36970 }, { "epoch": 2.914544737879385, "grad_norm": 0.7209786176681519, "learning_rate": 0.0005, "loss": 1.536698818206787, "step": 36971 }, { "epoch": 2.914623571147024, "grad_norm": 0.5219303369522095, "learning_rate": 0.0005, "loss": 1.3155385255813599, "step": 36972 }, { "epoch": 2.9147024044146628, "grad_norm": 0.4542820155620575, "learning_rate": 0.0005, "loss": 1.3161669969558716, "step": 36973 }, { "epoch": 2.914781237682302, "grad_norm": 0.484252393245697, "learning_rate": 0.0005, "loss": 1.326305627822876, "step": 36974 }, { "epoch": 2.914860070949941, "grad_norm": 0.4930354952812195, "learning_rate": 0.0005, "loss": 1.3657968044281006, "step": 36975 }, { "epoch": 2.9149389042175797, "grad_norm": 0.5101057291030884, "learning_rate": 0.0005, "loss": 1.2119355201721191, "step": 36976 }, { "epoch": 2.915017737485219, "grad_norm": 0.5133827924728394, "learning_rate": 0.0005, "loss": 1.3461031913757324, "step": 36977 }, { "epoch": 2.9150965707528576, "grad_norm": 0.6320619583129883, "learning_rate": 0.0005, "loss": 1.2694365978240967, "step": 36978 }, { "epoch": 2.9151754040204967, "grad_norm": 0.5033303499221802, "learning_rate": 0.0005, "loss": 1.3729146718978882, "step": 36979 }, { "epoch": 2.915254237288136, "grad_norm": 0.4354974329471588, "learning_rate": 0.0005, "loss": 1.3810715675354004, "step": 36980 }, { "epoch": 2.9153330705557745, "grad_norm": 0.5494214296340942, "learning_rate": 0.0005, "loss": 1.2650933265686035, "step": 36981 }, { "epoch": 2.9154119038234136, "grad_norm": 0.48085007071495056, "learning_rate": 0.0005, "loss": 1.5463676452636719, "step": 36982 }, { "epoch": 2.9154907370910523, "grad_norm": 0.4857863485813141, "learning_rate": 0.0005, "loss": 1.247083306312561, "step": 36983 }, { "epoch": 2.9155695703586915, "grad_norm": 0.575109601020813, "learning_rate": 0.0005, "loss": 1.3869221210479736, "step": 36984 }, { "epoch": 2.91564840362633, "grad_norm": 0.4186551868915558, "learning_rate": 0.0005, "loss": 1.2813730239868164, "step": 36985 }, { "epoch": 2.9157272368939693, "grad_norm": 0.4221450984477997, "learning_rate": 0.0005, "loss": 1.1564788818359375, "step": 36986 }, { "epoch": 2.915806070161608, "grad_norm": 0.5170528292655945, "learning_rate": 0.0005, "loss": 1.5556719303131104, "step": 36987 }, { "epoch": 2.915884903429247, "grad_norm": 0.5137656927108765, "learning_rate": 0.0005, "loss": 1.2947804927825928, "step": 36988 }, { "epoch": 2.9159637366968862, "grad_norm": 0.5698262453079224, "learning_rate": 0.0005, "loss": 1.4766826629638672, "step": 36989 }, { "epoch": 2.916042569964525, "grad_norm": 0.48098546266555786, "learning_rate": 0.0005, "loss": 1.3149774074554443, "step": 36990 }, { "epoch": 2.916121403232164, "grad_norm": 0.49446597695350647, "learning_rate": 0.0005, "loss": 1.3117382526397705, "step": 36991 }, { "epoch": 2.9162002364998028, "grad_norm": 0.54599928855896, "learning_rate": 0.0005, "loss": 1.5048828125, "step": 36992 }, { "epoch": 2.916279069767442, "grad_norm": 0.5420182347297668, "learning_rate": 0.0005, "loss": 1.2327485084533691, "step": 36993 }, { "epoch": 2.916357903035081, "grad_norm": 0.41671472787857056, "learning_rate": 0.0005, "loss": 1.2565584182739258, "step": 36994 }, { "epoch": 2.9164367363027197, "grad_norm": 0.5733807682991028, "learning_rate": 0.0005, "loss": 1.40772545337677, "step": 36995 }, { "epoch": 2.916515569570359, "grad_norm": 0.5048863887786865, "learning_rate": 0.0005, "loss": 1.3377876281738281, "step": 36996 }, { "epoch": 2.9165944028379975, "grad_norm": 0.6199180483818054, "learning_rate": 0.0005, "loss": 1.3933312892913818, "step": 36997 }, { "epoch": 2.9166732361056367, "grad_norm": 0.5510709881782532, "learning_rate": 0.0005, "loss": 1.507304310798645, "step": 36998 }, { "epoch": 2.9167520693732754, "grad_norm": 0.3885926604270935, "learning_rate": 0.0005, "loss": 1.2071657180786133, "step": 36999 }, { "epoch": 2.9168309026409145, "grad_norm": 0.5038968920707703, "learning_rate": 0.0005, "loss": 1.2944408655166626, "step": 37000 }, { "epoch": 2.916909735908553, "grad_norm": 0.4567616283893585, "learning_rate": 0.0005, "loss": 1.3394603729248047, "step": 37001 }, { "epoch": 2.9169885691761923, "grad_norm": 0.5274866819381714, "learning_rate": 0.0005, "loss": 1.596218466758728, "step": 37002 }, { "epoch": 2.9170674024438314, "grad_norm": 0.5185810327529907, "learning_rate": 0.0005, "loss": 1.309602975845337, "step": 37003 }, { "epoch": 2.91714623571147, "grad_norm": 0.4124067425727844, "learning_rate": 0.0005, "loss": 1.0114109516143799, "step": 37004 }, { "epoch": 2.9172250689791093, "grad_norm": 0.5147106647491455, "learning_rate": 0.0005, "loss": 1.5136133432388306, "step": 37005 }, { "epoch": 2.917303902246748, "grad_norm": 0.5802964568138123, "learning_rate": 0.0005, "loss": 1.5265731811523438, "step": 37006 }, { "epoch": 2.917382735514387, "grad_norm": 0.4966045320034027, "learning_rate": 0.0005, "loss": 1.418853759765625, "step": 37007 }, { "epoch": 2.9174615687820262, "grad_norm": 0.5714980959892273, "learning_rate": 0.0005, "loss": 1.3266267776489258, "step": 37008 }, { "epoch": 2.917540402049665, "grad_norm": 0.6643767952919006, "learning_rate": 0.0005, "loss": 1.4327237606048584, "step": 37009 }, { "epoch": 2.917619235317304, "grad_norm": 0.44268563389778137, "learning_rate": 0.0005, "loss": 1.176507592201233, "step": 37010 }, { "epoch": 2.9176980685849427, "grad_norm": 0.4347441792488098, "learning_rate": 0.0005, "loss": 1.1179190874099731, "step": 37011 }, { "epoch": 2.917776901852582, "grad_norm": 0.5301102995872498, "learning_rate": 0.0005, "loss": 1.5676329135894775, "step": 37012 }, { "epoch": 2.917855735120221, "grad_norm": 0.4919162690639496, "learning_rate": 0.0005, "loss": 1.2556681632995605, "step": 37013 }, { "epoch": 2.9179345683878597, "grad_norm": 0.4693661630153656, "learning_rate": 0.0005, "loss": 1.2624505758285522, "step": 37014 }, { "epoch": 2.9180134016554984, "grad_norm": 0.5899638533592224, "learning_rate": 0.0005, "loss": 1.411026954650879, "step": 37015 }, { "epoch": 2.9180922349231375, "grad_norm": 0.43282341957092285, "learning_rate": 0.0005, "loss": 1.300299882888794, "step": 37016 }, { "epoch": 2.9181710681907767, "grad_norm": 0.5798643827438354, "learning_rate": 0.0005, "loss": 1.3192410469055176, "step": 37017 }, { "epoch": 2.9182499014584153, "grad_norm": 0.4211958646774292, "learning_rate": 0.0005, "loss": 1.311682939529419, "step": 37018 }, { "epoch": 2.9183287347260545, "grad_norm": 0.5251580476760864, "learning_rate": 0.0005, "loss": 1.3584399223327637, "step": 37019 }, { "epoch": 2.918407567993693, "grad_norm": 0.4369482696056366, "learning_rate": 0.0005, "loss": 0.923346221446991, "step": 37020 }, { "epoch": 2.9184864012613323, "grad_norm": 0.5396785140037537, "learning_rate": 0.0005, "loss": 1.3975915908813477, "step": 37021 }, { "epoch": 2.9185652345289714, "grad_norm": 0.5303711891174316, "learning_rate": 0.0005, "loss": 1.4248442649841309, "step": 37022 }, { "epoch": 2.91864406779661, "grad_norm": 0.5173888206481934, "learning_rate": 0.0005, "loss": 1.3624777793884277, "step": 37023 }, { "epoch": 2.9187229010642493, "grad_norm": 0.4695943295955658, "learning_rate": 0.0005, "loss": 1.3204154968261719, "step": 37024 }, { "epoch": 2.918801734331888, "grad_norm": 0.5163505673408508, "learning_rate": 0.0005, "loss": 1.2825064659118652, "step": 37025 }, { "epoch": 2.918880567599527, "grad_norm": 0.5453382134437561, "learning_rate": 0.0005, "loss": 1.3505510091781616, "step": 37026 }, { "epoch": 2.918959400867166, "grad_norm": 0.4673963785171509, "learning_rate": 0.0005, "loss": 1.2215440273284912, "step": 37027 }, { "epoch": 2.919038234134805, "grad_norm": 0.4564964771270752, "learning_rate": 0.0005, "loss": 1.2297546863555908, "step": 37028 }, { "epoch": 2.9191170674024436, "grad_norm": 0.42107465863227844, "learning_rate": 0.0005, "loss": 1.2202574014663696, "step": 37029 }, { "epoch": 2.9191959006700827, "grad_norm": 0.48270711302757263, "learning_rate": 0.0005, "loss": 1.3288401365280151, "step": 37030 }, { "epoch": 2.919274733937722, "grad_norm": 0.4790329337120056, "learning_rate": 0.0005, "loss": 1.1689648628234863, "step": 37031 }, { "epoch": 2.9193535672053605, "grad_norm": 0.4616105556488037, "learning_rate": 0.0005, "loss": 1.3305246829986572, "step": 37032 }, { "epoch": 2.9194324004729997, "grad_norm": 0.46562373638153076, "learning_rate": 0.0005, "loss": 1.2404807806015015, "step": 37033 }, { "epoch": 2.9195112337406384, "grad_norm": 0.41864749789237976, "learning_rate": 0.0005, "loss": 1.2151005268096924, "step": 37034 }, { "epoch": 2.9195900670082775, "grad_norm": 0.500231146812439, "learning_rate": 0.0005, "loss": 1.4428435564041138, "step": 37035 }, { "epoch": 2.9196689002759166, "grad_norm": 0.4431237280368805, "learning_rate": 0.0005, "loss": 1.247208833694458, "step": 37036 }, { "epoch": 2.9197477335435553, "grad_norm": 0.7044850587844849, "learning_rate": 0.0005, "loss": 1.2299288511276245, "step": 37037 }, { "epoch": 2.9198265668111945, "grad_norm": 0.484625905752182, "learning_rate": 0.0005, "loss": 1.1073620319366455, "step": 37038 }, { "epoch": 2.919905400078833, "grad_norm": 0.5829653739929199, "learning_rate": 0.0005, "loss": 1.489316701889038, "step": 37039 }, { "epoch": 2.9199842333464723, "grad_norm": 0.5627983808517456, "learning_rate": 0.0005, "loss": 1.3886598348617554, "step": 37040 }, { "epoch": 2.9200630666141114, "grad_norm": 0.4668344557285309, "learning_rate": 0.0005, "loss": 1.3270599842071533, "step": 37041 }, { "epoch": 2.92014189988175, "grad_norm": 0.6599605083465576, "learning_rate": 0.0005, "loss": 1.5079963207244873, "step": 37042 }, { "epoch": 2.920220733149389, "grad_norm": 0.5690904855728149, "learning_rate": 0.0005, "loss": 1.425349473953247, "step": 37043 }, { "epoch": 2.920299566417028, "grad_norm": 0.5300254225730896, "learning_rate": 0.0005, "loss": 1.4710769653320312, "step": 37044 }, { "epoch": 2.920378399684667, "grad_norm": 0.5253704786300659, "learning_rate": 0.0005, "loss": 1.3712735176086426, "step": 37045 }, { "epoch": 2.9204572329523057, "grad_norm": 0.42431795597076416, "learning_rate": 0.0005, "loss": 1.24259352684021, "step": 37046 }, { "epoch": 2.920536066219945, "grad_norm": 0.548588216304779, "learning_rate": 0.0005, "loss": 1.560464859008789, "step": 37047 }, { "epoch": 2.9206148994875836, "grad_norm": 0.4336687922477722, "learning_rate": 0.0005, "loss": 1.2018563747406006, "step": 37048 }, { "epoch": 2.9206937327552227, "grad_norm": 0.4930295944213867, "learning_rate": 0.0005, "loss": 1.2984004020690918, "step": 37049 }, { "epoch": 2.920772566022862, "grad_norm": 1.0352643728256226, "learning_rate": 0.0005, "loss": 1.2795675992965698, "step": 37050 }, { "epoch": 2.9208513992905005, "grad_norm": 0.45211049914360046, "learning_rate": 0.0005, "loss": 1.2965149879455566, "step": 37051 }, { "epoch": 2.9209302325581397, "grad_norm": 0.49307841062545776, "learning_rate": 0.0005, "loss": 1.1648328304290771, "step": 37052 }, { "epoch": 2.9210090658257784, "grad_norm": 0.431084007024765, "learning_rate": 0.0005, "loss": 1.295543909072876, "step": 37053 }, { "epoch": 2.9210878990934175, "grad_norm": 0.7365328073501587, "learning_rate": 0.0005, "loss": 1.4688684940338135, "step": 37054 }, { "epoch": 2.9211667323610566, "grad_norm": 0.47521355748176575, "learning_rate": 0.0005, "loss": 1.1776520013809204, "step": 37055 }, { "epoch": 2.9212455656286953, "grad_norm": 0.4481161832809448, "learning_rate": 0.0005, "loss": 1.4036767482757568, "step": 37056 }, { "epoch": 2.921324398896334, "grad_norm": 0.8336414098739624, "learning_rate": 0.0005, "loss": 1.4135925769805908, "step": 37057 }, { "epoch": 2.921403232163973, "grad_norm": 0.7145709991455078, "learning_rate": 0.0005, "loss": 1.3576338291168213, "step": 37058 }, { "epoch": 2.9214820654316123, "grad_norm": 0.47540974617004395, "learning_rate": 0.0005, "loss": 1.148485541343689, "step": 37059 }, { "epoch": 2.921560898699251, "grad_norm": 0.4955041706562042, "learning_rate": 0.0005, "loss": 1.3408900499343872, "step": 37060 }, { "epoch": 2.92163973196689, "grad_norm": 0.5192373991012573, "learning_rate": 0.0005, "loss": 1.1769890785217285, "step": 37061 }, { "epoch": 2.9217185652345288, "grad_norm": 0.47534000873565674, "learning_rate": 0.0005, "loss": 1.2177822589874268, "step": 37062 }, { "epoch": 2.921797398502168, "grad_norm": 0.44705930352211, "learning_rate": 0.0005, "loss": 1.4418439865112305, "step": 37063 }, { "epoch": 2.921876231769807, "grad_norm": 0.6102970242500305, "learning_rate": 0.0005, "loss": 1.3874642848968506, "step": 37064 }, { "epoch": 2.9219550650374457, "grad_norm": 0.45829710364341736, "learning_rate": 0.0005, "loss": 1.2922818660736084, "step": 37065 }, { "epoch": 2.922033898305085, "grad_norm": 0.5990520715713501, "learning_rate": 0.0005, "loss": 1.490405797958374, "step": 37066 }, { "epoch": 2.9221127315727236, "grad_norm": 0.5407623648643494, "learning_rate": 0.0005, "loss": 1.3721888065338135, "step": 37067 }, { "epoch": 2.9221915648403627, "grad_norm": 0.43328502774238586, "learning_rate": 0.0005, "loss": 1.1821410655975342, "step": 37068 }, { "epoch": 2.922270398108002, "grad_norm": 0.4428284168243408, "learning_rate": 0.0005, "loss": 1.24334716796875, "step": 37069 }, { "epoch": 2.9223492313756405, "grad_norm": 0.4531257748603821, "learning_rate": 0.0005, "loss": 1.3651752471923828, "step": 37070 }, { "epoch": 2.922428064643279, "grad_norm": 0.46154752373695374, "learning_rate": 0.0005, "loss": 1.237304449081421, "step": 37071 }, { "epoch": 2.9225068979109183, "grad_norm": 0.6701400876045227, "learning_rate": 0.0005, "loss": 1.4111356735229492, "step": 37072 }, { "epoch": 2.9225857311785575, "grad_norm": 0.527035117149353, "learning_rate": 0.0005, "loss": 1.545750379562378, "step": 37073 }, { "epoch": 2.922664564446196, "grad_norm": 0.6051018834114075, "learning_rate": 0.0005, "loss": 1.3800697326660156, "step": 37074 }, { "epoch": 2.9227433977138353, "grad_norm": 0.5844812989234924, "learning_rate": 0.0005, "loss": 1.474501371383667, "step": 37075 }, { "epoch": 2.922822230981474, "grad_norm": 0.4465107321739197, "learning_rate": 0.0005, "loss": 1.2171211242675781, "step": 37076 }, { "epoch": 2.922901064249113, "grad_norm": 0.46238335967063904, "learning_rate": 0.0005, "loss": 1.0840976238250732, "step": 37077 }, { "epoch": 2.9229798975167522, "grad_norm": 0.5468716025352478, "learning_rate": 0.0005, "loss": 1.3052735328674316, "step": 37078 }, { "epoch": 2.923058730784391, "grad_norm": 0.5362783670425415, "learning_rate": 0.0005, "loss": 1.4046632051467896, "step": 37079 }, { "epoch": 2.92313756405203, "grad_norm": 0.638164758682251, "learning_rate": 0.0005, "loss": 1.5502548217773438, "step": 37080 }, { "epoch": 2.9232163973196688, "grad_norm": 0.5798524022102356, "learning_rate": 0.0005, "loss": 1.4543262720108032, "step": 37081 }, { "epoch": 2.923295230587308, "grad_norm": 0.6630875468254089, "learning_rate": 0.0005, "loss": 1.428640604019165, "step": 37082 }, { "epoch": 2.923374063854947, "grad_norm": 0.5323380827903748, "learning_rate": 0.0005, "loss": 1.2392053604125977, "step": 37083 }, { "epoch": 2.9234528971225857, "grad_norm": 0.4648243486881256, "learning_rate": 0.0005, "loss": 1.3372645378112793, "step": 37084 }, { "epoch": 2.9235317303902244, "grad_norm": 0.41273921728134155, "learning_rate": 0.0005, "loss": 1.0778923034667969, "step": 37085 }, { "epoch": 2.9236105636578635, "grad_norm": 0.5194581747055054, "learning_rate": 0.0005, "loss": 1.4903740882873535, "step": 37086 }, { "epoch": 2.9236893969255027, "grad_norm": 0.5807999968528748, "learning_rate": 0.0005, "loss": 1.4631770849227905, "step": 37087 }, { "epoch": 2.9237682301931414, "grad_norm": 0.4599182903766632, "learning_rate": 0.0005, "loss": 1.3369072675704956, "step": 37088 }, { "epoch": 2.9238470634607805, "grad_norm": 0.6453559398651123, "learning_rate": 0.0005, "loss": 1.338792324066162, "step": 37089 }, { "epoch": 2.923925896728419, "grad_norm": 0.6043188571929932, "learning_rate": 0.0005, "loss": 1.3892581462860107, "step": 37090 }, { "epoch": 2.9240047299960583, "grad_norm": 0.45119789242744446, "learning_rate": 0.0005, "loss": 1.1550836563110352, "step": 37091 }, { "epoch": 2.9240835632636974, "grad_norm": 0.5340854525566101, "learning_rate": 0.0005, "loss": 1.2086694240570068, "step": 37092 }, { "epoch": 2.924162396531336, "grad_norm": 0.48192089796066284, "learning_rate": 0.0005, "loss": 1.2226167917251587, "step": 37093 }, { "epoch": 2.9242412297989753, "grad_norm": 0.5893846154212952, "learning_rate": 0.0005, "loss": 1.3877215385437012, "step": 37094 }, { "epoch": 2.924320063066614, "grad_norm": 0.536146879196167, "learning_rate": 0.0005, "loss": 1.3889429569244385, "step": 37095 }, { "epoch": 2.924398896334253, "grad_norm": 0.5259897708892822, "learning_rate": 0.0005, "loss": 1.3617242574691772, "step": 37096 }, { "epoch": 2.9244777296018922, "grad_norm": 0.45891180634498596, "learning_rate": 0.0005, "loss": 1.1915000677108765, "step": 37097 }, { "epoch": 2.924556562869531, "grad_norm": 0.41200727224349976, "learning_rate": 0.0005, "loss": 1.4645999670028687, "step": 37098 }, { "epoch": 2.92463539613717, "grad_norm": 0.46895647048950195, "learning_rate": 0.0005, "loss": 1.4430420398712158, "step": 37099 }, { "epoch": 2.9247142294048087, "grad_norm": 0.4344503581523895, "learning_rate": 0.0005, "loss": 1.2048346996307373, "step": 37100 }, { "epoch": 2.924793062672448, "grad_norm": 0.4161474406719208, "learning_rate": 0.0005, "loss": 1.2067731618881226, "step": 37101 }, { "epoch": 2.9248718959400866, "grad_norm": 0.5162515640258789, "learning_rate": 0.0005, "loss": 1.6014823913574219, "step": 37102 }, { "epoch": 2.9249507292077257, "grad_norm": 0.5468837022781372, "learning_rate": 0.0005, "loss": 1.482527494430542, "step": 37103 }, { "epoch": 2.9250295624753644, "grad_norm": 0.4610632359981537, "learning_rate": 0.0005, "loss": 1.121326208114624, "step": 37104 }, { "epoch": 2.9251083957430035, "grad_norm": 0.4112664461135864, "learning_rate": 0.0005, "loss": 1.332477331161499, "step": 37105 }, { "epoch": 2.9251872290106427, "grad_norm": 0.48231562972068787, "learning_rate": 0.0005, "loss": 1.174212098121643, "step": 37106 }, { "epoch": 2.9252660622782813, "grad_norm": 0.4195958375930786, "learning_rate": 0.0005, "loss": 1.1806493997573853, "step": 37107 }, { "epoch": 2.9253448955459205, "grad_norm": 0.43805423378944397, "learning_rate": 0.0005, "loss": 1.32912015914917, "step": 37108 }, { "epoch": 2.925423728813559, "grad_norm": 0.7146281599998474, "learning_rate": 0.0005, "loss": 1.4676077365875244, "step": 37109 }, { "epoch": 2.9255025620811983, "grad_norm": 0.42287477850914, "learning_rate": 0.0005, "loss": 1.159832239151001, "step": 37110 }, { "epoch": 2.9255813953488374, "grad_norm": 0.5269594192504883, "learning_rate": 0.0005, "loss": 1.3002662658691406, "step": 37111 }, { "epoch": 2.925660228616476, "grad_norm": 0.4965255856513977, "learning_rate": 0.0005, "loss": 1.2166035175323486, "step": 37112 }, { "epoch": 2.9257390618841153, "grad_norm": 0.5154589414596558, "learning_rate": 0.0005, "loss": 1.5894086360931396, "step": 37113 }, { "epoch": 2.925817895151754, "grad_norm": 0.5878334641456604, "learning_rate": 0.0005, "loss": 1.5005624294281006, "step": 37114 }, { "epoch": 2.925896728419393, "grad_norm": 0.4614451229572296, "learning_rate": 0.0005, "loss": 1.3576023578643799, "step": 37115 }, { "epoch": 2.9259755616870318, "grad_norm": 0.4610863924026489, "learning_rate": 0.0005, "loss": 1.2923542261123657, "step": 37116 }, { "epoch": 2.926054394954671, "grad_norm": 0.426794171333313, "learning_rate": 0.0005, "loss": 1.1183805465698242, "step": 37117 }, { "epoch": 2.9261332282223096, "grad_norm": 0.49445411562919617, "learning_rate": 0.0005, "loss": 1.2015881538391113, "step": 37118 }, { "epoch": 2.9262120614899487, "grad_norm": 0.6025077700614929, "learning_rate": 0.0005, "loss": 1.3196818828582764, "step": 37119 }, { "epoch": 2.926290894757588, "grad_norm": 0.4335172474384308, "learning_rate": 0.0005, "loss": 1.1765928268432617, "step": 37120 }, { "epoch": 2.9263697280252265, "grad_norm": 0.5068265795707703, "learning_rate": 0.0005, "loss": 1.3751226663589478, "step": 37121 }, { "epoch": 2.9264485612928657, "grad_norm": 0.4613427519798279, "learning_rate": 0.0005, "loss": 1.130423665046692, "step": 37122 }, { "epoch": 2.9265273945605044, "grad_norm": 0.520601212978363, "learning_rate": 0.0005, "loss": 1.2427746057510376, "step": 37123 }, { "epoch": 2.9266062278281435, "grad_norm": 0.5663459300994873, "learning_rate": 0.0005, "loss": 1.2386668920516968, "step": 37124 }, { "epoch": 2.9266850610957826, "grad_norm": 0.6468981504440308, "learning_rate": 0.0005, "loss": 1.4114755392074585, "step": 37125 }, { "epoch": 2.9267638943634213, "grad_norm": 0.4794180691242218, "learning_rate": 0.0005, "loss": 1.2513439655303955, "step": 37126 }, { "epoch": 2.9268427276310605, "grad_norm": 0.4803503751754761, "learning_rate": 0.0005, "loss": 1.2672228813171387, "step": 37127 }, { "epoch": 2.926921560898699, "grad_norm": 0.44462940096855164, "learning_rate": 0.0005, "loss": 1.1392993927001953, "step": 37128 }, { "epoch": 2.9270003941663383, "grad_norm": 0.5345552563667297, "learning_rate": 0.0005, "loss": 1.3173401355743408, "step": 37129 }, { "epoch": 2.9270792274339774, "grad_norm": 0.5500726103782654, "learning_rate": 0.0005, "loss": 1.3881094455718994, "step": 37130 }, { "epoch": 2.927158060701616, "grad_norm": 0.48535189032554626, "learning_rate": 0.0005, "loss": 1.4546911716461182, "step": 37131 }, { "epoch": 2.927236893969255, "grad_norm": 0.4248813986778259, "learning_rate": 0.0005, "loss": 1.4489046335220337, "step": 37132 }, { "epoch": 2.927315727236894, "grad_norm": 0.5894340872764587, "learning_rate": 0.0005, "loss": 1.4235817193984985, "step": 37133 }, { "epoch": 2.927394560504533, "grad_norm": 0.5517193078994751, "learning_rate": 0.0005, "loss": 1.2534539699554443, "step": 37134 }, { "epoch": 2.9274733937721718, "grad_norm": 0.4684257507324219, "learning_rate": 0.0005, "loss": 1.2777066230773926, "step": 37135 }, { "epoch": 2.927552227039811, "grad_norm": 0.5434359908103943, "learning_rate": 0.0005, "loss": 1.4174988269805908, "step": 37136 }, { "epoch": 2.9276310603074496, "grad_norm": 0.5433798432350159, "learning_rate": 0.0005, "loss": 1.4824016094207764, "step": 37137 }, { "epoch": 2.9277098935750887, "grad_norm": 0.47177815437316895, "learning_rate": 0.0005, "loss": 1.2767916917800903, "step": 37138 }, { "epoch": 2.927788726842728, "grad_norm": 0.5078001618385315, "learning_rate": 0.0005, "loss": 1.3687137365341187, "step": 37139 }, { "epoch": 2.9278675601103665, "grad_norm": 0.6282058358192444, "learning_rate": 0.0005, "loss": 1.4356995820999146, "step": 37140 }, { "epoch": 2.9279463933780057, "grad_norm": 0.39212435483932495, "learning_rate": 0.0005, "loss": 1.020278811454773, "step": 37141 }, { "epoch": 2.9280252266456444, "grad_norm": 0.6144207715988159, "learning_rate": 0.0005, "loss": 1.2740237712860107, "step": 37142 }, { "epoch": 2.9281040599132835, "grad_norm": 0.5204131603240967, "learning_rate": 0.0005, "loss": 1.4103635549545288, "step": 37143 }, { "epoch": 2.9281828931809226, "grad_norm": 0.5053072571754456, "learning_rate": 0.0005, "loss": 1.3556917905807495, "step": 37144 }, { "epoch": 2.9282617264485613, "grad_norm": 0.4375908672809601, "learning_rate": 0.0005, "loss": 1.2167506217956543, "step": 37145 }, { "epoch": 2.9283405597162, "grad_norm": 0.5197266936302185, "learning_rate": 0.0005, "loss": 1.381803274154663, "step": 37146 }, { "epoch": 2.928419392983839, "grad_norm": 0.46808329224586487, "learning_rate": 0.0005, "loss": 1.5173287391662598, "step": 37147 }, { "epoch": 2.9284982262514783, "grad_norm": 0.6033085584640503, "learning_rate": 0.0005, "loss": 1.3661636114120483, "step": 37148 }, { "epoch": 2.928577059519117, "grad_norm": 0.4786939024925232, "learning_rate": 0.0005, "loss": 1.3245532512664795, "step": 37149 }, { "epoch": 2.928655892786756, "grad_norm": 0.4980742633342743, "learning_rate": 0.0005, "loss": 1.2369310855865479, "step": 37150 }, { "epoch": 2.9287347260543948, "grad_norm": 0.6769166588783264, "learning_rate": 0.0005, "loss": 1.2458837032318115, "step": 37151 }, { "epoch": 2.928813559322034, "grad_norm": 0.5855900645256042, "learning_rate": 0.0005, "loss": 1.382710337638855, "step": 37152 }, { "epoch": 2.928892392589673, "grad_norm": 0.47264671325683594, "learning_rate": 0.0005, "loss": 1.2826122045516968, "step": 37153 }, { "epoch": 2.9289712258573117, "grad_norm": 0.4472605288028717, "learning_rate": 0.0005, "loss": 1.3547197580337524, "step": 37154 }, { "epoch": 2.929050059124951, "grad_norm": 0.8165088295936584, "learning_rate": 0.0005, "loss": 1.4858368635177612, "step": 37155 }, { "epoch": 2.9291288923925896, "grad_norm": 0.5473906397819519, "learning_rate": 0.0005, "loss": 1.4349207878112793, "step": 37156 }, { "epoch": 2.9292077256602287, "grad_norm": 0.5876787900924683, "learning_rate": 0.0005, "loss": 1.4336549043655396, "step": 37157 }, { "epoch": 2.929286558927868, "grad_norm": 0.6003212332725525, "learning_rate": 0.0005, "loss": 1.448138952255249, "step": 37158 }, { "epoch": 2.9293653921955065, "grad_norm": 0.5714208483695984, "learning_rate": 0.0005, "loss": 1.3927443027496338, "step": 37159 }, { "epoch": 2.929444225463145, "grad_norm": 0.6903099417686462, "learning_rate": 0.0005, "loss": 1.3485355377197266, "step": 37160 }, { "epoch": 2.9295230587307843, "grad_norm": 0.507839560508728, "learning_rate": 0.0005, "loss": 1.255950689315796, "step": 37161 }, { "epoch": 2.9296018919984235, "grad_norm": 0.5306001901626587, "learning_rate": 0.0005, "loss": 1.4297853708267212, "step": 37162 }, { "epoch": 2.929680725266062, "grad_norm": 0.49935320019721985, "learning_rate": 0.0005, "loss": 1.3513002395629883, "step": 37163 }, { "epoch": 2.9297595585337013, "grad_norm": 0.48116543889045715, "learning_rate": 0.0005, "loss": 1.2351255416870117, "step": 37164 }, { "epoch": 2.92983839180134, "grad_norm": 0.4143216907978058, "learning_rate": 0.0005, "loss": 1.104630470275879, "step": 37165 }, { "epoch": 2.929917225068979, "grad_norm": 0.4379315972328186, "learning_rate": 0.0005, "loss": 1.1775336265563965, "step": 37166 }, { "epoch": 2.9299960583366182, "grad_norm": 0.5708898305892944, "learning_rate": 0.0005, "loss": 1.2320594787597656, "step": 37167 }, { "epoch": 2.930074891604257, "grad_norm": 0.6050118207931519, "learning_rate": 0.0005, "loss": 1.4196016788482666, "step": 37168 }, { "epoch": 2.930153724871896, "grad_norm": 0.4333004653453827, "learning_rate": 0.0005, "loss": 1.2332016229629517, "step": 37169 }, { "epoch": 2.9302325581395348, "grad_norm": 0.41126886010169983, "learning_rate": 0.0005, "loss": 1.2606772184371948, "step": 37170 }, { "epoch": 2.930311391407174, "grad_norm": 0.491947740316391, "learning_rate": 0.0005, "loss": 1.2946857213974, "step": 37171 }, { "epoch": 2.930390224674813, "grad_norm": 0.42417046427726746, "learning_rate": 0.0005, "loss": 1.2926039695739746, "step": 37172 }, { "epoch": 2.9304690579424517, "grad_norm": 0.4747915267944336, "learning_rate": 0.0005, "loss": 1.2467405796051025, "step": 37173 }, { "epoch": 2.9305478912100904, "grad_norm": 0.4860409200191498, "learning_rate": 0.0005, "loss": 1.3810818195343018, "step": 37174 }, { "epoch": 2.9306267244777295, "grad_norm": 0.47754722833633423, "learning_rate": 0.0005, "loss": 1.1662609577178955, "step": 37175 }, { "epoch": 2.9307055577453687, "grad_norm": 0.5431028604507446, "learning_rate": 0.0005, "loss": 1.2524151802062988, "step": 37176 }, { "epoch": 2.9307843910130074, "grad_norm": 0.4641863703727722, "learning_rate": 0.0005, "loss": 1.0965638160705566, "step": 37177 }, { "epoch": 2.9308632242806465, "grad_norm": 0.45639660954475403, "learning_rate": 0.0005, "loss": 1.0920169353485107, "step": 37178 }, { "epoch": 2.930942057548285, "grad_norm": 0.5916280150413513, "learning_rate": 0.0005, "loss": 1.3529969453811646, "step": 37179 }, { "epoch": 2.9310208908159243, "grad_norm": 0.5096551775932312, "learning_rate": 0.0005, "loss": 1.2618939876556396, "step": 37180 }, { "epoch": 2.9310997240835635, "grad_norm": 0.45534375309944153, "learning_rate": 0.0005, "loss": 1.3524596691131592, "step": 37181 }, { "epoch": 2.931178557351202, "grad_norm": 0.6399125456809998, "learning_rate": 0.0005, "loss": 1.4562079906463623, "step": 37182 }, { "epoch": 2.9312573906188413, "grad_norm": 0.48712587356567383, "learning_rate": 0.0005, "loss": 1.249867558479309, "step": 37183 }, { "epoch": 2.93133622388648, "grad_norm": 0.4562353193759918, "learning_rate": 0.0005, "loss": 1.394737720489502, "step": 37184 }, { "epoch": 2.931415057154119, "grad_norm": 0.5886565446853638, "learning_rate": 0.0005, "loss": 1.3234421014785767, "step": 37185 }, { "epoch": 2.9314938904217582, "grad_norm": 0.4266306161880493, "learning_rate": 0.0005, "loss": 1.5120725631713867, "step": 37186 }, { "epoch": 2.931572723689397, "grad_norm": 0.9003441333770752, "learning_rate": 0.0005, "loss": 1.516008973121643, "step": 37187 }, { "epoch": 2.9316515569570356, "grad_norm": 0.51206374168396, "learning_rate": 0.0005, "loss": 1.2461533546447754, "step": 37188 }, { "epoch": 2.9317303902246747, "grad_norm": 0.5014638900756836, "learning_rate": 0.0005, "loss": 1.266013741493225, "step": 37189 }, { "epoch": 2.931809223492314, "grad_norm": 0.4393025040626526, "learning_rate": 0.0005, "loss": 1.2423433065414429, "step": 37190 }, { "epoch": 2.9318880567599526, "grad_norm": 0.46344903111457825, "learning_rate": 0.0005, "loss": 1.266005277633667, "step": 37191 }, { "epoch": 2.9319668900275917, "grad_norm": 0.4730507433414459, "learning_rate": 0.0005, "loss": 1.4079694747924805, "step": 37192 }, { "epoch": 2.9320457232952304, "grad_norm": 0.48005470633506775, "learning_rate": 0.0005, "loss": 1.2415852546691895, "step": 37193 }, { "epoch": 2.9321245565628695, "grad_norm": 0.5145275592803955, "learning_rate": 0.0005, "loss": 1.3251023292541504, "step": 37194 }, { "epoch": 2.9322033898305087, "grad_norm": 0.5873146653175354, "learning_rate": 0.0005, "loss": 1.527759313583374, "step": 37195 }, { "epoch": 2.9322822230981473, "grad_norm": 0.4787253439426422, "learning_rate": 0.0005, "loss": 1.2404372692108154, "step": 37196 }, { "epoch": 2.9323610563657865, "grad_norm": 0.48927274346351624, "learning_rate": 0.0005, "loss": 1.38224196434021, "step": 37197 }, { "epoch": 2.932439889633425, "grad_norm": 0.4267368018627167, "learning_rate": 0.0005, "loss": 1.2919275760650635, "step": 37198 }, { "epoch": 2.9325187229010643, "grad_norm": 0.4636916518211365, "learning_rate": 0.0005, "loss": 1.4937562942504883, "step": 37199 }, { "epoch": 2.9325975561687034, "grad_norm": 0.5377157330513, "learning_rate": 0.0005, "loss": 1.5767353773117065, "step": 37200 }, { "epoch": 2.932676389436342, "grad_norm": 0.4686710238456726, "learning_rate": 0.0005, "loss": 1.137939691543579, "step": 37201 }, { "epoch": 2.932755222703981, "grad_norm": 0.48888206481933594, "learning_rate": 0.0005, "loss": 1.3598582744598389, "step": 37202 }, { "epoch": 2.93283405597162, "grad_norm": 0.5694830417633057, "learning_rate": 0.0005, "loss": 1.1899170875549316, "step": 37203 }, { "epoch": 2.932912889239259, "grad_norm": 0.4525144696235657, "learning_rate": 0.0005, "loss": 1.3157850503921509, "step": 37204 }, { "epoch": 2.9329917225068978, "grad_norm": 0.40995004773139954, "learning_rate": 0.0005, "loss": 1.2054345607757568, "step": 37205 }, { "epoch": 2.933070555774537, "grad_norm": 0.5735835433006287, "learning_rate": 0.0005, "loss": 1.3417633771896362, "step": 37206 }, { "epoch": 2.9331493890421756, "grad_norm": 0.41268789768218994, "learning_rate": 0.0005, "loss": 1.3047653436660767, "step": 37207 }, { "epoch": 2.9332282223098147, "grad_norm": 0.47975605726242065, "learning_rate": 0.0005, "loss": 1.3533425331115723, "step": 37208 }, { "epoch": 2.933307055577454, "grad_norm": 0.45632562041282654, "learning_rate": 0.0005, "loss": 1.3203763961791992, "step": 37209 }, { "epoch": 2.9333858888450925, "grad_norm": 0.44133472442626953, "learning_rate": 0.0005, "loss": 1.3323339223861694, "step": 37210 }, { "epoch": 2.9334647221127317, "grad_norm": 0.5887109637260437, "learning_rate": 0.0005, "loss": 1.4445157051086426, "step": 37211 }, { "epoch": 2.9335435553803704, "grad_norm": 0.5128540396690369, "learning_rate": 0.0005, "loss": 1.4363890886306763, "step": 37212 }, { "epoch": 2.9336223886480095, "grad_norm": 0.49442586302757263, "learning_rate": 0.0005, "loss": 1.2755073308944702, "step": 37213 }, { "epoch": 2.9337012219156486, "grad_norm": 0.6204273104667664, "learning_rate": 0.0005, "loss": 1.242724895477295, "step": 37214 }, { "epoch": 2.9337800551832873, "grad_norm": 0.4332626760005951, "learning_rate": 0.0005, "loss": 1.3446180820465088, "step": 37215 }, { "epoch": 2.933858888450926, "grad_norm": 0.5069271922111511, "learning_rate": 0.0005, "loss": 1.4643628597259521, "step": 37216 }, { "epoch": 2.933937721718565, "grad_norm": 0.5569908022880554, "learning_rate": 0.0005, "loss": 1.3701748847961426, "step": 37217 }, { "epoch": 2.9340165549862043, "grad_norm": 0.4868171513080597, "learning_rate": 0.0005, "loss": 1.227219581604004, "step": 37218 }, { "epoch": 2.934095388253843, "grad_norm": 0.5722386837005615, "learning_rate": 0.0005, "loss": 1.4034600257873535, "step": 37219 }, { "epoch": 2.934174221521482, "grad_norm": 0.618185818195343, "learning_rate": 0.0005, "loss": 1.309371829032898, "step": 37220 }, { "epoch": 2.934253054789121, "grad_norm": 0.6088345050811768, "learning_rate": 0.0005, "loss": 1.3169732093811035, "step": 37221 }, { "epoch": 2.93433188805676, "grad_norm": 0.8067331314086914, "learning_rate": 0.0005, "loss": 1.4861385822296143, "step": 37222 }, { "epoch": 2.934410721324399, "grad_norm": 0.4292163550853729, "learning_rate": 0.0005, "loss": 1.3860780000686646, "step": 37223 }, { "epoch": 2.9344895545920378, "grad_norm": 0.459086537361145, "learning_rate": 0.0005, "loss": 1.1864182949066162, "step": 37224 }, { "epoch": 2.934568387859677, "grad_norm": 0.48850351572036743, "learning_rate": 0.0005, "loss": 1.257588505744934, "step": 37225 }, { "epoch": 2.9346472211273156, "grad_norm": 0.3817974030971527, "learning_rate": 0.0005, "loss": 1.1854068040847778, "step": 37226 }, { "epoch": 2.9347260543949547, "grad_norm": 0.473195880651474, "learning_rate": 0.0005, "loss": 1.2903202772140503, "step": 37227 }, { "epoch": 2.934804887662594, "grad_norm": 0.4574164152145386, "learning_rate": 0.0005, "loss": 1.320494294166565, "step": 37228 }, { "epoch": 2.9348837209302325, "grad_norm": 0.535136342048645, "learning_rate": 0.0005, "loss": 1.4620742797851562, "step": 37229 }, { "epoch": 2.9349625541978717, "grad_norm": 0.4566954970359802, "learning_rate": 0.0005, "loss": 1.2411543130874634, "step": 37230 }, { "epoch": 2.9350413874655104, "grad_norm": 0.59474116563797, "learning_rate": 0.0005, "loss": 1.456344485282898, "step": 37231 }, { "epoch": 2.9351202207331495, "grad_norm": 0.6011256575584412, "learning_rate": 0.0005, "loss": 1.2215924263000488, "step": 37232 }, { "epoch": 2.935199054000788, "grad_norm": 0.5223492980003357, "learning_rate": 0.0005, "loss": 1.6051890850067139, "step": 37233 }, { "epoch": 2.9352778872684273, "grad_norm": 0.4822731018066406, "learning_rate": 0.0005, "loss": 1.2947402000427246, "step": 37234 }, { "epoch": 2.935356720536066, "grad_norm": 0.5562509894371033, "learning_rate": 0.0005, "loss": 1.2786449193954468, "step": 37235 }, { "epoch": 2.935435553803705, "grad_norm": 0.4678162932395935, "learning_rate": 0.0005, "loss": 1.3099207878112793, "step": 37236 }, { "epoch": 2.9355143870713443, "grad_norm": 0.4066970944404602, "learning_rate": 0.0005, "loss": 1.084060549736023, "step": 37237 }, { "epoch": 2.935593220338983, "grad_norm": 0.528339684009552, "learning_rate": 0.0005, "loss": 1.410762071609497, "step": 37238 }, { "epoch": 2.935672053606622, "grad_norm": 0.5176801085472107, "learning_rate": 0.0005, "loss": 1.4862959384918213, "step": 37239 }, { "epoch": 2.935750886874261, "grad_norm": 0.42084038257598877, "learning_rate": 0.0005, "loss": 1.3179008960723877, "step": 37240 }, { "epoch": 2.9358297201419, "grad_norm": 0.4398622512817383, "learning_rate": 0.0005, "loss": 1.2764997482299805, "step": 37241 }, { "epoch": 2.935908553409539, "grad_norm": 0.45016908645629883, "learning_rate": 0.0005, "loss": 1.3016952276229858, "step": 37242 }, { "epoch": 2.9359873866771777, "grad_norm": 0.45250728726387024, "learning_rate": 0.0005, "loss": 1.3127859830856323, "step": 37243 }, { "epoch": 2.936066219944817, "grad_norm": 0.4285429120063782, "learning_rate": 0.0005, "loss": 1.3100700378417969, "step": 37244 }, { "epoch": 2.9361450532124556, "grad_norm": 0.46793133020401, "learning_rate": 0.0005, "loss": 1.493593692779541, "step": 37245 }, { "epoch": 2.9362238864800947, "grad_norm": 0.4741816520690918, "learning_rate": 0.0005, "loss": 1.285111427307129, "step": 37246 }, { "epoch": 2.9363027197477334, "grad_norm": 0.4930048882961273, "learning_rate": 0.0005, "loss": 1.3936216831207275, "step": 37247 }, { "epoch": 2.9363815530153725, "grad_norm": 0.5264933705329895, "learning_rate": 0.0005, "loss": 1.3642268180847168, "step": 37248 }, { "epoch": 2.936460386283011, "grad_norm": 0.7273560166358948, "learning_rate": 0.0005, "loss": 1.4915884733200073, "step": 37249 }, { "epoch": 2.9365392195506503, "grad_norm": 0.5266973972320557, "learning_rate": 0.0005, "loss": 1.2888414859771729, "step": 37250 }, { "epoch": 2.9366180528182895, "grad_norm": 0.7977702021598816, "learning_rate": 0.0005, "loss": 1.439014196395874, "step": 37251 }, { "epoch": 2.936696886085928, "grad_norm": 0.5796211361885071, "learning_rate": 0.0005, "loss": 1.2607982158660889, "step": 37252 }, { "epoch": 2.9367757193535673, "grad_norm": 0.6108240485191345, "learning_rate": 0.0005, "loss": 1.3802893161773682, "step": 37253 }, { "epoch": 2.936854552621206, "grad_norm": 0.4601585268974304, "learning_rate": 0.0005, "loss": 1.3287463188171387, "step": 37254 }, { "epoch": 2.936933385888845, "grad_norm": 0.42439112067222595, "learning_rate": 0.0005, "loss": 1.1774091720581055, "step": 37255 }, { "epoch": 2.9370122191564842, "grad_norm": 0.4861759543418884, "learning_rate": 0.0005, "loss": 1.2989158630371094, "step": 37256 }, { "epoch": 2.937091052424123, "grad_norm": 0.511989414691925, "learning_rate": 0.0005, "loss": 1.3604159355163574, "step": 37257 }, { "epoch": 2.937169885691762, "grad_norm": 0.4841563403606415, "learning_rate": 0.0005, "loss": 1.2937169075012207, "step": 37258 }, { "epoch": 2.9372487189594008, "grad_norm": 0.5103445053100586, "learning_rate": 0.0005, "loss": 1.3255064487457275, "step": 37259 }, { "epoch": 2.93732755222704, "grad_norm": 0.6328389644622803, "learning_rate": 0.0005, "loss": 1.3348963260650635, "step": 37260 }, { "epoch": 2.937406385494679, "grad_norm": 0.5194947719573975, "learning_rate": 0.0005, "loss": 1.3276430368423462, "step": 37261 }, { "epoch": 2.9374852187623177, "grad_norm": 0.562743604183197, "learning_rate": 0.0005, "loss": 1.4223403930664062, "step": 37262 }, { "epoch": 2.9375640520299564, "grad_norm": 0.5670101046562195, "learning_rate": 0.0005, "loss": 1.5519440174102783, "step": 37263 }, { "epoch": 2.9376428852975955, "grad_norm": 0.5026163458824158, "learning_rate": 0.0005, "loss": 1.258479118347168, "step": 37264 }, { "epoch": 2.9377217185652347, "grad_norm": 0.5007374882698059, "learning_rate": 0.0005, "loss": 1.3376970291137695, "step": 37265 }, { "epoch": 2.9378005518328734, "grad_norm": 0.5167615413665771, "learning_rate": 0.0005, "loss": 1.4032163619995117, "step": 37266 }, { "epoch": 2.9378793851005125, "grad_norm": 0.4819912910461426, "learning_rate": 0.0005, "loss": 1.1509343385696411, "step": 37267 }, { "epoch": 2.937958218368151, "grad_norm": 0.5651017427444458, "learning_rate": 0.0005, "loss": 1.196603536605835, "step": 37268 }, { "epoch": 2.9380370516357903, "grad_norm": 0.4997085928916931, "learning_rate": 0.0005, "loss": 1.4755840301513672, "step": 37269 }, { "epoch": 2.9381158849034295, "grad_norm": 0.5289938449859619, "learning_rate": 0.0005, "loss": 1.4457303285598755, "step": 37270 }, { "epoch": 2.938194718171068, "grad_norm": 0.5424359440803528, "learning_rate": 0.0005, "loss": 1.387644648551941, "step": 37271 }, { "epoch": 2.9382735514387073, "grad_norm": 0.5711007714271545, "learning_rate": 0.0005, "loss": 1.4095004796981812, "step": 37272 }, { "epoch": 2.938352384706346, "grad_norm": 0.45977655053138733, "learning_rate": 0.0005, "loss": 1.1918939352035522, "step": 37273 }, { "epoch": 2.938431217973985, "grad_norm": 0.3833836317062378, "learning_rate": 0.0005, "loss": 1.3324625492095947, "step": 37274 }, { "epoch": 2.9385100512416242, "grad_norm": 0.4797762334346771, "learning_rate": 0.0005, "loss": 1.4263827800750732, "step": 37275 }, { "epoch": 2.938588884509263, "grad_norm": 0.4633890688419342, "learning_rate": 0.0005, "loss": 1.4317946434020996, "step": 37276 }, { "epoch": 2.9386677177769016, "grad_norm": 0.36630216240882874, "learning_rate": 0.0005, "loss": 1.1407506465911865, "step": 37277 }, { "epoch": 2.9387465510445407, "grad_norm": 0.6737692356109619, "learning_rate": 0.0005, "loss": 1.4711636304855347, "step": 37278 }, { "epoch": 2.93882538431218, "grad_norm": 0.4593823552131653, "learning_rate": 0.0005, "loss": 1.1304240226745605, "step": 37279 }, { "epoch": 2.9389042175798186, "grad_norm": 0.4337157905101776, "learning_rate": 0.0005, "loss": 1.2139959335327148, "step": 37280 }, { "epoch": 2.9389830508474577, "grad_norm": 0.5035087466239929, "learning_rate": 0.0005, "loss": 1.3060740232467651, "step": 37281 }, { "epoch": 2.9390618841150964, "grad_norm": 0.4108577072620392, "learning_rate": 0.0005, "loss": 1.1997578144073486, "step": 37282 }, { "epoch": 2.9391407173827355, "grad_norm": 0.6044961810112, "learning_rate": 0.0005, "loss": 1.4436753988265991, "step": 37283 }, { "epoch": 2.9392195506503747, "grad_norm": 0.4414765238761902, "learning_rate": 0.0005, "loss": 0.973490834236145, "step": 37284 }, { "epoch": 2.9392983839180133, "grad_norm": 0.5507776141166687, "learning_rate": 0.0005, "loss": 1.2415766716003418, "step": 37285 }, { "epoch": 2.9393772171856525, "grad_norm": 0.5002087354660034, "learning_rate": 0.0005, "loss": 1.4066495895385742, "step": 37286 }, { "epoch": 2.939456050453291, "grad_norm": 0.523774266242981, "learning_rate": 0.0005, "loss": 1.3485664129257202, "step": 37287 }, { "epoch": 2.9395348837209303, "grad_norm": 0.5936563611030579, "learning_rate": 0.0005, "loss": 1.457406759262085, "step": 37288 }, { "epoch": 2.9396137169885694, "grad_norm": 0.4211646318435669, "learning_rate": 0.0005, "loss": 1.18385648727417, "step": 37289 }, { "epoch": 2.939692550256208, "grad_norm": 0.4583413004875183, "learning_rate": 0.0005, "loss": 1.264534831047058, "step": 37290 }, { "epoch": 2.939771383523847, "grad_norm": 0.505450963973999, "learning_rate": 0.0005, "loss": 1.3709771633148193, "step": 37291 }, { "epoch": 2.939850216791486, "grad_norm": 0.5570904612541199, "learning_rate": 0.0005, "loss": 1.4581880569458008, "step": 37292 }, { "epoch": 2.939929050059125, "grad_norm": 0.4992079436779022, "learning_rate": 0.0005, "loss": 1.387450098991394, "step": 37293 }, { "epoch": 2.9400078833267638, "grad_norm": 0.414596289396286, "learning_rate": 0.0005, "loss": 1.1987754106521606, "step": 37294 }, { "epoch": 2.940086716594403, "grad_norm": 0.5673440098762512, "learning_rate": 0.0005, "loss": 1.288353681564331, "step": 37295 }, { "epoch": 2.9401655498620416, "grad_norm": 0.4239142835140228, "learning_rate": 0.0005, "loss": 1.2206521034240723, "step": 37296 }, { "epoch": 2.9402443831296807, "grad_norm": 0.4018714725971222, "learning_rate": 0.0005, "loss": 1.1696977615356445, "step": 37297 }, { "epoch": 2.94032321639732, "grad_norm": 0.50206458568573, "learning_rate": 0.0005, "loss": 1.6673221588134766, "step": 37298 }, { "epoch": 2.9404020496649586, "grad_norm": 0.5165249705314636, "learning_rate": 0.0005, "loss": 1.4486415386199951, "step": 37299 }, { "epoch": 2.9404808829325977, "grad_norm": 0.683713436126709, "learning_rate": 0.0005, "loss": 1.2902288436889648, "step": 37300 }, { "epoch": 2.9405597162002364, "grad_norm": 0.41535353660583496, "learning_rate": 0.0005, "loss": 1.3106694221496582, "step": 37301 }, { "epoch": 2.9406385494678755, "grad_norm": 0.44121885299682617, "learning_rate": 0.0005, "loss": 1.1084623336791992, "step": 37302 }, { "epoch": 2.9407173827355146, "grad_norm": 0.39966607093811035, "learning_rate": 0.0005, "loss": 1.2902989387512207, "step": 37303 }, { "epoch": 2.9407962160031533, "grad_norm": 0.4655057489871979, "learning_rate": 0.0005, "loss": 1.2025065422058105, "step": 37304 }, { "epoch": 2.940875049270792, "grad_norm": 0.5313922762870789, "learning_rate": 0.0005, "loss": 1.451745629310608, "step": 37305 }, { "epoch": 2.940953882538431, "grad_norm": 0.5507072806358337, "learning_rate": 0.0005, "loss": 1.3111181259155273, "step": 37306 }, { "epoch": 2.9410327158060703, "grad_norm": 0.4707888960838318, "learning_rate": 0.0005, "loss": 1.235013723373413, "step": 37307 }, { "epoch": 2.941111549073709, "grad_norm": 0.6927098035812378, "learning_rate": 0.0005, "loss": 1.3720357418060303, "step": 37308 }, { "epoch": 2.941190382341348, "grad_norm": 0.4192322790622711, "learning_rate": 0.0005, "loss": 1.145310640335083, "step": 37309 }, { "epoch": 2.941269215608987, "grad_norm": 0.499160498380661, "learning_rate": 0.0005, "loss": 1.2805485725402832, "step": 37310 }, { "epoch": 2.941348048876626, "grad_norm": 0.4531072676181793, "learning_rate": 0.0005, "loss": 1.1937220096588135, "step": 37311 }, { "epoch": 2.941426882144265, "grad_norm": 0.7548674941062927, "learning_rate": 0.0005, "loss": 1.4198501110076904, "step": 37312 }, { "epoch": 2.9415057154119038, "grad_norm": 0.5393959879875183, "learning_rate": 0.0005, "loss": 1.3425391912460327, "step": 37313 }, { "epoch": 2.941584548679543, "grad_norm": 0.5367763638496399, "learning_rate": 0.0005, "loss": 1.2632627487182617, "step": 37314 }, { "epoch": 2.9416633819471816, "grad_norm": 0.4883790910243988, "learning_rate": 0.0005, "loss": 1.5032825469970703, "step": 37315 }, { "epoch": 2.9417422152148207, "grad_norm": 0.4800208508968353, "learning_rate": 0.0005, "loss": 1.2107884883880615, "step": 37316 }, { "epoch": 2.94182104848246, "grad_norm": 0.5695030689239502, "learning_rate": 0.0005, "loss": 1.3228905200958252, "step": 37317 }, { "epoch": 2.9418998817500985, "grad_norm": 0.6881243586540222, "learning_rate": 0.0005, "loss": 1.4396932125091553, "step": 37318 }, { "epoch": 2.9419787150177372, "grad_norm": 0.5236849784851074, "learning_rate": 0.0005, "loss": 1.3376188278198242, "step": 37319 }, { "epoch": 2.9420575482853764, "grad_norm": 0.5536001324653625, "learning_rate": 0.0005, "loss": 1.2922720909118652, "step": 37320 }, { "epoch": 2.9421363815530155, "grad_norm": 1.0547009706497192, "learning_rate": 0.0005, "loss": 1.2761411666870117, "step": 37321 }, { "epoch": 2.942215214820654, "grad_norm": 0.5573486089706421, "learning_rate": 0.0005, "loss": 1.449711799621582, "step": 37322 }, { "epoch": 2.9422940480882933, "grad_norm": 0.6278600692749023, "learning_rate": 0.0005, "loss": 1.1874284744262695, "step": 37323 }, { "epoch": 2.942372881355932, "grad_norm": 0.6254339814186096, "learning_rate": 0.0005, "loss": 1.3187854290008545, "step": 37324 }, { "epoch": 2.942451714623571, "grad_norm": 0.5677266716957092, "learning_rate": 0.0005, "loss": 1.4411993026733398, "step": 37325 }, { "epoch": 2.9425305478912103, "grad_norm": 0.47634273767471313, "learning_rate": 0.0005, "loss": 1.2458879947662354, "step": 37326 }, { "epoch": 2.942609381158849, "grad_norm": 0.41006627678871155, "learning_rate": 0.0005, "loss": 1.2643754482269287, "step": 37327 }, { "epoch": 2.942688214426488, "grad_norm": 0.5283365249633789, "learning_rate": 0.0005, "loss": 1.1901655197143555, "step": 37328 }, { "epoch": 2.942767047694127, "grad_norm": 0.5543285608291626, "learning_rate": 0.0005, "loss": 1.4768658876419067, "step": 37329 }, { "epoch": 2.942845880961766, "grad_norm": 0.5536456108093262, "learning_rate": 0.0005, "loss": 1.2691516876220703, "step": 37330 }, { "epoch": 2.942924714229405, "grad_norm": 0.4218357503414154, "learning_rate": 0.0005, "loss": 1.063331961631775, "step": 37331 }, { "epoch": 2.9430035474970437, "grad_norm": 0.6240367889404297, "learning_rate": 0.0005, "loss": 1.3727160692214966, "step": 37332 }, { "epoch": 2.9430823807646824, "grad_norm": 0.45592930912971497, "learning_rate": 0.0005, "loss": 1.20904541015625, "step": 37333 }, { "epoch": 2.9431612140323216, "grad_norm": 0.4786480665206909, "learning_rate": 0.0005, "loss": 1.3287895917892456, "step": 37334 }, { "epoch": 2.9432400472999607, "grad_norm": 0.3963547646999359, "learning_rate": 0.0005, "loss": 1.143425464630127, "step": 37335 }, { "epoch": 2.9433188805675994, "grad_norm": 0.4451666474342346, "learning_rate": 0.0005, "loss": 1.1749613285064697, "step": 37336 }, { "epoch": 2.9433977138352385, "grad_norm": 0.5362713932991028, "learning_rate": 0.0005, "loss": 1.3392105102539062, "step": 37337 }, { "epoch": 2.943476547102877, "grad_norm": 0.4568313658237457, "learning_rate": 0.0005, "loss": 1.3864154815673828, "step": 37338 }, { "epoch": 2.9435553803705163, "grad_norm": 0.5087440609931946, "learning_rate": 0.0005, "loss": 1.2655290365219116, "step": 37339 }, { "epoch": 2.9436342136381555, "grad_norm": 0.4780508875846863, "learning_rate": 0.0005, "loss": 1.429150938987732, "step": 37340 }, { "epoch": 2.943713046905794, "grad_norm": 0.48023781180381775, "learning_rate": 0.0005, "loss": 1.2494919300079346, "step": 37341 }, { "epoch": 2.9437918801734333, "grad_norm": 0.491788387298584, "learning_rate": 0.0005, "loss": 1.2513926029205322, "step": 37342 }, { "epoch": 2.943870713441072, "grad_norm": 0.4276345670223236, "learning_rate": 0.0005, "loss": 1.275623083114624, "step": 37343 }, { "epoch": 2.943949546708711, "grad_norm": 0.41305404901504517, "learning_rate": 0.0005, "loss": 1.2342631816864014, "step": 37344 }, { "epoch": 2.9440283799763503, "grad_norm": 0.6731377243995667, "learning_rate": 0.0005, "loss": 1.590284824371338, "step": 37345 }, { "epoch": 2.944107213243989, "grad_norm": 0.452616423368454, "learning_rate": 0.0005, "loss": 1.2011456489562988, "step": 37346 }, { "epoch": 2.9441860465116276, "grad_norm": 0.5502145290374756, "learning_rate": 0.0005, "loss": 1.412003755569458, "step": 37347 }, { "epoch": 2.9442648797792668, "grad_norm": 0.5758440494537354, "learning_rate": 0.0005, "loss": 1.3769887685775757, "step": 37348 }, { "epoch": 2.944343713046906, "grad_norm": 0.5143270492553711, "learning_rate": 0.0005, "loss": 1.2895852327346802, "step": 37349 }, { "epoch": 2.9444225463145446, "grad_norm": 0.5219380259513855, "learning_rate": 0.0005, "loss": 1.5858044624328613, "step": 37350 }, { "epoch": 2.9445013795821837, "grad_norm": 0.41267508268356323, "learning_rate": 0.0005, "loss": 1.1864606142044067, "step": 37351 }, { "epoch": 2.9445802128498224, "grad_norm": 0.4928577244281769, "learning_rate": 0.0005, "loss": 1.1565945148468018, "step": 37352 }, { "epoch": 2.9446590461174615, "grad_norm": 0.47518908977508545, "learning_rate": 0.0005, "loss": 1.1715219020843506, "step": 37353 }, { "epoch": 2.9447378793851007, "grad_norm": 0.6083152294158936, "learning_rate": 0.0005, "loss": 1.4874141216278076, "step": 37354 }, { "epoch": 2.9448167126527394, "grad_norm": 0.5362884998321533, "learning_rate": 0.0005, "loss": 1.2389925718307495, "step": 37355 }, { "epoch": 2.9448955459203785, "grad_norm": 0.3574358820915222, "learning_rate": 0.0005, "loss": 1.0709296464920044, "step": 37356 }, { "epoch": 2.944974379188017, "grad_norm": 0.4180741012096405, "learning_rate": 0.0005, "loss": 1.1533870697021484, "step": 37357 }, { "epoch": 2.9450532124556563, "grad_norm": 0.4491770267486572, "learning_rate": 0.0005, "loss": 1.2517094612121582, "step": 37358 }, { "epoch": 2.9451320457232955, "grad_norm": 0.6157420873641968, "learning_rate": 0.0005, "loss": 1.330085277557373, "step": 37359 }, { "epoch": 2.945210878990934, "grad_norm": 0.42452189326286316, "learning_rate": 0.0005, "loss": 1.1199226379394531, "step": 37360 }, { "epoch": 2.9452897122585733, "grad_norm": 0.5665767192840576, "learning_rate": 0.0005, "loss": 1.464888572692871, "step": 37361 }, { "epoch": 2.945368545526212, "grad_norm": 0.45917484164237976, "learning_rate": 0.0005, "loss": 1.4479734897613525, "step": 37362 }, { "epoch": 2.945447378793851, "grad_norm": 0.42378687858581543, "learning_rate": 0.0005, "loss": 0.9761061668395996, "step": 37363 }, { "epoch": 2.94552621206149, "grad_norm": 0.7167550325393677, "learning_rate": 0.0005, "loss": 1.6667089462280273, "step": 37364 }, { "epoch": 2.945605045329129, "grad_norm": 0.604599118232727, "learning_rate": 0.0005, "loss": 1.273695468902588, "step": 37365 }, { "epoch": 2.9456838785967676, "grad_norm": 0.7242280840873718, "learning_rate": 0.0005, "loss": 1.2879858016967773, "step": 37366 }, { "epoch": 2.9457627118644067, "grad_norm": 1.0163788795471191, "learning_rate": 0.0005, "loss": 1.4820029735565186, "step": 37367 }, { "epoch": 2.945841545132046, "grad_norm": 0.7148561477661133, "learning_rate": 0.0005, "loss": 1.4874151945114136, "step": 37368 }, { "epoch": 2.9459203783996846, "grad_norm": 0.6411413550376892, "learning_rate": 0.0005, "loss": 1.4482214450836182, "step": 37369 }, { "epoch": 2.9459992116673237, "grad_norm": 1.4280389547348022, "learning_rate": 0.0005, "loss": 1.2512390613555908, "step": 37370 }, { "epoch": 2.9460780449349624, "grad_norm": 0.5475523471832275, "learning_rate": 0.0005, "loss": 1.2106449604034424, "step": 37371 }, { "epoch": 2.9461568782026015, "grad_norm": 0.5527087450027466, "learning_rate": 0.0005, "loss": 1.362572431564331, "step": 37372 }, { "epoch": 2.9462357114702407, "grad_norm": 0.5574802160263062, "learning_rate": 0.0005, "loss": 1.2901034355163574, "step": 37373 }, { "epoch": 2.9463145447378793, "grad_norm": 0.4060367941856384, "learning_rate": 0.0005, "loss": 1.1642130613327026, "step": 37374 }, { "epoch": 2.9463933780055185, "grad_norm": 0.5836485624313354, "learning_rate": 0.0005, "loss": 1.414128065109253, "step": 37375 }, { "epoch": 2.946472211273157, "grad_norm": 0.45581740140914917, "learning_rate": 0.0005, "loss": 1.344228982925415, "step": 37376 }, { "epoch": 2.9465510445407963, "grad_norm": 0.8542118072509766, "learning_rate": 0.0005, "loss": 1.3339734077453613, "step": 37377 }, { "epoch": 2.946629877808435, "grad_norm": 0.6380466818809509, "learning_rate": 0.0005, "loss": 1.2847168445587158, "step": 37378 }, { "epoch": 2.946708711076074, "grad_norm": 0.48567408323287964, "learning_rate": 0.0005, "loss": 1.2441023588180542, "step": 37379 }, { "epoch": 2.946787544343713, "grad_norm": 0.6125728487968445, "learning_rate": 0.0005, "loss": 1.506967306137085, "step": 37380 }, { "epoch": 2.946866377611352, "grad_norm": 0.4617007374763489, "learning_rate": 0.0005, "loss": 1.2384307384490967, "step": 37381 }, { "epoch": 2.946945210878991, "grad_norm": 0.48827630281448364, "learning_rate": 0.0005, "loss": 1.3650455474853516, "step": 37382 }, { "epoch": 2.9470240441466298, "grad_norm": 0.5236866474151611, "learning_rate": 0.0005, "loss": 1.3770148754119873, "step": 37383 }, { "epoch": 2.947102877414269, "grad_norm": 0.5258287191390991, "learning_rate": 0.0005, "loss": 1.391974687576294, "step": 37384 }, { "epoch": 2.9471817106819076, "grad_norm": 0.476532906293869, "learning_rate": 0.0005, "loss": 1.2945144176483154, "step": 37385 }, { "epoch": 2.9472605439495467, "grad_norm": 0.4436320960521698, "learning_rate": 0.0005, "loss": 1.0755980014801025, "step": 37386 }, { "epoch": 2.947339377217186, "grad_norm": 0.46746546030044556, "learning_rate": 0.0005, "loss": 1.152419090270996, "step": 37387 }, { "epoch": 2.9474182104848246, "grad_norm": 0.48206454515457153, "learning_rate": 0.0005, "loss": 1.4125816822052002, "step": 37388 }, { "epoch": 2.9474970437524637, "grad_norm": 0.4768860638141632, "learning_rate": 0.0005, "loss": 1.2134737968444824, "step": 37389 }, { "epoch": 2.9475758770201024, "grad_norm": 0.5716037154197693, "learning_rate": 0.0005, "loss": 1.2596725225448608, "step": 37390 }, { "epoch": 2.9476547102877415, "grad_norm": 0.454606294631958, "learning_rate": 0.0005, "loss": 1.3085124492645264, "step": 37391 }, { "epoch": 2.9477335435553806, "grad_norm": 0.8186169266700745, "learning_rate": 0.0005, "loss": 1.3204927444458008, "step": 37392 }, { "epoch": 2.9478123768230193, "grad_norm": 0.44064345955848694, "learning_rate": 0.0005, "loss": 1.266257643699646, "step": 37393 }, { "epoch": 2.947891210090658, "grad_norm": 0.7768656611442566, "learning_rate": 0.0005, "loss": 1.2151174545288086, "step": 37394 }, { "epoch": 2.947970043358297, "grad_norm": 0.7434076070785522, "learning_rate": 0.0005, "loss": 1.5391809940338135, "step": 37395 }, { "epoch": 2.9480488766259363, "grad_norm": 0.6512233018875122, "learning_rate": 0.0005, "loss": 1.451873540878296, "step": 37396 }, { "epoch": 2.948127709893575, "grad_norm": 0.5657356977462769, "learning_rate": 0.0005, "loss": 1.2430559396743774, "step": 37397 }, { "epoch": 2.948206543161214, "grad_norm": 1.1908899545669556, "learning_rate": 0.0005, "loss": 1.3917789459228516, "step": 37398 }, { "epoch": 2.948285376428853, "grad_norm": 0.5310236811637878, "learning_rate": 0.0005, "loss": 1.4186328649520874, "step": 37399 }, { "epoch": 2.948364209696492, "grad_norm": 0.46336570382118225, "learning_rate": 0.0005, "loss": 1.235318660736084, "step": 37400 }, { "epoch": 2.948443042964131, "grad_norm": 0.6009505391120911, "learning_rate": 0.0005, "loss": 1.4374868869781494, "step": 37401 }, { "epoch": 2.9485218762317698, "grad_norm": 0.4597631096839905, "learning_rate": 0.0005, "loss": 1.2811284065246582, "step": 37402 }, { "epoch": 2.948600709499409, "grad_norm": 0.5913975834846497, "learning_rate": 0.0005, "loss": 1.4739420413970947, "step": 37403 }, { "epoch": 2.9486795427670476, "grad_norm": 0.574256956577301, "learning_rate": 0.0005, "loss": 1.46928071975708, "step": 37404 }, { "epoch": 2.9487583760346867, "grad_norm": 0.8499171137809753, "learning_rate": 0.0005, "loss": 1.482170581817627, "step": 37405 }, { "epoch": 2.948837209302326, "grad_norm": 0.5103341937065125, "learning_rate": 0.0005, "loss": 1.4429423809051514, "step": 37406 }, { "epoch": 2.9489160425699645, "grad_norm": 0.5639970898628235, "learning_rate": 0.0005, "loss": 1.4667341709136963, "step": 37407 }, { "epoch": 2.9489948758376032, "grad_norm": 0.7564717531204224, "learning_rate": 0.0005, "loss": 1.3598792552947998, "step": 37408 }, { "epoch": 2.9490737091052424, "grad_norm": 0.5642161965370178, "learning_rate": 0.0005, "loss": 1.4645088911056519, "step": 37409 }, { "epoch": 2.9491525423728815, "grad_norm": 0.5715333819389343, "learning_rate": 0.0005, "loss": 1.3884756565093994, "step": 37410 }, { "epoch": 2.94923137564052, "grad_norm": 0.5573865175247192, "learning_rate": 0.0005, "loss": 1.3527565002441406, "step": 37411 }, { "epoch": 2.9493102089081593, "grad_norm": 0.4196464419364929, "learning_rate": 0.0005, "loss": 1.1481571197509766, "step": 37412 }, { "epoch": 2.949389042175798, "grad_norm": 0.6818985342979431, "learning_rate": 0.0005, "loss": 1.3697253465652466, "step": 37413 }, { "epoch": 2.949467875443437, "grad_norm": 0.6059028506278992, "learning_rate": 0.0005, "loss": 1.602503776550293, "step": 37414 }, { "epoch": 2.9495467087110763, "grad_norm": 0.42307862639427185, "learning_rate": 0.0005, "loss": 1.1244151592254639, "step": 37415 }, { "epoch": 2.949625541978715, "grad_norm": 0.6218199133872986, "learning_rate": 0.0005, "loss": 1.4741579294204712, "step": 37416 }, { "epoch": 2.949704375246354, "grad_norm": 0.46253347396850586, "learning_rate": 0.0005, "loss": 1.1452633142471313, "step": 37417 }, { "epoch": 2.949783208513993, "grad_norm": 0.47173357009887695, "learning_rate": 0.0005, "loss": 1.0952162742614746, "step": 37418 }, { "epoch": 2.949862041781632, "grad_norm": 0.4744516909122467, "learning_rate": 0.0005, "loss": 1.3487554788589478, "step": 37419 }, { "epoch": 2.949940875049271, "grad_norm": 0.4851638376712799, "learning_rate": 0.0005, "loss": 1.1699343919754028, "step": 37420 }, { "epoch": 2.9500197083169097, "grad_norm": 0.4282307028770447, "learning_rate": 0.0005, "loss": 1.2570137977600098, "step": 37421 }, { "epoch": 2.9500985415845484, "grad_norm": 0.4650927782058716, "learning_rate": 0.0005, "loss": 1.2709461450576782, "step": 37422 }, { "epoch": 2.9501773748521876, "grad_norm": 0.38101330399513245, "learning_rate": 0.0005, "loss": 1.1282175779342651, "step": 37423 }, { "epoch": 2.9502562081198267, "grad_norm": 0.7132886648178101, "learning_rate": 0.0005, "loss": 1.2346417903900146, "step": 37424 }, { "epoch": 2.9503350413874654, "grad_norm": 0.5160473585128784, "learning_rate": 0.0005, "loss": 1.4729714393615723, "step": 37425 }, { "epoch": 2.9504138746551045, "grad_norm": 0.467856228351593, "learning_rate": 0.0005, "loss": 1.0968307256698608, "step": 37426 }, { "epoch": 2.950492707922743, "grad_norm": 0.5563046932220459, "learning_rate": 0.0005, "loss": 1.4034104347229004, "step": 37427 }, { "epoch": 2.9505715411903823, "grad_norm": 0.820339024066925, "learning_rate": 0.0005, "loss": 1.280906081199646, "step": 37428 }, { "epoch": 2.9506503744580215, "grad_norm": 0.48547884821891785, "learning_rate": 0.0005, "loss": 1.2223674058914185, "step": 37429 }, { "epoch": 2.95072920772566, "grad_norm": 0.6238885521888733, "learning_rate": 0.0005, "loss": 1.3865808248519897, "step": 37430 }, { "epoch": 2.9508080409932993, "grad_norm": 0.49181145429611206, "learning_rate": 0.0005, "loss": 1.4039738178253174, "step": 37431 }, { "epoch": 2.950886874260938, "grad_norm": 0.45766133069992065, "learning_rate": 0.0005, "loss": 1.277606725692749, "step": 37432 }, { "epoch": 2.950965707528577, "grad_norm": 0.5438012480735779, "learning_rate": 0.0005, "loss": 1.3881261348724365, "step": 37433 }, { "epoch": 2.9510445407962163, "grad_norm": 0.4741272032260895, "learning_rate": 0.0005, "loss": 1.3084379434585571, "step": 37434 }, { "epoch": 2.951123374063855, "grad_norm": 0.7338312268257141, "learning_rate": 0.0005, "loss": 1.3932673931121826, "step": 37435 }, { "epoch": 2.9512022073314936, "grad_norm": 0.5522447824478149, "learning_rate": 0.0005, "loss": 1.5112507343292236, "step": 37436 }, { "epoch": 2.9512810405991328, "grad_norm": 0.4658310115337372, "learning_rate": 0.0005, "loss": 1.3925273418426514, "step": 37437 }, { "epoch": 2.951359873866772, "grad_norm": 0.5233344435691833, "learning_rate": 0.0005, "loss": 1.328052043914795, "step": 37438 }, { "epoch": 2.9514387071344106, "grad_norm": 0.5720542073249817, "learning_rate": 0.0005, "loss": 1.3204007148742676, "step": 37439 }, { "epoch": 2.9515175404020497, "grad_norm": 0.48296985030174255, "learning_rate": 0.0005, "loss": 1.3741672039031982, "step": 37440 }, { "epoch": 2.9515963736696884, "grad_norm": 0.510893702507019, "learning_rate": 0.0005, "loss": 1.3788714408874512, "step": 37441 }, { "epoch": 2.9516752069373275, "grad_norm": 0.5298483967781067, "learning_rate": 0.0005, "loss": 1.431459903717041, "step": 37442 }, { "epoch": 2.9517540402049667, "grad_norm": 0.4158026874065399, "learning_rate": 0.0005, "loss": 1.3675659894943237, "step": 37443 }, { "epoch": 2.9518328734726054, "grad_norm": 0.46432575583457947, "learning_rate": 0.0005, "loss": 1.2949862480163574, "step": 37444 }, { "epoch": 2.9519117067402445, "grad_norm": 0.5618082880973816, "learning_rate": 0.0005, "loss": 1.4661016464233398, "step": 37445 }, { "epoch": 2.951990540007883, "grad_norm": 0.43128523230552673, "learning_rate": 0.0005, "loss": 1.315540075302124, "step": 37446 }, { "epoch": 2.9520693732755223, "grad_norm": 0.5110418200492859, "learning_rate": 0.0005, "loss": 1.3451242446899414, "step": 37447 }, { "epoch": 2.9521482065431615, "grad_norm": 0.3823510408401489, "learning_rate": 0.0005, "loss": 1.2266143560409546, "step": 37448 }, { "epoch": 2.9522270398108, "grad_norm": 0.428175687789917, "learning_rate": 0.0005, "loss": 1.335517168045044, "step": 37449 }, { "epoch": 2.952305873078439, "grad_norm": 0.48680081963539124, "learning_rate": 0.0005, "loss": 1.400116205215454, "step": 37450 }, { "epoch": 2.952384706346078, "grad_norm": 0.45552244782447815, "learning_rate": 0.0005, "loss": 1.2104206085205078, "step": 37451 }, { "epoch": 2.952463539613717, "grad_norm": 0.48932376503944397, "learning_rate": 0.0005, "loss": 1.2652209997177124, "step": 37452 }, { "epoch": 2.952542372881356, "grad_norm": 0.45878851413726807, "learning_rate": 0.0005, "loss": 1.3805954456329346, "step": 37453 }, { "epoch": 2.952621206148995, "grad_norm": 0.45610690116882324, "learning_rate": 0.0005, "loss": 1.4261976480484009, "step": 37454 }, { "epoch": 2.9527000394166336, "grad_norm": 0.5058733820915222, "learning_rate": 0.0005, "loss": 1.3841984272003174, "step": 37455 }, { "epoch": 2.9527788726842727, "grad_norm": 0.49835067987442017, "learning_rate": 0.0005, "loss": 1.3042590618133545, "step": 37456 }, { "epoch": 2.952857705951912, "grad_norm": 0.5331882834434509, "learning_rate": 0.0005, "loss": 1.4996659755706787, "step": 37457 }, { "epoch": 2.9529365392195506, "grad_norm": 0.5337074398994446, "learning_rate": 0.0005, "loss": 1.2825326919555664, "step": 37458 }, { "epoch": 2.9530153724871897, "grad_norm": 0.5359159708023071, "learning_rate": 0.0005, "loss": 1.650681495666504, "step": 37459 }, { "epoch": 2.9530942057548284, "grad_norm": 0.3973054587841034, "learning_rate": 0.0005, "loss": 1.2694669961929321, "step": 37460 }, { "epoch": 2.9531730390224675, "grad_norm": 0.4487951099872589, "learning_rate": 0.0005, "loss": 1.34037446975708, "step": 37461 }, { "epoch": 2.9532518722901067, "grad_norm": 0.46024322509765625, "learning_rate": 0.0005, "loss": 1.4234286546707153, "step": 37462 }, { "epoch": 2.9533307055577454, "grad_norm": 0.6095305681228638, "learning_rate": 0.0005, "loss": 1.4277534484863281, "step": 37463 }, { "epoch": 2.953409538825384, "grad_norm": 0.4548524022102356, "learning_rate": 0.0005, "loss": 1.3266310691833496, "step": 37464 }, { "epoch": 2.953488372093023, "grad_norm": 0.5047696232795715, "learning_rate": 0.0005, "loss": 1.4144788980484009, "step": 37465 }, { "epoch": 2.9535672053606623, "grad_norm": 0.45488956570625305, "learning_rate": 0.0005, "loss": 1.1749536991119385, "step": 37466 }, { "epoch": 2.953646038628301, "grad_norm": 0.45124879479408264, "learning_rate": 0.0005, "loss": 1.4005401134490967, "step": 37467 }, { "epoch": 2.95372487189594, "grad_norm": 0.44088855385780334, "learning_rate": 0.0005, "loss": 1.4179706573486328, "step": 37468 }, { "epoch": 2.953803705163579, "grad_norm": 0.4343821704387665, "learning_rate": 0.0005, "loss": 1.260789155960083, "step": 37469 }, { "epoch": 2.953882538431218, "grad_norm": 0.4676319658756256, "learning_rate": 0.0005, "loss": 1.1696839332580566, "step": 37470 }, { "epoch": 2.953961371698857, "grad_norm": 0.6042037010192871, "learning_rate": 0.0005, "loss": 1.4078165292739868, "step": 37471 }, { "epoch": 2.9540402049664958, "grad_norm": 0.47551071643829346, "learning_rate": 0.0005, "loss": 1.3577101230621338, "step": 37472 }, { "epoch": 2.954119038234135, "grad_norm": 0.39495691657066345, "learning_rate": 0.0005, "loss": 1.2548131942749023, "step": 37473 }, { "epoch": 2.9541978715017736, "grad_norm": 0.5649358630180359, "learning_rate": 0.0005, "loss": 1.355931043624878, "step": 37474 }, { "epoch": 2.9542767047694127, "grad_norm": 0.40831196308135986, "learning_rate": 0.0005, "loss": 1.093292236328125, "step": 37475 }, { "epoch": 2.954355538037052, "grad_norm": 0.5712772011756897, "learning_rate": 0.0005, "loss": 1.3798024654388428, "step": 37476 }, { "epoch": 2.9544343713046906, "grad_norm": 0.5072566866874695, "learning_rate": 0.0005, "loss": 1.3751342296600342, "step": 37477 }, { "epoch": 2.9545132045723297, "grad_norm": 0.4863387942314148, "learning_rate": 0.0005, "loss": 1.1323856115341187, "step": 37478 }, { "epoch": 2.9545920378399684, "grad_norm": 0.5211537480354309, "learning_rate": 0.0005, "loss": 1.3831045627593994, "step": 37479 }, { "epoch": 2.9546708711076075, "grad_norm": 0.5195642709732056, "learning_rate": 0.0005, "loss": 1.1793631315231323, "step": 37480 }, { "epoch": 2.954749704375246, "grad_norm": 0.5762971639633179, "learning_rate": 0.0005, "loss": 1.3024158477783203, "step": 37481 }, { "epoch": 2.9548285376428853, "grad_norm": 0.5030107498168945, "learning_rate": 0.0005, "loss": 1.3274719715118408, "step": 37482 }, { "epoch": 2.954907370910524, "grad_norm": 0.496528297662735, "learning_rate": 0.0005, "loss": 1.1682072877883911, "step": 37483 }, { "epoch": 2.954986204178163, "grad_norm": 0.6986111402511597, "learning_rate": 0.0005, "loss": 1.4584360122680664, "step": 37484 }, { "epoch": 2.9550650374458023, "grad_norm": 0.5443312525749207, "learning_rate": 0.0005, "loss": 1.361884593963623, "step": 37485 }, { "epoch": 2.955143870713441, "grad_norm": 0.45022058486938477, "learning_rate": 0.0005, "loss": 1.3639423847198486, "step": 37486 }, { "epoch": 2.95522270398108, "grad_norm": 0.47169461846351624, "learning_rate": 0.0005, "loss": 1.0375516414642334, "step": 37487 }, { "epoch": 2.955301537248719, "grad_norm": 0.5437594056129456, "learning_rate": 0.0005, "loss": 1.4697740077972412, "step": 37488 }, { "epoch": 2.955380370516358, "grad_norm": 0.5026265978813171, "learning_rate": 0.0005, "loss": 1.6267738342285156, "step": 37489 }, { "epoch": 2.955459203783997, "grad_norm": 0.49737992882728577, "learning_rate": 0.0005, "loss": 1.3300964832305908, "step": 37490 }, { "epoch": 2.9555380370516358, "grad_norm": 0.47611790895462036, "learning_rate": 0.0005, "loss": 1.4997625350952148, "step": 37491 }, { "epoch": 2.955616870319275, "grad_norm": 0.5251145958900452, "learning_rate": 0.0005, "loss": 1.1970748901367188, "step": 37492 }, { "epoch": 2.9556957035869136, "grad_norm": 0.4799356758594513, "learning_rate": 0.0005, "loss": 1.3454891443252563, "step": 37493 }, { "epoch": 2.9557745368545527, "grad_norm": 0.4595882296562195, "learning_rate": 0.0005, "loss": 1.109462022781372, "step": 37494 }, { "epoch": 2.9558533701221914, "grad_norm": 0.5520192980766296, "learning_rate": 0.0005, "loss": 1.511361837387085, "step": 37495 }, { "epoch": 2.9559322033898305, "grad_norm": 0.4296691119670868, "learning_rate": 0.0005, "loss": 1.3177597522735596, "step": 37496 }, { "epoch": 2.9560110366574692, "grad_norm": 0.5163315534591675, "learning_rate": 0.0005, "loss": 1.2895499467849731, "step": 37497 }, { "epoch": 2.9560898699251084, "grad_norm": 0.5067040324211121, "learning_rate": 0.0005, "loss": 1.3770934343338013, "step": 37498 }, { "epoch": 2.9561687031927475, "grad_norm": 0.45195069909095764, "learning_rate": 0.0005, "loss": 1.3609914779663086, "step": 37499 }, { "epoch": 2.956247536460386, "grad_norm": 0.5082876086235046, "learning_rate": 0.0005, "loss": 1.25264573097229, "step": 37500 }, { "epoch": 2.9563263697280253, "grad_norm": 0.5005958080291748, "learning_rate": 0.0005, "loss": 1.3917222023010254, "step": 37501 }, { "epoch": 2.956405202995664, "grad_norm": 0.42491844296455383, "learning_rate": 0.0005, "loss": 1.2532058954238892, "step": 37502 }, { "epoch": 2.956484036263303, "grad_norm": 0.8314566016197205, "learning_rate": 0.0005, "loss": 1.4840705394744873, "step": 37503 }, { "epoch": 2.9565628695309423, "grad_norm": 0.4306584298610687, "learning_rate": 0.0005, "loss": 1.1616947650909424, "step": 37504 }, { "epoch": 2.956641702798581, "grad_norm": 0.5459445118904114, "learning_rate": 0.0005, "loss": 1.3380591869354248, "step": 37505 }, { "epoch": 2.95672053606622, "grad_norm": 0.4906367063522339, "learning_rate": 0.0005, "loss": 1.260389804840088, "step": 37506 }, { "epoch": 2.956799369333859, "grad_norm": 0.5525729656219482, "learning_rate": 0.0005, "loss": 1.5897424221038818, "step": 37507 }, { "epoch": 2.956878202601498, "grad_norm": 0.5174534916877747, "learning_rate": 0.0005, "loss": 1.326830267906189, "step": 37508 }, { "epoch": 2.956957035869137, "grad_norm": 0.4670041501522064, "learning_rate": 0.0005, "loss": 1.1865432262420654, "step": 37509 }, { "epoch": 2.9570358691367757, "grad_norm": 0.5369630455970764, "learning_rate": 0.0005, "loss": 1.5996155738830566, "step": 37510 }, { "epoch": 2.9571147024044144, "grad_norm": 0.800230860710144, "learning_rate": 0.0005, "loss": 1.365663766860962, "step": 37511 }, { "epoch": 2.9571935356720536, "grad_norm": 0.5487827062606812, "learning_rate": 0.0005, "loss": 1.3720496892929077, "step": 37512 }, { "epoch": 2.9572723689396927, "grad_norm": 0.4946277141571045, "learning_rate": 0.0005, "loss": 1.2059191465377808, "step": 37513 }, { "epoch": 2.9573512022073314, "grad_norm": 0.5201175808906555, "learning_rate": 0.0005, "loss": 1.3474539518356323, "step": 37514 }, { "epoch": 2.9574300354749705, "grad_norm": 0.48577606678009033, "learning_rate": 0.0005, "loss": 1.1550745964050293, "step": 37515 }, { "epoch": 2.957508868742609, "grad_norm": 0.41794613003730774, "learning_rate": 0.0005, "loss": 1.136111855506897, "step": 37516 }, { "epoch": 2.9575877020102483, "grad_norm": 0.7090915441513062, "learning_rate": 0.0005, "loss": 1.4987900257110596, "step": 37517 }, { "epoch": 2.9576665352778875, "grad_norm": 0.647500216960907, "learning_rate": 0.0005, "loss": 1.445472002029419, "step": 37518 }, { "epoch": 2.957745368545526, "grad_norm": 0.38123300671577454, "learning_rate": 0.0005, "loss": 1.1302070617675781, "step": 37519 }, { "epoch": 2.9578242018131653, "grad_norm": 0.5769757628440857, "learning_rate": 0.0005, "loss": 1.2247982025146484, "step": 37520 }, { "epoch": 2.957903035080804, "grad_norm": 0.5506055951118469, "learning_rate": 0.0005, "loss": 1.3232389688491821, "step": 37521 }, { "epoch": 2.957981868348443, "grad_norm": 0.5073963403701782, "learning_rate": 0.0005, "loss": 1.4613440036773682, "step": 37522 }, { "epoch": 2.9580607016160823, "grad_norm": 0.48279187083244324, "learning_rate": 0.0005, "loss": 1.33658766746521, "step": 37523 }, { "epoch": 2.958139534883721, "grad_norm": 0.49487242102622986, "learning_rate": 0.0005, "loss": 1.2862008810043335, "step": 37524 }, { "epoch": 2.9582183681513596, "grad_norm": 0.4829619228839874, "learning_rate": 0.0005, "loss": 1.219534158706665, "step": 37525 }, { "epoch": 2.9582972014189988, "grad_norm": 0.54265296459198, "learning_rate": 0.0005, "loss": 1.3743396997451782, "step": 37526 }, { "epoch": 2.958376034686638, "grad_norm": 0.4510952830314636, "learning_rate": 0.0005, "loss": 1.4226722717285156, "step": 37527 }, { "epoch": 2.9584548679542766, "grad_norm": 0.5635758638381958, "learning_rate": 0.0005, "loss": 1.440593957901001, "step": 37528 }, { "epoch": 2.9585337012219157, "grad_norm": 0.5204679369926453, "learning_rate": 0.0005, "loss": 1.2313107252120972, "step": 37529 }, { "epoch": 2.9586125344895544, "grad_norm": 0.40652501583099365, "learning_rate": 0.0005, "loss": 1.1296584606170654, "step": 37530 }, { "epoch": 2.9586913677571935, "grad_norm": 0.4419559836387634, "learning_rate": 0.0005, "loss": 1.4015157222747803, "step": 37531 }, { "epoch": 2.9587702010248327, "grad_norm": 0.5015789270401001, "learning_rate": 0.0005, "loss": 1.452755093574524, "step": 37532 }, { "epoch": 2.9588490342924714, "grad_norm": 0.4488150179386139, "learning_rate": 0.0005, "loss": 1.258096694946289, "step": 37533 }, { "epoch": 2.9589278675601105, "grad_norm": 0.45325419306755066, "learning_rate": 0.0005, "loss": 1.3569746017456055, "step": 37534 }, { "epoch": 2.959006700827749, "grad_norm": 0.5125746130943298, "learning_rate": 0.0005, "loss": 1.2340763807296753, "step": 37535 }, { "epoch": 2.9590855340953883, "grad_norm": 0.5764676332473755, "learning_rate": 0.0005, "loss": 1.4079220294952393, "step": 37536 }, { "epoch": 2.9591643673630275, "grad_norm": 0.41010767221450806, "learning_rate": 0.0005, "loss": 1.2473604679107666, "step": 37537 }, { "epoch": 2.959243200630666, "grad_norm": 0.5112113356590271, "learning_rate": 0.0005, "loss": 1.3864436149597168, "step": 37538 }, { "epoch": 2.959322033898305, "grad_norm": 0.6730899214744568, "learning_rate": 0.0005, "loss": 1.4925321340560913, "step": 37539 }, { "epoch": 2.959400867165944, "grad_norm": 0.4857730269432068, "learning_rate": 0.0005, "loss": 1.1564617156982422, "step": 37540 }, { "epoch": 2.959479700433583, "grad_norm": 0.4487212002277374, "learning_rate": 0.0005, "loss": 1.158754825592041, "step": 37541 }, { "epoch": 2.959558533701222, "grad_norm": 0.5793789625167847, "learning_rate": 0.0005, "loss": 1.4903905391693115, "step": 37542 }, { "epoch": 2.959637366968861, "grad_norm": 0.4966428577899933, "learning_rate": 0.0005, "loss": 1.3803563117980957, "step": 37543 }, { "epoch": 2.9597162002364996, "grad_norm": 0.44930559396743774, "learning_rate": 0.0005, "loss": 1.2274386882781982, "step": 37544 }, { "epoch": 2.9597950335041388, "grad_norm": 0.5819132328033447, "learning_rate": 0.0005, "loss": 1.369327187538147, "step": 37545 }, { "epoch": 2.959873866771778, "grad_norm": 0.44711872935295105, "learning_rate": 0.0005, "loss": 1.261596918106079, "step": 37546 }, { "epoch": 2.9599527000394166, "grad_norm": 0.4333173632621765, "learning_rate": 0.0005, "loss": 1.2945139408111572, "step": 37547 }, { "epoch": 2.9600315333070557, "grad_norm": 0.5513483881950378, "learning_rate": 0.0005, "loss": 1.4584417343139648, "step": 37548 }, { "epoch": 2.9601103665746944, "grad_norm": 0.4678405821323395, "learning_rate": 0.0005, "loss": 1.204482078552246, "step": 37549 }, { "epoch": 2.9601891998423335, "grad_norm": 0.5558786392211914, "learning_rate": 0.0005, "loss": 1.4388014078140259, "step": 37550 }, { "epoch": 2.9602680331099727, "grad_norm": 0.5017213225364685, "learning_rate": 0.0005, "loss": 1.370842695236206, "step": 37551 }, { "epoch": 2.9603468663776114, "grad_norm": 0.4505165219306946, "learning_rate": 0.0005, "loss": 1.1995081901550293, "step": 37552 }, { "epoch": 2.96042569964525, "grad_norm": 0.446237713098526, "learning_rate": 0.0005, "loss": 1.2707438468933105, "step": 37553 }, { "epoch": 2.960504532912889, "grad_norm": 0.44917115569114685, "learning_rate": 0.0005, "loss": 1.3467936515808105, "step": 37554 }, { "epoch": 2.9605833661805283, "grad_norm": 0.49480992555618286, "learning_rate": 0.0005, "loss": 1.3554837703704834, "step": 37555 }, { "epoch": 2.960662199448167, "grad_norm": 0.6489288210868835, "learning_rate": 0.0005, "loss": 1.3410327434539795, "step": 37556 }, { "epoch": 2.960741032715806, "grad_norm": 0.5322710871696472, "learning_rate": 0.0005, "loss": 1.417337417602539, "step": 37557 }, { "epoch": 2.960819865983445, "grad_norm": 0.530937671661377, "learning_rate": 0.0005, "loss": 1.469214916229248, "step": 37558 }, { "epoch": 2.960898699251084, "grad_norm": 0.4175611138343811, "learning_rate": 0.0005, "loss": 1.1502048969268799, "step": 37559 }, { "epoch": 2.960977532518723, "grad_norm": 0.47766968607902527, "learning_rate": 0.0005, "loss": 1.388993501663208, "step": 37560 }, { "epoch": 2.9610563657863618, "grad_norm": 0.3644985556602478, "learning_rate": 0.0005, "loss": 1.1786222457885742, "step": 37561 }, { "epoch": 2.961135199054001, "grad_norm": 0.6222735047340393, "learning_rate": 0.0005, "loss": 1.4022841453552246, "step": 37562 }, { "epoch": 2.9612140323216396, "grad_norm": 0.6230124235153198, "learning_rate": 0.0005, "loss": 1.429890513420105, "step": 37563 }, { "epoch": 2.9612928655892787, "grad_norm": 0.4976539611816406, "learning_rate": 0.0005, "loss": 1.4221965074539185, "step": 37564 }, { "epoch": 2.961371698856918, "grad_norm": 0.4874333441257477, "learning_rate": 0.0005, "loss": 1.3401498794555664, "step": 37565 }, { "epoch": 2.9614505321245566, "grad_norm": 0.5750374794006348, "learning_rate": 0.0005, "loss": 1.4125635623931885, "step": 37566 }, { "epoch": 2.9615293653921952, "grad_norm": 0.5447248220443726, "learning_rate": 0.0005, "loss": 1.4831593036651611, "step": 37567 }, { "epoch": 2.9616081986598344, "grad_norm": 0.5527049899101257, "learning_rate": 0.0005, "loss": 1.3102951049804688, "step": 37568 }, { "epoch": 2.9616870319274735, "grad_norm": 0.437802255153656, "learning_rate": 0.0005, "loss": 1.1469999551773071, "step": 37569 }, { "epoch": 2.961765865195112, "grad_norm": 0.46949848532676697, "learning_rate": 0.0005, "loss": 1.203563928604126, "step": 37570 }, { "epoch": 2.9618446984627513, "grad_norm": 0.38996633887290955, "learning_rate": 0.0005, "loss": 1.3374273777008057, "step": 37571 }, { "epoch": 2.96192353173039, "grad_norm": 0.5029418468475342, "learning_rate": 0.0005, "loss": 1.4777448177337646, "step": 37572 }, { "epoch": 2.962002364998029, "grad_norm": 0.6399614214897156, "learning_rate": 0.0005, "loss": 1.3551849126815796, "step": 37573 }, { "epoch": 2.9620811982656683, "grad_norm": 0.5706406235694885, "learning_rate": 0.0005, "loss": 1.2577258348464966, "step": 37574 }, { "epoch": 2.962160031533307, "grad_norm": 0.6074053049087524, "learning_rate": 0.0005, "loss": 1.4093819856643677, "step": 37575 }, { "epoch": 2.962238864800946, "grad_norm": 0.45704102516174316, "learning_rate": 0.0005, "loss": 1.5390373468399048, "step": 37576 }, { "epoch": 2.962317698068585, "grad_norm": 0.6045685410499573, "learning_rate": 0.0005, "loss": 1.29458749294281, "step": 37577 }, { "epoch": 2.962396531336224, "grad_norm": 0.4687945544719696, "learning_rate": 0.0005, "loss": 1.3829599618911743, "step": 37578 }, { "epoch": 2.962475364603863, "grad_norm": 0.5945901870727539, "learning_rate": 0.0005, "loss": 1.5337203741073608, "step": 37579 }, { "epoch": 2.9625541978715018, "grad_norm": 0.4637298285961151, "learning_rate": 0.0005, "loss": 1.127851128578186, "step": 37580 }, { "epoch": 2.9626330311391404, "grad_norm": 0.5576623678207397, "learning_rate": 0.0005, "loss": 1.3857903480529785, "step": 37581 }, { "epoch": 2.9627118644067796, "grad_norm": 0.709873616695404, "learning_rate": 0.0005, "loss": 1.531751275062561, "step": 37582 }, { "epoch": 2.9627906976744187, "grad_norm": 0.6489554643630981, "learning_rate": 0.0005, "loss": 1.412790298461914, "step": 37583 }, { "epoch": 2.9628695309420574, "grad_norm": 0.5069249272346497, "learning_rate": 0.0005, "loss": 1.1867823600769043, "step": 37584 }, { "epoch": 2.9629483642096965, "grad_norm": 0.5166312456130981, "learning_rate": 0.0005, "loss": 1.4157359600067139, "step": 37585 }, { "epoch": 2.9630271974773352, "grad_norm": 0.48570430278778076, "learning_rate": 0.0005, "loss": 1.1700966358184814, "step": 37586 }, { "epoch": 2.9631060307449744, "grad_norm": 0.44813165068626404, "learning_rate": 0.0005, "loss": 1.428598403930664, "step": 37587 }, { "epoch": 2.9631848640126135, "grad_norm": 0.47134682536125183, "learning_rate": 0.0005, "loss": 1.3125829696655273, "step": 37588 }, { "epoch": 2.963263697280252, "grad_norm": 0.397587388753891, "learning_rate": 0.0005, "loss": 1.1806888580322266, "step": 37589 }, { "epoch": 2.9633425305478913, "grad_norm": 0.5301365852355957, "learning_rate": 0.0005, "loss": 1.171606183052063, "step": 37590 }, { "epoch": 2.96342136381553, "grad_norm": 0.38958507776260376, "learning_rate": 0.0005, "loss": 1.24473237991333, "step": 37591 }, { "epoch": 2.963500197083169, "grad_norm": 0.5396920442581177, "learning_rate": 0.0005, "loss": 1.296028733253479, "step": 37592 }, { "epoch": 2.9635790303508083, "grad_norm": 0.6518628597259521, "learning_rate": 0.0005, "loss": 1.347343921661377, "step": 37593 }, { "epoch": 2.963657863618447, "grad_norm": 0.5147398710250854, "learning_rate": 0.0005, "loss": 1.347115397453308, "step": 37594 }, { "epoch": 2.9637366968860857, "grad_norm": 0.5181266069412231, "learning_rate": 0.0005, "loss": 1.3722867965698242, "step": 37595 }, { "epoch": 2.963815530153725, "grad_norm": 0.47971734404563904, "learning_rate": 0.0005, "loss": 1.232000708580017, "step": 37596 }, { "epoch": 2.963894363421364, "grad_norm": 0.5122696161270142, "learning_rate": 0.0005, "loss": 1.2923921346664429, "step": 37597 }, { "epoch": 2.9639731966890026, "grad_norm": 0.5327615737915039, "learning_rate": 0.0005, "loss": 1.3395512104034424, "step": 37598 }, { "epoch": 2.9640520299566417, "grad_norm": 0.41746824979782104, "learning_rate": 0.0005, "loss": 1.1287143230438232, "step": 37599 }, { "epoch": 2.9641308632242804, "grad_norm": 0.4355054497718811, "learning_rate": 0.0005, "loss": 1.307047963142395, "step": 37600 }, { "epoch": 2.9642096964919196, "grad_norm": 0.5103436708450317, "learning_rate": 0.0005, "loss": 1.4410593509674072, "step": 37601 }, { "epoch": 2.9642885297595587, "grad_norm": 0.47786882519721985, "learning_rate": 0.0005, "loss": 1.3113949298858643, "step": 37602 }, { "epoch": 2.9643673630271974, "grad_norm": 0.46220171451568604, "learning_rate": 0.0005, "loss": 1.1823315620422363, "step": 37603 }, { "epoch": 2.9644461962948365, "grad_norm": 0.5392252206802368, "learning_rate": 0.0005, "loss": 1.5560612678527832, "step": 37604 }, { "epoch": 2.964525029562475, "grad_norm": 0.5359453558921814, "learning_rate": 0.0005, "loss": 1.5595512390136719, "step": 37605 }, { "epoch": 2.9646038628301143, "grad_norm": 0.5962067246437073, "learning_rate": 0.0005, "loss": 1.5225648880004883, "step": 37606 }, { "epoch": 2.9646826960977535, "grad_norm": 0.38125455379486084, "learning_rate": 0.0005, "loss": 1.2286319732666016, "step": 37607 }, { "epoch": 2.964761529365392, "grad_norm": 0.42492812871932983, "learning_rate": 0.0005, "loss": 1.3609647750854492, "step": 37608 }, { "epoch": 2.9648403626330313, "grad_norm": 0.5101460218429565, "learning_rate": 0.0005, "loss": 1.6963469982147217, "step": 37609 }, { "epoch": 2.96491919590067, "grad_norm": 0.5219027996063232, "learning_rate": 0.0005, "loss": 1.3201665878295898, "step": 37610 }, { "epoch": 2.964998029168309, "grad_norm": 0.4686714708805084, "learning_rate": 0.0005, "loss": 1.3192167282104492, "step": 37611 }, { "epoch": 2.965076862435948, "grad_norm": 0.5726510286331177, "learning_rate": 0.0005, "loss": 1.2085905075073242, "step": 37612 }, { "epoch": 2.965155695703587, "grad_norm": 0.4534100890159607, "learning_rate": 0.0005, "loss": 1.2444676160812378, "step": 37613 }, { "epoch": 2.9652345289712256, "grad_norm": 0.4512803554534912, "learning_rate": 0.0005, "loss": 1.192558765411377, "step": 37614 }, { "epoch": 2.9653133622388648, "grad_norm": 0.3694343864917755, "learning_rate": 0.0005, "loss": 1.1207304000854492, "step": 37615 }, { "epoch": 2.965392195506504, "grad_norm": 0.4768759310245514, "learning_rate": 0.0005, "loss": 1.352038860321045, "step": 37616 }, { "epoch": 2.9654710287741426, "grad_norm": 0.5629090666770935, "learning_rate": 0.0005, "loss": 1.3212000131607056, "step": 37617 }, { "epoch": 2.9655498620417817, "grad_norm": 0.5868048071861267, "learning_rate": 0.0005, "loss": 1.2606940269470215, "step": 37618 }, { "epoch": 2.9656286953094204, "grad_norm": 0.49336668848991394, "learning_rate": 0.0005, "loss": 1.526147484779358, "step": 37619 }, { "epoch": 2.9657075285770595, "grad_norm": 0.5480028390884399, "learning_rate": 0.0005, "loss": 1.4163564443588257, "step": 37620 }, { "epoch": 2.9657863618446987, "grad_norm": 0.532843291759491, "learning_rate": 0.0005, "loss": 1.2572543621063232, "step": 37621 }, { "epoch": 2.9658651951123374, "grad_norm": 0.486571729183197, "learning_rate": 0.0005, "loss": 1.3189740180969238, "step": 37622 }, { "epoch": 2.9659440283799765, "grad_norm": 0.4411473274230957, "learning_rate": 0.0005, "loss": 1.1869382858276367, "step": 37623 }, { "epoch": 2.966022861647615, "grad_norm": 0.5725276470184326, "learning_rate": 0.0005, "loss": 1.233830451965332, "step": 37624 }, { "epoch": 2.9661016949152543, "grad_norm": 0.6490548253059387, "learning_rate": 0.0005, "loss": 1.2424685955047607, "step": 37625 }, { "epoch": 2.966180528182893, "grad_norm": 0.5611124634742737, "learning_rate": 0.0005, "loss": 1.3313969373703003, "step": 37626 }, { "epoch": 2.966259361450532, "grad_norm": 0.5105980634689331, "learning_rate": 0.0005, "loss": 1.322290062904358, "step": 37627 }, { "epoch": 2.966338194718171, "grad_norm": 0.6640895009040833, "learning_rate": 0.0005, "loss": 1.4664937257766724, "step": 37628 }, { "epoch": 2.96641702798581, "grad_norm": 0.4953853189945221, "learning_rate": 0.0005, "loss": 1.4700292348861694, "step": 37629 }, { "epoch": 2.966495861253449, "grad_norm": 0.5076181888580322, "learning_rate": 0.0005, "loss": 1.2774839401245117, "step": 37630 }, { "epoch": 2.966574694521088, "grad_norm": 0.5788204669952393, "learning_rate": 0.0005, "loss": 1.5354728698730469, "step": 37631 }, { "epoch": 2.966653527788727, "grad_norm": 0.5409718155860901, "learning_rate": 0.0005, "loss": 1.3224575519561768, "step": 37632 }, { "epoch": 2.9667323610563656, "grad_norm": 0.5342915058135986, "learning_rate": 0.0005, "loss": 1.4731495380401611, "step": 37633 }, { "epoch": 2.9668111943240048, "grad_norm": 0.4458799362182617, "learning_rate": 0.0005, "loss": 1.236743688583374, "step": 37634 }, { "epoch": 2.966890027591644, "grad_norm": 0.49228382110595703, "learning_rate": 0.0005, "loss": 1.4482721090316772, "step": 37635 }, { "epoch": 2.9669688608592826, "grad_norm": 0.4813539385795593, "learning_rate": 0.0005, "loss": 1.2726538181304932, "step": 37636 }, { "epoch": 2.9670476941269217, "grad_norm": 0.5645744800567627, "learning_rate": 0.0005, "loss": 1.4602409601211548, "step": 37637 }, { "epoch": 2.9671265273945604, "grad_norm": 0.4208195209503174, "learning_rate": 0.0005, "loss": 1.319846510887146, "step": 37638 }, { "epoch": 2.9672053606621995, "grad_norm": 0.4864834249019623, "learning_rate": 0.0005, "loss": 1.4445942640304565, "step": 37639 }, { "epoch": 2.9672841939298387, "grad_norm": 0.43126586079597473, "learning_rate": 0.0005, "loss": 1.2872569561004639, "step": 37640 }, { "epoch": 2.9673630271974774, "grad_norm": 0.6140730977058411, "learning_rate": 0.0005, "loss": 1.352862000465393, "step": 37641 }, { "epoch": 2.967441860465116, "grad_norm": 0.49082374572753906, "learning_rate": 0.0005, "loss": 1.3681305646896362, "step": 37642 }, { "epoch": 2.967520693732755, "grad_norm": 0.7375893592834473, "learning_rate": 0.0005, "loss": 1.3516335487365723, "step": 37643 }, { "epoch": 2.9675995270003943, "grad_norm": 0.4301815330982208, "learning_rate": 0.0005, "loss": 1.274770975112915, "step": 37644 }, { "epoch": 2.967678360268033, "grad_norm": 0.552017867565155, "learning_rate": 0.0005, "loss": 1.4089778661727905, "step": 37645 }, { "epoch": 2.967757193535672, "grad_norm": 0.6643384695053101, "learning_rate": 0.0005, "loss": 1.3842952251434326, "step": 37646 }, { "epoch": 2.967836026803311, "grad_norm": 0.7485542893409729, "learning_rate": 0.0005, "loss": 1.2099053859710693, "step": 37647 }, { "epoch": 2.96791486007095, "grad_norm": 0.4321528673171997, "learning_rate": 0.0005, "loss": 1.2338907718658447, "step": 37648 }, { "epoch": 2.967993693338589, "grad_norm": 0.5560771822929382, "learning_rate": 0.0005, "loss": 1.3486825227737427, "step": 37649 }, { "epoch": 2.968072526606228, "grad_norm": 0.5784648060798645, "learning_rate": 0.0005, "loss": 1.502783179283142, "step": 37650 }, { "epoch": 2.968151359873867, "grad_norm": 0.6335439682006836, "learning_rate": 0.0005, "loss": 1.4709117412567139, "step": 37651 }, { "epoch": 2.9682301931415056, "grad_norm": 0.46650370955467224, "learning_rate": 0.0005, "loss": 1.1504195928573608, "step": 37652 }, { "epoch": 2.9683090264091447, "grad_norm": 0.6370490789413452, "learning_rate": 0.0005, "loss": 1.3586995601654053, "step": 37653 }, { "epoch": 2.968387859676784, "grad_norm": 0.49101245403289795, "learning_rate": 0.0005, "loss": 1.356168270111084, "step": 37654 }, { "epoch": 2.9684666929444226, "grad_norm": 0.423312783241272, "learning_rate": 0.0005, "loss": 1.1648653745651245, "step": 37655 }, { "epoch": 2.9685455262120612, "grad_norm": 0.4945737421512604, "learning_rate": 0.0005, "loss": 1.2592580318450928, "step": 37656 }, { "epoch": 2.9686243594797004, "grad_norm": 0.5064235329627991, "learning_rate": 0.0005, "loss": 1.4037449359893799, "step": 37657 }, { "epoch": 2.9687031927473395, "grad_norm": 0.6773306727409363, "learning_rate": 0.0005, "loss": 1.51741623878479, "step": 37658 }, { "epoch": 2.968782026014978, "grad_norm": 0.4967958629131317, "learning_rate": 0.0005, "loss": 1.4705169200897217, "step": 37659 }, { "epoch": 2.9688608592826173, "grad_norm": 0.5212958455085754, "learning_rate": 0.0005, "loss": 1.5879974365234375, "step": 37660 }, { "epoch": 2.968939692550256, "grad_norm": 0.5500802993774414, "learning_rate": 0.0005, "loss": 1.2839761972427368, "step": 37661 }, { "epoch": 2.969018525817895, "grad_norm": 0.6374375224113464, "learning_rate": 0.0005, "loss": 1.338862657546997, "step": 37662 }, { "epoch": 2.9690973590855343, "grad_norm": 0.5873436331748962, "learning_rate": 0.0005, "loss": 1.4708110094070435, "step": 37663 }, { "epoch": 2.969176192353173, "grad_norm": 0.5066483020782471, "learning_rate": 0.0005, "loss": 1.2635836601257324, "step": 37664 }, { "epoch": 2.969255025620812, "grad_norm": 0.48698675632476807, "learning_rate": 0.0005, "loss": 1.3183231353759766, "step": 37665 }, { "epoch": 2.969333858888451, "grad_norm": 0.465366393327713, "learning_rate": 0.0005, "loss": 1.1382663249969482, "step": 37666 }, { "epoch": 2.96941269215609, "grad_norm": 0.7012698650360107, "learning_rate": 0.0005, "loss": 1.5406807661056519, "step": 37667 }, { "epoch": 2.969491525423729, "grad_norm": 0.6281461119651794, "learning_rate": 0.0005, "loss": 1.4167935848236084, "step": 37668 }, { "epoch": 2.9695703586913678, "grad_norm": 0.4151402413845062, "learning_rate": 0.0005, "loss": 1.273425579071045, "step": 37669 }, { "epoch": 2.9696491919590065, "grad_norm": 0.48587673902511597, "learning_rate": 0.0005, "loss": 1.282548189163208, "step": 37670 }, { "epoch": 2.9697280252266456, "grad_norm": 0.727833092212677, "learning_rate": 0.0005, "loss": 1.3710191249847412, "step": 37671 }, { "epoch": 2.9698068584942847, "grad_norm": 0.4626852869987488, "learning_rate": 0.0005, "loss": 1.1651082038879395, "step": 37672 }, { "epoch": 2.9698856917619234, "grad_norm": 0.4839429557323456, "learning_rate": 0.0005, "loss": 1.1808850765228271, "step": 37673 }, { "epoch": 2.9699645250295625, "grad_norm": 0.42217159271240234, "learning_rate": 0.0005, "loss": 1.192983627319336, "step": 37674 }, { "epoch": 2.9700433582972012, "grad_norm": 0.3950970470905304, "learning_rate": 0.0005, "loss": 1.200371503829956, "step": 37675 }, { "epoch": 2.9701221915648404, "grad_norm": 0.5268564820289612, "learning_rate": 0.0005, "loss": 1.3765549659729004, "step": 37676 }, { "epoch": 2.9702010248324795, "grad_norm": 0.4768327474594116, "learning_rate": 0.0005, "loss": 1.4703748226165771, "step": 37677 }, { "epoch": 2.970279858100118, "grad_norm": 0.5203197002410889, "learning_rate": 0.0005, "loss": 1.624157428741455, "step": 37678 }, { "epoch": 2.9703586913677573, "grad_norm": 0.6189981698989868, "learning_rate": 0.0005, "loss": 1.3674707412719727, "step": 37679 }, { "epoch": 2.970437524635396, "grad_norm": 0.42986056208610535, "learning_rate": 0.0005, "loss": 1.1473444700241089, "step": 37680 }, { "epoch": 2.970516357903035, "grad_norm": 0.43313390016555786, "learning_rate": 0.0005, "loss": 1.1989306211471558, "step": 37681 }, { "epoch": 2.9705951911706743, "grad_norm": 0.5247209668159485, "learning_rate": 0.0005, "loss": 1.589369297027588, "step": 37682 }, { "epoch": 2.970674024438313, "grad_norm": 0.4927487075328827, "learning_rate": 0.0005, "loss": 1.2294456958770752, "step": 37683 }, { "epoch": 2.9707528577059517, "grad_norm": 0.46248605847358704, "learning_rate": 0.0005, "loss": 1.2830369472503662, "step": 37684 }, { "epoch": 2.970831690973591, "grad_norm": 0.5137747526168823, "learning_rate": 0.0005, "loss": 1.1423146724700928, "step": 37685 }, { "epoch": 2.97091052424123, "grad_norm": 0.5312676429748535, "learning_rate": 0.0005, "loss": 1.2711602449417114, "step": 37686 }, { "epoch": 2.9709893575088686, "grad_norm": 0.6646453738212585, "learning_rate": 0.0005, "loss": 1.4588264226913452, "step": 37687 }, { "epoch": 2.9710681907765077, "grad_norm": 0.5539028644561768, "learning_rate": 0.0005, "loss": 1.490077018737793, "step": 37688 }, { "epoch": 2.9711470240441464, "grad_norm": 0.44249746203422546, "learning_rate": 0.0005, "loss": 1.1616730690002441, "step": 37689 }, { "epoch": 2.9712258573117856, "grad_norm": 0.42926254868507385, "learning_rate": 0.0005, "loss": 1.3324353694915771, "step": 37690 }, { "epoch": 2.9713046905794247, "grad_norm": 0.521039605140686, "learning_rate": 0.0005, "loss": 1.2739641666412354, "step": 37691 }, { "epoch": 2.9713835238470634, "grad_norm": 0.689311683177948, "learning_rate": 0.0005, "loss": 1.3165299892425537, "step": 37692 }, { "epoch": 2.9714623571147025, "grad_norm": 0.6180358529090881, "learning_rate": 0.0005, "loss": 1.2299283742904663, "step": 37693 }, { "epoch": 2.971541190382341, "grad_norm": 0.6378610134124756, "learning_rate": 0.0005, "loss": 1.4354315996170044, "step": 37694 }, { "epoch": 2.9716200236499803, "grad_norm": 0.4264107644557953, "learning_rate": 0.0005, "loss": 1.2672475576400757, "step": 37695 }, { "epoch": 2.9716988569176195, "grad_norm": 0.3920551836490631, "learning_rate": 0.0005, "loss": 1.3055222034454346, "step": 37696 }, { "epoch": 2.971777690185258, "grad_norm": 0.7500895857810974, "learning_rate": 0.0005, "loss": 1.4508967399597168, "step": 37697 }, { "epoch": 2.971856523452897, "grad_norm": 0.4630213975906372, "learning_rate": 0.0005, "loss": 1.3647358417510986, "step": 37698 }, { "epoch": 2.971935356720536, "grad_norm": 0.40568751096725464, "learning_rate": 0.0005, "loss": 1.1499450206756592, "step": 37699 }, { "epoch": 2.972014189988175, "grad_norm": 0.4856737554073334, "learning_rate": 0.0005, "loss": 1.3969240188598633, "step": 37700 }, { "epoch": 2.972093023255814, "grad_norm": 0.4990507960319519, "learning_rate": 0.0005, "loss": 1.2482577562332153, "step": 37701 }, { "epoch": 2.972171856523453, "grad_norm": 0.5132024884223938, "learning_rate": 0.0005, "loss": 1.4023699760437012, "step": 37702 }, { "epoch": 2.9722506897910916, "grad_norm": 0.4802340567111969, "learning_rate": 0.0005, "loss": 1.3958501815795898, "step": 37703 }, { "epoch": 2.9723295230587308, "grad_norm": 0.4288797676563263, "learning_rate": 0.0005, "loss": 1.250030517578125, "step": 37704 }, { "epoch": 2.97240835632637, "grad_norm": 0.48735135793685913, "learning_rate": 0.0005, "loss": 1.3767061233520508, "step": 37705 }, { "epoch": 2.9724871895940086, "grad_norm": 0.4589133858680725, "learning_rate": 0.0005, "loss": 1.298476219177246, "step": 37706 }, { "epoch": 2.9725660228616477, "grad_norm": 0.43228989839553833, "learning_rate": 0.0005, "loss": 1.1494946479797363, "step": 37707 }, { "epoch": 2.9726448561292864, "grad_norm": 0.5784137845039368, "learning_rate": 0.0005, "loss": 1.4100441932678223, "step": 37708 }, { "epoch": 2.9727236893969256, "grad_norm": 0.508453905582428, "learning_rate": 0.0005, "loss": 1.3043735027313232, "step": 37709 }, { "epoch": 2.9728025226645647, "grad_norm": 0.5974651575088501, "learning_rate": 0.0005, "loss": 1.4419671297073364, "step": 37710 }, { "epoch": 2.9728813559322034, "grad_norm": 0.40471601486206055, "learning_rate": 0.0005, "loss": 1.2585126161575317, "step": 37711 }, { "epoch": 2.972960189199842, "grad_norm": 0.48003309965133667, "learning_rate": 0.0005, "loss": 1.303358554840088, "step": 37712 }, { "epoch": 2.973039022467481, "grad_norm": 0.5542921423912048, "learning_rate": 0.0005, "loss": 1.3795967102050781, "step": 37713 }, { "epoch": 2.9731178557351203, "grad_norm": 0.443204402923584, "learning_rate": 0.0005, "loss": 1.1614482402801514, "step": 37714 }, { "epoch": 2.973196689002759, "grad_norm": 0.5537109375, "learning_rate": 0.0005, "loss": 1.2729713916778564, "step": 37715 }, { "epoch": 2.973275522270398, "grad_norm": 0.6312144994735718, "learning_rate": 0.0005, "loss": 1.5209254026412964, "step": 37716 }, { "epoch": 2.973354355538037, "grad_norm": 0.5694180727005005, "learning_rate": 0.0005, "loss": 1.4205355644226074, "step": 37717 }, { "epoch": 2.973433188805676, "grad_norm": 0.5111395716667175, "learning_rate": 0.0005, "loss": 1.2796684503555298, "step": 37718 }, { "epoch": 2.973512022073315, "grad_norm": 0.4139835834503174, "learning_rate": 0.0005, "loss": 1.2075486183166504, "step": 37719 }, { "epoch": 2.973590855340954, "grad_norm": 0.4093917906284332, "learning_rate": 0.0005, "loss": 1.345212459564209, "step": 37720 }, { "epoch": 2.973669688608593, "grad_norm": 0.4748329818248749, "learning_rate": 0.0005, "loss": 1.1664623022079468, "step": 37721 }, { "epoch": 2.9737485218762316, "grad_norm": 0.4922768473625183, "learning_rate": 0.0005, "loss": 1.3664216995239258, "step": 37722 }, { "epoch": 2.9738273551438708, "grad_norm": 0.5853511095046997, "learning_rate": 0.0005, "loss": 1.5233385562896729, "step": 37723 }, { "epoch": 2.97390618841151, "grad_norm": 0.5458770990371704, "learning_rate": 0.0005, "loss": 1.3552143573760986, "step": 37724 }, { "epoch": 2.9739850216791486, "grad_norm": 0.6246868371963501, "learning_rate": 0.0005, "loss": 1.430443525314331, "step": 37725 }, { "epoch": 2.9740638549467877, "grad_norm": 0.5196026563644409, "learning_rate": 0.0005, "loss": 1.4095999002456665, "step": 37726 }, { "epoch": 2.9741426882144264, "grad_norm": 0.6657677292823792, "learning_rate": 0.0005, "loss": 1.3787535429000854, "step": 37727 }, { "epoch": 2.9742215214820655, "grad_norm": 0.40391016006469727, "learning_rate": 0.0005, "loss": 1.2879037857055664, "step": 37728 }, { "epoch": 2.9743003547497042, "grad_norm": 0.4714762568473816, "learning_rate": 0.0005, "loss": 1.2760393619537354, "step": 37729 }, { "epoch": 2.9743791880173434, "grad_norm": 0.6429158449172974, "learning_rate": 0.0005, "loss": 1.334106206893921, "step": 37730 }, { "epoch": 2.974458021284982, "grad_norm": 0.5022372603416443, "learning_rate": 0.0005, "loss": 1.3148725032806396, "step": 37731 }, { "epoch": 2.974536854552621, "grad_norm": 0.45064499974250793, "learning_rate": 0.0005, "loss": 1.2726993560791016, "step": 37732 }, { "epoch": 2.9746156878202603, "grad_norm": 0.39662837982177734, "learning_rate": 0.0005, "loss": 1.23745858669281, "step": 37733 }, { "epoch": 2.974694521087899, "grad_norm": 0.5654743313789368, "learning_rate": 0.0005, "loss": 1.2791019678115845, "step": 37734 }, { "epoch": 2.974773354355538, "grad_norm": 0.4623188376426697, "learning_rate": 0.0005, "loss": 1.318028211593628, "step": 37735 }, { "epoch": 2.974852187623177, "grad_norm": 0.5595988631248474, "learning_rate": 0.0005, "loss": 1.3450034856796265, "step": 37736 }, { "epoch": 2.974931020890816, "grad_norm": 0.5632365942001343, "learning_rate": 0.0005, "loss": 1.4385013580322266, "step": 37737 }, { "epoch": 2.975009854158455, "grad_norm": 0.4549705386161804, "learning_rate": 0.0005, "loss": 1.1846266984939575, "step": 37738 }, { "epoch": 2.975088687426094, "grad_norm": 0.4782361090183258, "learning_rate": 0.0005, "loss": 1.4851758480072021, "step": 37739 }, { "epoch": 2.975167520693733, "grad_norm": 0.44892656803131104, "learning_rate": 0.0005, "loss": 1.298522710800171, "step": 37740 }, { "epoch": 2.9752463539613716, "grad_norm": 0.5035371780395508, "learning_rate": 0.0005, "loss": 1.1525111198425293, "step": 37741 }, { "epoch": 2.9753251872290107, "grad_norm": 0.46497777104377747, "learning_rate": 0.0005, "loss": 1.1623998880386353, "step": 37742 }, { "epoch": 2.9754040204966494, "grad_norm": 0.43875056505203247, "learning_rate": 0.0005, "loss": 1.1041045188903809, "step": 37743 }, { "epoch": 2.9754828537642886, "grad_norm": 0.45531973242759705, "learning_rate": 0.0005, "loss": 1.3797662258148193, "step": 37744 }, { "epoch": 2.9755616870319272, "grad_norm": 0.601791262626648, "learning_rate": 0.0005, "loss": 1.4340143203735352, "step": 37745 }, { "epoch": 2.9756405202995664, "grad_norm": 0.5449709892272949, "learning_rate": 0.0005, "loss": 1.0858262777328491, "step": 37746 }, { "epoch": 2.9757193535672055, "grad_norm": 0.46468585729599, "learning_rate": 0.0005, "loss": 1.322121024131775, "step": 37747 }, { "epoch": 2.975798186834844, "grad_norm": 0.49998611211776733, "learning_rate": 0.0005, "loss": 1.249186635017395, "step": 37748 }, { "epoch": 2.9758770201024833, "grad_norm": 0.520524799823761, "learning_rate": 0.0005, "loss": 1.225386381149292, "step": 37749 }, { "epoch": 2.975955853370122, "grad_norm": 0.5267033576965332, "learning_rate": 0.0005, "loss": 1.298279047012329, "step": 37750 }, { "epoch": 2.976034686637761, "grad_norm": 0.5120807886123657, "learning_rate": 0.0005, "loss": 1.4009840488433838, "step": 37751 }, { "epoch": 2.9761135199054003, "grad_norm": 0.7115445137023926, "learning_rate": 0.0005, "loss": 1.2980533838272095, "step": 37752 }, { "epoch": 2.976192353173039, "grad_norm": 0.4361858665943146, "learning_rate": 0.0005, "loss": 1.2634435892105103, "step": 37753 }, { "epoch": 2.976271186440678, "grad_norm": 0.44269195199012756, "learning_rate": 0.0005, "loss": 1.2811734676361084, "step": 37754 }, { "epoch": 2.976350019708317, "grad_norm": 0.5271441340446472, "learning_rate": 0.0005, "loss": 1.441220760345459, "step": 37755 }, { "epoch": 2.976428852975956, "grad_norm": 0.7670456171035767, "learning_rate": 0.0005, "loss": 1.436437726020813, "step": 37756 }, { "epoch": 2.976507686243595, "grad_norm": 0.639912486076355, "learning_rate": 0.0005, "loss": 1.5697176456451416, "step": 37757 }, { "epoch": 2.9765865195112338, "grad_norm": 0.539840579032898, "learning_rate": 0.0005, "loss": 1.3681821823120117, "step": 37758 }, { "epoch": 2.9766653527788725, "grad_norm": 0.5104287266731262, "learning_rate": 0.0005, "loss": 1.3132323026657104, "step": 37759 }, { "epoch": 2.9767441860465116, "grad_norm": 0.6139834523200989, "learning_rate": 0.0005, "loss": 1.4292185306549072, "step": 37760 }, { "epoch": 2.9768230193141507, "grad_norm": 0.5603967905044556, "learning_rate": 0.0005, "loss": 1.5798323154449463, "step": 37761 }, { "epoch": 2.9769018525817894, "grad_norm": 0.497747540473938, "learning_rate": 0.0005, "loss": 1.352079153060913, "step": 37762 }, { "epoch": 2.9769806858494285, "grad_norm": 0.49707937240600586, "learning_rate": 0.0005, "loss": 1.304160237312317, "step": 37763 }, { "epoch": 2.9770595191170672, "grad_norm": 0.4158312678337097, "learning_rate": 0.0005, "loss": 1.4718672037124634, "step": 37764 }, { "epoch": 2.9771383523847064, "grad_norm": 0.5365796089172363, "learning_rate": 0.0005, "loss": 1.603234052658081, "step": 37765 }, { "epoch": 2.9772171856523455, "grad_norm": 0.8360150456428528, "learning_rate": 0.0005, "loss": 1.467516303062439, "step": 37766 }, { "epoch": 2.977296018919984, "grad_norm": 0.4464235305786133, "learning_rate": 0.0005, "loss": 1.1798503398895264, "step": 37767 }, { "epoch": 2.9773748521876233, "grad_norm": 0.5780771374702454, "learning_rate": 0.0005, "loss": 1.3643443584442139, "step": 37768 }, { "epoch": 2.977453685455262, "grad_norm": 0.861347496509552, "learning_rate": 0.0005, "loss": 1.4312340021133423, "step": 37769 }, { "epoch": 2.977532518722901, "grad_norm": 0.6311069130897522, "learning_rate": 0.0005, "loss": 1.4645090103149414, "step": 37770 }, { "epoch": 2.9776113519905403, "grad_norm": 0.5265384316444397, "learning_rate": 0.0005, "loss": 1.2105802297592163, "step": 37771 }, { "epoch": 2.977690185258179, "grad_norm": 0.4939921796321869, "learning_rate": 0.0005, "loss": 1.2866075038909912, "step": 37772 }, { "epoch": 2.9777690185258177, "grad_norm": 0.492203950881958, "learning_rate": 0.0005, "loss": 1.3395581245422363, "step": 37773 }, { "epoch": 2.977847851793457, "grad_norm": 0.7119588255882263, "learning_rate": 0.0005, "loss": 1.1011550426483154, "step": 37774 }, { "epoch": 2.977926685061096, "grad_norm": 0.45051631331443787, "learning_rate": 0.0005, "loss": 1.316806674003601, "step": 37775 }, { "epoch": 2.9780055183287346, "grad_norm": 0.4346090257167816, "learning_rate": 0.0005, "loss": 1.1225049495697021, "step": 37776 }, { "epoch": 2.9780843515963737, "grad_norm": 0.536075234413147, "learning_rate": 0.0005, "loss": 1.3162540197372437, "step": 37777 }, { "epoch": 2.9781631848640124, "grad_norm": 0.48036083579063416, "learning_rate": 0.0005, "loss": 1.0939654111862183, "step": 37778 }, { "epoch": 2.9782420181316516, "grad_norm": 0.4729277789592743, "learning_rate": 0.0005, "loss": 1.2793829441070557, "step": 37779 }, { "epoch": 2.9783208513992907, "grad_norm": 0.4670003354549408, "learning_rate": 0.0005, "loss": 1.2513163089752197, "step": 37780 }, { "epoch": 2.9783996846669294, "grad_norm": 0.7233255505561829, "learning_rate": 0.0005, "loss": 1.5284843444824219, "step": 37781 }, { "epoch": 2.9784785179345685, "grad_norm": 0.48059117794036865, "learning_rate": 0.0005, "loss": 1.2870421409606934, "step": 37782 }, { "epoch": 2.978557351202207, "grad_norm": 0.39155855774879456, "learning_rate": 0.0005, "loss": 1.3411641120910645, "step": 37783 }, { "epoch": 2.9786361844698463, "grad_norm": 0.41194215416908264, "learning_rate": 0.0005, "loss": 1.2339067459106445, "step": 37784 }, { "epoch": 2.9787150177374855, "grad_norm": 0.5092242360115051, "learning_rate": 0.0005, "loss": 1.3813198804855347, "step": 37785 }, { "epoch": 2.978793851005124, "grad_norm": 0.49239200353622437, "learning_rate": 0.0005, "loss": 1.4184978008270264, "step": 37786 }, { "epoch": 2.978872684272763, "grad_norm": 0.4684874415397644, "learning_rate": 0.0005, "loss": 1.3244915008544922, "step": 37787 }, { "epoch": 2.978951517540402, "grad_norm": 0.5262724161148071, "learning_rate": 0.0005, "loss": 1.3222204446792603, "step": 37788 }, { "epoch": 2.979030350808041, "grad_norm": 0.5085456967353821, "learning_rate": 0.0005, "loss": 1.2930833101272583, "step": 37789 }, { "epoch": 2.97910918407568, "grad_norm": 0.6261137127876282, "learning_rate": 0.0005, "loss": 1.489539623260498, "step": 37790 }, { "epoch": 2.979188017343319, "grad_norm": 0.49143293499946594, "learning_rate": 0.0005, "loss": 1.2877647876739502, "step": 37791 }, { "epoch": 2.9792668506109576, "grad_norm": 0.43204233050346375, "learning_rate": 0.0005, "loss": 1.1198135614395142, "step": 37792 }, { "epoch": 2.9793456838785968, "grad_norm": 0.6145879626274109, "learning_rate": 0.0005, "loss": 1.5554496049880981, "step": 37793 }, { "epoch": 2.979424517146236, "grad_norm": 0.4596726894378662, "learning_rate": 0.0005, "loss": 1.343940019607544, "step": 37794 }, { "epoch": 2.9795033504138746, "grad_norm": 0.5010921359062195, "learning_rate": 0.0005, "loss": 1.4994096755981445, "step": 37795 }, { "epoch": 2.9795821836815137, "grad_norm": 0.5185250043869019, "learning_rate": 0.0005, "loss": 1.6594383716583252, "step": 37796 }, { "epoch": 2.9796610169491524, "grad_norm": 0.5764041543006897, "learning_rate": 0.0005, "loss": 1.3833742141723633, "step": 37797 }, { "epoch": 2.9797398502167916, "grad_norm": 0.576433002948761, "learning_rate": 0.0005, "loss": 1.3291940689086914, "step": 37798 }, { "epoch": 2.9798186834844307, "grad_norm": 0.44508877396583557, "learning_rate": 0.0005, "loss": 1.436962366104126, "step": 37799 }, { "epoch": 2.9798975167520694, "grad_norm": 0.4713291823863983, "learning_rate": 0.0005, "loss": 1.2428630590438843, "step": 37800 }, { "epoch": 2.979976350019708, "grad_norm": 0.5732701420783997, "learning_rate": 0.0005, "loss": 1.468048334121704, "step": 37801 }, { "epoch": 2.980055183287347, "grad_norm": 0.5096508860588074, "learning_rate": 0.0005, "loss": 1.247584581375122, "step": 37802 }, { "epoch": 2.9801340165549863, "grad_norm": 0.7499557733535767, "learning_rate": 0.0005, "loss": 1.206246256828308, "step": 37803 }, { "epoch": 2.980212849822625, "grad_norm": 0.472260445356369, "learning_rate": 0.0005, "loss": 1.2672672271728516, "step": 37804 }, { "epoch": 2.980291683090264, "grad_norm": 0.451128214597702, "learning_rate": 0.0005, "loss": 1.3295270204544067, "step": 37805 }, { "epoch": 2.980370516357903, "grad_norm": 0.4439225196838379, "learning_rate": 0.0005, "loss": 1.2806506156921387, "step": 37806 }, { "epoch": 2.980449349625542, "grad_norm": 0.5184781551361084, "learning_rate": 0.0005, "loss": 1.4434598684310913, "step": 37807 }, { "epoch": 2.980528182893181, "grad_norm": 0.5962735414505005, "learning_rate": 0.0005, "loss": 1.284096598625183, "step": 37808 }, { "epoch": 2.98060701616082, "grad_norm": 0.5060304403305054, "learning_rate": 0.0005, "loss": 1.2790942192077637, "step": 37809 }, { "epoch": 2.980685849428459, "grad_norm": 0.4895177483558655, "learning_rate": 0.0005, "loss": 1.429572582244873, "step": 37810 }, { "epoch": 2.9807646826960976, "grad_norm": 0.6285610795021057, "learning_rate": 0.0005, "loss": 1.1723170280456543, "step": 37811 }, { "epoch": 2.9808435159637368, "grad_norm": 0.4438744783401489, "learning_rate": 0.0005, "loss": 1.2018357515335083, "step": 37812 }, { "epoch": 2.980922349231376, "grad_norm": 0.47684773802757263, "learning_rate": 0.0005, "loss": 1.2497124671936035, "step": 37813 }, { "epoch": 2.9810011824990146, "grad_norm": 0.5686994194984436, "learning_rate": 0.0005, "loss": 1.3229944705963135, "step": 37814 }, { "epoch": 2.9810800157666533, "grad_norm": 0.4814355671405792, "learning_rate": 0.0005, "loss": 1.219372272491455, "step": 37815 }, { "epoch": 2.9811588490342924, "grad_norm": 0.44461655616760254, "learning_rate": 0.0005, "loss": 1.268787145614624, "step": 37816 }, { "epoch": 2.9812376823019315, "grad_norm": 0.506849467754364, "learning_rate": 0.0005, "loss": 1.1699509620666504, "step": 37817 }, { "epoch": 2.9813165155695702, "grad_norm": 0.4426959455013275, "learning_rate": 0.0005, "loss": 1.2749319076538086, "step": 37818 }, { "epoch": 2.9813953488372094, "grad_norm": 0.5499027967453003, "learning_rate": 0.0005, "loss": 1.431815266609192, "step": 37819 }, { "epoch": 2.981474182104848, "grad_norm": 0.6087557673454285, "learning_rate": 0.0005, "loss": 1.1067920923233032, "step": 37820 }, { "epoch": 2.981553015372487, "grad_norm": 0.5524484515190125, "learning_rate": 0.0005, "loss": 1.3385536670684814, "step": 37821 }, { "epoch": 2.9816318486401263, "grad_norm": 0.847946047782898, "learning_rate": 0.0005, "loss": 1.4499657154083252, "step": 37822 }, { "epoch": 2.981710681907765, "grad_norm": 0.5137755274772644, "learning_rate": 0.0005, "loss": 1.2879765033721924, "step": 37823 }, { "epoch": 2.981789515175404, "grad_norm": 0.48983362317085266, "learning_rate": 0.0005, "loss": 1.459937334060669, "step": 37824 }, { "epoch": 2.981868348443043, "grad_norm": 0.5392048954963684, "learning_rate": 0.0005, "loss": 1.5729129314422607, "step": 37825 }, { "epoch": 2.981947181710682, "grad_norm": 0.6509913802146912, "learning_rate": 0.0005, "loss": 1.197527527809143, "step": 37826 }, { "epoch": 2.982026014978321, "grad_norm": 0.602465808391571, "learning_rate": 0.0005, "loss": 1.2763714790344238, "step": 37827 }, { "epoch": 2.98210484824596, "grad_norm": 0.4586010277271271, "learning_rate": 0.0005, "loss": 1.3135727643966675, "step": 37828 }, { "epoch": 2.9821836815135985, "grad_norm": 0.47785258293151855, "learning_rate": 0.0005, "loss": 1.3175766468048096, "step": 37829 }, { "epoch": 2.9822625147812376, "grad_norm": 0.4858480393886566, "learning_rate": 0.0005, "loss": 1.3597657680511475, "step": 37830 }, { "epoch": 2.9823413480488767, "grad_norm": 0.4707207977771759, "learning_rate": 0.0005, "loss": 1.2760976552963257, "step": 37831 }, { "epoch": 2.9824201813165154, "grad_norm": 0.6535650491714478, "learning_rate": 0.0005, "loss": 1.3835136890411377, "step": 37832 }, { "epoch": 2.9824990145841546, "grad_norm": 0.5729506611824036, "learning_rate": 0.0005, "loss": 1.3264691829681396, "step": 37833 }, { "epoch": 2.9825778478517933, "grad_norm": 0.4818918704986572, "learning_rate": 0.0005, "loss": 1.4493794441223145, "step": 37834 }, { "epoch": 2.9826566811194324, "grad_norm": 0.591497004032135, "learning_rate": 0.0005, "loss": 1.4486461877822876, "step": 37835 }, { "epoch": 2.9827355143870715, "grad_norm": 0.46327027678489685, "learning_rate": 0.0005, "loss": 1.2014641761779785, "step": 37836 }, { "epoch": 2.98281434765471, "grad_norm": 0.4741029739379883, "learning_rate": 0.0005, "loss": 1.4471266269683838, "step": 37837 }, { "epoch": 2.9828931809223493, "grad_norm": 0.49369215965270996, "learning_rate": 0.0005, "loss": 1.4781110286712646, "step": 37838 }, { "epoch": 2.982972014189988, "grad_norm": 0.5244032144546509, "learning_rate": 0.0005, "loss": 1.3729028701782227, "step": 37839 }, { "epoch": 2.983050847457627, "grad_norm": 0.4566875100135803, "learning_rate": 0.0005, "loss": 1.2430956363677979, "step": 37840 }, { "epoch": 2.9831296807252663, "grad_norm": 0.5937605500221252, "learning_rate": 0.0005, "loss": 1.392404556274414, "step": 37841 }, { "epoch": 2.983208513992905, "grad_norm": 0.4493561387062073, "learning_rate": 0.0005, "loss": 1.3847005367279053, "step": 37842 }, { "epoch": 2.9832873472605437, "grad_norm": 0.5092639327049255, "learning_rate": 0.0005, "loss": 1.2647082805633545, "step": 37843 }, { "epoch": 2.983366180528183, "grad_norm": 0.4455958306789398, "learning_rate": 0.0005, "loss": 1.6071573495864868, "step": 37844 }, { "epoch": 2.983445013795822, "grad_norm": 0.4354688823223114, "learning_rate": 0.0005, "loss": 1.2279542684555054, "step": 37845 }, { "epoch": 2.9835238470634606, "grad_norm": 0.5384683609008789, "learning_rate": 0.0005, "loss": 1.3633575439453125, "step": 37846 }, { "epoch": 2.9836026803310998, "grad_norm": 0.5240153074264526, "learning_rate": 0.0005, "loss": 1.3743709325790405, "step": 37847 }, { "epoch": 2.9836815135987385, "grad_norm": 0.5735905170440674, "learning_rate": 0.0005, "loss": 1.2736907005310059, "step": 37848 }, { "epoch": 2.9837603468663776, "grad_norm": 0.6192960143089294, "learning_rate": 0.0005, "loss": 1.321622610092163, "step": 37849 }, { "epoch": 2.9838391801340167, "grad_norm": 0.4449801445007324, "learning_rate": 0.0005, "loss": 1.5099600553512573, "step": 37850 }, { "epoch": 2.9839180134016554, "grad_norm": 0.41600680351257324, "learning_rate": 0.0005, "loss": 1.150144100189209, "step": 37851 }, { "epoch": 2.9839968466692945, "grad_norm": 0.4882711172103882, "learning_rate": 0.0005, "loss": 1.2685457468032837, "step": 37852 }, { "epoch": 2.9840756799369332, "grad_norm": 0.7693792581558228, "learning_rate": 0.0005, "loss": 1.3534693717956543, "step": 37853 }, { "epoch": 2.9841545132045724, "grad_norm": 0.45844313502311707, "learning_rate": 0.0005, "loss": 1.1246684789657593, "step": 37854 }, { "epoch": 2.9842333464722115, "grad_norm": 0.43946629762649536, "learning_rate": 0.0005, "loss": 1.1987755298614502, "step": 37855 }, { "epoch": 2.98431217973985, "grad_norm": 0.51410973072052, "learning_rate": 0.0005, "loss": 1.4623219966888428, "step": 37856 }, { "epoch": 2.9843910130074893, "grad_norm": 0.539886474609375, "learning_rate": 0.0005, "loss": 1.303922414779663, "step": 37857 }, { "epoch": 2.984469846275128, "grad_norm": 0.491702675819397, "learning_rate": 0.0005, "loss": 1.3956966400146484, "step": 37858 }, { "epoch": 2.984548679542767, "grad_norm": 0.43993479013442993, "learning_rate": 0.0005, "loss": 1.2696058750152588, "step": 37859 }, { "epoch": 2.984627512810406, "grad_norm": 0.48417192697525024, "learning_rate": 0.0005, "loss": 1.2131577730178833, "step": 37860 }, { "epoch": 2.984706346078045, "grad_norm": 0.511181652545929, "learning_rate": 0.0005, "loss": 1.4075777530670166, "step": 37861 }, { "epoch": 2.9847851793456837, "grad_norm": 0.656215250492096, "learning_rate": 0.0005, "loss": 1.5891971588134766, "step": 37862 }, { "epoch": 2.984864012613323, "grad_norm": 0.5176712274551392, "learning_rate": 0.0005, "loss": 1.4583556652069092, "step": 37863 }, { "epoch": 2.984942845880962, "grad_norm": 0.5399909019470215, "learning_rate": 0.0005, "loss": 1.3518733978271484, "step": 37864 }, { "epoch": 2.9850216791486006, "grad_norm": 0.41429606080055237, "learning_rate": 0.0005, "loss": 1.1525824069976807, "step": 37865 }, { "epoch": 2.9851005124162397, "grad_norm": 0.4819020628929138, "learning_rate": 0.0005, "loss": 1.3182353973388672, "step": 37866 }, { "epoch": 2.9851793456838784, "grad_norm": 0.6178671717643738, "learning_rate": 0.0005, "loss": 1.3166301250457764, "step": 37867 }, { "epoch": 2.9852581789515176, "grad_norm": 0.4687419533729553, "learning_rate": 0.0005, "loss": 1.2084016799926758, "step": 37868 }, { "epoch": 2.9853370122191567, "grad_norm": 0.4369220435619354, "learning_rate": 0.0005, "loss": 1.1709550619125366, "step": 37869 }, { "epoch": 2.9854158454867954, "grad_norm": 0.7404733896255493, "learning_rate": 0.0005, "loss": 1.2248296737670898, "step": 37870 }, { "epoch": 2.9854946787544345, "grad_norm": 0.486631840467453, "learning_rate": 0.0005, "loss": 1.3571943044662476, "step": 37871 }, { "epoch": 2.985573512022073, "grad_norm": 0.5244007110595703, "learning_rate": 0.0005, "loss": 1.1989068984985352, "step": 37872 }, { "epoch": 2.9856523452897124, "grad_norm": 0.5020921230316162, "learning_rate": 0.0005, "loss": 1.1986926794052124, "step": 37873 }, { "epoch": 2.985731178557351, "grad_norm": 0.5254316926002502, "learning_rate": 0.0005, "loss": 1.2902419567108154, "step": 37874 }, { "epoch": 2.98581001182499, "grad_norm": 0.4487181603908539, "learning_rate": 0.0005, "loss": 1.2971280813217163, "step": 37875 }, { "epoch": 2.985888845092629, "grad_norm": 0.5085842609405518, "learning_rate": 0.0005, "loss": 1.3318685293197632, "step": 37876 }, { "epoch": 2.985967678360268, "grad_norm": 0.48550236225128174, "learning_rate": 0.0005, "loss": 1.3631342649459839, "step": 37877 }, { "epoch": 2.986046511627907, "grad_norm": 0.6157710552215576, "learning_rate": 0.0005, "loss": 1.3145405054092407, "step": 37878 }, { "epoch": 2.986125344895546, "grad_norm": 0.5046636462211609, "learning_rate": 0.0005, "loss": 1.3024719953536987, "step": 37879 }, { "epoch": 2.986204178163185, "grad_norm": 0.4220755398273468, "learning_rate": 0.0005, "loss": 1.28120756149292, "step": 37880 }, { "epoch": 2.9862830114308236, "grad_norm": 0.4311779737472534, "learning_rate": 0.0005, "loss": 1.3802924156188965, "step": 37881 }, { "epoch": 2.9863618446984628, "grad_norm": 0.5279558897018433, "learning_rate": 0.0005, "loss": 1.3272695541381836, "step": 37882 }, { "epoch": 2.986440677966102, "grad_norm": 0.6595558524131775, "learning_rate": 0.0005, "loss": 1.4005206823349, "step": 37883 }, { "epoch": 2.9865195112337406, "grad_norm": 0.650118887424469, "learning_rate": 0.0005, "loss": 1.421353816986084, "step": 37884 }, { "epoch": 2.9865983445013797, "grad_norm": 0.5022082924842834, "learning_rate": 0.0005, "loss": 1.313551664352417, "step": 37885 }, { "epoch": 2.9866771777690184, "grad_norm": 0.48172280192375183, "learning_rate": 0.0005, "loss": 1.3004558086395264, "step": 37886 }, { "epoch": 2.9867560110366576, "grad_norm": 0.719968318939209, "learning_rate": 0.0005, "loss": 1.421760082244873, "step": 37887 }, { "epoch": 2.9868348443042967, "grad_norm": 0.6631461977958679, "learning_rate": 0.0005, "loss": 1.3663344383239746, "step": 37888 }, { "epoch": 2.9869136775719354, "grad_norm": 0.5277864336967468, "learning_rate": 0.0005, "loss": 1.326442003250122, "step": 37889 }, { "epoch": 2.986992510839574, "grad_norm": 0.4933567941188812, "learning_rate": 0.0005, "loss": 1.3733654022216797, "step": 37890 }, { "epoch": 2.987071344107213, "grad_norm": 0.4336526691913605, "learning_rate": 0.0005, "loss": 1.4093255996704102, "step": 37891 }, { "epoch": 2.9871501773748523, "grad_norm": 0.4714505970478058, "learning_rate": 0.0005, "loss": 1.2890963554382324, "step": 37892 }, { "epoch": 2.987229010642491, "grad_norm": 0.7074570655822754, "learning_rate": 0.0005, "loss": 1.4629440307617188, "step": 37893 }, { "epoch": 2.98730784391013, "grad_norm": 0.4687076807022095, "learning_rate": 0.0005, "loss": 1.331011176109314, "step": 37894 }, { "epoch": 2.987386677177769, "grad_norm": 0.46756553649902344, "learning_rate": 0.0005, "loss": 1.2358424663543701, "step": 37895 }, { "epoch": 2.987465510445408, "grad_norm": 1.397051215171814, "learning_rate": 0.0005, "loss": 1.3808484077453613, "step": 37896 }, { "epoch": 2.987544343713047, "grad_norm": 0.4877570569515228, "learning_rate": 0.0005, "loss": 1.1986594200134277, "step": 37897 }, { "epoch": 2.987623176980686, "grad_norm": 0.48172280192375183, "learning_rate": 0.0005, "loss": 1.2568279504776, "step": 37898 }, { "epoch": 2.987702010248325, "grad_norm": 0.47742289304733276, "learning_rate": 0.0005, "loss": 1.2723159790039062, "step": 37899 }, { "epoch": 2.9877808435159636, "grad_norm": 0.4846588969230652, "learning_rate": 0.0005, "loss": 1.4172470569610596, "step": 37900 }, { "epoch": 2.9878596767836028, "grad_norm": 0.4958292543888092, "learning_rate": 0.0005, "loss": 1.280577540397644, "step": 37901 }, { "epoch": 2.987938510051242, "grad_norm": 0.4366641640663147, "learning_rate": 0.0005, "loss": 1.2786015272140503, "step": 37902 }, { "epoch": 2.9880173433188806, "grad_norm": 0.48327553272247314, "learning_rate": 0.0005, "loss": 1.4711353778839111, "step": 37903 }, { "epoch": 2.9880961765865193, "grad_norm": 0.5552796125411987, "learning_rate": 0.0005, "loss": 1.6092770099639893, "step": 37904 }, { "epoch": 2.9881750098541584, "grad_norm": 0.5890708565711975, "learning_rate": 0.0005, "loss": 1.4170150756835938, "step": 37905 }, { "epoch": 2.9882538431217975, "grad_norm": 0.5464926362037659, "learning_rate": 0.0005, "loss": 1.344449758529663, "step": 37906 }, { "epoch": 2.9883326763894362, "grad_norm": 0.4824848771095276, "learning_rate": 0.0005, "loss": 1.3357174396514893, "step": 37907 }, { "epoch": 2.9884115096570754, "grad_norm": 0.5088561177253723, "learning_rate": 0.0005, "loss": 1.5851225852966309, "step": 37908 }, { "epoch": 2.988490342924714, "grad_norm": 0.5320785641670227, "learning_rate": 0.0005, "loss": 1.3957555294036865, "step": 37909 }, { "epoch": 2.988569176192353, "grad_norm": 0.45232850313186646, "learning_rate": 0.0005, "loss": 1.4343154430389404, "step": 37910 }, { "epoch": 2.9886480094599923, "grad_norm": 0.5878507494926453, "learning_rate": 0.0005, "loss": 1.2508869171142578, "step": 37911 }, { "epoch": 2.988726842727631, "grad_norm": 0.671114444732666, "learning_rate": 0.0005, "loss": 1.325145959854126, "step": 37912 }, { "epoch": 2.98880567599527, "grad_norm": 0.47555285692214966, "learning_rate": 0.0005, "loss": 1.2573411464691162, "step": 37913 }, { "epoch": 2.988884509262909, "grad_norm": 0.4631212651729584, "learning_rate": 0.0005, "loss": 1.3588703870773315, "step": 37914 }, { "epoch": 2.988963342530548, "grad_norm": 0.5876234173774719, "learning_rate": 0.0005, "loss": 1.5651826858520508, "step": 37915 }, { "epoch": 2.989042175798187, "grad_norm": 0.5450136661529541, "learning_rate": 0.0005, "loss": 1.0143449306488037, "step": 37916 }, { "epoch": 2.989121009065826, "grad_norm": 0.4277699291706085, "learning_rate": 0.0005, "loss": 1.1536835432052612, "step": 37917 }, { "epoch": 2.9891998423334645, "grad_norm": 0.48625320196151733, "learning_rate": 0.0005, "loss": 1.2783875465393066, "step": 37918 }, { "epoch": 2.9892786756011036, "grad_norm": 0.4465433955192566, "learning_rate": 0.0005, "loss": 1.335141658782959, "step": 37919 }, { "epoch": 2.9893575088687427, "grad_norm": 0.9824240803718567, "learning_rate": 0.0005, "loss": 1.2733275890350342, "step": 37920 }, { "epoch": 2.9894363421363814, "grad_norm": 0.4548853635787964, "learning_rate": 0.0005, "loss": 1.248570442199707, "step": 37921 }, { "epoch": 2.9895151754040206, "grad_norm": 0.6136233806610107, "learning_rate": 0.0005, "loss": 1.3814306259155273, "step": 37922 }, { "epoch": 2.9895940086716593, "grad_norm": 0.4709720313549042, "learning_rate": 0.0005, "loss": 1.15403413772583, "step": 37923 }, { "epoch": 2.9896728419392984, "grad_norm": 0.4827895760536194, "learning_rate": 0.0005, "loss": 1.2671518325805664, "step": 37924 }, { "epoch": 2.9897516752069375, "grad_norm": 0.599524199962616, "learning_rate": 0.0005, "loss": 1.1410870552062988, "step": 37925 }, { "epoch": 2.989830508474576, "grad_norm": 0.4770634174346924, "learning_rate": 0.0005, "loss": 1.2582913637161255, "step": 37926 }, { "epoch": 2.9899093417422153, "grad_norm": 0.7110377550125122, "learning_rate": 0.0005, "loss": 1.2010419368743896, "step": 37927 }, { "epoch": 2.989988175009854, "grad_norm": 0.47584593296051025, "learning_rate": 0.0005, "loss": 1.3685133457183838, "step": 37928 }, { "epoch": 2.990067008277493, "grad_norm": 0.4206175208091736, "learning_rate": 0.0005, "loss": 1.213095784187317, "step": 37929 }, { "epoch": 2.9901458415451323, "grad_norm": 0.4460391402244568, "learning_rate": 0.0005, "loss": 1.340963363647461, "step": 37930 }, { "epoch": 2.990224674812771, "grad_norm": 0.5280495882034302, "learning_rate": 0.0005, "loss": 1.2165133953094482, "step": 37931 }, { "epoch": 2.9903035080804097, "grad_norm": 0.5696212649345398, "learning_rate": 0.0005, "loss": 1.320936918258667, "step": 37932 }, { "epoch": 2.990382341348049, "grad_norm": 0.412801057100296, "learning_rate": 0.0005, "loss": 1.176172137260437, "step": 37933 }, { "epoch": 2.990461174615688, "grad_norm": 0.4596409201622009, "learning_rate": 0.0005, "loss": 1.471209168434143, "step": 37934 }, { "epoch": 2.9905400078833266, "grad_norm": 0.46727338433265686, "learning_rate": 0.0005, "loss": 1.1509034633636475, "step": 37935 }, { "epoch": 2.9906188411509658, "grad_norm": 0.5490047335624695, "learning_rate": 0.0005, "loss": 1.2542974948883057, "step": 37936 }, { "epoch": 2.9906976744186045, "grad_norm": 0.48823925852775574, "learning_rate": 0.0005, "loss": 1.3735865354537964, "step": 37937 }, { "epoch": 2.9907765076862436, "grad_norm": 0.5249124765396118, "learning_rate": 0.0005, "loss": 1.2945027351379395, "step": 37938 }, { "epoch": 2.9908553409538827, "grad_norm": 0.5041276216506958, "learning_rate": 0.0005, "loss": 1.394115924835205, "step": 37939 }, { "epoch": 2.9909341742215214, "grad_norm": 0.4762544333934784, "learning_rate": 0.0005, "loss": 1.1311447620391846, "step": 37940 }, { "epoch": 2.9910130074891605, "grad_norm": 0.45073822140693665, "learning_rate": 0.0005, "loss": 1.4240953922271729, "step": 37941 }, { "epoch": 2.9910918407567992, "grad_norm": 0.5309779644012451, "learning_rate": 0.0005, "loss": 1.3892455101013184, "step": 37942 }, { "epoch": 2.9911706740244384, "grad_norm": 0.5983359217643738, "learning_rate": 0.0005, "loss": 1.4121755361557007, "step": 37943 }, { "epoch": 2.9912495072920775, "grad_norm": 0.4606142044067383, "learning_rate": 0.0005, "loss": 1.2727934122085571, "step": 37944 }, { "epoch": 2.991328340559716, "grad_norm": 0.46710485219955444, "learning_rate": 0.0005, "loss": 1.180153727531433, "step": 37945 }, { "epoch": 2.991407173827355, "grad_norm": 0.481792151927948, "learning_rate": 0.0005, "loss": 1.4390056133270264, "step": 37946 }, { "epoch": 2.991486007094994, "grad_norm": 0.5721816420555115, "learning_rate": 0.0005, "loss": 1.1381709575653076, "step": 37947 }, { "epoch": 2.991564840362633, "grad_norm": 0.47233131527900696, "learning_rate": 0.0005, "loss": 1.1984080076217651, "step": 37948 }, { "epoch": 2.991643673630272, "grad_norm": 0.49920469522476196, "learning_rate": 0.0005, "loss": 1.2422704696655273, "step": 37949 }, { "epoch": 2.991722506897911, "grad_norm": 0.6207696199417114, "learning_rate": 0.0005, "loss": 1.3916270732879639, "step": 37950 }, { "epoch": 2.9918013401655497, "grad_norm": 0.5302921533584595, "learning_rate": 0.0005, "loss": 1.2822489738464355, "step": 37951 }, { "epoch": 2.991880173433189, "grad_norm": 0.45075467228889465, "learning_rate": 0.0005, "loss": 1.334141492843628, "step": 37952 }, { "epoch": 2.991959006700828, "grad_norm": 1.4992941617965698, "learning_rate": 0.0005, "loss": 1.3060115575790405, "step": 37953 }, { "epoch": 2.9920378399684666, "grad_norm": 0.6221752762794495, "learning_rate": 0.0005, "loss": 1.4902989864349365, "step": 37954 }, { "epoch": 2.9921166732361058, "grad_norm": 0.5874463319778442, "learning_rate": 0.0005, "loss": 1.4045109748840332, "step": 37955 }, { "epoch": 2.9921955065037444, "grad_norm": 0.4881225824356079, "learning_rate": 0.0005, "loss": 1.3900113105773926, "step": 37956 }, { "epoch": 2.9922743397713836, "grad_norm": 0.471627414226532, "learning_rate": 0.0005, "loss": 1.2591749429702759, "step": 37957 }, { "epoch": 2.9923531730390227, "grad_norm": 0.46038585901260376, "learning_rate": 0.0005, "loss": 1.1717244386672974, "step": 37958 }, { "epoch": 2.9924320063066614, "grad_norm": 0.5886897444725037, "learning_rate": 0.0005, "loss": 1.3104795217514038, "step": 37959 }, { "epoch": 2.9925108395743, "grad_norm": 0.47952142357826233, "learning_rate": 0.0005, "loss": 1.3006534576416016, "step": 37960 }, { "epoch": 2.992589672841939, "grad_norm": 0.6555247902870178, "learning_rate": 0.0005, "loss": 1.3491415977478027, "step": 37961 }, { "epoch": 2.9926685061095784, "grad_norm": 0.4770698547363281, "learning_rate": 0.0005, "loss": 1.3715037107467651, "step": 37962 }, { "epoch": 2.992747339377217, "grad_norm": 0.5306077003479004, "learning_rate": 0.0005, "loss": 1.6630539894104004, "step": 37963 }, { "epoch": 2.992826172644856, "grad_norm": 0.6352501511573792, "learning_rate": 0.0005, "loss": 1.3672480583190918, "step": 37964 }, { "epoch": 2.992905005912495, "grad_norm": 0.5208380818367004, "learning_rate": 0.0005, "loss": 1.404760479927063, "step": 37965 }, { "epoch": 2.992983839180134, "grad_norm": 0.5647348165512085, "learning_rate": 0.0005, "loss": 1.4149843454360962, "step": 37966 }, { "epoch": 2.993062672447773, "grad_norm": 0.5135384798049927, "learning_rate": 0.0005, "loss": 1.249439001083374, "step": 37967 }, { "epoch": 2.993141505715412, "grad_norm": 0.5925917625427246, "learning_rate": 0.0005, "loss": 1.3330347537994385, "step": 37968 }, { "epoch": 2.993220338983051, "grad_norm": 0.620803952217102, "learning_rate": 0.0005, "loss": 1.3905973434448242, "step": 37969 }, { "epoch": 2.9932991722506896, "grad_norm": 0.5363916158676147, "learning_rate": 0.0005, "loss": 1.3215594291687012, "step": 37970 }, { "epoch": 2.9933780055183288, "grad_norm": 0.49420350790023804, "learning_rate": 0.0005, "loss": 1.243640661239624, "step": 37971 }, { "epoch": 2.993456838785968, "grad_norm": 0.4119705259799957, "learning_rate": 0.0005, "loss": 1.2011537551879883, "step": 37972 }, { "epoch": 2.9935356720536066, "grad_norm": 0.5397916436195374, "learning_rate": 0.0005, "loss": 1.2162981033325195, "step": 37973 }, { "epoch": 2.9936145053212453, "grad_norm": 0.4811769425868988, "learning_rate": 0.0005, "loss": 1.149728536605835, "step": 37974 }, { "epoch": 2.9936933385888844, "grad_norm": 0.43196484446525574, "learning_rate": 0.0005, "loss": 1.297163724899292, "step": 37975 }, { "epoch": 2.9937721718565236, "grad_norm": 0.5471370816230774, "learning_rate": 0.0005, "loss": 1.4339196681976318, "step": 37976 }, { "epoch": 2.9938510051241622, "grad_norm": 0.5380566120147705, "learning_rate": 0.0005, "loss": 1.5701179504394531, "step": 37977 }, { "epoch": 2.9939298383918014, "grad_norm": 0.6564227938652039, "learning_rate": 0.0005, "loss": 1.385722279548645, "step": 37978 }, { "epoch": 2.99400867165944, "grad_norm": 0.5356816649436951, "learning_rate": 0.0005, "loss": 1.4363998174667358, "step": 37979 }, { "epoch": 2.994087504927079, "grad_norm": 0.5935368537902832, "learning_rate": 0.0005, "loss": 1.1768791675567627, "step": 37980 }, { "epoch": 2.9941663381947183, "grad_norm": 0.4811026453971863, "learning_rate": 0.0005, "loss": 1.3192468881607056, "step": 37981 }, { "epoch": 2.994245171462357, "grad_norm": 0.46352049708366394, "learning_rate": 0.0005, "loss": 1.2341662645339966, "step": 37982 }, { "epoch": 2.994324004729996, "grad_norm": 0.5612407326698303, "learning_rate": 0.0005, "loss": 1.5323528051376343, "step": 37983 }, { "epoch": 2.994402837997635, "grad_norm": 0.5229946374893188, "learning_rate": 0.0005, "loss": 1.325432538986206, "step": 37984 }, { "epoch": 2.994481671265274, "grad_norm": 0.5081897377967834, "learning_rate": 0.0005, "loss": 1.2308484315872192, "step": 37985 }, { "epoch": 2.994560504532913, "grad_norm": 0.6120437979698181, "learning_rate": 0.0005, "loss": 1.3984993696212769, "step": 37986 }, { "epoch": 2.994639337800552, "grad_norm": 0.7167268395423889, "learning_rate": 0.0005, "loss": 1.3658568859100342, "step": 37987 }, { "epoch": 2.994718171068191, "grad_norm": 0.4307289719581604, "learning_rate": 0.0005, "loss": 1.4632185697555542, "step": 37988 }, { "epoch": 2.9947970043358296, "grad_norm": 0.5486027598381042, "learning_rate": 0.0005, "loss": 1.4323065280914307, "step": 37989 }, { "epoch": 2.9948758376034688, "grad_norm": 0.5107354521751404, "learning_rate": 0.0005, "loss": 1.3077538013458252, "step": 37990 }, { "epoch": 2.9949546708711074, "grad_norm": 0.5933847427368164, "learning_rate": 0.0005, "loss": 1.5352336168289185, "step": 37991 }, { "epoch": 2.9950335041387466, "grad_norm": 0.5845803618431091, "learning_rate": 0.0005, "loss": 1.4027695655822754, "step": 37992 }, { "epoch": 2.9951123374063853, "grad_norm": 0.5230991244316101, "learning_rate": 0.0005, "loss": 1.1934953927993774, "step": 37993 }, { "epoch": 2.9951911706740244, "grad_norm": 0.797806441783905, "learning_rate": 0.0005, "loss": 1.3956122398376465, "step": 37994 }, { "epoch": 2.9952700039416635, "grad_norm": 0.42934751510620117, "learning_rate": 0.0005, "loss": 1.462841272354126, "step": 37995 }, { "epoch": 2.9953488372093022, "grad_norm": 0.5750279426574707, "learning_rate": 0.0005, "loss": 1.3840365409851074, "step": 37996 }, { "epoch": 2.9954276704769414, "grad_norm": 0.6907225847244263, "learning_rate": 0.0005, "loss": 1.2921876907348633, "step": 37997 }, { "epoch": 2.99550650374458, "grad_norm": 0.5433467030525208, "learning_rate": 0.0005, "loss": 1.3817788362503052, "step": 37998 }, { "epoch": 2.995585337012219, "grad_norm": 0.516510546207428, "learning_rate": 0.0005, "loss": 1.3515632152557373, "step": 37999 }, { "epoch": 2.9956641702798583, "grad_norm": 0.5077895522117615, "learning_rate": 0.0005, "loss": 1.4885475635528564, "step": 38000 }, { "epoch": 2.995743003547497, "grad_norm": 0.5576490759849548, "learning_rate": 0.0005, "loss": 1.41080904006958, "step": 38001 }, { "epoch": 2.995821836815136, "grad_norm": 0.44953441619873047, "learning_rate": 0.0005, "loss": 1.1028947830200195, "step": 38002 }, { "epoch": 2.995900670082775, "grad_norm": 0.4608948528766632, "learning_rate": 0.0005, "loss": 1.3950672149658203, "step": 38003 }, { "epoch": 2.995979503350414, "grad_norm": 0.4685060381889343, "learning_rate": 0.0005, "loss": 1.2999248504638672, "step": 38004 }, { "epoch": 2.9960583366180527, "grad_norm": 0.5027127265930176, "learning_rate": 0.0005, "loss": 1.3580421209335327, "step": 38005 }, { "epoch": 2.996137169885692, "grad_norm": 0.4924803078174591, "learning_rate": 0.0005, "loss": 1.1373142004013062, "step": 38006 }, { "epoch": 2.9962160031533305, "grad_norm": 0.4445236027240753, "learning_rate": 0.0005, "loss": 1.1883323192596436, "step": 38007 }, { "epoch": 2.9962948364209696, "grad_norm": 0.572131872177124, "learning_rate": 0.0005, "loss": 1.594780683517456, "step": 38008 }, { "epoch": 2.9963736696886087, "grad_norm": 0.5083289742469788, "learning_rate": 0.0005, "loss": 1.294116735458374, "step": 38009 }, { "epoch": 2.9964525029562474, "grad_norm": 0.5793941617012024, "learning_rate": 0.0005, "loss": 1.4903593063354492, "step": 38010 }, { "epoch": 2.9965313362238866, "grad_norm": 0.44001030921936035, "learning_rate": 0.0005, "loss": 1.3443272113800049, "step": 38011 }, { "epoch": 2.9966101694915253, "grad_norm": 0.5118448734283447, "learning_rate": 0.0005, "loss": 1.257291555404663, "step": 38012 }, { "epoch": 2.9966890027591644, "grad_norm": 0.4333665668964386, "learning_rate": 0.0005, "loss": 1.2937783002853394, "step": 38013 }, { "epoch": 2.9967678360268035, "grad_norm": 0.4357512891292572, "learning_rate": 0.0005, "loss": 1.2193796634674072, "step": 38014 }, { "epoch": 2.996846669294442, "grad_norm": 0.45822909474372864, "learning_rate": 0.0005, "loss": 1.2964175939559937, "step": 38015 }, { "epoch": 2.9969255025620813, "grad_norm": 0.47870129346847534, "learning_rate": 0.0005, "loss": 1.4499969482421875, "step": 38016 }, { "epoch": 2.99700433582972, "grad_norm": 1.274158000946045, "learning_rate": 0.0005, "loss": 1.2606096267700195, "step": 38017 }, { "epoch": 2.997083169097359, "grad_norm": 0.5881277918815613, "learning_rate": 0.0005, "loss": 1.370527744293213, "step": 38018 }, { "epoch": 2.9971620023649983, "grad_norm": 0.4433802366256714, "learning_rate": 0.0005, "loss": 1.185387134552002, "step": 38019 }, { "epoch": 2.997240835632637, "grad_norm": 0.5129249691963196, "learning_rate": 0.0005, "loss": 1.265906572341919, "step": 38020 }, { "epoch": 2.9973196689002757, "grad_norm": 0.5972183346748352, "learning_rate": 0.0005, "loss": 1.372222661972046, "step": 38021 }, { "epoch": 2.997398502167915, "grad_norm": 0.49543797969818115, "learning_rate": 0.0005, "loss": 1.4341726303100586, "step": 38022 }, { "epoch": 2.997477335435554, "grad_norm": 0.4777059853076935, "learning_rate": 0.0005, "loss": 1.157549500465393, "step": 38023 }, { "epoch": 2.9975561687031926, "grad_norm": 0.5473328232765198, "learning_rate": 0.0005, "loss": 1.1058261394500732, "step": 38024 }, { "epoch": 2.9976350019708318, "grad_norm": 0.5287912487983704, "learning_rate": 0.0005, "loss": 1.2174971103668213, "step": 38025 }, { "epoch": 2.9977138352384705, "grad_norm": 0.47773441672325134, "learning_rate": 0.0005, "loss": 1.1525778770446777, "step": 38026 }, { "epoch": 2.9977926685061096, "grad_norm": 0.6196656227111816, "learning_rate": 0.0005, "loss": 1.5075130462646484, "step": 38027 }, { "epoch": 2.9978715017737487, "grad_norm": 0.5115458965301514, "learning_rate": 0.0005, "loss": 1.2383997440338135, "step": 38028 }, { "epoch": 2.9979503350413874, "grad_norm": 0.49518251419067383, "learning_rate": 0.0005, "loss": 1.2324957847595215, "step": 38029 }, { "epoch": 2.9980291683090265, "grad_norm": 0.5457821488380432, "learning_rate": 0.0005, "loss": 1.147982120513916, "step": 38030 }, { "epoch": 2.9981080015766652, "grad_norm": 0.44035419821739197, "learning_rate": 0.0005, "loss": 1.1677477359771729, "step": 38031 }, { "epoch": 2.9981868348443044, "grad_norm": 0.48320087790489197, "learning_rate": 0.0005, "loss": 1.2973897457122803, "step": 38032 }, { "epoch": 2.9982656681119435, "grad_norm": 0.5078787207603455, "learning_rate": 0.0005, "loss": 1.3652961254119873, "step": 38033 }, { "epoch": 2.998344501379582, "grad_norm": 0.47075775265693665, "learning_rate": 0.0005, "loss": 1.2242064476013184, "step": 38034 }, { "epoch": 2.998423334647221, "grad_norm": 0.5050289034843445, "learning_rate": 0.0005, "loss": 1.168178915977478, "step": 38035 }, { "epoch": 2.99850216791486, "grad_norm": 0.46929487586021423, "learning_rate": 0.0005, "loss": 1.2982802391052246, "step": 38036 }, { "epoch": 2.998581001182499, "grad_norm": 0.45954886078834534, "learning_rate": 0.0005, "loss": 1.3262414932250977, "step": 38037 }, { "epoch": 2.998659834450138, "grad_norm": 0.5061097741127014, "learning_rate": 0.0005, "loss": 1.3827606439590454, "step": 38038 }, { "epoch": 2.998738667717777, "grad_norm": 0.5840315818786621, "learning_rate": 0.0005, "loss": 1.2715561389923096, "step": 38039 }, { "epoch": 2.9988175009854157, "grad_norm": 0.5296372175216675, "learning_rate": 0.0005, "loss": 1.3522231578826904, "step": 38040 }, { "epoch": 2.998896334253055, "grad_norm": 0.47775962948799133, "learning_rate": 0.0005, "loss": 1.1958898305892944, "step": 38041 }, { "epoch": 2.998975167520694, "grad_norm": 0.4876762926578522, "learning_rate": 0.0005, "loss": 1.3820065259933472, "step": 38042 }, { "epoch": 2.9990540007883326, "grad_norm": 0.4635215997695923, "learning_rate": 0.0005, "loss": 1.3968158960342407, "step": 38043 }, { "epoch": 2.9991328340559718, "grad_norm": 0.41806337237358093, "learning_rate": 0.0005, "loss": 1.1517666578292847, "step": 38044 }, { "epoch": 2.9992116673236104, "grad_norm": 0.4917979836463928, "learning_rate": 0.0005, "loss": 1.2800548076629639, "step": 38045 }, { "epoch": 2.9992905005912496, "grad_norm": 0.4275851547718048, "learning_rate": 0.0005, "loss": 1.3259639739990234, "step": 38046 }, { "epoch": 2.9993693338588887, "grad_norm": 0.5180516839027405, "learning_rate": 0.0005, "loss": 1.6934837102890015, "step": 38047 }, { "epoch": 2.9994481671265274, "grad_norm": 0.8427051901817322, "learning_rate": 0.0005, "loss": 1.4579579830169678, "step": 38048 }, { "epoch": 2.999527000394166, "grad_norm": 0.5306013226509094, "learning_rate": 0.0005, "loss": 1.3966405391693115, "step": 38049 }, { "epoch": 2.999605833661805, "grad_norm": 0.6089358925819397, "learning_rate": 0.0005, "loss": 1.5962724685668945, "step": 38050 }, { "epoch": 2.9996846669294444, "grad_norm": 0.6269103288650513, "learning_rate": 0.0005, "loss": 1.3454458713531494, "step": 38051 }, { "epoch": 2.999763500197083, "grad_norm": 0.4778967499732971, "learning_rate": 0.0005, "loss": 1.397801160812378, "step": 38052 }, { "epoch": 2.999842333464722, "grad_norm": 0.5647357106208801, "learning_rate": 0.0005, "loss": 1.3576332330703735, "step": 38053 }, { "epoch": 2.999921166732361, "grad_norm": 0.5662685632705688, "learning_rate": 0.0005, "loss": 1.6109949350357056, "step": 38054 }, { "epoch": 3.0, "grad_norm": 0.43602100014686584, "learning_rate": 0.0005, "loss": 1.1965525150299072, "step": 38055 }, { "epoch": 3.000078833267639, "grad_norm": 0.5172255039215088, "learning_rate": 0.0005, "loss": 1.3545863628387451, "step": 38056 }, { "epoch": 3.000157666535278, "grad_norm": 0.41059207916259766, "learning_rate": 0.0005, "loss": 1.2548128366470337, "step": 38057 }, { "epoch": 3.000236499802917, "grad_norm": 0.4304202198982239, "learning_rate": 0.0005, "loss": 1.3177459239959717, "step": 38058 }, { "epoch": 3.0003153330705556, "grad_norm": 0.42955583333969116, "learning_rate": 0.0005, "loss": 1.1836788654327393, "step": 38059 }, { "epoch": 3.000394166338195, "grad_norm": 0.4405948221683502, "learning_rate": 0.0005, "loss": 1.297412395477295, "step": 38060 }, { "epoch": 3.0004729996058335, "grad_norm": 0.42439496517181396, "learning_rate": 0.0005, "loss": 1.216888189315796, "step": 38061 }, { "epoch": 3.0005518328734726, "grad_norm": 0.5153264403343201, "learning_rate": 0.0005, "loss": 1.315169095993042, "step": 38062 }, { "epoch": 3.0006306661411117, "grad_norm": 0.4057825803756714, "learning_rate": 0.0005, "loss": 1.2411847114562988, "step": 38063 }, { "epoch": 3.0007094994087504, "grad_norm": 0.47508785128593445, "learning_rate": 0.0005, "loss": 1.3015356063842773, "step": 38064 }, { "epoch": 3.0007883326763896, "grad_norm": 0.44200095534324646, "learning_rate": 0.0005, "loss": 1.2763731479644775, "step": 38065 }, { "epoch": 3.0008671659440282, "grad_norm": 0.4852166473865509, "learning_rate": 0.0005, "loss": 1.1978801488876343, "step": 38066 }, { "epoch": 3.0009459992116674, "grad_norm": 0.4915432929992676, "learning_rate": 0.0005, "loss": 1.441231369972229, "step": 38067 }, { "epoch": 3.001024832479306, "grad_norm": 0.4353177845478058, "learning_rate": 0.0005, "loss": 1.2225748300552368, "step": 38068 }, { "epoch": 3.001103665746945, "grad_norm": 0.4719703793525696, "learning_rate": 0.0005, "loss": 1.4264298677444458, "step": 38069 }, { "epoch": 3.0011824990145843, "grad_norm": 0.4909805357456207, "learning_rate": 0.0005, "loss": 1.3230371475219727, "step": 38070 }, { "epoch": 3.001261332282223, "grad_norm": 0.6335045695304871, "learning_rate": 0.0005, "loss": 1.5176029205322266, "step": 38071 }, { "epoch": 3.001340165549862, "grad_norm": 0.5409688353538513, "learning_rate": 0.0005, "loss": 1.5581475496292114, "step": 38072 }, { "epoch": 3.001418998817501, "grad_norm": 0.44178420305252075, "learning_rate": 0.0005, "loss": 1.2681001424789429, "step": 38073 }, { "epoch": 3.00149783208514, "grad_norm": 0.708337664604187, "learning_rate": 0.0005, "loss": 1.167382001876831, "step": 38074 }, { "epoch": 3.0015766653527787, "grad_norm": 0.4920859932899475, "learning_rate": 0.0005, "loss": 1.1467688083648682, "step": 38075 }, { "epoch": 3.001655498620418, "grad_norm": 0.3891891837120056, "learning_rate": 0.0005, "loss": 1.095047950744629, "step": 38076 }, { "epoch": 3.001734331888057, "grad_norm": 0.5269908308982849, "learning_rate": 0.0005, "loss": 1.3127979040145874, "step": 38077 }, { "epoch": 3.0018131651556956, "grad_norm": 0.5838935375213623, "learning_rate": 0.0005, "loss": 1.3574368953704834, "step": 38078 }, { "epoch": 3.0018919984233348, "grad_norm": 0.45476996898651123, "learning_rate": 0.0005, "loss": 1.1565356254577637, "step": 38079 }, { "epoch": 3.0019708316909735, "grad_norm": 0.4795803427696228, "learning_rate": 0.0005, "loss": 1.097564458847046, "step": 38080 }, { "epoch": 3.0020496649586126, "grad_norm": 0.42894309759140015, "learning_rate": 0.0005, "loss": 1.1819096803665161, "step": 38081 }, { "epoch": 3.0021284982262513, "grad_norm": 0.8226117491722107, "learning_rate": 0.0005, "loss": 1.443706750869751, "step": 38082 }, { "epoch": 3.0022073314938904, "grad_norm": 0.45651957392692566, "learning_rate": 0.0005, "loss": 1.2029693126678467, "step": 38083 }, { "epoch": 3.0022861647615295, "grad_norm": 0.686248242855072, "learning_rate": 0.0005, "loss": 1.1556925773620605, "step": 38084 }, { "epoch": 3.0023649980291682, "grad_norm": 0.501768946647644, "learning_rate": 0.0005, "loss": 1.1980942487716675, "step": 38085 }, { "epoch": 3.0024438312968074, "grad_norm": 0.4644217789173126, "learning_rate": 0.0005, "loss": 1.3094475269317627, "step": 38086 }, { "epoch": 3.002522664564446, "grad_norm": 0.4685550630092621, "learning_rate": 0.0005, "loss": 1.273277997970581, "step": 38087 }, { "epoch": 3.002601497832085, "grad_norm": 0.46221718192100525, "learning_rate": 0.0005, "loss": 1.2718981504440308, "step": 38088 }, { "epoch": 3.002680331099724, "grad_norm": 0.45714911818504333, "learning_rate": 0.0005, "loss": 1.0716594457626343, "step": 38089 }, { "epoch": 3.002759164367363, "grad_norm": 0.468082457780838, "learning_rate": 0.0005, "loss": 1.2771706581115723, "step": 38090 }, { "epoch": 3.002837997635002, "grad_norm": 0.47344261407852173, "learning_rate": 0.0005, "loss": 1.363579511642456, "step": 38091 }, { "epoch": 3.002916830902641, "grad_norm": 0.4576839804649353, "learning_rate": 0.0005, "loss": 1.2602474689483643, "step": 38092 }, { "epoch": 3.00299566417028, "grad_norm": 0.4149823784828186, "learning_rate": 0.0005, "loss": 1.2182159423828125, "step": 38093 }, { "epoch": 3.0030744974379187, "grad_norm": 0.4780523478984833, "learning_rate": 0.0005, "loss": 1.4997475147247314, "step": 38094 }, { "epoch": 3.003153330705558, "grad_norm": 0.5778812170028687, "learning_rate": 0.0005, "loss": 1.306782841682434, "step": 38095 }, { "epoch": 3.0032321639731965, "grad_norm": 0.4124795198440552, "learning_rate": 0.0005, "loss": 1.121260404586792, "step": 38096 }, { "epoch": 3.0033109972408356, "grad_norm": 0.4974762797355652, "learning_rate": 0.0005, "loss": 1.3123719692230225, "step": 38097 }, { "epoch": 3.0033898305084747, "grad_norm": 0.40751907229423523, "learning_rate": 0.0005, "loss": 1.1440097093582153, "step": 38098 }, { "epoch": 3.0034686637761134, "grad_norm": 0.6856984496116638, "learning_rate": 0.0005, "loss": 1.2868143320083618, "step": 38099 }, { "epoch": 3.0035474970437526, "grad_norm": 0.4962105453014374, "learning_rate": 0.0005, "loss": 1.1944478750228882, "step": 38100 }, { "epoch": 3.0036263303113913, "grad_norm": 0.3607669770717621, "learning_rate": 0.0005, "loss": 1.0991392135620117, "step": 38101 }, { "epoch": 3.0037051635790304, "grad_norm": 0.4439637064933777, "learning_rate": 0.0005, "loss": 1.240795373916626, "step": 38102 }, { "epoch": 3.003783996846669, "grad_norm": 0.5172920227050781, "learning_rate": 0.0005, "loss": 1.249295949935913, "step": 38103 }, { "epoch": 3.003862830114308, "grad_norm": 0.42850610613822937, "learning_rate": 0.0005, "loss": 1.332783818244934, "step": 38104 }, { "epoch": 3.0039416633819473, "grad_norm": 0.4153055250644684, "learning_rate": 0.0005, "loss": 1.3056788444519043, "step": 38105 }, { "epoch": 3.004020496649586, "grad_norm": 0.5213426351547241, "learning_rate": 0.0005, "loss": 1.3480279445648193, "step": 38106 }, { "epoch": 3.004099329917225, "grad_norm": 0.40891188383102417, "learning_rate": 0.0005, "loss": 1.1655198335647583, "step": 38107 }, { "epoch": 3.004178163184864, "grad_norm": 0.9279329180717468, "learning_rate": 0.0005, "loss": 1.356558918952942, "step": 38108 }, { "epoch": 3.004256996452503, "grad_norm": 0.34508293867111206, "learning_rate": 0.0005, "loss": 1.118791103363037, "step": 38109 }, { "epoch": 3.0043358297201417, "grad_norm": 0.4800216257572174, "learning_rate": 0.0005, "loss": 1.2014161348342896, "step": 38110 }, { "epoch": 3.004414662987781, "grad_norm": 0.4819624722003937, "learning_rate": 0.0005, "loss": 1.3858410120010376, "step": 38111 }, { "epoch": 3.00449349625542, "grad_norm": 0.4601002335548401, "learning_rate": 0.0005, "loss": 1.1420729160308838, "step": 38112 }, { "epoch": 3.0045723295230586, "grad_norm": 0.40661823749542236, "learning_rate": 0.0005, "loss": 1.0831502676010132, "step": 38113 }, { "epoch": 3.0046511627906978, "grad_norm": 0.5090672373771667, "learning_rate": 0.0005, "loss": 1.3151789903640747, "step": 38114 }, { "epoch": 3.0047299960583365, "grad_norm": 0.5105653405189514, "learning_rate": 0.0005, "loss": 1.4069006443023682, "step": 38115 }, { "epoch": 3.0048088293259756, "grad_norm": 0.4878351390361786, "learning_rate": 0.0005, "loss": 1.4060068130493164, "step": 38116 }, { "epoch": 3.0048876625936147, "grad_norm": 0.5933428406715393, "learning_rate": 0.0005, "loss": 1.3683758974075317, "step": 38117 }, { "epoch": 3.0049664958612534, "grad_norm": 0.6481956243515015, "learning_rate": 0.0005, "loss": 1.3108832836151123, "step": 38118 }, { "epoch": 3.0050453291288926, "grad_norm": 0.4882284700870514, "learning_rate": 0.0005, "loss": 1.4018194675445557, "step": 38119 }, { "epoch": 3.0051241623965312, "grad_norm": 0.44871655106544495, "learning_rate": 0.0005, "loss": 1.359512448310852, "step": 38120 }, { "epoch": 3.0052029956641704, "grad_norm": 0.5854924321174622, "learning_rate": 0.0005, "loss": 1.4372891187667847, "step": 38121 }, { "epoch": 3.005281828931809, "grad_norm": 0.45099055767059326, "learning_rate": 0.0005, "loss": 1.040327787399292, "step": 38122 }, { "epoch": 3.005360662199448, "grad_norm": 0.6370871663093567, "learning_rate": 0.0005, "loss": 1.318671703338623, "step": 38123 }, { "epoch": 3.0054394954670873, "grad_norm": 0.4602620601654053, "learning_rate": 0.0005, "loss": 1.4278264045715332, "step": 38124 }, { "epoch": 3.005518328734726, "grad_norm": 0.5124284029006958, "learning_rate": 0.0005, "loss": 1.2707246541976929, "step": 38125 }, { "epoch": 3.005597162002365, "grad_norm": 0.5033513903617859, "learning_rate": 0.0005, "loss": 1.2525626420974731, "step": 38126 }, { "epoch": 3.005675995270004, "grad_norm": 0.5960168838500977, "learning_rate": 0.0005, "loss": 1.3203628063201904, "step": 38127 }, { "epoch": 3.005754828537643, "grad_norm": 0.5537852644920349, "learning_rate": 0.0005, "loss": 1.3665781021118164, "step": 38128 }, { "epoch": 3.0058336618052817, "grad_norm": 0.4911436438560486, "learning_rate": 0.0005, "loss": 1.337953805923462, "step": 38129 }, { "epoch": 3.005912495072921, "grad_norm": 0.5000629425048828, "learning_rate": 0.0005, "loss": 1.1708574295043945, "step": 38130 }, { "epoch": 3.00599132834056, "grad_norm": 0.5026279091835022, "learning_rate": 0.0005, "loss": 1.1496012210845947, "step": 38131 }, { "epoch": 3.0060701616081986, "grad_norm": 0.4544657766819, "learning_rate": 0.0005, "loss": 1.268959879875183, "step": 38132 }, { "epoch": 3.0061489948758378, "grad_norm": 0.4778667390346527, "learning_rate": 0.0005, "loss": 1.3360583782196045, "step": 38133 }, { "epoch": 3.0062278281434764, "grad_norm": 0.44680526852607727, "learning_rate": 0.0005, "loss": 1.3104418516159058, "step": 38134 }, { "epoch": 3.0063066614111156, "grad_norm": 0.5280951261520386, "learning_rate": 0.0005, "loss": 1.2738711833953857, "step": 38135 }, { "epoch": 3.0063854946787543, "grad_norm": 0.3994695842266083, "learning_rate": 0.0005, "loss": 1.0743705034255981, "step": 38136 }, { "epoch": 3.0064643279463934, "grad_norm": 0.42298296093940735, "learning_rate": 0.0005, "loss": 1.1638708114624023, "step": 38137 }, { "epoch": 3.0065431612140325, "grad_norm": 0.37438955903053284, "learning_rate": 0.0005, "loss": 1.175703763961792, "step": 38138 }, { "epoch": 3.0066219944816712, "grad_norm": 0.5004730224609375, "learning_rate": 0.0005, "loss": 1.411717176437378, "step": 38139 }, { "epoch": 3.0067008277493104, "grad_norm": 0.48702138662338257, "learning_rate": 0.0005, "loss": 1.1562256813049316, "step": 38140 }, { "epoch": 3.006779661016949, "grad_norm": 0.6612535119056702, "learning_rate": 0.0005, "loss": 1.29154372215271, "step": 38141 }, { "epoch": 3.006858494284588, "grad_norm": 0.43447116017341614, "learning_rate": 0.0005, "loss": 1.2348039150238037, "step": 38142 }, { "epoch": 3.006937327552227, "grad_norm": 0.4494715631008148, "learning_rate": 0.0005, "loss": 1.3379149436950684, "step": 38143 }, { "epoch": 3.007016160819866, "grad_norm": 0.4127904772758484, "learning_rate": 0.0005, "loss": 1.1646978855133057, "step": 38144 }, { "epoch": 3.007094994087505, "grad_norm": 0.47240686416625977, "learning_rate": 0.0005, "loss": 1.3233766555786133, "step": 38145 }, { "epoch": 3.007173827355144, "grad_norm": 0.4955865144729614, "learning_rate": 0.0005, "loss": 1.303625464439392, "step": 38146 }, { "epoch": 3.007252660622783, "grad_norm": 0.5072349309921265, "learning_rate": 0.0005, "loss": 1.1913480758666992, "step": 38147 }, { "epoch": 3.0073314938904216, "grad_norm": 0.40100884437561035, "learning_rate": 0.0005, "loss": 1.153003215789795, "step": 38148 }, { "epoch": 3.007410327158061, "grad_norm": 0.4864492416381836, "learning_rate": 0.0005, "loss": 1.2763283252716064, "step": 38149 }, { "epoch": 3.0074891604256995, "grad_norm": 0.4645169675350189, "learning_rate": 0.0005, "loss": 1.1663087606430054, "step": 38150 }, { "epoch": 3.0075679936933386, "grad_norm": 0.4291418194770813, "learning_rate": 0.0005, "loss": 1.3435559272766113, "step": 38151 }, { "epoch": 3.0076468269609777, "grad_norm": 0.5006572604179382, "learning_rate": 0.0005, "loss": 1.2904789447784424, "step": 38152 }, { "epoch": 3.0077256602286164, "grad_norm": 0.42850542068481445, "learning_rate": 0.0005, "loss": 1.2210180759429932, "step": 38153 }, { "epoch": 3.0078044934962556, "grad_norm": 0.554655909538269, "learning_rate": 0.0005, "loss": 1.3943729400634766, "step": 38154 }, { "epoch": 3.0078833267638942, "grad_norm": 0.4888518452644348, "learning_rate": 0.0005, "loss": 1.26765775680542, "step": 38155 }, { "epoch": 3.0079621600315334, "grad_norm": 0.5561066269874573, "learning_rate": 0.0005, "loss": 1.3726557493209839, "step": 38156 }, { "epoch": 3.008040993299172, "grad_norm": 0.5335647463798523, "learning_rate": 0.0005, "loss": 1.403747797012329, "step": 38157 }, { "epoch": 3.008119826566811, "grad_norm": 0.42854776978492737, "learning_rate": 0.0005, "loss": 1.1726174354553223, "step": 38158 }, { "epoch": 3.0081986598344503, "grad_norm": 0.49884676933288574, "learning_rate": 0.0005, "loss": 1.2029536962509155, "step": 38159 }, { "epoch": 3.008277493102089, "grad_norm": 0.502119243144989, "learning_rate": 0.0005, "loss": 1.3175034523010254, "step": 38160 }, { "epoch": 3.008356326369728, "grad_norm": 0.5279127359390259, "learning_rate": 0.0005, "loss": 1.6784381866455078, "step": 38161 }, { "epoch": 3.008435159637367, "grad_norm": 0.4502151906490326, "learning_rate": 0.0005, "loss": 1.243877649307251, "step": 38162 }, { "epoch": 3.008513992905006, "grad_norm": 0.5660851001739502, "learning_rate": 0.0005, "loss": 1.417262077331543, "step": 38163 }, { "epoch": 3.0085928261726447, "grad_norm": 0.45285290479660034, "learning_rate": 0.0005, "loss": 1.1820570230484009, "step": 38164 }, { "epoch": 3.008671659440284, "grad_norm": 0.4214780032634735, "learning_rate": 0.0005, "loss": 1.2335524559020996, "step": 38165 }, { "epoch": 3.008750492707923, "grad_norm": 0.5015681982040405, "learning_rate": 0.0005, "loss": 1.2986317873001099, "step": 38166 }, { "epoch": 3.0088293259755616, "grad_norm": 0.4868106544017792, "learning_rate": 0.0005, "loss": 1.4409306049346924, "step": 38167 }, { "epoch": 3.0089081592432008, "grad_norm": 0.5017085671424866, "learning_rate": 0.0005, "loss": 1.1370540857315063, "step": 38168 }, { "epoch": 3.0089869925108395, "grad_norm": 0.4019657373428345, "learning_rate": 0.0005, "loss": 1.1491456031799316, "step": 38169 }, { "epoch": 3.0090658257784786, "grad_norm": 0.4305318295955658, "learning_rate": 0.0005, "loss": 1.4677863121032715, "step": 38170 }, { "epoch": 3.0091446590461173, "grad_norm": 0.38944026827812195, "learning_rate": 0.0005, "loss": 1.2319585084915161, "step": 38171 }, { "epoch": 3.0092234923137564, "grad_norm": 0.44393390417099, "learning_rate": 0.0005, "loss": 1.1600401401519775, "step": 38172 }, { "epoch": 3.0093023255813955, "grad_norm": 0.45180484652519226, "learning_rate": 0.0005, "loss": 1.3556885719299316, "step": 38173 }, { "epoch": 3.0093811588490342, "grad_norm": 0.41093552112579346, "learning_rate": 0.0005, "loss": 1.139050006866455, "step": 38174 }, { "epoch": 3.0094599921166734, "grad_norm": 0.44811058044433594, "learning_rate": 0.0005, "loss": 1.1756060123443604, "step": 38175 }, { "epoch": 3.009538825384312, "grad_norm": 0.4822646975517273, "learning_rate": 0.0005, "loss": 1.2615554332733154, "step": 38176 }, { "epoch": 3.009617658651951, "grad_norm": 0.43995895981788635, "learning_rate": 0.0005, "loss": 1.3928418159484863, "step": 38177 }, { "epoch": 3.00969649191959, "grad_norm": 0.5114942789077759, "learning_rate": 0.0005, "loss": 1.4275672435760498, "step": 38178 }, { "epoch": 3.009775325187229, "grad_norm": 0.5504796504974365, "learning_rate": 0.0005, "loss": 1.3104667663574219, "step": 38179 }, { "epoch": 3.009854158454868, "grad_norm": 0.48086079955101013, "learning_rate": 0.0005, "loss": 1.2510725259780884, "step": 38180 }, { "epoch": 3.009932991722507, "grad_norm": 0.4112483859062195, "learning_rate": 0.0005, "loss": 1.1645464897155762, "step": 38181 }, { "epoch": 3.010011824990146, "grad_norm": 0.39902153611183167, "learning_rate": 0.0005, "loss": 1.2108954191207886, "step": 38182 }, { "epoch": 3.0100906582577847, "grad_norm": 0.49620404839515686, "learning_rate": 0.0005, "loss": 1.098551630973816, "step": 38183 }, { "epoch": 3.010169491525424, "grad_norm": 0.6061134338378906, "learning_rate": 0.0005, "loss": 1.2980005741119385, "step": 38184 }, { "epoch": 3.0102483247930625, "grad_norm": 0.4414917230606079, "learning_rate": 0.0005, "loss": 1.187915563583374, "step": 38185 }, { "epoch": 3.0103271580607016, "grad_norm": 0.5502263307571411, "learning_rate": 0.0005, "loss": 1.351831078529358, "step": 38186 }, { "epoch": 3.0104059913283407, "grad_norm": 0.4193652868270874, "learning_rate": 0.0005, "loss": 1.2424925565719604, "step": 38187 }, { "epoch": 3.0104848245959794, "grad_norm": 0.540643572807312, "learning_rate": 0.0005, "loss": 1.1840327978134155, "step": 38188 }, { "epoch": 3.0105636578636186, "grad_norm": 0.44257691502571106, "learning_rate": 0.0005, "loss": 1.169384479522705, "step": 38189 }, { "epoch": 3.0106424911312573, "grad_norm": 0.47688987851142883, "learning_rate": 0.0005, "loss": 1.2788439989089966, "step": 38190 }, { "epoch": 3.0107213243988964, "grad_norm": 0.5903122425079346, "learning_rate": 0.0005, "loss": 1.3338936567306519, "step": 38191 }, { "epoch": 3.010800157666535, "grad_norm": 0.38271716237068176, "learning_rate": 0.0005, "loss": 1.139522671699524, "step": 38192 }, { "epoch": 3.010878990934174, "grad_norm": 0.49258342385292053, "learning_rate": 0.0005, "loss": 1.2959896326065063, "step": 38193 }, { "epoch": 3.0109578242018133, "grad_norm": 0.482013076543808, "learning_rate": 0.0005, "loss": 1.381333351135254, "step": 38194 }, { "epoch": 3.011036657469452, "grad_norm": 0.46467724442481995, "learning_rate": 0.0005, "loss": 1.3250694274902344, "step": 38195 }, { "epoch": 3.011115490737091, "grad_norm": 0.39871957898139954, "learning_rate": 0.0005, "loss": 1.1335992813110352, "step": 38196 }, { "epoch": 3.01119432400473, "grad_norm": 0.5181615352630615, "learning_rate": 0.0005, "loss": 1.3516697883605957, "step": 38197 }, { "epoch": 3.011273157272369, "grad_norm": 0.6804710626602173, "learning_rate": 0.0005, "loss": 1.368577003479004, "step": 38198 }, { "epoch": 3.0113519905400077, "grad_norm": 0.44214144349098206, "learning_rate": 0.0005, "loss": 1.3333559036254883, "step": 38199 }, { "epoch": 3.011430823807647, "grad_norm": 0.502357542514801, "learning_rate": 0.0005, "loss": 1.3431901931762695, "step": 38200 }, { "epoch": 3.011509657075286, "grad_norm": 0.5031762719154358, "learning_rate": 0.0005, "loss": 1.4442061185836792, "step": 38201 }, { "epoch": 3.0115884903429246, "grad_norm": 0.5662708282470703, "learning_rate": 0.0005, "loss": 1.2124180793762207, "step": 38202 }, { "epoch": 3.0116673236105638, "grad_norm": 0.572605311870575, "learning_rate": 0.0005, "loss": 1.3531439304351807, "step": 38203 }, { "epoch": 3.0117461568782025, "grad_norm": 0.44765791296958923, "learning_rate": 0.0005, "loss": 1.2976593971252441, "step": 38204 }, { "epoch": 3.0118249901458416, "grad_norm": 0.4315425455570221, "learning_rate": 0.0005, "loss": 1.2504966259002686, "step": 38205 }, { "epoch": 3.0119038234134803, "grad_norm": 0.5089066028594971, "learning_rate": 0.0005, "loss": 1.3848463296890259, "step": 38206 }, { "epoch": 3.0119826566811194, "grad_norm": 0.48847725987434387, "learning_rate": 0.0005, "loss": 1.3347175121307373, "step": 38207 }, { "epoch": 3.0120614899487586, "grad_norm": 0.4146876335144043, "learning_rate": 0.0005, "loss": 1.2940387725830078, "step": 38208 }, { "epoch": 3.0121403232163972, "grad_norm": 0.41872990131378174, "learning_rate": 0.0005, "loss": 1.1065473556518555, "step": 38209 }, { "epoch": 3.0122191564840364, "grad_norm": 0.4050520956516266, "learning_rate": 0.0005, "loss": 1.2626583576202393, "step": 38210 }, { "epoch": 3.012297989751675, "grad_norm": 0.4898287057876587, "learning_rate": 0.0005, "loss": 1.2863688468933105, "step": 38211 }, { "epoch": 3.012376823019314, "grad_norm": 0.4733835756778717, "learning_rate": 0.0005, "loss": 1.2848572731018066, "step": 38212 }, { "epoch": 3.012455656286953, "grad_norm": 0.3972705602645874, "learning_rate": 0.0005, "loss": 0.9962921142578125, "step": 38213 }, { "epoch": 3.012534489554592, "grad_norm": 1.8246774673461914, "learning_rate": 0.0005, "loss": 1.3490631580352783, "step": 38214 }, { "epoch": 3.012613322822231, "grad_norm": 0.42455020546913147, "learning_rate": 0.0005, "loss": 1.230210781097412, "step": 38215 }, { "epoch": 3.01269215608987, "grad_norm": 0.40404757857322693, "learning_rate": 0.0005, "loss": 1.291137933731079, "step": 38216 }, { "epoch": 3.012770989357509, "grad_norm": 0.4067671000957489, "learning_rate": 0.0005, "loss": 1.1454797983169556, "step": 38217 }, { "epoch": 3.0128498226251477, "grad_norm": 0.5472249388694763, "learning_rate": 0.0005, "loss": 1.4536209106445312, "step": 38218 }, { "epoch": 3.012928655892787, "grad_norm": 0.5197923183441162, "learning_rate": 0.0005, "loss": 1.313840627670288, "step": 38219 }, { "epoch": 3.0130074891604255, "grad_norm": 0.5345713496208191, "learning_rate": 0.0005, "loss": 1.5103785991668701, "step": 38220 }, { "epoch": 3.0130863224280646, "grad_norm": 0.4335557520389557, "learning_rate": 0.0005, "loss": 1.3329330682754517, "step": 38221 }, { "epoch": 3.0131651556957038, "grad_norm": 0.4096616804599762, "learning_rate": 0.0005, "loss": 1.2467100620269775, "step": 38222 }, { "epoch": 3.0132439889633424, "grad_norm": 0.5683577656745911, "learning_rate": 0.0005, "loss": 1.2319104671478271, "step": 38223 }, { "epoch": 3.0133228222309816, "grad_norm": 0.5945752859115601, "learning_rate": 0.0005, "loss": 1.4430060386657715, "step": 38224 }, { "epoch": 3.0134016554986203, "grad_norm": 0.445819228887558, "learning_rate": 0.0005, "loss": 1.1117141246795654, "step": 38225 }, { "epoch": 3.0134804887662594, "grad_norm": 0.5176886320114136, "learning_rate": 0.0005, "loss": 1.2242379188537598, "step": 38226 }, { "epoch": 3.013559322033898, "grad_norm": 0.4942239820957184, "learning_rate": 0.0005, "loss": 1.3400686979293823, "step": 38227 }, { "epoch": 3.0136381553015372, "grad_norm": 0.5311552882194519, "learning_rate": 0.0005, "loss": 1.2527215480804443, "step": 38228 }, { "epoch": 3.0137169885691764, "grad_norm": 0.545792818069458, "learning_rate": 0.0005, "loss": 1.5175986289978027, "step": 38229 }, { "epoch": 3.013795821836815, "grad_norm": 0.4557572305202484, "learning_rate": 0.0005, "loss": 1.158548355102539, "step": 38230 }, { "epoch": 3.013874655104454, "grad_norm": 0.5441306829452515, "learning_rate": 0.0005, "loss": 1.2018179893493652, "step": 38231 }, { "epoch": 3.013953488372093, "grad_norm": 0.5488260388374329, "learning_rate": 0.0005, "loss": 1.3744467496871948, "step": 38232 }, { "epoch": 3.014032321639732, "grad_norm": 0.4951225519180298, "learning_rate": 0.0005, "loss": 1.390800952911377, "step": 38233 }, { "epoch": 3.0141111549073707, "grad_norm": 0.3976086676120758, "learning_rate": 0.0005, "loss": 1.1682658195495605, "step": 38234 }, { "epoch": 3.01418998817501, "grad_norm": 0.5303226113319397, "learning_rate": 0.0005, "loss": 1.3582172393798828, "step": 38235 }, { "epoch": 3.014268821442649, "grad_norm": 0.4955292344093323, "learning_rate": 0.0005, "loss": 1.190610408782959, "step": 38236 }, { "epoch": 3.0143476547102876, "grad_norm": 0.43837258219718933, "learning_rate": 0.0005, "loss": 1.3297157287597656, "step": 38237 }, { "epoch": 3.014426487977927, "grad_norm": 0.40766558051109314, "learning_rate": 0.0005, "loss": 1.1485620737075806, "step": 38238 }, { "epoch": 3.0145053212455655, "grad_norm": 0.6218746304512024, "learning_rate": 0.0005, "loss": 1.301774501800537, "step": 38239 }, { "epoch": 3.0145841545132046, "grad_norm": 0.5888381004333496, "learning_rate": 0.0005, "loss": 1.3302143812179565, "step": 38240 }, { "epoch": 3.0146629877808433, "grad_norm": 0.4645770192146301, "learning_rate": 0.0005, "loss": 1.3304250240325928, "step": 38241 }, { "epoch": 3.0147418210484824, "grad_norm": 0.4626467227935791, "learning_rate": 0.0005, "loss": 1.1649764776229858, "step": 38242 }, { "epoch": 3.0148206543161216, "grad_norm": 0.45602941513061523, "learning_rate": 0.0005, "loss": 1.2055575847625732, "step": 38243 }, { "epoch": 3.0148994875837603, "grad_norm": 0.4492424428462982, "learning_rate": 0.0005, "loss": 1.3454880714416504, "step": 38244 }, { "epoch": 3.0149783208513994, "grad_norm": 0.5072872638702393, "learning_rate": 0.0005, "loss": 1.2665977478027344, "step": 38245 }, { "epoch": 3.015057154119038, "grad_norm": 0.4629589319229126, "learning_rate": 0.0005, "loss": 1.1089116334915161, "step": 38246 }, { "epoch": 3.015135987386677, "grad_norm": 0.45761168003082275, "learning_rate": 0.0005, "loss": 1.2917115688323975, "step": 38247 }, { "epoch": 3.0152148206543163, "grad_norm": 0.3989417254924774, "learning_rate": 0.0005, "loss": 1.162611961364746, "step": 38248 }, { "epoch": 3.015293653921955, "grad_norm": 0.4808862805366516, "learning_rate": 0.0005, "loss": 1.1757479906082153, "step": 38249 }, { "epoch": 3.015372487189594, "grad_norm": 0.5148239135742188, "learning_rate": 0.0005, "loss": 1.3620960712432861, "step": 38250 }, { "epoch": 3.015451320457233, "grad_norm": 0.4691721796989441, "learning_rate": 0.0005, "loss": 1.297074317932129, "step": 38251 }, { "epoch": 3.015530153724872, "grad_norm": 0.4947117567062378, "learning_rate": 0.0005, "loss": 1.2751264572143555, "step": 38252 }, { "epoch": 3.0156089869925107, "grad_norm": 0.6924787759780884, "learning_rate": 0.0005, "loss": 1.2478547096252441, "step": 38253 }, { "epoch": 3.01568782026015, "grad_norm": 0.44089436531066895, "learning_rate": 0.0005, "loss": 1.1184231042861938, "step": 38254 }, { "epoch": 3.015766653527789, "grad_norm": 0.5149108171463013, "learning_rate": 0.0005, "loss": 1.2118878364562988, "step": 38255 }, { "epoch": 3.0158454867954276, "grad_norm": 0.6390887498855591, "learning_rate": 0.0005, "loss": 1.2663006782531738, "step": 38256 }, { "epoch": 3.0159243200630668, "grad_norm": 0.49891549348831177, "learning_rate": 0.0005, "loss": 1.2182502746582031, "step": 38257 }, { "epoch": 3.0160031533307055, "grad_norm": 0.509588360786438, "learning_rate": 0.0005, "loss": 1.3451006412506104, "step": 38258 }, { "epoch": 3.0160819865983446, "grad_norm": 0.4378378093242645, "learning_rate": 0.0005, "loss": 1.1690553426742554, "step": 38259 }, { "epoch": 3.0161608198659833, "grad_norm": 0.471111536026001, "learning_rate": 0.0005, "loss": 1.2543845176696777, "step": 38260 }, { "epoch": 3.0162396531336224, "grad_norm": 0.4932173788547516, "learning_rate": 0.0005, "loss": 1.1759238243103027, "step": 38261 }, { "epoch": 3.0163184864012615, "grad_norm": 0.4781510531902313, "learning_rate": 0.0005, "loss": 1.3131489753723145, "step": 38262 }, { "epoch": 3.0163973196689002, "grad_norm": 0.5019590854644775, "learning_rate": 0.0005, "loss": 1.3094807863235474, "step": 38263 }, { "epoch": 3.0164761529365394, "grad_norm": 0.582025945186615, "learning_rate": 0.0005, "loss": 1.2413510084152222, "step": 38264 }, { "epoch": 3.016554986204178, "grad_norm": 0.48648974299430847, "learning_rate": 0.0005, "loss": 1.19645094871521, "step": 38265 }, { "epoch": 3.016633819471817, "grad_norm": 0.43070873618125916, "learning_rate": 0.0005, "loss": 1.3371944427490234, "step": 38266 }, { "epoch": 3.016712652739456, "grad_norm": 0.4332801103591919, "learning_rate": 0.0005, "loss": 1.2367385625839233, "step": 38267 }, { "epoch": 3.016791486007095, "grad_norm": 0.6078312993049622, "learning_rate": 0.0005, "loss": 1.237451195716858, "step": 38268 }, { "epoch": 3.016870319274734, "grad_norm": 0.4014740288257599, "learning_rate": 0.0005, "loss": 1.3070467710494995, "step": 38269 }, { "epoch": 3.016949152542373, "grad_norm": 0.43898093700408936, "learning_rate": 0.0005, "loss": 1.1690294742584229, "step": 38270 }, { "epoch": 3.017027985810012, "grad_norm": 0.5019550323486328, "learning_rate": 0.0005, "loss": 1.3017735481262207, "step": 38271 }, { "epoch": 3.0171068190776507, "grad_norm": 0.4742337763309479, "learning_rate": 0.0005, "loss": 1.107816457748413, "step": 38272 }, { "epoch": 3.01718565234529, "grad_norm": 0.4285162389278412, "learning_rate": 0.0005, "loss": 1.1507681608200073, "step": 38273 }, { "epoch": 3.0172644856129285, "grad_norm": 0.522327184677124, "learning_rate": 0.0005, "loss": 1.6548898220062256, "step": 38274 }, { "epoch": 3.0173433188805676, "grad_norm": 0.36846038699150085, "learning_rate": 0.0005, "loss": 1.0716911554336548, "step": 38275 }, { "epoch": 3.0174221521482067, "grad_norm": 0.5317521691322327, "learning_rate": 0.0005, "loss": 1.2740306854248047, "step": 38276 }, { "epoch": 3.0175009854158454, "grad_norm": 0.4972301125526428, "learning_rate": 0.0005, "loss": 1.2648441791534424, "step": 38277 }, { "epoch": 3.0175798186834846, "grad_norm": 0.45332983136177063, "learning_rate": 0.0005, "loss": 1.1198723316192627, "step": 38278 }, { "epoch": 3.0176586519511233, "grad_norm": 0.4371452033519745, "learning_rate": 0.0005, "loss": 1.2391066551208496, "step": 38279 }, { "epoch": 3.0177374852187624, "grad_norm": 0.5640547275543213, "learning_rate": 0.0005, "loss": 1.2589752674102783, "step": 38280 }, { "epoch": 3.017816318486401, "grad_norm": 0.518803060054779, "learning_rate": 0.0005, "loss": 1.3959801197052002, "step": 38281 }, { "epoch": 3.01789515175404, "grad_norm": 0.4751669466495514, "learning_rate": 0.0005, "loss": 1.2289958000183105, "step": 38282 }, { "epoch": 3.0179739850216794, "grad_norm": 0.4694336950778961, "learning_rate": 0.0005, "loss": 1.294443964958191, "step": 38283 }, { "epoch": 3.018052818289318, "grad_norm": 0.42322874069213867, "learning_rate": 0.0005, "loss": 1.2472498416900635, "step": 38284 }, { "epoch": 3.018131651556957, "grad_norm": 0.4426540434360504, "learning_rate": 0.0005, "loss": 1.1961874961853027, "step": 38285 }, { "epoch": 3.018210484824596, "grad_norm": 0.46686986088752747, "learning_rate": 0.0005, "loss": 1.434823751449585, "step": 38286 }, { "epoch": 3.018289318092235, "grad_norm": 0.44832924008369446, "learning_rate": 0.0005, "loss": 1.0486608743667603, "step": 38287 }, { "epoch": 3.0183681513598737, "grad_norm": 0.3802259862422943, "learning_rate": 0.0005, "loss": 1.0411051511764526, "step": 38288 }, { "epoch": 3.018446984627513, "grad_norm": 0.661003589630127, "learning_rate": 0.0005, "loss": 1.3096450567245483, "step": 38289 }, { "epoch": 3.018525817895152, "grad_norm": 0.5450421571731567, "learning_rate": 0.0005, "loss": 1.410789966583252, "step": 38290 }, { "epoch": 3.0186046511627906, "grad_norm": 0.436237633228302, "learning_rate": 0.0005, "loss": 1.2737178802490234, "step": 38291 }, { "epoch": 3.0186834844304298, "grad_norm": 0.49677303433418274, "learning_rate": 0.0005, "loss": 1.1401207447052002, "step": 38292 }, { "epoch": 3.0187623176980685, "grad_norm": 0.4839051067829132, "learning_rate": 0.0005, "loss": 1.3523595333099365, "step": 38293 }, { "epoch": 3.0188411509657076, "grad_norm": 0.518366813659668, "learning_rate": 0.0005, "loss": 1.319021463394165, "step": 38294 }, { "epoch": 3.0189199842333463, "grad_norm": 0.4044373631477356, "learning_rate": 0.0005, "loss": 1.1244359016418457, "step": 38295 }, { "epoch": 3.0189988175009854, "grad_norm": 0.4994047284126282, "learning_rate": 0.0005, "loss": 1.2974016666412354, "step": 38296 }, { "epoch": 3.0190776507686246, "grad_norm": 0.49882757663726807, "learning_rate": 0.0005, "loss": 1.312508463859558, "step": 38297 }, { "epoch": 3.0191564840362632, "grad_norm": 0.40521883964538574, "learning_rate": 0.0005, "loss": 1.1017119884490967, "step": 38298 }, { "epoch": 3.0192353173039024, "grad_norm": 0.5087136030197144, "learning_rate": 0.0005, "loss": 1.3850507736206055, "step": 38299 }, { "epoch": 3.019314150571541, "grad_norm": 0.47428587079048157, "learning_rate": 0.0005, "loss": 1.147186040878296, "step": 38300 }, { "epoch": 3.01939298383918, "grad_norm": 0.5523062944412231, "learning_rate": 0.0005, "loss": 1.262796401977539, "step": 38301 }, { "epoch": 3.019471817106819, "grad_norm": 0.40394893288612366, "learning_rate": 0.0005, "loss": 1.1270203590393066, "step": 38302 }, { "epoch": 3.019550650374458, "grad_norm": 0.5735059976577759, "learning_rate": 0.0005, "loss": 1.3064048290252686, "step": 38303 }, { "epoch": 3.019629483642097, "grad_norm": 0.471655935049057, "learning_rate": 0.0005, "loss": 1.2512280941009521, "step": 38304 }, { "epoch": 3.019708316909736, "grad_norm": 0.5052607655525208, "learning_rate": 0.0005, "loss": 1.2199430465698242, "step": 38305 }, { "epoch": 3.019787150177375, "grad_norm": 0.4820157289505005, "learning_rate": 0.0005, "loss": 1.2601814270019531, "step": 38306 }, { "epoch": 3.0198659834450137, "grad_norm": 0.5772440433502197, "learning_rate": 0.0005, "loss": 1.1804438829421997, "step": 38307 }, { "epoch": 3.019944816712653, "grad_norm": 0.592475175857544, "learning_rate": 0.0005, "loss": 1.360892415046692, "step": 38308 }, { "epoch": 3.0200236499802915, "grad_norm": 0.40539446473121643, "learning_rate": 0.0005, "loss": 1.235708475112915, "step": 38309 }, { "epoch": 3.0201024832479306, "grad_norm": 0.5090306997299194, "learning_rate": 0.0005, "loss": 1.2596428394317627, "step": 38310 }, { "epoch": 3.0201813165155698, "grad_norm": 0.4824944734573364, "learning_rate": 0.0005, "loss": 1.143846035003662, "step": 38311 }, { "epoch": 3.0202601497832084, "grad_norm": 0.4366331994533539, "learning_rate": 0.0005, "loss": 1.0685632228851318, "step": 38312 }, { "epoch": 3.0203389830508476, "grad_norm": 0.4366435408592224, "learning_rate": 0.0005, "loss": 1.1889539957046509, "step": 38313 }, { "epoch": 3.0204178163184863, "grad_norm": 0.4667447805404663, "learning_rate": 0.0005, "loss": 1.3606538772583008, "step": 38314 }, { "epoch": 3.0204966495861254, "grad_norm": 0.5049741268157959, "learning_rate": 0.0005, "loss": 1.2597309350967407, "step": 38315 }, { "epoch": 3.020575482853764, "grad_norm": 0.4163329005241394, "learning_rate": 0.0005, "loss": 1.1794583797454834, "step": 38316 }, { "epoch": 3.0206543161214032, "grad_norm": 0.4783265292644501, "learning_rate": 0.0005, "loss": 1.105346918106079, "step": 38317 }, { "epoch": 3.0207331493890424, "grad_norm": 0.4775916337966919, "learning_rate": 0.0005, "loss": 1.2202719449996948, "step": 38318 }, { "epoch": 3.020811982656681, "grad_norm": 0.4442775845527649, "learning_rate": 0.0005, "loss": 1.1805939674377441, "step": 38319 }, { "epoch": 3.02089081592432, "grad_norm": 0.6615048050880432, "learning_rate": 0.0005, "loss": 1.30265474319458, "step": 38320 }, { "epoch": 3.020969649191959, "grad_norm": 0.560406506061554, "learning_rate": 0.0005, "loss": 1.0772126913070679, "step": 38321 }, { "epoch": 3.021048482459598, "grad_norm": 0.46900323033332825, "learning_rate": 0.0005, "loss": 1.2326953411102295, "step": 38322 }, { "epoch": 3.0211273157272367, "grad_norm": 0.46131813526153564, "learning_rate": 0.0005, "loss": 1.124667763710022, "step": 38323 }, { "epoch": 3.021206148994876, "grad_norm": 0.40204504132270813, "learning_rate": 0.0005, "loss": 1.0448758602142334, "step": 38324 }, { "epoch": 3.021284982262515, "grad_norm": 0.4281640946865082, "learning_rate": 0.0005, "loss": 1.1352074146270752, "step": 38325 }, { "epoch": 3.0213638155301537, "grad_norm": 0.4763382375240326, "learning_rate": 0.0005, "loss": 1.413292646408081, "step": 38326 }, { "epoch": 3.021442648797793, "grad_norm": 0.4497564136981964, "learning_rate": 0.0005, "loss": 1.1206766366958618, "step": 38327 }, { "epoch": 3.0215214820654315, "grad_norm": 0.4780956506729126, "learning_rate": 0.0005, "loss": 1.3593679666519165, "step": 38328 }, { "epoch": 3.0216003153330706, "grad_norm": 0.41009122133255005, "learning_rate": 0.0005, "loss": 1.250837802886963, "step": 38329 }, { "epoch": 3.0216791486007093, "grad_norm": 0.4778228998184204, "learning_rate": 0.0005, "loss": 1.3344306945800781, "step": 38330 }, { "epoch": 3.0217579818683484, "grad_norm": 0.5419233441352844, "learning_rate": 0.0005, "loss": 1.1147232055664062, "step": 38331 }, { "epoch": 3.0218368151359876, "grad_norm": 0.45934998989105225, "learning_rate": 0.0005, "loss": 1.0978448390960693, "step": 38332 }, { "epoch": 3.0219156484036263, "grad_norm": 0.4794485569000244, "learning_rate": 0.0005, "loss": 1.282004952430725, "step": 38333 }, { "epoch": 3.0219944816712654, "grad_norm": 0.4403938353061676, "learning_rate": 0.0005, "loss": 1.3945930004119873, "step": 38334 }, { "epoch": 3.022073314938904, "grad_norm": 0.37460070848464966, "learning_rate": 0.0005, "loss": 1.27164626121521, "step": 38335 }, { "epoch": 3.022152148206543, "grad_norm": 0.3817312717437744, "learning_rate": 0.0005, "loss": 1.066949486732483, "step": 38336 }, { "epoch": 3.022230981474182, "grad_norm": 0.40745604038238525, "learning_rate": 0.0005, "loss": 1.2115681171417236, "step": 38337 }, { "epoch": 3.022309814741821, "grad_norm": 0.45202139019966125, "learning_rate": 0.0005, "loss": 1.1412100791931152, "step": 38338 }, { "epoch": 3.02238864800946, "grad_norm": 0.4738730192184448, "learning_rate": 0.0005, "loss": 1.3127455711364746, "step": 38339 }, { "epoch": 3.022467481277099, "grad_norm": 0.5701227188110352, "learning_rate": 0.0005, "loss": 1.3321702480316162, "step": 38340 }, { "epoch": 3.022546314544738, "grad_norm": 0.4472436010837555, "learning_rate": 0.0005, "loss": 1.22727370262146, "step": 38341 }, { "epoch": 3.0226251478123767, "grad_norm": 0.5484700798988342, "learning_rate": 0.0005, "loss": 1.3215610980987549, "step": 38342 }, { "epoch": 3.022703981080016, "grad_norm": 0.6309471726417542, "learning_rate": 0.0005, "loss": 1.3792510032653809, "step": 38343 }, { "epoch": 3.0227828143476545, "grad_norm": 0.46338048577308655, "learning_rate": 0.0005, "loss": 1.1917167901992798, "step": 38344 }, { "epoch": 3.0228616476152936, "grad_norm": 0.44813957810401917, "learning_rate": 0.0005, "loss": 1.3668906688690186, "step": 38345 }, { "epoch": 3.0229404808829328, "grad_norm": 0.3996381461620331, "learning_rate": 0.0005, "loss": 0.9880755543708801, "step": 38346 }, { "epoch": 3.0230193141505715, "grad_norm": 0.5149873495101929, "learning_rate": 0.0005, "loss": 1.285815954208374, "step": 38347 }, { "epoch": 3.0230981474182106, "grad_norm": 0.5071923732757568, "learning_rate": 0.0005, "loss": 1.2923765182495117, "step": 38348 }, { "epoch": 3.0231769806858493, "grad_norm": 0.4432812035083771, "learning_rate": 0.0005, "loss": 1.157562255859375, "step": 38349 }, { "epoch": 3.0232558139534884, "grad_norm": 0.44934579730033875, "learning_rate": 0.0005, "loss": 1.2466461658477783, "step": 38350 }, { "epoch": 3.023334647221127, "grad_norm": 0.4689502418041229, "learning_rate": 0.0005, "loss": 1.1524059772491455, "step": 38351 }, { "epoch": 3.0234134804887662, "grad_norm": 0.43167436122894287, "learning_rate": 0.0005, "loss": 1.0379302501678467, "step": 38352 }, { "epoch": 3.0234923137564054, "grad_norm": 0.45786386728286743, "learning_rate": 0.0005, "loss": 1.1952338218688965, "step": 38353 }, { "epoch": 3.023571147024044, "grad_norm": 0.42524033784866333, "learning_rate": 0.0005, "loss": 1.2498693466186523, "step": 38354 }, { "epoch": 3.023649980291683, "grad_norm": 0.4832327365875244, "learning_rate": 0.0005, "loss": 1.2607810497283936, "step": 38355 }, { "epoch": 3.023728813559322, "grad_norm": 0.5198434591293335, "learning_rate": 0.0005, "loss": 1.241790771484375, "step": 38356 }, { "epoch": 3.023807646826961, "grad_norm": 0.44083157181739807, "learning_rate": 0.0005, "loss": 1.2268964052200317, "step": 38357 }, { "epoch": 3.0238864800945997, "grad_norm": 0.470303475856781, "learning_rate": 0.0005, "loss": 1.1240880489349365, "step": 38358 }, { "epoch": 3.023965313362239, "grad_norm": 0.5390090942382812, "learning_rate": 0.0005, "loss": 1.283587098121643, "step": 38359 }, { "epoch": 3.024044146629878, "grad_norm": 0.6337927579879761, "learning_rate": 0.0005, "loss": 1.4094369411468506, "step": 38360 }, { "epoch": 3.0241229798975167, "grad_norm": 0.5104807019233704, "learning_rate": 0.0005, "loss": 1.3309383392333984, "step": 38361 }, { "epoch": 3.024201813165156, "grad_norm": 0.4715029299259186, "learning_rate": 0.0005, "loss": 1.4161067008972168, "step": 38362 }, { "epoch": 3.0242806464327945, "grad_norm": 0.5499156713485718, "learning_rate": 0.0005, "loss": 1.2300302982330322, "step": 38363 }, { "epoch": 3.0243594797004336, "grad_norm": 0.4604929983615875, "learning_rate": 0.0005, "loss": 1.26425302028656, "step": 38364 }, { "epoch": 3.0244383129680723, "grad_norm": 0.4145086109638214, "learning_rate": 0.0005, "loss": 1.2807364463806152, "step": 38365 }, { "epoch": 3.0245171462357114, "grad_norm": 0.47707387804985046, "learning_rate": 0.0005, "loss": 1.1650646924972534, "step": 38366 }, { "epoch": 3.0245959795033506, "grad_norm": 0.4523164629936218, "learning_rate": 0.0005, "loss": 1.1977052688598633, "step": 38367 }, { "epoch": 3.0246748127709893, "grad_norm": 0.5159420967102051, "learning_rate": 0.0005, "loss": 1.206111192703247, "step": 38368 }, { "epoch": 3.0247536460386284, "grad_norm": 0.5662567019462585, "learning_rate": 0.0005, "loss": 1.1896579265594482, "step": 38369 }, { "epoch": 3.024832479306267, "grad_norm": 0.4942394495010376, "learning_rate": 0.0005, "loss": 1.4234932661056519, "step": 38370 }, { "epoch": 3.024911312573906, "grad_norm": 0.5108106732368469, "learning_rate": 0.0005, "loss": 1.2314801216125488, "step": 38371 }, { "epoch": 3.024990145841545, "grad_norm": 0.3832027316093445, "learning_rate": 0.0005, "loss": 1.1088027954101562, "step": 38372 }, { "epoch": 3.025068979109184, "grad_norm": 0.4768335819244385, "learning_rate": 0.0005, "loss": 1.2253444194793701, "step": 38373 }, { "epoch": 3.025147812376823, "grad_norm": 0.4887617230415344, "learning_rate": 0.0005, "loss": 1.2462186813354492, "step": 38374 }, { "epoch": 3.025226645644462, "grad_norm": 0.5538120269775391, "learning_rate": 0.0005, "loss": 1.2607042789459229, "step": 38375 }, { "epoch": 3.025305478912101, "grad_norm": 0.4392106533050537, "learning_rate": 0.0005, "loss": 1.2363567352294922, "step": 38376 }, { "epoch": 3.0253843121797397, "grad_norm": 0.48508909344673157, "learning_rate": 0.0005, "loss": 1.5760859251022339, "step": 38377 }, { "epoch": 3.025463145447379, "grad_norm": 0.42897072434425354, "learning_rate": 0.0005, "loss": 1.3226338624954224, "step": 38378 }, { "epoch": 3.025541978715018, "grad_norm": 0.47861963510513306, "learning_rate": 0.0005, "loss": 1.0856997966766357, "step": 38379 }, { "epoch": 3.0256208119826566, "grad_norm": 0.4495581090450287, "learning_rate": 0.0005, "loss": 1.1926052570343018, "step": 38380 }, { "epoch": 3.0256996452502958, "grad_norm": 0.47633859515190125, "learning_rate": 0.0005, "loss": 1.2257041931152344, "step": 38381 }, { "epoch": 3.0257784785179345, "grad_norm": 0.5274145603179932, "learning_rate": 0.0005, "loss": 1.369706630706787, "step": 38382 }, { "epoch": 3.0258573117855736, "grad_norm": 0.48848432302474976, "learning_rate": 0.0005, "loss": 1.1840999126434326, "step": 38383 }, { "epoch": 3.0259361450532123, "grad_norm": 0.5115181803703308, "learning_rate": 0.0005, "loss": 1.5826044082641602, "step": 38384 }, { "epoch": 3.0260149783208514, "grad_norm": 0.4144842028617859, "learning_rate": 0.0005, "loss": 1.0981794595718384, "step": 38385 }, { "epoch": 3.0260938115884906, "grad_norm": 0.4602327346801758, "learning_rate": 0.0005, "loss": 1.0571794509887695, "step": 38386 }, { "epoch": 3.0261726448561292, "grad_norm": 0.40235599875450134, "learning_rate": 0.0005, "loss": 1.1091094017028809, "step": 38387 }, { "epoch": 3.0262514781237684, "grad_norm": 0.5784031748771667, "learning_rate": 0.0005, "loss": 1.4357106685638428, "step": 38388 }, { "epoch": 3.026330311391407, "grad_norm": 0.4061158299446106, "learning_rate": 0.0005, "loss": 1.1783620119094849, "step": 38389 }, { "epoch": 3.026409144659046, "grad_norm": 0.4870564043521881, "learning_rate": 0.0005, "loss": 1.3614246845245361, "step": 38390 }, { "epoch": 3.026487977926685, "grad_norm": 0.637012779712677, "learning_rate": 0.0005, "loss": 1.4402635097503662, "step": 38391 }, { "epoch": 3.026566811194324, "grad_norm": 0.5685805082321167, "learning_rate": 0.0005, "loss": 1.3121881484985352, "step": 38392 }, { "epoch": 3.026645644461963, "grad_norm": 0.5241433382034302, "learning_rate": 0.0005, "loss": 1.3473165035247803, "step": 38393 }, { "epoch": 3.026724477729602, "grad_norm": 0.46732237935066223, "learning_rate": 0.0005, "loss": 1.1112465858459473, "step": 38394 }, { "epoch": 3.026803310997241, "grad_norm": 0.6427475810050964, "learning_rate": 0.0005, "loss": 1.484946370124817, "step": 38395 }, { "epoch": 3.0268821442648797, "grad_norm": 0.6129369735717773, "learning_rate": 0.0005, "loss": 1.2756783962249756, "step": 38396 }, { "epoch": 3.026960977532519, "grad_norm": 0.4473821818828583, "learning_rate": 0.0005, "loss": 1.159879207611084, "step": 38397 }, { "epoch": 3.0270398108001575, "grad_norm": 0.467733770608902, "learning_rate": 0.0005, "loss": 1.2535208463668823, "step": 38398 }, { "epoch": 3.0271186440677966, "grad_norm": 0.5137246251106262, "learning_rate": 0.0005, "loss": 1.1731926202774048, "step": 38399 }, { "epoch": 3.0271974773354358, "grad_norm": 0.5995512008666992, "learning_rate": 0.0005, "loss": 1.4223897457122803, "step": 38400 }, { "epoch": 3.0272763106030744, "grad_norm": 0.4759600758552551, "learning_rate": 0.0005, "loss": 1.1644960641860962, "step": 38401 }, { "epoch": 3.0273551438707136, "grad_norm": 0.49876272678375244, "learning_rate": 0.0005, "loss": 1.2574992179870605, "step": 38402 }, { "epoch": 3.0274339771383523, "grad_norm": 0.4430388808250427, "learning_rate": 0.0005, "loss": 1.2730118036270142, "step": 38403 }, { "epoch": 3.0275128104059914, "grad_norm": 0.44602450728416443, "learning_rate": 0.0005, "loss": 1.3443076610565186, "step": 38404 }, { "epoch": 3.02759164367363, "grad_norm": 0.4655686318874359, "learning_rate": 0.0005, "loss": 1.2691251039505005, "step": 38405 }, { "epoch": 3.0276704769412692, "grad_norm": 0.5622986555099487, "learning_rate": 0.0005, "loss": 1.4062203168869019, "step": 38406 }, { "epoch": 3.0277493102089084, "grad_norm": 0.5848335027694702, "learning_rate": 0.0005, "loss": 1.2863394021987915, "step": 38407 }, { "epoch": 3.027828143476547, "grad_norm": 0.4781322479248047, "learning_rate": 0.0005, "loss": 1.3096033334732056, "step": 38408 }, { "epoch": 3.027906976744186, "grad_norm": 0.48617419600486755, "learning_rate": 0.0005, "loss": 1.2500743865966797, "step": 38409 }, { "epoch": 3.027985810011825, "grad_norm": 0.5409343242645264, "learning_rate": 0.0005, "loss": 1.2761776447296143, "step": 38410 }, { "epoch": 3.028064643279464, "grad_norm": 0.5321820974349976, "learning_rate": 0.0005, "loss": 1.2786765098571777, "step": 38411 }, { "epoch": 3.0281434765471027, "grad_norm": 0.5203651189804077, "learning_rate": 0.0005, "loss": 1.3178961277008057, "step": 38412 }, { "epoch": 3.028222309814742, "grad_norm": 0.43643680214881897, "learning_rate": 0.0005, "loss": 1.224790096282959, "step": 38413 }, { "epoch": 3.028301143082381, "grad_norm": 0.447444349527359, "learning_rate": 0.0005, "loss": 1.2465077638626099, "step": 38414 }, { "epoch": 3.0283799763500197, "grad_norm": 0.4604499340057373, "learning_rate": 0.0005, "loss": 1.1615444421768188, "step": 38415 }, { "epoch": 3.028458809617659, "grad_norm": 1.4313526153564453, "learning_rate": 0.0005, "loss": 1.2799972295761108, "step": 38416 }, { "epoch": 3.0285376428852975, "grad_norm": 0.5114001035690308, "learning_rate": 0.0005, "loss": 1.1260061264038086, "step": 38417 }, { "epoch": 3.0286164761529366, "grad_norm": 0.5164429545402527, "learning_rate": 0.0005, "loss": 1.2600864171981812, "step": 38418 }, { "epoch": 3.0286953094205753, "grad_norm": 0.523505687713623, "learning_rate": 0.0005, "loss": 1.2415492534637451, "step": 38419 }, { "epoch": 3.0287741426882144, "grad_norm": 0.4598449170589447, "learning_rate": 0.0005, "loss": 1.371960997581482, "step": 38420 }, { "epoch": 3.0288529759558536, "grad_norm": 0.5485953092575073, "learning_rate": 0.0005, "loss": 1.3536189794540405, "step": 38421 }, { "epoch": 3.0289318092234923, "grad_norm": 0.5794631242752075, "learning_rate": 0.0005, "loss": 1.2506439685821533, "step": 38422 }, { "epoch": 3.0290106424911314, "grad_norm": 0.5978306531906128, "learning_rate": 0.0005, "loss": 1.2965247631072998, "step": 38423 }, { "epoch": 3.02908947575877, "grad_norm": 0.5412899851799011, "learning_rate": 0.0005, "loss": 1.2296779155731201, "step": 38424 }, { "epoch": 3.029168309026409, "grad_norm": 0.5722755789756775, "learning_rate": 0.0005, "loss": 1.3749301433563232, "step": 38425 }, { "epoch": 3.029247142294048, "grad_norm": 0.5081954002380371, "learning_rate": 0.0005, "loss": 1.3911731243133545, "step": 38426 }, { "epoch": 3.029325975561687, "grad_norm": 0.6037060022354126, "learning_rate": 0.0005, "loss": 1.4289085865020752, "step": 38427 }, { "epoch": 3.029404808829326, "grad_norm": 0.4443906247615814, "learning_rate": 0.0005, "loss": 1.3241631984710693, "step": 38428 }, { "epoch": 3.029483642096965, "grad_norm": 0.45240870118141174, "learning_rate": 0.0005, "loss": 1.296573281288147, "step": 38429 }, { "epoch": 3.029562475364604, "grad_norm": 0.4837089478969574, "learning_rate": 0.0005, "loss": 1.0104689598083496, "step": 38430 }, { "epoch": 3.0296413086322427, "grad_norm": 0.3988146483898163, "learning_rate": 0.0005, "loss": 1.2331477403640747, "step": 38431 }, { "epoch": 3.029720141899882, "grad_norm": 0.44010332226753235, "learning_rate": 0.0005, "loss": 1.2160505056381226, "step": 38432 }, { "epoch": 3.0297989751675205, "grad_norm": 0.5046008825302124, "learning_rate": 0.0005, "loss": 1.4230198860168457, "step": 38433 }, { "epoch": 3.0298778084351596, "grad_norm": 0.479308545589447, "learning_rate": 0.0005, "loss": 1.3885884284973145, "step": 38434 }, { "epoch": 3.0299566417027988, "grad_norm": 0.4829701781272888, "learning_rate": 0.0005, "loss": 1.270087718963623, "step": 38435 }, { "epoch": 3.0300354749704375, "grad_norm": 0.4561765789985657, "learning_rate": 0.0005, "loss": 1.270298957824707, "step": 38436 }, { "epoch": 3.0301143082380766, "grad_norm": 0.5141663551330566, "learning_rate": 0.0005, "loss": 1.3006477355957031, "step": 38437 }, { "epoch": 3.0301931415057153, "grad_norm": 0.5145304799079895, "learning_rate": 0.0005, "loss": 1.4905027151107788, "step": 38438 }, { "epoch": 3.0302719747733544, "grad_norm": 0.5683323740959167, "learning_rate": 0.0005, "loss": 1.3222298622131348, "step": 38439 }, { "epoch": 3.030350808040993, "grad_norm": 0.41109344363212585, "learning_rate": 0.0005, "loss": 1.0945918560028076, "step": 38440 }, { "epoch": 3.0304296413086322, "grad_norm": 0.5152697563171387, "learning_rate": 0.0005, "loss": 1.2446415424346924, "step": 38441 }, { "epoch": 3.0305084745762714, "grad_norm": 0.5662177205085754, "learning_rate": 0.0005, "loss": 1.2132470607757568, "step": 38442 }, { "epoch": 3.03058730784391, "grad_norm": 0.5181043148040771, "learning_rate": 0.0005, "loss": 1.231361746788025, "step": 38443 }, { "epoch": 3.030666141111549, "grad_norm": 0.44805631041526794, "learning_rate": 0.0005, "loss": 1.0998644828796387, "step": 38444 }, { "epoch": 3.030744974379188, "grad_norm": 0.4027630090713501, "learning_rate": 0.0005, "loss": 1.102092981338501, "step": 38445 }, { "epoch": 3.030823807646827, "grad_norm": 0.45119810104370117, "learning_rate": 0.0005, "loss": 1.4468039274215698, "step": 38446 }, { "epoch": 3.0309026409144657, "grad_norm": 0.67012619972229, "learning_rate": 0.0005, "loss": 1.2012875080108643, "step": 38447 }, { "epoch": 3.030981474182105, "grad_norm": 0.4536828398704529, "learning_rate": 0.0005, "loss": 1.1127488613128662, "step": 38448 }, { "epoch": 3.031060307449744, "grad_norm": 0.4955998361110687, "learning_rate": 0.0005, "loss": 1.1872128248214722, "step": 38449 }, { "epoch": 3.0311391407173827, "grad_norm": 0.4179271459579468, "learning_rate": 0.0005, "loss": 1.2155060768127441, "step": 38450 }, { "epoch": 3.031217973985022, "grad_norm": 0.5741862654685974, "learning_rate": 0.0005, "loss": 1.302064299583435, "step": 38451 }, { "epoch": 3.0312968072526605, "grad_norm": 0.4966004490852356, "learning_rate": 0.0005, "loss": 1.2390224933624268, "step": 38452 }, { "epoch": 3.0313756405202996, "grad_norm": 0.6588488221168518, "learning_rate": 0.0005, "loss": 1.4030795097351074, "step": 38453 }, { "epoch": 3.0314544737879383, "grad_norm": 0.47474977374076843, "learning_rate": 0.0005, "loss": 1.2713961601257324, "step": 38454 }, { "epoch": 3.0315333070555774, "grad_norm": 0.6133233308792114, "learning_rate": 0.0005, "loss": 1.478991150856018, "step": 38455 }, { "epoch": 3.0316121403232166, "grad_norm": 0.5070640444755554, "learning_rate": 0.0005, "loss": 1.1969038248062134, "step": 38456 }, { "epoch": 3.0316909735908553, "grad_norm": 0.5200008153915405, "learning_rate": 0.0005, "loss": 1.5314176082611084, "step": 38457 }, { "epoch": 3.0317698068584944, "grad_norm": 0.4919328987598419, "learning_rate": 0.0005, "loss": 1.2032968997955322, "step": 38458 }, { "epoch": 3.031848640126133, "grad_norm": 0.43782705068588257, "learning_rate": 0.0005, "loss": 1.1907724142074585, "step": 38459 }, { "epoch": 3.031927473393772, "grad_norm": 0.45738303661346436, "learning_rate": 0.0005, "loss": 1.175781488418579, "step": 38460 }, { "epoch": 3.032006306661411, "grad_norm": 0.39405158162117004, "learning_rate": 0.0005, "loss": 1.1620230674743652, "step": 38461 }, { "epoch": 3.03208513992905, "grad_norm": 0.4309557378292084, "learning_rate": 0.0005, "loss": 1.1119354963302612, "step": 38462 }, { "epoch": 3.032163973196689, "grad_norm": 0.5157806873321533, "learning_rate": 0.0005, "loss": 1.3233630657196045, "step": 38463 }, { "epoch": 3.032242806464328, "grad_norm": 0.431936651468277, "learning_rate": 0.0005, "loss": 1.174045443534851, "step": 38464 }, { "epoch": 3.032321639731967, "grad_norm": 0.5232353210449219, "learning_rate": 0.0005, "loss": 1.2801547050476074, "step": 38465 }, { "epoch": 3.0324004729996057, "grad_norm": 0.47830939292907715, "learning_rate": 0.0005, "loss": 1.1647945642471313, "step": 38466 }, { "epoch": 3.032479306267245, "grad_norm": 0.5303729772567749, "learning_rate": 0.0005, "loss": 1.1871768236160278, "step": 38467 }, { "epoch": 3.0325581395348835, "grad_norm": 0.48857811093330383, "learning_rate": 0.0005, "loss": 1.1158069372177124, "step": 38468 }, { "epoch": 3.0326369728025226, "grad_norm": 0.4583672881126404, "learning_rate": 0.0005, "loss": 1.3531063795089722, "step": 38469 }, { "epoch": 3.032715806070162, "grad_norm": 0.47529804706573486, "learning_rate": 0.0005, "loss": 1.383743405342102, "step": 38470 }, { "epoch": 3.0327946393378005, "grad_norm": 0.48887118697166443, "learning_rate": 0.0005, "loss": 1.286618709564209, "step": 38471 }, { "epoch": 3.0328734726054396, "grad_norm": 0.5071263909339905, "learning_rate": 0.0005, "loss": 1.340214490890503, "step": 38472 }, { "epoch": 3.0329523058730783, "grad_norm": 0.47036030888557434, "learning_rate": 0.0005, "loss": 1.2004427909851074, "step": 38473 }, { "epoch": 3.0330311391407174, "grad_norm": 0.5924125909805298, "learning_rate": 0.0005, "loss": 1.2604401111602783, "step": 38474 }, { "epoch": 3.033109972408356, "grad_norm": 0.6294475793838501, "learning_rate": 0.0005, "loss": 1.2251758575439453, "step": 38475 }, { "epoch": 3.0331888056759952, "grad_norm": 0.5193201899528503, "learning_rate": 0.0005, "loss": 1.2662194967269897, "step": 38476 }, { "epoch": 3.0332676389436344, "grad_norm": 0.5055385231971741, "learning_rate": 0.0005, "loss": 1.1644151210784912, "step": 38477 }, { "epoch": 3.033346472211273, "grad_norm": 0.4494359791278839, "learning_rate": 0.0005, "loss": 1.2963976860046387, "step": 38478 }, { "epoch": 3.033425305478912, "grad_norm": 0.4397508203983307, "learning_rate": 0.0005, "loss": 1.3438127040863037, "step": 38479 }, { "epoch": 3.033504138746551, "grad_norm": 0.4988330900669098, "learning_rate": 0.0005, "loss": 1.3796591758728027, "step": 38480 }, { "epoch": 3.03358297201419, "grad_norm": 0.5232408046722412, "learning_rate": 0.0005, "loss": 1.3704816102981567, "step": 38481 }, { "epoch": 3.0336618052818287, "grad_norm": 0.5629637241363525, "learning_rate": 0.0005, "loss": 1.3832683563232422, "step": 38482 }, { "epoch": 3.033740638549468, "grad_norm": 0.4879716634750366, "learning_rate": 0.0005, "loss": 1.2033019065856934, "step": 38483 }, { "epoch": 3.033819471817107, "grad_norm": 0.384795218706131, "learning_rate": 0.0005, "loss": 1.1349296569824219, "step": 38484 }, { "epoch": 3.0338983050847457, "grad_norm": 0.46206530928611755, "learning_rate": 0.0005, "loss": 1.1641886234283447, "step": 38485 }, { "epoch": 3.033977138352385, "grad_norm": 0.46012458205223083, "learning_rate": 0.0005, "loss": 1.3646619319915771, "step": 38486 }, { "epoch": 3.0340559716200235, "grad_norm": 0.495309054851532, "learning_rate": 0.0005, "loss": 1.1208398342132568, "step": 38487 }, { "epoch": 3.0341348048876626, "grad_norm": 0.5152654051780701, "learning_rate": 0.0005, "loss": 1.2700276374816895, "step": 38488 }, { "epoch": 3.0342136381553013, "grad_norm": 0.6212447881698608, "learning_rate": 0.0005, "loss": 1.2408968210220337, "step": 38489 }, { "epoch": 3.0342924714229405, "grad_norm": 0.48737651109695435, "learning_rate": 0.0005, "loss": 1.438917875289917, "step": 38490 }, { "epoch": 3.0343713046905796, "grad_norm": 0.4740568697452545, "learning_rate": 0.0005, "loss": 1.3734846115112305, "step": 38491 }, { "epoch": 3.0344501379582183, "grad_norm": 0.4423596262931824, "learning_rate": 0.0005, "loss": 1.342329740524292, "step": 38492 }, { "epoch": 3.0345289712258574, "grad_norm": 0.4167326092720032, "learning_rate": 0.0005, "loss": 1.2058401107788086, "step": 38493 }, { "epoch": 3.034607804493496, "grad_norm": 0.5360651016235352, "learning_rate": 0.0005, "loss": 1.1750872135162354, "step": 38494 }, { "epoch": 3.0346866377611352, "grad_norm": 0.5477283596992493, "learning_rate": 0.0005, "loss": 1.4644688367843628, "step": 38495 }, { "epoch": 3.0347654710287744, "grad_norm": 0.3899651765823364, "learning_rate": 0.0005, "loss": 1.0239356756210327, "step": 38496 }, { "epoch": 3.034844304296413, "grad_norm": 0.5307832360267639, "learning_rate": 0.0005, "loss": 1.3370044231414795, "step": 38497 }, { "epoch": 3.034923137564052, "grad_norm": 0.5464719533920288, "learning_rate": 0.0005, "loss": 1.5491366386413574, "step": 38498 }, { "epoch": 3.035001970831691, "grad_norm": 0.41780421137809753, "learning_rate": 0.0005, "loss": 1.1066904067993164, "step": 38499 }, { "epoch": 3.03508080409933, "grad_norm": 0.5065969228744507, "learning_rate": 0.0005, "loss": 1.3184239864349365, "step": 38500 }, { "epoch": 3.0351596373669687, "grad_norm": 0.5449652075767517, "learning_rate": 0.0005, "loss": 1.3779817819595337, "step": 38501 }, { "epoch": 3.035238470634608, "grad_norm": 0.4660690128803253, "learning_rate": 0.0005, "loss": 1.2825498580932617, "step": 38502 }, { "epoch": 3.035317303902247, "grad_norm": 0.3420037031173706, "learning_rate": 0.0005, "loss": 1.09669828414917, "step": 38503 }, { "epoch": 3.0353961371698857, "grad_norm": 0.5904763340950012, "learning_rate": 0.0005, "loss": 1.300745964050293, "step": 38504 }, { "epoch": 3.035474970437525, "grad_norm": 0.39608678221702576, "learning_rate": 0.0005, "loss": 1.1754119396209717, "step": 38505 }, { "epoch": 3.0355538037051635, "grad_norm": 0.5141977071762085, "learning_rate": 0.0005, "loss": 1.3628695011138916, "step": 38506 }, { "epoch": 3.0356326369728026, "grad_norm": 0.6431477665901184, "learning_rate": 0.0005, "loss": 1.500380039215088, "step": 38507 }, { "epoch": 3.0357114702404413, "grad_norm": 0.5419459939002991, "learning_rate": 0.0005, "loss": 1.2841260433197021, "step": 38508 }, { "epoch": 3.0357903035080804, "grad_norm": 0.41591134667396545, "learning_rate": 0.0005, "loss": 1.1607351303100586, "step": 38509 }, { "epoch": 3.0358691367757196, "grad_norm": 0.39887988567352295, "learning_rate": 0.0005, "loss": 1.1811084747314453, "step": 38510 }, { "epoch": 3.0359479700433583, "grad_norm": 0.5298814177513123, "learning_rate": 0.0005, "loss": 1.1524043083190918, "step": 38511 }, { "epoch": 3.0360268033109974, "grad_norm": 0.5412434935569763, "learning_rate": 0.0005, "loss": 1.2760992050170898, "step": 38512 }, { "epoch": 3.036105636578636, "grad_norm": 0.7264583706855774, "learning_rate": 0.0005, "loss": 1.4547901153564453, "step": 38513 }, { "epoch": 3.036184469846275, "grad_norm": 0.5557476878166199, "learning_rate": 0.0005, "loss": 1.2693376541137695, "step": 38514 }, { "epoch": 3.036263303113914, "grad_norm": 0.42178231477737427, "learning_rate": 0.0005, "loss": 1.1991121768951416, "step": 38515 }, { "epoch": 3.036342136381553, "grad_norm": 0.4280041456222534, "learning_rate": 0.0005, "loss": 1.186699628829956, "step": 38516 }, { "epoch": 3.036420969649192, "grad_norm": 0.4301764965057373, "learning_rate": 0.0005, "loss": 1.2176522016525269, "step": 38517 }, { "epoch": 3.036499802916831, "grad_norm": 0.5248825550079346, "learning_rate": 0.0005, "loss": 1.3930987119674683, "step": 38518 }, { "epoch": 3.03657863618447, "grad_norm": 0.6379982233047485, "learning_rate": 0.0005, "loss": 1.3440423011779785, "step": 38519 }, { "epoch": 3.0366574694521087, "grad_norm": 0.45119914412498474, "learning_rate": 0.0005, "loss": 1.1672513484954834, "step": 38520 }, { "epoch": 3.036736302719748, "grad_norm": 0.4938011169433594, "learning_rate": 0.0005, "loss": 1.2036399841308594, "step": 38521 }, { "epoch": 3.0368151359873865, "grad_norm": 0.5691577196121216, "learning_rate": 0.0005, "loss": 1.3586199283599854, "step": 38522 }, { "epoch": 3.0368939692550256, "grad_norm": 0.36828336119651794, "learning_rate": 0.0005, "loss": 1.1644500494003296, "step": 38523 }, { "epoch": 3.0369728025226648, "grad_norm": 0.6232540607452393, "learning_rate": 0.0005, "loss": 1.3207738399505615, "step": 38524 }, { "epoch": 3.0370516357903035, "grad_norm": 1.4313639402389526, "learning_rate": 0.0005, "loss": 1.476491093635559, "step": 38525 }, { "epoch": 3.0371304690579426, "grad_norm": 0.6512470245361328, "learning_rate": 0.0005, "loss": 1.2461382150650024, "step": 38526 }, { "epoch": 3.0372093023255813, "grad_norm": 0.363617479801178, "learning_rate": 0.0005, "loss": 1.1397678852081299, "step": 38527 }, { "epoch": 3.0372881355932204, "grad_norm": 0.4094470739364624, "learning_rate": 0.0005, "loss": 1.3062145709991455, "step": 38528 }, { "epoch": 3.037366968860859, "grad_norm": 0.38656675815582275, "learning_rate": 0.0005, "loss": 1.071527361869812, "step": 38529 }, { "epoch": 3.0374458021284982, "grad_norm": 0.3893638551235199, "learning_rate": 0.0005, "loss": 1.3860671520233154, "step": 38530 }, { "epoch": 3.0375246353961374, "grad_norm": 0.3743937015533447, "learning_rate": 0.0005, "loss": 1.183847427368164, "step": 38531 }, { "epoch": 3.037603468663776, "grad_norm": 0.4583132565021515, "learning_rate": 0.0005, "loss": 1.1442959308624268, "step": 38532 }, { "epoch": 3.037682301931415, "grad_norm": 0.4728868007659912, "learning_rate": 0.0005, "loss": 1.1464781761169434, "step": 38533 }, { "epoch": 3.037761135199054, "grad_norm": 0.6552451252937317, "learning_rate": 0.0005, "loss": 1.2650370597839355, "step": 38534 }, { "epoch": 3.037839968466693, "grad_norm": 0.4845082461833954, "learning_rate": 0.0005, "loss": 1.1893970966339111, "step": 38535 }, { "epoch": 3.0379188017343317, "grad_norm": 0.522142231464386, "learning_rate": 0.0005, "loss": 1.2114959955215454, "step": 38536 }, { "epoch": 3.037997635001971, "grad_norm": 0.5513498783111572, "learning_rate": 0.0005, "loss": 1.2482080459594727, "step": 38537 }, { "epoch": 3.03807646826961, "grad_norm": 0.536553144454956, "learning_rate": 0.0005, "loss": 1.2043845653533936, "step": 38538 }, { "epoch": 3.0381553015372487, "grad_norm": 0.4948250651359558, "learning_rate": 0.0005, "loss": 1.0796093940734863, "step": 38539 }, { "epoch": 3.038234134804888, "grad_norm": 0.4888819754123688, "learning_rate": 0.0005, "loss": 1.2074646949768066, "step": 38540 }, { "epoch": 3.0383129680725265, "grad_norm": 0.43734100461006165, "learning_rate": 0.0005, "loss": 1.2649164199829102, "step": 38541 }, { "epoch": 3.0383918013401656, "grad_norm": 0.558753252029419, "learning_rate": 0.0005, "loss": 1.1239644289016724, "step": 38542 }, { "epoch": 3.0384706346078043, "grad_norm": 0.49975430965423584, "learning_rate": 0.0005, "loss": 1.496889352798462, "step": 38543 }, { "epoch": 3.0385494678754434, "grad_norm": 0.48487499356269836, "learning_rate": 0.0005, "loss": 1.1447609663009644, "step": 38544 }, { "epoch": 3.0386283011430826, "grad_norm": 0.4359559118747711, "learning_rate": 0.0005, "loss": 1.1962676048278809, "step": 38545 }, { "epoch": 3.0387071344107213, "grad_norm": 0.4021799564361572, "learning_rate": 0.0005, "loss": 1.1268881559371948, "step": 38546 }, { "epoch": 3.0387859676783604, "grad_norm": 0.74021977186203, "learning_rate": 0.0005, "loss": 1.1600598096847534, "step": 38547 }, { "epoch": 3.038864800945999, "grad_norm": 0.41386181116104126, "learning_rate": 0.0005, "loss": 1.1654934883117676, "step": 38548 }, { "epoch": 3.0389436342136382, "grad_norm": 0.45094114542007446, "learning_rate": 0.0005, "loss": 1.2759143114089966, "step": 38549 }, { "epoch": 3.039022467481277, "grad_norm": 0.46719658374786377, "learning_rate": 0.0005, "loss": 1.4640779495239258, "step": 38550 }, { "epoch": 3.039101300748916, "grad_norm": 0.3847014605998993, "learning_rate": 0.0005, "loss": 1.0448243618011475, "step": 38551 }, { "epoch": 3.039180134016555, "grad_norm": 0.46374770998954773, "learning_rate": 0.0005, "loss": 1.0795838832855225, "step": 38552 }, { "epoch": 3.039258967284194, "grad_norm": 0.45968928933143616, "learning_rate": 0.0005, "loss": 1.3011457920074463, "step": 38553 }, { "epoch": 3.039337800551833, "grad_norm": 0.509331464767456, "learning_rate": 0.0005, "loss": 1.278502345085144, "step": 38554 }, { "epoch": 3.0394166338194717, "grad_norm": 0.5583842396736145, "learning_rate": 0.0005, "loss": 1.3339587450027466, "step": 38555 }, { "epoch": 3.039495467087111, "grad_norm": 0.5404142737388611, "learning_rate": 0.0005, "loss": 1.4716517925262451, "step": 38556 }, { "epoch": 3.0395743003547495, "grad_norm": 1.1257187128067017, "learning_rate": 0.0005, "loss": 1.1378566026687622, "step": 38557 }, { "epoch": 3.0396531336223886, "grad_norm": 0.4572065472602844, "learning_rate": 0.0005, "loss": 1.1697556972503662, "step": 38558 }, { "epoch": 3.039731966890028, "grad_norm": 0.6116341352462769, "learning_rate": 0.0005, "loss": 1.4662798643112183, "step": 38559 }, { "epoch": 3.0398108001576665, "grad_norm": 0.4665769934654236, "learning_rate": 0.0005, "loss": 1.1945326328277588, "step": 38560 }, { "epoch": 3.0398896334253056, "grad_norm": 0.42821794748306274, "learning_rate": 0.0005, "loss": 1.1665456295013428, "step": 38561 }, { "epoch": 3.0399684666929443, "grad_norm": 0.4734460115432739, "learning_rate": 0.0005, "loss": 1.4108171463012695, "step": 38562 }, { "epoch": 3.0400472999605834, "grad_norm": 0.46344196796417236, "learning_rate": 0.0005, "loss": 1.294507622718811, "step": 38563 }, { "epoch": 3.040126133228222, "grad_norm": 0.5282376408576965, "learning_rate": 0.0005, "loss": 1.3256415128707886, "step": 38564 }, { "epoch": 3.0402049664958612, "grad_norm": 0.46020641922950745, "learning_rate": 0.0005, "loss": 1.2100489139556885, "step": 38565 }, { "epoch": 3.0402837997635004, "grad_norm": 0.4558367133140564, "learning_rate": 0.0005, "loss": 1.1653003692626953, "step": 38566 }, { "epoch": 3.040362633031139, "grad_norm": 0.405504047870636, "learning_rate": 0.0005, "loss": 1.1205476522445679, "step": 38567 }, { "epoch": 3.040441466298778, "grad_norm": 0.4896738827228546, "learning_rate": 0.0005, "loss": 1.2140471935272217, "step": 38568 }, { "epoch": 3.040520299566417, "grad_norm": 0.47099724411964417, "learning_rate": 0.0005, "loss": 1.1252694129943848, "step": 38569 }, { "epoch": 3.040599132834056, "grad_norm": 0.453070729970932, "learning_rate": 0.0005, "loss": 1.1141644716262817, "step": 38570 }, { "epoch": 3.0406779661016947, "grad_norm": 0.5128591060638428, "learning_rate": 0.0005, "loss": 1.380013346672058, "step": 38571 }, { "epoch": 3.040756799369334, "grad_norm": 0.5838190913200378, "learning_rate": 0.0005, "loss": 1.1040863990783691, "step": 38572 }, { "epoch": 3.040835632636973, "grad_norm": 0.5051816701889038, "learning_rate": 0.0005, "loss": 1.3543304204940796, "step": 38573 }, { "epoch": 3.0409144659046117, "grad_norm": 0.5313453674316406, "learning_rate": 0.0005, "loss": 1.3361685276031494, "step": 38574 }, { "epoch": 3.040993299172251, "grad_norm": 0.4592292606830597, "learning_rate": 0.0005, "loss": 1.3264188766479492, "step": 38575 }, { "epoch": 3.0410721324398895, "grad_norm": 0.5002912282943726, "learning_rate": 0.0005, "loss": 1.2558388710021973, "step": 38576 }, { "epoch": 3.0411509657075286, "grad_norm": 0.48239865899086, "learning_rate": 0.0005, "loss": 1.2218363285064697, "step": 38577 }, { "epoch": 3.0412297989751673, "grad_norm": 0.4473535716533661, "learning_rate": 0.0005, "loss": 1.2012836933135986, "step": 38578 }, { "epoch": 3.0413086322428065, "grad_norm": 0.4588888883590698, "learning_rate": 0.0005, "loss": 1.3874452114105225, "step": 38579 }, { "epoch": 3.0413874655104456, "grad_norm": 0.3917752206325531, "learning_rate": 0.0005, "loss": 1.2790852785110474, "step": 38580 }, { "epoch": 3.0414662987780843, "grad_norm": 0.44706636667251587, "learning_rate": 0.0005, "loss": 1.2863640785217285, "step": 38581 }, { "epoch": 3.0415451320457234, "grad_norm": 0.5906779766082764, "learning_rate": 0.0005, "loss": 1.3472449779510498, "step": 38582 }, { "epoch": 3.041623965313362, "grad_norm": 0.5342599153518677, "learning_rate": 0.0005, "loss": 1.2667508125305176, "step": 38583 }, { "epoch": 3.0417027985810012, "grad_norm": 0.4543016850948334, "learning_rate": 0.0005, "loss": 1.1868891716003418, "step": 38584 }, { "epoch": 3.04178163184864, "grad_norm": 0.6117770671844482, "learning_rate": 0.0005, "loss": 1.3948488235473633, "step": 38585 }, { "epoch": 3.041860465116279, "grad_norm": 0.8264676332473755, "learning_rate": 0.0005, "loss": 1.3980655670166016, "step": 38586 }, { "epoch": 3.041939298383918, "grad_norm": 0.5293194651603699, "learning_rate": 0.0005, "loss": 1.2149913311004639, "step": 38587 }, { "epoch": 3.042018131651557, "grad_norm": 0.5384728312492371, "learning_rate": 0.0005, "loss": 1.2226810455322266, "step": 38588 }, { "epoch": 3.042096964919196, "grad_norm": 1.0764198303222656, "learning_rate": 0.0005, "loss": 1.3305705785751343, "step": 38589 }, { "epoch": 3.0421757981868347, "grad_norm": 0.5796283483505249, "learning_rate": 0.0005, "loss": 1.3001734018325806, "step": 38590 }, { "epoch": 3.042254631454474, "grad_norm": 0.6056575775146484, "learning_rate": 0.0005, "loss": 1.2743147611618042, "step": 38591 }, { "epoch": 3.0423334647221125, "grad_norm": 0.44311434030532837, "learning_rate": 0.0005, "loss": 1.1666072607040405, "step": 38592 }, { "epoch": 3.0424122979897517, "grad_norm": 0.5128481984138489, "learning_rate": 0.0005, "loss": 1.3108558654785156, "step": 38593 }, { "epoch": 3.042491131257391, "grad_norm": 0.6024220585823059, "learning_rate": 0.0005, "loss": 1.3665403127670288, "step": 38594 }, { "epoch": 3.0425699645250295, "grad_norm": 0.5646086931228638, "learning_rate": 0.0005, "loss": 1.3923916816711426, "step": 38595 }, { "epoch": 3.0426487977926686, "grad_norm": 0.4207465946674347, "learning_rate": 0.0005, "loss": 1.0785638093948364, "step": 38596 }, { "epoch": 3.0427276310603073, "grad_norm": 0.4675338566303253, "learning_rate": 0.0005, "loss": 1.3195137977600098, "step": 38597 }, { "epoch": 3.0428064643279464, "grad_norm": 0.5807566046714783, "learning_rate": 0.0005, "loss": 1.442503571510315, "step": 38598 }, { "epoch": 3.042885297595585, "grad_norm": 0.4181891977787018, "learning_rate": 0.0005, "loss": 1.2025576829910278, "step": 38599 }, { "epoch": 3.0429641308632243, "grad_norm": 0.590350329875946, "learning_rate": 0.0005, "loss": 1.264145851135254, "step": 38600 }, { "epoch": 3.0430429641308634, "grad_norm": 0.5379387140274048, "learning_rate": 0.0005, "loss": 1.3914086818695068, "step": 38601 }, { "epoch": 3.043121797398502, "grad_norm": 0.48261597752571106, "learning_rate": 0.0005, "loss": 1.3173359632492065, "step": 38602 }, { "epoch": 3.043200630666141, "grad_norm": 0.4617786109447479, "learning_rate": 0.0005, "loss": 1.4431759119033813, "step": 38603 }, { "epoch": 3.04327946393378, "grad_norm": 0.3512524664402008, "learning_rate": 0.0005, "loss": 1.0969440937042236, "step": 38604 }, { "epoch": 3.043358297201419, "grad_norm": 0.5864971876144409, "learning_rate": 0.0005, "loss": 1.1614586114883423, "step": 38605 }, { "epoch": 3.0434371304690577, "grad_norm": 0.41370490193367004, "learning_rate": 0.0005, "loss": 1.0913958549499512, "step": 38606 }, { "epoch": 3.043515963736697, "grad_norm": 0.4544873535633087, "learning_rate": 0.0005, "loss": 1.1588091850280762, "step": 38607 }, { "epoch": 3.043594797004336, "grad_norm": 0.5618942975997925, "learning_rate": 0.0005, "loss": 1.368372917175293, "step": 38608 }, { "epoch": 3.0436736302719747, "grad_norm": 0.6039679050445557, "learning_rate": 0.0005, "loss": 1.2680480480194092, "step": 38609 }, { "epoch": 3.043752463539614, "grad_norm": 0.36966848373413086, "learning_rate": 0.0005, "loss": 1.145265817642212, "step": 38610 }, { "epoch": 3.0438312968072525, "grad_norm": 0.3921754062175751, "learning_rate": 0.0005, "loss": 1.169110655784607, "step": 38611 }, { "epoch": 3.0439101300748916, "grad_norm": 0.47957679629325867, "learning_rate": 0.0005, "loss": 1.2158342599868774, "step": 38612 }, { "epoch": 3.0439889633425303, "grad_norm": 0.5067515969276428, "learning_rate": 0.0005, "loss": 1.1973350048065186, "step": 38613 }, { "epoch": 3.0440677966101695, "grad_norm": 0.507087767124176, "learning_rate": 0.0005, "loss": 1.2927746772766113, "step": 38614 }, { "epoch": 3.0441466298778086, "grad_norm": 0.4862070381641388, "learning_rate": 0.0005, "loss": 1.4060076475143433, "step": 38615 }, { "epoch": 3.0442254631454473, "grad_norm": 0.4592137336730957, "learning_rate": 0.0005, "loss": 1.3615531921386719, "step": 38616 }, { "epoch": 3.0443042964130864, "grad_norm": 0.4735948145389557, "learning_rate": 0.0005, "loss": 1.1731919050216675, "step": 38617 }, { "epoch": 3.044383129680725, "grad_norm": 0.4646974503993988, "learning_rate": 0.0005, "loss": 1.2494497299194336, "step": 38618 }, { "epoch": 3.0444619629483642, "grad_norm": 0.46849629282951355, "learning_rate": 0.0005, "loss": 1.1663603782653809, "step": 38619 }, { "epoch": 3.044540796216003, "grad_norm": 0.40830451250076294, "learning_rate": 0.0005, "loss": 1.270328164100647, "step": 38620 }, { "epoch": 3.044619629483642, "grad_norm": 0.49909886717796326, "learning_rate": 0.0005, "loss": 1.3570806980133057, "step": 38621 }, { "epoch": 3.044698462751281, "grad_norm": 0.4405534267425537, "learning_rate": 0.0005, "loss": 1.180114507675171, "step": 38622 }, { "epoch": 3.04477729601892, "grad_norm": 0.6263576745986938, "learning_rate": 0.0005, "loss": 1.4938814640045166, "step": 38623 }, { "epoch": 3.044856129286559, "grad_norm": 0.6238422393798828, "learning_rate": 0.0005, "loss": 1.2556023597717285, "step": 38624 }, { "epoch": 3.0449349625541977, "grad_norm": 0.40735533833503723, "learning_rate": 0.0005, "loss": 1.0538785457611084, "step": 38625 }, { "epoch": 3.045013795821837, "grad_norm": 0.42446237802505493, "learning_rate": 0.0005, "loss": 1.2152262926101685, "step": 38626 }, { "epoch": 3.045092629089476, "grad_norm": 0.41743579506874084, "learning_rate": 0.0005, "loss": 1.2554633617401123, "step": 38627 }, { "epoch": 3.0451714623571147, "grad_norm": 0.42221948504447937, "learning_rate": 0.0005, "loss": 1.211653709411621, "step": 38628 }, { "epoch": 3.045250295624754, "grad_norm": 0.4344060719013214, "learning_rate": 0.0005, "loss": 1.3013105392456055, "step": 38629 }, { "epoch": 3.0453291288923925, "grad_norm": 0.44144371151924133, "learning_rate": 0.0005, "loss": 1.0996882915496826, "step": 38630 }, { "epoch": 3.0454079621600316, "grad_norm": 0.40263843536376953, "learning_rate": 0.0005, "loss": 1.2119804620742798, "step": 38631 }, { "epoch": 3.0454867954276703, "grad_norm": 0.43152543902397156, "learning_rate": 0.0005, "loss": 1.3030288219451904, "step": 38632 }, { "epoch": 3.0455656286953094, "grad_norm": 0.5291240811347961, "learning_rate": 0.0005, "loss": 1.298887014389038, "step": 38633 }, { "epoch": 3.0456444619629486, "grad_norm": 0.41057711839675903, "learning_rate": 0.0005, "loss": 1.3675603866577148, "step": 38634 }, { "epoch": 3.0457232952305873, "grad_norm": 0.4164160192012787, "learning_rate": 0.0005, "loss": 1.2443201541900635, "step": 38635 }, { "epoch": 3.0458021284982264, "grad_norm": 0.48776817321777344, "learning_rate": 0.0005, "loss": 1.3065838813781738, "step": 38636 }, { "epoch": 3.045880961765865, "grad_norm": 0.5237911343574524, "learning_rate": 0.0005, "loss": 1.187476396560669, "step": 38637 }, { "epoch": 3.0459597950335042, "grad_norm": 0.6104994416236877, "learning_rate": 0.0005, "loss": 1.2632852792739868, "step": 38638 }, { "epoch": 3.046038628301143, "grad_norm": 0.5814488530158997, "learning_rate": 0.0005, "loss": 1.2502570152282715, "step": 38639 }, { "epoch": 3.046117461568782, "grad_norm": 0.5047475695610046, "learning_rate": 0.0005, "loss": 1.3433587551116943, "step": 38640 }, { "epoch": 3.046196294836421, "grad_norm": 0.4959716200828552, "learning_rate": 0.0005, "loss": 1.369673728942871, "step": 38641 }, { "epoch": 3.04627512810406, "grad_norm": 0.5434342622756958, "learning_rate": 0.0005, "loss": 1.3363234996795654, "step": 38642 }, { "epoch": 3.046353961371699, "grad_norm": 0.391262024641037, "learning_rate": 0.0005, "loss": 1.082585096359253, "step": 38643 }, { "epoch": 3.0464327946393377, "grad_norm": 0.43478503823280334, "learning_rate": 0.0005, "loss": 1.3260276317596436, "step": 38644 }, { "epoch": 3.046511627906977, "grad_norm": 0.4700251519680023, "learning_rate": 0.0005, "loss": 1.3534743785858154, "step": 38645 }, { "epoch": 3.0465904611746155, "grad_norm": 0.40883317589759827, "learning_rate": 0.0005, "loss": 1.275871753692627, "step": 38646 }, { "epoch": 3.0466692944422546, "grad_norm": 0.9782808423042297, "learning_rate": 0.0005, "loss": 1.307342290878296, "step": 38647 }, { "epoch": 3.046748127709894, "grad_norm": 0.46892184019088745, "learning_rate": 0.0005, "loss": 1.2592363357543945, "step": 38648 }, { "epoch": 3.0468269609775325, "grad_norm": 0.45095109939575195, "learning_rate": 0.0005, "loss": 1.085792899131775, "step": 38649 }, { "epoch": 3.0469057942451716, "grad_norm": 0.4864087700843811, "learning_rate": 0.0005, "loss": 1.3032915592193604, "step": 38650 }, { "epoch": 3.0469846275128103, "grad_norm": 0.41610825061798096, "learning_rate": 0.0005, "loss": 1.2436299324035645, "step": 38651 }, { "epoch": 3.0470634607804494, "grad_norm": 0.5329388380050659, "learning_rate": 0.0005, "loss": 1.2554614543914795, "step": 38652 }, { "epoch": 3.047142294048088, "grad_norm": 0.45030081272125244, "learning_rate": 0.0005, "loss": 1.287063717842102, "step": 38653 }, { "epoch": 3.0472211273157273, "grad_norm": 0.5642667412757874, "learning_rate": 0.0005, "loss": 1.399641990661621, "step": 38654 }, { "epoch": 3.0472999605833664, "grad_norm": 0.5708595514297485, "learning_rate": 0.0005, "loss": 1.3512959480285645, "step": 38655 }, { "epoch": 3.047378793851005, "grad_norm": 0.4813443422317505, "learning_rate": 0.0005, "loss": 1.1970586776733398, "step": 38656 }, { "epoch": 3.047457627118644, "grad_norm": 0.4274310767650604, "learning_rate": 0.0005, "loss": 1.3741216659545898, "step": 38657 }, { "epoch": 3.047536460386283, "grad_norm": 0.4572593569755554, "learning_rate": 0.0005, "loss": 1.302454948425293, "step": 38658 }, { "epoch": 3.047615293653922, "grad_norm": 0.5813626050949097, "learning_rate": 0.0005, "loss": 1.3818843364715576, "step": 38659 }, { "epoch": 3.0476941269215607, "grad_norm": 0.4731156527996063, "learning_rate": 0.0005, "loss": 1.3253902196884155, "step": 38660 }, { "epoch": 3.0477729601892, "grad_norm": 0.4712454080581665, "learning_rate": 0.0005, "loss": 1.151880145072937, "step": 38661 }, { "epoch": 3.047851793456839, "grad_norm": 0.5128868222236633, "learning_rate": 0.0005, "loss": 1.2112481594085693, "step": 38662 }, { "epoch": 3.0479306267244777, "grad_norm": 0.5348307490348816, "learning_rate": 0.0005, "loss": 1.3518122434616089, "step": 38663 }, { "epoch": 3.048009459992117, "grad_norm": 0.5216495394706726, "learning_rate": 0.0005, "loss": 1.7450002431869507, "step": 38664 }, { "epoch": 3.0480882932597555, "grad_norm": 0.44482478499412537, "learning_rate": 0.0005, "loss": 1.214851975440979, "step": 38665 }, { "epoch": 3.0481671265273946, "grad_norm": 0.4759102165699005, "learning_rate": 0.0005, "loss": 1.420771598815918, "step": 38666 }, { "epoch": 3.0482459597950333, "grad_norm": 0.5413580536842346, "learning_rate": 0.0005, "loss": 1.5865497589111328, "step": 38667 }, { "epoch": 3.0483247930626725, "grad_norm": 0.4627858102321625, "learning_rate": 0.0005, "loss": 1.0796973705291748, "step": 38668 }, { "epoch": 3.0484036263303116, "grad_norm": 0.4794296324253082, "learning_rate": 0.0005, "loss": 1.2373762130737305, "step": 38669 }, { "epoch": 3.0484824595979503, "grad_norm": 0.547110915184021, "learning_rate": 0.0005, "loss": 1.3887916803359985, "step": 38670 }, { "epoch": 3.0485612928655894, "grad_norm": 0.4895302653312683, "learning_rate": 0.0005, "loss": 1.3401355743408203, "step": 38671 }, { "epoch": 3.048640126133228, "grad_norm": 0.5278130769729614, "learning_rate": 0.0005, "loss": 1.4259618520736694, "step": 38672 }, { "epoch": 3.0487189594008672, "grad_norm": 0.48179447650909424, "learning_rate": 0.0005, "loss": 1.3005578517913818, "step": 38673 }, { "epoch": 3.048797792668506, "grad_norm": 0.4310542941093445, "learning_rate": 0.0005, "loss": 1.2094308137893677, "step": 38674 }, { "epoch": 3.048876625936145, "grad_norm": 0.5042990446090698, "learning_rate": 0.0005, "loss": 1.3801178932189941, "step": 38675 }, { "epoch": 3.048955459203784, "grad_norm": 0.5554178357124329, "learning_rate": 0.0005, "loss": 1.473288893699646, "step": 38676 }, { "epoch": 3.049034292471423, "grad_norm": 0.4828067421913147, "learning_rate": 0.0005, "loss": 1.2623906135559082, "step": 38677 }, { "epoch": 3.049113125739062, "grad_norm": 0.515752911567688, "learning_rate": 0.0005, "loss": 1.4883630275726318, "step": 38678 }, { "epoch": 3.0491919590067007, "grad_norm": 0.3887334167957306, "learning_rate": 0.0005, "loss": 1.2777736186981201, "step": 38679 }, { "epoch": 3.04927079227434, "grad_norm": 0.47853147983551025, "learning_rate": 0.0005, "loss": 1.2795096635818481, "step": 38680 }, { "epoch": 3.0493496255419785, "grad_norm": 0.4107706844806671, "learning_rate": 0.0005, "loss": 1.1787834167480469, "step": 38681 }, { "epoch": 3.0494284588096177, "grad_norm": 0.4773777425289154, "learning_rate": 0.0005, "loss": 1.3580288887023926, "step": 38682 }, { "epoch": 3.049507292077257, "grad_norm": 0.3807441294193268, "learning_rate": 0.0005, "loss": 1.1872382164001465, "step": 38683 }, { "epoch": 3.0495861253448955, "grad_norm": 0.4735283851623535, "learning_rate": 0.0005, "loss": 1.3882137537002563, "step": 38684 }, { "epoch": 3.0496649586125346, "grad_norm": 0.5026901960372925, "learning_rate": 0.0005, "loss": 1.3252975940704346, "step": 38685 }, { "epoch": 3.0497437918801733, "grad_norm": 0.48087558150291443, "learning_rate": 0.0005, "loss": 1.1579015254974365, "step": 38686 }, { "epoch": 3.0498226251478124, "grad_norm": 0.44291138648986816, "learning_rate": 0.0005, "loss": 1.306368350982666, "step": 38687 }, { "epoch": 3.049901458415451, "grad_norm": 0.4699912667274475, "learning_rate": 0.0005, "loss": 1.3420379161834717, "step": 38688 }, { "epoch": 3.0499802916830903, "grad_norm": 0.4287806451320648, "learning_rate": 0.0005, "loss": 1.1500505208969116, "step": 38689 }, { "epoch": 3.0500591249507294, "grad_norm": 0.5329653024673462, "learning_rate": 0.0005, "loss": 1.4767482280731201, "step": 38690 }, { "epoch": 3.050137958218368, "grad_norm": 0.48322737216949463, "learning_rate": 0.0005, "loss": 1.2861887216567993, "step": 38691 }, { "epoch": 3.050216791486007, "grad_norm": 0.4382833242416382, "learning_rate": 0.0005, "loss": 1.2973095178604126, "step": 38692 }, { "epoch": 3.050295624753646, "grad_norm": 0.4807901382446289, "learning_rate": 0.0005, "loss": 1.2670482397079468, "step": 38693 }, { "epoch": 3.050374458021285, "grad_norm": 0.5689846873283386, "learning_rate": 0.0005, "loss": 1.3255581855773926, "step": 38694 }, { "epoch": 3.0504532912889237, "grad_norm": 0.5315099954605103, "learning_rate": 0.0005, "loss": 1.4335534572601318, "step": 38695 }, { "epoch": 3.050532124556563, "grad_norm": 0.4643932580947876, "learning_rate": 0.0005, "loss": 1.1839739084243774, "step": 38696 }, { "epoch": 3.050610957824202, "grad_norm": 0.46048516035079956, "learning_rate": 0.0005, "loss": 1.0835850238800049, "step": 38697 }, { "epoch": 3.0506897910918407, "grad_norm": 0.456268846988678, "learning_rate": 0.0005, "loss": 1.4012476205825806, "step": 38698 }, { "epoch": 3.05076862435948, "grad_norm": 0.4945274889469147, "learning_rate": 0.0005, "loss": 1.2566598653793335, "step": 38699 }, { "epoch": 3.0508474576271185, "grad_norm": 0.4558398425579071, "learning_rate": 0.0005, "loss": 1.1435456275939941, "step": 38700 }, { "epoch": 3.0509262908947576, "grad_norm": 0.4047398269176483, "learning_rate": 0.0005, "loss": 1.265372633934021, "step": 38701 }, { "epoch": 3.0510051241623963, "grad_norm": 0.44028404355049133, "learning_rate": 0.0005, "loss": 1.1590131521224976, "step": 38702 }, { "epoch": 3.0510839574300355, "grad_norm": 0.6212810277938843, "learning_rate": 0.0005, "loss": 1.210313320159912, "step": 38703 }, { "epoch": 3.0511627906976746, "grad_norm": 0.4095357358455658, "learning_rate": 0.0005, "loss": 1.2034621238708496, "step": 38704 }, { "epoch": 3.0512416239653133, "grad_norm": 0.5396894216537476, "learning_rate": 0.0005, "loss": 1.2894049882888794, "step": 38705 }, { "epoch": 3.0513204572329524, "grad_norm": 0.42067471146583557, "learning_rate": 0.0005, "loss": 1.1001611948013306, "step": 38706 }, { "epoch": 3.051399290500591, "grad_norm": 0.4911874830722809, "learning_rate": 0.0005, "loss": 1.3436702489852905, "step": 38707 }, { "epoch": 3.0514781237682302, "grad_norm": 0.5195749402046204, "learning_rate": 0.0005, "loss": 1.2416942119598389, "step": 38708 }, { "epoch": 3.051556957035869, "grad_norm": 0.47207072377204895, "learning_rate": 0.0005, "loss": 1.1663886308670044, "step": 38709 }, { "epoch": 3.051635790303508, "grad_norm": 0.413986474275589, "learning_rate": 0.0005, "loss": 1.185567855834961, "step": 38710 }, { "epoch": 3.051714623571147, "grad_norm": 0.491836279630661, "learning_rate": 0.0005, "loss": 1.3198397159576416, "step": 38711 }, { "epoch": 3.051793456838786, "grad_norm": 0.5604976415634155, "learning_rate": 0.0005, "loss": 1.3145751953125, "step": 38712 }, { "epoch": 3.051872290106425, "grad_norm": 0.39901432394981384, "learning_rate": 0.0005, "loss": 1.0239808559417725, "step": 38713 }, { "epoch": 3.0519511233740637, "grad_norm": 0.5007560849189758, "learning_rate": 0.0005, "loss": 1.0544863939285278, "step": 38714 }, { "epoch": 3.052029956641703, "grad_norm": 0.4752526879310608, "learning_rate": 0.0005, "loss": 1.3470828533172607, "step": 38715 }, { "epoch": 3.0521087899093415, "grad_norm": 0.3672970235347748, "learning_rate": 0.0005, "loss": 1.039646863937378, "step": 38716 }, { "epoch": 3.0521876231769807, "grad_norm": 0.45749980211257935, "learning_rate": 0.0005, "loss": 1.3027291297912598, "step": 38717 }, { "epoch": 3.05226645644462, "grad_norm": 0.44515490531921387, "learning_rate": 0.0005, "loss": 1.1875145435333252, "step": 38718 }, { "epoch": 3.0523452897122585, "grad_norm": 0.5292689204216003, "learning_rate": 0.0005, "loss": 1.328930139541626, "step": 38719 }, { "epoch": 3.0524241229798976, "grad_norm": 0.38189923763275146, "learning_rate": 0.0005, "loss": 1.2190799713134766, "step": 38720 }, { "epoch": 3.0525029562475363, "grad_norm": 0.44708797335624695, "learning_rate": 0.0005, "loss": 1.1513361930847168, "step": 38721 }, { "epoch": 3.0525817895151754, "grad_norm": 0.519799530506134, "learning_rate": 0.0005, "loss": 1.5237623453140259, "step": 38722 }, { "epoch": 3.052660622782814, "grad_norm": 0.5608494877815247, "learning_rate": 0.0005, "loss": 1.468430519104004, "step": 38723 }, { "epoch": 3.0527394560504533, "grad_norm": 0.7032254338264465, "learning_rate": 0.0005, "loss": 1.2575573921203613, "step": 38724 }, { "epoch": 3.0528182893180924, "grad_norm": 0.5407362580299377, "learning_rate": 0.0005, "loss": 1.1909852027893066, "step": 38725 }, { "epoch": 3.052897122585731, "grad_norm": 0.48982781171798706, "learning_rate": 0.0005, "loss": 1.3941679000854492, "step": 38726 }, { "epoch": 3.0529759558533702, "grad_norm": 0.4475078880786896, "learning_rate": 0.0005, "loss": 1.1975219249725342, "step": 38727 }, { "epoch": 3.053054789121009, "grad_norm": 0.5333338975906372, "learning_rate": 0.0005, "loss": 1.3412566184997559, "step": 38728 }, { "epoch": 3.053133622388648, "grad_norm": 0.5422744154930115, "learning_rate": 0.0005, "loss": 1.1401163339614868, "step": 38729 }, { "epoch": 3.0532124556562867, "grad_norm": 0.4826074540615082, "learning_rate": 0.0005, "loss": 1.2168960571289062, "step": 38730 }, { "epoch": 3.053291288923926, "grad_norm": 0.49765703082084656, "learning_rate": 0.0005, "loss": 1.4275760650634766, "step": 38731 }, { "epoch": 3.053370122191565, "grad_norm": 0.4368819296360016, "learning_rate": 0.0005, "loss": 1.082059621810913, "step": 38732 }, { "epoch": 3.0534489554592037, "grad_norm": 0.5907634496688843, "learning_rate": 0.0005, "loss": 1.1952471733093262, "step": 38733 }, { "epoch": 3.053527788726843, "grad_norm": 0.5602080225944519, "learning_rate": 0.0005, "loss": 1.3876569271087646, "step": 38734 }, { "epoch": 3.0536066219944815, "grad_norm": 0.43752437829971313, "learning_rate": 0.0005, "loss": 1.2804415225982666, "step": 38735 }, { "epoch": 3.0536854552621207, "grad_norm": 0.4493224620819092, "learning_rate": 0.0005, "loss": 1.1992907524108887, "step": 38736 }, { "epoch": 3.0537642885297593, "grad_norm": 0.5816624760627747, "learning_rate": 0.0005, "loss": 1.3936090469360352, "step": 38737 }, { "epoch": 3.0538431217973985, "grad_norm": 0.6551862955093384, "learning_rate": 0.0005, "loss": 1.2725586891174316, "step": 38738 }, { "epoch": 3.0539219550650376, "grad_norm": 0.41960492730140686, "learning_rate": 0.0005, "loss": 1.1932768821716309, "step": 38739 }, { "epoch": 3.0540007883326763, "grad_norm": 0.4617581069469452, "learning_rate": 0.0005, "loss": 1.2163050174713135, "step": 38740 }, { "epoch": 3.0540796216003154, "grad_norm": 0.549727201461792, "learning_rate": 0.0005, "loss": 1.3288424015045166, "step": 38741 }, { "epoch": 3.054158454867954, "grad_norm": 0.4660183787345886, "learning_rate": 0.0005, "loss": 1.2240021228790283, "step": 38742 }, { "epoch": 3.0542372881355933, "grad_norm": 0.43753936886787415, "learning_rate": 0.0005, "loss": 1.2075862884521484, "step": 38743 }, { "epoch": 3.0543161214032324, "grad_norm": 0.43602052330970764, "learning_rate": 0.0005, "loss": 1.183876395225525, "step": 38744 }, { "epoch": 3.054394954670871, "grad_norm": 0.5006428956985474, "learning_rate": 0.0005, "loss": 1.309419870376587, "step": 38745 }, { "epoch": 3.05447378793851, "grad_norm": 0.45522236824035645, "learning_rate": 0.0005, "loss": 1.1534013748168945, "step": 38746 }, { "epoch": 3.054552621206149, "grad_norm": 0.4102388918399811, "learning_rate": 0.0005, "loss": 1.1804596185684204, "step": 38747 }, { "epoch": 3.054631454473788, "grad_norm": 0.5027065873146057, "learning_rate": 0.0005, "loss": 1.313965082168579, "step": 38748 }, { "epoch": 3.0547102877414267, "grad_norm": 0.41021597385406494, "learning_rate": 0.0005, "loss": 1.2511341571807861, "step": 38749 }, { "epoch": 3.054789121009066, "grad_norm": 0.5143755078315735, "learning_rate": 0.0005, "loss": 1.4033465385437012, "step": 38750 }, { "epoch": 3.0548679542767045, "grad_norm": 0.4034670889377594, "learning_rate": 0.0005, "loss": 1.1771880388259888, "step": 38751 }, { "epoch": 3.0549467875443437, "grad_norm": 0.4416608214378357, "learning_rate": 0.0005, "loss": 1.1893326044082642, "step": 38752 }, { "epoch": 3.055025620811983, "grad_norm": 0.3532995283603668, "learning_rate": 0.0005, "loss": 1.1994750499725342, "step": 38753 }, { "epoch": 3.0551044540796215, "grad_norm": 0.4927957057952881, "learning_rate": 0.0005, "loss": 1.2373249530792236, "step": 38754 }, { "epoch": 3.0551832873472606, "grad_norm": 0.4685453772544861, "learning_rate": 0.0005, "loss": 1.1503269672393799, "step": 38755 }, { "epoch": 3.0552621206148993, "grad_norm": 0.6169057488441467, "learning_rate": 0.0005, "loss": 1.2082836627960205, "step": 38756 }, { "epoch": 3.0553409538825385, "grad_norm": 0.5450137257575989, "learning_rate": 0.0005, "loss": 1.2564984560012817, "step": 38757 }, { "epoch": 3.0554197871501776, "grad_norm": 0.37156936526298523, "learning_rate": 0.0005, "loss": 1.161562204360962, "step": 38758 }, { "epoch": 3.0554986204178163, "grad_norm": 0.57814621925354, "learning_rate": 0.0005, "loss": 1.281965732574463, "step": 38759 }, { "epoch": 3.0555774536854554, "grad_norm": 0.48270660638809204, "learning_rate": 0.0005, "loss": 1.2919032573699951, "step": 38760 }, { "epoch": 3.055656286953094, "grad_norm": 0.5088253021240234, "learning_rate": 0.0005, "loss": 1.3095641136169434, "step": 38761 }, { "epoch": 3.0557351202207332, "grad_norm": 0.5327481627464294, "learning_rate": 0.0005, "loss": 1.2091636657714844, "step": 38762 }, { "epoch": 3.055813953488372, "grad_norm": 0.41667208075523376, "learning_rate": 0.0005, "loss": 1.2319356203079224, "step": 38763 }, { "epoch": 3.055892786756011, "grad_norm": 0.4661937355995178, "learning_rate": 0.0005, "loss": 1.1745675802230835, "step": 38764 }, { "epoch": 3.05597162002365, "grad_norm": 0.6019800901412964, "learning_rate": 0.0005, "loss": 1.4209539890289307, "step": 38765 }, { "epoch": 3.056050453291289, "grad_norm": 0.4433797001838684, "learning_rate": 0.0005, "loss": 1.4522188901901245, "step": 38766 }, { "epoch": 3.056129286558928, "grad_norm": 0.5477941632270813, "learning_rate": 0.0005, "loss": 1.2380446195602417, "step": 38767 }, { "epoch": 3.0562081198265667, "grad_norm": 0.41242507100105286, "learning_rate": 0.0005, "loss": 1.2056806087493896, "step": 38768 }, { "epoch": 3.056286953094206, "grad_norm": 0.43652480840682983, "learning_rate": 0.0005, "loss": 1.2497823238372803, "step": 38769 }, { "epoch": 3.0563657863618445, "grad_norm": 0.5234698057174683, "learning_rate": 0.0005, "loss": 1.3770067691802979, "step": 38770 }, { "epoch": 3.0564446196294837, "grad_norm": 0.4792308509349823, "learning_rate": 0.0005, "loss": 1.309220790863037, "step": 38771 }, { "epoch": 3.056523452897123, "grad_norm": 0.46536311507225037, "learning_rate": 0.0005, "loss": 1.235242247581482, "step": 38772 }, { "epoch": 3.0566022861647615, "grad_norm": 0.4077453911304474, "learning_rate": 0.0005, "loss": 1.3015944957733154, "step": 38773 }, { "epoch": 3.0566811194324006, "grad_norm": 0.4146573543548584, "learning_rate": 0.0005, "loss": 1.1759281158447266, "step": 38774 }, { "epoch": 3.0567599527000393, "grad_norm": 0.434906005859375, "learning_rate": 0.0005, "loss": 1.2895069122314453, "step": 38775 }, { "epoch": 3.0568387859676784, "grad_norm": 0.4559861123561859, "learning_rate": 0.0005, "loss": 1.4043498039245605, "step": 38776 }, { "epoch": 3.056917619235317, "grad_norm": 0.45962727069854736, "learning_rate": 0.0005, "loss": 1.3759162425994873, "step": 38777 }, { "epoch": 3.0569964525029563, "grad_norm": 0.5348924994468689, "learning_rate": 0.0005, "loss": 1.2541728019714355, "step": 38778 }, { "epoch": 3.0570752857705954, "grad_norm": 0.42401421070098877, "learning_rate": 0.0005, "loss": 1.2038381099700928, "step": 38779 }, { "epoch": 3.057154119038234, "grad_norm": 0.4930373430252075, "learning_rate": 0.0005, "loss": 1.2808995246887207, "step": 38780 }, { "epoch": 3.057232952305873, "grad_norm": 0.4430539608001709, "learning_rate": 0.0005, "loss": 1.0540850162506104, "step": 38781 }, { "epoch": 3.057311785573512, "grad_norm": 0.4929957091808319, "learning_rate": 0.0005, "loss": 1.3678483963012695, "step": 38782 }, { "epoch": 3.057390618841151, "grad_norm": 0.4340328276157379, "learning_rate": 0.0005, "loss": 1.226752758026123, "step": 38783 }, { "epoch": 3.0574694521087897, "grad_norm": 0.5089324712753296, "learning_rate": 0.0005, "loss": 1.2137935161590576, "step": 38784 }, { "epoch": 3.057548285376429, "grad_norm": 0.4598979353904724, "learning_rate": 0.0005, "loss": 1.2161157131195068, "step": 38785 }, { "epoch": 3.057627118644068, "grad_norm": 0.516022264957428, "learning_rate": 0.0005, "loss": 1.4555035829544067, "step": 38786 }, { "epoch": 3.0577059519117067, "grad_norm": 0.5282260775566101, "learning_rate": 0.0005, "loss": 1.1565673351287842, "step": 38787 }, { "epoch": 3.057784785179346, "grad_norm": 8.31975269317627, "learning_rate": 0.0005, "loss": 1.32073974609375, "step": 38788 }, { "epoch": 3.0578636184469845, "grad_norm": 0.44482421875, "learning_rate": 0.0005, "loss": 1.145017385482788, "step": 38789 }, { "epoch": 3.0579424517146236, "grad_norm": 0.612695574760437, "learning_rate": 0.0005, "loss": 1.2359459400177002, "step": 38790 }, { "epoch": 3.0580212849822623, "grad_norm": 0.44029369950294495, "learning_rate": 0.0005, "loss": 1.256993293762207, "step": 38791 }, { "epoch": 3.0581001182499015, "grad_norm": 0.4877171516418457, "learning_rate": 0.0005, "loss": 1.381347894668579, "step": 38792 }, { "epoch": 3.0581789515175406, "grad_norm": 0.4680120646953583, "learning_rate": 0.0005, "loss": 1.3293200731277466, "step": 38793 }, { "epoch": 3.0582577847851793, "grad_norm": 0.4684521555900574, "learning_rate": 0.0005, "loss": 1.3172457218170166, "step": 38794 }, { "epoch": 3.0583366180528184, "grad_norm": 0.44362151622772217, "learning_rate": 0.0005, "loss": 1.333655595779419, "step": 38795 }, { "epoch": 3.058415451320457, "grad_norm": 0.4541197121143341, "learning_rate": 0.0005, "loss": 1.2006347179412842, "step": 38796 }, { "epoch": 3.0584942845880962, "grad_norm": 0.6252738833427429, "learning_rate": 0.0005, "loss": 1.3143515586853027, "step": 38797 }, { "epoch": 3.058573117855735, "grad_norm": 0.4939327538013458, "learning_rate": 0.0005, "loss": 1.2613276243209839, "step": 38798 }, { "epoch": 3.058651951123374, "grad_norm": 0.38181957602500916, "learning_rate": 0.0005, "loss": 1.0441982746124268, "step": 38799 }, { "epoch": 3.058730784391013, "grad_norm": 0.4813673496246338, "learning_rate": 0.0005, "loss": 1.3384442329406738, "step": 38800 }, { "epoch": 3.058809617658652, "grad_norm": 0.5763254165649414, "learning_rate": 0.0005, "loss": 1.3705077171325684, "step": 38801 }, { "epoch": 3.058888450926291, "grad_norm": 0.5399033427238464, "learning_rate": 0.0005, "loss": 1.3023171424865723, "step": 38802 }, { "epoch": 3.0589672841939297, "grad_norm": 0.41799643635749817, "learning_rate": 0.0005, "loss": 1.2134437561035156, "step": 38803 }, { "epoch": 3.059046117461569, "grad_norm": 0.44899651408195496, "learning_rate": 0.0005, "loss": 1.1523842811584473, "step": 38804 }, { "epoch": 3.0591249507292075, "grad_norm": 0.46470680832862854, "learning_rate": 0.0005, "loss": 1.165886640548706, "step": 38805 }, { "epoch": 3.0592037839968467, "grad_norm": 0.46751171350479126, "learning_rate": 0.0005, "loss": 1.444555401802063, "step": 38806 }, { "epoch": 3.059282617264486, "grad_norm": 0.38279712200164795, "learning_rate": 0.0005, "loss": 0.9049522876739502, "step": 38807 }, { "epoch": 3.0593614505321245, "grad_norm": 0.4660874009132385, "learning_rate": 0.0005, "loss": 1.3082544803619385, "step": 38808 }, { "epoch": 3.0594402837997636, "grad_norm": 0.5592897534370422, "learning_rate": 0.0005, "loss": 1.2368764877319336, "step": 38809 }, { "epoch": 3.0595191170674023, "grad_norm": 0.4341169595718384, "learning_rate": 0.0005, "loss": 1.3400124311447144, "step": 38810 }, { "epoch": 3.0595979503350414, "grad_norm": 0.4721355736255646, "learning_rate": 0.0005, "loss": 1.1998145580291748, "step": 38811 }, { "epoch": 3.05967678360268, "grad_norm": 0.4941612482070923, "learning_rate": 0.0005, "loss": 1.3228435516357422, "step": 38812 }, { "epoch": 3.0597556168703193, "grad_norm": 0.4502781927585602, "learning_rate": 0.0005, "loss": 1.1526519060134888, "step": 38813 }, { "epoch": 3.0598344501379584, "grad_norm": 0.43779242038726807, "learning_rate": 0.0005, "loss": 1.1657299995422363, "step": 38814 }, { "epoch": 3.059913283405597, "grad_norm": 0.48246490955352783, "learning_rate": 0.0005, "loss": 1.2299838066101074, "step": 38815 }, { "epoch": 3.0599921166732362, "grad_norm": 0.4464299976825714, "learning_rate": 0.0005, "loss": 1.2956876754760742, "step": 38816 }, { "epoch": 3.060070949940875, "grad_norm": 0.6507334113121033, "learning_rate": 0.0005, "loss": 1.2845942974090576, "step": 38817 }, { "epoch": 3.060149783208514, "grad_norm": 0.3979213833808899, "learning_rate": 0.0005, "loss": 1.0459083318710327, "step": 38818 }, { "epoch": 3.0602286164761527, "grad_norm": 0.5389127135276794, "learning_rate": 0.0005, "loss": 1.3433890342712402, "step": 38819 }, { "epoch": 3.060307449743792, "grad_norm": 0.5394101142883301, "learning_rate": 0.0005, "loss": 1.3252677917480469, "step": 38820 }, { "epoch": 3.060386283011431, "grad_norm": 0.570287823677063, "learning_rate": 0.0005, "loss": 1.264965534210205, "step": 38821 }, { "epoch": 3.0604651162790697, "grad_norm": 0.47396039962768555, "learning_rate": 0.0005, "loss": 1.1228046417236328, "step": 38822 }, { "epoch": 3.060543949546709, "grad_norm": 0.5863412022590637, "learning_rate": 0.0005, "loss": 1.3805285692214966, "step": 38823 }, { "epoch": 3.0606227828143475, "grad_norm": 0.5164576768875122, "learning_rate": 0.0005, "loss": 1.2001088857650757, "step": 38824 }, { "epoch": 3.0607016160819867, "grad_norm": 0.48678407073020935, "learning_rate": 0.0005, "loss": 1.399137258529663, "step": 38825 }, { "epoch": 3.0607804493496253, "grad_norm": 0.4268142580986023, "learning_rate": 0.0005, "loss": 1.1973271369934082, "step": 38826 }, { "epoch": 3.0608592826172645, "grad_norm": 0.48844021558761597, "learning_rate": 0.0005, "loss": 1.2283897399902344, "step": 38827 }, { "epoch": 3.0609381158849036, "grad_norm": 0.4614468812942505, "learning_rate": 0.0005, "loss": 1.217795968055725, "step": 38828 }, { "epoch": 3.0610169491525423, "grad_norm": 0.4916088879108429, "learning_rate": 0.0005, "loss": 1.1510868072509766, "step": 38829 }, { "epoch": 3.0610957824201814, "grad_norm": 0.4932372570037842, "learning_rate": 0.0005, "loss": 1.365553855895996, "step": 38830 }, { "epoch": 3.06117461568782, "grad_norm": 0.6620251536369324, "learning_rate": 0.0005, "loss": 1.229067325592041, "step": 38831 }, { "epoch": 3.0612534489554593, "grad_norm": 0.5550984740257263, "learning_rate": 0.0005, "loss": 1.239808440208435, "step": 38832 }, { "epoch": 3.061332282223098, "grad_norm": 0.48367932438850403, "learning_rate": 0.0005, "loss": 1.184892177581787, "step": 38833 }, { "epoch": 3.061411115490737, "grad_norm": 0.5348008275032043, "learning_rate": 0.0005, "loss": 1.358987808227539, "step": 38834 }, { "epoch": 3.061489948758376, "grad_norm": 0.5667091608047485, "learning_rate": 0.0005, "loss": 1.2040337324142456, "step": 38835 }, { "epoch": 3.061568782026015, "grad_norm": 0.5073720216751099, "learning_rate": 0.0005, "loss": 1.3683745861053467, "step": 38836 }, { "epoch": 3.061647615293654, "grad_norm": 0.4249161183834076, "learning_rate": 0.0005, "loss": 1.150076150894165, "step": 38837 }, { "epoch": 3.0617264485612927, "grad_norm": 0.4508042633533478, "learning_rate": 0.0005, "loss": 1.308042287826538, "step": 38838 }, { "epoch": 3.061805281828932, "grad_norm": 0.44367650151252747, "learning_rate": 0.0005, "loss": 1.318739414215088, "step": 38839 }, { "epoch": 3.0618841150965705, "grad_norm": 0.4468207061290741, "learning_rate": 0.0005, "loss": 1.2369530200958252, "step": 38840 }, { "epoch": 3.0619629483642097, "grad_norm": 0.4026443362236023, "learning_rate": 0.0005, "loss": 1.1938879489898682, "step": 38841 }, { "epoch": 3.062041781631849, "grad_norm": 0.5088368654251099, "learning_rate": 0.0005, "loss": 1.294037103652954, "step": 38842 }, { "epoch": 3.0621206148994875, "grad_norm": 0.4998241364955902, "learning_rate": 0.0005, "loss": 1.3544964790344238, "step": 38843 }, { "epoch": 3.0621994481671266, "grad_norm": 0.6453037858009338, "learning_rate": 0.0005, "loss": 1.1749578714370728, "step": 38844 }, { "epoch": 3.0622782814347653, "grad_norm": 0.3122086822986603, "learning_rate": 0.0005, "loss": 1.048264503479004, "step": 38845 }, { "epoch": 3.0623571147024045, "grad_norm": 0.4220406115055084, "learning_rate": 0.0005, "loss": 1.143620491027832, "step": 38846 }, { "epoch": 3.062435947970043, "grad_norm": 0.4161900579929352, "learning_rate": 0.0005, "loss": 1.2341924905776978, "step": 38847 }, { "epoch": 3.0625147812376823, "grad_norm": 0.49204784631729126, "learning_rate": 0.0005, "loss": 1.1811143159866333, "step": 38848 }, { "epoch": 3.0625936145053214, "grad_norm": 0.5498421788215637, "learning_rate": 0.0005, "loss": 1.3320417404174805, "step": 38849 }, { "epoch": 3.06267244777296, "grad_norm": 0.5998223423957825, "learning_rate": 0.0005, "loss": 1.249645471572876, "step": 38850 }, { "epoch": 3.0627512810405992, "grad_norm": 0.4793979823589325, "learning_rate": 0.0005, "loss": 1.2852591276168823, "step": 38851 }, { "epoch": 3.062830114308238, "grad_norm": 0.4899922013282776, "learning_rate": 0.0005, "loss": 1.329134464263916, "step": 38852 }, { "epoch": 3.062908947575877, "grad_norm": 0.46430325508117676, "learning_rate": 0.0005, "loss": 1.475912094116211, "step": 38853 }, { "epoch": 3.0629877808435158, "grad_norm": 0.5405142307281494, "learning_rate": 0.0005, "loss": 1.198137879371643, "step": 38854 }, { "epoch": 3.063066614111155, "grad_norm": 0.4378630220890045, "learning_rate": 0.0005, "loss": 1.3181061744689941, "step": 38855 }, { "epoch": 3.063145447378794, "grad_norm": 0.4137691557407379, "learning_rate": 0.0005, "loss": 1.145801067352295, "step": 38856 }, { "epoch": 3.0632242806464327, "grad_norm": 0.4707622528076172, "learning_rate": 0.0005, "loss": 1.1073602437973022, "step": 38857 }, { "epoch": 3.063303113914072, "grad_norm": 0.5563980340957642, "learning_rate": 0.0005, "loss": 1.2139620780944824, "step": 38858 }, { "epoch": 3.0633819471817105, "grad_norm": 0.8167272806167603, "learning_rate": 0.0005, "loss": 1.400341510772705, "step": 38859 }, { "epoch": 3.0634607804493497, "grad_norm": 0.4760298430919647, "learning_rate": 0.0005, "loss": 1.222090244293213, "step": 38860 }, { "epoch": 3.063539613716989, "grad_norm": 0.5400614142417908, "learning_rate": 0.0005, "loss": 1.2392382621765137, "step": 38861 }, { "epoch": 3.0636184469846275, "grad_norm": 0.4752773940563202, "learning_rate": 0.0005, "loss": 1.355592966079712, "step": 38862 }, { "epoch": 3.0636972802522666, "grad_norm": 0.544258177280426, "learning_rate": 0.0005, "loss": 1.2011966705322266, "step": 38863 }, { "epoch": 3.0637761135199053, "grad_norm": 0.535836398601532, "learning_rate": 0.0005, "loss": 1.3164703845977783, "step": 38864 }, { "epoch": 3.0638549467875444, "grad_norm": 0.6650379300117493, "learning_rate": 0.0005, "loss": 1.4979839324951172, "step": 38865 }, { "epoch": 3.063933780055183, "grad_norm": 0.43970611691474915, "learning_rate": 0.0005, "loss": 1.1160749197006226, "step": 38866 }, { "epoch": 3.0640126133228223, "grad_norm": 0.592138946056366, "learning_rate": 0.0005, "loss": 1.599548578262329, "step": 38867 }, { "epoch": 3.064091446590461, "grad_norm": 0.5797263383865356, "learning_rate": 0.0005, "loss": 1.2058758735656738, "step": 38868 }, { "epoch": 3.0641702798581, "grad_norm": 0.5739288926124573, "learning_rate": 0.0005, "loss": 1.0603010654449463, "step": 38869 }, { "epoch": 3.064249113125739, "grad_norm": 0.5733129978179932, "learning_rate": 0.0005, "loss": 1.3726850748062134, "step": 38870 }, { "epoch": 3.064327946393378, "grad_norm": 0.4387378990650177, "learning_rate": 0.0005, "loss": 1.1976163387298584, "step": 38871 }, { "epoch": 3.064406779661017, "grad_norm": 0.5995721220970154, "learning_rate": 0.0005, "loss": 1.3493297100067139, "step": 38872 }, { "epoch": 3.0644856129286557, "grad_norm": 0.5077950358390808, "learning_rate": 0.0005, "loss": 1.268999695777893, "step": 38873 }, { "epoch": 3.064564446196295, "grad_norm": 0.5050327777862549, "learning_rate": 0.0005, "loss": 1.2519493103027344, "step": 38874 }, { "epoch": 3.064643279463934, "grad_norm": 0.6175150871276855, "learning_rate": 0.0005, "loss": 1.2958955764770508, "step": 38875 }, { "epoch": 3.0647221127315727, "grad_norm": 0.503456175327301, "learning_rate": 0.0005, "loss": 1.3440134525299072, "step": 38876 }, { "epoch": 3.064800945999212, "grad_norm": 0.469325453042984, "learning_rate": 0.0005, "loss": 1.3113069534301758, "step": 38877 }, { "epoch": 3.0648797792668505, "grad_norm": 0.5100476145744324, "learning_rate": 0.0005, "loss": 1.2209230661392212, "step": 38878 }, { "epoch": 3.0649586125344896, "grad_norm": 0.6711524128913879, "learning_rate": 0.0005, "loss": 1.3012676239013672, "step": 38879 }, { "epoch": 3.0650374458021283, "grad_norm": 0.48681801557540894, "learning_rate": 0.0005, "loss": 1.2892004251480103, "step": 38880 }, { "epoch": 3.0651162790697675, "grad_norm": 0.5348912477493286, "learning_rate": 0.0005, "loss": 1.2730178833007812, "step": 38881 }, { "epoch": 3.065195112337406, "grad_norm": 0.48945072293281555, "learning_rate": 0.0005, "loss": 1.3750011920928955, "step": 38882 }, { "epoch": 3.0652739456050453, "grad_norm": 0.41338634490966797, "learning_rate": 0.0005, "loss": 1.3315069675445557, "step": 38883 }, { "epoch": 3.0653527788726844, "grad_norm": 0.44456884264945984, "learning_rate": 0.0005, "loss": 1.332697868347168, "step": 38884 }, { "epoch": 3.065431612140323, "grad_norm": 0.4989710748195648, "learning_rate": 0.0005, "loss": 1.2101038694381714, "step": 38885 }, { "epoch": 3.0655104454079622, "grad_norm": 0.5584672093391418, "learning_rate": 0.0005, "loss": 1.3824787139892578, "step": 38886 }, { "epoch": 3.065589278675601, "grad_norm": 0.5084114074707031, "learning_rate": 0.0005, "loss": 1.2548136711120605, "step": 38887 }, { "epoch": 3.06566811194324, "grad_norm": 0.4951121211051941, "learning_rate": 0.0005, "loss": 1.3033947944641113, "step": 38888 }, { "epoch": 3.065746945210879, "grad_norm": 0.4924905300140381, "learning_rate": 0.0005, "loss": 1.171302080154419, "step": 38889 }, { "epoch": 3.065825778478518, "grad_norm": 0.6031784415245056, "learning_rate": 0.0005, "loss": 1.5171763896942139, "step": 38890 }, { "epoch": 3.065904611746157, "grad_norm": 0.5660618543624878, "learning_rate": 0.0005, "loss": 1.3052752017974854, "step": 38891 }, { "epoch": 3.0659834450137957, "grad_norm": 0.6980299353599548, "learning_rate": 0.0005, "loss": 1.1226136684417725, "step": 38892 }, { "epoch": 3.066062278281435, "grad_norm": 0.6857502460479736, "learning_rate": 0.0005, "loss": 1.2083017826080322, "step": 38893 }, { "epoch": 3.0661411115490735, "grad_norm": 0.45727092027664185, "learning_rate": 0.0005, "loss": 1.257941722869873, "step": 38894 }, { "epoch": 3.0662199448167127, "grad_norm": 0.509003758430481, "learning_rate": 0.0005, "loss": 1.1160061359405518, "step": 38895 }, { "epoch": 3.066298778084352, "grad_norm": 0.45833489298820496, "learning_rate": 0.0005, "loss": 1.2260770797729492, "step": 38896 }, { "epoch": 3.0663776113519905, "grad_norm": 0.4630012512207031, "learning_rate": 0.0005, "loss": 1.2826166152954102, "step": 38897 }, { "epoch": 3.0664564446196296, "grad_norm": 0.40705040097236633, "learning_rate": 0.0005, "loss": 1.1919900178909302, "step": 38898 }, { "epoch": 3.0665352778872683, "grad_norm": 0.49798887968063354, "learning_rate": 0.0005, "loss": 1.2765254974365234, "step": 38899 }, { "epoch": 3.0666141111549075, "grad_norm": 0.5432992577552795, "learning_rate": 0.0005, "loss": 1.306082010269165, "step": 38900 }, { "epoch": 3.066692944422546, "grad_norm": 0.48949554562568665, "learning_rate": 0.0005, "loss": 1.2819864749908447, "step": 38901 }, { "epoch": 3.0667717776901853, "grad_norm": 0.430242121219635, "learning_rate": 0.0005, "loss": 1.1693882942199707, "step": 38902 }, { "epoch": 3.0668506109578244, "grad_norm": 0.4894055724143982, "learning_rate": 0.0005, "loss": 1.2685120105743408, "step": 38903 }, { "epoch": 3.066929444225463, "grad_norm": 0.36330974102020264, "learning_rate": 0.0005, "loss": 1.0343468189239502, "step": 38904 }, { "epoch": 3.0670082774931022, "grad_norm": 0.4422159790992737, "learning_rate": 0.0005, "loss": 1.265486478805542, "step": 38905 }, { "epoch": 3.067087110760741, "grad_norm": 0.4902992248535156, "learning_rate": 0.0005, "loss": 1.2573384046554565, "step": 38906 }, { "epoch": 3.06716594402838, "grad_norm": 0.45636844635009766, "learning_rate": 0.0005, "loss": 1.350192904472351, "step": 38907 }, { "epoch": 3.0672447772960187, "grad_norm": 0.4605485200881958, "learning_rate": 0.0005, "loss": 1.2763714790344238, "step": 38908 }, { "epoch": 3.067323610563658, "grad_norm": 0.5726245641708374, "learning_rate": 0.0005, "loss": 1.4412568807601929, "step": 38909 }, { "epoch": 3.067402443831297, "grad_norm": 0.5524728298187256, "learning_rate": 0.0005, "loss": 1.2411998510360718, "step": 38910 }, { "epoch": 3.0674812770989357, "grad_norm": 0.5415108799934387, "learning_rate": 0.0005, "loss": 1.2184315919876099, "step": 38911 }, { "epoch": 3.067560110366575, "grad_norm": 0.5963826179504395, "learning_rate": 0.0005, "loss": 1.4063019752502441, "step": 38912 }, { "epoch": 3.0676389436342135, "grad_norm": 0.4562038481235504, "learning_rate": 0.0005, "loss": 1.2339982986450195, "step": 38913 }, { "epoch": 3.0677177769018527, "grad_norm": 0.5260553359985352, "learning_rate": 0.0005, "loss": 1.0604277849197388, "step": 38914 }, { "epoch": 3.0677966101694913, "grad_norm": 0.48072540760040283, "learning_rate": 0.0005, "loss": 1.38563072681427, "step": 38915 }, { "epoch": 3.0678754434371305, "grad_norm": 0.4821900427341461, "learning_rate": 0.0005, "loss": 1.3220993280410767, "step": 38916 }, { "epoch": 3.0679542767047696, "grad_norm": 0.5222848057746887, "learning_rate": 0.0005, "loss": 1.1961575746536255, "step": 38917 }, { "epoch": 3.0680331099724083, "grad_norm": 0.5453267097473145, "learning_rate": 0.0005, "loss": 1.2488420009613037, "step": 38918 }, { "epoch": 3.0681119432400474, "grad_norm": 0.44864410161972046, "learning_rate": 0.0005, "loss": 0.9828225374221802, "step": 38919 }, { "epoch": 3.068190776507686, "grad_norm": 0.5962527394294739, "learning_rate": 0.0005, "loss": 1.19476318359375, "step": 38920 }, { "epoch": 3.0682696097753253, "grad_norm": 0.535286545753479, "learning_rate": 0.0005, "loss": 1.3539780378341675, "step": 38921 }, { "epoch": 3.068348443042964, "grad_norm": 0.474763959646225, "learning_rate": 0.0005, "loss": 1.20387864112854, "step": 38922 }, { "epoch": 3.068427276310603, "grad_norm": 0.47331368923187256, "learning_rate": 0.0005, "loss": 1.2582852840423584, "step": 38923 }, { "epoch": 3.068506109578242, "grad_norm": 0.5360419750213623, "learning_rate": 0.0005, "loss": 1.274154543876648, "step": 38924 }, { "epoch": 3.068584942845881, "grad_norm": 0.640243649482727, "learning_rate": 0.0005, "loss": 1.4253284931182861, "step": 38925 }, { "epoch": 3.06866377611352, "grad_norm": 0.5770308971405029, "learning_rate": 0.0005, "loss": 1.2489914894104004, "step": 38926 }, { "epoch": 3.0687426093811587, "grad_norm": 0.474679559469223, "learning_rate": 0.0005, "loss": 1.2757594585418701, "step": 38927 }, { "epoch": 3.068821442648798, "grad_norm": 0.5674602389335632, "learning_rate": 0.0005, "loss": 1.3955262899398804, "step": 38928 }, { "epoch": 3.0689002759164365, "grad_norm": 0.5167908072471619, "learning_rate": 0.0005, "loss": 1.2078449726104736, "step": 38929 }, { "epoch": 3.0689791091840757, "grad_norm": 0.5287320613861084, "learning_rate": 0.0005, "loss": 1.2099874019622803, "step": 38930 }, { "epoch": 3.069057942451715, "grad_norm": 0.42861199378967285, "learning_rate": 0.0005, "loss": 1.1586984395980835, "step": 38931 }, { "epoch": 3.0691367757193535, "grad_norm": 0.44671985507011414, "learning_rate": 0.0005, "loss": 1.2667880058288574, "step": 38932 }, { "epoch": 3.0692156089869926, "grad_norm": 0.47169196605682373, "learning_rate": 0.0005, "loss": 1.4700819253921509, "step": 38933 }, { "epoch": 3.0692944422546313, "grad_norm": 0.49417173862457275, "learning_rate": 0.0005, "loss": 1.2857511043548584, "step": 38934 }, { "epoch": 3.0693732755222705, "grad_norm": 0.6107233166694641, "learning_rate": 0.0005, "loss": 1.1596717834472656, "step": 38935 }, { "epoch": 3.069452108789909, "grad_norm": 0.44957008957862854, "learning_rate": 0.0005, "loss": 1.3197686672210693, "step": 38936 }, { "epoch": 3.0695309420575483, "grad_norm": 0.5398346781730652, "learning_rate": 0.0005, "loss": 1.4882018566131592, "step": 38937 }, { "epoch": 3.0696097753251874, "grad_norm": 0.6687116622924805, "learning_rate": 0.0005, "loss": 1.3290859460830688, "step": 38938 }, { "epoch": 3.069688608592826, "grad_norm": 0.41152095794677734, "learning_rate": 0.0005, "loss": 1.228734016418457, "step": 38939 }, { "epoch": 3.0697674418604652, "grad_norm": 0.42387571930885315, "learning_rate": 0.0005, "loss": 1.2360200881958008, "step": 38940 }, { "epoch": 3.069846275128104, "grad_norm": 0.5036365985870361, "learning_rate": 0.0005, "loss": 1.3146251440048218, "step": 38941 }, { "epoch": 3.069925108395743, "grad_norm": 0.4048135280609131, "learning_rate": 0.0005, "loss": 1.0955266952514648, "step": 38942 }, { "epoch": 3.0700039416633818, "grad_norm": 0.5288954377174377, "learning_rate": 0.0005, "loss": 1.320920705795288, "step": 38943 }, { "epoch": 3.070082774931021, "grad_norm": 0.5189542770385742, "learning_rate": 0.0005, "loss": 1.1687313318252563, "step": 38944 }, { "epoch": 3.07016160819866, "grad_norm": 0.5000877976417542, "learning_rate": 0.0005, "loss": 1.2725050449371338, "step": 38945 }, { "epoch": 3.0702404414662987, "grad_norm": 0.5476392507553101, "learning_rate": 0.0005, "loss": 1.3381245136260986, "step": 38946 }, { "epoch": 3.070319274733938, "grad_norm": 0.9024911522865295, "learning_rate": 0.0005, "loss": 1.3085579872131348, "step": 38947 }, { "epoch": 3.0703981080015765, "grad_norm": 0.39536064863204956, "learning_rate": 0.0005, "loss": 1.1941337585449219, "step": 38948 }, { "epoch": 3.0704769412692157, "grad_norm": 0.44566282629966736, "learning_rate": 0.0005, "loss": 1.1517119407653809, "step": 38949 }, { "epoch": 3.0705557745368544, "grad_norm": 0.41293811798095703, "learning_rate": 0.0005, "loss": 1.106116533279419, "step": 38950 }, { "epoch": 3.0706346078044935, "grad_norm": 0.4244997203350067, "learning_rate": 0.0005, "loss": 1.207923173904419, "step": 38951 }, { "epoch": 3.0707134410721326, "grad_norm": 0.4765937924385071, "learning_rate": 0.0005, "loss": 1.3089934587478638, "step": 38952 }, { "epoch": 3.0707922743397713, "grad_norm": 0.45452043414115906, "learning_rate": 0.0005, "loss": 1.2454965114593506, "step": 38953 }, { "epoch": 3.0708711076074104, "grad_norm": 0.43575069308280945, "learning_rate": 0.0005, "loss": 1.203067421913147, "step": 38954 }, { "epoch": 3.070949940875049, "grad_norm": 0.46356967091560364, "learning_rate": 0.0005, "loss": 1.3685201406478882, "step": 38955 }, { "epoch": 3.0710287741426883, "grad_norm": 0.6113157868385315, "learning_rate": 0.0005, "loss": 1.1288771629333496, "step": 38956 }, { "epoch": 3.071107607410327, "grad_norm": 0.49220919609069824, "learning_rate": 0.0005, "loss": 1.0369547605514526, "step": 38957 }, { "epoch": 3.071186440677966, "grad_norm": 0.43556591868400574, "learning_rate": 0.0005, "loss": 1.2628169059753418, "step": 38958 }, { "epoch": 3.0712652739456052, "grad_norm": 0.4803711473941803, "learning_rate": 0.0005, "loss": 1.2448303699493408, "step": 38959 }, { "epoch": 3.071344107213244, "grad_norm": 0.8871301412582397, "learning_rate": 0.0005, "loss": 1.3178154230117798, "step": 38960 }, { "epoch": 3.071422940480883, "grad_norm": 0.6376895308494568, "learning_rate": 0.0005, "loss": 1.2720046043395996, "step": 38961 }, { "epoch": 3.0715017737485217, "grad_norm": 0.4712105989456177, "learning_rate": 0.0005, "loss": 1.3556482791900635, "step": 38962 }, { "epoch": 3.071580607016161, "grad_norm": 0.47252213954925537, "learning_rate": 0.0005, "loss": 1.1602509021759033, "step": 38963 }, { "epoch": 3.0716594402837996, "grad_norm": 0.5260292887687683, "learning_rate": 0.0005, "loss": 1.2787004709243774, "step": 38964 }, { "epoch": 3.0717382735514387, "grad_norm": 0.5030308365821838, "learning_rate": 0.0005, "loss": 1.2570133209228516, "step": 38965 }, { "epoch": 3.071817106819078, "grad_norm": 0.5795015096664429, "learning_rate": 0.0005, "loss": 1.2364025115966797, "step": 38966 }, { "epoch": 3.0718959400867165, "grad_norm": 0.48831313848495483, "learning_rate": 0.0005, "loss": 1.1603645086288452, "step": 38967 }, { "epoch": 3.0719747733543556, "grad_norm": 0.43787387013435364, "learning_rate": 0.0005, "loss": 1.318669080734253, "step": 38968 }, { "epoch": 3.0720536066219943, "grad_norm": 0.3890395164489746, "learning_rate": 0.0005, "loss": 1.235795021057129, "step": 38969 }, { "epoch": 3.0721324398896335, "grad_norm": 0.44003158807754517, "learning_rate": 0.0005, "loss": 1.136338710784912, "step": 38970 }, { "epoch": 3.072211273157272, "grad_norm": 0.5593410730361938, "learning_rate": 0.0005, "loss": 1.2839241027832031, "step": 38971 }, { "epoch": 3.0722901064249113, "grad_norm": 0.4898250997066498, "learning_rate": 0.0005, "loss": 1.207939863204956, "step": 38972 }, { "epoch": 3.0723689396925504, "grad_norm": 0.5321543216705322, "learning_rate": 0.0005, "loss": 1.322059988975525, "step": 38973 }, { "epoch": 3.072447772960189, "grad_norm": 0.5638111233711243, "learning_rate": 0.0005, "loss": 1.2848608493804932, "step": 38974 }, { "epoch": 3.0725266062278282, "grad_norm": 0.4156438708305359, "learning_rate": 0.0005, "loss": 1.1324162483215332, "step": 38975 }, { "epoch": 3.072605439495467, "grad_norm": 0.6203200817108154, "learning_rate": 0.0005, "loss": 1.2250449657440186, "step": 38976 }, { "epoch": 3.072684272763106, "grad_norm": 0.37291449308395386, "learning_rate": 0.0005, "loss": 1.0508487224578857, "step": 38977 }, { "epoch": 3.0727631060307448, "grad_norm": 0.5248454213142395, "learning_rate": 0.0005, "loss": 1.175584077835083, "step": 38978 }, { "epoch": 3.072841939298384, "grad_norm": 0.5724268555641174, "learning_rate": 0.0005, "loss": 1.342209815979004, "step": 38979 }, { "epoch": 3.072920772566023, "grad_norm": 0.43511900305747986, "learning_rate": 0.0005, "loss": 1.1622154712677002, "step": 38980 }, { "epoch": 3.0729996058336617, "grad_norm": 0.4488365352153778, "learning_rate": 0.0005, "loss": 1.0647916793823242, "step": 38981 }, { "epoch": 3.073078439101301, "grad_norm": 0.4840516448020935, "learning_rate": 0.0005, "loss": 1.3802553415298462, "step": 38982 }, { "epoch": 3.0731572723689395, "grad_norm": 1.002147912979126, "learning_rate": 0.0005, "loss": 1.275376319885254, "step": 38983 }, { "epoch": 3.0732361056365787, "grad_norm": 0.3842715322971344, "learning_rate": 0.0005, "loss": 1.0941030979156494, "step": 38984 }, { "epoch": 3.0733149389042174, "grad_norm": 0.4146856963634491, "learning_rate": 0.0005, "loss": 1.157435655593872, "step": 38985 }, { "epoch": 3.0733937721718565, "grad_norm": 0.43904128670692444, "learning_rate": 0.0005, "loss": 1.3831803798675537, "step": 38986 }, { "epoch": 3.0734726054394956, "grad_norm": 0.4981675148010254, "learning_rate": 0.0005, "loss": 1.145566463470459, "step": 38987 }, { "epoch": 3.0735514387071343, "grad_norm": 0.5629326701164246, "learning_rate": 0.0005, "loss": 1.2157506942749023, "step": 38988 }, { "epoch": 3.0736302719747735, "grad_norm": 0.513113796710968, "learning_rate": 0.0005, "loss": 1.0712215900421143, "step": 38989 }, { "epoch": 3.073709105242412, "grad_norm": 0.5197942852973938, "learning_rate": 0.0005, "loss": 1.3030767440795898, "step": 38990 }, { "epoch": 3.0737879385100513, "grad_norm": 0.544543445110321, "learning_rate": 0.0005, "loss": 1.34712815284729, "step": 38991 }, { "epoch": 3.0738667717776904, "grad_norm": 0.4690501093864441, "learning_rate": 0.0005, "loss": 1.1346087455749512, "step": 38992 }, { "epoch": 3.073945605045329, "grad_norm": 2.916191339492798, "learning_rate": 0.0005, "loss": 1.2500873804092407, "step": 38993 }, { "epoch": 3.0740244383129682, "grad_norm": 0.4043516218662262, "learning_rate": 0.0005, "loss": 1.1597809791564941, "step": 38994 }, { "epoch": 3.074103271580607, "grad_norm": 0.5660451650619507, "learning_rate": 0.0005, "loss": 1.3070931434631348, "step": 38995 }, { "epoch": 3.074182104848246, "grad_norm": 0.4882361590862274, "learning_rate": 0.0005, "loss": 1.1555811166763306, "step": 38996 }, { "epoch": 3.0742609381158847, "grad_norm": 0.3862645626068115, "learning_rate": 0.0005, "loss": 1.094672441482544, "step": 38997 }, { "epoch": 3.074339771383524, "grad_norm": 0.4632544219493866, "learning_rate": 0.0005, "loss": 1.184504747390747, "step": 38998 }, { "epoch": 3.0744186046511626, "grad_norm": 0.5243950486183167, "learning_rate": 0.0005, "loss": 1.40315842628479, "step": 38999 }, { "epoch": 3.0744974379188017, "grad_norm": 0.5933886170387268, "learning_rate": 0.0005, "loss": 1.266621708869934, "step": 39000 }, { "epoch": 3.074576271186441, "grad_norm": 0.5575701594352722, "learning_rate": 0.0005, "loss": 1.4283006191253662, "step": 39001 }, { "epoch": 3.0746551044540795, "grad_norm": 0.4939846694469452, "learning_rate": 0.0005, "loss": 1.333122730255127, "step": 39002 }, { "epoch": 3.0747339377217187, "grad_norm": 0.40543097257614136, "learning_rate": 0.0005, "loss": 1.0318162441253662, "step": 39003 }, { "epoch": 3.0748127709893573, "grad_norm": 0.5103802680969238, "learning_rate": 0.0005, "loss": 1.322295904159546, "step": 39004 }, { "epoch": 3.0748916042569965, "grad_norm": 0.5763012170791626, "learning_rate": 0.0005, "loss": 1.287355661392212, "step": 39005 }, { "epoch": 3.0749704375246356, "grad_norm": 0.4249032735824585, "learning_rate": 0.0005, "loss": 1.2032372951507568, "step": 39006 }, { "epoch": 3.0750492707922743, "grad_norm": 0.5066051483154297, "learning_rate": 0.0005, "loss": 1.3808085918426514, "step": 39007 }, { "epoch": 3.0751281040599134, "grad_norm": 0.48219814896583557, "learning_rate": 0.0005, "loss": 1.3102256059646606, "step": 39008 }, { "epoch": 3.075206937327552, "grad_norm": 0.4534786343574524, "learning_rate": 0.0005, "loss": 1.318249225616455, "step": 39009 }, { "epoch": 3.0752857705951913, "grad_norm": 0.4126255214214325, "learning_rate": 0.0005, "loss": 1.2190768718719482, "step": 39010 }, { "epoch": 3.07536460386283, "grad_norm": 0.4516594707965851, "learning_rate": 0.0005, "loss": 1.2230398654937744, "step": 39011 }, { "epoch": 3.075443437130469, "grad_norm": 0.4260479211807251, "learning_rate": 0.0005, "loss": 1.2864325046539307, "step": 39012 }, { "epoch": 3.075522270398108, "grad_norm": 0.4935184717178345, "learning_rate": 0.0005, "loss": 1.3254491090774536, "step": 39013 }, { "epoch": 3.075601103665747, "grad_norm": 0.4328763782978058, "learning_rate": 0.0005, "loss": 1.216174840927124, "step": 39014 }, { "epoch": 3.075679936933386, "grad_norm": 0.47927770018577576, "learning_rate": 0.0005, "loss": 1.431679129600525, "step": 39015 }, { "epoch": 3.0757587702010247, "grad_norm": 0.4809252619743347, "learning_rate": 0.0005, "loss": 1.3678048849105835, "step": 39016 }, { "epoch": 3.075837603468664, "grad_norm": 0.4539242684841156, "learning_rate": 0.0005, "loss": 1.0491286516189575, "step": 39017 }, { "epoch": 3.0759164367363026, "grad_norm": 0.4457651674747467, "learning_rate": 0.0005, "loss": 1.0687354803085327, "step": 39018 }, { "epoch": 3.0759952700039417, "grad_norm": 0.4199627935886383, "learning_rate": 0.0005, "loss": 1.0770004987716675, "step": 39019 }, { "epoch": 3.076074103271581, "grad_norm": 0.38289469480514526, "learning_rate": 0.0005, "loss": 1.2662572860717773, "step": 39020 }, { "epoch": 3.0761529365392195, "grad_norm": 0.5174855589866638, "learning_rate": 0.0005, "loss": 1.112457275390625, "step": 39021 }, { "epoch": 3.0762317698068586, "grad_norm": 0.4185596704483032, "learning_rate": 0.0005, "loss": 1.2718861103057861, "step": 39022 }, { "epoch": 3.0763106030744973, "grad_norm": 0.4035126566886902, "learning_rate": 0.0005, "loss": 1.0934977531433105, "step": 39023 }, { "epoch": 3.0763894363421365, "grad_norm": 0.4062547981739044, "learning_rate": 0.0005, "loss": 1.1865530014038086, "step": 39024 }, { "epoch": 3.076468269609775, "grad_norm": 0.4513052999973297, "learning_rate": 0.0005, "loss": 1.200976848602295, "step": 39025 }, { "epoch": 3.0765471028774143, "grad_norm": 0.455849289894104, "learning_rate": 0.0005, "loss": 1.2812011241912842, "step": 39026 }, { "epoch": 3.0766259361450534, "grad_norm": 0.4097491204738617, "learning_rate": 0.0005, "loss": 1.1542657613754272, "step": 39027 }, { "epoch": 3.076704769412692, "grad_norm": 0.4671362340450287, "learning_rate": 0.0005, "loss": 1.1749881505966187, "step": 39028 }, { "epoch": 3.0767836026803312, "grad_norm": 0.7002843618392944, "learning_rate": 0.0005, "loss": 1.261536717414856, "step": 39029 }, { "epoch": 3.07686243594797, "grad_norm": 0.46731144189834595, "learning_rate": 0.0005, "loss": 1.279939889907837, "step": 39030 }, { "epoch": 3.076941269215609, "grad_norm": 0.7126699686050415, "learning_rate": 0.0005, "loss": 1.2069158554077148, "step": 39031 }, { "epoch": 3.0770201024832478, "grad_norm": 0.4997061789035797, "learning_rate": 0.0005, "loss": 1.3803741931915283, "step": 39032 }, { "epoch": 3.077098935750887, "grad_norm": 0.4279365837574005, "learning_rate": 0.0005, "loss": 1.3035261631011963, "step": 39033 }, { "epoch": 3.077177769018526, "grad_norm": 0.54860919713974, "learning_rate": 0.0005, "loss": 1.2648249864578247, "step": 39034 }, { "epoch": 3.0772566022861647, "grad_norm": 0.49216458201408386, "learning_rate": 0.0005, "loss": 1.2206255197525024, "step": 39035 }, { "epoch": 3.077335435553804, "grad_norm": 0.5499553680419922, "learning_rate": 0.0005, "loss": 1.1685545444488525, "step": 39036 }, { "epoch": 3.0774142688214425, "grad_norm": 0.43930619955062866, "learning_rate": 0.0005, "loss": 1.1451219320297241, "step": 39037 }, { "epoch": 3.0774931020890817, "grad_norm": 0.6683289408683777, "learning_rate": 0.0005, "loss": 1.1592137813568115, "step": 39038 }, { "epoch": 3.0775719353567204, "grad_norm": 0.5531504154205322, "learning_rate": 0.0005, "loss": 1.273473858833313, "step": 39039 }, { "epoch": 3.0776507686243595, "grad_norm": 0.4680805206298828, "learning_rate": 0.0005, "loss": 1.1486408710479736, "step": 39040 }, { "epoch": 3.0777296018919986, "grad_norm": 0.9963638186454773, "learning_rate": 0.0005, "loss": 1.370449185371399, "step": 39041 }, { "epoch": 3.0778084351596373, "grad_norm": 0.40156829357147217, "learning_rate": 0.0005, "loss": 1.1959539651870728, "step": 39042 }, { "epoch": 3.0778872684272764, "grad_norm": 0.48981723189353943, "learning_rate": 0.0005, "loss": 1.174532175064087, "step": 39043 }, { "epoch": 3.077966101694915, "grad_norm": 0.4919556677341461, "learning_rate": 0.0005, "loss": 1.31569504737854, "step": 39044 }, { "epoch": 3.0780449349625543, "grad_norm": 0.4736132323741913, "learning_rate": 0.0005, "loss": 1.3559980392456055, "step": 39045 }, { "epoch": 3.078123768230193, "grad_norm": 0.5320759415626526, "learning_rate": 0.0005, "loss": 1.3028825521469116, "step": 39046 }, { "epoch": 3.078202601497832, "grad_norm": 0.5211312770843506, "learning_rate": 0.0005, "loss": 1.4056854248046875, "step": 39047 }, { "epoch": 3.0782814347654712, "grad_norm": 0.46902942657470703, "learning_rate": 0.0005, "loss": 1.2555099725723267, "step": 39048 }, { "epoch": 3.07836026803311, "grad_norm": 0.43567290902137756, "learning_rate": 0.0005, "loss": 1.1513755321502686, "step": 39049 }, { "epoch": 3.078439101300749, "grad_norm": 0.4414464235305786, "learning_rate": 0.0005, "loss": 1.173425316810608, "step": 39050 }, { "epoch": 3.0785179345683877, "grad_norm": 0.543310821056366, "learning_rate": 0.0005, "loss": 1.3246651887893677, "step": 39051 }, { "epoch": 3.078596767836027, "grad_norm": 0.5495723485946655, "learning_rate": 0.0005, "loss": 1.1144438982009888, "step": 39052 }, { "epoch": 3.0786756011036656, "grad_norm": 0.7115514874458313, "learning_rate": 0.0005, "loss": 1.2525475025177002, "step": 39053 }, { "epoch": 3.0787544343713047, "grad_norm": 0.49594512581825256, "learning_rate": 0.0005, "loss": 1.1994953155517578, "step": 39054 }, { "epoch": 3.078833267638944, "grad_norm": 0.4241783618927002, "learning_rate": 0.0005, "loss": 1.360062599182129, "step": 39055 }, { "epoch": 3.0789121009065825, "grad_norm": 0.39064842462539673, "learning_rate": 0.0005, "loss": 1.0377109050750732, "step": 39056 }, { "epoch": 3.0789909341742216, "grad_norm": 0.459053099155426, "learning_rate": 0.0005, "loss": 1.202726125717163, "step": 39057 }, { "epoch": 3.0790697674418603, "grad_norm": 0.44700121879577637, "learning_rate": 0.0005, "loss": 1.2058017253875732, "step": 39058 }, { "epoch": 3.0791486007094995, "grad_norm": 0.47317320108413696, "learning_rate": 0.0005, "loss": 1.3632667064666748, "step": 39059 }, { "epoch": 3.079227433977138, "grad_norm": 0.4205184876918793, "learning_rate": 0.0005, "loss": 1.2246414422988892, "step": 39060 }, { "epoch": 3.0793062672447773, "grad_norm": 0.5461103320121765, "learning_rate": 0.0005, "loss": 1.3774510622024536, "step": 39061 }, { "epoch": 3.0793851005124164, "grad_norm": 0.5257154107093811, "learning_rate": 0.0005, "loss": 1.2634351253509521, "step": 39062 }, { "epoch": 3.079463933780055, "grad_norm": 0.49543440341949463, "learning_rate": 0.0005, "loss": 1.0896024703979492, "step": 39063 }, { "epoch": 3.0795427670476943, "grad_norm": 0.5657486319541931, "learning_rate": 0.0005, "loss": 1.3740830421447754, "step": 39064 }, { "epoch": 3.079621600315333, "grad_norm": 0.5336961150169373, "learning_rate": 0.0005, "loss": 1.2726351022720337, "step": 39065 }, { "epoch": 3.079700433582972, "grad_norm": 0.455023854970932, "learning_rate": 0.0005, "loss": 1.408595085144043, "step": 39066 }, { "epoch": 3.0797792668506108, "grad_norm": 0.7779815196990967, "learning_rate": 0.0005, "loss": 1.360599160194397, "step": 39067 }, { "epoch": 3.07985810011825, "grad_norm": 0.5147993564605713, "learning_rate": 0.0005, "loss": 1.1396745443344116, "step": 39068 }, { "epoch": 3.079936933385889, "grad_norm": 0.43374884128570557, "learning_rate": 0.0005, "loss": 1.0794376134872437, "step": 39069 }, { "epoch": 3.0800157666535277, "grad_norm": 0.5248311758041382, "learning_rate": 0.0005, "loss": 1.402733564376831, "step": 39070 }, { "epoch": 3.080094599921167, "grad_norm": 0.5400958061218262, "learning_rate": 0.0005, "loss": 1.3343485593795776, "step": 39071 }, { "epoch": 3.0801734331888055, "grad_norm": 0.4302334487438202, "learning_rate": 0.0005, "loss": 1.2616256475448608, "step": 39072 }, { "epoch": 3.0802522664564447, "grad_norm": 0.5207093358039856, "learning_rate": 0.0005, "loss": 1.6193501949310303, "step": 39073 }, { "epoch": 3.0803310997240834, "grad_norm": 0.45357850193977356, "learning_rate": 0.0005, "loss": 1.1847684383392334, "step": 39074 }, { "epoch": 3.0804099329917225, "grad_norm": 0.48208507895469666, "learning_rate": 0.0005, "loss": 1.267958402633667, "step": 39075 }, { "epoch": 3.0804887662593616, "grad_norm": 0.5476893186569214, "learning_rate": 0.0005, "loss": 1.4152384996414185, "step": 39076 }, { "epoch": 3.0805675995270003, "grad_norm": 0.5019116401672363, "learning_rate": 0.0005, "loss": 1.2000083923339844, "step": 39077 }, { "epoch": 3.0806464327946395, "grad_norm": 0.4520590305328369, "learning_rate": 0.0005, "loss": 1.1367518901824951, "step": 39078 }, { "epoch": 3.080725266062278, "grad_norm": 0.5238379240036011, "learning_rate": 0.0005, "loss": 1.2228899002075195, "step": 39079 }, { "epoch": 3.0808040993299173, "grad_norm": 0.4746509790420532, "learning_rate": 0.0005, "loss": 1.1441419124603271, "step": 39080 }, { "epoch": 3.080882932597556, "grad_norm": 0.4228378236293793, "learning_rate": 0.0005, "loss": 1.293948769569397, "step": 39081 }, { "epoch": 3.080961765865195, "grad_norm": 0.49568691849708557, "learning_rate": 0.0005, "loss": 1.3739876747131348, "step": 39082 }, { "epoch": 3.0810405991328342, "grad_norm": 0.5239235758781433, "learning_rate": 0.0005, "loss": 1.5280334949493408, "step": 39083 }, { "epoch": 3.081119432400473, "grad_norm": 0.47822004556655884, "learning_rate": 0.0005, "loss": 1.2057604789733887, "step": 39084 }, { "epoch": 3.081198265668112, "grad_norm": 0.5651688575744629, "learning_rate": 0.0005, "loss": 1.1633577346801758, "step": 39085 }, { "epoch": 3.0812770989357507, "grad_norm": 0.4126422107219696, "learning_rate": 0.0005, "loss": 1.124657392501831, "step": 39086 }, { "epoch": 3.08135593220339, "grad_norm": 0.49199172854423523, "learning_rate": 0.0005, "loss": 1.327793002128601, "step": 39087 }, { "epoch": 3.0814347654710286, "grad_norm": 0.5016091465950012, "learning_rate": 0.0005, "loss": 1.3926043510437012, "step": 39088 }, { "epoch": 3.0815135987386677, "grad_norm": 0.48021334409713745, "learning_rate": 0.0005, "loss": 1.0998892784118652, "step": 39089 }, { "epoch": 3.081592432006307, "grad_norm": 0.5055745840072632, "learning_rate": 0.0005, "loss": 1.242887020111084, "step": 39090 }, { "epoch": 3.0816712652739455, "grad_norm": 0.5312135219573975, "learning_rate": 0.0005, "loss": 1.1529779434204102, "step": 39091 }, { "epoch": 3.0817500985415847, "grad_norm": 0.5092077255249023, "learning_rate": 0.0005, "loss": 1.343877911567688, "step": 39092 }, { "epoch": 3.0818289318092233, "grad_norm": 0.4161212742328644, "learning_rate": 0.0005, "loss": 1.2209522724151611, "step": 39093 }, { "epoch": 3.0819077650768625, "grad_norm": 0.6023914217948914, "learning_rate": 0.0005, "loss": 1.2949615716934204, "step": 39094 }, { "epoch": 3.081986598344501, "grad_norm": 0.630893886089325, "learning_rate": 0.0005, "loss": 1.400458812713623, "step": 39095 }, { "epoch": 3.0820654316121403, "grad_norm": 0.5009772181510925, "learning_rate": 0.0005, "loss": 1.312385082244873, "step": 39096 }, { "epoch": 3.0821442648797794, "grad_norm": 0.42358124256134033, "learning_rate": 0.0005, "loss": 1.2960819005966187, "step": 39097 }, { "epoch": 3.082223098147418, "grad_norm": 0.562606155872345, "learning_rate": 0.0005, "loss": 1.3114060163497925, "step": 39098 }, { "epoch": 3.0823019314150573, "grad_norm": 0.4636075496673584, "learning_rate": 0.0005, "loss": 1.3409414291381836, "step": 39099 }, { "epoch": 3.082380764682696, "grad_norm": 0.36706870794296265, "learning_rate": 0.0005, "loss": 1.0014344453811646, "step": 39100 }, { "epoch": 3.082459597950335, "grad_norm": 0.5436671376228333, "learning_rate": 0.0005, "loss": 1.3998591899871826, "step": 39101 }, { "epoch": 3.0825384312179738, "grad_norm": 0.518330991268158, "learning_rate": 0.0005, "loss": 1.276153326034546, "step": 39102 }, { "epoch": 3.082617264485613, "grad_norm": 0.4653908312320709, "learning_rate": 0.0005, "loss": 1.1406806707382202, "step": 39103 }, { "epoch": 3.082696097753252, "grad_norm": 0.46991440653800964, "learning_rate": 0.0005, "loss": 1.2461366653442383, "step": 39104 }, { "epoch": 3.0827749310208907, "grad_norm": 0.462623655796051, "learning_rate": 0.0005, "loss": 1.408820390701294, "step": 39105 }, { "epoch": 3.08285376428853, "grad_norm": 0.39993155002593994, "learning_rate": 0.0005, "loss": 1.1043699979782104, "step": 39106 }, { "epoch": 3.0829325975561686, "grad_norm": 0.471920371055603, "learning_rate": 0.0005, "loss": 1.17833411693573, "step": 39107 }, { "epoch": 3.0830114308238077, "grad_norm": 0.5956815481185913, "learning_rate": 0.0005, "loss": 1.4397906064987183, "step": 39108 }, { "epoch": 3.083090264091447, "grad_norm": 0.5538866519927979, "learning_rate": 0.0005, "loss": 1.0060977935791016, "step": 39109 }, { "epoch": 3.0831690973590855, "grad_norm": 0.4886196255683899, "learning_rate": 0.0005, "loss": 1.2271122932434082, "step": 39110 }, { "epoch": 3.0832479306267246, "grad_norm": 0.5052747130393982, "learning_rate": 0.0005, "loss": 1.0814118385314941, "step": 39111 }, { "epoch": 3.0833267638943633, "grad_norm": 0.5669506192207336, "learning_rate": 0.0005, "loss": 1.3553940057754517, "step": 39112 }, { "epoch": 3.0834055971620025, "grad_norm": 0.4515964388847351, "learning_rate": 0.0005, "loss": 1.1294691562652588, "step": 39113 }, { "epoch": 3.083484430429641, "grad_norm": 0.4369240999221802, "learning_rate": 0.0005, "loss": 1.3179001808166504, "step": 39114 }, { "epoch": 3.0835632636972803, "grad_norm": 0.4792883098125458, "learning_rate": 0.0005, "loss": 1.3195433616638184, "step": 39115 }, { "epoch": 3.083642096964919, "grad_norm": 0.5816419720649719, "learning_rate": 0.0005, "loss": 1.2814496755599976, "step": 39116 }, { "epoch": 3.083720930232558, "grad_norm": 0.7918829917907715, "learning_rate": 0.0005, "loss": 1.3310422897338867, "step": 39117 }, { "epoch": 3.0837997635001972, "grad_norm": 0.4412184953689575, "learning_rate": 0.0005, "loss": 0.9954885244369507, "step": 39118 }, { "epoch": 3.083878596767836, "grad_norm": 0.7199451327323914, "learning_rate": 0.0005, "loss": 1.3542430400848389, "step": 39119 }, { "epoch": 3.083957430035475, "grad_norm": 0.5593336224555969, "learning_rate": 0.0005, "loss": 1.1648008823394775, "step": 39120 }, { "epoch": 3.0840362633031138, "grad_norm": 0.6256769299507141, "learning_rate": 0.0005, "loss": 1.278348445892334, "step": 39121 }, { "epoch": 3.084115096570753, "grad_norm": 0.48045679926872253, "learning_rate": 0.0005, "loss": 1.0304155349731445, "step": 39122 }, { "epoch": 3.084193929838392, "grad_norm": 0.5294955968856812, "learning_rate": 0.0005, "loss": 1.2536015510559082, "step": 39123 }, { "epoch": 3.0842727631060307, "grad_norm": 0.6265039443969727, "learning_rate": 0.0005, "loss": 1.3297812938690186, "step": 39124 }, { "epoch": 3.08435159637367, "grad_norm": 0.47389960289001465, "learning_rate": 0.0005, "loss": 1.207063913345337, "step": 39125 }, { "epoch": 3.0844304296413085, "grad_norm": 0.6618561148643494, "learning_rate": 0.0005, "loss": 1.6160558462142944, "step": 39126 }, { "epoch": 3.0845092629089477, "grad_norm": 0.6887626051902771, "learning_rate": 0.0005, "loss": 1.382063627243042, "step": 39127 }, { "epoch": 3.0845880961765864, "grad_norm": 0.5575161576271057, "learning_rate": 0.0005, "loss": 1.33561372756958, "step": 39128 }, { "epoch": 3.0846669294442255, "grad_norm": 0.5069162845611572, "learning_rate": 0.0005, "loss": 1.2413177490234375, "step": 39129 }, { "epoch": 3.084745762711864, "grad_norm": 0.487873911857605, "learning_rate": 0.0005, "loss": 1.2862542867660522, "step": 39130 }, { "epoch": 3.0848245959795033, "grad_norm": 0.5258761048316956, "learning_rate": 0.0005, "loss": 1.1557538509368896, "step": 39131 }, { "epoch": 3.0849034292471424, "grad_norm": 0.4437408745288849, "learning_rate": 0.0005, "loss": 1.2336454391479492, "step": 39132 }, { "epoch": 3.084982262514781, "grad_norm": 0.47319793701171875, "learning_rate": 0.0005, "loss": 1.2967395782470703, "step": 39133 }, { "epoch": 3.0850610957824203, "grad_norm": 0.5181519389152527, "learning_rate": 0.0005, "loss": 1.3830047845840454, "step": 39134 }, { "epoch": 3.085139929050059, "grad_norm": 0.6086652278900146, "learning_rate": 0.0005, "loss": 1.418327808380127, "step": 39135 }, { "epoch": 3.085218762317698, "grad_norm": 0.4783444404602051, "learning_rate": 0.0005, "loss": 1.3021039962768555, "step": 39136 }, { "epoch": 3.0852975955853372, "grad_norm": 0.45120224356651306, "learning_rate": 0.0005, "loss": 1.3170197010040283, "step": 39137 }, { "epoch": 3.085376428852976, "grad_norm": 0.5003499984741211, "learning_rate": 0.0005, "loss": 1.3477206230163574, "step": 39138 }, { "epoch": 3.085455262120615, "grad_norm": 0.44193771481513977, "learning_rate": 0.0005, "loss": 1.1700754165649414, "step": 39139 }, { "epoch": 3.0855340953882537, "grad_norm": 0.5807586908340454, "learning_rate": 0.0005, "loss": 1.346817135810852, "step": 39140 }, { "epoch": 3.085612928655893, "grad_norm": 0.3653697967529297, "learning_rate": 0.0005, "loss": 1.076068639755249, "step": 39141 }, { "epoch": 3.0856917619235316, "grad_norm": 0.47021472454071045, "learning_rate": 0.0005, "loss": 1.2095983028411865, "step": 39142 }, { "epoch": 3.0857705951911707, "grad_norm": 0.5363097786903381, "learning_rate": 0.0005, "loss": 1.327610731124878, "step": 39143 }, { "epoch": 3.08584942845881, "grad_norm": 0.42670202255249023, "learning_rate": 0.0005, "loss": 1.1809148788452148, "step": 39144 }, { "epoch": 3.0859282617264485, "grad_norm": 0.48897573351860046, "learning_rate": 0.0005, "loss": 1.3657803535461426, "step": 39145 }, { "epoch": 3.0860070949940877, "grad_norm": 0.3753773868083954, "learning_rate": 0.0005, "loss": 1.124521255493164, "step": 39146 }, { "epoch": 3.0860859282617263, "grad_norm": 0.6046682596206665, "learning_rate": 0.0005, "loss": 1.4498465061187744, "step": 39147 }, { "epoch": 3.0861647615293655, "grad_norm": 0.38576775789260864, "learning_rate": 0.0005, "loss": 1.17703378200531, "step": 39148 }, { "epoch": 3.086243594797004, "grad_norm": 0.48967015743255615, "learning_rate": 0.0005, "loss": 1.11942720413208, "step": 39149 }, { "epoch": 3.0863224280646433, "grad_norm": 0.6273025274276733, "learning_rate": 0.0005, "loss": 1.4615552425384521, "step": 39150 }, { "epoch": 3.0864012613322824, "grad_norm": 0.485236793756485, "learning_rate": 0.0005, "loss": 1.2750895023345947, "step": 39151 }, { "epoch": 3.086480094599921, "grad_norm": 0.5776492357254028, "learning_rate": 0.0005, "loss": 1.582087516784668, "step": 39152 }, { "epoch": 3.0865589278675603, "grad_norm": 0.5523198246955872, "learning_rate": 0.0005, "loss": 1.206064224243164, "step": 39153 }, { "epoch": 3.086637761135199, "grad_norm": 0.4144233763217926, "learning_rate": 0.0005, "loss": 1.2409536838531494, "step": 39154 }, { "epoch": 3.086716594402838, "grad_norm": 0.46898147463798523, "learning_rate": 0.0005, "loss": 1.128397822380066, "step": 39155 }, { "epoch": 3.0867954276704768, "grad_norm": 0.8611418604850769, "learning_rate": 0.0005, "loss": 1.380317211151123, "step": 39156 }, { "epoch": 3.086874260938116, "grad_norm": 0.4970347583293915, "learning_rate": 0.0005, "loss": 1.2409076690673828, "step": 39157 }, { "epoch": 3.086953094205755, "grad_norm": 0.5144018530845642, "learning_rate": 0.0005, "loss": 1.5206917524337769, "step": 39158 }, { "epoch": 3.0870319274733937, "grad_norm": 0.4927400052547455, "learning_rate": 0.0005, "loss": 1.2323133945465088, "step": 39159 }, { "epoch": 3.087110760741033, "grad_norm": 0.48676788806915283, "learning_rate": 0.0005, "loss": 1.140701174736023, "step": 39160 }, { "epoch": 3.0871895940086715, "grad_norm": 0.5599604249000549, "learning_rate": 0.0005, "loss": 1.4295499324798584, "step": 39161 }, { "epoch": 3.0872684272763107, "grad_norm": 0.48529595136642456, "learning_rate": 0.0005, "loss": 1.228273868560791, "step": 39162 }, { "epoch": 3.0873472605439494, "grad_norm": 0.45346134901046753, "learning_rate": 0.0005, "loss": 1.2234638929367065, "step": 39163 }, { "epoch": 3.0874260938115885, "grad_norm": 0.4397849440574646, "learning_rate": 0.0005, "loss": 1.1465816497802734, "step": 39164 }, { "epoch": 3.0875049270792276, "grad_norm": 0.44344794750213623, "learning_rate": 0.0005, "loss": 1.3095282316207886, "step": 39165 }, { "epoch": 3.0875837603468663, "grad_norm": 0.4174431264400482, "learning_rate": 0.0005, "loss": 1.1799439191818237, "step": 39166 }, { "epoch": 3.0876625936145055, "grad_norm": 0.3966579735279083, "learning_rate": 0.0005, "loss": 1.0736031532287598, "step": 39167 }, { "epoch": 3.087741426882144, "grad_norm": 0.4976462721824646, "learning_rate": 0.0005, "loss": 1.2946282625198364, "step": 39168 }, { "epoch": 3.0878202601497833, "grad_norm": 0.5112245082855225, "learning_rate": 0.0005, "loss": 1.4317545890808105, "step": 39169 }, { "epoch": 3.087899093417422, "grad_norm": 0.41057783365249634, "learning_rate": 0.0005, "loss": 1.275669813156128, "step": 39170 }, { "epoch": 3.087977926685061, "grad_norm": 0.663403332233429, "learning_rate": 0.0005, "loss": 1.2762763500213623, "step": 39171 }, { "epoch": 3.0880567599527002, "grad_norm": 0.4902200400829315, "learning_rate": 0.0005, "loss": 1.123402714729309, "step": 39172 }, { "epoch": 3.088135593220339, "grad_norm": 0.4898756146430969, "learning_rate": 0.0005, "loss": 1.2127106189727783, "step": 39173 }, { "epoch": 3.088214426487978, "grad_norm": 0.46600106358528137, "learning_rate": 0.0005, "loss": 1.1172064542770386, "step": 39174 }, { "epoch": 3.0882932597556167, "grad_norm": 0.37824270129203796, "learning_rate": 0.0005, "loss": 1.0878480672836304, "step": 39175 }, { "epoch": 3.088372093023256, "grad_norm": 0.44237950444221497, "learning_rate": 0.0005, "loss": 1.1111687421798706, "step": 39176 }, { "epoch": 3.0884509262908946, "grad_norm": 0.6072484254837036, "learning_rate": 0.0005, "loss": 1.478503942489624, "step": 39177 }, { "epoch": 3.0885297595585337, "grad_norm": 0.5472232103347778, "learning_rate": 0.0005, "loss": 1.2048195600509644, "step": 39178 }, { "epoch": 3.088608592826173, "grad_norm": 0.41119158267974854, "learning_rate": 0.0005, "loss": 1.0581483840942383, "step": 39179 }, { "epoch": 3.0886874260938115, "grad_norm": 0.4845554530620575, "learning_rate": 0.0005, "loss": 1.263197422027588, "step": 39180 }, { "epoch": 3.0887662593614507, "grad_norm": 0.4878229796886444, "learning_rate": 0.0005, "loss": 1.4143481254577637, "step": 39181 }, { "epoch": 3.0888450926290894, "grad_norm": 0.5388749241828918, "learning_rate": 0.0005, "loss": 1.1634910106658936, "step": 39182 }, { "epoch": 3.0889239258967285, "grad_norm": 0.47753357887268066, "learning_rate": 0.0005, "loss": 1.2869856357574463, "step": 39183 }, { "epoch": 3.089002759164367, "grad_norm": 0.599454939365387, "learning_rate": 0.0005, "loss": 1.2539851665496826, "step": 39184 }, { "epoch": 3.0890815924320063, "grad_norm": 0.3890388607978821, "learning_rate": 0.0005, "loss": 1.2248048782348633, "step": 39185 }, { "epoch": 3.0891604256996454, "grad_norm": 0.43140843510627747, "learning_rate": 0.0005, "loss": 1.1948072910308838, "step": 39186 }, { "epoch": 3.089239258967284, "grad_norm": 0.477542906999588, "learning_rate": 0.0005, "loss": 1.1530780792236328, "step": 39187 }, { "epoch": 3.0893180922349233, "grad_norm": 0.4538981020450592, "learning_rate": 0.0005, "loss": 1.339834213256836, "step": 39188 }, { "epoch": 3.089396925502562, "grad_norm": 0.49744951725006104, "learning_rate": 0.0005, "loss": 1.438936471939087, "step": 39189 }, { "epoch": 3.089475758770201, "grad_norm": 0.550007164478302, "learning_rate": 0.0005, "loss": 1.579636573791504, "step": 39190 }, { "epoch": 3.0895545920378398, "grad_norm": 0.48151203989982605, "learning_rate": 0.0005, "loss": 1.184957504272461, "step": 39191 }, { "epoch": 3.089633425305479, "grad_norm": 0.5201877951622009, "learning_rate": 0.0005, "loss": 1.4310553073883057, "step": 39192 }, { "epoch": 3.089712258573118, "grad_norm": 0.4849289655685425, "learning_rate": 0.0005, "loss": 1.4237065315246582, "step": 39193 }, { "epoch": 3.0897910918407567, "grad_norm": 0.5578452944755554, "learning_rate": 0.0005, "loss": 1.2169269323349, "step": 39194 }, { "epoch": 3.089869925108396, "grad_norm": 0.45151007175445557, "learning_rate": 0.0005, "loss": 1.341863751411438, "step": 39195 }, { "epoch": 3.0899487583760346, "grad_norm": 0.5916492342948914, "learning_rate": 0.0005, "loss": 1.436081051826477, "step": 39196 }, { "epoch": 3.0900275916436737, "grad_norm": 0.473552942276001, "learning_rate": 0.0005, "loss": 1.2202191352844238, "step": 39197 }, { "epoch": 3.0901064249113124, "grad_norm": 0.5065333247184753, "learning_rate": 0.0005, "loss": 1.5517189502716064, "step": 39198 }, { "epoch": 3.0901852581789515, "grad_norm": 0.4940241575241089, "learning_rate": 0.0005, "loss": 1.2392091751098633, "step": 39199 }, { "epoch": 3.0902640914465906, "grad_norm": 0.5002700090408325, "learning_rate": 0.0005, "loss": 1.2620289325714111, "step": 39200 }, { "epoch": 3.0903429247142293, "grad_norm": 0.49151846766471863, "learning_rate": 0.0005, "loss": 1.3382484912872314, "step": 39201 }, { "epoch": 3.0904217579818685, "grad_norm": 0.4343584477901459, "learning_rate": 0.0005, "loss": 1.3584834337234497, "step": 39202 }, { "epoch": 3.090500591249507, "grad_norm": 0.41906124353408813, "learning_rate": 0.0005, "loss": 1.143770456314087, "step": 39203 }, { "epoch": 3.0905794245171463, "grad_norm": 0.4558759927749634, "learning_rate": 0.0005, "loss": 1.0661449432373047, "step": 39204 }, { "epoch": 3.090658257784785, "grad_norm": 0.481014221906662, "learning_rate": 0.0005, "loss": 1.2112065553665161, "step": 39205 }, { "epoch": 3.090737091052424, "grad_norm": 0.4707048535346985, "learning_rate": 0.0005, "loss": 1.1579275131225586, "step": 39206 }, { "epoch": 3.0908159243200632, "grad_norm": 0.41227367520332336, "learning_rate": 0.0005, "loss": 1.0331484079360962, "step": 39207 }, { "epoch": 3.090894757587702, "grad_norm": 0.6954301595687866, "learning_rate": 0.0005, "loss": 1.1315996646881104, "step": 39208 }, { "epoch": 3.090973590855341, "grad_norm": 0.42062079906463623, "learning_rate": 0.0005, "loss": 1.160046935081482, "step": 39209 }, { "epoch": 3.0910524241229798, "grad_norm": 0.45738306641578674, "learning_rate": 0.0005, "loss": 1.2603659629821777, "step": 39210 }, { "epoch": 3.091131257390619, "grad_norm": 0.49092012643814087, "learning_rate": 0.0005, "loss": 1.170159935951233, "step": 39211 }, { "epoch": 3.0912100906582576, "grad_norm": 0.455220490694046, "learning_rate": 0.0005, "loss": 1.256831169128418, "step": 39212 }, { "epoch": 3.0912889239258967, "grad_norm": 0.5640748143196106, "learning_rate": 0.0005, "loss": 1.2933201789855957, "step": 39213 }, { "epoch": 3.091367757193536, "grad_norm": 0.5036972165107727, "learning_rate": 0.0005, "loss": 1.395585060119629, "step": 39214 }, { "epoch": 3.0914465904611745, "grad_norm": 0.4425426125526428, "learning_rate": 0.0005, "loss": 1.1188740730285645, "step": 39215 }, { "epoch": 3.0915254237288137, "grad_norm": 0.5504395961761475, "learning_rate": 0.0005, "loss": 1.4176924228668213, "step": 39216 }, { "epoch": 3.0916042569964524, "grad_norm": 0.600349485874176, "learning_rate": 0.0005, "loss": 1.4963951110839844, "step": 39217 }, { "epoch": 3.0916830902640915, "grad_norm": 0.4367353022098541, "learning_rate": 0.0005, "loss": 1.2862801551818848, "step": 39218 }, { "epoch": 3.09176192353173, "grad_norm": 0.7664047479629517, "learning_rate": 0.0005, "loss": 1.2958225011825562, "step": 39219 }, { "epoch": 3.0918407567993693, "grad_norm": 0.518417239189148, "learning_rate": 0.0005, "loss": 1.2158808708190918, "step": 39220 }, { "epoch": 3.0919195900670084, "grad_norm": 0.46286630630493164, "learning_rate": 0.0005, "loss": 1.3060321807861328, "step": 39221 }, { "epoch": 3.091998423334647, "grad_norm": 0.49372321367263794, "learning_rate": 0.0005, "loss": 1.3346439599990845, "step": 39222 }, { "epoch": 3.0920772566022863, "grad_norm": 0.4569106996059418, "learning_rate": 0.0005, "loss": 1.2318986654281616, "step": 39223 }, { "epoch": 3.092156089869925, "grad_norm": 0.528105616569519, "learning_rate": 0.0005, "loss": 1.4432201385498047, "step": 39224 }, { "epoch": 3.092234923137564, "grad_norm": 0.46685996651649475, "learning_rate": 0.0005, "loss": 1.3640905618667603, "step": 39225 }, { "epoch": 3.092313756405203, "grad_norm": 0.5183849930763245, "learning_rate": 0.0005, "loss": 1.162665605545044, "step": 39226 }, { "epoch": 3.092392589672842, "grad_norm": 0.5194159746170044, "learning_rate": 0.0005, "loss": 1.20280921459198, "step": 39227 }, { "epoch": 3.092471422940481, "grad_norm": 0.6246117949485779, "learning_rate": 0.0005, "loss": 1.2632882595062256, "step": 39228 }, { "epoch": 3.0925502562081197, "grad_norm": 0.4688594937324524, "learning_rate": 0.0005, "loss": 1.2065153121948242, "step": 39229 }, { "epoch": 3.092629089475759, "grad_norm": 0.4189336895942688, "learning_rate": 0.0005, "loss": 1.281809687614441, "step": 39230 }, { "epoch": 3.0927079227433976, "grad_norm": 0.43634098768234253, "learning_rate": 0.0005, "loss": 1.1415226459503174, "step": 39231 }, { "epoch": 3.0927867560110367, "grad_norm": 0.48216554522514343, "learning_rate": 0.0005, "loss": 1.0985279083251953, "step": 39232 }, { "epoch": 3.0928655892786754, "grad_norm": 0.44245830178260803, "learning_rate": 0.0005, "loss": 1.4058820009231567, "step": 39233 }, { "epoch": 3.0929444225463145, "grad_norm": 0.5296260714530945, "learning_rate": 0.0005, "loss": 1.1013954877853394, "step": 39234 }, { "epoch": 3.0930232558139537, "grad_norm": 0.4505157470703125, "learning_rate": 0.0005, "loss": 1.3327178955078125, "step": 39235 }, { "epoch": 3.0931020890815923, "grad_norm": 1.2160861492156982, "learning_rate": 0.0005, "loss": 1.3447638750076294, "step": 39236 }, { "epoch": 3.0931809223492315, "grad_norm": 0.41286078095436096, "learning_rate": 0.0005, "loss": 0.9721189737319946, "step": 39237 }, { "epoch": 3.09325975561687, "grad_norm": 0.4950298070907593, "learning_rate": 0.0005, "loss": 1.3659861087799072, "step": 39238 }, { "epoch": 3.0933385888845093, "grad_norm": 0.475027471780777, "learning_rate": 0.0005, "loss": 1.2527469396591187, "step": 39239 }, { "epoch": 3.0934174221521484, "grad_norm": 3.3669533729553223, "learning_rate": 0.0005, "loss": 2.1734561920166016, "step": 39240 }, { "epoch": 3.093496255419787, "grad_norm": 0.4604314863681793, "learning_rate": 0.0005, "loss": 1.3213378190994263, "step": 39241 }, { "epoch": 3.0935750886874263, "grad_norm": 0.4981360137462616, "learning_rate": 0.0005, "loss": 1.191352128982544, "step": 39242 }, { "epoch": 3.093653921955065, "grad_norm": 0.48983076214790344, "learning_rate": 0.0005, "loss": 1.3970134258270264, "step": 39243 }, { "epoch": 3.093732755222704, "grad_norm": 0.5228875875473022, "learning_rate": 0.0005, "loss": 1.4442018270492554, "step": 39244 }, { "epoch": 3.0938115884903428, "grad_norm": 0.5071526765823364, "learning_rate": 0.0005, "loss": 1.2694084644317627, "step": 39245 }, { "epoch": 3.093890421757982, "grad_norm": 0.477708101272583, "learning_rate": 0.0005, "loss": 1.193306565284729, "step": 39246 }, { "epoch": 3.0939692550256206, "grad_norm": 0.4392562806606293, "learning_rate": 0.0005, "loss": 1.1951215267181396, "step": 39247 }, { "epoch": 3.0940480882932597, "grad_norm": 0.46591246128082275, "learning_rate": 0.0005, "loss": 1.3415249586105347, "step": 39248 }, { "epoch": 3.094126921560899, "grad_norm": 0.4281461536884308, "learning_rate": 0.0005, "loss": 1.0027815103530884, "step": 39249 }, { "epoch": 3.0942057548285375, "grad_norm": 0.41536131501197815, "learning_rate": 0.0005, "loss": 1.0811165571212769, "step": 39250 }, { "epoch": 3.0942845880961767, "grad_norm": 0.423825740814209, "learning_rate": 0.0005, "loss": 1.1455707550048828, "step": 39251 }, { "epoch": 3.0943634213638154, "grad_norm": 0.5667056441307068, "learning_rate": 0.0005, "loss": 1.2729604244232178, "step": 39252 }, { "epoch": 3.0944422546314545, "grad_norm": 0.531478762626648, "learning_rate": 0.0005, "loss": 1.1476349830627441, "step": 39253 }, { "epoch": 3.0945210878990936, "grad_norm": 0.46359819173812866, "learning_rate": 0.0005, "loss": 1.3278857469558716, "step": 39254 }, { "epoch": 3.0945999211667323, "grad_norm": 0.4953582286834717, "learning_rate": 0.0005, "loss": 1.5081746578216553, "step": 39255 }, { "epoch": 3.0946787544343715, "grad_norm": 0.47253939509391785, "learning_rate": 0.0005, "loss": 1.2887043952941895, "step": 39256 }, { "epoch": 3.09475758770201, "grad_norm": 0.4150893986225128, "learning_rate": 0.0005, "loss": 1.067158579826355, "step": 39257 }, { "epoch": 3.0948364209696493, "grad_norm": 0.4654276669025421, "learning_rate": 0.0005, "loss": 1.1436691284179688, "step": 39258 }, { "epoch": 3.094915254237288, "grad_norm": 0.47937721014022827, "learning_rate": 0.0005, "loss": 1.126701831817627, "step": 39259 }, { "epoch": 3.094994087504927, "grad_norm": 0.5025377869606018, "learning_rate": 0.0005, "loss": 1.254364252090454, "step": 39260 }, { "epoch": 3.095072920772566, "grad_norm": 0.5300925970077515, "learning_rate": 0.0005, "loss": 1.25244140625, "step": 39261 }, { "epoch": 3.095151754040205, "grad_norm": 0.4727702736854553, "learning_rate": 0.0005, "loss": 1.3508906364440918, "step": 39262 }, { "epoch": 3.095230587307844, "grad_norm": 0.5919578075408936, "learning_rate": 0.0005, "loss": 1.4568419456481934, "step": 39263 }, { "epoch": 3.0953094205754828, "grad_norm": 0.4676871597766876, "learning_rate": 0.0005, "loss": 1.2341375350952148, "step": 39264 }, { "epoch": 3.095388253843122, "grad_norm": 1.101827621459961, "learning_rate": 0.0005, "loss": 1.2972091436386108, "step": 39265 }, { "epoch": 3.0954670871107606, "grad_norm": 0.5217928290367126, "learning_rate": 0.0005, "loss": 1.391042947769165, "step": 39266 }, { "epoch": 3.0955459203783997, "grad_norm": 0.6606528759002686, "learning_rate": 0.0005, "loss": 1.410637378692627, "step": 39267 }, { "epoch": 3.095624753646039, "grad_norm": 0.3702511787414551, "learning_rate": 0.0005, "loss": 1.1546475887298584, "step": 39268 }, { "epoch": 3.0957035869136775, "grad_norm": 0.5236912369728088, "learning_rate": 0.0005, "loss": 1.170520544052124, "step": 39269 }, { "epoch": 3.0957824201813167, "grad_norm": 0.420111745595932, "learning_rate": 0.0005, "loss": 1.1518737077713013, "step": 39270 }, { "epoch": 3.0958612534489554, "grad_norm": 0.4634600281715393, "learning_rate": 0.0005, "loss": 1.1911649703979492, "step": 39271 }, { "epoch": 3.0959400867165945, "grad_norm": 0.41551458835601807, "learning_rate": 0.0005, "loss": 1.2986738681793213, "step": 39272 }, { "epoch": 3.096018919984233, "grad_norm": 0.4437948167324066, "learning_rate": 0.0005, "loss": 1.1582674980163574, "step": 39273 }, { "epoch": 3.0960977532518723, "grad_norm": 0.48715952038764954, "learning_rate": 0.0005, "loss": 1.316493272781372, "step": 39274 }, { "epoch": 3.0961765865195114, "grad_norm": 0.4524422585964203, "learning_rate": 0.0005, "loss": 1.133763313293457, "step": 39275 }, { "epoch": 3.09625541978715, "grad_norm": 0.49421146512031555, "learning_rate": 0.0005, "loss": 1.267721176147461, "step": 39276 }, { "epoch": 3.0963342530547893, "grad_norm": 0.5567125678062439, "learning_rate": 0.0005, "loss": 1.2271969318389893, "step": 39277 }, { "epoch": 3.096413086322428, "grad_norm": 0.5037922263145447, "learning_rate": 0.0005, "loss": 1.2754781246185303, "step": 39278 }, { "epoch": 3.096491919590067, "grad_norm": 0.49631351232528687, "learning_rate": 0.0005, "loss": 1.4075407981872559, "step": 39279 }, { "epoch": 3.0965707528577058, "grad_norm": 0.45511189103126526, "learning_rate": 0.0005, "loss": 1.2679743766784668, "step": 39280 }, { "epoch": 3.096649586125345, "grad_norm": 0.5613793134689331, "learning_rate": 0.0005, "loss": 1.4269731044769287, "step": 39281 }, { "epoch": 3.096728419392984, "grad_norm": 0.5063499808311462, "learning_rate": 0.0005, "loss": 1.3924384117126465, "step": 39282 }, { "epoch": 3.0968072526606227, "grad_norm": 0.5114256739616394, "learning_rate": 0.0005, "loss": 1.3665627241134644, "step": 39283 }, { "epoch": 3.096886085928262, "grad_norm": 0.539503812789917, "learning_rate": 0.0005, "loss": 1.1695497035980225, "step": 39284 }, { "epoch": 3.0969649191959006, "grad_norm": 0.579808235168457, "learning_rate": 0.0005, "loss": 1.2554748058319092, "step": 39285 }, { "epoch": 3.0970437524635397, "grad_norm": 0.48381051421165466, "learning_rate": 0.0005, "loss": 1.3151123523712158, "step": 39286 }, { "epoch": 3.0971225857311784, "grad_norm": 0.570869505405426, "learning_rate": 0.0005, "loss": 1.1299927234649658, "step": 39287 }, { "epoch": 3.0972014189988175, "grad_norm": 0.47716182470321655, "learning_rate": 0.0005, "loss": 1.2904891967773438, "step": 39288 }, { "epoch": 3.0972802522664566, "grad_norm": 0.4897468388080597, "learning_rate": 0.0005, "loss": 1.2215720415115356, "step": 39289 }, { "epoch": 3.0973590855340953, "grad_norm": 0.4209783971309662, "learning_rate": 0.0005, "loss": 1.2759884595870972, "step": 39290 }, { "epoch": 3.0974379188017345, "grad_norm": 0.5879510641098022, "learning_rate": 0.0005, "loss": 1.3125627040863037, "step": 39291 }, { "epoch": 3.097516752069373, "grad_norm": 0.5108843445777893, "learning_rate": 0.0005, "loss": 1.4565269947052002, "step": 39292 }, { "epoch": 3.0975955853370123, "grad_norm": 0.46784281730651855, "learning_rate": 0.0005, "loss": 1.0931410789489746, "step": 39293 }, { "epoch": 3.097674418604651, "grad_norm": 0.4116418659687042, "learning_rate": 0.0005, "loss": 1.0494517087936401, "step": 39294 }, { "epoch": 3.09775325187229, "grad_norm": 0.5336835384368896, "learning_rate": 0.0005, "loss": 1.2716495990753174, "step": 39295 }, { "epoch": 3.0978320851399292, "grad_norm": 0.6050745248794556, "learning_rate": 0.0005, "loss": 1.4771404266357422, "step": 39296 }, { "epoch": 3.097910918407568, "grad_norm": 0.4504636228084564, "learning_rate": 0.0005, "loss": 1.2586561441421509, "step": 39297 }, { "epoch": 3.097989751675207, "grad_norm": 0.3691485822200775, "learning_rate": 0.0005, "loss": 0.9957917332649231, "step": 39298 }, { "epoch": 3.0980685849428458, "grad_norm": 0.4704388976097107, "learning_rate": 0.0005, "loss": 1.186105728149414, "step": 39299 }, { "epoch": 3.098147418210485, "grad_norm": 0.5188167095184326, "learning_rate": 0.0005, "loss": 1.273789405822754, "step": 39300 }, { "epoch": 3.0982262514781236, "grad_norm": 0.4622771739959717, "learning_rate": 0.0005, "loss": 1.19716215133667, "step": 39301 }, { "epoch": 3.0983050847457627, "grad_norm": 0.5471972823143005, "learning_rate": 0.0005, "loss": 1.3058080673217773, "step": 39302 }, { "epoch": 3.098383918013402, "grad_norm": 0.4064928889274597, "learning_rate": 0.0005, "loss": 1.2799791097640991, "step": 39303 }, { "epoch": 3.0984627512810405, "grad_norm": 0.4532792866230011, "learning_rate": 0.0005, "loss": 1.2377259731292725, "step": 39304 }, { "epoch": 3.0985415845486797, "grad_norm": 0.5458993911743164, "learning_rate": 0.0005, "loss": 1.2917407751083374, "step": 39305 }, { "epoch": 3.0986204178163184, "grad_norm": 0.4276975691318512, "learning_rate": 0.0005, "loss": 1.2499711513519287, "step": 39306 }, { "epoch": 3.0986992510839575, "grad_norm": 0.4767925441265106, "learning_rate": 0.0005, "loss": 1.2375789880752563, "step": 39307 }, { "epoch": 3.098778084351596, "grad_norm": 0.464487761259079, "learning_rate": 0.0005, "loss": 1.2233412265777588, "step": 39308 }, { "epoch": 3.0988569176192353, "grad_norm": 0.49085211753845215, "learning_rate": 0.0005, "loss": 1.2517776489257812, "step": 39309 }, { "epoch": 3.0989357508868745, "grad_norm": 0.5321225523948669, "learning_rate": 0.0005, "loss": 1.3619680404663086, "step": 39310 }, { "epoch": 3.099014584154513, "grad_norm": 0.47515255212783813, "learning_rate": 0.0005, "loss": 1.1912002563476562, "step": 39311 }, { "epoch": 3.0990934174221523, "grad_norm": 0.5217126607894897, "learning_rate": 0.0005, "loss": 1.4982820749282837, "step": 39312 }, { "epoch": 3.099172250689791, "grad_norm": 0.4443565309047699, "learning_rate": 0.0005, "loss": 1.1323192119598389, "step": 39313 }, { "epoch": 3.09925108395743, "grad_norm": 0.45285874605178833, "learning_rate": 0.0005, "loss": 1.2482049465179443, "step": 39314 }, { "epoch": 3.099329917225069, "grad_norm": 0.4487569332122803, "learning_rate": 0.0005, "loss": 1.2873742580413818, "step": 39315 }, { "epoch": 3.099408750492708, "grad_norm": 0.4600118100643158, "learning_rate": 0.0005, "loss": 1.2743455171585083, "step": 39316 }, { "epoch": 3.099487583760347, "grad_norm": 0.521182119846344, "learning_rate": 0.0005, "loss": 1.306554913520813, "step": 39317 }, { "epoch": 3.0995664170279857, "grad_norm": 0.5509605407714844, "learning_rate": 0.0005, "loss": 1.447422981262207, "step": 39318 }, { "epoch": 3.099645250295625, "grad_norm": 0.4834767282009125, "learning_rate": 0.0005, "loss": 1.1506962776184082, "step": 39319 }, { "epoch": 3.0997240835632636, "grad_norm": 0.4832778573036194, "learning_rate": 0.0005, "loss": 1.4686241149902344, "step": 39320 }, { "epoch": 3.0998029168309027, "grad_norm": 0.49752077460289, "learning_rate": 0.0005, "loss": 1.3314604759216309, "step": 39321 }, { "epoch": 3.0998817500985414, "grad_norm": 0.4657752811908722, "learning_rate": 0.0005, "loss": 1.2110261917114258, "step": 39322 }, { "epoch": 3.0999605833661805, "grad_norm": 0.5153312683105469, "learning_rate": 0.0005, "loss": 1.369367241859436, "step": 39323 }, { "epoch": 3.1000394166338197, "grad_norm": 0.41035810112953186, "learning_rate": 0.0005, "loss": 1.0831058025360107, "step": 39324 }, { "epoch": 3.1001182499014583, "grad_norm": 0.4921666383743286, "learning_rate": 0.0005, "loss": 1.1019532680511475, "step": 39325 }, { "epoch": 3.1001970831690975, "grad_norm": 0.6217090487480164, "learning_rate": 0.0005, "loss": 1.242064118385315, "step": 39326 }, { "epoch": 3.100275916436736, "grad_norm": 0.5664777755737305, "learning_rate": 0.0005, "loss": 1.4891862869262695, "step": 39327 }, { "epoch": 3.1003547497043753, "grad_norm": 0.6162574291229248, "learning_rate": 0.0005, "loss": 1.3950742483139038, "step": 39328 }, { "epoch": 3.100433582972014, "grad_norm": 0.5205617547035217, "learning_rate": 0.0005, "loss": 1.2537684440612793, "step": 39329 }, { "epoch": 3.100512416239653, "grad_norm": 0.43645429611206055, "learning_rate": 0.0005, "loss": 1.3705483675003052, "step": 39330 }, { "epoch": 3.1005912495072923, "grad_norm": 0.4549238681793213, "learning_rate": 0.0005, "loss": 1.3867368698120117, "step": 39331 }, { "epoch": 3.100670082774931, "grad_norm": 0.5652683973312378, "learning_rate": 0.0005, "loss": 1.3584766387939453, "step": 39332 }, { "epoch": 3.10074891604257, "grad_norm": 0.4758661091327667, "learning_rate": 0.0005, "loss": 1.2625467777252197, "step": 39333 }, { "epoch": 3.1008277493102088, "grad_norm": 0.43015187978744507, "learning_rate": 0.0005, "loss": 1.18209707736969, "step": 39334 }, { "epoch": 3.100906582577848, "grad_norm": 0.49405863881111145, "learning_rate": 0.0005, "loss": 1.3084065914154053, "step": 39335 }, { "epoch": 3.1009854158454866, "grad_norm": 0.5358275175094604, "learning_rate": 0.0005, "loss": 1.3726575374603271, "step": 39336 }, { "epoch": 3.1010642491131257, "grad_norm": 0.4365198612213135, "learning_rate": 0.0005, "loss": 1.1157851219177246, "step": 39337 }, { "epoch": 3.101143082380765, "grad_norm": 0.4275498390197754, "learning_rate": 0.0005, "loss": 1.1331135034561157, "step": 39338 }, { "epoch": 3.1012219156484035, "grad_norm": 0.5373244285583496, "learning_rate": 0.0005, "loss": 1.1763191223144531, "step": 39339 }, { "epoch": 3.1013007489160427, "grad_norm": 0.4850233495235443, "learning_rate": 0.0005, "loss": 1.2472021579742432, "step": 39340 }, { "epoch": 3.1013795821836814, "grad_norm": 0.40186476707458496, "learning_rate": 0.0005, "loss": 1.0708045959472656, "step": 39341 }, { "epoch": 3.1014584154513205, "grad_norm": 0.4382624328136444, "learning_rate": 0.0005, "loss": 1.2324904203414917, "step": 39342 }, { "epoch": 3.101537248718959, "grad_norm": 0.6986501216888428, "learning_rate": 0.0005, "loss": 1.263439416885376, "step": 39343 }, { "epoch": 3.1016160819865983, "grad_norm": 0.5894117951393127, "learning_rate": 0.0005, "loss": 1.395174503326416, "step": 39344 }, { "epoch": 3.1016949152542375, "grad_norm": 0.44071894884109497, "learning_rate": 0.0005, "loss": 1.1932110786437988, "step": 39345 }, { "epoch": 3.101773748521876, "grad_norm": 0.537376880645752, "learning_rate": 0.0005, "loss": 1.4381837844848633, "step": 39346 }, { "epoch": 3.1018525817895153, "grad_norm": 0.6201111078262329, "learning_rate": 0.0005, "loss": 1.4471101760864258, "step": 39347 }, { "epoch": 3.101931415057154, "grad_norm": 0.4573545455932617, "learning_rate": 0.0005, "loss": 1.2452452182769775, "step": 39348 }, { "epoch": 3.102010248324793, "grad_norm": 0.5276718735694885, "learning_rate": 0.0005, "loss": 1.347127079963684, "step": 39349 }, { "epoch": 3.102089081592432, "grad_norm": 0.4108153283596039, "learning_rate": 0.0005, "loss": 1.0743846893310547, "step": 39350 }, { "epoch": 3.102167914860071, "grad_norm": 0.5348408222198486, "learning_rate": 0.0005, "loss": 1.463594675064087, "step": 39351 }, { "epoch": 3.10224674812771, "grad_norm": 0.44636061787605286, "learning_rate": 0.0005, "loss": 1.2386137247085571, "step": 39352 }, { "epoch": 3.1023255813953488, "grad_norm": 0.48835432529449463, "learning_rate": 0.0005, "loss": 1.247327208518982, "step": 39353 }, { "epoch": 3.102404414662988, "grad_norm": 0.4276760518550873, "learning_rate": 0.0005, "loss": 1.1635985374450684, "step": 39354 }, { "epoch": 3.1024832479306266, "grad_norm": 0.6588287353515625, "learning_rate": 0.0005, "loss": 1.447355031967163, "step": 39355 }, { "epoch": 3.1025620811982657, "grad_norm": 0.5721825361251831, "learning_rate": 0.0005, "loss": 1.354243516921997, "step": 39356 }, { "epoch": 3.1026409144659044, "grad_norm": 0.42947423458099365, "learning_rate": 0.0005, "loss": 1.3721402883529663, "step": 39357 }, { "epoch": 3.1027197477335435, "grad_norm": 0.5567694902420044, "learning_rate": 0.0005, "loss": 1.1551966667175293, "step": 39358 }, { "epoch": 3.1027985810011827, "grad_norm": 0.5047075748443604, "learning_rate": 0.0005, "loss": 1.42802095413208, "step": 39359 }, { "epoch": 3.1028774142688214, "grad_norm": 0.4370732307434082, "learning_rate": 0.0005, "loss": 1.1761116981506348, "step": 39360 }, { "epoch": 3.1029562475364605, "grad_norm": 0.4748273193836212, "learning_rate": 0.0005, "loss": 1.1837679147720337, "step": 39361 }, { "epoch": 3.103035080804099, "grad_norm": 0.4351392984390259, "learning_rate": 0.0005, "loss": 1.2093448638916016, "step": 39362 }, { "epoch": 3.1031139140717383, "grad_norm": 0.7006819844245911, "learning_rate": 0.0005, "loss": 1.3463470935821533, "step": 39363 }, { "epoch": 3.103192747339377, "grad_norm": 0.45674261450767517, "learning_rate": 0.0005, "loss": 1.3315842151641846, "step": 39364 }, { "epoch": 3.103271580607016, "grad_norm": 1.1349740028381348, "learning_rate": 0.0005, "loss": 1.297980785369873, "step": 39365 }, { "epoch": 3.1033504138746553, "grad_norm": 0.5412163138389587, "learning_rate": 0.0005, "loss": 1.1544743776321411, "step": 39366 }, { "epoch": 3.103429247142294, "grad_norm": 0.5009206533432007, "learning_rate": 0.0005, "loss": 1.206453800201416, "step": 39367 }, { "epoch": 3.103508080409933, "grad_norm": 0.4618772864341736, "learning_rate": 0.0005, "loss": 1.1737444400787354, "step": 39368 }, { "epoch": 3.103586913677572, "grad_norm": 0.5093656778335571, "learning_rate": 0.0005, "loss": 1.3027245998382568, "step": 39369 }, { "epoch": 3.103665746945211, "grad_norm": 0.5942865014076233, "learning_rate": 0.0005, "loss": 1.3390566110610962, "step": 39370 }, { "epoch": 3.10374458021285, "grad_norm": 0.5012015104293823, "learning_rate": 0.0005, "loss": 1.2802276611328125, "step": 39371 }, { "epoch": 3.1038234134804887, "grad_norm": 0.46906256675720215, "learning_rate": 0.0005, "loss": 1.3405683040618896, "step": 39372 }, { "epoch": 3.103902246748128, "grad_norm": 0.4809010922908783, "learning_rate": 0.0005, "loss": 1.2402193546295166, "step": 39373 }, { "epoch": 3.1039810800157666, "grad_norm": 0.4500064551830292, "learning_rate": 0.0005, "loss": 1.078789472579956, "step": 39374 }, { "epoch": 3.1040599132834057, "grad_norm": 0.45866480469703674, "learning_rate": 0.0005, "loss": 1.0306329727172852, "step": 39375 }, { "epoch": 3.1041387465510444, "grad_norm": 0.4822998642921448, "learning_rate": 0.0005, "loss": 1.1087703704833984, "step": 39376 }, { "epoch": 3.1042175798186835, "grad_norm": 0.45636802911758423, "learning_rate": 0.0005, "loss": 1.226183295249939, "step": 39377 }, { "epoch": 3.104296413086322, "grad_norm": 0.4800741672515869, "learning_rate": 0.0005, "loss": 1.3436651229858398, "step": 39378 }, { "epoch": 3.1043752463539613, "grad_norm": 0.4948892891407013, "learning_rate": 0.0005, "loss": 1.2015042304992676, "step": 39379 }, { "epoch": 3.1044540796216005, "grad_norm": 0.6255939602851868, "learning_rate": 0.0005, "loss": 1.4221464395523071, "step": 39380 }, { "epoch": 3.104532912889239, "grad_norm": 0.6271708607673645, "learning_rate": 0.0005, "loss": 1.3500587940216064, "step": 39381 }, { "epoch": 3.1046117461568783, "grad_norm": 0.5546689629554749, "learning_rate": 0.0005, "loss": 1.2530744075775146, "step": 39382 }, { "epoch": 3.104690579424517, "grad_norm": 0.7332937121391296, "learning_rate": 0.0005, "loss": 1.224779486656189, "step": 39383 }, { "epoch": 3.104769412692156, "grad_norm": 0.5761423110961914, "learning_rate": 0.0005, "loss": 1.1715736389160156, "step": 39384 }, { "epoch": 3.1048482459597952, "grad_norm": 0.4405215382575989, "learning_rate": 0.0005, "loss": 1.2380837202072144, "step": 39385 }, { "epoch": 3.104927079227434, "grad_norm": 0.5199835300445557, "learning_rate": 0.0005, "loss": 1.3063194751739502, "step": 39386 }, { "epoch": 3.105005912495073, "grad_norm": 0.5660807490348816, "learning_rate": 0.0005, "loss": 1.289533257484436, "step": 39387 }, { "epoch": 3.1050847457627118, "grad_norm": 0.43035849928855896, "learning_rate": 0.0005, "loss": 1.1376579999923706, "step": 39388 }, { "epoch": 3.105163579030351, "grad_norm": 0.4893607199192047, "learning_rate": 0.0005, "loss": 1.0668964385986328, "step": 39389 }, { "epoch": 3.1052424122979896, "grad_norm": 0.5763993263244629, "learning_rate": 0.0005, "loss": 1.3994810581207275, "step": 39390 }, { "epoch": 3.1053212455656287, "grad_norm": 0.5137110352516174, "learning_rate": 0.0005, "loss": 1.0659739971160889, "step": 39391 }, { "epoch": 3.105400078833268, "grad_norm": 0.5371391177177429, "learning_rate": 0.0005, "loss": 1.1606476306915283, "step": 39392 }, { "epoch": 3.1054789121009065, "grad_norm": 0.7775342464447021, "learning_rate": 0.0005, "loss": 1.2209980487823486, "step": 39393 }, { "epoch": 3.1055577453685457, "grad_norm": 0.4860437512397766, "learning_rate": 0.0005, "loss": 1.300476312637329, "step": 39394 }, { "epoch": 3.1056365786361844, "grad_norm": 0.5440654158592224, "learning_rate": 0.0005, "loss": 1.2820820808410645, "step": 39395 }, { "epoch": 3.1057154119038235, "grad_norm": 0.4933505654335022, "learning_rate": 0.0005, "loss": 1.2576647996902466, "step": 39396 }, { "epoch": 3.105794245171462, "grad_norm": 0.5604109168052673, "learning_rate": 0.0005, "loss": 1.3195209503173828, "step": 39397 }, { "epoch": 3.1058730784391013, "grad_norm": 0.560382068157196, "learning_rate": 0.0005, "loss": 1.2096529006958008, "step": 39398 }, { "epoch": 3.1059519117067405, "grad_norm": 0.44344398379325867, "learning_rate": 0.0005, "loss": 1.3165326118469238, "step": 39399 }, { "epoch": 3.106030744974379, "grad_norm": 0.5047387480735779, "learning_rate": 0.0005, "loss": 1.283372402191162, "step": 39400 }, { "epoch": 3.1061095782420183, "grad_norm": 0.4891892373561859, "learning_rate": 0.0005, "loss": 1.2923634052276611, "step": 39401 }, { "epoch": 3.106188411509657, "grad_norm": 0.4453316330909729, "learning_rate": 0.0005, "loss": 1.2002323865890503, "step": 39402 }, { "epoch": 3.106267244777296, "grad_norm": 0.4394049048423767, "learning_rate": 0.0005, "loss": 1.1916048526763916, "step": 39403 }, { "epoch": 3.106346078044935, "grad_norm": 0.6163815259933472, "learning_rate": 0.0005, "loss": 1.5117805004119873, "step": 39404 }, { "epoch": 3.106424911312574, "grad_norm": 0.5168083906173706, "learning_rate": 0.0005, "loss": 1.295165777206421, "step": 39405 }, { "epoch": 3.106503744580213, "grad_norm": 0.4494140148162842, "learning_rate": 0.0005, "loss": 1.149336814880371, "step": 39406 }, { "epoch": 3.1065825778478517, "grad_norm": 0.5551671385765076, "learning_rate": 0.0005, "loss": 1.108142614364624, "step": 39407 }, { "epoch": 3.106661411115491, "grad_norm": 0.6625839471817017, "learning_rate": 0.0005, "loss": 1.2930386066436768, "step": 39408 }, { "epoch": 3.1067402443831296, "grad_norm": 0.5459152460098267, "learning_rate": 0.0005, "loss": 1.3768601417541504, "step": 39409 }, { "epoch": 3.1068190776507687, "grad_norm": 0.4922768771648407, "learning_rate": 0.0005, "loss": 1.0478830337524414, "step": 39410 }, { "epoch": 3.1068979109184074, "grad_norm": 0.49396467208862305, "learning_rate": 0.0005, "loss": 1.3014730215072632, "step": 39411 }, { "epoch": 3.1069767441860465, "grad_norm": 0.5554885864257812, "learning_rate": 0.0005, "loss": 1.1250817775726318, "step": 39412 }, { "epoch": 3.1070555774536857, "grad_norm": 0.4523577392101288, "learning_rate": 0.0005, "loss": 1.286022663116455, "step": 39413 }, { "epoch": 3.1071344107213243, "grad_norm": 0.4380037188529968, "learning_rate": 0.0005, "loss": 1.3650166988372803, "step": 39414 }, { "epoch": 3.1072132439889635, "grad_norm": 0.48364344239234924, "learning_rate": 0.0005, "loss": 1.1998376846313477, "step": 39415 }, { "epoch": 3.107292077256602, "grad_norm": 0.3944529592990875, "learning_rate": 0.0005, "loss": 1.14289391040802, "step": 39416 }, { "epoch": 3.1073709105242413, "grad_norm": 0.4360721707344055, "learning_rate": 0.0005, "loss": 1.2786281108856201, "step": 39417 }, { "epoch": 3.10744974379188, "grad_norm": 0.5060132741928101, "learning_rate": 0.0005, "loss": 1.1702519655227661, "step": 39418 }, { "epoch": 3.107528577059519, "grad_norm": 0.4271601140499115, "learning_rate": 0.0005, "loss": 1.2446608543395996, "step": 39419 }, { "epoch": 3.1076074103271583, "grad_norm": 0.5060343742370605, "learning_rate": 0.0005, "loss": 1.3566367626190186, "step": 39420 }, { "epoch": 3.107686243594797, "grad_norm": 1.062206506729126, "learning_rate": 0.0005, "loss": 1.2847039699554443, "step": 39421 }, { "epoch": 3.107765076862436, "grad_norm": 0.5073485970497131, "learning_rate": 0.0005, "loss": 1.4253005981445312, "step": 39422 }, { "epoch": 3.1078439101300748, "grad_norm": 0.4765856862068176, "learning_rate": 0.0005, "loss": 1.2812914848327637, "step": 39423 }, { "epoch": 3.107922743397714, "grad_norm": 0.4456115663051605, "learning_rate": 0.0005, "loss": 1.1534367799758911, "step": 39424 }, { "epoch": 3.1080015766653526, "grad_norm": 0.47712406516075134, "learning_rate": 0.0005, "loss": 1.3442611694335938, "step": 39425 }, { "epoch": 3.1080804099329917, "grad_norm": 0.6228588819503784, "learning_rate": 0.0005, "loss": 1.4625836610794067, "step": 39426 }, { "epoch": 3.108159243200631, "grad_norm": 0.5131491422653198, "learning_rate": 0.0005, "loss": 1.2054331302642822, "step": 39427 }, { "epoch": 3.1082380764682696, "grad_norm": 0.4995889365673065, "learning_rate": 0.0005, "loss": 1.2777736186981201, "step": 39428 }, { "epoch": 3.1083169097359087, "grad_norm": 0.45396900177001953, "learning_rate": 0.0005, "loss": 1.074044942855835, "step": 39429 }, { "epoch": 3.1083957430035474, "grad_norm": 0.5279074311256409, "learning_rate": 0.0005, "loss": 1.2855852842330933, "step": 39430 }, { "epoch": 3.1084745762711865, "grad_norm": 0.46880480647087097, "learning_rate": 0.0005, "loss": 1.1949372291564941, "step": 39431 }, { "epoch": 3.108553409538825, "grad_norm": 0.5062404274940491, "learning_rate": 0.0005, "loss": 1.3720043897628784, "step": 39432 }, { "epoch": 3.1086322428064643, "grad_norm": 0.5016666650772095, "learning_rate": 0.0005, "loss": 1.3253049850463867, "step": 39433 }, { "epoch": 3.1087110760741035, "grad_norm": 0.4478072226047516, "learning_rate": 0.0005, "loss": 1.0035648345947266, "step": 39434 }, { "epoch": 3.108789909341742, "grad_norm": 0.4413749575614929, "learning_rate": 0.0005, "loss": 1.1742974519729614, "step": 39435 }, { "epoch": 3.1088687426093813, "grad_norm": 0.6151784062385559, "learning_rate": 0.0005, "loss": 1.3412561416625977, "step": 39436 }, { "epoch": 3.10894757587702, "grad_norm": 0.5127355456352234, "learning_rate": 0.0005, "loss": 1.3337366580963135, "step": 39437 }, { "epoch": 3.109026409144659, "grad_norm": 0.42703354358673096, "learning_rate": 0.0005, "loss": 1.378753423690796, "step": 39438 }, { "epoch": 3.109105242412298, "grad_norm": 0.5434223413467407, "learning_rate": 0.0005, "loss": 1.374917984008789, "step": 39439 }, { "epoch": 3.109184075679937, "grad_norm": 0.5511744022369385, "learning_rate": 0.0005, "loss": 1.3342208862304688, "step": 39440 }, { "epoch": 3.109262908947576, "grad_norm": 0.670612633228302, "learning_rate": 0.0005, "loss": 1.4602267742156982, "step": 39441 }, { "epoch": 3.1093417422152148, "grad_norm": 0.466682493686676, "learning_rate": 0.0005, "loss": 1.3423724174499512, "step": 39442 }, { "epoch": 3.109420575482854, "grad_norm": 0.5167306661605835, "learning_rate": 0.0005, "loss": 1.4927871227264404, "step": 39443 }, { "epoch": 3.1094994087504926, "grad_norm": 0.4622834622859955, "learning_rate": 0.0005, "loss": 1.2104072570800781, "step": 39444 }, { "epoch": 3.1095782420181317, "grad_norm": 0.5500115752220154, "learning_rate": 0.0005, "loss": 1.297964096069336, "step": 39445 }, { "epoch": 3.1096570752857704, "grad_norm": 0.5288935303688049, "learning_rate": 0.0005, "loss": 1.2924717664718628, "step": 39446 }, { "epoch": 3.1097359085534095, "grad_norm": 0.46860045194625854, "learning_rate": 0.0005, "loss": 1.1523516178131104, "step": 39447 }, { "epoch": 3.1098147418210487, "grad_norm": 0.4680975675582886, "learning_rate": 0.0005, "loss": 1.1735612154006958, "step": 39448 }, { "epoch": 3.1098935750886874, "grad_norm": 0.44888556003570557, "learning_rate": 0.0005, "loss": 1.323462963104248, "step": 39449 }, { "epoch": 3.1099724083563265, "grad_norm": 0.6459269523620605, "learning_rate": 0.0005, "loss": 1.5417802333831787, "step": 39450 }, { "epoch": 3.110051241623965, "grad_norm": 0.5718587040901184, "learning_rate": 0.0005, "loss": 1.1783604621887207, "step": 39451 }, { "epoch": 3.1101300748916043, "grad_norm": 0.3623487651348114, "learning_rate": 0.0005, "loss": 1.0936133861541748, "step": 39452 }, { "epoch": 3.110208908159243, "grad_norm": 0.46421557664871216, "learning_rate": 0.0005, "loss": 1.1677274703979492, "step": 39453 }, { "epoch": 3.110287741426882, "grad_norm": 0.41382861137390137, "learning_rate": 0.0005, "loss": 1.1287792921066284, "step": 39454 }, { "epoch": 3.1103665746945213, "grad_norm": 0.4978891611099243, "learning_rate": 0.0005, "loss": 1.3585851192474365, "step": 39455 }, { "epoch": 3.11044540796216, "grad_norm": 0.5458316802978516, "learning_rate": 0.0005, "loss": 1.1698395013809204, "step": 39456 }, { "epoch": 3.110524241229799, "grad_norm": 0.37826594710350037, "learning_rate": 0.0005, "loss": 1.040323257446289, "step": 39457 }, { "epoch": 3.110603074497438, "grad_norm": 0.76798415184021, "learning_rate": 0.0005, "loss": 1.2843170166015625, "step": 39458 }, { "epoch": 3.110681907765077, "grad_norm": 0.39157211780548096, "learning_rate": 0.0005, "loss": 1.1251355409622192, "step": 39459 }, { "epoch": 3.1107607410327156, "grad_norm": 0.3744719326496124, "learning_rate": 0.0005, "loss": 1.160933494567871, "step": 39460 }, { "epoch": 3.1108395743003547, "grad_norm": 0.38353627920150757, "learning_rate": 0.0005, "loss": 1.1359570026397705, "step": 39461 }, { "epoch": 3.110918407567994, "grad_norm": 0.47994381189346313, "learning_rate": 0.0005, "loss": 1.3117144107818604, "step": 39462 }, { "epoch": 3.1109972408356326, "grad_norm": 0.5311474800109863, "learning_rate": 0.0005, "loss": 1.1931703090667725, "step": 39463 }, { "epoch": 3.1110760741032717, "grad_norm": 0.505544126033783, "learning_rate": 0.0005, "loss": 1.396405577659607, "step": 39464 }, { "epoch": 3.1111549073709104, "grad_norm": 0.4759344756603241, "learning_rate": 0.0005, "loss": 1.1616063117980957, "step": 39465 }, { "epoch": 3.1112337406385495, "grad_norm": 0.493355929851532, "learning_rate": 0.0005, "loss": 1.3407220840454102, "step": 39466 }, { "epoch": 3.111312573906188, "grad_norm": 0.5457290410995483, "learning_rate": 0.0005, "loss": 1.353948712348938, "step": 39467 }, { "epoch": 3.1113914071738273, "grad_norm": 0.45442628860473633, "learning_rate": 0.0005, "loss": 1.1638025045394897, "step": 39468 }, { "epoch": 3.1114702404414665, "grad_norm": 0.517027735710144, "learning_rate": 0.0005, "loss": 1.2833566665649414, "step": 39469 }, { "epoch": 3.111549073709105, "grad_norm": 0.5547226071357727, "learning_rate": 0.0005, "loss": 1.1497249603271484, "step": 39470 }, { "epoch": 3.1116279069767443, "grad_norm": 0.4676036834716797, "learning_rate": 0.0005, "loss": 1.3897819519042969, "step": 39471 }, { "epoch": 3.111706740244383, "grad_norm": 0.4673336446285248, "learning_rate": 0.0005, "loss": 1.2048702239990234, "step": 39472 }, { "epoch": 3.111785573512022, "grad_norm": 0.5319217443466187, "learning_rate": 0.0005, "loss": 1.0939178466796875, "step": 39473 }, { "epoch": 3.111864406779661, "grad_norm": 0.503730297088623, "learning_rate": 0.0005, "loss": 1.1864488124847412, "step": 39474 }, { "epoch": 3.1119432400473, "grad_norm": 0.5802656412124634, "learning_rate": 0.0005, "loss": 1.2646242380142212, "step": 39475 }, { "epoch": 3.112022073314939, "grad_norm": 0.42841681838035583, "learning_rate": 0.0005, "loss": 1.227358341217041, "step": 39476 }, { "epoch": 3.1121009065825778, "grad_norm": 0.5925626158714294, "learning_rate": 0.0005, "loss": 1.2358156442642212, "step": 39477 }, { "epoch": 3.112179739850217, "grad_norm": 0.47107046842575073, "learning_rate": 0.0005, "loss": 1.3574433326721191, "step": 39478 }, { "epoch": 3.1122585731178556, "grad_norm": 0.4620189666748047, "learning_rate": 0.0005, "loss": 1.2045555114746094, "step": 39479 }, { "epoch": 3.1123374063854947, "grad_norm": 0.5524142384529114, "learning_rate": 0.0005, "loss": 1.2531096935272217, "step": 39480 }, { "epoch": 3.1124162396531334, "grad_norm": 0.5024033188819885, "learning_rate": 0.0005, "loss": 1.3944146633148193, "step": 39481 }, { "epoch": 3.1124950729207725, "grad_norm": 0.41177934408187866, "learning_rate": 0.0005, "loss": 1.3263511657714844, "step": 39482 }, { "epoch": 3.1125739061884117, "grad_norm": 0.40624815225601196, "learning_rate": 0.0005, "loss": 1.1602097749710083, "step": 39483 }, { "epoch": 3.1126527394560504, "grad_norm": 0.42278194427490234, "learning_rate": 0.0005, "loss": 1.3130953311920166, "step": 39484 }, { "epoch": 3.1127315727236895, "grad_norm": 0.4956028163433075, "learning_rate": 0.0005, "loss": 1.2404398918151855, "step": 39485 }, { "epoch": 3.112810405991328, "grad_norm": 0.5505291223526001, "learning_rate": 0.0005, "loss": 1.2459681034088135, "step": 39486 }, { "epoch": 3.1128892392589673, "grad_norm": 0.5856680274009705, "learning_rate": 0.0005, "loss": 1.2687034606933594, "step": 39487 }, { "epoch": 3.1129680725266065, "grad_norm": 0.4922518730163574, "learning_rate": 0.0005, "loss": 1.3212485313415527, "step": 39488 }, { "epoch": 3.113046905794245, "grad_norm": 0.5192918181419373, "learning_rate": 0.0005, "loss": 1.3714091777801514, "step": 39489 }, { "epoch": 3.1131257390618843, "grad_norm": 0.43781062960624695, "learning_rate": 0.0005, "loss": 1.2358688116073608, "step": 39490 }, { "epoch": 3.113204572329523, "grad_norm": 0.4249979555606842, "learning_rate": 0.0005, "loss": 1.0844039916992188, "step": 39491 }, { "epoch": 3.113283405597162, "grad_norm": 0.6485137939453125, "learning_rate": 0.0005, "loss": 1.33211350440979, "step": 39492 }, { "epoch": 3.113362238864801, "grad_norm": 0.45203500986099243, "learning_rate": 0.0005, "loss": 1.153826355934143, "step": 39493 }, { "epoch": 3.11344107213244, "grad_norm": 0.46417900919914246, "learning_rate": 0.0005, "loss": 1.145895004272461, "step": 39494 }, { "epoch": 3.1135199054000786, "grad_norm": 0.545400083065033, "learning_rate": 0.0005, "loss": 1.2880688905715942, "step": 39495 }, { "epoch": 3.1135987386677177, "grad_norm": 0.41323864459991455, "learning_rate": 0.0005, "loss": 1.2057487964630127, "step": 39496 }, { "epoch": 3.113677571935357, "grad_norm": 0.6048996448516846, "learning_rate": 0.0005, "loss": 1.282585859298706, "step": 39497 }, { "epoch": 3.1137564052029956, "grad_norm": 0.41559499502182007, "learning_rate": 0.0005, "loss": 1.2615196704864502, "step": 39498 }, { "epoch": 3.1138352384706347, "grad_norm": 0.5653717517852783, "learning_rate": 0.0005, "loss": 1.2718443870544434, "step": 39499 }, { "epoch": 3.1139140717382734, "grad_norm": 0.430169016122818, "learning_rate": 0.0005, "loss": 1.2859902381896973, "step": 39500 }, { "epoch": 3.1139929050059125, "grad_norm": 0.5349664688110352, "learning_rate": 0.0005, "loss": 1.2830402851104736, "step": 39501 }, { "epoch": 3.1140717382735517, "grad_norm": 0.5743058323860168, "learning_rate": 0.0005, "loss": 1.2314790487289429, "step": 39502 }, { "epoch": 3.1141505715411903, "grad_norm": 0.5074831247329712, "learning_rate": 0.0005, "loss": 1.3867138624191284, "step": 39503 }, { "epoch": 3.1142294048088295, "grad_norm": 10.96536636352539, "learning_rate": 0.0005, "loss": 1.2613860368728638, "step": 39504 }, { "epoch": 3.114308238076468, "grad_norm": 0.5816332697868347, "learning_rate": 0.0005, "loss": 1.4537105560302734, "step": 39505 }, { "epoch": 3.1143870713441073, "grad_norm": 0.4619881808757782, "learning_rate": 0.0005, "loss": 1.2129617929458618, "step": 39506 }, { "epoch": 3.114465904611746, "grad_norm": 0.47201085090637207, "learning_rate": 0.0005, "loss": 1.251075267791748, "step": 39507 }, { "epoch": 3.114544737879385, "grad_norm": 0.49494364857673645, "learning_rate": 0.0005, "loss": 1.2829821109771729, "step": 39508 }, { "epoch": 3.114623571147024, "grad_norm": 0.4388059675693512, "learning_rate": 0.0005, "loss": 1.3096158504486084, "step": 39509 }, { "epoch": 3.114702404414663, "grad_norm": 0.436709463596344, "learning_rate": 0.0005, "loss": 1.1781995296478271, "step": 39510 }, { "epoch": 3.114781237682302, "grad_norm": 0.4346555471420288, "learning_rate": 0.0005, "loss": 1.3292161226272583, "step": 39511 }, { "epoch": 3.1148600709499408, "grad_norm": 0.4254816770553589, "learning_rate": 0.0005, "loss": 1.0907925367355347, "step": 39512 }, { "epoch": 3.11493890421758, "grad_norm": 0.5047460794448853, "learning_rate": 0.0005, "loss": 1.2560447454452515, "step": 39513 }, { "epoch": 3.1150177374852186, "grad_norm": 0.48683440685272217, "learning_rate": 0.0005, "loss": 1.2498133182525635, "step": 39514 }, { "epoch": 3.1150965707528577, "grad_norm": 0.4960048794746399, "learning_rate": 0.0005, "loss": 1.1320507526397705, "step": 39515 }, { "epoch": 3.115175404020497, "grad_norm": 0.6699831485748291, "learning_rate": 0.0005, "loss": 1.353875994682312, "step": 39516 }, { "epoch": 3.1152542372881356, "grad_norm": 0.4618304669857025, "learning_rate": 0.0005, "loss": 1.2357418537139893, "step": 39517 }, { "epoch": 3.1153330705557747, "grad_norm": 0.45623844861984253, "learning_rate": 0.0005, "loss": 1.2375580072402954, "step": 39518 }, { "epoch": 3.1154119038234134, "grad_norm": 0.5076389312744141, "learning_rate": 0.0005, "loss": 1.3769278526306152, "step": 39519 }, { "epoch": 3.1154907370910525, "grad_norm": 0.5559473037719727, "learning_rate": 0.0005, "loss": 1.3392153978347778, "step": 39520 }, { "epoch": 3.115569570358691, "grad_norm": 0.5161279439926147, "learning_rate": 0.0005, "loss": 1.2745951414108276, "step": 39521 }, { "epoch": 3.1156484036263303, "grad_norm": 0.41371259093284607, "learning_rate": 0.0005, "loss": 1.093116283416748, "step": 39522 }, { "epoch": 3.1157272368939695, "grad_norm": 0.6211155652999878, "learning_rate": 0.0005, "loss": 1.4931880235671997, "step": 39523 }, { "epoch": 3.115806070161608, "grad_norm": 0.48704054951667786, "learning_rate": 0.0005, "loss": 1.1483873128890991, "step": 39524 }, { "epoch": 3.1158849034292473, "grad_norm": 0.4961225390434265, "learning_rate": 0.0005, "loss": 1.3955190181732178, "step": 39525 }, { "epoch": 3.115963736696886, "grad_norm": 0.5015643835067749, "learning_rate": 0.0005, "loss": 1.2795302867889404, "step": 39526 }, { "epoch": 3.116042569964525, "grad_norm": 0.5246116518974304, "learning_rate": 0.0005, "loss": 1.2452149391174316, "step": 39527 }, { "epoch": 3.116121403232164, "grad_norm": 0.4759978652000427, "learning_rate": 0.0005, "loss": 1.2955658435821533, "step": 39528 }, { "epoch": 3.116200236499803, "grad_norm": 0.5507760643959045, "learning_rate": 0.0005, "loss": 1.2775837182998657, "step": 39529 }, { "epoch": 3.116279069767442, "grad_norm": 0.4967231750488281, "learning_rate": 0.0005, "loss": 1.3223319053649902, "step": 39530 }, { "epoch": 3.1163579030350808, "grad_norm": 0.5144562721252441, "learning_rate": 0.0005, "loss": 1.2320287227630615, "step": 39531 }, { "epoch": 3.11643673630272, "grad_norm": 0.4937226474285126, "learning_rate": 0.0005, "loss": 1.0900934934616089, "step": 39532 }, { "epoch": 3.1165155695703586, "grad_norm": 0.4125894606113434, "learning_rate": 0.0005, "loss": 1.1412184238433838, "step": 39533 }, { "epoch": 3.1165944028379977, "grad_norm": 0.4216351807117462, "learning_rate": 0.0005, "loss": 1.2214797735214233, "step": 39534 }, { "epoch": 3.1166732361056364, "grad_norm": 0.4172362685203552, "learning_rate": 0.0005, "loss": 1.1567742824554443, "step": 39535 }, { "epoch": 3.1167520693732755, "grad_norm": 0.4541739225387573, "learning_rate": 0.0005, "loss": 1.1162455081939697, "step": 39536 }, { "epoch": 3.1168309026409147, "grad_norm": 0.5971378684043884, "learning_rate": 0.0005, "loss": 1.4574236869812012, "step": 39537 }, { "epoch": 3.1169097359085534, "grad_norm": 0.49904894828796387, "learning_rate": 0.0005, "loss": 1.3089340925216675, "step": 39538 }, { "epoch": 3.1169885691761925, "grad_norm": 0.5039048790931702, "learning_rate": 0.0005, "loss": 1.2047367095947266, "step": 39539 }, { "epoch": 3.117067402443831, "grad_norm": 0.3785471022129059, "learning_rate": 0.0005, "loss": 1.1216928958892822, "step": 39540 }, { "epoch": 3.1171462357114703, "grad_norm": 0.547004222869873, "learning_rate": 0.0005, "loss": 1.3687841892242432, "step": 39541 }, { "epoch": 3.117225068979109, "grad_norm": 0.45542681217193604, "learning_rate": 0.0005, "loss": 1.2074391841888428, "step": 39542 }, { "epoch": 3.117303902246748, "grad_norm": 0.6752129793167114, "learning_rate": 0.0005, "loss": 1.3982765674591064, "step": 39543 }, { "epoch": 3.1173827355143873, "grad_norm": 0.6254521608352661, "learning_rate": 0.0005, "loss": 1.279498815536499, "step": 39544 }, { "epoch": 3.117461568782026, "grad_norm": 0.4601610004901886, "learning_rate": 0.0005, "loss": 1.0908094644546509, "step": 39545 }, { "epoch": 3.117540402049665, "grad_norm": 0.5539366006851196, "learning_rate": 0.0005, "loss": 1.245328664779663, "step": 39546 }, { "epoch": 3.117619235317304, "grad_norm": 0.40657779574394226, "learning_rate": 0.0005, "loss": 1.0580332279205322, "step": 39547 }, { "epoch": 3.117698068584943, "grad_norm": 0.4297974705696106, "learning_rate": 0.0005, "loss": 1.305593490600586, "step": 39548 }, { "epoch": 3.1177769018525816, "grad_norm": 0.5888925790786743, "learning_rate": 0.0005, "loss": 1.5245559215545654, "step": 39549 }, { "epoch": 3.1178557351202207, "grad_norm": 0.5799451470375061, "learning_rate": 0.0005, "loss": 1.4346920251846313, "step": 39550 }, { "epoch": 3.11793456838786, "grad_norm": 0.47809648513793945, "learning_rate": 0.0005, "loss": 1.2799897193908691, "step": 39551 }, { "epoch": 3.1180134016554986, "grad_norm": 0.5192083716392517, "learning_rate": 0.0005, "loss": 1.2035223245620728, "step": 39552 }, { "epoch": 3.1180922349231377, "grad_norm": 0.4620087444782257, "learning_rate": 0.0005, "loss": 1.3555161952972412, "step": 39553 }, { "epoch": 3.1181710681907764, "grad_norm": 0.5146397948265076, "learning_rate": 0.0005, "loss": 1.3219051361083984, "step": 39554 }, { "epoch": 3.1182499014584155, "grad_norm": 0.48018547892570496, "learning_rate": 0.0005, "loss": 1.3687680959701538, "step": 39555 }, { "epoch": 3.118328734726054, "grad_norm": 0.4198276400566101, "learning_rate": 0.0005, "loss": 1.1117480993270874, "step": 39556 }, { "epoch": 3.1184075679936933, "grad_norm": 0.44841432571411133, "learning_rate": 0.0005, "loss": 1.2148876190185547, "step": 39557 }, { "epoch": 3.1184864012613325, "grad_norm": 0.5404430031776428, "learning_rate": 0.0005, "loss": 1.2305982112884521, "step": 39558 }, { "epoch": 3.118565234528971, "grad_norm": 0.46197962760925293, "learning_rate": 0.0005, "loss": 1.2096021175384521, "step": 39559 }, { "epoch": 3.1186440677966103, "grad_norm": 0.4778299629688263, "learning_rate": 0.0005, "loss": 1.2264339923858643, "step": 39560 }, { "epoch": 3.118722901064249, "grad_norm": 0.4268759787082672, "learning_rate": 0.0005, "loss": 1.103240728378296, "step": 39561 }, { "epoch": 3.118801734331888, "grad_norm": 0.502385139465332, "learning_rate": 0.0005, "loss": 1.3743937015533447, "step": 39562 }, { "epoch": 3.118880567599527, "grad_norm": 0.586891233921051, "learning_rate": 0.0005, "loss": 1.279660701751709, "step": 39563 }, { "epoch": 3.118959400867166, "grad_norm": 0.4819672107696533, "learning_rate": 0.0005, "loss": 1.1913245916366577, "step": 39564 }, { "epoch": 3.119038234134805, "grad_norm": 0.5075849890708923, "learning_rate": 0.0005, "loss": 1.3116140365600586, "step": 39565 }, { "epoch": 3.1191170674024438, "grad_norm": 0.5327641367912292, "learning_rate": 0.0005, "loss": 1.347575306892395, "step": 39566 }, { "epoch": 3.119195900670083, "grad_norm": 0.4833544194698334, "learning_rate": 0.0005, "loss": 1.2816970348358154, "step": 39567 }, { "epoch": 3.1192747339377216, "grad_norm": 0.5093467235565186, "learning_rate": 0.0005, "loss": 1.4693410396575928, "step": 39568 }, { "epoch": 3.1193535672053607, "grad_norm": 0.5568950176239014, "learning_rate": 0.0005, "loss": 1.2465050220489502, "step": 39569 }, { "epoch": 3.1194324004729994, "grad_norm": 0.601751446723938, "learning_rate": 0.0005, "loss": 1.4579181671142578, "step": 39570 }, { "epoch": 3.1195112337406385, "grad_norm": 0.4832680821418762, "learning_rate": 0.0005, "loss": 1.2203173637390137, "step": 39571 }, { "epoch": 3.1195900670082777, "grad_norm": 0.5122175216674805, "learning_rate": 0.0005, "loss": 1.3272048234939575, "step": 39572 }, { "epoch": 3.1196689002759164, "grad_norm": 0.4941999018192291, "learning_rate": 0.0005, "loss": 1.2875186204910278, "step": 39573 }, { "epoch": 3.1197477335435555, "grad_norm": 0.42792242765426636, "learning_rate": 0.0005, "loss": 1.2194993495941162, "step": 39574 }, { "epoch": 3.119826566811194, "grad_norm": 0.568228006362915, "learning_rate": 0.0005, "loss": 1.2460196018218994, "step": 39575 }, { "epoch": 3.1199054000788333, "grad_norm": 0.5022338628768921, "learning_rate": 0.0005, "loss": 1.1419320106506348, "step": 39576 }, { "epoch": 3.119984233346472, "grad_norm": 0.5238350033760071, "learning_rate": 0.0005, "loss": 1.2299890518188477, "step": 39577 }, { "epoch": 3.120063066614111, "grad_norm": 0.5008108615875244, "learning_rate": 0.0005, "loss": 1.0348989963531494, "step": 39578 }, { "epoch": 3.1201418998817503, "grad_norm": 0.4746266305446625, "learning_rate": 0.0005, "loss": 1.1832420825958252, "step": 39579 }, { "epoch": 3.120220733149389, "grad_norm": 0.5063766837120056, "learning_rate": 0.0005, "loss": 1.2178865671157837, "step": 39580 }, { "epoch": 3.120299566417028, "grad_norm": 0.41491422057151794, "learning_rate": 0.0005, "loss": 1.19376802444458, "step": 39581 }, { "epoch": 3.120378399684667, "grad_norm": 0.5307129621505737, "learning_rate": 0.0005, "loss": 1.3742749691009521, "step": 39582 }, { "epoch": 3.120457232952306, "grad_norm": 0.6155792474746704, "learning_rate": 0.0005, "loss": 1.3965692520141602, "step": 39583 }, { "epoch": 3.1205360662199446, "grad_norm": 0.521040678024292, "learning_rate": 0.0005, "loss": 1.2492272853851318, "step": 39584 }, { "epoch": 3.1206148994875837, "grad_norm": 0.602062463760376, "learning_rate": 0.0005, "loss": 1.3957537412643433, "step": 39585 }, { "epoch": 3.120693732755223, "grad_norm": 0.48469486832618713, "learning_rate": 0.0005, "loss": 1.3145487308502197, "step": 39586 }, { "epoch": 3.1207725660228616, "grad_norm": 0.5697298049926758, "learning_rate": 0.0005, "loss": 1.3921629190444946, "step": 39587 }, { "epoch": 3.1208513992905007, "grad_norm": 0.4529956579208374, "learning_rate": 0.0005, "loss": 1.3851430416107178, "step": 39588 }, { "epoch": 3.1209302325581394, "grad_norm": 0.6167221665382385, "learning_rate": 0.0005, "loss": 1.28810453414917, "step": 39589 }, { "epoch": 3.1210090658257785, "grad_norm": 0.5590726733207703, "learning_rate": 0.0005, "loss": 1.420464277267456, "step": 39590 }, { "epoch": 3.121087899093417, "grad_norm": 0.4970751702785492, "learning_rate": 0.0005, "loss": 1.4005039930343628, "step": 39591 }, { "epoch": 3.1211667323610564, "grad_norm": 0.5344001054763794, "learning_rate": 0.0005, "loss": 1.213231086730957, "step": 39592 }, { "epoch": 3.1212455656286955, "grad_norm": 0.5388771891593933, "learning_rate": 0.0005, "loss": 1.340939998626709, "step": 39593 }, { "epoch": 3.121324398896334, "grad_norm": 0.7356370687484741, "learning_rate": 0.0005, "loss": 1.4109718799591064, "step": 39594 }, { "epoch": 3.1214032321639733, "grad_norm": 0.6267918944358826, "learning_rate": 0.0005, "loss": 1.353644847869873, "step": 39595 }, { "epoch": 3.121482065431612, "grad_norm": 0.518390417098999, "learning_rate": 0.0005, "loss": 1.4772151708602905, "step": 39596 }, { "epoch": 3.121560898699251, "grad_norm": 0.49294695258140564, "learning_rate": 0.0005, "loss": 1.1519607305526733, "step": 39597 }, { "epoch": 3.12163973196689, "grad_norm": 0.42916053533554077, "learning_rate": 0.0005, "loss": 1.2134788036346436, "step": 39598 }, { "epoch": 3.121718565234529, "grad_norm": 0.4170108139514923, "learning_rate": 0.0005, "loss": 1.1951017379760742, "step": 39599 }, { "epoch": 3.121797398502168, "grad_norm": 1.2396048307418823, "learning_rate": 0.0005, "loss": 1.3171017169952393, "step": 39600 }, { "epoch": 3.1218762317698068, "grad_norm": 0.4690518379211426, "learning_rate": 0.0005, "loss": 1.270298957824707, "step": 39601 }, { "epoch": 3.121955065037446, "grad_norm": 0.47995391488075256, "learning_rate": 0.0005, "loss": 1.2803170680999756, "step": 39602 }, { "epoch": 3.1220338983050846, "grad_norm": 0.5515641570091248, "learning_rate": 0.0005, "loss": 1.3738772869110107, "step": 39603 }, { "epoch": 3.1221127315727237, "grad_norm": 0.45411837100982666, "learning_rate": 0.0005, "loss": 1.314571738243103, "step": 39604 }, { "epoch": 3.1221915648403624, "grad_norm": 0.5776872038841248, "learning_rate": 0.0005, "loss": 1.3024277687072754, "step": 39605 }, { "epoch": 3.1222703981080016, "grad_norm": 0.5557652115821838, "learning_rate": 0.0005, "loss": 1.3109099864959717, "step": 39606 }, { "epoch": 3.1223492313756407, "grad_norm": 0.49849483370780945, "learning_rate": 0.0005, "loss": 1.1726129055023193, "step": 39607 }, { "epoch": 3.1224280646432794, "grad_norm": 0.544144868850708, "learning_rate": 0.0005, "loss": 1.2548187971115112, "step": 39608 }, { "epoch": 3.1225068979109185, "grad_norm": 0.4877639412879944, "learning_rate": 0.0005, "loss": 1.1191449165344238, "step": 39609 }, { "epoch": 3.122585731178557, "grad_norm": 0.5110722184181213, "learning_rate": 0.0005, "loss": 1.3835111856460571, "step": 39610 }, { "epoch": 3.1226645644461963, "grad_norm": 0.45563745498657227, "learning_rate": 0.0005, "loss": 1.3336787223815918, "step": 39611 }, { "epoch": 3.122743397713835, "grad_norm": 0.46572819352149963, "learning_rate": 0.0005, "loss": 1.1243822574615479, "step": 39612 }, { "epoch": 3.122822230981474, "grad_norm": 0.4231119155883789, "learning_rate": 0.0005, "loss": 1.094646692276001, "step": 39613 }, { "epoch": 3.1229010642491133, "grad_norm": 0.541603684425354, "learning_rate": 0.0005, "loss": 1.199437141418457, "step": 39614 }, { "epoch": 3.122979897516752, "grad_norm": 0.38274580240249634, "learning_rate": 0.0005, "loss": 1.1499769687652588, "step": 39615 }, { "epoch": 3.123058730784391, "grad_norm": 0.49104949831962585, "learning_rate": 0.0005, "loss": 1.2814836502075195, "step": 39616 }, { "epoch": 3.12313756405203, "grad_norm": 0.41107848286628723, "learning_rate": 0.0005, "loss": 1.2018249034881592, "step": 39617 }, { "epoch": 3.123216397319669, "grad_norm": 0.5023995637893677, "learning_rate": 0.0005, "loss": 1.1249306201934814, "step": 39618 }, { "epoch": 3.123295230587308, "grad_norm": 0.38325849175453186, "learning_rate": 0.0005, "loss": 1.2008841037750244, "step": 39619 }, { "epoch": 3.1233740638549468, "grad_norm": 0.4582204818725586, "learning_rate": 0.0005, "loss": 1.2133965492248535, "step": 39620 }, { "epoch": 3.123452897122586, "grad_norm": 0.5039604902267456, "learning_rate": 0.0005, "loss": 1.3424906730651855, "step": 39621 }, { "epoch": 3.1235317303902246, "grad_norm": 0.5549612045288086, "learning_rate": 0.0005, "loss": 1.0819411277770996, "step": 39622 }, { "epoch": 3.1236105636578637, "grad_norm": 0.478571355342865, "learning_rate": 0.0005, "loss": 1.2914931774139404, "step": 39623 }, { "epoch": 3.1236893969255024, "grad_norm": 0.5018674731254578, "learning_rate": 0.0005, "loss": 1.518579125404358, "step": 39624 }, { "epoch": 3.1237682301931415, "grad_norm": 0.42348188161849976, "learning_rate": 0.0005, "loss": 1.0360417366027832, "step": 39625 }, { "epoch": 3.1238470634607802, "grad_norm": 0.5666921734809875, "learning_rate": 0.0005, "loss": 1.210601806640625, "step": 39626 }, { "epoch": 3.1239258967284194, "grad_norm": 0.5019993185997009, "learning_rate": 0.0005, "loss": 1.27445387840271, "step": 39627 }, { "epoch": 3.1240047299960585, "grad_norm": 0.7426143288612366, "learning_rate": 0.0005, "loss": 1.4046504497528076, "step": 39628 }, { "epoch": 3.124083563263697, "grad_norm": 0.48908528685569763, "learning_rate": 0.0005, "loss": 1.5143766403198242, "step": 39629 }, { "epoch": 3.1241623965313363, "grad_norm": 0.4604223370552063, "learning_rate": 0.0005, "loss": 1.1709017753601074, "step": 39630 }, { "epoch": 3.124241229798975, "grad_norm": 0.5686215758323669, "learning_rate": 0.0005, "loss": 1.1970927715301514, "step": 39631 }, { "epoch": 3.124320063066614, "grad_norm": 0.5080866813659668, "learning_rate": 0.0005, "loss": 1.2899188995361328, "step": 39632 }, { "epoch": 3.1243988963342533, "grad_norm": 0.5888713598251343, "learning_rate": 0.0005, "loss": 1.323622226715088, "step": 39633 }, { "epoch": 3.124477729601892, "grad_norm": 0.4837344288825989, "learning_rate": 0.0005, "loss": 1.2145414352416992, "step": 39634 }, { "epoch": 3.124556562869531, "grad_norm": 0.5438107848167419, "learning_rate": 0.0005, "loss": 1.1695364713668823, "step": 39635 }, { "epoch": 3.12463539613717, "grad_norm": 0.5454672574996948, "learning_rate": 0.0005, "loss": 1.3477635383605957, "step": 39636 }, { "epoch": 3.124714229404809, "grad_norm": 0.4744795560836792, "learning_rate": 0.0005, "loss": 1.3198480606079102, "step": 39637 }, { "epoch": 3.1247930626724476, "grad_norm": 0.4538435637950897, "learning_rate": 0.0005, "loss": 1.2323293685913086, "step": 39638 }, { "epoch": 3.1248718959400867, "grad_norm": 0.5073833465576172, "learning_rate": 0.0005, "loss": 1.140514612197876, "step": 39639 }, { "epoch": 3.124950729207726, "grad_norm": 0.4989393949508667, "learning_rate": 0.0005, "loss": 1.3523555994033813, "step": 39640 }, { "epoch": 3.1250295624753646, "grad_norm": 0.5405762791633606, "learning_rate": 0.0005, "loss": 1.254944920539856, "step": 39641 }, { "epoch": 3.1251083957430037, "grad_norm": 0.46847620606422424, "learning_rate": 0.0005, "loss": 1.3772499561309814, "step": 39642 }, { "epoch": 3.1251872290106424, "grad_norm": 0.5441340804100037, "learning_rate": 0.0005, "loss": 1.3214607238769531, "step": 39643 }, { "epoch": 3.1252660622782815, "grad_norm": 0.5801217555999756, "learning_rate": 0.0005, "loss": 1.3985366821289062, "step": 39644 }, { "epoch": 3.12534489554592, "grad_norm": 0.3786030113697052, "learning_rate": 0.0005, "loss": 1.0233657360076904, "step": 39645 }, { "epoch": 3.1254237288135593, "grad_norm": 0.596105694770813, "learning_rate": 0.0005, "loss": 1.3820372819900513, "step": 39646 }, { "epoch": 3.1255025620811985, "grad_norm": 0.5060476064682007, "learning_rate": 0.0005, "loss": 1.1898597478866577, "step": 39647 }, { "epoch": 3.125581395348837, "grad_norm": 0.6407723426818848, "learning_rate": 0.0005, "loss": 1.1994965076446533, "step": 39648 }, { "epoch": 3.1256602286164763, "grad_norm": 0.45207899808883667, "learning_rate": 0.0005, "loss": 1.2668178081512451, "step": 39649 }, { "epoch": 3.125739061884115, "grad_norm": 0.5596973299980164, "learning_rate": 0.0005, "loss": 1.2759597301483154, "step": 39650 }, { "epoch": 3.125817895151754, "grad_norm": 0.41739869117736816, "learning_rate": 0.0005, "loss": 1.2408630847930908, "step": 39651 }, { "epoch": 3.125896728419393, "grad_norm": 0.4596366584300995, "learning_rate": 0.0005, "loss": 1.193255066871643, "step": 39652 }, { "epoch": 3.125975561687032, "grad_norm": 0.4316824972629547, "learning_rate": 0.0005, "loss": 1.1444921493530273, "step": 39653 }, { "epoch": 3.126054394954671, "grad_norm": 0.5774666666984558, "learning_rate": 0.0005, "loss": 1.4154475927352905, "step": 39654 }, { "epoch": 3.1261332282223098, "grad_norm": 0.5481775403022766, "learning_rate": 0.0005, "loss": 1.286900281906128, "step": 39655 }, { "epoch": 3.126212061489949, "grad_norm": 0.5591266751289368, "learning_rate": 0.0005, "loss": 1.1902703046798706, "step": 39656 }, { "epoch": 3.1262908947575876, "grad_norm": 0.49700331687927246, "learning_rate": 0.0005, "loss": 1.2565840482711792, "step": 39657 }, { "epoch": 3.1263697280252267, "grad_norm": 0.48201796412467957, "learning_rate": 0.0005, "loss": 1.346229910850525, "step": 39658 }, { "epoch": 3.1264485612928654, "grad_norm": 0.4126495122909546, "learning_rate": 0.0005, "loss": 1.2170383930206299, "step": 39659 }, { "epoch": 3.1265273945605045, "grad_norm": 0.47603315114974976, "learning_rate": 0.0005, "loss": 1.243001103401184, "step": 39660 }, { "epoch": 3.1266062278281437, "grad_norm": 0.5313109755516052, "learning_rate": 0.0005, "loss": 1.2545032501220703, "step": 39661 }, { "epoch": 3.1266850610957824, "grad_norm": 0.4987264573574066, "learning_rate": 0.0005, "loss": 1.1409357786178589, "step": 39662 }, { "epoch": 3.1267638943634215, "grad_norm": 0.42306458950042725, "learning_rate": 0.0005, "loss": 1.3874521255493164, "step": 39663 }, { "epoch": 3.12684272763106, "grad_norm": 0.4895158112049103, "learning_rate": 0.0005, "loss": 1.3447294235229492, "step": 39664 }, { "epoch": 3.1269215608986993, "grad_norm": 0.42526254057884216, "learning_rate": 0.0005, "loss": 1.13327956199646, "step": 39665 }, { "epoch": 3.127000394166338, "grad_norm": 0.5373322367668152, "learning_rate": 0.0005, "loss": 1.3083198070526123, "step": 39666 }, { "epoch": 3.127079227433977, "grad_norm": 0.478188693523407, "learning_rate": 0.0005, "loss": 1.0537961721420288, "step": 39667 }, { "epoch": 3.1271580607016163, "grad_norm": 0.5805370807647705, "learning_rate": 0.0005, "loss": 1.2515285015106201, "step": 39668 }, { "epoch": 3.127236893969255, "grad_norm": 0.5206645131111145, "learning_rate": 0.0005, "loss": 1.2649701833724976, "step": 39669 }, { "epoch": 3.127315727236894, "grad_norm": 0.4013196527957916, "learning_rate": 0.0005, "loss": 1.1649633646011353, "step": 39670 }, { "epoch": 3.127394560504533, "grad_norm": 0.5325573086738586, "learning_rate": 0.0005, "loss": 1.318080186843872, "step": 39671 }, { "epoch": 3.127473393772172, "grad_norm": 0.5278041958808899, "learning_rate": 0.0005, "loss": 1.2238101959228516, "step": 39672 }, { "epoch": 3.1275522270398106, "grad_norm": 0.4798932671546936, "learning_rate": 0.0005, "loss": 1.1347107887268066, "step": 39673 }, { "epoch": 3.1276310603074498, "grad_norm": 0.44395536184310913, "learning_rate": 0.0005, "loss": 1.334587574005127, "step": 39674 }, { "epoch": 3.127709893575089, "grad_norm": 0.5300804972648621, "learning_rate": 0.0005, "loss": 1.3790262937545776, "step": 39675 }, { "epoch": 3.1277887268427276, "grad_norm": 0.4955802857875824, "learning_rate": 0.0005, "loss": 1.422020673751831, "step": 39676 }, { "epoch": 3.1278675601103667, "grad_norm": 0.4509054720401764, "learning_rate": 0.0005, "loss": 1.0694820880889893, "step": 39677 }, { "epoch": 3.1279463933780054, "grad_norm": 0.44147950410842896, "learning_rate": 0.0005, "loss": 1.135770320892334, "step": 39678 }, { "epoch": 3.1280252266456445, "grad_norm": 0.5101730823516846, "learning_rate": 0.0005, "loss": 1.152700662612915, "step": 39679 }, { "epoch": 3.128104059913283, "grad_norm": 0.40941643714904785, "learning_rate": 0.0005, "loss": 1.1927682161331177, "step": 39680 }, { "epoch": 3.1281828931809224, "grad_norm": 0.45871302485466003, "learning_rate": 0.0005, "loss": 1.1377004384994507, "step": 39681 }, { "epoch": 3.1282617264485615, "grad_norm": 0.47581711411476135, "learning_rate": 0.0005, "loss": 1.3211889266967773, "step": 39682 }, { "epoch": 3.1283405597162, "grad_norm": 0.5350685715675354, "learning_rate": 0.0005, "loss": 1.243307113647461, "step": 39683 }, { "epoch": 3.1284193929838393, "grad_norm": 0.44864287972450256, "learning_rate": 0.0005, "loss": 1.2526419162750244, "step": 39684 }, { "epoch": 3.128498226251478, "grad_norm": 0.5220118165016174, "learning_rate": 0.0005, "loss": 1.5000290870666504, "step": 39685 }, { "epoch": 3.128577059519117, "grad_norm": 0.46669647097587585, "learning_rate": 0.0005, "loss": 1.1865136623382568, "step": 39686 }, { "epoch": 3.128655892786756, "grad_norm": 0.5794087648391724, "learning_rate": 0.0005, "loss": 1.2468774318695068, "step": 39687 }, { "epoch": 3.128734726054395, "grad_norm": 0.49629953503608704, "learning_rate": 0.0005, "loss": 1.4747672080993652, "step": 39688 }, { "epoch": 3.128813559322034, "grad_norm": 0.5585165023803711, "learning_rate": 0.0005, "loss": 1.1804945468902588, "step": 39689 }, { "epoch": 3.1288923925896728, "grad_norm": 0.5397582054138184, "learning_rate": 0.0005, "loss": 1.1886637210845947, "step": 39690 }, { "epoch": 3.128971225857312, "grad_norm": 0.5088427662849426, "learning_rate": 0.0005, "loss": 1.2363123893737793, "step": 39691 }, { "epoch": 3.1290500591249506, "grad_norm": 0.49260589480400085, "learning_rate": 0.0005, "loss": 1.3044191598892212, "step": 39692 }, { "epoch": 3.1291288923925897, "grad_norm": 0.43445634841918945, "learning_rate": 0.0005, "loss": 1.213484764099121, "step": 39693 }, { "epoch": 3.1292077256602284, "grad_norm": 0.4682696759700775, "learning_rate": 0.0005, "loss": 1.3480000495910645, "step": 39694 }, { "epoch": 3.1292865589278676, "grad_norm": 0.4479319453239441, "learning_rate": 0.0005, "loss": 1.473862886428833, "step": 39695 }, { "epoch": 3.1293653921955067, "grad_norm": 0.5385382175445557, "learning_rate": 0.0005, "loss": 1.3240753412246704, "step": 39696 }, { "epoch": 3.1294442254631454, "grad_norm": 0.4492053985595703, "learning_rate": 0.0005, "loss": 1.1675441265106201, "step": 39697 }, { "epoch": 3.1295230587307845, "grad_norm": 0.44519680738449097, "learning_rate": 0.0005, "loss": 1.2518603801727295, "step": 39698 }, { "epoch": 3.129601891998423, "grad_norm": 0.5044689178466797, "learning_rate": 0.0005, "loss": 1.4831798076629639, "step": 39699 }, { "epoch": 3.1296807252660623, "grad_norm": 0.5920885801315308, "learning_rate": 0.0005, "loss": 1.2926194667816162, "step": 39700 }, { "epoch": 3.129759558533701, "grad_norm": 0.54738450050354, "learning_rate": 0.0005, "loss": 1.435920238494873, "step": 39701 }, { "epoch": 3.12983839180134, "grad_norm": 0.46544116735458374, "learning_rate": 0.0005, "loss": 1.3076462745666504, "step": 39702 }, { "epoch": 3.1299172250689793, "grad_norm": 0.40942928194999695, "learning_rate": 0.0005, "loss": 1.1949505805969238, "step": 39703 }, { "epoch": 3.129996058336618, "grad_norm": 0.7000967860221863, "learning_rate": 0.0005, "loss": 1.2371479272842407, "step": 39704 }, { "epoch": 3.130074891604257, "grad_norm": 0.4678972363471985, "learning_rate": 0.0005, "loss": 1.3267918825149536, "step": 39705 }, { "epoch": 3.130153724871896, "grad_norm": 0.43598321080207825, "learning_rate": 0.0005, "loss": 1.1773512363433838, "step": 39706 }, { "epoch": 3.130232558139535, "grad_norm": 0.57728111743927, "learning_rate": 0.0005, "loss": 1.2581325769424438, "step": 39707 }, { "epoch": 3.1303113914071736, "grad_norm": 0.587094247341156, "learning_rate": 0.0005, "loss": 1.4928401708602905, "step": 39708 }, { "epoch": 3.1303902246748128, "grad_norm": 0.42767012119293213, "learning_rate": 0.0005, "loss": 1.1486997604370117, "step": 39709 }, { "epoch": 3.130469057942452, "grad_norm": 0.461692214012146, "learning_rate": 0.0005, "loss": 1.0599396228790283, "step": 39710 }, { "epoch": 3.1305478912100906, "grad_norm": 0.3845607042312622, "learning_rate": 0.0005, "loss": 1.2706928253173828, "step": 39711 }, { "epoch": 3.1306267244777297, "grad_norm": 0.610552191734314, "learning_rate": 0.0005, "loss": 1.4696354866027832, "step": 39712 }, { "epoch": 3.1307055577453684, "grad_norm": 0.5257070660591125, "learning_rate": 0.0005, "loss": 1.2076090574264526, "step": 39713 }, { "epoch": 3.1307843910130075, "grad_norm": 0.4582168161869049, "learning_rate": 0.0005, "loss": 1.258312463760376, "step": 39714 }, { "epoch": 3.1308632242806462, "grad_norm": 0.5434115529060364, "learning_rate": 0.0005, "loss": 1.227837324142456, "step": 39715 }, { "epoch": 3.1309420575482854, "grad_norm": 0.5114661455154419, "learning_rate": 0.0005, "loss": 1.4974335432052612, "step": 39716 }, { "epoch": 3.1310208908159245, "grad_norm": 0.5303900837898254, "learning_rate": 0.0005, "loss": 1.3794184923171997, "step": 39717 }, { "epoch": 3.131099724083563, "grad_norm": 0.6330380439758301, "learning_rate": 0.0005, "loss": 1.354163646697998, "step": 39718 }, { "epoch": 3.1311785573512023, "grad_norm": 0.4430636763572693, "learning_rate": 0.0005, "loss": 1.201442003250122, "step": 39719 }, { "epoch": 3.131257390618841, "grad_norm": 0.5092897415161133, "learning_rate": 0.0005, "loss": 1.446248173713684, "step": 39720 }, { "epoch": 3.13133622388648, "grad_norm": 0.5872414112091064, "learning_rate": 0.0005, "loss": 1.3296681642532349, "step": 39721 }, { "epoch": 3.131415057154119, "grad_norm": 0.5339453816413879, "learning_rate": 0.0005, "loss": 1.2042042016983032, "step": 39722 }, { "epoch": 3.131493890421758, "grad_norm": 0.5740318298339844, "learning_rate": 0.0005, "loss": 1.346815824508667, "step": 39723 }, { "epoch": 3.131572723689397, "grad_norm": 0.5409142971038818, "learning_rate": 0.0005, "loss": 1.2111302614212036, "step": 39724 }, { "epoch": 3.131651556957036, "grad_norm": 0.4195389151573181, "learning_rate": 0.0005, "loss": 1.136256456375122, "step": 39725 }, { "epoch": 3.131730390224675, "grad_norm": 0.41541457176208496, "learning_rate": 0.0005, "loss": 1.0610203742980957, "step": 39726 }, { "epoch": 3.1318092234923136, "grad_norm": 0.5597296953201294, "learning_rate": 0.0005, "loss": 1.2632396221160889, "step": 39727 }, { "epoch": 3.1318880567599527, "grad_norm": 0.44475701451301575, "learning_rate": 0.0005, "loss": 1.2640647888183594, "step": 39728 }, { "epoch": 3.1319668900275914, "grad_norm": 0.5317809581756592, "learning_rate": 0.0005, "loss": 1.265751600265503, "step": 39729 }, { "epoch": 3.1320457232952306, "grad_norm": 0.4981509745121002, "learning_rate": 0.0005, "loss": 1.1743497848510742, "step": 39730 }, { "epoch": 3.1321245565628697, "grad_norm": 0.5375617146492004, "learning_rate": 0.0005, "loss": 1.256036639213562, "step": 39731 }, { "epoch": 3.1322033898305084, "grad_norm": 0.5607829093933105, "learning_rate": 0.0005, "loss": 1.2103689908981323, "step": 39732 }, { "epoch": 3.1322822230981475, "grad_norm": 0.5194825530052185, "learning_rate": 0.0005, "loss": 1.4764479398727417, "step": 39733 }, { "epoch": 3.132361056365786, "grad_norm": 0.5163161158561707, "learning_rate": 0.0005, "loss": 1.2845048904418945, "step": 39734 }, { "epoch": 3.1324398896334253, "grad_norm": 0.5299049615859985, "learning_rate": 0.0005, "loss": 1.300363540649414, "step": 39735 }, { "epoch": 3.1325187229010645, "grad_norm": 0.4160817265510559, "learning_rate": 0.0005, "loss": 1.1903743743896484, "step": 39736 }, { "epoch": 3.132597556168703, "grad_norm": 0.4605136513710022, "learning_rate": 0.0005, "loss": 1.2603931427001953, "step": 39737 }, { "epoch": 3.1326763894363423, "grad_norm": 0.4932692348957062, "learning_rate": 0.0005, "loss": 1.3582699298858643, "step": 39738 }, { "epoch": 3.132755222703981, "grad_norm": 0.47444087266921997, "learning_rate": 0.0005, "loss": 1.3360271453857422, "step": 39739 }, { "epoch": 3.13283405597162, "grad_norm": 0.5226479768753052, "learning_rate": 0.0005, "loss": 1.2476167678833008, "step": 39740 }, { "epoch": 3.132912889239259, "grad_norm": 0.4356269836425781, "learning_rate": 0.0005, "loss": 1.2406256198883057, "step": 39741 }, { "epoch": 3.132991722506898, "grad_norm": 0.46238088607788086, "learning_rate": 0.0005, "loss": 1.3293261528015137, "step": 39742 }, { "epoch": 3.1330705557745366, "grad_norm": 0.5157813429832458, "learning_rate": 0.0005, "loss": 1.2056574821472168, "step": 39743 }, { "epoch": 3.1331493890421758, "grad_norm": 0.6297399997711182, "learning_rate": 0.0005, "loss": 1.2247822284698486, "step": 39744 }, { "epoch": 3.133228222309815, "grad_norm": 0.5537663698196411, "learning_rate": 0.0005, "loss": 1.2905892133712769, "step": 39745 }, { "epoch": 3.1333070555774536, "grad_norm": 0.40990495681762695, "learning_rate": 0.0005, "loss": 1.0940613746643066, "step": 39746 }, { "epoch": 3.1333858888450927, "grad_norm": 0.5565269589424133, "learning_rate": 0.0005, "loss": 1.3441022634506226, "step": 39747 }, { "epoch": 3.1334647221127314, "grad_norm": 0.5497156977653503, "learning_rate": 0.0005, "loss": 1.4634491205215454, "step": 39748 }, { "epoch": 3.1335435553803705, "grad_norm": 0.5147061944007874, "learning_rate": 0.0005, "loss": 1.3050956726074219, "step": 39749 }, { "epoch": 3.1336223886480097, "grad_norm": 0.41868117451667786, "learning_rate": 0.0005, "loss": 1.2268255949020386, "step": 39750 }, { "epoch": 3.1337012219156484, "grad_norm": 0.4383086562156677, "learning_rate": 0.0005, "loss": 1.1429526805877686, "step": 39751 }, { "epoch": 3.1337800551832875, "grad_norm": 0.46109309792518616, "learning_rate": 0.0005, "loss": 1.2072656154632568, "step": 39752 }, { "epoch": 3.133858888450926, "grad_norm": 0.570743978023529, "learning_rate": 0.0005, "loss": 1.3429261445999146, "step": 39753 }, { "epoch": 3.1339377217185653, "grad_norm": 4.504074573516846, "learning_rate": 0.0005, "loss": 1.2540547847747803, "step": 39754 }, { "epoch": 3.134016554986204, "grad_norm": 0.45133569836616516, "learning_rate": 0.0005, "loss": 1.224294662475586, "step": 39755 }, { "epoch": 3.134095388253843, "grad_norm": 0.53557950258255, "learning_rate": 0.0005, "loss": 1.453655481338501, "step": 39756 }, { "epoch": 3.134174221521482, "grad_norm": 0.5240284204483032, "learning_rate": 0.0005, "loss": 1.2641057968139648, "step": 39757 }, { "epoch": 3.134253054789121, "grad_norm": 0.5006718635559082, "learning_rate": 0.0005, "loss": 1.2878141403198242, "step": 39758 }, { "epoch": 3.13433188805676, "grad_norm": 0.481108158826828, "learning_rate": 0.0005, "loss": 1.3066349029541016, "step": 39759 }, { "epoch": 3.134410721324399, "grad_norm": 0.4495261013507843, "learning_rate": 0.0005, "loss": 1.3466994762420654, "step": 39760 }, { "epoch": 3.134489554592038, "grad_norm": 0.48068928718566895, "learning_rate": 0.0005, "loss": 1.2779624462127686, "step": 39761 }, { "epoch": 3.1345683878596766, "grad_norm": 0.4841587245464325, "learning_rate": 0.0005, "loss": 1.4659734964370728, "step": 39762 }, { "epoch": 3.1346472211273158, "grad_norm": 0.5375335216522217, "learning_rate": 0.0005, "loss": 1.2575743198394775, "step": 39763 }, { "epoch": 3.134726054394955, "grad_norm": 0.7639294862747192, "learning_rate": 0.0005, "loss": 1.3627681732177734, "step": 39764 }, { "epoch": 3.1348048876625936, "grad_norm": 0.4780145287513733, "learning_rate": 0.0005, "loss": 1.2986267805099487, "step": 39765 }, { "epoch": 3.1348837209302327, "grad_norm": 0.49754101037979126, "learning_rate": 0.0005, "loss": 1.3831188678741455, "step": 39766 }, { "epoch": 3.1349625541978714, "grad_norm": 0.8150331974029541, "learning_rate": 0.0005, "loss": 1.206728219985962, "step": 39767 }, { "epoch": 3.1350413874655105, "grad_norm": 0.7357121109962463, "learning_rate": 0.0005, "loss": 1.2552943229675293, "step": 39768 }, { "epoch": 3.135120220733149, "grad_norm": 0.49417078495025635, "learning_rate": 0.0005, "loss": 1.3268097639083862, "step": 39769 }, { "epoch": 3.1351990540007884, "grad_norm": 0.44206154346466064, "learning_rate": 0.0005, "loss": 1.2583502531051636, "step": 39770 }, { "epoch": 3.135277887268427, "grad_norm": 0.62650066614151, "learning_rate": 0.0005, "loss": 1.4466581344604492, "step": 39771 }, { "epoch": 3.135356720536066, "grad_norm": 0.4889332354068756, "learning_rate": 0.0005, "loss": 1.1688473224639893, "step": 39772 }, { "epoch": 3.1354355538037053, "grad_norm": 0.39081254601478577, "learning_rate": 0.0005, "loss": 1.1628705263137817, "step": 39773 }, { "epoch": 3.135514387071344, "grad_norm": 0.6213446259498596, "learning_rate": 0.0005, "loss": 1.198702096939087, "step": 39774 }, { "epoch": 3.135593220338983, "grad_norm": 0.500043511390686, "learning_rate": 0.0005, "loss": 1.2370505332946777, "step": 39775 }, { "epoch": 3.135672053606622, "grad_norm": 0.5842925310134888, "learning_rate": 0.0005, "loss": 1.4094645977020264, "step": 39776 }, { "epoch": 3.135750886874261, "grad_norm": 0.4993545114994049, "learning_rate": 0.0005, "loss": 1.2894597053527832, "step": 39777 }, { "epoch": 3.1358297201419, "grad_norm": 0.4691154658794403, "learning_rate": 0.0005, "loss": 1.1406666040420532, "step": 39778 }, { "epoch": 3.135908553409539, "grad_norm": 0.43714457750320435, "learning_rate": 0.0005, "loss": 1.1969318389892578, "step": 39779 }, { "epoch": 3.135987386677178, "grad_norm": 0.6118214726448059, "learning_rate": 0.0005, "loss": 1.3148682117462158, "step": 39780 }, { "epoch": 3.1360662199448166, "grad_norm": 0.5558173060417175, "learning_rate": 0.0005, "loss": 1.2729153633117676, "step": 39781 }, { "epoch": 3.1361450532124557, "grad_norm": 0.5593531131744385, "learning_rate": 0.0005, "loss": 1.27769136428833, "step": 39782 }, { "epoch": 3.1362238864800944, "grad_norm": 0.5182802081108093, "learning_rate": 0.0005, "loss": 1.2949542999267578, "step": 39783 }, { "epoch": 3.1363027197477336, "grad_norm": 0.5225841999053955, "learning_rate": 0.0005, "loss": 1.3956665992736816, "step": 39784 }, { "epoch": 3.1363815530153727, "grad_norm": 0.5665413737297058, "learning_rate": 0.0005, "loss": 1.1743247509002686, "step": 39785 }, { "epoch": 3.1364603862830114, "grad_norm": 0.4670252203941345, "learning_rate": 0.0005, "loss": 1.3365514278411865, "step": 39786 }, { "epoch": 3.1365392195506505, "grad_norm": 0.41148099303245544, "learning_rate": 0.0005, "loss": 1.099326252937317, "step": 39787 }, { "epoch": 3.136618052818289, "grad_norm": 0.5702115297317505, "learning_rate": 0.0005, "loss": 1.2434364557266235, "step": 39788 }, { "epoch": 3.1366968860859283, "grad_norm": 0.5216978192329407, "learning_rate": 0.0005, "loss": 1.4124562740325928, "step": 39789 }, { "epoch": 3.136775719353567, "grad_norm": 0.4598696529865265, "learning_rate": 0.0005, "loss": 1.082784652709961, "step": 39790 }, { "epoch": 3.136854552621206, "grad_norm": 0.5050743222236633, "learning_rate": 0.0005, "loss": 1.2651898860931396, "step": 39791 }, { "epoch": 3.1369333858888453, "grad_norm": 0.446439653635025, "learning_rate": 0.0005, "loss": 1.1894845962524414, "step": 39792 }, { "epoch": 3.137012219156484, "grad_norm": 0.47010087966918945, "learning_rate": 0.0005, "loss": 1.2820587158203125, "step": 39793 }, { "epoch": 3.137091052424123, "grad_norm": 0.41472160816192627, "learning_rate": 0.0005, "loss": 1.0309323072433472, "step": 39794 }, { "epoch": 3.137169885691762, "grad_norm": 0.5647326111793518, "learning_rate": 0.0005, "loss": 1.2626361846923828, "step": 39795 }, { "epoch": 3.137248718959401, "grad_norm": 0.5351423621177673, "learning_rate": 0.0005, "loss": 1.2290451526641846, "step": 39796 }, { "epoch": 3.1373275522270396, "grad_norm": 0.5140595436096191, "learning_rate": 0.0005, "loss": 1.2587494850158691, "step": 39797 }, { "epoch": 3.1374063854946788, "grad_norm": 0.6074240803718567, "learning_rate": 0.0005, "loss": 1.3374912738800049, "step": 39798 }, { "epoch": 3.137485218762318, "grad_norm": 0.5048567652702332, "learning_rate": 0.0005, "loss": 1.255890965461731, "step": 39799 }, { "epoch": 3.1375640520299566, "grad_norm": 0.42806747555732727, "learning_rate": 0.0005, "loss": 1.0406056642532349, "step": 39800 }, { "epoch": 3.1376428852975957, "grad_norm": 0.49613142013549805, "learning_rate": 0.0005, "loss": 1.1491564512252808, "step": 39801 }, { "epoch": 3.1377217185652344, "grad_norm": 0.470950722694397, "learning_rate": 0.0005, "loss": 1.3670048713684082, "step": 39802 }, { "epoch": 3.1378005518328735, "grad_norm": 0.43807730078697205, "learning_rate": 0.0005, "loss": 1.1419057846069336, "step": 39803 }, { "epoch": 3.1378793851005122, "grad_norm": 0.49671614170074463, "learning_rate": 0.0005, "loss": 1.2962913513183594, "step": 39804 }, { "epoch": 3.1379582183681514, "grad_norm": 0.5906836986541748, "learning_rate": 0.0005, "loss": 1.416002631187439, "step": 39805 }, { "epoch": 3.1380370516357905, "grad_norm": 0.5750124454498291, "learning_rate": 0.0005, "loss": 1.511432409286499, "step": 39806 }, { "epoch": 3.138115884903429, "grad_norm": 0.44441187381744385, "learning_rate": 0.0005, "loss": 1.141619086265564, "step": 39807 }, { "epoch": 3.1381947181710683, "grad_norm": 0.5574391484260559, "learning_rate": 0.0005, "loss": 1.4058867692947388, "step": 39808 }, { "epoch": 3.138273551438707, "grad_norm": 0.505477249622345, "learning_rate": 0.0005, "loss": 1.298375129699707, "step": 39809 }, { "epoch": 3.138352384706346, "grad_norm": 0.5149813890457153, "learning_rate": 0.0005, "loss": 1.173453450202942, "step": 39810 }, { "epoch": 3.138431217973985, "grad_norm": 0.5149548053741455, "learning_rate": 0.0005, "loss": 1.2740775346755981, "step": 39811 }, { "epoch": 3.138510051241624, "grad_norm": 0.4347096085548401, "learning_rate": 0.0005, "loss": 1.2656326293945312, "step": 39812 }, { "epoch": 3.138588884509263, "grad_norm": 0.6868494749069214, "learning_rate": 0.0005, "loss": 1.5136908292770386, "step": 39813 }, { "epoch": 3.138667717776902, "grad_norm": 0.49398306012153625, "learning_rate": 0.0005, "loss": 1.3843488693237305, "step": 39814 }, { "epoch": 3.138746551044541, "grad_norm": 0.4505859613418579, "learning_rate": 0.0005, "loss": 1.1669673919677734, "step": 39815 }, { "epoch": 3.1388253843121796, "grad_norm": 0.4949703514575958, "learning_rate": 0.0005, "loss": 1.0986577272415161, "step": 39816 }, { "epoch": 3.1389042175798187, "grad_norm": 0.40061914920806885, "learning_rate": 0.0005, "loss": 1.1263080835342407, "step": 39817 }, { "epoch": 3.1389830508474574, "grad_norm": 0.5509573817253113, "learning_rate": 0.0005, "loss": 1.3837919235229492, "step": 39818 }, { "epoch": 3.1390618841150966, "grad_norm": 0.4241085350513458, "learning_rate": 0.0005, "loss": 1.1079297065734863, "step": 39819 }, { "epoch": 3.1391407173827357, "grad_norm": 0.5822261571884155, "learning_rate": 0.0005, "loss": 1.5577964782714844, "step": 39820 }, { "epoch": 3.1392195506503744, "grad_norm": 0.5085211992263794, "learning_rate": 0.0005, "loss": 1.1833879947662354, "step": 39821 }, { "epoch": 3.1392983839180135, "grad_norm": 0.7132011651992798, "learning_rate": 0.0005, "loss": 1.3181087970733643, "step": 39822 }, { "epoch": 3.139377217185652, "grad_norm": 0.45939919352531433, "learning_rate": 0.0005, "loss": 1.126015543937683, "step": 39823 }, { "epoch": 3.1394560504532913, "grad_norm": 0.48261311650276184, "learning_rate": 0.0005, "loss": 1.2663466930389404, "step": 39824 }, { "epoch": 3.13953488372093, "grad_norm": 0.4574386477470398, "learning_rate": 0.0005, "loss": 1.2830160856246948, "step": 39825 }, { "epoch": 3.139613716988569, "grad_norm": 0.5521945357322693, "learning_rate": 0.0005, "loss": 1.4579159021377563, "step": 39826 }, { "epoch": 3.1396925502562083, "grad_norm": 0.4620998799800873, "learning_rate": 0.0005, "loss": 1.163590669631958, "step": 39827 }, { "epoch": 3.139771383523847, "grad_norm": 0.5573362708091736, "learning_rate": 0.0005, "loss": 1.2817199230194092, "step": 39828 }, { "epoch": 3.139850216791486, "grad_norm": 0.3928610384464264, "learning_rate": 0.0005, "loss": 1.342329502105713, "step": 39829 }, { "epoch": 3.139929050059125, "grad_norm": 0.49518364667892456, "learning_rate": 0.0005, "loss": 1.4324631690979004, "step": 39830 }, { "epoch": 3.140007883326764, "grad_norm": 0.5000633001327515, "learning_rate": 0.0005, "loss": 1.4267406463623047, "step": 39831 }, { "epoch": 3.1400867165944026, "grad_norm": 0.5557655096054077, "learning_rate": 0.0005, "loss": 1.153334379196167, "step": 39832 }, { "epoch": 3.1401655498620418, "grad_norm": 0.5217544436454773, "learning_rate": 0.0005, "loss": 1.2443424463272095, "step": 39833 }, { "epoch": 3.140244383129681, "grad_norm": 0.5439600348472595, "learning_rate": 0.0005, "loss": 1.3460872173309326, "step": 39834 }, { "epoch": 3.1403232163973196, "grad_norm": 0.39007288217544556, "learning_rate": 0.0005, "loss": 1.008910059928894, "step": 39835 }, { "epoch": 3.1404020496649587, "grad_norm": 0.5323079824447632, "learning_rate": 0.0005, "loss": 1.508619785308838, "step": 39836 }, { "epoch": 3.1404808829325974, "grad_norm": 0.536579966545105, "learning_rate": 0.0005, "loss": 1.3624368906021118, "step": 39837 }, { "epoch": 3.1405597162002366, "grad_norm": 0.4365614056587219, "learning_rate": 0.0005, "loss": 1.1754233837127686, "step": 39838 }, { "epoch": 3.1406385494678752, "grad_norm": 0.6674992442131042, "learning_rate": 0.0005, "loss": 1.3170573711395264, "step": 39839 }, { "epoch": 3.1407173827355144, "grad_norm": 0.4396044611930847, "learning_rate": 0.0005, "loss": 1.217003345489502, "step": 39840 }, { "epoch": 3.1407962160031535, "grad_norm": 0.46772193908691406, "learning_rate": 0.0005, "loss": 1.3872475624084473, "step": 39841 }, { "epoch": 3.140875049270792, "grad_norm": 0.6290957927703857, "learning_rate": 0.0005, "loss": 1.342625379562378, "step": 39842 }, { "epoch": 3.1409538825384313, "grad_norm": 0.5650720596313477, "learning_rate": 0.0005, "loss": 1.1806368827819824, "step": 39843 }, { "epoch": 3.14103271580607, "grad_norm": 0.5091869831085205, "learning_rate": 0.0005, "loss": 1.2397902011871338, "step": 39844 }, { "epoch": 3.141111549073709, "grad_norm": 0.5334578156471252, "learning_rate": 0.0005, "loss": 1.3148802518844604, "step": 39845 }, { "epoch": 3.141190382341348, "grad_norm": 0.4624142646789551, "learning_rate": 0.0005, "loss": 1.171325445175171, "step": 39846 }, { "epoch": 3.141269215608987, "grad_norm": 0.5103583931922913, "learning_rate": 0.0005, "loss": 1.294834852218628, "step": 39847 }, { "epoch": 3.141348048876626, "grad_norm": 0.45554864406585693, "learning_rate": 0.0005, "loss": 1.2354737520217896, "step": 39848 }, { "epoch": 3.141426882144265, "grad_norm": 0.49001964926719666, "learning_rate": 0.0005, "loss": 1.0427539348602295, "step": 39849 }, { "epoch": 3.141505715411904, "grad_norm": 0.4818524420261383, "learning_rate": 0.0005, "loss": 1.3716012239456177, "step": 39850 }, { "epoch": 3.1415845486795426, "grad_norm": 0.4946756660938263, "learning_rate": 0.0005, "loss": 1.3503332138061523, "step": 39851 }, { "epoch": 3.1416633819471818, "grad_norm": 0.44876018166542053, "learning_rate": 0.0005, "loss": 1.3111907243728638, "step": 39852 }, { "epoch": 3.141742215214821, "grad_norm": 0.5730226635932922, "learning_rate": 0.0005, "loss": 1.2612857818603516, "step": 39853 }, { "epoch": 3.1418210484824596, "grad_norm": 0.4522855579853058, "learning_rate": 0.0005, "loss": 1.1858980655670166, "step": 39854 }, { "epoch": 3.1418998817500987, "grad_norm": 0.5623651742935181, "learning_rate": 0.0005, "loss": 1.2827222347259521, "step": 39855 }, { "epoch": 3.1419787150177374, "grad_norm": 0.45467039942741394, "learning_rate": 0.0005, "loss": 1.338998794555664, "step": 39856 }, { "epoch": 3.1420575482853765, "grad_norm": 0.5034327507019043, "learning_rate": 0.0005, "loss": 1.2281231880187988, "step": 39857 }, { "epoch": 3.1421363815530152, "grad_norm": 0.5901978015899658, "learning_rate": 0.0005, "loss": 1.2833442687988281, "step": 39858 }, { "epoch": 3.1422152148206544, "grad_norm": 0.5377951264381409, "learning_rate": 0.0005, "loss": 1.289591908454895, "step": 39859 }, { "epoch": 3.142294048088293, "grad_norm": 0.47841134667396545, "learning_rate": 0.0005, "loss": 1.22995924949646, "step": 39860 }, { "epoch": 3.142372881355932, "grad_norm": 0.4670051336288452, "learning_rate": 0.0005, "loss": 1.2047576904296875, "step": 39861 }, { "epoch": 3.1424517146235713, "grad_norm": 0.41214096546173096, "learning_rate": 0.0005, "loss": 1.0170927047729492, "step": 39862 }, { "epoch": 3.14253054789121, "grad_norm": 0.47742316126823425, "learning_rate": 0.0005, "loss": 1.4815144538879395, "step": 39863 }, { "epoch": 3.142609381158849, "grad_norm": 0.4677462577819824, "learning_rate": 0.0005, "loss": 1.3861374855041504, "step": 39864 }, { "epoch": 3.142688214426488, "grad_norm": 0.5265710949897766, "learning_rate": 0.0005, "loss": 1.256697416305542, "step": 39865 }, { "epoch": 3.142767047694127, "grad_norm": 0.5545017719268799, "learning_rate": 0.0005, "loss": 1.2293778657913208, "step": 39866 }, { "epoch": 3.142845880961766, "grad_norm": 0.43607836961746216, "learning_rate": 0.0005, "loss": 1.247939944267273, "step": 39867 }, { "epoch": 3.142924714229405, "grad_norm": 0.4262433350086212, "learning_rate": 0.0005, "loss": 1.1125683784484863, "step": 39868 }, { "epoch": 3.143003547497044, "grad_norm": 0.534200131893158, "learning_rate": 0.0005, "loss": 1.5194189548492432, "step": 39869 }, { "epoch": 3.1430823807646826, "grad_norm": 0.5216259360313416, "learning_rate": 0.0005, "loss": 1.3571739196777344, "step": 39870 }, { "epoch": 3.1431612140323217, "grad_norm": 0.48461878299713135, "learning_rate": 0.0005, "loss": 1.186283826828003, "step": 39871 }, { "epoch": 3.1432400472999604, "grad_norm": 0.5519348382949829, "learning_rate": 0.0005, "loss": 1.431465744972229, "step": 39872 }, { "epoch": 3.1433188805675996, "grad_norm": 0.4671817719936371, "learning_rate": 0.0005, "loss": 1.289522409439087, "step": 39873 }, { "epoch": 3.1433977138352382, "grad_norm": 0.5001624226570129, "learning_rate": 0.0005, "loss": 1.3612240552902222, "step": 39874 }, { "epoch": 3.1434765471028774, "grad_norm": 0.5203068852424622, "learning_rate": 0.0005, "loss": 1.1278938055038452, "step": 39875 }, { "epoch": 3.1435553803705165, "grad_norm": 0.5487295389175415, "learning_rate": 0.0005, "loss": 1.1004972457885742, "step": 39876 }, { "epoch": 3.143634213638155, "grad_norm": 0.5071032047271729, "learning_rate": 0.0005, "loss": 1.337793231010437, "step": 39877 }, { "epoch": 3.1437130469057943, "grad_norm": 0.4648571312427521, "learning_rate": 0.0005, "loss": 1.252623200416565, "step": 39878 }, { "epoch": 3.143791880173433, "grad_norm": 0.4656762480735779, "learning_rate": 0.0005, "loss": 1.1843621730804443, "step": 39879 }, { "epoch": 3.143870713441072, "grad_norm": 0.4827611446380615, "learning_rate": 0.0005, "loss": 1.1997900009155273, "step": 39880 }, { "epoch": 3.1439495467087113, "grad_norm": 0.5065549612045288, "learning_rate": 0.0005, "loss": 1.2732338905334473, "step": 39881 }, { "epoch": 3.14402837997635, "grad_norm": 2.6185951232910156, "learning_rate": 0.0005, "loss": 1.4618151187896729, "step": 39882 }, { "epoch": 3.144107213243989, "grad_norm": 0.6207240223884583, "learning_rate": 0.0005, "loss": 1.5868732929229736, "step": 39883 }, { "epoch": 3.144186046511628, "grad_norm": 0.5509759187698364, "learning_rate": 0.0005, "loss": 1.1701195240020752, "step": 39884 }, { "epoch": 3.144264879779267, "grad_norm": 0.4813796579837799, "learning_rate": 0.0005, "loss": 1.074568271636963, "step": 39885 }, { "epoch": 3.1443437130469056, "grad_norm": 0.5515912771224976, "learning_rate": 0.0005, "loss": 1.3259949684143066, "step": 39886 }, { "epoch": 3.1444225463145448, "grad_norm": 0.6978266835212708, "learning_rate": 0.0005, "loss": 1.288344383239746, "step": 39887 }, { "epoch": 3.1445013795821835, "grad_norm": 0.6263989806175232, "learning_rate": 0.0005, "loss": 1.4518954753875732, "step": 39888 }, { "epoch": 3.1445802128498226, "grad_norm": 0.6662747859954834, "learning_rate": 0.0005, "loss": 1.3995695114135742, "step": 39889 }, { "epoch": 3.1446590461174617, "grad_norm": 0.5287757515907288, "learning_rate": 0.0005, "loss": 1.1387039422988892, "step": 39890 }, { "epoch": 3.1447378793851004, "grad_norm": 0.4418124258518219, "learning_rate": 0.0005, "loss": 1.2130616903305054, "step": 39891 }, { "epoch": 3.1448167126527395, "grad_norm": 0.5356543064117432, "learning_rate": 0.0005, "loss": 1.1145515441894531, "step": 39892 }, { "epoch": 3.1448955459203782, "grad_norm": 0.4743700623512268, "learning_rate": 0.0005, "loss": 1.1705979108810425, "step": 39893 }, { "epoch": 3.1449743791880174, "grad_norm": 0.43929317593574524, "learning_rate": 0.0005, "loss": 1.31197190284729, "step": 39894 }, { "epoch": 3.1450532124556565, "grad_norm": 0.40313804149627686, "learning_rate": 0.0005, "loss": 1.0891084671020508, "step": 39895 }, { "epoch": 3.145132045723295, "grad_norm": 0.5092272162437439, "learning_rate": 0.0005, "loss": 1.3410297632217407, "step": 39896 }, { "epoch": 3.1452108789909343, "grad_norm": 0.5575523376464844, "learning_rate": 0.0005, "loss": 1.241990089416504, "step": 39897 }, { "epoch": 3.145289712258573, "grad_norm": 0.45583686232566833, "learning_rate": 0.0005, "loss": 1.2293117046356201, "step": 39898 }, { "epoch": 3.145368545526212, "grad_norm": 0.5093246102333069, "learning_rate": 0.0005, "loss": 1.4252638816833496, "step": 39899 }, { "epoch": 3.145447378793851, "grad_norm": 0.40744754672050476, "learning_rate": 0.0005, "loss": 1.2062463760375977, "step": 39900 }, { "epoch": 3.14552621206149, "grad_norm": 0.567161500453949, "learning_rate": 0.0005, "loss": 1.310194492340088, "step": 39901 }, { "epoch": 3.1456050453291287, "grad_norm": 0.5910513997077942, "learning_rate": 0.0005, "loss": 1.2817585468292236, "step": 39902 }, { "epoch": 3.145683878596768, "grad_norm": 0.5329436659812927, "learning_rate": 0.0005, "loss": 1.0131080150604248, "step": 39903 }, { "epoch": 3.145762711864407, "grad_norm": 0.38397306203842163, "learning_rate": 0.0005, "loss": 1.2345037460327148, "step": 39904 }, { "epoch": 3.1458415451320456, "grad_norm": 0.6077719330787659, "learning_rate": 0.0005, "loss": 1.2597904205322266, "step": 39905 }, { "epoch": 3.1459203783996847, "grad_norm": 0.4858521819114685, "learning_rate": 0.0005, "loss": 1.3266955614089966, "step": 39906 }, { "epoch": 3.1459992116673234, "grad_norm": 0.5269766449928284, "learning_rate": 0.0005, "loss": 1.3818635940551758, "step": 39907 }, { "epoch": 3.1460780449349626, "grad_norm": 0.8021485209465027, "learning_rate": 0.0005, "loss": 1.4349303245544434, "step": 39908 }, { "epoch": 3.1461568782026017, "grad_norm": 0.524096667766571, "learning_rate": 0.0005, "loss": 1.422593116760254, "step": 39909 }, { "epoch": 3.1462357114702404, "grad_norm": 0.5110611915588379, "learning_rate": 0.0005, "loss": 1.1733384132385254, "step": 39910 }, { "epoch": 3.1463145447378795, "grad_norm": 0.41185522079467773, "learning_rate": 0.0005, "loss": 1.2952542304992676, "step": 39911 }, { "epoch": 3.146393378005518, "grad_norm": 0.42647629976272583, "learning_rate": 0.0005, "loss": 1.1280174255371094, "step": 39912 }, { "epoch": 3.1464722112731573, "grad_norm": 0.41018301248550415, "learning_rate": 0.0005, "loss": 1.2165446281433105, "step": 39913 }, { "epoch": 3.146551044540796, "grad_norm": 0.44696736335754395, "learning_rate": 0.0005, "loss": 1.2135019302368164, "step": 39914 }, { "epoch": 3.146629877808435, "grad_norm": 0.5341700911521912, "learning_rate": 0.0005, "loss": 1.1446003913879395, "step": 39915 }, { "epoch": 3.1467087110760743, "grad_norm": 0.4279029667377472, "learning_rate": 0.0005, "loss": 1.18505859375, "step": 39916 }, { "epoch": 3.146787544343713, "grad_norm": 0.815358579158783, "learning_rate": 0.0005, "loss": 1.3725734949111938, "step": 39917 }, { "epoch": 3.146866377611352, "grad_norm": 0.6249613761901855, "learning_rate": 0.0005, "loss": 1.2651833295822144, "step": 39918 }, { "epoch": 3.146945210878991, "grad_norm": 0.49140530824661255, "learning_rate": 0.0005, "loss": 1.148504376411438, "step": 39919 }, { "epoch": 3.14702404414663, "grad_norm": 0.522355318069458, "learning_rate": 0.0005, "loss": 1.317491054534912, "step": 39920 }, { "epoch": 3.1471028774142686, "grad_norm": 0.43632635474205017, "learning_rate": 0.0005, "loss": 1.3943603038787842, "step": 39921 }, { "epoch": 3.1471817106819078, "grad_norm": 0.45969536900520325, "learning_rate": 0.0005, "loss": 1.3177247047424316, "step": 39922 }, { "epoch": 3.147260543949547, "grad_norm": 0.535449743270874, "learning_rate": 0.0005, "loss": 1.450748324394226, "step": 39923 }, { "epoch": 3.1473393772171856, "grad_norm": 0.49302661418914795, "learning_rate": 0.0005, "loss": 1.3431637287139893, "step": 39924 }, { "epoch": 3.1474182104848247, "grad_norm": 0.4702833592891693, "learning_rate": 0.0005, "loss": 1.2505018711090088, "step": 39925 }, { "epoch": 3.1474970437524634, "grad_norm": 0.5547835826873779, "learning_rate": 0.0005, "loss": 1.2621440887451172, "step": 39926 }, { "epoch": 3.1475758770201026, "grad_norm": 0.49061495065689087, "learning_rate": 0.0005, "loss": 1.18495512008667, "step": 39927 }, { "epoch": 3.1476547102877412, "grad_norm": 0.4893087148666382, "learning_rate": 0.0005, "loss": 1.3386456966400146, "step": 39928 }, { "epoch": 3.1477335435553804, "grad_norm": 0.3948254883289337, "learning_rate": 0.0005, "loss": 1.0936543941497803, "step": 39929 }, { "epoch": 3.1478123768230195, "grad_norm": 0.4646036922931671, "learning_rate": 0.0005, "loss": 1.330530047416687, "step": 39930 }, { "epoch": 3.147891210090658, "grad_norm": 0.45092833042144775, "learning_rate": 0.0005, "loss": 1.0602127313613892, "step": 39931 }, { "epoch": 3.1479700433582973, "grad_norm": 0.6939958930015564, "learning_rate": 0.0005, "loss": 1.2635109424591064, "step": 39932 }, { "epoch": 3.148048876625936, "grad_norm": 0.4833165407180786, "learning_rate": 0.0005, "loss": 1.2270143032073975, "step": 39933 }, { "epoch": 3.148127709893575, "grad_norm": 0.4141255021095276, "learning_rate": 0.0005, "loss": 1.1671887636184692, "step": 39934 }, { "epoch": 3.148206543161214, "grad_norm": 0.5376659631729126, "learning_rate": 0.0005, "loss": 1.4521501064300537, "step": 39935 }, { "epoch": 3.148285376428853, "grad_norm": 0.44591379165649414, "learning_rate": 0.0005, "loss": 1.1979761123657227, "step": 39936 }, { "epoch": 3.148364209696492, "grad_norm": 0.4928877651691437, "learning_rate": 0.0005, "loss": 1.1661899089813232, "step": 39937 }, { "epoch": 3.148443042964131, "grad_norm": 0.4586421847343445, "learning_rate": 0.0005, "loss": 1.1595815420150757, "step": 39938 }, { "epoch": 3.14852187623177, "grad_norm": 0.4862096309661865, "learning_rate": 0.0005, "loss": 1.0566322803497314, "step": 39939 }, { "epoch": 3.1486007094994086, "grad_norm": 0.4867718815803528, "learning_rate": 0.0005, "loss": 1.1207499504089355, "step": 39940 }, { "epoch": 3.1486795427670478, "grad_norm": 0.4865862727165222, "learning_rate": 0.0005, "loss": 1.160988688468933, "step": 39941 }, { "epoch": 3.1487583760346864, "grad_norm": 0.5254025459289551, "learning_rate": 0.0005, "loss": 1.3459413051605225, "step": 39942 }, { "epoch": 3.1488372093023256, "grad_norm": 0.5171037316322327, "learning_rate": 0.0005, "loss": 1.4758648872375488, "step": 39943 }, { "epoch": 3.1489160425699647, "grad_norm": 0.5094932317733765, "learning_rate": 0.0005, "loss": 1.198873519897461, "step": 39944 }, { "epoch": 3.1489948758376034, "grad_norm": 0.5175991654396057, "learning_rate": 0.0005, "loss": 1.2847163677215576, "step": 39945 }, { "epoch": 3.1490737091052425, "grad_norm": 0.44008204340934753, "learning_rate": 0.0005, "loss": 1.2614431381225586, "step": 39946 }, { "epoch": 3.1491525423728812, "grad_norm": 0.42895326018333435, "learning_rate": 0.0005, "loss": 1.1503784656524658, "step": 39947 }, { "epoch": 3.1492313756405204, "grad_norm": 0.40057316422462463, "learning_rate": 0.0005, "loss": 1.0224409103393555, "step": 39948 }, { "epoch": 3.149310208908159, "grad_norm": 0.46658217906951904, "learning_rate": 0.0005, "loss": 1.2357356548309326, "step": 39949 }, { "epoch": 3.149389042175798, "grad_norm": 0.5494757890701294, "learning_rate": 0.0005, "loss": 1.3500466346740723, "step": 39950 }, { "epoch": 3.1494678754434373, "grad_norm": 0.4685058891773224, "learning_rate": 0.0005, "loss": 1.2788054943084717, "step": 39951 }, { "epoch": 3.149546708711076, "grad_norm": 0.47021448612213135, "learning_rate": 0.0005, "loss": 1.2309534549713135, "step": 39952 }, { "epoch": 3.149625541978715, "grad_norm": 0.4794260263442993, "learning_rate": 0.0005, "loss": 1.1304116249084473, "step": 39953 }, { "epoch": 3.149704375246354, "grad_norm": 0.5491377711296082, "learning_rate": 0.0005, "loss": 1.4586937427520752, "step": 39954 }, { "epoch": 3.149783208513993, "grad_norm": 0.5283344984054565, "learning_rate": 0.0005, "loss": 1.36944580078125, "step": 39955 }, { "epoch": 3.1498620417816316, "grad_norm": 0.5018125772476196, "learning_rate": 0.0005, "loss": 1.2504478693008423, "step": 39956 }, { "epoch": 3.149940875049271, "grad_norm": 0.44148916006088257, "learning_rate": 0.0005, "loss": 1.1017117500305176, "step": 39957 }, { "epoch": 3.15001970831691, "grad_norm": 0.668199896812439, "learning_rate": 0.0005, "loss": 1.4438116550445557, "step": 39958 }, { "epoch": 3.1500985415845486, "grad_norm": 0.5324940085411072, "learning_rate": 0.0005, "loss": 1.2790217399597168, "step": 39959 }, { "epoch": 3.1501773748521877, "grad_norm": 0.4768695831298828, "learning_rate": 0.0005, "loss": 1.3911103010177612, "step": 39960 }, { "epoch": 3.1502562081198264, "grad_norm": 0.38334327936172485, "learning_rate": 0.0005, "loss": 1.0748398303985596, "step": 39961 }, { "epoch": 3.1503350413874656, "grad_norm": 0.4519560635089874, "learning_rate": 0.0005, "loss": 1.3154287338256836, "step": 39962 }, { "epoch": 3.1504138746551043, "grad_norm": 0.6118413209915161, "learning_rate": 0.0005, "loss": 1.438657283782959, "step": 39963 }, { "epoch": 3.1504927079227434, "grad_norm": 0.40610578656196594, "learning_rate": 0.0005, "loss": 1.1727380752563477, "step": 39964 }, { "epoch": 3.1505715411903825, "grad_norm": 0.3828447759151459, "learning_rate": 0.0005, "loss": 1.2381770610809326, "step": 39965 }, { "epoch": 3.150650374458021, "grad_norm": 0.5198693871498108, "learning_rate": 0.0005, "loss": 1.4024596214294434, "step": 39966 }, { "epoch": 3.1507292077256603, "grad_norm": 0.4911203682422638, "learning_rate": 0.0005, "loss": 1.3867764472961426, "step": 39967 }, { "epoch": 3.150808040993299, "grad_norm": 0.45800289511680603, "learning_rate": 0.0005, "loss": 1.2457337379455566, "step": 39968 }, { "epoch": 3.150886874260938, "grad_norm": 0.5450226068496704, "learning_rate": 0.0005, "loss": 1.4434280395507812, "step": 39969 }, { "epoch": 3.150965707528577, "grad_norm": 0.38347128033638, "learning_rate": 0.0005, "loss": 1.0787296295166016, "step": 39970 }, { "epoch": 3.151044540796216, "grad_norm": 0.5110857486724854, "learning_rate": 0.0005, "loss": 1.3728609085083008, "step": 39971 }, { "epoch": 3.151123374063855, "grad_norm": 0.48055732250213623, "learning_rate": 0.0005, "loss": 1.5225592851638794, "step": 39972 }, { "epoch": 3.151202207331494, "grad_norm": 0.49178382754325867, "learning_rate": 0.0005, "loss": 1.2179508209228516, "step": 39973 }, { "epoch": 3.151281040599133, "grad_norm": 0.6793809533119202, "learning_rate": 0.0005, "loss": 1.1045082807540894, "step": 39974 }, { "epoch": 3.1513598738667716, "grad_norm": 0.7013352513313293, "learning_rate": 0.0005, "loss": 1.1562849283218384, "step": 39975 }, { "epoch": 3.1514387071344108, "grad_norm": 0.5297361016273499, "learning_rate": 0.0005, "loss": 1.3501079082489014, "step": 39976 }, { "epoch": 3.1515175404020495, "grad_norm": 0.4804772436618805, "learning_rate": 0.0005, "loss": 1.3126654624938965, "step": 39977 }, { "epoch": 3.1515963736696886, "grad_norm": 0.7174471020698547, "learning_rate": 0.0005, "loss": 1.278806447982788, "step": 39978 }, { "epoch": 3.1516752069373277, "grad_norm": 0.739685595035553, "learning_rate": 0.0005, "loss": 1.3178141117095947, "step": 39979 }, { "epoch": 3.1517540402049664, "grad_norm": 0.8778210282325745, "learning_rate": 0.0005, "loss": 1.335310935974121, "step": 39980 }, { "epoch": 3.1518328734726055, "grad_norm": 0.43704015016555786, "learning_rate": 0.0005, "loss": 1.3250501155853271, "step": 39981 }, { "epoch": 3.1519117067402442, "grad_norm": 0.5417104959487915, "learning_rate": 0.0005, "loss": 1.323401927947998, "step": 39982 }, { "epoch": 3.1519905400078834, "grad_norm": 0.5542714595794678, "learning_rate": 0.0005, "loss": 1.2699097394943237, "step": 39983 }, { "epoch": 3.1520693732755225, "grad_norm": 0.4538191258907318, "learning_rate": 0.0005, "loss": 1.2594521045684814, "step": 39984 }, { "epoch": 3.152148206543161, "grad_norm": 0.52961665391922, "learning_rate": 0.0005, "loss": 1.080214262008667, "step": 39985 }, { "epoch": 3.1522270398108003, "grad_norm": 0.530003011226654, "learning_rate": 0.0005, "loss": 1.3621020317077637, "step": 39986 }, { "epoch": 3.152305873078439, "grad_norm": 0.5241891145706177, "learning_rate": 0.0005, "loss": 1.2289974689483643, "step": 39987 }, { "epoch": 3.152384706346078, "grad_norm": 1.7014398574829102, "learning_rate": 0.0005, "loss": 1.358648419380188, "step": 39988 }, { "epoch": 3.152463539613717, "grad_norm": 0.45795437693595886, "learning_rate": 0.0005, "loss": 1.3362505435943604, "step": 39989 }, { "epoch": 3.152542372881356, "grad_norm": 0.3838466703891754, "learning_rate": 0.0005, "loss": 1.0936269760131836, "step": 39990 }, { "epoch": 3.1526212061489947, "grad_norm": 0.5639399886131287, "learning_rate": 0.0005, "loss": 1.582607626914978, "step": 39991 }, { "epoch": 3.152700039416634, "grad_norm": 0.40369874238967896, "learning_rate": 0.0005, "loss": 1.172027826309204, "step": 39992 }, { "epoch": 3.152778872684273, "grad_norm": 0.5160770416259766, "learning_rate": 0.0005, "loss": 1.314847707748413, "step": 39993 }, { "epoch": 3.1528577059519116, "grad_norm": 0.5498457551002502, "learning_rate": 0.0005, "loss": 1.38907790184021, "step": 39994 }, { "epoch": 3.1529365392195507, "grad_norm": 0.5507723689079285, "learning_rate": 0.0005, "loss": 1.2243385314941406, "step": 39995 }, { "epoch": 3.1530153724871894, "grad_norm": 0.511853039264679, "learning_rate": 0.0005, "loss": 1.390972375869751, "step": 39996 }, { "epoch": 3.1530942057548286, "grad_norm": 0.4060298502445221, "learning_rate": 0.0005, "loss": 1.173736810684204, "step": 39997 }, { "epoch": 3.1531730390224677, "grad_norm": 0.4175974130630493, "learning_rate": 0.0005, "loss": 1.2077782154083252, "step": 39998 }, { "epoch": 3.1532518722901064, "grad_norm": 0.4328242540359497, "learning_rate": 0.0005, "loss": 1.1133265495300293, "step": 39999 }, { "epoch": 3.1533307055577455, "grad_norm": 0.4843727648258209, "learning_rate": 0.0005, "loss": 1.4194180965423584, "step": 40000 }, { "epoch": 3.153409538825384, "grad_norm": 0.47171226143836975, "learning_rate": 0.0005, "loss": 1.380229115486145, "step": 40001 }, { "epoch": 3.1534883720930234, "grad_norm": 0.4913961589336395, "learning_rate": 0.0005, "loss": 1.080693006515503, "step": 40002 }, { "epoch": 3.153567205360662, "grad_norm": 0.46429598331451416, "learning_rate": 0.0005, "loss": 1.2846176624298096, "step": 40003 }, { "epoch": 3.153646038628301, "grad_norm": 0.5738230347633362, "learning_rate": 0.0005, "loss": 1.3629906177520752, "step": 40004 }, { "epoch": 3.15372487189594, "grad_norm": 0.552325963973999, "learning_rate": 0.0005, "loss": 1.298130750656128, "step": 40005 }, { "epoch": 3.153803705163579, "grad_norm": 0.39243754744529724, "learning_rate": 0.0005, "loss": 1.1948152780532837, "step": 40006 }, { "epoch": 3.153882538431218, "grad_norm": 0.4860674738883972, "learning_rate": 0.0005, "loss": 1.074361801147461, "step": 40007 }, { "epoch": 3.153961371698857, "grad_norm": 0.4517410099506378, "learning_rate": 0.0005, "loss": 1.2907012701034546, "step": 40008 }, { "epoch": 3.154040204966496, "grad_norm": 0.472294420003891, "learning_rate": 0.0005, "loss": 1.309089183807373, "step": 40009 }, { "epoch": 3.1541190382341346, "grad_norm": 0.4458678364753723, "learning_rate": 0.0005, "loss": 1.2607237100601196, "step": 40010 }, { "epoch": 3.1541978715017738, "grad_norm": 0.5548734664916992, "learning_rate": 0.0005, "loss": 1.4923312664031982, "step": 40011 }, { "epoch": 3.154276704769413, "grad_norm": 0.45453619956970215, "learning_rate": 0.0005, "loss": 1.191788911819458, "step": 40012 }, { "epoch": 3.1543555380370516, "grad_norm": 0.4404609203338623, "learning_rate": 0.0005, "loss": 1.2115193605422974, "step": 40013 }, { "epoch": 3.1544343713046907, "grad_norm": 0.6063752174377441, "learning_rate": 0.0005, "loss": 1.4035215377807617, "step": 40014 }, { "epoch": 3.1545132045723294, "grad_norm": 0.4926011562347412, "learning_rate": 0.0005, "loss": 1.347416877746582, "step": 40015 }, { "epoch": 3.1545920378399686, "grad_norm": 0.4212152063846588, "learning_rate": 0.0005, "loss": 1.263077974319458, "step": 40016 }, { "epoch": 3.1546708711076072, "grad_norm": 0.6254001259803772, "learning_rate": 0.0005, "loss": 1.3575067520141602, "step": 40017 }, { "epoch": 3.1547497043752464, "grad_norm": 0.5229492783546448, "learning_rate": 0.0005, "loss": 1.2165701389312744, "step": 40018 }, { "epoch": 3.154828537642885, "grad_norm": 0.46570566296577454, "learning_rate": 0.0005, "loss": 1.281247615814209, "step": 40019 }, { "epoch": 3.154907370910524, "grad_norm": 0.47610098123550415, "learning_rate": 0.0005, "loss": 1.1653549671173096, "step": 40020 }, { "epoch": 3.1549862041781633, "grad_norm": 0.47970643639564514, "learning_rate": 0.0005, "loss": 1.175478219985962, "step": 40021 }, { "epoch": 3.155065037445802, "grad_norm": 0.5642004609107971, "learning_rate": 0.0005, "loss": 1.1903793811798096, "step": 40022 }, { "epoch": 3.155143870713441, "grad_norm": 0.5214231014251709, "learning_rate": 0.0005, "loss": 1.339097023010254, "step": 40023 }, { "epoch": 3.15522270398108, "grad_norm": 0.911931574344635, "learning_rate": 0.0005, "loss": 1.4816348552703857, "step": 40024 }, { "epoch": 3.155301537248719, "grad_norm": 0.43707987666130066, "learning_rate": 0.0005, "loss": 1.1248009204864502, "step": 40025 }, { "epoch": 3.155380370516358, "grad_norm": 0.49727001786231995, "learning_rate": 0.0005, "loss": 1.3971388339996338, "step": 40026 }, { "epoch": 3.155459203783997, "grad_norm": 0.4198341369628906, "learning_rate": 0.0005, "loss": 1.1355855464935303, "step": 40027 }, { "epoch": 3.155538037051636, "grad_norm": 0.5817051529884338, "learning_rate": 0.0005, "loss": 1.3650099039077759, "step": 40028 }, { "epoch": 3.1556168703192746, "grad_norm": 0.42042791843414307, "learning_rate": 0.0005, "loss": 1.320287823677063, "step": 40029 }, { "epoch": 3.1556957035869138, "grad_norm": 0.446540892124176, "learning_rate": 0.0005, "loss": 1.1948580741882324, "step": 40030 }, { "epoch": 3.1557745368545524, "grad_norm": 0.4599778652191162, "learning_rate": 0.0005, "loss": 1.310490608215332, "step": 40031 }, { "epoch": 3.1558533701221916, "grad_norm": 0.4557128846645355, "learning_rate": 0.0005, "loss": 1.1536556482315063, "step": 40032 }, { "epoch": 3.1559322033898307, "grad_norm": 0.46355536580085754, "learning_rate": 0.0005, "loss": 1.2075660228729248, "step": 40033 }, { "epoch": 3.1560110366574694, "grad_norm": 0.46775364875793457, "learning_rate": 0.0005, "loss": 1.2062448263168335, "step": 40034 }, { "epoch": 3.1560898699251085, "grad_norm": 0.59791100025177, "learning_rate": 0.0005, "loss": 1.1633431911468506, "step": 40035 }, { "epoch": 3.1561687031927472, "grad_norm": 0.4355785548686981, "learning_rate": 0.0005, "loss": 1.2313956022262573, "step": 40036 }, { "epoch": 3.1562475364603864, "grad_norm": 0.527809202671051, "learning_rate": 0.0005, "loss": 1.2782195806503296, "step": 40037 }, { "epoch": 3.156326369728025, "grad_norm": 0.4984205663204193, "learning_rate": 0.0005, "loss": 1.191896677017212, "step": 40038 }, { "epoch": 3.156405202995664, "grad_norm": 0.46100324392318726, "learning_rate": 0.0005, "loss": 1.1882827281951904, "step": 40039 }, { "epoch": 3.1564840362633033, "grad_norm": 0.7945282459259033, "learning_rate": 0.0005, "loss": 1.2977664470672607, "step": 40040 }, { "epoch": 3.156562869530942, "grad_norm": 0.47027966380119324, "learning_rate": 0.0005, "loss": 1.2448610067367554, "step": 40041 }, { "epoch": 3.156641702798581, "grad_norm": 0.4043344259262085, "learning_rate": 0.0005, "loss": 1.1202373504638672, "step": 40042 }, { "epoch": 3.15672053606622, "grad_norm": 0.500794529914856, "learning_rate": 0.0005, "loss": 1.1721745729446411, "step": 40043 }, { "epoch": 3.156799369333859, "grad_norm": 0.47885051369667053, "learning_rate": 0.0005, "loss": 1.3104572296142578, "step": 40044 }, { "epoch": 3.1568782026014977, "grad_norm": 0.5451642870903015, "learning_rate": 0.0005, "loss": 1.1201138496398926, "step": 40045 }, { "epoch": 3.156957035869137, "grad_norm": 0.4014230966567993, "learning_rate": 0.0005, "loss": 1.21687912940979, "step": 40046 }, { "epoch": 3.157035869136776, "grad_norm": 0.39727815985679626, "learning_rate": 0.0005, "loss": 1.2148637771606445, "step": 40047 }, { "epoch": 3.1571147024044146, "grad_norm": 0.4526599943637848, "learning_rate": 0.0005, "loss": 1.1606969833374023, "step": 40048 }, { "epoch": 3.1571935356720537, "grad_norm": 0.7278921604156494, "learning_rate": 0.0005, "loss": 1.4514033794403076, "step": 40049 }, { "epoch": 3.1572723689396924, "grad_norm": 0.4601750671863556, "learning_rate": 0.0005, "loss": 1.295203447341919, "step": 40050 }, { "epoch": 3.1573512022073316, "grad_norm": 0.41379910707473755, "learning_rate": 0.0005, "loss": 1.0497591495513916, "step": 40051 }, { "epoch": 3.1574300354749703, "grad_norm": 0.5615060925483704, "learning_rate": 0.0005, "loss": 1.3040082454681396, "step": 40052 }, { "epoch": 3.1575088687426094, "grad_norm": 0.47272321581840515, "learning_rate": 0.0005, "loss": 1.254963994026184, "step": 40053 }, { "epoch": 3.1575877020102485, "grad_norm": 0.5462450385093689, "learning_rate": 0.0005, "loss": 1.2173309326171875, "step": 40054 }, { "epoch": 3.157666535277887, "grad_norm": 0.525511622428894, "learning_rate": 0.0005, "loss": 1.5403022766113281, "step": 40055 }, { "epoch": 3.1577453685455263, "grad_norm": 0.4989663362503052, "learning_rate": 0.0005, "loss": 1.3750537633895874, "step": 40056 }, { "epoch": 3.157824201813165, "grad_norm": 0.5034905672073364, "learning_rate": 0.0005, "loss": 1.3204550743103027, "step": 40057 }, { "epoch": 3.157903035080804, "grad_norm": 0.5279297232627869, "learning_rate": 0.0005, "loss": 1.3415160179138184, "step": 40058 }, { "epoch": 3.157981868348443, "grad_norm": 0.49468082189559937, "learning_rate": 0.0005, "loss": 1.2828590869903564, "step": 40059 }, { "epoch": 3.158060701616082, "grad_norm": 0.6583394408226013, "learning_rate": 0.0005, "loss": 1.315324068069458, "step": 40060 }, { "epoch": 3.158139534883721, "grad_norm": 0.538395881652832, "learning_rate": 0.0005, "loss": 1.2853493690490723, "step": 40061 }, { "epoch": 3.15821836815136, "grad_norm": 0.4901168644428253, "learning_rate": 0.0005, "loss": 1.1895384788513184, "step": 40062 }, { "epoch": 3.158297201418999, "grad_norm": 0.5109186172485352, "learning_rate": 0.0005, "loss": 1.2692897319793701, "step": 40063 }, { "epoch": 3.1583760346866376, "grad_norm": 0.5025815963745117, "learning_rate": 0.0005, "loss": 1.3112592697143555, "step": 40064 }, { "epoch": 3.1584548679542768, "grad_norm": 0.4905223846435547, "learning_rate": 0.0005, "loss": 1.4302202463150024, "step": 40065 }, { "epoch": 3.1585337012219155, "grad_norm": 0.6277375817298889, "learning_rate": 0.0005, "loss": 1.2583866119384766, "step": 40066 }, { "epoch": 3.1586125344895546, "grad_norm": 0.6191432476043701, "learning_rate": 0.0005, "loss": 1.4150428771972656, "step": 40067 }, { "epoch": 3.1586913677571937, "grad_norm": 0.388131320476532, "learning_rate": 0.0005, "loss": 1.2064614295959473, "step": 40068 }, { "epoch": 3.1587702010248324, "grad_norm": 0.6010722517967224, "learning_rate": 0.0005, "loss": 1.3622031211853027, "step": 40069 }, { "epoch": 3.1588490342924715, "grad_norm": 0.5542824864387512, "learning_rate": 0.0005, "loss": 1.2041091918945312, "step": 40070 }, { "epoch": 3.1589278675601102, "grad_norm": 0.5292606949806213, "learning_rate": 0.0005, "loss": 1.3269529342651367, "step": 40071 }, { "epoch": 3.1590067008277494, "grad_norm": 0.570719301700592, "learning_rate": 0.0005, "loss": 1.1815311908721924, "step": 40072 }, { "epoch": 3.159085534095388, "grad_norm": 0.46748265624046326, "learning_rate": 0.0005, "loss": 1.2027778625488281, "step": 40073 }, { "epoch": 3.159164367363027, "grad_norm": 0.43171700835227966, "learning_rate": 0.0005, "loss": 1.2447233200073242, "step": 40074 }, { "epoch": 3.1592432006306663, "grad_norm": 0.5796396136283875, "learning_rate": 0.0005, "loss": 1.2914081811904907, "step": 40075 }, { "epoch": 3.159322033898305, "grad_norm": 0.5330964922904968, "learning_rate": 0.0005, "loss": 1.2863661050796509, "step": 40076 }, { "epoch": 3.159400867165944, "grad_norm": 0.4970775544643402, "learning_rate": 0.0005, "loss": 1.322594165802002, "step": 40077 }, { "epoch": 3.159479700433583, "grad_norm": 0.5222551226615906, "learning_rate": 0.0005, "loss": 1.3869526386260986, "step": 40078 }, { "epoch": 3.159558533701222, "grad_norm": 0.5237962007522583, "learning_rate": 0.0005, "loss": 1.2692631483078003, "step": 40079 }, { "epoch": 3.1596373669688607, "grad_norm": 0.5334227681159973, "learning_rate": 0.0005, "loss": 1.514117956161499, "step": 40080 }, { "epoch": 3.1597162002365, "grad_norm": 0.5196874737739563, "learning_rate": 0.0005, "loss": 1.1659786701202393, "step": 40081 }, { "epoch": 3.159795033504139, "grad_norm": 0.39249515533447266, "learning_rate": 0.0005, "loss": 1.032181978225708, "step": 40082 }, { "epoch": 3.1598738667717776, "grad_norm": 0.4585658311843872, "learning_rate": 0.0005, "loss": 1.1294939517974854, "step": 40083 }, { "epoch": 3.1599527000394168, "grad_norm": 0.441449373960495, "learning_rate": 0.0005, "loss": 1.1644212007522583, "step": 40084 }, { "epoch": 3.1600315333070554, "grad_norm": 0.5129815340042114, "learning_rate": 0.0005, "loss": 1.1001546382904053, "step": 40085 }, { "epoch": 3.1601103665746946, "grad_norm": 0.4666096270084381, "learning_rate": 0.0005, "loss": 1.4178218841552734, "step": 40086 }, { "epoch": 3.1601891998423333, "grad_norm": 0.5344333648681641, "learning_rate": 0.0005, "loss": 1.5553605556488037, "step": 40087 }, { "epoch": 3.1602680331099724, "grad_norm": 0.4605514705181122, "learning_rate": 0.0005, "loss": 1.235302209854126, "step": 40088 }, { "epoch": 3.1603468663776115, "grad_norm": 0.4221894443035126, "learning_rate": 0.0005, "loss": 1.1816933155059814, "step": 40089 }, { "epoch": 3.16042569964525, "grad_norm": 0.4534260332584381, "learning_rate": 0.0005, "loss": 1.294575572013855, "step": 40090 }, { "epoch": 3.1605045329128894, "grad_norm": 0.503834068775177, "learning_rate": 0.0005, "loss": 1.2847177982330322, "step": 40091 }, { "epoch": 3.160583366180528, "grad_norm": 0.4687321186065674, "learning_rate": 0.0005, "loss": 1.4769189357757568, "step": 40092 }, { "epoch": 3.160662199448167, "grad_norm": 0.39448070526123047, "learning_rate": 0.0005, "loss": 1.224776029586792, "step": 40093 }, { "epoch": 3.160741032715806, "grad_norm": 0.7995356321334839, "learning_rate": 0.0005, "loss": 1.1322147846221924, "step": 40094 }, { "epoch": 3.160819865983445, "grad_norm": 0.49460965394973755, "learning_rate": 0.0005, "loss": 1.1509333848953247, "step": 40095 }, { "epoch": 3.160898699251084, "grad_norm": 0.4778231680393219, "learning_rate": 0.0005, "loss": 1.250266432762146, "step": 40096 }, { "epoch": 3.160977532518723, "grad_norm": 0.47160854935646057, "learning_rate": 0.0005, "loss": 1.2852907180786133, "step": 40097 }, { "epoch": 3.161056365786362, "grad_norm": 0.5314450860023499, "learning_rate": 0.0005, "loss": 1.460439682006836, "step": 40098 }, { "epoch": 3.1611351990540006, "grad_norm": 0.5746621489524841, "learning_rate": 0.0005, "loss": 1.02913498878479, "step": 40099 }, { "epoch": 3.1612140323216398, "grad_norm": 0.43790510296821594, "learning_rate": 0.0005, "loss": 1.2467247247695923, "step": 40100 }, { "epoch": 3.161292865589279, "grad_norm": 0.4231204688549042, "learning_rate": 0.0005, "loss": 1.2732059955596924, "step": 40101 }, { "epoch": 3.1613716988569176, "grad_norm": 0.4793110191822052, "learning_rate": 0.0005, "loss": 1.3501298427581787, "step": 40102 }, { "epoch": 3.1614505321245567, "grad_norm": 0.42806509137153625, "learning_rate": 0.0005, "loss": 1.2878742218017578, "step": 40103 }, { "epoch": 3.1615293653921954, "grad_norm": 0.5065123438835144, "learning_rate": 0.0005, "loss": 1.2296098470687866, "step": 40104 }, { "epoch": 3.1616081986598346, "grad_norm": 0.7962709069252014, "learning_rate": 0.0005, "loss": 1.3220043182373047, "step": 40105 }, { "epoch": 3.1616870319274732, "grad_norm": 0.4643732011318207, "learning_rate": 0.0005, "loss": 1.2082780599594116, "step": 40106 }, { "epoch": 3.1617658651951124, "grad_norm": 0.6602360010147095, "learning_rate": 0.0005, "loss": 1.1238453388214111, "step": 40107 }, { "epoch": 3.161844698462751, "grad_norm": 0.4134905934333801, "learning_rate": 0.0005, "loss": 1.139110803604126, "step": 40108 }, { "epoch": 3.16192353173039, "grad_norm": 0.46946829557418823, "learning_rate": 0.0005, "loss": 1.290299654006958, "step": 40109 }, { "epoch": 3.1620023649980293, "grad_norm": 0.4852382242679596, "learning_rate": 0.0005, "loss": 1.1657147407531738, "step": 40110 }, { "epoch": 3.162081198265668, "grad_norm": 0.4694349467754364, "learning_rate": 0.0005, "loss": 1.1036298274993896, "step": 40111 }, { "epoch": 3.162160031533307, "grad_norm": 0.7658026218414307, "learning_rate": 0.0005, "loss": 1.1984624862670898, "step": 40112 }, { "epoch": 3.162238864800946, "grad_norm": 0.4637118875980377, "learning_rate": 0.0005, "loss": 1.2012526988983154, "step": 40113 }, { "epoch": 3.162317698068585, "grad_norm": 0.48640429973602295, "learning_rate": 0.0005, "loss": 1.3841649293899536, "step": 40114 }, { "epoch": 3.162396531336224, "grad_norm": 0.41512173414230347, "learning_rate": 0.0005, "loss": 1.272176742553711, "step": 40115 }, { "epoch": 3.162475364603863, "grad_norm": 0.5251187682151794, "learning_rate": 0.0005, "loss": 1.2002677917480469, "step": 40116 }, { "epoch": 3.162554197871502, "grad_norm": 0.5017082095146179, "learning_rate": 0.0005, "loss": 1.2855093479156494, "step": 40117 }, { "epoch": 3.1626330311391406, "grad_norm": 0.45928606390953064, "learning_rate": 0.0005, "loss": 1.08381986618042, "step": 40118 }, { "epoch": 3.1627118644067798, "grad_norm": 0.6236018538475037, "learning_rate": 0.0005, "loss": 1.3600809574127197, "step": 40119 }, { "epoch": 3.1627906976744184, "grad_norm": 0.5007973909378052, "learning_rate": 0.0005, "loss": 1.3255157470703125, "step": 40120 }, { "epoch": 3.1628695309420576, "grad_norm": 0.5479406714439392, "learning_rate": 0.0005, "loss": 1.368308663368225, "step": 40121 }, { "epoch": 3.1629483642096963, "grad_norm": 0.4782806932926178, "learning_rate": 0.0005, "loss": 1.302143931388855, "step": 40122 }, { "epoch": 3.1630271974773354, "grad_norm": 0.6713998317718506, "learning_rate": 0.0005, "loss": 1.4039790630340576, "step": 40123 }, { "epoch": 3.1631060307449745, "grad_norm": 0.45648542046546936, "learning_rate": 0.0005, "loss": 1.2672752141952515, "step": 40124 }, { "epoch": 3.1631848640126132, "grad_norm": 0.6386526823043823, "learning_rate": 0.0005, "loss": 1.4590840339660645, "step": 40125 }, { "epoch": 3.1632636972802524, "grad_norm": 0.5218209028244019, "learning_rate": 0.0005, "loss": 1.3081138134002686, "step": 40126 }, { "epoch": 3.163342530547891, "grad_norm": 0.4546782970428467, "learning_rate": 0.0005, "loss": 1.3441216945648193, "step": 40127 }, { "epoch": 3.16342136381553, "grad_norm": 0.5447259545326233, "learning_rate": 0.0005, "loss": 1.2064155340194702, "step": 40128 }, { "epoch": 3.1635001970831693, "grad_norm": 0.4039882719516754, "learning_rate": 0.0005, "loss": 1.2484166622161865, "step": 40129 }, { "epoch": 3.163579030350808, "grad_norm": 0.4577229917049408, "learning_rate": 0.0005, "loss": 1.3263413906097412, "step": 40130 }, { "epoch": 3.163657863618447, "grad_norm": 0.5923717617988586, "learning_rate": 0.0005, "loss": 1.2095913887023926, "step": 40131 }, { "epoch": 3.163736696886086, "grad_norm": 0.5077263116836548, "learning_rate": 0.0005, "loss": 1.4352374076843262, "step": 40132 }, { "epoch": 3.163815530153725, "grad_norm": 0.5092824101448059, "learning_rate": 0.0005, "loss": 1.279405951499939, "step": 40133 }, { "epoch": 3.1638943634213637, "grad_norm": 0.4636360704898834, "learning_rate": 0.0005, "loss": 1.3390264511108398, "step": 40134 }, { "epoch": 3.163973196689003, "grad_norm": 0.4892065227031708, "learning_rate": 0.0005, "loss": 1.3774112462997437, "step": 40135 }, { "epoch": 3.1640520299566415, "grad_norm": 0.47466376423835754, "learning_rate": 0.0005, "loss": 1.1727044582366943, "step": 40136 }, { "epoch": 3.1641308632242806, "grad_norm": 0.5815948247909546, "learning_rate": 0.0005, "loss": 1.2936612367630005, "step": 40137 }, { "epoch": 3.1642096964919197, "grad_norm": 0.48564547300338745, "learning_rate": 0.0005, "loss": 1.1987895965576172, "step": 40138 }, { "epoch": 3.1642885297595584, "grad_norm": 0.48727285861968994, "learning_rate": 0.0005, "loss": 1.3606293201446533, "step": 40139 }, { "epoch": 3.1643673630271976, "grad_norm": 0.45583558082580566, "learning_rate": 0.0005, "loss": 1.2174053192138672, "step": 40140 }, { "epoch": 3.1644461962948363, "grad_norm": 0.49608397483825684, "learning_rate": 0.0005, "loss": 1.26433527469635, "step": 40141 }, { "epoch": 3.1645250295624754, "grad_norm": 1.094117283821106, "learning_rate": 0.0005, "loss": 1.1457864046096802, "step": 40142 }, { "epoch": 3.1646038628301145, "grad_norm": 0.43133604526519775, "learning_rate": 0.0005, "loss": 1.2074241638183594, "step": 40143 }, { "epoch": 3.164682696097753, "grad_norm": 0.42537152767181396, "learning_rate": 0.0005, "loss": 1.3964589834213257, "step": 40144 }, { "epoch": 3.1647615293653923, "grad_norm": 0.3965696394443512, "learning_rate": 0.0005, "loss": 1.1001713275909424, "step": 40145 }, { "epoch": 3.164840362633031, "grad_norm": 0.43206602334976196, "learning_rate": 0.0005, "loss": 1.1300032138824463, "step": 40146 }, { "epoch": 3.16491919590067, "grad_norm": 0.5617576837539673, "learning_rate": 0.0005, "loss": 1.266404390335083, "step": 40147 }, { "epoch": 3.164998029168309, "grad_norm": 0.7003365755081177, "learning_rate": 0.0005, "loss": 1.391018271446228, "step": 40148 }, { "epoch": 3.165076862435948, "grad_norm": 0.42419037222862244, "learning_rate": 0.0005, "loss": 1.2404851913452148, "step": 40149 }, { "epoch": 3.1651556957035867, "grad_norm": 0.5575304627418518, "learning_rate": 0.0005, "loss": 1.222365140914917, "step": 40150 }, { "epoch": 3.165234528971226, "grad_norm": 0.5116924047470093, "learning_rate": 0.0005, "loss": 1.293611764907837, "step": 40151 }, { "epoch": 3.165313362238865, "grad_norm": 0.5598807334899902, "learning_rate": 0.0005, "loss": 1.3094806671142578, "step": 40152 }, { "epoch": 3.1653921955065036, "grad_norm": 0.5941779017448425, "learning_rate": 0.0005, "loss": 1.2788572311401367, "step": 40153 }, { "epoch": 3.1654710287741428, "grad_norm": 0.5120550394058228, "learning_rate": 0.0005, "loss": 1.326028823852539, "step": 40154 }, { "epoch": 3.1655498620417815, "grad_norm": 0.44445300102233887, "learning_rate": 0.0005, "loss": 1.2686097621917725, "step": 40155 }, { "epoch": 3.1656286953094206, "grad_norm": 0.4511074423789978, "learning_rate": 0.0005, "loss": 1.216534972190857, "step": 40156 }, { "epoch": 3.1657075285770597, "grad_norm": 0.4896881580352783, "learning_rate": 0.0005, "loss": 1.28781259059906, "step": 40157 }, { "epoch": 3.1657863618446984, "grad_norm": 0.45295974612236023, "learning_rate": 0.0005, "loss": 1.1658172607421875, "step": 40158 }, { "epoch": 3.1658651951123375, "grad_norm": 0.5082204937934875, "learning_rate": 0.0005, "loss": 1.2734487056732178, "step": 40159 }, { "epoch": 3.1659440283799762, "grad_norm": 0.5937237739562988, "learning_rate": 0.0005, "loss": 1.3586318492889404, "step": 40160 }, { "epoch": 3.1660228616476154, "grad_norm": 0.4897784888744354, "learning_rate": 0.0005, "loss": 1.2916924953460693, "step": 40161 }, { "epoch": 3.166101694915254, "grad_norm": 0.43464696407318115, "learning_rate": 0.0005, "loss": 1.121651291847229, "step": 40162 }, { "epoch": 3.166180528182893, "grad_norm": 0.5037164092063904, "learning_rate": 0.0005, "loss": 1.465553879737854, "step": 40163 }, { "epoch": 3.1662593614505323, "grad_norm": 0.47007590532302856, "learning_rate": 0.0005, "loss": 1.181377649307251, "step": 40164 }, { "epoch": 3.166338194718171, "grad_norm": 0.5863820910453796, "learning_rate": 0.0005, "loss": 1.3045117855072021, "step": 40165 }, { "epoch": 3.16641702798581, "grad_norm": 0.48432233929634094, "learning_rate": 0.0005, "loss": 1.133763074874878, "step": 40166 }, { "epoch": 3.166495861253449, "grad_norm": 0.4861101508140564, "learning_rate": 0.0005, "loss": 1.4323694705963135, "step": 40167 }, { "epoch": 3.166574694521088, "grad_norm": 0.5744621753692627, "learning_rate": 0.0005, "loss": 1.4120997190475464, "step": 40168 }, { "epoch": 3.1666535277887267, "grad_norm": 0.4818229675292969, "learning_rate": 0.0005, "loss": 1.370936393737793, "step": 40169 }, { "epoch": 3.166732361056366, "grad_norm": 0.4634433388710022, "learning_rate": 0.0005, "loss": 1.3496043682098389, "step": 40170 }, { "epoch": 3.166811194324005, "grad_norm": 0.5216231942176819, "learning_rate": 0.0005, "loss": 1.3730461597442627, "step": 40171 }, { "epoch": 3.1668900275916436, "grad_norm": 0.4405357539653778, "learning_rate": 0.0005, "loss": 1.3459103107452393, "step": 40172 }, { "epoch": 3.1669688608592828, "grad_norm": 0.48742973804473877, "learning_rate": 0.0005, "loss": 1.2038007974624634, "step": 40173 }, { "epoch": 3.1670476941269214, "grad_norm": 0.38483861088752747, "learning_rate": 0.0005, "loss": 1.1127216815948486, "step": 40174 }, { "epoch": 3.1671265273945606, "grad_norm": 0.5498495101928711, "learning_rate": 0.0005, "loss": 1.3049136400222778, "step": 40175 }, { "epoch": 3.1672053606621993, "grad_norm": 0.43625178933143616, "learning_rate": 0.0005, "loss": 1.1170570850372314, "step": 40176 }, { "epoch": 3.1672841939298384, "grad_norm": 0.45629602670669556, "learning_rate": 0.0005, "loss": 1.2811315059661865, "step": 40177 }, { "epoch": 3.1673630271974775, "grad_norm": 0.5035341382026672, "learning_rate": 0.0005, "loss": 1.2045283317565918, "step": 40178 }, { "epoch": 3.167441860465116, "grad_norm": 0.6292280554771423, "learning_rate": 0.0005, "loss": 1.4809672832489014, "step": 40179 }, { "epoch": 3.1675206937327554, "grad_norm": 0.39136821031570435, "learning_rate": 0.0005, "loss": 1.2110142707824707, "step": 40180 }, { "epoch": 3.167599527000394, "grad_norm": 0.4165860712528229, "learning_rate": 0.0005, "loss": 1.1256351470947266, "step": 40181 }, { "epoch": 3.167678360268033, "grad_norm": 0.45092350244522095, "learning_rate": 0.0005, "loss": 1.148747205734253, "step": 40182 }, { "epoch": 3.167757193535672, "grad_norm": 0.4926093816757202, "learning_rate": 0.0005, "loss": 1.4762513637542725, "step": 40183 }, { "epoch": 3.167836026803311, "grad_norm": 0.6252551674842834, "learning_rate": 0.0005, "loss": 1.2666234970092773, "step": 40184 }, { "epoch": 3.16791486007095, "grad_norm": 0.5689696073532104, "learning_rate": 0.0005, "loss": 1.1712576150894165, "step": 40185 }, { "epoch": 3.167993693338589, "grad_norm": 0.4497593939304352, "learning_rate": 0.0005, "loss": 1.2306941747665405, "step": 40186 }, { "epoch": 3.168072526606228, "grad_norm": 0.6776871085166931, "learning_rate": 0.0005, "loss": 1.3751964569091797, "step": 40187 }, { "epoch": 3.1681513598738666, "grad_norm": 0.5914639830589294, "learning_rate": 0.0005, "loss": 1.2881908416748047, "step": 40188 }, { "epoch": 3.168230193141506, "grad_norm": 0.44617465138435364, "learning_rate": 0.0005, "loss": 1.2855722904205322, "step": 40189 }, { "epoch": 3.1683090264091445, "grad_norm": 0.45238643884658813, "learning_rate": 0.0005, "loss": 1.1264886856079102, "step": 40190 }, { "epoch": 3.1683878596767836, "grad_norm": 0.4879949986934662, "learning_rate": 0.0005, "loss": 1.2409262657165527, "step": 40191 }, { "epoch": 3.1684666929444227, "grad_norm": 0.5520868897438049, "learning_rate": 0.0005, "loss": 1.3868136405944824, "step": 40192 }, { "epoch": 3.1685455262120614, "grad_norm": 0.5165181756019592, "learning_rate": 0.0005, "loss": 1.2622246742248535, "step": 40193 }, { "epoch": 3.1686243594797006, "grad_norm": 0.5319223403930664, "learning_rate": 0.0005, "loss": 1.27388596534729, "step": 40194 }, { "epoch": 3.1687031927473392, "grad_norm": 0.4467349052429199, "learning_rate": 0.0005, "loss": 1.2924304008483887, "step": 40195 }, { "epoch": 3.1687820260149784, "grad_norm": 0.5149832367897034, "learning_rate": 0.0005, "loss": 1.3551957607269287, "step": 40196 }, { "epoch": 3.168860859282617, "grad_norm": 0.48951560258865356, "learning_rate": 0.0005, "loss": 1.2944867610931396, "step": 40197 }, { "epoch": 3.168939692550256, "grad_norm": 0.5923048257827759, "learning_rate": 0.0005, "loss": 1.3147038221359253, "step": 40198 }, { "epoch": 3.1690185258178953, "grad_norm": 0.3924154043197632, "learning_rate": 0.0005, "loss": 1.1455672979354858, "step": 40199 }, { "epoch": 3.169097359085534, "grad_norm": 0.45181378722190857, "learning_rate": 0.0005, "loss": 1.5242915153503418, "step": 40200 }, { "epoch": 3.169176192353173, "grad_norm": 0.5314849615097046, "learning_rate": 0.0005, "loss": 1.2718744277954102, "step": 40201 }, { "epoch": 3.169255025620812, "grad_norm": 0.5208360552787781, "learning_rate": 0.0005, "loss": 1.207153558731079, "step": 40202 }, { "epoch": 3.169333858888451, "grad_norm": 0.48926180601119995, "learning_rate": 0.0005, "loss": 1.3590099811553955, "step": 40203 }, { "epoch": 3.1694126921560897, "grad_norm": 0.6451011300086975, "learning_rate": 0.0005, "loss": 1.3814148902893066, "step": 40204 }, { "epoch": 3.169491525423729, "grad_norm": 0.45491349697113037, "learning_rate": 0.0005, "loss": 1.1400766372680664, "step": 40205 }, { "epoch": 3.169570358691368, "grad_norm": 0.511879563331604, "learning_rate": 0.0005, "loss": 1.270522117614746, "step": 40206 }, { "epoch": 3.1696491919590066, "grad_norm": 0.7238955497741699, "learning_rate": 0.0005, "loss": 1.2944023609161377, "step": 40207 }, { "epoch": 3.1697280252266458, "grad_norm": 0.599232017993927, "learning_rate": 0.0005, "loss": 1.2157111167907715, "step": 40208 }, { "epoch": 3.1698068584942845, "grad_norm": 0.48140469193458557, "learning_rate": 0.0005, "loss": 1.2933096885681152, "step": 40209 }, { "epoch": 3.1698856917619236, "grad_norm": 0.5405158996582031, "learning_rate": 0.0005, "loss": 1.341853380203247, "step": 40210 }, { "epoch": 3.1699645250295623, "grad_norm": 0.823630690574646, "learning_rate": 0.0005, "loss": 1.2951009273529053, "step": 40211 }, { "epoch": 3.1700433582972014, "grad_norm": 0.46758317947387695, "learning_rate": 0.0005, "loss": 1.245277762413025, "step": 40212 }, { "epoch": 3.1701221915648405, "grad_norm": 0.43538525700569153, "learning_rate": 0.0005, "loss": 1.2606450319290161, "step": 40213 }, { "epoch": 3.1702010248324792, "grad_norm": 0.464892715215683, "learning_rate": 0.0005, "loss": 1.180718183517456, "step": 40214 }, { "epoch": 3.1702798581001184, "grad_norm": 0.4891849458217621, "learning_rate": 0.0005, "loss": 1.191253662109375, "step": 40215 }, { "epoch": 3.170358691367757, "grad_norm": 1.6009670495986938, "learning_rate": 0.0005, "loss": 1.1997389793395996, "step": 40216 }, { "epoch": 3.170437524635396, "grad_norm": 0.46944552659988403, "learning_rate": 0.0005, "loss": 1.412271499633789, "step": 40217 }, { "epoch": 3.170516357903035, "grad_norm": 0.4831334054470062, "learning_rate": 0.0005, "loss": 1.2196645736694336, "step": 40218 }, { "epoch": 3.170595191170674, "grad_norm": 0.4640752077102661, "learning_rate": 0.0005, "loss": 1.3185606002807617, "step": 40219 }, { "epoch": 3.170674024438313, "grad_norm": 0.4400756359100342, "learning_rate": 0.0005, "loss": 1.1410037279129028, "step": 40220 }, { "epoch": 3.170752857705952, "grad_norm": 0.4752141535282135, "learning_rate": 0.0005, "loss": 1.212308645248413, "step": 40221 }, { "epoch": 3.170831690973591, "grad_norm": 0.6615800857543945, "learning_rate": 0.0005, "loss": 1.3105418682098389, "step": 40222 }, { "epoch": 3.1709105242412297, "grad_norm": 0.5143493413925171, "learning_rate": 0.0005, "loss": 1.159132480621338, "step": 40223 }, { "epoch": 3.170989357508869, "grad_norm": 0.47197744250297546, "learning_rate": 0.0005, "loss": 1.167036771774292, "step": 40224 }, { "epoch": 3.1710681907765075, "grad_norm": 1.1298972368240356, "learning_rate": 0.0005, "loss": 1.3817815780639648, "step": 40225 }, { "epoch": 3.1711470240441466, "grad_norm": 0.4412859380245209, "learning_rate": 0.0005, "loss": 1.2205654382705688, "step": 40226 }, { "epoch": 3.1712258573117857, "grad_norm": 0.6309797167778015, "learning_rate": 0.0005, "loss": 1.2432925701141357, "step": 40227 }, { "epoch": 3.1713046905794244, "grad_norm": 0.42121216654777527, "learning_rate": 0.0005, "loss": 1.1196553707122803, "step": 40228 }, { "epoch": 3.1713835238470636, "grad_norm": 0.46092694997787476, "learning_rate": 0.0005, "loss": 1.1547436714172363, "step": 40229 }, { "epoch": 3.1714623571147023, "grad_norm": 0.49643731117248535, "learning_rate": 0.0005, "loss": 1.3618967533111572, "step": 40230 }, { "epoch": 3.1715411903823414, "grad_norm": 0.46247243881225586, "learning_rate": 0.0005, "loss": 1.3413468599319458, "step": 40231 }, { "epoch": 3.1716200236499805, "grad_norm": 0.48538488149642944, "learning_rate": 0.0005, "loss": 1.1526880264282227, "step": 40232 }, { "epoch": 3.171698856917619, "grad_norm": 0.46002310514450073, "learning_rate": 0.0005, "loss": 1.2960827350616455, "step": 40233 }, { "epoch": 3.1717776901852583, "grad_norm": 0.5792748332023621, "learning_rate": 0.0005, "loss": 1.170632004737854, "step": 40234 }, { "epoch": 3.171856523452897, "grad_norm": 0.4479539096355438, "learning_rate": 0.0005, "loss": 1.2004053592681885, "step": 40235 }, { "epoch": 3.171935356720536, "grad_norm": 0.5902001261711121, "learning_rate": 0.0005, "loss": 1.3438117504119873, "step": 40236 }, { "epoch": 3.172014189988175, "grad_norm": 0.5474779605865479, "learning_rate": 0.0005, "loss": 1.4141566753387451, "step": 40237 }, { "epoch": 3.172093023255814, "grad_norm": 0.5385911464691162, "learning_rate": 0.0005, "loss": 1.29799485206604, "step": 40238 }, { "epoch": 3.1721718565234527, "grad_norm": 0.3972218632698059, "learning_rate": 0.0005, "loss": 1.0569552183151245, "step": 40239 }, { "epoch": 3.172250689791092, "grad_norm": 0.5869596600532532, "learning_rate": 0.0005, "loss": 1.4187917709350586, "step": 40240 }, { "epoch": 3.172329523058731, "grad_norm": 0.5045170783996582, "learning_rate": 0.0005, "loss": 1.1360886096954346, "step": 40241 }, { "epoch": 3.1724083563263696, "grad_norm": 0.4854266047477722, "learning_rate": 0.0005, "loss": 1.2816033363342285, "step": 40242 }, { "epoch": 3.1724871895940088, "grad_norm": 0.5064186453819275, "learning_rate": 0.0005, "loss": 1.1749298572540283, "step": 40243 }, { "epoch": 3.1725660228616475, "grad_norm": 0.49256786704063416, "learning_rate": 0.0005, "loss": 1.3788954019546509, "step": 40244 }, { "epoch": 3.1726448561292866, "grad_norm": 0.49055710434913635, "learning_rate": 0.0005, "loss": 1.1961244344711304, "step": 40245 }, { "epoch": 3.1727236893969257, "grad_norm": 0.3978559374809265, "learning_rate": 0.0005, "loss": 1.2712485790252686, "step": 40246 }, { "epoch": 3.1728025226645644, "grad_norm": 0.519285261631012, "learning_rate": 0.0005, "loss": 1.460279941558838, "step": 40247 }, { "epoch": 3.1728813559322036, "grad_norm": 0.5086483955383301, "learning_rate": 0.0005, "loss": 1.2204701900482178, "step": 40248 }, { "epoch": 3.1729601891998422, "grad_norm": 0.48658356070518494, "learning_rate": 0.0005, "loss": 1.116220235824585, "step": 40249 }, { "epoch": 3.1730390224674814, "grad_norm": 0.5189359784126282, "learning_rate": 0.0005, "loss": 1.2785141468048096, "step": 40250 }, { "epoch": 3.17311785573512, "grad_norm": 0.7640384435653687, "learning_rate": 0.0005, "loss": 1.3903725147247314, "step": 40251 }, { "epoch": 3.173196689002759, "grad_norm": 0.5228119492530823, "learning_rate": 0.0005, "loss": 1.2084962129592896, "step": 40252 }, { "epoch": 3.173275522270398, "grad_norm": 0.47419556975364685, "learning_rate": 0.0005, "loss": 1.2956116199493408, "step": 40253 }, { "epoch": 3.173354355538037, "grad_norm": 0.4356115162372589, "learning_rate": 0.0005, "loss": 1.2317733764648438, "step": 40254 }, { "epoch": 3.173433188805676, "grad_norm": 0.4038485884666443, "learning_rate": 0.0005, "loss": 1.1289756298065186, "step": 40255 }, { "epoch": 3.173512022073315, "grad_norm": 0.5733393430709839, "learning_rate": 0.0005, "loss": 1.245556354522705, "step": 40256 }, { "epoch": 3.173590855340954, "grad_norm": 0.475314199924469, "learning_rate": 0.0005, "loss": 1.3456487655639648, "step": 40257 }, { "epoch": 3.1736696886085927, "grad_norm": 0.461702823638916, "learning_rate": 0.0005, "loss": 1.3112354278564453, "step": 40258 }, { "epoch": 3.173748521876232, "grad_norm": 0.5228312611579895, "learning_rate": 0.0005, "loss": 1.1820684671401978, "step": 40259 }, { "epoch": 3.173827355143871, "grad_norm": 0.47787216305732727, "learning_rate": 0.0005, "loss": 1.15571928024292, "step": 40260 }, { "epoch": 3.1739061884115096, "grad_norm": 0.42547643184661865, "learning_rate": 0.0005, "loss": 1.072829008102417, "step": 40261 }, { "epoch": 3.1739850216791488, "grad_norm": 0.4638102054595947, "learning_rate": 0.0005, "loss": 1.2975038290023804, "step": 40262 }, { "epoch": 3.1740638549467874, "grad_norm": 0.4160965383052826, "learning_rate": 0.0005, "loss": 1.2834396362304688, "step": 40263 }, { "epoch": 3.1741426882144266, "grad_norm": 0.531014084815979, "learning_rate": 0.0005, "loss": 1.2455888986587524, "step": 40264 }, { "epoch": 3.1742215214820653, "grad_norm": 0.5042694807052612, "learning_rate": 0.0005, "loss": 1.286034345626831, "step": 40265 }, { "epoch": 3.1743003547497044, "grad_norm": 0.4876142740249634, "learning_rate": 0.0005, "loss": 1.2936365604400635, "step": 40266 }, { "epoch": 3.174379188017343, "grad_norm": 0.465096116065979, "learning_rate": 0.0005, "loss": 1.1768333911895752, "step": 40267 }, { "epoch": 3.1744580212849822, "grad_norm": 0.5212160348892212, "learning_rate": 0.0005, "loss": 1.2795183658599854, "step": 40268 }, { "epoch": 3.1745368545526214, "grad_norm": 0.48200470209121704, "learning_rate": 0.0005, "loss": 1.2429896593093872, "step": 40269 }, { "epoch": 3.17461568782026, "grad_norm": 0.4550228714942932, "learning_rate": 0.0005, "loss": 1.3614299297332764, "step": 40270 }, { "epoch": 3.174694521087899, "grad_norm": 0.5314401984214783, "learning_rate": 0.0005, "loss": 1.3232340812683105, "step": 40271 }, { "epoch": 3.174773354355538, "grad_norm": 0.5138487219810486, "learning_rate": 0.0005, "loss": 1.4715521335601807, "step": 40272 }, { "epoch": 3.174852187623177, "grad_norm": 0.4835594594478607, "learning_rate": 0.0005, "loss": 1.1954190731048584, "step": 40273 }, { "epoch": 3.174931020890816, "grad_norm": 0.6141279339790344, "learning_rate": 0.0005, "loss": 1.4448657035827637, "step": 40274 }, { "epoch": 3.175009854158455, "grad_norm": 0.44104695320129395, "learning_rate": 0.0005, "loss": 1.1215648651123047, "step": 40275 }, { "epoch": 3.175088687426094, "grad_norm": 0.6183782815933228, "learning_rate": 0.0005, "loss": 1.1853176355361938, "step": 40276 }, { "epoch": 3.1751675206937326, "grad_norm": 0.4818720817565918, "learning_rate": 0.0005, "loss": 1.1351717710494995, "step": 40277 }, { "epoch": 3.175246353961372, "grad_norm": 0.5171075463294983, "learning_rate": 0.0005, "loss": 1.2598838806152344, "step": 40278 }, { "epoch": 3.1753251872290105, "grad_norm": 0.42255720496177673, "learning_rate": 0.0005, "loss": 1.1929320096969604, "step": 40279 }, { "epoch": 3.1754040204966496, "grad_norm": 0.5139532089233398, "learning_rate": 0.0005, "loss": 1.1819238662719727, "step": 40280 }, { "epoch": 3.1754828537642887, "grad_norm": 0.3876491189002991, "learning_rate": 0.0005, "loss": 1.0867881774902344, "step": 40281 }, { "epoch": 3.1755616870319274, "grad_norm": 0.5728195309638977, "learning_rate": 0.0005, "loss": 1.4426391124725342, "step": 40282 }, { "epoch": 3.1756405202995666, "grad_norm": 0.4974670708179474, "learning_rate": 0.0005, "loss": 1.2259416580200195, "step": 40283 }, { "epoch": 3.1757193535672052, "grad_norm": 0.5048918724060059, "learning_rate": 0.0005, "loss": 1.1110167503356934, "step": 40284 }, { "epoch": 3.1757981868348444, "grad_norm": 0.5072154402732849, "learning_rate": 0.0005, "loss": 1.1609450578689575, "step": 40285 }, { "epoch": 3.175877020102483, "grad_norm": 0.5535417795181274, "learning_rate": 0.0005, "loss": 1.1945819854736328, "step": 40286 }, { "epoch": 3.175955853370122, "grad_norm": 0.5512798428535461, "learning_rate": 0.0005, "loss": 1.3489048480987549, "step": 40287 }, { "epoch": 3.1760346866377613, "grad_norm": 0.40721026062965393, "learning_rate": 0.0005, "loss": 1.1270098686218262, "step": 40288 }, { "epoch": 3.1761135199054, "grad_norm": 0.529388964176178, "learning_rate": 0.0005, "loss": 1.42596435546875, "step": 40289 }, { "epoch": 3.176192353173039, "grad_norm": 0.4693519175052643, "learning_rate": 0.0005, "loss": 1.1773364543914795, "step": 40290 }, { "epoch": 3.176271186440678, "grad_norm": 0.46533527970314026, "learning_rate": 0.0005, "loss": 1.1395145654678345, "step": 40291 }, { "epoch": 3.176350019708317, "grad_norm": 0.4661436676979065, "learning_rate": 0.0005, "loss": 1.3005088567733765, "step": 40292 }, { "epoch": 3.1764288529759557, "grad_norm": 0.8978545069694519, "learning_rate": 0.0005, "loss": 1.3903793096542358, "step": 40293 }, { "epoch": 3.176507686243595, "grad_norm": 0.47210434079170227, "learning_rate": 0.0005, "loss": 1.32672119140625, "step": 40294 }, { "epoch": 3.176586519511234, "grad_norm": 0.5161442160606384, "learning_rate": 0.0005, "loss": 1.1722606420516968, "step": 40295 }, { "epoch": 3.1766653527788726, "grad_norm": 0.5835766196250916, "learning_rate": 0.0005, "loss": 1.2622300386428833, "step": 40296 }, { "epoch": 3.1767441860465118, "grad_norm": 0.5112143754959106, "learning_rate": 0.0005, "loss": 1.1217514276504517, "step": 40297 }, { "epoch": 3.1768230193141505, "grad_norm": 0.621615469455719, "learning_rate": 0.0005, "loss": 1.2762970924377441, "step": 40298 }, { "epoch": 3.1769018525817896, "grad_norm": 0.4967564344406128, "learning_rate": 0.0005, "loss": 1.3042747974395752, "step": 40299 }, { "epoch": 3.1769806858494283, "grad_norm": 0.4711277186870575, "learning_rate": 0.0005, "loss": 1.1656627655029297, "step": 40300 }, { "epoch": 3.1770595191170674, "grad_norm": 0.4839176833629608, "learning_rate": 0.0005, "loss": 1.420186996459961, "step": 40301 }, { "epoch": 3.1771383523847065, "grad_norm": 0.45010700821876526, "learning_rate": 0.0005, "loss": 1.2445895671844482, "step": 40302 }, { "epoch": 3.1772171856523452, "grad_norm": 0.4318591058254242, "learning_rate": 0.0005, "loss": 1.3016061782836914, "step": 40303 }, { "epoch": 3.1772960189199844, "grad_norm": 0.5246853232383728, "learning_rate": 0.0005, "loss": 1.4266884326934814, "step": 40304 }, { "epoch": 3.177374852187623, "grad_norm": 0.4390321671962738, "learning_rate": 0.0005, "loss": 1.2213622331619263, "step": 40305 }, { "epoch": 3.177453685455262, "grad_norm": 0.5587955713272095, "learning_rate": 0.0005, "loss": 1.3160368204116821, "step": 40306 }, { "epoch": 3.177532518722901, "grad_norm": 0.4442296028137207, "learning_rate": 0.0005, "loss": 1.2403959035873413, "step": 40307 }, { "epoch": 3.17761135199054, "grad_norm": 0.4613252580165863, "learning_rate": 0.0005, "loss": 1.2473242282867432, "step": 40308 }, { "epoch": 3.177690185258179, "grad_norm": 0.5637354254722595, "learning_rate": 0.0005, "loss": 1.1339552402496338, "step": 40309 }, { "epoch": 3.177769018525818, "grad_norm": 0.47773611545562744, "learning_rate": 0.0005, "loss": 1.3154141902923584, "step": 40310 }, { "epoch": 3.177847851793457, "grad_norm": 0.46991080045700073, "learning_rate": 0.0005, "loss": 1.1204286813735962, "step": 40311 }, { "epoch": 3.1779266850610957, "grad_norm": 0.49777236580848694, "learning_rate": 0.0005, "loss": 1.3386436700820923, "step": 40312 }, { "epoch": 3.178005518328735, "grad_norm": 0.4968351125717163, "learning_rate": 0.0005, "loss": 1.0904269218444824, "step": 40313 }, { "epoch": 3.1780843515963735, "grad_norm": 0.5149489045143127, "learning_rate": 0.0005, "loss": 1.420578956604004, "step": 40314 }, { "epoch": 3.1781631848640126, "grad_norm": 0.43369367718696594, "learning_rate": 0.0005, "loss": 1.2618377208709717, "step": 40315 }, { "epoch": 3.1782420181316517, "grad_norm": 0.5289878845214844, "learning_rate": 0.0005, "loss": 1.3201955556869507, "step": 40316 }, { "epoch": 3.1783208513992904, "grad_norm": 0.5648883581161499, "learning_rate": 0.0005, "loss": 1.197082757949829, "step": 40317 }, { "epoch": 3.1783996846669296, "grad_norm": 0.46812230348587036, "learning_rate": 0.0005, "loss": 1.4457801580429077, "step": 40318 }, { "epoch": 3.1784785179345683, "grad_norm": 0.5101628303527832, "learning_rate": 0.0005, "loss": 1.2261298894882202, "step": 40319 }, { "epoch": 3.1785573512022074, "grad_norm": 0.41138142347335815, "learning_rate": 0.0005, "loss": 1.1483089923858643, "step": 40320 }, { "epoch": 3.178636184469846, "grad_norm": 0.5841816663742065, "learning_rate": 0.0005, "loss": 1.1160331964492798, "step": 40321 }, { "epoch": 3.178715017737485, "grad_norm": 0.5437864065170288, "learning_rate": 0.0005, "loss": 1.2788612842559814, "step": 40322 }, { "epoch": 3.1787938510051243, "grad_norm": 0.544398307800293, "learning_rate": 0.0005, "loss": 1.4661436080932617, "step": 40323 }, { "epoch": 3.178872684272763, "grad_norm": 0.45650017261505127, "learning_rate": 0.0005, "loss": 1.1142683029174805, "step": 40324 }, { "epoch": 3.178951517540402, "grad_norm": 0.5920630693435669, "learning_rate": 0.0005, "loss": 1.2285888195037842, "step": 40325 }, { "epoch": 3.179030350808041, "grad_norm": 0.4466896653175354, "learning_rate": 0.0005, "loss": 1.1086496114730835, "step": 40326 }, { "epoch": 3.17910918407568, "grad_norm": 0.3858180344104767, "learning_rate": 0.0005, "loss": 1.2130331993103027, "step": 40327 }, { "epoch": 3.1791880173433187, "grad_norm": 0.4507543742656708, "learning_rate": 0.0005, "loss": 1.1186991930007935, "step": 40328 }, { "epoch": 3.179266850610958, "grad_norm": 0.4807794392108917, "learning_rate": 0.0005, "loss": 1.0940604209899902, "step": 40329 }, { "epoch": 3.179345683878597, "grad_norm": 0.39296087622642517, "learning_rate": 0.0005, "loss": 1.1751601696014404, "step": 40330 }, { "epoch": 3.1794245171462356, "grad_norm": 0.45942747592926025, "learning_rate": 0.0005, "loss": 1.2064534425735474, "step": 40331 }, { "epoch": 3.1795033504138748, "grad_norm": 0.5079858303070068, "learning_rate": 0.0005, "loss": 1.3768951892852783, "step": 40332 }, { "epoch": 3.1795821836815135, "grad_norm": 0.49873727560043335, "learning_rate": 0.0005, "loss": 1.2635067701339722, "step": 40333 }, { "epoch": 3.1796610169491526, "grad_norm": 0.6084414720535278, "learning_rate": 0.0005, "loss": 1.434493899345398, "step": 40334 }, { "epoch": 3.1797398502167913, "grad_norm": 0.45547181367874146, "learning_rate": 0.0005, "loss": 1.2202948331832886, "step": 40335 }, { "epoch": 3.1798186834844304, "grad_norm": 0.4253889322280884, "learning_rate": 0.0005, "loss": 1.1258561611175537, "step": 40336 }, { "epoch": 3.1798975167520696, "grad_norm": 0.47116371989250183, "learning_rate": 0.0005, "loss": 1.2932116985321045, "step": 40337 }, { "epoch": 3.1799763500197082, "grad_norm": 0.5441059470176697, "learning_rate": 0.0005, "loss": 1.1699740886688232, "step": 40338 }, { "epoch": 3.1800551832873474, "grad_norm": 0.500846266746521, "learning_rate": 0.0005, "loss": 1.3207576274871826, "step": 40339 }, { "epoch": 3.180134016554986, "grad_norm": 0.5790202021598816, "learning_rate": 0.0005, "loss": 1.5895007848739624, "step": 40340 }, { "epoch": 3.180212849822625, "grad_norm": 0.43460991978645325, "learning_rate": 0.0005, "loss": 1.1756172180175781, "step": 40341 }, { "epoch": 3.180291683090264, "grad_norm": 0.5642719864845276, "learning_rate": 0.0005, "loss": 1.2146551609039307, "step": 40342 }, { "epoch": 3.180370516357903, "grad_norm": 0.6058875322341919, "learning_rate": 0.0005, "loss": 1.4941377639770508, "step": 40343 }, { "epoch": 3.180449349625542, "grad_norm": 0.6057689785957336, "learning_rate": 0.0005, "loss": 1.4586290121078491, "step": 40344 }, { "epoch": 3.180528182893181, "grad_norm": 0.4884141981601715, "learning_rate": 0.0005, "loss": 1.2853574752807617, "step": 40345 }, { "epoch": 3.18060701616082, "grad_norm": 0.4991169571876526, "learning_rate": 0.0005, "loss": 1.3420684337615967, "step": 40346 }, { "epoch": 3.1806858494284587, "grad_norm": 0.44460028409957886, "learning_rate": 0.0005, "loss": 1.1737463474273682, "step": 40347 }, { "epoch": 3.180764682696098, "grad_norm": 0.43580958247184753, "learning_rate": 0.0005, "loss": 1.2529776096343994, "step": 40348 }, { "epoch": 3.1808435159637365, "grad_norm": 0.5477449893951416, "learning_rate": 0.0005, "loss": 1.4171591997146606, "step": 40349 }, { "epoch": 3.1809223492313756, "grad_norm": 0.46604955196380615, "learning_rate": 0.0005, "loss": 1.2744693756103516, "step": 40350 }, { "epoch": 3.1810011824990148, "grad_norm": 0.4149838089942932, "learning_rate": 0.0005, "loss": 1.207674264907837, "step": 40351 }, { "epoch": 3.1810800157666534, "grad_norm": 0.5443339347839355, "learning_rate": 0.0005, "loss": 1.3114945888519287, "step": 40352 }, { "epoch": 3.1811588490342926, "grad_norm": 0.5254064798355103, "learning_rate": 0.0005, "loss": 1.1646716594696045, "step": 40353 }, { "epoch": 3.1812376823019313, "grad_norm": 0.5167798399925232, "learning_rate": 0.0005, "loss": 1.1519081592559814, "step": 40354 }, { "epoch": 3.1813165155695704, "grad_norm": 0.44488856196403503, "learning_rate": 0.0005, "loss": 1.3122975826263428, "step": 40355 }, { "epoch": 3.181395348837209, "grad_norm": 0.459987998008728, "learning_rate": 0.0005, "loss": 1.171281099319458, "step": 40356 }, { "epoch": 3.1814741821048482, "grad_norm": 0.4735908806324005, "learning_rate": 0.0005, "loss": 1.2706760168075562, "step": 40357 }, { "epoch": 3.1815530153724874, "grad_norm": 0.4522530436515808, "learning_rate": 0.0005, "loss": 1.2211058139801025, "step": 40358 }, { "epoch": 3.181631848640126, "grad_norm": 0.539609432220459, "learning_rate": 0.0005, "loss": 1.250382423400879, "step": 40359 }, { "epoch": 3.181710681907765, "grad_norm": 0.5297664403915405, "learning_rate": 0.0005, "loss": 1.3676674365997314, "step": 40360 }, { "epoch": 3.181789515175404, "grad_norm": 0.49438807368278503, "learning_rate": 0.0005, "loss": 1.4013322591781616, "step": 40361 }, { "epoch": 3.181868348443043, "grad_norm": 0.40648627281188965, "learning_rate": 0.0005, "loss": 1.2995185852050781, "step": 40362 }, { "epoch": 3.181947181710682, "grad_norm": 0.5167205333709717, "learning_rate": 0.0005, "loss": 1.2305479049682617, "step": 40363 }, { "epoch": 3.182026014978321, "grad_norm": 0.44015583395957947, "learning_rate": 0.0005, "loss": 1.1931378841400146, "step": 40364 }, { "epoch": 3.18210484824596, "grad_norm": 0.6874813437461853, "learning_rate": 0.0005, "loss": 1.2689802646636963, "step": 40365 }, { "epoch": 3.1821836815135986, "grad_norm": 0.49192002415657043, "learning_rate": 0.0005, "loss": 1.3712908029556274, "step": 40366 }, { "epoch": 3.182262514781238, "grad_norm": 0.3673006594181061, "learning_rate": 0.0005, "loss": 1.151466727256775, "step": 40367 }, { "epoch": 3.1823413480488765, "grad_norm": 0.5052450299263, "learning_rate": 0.0005, "loss": 1.358245611190796, "step": 40368 }, { "epoch": 3.1824201813165156, "grad_norm": 0.4940345883369446, "learning_rate": 0.0005, "loss": 1.1905615329742432, "step": 40369 }, { "epoch": 3.1824990145841543, "grad_norm": 0.6024975776672363, "learning_rate": 0.0005, "loss": 1.371104121208191, "step": 40370 }, { "epoch": 3.1825778478517934, "grad_norm": 0.5328425765037537, "learning_rate": 0.0005, "loss": 1.145660400390625, "step": 40371 }, { "epoch": 3.1826566811194326, "grad_norm": 0.46873268485069275, "learning_rate": 0.0005, "loss": 1.1954748630523682, "step": 40372 }, { "epoch": 3.1827355143870713, "grad_norm": 0.42076343297958374, "learning_rate": 0.0005, "loss": 1.1380205154418945, "step": 40373 }, { "epoch": 3.1828143476547104, "grad_norm": 0.4211304187774658, "learning_rate": 0.0005, "loss": 1.0156136751174927, "step": 40374 }, { "epoch": 3.182893180922349, "grad_norm": 0.4607575535774231, "learning_rate": 0.0005, "loss": 1.1596981287002563, "step": 40375 }, { "epoch": 3.182972014189988, "grad_norm": 0.43996813893318176, "learning_rate": 0.0005, "loss": 1.30634605884552, "step": 40376 }, { "epoch": 3.1830508474576273, "grad_norm": 0.5358755588531494, "learning_rate": 0.0005, "loss": 1.396028757095337, "step": 40377 }, { "epoch": 3.183129680725266, "grad_norm": 0.3869467079639435, "learning_rate": 0.0005, "loss": 1.2088438272476196, "step": 40378 }, { "epoch": 3.183208513992905, "grad_norm": 0.46117928624153137, "learning_rate": 0.0005, "loss": 1.1393020153045654, "step": 40379 }, { "epoch": 3.183287347260544, "grad_norm": 0.47824233770370483, "learning_rate": 0.0005, "loss": 1.183471918106079, "step": 40380 }, { "epoch": 3.183366180528183, "grad_norm": 0.47077226638793945, "learning_rate": 0.0005, "loss": 1.3256226778030396, "step": 40381 }, { "epoch": 3.1834450137958217, "grad_norm": 0.6898158192634583, "learning_rate": 0.0005, "loss": 1.1544640064239502, "step": 40382 }, { "epoch": 3.183523847063461, "grad_norm": 0.47102028131484985, "learning_rate": 0.0005, "loss": 1.3430166244506836, "step": 40383 }, { "epoch": 3.1836026803310995, "grad_norm": 0.4680933952331543, "learning_rate": 0.0005, "loss": 1.2527328729629517, "step": 40384 }, { "epoch": 3.1836815135987386, "grad_norm": 0.507692277431488, "learning_rate": 0.0005, "loss": 1.2133678197860718, "step": 40385 }, { "epoch": 3.1837603468663778, "grad_norm": 0.6313535571098328, "learning_rate": 0.0005, "loss": 1.320965051651001, "step": 40386 }, { "epoch": 3.1838391801340165, "grad_norm": 0.4726494550704956, "learning_rate": 0.0005, "loss": 1.3388817310333252, "step": 40387 }, { "epoch": 3.1839180134016556, "grad_norm": 0.5270622968673706, "learning_rate": 0.0005, "loss": 1.333693504333496, "step": 40388 }, { "epoch": 3.1839968466692943, "grad_norm": 0.4622407555580139, "learning_rate": 0.0005, "loss": 1.1548179388046265, "step": 40389 }, { "epoch": 3.1840756799369334, "grad_norm": 0.4139353334903717, "learning_rate": 0.0005, "loss": 1.1947839260101318, "step": 40390 }, { "epoch": 3.1841545132045725, "grad_norm": 0.5533555746078491, "learning_rate": 0.0005, "loss": 1.181344747543335, "step": 40391 }, { "epoch": 3.1842333464722112, "grad_norm": 0.44627606868743896, "learning_rate": 0.0005, "loss": 1.3135501146316528, "step": 40392 }, { "epoch": 3.1843121797398504, "grad_norm": 0.47846344113349915, "learning_rate": 0.0005, "loss": 1.4226831197738647, "step": 40393 }, { "epoch": 3.184391013007489, "grad_norm": 0.4104459881782532, "learning_rate": 0.0005, "loss": 1.2114492654800415, "step": 40394 }, { "epoch": 3.184469846275128, "grad_norm": 0.6262210607528687, "learning_rate": 0.0005, "loss": 1.305625081062317, "step": 40395 }, { "epoch": 3.184548679542767, "grad_norm": 0.5059931874275208, "learning_rate": 0.0005, "loss": 1.2698695659637451, "step": 40396 }, { "epoch": 3.184627512810406, "grad_norm": 0.5786153078079224, "learning_rate": 0.0005, "loss": 1.3711135387420654, "step": 40397 }, { "epoch": 3.1847063460780447, "grad_norm": 0.5061435699462891, "learning_rate": 0.0005, "loss": 1.388451337814331, "step": 40398 }, { "epoch": 3.184785179345684, "grad_norm": 0.5277723670005798, "learning_rate": 0.0005, "loss": 1.3888037204742432, "step": 40399 }, { "epoch": 3.184864012613323, "grad_norm": 0.43320101499557495, "learning_rate": 0.0005, "loss": 1.2707549333572388, "step": 40400 }, { "epoch": 3.1849428458809617, "grad_norm": 0.4801972210407257, "learning_rate": 0.0005, "loss": 1.1922850608825684, "step": 40401 }, { "epoch": 3.185021679148601, "grad_norm": 0.5640478730201721, "learning_rate": 0.0005, "loss": 1.2959468364715576, "step": 40402 }, { "epoch": 3.1851005124162395, "grad_norm": 0.5337197184562683, "learning_rate": 0.0005, "loss": 1.3113665580749512, "step": 40403 }, { "epoch": 3.1851793456838786, "grad_norm": 0.5574564337730408, "learning_rate": 0.0005, "loss": 1.348421335220337, "step": 40404 }, { "epoch": 3.1852581789515177, "grad_norm": 0.446697860956192, "learning_rate": 0.0005, "loss": 1.23871648311615, "step": 40405 }, { "epoch": 3.1853370122191564, "grad_norm": 0.5177463293075562, "learning_rate": 0.0005, "loss": 1.2017508745193481, "step": 40406 }, { "epoch": 3.1854158454867956, "grad_norm": 0.4896810054779053, "learning_rate": 0.0005, "loss": 1.1329436302185059, "step": 40407 }, { "epoch": 3.1854946787544343, "grad_norm": 0.6101018786430359, "learning_rate": 0.0005, "loss": 1.1577394008636475, "step": 40408 }, { "epoch": 3.1855735120220734, "grad_norm": 0.7969638109207153, "learning_rate": 0.0005, "loss": 1.3387377262115479, "step": 40409 }, { "epoch": 3.185652345289712, "grad_norm": 0.5360296964645386, "learning_rate": 0.0005, "loss": 1.220806360244751, "step": 40410 }, { "epoch": 3.185731178557351, "grad_norm": 0.4016564190387726, "learning_rate": 0.0005, "loss": 1.1648002862930298, "step": 40411 }, { "epoch": 3.1858100118249904, "grad_norm": 0.44237038493156433, "learning_rate": 0.0005, "loss": 1.2718486785888672, "step": 40412 }, { "epoch": 3.185888845092629, "grad_norm": 0.5501341819763184, "learning_rate": 0.0005, "loss": 1.3554624319076538, "step": 40413 }, { "epoch": 3.185967678360268, "grad_norm": 0.48478078842163086, "learning_rate": 0.0005, "loss": 1.3318555355072021, "step": 40414 }, { "epoch": 3.186046511627907, "grad_norm": 0.4232301414012909, "learning_rate": 0.0005, "loss": 1.1666502952575684, "step": 40415 }, { "epoch": 3.186125344895546, "grad_norm": 0.3942136764526367, "learning_rate": 0.0005, "loss": 1.185204267501831, "step": 40416 }, { "epoch": 3.1862041781631847, "grad_norm": 0.4626106023788452, "learning_rate": 0.0005, "loss": 1.3863741159439087, "step": 40417 }, { "epoch": 3.186283011430824, "grad_norm": 0.5360300540924072, "learning_rate": 0.0005, "loss": 1.379380226135254, "step": 40418 }, { "epoch": 3.186361844698463, "grad_norm": 0.39924779534339905, "learning_rate": 0.0005, "loss": 1.2181545495986938, "step": 40419 }, { "epoch": 3.1864406779661016, "grad_norm": 0.37755948305130005, "learning_rate": 0.0005, "loss": 1.1263964176177979, "step": 40420 }, { "epoch": 3.1865195112337408, "grad_norm": 0.39137551188468933, "learning_rate": 0.0005, "loss": 1.0154305696487427, "step": 40421 }, { "epoch": 3.1865983445013795, "grad_norm": 0.42586129903793335, "learning_rate": 0.0005, "loss": 1.0962612628936768, "step": 40422 }, { "epoch": 3.1866771777690186, "grad_norm": 0.5671952366828918, "learning_rate": 0.0005, "loss": 1.2205381393432617, "step": 40423 }, { "epoch": 3.1867560110366573, "grad_norm": 0.47744521498680115, "learning_rate": 0.0005, "loss": 1.1079190969467163, "step": 40424 }, { "epoch": 3.1868348443042964, "grad_norm": 0.8481087684631348, "learning_rate": 0.0005, "loss": 1.2625701427459717, "step": 40425 }, { "epoch": 3.1869136775719356, "grad_norm": 0.5023144483566284, "learning_rate": 0.0005, "loss": 1.1698801517486572, "step": 40426 }, { "epoch": 3.1869925108395742, "grad_norm": 0.6431583762168884, "learning_rate": 0.0005, "loss": 1.329185962677002, "step": 40427 }, { "epoch": 3.1870713441072134, "grad_norm": 0.41993996500968933, "learning_rate": 0.0005, "loss": 1.2845959663391113, "step": 40428 }, { "epoch": 3.187150177374852, "grad_norm": 0.45248210430145264, "learning_rate": 0.0005, "loss": 1.2288795709609985, "step": 40429 }, { "epoch": 3.187229010642491, "grad_norm": 0.647786021232605, "learning_rate": 0.0005, "loss": 1.4611449241638184, "step": 40430 }, { "epoch": 3.18730784391013, "grad_norm": 0.40758758783340454, "learning_rate": 0.0005, "loss": 1.1721371412277222, "step": 40431 }, { "epoch": 3.187386677177769, "grad_norm": 0.43507763743400574, "learning_rate": 0.0005, "loss": 1.3302643299102783, "step": 40432 }, { "epoch": 3.187465510445408, "grad_norm": 0.5725922584533691, "learning_rate": 0.0005, "loss": 1.383324146270752, "step": 40433 }, { "epoch": 3.187544343713047, "grad_norm": 0.3995359539985657, "learning_rate": 0.0005, "loss": 1.221184253692627, "step": 40434 }, { "epoch": 3.187623176980686, "grad_norm": 0.44582799077033997, "learning_rate": 0.0005, "loss": 1.1575167179107666, "step": 40435 }, { "epoch": 3.1877020102483247, "grad_norm": 0.43888482451438904, "learning_rate": 0.0005, "loss": 1.1856287717819214, "step": 40436 }, { "epoch": 3.187780843515964, "grad_norm": 0.6002542972564697, "learning_rate": 0.0005, "loss": 1.2108945846557617, "step": 40437 }, { "epoch": 3.1878596767836025, "grad_norm": 0.5215674042701721, "learning_rate": 0.0005, "loss": 1.2356107234954834, "step": 40438 }, { "epoch": 3.1879385100512416, "grad_norm": 0.5469601154327393, "learning_rate": 0.0005, "loss": 1.1564207077026367, "step": 40439 }, { "epoch": 3.1880173433188808, "grad_norm": 0.5182216167449951, "learning_rate": 0.0005, "loss": 1.1922767162322998, "step": 40440 }, { "epoch": 3.1880961765865194, "grad_norm": 0.6243627667427063, "learning_rate": 0.0005, "loss": 1.3643362522125244, "step": 40441 }, { "epoch": 3.1881750098541586, "grad_norm": 0.5207158923149109, "learning_rate": 0.0005, "loss": 1.2198282480239868, "step": 40442 }, { "epoch": 3.1882538431217973, "grad_norm": 0.4652569591999054, "learning_rate": 0.0005, "loss": 1.1210405826568604, "step": 40443 }, { "epoch": 3.1883326763894364, "grad_norm": 0.5515194535255432, "learning_rate": 0.0005, "loss": 1.374159336090088, "step": 40444 }, { "epoch": 3.188411509657075, "grad_norm": 0.4466422200202942, "learning_rate": 0.0005, "loss": 1.1726279258728027, "step": 40445 }, { "epoch": 3.1884903429247142, "grad_norm": 0.4419076442718506, "learning_rate": 0.0005, "loss": 1.163243293762207, "step": 40446 }, { "epoch": 3.1885691761923534, "grad_norm": 0.4447271227836609, "learning_rate": 0.0005, "loss": 1.1893415451049805, "step": 40447 }, { "epoch": 3.188648009459992, "grad_norm": 0.49342671036720276, "learning_rate": 0.0005, "loss": 1.292411208152771, "step": 40448 }, { "epoch": 3.188726842727631, "grad_norm": 0.5703242421150208, "learning_rate": 0.0005, "loss": 1.337085485458374, "step": 40449 }, { "epoch": 3.18880567599527, "grad_norm": 0.46221381425857544, "learning_rate": 0.0005, "loss": 1.3485807180404663, "step": 40450 }, { "epoch": 3.188884509262909, "grad_norm": 0.4099090099334717, "learning_rate": 0.0005, "loss": 1.1286386251449585, "step": 40451 }, { "epoch": 3.1889633425305477, "grad_norm": 0.7024586200714111, "learning_rate": 0.0005, "loss": 1.1829266548156738, "step": 40452 }, { "epoch": 3.189042175798187, "grad_norm": 0.6012717485427856, "learning_rate": 0.0005, "loss": 1.4604485034942627, "step": 40453 }, { "epoch": 3.189121009065826, "grad_norm": 0.4206457734107971, "learning_rate": 0.0005, "loss": 1.1313672065734863, "step": 40454 }, { "epoch": 3.1891998423334647, "grad_norm": 0.6383999586105347, "learning_rate": 0.0005, "loss": 1.4349322319030762, "step": 40455 }, { "epoch": 3.189278675601104, "grad_norm": 0.46737411618232727, "learning_rate": 0.0005, "loss": 1.3438133001327515, "step": 40456 }, { "epoch": 3.1893575088687425, "grad_norm": 0.4306635856628418, "learning_rate": 0.0005, "loss": 1.319098711013794, "step": 40457 }, { "epoch": 3.1894363421363816, "grad_norm": 0.5341996550559998, "learning_rate": 0.0005, "loss": 1.358850121498108, "step": 40458 }, { "epoch": 3.1895151754040203, "grad_norm": 0.4094924032688141, "learning_rate": 0.0005, "loss": 1.13466477394104, "step": 40459 }, { "epoch": 3.1895940086716594, "grad_norm": 0.4422849416732788, "learning_rate": 0.0005, "loss": 1.2258237600326538, "step": 40460 }, { "epoch": 3.1896728419392986, "grad_norm": 0.6057919263839722, "learning_rate": 0.0005, "loss": 1.176202416419983, "step": 40461 }, { "epoch": 3.1897516752069373, "grad_norm": 0.45859843492507935, "learning_rate": 0.0005, "loss": 1.2767856121063232, "step": 40462 }, { "epoch": 3.1898305084745764, "grad_norm": 0.409958153963089, "learning_rate": 0.0005, "loss": 1.0981829166412354, "step": 40463 }, { "epoch": 3.189909341742215, "grad_norm": 0.5661889314651489, "learning_rate": 0.0005, "loss": 1.1306519508361816, "step": 40464 }, { "epoch": 3.189988175009854, "grad_norm": 0.48159846663475037, "learning_rate": 0.0005, "loss": 1.1740880012512207, "step": 40465 }, { "epoch": 3.190067008277493, "grad_norm": 0.4311576187610626, "learning_rate": 0.0005, "loss": 1.1987411975860596, "step": 40466 }, { "epoch": 3.190145841545132, "grad_norm": 0.4071813225746155, "learning_rate": 0.0005, "loss": 1.0909221172332764, "step": 40467 }, { "epoch": 3.190224674812771, "grad_norm": 0.5030445456504822, "learning_rate": 0.0005, "loss": 1.1610583066940308, "step": 40468 }, { "epoch": 3.19030350808041, "grad_norm": 0.49784716963768005, "learning_rate": 0.0005, "loss": 1.3708784580230713, "step": 40469 }, { "epoch": 3.190382341348049, "grad_norm": 0.4607623815536499, "learning_rate": 0.0005, "loss": 1.2751917839050293, "step": 40470 }, { "epoch": 3.1904611746156877, "grad_norm": 0.503410816192627, "learning_rate": 0.0005, "loss": 1.1450979709625244, "step": 40471 }, { "epoch": 3.190540007883327, "grad_norm": 0.48194125294685364, "learning_rate": 0.0005, "loss": 1.1008944511413574, "step": 40472 }, { "epoch": 3.1906188411509655, "grad_norm": 0.5314260721206665, "learning_rate": 0.0005, "loss": 1.2012901306152344, "step": 40473 }, { "epoch": 3.1906976744186046, "grad_norm": 0.6236432790756226, "learning_rate": 0.0005, "loss": 1.26239812374115, "step": 40474 }, { "epoch": 3.1907765076862438, "grad_norm": 0.4882456362247467, "learning_rate": 0.0005, "loss": 1.3032673597335815, "step": 40475 }, { "epoch": 3.1908553409538825, "grad_norm": 0.5022043585777283, "learning_rate": 0.0005, "loss": 1.3417863845825195, "step": 40476 }, { "epoch": 3.1909341742215216, "grad_norm": 0.48078620433807373, "learning_rate": 0.0005, "loss": 1.0842194557189941, "step": 40477 }, { "epoch": 3.1910130074891603, "grad_norm": 0.5328305959701538, "learning_rate": 0.0005, "loss": 1.405449628829956, "step": 40478 }, { "epoch": 3.1910918407567994, "grad_norm": 0.5261589288711548, "learning_rate": 0.0005, "loss": 1.552417516708374, "step": 40479 }, { "epoch": 3.1911706740244385, "grad_norm": 0.4418122172355652, "learning_rate": 0.0005, "loss": 1.183638572692871, "step": 40480 }, { "epoch": 3.1912495072920772, "grad_norm": 0.44689804315567017, "learning_rate": 0.0005, "loss": 1.1303133964538574, "step": 40481 }, { "epoch": 3.1913283405597164, "grad_norm": 0.5101523399353027, "learning_rate": 0.0005, "loss": 1.2023565769195557, "step": 40482 }, { "epoch": 3.191407173827355, "grad_norm": 0.46999800205230713, "learning_rate": 0.0005, "loss": 1.2127625942230225, "step": 40483 }, { "epoch": 3.191486007094994, "grad_norm": 0.4917103946208954, "learning_rate": 0.0005, "loss": 1.3729853630065918, "step": 40484 }, { "epoch": 3.191564840362633, "grad_norm": 0.3755139112472534, "learning_rate": 0.0005, "loss": 1.1575931310653687, "step": 40485 }, { "epoch": 3.191643673630272, "grad_norm": 0.6425283551216125, "learning_rate": 0.0005, "loss": 1.3441221714019775, "step": 40486 }, { "epoch": 3.1917225068979107, "grad_norm": 0.40443721413612366, "learning_rate": 0.0005, "loss": 1.1623308658599854, "step": 40487 }, { "epoch": 3.19180134016555, "grad_norm": 0.48169830441474915, "learning_rate": 0.0005, "loss": 1.2548658847808838, "step": 40488 }, { "epoch": 3.191880173433189, "grad_norm": 0.5748241543769836, "learning_rate": 0.0005, "loss": 1.3289916515350342, "step": 40489 }, { "epoch": 3.1919590067008277, "grad_norm": 0.48500069975852966, "learning_rate": 0.0005, "loss": 1.18685781955719, "step": 40490 }, { "epoch": 3.192037839968467, "grad_norm": 0.43865150213241577, "learning_rate": 0.0005, "loss": 1.1904029846191406, "step": 40491 }, { "epoch": 3.1921166732361055, "grad_norm": 0.524355411529541, "learning_rate": 0.0005, "loss": 1.2573107481002808, "step": 40492 }, { "epoch": 3.1921955065037446, "grad_norm": 0.6167153716087341, "learning_rate": 0.0005, "loss": 1.2886221408843994, "step": 40493 }, { "epoch": 3.1922743397713838, "grad_norm": 0.45767727494239807, "learning_rate": 0.0005, "loss": 1.1458570957183838, "step": 40494 }, { "epoch": 3.1923531730390224, "grad_norm": 0.8230823874473572, "learning_rate": 0.0005, "loss": 1.2596547603607178, "step": 40495 }, { "epoch": 3.1924320063066616, "grad_norm": 0.47125867009162903, "learning_rate": 0.0005, "loss": 1.1742181777954102, "step": 40496 }, { "epoch": 3.1925108395743003, "grad_norm": 0.5409095883369446, "learning_rate": 0.0005, "loss": 1.243699550628662, "step": 40497 }, { "epoch": 3.1925896728419394, "grad_norm": 0.48425859212875366, "learning_rate": 0.0005, "loss": 1.2476484775543213, "step": 40498 }, { "epoch": 3.192668506109578, "grad_norm": 0.4690171480178833, "learning_rate": 0.0005, "loss": 1.424977421760559, "step": 40499 }, { "epoch": 3.192747339377217, "grad_norm": 0.4243658781051636, "learning_rate": 0.0005, "loss": 1.3348082304000854, "step": 40500 }, { "epoch": 3.192826172644856, "grad_norm": 0.42453011870384216, "learning_rate": 0.0005, "loss": 1.2814719676971436, "step": 40501 }, { "epoch": 3.192905005912495, "grad_norm": 0.4782258868217468, "learning_rate": 0.0005, "loss": 1.2645611763000488, "step": 40502 }, { "epoch": 3.192983839180134, "grad_norm": 0.4175955653190613, "learning_rate": 0.0005, "loss": 1.1568443775177002, "step": 40503 }, { "epoch": 3.193062672447773, "grad_norm": 0.4660438001155853, "learning_rate": 0.0005, "loss": 1.304227590560913, "step": 40504 }, { "epoch": 3.193141505715412, "grad_norm": 0.5414239168167114, "learning_rate": 0.0005, "loss": 1.322150468826294, "step": 40505 }, { "epoch": 3.1932203389830507, "grad_norm": 0.4355517625808716, "learning_rate": 0.0005, "loss": 1.1218822002410889, "step": 40506 }, { "epoch": 3.19329917225069, "grad_norm": 0.5223482251167297, "learning_rate": 0.0005, "loss": 1.4554262161254883, "step": 40507 }, { "epoch": 3.193378005518329, "grad_norm": 0.5416609048843384, "learning_rate": 0.0005, "loss": 1.3539496660232544, "step": 40508 }, { "epoch": 3.1934568387859676, "grad_norm": 0.4887795150279999, "learning_rate": 0.0005, "loss": 1.2217066287994385, "step": 40509 }, { "epoch": 3.1935356720536068, "grad_norm": 0.43235325813293457, "learning_rate": 0.0005, "loss": 1.2013494968414307, "step": 40510 }, { "epoch": 3.1936145053212455, "grad_norm": 0.5504346489906311, "learning_rate": 0.0005, "loss": 1.3635722398757935, "step": 40511 }, { "epoch": 3.1936933385888846, "grad_norm": 1.0535939931869507, "learning_rate": 0.0005, "loss": 1.2982953786849976, "step": 40512 }, { "epoch": 3.1937721718565233, "grad_norm": 0.5060712099075317, "learning_rate": 0.0005, "loss": 1.2116045951843262, "step": 40513 }, { "epoch": 3.1938510051241624, "grad_norm": 0.4820227026939392, "learning_rate": 0.0005, "loss": 1.3768188953399658, "step": 40514 }, { "epoch": 3.193929838391801, "grad_norm": 0.5744045972824097, "learning_rate": 0.0005, "loss": 1.2351819276809692, "step": 40515 }, { "epoch": 3.1940086716594402, "grad_norm": 0.39961177110671997, "learning_rate": 0.0005, "loss": 1.069710373878479, "step": 40516 }, { "epoch": 3.1940875049270794, "grad_norm": 0.49928876757621765, "learning_rate": 0.0005, "loss": 1.2918918132781982, "step": 40517 }, { "epoch": 3.194166338194718, "grad_norm": 0.7152036428451538, "learning_rate": 0.0005, "loss": 1.2352664470672607, "step": 40518 }, { "epoch": 3.194245171462357, "grad_norm": 0.4708118438720703, "learning_rate": 0.0005, "loss": 1.3568552732467651, "step": 40519 }, { "epoch": 3.194324004729996, "grad_norm": 0.5241134762763977, "learning_rate": 0.0005, "loss": 1.3535887002944946, "step": 40520 }, { "epoch": 3.194402837997635, "grad_norm": 0.4975242018699646, "learning_rate": 0.0005, "loss": 1.4848451614379883, "step": 40521 }, { "epoch": 3.194481671265274, "grad_norm": 0.49846625328063965, "learning_rate": 0.0005, "loss": 1.171800136566162, "step": 40522 }, { "epoch": 3.194560504532913, "grad_norm": 0.5712757110595703, "learning_rate": 0.0005, "loss": 1.4764864444732666, "step": 40523 }, { "epoch": 3.194639337800552, "grad_norm": 0.5281803011894226, "learning_rate": 0.0005, "loss": 1.3767120838165283, "step": 40524 }, { "epoch": 3.1947181710681907, "grad_norm": 0.5576261878013611, "learning_rate": 0.0005, "loss": 1.31342351436615, "step": 40525 }, { "epoch": 3.19479700433583, "grad_norm": 0.48936551809310913, "learning_rate": 0.0005, "loss": 1.2919142246246338, "step": 40526 }, { "epoch": 3.1948758376034685, "grad_norm": 0.4581144154071808, "learning_rate": 0.0005, "loss": 1.3615875244140625, "step": 40527 }, { "epoch": 3.1949546708711076, "grad_norm": 0.6121747493743896, "learning_rate": 0.0005, "loss": 1.347941517829895, "step": 40528 }, { "epoch": 3.1950335041387463, "grad_norm": 0.44803696870803833, "learning_rate": 0.0005, "loss": 1.0928877592086792, "step": 40529 }, { "epoch": 3.1951123374063854, "grad_norm": 0.5512735247612, "learning_rate": 0.0005, "loss": 1.337289571762085, "step": 40530 }, { "epoch": 3.1951911706740246, "grad_norm": 0.46217605471611023, "learning_rate": 0.0005, "loss": 1.2226078510284424, "step": 40531 }, { "epoch": 3.1952700039416633, "grad_norm": 0.5537824034690857, "learning_rate": 0.0005, "loss": 1.3728389739990234, "step": 40532 }, { "epoch": 3.1953488372093024, "grad_norm": 0.6274933815002441, "learning_rate": 0.0005, "loss": 1.1659555435180664, "step": 40533 }, { "epoch": 3.195427670476941, "grad_norm": 0.4135390818119049, "learning_rate": 0.0005, "loss": 1.1991850137710571, "step": 40534 }, { "epoch": 3.1955065037445802, "grad_norm": 0.38884779810905457, "learning_rate": 0.0005, "loss": 1.0923261642456055, "step": 40535 }, { "epoch": 3.1955853370122194, "grad_norm": 1.8138511180877686, "learning_rate": 0.0005, "loss": 1.176051139831543, "step": 40536 }, { "epoch": 3.195664170279858, "grad_norm": 0.5355877876281738, "learning_rate": 0.0005, "loss": 1.2389497756958008, "step": 40537 }, { "epoch": 3.195743003547497, "grad_norm": 0.4550693929195404, "learning_rate": 0.0005, "loss": 1.2621772289276123, "step": 40538 }, { "epoch": 3.195821836815136, "grad_norm": 0.5076927542686462, "learning_rate": 0.0005, "loss": 1.2782232761383057, "step": 40539 }, { "epoch": 3.195900670082775, "grad_norm": 0.3823372423648834, "learning_rate": 0.0005, "loss": 1.0932233333587646, "step": 40540 }, { "epoch": 3.1959795033504137, "grad_norm": 0.47922948002815247, "learning_rate": 0.0005, "loss": 1.2096126079559326, "step": 40541 }, { "epoch": 3.196058336618053, "grad_norm": 0.47594520449638367, "learning_rate": 0.0005, "loss": 1.4475665092468262, "step": 40542 }, { "epoch": 3.196137169885692, "grad_norm": 0.5513367652893066, "learning_rate": 0.0005, "loss": 1.3339776992797852, "step": 40543 }, { "epoch": 3.1962160031533307, "grad_norm": 0.523245096206665, "learning_rate": 0.0005, "loss": 1.2166380882263184, "step": 40544 }, { "epoch": 3.19629483642097, "grad_norm": 0.4177253246307373, "learning_rate": 0.0005, "loss": 1.0506372451782227, "step": 40545 }, { "epoch": 3.1963736696886085, "grad_norm": 0.5466887354850769, "learning_rate": 0.0005, "loss": 1.319204330444336, "step": 40546 }, { "epoch": 3.1964525029562476, "grad_norm": 0.5475627779960632, "learning_rate": 0.0005, "loss": 1.1055195331573486, "step": 40547 }, { "epoch": 3.1965313362238863, "grad_norm": 0.5807479023933411, "learning_rate": 0.0005, "loss": 1.253122091293335, "step": 40548 }, { "epoch": 3.1966101694915254, "grad_norm": 0.4716712534427643, "learning_rate": 0.0005, "loss": 1.2991054058074951, "step": 40549 }, { "epoch": 3.1966890027591646, "grad_norm": 0.5576531887054443, "learning_rate": 0.0005, "loss": 1.2946546077728271, "step": 40550 }, { "epoch": 3.1967678360268033, "grad_norm": 0.5360559225082397, "learning_rate": 0.0005, "loss": 1.4556039571762085, "step": 40551 }, { "epoch": 3.1968466692944424, "grad_norm": 0.5854495167732239, "learning_rate": 0.0005, "loss": 1.268498420715332, "step": 40552 }, { "epoch": 3.196925502562081, "grad_norm": 0.5137730240821838, "learning_rate": 0.0005, "loss": 1.2867168188095093, "step": 40553 }, { "epoch": 3.19700433582972, "grad_norm": 0.525212287902832, "learning_rate": 0.0005, "loss": 1.2991275787353516, "step": 40554 }, { "epoch": 3.197083169097359, "grad_norm": 0.47961190342903137, "learning_rate": 0.0005, "loss": 1.1513423919677734, "step": 40555 }, { "epoch": 3.197162002364998, "grad_norm": 0.4579901099205017, "learning_rate": 0.0005, "loss": 1.1930859088897705, "step": 40556 }, { "epoch": 3.197240835632637, "grad_norm": 0.4063017964363098, "learning_rate": 0.0005, "loss": 1.041961908340454, "step": 40557 }, { "epoch": 3.197319668900276, "grad_norm": 0.5547064542770386, "learning_rate": 0.0005, "loss": 1.2029712200164795, "step": 40558 }, { "epoch": 3.197398502167915, "grad_norm": 0.5472372174263, "learning_rate": 0.0005, "loss": 1.2894525527954102, "step": 40559 }, { "epoch": 3.1974773354355537, "grad_norm": 0.5154145359992981, "learning_rate": 0.0005, "loss": 1.3669540882110596, "step": 40560 }, { "epoch": 3.197556168703193, "grad_norm": 0.4635772705078125, "learning_rate": 0.0005, "loss": 1.119953989982605, "step": 40561 }, { "epoch": 3.1976350019708315, "grad_norm": 0.4292978346347809, "learning_rate": 0.0005, "loss": 1.2173172235488892, "step": 40562 }, { "epoch": 3.1977138352384706, "grad_norm": 0.5189384818077087, "learning_rate": 0.0005, "loss": 1.2994980812072754, "step": 40563 }, { "epoch": 3.1977926685061098, "grad_norm": 0.5534168481826782, "learning_rate": 0.0005, "loss": 1.337844967842102, "step": 40564 }, { "epoch": 3.1978715017737485, "grad_norm": 0.6219118237495422, "learning_rate": 0.0005, "loss": 1.4354616403579712, "step": 40565 }, { "epoch": 3.1979503350413876, "grad_norm": 0.5516032576560974, "learning_rate": 0.0005, "loss": 1.3117492198944092, "step": 40566 }, { "epoch": 3.1980291683090263, "grad_norm": 0.4893525242805481, "learning_rate": 0.0005, "loss": 1.2370240688323975, "step": 40567 }, { "epoch": 3.1981080015766654, "grad_norm": 0.45172959566116333, "learning_rate": 0.0005, "loss": 1.0768086910247803, "step": 40568 }, { "epoch": 3.198186834844304, "grad_norm": 0.5833358764648438, "learning_rate": 0.0005, "loss": 1.250650405883789, "step": 40569 }, { "epoch": 3.1982656681119432, "grad_norm": 0.5209783315658569, "learning_rate": 0.0005, "loss": 1.3650226593017578, "step": 40570 }, { "epoch": 3.1983445013795824, "grad_norm": 0.5526953935623169, "learning_rate": 0.0005, "loss": 1.4105188846588135, "step": 40571 }, { "epoch": 3.198423334647221, "grad_norm": 0.44192707538604736, "learning_rate": 0.0005, "loss": 1.242629051208496, "step": 40572 }, { "epoch": 3.19850216791486, "grad_norm": 0.5145906805992126, "learning_rate": 0.0005, "loss": 1.5010924339294434, "step": 40573 }, { "epoch": 3.198581001182499, "grad_norm": 0.4685594141483307, "learning_rate": 0.0005, "loss": 1.224776029586792, "step": 40574 }, { "epoch": 3.198659834450138, "grad_norm": 0.5540226697921753, "learning_rate": 0.0005, "loss": 1.213813304901123, "step": 40575 }, { "epoch": 3.1987386677177767, "grad_norm": 0.6330887675285339, "learning_rate": 0.0005, "loss": 1.3741402626037598, "step": 40576 }, { "epoch": 3.198817500985416, "grad_norm": 0.6177046298980713, "learning_rate": 0.0005, "loss": 1.2944040298461914, "step": 40577 }, { "epoch": 3.198896334253055, "grad_norm": 0.42755502462387085, "learning_rate": 0.0005, "loss": 1.2502180337905884, "step": 40578 }, { "epoch": 3.1989751675206937, "grad_norm": 3.963637351989746, "learning_rate": 0.0005, "loss": 1.2306780815124512, "step": 40579 }, { "epoch": 3.199054000788333, "grad_norm": 0.48665428161621094, "learning_rate": 0.0005, "loss": 1.1158270835876465, "step": 40580 }, { "epoch": 3.1991328340559715, "grad_norm": 0.5905230641365051, "learning_rate": 0.0005, "loss": 1.187648892402649, "step": 40581 }, { "epoch": 3.1992116673236106, "grad_norm": 0.49960562586784363, "learning_rate": 0.0005, "loss": 1.228922963142395, "step": 40582 }, { "epoch": 3.1992905005912493, "grad_norm": 0.6603894829750061, "learning_rate": 0.0005, "loss": 1.1298773288726807, "step": 40583 }, { "epoch": 3.1993693338588884, "grad_norm": 0.44930437207221985, "learning_rate": 0.0005, "loss": 1.2206696271896362, "step": 40584 }, { "epoch": 3.1994481671265276, "grad_norm": 0.5569816827774048, "learning_rate": 0.0005, "loss": 1.309085726737976, "step": 40585 }, { "epoch": 3.1995270003941663, "grad_norm": 0.5495468378067017, "learning_rate": 0.0005, "loss": 1.3518925905227661, "step": 40586 }, { "epoch": 3.1996058336618054, "grad_norm": 0.48938921093940735, "learning_rate": 0.0005, "loss": 1.0939263105392456, "step": 40587 }, { "epoch": 3.199684666929444, "grad_norm": 0.5355215072631836, "learning_rate": 0.0005, "loss": 1.3347041606903076, "step": 40588 }, { "epoch": 3.199763500197083, "grad_norm": 0.4525265693664551, "learning_rate": 0.0005, "loss": 1.1615500450134277, "step": 40589 }, { "epoch": 3.199842333464722, "grad_norm": 0.4571050703525543, "learning_rate": 0.0005, "loss": 1.2093709707260132, "step": 40590 }, { "epoch": 3.199921166732361, "grad_norm": 0.4741862416267395, "learning_rate": 0.0005, "loss": 1.1222583055496216, "step": 40591 }, { "epoch": 3.2, "grad_norm": 0.5095850229263306, "learning_rate": 0.0005, "loss": 1.2373898029327393, "step": 40592 }, { "epoch": 3.200078833267639, "grad_norm": 0.5306023359298706, "learning_rate": 0.0005, "loss": 1.3671669960021973, "step": 40593 }, { "epoch": 3.200157666535278, "grad_norm": 0.47149452567100525, "learning_rate": 0.0005, "loss": 1.142472505569458, "step": 40594 }, { "epoch": 3.2002364998029167, "grad_norm": 0.5998538136482239, "learning_rate": 0.0005, "loss": 1.451622486114502, "step": 40595 }, { "epoch": 3.200315333070556, "grad_norm": 0.4367547333240509, "learning_rate": 0.0005, "loss": 1.3164618015289307, "step": 40596 }, { "epoch": 3.2003941663381945, "grad_norm": 0.4533883333206177, "learning_rate": 0.0005, "loss": 1.2357873916625977, "step": 40597 }, { "epoch": 3.2004729996058336, "grad_norm": 0.43060797452926636, "learning_rate": 0.0005, "loss": 1.1456769704818726, "step": 40598 }, { "epoch": 3.200551832873473, "grad_norm": 0.461168110370636, "learning_rate": 0.0005, "loss": 1.362038016319275, "step": 40599 }, { "epoch": 3.2006306661411115, "grad_norm": 0.5106250047683716, "learning_rate": 0.0005, "loss": 1.330406665802002, "step": 40600 }, { "epoch": 3.2007094994087506, "grad_norm": 0.49859902262687683, "learning_rate": 0.0005, "loss": 1.1464745998382568, "step": 40601 }, { "epoch": 3.2007883326763893, "grad_norm": 0.5046437978744507, "learning_rate": 0.0005, "loss": 1.3482441902160645, "step": 40602 }, { "epoch": 3.2008671659440284, "grad_norm": 0.4178670644760132, "learning_rate": 0.0005, "loss": 1.3420960903167725, "step": 40603 }, { "epoch": 3.200945999211667, "grad_norm": 0.5163297653198242, "learning_rate": 0.0005, "loss": 1.4696550369262695, "step": 40604 }, { "epoch": 3.2010248324793062, "grad_norm": 0.5383772253990173, "learning_rate": 0.0005, "loss": 1.1297277212142944, "step": 40605 }, { "epoch": 3.2011036657469454, "grad_norm": 0.5867833495140076, "learning_rate": 0.0005, "loss": 1.3116344213485718, "step": 40606 }, { "epoch": 3.201182499014584, "grad_norm": 0.5855990648269653, "learning_rate": 0.0005, "loss": 1.2298457622528076, "step": 40607 }, { "epoch": 3.201261332282223, "grad_norm": 0.5375916361808777, "learning_rate": 0.0005, "loss": 1.1219968795776367, "step": 40608 }, { "epoch": 3.201340165549862, "grad_norm": 0.5014503598213196, "learning_rate": 0.0005, "loss": 1.3260624408721924, "step": 40609 }, { "epoch": 3.201418998817501, "grad_norm": 0.45998385548591614, "learning_rate": 0.0005, "loss": 1.2254314422607422, "step": 40610 }, { "epoch": 3.20149783208514, "grad_norm": 0.4872417449951172, "learning_rate": 0.0005, "loss": 1.0616449117660522, "step": 40611 }, { "epoch": 3.201576665352779, "grad_norm": 0.4437578320503235, "learning_rate": 0.0005, "loss": 1.141876220703125, "step": 40612 }, { "epoch": 3.201655498620418, "grad_norm": 0.4981484115123749, "learning_rate": 0.0005, "loss": 1.099291443824768, "step": 40613 }, { "epoch": 3.2017343318880567, "grad_norm": 0.48233717679977417, "learning_rate": 0.0005, "loss": 1.2517650127410889, "step": 40614 }, { "epoch": 3.201813165155696, "grad_norm": 0.7053465843200684, "learning_rate": 0.0005, "loss": 1.3223397731781006, "step": 40615 }, { "epoch": 3.2018919984233345, "grad_norm": 0.7441575527191162, "learning_rate": 0.0005, "loss": 1.236237645149231, "step": 40616 }, { "epoch": 3.2019708316909736, "grad_norm": 0.45591750741004944, "learning_rate": 0.0005, "loss": 1.2534791231155396, "step": 40617 }, { "epoch": 3.2020496649586123, "grad_norm": 0.6988064646720886, "learning_rate": 0.0005, "loss": 1.2275817394256592, "step": 40618 }, { "epoch": 3.2021284982262515, "grad_norm": 0.4449075162410736, "learning_rate": 0.0005, "loss": 1.0680088996887207, "step": 40619 }, { "epoch": 3.2022073314938906, "grad_norm": 1.6181923151016235, "learning_rate": 0.0005, "loss": 1.3019940853118896, "step": 40620 }, { "epoch": 3.2022861647615293, "grad_norm": 0.5170645713806152, "learning_rate": 0.0005, "loss": 1.2599624395370483, "step": 40621 }, { "epoch": 3.2023649980291684, "grad_norm": 0.5479228496551514, "learning_rate": 0.0005, "loss": 1.4819989204406738, "step": 40622 }, { "epoch": 3.202443831296807, "grad_norm": 0.4799591600894928, "learning_rate": 0.0005, "loss": 1.217883586883545, "step": 40623 }, { "epoch": 3.2025226645644462, "grad_norm": 0.5044324398040771, "learning_rate": 0.0005, "loss": 1.5321393013000488, "step": 40624 }, { "epoch": 3.2026014978320854, "grad_norm": 0.7444487810134888, "learning_rate": 0.0005, "loss": 1.407672643661499, "step": 40625 }, { "epoch": 3.202680331099724, "grad_norm": 0.48538893461227417, "learning_rate": 0.0005, "loss": 1.1116724014282227, "step": 40626 }, { "epoch": 3.202759164367363, "grad_norm": 0.425381064414978, "learning_rate": 0.0005, "loss": 1.0890663862228394, "step": 40627 }, { "epoch": 3.202837997635002, "grad_norm": 0.5181203484535217, "learning_rate": 0.0005, "loss": 1.4134113788604736, "step": 40628 }, { "epoch": 3.202916830902641, "grad_norm": 0.5550560355186462, "learning_rate": 0.0005, "loss": 1.3900071382522583, "step": 40629 }, { "epoch": 3.2029956641702797, "grad_norm": 0.5059005618095398, "learning_rate": 0.0005, "loss": 1.2598152160644531, "step": 40630 }, { "epoch": 3.203074497437919, "grad_norm": 0.5079686045646667, "learning_rate": 0.0005, "loss": 1.4402507543563843, "step": 40631 }, { "epoch": 3.2031533307055575, "grad_norm": 0.601824939250946, "learning_rate": 0.0005, "loss": 1.428990364074707, "step": 40632 }, { "epoch": 3.2032321639731967, "grad_norm": 0.4923834800720215, "learning_rate": 0.0005, "loss": 1.267663598060608, "step": 40633 }, { "epoch": 3.203310997240836, "grad_norm": 0.602982223033905, "learning_rate": 0.0005, "loss": 1.2656292915344238, "step": 40634 }, { "epoch": 3.2033898305084745, "grad_norm": 0.6188969016075134, "learning_rate": 0.0005, "loss": 1.3170125484466553, "step": 40635 }, { "epoch": 3.2034686637761136, "grad_norm": 0.46737512946128845, "learning_rate": 0.0005, "loss": 1.4020445346832275, "step": 40636 }, { "epoch": 3.2035474970437523, "grad_norm": 0.5162913203239441, "learning_rate": 0.0005, "loss": 1.2331159114837646, "step": 40637 }, { "epoch": 3.2036263303113914, "grad_norm": 0.3788471817970276, "learning_rate": 0.0005, "loss": 1.2164084911346436, "step": 40638 }, { "epoch": 3.2037051635790306, "grad_norm": 0.39792686700820923, "learning_rate": 0.0005, "loss": 1.2688028812408447, "step": 40639 }, { "epoch": 3.2037839968466693, "grad_norm": 0.5012977123260498, "learning_rate": 0.0005, "loss": 1.3900458812713623, "step": 40640 }, { "epoch": 3.2038628301143084, "grad_norm": 0.4441887140274048, "learning_rate": 0.0005, "loss": 1.1949588060379028, "step": 40641 }, { "epoch": 3.203941663381947, "grad_norm": 0.47415053844451904, "learning_rate": 0.0005, "loss": 1.207554578781128, "step": 40642 }, { "epoch": 3.204020496649586, "grad_norm": 0.49403947591781616, "learning_rate": 0.0005, "loss": 1.2686865329742432, "step": 40643 }, { "epoch": 3.204099329917225, "grad_norm": 0.4629196226596832, "learning_rate": 0.0005, "loss": 1.345819354057312, "step": 40644 }, { "epoch": 3.204178163184864, "grad_norm": 0.4632996618747711, "learning_rate": 0.0005, "loss": 1.286811113357544, "step": 40645 }, { "epoch": 3.2042569964525027, "grad_norm": 0.363303005695343, "learning_rate": 0.0005, "loss": 1.0866942405700684, "step": 40646 }, { "epoch": 3.204335829720142, "grad_norm": 0.47119274735450745, "learning_rate": 0.0005, "loss": 1.2788429260253906, "step": 40647 }, { "epoch": 3.204414662987781, "grad_norm": 0.6268979907035828, "learning_rate": 0.0005, "loss": 1.3218858242034912, "step": 40648 }, { "epoch": 3.2044934962554197, "grad_norm": 0.5208233594894409, "learning_rate": 0.0005, "loss": 1.2535006999969482, "step": 40649 }, { "epoch": 3.204572329523059, "grad_norm": 0.5889619588851929, "learning_rate": 0.0005, "loss": 1.3130970001220703, "step": 40650 }, { "epoch": 3.2046511627906975, "grad_norm": 0.38955143094062805, "learning_rate": 0.0005, "loss": 1.1042940616607666, "step": 40651 }, { "epoch": 3.2047299960583366, "grad_norm": 0.5277478694915771, "learning_rate": 0.0005, "loss": 1.330381155014038, "step": 40652 }, { "epoch": 3.2048088293259758, "grad_norm": 0.48687687516212463, "learning_rate": 0.0005, "loss": 1.3103229999542236, "step": 40653 }, { "epoch": 3.2048876625936145, "grad_norm": 0.44298291206359863, "learning_rate": 0.0005, "loss": 1.3978497982025146, "step": 40654 }, { "epoch": 3.2049664958612536, "grad_norm": 0.6666715741157532, "learning_rate": 0.0005, "loss": 1.1566379070281982, "step": 40655 }, { "epoch": 3.2050453291288923, "grad_norm": 0.4575776755809784, "learning_rate": 0.0005, "loss": 1.0946179628372192, "step": 40656 }, { "epoch": 3.2051241623965314, "grad_norm": 0.4964032471179962, "learning_rate": 0.0005, "loss": 1.2679115533828735, "step": 40657 }, { "epoch": 3.20520299566417, "grad_norm": 0.47501158714294434, "learning_rate": 0.0005, "loss": 1.2106345891952515, "step": 40658 }, { "epoch": 3.2052818289318092, "grad_norm": 0.5522921681404114, "learning_rate": 0.0005, "loss": 1.300243616104126, "step": 40659 }, { "epoch": 3.2053606621994484, "grad_norm": 0.45414310693740845, "learning_rate": 0.0005, "loss": 1.3652970790863037, "step": 40660 }, { "epoch": 3.205439495467087, "grad_norm": 0.65073162317276, "learning_rate": 0.0005, "loss": 1.3284547328948975, "step": 40661 }, { "epoch": 3.205518328734726, "grad_norm": 0.4163046181201935, "learning_rate": 0.0005, "loss": 1.0582070350646973, "step": 40662 }, { "epoch": 3.205597162002365, "grad_norm": 0.4680594503879547, "learning_rate": 0.0005, "loss": 1.2836580276489258, "step": 40663 }, { "epoch": 3.205675995270004, "grad_norm": 0.5256935358047485, "learning_rate": 0.0005, "loss": 1.2885336875915527, "step": 40664 }, { "epoch": 3.2057548285376427, "grad_norm": 0.5038384795188904, "learning_rate": 0.0005, "loss": 1.1938854455947876, "step": 40665 }, { "epoch": 3.205833661805282, "grad_norm": 0.5133437514305115, "learning_rate": 0.0005, "loss": 1.307403326034546, "step": 40666 }, { "epoch": 3.205912495072921, "grad_norm": 0.49144989252090454, "learning_rate": 0.0005, "loss": 1.1989200115203857, "step": 40667 }, { "epoch": 3.2059913283405597, "grad_norm": 0.5585909485816956, "learning_rate": 0.0005, "loss": 1.3545671701431274, "step": 40668 }, { "epoch": 3.206070161608199, "grad_norm": 0.5926771759986877, "learning_rate": 0.0005, "loss": 1.344557762145996, "step": 40669 }, { "epoch": 3.2061489948758375, "grad_norm": 0.544562816619873, "learning_rate": 0.0005, "loss": 1.1809743642807007, "step": 40670 }, { "epoch": 3.2062278281434766, "grad_norm": 0.437171995639801, "learning_rate": 0.0005, "loss": 1.1761682033538818, "step": 40671 }, { "epoch": 3.2063066614111153, "grad_norm": 0.4289693236351013, "learning_rate": 0.0005, "loss": 1.3429765701293945, "step": 40672 }, { "epoch": 3.2063854946787544, "grad_norm": 0.5416253805160522, "learning_rate": 0.0005, "loss": 1.3395640850067139, "step": 40673 }, { "epoch": 3.2064643279463936, "grad_norm": 0.52318274974823, "learning_rate": 0.0005, "loss": 1.2495286464691162, "step": 40674 }, { "epoch": 3.2065431612140323, "grad_norm": 0.8596315979957581, "learning_rate": 0.0005, "loss": 1.335492730140686, "step": 40675 }, { "epoch": 3.2066219944816714, "grad_norm": 0.5072177052497864, "learning_rate": 0.0005, "loss": 1.1539325714111328, "step": 40676 }, { "epoch": 3.20670082774931, "grad_norm": 0.3662404716014862, "learning_rate": 0.0005, "loss": 1.2188608646392822, "step": 40677 }, { "epoch": 3.2067796610169492, "grad_norm": 0.3931647837162018, "learning_rate": 0.0005, "loss": 1.1153528690338135, "step": 40678 }, { "epoch": 3.206858494284588, "grad_norm": 0.476098895072937, "learning_rate": 0.0005, "loss": 1.2312874794006348, "step": 40679 }, { "epoch": 3.206937327552227, "grad_norm": 0.4782717525959015, "learning_rate": 0.0005, "loss": 1.085288166999817, "step": 40680 }, { "epoch": 3.207016160819866, "grad_norm": 0.5046138167381287, "learning_rate": 0.0005, "loss": 1.2581281661987305, "step": 40681 }, { "epoch": 3.207094994087505, "grad_norm": 0.5941318273544312, "learning_rate": 0.0005, "loss": 1.2739989757537842, "step": 40682 }, { "epoch": 3.207173827355144, "grad_norm": 0.5268040299415588, "learning_rate": 0.0005, "loss": 1.156285285949707, "step": 40683 }, { "epoch": 3.2072526606227827, "grad_norm": 0.5022085905075073, "learning_rate": 0.0005, "loss": 1.286865472793579, "step": 40684 }, { "epoch": 3.207331493890422, "grad_norm": 0.5019497871398926, "learning_rate": 0.0005, "loss": 1.301439642906189, "step": 40685 }, { "epoch": 3.2074103271580605, "grad_norm": 0.5181148052215576, "learning_rate": 0.0005, "loss": 1.3805196285247803, "step": 40686 }, { "epoch": 3.2074891604256996, "grad_norm": 0.4732374846935272, "learning_rate": 0.0005, "loss": 1.2832061052322388, "step": 40687 }, { "epoch": 3.207567993693339, "grad_norm": 0.3778994679450989, "learning_rate": 0.0005, "loss": 1.175635814666748, "step": 40688 }, { "epoch": 3.2076468269609775, "grad_norm": 0.5905019640922546, "learning_rate": 0.0005, "loss": 1.235018014907837, "step": 40689 }, { "epoch": 3.2077256602286166, "grad_norm": 0.4549114406108856, "learning_rate": 0.0005, "loss": 1.302917718887329, "step": 40690 }, { "epoch": 3.2078044934962553, "grad_norm": 0.64287930727005, "learning_rate": 0.0005, "loss": 1.2235503196716309, "step": 40691 }, { "epoch": 3.2078833267638944, "grad_norm": 0.43311282992362976, "learning_rate": 0.0005, "loss": 1.1407749652862549, "step": 40692 }, { "epoch": 3.207962160031533, "grad_norm": 0.49904170632362366, "learning_rate": 0.0005, "loss": 1.1615657806396484, "step": 40693 }, { "epoch": 3.2080409932991722, "grad_norm": 0.522433876991272, "learning_rate": 0.0005, "loss": 1.2952678203582764, "step": 40694 }, { "epoch": 3.2081198265668114, "grad_norm": 0.4473838210105896, "learning_rate": 0.0005, "loss": 1.1096267700195312, "step": 40695 }, { "epoch": 3.20819865983445, "grad_norm": 0.6101608872413635, "learning_rate": 0.0005, "loss": 1.315018653869629, "step": 40696 }, { "epoch": 3.208277493102089, "grad_norm": 0.5289496779441833, "learning_rate": 0.0005, "loss": 1.1840705871582031, "step": 40697 }, { "epoch": 3.208356326369728, "grad_norm": 0.5530827641487122, "learning_rate": 0.0005, "loss": 1.2483899593353271, "step": 40698 }, { "epoch": 3.208435159637367, "grad_norm": 0.42022624611854553, "learning_rate": 0.0005, "loss": 1.0156251192092896, "step": 40699 }, { "epoch": 3.2085139929050057, "grad_norm": 0.5433399677276611, "learning_rate": 0.0005, "loss": 1.2906218767166138, "step": 40700 }, { "epoch": 3.208592826172645, "grad_norm": 0.5868489742279053, "learning_rate": 0.0005, "loss": 1.240829348564148, "step": 40701 }, { "epoch": 3.208671659440284, "grad_norm": 0.5210711359977722, "learning_rate": 0.0005, "loss": 1.118011713027954, "step": 40702 }, { "epoch": 3.2087504927079227, "grad_norm": 0.539951503276825, "learning_rate": 0.0005, "loss": 1.3576469421386719, "step": 40703 }, { "epoch": 3.208829325975562, "grad_norm": 0.4844878017902374, "learning_rate": 0.0005, "loss": 1.3250929117202759, "step": 40704 }, { "epoch": 3.2089081592432005, "grad_norm": 0.5644917488098145, "learning_rate": 0.0005, "loss": 1.2108995914459229, "step": 40705 }, { "epoch": 3.2089869925108396, "grad_norm": 0.5478568077087402, "learning_rate": 0.0005, "loss": 1.3302682638168335, "step": 40706 }, { "epoch": 3.2090658257784783, "grad_norm": 0.5321033000946045, "learning_rate": 0.0005, "loss": 1.3591703176498413, "step": 40707 }, { "epoch": 3.2091446590461175, "grad_norm": 0.6084369421005249, "learning_rate": 0.0005, "loss": 1.066530704498291, "step": 40708 }, { "epoch": 3.2092234923137566, "grad_norm": 0.6828036308288574, "learning_rate": 0.0005, "loss": 1.6477527618408203, "step": 40709 }, { "epoch": 3.2093023255813953, "grad_norm": 0.5083723664283752, "learning_rate": 0.0005, "loss": 1.3983641862869263, "step": 40710 }, { "epoch": 3.2093811588490344, "grad_norm": 0.5840211510658264, "learning_rate": 0.0005, "loss": 1.4153701066970825, "step": 40711 }, { "epoch": 3.209459992116673, "grad_norm": 0.48958495259284973, "learning_rate": 0.0005, "loss": 1.273774266242981, "step": 40712 }, { "epoch": 3.2095388253843122, "grad_norm": 0.5383874773979187, "learning_rate": 0.0005, "loss": 1.3415522575378418, "step": 40713 }, { "epoch": 3.209617658651951, "grad_norm": 0.5158714652061462, "learning_rate": 0.0005, "loss": 1.228916883468628, "step": 40714 }, { "epoch": 3.20969649191959, "grad_norm": 0.4755426347255707, "learning_rate": 0.0005, "loss": 1.228013277053833, "step": 40715 }, { "epoch": 3.209775325187229, "grad_norm": 0.5222628116607666, "learning_rate": 0.0005, "loss": 1.13726007938385, "step": 40716 }, { "epoch": 3.209854158454868, "grad_norm": 0.48231807351112366, "learning_rate": 0.0005, "loss": 1.3501338958740234, "step": 40717 }, { "epoch": 3.209932991722507, "grad_norm": 0.4454270005226135, "learning_rate": 0.0005, "loss": 1.1921327114105225, "step": 40718 }, { "epoch": 3.2100118249901457, "grad_norm": 0.5724688172340393, "learning_rate": 0.0005, "loss": 1.3340041637420654, "step": 40719 }, { "epoch": 3.210090658257785, "grad_norm": 0.508535623550415, "learning_rate": 0.0005, "loss": 1.2351269721984863, "step": 40720 }, { "epoch": 3.2101694915254235, "grad_norm": 0.4899470806121826, "learning_rate": 0.0005, "loss": 1.2458381652832031, "step": 40721 }, { "epoch": 3.2102483247930627, "grad_norm": 0.525529682636261, "learning_rate": 0.0005, "loss": 1.268144130706787, "step": 40722 }, { "epoch": 3.210327158060702, "grad_norm": 0.5095897912979126, "learning_rate": 0.0005, "loss": 1.2954797744750977, "step": 40723 }, { "epoch": 3.2104059913283405, "grad_norm": 0.5675371885299683, "learning_rate": 0.0005, "loss": 1.2769685983657837, "step": 40724 }, { "epoch": 3.2104848245959796, "grad_norm": 0.47638243436813354, "learning_rate": 0.0005, "loss": 1.3646814823150635, "step": 40725 }, { "epoch": 3.2105636578636183, "grad_norm": 0.4095519483089447, "learning_rate": 0.0005, "loss": 1.1736350059509277, "step": 40726 }, { "epoch": 3.2106424911312574, "grad_norm": 0.47222864627838135, "learning_rate": 0.0005, "loss": 1.3170510530471802, "step": 40727 }, { "epoch": 3.2107213243988966, "grad_norm": 0.5089887380599976, "learning_rate": 0.0005, "loss": 1.3307472467422485, "step": 40728 }, { "epoch": 3.2108001576665353, "grad_norm": 0.5250945687294006, "learning_rate": 0.0005, "loss": 1.2915347814559937, "step": 40729 }, { "epoch": 3.2108789909341744, "grad_norm": 0.6762149333953857, "learning_rate": 0.0005, "loss": 1.1737416982650757, "step": 40730 }, { "epoch": 3.210957824201813, "grad_norm": 0.4601719081401825, "learning_rate": 0.0005, "loss": 1.3294222354888916, "step": 40731 }, { "epoch": 3.211036657469452, "grad_norm": 0.49206146597862244, "learning_rate": 0.0005, "loss": 1.2663929462432861, "step": 40732 }, { "epoch": 3.211115490737091, "grad_norm": 0.5207393765449524, "learning_rate": 0.0005, "loss": 1.299536943435669, "step": 40733 }, { "epoch": 3.21119432400473, "grad_norm": 0.42327478528022766, "learning_rate": 0.0005, "loss": 1.2239291667938232, "step": 40734 }, { "epoch": 3.2112731572723687, "grad_norm": 0.44223111867904663, "learning_rate": 0.0005, "loss": 1.0940179824829102, "step": 40735 }, { "epoch": 3.211351990540008, "grad_norm": 0.419229120016098, "learning_rate": 0.0005, "loss": 1.1089410781860352, "step": 40736 }, { "epoch": 3.211430823807647, "grad_norm": 0.6909213066101074, "learning_rate": 0.0005, "loss": 1.1879353523254395, "step": 40737 }, { "epoch": 3.2115096570752857, "grad_norm": 0.4770931005477905, "learning_rate": 0.0005, "loss": 1.3058987855911255, "step": 40738 }, { "epoch": 3.211588490342925, "grad_norm": 0.5094619393348694, "learning_rate": 0.0005, "loss": 1.5243030786514282, "step": 40739 }, { "epoch": 3.2116673236105635, "grad_norm": 0.5371541380882263, "learning_rate": 0.0005, "loss": 1.2289152145385742, "step": 40740 }, { "epoch": 3.2117461568782026, "grad_norm": 0.442872554063797, "learning_rate": 0.0005, "loss": 1.136893391609192, "step": 40741 }, { "epoch": 3.2118249901458418, "grad_norm": 0.6012669205665588, "learning_rate": 0.0005, "loss": 1.3542895317077637, "step": 40742 }, { "epoch": 3.2119038234134805, "grad_norm": 0.6350197792053223, "learning_rate": 0.0005, "loss": 1.3191735744476318, "step": 40743 }, { "epoch": 3.2119826566811196, "grad_norm": 0.4468466639518738, "learning_rate": 0.0005, "loss": 1.2798030376434326, "step": 40744 }, { "epoch": 3.2120614899487583, "grad_norm": 0.4969068765640259, "learning_rate": 0.0005, "loss": 1.1832118034362793, "step": 40745 }, { "epoch": 3.2121403232163974, "grad_norm": 0.45095914602279663, "learning_rate": 0.0005, "loss": 1.1981019973754883, "step": 40746 }, { "epoch": 3.212219156484036, "grad_norm": 0.4504956901073456, "learning_rate": 0.0005, "loss": 1.493161678314209, "step": 40747 }, { "epoch": 3.2122979897516752, "grad_norm": 0.5442169904708862, "learning_rate": 0.0005, "loss": 1.492973804473877, "step": 40748 }, { "epoch": 3.212376823019314, "grad_norm": 0.5007144212722778, "learning_rate": 0.0005, "loss": 1.053500771522522, "step": 40749 }, { "epoch": 3.212455656286953, "grad_norm": 0.399995893239975, "learning_rate": 0.0005, "loss": 1.2837200164794922, "step": 40750 }, { "epoch": 3.212534489554592, "grad_norm": 0.5269545316696167, "learning_rate": 0.0005, "loss": 1.4379842281341553, "step": 40751 }, { "epoch": 3.212613322822231, "grad_norm": 0.4814302325248718, "learning_rate": 0.0005, "loss": 1.3924384117126465, "step": 40752 }, { "epoch": 3.21269215608987, "grad_norm": 0.3937493860721588, "learning_rate": 0.0005, "loss": 1.2658805847167969, "step": 40753 }, { "epoch": 3.2127709893575087, "grad_norm": 0.490323930978775, "learning_rate": 0.0005, "loss": 1.2614701986312866, "step": 40754 }, { "epoch": 3.212849822625148, "grad_norm": 0.47466525435447693, "learning_rate": 0.0005, "loss": 1.1709136962890625, "step": 40755 }, { "epoch": 3.212928655892787, "grad_norm": 0.5821192860603333, "learning_rate": 0.0005, "loss": 1.3596701622009277, "step": 40756 }, { "epoch": 3.2130074891604257, "grad_norm": 0.5675939321517944, "learning_rate": 0.0005, "loss": 1.3312278985977173, "step": 40757 }, { "epoch": 3.213086322428065, "grad_norm": 0.40660396218299866, "learning_rate": 0.0005, "loss": 1.136598825454712, "step": 40758 }, { "epoch": 3.2131651556957035, "grad_norm": 0.4480671286582947, "learning_rate": 0.0005, "loss": 1.1572378873825073, "step": 40759 }, { "epoch": 3.2132439889633426, "grad_norm": 0.513364851474762, "learning_rate": 0.0005, "loss": 1.3242592811584473, "step": 40760 }, { "epoch": 3.2133228222309813, "grad_norm": 0.460325688123703, "learning_rate": 0.0005, "loss": 1.3449747562408447, "step": 40761 }, { "epoch": 3.2134016554986204, "grad_norm": 0.41700392961502075, "learning_rate": 0.0005, "loss": 1.2719175815582275, "step": 40762 }, { "epoch": 3.213480488766259, "grad_norm": 0.4099886119365692, "learning_rate": 0.0005, "loss": 1.2098218202590942, "step": 40763 }, { "epoch": 3.2135593220338983, "grad_norm": 0.4761386811733246, "learning_rate": 0.0005, "loss": 1.1688134670257568, "step": 40764 }, { "epoch": 3.2136381553015374, "grad_norm": 0.5943678021430969, "learning_rate": 0.0005, "loss": 1.282261848449707, "step": 40765 }, { "epoch": 3.213716988569176, "grad_norm": 0.6155529618263245, "learning_rate": 0.0005, "loss": 1.446380853652954, "step": 40766 }, { "epoch": 3.2137958218368152, "grad_norm": 0.43149039149284363, "learning_rate": 0.0005, "loss": 1.2462257146835327, "step": 40767 }, { "epoch": 3.213874655104454, "grad_norm": 0.46200141310691833, "learning_rate": 0.0005, "loss": 1.1861340999603271, "step": 40768 }, { "epoch": 3.213953488372093, "grad_norm": 0.5640994906425476, "learning_rate": 0.0005, "loss": 1.1974422931671143, "step": 40769 }, { "epoch": 3.214032321639732, "grad_norm": 0.5192925333976746, "learning_rate": 0.0005, "loss": 1.2813608646392822, "step": 40770 }, { "epoch": 3.214111154907371, "grad_norm": 0.47797343134880066, "learning_rate": 0.0005, "loss": 1.1611628532409668, "step": 40771 }, { "epoch": 3.21418998817501, "grad_norm": 0.5127589106559753, "learning_rate": 0.0005, "loss": 1.2778724431991577, "step": 40772 }, { "epoch": 3.2142688214426487, "grad_norm": 0.633574903011322, "learning_rate": 0.0005, "loss": 1.4160676002502441, "step": 40773 }, { "epoch": 3.214347654710288, "grad_norm": 0.5362206697463989, "learning_rate": 0.0005, "loss": 1.319478988647461, "step": 40774 }, { "epoch": 3.2144264879779265, "grad_norm": 0.5409450531005859, "learning_rate": 0.0005, "loss": 1.241410732269287, "step": 40775 }, { "epoch": 3.2145053212455656, "grad_norm": 0.4263077676296234, "learning_rate": 0.0005, "loss": 1.311805009841919, "step": 40776 }, { "epoch": 3.2145841545132043, "grad_norm": 0.5075344443321228, "learning_rate": 0.0005, "loss": 1.4038519859313965, "step": 40777 }, { "epoch": 3.2146629877808435, "grad_norm": 0.5361973643302917, "learning_rate": 0.0005, "loss": 1.3909502029418945, "step": 40778 }, { "epoch": 3.2147418210484826, "grad_norm": 0.5555140376091003, "learning_rate": 0.0005, "loss": 1.426256775856018, "step": 40779 }, { "epoch": 3.2148206543161213, "grad_norm": 0.4713233709335327, "learning_rate": 0.0005, "loss": 1.1728127002716064, "step": 40780 }, { "epoch": 3.2148994875837604, "grad_norm": 0.4926002323627472, "learning_rate": 0.0005, "loss": 1.142921805381775, "step": 40781 }, { "epoch": 3.214978320851399, "grad_norm": 0.4840768575668335, "learning_rate": 0.0005, "loss": 1.2204704284667969, "step": 40782 }, { "epoch": 3.2150571541190383, "grad_norm": 0.4015595316886902, "learning_rate": 0.0005, "loss": 1.095863699913025, "step": 40783 }, { "epoch": 3.2151359873866774, "grad_norm": 0.5235597491264343, "learning_rate": 0.0005, "loss": 1.3524255752563477, "step": 40784 }, { "epoch": 3.215214820654316, "grad_norm": 0.5206422805786133, "learning_rate": 0.0005, "loss": 1.500978708267212, "step": 40785 }, { "epoch": 3.215293653921955, "grad_norm": 0.47721007466316223, "learning_rate": 0.0005, "loss": 1.4404826164245605, "step": 40786 }, { "epoch": 3.215372487189594, "grad_norm": 0.4778573513031006, "learning_rate": 0.0005, "loss": 1.190277099609375, "step": 40787 }, { "epoch": 3.215451320457233, "grad_norm": 0.49974414706230164, "learning_rate": 0.0005, "loss": 1.3126710653305054, "step": 40788 }, { "epoch": 3.2155301537248717, "grad_norm": 0.48546454310417175, "learning_rate": 0.0005, "loss": 1.302103877067566, "step": 40789 }, { "epoch": 3.215608986992511, "grad_norm": 0.5886396765708923, "learning_rate": 0.0005, "loss": 1.406283974647522, "step": 40790 }, { "epoch": 3.21568782026015, "grad_norm": 0.5185992121696472, "learning_rate": 0.0005, "loss": 1.4877386093139648, "step": 40791 }, { "epoch": 3.2157666535277887, "grad_norm": 0.5050023794174194, "learning_rate": 0.0005, "loss": 1.3389369249343872, "step": 40792 }, { "epoch": 3.215845486795428, "grad_norm": 0.48454931378364563, "learning_rate": 0.0005, "loss": 1.2590374946594238, "step": 40793 }, { "epoch": 3.2159243200630665, "grad_norm": 0.5087268352508545, "learning_rate": 0.0005, "loss": 1.2923095226287842, "step": 40794 }, { "epoch": 3.2160031533307056, "grad_norm": 0.47439637780189514, "learning_rate": 0.0005, "loss": 1.0993993282318115, "step": 40795 }, { "epoch": 3.2160819865983443, "grad_norm": 0.49753814935684204, "learning_rate": 0.0005, "loss": 1.3381292819976807, "step": 40796 }, { "epoch": 3.2161608198659835, "grad_norm": 0.47305676341056824, "learning_rate": 0.0005, "loss": 1.434260606765747, "step": 40797 }, { "epoch": 3.2162396531336226, "grad_norm": 0.6103711128234863, "learning_rate": 0.0005, "loss": 1.2675085067749023, "step": 40798 }, { "epoch": 3.2163184864012613, "grad_norm": 0.5112127661705017, "learning_rate": 0.0005, "loss": 1.3239362239837646, "step": 40799 }, { "epoch": 3.2163973196689004, "grad_norm": 0.48520854115486145, "learning_rate": 0.0005, "loss": 1.5017746686935425, "step": 40800 }, { "epoch": 3.216476152936539, "grad_norm": 0.6403666734695435, "learning_rate": 0.0005, "loss": 1.3119897842407227, "step": 40801 }, { "epoch": 3.2165549862041782, "grad_norm": 0.48647886514663696, "learning_rate": 0.0005, "loss": 1.242140293121338, "step": 40802 }, { "epoch": 3.216633819471817, "grad_norm": 0.4356079697608948, "learning_rate": 0.0005, "loss": 1.1823365688323975, "step": 40803 }, { "epoch": 3.216712652739456, "grad_norm": 0.44858139753341675, "learning_rate": 0.0005, "loss": 1.2488770484924316, "step": 40804 }, { "epoch": 3.216791486007095, "grad_norm": 0.4619700610637665, "learning_rate": 0.0005, "loss": 1.2287853956222534, "step": 40805 }, { "epoch": 3.216870319274734, "grad_norm": 0.49494948983192444, "learning_rate": 0.0005, "loss": 1.379097819328308, "step": 40806 }, { "epoch": 3.216949152542373, "grad_norm": 0.4275154173374176, "learning_rate": 0.0005, "loss": 1.1428070068359375, "step": 40807 }, { "epoch": 3.2170279858100117, "grad_norm": 0.39644870162010193, "learning_rate": 0.0005, "loss": 1.1695598363876343, "step": 40808 }, { "epoch": 3.217106819077651, "grad_norm": 0.508237898349762, "learning_rate": 0.0005, "loss": 1.4503833055496216, "step": 40809 }, { "epoch": 3.2171856523452895, "grad_norm": 1.0279220342636108, "learning_rate": 0.0005, "loss": 1.311084270477295, "step": 40810 }, { "epoch": 3.2172644856129287, "grad_norm": 0.6279381513595581, "learning_rate": 0.0005, "loss": 1.4001014232635498, "step": 40811 }, { "epoch": 3.217343318880568, "grad_norm": 0.5819089412689209, "learning_rate": 0.0005, "loss": 1.527130126953125, "step": 40812 }, { "epoch": 3.2174221521482065, "grad_norm": 0.5614640116691589, "learning_rate": 0.0005, "loss": 1.182490587234497, "step": 40813 }, { "epoch": 3.2175009854158456, "grad_norm": 0.5523318648338318, "learning_rate": 0.0005, "loss": 1.2921128273010254, "step": 40814 }, { "epoch": 3.2175798186834843, "grad_norm": 0.5724701881408691, "learning_rate": 0.0005, "loss": 1.3645516633987427, "step": 40815 }, { "epoch": 3.2176586519511234, "grad_norm": 0.6110931038856506, "learning_rate": 0.0005, "loss": 1.4270670413970947, "step": 40816 }, { "epoch": 3.217737485218762, "grad_norm": 0.47974467277526855, "learning_rate": 0.0005, "loss": 1.327822208404541, "step": 40817 }, { "epoch": 3.2178163184864013, "grad_norm": 0.7181637287139893, "learning_rate": 0.0005, "loss": 1.3270950317382812, "step": 40818 }, { "epoch": 3.2178951517540404, "grad_norm": 0.4671211242675781, "learning_rate": 0.0005, "loss": 1.4181196689605713, "step": 40819 }, { "epoch": 3.217973985021679, "grad_norm": 0.41108575463294983, "learning_rate": 0.0005, "loss": 1.183306097984314, "step": 40820 }, { "epoch": 3.218052818289318, "grad_norm": 0.4808993339538574, "learning_rate": 0.0005, "loss": 1.2611690759658813, "step": 40821 }, { "epoch": 3.218131651556957, "grad_norm": 0.44092774391174316, "learning_rate": 0.0005, "loss": 1.2052795886993408, "step": 40822 }, { "epoch": 3.218210484824596, "grad_norm": 0.5178834199905396, "learning_rate": 0.0005, "loss": 1.266247034072876, "step": 40823 }, { "epoch": 3.2182893180922347, "grad_norm": 1.09091317653656, "learning_rate": 0.0005, "loss": 1.3174526691436768, "step": 40824 }, { "epoch": 3.218368151359874, "grad_norm": 0.5079439282417297, "learning_rate": 0.0005, "loss": 1.223810076713562, "step": 40825 }, { "epoch": 3.218446984627513, "grad_norm": 0.47735491394996643, "learning_rate": 0.0005, "loss": 1.2942039966583252, "step": 40826 }, { "epoch": 3.2185258178951517, "grad_norm": 0.6080096960067749, "learning_rate": 0.0005, "loss": 1.5355157852172852, "step": 40827 }, { "epoch": 3.218604651162791, "grad_norm": 0.5057768821716309, "learning_rate": 0.0005, "loss": 1.236592411994934, "step": 40828 }, { "epoch": 3.2186834844304295, "grad_norm": 0.4104818105697632, "learning_rate": 0.0005, "loss": 1.1261684894561768, "step": 40829 }, { "epoch": 3.2187623176980686, "grad_norm": 0.5214032530784607, "learning_rate": 0.0005, "loss": 1.2510099411010742, "step": 40830 }, { "epoch": 3.2188411509657073, "grad_norm": 0.49205538630485535, "learning_rate": 0.0005, "loss": 1.319146752357483, "step": 40831 }, { "epoch": 3.2189199842333465, "grad_norm": 0.5145238637924194, "learning_rate": 0.0005, "loss": 1.282620906829834, "step": 40832 }, { "epoch": 3.2189988175009856, "grad_norm": 0.5301628708839417, "learning_rate": 0.0005, "loss": 1.1799626350402832, "step": 40833 }, { "epoch": 3.2190776507686243, "grad_norm": 0.5670973658561707, "learning_rate": 0.0005, "loss": 1.3473258018493652, "step": 40834 }, { "epoch": 3.2191564840362634, "grad_norm": 0.4880748391151428, "learning_rate": 0.0005, "loss": 1.2269554138183594, "step": 40835 }, { "epoch": 3.219235317303902, "grad_norm": 0.5281334519386292, "learning_rate": 0.0005, "loss": 1.3310818672180176, "step": 40836 }, { "epoch": 3.2193141505715412, "grad_norm": 0.4939180910587311, "learning_rate": 0.0005, "loss": 1.203050136566162, "step": 40837 }, { "epoch": 3.21939298383918, "grad_norm": 0.5053716897964478, "learning_rate": 0.0005, "loss": 1.3696386814117432, "step": 40838 }, { "epoch": 3.219471817106819, "grad_norm": 0.47407010197639465, "learning_rate": 0.0005, "loss": 1.1680195331573486, "step": 40839 }, { "epoch": 3.219550650374458, "grad_norm": 0.4453144967556, "learning_rate": 0.0005, "loss": 1.0906651020050049, "step": 40840 }, { "epoch": 3.219629483642097, "grad_norm": 0.5304099321365356, "learning_rate": 0.0005, "loss": 1.4683644771575928, "step": 40841 }, { "epoch": 3.219708316909736, "grad_norm": 0.774619996547699, "learning_rate": 0.0005, "loss": 1.3866307735443115, "step": 40842 }, { "epoch": 3.2197871501773747, "grad_norm": 0.4025207757949829, "learning_rate": 0.0005, "loss": 1.1752551794052124, "step": 40843 }, { "epoch": 3.219865983445014, "grad_norm": 1.0210888385772705, "learning_rate": 0.0005, "loss": 1.2196624279022217, "step": 40844 }, { "epoch": 3.2199448167126525, "grad_norm": 0.5130692720413208, "learning_rate": 0.0005, "loss": 1.2988770008087158, "step": 40845 }, { "epoch": 3.2200236499802917, "grad_norm": 0.5726335048675537, "learning_rate": 0.0005, "loss": 1.3941445350646973, "step": 40846 }, { "epoch": 3.220102483247931, "grad_norm": 0.8140377998352051, "learning_rate": 0.0005, "loss": 1.270442247390747, "step": 40847 }, { "epoch": 3.2201813165155695, "grad_norm": 0.5108324885368347, "learning_rate": 0.0005, "loss": 1.3427001237869263, "step": 40848 }, { "epoch": 3.2202601497832086, "grad_norm": 0.4456924796104431, "learning_rate": 0.0005, "loss": 1.0686967372894287, "step": 40849 }, { "epoch": 3.2203389830508473, "grad_norm": 0.41646191477775574, "learning_rate": 0.0005, "loss": 1.1392483711242676, "step": 40850 }, { "epoch": 3.2204178163184864, "grad_norm": 0.44294214248657227, "learning_rate": 0.0005, "loss": 1.2402304410934448, "step": 40851 }, { "epoch": 3.220496649586125, "grad_norm": 0.4961010813713074, "learning_rate": 0.0005, "loss": 1.140300989151001, "step": 40852 }, { "epoch": 3.2205754828537643, "grad_norm": 0.48680153489112854, "learning_rate": 0.0005, "loss": 1.2857242822647095, "step": 40853 }, { "epoch": 3.2206543161214034, "grad_norm": 0.5011616349220276, "learning_rate": 0.0005, "loss": 1.134428858757019, "step": 40854 }, { "epoch": 3.220733149389042, "grad_norm": 0.44016233086586, "learning_rate": 0.0005, "loss": 1.2798782587051392, "step": 40855 }, { "epoch": 3.2208119826566812, "grad_norm": 0.44757065176963806, "learning_rate": 0.0005, "loss": 1.3938100337982178, "step": 40856 }, { "epoch": 3.22089081592432, "grad_norm": 0.5727051496505737, "learning_rate": 0.0005, "loss": 1.323127031326294, "step": 40857 }, { "epoch": 3.220969649191959, "grad_norm": 0.6535202860832214, "learning_rate": 0.0005, "loss": 1.4100515842437744, "step": 40858 }, { "epoch": 3.221048482459598, "grad_norm": 0.5212929844856262, "learning_rate": 0.0005, "loss": 1.159315824508667, "step": 40859 }, { "epoch": 3.221127315727237, "grad_norm": 0.5203512907028198, "learning_rate": 0.0005, "loss": 1.3726850748062134, "step": 40860 }, { "epoch": 3.221206148994876, "grad_norm": 0.44626617431640625, "learning_rate": 0.0005, "loss": 1.2634010314941406, "step": 40861 }, { "epoch": 3.2212849822625147, "grad_norm": 0.5212575793266296, "learning_rate": 0.0005, "loss": 1.1809945106506348, "step": 40862 }, { "epoch": 3.221363815530154, "grad_norm": 0.5053447484970093, "learning_rate": 0.0005, "loss": 1.2809576988220215, "step": 40863 }, { "epoch": 3.2214426487977925, "grad_norm": 0.5869295001029968, "learning_rate": 0.0005, "loss": 1.5161573886871338, "step": 40864 }, { "epoch": 3.2215214820654317, "grad_norm": 0.5307978391647339, "learning_rate": 0.0005, "loss": 1.2761492729187012, "step": 40865 }, { "epoch": 3.2216003153330703, "grad_norm": 0.42161768674850464, "learning_rate": 0.0005, "loss": 1.0983986854553223, "step": 40866 }, { "epoch": 3.2216791486007095, "grad_norm": 0.506656289100647, "learning_rate": 0.0005, "loss": 1.2404179573059082, "step": 40867 }, { "epoch": 3.2217579818683486, "grad_norm": 0.8938583731651306, "learning_rate": 0.0005, "loss": 1.4589221477508545, "step": 40868 }, { "epoch": 3.2218368151359873, "grad_norm": 0.4596725404262543, "learning_rate": 0.0005, "loss": 1.1493334770202637, "step": 40869 }, { "epoch": 3.2219156484036264, "grad_norm": 0.4695694148540497, "learning_rate": 0.0005, "loss": 1.2573529481887817, "step": 40870 }, { "epoch": 3.221994481671265, "grad_norm": 0.5759671926498413, "learning_rate": 0.0005, "loss": 1.5266103744506836, "step": 40871 }, { "epoch": 3.2220733149389043, "grad_norm": 0.5749031901359558, "learning_rate": 0.0005, "loss": 1.4161128997802734, "step": 40872 }, { "epoch": 3.2221521482065434, "grad_norm": 0.502007007598877, "learning_rate": 0.0005, "loss": 1.2036854028701782, "step": 40873 }, { "epoch": 3.222230981474182, "grad_norm": 0.424230694770813, "learning_rate": 0.0005, "loss": 1.1747386455535889, "step": 40874 }, { "epoch": 3.222309814741821, "grad_norm": 0.5118686556816101, "learning_rate": 0.0005, "loss": 1.1785149574279785, "step": 40875 }, { "epoch": 3.22238864800946, "grad_norm": 0.4113151431083679, "learning_rate": 0.0005, "loss": 1.1244934797286987, "step": 40876 }, { "epoch": 3.222467481277099, "grad_norm": 0.6201362609863281, "learning_rate": 0.0005, "loss": 1.4808964729309082, "step": 40877 }, { "epoch": 3.2225463145447377, "grad_norm": 0.5696319937705994, "learning_rate": 0.0005, "loss": 1.4277338981628418, "step": 40878 }, { "epoch": 3.222625147812377, "grad_norm": 0.4386507570743561, "learning_rate": 0.0005, "loss": 1.1788426637649536, "step": 40879 }, { "epoch": 3.2227039810800155, "grad_norm": 0.458072692155838, "learning_rate": 0.0005, "loss": 1.112166404724121, "step": 40880 }, { "epoch": 3.2227828143476547, "grad_norm": 0.5360956788063049, "learning_rate": 0.0005, "loss": 1.2720966339111328, "step": 40881 }, { "epoch": 3.222861647615294, "grad_norm": 0.5355584025382996, "learning_rate": 0.0005, "loss": 1.404817819595337, "step": 40882 }, { "epoch": 3.2229404808829325, "grad_norm": 0.5448899865150452, "learning_rate": 0.0005, "loss": 1.2467671632766724, "step": 40883 }, { "epoch": 3.2230193141505716, "grad_norm": 0.47924715280532837, "learning_rate": 0.0005, "loss": 1.3802204132080078, "step": 40884 }, { "epoch": 3.2230981474182103, "grad_norm": 0.5497525930404663, "learning_rate": 0.0005, "loss": 1.256594181060791, "step": 40885 }, { "epoch": 3.2231769806858495, "grad_norm": 0.4606284499168396, "learning_rate": 0.0005, "loss": 1.3725886344909668, "step": 40886 }, { "epoch": 3.2232558139534886, "grad_norm": 0.6642739176750183, "learning_rate": 0.0005, "loss": 1.4871604442596436, "step": 40887 }, { "epoch": 3.2233346472211273, "grad_norm": 0.48385411500930786, "learning_rate": 0.0005, "loss": 1.2072376012802124, "step": 40888 }, { "epoch": 3.2234134804887664, "grad_norm": 0.49985072016716003, "learning_rate": 0.0005, "loss": 1.400475025177002, "step": 40889 }, { "epoch": 3.223492313756405, "grad_norm": 0.4560217559337616, "learning_rate": 0.0005, "loss": 1.1650971174240112, "step": 40890 }, { "epoch": 3.2235711470240442, "grad_norm": 0.4687442481517792, "learning_rate": 0.0005, "loss": 1.1708648204803467, "step": 40891 }, { "epoch": 3.223649980291683, "grad_norm": 0.46272701025009155, "learning_rate": 0.0005, "loss": 1.2787201404571533, "step": 40892 }, { "epoch": 3.223728813559322, "grad_norm": 0.5013549327850342, "learning_rate": 0.0005, "loss": 1.3059773445129395, "step": 40893 }, { "epoch": 3.2238076468269607, "grad_norm": 0.5316804051399231, "learning_rate": 0.0005, "loss": 1.3892414569854736, "step": 40894 }, { "epoch": 3.2238864800946, "grad_norm": 0.43932250142097473, "learning_rate": 0.0005, "loss": 1.2085528373718262, "step": 40895 }, { "epoch": 3.223965313362239, "grad_norm": 0.4408723711967468, "learning_rate": 0.0005, "loss": 1.389549970626831, "step": 40896 }, { "epoch": 3.2240441466298777, "grad_norm": 0.6713367104530334, "learning_rate": 0.0005, "loss": 1.4068033695220947, "step": 40897 }, { "epoch": 3.224122979897517, "grad_norm": 0.44390934705734253, "learning_rate": 0.0005, "loss": 1.2366318702697754, "step": 40898 }, { "epoch": 3.2242018131651555, "grad_norm": 0.41283732652664185, "learning_rate": 0.0005, "loss": 1.0703805685043335, "step": 40899 }, { "epoch": 3.2242806464327947, "grad_norm": 0.6102150678634644, "learning_rate": 0.0005, "loss": 1.3270673751831055, "step": 40900 }, { "epoch": 3.224359479700434, "grad_norm": 0.4869033396244049, "learning_rate": 0.0005, "loss": 1.214902639389038, "step": 40901 }, { "epoch": 3.2244383129680725, "grad_norm": 0.437767893075943, "learning_rate": 0.0005, "loss": 1.0845909118652344, "step": 40902 }, { "epoch": 3.2245171462357116, "grad_norm": 0.5148182511329651, "learning_rate": 0.0005, "loss": 1.3934729099273682, "step": 40903 }, { "epoch": 3.2245959795033503, "grad_norm": 0.5143916010856628, "learning_rate": 0.0005, "loss": 1.3552138805389404, "step": 40904 }, { "epoch": 3.2246748127709894, "grad_norm": 0.4559633731842041, "learning_rate": 0.0005, "loss": 1.2640442848205566, "step": 40905 }, { "epoch": 3.224753646038628, "grad_norm": 0.46760526299476624, "learning_rate": 0.0005, "loss": 1.363406777381897, "step": 40906 }, { "epoch": 3.2248324793062673, "grad_norm": 0.5158870220184326, "learning_rate": 0.0005, "loss": 1.2827590703964233, "step": 40907 }, { "epoch": 3.224911312573906, "grad_norm": 0.5642061829566956, "learning_rate": 0.0005, "loss": 1.303371548652649, "step": 40908 }, { "epoch": 3.224990145841545, "grad_norm": 0.4352210462093353, "learning_rate": 0.0005, "loss": 1.1629507541656494, "step": 40909 }, { "epoch": 3.225068979109184, "grad_norm": 0.4970289468765259, "learning_rate": 0.0005, "loss": 1.420644760131836, "step": 40910 }, { "epoch": 3.225147812376823, "grad_norm": 0.506496787071228, "learning_rate": 0.0005, "loss": 1.3754349946975708, "step": 40911 }, { "epoch": 3.225226645644462, "grad_norm": 0.46096161007881165, "learning_rate": 0.0005, "loss": 1.2322657108306885, "step": 40912 }, { "epoch": 3.2253054789121007, "grad_norm": 0.51638263463974, "learning_rate": 0.0005, "loss": 1.259350299835205, "step": 40913 }, { "epoch": 3.22538431217974, "grad_norm": 0.4314644932746887, "learning_rate": 0.0005, "loss": 1.2086784839630127, "step": 40914 }, { "epoch": 3.225463145447379, "grad_norm": 0.39806246757507324, "learning_rate": 0.0005, "loss": 1.1269586086273193, "step": 40915 }, { "epoch": 3.2255419787150177, "grad_norm": 1.4405041933059692, "learning_rate": 0.0005, "loss": 1.4300034046173096, "step": 40916 }, { "epoch": 3.225620811982657, "grad_norm": 0.49960172176361084, "learning_rate": 0.0005, "loss": 1.200451374053955, "step": 40917 }, { "epoch": 3.2256996452502955, "grad_norm": 0.6759348511695862, "learning_rate": 0.0005, "loss": 1.4823365211486816, "step": 40918 }, { "epoch": 3.2257784785179346, "grad_norm": 0.5529398918151855, "learning_rate": 0.0005, "loss": 1.3494527339935303, "step": 40919 }, { "epoch": 3.2258573117855733, "grad_norm": 0.5175082087516785, "learning_rate": 0.0005, "loss": 1.1879198551177979, "step": 40920 }, { "epoch": 3.2259361450532125, "grad_norm": 0.5753660798072815, "learning_rate": 0.0005, "loss": 1.3263647556304932, "step": 40921 }, { "epoch": 3.2260149783208516, "grad_norm": 0.5648173093795776, "learning_rate": 0.0005, "loss": 1.4710702896118164, "step": 40922 }, { "epoch": 3.2260938115884903, "grad_norm": 0.4484966993331909, "learning_rate": 0.0005, "loss": 1.0764124393463135, "step": 40923 }, { "epoch": 3.2261726448561294, "grad_norm": 0.4290018677711487, "learning_rate": 0.0005, "loss": 1.1650866270065308, "step": 40924 }, { "epoch": 3.226251478123768, "grad_norm": 0.4441242516040802, "learning_rate": 0.0005, "loss": 1.1265933513641357, "step": 40925 }, { "epoch": 3.2263303113914072, "grad_norm": 0.5031655430793762, "learning_rate": 0.0005, "loss": 1.2006518840789795, "step": 40926 }, { "epoch": 3.226409144659046, "grad_norm": 0.6047810912132263, "learning_rate": 0.0005, "loss": 1.271759033203125, "step": 40927 }, { "epoch": 3.226487977926685, "grad_norm": 0.38762909173965454, "learning_rate": 0.0005, "loss": 1.1172351837158203, "step": 40928 }, { "epoch": 3.226566811194324, "grad_norm": 0.47661271691322327, "learning_rate": 0.0005, "loss": 1.2962721586227417, "step": 40929 }, { "epoch": 3.226645644461963, "grad_norm": 0.9827678799629211, "learning_rate": 0.0005, "loss": 1.3424656391143799, "step": 40930 }, { "epoch": 3.226724477729602, "grad_norm": 0.4276493191719055, "learning_rate": 0.0005, "loss": 1.25411057472229, "step": 40931 }, { "epoch": 3.2268033109972407, "grad_norm": 0.5553174018859863, "learning_rate": 0.0005, "loss": 1.3491921424865723, "step": 40932 }, { "epoch": 3.22688214426488, "grad_norm": 0.5685228705406189, "learning_rate": 0.0005, "loss": 1.4660863876342773, "step": 40933 }, { "epoch": 3.2269609775325185, "grad_norm": 0.600203275680542, "learning_rate": 0.0005, "loss": 1.5122387409210205, "step": 40934 }, { "epoch": 3.2270398108001577, "grad_norm": 0.46631020307540894, "learning_rate": 0.0005, "loss": 1.180431604385376, "step": 40935 }, { "epoch": 3.227118644067797, "grad_norm": 0.4873506724834442, "learning_rate": 0.0005, "loss": 1.203260898590088, "step": 40936 }, { "epoch": 3.2271974773354355, "grad_norm": 0.5032846927642822, "learning_rate": 0.0005, "loss": 1.4315969944000244, "step": 40937 }, { "epoch": 3.2272763106030746, "grad_norm": 0.46396440267562866, "learning_rate": 0.0005, "loss": 1.1291924715042114, "step": 40938 }, { "epoch": 3.2273551438707133, "grad_norm": 0.5734161734580994, "learning_rate": 0.0005, "loss": 1.4536826610565186, "step": 40939 }, { "epoch": 3.2274339771383524, "grad_norm": 0.5276404023170471, "learning_rate": 0.0005, "loss": 1.3686034679412842, "step": 40940 }, { "epoch": 3.227512810405991, "grad_norm": 0.470945805311203, "learning_rate": 0.0005, "loss": 1.2883427143096924, "step": 40941 }, { "epoch": 3.2275916436736303, "grad_norm": 0.48436176776885986, "learning_rate": 0.0005, "loss": 1.3383381366729736, "step": 40942 }, { "epoch": 3.2276704769412694, "grad_norm": 0.41221916675567627, "learning_rate": 0.0005, "loss": 1.116983413696289, "step": 40943 }, { "epoch": 3.227749310208908, "grad_norm": 0.48835068941116333, "learning_rate": 0.0005, "loss": 1.2913051843643188, "step": 40944 }, { "epoch": 3.2278281434765472, "grad_norm": 0.6393234729766846, "learning_rate": 0.0005, "loss": 1.4229815006256104, "step": 40945 }, { "epoch": 3.227906976744186, "grad_norm": 0.5076099038124084, "learning_rate": 0.0005, "loss": 1.3775771856307983, "step": 40946 }, { "epoch": 3.227985810011825, "grad_norm": 0.5141657590866089, "learning_rate": 0.0005, "loss": 1.2507450580596924, "step": 40947 }, { "epoch": 3.2280646432794637, "grad_norm": 0.45770642161369324, "learning_rate": 0.0005, "loss": 1.3249452114105225, "step": 40948 }, { "epoch": 3.228143476547103, "grad_norm": 0.4156338572502136, "learning_rate": 0.0005, "loss": 1.141560435295105, "step": 40949 }, { "epoch": 3.228222309814742, "grad_norm": 0.5222991704940796, "learning_rate": 0.0005, "loss": 1.3365553617477417, "step": 40950 }, { "epoch": 3.2283011430823807, "grad_norm": 0.4454253315925598, "learning_rate": 0.0005, "loss": 1.308384895324707, "step": 40951 }, { "epoch": 3.22837997635002, "grad_norm": 0.46638357639312744, "learning_rate": 0.0005, "loss": 1.238686442375183, "step": 40952 }, { "epoch": 3.2284588096176585, "grad_norm": 0.4668530225753784, "learning_rate": 0.0005, "loss": 1.3371179103851318, "step": 40953 }, { "epoch": 3.2285376428852977, "grad_norm": 0.42634785175323486, "learning_rate": 0.0005, "loss": 1.117242455482483, "step": 40954 }, { "epoch": 3.2286164761529363, "grad_norm": 0.5335004925727844, "learning_rate": 0.0005, "loss": 1.2444013357162476, "step": 40955 }, { "epoch": 3.2286953094205755, "grad_norm": 0.4989413321018219, "learning_rate": 0.0005, "loss": 1.3521687984466553, "step": 40956 }, { "epoch": 3.2287741426882146, "grad_norm": 0.5467063784599304, "learning_rate": 0.0005, "loss": 1.4053804874420166, "step": 40957 }, { "epoch": 3.2288529759558533, "grad_norm": 0.5688581466674805, "learning_rate": 0.0005, "loss": 1.3563591241836548, "step": 40958 }, { "epoch": 3.2289318092234924, "grad_norm": 0.35731279850006104, "learning_rate": 0.0005, "loss": 1.112658977508545, "step": 40959 }, { "epoch": 3.229010642491131, "grad_norm": 0.38663771748542786, "learning_rate": 0.0005, "loss": 1.1723246574401855, "step": 40960 }, { "epoch": 3.2290894757587703, "grad_norm": 0.4736940860748291, "learning_rate": 0.0005, "loss": 1.2310583591461182, "step": 40961 }, { "epoch": 3.229168309026409, "grad_norm": 0.4846792221069336, "learning_rate": 0.0005, "loss": 1.1594974994659424, "step": 40962 }, { "epoch": 3.229247142294048, "grad_norm": 0.46986761689186096, "learning_rate": 0.0005, "loss": 1.4084407091140747, "step": 40963 }, { "epoch": 3.229325975561687, "grad_norm": 0.4804151654243469, "learning_rate": 0.0005, "loss": 1.358962059020996, "step": 40964 }, { "epoch": 3.229404808829326, "grad_norm": 0.4710003435611725, "learning_rate": 0.0005, "loss": 1.2912623882293701, "step": 40965 }, { "epoch": 3.229483642096965, "grad_norm": 0.4794357419013977, "learning_rate": 0.0005, "loss": 1.18022620677948, "step": 40966 }, { "epoch": 3.2295624753646037, "grad_norm": 0.42971011996269226, "learning_rate": 0.0005, "loss": 1.3348698616027832, "step": 40967 }, { "epoch": 3.229641308632243, "grad_norm": 0.42858415842056274, "learning_rate": 0.0005, "loss": 1.2343711853027344, "step": 40968 }, { "epoch": 3.2297201418998815, "grad_norm": 0.46424582600593567, "learning_rate": 0.0005, "loss": 1.240505337715149, "step": 40969 }, { "epoch": 3.2297989751675207, "grad_norm": 0.47259095311164856, "learning_rate": 0.0005, "loss": 1.1272642612457275, "step": 40970 }, { "epoch": 3.22987780843516, "grad_norm": 0.5006408095359802, "learning_rate": 0.0005, "loss": 1.3203550577163696, "step": 40971 }, { "epoch": 3.2299566417027985, "grad_norm": 0.517471969127655, "learning_rate": 0.0005, "loss": 1.1604876518249512, "step": 40972 }, { "epoch": 3.2300354749704376, "grad_norm": 0.44236791133880615, "learning_rate": 0.0005, "loss": 1.3740723133087158, "step": 40973 }, { "epoch": 3.2301143082380763, "grad_norm": 0.4467455744743347, "learning_rate": 0.0005, "loss": 1.1287585496902466, "step": 40974 }, { "epoch": 3.2301931415057155, "grad_norm": 0.5006616115570068, "learning_rate": 0.0005, "loss": 1.4912526607513428, "step": 40975 }, { "epoch": 3.230271974773354, "grad_norm": 0.4544927179813385, "learning_rate": 0.0005, "loss": 1.226128101348877, "step": 40976 }, { "epoch": 3.2303508080409933, "grad_norm": 0.5514025092124939, "learning_rate": 0.0005, "loss": 1.285352110862732, "step": 40977 }, { "epoch": 3.2304296413086324, "grad_norm": 0.4618047773838043, "learning_rate": 0.0005, "loss": 1.1946682929992676, "step": 40978 }, { "epoch": 3.230508474576271, "grad_norm": 0.5031978487968445, "learning_rate": 0.0005, "loss": 1.3888503313064575, "step": 40979 }, { "epoch": 3.2305873078439102, "grad_norm": 0.46352872252464294, "learning_rate": 0.0005, "loss": 1.2951405048370361, "step": 40980 }, { "epoch": 3.230666141111549, "grad_norm": 0.6088129878044128, "learning_rate": 0.0005, "loss": 1.3951044082641602, "step": 40981 }, { "epoch": 3.230744974379188, "grad_norm": 0.44217368960380554, "learning_rate": 0.0005, "loss": 1.233741044998169, "step": 40982 }, { "epoch": 3.2308238076468268, "grad_norm": 0.5198853611946106, "learning_rate": 0.0005, "loss": 1.400343894958496, "step": 40983 }, { "epoch": 3.230902640914466, "grad_norm": 0.48442813754081726, "learning_rate": 0.0005, "loss": 1.2689189910888672, "step": 40984 }, { "epoch": 3.230981474182105, "grad_norm": 0.4697689414024353, "learning_rate": 0.0005, "loss": 1.3069030046463013, "step": 40985 }, { "epoch": 3.2310603074497437, "grad_norm": 0.5965155363082886, "learning_rate": 0.0005, "loss": 1.471224069595337, "step": 40986 }, { "epoch": 3.231139140717383, "grad_norm": 0.4717958867549896, "learning_rate": 0.0005, "loss": 1.389009714126587, "step": 40987 }, { "epoch": 3.2312179739850215, "grad_norm": 0.5188894867897034, "learning_rate": 0.0005, "loss": 1.449406623840332, "step": 40988 }, { "epoch": 3.2312968072526607, "grad_norm": 0.4158378839492798, "learning_rate": 0.0005, "loss": 1.209627389907837, "step": 40989 }, { "epoch": 3.2313756405203, "grad_norm": 0.5409836769104004, "learning_rate": 0.0005, "loss": 1.3272340297698975, "step": 40990 }, { "epoch": 3.2314544737879385, "grad_norm": 0.4994296431541443, "learning_rate": 0.0005, "loss": 1.2423357963562012, "step": 40991 }, { "epoch": 3.2315333070555776, "grad_norm": 0.5117178559303284, "learning_rate": 0.0005, "loss": 1.4454584121704102, "step": 40992 }, { "epoch": 3.2316121403232163, "grad_norm": 0.4421779215335846, "learning_rate": 0.0005, "loss": 1.260688304901123, "step": 40993 }, { "epoch": 3.2316909735908554, "grad_norm": 0.4641592800617218, "learning_rate": 0.0005, "loss": 1.201174259185791, "step": 40994 }, { "epoch": 3.231769806858494, "grad_norm": 0.5898345112800598, "learning_rate": 0.0005, "loss": 1.2377722263336182, "step": 40995 }, { "epoch": 3.2318486401261333, "grad_norm": 0.5199950337409973, "learning_rate": 0.0005, "loss": 1.4669873714447021, "step": 40996 }, { "epoch": 3.231927473393772, "grad_norm": 0.5171641111373901, "learning_rate": 0.0005, "loss": 1.2071211338043213, "step": 40997 }, { "epoch": 3.232006306661411, "grad_norm": 0.45611608028411865, "learning_rate": 0.0005, "loss": 1.2472845315933228, "step": 40998 }, { "epoch": 3.23208513992905, "grad_norm": 0.501116931438446, "learning_rate": 0.0005, "loss": 1.06765615940094, "step": 40999 }, { "epoch": 3.232163973196689, "grad_norm": 0.5628301501274109, "learning_rate": 0.0005, "loss": 1.4937410354614258, "step": 41000 }, { "epoch": 3.232242806464328, "grad_norm": 0.4873047173023224, "learning_rate": 0.0005, "loss": 1.2229278087615967, "step": 41001 }, { "epoch": 3.2323216397319667, "grad_norm": 0.5507165193557739, "learning_rate": 0.0005, "loss": 1.3454129695892334, "step": 41002 }, { "epoch": 3.232400472999606, "grad_norm": 0.5813633799552917, "learning_rate": 0.0005, "loss": 1.3377416133880615, "step": 41003 }, { "epoch": 3.232479306267245, "grad_norm": 0.5428197979927063, "learning_rate": 0.0005, "loss": 1.2575525045394897, "step": 41004 }, { "epoch": 3.2325581395348837, "grad_norm": 0.481627494096756, "learning_rate": 0.0005, "loss": 1.1985957622528076, "step": 41005 }, { "epoch": 3.232636972802523, "grad_norm": 0.4468124508857727, "learning_rate": 0.0005, "loss": 1.1141129732131958, "step": 41006 }, { "epoch": 3.2327158060701615, "grad_norm": 0.5304942727088928, "learning_rate": 0.0005, "loss": 1.3360153436660767, "step": 41007 }, { "epoch": 3.2327946393378006, "grad_norm": 0.46441251039505005, "learning_rate": 0.0005, "loss": 1.3986616134643555, "step": 41008 }, { "epoch": 3.2328734726054393, "grad_norm": 0.4087899327278137, "learning_rate": 0.0005, "loss": 1.223435640335083, "step": 41009 }, { "epoch": 3.2329523058730785, "grad_norm": 0.4120914041996002, "learning_rate": 0.0005, "loss": 1.130018711090088, "step": 41010 }, { "epoch": 3.233031139140717, "grad_norm": 0.5851173996925354, "learning_rate": 0.0005, "loss": 1.2955949306488037, "step": 41011 }, { "epoch": 3.2331099724083563, "grad_norm": 0.39948099851608276, "learning_rate": 0.0005, "loss": 1.1621077060699463, "step": 41012 }, { "epoch": 3.2331888056759954, "grad_norm": 0.45252907276153564, "learning_rate": 0.0005, "loss": 1.2961479425430298, "step": 41013 }, { "epoch": 3.233267638943634, "grad_norm": 0.44938382506370544, "learning_rate": 0.0005, "loss": 1.3508363962173462, "step": 41014 }, { "epoch": 3.2333464722112732, "grad_norm": 0.49153420329093933, "learning_rate": 0.0005, "loss": 1.371678113937378, "step": 41015 }, { "epoch": 3.233425305478912, "grad_norm": 0.49851545691490173, "learning_rate": 0.0005, "loss": 1.2288267612457275, "step": 41016 }, { "epoch": 3.233504138746551, "grad_norm": 0.5094386339187622, "learning_rate": 0.0005, "loss": 1.3870422840118408, "step": 41017 }, { "epoch": 3.23358297201419, "grad_norm": 0.4306117296218872, "learning_rate": 0.0005, "loss": 1.2764266729354858, "step": 41018 }, { "epoch": 3.233661805281829, "grad_norm": 0.5444765090942383, "learning_rate": 0.0005, "loss": 1.2049202919006348, "step": 41019 }, { "epoch": 3.233740638549468, "grad_norm": 0.42296311259269714, "learning_rate": 0.0005, "loss": 1.1868280172348022, "step": 41020 }, { "epoch": 3.2338194718171067, "grad_norm": 0.4442746341228485, "learning_rate": 0.0005, "loss": 1.1188526153564453, "step": 41021 }, { "epoch": 3.233898305084746, "grad_norm": 0.4859806299209595, "learning_rate": 0.0005, "loss": 1.3558628559112549, "step": 41022 }, { "epoch": 3.2339771383523845, "grad_norm": 0.5193735361099243, "learning_rate": 0.0005, "loss": 1.2966415882110596, "step": 41023 }, { "epoch": 3.2340559716200237, "grad_norm": 0.5313476324081421, "learning_rate": 0.0005, "loss": 1.2281477451324463, "step": 41024 }, { "epoch": 3.2341348048876624, "grad_norm": 0.371060848236084, "learning_rate": 0.0005, "loss": 1.0792889595031738, "step": 41025 }, { "epoch": 3.2342136381553015, "grad_norm": 0.47182050347328186, "learning_rate": 0.0005, "loss": 1.3220021724700928, "step": 41026 }, { "epoch": 3.2342924714229406, "grad_norm": 0.4257854223251343, "learning_rate": 0.0005, "loss": 1.25868821144104, "step": 41027 }, { "epoch": 3.2343713046905793, "grad_norm": 0.4599311053752899, "learning_rate": 0.0005, "loss": 1.2746137380599976, "step": 41028 }, { "epoch": 3.2344501379582185, "grad_norm": 0.4508652985095978, "learning_rate": 0.0005, "loss": 1.1061931848526, "step": 41029 }, { "epoch": 3.234528971225857, "grad_norm": 0.5136094093322754, "learning_rate": 0.0005, "loss": 1.2055151462554932, "step": 41030 }, { "epoch": 3.2346078044934963, "grad_norm": 0.4285195469856262, "learning_rate": 0.0005, "loss": 1.3683490753173828, "step": 41031 }, { "epoch": 3.2346866377611354, "grad_norm": 0.5570288300514221, "learning_rate": 0.0005, "loss": 1.2288568019866943, "step": 41032 }, { "epoch": 3.234765471028774, "grad_norm": 0.6198702454566956, "learning_rate": 0.0005, "loss": 1.3695814609527588, "step": 41033 }, { "epoch": 3.2348443042964132, "grad_norm": 0.4559633731842041, "learning_rate": 0.0005, "loss": 1.2840797901153564, "step": 41034 }, { "epoch": 3.234923137564052, "grad_norm": 0.5156530141830444, "learning_rate": 0.0005, "loss": 1.359931230545044, "step": 41035 }, { "epoch": 3.235001970831691, "grad_norm": 0.4223199784755707, "learning_rate": 0.0005, "loss": 1.1128512620925903, "step": 41036 }, { "epoch": 3.2350808040993297, "grad_norm": 0.43794822692871094, "learning_rate": 0.0005, "loss": 1.3614938259124756, "step": 41037 }, { "epoch": 3.235159637366969, "grad_norm": 0.5145030617713928, "learning_rate": 0.0005, "loss": 1.3657212257385254, "step": 41038 }, { "epoch": 3.235238470634608, "grad_norm": 0.49977219104766846, "learning_rate": 0.0005, "loss": 1.311551809310913, "step": 41039 }, { "epoch": 3.2353173039022467, "grad_norm": 0.4240473806858063, "learning_rate": 0.0005, "loss": 1.1568292379379272, "step": 41040 }, { "epoch": 3.235396137169886, "grad_norm": 0.42108041048049927, "learning_rate": 0.0005, "loss": 1.1362897157669067, "step": 41041 }, { "epoch": 3.2354749704375245, "grad_norm": 0.5092315077781677, "learning_rate": 0.0005, "loss": 1.1833020448684692, "step": 41042 }, { "epoch": 3.2355538037051637, "grad_norm": 0.4152328073978424, "learning_rate": 0.0005, "loss": 1.0683385133743286, "step": 41043 }, { "epoch": 3.2356326369728023, "grad_norm": 0.7912717461585999, "learning_rate": 0.0005, "loss": 1.4329057931900024, "step": 41044 }, { "epoch": 3.2357114702404415, "grad_norm": 0.5796415209770203, "learning_rate": 0.0005, "loss": 1.415204644203186, "step": 41045 }, { "epoch": 3.2357903035080806, "grad_norm": 0.5569447875022888, "learning_rate": 0.0005, "loss": 1.302048921585083, "step": 41046 }, { "epoch": 3.2358691367757193, "grad_norm": 0.43522265553474426, "learning_rate": 0.0005, "loss": 1.1350855827331543, "step": 41047 }, { "epoch": 3.2359479700433584, "grad_norm": 0.4868896007537842, "learning_rate": 0.0005, "loss": 1.2848012447357178, "step": 41048 }, { "epoch": 3.236026803310997, "grad_norm": 0.5229173898696899, "learning_rate": 0.0005, "loss": 1.328205943107605, "step": 41049 }, { "epoch": 3.2361056365786363, "grad_norm": 0.40339648723602295, "learning_rate": 0.0005, "loss": 1.1732773780822754, "step": 41050 }, { "epoch": 3.236184469846275, "grad_norm": 0.40275344252586365, "learning_rate": 0.0005, "loss": 1.2654401063919067, "step": 41051 }, { "epoch": 3.236263303113914, "grad_norm": 0.402140736579895, "learning_rate": 0.0005, "loss": 1.2394626140594482, "step": 41052 }, { "epoch": 3.236342136381553, "grad_norm": 1.3840396404266357, "learning_rate": 0.0005, "loss": 1.4384623765945435, "step": 41053 }, { "epoch": 3.236420969649192, "grad_norm": 0.5591399669647217, "learning_rate": 0.0005, "loss": 1.3631386756896973, "step": 41054 }, { "epoch": 3.236499802916831, "grad_norm": 0.49438658356666565, "learning_rate": 0.0005, "loss": 1.2916096448898315, "step": 41055 }, { "epoch": 3.2365786361844697, "grad_norm": 0.5295307040214539, "learning_rate": 0.0005, "loss": 1.2697683572769165, "step": 41056 }, { "epoch": 3.236657469452109, "grad_norm": 0.48495233058929443, "learning_rate": 0.0005, "loss": 1.3400282859802246, "step": 41057 }, { "epoch": 3.2367363027197475, "grad_norm": 0.5225116610527039, "learning_rate": 0.0005, "loss": 1.2558989524841309, "step": 41058 }, { "epoch": 3.2368151359873867, "grad_norm": 0.4159237742424011, "learning_rate": 0.0005, "loss": 1.1407926082611084, "step": 41059 }, { "epoch": 3.236893969255026, "grad_norm": 0.38021379709243774, "learning_rate": 0.0005, "loss": 1.1672279834747314, "step": 41060 }, { "epoch": 3.2369728025226645, "grad_norm": 0.647892415523529, "learning_rate": 0.0005, "loss": 1.427055835723877, "step": 41061 }, { "epoch": 3.2370516357903036, "grad_norm": 0.4234013557434082, "learning_rate": 0.0005, "loss": 1.2586441040039062, "step": 41062 }, { "epoch": 3.2371304690579423, "grad_norm": 0.5090274214744568, "learning_rate": 0.0005, "loss": 1.1913485527038574, "step": 41063 }, { "epoch": 3.2372093023255815, "grad_norm": 0.519709587097168, "learning_rate": 0.0005, "loss": 1.3729482889175415, "step": 41064 }, { "epoch": 3.23728813559322, "grad_norm": 0.46673649549484253, "learning_rate": 0.0005, "loss": 1.2039403915405273, "step": 41065 }, { "epoch": 3.2373669688608593, "grad_norm": 0.4326067566871643, "learning_rate": 0.0005, "loss": 1.256422996520996, "step": 41066 }, { "epoch": 3.2374458021284984, "grad_norm": 0.4352048337459564, "learning_rate": 0.0005, "loss": 1.0265390872955322, "step": 41067 }, { "epoch": 3.237524635396137, "grad_norm": 0.49792253971099854, "learning_rate": 0.0005, "loss": 1.2562103271484375, "step": 41068 }, { "epoch": 3.2376034686637762, "grad_norm": 0.556244969367981, "learning_rate": 0.0005, "loss": 1.0966756343841553, "step": 41069 }, { "epoch": 3.237682301931415, "grad_norm": 0.6030845642089844, "learning_rate": 0.0005, "loss": 1.3812739849090576, "step": 41070 }, { "epoch": 3.237761135199054, "grad_norm": 0.4943009614944458, "learning_rate": 0.0005, "loss": 1.4112000465393066, "step": 41071 }, { "epoch": 3.2378399684666928, "grad_norm": 0.5603412389755249, "learning_rate": 0.0005, "loss": 1.368852138519287, "step": 41072 }, { "epoch": 3.237918801734332, "grad_norm": 1.009977102279663, "learning_rate": 0.0005, "loss": 1.150108814239502, "step": 41073 }, { "epoch": 3.237997635001971, "grad_norm": 0.4199160933494568, "learning_rate": 0.0005, "loss": 1.2205142974853516, "step": 41074 }, { "epoch": 3.2380764682696097, "grad_norm": 0.6915788054466248, "learning_rate": 0.0005, "loss": 1.3865011930465698, "step": 41075 }, { "epoch": 3.238155301537249, "grad_norm": 0.4489811360836029, "learning_rate": 0.0005, "loss": 1.169645071029663, "step": 41076 }, { "epoch": 3.2382341348048875, "grad_norm": 0.4562015235424042, "learning_rate": 0.0005, "loss": 1.345794439315796, "step": 41077 }, { "epoch": 3.2383129680725267, "grad_norm": 0.5057214498519897, "learning_rate": 0.0005, "loss": 1.2219781875610352, "step": 41078 }, { "epoch": 3.2383918013401654, "grad_norm": 0.658223032951355, "learning_rate": 0.0005, "loss": 1.173285722732544, "step": 41079 }, { "epoch": 3.2384706346078045, "grad_norm": 0.5235193371772766, "learning_rate": 0.0005, "loss": 1.29423189163208, "step": 41080 }, { "epoch": 3.2385494678754436, "grad_norm": 0.669804036617279, "learning_rate": 0.0005, "loss": 1.233865737915039, "step": 41081 }, { "epoch": 3.2386283011430823, "grad_norm": 0.5103066563606262, "learning_rate": 0.0005, "loss": 1.2987494468688965, "step": 41082 }, { "epoch": 3.2387071344107214, "grad_norm": 0.5121967196464539, "learning_rate": 0.0005, "loss": 1.3965907096862793, "step": 41083 }, { "epoch": 3.23878596767836, "grad_norm": 0.5156020522117615, "learning_rate": 0.0005, "loss": 1.322723150253296, "step": 41084 }, { "epoch": 3.2388648009459993, "grad_norm": 0.5646011233329773, "learning_rate": 0.0005, "loss": 1.0410428047180176, "step": 41085 }, { "epoch": 3.238943634213638, "grad_norm": 0.5100728273391724, "learning_rate": 0.0005, "loss": 1.2309218645095825, "step": 41086 }, { "epoch": 3.239022467481277, "grad_norm": 0.4144975244998932, "learning_rate": 0.0005, "loss": 1.2292087078094482, "step": 41087 }, { "epoch": 3.2391013007489162, "grad_norm": 0.5325872302055359, "learning_rate": 0.0005, "loss": 1.1862753629684448, "step": 41088 }, { "epoch": 3.239180134016555, "grad_norm": 0.5085212588310242, "learning_rate": 0.0005, "loss": 1.1778805255889893, "step": 41089 }, { "epoch": 3.239258967284194, "grad_norm": 0.6029416918754578, "learning_rate": 0.0005, "loss": 1.3277740478515625, "step": 41090 }, { "epoch": 3.2393378005518327, "grad_norm": 0.5442660450935364, "learning_rate": 0.0005, "loss": 1.0835002660751343, "step": 41091 }, { "epoch": 3.239416633819472, "grad_norm": 0.42579635977745056, "learning_rate": 0.0005, "loss": 1.1710889339447021, "step": 41092 }, { "epoch": 3.2394954670871106, "grad_norm": 0.6128968596458435, "learning_rate": 0.0005, "loss": 1.2139681577682495, "step": 41093 }, { "epoch": 3.2395743003547497, "grad_norm": 0.48391881585121155, "learning_rate": 0.0005, "loss": 1.2525017261505127, "step": 41094 }, { "epoch": 3.239653133622389, "grad_norm": 0.45918893814086914, "learning_rate": 0.0005, "loss": 1.285020112991333, "step": 41095 }, { "epoch": 3.2397319668900275, "grad_norm": 0.5956730842590332, "learning_rate": 0.0005, "loss": 1.3893320560455322, "step": 41096 }, { "epoch": 3.2398108001576666, "grad_norm": 0.4319465458393097, "learning_rate": 0.0005, "loss": 1.1940321922302246, "step": 41097 }, { "epoch": 3.2398896334253053, "grad_norm": 0.4545774459838867, "learning_rate": 0.0005, "loss": 1.131246566772461, "step": 41098 }, { "epoch": 3.2399684666929445, "grad_norm": 0.48691561818122864, "learning_rate": 0.0005, "loss": 1.2723445892333984, "step": 41099 }, { "epoch": 3.240047299960583, "grad_norm": 0.5811123251914978, "learning_rate": 0.0005, "loss": 1.2566320896148682, "step": 41100 }, { "epoch": 3.2401261332282223, "grad_norm": 0.40112122893333435, "learning_rate": 0.0005, "loss": 1.125367522239685, "step": 41101 }, { "epoch": 3.2402049664958614, "grad_norm": 0.42077717185020447, "learning_rate": 0.0005, "loss": 1.1646816730499268, "step": 41102 }, { "epoch": 3.2402837997635, "grad_norm": 0.5611875653266907, "learning_rate": 0.0005, "loss": 1.2542011737823486, "step": 41103 }, { "epoch": 3.2403626330311393, "grad_norm": 0.6034494042396545, "learning_rate": 0.0005, "loss": 1.425809621810913, "step": 41104 }, { "epoch": 3.240441466298778, "grad_norm": 0.5104132294654846, "learning_rate": 0.0005, "loss": 1.2084399461746216, "step": 41105 }, { "epoch": 3.240520299566417, "grad_norm": 0.6596974730491638, "learning_rate": 0.0005, "loss": 1.2103402614593506, "step": 41106 }, { "epoch": 3.240599132834056, "grad_norm": 0.5416932702064514, "learning_rate": 0.0005, "loss": 1.1873172521591187, "step": 41107 }, { "epoch": 3.240677966101695, "grad_norm": 0.6337504982948303, "learning_rate": 0.0005, "loss": 1.3238409757614136, "step": 41108 }, { "epoch": 3.240756799369334, "grad_norm": 0.5291714072227478, "learning_rate": 0.0005, "loss": 1.4090023040771484, "step": 41109 }, { "epoch": 3.2408356326369727, "grad_norm": 0.5730088949203491, "learning_rate": 0.0005, "loss": 1.380946159362793, "step": 41110 }, { "epoch": 3.240914465904612, "grad_norm": 0.6059173345565796, "learning_rate": 0.0005, "loss": 1.3480143547058105, "step": 41111 }, { "epoch": 3.2409932991722505, "grad_norm": 0.4963993728160858, "learning_rate": 0.0005, "loss": 1.1138497591018677, "step": 41112 }, { "epoch": 3.2410721324398897, "grad_norm": 0.41950786113739014, "learning_rate": 0.0005, "loss": 1.3055908679962158, "step": 41113 }, { "epoch": 3.2411509657075284, "grad_norm": 0.4529964327812195, "learning_rate": 0.0005, "loss": 1.26863694190979, "step": 41114 }, { "epoch": 3.2412297989751675, "grad_norm": 0.48499077558517456, "learning_rate": 0.0005, "loss": 1.2393068075180054, "step": 41115 }, { "epoch": 3.2413086322428066, "grad_norm": 0.6391645669937134, "learning_rate": 0.0005, "loss": 1.5672833919525146, "step": 41116 }, { "epoch": 3.2413874655104453, "grad_norm": 0.48296651244163513, "learning_rate": 0.0005, "loss": 1.164677619934082, "step": 41117 }, { "epoch": 3.2414662987780845, "grad_norm": 0.5096278190612793, "learning_rate": 0.0005, "loss": 1.0361229181289673, "step": 41118 }, { "epoch": 3.241545132045723, "grad_norm": 0.46723058819770813, "learning_rate": 0.0005, "loss": 1.1445178985595703, "step": 41119 }, { "epoch": 3.2416239653133623, "grad_norm": 0.613327145576477, "learning_rate": 0.0005, "loss": 1.3273636102676392, "step": 41120 }, { "epoch": 3.2417027985810014, "grad_norm": 0.5731262564659119, "learning_rate": 0.0005, "loss": 1.2580326795578003, "step": 41121 }, { "epoch": 3.24178163184864, "grad_norm": 0.4894593060016632, "learning_rate": 0.0005, "loss": 1.2053178548812866, "step": 41122 }, { "epoch": 3.2418604651162792, "grad_norm": 0.5242700576782227, "learning_rate": 0.0005, "loss": 1.2309433221817017, "step": 41123 }, { "epoch": 3.241939298383918, "grad_norm": 0.5388328433036804, "learning_rate": 0.0005, "loss": 1.3903180360794067, "step": 41124 }, { "epoch": 3.242018131651557, "grad_norm": 0.5323700904846191, "learning_rate": 0.0005, "loss": 1.343674659729004, "step": 41125 }, { "epoch": 3.2420969649191957, "grad_norm": 0.5241785645484924, "learning_rate": 0.0005, "loss": 1.3838350772857666, "step": 41126 }, { "epoch": 3.242175798186835, "grad_norm": 0.4704020619392395, "learning_rate": 0.0005, "loss": 1.2225795984268188, "step": 41127 }, { "epoch": 3.2422546314544736, "grad_norm": 0.5225622653961182, "learning_rate": 0.0005, "loss": 1.1268534660339355, "step": 41128 }, { "epoch": 3.2423334647221127, "grad_norm": 0.46428874135017395, "learning_rate": 0.0005, "loss": 1.187796711921692, "step": 41129 }, { "epoch": 3.242412297989752, "grad_norm": 0.5982572436332703, "learning_rate": 0.0005, "loss": 1.424540638923645, "step": 41130 }, { "epoch": 3.2424911312573905, "grad_norm": 0.5461515188217163, "learning_rate": 0.0005, "loss": 1.3602428436279297, "step": 41131 }, { "epoch": 3.2425699645250297, "grad_norm": 0.5779017210006714, "learning_rate": 0.0005, "loss": 1.349544644355774, "step": 41132 }, { "epoch": 3.2426487977926683, "grad_norm": 0.4049364924430847, "learning_rate": 0.0005, "loss": 1.141648530960083, "step": 41133 }, { "epoch": 3.2427276310603075, "grad_norm": 0.6108742952346802, "learning_rate": 0.0005, "loss": 1.1655009984970093, "step": 41134 }, { "epoch": 3.2428064643279466, "grad_norm": 0.5816119313240051, "learning_rate": 0.0005, "loss": 1.1241939067840576, "step": 41135 }, { "epoch": 3.2428852975955853, "grad_norm": 0.6724477410316467, "learning_rate": 0.0005, "loss": 1.3575148582458496, "step": 41136 }, { "epoch": 3.2429641308632244, "grad_norm": 0.708691418170929, "learning_rate": 0.0005, "loss": 1.380783200263977, "step": 41137 }, { "epoch": 3.243042964130863, "grad_norm": 0.4751039743423462, "learning_rate": 0.0005, "loss": 1.2939724922180176, "step": 41138 }, { "epoch": 3.2431217973985023, "grad_norm": 0.4774065613746643, "learning_rate": 0.0005, "loss": 1.2321878671646118, "step": 41139 }, { "epoch": 3.243200630666141, "grad_norm": 0.4777087867259979, "learning_rate": 0.0005, "loss": 1.25950288772583, "step": 41140 }, { "epoch": 3.24327946393378, "grad_norm": 0.5024839043617249, "learning_rate": 0.0005, "loss": 1.3711196184158325, "step": 41141 }, { "epoch": 3.2433582972014188, "grad_norm": 0.419479101896286, "learning_rate": 0.0005, "loss": 1.3035281896591187, "step": 41142 }, { "epoch": 3.243437130469058, "grad_norm": 0.56749027967453, "learning_rate": 0.0005, "loss": 1.3594200611114502, "step": 41143 }, { "epoch": 3.243515963736697, "grad_norm": 0.4650183320045471, "learning_rate": 0.0005, "loss": 1.2087304592132568, "step": 41144 }, { "epoch": 3.2435947970043357, "grad_norm": 0.4608229100704193, "learning_rate": 0.0005, "loss": 1.2281475067138672, "step": 41145 }, { "epoch": 3.243673630271975, "grad_norm": 0.5549835562705994, "learning_rate": 0.0005, "loss": 1.2754526138305664, "step": 41146 }, { "epoch": 3.2437524635396136, "grad_norm": 0.5403249859809875, "learning_rate": 0.0005, "loss": 1.4007230997085571, "step": 41147 }, { "epoch": 3.2438312968072527, "grad_norm": 0.47974997758865356, "learning_rate": 0.0005, "loss": 1.3041651248931885, "step": 41148 }, { "epoch": 3.243910130074892, "grad_norm": 0.4384171962738037, "learning_rate": 0.0005, "loss": 1.0533138513565063, "step": 41149 }, { "epoch": 3.2439889633425305, "grad_norm": 0.5690820813179016, "learning_rate": 0.0005, "loss": 1.3678405284881592, "step": 41150 }, { "epoch": 3.2440677966101696, "grad_norm": 0.533080518245697, "learning_rate": 0.0005, "loss": 1.271950364112854, "step": 41151 }, { "epoch": 3.2441466298778083, "grad_norm": 0.5566679835319519, "learning_rate": 0.0005, "loss": 1.329636573791504, "step": 41152 }, { "epoch": 3.2442254631454475, "grad_norm": 0.5050093531608582, "learning_rate": 0.0005, "loss": 1.2380764484405518, "step": 41153 }, { "epoch": 3.244304296413086, "grad_norm": 0.46766069531440735, "learning_rate": 0.0005, "loss": 1.1982638835906982, "step": 41154 }, { "epoch": 3.2443831296807253, "grad_norm": 0.4110772907733917, "learning_rate": 0.0005, "loss": 1.2274441719055176, "step": 41155 }, { "epoch": 3.244461962948364, "grad_norm": 0.49179044365882874, "learning_rate": 0.0005, "loss": 1.2687654495239258, "step": 41156 }, { "epoch": 3.244540796216003, "grad_norm": 0.49301964044570923, "learning_rate": 0.0005, "loss": 1.293985366821289, "step": 41157 }, { "epoch": 3.2446196294836422, "grad_norm": 0.5926015973091125, "learning_rate": 0.0005, "loss": 1.3023760318756104, "step": 41158 }, { "epoch": 3.244698462751281, "grad_norm": 0.5028121471405029, "learning_rate": 0.0005, "loss": 1.3487852811813354, "step": 41159 }, { "epoch": 3.24477729601892, "grad_norm": 0.43977123498916626, "learning_rate": 0.0005, "loss": 1.317923903465271, "step": 41160 }, { "epoch": 3.2448561292865588, "grad_norm": 0.4685337543487549, "learning_rate": 0.0005, "loss": 1.332154631614685, "step": 41161 }, { "epoch": 3.244934962554198, "grad_norm": 0.4879566431045532, "learning_rate": 0.0005, "loss": 1.3494389057159424, "step": 41162 }, { "epoch": 3.245013795821837, "grad_norm": 0.6113911271095276, "learning_rate": 0.0005, "loss": 1.2473926544189453, "step": 41163 }, { "epoch": 3.2450926290894757, "grad_norm": 0.5082280039787292, "learning_rate": 0.0005, "loss": 1.472517728805542, "step": 41164 }, { "epoch": 3.245171462357115, "grad_norm": 0.42678412795066833, "learning_rate": 0.0005, "loss": 1.1851696968078613, "step": 41165 }, { "epoch": 3.2452502956247535, "grad_norm": 0.4957382380962372, "learning_rate": 0.0005, "loss": 1.306965708732605, "step": 41166 }, { "epoch": 3.2453291288923927, "grad_norm": 0.45778688788414, "learning_rate": 0.0005, "loss": 1.2950780391693115, "step": 41167 }, { "epoch": 3.2454079621600314, "grad_norm": 0.46633487939834595, "learning_rate": 0.0005, "loss": 1.373844861984253, "step": 41168 }, { "epoch": 3.2454867954276705, "grad_norm": 0.407785564661026, "learning_rate": 0.0005, "loss": 1.155517578125, "step": 41169 }, { "epoch": 3.2455656286953096, "grad_norm": 0.4284050166606903, "learning_rate": 0.0005, "loss": 1.2957756519317627, "step": 41170 }, { "epoch": 3.2456444619629483, "grad_norm": 0.4270857572555542, "learning_rate": 0.0005, "loss": 1.1219995021820068, "step": 41171 }, { "epoch": 3.2457232952305874, "grad_norm": 0.5522539615631104, "learning_rate": 0.0005, "loss": 1.282012701034546, "step": 41172 }, { "epoch": 3.245802128498226, "grad_norm": 0.5644921660423279, "learning_rate": 0.0005, "loss": 1.2975574731826782, "step": 41173 }, { "epoch": 3.2458809617658653, "grad_norm": 0.5272038578987122, "learning_rate": 0.0005, "loss": 1.1736499071121216, "step": 41174 }, { "epoch": 3.245959795033504, "grad_norm": 0.47466668486595154, "learning_rate": 0.0005, "loss": 1.2775275707244873, "step": 41175 }, { "epoch": 3.246038628301143, "grad_norm": 0.4403704106807709, "learning_rate": 0.0005, "loss": 1.0740841627120972, "step": 41176 }, { "epoch": 3.2461174615687822, "grad_norm": 0.44449496269226074, "learning_rate": 0.0005, "loss": 1.1180343627929688, "step": 41177 }, { "epoch": 3.246196294836421, "grad_norm": 0.45986631512641907, "learning_rate": 0.0005, "loss": 1.299831509590149, "step": 41178 }, { "epoch": 3.24627512810406, "grad_norm": 0.4295060634613037, "learning_rate": 0.0005, "loss": 1.3468480110168457, "step": 41179 }, { "epoch": 3.2463539613716987, "grad_norm": 0.41570988297462463, "learning_rate": 0.0005, "loss": 1.1398035287857056, "step": 41180 }, { "epoch": 3.246432794639338, "grad_norm": 0.5025327801704407, "learning_rate": 0.0005, "loss": 1.192793846130371, "step": 41181 }, { "epoch": 3.2465116279069766, "grad_norm": 0.5575039386749268, "learning_rate": 0.0005, "loss": 1.231771469116211, "step": 41182 }, { "epoch": 3.2465904611746157, "grad_norm": 0.4113818109035492, "learning_rate": 0.0005, "loss": 1.3461017608642578, "step": 41183 }, { "epoch": 3.246669294442255, "grad_norm": 0.5747336745262146, "learning_rate": 0.0005, "loss": 1.2420167922973633, "step": 41184 }, { "epoch": 3.2467481277098935, "grad_norm": 0.5400230884552002, "learning_rate": 0.0005, "loss": 1.441267728805542, "step": 41185 }, { "epoch": 3.2468269609775327, "grad_norm": 0.5002279877662659, "learning_rate": 0.0005, "loss": 1.2865607738494873, "step": 41186 }, { "epoch": 3.2469057942451713, "grad_norm": 0.4720093011856079, "learning_rate": 0.0005, "loss": 1.3020161390304565, "step": 41187 }, { "epoch": 3.2469846275128105, "grad_norm": 0.4208812713623047, "learning_rate": 0.0005, "loss": 1.2577216625213623, "step": 41188 }, { "epoch": 3.247063460780449, "grad_norm": 0.5121655464172363, "learning_rate": 0.0005, "loss": 1.1563010215759277, "step": 41189 }, { "epoch": 3.2471422940480883, "grad_norm": 0.3452637195587158, "learning_rate": 0.0005, "loss": 1.0789850950241089, "step": 41190 }, { "epoch": 3.2472211273157274, "grad_norm": 0.6201347708702087, "learning_rate": 0.0005, "loss": 1.3875685930252075, "step": 41191 }, { "epoch": 3.247299960583366, "grad_norm": 0.458840548992157, "learning_rate": 0.0005, "loss": 1.173257827758789, "step": 41192 }, { "epoch": 3.2473787938510053, "grad_norm": 0.4838944971561432, "learning_rate": 0.0005, "loss": 1.1707117557525635, "step": 41193 }, { "epoch": 3.247457627118644, "grad_norm": 0.4515765309333801, "learning_rate": 0.0005, "loss": 1.1924139261245728, "step": 41194 }, { "epoch": 3.247536460386283, "grad_norm": 1.0097880363464355, "learning_rate": 0.0005, "loss": 1.1507487297058105, "step": 41195 }, { "epoch": 3.2476152936539218, "grad_norm": 0.7533256411552429, "learning_rate": 0.0005, "loss": 1.3703151941299438, "step": 41196 }, { "epoch": 3.247694126921561, "grad_norm": 0.5024777054786682, "learning_rate": 0.0005, "loss": 1.2253254652023315, "step": 41197 }, { "epoch": 3.2477729601892, "grad_norm": 0.7514162063598633, "learning_rate": 0.0005, "loss": 1.281158447265625, "step": 41198 }, { "epoch": 3.2478517934568387, "grad_norm": 0.6016988754272461, "learning_rate": 0.0005, "loss": 1.2035727500915527, "step": 41199 }, { "epoch": 3.247930626724478, "grad_norm": 0.4943963289260864, "learning_rate": 0.0005, "loss": 1.363116979598999, "step": 41200 }, { "epoch": 3.2480094599921165, "grad_norm": 1.988366961479187, "learning_rate": 0.0005, "loss": 1.3810595273971558, "step": 41201 }, { "epoch": 3.2480882932597557, "grad_norm": 0.5080608129501343, "learning_rate": 0.0005, "loss": 1.1075329780578613, "step": 41202 }, { "epoch": 3.2481671265273944, "grad_norm": 0.45474734902381897, "learning_rate": 0.0005, "loss": 1.2382007837295532, "step": 41203 }, { "epoch": 3.2482459597950335, "grad_norm": 0.48355692625045776, "learning_rate": 0.0005, "loss": 1.1955054998397827, "step": 41204 }, { "epoch": 3.2483247930626726, "grad_norm": 0.5216706991195679, "learning_rate": 0.0005, "loss": 1.3521194458007812, "step": 41205 }, { "epoch": 3.2484036263303113, "grad_norm": 0.5151782631874084, "learning_rate": 0.0005, "loss": 1.1178733110427856, "step": 41206 }, { "epoch": 3.2484824595979505, "grad_norm": 0.5762561559677124, "learning_rate": 0.0005, "loss": 1.3945908546447754, "step": 41207 }, { "epoch": 3.248561292865589, "grad_norm": 0.41202062368392944, "learning_rate": 0.0005, "loss": 1.2222546339035034, "step": 41208 }, { "epoch": 3.2486401261332283, "grad_norm": 0.7123037576675415, "learning_rate": 0.0005, "loss": 1.5253616571426392, "step": 41209 }, { "epoch": 3.248718959400867, "grad_norm": 0.5328428149223328, "learning_rate": 0.0005, "loss": 1.3570404052734375, "step": 41210 }, { "epoch": 3.248797792668506, "grad_norm": 0.7403106689453125, "learning_rate": 0.0005, "loss": 1.4962152242660522, "step": 41211 }, { "epoch": 3.2488766259361452, "grad_norm": 0.5037489533424377, "learning_rate": 0.0005, "loss": 1.1946396827697754, "step": 41212 }, { "epoch": 3.248955459203784, "grad_norm": 0.6168802380561829, "learning_rate": 0.0005, "loss": 1.3281841278076172, "step": 41213 }, { "epoch": 3.249034292471423, "grad_norm": 0.5331000685691833, "learning_rate": 0.0005, "loss": 1.305604338645935, "step": 41214 }, { "epoch": 3.2491131257390617, "grad_norm": 0.46111413836479187, "learning_rate": 0.0005, "loss": 1.3496506214141846, "step": 41215 }, { "epoch": 3.249191959006701, "grad_norm": 0.5605401396751404, "learning_rate": 0.0005, "loss": 1.3689262866973877, "step": 41216 }, { "epoch": 3.2492707922743396, "grad_norm": 0.5889496207237244, "learning_rate": 0.0005, "loss": 1.3446335792541504, "step": 41217 }, { "epoch": 3.2493496255419787, "grad_norm": 0.5330469012260437, "learning_rate": 0.0005, "loss": 1.1188641786575317, "step": 41218 }, { "epoch": 3.249428458809618, "grad_norm": 0.4955160617828369, "learning_rate": 0.0005, "loss": 1.2858562469482422, "step": 41219 }, { "epoch": 3.2495072920772565, "grad_norm": 0.48726174235343933, "learning_rate": 0.0005, "loss": 1.2575395107269287, "step": 41220 }, { "epoch": 3.2495861253448957, "grad_norm": 0.47620800137519836, "learning_rate": 0.0005, "loss": 1.159561038017273, "step": 41221 }, { "epoch": 3.2496649586125343, "grad_norm": 0.5292633175849915, "learning_rate": 0.0005, "loss": 1.1214394569396973, "step": 41222 }, { "epoch": 3.2497437918801735, "grad_norm": 0.4476287364959717, "learning_rate": 0.0005, "loss": 1.3239580392837524, "step": 41223 }, { "epoch": 3.249822625147812, "grad_norm": 0.6172661185264587, "learning_rate": 0.0005, "loss": 1.3135473728179932, "step": 41224 }, { "epoch": 3.2499014584154513, "grad_norm": 0.49666810035705566, "learning_rate": 0.0005, "loss": 1.27577805519104, "step": 41225 }, { "epoch": 3.2499802916830904, "grad_norm": 0.5398963093757629, "learning_rate": 0.0005, "loss": 1.3680304288864136, "step": 41226 }, { "epoch": 3.250059124950729, "grad_norm": 0.493581086397171, "learning_rate": 0.0005, "loss": 1.2748291492462158, "step": 41227 }, { "epoch": 3.2501379582183683, "grad_norm": 0.5101732611656189, "learning_rate": 0.0005, "loss": 1.2645264863967896, "step": 41228 }, { "epoch": 3.250216791486007, "grad_norm": 0.5048116445541382, "learning_rate": 0.0005, "loss": 1.1986312866210938, "step": 41229 }, { "epoch": 3.250295624753646, "grad_norm": 0.5532713532447815, "learning_rate": 0.0005, "loss": 1.4762964248657227, "step": 41230 }, { "epoch": 3.2503744580212848, "grad_norm": 0.4019574522972107, "learning_rate": 0.0005, "loss": 1.1812708377838135, "step": 41231 }, { "epoch": 3.250453291288924, "grad_norm": 0.7318173050880432, "learning_rate": 0.0005, "loss": 1.4380775690078735, "step": 41232 }, { "epoch": 3.250532124556563, "grad_norm": 0.5016492605209351, "learning_rate": 0.0005, "loss": 1.3121795654296875, "step": 41233 }, { "epoch": 3.2506109578242017, "grad_norm": 0.5326088070869446, "learning_rate": 0.0005, "loss": 1.297546148300171, "step": 41234 }, { "epoch": 3.250689791091841, "grad_norm": 0.620925784111023, "learning_rate": 0.0005, "loss": 1.4452763795852661, "step": 41235 }, { "epoch": 3.2507686243594796, "grad_norm": 0.5518088936805725, "learning_rate": 0.0005, "loss": 1.266813039779663, "step": 41236 }, { "epoch": 3.2508474576271187, "grad_norm": 0.4714052379131317, "learning_rate": 0.0005, "loss": 1.2970868349075317, "step": 41237 }, { "epoch": 3.250926290894758, "grad_norm": 0.4930787682533264, "learning_rate": 0.0005, "loss": 1.0742394924163818, "step": 41238 }, { "epoch": 3.2510051241623965, "grad_norm": 0.5363984704017639, "learning_rate": 0.0005, "loss": 1.2987844944000244, "step": 41239 }, { "epoch": 3.2510839574300356, "grad_norm": 0.48355749249458313, "learning_rate": 0.0005, "loss": 1.2716249227523804, "step": 41240 }, { "epoch": 3.2511627906976743, "grad_norm": 0.5073627829551697, "learning_rate": 0.0005, "loss": 1.19820237159729, "step": 41241 }, { "epoch": 3.2512416239653135, "grad_norm": 0.4788415729999542, "learning_rate": 0.0005, "loss": 1.2783207893371582, "step": 41242 }, { "epoch": 3.251320457232952, "grad_norm": 0.547149121761322, "learning_rate": 0.0005, "loss": 1.217618465423584, "step": 41243 }, { "epoch": 3.2513992905005913, "grad_norm": 0.5002129077911377, "learning_rate": 0.0005, "loss": 1.1787232160568237, "step": 41244 }, { "epoch": 3.25147812376823, "grad_norm": 0.6249738335609436, "learning_rate": 0.0005, "loss": 1.5241050720214844, "step": 41245 }, { "epoch": 3.251556957035869, "grad_norm": 0.4424053728580475, "learning_rate": 0.0005, "loss": 1.1431289911270142, "step": 41246 }, { "epoch": 3.2516357903035082, "grad_norm": 0.9063099026679993, "learning_rate": 0.0005, "loss": 1.41536283493042, "step": 41247 }, { "epoch": 3.251714623571147, "grad_norm": 0.6993857026100159, "learning_rate": 0.0005, "loss": 1.2298195362091064, "step": 41248 }, { "epoch": 3.251793456838786, "grad_norm": 0.4855272173881531, "learning_rate": 0.0005, "loss": 1.2270760536193848, "step": 41249 }, { "epoch": 3.2518722901064248, "grad_norm": 0.4984653890132904, "learning_rate": 0.0005, "loss": 1.2376655340194702, "step": 41250 }, { "epoch": 3.251951123374064, "grad_norm": 0.46208956837654114, "learning_rate": 0.0005, "loss": 1.2719695568084717, "step": 41251 }, { "epoch": 3.252029956641703, "grad_norm": 0.5168904066085815, "learning_rate": 0.0005, "loss": 1.3637266159057617, "step": 41252 }, { "epoch": 3.2521087899093417, "grad_norm": 0.4859212636947632, "learning_rate": 0.0005, "loss": 1.1721941232681274, "step": 41253 }, { "epoch": 3.252187623176981, "grad_norm": 0.5035250782966614, "learning_rate": 0.0005, "loss": 1.1774914264678955, "step": 41254 }, { "epoch": 3.2522664564446195, "grad_norm": 0.5075015425682068, "learning_rate": 0.0005, "loss": 1.1834332942962646, "step": 41255 }, { "epoch": 3.2523452897122587, "grad_norm": 0.4314023554325104, "learning_rate": 0.0005, "loss": 1.0757887363433838, "step": 41256 }, { "epoch": 3.2524241229798974, "grad_norm": 0.4962129294872284, "learning_rate": 0.0005, "loss": 1.210559606552124, "step": 41257 }, { "epoch": 3.2525029562475365, "grad_norm": 0.44304776191711426, "learning_rate": 0.0005, "loss": 1.2292308807373047, "step": 41258 }, { "epoch": 3.252581789515175, "grad_norm": 0.41245466470718384, "learning_rate": 0.0005, "loss": 1.273241639137268, "step": 41259 }, { "epoch": 3.2526606227828143, "grad_norm": 0.48448634147644043, "learning_rate": 0.0005, "loss": 1.3737568855285645, "step": 41260 }, { "epoch": 3.2527394560504534, "grad_norm": 0.49325138330459595, "learning_rate": 0.0005, "loss": 1.2339420318603516, "step": 41261 }, { "epoch": 3.252818289318092, "grad_norm": 0.5170422196388245, "learning_rate": 0.0005, "loss": 1.3293390274047852, "step": 41262 }, { "epoch": 3.2528971225857313, "grad_norm": 0.4441342055797577, "learning_rate": 0.0005, "loss": 1.1160086393356323, "step": 41263 }, { "epoch": 3.25297595585337, "grad_norm": 0.4768804609775543, "learning_rate": 0.0005, "loss": 1.203073263168335, "step": 41264 }, { "epoch": 3.253054789121009, "grad_norm": 0.430630624294281, "learning_rate": 0.0005, "loss": 1.0758030414581299, "step": 41265 }, { "epoch": 3.2531336223886482, "grad_norm": 0.5210658311843872, "learning_rate": 0.0005, "loss": 1.2625112533569336, "step": 41266 }, { "epoch": 3.253212455656287, "grad_norm": 0.5979775190353394, "learning_rate": 0.0005, "loss": 1.1644983291625977, "step": 41267 }, { "epoch": 3.253291288923926, "grad_norm": 0.5142203569412231, "learning_rate": 0.0005, "loss": 1.2528040409088135, "step": 41268 }, { "epoch": 3.2533701221915647, "grad_norm": 0.465583860874176, "learning_rate": 0.0005, "loss": 1.3061941862106323, "step": 41269 }, { "epoch": 3.253448955459204, "grad_norm": 0.5792616009712219, "learning_rate": 0.0005, "loss": 1.2441210746765137, "step": 41270 }, { "epoch": 3.2535277887268426, "grad_norm": 0.43174058198928833, "learning_rate": 0.0005, "loss": 1.179365873336792, "step": 41271 }, { "epoch": 3.2536066219944817, "grad_norm": 0.386227548122406, "learning_rate": 0.0005, "loss": 1.1220076084136963, "step": 41272 }, { "epoch": 3.2536854552621204, "grad_norm": 0.49750474095344543, "learning_rate": 0.0005, "loss": 1.2440266609191895, "step": 41273 }, { "epoch": 3.2537642885297595, "grad_norm": 0.5674493312835693, "learning_rate": 0.0005, "loss": 1.292635202407837, "step": 41274 }, { "epoch": 3.2538431217973987, "grad_norm": 0.4646761119365692, "learning_rate": 0.0005, "loss": 1.287408709526062, "step": 41275 }, { "epoch": 3.2539219550650373, "grad_norm": 0.48794427514076233, "learning_rate": 0.0005, "loss": 1.4229971170425415, "step": 41276 }, { "epoch": 3.2540007883326765, "grad_norm": 0.5670682787895203, "learning_rate": 0.0005, "loss": 1.5048296451568604, "step": 41277 }, { "epoch": 3.254079621600315, "grad_norm": 0.6044489741325378, "learning_rate": 0.0005, "loss": 1.479354739189148, "step": 41278 }, { "epoch": 3.2541584548679543, "grad_norm": 0.4292144179344177, "learning_rate": 0.0005, "loss": 1.1215980052947998, "step": 41279 }, { "epoch": 3.2542372881355934, "grad_norm": 0.37501153349876404, "learning_rate": 0.0005, "loss": 1.0648441314697266, "step": 41280 }, { "epoch": 3.254316121403232, "grad_norm": 0.4665220379829407, "learning_rate": 0.0005, "loss": 1.1206612586975098, "step": 41281 }, { "epoch": 3.2543949546708713, "grad_norm": 0.5902922749519348, "learning_rate": 0.0005, "loss": 1.3530302047729492, "step": 41282 }, { "epoch": 3.25447378793851, "grad_norm": 0.6207172274589539, "learning_rate": 0.0005, "loss": 1.3863894939422607, "step": 41283 }, { "epoch": 3.254552621206149, "grad_norm": 0.47332578897476196, "learning_rate": 0.0005, "loss": 1.1916298866271973, "step": 41284 }, { "epoch": 3.2546314544737878, "grad_norm": 0.3990423381328583, "learning_rate": 0.0005, "loss": 1.1095788478851318, "step": 41285 }, { "epoch": 3.254710287741427, "grad_norm": 0.5712389945983887, "learning_rate": 0.0005, "loss": 1.3492488861083984, "step": 41286 }, { "epoch": 3.2547891210090656, "grad_norm": 0.4415302276611328, "learning_rate": 0.0005, "loss": 1.1702079772949219, "step": 41287 }, { "epoch": 3.2548679542767047, "grad_norm": 0.5221429467201233, "learning_rate": 0.0005, "loss": 1.4968607425689697, "step": 41288 }, { "epoch": 3.254946787544344, "grad_norm": 0.5121752619743347, "learning_rate": 0.0005, "loss": 1.326019525527954, "step": 41289 }, { "epoch": 3.2550256208119825, "grad_norm": 0.47403380274772644, "learning_rate": 0.0005, "loss": 1.2844092845916748, "step": 41290 }, { "epoch": 3.2551044540796217, "grad_norm": 0.6228576898574829, "learning_rate": 0.0005, "loss": 1.5343711376190186, "step": 41291 }, { "epoch": 3.2551832873472604, "grad_norm": 0.5256320834159851, "learning_rate": 0.0005, "loss": 1.3100464344024658, "step": 41292 }, { "epoch": 3.2552621206148995, "grad_norm": 0.6667754054069519, "learning_rate": 0.0005, "loss": 1.278864860534668, "step": 41293 }, { "epoch": 3.2553409538825386, "grad_norm": 0.4562358260154724, "learning_rate": 0.0005, "loss": 1.137761116027832, "step": 41294 }, { "epoch": 3.2554197871501773, "grad_norm": 0.4704534411430359, "learning_rate": 0.0005, "loss": 1.119810700416565, "step": 41295 }, { "epoch": 3.2554986204178165, "grad_norm": 0.5505608916282654, "learning_rate": 0.0005, "loss": 1.4662880897521973, "step": 41296 }, { "epoch": 3.255577453685455, "grad_norm": 0.4743283987045288, "learning_rate": 0.0005, "loss": 1.2791615724563599, "step": 41297 }, { "epoch": 3.2556562869530943, "grad_norm": 0.7146018743515015, "learning_rate": 0.0005, "loss": 1.2356934547424316, "step": 41298 }, { "epoch": 3.255735120220733, "grad_norm": 0.5737740993499756, "learning_rate": 0.0005, "loss": 1.3408472537994385, "step": 41299 }, { "epoch": 3.255813953488372, "grad_norm": 0.5070478916168213, "learning_rate": 0.0005, "loss": 1.3549036979675293, "step": 41300 }, { "epoch": 3.255892786756011, "grad_norm": 0.44076982140541077, "learning_rate": 0.0005, "loss": 1.2378824949264526, "step": 41301 }, { "epoch": 3.25597162002365, "grad_norm": 0.5706801414489746, "learning_rate": 0.0005, "loss": 1.5316174030303955, "step": 41302 }, { "epoch": 3.256050453291289, "grad_norm": 0.5692708492279053, "learning_rate": 0.0005, "loss": 1.3601747751235962, "step": 41303 }, { "epoch": 3.2561292865589277, "grad_norm": 0.5237729549407959, "learning_rate": 0.0005, "loss": 1.276020884513855, "step": 41304 }, { "epoch": 3.256208119826567, "grad_norm": 0.5461746454238892, "learning_rate": 0.0005, "loss": 1.4493076801300049, "step": 41305 }, { "epoch": 3.2562869530942056, "grad_norm": 0.43112146854400635, "learning_rate": 0.0005, "loss": 1.1595752239227295, "step": 41306 }, { "epoch": 3.2563657863618447, "grad_norm": 0.5348663926124573, "learning_rate": 0.0005, "loss": 1.2952206134796143, "step": 41307 }, { "epoch": 3.256444619629484, "grad_norm": 0.4246830642223358, "learning_rate": 0.0005, "loss": 1.0518007278442383, "step": 41308 }, { "epoch": 3.2565234528971225, "grad_norm": 0.4594987630844116, "learning_rate": 0.0005, "loss": 1.2572894096374512, "step": 41309 }, { "epoch": 3.2566022861647617, "grad_norm": 0.45993179082870483, "learning_rate": 0.0005, "loss": 1.1798667907714844, "step": 41310 }, { "epoch": 3.2566811194324004, "grad_norm": 0.4458373785018921, "learning_rate": 0.0005, "loss": 1.3417530059814453, "step": 41311 }, { "epoch": 3.2567599527000395, "grad_norm": 0.4997309148311615, "learning_rate": 0.0005, "loss": 1.1312596797943115, "step": 41312 }, { "epoch": 3.256838785967678, "grad_norm": 0.5190314650535583, "learning_rate": 0.0005, "loss": 1.2120757102966309, "step": 41313 }, { "epoch": 3.2569176192353173, "grad_norm": 0.419391393661499, "learning_rate": 0.0005, "loss": 1.2652881145477295, "step": 41314 }, { "epoch": 3.2569964525029564, "grad_norm": 0.5186970829963684, "learning_rate": 0.0005, "loss": 1.42916738986969, "step": 41315 }, { "epoch": 3.257075285770595, "grad_norm": 0.4500555992126465, "learning_rate": 0.0005, "loss": 1.228025197982788, "step": 41316 }, { "epoch": 3.2571541190382343, "grad_norm": 0.5422782301902771, "learning_rate": 0.0005, "loss": 1.3004223108291626, "step": 41317 }, { "epoch": 3.257232952305873, "grad_norm": 0.5205091238021851, "learning_rate": 0.0005, "loss": 1.3200490474700928, "step": 41318 }, { "epoch": 3.257311785573512, "grad_norm": 0.4891775846481323, "learning_rate": 0.0005, "loss": 1.3500409126281738, "step": 41319 }, { "epoch": 3.2573906188411508, "grad_norm": 0.4565182030200958, "learning_rate": 0.0005, "loss": 1.183782935142517, "step": 41320 }, { "epoch": 3.25746945210879, "grad_norm": 0.479397714138031, "learning_rate": 0.0005, "loss": 1.2362377643585205, "step": 41321 }, { "epoch": 3.257548285376429, "grad_norm": 0.5195869207382202, "learning_rate": 0.0005, "loss": 1.171806812286377, "step": 41322 }, { "epoch": 3.2576271186440677, "grad_norm": 0.45364439487457275, "learning_rate": 0.0005, "loss": 1.2206921577453613, "step": 41323 }, { "epoch": 3.257705951911707, "grad_norm": 0.4252017140388489, "learning_rate": 0.0005, "loss": 1.3021725416183472, "step": 41324 }, { "epoch": 3.2577847851793456, "grad_norm": 0.515052318572998, "learning_rate": 0.0005, "loss": 1.2404577732086182, "step": 41325 }, { "epoch": 3.2578636184469847, "grad_norm": 0.4776219427585602, "learning_rate": 0.0005, "loss": 1.3421051502227783, "step": 41326 }, { "epoch": 3.2579424517146234, "grad_norm": 0.40631163120269775, "learning_rate": 0.0005, "loss": 1.105337142944336, "step": 41327 }, { "epoch": 3.2580212849822625, "grad_norm": 0.48647522926330566, "learning_rate": 0.0005, "loss": 1.2742550373077393, "step": 41328 }, { "epoch": 3.2581001182499016, "grad_norm": 0.48768821358680725, "learning_rate": 0.0005, "loss": 1.4013687372207642, "step": 41329 }, { "epoch": 3.2581789515175403, "grad_norm": 0.5822628140449524, "learning_rate": 0.0005, "loss": 1.2340670824050903, "step": 41330 }, { "epoch": 3.2582577847851795, "grad_norm": 0.48323285579681396, "learning_rate": 0.0005, "loss": 1.1435472965240479, "step": 41331 }, { "epoch": 3.258336618052818, "grad_norm": 0.4700906574726105, "learning_rate": 0.0005, "loss": 1.1932990550994873, "step": 41332 }, { "epoch": 3.2584154513204573, "grad_norm": 0.6577528715133667, "learning_rate": 0.0005, "loss": 1.233954668045044, "step": 41333 }, { "epoch": 3.258494284588096, "grad_norm": 0.4623844623565674, "learning_rate": 0.0005, "loss": 1.3579282760620117, "step": 41334 }, { "epoch": 3.258573117855735, "grad_norm": 0.45040833950042725, "learning_rate": 0.0005, "loss": 1.2595429420471191, "step": 41335 }, { "epoch": 3.2586519511233742, "grad_norm": 0.5934809446334839, "learning_rate": 0.0005, "loss": 1.3622523546218872, "step": 41336 }, { "epoch": 3.258730784391013, "grad_norm": 0.4622778296470642, "learning_rate": 0.0005, "loss": 1.1432700157165527, "step": 41337 }, { "epoch": 3.258809617658652, "grad_norm": 0.43818944692611694, "learning_rate": 0.0005, "loss": 1.2571377754211426, "step": 41338 }, { "epoch": 3.2588884509262908, "grad_norm": 0.5135265588760376, "learning_rate": 0.0005, "loss": 1.5176656246185303, "step": 41339 }, { "epoch": 3.25896728419393, "grad_norm": 0.5483214855194092, "learning_rate": 0.0005, "loss": 1.53554368019104, "step": 41340 }, { "epoch": 3.259046117461569, "grad_norm": 0.5528229475021362, "learning_rate": 0.0005, "loss": 1.3195011615753174, "step": 41341 }, { "epoch": 3.2591249507292077, "grad_norm": 0.48168307542800903, "learning_rate": 0.0005, "loss": 1.3205232620239258, "step": 41342 }, { "epoch": 3.259203783996847, "grad_norm": 0.46321243047714233, "learning_rate": 0.0005, "loss": 1.1337909698486328, "step": 41343 }, { "epoch": 3.2592826172644855, "grad_norm": 0.48772403597831726, "learning_rate": 0.0005, "loss": 1.2021015882492065, "step": 41344 }, { "epoch": 3.2593614505321247, "grad_norm": 0.43753308057785034, "learning_rate": 0.0005, "loss": 1.213818907737732, "step": 41345 }, { "epoch": 3.2594402837997634, "grad_norm": 0.47988027334213257, "learning_rate": 0.0005, "loss": 1.384192705154419, "step": 41346 }, { "epoch": 3.2595191170674025, "grad_norm": 0.39977407455444336, "learning_rate": 0.0005, "loss": 1.1626248359680176, "step": 41347 }, { "epoch": 3.259597950335041, "grad_norm": 0.5321803689002991, "learning_rate": 0.0005, "loss": 1.3887875080108643, "step": 41348 }, { "epoch": 3.2596767836026803, "grad_norm": 0.5354970097541809, "learning_rate": 0.0005, "loss": 1.3922603130340576, "step": 41349 }, { "epoch": 3.2597556168703195, "grad_norm": 0.5182212591171265, "learning_rate": 0.0005, "loss": 1.45097017288208, "step": 41350 }, { "epoch": 3.259834450137958, "grad_norm": 0.434685617685318, "learning_rate": 0.0005, "loss": 1.189119577407837, "step": 41351 }, { "epoch": 3.2599132834055973, "grad_norm": 0.4993115961551666, "learning_rate": 0.0005, "loss": 1.3116850852966309, "step": 41352 }, { "epoch": 3.259992116673236, "grad_norm": 0.4949870705604553, "learning_rate": 0.0005, "loss": 1.4437317848205566, "step": 41353 }, { "epoch": 3.260070949940875, "grad_norm": 0.5002944469451904, "learning_rate": 0.0005, "loss": 1.269087791442871, "step": 41354 }, { "epoch": 3.2601497832085142, "grad_norm": 0.5186412334442139, "learning_rate": 0.0005, "loss": 1.4105134010314941, "step": 41355 }, { "epoch": 3.260228616476153, "grad_norm": 0.48407888412475586, "learning_rate": 0.0005, "loss": 1.3285713195800781, "step": 41356 }, { "epoch": 3.260307449743792, "grad_norm": 0.47099608182907104, "learning_rate": 0.0005, "loss": 1.196699857711792, "step": 41357 }, { "epoch": 3.2603862830114307, "grad_norm": 0.4569717347621918, "learning_rate": 0.0005, "loss": 1.3388729095458984, "step": 41358 }, { "epoch": 3.26046511627907, "grad_norm": 0.5316867828369141, "learning_rate": 0.0005, "loss": 1.333269476890564, "step": 41359 }, { "epoch": 3.2605439495467086, "grad_norm": 0.461051881313324, "learning_rate": 0.0005, "loss": 1.2124273777008057, "step": 41360 }, { "epoch": 3.2606227828143477, "grad_norm": 0.44837337732315063, "learning_rate": 0.0005, "loss": 1.0711712837219238, "step": 41361 }, { "epoch": 3.2607016160819864, "grad_norm": 0.4921617805957794, "learning_rate": 0.0005, "loss": 1.2594858407974243, "step": 41362 }, { "epoch": 3.2607804493496255, "grad_norm": 0.5828564167022705, "learning_rate": 0.0005, "loss": 1.433159351348877, "step": 41363 }, { "epoch": 3.2608592826172647, "grad_norm": 0.5031640529632568, "learning_rate": 0.0005, "loss": 1.3974276781082153, "step": 41364 }, { "epoch": 3.2609381158849033, "grad_norm": 0.5430380702018738, "learning_rate": 0.0005, "loss": 1.4100565910339355, "step": 41365 }, { "epoch": 3.2610169491525425, "grad_norm": 0.432608038187027, "learning_rate": 0.0005, "loss": 1.2573246955871582, "step": 41366 }, { "epoch": 3.261095782420181, "grad_norm": 0.47992855310440063, "learning_rate": 0.0005, "loss": 1.2059847116470337, "step": 41367 }, { "epoch": 3.2611746156878203, "grad_norm": 0.46168604493141174, "learning_rate": 0.0005, "loss": 1.095402717590332, "step": 41368 }, { "epoch": 3.2612534489554594, "grad_norm": 0.4740791618824005, "learning_rate": 0.0005, "loss": 1.4095957279205322, "step": 41369 }, { "epoch": 3.261332282223098, "grad_norm": 0.49668896198272705, "learning_rate": 0.0005, "loss": 1.179880142211914, "step": 41370 }, { "epoch": 3.2614111154907373, "grad_norm": 0.4874369502067566, "learning_rate": 0.0005, "loss": 1.3763747215270996, "step": 41371 }, { "epoch": 3.261489948758376, "grad_norm": 0.42809128761291504, "learning_rate": 0.0005, "loss": 1.2157039642333984, "step": 41372 }, { "epoch": 3.261568782026015, "grad_norm": 0.5379252433776855, "learning_rate": 0.0005, "loss": 1.333910346031189, "step": 41373 }, { "epoch": 3.2616476152936538, "grad_norm": 0.459793359041214, "learning_rate": 0.0005, "loss": 1.3873755931854248, "step": 41374 }, { "epoch": 3.261726448561293, "grad_norm": 0.48097455501556396, "learning_rate": 0.0005, "loss": 1.339951992034912, "step": 41375 }, { "epoch": 3.2618052818289316, "grad_norm": 0.5321443676948547, "learning_rate": 0.0005, "loss": 1.2078945636749268, "step": 41376 }, { "epoch": 3.2618841150965707, "grad_norm": 0.4784199297428131, "learning_rate": 0.0005, "loss": 1.3291782140731812, "step": 41377 }, { "epoch": 3.26196294836421, "grad_norm": 0.5057310461997986, "learning_rate": 0.0005, "loss": 1.237588882446289, "step": 41378 }, { "epoch": 3.2620417816318485, "grad_norm": 0.4495086669921875, "learning_rate": 0.0005, "loss": 1.190819501876831, "step": 41379 }, { "epoch": 3.2621206148994877, "grad_norm": 0.5292999148368835, "learning_rate": 0.0005, "loss": 1.2384130954742432, "step": 41380 }, { "epoch": 3.2621994481671264, "grad_norm": 0.46079325675964355, "learning_rate": 0.0005, "loss": 1.3634967803955078, "step": 41381 }, { "epoch": 3.2622782814347655, "grad_norm": 0.48320072889328003, "learning_rate": 0.0005, "loss": 1.226747751235962, "step": 41382 }, { "epoch": 3.2623571147024046, "grad_norm": 0.4649425446987152, "learning_rate": 0.0005, "loss": 1.1415562629699707, "step": 41383 }, { "epoch": 3.2624359479700433, "grad_norm": 0.5183498859405518, "learning_rate": 0.0005, "loss": 1.327425479888916, "step": 41384 }, { "epoch": 3.2625147812376825, "grad_norm": 0.5578646063804626, "learning_rate": 0.0005, "loss": 1.1875967979431152, "step": 41385 }, { "epoch": 3.262593614505321, "grad_norm": 0.5034525394439697, "learning_rate": 0.0005, "loss": 1.2961955070495605, "step": 41386 }, { "epoch": 3.2626724477729603, "grad_norm": 0.445152223110199, "learning_rate": 0.0005, "loss": 1.2018145322799683, "step": 41387 }, { "epoch": 3.262751281040599, "grad_norm": 0.41555294394493103, "learning_rate": 0.0005, "loss": 1.1772692203521729, "step": 41388 }, { "epoch": 3.262830114308238, "grad_norm": 0.5107911229133606, "learning_rate": 0.0005, "loss": 1.2575080394744873, "step": 41389 }, { "epoch": 3.262908947575877, "grad_norm": 0.6660021543502808, "learning_rate": 0.0005, "loss": 1.344473123550415, "step": 41390 }, { "epoch": 3.262987780843516, "grad_norm": 1.0710415840148926, "learning_rate": 0.0005, "loss": 1.2402606010437012, "step": 41391 }, { "epoch": 3.263066614111155, "grad_norm": 0.4959636628627777, "learning_rate": 0.0005, "loss": 1.4231480360031128, "step": 41392 }, { "epoch": 3.2631454473787938, "grad_norm": 0.5071349740028381, "learning_rate": 0.0005, "loss": 1.2666875123977661, "step": 41393 }, { "epoch": 3.263224280646433, "grad_norm": 0.7209861874580383, "learning_rate": 0.0005, "loss": 1.3243756294250488, "step": 41394 }, { "epoch": 3.2633031139140716, "grad_norm": 0.6022376418113708, "learning_rate": 0.0005, "loss": 1.2787026166915894, "step": 41395 }, { "epoch": 3.2633819471817107, "grad_norm": 0.5581433773040771, "learning_rate": 0.0005, "loss": 1.2722954750061035, "step": 41396 }, { "epoch": 3.26346078044935, "grad_norm": 0.5545929074287415, "learning_rate": 0.0005, "loss": 1.2616522312164307, "step": 41397 }, { "epoch": 3.2635396137169885, "grad_norm": 0.4698554277420044, "learning_rate": 0.0005, "loss": 1.1190335750579834, "step": 41398 }, { "epoch": 3.2636184469846277, "grad_norm": 0.5023074746131897, "learning_rate": 0.0005, "loss": 1.313902735710144, "step": 41399 }, { "epoch": 3.2636972802522664, "grad_norm": 0.45766863226890564, "learning_rate": 0.0005, "loss": 1.3223531246185303, "step": 41400 }, { "epoch": 3.2637761135199055, "grad_norm": 0.43175673484802246, "learning_rate": 0.0005, "loss": 1.1074143648147583, "step": 41401 }, { "epoch": 3.263854946787544, "grad_norm": 0.6505957841873169, "learning_rate": 0.0005, "loss": 1.3657257556915283, "step": 41402 }, { "epoch": 3.2639337800551833, "grad_norm": 0.7760997414588928, "learning_rate": 0.0005, "loss": 1.1922268867492676, "step": 41403 }, { "epoch": 3.264012613322822, "grad_norm": 0.3963198661804199, "learning_rate": 0.0005, "loss": 1.0903786420822144, "step": 41404 }, { "epoch": 3.264091446590461, "grad_norm": 0.4586145281791687, "learning_rate": 0.0005, "loss": 1.2302641868591309, "step": 41405 }, { "epoch": 3.2641702798581003, "grad_norm": 0.46467044949531555, "learning_rate": 0.0005, "loss": 1.1068181991577148, "step": 41406 }, { "epoch": 3.264249113125739, "grad_norm": 0.4191123843193054, "learning_rate": 0.0005, "loss": 1.07279372215271, "step": 41407 }, { "epoch": 3.264327946393378, "grad_norm": 0.5070210099220276, "learning_rate": 0.0005, "loss": 1.4625368118286133, "step": 41408 }, { "epoch": 3.2644067796610168, "grad_norm": 0.5095192193984985, "learning_rate": 0.0005, "loss": 1.1283636093139648, "step": 41409 }, { "epoch": 3.264485612928656, "grad_norm": 0.5338627099990845, "learning_rate": 0.0005, "loss": 1.1934576034545898, "step": 41410 }, { "epoch": 3.264564446196295, "grad_norm": 0.4912402033805847, "learning_rate": 0.0005, "loss": 1.115457534790039, "step": 41411 }, { "epoch": 3.2646432794639337, "grad_norm": 0.5623398423194885, "learning_rate": 0.0005, "loss": 1.2347748279571533, "step": 41412 }, { "epoch": 3.264722112731573, "grad_norm": 0.4259934723377228, "learning_rate": 0.0005, "loss": 1.083320140838623, "step": 41413 }, { "epoch": 3.2648009459992116, "grad_norm": 0.4275723993778229, "learning_rate": 0.0005, "loss": 1.0853232145309448, "step": 41414 }, { "epoch": 3.2648797792668507, "grad_norm": 0.4514773488044739, "learning_rate": 0.0005, "loss": 1.2079321146011353, "step": 41415 }, { "epoch": 3.2649586125344894, "grad_norm": 0.5754466652870178, "learning_rate": 0.0005, "loss": 1.261570930480957, "step": 41416 }, { "epoch": 3.2650374458021285, "grad_norm": 0.527813732624054, "learning_rate": 0.0005, "loss": 1.2815569639205933, "step": 41417 }, { "epoch": 3.265116279069767, "grad_norm": 0.5818617939949036, "learning_rate": 0.0005, "loss": 1.3874235153198242, "step": 41418 }, { "epoch": 3.2651951123374063, "grad_norm": 0.599280059337616, "learning_rate": 0.0005, "loss": 1.341775894165039, "step": 41419 }, { "epoch": 3.2652739456050455, "grad_norm": 0.4822530746459961, "learning_rate": 0.0005, "loss": 1.2167024612426758, "step": 41420 }, { "epoch": 3.265352778872684, "grad_norm": 0.5255945324897766, "learning_rate": 0.0005, "loss": 1.2007131576538086, "step": 41421 }, { "epoch": 3.2654316121403233, "grad_norm": 0.48339200019836426, "learning_rate": 0.0005, "loss": 1.2770026922225952, "step": 41422 }, { "epoch": 3.265510445407962, "grad_norm": 0.5318437218666077, "learning_rate": 0.0005, "loss": 1.3817421197891235, "step": 41423 }, { "epoch": 3.265589278675601, "grad_norm": 0.46052929759025574, "learning_rate": 0.0005, "loss": 1.2249820232391357, "step": 41424 }, { "epoch": 3.2656681119432402, "grad_norm": 0.44303521513938904, "learning_rate": 0.0005, "loss": 1.2248154878616333, "step": 41425 }, { "epoch": 3.265746945210879, "grad_norm": 0.423152893781662, "learning_rate": 0.0005, "loss": 1.065161943435669, "step": 41426 }, { "epoch": 3.265825778478518, "grad_norm": 0.5141023397445679, "learning_rate": 0.0005, "loss": 1.360818862915039, "step": 41427 }, { "epoch": 3.2659046117461568, "grad_norm": 0.4302081763744354, "learning_rate": 0.0005, "loss": 1.3419647216796875, "step": 41428 }, { "epoch": 3.265983445013796, "grad_norm": 0.4627336263656616, "learning_rate": 0.0005, "loss": 1.2271158695220947, "step": 41429 }, { "epoch": 3.2660622782814346, "grad_norm": 0.5680274367332458, "learning_rate": 0.0005, "loss": 1.4341330528259277, "step": 41430 }, { "epoch": 3.2661411115490737, "grad_norm": 0.4574672281742096, "learning_rate": 0.0005, "loss": 1.2707056999206543, "step": 41431 }, { "epoch": 3.2662199448167124, "grad_norm": 0.5080463886260986, "learning_rate": 0.0005, "loss": 1.380725383758545, "step": 41432 }, { "epoch": 3.2662987780843515, "grad_norm": 0.4418979287147522, "learning_rate": 0.0005, "loss": 1.2391116619110107, "step": 41433 }, { "epoch": 3.2663776113519907, "grad_norm": 0.4010918140411377, "learning_rate": 0.0005, "loss": 1.1109422445297241, "step": 41434 }, { "epoch": 3.2664564446196294, "grad_norm": 0.5356996059417725, "learning_rate": 0.0005, "loss": 1.1499615907669067, "step": 41435 }, { "epoch": 3.2665352778872685, "grad_norm": 0.40352964401245117, "learning_rate": 0.0005, "loss": 1.2226977348327637, "step": 41436 }, { "epoch": 3.266614111154907, "grad_norm": 0.8094921708106995, "learning_rate": 0.0005, "loss": 1.168229103088379, "step": 41437 }, { "epoch": 3.2666929444225463, "grad_norm": 0.5104655623435974, "learning_rate": 0.0005, "loss": 1.2916197776794434, "step": 41438 }, { "epoch": 3.2667717776901855, "grad_norm": 0.5418426394462585, "learning_rate": 0.0005, "loss": 1.198710560798645, "step": 41439 }, { "epoch": 3.266850610957824, "grad_norm": 0.5042429566383362, "learning_rate": 0.0005, "loss": 1.3515143394470215, "step": 41440 }, { "epoch": 3.2669294442254633, "grad_norm": 0.5421886444091797, "learning_rate": 0.0005, "loss": 1.54097318649292, "step": 41441 }, { "epoch": 3.267008277493102, "grad_norm": 0.5233182907104492, "learning_rate": 0.0005, "loss": 1.1596046686172485, "step": 41442 }, { "epoch": 3.267087110760741, "grad_norm": 0.6582996249198914, "learning_rate": 0.0005, "loss": 1.1406205892562866, "step": 41443 }, { "epoch": 3.26716594402838, "grad_norm": 0.38909831643104553, "learning_rate": 0.0005, "loss": 1.1575231552124023, "step": 41444 }, { "epoch": 3.267244777296019, "grad_norm": 0.4437812268733978, "learning_rate": 0.0005, "loss": 1.2532951831817627, "step": 41445 }, { "epoch": 3.267323610563658, "grad_norm": 0.47057172656059265, "learning_rate": 0.0005, "loss": 1.2709448337554932, "step": 41446 }, { "epoch": 3.2674024438312967, "grad_norm": 0.4593382775783539, "learning_rate": 0.0005, "loss": 1.2309863567352295, "step": 41447 }, { "epoch": 3.267481277098936, "grad_norm": 0.5941646695137024, "learning_rate": 0.0005, "loss": 1.2873998880386353, "step": 41448 }, { "epoch": 3.2675601103665746, "grad_norm": 0.614964485168457, "learning_rate": 0.0005, "loss": 1.125678539276123, "step": 41449 }, { "epoch": 3.2676389436342137, "grad_norm": 0.5244522094726562, "learning_rate": 0.0005, "loss": 1.379681944847107, "step": 41450 }, { "epoch": 3.2677177769018524, "grad_norm": 0.40640777349472046, "learning_rate": 0.0005, "loss": 1.2389154434204102, "step": 41451 }, { "epoch": 3.2677966101694915, "grad_norm": 0.49756425619125366, "learning_rate": 0.0005, "loss": 1.3653337955474854, "step": 41452 }, { "epoch": 3.2678754434371307, "grad_norm": 0.4583541452884674, "learning_rate": 0.0005, "loss": 1.3052362203598022, "step": 41453 }, { "epoch": 3.2679542767047693, "grad_norm": 0.6027761101722717, "learning_rate": 0.0005, "loss": 1.29683518409729, "step": 41454 }, { "epoch": 3.2680331099724085, "grad_norm": 0.49362361431121826, "learning_rate": 0.0005, "loss": 1.2621805667877197, "step": 41455 }, { "epoch": 3.268111943240047, "grad_norm": 0.4237828254699707, "learning_rate": 0.0005, "loss": 1.277545690536499, "step": 41456 }, { "epoch": 3.2681907765076863, "grad_norm": 0.3695506751537323, "learning_rate": 0.0005, "loss": 0.9655591249465942, "step": 41457 }, { "epoch": 3.2682696097753254, "grad_norm": 0.4807402193546295, "learning_rate": 0.0005, "loss": 1.2804608345031738, "step": 41458 }, { "epoch": 3.268348443042964, "grad_norm": 0.6076592803001404, "learning_rate": 0.0005, "loss": 1.209150791168213, "step": 41459 }, { "epoch": 3.2684272763106033, "grad_norm": 0.4491274654865265, "learning_rate": 0.0005, "loss": 1.1135797500610352, "step": 41460 }, { "epoch": 3.268506109578242, "grad_norm": 0.4768807291984558, "learning_rate": 0.0005, "loss": 1.2510100603103638, "step": 41461 }, { "epoch": 3.268584942845881, "grad_norm": 0.4696860611438751, "learning_rate": 0.0005, "loss": 1.2209559679031372, "step": 41462 }, { "epoch": 3.2686637761135198, "grad_norm": 0.4296741783618927, "learning_rate": 0.0005, "loss": 1.0438203811645508, "step": 41463 }, { "epoch": 3.268742609381159, "grad_norm": 0.5375626683235168, "learning_rate": 0.0005, "loss": 1.5193616151809692, "step": 41464 }, { "epoch": 3.2688214426487976, "grad_norm": 0.45668235421180725, "learning_rate": 0.0005, "loss": 1.2106070518493652, "step": 41465 }, { "epoch": 3.2689002759164367, "grad_norm": 0.47123754024505615, "learning_rate": 0.0005, "loss": 1.1095707416534424, "step": 41466 }, { "epoch": 3.268979109184076, "grad_norm": 0.5568802952766418, "learning_rate": 0.0005, "loss": 1.3780388832092285, "step": 41467 }, { "epoch": 3.2690579424517145, "grad_norm": 0.6232233047485352, "learning_rate": 0.0005, "loss": 1.249795913696289, "step": 41468 }, { "epoch": 3.2691367757193537, "grad_norm": 0.45853355526924133, "learning_rate": 0.0005, "loss": 1.1093202829360962, "step": 41469 }, { "epoch": 3.2692156089869924, "grad_norm": 0.6081904768943787, "learning_rate": 0.0005, "loss": 1.5187079906463623, "step": 41470 }, { "epoch": 3.2692944422546315, "grad_norm": 0.5274895429611206, "learning_rate": 0.0005, "loss": 1.1177070140838623, "step": 41471 }, { "epoch": 3.2693732755222706, "grad_norm": 0.6067177057266235, "learning_rate": 0.0005, "loss": 1.5242094993591309, "step": 41472 }, { "epoch": 3.2694521087899093, "grad_norm": 0.6758847832679749, "learning_rate": 0.0005, "loss": 1.3830608129501343, "step": 41473 }, { "epoch": 3.2695309420575485, "grad_norm": 0.5443260073661804, "learning_rate": 0.0005, "loss": 1.186607003211975, "step": 41474 }, { "epoch": 3.269609775325187, "grad_norm": 0.5419091582298279, "learning_rate": 0.0005, "loss": 1.325563907623291, "step": 41475 }, { "epoch": 3.2696886085928263, "grad_norm": 0.6799964904785156, "learning_rate": 0.0005, "loss": 1.3414783477783203, "step": 41476 }, { "epoch": 3.269767441860465, "grad_norm": 0.549360454082489, "learning_rate": 0.0005, "loss": 1.4091098308563232, "step": 41477 }, { "epoch": 3.269846275128104, "grad_norm": 0.4749465584754944, "learning_rate": 0.0005, "loss": 1.1443145275115967, "step": 41478 }, { "epoch": 3.269925108395743, "grad_norm": 0.5056540369987488, "learning_rate": 0.0005, "loss": 1.41731595993042, "step": 41479 }, { "epoch": 3.270003941663382, "grad_norm": 0.5185035467147827, "learning_rate": 0.0005, "loss": 1.2487962245941162, "step": 41480 }, { "epoch": 3.270082774931021, "grad_norm": 0.38727644085884094, "learning_rate": 0.0005, "loss": 1.0923951864242554, "step": 41481 }, { "epoch": 3.2701616081986598, "grad_norm": 0.43712836503982544, "learning_rate": 0.0005, "loss": 1.1269164085388184, "step": 41482 }, { "epoch": 3.270240441466299, "grad_norm": 0.5434361100196838, "learning_rate": 0.0005, "loss": 1.2879135608673096, "step": 41483 }, { "epoch": 3.2703192747339376, "grad_norm": 0.5933526158332825, "learning_rate": 0.0005, "loss": 1.2815607786178589, "step": 41484 }, { "epoch": 3.2703981080015767, "grad_norm": 0.5027111768722534, "learning_rate": 0.0005, "loss": 1.3358917236328125, "step": 41485 }, { "epoch": 3.270476941269216, "grad_norm": 0.5328910946846008, "learning_rate": 0.0005, "loss": 1.0970709323883057, "step": 41486 }, { "epoch": 3.2705557745368545, "grad_norm": 0.4898638129234314, "learning_rate": 0.0005, "loss": 1.146471619606018, "step": 41487 }, { "epoch": 3.2706346078044937, "grad_norm": 0.5551988482475281, "learning_rate": 0.0005, "loss": 1.214486002922058, "step": 41488 }, { "epoch": 3.2707134410721324, "grad_norm": 0.4247148036956787, "learning_rate": 0.0005, "loss": 1.1717854738235474, "step": 41489 }, { "epoch": 3.2707922743397715, "grad_norm": 0.5281710624694824, "learning_rate": 0.0005, "loss": 1.321343183517456, "step": 41490 }, { "epoch": 3.27087110760741, "grad_norm": 0.5850325226783752, "learning_rate": 0.0005, "loss": 1.465458869934082, "step": 41491 }, { "epoch": 3.2709499408750493, "grad_norm": 0.4897136092185974, "learning_rate": 0.0005, "loss": 1.3213889598846436, "step": 41492 }, { "epoch": 3.271028774142688, "grad_norm": 0.757268488407135, "learning_rate": 0.0005, "loss": 1.2205145359039307, "step": 41493 }, { "epoch": 3.271107607410327, "grad_norm": 0.5190909504890442, "learning_rate": 0.0005, "loss": 1.5134427547454834, "step": 41494 }, { "epoch": 3.2711864406779663, "grad_norm": 0.5148150324821472, "learning_rate": 0.0005, "loss": 1.2415951490402222, "step": 41495 }, { "epoch": 3.271265273945605, "grad_norm": 0.5625377893447876, "learning_rate": 0.0005, "loss": 1.3409903049468994, "step": 41496 }, { "epoch": 3.271344107213244, "grad_norm": 0.430771142244339, "learning_rate": 0.0005, "loss": 1.2739455699920654, "step": 41497 }, { "epoch": 3.271422940480883, "grad_norm": 0.538405179977417, "learning_rate": 0.0005, "loss": 1.340102195739746, "step": 41498 }, { "epoch": 3.271501773748522, "grad_norm": 0.4180041253566742, "learning_rate": 0.0005, "loss": 1.0643088817596436, "step": 41499 }, { "epoch": 3.271580607016161, "grad_norm": 0.44429442286491394, "learning_rate": 0.0005, "loss": 1.2814372777938843, "step": 41500 }, { "epoch": 3.2716594402837997, "grad_norm": 0.39194387197494507, "learning_rate": 0.0005, "loss": 1.040155291557312, "step": 41501 }, { "epoch": 3.271738273551439, "grad_norm": 0.552337110042572, "learning_rate": 0.0005, "loss": 1.2316570281982422, "step": 41502 }, { "epoch": 3.2718171068190776, "grad_norm": 0.41912031173706055, "learning_rate": 0.0005, "loss": 1.1383616924285889, "step": 41503 }, { "epoch": 3.2718959400867167, "grad_norm": 0.7153028845787048, "learning_rate": 0.0005, "loss": 1.391257643699646, "step": 41504 }, { "epoch": 3.2719747733543554, "grad_norm": 0.5664421916007996, "learning_rate": 0.0005, "loss": 1.4065442085266113, "step": 41505 }, { "epoch": 3.2720536066219945, "grad_norm": 0.4463678300380707, "learning_rate": 0.0005, "loss": 1.1212029457092285, "step": 41506 }, { "epoch": 3.272132439889633, "grad_norm": 0.41719383001327515, "learning_rate": 0.0005, "loss": 1.1774570941925049, "step": 41507 }, { "epoch": 3.2722112731572723, "grad_norm": 0.5268654227256775, "learning_rate": 0.0005, "loss": 1.2563269138336182, "step": 41508 }, { "epoch": 3.2722901064249115, "grad_norm": 0.4004519283771515, "learning_rate": 0.0005, "loss": 1.2169289588928223, "step": 41509 }, { "epoch": 3.27236893969255, "grad_norm": 0.4885542690753937, "learning_rate": 0.0005, "loss": 1.2719377279281616, "step": 41510 }, { "epoch": 3.2724477729601893, "grad_norm": 0.5149680972099304, "learning_rate": 0.0005, "loss": 1.348384141921997, "step": 41511 }, { "epoch": 3.272526606227828, "grad_norm": 0.5200551748275757, "learning_rate": 0.0005, "loss": 1.3158271312713623, "step": 41512 }, { "epoch": 3.272605439495467, "grad_norm": 0.49897754192352295, "learning_rate": 0.0005, "loss": 1.474135160446167, "step": 41513 }, { "epoch": 3.2726842727631063, "grad_norm": 0.6431525945663452, "learning_rate": 0.0005, "loss": 1.359282374382019, "step": 41514 }, { "epoch": 3.272763106030745, "grad_norm": 0.4130961000919342, "learning_rate": 0.0005, "loss": 1.1804587841033936, "step": 41515 }, { "epoch": 3.272841939298384, "grad_norm": 0.5875210165977478, "learning_rate": 0.0005, "loss": 1.3049492835998535, "step": 41516 }, { "epoch": 3.2729207725660228, "grad_norm": 0.5165806412696838, "learning_rate": 0.0005, "loss": 1.4030756950378418, "step": 41517 }, { "epoch": 3.272999605833662, "grad_norm": 0.4796273112297058, "learning_rate": 0.0005, "loss": 1.4197449684143066, "step": 41518 }, { "epoch": 3.2730784391013006, "grad_norm": 0.4251416027545929, "learning_rate": 0.0005, "loss": 1.16522216796875, "step": 41519 }, { "epoch": 3.2731572723689397, "grad_norm": 0.4741961658000946, "learning_rate": 0.0005, "loss": 1.2896409034729004, "step": 41520 }, { "epoch": 3.2732361056365784, "grad_norm": 0.6822875142097473, "learning_rate": 0.0005, "loss": 1.3686524629592896, "step": 41521 }, { "epoch": 3.2733149389042175, "grad_norm": 0.48473218083381653, "learning_rate": 0.0005, "loss": 1.2264549732208252, "step": 41522 }, { "epoch": 3.2733937721718567, "grad_norm": 1.192671298980713, "learning_rate": 0.0005, "loss": 1.2224984169006348, "step": 41523 }, { "epoch": 3.2734726054394954, "grad_norm": 0.565972089767456, "learning_rate": 0.0005, "loss": 1.3040547370910645, "step": 41524 }, { "epoch": 3.2735514387071345, "grad_norm": 0.49907898902893066, "learning_rate": 0.0005, "loss": 1.3641406297683716, "step": 41525 }, { "epoch": 3.273630271974773, "grad_norm": 0.5038214325904846, "learning_rate": 0.0005, "loss": 1.2664453983306885, "step": 41526 }, { "epoch": 3.2737091052424123, "grad_norm": 0.5707038640975952, "learning_rate": 0.0005, "loss": 1.5060670375823975, "step": 41527 }, { "epoch": 3.2737879385100515, "grad_norm": 0.5284197330474854, "learning_rate": 0.0005, "loss": 1.2741608619689941, "step": 41528 }, { "epoch": 3.27386677177769, "grad_norm": 0.6128733158111572, "learning_rate": 0.0005, "loss": 1.335343599319458, "step": 41529 }, { "epoch": 3.2739456050453293, "grad_norm": 0.49640142917633057, "learning_rate": 0.0005, "loss": 1.2125717401504517, "step": 41530 }, { "epoch": 3.274024438312968, "grad_norm": 0.4391765296459198, "learning_rate": 0.0005, "loss": 1.1480541229248047, "step": 41531 }, { "epoch": 3.274103271580607, "grad_norm": 0.5452277064323425, "learning_rate": 0.0005, "loss": 1.3226218223571777, "step": 41532 }, { "epoch": 3.274182104848246, "grad_norm": 0.43211910128593445, "learning_rate": 0.0005, "loss": 1.1919300556182861, "step": 41533 }, { "epoch": 3.274260938115885, "grad_norm": 0.6702319383621216, "learning_rate": 0.0005, "loss": 1.3968645334243774, "step": 41534 }, { "epoch": 3.2743397713835236, "grad_norm": 0.45988282561302185, "learning_rate": 0.0005, "loss": 1.1489646434783936, "step": 41535 }, { "epoch": 3.2744186046511627, "grad_norm": 0.3878730833530426, "learning_rate": 0.0005, "loss": 1.1407068967819214, "step": 41536 }, { "epoch": 3.274497437918802, "grad_norm": 0.8272055983543396, "learning_rate": 0.0005, "loss": 1.6088670492172241, "step": 41537 }, { "epoch": 3.2745762711864406, "grad_norm": 0.5000643134117126, "learning_rate": 0.0005, "loss": 1.4099972248077393, "step": 41538 }, { "epoch": 3.2746551044540797, "grad_norm": 0.45004209876060486, "learning_rate": 0.0005, "loss": 1.1465227603912354, "step": 41539 }, { "epoch": 3.2747339377217184, "grad_norm": 0.6150340437889099, "learning_rate": 0.0005, "loss": 1.2217926979064941, "step": 41540 }, { "epoch": 3.2748127709893575, "grad_norm": 0.514006495475769, "learning_rate": 0.0005, "loss": 1.2476856708526611, "step": 41541 }, { "epoch": 3.2748916042569967, "grad_norm": 0.5150729417800903, "learning_rate": 0.0005, "loss": 1.1898753643035889, "step": 41542 }, { "epoch": 3.2749704375246353, "grad_norm": 0.5968907475471497, "learning_rate": 0.0005, "loss": 1.2420907020568848, "step": 41543 }, { "epoch": 3.2750492707922745, "grad_norm": 0.49032577872276306, "learning_rate": 0.0005, "loss": 1.3587687015533447, "step": 41544 }, { "epoch": 3.275128104059913, "grad_norm": 0.559961199760437, "learning_rate": 0.0005, "loss": 1.3318450450897217, "step": 41545 }, { "epoch": 3.2752069373275523, "grad_norm": 0.4365476369857788, "learning_rate": 0.0005, "loss": 1.1314842700958252, "step": 41546 }, { "epoch": 3.275285770595191, "grad_norm": 0.4402199387550354, "learning_rate": 0.0005, "loss": 1.0882863998413086, "step": 41547 }, { "epoch": 3.27536460386283, "grad_norm": 0.46706244349479675, "learning_rate": 0.0005, "loss": 1.1941630840301514, "step": 41548 }, { "epoch": 3.275443437130469, "grad_norm": 0.4225579500198364, "learning_rate": 0.0005, "loss": 1.2209522724151611, "step": 41549 }, { "epoch": 3.275522270398108, "grad_norm": 0.4450872540473938, "learning_rate": 0.0005, "loss": 1.2679338455200195, "step": 41550 }, { "epoch": 3.275601103665747, "grad_norm": 0.4647994637489319, "learning_rate": 0.0005, "loss": 1.1795552968978882, "step": 41551 }, { "epoch": 3.2756799369333858, "grad_norm": 0.5162982940673828, "learning_rate": 0.0005, "loss": 1.219031572341919, "step": 41552 }, { "epoch": 3.275758770201025, "grad_norm": 0.449991911649704, "learning_rate": 0.0005, "loss": 1.2737274169921875, "step": 41553 }, { "epoch": 3.2758376034686636, "grad_norm": 0.4637773633003235, "learning_rate": 0.0005, "loss": 1.2206623554229736, "step": 41554 }, { "epoch": 3.2759164367363027, "grad_norm": 0.5152782797813416, "learning_rate": 0.0005, "loss": 1.3357369899749756, "step": 41555 }, { "epoch": 3.275995270003942, "grad_norm": 0.42917299270629883, "learning_rate": 0.0005, "loss": 1.1995950937271118, "step": 41556 }, { "epoch": 3.2760741032715806, "grad_norm": 0.5048410296440125, "learning_rate": 0.0005, "loss": 1.3663935661315918, "step": 41557 }, { "epoch": 3.2761529365392197, "grad_norm": 0.5770251750946045, "learning_rate": 0.0005, "loss": 1.3046419620513916, "step": 41558 }, { "epoch": 3.2762317698068584, "grad_norm": 0.5123456120491028, "learning_rate": 0.0005, "loss": 1.2593210935592651, "step": 41559 }, { "epoch": 3.2763106030744975, "grad_norm": 0.40642625093460083, "learning_rate": 0.0005, "loss": 1.0898674726486206, "step": 41560 }, { "epoch": 3.276389436342136, "grad_norm": 0.5566287040710449, "learning_rate": 0.0005, "loss": 1.407031536102295, "step": 41561 }, { "epoch": 3.2764682696097753, "grad_norm": 0.48389872908592224, "learning_rate": 0.0005, "loss": 1.2121567726135254, "step": 41562 }, { "epoch": 3.276547102877414, "grad_norm": 0.5187577605247498, "learning_rate": 0.0005, "loss": 1.4361932277679443, "step": 41563 }, { "epoch": 3.276625936145053, "grad_norm": 0.5175681710243225, "learning_rate": 0.0005, "loss": 1.2435932159423828, "step": 41564 }, { "epoch": 3.2767047694126923, "grad_norm": 0.43748798966407776, "learning_rate": 0.0005, "loss": 1.249077558517456, "step": 41565 }, { "epoch": 3.276783602680331, "grad_norm": 0.43768835067749023, "learning_rate": 0.0005, "loss": 1.230251669883728, "step": 41566 }, { "epoch": 3.27686243594797, "grad_norm": 0.5572852492332458, "learning_rate": 0.0005, "loss": 1.4664428234100342, "step": 41567 }, { "epoch": 3.276941269215609, "grad_norm": 0.5925858616828918, "learning_rate": 0.0005, "loss": 1.2858775854110718, "step": 41568 }, { "epoch": 3.277020102483248, "grad_norm": 0.5109912753105164, "learning_rate": 0.0005, "loss": 1.1706838607788086, "step": 41569 }, { "epoch": 3.277098935750887, "grad_norm": 0.6381964087486267, "learning_rate": 0.0005, "loss": 1.3136295080184937, "step": 41570 }, { "epoch": 3.2771777690185258, "grad_norm": 0.4665488302707672, "learning_rate": 0.0005, "loss": 1.1995137929916382, "step": 41571 }, { "epoch": 3.277256602286165, "grad_norm": 0.5033184289932251, "learning_rate": 0.0005, "loss": 1.2454907894134521, "step": 41572 }, { "epoch": 3.2773354355538036, "grad_norm": 0.6286138892173767, "learning_rate": 0.0005, "loss": 1.4783717393875122, "step": 41573 }, { "epoch": 3.2774142688214427, "grad_norm": 0.5466206669807434, "learning_rate": 0.0005, "loss": 1.3934754133224487, "step": 41574 }, { "epoch": 3.2774931020890814, "grad_norm": 0.5286248326301575, "learning_rate": 0.0005, "loss": 1.3855063915252686, "step": 41575 }, { "epoch": 3.2775719353567205, "grad_norm": 0.6171691417694092, "learning_rate": 0.0005, "loss": 1.3716728687286377, "step": 41576 }, { "epoch": 3.2776507686243597, "grad_norm": 0.4394349753856659, "learning_rate": 0.0005, "loss": 1.1998567581176758, "step": 41577 }, { "epoch": 3.2777296018919984, "grad_norm": 0.5349592566490173, "learning_rate": 0.0005, "loss": 1.3446637392044067, "step": 41578 }, { "epoch": 3.2778084351596375, "grad_norm": 0.49064311385154724, "learning_rate": 0.0005, "loss": 1.2938337326049805, "step": 41579 }, { "epoch": 3.277887268427276, "grad_norm": 0.5240383148193359, "learning_rate": 0.0005, "loss": 1.2741554975509644, "step": 41580 }, { "epoch": 3.2779661016949153, "grad_norm": 0.4670637845993042, "learning_rate": 0.0005, "loss": 1.2988476753234863, "step": 41581 }, { "epoch": 3.278044934962554, "grad_norm": 0.5135108232498169, "learning_rate": 0.0005, "loss": 1.2541279792785645, "step": 41582 }, { "epoch": 3.278123768230193, "grad_norm": 0.5240087509155273, "learning_rate": 0.0005, "loss": 1.1863199472427368, "step": 41583 }, { "epoch": 3.2782026014978323, "grad_norm": 0.47000983357429504, "learning_rate": 0.0005, "loss": 1.1841636896133423, "step": 41584 }, { "epoch": 3.278281434765471, "grad_norm": 0.4769038259983063, "learning_rate": 0.0005, "loss": 1.1852853298187256, "step": 41585 }, { "epoch": 3.27836026803311, "grad_norm": 0.4735923409461975, "learning_rate": 0.0005, "loss": 1.4057245254516602, "step": 41586 }, { "epoch": 3.278439101300749, "grad_norm": 0.5419003367424011, "learning_rate": 0.0005, "loss": 0.9663923978805542, "step": 41587 }, { "epoch": 3.278517934568388, "grad_norm": 0.5103229880332947, "learning_rate": 0.0005, "loss": 1.2054708003997803, "step": 41588 }, { "epoch": 3.278596767836027, "grad_norm": 0.5411425232887268, "learning_rate": 0.0005, "loss": 1.4262157678604126, "step": 41589 }, { "epoch": 3.2786756011036657, "grad_norm": 0.5215681791305542, "learning_rate": 0.0005, "loss": 1.243841290473938, "step": 41590 }, { "epoch": 3.278754434371305, "grad_norm": 0.39757251739501953, "learning_rate": 0.0005, "loss": 1.179768443107605, "step": 41591 }, { "epoch": 3.2788332676389436, "grad_norm": 0.5532304048538208, "learning_rate": 0.0005, "loss": 1.1279720067977905, "step": 41592 }, { "epoch": 3.2789121009065827, "grad_norm": 0.58103346824646, "learning_rate": 0.0005, "loss": 1.2230334281921387, "step": 41593 }, { "epoch": 3.2789909341742214, "grad_norm": 0.47104620933532715, "learning_rate": 0.0005, "loss": 1.2336053848266602, "step": 41594 }, { "epoch": 3.2790697674418605, "grad_norm": 0.4817483127117157, "learning_rate": 0.0005, "loss": 1.2395589351654053, "step": 41595 }, { "epoch": 3.279148600709499, "grad_norm": 0.44325071573257446, "learning_rate": 0.0005, "loss": 1.220460295677185, "step": 41596 }, { "epoch": 3.2792274339771383, "grad_norm": 0.5162243247032166, "learning_rate": 0.0005, "loss": 1.2270870208740234, "step": 41597 }, { "epoch": 3.2793062672447775, "grad_norm": 0.517780601978302, "learning_rate": 0.0005, "loss": 1.3033127784729004, "step": 41598 }, { "epoch": 3.279385100512416, "grad_norm": 0.45099520683288574, "learning_rate": 0.0005, "loss": 1.2285263538360596, "step": 41599 }, { "epoch": 3.2794639337800553, "grad_norm": 0.5868814587593079, "learning_rate": 0.0005, "loss": 1.1534019708633423, "step": 41600 }, { "epoch": 3.279542767047694, "grad_norm": 0.5457628965377808, "learning_rate": 0.0005, "loss": 1.244834065437317, "step": 41601 }, { "epoch": 3.279621600315333, "grad_norm": 0.46044477820396423, "learning_rate": 0.0005, "loss": 1.4170899391174316, "step": 41602 }, { "epoch": 3.2797004335829723, "grad_norm": 0.4276989996433258, "learning_rate": 0.0005, "loss": 0.9787282347679138, "step": 41603 }, { "epoch": 3.279779266850611, "grad_norm": 0.6460863947868347, "learning_rate": 0.0005, "loss": 1.4362438917160034, "step": 41604 }, { "epoch": 3.27985810011825, "grad_norm": 0.5889380574226379, "learning_rate": 0.0005, "loss": 1.4797292947769165, "step": 41605 }, { "epoch": 3.2799369333858888, "grad_norm": 0.4898006319999695, "learning_rate": 0.0005, "loss": 1.3372708559036255, "step": 41606 }, { "epoch": 3.280015766653528, "grad_norm": 0.5253031253814697, "learning_rate": 0.0005, "loss": 1.187548041343689, "step": 41607 }, { "epoch": 3.2800945999211666, "grad_norm": 0.46261683106422424, "learning_rate": 0.0005, "loss": 1.3911492824554443, "step": 41608 }, { "epoch": 3.2801734331888057, "grad_norm": 0.6111916303634644, "learning_rate": 0.0005, "loss": 1.3273835182189941, "step": 41609 }, { "epoch": 3.2802522664564444, "grad_norm": 0.43787071108818054, "learning_rate": 0.0005, "loss": 1.211978793144226, "step": 41610 }, { "epoch": 3.2803310997240835, "grad_norm": 0.44912847876548767, "learning_rate": 0.0005, "loss": 1.192911148071289, "step": 41611 }, { "epoch": 3.2804099329917227, "grad_norm": 0.4755060076713562, "learning_rate": 0.0005, "loss": 1.128144383430481, "step": 41612 }, { "epoch": 3.2804887662593614, "grad_norm": 0.46208739280700684, "learning_rate": 0.0005, "loss": 1.2612452507019043, "step": 41613 }, { "epoch": 3.2805675995270005, "grad_norm": 0.4822307229042053, "learning_rate": 0.0005, "loss": 1.2419098615646362, "step": 41614 }, { "epoch": 3.280646432794639, "grad_norm": 0.5150607824325562, "learning_rate": 0.0005, "loss": 1.29887056350708, "step": 41615 }, { "epoch": 3.2807252660622783, "grad_norm": 0.5648275017738342, "learning_rate": 0.0005, "loss": 1.3305193185806274, "step": 41616 }, { "epoch": 3.2808040993299175, "grad_norm": 0.5349311232566833, "learning_rate": 0.0005, "loss": 1.3438332080841064, "step": 41617 }, { "epoch": 3.280882932597556, "grad_norm": 0.9323990941047668, "learning_rate": 0.0005, "loss": 1.4233025312423706, "step": 41618 }, { "epoch": 3.2809617658651953, "grad_norm": 0.4914523959159851, "learning_rate": 0.0005, "loss": 1.218595266342163, "step": 41619 }, { "epoch": 3.281040599132834, "grad_norm": 0.5360047221183777, "learning_rate": 0.0005, "loss": 1.1371147632598877, "step": 41620 }, { "epoch": 3.281119432400473, "grad_norm": 0.44743674993515015, "learning_rate": 0.0005, "loss": 1.2655644416809082, "step": 41621 }, { "epoch": 3.281198265668112, "grad_norm": 0.48663321137428284, "learning_rate": 0.0005, "loss": 1.2571004629135132, "step": 41622 }, { "epoch": 3.281277098935751, "grad_norm": 0.5087583661079407, "learning_rate": 0.0005, "loss": 1.404484510421753, "step": 41623 }, { "epoch": 3.2813559322033896, "grad_norm": 0.5055968165397644, "learning_rate": 0.0005, "loss": 1.341991901397705, "step": 41624 }, { "epoch": 3.2814347654710287, "grad_norm": 0.4755132794380188, "learning_rate": 0.0005, "loss": 1.2024840116500854, "step": 41625 }, { "epoch": 3.281513598738668, "grad_norm": 0.47781625390052795, "learning_rate": 0.0005, "loss": 1.3213313817977905, "step": 41626 }, { "epoch": 3.2815924320063066, "grad_norm": 0.4870625138282776, "learning_rate": 0.0005, "loss": 1.3461768627166748, "step": 41627 }, { "epoch": 3.2816712652739457, "grad_norm": 0.5769914388656616, "learning_rate": 0.0005, "loss": 1.4266557693481445, "step": 41628 }, { "epoch": 3.2817500985415844, "grad_norm": 0.4373149573802948, "learning_rate": 0.0005, "loss": 1.327979564666748, "step": 41629 }, { "epoch": 3.2818289318092235, "grad_norm": 0.43304914236068726, "learning_rate": 0.0005, "loss": 1.332096815109253, "step": 41630 }, { "epoch": 3.2819077650768627, "grad_norm": 0.47067180275917053, "learning_rate": 0.0005, "loss": 1.2171826362609863, "step": 41631 }, { "epoch": 3.2819865983445013, "grad_norm": 0.5751140713691711, "learning_rate": 0.0005, "loss": 1.3632357120513916, "step": 41632 }, { "epoch": 3.2820654316121405, "grad_norm": 0.6249369382858276, "learning_rate": 0.0005, "loss": 1.2875607013702393, "step": 41633 }, { "epoch": 3.282144264879779, "grad_norm": 0.518699049949646, "learning_rate": 0.0005, "loss": 1.329986333847046, "step": 41634 }, { "epoch": 3.2822230981474183, "grad_norm": 0.5559068918228149, "learning_rate": 0.0005, "loss": 1.2654732465744019, "step": 41635 }, { "epoch": 3.282301931415057, "grad_norm": 0.43761196732521057, "learning_rate": 0.0005, "loss": 1.138207197189331, "step": 41636 }, { "epoch": 3.282380764682696, "grad_norm": 0.44343215227127075, "learning_rate": 0.0005, "loss": 1.3053046464920044, "step": 41637 }, { "epoch": 3.282459597950335, "grad_norm": 0.5579126477241516, "learning_rate": 0.0005, "loss": 1.2301976680755615, "step": 41638 }, { "epoch": 3.282538431217974, "grad_norm": 0.5226957201957703, "learning_rate": 0.0005, "loss": 1.327510118484497, "step": 41639 }, { "epoch": 3.282617264485613, "grad_norm": 0.8215910196304321, "learning_rate": 0.0005, "loss": 1.3761286735534668, "step": 41640 }, { "epoch": 3.2826960977532518, "grad_norm": 0.6208674311637878, "learning_rate": 0.0005, "loss": 1.230411171913147, "step": 41641 }, { "epoch": 3.282774931020891, "grad_norm": 0.5564749836921692, "learning_rate": 0.0005, "loss": 1.2063264846801758, "step": 41642 }, { "epoch": 3.2828537642885296, "grad_norm": 0.49458107352256775, "learning_rate": 0.0005, "loss": 1.0893763303756714, "step": 41643 }, { "epoch": 3.2829325975561687, "grad_norm": 0.6897500157356262, "learning_rate": 0.0005, "loss": 1.2753334045410156, "step": 41644 }, { "epoch": 3.283011430823808, "grad_norm": 0.4167044162750244, "learning_rate": 0.0005, "loss": 1.1757168769836426, "step": 41645 }, { "epoch": 3.2830902640914466, "grad_norm": 0.5720968246459961, "learning_rate": 0.0005, "loss": 1.3765873908996582, "step": 41646 }, { "epoch": 3.2831690973590857, "grad_norm": 0.45961523056030273, "learning_rate": 0.0005, "loss": 1.2483798265457153, "step": 41647 }, { "epoch": 3.2832479306267244, "grad_norm": 0.5760602951049805, "learning_rate": 0.0005, "loss": 1.1927611827850342, "step": 41648 }, { "epoch": 3.2833267638943635, "grad_norm": 0.5351519584655762, "learning_rate": 0.0005, "loss": 1.2823679447174072, "step": 41649 }, { "epoch": 3.283405597162002, "grad_norm": 0.4379490613937378, "learning_rate": 0.0005, "loss": 1.2038030624389648, "step": 41650 }, { "epoch": 3.2834844304296413, "grad_norm": 0.5623407363891602, "learning_rate": 0.0005, "loss": 1.2284492254257202, "step": 41651 }, { "epoch": 3.28356326369728, "grad_norm": 0.4261610209941864, "learning_rate": 0.0005, "loss": 1.1191024780273438, "step": 41652 }, { "epoch": 3.283642096964919, "grad_norm": 0.5112648606300354, "learning_rate": 0.0005, "loss": 1.3071913719177246, "step": 41653 }, { "epoch": 3.2837209302325583, "grad_norm": 0.5202212929725647, "learning_rate": 0.0005, "loss": 1.3259295225143433, "step": 41654 }, { "epoch": 3.283799763500197, "grad_norm": 0.6219930052757263, "learning_rate": 0.0005, "loss": 1.3502976894378662, "step": 41655 }, { "epoch": 3.283878596767836, "grad_norm": 0.4923948049545288, "learning_rate": 0.0005, "loss": 1.344630241394043, "step": 41656 }, { "epoch": 3.283957430035475, "grad_norm": 0.4454367160797119, "learning_rate": 0.0005, "loss": 1.1435520648956299, "step": 41657 }, { "epoch": 3.284036263303114, "grad_norm": 0.5033500790596008, "learning_rate": 0.0005, "loss": 1.431938886642456, "step": 41658 }, { "epoch": 3.284115096570753, "grad_norm": 0.518852949142456, "learning_rate": 0.0005, "loss": 1.3178420066833496, "step": 41659 }, { "epoch": 3.2841939298383918, "grad_norm": 0.5576997995376587, "learning_rate": 0.0005, "loss": 1.3803160190582275, "step": 41660 }, { "epoch": 3.284272763106031, "grad_norm": 0.516165554523468, "learning_rate": 0.0005, "loss": 1.3192493915557861, "step": 41661 }, { "epoch": 3.2843515963736696, "grad_norm": 0.7282276749610901, "learning_rate": 0.0005, "loss": 1.2064368724822998, "step": 41662 }, { "epoch": 3.2844304296413087, "grad_norm": 0.4331624507904053, "learning_rate": 0.0005, "loss": 1.332215666770935, "step": 41663 }, { "epoch": 3.2845092629089474, "grad_norm": 0.4069419205188751, "learning_rate": 0.0005, "loss": 1.1518065929412842, "step": 41664 }, { "epoch": 3.2845880961765865, "grad_norm": 0.47542646527290344, "learning_rate": 0.0005, "loss": 1.3082435131072998, "step": 41665 }, { "epoch": 3.2846669294442252, "grad_norm": 0.5019206404685974, "learning_rate": 0.0005, "loss": 1.2434505224227905, "step": 41666 }, { "epoch": 3.2847457627118644, "grad_norm": 0.45126843452453613, "learning_rate": 0.0005, "loss": 1.3597040176391602, "step": 41667 }, { "epoch": 3.2848245959795035, "grad_norm": 0.481452077627182, "learning_rate": 0.0005, "loss": 1.171805739402771, "step": 41668 }, { "epoch": 3.284903429247142, "grad_norm": 0.4370932877063751, "learning_rate": 0.0005, "loss": 1.1442079544067383, "step": 41669 }, { "epoch": 3.2849822625147813, "grad_norm": 0.41948407888412476, "learning_rate": 0.0005, "loss": 1.1332149505615234, "step": 41670 }, { "epoch": 3.28506109578242, "grad_norm": 0.5242394804954529, "learning_rate": 0.0005, "loss": 1.3901512622833252, "step": 41671 }, { "epoch": 3.285139929050059, "grad_norm": 0.540686309337616, "learning_rate": 0.0005, "loss": 1.2366981506347656, "step": 41672 }, { "epoch": 3.2852187623176983, "grad_norm": 0.45276308059692383, "learning_rate": 0.0005, "loss": 1.439781665802002, "step": 41673 }, { "epoch": 3.285297595585337, "grad_norm": 0.5800182819366455, "learning_rate": 0.0005, "loss": 1.244666576385498, "step": 41674 }, { "epoch": 3.285376428852976, "grad_norm": 0.4986141324043274, "learning_rate": 0.0005, "loss": 1.3340481519699097, "step": 41675 }, { "epoch": 3.285455262120615, "grad_norm": 0.5503829717636108, "learning_rate": 0.0005, "loss": 1.3019789457321167, "step": 41676 }, { "epoch": 3.285534095388254, "grad_norm": 0.4496835172176361, "learning_rate": 0.0005, "loss": 1.310183048248291, "step": 41677 }, { "epoch": 3.2856129286558926, "grad_norm": 0.45748475193977356, "learning_rate": 0.0005, "loss": 1.406698226928711, "step": 41678 }, { "epoch": 3.2856917619235317, "grad_norm": 0.4270932078361511, "learning_rate": 0.0005, "loss": 1.320944905281067, "step": 41679 }, { "epoch": 3.2857705951911704, "grad_norm": 0.5039002895355225, "learning_rate": 0.0005, "loss": 1.2998459339141846, "step": 41680 }, { "epoch": 3.2858494284588096, "grad_norm": 0.5226619839668274, "learning_rate": 0.0005, "loss": 1.2713990211486816, "step": 41681 }, { "epoch": 3.2859282617264487, "grad_norm": 0.6047051548957825, "learning_rate": 0.0005, "loss": 1.0862644910812378, "step": 41682 }, { "epoch": 3.2860070949940874, "grad_norm": 0.6015435457229614, "learning_rate": 0.0005, "loss": 1.2724990844726562, "step": 41683 }, { "epoch": 3.2860859282617265, "grad_norm": 0.7873519659042358, "learning_rate": 0.0005, "loss": 1.1443076133728027, "step": 41684 }, { "epoch": 3.286164761529365, "grad_norm": 0.42588359117507935, "learning_rate": 0.0005, "loss": 1.3347876071929932, "step": 41685 }, { "epoch": 3.2862435947970043, "grad_norm": 0.582648754119873, "learning_rate": 0.0005, "loss": 1.3394360542297363, "step": 41686 }, { "epoch": 3.2863224280646435, "grad_norm": 0.5697834491729736, "learning_rate": 0.0005, "loss": 1.5013909339904785, "step": 41687 }, { "epoch": 3.286401261332282, "grad_norm": 0.42087557911872864, "learning_rate": 0.0005, "loss": 1.3207951784133911, "step": 41688 }, { "epoch": 3.2864800945999213, "grad_norm": 0.43939679861068726, "learning_rate": 0.0005, "loss": 1.3563461303710938, "step": 41689 }, { "epoch": 3.28655892786756, "grad_norm": 0.43161332607269287, "learning_rate": 0.0005, "loss": 1.1647088527679443, "step": 41690 }, { "epoch": 3.286637761135199, "grad_norm": 0.5412992238998413, "learning_rate": 0.0005, "loss": 1.265758752822876, "step": 41691 }, { "epoch": 3.286716594402838, "grad_norm": 0.5231263041496277, "learning_rate": 0.0005, "loss": 1.2863388061523438, "step": 41692 }, { "epoch": 3.286795427670477, "grad_norm": 0.4689294099807739, "learning_rate": 0.0005, "loss": 1.1523123979568481, "step": 41693 }, { "epoch": 3.286874260938116, "grad_norm": 0.5169239044189453, "learning_rate": 0.0005, "loss": 1.219849705696106, "step": 41694 }, { "epoch": 3.2869530942057548, "grad_norm": 0.5665696263313293, "learning_rate": 0.0005, "loss": 1.1989710330963135, "step": 41695 }, { "epoch": 3.287031927473394, "grad_norm": 0.5132114291191101, "learning_rate": 0.0005, "loss": 1.2017982006072998, "step": 41696 }, { "epoch": 3.2871107607410326, "grad_norm": 0.5103265643119812, "learning_rate": 0.0005, "loss": 1.4413855075836182, "step": 41697 }, { "epoch": 3.2871895940086717, "grad_norm": 0.5854403972625732, "learning_rate": 0.0005, "loss": 1.2620165348052979, "step": 41698 }, { "epoch": 3.2872684272763104, "grad_norm": 0.46964406967163086, "learning_rate": 0.0005, "loss": 1.2302964925765991, "step": 41699 }, { "epoch": 3.2873472605439495, "grad_norm": 0.5489677786827087, "learning_rate": 0.0005, "loss": 1.2393250465393066, "step": 41700 }, { "epoch": 3.2874260938115887, "grad_norm": 0.5137243270874023, "learning_rate": 0.0005, "loss": 1.250685214996338, "step": 41701 }, { "epoch": 3.2875049270792274, "grad_norm": 0.4099727272987366, "learning_rate": 0.0005, "loss": 1.1250081062316895, "step": 41702 }, { "epoch": 3.2875837603468665, "grad_norm": 0.5306157469749451, "learning_rate": 0.0005, "loss": 1.1483094692230225, "step": 41703 }, { "epoch": 3.287662593614505, "grad_norm": 0.40551960468292236, "learning_rate": 0.0005, "loss": 1.1859443187713623, "step": 41704 }, { "epoch": 3.2877414268821443, "grad_norm": 0.46704182028770447, "learning_rate": 0.0005, "loss": 1.3075058460235596, "step": 41705 }, { "epoch": 3.2878202601497835, "grad_norm": 0.4880896210670471, "learning_rate": 0.0005, "loss": 1.3086600303649902, "step": 41706 }, { "epoch": 3.287899093417422, "grad_norm": 0.44244149327278137, "learning_rate": 0.0005, "loss": 1.238840937614441, "step": 41707 }, { "epoch": 3.2879779266850613, "grad_norm": 0.4221741855144501, "learning_rate": 0.0005, "loss": 1.1883639097213745, "step": 41708 }, { "epoch": 3.2880567599527, "grad_norm": 0.6176294088363647, "learning_rate": 0.0005, "loss": 1.2852363586425781, "step": 41709 }, { "epoch": 3.288135593220339, "grad_norm": 0.5180298686027527, "learning_rate": 0.0005, "loss": 1.3061555624008179, "step": 41710 }, { "epoch": 3.288214426487978, "grad_norm": 0.43173617124557495, "learning_rate": 0.0005, "loss": 1.234483003616333, "step": 41711 }, { "epoch": 3.288293259755617, "grad_norm": 0.5387718677520752, "learning_rate": 0.0005, "loss": 1.354083776473999, "step": 41712 }, { "epoch": 3.2883720930232556, "grad_norm": 0.5109738111495972, "learning_rate": 0.0005, "loss": 1.2153942584991455, "step": 41713 }, { "epoch": 3.2884509262908947, "grad_norm": 0.35202792286872864, "learning_rate": 0.0005, "loss": 1.1411056518554688, "step": 41714 }, { "epoch": 3.288529759558534, "grad_norm": 0.47397640347480774, "learning_rate": 0.0005, "loss": 1.2164252996444702, "step": 41715 }, { "epoch": 3.2886085928261726, "grad_norm": 0.5373311042785645, "learning_rate": 0.0005, "loss": 1.2522474527359009, "step": 41716 }, { "epoch": 3.2886874260938117, "grad_norm": 0.48222678899765015, "learning_rate": 0.0005, "loss": 1.2286255359649658, "step": 41717 }, { "epoch": 3.2887662593614504, "grad_norm": 0.515740156173706, "learning_rate": 0.0005, "loss": 1.2122883796691895, "step": 41718 }, { "epoch": 3.2888450926290895, "grad_norm": 0.5157344937324524, "learning_rate": 0.0005, "loss": 1.3720815181732178, "step": 41719 }, { "epoch": 3.2889239258967287, "grad_norm": 0.5308725833892822, "learning_rate": 0.0005, "loss": 1.3905234336853027, "step": 41720 }, { "epoch": 3.2890027591643674, "grad_norm": 0.5485396385192871, "learning_rate": 0.0005, "loss": 1.473556399345398, "step": 41721 }, { "epoch": 3.2890815924320065, "grad_norm": 0.4716918170452118, "learning_rate": 0.0005, "loss": 1.2937034368515015, "step": 41722 }, { "epoch": 3.289160425699645, "grad_norm": 0.4612436294555664, "learning_rate": 0.0005, "loss": 1.247507095336914, "step": 41723 }, { "epoch": 3.2892392589672843, "grad_norm": 0.4825727343559265, "learning_rate": 0.0005, "loss": 1.2311608791351318, "step": 41724 }, { "epoch": 3.289318092234923, "grad_norm": 0.3614436089992523, "learning_rate": 0.0005, "loss": 1.1212629079818726, "step": 41725 }, { "epoch": 3.289396925502562, "grad_norm": 0.43149152398109436, "learning_rate": 0.0005, "loss": 1.3214601278305054, "step": 41726 }, { "epoch": 3.289475758770201, "grad_norm": 0.5585096478462219, "learning_rate": 0.0005, "loss": 1.29047429561615, "step": 41727 }, { "epoch": 3.28955459203784, "grad_norm": 0.3973848819732666, "learning_rate": 0.0005, "loss": 1.2442303895950317, "step": 41728 }, { "epoch": 3.289633425305479, "grad_norm": 0.47654905915260315, "learning_rate": 0.0005, "loss": 1.1170587539672852, "step": 41729 }, { "epoch": 3.2897122585731178, "grad_norm": 0.5805005431175232, "learning_rate": 0.0005, "loss": 1.4567432403564453, "step": 41730 }, { "epoch": 3.289791091840757, "grad_norm": 0.40257784724235535, "learning_rate": 0.0005, "loss": 1.1405115127563477, "step": 41731 }, { "epoch": 3.2898699251083956, "grad_norm": 0.40588730573654175, "learning_rate": 0.0005, "loss": 1.0188932418823242, "step": 41732 }, { "epoch": 3.2899487583760347, "grad_norm": 0.39721542596817017, "learning_rate": 0.0005, "loss": 1.0029059648513794, "step": 41733 }, { "epoch": 3.290027591643674, "grad_norm": 0.46156010031700134, "learning_rate": 0.0005, "loss": 1.1860063076019287, "step": 41734 }, { "epoch": 3.2901064249113126, "grad_norm": 0.4635683596134186, "learning_rate": 0.0005, "loss": 1.2672375440597534, "step": 41735 }, { "epoch": 3.2901852581789517, "grad_norm": 0.49160388112068176, "learning_rate": 0.0005, "loss": 1.4640402793884277, "step": 41736 }, { "epoch": 3.2902640914465904, "grad_norm": 0.5415918231010437, "learning_rate": 0.0005, "loss": 1.2573652267456055, "step": 41737 }, { "epoch": 3.2903429247142295, "grad_norm": 0.7133769989013672, "learning_rate": 0.0005, "loss": 1.1490941047668457, "step": 41738 }, { "epoch": 3.290421757981868, "grad_norm": 0.46514979004859924, "learning_rate": 0.0005, "loss": 1.27432119846344, "step": 41739 }, { "epoch": 3.2905005912495073, "grad_norm": 0.5457512140274048, "learning_rate": 0.0005, "loss": 1.224785566329956, "step": 41740 }, { "epoch": 3.290579424517146, "grad_norm": 0.5517889857292175, "learning_rate": 0.0005, "loss": 1.2833991050720215, "step": 41741 }, { "epoch": 3.290658257784785, "grad_norm": 0.5766763091087341, "learning_rate": 0.0005, "loss": 1.2587649822235107, "step": 41742 }, { "epoch": 3.2907370910524243, "grad_norm": 0.5085377097129822, "learning_rate": 0.0005, "loss": 1.2505382299423218, "step": 41743 }, { "epoch": 3.290815924320063, "grad_norm": 0.5264818668365479, "learning_rate": 0.0005, "loss": 1.1471027135849, "step": 41744 }, { "epoch": 3.290894757587702, "grad_norm": 0.5362164378166199, "learning_rate": 0.0005, "loss": 1.2304686307907104, "step": 41745 }, { "epoch": 3.290973590855341, "grad_norm": 0.480459064245224, "learning_rate": 0.0005, "loss": 1.363121747970581, "step": 41746 }, { "epoch": 3.29105242412298, "grad_norm": 0.46922188997268677, "learning_rate": 0.0005, "loss": 1.2070395946502686, "step": 41747 }, { "epoch": 3.291131257390619, "grad_norm": 0.5399829745292664, "learning_rate": 0.0005, "loss": 1.2396373748779297, "step": 41748 }, { "epoch": 3.2912100906582578, "grad_norm": 0.4161105751991272, "learning_rate": 0.0005, "loss": 1.1029040813446045, "step": 41749 }, { "epoch": 3.291288923925897, "grad_norm": 0.4137319326400757, "learning_rate": 0.0005, "loss": 1.1896569728851318, "step": 41750 }, { "epoch": 3.2913677571935356, "grad_norm": 0.5340877771377563, "learning_rate": 0.0005, "loss": 1.1730067729949951, "step": 41751 }, { "epoch": 3.2914465904611747, "grad_norm": 0.4293448328971863, "learning_rate": 0.0005, "loss": 1.1682095527648926, "step": 41752 }, { "epoch": 3.2915254237288134, "grad_norm": 0.5201767683029175, "learning_rate": 0.0005, "loss": 1.5791339874267578, "step": 41753 }, { "epoch": 3.2916042569964525, "grad_norm": 0.49716341495513916, "learning_rate": 0.0005, "loss": 1.3557593822479248, "step": 41754 }, { "epoch": 3.2916830902640912, "grad_norm": 0.542344331741333, "learning_rate": 0.0005, "loss": 1.4452091455459595, "step": 41755 }, { "epoch": 3.2917619235317304, "grad_norm": 0.488413006067276, "learning_rate": 0.0005, "loss": 1.1910924911499023, "step": 41756 }, { "epoch": 3.2918407567993695, "grad_norm": 0.49594640731811523, "learning_rate": 0.0005, "loss": 1.2064666748046875, "step": 41757 }, { "epoch": 3.291919590067008, "grad_norm": 0.6219992637634277, "learning_rate": 0.0005, "loss": 1.341442584991455, "step": 41758 }, { "epoch": 3.2919984233346473, "grad_norm": 0.6774163842201233, "learning_rate": 0.0005, "loss": 1.5438740253448486, "step": 41759 }, { "epoch": 3.292077256602286, "grad_norm": 0.5185214281082153, "learning_rate": 0.0005, "loss": 1.234466552734375, "step": 41760 }, { "epoch": 3.292156089869925, "grad_norm": 0.4777226150035858, "learning_rate": 0.0005, "loss": 1.260406732559204, "step": 41761 }, { "epoch": 3.2922349231375643, "grad_norm": 0.5154137015342712, "learning_rate": 0.0005, "loss": 1.242393970489502, "step": 41762 }, { "epoch": 3.292313756405203, "grad_norm": 0.4767739474773407, "learning_rate": 0.0005, "loss": 1.136068344116211, "step": 41763 }, { "epoch": 3.292392589672842, "grad_norm": 0.5481808185577393, "learning_rate": 0.0005, "loss": 1.3046189546585083, "step": 41764 }, { "epoch": 3.292471422940481, "grad_norm": 0.5299889445304871, "learning_rate": 0.0005, "loss": 1.3032951354980469, "step": 41765 }, { "epoch": 3.29255025620812, "grad_norm": 0.5068818926811218, "learning_rate": 0.0005, "loss": 1.3249320983886719, "step": 41766 }, { "epoch": 3.2926290894757586, "grad_norm": 0.5484403967857361, "learning_rate": 0.0005, "loss": 1.3784739971160889, "step": 41767 }, { "epoch": 3.2927079227433977, "grad_norm": 0.5384910106658936, "learning_rate": 0.0005, "loss": 1.1696679592132568, "step": 41768 }, { "epoch": 3.2927867560110364, "grad_norm": 0.450460821390152, "learning_rate": 0.0005, "loss": 1.387204647064209, "step": 41769 }, { "epoch": 3.2928655892786756, "grad_norm": 0.47456490993499756, "learning_rate": 0.0005, "loss": 1.1450334787368774, "step": 41770 }, { "epoch": 3.2929444225463147, "grad_norm": 0.4964490830898285, "learning_rate": 0.0005, "loss": 1.2904001474380493, "step": 41771 }, { "epoch": 3.2930232558139534, "grad_norm": 0.4905734062194824, "learning_rate": 0.0005, "loss": 1.1687133312225342, "step": 41772 }, { "epoch": 3.2931020890815925, "grad_norm": 0.512662947177887, "learning_rate": 0.0005, "loss": 1.3680150508880615, "step": 41773 }, { "epoch": 3.293180922349231, "grad_norm": 0.4674578309059143, "learning_rate": 0.0005, "loss": 1.3181318044662476, "step": 41774 }, { "epoch": 3.2932597556168703, "grad_norm": 0.45549124479293823, "learning_rate": 0.0005, "loss": 1.3342469930648804, "step": 41775 }, { "epoch": 3.2933385888845095, "grad_norm": 0.5272610187530518, "learning_rate": 0.0005, "loss": 1.5399035215377808, "step": 41776 }, { "epoch": 3.293417422152148, "grad_norm": 0.5424755215644836, "learning_rate": 0.0005, "loss": 1.2949669361114502, "step": 41777 }, { "epoch": 3.2934962554197873, "grad_norm": 0.5653389096260071, "learning_rate": 0.0005, "loss": 1.2234055995941162, "step": 41778 }, { "epoch": 3.293575088687426, "grad_norm": 0.5654834508895874, "learning_rate": 0.0005, "loss": 1.3549097776412964, "step": 41779 }, { "epoch": 3.293653921955065, "grad_norm": 0.521600067615509, "learning_rate": 0.0005, "loss": 1.2788161039352417, "step": 41780 }, { "epoch": 3.293732755222704, "grad_norm": 0.5292627215385437, "learning_rate": 0.0005, "loss": 1.3211095333099365, "step": 41781 }, { "epoch": 3.293811588490343, "grad_norm": 0.5134286880493164, "learning_rate": 0.0005, "loss": 1.345203161239624, "step": 41782 }, { "epoch": 3.2938904217579816, "grad_norm": 0.46252334117889404, "learning_rate": 0.0005, "loss": 1.2398707866668701, "step": 41783 }, { "epoch": 3.2939692550256208, "grad_norm": 0.5270535349845886, "learning_rate": 0.0005, "loss": 1.2128326892852783, "step": 41784 }, { "epoch": 3.29404808829326, "grad_norm": 0.44366636872291565, "learning_rate": 0.0005, "loss": 1.2855587005615234, "step": 41785 }, { "epoch": 3.2941269215608986, "grad_norm": 0.49155083298683167, "learning_rate": 0.0005, "loss": 1.1821506023406982, "step": 41786 }, { "epoch": 3.2942057548285377, "grad_norm": 0.6487009525299072, "learning_rate": 0.0005, "loss": 1.443598985671997, "step": 41787 }, { "epoch": 3.2942845880961764, "grad_norm": 0.45999395847320557, "learning_rate": 0.0005, "loss": 1.1663179397583008, "step": 41788 }, { "epoch": 3.2943634213638155, "grad_norm": 0.4838302433490753, "learning_rate": 0.0005, "loss": 1.158003807067871, "step": 41789 }, { "epoch": 3.2944422546314547, "grad_norm": 0.49880459904670715, "learning_rate": 0.0005, "loss": 1.2909154891967773, "step": 41790 }, { "epoch": 3.2945210878990934, "grad_norm": 0.48148199915885925, "learning_rate": 0.0005, "loss": 1.2901661396026611, "step": 41791 }, { "epoch": 3.2945999211667325, "grad_norm": 0.49706411361694336, "learning_rate": 0.0005, "loss": 1.298561930656433, "step": 41792 }, { "epoch": 3.294678754434371, "grad_norm": 0.46432629227638245, "learning_rate": 0.0005, "loss": 1.26047945022583, "step": 41793 }, { "epoch": 3.2947575877020103, "grad_norm": 0.5891541838645935, "learning_rate": 0.0005, "loss": 1.209114909172058, "step": 41794 }, { "epoch": 3.294836420969649, "grad_norm": 0.42917388677597046, "learning_rate": 0.0005, "loss": 1.1504111289978027, "step": 41795 }, { "epoch": 3.294915254237288, "grad_norm": 0.5098594427108765, "learning_rate": 0.0005, "loss": 1.3601149320602417, "step": 41796 }, { "epoch": 3.294994087504927, "grad_norm": 0.4671260714530945, "learning_rate": 0.0005, "loss": 1.2701033353805542, "step": 41797 }, { "epoch": 3.295072920772566, "grad_norm": 0.4332502484321594, "learning_rate": 0.0005, "loss": 1.1704094409942627, "step": 41798 }, { "epoch": 3.295151754040205, "grad_norm": 0.4222606420516968, "learning_rate": 0.0005, "loss": 1.1372902393341064, "step": 41799 }, { "epoch": 3.295230587307844, "grad_norm": 0.572769284248352, "learning_rate": 0.0005, "loss": 1.6044082641601562, "step": 41800 }, { "epoch": 3.295309420575483, "grad_norm": 0.44632211327552795, "learning_rate": 0.0005, "loss": 1.2417924404144287, "step": 41801 }, { "epoch": 3.2953882538431216, "grad_norm": 0.6748464703559875, "learning_rate": 0.0005, "loss": 1.261936902999878, "step": 41802 }, { "epoch": 3.2954670871107608, "grad_norm": 0.4786561131477356, "learning_rate": 0.0005, "loss": 1.3371609449386597, "step": 41803 }, { "epoch": 3.2955459203784, "grad_norm": 0.48772740364074707, "learning_rate": 0.0005, "loss": 1.219496488571167, "step": 41804 }, { "epoch": 3.2956247536460386, "grad_norm": 0.500397801399231, "learning_rate": 0.0005, "loss": 1.1513111591339111, "step": 41805 }, { "epoch": 3.2957035869136777, "grad_norm": 0.45187175273895264, "learning_rate": 0.0005, "loss": 1.3277629613876343, "step": 41806 }, { "epoch": 3.2957824201813164, "grad_norm": 0.6153271198272705, "learning_rate": 0.0005, "loss": 1.2943897247314453, "step": 41807 }, { "epoch": 3.2958612534489555, "grad_norm": 0.47856417298316956, "learning_rate": 0.0005, "loss": 1.1020636558532715, "step": 41808 }, { "epoch": 3.295940086716594, "grad_norm": 1.06929349899292, "learning_rate": 0.0005, "loss": 1.5808606147766113, "step": 41809 }, { "epoch": 3.2960189199842334, "grad_norm": 0.45125818252563477, "learning_rate": 0.0005, "loss": 1.1592004299163818, "step": 41810 }, { "epoch": 3.296097753251872, "grad_norm": 0.5511773228645325, "learning_rate": 0.0005, "loss": 1.2415940761566162, "step": 41811 }, { "epoch": 3.296176586519511, "grad_norm": 0.5279024839401245, "learning_rate": 0.0005, "loss": 1.3178883790969849, "step": 41812 }, { "epoch": 3.2962554197871503, "grad_norm": 0.5143172144889832, "learning_rate": 0.0005, "loss": 1.214759349822998, "step": 41813 }, { "epoch": 3.296334253054789, "grad_norm": 1.2267258167266846, "learning_rate": 0.0005, "loss": 1.276900291442871, "step": 41814 }, { "epoch": 3.296413086322428, "grad_norm": 0.6112890243530273, "learning_rate": 0.0005, "loss": 1.6449742317199707, "step": 41815 }, { "epoch": 3.296491919590067, "grad_norm": 0.41716501116752625, "learning_rate": 0.0005, "loss": 1.175669550895691, "step": 41816 }, { "epoch": 3.296570752857706, "grad_norm": 0.5165693759918213, "learning_rate": 0.0005, "loss": 1.3639918565750122, "step": 41817 }, { "epoch": 3.296649586125345, "grad_norm": 0.5052714943885803, "learning_rate": 0.0005, "loss": 1.2524006366729736, "step": 41818 }, { "epoch": 3.2967284193929838, "grad_norm": 0.603708803653717, "learning_rate": 0.0005, "loss": 1.2901332378387451, "step": 41819 }, { "epoch": 3.296807252660623, "grad_norm": 0.44713979959487915, "learning_rate": 0.0005, "loss": 1.2674059867858887, "step": 41820 }, { "epoch": 3.2968860859282616, "grad_norm": 0.4118892252445221, "learning_rate": 0.0005, "loss": 1.1603482961654663, "step": 41821 }, { "epoch": 3.2969649191959007, "grad_norm": 0.5973157286643982, "learning_rate": 0.0005, "loss": 1.4114545583724976, "step": 41822 }, { "epoch": 3.2970437524635394, "grad_norm": 0.5536631345748901, "learning_rate": 0.0005, "loss": 1.3541665077209473, "step": 41823 }, { "epoch": 3.2971225857311786, "grad_norm": 0.5214748382568359, "learning_rate": 0.0005, "loss": 1.1718885898590088, "step": 41824 }, { "epoch": 3.2972014189988177, "grad_norm": 0.563778817653656, "learning_rate": 0.0005, "loss": 1.1762464046478271, "step": 41825 }, { "epoch": 3.2972802522664564, "grad_norm": 0.6876708269119263, "learning_rate": 0.0005, "loss": 1.3333684206008911, "step": 41826 }, { "epoch": 3.2973590855340955, "grad_norm": 0.47473230957984924, "learning_rate": 0.0005, "loss": 1.1935091018676758, "step": 41827 }, { "epoch": 3.297437918801734, "grad_norm": 0.6665705442428589, "learning_rate": 0.0005, "loss": 1.3461848497390747, "step": 41828 }, { "epoch": 3.2975167520693733, "grad_norm": 0.47693875432014465, "learning_rate": 0.0005, "loss": 1.4041900634765625, "step": 41829 }, { "epoch": 3.297595585337012, "grad_norm": 0.675572395324707, "learning_rate": 0.0005, "loss": 1.2714210748672485, "step": 41830 }, { "epoch": 3.297674418604651, "grad_norm": 0.46503913402557373, "learning_rate": 0.0005, "loss": 1.1766189336776733, "step": 41831 }, { "epoch": 3.2977532518722903, "grad_norm": 0.5346975326538086, "learning_rate": 0.0005, "loss": 1.4085748195648193, "step": 41832 }, { "epoch": 3.297832085139929, "grad_norm": 0.5430369973182678, "learning_rate": 0.0005, "loss": 1.5740821361541748, "step": 41833 }, { "epoch": 3.297910918407568, "grad_norm": 0.504619300365448, "learning_rate": 0.0005, "loss": 1.2617671489715576, "step": 41834 }, { "epoch": 3.297989751675207, "grad_norm": 0.5223215222358704, "learning_rate": 0.0005, "loss": 1.4317667484283447, "step": 41835 }, { "epoch": 3.298068584942846, "grad_norm": 0.5338739156723022, "learning_rate": 0.0005, "loss": 1.1081880331039429, "step": 41836 }, { "epoch": 3.298147418210485, "grad_norm": 0.4603341519832611, "learning_rate": 0.0005, "loss": 1.1971967220306396, "step": 41837 }, { "epoch": 3.2982262514781238, "grad_norm": 0.47762948274612427, "learning_rate": 0.0005, "loss": 1.4061501026153564, "step": 41838 }, { "epoch": 3.298305084745763, "grad_norm": 0.5312456488609314, "learning_rate": 0.0005, "loss": 1.4077441692352295, "step": 41839 }, { "epoch": 3.2983839180134016, "grad_norm": 0.41636061668395996, "learning_rate": 0.0005, "loss": 1.0517137050628662, "step": 41840 }, { "epoch": 3.2984627512810407, "grad_norm": 0.6049072742462158, "learning_rate": 0.0005, "loss": 1.2304320335388184, "step": 41841 }, { "epoch": 3.2985415845486794, "grad_norm": 0.4176315367221832, "learning_rate": 0.0005, "loss": 1.1231873035430908, "step": 41842 }, { "epoch": 3.2986204178163185, "grad_norm": 0.49920153617858887, "learning_rate": 0.0005, "loss": 1.3643887042999268, "step": 41843 }, { "epoch": 3.2986992510839572, "grad_norm": 0.3964249789714813, "learning_rate": 0.0005, "loss": 1.0423468351364136, "step": 41844 }, { "epoch": 3.2987780843515964, "grad_norm": 0.5031570196151733, "learning_rate": 0.0005, "loss": 1.2293334007263184, "step": 41845 }, { "epoch": 3.2988569176192355, "grad_norm": 0.7191113829612732, "learning_rate": 0.0005, "loss": 1.396061658859253, "step": 41846 }, { "epoch": 3.298935750886874, "grad_norm": 0.5313044786453247, "learning_rate": 0.0005, "loss": 1.2582192420959473, "step": 41847 }, { "epoch": 3.2990145841545133, "grad_norm": 0.5856660008430481, "learning_rate": 0.0005, "loss": 1.519425630569458, "step": 41848 }, { "epoch": 3.299093417422152, "grad_norm": 0.4384375512599945, "learning_rate": 0.0005, "loss": 1.005298376083374, "step": 41849 }, { "epoch": 3.299172250689791, "grad_norm": 0.4672095775604248, "learning_rate": 0.0005, "loss": 1.3292094469070435, "step": 41850 }, { "epoch": 3.2992510839574303, "grad_norm": 0.5099632143974304, "learning_rate": 0.0005, "loss": 1.2004731893539429, "step": 41851 }, { "epoch": 3.299329917225069, "grad_norm": 0.3805859088897705, "learning_rate": 0.0005, "loss": 1.067939281463623, "step": 41852 }, { "epoch": 3.299408750492708, "grad_norm": 0.5386814475059509, "learning_rate": 0.0005, "loss": 1.2558434009552002, "step": 41853 }, { "epoch": 3.299487583760347, "grad_norm": 0.5010942816734314, "learning_rate": 0.0005, "loss": 1.0731689929962158, "step": 41854 }, { "epoch": 3.299566417027986, "grad_norm": 0.47635966539382935, "learning_rate": 0.0005, "loss": 1.1833455562591553, "step": 41855 }, { "epoch": 3.2996452502956246, "grad_norm": 0.8529984951019287, "learning_rate": 0.0005, "loss": 1.3876643180847168, "step": 41856 }, { "epoch": 3.2997240835632637, "grad_norm": 0.5080602169036865, "learning_rate": 0.0005, "loss": 1.3132076263427734, "step": 41857 }, { "epoch": 3.2998029168309024, "grad_norm": 0.45393216609954834, "learning_rate": 0.0005, "loss": 1.1930487155914307, "step": 41858 }, { "epoch": 3.2998817500985416, "grad_norm": 0.5497251749038696, "learning_rate": 0.0005, "loss": 1.1253756284713745, "step": 41859 }, { "epoch": 3.2999605833661807, "grad_norm": 0.5113169550895691, "learning_rate": 0.0005, "loss": 1.3312395811080933, "step": 41860 }, { "epoch": 3.3000394166338194, "grad_norm": 0.4716039001941681, "learning_rate": 0.0005, "loss": 1.2982200384140015, "step": 41861 }, { "epoch": 3.3001182499014585, "grad_norm": 0.535132884979248, "learning_rate": 0.0005, "loss": 1.2410467863082886, "step": 41862 }, { "epoch": 3.300197083169097, "grad_norm": 0.4491260051727295, "learning_rate": 0.0005, "loss": 1.2313642501831055, "step": 41863 }, { "epoch": 3.3002759164367363, "grad_norm": 0.4590247869491577, "learning_rate": 0.0005, "loss": 1.1560389995574951, "step": 41864 }, { "epoch": 3.3003547497043755, "grad_norm": 0.5057417750358582, "learning_rate": 0.0005, "loss": 1.2836235761642456, "step": 41865 }, { "epoch": 3.300433582972014, "grad_norm": 0.4944101572036743, "learning_rate": 0.0005, "loss": 1.2397456169128418, "step": 41866 }, { "epoch": 3.3005124162396533, "grad_norm": 0.5873635411262512, "learning_rate": 0.0005, "loss": 1.2346984148025513, "step": 41867 }, { "epoch": 3.300591249507292, "grad_norm": 0.45766156911849976, "learning_rate": 0.0005, "loss": 1.2379894256591797, "step": 41868 }, { "epoch": 3.300670082774931, "grad_norm": 0.4302741587162018, "learning_rate": 0.0005, "loss": 1.2277653217315674, "step": 41869 }, { "epoch": 3.30074891604257, "grad_norm": 0.5556637644767761, "learning_rate": 0.0005, "loss": 1.542435646057129, "step": 41870 }, { "epoch": 3.300827749310209, "grad_norm": 0.5078682899475098, "learning_rate": 0.0005, "loss": 1.1159340143203735, "step": 41871 }, { "epoch": 3.3009065825778476, "grad_norm": 0.5895577073097229, "learning_rate": 0.0005, "loss": 1.3986015319824219, "step": 41872 }, { "epoch": 3.3009854158454868, "grad_norm": 0.533717930316925, "learning_rate": 0.0005, "loss": 1.2272450923919678, "step": 41873 }, { "epoch": 3.301064249113126, "grad_norm": 0.6632815599441528, "learning_rate": 0.0005, "loss": 1.3580585718154907, "step": 41874 }, { "epoch": 3.3011430823807646, "grad_norm": 0.5143144130706787, "learning_rate": 0.0005, "loss": 1.3055108785629272, "step": 41875 }, { "epoch": 3.3012219156484037, "grad_norm": 0.5315781831741333, "learning_rate": 0.0005, "loss": 1.3837882280349731, "step": 41876 }, { "epoch": 3.3013007489160424, "grad_norm": 0.5174258947372437, "learning_rate": 0.0005, "loss": 1.2202023267745972, "step": 41877 }, { "epoch": 3.3013795821836815, "grad_norm": 0.4732658863067627, "learning_rate": 0.0005, "loss": 1.3956513404846191, "step": 41878 }, { "epoch": 3.3014584154513207, "grad_norm": 0.5761735439300537, "learning_rate": 0.0005, "loss": 1.2600284814834595, "step": 41879 }, { "epoch": 3.3015372487189594, "grad_norm": 0.5764456987380981, "learning_rate": 0.0005, "loss": 1.2091361284255981, "step": 41880 }, { "epoch": 3.3016160819865985, "grad_norm": 0.5735338926315308, "learning_rate": 0.0005, "loss": 1.4807171821594238, "step": 41881 }, { "epoch": 3.301694915254237, "grad_norm": 0.5548096895217896, "learning_rate": 0.0005, "loss": 1.2419030666351318, "step": 41882 }, { "epoch": 3.3017737485218763, "grad_norm": 0.5736487507820129, "learning_rate": 0.0005, "loss": 1.2623897790908813, "step": 41883 }, { "epoch": 3.301852581789515, "grad_norm": 0.5170615315437317, "learning_rate": 0.0005, "loss": 1.372476577758789, "step": 41884 }, { "epoch": 3.301931415057154, "grad_norm": 0.3795608580112457, "learning_rate": 0.0005, "loss": 1.20823073387146, "step": 41885 }, { "epoch": 3.302010248324793, "grad_norm": 0.40131866931915283, "learning_rate": 0.0005, "loss": 1.1191065311431885, "step": 41886 }, { "epoch": 3.302089081592432, "grad_norm": 0.4032849371433258, "learning_rate": 0.0005, "loss": 1.1813404560089111, "step": 41887 }, { "epoch": 3.302167914860071, "grad_norm": 0.5546548366546631, "learning_rate": 0.0005, "loss": 1.3049383163452148, "step": 41888 }, { "epoch": 3.30224674812771, "grad_norm": 0.5171524882316589, "learning_rate": 0.0005, "loss": 1.3535447120666504, "step": 41889 }, { "epoch": 3.302325581395349, "grad_norm": 0.5159822702407837, "learning_rate": 0.0005, "loss": 1.2101205587387085, "step": 41890 }, { "epoch": 3.3024044146629876, "grad_norm": 0.5329512357711792, "learning_rate": 0.0005, "loss": 1.4384560585021973, "step": 41891 }, { "epoch": 3.3024832479306268, "grad_norm": 0.4929252564907074, "learning_rate": 0.0005, "loss": 1.1700522899627686, "step": 41892 }, { "epoch": 3.302562081198266, "grad_norm": 0.4683291018009186, "learning_rate": 0.0005, "loss": 1.0723261833190918, "step": 41893 }, { "epoch": 3.3026409144659046, "grad_norm": 0.4402637481689453, "learning_rate": 0.0005, "loss": 1.2101022005081177, "step": 41894 }, { "epoch": 3.3027197477335437, "grad_norm": 0.47329580783843994, "learning_rate": 0.0005, "loss": 1.2958250045776367, "step": 41895 }, { "epoch": 3.3027985810011824, "grad_norm": 1.3275924921035767, "learning_rate": 0.0005, "loss": 1.1187196969985962, "step": 41896 }, { "epoch": 3.3028774142688215, "grad_norm": 0.4434497356414795, "learning_rate": 0.0005, "loss": 1.177448034286499, "step": 41897 }, { "epoch": 3.30295624753646, "grad_norm": 0.43319910764694214, "learning_rate": 0.0005, "loss": 1.1557286977767944, "step": 41898 }, { "epoch": 3.3030350808040994, "grad_norm": 0.500026524066925, "learning_rate": 0.0005, "loss": 1.3341846466064453, "step": 41899 }, { "epoch": 3.303113914071738, "grad_norm": 0.4727371335029602, "learning_rate": 0.0005, "loss": 1.2792448997497559, "step": 41900 }, { "epoch": 3.303192747339377, "grad_norm": 0.5022775530815125, "learning_rate": 0.0005, "loss": 1.3099708557128906, "step": 41901 }, { "epoch": 3.3032715806070163, "grad_norm": 0.46157941222190857, "learning_rate": 0.0005, "loss": 1.2876169681549072, "step": 41902 }, { "epoch": 3.303350413874655, "grad_norm": 0.4573337137699127, "learning_rate": 0.0005, "loss": 1.3198304176330566, "step": 41903 }, { "epoch": 3.303429247142294, "grad_norm": 0.5206870436668396, "learning_rate": 0.0005, "loss": 1.1926273107528687, "step": 41904 }, { "epoch": 3.303508080409933, "grad_norm": 0.4518970251083374, "learning_rate": 0.0005, "loss": 1.1570978164672852, "step": 41905 }, { "epoch": 3.303586913677572, "grad_norm": 2.3292834758758545, "learning_rate": 0.0005, "loss": 1.307976245880127, "step": 41906 }, { "epoch": 3.303665746945211, "grad_norm": 0.44559258222579956, "learning_rate": 0.0005, "loss": 1.268445372581482, "step": 41907 }, { "epoch": 3.30374458021285, "grad_norm": 0.5669609904289246, "learning_rate": 0.0005, "loss": 1.3252187967300415, "step": 41908 }, { "epoch": 3.303823413480489, "grad_norm": 0.5066503286361694, "learning_rate": 0.0005, "loss": 1.3524962663650513, "step": 41909 }, { "epoch": 3.3039022467481276, "grad_norm": 0.710760772228241, "learning_rate": 0.0005, "loss": 1.380415916442871, "step": 41910 }, { "epoch": 3.3039810800157667, "grad_norm": 0.5738423466682434, "learning_rate": 0.0005, "loss": 1.2053024768829346, "step": 41911 }, { "epoch": 3.3040599132834054, "grad_norm": 0.5674883723258972, "learning_rate": 0.0005, "loss": 1.2892065048217773, "step": 41912 }, { "epoch": 3.3041387465510446, "grad_norm": 0.3870335519313812, "learning_rate": 0.0005, "loss": 1.3154226541519165, "step": 41913 }, { "epoch": 3.3042175798186832, "grad_norm": 0.44048354029655457, "learning_rate": 0.0005, "loss": 1.278269648551941, "step": 41914 }, { "epoch": 3.3042964130863224, "grad_norm": 0.509925127029419, "learning_rate": 0.0005, "loss": 1.3896077871322632, "step": 41915 }, { "epoch": 3.3043752463539615, "grad_norm": 0.5178061127662659, "learning_rate": 0.0005, "loss": 1.2474429607391357, "step": 41916 }, { "epoch": 3.3044540796216, "grad_norm": 0.43152979016304016, "learning_rate": 0.0005, "loss": 1.3039755821228027, "step": 41917 }, { "epoch": 3.3045329128892393, "grad_norm": 0.45352765917778015, "learning_rate": 0.0005, "loss": 1.3360605239868164, "step": 41918 }, { "epoch": 3.304611746156878, "grad_norm": 0.4357028305530548, "learning_rate": 0.0005, "loss": 1.2996375560760498, "step": 41919 }, { "epoch": 3.304690579424517, "grad_norm": 0.42459964752197266, "learning_rate": 0.0005, "loss": 1.3310213088989258, "step": 41920 }, { "epoch": 3.3047694126921563, "grad_norm": 0.42470622062683105, "learning_rate": 0.0005, "loss": 1.1528620719909668, "step": 41921 }, { "epoch": 3.304848245959795, "grad_norm": 0.5503382682800293, "learning_rate": 0.0005, "loss": 1.2253693342208862, "step": 41922 }, { "epoch": 3.304927079227434, "grad_norm": 1.1831204891204834, "learning_rate": 0.0005, "loss": 1.3004343509674072, "step": 41923 }, { "epoch": 3.305005912495073, "grad_norm": 0.5524369478225708, "learning_rate": 0.0005, "loss": 1.2084262371063232, "step": 41924 }, { "epoch": 3.305084745762712, "grad_norm": 0.5761425495147705, "learning_rate": 0.0005, "loss": 1.3865388631820679, "step": 41925 }, { "epoch": 3.3051635790303506, "grad_norm": 0.4700186550617218, "learning_rate": 0.0005, "loss": 1.1998858451843262, "step": 41926 }, { "epoch": 3.3052424122979898, "grad_norm": 0.48319926857948303, "learning_rate": 0.0005, "loss": 1.361744999885559, "step": 41927 }, { "epoch": 3.3053212455656285, "grad_norm": 0.5875772833824158, "learning_rate": 0.0005, "loss": 1.4921963214874268, "step": 41928 }, { "epoch": 3.3054000788332676, "grad_norm": 0.5424760580062866, "learning_rate": 0.0005, "loss": 1.4784572124481201, "step": 41929 }, { "epoch": 3.3054789121009067, "grad_norm": 0.48517587780952454, "learning_rate": 0.0005, "loss": 1.197364330291748, "step": 41930 }, { "epoch": 3.3055577453685454, "grad_norm": 0.398144006729126, "learning_rate": 0.0005, "loss": 1.2547898292541504, "step": 41931 }, { "epoch": 3.3056365786361845, "grad_norm": 0.5159303545951843, "learning_rate": 0.0005, "loss": 1.5018383264541626, "step": 41932 }, { "epoch": 3.3057154119038232, "grad_norm": 0.4400293231010437, "learning_rate": 0.0005, "loss": 1.1901516914367676, "step": 41933 }, { "epoch": 3.3057942451714624, "grad_norm": 0.4305919110774994, "learning_rate": 0.0005, "loss": 1.2088522911071777, "step": 41934 }, { "epoch": 3.3058730784391015, "grad_norm": 0.42705920338630676, "learning_rate": 0.0005, "loss": 1.13735830783844, "step": 41935 }, { "epoch": 3.30595191170674, "grad_norm": 0.45732927322387695, "learning_rate": 0.0005, "loss": 1.1402171850204468, "step": 41936 }, { "epoch": 3.3060307449743793, "grad_norm": 0.42304277420043945, "learning_rate": 0.0005, "loss": 1.3868076801300049, "step": 41937 }, { "epoch": 3.306109578242018, "grad_norm": 0.6479952931404114, "learning_rate": 0.0005, "loss": 1.4138357639312744, "step": 41938 }, { "epoch": 3.306188411509657, "grad_norm": 0.5274357795715332, "learning_rate": 0.0005, "loss": 1.4159191846847534, "step": 41939 }, { "epoch": 3.306267244777296, "grad_norm": 0.5136909484863281, "learning_rate": 0.0005, "loss": 1.189521074295044, "step": 41940 }, { "epoch": 3.306346078044935, "grad_norm": 0.5126540660858154, "learning_rate": 0.0005, "loss": 1.3599395751953125, "step": 41941 }, { "epoch": 3.306424911312574, "grad_norm": 0.48874029517173767, "learning_rate": 0.0005, "loss": 1.269168734550476, "step": 41942 }, { "epoch": 3.306503744580213, "grad_norm": 0.4051972031593323, "learning_rate": 0.0005, "loss": 1.2058931589126587, "step": 41943 }, { "epoch": 3.306582577847852, "grad_norm": 0.5702098608016968, "learning_rate": 0.0005, "loss": 1.3743157386779785, "step": 41944 }, { "epoch": 3.3066614111154906, "grad_norm": 0.5451082587242126, "learning_rate": 0.0005, "loss": 1.3142427206039429, "step": 41945 }, { "epoch": 3.3067402443831297, "grad_norm": 0.5908944606781006, "learning_rate": 0.0005, "loss": 1.3606475591659546, "step": 41946 }, { "epoch": 3.3068190776507684, "grad_norm": 0.49508947134017944, "learning_rate": 0.0005, "loss": 1.1550819873809814, "step": 41947 }, { "epoch": 3.3068979109184076, "grad_norm": 0.5263635516166687, "learning_rate": 0.0005, "loss": 1.2665154933929443, "step": 41948 }, { "epoch": 3.3069767441860467, "grad_norm": 0.520010232925415, "learning_rate": 0.0005, "loss": 1.1617999076843262, "step": 41949 }, { "epoch": 3.3070555774536854, "grad_norm": 0.41107872128486633, "learning_rate": 0.0005, "loss": 1.1168456077575684, "step": 41950 }, { "epoch": 3.3071344107213245, "grad_norm": 0.45715466141700745, "learning_rate": 0.0005, "loss": 1.1698520183563232, "step": 41951 }, { "epoch": 3.307213243988963, "grad_norm": 0.497233122587204, "learning_rate": 0.0005, "loss": 1.2479374408721924, "step": 41952 }, { "epoch": 3.3072920772566023, "grad_norm": 0.7082249522209167, "learning_rate": 0.0005, "loss": 1.2696760892868042, "step": 41953 }, { "epoch": 3.307370910524241, "grad_norm": 0.49815505743026733, "learning_rate": 0.0005, "loss": 1.3009388446807861, "step": 41954 }, { "epoch": 3.30744974379188, "grad_norm": 0.46848639845848083, "learning_rate": 0.0005, "loss": 1.308410406112671, "step": 41955 }, { "epoch": 3.3075285770595193, "grad_norm": 0.462158203125, "learning_rate": 0.0005, "loss": 1.1946208477020264, "step": 41956 }, { "epoch": 3.307607410327158, "grad_norm": 0.5504181981086731, "learning_rate": 0.0005, "loss": 1.3193154335021973, "step": 41957 }, { "epoch": 3.307686243594797, "grad_norm": 0.5218682289123535, "learning_rate": 0.0005, "loss": 1.2711141109466553, "step": 41958 }, { "epoch": 3.307765076862436, "grad_norm": 0.44344818592071533, "learning_rate": 0.0005, "loss": 1.3035314083099365, "step": 41959 }, { "epoch": 3.307843910130075, "grad_norm": 0.5348538160324097, "learning_rate": 0.0005, "loss": 1.312483310699463, "step": 41960 }, { "epoch": 3.3079227433977136, "grad_norm": 0.4654194712638855, "learning_rate": 0.0005, "loss": 1.3087708950042725, "step": 41961 }, { "epoch": 3.3080015766653528, "grad_norm": 0.5372716784477234, "learning_rate": 0.0005, "loss": 1.4474380016326904, "step": 41962 }, { "epoch": 3.308080409932992, "grad_norm": 0.42599281668663025, "learning_rate": 0.0005, "loss": 1.1836214065551758, "step": 41963 }, { "epoch": 3.3081592432006306, "grad_norm": 0.4542127847671509, "learning_rate": 0.0005, "loss": 1.2626625299453735, "step": 41964 }, { "epoch": 3.3082380764682697, "grad_norm": 0.660170316696167, "learning_rate": 0.0005, "loss": 1.3079291582107544, "step": 41965 }, { "epoch": 3.3083169097359084, "grad_norm": 0.5676453113555908, "learning_rate": 0.0005, "loss": 1.454423427581787, "step": 41966 }, { "epoch": 3.3083957430035476, "grad_norm": 0.43809863924980164, "learning_rate": 0.0005, "loss": 1.2255735397338867, "step": 41967 }, { "epoch": 3.3084745762711867, "grad_norm": 0.5298854112625122, "learning_rate": 0.0005, "loss": 1.4364123344421387, "step": 41968 }, { "epoch": 3.3085534095388254, "grad_norm": 0.5391120910644531, "learning_rate": 0.0005, "loss": 1.285109519958496, "step": 41969 }, { "epoch": 3.3086322428064645, "grad_norm": 0.5208351016044617, "learning_rate": 0.0005, "loss": 1.4362847805023193, "step": 41970 }, { "epoch": 3.308711076074103, "grad_norm": 0.5025407075881958, "learning_rate": 0.0005, "loss": 1.2792657613754272, "step": 41971 }, { "epoch": 3.3087899093417423, "grad_norm": 0.49360594153404236, "learning_rate": 0.0005, "loss": 1.3033610582351685, "step": 41972 }, { "epoch": 3.308868742609381, "grad_norm": 0.42561450600624084, "learning_rate": 0.0005, "loss": 1.019817590713501, "step": 41973 }, { "epoch": 3.30894757587702, "grad_norm": 0.4570317268371582, "learning_rate": 0.0005, "loss": 1.1284961700439453, "step": 41974 }, { "epoch": 3.309026409144659, "grad_norm": 0.5082374215126038, "learning_rate": 0.0005, "loss": 1.2380805015563965, "step": 41975 }, { "epoch": 3.309105242412298, "grad_norm": 0.6179404258728027, "learning_rate": 0.0005, "loss": 1.3205080032348633, "step": 41976 }, { "epoch": 3.309184075679937, "grad_norm": 0.4844629466533661, "learning_rate": 0.0005, "loss": 1.2968109846115112, "step": 41977 }, { "epoch": 3.309262908947576, "grad_norm": 0.45296284556388855, "learning_rate": 0.0005, "loss": 1.1613595485687256, "step": 41978 }, { "epoch": 3.309341742215215, "grad_norm": 0.4830847680568695, "learning_rate": 0.0005, "loss": 1.4290423393249512, "step": 41979 }, { "epoch": 3.3094205754828536, "grad_norm": 0.44516482949256897, "learning_rate": 0.0005, "loss": 1.2070341110229492, "step": 41980 }, { "epoch": 3.3094994087504928, "grad_norm": 0.5468893647193909, "learning_rate": 0.0005, "loss": 1.346147060394287, "step": 41981 }, { "epoch": 3.309578242018132, "grad_norm": 0.7181835174560547, "learning_rate": 0.0005, "loss": 1.3627420663833618, "step": 41982 }, { "epoch": 3.3096570752857706, "grad_norm": 0.47689491510391235, "learning_rate": 0.0005, "loss": 1.1854768991470337, "step": 41983 }, { "epoch": 3.3097359085534097, "grad_norm": 0.47511109709739685, "learning_rate": 0.0005, "loss": 1.3247885704040527, "step": 41984 }, { "epoch": 3.3098147418210484, "grad_norm": 0.4732417166233063, "learning_rate": 0.0005, "loss": 1.151748776435852, "step": 41985 }, { "epoch": 3.3098935750886875, "grad_norm": 0.44549062848091125, "learning_rate": 0.0005, "loss": 1.2754945755004883, "step": 41986 }, { "epoch": 3.3099724083563262, "grad_norm": 0.5021743178367615, "learning_rate": 0.0005, "loss": 1.138929843902588, "step": 41987 }, { "epoch": 3.3100512416239654, "grad_norm": 0.5799869894981384, "learning_rate": 0.0005, "loss": 1.4703998565673828, "step": 41988 }, { "epoch": 3.310130074891604, "grad_norm": 0.6396578550338745, "learning_rate": 0.0005, "loss": 1.3815257549285889, "step": 41989 }, { "epoch": 3.310208908159243, "grad_norm": 0.45689746737480164, "learning_rate": 0.0005, "loss": 1.1320180892944336, "step": 41990 }, { "epoch": 3.3102877414268823, "grad_norm": 1.2571098804473877, "learning_rate": 0.0005, "loss": 1.180165410041809, "step": 41991 }, { "epoch": 3.310366574694521, "grad_norm": 0.46568459272384644, "learning_rate": 0.0005, "loss": 1.1636481285095215, "step": 41992 }, { "epoch": 3.31044540796216, "grad_norm": 0.4897395074367523, "learning_rate": 0.0005, "loss": 1.458134651184082, "step": 41993 }, { "epoch": 3.310524241229799, "grad_norm": 0.5093777775764465, "learning_rate": 0.0005, "loss": 1.3458454608917236, "step": 41994 }, { "epoch": 3.310603074497438, "grad_norm": 0.42131954431533813, "learning_rate": 0.0005, "loss": 1.042020320892334, "step": 41995 }, { "epoch": 3.310681907765077, "grad_norm": 0.4344384968280792, "learning_rate": 0.0005, "loss": 1.2827285528182983, "step": 41996 }, { "epoch": 3.310760741032716, "grad_norm": 0.608624279499054, "learning_rate": 0.0005, "loss": 1.2965738773345947, "step": 41997 }, { "epoch": 3.310839574300355, "grad_norm": 0.4723442494869232, "learning_rate": 0.0005, "loss": 1.230270504951477, "step": 41998 }, { "epoch": 3.3109184075679936, "grad_norm": 0.5777134299278259, "learning_rate": 0.0005, "loss": 1.361314058303833, "step": 41999 }, { "epoch": 3.3109972408356327, "grad_norm": 0.5476092100143433, "learning_rate": 0.0005, "loss": 1.392516851425171, "step": 42000 }, { "epoch": 3.3110760741032714, "grad_norm": 0.44177961349487305, "learning_rate": 0.0005, "loss": 1.2231663465499878, "step": 42001 }, { "epoch": 3.3111549073709106, "grad_norm": 0.5655843019485474, "learning_rate": 0.0005, "loss": 1.5287050008773804, "step": 42002 }, { "epoch": 3.3112337406385492, "grad_norm": 0.4927029311656952, "learning_rate": 0.0005, "loss": 1.2550181150436401, "step": 42003 }, { "epoch": 3.3113125739061884, "grad_norm": 0.5015349388122559, "learning_rate": 0.0005, "loss": 1.252601146697998, "step": 42004 }, { "epoch": 3.3113914071738275, "grad_norm": 0.4578339457511902, "learning_rate": 0.0005, "loss": 1.2219997644424438, "step": 42005 }, { "epoch": 3.311470240441466, "grad_norm": 0.48051610589027405, "learning_rate": 0.0005, "loss": 1.4688812494277954, "step": 42006 }, { "epoch": 3.3115490737091053, "grad_norm": 1.412675380706787, "learning_rate": 0.0005, "loss": 1.4810433387756348, "step": 42007 }, { "epoch": 3.311627906976744, "grad_norm": 0.48699942231178284, "learning_rate": 0.0005, "loss": 1.3808929920196533, "step": 42008 }, { "epoch": 3.311706740244383, "grad_norm": 0.720150351524353, "learning_rate": 0.0005, "loss": 1.2351467609405518, "step": 42009 }, { "epoch": 3.3117855735120223, "grad_norm": 0.5877724885940552, "learning_rate": 0.0005, "loss": 1.3770883083343506, "step": 42010 }, { "epoch": 3.311864406779661, "grad_norm": 0.5199798941612244, "learning_rate": 0.0005, "loss": 1.3308991193771362, "step": 42011 }, { "epoch": 3.3119432400473, "grad_norm": 0.3938472867012024, "learning_rate": 0.0005, "loss": 1.1019909381866455, "step": 42012 }, { "epoch": 3.312022073314939, "grad_norm": 0.4442683160305023, "learning_rate": 0.0005, "loss": 1.089773416519165, "step": 42013 }, { "epoch": 3.312100906582578, "grad_norm": 0.5062145590782166, "learning_rate": 0.0005, "loss": 1.3032915592193604, "step": 42014 }, { "epoch": 3.3121797398502166, "grad_norm": 0.4320179224014282, "learning_rate": 0.0005, "loss": 1.1769883632659912, "step": 42015 }, { "epoch": 3.3122585731178558, "grad_norm": 0.5242143273353577, "learning_rate": 0.0005, "loss": 1.2329128980636597, "step": 42016 }, { "epoch": 3.3123374063854945, "grad_norm": 0.6133084893226624, "learning_rate": 0.0005, "loss": 1.3913962841033936, "step": 42017 }, { "epoch": 3.3124162396531336, "grad_norm": 0.41590946912765503, "learning_rate": 0.0005, "loss": 1.1050621271133423, "step": 42018 }, { "epoch": 3.3124950729207727, "grad_norm": 0.43010830879211426, "learning_rate": 0.0005, "loss": 1.3205363750457764, "step": 42019 }, { "epoch": 3.3125739061884114, "grad_norm": 0.5601261258125305, "learning_rate": 0.0005, "loss": 1.3472038507461548, "step": 42020 }, { "epoch": 3.3126527394560505, "grad_norm": 0.5132040977478027, "learning_rate": 0.0005, "loss": 1.2849297523498535, "step": 42021 }, { "epoch": 3.3127315727236892, "grad_norm": 0.4674268066883087, "learning_rate": 0.0005, "loss": 1.2525231838226318, "step": 42022 }, { "epoch": 3.3128104059913284, "grad_norm": 0.5922921299934387, "learning_rate": 0.0005, "loss": 1.3551831245422363, "step": 42023 }, { "epoch": 3.3128892392589675, "grad_norm": 0.48321533203125, "learning_rate": 0.0005, "loss": 1.3330916166305542, "step": 42024 }, { "epoch": 3.312968072526606, "grad_norm": 0.5479176640510559, "learning_rate": 0.0005, "loss": 1.3322017192840576, "step": 42025 }, { "epoch": 3.3130469057942453, "grad_norm": 0.4504791796207428, "learning_rate": 0.0005, "loss": 1.1966447830200195, "step": 42026 }, { "epoch": 3.313125739061884, "grad_norm": 0.45558875799179077, "learning_rate": 0.0005, "loss": 1.3368091583251953, "step": 42027 }, { "epoch": 3.313204572329523, "grad_norm": 0.7349852323532104, "learning_rate": 0.0005, "loss": 1.132364273071289, "step": 42028 }, { "epoch": 3.313283405597162, "grad_norm": 0.41940948367118835, "learning_rate": 0.0005, "loss": 1.1450594663619995, "step": 42029 }, { "epoch": 3.313362238864801, "grad_norm": 0.5538318157196045, "learning_rate": 0.0005, "loss": 1.485581398010254, "step": 42030 }, { "epoch": 3.3134410721324397, "grad_norm": 0.5673356652259827, "learning_rate": 0.0005, "loss": 1.1750664710998535, "step": 42031 }, { "epoch": 3.313519905400079, "grad_norm": 0.600793719291687, "learning_rate": 0.0005, "loss": 1.4029113054275513, "step": 42032 }, { "epoch": 3.313598738667718, "grad_norm": 0.5390565991401672, "learning_rate": 0.0005, "loss": 1.4540586471557617, "step": 42033 }, { "epoch": 3.3136775719353566, "grad_norm": 0.42218565940856934, "learning_rate": 0.0005, "loss": 0.9963094592094421, "step": 42034 }, { "epoch": 3.3137564052029957, "grad_norm": 0.5278711318969727, "learning_rate": 0.0005, "loss": 1.233825445175171, "step": 42035 }, { "epoch": 3.3138352384706344, "grad_norm": 0.45876508951187134, "learning_rate": 0.0005, "loss": 1.2793775796890259, "step": 42036 }, { "epoch": 3.3139140717382736, "grad_norm": 0.48531362414360046, "learning_rate": 0.0005, "loss": 1.2603955268859863, "step": 42037 }, { "epoch": 3.3139929050059127, "grad_norm": 0.47501641511917114, "learning_rate": 0.0005, "loss": 1.265389323234558, "step": 42038 }, { "epoch": 3.3140717382735514, "grad_norm": 0.5090013742446899, "learning_rate": 0.0005, "loss": 1.1933352947235107, "step": 42039 }, { "epoch": 3.3141505715411905, "grad_norm": 0.5239922404289246, "learning_rate": 0.0005, "loss": 1.2767024040222168, "step": 42040 }, { "epoch": 3.314229404808829, "grad_norm": 0.45172157883644104, "learning_rate": 0.0005, "loss": 1.2564976215362549, "step": 42041 }, { "epoch": 3.3143082380764683, "grad_norm": 0.5654228925704956, "learning_rate": 0.0005, "loss": 1.3529411554336548, "step": 42042 }, { "epoch": 3.314387071344107, "grad_norm": 0.5601305961608887, "learning_rate": 0.0005, "loss": 1.418510913848877, "step": 42043 }, { "epoch": 3.314465904611746, "grad_norm": 0.3968908488750458, "learning_rate": 0.0005, "loss": 1.1815569400787354, "step": 42044 }, { "epoch": 3.314544737879385, "grad_norm": 0.39471718668937683, "learning_rate": 0.0005, "loss": 1.1371045112609863, "step": 42045 }, { "epoch": 3.314623571147024, "grad_norm": 0.7466753721237183, "learning_rate": 0.0005, "loss": 1.3991105556488037, "step": 42046 }, { "epoch": 3.314702404414663, "grad_norm": 0.4625467360019684, "learning_rate": 0.0005, "loss": 1.3484692573547363, "step": 42047 }, { "epoch": 3.314781237682302, "grad_norm": 0.6875941157341003, "learning_rate": 0.0005, "loss": 1.3092052936553955, "step": 42048 }, { "epoch": 3.314860070949941, "grad_norm": 0.4689771234989166, "learning_rate": 0.0005, "loss": 1.3601934909820557, "step": 42049 }, { "epoch": 3.3149389042175796, "grad_norm": 0.5085232257843018, "learning_rate": 0.0005, "loss": 1.2327921390533447, "step": 42050 }, { "epoch": 3.3150177374852188, "grad_norm": 0.5665255188941956, "learning_rate": 0.0005, "loss": 1.2248132228851318, "step": 42051 }, { "epoch": 3.315096570752858, "grad_norm": 0.4890805184841156, "learning_rate": 0.0005, "loss": 1.274398684501648, "step": 42052 }, { "epoch": 3.3151754040204966, "grad_norm": 0.47321170568466187, "learning_rate": 0.0005, "loss": 1.3240728378295898, "step": 42053 }, { "epoch": 3.3152542372881357, "grad_norm": 0.523737370967865, "learning_rate": 0.0005, "loss": 1.1063584089279175, "step": 42054 }, { "epoch": 3.3153330705557744, "grad_norm": 0.47290441393852234, "learning_rate": 0.0005, "loss": 1.2274081707000732, "step": 42055 }, { "epoch": 3.3154119038234136, "grad_norm": 0.5467363595962524, "learning_rate": 0.0005, "loss": 1.224570870399475, "step": 42056 }, { "epoch": 3.3154907370910522, "grad_norm": 0.39457911252975464, "learning_rate": 0.0005, "loss": 1.1740319728851318, "step": 42057 }, { "epoch": 3.3155695703586914, "grad_norm": 0.5379483699798584, "learning_rate": 0.0005, "loss": 1.3219150304794312, "step": 42058 }, { "epoch": 3.31564840362633, "grad_norm": 0.44933637976646423, "learning_rate": 0.0005, "loss": 1.128340482711792, "step": 42059 }, { "epoch": 3.315727236893969, "grad_norm": 0.4122238755226135, "learning_rate": 0.0005, "loss": 1.1354057788848877, "step": 42060 }, { "epoch": 3.3158060701616083, "grad_norm": 0.4839823842048645, "learning_rate": 0.0005, "loss": 1.1891207695007324, "step": 42061 }, { "epoch": 3.315884903429247, "grad_norm": 0.48454755544662476, "learning_rate": 0.0005, "loss": 1.1049798727035522, "step": 42062 }, { "epoch": 3.315963736696886, "grad_norm": 0.5579082369804382, "learning_rate": 0.0005, "loss": 1.5350348949432373, "step": 42063 }, { "epoch": 3.316042569964525, "grad_norm": 0.4369601011276245, "learning_rate": 0.0005, "loss": 1.1575684547424316, "step": 42064 }, { "epoch": 3.316121403232164, "grad_norm": 0.4358580708503723, "learning_rate": 0.0005, "loss": 1.3550388813018799, "step": 42065 }, { "epoch": 3.316200236499803, "grad_norm": 0.40494534373283386, "learning_rate": 0.0005, "loss": 1.1811935901641846, "step": 42066 }, { "epoch": 3.316279069767442, "grad_norm": 0.45669984817504883, "learning_rate": 0.0005, "loss": 1.2854387760162354, "step": 42067 }, { "epoch": 3.316357903035081, "grad_norm": 0.5090774893760681, "learning_rate": 0.0005, "loss": 1.2425565719604492, "step": 42068 }, { "epoch": 3.3164367363027196, "grad_norm": 0.4925302267074585, "learning_rate": 0.0005, "loss": 1.4015600681304932, "step": 42069 }, { "epoch": 3.3165155695703588, "grad_norm": 0.4505061209201813, "learning_rate": 0.0005, "loss": 1.213371753692627, "step": 42070 }, { "epoch": 3.3165944028379974, "grad_norm": 0.47871384024620056, "learning_rate": 0.0005, "loss": 1.19317626953125, "step": 42071 }, { "epoch": 3.3166732361056366, "grad_norm": 0.4439511001110077, "learning_rate": 0.0005, "loss": 1.2555702924728394, "step": 42072 }, { "epoch": 3.3167520693732757, "grad_norm": 0.47599169611930847, "learning_rate": 0.0005, "loss": 1.1299591064453125, "step": 42073 }, { "epoch": 3.3168309026409144, "grad_norm": 0.49568822979927063, "learning_rate": 0.0005, "loss": 1.4529306888580322, "step": 42074 }, { "epoch": 3.3169097359085535, "grad_norm": 0.39446014165878296, "learning_rate": 0.0005, "loss": 1.2847532033920288, "step": 42075 }, { "epoch": 3.3169885691761922, "grad_norm": 0.4762403964996338, "learning_rate": 0.0005, "loss": 1.3393762111663818, "step": 42076 }, { "epoch": 3.3170674024438314, "grad_norm": 0.4006158709526062, "learning_rate": 0.0005, "loss": 1.184471607208252, "step": 42077 }, { "epoch": 3.31714623571147, "grad_norm": 0.5199236869812012, "learning_rate": 0.0005, "loss": 1.4448776245117188, "step": 42078 }, { "epoch": 3.317225068979109, "grad_norm": 0.4990435242652893, "learning_rate": 0.0005, "loss": 1.405211329460144, "step": 42079 }, { "epoch": 3.3173039022467483, "grad_norm": 0.5190753936767578, "learning_rate": 0.0005, "loss": 1.2857837677001953, "step": 42080 }, { "epoch": 3.317382735514387, "grad_norm": 0.4645136892795563, "learning_rate": 0.0005, "loss": 1.162135362625122, "step": 42081 }, { "epoch": 3.317461568782026, "grad_norm": 0.5032514333724976, "learning_rate": 0.0005, "loss": 1.3658720254898071, "step": 42082 }, { "epoch": 3.317540402049665, "grad_norm": 0.5286292433738708, "learning_rate": 0.0005, "loss": 1.3613314628601074, "step": 42083 }, { "epoch": 3.317619235317304, "grad_norm": 0.3692351281642914, "learning_rate": 0.0005, "loss": 1.0680969953536987, "step": 42084 }, { "epoch": 3.317698068584943, "grad_norm": 0.5055104494094849, "learning_rate": 0.0005, "loss": 1.2173101902008057, "step": 42085 }, { "epoch": 3.317776901852582, "grad_norm": 0.5256898403167725, "learning_rate": 0.0005, "loss": 1.1767159700393677, "step": 42086 }, { "epoch": 3.317855735120221, "grad_norm": 0.5595043897628784, "learning_rate": 0.0005, "loss": 1.3974213600158691, "step": 42087 }, { "epoch": 3.3179345683878596, "grad_norm": 0.49472689628601074, "learning_rate": 0.0005, "loss": 1.2629425525665283, "step": 42088 }, { "epoch": 3.3180134016554987, "grad_norm": 0.6350935101509094, "learning_rate": 0.0005, "loss": 1.322073221206665, "step": 42089 }, { "epoch": 3.3180922349231374, "grad_norm": 0.43218186497688293, "learning_rate": 0.0005, "loss": 1.2243266105651855, "step": 42090 }, { "epoch": 3.3181710681907766, "grad_norm": 0.5624625086784363, "learning_rate": 0.0005, "loss": 1.2514468431472778, "step": 42091 }, { "epoch": 3.3182499014584153, "grad_norm": 0.567854642868042, "learning_rate": 0.0005, "loss": 1.382500171661377, "step": 42092 }, { "epoch": 3.3183287347260544, "grad_norm": 0.4359531104564667, "learning_rate": 0.0005, "loss": 1.1610219478607178, "step": 42093 }, { "epoch": 3.3184075679936935, "grad_norm": 0.5346300601959229, "learning_rate": 0.0005, "loss": 1.2617902755737305, "step": 42094 }, { "epoch": 3.318486401261332, "grad_norm": 0.4279015064239502, "learning_rate": 0.0005, "loss": 1.0447676181793213, "step": 42095 }, { "epoch": 3.3185652345289713, "grad_norm": 0.45838668942451477, "learning_rate": 0.0005, "loss": 1.0733363628387451, "step": 42096 }, { "epoch": 3.31864406779661, "grad_norm": 0.4354463517665863, "learning_rate": 0.0005, "loss": 1.1366326808929443, "step": 42097 }, { "epoch": 3.318722901064249, "grad_norm": 0.6695286631584167, "learning_rate": 0.0005, "loss": 1.32099449634552, "step": 42098 }, { "epoch": 3.3188017343318883, "grad_norm": 0.6121382117271423, "learning_rate": 0.0005, "loss": 1.3724349737167358, "step": 42099 }, { "epoch": 3.318880567599527, "grad_norm": 0.5747929811477661, "learning_rate": 0.0005, "loss": 1.286062240600586, "step": 42100 }, { "epoch": 3.318959400867166, "grad_norm": 0.5155048370361328, "learning_rate": 0.0005, "loss": 1.3649917840957642, "step": 42101 }, { "epoch": 3.319038234134805, "grad_norm": 0.46782490611076355, "learning_rate": 0.0005, "loss": 1.1870436668395996, "step": 42102 }, { "epoch": 3.319117067402444, "grad_norm": 0.66013103723526, "learning_rate": 0.0005, "loss": 1.452290654182434, "step": 42103 }, { "epoch": 3.3191959006700826, "grad_norm": 0.45484253764152527, "learning_rate": 0.0005, "loss": 1.2777657508850098, "step": 42104 }, { "epoch": 3.3192747339377218, "grad_norm": 0.758335530757904, "learning_rate": 0.0005, "loss": 1.3226168155670166, "step": 42105 }, { "epoch": 3.3193535672053605, "grad_norm": 0.6220456957817078, "learning_rate": 0.0005, "loss": 1.3944053649902344, "step": 42106 }, { "epoch": 3.3194324004729996, "grad_norm": 0.4557458460330963, "learning_rate": 0.0005, "loss": 1.2070032358169556, "step": 42107 }, { "epoch": 3.3195112337406387, "grad_norm": 0.4548094570636749, "learning_rate": 0.0005, "loss": 1.238399863243103, "step": 42108 }, { "epoch": 3.3195900670082774, "grad_norm": 0.532184898853302, "learning_rate": 0.0005, "loss": 1.447709560394287, "step": 42109 }, { "epoch": 3.3196689002759165, "grad_norm": 0.5296492576599121, "learning_rate": 0.0005, "loss": 1.3333920240402222, "step": 42110 }, { "epoch": 3.3197477335435552, "grad_norm": 0.4634665548801422, "learning_rate": 0.0005, "loss": 1.118072509765625, "step": 42111 }, { "epoch": 3.3198265668111944, "grad_norm": 0.46573758125305176, "learning_rate": 0.0005, "loss": 1.1482213735580444, "step": 42112 }, { "epoch": 3.3199054000788335, "grad_norm": 0.6417176127433777, "learning_rate": 0.0005, "loss": 1.131424903869629, "step": 42113 }, { "epoch": 3.319984233346472, "grad_norm": 0.5035547614097595, "learning_rate": 0.0005, "loss": 1.3146789073944092, "step": 42114 }, { "epoch": 3.3200630666141113, "grad_norm": 0.5941991209983826, "learning_rate": 0.0005, "loss": 1.4864251613616943, "step": 42115 }, { "epoch": 3.32014189988175, "grad_norm": 0.4956797659397125, "learning_rate": 0.0005, "loss": 1.4231499433517456, "step": 42116 }, { "epoch": 3.320220733149389, "grad_norm": 0.3419398069381714, "learning_rate": 0.0005, "loss": 1.161808729171753, "step": 42117 }, { "epoch": 3.320299566417028, "grad_norm": 0.483937531709671, "learning_rate": 0.0005, "loss": 1.3072105646133423, "step": 42118 }, { "epoch": 3.320378399684667, "grad_norm": 0.4585341215133667, "learning_rate": 0.0005, "loss": 1.17162024974823, "step": 42119 }, { "epoch": 3.3204572329523057, "grad_norm": 0.4745255708694458, "learning_rate": 0.0005, "loss": 1.1874366998672485, "step": 42120 }, { "epoch": 3.320536066219945, "grad_norm": 0.5510884523391724, "learning_rate": 0.0005, "loss": 1.2000329494476318, "step": 42121 }, { "epoch": 3.320614899487584, "grad_norm": 0.45206114649772644, "learning_rate": 0.0005, "loss": 1.1593592166900635, "step": 42122 }, { "epoch": 3.3206937327552226, "grad_norm": 0.40807202458381653, "learning_rate": 0.0005, "loss": 1.1925169229507446, "step": 42123 }, { "epoch": 3.3207725660228617, "grad_norm": 0.46859633922576904, "learning_rate": 0.0005, "loss": 1.150563359260559, "step": 42124 }, { "epoch": 3.3208513992905004, "grad_norm": 0.5822025537490845, "learning_rate": 0.0005, "loss": 1.3949484825134277, "step": 42125 }, { "epoch": 3.3209302325581396, "grad_norm": 0.49455803632736206, "learning_rate": 0.0005, "loss": 1.193544864654541, "step": 42126 }, { "epoch": 3.3210090658257787, "grad_norm": 0.6668804883956909, "learning_rate": 0.0005, "loss": 1.5369877815246582, "step": 42127 }, { "epoch": 3.3210878990934174, "grad_norm": 0.41667360067367554, "learning_rate": 0.0005, "loss": 1.1342597007751465, "step": 42128 }, { "epoch": 3.3211667323610565, "grad_norm": 0.4093687832355499, "learning_rate": 0.0005, "loss": 1.116858720779419, "step": 42129 }, { "epoch": 3.321245565628695, "grad_norm": 0.45306330919265747, "learning_rate": 0.0005, "loss": 1.0070111751556396, "step": 42130 }, { "epoch": 3.3213243988963344, "grad_norm": 0.3963058590888977, "learning_rate": 0.0005, "loss": 1.1883008480072021, "step": 42131 }, { "epoch": 3.321403232163973, "grad_norm": 0.41844677925109863, "learning_rate": 0.0005, "loss": 1.2449140548706055, "step": 42132 }, { "epoch": 3.321482065431612, "grad_norm": 0.44374075531959534, "learning_rate": 0.0005, "loss": 1.2803889513015747, "step": 42133 }, { "epoch": 3.321560898699251, "grad_norm": 0.528725266456604, "learning_rate": 0.0005, "loss": 1.2469971179962158, "step": 42134 }, { "epoch": 3.32163973196689, "grad_norm": 0.5496764779090881, "learning_rate": 0.0005, "loss": 1.2932672500610352, "step": 42135 }, { "epoch": 3.321718565234529, "grad_norm": 0.553315281867981, "learning_rate": 0.0005, "loss": 1.1853744983673096, "step": 42136 }, { "epoch": 3.321797398502168, "grad_norm": 0.5404039621353149, "learning_rate": 0.0005, "loss": 1.4325556755065918, "step": 42137 }, { "epoch": 3.321876231769807, "grad_norm": 0.4402928054332733, "learning_rate": 0.0005, "loss": 1.179224967956543, "step": 42138 }, { "epoch": 3.3219550650374456, "grad_norm": 0.4953222870826721, "learning_rate": 0.0005, "loss": 1.2268791198730469, "step": 42139 }, { "epoch": 3.3220338983050848, "grad_norm": 0.4578770101070404, "learning_rate": 0.0005, "loss": 1.0550992488861084, "step": 42140 }, { "epoch": 3.322112731572724, "grad_norm": 0.4667879343032837, "learning_rate": 0.0005, "loss": 1.2203859090805054, "step": 42141 }, { "epoch": 3.3221915648403626, "grad_norm": 0.48073264956474304, "learning_rate": 0.0005, "loss": 1.2101454734802246, "step": 42142 }, { "epoch": 3.3222703981080017, "grad_norm": 0.4990485608577728, "learning_rate": 0.0005, "loss": 1.3432958126068115, "step": 42143 }, { "epoch": 3.3223492313756404, "grad_norm": 0.4720006585121155, "learning_rate": 0.0005, "loss": 1.2401392459869385, "step": 42144 }, { "epoch": 3.3224280646432796, "grad_norm": 0.5743334293365479, "learning_rate": 0.0005, "loss": 1.5166733264923096, "step": 42145 }, { "epoch": 3.3225068979109182, "grad_norm": 0.5792993307113647, "learning_rate": 0.0005, "loss": 1.3866502046585083, "step": 42146 }, { "epoch": 3.3225857311785574, "grad_norm": 0.5154690146446228, "learning_rate": 0.0005, "loss": 1.2347509860992432, "step": 42147 }, { "epoch": 3.322664564446196, "grad_norm": 0.41572538018226624, "learning_rate": 0.0005, "loss": 1.134830117225647, "step": 42148 }, { "epoch": 3.322743397713835, "grad_norm": 0.46319347620010376, "learning_rate": 0.0005, "loss": 1.3032255172729492, "step": 42149 }, { "epoch": 3.3228222309814743, "grad_norm": 0.5375227332115173, "learning_rate": 0.0005, "loss": 1.3466715812683105, "step": 42150 }, { "epoch": 3.322901064249113, "grad_norm": 0.41697633266448975, "learning_rate": 0.0005, "loss": 1.2475707530975342, "step": 42151 }, { "epoch": 3.322979897516752, "grad_norm": 0.4733806252479553, "learning_rate": 0.0005, "loss": 1.2645974159240723, "step": 42152 }, { "epoch": 3.323058730784391, "grad_norm": 0.523132860660553, "learning_rate": 0.0005, "loss": 1.2688299417495728, "step": 42153 }, { "epoch": 3.32313756405203, "grad_norm": 0.5443792343139648, "learning_rate": 0.0005, "loss": 1.3985083103179932, "step": 42154 }, { "epoch": 3.323216397319669, "grad_norm": 0.4621083438396454, "learning_rate": 0.0005, "loss": 1.0456749200820923, "step": 42155 }, { "epoch": 3.323295230587308, "grad_norm": 0.5137351751327515, "learning_rate": 0.0005, "loss": 1.3570103645324707, "step": 42156 }, { "epoch": 3.323374063854947, "grad_norm": 0.4763430655002594, "learning_rate": 0.0005, "loss": 1.2434202432632446, "step": 42157 }, { "epoch": 3.3234528971225856, "grad_norm": 0.5002345442771912, "learning_rate": 0.0005, "loss": 1.2761317491531372, "step": 42158 }, { "epoch": 3.3235317303902248, "grad_norm": 0.4914954602718353, "learning_rate": 0.0005, "loss": 1.2776670455932617, "step": 42159 }, { "epoch": 3.3236105636578634, "grad_norm": 0.4259393811225891, "learning_rate": 0.0005, "loss": 1.4076985120773315, "step": 42160 }, { "epoch": 3.3236893969255026, "grad_norm": 0.5719548463821411, "learning_rate": 0.0005, "loss": 1.353471040725708, "step": 42161 }, { "epoch": 3.3237682301931413, "grad_norm": 0.5132163166999817, "learning_rate": 0.0005, "loss": 1.2064762115478516, "step": 42162 }, { "epoch": 3.3238470634607804, "grad_norm": 0.5094379782676697, "learning_rate": 0.0005, "loss": 1.1390256881713867, "step": 42163 }, { "epoch": 3.3239258967284195, "grad_norm": 1.1971558332443237, "learning_rate": 0.0005, "loss": 1.3703858852386475, "step": 42164 }, { "epoch": 3.3240047299960582, "grad_norm": 0.5099484324455261, "learning_rate": 0.0005, "loss": 1.3820098638534546, "step": 42165 }, { "epoch": 3.3240835632636974, "grad_norm": 0.4788210690021515, "learning_rate": 0.0005, "loss": 1.4615730047225952, "step": 42166 }, { "epoch": 3.324162396531336, "grad_norm": 0.44040825963020325, "learning_rate": 0.0005, "loss": 1.1740248203277588, "step": 42167 }, { "epoch": 3.324241229798975, "grad_norm": 0.5596740245819092, "learning_rate": 0.0005, "loss": 1.333481788635254, "step": 42168 }, { "epoch": 3.3243200630666143, "grad_norm": 0.45605701208114624, "learning_rate": 0.0005, "loss": 1.2728813886642456, "step": 42169 }, { "epoch": 3.324398896334253, "grad_norm": 0.5313238501548767, "learning_rate": 0.0005, "loss": 1.171556830406189, "step": 42170 }, { "epoch": 3.324477729601892, "grad_norm": 0.49046602845191956, "learning_rate": 0.0005, "loss": 1.3319265842437744, "step": 42171 }, { "epoch": 3.324556562869531, "grad_norm": 0.5642704963684082, "learning_rate": 0.0005, "loss": 1.4214560985565186, "step": 42172 }, { "epoch": 3.32463539613717, "grad_norm": 0.47142231464385986, "learning_rate": 0.0005, "loss": 1.1560016870498657, "step": 42173 }, { "epoch": 3.3247142294048087, "grad_norm": 0.5162842273712158, "learning_rate": 0.0005, "loss": 1.3131989240646362, "step": 42174 }, { "epoch": 3.324793062672448, "grad_norm": 2.574761390686035, "learning_rate": 0.0005, "loss": 1.1803330183029175, "step": 42175 }, { "epoch": 3.3248718959400865, "grad_norm": 0.5520728230476379, "learning_rate": 0.0005, "loss": 1.3818055391311646, "step": 42176 }, { "epoch": 3.3249507292077256, "grad_norm": 0.5236502289772034, "learning_rate": 0.0005, "loss": 1.2540500164031982, "step": 42177 }, { "epoch": 3.3250295624753647, "grad_norm": 0.6609927415847778, "learning_rate": 0.0005, "loss": 1.4637362957000732, "step": 42178 }, { "epoch": 3.3251083957430034, "grad_norm": 0.5586294531822205, "learning_rate": 0.0005, "loss": 1.26377272605896, "step": 42179 }, { "epoch": 3.3251872290106426, "grad_norm": 0.4611533284187317, "learning_rate": 0.0005, "loss": 1.166419267654419, "step": 42180 }, { "epoch": 3.3252660622782813, "grad_norm": 0.42325904965400696, "learning_rate": 0.0005, "loss": 1.3518478870391846, "step": 42181 }, { "epoch": 3.3253448955459204, "grad_norm": 0.4790531098842621, "learning_rate": 0.0005, "loss": 1.263221025466919, "step": 42182 }, { "epoch": 3.3254237288135595, "grad_norm": 0.5113536715507507, "learning_rate": 0.0005, "loss": 1.3606680631637573, "step": 42183 }, { "epoch": 3.325502562081198, "grad_norm": 0.49166131019592285, "learning_rate": 0.0005, "loss": 1.2883169651031494, "step": 42184 }, { "epoch": 3.3255813953488373, "grad_norm": 0.5126718282699585, "learning_rate": 0.0005, "loss": 1.3218257427215576, "step": 42185 }, { "epoch": 3.325660228616476, "grad_norm": 0.48624590039253235, "learning_rate": 0.0005, "loss": 1.1797958612442017, "step": 42186 }, { "epoch": 3.325739061884115, "grad_norm": 0.398248553276062, "learning_rate": 0.0005, "loss": 1.1734217405319214, "step": 42187 }, { "epoch": 3.325817895151754, "grad_norm": 0.4266739785671234, "learning_rate": 0.0005, "loss": 1.2345236539840698, "step": 42188 }, { "epoch": 3.325896728419393, "grad_norm": 0.46482110023498535, "learning_rate": 0.0005, "loss": 1.3297827243804932, "step": 42189 }, { "epoch": 3.3259755616870317, "grad_norm": 0.5317726731300354, "learning_rate": 0.0005, "loss": 1.3292436599731445, "step": 42190 }, { "epoch": 3.326054394954671, "grad_norm": 0.531340479850769, "learning_rate": 0.0005, "loss": 1.1863067150115967, "step": 42191 }, { "epoch": 3.32613322822231, "grad_norm": 0.40181219577789307, "learning_rate": 0.0005, "loss": 1.0928785800933838, "step": 42192 }, { "epoch": 3.3262120614899486, "grad_norm": 0.46025869250297546, "learning_rate": 0.0005, "loss": 1.3192596435546875, "step": 42193 }, { "epoch": 3.3262908947575878, "grad_norm": 0.568795919418335, "learning_rate": 0.0005, "loss": 1.1783537864685059, "step": 42194 }, { "epoch": 3.3263697280252265, "grad_norm": 0.6431893706321716, "learning_rate": 0.0005, "loss": 1.437475323677063, "step": 42195 }, { "epoch": 3.3264485612928656, "grad_norm": 0.4683230221271515, "learning_rate": 0.0005, "loss": 1.3469717502593994, "step": 42196 }, { "epoch": 3.3265273945605047, "grad_norm": 0.4684154689311981, "learning_rate": 0.0005, "loss": 1.2855117321014404, "step": 42197 }, { "epoch": 3.3266062278281434, "grad_norm": 0.47229671478271484, "learning_rate": 0.0005, "loss": 1.193702220916748, "step": 42198 }, { "epoch": 3.3266850610957825, "grad_norm": 0.41941624879837036, "learning_rate": 0.0005, "loss": 1.2320594787597656, "step": 42199 }, { "epoch": 3.3267638943634212, "grad_norm": 0.49238982796669006, "learning_rate": 0.0005, "loss": 1.1411151885986328, "step": 42200 }, { "epoch": 3.3268427276310604, "grad_norm": 0.4019154906272888, "learning_rate": 0.0005, "loss": 1.2003085613250732, "step": 42201 }, { "epoch": 3.326921560898699, "grad_norm": 0.5479142665863037, "learning_rate": 0.0005, "loss": 1.2574832439422607, "step": 42202 }, { "epoch": 3.327000394166338, "grad_norm": 0.4918946623802185, "learning_rate": 0.0005, "loss": 1.3746013641357422, "step": 42203 }, { "epoch": 3.3270792274339773, "grad_norm": 0.4436609745025635, "learning_rate": 0.0005, "loss": 1.1938402652740479, "step": 42204 }, { "epoch": 3.327158060701616, "grad_norm": 0.5289738178253174, "learning_rate": 0.0005, "loss": 1.2575825452804565, "step": 42205 }, { "epoch": 3.327236893969255, "grad_norm": 0.48405227065086365, "learning_rate": 0.0005, "loss": 1.3304126262664795, "step": 42206 }, { "epoch": 3.327315727236894, "grad_norm": 0.45047956705093384, "learning_rate": 0.0005, "loss": 1.2103623151779175, "step": 42207 }, { "epoch": 3.327394560504533, "grad_norm": 0.5083404779434204, "learning_rate": 0.0005, "loss": 1.29458487033844, "step": 42208 }, { "epoch": 3.3274733937721717, "grad_norm": 0.568210244178772, "learning_rate": 0.0005, "loss": 1.2993121147155762, "step": 42209 }, { "epoch": 3.327552227039811, "grad_norm": 0.5633370280265808, "learning_rate": 0.0005, "loss": 1.3359148502349854, "step": 42210 }, { "epoch": 3.32763106030745, "grad_norm": 0.40285825729370117, "learning_rate": 0.0005, "loss": 1.198042392730713, "step": 42211 }, { "epoch": 3.3277098935750886, "grad_norm": 0.4592476189136505, "learning_rate": 0.0005, "loss": 1.1127309799194336, "step": 42212 }, { "epoch": 3.3277887268427278, "grad_norm": 0.5447374582290649, "learning_rate": 0.0005, "loss": 1.2819888591766357, "step": 42213 }, { "epoch": 3.3278675601103664, "grad_norm": 0.3995102047920227, "learning_rate": 0.0005, "loss": 1.1249902248382568, "step": 42214 }, { "epoch": 3.3279463933780056, "grad_norm": 0.37979021668434143, "learning_rate": 0.0005, "loss": 1.0855607986450195, "step": 42215 }, { "epoch": 3.3280252266456447, "grad_norm": 0.46279481053352356, "learning_rate": 0.0005, "loss": 1.318159818649292, "step": 42216 }, { "epoch": 3.3281040599132834, "grad_norm": 0.5065648555755615, "learning_rate": 0.0005, "loss": 1.2049132585525513, "step": 42217 }, { "epoch": 3.3281828931809225, "grad_norm": 0.5493101477622986, "learning_rate": 0.0005, "loss": 1.200376033782959, "step": 42218 }, { "epoch": 3.328261726448561, "grad_norm": 0.5410277247428894, "learning_rate": 0.0005, "loss": 1.3236072063446045, "step": 42219 }, { "epoch": 3.3283405597162004, "grad_norm": 0.40959250926971436, "learning_rate": 0.0005, "loss": 1.2226734161376953, "step": 42220 }, { "epoch": 3.328419392983839, "grad_norm": 0.39543333649635315, "learning_rate": 0.0005, "loss": 1.2355859279632568, "step": 42221 }, { "epoch": 3.328498226251478, "grad_norm": 0.5061740875244141, "learning_rate": 0.0005, "loss": 1.343308448791504, "step": 42222 }, { "epoch": 3.328577059519117, "grad_norm": 0.4256511330604553, "learning_rate": 0.0005, "loss": 1.194975733757019, "step": 42223 }, { "epoch": 3.328655892786756, "grad_norm": 0.44547128677368164, "learning_rate": 0.0005, "loss": 1.2121200561523438, "step": 42224 }, { "epoch": 3.328734726054395, "grad_norm": 0.48664379119873047, "learning_rate": 0.0005, "loss": 1.3047168254852295, "step": 42225 }, { "epoch": 3.328813559322034, "grad_norm": 0.7388359904289246, "learning_rate": 0.0005, "loss": 1.3603265285491943, "step": 42226 }, { "epoch": 3.328892392589673, "grad_norm": 0.5351418256759644, "learning_rate": 0.0005, "loss": 1.3588883876800537, "step": 42227 }, { "epoch": 3.3289712258573116, "grad_norm": 0.4605548679828644, "learning_rate": 0.0005, "loss": 1.2319798469543457, "step": 42228 }, { "epoch": 3.3290500591249508, "grad_norm": 0.5562503933906555, "learning_rate": 0.0005, "loss": 1.2283413410186768, "step": 42229 }, { "epoch": 3.32912889239259, "grad_norm": 0.611169159412384, "learning_rate": 0.0005, "loss": 1.062504529953003, "step": 42230 }, { "epoch": 3.3292077256602286, "grad_norm": 0.4314095377922058, "learning_rate": 0.0005, "loss": 1.256737232208252, "step": 42231 }, { "epoch": 3.3292865589278677, "grad_norm": 0.4202043116092682, "learning_rate": 0.0005, "loss": 1.2570092678070068, "step": 42232 }, { "epoch": 3.3293653921955064, "grad_norm": 0.48014649748802185, "learning_rate": 0.0005, "loss": 1.2934906482696533, "step": 42233 }, { "epoch": 3.3294442254631456, "grad_norm": 0.501439094543457, "learning_rate": 0.0005, "loss": 1.238781452178955, "step": 42234 }, { "epoch": 3.3295230587307842, "grad_norm": 0.456326425075531, "learning_rate": 0.0005, "loss": 1.2583564519882202, "step": 42235 }, { "epoch": 3.3296018919984234, "grad_norm": 0.5445266962051392, "learning_rate": 0.0005, "loss": 1.1591475009918213, "step": 42236 }, { "epoch": 3.329680725266062, "grad_norm": 0.40431004762649536, "learning_rate": 0.0005, "loss": 1.173210859298706, "step": 42237 }, { "epoch": 3.329759558533701, "grad_norm": 0.53940749168396, "learning_rate": 0.0005, "loss": 1.2835822105407715, "step": 42238 }, { "epoch": 3.3298383918013403, "grad_norm": 0.5279499292373657, "learning_rate": 0.0005, "loss": 1.2432928085327148, "step": 42239 }, { "epoch": 3.329917225068979, "grad_norm": 0.4737377464771271, "learning_rate": 0.0005, "loss": 1.351858139038086, "step": 42240 }, { "epoch": 3.329996058336618, "grad_norm": 0.5369511246681213, "learning_rate": 0.0005, "loss": 1.4164717197418213, "step": 42241 }, { "epoch": 3.330074891604257, "grad_norm": 0.4359913170337677, "learning_rate": 0.0005, "loss": 1.2507593631744385, "step": 42242 }, { "epoch": 3.330153724871896, "grad_norm": 0.7104524374008179, "learning_rate": 0.0005, "loss": 1.1777255535125732, "step": 42243 }, { "epoch": 3.330232558139535, "grad_norm": 0.5561316013336182, "learning_rate": 0.0005, "loss": 1.293837308883667, "step": 42244 }, { "epoch": 3.330311391407174, "grad_norm": 0.549769937992096, "learning_rate": 0.0005, "loss": 1.206296682357788, "step": 42245 }, { "epoch": 3.330390224674813, "grad_norm": 0.6078068017959595, "learning_rate": 0.0005, "loss": 1.25114905834198, "step": 42246 }, { "epoch": 3.3304690579424516, "grad_norm": 0.4937349855899811, "learning_rate": 0.0005, "loss": 1.3136364221572876, "step": 42247 }, { "epoch": 3.3305478912100908, "grad_norm": 0.4891364872455597, "learning_rate": 0.0005, "loss": 1.2117159366607666, "step": 42248 }, { "epoch": 3.3306267244777295, "grad_norm": 0.4264936149120331, "learning_rate": 0.0005, "loss": 1.0185937881469727, "step": 42249 }, { "epoch": 3.3307055577453686, "grad_norm": 0.48137223720550537, "learning_rate": 0.0005, "loss": 1.2345201969146729, "step": 42250 }, { "epoch": 3.3307843910130073, "grad_norm": 0.530651867389679, "learning_rate": 0.0005, "loss": 1.3720697164535522, "step": 42251 }, { "epoch": 3.3308632242806464, "grad_norm": 0.42562854290008545, "learning_rate": 0.0005, "loss": 1.2100961208343506, "step": 42252 }, { "epoch": 3.3309420575482855, "grad_norm": 0.687297523021698, "learning_rate": 0.0005, "loss": 1.4963421821594238, "step": 42253 }, { "epoch": 3.3310208908159242, "grad_norm": 0.5615897178649902, "learning_rate": 0.0005, "loss": 1.2495272159576416, "step": 42254 }, { "epoch": 3.3310997240835634, "grad_norm": 0.5119183659553528, "learning_rate": 0.0005, "loss": 1.3077492713928223, "step": 42255 }, { "epoch": 3.331178557351202, "grad_norm": 0.42091527581214905, "learning_rate": 0.0005, "loss": 1.1751474142074585, "step": 42256 }, { "epoch": 3.331257390618841, "grad_norm": 0.601911723613739, "learning_rate": 0.0005, "loss": 1.1767317056655884, "step": 42257 }, { "epoch": 3.3313362238864803, "grad_norm": 0.3993307948112488, "learning_rate": 0.0005, "loss": 1.120989203453064, "step": 42258 }, { "epoch": 3.331415057154119, "grad_norm": 0.5039916634559631, "learning_rate": 0.0005, "loss": 1.369771957397461, "step": 42259 }, { "epoch": 3.331493890421758, "grad_norm": 0.41826483607292175, "learning_rate": 0.0005, "loss": 1.1753864288330078, "step": 42260 }, { "epoch": 3.331572723689397, "grad_norm": 0.4236771762371063, "learning_rate": 0.0005, "loss": 1.0396757125854492, "step": 42261 }, { "epoch": 3.331651556957036, "grad_norm": 1.5474700927734375, "learning_rate": 0.0005, "loss": 1.1936489343643188, "step": 42262 }, { "epoch": 3.3317303902246747, "grad_norm": 0.6304023861885071, "learning_rate": 0.0005, "loss": 1.3723421096801758, "step": 42263 }, { "epoch": 3.331809223492314, "grad_norm": 0.4982881247997284, "learning_rate": 0.0005, "loss": 1.1837102174758911, "step": 42264 }, { "epoch": 3.3318880567599525, "grad_norm": 0.7976506352424622, "learning_rate": 0.0005, "loss": 1.196420431137085, "step": 42265 }, { "epoch": 3.3319668900275916, "grad_norm": 0.44261619448661804, "learning_rate": 0.0005, "loss": 1.2688792943954468, "step": 42266 }, { "epoch": 3.3320457232952307, "grad_norm": 0.5183844566345215, "learning_rate": 0.0005, "loss": 1.3240166902542114, "step": 42267 }, { "epoch": 3.3321245565628694, "grad_norm": 0.7651805877685547, "learning_rate": 0.0005, "loss": 1.5175538063049316, "step": 42268 }, { "epoch": 3.3322033898305086, "grad_norm": 0.38399359583854675, "learning_rate": 0.0005, "loss": 1.1154500246047974, "step": 42269 }, { "epoch": 3.3322822230981473, "grad_norm": 0.5029155611991882, "learning_rate": 0.0005, "loss": 1.357276201248169, "step": 42270 }, { "epoch": 3.3323610563657864, "grad_norm": 0.5076587796211243, "learning_rate": 0.0005, "loss": 1.3063199520111084, "step": 42271 }, { "epoch": 3.3324398896334255, "grad_norm": 0.44968247413635254, "learning_rate": 0.0005, "loss": 1.1075321435928345, "step": 42272 }, { "epoch": 3.332518722901064, "grad_norm": 0.6601288914680481, "learning_rate": 0.0005, "loss": 1.3286265134811401, "step": 42273 }, { "epoch": 3.3325975561687033, "grad_norm": 0.49932318925857544, "learning_rate": 0.0005, "loss": 1.2515110969543457, "step": 42274 }, { "epoch": 3.332676389436342, "grad_norm": 0.562858521938324, "learning_rate": 0.0005, "loss": 1.3714830875396729, "step": 42275 }, { "epoch": 3.332755222703981, "grad_norm": 0.5437076687812805, "learning_rate": 0.0005, "loss": 1.3896946907043457, "step": 42276 }, { "epoch": 3.33283405597162, "grad_norm": 0.5428492426872253, "learning_rate": 0.0005, "loss": 1.3316963911056519, "step": 42277 }, { "epoch": 3.332912889239259, "grad_norm": 0.5076978206634521, "learning_rate": 0.0005, "loss": 1.154998540878296, "step": 42278 }, { "epoch": 3.3329917225068977, "grad_norm": 1.037615418434143, "learning_rate": 0.0005, "loss": 1.2670845985412598, "step": 42279 }, { "epoch": 3.333070555774537, "grad_norm": 0.472928911447525, "learning_rate": 0.0005, "loss": 1.2224481105804443, "step": 42280 }, { "epoch": 3.333149389042176, "grad_norm": 0.5329033732414246, "learning_rate": 0.0005, "loss": 1.291746973991394, "step": 42281 }, { "epoch": 3.3332282223098146, "grad_norm": 0.4052448868751526, "learning_rate": 0.0005, "loss": 1.0560448169708252, "step": 42282 }, { "epoch": 3.3333070555774538, "grad_norm": 0.6118029952049255, "learning_rate": 0.0005, "loss": 1.2567455768585205, "step": 42283 }, { "epoch": 3.3333858888450925, "grad_norm": 0.5492368936538696, "learning_rate": 0.0005, "loss": 1.2795579433441162, "step": 42284 }, { "epoch": 3.3334647221127316, "grad_norm": 0.466457724571228, "learning_rate": 0.0005, "loss": 1.1656416654586792, "step": 42285 }, { "epoch": 3.3335435553803707, "grad_norm": 0.5048523545265198, "learning_rate": 0.0005, "loss": 1.3527199029922485, "step": 42286 }, { "epoch": 3.3336223886480094, "grad_norm": 0.5270387530326843, "learning_rate": 0.0005, "loss": 1.2261039018630981, "step": 42287 }, { "epoch": 3.3337012219156485, "grad_norm": 0.6695361733436584, "learning_rate": 0.0005, "loss": 1.2102124691009521, "step": 42288 }, { "epoch": 3.3337800551832872, "grad_norm": 0.5905041098594666, "learning_rate": 0.0005, "loss": 1.3378088474273682, "step": 42289 }, { "epoch": 3.3338588884509264, "grad_norm": 0.4577033817768097, "learning_rate": 0.0005, "loss": 1.3433167934417725, "step": 42290 }, { "epoch": 3.333937721718565, "grad_norm": 0.6330180168151855, "learning_rate": 0.0005, "loss": 1.2771339416503906, "step": 42291 }, { "epoch": 3.334016554986204, "grad_norm": 0.5297958850860596, "learning_rate": 0.0005, "loss": 1.2558749914169312, "step": 42292 }, { "epoch": 3.334095388253843, "grad_norm": 0.5025619864463806, "learning_rate": 0.0005, "loss": 1.233741283416748, "step": 42293 }, { "epoch": 3.334174221521482, "grad_norm": 0.44977277517318726, "learning_rate": 0.0005, "loss": 1.2361891269683838, "step": 42294 }, { "epoch": 3.334253054789121, "grad_norm": 0.4871993660926819, "learning_rate": 0.0005, "loss": 1.3703123331069946, "step": 42295 }, { "epoch": 3.33433188805676, "grad_norm": 0.5100767612457275, "learning_rate": 0.0005, "loss": 1.2994277477264404, "step": 42296 }, { "epoch": 3.334410721324399, "grad_norm": 0.47488686442375183, "learning_rate": 0.0005, "loss": 1.146148443222046, "step": 42297 }, { "epoch": 3.3344895545920377, "grad_norm": 0.5571632981300354, "learning_rate": 0.0005, "loss": 1.3115686178207397, "step": 42298 }, { "epoch": 3.334568387859677, "grad_norm": 0.556776225566864, "learning_rate": 0.0005, "loss": 1.4332740306854248, "step": 42299 }, { "epoch": 3.334647221127316, "grad_norm": 0.5345016717910767, "learning_rate": 0.0005, "loss": 1.21543550491333, "step": 42300 }, { "epoch": 3.3347260543949546, "grad_norm": 0.4316347539424896, "learning_rate": 0.0005, "loss": 1.340553641319275, "step": 42301 }, { "epoch": 3.3348048876625938, "grad_norm": 0.5286656022071838, "learning_rate": 0.0005, "loss": 1.331904649734497, "step": 42302 }, { "epoch": 3.3348837209302324, "grad_norm": 0.656341552734375, "learning_rate": 0.0005, "loss": 1.4276989698410034, "step": 42303 }, { "epoch": 3.3349625541978716, "grad_norm": 0.42136627435684204, "learning_rate": 0.0005, "loss": 1.3316235542297363, "step": 42304 }, { "epoch": 3.3350413874655103, "grad_norm": 0.623185396194458, "learning_rate": 0.0005, "loss": 1.418392300605774, "step": 42305 }, { "epoch": 3.3351202207331494, "grad_norm": 0.6609506011009216, "learning_rate": 0.0005, "loss": 1.2498966455459595, "step": 42306 }, { "epoch": 3.335199054000788, "grad_norm": 0.4343056380748749, "learning_rate": 0.0005, "loss": 1.1867440938949585, "step": 42307 }, { "epoch": 3.335277887268427, "grad_norm": 0.5122261047363281, "learning_rate": 0.0005, "loss": 1.2661741971969604, "step": 42308 }, { "epoch": 3.3353567205360664, "grad_norm": 0.5679110288619995, "learning_rate": 0.0005, "loss": 1.3242608308792114, "step": 42309 }, { "epoch": 3.335435553803705, "grad_norm": 0.4420528709888458, "learning_rate": 0.0005, "loss": 1.0589792728424072, "step": 42310 }, { "epoch": 3.335514387071344, "grad_norm": 0.6934396624565125, "learning_rate": 0.0005, "loss": 1.140365481376648, "step": 42311 }, { "epoch": 3.335593220338983, "grad_norm": 0.6739552021026611, "learning_rate": 0.0005, "loss": 1.3832108974456787, "step": 42312 }, { "epoch": 3.335672053606622, "grad_norm": 0.43544861674308777, "learning_rate": 0.0005, "loss": 1.2579716444015503, "step": 42313 }, { "epoch": 3.335750886874261, "grad_norm": 0.5173463821411133, "learning_rate": 0.0005, "loss": 1.1186785697937012, "step": 42314 }, { "epoch": 3.3358297201419, "grad_norm": 0.49587681889533997, "learning_rate": 0.0005, "loss": 1.102308750152588, "step": 42315 }, { "epoch": 3.335908553409539, "grad_norm": 0.473529189825058, "learning_rate": 0.0005, "loss": 1.3415409326553345, "step": 42316 }, { "epoch": 3.3359873866771776, "grad_norm": 0.5726492404937744, "learning_rate": 0.0005, "loss": 1.3601880073547363, "step": 42317 }, { "epoch": 3.336066219944817, "grad_norm": 0.5583919286727905, "learning_rate": 0.0005, "loss": 1.4404622316360474, "step": 42318 }, { "epoch": 3.3361450532124555, "grad_norm": 0.6033199429512024, "learning_rate": 0.0005, "loss": 1.2239735126495361, "step": 42319 }, { "epoch": 3.3362238864800946, "grad_norm": 0.46163085103034973, "learning_rate": 0.0005, "loss": 1.1044764518737793, "step": 42320 }, { "epoch": 3.3363027197477337, "grad_norm": 0.5797625780105591, "learning_rate": 0.0005, "loss": 1.5722637176513672, "step": 42321 }, { "epoch": 3.3363815530153724, "grad_norm": 0.5419372320175171, "learning_rate": 0.0005, "loss": 1.3171837329864502, "step": 42322 }, { "epoch": 3.3364603862830116, "grad_norm": 0.48554548621177673, "learning_rate": 0.0005, "loss": 1.2461869716644287, "step": 42323 }, { "epoch": 3.3365392195506502, "grad_norm": 0.44623881578445435, "learning_rate": 0.0005, "loss": 1.3361563682556152, "step": 42324 }, { "epoch": 3.3366180528182894, "grad_norm": 0.41275128722190857, "learning_rate": 0.0005, "loss": 1.1628727912902832, "step": 42325 }, { "epoch": 3.336696886085928, "grad_norm": 0.5271611213684082, "learning_rate": 0.0005, "loss": 1.2795451879501343, "step": 42326 }, { "epoch": 3.336775719353567, "grad_norm": 0.5101765990257263, "learning_rate": 0.0005, "loss": 1.3361804485321045, "step": 42327 }, { "epoch": 3.3368545526212063, "grad_norm": 0.46866634488105774, "learning_rate": 0.0005, "loss": 1.2945027351379395, "step": 42328 }, { "epoch": 3.336933385888845, "grad_norm": 0.52643883228302, "learning_rate": 0.0005, "loss": 1.30197274684906, "step": 42329 }, { "epoch": 3.337012219156484, "grad_norm": 0.36598047614097595, "learning_rate": 0.0005, "loss": 1.1355741024017334, "step": 42330 }, { "epoch": 3.337091052424123, "grad_norm": 0.5622857809066772, "learning_rate": 0.0005, "loss": 1.2285518646240234, "step": 42331 }, { "epoch": 3.337169885691762, "grad_norm": 0.604636549949646, "learning_rate": 0.0005, "loss": 1.4252371788024902, "step": 42332 }, { "epoch": 3.3372487189594007, "grad_norm": 0.4009700417518616, "learning_rate": 0.0005, "loss": 1.2031214237213135, "step": 42333 }, { "epoch": 3.33732755222704, "grad_norm": 0.4130699634552002, "learning_rate": 0.0005, "loss": 1.2157409191131592, "step": 42334 }, { "epoch": 3.337406385494679, "grad_norm": 0.43089184165000916, "learning_rate": 0.0005, "loss": 1.1505380868911743, "step": 42335 }, { "epoch": 3.3374852187623176, "grad_norm": 0.40108269453048706, "learning_rate": 0.0005, "loss": 1.128105878829956, "step": 42336 }, { "epoch": 3.3375640520299568, "grad_norm": 0.5539329051971436, "learning_rate": 0.0005, "loss": 1.3582725524902344, "step": 42337 }, { "epoch": 3.3376428852975955, "grad_norm": 0.4940076172351837, "learning_rate": 0.0005, "loss": 1.2829080820083618, "step": 42338 }, { "epoch": 3.3377217185652346, "grad_norm": 0.4455713927745819, "learning_rate": 0.0005, "loss": 1.2894848585128784, "step": 42339 }, { "epoch": 3.3378005518328733, "grad_norm": 0.5064209699630737, "learning_rate": 0.0005, "loss": 1.2354154586791992, "step": 42340 }, { "epoch": 3.3378793851005124, "grad_norm": 0.407250314950943, "learning_rate": 0.0005, "loss": 1.154184103012085, "step": 42341 }, { "epoch": 3.3379582183681515, "grad_norm": 2.7040297985076904, "learning_rate": 0.0005, "loss": 1.2238914966583252, "step": 42342 }, { "epoch": 3.3380370516357902, "grad_norm": 0.5179610252380371, "learning_rate": 0.0005, "loss": 1.2279438972473145, "step": 42343 }, { "epoch": 3.3381158849034294, "grad_norm": 0.6229661703109741, "learning_rate": 0.0005, "loss": 1.1178628206253052, "step": 42344 }, { "epoch": 3.338194718171068, "grad_norm": 0.521185040473938, "learning_rate": 0.0005, "loss": 1.2161500453948975, "step": 42345 }, { "epoch": 3.338273551438707, "grad_norm": 0.6495965719223022, "learning_rate": 0.0005, "loss": 1.5724947452545166, "step": 42346 }, { "epoch": 3.3383523847063463, "grad_norm": 0.5650544762611389, "learning_rate": 0.0005, "loss": 1.1817774772644043, "step": 42347 }, { "epoch": 3.338431217973985, "grad_norm": 0.5439162254333496, "learning_rate": 0.0005, "loss": 1.1904823780059814, "step": 42348 }, { "epoch": 3.338510051241624, "grad_norm": 0.5540597438812256, "learning_rate": 0.0005, "loss": 1.2153124809265137, "step": 42349 }, { "epoch": 3.338588884509263, "grad_norm": 0.5228292346000671, "learning_rate": 0.0005, "loss": 1.3824522495269775, "step": 42350 }, { "epoch": 3.338667717776902, "grad_norm": 0.4991127550601959, "learning_rate": 0.0005, "loss": 1.3717952966690063, "step": 42351 }, { "epoch": 3.3387465510445407, "grad_norm": 0.5061830282211304, "learning_rate": 0.0005, "loss": 1.2868924140930176, "step": 42352 }, { "epoch": 3.33882538431218, "grad_norm": 0.4802427291870117, "learning_rate": 0.0005, "loss": 1.3280949592590332, "step": 42353 }, { "epoch": 3.3389042175798185, "grad_norm": 0.41707324981689453, "learning_rate": 0.0005, "loss": 1.2568037509918213, "step": 42354 }, { "epoch": 3.3389830508474576, "grad_norm": 0.4496760368347168, "learning_rate": 0.0005, "loss": 1.0602200031280518, "step": 42355 }, { "epoch": 3.3390618841150967, "grad_norm": 0.47655320167541504, "learning_rate": 0.0005, "loss": 1.2589401006698608, "step": 42356 }, { "epoch": 3.3391407173827354, "grad_norm": 0.5364562273025513, "learning_rate": 0.0005, "loss": 1.1591863632202148, "step": 42357 }, { "epoch": 3.3392195506503746, "grad_norm": 0.3736075162887573, "learning_rate": 0.0005, "loss": 1.089773416519165, "step": 42358 }, { "epoch": 3.3392983839180133, "grad_norm": 0.4727809429168701, "learning_rate": 0.0005, "loss": 1.213719367980957, "step": 42359 }, { "epoch": 3.3393772171856524, "grad_norm": 0.4996410012245178, "learning_rate": 0.0005, "loss": 1.4802345037460327, "step": 42360 }, { "epoch": 3.3394560504532915, "grad_norm": 0.4649507403373718, "learning_rate": 0.0005, "loss": 1.189777135848999, "step": 42361 }, { "epoch": 3.33953488372093, "grad_norm": 0.4371929168701172, "learning_rate": 0.0005, "loss": 1.0597186088562012, "step": 42362 }, { "epoch": 3.3396137169885693, "grad_norm": 0.4962832033634186, "learning_rate": 0.0005, "loss": 1.3198432922363281, "step": 42363 }, { "epoch": 3.339692550256208, "grad_norm": 0.6920295357704163, "learning_rate": 0.0005, "loss": 1.2377849817276, "step": 42364 }, { "epoch": 3.339771383523847, "grad_norm": 0.4531592130661011, "learning_rate": 0.0005, "loss": 1.3035938739776611, "step": 42365 }, { "epoch": 3.339850216791486, "grad_norm": 0.44040051102638245, "learning_rate": 0.0005, "loss": 1.2797948122024536, "step": 42366 }, { "epoch": 3.339929050059125, "grad_norm": 0.4556228518486023, "learning_rate": 0.0005, "loss": 1.262943983078003, "step": 42367 }, { "epoch": 3.3400078833267637, "grad_norm": 0.4459753930568695, "learning_rate": 0.0005, "loss": 1.3463001251220703, "step": 42368 }, { "epoch": 3.340086716594403, "grad_norm": 0.3994152545928955, "learning_rate": 0.0005, "loss": 1.1171319484710693, "step": 42369 }, { "epoch": 3.340165549862042, "grad_norm": 0.45524299144744873, "learning_rate": 0.0005, "loss": 1.2514017820358276, "step": 42370 }, { "epoch": 3.3402443831296806, "grad_norm": 0.4736544191837311, "learning_rate": 0.0005, "loss": 1.2493224143981934, "step": 42371 }, { "epoch": 3.3403232163973198, "grad_norm": 0.52191561460495, "learning_rate": 0.0005, "loss": 1.29662024974823, "step": 42372 }, { "epoch": 3.3404020496649585, "grad_norm": 0.49350032210350037, "learning_rate": 0.0005, "loss": 1.3177138566970825, "step": 42373 }, { "epoch": 3.3404808829325976, "grad_norm": 0.5002524852752686, "learning_rate": 0.0005, "loss": 1.3310173749923706, "step": 42374 }, { "epoch": 3.3405597162002367, "grad_norm": 0.5710030198097229, "learning_rate": 0.0005, "loss": 1.366779088973999, "step": 42375 }, { "epoch": 3.3406385494678754, "grad_norm": 0.48054423928260803, "learning_rate": 0.0005, "loss": 1.2002382278442383, "step": 42376 }, { "epoch": 3.3407173827355146, "grad_norm": 0.5783564448356628, "learning_rate": 0.0005, "loss": 1.6117873191833496, "step": 42377 }, { "epoch": 3.3407962160031532, "grad_norm": 0.41544458270072937, "learning_rate": 0.0005, "loss": 1.0803985595703125, "step": 42378 }, { "epoch": 3.3408750492707924, "grad_norm": 4.9608588218688965, "learning_rate": 0.0005, "loss": 1.212355375289917, "step": 42379 }, { "epoch": 3.340953882538431, "grad_norm": 0.5225316286087036, "learning_rate": 0.0005, "loss": 1.1854220628738403, "step": 42380 }, { "epoch": 3.34103271580607, "grad_norm": 0.4712127149105072, "learning_rate": 0.0005, "loss": 1.30283522605896, "step": 42381 }, { "epoch": 3.341111549073709, "grad_norm": 0.4892020523548126, "learning_rate": 0.0005, "loss": 1.3164708614349365, "step": 42382 }, { "epoch": 3.341190382341348, "grad_norm": 0.5178771018981934, "learning_rate": 0.0005, "loss": 1.3142402172088623, "step": 42383 }, { "epoch": 3.341269215608987, "grad_norm": 0.46798062324523926, "learning_rate": 0.0005, "loss": 1.3557038307189941, "step": 42384 }, { "epoch": 3.341348048876626, "grad_norm": 0.6425992846488953, "learning_rate": 0.0005, "loss": 1.2516509294509888, "step": 42385 }, { "epoch": 3.341426882144265, "grad_norm": 0.4733644425868988, "learning_rate": 0.0005, "loss": 1.1724681854248047, "step": 42386 }, { "epoch": 3.3415057154119037, "grad_norm": 0.4202762842178345, "learning_rate": 0.0005, "loss": 1.1760356426239014, "step": 42387 }, { "epoch": 3.341584548679543, "grad_norm": 0.5315889716148376, "learning_rate": 0.0005, "loss": 1.259069800376892, "step": 42388 }, { "epoch": 3.341663381947182, "grad_norm": 0.45407459139823914, "learning_rate": 0.0005, "loss": 1.2206450700759888, "step": 42389 }, { "epoch": 3.3417422152148206, "grad_norm": 0.498420774936676, "learning_rate": 0.0005, "loss": 1.302747368812561, "step": 42390 }, { "epoch": 3.3418210484824598, "grad_norm": 0.48985448479652405, "learning_rate": 0.0005, "loss": 1.3642699718475342, "step": 42391 }, { "epoch": 3.3418998817500984, "grad_norm": 0.6324964761734009, "learning_rate": 0.0005, "loss": 1.3936105966567993, "step": 42392 }, { "epoch": 3.3419787150177376, "grad_norm": 0.6058117151260376, "learning_rate": 0.0005, "loss": 1.2888836860656738, "step": 42393 }, { "epoch": 3.3420575482853763, "grad_norm": 0.7325007915496826, "learning_rate": 0.0005, "loss": 1.2066245079040527, "step": 42394 }, { "epoch": 3.3421363815530154, "grad_norm": 0.501948356628418, "learning_rate": 0.0005, "loss": 1.3734726905822754, "step": 42395 }, { "epoch": 3.342215214820654, "grad_norm": 0.3713783323764801, "learning_rate": 0.0005, "loss": 1.2926746606826782, "step": 42396 }, { "epoch": 3.3422940480882932, "grad_norm": 0.43322235345840454, "learning_rate": 0.0005, "loss": 1.0958142280578613, "step": 42397 }, { "epoch": 3.3423728813559324, "grad_norm": 0.47243595123291016, "learning_rate": 0.0005, "loss": 1.1566293239593506, "step": 42398 }, { "epoch": 3.342451714623571, "grad_norm": 0.46591439843177795, "learning_rate": 0.0005, "loss": 1.1206399202346802, "step": 42399 }, { "epoch": 3.34253054789121, "grad_norm": 0.550294041633606, "learning_rate": 0.0005, "loss": 1.24928879737854, "step": 42400 }, { "epoch": 3.342609381158849, "grad_norm": 0.5333355069160461, "learning_rate": 0.0005, "loss": 1.1438333988189697, "step": 42401 }, { "epoch": 3.342688214426488, "grad_norm": 0.45147132873535156, "learning_rate": 0.0005, "loss": 1.3504424095153809, "step": 42402 }, { "epoch": 3.342767047694127, "grad_norm": 0.44815799593925476, "learning_rate": 0.0005, "loss": 1.2969260215759277, "step": 42403 }, { "epoch": 3.342845880961766, "grad_norm": 0.46833494305610657, "learning_rate": 0.0005, "loss": 1.2473578453063965, "step": 42404 }, { "epoch": 3.342924714229405, "grad_norm": 0.6370537877082825, "learning_rate": 0.0005, "loss": 1.1643092632293701, "step": 42405 }, { "epoch": 3.3430035474970436, "grad_norm": 0.7064334154129028, "learning_rate": 0.0005, "loss": 1.5162333250045776, "step": 42406 }, { "epoch": 3.343082380764683, "grad_norm": 0.4985773265361786, "learning_rate": 0.0005, "loss": 1.2069661617279053, "step": 42407 }, { "epoch": 3.3431612140323215, "grad_norm": 0.45713794231414795, "learning_rate": 0.0005, "loss": 1.23038911819458, "step": 42408 }, { "epoch": 3.3432400472999606, "grad_norm": 0.570200502872467, "learning_rate": 0.0005, "loss": 1.148747444152832, "step": 42409 }, { "epoch": 3.3433188805675993, "grad_norm": 0.48538509011268616, "learning_rate": 0.0005, "loss": 1.2483354806900024, "step": 42410 }, { "epoch": 3.3433977138352384, "grad_norm": 0.46166589856147766, "learning_rate": 0.0005, "loss": 1.1938602924346924, "step": 42411 }, { "epoch": 3.3434765471028776, "grad_norm": 0.4891231656074524, "learning_rate": 0.0005, "loss": 1.2256271839141846, "step": 42412 }, { "epoch": 3.3435553803705163, "grad_norm": 0.4439363181591034, "learning_rate": 0.0005, "loss": 1.3341631889343262, "step": 42413 }, { "epoch": 3.3436342136381554, "grad_norm": 0.45214754343032837, "learning_rate": 0.0005, "loss": 1.1677935123443604, "step": 42414 }, { "epoch": 3.343713046905794, "grad_norm": 0.4525139033794403, "learning_rate": 0.0005, "loss": 1.3924683332443237, "step": 42415 }, { "epoch": 3.343791880173433, "grad_norm": 0.3705638349056244, "learning_rate": 0.0005, "loss": 1.0771714448928833, "step": 42416 }, { "epoch": 3.3438707134410723, "grad_norm": 0.5142534971237183, "learning_rate": 0.0005, "loss": 1.094417691230774, "step": 42417 }, { "epoch": 3.343949546708711, "grad_norm": 0.4063853919506073, "learning_rate": 0.0005, "loss": 1.2862999439239502, "step": 42418 }, { "epoch": 3.34402837997635, "grad_norm": 0.428224116563797, "learning_rate": 0.0005, "loss": 1.2651649713516235, "step": 42419 }, { "epoch": 3.344107213243989, "grad_norm": 0.6456630229949951, "learning_rate": 0.0005, "loss": 1.3117129802703857, "step": 42420 }, { "epoch": 3.344186046511628, "grad_norm": 0.5715801119804382, "learning_rate": 0.0005, "loss": 1.4610927104949951, "step": 42421 }, { "epoch": 3.3442648797792667, "grad_norm": 0.43698328733444214, "learning_rate": 0.0005, "loss": 1.0697516202926636, "step": 42422 }, { "epoch": 3.344343713046906, "grad_norm": 0.41934990882873535, "learning_rate": 0.0005, "loss": 1.1468124389648438, "step": 42423 }, { "epoch": 3.3444225463145445, "grad_norm": 0.47200924158096313, "learning_rate": 0.0005, "loss": 1.2154366970062256, "step": 42424 }, { "epoch": 3.3445013795821836, "grad_norm": 0.8321899771690369, "learning_rate": 0.0005, "loss": 1.0564064979553223, "step": 42425 }, { "epoch": 3.3445802128498228, "grad_norm": 0.5647403001785278, "learning_rate": 0.0005, "loss": 1.293210506439209, "step": 42426 }, { "epoch": 3.3446590461174615, "grad_norm": 0.4336339831352234, "learning_rate": 0.0005, "loss": 1.3857383728027344, "step": 42427 }, { "epoch": 3.3447378793851006, "grad_norm": 0.5430848598480225, "learning_rate": 0.0005, "loss": 1.5000581741333008, "step": 42428 }, { "epoch": 3.3448167126527393, "grad_norm": 0.5943288207054138, "learning_rate": 0.0005, "loss": 1.1912548542022705, "step": 42429 }, { "epoch": 3.3448955459203784, "grad_norm": 0.5008184909820557, "learning_rate": 0.0005, "loss": 1.2254598140716553, "step": 42430 }, { "epoch": 3.3449743791880175, "grad_norm": 0.4856315553188324, "learning_rate": 0.0005, "loss": 1.3109476566314697, "step": 42431 }, { "epoch": 3.3450532124556562, "grad_norm": 0.5508681535720825, "learning_rate": 0.0005, "loss": 1.2231918573379517, "step": 42432 }, { "epoch": 3.3451320457232954, "grad_norm": 0.545710563659668, "learning_rate": 0.0005, "loss": 1.2751033306121826, "step": 42433 }, { "epoch": 3.345210878990934, "grad_norm": 0.5822629332542419, "learning_rate": 0.0005, "loss": 1.4226834774017334, "step": 42434 }, { "epoch": 3.345289712258573, "grad_norm": 0.4462313652038574, "learning_rate": 0.0005, "loss": 1.2265535593032837, "step": 42435 }, { "epoch": 3.345368545526212, "grad_norm": 0.44229039549827576, "learning_rate": 0.0005, "loss": 1.3299076557159424, "step": 42436 }, { "epoch": 3.345447378793851, "grad_norm": 0.49575114250183105, "learning_rate": 0.0005, "loss": 1.3021342754364014, "step": 42437 }, { "epoch": 3.3455262120614897, "grad_norm": 0.5423589944839478, "learning_rate": 0.0005, "loss": 1.180978775024414, "step": 42438 }, { "epoch": 3.345605045329129, "grad_norm": 0.406369686126709, "learning_rate": 0.0005, "loss": 1.110373854637146, "step": 42439 }, { "epoch": 3.345683878596768, "grad_norm": 0.5741691589355469, "learning_rate": 0.0005, "loss": 1.2110368013381958, "step": 42440 }, { "epoch": 3.3457627118644067, "grad_norm": 0.5618547797203064, "learning_rate": 0.0005, "loss": 1.3855030536651611, "step": 42441 }, { "epoch": 3.345841545132046, "grad_norm": 0.44202858209609985, "learning_rate": 0.0005, "loss": 1.2297537326812744, "step": 42442 }, { "epoch": 3.3459203783996845, "grad_norm": 0.4403773844242096, "learning_rate": 0.0005, "loss": 1.310010313987732, "step": 42443 }, { "epoch": 3.3459992116673236, "grad_norm": 0.4721180498600006, "learning_rate": 0.0005, "loss": 1.2101006507873535, "step": 42444 }, { "epoch": 3.3460780449349627, "grad_norm": 0.5009961128234863, "learning_rate": 0.0005, "loss": 1.2408971786499023, "step": 42445 }, { "epoch": 3.3461568782026014, "grad_norm": 0.3709256052970886, "learning_rate": 0.0005, "loss": 1.0605885982513428, "step": 42446 }, { "epoch": 3.3462357114702406, "grad_norm": 0.5457291603088379, "learning_rate": 0.0005, "loss": 1.2522172927856445, "step": 42447 }, { "epoch": 3.3463145447378793, "grad_norm": 0.4846973717212677, "learning_rate": 0.0005, "loss": 1.4069976806640625, "step": 42448 }, { "epoch": 3.3463933780055184, "grad_norm": 0.47667789459228516, "learning_rate": 0.0005, "loss": 1.4162813425064087, "step": 42449 }, { "epoch": 3.346472211273157, "grad_norm": 0.44060006737709045, "learning_rate": 0.0005, "loss": 1.2584826946258545, "step": 42450 }, { "epoch": 3.346551044540796, "grad_norm": 0.6362452507019043, "learning_rate": 0.0005, "loss": 1.5486345291137695, "step": 42451 }, { "epoch": 3.3466298778084353, "grad_norm": 0.81242835521698, "learning_rate": 0.0005, "loss": 1.3999357223510742, "step": 42452 }, { "epoch": 3.346708711076074, "grad_norm": 0.5471291542053223, "learning_rate": 0.0005, "loss": 1.247759222984314, "step": 42453 }, { "epoch": 3.346787544343713, "grad_norm": 0.5320017337799072, "learning_rate": 0.0005, "loss": 1.3891069889068604, "step": 42454 }, { "epoch": 3.346866377611352, "grad_norm": 0.4874987006187439, "learning_rate": 0.0005, "loss": 1.261838436126709, "step": 42455 }, { "epoch": 3.346945210878991, "grad_norm": 0.364958256483078, "learning_rate": 0.0005, "loss": 1.0744411945343018, "step": 42456 }, { "epoch": 3.3470240441466297, "grad_norm": 0.479402631521225, "learning_rate": 0.0005, "loss": 1.3203585147857666, "step": 42457 }, { "epoch": 3.347102877414269, "grad_norm": 0.4611596465110779, "learning_rate": 0.0005, "loss": 1.3060359954833984, "step": 42458 }, { "epoch": 3.347181710681908, "grad_norm": 0.47585588693618774, "learning_rate": 0.0005, "loss": 1.3943426609039307, "step": 42459 }, { "epoch": 3.3472605439495466, "grad_norm": 0.49625563621520996, "learning_rate": 0.0005, "loss": 1.2477123737335205, "step": 42460 }, { "epoch": 3.3473393772171858, "grad_norm": 0.58412104845047, "learning_rate": 0.0005, "loss": 1.3419183492660522, "step": 42461 }, { "epoch": 3.3474182104848245, "grad_norm": 0.6644004583358765, "learning_rate": 0.0005, "loss": 1.5704431533813477, "step": 42462 }, { "epoch": 3.3474970437524636, "grad_norm": 0.604292631149292, "learning_rate": 0.0005, "loss": 1.2814488410949707, "step": 42463 }, { "epoch": 3.3475758770201027, "grad_norm": 0.5870885848999023, "learning_rate": 0.0005, "loss": 1.4100416898727417, "step": 42464 }, { "epoch": 3.3476547102877414, "grad_norm": 0.5081024765968323, "learning_rate": 0.0005, "loss": 1.0905390977859497, "step": 42465 }, { "epoch": 3.3477335435553806, "grad_norm": 0.545538067817688, "learning_rate": 0.0005, "loss": 1.4047942161560059, "step": 42466 }, { "epoch": 3.3478123768230192, "grad_norm": 0.4470517933368683, "learning_rate": 0.0005, "loss": 1.1957831382751465, "step": 42467 }, { "epoch": 3.3478912100906584, "grad_norm": 0.561817467212677, "learning_rate": 0.0005, "loss": 1.4927287101745605, "step": 42468 }, { "epoch": 3.347970043358297, "grad_norm": 0.5145952105522156, "learning_rate": 0.0005, "loss": 1.2175302505493164, "step": 42469 }, { "epoch": 3.348048876625936, "grad_norm": 0.4651070237159729, "learning_rate": 0.0005, "loss": 1.097002625465393, "step": 42470 }, { "epoch": 3.348127709893575, "grad_norm": 0.5360474586486816, "learning_rate": 0.0005, "loss": 1.233659267425537, "step": 42471 }, { "epoch": 3.348206543161214, "grad_norm": 0.6371740102767944, "learning_rate": 0.0005, "loss": 1.239217758178711, "step": 42472 }, { "epoch": 3.348285376428853, "grad_norm": 0.5047164559364319, "learning_rate": 0.0005, "loss": 1.0434396266937256, "step": 42473 }, { "epoch": 3.348364209696492, "grad_norm": 0.46636536717414856, "learning_rate": 0.0005, "loss": 1.264232873916626, "step": 42474 }, { "epoch": 3.348443042964131, "grad_norm": 0.5566873550415039, "learning_rate": 0.0005, "loss": 1.235440969467163, "step": 42475 }, { "epoch": 3.3485218762317697, "grad_norm": 0.5484682321548462, "learning_rate": 0.0005, "loss": 1.2765066623687744, "step": 42476 }, { "epoch": 3.348600709499409, "grad_norm": 0.48475608229637146, "learning_rate": 0.0005, "loss": 1.45707106590271, "step": 42477 }, { "epoch": 3.348679542767048, "grad_norm": 0.43489935994148254, "learning_rate": 0.0005, "loss": 1.2755372524261475, "step": 42478 }, { "epoch": 3.3487583760346866, "grad_norm": 0.4126003682613373, "learning_rate": 0.0005, "loss": 1.2568548917770386, "step": 42479 }, { "epoch": 3.3488372093023258, "grad_norm": 0.59727543592453, "learning_rate": 0.0005, "loss": 1.449302077293396, "step": 42480 }, { "epoch": 3.3489160425699644, "grad_norm": 0.49610477685928345, "learning_rate": 0.0005, "loss": 1.171386957168579, "step": 42481 }, { "epoch": 3.3489948758376036, "grad_norm": 0.4156876504421234, "learning_rate": 0.0005, "loss": 1.1045894622802734, "step": 42482 }, { "epoch": 3.3490737091052423, "grad_norm": 0.4425120949745178, "learning_rate": 0.0005, "loss": 1.2584044933319092, "step": 42483 }, { "epoch": 3.3491525423728814, "grad_norm": 0.4671076536178589, "learning_rate": 0.0005, "loss": 0.974591851234436, "step": 42484 }, { "epoch": 3.34923137564052, "grad_norm": 0.6820195317268372, "learning_rate": 0.0005, "loss": 1.3191137313842773, "step": 42485 }, { "epoch": 3.3493102089081592, "grad_norm": 0.5159589648246765, "learning_rate": 0.0005, "loss": 1.2033483982086182, "step": 42486 }, { "epoch": 3.3493890421757984, "grad_norm": 0.5635720491409302, "learning_rate": 0.0005, "loss": 1.2624585628509521, "step": 42487 }, { "epoch": 3.349467875443437, "grad_norm": 0.5726202726364136, "learning_rate": 0.0005, "loss": 1.424241304397583, "step": 42488 }, { "epoch": 3.349546708711076, "grad_norm": 0.5111655592918396, "learning_rate": 0.0005, "loss": 1.287771224975586, "step": 42489 }, { "epoch": 3.349625541978715, "grad_norm": 0.44397085905075073, "learning_rate": 0.0005, "loss": 1.1887446641921997, "step": 42490 }, { "epoch": 3.349704375246354, "grad_norm": 0.5313585996627808, "learning_rate": 0.0005, "loss": 1.4296891689300537, "step": 42491 }, { "epoch": 3.349783208513993, "grad_norm": 0.572935163974762, "learning_rate": 0.0005, "loss": 1.1038639545440674, "step": 42492 }, { "epoch": 3.349862041781632, "grad_norm": 0.4766523241996765, "learning_rate": 0.0005, "loss": 1.3407669067382812, "step": 42493 }, { "epoch": 3.349940875049271, "grad_norm": 0.40346643328666687, "learning_rate": 0.0005, "loss": 1.128145694732666, "step": 42494 }, { "epoch": 3.3500197083169097, "grad_norm": 0.6327558755874634, "learning_rate": 0.0005, "loss": 1.3333476781845093, "step": 42495 }, { "epoch": 3.350098541584549, "grad_norm": 0.4867997467517853, "learning_rate": 0.0005, "loss": 1.3148057460784912, "step": 42496 }, { "epoch": 3.3501773748521875, "grad_norm": 0.5102829933166504, "learning_rate": 0.0005, "loss": 1.2089076042175293, "step": 42497 }, { "epoch": 3.3502562081198266, "grad_norm": 0.5526378750801086, "learning_rate": 0.0005, "loss": 1.4008283615112305, "step": 42498 }, { "epoch": 3.3503350413874653, "grad_norm": 0.5311782360076904, "learning_rate": 0.0005, "loss": 1.1875567436218262, "step": 42499 }, { "epoch": 3.3504138746551044, "grad_norm": 0.49309778213500977, "learning_rate": 0.0005, "loss": 1.3525599241256714, "step": 42500 }, { "epoch": 3.3504927079227436, "grad_norm": 0.4822326600551605, "learning_rate": 0.0005, "loss": 1.2280163764953613, "step": 42501 }, { "epoch": 3.3505715411903823, "grad_norm": 0.5732434988021851, "learning_rate": 0.0005, "loss": 1.3576686382293701, "step": 42502 }, { "epoch": 3.3506503744580214, "grad_norm": 0.39630430936813354, "learning_rate": 0.0005, "loss": 1.26460862159729, "step": 42503 }, { "epoch": 3.35072920772566, "grad_norm": 0.5740953683853149, "learning_rate": 0.0005, "loss": 1.4912993907928467, "step": 42504 }, { "epoch": 3.350808040993299, "grad_norm": 0.617511510848999, "learning_rate": 0.0005, "loss": 1.2444183826446533, "step": 42505 }, { "epoch": 3.3508868742609383, "grad_norm": 0.7013572454452515, "learning_rate": 0.0005, "loss": 1.2227756977081299, "step": 42506 }, { "epoch": 3.350965707528577, "grad_norm": 0.5632012486457825, "learning_rate": 0.0005, "loss": 1.4623819589614868, "step": 42507 }, { "epoch": 3.351044540796216, "grad_norm": 0.46502941846847534, "learning_rate": 0.0005, "loss": 1.309446930885315, "step": 42508 }, { "epoch": 3.351123374063855, "grad_norm": 0.48685234785079956, "learning_rate": 0.0005, "loss": 1.1590275764465332, "step": 42509 }, { "epoch": 3.351202207331494, "grad_norm": 0.4442802369594574, "learning_rate": 0.0005, "loss": 1.292426347732544, "step": 42510 }, { "epoch": 3.3512810405991327, "grad_norm": 0.49298954010009766, "learning_rate": 0.0005, "loss": 1.3335087299346924, "step": 42511 }, { "epoch": 3.351359873866772, "grad_norm": 0.6025272607803345, "learning_rate": 0.0005, "loss": 1.346366047859192, "step": 42512 }, { "epoch": 3.3514387071344105, "grad_norm": 0.5124161243438721, "learning_rate": 0.0005, "loss": 1.3244049549102783, "step": 42513 }, { "epoch": 3.3515175404020496, "grad_norm": 0.6020404100418091, "learning_rate": 0.0005, "loss": 1.3114328384399414, "step": 42514 }, { "epoch": 3.3515963736696888, "grad_norm": 0.5578948259353638, "learning_rate": 0.0005, "loss": 1.2145425081253052, "step": 42515 }, { "epoch": 3.3516752069373275, "grad_norm": 0.4213910400867462, "learning_rate": 0.0005, "loss": 1.2113487720489502, "step": 42516 }, { "epoch": 3.3517540402049666, "grad_norm": 0.4077582359313965, "learning_rate": 0.0005, "loss": 1.287518858909607, "step": 42517 }, { "epoch": 3.3518328734726053, "grad_norm": 0.43599143624305725, "learning_rate": 0.0005, "loss": 1.1431010961532593, "step": 42518 }, { "epoch": 3.3519117067402444, "grad_norm": 0.6348451972007751, "learning_rate": 0.0005, "loss": 1.286489725112915, "step": 42519 }, { "epoch": 3.3519905400078835, "grad_norm": 0.5874478220939636, "learning_rate": 0.0005, "loss": 1.1864070892333984, "step": 42520 }, { "epoch": 3.3520693732755222, "grad_norm": 0.5467912554740906, "learning_rate": 0.0005, "loss": 1.273460030555725, "step": 42521 }, { "epoch": 3.3521482065431614, "grad_norm": 0.4938220977783203, "learning_rate": 0.0005, "loss": 1.1435279846191406, "step": 42522 }, { "epoch": 3.3522270398108, "grad_norm": 0.4704279899597168, "learning_rate": 0.0005, "loss": 1.1486146450042725, "step": 42523 }, { "epoch": 3.352305873078439, "grad_norm": 0.5386659502983093, "learning_rate": 0.0005, "loss": 1.1572808027267456, "step": 42524 }, { "epoch": 3.352384706346078, "grad_norm": 0.520433783531189, "learning_rate": 0.0005, "loss": 1.1311969757080078, "step": 42525 }, { "epoch": 3.352463539613717, "grad_norm": 0.6783676743507385, "learning_rate": 0.0005, "loss": 1.42312490940094, "step": 42526 }, { "epoch": 3.3525423728813557, "grad_norm": 0.5056478977203369, "learning_rate": 0.0005, "loss": 1.3864850997924805, "step": 42527 }, { "epoch": 3.352621206148995, "grad_norm": 0.47727182507514954, "learning_rate": 0.0005, "loss": 1.1652995347976685, "step": 42528 }, { "epoch": 3.352700039416634, "grad_norm": 0.4605635702610016, "learning_rate": 0.0005, "loss": 1.3380171060562134, "step": 42529 }, { "epoch": 3.3527788726842727, "grad_norm": 0.532476007938385, "learning_rate": 0.0005, "loss": 1.4032888412475586, "step": 42530 }, { "epoch": 3.352857705951912, "grad_norm": 0.5090228915214539, "learning_rate": 0.0005, "loss": 1.439924955368042, "step": 42531 }, { "epoch": 3.3529365392195505, "grad_norm": 0.40284213423728943, "learning_rate": 0.0005, "loss": 1.251326084136963, "step": 42532 }, { "epoch": 3.3530153724871896, "grad_norm": 0.507203221321106, "learning_rate": 0.0005, "loss": 1.4203541278839111, "step": 42533 }, { "epoch": 3.3530942057548287, "grad_norm": 0.4265500009059906, "learning_rate": 0.0005, "loss": 1.2188334465026855, "step": 42534 }, { "epoch": 3.3531730390224674, "grad_norm": 0.43219026923179626, "learning_rate": 0.0005, "loss": 1.1474615335464478, "step": 42535 }, { "epoch": 3.3532518722901066, "grad_norm": 1.5512803792953491, "learning_rate": 0.0005, "loss": 1.2965452671051025, "step": 42536 }, { "epoch": 3.3533307055577453, "grad_norm": 0.5659741759300232, "learning_rate": 0.0005, "loss": 1.3255499601364136, "step": 42537 }, { "epoch": 3.3534095388253844, "grad_norm": 0.4983028173446655, "learning_rate": 0.0005, "loss": 1.3836913108825684, "step": 42538 }, { "epoch": 3.353488372093023, "grad_norm": 0.6419510245323181, "learning_rate": 0.0005, "loss": 1.4423270225524902, "step": 42539 }, { "epoch": 3.353567205360662, "grad_norm": 0.4787171185016632, "learning_rate": 0.0005, "loss": 1.20639967918396, "step": 42540 }, { "epoch": 3.353646038628301, "grad_norm": 0.4599968492984772, "learning_rate": 0.0005, "loss": 1.0029629468917847, "step": 42541 }, { "epoch": 3.35372487189594, "grad_norm": 0.44016677141189575, "learning_rate": 0.0005, "loss": 1.1833480596542358, "step": 42542 }, { "epoch": 3.353803705163579, "grad_norm": 0.519207775592804, "learning_rate": 0.0005, "loss": 1.2193689346313477, "step": 42543 }, { "epoch": 3.353882538431218, "grad_norm": 0.5412008762359619, "learning_rate": 0.0005, "loss": 1.2621710300445557, "step": 42544 }, { "epoch": 3.353961371698857, "grad_norm": 0.5723956823348999, "learning_rate": 0.0005, "loss": 1.4970065355300903, "step": 42545 }, { "epoch": 3.3540402049664957, "grad_norm": 0.5162366032600403, "learning_rate": 0.0005, "loss": 1.4330588579177856, "step": 42546 }, { "epoch": 3.354119038234135, "grad_norm": 0.5514576435089111, "learning_rate": 0.0005, "loss": 1.2374095916748047, "step": 42547 }, { "epoch": 3.354197871501774, "grad_norm": 0.48078858852386475, "learning_rate": 0.0005, "loss": 1.0999031066894531, "step": 42548 }, { "epoch": 3.3542767047694126, "grad_norm": 0.46552547812461853, "learning_rate": 0.0005, "loss": 1.2871596813201904, "step": 42549 }, { "epoch": 3.3543555380370518, "grad_norm": 0.47301018238067627, "learning_rate": 0.0005, "loss": 1.2916712760925293, "step": 42550 }, { "epoch": 3.3544343713046905, "grad_norm": 0.5261944532394409, "learning_rate": 0.0005, "loss": 1.219348669052124, "step": 42551 }, { "epoch": 3.3545132045723296, "grad_norm": 0.5131217241287231, "learning_rate": 0.0005, "loss": 1.4635664224624634, "step": 42552 }, { "epoch": 3.3545920378399683, "grad_norm": 0.5853961706161499, "learning_rate": 0.0005, "loss": 1.2843246459960938, "step": 42553 }, { "epoch": 3.3546708711076074, "grad_norm": 0.5430203676223755, "learning_rate": 0.0005, "loss": 1.2506890296936035, "step": 42554 }, { "epoch": 3.354749704375246, "grad_norm": 0.44340652227401733, "learning_rate": 0.0005, "loss": 1.1773254871368408, "step": 42555 }, { "epoch": 3.3548285376428852, "grad_norm": 0.5458568334579468, "learning_rate": 0.0005, "loss": 1.2962820529937744, "step": 42556 }, { "epoch": 3.3549073709105244, "grad_norm": 0.43479785323143005, "learning_rate": 0.0005, "loss": 1.190309762954712, "step": 42557 }, { "epoch": 3.354986204178163, "grad_norm": 0.5701841115951538, "learning_rate": 0.0005, "loss": 1.3493034839630127, "step": 42558 }, { "epoch": 3.355065037445802, "grad_norm": 0.555820643901825, "learning_rate": 0.0005, "loss": 1.3075647354125977, "step": 42559 }, { "epoch": 3.355143870713441, "grad_norm": 0.45093438029289246, "learning_rate": 0.0005, "loss": 1.198004961013794, "step": 42560 }, { "epoch": 3.35522270398108, "grad_norm": 0.4877055585384369, "learning_rate": 0.0005, "loss": 1.2858210802078247, "step": 42561 }, { "epoch": 3.355301537248719, "grad_norm": 0.5274288058280945, "learning_rate": 0.0005, "loss": 1.2050204277038574, "step": 42562 }, { "epoch": 3.355380370516358, "grad_norm": 0.522805392742157, "learning_rate": 0.0005, "loss": 1.3420352935791016, "step": 42563 }, { "epoch": 3.355459203783997, "grad_norm": 0.4134247601032257, "learning_rate": 0.0005, "loss": 1.2615844011306763, "step": 42564 }, { "epoch": 3.3555380370516357, "grad_norm": 0.4200349748134613, "learning_rate": 0.0005, "loss": 1.2446739673614502, "step": 42565 }, { "epoch": 3.355616870319275, "grad_norm": 0.5131360292434692, "learning_rate": 0.0005, "loss": 1.1772130727767944, "step": 42566 }, { "epoch": 3.3556957035869135, "grad_norm": 0.4649904668331146, "learning_rate": 0.0005, "loss": 1.2829363346099854, "step": 42567 }, { "epoch": 3.3557745368545526, "grad_norm": 0.4534517228603363, "learning_rate": 0.0005, "loss": 1.3466836214065552, "step": 42568 }, { "epoch": 3.3558533701221918, "grad_norm": 0.48144254088401794, "learning_rate": 0.0005, "loss": 1.3781031370162964, "step": 42569 }, { "epoch": 3.3559322033898304, "grad_norm": 0.40721216797828674, "learning_rate": 0.0005, "loss": 1.255256175994873, "step": 42570 }, { "epoch": 3.3560110366574696, "grad_norm": 0.4693513810634613, "learning_rate": 0.0005, "loss": 1.133819818496704, "step": 42571 }, { "epoch": 3.3560898699251083, "grad_norm": 0.5438830852508545, "learning_rate": 0.0005, "loss": 1.454430341720581, "step": 42572 }, { "epoch": 3.3561687031927474, "grad_norm": 0.42374715209007263, "learning_rate": 0.0005, "loss": 1.3175673484802246, "step": 42573 }, { "epoch": 3.356247536460386, "grad_norm": 0.4854324758052826, "learning_rate": 0.0005, "loss": 1.2571074962615967, "step": 42574 }, { "epoch": 3.3563263697280252, "grad_norm": 0.5164278745651245, "learning_rate": 0.0005, "loss": 1.3452939987182617, "step": 42575 }, { "epoch": 3.3564052029956644, "grad_norm": 0.44590699672698975, "learning_rate": 0.0005, "loss": 1.256988286972046, "step": 42576 }, { "epoch": 3.356484036263303, "grad_norm": 0.6867178678512573, "learning_rate": 0.0005, "loss": 1.263224720954895, "step": 42577 }, { "epoch": 3.356562869530942, "grad_norm": 0.47209569811820984, "learning_rate": 0.0005, "loss": 1.2985739707946777, "step": 42578 }, { "epoch": 3.356641702798581, "grad_norm": 0.47188425064086914, "learning_rate": 0.0005, "loss": 1.2669509649276733, "step": 42579 }, { "epoch": 3.35672053606622, "grad_norm": 0.4379241168498993, "learning_rate": 0.0005, "loss": 1.2501499652862549, "step": 42580 }, { "epoch": 3.3567993693338587, "grad_norm": 0.476429283618927, "learning_rate": 0.0005, "loss": 1.1945043802261353, "step": 42581 }, { "epoch": 3.356878202601498, "grad_norm": 0.48316267132759094, "learning_rate": 0.0005, "loss": 1.2959926128387451, "step": 42582 }, { "epoch": 3.356957035869137, "grad_norm": 0.5199968814849854, "learning_rate": 0.0005, "loss": 1.2134147882461548, "step": 42583 }, { "epoch": 3.3570358691367757, "grad_norm": 0.4014134109020233, "learning_rate": 0.0005, "loss": 1.126596450805664, "step": 42584 }, { "epoch": 3.357114702404415, "grad_norm": 1.5738998651504517, "learning_rate": 0.0005, "loss": 1.393951654434204, "step": 42585 }, { "epoch": 3.3571935356720535, "grad_norm": 0.44119730591773987, "learning_rate": 0.0005, "loss": 1.3054615259170532, "step": 42586 }, { "epoch": 3.3572723689396926, "grad_norm": 0.487751305103302, "learning_rate": 0.0005, "loss": 1.201554536819458, "step": 42587 }, { "epoch": 3.3573512022073313, "grad_norm": 0.5071004033088684, "learning_rate": 0.0005, "loss": 1.3384356498718262, "step": 42588 }, { "epoch": 3.3574300354749704, "grad_norm": 0.5454431772232056, "learning_rate": 0.0005, "loss": 1.397212266921997, "step": 42589 }, { "epoch": 3.3575088687426096, "grad_norm": 0.461275190114975, "learning_rate": 0.0005, "loss": 1.2539429664611816, "step": 42590 }, { "epoch": 3.3575877020102483, "grad_norm": 0.5256936550140381, "learning_rate": 0.0005, "loss": 1.2644670009613037, "step": 42591 }, { "epoch": 3.3576665352778874, "grad_norm": 0.48461610078811646, "learning_rate": 0.0005, "loss": 1.2836555242538452, "step": 42592 }, { "epoch": 3.357745368545526, "grad_norm": 0.4552992582321167, "learning_rate": 0.0005, "loss": 1.1373066902160645, "step": 42593 }, { "epoch": 3.357824201813165, "grad_norm": 0.5619865655899048, "learning_rate": 0.0005, "loss": 1.2861994504928589, "step": 42594 }, { "epoch": 3.3579030350808043, "grad_norm": 0.49499401450157166, "learning_rate": 0.0005, "loss": 1.2990124225616455, "step": 42595 }, { "epoch": 3.357981868348443, "grad_norm": 0.6202576756477356, "learning_rate": 0.0005, "loss": 1.155917763710022, "step": 42596 }, { "epoch": 3.358060701616082, "grad_norm": 0.4454291760921478, "learning_rate": 0.0005, "loss": 1.3544042110443115, "step": 42597 }, { "epoch": 3.358139534883721, "grad_norm": 0.43373507261276245, "learning_rate": 0.0005, "loss": 1.0816820859909058, "step": 42598 }, { "epoch": 3.35821836815136, "grad_norm": 0.3838677406311035, "learning_rate": 0.0005, "loss": 1.1319199800491333, "step": 42599 }, { "epoch": 3.3582972014189987, "grad_norm": 0.4804266393184662, "learning_rate": 0.0005, "loss": 1.4185268878936768, "step": 42600 }, { "epoch": 3.358376034686638, "grad_norm": 0.5062828660011292, "learning_rate": 0.0005, "loss": 1.2081360816955566, "step": 42601 }, { "epoch": 3.3584548679542765, "grad_norm": 0.5694402456283569, "learning_rate": 0.0005, "loss": 1.1970627307891846, "step": 42602 }, { "epoch": 3.3585337012219156, "grad_norm": 0.5061668157577515, "learning_rate": 0.0005, "loss": 1.3323371410369873, "step": 42603 }, { "epoch": 3.3586125344895548, "grad_norm": 0.4722276031970978, "learning_rate": 0.0005, "loss": 1.438768982887268, "step": 42604 }, { "epoch": 3.3586913677571935, "grad_norm": 0.5230743885040283, "learning_rate": 0.0005, "loss": 1.0991603136062622, "step": 42605 }, { "epoch": 3.3587702010248326, "grad_norm": 0.4672980010509491, "learning_rate": 0.0005, "loss": 1.1050680875778198, "step": 42606 }, { "epoch": 3.3588490342924713, "grad_norm": 0.456846684217453, "learning_rate": 0.0005, "loss": 1.1716113090515137, "step": 42607 }, { "epoch": 3.3589278675601104, "grad_norm": 0.5892602205276489, "learning_rate": 0.0005, "loss": 1.2521352767944336, "step": 42608 }, { "epoch": 3.3590067008277495, "grad_norm": 0.5246674418449402, "learning_rate": 0.0005, "loss": 1.266381025314331, "step": 42609 }, { "epoch": 3.3590855340953882, "grad_norm": 0.5290395617485046, "learning_rate": 0.0005, "loss": 1.426844835281372, "step": 42610 }, { "epoch": 3.3591643673630274, "grad_norm": 0.49102434515953064, "learning_rate": 0.0005, "loss": 1.2273660898208618, "step": 42611 }, { "epoch": 3.359243200630666, "grad_norm": 0.5081270933151245, "learning_rate": 0.0005, "loss": 1.096758246421814, "step": 42612 }, { "epoch": 3.359322033898305, "grad_norm": 0.5052531361579895, "learning_rate": 0.0005, "loss": 1.1427559852600098, "step": 42613 }, { "epoch": 3.359400867165944, "grad_norm": 0.543298602104187, "learning_rate": 0.0005, "loss": 1.3415477275848389, "step": 42614 }, { "epoch": 3.359479700433583, "grad_norm": 0.5727959871292114, "learning_rate": 0.0005, "loss": 1.3214269876480103, "step": 42615 }, { "epoch": 3.3595585337012217, "grad_norm": 0.543392539024353, "learning_rate": 0.0005, "loss": 1.2216429710388184, "step": 42616 }, { "epoch": 3.359637366968861, "grad_norm": 1.01414954662323, "learning_rate": 0.0005, "loss": 1.438513994216919, "step": 42617 }, { "epoch": 3.3597162002365, "grad_norm": 0.42415422201156616, "learning_rate": 0.0005, "loss": 1.3038830757141113, "step": 42618 }, { "epoch": 3.3597950335041387, "grad_norm": 0.46490031480789185, "learning_rate": 0.0005, "loss": 1.3370110988616943, "step": 42619 }, { "epoch": 3.359873866771778, "grad_norm": 0.5314767360687256, "learning_rate": 0.0005, "loss": 1.1841750144958496, "step": 42620 }, { "epoch": 3.3599527000394165, "grad_norm": 0.4826192855834961, "learning_rate": 0.0005, "loss": 1.2321586608886719, "step": 42621 }, { "epoch": 3.3600315333070556, "grad_norm": 0.4812522232532501, "learning_rate": 0.0005, "loss": 1.19792640209198, "step": 42622 }, { "epoch": 3.3601103665746948, "grad_norm": 0.580966591835022, "learning_rate": 0.0005, "loss": 1.27219557762146, "step": 42623 }, { "epoch": 3.3601891998423334, "grad_norm": 0.4504327178001404, "learning_rate": 0.0005, "loss": 1.1936914920806885, "step": 42624 }, { "epoch": 3.3602680331099726, "grad_norm": 0.49724647402763367, "learning_rate": 0.0005, "loss": 1.1907649040222168, "step": 42625 }, { "epoch": 3.3603468663776113, "grad_norm": 0.5680786967277527, "learning_rate": 0.0005, "loss": 1.2880580425262451, "step": 42626 }, { "epoch": 3.3604256996452504, "grad_norm": 0.4526606798171997, "learning_rate": 0.0005, "loss": 1.1606664657592773, "step": 42627 }, { "epoch": 3.360504532912889, "grad_norm": 0.5936855673789978, "learning_rate": 0.0005, "loss": 1.3582209348678589, "step": 42628 }, { "epoch": 3.360583366180528, "grad_norm": 0.4814375638961792, "learning_rate": 0.0005, "loss": 1.3614095449447632, "step": 42629 }, { "epoch": 3.360662199448167, "grad_norm": 0.509188175201416, "learning_rate": 0.0005, "loss": 1.285315752029419, "step": 42630 }, { "epoch": 3.360741032715806, "grad_norm": 0.48705363273620605, "learning_rate": 0.0005, "loss": 1.2121678590774536, "step": 42631 }, { "epoch": 3.360819865983445, "grad_norm": 0.6202617883682251, "learning_rate": 0.0005, "loss": 1.135115385055542, "step": 42632 }, { "epoch": 3.360898699251084, "grad_norm": 0.44153326749801636, "learning_rate": 0.0005, "loss": 1.162695288658142, "step": 42633 }, { "epoch": 3.360977532518723, "grad_norm": 0.38469403982162476, "learning_rate": 0.0005, "loss": 1.1694368124008179, "step": 42634 }, { "epoch": 3.3610563657863617, "grad_norm": 0.40511584281921387, "learning_rate": 0.0005, "loss": 1.0564415454864502, "step": 42635 }, { "epoch": 3.361135199054001, "grad_norm": 0.38680535554885864, "learning_rate": 0.0005, "loss": 1.020140528678894, "step": 42636 }, { "epoch": 3.36121403232164, "grad_norm": 0.4711974561214447, "learning_rate": 0.0005, "loss": 1.2679938077926636, "step": 42637 }, { "epoch": 3.3612928655892786, "grad_norm": 0.6707350611686707, "learning_rate": 0.0005, "loss": 1.3918981552124023, "step": 42638 }, { "epoch": 3.3613716988569178, "grad_norm": 0.5689778923988342, "learning_rate": 0.0005, "loss": 1.2993601560592651, "step": 42639 }, { "epoch": 3.3614505321245565, "grad_norm": 0.6644056439399719, "learning_rate": 0.0005, "loss": 1.2272988557815552, "step": 42640 }, { "epoch": 3.3615293653921956, "grad_norm": 0.5497466921806335, "learning_rate": 0.0005, "loss": 1.264762282371521, "step": 42641 }, { "epoch": 3.3616081986598343, "grad_norm": 0.43554604053497314, "learning_rate": 0.0005, "loss": 1.2102727890014648, "step": 42642 }, { "epoch": 3.3616870319274734, "grad_norm": 0.49717164039611816, "learning_rate": 0.0005, "loss": 1.2174873352050781, "step": 42643 }, { "epoch": 3.361765865195112, "grad_norm": 0.43715164065361023, "learning_rate": 0.0005, "loss": 1.182060956954956, "step": 42644 }, { "epoch": 3.3618446984627512, "grad_norm": 0.6873751282691956, "learning_rate": 0.0005, "loss": 1.5232218503952026, "step": 42645 }, { "epoch": 3.3619235317303904, "grad_norm": 0.4443782866001129, "learning_rate": 0.0005, "loss": 1.1761703491210938, "step": 42646 }, { "epoch": 3.362002364998029, "grad_norm": 0.5539326071739197, "learning_rate": 0.0005, "loss": 1.2094670534133911, "step": 42647 }, { "epoch": 3.362081198265668, "grad_norm": 0.6627309322357178, "learning_rate": 0.0005, "loss": 1.3073862791061401, "step": 42648 }, { "epoch": 3.362160031533307, "grad_norm": 0.4815302789211273, "learning_rate": 0.0005, "loss": 1.4295403957366943, "step": 42649 }, { "epoch": 3.362238864800946, "grad_norm": 0.45070070028305054, "learning_rate": 0.0005, "loss": 1.2171249389648438, "step": 42650 }, { "epoch": 3.362317698068585, "grad_norm": 0.4997197687625885, "learning_rate": 0.0005, "loss": 1.210494875907898, "step": 42651 }, { "epoch": 3.362396531336224, "grad_norm": 0.45121002197265625, "learning_rate": 0.0005, "loss": 1.1231318712234497, "step": 42652 }, { "epoch": 3.362475364603863, "grad_norm": 0.5398979187011719, "learning_rate": 0.0005, "loss": 1.113215684890747, "step": 42653 }, { "epoch": 3.3625541978715017, "grad_norm": 0.7360293865203857, "learning_rate": 0.0005, "loss": 1.1360828876495361, "step": 42654 }, { "epoch": 3.362633031139141, "grad_norm": 0.3440213203430176, "learning_rate": 0.0005, "loss": 0.974789023399353, "step": 42655 }, { "epoch": 3.3627118644067795, "grad_norm": 0.5196771621704102, "learning_rate": 0.0005, "loss": 1.296372890472412, "step": 42656 }, { "epoch": 3.3627906976744186, "grad_norm": 0.46373337507247925, "learning_rate": 0.0005, "loss": 1.297523021697998, "step": 42657 }, { "epoch": 3.3628695309420573, "grad_norm": 0.44239580631256104, "learning_rate": 0.0005, "loss": 1.240074634552002, "step": 42658 }, { "epoch": 3.3629483642096965, "grad_norm": 0.5156482458114624, "learning_rate": 0.0005, "loss": 1.2545372247695923, "step": 42659 }, { "epoch": 3.3630271974773356, "grad_norm": 0.5565652251243591, "learning_rate": 0.0005, "loss": 1.4130107164382935, "step": 42660 }, { "epoch": 3.3631060307449743, "grad_norm": 0.4743078052997589, "learning_rate": 0.0005, "loss": 1.3723136186599731, "step": 42661 }, { "epoch": 3.3631848640126134, "grad_norm": 0.4756978154182434, "learning_rate": 0.0005, "loss": 1.2594410181045532, "step": 42662 }, { "epoch": 3.363263697280252, "grad_norm": 0.5053350925445557, "learning_rate": 0.0005, "loss": 1.3170496225357056, "step": 42663 }, { "epoch": 3.3633425305478912, "grad_norm": 0.4895392954349518, "learning_rate": 0.0005, "loss": 1.3514318466186523, "step": 42664 }, { "epoch": 3.3634213638155304, "grad_norm": 0.43150195479393005, "learning_rate": 0.0005, "loss": 1.1792361736297607, "step": 42665 }, { "epoch": 3.363500197083169, "grad_norm": 0.4781048595905304, "learning_rate": 0.0005, "loss": 1.1100881099700928, "step": 42666 }, { "epoch": 3.363579030350808, "grad_norm": 0.4815339744091034, "learning_rate": 0.0005, "loss": 1.2156224250793457, "step": 42667 }, { "epoch": 3.363657863618447, "grad_norm": 0.4559330642223358, "learning_rate": 0.0005, "loss": 1.3236254453659058, "step": 42668 }, { "epoch": 3.363736696886086, "grad_norm": 0.426220178604126, "learning_rate": 0.0005, "loss": 1.1082139015197754, "step": 42669 }, { "epoch": 3.3638155301537247, "grad_norm": 0.47969263792037964, "learning_rate": 0.0005, "loss": 1.2341675758361816, "step": 42670 }, { "epoch": 3.363894363421364, "grad_norm": 0.48585522174835205, "learning_rate": 0.0005, "loss": 1.2231663465499878, "step": 42671 }, { "epoch": 3.3639731966890025, "grad_norm": 0.47255730628967285, "learning_rate": 0.0005, "loss": 1.3869261741638184, "step": 42672 }, { "epoch": 3.3640520299566417, "grad_norm": 0.4111787974834442, "learning_rate": 0.0005, "loss": 1.2306127548217773, "step": 42673 }, { "epoch": 3.364130863224281, "grad_norm": 0.4783375561237335, "learning_rate": 0.0005, "loss": 1.2593632936477661, "step": 42674 }, { "epoch": 3.3642096964919195, "grad_norm": 0.47754693031311035, "learning_rate": 0.0005, "loss": 1.4957144260406494, "step": 42675 }, { "epoch": 3.3642885297595586, "grad_norm": 0.6147499084472656, "learning_rate": 0.0005, "loss": 1.4819501638412476, "step": 42676 }, { "epoch": 3.3643673630271973, "grad_norm": 0.5294790267944336, "learning_rate": 0.0005, "loss": 1.1601052284240723, "step": 42677 }, { "epoch": 3.3644461962948364, "grad_norm": 0.5095248818397522, "learning_rate": 0.0005, "loss": 1.2775187492370605, "step": 42678 }, { "epoch": 3.3645250295624756, "grad_norm": 0.4698661267757416, "learning_rate": 0.0005, "loss": 1.256331205368042, "step": 42679 }, { "epoch": 3.3646038628301143, "grad_norm": 0.4587051570415497, "learning_rate": 0.0005, "loss": 1.360696792602539, "step": 42680 }, { "epoch": 3.3646826960977534, "grad_norm": 0.5164910554885864, "learning_rate": 0.0005, "loss": 1.250150442123413, "step": 42681 }, { "epoch": 3.364761529365392, "grad_norm": 0.4733499586582184, "learning_rate": 0.0005, "loss": 1.3119730949401855, "step": 42682 }, { "epoch": 3.364840362633031, "grad_norm": 0.36193281412124634, "learning_rate": 0.0005, "loss": 1.1736810207366943, "step": 42683 }, { "epoch": 3.36491919590067, "grad_norm": 0.48914942145347595, "learning_rate": 0.0005, "loss": 1.2634981870651245, "step": 42684 }, { "epoch": 3.364998029168309, "grad_norm": 0.6341032981872559, "learning_rate": 0.0005, "loss": 1.2536797523498535, "step": 42685 }, { "epoch": 3.3650768624359477, "grad_norm": 0.5626780390739441, "learning_rate": 0.0005, "loss": 1.2395178079605103, "step": 42686 }, { "epoch": 3.365155695703587, "grad_norm": 0.5265045166015625, "learning_rate": 0.0005, "loss": 1.5322344303131104, "step": 42687 }, { "epoch": 3.365234528971226, "grad_norm": 0.4675958454608917, "learning_rate": 0.0005, "loss": 1.2730364799499512, "step": 42688 }, { "epoch": 3.3653133622388647, "grad_norm": 0.42037662863731384, "learning_rate": 0.0005, "loss": 1.2594683170318604, "step": 42689 }, { "epoch": 3.365392195506504, "grad_norm": 0.44871461391448975, "learning_rate": 0.0005, "loss": 1.2132411003112793, "step": 42690 }, { "epoch": 3.3654710287741425, "grad_norm": 0.7223486304283142, "learning_rate": 0.0005, "loss": 1.2595019340515137, "step": 42691 }, { "epoch": 3.3655498620417816, "grad_norm": 0.4165685474872589, "learning_rate": 0.0005, "loss": 1.163686752319336, "step": 42692 }, { "epoch": 3.3656286953094208, "grad_norm": 0.45776116847991943, "learning_rate": 0.0005, "loss": 1.3745832443237305, "step": 42693 }, { "epoch": 3.3657075285770595, "grad_norm": 0.4976176619529724, "learning_rate": 0.0005, "loss": 1.2250044345855713, "step": 42694 }, { "epoch": 3.3657863618446986, "grad_norm": 0.4897865355014801, "learning_rate": 0.0005, "loss": 1.2554537057876587, "step": 42695 }, { "epoch": 3.3658651951123373, "grad_norm": 0.38004934787750244, "learning_rate": 0.0005, "loss": 1.1511476039886475, "step": 42696 }, { "epoch": 3.3659440283799764, "grad_norm": 0.5645003914833069, "learning_rate": 0.0005, "loss": 1.440367579460144, "step": 42697 }, { "epoch": 3.366022861647615, "grad_norm": 0.49168720841407776, "learning_rate": 0.0005, "loss": 1.0626468658447266, "step": 42698 }, { "epoch": 3.3661016949152542, "grad_norm": 0.5073549151420593, "learning_rate": 0.0005, "loss": 1.262709379196167, "step": 42699 }, { "epoch": 3.3661805281828934, "grad_norm": 0.5523815751075745, "learning_rate": 0.0005, "loss": 1.0601801872253418, "step": 42700 }, { "epoch": 3.366259361450532, "grad_norm": 0.45704254508018494, "learning_rate": 0.0005, "loss": 1.2219300270080566, "step": 42701 }, { "epoch": 3.366338194718171, "grad_norm": 0.5689232349395752, "learning_rate": 0.0005, "loss": 1.330254077911377, "step": 42702 }, { "epoch": 3.36641702798581, "grad_norm": 0.419935017824173, "learning_rate": 0.0005, "loss": 1.209850788116455, "step": 42703 }, { "epoch": 3.366495861253449, "grad_norm": 0.43361547589302063, "learning_rate": 0.0005, "loss": 1.1337659358978271, "step": 42704 }, { "epoch": 3.3665746945210877, "grad_norm": 0.7639635801315308, "learning_rate": 0.0005, "loss": 1.283616065979004, "step": 42705 }, { "epoch": 3.366653527788727, "grad_norm": 0.5554251074790955, "learning_rate": 0.0005, "loss": 1.2345292568206787, "step": 42706 }, { "epoch": 3.366732361056366, "grad_norm": 0.5881954431533813, "learning_rate": 0.0005, "loss": 1.3859443664550781, "step": 42707 }, { "epoch": 3.3668111943240047, "grad_norm": 0.5034611225128174, "learning_rate": 0.0005, "loss": 1.1659770011901855, "step": 42708 }, { "epoch": 3.366890027591644, "grad_norm": 0.5436041355133057, "learning_rate": 0.0005, "loss": 1.3755792379379272, "step": 42709 }, { "epoch": 3.3669688608592825, "grad_norm": 0.4388263523578644, "learning_rate": 0.0005, "loss": 1.1372244358062744, "step": 42710 }, { "epoch": 3.3670476941269216, "grad_norm": 0.5155458450317383, "learning_rate": 0.0005, "loss": 1.506619930267334, "step": 42711 }, { "epoch": 3.3671265273945608, "grad_norm": 0.43318143486976624, "learning_rate": 0.0005, "loss": 1.0450278520584106, "step": 42712 }, { "epoch": 3.3672053606621994, "grad_norm": 0.3840367794036865, "learning_rate": 0.0005, "loss": 1.1254236698150635, "step": 42713 }, { "epoch": 3.3672841939298386, "grad_norm": 0.5252423286437988, "learning_rate": 0.0005, "loss": 1.3074767589569092, "step": 42714 }, { "epoch": 3.3673630271974773, "grad_norm": 0.5388528108596802, "learning_rate": 0.0005, "loss": 1.16232168674469, "step": 42715 }, { "epoch": 3.3674418604651164, "grad_norm": 0.6151436567306519, "learning_rate": 0.0005, "loss": 1.259880542755127, "step": 42716 }, { "epoch": 3.367520693732755, "grad_norm": 0.5505131483078003, "learning_rate": 0.0005, "loss": 1.2324261665344238, "step": 42717 }, { "epoch": 3.367599527000394, "grad_norm": 0.5246165990829468, "learning_rate": 0.0005, "loss": 1.2501897811889648, "step": 42718 }, { "epoch": 3.367678360268033, "grad_norm": 0.4166138172149658, "learning_rate": 0.0005, "loss": 1.1728835105895996, "step": 42719 }, { "epoch": 3.367757193535672, "grad_norm": 0.586530864238739, "learning_rate": 0.0005, "loss": 1.3008726835250854, "step": 42720 }, { "epoch": 3.367836026803311, "grad_norm": 0.5485777854919434, "learning_rate": 0.0005, "loss": 1.2891454696655273, "step": 42721 }, { "epoch": 3.36791486007095, "grad_norm": 0.5883936882019043, "learning_rate": 0.0005, "loss": 1.3342671394348145, "step": 42722 }, { "epoch": 3.367993693338589, "grad_norm": 0.5187005996704102, "learning_rate": 0.0005, "loss": 1.2226814031600952, "step": 42723 }, { "epoch": 3.3680725266062277, "grad_norm": 0.5719951391220093, "learning_rate": 0.0005, "loss": 1.409011721611023, "step": 42724 }, { "epoch": 3.368151359873867, "grad_norm": 0.4390251934528351, "learning_rate": 0.0005, "loss": 1.2507479190826416, "step": 42725 }, { "epoch": 3.368230193141506, "grad_norm": 0.4387512505054474, "learning_rate": 0.0005, "loss": 1.184254765510559, "step": 42726 }, { "epoch": 3.3683090264091446, "grad_norm": 0.47934895753860474, "learning_rate": 0.0005, "loss": 1.332252025604248, "step": 42727 }, { "epoch": 3.368387859676784, "grad_norm": 0.6258806586265564, "learning_rate": 0.0005, "loss": 1.2360460758209229, "step": 42728 }, { "epoch": 3.3684666929444225, "grad_norm": 0.5174769163131714, "learning_rate": 0.0005, "loss": 1.4156330823898315, "step": 42729 }, { "epoch": 3.3685455262120616, "grad_norm": 0.49932661652565, "learning_rate": 0.0005, "loss": 1.1838972568511963, "step": 42730 }, { "epoch": 3.3686243594797003, "grad_norm": 0.5134908556938171, "learning_rate": 0.0005, "loss": 1.3113200664520264, "step": 42731 }, { "epoch": 3.3687031927473394, "grad_norm": 0.5646105408668518, "learning_rate": 0.0005, "loss": 1.2466559410095215, "step": 42732 }, { "epoch": 3.368782026014978, "grad_norm": 0.47157955169677734, "learning_rate": 0.0005, "loss": 1.1618846654891968, "step": 42733 }, { "epoch": 3.3688608592826172, "grad_norm": 0.5555227994918823, "learning_rate": 0.0005, "loss": 1.3175066709518433, "step": 42734 }, { "epoch": 3.3689396925502564, "grad_norm": 0.5475462675094604, "learning_rate": 0.0005, "loss": 1.2119379043579102, "step": 42735 }, { "epoch": 3.369018525817895, "grad_norm": 0.45017555356025696, "learning_rate": 0.0005, "loss": 1.117638111114502, "step": 42736 }, { "epoch": 3.369097359085534, "grad_norm": 0.43165498971939087, "learning_rate": 0.0005, "loss": 1.2134617567062378, "step": 42737 }, { "epoch": 3.369176192353173, "grad_norm": 0.4319680333137512, "learning_rate": 0.0005, "loss": 1.2319955825805664, "step": 42738 }, { "epoch": 3.369255025620812, "grad_norm": 0.5996122360229492, "learning_rate": 0.0005, "loss": 1.3071680068969727, "step": 42739 }, { "epoch": 3.369333858888451, "grad_norm": 0.5207662582397461, "learning_rate": 0.0005, "loss": 1.189711093902588, "step": 42740 }, { "epoch": 3.36941269215609, "grad_norm": 0.6647897362709045, "learning_rate": 0.0005, "loss": 1.4939603805541992, "step": 42741 }, { "epoch": 3.369491525423729, "grad_norm": 0.4489682912826538, "learning_rate": 0.0005, "loss": 1.0167560577392578, "step": 42742 }, { "epoch": 3.3695703586913677, "grad_norm": 0.523892879486084, "learning_rate": 0.0005, "loss": 1.3714749813079834, "step": 42743 }, { "epoch": 3.369649191959007, "grad_norm": 0.5270841717720032, "learning_rate": 0.0005, "loss": 1.2866957187652588, "step": 42744 }, { "epoch": 3.3697280252266455, "grad_norm": 0.43048912286758423, "learning_rate": 0.0005, "loss": 1.4011616706848145, "step": 42745 }, { "epoch": 3.3698068584942846, "grad_norm": 0.5687236785888672, "learning_rate": 0.0005, "loss": 1.402856469154358, "step": 42746 }, { "epoch": 3.3698856917619233, "grad_norm": 0.5080990791320801, "learning_rate": 0.0005, "loss": 1.4134653806686401, "step": 42747 }, { "epoch": 3.3699645250295625, "grad_norm": 0.45467108488082886, "learning_rate": 0.0005, "loss": 1.3715882301330566, "step": 42748 }, { "epoch": 3.3700433582972016, "grad_norm": 0.5469592809677124, "learning_rate": 0.0005, "loss": 1.3651955127716064, "step": 42749 }, { "epoch": 3.3701221915648403, "grad_norm": 0.5367214679718018, "learning_rate": 0.0005, "loss": 1.4645564556121826, "step": 42750 }, { "epoch": 3.3702010248324794, "grad_norm": 0.47985514998435974, "learning_rate": 0.0005, "loss": 1.101765513420105, "step": 42751 }, { "epoch": 3.370279858100118, "grad_norm": 0.47543567419052124, "learning_rate": 0.0005, "loss": 1.4240217208862305, "step": 42752 }, { "epoch": 3.3703586913677572, "grad_norm": 0.5311878323554993, "learning_rate": 0.0005, "loss": 1.14536714553833, "step": 42753 }, { "epoch": 3.3704375246353964, "grad_norm": 0.5482659935951233, "learning_rate": 0.0005, "loss": 1.3307403326034546, "step": 42754 }, { "epoch": 3.370516357903035, "grad_norm": 0.4788704812526703, "learning_rate": 0.0005, "loss": 1.2844860553741455, "step": 42755 }, { "epoch": 3.370595191170674, "grad_norm": 0.5273629426956177, "learning_rate": 0.0005, "loss": 1.2010059356689453, "step": 42756 }, { "epoch": 3.370674024438313, "grad_norm": 0.6133926510810852, "learning_rate": 0.0005, "loss": 1.2802919149398804, "step": 42757 }, { "epoch": 3.370752857705952, "grad_norm": 0.3918972611427307, "learning_rate": 0.0005, "loss": 1.2200193405151367, "step": 42758 }, { "epoch": 3.3708316909735907, "grad_norm": 0.467598557472229, "learning_rate": 0.0005, "loss": 1.1304227113723755, "step": 42759 }, { "epoch": 3.37091052424123, "grad_norm": 0.5105477571487427, "learning_rate": 0.0005, "loss": 1.2717382907867432, "step": 42760 }, { "epoch": 3.3709893575088685, "grad_norm": 0.63238525390625, "learning_rate": 0.0005, "loss": 1.195911169052124, "step": 42761 }, { "epoch": 3.3710681907765077, "grad_norm": 0.8153982758522034, "learning_rate": 0.0005, "loss": 1.4093685150146484, "step": 42762 }, { "epoch": 3.371147024044147, "grad_norm": 0.5498484373092651, "learning_rate": 0.0005, "loss": 1.4170148372650146, "step": 42763 }, { "epoch": 3.3712258573117855, "grad_norm": 0.5318129062652588, "learning_rate": 0.0005, "loss": 1.2734915018081665, "step": 42764 }, { "epoch": 3.3713046905794246, "grad_norm": 0.4593013823032379, "learning_rate": 0.0005, "loss": 1.209279179573059, "step": 42765 }, { "epoch": 3.3713835238470633, "grad_norm": 0.4470321238040924, "learning_rate": 0.0005, "loss": 1.154147744178772, "step": 42766 }, { "epoch": 3.3714623571147024, "grad_norm": 0.5336978435516357, "learning_rate": 0.0005, "loss": 1.1496555805206299, "step": 42767 }, { "epoch": 3.3715411903823416, "grad_norm": 0.4664359390735626, "learning_rate": 0.0005, "loss": 1.2345335483551025, "step": 42768 }, { "epoch": 3.3716200236499803, "grad_norm": 0.5083812475204468, "learning_rate": 0.0005, "loss": 1.3059520721435547, "step": 42769 }, { "epoch": 3.3716988569176194, "grad_norm": 0.46298947930336, "learning_rate": 0.0005, "loss": 1.2311437129974365, "step": 42770 }, { "epoch": 3.371777690185258, "grad_norm": 0.5323328971862793, "learning_rate": 0.0005, "loss": 1.2766114473342896, "step": 42771 }, { "epoch": 3.371856523452897, "grad_norm": 0.6708410382270813, "learning_rate": 0.0005, "loss": 1.4217617511749268, "step": 42772 }, { "epoch": 3.371935356720536, "grad_norm": 0.5187616944313049, "learning_rate": 0.0005, "loss": 1.280672311782837, "step": 42773 }, { "epoch": 3.372014189988175, "grad_norm": 0.41442862153053284, "learning_rate": 0.0005, "loss": 1.2884440422058105, "step": 42774 }, { "epoch": 3.3720930232558137, "grad_norm": 0.4962190091609955, "learning_rate": 0.0005, "loss": 1.480453610420227, "step": 42775 }, { "epoch": 3.372171856523453, "grad_norm": 0.4237549304962158, "learning_rate": 0.0005, "loss": 1.2636704444885254, "step": 42776 }, { "epoch": 3.372250689791092, "grad_norm": 0.3645356297492981, "learning_rate": 0.0005, "loss": 1.0172054767608643, "step": 42777 }, { "epoch": 3.3723295230587307, "grad_norm": 0.5637170076370239, "learning_rate": 0.0005, "loss": 1.14998459815979, "step": 42778 }, { "epoch": 3.37240835632637, "grad_norm": 0.41562002897262573, "learning_rate": 0.0005, "loss": 1.2139838933944702, "step": 42779 }, { "epoch": 3.3724871895940085, "grad_norm": 0.41756582260131836, "learning_rate": 0.0005, "loss": 1.049862265586853, "step": 42780 }, { "epoch": 3.3725660228616476, "grad_norm": 0.47430047392845154, "learning_rate": 0.0005, "loss": 1.2831497192382812, "step": 42781 }, { "epoch": 3.3726448561292868, "grad_norm": 0.5228739976882935, "learning_rate": 0.0005, "loss": 1.2270268201828003, "step": 42782 }, { "epoch": 3.3727236893969255, "grad_norm": 0.44267717003822327, "learning_rate": 0.0005, "loss": 1.335407018661499, "step": 42783 }, { "epoch": 3.3728025226645646, "grad_norm": 0.44082242250442505, "learning_rate": 0.0005, "loss": 1.342983603477478, "step": 42784 }, { "epoch": 3.3728813559322033, "grad_norm": 0.5093362927436829, "learning_rate": 0.0005, "loss": 1.4474096298217773, "step": 42785 }, { "epoch": 3.3729601891998424, "grad_norm": 0.44448596239089966, "learning_rate": 0.0005, "loss": 1.198551893234253, "step": 42786 }, { "epoch": 3.373039022467481, "grad_norm": 0.48142775893211365, "learning_rate": 0.0005, "loss": 1.337808609008789, "step": 42787 }, { "epoch": 3.3731178557351202, "grad_norm": 0.42334142327308655, "learning_rate": 0.0005, "loss": 1.33787202835083, "step": 42788 }, { "epoch": 3.373196689002759, "grad_norm": 0.6311774849891663, "learning_rate": 0.0005, "loss": 1.3140294551849365, "step": 42789 }, { "epoch": 3.373275522270398, "grad_norm": 0.6559377312660217, "learning_rate": 0.0005, "loss": 1.4328532218933105, "step": 42790 }, { "epoch": 3.373354355538037, "grad_norm": 0.488839328289032, "learning_rate": 0.0005, "loss": 1.3477776050567627, "step": 42791 }, { "epoch": 3.373433188805676, "grad_norm": 0.515553891658783, "learning_rate": 0.0005, "loss": 1.4563922882080078, "step": 42792 }, { "epoch": 3.373512022073315, "grad_norm": 0.4441821277141571, "learning_rate": 0.0005, "loss": 1.2295745611190796, "step": 42793 }, { "epoch": 3.3735908553409537, "grad_norm": 0.4890415668487549, "learning_rate": 0.0005, "loss": 1.357094168663025, "step": 42794 }, { "epoch": 3.373669688608593, "grad_norm": 0.47839412093162537, "learning_rate": 0.0005, "loss": 1.1638929843902588, "step": 42795 }, { "epoch": 3.373748521876232, "grad_norm": 0.5529443025588989, "learning_rate": 0.0005, "loss": 1.3884788751602173, "step": 42796 }, { "epoch": 3.3738273551438707, "grad_norm": 0.44054171442985535, "learning_rate": 0.0005, "loss": 1.2014423608779907, "step": 42797 }, { "epoch": 3.37390618841151, "grad_norm": 0.48361295461654663, "learning_rate": 0.0005, "loss": 1.3259309530258179, "step": 42798 }, { "epoch": 3.3739850216791485, "grad_norm": 0.5527488589286804, "learning_rate": 0.0005, "loss": 1.3503401279449463, "step": 42799 }, { "epoch": 3.3740638549467876, "grad_norm": 0.44753265380859375, "learning_rate": 0.0005, "loss": 1.120535135269165, "step": 42800 }, { "epoch": 3.3741426882144263, "grad_norm": 0.46062150597572327, "learning_rate": 0.0005, "loss": 1.201432228088379, "step": 42801 }, { "epoch": 3.3742215214820654, "grad_norm": 0.5317683815956116, "learning_rate": 0.0005, "loss": 1.3654000759124756, "step": 42802 }, { "epoch": 3.374300354749704, "grad_norm": 0.46212124824523926, "learning_rate": 0.0005, "loss": 1.1860288381576538, "step": 42803 }, { "epoch": 3.3743791880173433, "grad_norm": 0.5459513068199158, "learning_rate": 0.0005, "loss": 1.472851037979126, "step": 42804 }, { "epoch": 3.3744580212849824, "grad_norm": 0.4765467345714569, "learning_rate": 0.0005, "loss": 1.1349146366119385, "step": 42805 }, { "epoch": 3.374536854552621, "grad_norm": 0.42061445116996765, "learning_rate": 0.0005, "loss": 1.0178886651992798, "step": 42806 }, { "epoch": 3.3746156878202602, "grad_norm": 0.6424551010131836, "learning_rate": 0.0005, "loss": 1.2886137962341309, "step": 42807 }, { "epoch": 3.374694521087899, "grad_norm": 0.6036191582679749, "learning_rate": 0.0005, "loss": 1.4006767272949219, "step": 42808 }, { "epoch": 3.374773354355538, "grad_norm": 0.4940524995326996, "learning_rate": 0.0005, "loss": 1.331521987915039, "step": 42809 }, { "epoch": 3.374852187623177, "grad_norm": 0.5501613020896912, "learning_rate": 0.0005, "loss": 1.3973908424377441, "step": 42810 }, { "epoch": 3.374931020890816, "grad_norm": 0.48393014073371887, "learning_rate": 0.0005, "loss": 1.1015985012054443, "step": 42811 }, { "epoch": 3.375009854158455, "grad_norm": 0.5356689691543579, "learning_rate": 0.0005, "loss": 1.2038524150848389, "step": 42812 }, { "epoch": 3.3750886874260937, "grad_norm": 0.5107334852218628, "learning_rate": 0.0005, "loss": 1.1711491346359253, "step": 42813 }, { "epoch": 3.375167520693733, "grad_norm": 0.4566698372364044, "learning_rate": 0.0005, "loss": 1.3666363954544067, "step": 42814 }, { "epoch": 3.3752463539613715, "grad_norm": 0.49790245294570923, "learning_rate": 0.0005, "loss": 1.2800908088684082, "step": 42815 }, { "epoch": 3.3753251872290106, "grad_norm": 0.5511098504066467, "learning_rate": 0.0005, "loss": 1.3071324825286865, "step": 42816 }, { "epoch": 3.3754040204966493, "grad_norm": 0.45330753922462463, "learning_rate": 0.0005, "loss": 1.2211737632751465, "step": 42817 }, { "epoch": 3.3754828537642885, "grad_norm": 0.43981221318244934, "learning_rate": 0.0005, "loss": 1.2635455131530762, "step": 42818 }, { "epoch": 3.3755616870319276, "grad_norm": 0.47840428352355957, "learning_rate": 0.0005, "loss": 1.2890397310256958, "step": 42819 }, { "epoch": 3.3756405202995663, "grad_norm": 0.4505106508731842, "learning_rate": 0.0005, "loss": 1.2911796569824219, "step": 42820 }, { "epoch": 3.3757193535672054, "grad_norm": 0.5972142815589905, "learning_rate": 0.0005, "loss": 1.3318955898284912, "step": 42821 }, { "epoch": 3.375798186834844, "grad_norm": 0.5480053424835205, "learning_rate": 0.0005, "loss": 1.3835585117340088, "step": 42822 }, { "epoch": 3.3758770201024833, "grad_norm": 0.5415819883346558, "learning_rate": 0.0005, "loss": 1.3574612140655518, "step": 42823 }, { "epoch": 3.3759558533701224, "grad_norm": 0.422006219625473, "learning_rate": 0.0005, "loss": 1.149503469467163, "step": 42824 }, { "epoch": 3.376034686637761, "grad_norm": 0.5420789122581482, "learning_rate": 0.0005, "loss": 1.3481152057647705, "step": 42825 }, { "epoch": 3.3761135199054, "grad_norm": 0.4639992117881775, "learning_rate": 0.0005, "loss": 1.2497022151947021, "step": 42826 }, { "epoch": 3.376192353173039, "grad_norm": 0.46759840846061707, "learning_rate": 0.0005, "loss": 1.256515383720398, "step": 42827 }, { "epoch": 3.376271186440678, "grad_norm": 0.45526883006095886, "learning_rate": 0.0005, "loss": 1.2594964504241943, "step": 42828 }, { "epoch": 3.3763500197083167, "grad_norm": 0.5108746290206909, "learning_rate": 0.0005, "loss": 1.3625974655151367, "step": 42829 }, { "epoch": 3.376428852975956, "grad_norm": 0.4445032775402069, "learning_rate": 0.0005, "loss": 1.2260551452636719, "step": 42830 }, { "epoch": 3.376507686243595, "grad_norm": 0.4996316730976105, "learning_rate": 0.0005, "loss": 1.541151762008667, "step": 42831 }, { "epoch": 3.3765865195112337, "grad_norm": 0.4364610016345978, "learning_rate": 0.0005, "loss": 1.3146061897277832, "step": 42832 }, { "epoch": 3.376665352778873, "grad_norm": 0.5269635915756226, "learning_rate": 0.0005, "loss": 1.1798679828643799, "step": 42833 }, { "epoch": 3.3767441860465115, "grad_norm": 0.4787026047706604, "learning_rate": 0.0005, "loss": 1.2452456951141357, "step": 42834 }, { "epoch": 3.3768230193141506, "grad_norm": 0.6281418204307556, "learning_rate": 0.0005, "loss": 1.3502700328826904, "step": 42835 }, { "epoch": 3.3769018525817893, "grad_norm": 0.4259750545024872, "learning_rate": 0.0005, "loss": 1.1633979082107544, "step": 42836 }, { "epoch": 3.3769806858494285, "grad_norm": 0.4712848663330078, "learning_rate": 0.0005, "loss": 1.2694059610366821, "step": 42837 }, { "epoch": 3.3770595191170676, "grad_norm": 0.44155630469322205, "learning_rate": 0.0005, "loss": 1.141998052597046, "step": 42838 }, { "epoch": 3.3771383523847063, "grad_norm": 0.4190366268157959, "learning_rate": 0.0005, "loss": 1.1478880643844604, "step": 42839 }, { "epoch": 3.3772171856523454, "grad_norm": 0.6171144247055054, "learning_rate": 0.0005, "loss": 1.357060194015503, "step": 42840 }, { "epoch": 3.377296018919984, "grad_norm": 0.5394687056541443, "learning_rate": 0.0005, "loss": 1.1611440181732178, "step": 42841 }, { "epoch": 3.3773748521876232, "grad_norm": 0.4728753864765167, "learning_rate": 0.0005, "loss": 1.2199385166168213, "step": 42842 }, { "epoch": 3.3774536854552624, "grad_norm": 0.46291980147361755, "learning_rate": 0.0005, "loss": 1.360127568244934, "step": 42843 }, { "epoch": 3.377532518722901, "grad_norm": 0.57915860414505, "learning_rate": 0.0005, "loss": 1.2978156805038452, "step": 42844 }, { "epoch": 3.37761135199054, "grad_norm": 0.5113946199417114, "learning_rate": 0.0005, "loss": 1.4204692840576172, "step": 42845 }, { "epoch": 3.377690185258179, "grad_norm": 0.49936312437057495, "learning_rate": 0.0005, "loss": 1.281846046447754, "step": 42846 }, { "epoch": 3.377769018525818, "grad_norm": 0.5263186693191528, "learning_rate": 0.0005, "loss": 1.4433852434158325, "step": 42847 }, { "epoch": 3.3778478517934567, "grad_norm": 0.47524717450141907, "learning_rate": 0.0005, "loss": 1.2324374914169312, "step": 42848 }, { "epoch": 3.377926685061096, "grad_norm": 0.44961124658584595, "learning_rate": 0.0005, "loss": 1.3266243934631348, "step": 42849 }, { "epoch": 3.3780055183287345, "grad_norm": 0.47695955634117126, "learning_rate": 0.0005, "loss": 1.1410961151123047, "step": 42850 }, { "epoch": 3.3780843515963737, "grad_norm": 0.5139963626861572, "learning_rate": 0.0005, "loss": 1.2540576457977295, "step": 42851 }, { "epoch": 3.378163184864013, "grad_norm": 0.5133485198020935, "learning_rate": 0.0005, "loss": 1.2720866203308105, "step": 42852 }, { "epoch": 3.3782420181316515, "grad_norm": 0.6609495878219604, "learning_rate": 0.0005, "loss": 1.2936184406280518, "step": 42853 }, { "epoch": 3.3783208513992906, "grad_norm": 0.43140363693237305, "learning_rate": 0.0005, "loss": 1.156329870223999, "step": 42854 }, { "epoch": 3.3783996846669293, "grad_norm": 0.3909825086593628, "learning_rate": 0.0005, "loss": 1.0846514701843262, "step": 42855 }, { "epoch": 3.3784785179345684, "grad_norm": 0.4138539135456085, "learning_rate": 0.0005, "loss": 1.1095702648162842, "step": 42856 }, { "epoch": 3.3785573512022076, "grad_norm": 0.5996120572090149, "learning_rate": 0.0005, "loss": 1.3059988021850586, "step": 42857 }, { "epoch": 3.3786361844698463, "grad_norm": 0.5881601572036743, "learning_rate": 0.0005, "loss": 1.2333400249481201, "step": 42858 }, { "epoch": 3.3787150177374854, "grad_norm": 0.4995691478252411, "learning_rate": 0.0005, "loss": 1.323555827140808, "step": 42859 }, { "epoch": 3.378793851005124, "grad_norm": 0.4871627986431122, "learning_rate": 0.0005, "loss": 1.3140785694122314, "step": 42860 }, { "epoch": 3.378872684272763, "grad_norm": 0.5886855721473694, "learning_rate": 0.0005, "loss": 1.3296728134155273, "step": 42861 }, { "epoch": 3.378951517540402, "grad_norm": 0.5099838376045227, "learning_rate": 0.0005, "loss": 1.2047864198684692, "step": 42862 }, { "epoch": 3.379030350808041, "grad_norm": 0.696169912815094, "learning_rate": 0.0005, "loss": 1.3312020301818848, "step": 42863 }, { "epoch": 3.3791091840756797, "grad_norm": 0.45464956760406494, "learning_rate": 0.0005, "loss": 1.2165578603744507, "step": 42864 }, { "epoch": 3.379188017343319, "grad_norm": 0.5420998930931091, "learning_rate": 0.0005, "loss": 1.264709234237671, "step": 42865 }, { "epoch": 3.379266850610958, "grad_norm": 0.46000251173973083, "learning_rate": 0.0005, "loss": 1.336728572845459, "step": 42866 }, { "epoch": 3.3793456838785967, "grad_norm": 0.4990459978580475, "learning_rate": 0.0005, "loss": 1.3565726280212402, "step": 42867 }, { "epoch": 3.379424517146236, "grad_norm": 0.44284898042678833, "learning_rate": 0.0005, "loss": 1.1539262533187866, "step": 42868 }, { "epoch": 3.3795033504138745, "grad_norm": 0.6354230642318726, "learning_rate": 0.0005, "loss": 1.2586474418640137, "step": 42869 }, { "epoch": 3.3795821836815136, "grad_norm": 0.5144823789596558, "learning_rate": 0.0005, "loss": 1.2123782634735107, "step": 42870 }, { "epoch": 3.3796610169491528, "grad_norm": 0.5990080237388611, "learning_rate": 0.0005, "loss": 1.4102293252944946, "step": 42871 }, { "epoch": 3.3797398502167915, "grad_norm": 0.44201546907424927, "learning_rate": 0.0005, "loss": 1.3320038318634033, "step": 42872 }, { "epoch": 3.3798186834844306, "grad_norm": 0.5438020825386047, "learning_rate": 0.0005, "loss": 1.2250851392745972, "step": 42873 }, { "epoch": 3.3798975167520693, "grad_norm": 0.5183516144752502, "learning_rate": 0.0005, "loss": 1.4879062175750732, "step": 42874 }, { "epoch": 3.3799763500197084, "grad_norm": 0.5287479758262634, "learning_rate": 0.0005, "loss": 1.162198543548584, "step": 42875 }, { "epoch": 3.380055183287347, "grad_norm": 0.46451547741889954, "learning_rate": 0.0005, "loss": 1.324660062789917, "step": 42876 }, { "epoch": 3.3801340165549862, "grad_norm": 0.5380974411964417, "learning_rate": 0.0005, "loss": 1.1129224300384521, "step": 42877 }, { "epoch": 3.380212849822625, "grad_norm": 0.6042094230651855, "learning_rate": 0.0005, "loss": 1.1739054918289185, "step": 42878 }, { "epoch": 3.380291683090264, "grad_norm": 0.4952510595321655, "learning_rate": 0.0005, "loss": 1.3501410484313965, "step": 42879 }, { "epoch": 3.380370516357903, "grad_norm": 0.5227723717689514, "learning_rate": 0.0005, "loss": 1.2236969470977783, "step": 42880 }, { "epoch": 3.380449349625542, "grad_norm": 0.5257379412651062, "learning_rate": 0.0005, "loss": 1.3271231651306152, "step": 42881 }, { "epoch": 3.380528182893181, "grad_norm": 0.5114490985870361, "learning_rate": 0.0005, "loss": 1.3269689083099365, "step": 42882 }, { "epoch": 3.3806070161608197, "grad_norm": 0.6126664876937866, "learning_rate": 0.0005, "loss": 1.4509780406951904, "step": 42883 }, { "epoch": 3.380685849428459, "grad_norm": 0.4814017117023468, "learning_rate": 0.0005, "loss": 1.1692466735839844, "step": 42884 }, { "epoch": 3.380764682696098, "grad_norm": 0.5138307809829712, "learning_rate": 0.0005, "loss": 1.363472819328308, "step": 42885 }, { "epoch": 3.3808435159637367, "grad_norm": 0.508761465549469, "learning_rate": 0.0005, "loss": 1.2337857484817505, "step": 42886 }, { "epoch": 3.380922349231376, "grad_norm": 0.4780846834182739, "learning_rate": 0.0005, "loss": 1.217907428741455, "step": 42887 }, { "epoch": 3.3810011824990145, "grad_norm": 0.5064906477928162, "learning_rate": 0.0005, "loss": 1.1913338899612427, "step": 42888 }, { "epoch": 3.3810800157666536, "grad_norm": 0.5975874066352844, "learning_rate": 0.0005, "loss": 1.3747847080230713, "step": 42889 }, { "epoch": 3.3811588490342923, "grad_norm": 0.4398951232433319, "learning_rate": 0.0005, "loss": 1.1869721412658691, "step": 42890 }, { "epoch": 3.3812376823019314, "grad_norm": 0.4590243399143219, "learning_rate": 0.0005, "loss": 1.2005215883255005, "step": 42891 }, { "epoch": 3.38131651556957, "grad_norm": 0.4110982418060303, "learning_rate": 0.0005, "loss": 1.3311529159545898, "step": 42892 }, { "epoch": 3.3813953488372093, "grad_norm": 0.43194228410720825, "learning_rate": 0.0005, "loss": 1.2706037759780884, "step": 42893 }, { "epoch": 3.3814741821048484, "grad_norm": 0.484311044216156, "learning_rate": 0.0005, "loss": 1.317152976989746, "step": 42894 }, { "epoch": 3.381553015372487, "grad_norm": 0.5456624627113342, "learning_rate": 0.0005, "loss": 1.4094114303588867, "step": 42895 }, { "epoch": 3.3816318486401262, "grad_norm": 0.515613853931427, "learning_rate": 0.0005, "loss": 1.2672532796859741, "step": 42896 }, { "epoch": 3.381710681907765, "grad_norm": 0.46437591314315796, "learning_rate": 0.0005, "loss": 1.2805418968200684, "step": 42897 }, { "epoch": 3.381789515175404, "grad_norm": 0.5128554105758667, "learning_rate": 0.0005, "loss": 1.3243563175201416, "step": 42898 }, { "epoch": 3.381868348443043, "grad_norm": 0.5240716338157654, "learning_rate": 0.0005, "loss": 1.1425738334655762, "step": 42899 }, { "epoch": 3.381947181710682, "grad_norm": 0.5185249447822571, "learning_rate": 0.0005, "loss": 1.5220870971679688, "step": 42900 }, { "epoch": 3.382026014978321, "grad_norm": 0.6433537602424622, "learning_rate": 0.0005, "loss": 1.140296220779419, "step": 42901 }, { "epoch": 3.3821048482459597, "grad_norm": 0.5540969371795654, "learning_rate": 0.0005, "loss": 1.357919454574585, "step": 42902 }, { "epoch": 3.382183681513599, "grad_norm": 0.47773393988609314, "learning_rate": 0.0005, "loss": 1.253182291984558, "step": 42903 }, { "epoch": 3.3822625147812375, "grad_norm": 0.46942979097366333, "learning_rate": 0.0005, "loss": 1.3422164916992188, "step": 42904 }, { "epoch": 3.3823413480488767, "grad_norm": 0.4979836940765381, "learning_rate": 0.0005, "loss": 1.1730549335479736, "step": 42905 }, { "epoch": 3.3824201813165153, "grad_norm": 0.8690134882926941, "learning_rate": 0.0005, "loss": 1.218980073928833, "step": 42906 }, { "epoch": 3.3824990145841545, "grad_norm": 0.4391475021839142, "learning_rate": 0.0005, "loss": 1.1302675008773804, "step": 42907 }, { "epoch": 3.3825778478517936, "grad_norm": 0.5739551782608032, "learning_rate": 0.0005, "loss": 1.3768985271453857, "step": 42908 }, { "epoch": 3.3826566811194323, "grad_norm": 0.4228488802909851, "learning_rate": 0.0005, "loss": 1.261170744895935, "step": 42909 }, { "epoch": 3.3827355143870714, "grad_norm": 0.5253121256828308, "learning_rate": 0.0005, "loss": 1.2740241289138794, "step": 42910 }, { "epoch": 3.38281434765471, "grad_norm": 0.47046366333961487, "learning_rate": 0.0005, "loss": 1.262708067893982, "step": 42911 }, { "epoch": 3.3828931809223493, "grad_norm": 0.4947512745857239, "learning_rate": 0.0005, "loss": 1.2787630558013916, "step": 42912 }, { "epoch": 3.3829720141899884, "grad_norm": 0.4086579382419586, "learning_rate": 0.0005, "loss": 1.212930679321289, "step": 42913 }, { "epoch": 3.383050847457627, "grad_norm": 0.5280857086181641, "learning_rate": 0.0005, "loss": 1.3096426725387573, "step": 42914 }, { "epoch": 3.383129680725266, "grad_norm": 0.3993413746356964, "learning_rate": 0.0005, "loss": 1.186711311340332, "step": 42915 }, { "epoch": 3.383208513992905, "grad_norm": 0.4954410791397095, "learning_rate": 0.0005, "loss": 1.25325345993042, "step": 42916 }, { "epoch": 3.383287347260544, "grad_norm": 0.538329541683197, "learning_rate": 0.0005, "loss": 1.324583649635315, "step": 42917 }, { "epoch": 3.3833661805281827, "grad_norm": 0.4508334696292877, "learning_rate": 0.0005, "loss": 1.1452898979187012, "step": 42918 }, { "epoch": 3.383445013795822, "grad_norm": 0.5303165316581726, "learning_rate": 0.0005, "loss": 1.1993601322174072, "step": 42919 }, { "epoch": 3.3835238470634605, "grad_norm": 0.5069688558578491, "learning_rate": 0.0005, "loss": 1.2239060401916504, "step": 42920 }, { "epoch": 3.3836026803310997, "grad_norm": 0.48063239455223083, "learning_rate": 0.0005, "loss": 1.2914764881134033, "step": 42921 }, { "epoch": 3.383681513598739, "grad_norm": 0.6211482882499695, "learning_rate": 0.0005, "loss": 1.4075225591659546, "step": 42922 }, { "epoch": 3.3837603468663775, "grad_norm": 0.5903183221817017, "learning_rate": 0.0005, "loss": 1.5337107181549072, "step": 42923 }, { "epoch": 3.3838391801340166, "grad_norm": 0.47365331649780273, "learning_rate": 0.0005, "loss": 1.4819207191467285, "step": 42924 }, { "epoch": 3.3839180134016553, "grad_norm": 0.4855276942253113, "learning_rate": 0.0005, "loss": 1.3399791717529297, "step": 42925 }, { "epoch": 3.3839968466692945, "grad_norm": 0.4394040107727051, "learning_rate": 0.0005, "loss": 1.2184447050094604, "step": 42926 }, { "epoch": 3.3840756799369336, "grad_norm": 0.4569913148880005, "learning_rate": 0.0005, "loss": 1.1546449661254883, "step": 42927 }, { "epoch": 3.3841545132045723, "grad_norm": 0.45420917868614197, "learning_rate": 0.0005, "loss": 1.1178669929504395, "step": 42928 }, { "epoch": 3.3842333464722114, "grad_norm": 0.4602128565311432, "learning_rate": 0.0005, "loss": 1.2321850061416626, "step": 42929 }, { "epoch": 3.38431217973985, "grad_norm": 0.6433141231536865, "learning_rate": 0.0005, "loss": 1.398860216140747, "step": 42930 }, { "epoch": 3.3843910130074892, "grad_norm": 0.5009050369262695, "learning_rate": 0.0005, "loss": 1.13065767288208, "step": 42931 }, { "epoch": 3.384469846275128, "grad_norm": 0.5160729289054871, "learning_rate": 0.0005, "loss": 1.1670838594436646, "step": 42932 }, { "epoch": 3.384548679542767, "grad_norm": 0.5606216192245483, "learning_rate": 0.0005, "loss": 1.2936594486236572, "step": 42933 }, { "epoch": 3.3846275128104057, "grad_norm": 0.4594409167766571, "learning_rate": 0.0005, "loss": 1.1827208995819092, "step": 42934 }, { "epoch": 3.384706346078045, "grad_norm": 0.4184475541114807, "learning_rate": 0.0005, "loss": 1.1313084363937378, "step": 42935 }, { "epoch": 3.384785179345684, "grad_norm": 0.4794382154941559, "learning_rate": 0.0005, "loss": 1.2831881046295166, "step": 42936 }, { "epoch": 3.3848640126133227, "grad_norm": 0.46876993775367737, "learning_rate": 0.0005, "loss": 1.4484312534332275, "step": 42937 }, { "epoch": 3.384942845880962, "grad_norm": 0.45917582511901855, "learning_rate": 0.0005, "loss": 1.1601462364196777, "step": 42938 }, { "epoch": 3.3850216791486005, "grad_norm": 0.4759270250797272, "learning_rate": 0.0005, "loss": 1.1656787395477295, "step": 42939 }, { "epoch": 3.3851005124162397, "grad_norm": 0.41277799010276794, "learning_rate": 0.0005, "loss": 1.2594730854034424, "step": 42940 }, { "epoch": 3.385179345683879, "grad_norm": 0.4741520583629608, "learning_rate": 0.0005, "loss": 1.202451229095459, "step": 42941 }, { "epoch": 3.3852581789515175, "grad_norm": 0.48170405626296997, "learning_rate": 0.0005, "loss": 1.333081841468811, "step": 42942 }, { "epoch": 3.3853370122191566, "grad_norm": 0.489498496055603, "learning_rate": 0.0005, "loss": 1.4130537509918213, "step": 42943 }, { "epoch": 3.3854158454867953, "grad_norm": 0.4432399868965149, "learning_rate": 0.0005, "loss": 1.2975263595581055, "step": 42944 }, { "epoch": 3.3854946787544344, "grad_norm": 0.5329186916351318, "learning_rate": 0.0005, "loss": 1.260764718055725, "step": 42945 }, { "epoch": 3.385573512022073, "grad_norm": 0.6379342675209045, "learning_rate": 0.0005, "loss": 1.3596585988998413, "step": 42946 }, { "epoch": 3.3856523452897123, "grad_norm": 0.4463348984718323, "learning_rate": 0.0005, "loss": 1.1701104640960693, "step": 42947 }, { "epoch": 3.3857311785573514, "grad_norm": 0.5027973055839539, "learning_rate": 0.0005, "loss": 1.1768429279327393, "step": 42948 }, { "epoch": 3.38581001182499, "grad_norm": 0.5995754599571228, "learning_rate": 0.0005, "loss": 1.3096282482147217, "step": 42949 }, { "epoch": 3.385888845092629, "grad_norm": 0.7183791399002075, "learning_rate": 0.0005, "loss": 1.2382643222808838, "step": 42950 }, { "epoch": 3.385967678360268, "grad_norm": 0.5986639857292175, "learning_rate": 0.0005, "loss": 1.2787503004074097, "step": 42951 }, { "epoch": 3.386046511627907, "grad_norm": 0.4699837565422058, "learning_rate": 0.0005, "loss": 1.1927939653396606, "step": 42952 }, { "epoch": 3.3861253448955457, "grad_norm": 0.4926482141017914, "learning_rate": 0.0005, "loss": 1.2833709716796875, "step": 42953 }, { "epoch": 3.386204178163185, "grad_norm": 0.44846516847610474, "learning_rate": 0.0005, "loss": 1.1317944526672363, "step": 42954 }, { "epoch": 3.386283011430824, "grad_norm": 0.6004607081413269, "learning_rate": 0.0005, "loss": 1.0991846323013306, "step": 42955 }, { "epoch": 3.3863618446984627, "grad_norm": 0.44134533405303955, "learning_rate": 0.0005, "loss": 1.2598662376403809, "step": 42956 }, { "epoch": 3.386440677966102, "grad_norm": 0.4671059250831604, "learning_rate": 0.0005, "loss": 1.1474051475524902, "step": 42957 }, { "epoch": 3.3865195112337405, "grad_norm": 0.40538978576660156, "learning_rate": 0.0005, "loss": 1.2611737251281738, "step": 42958 }, { "epoch": 3.3865983445013796, "grad_norm": 0.46888795495033264, "learning_rate": 0.0005, "loss": 1.242919683456421, "step": 42959 }, { "epoch": 3.3866771777690183, "grad_norm": 0.5342541337013245, "learning_rate": 0.0005, "loss": 1.2816156148910522, "step": 42960 }, { "epoch": 3.3867560110366575, "grad_norm": 0.5308070778846741, "learning_rate": 0.0005, "loss": 1.3441379070281982, "step": 42961 }, { "epoch": 3.3868348443042966, "grad_norm": 0.49677392840385437, "learning_rate": 0.0005, "loss": 1.208540439605713, "step": 42962 }, { "epoch": 3.3869136775719353, "grad_norm": 0.4675261080265045, "learning_rate": 0.0005, "loss": 1.3727359771728516, "step": 42963 }, { "epoch": 3.3869925108395744, "grad_norm": 0.44637706875801086, "learning_rate": 0.0005, "loss": 1.0806260108947754, "step": 42964 }, { "epoch": 3.387071344107213, "grad_norm": 0.519047737121582, "learning_rate": 0.0005, "loss": 1.1747909784317017, "step": 42965 }, { "epoch": 3.3871501773748522, "grad_norm": 0.5348745584487915, "learning_rate": 0.0005, "loss": 1.4690196514129639, "step": 42966 }, { "epoch": 3.387229010642491, "grad_norm": 0.6364859342575073, "learning_rate": 0.0005, "loss": 1.293869972229004, "step": 42967 }, { "epoch": 3.38730784391013, "grad_norm": 0.46796154975891113, "learning_rate": 0.0005, "loss": 1.2633626461029053, "step": 42968 }, { "epoch": 3.387386677177769, "grad_norm": 0.45918264985084534, "learning_rate": 0.0005, "loss": 1.255361795425415, "step": 42969 }, { "epoch": 3.387465510445408, "grad_norm": 0.4809223413467407, "learning_rate": 0.0005, "loss": 1.2038793563842773, "step": 42970 }, { "epoch": 3.387544343713047, "grad_norm": 0.4248284697532654, "learning_rate": 0.0005, "loss": 1.303848385810852, "step": 42971 }, { "epoch": 3.3876231769806857, "grad_norm": 2.0107316970825195, "learning_rate": 0.0005, "loss": 1.2608534097671509, "step": 42972 }, { "epoch": 3.387702010248325, "grad_norm": 0.4597564935684204, "learning_rate": 0.0005, "loss": 1.258497953414917, "step": 42973 }, { "epoch": 3.387780843515964, "grad_norm": 0.43492308259010315, "learning_rate": 0.0005, "loss": 1.2865387201309204, "step": 42974 }, { "epoch": 3.3878596767836027, "grad_norm": 0.6324629783630371, "learning_rate": 0.0005, "loss": 1.3113716840744019, "step": 42975 }, { "epoch": 3.387938510051242, "grad_norm": 0.5448060631752014, "learning_rate": 0.0005, "loss": 1.2489359378814697, "step": 42976 }, { "epoch": 3.3880173433188805, "grad_norm": 0.5294236540794373, "learning_rate": 0.0005, "loss": 1.2824856042861938, "step": 42977 }, { "epoch": 3.3880961765865196, "grad_norm": 0.4514388144016266, "learning_rate": 0.0005, "loss": 1.245897889137268, "step": 42978 }, { "epoch": 3.3881750098541583, "grad_norm": 0.4351203739643097, "learning_rate": 0.0005, "loss": 1.2152303457260132, "step": 42979 }, { "epoch": 3.3882538431217974, "grad_norm": 0.4931900203227997, "learning_rate": 0.0005, "loss": 1.2923940420150757, "step": 42980 }, { "epoch": 3.388332676389436, "grad_norm": 0.5818624496459961, "learning_rate": 0.0005, "loss": 1.3699686527252197, "step": 42981 }, { "epoch": 3.3884115096570753, "grad_norm": 0.49640753865242004, "learning_rate": 0.0005, "loss": 1.4160078763961792, "step": 42982 }, { "epoch": 3.3884903429247144, "grad_norm": 0.4454677104949951, "learning_rate": 0.0005, "loss": 1.2840490341186523, "step": 42983 }, { "epoch": 3.388569176192353, "grad_norm": 0.48881015181541443, "learning_rate": 0.0005, "loss": 1.3476519584655762, "step": 42984 }, { "epoch": 3.3886480094599922, "grad_norm": 0.5144036412239075, "learning_rate": 0.0005, "loss": 1.1408151388168335, "step": 42985 }, { "epoch": 3.388726842727631, "grad_norm": 0.4959622025489807, "learning_rate": 0.0005, "loss": 1.226448893547058, "step": 42986 }, { "epoch": 3.38880567599527, "grad_norm": 0.5274341702461243, "learning_rate": 0.0005, "loss": 1.2188160419464111, "step": 42987 }, { "epoch": 3.388884509262909, "grad_norm": 0.49901100993156433, "learning_rate": 0.0005, "loss": 1.3446288108825684, "step": 42988 }, { "epoch": 3.388963342530548, "grad_norm": 0.537609338760376, "learning_rate": 0.0005, "loss": 1.1985888481140137, "step": 42989 }, { "epoch": 3.389042175798187, "grad_norm": 0.48212799429893494, "learning_rate": 0.0005, "loss": 1.2432324886322021, "step": 42990 }, { "epoch": 3.3891210090658257, "grad_norm": 0.47148042917251587, "learning_rate": 0.0005, "loss": 1.278635025024414, "step": 42991 }, { "epoch": 3.389199842333465, "grad_norm": 0.49152204394340515, "learning_rate": 0.0005, "loss": 1.140925407409668, "step": 42992 }, { "epoch": 3.3892786756011035, "grad_norm": 0.44610801339149475, "learning_rate": 0.0005, "loss": 1.2703948020935059, "step": 42993 }, { "epoch": 3.3893575088687427, "grad_norm": 0.45522254705429077, "learning_rate": 0.0005, "loss": 1.3401217460632324, "step": 42994 }, { "epoch": 3.3894363421363813, "grad_norm": 0.49322113394737244, "learning_rate": 0.0005, "loss": 1.3290042877197266, "step": 42995 }, { "epoch": 3.3895151754040205, "grad_norm": 0.4769757091999054, "learning_rate": 0.0005, "loss": 1.4196885824203491, "step": 42996 }, { "epoch": 3.3895940086716596, "grad_norm": 0.4891662895679474, "learning_rate": 0.0005, "loss": 1.072469711303711, "step": 42997 }, { "epoch": 3.3896728419392983, "grad_norm": 0.4308532178401947, "learning_rate": 0.0005, "loss": 1.0943801403045654, "step": 42998 }, { "epoch": 3.3897516752069374, "grad_norm": 0.5360533595085144, "learning_rate": 0.0005, "loss": 1.2972984313964844, "step": 42999 }, { "epoch": 3.389830508474576, "grad_norm": 0.4296112358570099, "learning_rate": 0.0005, "loss": 1.0363867282867432, "step": 43000 }, { "epoch": 3.3899093417422153, "grad_norm": 0.5319511890411377, "learning_rate": 0.0005, "loss": 1.4444477558135986, "step": 43001 }, { "epoch": 3.3899881750098544, "grad_norm": 0.7304995059967041, "learning_rate": 0.0005, "loss": 1.3503012657165527, "step": 43002 }, { "epoch": 3.390067008277493, "grad_norm": 0.4831066429615021, "learning_rate": 0.0005, "loss": 1.223555564880371, "step": 43003 }, { "epoch": 3.390145841545132, "grad_norm": 0.4610758125782013, "learning_rate": 0.0005, "loss": 1.1733195781707764, "step": 43004 }, { "epoch": 3.390224674812771, "grad_norm": 0.5321521162986755, "learning_rate": 0.0005, "loss": 1.2335625886917114, "step": 43005 }, { "epoch": 3.39030350808041, "grad_norm": 0.5799230933189392, "learning_rate": 0.0005, "loss": 1.2829875946044922, "step": 43006 }, { "epoch": 3.3903823413480487, "grad_norm": 0.4441988468170166, "learning_rate": 0.0005, "loss": 1.1514816284179688, "step": 43007 }, { "epoch": 3.390461174615688, "grad_norm": 0.5307936072349548, "learning_rate": 0.0005, "loss": 1.2551279067993164, "step": 43008 }, { "epoch": 3.3905400078833265, "grad_norm": 0.4529013931751251, "learning_rate": 0.0005, "loss": 1.2524305582046509, "step": 43009 }, { "epoch": 3.3906188411509657, "grad_norm": 0.4784030616283417, "learning_rate": 0.0005, "loss": 1.5105048418045044, "step": 43010 }, { "epoch": 3.390697674418605, "grad_norm": 0.43424922227859497, "learning_rate": 0.0005, "loss": 1.2500327825546265, "step": 43011 }, { "epoch": 3.3907765076862435, "grad_norm": 0.45295068621635437, "learning_rate": 0.0005, "loss": 1.2875268459320068, "step": 43012 }, { "epoch": 3.3908553409538826, "grad_norm": 0.4739123582839966, "learning_rate": 0.0005, "loss": 1.2167813777923584, "step": 43013 }, { "epoch": 3.3909341742215213, "grad_norm": 0.4856279194355011, "learning_rate": 0.0005, "loss": 1.3170125484466553, "step": 43014 }, { "epoch": 3.3910130074891605, "grad_norm": 0.5250580310821533, "learning_rate": 0.0005, "loss": 1.2165716886520386, "step": 43015 }, { "epoch": 3.3910918407567996, "grad_norm": 0.5334176421165466, "learning_rate": 0.0005, "loss": 1.7002724409103394, "step": 43016 }, { "epoch": 3.3911706740244383, "grad_norm": 0.43595045804977417, "learning_rate": 0.0005, "loss": 1.3124330043792725, "step": 43017 }, { "epoch": 3.3912495072920774, "grad_norm": 0.4470703899860382, "learning_rate": 0.0005, "loss": 1.3712022304534912, "step": 43018 }, { "epoch": 3.391328340559716, "grad_norm": 0.4765996038913727, "learning_rate": 0.0005, "loss": 1.2485190629959106, "step": 43019 }, { "epoch": 3.3914071738273552, "grad_norm": 0.508426308631897, "learning_rate": 0.0005, "loss": 1.2665512561798096, "step": 43020 }, { "epoch": 3.391486007094994, "grad_norm": 0.488626092672348, "learning_rate": 0.0005, "loss": 1.306536316871643, "step": 43021 }, { "epoch": 3.391564840362633, "grad_norm": 0.5643113851547241, "learning_rate": 0.0005, "loss": 1.2898917198181152, "step": 43022 }, { "epoch": 3.3916436736302717, "grad_norm": 0.4852949380874634, "learning_rate": 0.0005, "loss": 1.3747122287750244, "step": 43023 }, { "epoch": 3.391722506897911, "grad_norm": 0.5511708855628967, "learning_rate": 0.0005, "loss": 1.277306318283081, "step": 43024 }, { "epoch": 3.39180134016555, "grad_norm": 0.5311281085014343, "learning_rate": 0.0005, "loss": 1.285245418548584, "step": 43025 }, { "epoch": 3.3918801734331887, "grad_norm": 0.5046687722206116, "learning_rate": 0.0005, "loss": 1.1663075685501099, "step": 43026 }, { "epoch": 3.391959006700828, "grad_norm": 0.45308560132980347, "learning_rate": 0.0005, "loss": 1.219616413116455, "step": 43027 }, { "epoch": 3.3920378399684665, "grad_norm": 0.43412643671035767, "learning_rate": 0.0005, "loss": 1.0643737316131592, "step": 43028 }, { "epoch": 3.3921166732361057, "grad_norm": 0.40689700841903687, "learning_rate": 0.0005, "loss": 1.0931706428527832, "step": 43029 }, { "epoch": 3.392195506503745, "grad_norm": 0.46808162331581116, "learning_rate": 0.0005, "loss": 1.0225625038146973, "step": 43030 }, { "epoch": 3.3922743397713835, "grad_norm": 0.5904189348220825, "learning_rate": 0.0005, "loss": 1.4046368598937988, "step": 43031 }, { "epoch": 3.3923531730390226, "grad_norm": 0.5326830148696899, "learning_rate": 0.0005, "loss": 1.2859574556350708, "step": 43032 }, { "epoch": 3.3924320063066613, "grad_norm": 0.42079034447669983, "learning_rate": 0.0005, "loss": 1.1527762413024902, "step": 43033 }, { "epoch": 3.3925108395743004, "grad_norm": 0.5842688679695129, "learning_rate": 0.0005, "loss": 1.4635140895843506, "step": 43034 }, { "epoch": 3.392589672841939, "grad_norm": 0.46538060903549194, "learning_rate": 0.0005, "loss": 1.3000651597976685, "step": 43035 }, { "epoch": 3.3926685061095783, "grad_norm": 0.448314905166626, "learning_rate": 0.0005, "loss": 1.233133316040039, "step": 43036 }, { "epoch": 3.392747339377217, "grad_norm": 0.45866745710372925, "learning_rate": 0.0005, "loss": 1.2543349266052246, "step": 43037 }, { "epoch": 3.392826172644856, "grad_norm": 0.4241413474082947, "learning_rate": 0.0005, "loss": 1.1650657653808594, "step": 43038 }, { "epoch": 3.392905005912495, "grad_norm": 0.5040099024772644, "learning_rate": 0.0005, "loss": 1.1165351867675781, "step": 43039 }, { "epoch": 3.392983839180134, "grad_norm": 0.5100919604301453, "learning_rate": 0.0005, "loss": 1.1568856239318848, "step": 43040 }, { "epoch": 3.393062672447773, "grad_norm": 0.49700307846069336, "learning_rate": 0.0005, "loss": 1.2701435089111328, "step": 43041 }, { "epoch": 3.3931415057154117, "grad_norm": 0.4747270345687866, "learning_rate": 0.0005, "loss": 1.3208905458450317, "step": 43042 }, { "epoch": 3.393220338983051, "grad_norm": 0.5912426114082336, "learning_rate": 0.0005, "loss": 1.2424967288970947, "step": 43043 }, { "epoch": 3.39329917225069, "grad_norm": 0.5265390872955322, "learning_rate": 0.0005, "loss": 1.3595540523529053, "step": 43044 }, { "epoch": 3.3933780055183287, "grad_norm": 0.46084222197532654, "learning_rate": 0.0005, "loss": 1.1598767042160034, "step": 43045 }, { "epoch": 3.393456838785968, "grad_norm": 0.4486065208911896, "learning_rate": 0.0005, "loss": 1.3386948108673096, "step": 43046 }, { "epoch": 3.3935356720536065, "grad_norm": 0.4324910342693329, "learning_rate": 0.0005, "loss": 1.2350002527236938, "step": 43047 }, { "epoch": 3.3936145053212456, "grad_norm": 0.7243903875350952, "learning_rate": 0.0005, "loss": 1.2131807804107666, "step": 43048 }, { "epoch": 3.3936933385888843, "grad_norm": 0.48492130637168884, "learning_rate": 0.0005, "loss": 1.1040669679641724, "step": 43049 }, { "epoch": 3.3937721718565235, "grad_norm": 0.4828647971153259, "learning_rate": 0.0005, "loss": 1.40885329246521, "step": 43050 }, { "epoch": 3.393851005124162, "grad_norm": 0.4020845890045166, "learning_rate": 0.0005, "loss": 1.1067705154418945, "step": 43051 }, { "epoch": 3.3939298383918013, "grad_norm": 0.44597944617271423, "learning_rate": 0.0005, "loss": 1.357936143875122, "step": 43052 }, { "epoch": 3.3940086716594404, "grad_norm": 0.5886920094490051, "learning_rate": 0.0005, "loss": 1.4325218200683594, "step": 43053 }, { "epoch": 3.394087504927079, "grad_norm": 0.3785911798477173, "learning_rate": 0.0005, "loss": 1.066876769065857, "step": 43054 }, { "epoch": 3.3941663381947182, "grad_norm": 0.46366655826568604, "learning_rate": 0.0005, "loss": 1.2873237133026123, "step": 43055 }, { "epoch": 3.394245171462357, "grad_norm": 0.5266091227531433, "learning_rate": 0.0005, "loss": 1.3925058841705322, "step": 43056 }, { "epoch": 3.394324004729996, "grad_norm": 0.46059271693229675, "learning_rate": 0.0005, "loss": 1.1501286029815674, "step": 43057 }, { "epoch": 3.394402837997635, "grad_norm": 0.49920031428337097, "learning_rate": 0.0005, "loss": 1.414326786994934, "step": 43058 }, { "epoch": 3.394481671265274, "grad_norm": 0.4778110980987549, "learning_rate": 0.0005, "loss": 1.1376674175262451, "step": 43059 }, { "epoch": 3.394560504532913, "grad_norm": 0.6241567730903625, "learning_rate": 0.0005, "loss": 1.3279998302459717, "step": 43060 }, { "epoch": 3.3946393378005517, "grad_norm": 0.628321647644043, "learning_rate": 0.0005, "loss": 1.291524052619934, "step": 43061 }, { "epoch": 3.394718171068191, "grad_norm": 0.4849708378314972, "learning_rate": 0.0005, "loss": 1.2087430953979492, "step": 43062 }, { "epoch": 3.3947970043358295, "grad_norm": 0.43647998571395874, "learning_rate": 0.0005, "loss": 1.1487293243408203, "step": 43063 }, { "epoch": 3.3948758376034687, "grad_norm": 0.6418412327766418, "learning_rate": 0.0005, "loss": 1.1039085388183594, "step": 43064 }, { "epoch": 3.3949546708711074, "grad_norm": 0.48030513525009155, "learning_rate": 0.0005, "loss": 1.2839455604553223, "step": 43065 }, { "epoch": 3.3950335041387465, "grad_norm": 0.4783203601837158, "learning_rate": 0.0005, "loss": 1.3634531497955322, "step": 43066 }, { "epoch": 3.3951123374063856, "grad_norm": 0.38026440143585205, "learning_rate": 0.0005, "loss": 1.1397429704666138, "step": 43067 }, { "epoch": 3.3951911706740243, "grad_norm": 0.5188527703285217, "learning_rate": 0.0005, "loss": 1.4559223651885986, "step": 43068 }, { "epoch": 3.3952700039416635, "grad_norm": 0.5633111596107483, "learning_rate": 0.0005, "loss": 1.404477834701538, "step": 43069 }, { "epoch": 3.395348837209302, "grad_norm": 0.5345276594161987, "learning_rate": 0.0005, "loss": 1.4107353687286377, "step": 43070 }, { "epoch": 3.3954276704769413, "grad_norm": 0.48649272322654724, "learning_rate": 0.0005, "loss": 1.2318143844604492, "step": 43071 }, { "epoch": 3.3955065037445804, "grad_norm": 0.432620108127594, "learning_rate": 0.0005, "loss": 1.085463047027588, "step": 43072 }, { "epoch": 3.395585337012219, "grad_norm": 0.6247115731239319, "learning_rate": 0.0005, "loss": 1.2850933074951172, "step": 43073 }, { "epoch": 3.3956641702798582, "grad_norm": 0.41960445046424866, "learning_rate": 0.0005, "loss": 1.139694094657898, "step": 43074 }, { "epoch": 3.395743003547497, "grad_norm": 0.5062206387519836, "learning_rate": 0.0005, "loss": 1.3167731761932373, "step": 43075 }, { "epoch": 3.395821836815136, "grad_norm": 0.5189768075942993, "learning_rate": 0.0005, "loss": 1.400637149810791, "step": 43076 }, { "epoch": 3.3959006700827747, "grad_norm": 0.4485537111759186, "learning_rate": 0.0005, "loss": 1.3338820934295654, "step": 43077 }, { "epoch": 3.395979503350414, "grad_norm": 0.5563005805015564, "learning_rate": 0.0005, "loss": 1.1110121011734009, "step": 43078 }, { "epoch": 3.396058336618053, "grad_norm": 0.5598925352096558, "learning_rate": 0.0005, "loss": 1.4119421243667603, "step": 43079 }, { "epoch": 3.3961371698856917, "grad_norm": 0.7195712327957153, "learning_rate": 0.0005, "loss": 1.408370018005371, "step": 43080 }, { "epoch": 3.396216003153331, "grad_norm": 0.44627198576927185, "learning_rate": 0.0005, "loss": 1.2002428770065308, "step": 43081 }, { "epoch": 3.3962948364209695, "grad_norm": 0.4514961838722229, "learning_rate": 0.0005, "loss": 1.1935875415802002, "step": 43082 }, { "epoch": 3.3963736696886087, "grad_norm": 0.4590444564819336, "learning_rate": 0.0005, "loss": 1.306006908416748, "step": 43083 }, { "epoch": 3.3964525029562473, "grad_norm": 0.5022982954978943, "learning_rate": 0.0005, "loss": 1.3533084392547607, "step": 43084 }, { "epoch": 3.3965313362238865, "grad_norm": 0.48064106702804565, "learning_rate": 0.0005, "loss": 1.2666103839874268, "step": 43085 }, { "epoch": 3.3966101694915256, "grad_norm": 0.5452141761779785, "learning_rate": 0.0005, "loss": 1.2902209758758545, "step": 43086 }, { "epoch": 3.3966890027591643, "grad_norm": 0.5503678917884827, "learning_rate": 0.0005, "loss": 1.4186558723449707, "step": 43087 }, { "epoch": 3.3967678360268034, "grad_norm": 0.40691378712654114, "learning_rate": 0.0005, "loss": 1.270787000656128, "step": 43088 }, { "epoch": 3.396846669294442, "grad_norm": 0.5038998126983643, "learning_rate": 0.0005, "loss": 1.1967453956604004, "step": 43089 }, { "epoch": 3.3969255025620813, "grad_norm": 0.5634019374847412, "learning_rate": 0.0005, "loss": 1.2279855012893677, "step": 43090 }, { "epoch": 3.3970043358297204, "grad_norm": 0.4721539616584778, "learning_rate": 0.0005, "loss": 1.2156920433044434, "step": 43091 }, { "epoch": 3.397083169097359, "grad_norm": 0.48590943217277527, "learning_rate": 0.0005, "loss": 1.1263668537139893, "step": 43092 }, { "epoch": 3.397162002364998, "grad_norm": 0.5240930318832397, "learning_rate": 0.0005, "loss": 1.323089361190796, "step": 43093 }, { "epoch": 3.397240835632637, "grad_norm": 0.41750568151474, "learning_rate": 0.0005, "loss": 1.1948628425598145, "step": 43094 }, { "epoch": 3.397319668900276, "grad_norm": 0.46351757645606995, "learning_rate": 0.0005, "loss": 1.255082130432129, "step": 43095 }, { "epoch": 3.3973985021679147, "grad_norm": 0.5729573965072632, "learning_rate": 0.0005, "loss": 1.4309070110321045, "step": 43096 }, { "epoch": 3.397477335435554, "grad_norm": 0.6884874701499939, "learning_rate": 0.0005, "loss": 1.204970121383667, "step": 43097 }, { "epoch": 3.3975561687031925, "grad_norm": 0.5264874696731567, "learning_rate": 0.0005, "loss": 1.2272365093231201, "step": 43098 }, { "epoch": 3.3976350019708317, "grad_norm": 0.6031551957130432, "learning_rate": 0.0005, "loss": 1.2811626195907593, "step": 43099 }, { "epoch": 3.397713835238471, "grad_norm": 0.5604971647262573, "learning_rate": 0.0005, "loss": 1.3150596618652344, "step": 43100 }, { "epoch": 3.3977926685061095, "grad_norm": 0.5103268027305603, "learning_rate": 0.0005, "loss": 1.1751186847686768, "step": 43101 }, { "epoch": 3.3978715017737486, "grad_norm": 0.6389739513397217, "learning_rate": 0.0005, "loss": 1.4040782451629639, "step": 43102 }, { "epoch": 3.3979503350413873, "grad_norm": 0.5680297017097473, "learning_rate": 0.0005, "loss": 1.2658452987670898, "step": 43103 }, { "epoch": 3.3980291683090265, "grad_norm": 0.4795224070549011, "learning_rate": 0.0005, "loss": 1.2397000789642334, "step": 43104 }, { "epoch": 3.3981080015766656, "grad_norm": 0.4639188051223755, "learning_rate": 0.0005, "loss": 1.2150424718856812, "step": 43105 }, { "epoch": 3.3981868348443043, "grad_norm": 1.3010094165802002, "learning_rate": 0.0005, "loss": 1.3053276538848877, "step": 43106 }, { "epoch": 3.3982656681119434, "grad_norm": 0.4583951234817505, "learning_rate": 0.0005, "loss": 1.3004671335220337, "step": 43107 }, { "epoch": 3.398344501379582, "grad_norm": 0.45090150833129883, "learning_rate": 0.0005, "loss": 1.277626395225525, "step": 43108 }, { "epoch": 3.3984233346472212, "grad_norm": 0.539652943611145, "learning_rate": 0.0005, "loss": 1.3626128435134888, "step": 43109 }, { "epoch": 3.39850216791486, "grad_norm": 0.4974725544452667, "learning_rate": 0.0005, "loss": 1.3749451637268066, "step": 43110 }, { "epoch": 3.398581001182499, "grad_norm": 0.5150968432426453, "learning_rate": 0.0005, "loss": 1.1496400833129883, "step": 43111 }, { "epoch": 3.3986598344501378, "grad_norm": 0.44783955812454224, "learning_rate": 0.0005, "loss": 1.1124539375305176, "step": 43112 }, { "epoch": 3.398738667717777, "grad_norm": 0.5421816110610962, "learning_rate": 0.0005, "loss": 1.3256430625915527, "step": 43113 }, { "epoch": 3.398817500985416, "grad_norm": 0.44179990887641907, "learning_rate": 0.0005, "loss": 1.1817530393600464, "step": 43114 }, { "epoch": 3.3988963342530547, "grad_norm": 0.5769544243812561, "learning_rate": 0.0005, "loss": 1.4409167766571045, "step": 43115 }, { "epoch": 3.398975167520694, "grad_norm": 0.5366190671920776, "learning_rate": 0.0005, "loss": 1.2606443166732788, "step": 43116 }, { "epoch": 3.3990540007883325, "grad_norm": 0.48972994089126587, "learning_rate": 0.0005, "loss": 1.2132848501205444, "step": 43117 }, { "epoch": 3.3991328340559717, "grad_norm": 0.5023558139801025, "learning_rate": 0.0005, "loss": 1.1538063287734985, "step": 43118 }, { "epoch": 3.399211667323611, "grad_norm": 0.5581278800964355, "learning_rate": 0.0005, "loss": 1.3721680641174316, "step": 43119 }, { "epoch": 3.3992905005912495, "grad_norm": 0.5702700614929199, "learning_rate": 0.0005, "loss": 1.251692533493042, "step": 43120 }, { "epoch": 3.3993693338588886, "grad_norm": 0.45294031500816345, "learning_rate": 0.0005, "loss": 1.1645110845565796, "step": 43121 }, { "epoch": 3.3994481671265273, "grad_norm": 0.4556853473186493, "learning_rate": 0.0005, "loss": 1.263972520828247, "step": 43122 }, { "epoch": 3.3995270003941664, "grad_norm": 0.5209488868713379, "learning_rate": 0.0005, "loss": 1.3098610639572144, "step": 43123 }, { "epoch": 3.399605833661805, "grad_norm": 0.42414143681526184, "learning_rate": 0.0005, "loss": 1.135055422782898, "step": 43124 }, { "epoch": 3.3996846669294443, "grad_norm": 0.5863847136497498, "learning_rate": 0.0005, "loss": 1.2382763624191284, "step": 43125 }, { "epoch": 3.399763500197083, "grad_norm": 0.512074887752533, "learning_rate": 0.0005, "loss": 1.201033353805542, "step": 43126 }, { "epoch": 3.399842333464722, "grad_norm": 0.4952307343482971, "learning_rate": 0.0005, "loss": 1.3247498273849487, "step": 43127 }, { "epoch": 3.399921166732361, "grad_norm": 0.39877623319625854, "learning_rate": 0.0005, "loss": 1.1654994487762451, "step": 43128 }, { "epoch": 3.4, "grad_norm": 0.5222644209861755, "learning_rate": 0.0005, "loss": 1.22652268409729, "step": 43129 }, { "epoch": 3.400078833267639, "grad_norm": 0.4179462790489197, "learning_rate": 0.0005, "loss": 1.2366998195648193, "step": 43130 }, { "epoch": 3.4001576665352777, "grad_norm": 0.45438894629478455, "learning_rate": 0.0005, "loss": 1.3454850912094116, "step": 43131 }, { "epoch": 3.400236499802917, "grad_norm": 0.5760956406593323, "learning_rate": 0.0005, "loss": 1.4603610038757324, "step": 43132 }, { "epoch": 3.400315333070556, "grad_norm": 0.48151567578315735, "learning_rate": 0.0005, "loss": 1.3365161418914795, "step": 43133 }, { "epoch": 3.4003941663381947, "grad_norm": 0.6124932169914246, "learning_rate": 0.0005, "loss": 1.4233286380767822, "step": 43134 }, { "epoch": 3.400472999605834, "grad_norm": 0.46739062666893005, "learning_rate": 0.0005, "loss": 1.219480276107788, "step": 43135 }, { "epoch": 3.4005518328734725, "grad_norm": 0.46931666135787964, "learning_rate": 0.0005, "loss": 1.1775426864624023, "step": 43136 }, { "epoch": 3.4006306661411116, "grad_norm": 0.6289806962013245, "learning_rate": 0.0005, "loss": 1.1705594062805176, "step": 43137 }, { "epoch": 3.4007094994087503, "grad_norm": 0.4665806293487549, "learning_rate": 0.0005, "loss": 1.1612416505813599, "step": 43138 }, { "epoch": 3.4007883326763895, "grad_norm": 0.4594792127609253, "learning_rate": 0.0005, "loss": 1.147134780883789, "step": 43139 }, { "epoch": 3.400867165944028, "grad_norm": 0.47995424270629883, "learning_rate": 0.0005, "loss": 1.1844274997711182, "step": 43140 }, { "epoch": 3.4009459992116673, "grad_norm": 0.5035226941108704, "learning_rate": 0.0005, "loss": 1.2289371490478516, "step": 43141 }, { "epoch": 3.4010248324793064, "grad_norm": 0.46939975023269653, "learning_rate": 0.0005, "loss": 1.2400152683258057, "step": 43142 }, { "epoch": 3.401103665746945, "grad_norm": 0.5746283531188965, "learning_rate": 0.0005, "loss": 1.363795518875122, "step": 43143 }, { "epoch": 3.4011824990145842, "grad_norm": 0.5195564031600952, "learning_rate": 0.0005, "loss": 1.3502061367034912, "step": 43144 }, { "epoch": 3.401261332282223, "grad_norm": 0.511913537979126, "learning_rate": 0.0005, "loss": 1.2367515563964844, "step": 43145 }, { "epoch": 3.401340165549862, "grad_norm": 0.411700576543808, "learning_rate": 0.0005, "loss": 1.2151527404785156, "step": 43146 }, { "epoch": 3.401418998817501, "grad_norm": 0.4017581343650818, "learning_rate": 0.0005, "loss": 1.1187912225723267, "step": 43147 }, { "epoch": 3.40149783208514, "grad_norm": 0.48031359910964966, "learning_rate": 0.0005, "loss": 1.2738008499145508, "step": 43148 }, { "epoch": 3.401576665352779, "grad_norm": 0.4417458176612854, "learning_rate": 0.0005, "loss": 1.1416993141174316, "step": 43149 }, { "epoch": 3.4016554986204177, "grad_norm": 0.4954468011856079, "learning_rate": 0.0005, "loss": 1.2824310064315796, "step": 43150 }, { "epoch": 3.401734331888057, "grad_norm": 0.5240968465805054, "learning_rate": 0.0005, "loss": 1.3221179246902466, "step": 43151 }, { "epoch": 3.4018131651556955, "grad_norm": 0.4368787407875061, "learning_rate": 0.0005, "loss": 1.0632867813110352, "step": 43152 }, { "epoch": 3.4018919984233347, "grad_norm": 0.4743981659412384, "learning_rate": 0.0005, "loss": 1.2790229320526123, "step": 43153 }, { "epoch": 3.4019708316909734, "grad_norm": 0.42822715640068054, "learning_rate": 0.0005, "loss": 1.1363884210586548, "step": 43154 }, { "epoch": 3.4020496649586125, "grad_norm": 0.4651559591293335, "learning_rate": 0.0005, "loss": 1.1818803548812866, "step": 43155 }, { "epoch": 3.4021284982262516, "grad_norm": 0.49408969283103943, "learning_rate": 0.0005, "loss": 1.338665246963501, "step": 43156 }, { "epoch": 3.4022073314938903, "grad_norm": 0.4865461587905884, "learning_rate": 0.0005, "loss": 1.2423096895217896, "step": 43157 }, { "epoch": 3.4022861647615295, "grad_norm": 0.5309321880340576, "learning_rate": 0.0005, "loss": 1.3998626470565796, "step": 43158 }, { "epoch": 3.402364998029168, "grad_norm": 0.6556047201156616, "learning_rate": 0.0005, "loss": 1.2102446556091309, "step": 43159 }, { "epoch": 3.4024438312968073, "grad_norm": 0.5074491500854492, "learning_rate": 0.0005, "loss": 1.5169254541397095, "step": 43160 }, { "epoch": 3.4025226645644464, "grad_norm": 0.5535673499107361, "learning_rate": 0.0005, "loss": 1.3594025373458862, "step": 43161 }, { "epoch": 3.402601497832085, "grad_norm": 0.4724459946155548, "learning_rate": 0.0005, "loss": 1.184880256652832, "step": 43162 }, { "epoch": 3.4026803310997242, "grad_norm": 0.4964372515678406, "learning_rate": 0.0005, "loss": 1.3515719175338745, "step": 43163 }, { "epoch": 3.402759164367363, "grad_norm": 0.7352624535560608, "learning_rate": 0.0005, "loss": 1.4779506921768188, "step": 43164 }, { "epoch": 3.402837997635002, "grad_norm": 0.5396608710289001, "learning_rate": 0.0005, "loss": 1.3525066375732422, "step": 43165 }, { "epoch": 3.4029168309026407, "grad_norm": 0.41166090965270996, "learning_rate": 0.0005, "loss": 1.1737085580825806, "step": 43166 }, { "epoch": 3.40299566417028, "grad_norm": 0.5465892553329468, "learning_rate": 0.0005, "loss": 1.187556266784668, "step": 43167 }, { "epoch": 3.4030744974379186, "grad_norm": 0.4123813509941101, "learning_rate": 0.0005, "loss": 1.1430366039276123, "step": 43168 }, { "epoch": 3.4031533307055577, "grad_norm": 0.5035878419876099, "learning_rate": 0.0005, "loss": 1.3533108234405518, "step": 43169 }, { "epoch": 3.403232163973197, "grad_norm": 0.5513724684715271, "learning_rate": 0.0005, "loss": 1.3635780811309814, "step": 43170 }, { "epoch": 3.4033109972408355, "grad_norm": 0.4296514093875885, "learning_rate": 0.0005, "loss": 1.105316400527954, "step": 43171 }, { "epoch": 3.4033898305084747, "grad_norm": 0.41602855920791626, "learning_rate": 0.0005, "loss": 1.274237871170044, "step": 43172 }, { "epoch": 3.4034686637761133, "grad_norm": 0.5186355113983154, "learning_rate": 0.0005, "loss": 1.34931480884552, "step": 43173 }, { "epoch": 3.4035474970437525, "grad_norm": 0.45837363600730896, "learning_rate": 0.0005, "loss": 1.169926643371582, "step": 43174 }, { "epoch": 3.4036263303113916, "grad_norm": 0.5498846173286438, "learning_rate": 0.0005, "loss": 1.2267324924468994, "step": 43175 }, { "epoch": 3.4037051635790303, "grad_norm": 0.6061344146728516, "learning_rate": 0.0005, "loss": 1.266991376876831, "step": 43176 }, { "epoch": 3.4037839968466694, "grad_norm": 0.5214206576347351, "learning_rate": 0.0005, "loss": 1.381730318069458, "step": 43177 }, { "epoch": 3.403862830114308, "grad_norm": 0.7651605606079102, "learning_rate": 0.0005, "loss": 1.3505885601043701, "step": 43178 }, { "epoch": 3.4039416633819473, "grad_norm": 0.5332364439964294, "learning_rate": 0.0005, "loss": 1.3475639820098877, "step": 43179 }, { "epoch": 3.404020496649586, "grad_norm": 0.626509428024292, "learning_rate": 0.0005, "loss": 1.2808163166046143, "step": 43180 }, { "epoch": 3.404099329917225, "grad_norm": 0.5807203054428101, "learning_rate": 0.0005, "loss": 1.5038154125213623, "step": 43181 }, { "epoch": 3.4041781631848638, "grad_norm": 0.5129910111427307, "learning_rate": 0.0005, "loss": 1.2840518951416016, "step": 43182 }, { "epoch": 3.404256996452503, "grad_norm": 0.4747876822948456, "learning_rate": 0.0005, "loss": 1.4120521545410156, "step": 43183 }, { "epoch": 3.404335829720142, "grad_norm": 0.4799458384513855, "learning_rate": 0.0005, "loss": 1.0978397130966187, "step": 43184 }, { "epoch": 3.4044146629877807, "grad_norm": 0.6267547607421875, "learning_rate": 0.0005, "loss": 1.2781445980072021, "step": 43185 }, { "epoch": 3.40449349625542, "grad_norm": 0.5551178455352783, "learning_rate": 0.0005, "loss": 1.4107136726379395, "step": 43186 }, { "epoch": 3.4045723295230585, "grad_norm": 0.8332844972610474, "learning_rate": 0.0005, "loss": 1.365410566329956, "step": 43187 }, { "epoch": 3.4046511627906977, "grad_norm": 0.5536497831344604, "learning_rate": 0.0005, "loss": 1.3176382780075073, "step": 43188 }, { "epoch": 3.404729996058337, "grad_norm": 0.538564920425415, "learning_rate": 0.0005, "loss": 1.1832199096679688, "step": 43189 }, { "epoch": 3.4048088293259755, "grad_norm": 0.47075140476226807, "learning_rate": 0.0005, "loss": 1.1684861183166504, "step": 43190 }, { "epoch": 3.4048876625936146, "grad_norm": 0.7298790216445923, "learning_rate": 0.0005, "loss": 1.388615369796753, "step": 43191 }, { "epoch": 3.4049664958612533, "grad_norm": 0.5527151823043823, "learning_rate": 0.0005, "loss": 1.20563542842865, "step": 43192 }, { "epoch": 3.4050453291288925, "grad_norm": 0.6362916231155396, "learning_rate": 0.0005, "loss": 1.1769973039627075, "step": 43193 }, { "epoch": 3.405124162396531, "grad_norm": 0.4748608469963074, "learning_rate": 0.0005, "loss": 1.1868396997451782, "step": 43194 }, { "epoch": 3.4052029956641703, "grad_norm": 0.5246323347091675, "learning_rate": 0.0005, "loss": 1.216733694076538, "step": 43195 }, { "epoch": 3.405281828931809, "grad_norm": 0.5286344885826111, "learning_rate": 0.0005, "loss": 1.4029484987258911, "step": 43196 }, { "epoch": 3.405360662199448, "grad_norm": 0.44893378019332886, "learning_rate": 0.0005, "loss": 1.3592690229415894, "step": 43197 }, { "epoch": 3.4054394954670872, "grad_norm": 0.4873732328414917, "learning_rate": 0.0005, "loss": 1.1760916709899902, "step": 43198 }, { "epoch": 3.405518328734726, "grad_norm": 0.448438823223114, "learning_rate": 0.0005, "loss": 1.2256028652191162, "step": 43199 }, { "epoch": 3.405597162002365, "grad_norm": 0.502758800983429, "learning_rate": 0.0005, "loss": 1.2704694271087646, "step": 43200 }, { "epoch": 3.4056759952700038, "grad_norm": 0.6469139456748962, "learning_rate": 0.0005, "loss": 1.2529131174087524, "step": 43201 }, { "epoch": 3.405754828537643, "grad_norm": 0.5182073712348938, "learning_rate": 0.0005, "loss": 1.3364847898483276, "step": 43202 }, { "epoch": 3.405833661805282, "grad_norm": 0.5679963827133179, "learning_rate": 0.0005, "loss": 1.4712722301483154, "step": 43203 }, { "epoch": 3.4059124950729207, "grad_norm": 0.5060663819313049, "learning_rate": 0.0005, "loss": 1.3038980960845947, "step": 43204 }, { "epoch": 3.40599132834056, "grad_norm": 0.4909553527832031, "learning_rate": 0.0005, "loss": 1.2476192712783813, "step": 43205 }, { "epoch": 3.4060701616081985, "grad_norm": 0.5309960246086121, "learning_rate": 0.0005, "loss": 1.301465630531311, "step": 43206 }, { "epoch": 3.4061489948758377, "grad_norm": 0.4662777781486511, "learning_rate": 0.0005, "loss": 1.3436659574508667, "step": 43207 }, { "epoch": 3.4062278281434764, "grad_norm": 0.5272956490516663, "learning_rate": 0.0005, "loss": 1.2236663103103638, "step": 43208 }, { "epoch": 3.4063066614111155, "grad_norm": 0.576332151889801, "learning_rate": 0.0005, "loss": 1.261699914932251, "step": 43209 }, { "epoch": 3.4063854946787546, "grad_norm": 0.5086657404899597, "learning_rate": 0.0005, "loss": 1.3111047744750977, "step": 43210 }, { "epoch": 3.4064643279463933, "grad_norm": 0.5208643674850464, "learning_rate": 0.0005, "loss": 1.133629560470581, "step": 43211 }, { "epoch": 3.4065431612140324, "grad_norm": 0.5426317453384399, "learning_rate": 0.0005, "loss": 1.3557072877883911, "step": 43212 }, { "epoch": 3.406621994481671, "grad_norm": 0.4555416703224182, "learning_rate": 0.0005, "loss": 1.326690673828125, "step": 43213 }, { "epoch": 3.4067008277493103, "grad_norm": 0.4283541738986969, "learning_rate": 0.0005, "loss": 1.1227270364761353, "step": 43214 }, { "epoch": 3.406779661016949, "grad_norm": 0.4351079761981964, "learning_rate": 0.0005, "loss": 1.2628227472305298, "step": 43215 }, { "epoch": 3.406858494284588, "grad_norm": 0.44901591539382935, "learning_rate": 0.0005, "loss": 1.1862527132034302, "step": 43216 }, { "epoch": 3.4069373275522272, "grad_norm": 0.6203770637512207, "learning_rate": 0.0005, "loss": 1.2313193082809448, "step": 43217 }, { "epoch": 3.407016160819866, "grad_norm": 0.5091573596000671, "learning_rate": 0.0005, "loss": 1.3554697036743164, "step": 43218 }, { "epoch": 3.407094994087505, "grad_norm": 0.8393182158470154, "learning_rate": 0.0005, "loss": 1.2760841846466064, "step": 43219 }, { "epoch": 3.4071738273551437, "grad_norm": 0.49308231472969055, "learning_rate": 0.0005, "loss": 1.1734912395477295, "step": 43220 }, { "epoch": 3.407252660622783, "grad_norm": 0.37438908219337463, "learning_rate": 0.0005, "loss": 1.2101085186004639, "step": 43221 }, { "epoch": 3.407331493890422, "grad_norm": 0.5649629235267639, "learning_rate": 0.0005, "loss": 1.2743252515792847, "step": 43222 }, { "epoch": 3.4074103271580607, "grad_norm": 0.36644282937049866, "learning_rate": 0.0005, "loss": 1.1529039144515991, "step": 43223 }, { "epoch": 3.4074891604257, "grad_norm": 0.44590431451797485, "learning_rate": 0.0005, "loss": 1.3108073472976685, "step": 43224 }, { "epoch": 3.4075679936933385, "grad_norm": 0.4463353157043457, "learning_rate": 0.0005, "loss": 1.3087894916534424, "step": 43225 }, { "epoch": 3.4076468269609776, "grad_norm": 0.7014824151992798, "learning_rate": 0.0005, "loss": 1.3878543376922607, "step": 43226 }, { "epoch": 3.4077256602286163, "grad_norm": 0.6466116905212402, "learning_rate": 0.0005, "loss": 1.430259108543396, "step": 43227 }, { "epoch": 3.4078044934962555, "grad_norm": 0.48939061164855957, "learning_rate": 0.0005, "loss": 1.4726256132125854, "step": 43228 }, { "epoch": 3.407883326763894, "grad_norm": 0.543843150138855, "learning_rate": 0.0005, "loss": 1.3225102424621582, "step": 43229 }, { "epoch": 3.4079621600315333, "grad_norm": 0.5119138360023499, "learning_rate": 0.0005, "loss": 1.193932294845581, "step": 43230 }, { "epoch": 3.4080409932991724, "grad_norm": 0.5319437384605408, "learning_rate": 0.0005, "loss": 1.032399296760559, "step": 43231 }, { "epoch": 3.408119826566811, "grad_norm": 0.546912670135498, "learning_rate": 0.0005, "loss": 1.4391282796859741, "step": 43232 }, { "epoch": 3.4081986598344503, "grad_norm": 0.4477301239967346, "learning_rate": 0.0005, "loss": 1.2305901050567627, "step": 43233 }, { "epoch": 3.408277493102089, "grad_norm": 0.5727732181549072, "learning_rate": 0.0005, "loss": 1.1832355260849, "step": 43234 }, { "epoch": 3.408356326369728, "grad_norm": 0.5150095224380493, "learning_rate": 0.0005, "loss": 1.022723913192749, "step": 43235 }, { "epoch": 3.408435159637367, "grad_norm": 0.4685650169849396, "learning_rate": 0.0005, "loss": 1.1655588150024414, "step": 43236 }, { "epoch": 3.408513992905006, "grad_norm": 0.5258322358131409, "learning_rate": 0.0005, "loss": 1.4982941150665283, "step": 43237 }, { "epoch": 3.408592826172645, "grad_norm": 0.5602338314056396, "learning_rate": 0.0005, "loss": 1.2950801849365234, "step": 43238 }, { "epoch": 3.4086716594402837, "grad_norm": 0.6627165675163269, "learning_rate": 0.0005, "loss": 1.5310512781143188, "step": 43239 }, { "epoch": 3.408750492707923, "grad_norm": 0.4596858620643616, "learning_rate": 0.0005, "loss": 1.135974407196045, "step": 43240 }, { "epoch": 3.4088293259755615, "grad_norm": 0.5972371101379395, "learning_rate": 0.0005, "loss": 1.3595232963562012, "step": 43241 }, { "epoch": 3.4089081592432007, "grad_norm": 0.47583508491516113, "learning_rate": 0.0005, "loss": 1.2973144054412842, "step": 43242 }, { "epoch": 3.4089869925108394, "grad_norm": 0.5378425717353821, "learning_rate": 0.0005, "loss": 1.3419768810272217, "step": 43243 }, { "epoch": 3.4090658257784785, "grad_norm": 0.45616576075553894, "learning_rate": 0.0005, "loss": 1.157231092453003, "step": 43244 }, { "epoch": 3.4091446590461176, "grad_norm": 0.5451488494873047, "learning_rate": 0.0005, "loss": 1.3689486980438232, "step": 43245 }, { "epoch": 3.4092234923137563, "grad_norm": 0.49603888392448425, "learning_rate": 0.0005, "loss": 1.3644458055496216, "step": 43246 }, { "epoch": 3.4093023255813955, "grad_norm": 0.7467665672302246, "learning_rate": 0.0005, "loss": 1.246105670928955, "step": 43247 }, { "epoch": 3.409381158849034, "grad_norm": 0.48633477091789246, "learning_rate": 0.0005, "loss": 1.2625097036361694, "step": 43248 }, { "epoch": 3.4094599921166733, "grad_norm": 0.4602932929992676, "learning_rate": 0.0005, "loss": 1.173510193824768, "step": 43249 }, { "epoch": 3.4095388253843124, "grad_norm": 0.47578224539756775, "learning_rate": 0.0005, "loss": 1.1983141899108887, "step": 43250 }, { "epoch": 3.409617658651951, "grad_norm": 0.6848851442337036, "learning_rate": 0.0005, "loss": 1.283352017402649, "step": 43251 }, { "epoch": 3.4096964919195902, "grad_norm": 0.3911362886428833, "learning_rate": 0.0005, "loss": 1.0561301708221436, "step": 43252 }, { "epoch": 3.409775325187229, "grad_norm": 0.4863554835319519, "learning_rate": 0.0005, "loss": 1.255308985710144, "step": 43253 }, { "epoch": 3.409854158454868, "grad_norm": 0.4457938075065613, "learning_rate": 0.0005, "loss": 1.2623231410980225, "step": 43254 }, { "epoch": 3.4099329917225067, "grad_norm": 0.4847605526447296, "learning_rate": 0.0005, "loss": 1.3779903650283813, "step": 43255 }, { "epoch": 3.410011824990146, "grad_norm": 0.4480649530887604, "learning_rate": 0.0005, "loss": 1.2373771667480469, "step": 43256 }, { "epoch": 3.4100906582577846, "grad_norm": 0.46006324887275696, "learning_rate": 0.0005, "loss": 1.1278016567230225, "step": 43257 }, { "epoch": 3.4101694915254237, "grad_norm": 0.5434769988059998, "learning_rate": 0.0005, "loss": 1.5725739002227783, "step": 43258 }, { "epoch": 3.410248324793063, "grad_norm": 0.4590906798839569, "learning_rate": 0.0005, "loss": 1.2574249505996704, "step": 43259 }, { "epoch": 3.4103271580607015, "grad_norm": 0.4790065884590149, "learning_rate": 0.0005, "loss": 1.3064874410629272, "step": 43260 }, { "epoch": 3.4104059913283407, "grad_norm": 0.4858627915382385, "learning_rate": 0.0005, "loss": 1.2699484825134277, "step": 43261 }, { "epoch": 3.4104848245959793, "grad_norm": 0.7333876490592957, "learning_rate": 0.0005, "loss": 1.370261788368225, "step": 43262 }, { "epoch": 3.4105636578636185, "grad_norm": 0.5066922903060913, "learning_rate": 0.0005, "loss": 1.333094835281372, "step": 43263 }, { "epoch": 3.4106424911312576, "grad_norm": 0.43375536799430847, "learning_rate": 0.0005, "loss": 1.196824550628662, "step": 43264 }, { "epoch": 3.4107213243988963, "grad_norm": 0.822099506855011, "learning_rate": 0.0005, "loss": 1.4816458225250244, "step": 43265 }, { "epoch": 3.4108001576665354, "grad_norm": 0.5106678605079651, "learning_rate": 0.0005, "loss": 1.130528450012207, "step": 43266 }, { "epoch": 3.410878990934174, "grad_norm": 0.6889308094978333, "learning_rate": 0.0005, "loss": 1.153183937072754, "step": 43267 }, { "epoch": 3.4109578242018133, "grad_norm": 0.44473811984062195, "learning_rate": 0.0005, "loss": 1.2829909324645996, "step": 43268 }, { "epoch": 3.411036657469452, "grad_norm": 0.5137028694152832, "learning_rate": 0.0005, "loss": 1.3796706199645996, "step": 43269 }, { "epoch": 3.411115490737091, "grad_norm": 0.5347448587417603, "learning_rate": 0.0005, "loss": 1.4136900901794434, "step": 43270 }, { "epoch": 3.4111943240047298, "grad_norm": 0.5015246272087097, "learning_rate": 0.0005, "loss": 1.3695294857025146, "step": 43271 }, { "epoch": 3.411273157272369, "grad_norm": 0.4694265127182007, "learning_rate": 0.0005, "loss": 1.089254379272461, "step": 43272 }, { "epoch": 3.411351990540008, "grad_norm": 0.4653729498386383, "learning_rate": 0.0005, "loss": 1.3488394021987915, "step": 43273 }, { "epoch": 3.4114308238076467, "grad_norm": 0.4488559663295746, "learning_rate": 0.0005, "loss": 1.1482911109924316, "step": 43274 }, { "epoch": 3.411509657075286, "grad_norm": 0.5703432559967041, "learning_rate": 0.0005, "loss": 1.3992023468017578, "step": 43275 }, { "epoch": 3.4115884903429246, "grad_norm": 0.5037311315536499, "learning_rate": 0.0005, "loss": 1.2751703262329102, "step": 43276 }, { "epoch": 3.4116673236105637, "grad_norm": 0.4602081775665283, "learning_rate": 0.0005, "loss": 1.2617601156234741, "step": 43277 }, { "epoch": 3.411746156878203, "grad_norm": 0.5072439908981323, "learning_rate": 0.0005, "loss": 1.09953773021698, "step": 43278 }, { "epoch": 3.4118249901458415, "grad_norm": 0.48537391424179077, "learning_rate": 0.0005, "loss": 1.0691416263580322, "step": 43279 }, { "epoch": 3.4119038234134806, "grad_norm": 0.5251187682151794, "learning_rate": 0.0005, "loss": 1.1400880813598633, "step": 43280 }, { "epoch": 3.4119826566811193, "grad_norm": 0.5042731165885925, "learning_rate": 0.0005, "loss": 1.2078443765640259, "step": 43281 }, { "epoch": 3.4120614899487585, "grad_norm": 0.589286744594574, "learning_rate": 0.0005, "loss": 1.3490149974822998, "step": 43282 }, { "epoch": 3.412140323216397, "grad_norm": 0.48215949535369873, "learning_rate": 0.0005, "loss": 1.2769110202789307, "step": 43283 }, { "epoch": 3.4122191564840363, "grad_norm": 0.5370777249336243, "learning_rate": 0.0005, "loss": 1.3735333681106567, "step": 43284 }, { "epoch": 3.412297989751675, "grad_norm": 0.484984815120697, "learning_rate": 0.0005, "loss": 1.335404872894287, "step": 43285 }, { "epoch": 3.412376823019314, "grad_norm": 0.4856882691383362, "learning_rate": 0.0005, "loss": 1.4358947277069092, "step": 43286 }, { "epoch": 3.4124556562869532, "grad_norm": 0.48828449845314026, "learning_rate": 0.0005, "loss": 1.2647407054901123, "step": 43287 }, { "epoch": 3.412534489554592, "grad_norm": 0.47926944494247437, "learning_rate": 0.0005, "loss": 1.2763508558273315, "step": 43288 }, { "epoch": 3.412613322822231, "grad_norm": 0.5559912323951721, "learning_rate": 0.0005, "loss": 1.4012987613677979, "step": 43289 }, { "epoch": 3.4126921560898698, "grad_norm": 0.41974833607673645, "learning_rate": 0.0005, "loss": 1.2707699537277222, "step": 43290 }, { "epoch": 3.412770989357509, "grad_norm": 0.44604042172431946, "learning_rate": 0.0005, "loss": 1.2132463455200195, "step": 43291 }, { "epoch": 3.412849822625148, "grad_norm": 0.5421559810638428, "learning_rate": 0.0005, "loss": 1.5346003770828247, "step": 43292 }, { "epoch": 3.4129286558927867, "grad_norm": 0.39282768964767456, "learning_rate": 0.0005, "loss": 1.1240389347076416, "step": 43293 }, { "epoch": 3.413007489160426, "grad_norm": 0.4734289050102234, "learning_rate": 0.0005, "loss": 1.199683427810669, "step": 43294 }, { "epoch": 3.4130863224280645, "grad_norm": 0.6374927759170532, "learning_rate": 0.0005, "loss": 1.2647638320922852, "step": 43295 }, { "epoch": 3.4131651556957037, "grad_norm": 0.8702766299247742, "learning_rate": 0.0005, "loss": 1.426741600036621, "step": 43296 }, { "epoch": 3.4132439889633424, "grad_norm": 0.48422789573669434, "learning_rate": 0.0005, "loss": 1.2283830642700195, "step": 43297 }, { "epoch": 3.4133228222309815, "grad_norm": 0.465317964553833, "learning_rate": 0.0005, "loss": 1.1942529678344727, "step": 43298 }, { "epoch": 3.41340165549862, "grad_norm": 0.4575233459472656, "learning_rate": 0.0005, "loss": 1.141023874282837, "step": 43299 }, { "epoch": 3.4134804887662593, "grad_norm": 0.430848628282547, "learning_rate": 0.0005, "loss": 1.2655872106552124, "step": 43300 }, { "epoch": 3.4135593220338984, "grad_norm": 0.5508869290351868, "learning_rate": 0.0005, "loss": 1.2738170623779297, "step": 43301 }, { "epoch": 3.413638155301537, "grad_norm": 0.5397567749023438, "learning_rate": 0.0005, "loss": 1.264992356300354, "step": 43302 }, { "epoch": 3.4137169885691763, "grad_norm": 1.4015860557556152, "learning_rate": 0.0005, "loss": 1.3723509311676025, "step": 43303 }, { "epoch": 3.413795821836815, "grad_norm": 0.5791372060775757, "learning_rate": 0.0005, "loss": 1.2276867628097534, "step": 43304 }, { "epoch": 3.413874655104454, "grad_norm": 0.5676227807998657, "learning_rate": 0.0005, "loss": 1.305124044418335, "step": 43305 }, { "epoch": 3.4139534883720932, "grad_norm": 0.61164391040802, "learning_rate": 0.0005, "loss": 1.3196500539779663, "step": 43306 }, { "epoch": 3.414032321639732, "grad_norm": 0.5388994812965393, "learning_rate": 0.0005, "loss": 1.3155052661895752, "step": 43307 }, { "epoch": 3.414111154907371, "grad_norm": 0.41943222284317017, "learning_rate": 0.0005, "loss": 0.9882556796073914, "step": 43308 }, { "epoch": 3.4141899881750097, "grad_norm": 0.48088985681533813, "learning_rate": 0.0005, "loss": 1.2296051979064941, "step": 43309 }, { "epoch": 3.414268821442649, "grad_norm": 0.5894312262535095, "learning_rate": 0.0005, "loss": 1.3278321027755737, "step": 43310 }, { "epoch": 3.4143476547102876, "grad_norm": 0.553566575050354, "learning_rate": 0.0005, "loss": 1.3618402481079102, "step": 43311 }, { "epoch": 3.4144264879779267, "grad_norm": 0.5788397789001465, "learning_rate": 0.0005, "loss": 1.3619959354400635, "step": 43312 }, { "epoch": 3.4145053212455654, "grad_norm": 0.44648870825767517, "learning_rate": 0.0005, "loss": 1.2934688329696655, "step": 43313 }, { "epoch": 3.4145841545132045, "grad_norm": 0.448108434677124, "learning_rate": 0.0005, "loss": 1.2231836318969727, "step": 43314 }, { "epoch": 3.4146629877808437, "grad_norm": 0.48381897807121277, "learning_rate": 0.0005, "loss": 1.1216599941253662, "step": 43315 }, { "epoch": 3.4147418210484823, "grad_norm": 0.48669567704200745, "learning_rate": 0.0005, "loss": 1.167050838470459, "step": 43316 }, { "epoch": 3.4148206543161215, "grad_norm": 0.5328090190887451, "learning_rate": 0.0005, "loss": 1.1960129737854004, "step": 43317 }, { "epoch": 3.41489948758376, "grad_norm": 0.5083831548690796, "learning_rate": 0.0005, "loss": 1.219738245010376, "step": 43318 }, { "epoch": 3.4149783208513993, "grad_norm": 0.5070359110832214, "learning_rate": 0.0005, "loss": 1.3960647583007812, "step": 43319 }, { "epoch": 3.4150571541190384, "grad_norm": 0.5166724324226379, "learning_rate": 0.0005, "loss": 1.245471715927124, "step": 43320 }, { "epoch": 3.415135987386677, "grad_norm": 19.438966751098633, "learning_rate": 0.0005, "loss": 1.5447359085083008, "step": 43321 }, { "epoch": 3.4152148206543163, "grad_norm": 0.4821630120277405, "learning_rate": 0.0005, "loss": 1.2688053846359253, "step": 43322 }, { "epoch": 3.415293653921955, "grad_norm": 0.4394901990890503, "learning_rate": 0.0005, "loss": 1.29827880859375, "step": 43323 }, { "epoch": 3.415372487189594, "grad_norm": 0.5796383023262024, "learning_rate": 0.0005, "loss": 1.2593464851379395, "step": 43324 }, { "epoch": 3.4154513204572328, "grad_norm": 0.4504013955593109, "learning_rate": 0.0005, "loss": 1.109417200088501, "step": 43325 }, { "epoch": 3.415530153724872, "grad_norm": 0.5119984149932861, "learning_rate": 0.0005, "loss": 1.1517837047576904, "step": 43326 }, { "epoch": 3.415608986992511, "grad_norm": 0.44478070735931396, "learning_rate": 0.0005, "loss": 1.1721036434173584, "step": 43327 }, { "epoch": 3.4156878202601497, "grad_norm": 0.5517787933349609, "learning_rate": 0.0005, "loss": 1.3643825054168701, "step": 43328 }, { "epoch": 3.415766653527789, "grad_norm": 0.47911423444747925, "learning_rate": 0.0005, "loss": 1.1476733684539795, "step": 43329 }, { "epoch": 3.4158454867954275, "grad_norm": 0.47293931245803833, "learning_rate": 0.0005, "loss": 1.2655102014541626, "step": 43330 }, { "epoch": 3.4159243200630667, "grad_norm": 0.3727693557739258, "learning_rate": 0.0005, "loss": 1.091073989868164, "step": 43331 }, { "epoch": 3.4160031533307054, "grad_norm": 0.5470274686813354, "learning_rate": 0.0005, "loss": 1.2248591184616089, "step": 43332 }, { "epoch": 3.4160819865983445, "grad_norm": 0.5351728200912476, "learning_rate": 0.0005, "loss": 1.1829394102096558, "step": 43333 }, { "epoch": 3.4161608198659836, "grad_norm": 0.5256425142288208, "learning_rate": 0.0005, "loss": 1.390117883682251, "step": 43334 }, { "epoch": 3.4162396531336223, "grad_norm": 0.4367673695087433, "learning_rate": 0.0005, "loss": 1.1462680101394653, "step": 43335 }, { "epoch": 3.4163184864012615, "grad_norm": 0.4550340473651886, "learning_rate": 0.0005, "loss": 1.394979476928711, "step": 43336 }, { "epoch": 3.4163973196689, "grad_norm": 0.46919307112693787, "learning_rate": 0.0005, "loss": 1.315136432647705, "step": 43337 }, { "epoch": 3.4164761529365393, "grad_norm": 0.5219528079032898, "learning_rate": 0.0005, "loss": 1.251110553741455, "step": 43338 }, { "epoch": 3.4165549862041784, "grad_norm": 0.5145066380500793, "learning_rate": 0.0005, "loss": 1.3227159976959229, "step": 43339 }, { "epoch": 3.416633819471817, "grad_norm": 0.5282282829284668, "learning_rate": 0.0005, "loss": 1.2640767097473145, "step": 43340 }, { "epoch": 3.4167126527394562, "grad_norm": 0.4769418239593506, "learning_rate": 0.0005, "loss": 1.2711336612701416, "step": 43341 }, { "epoch": 3.416791486007095, "grad_norm": 0.46201545000076294, "learning_rate": 0.0005, "loss": 1.270702600479126, "step": 43342 }, { "epoch": 3.416870319274734, "grad_norm": 0.5397853255271912, "learning_rate": 0.0005, "loss": 1.233446717262268, "step": 43343 }, { "epoch": 3.4169491525423727, "grad_norm": 0.4989340603351593, "learning_rate": 0.0005, "loss": 1.3638014793395996, "step": 43344 }, { "epoch": 3.417027985810012, "grad_norm": 0.5548025369644165, "learning_rate": 0.0005, "loss": 1.2711577415466309, "step": 43345 }, { "epoch": 3.4171068190776506, "grad_norm": 0.514708399772644, "learning_rate": 0.0005, "loss": 1.4447908401489258, "step": 43346 }, { "epoch": 3.4171856523452897, "grad_norm": 0.4681989252567291, "learning_rate": 0.0005, "loss": 1.3228212594985962, "step": 43347 }, { "epoch": 3.417264485612929, "grad_norm": 0.46820491552352905, "learning_rate": 0.0005, "loss": 1.3532617092132568, "step": 43348 }, { "epoch": 3.4173433188805675, "grad_norm": 0.4284515380859375, "learning_rate": 0.0005, "loss": 1.125029444694519, "step": 43349 }, { "epoch": 3.4174221521482067, "grad_norm": 0.5862040519714355, "learning_rate": 0.0005, "loss": 1.3194770812988281, "step": 43350 }, { "epoch": 3.4175009854158453, "grad_norm": 0.4915212392807007, "learning_rate": 0.0005, "loss": 1.302133321762085, "step": 43351 }, { "epoch": 3.4175798186834845, "grad_norm": 0.44983965158462524, "learning_rate": 0.0005, "loss": 1.2759004831314087, "step": 43352 }, { "epoch": 3.4176586519511236, "grad_norm": 0.6067289710044861, "learning_rate": 0.0005, "loss": 1.2271461486816406, "step": 43353 }, { "epoch": 3.4177374852187623, "grad_norm": 0.5400413870811462, "learning_rate": 0.0005, "loss": 1.4027388095855713, "step": 43354 }, { "epoch": 3.4178163184864014, "grad_norm": 0.5249491930007935, "learning_rate": 0.0005, "loss": 1.2044398784637451, "step": 43355 }, { "epoch": 3.41789515175404, "grad_norm": 0.6934369802474976, "learning_rate": 0.0005, "loss": 1.3695132732391357, "step": 43356 }, { "epoch": 3.4179739850216793, "grad_norm": 0.5146216154098511, "learning_rate": 0.0005, "loss": 1.515318512916565, "step": 43357 }, { "epoch": 3.418052818289318, "grad_norm": 0.5245826244354248, "learning_rate": 0.0005, "loss": 1.3125760555267334, "step": 43358 }, { "epoch": 3.418131651556957, "grad_norm": 0.39751920104026794, "learning_rate": 0.0005, "loss": 1.1090648174285889, "step": 43359 }, { "epoch": 3.4182104848245958, "grad_norm": 0.5377546548843384, "learning_rate": 0.0005, "loss": 1.2030006647109985, "step": 43360 }, { "epoch": 3.418289318092235, "grad_norm": 0.47197815775871277, "learning_rate": 0.0005, "loss": 1.0530107021331787, "step": 43361 }, { "epoch": 3.418368151359874, "grad_norm": 0.5862656235694885, "learning_rate": 0.0005, "loss": 1.4226188659667969, "step": 43362 }, { "epoch": 3.4184469846275127, "grad_norm": 0.4505815804004669, "learning_rate": 0.0005, "loss": 1.3063771724700928, "step": 43363 }, { "epoch": 3.418525817895152, "grad_norm": 0.5214923024177551, "learning_rate": 0.0005, "loss": 1.3832842111587524, "step": 43364 }, { "epoch": 3.4186046511627906, "grad_norm": 0.5102865099906921, "learning_rate": 0.0005, "loss": 1.270359992980957, "step": 43365 }, { "epoch": 3.4186834844304297, "grad_norm": 0.4836403429508209, "learning_rate": 0.0005, "loss": 1.1643362045288086, "step": 43366 }, { "epoch": 3.418762317698069, "grad_norm": 0.44092118740081787, "learning_rate": 0.0005, "loss": 1.1636784076690674, "step": 43367 }, { "epoch": 3.4188411509657075, "grad_norm": 0.4879401624202728, "learning_rate": 0.0005, "loss": 1.204143762588501, "step": 43368 }, { "epoch": 3.4189199842333466, "grad_norm": 0.8100945949554443, "learning_rate": 0.0005, "loss": 1.284116268157959, "step": 43369 }, { "epoch": 3.4189988175009853, "grad_norm": 0.518483579158783, "learning_rate": 0.0005, "loss": 1.370030164718628, "step": 43370 }, { "epoch": 3.4190776507686245, "grad_norm": 0.5333564877510071, "learning_rate": 0.0005, "loss": 1.4316072463989258, "step": 43371 }, { "epoch": 3.419156484036263, "grad_norm": 0.49226847290992737, "learning_rate": 0.0005, "loss": 1.3661959171295166, "step": 43372 }, { "epoch": 3.4192353173039023, "grad_norm": 0.5098966360092163, "learning_rate": 0.0005, "loss": 1.2383060455322266, "step": 43373 }, { "epoch": 3.419314150571541, "grad_norm": 0.4690035879611969, "learning_rate": 0.0005, "loss": 1.5377569198608398, "step": 43374 }, { "epoch": 3.41939298383918, "grad_norm": 0.5056219696998596, "learning_rate": 0.0005, "loss": 1.1547974348068237, "step": 43375 }, { "epoch": 3.4194718171068192, "grad_norm": 0.5098615884780884, "learning_rate": 0.0005, "loss": 1.2067487239837646, "step": 43376 }, { "epoch": 3.419550650374458, "grad_norm": 0.6120581030845642, "learning_rate": 0.0005, "loss": 1.3964896202087402, "step": 43377 }, { "epoch": 3.419629483642097, "grad_norm": 0.5815962553024292, "learning_rate": 0.0005, "loss": 1.260872721672058, "step": 43378 }, { "epoch": 3.4197083169097358, "grad_norm": 0.6668867468833923, "learning_rate": 0.0005, "loss": 1.419434666633606, "step": 43379 }, { "epoch": 3.419787150177375, "grad_norm": 0.6484584808349609, "learning_rate": 0.0005, "loss": 1.4364643096923828, "step": 43380 }, { "epoch": 3.419865983445014, "grad_norm": 0.48116248846054077, "learning_rate": 0.0005, "loss": 1.3161671161651611, "step": 43381 }, { "epoch": 3.4199448167126527, "grad_norm": 0.4592727720737457, "learning_rate": 0.0005, "loss": 1.4330884218215942, "step": 43382 }, { "epoch": 3.420023649980292, "grad_norm": 0.5891196131706238, "learning_rate": 0.0005, "loss": 1.3550806045532227, "step": 43383 }, { "epoch": 3.4201024832479305, "grad_norm": 0.6836613416671753, "learning_rate": 0.0005, "loss": 1.5629222393035889, "step": 43384 }, { "epoch": 3.4201813165155697, "grad_norm": 0.5204800963401794, "learning_rate": 0.0005, "loss": 1.3143919706344604, "step": 43385 }, { "epoch": 3.4202601497832084, "grad_norm": 0.3912294805049896, "learning_rate": 0.0005, "loss": 1.0810911655426025, "step": 43386 }, { "epoch": 3.4203389830508475, "grad_norm": 0.4711891710758209, "learning_rate": 0.0005, "loss": 1.3070257902145386, "step": 43387 }, { "epoch": 3.420417816318486, "grad_norm": 0.6051324009895325, "learning_rate": 0.0005, "loss": 1.428093671798706, "step": 43388 }, { "epoch": 3.4204966495861253, "grad_norm": 0.5317126512527466, "learning_rate": 0.0005, "loss": 1.5592608451843262, "step": 43389 }, { "epoch": 3.4205754828537644, "grad_norm": 0.5144670605659485, "learning_rate": 0.0005, "loss": 1.2005119323730469, "step": 43390 }, { "epoch": 3.420654316121403, "grad_norm": 0.49734967947006226, "learning_rate": 0.0005, "loss": 1.2762335538864136, "step": 43391 }, { "epoch": 3.4207331493890423, "grad_norm": 0.4882102608680725, "learning_rate": 0.0005, "loss": 1.4125008583068848, "step": 43392 }, { "epoch": 3.420811982656681, "grad_norm": 0.4540528953075409, "learning_rate": 0.0005, "loss": 1.0128083229064941, "step": 43393 }, { "epoch": 3.42089081592432, "grad_norm": 0.4267764985561371, "learning_rate": 0.0005, "loss": 1.2013030052185059, "step": 43394 }, { "epoch": 3.4209696491919592, "grad_norm": 0.42982324957847595, "learning_rate": 0.0005, "loss": 1.2516133785247803, "step": 43395 }, { "epoch": 3.421048482459598, "grad_norm": 0.5007809996604919, "learning_rate": 0.0005, "loss": 1.1776988506317139, "step": 43396 }, { "epoch": 3.421127315727237, "grad_norm": 0.5700383186340332, "learning_rate": 0.0005, "loss": 1.1105704307556152, "step": 43397 }, { "epoch": 3.4212061489948757, "grad_norm": 0.4477321207523346, "learning_rate": 0.0005, "loss": 1.1516613960266113, "step": 43398 }, { "epoch": 3.421284982262515, "grad_norm": 0.5656415820121765, "learning_rate": 0.0005, "loss": 1.4195730686187744, "step": 43399 }, { "epoch": 3.4213638155301536, "grad_norm": 0.39664164185523987, "learning_rate": 0.0005, "loss": 1.0282692909240723, "step": 43400 }, { "epoch": 3.4214426487977927, "grad_norm": 0.44614481925964355, "learning_rate": 0.0005, "loss": 1.4208016395568848, "step": 43401 }, { "epoch": 3.4215214820654314, "grad_norm": 0.48278191685676575, "learning_rate": 0.0005, "loss": 1.2871277332305908, "step": 43402 }, { "epoch": 3.4216003153330705, "grad_norm": 1.2754786014556885, "learning_rate": 0.0005, "loss": 1.1761786937713623, "step": 43403 }, { "epoch": 3.4216791486007097, "grad_norm": 0.4829910397529602, "learning_rate": 0.0005, "loss": 1.221142053604126, "step": 43404 }, { "epoch": 3.4217579818683483, "grad_norm": 0.45714253187179565, "learning_rate": 0.0005, "loss": 1.4484304189682007, "step": 43405 }, { "epoch": 3.4218368151359875, "grad_norm": 0.4726479649543762, "learning_rate": 0.0005, "loss": 1.1951322555541992, "step": 43406 }, { "epoch": 3.421915648403626, "grad_norm": 0.442759245634079, "learning_rate": 0.0005, "loss": 1.2305657863616943, "step": 43407 }, { "epoch": 3.4219944816712653, "grad_norm": 2.3438780307769775, "learning_rate": 0.0005, "loss": 1.395880103111267, "step": 43408 }, { "epoch": 3.4220733149389044, "grad_norm": 0.6652909517288208, "learning_rate": 0.0005, "loss": 1.065099835395813, "step": 43409 }, { "epoch": 3.422152148206543, "grad_norm": 0.5960903763771057, "learning_rate": 0.0005, "loss": 1.3320757150650024, "step": 43410 }, { "epoch": 3.4222309814741823, "grad_norm": 0.47540706396102905, "learning_rate": 0.0005, "loss": 0.979780912399292, "step": 43411 }, { "epoch": 3.422309814741821, "grad_norm": 0.4924483001232147, "learning_rate": 0.0005, "loss": 1.3138508796691895, "step": 43412 }, { "epoch": 3.42238864800946, "grad_norm": 0.5927886962890625, "learning_rate": 0.0005, "loss": 1.3326528072357178, "step": 43413 }, { "epoch": 3.4224674812770988, "grad_norm": 0.5422759056091309, "learning_rate": 0.0005, "loss": 1.2496676445007324, "step": 43414 }, { "epoch": 3.422546314544738, "grad_norm": 0.4918927848339081, "learning_rate": 0.0005, "loss": 1.3730145692825317, "step": 43415 }, { "epoch": 3.4226251478123766, "grad_norm": 0.44032374024391174, "learning_rate": 0.0005, "loss": 1.1537138223648071, "step": 43416 }, { "epoch": 3.4227039810800157, "grad_norm": 0.5117684006690979, "learning_rate": 0.0005, "loss": 1.0810273885726929, "step": 43417 }, { "epoch": 3.422782814347655, "grad_norm": 0.5498191118240356, "learning_rate": 0.0005, "loss": 1.4367948770523071, "step": 43418 }, { "epoch": 3.4228616476152935, "grad_norm": 0.5838435292243958, "learning_rate": 0.0005, "loss": 1.4129962921142578, "step": 43419 }, { "epoch": 3.4229404808829327, "grad_norm": 0.46633878350257874, "learning_rate": 0.0005, "loss": 1.2582600116729736, "step": 43420 }, { "epoch": 3.4230193141505714, "grad_norm": 0.5066664218902588, "learning_rate": 0.0005, "loss": 1.4731590747833252, "step": 43421 }, { "epoch": 3.4230981474182105, "grad_norm": 0.6577362418174744, "learning_rate": 0.0005, "loss": 1.329310655593872, "step": 43422 }, { "epoch": 3.4231769806858496, "grad_norm": 0.479571670293808, "learning_rate": 0.0005, "loss": 1.2782132625579834, "step": 43423 }, { "epoch": 3.4232558139534883, "grad_norm": 0.4810771346092224, "learning_rate": 0.0005, "loss": 1.2031171321868896, "step": 43424 }, { "epoch": 3.4233346472211275, "grad_norm": 0.5315243601799011, "learning_rate": 0.0005, "loss": 1.5793565511703491, "step": 43425 }, { "epoch": 3.423413480488766, "grad_norm": 0.512362003326416, "learning_rate": 0.0005, "loss": 1.3219736814498901, "step": 43426 }, { "epoch": 3.4234923137564053, "grad_norm": 0.5477590560913086, "learning_rate": 0.0005, "loss": 1.2257739305496216, "step": 43427 }, { "epoch": 3.423571147024044, "grad_norm": 0.6473680734634399, "learning_rate": 0.0005, "loss": 1.2850700616836548, "step": 43428 }, { "epoch": 3.423649980291683, "grad_norm": 0.46353960037231445, "learning_rate": 0.0005, "loss": 1.2958018779754639, "step": 43429 }, { "epoch": 3.423728813559322, "grad_norm": 0.47774356603622437, "learning_rate": 0.0005, "loss": 1.3565335273742676, "step": 43430 }, { "epoch": 3.423807646826961, "grad_norm": 0.4688195586204529, "learning_rate": 0.0005, "loss": 1.0519840717315674, "step": 43431 }, { "epoch": 3.4238864800946, "grad_norm": 0.5218116044998169, "learning_rate": 0.0005, "loss": 1.2881543636322021, "step": 43432 }, { "epoch": 3.4239653133622387, "grad_norm": 0.5335556268692017, "learning_rate": 0.0005, "loss": 1.3399524688720703, "step": 43433 }, { "epoch": 3.424044146629878, "grad_norm": 0.5286717414855957, "learning_rate": 0.0005, "loss": 1.3956584930419922, "step": 43434 }, { "epoch": 3.4241229798975166, "grad_norm": 0.43038251996040344, "learning_rate": 0.0005, "loss": 1.1782606840133667, "step": 43435 }, { "epoch": 3.4242018131651557, "grad_norm": 0.5515519380569458, "learning_rate": 0.0005, "loss": 1.3163701295852661, "step": 43436 }, { "epoch": 3.424280646432795, "grad_norm": 0.5441070795059204, "learning_rate": 0.0005, "loss": 1.4066333770751953, "step": 43437 }, { "epoch": 3.4243594797004335, "grad_norm": 0.48618587851524353, "learning_rate": 0.0005, "loss": 1.3184324502944946, "step": 43438 }, { "epoch": 3.4244383129680727, "grad_norm": 0.35903576016426086, "learning_rate": 0.0005, "loss": 1.1079241037368774, "step": 43439 }, { "epoch": 3.4245171462357114, "grad_norm": 0.5642359256744385, "learning_rate": 0.0005, "loss": 1.274023175239563, "step": 43440 }, { "epoch": 3.4245959795033505, "grad_norm": 0.4664187729358673, "learning_rate": 0.0005, "loss": 1.1760227680206299, "step": 43441 }, { "epoch": 3.424674812770989, "grad_norm": 0.6061961054801941, "learning_rate": 0.0005, "loss": 1.41502046585083, "step": 43442 }, { "epoch": 3.4247536460386283, "grad_norm": 0.4046594202518463, "learning_rate": 0.0005, "loss": 1.0306367874145508, "step": 43443 }, { "epoch": 3.424832479306267, "grad_norm": 0.4424246549606323, "learning_rate": 0.0005, "loss": 1.1568118333816528, "step": 43444 }, { "epoch": 3.424911312573906, "grad_norm": 0.46782130002975464, "learning_rate": 0.0005, "loss": 1.298475742340088, "step": 43445 }, { "epoch": 3.4249901458415453, "grad_norm": 0.5608625411987305, "learning_rate": 0.0005, "loss": 1.3257458209991455, "step": 43446 }, { "epoch": 3.425068979109184, "grad_norm": 0.5498725175857544, "learning_rate": 0.0005, "loss": 1.3009631633758545, "step": 43447 }, { "epoch": 3.425147812376823, "grad_norm": 0.43403416872024536, "learning_rate": 0.0005, "loss": 1.217219591140747, "step": 43448 }, { "epoch": 3.4252266456444618, "grad_norm": 0.4967885911464691, "learning_rate": 0.0005, "loss": 1.1419858932495117, "step": 43449 }, { "epoch": 3.425305478912101, "grad_norm": 0.6497843265533447, "learning_rate": 0.0005, "loss": 1.4017583131790161, "step": 43450 }, { "epoch": 3.42538431217974, "grad_norm": 0.4835027754306793, "learning_rate": 0.0005, "loss": 1.2712061405181885, "step": 43451 }, { "epoch": 3.4254631454473787, "grad_norm": 0.5412828326225281, "learning_rate": 0.0005, "loss": 1.4342570304870605, "step": 43452 }, { "epoch": 3.425541978715018, "grad_norm": 0.539591372013092, "learning_rate": 0.0005, "loss": 1.2769317626953125, "step": 43453 }, { "epoch": 3.4256208119826566, "grad_norm": 0.7422391176223755, "learning_rate": 0.0005, "loss": 1.3157343864440918, "step": 43454 }, { "epoch": 3.4256996452502957, "grad_norm": 2.4612679481506348, "learning_rate": 0.0005, "loss": 1.402788758277893, "step": 43455 }, { "epoch": 3.4257784785179344, "grad_norm": 0.5398597121238708, "learning_rate": 0.0005, "loss": 1.259788990020752, "step": 43456 }, { "epoch": 3.4258573117855735, "grad_norm": 0.6335151195526123, "learning_rate": 0.0005, "loss": 1.2805812358856201, "step": 43457 }, { "epoch": 3.4259361450532126, "grad_norm": 0.4520660936832428, "learning_rate": 0.0005, "loss": 1.2583487033843994, "step": 43458 }, { "epoch": 3.4260149783208513, "grad_norm": 0.6824857592582703, "learning_rate": 0.0005, "loss": 1.170520305633545, "step": 43459 }, { "epoch": 3.4260938115884905, "grad_norm": 0.5948206782341003, "learning_rate": 0.0005, "loss": 1.2344632148742676, "step": 43460 }, { "epoch": 3.426172644856129, "grad_norm": 0.5373913645744324, "learning_rate": 0.0005, "loss": 1.2403148412704468, "step": 43461 }, { "epoch": 3.4262514781237683, "grad_norm": 0.5886154770851135, "learning_rate": 0.0005, "loss": 1.4011082649230957, "step": 43462 }, { "epoch": 3.426330311391407, "grad_norm": 0.5020747780799866, "learning_rate": 0.0005, "loss": 1.2621750831604004, "step": 43463 }, { "epoch": 3.426409144659046, "grad_norm": 0.4310076832771301, "learning_rate": 0.0005, "loss": 1.1318540573120117, "step": 43464 }, { "epoch": 3.4264879779266852, "grad_norm": 0.5072072148323059, "learning_rate": 0.0005, "loss": 1.137128233909607, "step": 43465 }, { "epoch": 3.426566811194324, "grad_norm": 0.5820508599281311, "learning_rate": 0.0005, "loss": 1.296635389328003, "step": 43466 }, { "epoch": 3.426645644461963, "grad_norm": 0.41800186038017273, "learning_rate": 0.0005, "loss": 1.2345845699310303, "step": 43467 }, { "epoch": 3.4267244777296018, "grad_norm": 0.5763178467750549, "learning_rate": 0.0005, "loss": 1.3066517114639282, "step": 43468 }, { "epoch": 3.426803310997241, "grad_norm": 0.765401303768158, "learning_rate": 0.0005, "loss": 1.4045356512069702, "step": 43469 }, { "epoch": 3.42688214426488, "grad_norm": 0.5342881083488464, "learning_rate": 0.0005, "loss": 1.1991935968399048, "step": 43470 }, { "epoch": 3.4269609775325187, "grad_norm": 0.581147313117981, "learning_rate": 0.0005, "loss": 1.323378324508667, "step": 43471 }, { "epoch": 3.427039810800158, "grad_norm": 0.5492790341377258, "learning_rate": 0.0005, "loss": 1.1433058977127075, "step": 43472 }, { "epoch": 3.4271186440677965, "grad_norm": 0.7032603621482849, "learning_rate": 0.0005, "loss": 1.4151755571365356, "step": 43473 }, { "epoch": 3.4271974773354357, "grad_norm": 0.5812634229660034, "learning_rate": 0.0005, "loss": 1.2390716075897217, "step": 43474 }, { "epoch": 3.4272763106030744, "grad_norm": 0.48159945011138916, "learning_rate": 0.0005, "loss": 1.2218626737594604, "step": 43475 }, { "epoch": 3.4273551438707135, "grad_norm": 0.6250247955322266, "learning_rate": 0.0005, "loss": 1.2332186698913574, "step": 43476 }, { "epoch": 3.427433977138352, "grad_norm": 0.43959835171699524, "learning_rate": 0.0005, "loss": 1.2288200855255127, "step": 43477 }, { "epoch": 3.4275128104059913, "grad_norm": 0.42720654606819153, "learning_rate": 0.0005, "loss": 1.3040125370025635, "step": 43478 }, { "epoch": 3.4275916436736305, "grad_norm": 0.4705981910228729, "learning_rate": 0.0005, "loss": 1.2467325925827026, "step": 43479 }, { "epoch": 3.427670476941269, "grad_norm": 0.4176853597164154, "learning_rate": 0.0005, "loss": 1.0682390928268433, "step": 43480 }, { "epoch": 3.4277493102089083, "grad_norm": 0.4386184811592102, "learning_rate": 0.0005, "loss": 1.2589876651763916, "step": 43481 }, { "epoch": 3.427828143476547, "grad_norm": 0.4565248191356659, "learning_rate": 0.0005, "loss": 1.201645851135254, "step": 43482 }, { "epoch": 3.427906976744186, "grad_norm": 0.4000530540943146, "learning_rate": 0.0005, "loss": 1.1402294635772705, "step": 43483 }, { "epoch": 3.4279858100118252, "grad_norm": 0.3719308078289032, "learning_rate": 0.0005, "loss": 1.1381561756134033, "step": 43484 }, { "epoch": 3.428064643279464, "grad_norm": 0.5123340487480164, "learning_rate": 0.0005, "loss": 1.2263139486312866, "step": 43485 }, { "epoch": 3.428143476547103, "grad_norm": 0.6007055044174194, "learning_rate": 0.0005, "loss": 1.2189855575561523, "step": 43486 }, { "epoch": 3.4282223098147417, "grad_norm": 0.5723974704742432, "learning_rate": 0.0005, "loss": 1.200189471244812, "step": 43487 }, { "epoch": 3.428301143082381, "grad_norm": 0.8603023290634155, "learning_rate": 0.0005, "loss": 1.361051082611084, "step": 43488 }, { "epoch": 3.4283799763500196, "grad_norm": 0.4909905195236206, "learning_rate": 0.0005, "loss": 1.3072034120559692, "step": 43489 }, { "epoch": 3.4284588096176587, "grad_norm": 0.4212280511856079, "learning_rate": 0.0005, "loss": 1.174278736114502, "step": 43490 }, { "epoch": 3.4285376428852974, "grad_norm": 0.502111554145813, "learning_rate": 0.0005, "loss": 1.210913062095642, "step": 43491 }, { "epoch": 3.4286164761529365, "grad_norm": 0.48572060465812683, "learning_rate": 0.0005, "loss": 1.2912161350250244, "step": 43492 }, { "epoch": 3.4286953094205757, "grad_norm": 0.599881112575531, "learning_rate": 0.0005, "loss": 1.2328060865402222, "step": 43493 }, { "epoch": 3.4287741426882143, "grad_norm": 0.42169293761253357, "learning_rate": 0.0005, "loss": 1.2131810188293457, "step": 43494 }, { "epoch": 3.4288529759558535, "grad_norm": 0.577588677406311, "learning_rate": 0.0005, "loss": 1.5260590314865112, "step": 43495 }, { "epoch": 3.428931809223492, "grad_norm": 0.6642958521842957, "learning_rate": 0.0005, "loss": 1.1555566787719727, "step": 43496 }, { "epoch": 3.4290106424911313, "grad_norm": 0.454845130443573, "learning_rate": 0.0005, "loss": 1.2242937088012695, "step": 43497 }, { "epoch": 3.4290894757587704, "grad_norm": 0.5165624618530273, "learning_rate": 0.0005, "loss": 1.3433191776275635, "step": 43498 }, { "epoch": 3.429168309026409, "grad_norm": 0.5814336538314819, "learning_rate": 0.0005, "loss": 1.2741847038269043, "step": 43499 }, { "epoch": 3.4292471422940483, "grad_norm": 0.42743605375289917, "learning_rate": 0.0005, "loss": 1.19417405128479, "step": 43500 }, { "epoch": 3.429325975561687, "grad_norm": 0.4617914855480194, "learning_rate": 0.0005, "loss": 1.316582441329956, "step": 43501 }, { "epoch": 3.429404808829326, "grad_norm": 0.4872722923755646, "learning_rate": 0.0005, "loss": 1.2991281747817993, "step": 43502 }, { "epoch": 3.4294836420969648, "grad_norm": 0.5267223715782166, "learning_rate": 0.0005, "loss": 1.3088257312774658, "step": 43503 }, { "epoch": 3.429562475364604, "grad_norm": 0.44979068636894226, "learning_rate": 0.0005, "loss": 1.3760159015655518, "step": 43504 }, { "epoch": 3.4296413086322426, "grad_norm": 0.511326789855957, "learning_rate": 0.0005, "loss": 1.25826096534729, "step": 43505 }, { "epoch": 3.4297201418998817, "grad_norm": 0.6083822846412659, "learning_rate": 0.0005, "loss": 1.2569984197616577, "step": 43506 }, { "epoch": 3.429798975167521, "grad_norm": 0.4958832561969757, "learning_rate": 0.0005, "loss": 1.1920989751815796, "step": 43507 }, { "epoch": 3.4298778084351595, "grad_norm": 0.3707607686519623, "learning_rate": 0.0005, "loss": 1.1182985305786133, "step": 43508 }, { "epoch": 3.4299566417027987, "grad_norm": 0.5435582399368286, "learning_rate": 0.0005, "loss": 1.428367018699646, "step": 43509 }, { "epoch": 3.4300354749704374, "grad_norm": 0.4867306649684906, "learning_rate": 0.0005, "loss": 1.2905079126358032, "step": 43510 }, { "epoch": 3.4301143082380765, "grad_norm": 0.4629400968551636, "learning_rate": 0.0005, "loss": 1.3034799098968506, "step": 43511 }, { "epoch": 3.4301931415057156, "grad_norm": 0.6131288409233093, "learning_rate": 0.0005, "loss": 1.279266357421875, "step": 43512 }, { "epoch": 3.4302719747733543, "grad_norm": 0.5492829084396362, "learning_rate": 0.0005, "loss": 1.3607898950576782, "step": 43513 }, { "epoch": 3.4303508080409935, "grad_norm": 0.4251355230808258, "learning_rate": 0.0005, "loss": 1.279071569442749, "step": 43514 }, { "epoch": 3.430429641308632, "grad_norm": 0.4973325729370117, "learning_rate": 0.0005, "loss": 1.1850619316101074, "step": 43515 }, { "epoch": 3.4305084745762713, "grad_norm": 0.4483499228954315, "learning_rate": 0.0005, "loss": 1.2098078727722168, "step": 43516 }, { "epoch": 3.43058730784391, "grad_norm": 0.43703725934028625, "learning_rate": 0.0005, "loss": 1.3048796653747559, "step": 43517 }, { "epoch": 3.430666141111549, "grad_norm": 0.48492223024368286, "learning_rate": 0.0005, "loss": 1.3378262519836426, "step": 43518 }, { "epoch": 3.430744974379188, "grad_norm": 0.4063245356082916, "learning_rate": 0.0005, "loss": 1.1830356121063232, "step": 43519 }, { "epoch": 3.430823807646827, "grad_norm": 0.4596664011478424, "learning_rate": 0.0005, "loss": 1.4781337976455688, "step": 43520 }, { "epoch": 3.430902640914466, "grad_norm": 0.45034217834472656, "learning_rate": 0.0005, "loss": 1.442702054977417, "step": 43521 }, { "epoch": 3.4309814741821048, "grad_norm": 0.5353452563285828, "learning_rate": 0.0005, "loss": 1.4526095390319824, "step": 43522 }, { "epoch": 3.431060307449744, "grad_norm": 0.5545448660850525, "learning_rate": 0.0005, "loss": 1.3962243795394897, "step": 43523 }, { "epoch": 3.4311391407173826, "grad_norm": 0.4893529713153839, "learning_rate": 0.0005, "loss": 1.1983568668365479, "step": 43524 }, { "epoch": 3.4312179739850217, "grad_norm": 0.46353232860565186, "learning_rate": 0.0005, "loss": 1.3693013191223145, "step": 43525 }, { "epoch": 3.431296807252661, "grad_norm": 0.5353235602378845, "learning_rate": 0.0005, "loss": 1.2077863216400146, "step": 43526 }, { "epoch": 3.4313756405202995, "grad_norm": 0.48173239827156067, "learning_rate": 0.0005, "loss": 1.2802166938781738, "step": 43527 }, { "epoch": 3.4314544737879387, "grad_norm": 0.5392845273017883, "learning_rate": 0.0005, "loss": 1.218416690826416, "step": 43528 }, { "epoch": 3.4315333070555774, "grad_norm": 0.4799399673938751, "learning_rate": 0.0005, "loss": 1.3337969779968262, "step": 43529 }, { "epoch": 3.4316121403232165, "grad_norm": 0.5077009201049805, "learning_rate": 0.0005, "loss": 1.2575836181640625, "step": 43530 }, { "epoch": 3.431690973590855, "grad_norm": 0.4849899411201477, "learning_rate": 0.0005, "loss": 1.2659752368927002, "step": 43531 }, { "epoch": 3.4317698068584943, "grad_norm": 0.5722626447677612, "learning_rate": 0.0005, "loss": 1.161274790763855, "step": 43532 }, { "epoch": 3.431848640126133, "grad_norm": 0.5804334282875061, "learning_rate": 0.0005, "loss": 1.374617576599121, "step": 43533 }, { "epoch": 3.431927473393772, "grad_norm": 0.4820975661277771, "learning_rate": 0.0005, "loss": 1.2898738384246826, "step": 43534 }, { "epoch": 3.4320063066614113, "grad_norm": 0.44425907731056213, "learning_rate": 0.0005, "loss": 1.1955091953277588, "step": 43535 }, { "epoch": 3.43208513992905, "grad_norm": 0.5124627947807312, "learning_rate": 0.0005, "loss": 1.3674349784851074, "step": 43536 }, { "epoch": 3.432163973196689, "grad_norm": 0.4310418665409088, "learning_rate": 0.0005, "loss": 1.207883596420288, "step": 43537 }, { "epoch": 3.4322428064643278, "grad_norm": 0.4791730046272278, "learning_rate": 0.0005, "loss": 1.256032109260559, "step": 43538 }, { "epoch": 3.432321639731967, "grad_norm": 0.40935787558555603, "learning_rate": 0.0005, "loss": 1.270811676979065, "step": 43539 }, { "epoch": 3.432400472999606, "grad_norm": 0.552908718585968, "learning_rate": 0.0005, "loss": 1.2518216371536255, "step": 43540 }, { "epoch": 3.4324793062672447, "grad_norm": 0.4530423581600189, "learning_rate": 0.0005, "loss": 1.3126267194747925, "step": 43541 }, { "epoch": 3.432558139534884, "grad_norm": 0.3571609556674957, "learning_rate": 0.0005, "loss": 0.9119449853897095, "step": 43542 }, { "epoch": 3.4326369728025226, "grad_norm": 0.5169931054115295, "learning_rate": 0.0005, "loss": 1.3692421913146973, "step": 43543 }, { "epoch": 3.4327158060701617, "grad_norm": 0.45588648319244385, "learning_rate": 0.0005, "loss": 1.222630500793457, "step": 43544 }, { "epoch": 3.4327946393378004, "grad_norm": 0.3909289538860321, "learning_rate": 0.0005, "loss": 1.0532499551773071, "step": 43545 }, { "epoch": 3.4328734726054395, "grad_norm": 0.42720356583595276, "learning_rate": 0.0005, "loss": 1.2966811656951904, "step": 43546 }, { "epoch": 3.432952305873078, "grad_norm": 0.5347101092338562, "learning_rate": 0.0005, "loss": 1.3591570854187012, "step": 43547 }, { "epoch": 3.4330311391407173, "grad_norm": 0.46940377354621887, "learning_rate": 0.0005, "loss": 1.3875555992126465, "step": 43548 }, { "epoch": 3.4331099724083565, "grad_norm": 0.5096518397331238, "learning_rate": 0.0005, "loss": 1.218543291091919, "step": 43549 }, { "epoch": 3.433188805675995, "grad_norm": 0.5192427635192871, "learning_rate": 0.0005, "loss": 1.3112597465515137, "step": 43550 }, { "epoch": 3.4332676389436343, "grad_norm": 0.4053587317466736, "learning_rate": 0.0005, "loss": 1.198757290840149, "step": 43551 }, { "epoch": 3.433346472211273, "grad_norm": 0.4810882806777954, "learning_rate": 0.0005, "loss": 1.3110418319702148, "step": 43552 }, { "epoch": 3.433425305478912, "grad_norm": 0.4935343265533447, "learning_rate": 0.0005, "loss": 1.388061285018921, "step": 43553 }, { "epoch": 3.4335041387465512, "grad_norm": 0.5294564366340637, "learning_rate": 0.0005, "loss": 1.1346948146820068, "step": 43554 }, { "epoch": 3.43358297201419, "grad_norm": 0.528614342212677, "learning_rate": 0.0005, "loss": 1.3913918733596802, "step": 43555 }, { "epoch": 3.433661805281829, "grad_norm": 0.45330682396888733, "learning_rate": 0.0005, "loss": 1.0313353538513184, "step": 43556 }, { "epoch": 3.4337406385494678, "grad_norm": 0.5083464980125427, "learning_rate": 0.0005, "loss": 1.374301791191101, "step": 43557 }, { "epoch": 3.433819471817107, "grad_norm": 0.5516804456710815, "learning_rate": 0.0005, "loss": 1.1721867322921753, "step": 43558 }, { "epoch": 3.4338983050847456, "grad_norm": 0.5237840414047241, "learning_rate": 0.0005, "loss": 1.24713933467865, "step": 43559 }, { "epoch": 3.4339771383523847, "grad_norm": 0.48883551359176636, "learning_rate": 0.0005, "loss": 1.160031795501709, "step": 43560 }, { "epoch": 3.4340559716200234, "grad_norm": 0.492453932762146, "learning_rate": 0.0005, "loss": 1.3706152439117432, "step": 43561 }, { "epoch": 3.4341348048876625, "grad_norm": 0.45834043622016907, "learning_rate": 0.0005, "loss": 1.4150588512420654, "step": 43562 }, { "epoch": 3.4342136381553017, "grad_norm": 0.43412211537361145, "learning_rate": 0.0005, "loss": 1.369856357574463, "step": 43563 }, { "epoch": 3.4342924714229404, "grad_norm": 0.4613443911075592, "learning_rate": 0.0005, "loss": 1.217240571975708, "step": 43564 }, { "epoch": 3.4343713046905795, "grad_norm": 0.5125871300697327, "learning_rate": 0.0005, "loss": 1.2240749597549438, "step": 43565 }, { "epoch": 3.434450137958218, "grad_norm": 0.41361984610557556, "learning_rate": 0.0005, "loss": 1.3214740753173828, "step": 43566 }, { "epoch": 3.4345289712258573, "grad_norm": 0.4527488350868225, "learning_rate": 0.0005, "loss": 1.1795549392700195, "step": 43567 }, { "epoch": 3.4346078044934965, "grad_norm": 0.5264054536819458, "learning_rate": 0.0005, "loss": 1.2058730125427246, "step": 43568 }, { "epoch": 3.434686637761135, "grad_norm": 0.42903152108192444, "learning_rate": 0.0005, "loss": 1.2530076503753662, "step": 43569 }, { "epoch": 3.4347654710287743, "grad_norm": 0.4966459274291992, "learning_rate": 0.0005, "loss": 1.31966233253479, "step": 43570 }, { "epoch": 3.434844304296413, "grad_norm": 0.547174870967865, "learning_rate": 0.0005, "loss": 1.4281699657440186, "step": 43571 }, { "epoch": 3.434923137564052, "grad_norm": 0.46032029390335083, "learning_rate": 0.0005, "loss": 1.2750754356384277, "step": 43572 }, { "epoch": 3.435001970831691, "grad_norm": 0.4604875147342682, "learning_rate": 0.0005, "loss": 1.2387313842773438, "step": 43573 }, { "epoch": 3.43508080409933, "grad_norm": 1.1312321424484253, "learning_rate": 0.0005, "loss": 1.1340675354003906, "step": 43574 }, { "epoch": 3.435159637366969, "grad_norm": 0.48489007353782654, "learning_rate": 0.0005, "loss": 1.3059217929840088, "step": 43575 }, { "epoch": 3.4352384706346077, "grad_norm": 0.4938885569572449, "learning_rate": 0.0005, "loss": 1.232797622680664, "step": 43576 }, { "epoch": 3.435317303902247, "grad_norm": 0.47927412390708923, "learning_rate": 0.0005, "loss": 1.2856390476226807, "step": 43577 }, { "epoch": 3.4353961371698856, "grad_norm": 0.5989406704902649, "learning_rate": 0.0005, "loss": 1.3593344688415527, "step": 43578 }, { "epoch": 3.4354749704375247, "grad_norm": 0.5081428289413452, "learning_rate": 0.0005, "loss": 1.3458642959594727, "step": 43579 }, { "epoch": 3.4355538037051634, "grad_norm": 0.7104975581169128, "learning_rate": 0.0005, "loss": 1.316091537475586, "step": 43580 }, { "epoch": 3.4356326369728025, "grad_norm": 0.42663586139678955, "learning_rate": 0.0005, "loss": 1.2302238941192627, "step": 43581 }, { "epoch": 3.4357114702404417, "grad_norm": 0.3792119324207306, "learning_rate": 0.0005, "loss": 1.1190943717956543, "step": 43582 }, { "epoch": 3.4357903035080803, "grad_norm": 0.41071459650993347, "learning_rate": 0.0005, "loss": 1.1933010816574097, "step": 43583 }, { "epoch": 3.4358691367757195, "grad_norm": 0.48512646555900574, "learning_rate": 0.0005, "loss": 1.2951676845550537, "step": 43584 }, { "epoch": 3.435947970043358, "grad_norm": 0.4793699383735657, "learning_rate": 0.0005, "loss": 1.1600908041000366, "step": 43585 }, { "epoch": 3.4360268033109973, "grad_norm": 0.5005134344100952, "learning_rate": 0.0005, "loss": 1.2254441976547241, "step": 43586 }, { "epoch": 3.436105636578636, "grad_norm": 0.41834312677383423, "learning_rate": 0.0005, "loss": 1.3171651363372803, "step": 43587 }, { "epoch": 3.436184469846275, "grad_norm": 0.4224925637245178, "learning_rate": 0.0005, "loss": 1.163655400276184, "step": 43588 }, { "epoch": 3.4362633031139143, "grad_norm": 0.5701112747192383, "learning_rate": 0.0005, "loss": 1.210972785949707, "step": 43589 }, { "epoch": 3.436342136381553, "grad_norm": 0.7049112319946289, "learning_rate": 0.0005, "loss": 1.3110733032226562, "step": 43590 }, { "epoch": 3.436420969649192, "grad_norm": 0.39892271161079407, "learning_rate": 0.0005, "loss": 1.1021744012832642, "step": 43591 }, { "epoch": 3.4364998029168308, "grad_norm": 0.4738233983516693, "learning_rate": 0.0005, "loss": 1.3793108463287354, "step": 43592 }, { "epoch": 3.43657863618447, "grad_norm": 0.48931485414505005, "learning_rate": 0.0005, "loss": 1.3488725423812866, "step": 43593 }, { "epoch": 3.4366574694521086, "grad_norm": 0.4114474058151245, "learning_rate": 0.0005, "loss": 1.3441905975341797, "step": 43594 }, { "epoch": 3.4367363027197477, "grad_norm": 0.4343203008174896, "learning_rate": 0.0005, "loss": 1.0831718444824219, "step": 43595 }, { "epoch": 3.436815135987387, "grad_norm": 0.46438100934028625, "learning_rate": 0.0005, "loss": 1.154151201248169, "step": 43596 }, { "epoch": 3.4368939692550255, "grad_norm": 0.5030050277709961, "learning_rate": 0.0005, "loss": 1.2121021747589111, "step": 43597 }, { "epoch": 3.4369728025226647, "grad_norm": 0.5036848783493042, "learning_rate": 0.0005, "loss": 1.1863746643066406, "step": 43598 }, { "epoch": 3.4370516357903034, "grad_norm": 0.47768041491508484, "learning_rate": 0.0005, "loss": 1.3093836307525635, "step": 43599 }, { "epoch": 3.4371304690579425, "grad_norm": 0.5000249147415161, "learning_rate": 0.0005, "loss": 1.24664306640625, "step": 43600 }, { "epoch": 3.4372093023255816, "grad_norm": 0.5551121234893799, "learning_rate": 0.0005, "loss": 1.4204320907592773, "step": 43601 }, { "epoch": 3.4372881355932203, "grad_norm": 1.0393869876861572, "learning_rate": 0.0005, "loss": 1.441880702972412, "step": 43602 }, { "epoch": 3.4373669688608595, "grad_norm": 0.4663856029510498, "learning_rate": 0.0005, "loss": 1.2408013343811035, "step": 43603 }, { "epoch": 3.437445802128498, "grad_norm": 0.5859962701797485, "learning_rate": 0.0005, "loss": 1.4127718210220337, "step": 43604 }, { "epoch": 3.4375246353961373, "grad_norm": 0.46187207102775574, "learning_rate": 0.0005, "loss": 1.11631441116333, "step": 43605 }, { "epoch": 3.437603468663776, "grad_norm": 0.6543595790863037, "learning_rate": 0.0005, "loss": 1.2828044891357422, "step": 43606 }, { "epoch": 3.437682301931415, "grad_norm": 0.4743420481681824, "learning_rate": 0.0005, "loss": 1.2169502973556519, "step": 43607 }, { "epoch": 3.437761135199054, "grad_norm": 0.543117344379425, "learning_rate": 0.0005, "loss": 1.3963491916656494, "step": 43608 }, { "epoch": 3.437839968466693, "grad_norm": 0.6883164048194885, "learning_rate": 0.0005, "loss": 1.3287224769592285, "step": 43609 }, { "epoch": 3.437918801734332, "grad_norm": 0.41218313574790955, "learning_rate": 0.0005, "loss": 1.06107497215271, "step": 43610 }, { "epoch": 3.4379976350019708, "grad_norm": 0.4674173891544342, "learning_rate": 0.0005, "loss": 1.2552165985107422, "step": 43611 }, { "epoch": 3.43807646826961, "grad_norm": 0.5689637064933777, "learning_rate": 0.0005, "loss": 1.3364109992980957, "step": 43612 }, { "epoch": 3.4381553015372486, "grad_norm": 0.576176106929779, "learning_rate": 0.0005, "loss": 1.5061798095703125, "step": 43613 }, { "epoch": 3.4382341348048877, "grad_norm": 0.43273627758026123, "learning_rate": 0.0005, "loss": 1.1675101518630981, "step": 43614 }, { "epoch": 3.438312968072527, "grad_norm": 0.5675897002220154, "learning_rate": 0.0005, "loss": 1.2623062133789062, "step": 43615 }, { "epoch": 3.4383918013401655, "grad_norm": 0.5879013538360596, "learning_rate": 0.0005, "loss": 1.3168394565582275, "step": 43616 }, { "epoch": 3.4384706346078047, "grad_norm": 0.47131314873695374, "learning_rate": 0.0005, "loss": 1.4180042743682861, "step": 43617 }, { "epoch": 3.4385494678754434, "grad_norm": 0.4979723393917084, "learning_rate": 0.0005, "loss": 1.2033557891845703, "step": 43618 }, { "epoch": 3.4386283011430825, "grad_norm": 0.4652993977069855, "learning_rate": 0.0005, "loss": 1.2940523624420166, "step": 43619 }, { "epoch": 3.438707134410721, "grad_norm": 0.5947270393371582, "learning_rate": 0.0005, "loss": 1.18000328540802, "step": 43620 }, { "epoch": 3.4387859676783603, "grad_norm": 0.7067752480506897, "learning_rate": 0.0005, "loss": 1.1978862285614014, "step": 43621 }, { "epoch": 3.438864800945999, "grad_norm": 0.4344562888145447, "learning_rate": 0.0005, "loss": 1.1068271398544312, "step": 43622 }, { "epoch": 3.438943634213638, "grad_norm": 0.4562564194202423, "learning_rate": 0.0005, "loss": 1.2808036804199219, "step": 43623 }, { "epoch": 3.4390224674812773, "grad_norm": 0.5834528207778931, "learning_rate": 0.0005, "loss": 1.220762848854065, "step": 43624 }, { "epoch": 3.439101300748916, "grad_norm": 0.5193427205085754, "learning_rate": 0.0005, "loss": 1.4796606302261353, "step": 43625 }, { "epoch": 3.439180134016555, "grad_norm": 0.6353371739387512, "learning_rate": 0.0005, "loss": 1.267261266708374, "step": 43626 }, { "epoch": 3.439258967284194, "grad_norm": 0.4962109327316284, "learning_rate": 0.0005, "loss": 1.262668251991272, "step": 43627 }, { "epoch": 3.439337800551833, "grad_norm": 0.397308349609375, "learning_rate": 0.0005, "loss": 1.1028947830200195, "step": 43628 }, { "epoch": 3.439416633819472, "grad_norm": 0.4472196400165558, "learning_rate": 0.0005, "loss": 1.1079604625701904, "step": 43629 }, { "epoch": 3.4394954670871107, "grad_norm": 0.5660980939865112, "learning_rate": 0.0005, "loss": 1.3451991081237793, "step": 43630 }, { "epoch": 3.43957430035475, "grad_norm": 0.49120351672172546, "learning_rate": 0.0005, "loss": 1.2452714443206787, "step": 43631 }, { "epoch": 3.4396531336223886, "grad_norm": 0.6011832356452942, "learning_rate": 0.0005, "loss": 1.2301908731460571, "step": 43632 }, { "epoch": 3.4397319668900277, "grad_norm": 0.4948403835296631, "learning_rate": 0.0005, "loss": 1.201143741607666, "step": 43633 }, { "epoch": 3.4398108001576664, "grad_norm": 0.42461758852005005, "learning_rate": 0.0005, "loss": 1.1933250427246094, "step": 43634 }, { "epoch": 3.4398896334253055, "grad_norm": 0.4801226258277893, "learning_rate": 0.0005, "loss": 1.473893404006958, "step": 43635 }, { "epoch": 3.439968466692944, "grad_norm": 0.409798264503479, "learning_rate": 0.0005, "loss": 1.1340982913970947, "step": 43636 }, { "epoch": 3.4400472999605833, "grad_norm": 0.4249020516872406, "learning_rate": 0.0005, "loss": 1.0727418661117554, "step": 43637 }, { "epoch": 3.4401261332282225, "grad_norm": 0.47857925295829773, "learning_rate": 0.0005, "loss": 1.32232666015625, "step": 43638 }, { "epoch": 3.440204966495861, "grad_norm": 0.5306569337844849, "learning_rate": 0.0005, "loss": 1.1661752462387085, "step": 43639 }, { "epoch": 3.4402837997635003, "grad_norm": 0.5485978722572327, "learning_rate": 0.0005, "loss": 1.3357456922531128, "step": 43640 }, { "epoch": 3.440362633031139, "grad_norm": 4.590035915374756, "learning_rate": 0.0005, "loss": 1.1505662202835083, "step": 43641 }, { "epoch": 3.440441466298778, "grad_norm": 0.46183931827545166, "learning_rate": 0.0005, "loss": 1.169447422027588, "step": 43642 }, { "epoch": 3.4405202995664173, "grad_norm": 0.47544512152671814, "learning_rate": 0.0005, "loss": 1.303919792175293, "step": 43643 }, { "epoch": 3.440599132834056, "grad_norm": 0.604429304599762, "learning_rate": 0.0005, "loss": 1.2368199825286865, "step": 43644 }, { "epoch": 3.440677966101695, "grad_norm": 0.48336148262023926, "learning_rate": 0.0005, "loss": 1.4374642372131348, "step": 43645 }, { "epoch": 3.4407567993693338, "grad_norm": 0.49467387795448303, "learning_rate": 0.0005, "loss": 1.3867204189300537, "step": 43646 }, { "epoch": 3.440835632636973, "grad_norm": 0.46437743306159973, "learning_rate": 0.0005, "loss": 1.2735600471496582, "step": 43647 }, { "epoch": 3.4409144659046116, "grad_norm": 0.5709671378135681, "learning_rate": 0.0005, "loss": 1.2473950386047363, "step": 43648 }, { "epoch": 3.4409932991722507, "grad_norm": 0.5038465857505798, "learning_rate": 0.0005, "loss": 1.1578971147537231, "step": 43649 }, { "epoch": 3.4410721324398894, "grad_norm": 0.5125970244407654, "learning_rate": 0.0005, "loss": 1.2125704288482666, "step": 43650 }, { "epoch": 3.4411509657075285, "grad_norm": 0.5722680687904358, "learning_rate": 0.0005, "loss": 1.3097991943359375, "step": 43651 }, { "epoch": 3.4412297989751677, "grad_norm": 1.3541905879974365, "learning_rate": 0.0005, "loss": 1.305817723274231, "step": 43652 }, { "epoch": 3.4413086322428064, "grad_norm": 0.5388187170028687, "learning_rate": 0.0005, "loss": 1.511073112487793, "step": 43653 }, { "epoch": 3.4413874655104455, "grad_norm": 0.5197423100471497, "learning_rate": 0.0005, "loss": 1.3136683702468872, "step": 43654 }, { "epoch": 3.441466298778084, "grad_norm": 0.46869292855262756, "learning_rate": 0.0005, "loss": 1.258360505104065, "step": 43655 }, { "epoch": 3.4415451320457233, "grad_norm": 0.4939497113227844, "learning_rate": 0.0005, "loss": 1.2320530414581299, "step": 43656 }, { "epoch": 3.4416239653133625, "grad_norm": 0.6514589786529541, "learning_rate": 0.0005, "loss": 1.4575269222259521, "step": 43657 }, { "epoch": 3.441702798581001, "grad_norm": 0.5084625482559204, "learning_rate": 0.0005, "loss": 1.2553083896636963, "step": 43658 }, { "epoch": 3.4417816318486403, "grad_norm": 0.5215389728546143, "learning_rate": 0.0005, "loss": 1.2310727834701538, "step": 43659 }, { "epoch": 3.441860465116279, "grad_norm": 0.5696213841438293, "learning_rate": 0.0005, "loss": 1.3448346853256226, "step": 43660 }, { "epoch": 3.441939298383918, "grad_norm": 0.3778095841407776, "learning_rate": 0.0005, "loss": 1.22784423828125, "step": 43661 }, { "epoch": 3.442018131651557, "grad_norm": 0.4829685091972351, "learning_rate": 0.0005, "loss": 1.2516088485717773, "step": 43662 }, { "epoch": 3.442096964919196, "grad_norm": 0.49805012345314026, "learning_rate": 0.0005, "loss": 1.2931859493255615, "step": 43663 }, { "epoch": 3.4421757981868346, "grad_norm": 0.4776909053325653, "learning_rate": 0.0005, "loss": 1.466339349746704, "step": 43664 }, { "epoch": 3.4422546314544737, "grad_norm": 0.4887171983718872, "learning_rate": 0.0005, "loss": 1.1913509368896484, "step": 43665 }, { "epoch": 3.442333464722113, "grad_norm": 0.5472883582115173, "learning_rate": 0.0005, "loss": 1.2743936777114868, "step": 43666 }, { "epoch": 3.4424122979897516, "grad_norm": 0.4720466434955597, "learning_rate": 0.0005, "loss": 1.0852679014205933, "step": 43667 }, { "epoch": 3.4424911312573907, "grad_norm": 0.6003636121749878, "learning_rate": 0.0005, "loss": 1.19492769241333, "step": 43668 }, { "epoch": 3.4425699645250294, "grad_norm": 0.4467200040817261, "learning_rate": 0.0005, "loss": 1.213687777519226, "step": 43669 }, { "epoch": 3.4426487977926685, "grad_norm": 0.5439389944076538, "learning_rate": 0.0005, "loss": 1.3796299695968628, "step": 43670 }, { "epoch": 3.4427276310603077, "grad_norm": 0.5076042413711548, "learning_rate": 0.0005, "loss": 1.0978333950042725, "step": 43671 }, { "epoch": 3.4428064643279463, "grad_norm": 0.4832623600959778, "learning_rate": 0.0005, "loss": 1.277155876159668, "step": 43672 }, { "epoch": 3.4428852975955855, "grad_norm": 0.5234571099281311, "learning_rate": 0.0005, "loss": 1.218309760093689, "step": 43673 }, { "epoch": 3.442964130863224, "grad_norm": 0.5443931818008423, "learning_rate": 0.0005, "loss": 1.3610200881958008, "step": 43674 }, { "epoch": 3.4430429641308633, "grad_norm": 0.5038202404975891, "learning_rate": 0.0005, "loss": 1.428662657737732, "step": 43675 }, { "epoch": 3.443121797398502, "grad_norm": 0.6454592943191528, "learning_rate": 0.0005, "loss": 1.3815358877182007, "step": 43676 }, { "epoch": 3.443200630666141, "grad_norm": 0.6978784799575806, "learning_rate": 0.0005, "loss": 1.4008159637451172, "step": 43677 }, { "epoch": 3.44327946393378, "grad_norm": 0.5079542994499207, "learning_rate": 0.0005, "loss": 1.301867961883545, "step": 43678 }, { "epoch": 3.443358297201419, "grad_norm": 0.4832729995250702, "learning_rate": 0.0005, "loss": 1.3441483974456787, "step": 43679 }, { "epoch": 3.443437130469058, "grad_norm": 0.454116553068161, "learning_rate": 0.0005, "loss": 1.2517168521881104, "step": 43680 }, { "epoch": 3.4435159637366968, "grad_norm": 0.5511747002601624, "learning_rate": 0.0005, "loss": 1.185707926750183, "step": 43681 }, { "epoch": 3.443594797004336, "grad_norm": 0.48151731491088867, "learning_rate": 0.0005, "loss": 1.2905526161193848, "step": 43682 }, { "epoch": 3.4436736302719746, "grad_norm": 0.498606413602829, "learning_rate": 0.0005, "loss": 1.2967942953109741, "step": 43683 }, { "epoch": 3.4437524635396137, "grad_norm": 0.5306023955345154, "learning_rate": 0.0005, "loss": 1.2142126560211182, "step": 43684 }, { "epoch": 3.443831296807253, "grad_norm": 0.4654114842414856, "learning_rate": 0.0005, "loss": 1.3782172203063965, "step": 43685 }, { "epoch": 3.4439101300748916, "grad_norm": 0.47570300102233887, "learning_rate": 0.0005, "loss": 1.2509738206863403, "step": 43686 }, { "epoch": 3.4439889633425307, "grad_norm": 0.4589250683784485, "learning_rate": 0.0005, "loss": 1.1634576320648193, "step": 43687 }, { "epoch": 3.4440677966101694, "grad_norm": 0.49753063917160034, "learning_rate": 0.0005, "loss": 1.3739995956420898, "step": 43688 }, { "epoch": 3.4441466298778085, "grad_norm": 0.5106716752052307, "learning_rate": 0.0005, "loss": 1.3189644813537598, "step": 43689 }, { "epoch": 3.444225463145447, "grad_norm": 0.4877687692642212, "learning_rate": 0.0005, "loss": 1.3724262714385986, "step": 43690 }, { "epoch": 3.4443042964130863, "grad_norm": 0.4883635342121124, "learning_rate": 0.0005, "loss": 1.3262913227081299, "step": 43691 }, { "epoch": 3.444383129680725, "grad_norm": 0.5047471523284912, "learning_rate": 0.0005, "loss": 1.1716290712356567, "step": 43692 }, { "epoch": 3.444461962948364, "grad_norm": 0.5027888417243958, "learning_rate": 0.0005, "loss": 1.2131571769714355, "step": 43693 }, { "epoch": 3.4445407962160033, "grad_norm": 0.3656828701496124, "learning_rate": 0.0005, "loss": 1.183549404144287, "step": 43694 }, { "epoch": 3.444619629483642, "grad_norm": 0.4732455313205719, "learning_rate": 0.0005, "loss": 1.4622857570648193, "step": 43695 }, { "epoch": 3.444698462751281, "grad_norm": 0.5896660089492798, "learning_rate": 0.0005, "loss": 1.3573569059371948, "step": 43696 }, { "epoch": 3.44477729601892, "grad_norm": 0.4406907856464386, "learning_rate": 0.0005, "loss": 1.4617135524749756, "step": 43697 }, { "epoch": 3.444856129286559, "grad_norm": 0.5186634659767151, "learning_rate": 0.0005, "loss": 1.2049163579940796, "step": 43698 }, { "epoch": 3.444934962554198, "grad_norm": 0.5064815878868103, "learning_rate": 0.0005, "loss": 1.1400681734085083, "step": 43699 }, { "epoch": 3.4450137958218368, "grad_norm": 0.46577346324920654, "learning_rate": 0.0005, "loss": 1.2911884784698486, "step": 43700 }, { "epoch": 3.445092629089476, "grad_norm": 0.6741596460342407, "learning_rate": 0.0005, "loss": 1.291825771331787, "step": 43701 }, { "epoch": 3.4451714623571146, "grad_norm": 0.5520520806312561, "learning_rate": 0.0005, "loss": 1.2136800289154053, "step": 43702 }, { "epoch": 3.4452502956247537, "grad_norm": 0.4883486032485962, "learning_rate": 0.0005, "loss": 1.3549370765686035, "step": 43703 }, { "epoch": 3.4453291288923924, "grad_norm": 0.598175048828125, "learning_rate": 0.0005, "loss": 1.1821290254592896, "step": 43704 }, { "epoch": 3.4454079621600315, "grad_norm": 0.5206419825553894, "learning_rate": 0.0005, "loss": 1.2478609085083008, "step": 43705 }, { "epoch": 3.4454867954276707, "grad_norm": 0.4864487648010254, "learning_rate": 0.0005, "loss": 1.1886630058288574, "step": 43706 }, { "epoch": 3.4455656286953094, "grad_norm": 0.5152319073677063, "learning_rate": 0.0005, "loss": 1.3173261880874634, "step": 43707 }, { "epoch": 3.4456444619629485, "grad_norm": 0.4444006085395813, "learning_rate": 0.0005, "loss": 1.192507028579712, "step": 43708 }, { "epoch": 3.445723295230587, "grad_norm": 0.4974558353424072, "learning_rate": 0.0005, "loss": 1.3766402006149292, "step": 43709 }, { "epoch": 3.4458021284982263, "grad_norm": 0.5551294684410095, "learning_rate": 0.0005, "loss": 1.2679791450500488, "step": 43710 }, { "epoch": 3.445880961765865, "grad_norm": 0.6448603272438049, "learning_rate": 0.0005, "loss": 1.5416123867034912, "step": 43711 }, { "epoch": 3.445959795033504, "grad_norm": 0.48512688279151917, "learning_rate": 0.0005, "loss": 1.2493054866790771, "step": 43712 }, { "epoch": 3.4460386283011433, "grad_norm": 0.572147786617279, "learning_rate": 0.0005, "loss": 1.2929015159606934, "step": 43713 }, { "epoch": 3.446117461568782, "grad_norm": 0.48389917612075806, "learning_rate": 0.0005, "loss": 1.4158042669296265, "step": 43714 }, { "epoch": 3.446196294836421, "grad_norm": 0.47525304555892944, "learning_rate": 0.0005, "loss": 1.356130599975586, "step": 43715 }, { "epoch": 3.44627512810406, "grad_norm": 0.4595698118209839, "learning_rate": 0.0005, "loss": 1.2959001064300537, "step": 43716 }, { "epoch": 3.446353961371699, "grad_norm": 0.5687461495399475, "learning_rate": 0.0005, "loss": 1.302316665649414, "step": 43717 }, { "epoch": 3.446432794639338, "grad_norm": 0.5123456716537476, "learning_rate": 0.0005, "loss": 1.440299153327942, "step": 43718 }, { "epoch": 3.4465116279069767, "grad_norm": 0.620614767074585, "learning_rate": 0.0005, "loss": 1.3983268737792969, "step": 43719 }, { "epoch": 3.446590461174616, "grad_norm": 0.6639484167098999, "learning_rate": 0.0005, "loss": 1.3169724941253662, "step": 43720 }, { "epoch": 3.4466692944422546, "grad_norm": 0.4644039273262024, "learning_rate": 0.0005, "loss": 1.0848580598831177, "step": 43721 }, { "epoch": 3.4467481277098937, "grad_norm": 0.5885260701179504, "learning_rate": 0.0005, "loss": 1.323127269744873, "step": 43722 }, { "epoch": 3.4468269609775324, "grad_norm": 0.504862904548645, "learning_rate": 0.0005, "loss": 1.2500462532043457, "step": 43723 }, { "epoch": 3.4469057942451715, "grad_norm": 1.601050853729248, "learning_rate": 0.0005, "loss": 1.1373403072357178, "step": 43724 }, { "epoch": 3.44698462751281, "grad_norm": 0.5049881339073181, "learning_rate": 0.0005, "loss": 1.292389154434204, "step": 43725 }, { "epoch": 3.4470634607804493, "grad_norm": 0.5373679399490356, "learning_rate": 0.0005, "loss": 1.2579975128173828, "step": 43726 }, { "epoch": 3.4471422940480885, "grad_norm": 0.5794665217399597, "learning_rate": 0.0005, "loss": 1.2047278881072998, "step": 43727 }, { "epoch": 3.447221127315727, "grad_norm": 0.6348727345466614, "learning_rate": 0.0005, "loss": 1.308746337890625, "step": 43728 }, { "epoch": 3.4472999605833663, "grad_norm": 0.4936203956604004, "learning_rate": 0.0005, "loss": 1.2828800678253174, "step": 43729 }, { "epoch": 3.447378793851005, "grad_norm": 0.46671050786972046, "learning_rate": 0.0005, "loss": 1.2499473094940186, "step": 43730 }, { "epoch": 3.447457627118644, "grad_norm": 0.4613213539123535, "learning_rate": 0.0005, "loss": 1.2604570388793945, "step": 43731 }, { "epoch": 3.4475364603862833, "grad_norm": 0.4272841215133667, "learning_rate": 0.0005, "loss": 1.3171570301055908, "step": 43732 }, { "epoch": 3.447615293653922, "grad_norm": 0.4802916646003723, "learning_rate": 0.0005, "loss": 1.2915754318237305, "step": 43733 }, { "epoch": 3.447694126921561, "grad_norm": 0.47407519817352295, "learning_rate": 0.0005, "loss": 1.1339493989944458, "step": 43734 }, { "epoch": 3.4477729601891998, "grad_norm": 0.550498366355896, "learning_rate": 0.0005, "loss": 1.3190608024597168, "step": 43735 }, { "epoch": 3.447851793456839, "grad_norm": 0.5961028933525085, "learning_rate": 0.0005, "loss": 1.402818202972412, "step": 43736 }, { "epoch": 3.4479306267244776, "grad_norm": 0.4854212999343872, "learning_rate": 0.0005, "loss": 1.1843013763427734, "step": 43737 }, { "epoch": 3.4480094599921167, "grad_norm": 0.42849040031433105, "learning_rate": 0.0005, "loss": 1.273390293121338, "step": 43738 }, { "epoch": 3.4480882932597554, "grad_norm": 0.48825037479400635, "learning_rate": 0.0005, "loss": 1.4436067342758179, "step": 43739 }, { "epoch": 3.4481671265273945, "grad_norm": 0.5354508757591248, "learning_rate": 0.0005, "loss": 1.1409145593643188, "step": 43740 }, { "epoch": 3.4482459597950337, "grad_norm": 0.5179951190948486, "learning_rate": 0.0005, "loss": 1.2343767881393433, "step": 43741 }, { "epoch": 3.4483247930626724, "grad_norm": 0.45106858015060425, "learning_rate": 0.0005, "loss": 1.2735673189163208, "step": 43742 }, { "epoch": 3.4484036263303115, "grad_norm": 0.4150621294975281, "learning_rate": 0.0005, "loss": 1.238560438156128, "step": 43743 }, { "epoch": 3.44848245959795, "grad_norm": 0.49489855766296387, "learning_rate": 0.0005, "loss": 1.199102520942688, "step": 43744 }, { "epoch": 3.4485612928655893, "grad_norm": 0.4969518780708313, "learning_rate": 0.0005, "loss": 1.1950291395187378, "step": 43745 }, { "epoch": 3.4486401261332285, "grad_norm": 0.48919302225112915, "learning_rate": 0.0005, "loss": 1.3085896968841553, "step": 43746 }, { "epoch": 3.448718959400867, "grad_norm": 0.4548933207988739, "learning_rate": 0.0005, "loss": 1.2564213275909424, "step": 43747 }, { "epoch": 3.4487977926685063, "grad_norm": 0.5315878391265869, "learning_rate": 0.0005, "loss": 1.3386651277542114, "step": 43748 }, { "epoch": 3.448876625936145, "grad_norm": 0.5675352811813354, "learning_rate": 0.0005, "loss": 1.3110430240631104, "step": 43749 }, { "epoch": 3.448955459203784, "grad_norm": 0.4518357515335083, "learning_rate": 0.0005, "loss": 1.373713493347168, "step": 43750 }, { "epoch": 3.449034292471423, "grad_norm": 0.42257559299468994, "learning_rate": 0.0005, "loss": 1.2970707416534424, "step": 43751 }, { "epoch": 3.449113125739062, "grad_norm": 0.4467068910598755, "learning_rate": 0.0005, "loss": 1.1367387771606445, "step": 43752 }, { "epoch": 3.4491919590067006, "grad_norm": 0.6279106736183167, "learning_rate": 0.0005, "loss": 1.1534881591796875, "step": 43753 }, { "epoch": 3.4492707922743397, "grad_norm": 0.5342279076576233, "learning_rate": 0.0005, "loss": 1.3466417789459229, "step": 43754 }, { "epoch": 3.449349625541979, "grad_norm": 0.5048367977142334, "learning_rate": 0.0005, "loss": 1.1678063869476318, "step": 43755 }, { "epoch": 3.4494284588096176, "grad_norm": 0.46654975414276123, "learning_rate": 0.0005, "loss": 1.352990984916687, "step": 43756 }, { "epoch": 3.4495072920772567, "grad_norm": 0.5300517678260803, "learning_rate": 0.0005, "loss": 1.160538911819458, "step": 43757 }, { "epoch": 3.4495861253448954, "grad_norm": 0.6673579812049866, "learning_rate": 0.0005, "loss": 1.424926996231079, "step": 43758 }, { "epoch": 3.4496649586125345, "grad_norm": 0.48366618156433105, "learning_rate": 0.0005, "loss": 1.1921117305755615, "step": 43759 }, { "epoch": 3.4497437918801737, "grad_norm": 0.4254242777824402, "learning_rate": 0.0005, "loss": 1.2003990411758423, "step": 43760 }, { "epoch": 3.4498226251478123, "grad_norm": 0.6009359955787659, "learning_rate": 0.0005, "loss": 1.4761841297149658, "step": 43761 }, { "epoch": 3.4499014584154515, "grad_norm": 0.4783530533313751, "learning_rate": 0.0005, "loss": 1.3247456550598145, "step": 43762 }, { "epoch": 3.44998029168309, "grad_norm": 0.4471220374107361, "learning_rate": 0.0005, "loss": 1.3347644805908203, "step": 43763 }, { "epoch": 3.4500591249507293, "grad_norm": 0.4461571276187897, "learning_rate": 0.0005, "loss": 1.1570262908935547, "step": 43764 }, { "epoch": 3.450137958218368, "grad_norm": 0.480812668800354, "learning_rate": 0.0005, "loss": 1.321843147277832, "step": 43765 }, { "epoch": 3.450216791486007, "grad_norm": 0.5181843042373657, "learning_rate": 0.0005, "loss": 1.305985689163208, "step": 43766 }, { "epoch": 3.450295624753646, "grad_norm": 0.43503478169441223, "learning_rate": 0.0005, "loss": 1.0756118297576904, "step": 43767 }, { "epoch": 3.450374458021285, "grad_norm": 0.5697928071022034, "learning_rate": 0.0005, "loss": 1.3353952169418335, "step": 43768 }, { "epoch": 3.450453291288924, "grad_norm": 0.7717330455780029, "learning_rate": 0.0005, "loss": 1.310896396636963, "step": 43769 }, { "epoch": 3.4505321245565628, "grad_norm": 0.4764523506164551, "learning_rate": 0.0005, "loss": 1.254429817199707, "step": 43770 }, { "epoch": 3.450610957824202, "grad_norm": 0.5094406604766846, "learning_rate": 0.0005, "loss": 1.2664248943328857, "step": 43771 }, { "epoch": 3.4506897910918406, "grad_norm": 0.4865184426307678, "learning_rate": 0.0005, "loss": 1.1769862174987793, "step": 43772 }, { "epoch": 3.4507686243594797, "grad_norm": 0.5414383411407471, "learning_rate": 0.0005, "loss": 1.4387599229812622, "step": 43773 }, { "epoch": 3.450847457627119, "grad_norm": 0.4489314556121826, "learning_rate": 0.0005, "loss": 1.2184844017028809, "step": 43774 }, { "epoch": 3.4509262908947576, "grad_norm": 0.6566852331161499, "learning_rate": 0.0005, "loss": 1.4221711158752441, "step": 43775 }, { "epoch": 3.4510051241623967, "grad_norm": 0.435133159160614, "learning_rate": 0.0005, "loss": 1.169726848602295, "step": 43776 }, { "epoch": 3.4510839574300354, "grad_norm": 0.9004863500595093, "learning_rate": 0.0005, "loss": 1.2553621530532837, "step": 43777 }, { "epoch": 3.4511627906976745, "grad_norm": 0.4655000865459442, "learning_rate": 0.0005, "loss": 1.3362690210342407, "step": 43778 }, { "epoch": 3.451241623965313, "grad_norm": 0.4854734539985657, "learning_rate": 0.0005, "loss": 1.2711433172225952, "step": 43779 }, { "epoch": 3.4513204572329523, "grad_norm": 0.48329538106918335, "learning_rate": 0.0005, "loss": 1.298656940460205, "step": 43780 }, { "epoch": 3.451399290500591, "grad_norm": 0.42474842071533203, "learning_rate": 0.0005, "loss": 1.1491472721099854, "step": 43781 }, { "epoch": 3.45147812376823, "grad_norm": 0.5240734219551086, "learning_rate": 0.0005, "loss": 1.2811458110809326, "step": 43782 }, { "epoch": 3.4515569570358693, "grad_norm": 0.4303706884384155, "learning_rate": 0.0005, "loss": 1.2303752899169922, "step": 43783 }, { "epoch": 3.451635790303508, "grad_norm": 0.5871094465255737, "learning_rate": 0.0005, "loss": 1.4611968994140625, "step": 43784 }, { "epoch": 3.451714623571147, "grad_norm": 0.49235665798187256, "learning_rate": 0.0005, "loss": 1.3544033765792847, "step": 43785 }, { "epoch": 3.451793456838786, "grad_norm": 0.5497705936431885, "learning_rate": 0.0005, "loss": 1.3910574913024902, "step": 43786 }, { "epoch": 3.451872290106425, "grad_norm": 0.5149058103561401, "learning_rate": 0.0005, "loss": 1.2811402082443237, "step": 43787 }, { "epoch": 3.451951123374064, "grad_norm": 0.3626059591770172, "learning_rate": 0.0005, "loss": 1.0432332754135132, "step": 43788 }, { "epoch": 3.4520299566417028, "grad_norm": 0.452690452337265, "learning_rate": 0.0005, "loss": 1.158758521080017, "step": 43789 }, { "epoch": 3.452108789909342, "grad_norm": 0.5581976771354675, "learning_rate": 0.0005, "loss": 1.227492332458496, "step": 43790 }, { "epoch": 3.4521876231769806, "grad_norm": 0.4591301679611206, "learning_rate": 0.0005, "loss": 1.1078417301177979, "step": 43791 }, { "epoch": 3.4522664564446197, "grad_norm": 0.5997911095619202, "learning_rate": 0.0005, "loss": 1.3562538623809814, "step": 43792 }, { "epoch": 3.4523452897122584, "grad_norm": 0.520743191242218, "learning_rate": 0.0005, "loss": 1.3840919733047485, "step": 43793 }, { "epoch": 3.4524241229798975, "grad_norm": 0.4658852815628052, "learning_rate": 0.0005, "loss": 1.3947672843933105, "step": 43794 }, { "epoch": 3.4525029562475362, "grad_norm": 0.4419039189815521, "learning_rate": 0.0005, "loss": 1.1174347400665283, "step": 43795 }, { "epoch": 3.4525817895151754, "grad_norm": 0.5072311758995056, "learning_rate": 0.0005, "loss": 1.3664062023162842, "step": 43796 }, { "epoch": 3.4526606227828145, "grad_norm": 0.5250522494316101, "learning_rate": 0.0005, "loss": 1.336594820022583, "step": 43797 }, { "epoch": 3.452739456050453, "grad_norm": 0.4196341931819916, "learning_rate": 0.0005, "loss": 1.0282557010650635, "step": 43798 }, { "epoch": 3.4528182893180923, "grad_norm": 0.5217872262001038, "learning_rate": 0.0005, "loss": 1.3079452514648438, "step": 43799 }, { "epoch": 3.452897122585731, "grad_norm": 0.4585111141204834, "learning_rate": 0.0005, "loss": 1.1576834917068481, "step": 43800 }, { "epoch": 3.45297595585337, "grad_norm": 0.706707239151001, "learning_rate": 0.0005, "loss": 1.3815937042236328, "step": 43801 }, { "epoch": 3.4530547891210093, "grad_norm": 0.49818089604377747, "learning_rate": 0.0005, "loss": 1.1951119899749756, "step": 43802 }, { "epoch": 3.453133622388648, "grad_norm": 0.39750349521636963, "learning_rate": 0.0005, "loss": 1.121811866760254, "step": 43803 }, { "epoch": 3.453212455656287, "grad_norm": 0.5982358455657959, "learning_rate": 0.0005, "loss": 1.337576150894165, "step": 43804 }, { "epoch": 3.453291288923926, "grad_norm": 0.4929872453212738, "learning_rate": 0.0005, "loss": 1.3081040382385254, "step": 43805 }, { "epoch": 3.453370122191565, "grad_norm": 0.47465717792510986, "learning_rate": 0.0005, "loss": 1.223488688468933, "step": 43806 }, { "epoch": 3.4534489554592036, "grad_norm": 0.46720999479293823, "learning_rate": 0.0005, "loss": 1.4193246364593506, "step": 43807 }, { "epoch": 3.4535277887268427, "grad_norm": 0.5500472187995911, "learning_rate": 0.0005, "loss": 1.3126293420791626, "step": 43808 }, { "epoch": 3.4536066219944814, "grad_norm": 0.6767479181289673, "learning_rate": 0.0005, "loss": 1.273970127105713, "step": 43809 }, { "epoch": 3.4536854552621206, "grad_norm": 0.47235971689224243, "learning_rate": 0.0005, "loss": 1.2474268674850464, "step": 43810 }, { "epoch": 3.4537642885297597, "grad_norm": 1.5716972351074219, "learning_rate": 0.0005, "loss": 1.169691801071167, "step": 43811 }, { "epoch": 3.4538431217973984, "grad_norm": 0.473375141620636, "learning_rate": 0.0005, "loss": 1.2261955738067627, "step": 43812 }, { "epoch": 3.4539219550650375, "grad_norm": 0.5778483748435974, "learning_rate": 0.0005, "loss": 1.2246966361999512, "step": 43813 }, { "epoch": 3.454000788332676, "grad_norm": 0.46735313534736633, "learning_rate": 0.0005, "loss": 1.0438144207000732, "step": 43814 }, { "epoch": 3.4540796216003153, "grad_norm": 0.7614744901657104, "learning_rate": 0.0005, "loss": 1.4430526494979858, "step": 43815 }, { "epoch": 3.4541584548679545, "grad_norm": 0.4938899576663971, "learning_rate": 0.0005, "loss": 1.4619908332824707, "step": 43816 }, { "epoch": 3.454237288135593, "grad_norm": 0.42833930253982544, "learning_rate": 0.0005, "loss": 1.1721071004867554, "step": 43817 }, { "epoch": 3.4543161214032323, "grad_norm": 0.4111832082271576, "learning_rate": 0.0005, "loss": 1.0701453685760498, "step": 43818 }, { "epoch": 3.454394954670871, "grad_norm": 0.5754592418670654, "learning_rate": 0.0005, "loss": 1.270618200302124, "step": 43819 }, { "epoch": 3.45447378793851, "grad_norm": 0.564616858959198, "learning_rate": 0.0005, "loss": 1.1682145595550537, "step": 43820 }, { "epoch": 3.454552621206149, "grad_norm": 0.5744191408157349, "learning_rate": 0.0005, "loss": 1.38565194606781, "step": 43821 }, { "epoch": 3.454631454473788, "grad_norm": 0.5159692168235779, "learning_rate": 0.0005, "loss": 1.273038625717163, "step": 43822 }, { "epoch": 3.4547102877414266, "grad_norm": 0.5438783764839172, "learning_rate": 0.0005, "loss": 1.1909717321395874, "step": 43823 }, { "epoch": 3.4547891210090658, "grad_norm": 0.45800045132637024, "learning_rate": 0.0005, "loss": 1.1662620306015015, "step": 43824 }, { "epoch": 3.454867954276705, "grad_norm": 0.5135069489479065, "learning_rate": 0.0005, "loss": 1.2017290592193604, "step": 43825 }, { "epoch": 3.4549467875443436, "grad_norm": 0.5096903443336487, "learning_rate": 0.0005, "loss": 1.195009708404541, "step": 43826 }, { "epoch": 3.4550256208119827, "grad_norm": 0.5022465586662292, "learning_rate": 0.0005, "loss": 1.2576627731323242, "step": 43827 }, { "epoch": 3.4551044540796214, "grad_norm": 0.4964195191860199, "learning_rate": 0.0005, "loss": 1.2197821140289307, "step": 43828 }, { "epoch": 3.4551832873472605, "grad_norm": 0.5102643966674805, "learning_rate": 0.0005, "loss": 1.322819709777832, "step": 43829 }, { "epoch": 3.4552621206148997, "grad_norm": 0.5056937336921692, "learning_rate": 0.0005, "loss": 1.2146995067596436, "step": 43830 }, { "epoch": 3.4553409538825384, "grad_norm": 0.506433367729187, "learning_rate": 0.0005, "loss": 1.127708911895752, "step": 43831 }, { "epoch": 3.4554197871501775, "grad_norm": 0.5581032037734985, "learning_rate": 0.0005, "loss": 1.3657413721084595, "step": 43832 }, { "epoch": 3.455498620417816, "grad_norm": 0.4794486165046692, "learning_rate": 0.0005, "loss": 1.2871599197387695, "step": 43833 }, { "epoch": 3.4555774536854553, "grad_norm": 0.5817059874534607, "learning_rate": 0.0005, "loss": 1.520534873008728, "step": 43834 }, { "epoch": 3.455656286953094, "grad_norm": 0.44231942296028137, "learning_rate": 0.0005, "loss": 1.1287599802017212, "step": 43835 }, { "epoch": 3.455735120220733, "grad_norm": 0.5035330653190613, "learning_rate": 0.0005, "loss": 1.339063048362732, "step": 43836 }, { "epoch": 3.4558139534883723, "grad_norm": 0.5155431032180786, "learning_rate": 0.0005, "loss": 1.2883527278900146, "step": 43837 }, { "epoch": 3.455892786756011, "grad_norm": 0.5055833458900452, "learning_rate": 0.0005, "loss": 1.2457144260406494, "step": 43838 }, { "epoch": 3.45597162002365, "grad_norm": 0.4927370846271515, "learning_rate": 0.0005, "loss": 1.2021864652633667, "step": 43839 }, { "epoch": 3.456050453291289, "grad_norm": 0.5228103995323181, "learning_rate": 0.0005, "loss": 1.2830522060394287, "step": 43840 }, { "epoch": 3.456129286558928, "grad_norm": 0.4780651926994324, "learning_rate": 0.0005, "loss": 1.1443157196044922, "step": 43841 }, { "epoch": 3.4562081198265666, "grad_norm": 0.48311764001846313, "learning_rate": 0.0005, "loss": 1.2671643495559692, "step": 43842 }, { "epoch": 3.4562869530942057, "grad_norm": 0.4582333564758301, "learning_rate": 0.0005, "loss": 1.178335189819336, "step": 43843 }, { "epoch": 3.456365786361845, "grad_norm": 0.4129616916179657, "learning_rate": 0.0005, "loss": 1.0973972082138062, "step": 43844 }, { "epoch": 3.4564446196294836, "grad_norm": 0.5062204599380493, "learning_rate": 0.0005, "loss": 1.3601419925689697, "step": 43845 }, { "epoch": 3.4565234528971227, "grad_norm": 0.5002597570419312, "learning_rate": 0.0005, "loss": 1.362540364265442, "step": 43846 }, { "epoch": 3.4566022861647614, "grad_norm": 0.4608440399169922, "learning_rate": 0.0005, "loss": 1.281070590019226, "step": 43847 }, { "epoch": 3.4566811194324005, "grad_norm": 0.4590419828891754, "learning_rate": 0.0005, "loss": 1.239074468612671, "step": 43848 }, { "epoch": 3.4567599527000397, "grad_norm": 0.4437101185321808, "learning_rate": 0.0005, "loss": 1.1143028736114502, "step": 43849 }, { "epoch": 3.4568387859676784, "grad_norm": 0.4197205603122711, "learning_rate": 0.0005, "loss": 1.0585509538650513, "step": 43850 }, { "epoch": 3.4569176192353175, "grad_norm": 0.5122214555740356, "learning_rate": 0.0005, "loss": 1.3268229961395264, "step": 43851 }, { "epoch": 3.456996452502956, "grad_norm": 0.41277337074279785, "learning_rate": 0.0005, "loss": 1.0992851257324219, "step": 43852 }, { "epoch": 3.4570752857705953, "grad_norm": 0.4408724904060364, "learning_rate": 0.0005, "loss": 1.2628202438354492, "step": 43853 }, { "epoch": 3.457154119038234, "grad_norm": 0.6001467704772949, "learning_rate": 0.0005, "loss": 1.0762333869934082, "step": 43854 }, { "epoch": 3.457232952305873, "grad_norm": 0.4654134511947632, "learning_rate": 0.0005, "loss": 1.3145828247070312, "step": 43855 }, { "epoch": 3.457311785573512, "grad_norm": 0.4290994107723236, "learning_rate": 0.0005, "loss": 1.209053874015808, "step": 43856 }, { "epoch": 3.457390618841151, "grad_norm": 0.6423136591911316, "learning_rate": 0.0005, "loss": 1.3477518558502197, "step": 43857 }, { "epoch": 3.45746945210879, "grad_norm": 0.5541141033172607, "learning_rate": 0.0005, "loss": 1.4687767028808594, "step": 43858 }, { "epoch": 3.4575482853764288, "grad_norm": 0.4256178140640259, "learning_rate": 0.0005, "loss": 1.0760374069213867, "step": 43859 }, { "epoch": 3.457627118644068, "grad_norm": 0.48502251505851746, "learning_rate": 0.0005, "loss": 1.2331374883651733, "step": 43860 }, { "epoch": 3.4577059519117066, "grad_norm": 0.48469996452331543, "learning_rate": 0.0005, "loss": 1.1242995262145996, "step": 43861 }, { "epoch": 3.4577847851793457, "grad_norm": 0.45883235335350037, "learning_rate": 0.0005, "loss": 1.116687536239624, "step": 43862 }, { "epoch": 3.457863618446985, "grad_norm": 0.5560346841812134, "learning_rate": 0.0005, "loss": 1.1816620826721191, "step": 43863 }, { "epoch": 3.4579424517146236, "grad_norm": 0.5747630000114441, "learning_rate": 0.0005, "loss": 1.3298654556274414, "step": 43864 }, { "epoch": 3.4580212849822627, "grad_norm": 0.41306519508361816, "learning_rate": 0.0005, "loss": 0.9793171882629395, "step": 43865 }, { "epoch": 3.4581001182499014, "grad_norm": 0.43189576268196106, "learning_rate": 0.0005, "loss": 1.2596851587295532, "step": 43866 }, { "epoch": 3.4581789515175405, "grad_norm": 0.4816450774669647, "learning_rate": 0.0005, "loss": 1.3253562450408936, "step": 43867 }, { "epoch": 3.458257784785179, "grad_norm": 0.39342615008354187, "learning_rate": 0.0005, "loss": 1.1831305027008057, "step": 43868 }, { "epoch": 3.4583366180528183, "grad_norm": 0.4425601363182068, "learning_rate": 0.0005, "loss": 1.201378345489502, "step": 43869 }, { "epoch": 3.458415451320457, "grad_norm": 0.5032881498336792, "learning_rate": 0.0005, "loss": 1.4432642459869385, "step": 43870 }, { "epoch": 3.458494284588096, "grad_norm": 0.44192129373550415, "learning_rate": 0.0005, "loss": 1.2630600929260254, "step": 43871 }, { "epoch": 3.4585731178557353, "grad_norm": 0.5710341334342957, "learning_rate": 0.0005, "loss": 1.4949990510940552, "step": 43872 }, { "epoch": 3.458651951123374, "grad_norm": 0.43987739086151123, "learning_rate": 0.0005, "loss": 1.3129860162734985, "step": 43873 }, { "epoch": 3.458730784391013, "grad_norm": 0.5310162305831909, "learning_rate": 0.0005, "loss": 1.4791520833969116, "step": 43874 }, { "epoch": 3.458809617658652, "grad_norm": 0.4674389958381653, "learning_rate": 0.0005, "loss": 1.1377158164978027, "step": 43875 }, { "epoch": 3.458888450926291, "grad_norm": 0.5971294045448303, "learning_rate": 0.0005, "loss": 1.3902353048324585, "step": 43876 }, { "epoch": 3.45896728419393, "grad_norm": 0.4795667231082916, "learning_rate": 0.0005, "loss": 1.2441091537475586, "step": 43877 }, { "epoch": 3.4590461174615688, "grad_norm": 0.5258584022521973, "learning_rate": 0.0005, "loss": 1.2575838565826416, "step": 43878 }, { "epoch": 3.459124950729208, "grad_norm": 0.47062981128692627, "learning_rate": 0.0005, "loss": 1.2584255933761597, "step": 43879 }, { "epoch": 3.4592037839968466, "grad_norm": 0.47256341576576233, "learning_rate": 0.0005, "loss": 1.3052778244018555, "step": 43880 }, { "epoch": 3.4592826172644857, "grad_norm": 0.4854395389556885, "learning_rate": 0.0005, "loss": 1.3133121728897095, "step": 43881 }, { "epoch": 3.4593614505321244, "grad_norm": 0.48478925228118896, "learning_rate": 0.0005, "loss": 1.3998620510101318, "step": 43882 }, { "epoch": 3.4594402837997635, "grad_norm": 0.5418775677680969, "learning_rate": 0.0005, "loss": 1.4496519565582275, "step": 43883 }, { "epoch": 3.4595191170674022, "grad_norm": 0.437349408864975, "learning_rate": 0.0005, "loss": 1.434457540512085, "step": 43884 }, { "epoch": 3.4595979503350414, "grad_norm": 0.4946899712085724, "learning_rate": 0.0005, "loss": 1.313520908355713, "step": 43885 }, { "epoch": 3.4596767836026805, "grad_norm": 0.4999677240848541, "learning_rate": 0.0005, "loss": 1.3400057554244995, "step": 43886 }, { "epoch": 3.459755616870319, "grad_norm": 0.4157850742340088, "learning_rate": 0.0005, "loss": 1.150496482849121, "step": 43887 }, { "epoch": 3.4598344501379583, "grad_norm": 0.5880365371704102, "learning_rate": 0.0005, "loss": 1.3889600038528442, "step": 43888 }, { "epoch": 3.459913283405597, "grad_norm": 0.39274534583091736, "learning_rate": 0.0005, "loss": 1.275599479675293, "step": 43889 }, { "epoch": 3.459992116673236, "grad_norm": 0.5005271434783936, "learning_rate": 0.0005, "loss": 1.2707278728485107, "step": 43890 }, { "epoch": 3.4600709499408753, "grad_norm": 0.4780334532260895, "learning_rate": 0.0005, "loss": 1.1576197147369385, "step": 43891 }, { "epoch": 3.460149783208514, "grad_norm": 0.47097697854042053, "learning_rate": 0.0005, "loss": 1.2377903461456299, "step": 43892 }, { "epoch": 3.460228616476153, "grad_norm": 0.51385098695755, "learning_rate": 0.0005, "loss": 1.3320870399475098, "step": 43893 }, { "epoch": 3.460307449743792, "grad_norm": 0.5367034077644348, "learning_rate": 0.0005, "loss": 1.3847994804382324, "step": 43894 }, { "epoch": 3.460386283011431, "grad_norm": 0.5320606231689453, "learning_rate": 0.0005, "loss": 1.205899715423584, "step": 43895 }, { "epoch": 3.4604651162790696, "grad_norm": 0.5945872068405151, "learning_rate": 0.0005, "loss": 1.2767442464828491, "step": 43896 }, { "epoch": 3.4605439495467087, "grad_norm": 0.49000445008277893, "learning_rate": 0.0005, "loss": 1.1602305173873901, "step": 43897 }, { "epoch": 3.4606227828143474, "grad_norm": 0.5427080392837524, "learning_rate": 0.0005, "loss": 1.2627055644989014, "step": 43898 }, { "epoch": 3.4607016160819866, "grad_norm": 0.4902276396751404, "learning_rate": 0.0005, "loss": 1.2512567043304443, "step": 43899 }, { "epoch": 3.4607804493496257, "grad_norm": 0.5277472734451294, "learning_rate": 0.0005, "loss": 1.1619963645935059, "step": 43900 }, { "epoch": 3.4608592826172644, "grad_norm": 0.443086177110672, "learning_rate": 0.0005, "loss": 1.258399486541748, "step": 43901 }, { "epoch": 3.4609381158849035, "grad_norm": 0.4236125349998474, "learning_rate": 0.0005, "loss": 1.2001093626022339, "step": 43902 }, { "epoch": 3.461016949152542, "grad_norm": 0.5368687510490417, "learning_rate": 0.0005, "loss": 1.257911205291748, "step": 43903 }, { "epoch": 3.4610957824201813, "grad_norm": 0.5263621807098389, "learning_rate": 0.0005, "loss": 1.2356863021850586, "step": 43904 }, { "epoch": 3.4611746156878205, "grad_norm": 0.4442151188850403, "learning_rate": 0.0005, "loss": 1.2823565006256104, "step": 43905 }, { "epoch": 3.461253448955459, "grad_norm": 0.5330576300621033, "learning_rate": 0.0005, "loss": 1.340409517288208, "step": 43906 }, { "epoch": 3.4613322822230983, "grad_norm": 0.5366117358207703, "learning_rate": 0.0005, "loss": 1.3642494678497314, "step": 43907 }, { "epoch": 3.461411115490737, "grad_norm": 0.5105454325675964, "learning_rate": 0.0005, "loss": 1.3056867122650146, "step": 43908 }, { "epoch": 3.461489948758376, "grad_norm": 0.6413982510566711, "learning_rate": 0.0005, "loss": 1.3891257047653198, "step": 43909 }, { "epoch": 3.461568782026015, "grad_norm": 0.49596449732780457, "learning_rate": 0.0005, "loss": 1.4443371295928955, "step": 43910 }, { "epoch": 3.461647615293654, "grad_norm": 0.5247153043746948, "learning_rate": 0.0005, "loss": 1.236129641532898, "step": 43911 }, { "epoch": 3.4617264485612926, "grad_norm": 0.47697773575782776, "learning_rate": 0.0005, "loss": 1.2608283758163452, "step": 43912 }, { "epoch": 3.4618052818289318, "grad_norm": 0.47703275084495544, "learning_rate": 0.0005, "loss": 1.3644286394119263, "step": 43913 }, { "epoch": 3.461884115096571, "grad_norm": 0.4348028898239136, "learning_rate": 0.0005, "loss": 1.2486779689788818, "step": 43914 }, { "epoch": 3.4619629483642096, "grad_norm": 0.42607370018959045, "learning_rate": 0.0005, "loss": 1.1504452228546143, "step": 43915 }, { "epoch": 3.4620417816318487, "grad_norm": 0.5953100919723511, "learning_rate": 0.0005, "loss": 1.4271039962768555, "step": 43916 }, { "epoch": 3.4621206148994874, "grad_norm": 0.44427865743637085, "learning_rate": 0.0005, "loss": 1.22882080078125, "step": 43917 }, { "epoch": 3.4621994481671265, "grad_norm": 0.35929933190345764, "learning_rate": 0.0005, "loss": 1.096308708190918, "step": 43918 }, { "epoch": 3.4622782814347657, "grad_norm": 0.5032778382301331, "learning_rate": 0.0005, "loss": 1.3291963338851929, "step": 43919 }, { "epoch": 3.4623571147024044, "grad_norm": 0.669076681137085, "learning_rate": 0.0005, "loss": 1.141786813735962, "step": 43920 }, { "epoch": 3.4624359479700435, "grad_norm": 0.4680408537387848, "learning_rate": 0.0005, "loss": 1.125416874885559, "step": 43921 }, { "epoch": 3.462514781237682, "grad_norm": 0.49869927763938904, "learning_rate": 0.0005, "loss": 1.147314190864563, "step": 43922 }, { "epoch": 3.4625936145053213, "grad_norm": 0.486385703086853, "learning_rate": 0.0005, "loss": 1.3124291896820068, "step": 43923 }, { "epoch": 3.46267244777296, "grad_norm": 0.5739811658859253, "learning_rate": 0.0005, "loss": 1.357689619064331, "step": 43924 }, { "epoch": 3.462751281040599, "grad_norm": 0.5757973194122314, "learning_rate": 0.0005, "loss": 1.3722704648971558, "step": 43925 }, { "epoch": 3.462830114308238, "grad_norm": 0.403140664100647, "learning_rate": 0.0005, "loss": 1.137131929397583, "step": 43926 }, { "epoch": 3.462908947575877, "grad_norm": 0.37970253825187683, "learning_rate": 0.0005, "loss": 1.1779077053070068, "step": 43927 }, { "epoch": 3.462987780843516, "grad_norm": 0.49942150712013245, "learning_rate": 0.0005, "loss": 1.3538504838943481, "step": 43928 }, { "epoch": 3.463066614111155, "grad_norm": 0.4780535101890564, "learning_rate": 0.0005, "loss": 1.0930975675582886, "step": 43929 }, { "epoch": 3.463145447378794, "grad_norm": 0.5061644315719604, "learning_rate": 0.0005, "loss": 1.3240010738372803, "step": 43930 }, { "epoch": 3.4632242806464326, "grad_norm": 0.5259237885475159, "learning_rate": 0.0005, "loss": 1.2478153705596924, "step": 43931 }, { "epoch": 3.4633031139140718, "grad_norm": 0.6462864279747009, "learning_rate": 0.0005, "loss": 1.2293686866760254, "step": 43932 }, { "epoch": 3.463381947181711, "grad_norm": 0.5419095754623413, "learning_rate": 0.0005, "loss": 1.3790839910507202, "step": 43933 }, { "epoch": 3.4634607804493496, "grad_norm": 0.49330130219459534, "learning_rate": 0.0005, "loss": 1.1617932319641113, "step": 43934 }, { "epoch": 3.4635396137169887, "grad_norm": 0.6622524261474609, "learning_rate": 0.0005, "loss": 1.1378473043441772, "step": 43935 }, { "epoch": 3.4636184469846274, "grad_norm": 0.4726267457008362, "learning_rate": 0.0005, "loss": 1.0985242128372192, "step": 43936 }, { "epoch": 3.4636972802522665, "grad_norm": 0.5119475722312927, "learning_rate": 0.0005, "loss": 1.1942628622055054, "step": 43937 }, { "epoch": 3.463776113519905, "grad_norm": 0.6262466311454773, "learning_rate": 0.0005, "loss": 1.341771125793457, "step": 43938 }, { "epoch": 3.4638549467875444, "grad_norm": 0.4384537637233734, "learning_rate": 0.0005, "loss": 1.0996921062469482, "step": 43939 }, { "epoch": 3.463933780055183, "grad_norm": 0.4924084544181824, "learning_rate": 0.0005, "loss": 1.2491345405578613, "step": 43940 }, { "epoch": 3.464012613322822, "grad_norm": 0.4136383831501007, "learning_rate": 0.0005, "loss": 1.069718599319458, "step": 43941 }, { "epoch": 3.4640914465904613, "grad_norm": 0.6870400905609131, "learning_rate": 0.0005, "loss": 1.2724494934082031, "step": 43942 }, { "epoch": 3.4641702798581, "grad_norm": 0.4545891582965851, "learning_rate": 0.0005, "loss": 1.148930549621582, "step": 43943 }, { "epoch": 3.464249113125739, "grad_norm": 0.5829932689666748, "learning_rate": 0.0005, "loss": 1.3620920181274414, "step": 43944 }, { "epoch": 3.464327946393378, "grad_norm": 0.4988032579421997, "learning_rate": 0.0005, "loss": 1.0620572566986084, "step": 43945 }, { "epoch": 3.464406779661017, "grad_norm": 0.5527672171592712, "learning_rate": 0.0005, "loss": 1.3901904821395874, "step": 43946 }, { "epoch": 3.464485612928656, "grad_norm": 0.5128871202468872, "learning_rate": 0.0005, "loss": 1.1498901844024658, "step": 43947 }, { "epoch": 3.464564446196295, "grad_norm": 0.5610076189041138, "learning_rate": 0.0005, "loss": 1.4927340745925903, "step": 43948 }, { "epoch": 3.464643279463934, "grad_norm": 0.43839433789253235, "learning_rate": 0.0005, "loss": 1.118835687637329, "step": 43949 }, { "epoch": 3.4647221127315726, "grad_norm": 0.45657792687416077, "learning_rate": 0.0005, "loss": 1.1197965145111084, "step": 43950 }, { "epoch": 3.4648009459992117, "grad_norm": 0.5982752442359924, "learning_rate": 0.0005, "loss": 1.3033950328826904, "step": 43951 }, { "epoch": 3.4648797792668504, "grad_norm": 0.5698488354682922, "learning_rate": 0.0005, "loss": 1.367027759552002, "step": 43952 }, { "epoch": 3.4649586125344896, "grad_norm": 0.45434659719467163, "learning_rate": 0.0005, "loss": 1.1853641271591187, "step": 43953 }, { "epoch": 3.4650374458021287, "grad_norm": 0.4327104389667511, "learning_rate": 0.0005, "loss": 1.119166374206543, "step": 43954 }, { "epoch": 3.4651162790697674, "grad_norm": 0.5209916830062866, "learning_rate": 0.0005, "loss": 1.1550207138061523, "step": 43955 }, { "epoch": 3.4651951123374065, "grad_norm": 0.496111124753952, "learning_rate": 0.0005, "loss": 1.1458806991577148, "step": 43956 }, { "epoch": 3.465273945605045, "grad_norm": 0.4431871175765991, "learning_rate": 0.0005, "loss": 1.271702527999878, "step": 43957 }, { "epoch": 3.4653527788726843, "grad_norm": 0.5690943598747253, "learning_rate": 0.0005, "loss": 1.3319413661956787, "step": 43958 }, { "epoch": 3.465431612140323, "grad_norm": 0.512492835521698, "learning_rate": 0.0005, "loss": 1.3434674739837646, "step": 43959 }, { "epoch": 3.465510445407962, "grad_norm": 0.4406174421310425, "learning_rate": 0.0005, "loss": 1.195059061050415, "step": 43960 }, { "epoch": 3.4655892786756013, "grad_norm": 0.42350247502326965, "learning_rate": 0.0005, "loss": 1.3331563472747803, "step": 43961 }, { "epoch": 3.46566811194324, "grad_norm": 0.4642043113708496, "learning_rate": 0.0005, "loss": 1.2641077041625977, "step": 43962 }, { "epoch": 3.465746945210879, "grad_norm": 0.444421648979187, "learning_rate": 0.0005, "loss": 1.1103224754333496, "step": 43963 }, { "epoch": 3.465825778478518, "grad_norm": 0.622886598110199, "learning_rate": 0.0005, "loss": 1.3496389389038086, "step": 43964 }, { "epoch": 3.465904611746157, "grad_norm": 0.39309027791023254, "learning_rate": 0.0005, "loss": 1.1531336307525635, "step": 43965 }, { "epoch": 3.465983445013796, "grad_norm": 0.8090689778327942, "learning_rate": 0.0005, "loss": 1.5796966552734375, "step": 43966 }, { "epoch": 3.4660622782814348, "grad_norm": 0.5028672218322754, "learning_rate": 0.0005, "loss": 1.2361626625061035, "step": 43967 }, { "epoch": 3.466141111549074, "grad_norm": 0.5562840104103088, "learning_rate": 0.0005, "loss": 1.2977571487426758, "step": 43968 }, { "epoch": 3.4662199448167126, "grad_norm": 0.5315565466880798, "learning_rate": 0.0005, "loss": 1.3067471981048584, "step": 43969 }, { "epoch": 3.4662987780843517, "grad_norm": 0.6612315773963928, "learning_rate": 0.0005, "loss": 1.247093915939331, "step": 43970 }, { "epoch": 3.4663776113519904, "grad_norm": 0.5640851855278015, "learning_rate": 0.0005, "loss": 1.3786439895629883, "step": 43971 }, { "epoch": 3.4664564446196295, "grad_norm": 0.4864377975463867, "learning_rate": 0.0005, "loss": 1.2958333492279053, "step": 43972 }, { "epoch": 3.4665352778872682, "grad_norm": 0.6756805777549744, "learning_rate": 0.0005, "loss": 1.306325078010559, "step": 43973 }, { "epoch": 3.4666141111549074, "grad_norm": 0.5838222503662109, "learning_rate": 0.0005, "loss": 1.1768214702606201, "step": 43974 }, { "epoch": 3.4666929444225465, "grad_norm": 0.49623560905456543, "learning_rate": 0.0005, "loss": 1.144906759262085, "step": 43975 }, { "epoch": 3.466771777690185, "grad_norm": 0.40094703435897827, "learning_rate": 0.0005, "loss": 1.1607939004898071, "step": 43976 }, { "epoch": 3.4668506109578243, "grad_norm": 0.5368673205375671, "learning_rate": 0.0005, "loss": 1.5211732387542725, "step": 43977 }, { "epoch": 3.466929444225463, "grad_norm": 0.44251009821891785, "learning_rate": 0.0005, "loss": 1.038028359413147, "step": 43978 }, { "epoch": 3.467008277493102, "grad_norm": 0.5486521124839783, "learning_rate": 0.0005, "loss": 1.4807547330856323, "step": 43979 }, { "epoch": 3.4670871107607413, "grad_norm": 0.4802781939506531, "learning_rate": 0.0005, "loss": 1.2820615768432617, "step": 43980 }, { "epoch": 3.46716594402838, "grad_norm": 0.48795899748802185, "learning_rate": 0.0005, "loss": 1.2882614135742188, "step": 43981 }, { "epoch": 3.467244777296019, "grad_norm": 0.48269981145858765, "learning_rate": 0.0005, "loss": 1.3460144996643066, "step": 43982 }, { "epoch": 3.467323610563658, "grad_norm": 0.664647102355957, "learning_rate": 0.0005, "loss": 1.3342759609222412, "step": 43983 }, { "epoch": 3.467402443831297, "grad_norm": 0.46685791015625, "learning_rate": 0.0005, "loss": 1.2311437129974365, "step": 43984 }, { "epoch": 3.4674812770989356, "grad_norm": 0.5266059637069702, "learning_rate": 0.0005, "loss": 1.3220328092575073, "step": 43985 }, { "epoch": 3.4675601103665747, "grad_norm": 0.5994249582290649, "learning_rate": 0.0005, "loss": 1.3314414024353027, "step": 43986 }, { "epoch": 3.4676389436342134, "grad_norm": 0.5073498487472534, "learning_rate": 0.0005, "loss": 1.3594627380371094, "step": 43987 }, { "epoch": 3.4677177769018526, "grad_norm": 0.4605572521686554, "learning_rate": 0.0005, "loss": 1.3305912017822266, "step": 43988 }, { "epoch": 3.4677966101694917, "grad_norm": 0.48662978410720825, "learning_rate": 0.0005, "loss": 1.3099446296691895, "step": 43989 }, { "epoch": 3.4678754434371304, "grad_norm": 0.5823535919189453, "learning_rate": 0.0005, "loss": 1.225848913192749, "step": 43990 }, { "epoch": 3.4679542767047695, "grad_norm": 0.5542646050453186, "learning_rate": 0.0005, "loss": 1.2806416749954224, "step": 43991 }, { "epoch": 3.468033109972408, "grad_norm": 0.4956663250923157, "learning_rate": 0.0005, "loss": 1.3450230360031128, "step": 43992 }, { "epoch": 3.4681119432400473, "grad_norm": 0.4854719340801239, "learning_rate": 0.0005, "loss": 1.417221188545227, "step": 43993 }, { "epoch": 3.4681907765076865, "grad_norm": 0.43526169657707214, "learning_rate": 0.0005, "loss": 1.0720577239990234, "step": 43994 }, { "epoch": 3.468269609775325, "grad_norm": 0.5088680386543274, "learning_rate": 0.0005, "loss": 1.2202495336532593, "step": 43995 }, { "epoch": 3.4683484430429643, "grad_norm": 0.4479180872440338, "learning_rate": 0.0005, "loss": 1.2505358457565308, "step": 43996 }, { "epoch": 3.468427276310603, "grad_norm": 0.5383909344673157, "learning_rate": 0.0005, "loss": 1.3234950304031372, "step": 43997 }, { "epoch": 3.468506109578242, "grad_norm": 0.4410642385482788, "learning_rate": 0.0005, "loss": 1.2775278091430664, "step": 43998 }, { "epoch": 3.468584942845881, "grad_norm": 0.4969271123409271, "learning_rate": 0.0005, "loss": 1.1872613430023193, "step": 43999 }, { "epoch": 3.46866377611352, "grad_norm": 0.3988886773586273, "learning_rate": 0.0005, "loss": 1.3860979080200195, "step": 44000 }, { "epoch": 3.4687426093811586, "grad_norm": 0.5287735462188721, "learning_rate": 0.0005, "loss": 1.272078514099121, "step": 44001 }, { "epoch": 3.4688214426487978, "grad_norm": 0.5119079351425171, "learning_rate": 0.0005, "loss": 1.3146252632141113, "step": 44002 }, { "epoch": 3.468900275916437, "grad_norm": 0.5275547504425049, "learning_rate": 0.0005, "loss": 1.2991132736206055, "step": 44003 }, { "epoch": 3.4689791091840756, "grad_norm": 0.5459595322608948, "learning_rate": 0.0005, "loss": 1.2425942420959473, "step": 44004 }, { "epoch": 3.4690579424517147, "grad_norm": 0.6758235096931458, "learning_rate": 0.0005, "loss": 1.1185646057128906, "step": 44005 }, { "epoch": 3.4691367757193534, "grad_norm": 0.4355527460575104, "learning_rate": 0.0005, "loss": 1.177842378616333, "step": 44006 }, { "epoch": 3.4692156089869925, "grad_norm": 0.47436749935150146, "learning_rate": 0.0005, "loss": 1.1542251110076904, "step": 44007 }, { "epoch": 3.4692944422546317, "grad_norm": 0.48449233174324036, "learning_rate": 0.0005, "loss": 1.3455417156219482, "step": 44008 }, { "epoch": 3.4693732755222704, "grad_norm": 0.5746595859527588, "learning_rate": 0.0005, "loss": 1.1710851192474365, "step": 44009 }, { "epoch": 3.4694521087899095, "grad_norm": 0.5564274191856384, "learning_rate": 0.0005, "loss": 1.5457395315170288, "step": 44010 }, { "epoch": 3.469530942057548, "grad_norm": 0.8053482174873352, "learning_rate": 0.0005, "loss": 1.2352149486541748, "step": 44011 }, { "epoch": 3.4696097753251873, "grad_norm": 0.4956526756286621, "learning_rate": 0.0005, "loss": 1.266063928604126, "step": 44012 }, { "epoch": 3.469688608592826, "grad_norm": 0.4440556764602661, "learning_rate": 0.0005, "loss": 1.2310552597045898, "step": 44013 }, { "epoch": 3.469767441860465, "grad_norm": 0.600740909576416, "learning_rate": 0.0005, "loss": 1.4400579929351807, "step": 44014 }, { "epoch": 3.469846275128104, "grad_norm": 0.4519367516040802, "learning_rate": 0.0005, "loss": 1.039583683013916, "step": 44015 }, { "epoch": 3.469925108395743, "grad_norm": 0.5391834378242493, "learning_rate": 0.0005, "loss": 1.4197027683258057, "step": 44016 }, { "epoch": 3.470003941663382, "grad_norm": 0.6911340355873108, "learning_rate": 0.0005, "loss": 1.445645809173584, "step": 44017 }, { "epoch": 3.470082774931021, "grad_norm": 0.4752434194087982, "learning_rate": 0.0005, "loss": 1.2130213975906372, "step": 44018 }, { "epoch": 3.47016160819866, "grad_norm": 0.47661975026130676, "learning_rate": 0.0005, "loss": 1.217000961303711, "step": 44019 }, { "epoch": 3.4702404414662986, "grad_norm": 0.4411703050136566, "learning_rate": 0.0005, "loss": 1.342348337173462, "step": 44020 }, { "epoch": 3.4703192747339378, "grad_norm": 0.5248069167137146, "learning_rate": 0.0005, "loss": 1.3817155361175537, "step": 44021 }, { "epoch": 3.470398108001577, "grad_norm": 0.5133796334266663, "learning_rate": 0.0005, "loss": 1.2632101774215698, "step": 44022 }, { "epoch": 3.4704769412692156, "grad_norm": 0.4555703401565552, "learning_rate": 0.0005, "loss": 1.1687195301055908, "step": 44023 }, { "epoch": 3.4705557745368547, "grad_norm": 0.4415247142314911, "learning_rate": 0.0005, "loss": 1.150731086730957, "step": 44024 }, { "epoch": 3.4706346078044934, "grad_norm": 0.4978642463684082, "learning_rate": 0.0005, "loss": 1.3851842880249023, "step": 44025 }, { "epoch": 3.4707134410721325, "grad_norm": 0.5014347434043884, "learning_rate": 0.0005, "loss": 1.2340232133865356, "step": 44026 }, { "epoch": 3.470792274339771, "grad_norm": 0.5330682992935181, "learning_rate": 0.0005, "loss": 1.3308866024017334, "step": 44027 }, { "epoch": 3.4708711076074104, "grad_norm": 0.41249939799308777, "learning_rate": 0.0005, "loss": 1.0528514385223389, "step": 44028 }, { "epoch": 3.470949940875049, "grad_norm": 0.4190421402454376, "learning_rate": 0.0005, "loss": 1.1662286520004272, "step": 44029 }, { "epoch": 3.471028774142688, "grad_norm": 0.5639238953590393, "learning_rate": 0.0005, "loss": 1.35065495967865, "step": 44030 }, { "epoch": 3.4711076074103273, "grad_norm": 0.5359704494476318, "learning_rate": 0.0005, "loss": 1.3308907747268677, "step": 44031 }, { "epoch": 3.471186440677966, "grad_norm": 0.5079149007797241, "learning_rate": 0.0005, "loss": 1.3855242729187012, "step": 44032 }, { "epoch": 3.471265273945605, "grad_norm": 0.6276233196258545, "learning_rate": 0.0005, "loss": 1.215773105621338, "step": 44033 }, { "epoch": 3.471344107213244, "grad_norm": 0.5163276195526123, "learning_rate": 0.0005, "loss": 1.3431246280670166, "step": 44034 }, { "epoch": 3.471422940480883, "grad_norm": 0.4503610134124756, "learning_rate": 0.0005, "loss": 1.3125622272491455, "step": 44035 }, { "epoch": 3.471501773748522, "grad_norm": 0.6504111886024475, "learning_rate": 0.0005, "loss": 1.2801671028137207, "step": 44036 }, { "epoch": 3.471580607016161, "grad_norm": 0.5795455574989319, "learning_rate": 0.0005, "loss": 1.195300579071045, "step": 44037 }, { "epoch": 3.4716594402838, "grad_norm": 0.4528736174106598, "learning_rate": 0.0005, "loss": 1.3095650672912598, "step": 44038 }, { "epoch": 3.4717382735514386, "grad_norm": 0.37509864568710327, "learning_rate": 0.0005, "loss": 1.1852598190307617, "step": 44039 }, { "epoch": 3.4718171068190777, "grad_norm": 0.43858444690704346, "learning_rate": 0.0005, "loss": 1.3384097814559937, "step": 44040 }, { "epoch": 3.4718959400867164, "grad_norm": 0.6223618984222412, "learning_rate": 0.0005, "loss": 1.4929120540618896, "step": 44041 }, { "epoch": 3.4719747733543556, "grad_norm": 0.5261398553848267, "learning_rate": 0.0005, "loss": 1.2276995182037354, "step": 44042 }, { "epoch": 3.4720536066219942, "grad_norm": 0.45127367973327637, "learning_rate": 0.0005, "loss": 1.310782790184021, "step": 44043 }, { "epoch": 3.4721324398896334, "grad_norm": 0.5011959671974182, "learning_rate": 0.0005, "loss": 1.3927810192108154, "step": 44044 }, { "epoch": 3.4722112731572725, "grad_norm": 0.4995560944080353, "learning_rate": 0.0005, "loss": 1.2905683517456055, "step": 44045 }, { "epoch": 3.472290106424911, "grad_norm": 0.4844280481338501, "learning_rate": 0.0005, "loss": 1.17847740650177, "step": 44046 }, { "epoch": 3.4723689396925503, "grad_norm": 0.4656869173049927, "learning_rate": 0.0005, "loss": 1.113938808441162, "step": 44047 }, { "epoch": 3.472447772960189, "grad_norm": 0.5105394124984741, "learning_rate": 0.0005, "loss": 1.18971848487854, "step": 44048 }, { "epoch": 3.472526606227828, "grad_norm": 0.4694296419620514, "learning_rate": 0.0005, "loss": 1.4619348049163818, "step": 44049 }, { "epoch": 3.4726054394954673, "grad_norm": 0.4769342541694641, "learning_rate": 0.0005, "loss": 1.1726789474487305, "step": 44050 }, { "epoch": 3.472684272763106, "grad_norm": 0.5236493349075317, "learning_rate": 0.0005, "loss": 1.3360371589660645, "step": 44051 }, { "epoch": 3.472763106030745, "grad_norm": 0.4836748242378235, "learning_rate": 0.0005, "loss": 1.2848435640335083, "step": 44052 }, { "epoch": 3.472841939298384, "grad_norm": 0.5144875645637512, "learning_rate": 0.0005, "loss": 1.453824758529663, "step": 44053 }, { "epoch": 3.472920772566023, "grad_norm": 0.49655601382255554, "learning_rate": 0.0005, "loss": 1.2439768314361572, "step": 44054 }, { "epoch": 3.4729996058336616, "grad_norm": 0.48482853174209595, "learning_rate": 0.0005, "loss": 1.3510994911193848, "step": 44055 }, { "epoch": 3.4730784391013008, "grad_norm": 0.6108825206756592, "learning_rate": 0.0005, "loss": 1.3317985534667969, "step": 44056 }, { "epoch": 3.4731572723689395, "grad_norm": 0.4008643627166748, "learning_rate": 0.0005, "loss": 0.9995501041412354, "step": 44057 }, { "epoch": 3.4732361056365786, "grad_norm": 0.5559929609298706, "learning_rate": 0.0005, "loss": 1.2476403713226318, "step": 44058 }, { "epoch": 3.4733149389042177, "grad_norm": 0.5369976758956909, "learning_rate": 0.0005, "loss": 1.2566184997558594, "step": 44059 }, { "epoch": 3.4733937721718564, "grad_norm": 0.530674397945404, "learning_rate": 0.0005, "loss": 1.4068918228149414, "step": 44060 }, { "epoch": 3.4734726054394955, "grad_norm": 0.5389852523803711, "learning_rate": 0.0005, "loss": 1.2617820501327515, "step": 44061 }, { "epoch": 3.4735514387071342, "grad_norm": 0.39732417464256287, "learning_rate": 0.0005, "loss": 1.184638261795044, "step": 44062 }, { "epoch": 3.4736302719747734, "grad_norm": 0.5264238119125366, "learning_rate": 0.0005, "loss": 1.5337203741073608, "step": 44063 }, { "epoch": 3.4737091052424125, "grad_norm": 0.48485010862350464, "learning_rate": 0.0005, "loss": 1.2159721851348877, "step": 44064 }, { "epoch": 3.473787938510051, "grad_norm": 0.4620295464992523, "learning_rate": 0.0005, "loss": 1.3475340604782104, "step": 44065 }, { "epoch": 3.4738667717776903, "grad_norm": 0.4540955126285553, "learning_rate": 0.0005, "loss": 1.45603609085083, "step": 44066 }, { "epoch": 3.473945605045329, "grad_norm": 0.49516135454177856, "learning_rate": 0.0005, "loss": 1.0938975811004639, "step": 44067 }, { "epoch": 3.474024438312968, "grad_norm": 0.6175971031188965, "learning_rate": 0.0005, "loss": 1.2508618831634521, "step": 44068 }, { "epoch": 3.474103271580607, "grad_norm": 0.47671082615852356, "learning_rate": 0.0005, "loss": 1.3237318992614746, "step": 44069 }, { "epoch": 3.474182104848246, "grad_norm": 0.47017771005630493, "learning_rate": 0.0005, "loss": 1.2886714935302734, "step": 44070 }, { "epoch": 3.4742609381158847, "grad_norm": 0.46370595693588257, "learning_rate": 0.0005, "loss": 1.082580327987671, "step": 44071 }, { "epoch": 3.474339771383524, "grad_norm": 0.5519661903381348, "learning_rate": 0.0005, "loss": 1.4489712715148926, "step": 44072 }, { "epoch": 3.474418604651163, "grad_norm": 0.3746091425418854, "learning_rate": 0.0005, "loss": 1.0410430431365967, "step": 44073 }, { "epoch": 3.4744974379188016, "grad_norm": 0.702418327331543, "learning_rate": 0.0005, "loss": 1.1912024021148682, "step": 44074 }, { "epoch": 3.4745762711864407, "grad_norm": 0.6293454170227051, "learning_rate": 0.0005, "loss": 1.3892261981964111, "step": 44075 }, { "epoch": 3.4746551044540794, "grad_norm": 0.4499078392982483, "learning_rate": 0.0005, "loss": 1.2354427576065063, "step": 44076 }, { "epoch": 3.4747339377217186, "grad_norm": 0.4819203317165375, "learning_rate": 0.0005, "loss": 1.423983097076416, "step": 44077 }, { "epoch": 3.4748127709893577, "grad_norm": 0.45233625173568726, "learning_rate": 0.0005, "loss": 1.0940790176391602, "step": 44078 }, { "epoch": 3.4748916042569964, "grad_norm": 0.5886359810829163, "learning_rate": 0.0005, "loss": 1.3040831089019775, "step": 44079 }, { "epoch": 3.4749704375246355, "grad_norm": 0.5462557673454285, "learning_rate": 0.0005, "loss": 1.3896955251693726, "step": 44080 }, { "epoch": 3.475049270792274, "grad_norm": 0.8299410939216614, "learning_rate": 0.0005, "loss": 1.2955245971679688, "step": 44081 }, { "epoch": 3.4751281040599133, "grad_norm": 0.48111921548843384, "learning_rate": 0.0005, "loss": 1.2645598649978638, "step": 44082 }, { "epoch": 3.475206937327552, "grad_norm": 0.5382553935050964, "learning_rate": 0.0005, "loss": 1.419614553451538, "step": 44083 }, { "epoch": 3.475285770595191, "grad_norm": 0.5240707993507385, "learning_rate": 0.0005, "loss": 1.3516106605529785, "step": 44084 }, { "epoch": 3.4753646038628303, "grad_norm": 0.6060739755630493, "learning_rate": 0.0005, "loss": 1.3593690395355225, "step": 44085 }, { "epoch": 3.475443437130469, "grad_norm": 0.4497324824333191, "learning_rate": 0.0005, "loss": 1.2915323972702026, "step": 44086 }, { "epoch": 3.475522270398108, "grad_norm": 0.4960666298866272, "learning_rate": 0.0005, "loss": 1.3223243951797485, "step": 44087 }, { "epoch": 3.475601103665747, "grad_norm": 0.5429546236991882, "learning_rate": 0.0005, "loss": 1.3192944526672363, "step": 44088 }, { "epoch": 3.475679936933386, "grad_norm": 0.5825145244598389, "learning_rate": 0.0005, "loss": 1.3193421363830566, "step": 44089 }, { "epoch": 3.4757587702010246, "grad_norm": 0.5082308053970337, "learning_rate": 0.0005, "loss": 1.1704282760620117, "step": 44090 }, { "epoch": 3.4758376034686638, "grad_norm": 0.4068968594074249, "learning_rate": 0.0005, "loss": 1.1272289752960205, "step": 44091 }, { "epoch": 3.475916436736303, "grad_norm": 0.43876707553863525, "learning_rate": 0.0005, "loss": 1.2814879417419434, "step": 44092 }, { "epoch": 3.4759952700039416, "grad_norm": 0.46115878224372864, "learning_rate": 0.0005, "loss": 1.1800456047058105, "step": 44093 }, { "epoch": 3.4760741032715807, "grad_norm": 0.5586817860603333, "learning_rate": 0.0005, "loss": 1.3185803890228271, "step": 44094 }, { "epoch": 3.4761529365392194, "grad_norm": 0.42605820298194885, "learning_rate": 0.0005, "loss": 1.2864511013031006, "step": 44095 }, { "epoch": 3.4762317698068586, "grad_norm": 0.6113370060920715, "learning_rate": 0.0005, "loss": 1.392266869544983, "step": 44096 }, { "epoch": 3.4763106030744977, "grad_norm": 0.4466613531112671, "learning_rate": 0.0005, "loss": 1.1648329496383667, "step": 44097 }, { "epoch": 3.4763894363421364, "grad_norm": 0.5623279213905334, "learning_rate": 0.0005, "loss": 1.4868216514587402, "step": 44098 }, { "epoch": 3.4764682696097755, "grad_norm": 0.5326970815658569, "learning_rate": 0.0005, "loss": 1.2175087928771973, "step": 44099 }, { "epoch": 3.476547102877414, "grad_norm": 0.5224456787109375, "learning_rate": 0.0005, "loss": 1.32743239402771, "step": 44100 }, { "epoch": 3.4766259361450533, "grad_norm": 0.6269277334213257, "learning_rate": 0.0005, "loss": 1.3023799657821655, "step": 44101 }, { "epoch": 3.476704769412692, "grad_norm": 0.4526948928833008, "learning_rate": 0.0005, "loss": 1.3266409635543823, "step": 44102 }, { "epoch": 3.476783602680331, "grad_norm": 0.45473581552505493, "learning_rate": 0.0005, "loss": 1.1558948755264282, "step": 44103 }, { "epoch": 3.47686243594797, "grad_norm": 0.4788474142551422, "learning_rate": 0.0005, "loss": 1.2640838623046875, "step": 44104 }, { "epoch": 3.476941269215609, "grad_norm": 0.4355742335319519, "learning_rate": 0.0005, "loss": 1.2391531467437744, "step": 44105 }, { "epoch": 3.477020102483248, "grad_norm": 0.5609140992164612, "learning_rate": 0.0005, "loss": 1.265041470527649, "step": 44106 }, { "epoch": 3.477098935750887, "grad_norm": 0.4848872721195221, "learning_rate": 0.0005, "loss": 1.242489218711853, "step": 44107 }, { "epoch": 3.477177769018526, "grad_norm": 0.43298301100730896, "learning_rate": 0.0005, "loss": 1.2461204528808594, "step": 44108 }, { "epoch": 3.4772566022861646, "grad_norm": 0.4803328812122345, "learning_rate": 0.0005, "loss": 1.446925401687622, "step": 44109 }, { "epoch": 3.4773354355538038, "grad_norm": 0.45378580689430237, "learning_rate": 0.0005, "loss": 1.2911241054534912, "step": 44110 }, { "epoch": 3.477414268821443, "grad_norm": 0.5404772162437439, "learning_rate": 0.0005, "loss": 1.3249917030334473, "step": 44111 }, { "epoch": 3.4774931020890816, "grad_norm": 0.5489528775215149, "learning_rate": 0.0005, "loss": 1.317474365234375, "step": 44112 }, { "epoch": 3.4775719353567207, "grad_norm": 0.46042782068252563, "learning_rate": 0.0005, "loss": 1.2432563304901123, "step": 44113 }, { "epoch": 3.4776507686243594, "grad_norm": 0.4578711688518524, "learning_rate": 0.0005, "loss": 1.290863275527954, "step": 44114 }, { "epoch": 3.4777296018919985, "grad_norm": 0.4517608880996704, "learning_rate": 0.0005, "loss": 1.1686880588531494, "step": 44115 }, { "epoch": 3.4778084351596372, "grad_norm": 0.5365732908248901, "learning_rate": 0.0005, "loss": 1.4044889211654663, "step": 44116 }, { "epoch": 3.4778872684272764, "grad_norm": 0.4656599760055542, "learning_rate": 0.0005, "loss": 1.3231778144836426, "step": 44117 }, { "epoch": 3.477966101694915, "grad_norm": 0.5145534873008728, "learning_rate": 0.0005, "loss": 1.311644196510315, "step": 44118 }, { "epoch": 3.478044934962554, "grad_norm": 0.5471598505973816, "learning_rate": 0.0005, "loss": 1.3623250722885132, "step": 44119 }, { "epoch": 3.4781237682301933, "grad_norm": 0.4364493489265442, "learning_rate": 0.0005, "loss": 1.1251680850982666, "step": 44120 }, { "epoch": 3.478202601497832, "grad_norm": 0.6282127499580383, "learning_rate": 0.0005, "loss": 1.2782108783721924, "step": 44121 }, { "epoch": 3.478281434765471, "grad_norm": 0.523580014705658, "learning_rate": 0.0005, "loss": 1.4941614866256714, "step": 44122 }, { "epoch": 3.47836026803311, "grad_norm": 0.397883802652359, "learning_rate": 0.0005, "loss": 1.0965909957885742, "step": 44123 }, { "epoch": 3.478439101300749, "grad_norm": 0.6065942645072937, "learning_rate": 0.0005, "loss": 1.2645164728164673, "step": 44124 }, { "epoch": 3.478517934568388, "grad_norm": 0.4666815400123596, "learning_rate": 0.0005, "loss": 1.4036533832550049, "step": 44125 }, { "epoch": 3.478596767836027, "grad_norm": 0.44726496934890747, "learning_rate": 0.0005, "loss": 1.3124090433120728, "step": 44126 }, { "epoch": 3.478675601103666, "grad_norm": 0.563503623008728, "learning_rate": 0.0005, "loss": 1.3129054307937622, "step": 44127 }, { "epoch": 3.4787544343713046, "grad_norm": 0.4704107344150543, "learning_rate": 0.0005, "loss": 1.1111955642700195, "step": 44128 }, { "epoch": 3.4788332676389437, "grad_norm": 0.630415141582489, "learning_rate": 0.0005, "loss": 1.2807878255844116, "step": 44129 }, { "epoch": 3.4789121009065824, "grad_norm": 0.5784383416175842, "learning_rate": 0.0005, "loss": 1.522167682647705, "step": 44130 }, { "epoch": 3.4789909341742216, "grad_norm": 0.5047422051429749, "learning_rate": 0.0005, "loss": 1.4106236696243286, "step": 44131 }, { "epoch": 3.4790697674418603, "grad_norm": 0.5463342666625977, "learning_rate": 0.0005, "loss": 1.2827214002609253, "step": 44132 }, { "epoch": 3.4791486007094994, "grad_norm": 0.5304962992668152, "learning_rate": 0.0005, "loss": 1.3216049671173096, "step": 44133 }, { "epoch": 3.4792274339771385, "grad_norm": 0.4344012439250946, "learning_rate": 0.0005, "loss": 1.2115564346313477, "step": 44134 }, { "epoch": 3.479306267244777, "grad_norm": 0.46458327770233154, "learning_rate": 0.0005, "loss": 1.2607715129852295, "step": 44135 }, { "epoch": 3.4793851005124163, "grad_norm": 0.4543386697769165, "learning_rate": 0.0005, "loss": 1.2103549242019653, "step": 44136 }, { "epoch": 3.479463933780055, "grad_norm": 1.8061360120773315, "learning_rate": 0.0005, "loss": 1.4147478342056274, "step": 44137 }, { "epoch": 3.479542767047694, "grad_norm": 0.522785484790802, "learning_rate": 0.0005, "loss": 1.2997684478759766, "step": 44138 }, { "epoch": 3.4796216003153333, "grad_norm": 0.49537304043769836, "learning_rate": 0.0005, "loss": 1.3485169410705566, "step": 44139 }, { "epoch": 3.479700433582972, "grad_norm": 0.46014490723609924, "learning_rate": 0.0005, "loss": 1.441424012184143, "step": 44140 }, { "epoch": 3.479779266850611, "grad_norm": 0.6315468549728394, "learning_rate": 0.0005, "loss": 1.3746685981750488, "step": 44141 }, { "epoch": 3.47985810011825, "grad_norm": 0.4743790924549103, "learning_rate": 0.0005, "loss": 1.3789998292922974, "step": 44142 }, { "epoch": 3.479936933385889, "grad_norm": 0.4782256484031677, "learning_rate": 0.0005, "loss": 1.2605664730072021, "step": 44143 }, { "epoch": 3.4800157666535276, "grad_norm": 2.820340156555176, "learning_rate": 0.0005, "loss": 1.1183825731277466, "step": 44144 }, { "epoch": 3.4800945999211668, "grad_norm": 0.49852702021598816, "learning_rate": 0.0005, "loss": 1.4005804061889648, "step": 44145 }, { "epoch": 3.4801734331888055, "grad_norm": 0.5189117789268494, "learning_rate": 0.0005, "loss": 1.3414053916931152, "step": 44146 }, { "epoch": 3.4802522664564446, "grad_norm": 0.46385645866394043, "learning_rate": 0.0005, "loss": 1.242016315460205, "step": 44147 }, { "epoch": 3.4803310997240837, "grad_norm": 0.4932626783847809, "learning_rate": 0.0005, "loss": 1.1423752307891846, "step": 44148 }, { "epoch": 3.4804099329917224, "grad_norm": 0.4682006537914276, "learning_rate": 0.0005, "loss": 1.2320804595947266, "step": 44149 }, { "epoch": 3.4804887662593615, "grad_norm": 0.49508535861968994, "learning_rate": 0.0005, "loss": 1.485501766204834, "step": 44150 }, { "epoch": 3.4805675995270002, "grad_norm": 0.424511581659317, "learning_rate": 0.0005, "loss": 1.2579963207244873, "step": 44151 }, { "epoch": 3.4806464327946394, "grad_norm": 0.43174228072166443, "learning_rate": 0.0005, "loss": 1.198180079460144, "step": 44152 }, { "epoch": 3.4807252660622785, "grad_norm": 0.47711604833602905, "learning_rate": 0.0005, "loss": 1.0958552360534668, "step": 44153 }, { "epoch": 3.480804099329917, "grad_norm": 0.5659869313240051, "learning_rate": 0.0005, "loss": 1.3722879886627197, "step": 44154 }, { "epoch": 3.4808829325975563, "grad_norm": 0.49582549929618835, "learning_rate": 0.0005, "loss": 1.348372220993042, "step": 44155 }, { "epoch": 3.480961765865195, "grad_norm": 0.512190580368042, "learning_rate": 0.0005, "loss": 1.4702067375183105, "step": 44156 }, { "epoch": 3.481040599132834, "grad_norm": 0.5121526718139648, "learning_rate": 0.0005, "loss": 1.3173915147781372, "step": 44157 }, { "epoch": 3.481119432400473, "grad_norm": 0.46691641211509705, "learning_rate": 0.0005, "loss": 1.1460583209991455, "step": 44158 }, { "epoch": 3.481198265668112, "grad_norm": 0.39642533659935, "learning_rate": 0.0005, "loss": 1.2315689325332642, "step": 44159 }, { "epoch": 3.4812770989357507, "grad_norm": 0.45362555980682373, "learning_rate": 0.0005, "loss": 1.1532843112945557, "step": 44160 }, { "epoch": 3.48135593220339, "grad_norm": 0.49181288480758667, "learning_rate": 0.0005, "loss": 1.2192211151123047, "step": 44161 }, { "epoch": 3.481434765471029, "grad_norm": 0.5182312726974487, "learning_rate": 0.0005, "loss": 1.1404931545257568, "step": 44162 }, { "epoch": 3.4815135987386676, "grad_norm": 0.48589810729026794, "learning_rate": 0.0005, "loss": 1.2860608100891113, "step": 44163 }, { "epoch": 3.4815924320063067, "grad_norm": 0.4144526720046997, "learning_rate": 0.0005, "loss": 0.9649994373321533, "step": 44164 }, { "epoch": 3.4816712652739454, "grad_norm": 0.47370627522468567, "learning_rate": 0.0005, "loss": 1.2329957485198975, "step": 44165 }, { "epoch": 3.4817500985415846, "grad_norm": 0.4671262204647064, "learning_rate": 0.0005, "loss": 1.1800651550292969, "step": 44166 }, { "epoch": 3.4818289318092237, "grad_norm": 0.5203824639320374, "learning_rate": 0.0005, "loss": 1.3829408884048462, "step": 44167 }, { "epoch": 3.4819077650768624, "grad_norm": 0.487542986869812, "learning_rate": 0.0005, "loss": 1.3524763584136963, "step": 44168 }, { "epoch": 3.4819865983445015, "grad_norm": 0.46466875076293945, "learning_rate": 0.0005, "loss": 1.3903651237487793, "step": 44169 }, { "epoch": 3.48206543161214, "grad_norm": 0.5414243340492249, "learning_rate": 0.0005, "loss": 1.4786813259124756, "step": 44170 }, { "epoch": 3.4821442648797793, "grad_norm": 0.6949797868728638, "learning_rate": 0.0005, "loss": 1.2898770570755005, "step": 44171 }, { "epoch": 3.482223098147418, "grad_norm": 0.46518608927726746, "learning_rate": 0.0005, "loss": 1.403482437133789, "step": 44172 }, { "epoch": 3.482301931415057, "grad_norm": 0.44079381227493286, "learning_rate": 0.0005, "loss": 1.173985242843628, "step": 44173 }, { "epoch": 3.482380764682696, "grad_norm": 0.5581363439559937, "learning_rate": 0.0005, "loss": 1.3028771877288818, "step": 44174 }, { "epoch": 3.482459597950335, "grad_norm": 0.5756044387817383, "learning_rate": 0.0005, "loss": 1.3416104316711426, "step": 44175 }, { "epoch": 3.482538431217974, "grad_norm": 0.5332006812095642, "learning_rate": 0.0005, "loss": 1.2619434595108032, "step": 44176 }, { "epoch": 3.482617264485613, "grad_norm": 0.4451294541358948, "learning_rate": 0.0005, "loss": 1.1317873001098633, "step": 44177 }, { "epoch": 3.482696097753252, "grad_norm": 0.43406975269317627, "learning_rate": 0.0005, "loss": 1.1490556001663208, "step": 44178 }, { "epoch": 3.4827749310208906, "grad_norm": 0.4995991289615631, "learning_rate": 0.0005, "loss": 1.2088621854782104, "step": 44179 }, { "epoch": 3.4828537642885298, "grad_norm": 0.4707922041416168, "learning_rate": 0.0005, "loss": 1.207301378250122, "step": 44180 }, { "epoch": 3.482932597556169, "grad_norm": 0.5591850876808167, "learning_rate": 0.0005, "loss": 1.422544240951538, "step": 44181 }, { "epoch": 3.4830114308238076, "grad_norm": 0.5912048816680908, "learning_rate": 0.0005, "loss": 1.277963399887085, "step": 44182 }, { "epoch": 3.4830902640914467, "grad_norm": 0.45869776606559753, "learning_rate": 0.0005, "loss": 1.121181845664978, "step": 44183 }, { "epoch": 3.4831690973590854, "grad_norm": 0.7519627213478088, "learning_rate": 0.0005, "loss": 1.3873114585876465, "step": 44184 }, { "epoch": 3.4832479306267246, "grad_norm": 0.5122584700584412, "learning_rate": 0.0005, "loss": 1.2368221282958984, "step": 44185 }, { "epoch": 3.4833267638943632, "grad_norm": 0.49374815821647644, "learning_rate": 0.0005, "loss": 1.300377368927002, "step": 44186 }, { "epoch": 3.4834055971620024, "grad_norm": 0.4801187217235565, "learning_rate": 0.0005, "loss": 1.2073744535446167, "step": 44187 }, { "epoch": 3.483484430429641, "grad_norm": 0.4746382534503937, "learning_rate": 0.0005, "loss": 1.303282618522644, "step": 44188 }, { "epoch": 3.48356326369728, "grad_norm": 0.4333014488220215, "learning_rate": 0.0005, "loss": 1.1050889492034912, "step": 44189 }, { "epoch": 3.4836420969649193, "grad_norm": 0.5415758490562439, "learning_rate": 0.0005, "loss": 1.1547627449035645, "step": 44190 }, { "epoch": 3.483720930232558, "grad_norm": 0.43933144211769104, "learning_rate": 0.0005, "loss": 1.321742296218872, "step": 44191 }, { "epoch": 3.483799763500197, "grad_norm": 0.44614580273628235, "learning_rate": 0.0005, "loss": 1.353164553642273, "step": 44192 }, { "epoch": 3.483878596767836, "grad_norm": 0.6980538964271545, "learning_rate": 0.0005, "loss": 1.3192816972732544, "step": 44193 }, { "epoch": 3.483957430035475, "grad_norm": 0.42394551634788513, "learning_rate": 0.0005, "loss": 1.2238924503326416, "step": 44194 }, { "epoch": 3.484036263303114, "grad_norm": 0.5213935375213623, "learning_rate": 0.0005, "loss": 1.1841583251953125, "step": 44195 }, { "epoch": 3.484115096570753, "grad_norm": 0.43098175525665283, "learning_rate": 0.0005, "loss": 1.1983757019042969, "step": 44196 }, { "epoch": 3.484193929838392, "grad_norm": 0.4685122072696686, "learning_rate": 0.0005, "loss": 1.2082901000976562, "step": 44197 }, { "epoch": 3.4842727631060306, "grad_norm": 0.49727723002433777, "learning_rate": 0.0005, "loss": 1.0818524360656738, "step": 44198 }, { "epoch": 3.4843515963736698, "grad_norm": 0.5621758699417114, "learning_rate": 0.0005, "loss": 1.3477041721343994, "step": 44199 }, { "epoch": 3.4844304296413084, "grad_norm": 0.41534870862960815, "learning_rate": 0.0005, "loss": 1.286872386932373, "step": 44200 }, { "epoch": 3.4845092629089476, "grad_norm": 0.5549207925796509, "learning_rate": 0.0005, "loss": 1.3981331586837769, "step": 44201 }, { "epoch": 3.4845880961765867, "grad_norm": 0.617201030254364, "learning_rate": 0.0005, "loss": 1.4878361225128174, "step": 44202 }, { "epoch": 3.4846669294442254, "grad_norm": 0.49457913637161255, "learning_rate": 0.0005, "loss": 1.3006327152252197, "step": 44203 }, { "epoch": 3.4847457627118645, "grad_norm": 0.6761574149131775, "learning_rate": 0.0005, "loss": 1.4855002164840698, "step": 44204 }, { "epoch": 3.4848245959795032, "grad_norm": 0.5415863990783691, "learning_rate": 0.0005, "loss": 1.405483603477478, "step": 44205 }, { "epoch": 3.4849034292471424, "grad_norm": 0.5018265247344971, "learning_rate": 0.0005, "loss": 1.1043410301208496, "step": 44206 }, { "epoch": 3.484982262514781, "grad_norm": 0.5635997653007507, "learning_rate": 0.0005, "loss": 1.4769690036773682, "step": 44207 }, { "epoch": 3.48506109578242, "grad_norm": 0.9048661589622498, "learning_rate": 0.0005, "loss": 1.3030383586883545, "step": 44208 }, { "epoch": 3.4851399290500593, "grad_norm": 0.529050350189209, "learning_rate": 0.0005, "loss": 1.261720895767212, "step": 44209 }, { "epoch": 3.485218762317698, "grad_norm": 0.4912360906600952, "learning_rate": 0.0005, "loss": 1.2051811218261719, "step": 44210 }, { "epoch": 3.485297595585337, "grad_norm": 0.4858928322792053, "learning_rate": 0.0005, "loss": 1.3533698320388794, "step": 44211 }, { "epoch": 3.485376428852976, "grad_norm": 0.49590086936950684, "learning_rate": 0.0005, "loss": 1.163379430770874, "step": 44212 }, { "epoch": 3.485455262120615, "grad_norm": 0.47765809297561646, "learning_rate": 0.0005, "loss": 1.3934667110443115, "step": 44213 }, { "epoch": 3.4855340953882537, "grad_norm": 0.5091080665588379, "learning_rate": 0.0005, "loss": 1.3581023216247559, "step": 44214 }, { "epoch": 3.485612928655893, "grad_norm": 0.39397844672203064, "learning_rate": 0.0005, "loss": 0.9919247627258301, "step": 44215 }, { "epoch": 3.485691761923532, "grad_norm": 0.5303303599357605, "learning_rate": 0.0005, "loss": 1.3164877891540527, "step": 44216 }, { "epoch": 3.4857705951911706, "grad_norm": 0.49566641449928284, "learning_rate": 0.0005, "loss": 1.2804350852966309, "step": 44217 }, { "epoch": 3.4858494284588097, "grad_norm": 0.5609853267669678, "learning_rate": 0.0005, "loss": 1.3844170570373535, "step": 44218 }, { "epoch": 3.4859282617264484, "grad_norm": 0.5356217622756958, "learning_rate": 0.0005, "loss": 1.2105387449264526, "step": 44219 }, { "epoch": 3.4860070949940876, "grad_norm": 0.5360509157180786, "learning_rate": 0.0005, "loss": 1.4655957221984863, "step": 44220 }, { "epoch": 3.4860859282617263, "grad_norm": 0.524792492389679, "learning_rate": 0.0005, "loss": 1.256117820739746, "step": 44221 }, { "epoch": 3.4861647615293654, "grad_norm": 0.966243326663971, "learning_rate": 0.0005, "loss": 1.5014532804489136, "step": 44222 }, { "epoch": 3.4862435947970045, "grad_norm": 0.49699655175209045, "learning_rate": 0.0005, "loss": 1.2211344242095947, "step": 44223 }, { "epoch": 3.486322428064643, "grad_norm": 0.5571328997612, "learning_rate": 0.0005, "loss": 1.4175610542297363, "step": 44224 }, { "epoch": 3.4864012613322823, "grad_norm": 0.4899772107601166, "learning_rate": 0.0005, "loss": 1.1571131944656372, "step": 44225 }, { "epoch": 3.486480094599921, "grad_norm": 0.45844706892967224, "learning_rate": 0.0005, "loss": 1.3636430501937866, "step": 44226 }, { "epoch": 3.48655892786756, "grad_norm": 0.5833799242973328, "learning_rate": 0.0005, "loss": 1.2798383235931396, "step": 44227 }, { "epoch": 3.4866377611351993, "grad_norm": 0.5947113037109375, "learning_rate": 0.0005, "loss": 1.4641430377960205, "step": 44228 }, { "epoch": 3.486716594402838, "grad_norm": 0.5411108136177063, "learning_rate": 0.0005, "loss": 1.2785775661468506, "step": 44229 }, { "epoch": 3.486795427670477, "grad_norm": 0.43417561054229736, "learning_rate": 0.0005, "loss": 1.1570791006088257, "step": 44230 }, { "epoch": 3.486874260938116, "grad_norm": 0.4600944519042969, "learning_rate": 0.0005, "loss": 1.2082797288894653, "step": 44231 }, { "epoch": 3.486953094205755, "grad_norm": 0.49167361855506897, "learning_rate": 0.0005, "loss": 1.2317450046539307, "step": 44232 }, { "epoch": 3.4870319274733936, "grad_norm": 0.6172524690628052, "learning_rate": 0.0005, "loss": 1.4804739952087402, "step": 44233 }, { "epoch": 3.4871107607410328, "grad_norm": 0.37182140350341797, "learning_rate": 0.0005, "loss": 1.1223441362380981, "step": 44234 }, { "epoch": 3.4871895940086715, "grad_norm": 0.468504399061203, "learning_rate": 0.0005, "loss": 1.3529376983642578, "step": 44235 }, { "epoch": 3.4872684272763106, "grad_norm": 0.46468469500541687, "learning_rate": 0.0005, "loss": 1.3914482593536377, "step": 44236 }, { "epoch": 3.4873472605439497, "grad_norm": 0.4460519254207611, "learning_rate": 0.0005, "loss": 1.2058196067810059, "step": 44237 }, { "epoch": 3.4874260938115884, "grad_norm": 0.4376492202281952, "learning_rate": 0.0005, "loss": 1.1773182153701782, "step": 44238 }, { "epoch": 3.4875049270792275, "grad_norm": 0.6940970420837402, "learning_rate": 0.0005, "loss": 1.27024245262146, "step": 44239 }, { "epoch": 3.4875837603468662, "grad_norm": 0.5408843755722046, "learning_rate": 0.0005, "loss": 1.1861226558685303, "step": 44240 }, { "epoch": 3.4876625936145054, "grad_norm": 0.46009618043899536, "learning_rate": 0.0005, "loss": 1.1780530214309692, "step": 44241 }, { "epoch": 3.4877414268821445, "grad_norm": 0.5537482500076294, "learning_rate": 0.0005, "loss": 1.4163880348205566, "step": 44242 }, { "epoch": 3.487820260149783, "grad_norm": 0.5837041139602661, "learning_rate": 0.0005, "loss": 1.3218495845794678, "step": 44243 }, { "epoch": 3.4878990934174223, "grad_norm": 0.6168889403343201, "learning_rate": 0.0005, "loss": 1.3415944576263428, "step": 44244 }, { "epoch": 3.487977926685061, "grad_norm": 0.47908100485801697, "learning_rate": 0.0005, "loss": 1.1381865739822388, "step": 44245 }, { "epoch": 3.4880567599527, "grad_norm": 0.5037289261817932, "learning_rate": 0.0005, "loss": 1.4590675830841064, "step": 44246 }, { "epoch": 3.488135593220339, "grad_norm": 0.5994060635566711, "learning_rate": 0.0005, "loss": 1.4658591747283936, "step": 44247 }, { "epoch": 3.488214426487978, "grad_norm": 0.4561416208744049, "learning_rate": 0.0005, "loss": 1.1527199745178223, "step": 44248 }, { "epoch": 3.4882932597556167, "grad_norm": 0.5860366225242615, "learning_rate": 0.0005, "loss": 1.2754747867584229, "step": 44249 }, { "epoch": 3.488372093023256, "grad_norm": 0.4894123077392578, "learning_rate": 0.0005, "loss": 1.3213722705841064, "step": 44250 }, { "epoch": 3.488450926290895, "grad_norm": 0.46663275361061096, "learning_rate": 0.0005, "loss": 1.2421698570251465, "step": 44251 }, { "epoch": 3.4885297595585336, "grad_norm": 0.4422776699066162, "learning_rate": 0.0005, "loss": 1.1389827728271484, "step": 44252 }, { "epoch": 3.4886085928261727, "grad_norm": 0.6110090017318726, "learning_rate": 0.0005, "loss": 1.211732029914856, "step": 44253 }, { "epoch": 3.4886874260938114, "grad_norm": 0.4504426121711731, "learning_rate": 0.0005, "loss": 1.1214077472686768, "step": 44254 }, { "epoch": 3.4887662593614506, "grad_norm": 0.5179266333580017, "learning_rate": 0.0005, "loss": 1.2909948825836182, "step": 44255 }, { "epoch": 3.4888450926290897, "grad_norm": 0.8640015125274658, "learning_rate": 0.0005, "loss": 1.2439332008361816, "step": 44256 }, { "epoch": 3.4889239258967284, "grad_norm": 0.45543867349624634, "learning_rate": 0.0005, "loss": 1.308074951171875, "step": 44257 }, { "epoch": 3.4890027591643675, "grad_norm": 0.39239969849586487, "learning_rate": 0.0005, "loss": 0.9352260231971741, "step": 44258 }, { "epoch": 3.489081592432006, "grad_norm": 0.6514652967453003, "learning_rate": 0.0005, "loss": 1.3978389501571655, "step": 44259 }, { "epoch": 3.4891604256996454, "grad_norm": 0.44418808817863464, "learning_rate": 0.0005, "loss": 1.127500295639038, "step": 44260 }, { "epoch": 3.489239258967284, "grad_norm": 0.5132675170898438, "learning_rate": 0.0005, "loss": 1.4013657569885254, "step": 44261 }, { "epoch": 3.489318092234923, "grad_norm": 0.5102426409721375, "learning_rate": 0.0005, "loss": 1.1176879405975342, "step": 44262 }, { "epoch": 3.489396925502562, "grad_norm": 0.5934184789657593, "learning_rate": 0.0005, "loss": 1.3992555141448975, "step": 44263 }, { "epoch": 3.489475758770201, "grad_norm": 0.6459006667137146, "learning_rate": 0.0005, "loss": 1.4001957178115845, "step": 44264 }, { "epoch": 3.48955459203784, "grad_norm": 0.564344048500061, "learning_rate": 0.0005, "loss": 1.328603744506836, "step": 44265 }, { "epoch": 3.489633425305479, "grad_norm": 0.5388319492340088, "learning_rate": 0.0005, "loss": 1.3301472663879395, "step": 44266 }, { "epoch": 3.489712258573118, "grad_norm": 0.5245396494865417, "learning_rate": 0.0005, "loss": 1.1591389179229736, "step": 44267 }, { "epoch": 3.4897910918407566, "grad_norm": 0.5049525499343872, "learning_rate": 0.0005, "loss": 1.2619563341140747, "step": 44268 }, { "epoch": 3.4898699251083958, "grad_norm": 0.6886921525001526, "learning_rate": 0.0005, "loss": 1.394723892211914, "step": 44269 }, { "epoch": 3.489948758376035, "grad_norm": 0.5229921340942383, "learning_rate": 0.0005, "loss": 1.1924253702163696, "step": 44270 }, { "epoch": 3.4900275916436736, "grad_norm": 0.5469655394554138, "learning_rate": 0.0005, "loss": 1.3142175674438477, "step": 44271 }, { "epoch": 3.4901064249113127, "grad_norm": 0.6261155605316162, "learning_rate": 0.0005, "loss": 1.2721104621887207, "step": 44272 }, { "epoch": 3.4901852581789514, "grad_norm": 0.6151033043861389, "learning_rate": 0.0005, "loss": 1.57529878616333, "step": 44273 }, { "epoch": 3.4902640914465906, "grad_norm": 0.5394106507301331, "learning_rate": 0.0005, "loss": 1.4077117443084717, "step": 44274 }, { "epoch": 3.4903429247142292, "grad_norm": 0.5641220808029175, "learning_rate": 0.0005, "loss": 1.2878531217575073, "step": 44275 }, { "epoch": 3.4904217579818684, "grad_norm": 0.5024354457855225, "learning_rate": 0.0005, "loss": 1.4155434370040894, "step": 44276 }, { "epoch": 3.490500591249507, "grad_norm": 0.4968448579311371, "learning_rate": 0.0005, "loss": 1.330612063407898, "step": 44277 }, { "epoch": 3.490579424517146, "grad_norm": 0.4428018629550934, "learning_rate": 0.0005, "loss": 1.24473237991333, "step": 44278 }, { "epoch": 3.4906582577847853, "grad_norm": 0.4389900863170624, "learning_rate": 0.0005, "loss": 1.3308652639389038, "step": 44279 }, { "epoch": 3.490737091052424, "grad_norm": 0.5596653819084167, "learning_rate": 0.0005, "loss": 1.2747567892074585, "step": 44280 }, { "epoch": 3.490815924320063, "grad_norm": 0.4830167591571808, "learning_rate": 0.0005, "loss": 1.271946907043457, "step": 44281 }, { "epoch": 3.490894757587702, "grad_norm": 0.6180931925773621, "learning_rate": 0.0005, "loss": 1.2773876190185547, "step": 44282 }, { "epoch": 3.490973590855341, "grad_norm": 0.4916451871395111, "learning_rate": 0.0005, "loss": 1.3497130870819092, "step": 44283 }, { "epoch": 3.49105242412298, "grad_norm": 0.47184252738952637, "learning_rate": 0.0005, "loss": 1.3107829093933105, "step": 44284 }, { "epoch": 3.491131257390619, "grad_norm": 0.45394471287727356, "learning_rate": 0.0005, "loss": 1.0960240364074707, "step": 44285 }, { "epoch": 3.491210090658258, "grad_norm": 0.4713834524154663, "learning_rate": 0.0005, "loss": 1.3577466011047363, "step": 44286 }, { "epoch": 3.4912889239258966, "grad_norm": 0.37329891324043274, "learning_rate": 0.0005, "loss": 1.3315179347991943, "step": 44287 }, { "epoch": 3.4913677571935358, "grad_norm": 0.4305056929588318, "learning_rate": 0.0005, "loss": 1.1639318466186523, "step": 44288 }, { "epoch": 3.4914465904611744, "grad_norm": 0.7873548269271851, "learning_rate": 0.0005, "loss": 1.551262378692627, "step": 44289 }, { "epoch": 3.4915254237288136, "grad_norm": 0.49437215924263, "learning_rate": 0.0005, "loss": 1.2418296337127686, "step": 44290 }, { "epoch": 3.4916042569964523, "grad_norm": 0.5388768911361694, "learning_rate": 0.0005, "loss": 1.590428352355957, "step": 44291 }, { "epoch": 3.4916830902640914, "grad_norm": 0.3970334529876709, "learning_rate": 0.0005, "loss": 1.1436213254928589, "step": 44292 }, { "epoch": 3.4917619235317305, "grad_norm": 0.5518372654914856, "learning_rate": 0.0005, "loss": 1.353054404258728, "step": 44293 }, { "epoch": 3.4918407567993692, "grad_norm": 0.4804656207561493, "learning_rate": 0.0005, "loss": 1.259535789489746, "step": 44294 }, { "epoch": 3.4919195900670084, "grad_norm": 0.4433988034725189, "learning_rate": 0.0005, "loss": 1.1272566318511963, "step": 44295 }, { "epoch": 3.491998423334647, "grad_norm": 0.44000253081321716, "learning_rate": 0.0005, "loss": 1.1256372928619385, "step": 44296 }, { "epoch": 3.492077256602286, "grad_norm": 0.5724027156829834, "learning_rate": 0.0005, "loss": 1.330965280532837, "step": 44297 }, { "epoch": 3.4921560898699253, "grad_norm": 0.4713880717754364, "learning_rate": 0.0005, "loss": 1.2481786012649536, "step": 44298 }, { "epoch": 3.492234923137564, "grad_norm": 0.46418219804763794, "learning_rate": 0.0005, "loss": 1.4864609241485596, "step": 44299 }, { "epoch": 3.492313756405203, "grad_norm": 0.4414752125740051, "learning_rate": 0.0005, "loss": 1.251083493232727, "step": 44300 }, { "epoch": 3.492392589672842, "grad_norm": 0.499362975358963, "learning_rate": 0.0005, "loss": 1.1920615434646606, "step": 44301 }, { "epoch": 3.492471422940481, "grad_norm": 0.47345513105392456, "learning_rate": 0.0005, "loss": 1.234512448310852, "step": 44302 }, { "epoch": 3.4925502562081197, "grad_norm": 0.5192893743515015, "learning_rate": 0.0005, "loss": 1.3379064798355103, "step": 44303 }, { "epoch": 3.492629089475759, "grad_norm": 0.5738046169281006, "learning_rate": 0.0005, "loss": 1.2931184768676758, "step": 44304 }, { "epoch": 3.4927079227433975, "grad_norm": 0.46125397086143494, "learning_rate": 0.0005, "loss": 1.2708525657653809, "step": 44305 }, { "epoch": 3.4927867560110366, "grad_norm": 0.5451238751411438, "learning_rate": 0.0005, "loss": 1.286603331565857, "step": 44306 }, { "epoch": 3.4928655892786757, "grad_norm": 0.5151034593582153, "learning_rate": 0.0005, "loss": 1.30344557762146, "step": 44307 }, { "epoch": 3.4929444225463144, "grad_norm": 0.4477463960647583, "learning_rate": 0.0005, "loss": 1.1921849250793457, "step": 44308 }, { "epoch": 3.4930232558139536, "grad_norm": 0.6344972252845764, "learning_rate": 0.0005, "loss": 1.5774877071380615, "step": 44309 }, { "epoch": 3.4931020890815923, "grad_norm": 0.5101532936096191, "learning_rate": 0.0005, "loss": 1.1936969757080078, "step": 44310 }, { "epoch": 3.4931809223492314, "grad_norm": 0.5965843796730042, "learning_rate": 0.0005, "loss": 1.337509274482727, "step": 44311 }, { "epoch": 3.4932597556168705, "grad_norm": 0.5241315364837646, "learning_rate": 0.0005, "loss": 1.2617499828338623, "step": 44312 }, { "epoch": 3.493338588884509, "grad_norm": 0.47662806510925293, "learning_rate": 0.0005, "loss": 1.171041488647461, "step": 44313 }, { "epoch": 3.4934174221521483, "grad_norm": 0.7135937809944153, "learning_rate": 0.0005, "loss": 1.2880425453186035, "step": 44314 }, { "epoch": 3.493496255419787, "grad_norm": 0.5595747828483582, "learning_rate": 0.0005, "loss": 1.2753654718399048, "step": 44315 }, { "epoch": 3.493575088687426, "grad_norm": 0.4794878661632538, "learning_rate": 0.0005, "loss": 1.2023364305496216, "step": 44316 }, { "epoch": 3.493653921955065, "grad_norm": 0.7672681212425232, "learning_rate": 0.0005, "loss": 1.2100605964660645, "step": 44317 }, { "epoch": 3.493732755222704, "grad_norm": 0.5474646687507629, "learning_rate": 0.0005, "loss": 1.3465383052825928, "step": 44318 }, { "epoch": 3.4938115884903427, "grad_norm": 0.48668375611305237, "learning_rate": 0.0005, "loss": 1.2788052558898926, "step": 44319 }, { "epoch": 3.493890421757982, "grad_norm": 0.4478723704814911, "learning_rate": 0.0005, "loss": 1.2640737295150757, "step": 44320 }, { "epoch": 3.493969255025621, "grad_norm": 0.4462355077266693, "learning_rate": 0.0005, "loss": 1.2004191875457764, "step": 44321 }, { "epoch": 3.4940480882932596, "grad_norm": 0.5629804134368896, "learning_rate": 0.0005, "loss": 1.337852954864502, "step": 44322 }, { "epoch": 3.4941269215608988, "grad_norm": 0.5148769021034241, "learning_rate": 0.0005, "loss": 1.241895079612732, "step": 44323 }, { "epoch": 3.4942057548285375, "grad_norm": 0.6119059920310974, "learning_rate": 0.0005, "loss": 1.1932926177978516, "step": 44324 }, { "epoch": 3.4942845880961766, "grad_norm": 0.416105717420578, "learning_rate": 0.0005, "loss": 1.1792958974838257, "step": 44325 }, { "epoch": 3.4943634213638157, "grad_norm": 0.5048575401306152, "learning_rate": 0.0005, "loss": 1.2491655349731445, "step": 44326 }, { "epoch": 3.4944422546314544, "grad_norm": 0.4943040907382965, "learning_rate": 0.0005, "loss": 1.2380123138427734, "step": 44327 }, { "epoch": 3.4945210878990935, "grad_norm": 0.4536762535572052, "learning_rate": 0.0005, "loss": 1.10715913772583, "step": 44328 }, { "epoch": 3.4945999211667322, "grad_norm": 0.4497388005256653, "learning_rate": 0.0005, "loss": 1.2449641227722168, "step": 44329 }, { "epoch": 3.4946787544343714, "grad_norm": 0.4162633717060089, "learning_rate": 0.0005, "loss": 1.1388225555419922, "step": 44330 }, { "epoch": 3.49475758770201, "grad_norm": 0.5025551319122314, "learning_rate": 0.0005, "loss": 1.1430643796920776, "step": 44331 }, { "epoch": 3.494836420969649, "grad_norm": 0.5867410898208618, "learning_rate": 0.0005, "loss": 1.3369625806808472, "step": 44332 }, { "epoch": 3.4949152542372883, "grad_norm": 0.4202621579170227, "learning_rate": 0.0005, "loss": 1.1070141792297363, "step": 44333 }, { "epoch": 3.494994087504927, "grad_norm": 0.42532727122306824, "learning_rate": 0.0005, "loss": 1.109890341758728, "step": 44334 }, { "epoch": 3.495072920772566, "grad_norm": 0.6661102771759033, "learning_rate": 0.0005, "loss": 1.3123688697814941, "step": 44335 }, { "epoch": 3.495151754040205, "grad_norm": 0.5082806348800659, "learning_rate": 0.0005, "loss": 1.3060052394866943, "step": 44336 }, { "epoch": 3.495230587307844, "grad_norm": 0.4262036085128784, "learning_rate": 0.0005, "loss": 1.240134835243225, "step": 44337 }, { "epoch": 3.4953094205754827, "grad_norm": 0.5060474872589111, "learning_rate": 0.0005, "loss": 1.2320563793182373, "step": 44338 }, { "epoch": 3.495388253843122, "grad_norm": 0.4207432270050049, "learning_rate": 0.0005, "loss": 1.103682279586792, "step": 44339 }, { "epoch": 3.495467087110761, "grad_norm": 0.5697817802429199, "learning_rate": 0.0005, "loss": 1.3480165004730225, "step": 44340 }, { "epoch": 3.4955459203783996, "grad_norm": 0.6777956485748291, "learning_rate": 0.0005, "loss": 1.3240941762924194, "step": 44341 }, { "epoch": 3.4956247536460388, "grad_norm": 0.5323829054832458, "learning_rate": 0.0005, "loss": 1.2475930452346802, "step": 44342 }, { "epoch": 3.4957035869136774, "grad_norm": 0.47864893078804016, "learning_rate": 0.0005, "loss": 1.4522188901901245, "step": 44343 }, { "epoch": 3.4957824201813166, "grad_norm": 0.5490669012069702, "learning_rate": 0.0005, "loss": 1.2880465984344482, "step": 44344 }, { "epoch": 3.4958612534489557, "grad_norm": 0.4444907009601593, "learning_rate": 0.0005, "loss": 1.2826656103134155, "step": 44345 }, { "epoch": 3.4959400867165944, "grad_norm": 0.5286296010017395, "learning_rate": 0.0005, "loss": 1.500462532043457, "step": 44346 }, { "epoch": 3.4960189199842335, "grad_norm": 0.5400375127792358, "learning_rate": 0.0005, "loss": 1.4351389408111572, "step": 44347 }, { "epoch": 3.496097753251872, "grad_norm": 0.4794093668460846, "learning_rate": 0.0005, "loss": 1.1005468368530273, "step": 44348 }, { "epoch": 3.4961765865195114, "grad_norm": 0.4391908049583435, "learning_rate": 0.0005, "loss": 1.145662546157837, "step": 44349 }, { "epoch": 3.49625541978715, "grad_norm": 0.44223904609680176, "learning_rate": 0.0005, "loss": 1.3320255279541016, "step": 44350 }, { "epoch": 3.496334253054789, "grad_norm": 0.41133975982666016, "learning_rate": 0.0005, "loss": 1.1833864450454712, "step": 44351 }, { "epoch": 3.496413086322428, "grad_norm": 0.42691469192504883, "learning_rate": 0.0005, "loss": 1.3609164953231812, "step": 44352 }, { "epoch": 3.496491919590067, "grad_norm": 0.4900828003883362, "learning_rate": 0.0005, "loss": 1.3482729196548462, "step": 44353 }, { "epoch": 3.496570752857706, "grad_norm": 0.4448995888233185, "learning_rate": 0.0005, "loss": 1.2177472114562988, "step": 44354 }, { "epoch": 3.496649586125345, "grad_norm": 0.5164656639099121, "learning_rate": 0.0005, "loss": 1.3881494998931885, "step": 44355 }, { "epoch": 3.496728419392984, "grad_norm": 0.40822574496269226, "learning_rate": 0.0005, "loss": 1.2617194652557373, "step": 44356 }, { "epoch": 3.4968072526606226, "grad_norm": 0.47539934515953064, "learning_rate": 0.0005, "loss": 1.219170093536377, "step": 44357 }, { "epoch": 3.496886085928262, "grad_norm": 0.5434710383415222, "learning_rate": 0.0005, "loss": 1.2226780652999878, "step": 44358 }, { "epoch": 3.496964919195901, "grad_norm": 0.4545345902442932, "learning_rate": 0.0005, "loss": 1.1877484321594238, "step": 44359 }, { "epoch": 3.4970437524635396, "grad_norm": 0.5636056065559387, "learning_rate": 0.0005, "loss": 1.4825685024261475, "step": 44360 }, { "epoch": 3.4971225857311787, "grad_norm": 0.5109183192253113, "learning_rate": 0.0005, "loss": 1.1958460807800293, "step": 44361 }, { "epoch": 3.4972014189988174, "grad_norm": 0.4793962836265564, "learning_rate": 0.0005, "loss": 1.334839105606079, "step": 44362 }, { "epoch": 3.4972802522664566, "grad_norm": 0.5519615411758423, "learning_rate": 0.0005, "loss": 1.3459084033966064, "step": 44363 }, { "epoch": 3.4973590855340952, "grad_norm": 0.5578660368919373, "learning_rate": 0.0005, "loss": 1.3773233890533447, "step": 44364 }, { "epoch": 3.4974379188017344, "grad_norm": 0.5348013639450073, "learning_rate": 0.0005, "loss": 1.2419703006744385, "step": 44365 }, { "epoch": 3.497516752069373, "grad_norm": 0.5714855790138245, "learning_rate": 0.0005, "loss": 1.349615454673767, "step": 44366 }, { "epoch": 3.497595585337012, "grad_norm": 0.46615445613861084, "learning_rate": 0.0005, "loss": 1.160416603088379, "step": 44367 }, { "epoch": 3.4976744186046513, "grad_norm": 0.568857729434967, "learning_rate": 0.0005, "loss": 1.365412712097168, "step": 44368 }, { "epoch": 3.49775325187229, "grad_norm": 0.6020506620407104, "learning_rate": 0.0005, "loss": 1.249292254447937, "step": 44369 }, { "epoch": 3.497832085139929, "grad_norm": 0.511961042881012, "learning_rate": 0.0005, "loss": 1.2898070812225342, "step": 44370 }, { "epoch": 3.497910918407568, "grad_norm": 0.43883851170539856, "learning_rate": 0.0005, "loss": 1.2264052629470825, "step": 44371 }, { "epoch": 3.497989751675207, "grad_norm": 0.46949949860572815, "learning_rate": 0.0005, "loss": 1.2499964237213135, "step": 44372 }, { "epoch": 3.498068584942846, "grad_norm": 0.5473625659942627, "learning_rate": 0.0005, "loss": 1.354202389717102, "step": 44373 }, { "epoch": 3.498147418210485, "grad_norm": 0.47900906205177307, "learning_rate": 0.0005, "loss": 1.1520460844039917, "step": 44374 }, { "epoch": 3.498226251478124, "grad_norm": 0.41811251640319824, "learning_rate": 0.0005, "loss": 1.1549460887908936, "step": 44375 }, { "epoch": 3.4983050847457626, "grad_norm": 0.4558243453502655, "learning_rate": 0.0005, "loss": 1.400622010231018, "step": 44376 }, { "epoch": 3.4983839180134018, "grad_norm": 0.5738328099250793, "learning_rate": 0.0005, "loss": 1.2726733684539795, "step": 44377 }, { "epoch": 3.4984627512810405, "grad_norm": 0.522406280040741, "learning_rate": 0.0005, "loss": 1.3697477579116821, "step": 44378 }, { "epoch": 3.4985415845486796, "grad_norm": 0.4408237934112549, "learning_rate": 0.0005, "loss": 1.264073133468628, "step": 44379 }, { "epoch": 3.4986204178163183, "grad_norm": 3.6674160957336426, "learning_rate": 0.0005, "loss": 1.3445500135421753, "step": 44380 }, { "epoch": 3.4986992510839574, "grad_norm": 0.46807026863098145, "learning_rate": 0.0005, "loss": 1.1564605236053467, "step": 44381 }, { "epoch": 3.4987780843515965, "grad_norm": 0.479879230260849, "learning_rate": 0.0005, "loss": 1.332977056503296, "step": 44382 }, { "epoch": 3.4988569176192352, "grad_norm": 0.550933837890625, "learning_rate": 0.0005, "loss": 1.2532484531402588, "step": 44383 }, { "epoch": 3.4989357508868744, "grad_norm": 0.5158593654632568, "learning_rate": 0.0005, "loss": 1.3895447254180908, "step": 44384 }, { "epoch": 3.499014584154513, "grad_norm": 0.40580517053604126, "learning_rate": 0.0005, "loss": 1.1656229496002197, "step": 44385 }, { "epoch": 3.499093417422152, "grad_norm": 0.4078115224838257, "learning_rate": 0.0005, "loss": 1.3284504413604736, "step": 44386 }, { "epoch": 3.4991722506897913, "grad_norm": 0.7590762376785278, "learning_rate": 0.0005, "loss": 1.0942180156707764, "step": 44387 }, { "epoch": 3.49925108395743, "grad_norm": 0.41009360551834106, "learning_rate": 0.0005, "loss": 1.1230604648590088, "step": 44388 }, { "epoch": 3.499329917225069, "grad_norm": 0.5358414649963379, "learning_rate": 0.0005, "loss": 1.3897954225540161, "step": 44389 }, { "epoch": 3.499408750492708, "grad_norm": 0.5392847061157227, "learning_rate": 0.0005, "loss": 1.3064637184143066, "step": 44390 }, { "epoch": 3.499487583760347, "grad_norm": 0.6083241105079651, "learning_rate": 0.0005, "loss": 1.3107750415802002, "step": 44391 }, { "epoch": 3.4995664170279857, "grad_norm": 0.40790054202079773, "learning_rate": 0.0005, "loss": 1.1061313152313232, "step": 44392 }, { "epoch": 3.499645250295625, "grad_norm": 0.43766170740127563, "learning_rate": 0.0005, "loss": 1.3528683185577393, "step": 44393 }, { "epoch": 3.4997240835632635, "grad_norm": 0.49082282185554504, "learning_rate": 0.0005, "loss": 1.276902675628662, "step": 44394 }, { "epoch": 3.4998029168309026, "grad_norm": 0.5299040675163269, "learning_rate": 0.0005, "loss": 1.3871972560882568, "step": 44395 }, { "epoch": 3.4998817500985417, "grad_norm": 0.5913461446762085, "learning_rate": 0.0005, "loss": 1.2165371179580688, "step": 44396 }, { "epoch": 3.4999605833661804, "grad_norm": 0.728201687335968, "learning_rate": 0.0005, "loss": 1.2695666551589966, "step": 44397 }, { "epoch": 3.5000394166338196, "grad_norm": 0.5495792627334595, "learning_rate": 0.0005, "loss": 1.4122304916381836, "step": 44398 }, { "epoch": 3.5001182499014583, "grad_norm": 0.5177971720695496, "learning_rate": 0.0005, "loss": 1.1215791702270508, "step": 44399 }, { "epoch": 3.5001970831690974, "grad_norm": 0.7091130614280701, "learning_rate": 0.0005, "loss": 1.2925665378570557, "step": 44400 }, { "epoch": 3.5002759164367365, "grad_norm": 0.5759328007698059, "learning_rate": 0.0005, "loss": 1.5884379148483276, "step": 44401 }, { "epoch": 3.500354749704375, "grad_norm": 0.5645767450332642, "learning_rate": 0.0005, "loss": 1.3333903551101685, "step": 44402 }, { "epoch": 3.5004335829720143, "grad_norm": 0.5360730290412903, "learning_rate": 0.0005, "loss": 1.380666732788086, "step": 44403 }, { "epoch": 3.500512416239653, "grad_norm": 0.5658940076828003, "learning_rate": 0.0005, "loss": 1.1858357191085815, "step": 44404 }, { "epoch": 3.500591249507292, "grad_norm": 0.4804772734642029, "learning_rate": 0.0005, "loss": 1.365801215171814, "step": 44405 }, { "epoch": 3.500670082774931, "grad_norm": 0.5393449664115906, "learning_rate": 0.0005, "loss": 1.2741596698760986, "step": 44406 }, { "epoch": 3.50074891604257, "grad_norm": 0.6805107593536377, "learning_rate": 0.0005, "loss": 1.2885551452636719, "step": 44407 }, { "epoch": 3.5008277493102087, "grad_norm": 0.4208737015724182, "learning_rate": 0.0005, "loss": 1.0855299234390259, "step": 44408 }, { "epoch": 3.500906582577848, "grad_norm": 0.4348449110984802, "learning_rate": 0.0005, "loss": 1.2485225200653076, "step": 44409 }, { "epoch": 3.500985415845487, "grad_norm": 0.5672887563705444, "learning_rate": 0.0005, "loss": 1.3449947834014893, "step": 44410 }, { "epoch": 3.5010642491131256, "grad_norm": 0.5397998690605164, "learning_rate": 0.0005, "loss": 1.1930749416351318, "step": 44411 }, { "epoch": 3.5011430823807648, "grad_norm": 0.4451470971107483, "learning_rate": 0.0005, "loss": 1.2609540224075317, "step": 44412 }, { "epoch": 3.5012219156484035, "grad_norm": 0.629694402217865, "learning_rate": 0.0005, "loss": 1.461031436920166, "step": 44413 }, { "epoch": 3.5013007489160426, "grad_norm": 0.48213252425193787, "learning_rate": 0.0005, "loss": 1.1322449445724487, "step": 44414 }, { "epoch": 3.5013795821836817, "grad_norm": 0.6509042382240295, "learning_rate": 0.0005, "loss": 1.208349347114563, "step": 44415 }, { "epoch": 3.5014584154513204, "grad_norm": 0.5068619251251221, "learning_rate": 0.0005, "loss": 1.2538940906524658, "step": 44416 }, { "epoch": 3.5015372487189595, "grad_norm": 0.5294203758239746, "learning_rate": 0.0005, "loss": 1.2861154079437256, "step": 44417 }, { "epoch": 3.5016160819865982, "grad_norm": 0.4703128933906555, "learning_rate": 0.0005, "loss": 1.289681315422058, "step": 44418 }, { "epoch": 3.5016949152542374, "grad_norm": 0.5769262909889221, "learning_rate": 0.0005, "loss": 1.3537671566009521, "step": 44419 }, { "epoch": 3.5017737485218765, "grad_norm": 0.4765819311141968, "learning_rate": 0.0005, "loss": 1.1965293884277344, "step": 44420 }, { "epoch": 3.501852581789515, "grad_norm": 0.661038339138031, "learning_rate": 0.0005, "loss": 1.2337589263916016, "step": 44421 }, { "epoch": 3.501931415057154, "grad_norm": 0.467708021402359, "learning_rate": 0.0005, "loss": 1.1679317951202393, "step": 44422 }, { "epoch": 3.502010248324793, "grad_norm": 0.6054033637046814, "learning_rate": 0.0005, "loss": 1.2412563562393188, "step": 44423 }, { "epoch": 3.502089081592432, "grad_norm": 0.5323799252510071, "learning_rate": 0.0005, "loss": 1.2309348583221436, "step": 44424 }, { "epoch": 3.502167914860071, "grad_norm": 0.5323849320411682, "learning_rate": 0.0005, "loss": 1.2465448379516602, "step": 44425 }, { "epoch": 3.50224674812771, "grad_norm": 0.4954186677932739, "learning_rate": 0.0005, "loss": 1.317109227180481, "step": 44426 }, { "epoch": 3.5023255813953487, "grad_norm": 0.4258725941181183, "learning_rate": 0.0005, "loss": 1.0741641521453857, "step": 44427 }, { "epoch": 3.502404414662988, "grad_norm": 0.4670664966106415, "learning_rate": 0.0005, "loss": 1.2087922096252441, "step": 44428 }, { "epoch": 3.502483247930627, "grad_norm": 0.5800259709358215, "learning_rate": 0.0005, "loss": 1.3043780326843262, "step": 44429 }, { "epoch": 3.5025620811982656, "grad_norm": 0.6055854558944702, "learning_rate": 0.0005, "loss": 1.6052438020706177, "step": 44430 }, { "epoch": 3.5026409144659048, "grad_norm": 0.4396646022796631, "learning_rate": 0.0005, "loss": 1.1913609504699707, "step": 44431 }, { "epoch": 3.5027197477335434, "grad_norm": 0.48704785108566284, "learning_rate": 0.0005, "loss": 1.2096718549728394, "step": 44432 }, { "epoch": 3.5027985810011826, "grad_norm": 0.5846884846687317, "learning_rate": 0.0005, "loss": 1.3633347749710083, "step": 44433 }, { "epoch": 3.5028774142688217, "grad_norm": 0.46382224559783936, "learning_rate": 0.0005, "loss": 1.47076416015625, "step": 44434 }, { "epoch": 3.5029562475364604, "grad_norm": 0.7112334370613098, "learning_rate": 0.0005, "loss": 1.3566303253173828, "step": 44435 }, { "epoch": 3.503035080804099, "grad_norm": 0.8993774056434631, "learning_rate": 0.0005, "loss": 1.3216497898101807, "step": 44436 }, { "epoch": 3.503113914071738, "grad_norm": 0.6205883026123047, "learning_rate": 0.0005, "loss": 1.2580077648162842, "step": 44437 }, { "epoch": 3.5031927473393774, "grad_norm": 0.4483308494091034, "learning_rate": 0.0005, "loss": 1.3941457271575928, "step": 44438 }, { "epoch": 3.503271580607016, "grad_norm": 0.45601004362106323, "learning_rate": 0.0005, "loss": 1.0939747095108032, "step": 44439 }, { "epoch": 3.503350413874655, "grad_norm": 0.4438018798828125, "learning_rate": 0.0005, "loss": 1.2565503120422363, "step": 44440 }, { "epoch": 3.503429247142294, "grad_norm": 0.4404948055744171, "learning_rate": 0.0005, "loss": 1.1096453666687012, "step": 44441 }, { "epoch": 3.503508080409933, "grad_norm": 0.49845466017723083, "learning_rate": 0.0005, "loss": 1.280943512916565, "step": 44442 }, { "epoch": 3.503586913677572, "grad_norm": 0.4840684235095978, "learning_rate": 0.0005, "loss": 1.2030411958694458, "step": 44443 }, { "epoch": 3.503665746945211, "grad_norm": 0.6235917210578918, "learning_rate": 0.0005, "loss": 1.2185485363006592, "step": 44444 }, { "epoch": 3.50374458021285, "grad_norm": 0.4504530727863312, "learning_rate": 0.0005, "loss": 1.403271198272705, "step": 44445 }, { "epoch": 3.5038234134804886, "grad_norm": 0.6943440437316895, "learning_rate": 0.0005, "loss": 1.3633782863616943, "step": 44446 }, { "epoch": 3.503902246748128, "grad_norm": 0.5896892547607422, "learning_rate": 0.0005, "loss": 1.4060708284378052, "step": 44447 }, { "epoch": 3.503981080015767, "grad_norm": 0.47967639565467834, "learning_rate": 0.0005, "loss": 1.2827361822128296, "step": 44448 }, { "epoch": 3.5040599132834056, "grad_norm": 0.5617758631706238, "learning_rate": 0.0005, "loss": 1.1806169748306274, "step": 44449 }, { "epoch": 3.5041387465510443, "grad_norm": 0.5632055997848511, "learning_rate": 0.0005, "loss": 1.2070186138153076, "step": 44450 }, { "epoch": 3.5042175798186834, "grad_norm": 0.5654571652412415, "learning_rate": 0.0005, "loss": 1.3743088245391846, "step": 44451 }, { "epoch": 3.5042964130863226, "grad_norm": 0.4479992389678955, "learning_rate": 0.0005, "loss": 1.0934205055236816, "step": 44452 }, { "epoch": 3.5043752463539612, "grad_norm": 0.5388084650039673, "learning_rate": 0.0005, "loss": 1.4412715435028076, "step": 44453 }, { "epoch": 3.5044540796216004, "grad_norm": 0.48463648557662964, "learning_rate": 0.0005, "loss": 1.2905871868133545, "step": 44454 }, { "epoch": 3.504532912889239, "grad_norm": 0.4624309837818146, "learning_rate": 0.0005, "loss": 1.1908230781555176, "step": 44455 }, { "epoch": 3.504611746156878, "grad_norm": 0.4539285898208618, "learning_rate": 0.0005, "loss": 1.2576253414154053, "step": 44456 }, { "epoch": 3.5046905794245173, "grad_norm": 0.4722810685634613, "learning_rate": 0.0005, "loss": 1.1559085845947266, "step": 44457 }, { "epoch": 3.504769412692156, "grad_norm": 0.5273647308349609, "learning_rate": 0.0005, "loss": 1.2757036685943604, "step": 44458 }, { "epoch": 3.504848245959795, "grad_norm": 0.5354132056236267, "learning_rate": 0.0005, "loss": 1.2964626550674438, "step": 44459 }, { "epoch": 3.504927079227434, "grad_norm": 0.5536348223686218, "learning_rate": 0.0005, "loss": 1.3199341297149658, "step": 44460 }, { "epoch": 3.505005912495073, "grad_norm": 0.41202837228775024, "learning_rate": 0.0005, "loss": 1.21401047706604, "step": 44461 }, { "epoch": 3.505084745762712, "grad_norm": 0.5575760006904602, "learning_rate": 0.0005, "loss": 1.3480356931686401, "step": 44462 }, { "epoch": 3.505163579030351, "grad_norm": 0.39437851309776306, "learning_rate": 0.0005, "loss": 1.2925364971160889, "step": 44463 }, { "epoch": 3.5052424122979895, "grad_norm": 0.5859208703041077, "learning_rate": 0.0005, "loss": 1.2643547058105469, "step": 44464 }, { "epoch": 3.5053212455656286, "grad_norm": 0.4701937437057495, "learning_rate": 0.0005, "loss": 1.300875186920166, "step": 44465 }, { "epoch": 3.5054000788332678, "grad_norm": 0.4474843144416809, "learning_rate": 0.0005, "loss": 1.0078829526901245, "step": 44466 }, { "epoch": 3.5054789121009065, "grad_norm": 0.4273213744163513, "learning_rate": 0.0005, "loss": 0.9937455058097839, "step": 44467 }, { "epoch": 3.5055577453685456, "grad_norm": 0.4722049832344055, "learning_rate": 0.0005, "loss": 1.424346685409546, "step": 44468 }, { "epoch": 3.5056365786361843, "grad_norm": 0.5188692212104797, "learning_rate": 0.0005, "loss": 1.2705037593841553, "step": 44469 }, { "epoch": 3.5057154119038234, "grad_norm": 0.4607539772987366, "learning_rate": 0.0005, "loss": 1.1138896942138672, "step": 44470 }, { "epoch": 3.5057942451714625, "grad_norm": 0.40386512875556946, "learning_rate": 0.0005, "loss": 1.1223647594451904, "step": 44471 }, { "epoch": 3.5058730784391012, "grad_norm": 0.4146333932876587, "learning_rate": 0.0005, "loss": 1.1788368225097656, "step": 44472 }, { "epoch": 3.5059519117067404, "grad_norm": 0.5663114786148071, "learning_rate": 0.0005, "loss": 1.3760082721710205, "step": 44473 }, { "epoch": 3.506030744974379, "grad_norm": 0.5469413995742798, "learning_rate": 0.0005, "loss": 1.2659128904342651, "step": 44474 }, { "epoch": 3.506109578242018, "grad_norm": 0.47416824102401733, "learning_rate": 0.0005, "loss": 1.3125405311584473, "step": 44475 }, { "epoch": 3.5061884115096573, "grad_norm": 0.41306421160697937, "learning_rate": 0.0005, "loss": 1.2019355297088623, "step": 44476 }, { "epoch": 3.506267244777296, "grad_norm": 0.4384831190109253, "learning_rate": 0.0005, "loss": 1.202483892440796, "step": 44477 }, { "epoch": 3.5063460780449347, "grad_norm": 0.4336482584476471, "learning_rate": 0.0005, "loss": 1.2021746635437012, "step": 44478 }, { "epoch": 3.506424911312574, "grad_norm": 0.4349011480808258, "learning_rate": 0.0005, "loss": 1.2192713022232056, "step": 44479 }, { "epoch": 3.506503744580213, "grad_norm": 0.39862579107284546, "learning_rate": 0.0005, "loss": 1.1837360858917236, "step": 44480 }, { "epoch": 3.5065825778478517, "grad_norm": 0.49039772152900696, "learning_rate": 0.0005, "loss": 1.1793396472930908, "step": 44481 }, { "epoch": 3.506661411115491, "grad_norm": 0.4893423318862915, "learning_rate": 0.0005, "loss": 1.3072112798690796, "step": 44482 }, { "epoch": 3.5067402443831295, "grad_norm": 0.4656248390674591, "learning_rate": 0.0005, "loss": 1.2550933361053467, "step": 44483 }, { "epoch": 3.5068190776507686, "grad_norm": 0.43282589316368103, "learning_rate": 0.0005, "loss": 1.1998934745788574, "step": 44484 }, { "epoch": 3.5068979109184077, "grad_norm": 0.5809460878372192, "learning_rate": 0.0005, "loss": 1.2454357147216797, "step": 44485 }, { "epoch": 3.5069767441860464, "grad_norm": 0.571927547454834, "learning_rate": 0.0005, "loss": 1.328035831451416, "step": 44486 }, { "epoch": 3.5070555774536856, "grad_norm": 0.6622436046600342, "learning_rate": 0.0005, "loss": 1.4106366634368896, "step": 44487 }, { "epoch": 3.5071344107213243, "grad_norm": 0.6647752523422241, "learning_rate": 0.0005, "loss": 1.1976547241210938, "step": 44488 }, { "epoch": 3.5072132439889634, "grad_norm": 0.5101026296615601, "learning_rate": 0.0005, "loss": 1.1438226699829102, "step": 44489 }, { "epoch": 3.5072920772566025, "grad_norm": 0.5474787354469299, "learning_rate": 0.0005, "loss": 1.2380266189575195, "step": 44490 }, { "epoch": 3.507370910524241, "grad_norm": 0.4831570088863373, "learning_rate": 0.0005, "loss": 1.225003957748413, "step": 44491 }, { "epoch": 3.50744974379188, "grad_norm": 0.47814440727233887, "learning_rate": 0.0005, "loss": 1.249830722808838, "step": 44492 }, { "epoch": 3.507528577059519, "grad_norm": 0.5141348242759705, "learning_rate": 0.0005, "loss": 1.3746497631072998, "step": 44493 }, { "epoch": 3.507607410327158, "grad_norm": 0.4508725106716156, "learning_rate": 0.0005, "loss": 1.108344554901123, "step": 44494 }, { "epoch": 3.507686243594797, "grad_norm": 0.4855518043041229, "learning_rate": 0.0005, "loss": 1.2160428762435913, "step": 44495 }, { "epoch": 3.507765076862436, "grad_norm": 0.4537811875343323, "learning_rate": 0.0005, "loss": 1.1950554847717285, "step": 44496 }, { "epoch": 3.5078439101300747, "grad_norm": 0.5161044597625732, "learning_rate": 0.0005, "loss": 1.472007393836975, "step": 44497 }, { "epoch": 3.507922743397714, "grad_norm": 0.471700519323349, "learning_rate": 0.0005, "loss": 1.2936922311782837, "step": 44498 }, { "epoch": 3.508001576665353, "grad_norm": 0.403438001871109, "learning_rate": 0.0005, "loss": 1.188244342803955, "step": 44499 }, { "epoch": 3.5080804099329916, "grad_norm": 0.4481945037841797, "learning_rate": 0.0005, "loss": 1.2652127742767334, "step": 44500 }, { "epoch": 3.5081592432006308, "grad_norm": 0.5368395447731018, "learning_rate": 0.0005, "loss": 1.282059907913208, "step": 44501 }, { "epoch": 3.5082380764682695, "grad_norm": 0.5056275129318237, "learning_rate": 0.0005, "loss": 1.438126564025879, "step": 44502 }, { "epoch": 3.5083169097359086, "grad_norm": 0.47441837191581726, "learning_rate": 0.0005, "loss": 1.044216275215149, "step": 44503 }, { "epoch": 3.5083957430035477, "grad_norm": 0.47334378957748413, "learning_rate": 0.0005, "loss": 1.200361967086792, "step": 44504 }, { "epoch": 3.5084745762711864, "grad_norm": 0.4810355603694916, "learning_rate": 0.0005, "loss": 1.3549671173095703, "step": 44505 }, { "epoch": 3.508553409538825, "grad_norm": 0.598784327507019, "learning_rate": 0.0005, "loss": 1.2366666793823242, "step": 44506 }, { "epoch": 3.5086322428064642, "grad_norm": 0.4922724664211273, "learning_rate": 0.0005, "loss": 1.2012605667114258, "step": 44507 }, { "epoch": 3.5087110760741034, "grad_norm": 0.4337288737297058, "learning_rate": 0.0005, "loss": 1.1044647693634033, "step": 44508 }, { "epoch": 3.508789909341742, "grad_norm": 0.4480409324169159, "learning_rate": 0.0005, "loss": 1.3512741327285767, "step": 44509 }, { "epoch": 3.508868742609381, "grad_norm": 0.5149686336517334, "learning_rate": 0.0005, "loss": 1.2651952505111694, "step": 44510 }, { "epoch": 3.50894757587702, "grad_norm": 0.40681156516075134, "learning_rate": 0.0005, "loss": 1.208730936050415, "step": 44511 }, { "epoch": 3.509026409144659, "grad_norm": 0.6724463701248169, "learning_rate": 0.0005, "loss": 1.3401820659637451, "step": 44512 }, { "epoch": 3.509105242412298, "grad_norm": 0.46983861923217773, "learning_rate": 0.0005, "loss": 1.3260256052017212, "step": 44513 }, { "epoch": 3.509184075679937, "grad_norm": 0.4300062656402588, "learning_rate": 0.0005, "loss": 1.1389305591583252, "step": 44514 }, { "epoch": 3.509262908947576, "grad_norm": 0.4757913053035736, "learning_rate": 0.0005, "loss": 1.2854666709899902, "step": 44515 }, { "epoch": 3.5093417422152147, "grad_norm": 0.4606381952762604, "learning_rate": 0.0005, "loss": 1.0709258317947388, "step": 44516 }, { "epoch": 3.509420575482854, "grad_norm": 0.46382978558540344, "learning_rate": 0.0005, "loss": 1.2617676258087158, "step": 44517 }, { "epoch": 3.509499408750493, "grad_norm": 0.4583255648612976, "learning_rate": 0.0005, "loss": 1.2477964162826538, "step": 44518 }, { "epoch": 3.5095782420181316, "grad_norm": 0.4947040379047394, "learning_rate": 0.0005, "loss": 1.3281773328781128, "step": 44519 }, { "epoch": 3.5096570752857708, "grad_norm": 0.3868601322174072, "learning_rate": 0.0005, "loss": 1.054732322692871, "step": 44520 }, { "epoch": 3.5097359085534094, "grad_norm": 0.43666473031044006, "learning_rate": 0.0005, "loss": 1.2765791416168213, "step": 44521 }, { "epoch": 3.5098147418210486, "grad_norm": 0.4963330328464508, "learning_rate": 0.0005, "loss": 1.2801194190979004, "step": 44522 }, { "epoch": 3.5098935750886873, "grad_norm": 0.600666344165802, "learning_rate": 0.0005, "loss": 1.382859230041504, "step": 44523 }, { "epoch": 3.5099724083563264, "grad_norm": 0.4208693206310272, "learning_rate": 0.0005, "loss": 1.3364696502685547, "step": 44524 }, { "epoch": 3.510051241623965, "grad_norm": 0.43610885739326477, "learning_rate": 0.0005, "loss": 1.241985559463501, "step": 44525 }, { "epoch": 3.5101300748916042, "grad_norm": 0.4389197826385498, "learning_rate": 0.0005, "loss": 1.1882717609405518, "step": 44526 }, { "epoch": 3.5102089081592434, "grad_norm": 0.5684213638305664, "learning_rate": 0.0005, "loss": 1.1945966482162476, "step": 44527 }, { "epoch": 3.510287741426882, "grad_norm": 0.5256818532943726, "learning_rate": 0.0005, "loss": 1.2047088146209717, "step": 44528 }, { "epoch": 3.510366574694521, "grad_norm": 0.4133375287055969, "learning_rate": 0.0005, "loss": 1.1052968502044678, "step": 44529 }, { "epoch": 3.51044540796216, "grad_norm": 0.5504164695739746, "learning_rate": 0.0005, "loss": 1.3933149576187134, "step": 44530 }, { "epoch": 3.510524241229799, "grad_norm": 0.48379647731781006, "learning_rate": 0.0005, "loss": 1.270625114440918, "step": 44531 }, { "epoch": 3.510603074497438, "grad_norm": 0.570371150970459, "learning_rate": 0.0005, "loss": 1.0711450576782227, "step": 44532 }, { "epoch": 3.510681907765077, "grad_norm": 0.5104467868804932, "learning_rate": 0.0005, "loss": 1.2901376485824585, "step": 44533 }, { "epoch": 3.510760741032716, "grad_norm": 0.47405239939689636, "learning_rate": 0.0005, "loss": 1.3934814929962158, "step": 44534 }, { "epoch": 3.5108395743003546, "grad_norm": 0.569187581539154, "learning_rate": 0.0005, "loss": 1.3467259407043457, "step": 44535 }, { "epoch": 3.510918407567994, "grad_norm": 0.5293806195259094, "learning_rate": 0.0005, "loss": 1.4407916069030762, "step": 44536 }, { "epoch": 3.5109972408356325, "grad_norm": 0.5129858255386353, "learning_rate": 0.0005, "loss": 1.0900633335113525, "step": 44537 }, { "epoch": 3.5110760741032716, "grad_norm": 0.42674094438552856, "learning_rate": 0.0005, "loss": 1.1114037036895752, "step": 44538 }, { "epoch": 3.5111549073709103, "grad_norm": 0.41174766421318054, "learning_rate": 0.0005, "loss": 1.2093229293823242, "step": 44539 }, { "epoch": 3.5112337406385494, "grad_norm": 0.5391333103179932, "learning_rate": 0.0005, "loss": 1.3108067512512207, "step": 44540 }, { "epoch": 3.5113125739061886, "grad_norm": 0.6035560369491577, "learning_rate": 0.0005, "loss": 1.3948925733566284, "step": 44541 }, { "epoch": 3.5113914071738273, "grad_norm": 0.45651984214782715, "learning_rate": 0.0005, "loss": 1.073955774307251, "step": 44542 }, { "epoch": 3.5114702404414664, "grad_norm": 0.5012787580490112, "learning_rate": 0.0005, "loss": 1.0892555713653564, "step": 44543 }, { "epoch": 3.511549073709105, "grad_norm": 0.4555283188819885, "learning_rate": 0.0005, "loss": 1.0736620426177979, "step": 44544 }, { "epoch": 3.511627906976744, "grad_norm": 0.5066616535186768, "learning_rate": 0.0005, "loss": 1.3810515403747559, "step": 44545 }, { "epoch": 3.5117067402443833, "grad_norm": 0.46648716926574707, "learning_rate": 0.0005, "loss": 1.254774808883667, "step": 44546 }, { "epoch": 3.511785573512022, "grad_norm": 0.517118513584137, "learning_rate": 0.0005, "loss": 1.1523655652999878, "step": 44547 }, { "epoch": 3.511864406779661, "grad_norm": 0.6031858325004578, "learning_rate": 0.0005, "loss": 1.2868834733963013, "step": 44548 }, { "epoch": 3.5119432400473, "grad_norm": 0.4227268695831299, "learning_rate": 0.0005, "loss": 1.3290815353393555, "step": 44549 }, { "epoch": 3.512022073314939, "grad_norm": 0.4667085111141205, "learning_rate": 0.0005, "loss": 1.3424339294433594, "step": 44550 }, { "epoch": 3.512100906582578, "grad_norm": 0.4841342866420746, "learning_rate": 0.0005, "loss": 1.1739575862884521, "step": 44551 }, { "epoch": 3.512179739850217, "grad_norm": 0.4542592465877533, "learning_rate": 0.0005, "loss": 1.3462181091308594, "step": 44552 }, { "epoch": 3.5122585731178555, "grad_norm": 0.5094274282455444, "learning_rate": 0.0005, "loss": 1.2188200950622559, "step": 44553 }, { "epoch": 3.5123374063854946, "grad_norm": 0.5446190237998962, "learning_rate": 0.0005, "loss": 1.3390843868255615, "step": 44554 }, { "epoch": 3.5124162396531338, "grad_norm": 0.47910410165786743, "learning_rate": 0.0005, "loss": 1.1576216220855713, "step": 44555 }, { "epoch": 3.5124950729207725, "grad_norm": 0.5891231894493103, "learning_rate": 0.0005, "loss": 1.3138196468353271, "step": 44556 }, { "epoch": 3.5125739061884116, "grad_norm": 0.4836053252220154, "learning_rate": 0.0005, "loss": 1.2917354106903076, "step": 44557 }, { "epoch": 3.5126527394560503, "grad_norm": 0.48472103476524353, "learning_rate": 0.0005, "loss": 1.3917236328125, "step": 44558 }, { "epoch": 3.5127315727236894, "grad_norm": 0.44533205032348633, "learning_rate": 0.0005, "loss": 1.1638333797454834, "step": 44559 }, { "epoch": 3.5128104059913285, "grad_norm": 0.5009151697158813, "learning_rate": 0.0005, "loss": 1.1956770420074463, "step": 44560 }, { "epoch": 3.5128892392589672, "grad_norm": 0.434909462928772, "learning_rate": 0.0005, "loss": 1.2815862894058228, "step": 44561 }, { "epoch": 3.5129680725266064, "grad_norm": 0.6837286949157715, "learning_rate": 0.0005, "loss": 1.396215558052063, "step": 44562 }, { "epoch": 3.513046905794245, "grad_norm": 0.8756442666053772, "learning_rate": 0.0005, "loss": 1.1857256889343262, "step": 44563 }, { "epoch": 3.513125739061884, "grad_norm": 0.5677109956741333, "learning_rate": 0.0005, "loss": 1.5228912830352783, "step": 44564 }, { "epoch": 3.5132045723295233, "grad_norm": 0.4646221101284027, "learning_rate": 0.0005, "loss": 1.2456107139587402, "step": 44565 }, { "epoch": 3.513283405597162, "grad_norm": 0.5264571905136108, "learning_rate": 0.0005, "loss": 1.037970781326294, "step": 44566 }, { "epoch": 3.5133622388648007, "grad_norm": 0.5526679158210754, "learning_rate": 0.0005, "loss": 1.3633100986480713, "step": 44567 }, { "epoch": 3.51344107213244, "grad_norm": 0.6080155372619629, "learning_rate": 0.0005, "loss": 1.4066083431243896, "step": 44568 }, { "epoch": 3.513519905400079, "grad_norm": 0.4820728600025177, "learning_rate": 0.0005, "loss": 1.3412396907806396, "step": 44569 }, { "epoch": 3.5135987386677177, "grad_norm": 0.6138069033622742, "learning_rate": 0.0005, "loss": 1.2834866046905518, "step": 44570 }, { "epoch": 3.513677571935357, "grad_norm": 0.4165630042552948, "learning_rate": 0.0005, "loss": 1.312591552734375, "step": 44571 }, { "epoch": 3.5137564052029955, "grad_norm": 0.5244012475013733, "learning_rate": 0.0005, "loss": 1.3134807348251343, "step": 44572 }, { "epoch": 3.5138352384706346, "grad_norm": 0.4418822228908539, "learning_rate": 0.0005, "loss": 1.173351526260376, "step": 44573 }, { "epoch": 3.5139140717382737, "grad_norm": 0.4179478883743286, "learning_rate": 0.0005, "loss": 1.3715755939483643, "step": 44574 }, { "epoch": 3.5139929050059124, "grad_norm": 0.449228972196579, "learning_rate": 0.0005, "loss": 1.451535701751709, "step": 44575 }, { "epoch": 3.5140717382735516, "grad_norm": 0.5131787657737732, "learning_rate": 0.0005, "loss": 1.263366937637329, "step": 44576 }, { "epoch": 3.5141505715411903, "grad_norm": 0.5265375971794128, "learning_rate": 0.0005, "loss": 1.3123809099197388, "step": 44577 }, { "epoch": 3.5142294048088294, "grad_norm": 0.4924761950969696, "learning_rate": 0.0005, "loss": 1.2697372436523438, "step": 44578 }, { "epoch": 3.5143082380764685, "grad_norm": 0.5362016558647156, "learning_rate": 0.0005, "loss": 1.221357822418213, "step": 44579 }, { "epoch": 3.514387071344107, "grad_norm": 0.49667102098464966, "learning_rate": 0.0005, "loss": 1.3940024375915527, "step": 44580 }, { "epoch": 3.514465904611746, "grad_norm": 0.5286452174186707, "learning_rate": 0.0005, "loss": 1.1071226596832275, "step": 44581 }, { "epoch": 3.514544737879385, "grad_norm": 0.47722092270851135, "learning_rate": 0.0005, "loss": 1.2492018938064575, "step": 44582 }, { "epoch": 3.514623571147024, "grad_norm": 0.5712803602218628, "learning_rate": 0.0005, "loss": 1.3529739379882812, "step": 44583 }, { "epoch": 3.514702404414663, "grad_norm": 0.46686381101608276, "learning_rate": 0.0005, "loss": 1.1685547828674316, "step": 44584 }, { "epoch": 3.514781237682302, "grad_norm": 0.44820114970207214, "learning_rate": 0.0005, "loss": 1.1954965591430664, "step": 44585 }, { "epoch": 3.5148600709499407, "grad_norm": 0.5887491703033447, "learning_rate": 0.0005, "loss": 1.1810435056686401, "step": 44586 }, { "epoch": 3.51493890421758, "grad_norm": 0.5165965557098389, "learning_rate": 0.0005, "loss": 1.1440486907958984, "step": 44587 }, { "epoch": 3.515017737485219, "grad_norm": 0.5039072036743164, "learning_rate": 0.0005, "loss": 1.4005005359649658, "step": 44588 }, { "epoch": 3.5150965707528576, "grad_norm": 0.5062404274940491, "learning_rate": 0.0005, "loss": 1.146073579788208, "step": 44589 }, { "epoch": 3.5151754040204968, "grad_norm": 0.5143190622329712, "learning_rate": 0.0005, "loss": 1.3537052869796753, "step": 44590 }, { "epoch": 3.5152542372881355, "grad_norm": 0.44325074553489685, "learning_rate": 0.0005, "loss": 1.1719415187835693, "step": 44591 }, { "epoch": 3.5153330705557746, "grad_norm": 0.5289230346679688, "learning_rate": 0.0005, "loss": 1.2065019607543945, "step": 44592 }, { "epoch": 3.5154119038234137, "grad_norm": 0.4619726538658142, "learning_rate": 0.0005, "loss": 1.170379877090454, "step": 44593 }, { "epoch": 3.5154907370910524, "grad_norm": 0.46308931708335876, "learning_rate": 0.0005, "loss": 1.319284200668335, "step": 44594 }, { "epoch": 3.515569570358691, "grad_norm": 0.5691713094711304, "learning_rate": 0.0005, "loss": 1.3971590995788574, "step": 44595 }, { "epoch": 3.5156484036263302, "grad_norm": 0.5871337652206421, "learning_rate": 0.0005, "loss": 1.2711460590362549, "step": 44596 }, { "epoch": 3.5157272368939694, "grad_norm": 0.4325733482837677, "learning_rate": 0.0005, "loss": 1.111464500427246, "step": 44597 }, { "epoch": 3.515806070161608, "grad_norm": 0.5174906849861145, "learning_rate": 0.0005, "loss": 1.2931647300720215, "step": 44598 }, { "epoch": 3.515884903429247, "grad_norm": 0.49246329069137573, "learning_rate": 0.0005, "loss": 1.2422199249267578, "step": 44599 }, { "epoch": 3.515963736696886, "grad_norm": 0.44870585203170776, "learning_rate": 0.0005, "loss": 1.1415804624557495, "step": 44600 }, { "epoch": 3.516042569964525, "grad_norm": 0.46918851137161255, "learning_rate": 0.0005, "loss": 1.3550875186920166, "step": 44601 }, { "epoch": 3.516121403232164, "grad_norm": 0.999474287033081, "learning_rate": 0.0005, "loss": 1.1787136793136597, "step": 44602 }, { "epoch": 3.516200236499803, "grad_norm": 0.5490060448646545, "learning_rate": 0.0005, "loss": 1.4587221145629883, "step": 44603 }, { "epoch": 3.516279069767442, "grad_norm": 0.47591859102249146, "learning_rate": 0.0005, "loss": 1.4133214950561523, "step": 44604 }, { "epoch": 3.5163579030350807, "grad_norm": 0.5196912288665771, "learning_rate": 0.0005, "loss": 1.2586770057678223, "step": 44605 }, { "epoch": 3.51643673630272, "grad_norm": 0.5308253765106201, "learning_rate": 0.0005, "loss": 1.2574470043182373, "step": 44606 }, { "epoch": 3.516515569570359, "grad_norm": 0.48938795924186707, "learning_rate": 0.0005, "loss": 1.2901859283447266, "step": 44607 }, { "epoch": 3.5165944028379976, "grad_norm": 0.5728534460067749, "learning_rate": 0.0005, "loss": 1.3623592853546143, "step": 44608 }, { "epoch": 3.5166732361056363, "grad_norm": 0.46415913105010986, "learning_rate": 0.0005, "loss": 1.1579575538635254, "step": 44609 }, { "epoch": 3.5167520693732754, "grad_norm": 0.5772460699081421, "learning_rate": 0.0005, "loss": 1.3362452983856201, "step": 44610 }, { "epoch": 3.5168309026409146, "grad_norm": 0.42771974205970764, "learning_rate": 0.0005, "loss": 1.183853268623352, "step": 44611 }, { "epoch": 3.5169097359085533, "grad_norm": 0.46870511770248413, "learning_rate": 0.0005, "loss": 1.0779061317443848, "step": 44612 }, { "epoch": 3.5169885691761924, "grad_norm": 0.5467145442962646, "learning_rate": 0.0005, "loss": 1.1798179149627686, "step": 44613 }, { "epoch": 3.517067402443831, "grad_norm": 0.9553254842758179, "learning_rate": 0.0005, "loss": 1.3718948364257812, "step": 44614 }, { "epoch": 3.5171462357114702, "grad_norm": 0.46684983372688293, "learning_rate": 0.0005, "loss": 1.2542357444763184, "step": 44615 }, { "epoch": 3.5172250689791094, "grad_norm": 0.5333155989646912, "learning_rate": 0.0005, "loss": 1.3881529569625854, "step": 44616 }, { "epoch": 3.517303902246748, "grad_norm": 0.4752059578895569, "learning_rate": 0.0005, "loss": 1.1304643154144287, "step": 44617 }, { "epoch": 3.517382735514387, "grad_norm": 0.5274489521980286, "learning_rate": 0.0005, "loss": 1.2504687309265137, "step": 44618 }, { "epoch": 3.517461568782026, "grad_norm": 0.5618811249732971, "learning_rate": 0.0005, "loss": 1.2274565696716309, "step": 44619 }, { "epoch": 3.517540402049665, "grad_norm": 0.7106732726097107, "learning_rate": 0.0005, "loss": 1.567183256149292, "step": 44620 }, { "epoch": 3.517619235317304, "grad_norm": 0.48427167534828186, "learning_rate": 0.0005, "loss": 1.31540846824646, "step": 44621 }, { "epoch": 3.517698068584943, "grad_norm": 0.7803435921669006, "learning_rate": 0.0005, "loss": 1.3462131023406982, "step": 44622 }, { "epoch": 3.5177769018525815, "grad_norm": 0.6479778289794922, "learning_rate": 0.0005, "loss": 1.4637013673782349, "step": 44623 }, { "epoch": 3.5178557351202207, "grad_norm": 0.4999139904975891, "learning_rate": 0.0005, "loss": 1.2745516300201416, "step": 44624 }, { "epoch": 3.51793456838786, "grad_norm": 0.6967765092849731, "learning_rate": 0.0005, "loss": 1.3541128635406494, "step": 44625 }, { "epoch": 3.5180134016554985, "grad_norm": 0.43157631158828735, "learning_rate": 0.0005, "loss": 1.1802377700805664, "step": 44626 }, { "epoch": 3.5180922349231376, "grad_norm": 0.49243032932281494, "learning_rate": 0.0005, "loss": 1.2763266563415527, "step": 44627 }, { "epoch": 3.5181710681907763, "grad_norm": 0.44331657886505127, "learning_rate": 0.0005, "loss": 1.205341100692749, "step": 44628 }, { "epoch": 3.5182499014584154, "grad_norm": 0.47096171975135803, "learning_rate": 0.0005, "loss": 1.2826714515686035, "step": 44629 }, { "epoch": 3.5183287347260546, "grad_norm": 0.3963433802127838, "learning_rate": 0.0005, "loss": 1.1612293720245361, "step": 44630 }, { "epoch": 3.5184075679936933, "grad_norm": 0.549405038356781, "learning_rate": 0.0005, "loss": 1.2375752925872803, "step": 44631 }, { "epoch": 3.5184864012613324, "grad_norm": 0.7074840068817139, "learning_rate": 0.0005, "loss": 1.4883959293365479, "step": 44632 }, { "epoch": 3.518565234528971, "grad_norm": 0.46650564670562744, "learning_rate": 0.0005, "loss": 1.3550965785980225, "step": 44633 }, { "epoch": 3.51864406779661, "grad_norm": 0.5475813150405884, "learning_rate": 0.0005, "loss": 1.3886470794677734, "step": 44634 }, { "epoch": 3.5187229010642493, "grad_norm": 0.5837796330451965, "learning_rate": 0.0005, "loss": 1.5126335620880127, "step": 44635 }, { "epoch": 3.518801734331888, "grad_norm": 0.5735544562339783, "learning_rate": 0.0005, "loss": 1.3260403871536255, "step": 44636 }, { "epoch": 3.518880567599527, "grad_norm": 0.5058629512786865, "learning_rate": 0.0005, "loss": 1.4079771041870117, "step": 44637 }, { "epoch": 3.518959400867166, "grad_norm": 0.4807615280151367, "learning_rate": 0.0005, "loss": 1.3015539646148682, "step": 44638 }, { "epoch": 3.519038234134805, "grad_norm": 0.47198110818862915, "learning_rate": 0.0005, "loss": 1.363797903060913, "step": 44639 }, { "epoch": 3.5191170674024437, "grad_norm": 0.427362322807312, "learning_rate": 0.0005, "loss": 1.1739001274108887, "step": 44640 }, { "epoch": 3.519195900670083, "grad_norm": 0.48680925369262695, "learning_rate": 0.0005, "loss": 1.2576026916503906, "step": 44641 }, { "epoch": 3.5192747339377215, "grad_norm": 0.48554372787475586, "learning_rate": 0.0005, "loss": 1.3281655311584473, "step": 44642 }, { "epoch": 3.5193535672053606, "grad_norm": 0.597663402557373, "learning_rate": 0.0005, "loss": 1.2363660335540771, "step": 44643 }, { "epoch": 3.5194324004729998, "grad_norm": 0.4681699275970459, "learning_rate": 0.0005, "loss": 1.3757851123809814, "step": 44644 }, { "epoch": 3.5195112337406385, "grad_norm": 0.6000415682792664, "learning_rate": 0.0005, "loss": 1.2420334815979004, "step": 44645 }, { "epoch": 3.5195900670082776, "grad_norm": 0.47809427976608276, "learning_rate": 0.0005, "loss": 1.223360300064087, "step": 44646 }, { "epoch": 3.5196689002759163, "grad_norm": 3.5893712043762207, "learning_rate": 0.0005, "loss": 1.1117316484451294, "step": 44647 }, { "epoch": 3.5197477335435554, "grad_norm": 0.5640903115272522, "learning_rate": 0.0005, "loss": 1.2347702980041504, "step": 44648 }, { "epoch": 3.5198265668111945, "grad_norm": 0.45574161410331726, "learning_rate": 0.0005, "loss": 1.2065346240997314, "step": 44649 }, { "epoch": 3.5199054000788332, "grad_norm": 0.5368057489395142, "learning_rate": 0.0005, "loss": 1.390061616897583, "step": 44650 }, { "epoch": 3.5199842333464724, "grad_norm": 0.5461975932121277, "learning_rate": 0.0005, "loss": 1.4230592250823975, "step": 44651 }, { "epoch": 3.520063066614111, "grad_norm": 0.49190065264701843, "learning_rate": 0.0005, "loss": 1.3170640468597412, "step": 44652 }, { "epoch": 3.52014189988175, "grad_norm": 0.5077310800552368, "learning_rate": 0.0005, "loss": 1.3440731763839722, "step": 44653 }, { "epoch": 3.520220733149389, "grad_norm": 0.6509234309196472, "learning_rate": 0.0005, "loss": 1.2398072481155396, "step": 44654 }, { "epoch": 3.520299566417028, "grad_norm": 0.44564276933670044, "learning_rate": 0.0005, "loss": 1.2247017621994019, "step": 44655 }, { "epoch": 3.5203783996846667, "grad_norm": 0.50083988904953, "learning_rate": 0.0005, "loss": 1.2830486297607422, "step": 44656 }, { "epoch": 3.520457232952306, "grad_norm": 0.46616411209106445, "learning_rate": 0.0005, "loss": 1.2680991888046265, "step": 44657 }, { "epoch": 3.520536066219945, "grad_norm": 0.4003850519657135, "learning_rate": 0.0005, "loss": 1.087880253791809, "step": 44658 }, { "epoch": 3.5206148994875837, "grad_norm": 0.4909180998802185, "learning_rate": 0.0005, "loss": 1.1810816526412964, "step": 44659 }, { "epoch": 3.520693732755223, "grad_norm": 0.494884192943573, "learning_rate": 0.0005, "loss": 1.4118307828903198, "step": 44660 }, { "epoch": 3.5207725660228615, "grad_norm": 0.539596676826477, "learning_rate": 0.0005, "loss": 1.3230341672897339, "step": 44661 }, { "epoch": 3.5208513992905006, "grad_norm": 0.5630778074264526, "learning_rate": 0.0005, "loss": 1.3020387887954712, "step": 44662 }, { "epoch": 3.5209302325581397, "grad_norm": 0.41993412375450134, "learning_rate": 0.0005, "loss": 1.1902514696121216, "step": 44663 }, { "epoch": 3.5210090658257784, "grad_norm": 0.41217780113220215, "learning_rate": 0.0005, "loss": 1.1472342014312744, "step": 44664 }, { "epoch": 3.5210878990934176, "grad_norm": 0.3538380265235901, "learning_rate": 0.0005, "loss": 1.0978729724884033, "step": 44665 }, { "epoch": 3.5211667323610563, "grad_norm": 0.576420783996582, "learning_rate": 0.0005, "loss": 1.3889460563659668, "step": 44666 }, { "epoch": 3.5212455656286954, "grad_norm": 0.5066046118736267, "learning_rate": 0.0005, "loss": 1.3463810682296753, "step": 44667 }, { "epoch": 3.5213243988963345, "grad_norm": 0.5711910724639893, "learning_rate": 0.0005, "loss": 1.3213952779769897, "step": 44668 }, { "epoch": 3.521403232163973, "grad_norm": 0.48388150334358215, "learning_rate": 0.0005, "loss": 1.115187168121338, "step": 44669 }, { "epoch": 3.521482065431612, "grad_norm": 0.5125018358230591, "learning_rate": 0.0005, "loss": 1.3096930980682373, "step": 44670 }, { "epoch": 3.521560898699251, "grad_norm": 0.5345889329910278, "learning_rate": 0.0005, "loss": 1.2045788764953613, "step": 44671 }, { "epoch": 3.52163973196689, "grad_norm": 0.49777790904045105, "learning_rate": 0.0005, "loss": 1.4975146055221558, "step": 44672 }, { "epoch": 3.521718565234529, "grad_norm": 0.4799497723579407, "learning_rate": 0.0005, "loss": 1.4069862365722656, "step": 44673 }, { "epoch": 3.521797398502168, "grad_norm": 0.46182939410209656, "learning_rate": 0.0005, "loss": 1.1257784366607666, "step": 44674 }, { "epoch": 3.5218762317698067, "grad_norm": 0.4597342610359192, "learning_rate": 0.0005, "loss": 1.1868400573730469, "step": 44675 }, { "epoch": 3.521955065037446, "grad_norm": 0.5210580229759216, "learning_rate": 0.0005, "loss": 1.3130059242248535, "step": 44676 }, { "epoch": 3.522033898305085, "grad_norm": 0.3851243853569031, "learning_rate": 0.0005, "loss": 1.02614164352417, "step": 44677 }, { "epoch": 3.5221127315727236, "grad_norm": 0.4804248809814453, "learning_rate": 0.0005, "loss": 1.2264227867126465, "step": 44678 }, { "epoch": 3.5221915648403628, "grad_norm": 0.42444488406181335, "learning_rate": 0.0005, "loss": 1.2005445957183838, "step": 44679 }, { "epoch": 3.5222703981080015, "grad_norm": 0.42800021171569824, "learning_rate": 0.0005, "loss": 1.2007108926773071, "step": 44680 }, { "epoch": 3.5223492313756406, "grad_norm": 0.38158586621284485, "learning_rate": 0.0005, "loss": 1.186763048171997, "step": 44681 }, { "epoch": 3.5224280646432797, "grad_norm": 0.4160725772380829, "learning_rate": 0.0005, "loss": 1.3073959350585938, "step": 44682 }, { "epoch": 3.5225068979109184, "grad_norm": 0.5108923316001892, "learning_rate": 0.0005, "loss": 1.1878247261047363, "step": 44683 }, { "epoch": 3.522585731178557, "grad_norm": 0.44607698917388916, "learning_rate": 0.0005, "loss": 1.1295051574707031, "step": 44684 }, { "epoch": 3.5226645644461962, "grad_norm": 0.47132325172424316, "learning_rate": 0.0005, "loss": 1.414888858795166, "step": 44685 }, { "epoch": 3.5227433977138354, "grad_norm": 0.4639466404914856, "learning_rate": 0.0005, "loss": 1.2367377281188965, "step": 44686 }, { "epoch": 3.522822230981474, "grad_norm": 0.5205345153808594, "learning_rate": 0.0005, "loss": 1.387939214706421, "step": 44687 }, { "epoch": 3.522901064249113, "grad_norm": 0.6155309677124023, "learning_rate": 0.0005, "loss": 1.184596300125122, "step": 44688 }, { "epoch": 3.522979897516752, "grad_norm": 0.45072194933891296, "learning_rate": 0.0005, "loss": 1.1619993448257446, "step": 44689 }, { "epoch": 3.523058730784391, "grad_norm": 0.48596513271331787, "learning_rate": 0.0005, "loss": 1.1121609210968018, "step": 44690 }, { "epoch": 3.52313756405203, "grad_norm": 0.4962114095687866, "learning_rate": 0.0005, "loss": 1.240014672279358, "step": 44691 }, { "epoch": 3.523216397319669, "grad_norm": 0.7536728382110596, "learning_rate": 0.0005, "loss": 1.2241604328155518, "step": 44692 }, { "epoch": 3.523295230587308, "grad_norm": 0.5514382123947144, "learning_rate": 0.0005, "loss": 1.4453847408294678, "step": 44693 }, { "epoch": 3.5233740638549467, "grad_norm": 0.4577955901622772, "learning_rate": 0.0005, "loss": 1.246556043624878, "step": 44694 }, { "epoch": 3.523452897122586, "grad_norm": 0.43155714869499207, "learning_rate": 0.0005, "loss": 1.041677474975586, "step": 44695 }, { "epoch": 3.523531730390225, "grad_norm": 0.47709420323371887, "learning_rate": 0.0005, "loss": 1.2606242895126343, "step": 44696 }, { "epoch": 3.5236105636578636, "grad_norm": 0.5002651810646057, "learning_rate": 0.0005, "loss": 1.3325345516204834, "step": 44697 }, { "epoch": 3.5236893969255023, "grad_norm": 0.45821547508239746, "learning_rate": 0.0005, "loss": 1.2073512077331543, "step": 44698 }, { "epoch": 3.5237682301931414, "grad_norm": 0.5126950740814209, "learning_rate": 0.0005, "loss": 1.5088424682617188, "step": 44699 }, { "epoch": 3.5238470634607806, "grad_norm": 0.5076852440834045, "learning_rate": 0.0005, "loss": 1.2262078523635864, "step": 44700 }, { "epoch": 3.5239258967284193, "grad_norm": 0.5218154191970825, "learning_rate": 0.0005, "loss": 1.38035249710083, "step": 44701 }, { "epoch": 3.5240047299960584, "grad_norm": 0.4409809410572052, "learning_rate": 0.0005, "loss": 1.1706682443618774, "step": 44702 }, { "epoch": 3.524083563263697, "grad_norm": 0.5339738130569458, "learning_rate": 0.0005, "loss": 1.2443925142288208, "step": 44703 }, { "epoch": 3.5241623965313362, "grad_norm": 0.5549371242523193, "learning_rate": 0.0005, "loss": 1.267279863357544, "step": 44704 }, { "epoch": 3.5242412297989754, "grad_norm": 0.4795302748680115, "learning_rate": 0.0005, "loss": 1.27614426612854, "step": 44705 }, { "epoch": 3.524320063066614, "grad_norm": 0.5416979789733887, "learning_rate": 0.0005, "loss": 1.1537734270095825, "step": 44706 }, { "epoch": 3.524398896334253, "grad_norm": 0.5610831379890442, "learning_rate": 0.0005, "loss": 1.2353858947753906, "step": 44707 }, { "epoch": 3.524477729601892, "grad_norm": 0.7010200023651123, "learning_rate": 0.0005, "loss": 1.2330458164215088, "step": 44708 }, { "epoch": 3.524556562869531, "grad_norm": 0.6669092774391174, "learning_rate": 0.0005, "loss": 1.335146188735962, "step": 44709 }, { "epoch": 3.52463539613717, "grad_norm": 0.522818386554718, "learning_rate": 0.0005, "loss": 1.2153856754302979, "step": 44710 }, { "epoch": 3.524714229404809, "grad_norm": 0.4779660701751709, "learning_rate": 0.0005, "loss": 1.3354731798171997, "step": 44711 }, { "epoch": 3.5247930626724475, "grad_norm": 0.4365154802799225, "learning_rate": 0.0005, "loss": 1.2609071731567383, "step": 44712 }, { "epoch": 3.5248718959400867, "grad_norm": 0.4070262908935547, "learning_rate": 0.0005, "loss": 1.1969817876815796, "step": 44713 }, { "epoch": 3.524950729207726, "grad_norm": 0.4493568241596222, "learning_rate": 0.0005, "loss": 1.1801998615264893, "step": 44714 }, { "epoch": 3.5250295624753645, "grad_norm": 0.4181433618068695, "learning_rate": 0.0005, "loss": 1.1562670469284058, "step": 44715 }, { "epoch": 3.5251083957430036, "grad_norm": 0.4735265374183655, "learning_rate": 0.0005, "loss": 1.1517380475997925, "step": 44716 }, { "epoch": 3.5251872290106423, "grad_norm": 0.4103820025920868, "learning_rate": 0.0005, "loss": 1.0594451427459717, "step": 44717 }, { "epoch": 3.5252660622782814, "grad_norm": 0.6050926446914673, "learning_rate": 0.0005, "loss": 1.4752479791641235, "step": 44718 }, { "epoch": 3.5253448955459206, "grad_norm": 0.6647223234176636, "learning_rate": 0.0005, "loss": 1.3571481704711914, "step": 44719 }, { "epoch": 3.5254237288135593, "grad_norm": 0.5404068231582642, "learning_rate": 0.0005, "loss": 1.404515266418457, "step": 44720 }, { "epoch": 3.5255025620811984, "grad_norm": 0.4696350395679474, "learning_rate": 0.0005, "loss": 1.0521271228790283, "step": 44721 }, { "epoch": 3.525581395348837, "grad_norm": 0.48208704590797424, "learning_rate": 0.0005, "loss": 1.1994786262512207, "step": 44722 }, { "epoch": 3.525660228616476, "grad_norm": 0.5184401273727417, "learning_rate": 0.0005, "loss": 1.1725648641586304, "step": 44723 }, { "epoch": 3.5257390618841153, "grad_norm": 0.5350613594055176, "learning_rate": 0.0005, "loss": 1.2580037117004395, "step": 44724 }, { "epoch": 3.525817895151754, "grad_norm": 0.461798757314682, "learning_rate": 0.0005, "loss": 1.1358973979949951, "step": 44725 }, { "epoch": 3.5258967284193927, "grad_norm": 0.5450132489204407, "learning_rate": 0.0005, "loss": 1.2052632570266724, "step": 44726 }, { "epoch": 3.525975561687032, "grad_norm": 0.5361846089363098, "learning_rate": 0.0005, "loss": 1.4465900659561157, "step": 44727 }, { "epoch": 3.526054394954671, "grad_norm": 0.5043549537658691, "learning_rate": 0.0005, "loss": 1.2908705472946167, "step": 44728 }, { "epoch": 3.5261332282223097, "grad_norm": 0.494295597076416, "learning_rate": 0.0005, "loss": 1.3986924886703491, "step": 44729 }, { "epoch": 3.526212061489949, "grad_norm": 0.4657207429409027, "learning_rate": 0.0005, "loss": 1.1259056329727173, "step": 44730 }, { "epoch": 3.5262908947575875, "grad_norm": 0.5559113025665283, "learning_rate": 0.0005, "loss": 1.1917695999145508, "step": 44731 }, { "epoch": 3.5263697280252266, "grad_norm": 0.5002247095108032, "learning_rate": 0.0005, "loss": 1.1589765548706055, "step": 44732 }, { "epoch": 3.5264485612928658, "grad_norm": 0.4630691707134247, "learning_rate": 0.0005, "loss": 1.2660331726074219, "step": 44733 }, { "epoch": 3.5265273945605045, "grad_norm": 0.5886719822883606, "learning_rate": 0.0005, "loss": 1.371647834777832, "step": 44734 }, { "epoch": 3.5266062278281436, "grad_norm": 0.5036252737045288, "learning_rate": 0.0005, "loss": 1.1231193542480469, "step": 44735 }, { "epoch": 3.5266850610957823, "grad_norm": 0.7558600902557373, "learning_rate": 0.0005, "loss": 1.4089984893798828, "step": 44736 }, { "epoch": 3.5267638943634214, "grad_norm": 0.49136456847190857, "learning_rate": 0.0005, "loss": 1.34709632396698, "step": 44737 }, { "epoch": 3.5268427276310605, "grad_norm": 0.5928111672401428, "learning_rate": 0.0005, "loss": 1.2130613327026367, "step": 44738 }, { "epoch": 3.5269215608986992, "grad_norm": 0.44745126366615295, "learning_rate": 0.0005, "loss": 1.3068320751190186, "step": 44739 }, { "epoch": 3.527000394166338, "grad_norm": 0.46761757135391235, "learning_rate": 0.0005, "loss": 1.1092629432678223, "step": 44740 }, { "epoch": 3.527079227433977, "grad_norm": 0.454906165599823, "learning_rate": 0.0005, "loss": 1.3429111242294312, "step": 44741 }, { "epoch": 3.527158060701616, "grad_norm": 0.46362560987472534, "learning_rate": 0.0005, "loss": 1.284580945968628, "step": 44742 }, { "epoch": 3.527236893969255, "grad_norm": 0.4867109954357147, "learning_rate": 0.0005, "loss": 1.4640803337097168, "step": 44743 }, { "epoch": 3.527315727236894, "grad_norm": 0.8907666802406311, "learning_rate": 0.0005, "loss": 1.2957323789596558, "step": 44744 }, { "epoch": 3.5273945605045327, "grad_norm": 0.5172727704048157, "learning_rate": 0.0005, "loss": 1.341807246208191, "step": 44745 }, { "epoch": 3.527473393772172, "grad_norm": 0.48645153641700745, "learning_rate": 0.0005, "loss": 1.1571195125579834, "step": 44746 }, { "epoch": 3.527552227039811, "grad_norm": 0.659098744392395, "learning_rate": 0.0005, "loss": 1.32891845703125, "step": 44747 }, { "epoch": 3.5276310603074497, "grad_norm": 0.46669986844062805, "learning_rate": 0.0005, "loss": 1.1042786836624146, "step": 44748 }, { "epoch": 3.527709893575089, "grad_norm": 0.45063379406929016, "learning_rate": 0.0005, "loss": 1.2617385387420654, "step": 44749 }, { "epoch": 3.5277887268427275, "grad_norm": 0.5155157446861267, "learning_rate": 0.0005, "loss": 1.3589329719543457, "step": 44750 }, { "epoch": 3.5278675601103666, "grad_norm": 0.4553842842578888, "learning_rate": 0.0005, "loss": 1.3385273218154907, "step": 44751 }, { "epoch": 3.5279463933780058, "grad_norm": 0.43912163376808167, "learning_rate": 0.0005, "loss": 1.2076234817504883, "step": 44752 }, { "epoch": 3.5280252266456444, "grad_norm": 1.0091897249221802, "learning_rate": 0.0005, "loss": 1.2690128087997437, "step": 44753 }, { "epoch": 3.528104059913283, "grad_norm": 0.48181021213531494, "learning_rate": 0.0005, "loss": 1.2529497146606445, "step": 44754 }, { "epoch": 3.5281828931809223, "grad_norm": 0.41646379232406616, "learning_rate": 0.0005, "loss": 1.1099090576171875, "step": 44755 }, { "epoch": 3.5282617264485614, "grad_norm": 0.5138262510299683, "learning_rate": 0.0005, "loss": 1.4191348552703857, "step": 44756 }, { "epoch": 3.5283405597162, "grad_norm": 0.4618140161037445, "learning_rate": 0.0005, "loss": 1.263901710510254, "step": 44757 }, { "epoch": 3.528419392983839, "grad_norm": 0.4203506410121918, "learning_rate": 0.0005, "loss": 1.1639714241027832, "step": 44758 }, { "epoch": 3.528498226251478, "grad_norm": 0.45839881896972656, "learning_rate": 0.0005, "loss": 1.1716322898864746, "step": 44759 }, { "epoch": 3.528577059519117, "grad_norm": 0.5076318383216858, "learning_rate": 0.0005, "loss": 1.1993985176086426, "step": 44760 }, { "epoch": 3.528655892786756, "grad_norm": 0.4354857802391052, "learning_rate": 0.0005, "loss": 1.336599349975586, "step": 44761 }, { "epoch": 3.528734726054395, "grad_norm": 0.587745189666748, "learning_rate": 0.0005, "loss": 1.2626080513000488, "step": 44762 }, { "epoch": 3.528813559322034, "grad_norm": 0.5135249495506287, "learning_rate": 0.0005, "loss": 1.1437863111495972, "step": 44763 }, { "epoch": 3.5288923925896727, "grad_norm": 0.4750615656375885, "learning_rate": 0.0005, "loss": 1.447461485862732, "step": 44764 }, { "epoch": 3.528971225857312, "grad_norm": 0.5035244226455688, "learning_rate": 0.0005, "loss": 1.516066312789917, "step": 44765 }, { "epoch": 3.529050059124951, "grad_norm": 0.5546538233757019, "learning_rate": 0.0005, "loss": 1.3864573240280151, "step": 44766 }, { "epoch": 3.5291288923925896, "grad_norm": 0.5068225860595703, "learning_rate": 0.0005, "loss": 1.2374815940856934, "step": 44767 }, { "epoch": 3.529207725660229, "grad_norm": 0.5606104135513306, "learning_rate": 0.0005, "loss": 1.339759111404419, "step": 44768 }, { "epoch": 3.5292865589278675, "grad_norm": 0.5486425161361694, "learning_rate": 0.0005, "loss": 1.406557321548462, "step": 44769 }, { "epoch": 3.5293653921955066, "grad_norm": 0.5389549732208252, "learning_rate": 0.0005, "loss": 1.1569663286209106, "step": 44770 }, { "epoch": 3.5294442254631453, "grad_norm": 0.6078627705574036, "learning_rate": 0.0005, "loss": 1.3327724933624268, "step": 44771 }, { "epoch": 3.5295230587307844, "grad_norm": 0.4716689884662628, "learning_rate": 0.0005, "loss": 1.2259522676467896, "step": 44772 }, { "epoch": 3.529601891998423, "grad_norm": 0.521706759929657, "learning_rate": 0.0005, "loss": 1.3242875337600708, "step": 44773 }, { "epoch": 3.5296807252660622, "grad_norm": 0.6345710754394531, "learning_rate": 0.0005, "loss": 1.3452374935150146, "step": 44774 }, { "epoch": 3.5297595585337014, "grad_norm": 0.691030740737915, "learning_rate": 0.0005, "loss": 1.4712400436401367, "step": 44775 }, { "epoch": 3.52983839180134, "grad_norm": 0.538523256778717, "learning_rate": 0.0005, "loss": 1.1667793989181519, "step": 44776 }, { "epoch": 3.529917225068979, "grad_norm": 0.4276440739631653, "learning_rate": 0.0005, "loss": 1.1124024391174316, "step": 44777 }, { "epoch": 3.529996058336618, "grad_norm": 0.51897794008255, "learning_rate": 0.0005, "loss": 1.309124231338501, "step": 44778 }, { "epoch": 3.530074891604257, "grad_norm": 0.5134025812149048, "learning_rate": 0.0005, "loss": 1.1385717391967773, "step": 44779 }, { "epoch": 3.530153724871896, "grad_norm": 0.4836365282535553, "learning_rate": 0.0005, "loss": 1.1415985822677612, "step": 44780 }, { "epoch": 3.530232558139535, "grad_norm": 0.47231870889663696, "learning_rate": 0.0005, "loss": 1.3345930576324463, "step": 44781 }, { "epoch": 3.530311391407174, "grad_norm": 0.5458345413208008, "learning_rate": 0.0005, "loss": 1.4760069847106934, "step": 44782 }, { "epoch": 3.5303902246748127, "grad_norm": 0.4254744350910187, "learning_rate": 0.0005, "loss": 1.1560925245285034, "step": 44783 }, { "epoch": 3.530469057942452, "grad_norm": 0.5860439538955688, "learning_rate": 0.0005, "loss": 1.3840428590774536, "step": 44784 }, { "epoch": 3.5305478912100905, "grad_norm": 0.504065215587616, "learning_rate": 0.0005, "loss": 1.0953269004821777, "step": 44785 }, { "epoch": 3.5306267244777296, "grad_norm": 0.4327642023563385, "learning_rate": 0.0005, "loss": 1.1200262308120728, "step": 44786 }, { "epoch": 3.5307055577453683, "grad_norm": 0.4079342484474182, "learning_rate": 0.0005, "loss": 1.1552592515945435, "step": 44787 }, { "epoch": 3.5307843910130075, "grad_norm": 0.3998955190181732, "learning_rate": 0.0005, "loss": 1.409529685974121, "step": 44788 }, { "epoch": 3.5308632242806466, "grad_norm": 0.614245593547821, "learning_rate": 0.0005, "loss": 1.5223437547683716, "step": 44789 }, { "epoch": 3.5309420575482853, "grad_norm": 0.4568055272102356, "learning_rate": 0.0005, "loss": 1.385784387588501, "step": 44790 }, { "epoch": 3.5310208908159244, "grad_norm": 0.5444682240486145, "learning_rate": 0.0005, "loss": 1.240400791168213, "step": 44791 }, { "epoch": 3.531099724083563, "grad_norm": 0.5083802938461304, "learning_rate": 0.0005, "loss": 1.4179434776306152, "step": 44792 }, { "epoch": 3.5311785573512022, "grad_norm": 0.5150500535964966, "learning_rate": 0.0005, "loss": 1.1406245231628418, "step": 44793 }, { "epoch": 3.5312573906188414, "grad_norm": 0.4590303301811218, "learning_rate": 0.0005, "loss": 1.158980369567871, "step": 44794 }, { "epoch": 3.53133622388648, "grad_norm": 0.4305022656917572, "learning_rate": 0.0005, "loss": 1.1831398010253906, "step": 44795 }, { "epoch": 3.531415057154119, "grad_norm": 0.48650169372558594, "learning_rate": 0.0005, "loss": 1.3595905303955078, "step": 44796 }, { "epoch": 3.531493890421758, "grad_norm": 0.45609092712402344, "learning_rate": 0.0005, "loss": 1.2731038331985474, "step": 44797 }, { "epoch": 3.531572723689397, "grad_norm": 0.5022228956222534, "learning_rate": 0.0005, "loss": 1.2951769828796387, "step": 44798 }, { "epoch": 3.531651556957036, "grad_norm": 0.5209406614303589, "learning_rate": 0.0005, "loss": 1.2171307802200317, "step": 44799 }, { "epoch": 3.531730390224675, "grad_norm": 0.4948275089263916, "learning_rate": 0.0005, "loss": 1.2988512516021729, "step": 44800 }, { "epoch": 3.5318092234923135, "grad_norm": 0.48168256878852844, "learning_rate": 0.0005, "loss": 1.2804375886917114, "step": 44801 }, { "epoch": 3.5318880567599527, "grad_norm": 0.4008319675922394, "learning_rate": 0.0005, "loss": 1.2436003684997559, "step": 44802 }, { "epoch": 3.531966890027592, "grad_norm": 0.4619917571544647, "learning_rate": 0.0005, "loss": 1.3569974899291992, "step": 44803 }, { "epoch": 3.5320457232952305, "grad_norm": 0.4642999470233917, "learning_rate": 0.0005, "loss": 1.1929035186767578, "step": 44804 }, { "epoch": 3.5321245565628696, "grad_norm": 0.48531439900398254, "learning_rate": 0.0005, "loss": 1.3227040767669678, "step": 44805 }, { "epoch": 3.5322033898305083, "grad_norm": 0.5417323112487793, "learning_rate": 0.0005, "loss": 1.276534080505371, "step": 44806 }, { "epoch": 3.5322822230981474, "grad_norm": 0.46969205141067505, "learning_rate": 0.0005, "loss": 1.1997442245483398, "step": 44807 }, { "epoch": 3.5323610563657866, "grad_norm": 0.4820765554904938, "learning_rate": 0.0005, "loss": 1.3382468223571777, "step": 44808 }, { "epoch": 3.5324398896334253, "grad_norm": 0.4827731251716614, "learning_rate": 0.0005, "loss": 1.2396330833435059, "step": 44809 }, { "epoch": 3.5325187229010644, "grad_norm": 0.5363102555274963, "learning_rate": 0.0005, "loss": 1.5088906288146973, "step": 44810 }, { "epoch": 3.532597556168703, "grad_norm": 0.47157129645347595, "learning_rate": 0.0005, "loss": 1.1808130741119385, "step": 44811 }, { "epoch": 3.532676389436342, "grad_norm": 0.8019326329231262, "learning_rate": 0.0005, "loss": 1.2928307056427002, "step": 44812 }, { "epoch": 3.5327552227039813, "grad_norm": 0.5164316296577454, "learning_rate": 0.0005, "loss": 1.3883552551269531, "step": 44813 }, { "epoch": 3.53283405597162, "grad_norm": 0.4444841742515564, "learning_rate": 0.0005, "loss": 1.2044181823730469, "step": 44814 }, { "epoch": 3.5329128892392587, "grad_norm": 0.5327981114387512, "learning_rate": 0.0005, "loss": 1.21705162525177, "step": 44815 }, { "epoch": 3.532991722506898, "grad_norm": 0.4769997298717499, "learning_rate": 0.0005, "loss": 1.1650404930114746, "step": 44816 }, { "epoch": 3.533070555774537, "grad_norm": 0.4564444422721863, "learning_rate": 0.0005, "loss": 1.2242889404296875, "step": 44817 }, { "epoch": 3.5331493890421757, "grad_norm": 0.595165491104126, "learning_rate": 0.0005, "loss": 1.29318368434906, "step": 44818 }, { "epoch": 3.533228222309815, "grad_norm": 0.4125991463661194, "learning_rate": 0.0005, "loss": 1.0898438692092896, "step": 44819 }, { "epoch": 3.5333070555774535, "grad_norm": 0.5924015641212463, "learning_rate": 0.0005, "loss": 1.345137357711792, "step": 44820 }, { "epoch": 3.5333858888450926, "grad_norm": 0.414568156003952, "learning_rate": 0.0005, "loss": 1.28028404712677, "step": 44821 }, { "epoch": 3.5334647221127318, "grad_norm": 0.6420741081237793, "learning_rate": 0.0005, "loss": 1.3216387033462524, "step": 44822 }, { "epoch": 3.5335435553803705, "grad_norm": 0.510370135307312, "learning_rate": 0.0005, "loss": 1.2996068000793457, "step": 44823 }, { "epoch": 3.5336223886480096, "grad_norm": 0.5135213136672974, "learning_rate": 0.0005, "loss": 1.2800211906433105, "step": 44824 }, { "epoch": 3.5337012219156483, "grad_norm": 0.5174078941345215, "learning_rate": 0.0005, "loss": 1.3078471422195435, "step": 44825 }, { "epoch": 3.5337800551832874, "grad_norm": 0.4773821234703064, "learning_rate": 0.0005, "loss": 1.3829028606414795, "step": 44826 }, { "epoch": 3.5338588884509265, "grad_norm": 0.6957681179046631, "learning_rate": 0.0005, "loss": 1.3653852939605713, "step": 44827 }, { "epoch": 3.5339377217185652, "grad_norm": 0.43347758054733276, "learning_rate": 0.0005, "loss": 1.2305996417999268, "step": 44828 }, { "epoch": 3.534016554986204, "grad_norm": 0.4286946654319763, "learning_rate": 0.0005, "loss": 1.1619553565979004, "step": 44829 }, { "epoch": 3.534095388253843, "grad_norm": 0.43941354751586914, "learning_rate": 0.0005, "loss": 1.2842260599136353, "step": 44830 }, { "epoch": 3.534174221521482, "grad_norm": 0.4973810315132141, "learning_rate": 0.0005, "loss": 1.2089636325836182, "step": 44831 }, { "epoch": 3.534253054789121, "grad_norm": 0.5738926529884338, "learning_rate": 0.0005, "loss": 1.2877278327941895, "step": 44832 }, { "epoch": 3.53433188805676, "grad_norm": 0.4132198691368103, "learning_rate": 0.0005, "loss": 1.1080667972564697, "step": 44833 }, { "epoch": 3.5344107213243987, "grad_norm": 0.3997679352760315, "learning_rate": 0.0005, "loss": 1.1430096626281738, "step": 44834 }, { "epoch": 3.534489554592038, "grad_norm": 0.745905339717865, "learning_rate": 0.0005, "loss": 1.337465524673462, "step": 44835 }, { "epoch": 3.534568387859677, "grad_norm": 0.7672492861747742, "learning_rate": 0.0005, "loss": 1.229245901107788, "step": 44836 }, { "epoch": 3.5346472211273157, "grad_norm": 0.46318987011909485, "learning_rate": 0.0005, "loss": 1.370640516281128, "step": 44837 }, { "epoch": 3.534726054394955, "grad_norm": 0.6562967300415039, "learning_rate": 0.0005, "loss": 1.2953916788101196, "step": 44838 }, { "epoch": 3.5348048876625935, "grad_norm": 0.47061848640441895, "learning_rate": 0.0005, "loss": 1.3018385171890259, "step": 44839 }, { "epoch": 3.5348837209302326, "grad_norm": 0.5700157880783081, "learning_rate": 0.0005, "loss": 1.345289707183838, "step": 44840 }, { "epoch": 3.5349625541978718, "grad_norm": 0.5099900364875793, "learning_rate": 0.0005, "loss": 1.2835925817489624, "step": 44841 }, { "epoch": 3.5350413874655104, "grad_norm": 0.45530667901039124, "learning_rate": 0.0005, "loss": 1.2961845397949219, "step": 44842 }, { "epoch": 3.535120220733149, "grad_norm": 0.4921623766422272, "learning_rate": 0.0005, "loss": 1.1801645755767822, "step": 44843 }, { "epoch": 3.5351990540007883, "grad_norm": 0.5774579048156738, "learning_rate": 0.0005, "loss": 1.1857718229293823, "step": 44844 }, { "epoch": 3.5352778872684274, "grad_norm": 0.7204754948616028, "learning_rate": 0.0005, "loss": 1.2200556993484497, "step": 44845 }, { "epoch": 3.535356720536066, "grad_norm": 0.4514916241168976, "learning_rate": 0.0005, "loss": 1.3155186176300049, "step": 44846 }, { "epoch": 3.535435553803705, "grad_norm": 0.417091965675354, "learning_rate": 0.0005, "loss": 1.1491248607635498, "step": 44847 }, { "epoch": 3.535514387071344, "grad_norm": 0.5786420702934265, "learning_rate": 0.0005, "loss": 1.1929781436920166, "step": 44848 }, { "epoch": 3.535593220338983, "grad_norm": 0.38973093032836914, "learning_rate": 0.0005, "loss": 1.0844967365264893, "step": 44849 }, { "epoch": 3.535672053606622, "grad_norm": 0.42820024490356445, "learning_rate": 0.0005, "loss": 1.1527864933013916, "step": 44850 }, { "epoch": 3.535750886874261, "grad_norm": 0.6400585770606995, "learning_rate": 0.0005, "loss": 1.3379993438720703, "step": 44851 }, { "epoch": 3.5358297201419, "grad_norm": 0.6746973395347595, "learning_rate": 0.0005, "loss": 1.3260602951049805, "step": 44852 }, { "epoch": 3.5359085534095387, "grad_norm": 0.4987435042858124, "learning_rate": 0.0005, "loss": 1.1638569831848145, "step": 44853 }, { "epoch": 3.535987386677178, "grad_norm": 0.5207794308662415, "learning_rate": 0.0005, "loss": 1.2622101306915283, "step": 44854 }, { "epoch": 3.536066219944817, "grad_norm": 0.5597882866859436, "learning_rate": 0.0005, "loss": 1.3770612478256226, "step": 44855 }, { "epoch": 3.5361450532124556, "grad_norm": 0.5836918950080872, "learning_rate": 0.0005, "loss": 1.212001085281372, "step": 44856 }, { "epoch": 3.5362238864800943, "grad_norm": 0.595779538154602, "learning_rate": 0.0005, "loss": 1.2108652591705322, "step": 44857 }, { "epoch": 3.5363027197477335, "grad_norm": 0.4129418134689331, "learning_rate": 0.0005, "loss": 1.1867111921310425, "step": 44858 }, { "epoch": 3.5363815530153726, "grad_norm": 0.4371556043624878, "learning_rate": 0.0005, "loss": 1.293694257736206, "step": 44859 }, { "epoch": 3.5364603862830113, "grad_norm": 0.5177008509635925, "learning_rate": 0.0005, "loss": 1.2664315700531006, "step": 44860 }, { "epoch": 3.5365392195506504, "grad_norm": 0.5091717839241028, "learning_rate": 0.0005, "loss": 1.3664652109146118, "step": 44861 }, { "epoch": 3.536618052818289, "grad_norm": 0.5569024682044983, "learning_rate": 0.0005, "loss": 1.229142427444458, "step": 44862 }, { "epoch": 3.5366968860859282, "grad_norm": 0.4886809289455414, "learning_rate": 0.0005, "loss": 1.0845001935958862, "step": 44863 }, { "epoch": 3.5367757193535674, "grad_norm": 0.4450102746486664, "learning_rate": 0.0005, "loss": 1.0648505687713623, "step": 44864 }, { "epoch": 3.536854552621206, "grad_norm": 0.544039785861969, "learning_rate": 0.0005, "loss": 1.3345885276794434, "step": 44865 }, { "epoch": 3.536933385888845, "grad_norm": 0.45489785075187683, "learning_rate": 0.0005, "loss": 1.2556637525558472, "step": 44866 }, { "epoch": 3.537012219156484, "grad_norm": 0.5053659081459045, "learning_rate": 0.0005, "loss": 1.3983230590820312, "step": 44867 }, { "epoch": 3.537091052424123, "grad_norm": 0.46316319704055786, "learning_rate": 0.0005, "loss": 1.331229329109192, "step": 44868 }, { "epoch": 3.537169885691762, "grad_norm": 0.5273805856704712, "learning_rate": 0.0005, "loss": 1.273391604423523, "step": 44869 }, { "epoch": 3.537248718959401, "grad_norm": 0.5021863579750061, "learning_rate": 0.0005, "loss": 1.3230401277542114, "step": 44870 }, { "epoch": 3.5373275522270395, "grad_norm": 0.4830264449119568, "learning_rate": 0.0005, "loss": 1.2080730199813843, "step": 44871 }, { "epoch": 3.5374063854946787, "grad_norm": 0.4761058986186981, "learning_rate": 0.0005, "loss": 1.2220463752746582, "step": 44872 }, { "epoch": 3.537485218762318, "grad_norm": 0.6103178262710571, "learning_rate": 0.0005, "loss": 1.344240427017212, "step": 44873 }, { "epoch": 3.5375640520299565, "grad_norm": 0.4351847469806671, "learning_rate": 0.0005, "loss": 1.082375407218933, "step": 44874 }, { "epoch": 3.5376428852975956, "grad_norm": 0.3996249735355377, "learning_rate": 0.0005, "loss": 1.05665922164917, "step": 44875 }, { "epoch": 3.5377217185652343, "grad_norm": 0.4823113977909088, "learning_rate": 0.0005, "loss": 1.3372461795806885, "step": 44876 }, { "epoch": 3.5378005518328735, "grad_norm": 0.4771958291530609, "learning_rate": 0.0005, "loss": 1.1831574440002441, "step": 44877 }, { "epoch": 3.5378793851005126, "grad_norm": 0.6059201955795288, "learning_rate": 0.0005, "loss": 1.239565134048462, "step": 44878 }, { "epoch": 3.5379582183681513, "grad_norm": 0.4700435698032379, "learning_rate": 0.0005, "loss": 1.1818804740905762, "step": 44879 }, { "epoch": 3.5380370516357904, "grad_norm": 0.5503418445587158, "learning_rate": 0.0005, "loss": 1.3221385478973389, "step": 44880 }, { "epoch": 3.538115884903429, "grad_norm": 0.4348927140235901, "learning_rate": 0.0005, "loss": 1.394583821296692, "step": 44881 }, { "epoch": 3.5381947181710682, "grad_norm": 0.6113967299461365, "learning_rate": 0.0005, "loss": 1.4025992155075073, "step": 44882 }, { "epoch": 3.5382735514387074, "grad_norm": 0.5170404314994812, "learning_rate": 0.0005, "loss": 1.4102301597595215, "step": 44883 }, { "epoch": 3.538352384706346, "grad_norm": 0.43388205766677856, "learning_rate": 0.0005, "loss": 1.3658579587936401, "step": 44884 }, { "epoch": 3.538431217973985, "grad_norm": 0.4402935802936554, "learning_rate": 0.0005, "loss": 1.179660439491272, "step": 44885 }, { "epoch": 3.538510051241624, "grad_norm": 0.4916670322418213, "learning_rate": 0.0005, "loss": 1.3413949012756348, "step": 44886 }, { "epoch": 3.538588884509263, "grad_norm": 0.4272078573703766, "learning_rate": 0.0005, "loss": 1.145640254020691, "step": 44887 }, { "epoch": 3.5386677177769017, "grad_norm": 0.544792652130127, "learning_rate": 0.0005, "loss": 1.295133352279663, "step": 44888 }, { "epoch": 3.538746551044541, "grad_norm": 0.5259416699409485, "learning_rate": 0.0005, "loss": 1.2084673643112183, "step": 44889 }, { "epoch": 3.5388253843121795, "grad_norm": 0.4731505811214447, "learning_rate": 0.0005, "loss": 1.3023548126220703, "step": 44890 }, { "epoch": 3.5389042175798187, "grad_norm": 0.5741350054740906, "learning_rate": 0.0005, "loss": 1.3111677169799805, "step": 44891 }, { "epoch": 3.538983050847458, "grad_norm": 0.5239904522895813, "learning_rate": 0.0005, "loss": 1.2992963790893555, "step": 44892 }, { "epoch": 3.5390618841150965, "grad_norm": 0.6307811737060547, "learning_rate": 0.0005, "loss": 1.3365474939346313, "step": 44893 }, { "epoch": 3.5391407173827356, "grad_norm": 0.5766117572784424, "learning_rate": 0.0005, "loss": 1.2954453229904175, "step": 44894 }, { "epoch": 3.5392195506503743, "grad_norm": 0.5504209399223328, "learning_rate": 0.0005, "loss": 1.4605705738067627, "step": 44895 }, { "epoch": 3.5392983839180134, "grad_norm": 0.5073075890541077, "learning_rate": 0.0005, "loss": 1.186511516571045, "step": 44896 }, { "epoch": 3.5393772171856526, "grad_norm": 0.44277510046958923, "learning_rate": 0.0005, "loss": 1.214829921722412, "step": 44897 }, { "epoch": 3.5394560504532913, "grad_norm": 0.5097548365592957, "learning_rate": 0.0005, "loss": 1.2005503177642822, "step": 44898 }, { "epoch": 3.5395348837209304, "grad_norm": 0.6422630548477173, "learning_rate": 0.0005, "loss": 1.5753681659698486, "step": 44899 }, { "epoch": 3.539613716988569, "grad_norm": 0.5388501882553101, "learning_rate": 0.0005, "loss": 1.2216517925262451, "step": 44900 }, { "epoch": 3.539692550256208, "grad_norm": 0.4873548746109009, "learning_rate": 0.0005, "loss": 1.3835386037826538, "step": 44901 }, { "epoch": 3.539771383523847, "grad_norm": 0.4660564363002777, "learning_rate": 0.0005, "loss": 1.2163676023483276, "step": 44902 }, { "epoch": 3.539850216791486, "grad_norm": 0.4962223470211029, "learning_rate": 0.0005, "loss": 1.281701683998108, "step": 44903 }, { "epoch": 3.5399290500591247, "grad_norm": 0.44602060317993164, "learning_rate": 0.0005, "loss": 1.3006742000579834, "step": 44904 }, { "epoch": 3.540007883326764, "grad_norm": 0.509215772151947, "learning_rate": 0.0005, "loss": 1.2555747032165527, "step": 44905 }, { "epoch": 3.540086716594403, "grad_norm": 0.610963761806488, "learning_rate": 0.0005, "loss": 1.3110032081604004, "step": 44906 }, { "epoch": 3.5401655498620417, "grad_norm": 0.4151895046234131, "learning_rate": 0.0005, "loss": 1.1720951795578003, "step": 44907 }, { "epoch": 3.540244383129681, "grad_norm": 0.5473874807357788, "learning_rate": 0.0005, "loss": 1.2356452941894531, "step": 44908 }, { "epoch": 3.5403232163973195, "grad_norm": 0.4381227493286133, "learning_rate": 0.0005, "loss": 1.2724103927612305, "step": 44909 }, { "epoch": 3.5404020496649586, "grad_norm": 0.6024817228317261, "learning_rate": 0.0005, "loss": 1.3459324836730957, "step": 44910 }, { "epoch": 3.5404808829325978, "grad_norm": 0.4732173681259155, "learning_rate": 0.0005, "loss": 1.3107774257659912, "step": 44911 }, { "epoch": 3.5405597162002365, "grad_norm": 0.5304021835327148, "learning_rate": 0.0005, "loss": 1.3518996238708496, "step": 44912 }, { "epoch": 3.5406385494678756, "grad_norm": 0.5432363152503967, "learning_rate": 0.0005, "loss": 1.3665680885314941, "step": 44913 }, { "epoch": 3.5407173827355143, "grad_norm": 0.5667340755462646, "learning_rate": 0.0005, "loss": 1.2505080699920654, "step": 44914 }, { "epoch": 3.5407962160031534, "grad_norm": 0.6289016604423523, "learning_rate": 0.0005, "loss": 1.299940824508667, "step": 44915 }, { "epoch": 3.5408750492707926, "grad_norm": 0.6310091018676758, "learning_rate": 0.0005, "loss": 1.3751189708709717, "step": 44916 }, { "epoch": 3.5409538825384312, "grad_norm": 0.4881187975406647, "learning_rate": 0.0005, "loss": 1.197299838066101, "step": 44917 }, { "epoch": 3.54103271580607, "grad_norm": 0.49677857756614685, "learning_rate": 0.0005, "loss": 1.3757832050323486, "step": 44918 }, { "epoch": 3.541111549073709, "grad_norm": 0.4865146577358246, "learning_rate": 0.0005, "loss": 1.2937641143798828, "step": 44919 }, { "epoch": 3.541190382341348, "grad_norm": 0.48815762996673584, "learning_rate": 0.0005, "loss": 1.141249179840088, "step": 44920 }, { "epoch": 3.541269215608987, "grad_norm": 0.5625582933425903, "learning_rate": 0.0005, "loss": 1.406372308731079, "step": 44921 }, { "epoch": 3.541348048876626, "grad_norm": 0.545325517654419, "learning_rate": 0.0005, "loss": 1.2919477224349976, "step": 44922 }, { "epoch": 3.5414268821442647, "grad_norm": 0.4736238420009613, "learning_rate": 0.0005, "loss": 1.124530553817749, "step": 44923 }, { "epoch": 3.541505715411904, "grad_norm": 0.7008804082870483, "learning_rate": 0.0005, "loss": 1.3678905963897705, "step": 44924 }, { "epoch": 3.541584548679543, "grad_norm": 0.5382319688796997, "learning_rate": 0.0005, "loss": 1.3261370658874512, "step": 44925 }, { "epoch": 3.5416633819471817, "grad_norm": 0.6072627305984497, "learning_rate": 0.0005, "loss": 1.2958552837371826, "step": 44926 }, { "epoch": 3.541742215214821, "grad_norm": 0.49284467101097107, "learning_rate": 0.0005, "loss": 1.089312195777893, "step": 44927 }, { "epoch": 3.5418210484824595, "grad_norm": 0.7404364347457886, "learning_rate": 0.0005, "loss": 1.3222655057907104, "step": 44928 }, { "epoch": 3.5418998817500986, "grad_norm": 0.8869196772575378, "learning_rate": 0.0005, "loss": 1.2197237014770508, "step": 44929 }, { "epoch": 3.5419787150177378, "grad_norm": 0.38952383399009705, "learning_rate": 0.0005, "loss": 1.2128993272781372, "step": 44930 }, { "epoch": 3.5420575482853764, "grad_norm": 0.4698933959007263, "learning_rate": 0.0005, "loss": 1.1559579372406006, "step": 44931 }, { "epoch": 3.542136381553015, "grad_norm": 0.5898988842964172, "learning_rate": 0.0005, "loss": 1.4747540950775146, "step": 44932 }, { "epoch": 3.5422152148206543, "grad_norm": 0.43487703800201416, "learning_rate": 0.0005, "loss": 1.162764310836792, "step": 44933 }, { "epoch": 3.5422940480882934, "grad_norm": 0.6242796778678894, "learning_rate": 0.0005, "loss": 1.3648262023925781, "step": 44934 }, { "epoch": 3.542372881355932, "grad_norm": 0.5969739556312561, "learning_rate": 0.0005, "loss": 1.2076470851898193, "step": 44935 }, { "epoch": 3.5424517146235712, "grad_norm": 0.5144575834274292, "learning_rate": 0.0005, "loss": 1.3038699626922607, "step": 44936 }, { "epoch": 3.54253054789121, "grad_norm": 0.47791656851768494, "learning_rate": 0.0005, "loss": 1.235339879989624, "step": 44937 }, { "epoch": 3.542609381158849, "grad_norm": 0.4709607660770416, "learning_rate": 0.0005, "loss": 1.0278713703155518, "step": 44938 }, { "epoch": 3.542688214426488, "grad_norm": 0.4056906998157501, "learning_rate": 0.0005, "loss": 1.0849262475967407, "step": 44939 }, { "epoch": 3.542767047694127, "grad_norm": 0.794424831867218, "learning_rate": 0.0005, "loss": 1.4239017963409424, "step": 44940 }, { "epoch": 3.542845880961766, "grad_norm": 0.6121653914451599, "learning_rate": 0.0005, "loss": 1.3953931331634521, "step": 44941 }, { "epoch": 3.5429247142294047, "grad_norm": 0.5181073546409607, "learning_rate": 0.0005, "loss": 1.2545472383499146, "step": 44942 }, { "epoch": 3.543003547497044, "grad_norm": 0.46898338198661804, "learning_rate": 0.0005, "loss": 1.158766746520996, "step": 44943 }, { "epoch": 3.543082380764683, "grad_norm": 0.4846633970737457, "learning_rate": 0.0005, "loss": 1.1158169507980347, "step": 44944 }, { "epoch": 3.5431612140323216, "grad_norm": 0.5731545090675354, "learning_rate": 0.0005, "loss": 1.3320138454437256, "step": 44945 }, { "epoch": 3.5432400472999603, "grad_norm": 0.5438924431800842, "learning_rate": 0.0005, "loss": 1.2665623426437378, "step": 44946 }, { "epoch": 3.5433188805675995, "grad_norm": 0.40726301074028015, "learning_rate": 0.0005, "loss": 1.1014204025268555, "step": 44947 }, { "epoch": 3.5433977138352386, "grad_norm": 0.4280203878879547, "learning_rate": 0.0005, "loss": 1.2226470708847046, "step": 44948 }, { "epoch": 3.5434765471028773, "grad_norm": 0.620817244052887, "learning_rate": 0.0005, "loss": 1.270385503768921, "step": 44949 }, { "epoch": 3.5435553803705164, "grad_norm": 0.6541272401809692, "learning_rate": 0.0005, "loss": 1.2482450008392334, "step": 44950 }, { "epoch": 3.543634213638155, "grad_norm": 0.5258820056915283, "learning_rate": 0.0005, "loss": 1.091740369796753, "step": 44951 }, { "epoch": 3.5437130469057943, "grad_norm": 0.5063846111297607, "learning_rate": 0.0005, "loss": 1.3574392795562744, "step": 44952 }, { "epoch": 3.5437918801734334, "grad_norm": 0.4412768483161926, "learning_rate": 0.0005, "loss": 1.213847279548645, "step": 44953 }, { "epoch": 3.543870713441072, "grad_norm": 0.4303953945636749, "learning_rate": 0.0005, "loss": 1.1244561672210693, "step": 44954 }, { "epoch": 3.543949546708711, "grad_norm": 0.4719688296318054, "learning_rate": 0.0005, "loss": 1.2255607843399048, "step": 44955 }, { "epoch": 3.54402837997635, "grad_norm": 0.45615217089653015, "learning_rate": 0.0005, "loss": 1.2107489109039307, "step": 44956 }, { "epoch": 3.544107213243989, "grad_norm": 0.519476056098938, "learning_rate": 0.0005, "loss": 1.3530869483947754, "step": 44957 }, { "epoch": 3.544186046511628, "grad_norm": 0.4770585298538208, "learning_rate": 0.0005, "loss": 1.1966557502746582, "step": 44958 }, { "epoch": 3.544264879779267, "grad_norm": 0.41788044571876526, "learning_rate": 0.0005, "loss": 1.1695666313171387, "step": 44959 }, { "epoch": 3.5443437130469055, "grad_norm": 0.5102227926254272, "learning_rate": 0.0005, "loss": 1.2429835796356201, "step": 44960 }, { "epoch": 3.5444225463145447, "grad_norm": 0.7904093861579895, "learning_rate": 0.0005, "loss": 1.1939035654067993, "step": 44961 }, { "epoch": 3.544501379582184, "grad_norm": 0.4610939621925354, "learning_rate": 0.0005, "loss": 1.3306540250778198, "step": 44962 }, { "epoch": 3.5445802128498225, "grad_norm": 0.43476420640945435, "learning_rate": 0.0005, "loss": 1.1040401458740234, "step": 44963 }, { "epoch": 3.5446590461174616, "grad_norm": 0.41274476051330566, "learning_rate": 0.0005, "loss": 1.0962505340576172, "step": 44964 }, { "epoch": 3.5447378793851003, "grad_norm": 0.5188060998916626, "learning_rate": 0.0005, "loss": 1.4840373992919922, "step": 44965 }, { "epoch": 3.5448167126527395, "grad_norm": 0.6051977872848511, "learning_rate": 0.0005, "loss": 1.3098535537719727, "step": 44966 }, { "epoch": 3.5448955459203786, "grad_norm": 0.5128390789031982, "learning_rate": 0.0005, "loss": 1.2520874738693237, "step": 44967 }, { "epoch": 3.5449743791880173, "grad_norm": 0.44389045238494873, "learning_rate": 0.0005, "loss": 1.1904501914978027, "step": 44968 }, { "epoch": 3.5450532124556564, "grad_norm": 0.467491090297699, "learning_rate": 0.0005, "loss": 1.4327081441879272, "step": 44969 }, { "epoch": 3.545132045723295, "grad_norm": 0.5140039920806885, "learning_rate": 0.0005, "loss": 1.2913484573364258, "step": 44970 }, { "epoch": 3.5452108789909342, "grad_norm": 0.47050151228904724, "learning_rate": 0.0005, "loss": 1.4025604724884033, "step": 44971 }, { "epoch": 3.5452897122585734, "grad_norm": 0.5635112524032593, "learning_rate": 0.0005, "loss": 1.4309945106506348, "step": 44972 }, { "epoch": 3.545368545526212, "grad_norm": 0.47301626205444336, "learning_rate": 0.0005, "loss": 1.1419742107391357, "step": 44973 }, { "epoch": 3.5454473787938507, "grad_norm": 0.4564254879951477, "learning_rate": 0.0005, "loss": 1.0885791778564453, "step": 44974 }, { "epoch": 3.54552621206149, "grad_norm": 0.4797138571739197, "learning_rate": 0.0005, "loss": 1.2414320707321167, "step": 44975 }, { "epoch": 3.545605045329129, "grad_norm": 0.40709570050239563, "learning_rate": 0.0005, "loss": 1.1350619792938232, "step": 44976 }, { "epoch": 3.5456838785967677, "grad_norm": 0.6438767910003662, "learning_rate": 0.0005, "loss": 1.4077816009521484, "step": 44977 }, { "epoch": 3.545762711864407, "grad_norm": 0.5544072389602661, "learning_rate": 0.0005, "loss": 1.1729986667633057, "step": 44978 }, { "epoch": 3.5458415451320455, "grad_norm": 0.5252721309661865, "learning_rate": 0.0005, "loss": 1.6860969066619873, "step": 44979 }, { "epoch": 3.5459203783996847, "grad_norm": 0.5736954212188721, "learning_rate": 0.0005, "loss": 1.2175991535186768, "step": 44980 }, { "epoch": 3.545999211667324, "grad_norm": 0.41848912835121155, "learning_rate": 0.0005, "loss": 1.1146039962768555, "step": 44981 }, { "epoch": 3.5460780449349625, "grad_norm": 0.5127090811729431, "learning_rate": 0.0005, "loss": 1.2307803630828857, "step": 44982 }, { "epoch": 3.5461568782026016, "grad_norm": 0.42349711060523987, "learning_rate": 0.0005, "loss": 1.2290335893630981, "step": 44983 }, { "epoch": 3.5462357114702403, "grad_norm": 0.5254520177841187, "learning_rate": 0.0005, "loss": 1.2426741123199463, "step": 44984 }, { "epoch": 3.5463145447378794, "grad_norm": 0.4641270935535431, "learning_rate": 0.0005, "loss": 1.3783336877822876, "step": 44985 }, { "epoch": 3.5463933780055186, "grad_norm": 0.7781263589859009, "learning_rate": 0.0005, "loss": 1.2841565608978271, "step": 44986 }, { "epoch": 3.5464722112731573, "grad_norm": 0.6602312326431274, "learning_rate": 0.0005, "loss": 1.4415050745010376, "step": 44987 }, { "epoch": 3.546551044540796, "grad_norm": 0.5112355351448059, "learning_rate": 0.0005, "loss": 1.444257378578186, "step": 44988 }, { "epoch": 3.546629877808435, "grad_norm": 0.6436572670936584, "learning_rate": 0.0005, "loss": 1.5113732814788818, "step": 44989 }, { "epoch": 3.546708711076074, "grad_norm": 0.4182049036026001, "learning_rate": 0.0005, "loss": 1.2531068325042725, "step": 44990 }, { "epoch": 3.546787544343713, "grad_norm": 0.6005423665046692, "learning_rate": 0.0005, "loss": 1.552735686302185, "step": 44991 }, { "epoch": 3.546866377611352, "grad_norm": 0.4637145400047302, "learning_rate": 0.0005, "loss": 1.175987720489502, "step": 44992 }, { "epoch": 3.5469452108789907, "grad_norm": 0.5279605388641357, "learning_rate": 0.0005, "loss": 1.4952638149261475, "step": 44993 }, { "epoch": 3.54702404414663, "grad_norm": 0.5083686709403992, "learning_rate": 0.0005, "loss": 1.2699419260025024, "step": 44994 }, { "epoch": 3.547102877414269, "grad_norm": 0.3928317725658417, "learning_rate": 0.0005, "loss": 1.0039387941360474, "step": 44995 }, { "epoch": 3.5471817106819077, "grad_norm": 0.43076953291893005, "learning_rate": 0.0005, "loss": 1.4105181694030762, "step": 44996 }, { "epoch": 3.547260543949547, "grad_norm": 0.46764805912971497, "learning_rate": 0.0005, "loss": 1.2690353393554688, "step": 44997 }, { "epoch": 3.5473393772171855, "grad_norm": 0.4202384352684021, "learning_rate": 0.0005, "loss": 1.3089394569396973, "step": 44998 }, { "epoch": 3.5474182104848246, "grad_norm": 0.46254321932792664, "learning_rate": 0.0005, "loss": 1.287567138671875, "step": 44999 }, { "epoch": 3.5474970437524638, "grad_norm": 0.4566599726676941, "learning_rate": 0.0005, "loss": 1.2870633602142334, "step": 45000 }, { "epoch": 3.5475758770201025, "grad_norm": 0.4952295124530792, "learning_rate": 0.0005, "loss": 1.2218981981277466, "step": 45001 }, { "epoch": 3.547654710287741, "grad_norm": 0.5231389403343201, "learning_rate": 0.0005, "loss": 1.319441318511963, "step": 45002 }, { "epoch": 3.5477335435553803, "grad_norm": 0.5477681756019592, "learning_rate": 0.0005, "loss": 1.0789000988006592, "step": 45003 }, { "epoch": 3.5478123768230194, "grad_norm": 0.512557864189148, "learning_rate": 0.0005, "loss": 1.2082278728485107, "step": 45004 }, { "epoch": 3.547891210090658, "grad_norm": 0.4187367558479309, "learning_rate": 0.0005, "loss": 1.1755777597427368, "step": 45005 }, { "epoch": 3.5479700433582972, "grad_norm": 0.6133053302764893, "learning_rate": 0.0005, "loss": 1.2776751518249512, "step": 45006 }, { "epoch": 3.548048876625936, "grad_norm": 0.5518965721130371, "learning_rate": 0.0005, "loss": 1.386067509651184, "step": 45007 }, { "epoch": 3.548127709893575, "grad_norm": 0.5174204111099243, "learning_rate": 0.0005, "loss": 1.1723763942718506, "step": 45008 }, { "epoch": 3.548206543161214, "grad_norm": 0.5844295620918274, "learning_rate": 0.0005, "loss": 1.22953200340271, "step": 45009 }, { "epoch": 3.548285376428853, "grad_norm": 0.5204710364341736, "learning_rate": 0.0005, "loss": 1.285369634628296, "step": 45010 }, { "epoch": 3.548364209696492, "grad_norm": 0.46373310685157776, "learning_rate": 0.0005, "loss": 1.0913658142089844, "step": 45011 }, { "epoch": 3.5484430429641307, "grad_norm": 0.4113510847091675, "learning_rate": 0.0005, "loss": 1.0657579898834229, "step": 45012 }, { "epoch": 3.54852187623177, "grad_norm": 0.4813438355922699, "learning_rate": 0.0005, "loss": 1.2386096715927124, "step": 45013 }, { "epoch": 3.548600709499409, "grad_norm": 0.46262896060943604, "learning_rate": 0.0005, "loss": 1.338014841079712, "step": 45014 }, { "epoch": 3.5486795427670477, "grad_norm": 0.514866292476654, "learning_rate": 0.0005, "loss": 1.3215172290802002, "step": 45015 }, { "epoch": 3.548758376034687, "grad_norm": 0.6576724052429199, "learning_rate": 0.0005, "loss": 1.458448052406311, "step": 45016 }, { "epoch": 3.5488372093023255, "grad_norm": 0.5686200261116028, "learning_rate": 0.0005, "loss": 1.2280352115631104, "step": 45017 }, { "epoch": 3.5489160425699646, "grad_norm": 0.614853024482727, "learning_rate": 0.0005, "loss": 1.4901423454284668, "step": 45018 }, { "epoch": 3.5489948758376033, "grad_norm": 0.5564982295036316, "learning_rate": 0.0005, "loss": 1.1063597202301025, "step": 45019 }, { "epoch": 3.5490737091052424, "grad_norm": 0.4952833950519562, "learning_rate": 0.0005, "loss": 1.2675285339355469, "step": 45020 }, { "epoch": 3.549152542372881, "grad_norm": 0.4442031681537628, "learning_rate": 0.0005, "loss": 1.2158077955245972, "step": 45021 }, { "epoch": 3.5492313756405203, "grad_norm": 0.46220293641090393, "learning_rate": 0.0005, "loss": 1.1618038415908813, "step": 45022 }, { "epoch": 3.5493102089081594, "grad_norm": 0.5240673422813416, "learning_rate": 0.0005, "loss": 1.4825239181518555, "step": 45023 }, { "epoch": 3.549389042175798, "grad_norm": 0.5197438597679138, "learning_rate": 0.0005, "loss": 1.3026372194290161, "step": 45024 }, { "epoch": 3.5494678754434372, "grad_norm": 0.568203330039978, "learning_rate": 0.0005, "loss": 1.4985260963439941, "step": 45025 }, { "epoch": 3.549546708711076, "grad_norm": 0.5048672556877136, "learning_rate": 0.0005, "loss": 1.3080761432647705, "step": 45026 }, { "epoch": 3.549625541978715, "grad_norm": 0.4614810347557068, "learning_rate": 0.0005, "loss": 1.2841953039169312, "step": 45027 }, { "epoch": 3.549704375246354, "grad_norm": 0.5402119159698486, "learning_rate": 0.0005, "loss": 1.217536211013794, "step": 45028 }, { "epoch": 3.549783208513993, "grad_norm": 0.8747313618659973, "learning_rate": 0.0005, "loss": 1.163834810256958, "step": 45029 }, { "epoch": 3.549862041781632, "grad_norm": 0.4780193269252777, "learning_rate": 0.0005, "loss": 1.1638805866241455, "step": 45030 }, { "epoch": 3.5499408750492707, "grad_norm": 0.616423487663269, "learning_rate": 0.0005, "loss": 1.3676648139953613, "step": 45031 }, { "epoch": 3.55001970831691, "grad_norm": 0.5758566856384277, "learning_rate": 0.0005, "loss": 1.3862494230270386, "step": 45032 }, { "epoch": 3.5500985415845485, "grad_norm": 0.4400688707828522, "learning_rate": 0.0005, "loss": 1.138129711151123, "step": 45033 }, { "epoch": 3.5501773748521877, "grad_norm": 0.5852007865905762, "learning_rate": 0.0005, "loss": 1.3640851974487305, "step": 45034 }, { "epoch": 3.5502562081198263, "grad_norm": 0.5404772162437439, "learning_rate": 0.0005, "loss": 1.3380568027496338, "step": 45035 }, { "epoch": 3.5503350413874655, "grad_norm": 0.6013800501823425, "learning_rate": 0.0005, "loss": 1.460200309753418, "step": 45036 }, { "epoch": 3.5504138746551046, "grad_norm": 0.5210067629814148, "learning_rate": 0.0005, "loss": 1.3098886013031006, "step": 45037 }, { "epoch": 3.5504927079227433, "grad_norm": 0.4261118769645691, "learning_rate": 0.0005, "loss": 1.26690673828125, "step": 45038 }, { "epoch": 3.5505715411903824, "grad_norm": 0.41763192415237427, "learning_rate": 0.0005, "loss": 1.1811292171478271, "step": 45039 }, { "epoch": 3.550650374458021, "grad_norm": 0.4500376880168915, "learning_rate": 0.0005, "loss": 1.4074435234069824, "step": 45040 }, { "epoch": 3.5507292077256603, "grad_norm": 0.4064268171787262, "learning_rate": 0.0005, "loss": 1.1581597328186035, "step": 45041 }, { "epoch": 3.5508080409932994, "grad_norm": 0.43158483505249023, "learning_rate": 0.0005, "loss": 1.3001885414123535, "step": 45042 }, { "epoch": 3.550886874260938, "grad_norm": 0.5335053205490112, "learning_rate": 0.0005, "loss": 1.3636358976364136, "step": 45043 }, { "epoch": 3.550965707528577, "grad_norm": 0.47131505608558655, "learning_rate": 0.0005, "loss": 1.1432385444641113, "step": 45044 }, { "epoch": 3.551044540796216, "grad_norm": 0.45851173996925354, "learning_rate": 0.0005, "loss": 1.304927110671997, "step": 45045 }, { "epoch": 3.551123374063855, "grad_norm": 0.752923846244812, "learning_rate": 0.0005, "loss": 1.3424758911132812, "step": 45046 }, { "epoch": 3.551202207331494, "grad_norm": 0.4267685115337372, "learning_rate": 0.0005, "loss": 1.2484276294708252, "step": 45047 }, { "epoch": 3.551281040599133, "grad_norm": 0.4328228533267975, "learning_rate": 0.0005, "loss": 1.1607028245925903, "step": 45048 }, { "epoch": 3.5513598738667715, "grad_norm": 0.6297481060028076, "learning_rate": 0.0005, "loss": 1.3717535734176636, "step": 45049 }, { "epoch": 3.5514387071344107, "grad_norm": 0.423749178647995, "learning_rate": 0.0005, "loss": 1.100687026977539, "step": 45050 }, { "epoch": 3.55151754040205, "grad_norm": 0.4798382818698883, "learning_rate": 0.0005, "loss": 1.1445053815841675, "step": 45051 }, { "epoch": 3.5515963736696885, "grad_norm": 0.48208263516426086, "learning_rate": 0.0005, "loss": 1.3321175575256348, "step": 45052 }, { "epoch": 3.5516752069373276, "grad_norm": 0.4281291365623474, "learning_rate": 0.0005, "loss": 1.0994725227355957, "step": 45053 }, { "epoch": 3.5517540402049663, "grad_norm": 0.41977134346961975, "learning_rate": 0.0005, "loss": 1.1933305263519287, "step": 45054 }, { "epoch": 3.5518328734726055, "grad_norm": 0.6043084859848022, "learning_rate": 0.0005, "loss": 1.3697212934494019, "step": 45055 }, { "epoch": 3.5519117067402446, "grad_norm": 0.5503479838371277, "learning_rate": 0.0005, "loss": 1.236129641532898, "step": 45056 }, { "epoch": 3.5519905400078833, "grad_norm": 0.6278415322303772, "learning_rate": 0.0005, "loss": 1.4649357795715332, "step": 45057 }, { "epoch": 3.5520693732755224, "grad_norm": 0.47380250692367554, "learning_rate": 0.0005, "loss": 1.2434251308441162, "step": 45058 }, { "epoch": 3.552148206543161, "grad_norm": 0.5533325672149658, "learning_rate": 0.0005, "loss": 1.3340449333190918, "step": 45059 }, { "epoch": 3.5522270398108002, "grad_norm": 0.4724031686782837, "learning_rate": 0.0005, "loss": 1.2490473985671997, "step": 45060 }, { "epoch": 3.5523058730784394, "grad_norm": 0.47086772322654724, "learning_rate": 0.0005, "loss": 1.2652692794799805, "step": 45061 }, { "epoch": 3.552384706346078, "grad_norm": 0.676795482635498, "learning_rate": 0.0005, "loss": 1.409902572631836, "step": 45062 }, { "epoch": 3.5524635396137167, "grad_norm": 0.4267607629299164, "learning_rate": 0.0005, "loss": 1.2997055053710938, "step": 45063 }, { "epoch": 3.552542372881356, "grad_norm": 0.6790305972099304, "learning_rate": 0.0005, "loss": 1.4244165420532227, "step": 45064 }, { "epoch": 3.552621206148995, "grad_norm": 1.1720606088638306, "learning_rate": 0.0005, "loss": 1.3771765232086182, "step": 45065 }, { "epoch": 3.5527000394166337, "grad_norm": 0.5848211646080017, "learning_rate": 0.0005, "loss": 1.340609073638916, "step": 45066 }, { "epoch": 3.552778872684273, "grad_norm": 0.5614550709724426, "learning_rate": 0.0005, "loss": 1.2955055236816406, "step": 45067 }, { "epoch": 3.5528577059519115, "grad_norm": 0.5047444105148315, "learning_rate": 0.0005, "loss": 1.3423552513122559, "step": 45068 }, { "epoch": 3.5529365392195507, "grad_norm": 0.6244706511497498, "learning_rate": 0.0005, "loss": 1.222090244293213, "step": 45069 }, { "epoch": 3.55301537248719, "grad_norm": 0.5021032094955444, "learning_rate": 0.0005, "loss": 1.2237229347229004, "step": 45070 }, { "epoch": 3.5530942057548285, "grad_norm": 0.5912142992019653, "learning_rate": 0.0005, "loss": 1.2560057640075684, "step": 45071 }, { "epoch": 3.5531730390224676, "grad_norm": 0.5566813945770264, "learning_rate": 0.0005, "loss": 1.3143486976623535, "step": 45072 }, { "epoch": 3.5532518722901063, "grad_norm": 0.5319991111755371, "learning_rate": 0.0005, "loss": 1.2654752731323242, "step": 45073 }, { "epoch": 3.5533307055577454, "grad_norm": 0.4868931770324707, "learning_rate": 0.0005, "loss": 1.329600214958191, "step": 45074 }, { "epoch": 3.5534095388253846, "grad_norm": 0.6706181168556213, "learning_rate": 0.0005, "loss": 1.463851809501648, "step": 45075 }, { "epoch": 3.5534883720930233, "grad_norm": 0.5439965128898621, "learning_rate": 0.0005, "loss": 1.2797250747680664, "step": 45076 }, { "epoch": 3.553567205360662, "grad_norm": 0.549312174320221, "learning_rate": 0.0005, "loss": 1.4319212436676025, "step": 45077 }, { "epoch": 3.553646038628301, "grad_norm": 0.4240540564060211, "learning_rate": 0.0005, "loss": 1.084676742553711, "step": 45078 }, { "epoch": 3.55372487189594, "grad_norm": 0.45359644293785095, "learning_rate": 0.0005, "loss": 1.204298496246338, "step": 45079 }, { "epoch": 3.553803705163579, "grad_norm": 0.4873272180557251, "learning_rate": 0.0005, "loss": 1.3794184923171997, "step": 45080 }, { "epoch": 3.553882538431218, "grad_norm": 0.45315179228782654, "learning_rate": 0.0005, "loss": 1.3849388360977173, "step": 45081 }, { "epoch": 3.5539613716988567, "grad_norm": 0.42628711462020874, "learning_rate": 0.0005, "loss": 1.0978589057922363, "step": 45082 }, { "epoch": 3.554040204966496, "grad_norm": 0.5572453737258911, "learning_rate": 0.0005, "loss": 1.245295524597168, "step": 45083 }, { "epoch": 3.554119038234135, "grad_norm": 0.5178634524345398, "learning_rate": 0.0005, "loss": 1.4080700874328613, "step": 45084 }, { "epoch": 3.5541978715017737, "grad_norm": 0.529926598072052, "learning_rate": 0.0005, "loss": 1.4512933492660522, "step": 45085 }, { "epoch": 3.554276704769413, "grad_norm": 0.4301275312900543, "learning_rate": 0.0005, "loss": 1.1863398551940918, "step": 45086 }, { "epoch": 3.5543555380370515, "grad_norm": 0.48028743267059326, "learning_rate": 0.0005, "loss": 1.3172639608383179, "step": 45087 }, { "epoch": 3.5544343713046906, "grad_norm": 0.6512249708175659, "learning_rate": 0.0005, "loss": 1.4742099046707153, "step": 45088 }, { "epoch": 3.5545132045723298, "grad_norm": 0.506247878074646, "learning_rate": 0.0005, "loss": 1.537987470626831, "step": 45089 }, { "epoch": 3.5545920378399685, "grad_norm": 0.5102134943008423, "learning_rate": 0.0005, "loss": 1.3712971210479736, "step": 45090 }, { "epoch": 3.554670871107607, "grad_norm": 0.7215264439582825, "learning_rate": 0.0005, "loss": 1.402930736541748, "step": 45091 }, { "epoch": 3.5547497043752463, "grad_norm": 0.40270721912384033, "learning_rate": 0.0005, "loss": 1.124166488647461, "step": 45092 }, { "epoch": 3.5548285376428854, "grad_norm": 0.4889290928840637, "learning_rate": 0.0005, "loss": 1.182498812675476, "step": 45093 }, { "epoch": 3.554907370910524, "grad_norm": 0.4620359539985657, "learning_rate": 0.0005, "loss": 1.1640446186065674, "step": 45094 }, { "epoch": 3.5549862041781632, "grad_norm": 0.42927858233451843, "learning_rate": 0.0005, "loss": 1.2642078399658203, "step": 45095 }, { "epoch": 3.555065037445802, "grad_norm": 0.43830275535583496, "learning_rate": 0.0005, "loss": 1.1755226850509644, "step": 45096 }, { "epoch": 3.555143870713441, "grad_norm": 0.6051668524742126, "learning_rate": 0.0005, "loss": 1.0835041999816895, "step": 45097 }, { "epoch": 3.55522270398108, "grad_norm": 0.4461215138435364, "learning_rate": 0.0005, "loss": 1.1883418560028076, "step": 45098 }, { "epoch": 3.555301537248719, "grad_norm": 0.5501947999000549, "learning_rate": 0.0005, "loss": 1.3238999843597412, "step": 45099 }, { "epoch": 3.555380370516358, "grad_norm": 0.4519168734550476, "learning_rate": 0.0005, "loss": 1.329245686531067, "step": 45100 }, { "epoch": 3.5554592037839967, "grad_norm": 0.4356434941291809, "learning_rate": 0.0005, "loss": 1.2228741645812988, "step": 45101 }, { "epoch": 3.555538037051636, "grad_norm": 0.6516059041023254, "learning_rate": 0.0005, "loss": 1.1710171699523926, "step": 45102 }, { "epoch": 3.555616870319275, "grad_norm": 0.44552287459373474, "learning_rate": 0.0005, "loss": 1.2746167182922363, "step": 45103 }, { "epoch": 3.5556957035869137, "grad_norm": 0.5131562948226929, "learning_rate": 0.0005, "loss": 1.1080249547958374, "step": 45104 }, { "epoch": 3.5557745368545524, "grad_norm": 0.48300689458847046, "learning_rate": 0.0005, "loss": 1.0525377988815308, "step": 45105 }, { "epoch": 3.5558533701221915, "grad_norm": 0.49058276414871216, "learning_rate": 0.0005, "loss": 1.2390539646148682, "step": 45106 }, { "epoch": 3.5559322033898306, "grad_norm": 0.4045112431049347, "learning_rate": 0.0005, "loss": 1.1611990928649902, "step": 45107 }, { "epoch": 3.5560110366574693, "grad_norm": 0.5585899949073792, "learning_rate": 0.0005, "loss": 1.4521015882492065, "step": 45108 }, { "epoch": 3.5560898699251084, "grad_norm": 0.6851410865783691, "learning_rate": 0.0005, "loss": 1.4671682119369507, "step": 45109 }, { "epoch": 3.556168703192747, "grad_norm": 0.4964209496974945, "learning_rate": 0.0005, "loss": 1.4474260807037354, "step": 45110 }, { "epoch": 3.5562475364603863, "grad_norm": 0.5620427131652832, "learning_rate": 0.0005, "loss": 1.441184401512146, "step": 45111 }, { "epoch": 3.5563263697280254, "grad_norm": 0.41713812947273254, "learning_rate": 0.0005, "loss": 1.1204239130020142, "step": 45112 }, { "epoch": 3.556405202995664, "grad_norm": 0.46176284551620483, "learning_rate": 0.0005, "loss": 1.1796523332595825, "step": 45113 }, { "epoch": 3.5564840362633032, "grad_norm": 0.588527262210846, "learning_rate": 0.0005, "loss": 1.350699782371521, "step": 45114 }, { "epoch": 3.556562869530942, "grad_norm": 0.5085123181343079, "learning_rate": 0.0005, "loss": 1.077441930770874, "step": 45115 }, { "epoch": 3.556641702798581, "grad_norm": 0.5020694136619568, "learning_rate": 0.0005, "loss": 1.1336578130722046, "step": 45116 }, { "epoch": 3.55672053606622, "grad_norm": 0.5412705540657043, "learning_rate": 0.0005, "loss": 1.3024755716323853, "step": 45117 }, { "epoch": 3.556799369333859, "grad_norm": 0.5525301694869995, "learning_rate": 0.0005, "loss": 1.2722830772399902, "step": 45118 }, { "epoch": 3.5568782026014976, "grad_norm": 0.5282907485961914, "learning_rate": 0.0005, "loss": 1.138701319694519, "step": 45119 }, { "epoch": 3.5569570358691367, "grad_norm": 0.5463440418243408, "learning_rate": 0.0005, "loss": 1.4852616786956787, "step": 45120 }, { "epoch": 3.557035869136776, "grad_norm": 0.5547150373458862, "learning_rate": 0.0005, "loss": 1.1093511581420898, "step": 45121 }, { "epoch": 3.5571147024044145, "grad_norm": 0.4837704598903656, "learning_rate": 0.0005, "loss": 1.448015570640564, "step": 45122 }, { "epoch": 3.5571935356720537, "grad_norm": 0.686010479927063, "learning_rate": 0.0005, "loss": 1.345609426498413, "step": 45123 }, { "epoch": 3.5572723689396923, "grad_norm": 0.6230019330978394, "learning_rate": 0.0005, "loss": 1.431509017944336, "step": 45124 }, { "epoch": 3.5573512022073315, "grad_norm": 0.4861557185649872, "learning_rate": 0.0005, "loss": 1.4522879123687744, "step": 45125 }, { "epoch": 3.5574300354749706, "grad_norm": 0.4550827741622925, "learning_rate": 0.0005, "loss": 1.2108147144317627, "step": 45126 }, { "epoch": 3.5575088687426093, "grad_norm": 0.6732864379882812, "learning_rate": 0.0005, "loss": 1.171116828918457, "step": 45127 }, { "epoch": 3.5575877020102484, "grad_norm": 0.4229256212711334, "learning_rate": 0.0005, "loss": 1.1747620105743408, "step": 45128 }, { "epoch": 3.557666535277887, "grad_norm": 0.45566996932029724, "learning_rate": 0.0005, "loss": 1.2608085870742798, "step": 45129 }, { "epoch": 3.5577453685455263, "grad_norm": 0.5617508292198181, "learning_rate": 0.0005, "loss": 1.3433339595794678, "step": 45130 }, { "epoch": 3.5578242018131654, "grad_norm": 0.569222092628479, "learning_rate": 0.0005, "loss": 1.1436210870742798, "step": 45131 }, { "epoch": 3.557903035080804, "grad_norm": 0.5295220613479614, "learning_rate": 0.0005, "loss": 1.3629944324493408, "step": 45132 }, { "epoch": 3.5579818683484428, "grad_norm": 0.457011342048645, "learning_rate": 0.0005, "loss": 1.4435508251190186, "step": 45133 }, { "epoch": 3.558060701616082, "grad_norm": 0.4487841725349426, "learning_rate": 0.0005, "loss": 1.272512674331665, "step": 45134 }, { "epoch": 3.558139534883721, "grad_norm": 0.5522942543029785, "learning_rate": 0.0005, "loss": 1.4881322383880615, "step": 45135 }, { "epoch": 3.5582183681513597, "grad_norm": 0.5381942987442017, "learning_rate": 0.0005, "loss": 1.2515517473220825, "step": 45136 }, { "epoch": 3.558297201418999, "grad_norm": 0.536293089389801, "learning_rate": 0.0005, "loss": 1.3901102542877197, "step": 45137 }, { "epoch": 3.5583760346866375, "grad_norm": 0.5509329438209534, "learning_rate": 0.0005, "loss": 1.3412059545516968, "step": 45138 }, { "epoch": 3.5584548679542767, "grad_norm": 0.6344010829925537, "learning_rate": 0.0005, "loss": 1.4061484336853027, "step": 45139 }, { "epoch": 3.558533701221916, "grad_norm": 0.45673561096191406, "learning_rate": 0.0005, "loss": 1.2827649116516113, "step": 45140 }, { "epoch": 3.5586125344895545, "grad_norm": 0.5055440664291382, "learning_rate": 0.0005, "loss": 1.2940508127212524, "step": 45141 }, { "epoch": 3.5586913677571936, "grad_norm": 0.6266221404075623, "learning_rate": 0.0005, "loss": 1.3581228256225586, "step": 45142 }, { "epoch": 3.5587702010248323, "grad_norm": 0.5350914001464844, "learning_rate": 0.0005, "loss": 1.1772347688674927, "step": 45143 }, { "epoch": 3.5588490342924715, "grad_norm": 0.4637223184108734, "learning_rate": 0.0005, "loss": 1.1711822748184204, "step": 45144 }, { "epoch": 3.5589278675601106, "grad_norm": 0.53917396068573, "learning_rate": 0.0005, "loss": 1.361875057220459, "step": 45145 }, { "epoch": 3.5590067008277493, "grad_norm": 0.43709906935691833, "learning_rate": 0.0005, "loss": 1.2216039896011353, "step": 45146 }, { "epoch": 3.5590855340953884, "grad_norm": 0.4724973738193512, "learning_rate": 0.0005, "loss": 1.2149766683578491, "step": 45147 }, { "epoch": 3.559164367363027, "grad_norm": 0.5832125544548035, "learning_rate": 0.0005, "loss": 1.2691023349761963, "step": 45148 }, { "epoch": 3.5592432006306662, "grad_norm": 0.6039808392524719, "learning_rate": 0.0005, "loss": 1.3175115585327148, "step": 45149 }, { "epoch": 3.559322033898305, "grad_norm": 0.5274396538734436, "learning_rate": 0.0005, "loss": 1.582348346710205, "step": 45150 }, { "epoch": 3.559400867165944, "grad_norm": 0.6642350554466248, "learning_rate": 0.0005, "loss": 1.4870637655258179, "step": 45151 }, { "epoch": 3.5594797004335827, "grad_norm": 0.4542836546897888, "learning_rate": 0.0005, "loss": 1.3611197471618652, "step": 45152 }, { "epoch": 3.559558533701222, "grad_norm": 0.4846554398536682, "learning_rate": 0.0005, "loss": 1.2110648155212402, "step": 45153 }, { "epoch": 3.559637366968861, "grad_norm": 0.40574491024017334, "learning_rate": 0.0005, "loss": 1.0804554224014282, "step": 45154 }, { "epoch": 3.5597162002364997, "grad_norm": 0.49441438913345337, "learning_rate": 0.0005, "loss": 1.1724605560302734, "step": 45155 }, { "epoch": 3.559795033504139, "grad_norm": 0.5848673582077026, "learning_rate": 0.0005, "loss": 1.3608087301254272, "step": 45156 }, { "epoch": 3.5598738667717775, "grad_norm": 0.544625997543335, "learning_rate": 0.0005, "loss": 1.3072896003723145, "step": 45157 }, { "epoch": 3.5599527000394167, "grad_norm": 0.5014657378196716, "learning_rate": 0.0005, "loss": 1.53641676902771, "step": 45158 }, { "epoch": 3.560031533307056, "grad_norm": 0.4735645055770874, "learning_rate": 0.0005, "loss": 1.1015857458114624, "step": 45159 }, { "epoch": 3.5601103665746945, "grad_norm": 0.4874095022678375, "learning_rate": 0.0005, "loss": 1.2177540063858032, "step": 45160 }, { "epoch": 3.5601891998423336, "grad_norm": 0.4851941466331482, "learning_rate": 0.0005, "loss": 1.2087445259094238, "step": 45161 }, { "epoch": 3.5602680331099723, "grad_norm": 0.4499934911727905, "learning_rate": 0.0005, "loss": 1.1628391742706299, "step": 45162 }, { "epoch": 3.5603468663776114, "grad_norm": 0.5284795165061951, "learning_rate": 0.0005, "loss": 1.3707213401794434, "step": 45163 }, { "epoch": 3.56042569964525, "grad_norm": 0.46784383058547974, "learning_rate": 0.0005, "loss": 1.3748259544372559, "step": 45164 }, { "epoch": 3.5605045329128893, "grad_norm": 0.5811028480529785, "learning_rate": 0.0005, "loss": 1.3338913917541504, "step": 45165 }, { "epoch": 3.560583366180528, "grad_norm": 0.5278616547584534, "learning_rate": 0.0005, "loss": 1.3824008703231812, "step": 45166 }, { "epoch": 3.560662199448167, "grad_norm": 0.5326409935951233, "learning_rate": 0.0005, "loss": 1.1513913869857788, "step": 45167 }, { "epoch": 3.560741032715806, "grad_norm": 0.5278043746948242, "learning_rate": 0.0005, "loss": 1.0533666610717773, "step": 45168 }, { "epoch": 3.560819865983445, "grad_norm": 0.5745352506637573, "learning_rate": 0.0005, "loss": 1.2790539264678955, "step": 45169 }, { "epoch": 3.560898699251084, "grad_norm": 0.53709876537323, "learning_rate": 0.0005, "loss": 1.296314001083374, "step": 45170 }, { "epoch": 3.5609775325187227, "grad_norm": 0.5220707654953003, "learning_rate": 0.0005, "loss": 1.336432695388794, "step": 45171 }, { "epoch": 3.561056365786362, "grad_norm": 0.40593206882476807, "learning_rate": 0.0005, "loss": 1.1508166790008545, "step": 45172 }, { "epoch": 3.561135199054001, "grad_norm": 0.5926831960678101, "learning_rate": 0.0005, "loss": 1.427262783050537, "step": 45173 }, { "epoch": 3.5612140323216397, "grad_norm": 0.6195935606956482, "learning_rate": 0.0005, "loss": 1.205034613609314, "step": 45174 }, { "epoch": 3.561292865589279, "grad_norm": 0.6754181981086731, "learning_rate": 0.0005, "loss": 1.4082815647125244, "step": 45175 }, { "epoch": 3.5613716988569175, "grad_norm": 0.49369141459465027, "learning_rate": 0.0005, "loss": 1.1437488794326782, "step": 45176 }, { "epoch": 3.5614505321245566, "grad_norm": 0.5868231654167175, "learning_rate": 0.0005, "loss": 1.3548011779785156, "step": 45177 }, { "epoch": 3.561529365392196, "grad_norm": 0.4636881649494171, "learning_rate": 0.0005, "loss": 1.1025526523590088, "step": 45178 }, { "epoch": 3.5616081986598345, "grad_norm": 0.4645721912384033, "learning_rate": 0.0005, "loss": 1.1545729637145996, "step": 45179 }, { "epoch": 3.561687031927473, "grad_norm": 0.7221872806549072, "learning_rate": 0.0005, "loss": 1.5342081785202026, "step": 45180 }, { "epoch": 3.5617658651951123, "grad_norm": 0.5938200354576111, "learning_rate": 0.0005, "loss": 1.328195571899414, "step": 45181 }, { "epoch": 3.5618446984627514, "grad_norm": 0.4272399842739105, "learning_rate": 0.0005, "loss": 1.2778230905532837, "step": 45182 }, { "epoch": 3.56192353173039, "grad_norm": 0.5084415078163147, "learning_rate": 0.0005, "loss": 1.2418193817138672, "step": 45183 }, { "epoch": 3.5620023649980292, "grad_norm": 0.46844667196273804, "learning_rate": 0.0005, "loss": 1.365983486175537, "step": 45184 }, { "epoch": 3.562081198265668, "grad_norm": 0.5786079168319702, "learning_rate": 0.0005, "loss": 1.3119101524353027, "step": 45185 }, { "epoch": 3.562160031533307, "grad_norm": 0.5226276516914368, "learning_rate": 0.0005, "loss": 1.4490630626678467, "step": 45186 }, { "epoch": 3.562238864800946, "grad_norm": 0.5867170095443726, "learning_rate": 0.0005, "loss": 1.4280977249145508, "step": 45187 }, { "epoch": 3.562317698068585, "grad_norm": 0.48806747794151306, "learning_rate": 0.0005, "loss": 1.2411774396896362, "step": 45188 }, { "epoch": 3.562396531336224, "grad_norm": 0.4205285906791687, "learning_rate": 0.0005, "loss": 1.1721925735473633, "step": 45189 }, { "epoch": 3.5624753646038627, "grad_norm": 0.4754857122898102, "learning_rate": 0.0005, "loss": 1.3105515241622925, "step": 45190 }, { "epoch": 3.562554197871502, "grad_norm": 0.5061383247375488, "learning_rate": 0.0005, "loss": 1.1242784261703491, "step": 45191 }, { "epoch": 3.562633031139141, "grad_norm": 0.7682978510856628, "learning_rate": 0.0005, "loss": 1.2651495933532715, "step": 45192 }, { "epoch": 3.5627118644067797, "grad_norm": 0.4508899450302124, "learning_rate": 0.0005, "loss": 1.3006556034088135, "step": 45193 }, { "epoch": 3.5627906976744184, "grad_norm": 0.42809465527534485, "learning_rate": 0.0005, "loss": 1.3624694347381592, "step": 45194 }, { "epoch": 3.5628695309420575, "grad_norm": 0.5963426232337952, "learning_rate": 0.0005, "loss": 1.4718003273010254, "step": 45195 }, { "epoch": 3.5629483642096966, "grad_norm": 0.5415094494819641, "learning_rate": 0.0005, "loss": 1.2940986156463623, "step": 45196 }, { "epoch": 3.5630271974773353, "grad_norm": 0.7418596148490906, "learning_rate": 0.0005, "loss": 1.1189576387405396, "step": 45197 }, { "epoch": 3.5631060307449745, "grad_norm": 0.6311092376708984, "learning_rate": 0.0005, "loss": 1.4132795333862305, "step": 45198 }, { "epoch": 3.563184864012613, "grad_norm": 0.5692657232284546, "learning_rate": 0.0005, "loss": 1.2597804069519043, "step": 45199 }, { "epoch": 3.5632636972802523, "grad_norm": 0.44093161821365356, "learning_rate": 0.0005, "loss": 1.267263412475586, "step": 45200 }, { "epoch": 3.5633425305478914, "grad_norm": 0.47838684916496277, "learning_rate": 0.0005, "loss": 1.3212398290634155, "step": 45201 }, { "epoch": 3.56342136381553, "grad_norm": 0.4947865605354309, "learning_rate": 0.0005, "loss": 1.2921490669250488, "step": 45202 }, { "epoch": 3.5635001970831692, "grad_norm": 0.4555822014808655, "learning_rate": 0.0005, "loss": 1.3102258443832397, "step": 45203 }, { "epoch": 3.563579030350808, "grad_norm": 0.6608754992485046, "learning_rate": 0.0005, "loss": 1.2645140886306763, "step": 45204 }, { "epoch": 3.563657863618447, "grad_norm": 0.5168376564979553, "learning_rate": 0.0005, "loss": 1.3013019561767578, "step": 45205 }, { "epoch": 3.563736696886086, "grad_norm": 0.49473854899406433, "learning_rate": 0.0005, "loss": 1.3360183238983154, "step": 45206 }, { "epoch": 3.563815530153725, "grad_norm": 0.4821803867816925, "learning_rate": 0.0005, "loss": 1.3037399053573608, "step": 45207 }, { "epoch": 3.5638943634213636, "grad_norm": 0.5218814611434937, "learning_rate": 0.0005, "loss": 1.3360360860824585, "step": 45208 }, { "epoch": 3.5639731966890027, "grad_norm": 0.45477449893951416, "learning_rate": 0.0005, "loss": 1.0665569305419922, "step": 45209 }, { "epoch": 3.564052029956642, "grad_norm": 0.4933435022830963, "learning_rate": 0.0005, "loss": 1.2954224348068237, "step": 45210 }, { "epoch": 3.5641308632242805, "grad_norm": 0.6260561943054199, "learning_rate": 0.0005, "loss": 1.4898414611816406, "step": 45211 }, { "epoch": 3.5642096964919197, "grad_norm": 0.5030451416969299, "learning_rate": 0.0005, "loss": 1.1993424892425537, "step": 45212 }, { "epoch": 3.5642885297595583, "grad_norm": 0.5329750776290894, "learning_rate": 0.0005, "loss": 1.3933883905410767, "step": 45213 }, { "epoch": 3.5643673630271975, "grad_norm": 1.4289214611053467, "learning_rate": 0.0005, "loss": 1.2689342498779297, "step": 45214 }, { "epoch": 3.5644461962948366, "grad_norm": 0.44615277647972107, "learning_rate": 0.0005, "loss": 1.3320949077606201, "step": 45215 }, { "epoch": 3.5645250295624753, "grad_norm": 0.4041118025779724, "learning_rate": 0.0005, "loss": 1.218078374862671, "step": 45216 }, { "epoch": 3.5646038628301144, "grad_norm": 0.5459672212600708, "learning_rate": 0.0005, "loss": 1.429681420326233, "step": 45217 }, { "epoch": 3.564682696097753, "grad_norm": 0.7047765254974365, "learning_rate": 0.0005, "loss": 1.4312596321105957, "step": 45218 }, { "epoch": 3.5647615293653923, "grad_norm": 0.438377320766449, "learning_rate": 0.0005, "loss": 1.0661907196044922, "step": 45219 }, { "epoch": 3.5648403626330314, "grad_norm": 0.5621938109397888, "learning_rate": 0.0005, "loss": 1.1855952739715576, "step": 45220 }, { "epoch": 3.56491919590067, "grad_norm": 0.4500686228275299, "learning_rate": 0.0005, "loss": 1.3063735961914062, "step": 45221 }, { "epoch": 3.5649980291683088, "grad_norm": 0.4741863012313843, "learning_rate": 0.0005, "loss": 1.337113618850708, "step": 45222 }, { "epoch": 3.565076862435948, "grad_norm": 0.4334835708141327, "learning_rate": 0.0005, "loss": 1.2229119539260864, "step": 45223 }, { "epoch": 3.565155695703587, "grad_norm": 0.4104450047016144, "learning_rate": 0.0005, "loss": 1.2805652618408203, "step": 45224 }, { "epoch": 3.5652345289712257, "grad_norm": 0.537163496017456, "learning_rate": 0.0005, "loss": 1.3344335556030273, "step": 45225 }, { "epoch": 3.565313362238865, "grad_norm": 0.49024102091789246, "learning_rate": 0.0005, "loss": 1.1862894296646118, "step": 45226 }, { "epoch": 3.5653921955065035, "grad_norm": 0.5499979853630066, "learning_rate": 0.0005, "loss": 1.2164287567138672, "step": 45227 }, { "epoch": 3.5654710287741427, "grad_norm": 0.548349142074585, "learning_rate": 0.0005, "loss": 1.177367925643921, "step": 45228 }, { "epoch": 3.565549862041782, "grad_norm": 0.43923261761665344, "learning_rate": 0.0005, "loss": 1.0674880743026733, "step": 45229 }, { "epoch": 3.5656286953094205, "grad_norm": 0.3871088922023773, "learning_rate": 0.0005, "loss": 1.0801142454147339, "step": 45230 }, { "epoch": 3.5657075285770596, "grad_norm": 0.41049274802207947, "learning_rate": 0.0005, "loss": 1.093259572982788, "step": 45231 }, { "epoch": 3.5657863618446983, "grad_norm": 0.5906007885932922, "learning_rate": 0.0005, "loss": 1.355938196182251, "step": 45232 }, { "epoch": 3.5658651951123375, "grad_norm": 0.4876682162284851, "learning_rate": 0.0005, "loss": 1.383928656578064, "step": 45233 }, { "epoch": 3.5659440283799766, "grad_norm": 0.46908217668533325, "learning_rate": 0.0005, "loss": 1.1381456851959229, "step": 45234 }, { "epoch": 3.5660228616476153, "grad_norm": 0.410416841506958, "learning_rate": 0.0005, "loss": 1.0192899703979492, "step": 45235 }, { "epoch": 3.566101694915254, "grad_norm": 0.4351522922515869, "learning_rate": 0.0005, "loss": 1.2081857919692993, "step": 45236 }, { "epoch": 3.566180528182893, "grad_norm": 0.5598049759864807, "learning_rate": 0.0005, "loss": 1.2705692052841187, "step": 45237 }, { "epoch": 3.5662593614505322, "grad_norm": 0.5380088686943054, "learning_rate": 0.0005, "loss": 1.3184599876403809, "step": 45238 }, { "epoch": 3.566338194718171, "grad_norm": 0.445100873708725, "learning_rate": 0.0005, "loss": 1.1791682243347168, "step": 45239 }, { "epoch": 3.56641702798581, "grad_norm": 0.7226912975311279, "learning_rate": 0.0005, "loss": 1.2432951927185059, "step": 45240 }, { "epoch": 3.5664958612534488, "grad_norm": 0.4157835841178894, "learning_rate": 0.0005, "loss": 1.027052640914917, "step": 45241 }, { "epoch": 3.566574694521088, "grad_norm": 0.47589731216430664, "learning_rate": 0.0005, "loss": 1.145977258682251, "step": 45242 }, { "epoch": 3.566653527788727, "grad_norm": 0.5579482316970825, "learning_rate": 0.0005, "loss": 1.3468818664550781, "step": 45243 }, { "epoch": 3.5667323610563657, "grad_norm": 0.4770376682281494, "learning_rate": 0.0005, "loss": 1.347692847251892, "step": 45244 }, { "epoch": 3.566811194324005, "grad_norm": 0.7169802188873291, "learning_rate": 0.0005, "loss": 1.3157694339752197, "step": 45245 }, { "epoch": 3.5668900275916435, "grad_norm": 0.507165789604187, "learning_rate": 0.0005, "loss": 1.3064680099487305, "step": 45246 }, { "epoch": 3.5669688608592827, "grad_norm": 0.33477091789245605, "learning_rate": 0.0005, "loss": 1.0249037742614746, "step": 45247 }, { "epoch": 3.567047694126922, "grad_norm": 0.46454447507858276, "learning_rate": 0.0005, "loss": 1.3799887895584106, "step": 45248 }, { "epoch": 3.5671265273945605, "grad_norm": 0.4422523081302643, "learning_rate": 0.0005, "loss": 1.1546070575714111, "step": 45249 }, { "epoch": 3.567205360662199, "grad_norm": 0.46553531289100647, "learning_rate": 0.0005, "loss": 1.456045389175415, "step": 45250 }, { "epoch": 3.5672841939298383, "grad_norm": 0.44020503759384155, "learning_rate": 0.0005, "loss": 1.248932123184204, "step": 45251 }, { "epoch": 3.5673630271974774, "grad_norm": 1.1670697927474976, "learning_rate": 0.0005, "loss": 1.158825397491455, "step": 45252 }, { "epoch": 3.567441860465116, "grad_norm": 0.514388382434845, "learning_rate": 0.0005, "loss": 1.3258261680603027, "step": 45253 }, { "epoch": 3.5675206937327553, "grad_norm": 0.564494252204895, "learning_rate": 0.0005, "loss": 1.3294652700424194, "step": 45254 }, { "epoch": 3.567599527000394, "grad_norm": 0.5248504281044006, "learning_rate": 0.0005, "loss": 0.9832471013069153, "step": 45255 }, { "epoch": 3.567678360268033, "grad_norm": 0.4947240650653839, "learning_rate": 0.0005, "loss": 1.438279390335083, "step": 45256 }, { "epoch": 3.5677571935356722, "grad_norm": 0.4306790828704834, "learning_rate": 0.0005, "loss": 1.205732822418213, "step": 45257 }, { "epoch": 3.567836026803311, "grad_norm": 0.3943689465522766, "learning_rate": 0.0005, "loss": 1.1073651313781738, "step": 45258 }, { "epoch": 3.56791486007095, "grad_norm": 0.5105262994766235, "learning_rate": 0.0005, "loss": 1.3562147617340088, "step": 45259 }, { "epoch": 3.5679936933385887, "grad_norm": 0.5430404543876648, "learning_rate": 0.0005, "loss": 1.2827587127685547, "step": 45260 }, { "epoch": 3.568072526606228, "grad_norm": 0.4562393128871918, "learning_rate": 0.0005, "loss": 1.2086970806121826, "step": 45261 }, { "epoch": 3.568151359873867, "grad_norm": 0.49436697363853455, "learning_rate": 0.0005, "loss": 1.2843130826950073, "step": 45262 }, { "epoch": 3.5682301931415057, "grad_norm": 0.41522642970085144, "learning_rate": 0.0005, "loss": 1.196022868156433, "step": 45263 }, { "epoch": 3.568309026409145, "grad_norm": 0.42119285464286804, "learning_rate": 0.0005, "loss": 1.1402356624603271, "step": 45264 }, { "epoch": 3.5683878596767835, "grad_norm": 0.45072200894355774, "learning_rate": 0.0005, "loss": 1.309807538986206, "step": 45265 }, { "epoch": 3.5684666929444226, "grad_norm": 0.4879697263240814, "learning_rate": 0.0005, "loss": 1.1851218938827515, "step": 45266 }, { "epoch": 3.5685455262120613, "grad_norm": 0.49821481108665466, "learning_rate": 0.0005, "loss": 1.2130478620529175, "step": 45267 }, { "epoch": 3.5686243594797005, "grad_norm": 0.6827571392059326, "learning_rate": 0.0005, "loss": 1.1145515441894531, "step": 45268 }, { "epoch": 3.568703192747339, "grad_norm": 0.4889225363731384, "learning_rate": 0.0005, "loss": 1.3667668104171753, "step": 45269 }, { "epoch": 3.5687820260149783, "grad_norm": 0.5147418975830078, "learning_rate": 0.0005, "loss": 1.2040077447891235, "step": 45270 }, { "epoch": 3.5688608592826174, "grad_norm": 0.5104078650474548, "learning_rate": 0.0005, "loss": 1.3542604446411133, "step": 45271 }, { "epoch": 3.568939692550256, "grad_norm": 0.4969497621059418, "learning_rate": 0.0005, "loss": 1.2276825904846191, "step": 45272 }, { "epoch": 3.5690185258178952, "grad_norm": 0.46864667534828186, "learning_rate": 0.0005, "loss": 1.4085428714752197, "step": 45273 }, { "epoch": 3.569097359085534, "grad_norm": 0.6084023118019104, "learning_rate": 0.0005, "loss": 1.4587230682373047, "step": 45274 }, { "epoch": 3.569176192353173, "grad_norm": 0.5308482646942139, "learning_rate": 0.0005, "loss": 1.2958526611328125, "step": 45275 }, { "epoch": 3.569255025620812, "grad_norm": 0.5199843049049377, "learning_rate": 0.0005, "loss": 1.3148107528686523, "step": 45276 }, { "epoch": 3.569333858888451, "grad_norm": 0.5890226364135742, "learning_rate": 0.0005, "loss": 1.4445456266403198, "step": 45277 }, { "epoch": 3.56941269215609, "grad_norm": 0.476321280002594, "learning_rate": 0.0005, "loss": 1.1285984516143799, "step": 45278 }, { "epoch": 3.5694915254237287, "grad_norm": 0.4610666334629059, "learning_rate": 0.0005, "loss": 1.2302768230438232, "step": 45279 }, { "epoch": 3.569570358691368, "grad_norm": 0.4943952262401581, "learning_rate": 0.0005, "loss": 1.1967202425003052, "step": 45280 }, { "epoch": 3.5696491919590065, "grad_norm": 0.6478649973869324, "learning_rate": 0.0005, "loss": 1.3129212856292725, "step": 45281 }, { "epoch": 3.5697280252266457, "grad_norm": 0.600059449672699, "learning_rate": 0.0005, "loss": 1.4976493120193481, "step": 45282 }, { "epoch": 3.5698068584942844, "grad_norm": 0.5103821158409119, "learning_rate": 0.0005, "loss": 1.2205314636230469, "step": 45283 }, { "epoch": 3.5698856917619235, "grad_norm": 0.4845653474330902, "learning_rate": 0.0005, "loss": 1.2195849418640137, "step": 45284 }, { "epoch": 3.5699645250295626, "grad_norm": 0.5614420175552368, "learning_rate": 0.0005, "loss": 1.450602412223816, "step": 45285 }, { "epoch": 3.5700433582972013, "grad_norm": 0.5281779766082764, "learning_rate": 0.0005, "loss": 1.5687025785446167, "step": 45286 }, { "epoch": 3.5701221915648405, "grad_norm": 0.47533565759658813, "learning_rate": 0.0005, "loss": 1.2275214195251465, "step": 45287 }, { "epoch": 3.570201024832479, "grad_norm": 0.43895581364631653, "learning_rate": 0.0005, "loss": 1.1331686973571777, "step": 45288 }, { "epoch": 3.5702798581001183, "grad_norm": 0.4506877064704895, "learning_rate": 0.0005, "loss": 1.3610703945159912, "step": 45289 }, { "epoch": 3.5703586913677574, "grad_norm": 0.44530951976776123, "learning_rate": 0.0005, "loss": 1.2762980461120605, "step": 45290 }, { "epoch": 3.570437524635396, "grad_norm": 2.458470106124878, "learning_rate": 0.0005, "loss": 1.1404370069503784, "step": 45291 }, { "epoch": 3.5705163579030352, "grad_norm": 0.4336091876029968, "learning_rate": 0.0005, "loss": 1.1756210327148438, "step": 45292 }, { "epoch": 3.570595191170674, "grad_norm": 0.5305756330490112, "learning_rate": 0.0005, "loss": 1.3316304683685303, "step": 45293 }, { "epoch": 3.570674024438313, "grad_norm": 0.5412078499794006, "learning_rate": 0.0005, "loss": 1.313296914100647, "step": 45294 }, { "epoch": 3.570752857705952, "grad_norm": 0.5438179969787598, "learning_rate": 0.0005, "loss": 1.4053475856781006, "step": 45295 }, { "epoch": 3.570831690973591, "grad_norm": 0.48217281699180603, "learning_rate": 0.0005, "loss": 1.2095003128051758, "step": 45296 }, { "epoch": 3.5709105242412296, "grad_norm": 0.4529084265232086, "learning_rate": 0.0005, "loss": 1.0849767923355103, "step": 45297 }, { "epoch": 3.5709893575088687, "grad_norm": 0.45501509308815, "learning_rate": 0.0005, "loss": 1.2607202529907227, "step": 45298 }, { "epoch": 3.571068190776508, "grad_norm": 0.5798186659812927, "learning_rate": 0.0005, "loss": 1.433593988418579, "step": 45299 }, { "epoch": 3.5711470240441465, "grad_norm": 0.4556259214878082, "learning_rate": 0.0005, "loss": 1.3497064113616943, "step": 45300 }, { "epoch": 3.5712258573117857, "grad_norm": 0.5027058720588684, "learning_rate": 0.0005, "loss": 1.2710999250411987, "step": 45301 }, { "epoch": 3.5713046905794243, "grad_norm": 0.45577487349510193, "learning_rate": 0.0005, "loss": 1.3247389793395996, "step": 45302 }, { "epoch": 3.5713835238470635, "grad_norm": 0.5034227967262268, "learning_rate": 0.0005, "loss": 1.3067681789398193, "step": 45303 }, { "epoch": 3.5714623571147026, "grad_norm": 0.4299364686012268, "learning_rate": 0.0005, "loss": 1.3221633434295654, "step": 45304 }, { "epoch": 3.5715411903823413, "grad_norm": 0.395571768283844, "learning_rate": 0.0005, "loss": 1.0765751600265503, "step": 45305 }, { "epoch": 3.5716200236499804, "grad_norm": 0.4265781342983246, "learning_rate": 0.0005, "loss": 1.261396050453186, "step": 45306 }, { "epoch": 3.571698856917619, "grad_norm": 0.5235298275947571, "learning_rate": 0.0005, "loss": 1.321744680404663, "step": 45307 }, { "epoch": 3.5717776901852583, "grad_norm": 0.5220184922218323, "learning_rate": 0.0005, "loss": 1.349283218383789, "step": 45308 }, { "epoch": 3.5718565234528974, "grad_norm": 0.4470118284225464, "learning_rate": 0.0005, "loss": 1.02190101146698, "step": 45309 }, { "epoch": 3.571935356720536, "grad_norm": 0.43107497692108154, "learning_rate": 0.0005, "loss": 1.3114778995513916, "step": 45310 }, { "epoch": 3.5720141899881748, "grad_norm": 0.4400104880332947, "learning_rate": 0.0005, "loss": 1.2359950542449951, "step": 45311 }, { "epoch": 3.572093023255814, "grad_norm": 0.44929471611976624, "learning_rate": 0.0005, "loss": 1.347976565361023, "step": 45312 }, { "epoch": 3.572171856523453, "grad_norm": 0.4592088460922241, "learning_rate": 0.0005, "loss": 1.2864054441452026, "step": 45313 }, { "epoch": 3.5722506897910917, "grad_norm": 0.562370777130127, "learning_rate": 0.0005, "loss": 1.220481038093567, "step": 45314 }, { "epoch": 3.572329523058731, "grad_norm": 0.608647346496582, "learning_rate": 0.0005, "loss": 1.2763314247131348, "step": 45315 }, { "epoch": 3.5724083563263695, "grad_norm": 0.48140430450439453, "learning_rate": 0.0005, "loss": 1.1639666557312012, "step": 45316 }, { "epoch": 3.5724871895940087, "grad_norm": 0.4195913076400757, "learning_rate": 0.0005, "loss": 1.1774775981903076, "step": 45317 }, { "epoch": 3.572566022861648, "grad_norm": 0.4304448366165161, "learning_rate": 0.0005, "loss": 1.3065325021743774, "step": 45318 }, { "epoch": 3.5726448561292865, "grad_norm": 0.5790470242500305, "learning_rate": 0.0005, "loss": 1.3868529796600342, "step": 45319 }, { "epoch": 3.5727236893969256, "grad_norm": 0.5095722079277039, "learning_rate": 0.0005, "loss": 1.1967113018035889, "step": 45320 }, { "epoch": 3.5728025226645643, "grad_norm": 0.5629351139068604, "learning_rate": 0.0005, "loss": 1.299597144126892, "step": 45321 }, { "epoch": 3.5728813559322035, "grad_norm": 0.5490640997886658, "learning_rate": 0.0005, "loss": 1.1922019720077515, "step": 45322 }, { "epoch": 3.5729601891998426, "grad_norm": 0.4966754615306854, "learning_rate": 0.0005, "loss": 1.5055978298187256, "step": 45323 }, { "epoch": 3.5730390224674813, "grad_norm": 0.4572281837463379, "learning_rate": 0.0005, "loss": 1.1700924634933472, "step": 45324 }, { "epoch": 3.57311785573512, "grad_norm": 1.205889344215393, "learning_rate": 0.0005, "loss": 1.1081697940826416, "step": 45325 }, { "epoch": 3.573196689002759, "grad_norm": 0.6549040079116821, "learning_rate": 0.0005, "loss": 1.279325246810913, "step": 45326 }, { "epoch": 3.5732755222703982, "grad_norm": 0.44254186749458313, "learning_rate": 0.0005, "loss": 1.2291408777236938, "step": 45327 }, { "epoch": 3.573354355538037, "grad_norm": 0.4764098823070526, "learning_rate": 0.0005, "loss": 1.3697593212127686, "step": 45328 }, { "epoch": 3.573433188805676, "grad_norm": 0.5390852093696594, "learning_rate": 0.0005, "loss": 1.3242785930633545, "step": 45329 }, { "epoch": 3.5735120220733148, "grad_norm": 0.6597087383270264, "learning_rate": 0.0005, "loss": 1.4287183284759521, "step": 45330 }, { "epoch": 3.573590855340954, "grad_norm": 0.5843197703361511, "learning_rate": 0.0005, "loss": 1.186037540435791, "step": 45331 }, { "epoch": 3.573669688608593, "grad_norm": 0.4797787070274353, "learning_rate": 0.0005, "loss": 1.3170573711395264, "step": 45332 }, { "epoch": 3.5737485218762317, "grad_norm": 0.5228732824325562, "learning_rate": 0.0005, "loss": 1.2884070873260498, "step": 45333 }, { "epoch": 3.573827355143871, "grad_norm": 0.40828606486320496, "learning_rate": 0.0005, "loss": 1.164902687072754, "step": 45334 }, { "epoch": 3.5739061884115095, "grad_norm": 0.4986039102077484, "learning_rate": 0.0005, "loss": 1.1002452373504639, "step": 45335 }, { "epoch": 3.5739850216791487, "grad_norm": 0.667829692363739, "learning_rate": 0.0005, "loss": 1.3550283908843994, "step": 45336 }, { "epoch": 3.574063854946788, "grad_norm": 0.5091684460639954, "learning_rate": 0.0005, "loss": 1.2239142656326294, "step": 45337 }, { "epoch": 3.5741426882144265, "grad_norm": 0.6134393811225891, "learning_rate": 0.0005, "loss": 1.4531424045562744, "step": 45338 }, { "epoch": 3.574221521482065, "grad_norm": 0.5762622356414795, "learning_rate": 0.0005, "loss": 1.3615896701812744, "step": 45339 }, { "epoch": 3.5743003547497043, "grad_norm": 0.42715221643447876, "learning_rate": 0.0005, "loss": 1.2175228595733643, "step": 45340 }, { "epoch": 3.5743791880173434, "grad_norm": 0.528055727481842, "learning_rate": 0.0005, "loss": 1.2132229804992676, "step": 45341 }, { "epoch": 3.574458021284982, "grad_norm": 0.4585726261138916, "learning_rate": 0.0005, "loss": 1.2676829099655151, "step": 45342 }, { "epoch": 3.5745368545526213, "grad_norm": 0.535606861114502, "learning_rate": 0.0005, "loss": 1.4300920963287354, "step": 45343 }, { "epoch": 3.57461568782026, "grad_norm": 1.9363194704055786, "learning_rate": 0.0005, "loss": 1.3348169326782227, "step": 45344 }, { "epoch": 3.574694521087899, "grad_norm": 0.5032480955123901, "learning_rate": 0.0005, "loss": 1.2828978300094604, "step": 45345 }, { "epoch": 3.5747733543555382, "grad_norm": 0.4652049243450165, "learning_rate": 0.0005, "loss": 1.2343765497207642, "step": 45346 }, { "epoch": 3.574852187623177, "grad_norm": 0.4789966940879822, "learning_rate": 0.0005, "loss": 1.2407785654067993, "step": 45347 }, { "epoch": 3.574931020890816, "grad_norm": 0.6835442781448364, "learning_rate": 0.0005, "loss": 1.3424116373062134, "step": 45348 }, { "epoch": 3.5750098541584547, "grad_norm": 0.6371070146560669, "learning_rate": 0.0005, "loss": 1.346115231513977, "step": 45349 }, { "epoch": 3.575088687426094, "grad_norm": 0.49011915922164917, "learning_rate": 0.0005, "loss": 1.1849334239959717, "step": 45350 }, { "epoch": 3.575167520693733, "grad_norm": 0.7148013114929199, "learning_rate": 0.0005, "loss": 1.4862293004989624, "step": 45351 }, { "epoch": 3.5752463539613717, "grad_norm": 0.552703857421875, "learning_rate": 0.0005, "loss": 1.1632781028747559, "step": 45352 }, { "epoch": 3.5753251872290104, "grad_norm": 0.5789190530776978, "learning_rate": 0.0005, "loss": 1.4886583089828491, "step": 45353 }, { "epoch": 3.5754040204966495, "grad_norm": 1.2340102195739746, "learning_rate": 0.0005, "loss": 1.35858154296875, "step": 45354 }, { "epoch": 3.5754828537642886, "grad_norm": 0.5854961276054382, "learning_rate": 0.0005, "loss": 1.3413017988204956, "step": 45355 }, { "epoch": 3.5755616870319273, "grad_norm": 0.4572935700416565, "learning_rate": 0.0005, "loss": 1.2966570854187012, "step": 45356 }, { "epoch": 3.5756405202995665, "grad_norm": 0.7108116745948792, "learning_rate": 0.0005, "loss": 1.2065818309783936, "step": 45357 }, { "epoch": 3.575719353567205, "grad_norm": 0.4431653916835785, "learning_rate": 0.0005, "loss": 1.2209100723266602, "step": 45358 }, { "epoch": 3.5757981868348443, "grad_norm": 0.607379674911499, "learning_rate": 0.0005, "loss": 1.3703862428665161, "step": 45359 }, { "epoch": 3.5758770201024834, "grad_norm": 0.44918039441108704, "learning_rate": 0.0005, "loss": 1.2155390977859497, "step": 45360 }, { "epoch": 3.575955853370122, "grad_norm": 0.5036497712135315, "learning_rate": 0.0005, "loss": 1.2601261138916016, "step": 45361 }, { "epoch": 3.5760346866377613, "grad_norm": 0.44933995604515076, "learning_rate": 0.0005, "loss": 1.3081555366516113, "step": 45362 }, { "epoch": 3.5761135199054, "grad_norm": 0.4942130446434021, "learning_rate": 0.0005, "loss": 1.2118538618087769, "step": 45363 }, { "epoch": 3.576192353173039, "grad_norm": 0.5099126100540161, "learning_rate": 0.0005, "loss": 1.3235297203063965, "step": 45364 }, { "epoch": 3.576271186440678, "grad_norm": 0.5274810791015625, "learning_rate": 0.0005, "loss": 1.4052221775054932, "step": 45365 }, { "epoch": 3.576350019708317, "grad_norm": 0.550609290599823, "learning_rate": 0.0005, "loss": 1.4315582513809204, "step": 45366 }, { "epoch": 3.5764288529759556, "grad_norm": 0.7100643515586853, "learning_rate": 0.0005, "loss": 1.120758056640625, "step": 45367 }, { "epoch": 3.5765076862435947, "grad_norm": 0.4869360625743866, "learning_rate": 0.0005, "loss": 1.3806843757629395, "step": 45368 }, { "epoch": 3.576586519511234, "grad_norm": 0.49999383091926575, "learning_rate": 0.0005, "loss": 1.2095452547073364, "step": 45369 }, { "epoch": 3.5766653527788725, "grad_norm": 0.49752873182296753, "learning_rate": 0.0005, "loss": 1.25530207157135, "step": 45370 }, { "epoch": 3.5767441860465117, "grad_norm": 0.4410158097743988, "learning_rate": 0.0005, "loss": 1.275869369506836, "step": 45371 }, { "epoch": 3.5768230193141504, "grad_norm": 0.5302706956863403, "learning_rate": 0.0005, "loss": 1.3215672969818115, "step": 45372 }, { "epoch": 3.5769018525817895, "grad_norm": 0.43607965111732483, "learning_rate": 0.0005, "loss": 1.1955697536468506, "step": 45373 }, { "epoch": 3.5769806858494286, "grad_norm": 0.5550905466079712, "learning_rate": 0.0005, "loss": 1.298419713973999, "step": 45374 }, { "epoch": 3.5770595191170673, "grad_norm": 0.4348103702068329, "learning_rate": 0.0005, "loss": 1.2114681005477905, "step": 45375 }, { "epoch": 3.5771383523847065, "grad_norm": 0.5239408016204834, "learning_rate": 0.0005, "loss": 1.2581223249435425, "step": 45376 }, { "epoch": 3.577217185652345, "grad_norm": 0.5229962468147278, "learning_rate": 0.0005, "loss": 1.3755080699920654, "step": 45377 }, { "epoch": 3.5772960189199843, "grad_norm": 0.5607466697692871, "learning_rate": 0.0005, "loss": 1.5309982299804688, "step": 45378 }, { "epoch": 3.5773748521876234, "grad_norm": 0.5276429653167725, "learning_rate": 0.0005, "loss": 1.3554306030273438, "step": 45379 }, { "epoch": 3.577453685455262, "grad_norm": 0.4378189146518707, "learning_rate": 0.0005, "loss": 1.2036781311035156, "step": 45380 }, { "epoch": 3.577532518722901, "grad_norm": 0.5017520189285278, "learning_rate": 0.0005, "loss": 1.3801802396774292, "step": 45381 }, { "epoch": 3.57761135199054, "grad_norm": 0.520917534828186, "learning_rate": 0.0005, "loss": 1.3154370784759521, "step": 45382 }, { "epoch": 3.577690185258179, "grad_norm": 0.45158541202545166, "learning_rate": 0.0005, "loss": 1.2836053371429443, "step": 45383 }, { "epoch": 3.5777690185258177, "grad_norm": 0.41216975450515747, "learning_rate": 0.0005, "loss": 1.3783122301101685, "step": 45384 }, { "epoch": 3.577847851793457, "grad_norm": 0.5285618305206299, "learning_rate": 0.0005, "loss": 1.3821512460708618, "step": 45385 }, { "epoch": 3.5779266850610956, "grad_norm": 0.5619550943374634, "learning_rate": 0.0005, "loss": 1.364012598991394, "step": 45386 }, { "epoch": 3.5780055183287347, "grad_norm": 0.5774295330047607, "learning_rate": 0.0005, "loss": 1.4398844242095947, "step": 45387 }, { "epoch": 3.578084351596374, "grad_norm": 0.4461331367492676, "learning_rate": 0.0005, "loss": 1.1726608276367188, "step": 45388 }, { "epoch": 3.5781631848640125, "grad_norm": 0.4691827893257141, "learning_rate": 0.0005, "loss": 1.1909013986587524, "step": 45389 }, { "epoch": 3.5782420181316517, "grad_norm": 0.4345785677433014, "learning_rate": 0.0005, "loss": 1.2790154218673706, "step": 45390 }, { "epoch": 3.5783208513992903, "grad_norm": 0.5404095649719238, "learning_rate": 0.0005, "loss": 1.2598503828048706, "step": 45391 }, { "epoch": 3.5783996846669295, "grad_norm": 0.44221267104148865, "learning_rate": 0.0005, "loss": 1.2150523662567139, "step": 45392 }, { "epoch": 3.5784785179345686, "grad_norm": 0.43314042687416077, "learning_rate": 0.0005, "loss": 1.2193388938903809, "step": 45393 }, { "epoch": 3.5785573512022073, "grad_norm": 0.3867049813270569, "learning_rate": 0.0005, "loss": 1.249867558479309, "step": 45394 }, { "epoch": 3.5786361844698464, "grad_norm": 0.43322715163230896, "learning_rate": 0.0005, "loss": 1.062787652015686, "step": 45395 }, { "epoch": 3.578715017737485, "grad_norm": 0.47869399189949036, "learning_rate": 0.0005, "loss": 1.3104071617126465, "step": 45396 }, { "epoch": 3.5787938510051243, "grad_norm": 0.4948878586292267, "learning_rate": 0.0005, "loss": 1.2627285718917847, "step": 45397 }, { "epoch": 3.578872684272763, "grad_norm": 0.4912055730819702, "learning_rate": 0.0005, "loss": 1.3955628871917725, "step": 45398 }, { "epoch": 3.578951517540402, "grad_norm": 0.5598193407058716, "learning_rate": 0.0005, "loss": 1.4495724439620972, "step": 45399 }, { "epoch": 3.5790303508080408, "grad_norm": 0.4401855766773224, "learning_rate": 0.0005, "loss": 1.226430058479309, "step": 45400 }, { "epoch": 3.57910918407568, "grad_norm": 0.49608877301216125, "learning_rate": 0.0005, "loss": 1.2255859375, "step": 45401 }, { "epoch": 3.579188017343319, "grad_norm": 0.5021622180938721, "learning_rate": 0.0005, "loss": 1.336445927619934, "step": 45402 }, { "epoch": 3.5792668506109577, "grad_norm": 0.5309399962425232, "learning_rate": 0.0005, "loss": 1.2338683605194092, "step": 45403 }, { "epoch": 3.579345683878597, "grad_norm": 0.5599076151847839, "learning_rate": 0.0005, "loss": 1.351836919784546, "step": 45404 }, { "epoch": 3.5794245171462356, "grad_norm": 0.4370618164539337, "learning_rate": 0.0005, "loss": 1.2285786867141724, "step": 45405 }, { "epoch": 3.5795033504138747, "grad_norm": 0.5378546714782715, "learning_rate": 0.0005, "loss": 1.2203583717346191, "step": 45406 }, { "epoch": 3.579582183681514, "grad_norm": 0.4214490056037903, "learning_rate": 0.0005, "loss": 1.4021039009094238, "step": 45407 }, { "epoch": 3.5796610169491525, "grad_norm": 0.46370556950569153, "learning_rate": 0.0005, "loss": 1.4458895921707153, "step": 45408 }, { "epoch": 3.5797398502167916, "grad_norm": 0.41217201948165894, "learning_rate": 0.0005, "loss": 1.169553518295288, "step": 45409 }, { "epoch": 3.5798186834844303, "grad_norm": 0.44097110629081726, "learning_rate": 0.0005, "loss": 1.2199469804763794, "step": 45410 }, { "epoch": 3.5798975167520695, "grad_norm": 0.3945016860961914, "learning_rate": 0.0005, "loss": 1.233506441116333, "step": 45411 }, { "epoch": 3.579976350019708, "grad_norm": 0.47408175468444824, "learning_rate": 0.0005, "loss": 1.139363408088684, "step": 45412 }, { "epoch": 3.5800551832873473, "grad_norm": 0.5266460180282593, "learning_rate": 0.0005, "loss": 1.3238754272460938, "step": 45413 }, { "epoch": 3.580134016554986, "grad_norm": 0.5679187774658203, "learning_rate": 0.0005, "loss": 1.4555764198303223, "step": 45414 }, { "epoch": 3.580212849822625, "grad_norm": 0.5749555826187134, "learning_rate": 0.0005, "loss": 1.2461731433868408, "step": 45415 }, { "epoch": 3.5802916830902642, "grad_norm": 0.47919002175331116, "learning_rate": 0.0005, "loss": 1.1431175470352173, "step": 45416 }, { "epoch": 3.580370516357903, "grad_norm": 0.5525847673416138, "learning_rate": 0.0005, "loss": 1.3436843156814575, "step": 45417 }, { "epoch": 3.580449349625542, "grad_norm": 0.4264158606529236, "learning_rate": 0.0005, "loss": 1.20667564868927, "step": 45418 }, { "epoch": 3.5805281828931808, "grad_norm": 0.5227060914039612, "learning_rate": 0.0005, "loss": 1.4376081228256226, "step": 45419 }, { "epoch": 3.58060701616082, "grad_norm": 0.599452018737793, "learning_rate": 0.0005, "loss": 1.3031513690948486, "step": 45420 }, { "epoch": 3.580685849428459, "grad_norm": 0.5237711668014526, "learning_rate": 0.0005, "loss": 1.291198968887329, "step": 45421 }, { "epoch": 3.5807646826960977, "grad_norm": 0.5421304702758789, "learning_rate": 0.0005, "loss": 1.2927258014678955, "step": 45422 }, { "epoch": 3.580843515963737, "grad_norm": 0.4990127682685852, "learning_rate": 0.0005, "loss": 1.5580335855484009, "step": 45423 }, { "epoch": 3.5809223492313755, "grad_norm": 0.42607712745666504, "learning_rate": 0.0005, "loss": 1.2759735584259033, "step": 45424 }, { "epoch": 3.5810011824990147, "grad_norm": 0.4076789319515228, "learning_rate": 0.0005, "loss": 1.1631073951721191, "step": 45425 }, { "epoch": 3.581080015766654, "grad_norm": 0.5527409911155701, "learning_rate": 0.0005, "loss": 1.391591191291809, "step": 45426 }, { "epoch": 3.5811588490342925, "grad_norm": 0.6082825064659119, "learning_rate": 0.0005, "loss": 1.3412829637527466, "step": 45427 }, { "epoch": 3.581237682301931, "grad_norm": 0.4575500786304474, "learning_rate": 0.0005, "loss": 1.1360245943069458, "step": 45428 }, { "epoch": 3.5813165155695703, "grad_norm": 0.46552908420562744, "learning_rate": 0.0005, "loss": 1.2117899656295776, "step": 45429 }, { "epoch": 3.5813953488372094, "grad_norm": 0.4681676924228668, "learning_rate": 0.0005, "loss": 1.4042384624481201, "step": 45430 }, { "epoch": 3.581474182104848, "grad_norm": 0.6465947031974792, "learning_rate": 0.0005, "loss": 1.2775566577911377, "step": 45431 }, { "epoch": 3.5815530153724873, "grad_norm": 0.4739609658718109, "learning_rate": 0.0005, "loss": 1.2673548460006714, "step": 45432 }, { "epoch": 3.581631848640126, "grad_norm": 0.44021791219711304, "learning_rate": 0.0005, "loss": 1.2650914192199707, "step": 45433 }, { "epoch": 3.581710681907765, "grad_norm": 0.5026242733001709, "learning_rate": 0.0005, "loss": 1.2046977281570435, "step": 45434 }, { "epoch": 3.5817895151754042, "grad_norm": 0.48209232091903687, "learning_rate": 0.0005, "loss": 1.2825820446014404, "step": 45435 }, { "epoch": 3.581868348443043, "grad_norm": 0.504200279712677, "learning_rate": 0.0005, "loss": 1.1901891231536865, "step": 45436 }, { "epoch": 3.581947181710682, "grad_norm": 0.5674180388450623, "learning_rate": 0.0005, "loss": 1.381899118423462, "step": 45437 }, { "epoch": 3.5820260149783207, "grad_norm": 0.5772111415863037, "learning_rate": 0.0005, "loss": 1.3756825923919678, "step": 45438 }, { "epoch": 3.58210484824596, "grad_norm": 0.5050142407417297, "learning_rate": 0.0005, "loss": 1.250128149986267, "step": 45439 }, { "epoch": 3.582183681513599, "grad_norm": 0.5385072827339172, "learning_rate": 0.0005, "loss": 1.3192479610443115, "step": 45440 }, { "epoch": 3.5822625147812377, "grad_norm": 0.509751558303833, "learning_rate": 0.0005, "loss": 1.2161328792572021, "step": 45441 }, { "epoch": 3.5823413480488764, "grad_norm": 0.4293012320995331, "learning_rate": 0.0005, "loss": 1.1694419384002686, "step": 45442 }, { "epoch": 3.5824201813165155, "grad_norm": 0.5955722332000732, "learning_rate": 0.0005, "loss": 1.2691377401351929, "step": 45443 }, { "epoch": 3.5824990145841547, "grad_norm": 0.38729026913642883, "learning_rate": 0.0005, "loss": 1.1690592765808105, "step": 45444 }, { "epoch": 3.5825778478517933, "grad_norm": 0.6639812588691711, "learning_rate": 0.0005, "loss": 1.436297059059143, "step": 45445 }, { "epoch": 3.5826566811194325, "grad_norm": 0.4298672080039978, "learning_rate": 0.0005, "loss": 1.354229211807251, "step": 45446 }, { "epoch": 3.582735514387071, "grad_norm": 0.49894365668296814, "learning_rate": 0.0005, "loss": 1.1641814708709717, "step": 45447 }, { "epoch": 3.5828143476547103, "grad_norm": 0.5760447978973389, "learning_rate": 0.0005, "loss": 1.282752275466919, "step": 45448 }, { "epoch": 3.5828931809223494, "grad_norm": 0.45700353384017944, "learning_rate": 0.0005, "loss": 1.30942702293396, "step": 45449 }, { "epoch": 3.582972014189988, "grad_norm": 0.5586785674095154, "learning_rate": 0.0005, "loss": 1.275684118270874, "step": 45450 }, { "epoch": 3.5830508474576273, "grad_norm": 0.5327780246734619, "learning_rate": 0.0005, "loss": 1.4120416641235352, "step": 45451 }, { "epoch": 3.583129680725266, "grad_norm": 0.5347539186477661, "learning_rate": 0.0005, "loss": 1.0687077045440674, "step": 45452 }, { "epoch": 3.583208513992905, "grad_norm": 0.47279733419418335, "learning_rate": 0.0005, "loss": 1.3027660846710205, "step": 45453 }, { "epoch": 3.583287347260544, "grad_norm": 0.45222774147987366, "learning_rate": 0.0005, "loss": 1.1704257726669312, "step": 45454 }, { "epoch": 3.583366180528183, "grad_norm": 0.500872015953064, "learning_rate": 0.0005, "loss": 1.2950456142425537, "step": 45455 }, { "epoch": 3.5834450137958216, "grad_norm": 0.6931667923927307, "learning_rate": 0.0005, "loss": 1.221466064453125, "step": 45456 }, { "epoch": 3.5835238470634607, "grad_norm": 0.47043243050575256, "learning_rate": 0.0005, "loss": 1.2366199493408203, "step": 45457 }, { "epoch": 3.5836026803311, "grad_norm": 0.5638597011566162, "learning_rate": 0.0005, "loss": 1.4308325052261353, "step": 45458 }, { "epoch": 3.5836815135987385, "grad_norm": 0.4912787079811096, "learning_rate": 0.0005, "loss": 1.2630908489227295, "step": 45459 }, { "epoch": 3.5837603468663777, "grad_norm": 0.457076758146286, "learning_rate": 0.0005, "loss": 1.3138459920883179, "step": 45460 }, { "epoch": 3.5838391801340164, "grad_norm": 0.4466817378997803, "learning_rate": 0.0005, "loss": 1.302025318145752, "step": 45461 }, { "epoch": 3.5839180134016555, "grad_norm": 0.5223470330238342, "learning_rate": 0.0005, "loss": 1.348302960395813, "step": 45462 }, { "epoch": 3.5839968466692946, "grad_norm": 0.4566299021244049, "learning_rate": 0.0005, "loss": 1.1706351041793823, "step": 45463 }, { "epoch": 3.5840756799369333, "grad_norm": 0.4455818831920624, "learning_rate": 0.0005, "loss": 1.1846843957901, "step": 45464 }, { "epoch": 3.5841545132045725, "grad_norm": 0.4374377727508545, "learning_rate": 0.0005, "loss": 1.2643886804580688, "step": 45465 }, { "epoch": 3.584233346472211, "grad_norm": 0.44592344760894775, "learning_rate": 0.0005, "loss": 1.253477692604065, "step": 45466 }, { "epoch": 3.5843121797398503, "grad_norm": 0.5640086531639099, "learning_rate": 0.0005, "loss": 1.451547622680664, "step": 45467 }, { "epoch": 3.5843910130074894, "grad_norm": 0.4892263114452362, "learning_rate": 0.0005, "loss": 1.2898640632629395, "step": 45468 }, { "epoch": 3.584469846275128, "grad_norm": 0.5089342594146729, "learning_rate": 0.0005, "loss": 1.1893374919891357, "step": 45469 }, { "epoch": 3.584548679542767, "grad_norm": 0.5189995169639587, "learning_rate": 0.0005, "loss": 1.2732160091400146, "step": 45470 }, { "epoch": 3.584627512810406, "grad_norm": 0.4067547023296356, "learning_rate": 0.0005, "loss": 1.1600937843322754, "step": 45471 }, { "epoch": 3.584706346078045, "grad_norm": 0.39437800645828247, "learning_rate": 0.0005, "loss": 1.1070950031280518, "step": 45472 }, { "epoch": 3.5847851793456837, "grad_norm": 0.598099410533905, "learning_rate": 0.0005, "loss": 1.3838415145874023, "step": 45473 }, { "epoch": 3.584864012613323, "grad_norm": 0.5865551829338074, "learning_rate": 0.0005, "loss": 1.2604374885559082, "step": 45474 }, { "epoch": 3.5849428458809616, "grad_norm": 0.4682390093803406, "learning_rate": 0.0005, "loss": 1.1810271739959717, "step": 45475 }, { "epoch": 3.5850216791486007, "grad_norm": 0.6123450398445129, "learning_rate": 0.0005, "loss": 1.4266319274902344, "step": 45476 }, { "epoch": 3.58510051241624, "grad_norm": 0.44440287351608276, "learning_rate": 0.0005, "loss": 1.235154151916504, "step": 45477 }, { "epoch": 3.5851793456838785, "grad_norm": 0.41201645135879517, "learning_rate": 0.0005, "loss": 1.0901824235916138, "step": 45478 }, { "epoch": 3.5852581789515177, "grad_norm": 0.5096240639686584, "learning_rate": 0.0005, "loss": 1.3739802837371826, "step": 45479 }, { "epoch": 3.5853370122191563, "grad_norm": 0.6146255135536194, "learning_rate": 0.0005, "loss": 1.3134634494781494, "step": 45480 }, { "epoch": 3.5854158454867955, "grad_norm": 0.621833860874176, "learning_rate": 0.0005, "loss": 1.263318657875061, "step": 45481 }, { "epoch": 3.5854946787544346, "grad_norm": 0.5364928841590881, "learning_rate": 0.0005, "loss": 1.5517616271972656, "step": 45482 }, { "epoch": 3.5855735120220733, "grad_norm": 0.524772047996521, "learning_rate": 0.0005, "loss": 1.2794357538223267, "step": 45483 }, { "epoch": 3.585652345289712, "grad_norm": 0.45462387800216675, "learning_rate": 0.0005, "loss": 1.1029555797576904, "step": 45484 }, { "epoch": 3.585731178557351, "grad_norm": 0.41741856932640076, "learning_rate": 0.0005, "loss": 1.1690011024475098, "step": 45485 }, { "epoch": 3.5858100118249903, "grad_norm": 0.6079079508781433, "learning_rate": 0.0005, "loss": 1.2790822982788086, "step": 45486 }, { "epoch": 3.585888845092629, "grad_norm": 0.487872838973999, "learning_rate": 0.0005, "loss": 1.2371125221252441, "step": 45487 }, { "epoch": 3.585967678360268, "grad_norm": 0.5677239298820496, "learning_rate": 0.0005, "loss": 1.2053271532058716, "step": 45488 }, { "epoch": 3.5860465116279068, "grad_norm": 0.4470195770263672, "learning_rate": 0.0005, "loss": 1.188941240310669, "step": 45489 }, { "epoch": 3.586125344895546, "grad_norm": 0.5787878632545471, "learning_rate": 0.0005, "loss": 1.3947789669036865, "step": 45490 }, { "epoch": 3.586204178163185, "grad_norm": 0.4271835684776306, "learning_rate": 0.0005, "loss": 1.129306674003601, "step": 45491 }, { "epoch": 3.5862830114308237, "grad_norm": 0.48879384994506836, "learning_rate": 0.0005, "loss": 1.3741309642791748, "step": 45492 }, { "epoch": 3.586361844698463, "grad_norm": 0.49111396074295044, "learning_rate": 0.0005, "loss": 1.22950279712677, "step": 45493 }, { "epoch": 3.5864406779661016, "grad_norm": 0.5945091843605042, "learning_rate": 0.0005, "loss": 1.583863377571106, "step": 45494 }, { "epoch": 3.5865195112337407, "grad_norm": 0.5759276747703552, "learning_rate": 0.0005, "loss": 1.270784854888916, "step": 45495 }, { "epoch": 3.58659834450138, "grad_norm": 0.46881088614463806, "learning_rate": 0.0005, "loss": 1.178105115890503, "step": 45496 }, { "epoch": 3.5866771777690185, "grad_norm": 0.5346462726593018, "learning_rate": 0.0005, "loss": 1.2270052433013916, "step": 45497 }, { "epoch": 3.586756011036657, "grad_norm": 0.5370460748672485, "learning_rate": 0.0005, "loss": 1.3828904628753662, "step": 45498 }, { "epoch": 3.5868348443042963, "grad_norm": 0.532934844493866, "learning_rate": 0.0005, "loss": 1.4305658340454102, "step": 45499 }, { "epoch": 3.5869136775719355, "grad_norm": 0.4276282489299774, "learning_rate": 0.0005, "loss": 1.2849043607711792, "step": 45500 }, { "epoch": 3.586992510839574, "grad_norm": 0.46818462014198303, "learning_rate": 0.0005, "loss": 1.1469905376434326, "step": 45501 }, { "epoch": 3.5870713441072133, "grad_norm": 0.3960469365119934, "learning_rate": 0.0005, "loss": 0.9341396689414978, "step": 45502 }, { "epoch": 3.587150177374852, "grad_norm": 0.3541354537010193, "learning_rate": 0.0005, "loss": 0.963657021522522, "step": 45503 }, { "epoch": 3.587229010642491, "grad_norm": 0.4762263000011444, "learning_rate": 0.0005, "loss": 1.0712621212005615, "step": 45504 }, { "epoch": 3.5873078439101302, "grad_norm": 0.5937535166740417, "learning_rate": 0.0005, "loss": 1.262606143951416, "step": 45505 }, { "epoch": 3.587386677177769, "grad_norm": 0.5314393639564514, "learning_rate": 0.0005, "loss": 1.3228764533996582, "step": 45506 }, { "epoch": 3.587465510445408, "grad_norm": 0.5043328404426575, "learning_rate": 0.0005, "loss": 1.3138983249664307, "step": 45507 }, { "epoch": 3.5875443437130468, "grad_norm": 0.46700161695480347, "learning_rate": 0.0005, "loss": 1.1286503076553345, "step": 45508 }, { "epoch": 3.587623176980686, "grad_norm": 0.48471617698669434, "learning_rate": 0.0005, "loss": 1.31292724609375, "step": 45509 }, { "epoch": 3.587702010248325, "grad_norm": 0.4853965938091278, "learning_rate": 0.0005, "loss": 1.1612731218338013, "step": 45510 }, { "epoch": 3.5877808435159637, "grad_norm": 0.46681520342826843, "learning_rate": 0.0005, "loss": 1.2980027198791504, "step": 45511 }, { "epoch": 3.5878596767836024, "grad_norm": 0.5269075632095337, "learning_rate": 0.0005, "loss": 1.2246278524398804, "step": 45512 }, { "epoch": 3.5879385100512415, "grad_norm": 0.4320446848869324, "learning_rate": 0.0005, "loss": 1.213039517402649, "step": 45513 }, { "epoch": 3.5880173433188807, "grad_norm": 0.4317951798439026, "learning_rate": 0.0005, "loss": 1.392883062362671, "step": 45514 }, { "epoch": 3.5880961765865194, "grad_norm": 0.5611048936843872, "learning_rate": 0.0005, "loss": 1.3217910528182983, "step": 45515 }, { "epoch": 3.5881750098541585, "grad_norm": 0.44185712933540344, "learning_rate": 0.0005, "loss": 1.252420425415039, "step": 45516 }, { "epoch": 3.588253843121797, "grad_norm": 0.465969979763031, "learning_rate": 0.0005, "loss": 1.1183254718780518, "step": 45517 }, { "epoch": 3.5883326763894363, "grad_norm": 0.4830148220062256, "learning_rate": 0.0005, "loss": 1.1941051483154297, "step": 45518 }, { "epoch": 3.5884115096570754, "grad_norm": 0.5070793032646179, "learning_rate": 0.0005, "loss": 1.4162344932556152, "step": 45519 }, { "epoch": 3.588490342924714, "grad_norm": 0.45110687613487244, "learning_rate": 0.0005, "loss": 1.1883103847503662, "step": 45520 }, { "epoch": 3.5885691761923533, "grad_norm": 0.6149730682373047, "learning_rate": 0.0005, "loss": 1.337856411933899, "step": 45521 }, { "epoch": 3.588648009459992, "grad_norm": 0.485667884349823, "learning_rate": 0.0005, "loss": 1.2594221830368042, "step": 45522 }, { "epoch": 3.588726842727631, "grad_norm": 0.46244317293167114, "learning_rate": 0.0005, "loss": 1.1169726848602295, "step": 45523 }, { "epoch": 3.5888056759952702, "grad_norm": 0.519569993019104, "learning_rate": 0.0005, "loss": 1.4138970375061035, "step": 45524 }, { "epoch": 3.588884509262909, "grad_norm": 0.49592599272727966, "learning_rate": 0.0005, "loss": 1.1462182998657227, "step": 45525 }, { "epoch": 3.588963342530548, "grad_norm": 0.4517197608947754, "learning_rate": 0.0005, "loss": 1.2020957469940186, "step": 45526 }, { "epoch": 3.5890421757981867, "grad_norm": 0.46413689851760864, "learning_rate": 0.0005, "loss": 1.18589448928833, "step": 45527 }, { "epoch": 3.589121009065826, "grad_norm": 0.5476076006889343, "learning_rate": 0.0005, "loss": 1.3415756225585938, "step": 45528 }, { "epoch": 3.5891998423334646, "grad_norm": 0.4914010763168335, "learning_rate": 0.0005, "loss": 1.3995517492294312, "step": 45529 }, { "epoch": 3.5892786756011037, "grad_norm": 0.49757248163223267, "learning_rate": 0.0005, "loss": 1.359354853630066, "step": 45530 }, { "epoch": 3.5893575088687424, "grad_norm": 0.5591627359390259, "learning_rate": 0.0005, "loss": 1.2216556072235107, "step": 45531 }, { "epoch": 3.5894363421363815, "grad_norm": 0.5122225880622864, "learning_rate": 0.0005, "loss": 1.4341764450073242, "step": 45532 }, { "epoch": 3.5895151754040207, "grad_norm": 0.5030372142791748, "learning_rate": 0.0005, "loss": 1.2466169595718384, "step": 45533 }, { "epoch": 3.5895940086716593, "grad_norm": 0.4303937554359436, "learning_rate": 0.0005, "loss": 1.233310580253601, "step": 45534 }, { "epoch": 3.5896728419392985, "grad_norm": 0.46638426184654236, "learning_rate": 0.0005, "loss": 1.2790377140045166, "step": 45535 }, { "epoch": 3.589751675206937, "grad_norm": 0.6472895741462708, "learning_rate": 0.0005, "loss": 1.4467111825942993, "step": 45536 }, { "epoch": 3.5898305084745763, "grad_norm": 0.49679675698280334, "learning_rate": 0.0005, "loss": 1.281295657157898, "step": 45537 }, { "epoch": 3.5899093417422154, "grad_norm": 0.48592501878738403, "learning_rate": 0.0005, "loss": 1.3495434522628784, "step": 45538 }, { "epoch": 3.589988175009854, "grad_norm": 0.4714549779891968, "learning_rate": 0.0005, "loss": 1.2128932476043701, "step": 45539 }, { "epoch": 3.5900670082774933, "grad_norm": 0.453643262386322, "learning_rate": 0.0005, "loss": 1.4411818981170654, "step": 45540 }, { "epoch": 3.590145841545132, "grad_norm": 0.4879900813102722, "learning_rate": 0.0005, "loss": 1.2206952571868896, "step": 45541 }, { "epoch": 3.590224674812771, "grad_norm": 0.48517659306526184, "learning_rate": 0.0005, "loss": 1.1747643947601318, "step": 45542 }, { "epoch": 3.5903035080804098, "grad_norm": 0.5937714576721191, "learning_rate": 0.0005, "loss": 1.2021100521087646, "step": 45543 }, { "epoch": 3.590382341348049, "grad_norm": 0.5143617987632751, "learning_rate": 0.0005, "loss": 1.4097590446472168, "step": 45544 }, { "epoch": 3.5904611746156876, "grad_norm": 0.5874407291412354, "learning_rate": 0.0005, "loss": 1.293550968170166, "step": 45545 }, { "epoch": 3.5905400078833267, "grad_norm": 0.5204512476921082, "learning_rate": 0.0005, "loss": 1.2329492568969727, "step": 45546 }, { "epoch": 3.590618841150966, "grad_norm": 0.47072023153305054, "learning_rate": 0.0005, "loss": 1.2249239683151245, "step": 45547 }, { "epoch": 3.5906976744186045, "grad_norm": 0.5482580661773682, "learning_rate": 0.0005, "loss": 1.4302830696105957, "step": 45548 }, { "epoch": 3.5907765076862437, "grad_norm": 0.48479899764060974, "learning_rate": 0.0005, "loss": 1.2156769037246704, "step": 45549 }, { "epoch": 3.5908553409538824, "grad_norm": 0.3989731967449188, "learning_rate": 0.0005, "loss": 1.2030434608459473, "step": 45550 }, { "epoch": 3.5909341742215215, "grad_norm": 0.5002760291099548, "learning_rate": 0.0005, "loss": 1.4019739627838135, "step": 45551 }, { "epoch": 3.5910130074891606, "grad_norm": 0.5154070258140564, "learning_rate": 0.0005, "loss": 1.1702334880828857, "step": 45552 }, { "epoch": 3.5910918407567993, "grad_norm": 0.4883604645729065, "learning_rate": 0.0005, "loss": 1.2235734462738037, "step": 45553 }, { "epoch": 3.5911706740244385, "grad_norm": 0.4453212320804596, "learning_rate": 0.0005, "loss": 1.1805684566497803, "step": 45554 }, { "epoch": 3.591249507292077, "grad_norm": 0.4255988895893097, "learning_rate": 0.0005, "loss": 1.317915439605713, "step": 45555 }, { "epoch": 3.5913283405597163, "grad_norm": 0.42640358209609985, "learning_rate": 0.0005, "loss": 1.07227623462677, "step": 45556 }, { "epoch": 3.5914071738273554, "grad_norm": 0.7157442569732666, "learning_rate": 0.0005, "loss": 1.2896591424942017, "step": 45557 }, { "epoch": 3.591486007094994, "grad_norm": 0.4647504389286041, "learning_rate": 0.0005, "loss": 1.4405288696289062, "step": 45558 }, { "epoch": 3.591564840362633, "grad_norm": 0.40621036291122437, "learning_rate": 0.0005, "loss": 1.170662522315979, "step": 45559 }, { "epoch": 3.591643673630272, "grad_norm": 0.6638638377189636, "learning_rate": 0.0005, "loss": 1.4113942384719849, "step": 45560 }, { "epoch": 3.591722506897911, "grad_norm": 0.473025381565094, "learning_rate": 0.0005, "loss": 1.269526720046997, "step": 45561 }, { "epoch": 3.5918013401655497, "grad_norm": 0.5630242228507996, "learning_rate": 0.0005, "loss": 1.2824375629425049, "step": 45562 }, { "epoch": 3.591880173433189, "grad_norm": 0.5970638990402222, "learning_rate": 0.0005, "loss": 1.4603025913238525, "step": 45563 }, { "epoch": 3.5919590067008276, "grad_norm": 0.5197572708129883, "learning_rate": 0.0005, "loss": 1.2561737298965454, "step": 45564 }, { "epoch": 3.5920378399684667, "grad_norm": 0.4885713458061218, "learning_rate": 0.0005, "loss": 1.2779052257537842, "step": 45565 }, { "epoch": 3.592116673236106, "grad_norm": 0.45857468247413635, "learning_rate": 0.0005, "loss": 1.1441810131072998, "step": 45566 }, { "epoch": 3.5921955065037445, "grad_norm": 0.5469435453414917, "learning_rate": 0.0005, "loss": 1.3594821691513062, "step": 45567 }, { "epoch": 3.5922743397713837, "grad_norm": 2.333705425262451, "learning_rate": 0.0005, "loss": 1.3296806812286377, "step": 45568 }, { "epoch": 3.5923531730390224, "grad_norm": 0.5230145454406738, "learning_rate": 0.0005, "loss": 1.2385849952697754, "step": 45569 }, { "epoch": 3.5924320063066615, "grad_norm": 0.4879568815231323, "learning_rate": 0.0005, "loss": 1.3439338207244873, "step": 45570 }, { "epoch": 3.5925108395743006, "grad_norm": 0.502714216709137, "learning_rate": 0.0005, "loss": 1.2887613773345947, "step": 45571 }, { "epoch": 3.5925896728419393, "grad_norm": 0.5111032128334045, "learning_rate": 0.0005, "loss": 1.3230679035186768, "step": 45572 }, { "epoch": 3.592668506109578, "grad_norm": 0.483985036611557, "learning_rate": 0.0005, "loss": 1.3277742862701416, "step": 45573 }, { "epoch": 3.592747339377217, "grad_norm": 0.516488790512085, "learning_rate": 0.0005, "loss": 1.3135466575622559, "step": 45574 }, { "epoch": 3.5928261726448563, "grad_norm": 0.5080379843711853, "learning_rate": 0.0005, "loss": 1.4618788957595825, "step": 45575 }, { "epoch": 3.592905005912495, "grad_norm": 0.6840089559555054, "learning_rate": 0.0005, "loss": 1.2284560203552246, "step": 45576 }, { "epoch": 3.592983839180134, "grad_norm": 0.4935677945613861, "learning_rate": 0.0005, "loss": 1.073888897895813, "step": 45577 }, { "epoch": 3.5930626724477728, "grad_norm": 0.5365946292877197, "learning_rate": 0.0005, "loss": 1.343984842300415, "step": 45578 }, { "epoch": 3.593141505715412, "grad_norm": 0.5192419290542603, "learning_rate": 0.0005, "loss": 1.268846035003662, "step": 45579 }, { "epoch": 3.593220338983051, "grad_norm": 0.5200296640396118, "learning_rate": 0.0005, "loss": 1.2625439167022705, "step": 45580 }, { "epoch": 3.5932991722506897, "grad_norm": 0.48527416586875916, "learning_rate": 0.0005, "loss": 1.2371692657470703, "step": 45581 }, { "epoch": 3.593378005518329, "grad_norm": 0.5194915533065796, "learning_rate": 0.0005, "loss": 1.2066354751586914, "step": 45582 }, { "epoch": 3.5934568387859676, "grad_norm": 0.5301704406738281, "learning_rate": 0.0005, "loss": 1.263770341873169, "step": 45583 }, { "epoch": 3.5935356720536067, "grad_norm": 0.5325271487236023, "learning_rate": 0.0005, "loss": 1.2936646938323975, "step": 45584 }, { "epoch": 3.593614505321246, "grad_norm": 0.5030555129051208, "learning_rate": 0.0005, "loss": 1.2010213136672974, "step": 45585 }, { "epoch": 3.5936933385888845, "grad_norm": 0.3981916308403015, "learning_rate": 0.0005, "loss": 1.095172643661499, "step": 45586 }, { "epoch": 3.593772171856523, "grad_norm": 0.5040687918663025, "learning_rate": 0.0005, "loss": 1.2949352264404297, "step": 45587 }, { "epoch": 3.5938510051241623, "grad_norm": 0.44161131978034973, "learning_rate": 0.0005, "loss": 1.278261661529541, "step": 45588 }, { "epoch": 3.5939298383918015, "grad_norm": 0.5568061470985413, "learning_rate": 0.0005, "loss": 1.391880989074707, "step": 45589 }, { "epoch": 3.59400867165944, "grad_norm": 0.41530710458755493, "learning_rate": 0.0005, "loss": 1.2146704196929932, "step": 45590 }, { "epoch": 3.5940875049270793, "grad_norm": 0.5433704853057861, "learning_rate": 0.0005, "loss": 1.376133680343628, "step": 45591 }, { "epoch": 3.594166338194718, "grad_norm": 0.5394404530525208, "learning_rate": 0.0005, "loss": 1.346557378768921, "step": 45592 }, { "epoch": 3.594245171462357, "grad_norm": 0.441829651594162, "learning_rate": 0.0005, "loss": 1.255059003829956, "step": 45593 }, { "epoch": 3.5943240047299962, "grad_norm": 0.42052268981933594, "learning_rate": 0.0005, "loss": 1.1382914781570435, "step": 45594 }, { "epoch": 3.594402837997635, "grad_norm": 0.47095558047294617, "learning_rate": 0.0005, "loss": 1.1533026695251465, "step": 45595 }, { "epoch": 3.594481671265274, "grad_norm": 0.4170759916305542, "learning_rate": 0.0005, "loss": 1.199343204498291, "step": 45596 }, { "epoch": 3.5945605045329128, "grad_norm": 0.49270036816596985, "learning_rate": 0.0005, "loss": 1.3448095321655273, "step": 45597 }, { "epoch": 3.594639337800552, "grad_norm": 0.3963833451271057, "learning_rate": 0.0005, "loss": 1.0403246879577637, "step": 45598 }, { "epoch": 3.594718171068191, "grad_norm": 0.521946907043457, "learning_rate": 0.0005, "loss": 1.4476040601730347, "step": 45599 }, { "epoch": 3.5947970043358297, "grad_norm": 0.5388450026512146, "learning_rate": 0.0005, "loss": 1.3197176456451416, "step": 45600 }, { "epoch": 3.5948758376034684, "grad_norm": 0.5456622838973999, "learning_rate": 0.0005, "loss": 1.2801620960235596, "step": 45601 }, { "epoch": 3.5949546708711075, "grad_norm": 0.47481077909469604, "learning_rate": 0.0005, "loss": 1.399564266204834, "step": 45602 }, { "epoch": 3.5950335041387467, "grad_norm": 0.5160127282142639, "learning_rate": 0.0005, "loss": 1.4436640739440918, "step": 45603 }, { "epoch": 3.5951123374063854, "grad_norm": 0.42027226090431213, "learning_rate": 0.0005, "loss": 1.3495219945907593, "step": 45604 }, { "epoch": 3.5951911706740245, "grad_norm": 0.39767032861709595, "learning_rate": 0.0005, "loss": 1.0835201740264893, "step": 45605 }, { "epoch": 3.595270003941663, "grad_norm": 0.4074316918849945, "learning_rate": 0.0005, "loss": 1.1829297542572021, "step": 45606 }, { "epoch": 3.5953488372093023, "grad_norm": 0.47830840945243835, "learning_rate": 0.0005, "loss": 1.3329966068267822, "step": 45607 }, { "epoch": 3.5954276704769415, "grad_norm": 0.487676739692688, "learning_rate": 0.0005, "loss": 1.404198408126831, "step": 45608 }, { "epoch": 3.59550650374458, "grad_norm": 0.446491003036499, "learning_rate": 0.0005, "loss": 1.2745177745819092, "step": 45609 }, { "epoch": 3.5955853370122193, "grad_norm": 0.4355182349681854, "learning_rate": 0.0005, "loss": 1.3726290464401245, "step": 45610 }, { "epoch": 3.595664170279858, "grad_norm": 0.5319410562515259, "learning_rate": 0.0005, "loss": 1.2711875438690186, "step": 45611 }, { "epoch": 3.595743003547497, "grad_norm": 0.5086252689361572, "learning_rate": 0.0005, "loss": 1.2085962295532227, "step": 45612 }, { "epoch": 3.5958218368151362, "grad_norm": 0.5730056762695312, "learning_rate": 0.0005, "loss": 1.3053313493728638, "step": 45613 }, { "epoch": 3.595900670082775, "grad_norm": 0.5680270195007324, "learning_rate": 0.0005, "loss": 1.2612563371658325, "step": 45614 }, { "epoch": 3.5959795033504136, "grad_norm": 0.7723172307014465, "learning_rate": 0.0005, "loss": 1.2636723518371582, "step": 45615 }, { "epoch": 3.5960583366180527, "grad_norm": 0.5044211745262146, "learning_rate": 0.0005, "loss": 1.267310380935669, "step": 45616 }, { "epoch": 3.596137169885692, "grad_norm": 0.5025750398635864, "learning_rate": 0.0005, "loss": 1.1528408527374268, "step": 45617 }, { "epoch": 3.5962160031533306, "grad_norm": 0.4610293209552765, "learning_rate": 0.0005, "loss": 1.3484206199645996, "step": 45618 }, { "epoch": 3.5962948364209697, "grad_norm": 0.40523046255111694, "learning_rate": 0.0005, "loss": 1.1940828561782837, "step": 45619 }, { "epoch": 3.5963736696886084, "grad_norm": 0.4567095637321472, "learning_rate": 0.0005, "loss": 1.230882167816162, "step": 45620 }, { "epoch": 3.5964525029562475, "grad_norm": 0.4853253960609436, "learning_rate": 0.0005, "loss": 1.3248218297958374, "step": 45621 }, { "epoch": 3.5965313362238867, "grad_norm": 0.5556084513664246, "learning_rate": 0.0005, "loss": 1.2511752843856812, "step": 45622 }, { "epoch": 3.5966101694915253, "grad_norm": 0.5590431690216064, "learning_rate": 0.0005, "loss": 1.4227790832519531, "step": 45623 }, { "epoch": 3.5966890027591645, "grad_norm": 0.47729581594467163, "learning_rate": 0.0005, "loss": 1.2059372663497925, "step": 45624 }, { "epoch": 3.596767836026803, "grad_norm": 0.518942654132843, "learning_rate": 0.0005, "loss": 1.38265860080719, "step": 45625 }, { "epoch": 3.5968466692944423, "grad_norm": 0.47073182463645935, "learning_rate": 0.0005, "loss": 1.3476922512054443, "step": 45626 }, { "epoch": 3.5969255025620814, "grad_norm": 0.4409204423427582, "learning_rate": 0.0005, "loss": 1.0664290189743042, "step": 45627 }, { "epoch": 3.59700433582972, "grad_norm": 0.7008553743362427, "learning_rate": 0.0005, "loss": 1.174698829650879, "step": 45628 }, { "epoch": 3.597083169097359, "grad_norm": 0.510417103767395, "learning_rate": 0.0005, "loss": 1.327322006225586, "step": 45629 }, { "epoch": 3.597162002364998, "grad_norm": 0.6429217457771301, "learning_rate": 0.0005, "loss": 1.381960153579712, "step": 45630 }, { "epoch": 3.597240835632637, "grad_norm": 0.5290387272834778, "learning_rate": 0.0005, "loss": 1.2215120792388916, "step": 45631 }, { "epoch": 3.5973196689002758, "grad_norm": 0.5176438689231873, "learning_rate": 0.0005, "loss": 1.2377092838287354, "step": 45632 }, { "epoch": 3.597398502167915, "grad_norm": 0.4232587218284607, "learning_rate": 0.0005, "loss": 1.1074047088623047, "step": 45633 }, { "epoch": 3.5974773354355536, "grad_norm": 1.1131573915481567, "learning_rate": 0.0005, "loss": 1.2796119451522827, "step": 45634 }, { "epoch": 3.5975561687031927, "grad_norm": 0.5773780345916748, "learning_rate": 0.0005, "loss": 1.4088152647018433, "step": 45635 }, { "epoch": 3.597635001970832, "grad_norm": 0.40061649680137634, "learning_rate": 0.0005, "loss": 1.181950330734253, "step": 45636 }, { "epoch": 3.5977138352384705, "grad_norm": 0.532589316368103, "learning_rate": 0.0005, "loss": 1.391171932220459, "step": 45637 }, { "epoch": 3.5977926685061097, "grad_norm": 0.60683274269104, "learning_rate": 0.0005, "loss": 1.4294782876968384, "step": 45638 }, { "epoch": 3.5978715017737484, "grad_norm": 0.43857377767562866, "learning_rate": 0.0005, "loss": 1.2932629585266113, "step": 45639 }, { "epoch": 3.5979503350413875, "grad_norm": 0.5270175337791443, "learning_rate": 0.0005, "loss": 1.1772868633270264, "step": 45640 }, { "epoch": 3.5980291683090266, "grad_norm": 0.4281996488571167, "learning_rate": 0.0005, "loss": 1.3474924564361572, "step": 45641 }, { "epoch": 3.5981080015766653, "grad_norm": 0.6881378293037415, "learning_rate": 0.0005, "loss": 1.464634895324707, "step": 45642 }, { "epoch": 3.5981868348443045, "grad_norm": 0.6218445897102356, "learning_rate": 0.0005, "loss": 1.2355172634124756, "step": 45643 }, { "epoch": 3.598265668111943, "grad_norm": 0.5615912675857544, "learning_rate": 0.0005, "loss": 1.262290596961975, "step": 45644 }, { "epoch": 3.5983445013795823, "grad_norm": 0.6036860942840576, "learning_rate": 0.0005, "loss": 1.2960436344146729, "step": 45645 }, { "epoch": 3.598423334647221, "grad_norm": 0.4994826316833496, "learning_rate": 0.0005, "loss": 1.2414642572402954, "step": 45646 }, { "epoch": 3.59850216791486, "grad_norm": 0.45184558629989624, "learning_rate": 0.0005, "loss": 1.2451297044754028, "step": 45647 }, { "epoch": 3.598581001182499, "grad_norm": 0.50020432472229, "learning_rate": 0.0005, "loss": 1.17277193069458, "step": 45648 }, { "epoch": 3.598659834450138, "grad_norm": 0.5469088554382324, "learning_rate": 0.0005, "loss": 1.4680590629577637, "step": 45649 }, { "epoch": 3.598738667717777, "grad_norm": 0.5371556878089905, "learning_rate": 0.0005, "loss": 1.2892967462539673, "step": 45650 }, { "epoch": 3.5988175009854158, "grad_norm": 0.40436697006225586, "learning_rate": 0.0005, "loss": 1.118757963180542, "step": 45651 }, { "epoch": 3.598896334253055, "grad_norm": 0.5184750556945801, "learning_rate": 0.0005, "loss": 1.2256231307983398, "step": 45652 }, { "epoch": 3.5989751675206936, "grad_norm": 0.5357573628425598, "learning_rate": 0.0005, "loss": 1.4262113571166992, "step": 45653 }, { "epoch": 3.5990540007883327, "grad_norm": 0.5491665601730347, "learning_rate": 0.0005, "loss": 1.328577995300293, "step": 45654 }, { "epoch": 3.599132834055972, "grad_norm": 0.4180278778076172, "learning_rate": 0.0005, "loss": 1.1964737176895142, "step": 45655 }, { "epoch": 3.5992116673236105, "grad_norm": 0.6391097903251648, "learning_rate": 0.0005, "loss": 1.306380033493042, "step": 45656 }, { "epoch": 3.5992905005912497, "grad_norm": 0.4214271008968353, "learning_rate": 0.0005, "loss": 1.2048810720443726, "step": 45657 }, { "epoch": 3.5993693338588884, "grad_norm": 0.4815274775028229, "learning_rate": 0.0005, "loss": 1.1211414337158203, "step": 45658 }, { "epoch": 3.5994481671265275, "grad_norm": 0.5796405673027039, "learning_rate": 0.0005, "loss": 1.3687613010406494, "step": 45659 }, { "epoch": 3.599527000394166, "grad_norm": 0.465366929769516, "learning_rate": 0.0005, "loss": 1.2490296363830566, "step": 45660 }, { "epoch": 3.5996058336618053, "grad_norm": 0.5207885503768921, "learning_rate": 0.0005, "loss": 1.3395719528198242, "step": 45661 }, { "epoch": 3.599684666929444, "grad_norm": 0.45786458253860474, "learning_rate": 0.0005, "loss": 1.2763193845748901, "step": 45662 }, { "epoch": 3.599763500197083, "grad_norm": 0.45477986335754395, "learning_rate": 0.0005, "loss": 1.3010492324829102, "step": 45663 }, { "epoch": 3.5998423334647223, "grad_norm": 0.4364320635795593, "learning_rate": 0.0005, "loss": 1.1663531064987183, "step": 45664 }, { "epoch": 3.599921166732361, "grad_norm": 0.4600735306739807, "learning_rate": 0.0005, "loss": 1.1427148580551147, "step": 45665 }, { "epoch": 3.6, "grad_norm": 0.45923134684562683, "learning_rate": 0.0005, "loss": 1.3320903778076172, "step": 45666 }, { "epoch": 3.600078833267639, "grad_norm": 0.547665536403656, "learning_rate": 0.0005, "loss": 1.3771228790283203, "step": 45667 }, { "epoch": 3.600157666535278, "grad_norm": 0.4128832221031189, "learning_rate": 0.0005, "loss": 1.281674861907959, "step": 45668 }, { "epoch": 3.600236499802917, "grad_norm": 0.48509785532951355, "learning_rate": 0.0005, "loss": 1.2864634990692139, "step": 45669 }, { "epoch": 3.6003153330705557, "grad_norm": 0.4314589202404022, "learning_rate": 0.0005, "loss": 1.1688063144683838, "step": 45670 }, { "epoch": 3.600394166338195, "grad_norm": 0.42876648902893066, "learning_rate": 0.0005, "loss": 1.2877557277679443, "step": 45671 }, { "epoch": 3.6004729996058336, "grad_norm": 0.4815070331096649, "learning_rate": 0.0005, "loss": 1.1557979583740234, "step": 45672 }, { "epoch": 3.6005518328734727, "grad_norm": 0.44556209444999695, "learning_rate": 0.0005, "loss": 1.236672043800354, "step": 45673 }, { "epoch": 3.600630666141112, "grad_norm": 0.47721242904663086, "learning_rate": 0.0005, "loss": 1.1783201694488525, "step": 45674 }, { "epoch": 3.6007094994087505, "grad_norm": 0.49101120233535767, "learning_rate": 0.0005, "loss": 1.2537832260131836, "step": 45675 }, { "epoch": 3.600788332676389, "grad_norm": 0.46399959921836853, "learning_rate": 0.0005, "loss": 1.3508532047271729, "step": 45676 }, { "epoch": 3.6008671659440283, "grad_norm": 0.5560411214828491, "learning_rate": 0.0005, "loss": 1.1474934816360474, "step": 45677 }, { "epoch": 3.6009459992116675, "grad_norm": 0.5172504186630249, "learning_rate": 0.0005, "loss": 1.163179636001587, "step": 45678 }, { "epoch": 3.601024832479306, "grad_norm": 0.4978439211845398, "learning_rate": 0.0005, "loss": 1.3759396076202393, "step": 45679 }, { "epoch": 3.6011036657469453, "grad_norm": 0.4241500794887543, "learning_rate": 0.0005, "loss": 1.1760097742080688, "step": 45680 }, { "epoch": 3.601182499014584, "grad_norm": 0.5174676775932312, "learning_rate": 0.0005, "loss": 1.1773532629013062, "step": 45681 }, { "epoch": 3.601261332282223, "grad_norm": 0.3979523181915283, "learning_rate": 0.0005, "loss": 1.1355438232421875, "step": 45682 }, { "epoch": 3.6013401655498622, "grad_norm": 0.4867917001247406, "learning_rate": 0.0005, "loss": 1.2654975652694702, "step": 45683 }, { "epoch": 3.601418998817501, "grad_norm": 0.47241681814193726, "learning_rate": 0.0005, "loss": 1.1463309526443481, "step": 45684 }, { "epoch": 3.60149783208514, "grad_norm": 0.5009178519248962, "learning_rate": 0.0005, "loss": 1.3501607179641724, "step": 45685 }, { "epoch": 3.6015766653527788, "grad_norm": 0.5059217810630798, "learning_rate": 0.0005, "loss": 1.2223131656646729, "step": 45686 }, { "epoch": 3.601655498620418, "grad_norm": 0.49265256524086, "learning_rate": 0.0005, "loss": 1.3602657318115234, "step": 45687 }, { "epoch": 3.601734331888057, "grad_norm": 0.5393773317337036, "learning_rate": 0.0005, "loss": 1.220975399017334, "step": 45688 }, { "epoch": 3.6018131651556957, "grad_norm": 0.4473907947540283, "learning_rate": 0.0005, "loss": 1.2212228775024414, "step": 45689 }, { "epoch": 3.6018919984233344, "grad_norm": 0.38667309284210205, "learning_rate": 0.0005, "loss": 1.1257303953170776, "step": 45690 }, { "epoch": 3.6019708316909735, "grad_norm": 0.48421359062194824, "learning_rate": 0.0005, "loss": 1.3138322830200195, "step": 45691 }, { "epoch": 3.6020496649586127, "grad_norm": 0.4094870686531067, "learning_rate": 0.0005, "loss": 1.1763057708740234, "step": 45692 }, { "epoch": 3.6021284982262514, "grad_norm": 0.5816775560379028, "learning_rate": 0.0005, "loss": 1.2548295259475708, "step": 45693 }, { "epoch": 3.6022073314938905, "grad_norm": 0.4611140489578247, "learning_rate": 0.0005, "loss": 1.311720848083496, "step": 45694 }, { "epoch": 3.602286164761529, "grad_norm": 1.2771607637405396, "learning_rate": 0.0005, "loss": 1.1505405902862549, "step": 45695 }, { "epoch": 3.6023649980291683, "grad_norm": 0.5423324108123779, "learning_rate": 0.0005, "loss": 1.35590398311615, "step": 45696 }, { "epoch": 3.6024438312968075, "grad_norm": 0.47950640320777893, "learning_rate": 0.0005, "loss": 1.1545498371124268, "step": 45697 }, { "epoch": 3.602522664564446, "grad_norm": 0.48288702964782715, "learning_rate": 0.0005, "loss": 1.3125452995300293, "step": 45698 }, { "epoch": 3.6026014978320853, "grad_norm": 0.4670730531215668, "learning_rate": 0.0005, "loss": 1.2397128343582153, "step": 45699 }, { "epoch": 3.602680331099724, "grad_norm": 0.4134332239627838, "learning_rate": 0.0005, "loss": 1.1081217527389526, "step": 45700 }, { "epoch": 3.602759164367363, "grad_norm": 0.4199460446834564, "learning_rate": 0.0005, "loss": 1.235605239868164, "step": 45701 }, { "epoch": 3.6028379976350022, "grad_norm": 0.5279366970062256, "learning_rate": 0.0005, "loss": 1.535345196723938, "step": 45702 }, { "epoch": 3.602916830902641, "grad_norm": 0.5484155416488647, "learning_rate": 0.0005, "loss": 1.1649049520492554, "step": 45703 }, { "epoch": 3.6029956641702796, "grad_norm": 0.5976930260658264, "learning_rate": 0.0005, "loss": 1.3853532075881958, "step": 45704 }, { "epoch": 3.6030744974379187, "grad_norm": 0.4974695146083832, "learning_rate": 0.0005, "loss": 1.2632166147232056, "step": 45705 }, { "epoch": 3.603153330705558, "grad_norm": 0.7297894358634949, "learning_rate": 0.0005, "loss": 1.3303166627883911, "step": 45706 }, { "epoch": 3.6032321639731966, "grad_norm": 0.5809283256530762, "learning_rate": 0.0005, "loss": 1.3865578174591064, "step": 45707 }, { "epoch": 3.6033109972408357, "grad_norm": 0.602552592754364, "learning_rate": 0.0005, "loss": 1.3415563106536865, "step": 45708 }, { "epoch": 3.6033898305084744, "grad_norm": 0.5432518124580383, "learning_rate": 0.0005, "loss": 1.4612228870391846, "step": 45709 }, { "epoch": 3.6034686637761135, "grad_norm": 0.45381462574005127, "learning_rate": 0.0005, "loss": 1.183068037033081, "step": 45710 }, { "epoch": 3.6035474970437527, "grad_norm": 0.47468096017837524, "learning_rate": 0.0005, "loss": 1.2366688251495361, "step": 45711 }, { "epoch": 3.6036263303113913, "grad_norm": 0.42966926097869873, "learning_rate": 0.0005, "loss": 1.1690595149993896, "step": 45712 }, { "epoch": 3.6037051635790305, "grad_norm": 0.5000742077827454, "learning_rate": 0.0005, "loss": 1.3604506254196167, "step": 45713 }, { "epoch": 3.603783996846669, "grad_norm": 0.5764767527580261, "learning_rate": 0.0005, "loss": 1.331712007522583, "step": 45714 }, { "epoch": 3.6038628301143083, "grad_norm": 0.4765510559082031, "learning_rate": 0.0005, "loss": 1.081678032875061, "step": 45715 }, { "epoch": 3.6039416633819474, "grad_norm": 0.48670634627342224, "learning_rate": 0.0005, "loss": 1.3649908304214478, "step": 45716 }, { "epoch": 3.604020496649586, "grad_norm": 0.48957276344299316, "learning_rate": 0.0005, "loss": 1.185054063796997, "step": 45717 }, { "epoch": 3.604099329917225, "grad_norm": 0.45713815093040466, "learning_rate": 0.0005, "loss": 1.2360007762908936, "step": 45718 }, { "epoch": 3.604178163184864, "grad_norm": 0.5012409687042236, "learning_rate": 0.0005, "loss": 1.2487316131591797, "step": 45719 }, { "epoch": 3.604256996452503, "grad_norm": 0.4449409246444702, "learning_rate": 0.0005, "loss": 1.3988943099975586, "step": 45720 }, { "epoch": 3.6043358297201418, "grad_norm": 0.513364315032959, "learning_rate": 0.0005, "loss": 1.2201709747314453, "step": 45721 }, { "epoch": 3.604414662987781, "grad_norm": 0.4609218239784241, "learning_rate": 0.0005, "loss": 1.2226250171661377, "step": 45722 }, { "epoch": 3.6044934962554196, "grad_norm": 0.5654221773147583, "learning_rate": 0.0005, "loss": 1.4695453643798828, "step": 45723 }, { "epoch": 3.6045723295230587, "grad_norm": 0.4210575222969055, "learning_rate": 0.0005, "loss": 1.1454648971557617, "step": 45724 }, { "epoch": 3.604651162790698, "grad_norm": 0.45597970485687256, "learning_rate": 0.0005, "loss": 1.2267050743103027, "step": 45725 }, { "epoch": 3.6047299960583365, "grad_norm": 0.4231503903865814, "learning_rate": 0.0005, "loss": 1.1984055042266846, "step": 45726 }, { "epoch": 3.6048088293259757, "grad_norm": 0.5794913172721863, "learning_rate": 0.0005, "loss": 1.3742974996566772, "step": 45727 }, { "epoch": 3.6048876625936144, "grad_norm": 0.526130735874176, "learning_rate": 0.0005, "loss": 1.2895331382751465, "step": 45728 }, { "epoch": 3.6049664958612535, "grad_norm": 0.4658701717853546, "learning_rate": 0.0005, "loss": 1.25624680519104, "step": 45729 }, { "epoch": 3.6050453291288926, "grad_norm": 0.48233041167259216, "learning_rate": 0.0005, "loss": 1.2723060846328735, "step": 45730 }, { "epoch": 3.6051241623965313, "grad_norm": 0.41072532534599304, "learning_rate": 0.0005, "loss": 1.131089210510254, "step": 45731 }, { "epoch": 3.60520299566417, "grad_norm": 0.3867538571357727, "learning_rate": 0.0005, "loss": 1.2375308275222778, "step": 45732 }, { "epoch": 3.605281828931809, "grad_norm": 0.42501598596572876, "learning_rate": 0.0005, "loss": 1.1843841075897217, "step": 45733 }, { "epoch": 3.6053606621994483, "grad_norm": 0.5155185461044312, "learning_rate": 0.0005, "loss": 1.2150094509124756, "step": 45734 }, { "epoch": 3.605439495467087, "grad_norm": 0.5034691095352173, "learning_rate": 0.0005, "loss": 1.2406764030456543, "step": 45735 }, { "epoch": 3.605518328734726, "grad_norm": 0.5546602606773376, "learning_rate": 0.0005, "loss": 1.4038515090942383, "step": 45736 }, { "epoch": 3.605597162002365, "grad_norm": 0.571823000907898, "learning_rate": 0.0005, "loss": 1.4490396976470947, "step": 45737 }, { "epoch": 3.605675995270004, "grad_norm": 0.4819803237915039, "learning_rate": 0.0005, "loss": 1.1913295984268188, "step": 45738 }, { "epoch": 3.605754828537643, "grad_norm": 0.5658876299858093, "learning_rate": 0.0005, "loss": 1.4122059345245361, "step": 45739 }, { "epoch": 3.6058336618052818, "grad_norm": 0.5828507542610168, "learning_rate": 0.0005, "loss": 1.1826070547103882, "step": 45740 }, { "epoch": 3.605912495072921, "grad_norm": 0.49005943536758423, "learning_rate": 0.0005, "loss": 1.2639514207839966, "step": 45741 }, { "epoch": 3.6059913283405596, "grad_norm": 0.5443031191825867, "learning_rate": 0.0005, "loss": 1.3040450811386108, "step": 45742 }, { "epoch": 3.6060701616081987, "grad_norm": 0.4635031819343567, "learning_rate": 0.0005, "loss": 1.2542684078216553, "step": 45743 }, { "epoch": 3.606148994875838, "grad_norm": 0.5085243582725525, "learning_rate": 0.0005, "loss": 1.2384355068206787, "step": 45744 }, { "epoch": 3.6062278281434765, "grad_norm": 0.48629793524742126, "learning_rate": 0.0005, "loss": 1.179107904434204, "step": 45745 }, { "epoch": 3.606306661411115, "grad_norm": 0.4850103259086609, "learning_rate": 0.0005, "loss": 1.1882562637329102, "step": 45746 }, { "epoch": 3.6063854946787544, "grad_norm": 0.5045951008796692, "learning_rate": 0.0005, "loss": 1.238855004310608, "step": 45747 }, { "epoch": 3.6064643279463935, "grad_norm": 0.5150193572044373, "learning_rate": 0.0005, "loss": 1.1413037776947021, "step": 45748 }, { "epoch": 3.606543161214032, "grad_norm": 0.4568234384059906, "learning_rate": 0.0005, "loss": 1.1296557188034058, "step": 45749 }, { "epoch": 3.6066219944816713, "grad_norm": 0.6046420335769653, "learning_rate": 0.0005, "loss": 1.5548808574676514, "step": 45750 }, { "epoch": 3.60670082774931, "grad_norm": 0.5535560250282288, "learning_rate": 0.0005, "loss": 1.1562259197235107, "step": 45751 }, { "epoch": 3.606779661016949, "grad_norm": 0.574021577835083, "learning_rate": 0.0005, "loss": 1.1701949834823608, "step": 45752 }, { "epoch": 3.6068584942845883, "grad_norm": 0.425038605928421, "learning_rate": 0.0005, "loss": 1.2538493871688843, "step": 45753 }, { "epoch": 3.606937327552227, "grad_norm": 0.4945445656776428, "learning_rate": 0.0005, "loss": 1.240157127380371, "step": 45754 }, { "epoch": 3.607016160819866, "grad_norm": 0.6715453267097473, "learning_rate": 0.0005, "loss": 1.1802515983581543, "step": 45755 }, { "epoch": 3.607094994087505, "grad_norm": 0.5687282681465149, "learning_rate": 0.0005, "loss": 1.3521349430084229, "step": 45756 }, { "epoch": 3.607173827355144, "grad_norm": 0.5234755277633667, "learning_rate": 0.0005, "loss": 1.4129164218902588, "step": 45757 }, { "epoch": 3.607252660622783, "grad_norm": 0.5138328075408936, "learning_rate": 0.0005, "loss": 1.1953251361846924, "step": 45758 }, { "epoch": 3.6073314938904217, "grad_norm": 0.4075775146484375, "learning_rate": 0.0005, "loss": 1.2948596477508545, "step": 45759 }, { "epoch": 3.6074103271580604, "grad_norm": 0.5283459424972534, "learning_rate": 0.0005, "loss": 1.4051142930984497, "step": 45760 }, { "epoch": 3.6074891604256996, "grad_norm": 0.5215985178947449, "learning_rate": 0.0005, "loss": 1.2285289764404297, "step": 45761 }, { "epoch": 3.6075679936933387, "grad_norm": 0.4921421706676483, "learning_rate": 0.0005, "loss": 1.413631796836853, "step": 45762 }, { "epoch": 3.6076468269609774, "grad_norm": 0.5411540269851685, "learning_rate": 0.0005, "loss": 1.2473400831222534, "step": 45763 }, { "epoch": 3.6077256602286165, "grad_norm": 0.5182314515113831, "learning_rate": 0.0005, "loss": 1.4779846668243408, "step": 45764 }, { "epoch": 3.607804493496255, "grad_norm": 0.595694899559021, "learning_rate": 0.0005, "loss": 1.3846263885498047, "step": 45765 }, { "epoch": 3.6078833267638943, "grad_norm": 0.43278831243515015, "learning_rate": 0.0005, "loss": 1.3278627395629883, "step": 45766 }, { "epoch": 3.6079621600315335, "grad_norm": 0.41710445284843445, "learning_rate": 0.0005, "loss": 1.2043194770812988, "step": 45767 }, { "epoch": 3.608040993299172, "grad_norm": 0.4732602536678314, "learning_rate": 0.0005, "loss": 1.1672489643096924, "step": 45768 }, { "epoch": 3.6081198265668113, "grad_norm": 0.44247761368751526, "learning_rate": 0.0005, "loss": 1.2450275421142578, "step": 45769 }, { "epoch": 3.60819865983445, "grad_norm": 0.44521641731262207, "learning_rate": 0.0005, "loss": 1.193792462348938, "step": 45770 }, { "epoch": 3.608277493102089, "grad_norm": 0.5313852429389954, "learning_rate": 0.0005, "loss": 1.030549168586731, "step": 45771 }, { "epoch": 3.6083563263697283, "grad_norm": 0.6597958207130432, "learning_rate": 0.0005, "loss": 1.3592729568481445, "step": 45772 }, { "epoch": 3.608435159637367, "grad_norm": 0.5839212536811829, "learning_rate": 0.0005, "loss": 1.4194543361663818, "step": 45773 }, { "epoch": 3.608513992905006, "grad_norm": 0.646911084651947, "learning_rate": 0.0005, "loss": 1.3771430253982544, "step": 45774 }, { "epoch": 3.6085928261726448, "grad_norm": 0.5923305153846741, "learning_rate": 0.0005, "loss": 1.3385581970214844, "step": 45775 }, { "epoch": 3.608671659440284, "grad_norm": 0.5758959054946899, "learning_rate": 0.0005, "loss": 1.3181860446929932, "step": 45776 }, { "epoch": 3.6087504927079226, "grad_norm": 0.6766641139984131, "learning_rate": 0.0005, "loss": 1.3197851181030273, "step": 45777 }, { "epoch": 3.6088293259755617, "grad_norm": 0.4501276910305023, "learning_rate": 0.0005, "loss": 1.2223880290985107, "step": 45778 }, { "epoch": 3.6089081592432004, "grad_norm": 0.5695026516914368, "learning_rate": 0.0005, "loss": 1.1613578796386719, "step": 45779 }, { "epoch": 3.6089869925108395, "grad_norm": 0.43775415420532227, "learning_rate": 0.0005, "loss": 1.17158842086792, "step": 45780 }, { "epoch": 3.6090658257784787, "grad_norm": 0.5120850205421448, "learning_rate": 0.0005, "loss": 1.3201361894607544, "step": 45781 }, { "epoch": 3.6091446590461174, "grad_norm": 0.4273854196071625, "learning_rate": 0.0005, "loss": 1.1415995359420776, "step": 45782 }, { "epoch": 3.6092234923137565, "grad_norm": 0.4904845058917999, "learning_rate": 0.0005, "loss": 1.2702233791351318, "step": 45783 }, { "epoch": 3.609302325581395, "grad_norm": 0.4686800539493561, "learning_rate": 0.0005, "loss": 1.2650829553604126, "step": 45784 }, { "epoch": 3.6093811588490343, "grad_norm": 0.5084258317947388, "learning_rate": 0.0005, "loss": 1.252855896949768, "step": 45785 }, { "epoch": 3.6094599921166735, "grad_norm": 0.5140952467918396, "learning_rate": 0.0005, "loss": 1.418750286102295, "step": 45786 }, { "epoch": 3.609538825384312, "grad_norm": 0.4386339485645294, "learning_rate": 0.0005, "loss": 1.1564116477966309, "step": 45787 }, { "epoch": 3.6096176586519513, "grad_norm": 0.491591215133667, "learning_rate": 0.0005, "loss": 1.340254545211792, "step": 45788 }, { "epoch": 3.60969649191959, "grad_norm": 0.4383680820465088, "learning_rate": 0.0005, "loss": 1.2769408226013184, "step": 45789 }, { "epoch": 3.609775325187229, "grad_norm": 0.4765050709247589, "learning_rate": 0.0005, "loss": 1.2022687196731567, "step": 45790 }, { "epoch": 3.609854158454868, "grad_norm": 0.3751142919063568, "learning_rate": 0.0005, "loss": 1.0181362628936768, "step": 45791 }, { "epoch": 3.609932991722507, "grad_norm": 0.5310887098312378, "learning_rate": 0.0005, "loss": 1.1672821044921875, "step": 45792 }, { "epoch": 3.6100118249901456, "grad_norm": 0.4566463232040405, "learning_rate": 0.0005, "loss": 1.1305310726165771, "step": 45793 }, { "epoch": 3.6100906582577847, "grad_norm": 0.5081789493560791, "learning_rate": 0.0005, "loss": 1.1446833610534668, "step": 45794 }, { "epoch": 3.610169491525424, "grad_norm": 0.3812989294528961, "learning_rate": 0.0005, "loss": 1.1759470701217651, "step": 45795 }, { "epoch": 3.6102483247930626, "grad_norm": 0.5826546549797058, "learning_rate": 0.0005, "loss": 1.2311346530914307, "step": 45796 }, { "epoch": 3.6103271580607017, "grad_norm": 0.41540202498435974, "learning_rate": 0.0005, "loss": 1.2620327472686768, "step": 45797 }, { "epoch": 3.6104059913283404, "grad_norm": 0.4521394968032837, "learning_rate": 0.0005, "loss": 1.3048481941223145, "step": 45798 }, { "epoch": 3.6104848245959795, "grad_norm": 0.5943240523338318, "learning_rate": 0.0005, "loss": 1.1907129287719727, "step": 45799 }, { "epoch": 3.6105636578636187, "grad_norm": 0.47073090076446533, "learning_rate": 0.0005, "loss": 1.2763333320617676, "step": 45800 }, { "epoch": 3.6106424911312573, "grad_norm": 0.4789015054702759, "learning_rate": 0.0005, "loss": 1.2892465591430664, "step": 45801 }, { "epoch": 3.6107213243988965, "grad_norm": 0.5158909559249878, "learning_rate": 0.0005, "loss": 1.3095042705535889, "step": 45802 }, { "epoch": 3.610800157666535, "grad_norm": 0.6073617935180664, "learning_rate": 0.0005, "loss": 1.1890631914138794, "step": 45803 }, { "epoch": 3.6108789909341743, "grad_norm": 0.476239413022995, "learning_rate": 0.0005, "loss": 1.2986623048782349, "step": 45804 }, { "epoch": 3.6109578242018134, "grad_norm": 0.5093705654144287, "learning_rate": 0.0005, "loss": 1.1356492042541504, "step": 45805 }, { "epoch": 3.611036657469452, "grad_norm": 0.49145135283470154, "learning_rate": 0.0005, "loss": 1.3723232746124268, "step": 45806 }, { "epoch": 3.611115490737091, "grad_norm": 0.47865548729896545, "learning_rate": 0.0005, "loss": 1.2217950820922852, "step": 45807 }, { "epoch": 3.61119432400473, "grad_norm": 0.6306712031364441, "learning_rate": 0.0005, "loss": 1.3785796165466309, "step": 45808 }, { "epoch": 3.611273157272369, "grad_norm": 0.5585592985153198, "learning_rate": 0.0005, "loss": 1.2371747493743896, "step": 45809 }, { "epoch": 3.6113519905400078, "grad_norm": 0.5301547646522522, "learning_rate": 0.0005, "loss": 1.2857592105865479, "step": 45810 }, { "epoch": 3.611430823807647, "grad_norm": 0.787341833114624, "learning_rate": 0.0005, "loss": 1.510366439819336, "step": 45811 }, { "epoch": 3.6115096570752856, "grad_norm": 0.5024363398551941, "learning_rate": 0.0005, "loss": 1.1989377737045288, "step": 45812 }, { "epoch": 3.6115884903429247, "grad_norm": 0.4956853985786438, "learning_rate": 0.0005, "loss": 1.3224154710769653, "step": 45813 }, { "epoch": 3.611667323610564, "grad_norm": 0.48765328526496887, "learning_rate": 0.0005, "loss": 1.2203892469406128, "step": 45814 }, { "epoch": 3.6117461568782026, "grad_norm": 0.35931384563446045, "learning_rate": 0.0005, "loss": 0.9879898428916931, "step": 45815 }, { "epoch": 3.6118249901458417, "grad_norm": 0.548092782497406, "learning_rate": 0.0005, "loss": 1.3830102682113647, "step": 45816 }, { "epoch": 3.6119038234134804, "grad_norm": 0.5480029582977295, "learning_rate": 0.0005, "loss": 1.2031227350234985, "step": 45817 }, { "epoch": 3.6119826566811195, "grad_norm": 0.4984031915664673, "learning_rate": 0.0005, "loss": 1.2999707460403442, "step": 45818 }, { "epoch": 3.6120614899487586, "grad_norm": 0.5279730558395386, "learning_rate": 0.0005, "loss": 1.1843862533569336, "step": 45819 }, { "epoch": 3.6121403232163973, "grad_norm": 0.4749000072479248, "learning_rate": 0.0005, "loss": 1.1409538984298706, "step": 45820 }, { "epoch": 3.612219156484036, "grad_norm": 0.45554086565971375, "learning_rate": 0.0005, "loss": 1.3618905544281006, "step": 45821 }, { "epoch": 3.612297989751675, "grad_norm": 0.47009530663490295, "learning_rate": 0.0005, "loss": 1.3039602041244507, "step": 45822 }, { "epoch": 3.6123768230193143, "grad_norm": 0.5242295861244202, "learning_rate": 0.0005, "loss": 1.3271734714508057, "step": 45823 }, { "epoch": 3.612455656286953, "grad_norm": 0.4157746136188507, "learning_rate": 0.0005, "loss": 1.2636997699737549, "step": 45824 }, { "epoch": 3.612534489554592, "grad_norm": 0.4880211651325226, "learning_rate": 0.0005, "loss": 1.2161576747894287, "step": 45825 }, { "epoch": 3.612613322822231, "grad_norm": 0.6075988411903381, "learning_rate": 0.0005, "loss": 1.41250479221344, "step": 45826 }, { "epoch": 3.61269215608987, "grad_norm": 0.5858222246170044, "learning_rate": 0.0005, "loss": 1.2849236726760864, "step": 45827 }, { "epoch": 3.612770989357509, "grad_norm": 0.4616021513938904, "learning_rate": 0.0005, "loss": 1.2144508361816406, "step": 45828 }, { "epoch": 3.6128498226251478, "grad_norm": 0.5518797636032104, "learning_rate": 0.0005, "loss": 1.3337039947509766, "step": 45829 }, { "epoch": 3.612928655892787, "grad_norm": 0.491303414106369, "learning_rate": 0.0005, "loss": 1.3034816980361938, "step": 45830 }, { "epoch": 3.6130074891604256, "grad_norm": 0.4482751190662384, "learning_rate": 0.0005, "loss": 1.1981282234191895, "step": 45831 }, { "epoch": 3.6130863224280647, "grad_norm": 0.4998152554035187, "learning_rate": 0.0005, "loss": 1.3073747158050537, "step": 45832 }, { "epoch": 3.613165155695704, "grad_norm": 0.42586007714271545, "learning_rate": 0.0005, "loss": 1.0979396104812622, "step": 45833 }, { "epoch": 3.6132439889633425, "grad_norm": 0.4914563298225403, "learning_rate": 0.0005, "loss": 1.1267080307006836, "step": 45834 }, { "epoch": 3.6133228222309812, "grad_norm": 0.6964504718780518, "learning_rate": 0.0005, "loss": 1.416275978088379, "step": 45835 }, { "epoch": 3.6134016554986204, "grad_norm": 0.44424089789390564, "learning_rate": 0.0005, "loss": 1.1496305465698242, "step": 45836 }, { "epoch": 3.6134804887662595, "grad_norm": 0.5254098773002625, "learning_rate": 0.0005, "loss": 1.2844352722167969, "step": 45837 }, { "epoch": 3.613559322033898, "grad_norm": 0.48541656136512756, "learning_rate": 0.0005, "loss": 1.2418556213378906, "step": 45838 }, { "epoch": 3.6136381553015373, "grad_norm": 0.48678058385849, "learning_rate": 0.0005, "loss": 1.2353675365447998, "step": 45839 }, { "epoch": 3.613716988569176, "grad_norm": 0.49703845381736755, "learning_rate": 0.0005, "loss": 1.2240393161773682, "step": 45840 }, { "epoch": 3.613795821836815, "grad_norm": 0.40631580352783203, "learning_rate": 0.0005, "loss": 1.048504114151001, "step": 45841 }, { "epoch": 3.6138746551044543, "grad_norm": 0.4681609869003296, "learning_rate": 0.0005, "loss": 1.2789864540100098, "step": 45842 }, { "epoch": 3.613953488372093, "grad_norm": 0.3874270021915436, "learning_rate": 0.0005, "loss": 1.1152966022491455, "step": 45843 }, { "epoch": 3.614032321639732, "grad_norm": 0.47501006722450256, "learning_rate": 0.0005, "loss": 1.3071459531784058, "step": 45844 }, { "epoch": 3.614111154907371, "grad_norm": 0.4338400065898895, "learning_rate": 0.0005, "loss": 1.11641263961792, "step": 45845 }, { "epoch": 3.61418998817501, "grad_norm": 0.6814366579055786, "learning_rate": 0.0005, "loss": 1.4449241161346436, "step": 45846 }, { "epoch": 3.614268821442649, "grad_norm": 0.5075387358665466, "learning_rate": 0.0005, "loss": 1.2163457870483398, "step": 45847 }, { "epoch": 3.6143476547102877, "grad_norm": 0.48091965913772583, "learning_rate": 0.0005, "loss": 1.2040833234786987, "step": 45848 }, { "epoch": 3.6144264879779264, "grad_norm": 0.4543597400188446, "learning_rate": 0.0005, "loss": 1.2568140029907227, "step": 45849 }, { "epoch": 3.6145053212455656, "grad_norm": 0.5392129421234131, "learning_rate": 0.0005, "loss": 1.405626893043518, "step": 45850 }, { "epoch": 3.6145841545132047, "grad_norm": 0.5637416243553162, "learning_rate": 0.0005, "loss": 1.3383616209030151, "step": 45851 }, { "epoch": 3.6146629877808434, "grad_norm": 0.49066540598869324, "learning_rate": 0.0005, "loss": 1.4106613397598267, "step": 45852 }, { "epoch": 3.6147418210484825, "grad_norm": 0.5005117058753967, "learning_rate": 0.0005, "loss": 1.25919508934021, "step": 45853 }, { "epoch": 3.614820654316121, "grad_norm": 0.46599122881889343, "learning_rate": 0.0005, "loss": 1.096869707107544, "step": 45854 }, { "epoch": 3.6148994875837603, "grad_norm": 0.4388454258441925, "learning_rate": 0.0005, "loss": 1.1551096439361572, "step": 45855 }, { "epoch": 3.6149783208513995, "grad_norm": 0.39149588346481323, "learning_rate": 0.0005, "loss": 1.1291162967681885, "step": 45856 }, { "epoch": 3.615057154119038, "grad_norm": 0.478617399930954, "learning_rate": 0.0005, "loss": 1.2845765352249146, "step": 45857 }, { "epoch": 3.6151359873866773, "grad_norm": 0.6347068548202515, "learning_rate": 0.0005, "loss": 1.3873404264450073, "step": 45858 }, { "epoch": 3.615214820654316, "grad_norm": 0.604751706123352, "learning_rate": 0.0005, "loss": 1.312315821647644, "step": 45859 }, { "epoch": 3.615293653921955, "grad_norm": 0.5354259014129639, "learning_rate": 0.0005, "loss": 1.2905712127685547, "step": 45860 }, { "epoch": 3.6153724871895943, "grad_norm": 0.4789677858352661, "learning_rate": 0.0005, "loss": 1.3576678037643433, "step": 45861 }, { "epoch": 3.615451320457233, "grad_norm": 0.4039301574230194, "learning_rate": 0.0005, "loss": 1.0633972883224487, "step": 45862 }, { "epoch": 3.6155301537248716, "grad_norm": 0.554878294467926, "learning_rate": 0.0005, "loss": 1.2307755947113037, "step": 45863 }, { "epoch": 3.6156089869925108, "grad_norm": 0.6020869016647339, "learning_rate": 0.0005, "loss": 1.2641583681106567, "step": 45864 }, { "epoch": 3.61568782026015, "grad_norm": 0.489791601896286, "learning_rate": 0.0005, "loss": 1.3942328691482544, "step": 45865 }, { "epoch": 3.6157666535277886, "grad_norm": 0.4068772494792938, "learning_rate": 0.0005, "loss": 1.2416574954986572, "step": 45866 }, { "epoch": 3.6158454867954277, "grad_norm": 0.5648940801620483, "learning_rate": 0.0005, "loss": 1.2385120391845703, "step": 45867 }, { "epoch": 3.6159243200630664, "grad_norm": 0.46390604972839355, "learning_rate": 0.0005, "loss": 1.2336370944976807, "step": 45868 }, { "epoch": 3.6160031533307055, "grad_norm": 0.545745313167572, "learning_rate": 0.0005, "loss": 1.379106044769287, "step": 45869 }, { "epoch": 3.6160819865983447, "grad_norm": 0.5721821188926697, "learning_rate": 0.0005, "loss": 1.3982229232788086, "step": 45870 }, { "epoch": 3.6161608198659834, "grad_norm": 0.42259785532951355, "learning_rate": 0.0005, "loss": 1.2484383583068848, "step": 45871 }, { "epoch": 3.6162396531336225, "grad_norm": 0.49577608704566956, "learning_rate": 0.0005, "loss": 1.2434720993041992, "step": 45872 }, { "epoch": 3.616318486401261, "grad_norm": 0.5245388746261597, "learning_rate": 0.0005, "loss": 1.2382901906967163, "step": 45873 }, { "epoch": 3.6163973196689003, "grad_norm": 0.5146515965461731, "learning_rate": 0.0005, "loss": 1.4044315814971924, "step": 45874 }, { "epoch": 3.6164761529365395, "grad_norm": 0.44190508127212524, "learning_rate": 0.0005, "loss": 1.1044777631759644, "step": 45875 }, { "epoch": 3.616554986204178, "grad_norm": 0.519976019859314, "learning_rate": 0.0005, "loss": 1.254955530166626, "step": 45876 }, { "epoch": 3.616633819471817, "grad_norm": 0.4601161479949951, "learning_rate": 0.0005, "loss": 1.1775424480438232, "step": 45877 }, { "epoch": 3.616712652739456, "grad_norm": 0.5800990462303162, "learning_rate": 0.0005, "loss": 1.2119685411453247, "step": 45878 }, { "epoch": 3.616791486007095, "grad_norm": 0.4678443670272827, "learning_rate": 0.0005, "loss": 1.312562346458435, "step": 45879 }, { "epoch": 3.616870319274734, "grad_norm": 0.5636041164398193, "learning_rate": 0.0005, "loss": 1.3982760906219482, "step": 45880 }, { "epoch": 3.616949152542373, "grad_norm": 0.4563373327255249, "learning_rate": 0.0005, "loss": 1.1099424362182617, "step": 45881 }, { "epoch": 3.6170279858100116, "grad_norm": 0.5477206110954285, "learning_rate": 0.0005, "loss": 1.1960456371307373, "step": 45882 }, { "epoch": 3.6171068190776507, "grad_norm": 0.43246322870254517, "learning_rate": 0.0005, "loss": 1.227111577987671, "step": 45883 }, { "epoch": 3.61718565234529, "grad_norm": 0.5117672681808472, "learning_rate": 0.0005, "loss": 1.2439002990722656, "step": 45884 }, { "epoch": 3.6172644856129286, "grad_norm": 0.7244189381599426, "learning_rate": 0.0005, "loss": 1.525134563446045, "step": 45885 }, { "epoch": 3.6173433188805677, "grad_norm": 0.42528384923934937, "learning_rate": 0.0005, "loss": 1.213024616241455, "step": 45886 }, { "epoch": 3.6174221521482064, "grad_norm": 0.45613640546798706, "learning_rate": 0.0005, "loss": 1.2723582983016968, "step": 45887 }, { "epoch": 3.6175009854158455, "grad_norm": 0.4467901587486267, "learning_rate": 0.0005, "loss": 1.2758119106292725, "step": 45888 }, { "epoch": 3.6175798186834847, "grad_norm": 0.5153112411499023, "learning_rate": 0.0005, "loss": 1.2958273887634277, "step": 45889 }, { "epoch": 3.6176586519511233, "grad_norm": 0.4985107481479645, "learning_rate": 0.0005, "loss": 1.1969181299209595, "step": 45890 }, { "epoch": 3.6177374852187625, "grad_norm": 0.45701542496681213, "learning_rate": 0.0005, "loss": 1.1738073825836182, "step": 45891 }, { "epoch": 3.617816318486401, "grad_norm": 0.5467339754104614, "learning_rate": 0.0005, "loss": 1.4242353439331055, "step": 45892 }, { "epoch": 3.6178951517540403, "grad_norm": 0.5099837183952332, "learning_rate": 0.0005, "loss": 1.4479937553405762, "step": 45893 }, { "epoch": 3.617973985021679, "grad_norm": 0.4114834666252136, "learning_rate": 0.0005, "loss": 1.2285038232803345, "step": 45894 }, { "epoch": 3.618052818289318, "grad_norm": 0.39696004986763, "learning_rate": 0.0005, "loss": 1.1775914430618286, "step": 45895 }, { "epoch": 3.618131651556957, "grad_norm": 0.47671714425086975, "learning_rate": 0.0005, "loss": 1.239446997642517, "step": 45896 }, { "epoch": 3.618210484824596, "grad_norm": 0.41311198472976685, "learning_rate": 0.0005, "loss": 1.2708382606506348, "step": 45897 }, { "epoch": 3.618289318092235, "grad_norm": 0.5089972615242004, "learning_rate": 0.0005, "loss": 1.3044965267181396, "step": 45898 }, { "epoch": 3.6183681513598738, "grad_norm": 0.4440886378288269, "learning_rate": 0.0005, "loss": 1.2206964492797852, "step": 45899 }, { "epoch": 3.618446984627513, "grad_norm": 0.4541757106781006, "learning_rate": 0.0005, "loss": 1.1556711196899414, "step": 45900 }, { "epoch": 3.6185258178951516, "grad_norm": 0.4729645550251007, "learning_rate": 0.0005, "loss": 1.394117832183838, "step": 45901 }, { "epoch": 3.6186046511627907, "grad_norm": 0.48178988695144653, "learning_rate": 0.0005, "loss": 1.381003975868225, "step": 45902 }, { "epoch": 3.61868348443043, "grad_norm": 0.4244893193244934, "learning_rate": 0.0005, "loss": 1.22590970993042, "step": 45903 }, { "epoch": 3.6187623176980686, "grad_norm": 0.5024133324623108, "learning_rate": 0.0005, "loss": 1.2544186115264893, "step": 45904 }, { "epoch": 3.6188411509657077, "grad_norm": 0.47785791754722595, "learning_rate": 0.0005, "loss": 1.379063367843628, "step": 45905 }, { "epoch": 3.6189199842333464, "grad_norm": 0.5155704021453857, "learning_rate": 0.0005, "loss": 1.3361015319824219, "step": 45906 }, { "epoch": 3.6189988175009855, "grad_norm": 0.46846091747283936, "learning_rate": 0.0005, "loss": 1.2601397037506104, "step": 45907 }, { "epoch": 3.619077650768624, "grad_norm": 0.4843665361404419, "learning_rate": 0.0005, "loss": 1.1928503513336182, "step": 45908 }, { "epoch": 3.6191564840362633, "grad_norm": 0.4694751501083374, "learning_rate": 0.0005, "loss": 1.2677793502807617, "step": 45909 }, { "epoch": 3.619235317303902, "grad_norm": 0.5418984293937683, "learning_rate": 0.0005, "loss": 1.311011552810669, "step": 45910 }, { "epoch": 3.619314150571541, "grad_norm": 0.5077829957008362, "learning_rate": 0.0005, "loss": 1.3786418437957764, "step": 45911 }, { "epoch": 3.6193929838391803, "grad_norm": 0.9533165693283081, "learning_rate": 0.0005, "loss": 1.3541638851165771, "step": 45912 }, { "epoch": 3.619471817106819, "grad_norm": 0.47081607580184937, "learning_rate": 0.0005, "loss": 1.217172622680664, "step": 45913 }, { "epoch": 3.619550650374458, "grad_norm": 0.6429731249809265, "learning_rate": 0.0005, "loss": 1.1940369606018066, "step": 45914 }, { "epoch": 3.619629483642097, "grad_norm": 0.48368600010871887, "learning_rate": 0.0005, "loss": 1.068752646446228, "step": 45915 }, { "epoch": 3.619708316909736, "grad_norm": 0.5208361744880676, "learning_rate": 0.0005, "loss": 1.332547903060913, "step": 45916 }, { "epoch": 3.619787150177375, "grad_norm": 0.40711790323257446, "learning_rate": 0.0005, "loss": 1.116931438446045, "step": 45917 }, { "epoch": 3.6198659834450138, "grad_norm": 0.5196937322616577, "learning_rate": 0.0005, "loss": 1.2129946947097778, "step": 45918 }, { "epoch": 3.619944816712653, "grad_norm": 0.5678977370262146, "learning_rate": 0.0005, "loss": 1.3885612487792969, "step": 45919 }, { "epoch": 3.6200236499802916, "grad_norm": 0.5864879488945007, "learning_rate": 0.0005, "loss": 1.4037837982177734, "step": 45920 }, { "epoch": 3.6201024832479307, "grad_norm": 0.3904871344566345, "learning_rate": 0.0005, "loss": 1.0365939140319824, "step": 45921 }, { "epoch": 3.62018131651557, "grad_norm": 0.524574339389801, "learning_rate": 0.0005, "loss": 1.4150135517120361, "step": 45922 }, { "epoch": 3.6202601497832085, "grad_norm": 0.41572126746177673, "learning_rate": 0.0005, "loss": 1.0967044830322266, "step": 45923 }, { "epoch": 3.6203389830508472, "grad_norm": 0.5851784944534302, "learning_rate": 0.0005, "loss": 1.4496433734893799, "step": 45924 }, { "epoch": 3.6204178163184864, "grad_norm": 0.45850032567977905, "learning_rate": 0.0005, "loss": 1.2071454524993896, "step": 45925 }, { "epoch": 3.6204966495861255, "grad_norm": 0.4583226442337036, "learning_rate": 0.0005, "loss": 1.2865946292877197, "step": 45926 }, { "epoch": 3.620575482853764, "grad_norm": 0.551427960395813, "learning_rate": 0.0005, "loss": 1.342533826828003, "step": 45927 }, { "epoch": 3.6206543161214033, "grad_norm": 0.4560401737689972, "learning_rate": 0.0005, "loss": 1.2223570346832275, "step": 45928 }, { "epoch": 3.620733149389042, "grad_norm": 0.5580688118934631, "learning_rate": 0.0005, "loss": 1.297060489654541, "step": 45929 }, { "epoch": 3.620811982656681, "grad_norm": 0.43504342436790466, "learning_rate": 0.0005, "loss": 1.0636671781539917, "step": 45930 }, { "epoch": 3.6208908159243203, "grad_norm": 0.46932607889175415, "learning_rate": 0.0005, "loss": 1.1696683168411255, "step": 45931 }, { "epoch": 3.620969649191959, "grad_norm": 0.44976624846458435, "learning_rate": 0.0005, "loss": 1.1297786235809326, "step": 45932 }, { "epoch": 3.621048482459598, "grad_norm": 0.4801890254020691, "learning_rate": 0.0005, "loss": 1.2183295488357544, "step": 45933 }, { "epoch": 3.621127315727237, "grad_norm": 0.591163694858551, "learning_rate": 0.0005, "loss": 1.2511506080627441, "step": 45934 }, { "epoch": 3.621206148994876, "grad_norm": 0.4739866256713867, "learning_rate": 0.0005, "loss": 1.1310315132141113, "step": 45935 }, { "epoch": 3.621284982262515, "grad_norm": 0.4162527918815613, "learning_rate": 0.0005, "loss": 1.1498640775680542, "step": 45936 }, { "epoch": 3.6213638155301537, "grad_norm": 0.4812582731246948, "learning_rate": 0.0005, "loss": 1.24480402469635, "step": 45937 }, { "epoch": 3.6214426487977924, "grad_norm": 0.4489455819129944, "learning_rate": 0.0005, "loss": 1.339597225189209, "step": 45938 }, { "epoch": 3.6215214820654316, "grad_norm": 0.4495399296283722, "learning_rate": 0.0005, "loss": 1.3079383373260498, "step": 45939 }, { "epoch": 3.6216003153330707, "grad_norm": 0.3755642771720886, "learning_rate": 0.0005, "loss": 1.0134168863296509, "step": 45940 }, { "epoch": 3.6216791486007094, "grad_norm": 0.5765043497085571, "learning_rate": 0.0005, "loss": 1.2198662757873535, "step": 45941 }, { "epoch": 3.6217579818683485, "grad_norm": 0.5556697249412537, "learning_rate": 0.0005, "loss": 1.3968379497528076, "step": 45942 }, { "epoch": 3.621836815135987, "grad_norm": 0.5948308110237122, "learning_rate": 0.0005, "loss": 1.1910775899887085, "step": 45943 }, { "epoch": 3.6219156484036263, "grad_norm": 0.553699254989624, "learning_rate": 0.0005, "loss": 1.333829641342163, "step": 45944 }, { "epoch": 3.6219944816712655, "grad_norm": 0.5310903787612915, "learning_rate": 0.0005, "loss": 1.1199052333831787, "step": 45945 }, { "epoch": 3.622073314938904, "grad_norm": 0.5979693531990051, "learning_rate": 0.0005, "loss": 1.3059008121490479, "step": 45946 }, { "epoch": 3.6221521482065433, "grad_norm": 0.4809453785419464, "learning_rate": 0.0005, "loss": 1.3096699714660645, "step": 45947 }, { "epoch": 3.622230981474182, "grad_norm": 0.49660342931747437, "learning_rate": 0.0005, "loss": 1.2838927507400513, "step": 45948 }, { "epoch": 3.622309814741821, "grad_norm": 0.45373642444610596, "learning_rate": 0.0005, "loss": 1.1733888387680054, "step": 45949 }, { "epoch": 3.6223886480094603, "grad_norm": 0.4386715888977051, "learning_rate": 0.0005, "loss": 1.21156644821167, "step": 45950 }, { "epoch": 3.622467481277099, "grad_norm": 0.4915747046470642, "learning_rate": 0.0005, "loss": 1.1911628246307373, "step": 45951 }, { "epoch": 3.6225463145447376, "grad_norm": 0.5366578102111816, "learning_rate": 0.0005, "loss": 1.3827040195465088, "step": 45952 }, { "epoch": 3.6226251478123768, "grad_norm": 0.4988233745098114, "learning_rate": 0.0005, "loss": 1.362855076789856, "step": 45953 }, { "epoch": 3.622703981080016, "grad_norm": 0.42456239461898804, "learning_rate": 0.0005, "loss": 1.0728353261947632, "step": 45954 }, { "epoch": 3.6227828143476546, "grad_norm": 0.6985148787498474, "learning_rate": 0.0005, "loss": 1.3438490629196167, "step": 45955 }, { "epoch": 3.6228616476152937, "grad_norm": 0.47192543745040894, "learning_rate": 0.0005, "loss": 1.3318836688995361, "step": 45956 }, { "epoch": 3.6229404808829324, "grad_norm": 0.4731529951095581, "learning_rate": 0.0005, "loss": 1.189098834991455, "step": 45957 }, { "epoch": 3.6230193141505715, "grad_norm": 0.4480541944503784, "learning_rate": 0.0005, "loss": 1.3342915773391724, "step": 45958 }, { "epoch": 3.6230981474182107, "grad_norm": 0.452337384223938, "learning_rate": 0.0005, "loss": 1.2179516553878784, "step": 45959 }, { "epoch": 3.6231769806858494, "grad_norm": 0.4963824152946472, "learning_rate": 0.0005, "loss": 1.4370484352111816, "step": 45960 }, { "epoch": 3.6232558139534885, "grad_norm": 0.4890309274196625, "learning_rate": 0.0005, "loss": 1.19303560256958, "step": 45961 }, { "epoch": 3.623334647221127, "grad_norm": 0.4329412281513214, "learning_rate": 0.0005, "loss": 1.2033066749572754, "step": 45962 }, { "epoch": 3.6234134804887663, "grad_norm": 0.45432525873184204, "learning_rate": 0.0005, "loss": 1.2594698667526245, "step": 45963 }, { "epoch": 3.6234923137564055, "grad_norm": 0.45837458968162537, "learning_rate": 0.0005, "loss": 1.4235397577285767, "step": 45964 }, { "epoch": 3.623571147024044, "grad_norm": 0.4970240592956543, "learning_rate": 0.0005, "loss": 1.2771384716033936, "step": 45965 }, { "epoch": 3.623649980291683, "grad_norm": 0.49450191855430603, "learning_rate": 0.0005, "loss": 1.2385588884353638, "step": 45966 }, { "epoch": 3.623728813559322, "grad_norm": 0.47907376289367676, "learning_rate": 0.0005, "loss": 1.3424859046936035, "step": 45967 }, { "epoch": 3.623807646826961, "grad_norm": 0.5226465463638306, "learning_rate": 0.0005, "loss": 1.311179518699646, "step": 45968 }, { "epoch": 3.6238864800946, "grad_norm": 0.42223426699638367, "learning_rate": 0.0005, "loss": 1.2217175960540771, "step": 45969 }, { "epoch": 3.623965313362239, "grad_norm": 0.4931749403476715, "learning_rate": 0.0005, "loss": 1.4311211109161377, "step": 45970 }, { "epoch": 3.6240441466298776, "grad_norm": 0.458543062210083, "learning_rate": 0.0005, "loss": 1.2647831439971924, "step": 45971 }, { "epoch": 3.6241229798975167, "grad_norm": 0.653062105178833, "learning_rate": 0.0005, "loss": 1.283339262008667, "step": 45972 }, { "epoch": 3.624201813165156, "grad_norm": 0.5093584060668945, "learning_rate": 0.0005, "loss": 1.3334274291992188, "step": 45973 }, { "epoch": 3.6242806464327946, "grad_norm": 0.49196508526802063, "learning_rate": 0.0005, "loss": 1.2338948249816895, "step": 45974 }, { "epoch": 3.6243594797004337, "grad_norm": 0.5063489079475403, "learning_rate": 0.0005, "loss": 1.2769670486450195, "step": 45975 }, { "epoch": 3.6244383129680724, "grad_norm": 0.6543463468551636, "learning_rate": 0.0005, "loss": 1.1992392539978027, "step": 45976 }, { "epoch": 3.6245171462357115, "grad_norm": 0.468456894159317, "learning_rate": 0.0005, "loss": 1.1468446254730225, "step": 45977 }, { "epoch": 3.6245959795033507, "grad_norm": 0.5094833970069885, "learning_rate": 0.0005, "loss": 1.2684016227722168, "step": 45978 }, { "epoch": 3.6246748127709894, "grad_norm": 0.537068247795105, "learning_rate": 0.0005, "loss": 1.2861158847808838, "step": 45979 }, { "epoch": 3.624753646038628, "grad_norm": 0.38496795296669006, "learning_rate": 0.0005, "loss": 1.134294033050537, "step": 45980 }, { "epoch": 3.624832479306267, "grad_norm": 0.9501681327819824, "learning_rate": 0.0005, "loss": 1.1520472764968872, "step": 45981 }, { "epoch": 3.6249113125739063, "grad_norm": 0.42362770438194275, "learning_rate": 0.0005, "loss": 1.0444645881652832, "step": 45982 }, { "epoch": 3.624990145841545, "grad_norm": 0.6129242181777954, "learning_rate": 0.0005, "loss": 1.410497784614563, "step": 45983 }, { "epoch": 3.625068979109184, "grad_norm": 0.44954144954681396, "learning_rate": 0.0005, "loss": 1.363467812538147, "step": 45984 }, { "epoch": 3.625147812376823, "grad_norm": 0.5060535073280334, "learning_rate": 0.0005, "loss": 1.254594087600708, "step": 45985 }, { "epoch": 3.625226645644462, "grad_norm": 0.41246914863586426, "learning_rate": 0.0005, "loss": 1.0502629280090332, "step": 45986 }, { "epoch": 3.625305478912101, "grad_norm": 0.5679623484611511, "learning_rate": 0.0005, "loss": 1.3390185832977295, "step": 45987 }, { "epoch": 3.6253843121797398, "grad_norm": 0.6522738933563232, "learning_rate": 0.0005, "loss": 1.2790660858154297, "step": 45988 }, { "epoch": 3.625463145447379, "grad_norm": 0.4764975309371948, "learning_rate": 0.0005, "loss": 1.3008487224578857, "step": 45989 }, { "epoch": 3.6255419787150176, "grad_norm": 0.5263963937759399, "learning_rate": 0.0005, "loss": 1.4902840852737427, "step": 45990 }, { "epoch": 3.6256208119826567, "grad_norm": 0.5381773710250854, "learning_rate": 0.0005, "loss": 1.2497328519821167, "step": 45991 }, { "epoch": 3.625699645250296, "grad_norm": 0.5031587481498718, "learning_rate": 0.0005, "loss": 1.3345037698745728, "step": 45992 }, { "epoch": 3.6257784785179346, "grad_norm": 0.44262510538101196, "learning_rate": 0.0005, "loss": 1.2353243827819824, "step": 45993 }, { "epoch": 3.6258573117855732, "grad_norm": 0.40057116746902466, "learning_rate": 0.0005, "loss": 1.2238099575042725, "step": 45994 }, { "epoch": 3.6259361450532124, "grad_norm": 0.5154539942741394, "learning_rate": 0.0005, "loss": 1.223231554031372, "step": 45995 }, { "epoch": 3.6260149783208515, "grad_norm": 0.4661569893360138, "learning_rate": 0.0005, "loss": 1.1879827976226807, "step": 45996 }, { "epoch": 3.62609381158849, "grad_norm": 0.6689009070396423, "learning_rate": 0.0005, "loss": 1.2623902559280396, "step": 45997 }, { "epoch": 3.6261726448561293, "grad_norm": 0.555435836315155, "learning_rate": 0.0005, "loss": 1.1951340436935425, "step": 45998 }, { "epoch": 3.626251478123768, "grad_norm": 0.5972508788108826, "learning_rate": 0.0005, "loss": 1.1128864288330078, "step": 45999 }, { "epoch": 3.626330311391407, "grad_norm": 0.543017566204071, "learning_rate": 0.0005, "loss": 1.2160577774047852, "step": 46000 }, { "epoch": 3.6264091446590463, "grad_norm": 0.469245582818985, "learning_rate": 0.0005, "loss": 1.0349159240722656, "step": 46001 }, { "epoch": 3.626487977926685, "grad_norm": 0.561113715171814, "learning_rate": 0.0005, "loss": 1.376357913017273, "step": 46002 }, { "epoch": 3.626566811194324, "grad_norm": 0.5420067310333252, "learning_rate": 0.0005, "loss": 1.2052867412567139, "step": 46003 }, { "epoch": 3.626645644461963, "grad_norm": 0.821384608745575, "learning_rate": 0.0005, "loss": 1.1340793371200562, "step": 46004 }, { "epoch": 3.626724477729602, "grad_norm": 0.6593108773231506, "learning_rate": 0.0005, "loss": 1.2759742736816406, "step": 46005 }, { "epoch": 3.626803310997241, "grad_norm": 0.5732308030128479, "learning_rate": 0.0005, "loss": 1.1097356081008911, "step": 46006 }, { "epoch": 3.6268821442648798, "grad_norm": 0.5130834579467773, "learning_rate": 0.0005, "loss": 1.1323204040527344, "step": 46007 }, { "epoch": 3.6269609775325184, "grad_norm": 0.4380068778991699, "learning_rate": 0.0005, "loss": 1.1372730731964111, "step": 46008 }, { "epoch": 3.6270398108001576, "grad_norm": 0.5320486426353455, "learning_rate": 0.0005, "loss": 1.203556776046753, "step": 46009 }, { "epoch": 3.6271186440677967, "grad_norm": 0.624747633934021, "learning_rate": 0.0005, "loss": 1.276625394821167, "step": 46010 }, { "epoch": 3.6271974773354354, "grad_norm": 0.3938767910003662, "learning_rate": 0.0005, "loss": 1.0250108242034912, "step": 46011 }, { "epoch": 3.6272763106030745, "grad_norm": 0.3810824751853943, "learning_rate": 0.0005, "loss": 1.0456233024597168, "step": 46012 }, { "epoch": 3.6273551438707132, "grad_norm": 0.8525087237358093, "learning_rate": 0.0005, "loss": 1.3405778408050537, "step": 46013 }, { "epoch": 3.6274339771383524, "grad_norm": 0.572952926158905, "learning_rate": 0.0005, "loss": 1.2286232709884644, "step": 46014 }, { "epoch": 3.6275128104059915, "grad_norm": 0.69681316614151, "learning_rate": 0.0005, "loss": 1.1793608665466309, "step": 46015 }, { "epoch": 3.62759164367363, "grad_norm": 0.47317737340927124, "learning_rate": 0.0005, "loss": 1.1759849786758423, "step": 46016 }, { "epoch": 3.6276704769412693, "grad_norm": 0.4626159071922302, "learning_rate": 0.0005, "loss": 1.1878705024719238, "step": 46017 }, { "epoch": 3.627749310208908, "grad_norm": 0.8566371202468872, "learning_rate": 0.0005, "loss": 1.4796473979949951, "step": 46018 }, { "epoch": 3.627828143476547, "grad_norm": 0.5036817193031311, "learning_rate": 0.0005, "loss": 1.2556324005126953, "step": 46019 }, { "epoch": 3.6279069767441863, "grad_norm": 0.7027729153633118, "learning_rate": 0.0005, "loss": 1.2688686847686768, "step": 46020 }, { "epoch": 3.627985810011825, "grad_norm": 0.5193257331848145, "learning_rate": 0.0005, "loss": 1.1505107879638672, "step": 46021 }, { "epoch": 3.628064643279464, "grad_norm": 0.43280985951423645, "learning_rate": 0.0005, "loss": 1.0273563861846924, "step": 46022 }, { "epoch": 3.628143476547103, "grad_norm": 0.6972743272781372, "learning_rate": 0.0005, "loss": 1.258671760559082, "step": 46023 }, { "epoch": 3.628222309814742, "grad_norm": 0.5421479344367981, "learning_rate": 0.0005, "loss": 1.4510010480880737, "step": 46024 }, { "epoch": 3.6283011430823806, "grad_norm": 0.40135735273361206, "learning_rate": 0.0005, "loss": 1.1738053560256958, "step": 46025 }, { "epoch": 3.6283799763500197, "grad_norm": 0.6626793146133423, "learning_rate": 0.0005, "loss": 1.169975757598877, "step": 46026 }, { "epoch": 3.6284588096176584, "grad_norm": 0.5833564400672913, "learning_rate": 0.0005, "loss": 1.3149170875549316, "step": 46027 }, { "epoch": 3.6285376428852976, "grad_norm": 0.7315998673439026, "learning_rate": 0.0005, "loss": 1.3604464530944824, "step": 46028 }, { "epoch": 3.6286164761529367, "grad_norm": 0.41223111748695374, "learning_rate": 0.0005, "loss": 1.1229147911071777, "step": 46029 }, { "epoch": 3.6286953094205754, "grad_norm": 0.5058580040931702, "learning_rate": 0.0005, "loss": 1.1764113903045654, "step": 46030 }, { "epoch": 3.6287741426882145, "grad_norm": 0.7404573559761047, "learning_rate": 0.0005, "loss": 1.4565469026565552, "step": 46031 }, { "epoch": 3.628852975955853, "grad_norm": 0.674503743648529, "learning_rate": 0.0005, "loss": 1.2166690826416016, "step": 46032 }, { "epoch": 3.6289318092234923, "grad_norm": 0.5151107311248779, "learning_rate": 0.0005, "loss": 1.3108594417572021, "step": 46033 }, { "epoch": 3.6290106424911315, "grad_norm": 0.552315890789032, "learning_rate": 0.0005, "loss": 1.1125801801681519, "step": 46034 }, { "epoch": 3.62908947575877, "grad_norm": 0.4721320569515228, "learning_rate": 0.0005, "loss": 1.102765440940857, "step": 46035 }, { "epoch": 3.6291683090264093, "grad_norm": 0.4556034505367279, "learning_rate": 0.0005, "loss": 1.2875335216522217, "step": 46036 }, { "epoch": 3.629247142294048, "grad_norm": 0.3451801836490631, "learning_rate": 0.0005, "loss": 1.1383670568466187, "step": 46037 }, { "epoch": 3.629325975561687, "grad_norm": 0.5326777696609497, "learning_rate": 0.0005, "loss": 1.082789659500122, "step": 46038 }, { "epoch": 3.629404808829326, "grad_norm": 0.5084449052810669, "learning_rate": 0.0005, "loss": 1.0296286344528198, "step": 46039 }, { "epoch": 3.629483642096965, "grad_norm": 0.5156054496765137, "learning_rate": 0.0005, "loss": 1.073333978652954, "step": 46040 }, { "epoch": 3.6295624753646036, "grad_norm": 0.5252724885940552, "learning_rate": 0.0005, "loss": 1.2034342288970947, "step": 46041 }, { "epoch": 3.6296413086322428, "grad_norm": 0.6363316774368286, "learning_rate": 0.0005, "loss": 1.3346021175384521, "step": 46042 }, { "epoch": 3.629720141899882, "grad_norm": 0.5075132846832275, "learning_rate": 0.0005, "loss": 1.1632080078125, "step": 46043 }, { "epoch": 3.6297989751675206, "grad_norm": 0.5017642378807068, "learning_rate": 0.0005, "loss": 1.0102097988128662, "step": 46044 }, { "epoch": 3.6298778084351597, "grad_norm": 0.5926282405853271, "learning_rate": 0.0005, "loss": 1.2936632633209229, "step": 46045 }, { "epoch": 3.6299566417027984, "grad_norm": 0.4882802963256836, "learning_rate": 0.0005, "loss": 1.142225742340088, "step": 46046 }, { "epoch": 3.6300354749704375, "grad_norm": 0.626794159412384, "learning_rate": 0.0005, "loss": 1.3119335174560547, "step": 46047 }, { "epoch": 3.6301143082380767, "grad_norm": 0.39856189489364624, "learning_rate": 0.0005, "loss": 1.0426548719406128, "step": 46048 }, { "epoch": 3.6301931415057154, "grad_norm": 0.4311126172542572, "learning_rate": 0.0005, "loss": 1.1191179752349854, "step": 46049 }, { "epoch": 3.6302719747733545, "grad_norm": 0.4427367150783539, "learning_rate": 0.0005, "loss": 1.0866526365280151, "step": 46050 }, { "epoch": 3.630350808040993, "grad_norm": 0.358725368976593, "learning_rate": 0.0005, "loss": 1.0757291316986084, "step": 46051 }, { "epoch": 3.6304296413086323, "grad_norm": 0.4720592498779297, "learning_rate": 0.0005, "loss": 1.2213479280471802, "step": 46052 }, { "epoch": 3.6305084745762715, "grad_norm": 0.5369981527328491, "learning_rate": 0.0005, "loss": 1.1690424680709839, "step": 46053 }, { "epoch": 3.63058730784391, "grad_norm": 0.4115723967552185, "learning_rate": 0.0005, "loss": 1.074109435081482, "step": 46054 }, { "epoch": 3.630666141111549, "grad_norm": 0.6095196008682251, "learning_rate": 0.0005, "loss": 1.265152931213379, "step": 46055 }, { "epoch": 3.630744974379188, "grad_norm": 0.5141322016716003, "learning_rate": 0.0005, "loss": 1.1214122772216797, "step": 46056 }, { "epoch": 3.630823807646827, "grad_norm": 0.4402689039707184, "learning_rate": 0.0005, "loss": 1.2749741077423096, "step": 46057 }, { "epoch": 3.630902640914466, "grad_norm": 0.43826454877853394, "learning_rate": 0.0005, "loss": 1.1725306510925293, "step": 46058 }, { "epoch": 3.630981474182105, "grad_norm": 0.3988533020019531, "learning_rate": 0.0005, "loss": 1.0379371643066406, "step": 46059 }, { "epoch": 3.6310603074497436, "grad_norm": 0.43525004386901855, "learning_rate": 0.0005, "loss": 1.0297309160232544, "step": 46060 }, { "epoch": 3.6311391407173828, "grad_norm": 0.5832394361495972, "learning_rate": 0.0005, "loss": 1.2171375751495361, "step": 46061 }, { "epoch": 3.631217973985022, "grad_norm": 0.5781229138374329, "learning_rate": 0.0005, "loss": 1.1569607257843018, "step": 46062 }, { "epoch": 3.6312968072526606, "grad_norm": 0.4420260787010193, "learning_rate": 0.0005, "loss": 1.267845630645752, "step": 46063 }, { "epoch": 3.6313756405202997, "grad_norm": 0.4378724992275238, "learning_rate": 0.0005, "loss": 0.9497401118278503, "step": 46064 }, { "epoch": 3.6314544737879384, "grad_norm": 0.5223031640052795, "learning_rate": 0.0005, "loss": 1.1787192821502686, "step": 46065 }, { "epoch": 3.6315333070555775, "grad_norm": 0.4040414094924927, "learning_rate": 0.0005, "loss": 1.0009894371032715, "step": 46066 }, { "epoch": 3.6316121403232167, "grad_norm": 0.4817122220993042, "learning_rate": 0.0005, "loss": 1.264380693435669, "step": 46067 }, { "epoch": 3.6316909735908554, "grad_norm": 0.38942718505859375, "learning_rate": 0.0005, "loss": 1.034982442855835, "step": 46068 }, { "epoch": 3.631769806858494, "grad_norm": 0.5484592914581299, "learning_rate": 0.0005, "loss": 1.260555624961853, "step": 46069 }, { "epoch": 3.631848640126133, "grad_norm": 0.5918149948120117, "learning_rate": 0.0005, "loss": 1.385305643081665, "step": 46070 }, { "epoch": 3.6319274733937723, "grad_norm": 0.5594347715377808, "learning_rate": 0.0005, "loss": 1.2318731546401978, "step": 46071 }, { "epoch": 3.632006306661411, "grad_norm": 0.5181162357330322, "learning_rate": 0.0005, "loss": 1.2538073062896729, "step": 46072 }, { "epoch": 3.63208513992905, "grad_norm": 0.44447121024131775, "learning_rate": 0.0005, "loss": 1.0783522129058838, "step": 46073 }, { "epoch": 3.632163973196689, "grad_norm": 0.5561791658401489, "learning_rate": 0.0005, "loss": 1.1484711170196533, "step": 46074 }, { "epoch": 3.632242806464328, "grad_norm": 0.4834081530570984, "learning_rate": 0.0005, "loss": 1.3974087238311768, "step": 46075 }, { "epoch": 3.632321639731967, "grad_norm": 0.39906078577041626, "learning_rate": 0.0005, "loss": 1.0805463790893555, "step": 46076 }, { "epoch": 3.632400472999606, "grad_norm": 0.6096461415290833, "learning_rate": 0.0005, "loss": 1.2943058013916016, "step": 46077 }, { "epoch": 3.632479306267245, "grad_norm": 0.5167800784111023, "learning_rate": 0.0005, "loss": 1.0718920230865479, "step": 46078 }, { "epoch": 3.6325581395348836, "grad_norm": 0.44450950622558594, "learning_rate": 0.0005, "loss": 1.145853042602539, "step": 46079 }, { "epoch": 3.6326369728025227, "grad_norm": 0.42189037799835205, "learning_rate": 0.0005, "loss": 0.9868924617767334, "step": 46080 }, { "epoch": 3.632715806070162, "grad_norm": 1.3537030220031738, "learning_rate": 0.0005, "loss": 1.2063441276550293, "step": 46081 }, { "epoch": 3.6327946393378006, "grad_norm": 0.5255966186523438, "learning_rate": 0.0005, "loss": 1.2124176025390625, "step": 46082 }, { "epoch": 3.6328734726054392, "grad_norm": 0.44299399852752686, "learning_rate": 0.0005, "loss": 1.1540857553482056, "step": 46083 }, { "epoch": 3.6329523058730784, "grad_norm": 0.4591517746448517, "learning_rate": 0.0005, "loss": 1.2326467037200928, "step": 46084 }, { "epoch": 3.6330311391407175, "grad_norm": 0.4746222496032715, "learning_rate": 0.0005, "loss": 1.1071866750717163, "step": 46085 }, { "epoch": 3.633109972408356, "grad_norm": 0.4337271749973297, "learning_rate": 0.0005, "loss": 0.9667901992797852, "step": 46086 }, { "epoch": 3.6331888056759953, "grad_norm": 0.4310187101364136, "learning_rate": 0.0005, "loss": 0.9625717401504517, "step": 46087 }, { "epoch": 3.633267638943634, "grad_norm": 0.9444212913513184, "learning_rate": 0.0005, "loss": 1.0857988595962524, "step": 46088 }, { "epoch": 3.633346472211273, "grad_norm": 0.4113227128982544, "learning_rate": 0.0005, "loss": 1.0754528045654297, "step": 46089 }, { "epoch": 3.6334253054789123, "grad_norm": 0.48772573471069336, "learning_rate": 0.0005, "loss": 1.1489336490631104, "step": 46090 }, { "epoch": 3.633504138746551, "grad_norm": 0.45662567019462585, "learning_rate": 0.0005, "loss": 1.1120734214782715, "step": 46091 }, { "epoch": 3.63358297201419, "grad_norm": 0.5850002765655518, "learning_rate": 0.0005, "loss": 1.3163483142852783, "step": 46092 }, { "epoch": 3.633661805281829, "grad_norm": 0.5785184502601624, "learning_rate": 0.0005, "loss": 1.2436153888702393, "step": 46093 }, { "epoch": 3.633740638549468, "grad_norm": 0.6993347406387329, "learning_rate": 0.0005, "loss": 1.2101558446884155, "step": 46094 }, { "epoch": 3.633819471817107, "grad_norm": 0.46105635166168213, "learning_rate": 0.0005, "loss": 1.3313937187194824, "step": 46095 }, { "epoch": 3.6338983050847458, "grad_norm": 0.44314324855804443, "learning_rate": 0.0005, "loss": 1.3089712858200073, "step": 46096 }, { "epoch": 3.6339771383523845, "grad_norm": 0.49072012305259705, "learning_rate": 0.0005, "loss": 1.1261825561523438, "step": 46097 }, { "epoch": 3.6340559716200236, "grad_norm": 0.38899683952331543, "learning_rate": 0.0005, "loss": 1.0010446310043335, "step": 46098 }, { "epoch": 3.6341348048876627, "grad_norm": 0.6824069023132324, "learning_rate": 0.0005, "loss": 1.3268263339996338, "step": 46099 }, { "epoch": 3.6342136381553014, "grad_norm": 0.49178776144981384, "learning_rate": 0.0005, "loss": 1.2662301063537598, "step": 46100 }, { "epoch": 3.6342924714229405, "grad_norm": 0.4202243983745575, "learning_rate": 0.0005, "loss": 1.0025361776351929, "step": 46101 }, { "epoch": 3.6343713046905792, "grad_norm": 0.47265198826789856, "learning_rate": 0.0005, "loss": 1.2027264833450317, "step": 46102 }, { "epoch": 3.6344501379582184, "grad_norm": 0.4854518473148346, "learning_rate": 0.0005, "loss": 1.1318073272705078, "step": 46103 }, { "epoch": 3.6345289712258575, "grad_norm": 0.5262560844421387, "learning_rate": 0.0005, "loss": 1.266237735748291, "step": 46104 }, { "epoch": 3.634607804493496, "grad_norm": 0.45340779423713684, "learning_rate": 0.0005, "loss": 0.9968645572662354, "step": 46105 }, { "epoch": 3.6346866377611353, "grad_norm": 0.41093453764915466, "learning_rate": 0.0005, "loss": 0.9304891228675842, "step": 46106 }, { "epoch": 3.634765471028774, "grad_norm": 0.45155301690101624, "learning_rate": 0.0005, "loss": 1.2186307907104492, "step": 46107 }, { "epoch": 3.634844304296413, "grad_norm": 0.5613195300102234, "learning_rate": 0.0005, "loss": 1.2497153282165527, "step": 46108 }, { "epoch": 3.6349231375640523, "grad_norm": 0.3768059313297272, "learning_rate": 0.0005, "loss": 1.0471928119659424, "step": 46109 }, { "epoch": 3.635001970831691, "grad_norm": 0.5141207575798035, "learning_rate": 0.0005, "loss": 1.2045800685882568, "step": 46110 }, { "epoch": 3.6350808040993297, "grad_norm": 0.3899630904197693, "learning_rate": 0.0005, "loss": 1.0577287673950195, "step": 46111 }, { "epoch": 3.635159637366969, "grad_norm": 0.5280046463012695, "learning_rate": 0.0005, "loss": 1.1258920431137085, "step": 46112 }, { "epoch": 3.635238470634608, "grad_norm": 0.4785390794277191, "learning_rate": 0.0005, "loss": 1.0546424388885498, "step": 46113 }, { "epoch": 3.6353173039022466, "grad_norm": 0.5008459687232971, "learning_rate": 0.0005, "loss": 1.0335257053375244, "step": 46114 }, { "epoch": 3.6353961371698857, "grad_norm": 0.6973490715026855, "learning_rate": 0.0005, "loss": 1.115180492401123, "step": 46115 }, { "epoch": 3.6354749704375244, "grad_norm": 0.4119936525821686, "learning_rate": 0.0005, "loss": 1.1646761894226074, "step": 46116 }, { "epoch": 3.6355538037051636, "grad_norm": 0.4609147012233734, "learning_rate": 0.0005, "loss": 0.9710535407066345, "step": 46117 }, { "epoch": 3.6356326369728027, "grad_norm": 0.437212198972702, "learning_rate": 0.0005, "loss": 1.3425865173339844, "step": 46118 }, { "epoch": 3.6357114702404414, "grad_norm": 0.5291569828987122, "learning_rate": 0.0005, "loss": 1.1074378490447998, "step": 46119 }, { "epoch": 3.6357903035080805, "grad_norm": 0.4573465883731842, "learning_rate": 0.0005, "loss": 1.0083308219909668, "step": 46120 }, { "epoch": 3.635869136775719, "grad_norm": 0.43865469098091125, "learning_rate": 0.0005, "loss": 1.0273269414901733, "step": 46121 }, { "epoch": 3.6359479700433583, "grad_norm": 0.5515872240066528, "learning_rate": 0.0005, "loss": 1.2291613817214966, "step": 46122 }, { "epoch": 3.6360268033109975, "grad_norm": 0.4912876784801483, "learning_rate": 0.0005, "loss": 1.1723287105560303, "step": 46123 }, { "epoch": 3.636105636578636, "grad_norm": 0.45854878425598145, "learning_rate": 0.0005, "loss": 1.0533267259597778, "step": 46124 }, { "epoch": 3.636184469846275, "grad_norm": 0.5077431797981262, "learning_rate": 0.0005, "loss": 1.3609538078308105, "step": 46125 }, { "epoch": 3.636263303113914, "grad_norm": 0.5462519526481628, "learning_rate": 0.0005, "loss": 1.2165471315383911, "step": 46126 }, { "epoch": 3.636342136381553, "grad_norm": 0.5039799213409424, "learning_rate": 0.0005, "loss": 0.9227609634399414, "step": 46127 }, { "epoch": 3.636420969649192, "grad_norm": 0.5082367658615112, "learning_rate": 0.0005, "loss": 1.1022213697433472, "step": 46128 }, { "epoch": 3.636499802916831, "grad_norm": 0.5422670245170593, "learning_rate": 0.0005, "loss": 1.1607662439346313, "step": 46129 }, { "epoch": 3.6365786361844696, "grad_norm": 0.4837617874145508, "learning_rate": 0.0005, "loss": 1.095237374305725, "step": 46130 }, { "epoch": 3.6366574694521088, "grad_norm": 0.4812976121902466, "learning_rate": 0.0005, "loss": 1.071933388710022, "step": 46131 }, { "epoch": 3.636736302719748, "grad_norm": 0.4396720826625824, "learning_rate": 0.0005, "loss": 1.0474843978881836, "step": 46132 }, { "epoch": 3.6368151359873866, "grad_norm": 0.4878833591938019, "learning_rate": 0.0005, "loss": 0.9933179616928101, "step": 46133 }, { "epoch": 3.6368939692550257, "grad_norm": 0.49790236353874207, "learning_rate": 0.0005, "loss": 1.3335062265396118, "step": 46134 }, { "epoch": 3.6369728025226644, "grad_norm": 0.45835453271865845, "learning_rate": 0.0005, "loss": 1.1524730920791626, "step": 46135 }, { "epoch": 3.6370516357903035, "grad_norm": 0.6375622749328613, "learning_rate": 0.0005, "loss": 1.093336582183838, "step": 46136 }, { "epoch": 3.6371304690579427, "grad_norm": 0.45995476841926575, "learning_rate": 0.0005, "loss": 1.2175250053405762, "step": 46137 }, { "epoch": 3.6372093023255814, "grad_norm": 0.5359408855438232, "learning_rate": 0.0005, "loss": 1.138637900352478, "step": 46138 }, { "epoch": 3.63728813559322, "grad_norm": 0.4647688865661621, "learning_rate": 0.0005, "loss": 1.2825579643249512, "step": 46139 }, { "epoch": 3.637366968860859, "grad_norm": 0.46260538697242737, "learning_rate": 0.0005, "loss": 1.1796218156814575, "step": 46140 }, { "epoch": 3.6374458021284983, "grad_norm": 0.47423022985458374, "learning_rate": 0.0005, "loss": 1.0862228870391846, "step": 46141 }, { "epoch": 3.637524635396137, "grad_norm": 0.5650966167449951, "learning_rate": 0.0005, "loss": 1.3445985317230225, "step": 46142 }, { "epoch": 3.637603468663776, "grad_norm": 0.43251094222068787, "learning_rate": 0.0005, "loss": 1.1844770908355713, "step": 46143 }, { "epoch": 3.637682301931415, "grad_norm": 0.4673607051372528, "learning_rate": 0.0005, "loss": 1.1356885433197021, "step": 46144 }, { "epoch": 3.637761135199054, "grad_norm": 0.5502567291259766, "learning_rate": 0.0005, "loss": 1.3861253261566162, "step": 46145 }, { "epoch": 3.637839968466693, "grad_norm": 0.5660567879676819, "learning_rate": 0.0005, "loss": 1.284827470779419, "step": 46146 }, { "epoch": 3.637918801734332, "grad_norm": 0.6571692228317261, "learning_rate": 0.0005, "loss": 1.2369986772537231, "step": 46147 }, { "epoch": 3.637997635001971, "grad_norm": 0.523508608341217, "learning_rate": 0.0005, "loss": 1.1470954418182373, "step": 46148 }, { "epoch": 3.6380764682696096, "grad_norm": 0.5215460062026978, "learning_rate": 0.0005, "loss": 1.1734675168991089, "step": 46149 }, { "epoch": 3.6381553015372488, "grad_norm": 0.5623976588249207, "learning_rate": 0.0005, "loss": 1.269365668296814, "step": 46150 }, { "epoch": 3.638234134804888, "grad_norm": 0.6259650588035583, "learning_rate": 0.0005, "loss": 1.2839267253875732, "step": 46151 }, { "epoch": 3.6383129680725266, "grad_norm": 0.6620891690254211, "learning_rate": 0.0005, "loss": 1.0496201515197754, "step": 46152 }, { "epoch": 3.6383918013401657, "grad_norm": 0.6518326997756958, "learning_rate": 0.0005, "loss": 1.2680799961090088, "step": 46153 }, { "epoch": 3.6384706346078044, "grad_norm": 0.628768265247345, "learning_rate": 0.0005, "loss": 1.4262142181396484, "step": 46154 }, { "epoch": 3.6385494678754435, "grad_norm": 0.5037964582443237, "learning_rate": 0.0005, "loss": 1.291679859161377, "step": 46155 }, { "epoch": 3.6386283011430822, "grad_norm": 0.43203312158584595, "learning_rate": 0.0005, "loss": 1.177330493927002, "step": 46156 }, { "epoch": 3.6387071344107214, "grad_norm": 0.48884063959121704, "learning_rate": 0.0005, "loss": 1.1213555335998535, "step": 46157 }, { "epoch": 3.63878596767836, "grad_norm": 0.5139110088348389, "learning_rate": 0.0005, "loss": 1.235222578048706, "step": 46158 }, { "epoch": 3.638864800945999, "grad_norm": 0.46939000487327576, "learning_rate": 0.0005, "loss": 1.2464631795883179, "step": 46159 }, { "epoch": 3.6389436342136383, "grad_norm": 0.4375802278518677, "learning_rate": 0.0005, "loss": 1.0821187496185303, "step": 46160 }, { "epoch": 3.639022467481277, "grad_norm": 0.5342119336128235, "learning_rate": 0.0005, "loss": 1.173205852508545, "step": 46161 }, { "epoch": 3.639101300748916, "grad_norm": 0.5632505416870117, "learning_rate": 0.0005, "loss": 1.0785527229309082, "step": 46162 }, { "epoch": 3.639180134016555, "grad_norm": 0.4506508409976959, "learning_rate": 0.0005, "loss": 1.2130036354064941, "step": 46163 }, { "epoch": 3.639258967284194, "grad_norm": 0.5608935356140137, "learning_rate": 0.0005, "loss": 1.2098323106765747, "step": 46164 }, { "epoch": 3.639337800551833, "grad_norm": 0.4791201055049896, "learning_rate": 0.0005, "loss": 1.0476694107055664, "step": 46165 }, { "epoch": 3.639416633819472, "grad_norm": 0.5514375567436218, "learning_rate": 0.0005, "loss": 1.3487539291381836, "step": 46166 }, { "epoch": 3.639495467087111, "grad_norm": 0.44415614008903503, "learning_rate": 0.0005, "loss": 1.1637115478515625, "step": 46167 }, { "epoch": 3.6395743003547496, "grad_norm": 0.4344453811645508, "learning_rate": 0.0005, "loss": 1.2796858549118042, "step": 46168 }, { "epoch": 3.6396531336223887, "grad_norm": 0.6356104016304016, "learning_rate": 0.0005, "loss": 1.2706208229064941, "step": 46169 }, { "epoch": 3.6397319668900274, "grad_norm": 0.43749693036079407, "learning_rate": 0.0005, "loss": 1.3224684000015259, "step": 46170 }, { "epoch": 3.6398108001576666, "grad_norm": 0.47135186195373535, "learning_rate": 0.0005, "loss": 1.186755657196045, "step": 46171 }, { "epoch": 3.6398896334253052, "grad_norm": 0.4397342801094055, "learning_rate": 0.0005, "loss": 1.1149604320526123, "step": 46172 }, { "epoch": 3.6399684666929444, "grad_norm": 0.4506268799304962, "learning_rate": 0.0005, "loss": 1.2782464027404785, "step": 46173 }, { "epoch": 3.6400472999605835, "grad_norm": 0.4538976550102234, "learning_rate": 0.0005, "loss": 1.195307731628418, "step": 46174 }, { "epoch": 3.640126133228222, "grad_norm": 0.4756261706352234, "learning_rate": 0.0005, "loss": 1.157386064529419, "step": 46175 }, { "epoch": 3.6402049664958613, "grad_norm": 0.48364388942718506, "learning_rate": 0.0005, "loss": 1.289825439453125, "step": 46176 }, { "epoch": 3.6402837997635, "grad_norm": 0.5644857287406921, "learning_rate": 0.0005, "loss": 1.1033228635787964, "step": 46177 }, { "epoch": 3.640362633031139, "grad_norm": 0.4437544345855713, "learning_rate": 0.0005, "loss": 1.3306164741516113, "step": 46178 }, { "epoch": 3.6404414662987783, "grad_norm": 0.654058575630188, "learning_rate": 0.0005, "loss": 1.2362231016159058, "step": 46179 }, { "epoch": 3.640520299566417, "grad_norm": 0.5886818766593933, "learning_rate": 0.0005, "loss": 1.2817304134368896, "step": 46180 }, { "epoch": 3.640599132834056, "grad_norm": 0.47936418652534485, "learning_rate": 0.0005, "loss": 0.955308735370636, "step": 46181 }, { "epoch": 3.640677966101695, "grad_norm": 0.5557145476341248, "learning_rate": 0.0005, "loss": 1.2086198329925537, "step": 46182 }, { "epoch": 3.640756799369334, "grad_norm": 0.4981915354728699, "learning_rate": 0.0005, "loss": 1.1411131620407104, "step": 46183 }, { "epoch": 3.640835632636973, "grad_norm": 0.5075625777244568, "learning_rate": 0.0005, "loss": 1.1767752170562744, "step": 46184 }, { "epoch": 3.6409144659046118, "grad_norm": 0.5289866328239441, "learning_rate": 0.0005, "loss": 1.1502892971038818, "step": 46185 }, { "epoch": 3.6409932991722505, "grad_norm": 0.5407469868659973, "learning_rate": 0.0005, "loss": 1.223585844039917, "step": 46186 }, { "epoch": 3.6410721324398896, "grad_norm": 0.4768836796283722, "learning_rate": 0.0005, "loss": 1.217649221420288, "step": 46187 }, { "epoch": 3.6411509657075287, "grad_norm": 0.5381337404251099, "learning_rate": 0.0005, "loss": 1.2409257888793945, "step": 46188 }, { "epoch": 3.6412297989751674, "grad_norm": 0.4593011438846588, "learning_rate": 0.0005, "loss": 1.116984248161316, "step": 46189 }, { "epoch": 3.6413086322428065, "grad_norm": 0.42071136832237244, "learning_rate": 0.0005, "loss": 0.9848960638046265, "step": 46190 }, { "epoch": 3.6413874655104452, "grad_norm": 0.4314252436161041, "learning_rate": 0.0005, "loss": 1.1571179628372192, "step": 46191 }, { "epoch": 3.6414662987780844, "grad_norm": 0.5223487019538879, "learning_rate": 0.0005, "loss": 0.9803374409675598, "step": 46192 }, { "epoch": 3.6415451320457235, "grad_norm": 0.4009952247142792, "learning_rate": 0.0005, "loss": 1.0146677494049072, "step": 46193 }, { "epoch": 3.641623965313362, "grad_norm": 0.4161079525947571, "learning_rate": 0.0005, "loss": 1.0947548151016235, "step": 46194 }, { "epoch": 3.6417027985810013, "grad_norm": 0.4615621566772461, "learning_rate": 0.0005, "loss": 1.0500969886779785, "step": 46195 }, { "epoch": 3.64178163184864, "grad_norm": 0.5000073313713074, "learning_rate": 0.0005, "loss": 1.220078468322754, "step": 46196 }, { "epoch": 3.641860465116279, "grad_norm": 0.520083487033844, "learning_rate": 0.0005, "loss": 1.2433212995529175, "step": 46197 }, { "epoch": 3.6419392983839183, "grad_norm": 0.500288724899292, "learning_rate": 0.0005, "loss": 1.2141671180725098, "step": 46198 }, { "epoch": 3.642018131651557, "grad_norm": 0.43527287244796753, "learning_rate": 0.0005, "loss": 1.1013882160186768, "step": 46199 }, { "epoch": 3.6420969649191957, "grad_norm": 0.4861188232898712, "learning_rate": 0.0005, "loss": 0.9844220280647278, "step": 46200 }, { "epoch": 3.642175798186835, "grad_norm": 0.44078174233436584, "learning_rate": 0.0005, "loss": 1.1655192375183105, "step": 46201 }, { "epoch": 3.642254631454474, "grad_norm": 0.4765525460243225, "learning_rate": 0.0005, "loss": 1.247307300567627, "step": 46202 }, { "epoch": 3.6423334647221126, "grad_norm": 0.45075368881225586, "learning_rate": 0.0005, "loss": 1.1090492010116577, "step": 46203 }, { "epoch": 3.6424122979897517, "grad_norm": 0.43306368589401245, "learning_rate": 0.0005, "loss": 1.2138512134552002, "step": 46204 }, { "epoch": 3.6424911312573904, "grad_norm": 0.49791663885116577, "learning_rate": 0.0005, "loss": 1.2475509643554688, "step": 46205 }, { "epoch": 3.6425699645250296, "grad_norm": 0.459373414516449, "learning_rate": 0.0005, "loss": 1.2341481447219849, "step": 46206 }, { "epoch": 3.6426487977926687, "grad_norm": 0.42293208837509155, "learning_rate": 0.0005, "loss": 1.3038088083267212, "step": 46207 }, { "epoch": 3.6427276310603074, "grad_norm": 0.46580255031585693, "learning_rate": 0.0005, "loss": 1.11541748046875, "step": 46208 }, { "epoch": 3.6428064643279465, "grad_norm": 0.5160558819770813, "learning_rate": 0.0005, "loss": 1.0915833711624146, "step": 46209 }, { "epoch": 3.642885297595585, "grad_norm": 0.5589351654052734, "learning_rate": 0.0005, "loss": 1.17405366897583, "step": 46210 }, { "epoch": 3.6429641308632243, "grad_norm": 0.5110133290290833, "learning_rate": 0.0005, "loss": 1.1026160717010498, "step": 46211 }, { "epoch": 3.6430429641308635, "grad_norm": 0.6309652924537659, "learning_rate": 0.0005, "loss": 1.1088402271270752, "step": 46212 }, { "epoch": 3.643121797398502, "grad_norm": 0.4440753757953644, "learning_rate": 0.0005, "loss": 1.1220300197601318, "step": 46213 }, { "epoch": 3.643200630666141, "grad_norm": 0.5002050399780273, "learning_rate": 0.0005, "loss": 1.2554818391799927, "step": 46214 }, { "epoch": 3.64327946393378, "grad_norm": 0.4877324104309082, "learning_rate": 0.0005, "loss": 1.2435463666915894, "step": 46215 }, { "epoch": 3.643358297201419, "grad_norm": 0.5122839212417603, "learning_rate": 0.0005, "loss": 1.265730381011963, "step": 46216 }, { "epoch": 3.643437130469058, "grad_norm": 0.49702128767967224, "learning_rate": 0.0005, "loss": 1.0662505626678467, "step": 46217 }, { "epoch": 3.643515963736697, "grad_norm": 0.6238313317298889, "learning_rate": 0.0005, "loss": 1.1043678522109985, "step": 46218 }, { "epoch": 3.6435947970043356, "grad_norm": 0.5524874925613403, "learning_rate": 0.0005, "loss": 1.5065257549285889, "step": 46219 }, { "epoch": 3.6436736302719748, "grad_norm": 0.5253044962882996, "learning_rate": 0.0005, "loss": 1.5043913125991821, "step": 46220 }, { "epoch": 3.643752463539614, "grad_norm": 0.40080559253692627, "learning_rate": 0.0005, "loss": 1.1628365516662598, "step": 46221 }, { "epoch": 3.6438312968072526, "grad_norm": 0.45320722460746765, "learning_rate": 0.0005, "loss": 1.1168696880340576, "step": 46222 }, { "epoch": 3.6439101300748917, "grad_norm": 0.6199347972869873, "learning_rate": 0.0005, "loss": 1.602338433265686, "step": 46223 }, { "epoch": 3.6439889633425304, "grad_norm": 0.451922208070755, "learning_rate": 0.0005, "loss": 0.9625259637832642, "step": 46224 }, { "epoch": 3.6440677966101696, "grad_norm": 0.44284170866012573, "learning_rate": 0.0005, "loss": 1.264978051185608, "step": 46225 }, { "epoch": 3.6441466298778087, "grad_norm": 0.7416330575942993, "learning_rate": 0.0005, "loss": 1.2864317893981934, "step": 46226 }, { "epoch": 3.6442254631454474, "grad_norm": 0.5134961605072021, "learning_rate": 0.0005, "loss": 1.1000299453735352, "step": 46227 }, { "epoch": 3.644304296413086, "grad_norm": 0.47202038764953613, "learning_rate": 0.0005, "loss": 1.1289241313934326, "step": 46228 }, { "epoch": 3.644383129680725, "grad_norm": 0.4427843391895294, "learning_rate": 0.0005, "loss": 1.0087451934814453, "step": 46229 }, { "epoch": 3.6444619629483643, "grad_norm": 10.07833194732666, "learning_rate": 0.0005, "loss": 1.1417988538742065, "step": 46230 }, { "epoch": 3.644540796216003, "grad_norm": 0.5085926055908203, "learning_rate": 0.0005, "loss": 1.4202816486358643, "step": 46231 }, { "epoch": 3.644619629483642, "grad_norm": 0.4675804674625397, "learning_rate": 0.0005, "loss": 1.1132292747497559, "step": 46232 }, { "epoch": 3.644698462751281, "grad_norm": 0.45357728004455566, "learning_rate": 0.0005, "loss": 1.0729835033416748, "step": 46233 }, { "epoch": 3.64477729601892, "grad_norm": 0.4809906482696533, "learning_rate": 0.0005, "loss": 1.0737841129302979, "step": 46234 }, { "epoch": 3.644856129286559, "grad_norm": 0.5023313760757446, "learning_rate": 0.0005, "loss": 1.1898378133773804, "step": 46235 }, { "epoch": 3.644934962554198, "grad_norm": 0.5467884540557861, "learning_rate": 0.0005, "loss": 1.2772411108016968, "step": 46236 }, { "epoch": 3.645013795821837, "grad_norm": 0.49023133516311646, "learning_rate": 0.0005, "loss": 1.2235755920410156, "step": 46237 }, { "epoch": 3.6450926290894756, "grad_norm": 0.444494366645813, "learning_rate": 0.0005, "loss": 1.1150147914886475, "step": 46238 }, { "epoch": 3.6451714623571148, "grad_norm": 0.469669908285141, "learning_rate": 0.0005, "loss": 1.076810359954834, "step": 46239 }, { "epoch": 3.645250295624754, "grad_norm": 0.4593539237976074, "learning_rate": 0.0005, "loss": 1.042121171951294, "step": 46240 }, { "epoch": 3.6453291288923926, "grad_norm": 0.4794800877571106, "learning_rate": 0.0005, "loss": 1.082157850265503, "step": 46241 }, { "epoch": 3.6454079621600313, "grad_norm": 0.5188964605331421, "learning_rate": 0.0005, "loss": 1.0732088088989258, "step": 46242 }, { "epoch": 3.6454867954276704, "grad_norm": 0.6324275732040405, "learning_rate": 0.0005, "loss": 1.2161340713500977, "step": 46243 }, { "epoch": 3.6455656286953095, "grad_norm": 0.5455074906349182, "learning_rate": 0.0005, "loss": 1.2039375305175781, "step": 46244 }, { "epoch": 3.6456444619629482, "grad_norm": 0.5574131608009338, "learning_rate": 0.0005, "loss": 1.1460256576538086, "step": 46245 }, { "epoch": 3.6457232952305874, "grad_norm": 0.4374275207519531, "learning_rate": 0.0005, "loss": 1.1588832139968872, "step": 46246 }, { "epoch": 3.645802128498226, "grad_norm": 0.6413246393203735, "learning_rate": 0.0005, "loss": 1.1418066024780273, "step": 46247 }, { "epoch": 3.645880961765865, "grad_norm": 0.5057283043861389, "learning_rate": 0.0005, "loss": 1.2182443141937256, "step": 46248 }, { "epoch": 3.6459597950335043, "grad_norm": 0.48263582587242126, "learning_rate": 0.0005, "loss": 1.2022852897644043, "step": 46249 }, { "epoch": 3.646038628301143, "grad_norm": 0.47020670771598816, "learning_rate": 0.0005, "loss": 1.1030921936035156, "step": 46250 }, { "epoch": 3.646117461568782, "grad_norm": 0.4656805694103241, "learning_rate": 0.0005, "loss": 1.0597847700119019, "step": 46251 }, { "epoch": 3.646196294836421, "grad_norm": 0.5504574179649353, "learning_rate": 0.0005, "loss": 1.2615779638290405, "step": 46252 }, { "epoch": 3.64627512810406, "grad_norm": 0.5839623212814331, "learning_rate": 0.0005, "loss": 1.3899362087249756, "step": 46253 }, { "epoch": 3.646353961371699, "grad_norm": 0.45074793696403503, "learning_rate": 0.0005, "loss": 1.0428502559661865, "step": 46254 }, { "epoch": 3.646432794639338, "grad_norm": 0.577097475528717, "learning_rate": 0.0005, "loss": 1.2154197692871094, "step": 46255 }, { "epoch": 3.6465116279069765, "grad_norm": 0.4059605002403259, "learning_rate": 0.0005, "loss": 1.0869832038879395, "step": 46256 }, { "epoch": 3.6465904611746156, "grad_norm": 0.4497803747653961, "learning_rate": 0.0005, "loss": 1.0819509029388428, "step": 46257 }, { "epoch": 3.6466692944422547, "grad_norm": 0.4880330264568329, "learning_rate": 0.0005, "loss": 1.074103593826294, "step": 46258 }, { "epoch": 3.6467481277098934, "grad_norm": 0.5068318843841553, "learning_rate": 0.0005, "loss": 1.249179720878601, "step": 46259 }, { "epoch": 3.6468269609775326, "grad_norm": 0.4681592881679535, "learning_rate": 0.0005, "loss": 1.1615073680877686, "step": 46260 }, { "epoch": 3.6469057942451713, "grad_norm": 0.43779948353767395, "learning_rate": 0.0005, "loss": 1.1299203634262085, "step": 46261 }, { "epoch": 3.6469846275128104, "grad_norm": 0.5171858668327332, "learning_rate": 0.0005, "loss": 1.3592113256454468, "step": 46262 }, { "epoch": 3.6470634607804495, "grad_norm": 0.5130707025527954, "learning_rate": 0.0005, "loss": 1.250365972518921, "step": 46263 }, { "epoch": 3.647142294048088, "grad_norm": 0.6807295680046082, "learning_rate": 0.0005, "loss": 1.1268123388290405, "step": 46264 }, { "epoch": 3.6472211273157273, "grad_norm": 0.4795200228691101, "learning_rate": 0.0005, "loss": 1.053110122680664, "step": 46265 }, { "epoch": 3.647299960583366, "grad_norm": 0.6244438886642456, "learning_rate": 0.0005, "loss": 1.1647453308105469, "step": 46266 }, { "epoch": 3.647378793851005, "grad_norm": 0.4672386944293976, "learning_rate": 0.0005, "loss": 1.1018954515457153, "step": 46267 }, { "epoch": 3.6474576271186443, "grad_norm": 0.505674421787262, "learning_rate": 0.0005, "loss": 1.1215535402297974, "step": 46268 }, { "epoch": 3.647536460386283, "grad_norm": 0.4954613745212555, "learning_rate": 0.0005, "loss": 1.0554111003875732, "step": 46269 }, { "epoch": 3.647615293653922, "grad_norm": 0.4387138783931732, "learning_rate": 0.0005, "loss": 1.1653412580490112, "step": 46270 }, { "epoch": 3.647694126921561, "grad_norm": 0.5366362929344177, "learning_rate": 0.0005, "loss": 1.3004124164581299, "step": 46271 }, { "epoch": 3.6477729601892, "grad_norm": 0.4114678204059601, "learning_rate": 0.0005, "loss": 0.9983065128326416, "step": 46272 }, { "epoch": 3.6478517934568386, "grad_norm": 0.6494480967521667, "learning_rate": 0.0005, "loss": 1.3278369903564453, "step": 46273 }, { "epoch": 3.6479306267244778, "grad_norm": 0.5168892741203308, "learning_rate": 0.0005, "loss": 1.204564094543457, "step": 46274 }, { "epoch": 3.6480094599921165, "grad_norm": 0.5021420121192932, "learning_rate": 0.0005, "loss": 1.1625901460647583, "step": 46275 }, { "epoch": 3.6480882932597556, "grad_norm": 0.480348140001297, "learning_rate": 0.0005, "loss": 0.9891635179519653, "step": 46276 }, { "epoch": 3.6481671265273947, "grad_norm": 0.3855845034122467, "learning_rate": 0.0005, "loss": 0.9804059863090515, "step": 46277 }, { "epoch": 3.6482459597950334, "grad_norm": 0.5508798956871033, "learning_rate": 0.0005, "loss": 1.225914478302002, "step": 46278 }, { "epoch": 3.6483247930626725, "grad_norm": 0.5250518321990967, "learning_rate": 0.0005, "loss": 1.0692696571350098, "step": 46279 }, { "epoch": 3.6484036263303112, "grad_norm": 0.41868269443511963, "learning_rate": 0.0005, "loss": 1.174619197845459, "step": 46280 }, { "epoch": 3.6484824595979504, "grad_norm": 0.43901684880256653, "learning_rate": 0.0005, "loss": 1.2472256422042847, "step": 46281 }, { "epoch": 3.6485612928655895, "grad_norm": 0.5132541656494141, "learning_rate": 0.0005, "loss": 1.3630683422088623, "step": 46282 }, { "epoch": 3.648640126133228, "grad_norm": 0.5214169025421143, "learning_rate": 0.0005, "loss": 1.1254609823226929, "step": 46283 }, { "epoch": 3.6487189594008673, "grad_norm": 0.46852272748947144, "learning_rate": 0.0005, "loss": 1.121896505355835, "step": 46284 }, { "epoch": 3.648797792668506, "grad_norm": 0.679145097732544, "learning_rate": 0.0005, "loss": 1.3114756345748901, "step": 46285 }, { "epoch": 3.648876625936145, "grad_norm": 0.47058936953544617, "learning_rate": 0.0005, "loss": 1.1631896495819092, "step": 46286 }, { "epoch": 3.648955459203784, "grad_norm": 0.521674633026123, "learning_rate": 0.0005, "loss": 1.3386247158050537, "step": 46287 }, { "epoch": 3.649034292471423, "grad_norm": 0.4397425353527069, "learning_rate": 0.0005, "loss": 1.296762228012085, "step": 46288 }, { "epoch": 3.6491131257390617, "grad_norm": 0.5943349599838257, "learning_rate": 0.0005, "loss": 1.2686896324157715, "step": 46289 }, { "epoch": 3.649191959006701, "grad_norm": 0.5388730764389038, "learning_rate": 0.0005, "loss": 0.966695249080658, "step": 46290 }, { "epoch": 3.64927079227434, "grad_norm": 0.6030099987983704, "learning_rate": 0.0005, "loss": 1.382813811302185, "step": 46291 }, { "epoch": 3.6493496255419786, "grad_norm": 0.5542486310005188, "learning_rate": 0.0005, "loss": 1.2271490097045898, "step": 46292 }, { "epoch": 3.6494284588096177, "grad_norm": 0.6530491709709167, "learning_rate": 0.0005, "loss": 1.3226293325424194, "step": 46293 }, { "epoch": 3.6495072920772564, "grad_norm": 0.5056484341621399, "learning_rate": 0.0005, "loss": 1.0841779708862305, "step": 46294 }, { "epoch": 3.6495861253448956, "grad_norm": 0.5386248826980591, "learning_rate": 0.0005, "loss": 1.1638259887695312, "step": 46295 }, { "epoch": 3.6496649586125347, "grad_norm": 0.5669705867767334, "learning_rate": 0.0005, "loss": 1.358144998550415, "step": 46296 }, { "epoch": 3.6497437918801734, "grad_norm": 0.48263925313949585, "learning_rate": 0.0005, "loss": 1.074367642402649, "step": 46297 }, { "epoch": 3.6498226251478125, "grad_norm": 0.5911639928817749, "learning_rate": 0.0005, "loss": 1.2636302709579468, "step": 46298 }, { "epoch": 3.649901458415451, "grad_norm": 0.5555673837661743, "learning_rate": 0.0005, "loss": 1.2309372425079346, "step": 46299 }, { "epoch": 3.6499802916830903, "grad_norm": 0.46973690390586853, "learning_rate": 0.0005, "loss": 1.323866605758667, "step": 46300 }, { "epoch": 3.6500591249507295, "grad_norm": 0.43143191933631897, "learning_rate": 0.0005, "loss": 1.0666518211364746, "step": 46301 }, { "epoch": 3.650137958218368, "grad_norm": 0.5026876330375671, "learning_rate": 0.0005, "loss": 1.17994225025177, "step": 46302 }, { "epoch": 3.650216791486007, "grad_norm": 0.6816680431365967, "learning_rate": 0.0005, "loss": 1.3523828983306885, "step": 46303 }, { "epoch": 3.650295624753646, "grad_norm": 0.4013863801956177, "learning_rate": 0.0005, "loss": 1.108654499053955, "step": 46304 }, { "epoch": 3.650374458021285, "grad_norm": 0.5869361758232117, "learning_rate": 0.0005, "loss": 1.0675588846206665, "step": 46305 }, { "epoch": 3.650453291288924, "grad_norm": 0.6852049231529236, "learning_rate": 0.0005, "loss": 1.4095182418823242, "step": 46306 }, { "epoch": 3.650532124556563, "grad_norm": 0.5448091626167297, "learning_rate": 0.0005, "loss": 1.1433136463165283, "step": 46307 }, { "epoch": 3.6506109578242016, "grad_norm": 0.5728530287742615, "learning_rate": 0.0005, "loss": 1.0821385383605957, "step": 46308 }, { "epoch": 3.6506897910918408, "grad_norm": 0.465180367231369, "learning_rate": 0.0005, "loss": 1.182873249053955, "step": 46309 }, { "epoch": 3.65076862435948, "grad_norm": 0.8826100826263428, "learning_rate": 0.0005, "loss": 1.438612461090088, "step": 46310 }, { "epoch": 3.6508474576271186, "grad_norm": 0.4937165081501007, "learning_rate": 0.0005, "loss": 1.2073912620544434, "step": 46311 }, { "epoch": 3.6509262908947577, "grad_norm": 0.6208871603012085, "learning_rate": 0.0005, "loss": 1.3958739042282104, "step": 46312 }, { "epoch": 3.6510051241623964, "grad_norm": 0.5570685267448425, "learning_rate": 0.0005, "loss": 1.2169111967086792, "step": 46313 }, { "epoch": 3.6510839574300356, "grad_norm": 0.4655509889125824, "learning_rate": 0.0005, "loss": 1.095147967338562, "step": 46314 }, { "epoch": 3.6511627906976747, "grad_norm": 0.542096734046936, "learning_rate": 0.0005, "loss": 1.4002983570098877, "step": 46315 }, { "epoch": 3.6512416239653134, "grad_norm": 0.4757806360721588, "learning_rate": 0.0005, "loss": 1.1169838905334473, "step": 46316 }, { "epoch": 3.651320457232952, "grad_norm": 0.4690834581851959, "learning_rate": 0.0005, "loss": 1.0503184795379639, "step": 46317 }, { "epoch": 3.651399290500591, "grad_norm": 0.4656909704208374, "learning_rate": 0.0005, "loss": 1.100819706916809, "step": 46318 }, { "epoch": 3.6514781237682303, "grad_norm": 0.5875913500785828, "learning_rate": 0.0005, "loss": 1.406280517578125, "step": 46319 }, { "epoch": 3.651556957035869, "grad_norm": 0.5284146070480347, "learning_rate": 0.0005, "loss": 1.2540701627731323, "step": 46320 }, { "epoch": 3.651635790303508, "grad_norm": 0.4090999364852905, "learning_rate": 0.0005, "loss": 1.0151526927947998, "step": 46321 }, { "epoch": 3.651714623571147, "grad_norm": 0.3997728228569031, "learning_rate": 0.0005, "loss": 1.1790965795516968, "step": 46322 }, { "epoch": 3.651793456838786, "grad_norm": 0.46914052963256836, "learning_rate": 0.0005, "loss": 1.340715765953064, "step": 46323 }, { "epoch": 3.651872290106425, "grad_norm": 0.48501357436180115, "learning_rate": 0.0005, "loss": 1.1809828281402588, "step": 46324 }, { "epoch": 3.651951123374064, "grad_norm": 0.5127187371253967, "learning_rate": 0.0005, "loss": 1.1384570598602295, "step": 46325 }, { "epoch": 3.652029956641703, "grad_norm": 0.3946731984615326, "learning_rate": 0.0005, "loss": 0.9760780334472656, "step": 46326 }, { "epoch": 3.6521087899093416, "grad_norm": 0.4242056608200073, "learning_rate": 0.0005, "loss": 1.1805189847946167, "step": 46327 }, { "epoch": 3.6521876231769808, "grad_norm": 0.47170114517211914, "learning_rate": 0.0005, "loss": 1.3809688091278076, "step": 46328 }, { "epoch": 3.65226645644462, "grad_norm": 0.5398576855659485, "learning_rate": 0.0005, "loss": 1.1272183656692505, "step": 46329 }, { "epoch": 3.6523452897122586, "grad_norm": 0.43359512090682983, "learning_rate": 0.0005, "loss": 1.1841599941253662, "step": 46330 }, { "epoch": 3.6524241229798973, "grad_norm": 0.5086712837219238, "learning_rate": 0.0005, "loss": 1.226739764213562, "step": 46331 }, { "epoch": 3.6525029562475364, "grad_norm": 0.6074042916297913, "learning_rate": 0.0005, "loss": 1.2351984977722168, "step": 46332 }, { "epoch": 3.6525817895151755, "grad_norm": 0.5321926474571228, "learning_rate": 0.0005, "loss": 1.2385191917419434, "step": 46333 }, { "epoch": 3.6526606227828142, "grad_norm": 0.487882137298584, "learning_rate": 0.0005, "loss": 1.2591307163238525, "step": 46334 }, { "epoch": 3.6527394560504534, "grad_norm": 0.5358293056488037, "learning_rate": 0.0005, "loss": 1.2851570844650269, "step": 46335 }, { "epoch": 3.652818289318092, "grad_norm": 0.5068095326423645, "learning_rate": 0.0005, "loss": 1.2223384380340576, "step": 46336 }, { "epoch": 3.652897122585731, "grad_norm": 0.4411281943321228, "learning_rate": 0.0005, "loss": 1.1219972372055054, "step": 46337 }, { "epoch": 3.6529759558533703, "grad_norm": 0.4458535313606262, "learning_rate": 0.0005, "loss": 1.178094744682312, "step": 46338 }, { "epoch": 3.653054789121009, "grad_norm": 0.46706944704055786, "learning_rate": 0.0005, "loss": 1.1435624361038208, "step": 46339 }, { "epoch": 3.653133622388648, "grad_norm": 0.5236262679100037, "learning_rate": 0.0005, "loss": 1.13853120803833, "step": 46340 }, { "epoch": 3.653212455656287, "grad_norm": 0.4555429220199585, "learning_rate": 0.0005, "loss": 1.110056757926941, "step": 46341 }, { "epoch": 3.653291288923926, "grad_norm": 0.4924401640892029, "learning_rate": 0.0005, "loss": 1.0708887577056885, "step": 46342 }, { "epoch": 3.653370122191565, "grad_norm": 0.42328858375549316, "learning_rate": 0.0005, "loss": 1.1757957935333252, "step": 46343 }, { "epoch": 3.653448955459204, "grad_norm": 0.5219388008117676, "learning_rate": 0.0005, "loss": 1.149538278579712, "step": 46344 }, { "epoch": 3.6535277887268425, "grad_norm": 0.4962308704853058, "learning_rate": 0.0005, "loss": 1.457326889038086, "step": 46345 }, { "epoch": 3.6536066219944816, "grad_norm": 0.4539641737937927, "learning_rate": 0.0005, "loss": 1.1229643821716309, "step": 46346 }, { "epoch": 3.6536854552621207, "grad_norm": 0.5783478617668152, "learning_rate": 0.0005, "loss": 1.1036394834518433, "step": 46347 }, { "epoch": 3.6537642885297594, "grad_norm": 0.5740907788276672, "learning_rate": 0.0005, "loss": 1.1312198638916016, "step": 46348 }, { "epoch": 3.6538431217973986, "grad_norm": 0.8335264921188354, "learning_rate": 0.0005, "loss": 1.5998294353485107, "step": 46349 }, { "epoch": 3.6539219550650373, "grad_norm": 0.48683804273605347, "learning_rate": 0.0005, "loss": 1.242000699043274, "step": 46350 }, { "epoch": 3.6540007883326764, "grad_norm": 0.465670645236969, "learning_rate": 0.0005, "loss": 1.165781021118164, "step": 46351 }, { "epoch": 3.6540796216003155, "grad_norm": 0.42482122778892517, "learning_rate": 0.0005, "loss": 0.9227443933486938, "step": 46352 }, { "epoch": 3.654158454867954, "grad_norm": 0.5251588821411133, "learning_rate": 0.0005, "loss": 1.1861248016357422, "step": 46353 }, { "epoch": 3.6542372881355933, "grad_norm": 0.39246320724487305, "learning_rate": 0.0005, "loss": 1.1473565101623535, "step": 46354 }, { "epoch": 3.654316121403232, "grad_norm": 0.35932523012161255, "learning_rate": 0.0005, "loss": 0.952118992805481, "step": 46355 }, { "epoch": 3.654394954670871, "grad_norm": 0.541268527507782, "learning_rate": 0.0005, "loss": 1.0329705476760864, "step": 46356 }, { "epoch": 3.6544737879385103, "grad_norm": 0.4451134502887726, "learning_rate": 0.0005, "loss": 1.1027655601501465, "step": 46357 }, { "epoch": 3.654552621206149, "grad_norm": 0.5917876362800598, "learning_rate": 0.0005, "loss": 1.139082431793213, "step": 46358 }, { "epoch": 3.6546314544737877, "grad_norm": 0.4851183295249939, "learning_rate": 0.0005, "loss": 1.196204423904419, "step": 46359 }, { "epoch": 3.654710287741427, "grad_norm": 0.5825375914573669, "learning_rate": 0.0005, "loss": 1.0182636976242065, "step": 46360 }, { "epoch": 3.654789121009066, "grad_norm": 0.49513664841651917, "learning_rate": 0.0005, "loss": 1.201226830482483, "step": 46361 }, { "epoch": 3.6548679542767046, "grad_norm": 0.4792111814022064, "learning_rate": 0.0005, "loss": 1.220036506652832, "step": 46362 }, { "epoch": 3.6549467875443438, "grad_norm": 0.540777325630188, "learning_rate": 0.0005, "loss": 1.2671778202056885, "step": 46363 }, { "epoch": 3.6550256208119825, "grad_norm": 0.5153986215591431, "learning_rate": 0.0005, "loss": 1.1896857023239136, "step": 46364 }, { "epoch": 3.6551044540796216, "grad_norm": 0.4533141255378723, "learning_rate": 0.0005, "loss": 1.1934282779693604, "step": 46365 }, { "epoch": 3.6551832873472607, "grad_norm": 0.42374834418296814, "learning_rate": 0.0005, "loss": 1.0275955200195312, "step": 46366 }, { "epoch": 3.6552621206148994, "grad_norm": 0.5919808745384216, "learning_rate": 0.0005, "loss": 1.2667791843414307, "step": 46367 }, { "epoch": 3.6553409538825385, "grad_norm": 0.44603392481803894, "learning_rate": 0.0005, "loss": 1.0886478424072266, "step": 46368 }, { "epoch": 3.6554197871501772, "grad_norm": 0.5182394981384277, "learning_rate": 0.0005, "loss": 1.1612672805786133, "step": 46369 }, { "epoch": 3.6554986204178164, "grad_norm": 0.5198284387588501, "learning_rate": 0.0005, "loss": 1.2698190212249756, "step": 46370 }, { "epoch": 3.6555774536854555, "grad_norm": 0.48082083463668823, "learning_rate": 0.0005, "loss": 1.120326042175293, "step": 46371 }, { "epoch": 3.655656286953094, "grad_norm": 0.4465419054031372, "learning_rate": 0.0005, "loss": 1.0557539463043213, "step": 46372 }, { "epoch": 3.655735120220733, "grad_norm": 0.5145940184593201, "learning_rate": 0.0005, "loss": 1.0089004039764404, "step": 46373 }, { "epoch": 3.655813953488372, "grad_norm": 0.5364261269569397, "learning_rate": 0.0005, "loss": 1.3003438711166382, "step": 46374 }, { "epoch": 3.655892786756011, "grad_norm": 0.5074783563613892, "learning_rate": 0.0005, "loss": 1.3330047130584717, "step": 46375 }, { "epoch": 3.65597162002365, "grad_norm": 0.4113892912864685, "learning_rate": 0.0005, "loss": 1.0446867942810059, "step": 46376 }, { "epoch": 3.656050453291289, "grad_norm": 0.4579208493232727, "learning_rate": 0.0005, "loss": 1.1171702146530151, "step": 46377 }, { "epoch": 3.6561292865589277, "grad_norm": 0.5255722999572754, "learning_rate": 0.0005, "loss": 1.220261573791504, "step": 46378 }, { "epoch": 3.656208119826567, "grad_norm": 0.47620052099227905, "learning_rate": 0.0005, "loss": 1.235085368156433, "step": 46379 }, { "epoch": 3.656286953094206, "grad_norm": 0.4789087474346161, "learning_rate": 0.0005, "loss": 1.174424648284912, "step": 46380 }, { "epoch": 3.6563657863618446, "grad_norm": 0.524868369102478, "learning_rate": 0.0005, "loss": 1.0710585117340088, "step": 46381 }, { "epoch": 3.6564446196294837, "grad_norm": 0.4911907911300659, "learning_rate": 0.0005, "loss": 1.0747475624084473, "step": 46382 }, { "epoch": 3.6565234528971224, "grad_norm": 0.5951114296913147, "learning_rate": 0.0005, "loss": 1.4071108102798462, "step": 46383 }, { "epoch": 3.6566022861647616, "grad_norm": 0.4680987298488617, "learning_rate": 0.0005, "loss": 1.017759084701538, "step": 46384 }, { "epoch": 3.6566811194324007, "grad_norm": 0.47718825936317444, "learning_rate": 0.0005, "loss": 1.1952743530273438, "step": 46385 }, { "epoch": 3.6567599527000394, "grad_norm": 0.4703616499900818, "learning_rate": 0.0005, "loss": 1.0569367408752441, "step": 46386 }, { "epoch": 3.656838785967678, "grad_norm": 0.4381409287452698, "learning_rate": 0.0005, "loss": 1.1547095775604248, "step": 46387 }, { "epoch": 3.656917619235317, "grad_norm": 0.5314014554023743, "learning_rate": 0.0005, "loss": 1.2736296653747559, "step": 46388 }, { "epoch": 3.6569964525029564, "grad_norm": 0.6127570271492004, "learning_rate": 0.0005, "loss": 1.1778026819229126, "step": 46389 }, { "epoch": 3.657075285770595, "grad_norm": 0.5227131843566895, "learning_rate": 0.0005, "loss": 1.1418732404708862, "step": 46390 }, { "epoch": 3.657154119038234, "grad_norm": 0.5406356453895569, "learning_rate": 0.0005, "loss": 1.233567237854004, "step": 46391 }, { "epoch": 3.657232952305873, "grad_norm": 0.5425230860710144, "learning_rate": 0.0005, "loss": 1.4124504327774048, "step": 46392 }, { "epoch": 3.657311785573512, "grad_norm": 0.617622435092926, "learning_rate": 0.0005, "loss": 1.1549361944198608, "step": 46393 }, { "epoch": 3.657390618841151, "grad_norm": 0.44742587208747864, "learning_rate": 0.0005, "loss": 1.1885299682617188, "step": 46394 }, { "epoch": 3.65746945210879, "grad_norm": 0.5126245021820068, "learning_rate": 0.0005, "loss": 1.0418771505355835, "step": 46395 }, { "epoch": 3.657548285376429, "grad_norm": 0.5259259343147278, "learning_rate": 0.0005, "loss": 1.2506027221679688, "step": 46396 }, { "epoch": 3.6576271186440676, "grad_norm": 0.48677292466163635, "learning_rate": 0.0005, "loss": 1.2253648042678833, "step": 46397 }, { "epoch": 3.6577059519117068, "grad_norm": 0.5871817469596863, "learning_rate": 0.0005, "loss": 1.1408216953277588, "step": 46398 }, { "epoch": 3.657784785179346, "grad_norm": 0.44386446475982666, "learning_rate": 0.0005, "loss": 1.2239015102386475, "step": 46399 }, { "epoch": 3.6578636184469846, "grad_norm": 0.4207960069179535, "learning_rate": 0.0005, "loss": 1.0870165824890137, "step": 46400 }, { "epoch": 3.6579424517146237, "grad_norm": 0.5446878671646118, "learning_rate": 0.0005, "loss": 1.0472502708435059, "step": 46401 }, { "epoch": 3.6580212849822624, "grad_norm": 0.42731937766075134, "learning_rate": 0.0005, "loss": 1.10843026638031, "step": 46402 }, { "epoch": 3.6581001182499016, "grad_norm": 0.5621083974838257, "learning_rate": 0.0005, "loss": 1.3239229917526245, "step": 46403 }, { "epoch": 3.6581789515175402, "grad_norm": 0.4868457615375519, "learning_rate": 0.0005, "loss": 1.2265502214431763, "step": 46404 }, { "epoch": 3.6582577847851794, "grad_norm": 0.5389367938041687, "learning_rate": 0.0005, "loss": 1.3315041065216064, "step": 46405 }, { "epoch": 3.658336618052818, "grad_norm": 0.6268399357795715, "learning_rate": 0.0005, "loss": 1.2581688165664673, "step": 46406 }, { "epoch": 3.658415451320457, "grad_norm": 0.4197821021080017, "learning_rate": 0.0005, "loss": 1.1748284101486206, "step": 46407 }, { "epoch": 3.6584942845880963, "grad_norm": 0.5105742812156677, "learning_rate": 0.0005, "loss": 1.1382951736450195, "step": 46408 }, { "epoch": 3.658573117855735, "grad_norm": 0.4278769791126251, "learning_rate": 0.0005, "loss": 1.1361695528030396, "step": 46409 }, { "epoch": 3.658651951123374, "grad_norm": 0.5324357151985168, "learning_rate": 0.0005, "loss": 1.1560139656066895, "step": 46410 }, { "epoch": 3.658730784391013, "grad_norm": 0.4367132782936096, "learning_rate": 0.0005, "loss": 1.02056884765625, "step": 46411 }, { "epoch": 3.658809617658652, "grad_norm": 0.5423095226287842, "learning_rate": 0.0005, "loss": 1.3799346685409546, "step": 46412 }, { "epoch": 3.658888450926291, "grad_norm": 0.56480473279953, "learning_rate": 0.0005, "loss": 1.2130351066589355, "step": 46413 }, { "epoch": 3.65896728419393, "grad_norm": 0.4990917444229126, "learning_rate": 0.0005, "loss": 1.0790786743164062, "step": 46414 }, { "epoch": 3.659046117461569, "grad_norm": 0.38772666454315186, "learning_rate": 0.0005, "loss": 1.045514702796936, "step": 46415 }, { "epoch": 3.6591249507292076, "grad_norm": 0.41848281025886536, "learning_rate": 0.0005, "loss": 1.1731640100479126, "step": 46416 }, { "epoch": 3.6592037839968468, "grad_norm": 0.5208784937858582, "learning_rate": 0.0005, "loss": 1.276733636856079, "step": 46417 }, { "epoch": 3.6592826172644854, "grad_norm": 0.44886812567710876, "learning_rate": 0.0005, "loss": 1.0364495515823364, "step": 46418 }, { "epoch": 3.6593614505321246, "grad_norm": 0.5194827318191528, "learning_rate": 0.0005, "loss": 1.314860463142395, "step": 46419 }, { "epoch": 3.6594402837997633, "grad_norm": 0.575395405292511, "learning_rate": 0.0005, "loss": 1.1600184440612793, "step": 46420 }, { "epoch": 3.6595191170674024, "grad_norm": 0.4573901891708374, "learning_rate": 0.0005, "loss": 1.1095964908599854, "step": 46421 }, { "epoch": 3.6595979503350415, "grad_norm": 0.6451182961463928, "learning_rate": 0.0005, "loss": 1.2710765600204468, "step": 46422 }, { "epoch": 3.6596767836026802, "grad_norm": 0.5190736055374146, "learning_rate": 0.0005, "loss": 1.1252334117889404, "step": 46423 }, { "epoch": 3.6597556168703194, "grad_norm": 0.5647477507591248, "learning_rate": 0.0005, "loss": 1.2048735618591309, "step": 46424 }, { "epoch": 3.659834450137958, "grad_norm": 0.4648444652557373, "learning_rate": 0.0005, "loss": 0.9996271729469299, "step": 46425 }, { "epoch": 3.659913283405597, "grad_norm": 0.6780761480331421, "learning_rate": 0.0005, "loss": 1.3518946170806885, "step": 46426 }, { "epoch": 3.6599921166732363, "grad_norm": 0.45990362763404846, "learning_rate": 0.0005, "loss": 1.2889621257781982, "step": 46427 }, { "epoch": 3.660070949940875, "grad_norm": 0.592395007610321, "learning_rate": 0.0005, "loss": 1.051903486251831, "step": 46428 }, { "epoch": 3.660149783208514, "grad_norm": 0.5120651125907898, "learning_rate": 0.0005, "loss": 1.233644962310791, "step": 46429 }, { "epoch": 3.660228616476153, "grad_norm": 0.514212429523468, "learning_rate": 0.0005, "loss": 1.21761155128479, "step": 46430 }, { "epoch": 3.660307449743792, "grad_norm": 0.5447637438774109, "learning_rate": 0.0005, "loss": 1.36846125125885, "step": 46431 }, { "epoch": 3.660386283011431, "grad_norm": 0.428117960691452, "learning_rate": 0.0005, "loss": 1.0729930400848389, "step": 46432 }, { "epoch": 3.66046511627907, "grad_norm": 0.5563521385192871, "learning_rate": 0.0005, "loss": 1.111837387084961, "step": 46433 }, { "epoch": 3.6605439495467085, "grad_norm": 0.5103297233581543, "learning_rate": 0.0005, "loss": 1.1957749128341675, "step": 46434 }, { "epoch": 3.6606227828143476, "grad_norm": 0.5094735026359558, "learning_rate": 0.0005, "loss": 1.25057053565979, "step": 46435 }, { "epoch": 3.6607016160819867, "grad_norm": 0.5861404538154602, "learning_rate": 0.0005, "loss": 1.240684151649475, "step": 46436 }, { "epoch": 3.6607804493496254, "grad_norm": 0.8678625822067261, "learning_rate": 0.0005, "loss": 1.140365719795227, "step": 46437 }, { "epoch": 3.6608592826172646, "grad_norm": 0.46781644225120544, "learning_rate": 0.0005, "loss": 1.0527584552764893, "step": 46438 }, { "epoch": 3.6609381158849033, "grad_norm": 0.48899638652801514, "learning_rate": 0.0005, "loss": 1.2188966274261475, "step": 46439 }, { "epoch": 3.6610169491525424, "grad_norm": 0.5105384588241577, "learning_rate": 0.0005, "loss": 1.0401145219802856, "step": 46440 }, { "epoch": 3.6610957824201815, "grad_norm": 0.6039490103721619, "learning_rate": 0.0005, "loss": 1.2831032276153564, "step": 46441 }, { "epoch": 3.66117461568782, "grad_norm": 0.4305097758769989, "learning_rate": 0.0005, "loss": 1.103588581085205, "step": 46442 }, { "epoch": 3.6612534489554593, "grad_norm": 0.5145366191864014, "learning_rate": 0.0005, "loss": 1.0858269929885864, "step": 46443 }, { "epoch": 3.661332282223098, "grad_norm": 0.6857478022575378, "learning_rate": 0.0005, "loss": 1.2577722072601318, "step": 46444 }, { "epoch": 3.661411115490737, "grad_norm": 0.5267702341079712, "learning_rate": 0.0005, "loss": 1.297473430633545, "step": 46445 }, { "epoch": 3.6614899487583763, "grad_norm": 0.5221107006072998, "learning_rate": 0.0005, "loss": 1.1677616834640503, "step": 46446 }, { "epoch": 3.661568782026015, "grad_norm": 0.4124988913536072, "learning_rate": 0.0005, "loss": 1.0733006000518799, "step": 46447 }, { "epoch": 3.6616476152936537, "grad_norm": 0.5430581569671631, "learning_rate": 0.0005, "loss": 1.3141000270843506, "step": 46448 }, { "epoch": 3.661726448561293, "grad_norm": 0.5406635999679565, "learning_rate": 0.0005, "loss": 1.0757801532745361, "step": 46449 }, { "epoch": 3.661805281828932, "grad_norm": 0.5210402607917786, "learning_rate": 0.0005, "loss": 1.1819303035736084, "step": 46450 }, { "epoch": 3.6618841150965706, "grad_norm": 0.4733262360095978, "learning_rate": 0.0005, "loss": 1.0142872333526611, "step": 46451 }, { "epoch": 3.6619629483642098, "grad_norm": 0.5393936038017273, "learning_rate": 0.0005, "loss": 1.2766289710998535, "step": 46452 }, { "epoch": 3.6620417816318485, "grad_norm": 0.558923602104187, "learning_rate": 0.0005, "loss": 1.126790165901184, "step": 46453 }, { "epoch": 3.6621206148994876, "grad_norm": 0.4382665157318115, "learning_rate": 0.0005, "loss": 1.1859114170074463, "step": 46454 }, { "epoch": 3.6621994481671267, "grad_norm": 0.44832226634025574, "learning_rate": 0.0005, "loss": 1.0302650928497314, "step": 46455 }, { "epoch": 3.6622782814347654, "grad_norm": 0.5187481045722961, "learning_rate": 0.0005, "loss": 1.1751911640167236, "step": 46456 }, { "epoch": 3.6623571147024045, "grad_norm": 0.4818568825721741, "learning_rate": 0.0005, "loss": 1.0040183067321777, "step": 46457 }, { "epoch": 3.6624359479700432, "grad_norm": 0.5261511206626892, "learning_rate": 0.0005, "loss": 1.2024081945419312, "step": 46458 }, { "epoch": 3.6625147812376824, "grad_norm": 0.5904791951179504, "learning_rate": 0.0005, "loss": 1.106614351272583, "step": 46459 }, { "epoch": 3.6625936145053215, "grad_norm": 0.49045154452323914, "learning_rate": 0.0005, "loss": 1.039720892906189, "step": 46460 }, { "epoch": 3.66267244777296, "grad_norm": 0.7405903339385986, "learning_rate": 0.0005, "loss": 1.4417821168899536, "step": 46461 }, { "epoch": 3.662751281040599, "grad_norm": 0.5167423486709595, "learning_rate": 0.0005, "loss": 1.4202067852020264, "step": 46462 }, { "epoch": 3.662830114308238, "grad_norm": 0.48655837774276733, "learning_rate": 0.0005, "loss": 1.1075255870819092, "step": 46463 }, { "epoch": 3.662908947575877, "grad_norm": 0.5655698776245117, "learning_rate": 0.0005, "loss": 1.1150317192077637, "step": 46464 }, { "epoch": 3.662987780843516, "grad_norm": 0.46303680539131165, "learning_rate": 0.0005, "loss": 1.2527120113372803, "step": 46465 }, { "epoch": 3.663066614111155, "grad_norm": 0.5974205732345581, "learning_rate": 0.0005, "loss": 1.067613124847412, "step": 46466 }, { "epoch": 3.6631454473787937, "grad_norm": 0.43408942222595215, "learning_rate": 0.0005, "loss": 1.202863335609436, "step": 46467 }, { "epoch": 3.663224280646433, "grad_norm": 0.6205035448074341, "learning_rate": 0.0005, "loss": 1.112754464149475, "step": 46468 }, { "epoch": 3.663303113914072, "grad_norm": 0.563845694065094, "learning_rate": 0.0005, "loss": 1.4006317853927612, "step": 46469 }, { "epoch": 3.6633819471817106, "grad_norm": 0.4977625012397766, "learning_rate": 0.0005, "loss": 1.0274195671081543, "step": 46470 }, { "epoch": 3.6634607804493498, "grad_norm": 0.6146444082260132, "learning_rate": 0.0005, "loss": 1.2397245168685913, "step": 46471 }, { "epoch": 3.6635396137169884, "grad_norm": 0.45039841532707214, "learning_rate": 0.0005, "loss": 1.0864112377166748, "step": 46472 }, { "epoch": 3.6636184469846276, "grad_norm": 0.6352031826972961, "learning_rate": 0.0005, "loss": 1.2609866857528687, "step": 46473 }, { "epoch": 3.6636972802522667, "grad_norm": 0.6107380986213684, "learning_rate": 0.0005, "loss": 1.3959959745407104, "step": 46474 }, { "epoch": 3.6637761135199054, "grad_norm": 0.44268298149108887, "learning_rate": 0.0005, "loss": 1.2440340518951416, "step": 46475 }, { "epoch": 3.663854946787544, "grad_norm": 0.42917314171791077, "learning_rate": 0.0005, "loss": 0.9206855297088623, "step": 46476 }, { "epoch": 3.663933780055183, "grad_norm": 0.44889745116233826, "learning_rate": 0.0005, "loss": 1.0295687913894653, "step": 46477 }, { "epoch": 3.6640126133228224, "grad_norm": 0.4493635892868042, "learning_rate": 0.0005, "loss": 0.8684195876121521, "step": 46478 }, { "epoch": 3.664091446590461, "grad_norm": 0.5912221074104309, "learning_rate": 0.0005, "loss": 1.1722530126571655, "step": 46479 }, { "epoch": 3.6641702798581, "grad_norm": 0.45785829424858093, "learning_rate": 0.0005, "loss": 1.1363229751586914, "step": 46480 }, { "epoch": 3.664249113125739, "grad_norm": 0.44513726234436035, "learning_rate": 0.0005, "loss": 1.203627586364746, "step": 46481 }, { "epoch": 3.664327946393378, "grad_norm": 0.5508202314376831, "learning_rate": 0.0005, "loss": 1.2018595933914185, "step": 46482 }, { "epoch": 3.664406779661017, "grad_norm": 0.48095306754112244, "learning_rate": 0.0005, "loss": 1.1052746772766113, "step": 46483 }, { "epoch": 3.664485612928656, "grad_norm": 0.5477583408355713, "learning_rate": 0.0005, "loss": 1.2331925630569458, "step": 46484 }, { "epoch": 3.664564446196295, "grad_norm": 0.5238315463066101, "learning_rate": 0.0005, "loss": 1.2852420806884766, "step": 46485 }, { "epoch": 3.6646432794639336, "grad_norm": 0.5785328149795532, "learning_rate": 0.0005, "loss": 1.24410879611969, "step": 46486 }, { "epoch": 3.664722112731573, "grad_norm": 0.6525012850761414, "learning_rate": 0.0005, "loss": 1.3619154691696167, "step": 46487 }, { "epoch": 3.664800945999212, "grad_norm": 0.5909388661384583, "learning_rate": 0.0005, "loss": 1.4359548091888428, "step": 46488 }, { "epoch": 3.6648797792668506, "grad_norm": 0.4425332844257355, "learning_rate": 0.0005, "loss": 1.285530686378479, "step": 46489 }, { "epoch": 3.6649586125344893, "grad_norm": 0.4656549096107483, "learning_rate": 0.0005, "loss": 1.0009537935256958, "step": 46490 }, { "epoch": 3.6650374458021284, "grad_norm": 0.5815526247024536, "learning_rate": 0.0005, "loss": 1.3343102931976318, "step": 46491 }, { "epoch": 3.6651162790697676, "grad_norm": 0.509303092956543, "learning_rate": 0.0005, "loss": 1.1190208196640015, "step": 46492 }, { "epoch": 3.6651951123374062, "grad_norm": 0.4763169586658478, "learning_rate": 0.0005, "loss": 1.324671745300293, "step": 46493 }, { "epoch": 3.6652739456050454, "grad_norm": 0.4961045980453491, "learning_rate": 0.0005, "loss": 1.4118419885635376, "step": 46494 }, { "epoch": 3.665352778872684, "grad_norm": 0.5225496292114258, "learning_rate": 0.0005, "loss": 1.1449530124664307, "step": 46495 }, { "epoch": 3.665431612140323, "grad_norm": 0.5303519368171692, "learning_rate": 0.0005, "loss": 1.1654322147369385, "step": 46496 }, { "epoch": 3.6655104454079623, "grad_norm": 0.42706099152565, "learning_rate": 0.0005, "loss": 1.0374397039413452, "step": 46497 }, { "epoch": 3.665589278675601, "grad_norm": 0.4630277454853058, "learning_rate": 0.0005, "loss": 1.16618013381958, "step": 46498 }, { "epoch": 3.66566811194324, "grad_norm": 0.512614369392395, "learning_rate": 0.0005, "loss": 1.1483798027038574, "step": 46499 }, { "epoch": 3.665746945210879, "grad_norm": 0.5023557543754578, "learning_rate": 0.0005, "loss": 1.1619377136230469, "step": 46500 }, { "epoch": 3.665825778478518, "grad_norm": 0.43519842624664307, "learning_rate": 0.0005, "loss": 1.2250804901123047, "step": 46501 }, { "epoch": 3.665904611746157, "grad_norm": 0.4080950617790222, "learning_rate": 0.0005, "loss": 1.0251506567001343, "step": 46502 }, { "epoch": 3.665983445013796, "grad_norm": 0.47888246178627014, "learning_rate": 0.0005, "loss": 1.1167186498641968, "step": 46503 }, { "epoch": 3.6660622782814345, "grad_norm": 0.5244226455688477, "learning_rate": 0.0005, "loss": 1.2620375156402588, "step": 46504 }, { "epoch": 3.6661411115490736, "grad_norm": 0.5193426012992859, "learning_rate": 0.0005, "loss": 1.0553929805755615, "step": 46505 }, { "epoch": 3.6662199448167128, "grad_norm": 0.4265753924846649, "learning_rate": 0.0005, "loss": 1.107362151145935, "step": 46506 }, { "epoch": 3.6662987780843515, "grad_norm": 0.5274451375007629, "learning_rate": 0.0005, "loss": 1.10365891456604, "step": 46507 }, { "epoch": 3.6663776113519906, "grad_norm": 0.542218029499054, "learning_rate": 0.0005, "loss": 1.2945176362991333, "step": 46508 }, { "epoch": 3.6664564446196293, "grad_norm": 0.5532825589179993, "learning_rate": 0.0005, "loss": 1.2028279304504395, "step": 46509 }, { "epoch": 3.6665352778872684, "grad_norm": 0.4789067506790161, "learning_rate": 0.0005, "loss": 1.190636396408081, "step": 46510 }, { "epoch": 3.6666141111549075, "grad_norm": 0.4273386597633362, "learning_rate": 0.0005, "loss": 1.159049391746521, "step": 46511 }, { "epoch": 3.6666929444225462, "grad_norm": 0.5530059933662415, "learning_rate": 0.0005, "loss": 1.3093817234039307, "step": 46512 }, { "epoch": 3.6667717776901854, "grad_norm": 0.4555676281452179, "learning_rate": 0.0005, "loss": 0.9581876993179321, "step": 46513 }, { "epoch": 3.666850610957824, "grad_norm": 0.43382012844085693, "learning_rate": 0.0005, "loss": 1.1056772470474243, "step": 46514 }, { "epoch": 3.666929444225463, "grad_norm": 0.4098105728626251, "learning_rate": 0.0005, "loss": 1.077417016029358, "step": 46515 }, { "epoch": 3.6670082774931023, "grad_norm": 0.5280797481536865, "learning_rate": 0.0005, "loss": 1.2533166408538818, "step": 46516 }, { "epoch": 3.667087110760741, "grad_norm": 0.5262672901153564, "learning_rate": 0.0005, "loss": 1.3911893367767334, "step": 46517 }, { "epoch": 3.66716594402838, "grad_norm": 0.4752308428287506, "learning_rate": 0.0005, "loss": 1.3859189748764038, "step": 46518 }, { "epoch": 3.667244777296019, "grad_norm": 0.41513699293136597, "learning_rate": 0.0005, "loss": 1.0840308666229248, "step": 46519 }, { "epoch": 3.667323610563658, "grad_norm": 0.4555183947086334, "learning_rate": 0.0005, "loss": 1.0601813793182373, "step": 46520 }, { "epoch": 3.6674024438312967, "grad_norm": 0.5916547775268555, "learning_rate": 0.0005, "loss": 1.137695074081421, "step": 46521 }, { "epoch": 3.667481277098936, "grad_norm": 0.42486417293548584, "learning_rate": 0.0005, "loss": 1.2148163318634033, "step": 46522 }, { "epoch": 3.6675601103665745, "grad_norm": 0.415990948677063, "learning_rate": 0.0005, "loss": 1.0555734634399414, "step": 46523 }, { "epoch": 3.6676389436342136, "grad_norm": 0.7189897298812866, "learning_rate": 0.0005, "loss": 1.2190330028533936, "step": 46524 }, { "epoch": 3.6677177769018527, "grad_norm": 0.4351230561733246, "learning_rate": 0.0005, "loss": 1.0655295848846436, "step": 46525 }, { "epoch": 3.6677966101694914, "grad_norm": 0.5586251616477966, "learning_rate": 0.0005, "loss": 0.9670956134796143, "step": 46526 }, { "epoch": 3.6678754434371306, "grad_norm": 0.4498153328895569, "learning_rate": 0.0005, "loss": 1.0006210803985596, "step": 46527 }, { "epoch": 3.6679542767047693, "grad_norm": 0.5022490620613098, "learning_rate": 0.0005, "loss": 1.3810019493103027, "step": 46528 }, { "epoch": 3.6680331099724084, "grad_norm": 0.4416450262069702, "learning_rate": 0.0005, "loss": 1.2730369567871094, "step": 46529 }, { "epoch": 3.6681119432400475, "grad_norm": 0.4916452467441559, "learning_rate": 0.0005, "loss": 1.2006173133850098, "step": 46530 }, { "epoch": 3.668190776507686, "grad_norm": 0.5321526527404785, "learning_rate": 0.0005, "loss": 1.176278829574585, "step": 46531 }, { "epoch": 3.6682696097753253, "grad_norm": 0.5114721655845642, "learning_rate": 0.0005, "loss": 1.1856859922409058, "step": 46532 }, { "epoch": 3.668348443042964, "grad_norm": 0.4100472927093506, "learning_rate": 0.0005, "loss": 1.180772066116333, "step": 46533 }, { "epoch": 3.668427276310603, "grad_norm": 0.52713543176651, "learning_rate": 0.0005, "loss": 1.2266048192977905, "step": 46534 }, { "epoch": 3.668506109578242, "grad_norm": 0.4330287277698517, "learning_rate": 0.0005, "loss": 1.1376006603240967, "step": 46535 }, { "epoch": 3.668584942845881, "grad_norm": 0.4660966098308563, "learning_rate": 0.0005, "loss": 1.0099549293518066, "step": 46536 }, { "epoch": 3.6686637761135197, "grad_norm": 0.5245071649551392, "learning_rate": 0.0005, "loss": 1.5194628238677979, "step": 46537 }, { "epoch": 3.668742609381159, "grad_norm": 0.4098803997039795, "learning_rate": 0.0005, "loss": 1.1617894172668457, "step": 46538 }, { "epoch": 3.668821442648798, "grad_norm": 0.5057011246681213, "learning_rate": 0.0005, "loss": 1.3236247301101685, "step": 46539 }, { "epoch": 3.6689002759164366, "grad_norm": 0.4874818027019501, "learning_rate": 0.0005, "loss": 1.117114543914795, "step": 46540 }, { "epoch": 3.6689791091840758, "grad_norm": 0.5542557835578918, "learning_rate": 0.0005, "loss": 1.1504850387573242, "step": 46541 }, { "epoch": 3.6690579424517145, "grad_norm": 0.5347785353660583, "learning_rate": 0.0005, "loss": 1.239790916442871, "step": 46542 }, { "epoch": 3.6691367757193536, "grad_norm": 0.4209137260913849, "learning_rate": 0.0005, "loss": 1.0554823875427246, "step": 46543 }, { "epoch": 3.6692156089869927, "grad_norm": 0.45172742009162903, "learning_rate": 0.0005, "loss": 1.0472500324249268, "step": 46544 }, { "epoch": 3.6692944422546314, "grad_norm": 0.4005625545978546, "learning_rate": 0.0005, "loss": 1.065589189529419, "step": 46545 }, { "epoch": 3.6693732755222705, "grad_norm": 0.5290811657905579, "learning_rate": 0.0005, "loss": 1.127733826637268, "step": 46546 }, { "epoch": 3.6694521087899092, "grad_norm": 0.48504677414894104, "learning_rate": 0.0005, "loss": 1.2518755197525024, "step": 46547 }, { "epoch": 3.6695309420575484, "grad_norm": 0.4296618103981018, "learning_rate": 0.0005, "loss": 1.0254335403442383, "step": 46548 }, { "epoch": 3.6696097753251875, "grad_norm": 0.8356395363807678, "learning_rate": 0.0005, "loss": 1.109748363494873, "step": 46549 }, { "epoch": 3.669688608592826, "grad_norm": 0.4304676055908203, "learning_rate": 0.0005, "loss": 1.122079849243164, "step": 46550 }, { "epoch": 3.669767441860465, "grad_norm": 0.487108051776886, "learning_rate": 0.0005, "loss": 1.0403118133544922, "step": 46551 }, { "epoch": 3.669846275128104, "grad_norm": 0.4565827548503876, "learning_rate": 0.0005, "loss": 1.1209684610366821, "step": 46552 }, { "epoch": 3.669925108395743, "grad_norm": 0.4224412143230438, "learning_rate": 0.0005, "loss": 1.0532833337783813, "step": 46553 }, { "epoch": 3.670003941663382, "grad_norm": 0.49070101976394653, "learning_rate": 0.0005, "loss": 1.2714662551879883, "step": 46554 }, { "epoch": 3.670082774931021, "grad_norm": 0.6264057755470276, "learning_rate": 0.0005, "loss": 1.2523752450942993, "step": 46555 }, { "epoch": 3.6701616081986597, "grad_norm": 0.5620990991592407, "learning_rate": 0.0005, "loss": 1.0518414974212646, "step": 46556 }, { "epoch": 3.670240441466299, "grad_norm": 0.48739093542099, "learning_rate": 0.0005, "loss": 1.2488504648208618, "step": 46557 }, { "epoch": 3.670319274733938, "grad_norm": 0.4057793915271759, "learning_rate": 0.0005, "loss": 1.0452018976211548, "step": 46558 }, { "epoch": 3.6703981080015766, "grad_norm": 0.4694465696811676, "learning_rate": 0.0005, "loss": 1.155116081237793, "step": 46559 }, { "epoch": 3.6704769412692158, "grad_norm": 0.6275566816329956, "learning_rate": 0.0005, "loss": 1.3597979545593262, "step": 46560 }, { "epoch": 3.6705557745368544, "grad_norm": 0.47413015365600586, "learning_rate": 0.0005, "loss": 1.0879249572753906, "step": 46561 }, { "epoch": 3.6706346078044936, "grad_norm": 0.47700703144073486, "learning_rate": 0.0005, "loss": 1.0884151458740234, "step": 46562 }, { "epoch": 3.6707134410721327, "grad_norm": 0.5038704872131348, "learning_rate": 0.0005, "loss": 1.3255813121795654, "step": 46563 }, { "epoch": 3.6707922743397714, "grad_norm": 0.45846495032310486, "learning_rate": 0.0005, "loss": 1.1516246795654297, "step": 46564 }, { "epoch": 3.67087110760741, "grad_norm": 0.4833492636680603, "learning_rate": 0.0005, "loss": 1.1511774063110352, "step": 46565 }, { "epoch": 3.6709499408750492, "grad_norm": 0.5025455951690674, "learning_rate": 0.0005, "loss": 1.2063934803009033, "step": 46566 }, { "epoch": 3.6710287741426884, "grad_norm": 0.4462474286556244, "learning_rate": 0.0005, "loss": 1.133608341217041, "step": 46567 }, { "epoch": 3.671107607410327, "grad_norm": 0.5058015584945679, "learning_rate": 0.0005, "loss": 1.261659860610962, "step": 46568 }, { "epoch": 3.671186440677966, "grad_norm": 0.4979894459247589, "learning_rate": 0.0005, "loss": 1.2193822860717773, "step": 46569 }, { "epoch": 3.671265273945605, "grad_norm": 0.43549656867980957, "learning_rate": 0.0005, "loss": 1.1292221546173096, "step": 46570 }, { "epoch": 3.671344107213244, "grad_norm": 0.676784336566925, "learning_rate": 0.0005, "loss": 1.3277525901794434, "step": 46571 }, { "epoch": 3.671422940480883, "grad_norm": 0.4916597902774811, "learning_rate": 0.0005, "loss": 1.1734299659729004, "step": 46572 }, { "epoch": 3.671501773748522, "grad_norm": 0.4662167727947235, "learning_rate": 0.0005, "loss": 1.2656104564666748, "step": 46573 }, { "epoch": 3.671580607016161, "grad_norm": 0.46704626083374023, "learning_rate": 0.0005, "loss": 1.1683282852172852, "step": 46574 }, { "epoch": 3.6716594402837996, "grad_norm": 0.6636156439781189, "learning_rate": 0.0005, "loss": 1.0685341358184814, "step": 46575 }, { "epoch": 3.671738273551439, "grad_norm": 0.46316149830818176, "learning_rate": 0.0005, "loss": 1.1883965730667114, "step": 46576 }, { "epoch": 3.671817106819078, "grad_norm": 0.5693379044532776, "learning_rate": 0.0005, "loss": 1.1658775806427002, "step": 46577 }, { "epoch": 3.6718959400867166, "grad_norm": 0.5704672336578369, "learning_rate": 0.0005, "loss": 1.127548336982727, "step": 46578 }, { "epoch": 3.6719747733543553, "grad_norm": 0.4705476462841034, "learning_rate": 0.0005, "loss": 1.065563440322876, "step": 46579 }, { "epoch": 3.6720536066219944, "grad_norm": 0.5186314582824707, "learning_rate": 0.0005, "loss": 1.2020690441131592, "step": 46580 }, { "epoch": 3.6721324398896336, "grad_norm": 0.5976051092147827, "learning_rate": 0.0005, "loss": 1.1706129312515259, "step": 46581 }, { "epoch": 3.6722112731572722, "grad_norm": 0.40382906794548035, "learning_rate": 0.0005, "loss": 1.0850937366485596, "step": 46582 }, { "epoch": 3.6722901064249114, "grad_norm": 0.534530520439148, "learning_rate": 0.0005, "loss": 1.109731912612915, "step": 46583 }, { "epoch": 3.67236893969255, "grad_norm": 0.5389239192008972, "learning_rate": 0.0005, "loss": 1.1867623329162598, "step": 46584 }, { "epoch": 3.672447772960189, "grad_norm": 0.5931677222251892, "learning_rate": 0.0005, "loss": 1.3926036357879639, "step": 46585 }, { "epoch": 3.6725266062278283, "grad_norm": 0.5383573174476624, "learning_rate": 0.0005, "loss": 1.127317190170288, "step": 46586 }, { "epoch": 3.672605439495467, "grad_norm": 0.4669741690158844, "learning_rate": 0.0005, "loss": 1.207931637763977, "step": 46587 }, { "epoch": 3.672684272763106, "grad_norm": 0.44716599583625793, "learning_rate": 0.0005, "loss": 1.196187138557434, "step": 46588 }, { "epoch": 3.672763106030745, "grad_norm": 0.4668916165828705, "learning_rate": 0.0005, "loss": 1.0757954120635986, "step": 46589 }, { "epoch": 3.672841939298384, "grad_norm": 0.43117961287498474, "learning_rate": 0.0005, "loss": 1.0808193683624268, "step": 46590 }, { "epoch": 3.672920772566023, "grad_norm": 0.42174991965293884, "learning_rate": 0.0005, "loss": 1.1745494604110718, "step": 46591 }, { "epoch": 3.672999605833662, "grad_norm": 0.5004523396492004, "learning_rate": 0.0005, "loss": 1.1258115768432617, "step": 46592 }, { "epoch": 3.6730784391013005, "grad_norm": 0.4012010395526886, "learning_rate": 0.0005, "loss": 0.9683579206466675, "step": 46593 }, { "epoch": 3.6731572723689396, "grad_norm": 0.45651775598526, "learning_rate": 0.0005, "loss": 1.0818939208984375, "step": 46594 }, { "epoch": 3.6732361056365788, "grad_norm": 0.5868698358535767, "learning_rate": 0.0005, "loss": 1.3274863958358765, "step": 46595 }, { "epoch": 3.6733149389042175, "grad_norm": 0.461457759141922, "learning_rate": 0.0005, "loss": 1.1948535442352295, "step": 46596 }, { "epoch": 3.6733937721718566, "grad_norm": 0.4588591456413269, "learning_rate": 0.0005, "loss": 1.1721546649932861, "step": 46597 }, { "epoch": 3.6734726054394953, "grad_norm": 0.48479917645454407, "learning_rate": 0.0005, "loss": 1.041945219039917, "step": 46598 }, { "epoch": 3.6735514387071344, "grad_norm": 0.5116682052612305, "learning_rate": 0.0005, "loss": 1.2645752429962158, "step": 46599 }, { "epoch": 3.6736302719747735, "grad_norm": 0.5833112001419067, "learning_rate": 0.0005, "loss": 1.3686084747314453, "step": 46600 }, { "epoch": 3.6737091052424122, "grad_norm": 0.4885164797306061, "learning_rate": 0.0005, "loss": 1.242803692817688, "step": 46601 }, { "epoch": 3.6737879385100514, "grad_norm": 0.5088920593261719, "learning_rate": 0.0005, "loss": 1.250566005706787, "step": 46602 }, { "epoch": 3.67386677177769, "grad_norm": 0.5938891768455505, "learning_rate": 0.0005, "loss": 1.142837405204773, "step": 46603 }, { "epoch": 3.673945605045329, "grad_norm": 0.5022438168525696, "learning_rate": 0.0005, "loss": 1.3105515241622925, "step": 46604 }, { "epoch": 3.6740244383129683, "grad_norm": 0.5214592218399048, "learning_rate": 0.0005, "loss": 1.3121964931488037, "step": 46605 }, { "epoch": 3.674103271580607, "grad_norm": 0.5002555251121521, "learning_rate": 0.0005, "loss": 1.0299606323242188, "step": 46606 }, { "epoch": 3.6741821048482457, "grad_norm": 0.47508174180984497, "learning_rate": 0.0005, "loss": 1.0426721572875977, "step": 46607 }, { "epoch": 3.674260938115885, "grad_norm": 0.45235177874565125, "learning_rate": 0.0005, "loss": 1.0555356740951538, "step": 46608 }, { "epoch": 3.674339771383524, "grad_norm": 0.5651640295982361, "learning_rate": 0.0005, "loss": 1.395583152770996, "step": 46609 }, { "epoch": 3.6744186046511627, "grad_norm": 0.4637162685394287, "learning_rate": 0.0005, "loss": 1.1150052547454834, "step": 46610 }, { "epoch": 3.674497437918802, "grad_norm": 0.42033517360687256, "learning_rate": 0.0005, "loss": 1.1684277057647705, "step": 46611 }, { "epoch": 3.6745762711864405, "grad_norm": 0.548331081867218, "learning_rate": 0.0005, "loss": 1.2696597576141357, "step": 46612 }, { "epoch": 3.6746551044540796, "grad_norm": 0.44332897663116455, "learning_rate": 0.0005, "loss": 1.1443333625793457, "step": 46613 }, { "epoch": 3.6747339377217187, "grad_norm": 0.4989486336708069, "learning_rate": 0.0005, "loss": 1.2261356115341187, "step": 46614 }, { "epoch": 3.6748127709893574, "grad_norm": 0.41433531045913696, "learning_rate": 0.0005, "loss": 0.9427897334098816, "step": 46615 }, { "epoch": 3.6748916042569966, "grad_norm": 0.513552188873291, "learning_rate": 0.0005, "loss": 0.9755520820617676, "step": 46616 }, { "epoch": 3.6749704375246353, "grad_norm": 0.4325304329395294, "learning_rate": 0.0005, "loss": 1.2149958610534668, "step": 46617 }, { "epoch": 3.6750492707922744, "grad_norm": 0.5464634895324707, "learning_rate": 0.0005, "loss": 1.1035611629486084, "step": 46618 }, { "epoch": 3.6751281040599135, "grad_norm": 0.5170043110847473, "learning_rate": 0.0005, "loss": 1.2415516376495361, "step": 46619 }, { "epoch": 3.675206937327552, "grad_norm": 0.5711672306060791, "learning_rate": 0.0005, "loss": 0.9376850128173828, "step": 46620 }, { "epoch": 3.675285770595191, "grad_norm": 0.45699217915534973, "learning_rate": 0.0005, "loss": 1.1501338481903076, "step": 46621 }, { "epoch": 3.67536460386283, "grad_norm": 0.5290154218673706, "learning_rate": 0.0005, "loss": 1.2149409055709839, "step": 46622 }, { "epoch": 3.675443437130469, "grad_norm": 0.3984929919242859, "learning_rate": 0.0005, "loss": 0.9494004249572754, "step": 46623 }, { "epoch": 3.675522270398108, "grad_norm": 0.45090630650520325, "learning_rate": 0.0005, "loss": 1.1727999448776245, "step": 46624 }, { "epoch": 3.675601103665747, "grad_norm": 0.5594514012336731, "learning_rate": 0.0005, "loss": 1.2443400621414185, "step": 46625 }, { "epoch": 3.6756799369333857, "grad_norm": 0.5858643651008606, "learning_rate": 0.0005, "loss": 1.243186593055725, "step": 46626 }, { "epoch": 3.675758770201025, "grad_norm": 0.42764419317245483, "learning_rate": 0.0005, "loss": 0.980266809463501, "step": 46627 }, { "epoch": 3.675837603468664, "grad_norm": 0.45620954036712646, "learning_rate": 0.0005, "loss": 1.094808578491211, "step": 46628 }, { "epoch": 3.6759164367363026, "grad_norm": 0.49534744024276733, "learning_rate": 0.0005, "loss": 1.0618339776992798, "step": 46629 }, { "epoch": 3.6759952700039418, "grad_norm": 0.4635579288005829, "learning_rate": 0.0005, "loss": 1.2036125659942627, "step": 46630 }, { "epoch": 3.6760741032715805, "grad_norm": 0.4844565689563751, "learning_rate": 0.0005, "loss": 1.2725613117218018, "step": 46631 }, { "epoch": 3.6761529365392196, "grad_norm": 0.619008481502533, "learning_rate": 0.0005, "loss": 1.1485216617584229, "step": 46632 }, { "epoch": 3.6762317698068587, "grad_norm": 0.4034017026424408, "learning_rate": 0.0005, "loss": 1.0753668546676636, "step": 46633 }, { "epoch": 3.6763106030744974, "grad_norm": 0.5364486575126648, "learning_rate": 0.0005, "loss": 1.0503343343734741, "step": 46634 }, { "epoch": 3.676389436342136, "grad_norm": 0.5905762910842896, "learning_rate": 0.0005, "loss": 1.204364538192749, "step": 46635 }, { "epoch": 3.6764682696097752, "grad_norm": 0.4425184726715088, "learning_rate": 0.0005, "loss": 1.0573744773864746, "step": 46636 }, { "epoch": 3.6765471028774144, "grad_norm": 0.5191890001296997, "learning_rate": 0.0005, "loss": 1.246595025062561, "step": 46637 }, { "epoch": 3.676625936145053, "grad_norm": 0.47102090716362, "learning_rate": 0.0005, "loss": 1.047364354133606, "step": 46638 }, { "epoch": 3.676704769412692, "grad_norm": 0.6531123518943787, "learning_rate": 0.0005, "loss": 1.333262324333191, "step": 46639 }, { "epoch": 3.676783602680331, "grad_norm": 0.46551451086997986, "learning_rate": 0.0005, "loss": 1.188033103942871, "step": 46640 }, { "epoch": 3.67686243594797, "grad_norm": 0.45774170756340027, "learning_rate": 0.0005, "loss": 1.1299793720245361, "step": 46641 }, { "epoch": 3.676941269215609, "grad_norm": 0.5111827254295349, "learning_rate": 0.0005, "loss": 1.0319708585739136, "step": 46642 }, { "epoch": 3.677020102483248, "grad_norm": 0.4195496141910553, "learning_rate": 0.0005, "loss": 0.9974410533905029, "step": 46643 }, { "epoch": 3.677098935750887, "grad_norm": 0.5090094208717346, "learning_rate": 0.0005, "loss": 1.2302594184875488, "step": 46644 }, { "epoch": 3.6771777690185257, "grad_norm": 0.45569926500320435, "learning_rate": 0.0005, "loss": 1.0036590099334717, "step": 46645 }, { "epoch": 3.677256602286165, "grad_norm": 0.45485207438468933, "learning_rate": 0.0005, "loss": 1.1079401969909668, "step": 46646 }, { "epoch": 3.677335435553804, "grad_norm": 0.3986639380455017, "learning_rate": 0.0005, "loss": 0.8724422454833984, "step": 46647 }, { "epoch": 3.6774142688214426, "grad_norm": 0.44299253821372986, "learning_rate": 0.0005, "loss": 1.1054856777191162, "step": 46648 }, { "epoch": 3.6774931020890818, "grad_norm": 0.42207270860671997, "learning_rate": 0.0005, "loss": 1.0956308841705322, "step": 46649 }, { "epoch": 3.6775719353567204, "grad_norm": 0.6094698309898376, "learning_rate": 0.0005, "loss": 1.3970675468444824, "step": 46650 }, { "epoch": 3.6776507686243596, "grad_norm": 0.4279601275920868, "learning_rate": 0.0005, "loss": 1.0314717292785645, "step": 46651 }, { "epoch": 3.6777296018919983, "grad_norm": 0.6000016331672668, "learning_rate": 0.0005, "loss": 1.1581264734268188, "step": 46652 }, { "epoch": 3.6778084351596374, "grad_norm": 0.5034236907958984, "learning_rate": 0.0005, "loss": 1.1901392936706543, "step": 46653 }, { "epoch": 3.677887268427276, "grad_norm": 0.4663011133670807, "learning_rate": 0.0005, "loss": 1.1134381294250488, "step": 46654 }, { "epoch": 3.6779661016949152, "grad_norm": 0.42718562483787537, "learning_rate": 0.0005, "loss": 1.0490729808807373, "step": 46655 }, { "epoch": 3.6780449349625544, "grad_norm": 0.4327833354473114, "learning_rate": 0.0005, "loss": 1.0082011222839355, "step": 46656 }, { "epoch": 3.678123768230193, "grad_norm": 0.4456517994403839, "learning_rate": 0.0005, "loss": 1.0292911529541016, "step": 46657 }, { "epoch": 3.678202601497832, "grad_norm": 0.5212200284004211, "learning_rate": 0.0005, "loss": 1.1110209226608276, "step": 46658 }, { "epoch": 3.678281434765471, "grad_norm": 0.5036336779594421, "learning_rate": 0.0005, "loss": 1.143338680267334, "step": 46659 }, { "epoch": 3.67836026803311, "grad_norm": 0.49424976110458374, "learning_rate": 0.0005, "loss": 1.0654796361923218, "step": 46660 }, { "epoch": 3.678439101300749, "grad_norm": 0.6083377003669739, "learning_rate": 0.0005, "loss": 1.3760547637939453, "step": 46661 }, { "epoch": 3.678517934568388, "grad_norm": 0.4435303509235382, "learning_rate": 0.0005, "loss": 1.0344865322113037, "step": 46662 }, { "epoch": 3.678596767836027, "grad_norm": 0.4699949622154236, "learning_rate": 0.0005, "loss": 1.1305533647537231, "step": 46663 }, { "epoch": 3.6786756011036656, "grad_norm": 0.41012710332870483, "learning_rate": 0.0005, "loss": 0.9899945259094238, "step": 46664 }, { "epoch": 3.678754434371305, "grad_norm": 0.4432474672794342, "learning_rate": 0.0005, "loss": 1.1621819734573364, "step": 46665 }, { "epoch": 3.6788332676389435, "grad_norm": 0.4730214774608612, "learning_rate": 0.0005, "loss": 1.1794064044952393, "step": 46666 }, { "epoch": 3.6789121009065826, "grad_norm": 0.5116764903068542, "learning_rate": 0.0005, "loss": 1.1328215599060059, "step": 46667 }, { "epoch": 3.6789909341742213, "grad_norm": 0.6089425683021545, "learning_rate": 0.0005, "loss": 1.0977076292037964, "step": 46668 }, { "epoch": 3.6790697674418604, "grad_norm": 0.42335474491119385, "learning_rate": 0.0005, "loss": 0.999370813369751, "step": 46669 }, { "epoch": 3.6791486007094996, "grad_norm": 0.5787644386291504, "learning_rate": 0.0005, "loss": 1.1232279539108276, "step": 46670 }, { "epoch": 3.6792274339771383, "grad_norm": 0.5399438142776489, "learning_rate": 0.0005, "loss": 1.4230003356933594, "step": 46671 }, { "epoch": 3.6793062672447774, "grad_norm": 0.47556084394454956, "learning_rate": 0.0005, "loss": 1.2071572542190552, "step": 46672 }, { "epoch": 3.679385100512416, "grad_norm": 0.47874999046325684, "learning_rate": 0.0005, "loss": 1.1729377508163452, "step": 46673 }, { "epoch": 3.679463933780055, "grad_norm": 0.4715729057788849, "learning_rate": 0.0005, "loss": 1.2092373371124268, "step": 46674 }, { "epoch": 3.6795427670476943, "grad_norm": 0.42453017830848694, "learning_rate": 0.0005, "loss": 1.0691558122634888, "step": 46675 }, { "epoch": 3.679621600315333, "grad_norm": 0.49299341440200806, "learning_rate": 0.0005, "loss": 1.1321395635604858, "step": 46676 }, { "epoch": 3.679700433582972, "grad_norm": 0.5762041807174683, "learning_rate": 0.0005, "loss": 1.2469115257263184, "step": 46677 }, { "epoch": 3.679779266850611, "grad_norm": 0.48749810457229614, "learning_rate": 0.0005, "loss": 1.2111982107162476, "step": 46678 }, { "epoch": 3.67985810011825, "grad_norm": 0.3946404755115509, "learning_rate": 0.0005, "loss": 1.0986430644989014, "step": 46679 }, { "epoch": 3.679936933385889, "grad_norm": 0.45416197180747986, "learning_rate": 0.0005, "loss": 1.1182312965393066, "step": 46680 }, { "epoch": 3.680015766653528, "grad_norm": 0.42742919921875, "learning_rate": 0.0005, "loss": 1.075079321861267, "step": 46681 }, { "epoch": 3.6800945999211665, "grad_norm": 0.563583254814148, "learning_rate": 0.0005, "loss": 1.1749110221862793, "step": 46682 }, { "epoch": 3.6801734331888056, "grad_norm": 0.4716395437717438, "learning_rate": 0.0005, "loss": 1.0964183807373047, "step": 46683 }, { "epoch": 3.6802522664564448, "grad_norm": 0.527633011341095, "learning_rate": 0.0005, "loss": 1.138891577720642, "step": 46684 }, { "epoch": 3.6803310997240835, "grad_norm": 0.4620913863182068, "learning_rate": 0.0005, "loss": 1.0182945728302002, "step": 46685 }, { "epoch": 3.6804099329917226, "grad_norm": 0.531395435333252, "learning_rate": 0.0005, "loss": 1.1516509056091309, "step": 46686 }, { "epoch": 3.6804887662593613, "grad_norm": 0.4964684247970581, "learning_rate": 0.0005, "loss": 1.0486143827438354, "step": 46687 }, { "epoch": 3.6805675995270004, "grad_norm": 0.4932762086391449, "learning_rate": 0.0005, "loss": 1.266387939453125, "step": 46688 }, { "epoch": 3.6806464327946395, "grad_norm": 0.39219579100608826, "learning_rate": 0.0005, "loss": 0.9284707307815552, "step": 46689 }, { "epoch": 3.6807252660622782, "grad_norm": 0.4568368196487427, "learning_rate": 0.0005, "loss": 1.188800573348999, "step": 46690 }, { "epoch": 3.6808040993299174, "grad_norm": 0.5721930861473083, "learning_rate": 0.0005, "loss": 1.369715929031372, "step": 46691 }, { "epoch": 3.680882932597556, "grad_norm": 0.47035685181617737, "learning_rate": 0.0005, "loss": 1.395477533340454, "step": 46692 }, { "epoch": 3.680961765865195, "grad_norm": 0.5351433157920837, "learning_rate": 0.0005, "loss": 1.2147374153137207, "step": 46693 }, { "epoch": 3.6810405991328343, "grad_norm": 0.4601050913333893, "learning_rate": 0.0005, "loss": 1.114124059677124, "step": 46694 }, { "epoch": 3.681119432400473, "grad_norm": 0.5302593111991882, "learning_rate": 0.0005, "loss": 1.2300193309783936, "step": 46695 }, { "epoch": 3.6811982656681117, "grad_norm": 0.7215975522994995, "learning_rate": 0.0005, "loss": 1.4612061977386475, "step": 46696 }, { "epoch": 3.681277098935751, "grad_norm": 0.5393166542053223, "learning_rate": 0.0005, "loss": 1.218990445137024, "step": 46697 }, { "epoch": 3.68135593220339, "grad_norm": 0.4830215275287628, "learning_rate": 0.0005, "loss": 1.1118487119674683, "step": 46698 }, { "epoch": 3.6814347654710287, "grad_norm": 0.43839892745018005, "learning_rate": 0.0005, "loss": 1.1158406734466553, "step": 46699 }, { "epoch": 3.681513598738668, "grad_norm": 0.47836241126060486, "learning_rate": 0.0005, "loss": 1.2106266021728516, "step": 46700 }, { "epoch": 3.6815924320063065, "grad_norm": 0.6425345540046692, "learning_rate": 0.0005, "loss": 1.256864070892334, "step": 46701 }, { "epoch": 3.6816712652739456, "grad_norm": 0.46545273065567017, "learning_rate": 0.0005, "loss": 1.0844924449920654, "step": 46702 }, { "epoch": 3.6817500985415847, "grad_norm": 0.4633811116218567, "learning_rate": 0.0005, "loss": 1.080335021018982, "step": 46703 }, { "epoch": 3.6818289318092234, "grad_norm": 0.5706621408462524, "learning_rate": 0.0005, "loss": 1.2453150749206543, "step": 46704 }, { "epoch": 3.6819077650768626, "grad_norm": 0.5869265198707581, "learning_rate": 0.0005, "loss": 1.4105037450790405, "step": 46705 }, { "epoch": 3.6819865983445013, "grad_norm": 0.3877929747104645, "learning_rate": 0.0005, "loss": 1.0211178064346313, "step": 46706 }, { "epoch": 3.6820654316121404, "grad_norm": 0.4867314398288727, "learning_rate": 0.0005, "loss": 1.3527055978775024, "step": 46707 }, { "epoch": 3.6821442648797795, "grad_norm": 0.47013184428215027, "learning_rate": 0.0005, "loss": 1.090567946434021, "step": 46708 }, { "epoch": 3.682223098147418, "grad_norm": 0.6085910201072693, "learning_rate": 0.0005, "loss": 1.1474730968475342, "step": 46709 }, { "epoch": 3.682301931415057, "grad_norm": 0.4364602565765381, "learning_rate": 0.0005, "loss": 0.9883742928504944, "step": 46710 }, { "epoch": 3.682380764682696, "grad_norm": 0.4784218370914459, "learning_rate": 0.0005, "loss": 1.2125170230865479, "step": 46711 }, { "epoch": 3.682459597950335, "grad_norm": 0.5263304710388184, "learning_rate": 0.0005, "loss": 1.2781493663787842, "step": 46712 }, { "epoch": 3.682538431217974, "grad_norm": 0.5148542523384094, "learning_rate": 0.0005, "loss": 1.057806134223938, "step": 46713 }, { "epoch": 3.682617264485613, "grad_norm": 0.6028335094451904, "learning_rate": 0.0005, "loss": 1.1882307529449463, "step": 46714 }, { "epoch": 3.6826960977532517, "grad_norm": 0.448434442281723, "learning_rate": 0.0005, "loss": 1.1868171691894531, "step": 46715 }, { "epoch": 3.682774931020891, "grad_norm": 0.4109284281730652, "learning_rate": 0.0005, "loss": 1.0893487930297852, "step": 46716 }, { "epoch": 3.68285376428853, "grad_norm": 0.5597546100616455, "learning_rate": 0.0005, "loss": 1.1790419816970825, "step": 46717 }, { "epoch": 3.6829325975561686, "grad_norm": 0.6349406242370605, "learning_rate": 0.0005, "loss": 1.1802259683609009, "step": 46718 }, { "epoch": 3.6830114308238078, "grad_norm": 0.520950973033905, "learning_rate": 0.0005, "loss": 1.1501227617263794, "step": 46719 }, { "epoch": 3.6830902640914465, "grad_norm": 0.4466226398944855, "learning_rate": 0.0005, "loss": 1.2322462797164917, "step": 46720 }, { "epoch": 3.6831690973590856, "grad_norm": 0.44545891880989075, "learning_rate": 0.0005, "loss": 1.1571946144104004, "step": 46721 }, { "epoch": 3.6832479306267247, "grad_norm": 0.46875452995300293, "learning_rate": 0.0005, "loss": 1.1278190612792969, "step": 46722 }, { "epoch": 3.6833267638943634, "grad_norm": 0.3722717761993408, "learning_rate": 0.0005, "loss": 0.8703758120536804, "step": 46723 }, { "epoch": 3.683405597162002, "grad_norm": 0.5254897475242615, "learning_rate": 0.0005, "loss": 1.1282942295074463, "step": 46724 }, { "epoch": 3.6834844304296412, "grad_norm": 0.5291408896446228, "learning_rate": 0.0005, "loss": 1.2314302921295166, "step": 46725 }, { "epoch": 3.6835632636972804, "grad_norm": 0.5530027747154236, "learning_rate": 0.0005, "loss": 1.1431317329406738, "step": 46726 }, { "epoch": 3.683642096964919, "grad_norm": 0.4055531919002533, "learning_rate": 0.0005, "loss": 1.0586540699005127, "step": 46727 }, { "epoch": 3.683720930232558, "grad_norm": 0.6637369990348816, "learning_rate": 0.0005, "loss": 1.1674405336380005, "step": 46728 }, { "epoch": 3.683799763500197, "grad_norm": 0.4451940953731537, "learning_rate": 0.0005, "loss": 1.0632855892181396, "step": 46729 }, { "epoch": 3.683878596767836, "grad_norm": 0.501075804233551, "learning_rate": 0.0005, "loss": 1.2273471355438232, "step": 46730 }, { "epoch": 3.683957430035475, "grad_norm": 0.4900003969669342, "learning_rate": 0.0005, "loss": 1.1560724973678589, "step": 46731 }, { "epoch": 3.684036263303114, "grad_norm": 0.7885997891426086, "learning_rate": 0.0005, "loss": 1.2095818519592285, "step": 46732 }, { "epoch": 3.684115096570753, "grad_norm": 0.49145007133483887, "learning_rate": 0.0005, "loss": 1.2669073343276978, "step": 46733 }, { "epoch": 3.6841939298383917, "grad_norm": 0.5758306980133057, "learning_rate": 0.0005, "loss": 0.992806613445282, "step": 46734 }, { "epoch": 3.684272763106031, "grad_norm": 0.4758110046386719, "learning_rate": 0.0005, "loss": 1.2864314317703247, "step": 46735 }, { "epoch": 3.68435159637367, "grad_norm": 0.6039114594459534, "learning_rate": 0.0005, "loss": 1.1800613403320312, "step": 46736 }, { "epoch": 3.6844304296413086, "grad_norm": 0.45394572615623474, "learning_rate": 0.0005, "loss": 1.198014497756958, "step": 46737 }, { "epoch": 3.6845092629089473, "grad_norm": 0.47989410161972046, "learning_rate": 0.0005, "loss": 1.2834196090698242, "step": 46738 }, { "epoch": 3.6845880961765864, "grad_norm": 0.501542866230011, "learning_rate": 0.0005, "loss": 1.121055006980896, "step": 46739 }, { "epoch": 3.6846669294442256, "grad_norm": 0.45550617575645447, "learning_rate": 0.0005, "loss": 1.2075029611587524, "step": 46740 }, { "epoch": 3.6847457627118643, "grad_norm": 0.5066098570823669, "learning_rate": 0.0005, "loss": 1.2727859020233154, "step": 46741 }, { "epoch": 3.6848245959795034, "grad_norm": 0.528694212436676, "learning_rate": 0.0005, "loss": 1.1596777439117432, "step": 46742 }, { "epoch": 3.684903429247142, "grad_norm": 0.5404538512229919, "learning_rate": 0.0005, "loss": 1.110191822052002, "step": 46743 }, { "epoch": 3.6849822625147812, "grad_norm": 0.4694066047668457, "learning_rate": 0.0005, "loss": 1.084141731262207, "step": 46744 }, { "epoch": 3.6850610957824204, "grad_norm": 0.3891008794307709, "learning_rate": 0.0005, "loss": 1.1041004657745361, "step": 46745 }, { "epoch": 3.685139929050059, "grad_norm": 0.5058455467224121, "learning_rate": 0.0005, "loss": 1.1160171031951904, "step": 46746 }, { "epoch": 3.685218762317698, "grad_norm": 0.4633110761642456, "learning_rate": 0.0005, "loss": 1.2460575103759766, "step": 46747 }, { "epoch": 3.685297595585337, "grad_norm": 0.7111176252365112, "learning_rate": 0.0005, "loss": 1.2726075649261475, "step": 46748 }, { "epoch": 3.685376428852976, "grad_norm": 0.4582192003726959, "learning_rate": 0.0005, "loss": 1.1781737804412842, "step": 46749 }, { "epoch": 3.685455262120615, "grad_norm": 0.5052165985107422, "learning_rate": 0.0005, "loss": 1.2802577018737793, "step": 46750 }, { "epoch": 3.685534095388254, "grad_norm": 0.41507431864738464, "learning_rate": 0.0005, "loss": 1.1270122528076172, "step": 46751 }, { "epoch": 3.6856129286558925, "grad_norm": 0.3582753837108612, "learning_rate": 0.0005, "loss": 0.9638671875, "step": 46752 }, { "epoch": 3.6856917619235317, "grad_norm": 0.40087345242500305, "learning_rate": 0.0005, "loss": 1.043163776397705, "step": 46753 }, { "epoch": 3.685770595191171, "grad_norm": 0.5089718699455261, "learning_rate": 0.0005, "loss": 1.0845346450805664, "step": 46754 }, { "epoch": 3.6858494284588095, "grad_norm": 0.41591665148735046, "learning_rate": 0.0005, "loss": 1.1390495300292969, "step": 46755 }, { "epoch": 3.6859282617264486, "grad_norm": 0.6322845816612244, "learning_rate": 0.0005, "loss": 1.1245752573013306, "step": 46756 }, { "epoch": 3.6860070949940873, "grad_norm": 0.5005258917808533, "learning_rate": 0.0005, "loss": 1.1474645137786865, "step": 46757 }, { "epoch": 3.6860859282617264, "grad_norm": 0.39125996828079224, "learning_rate": 0.0005, "loss": 1.1042709350585938, "step": 46758 }, { "epoch": 3.6861647615293656, "grad_norm": 0.4311128854751587, "learning_rate": 0.0005, "loss": 1.0146667957305908, "step": 46759 }, { "epoch": 3.6862435947970043, "grad_norm": 0.43064969778060913, "learning_rate": 0.0005, "loss": 0.9052634239196777, "step": 46760 }, { "epoch": 3.6863224280646434, "grad_norm": 0.5111259818077087, "learning_rate": 0.0005, "loss": 1.3017868995666504, "step": 46761 }, { "epoch": 3.686401261332282, "grad_norm": 0.5689262747764587, "learning_rate": 0.0005, "loss": 1.1577777862548828, "step": 46762 }, { "epoch": 3.686480094599921, "grad_norm": 0.6091363430023193, "learning_rate": 0.0005, "loss": 1.4373891353607178, "step": 46763 }, { "epoch": 3.6865589278675603, "grad_norm": 0.4839254915714264, "learning_rate": 0.0005, "loss": 1.1674070358276367, "step": 46764 }, { "epoch": 3.686637761135199, "grad_norm": 0.463064581155777, "learning_rate": 0.0005, "loss": 1.0682770013809204, "step": 46765 }, { "epoch": 3.6867165944028377, "grad_norm": 0.45468229055404663, "learning_rate": 0.0005, "loss": 0.9542669057846069, "step": 46766 }, { "epoch": 3.686795427670477, "grad_norm": 0.4445118308067322, "learning_rate": 0.0005, "loss": 1.135019302368164, "step": 46767 }, { "epoch": 3.686874260938116, "grad_norm": 0.42515742778778076, "learning_rate": 0.0005, "loss": 1.1894959211349487, "step": 46768 }, { "epoch": 3.6869530942057547, "grad_norm": 0.5216279625892639, "learning_rate": 0.0005, "loss": 1.3355249166488647, "step": 46769 }, { "epoch": 3.687031927473394, "grad_norm": 0.5007055997848511, "learning_rate": 0.0005, "loss": 1.2200549840927124, "step": 46770 }, { "epoch": 3.6871107607410325, "grad_norm": 0.36549434065818787, "learning_rate": 0.0005, "loss": 0.9330268502235413, "step": 46771 }, { "epoch": 3.6871895940086716, "grad_norm": 0.4568188488483429, "learning_rate": 0.0005, "loss": 1.2940309047698975, "step": 46772 }, { "epoch": 3.6872684272763108, "grad_norm": 0.40764686465263367, "learning_rate": 0.0005, "loss": 0.9871317744255066, "step": 46773 }, { "epoch": 3.6873472605439495, "grad_norm": 0.5480743646621704, "learning_rate": 0.0005, "loss": 1.2877707481384277, "step": 46774 }, { "epoch": 3.6874260938115886, "grad_norm": 0.588726282119751, "learning_rate": 0.0005, "loss": 1.3749052286148071, "step": 46775 }, { "epoch": 3.6875049270792273, "grad_norm": 0.42164987325668335, "learning_rate": 0.0005, "loss": 1.0579026937484741, "step": 46776 }, { "epoch": 3.6875837603468664, "grad_norm": 0.5983657240867615, "learning_rate": 0.0005, "loss": 1.1779676675796509, "step": 46777 }, { "epoch": 3.6876625936145055, "grad_norm": 0.5702983736991882, "learning_rate": 0.0005, "loss": 1.1675810813903809, "step": 46778 }, { "epoch": 3.6877414268821442, "grad_norm": 0.4966394603252411, "learning_rate": 0.0005, "loss": 1.1634438037872314, "step": 46779 }, { "epoch": 3.6878202601497834, "grad_norm": 0.5116422176361084, "learning_rate": 0.0005, "loss": 1.1029176712036133, "step": 46780 }, { "epoch": 3.687899093417422, "grad_norm": 0.47507715225219727, "learning_rate": 0.0005, "loss": 0.9433661699295044, "step": 46781 }, { "epoch": 3.687977926685061, "grad_norm": 0.3825339674949646, "learning_rate": 0.0005, "loss": 1.020024061203003, "step": 46782 }, { "epoch": 3.6880567599527, "grad_norm": 0.5449747443199158, "learning_rate": 0.0005, "loss": 1.0278416872024536, "step": 46783 }, { "epoch": 3.688135593220339, "grad_norm": 0.5055089592933655, "learning_rate": 0.0005, "loss": 1.2293150424957275, "step": 46784 }, { "epoch": 3.6882144264879777, "grad_norm": 0.5734313130378723, "learning_rate": 0.0005, "loss": 1.0408121347427368, "step": 46785 }, { "epoch": 3.688293259755617, "grad_norm": 0.5904906392097473, "learning_rate": 0.0005, "loss": 1.3213465213775635, "step": 46786 }, { "epoch": 3.688372093023256, "grad_norm": 0.6207289695739746, "learning_rate": 0.0005, "loss": 1.2181599140167236, "step": 46787 }, { "epoch": 3.6884509262908947, "grad_norm": 0.6206756830215454, "learning_rate": 0.0005, "loss": 1.2004754543304443, "step": 46788 }, { "epoch": 3.688529759558534, "grad_norm": 0.5411120057106018, "learning_rate": 0.0005, "loss": 1.113102674484253, "step": 46789 }, { "epoch": 3.6886085928261725, "grad_norm": 0.40748924016952515, "learning_rate": 0.0005, "loss": 1.1958318948745728, "step": 46790 }, { "epoch": 3.6886874260938116, "grad_norm": 0.5163094997406006, "learning_rate": 0.0005, "loss": 1.1926896572113037, "step": 46791 }, { "epoch": 3.6887662593614508, "grad_norm": 0.4921269416809082, "learning_rate": 0.0005, "loss": 1.2325103282928467, "step": 46792 }, { "epoch": 3.6888450926290894, "grad_norm": 0.44322600960731506, "learning_rate": 0.0005, "loss": 1.0933837890625, "step": 46793 }, { "epoch": 3.6889239258967286, "grad_norm": 0.5158585906028748, "learning_rate": 0.0005, "loss": 1.269463062286377, "step": 46794 }, { "epoch": 3.6890027591643673, "grad_norm": 0.4723769426345825, "learning_rate": 0.0005, "loss": 1.2221496105194092, "step": 46795 }, { "epoch": 3.6890815924320064, "grad_norm": 0.48661795258522034, "learning_rate": 0.0005, "loss": 1.078192949295044, "step": 46796 }, { "epoch": 3.689160425699645, "grad_norm": 0.461232453584671, "learning_rate": 0.0005, "loss": 1.234543800354004, "step": 46797 }, { "epoch": 3.689239258967284, "grad_norm": 0.43482401967048645, "learning_rate": 0.0005, "loss": 1.0298123359680176, "step": 46798 }, { "epoch": 3.689318092234923, "grad_norm": 0.6148137450218201, "learning_rate": 0.0005, "loss": 1.1311981678009033, "step": 46799 }, { "epoch": 3.689396925502562, "grad_norm": 0.4736555218696594, "learning_rate": 0.0005, "loss": 1.1859523057937622, "step": 46800 }, { "epoch": 3.689475758770201, "grad_norm": 0.42503970861434937, "learning_rate": 0.0005, "loss": 1.1737104654312134, "step": 46801 }, { "epoch": 3.68955459203784, "grad_norm": 0.46813467144966125, "learning_rate": 0.0005, "loss": 1.0845084190368652, "step": 46802 }, { "epoch": 3.689633425305479, "grad_norm": 0.6358416676521301, "learning_rate": 0.0005, "loss": 1.4889509677886963, "step": 46803 }, { "epoch": 3.6897122585731177, "grad_norm": 0.4497186839580536, "learning_rate": 0.0005, "loss": 1.1930326223373413, "step": 46804 }, { "epoch": 3.689791091840757, "grad_norm": 0.49204713106155396, "learning_rate": 0.0005, "loss": 1.271732211112976, "step": 46805 }, { "epoch": 3.689869925108396, "grad_norm": 0.5432450175285339, "learning_rate": 0.0005, "loss": 1.2304613590240479, "step": 46806 }, { "epoch": 3.6899487583760346, "grad_norm": 0.7685779333114624, "learning_rate": 0.0005, "loss": 1.2501246929168701, "step": 46807 }, { "epoch": 3.6900275916436738, "grad_norm": 0.502900242805481, "learning_rate": 0.0005, "loss": 1.3904588222503662, "step": 46808 }, { "epoch": 3.6901064249113125, "grad_norm": 0.49166032671928406, "learning_rate": 0.0005, "loss": 1.0753902196884155, "step": 46809 }, { "epoch": 3.6901852581789516, "grad_norm": 0.48640307784080505, "learning_rate": 0.0005, "loss": 1.1360552310943604, "step": 46810 }, { "epoch": 3.6902640914465907, "grad_norm": 0.5443056225776672, "learning_rate": 0.0005, "loss": 1.0616378784179688, "step": 46811 }, { "epoch": 3.6903429247142294, "grad_norm": 0.5225563645362854, "learning_rate": 0.0005, "loss": 1.1752972602844238, "step": 46812 }, { "epoch": 3.690421757981868, "grad_norm": 0.4184797704219818, "learning_rate": 0.0005, "loss": 1.1282061338424683, "step": 46813 }, { "epoch": 3.6905005912495072, "grad_norm": 0.4811045229434967, "learning_rate": 0.0005, "loss": 1.1900391578674316, "step": 46814 }, { "epoch": 3.6905794245171464, "grad_norm": 0.5238629579544067, "learning_rate": 0.0005, "loss": 1.149918556213379, "step": 46815 }, { "epoch": 3.690658257784785, "grad_norm": 0.43746933341026306, "learning_rate": 0.0005, "loss": 1.0954818725585938, "step": 46816 }, { "epoch": 3.690737091052424, "grad_norm": 0.46884483098983765, "learning_rate": 0.0005, "loss": 1.2069897651672363, "step": 46817 }, { "epoch": 3.690815924320063, "grad_norm": 0.4575648605823517, "learning_rate": 0.0005, "loss": 1.2045083045959473, "step": 46818 }, { "epoch": 3.690894757587702, "grad_norm": 0.4435890316963196, "learning_rate": 0.0005, "loss": 1.0917344093322754, "step": 46819 }, { "epoch": 3.690973590855341, "grad_norm": 0.529194176197052, "learning_rate": 0.0005, "loss": 1.0578923225402832, "step": 46820 }, { "epoch": 3.69105242412298, "grad_norm": 0.42216557264328003, "learning_rate": 0.0005, "loss": 1.180466890335083, "step": 46821 }, { "epoch": 3.691131257390619, "grad_norm": 0.4700736403465271, "learning_rate": 0.0005, "loss": 1.0895986557006836, "step": 46822 }, { "epoch": 3.6912100906582577, "grad_norm": 0.5201564431190491, "learning_rate": 0.0005, "loss": 1.3133399486541748, "step": 46823 }, { "epoch": 3.691288923925897, "grad_norm": 0.46339064836502075, "learning_rate": 0.0005, "loss": 1.100653886795044, "step": 46824 }, { "epoch": 3.691367757193536, "grad_norm": 0.5147485136985779, "learning_rate": 0.0005, "loss": 1.1180262565612793, "step": 46825 }, { "epoch": 3.6914465904611746, "grad_norm": 0.46190130710601807, "learning_rate": 0.0005, "loss": 1.1009776592254639, "step": 46826 }, { "epoch": 3.6915254237288133, "grad_norm": 0.655579149723053, "learning_rate": 0.0005, "loss": 1.4283287525177002, "step": 46827 }, { "epoch": 3.6916042569964524, "grad_norm": 0.6171278357505798, "learning_rate": 0.0005, "loss": 1.4736144542694092, "step": 46828 }, { "epoch": 3.6916830902640916, "grad_norm": 0.48981568217277527, "learning_rate": 0.0005, "loss": 1.2442289590835571, "step": 46829 }, { "epoch": 3.6917619235317303, "grad_norm": 0.5039255023002625, "learning_rate": 0.0005, "loss": 1.234696865081787, "step": 46830 }, { "epoch": 3.6918407567993694, "grad_norm": 0.4446125626564026, "learning_rate": 0.0005, "loss": 1.0844000577926636, "step": 46831 }, { "epoch": 3.691919590067008, "grad_norm": 0.6050732731819153, "learning_rate": 0.0005, "loss": 1.2890357971191406, "step": 46832 }, { "epoch": 3.6919984233346472, "grad_norm": 0.4869295656681061, "learning_rate": 0.0005, "loss": 1.37676203250885, "step": 46833 }, { "epoch": 3.6920772566022864, "grad_norm": 0.46659019589424133, "learning_rate": 0.0005, "loss": 1.1684424877166748, "step": 46834 }, { "epoch": 3.692156089869925, "grad_norm": 0.5009894967079163, "learning_rate": 0.0005, "loss": 1.1744203567504883, "step": 46835 }, { "epoch": 3.692234923137564, "grad_norm": 0.45070114731788635, "learning_rate": 0.0005, "loss": 1.0529577732086182, "step": 46836 }, { "epoch": 3.692313756405203, "grad_norm": 0.44105905294418335, "learning_rate": 0.0005, "loss": 0.9929997324943542, "step": 46837 }, { "epoch": 3.692392589672842, "grad_norm": 0.47939035296440125, "learning_rate": 0.0005, "loss": 1.0229746103286743, "step": 46838 }, { "epoch": 3.692471422940481, "grad_norm": 0.4723515510559082, "learning_rate": 0.0005, "loss": 1.1395251750946045, "step": 46839 }, { "epoch": 3.69255025620812, "grad_norm": 0.45916107296943665, "learning_rate": 0.0005, "loss": 0.9890530109405518, "step": 46840 }, { "epoch": 3.6926290894757585, "grad_norm": 0.5443209409713745, "learning_rate": 0.0005, "loss": 1.3665235042572021, "step": 46841 }, { "epoch": 3.6927079227433977, "grad_norm": 0.5215277671813965, "learning_rate": 0.0005, "loss": 1.2053508758544922, "step": 46842 }, { "epoch": 3.692786756011037, "grad_norm": 0.45255497097969055, "learning_rate": 0.0005, "loss": 1.1456599235534668, "step": 46843 }, { "epoch": 3.6928655892786755, "grad_norm": 0.5102820992469788, "learning_rate": 0.0005, "loss": 1.3324717283248901, "step": 46844 }, { "epoch": 3.6929444225463146, "grad_norm": 0.380426287651062, "learning_rate": 0.0005, "loss": 1.3205360174179077, "step": 46845 }, { "epoch": 3.6930232558139533, "grad_norm": 0.4859904944896698, "learning_rate": 0.0005, "loss": 1.0020339488983154, "step": 46846 }, { "epoch": 3.6931020890815924, "grad_norm": 0.5066293478012085, "learning_rate": 0.0005, "loss": 1.1348745822906494, "step": 46847 }, { "epoch": 3.6931809223492316, "grad_norm": 0.6370750665664673, "learning_rate": 0.0005, "loss": 1.2958002090454102, "step": 46848 }, { "epoch": 3.6932597556168703, "grad_norm": 0.45609137415885925, "learning_rate": 0.0005, "loss": 1.1825008392333984, "step": 46849 }, { "epoch": 3.6933385888845094, "grad_norm": 0.5101833939552307, "learning_rate": 0.0005, "loss": 1.1455292701721191, "step": 46850 }, { "epoch": 3.693417422152148, "grad_norm": 0.5378990769386292, "learning_rate": 0.0005, "loss": 1.0995182991027832, "step": 46851 }, { "epoch": 3.693496255419787, "grad_norm": 0.6095654368400574, "learning_rate": 0.0005, "loss": 1.3616983890533447, "step": 46852 }, { "epoch": 3.6935750886874263, "grad_norm": 0.5684174299240112, "learning_rate": 0.0005, "loss": 1.257641315460205, "step": 46853 }, { "epoch": 3.693653921955065, "grad_norm": 0.5219075083732605, "learning_rate": 0.0005, "loss": 1.2412919998168945, "step": 46854 }, { "epoch": 3.6937327552227037, "grad_norm": 0.46916255354881287, "learning_rate": 0.0005, "loss": 1.2505974769592285, "step": 46855 }, { "epoch": 3.693811588490343, "grad_norm": 0.46243220567703247, "learning_rate": 0.0005, "loss": 1.1501439809799194, "step": 46856 }, { "epoch": 3.693890421757982, "grad_norm": 0.4750252366065979, "learning_rate": 0.0005, "loss": 1.3418937921524048, "step": 46857 }, { "epoch": 3.6939692550256207, "grad_norm": 0.43902450799942017, "learning_rate": 0.0005, "loss": 1.1179375648498535, "step": 46858 }, { "epoch": 3.69404808829326, "grad_norm": 0.45844393968582153, "learning_rate": 0.0005, "loss": 1.2177488803863525, "step": 46859 }, { "epoch": 3.6941269215608985, "grad_norm": 0.6969371438026428, "learning_rate": 0.0005, "loss": 1.0374934673309326, "step": 46860 }, { "epoch": 3.6942057548285376, "grad_norm": 0.5101849436759949, "learning_rate": 0.0005, "loss": 1.166253924369812, "step": 46861 }, { "epoch": 3.6942845880961768, "grad_norm": 0.44739750027656555, "learning_rate": 0.0005, "loss": 1.2306137084960938, "step": 46862 }, { "epoch": 3.6943634213638155, "grad_norm": 0.4886556565761566, "learning_rate": 0.0005, "loss": 1.25152587890625, "step": 46863 }, { "epoch": 3.6944422546314546, "grad_norm": 0.4313182830810547, "learning_rate": 0.0005, "loss": 1.146972894668579, "step": 46864 }, { "epoch": 3.6945210878990933, "grad_norm": 0.5926545858383179, "learning_rate": 0.0005, "loss": 1.229952096939087, "step": 46865 }, { "epoch": 3.6945999211667324, "grad_norm": 0.6015678644180298, "learning_rate": 0.0005, "loss": 1.2135636806488037, "step": 46866 }, { "epoch": 3.6946787544343715, "grad_norm": 0.5468148589134216, "learning_rate": 0.0005, "loss": 1.1581865549087524, "step": 46867 }, { "epoch": 3.6947575877020102, "grad_norm": 0.6249803304672241, "learning_rate": 0.0005, "loss": 1.2364104986190796, "step": 46868 }, { "epoch": 3.694836420969649, "grad_norm": 0.49980148673057556, "learning_rate": 0.0005, "loss": 1.0578906536102295, "step": 46869 }, { "epoch": 3.694915254237288, "grad_norm": 0.4366393983364105, "learning_rate": 0.0005, "loss": 1.1723096370697021, "step": 46870 }, { "epoch": 3.694994087504927, "grad_norm": 0.4130253195762634, "learning_rate": 0.0005, "loss": 1.0108966827392578, "step": 46871 }, { "epoch": 3.695072920772566, "grad_norm": 0.46540889143943787, "learning_rate": 0.0005, "loss": 1.2358075380325317, "step": 46872 }, { "epoch": 3.695151754040205, "grad_norm": 0.5522025227546692, "learning_rate": 0.0005, "loss": 1.0732614994049072, "step": 46873 }, { "epoch": 3.6952305873078437, "grad_norm": 0.5496219992637634, "learning_rate": 0.0005, "loss": 1.2082257270812988, "step": 46874 }, { "epoch": 3.695309420575483, "grad_norm": 0.49426016211509705, "learning_rate": 0.0005, "loss": 1.1432936191558838, "step": 46875 }, { "epoch": 3.695388253843122, "grad_norm": 0.510155439376831, "learning_rate": 0.0005, "loss": 1.0774434804916382, "step": 46876 }, { "epoch": 3.6954670871107607, "grad_norm": 0.7111227512359619, "learning_rate": 0.0005, "loss": 1.138943076133728, "step": 46877 }, { "epoch": 3.6955459203784, "grad_norm": 0.4537406265735626, "learning_rate": 0.0005, "loss": 1.1604654788970947, "step": 46878 }, { "epoch": 3.6956247536460385, "grad_norm": 0.6542945504188538, "learning_rate": 0.0005, "loss": 1.2141482830047607, "step": 46879 }, { "epoch": 3.6957035869136776, "grad_norm": 0.5000867247581482, "learning_rate": 0.0005, "loss": 1.2266085147857666, "step": 46880 }, { "epoch": 3.6957824201813168, "grad_norm": 0.42947399616241455, "learning_rate": 0.0005, "loss": 1.2499895095825195, "step": 46881 }, { "epoch": 3.6958612534489554, "grad_norm": 0.5171281695365906, "learning_rate": 0.0005, "loss": 1.1016570329666138, "step": 46882 }, { "epoch": 3.695940086716594, "grad_norm": 0.8703033328056335, "learning_rate": 0.0005, "loss": 1.329904556274414, "step": 46883 }, { "epoch": 3.6960189199842333, "grad_norm": 0.6067387461662292, "learning_rate": 0.0005, "loss": 1.1712861061096191, "step": 46884 }, { "epoch": 3.6960977532518724, "grad_norm": 0.4618043601512909, "learning_rate": 0.0005, "loss": 1.2117178440093994, "step": 46885 }, { "epoch": 3.696176586519511, "grad_norm": 0.5051960349082947, "learning_rate": 0.0005, "loss": 1.2316534519195557, "step": 46886 }, { "epoch": 3.69625541978715, "grad_norm": 0.5171363353729248, "learning_rate": 0.0005, "loss": 1.1285521984100342, "step": 46887 }, { "epoch": 3.696334253054789, "grad_norm": 0.5342144966125488, "learning_rate": 0.0005, "loss": 1.1747714281082153, "step": 46888 }, { "epoch": 3.696413086322428, "grad_norm": 0.413585901260376, "learning_rate": 0.0005, "loss": 1.022792935371399, "step": 46889 }, { "epoch": 3.696491919590067, "grad_norm": 0.43883243203163147, "learning_rate": 0.0005, "loss": 1.1420804262161255, "step": 46890 }, { "epoch": 3.696570752857706, "grad_norm": 0.46107426285743713, "learning_rate": 0.0005, "loss": 1.192105770111084, "step": 46891 }, { "epoch": 3.696649586125345, "grad_norm": 0.6029345393180847, "learning_rate": 0.0005, "loss": 1.2596192359924316, "step": 46892 }, { "epoch": 3.6967284193929837, "grad_norm": 0.357073575258255, "learning_rate": 0.0005, "loss": 0.9643843173980713, "step": 46893 }, { "epoch": 3.696807252660623, "grad_norm": 0.42385420203208923, "learning_rate": 0.0005, "loss": 1.147864818572998, "step": 46894 }, { "epoch": 3.696886085928262, "grad_norm": 0.6121686697006226, "learning_rate": 0.0005, "loss": 1.204426884651184, "step": 46895 }, { "epoch": 3.6969649191959006, "grad_norm": 0.4322160482406616, "learning_rate": 0.0005, "loss": 1.0552939176559448, "step": 46896 }, { "epoch": 3.69704375246354, "grad_norm": 0.5859193205833435, "learning_rate": 0.0005, "loss": 1.1763044595718384, "step": 46897 }, { "epoch": 3.6971225857311785, "grad_norm": 0.5469863414764404, "learning_rate": 0.0005, "loss": 1.1724052429199219, "step": 46898 }, { "epoch": 3.6972014189988176, "grad_norm": 0.49974533915519714, "learning_rate": 0.0005, "loss": 1.1088240146636963, "step": 46899 }, { "epoch": 3.6972802522664563, "grad_norm": 0.5169763565063477, "learning_rate": 0.0005, "loss": 1.305012583732605, "step": 46900 }, { "epoch": 3.6973590855340954, "grad_norm": 0.4401402771472931, "learning_rate": 0.0005, "loss": 1.1813174486160278, "step": 46901 }, { "epoch": 3.697437918801734, "grad_norm": 0.6258499622344971, "learning_rate": 0.0005, "loss": 1.2600376605987549, "step": 46902 }, { "epoch": 3.6975167520693732, "grad_norm": 0.5993801951408386, "learning_rate": 0.0005, "loss": 1.0481688976287842, "step": 46903 }, { "epoch": 3.6975955853370124, "grad_norm": 0.48064538836479187, "learning_rate": 0.0005, "loss": 1.1253163814544678, "step": 46904 }, { "epoch": 3.697674418604651, "grad_norm": 0.42195942997932434, "learning_rate": 0.0005, "loss": 0.9943241477012634, "step": 46905 }, { "epoch": 3.69775325187229, "grad_norm": 0.449874609708786, "learning_rate": 0.0005, "loss": 1.1660001277923584, "step": 46906 }, { "epoch": 3.697832085139929, "grad_norm": 0.48760277032852173, "learning_rate": 0.0005, "loss": 1.0781313180923462, "step": 46907 }, { "epoch": 3.697910918407568, "grad_norm": 0.4769480228424072, "learning_rate": 0.0005, "loss": 1.247257947921753, "step": 46908 }, { "epoch": 3.697989751675207, "grad_norm": 0.3744013011455536, "learning_rate": 0.0005, "loss": 1.020205020904541, "step": 46909 }, { "epoch": 3.698068584942846, "grad_norm": 0.5428759455680847, "learning_rate": 0.0005, "loss": 1.0506736040115356, "step": 46910 }, { "epoch": 3.698147418210485, "grad_norm": 0.44251546263694763, "learning_rate": 0.0005, "loss": 1.185344934463501, "step": 46911 }, { "epoch": 3.6982262514781237, "grad_norm": 0.49325501918792725, "learning_rate": 0.0005, "loss": 1.1777070760726929, "step": 46912 }, { "epoch": 3.698305084745763, "grad_norm": 0.48356977105140686, "learning_rate": 0.0005, "loss": 1.0789101123809814, "step": 46913 }, { "epoch": 3.6983839180134015, "grad_norm": 0.4414883553981781, "learning_rate": 0.0005, "loss": 1.1390008926391602, "step": 46914 }, { "epoch": 3.6984627512810406, "grad_norm": 0.48840880393981934, "learning_rate": 0.0005, "loss": 1.1723384857177734, "step": 46915 }, { "epoch": 3.6985415845486793, "grad_norm": 0.42062902450561523, "learning_rate": 0.0005, "loss": 1.0038493871688843, "step": 46916 }, { "epoch": 3.6986204178163185, "grad_norm": 0.35910874605178833, "learning_rate": 0.0005, "loss": 1.1133239269256592, "step": 46917 }, { "epoch": 3.6986992510839576, "grad_norm": 0.5210565328598022, "learning_rate": 0.0005, "loss": 1.2003281116485596, "step": 46918 }, { "epoch": 3.6987780843515963, "grad_norm": 0.548124372959137, "learning_rate": 0.0005, "loss": 1.0356950759887695, "step": 46919 }, { "epoch": 3.6988569176192354, "grad_norm": 0.5208722352981567, "learning_rate": 0.0005, "loss": 1.122938632965088, "step": 46920 }, { "epoch": 3.698935750886874, "grad_norm": 0.42640095949172974, "learning_rate": 0.0005, "loss": 1.1170138120651245, "step": 46921 }, { "epoch": 3.6990145841545132, "grad_norm": 0.5035885572433472, "learning_rate": 0.0005, "loss": 1.2908601760864258, "step": 46922 }, { "epoch": 3.6990934174221524, "grad_norm": 0.5141063332557678, "learning_rate": 0.0005, "loss": 1.131946325302124, "step": 46923 }, { "epoch": 3.699172250689791, "grad_norm": 0.5106687545776367, "learning_rate": 0.0005, "loss": 1.2188241481781006, "step": 46924 }, { "epoch": 3.69925108395743, "grad_norm": 0.5785642266273499, "learning_rate": 0.0005, "loss": 1.2128891944885254, "step": 46925 }, { "epoch": 3.699329917225069, "grad_norm": 0.41936036944389343, "learning_rate": 0.0005, "loss": 1.0117989778518677, "step": 46926 }, { "epoch": 3.699408750492708, "grad_norm": 0.42644762992858887, "learning_rate": 0.0005, "loss": 1.1945552825927734, "step": 46927 }, { "epoch": 3.699487583760347, "grad_norm": 0.48700764775276184, "learning_rate": 0.0005, "loss": 1.343623161315918, "step": 46928 }, { "epoch": 3.699566417027986, "grad_norm": 0.5013893842697144, "learning_rate": 0.0005, "loss": 1.214486837387085, "step": 46929 }, { "epoch": 3.6996452502956245, "grad_norm": 0.7931612133979797, "learning_rate": 0.0005, "loss": 1.244145393371582, "step": 46930 }, { "epoch": 3.6997240835632637, "grad_norm": 0.46901610493659973, "learning_rate": 0.0005, "loss": 1.038740634918213, "step": 46931 }, { "epoch": 3.699802916830903, "grad_norm": 0.45990413427352905, "learning_rate": 0.0005, "loss": 1.151721477508545, "step": 46932 }, { "epoch": 3.6998817500985415, "grad_norm": 0.5757656693458557, "learning_rate": 0.0005, "loss": 1.2588444948196411, "step": 46933 }, { "epoch": 3.6999605833661806, "grad_norm": 0.4598183035850525, "learning_rate": 0.0005, "loss": 1.1628230810165405, "step": 46934 }, { "epoch": 3.7000394166338193, "grad_norm": 0.5314276218414307, "learning_rate": 0.0005, "loss": 1.1877752542495728, "step": 46935 }, { "epoch": 3.7001182499014584, "grad_norm": 0.4852480888366699, "learning_rate": 0.0005, "loss": 1.1728529930114746, "step": 46936 }, { "epoch": 3.7001970831690976, "grad_norm": 0.4606856107711792, "learning_rate": 0.0005, "loss": 1.0371532440185547, "step": 46937 }, { "epoch": 3.7002759164367363, "grad_norm": 0.40784379839897156, "learning_rate": 0.0005, "loss": 1.1238774061203003, "step": 46938 }, { "epoch": 3.7003547497043754, "grad_norm": 0.5620031356811523, "learning_rate": 0.0005, "loss": 1.3086440563201904, "step": 46939 }, { "epoch": 3.700433582972014, "grad_norm": 0.6986834406852722, "learning_rate": 0.0005, "loss": 1.2323750257492065, "step": 46940 }, { "epoch": 3.700512416239653, "grad_norm": 0.5834417343139648, "learning_rate": 0.0005, "loss": 1.3995051383972168, "step": 46941 }, { "epoch": 3.7005912495072923, "grad_norm": 0.5545914173126221, "learning_rate": 0.0005, "loss": 1.1919814348220825, "step": 46942 }, { "epoch": 3.700670082774931, "grad_norm": 0.4382772743701935, "learning_rate": 0.0005, "loss": 1.0202062129974365, "step": 46943 }, { "epoch": 3.7007489160425697, "grad_norm": 0.7229049801826477, "learning_rate": 0.0005, "loss": 1.1700762510299683, "step": 46944 }, { "epoch": 3.700827749310209, "grad_norm": 0.5482728481292725, "learning_rate": 0.0005, "loss": 1.1613070964813232, "step": 46945 }, { "epoch": 3.700906582577848, "grad_norm": 0.386138379573822, "learning_rate": 0.0005, "loss": 0.9692613482475281, "step": 46946 }, { "epoch": 3.7009854158454867, "grad_norm": 0.4717428684234619, "learning_rate": 0.0005, "loss": 1.2225602865219116, "step": 46947 }, { "epoch": 3.701064249113126, "grad_norm": 0.5928312540054321, "learning_rate": 0.0005, "loss": 1.3320865631103516, "step": 46948 }, { "epoch": 3.7011430823807645, "grad_norm": 0.6713511347770691, "learning_rate": 0.0005, "loss": 1.1332919597625732, "step": 46949 }, { "epoch": 3.7012219156484036, "grad_norm": 0.48682233691215515, "learning_rate": 0.0005, "loss": 1.2112274169921875, "step": 46950 }, { "epoch": 3.7013007489160428, "grad_norm": 0.5118873119354248, "learning_rate": 0.0005, "loss": 1.1463572978973389, "step": 46951 }, { "epoch": 3.7013795821836815, "grad_norm": 0.503394603729248, "learning_rate": 0.0005, "loss": 1.1876311302185059, "step": 46952 }, { "epoch": 3.7014584154513206, "grad_norm": 0.6163498163223267, "learning_rate": 0.0005, "loss": 1.2753033638000488, "step": 46953 }, { "epoch": 3.7015372487189593, "grad_norm": 0.5021976828575134, "learning_rate": 0.0005, "loss": 1.1091666221618652, "step": 46954 }, { "epoch": 3.7016160819865984, "grad_norm": 0.5044662356376648, "learning_rate": 0.0005, "loss": 1.0867893695831299, "step": 46955 }, { "epoch": 3.7016949152542376, "grad_norm": 0.5583498477935791, "learning_rate": 0.0005, "loss": 1.2005561590194702, "step": 46956 }, { "epoch": 3.7017737485218762, "grad_norm": 0.507831871509552, "learning_rate": 0.0005, "loss": 1.3054041862487793, "step": 46957 }, { "epoch": 3.701852581789515, "grad_norm": 0.4880254864692688, "learning_rate": 0.0005, "loss": 1.170676827430725, "step": 46958 }, { "epoch": 3.701931415057154, "grad_norm": 0.4725780487060547, "learning_rate": 0.0005, "loss": 1.172612190246582, "step": 46959 }, { "epoch": 3.702010248324793, "grad_norm": 0.4965968132019043, "learning_rate": 0.0005, "loss": 1.2474225759506226, "step": 46960 }, { "epoch": 3.702089081592432, "grad_norm": 0.5860884189605713, "learning_rate": 0.0005, "loss": 1.2268412113189697, "step": 46961 }, { "epoch": 3.702167914860071, "grad_norm": 0.4249013364315033, "learning_rate": 0.0005, "loss": 1.1005185842514038, "step": 46962 }, { "epoch": 3.7022467481277097, "grad_norm": 0.5570046901702881, "learning_rate": 0.0005, "loss": 1.0982379913330078, "step": 46963 }, { "epoch": 3.702325581395349, "grad_norm": 0.5005532503128052, "learning_rate": 0.0005, "loss": 1.1985753774642944, "step": 46964 }, { "epoch": 3.702404414662988, "grad_norm": 0.5051524639129639, "learning_rate": 0.0005, "loss": 1.2016148567199707, "step": 46965 }, { "epoch": 3.7024832479306267, "grad_norm": 0.8324722051620483, "learning_rate": 0.0005, "loss": 1.0913887023925781, "step": 46966 }, { "epoch": 3.702562081198266, "grad_norm": 0.39703553915023804, "learning_rate": 0.0005, "loss": 1.161617636680603, "step": 46967 }, { "epoch": 3.7026409144659045, "grad_norm": 0.5659434795379639, "learning_rate": 0.0005, "loss": 0.9561223983764648, "step": 46968 }, { "epoch": 3.7027197477335436, "grad_norm": 0.5542515516281128, "learning_rate": 0.0005, "loss": 1.1208961009979248, "step": 46969 }, { "epoch": 3.7027985810011828, "grad_norm": 0.43619152903556824, "learning_rate": 0.0005, "loss": 0.9184815883636475, "step": 46970 }, { "epoch": 3.7028774142688214, "grad_norm": 0.6515117883682251, "learning_rate": 0.0005, "loss": 1.176405668258667, "step": 46971 }, { "epoch": 3.70295624753646, "grad_norm": 0.5561482906341553, "learning_rate": 0.0005, "loss": 1.218985915184021, "step": 46972 }, { "epoch": 3.7030350808040993, "grad_norm": 0.49596676230430603, "learning_rate": 0.0005, "loss": 1.1810638904571533, "step": 46973 }, { "epoch": 3.7031139140717384, "grad_norm": 0.40860557556152344, "learning_rate": 0.0005, "loss": 1.0265940427780151, "step": 46974 }, { "epoch": 3.703192747339377, "grad_norm": 0.651763379573822, "learning_rate": 0.0005, "loss": 1.1905722618103027, "step": 46975 }, { "epoch": 3.7032715806070162, "grad_norm": 0.434414803981781, "learning_rate": 0.0005, "loss": 1.042330026626587, "step": 46976 }, { "epoch": 3.703350413874655, "grad_norm": 0.5848256349563599, "learning_rate": 0.0005, "loss": 1.1607272624969482, "step": 46977 }, { "epoch": 3.703429247142294, "grad_norm": 0.6478274464607239, "learning_rate": 0.0005, "loss": 1.0053989887237549, "step": 46978 }, { "epoch": 3.703508080409933, "grad_norm": 0.5365073680877686, "learning_rate": 0.0005, "loss": 1.3010138273239136, "step": 46979 }, { "epoch": 3.703586913677572, "grad_norm": 0.4695917069911957, "learning_rate": 0.0005, "loss": 1.1893304586410522, "step": 46980 }, { "epoch": 3.703665746945211, "grad_norm": 0.5509558320045471, "learning_rate": 0.0005, "loss": 1.1593236923217773, "step": 46981 }, { "epoch": 3.7037445802128497, "grad_norm": 0.43502989411354065, "learning_rate": 0.0005, "loss": 1.0477650165557861, "step": 46982 }, { "epoch": 3.703823413480489, "grad_norm": 0.4398915469646454, "learning_rate": 0.0005, "loss": 1.2448022365570068, "step": 46983 }, { "epoch": 3.703902246748128, "grad_norm": 0.5450535416603088, "learning_rate": 0.0005, "loss": 1.280555009841919, "step": 46984 }, { "epoch": 3.7039810800157666, "grad_norm": 0.5820955038070679, "learning_rate": 0.0005, "loss": 1.4143248796463013, "step": 46985 }, { "epoch": 3.7040599132834053, "grad_norm": 0.4748273491859436, "learning_rate": 0.0005, "loss": 1.371667742729187, "step": 46986 }, { "epoch": 3.7041387465510445, "grad_norm": 0.5100811123847961, "learning_rate": 0.0005, "loss": 1.2789298295974731, "step": 46987 }, { "epoch": 3.7042175798186836, "grad_norm": 0.4572702944278717, "learning_rate": 0.0005, "loss": 1.216911792755127, "step": 46988 }, { "epoch": 3.7042964130863223, "grad_norm": 0.596622884273529, "learning_rate": 0.0005, "loss": 1.2864595651626587, "step": 46989 }, { "epoch": 3.7043752463539614, "grad_norm": 0.40567299723625183, "learning_rate": 0.0005, "loss": 1.1243715286254883, "step": 46990 }, { "epoch": 3.7044540796216, "grad_norm": 0.42364269495010376, "learning_rate": 0.0005, "loss": 1.1042282581329346, "step": 46991 }, { "epoch": 3.7045329128892392, "grad_norm": 0.49382948875427246, "learning_rate": 0.0005, "loss": 1.0414237976074219, "step": 46992 }, { "epoch": 3.7046117461568784, "grad_norm": 0.5164781212806702, "learning_rate": 0.0005, "loss": 1.3930881023406982, "step": 46993 }, { "epoch": 3.704690579424517, "grad_norm": 0.44991374015808105, "learning_rate": 0.0005, "loss": 1.2484437227249146, "step": 46994 }, { "epoch": 3.704769412692156, "grad_norm": 0.49543297290802, "learning_rate": 0.0005, "loss": 1.1631786823272705, "step": 46995 }, { "epoch": 3.704848245959795, "grad_norm": 0.45678654313087463, "learning_rate": 0.0005, "loss": 1.0236256122589111, "step": 46996 }, { "epoch": 3.704927079227434, "grad_norm": 0.6170737743377686, "learning_rate": 0.0005, "loss": 1.4022815227508545, "step": 46997 }, { "epoch": 3.705005912495073, "grad_norm": 0.6726410984992981, "learning_rate": 0.0005, "loss": 1.173786997795105, "step": 46998 }, { "epoch": 3.705084745762712, "grad_norm": 0.39759939908981323, "learning_rate": 0.0005, "loss": 1.001163363456726, "step": 46999 }, { "epoch": 3.7051635790303505, "grad_norm": 0.5310885310173035, "learning_rate": 0.0005, "loss": 1.295653223991394, "step": 47000 }, { "epoch": 3.7052424122979897, "grad_norm": 0.5079963207244873, "learning_rate": 0.0005, "loss": 1.0026814937591553, "step": 47001 }, { "epoch": 3.705321245565629, "grad_norm": 0.5345878005027771, "learning_rate": 0.0005, "loss": 1.1929924488067627, "step": 47002 }, { "epoch": 3.7054000788332675, "grad_norm": 0.474058598279953, "learning_rate": 0.0005, "loss": 1.2851853370666504, "step": 47003 }, { "epoch": 3.7054789121009066, "grad_norm": 0.452029824256897, "learning_rate": 0.0005, "loss": 1.1106703281402588, "step": 47004 }, { "epoch": 3.7055577453685453, "grad_norm": 0.537594735622406, "learning_rate": 0.0005, "loss": 1.1682095527648926, "step": 47005 }, { "epoch": 3.7056365786361845, "grad_norm": 0.4802529513835907, "learning_rate": 0.0005, "loss": 1.1060142517089844, "step": 47006 }, { "epoch": 3.7057154119038236, "grad_norm": 0.4823659062385559, "learning_rate": 0.0005, "loss": 1.1020255088806152, "step": 47007 }, { "epoch": 3.7057942451714623, "grad_norm": 0.46048304438591003, "learning_rate": 0.0005, "loss": 1.036600947380066, "step": 47008 }, { "epoch": 3.7058730784391014, "grad_norm": 0.513748824596405, "learning_rate": 0.0005, "loss": 1.1951802968978882, "step": 47009 }, { "epoch": 3.70595191170674, "grad_norm": 0.4896095097064972, "learning_rate": 0.0005, "loss": 1.103014588356018, "step": 47010 }, { "epoch": 3.7060307449743792, "grad_norm": 0.5527665019035339, "learning_rate": 0.0005, "loss": 1.2643567323684692, "step": 47011 }, { "epoch": 3.7061095782420184, "grad_norm": 0.44969573616981506, "learning_rate": 0.0005, "loss": 1.1841371059417725, "step": 47012 }, { "epoch": 3.706188411509657, "grad_norm": 0.5603436231613159, "learning_rate": 0.0005, "loss": 1.283089280128479, "step": 47013 }, { "epoch": 3.7062672447772957, "grad_norm": 0.6706597805023193, "learning_rate": 0.0005, "loss": 1.3218210935592651, "step": 47014 }, { "epoch": 3.706346078044935, "grad_norm": 0.34998390078544617, "learning_rate": 0.0005, "loss": 1.0282557010650635, "step": 47015 }, { "epoch": 3.706424911312574, "grad_norm": 0.4459271728992462, "learning_rate": 0.0005, "loss": 1.1456375122070312, "step": 47016 }, { "epoch": 3.7065037445802127, "grad_norm": 0.4293460547924042, "learning_rate": 0.0005, "loss": 1.1413904428482056, "step": 47017 }, { "epoch": 3.706582577847852, "grad_norm": 0.4141087830066681, "learning_rate": 0.0005, "loss": 1.0271780490875244, "step": 47018 }, { "epoch": 3.7066614111154905, "grad_norm": 0.5546520948410034, "learning_rate": 0.0005, "loss": 1.259765386581421, "step": 47019 }, { "epoch": 3.7067402443831297, "grad_norm": 0.5851903557777405, "learning_rate": 0.0005, "loss": 0.9894835352897644, "step": 47020 }, { "epoch": 3.706819077650769, "grad_norm": 0.5248143076896667, "learning_rate": 0.0005, "loss": 1.1119418144226074, "step": 47021 }, { "epoch": 3.7068979109184075, "grad_norm": 0.43693941831588745, "learning_rate": 0.0005, "loss": 1.1090281009674072, "step": 47022 }, { "epoch": 3.7069767441860466, "grad_norm": 0.5325188040733337, "learning_rate": 0.0005, "loss": 1.3138402700424194, "step": 47023 }, { "epoch": 3.7070555774536853, "grad_norm": 0.565586268901825, "learning_rate": 0.0005, "loss": 1.3882161378860474, "step": 47024 }, { "epoch": 3.7071344107213244, "grad_norm": 0.48935258388519287, "learning_rate": 0.0005, "loss": 1.3305354118347168, "step": 47025 }, { "epoch": 3.7072132439889636, "grad_norm": 0.48108556866645813, "learning_rate": 0.0005, "loss": 1.2267186641693115, "step": 47026 }, { "epoch": 3.7072920772566023, "grad_norm": 0.6328234672546387, "learning_rate": 0.0005, "loss": 1.1124873161315918, "step": 47027 }, { "epoch": 3.7073709105242414, "grad_norm": 0.4356291890144348, "learning_rate": 0.0005, "loss": 1.0891705751419067, "step": 47028 }, { "epoch": 3.70744974379188, "grad_norm": 0.5372101068496704, "learning_rate": 0.0005, "loss": 1.128502607345581, "step": 47029 }, { "epoch": 3.707528577059519, "grad_norm": 0.46015113592147827, "learning_rate": 0.0005, "loss": 1.0536656379699707, "step": 47030 }, { "epoch": 3.707607410327158, "grad_norm": 0.45781728625297546, "learning_rate": 0.0005, "loss": 1.1374378204345703, "step": 47031 }, { "epoch": 3.707686243594797, "grad_norm": 0.5389949083328247, "learning_rate": 0.0005, "loss": 1.2835729122161865, "step": 47032 }, { "epoch": 3.7077650768624357, "grad_norm": 0.4519518315792084, "learning_rate": 0.0005, "loss": 1.2605386972427368, "step": 47033 }, { "epoch": 3.707843910130075, "grad_norm": 0.4796331524848938, "learning_rate": 0.0005, "loss": 1.0779699087142944, "step": 47034 }, { "epoch": 3.707922743397714, "grad_norm": 0.47666460275650024, "learning_rate": 0.0005, "loss": 1.1502132415771484, "step": 47035 }, { "epoch": 3.7080015766653527, "grad_norm": 0.5661601424217224, "learning_rate": 0.0005, "loss": 1.2767579555511475, "step": 47036 }, { "epoch": 3.708080409932992, "grad_norm": 0.5075061321258545, "learning_rate": 0.0005, "loss": 1.159541130065918, "step": 47037 }, { "epoch": 3.7081592432006305, "grad_norm": 0.4436064064502716, "learning_rate": 0.0005, "loss": 1.1468348503112793, "step": 47038 }, { "epoch": 3.7082380764682696, "grad_norm": 0.5248683094978333, "learning_rate": 0.0005, "loss": 1.2062721252441406, "step": 47039 }, { "epoch": 3.7083169097359088, "grad_norm": 0.5407711863517761, "learning_rate": 0.0005, "loss": 1.1712924242019653, "step": 47040 }, { "epoch": 3.7083957430035475, "grad_norm": 0.4647124409675598, "learning_rate": 0.0005, "loss": 1.0127757787704468, "step": 47041 }, { "epoch": 3.7084745762711866, "grad_norm": 0.5219649076461792, "learning_rate": 0.0005, "loss": 1.1082566976547241, "step": 47042 }, { "epoch": 3.7085534095388253, "grad_norm": 0.5119364857673645, "learning_rate": 0.0005, "loss": 1.0835552215576172, "step": 47043 }, { "epoch": 3.7086322428064644, "grad_norm": 0.52278733253479, "learning_rate": 0.0005, "loss": 1.1370906829833984, "step": 47044 }, { "epoch": 3.708711076074103, "grad_norm": 0.5631520748138428, "learning_rate": 0.0005, "loss": 1.011475682258606, "step": 47045 }, { "epoch": 3.7087899093417422, "grad_norm": 0.41617417335510254, "learning_rate": 0.0005, "loss": 0.9598861336708069, "step": 47046 }, { "epoch": 3.708868742609381, "grad_norm": 0.5220712423324585, "learning_rate": 0.0005, "loss": 1.3123631477355957, "step": 47047 }, { "epoch": 3.70894757587702, "grad_norm": 0.49373120069503784, "learning_rate": 0.0005, "loss": 1.2865798473358154, "step": 47048 }, { "epoch": 3.709026409144659, "grad_norm": 0.41359439492225647, "learning_rate": 0.0005, "loss": 1.1308555603027344, "step": 47049 }, { "epoch": 3.709105242412298, "grad_norm": 0.41797935962677, "learning_rate": 0.0005, "loss": 1.1699597835540771, "step": 47050 }, { "epoch": 3.709184075679937, "grad_norm": 0.4827440083026886, "learning_rate": 0.0005, "loss": 1.287153720855713, "step": 47051 }, { "epoch": 3.7092629089475757, "grad_norm": 0.4077005982398987, "learning_rate": 0.0005, "loss": 1.108738660812378, "step": 47052 }, { "epoch": 3.709341742215215, "grad_norm": 0.41462022066116333, "learning_rate": 0.0005, "loss": 1.087207555770874, "step": 47053 }, { "epoch": 3.709420575482854, "grad_norm": 0.4626845121383667, "learning_rate": 0.0005, "loss": 1.1189675331115723, "step": 47054 }, { "epoch": 3.7094994087504927, "grad_norm": 0.4554724395275116, "learning_rate": 0.0005, "loss": 1.0899114608764648, "step": 47055 }, { "epoch": 3.709578242018132, "grad_norm": 0.4823145866394043, "learning_rate": 0.0005, "loss": 1.2033320665359497, "step": 47056 }, { "epoch": 3.7096570752857705, "grad_norm": 0.5068240165710449, "learning_rate": 0.0005, "loss": 1.1280572414398193, "step": 47057 }, { "epoch": 3.7097359085534096, "grad_norm": 0.46618062257766724, "learning_rate": 0.0005, "loss": 1.2286489009857178, "step": 47058 }, { "epoch": 3.7098147418210488, "grad_norm": 0.5372463464736938, "learning_rate": 0.0005, "loss": 1.264240026473999, "step": 47059 }, { "epoch": 3.7098935750886874, "grad_norm": 0.4749488830566406, "learning_rate": 0.0005, "loss": 1.075429081916809, "step": 47060 }, { "epoch": 3.709972408356326, "grad_norm": 0.47266295552253723, "learning_rate": 0.0005, "loss": 1.1184866428375244, "step": 47061 }, { "epoch": 3.7100512416239653, "grad_norm": 0.44925037026405334, "learning_rate": 0.0005, "loss": 0.9777132272720337, "step": 47062 }, { "epoch": 3.7101300748916044, "grad_norm": 0.45165300369262695, "learning_rate": 0.0005, "loss": 1.138324499130249, "step": 47063 }, { "epoch": 3.710208908159243, "grad_norm": 0.7779543399810791, "learning_rate": 0.0005, "loss": 1.2024481296539307, "step": 47064 }, { "epoch": 3.7102877414268822, "grad_norm": 0.4241328537464142, "learning_rate": 0.0005, "loss": 1.1390042304992676, "step": 47065 }, { "epoch": 3.710366574694521, "grad_norm": 0.42809391021728516, "learning_rate": 0.0005, "loss": 0.944078803062439, "step": 47066 }, { "epoch": 3.71044540796216, "grad_norm": 0.4885260760784149, "learning_rate": 0.0005, "loss": 1.1552529335021973, "step": 47067 }, { "epoch": 3.710524241229799, "grad_norm": 0.5015714168548584, "learning_rate": 0.0005, "loss": 1.1537423133850098, "step": 47068 }, { "epoch": 3.710603074497438, "grad_norm": 0.47942978143692017, "learning_rate": 0.0005, "loss": 1.1419109106063843, "step": 47069 }, { "epoch": 3.710681907765077, "grad_norm": 0.4478766620159149, "learning_rate": 0.0005, "loss": 1.012312650680542, "step": 47070 }, { "epoch": 3.7107607410327157, "grad_norm": 0.5074489116668701, "learning_rate": 0.0005, "loss": 1.1133815050125122, "step": 47071 }, { "epoch": 3.710839574300355, "grad_norm": 0.4504036605358124, "learning_rate": 0.0005, "loss": 1.068755030632019, "step": 47072 }, { "epoch": 3.710918407567994, "grad_norm": 0.5012994408607483, "learning_rate": 0.0005, "loss": 1.0878337621688843, "step": 47073 }, { "epoch": 3.7109972408356326, "grad_norm": 0.45348605513572693, "learning_rate": 0.0005, "loss": 1.0965073108673096, "step": 47074 }, { "epoch": 3.7110760741032713, "grad_norm": 0.5545541048049927, "learning_rate": 0.0005, "loss": 1.2274384498596191, "step": 47075 }, { "epoch": 3.7111549073709105, "grad_norm": 0.47808119654655457, "learning_rate": 0.0005, "loss": 1.111280083656311, "step": 47076 }, { "epoch": 3.7112337406385496, "grad_norm": 0.4650072455406189, "learning_rate": 0.0005, "loss": 1.160574197769165, "step": 47077 }, { "epoch": 3.7113125739061883, "grad_norm": 0.49952077865600586, "learning_rate": 0.0005, "loss": 1.2556936740875244, "step": 47078 }, { "epoch": 3.7113914071738274, "grad_norm": 0.440999835729599, "learning_rate": 0.0005, "loss": 1.1592073440551758, "step": 47079 }, { "epoch": 3.711470240441466, "grad_norm": 0.5053477883338928, "learning_rate": 0.0005, "loss": 1.1610780954360962, "step": 47080 }, { "epoch": 3.7115490737091053, "grad_norm": 0.5136944651603699, "learning_rate": 0.0005, "loss": 1.0988456010818481, "step": 47081 }, { "epoch": 3.7116279069767444, "grad_norm": 0.6763281226158142, "learning_rate": 0.0005, "loss": 1.232569694519043, "step": 47082 }, { "epoch": 3.711706740244383, "grad_norm": 0.8560540676116943, "learning_rate": 0.0005, "loss": 1.371056318283081, "step": 47083 }, { "epoch": 3.711785573512022, "grad_norm": 0.5562229156494141, "learning_rate": 0.0005, "loss": 1.140238642692566, "step": 47084 }, { "epoch": 3.711864406779661, "grad_norm": 0.7475435137748718, "learning_rate": 0.0005, "loss": 1.3196364641189575, "step": 47085 }, { "epoch": 3.7119432400473, "grad_norm": 0.5461942553520203, "learning_rate": 0.0005, "loss": 1.248408555984497, "step": 47086 }, { "epoch": 3.712022073314939, "grad_norm": 0.5417335033416748, "learning_rate": 0.0005, "loss": 0.9937793612480164, "step": 47087 }, { "epoch": 3.712100906582578, "grad_norm": 0.5866767168045044, "learning_rate": 0.0005, "loss": 1.3161797523498535, "step": 47088 }, { "epoch": 3.7121797398502165, "grad_norm": 0.39738744497299194, "learning_rate": 0.0005, "loss": 0.9689101576805115, "step": 47089 }, { "epoch": 3.7122585731178557, "grad_norm": 0.5668476819992065, "learning_rate": 0.0005, "loss": 1.1995794773101807, "step": 47090 }, { "epoch": 3.712337406385495, "grad_norm": 0.5653735399246216, "learning_rate": 0.0005, "loss": 1.1266319751739502, "step": 47091 }, { "epoch": 3.7124162396531335, "grad_norm": 0.41775915026664734, "learning_rate": 0.0005, "loss": 1.2301654815673828, "step": 47092 }, { "epoch": 3.7124950729207726, "grad_norm": 0.6298370361328125, "learning_rate": 0.0005, "loss": 1.2588856220245361, "step": 47093 }, { "epoch": 3.7125739061884113, "grad_norm": 0.4073375165462494, "learning_rate": 0.0005, "loss": 0.9573907256126404, "step": 47094 }, { "epoch": 3.7126527394560505, "grad_norm": 0.48932206630706787, "learning_rate": 0.0005, "loss": 1.1867462396621704, "step": 47095 }, { "epoch": 3.7127315727236896, "grad_norm": 0.6905373930931091, "learning_rate": 0.0005, "loss": 1.1466701030731201, "step": 47096 }, { "epoch": 3.7128104059913283, "grad_norm": 0.5242683291435242, "learning_rate": 0.0005, "loss": 1.0143699645996094, "step": 47097 }, { "epoch": 3.7128892392589674, "grad_norm": 0.5531546473503113, "learning_rate": 0.0005, "loss": 1.2654945850372314, "step": 47098 }, { "epoch": 3.712968072526606, "grad_norm": 0.5746904611587524, "learning_rate": 0.0005, "loss": 1.1826064586639404, "step": 47099 }, { "epoch": 3.7130469057942452, "grad_norm": 0.7466166019439697, "learning_rate": 0.0005, "loss": 1.2378630638122559, "step": 47100 }, { "epoch": 3.7131257390618844, "grad_norm": 0.47730645537376404, "learning_rate": 0.0005, "loss": 1.1073280572891235, "step": 47101 }, { "epoch": 3.713204572329523, "grad_norm": 0.6404744386672974, "learning_rate": 0.0005, "loss": 1.3858346939086914, "step": 47102 }, { "epoch": 3.7132834055971617, "grad_norm": 0.49770647287368774, "learning_rate": 0.0005, "loss": 1.1404588222503662, "step": 47103 }, { "epoch": 3.713362238864801, "grad_norm": 0.40252941846847534, "learning_rate": 0.0005, "loss": 0.9470983743667603, "step": 47104 }, { "epoch": 3.71344107213244, "grad_norm": 0.5018842816352844, "learning_rate": 0.0005, "loss": 1.3004027605056763, "step": 47105 }, { "epoch": 3.7135199054000787, "grad_norm": 0.5079991221427917, "learning_rate": 0.0005, "loss": 1.26426100730896, "step": 47106 }, { "epoch": 3.713598738667718, "grad_norm": 0.3478645384311676, "learning_rate": 0.0005, "loss": 0.8791888952255249, "step": 47107 }, { "epoch": 3.7136775719353565, "grad_norm": 0.5051834583282471, "learning_rate": 0.0005, "loss": 1.2211298942565918, "step": 47108 }, { "epoch": 3.7137564052029957, "grad_norm": 0.5312674045562744, "learning_rate": 0.0005, "loss": 1.2503224611282349, "step": 47109 }, { "epoch": 3.713835238470635, "grad_norm": 0.5065585970878601, "learning_rate": 0.0005, "loss": 1.176177978515625, "step": 47110 }, { "epoch": 3.7139140717382735, "grad_norm": 0.36676034331321716, "learning_rate": 0.0005, "loss": 1.0238441228866577, "step": 47111 }, { "epoch": 3.7139929050059126, "grad_norm": 1.1194019317626953, "learning_rate": 0.0005, "loss": 1.5760414600372314, "step": 47112 }, { "epoch": 3.7140717382735513, "grad_norm": 0.45023271441459656, "learning_rate": 0.0005, "loss": 1.0625749826431274, "step": 47113 }, { "epoch": 3.7141505715411904, "grad_norm": 0.5437976121902466, "learning_rate": 0.0005, "loss": 1.1076619625091553, "step": 47114 }, { "epoch": 3.7142294048088296, "grad_norm": 0.5508040189743042, "learning_rate": 0.0005, "loss": 1.210730791091919, "step": 47115 }, { "epoch": 3.7143082380764683, "grad_norm": 5.164280891418457, "learning_rate": 0.0005, "loss": 1.2522432804107666, "step": 47116 }, { "epoch": 3.714387071344107, "grad_norm": 0.49520939588546753, "learning_rate": 0.0005, "loss": 1.0624977350234985, "step": 47117 }, { "epoch": 3.714465904611746, "grad_norm": 0.4210755527019501, "learning_rate": 0.0005, "loss": 1.1646941900253296, "step": 47118 }, { "epoch": 3.714544737879385, "grad_norm": 0.4925489127635956, "learning_rate": 0.0005, "loss": 1.0366936922073364, "step": 47119 }, { "epoch": 3.714623571147024, "grad_norm": 0.607853889465332, "learning_rate": 0.0005, "loss": 1.1320151090621948, "step": 47120 }, { "epoch": 3.714702404414663, "grad_norm": 0.41151663661003113, "learning_rate": 0.0005, "loss": 0.990210771560669, "step": 47121 }, { "epoch": 3.7147812376823017, "grad_norm": 0.5080411434173584, "learning_rate": 0.0005, "loss": 1.0584206581115723, "step": 47122 }, { "epoch": 3.714860070949941, "grad_norm": 0.5849586725234985, "learning_rate": 0.0005, "loss": 1.251791000366211, "step": 47123 }, { "epoch": 3.71493890421758, "grad_norm": 0.49773576855659485, "learning_rate": 0.0005, "loss": 1.111998200416565, "step": 47124 }, { "epoch": 3.7150177374852187, "grad_norm": 0.4610898494720459, "learning_rate": 0.0005, "loss": 1.0643818378448486, "step": 47125 }, { "epoch": 3.715096570752858, "grad_norm": 0.3901474177837372, "learning_rate": 0.0005, "loss": 1.0877060890197754, "step": 47126 }, { "epoch": 3.7151754040204965, "grad_norm": 0.4755585193634033, "learning_rate": 0.0005, "loss": 1.0597481727600098, "step": 47127 }, { "epoch": 3.7152542372881356, "grad_norm": 0.48505720496177673, "learning_rate": 0.0005, "loss": 1.1332532167434692, "step": 47128 }, { "epoch": 3.7153330705557748, "grad_norm": 0.4179072976112366, "learning_rate": 0.0005, "loss": 1.1071255207061768, "step": 47129 }, { "epoch": 3.7154119038234135, "grad_norm": 0.5826844573020935, "learning_rate": 0.0005, "loss": 1.1530265808105469, "step": 47130 }, { "epoch": 3.715490737091052, "grad_norm": 0.42783915996551514, "learning_rate": 0.0005, "loss": 1.1161692142486572, "step": 47131 }, { "epoch": 3.7155695703586913, "grad_norm": 0.46450352668762207, "learning_rate": 0.0005, "loss": 1.0081639289855957, "step": 47132 }, { "epoch": 3.7156484036263304, "grad_norm": 0.5411396622657776, "learning_rate": 0.0005, "loss": 0.9961932897567749, "step": 47133 }, { "epoch": 3.715727236893969, "grad_norm": 0.5048567056655884, "learning_rate": 0.0005, "loss": 1.1815717220306396, "step": 47134 }, { "epoch": 3.7158060701616082, "grad_norm": 0.583690345287323, "learning_rate": 0.0005, "loss": 1.2080024480819702, "step": 47135 }, { "epoch": 3.715884903429247, "grad_norm": 0.4499274790287018, "learning_rate": 0.0005, "loss": 1.1954929828643799, "step": 47136 }, { "epoch": 3.715963736696886, "grad_norm": 0.6380215287208557, "learning_rate": 0.0005, "loss": 1.2048050165176392, "step": 47137 }, { "epoch": 3.716042569964525, "grad_norm": 0.4792526364326477, "learning_rate": 0.0005, "loss": 1.1737267971038818, "step": 47138 }, { "epoch": 3.716121403232164, "grad_norm": 0.5039910078048706, "learning_rate": 0.0005, "loss": 1.0990912914276123, "step": 47139 }, { "epoch": 3.716200236499803, "grad_norm": 0.5214114189147949, "learning_rate": 0.0005, "loss": 1.1296076774597168, "step": 47140 }, { "epoch": 3.7162790697674417, "grad_norm": 0.6429108381271362, "learning_rate": 0.0005, "loss": 1.3543086051940918, "step": 47141 }, { "epoch": 3.716357903035081, "grad_norm": 0.47894150018692017, "learning_rate": 0.0005, "loss": 1.2726380825042725, "step": 47142 }, { "epoch": 3.71643673630272, "grad_norm": 0.5776538848876953, "learning_rate": 0.0005, "loss": 1.2225067615509033, "step": 47143 }, { "epoch": 3.7165155695703587, "grad_norm": 0.5711541175842285, "learning_rate": 0.0005, "loss": 1.2457759380340576, "step": 47144 }, { "epoch": 3.716594402837998, "grad_norm": 0.3830343186855316, "learning_rate": 0.0005, "loss": 0.9813634157180786, "step": 47145 }, { "epoch": 3.7166732361056365, "grad_norm": 0.48419538140296936, "learning_rate": 0.0005, "loss": 1.3485904932022095, "step": 47146 }, { "epoch": 3.7167520693732756, "grad_norm": 0.594465970993042, "learning_rate": 0.0005, "loss": 1.3407683372497559, "step": 47147 }, { "epoch": 3.7168309026409143, "grad_norm": 0.44235262274742126, "learning_rate": 0.0005, "loss": 1.1543821096420288, "step": 47148 }, { "epoch": 3.7169097359085534, "grad_norm": 0.49265044927597046, "learning_rate": 0.0005, "loss": 1.198747158050537, "step": 47149 }, { "epoch": 3.716988569176192, "grad_norm": 0.7303048968315125, "learning_rate": 0.0005, "loss": 1.1539454460144043, "step": 47150 }, { "epoch": 3.7170674024438313, "grad_norm": 0.4276503920555115, "learning_rate": 0.0005, "loss": 1.0285298824310303, "step": 47151 }, { "epoch": 3.7171462357114704, "grad_norm": 0.42201128602027893, "learning_rate": 0.0005, "loss": 0.9884116649627686, "step": 47152 }, { "epoch": 3.717225068979109, "grad_norm": 0.4875768721103668, "learning_rate": 0.0005, "loss": 1.203844666481018, "step": 47153 }, { "epoch": 3.7173039022467482, "grad_norm": 0.5323154330253601, "learning_rate": 0.0005, "loss": 1.2906548976898193, "step": 47154 }, { "epoch": 3.717382735514387, "grad_norm": 0.5096399784088135, "learning_rate": 0.0005, "loss": 1.1476882696151733, "step": 47155 }, { "epoch": 3.717461568782026, "grad_norm": 0.5641473531723022, "learning_rate": 0.0005, "loss": 1.2642109394073486, "step": 47156 }, { "epoch": 3.717540402049665, "grad_norm": 0.4843803346157074, "learning_rate": 0.0005, "loss": 1.3403311967849731, "step": 47157 }, { "epoch": 3.717619235317304, "grad_norm": 0.5575552582740784, "learning_rate": 0.0005, "loss": 1.2048509120941162, "step": 47158 }, { "epoch": 3.717698068584943, "grad_norm": 0.5243378281593323, "learning_rate": 0.0005, "loss": 1.280613899230957, "step": 47159 }, { "epoch": 3.7177769018525817, "grad_norm": 0.5658305287361145, "learning_rate": 0.0005, "loss": 1.0594329833984375, "step": 47160 }, { "epoch": 3.717855735120221, "grad_norm": 0.5727092027664185, "learning_rate": 0.0005, "loss": 1.1514661312103271, "step": 47161 }, { "epoch": 3.7179345683878595, "grad_norm": 0.45811140537261963, "learning_rate": 0.0005, "loss": 1.1433203220367432, "step": 47162 }, { "epoch": 3.7180134016554987, "grad_norm": 0.45574644207954407, "learning_rate": 0.0005, "loss": 1.1693342924118042, "step": 47163 }, { "epoch": 3.7180922349231373, "grad_norm": 0.5088109374046326, "learning_rate": 0.0005, "loss": 1.3624221086502075, "step": 47164 }, { "epoch": 3.7181710681907765, "grad_norm": 0.5743856430053711, "learning_rate": 0.0005, "loss": 1.3040642738342285, "step": 47165 }, { "epoch": 3.7182499014584156, "grad_norm": 0.5510872602462769, "learning_rate": 0.0005, "loss": 1.1333614587783813, "step": 47166 }, { "epoch": 3.7183287347260543, "grad_norm": 0.45000550150871277, "learning_rate": 0.0005, "loss": 1.1846950054168701, "step": 47167 }, { "epoch": 3.7184075679936934, "grad_norm": 0.5270125865936279, "learning_rate": 0.0005, "loss": 1.2963613271713257, "step": 47168 }, { "epoch": 3.718486401261332, "grad_norm": 0.45244571566581726, "learning_rate": 0.0005, "loss": 1.2295175790786743, "step": 47169 }, { "epoch": 3.7185652345289713, "grad_norm": 0.5163474082946777, "learning_rate": 0.0005, "loss": 1.1891062259674072, "step": 47170 }, { "epoch": 3.7186440677966104, "grad_norm": 0.4446788430213928, "learning_rate": 0.0005, "loss": 1.0739285945892334, "step": 47171 }, { "epoch": 3.718722901064249, "grad_norm": 0.4049566686153412, "learning_rate": 0.0005, "loss": 1.0606365203857422, "step": 47172 }, { "epoch": 3.718801734331888, "grad_norm": 0.4315853714942932, "learning_rate": 0.0005, "loss": 1.0372624397277832, "step": 47173 }, { "epoch": 3.718880567599527, "grad_norm": 0.4936140477657318, "learning_rate": 0.0005, "loss": 1.1629703044891357, "step": 47174 }, { "epoch": 3.718959400867166, "grad_norm": 0.5668266415596008, "learning_rate": 0.0005, "loss": 1.35782790184021, "step": 47175 }, { "epoch": 3.7190382341348047, "grad_norm": 0.536789059638977, "learning_rate": 0.0005, "loss": 1.2848708629608154, "step": 47176 }, { "epoch": 3.719117067402444, "grad_norm": 0.47837451100349426, "learning_rate": 0.0005, "loss": 1.237289547920227, "step": 47177 }, { "epoch": 3.7191959006700825, "grad_norm": 0.5130124688148499, "learning_rate": 0.0005, "loss": 1.366426944732666, "step": 47178 }, { "epoch": 3.7192747339377217, "grad_norm": 0.5085543990135193, "learning_rate": 0.0005, "loss": 1.228701114654541, "step": 47179 }, { "epoch": 3.719353567205361, "grad_norm": 0.600318193435669, "learning_rate": 0.0005, "loss": 1.0601168870925903, "step": 47180 }, { "epoch": 3.7194324004729995, "grad_norm": 0.48652979731559753, "learning_rate": 0.0005, "loss": 1.3018338680267334, "step": 47181 }, { "epoch": 3.7195112337406386, "grad_norm": 0.45406895875930786, "learning_rate": 0.0005, "loss": 1.1073992252349854, "step": 47182 }, { "epoch": 3.7195900670082773, "grad_norm": 0.4673353135585785, "learning_rate": 0.0005, "loss": 1.0464262962341309, "step": 47183 }, { "epoch": 3.7196689002759165, "grad_norm": 0.5789265036582947, "learning_rate": 0.0005, "loss": 1.3189210891723633, "step": 47184 }, { "epoch": 3.7197477335435556, "grad_norm": 0.6181321740150452, "learning_rate": 0.0005, "loss": 1.105615258216858, "step": 47185 }, { "epoch": 3.7198265668111943, "grad_norm": 0.4806452691555023, "learning_rate": 0.0005, "loss": 1.1536831855773926, "step": 47186 }, { "epoch": 3.7199054000788334, "grad_norm": 0.4973941743373871, "learning_rate": 0.0005, "loss": 1.1704661846160889, "step": 47187 }, { "epoch": 3.719984233346472, "grad_norm": 0.4755565822124481, "learning_rate": 0.0005, "loss": 1.195529580116272, "step": 47188 }, { "epoch": 3.7200630666141112, "grad_norm": 0.4546929895877838, "learning_rate": 0.0005, "loss": 1.2227532863616943, "step": 47189 }, { "epoch": 3.7201418998817504, "grad_norm": 0.5060598850250244, "learning_rate": 0.0005, "loss": 1.1847941875457764, "step": 47190 }, { "epoch": 3.720220733149389, "grad_norm": 0.426480233669281, "learning_rate": 0.0005, "loss": 1.1524606943130493, "step": 47191 }, { "epoch": 3.7202995664170277, "grad_norm": 0.5233043432235718, "learning_rate": 0.0005, "loss": 1.2080714702606201, "step": 47192 }, { "epoch": 3.720378399684667, "grad_norm": 0.43001535534858704, "learning_rate": 0.0005, "loss": 1.1376076936721802, "step": 47193 }, { "epoch": 3.720457232952306, "grad_norm": 0.4604358971118927, "learning_rate": 0.0005, "loss": 1.1311519145965576, "step": 47194 }, { "epoch": 3.7205360662199447, "grad_norm": 0.4146542549133301, "learning_rate": 0.0005, "loss": 1.0198968648910522, "step": 47195 }, { "epoch": 3.720614899487584, "grad_norm": 0.4743072986602783, "learning_rate": 0.0005, "loss": 1.055935025215149, "step": 47196 }, { "epoch": 3.7206937327552225, "grad_norm": 0.4600590765476227, "learning_rate": 0.0005, "loss": 0.9992327094078064, "step": 47197 }, { "epoch": 3.7207725660228617, "grad_norm": 0.5305862426757812, "learning_rate": 0.0005, "loss": 1.1295918226242065, "step": 47198 }, { "epoch": 3.720851399290501, "grad_norm": 0.48574215173721313, "learning_rate": 0.0005, "loss": 1.2195651531219482, "step": 47199 }, { "epoch": 3.7209302325581395, "grad_norm": 0.7441205382347107, "learning_rate": 0.0005, "loss": 1.2382732629776, "step": 47200 }, { "epoch": 3.7210090658257786, "grad_norm": 0.5195163488388062, "learning_rate": 0.0005, "loss": 1.219417929649353, "step": 47201 }, { "epoch": 3.7210878990934173, "grad_norm": 0.5936300754547119, "learning_rate": 0.0005, "loss": 1.368036150932312, "step": 47202 }, { "epoch": 3.7211667323610564, "grad_norm": 0.5283398032188416, "learning_rate": 0.0005, "loss": 1.0689605474472046, "step": 47203 }, { "epoch": 3.7212455656286956, "grad_norm": 0.45587074756622314, "learning_rate": 0.0005, "loss": 1.133786916732788, "step": 47204 }, { "epoch": 3.7213243988963343, "grad_norm": 0.35649704933166504, "learning_rate": 0.0005, "loss": 0.9575182199478149, "step": 47205 }, { "epoch": 3.721403232163973, "grad_norm": 0.42928874492645264, "learning_rate": 0.0005, "loss": 1.152055263519287, "step": 47206 }, { "epoch": 3.721482065431612, "grad_norm": 0.42741554975509644, "learning_rate": 0.0005, "loss": 1.1258914470672607, "step": 47207 }, { "epoch": 3.721560898699251, "grad_norm": 0.591360330581665, "learning_rate": 0.0005, "loss": 1.0874978303909302, "step": 47208 }, { "epoch": 3.72163973196689, "grad_norm": 0.5638784766197205, "learning_rate": 0.0005, "loss": 1.190371036529541, "step": 47209 }, { "epoch": 3.721718565234529, "grad_norm": 0.44078966975212097, "learning_rate": 0.0005, "loss": 0.9528850317001343, "step": 47210 }, { "epoch": 3.7217973985021677, "grad_norm": 0.414730966091156, "learning_rate": 0.0005, "loss": 1.1311150789260864, "step": 47211 }, { "epoch": 3.721876231769807, "grad_norm": 0.4553992450237274, "learning_rate": 0.0005, "loss": 1.2013945579528809, "step": 47212 }, { "epoch": 3.721955065037446, "grad_norm": 0.5928716659545898, "learning_rate": 0.0005, "loss": 1.1996722221374512, "step": 47213 }, { "epoch": 3.7220338983050847, "grad_norm": 0.5074148178100586, "learning_rate": 0.0005, "loss": 1.3201956748962402, "step": 47214 }, { "epoch": 3.722112731572724, "grad_norm": 0.39707332849502563, "learning_rate": 0.0005, "loss": 1.0545765161514282, "step": 47215 }, { "epoch": 3.7221915648403625, "grad_norm": 0.4397980272769928, "learning_rate": 0.0005, "loss": 1.2204720973968506, "step": 47216 }, { "epoch": 3.7222703981080016, "grad_norm": 0.425615519285202, "learning_rate": 0.0005, "loss": 1.0974491834640503, "step": 47217 }, { "epoch": 3.7223492313756408, "grad_norm": 0.41203808784484863, "learning_rate": 0.0005, "loss": 1.0487751960754395, "step": 47218 }, { "epoch": 3.7224280646432795, "grad_norm": 0.46085503697395325, "learning_rate": 0.0005, "loss": 1.0665348768234253, "step": 47219 }, { "epoch": 3.722506897910918, "grad_norm": 0.4354979395866394, "learning_rate": 0.0005, "loss": 1.1717941761016846, "step": 47220 }, { "epoch": 3.7225857311785573, "grad_norm": 0.4483086168766022, "learning_rate": 0.0005, "loss": 1.1414332389831543, "step": 47221 }, { "epoch": 3.7226645644461964, "grad_norm": 0.4796055555343628, "learning_rate": 0.0005, "loss": 1.2407397031784058, "step": 47222 }, { "epoch": 3.722743397713835, "grad_norm": 0.44643914699554443, "learning_rate": 0.0005, "loss": 1.035569429397583, "step": 47223 }, { "epoch": 3.7228222309814742, "grad_norm": 0.4501679241657257, "learning_rate": 0.0005, "loss": 1.2405447959899902, "step": 47224 }, { "epoch": 3.722901064249113, "grad_norm": 1.2577366828918457, "learning_rate": 0.0005, "loss": 1.2324459552764893, "step": 47225 }, { "epoch": 3.722979897516752, "grad_norm": 0.5429232716560364, "learning_rate": 0.0005, "loss": 1.1295809745788574, "step": 47226 }, { "epoch": 3.723058730784391, "grad_norm": 0.534382700920105, "learning_rate": 0.0005, "loss": 1.0036218166351318, "step": 47227 }, { "epoch": 3.72313756405203, "grad_norm": 0.48769229650497437, "learning_rate": 0.0005, "loss": 1.1593754291534424, "step": 47228 }, { "epoch": 3.723216397319669, "grad_norm": 0.41263869404792786, "learning_rate": 0.0005, "loss": 1.0142379999160767, "step": 47229 }, { "epoch": 3.7232952305873077, "grad_norm": 0.4621187448501587, "learning_rate": 0.0005, "loss": 1.0313146114349365, "step": 47230 }, { "epoch": 3.723374063854947, "grad_norm": 0.4890674948692322, "learning_rate": 0.0005, "loss": 1.3102285861968994, "step": 47231 }, { "epoch": 3.723452897122586, "grad_norm": 0.6320035457611084, "learning_rate": 0.0005, "loss": 1.249448299407959, "step": 47232 }, { "epoch": 3.7235317303902247, "grad_norm": 0.6439370512962341, "learning_rate": 0.0005, "loss": 1.1656951904296875, "step": 47233 }, { "epoch": 3.7236105636578634, "grad_norm": 0.5394852757453918, "learning_rate": 0.0005, "loss": 1.230010747909546, "step": 47234 }, { "epoch": 3.7236893969255025, "grad_norm": 0.5251389741897583, "learning_rate": 0.0005, "loss": 1.1053245067596436, "step": 47235 }, { "epoch": 3.7237682301931416, "grad_norm": 0.8071495294570923, "learning_rate": 0.0005, "loss": 1.1901700496673584, "step": 47236 }, { "epoch": 3.7238470634607803, "grad_norm": 0.5707032680511475, "learning_rate": 0.0005, "loss": 1.1027719974517822, "step": 47237 }, { "epoch": 3.7239258967284194, "grad_norm": 0.5615478754043579, "learning_rate": 0.0005, "loss": 1.1906676292419434, "step": 47238 }, { "epoch": 3.724004729996058, "grad_norm": 0.5237498879432678, "learning_rate": 0.0005, "loss": 1.2244288921356201, "step": 47239 }, { "epoch": 3.7240835632636973, "grad_norm": 0.5115581750869751, "learning_rate": 0.0005, "loss": 1.2270655632019043, "step": 47240 }, { "epoch": 3.7241623965313364, "grad_norm": 0.4623621106147766, "learning_rate": 0.0005, "loss": 1.1359667778015137, "step": 47241 }, { "epoch": 3.724241229798975, "grad_norm": 0.4494568109512329, "learning_rate": 0.0005, "loss": 1.197228193283081, "step": 47242 }, { "epoch": 3.7243200630666142, "grad_norm": 0.49053046107292175, "learning_rate": 0.0005, "loss": 1.2631096839904785, "step": 47243 }, { "epoch": 3.724398896334253, "grad_norm": 0.48105791211128235, "learning_rate": 0.0005, "loss": 1.1311299800872803, "step": 47244 }, { "epoch": 3.724477729601892, "grad_norm": 0.45758211612701416, "learning_rate": 0.0005, "loss": 1.115078330039978, "step": 47245 }, { "epoch": 3.724556562869531, "grad_norm": 0.7251213192939758, "learning_rate": 0.0005, "loss": 1.1655203104019165, "step": 47246 }, { "epoch": 3.72463539613717, "grad_norm": 0.47789424657821655, "learning_rate": 0.0005, "loss": 1.0954068899154663, "step": 47247 }, { "epoch": 3.7247142294048086, "grad_norm": 0.5380071997642517, "learning_rate": 0.0005, "loss": 1.2495232820510864, "step": 47248 }, { "epoch": 3.7247930626724477, "grad_norm": 0.4734378755092621, "learning_rate": 0.0005, "loss": 1.125869631767273, "step": 47249 }, { "epoch": 3.724871895940087, "grad_norm": 0.44279196858406067, "learning_rate": 0.0005, "loss": 1.1443994045257568, "step": 47250 }, { "epoch": 3.7249507292077255, "grad_norm": 0.521606981754303, "learning_rate": 0.0005, "loss": 1.2009203433990479, "step": 47251 }, { "epoch": 3.7250295624753647, "grad_norm": 0.5966733694076538, "learning_rate": 0.0005, "loss": 1.1844635009765625, "step": 47252 }, { "epoch": 3.7251083957430033, "grad_norm": 0.508224368095398, "learning_rate": 0.0005, "loss": 1.1361784934997559, "step": 47253 }, { "epoch": 3.7251872290106425, "grad_norm": 0.49447402358055115, "learning_rate": 0.0005, "loss": 1.0400294065475464, "step": 47254 }, { "epoch": 3.7252660622782816, "grad_norm": 0.5536226034164429, "learning_rate": 0.0005, "loss": 1.205960750579834, "step": 47255 }, { "epoch": 3.7253448955459203, "grad_norm": 0.42774659395217896, "learning_rate": 0.0005, "loss": 1.1114399433135986, "step": 47256 }, { "epoch": 3.7254237288135594, "grad_norm": 0.5583769083023071, "learning_rate": 0.0005, "loss": 1.1073203086853027, "step": 47257 }, { "epoch": 3.725502562081198, "grad_norm": 0.48123666644096375, "learning_rate": 0.0005, "loss": 1.0805402994155884, "step": 47258 }, { "epoch": 3.7255813953488373, "grad_norm": 0.4185565710067749, "learning_rate": 0.0005, "loss": 0.9722195863723755, "step": 47259 }, { "epoch": 3.7256602286164764, "grad_norm": 0.5147371888160706, "learning_rate": 0.0005, "loss": 1.0692546367645264, "step": 47260 }, { "epoch": 3.725739061884115, "grad_norm": 0.4442807137966156, "learning_rate": 0.0005, "loss": 1.182840347290039, "step": 47261 }, { "epoch": 3.7258178951517538, "grad_norm": 0.4482479989528656, "learning_rate": 0.0005, "loss": 1.2132930755615234, "step": 47262 }, { "epoch": 3.725896728419393, "grad_norm": 0.5944738388061523, "learning_rate": 0.0005, "loss": 1.1376981735229492, "step": 47263 }, { "epoch": 3.725975561687032, "grad_norm": 0.5736440420150757, "learning_rate": 0.0005, "loss": 1.1698540449142456, "step": 47264 }, { "epoch": 3.7260543949546707, "grad_norm": 0.4461713433265686, "learning_rate": 0.0005, "loss": 1.1135531663894653, "step": 47265 }, { "epoch": 3.72613322822231, "grad_norm": 0.4350888133049011, "learning_rate": 0.0005, "loss": 1.1705679893493652, "step": 47266 }, { "epoch": 3.7262120614899485, "grad_norm": 0.49060434103012085, "learning_rate": 0.0005, "loss": 1.071965217590332, "step": 47267 }, { "epoch": 3.7262908947575877, "grad_norm": 0.3932003974914551, "learning_rate": 0.0005, "loss": 0.9276092052459717, "step": 47268 }, { "epoch": 3.726369728025227, "grad_norm": 0.46000856161117554, "learning_rate": 0.0005, "loss": 1.0246529579162598, "step": 47269 }, { "epoch": 3.7264485612928655, "grad_norm": 0.5433597564697266, "learning_rate": 0.0005, "loss": 1.188971757888794, "step": 47270 }, { "epoch": 3.7265273945605046, "grad_norm": 0.3792213499546051, "learning_rate": 0.0005, "loss": 0.9423677325248718, "step": 47271 }, { "epoch": 3.7266062278281433, "grad_norm": 0.5830658078193665, "learning_rate": 0.0005, "loss": 1.0954514741897583, "step": 47272 }, { "epoch": 3.7266850610957825, "grad_norm": 0.473515123128891, "learning_rate": 0.0005, "loss": 1.1113214492797852, "step": 47273 }, { "epoch": 3.7267638943634216, "grad_norm": 0.4182055592536926, "learning_rate": 0.0005, "loss": 1.0925791263580322, "step": 47274 }, { "epoch": 3.7268427276310603, "grad_norm": 0.4372805953025818, "learning_rate": 0.0005, "loss": 1.0253229141235352, "step": 47275 }, { "epoch": 3.7269215608986994, "grad_norm": 0.4036836326122284, "learning_rate": 0.0005, "loss": 1.0807451009750366, "step": 47276 }, { "epoch": 3.727000394166338, "grad_norm": 0.42785805463790894, "learning_rate": 0.0005, "loss": 0.9375964999198914, "step": 47277 }, { "epoch": 3.7270792274339772, "grad_norm": 0.4816080927848816, "learning_rate": 0.0005, "loss": 1.1448335647583008, "step": 47278 }, { "epoch": 3.727158060701616, "grad_norm": 0.6486508846282959, "learning_rate": 0.0005, "loss": 1.0946695804595947, "step": 47279 }, { "epoch": 3.727236893969255, "grad_norm": 0.6210126280784607, "learning_rate": 0.0005, "loss": 1.1211649179458618, "step": 47280 }, { "epoch": 3.7273157272368937, "grad_norm": 0.44580528140068054, "learning_rate": 0.0005, "loss": 1.1850615739822388, "step": 47281 }, { "epoch": 3.727394560504533, "grad_norm": 0.5417002439498901, "learning_rate": 0.0005, "loss": 1.370849370956421, "step": 47282 }, { "epoch": 3.727473393772172, "grad_norm": 0.6439396142959595, "learning_rate": 0.0005, "loss": 1.2611777782440186, "step": 47283 }, { "epoch": 3.7275522270398107, "grad_norm": 0.45964449644088745, "learning_rate": 0.0005, "loss": 1.0844690799713135, "step": 47284 }, { "epoch": 3.72763106030745, "grad_norm": 0.3912813365459442, "learning_rate": 0.0005, "loss": 1.0711040496826172, "step": 47285 }, { "epoch": 3.7277098935750885, "grad_norm": 0.7785072326660156, "learning_rate": 0.0005, "loss": 1.3386285305023193, "step": 47286 }, { "epoch": 3.7277887268427277, "grad_norm": 0.5956647396087646, "learning_rate": 0.0005, "loss": 1.1605679988861084, "step": 47287 }, { "epoch": 3.727867560110367, "grad_norm": 0.4750281572341919, "learning_rate": 0.0005, "loss": 1.0814197063446045, "step": 47288 }, { "epoch": 3.7279463933780055, "grad_norm": 0.5076045393943787, "learning_rate": 0.0005, "loss": 1.099622130393982, "step": 47289 }, { "epoch": 3.7280252266456446, "grad_norm": 0.5012047290802002, "learning_rate": 0.0005, "loss": 1.30051851272583, "step": 47290 }, { "epoch": 3.7281040599132833, "grad_norm": 0.41345450282096863, "learning_rate": 0.0005, "loss": 1.0492480993270874, "step": 47291 }, { "epoch": 3.7281828931809224, "grad_norm": 0.6658355593681335, "learning_rate": 0.0005, "loss": 1.349002718925476, "step": 47292 }, { "epoch": 3.728261726448561, "grad_norm": 0.40664568543434143, "learning_rate": 0.0005, "loss": 1.089518427848816, "step": 47293 }, { "epoch": 3.7283405597162003, "grad_norm": 0.4771116375923157, "learning_rate": 0.0005, "loss": 1.1950318813323975, "step": 47294 }, { "epoch": 3.728419392983839, "grad_norm": 0.4758749306201935, "learning_rate": 0.0005, "loss": 1.1998567581176758, "step": 47295 }, { "epoch": 3.728498226251478, "grad_norm": 0.45758292078971863, "learning_rate": 0.0005, "loss": 1.3123716115951538, "step": 47296 }, { "epoch": 3.728577059519117, "grad_norm": 0.5819057822227478, "learning_rate": 0.0005, "loss": 1.3642868995666504, "step": 47297 }, { "epoch": 3.728655892786756, "grad_norm": 0.41707926988601685, "learning_rate": 0.0005, "loss": 1.3178541660308838, "step": 47298 }, { "epoch": 3.728734726054395, "grad_norm": 0.5956693887710571, "learning_rate": 0.0005, "loss": 1.3302608728408813, "step": 47299 }, { "epoch": 3.7288135593220337, "grad_norm": 0.47658222913742065, "learning_rate": 0.0005, "loss": 1.082478404045105, "step": 47300 }, { "epoch": 3.728892392589673, "grad_norm": 0.45580819249153137, "learning_rate": 0.0005, "loss": 1.152735948562622, "step": 47301 }, { "epoch": 3.728971225857312, "grad_norm": 0.4814494252204895, "learning_rate": 0.0005, "loss": 1.2154510021209717, "step": 47302 }, { "epoch": 3.7290500591249507, "grad_norm": 0.5362802743911743, "learning_rate": 0.0005, "loss": 1.2845878601074219, "step": 47303 }, { "epoch": 3.72912889239259, "grad_norm": 0.41623634099960327, "learning_rate": 0.0005, "loss": 1.0291715860366821, "step": 47304 }, { "epoch": 3.7292077256602285, "grad_norm": 0.45702216029167175, "learning_rate": 0.0005, "loss": 1.0112063884735107, "step": 47305 }, { "epoch": 3.7292865589278676, "grad_norm": 0.57700514793396, "learning_rate": 0.0005, "loss": 1.2006607055664062, "step": 47306 }, { "epoch": 3.729365392195507, "grad_norm": 0.49590250849723816, "learning_rate": 0.0005, "loss": 1.2031519412994385, "step": 47307 }, { "epoch": 3.7294442254631455, "grad_norm": 0.4741141200065613, "learning_rate": 0.0005, "loss": 1.2014808654785156, "step": 47308 }, { "epoch": 3.729523058730784, "grad_norm": 0.39126133918762207, "learning_rate": 0.0005, "loss": 1.1030900478363037, "step": 47309 }, { "epoch": 3.7296018919984233, "grad_norm": 0.4901859760284424, "learning_rate": 0.0005, "loss": 1.0641056299209595, "step": 47310 }, { "epoch": 3.7296807252660624, "grad_norm": 0.6150422692298889, "learning_rate": 0.0005, "loss": 1.0674806833267212, "step": 47311 }, { "epoch": 3.729759558533701, "grad_norm": 0.42686423659324646, "learning_rate": 0.0005, "loss": 1.0347366333007812, "step": 47312 }, { "epoch": 3.7298383918013402, "grad_norm": 0.5357900857925415, "learning_rate": 0.0005, "loss": 1.1741588115692139, "step": 47313 }, { "epoch": 3.729917225068979, "grad_norm": 0.657514750957489, "learning_rate": 0.0005, "loss": 1.188735008239746, "step": 47314 }, { "epoch": 3.729996058336618, "grad_norm": 0.49579280614852905, "learning_rate": 0.0005, "loss": 1.3064026832580566, "step": 47315 }, { "epoch": 3.730074891604257, "grad_norm": 0.46280863881111145, "learning_rate": 0.0005, "loss": 1.3074930906295776, "step": 47316 }, { "epoch": 3.730153724871896, "grad_norm": 2.10901141166687, "learning_rate": 0.0005, "loss": 1.4510605335235596, "step": 47317 }, { "epoch": 3.730232558139535, "grad_norm": 0.5008848905563354, "learning_rate": 0.0005, "loss": 1.2600412368774414, "step": 47318 }, { "epoch": 3.7303113914071737, "grad_norm": 0.4716068506240845, "learning_rate": 0.0005, "loss": 1.1141774654388428, "step": 47319 }, { "epoch": 3.730390224674813, "grad_norm": 0.38438817858695984, "learning_rate": 0.0005, "loss": 1.1517155170440674, "step": 47320 }, { "epoch": 3.730469057942452, "grad_norm": 0.4386889338493347, "learning_rate": 0.0005, "loss": 1.1769118309020996, "step": 47321 }, { "epoch": 3.7305478912100907, "grad_norm": 0.4654213786125183, "learning_rate": 0.0005, "loss": 1.287510871887207, "step": 47322 }, { "epoch": 3.7306267244777294, "grad_norm": 0.513588547706604, "learning_rate": 0.0005, "loss": 1.0993114709854126, "step": 47323 }, { "epoch": 3.7307055577453685, "grad_norm": 0.4220348000526428, "learning_rate": 0.0005, "loss": 1.0772794485092163, "step": 47324 }, { "epoch": 3.7307843910130076, "grad_norm": 0.448770672082901, "learning_rate": 0.0005, "loss": 1.0185283422470093, "step": 47325 }, { "epoch": 3.7308632242806463, "grad_norm": 0.5129749774932861, "learning_rate": 0.0005, "loss": 1.2893024682998657, "step": 47326 }, { "epoch": 3.7309420575482855, "grad_norm": 0.44258445501327515, "learning_rate": 0.0005, "loss": 1.054833173751831, "step": 47327 }, { "epoch": 3.731020890815924, "grad_norm": 0.5111307501792908, "learning_rate": 0.0005, "loss": 1.2145147323608398, "step": 47328 }, { "epoch": 3.7310997240835633, "grad_norm": 0.6175457835197449, "learning_rate": 0.0005, "loss": 1.3563727140426636, "step": 47329 }, { "epoch": 3.7311785573512024, "grad_norm": 0.792249321937561, "learning_rate": 0.0005, "loss": 1.2623602151870728, "step": 47330 }, { "epoch": 3.731257390618841, "grad_norm": 0.5125011205673218, "learning_rate": 0.0005, "loss": 1.082540512084961, "step": 47331 }, { "epoch": 3.7313362238864802, "grad_norm": 0.521169900894165, "learning_rate": 0.0005, "loss": 1.1015713214874268, "step": 47332 }, { "epoch": 3.731415057154119, "grad_norm": 0.5909882187843323, "learning_rate": 0.0005, "loss": 1.1843059062957764, "step": 47333 }, { "epoch": 3.731493890421758, "grad_norm": 0.5190386176109314, "learning_rate": 0.0005, "loss": 1.0599209070205688, "step": 47334 }, { "epoch": 3.731572723689397, "grad_norm": 0.5597490668296814, "learning_rate": 0.0005, "loss": 1.2198059558868408, "step": 47335 }, { "epoch": 3.731651556957036, "grad_norm": 0.6918147802352905, "learning_rate": 0.0005, "loss": 1.1354529857635498, "step": 47336 }, { "epoch": 3.7317303902246746, "grad_norm": 0.410281777381897, "learning_rate": 0.0005, "loss": 1.1197237968444824, "step": 47337 }, { "epoch": 3.7318092234923137, "grad_norm": 0.5633116960525513, "learning_rate": 0.0005, "loss": 1.1513478755950928, "step": 47338 }, { "epoch": 3.731888056759953, "grad_norm": 0.45792287588119507, "learning_rate": 0.0005, "loss": 1.1198896169662476, "step": 47339 }, { "epoch": 3.7319668900275915, "grad_norm": 0.4549163579940796, "learning_rate": 0.0005, "loss": 1.1018069982528687, "step": 47340 }, { "epoch": 3.7320457232952307, "grad_norm": 0.43534019589424133, "learning_rate": 0.0005, "loss": 1.0724873542785645, "step": 47341 }, { "epoch": 3.7321245565628693, "grad_norm": 0.6255948543548584, "learning_rate": 0.0005, "loss": 1.0324711799621582, "step": 47342 }, { "epoch": 3.7322033898305085, "grad_norm": 0.49967142939567566, "learning_rate": 0.0005, "loss": 1.0121798515319824, "step": 47343 }, { "epoch": 3.7322822230981476, "grad_norm": 0.41005298495292664, "learning_rate": 0.0005, "loss": 1.1372463703155518, "step": 47344 }, { "epoch": 3.7323610563657863, "grad_norm": 0.6369946002960205, "learning_rate": 0.0005, "loss": 1.35499906539917, "step": 47345 }, { "epoch": 3.7324398896334254, "grad_norm": 0.5107620358467102, "learning_rate": 0.0005, "loss": 1.2169816493988037, "step": 47346 }, { "epoch": 3.732518722901064, "grad_norm": 0.6465522646903992, "learning_rate": 0.0005, "loss": 1.1138560771942139, "step": 47347 }, { "epoch": 3.7325975561687033, "grad_norm": 0.5080697536468506, "learning_rate": 0.0005, "loss": 1.2855030298233032, "step": 47348 }, { "epoch": 3.7326763894363424, "grad_norm": 0.38875219225883484, "learning_rate": 0.0005, "loss": 1.1335679292678833, "step": 47349 }, { "epoch": 3.732755222703981, "grad_norm": 0.47774428129196167, "learning_rate": 0.0005, "loss": 1.189544916152954, "step": 47350 }, { "epoch": 3.7328340559716198, "grad_norm": 0.4806383550167084, "learning_rate": 0.0005, "loss": 1.1521104574203491, "step": 47351 }, { "epoch": 3.732912889239259, "grad_norm": 0.5165848731994629, "learning_rate": 0.0005, "loss": 1.2114448547363281, "step": 47352 }, { "epoch": 3.732991722506898, "grad_norm": 0.44174066185951233, "learning_rate": 0.0005, "loss": 1.128915786743164, "step": 47353 }, { "epoch": 3.7330705557745367, "grad_norm": 0.637988269329071, "learning_rate": 0.0005, "loss": 1.3318660259246826, "step": 47354 }, { "epoch": 3.733149389042176, "grad_norm": 0.4323825240135193, "learning_rate": 0.0005, "loss": 1.2159600257873535, "step": 47355 }, { "epoch": 3.7332282223098145, "grad_norm": 0.4846901595592499, "learning_rate": 0.0005, "loss": 1.1240370273590088, "step": 47356 }, { "epoch": 3.7333070555774537, "grad_norm": 0.42394396662712097, "learning_rate": 0.0005, "loss": 0.8922114968299866, "step": 47357 }, { "epoch": 3.733385888845093, "grad_norm": 0.6022504568099976, "learning_rate": 0.0005, "loss": 1.1359738111495972, "step": 47358 }, { "epoch": 3.7334647221127315, "grad_norm": 0.5100480914115906, "learning_rate": 0.0005, "loss": 1.1617848873138428, "step": 47359 }, { "epoch": 3.7335435553803706, "grad_norm": 0.49056077003479004, "learning_rate": 0.0005, "loss": 1.1521365642547607, "step": 47360 }, { "epoch": 3.7336223886480093, "grad_norm": 0.5578051805496216, "learning_rate": 0.0005, "loss": 1.2594835758209229, "step": 47361 }, { "epoch": 3.7337012219156485, "grad_norm": 0.5163983702659607, "learning_rate": 0.0005, "loss": 1.3049876689910889, "step": 47362 }, { "epoch": 3.7337800551832876, "grad_norm": 0.5165190696716309, "learning_rate": 0.0005, "loss": 1.0037696361541748, "step": 47363 }, { "epoch": 3.7338588884509263, "grad_norm": 0.6102160215377808, "learning_rate": 0.0005, "loss": 1.2996288537979126, "step": 47364 }, { "epoch": 3.733937721718565, "grad_norm": 0.4850224554538727, "learning_rate": 0.0005, "loss": 1.2555527687072754, "step": 47365 }, { "epoch": 3.734016554986204, "grad_norm": 0.5022338628768921, "learning_rate": 0.0005, "loss": 1.2602704763412476, "step": 47366 }, { "epoch": 3.7340953882538432, "grad_norm": 0.47684594988822937, "learning_rate": 0.0005, "loss": 1.1841952800750732, "step": 47367 }, { "epoch": 3.734174221521482, "grad_norm": 0.5090296864509583, "learning_rate": 0.0005, "loss": 1.1945525407791138, "step": 47368 }, { "epoch": 3.734253054789121, "grad_norm": 0.4812265634536743, "learning_rate": 0.0005, "loss": 1.2705395221710205, "step": 47369 }, { "epoch": 3.7343318880567598, "grad_norm": 0.5369144082069397, "learning_rate": 0.0005, "loss": 1.2679921388626099, "step": 47370 }, { "epoch": 3.734410721324399, "grad_norm": 0.5724110007286072, "learning_rate": 0.0005, "loss": 1.2643804550170898, "step": 47371 }, { "epoch": 3.734489554592038, "grad_norm": 0.4799180030822754, "learning_rate": 0.0005, "loss": 1.2429498434066772, "step": 47372 }, { "epoch": 3.7345683878596767, "grad_norm": 0.5683801174163818, "learning_rate": 0.0005, "loss": 1.2527203559875488, "step": 47373 }, { "epoch": 3.734647221127316, "grad_norm": 0.44224122166633606, "learning_rate": 0.0005, "loss": 1.1971256732940674, "step": 47374 }, { "epoch": 3.7347260543949545, "grad_norm": 0.4455844461917877, "learning_rate": 0.0005, "loss": 1.2396011352539062, "step": 47375 }, { "epoch": 3.7348048876625937, "grad_norm": 0.4967523515224457, "learning_rate": 0.0005, "loss": 1.1347036361694336, "step": 47376 }, { "epoch": 3.734883720930233, "grad_norm": 0.592180073261261, "learning_rate": 0.0005, "loss": 1.11027193069458, "step": 47377 }, { "epoch": 3.7349625541978715, "grad_norm": 0.5401397347450256, "learning_rate": 0.0005, "loss": 1.0447309017181396, "step": 47378 }, { "epoch": 3.73504138746551, "grad_norm": 0.4681609869003296, "learning_rate": 0.0005, "loss": 1.0851868391036987, "step": 47379 }, { "epoch": 3.7351202207331493, "grad_norm": 0.4293742775917053, "learning_rate": 0.0005, "loss": 0.8849039077758789, "step": 47380 }, { "epoch": 3.7351990540007884, "grad_norm": 0.4758363664150238, "learning_rate": 0.0005, "loss": 1.1265078783035278, "step": 47381 }, { "epoch": 3.735277887268427, "grad_norm": 0.5145348906517029, "learning_rate": 0.0005, "loss": 1.3214360475540161, "step": 47382 }, { "epoch": 3.7353567205360663, "grad_norm": 0.430963397026062, "learning_rate": 0.0005, "loss": 1.135373592376709, "step": 47383 }, { "epoch": 3.735435553803705, "grad_norm": 0.4938531816005707, "learning_rate": 0.0005, "loss": 0.980175256729126, "step": 47384 }, { "epoch": 3.735514387071344, "grad_norm": 0.5299792289733887, "learning_rate": 0.0005, "loss": 1.0778396129608154, "step": 47385 }, { "epoch": 3.7355932203389832, "grad_norm": 0.562032163143158, "learning_rate": 0.0005, "loss": 1.2897610664367676, "step": 47386 }, { "epoch": 3.735672053606622, "grad_norm": 0.5240828394889832, "learning_rate": 0.0005, "loss": 1.1887098550796509, "step": 47387 }, { "epoch": 3.735750886874261, "grad_norm": 0.5515140891075134, "learning_rate": 0.0005, "loss": 1.2399260997772217, "step": 47388 }, { "epoch": 3.7358297201418997, "grad_norm": 0.42944782972335815, "learning_rate": 0.0005, "loss": 1.1207159757614136, "step": 47389 }, { "epoch": 3.735908553409539, "grad_norm": 0.45799049735069275, "learning_rate": 0.0005, "loss": 1.0865373611450195, "step": 47390 }, { "epoch": 3.735987386677178, "grad_norm": 0.5108734965324402, "learning_rate": 0.0005, "loss": 1.348875880241394, "step": 47391 }, { "epoch": 3.7360662199448167, "grad_norm": 0.43932560086250305, "learning_rate": 0.0005, "loss": 1.0883066654205322, "step": 47392 }, { "epoch": 3.7361450532124554, "grad_norm": 0.5612978339195251, "learning_rate": 0.0005, "loss": 1.097287893295288, "step": 47393 }, { "epoch": 3.7362238864800945, "grad_norm": 0.5804040431976318, "learning_rate": 0.0005, "loss": 1.1621103286743164, "step": 47394 }, { "epoch": 3.7363027197477336, "grad_norm": 0.5012611746788025, "learning_rate": 0.0005, "loss": 1.1098573207855225, "step": 47395 }, { "epoch": 3.7363815530153723, "grad_norm": 0.5011104941368103, "learning_rate": 0.0005, "loss": 1.246167778968811, "step": 47396 }, { "epoch": 3.7364603862830115, "grad_norm": 0.5320253372192383, "learning_rate": 0.0005, "loss": 1.1871368885040283, "step": 47397 }, { "epoch": 3.73653921955065, "grad_norm": 0.5061941146850586, "learning_rate": 0.0005, "loss": 1.362126350402832, "step": 47398 }, { "epoch": 3.7366180528182893, "grad_norm": 0.4056590497493744, "learning_rate": 0.0005, "loss": 0.9009342789649963, "step": 47399 }, { "epoch": 3.7366968860859284, "grad_norm": 0.4796268045902252, "learning_rate": 0.0005, "loss": 1.1959351301193237, "step": 47400 }, { "epoch": 3.736775719353567, "grad_norm": 0.4492042064666748, "learning_rate": 0.0005, "loss": 1.184647798538208, "step": 47401 }, { "epoch": 3.7368545526212062, "grad_norm": 0.5656758546829224, "learning_rate": 0.0005, "loss": 1.4439690113067627, "step": 47402 }, { "epoch": 3.736933385888845, "grad_norm": 0.48494818806648254, "learning_rate": 0.0005, "loss": 1.0931274890899658, "step": 47403 }, { "epoch": 3.737012219156484, "grad_norm": 0.5931583046913147, "learning_rate": 0.0005, "loss": 1.3019804954528809, "step": 47404 }, { "epoch": 3.737091052424123, "grad_norm": 0.5500408411026001, "learning_rate": 0.0005, "loss": 1.2583234310150146, "step": 47405 }, { "epoch": 3.737169885691762, "grad_norm": 0.5013110637664795, "learning_rate": 0.0005, "loss": 1.1545178890228271, "step": 47406 }, { "epoch": 3.737248718959401, "grad_norm": 0.6035080552101135, "learning_rate": 0.0005, "loss": 1.4095181226730347, "step": 47407 }, { "epoch": 3.7373275522270397, "grad_norm": 0.5299050211906433, "learning_rate": 0.0005, "loss": 1.1746535301208496, "step": 47408 }, { "epoch": 3.737406385494679, "grad_norm": 0.49836304783821106, "learning_rate": 0.0005, "loss": 1.1617379188537598, "step": 47409 }, { "epoch": 3.7374852187623175, "grad_norm": 0.43127840757369995, "learning_rate": 0.0005, "loss": 1.0587972402572632, "step": 47410 }, { "epoch": 3.7375640520299567, "grad_norm": 0.4472877085208893, "learning_rate": 0.0005, "loss": 1.2669448852539062, "step": 47411 }, { "epoch": 3.7376428852975954, "grad_norm": 0.5270978808403015, "learning_rate": 0.0005, "loss": 1.1448949575424194, "step": 47412 }, { "epoch": 3.7377217185652345, "grad_norm": 0.772673487663269, "learning_rate": 0.0005, "loss": 1.2630131244659424, "step": 47413 }, { "epoch": 3.7378005518328736, "grad_norm": 0.506884753704071, "learning_rate": 0.0005, "loss": 1.2424688339233398, "step": 47414 }, { "epoch": 3.7378793851005123, "grad_norm": 0.4433650076389313, "learning_rate": 0.0005, "loss": 1.2558975219726562, "step": 47415 }, { "epoch": 3.7379582183681515, "grad_norm": 0.4917128384113312, "learning_rate": 0.0005, "loss": 1.2208521366119385, "step": 47416 }, { "epoch": 3.73803705163579, "grad_norm": 0.4653043746948242, "learning_rate": 0.0005, "loss": 1.0563015937805176, "step": 47417 }, { "epoch": 3.7381158849034293, "grad_norm": 0.44868943095207214, "learning_rate": 0.0005, "loss": 1.1338633298873901, "step": 47418 }, { "epoch": 3.7381947181710684, "grad_norm": 0.43224620819091797, "learning_rate": 0.0005, "loss": 1.0444777011871338, "step": 47419 }, { "epoch": 3.738273551438707, "grad_norm": 0.49426209926605225, "learning_rate": 0.0005, "loss": 1.15195894241333, "step": 47420 }, { "epoch": 3.7383523847063462, "grad_norm": 0.4601420760154724, "learning_rate": 0.0005, "loss": 1.1738436222076416, "step": 47421 }, { "epoch": 3.738431217973985, "grad_norm": 0.4060986042022705, "learning_rate": 0.0005, "loss": 0.9381939172744751, "step": 47422 }, { "epoch": 3.738510051241624, "grad_norm": 0.5772339105606079, "learning_rate": 0.0005, "loss": 1.0959508419036865, "step": 47423 }, { "epoch": 3.7385888845092627, "grad_norm": 0.455089271068573, "learning_rate": 0.0005, "loss": 1.242190957069397, "step": 47424 }, { "epoch": 3.738667717776902, "grad_norm": 0.49940550327301025, "learning_rate": 0.0005, "loss": 1.2563796043395996, "step": 47425 }, { "epoch": 3.7387465510445406, "grad_norm": 0.42852115631103516, "learning_rate": 0.0005, "loss": 1.0866317749023438, "step": 47426 }, { "epoch": 3.7388253843121797, "grad_norm": 0.5542433857917786, "learning_rate": 0.0005, "loss": 1.2333884239196777, "step": 47427 }, { "epoch": 3.738904217579819, "grad_norm": 0.4589719772338867, "learning_rate": 0.0005, "loss": 1.2043421268463135, "step": 47428 }, { "epoch": 3.7389830508474575, "grad_norm": 0.45646241307258606, "learning_rate": 0.0005, "loss": 1.1724393367767334, "step": 47429 }, { "epoch": 3.7390618841150967, "grad_norm": 0.5370256900787354, "learning_rate": 0.0005, "loss": 1.1412965059280396, "step": 47430 }, { "epoch": 3.7391407173827353, "grad_norm": 0.5632538795471191, "learning_rate": 0.0005, "loss": 1.2316442728042603, "step": 47431 }, { "epoch": 3.7392195506503745, "grad_norm": 0.5206412076950073, "learning_rate": 0.0005, "loss": 1.1316685676574707, "step": 47432 }, { "epoch": 3.7392983839180136, "grad_norm": 0.4728887677192688, "learning_rate": 0.0005, "loss": 1.2098956108093262, "step": 47433 }, { "epoch": 3.7393772171856523, "grad_norm": 0.5392858386039734, "learning_rate": 0.0005, "loss": 1.186598300933838, "step": 47434 }, { "epoch": 3.7394560504532914, "grad_norm": 0.3946378827095032, "learning_rate": 0.0005, "loss": 1.2452573776245117, "step": 47435 }, { "epoch": 3.73953488372093, "grad_norm": 0.3771267831325531, "learning_rate": 0.0005, "loss": 0.9969339966773987, "step": 47436 }, { "epoch": 3.7396137169885693, "grad_norm": 0.5383622050285339, "learning_rate": 0.0005, "loss": 1.0918442010879517, "step": 47437 }, { "epoch": 3.7396925502562084, "grad_norm": 0.4781157076358795, "learning_rate": 0.0005, "loss": 1.0913316011428833, "step": 47438 }, { "epoch": 3.739771383523847, "grad_norm": 0.433667927980423, "learning_rate": 0.0005, "loss": 1.0591013431549072, "step": 47439 }, { "epoch": 3.7398502167914858, "grad_norm": 0.4764501750469208, "learning_rate": 0.0005, "loss": 1.0397255420684814, "step": 47440 }, { "epoch": 3.739929050059125, "grad_norm": 0.49490591883659363, "learning_rate": 0.0005, "loss": 1.2393220663070679, "step": 47441 }, { "epoch": 3.740007883326764, "grad_norm": 0.5473254919052124, "learning_rate": 0.0005, "loss": 1.1889499425888062, "step": 47442 }, { "epoch": 3.7400867165944027, "grad_norm": 0.48242029547691345, "learning_rate": 0.0005, "loss": 1.260911464691162, "step": 47443 }, { "epoch": 3.740165549862042, "grad_norm": 0.47759103775024414, "learning_rate": 0.0005, "loss": 1.22165846824646, "step": 47444 }, { "epoch": 3.7402443831296805, "grad_norm": 0.5115062594413757, "learning_rate": 0.0005, "loss": 1.0910911560058594, "step": 47445 }, { "epoch": 3.7403232163973197, "grad_norm": 0.48480916023254395, "learning_rate": 0.0005, "loss": 1.113121509552002, "step": 47446 }, { "epoch": 3.740402049664959, "grad_norm": 0.4260360598564148, "learning_rate": 0.0005, "loss": 1.14186692237854, "step": 47447 }, { "epoch": 3.7404808829325975, "grad_norm": 0.5036731362342834, "learning_rate": 0.0005, "loss": 1.2008785009384155, "step": 47448 }, { "epoch": 3.7405597162002366, "grad_norm": 0.399440199136734, "learning_rate": 0.0005, "loss": 1.192350149154663, "step": 47449 }, { "epoch": 3.7406385494678753, "grad_norm": 0.44375109672546387, "learning_rate": 0.0005, "loss": 1.0956693887710571, "step": 47450 }, { "epoch": 3.7407173827355145, "grad_norm": 0.48540249466896057, "learning_rate": 0.0005, "loss": 1.0763285160064697, "step": 47451 }, { "epoch": 3.7407962160031536, "grad_norm": 0.4821794629096985, "learning_rate": 0.0005, "loss": 1.2151970863342285, "step": 47452 }, { "epoch": 3.7408750492707923, "grad_norm": 0.5299864411354065, "learning_rate": 0.0005, "loss": 1.1508656740188599, "step": 47453 }, { "epoch": 3.740953882538431, "grad_norm": 0.46434029936790466, "learning_rate": 0.0005, "loss": 1.1116032600402832, "step": 47454 }, { "epoch": 3.74103271580607, "grad_norm": 0.4360203444957733, "learning_rate": 0.0005, "loss": 1.3293325901031494, "step": 47455 }, { "epoch": 3.7411115490737092, "grad_norm": 0.5979428887367249, "learning_rate": 0.0005, "loss": 1.1240888833999634, "step": 47456 }, { "epoch": 3.741190382341348, "grad_norm": 0.45254582166671753, "learning_rate": 0.0005, "loss": 1.1197781562805176, "step": 47457 }, { "epoch": 3.741269215608987, "grad_norm": 0.3917075991630554, "learning_rate": 0.0005, "loss": 0.9742978811264038, "step": 47458 }, { "epoch": 3.7413480488766258, "grad_norm": 0.4334771931171417, "learning_rate": 0.0005, "loss": 1.087822675704956, "step": 47459 }, { "epoch": 3.741426882144265, "grad_norm": 0.5292410850524902, "learning_rate": 0.0005, "loss": 1.1383053064346313, "step": 47460 }, { "epoch": 3.741505715411904, "grad_norm": 0.5622109174728394, "learning_rate": 0.0005, "loss": 1.0637725591659546, "step": 47461 }, { "epoch": 3.7415845486795427, "grad_norm": 0.551202654838562, "learning_rate": 0.0005, "loss": 1.268051266670227, "step": 47462 }, { "epoch": 3.741663381947182, "grad_norm": 0.4157787561416626, "learning_rate": 0.0005, "loss": 1.1226094961166382, "step": 47463 }, { "epoch": 3.7417422152148205, "grad_norm": 0.43690013885498047, "learning_rate": 0.0005, "loss": 0.9931702613830566, "step": 47464 }, { "epoch": 3.7418210484824597, "grad_norm": 0.4339562654495239, "learning_rate": 0.0005, "loss": 1.1095404624938965, "step": 47465 }, { "epoch": 3.741899881750099, "grad_norm": 0.4063182473182678, "learning_rate": 0.0005, "loss": 1.0375028848648071, "step": 47466 }, { "epoch": 3.7419787150177375, "grad_norm": 0.39193177223205566, "learning_rate": 0.0005, "loss": 1.0479612350463867, "step": 47467 }, { "epoch": 3.742057548285376, "grad_norm": 0.508849561214447, "learning_rate": 0.0005, "loss": 1.014690637588501, "step": 47468 }, { "epoch": 3.7421363815530153, "grad_norm": 0.3982994258403778, "learning_rate": 0.0005, "loss": 0.9373323321342468, "step": 47469 }, { "epoch": 3.7422152148206544, "grad_norm": 0.6746386289596558, "learning_rate": 0.0005, "loss": 1.159627079963684, "step": 47470 }, { "epoch": 3.742294048088293, "grad_norm": 0.5020285248756409, "learning_rate": 0.0005, "loss": 1.1496998071670532, "step": 47471 }, { "epoch": 3.7423728813559323, "grad_norm": 0.5058691501617432, "learning_rate": 0.0005, "loss": 1.1843609809875488, "step": 47472 }, { "epoch": 3.742451714623571, "grad_norm": 0.45417702198028564, "learning_rate": 0.0005, "loss": 1.1863404512405396, "step": 47473 }, { "epoch": 3.74253054789121, "grad_norm": 0.554619312286377, "learning_rate": 0.0005, "loss": 1.1681681871414185, "step": 47474 }, { "epoch": 3.7426093811588492, "grad_norm": 0.5954115390777588, "learning_rate": 0.0005, "loss": 1.3104759454727173, "step": 47475 }, { "epoch": 3.742688214426488, "grad_norm": 0.5757114887237549, "learning_rate": 0.0005, "loss": 1.1558645963668823, "step": 47476 }, { "epoch": 3.742767047694127, "grad_norm": 0.5965192317962646, "learning_rate": 0.0005, "loss": 1.342221736907959, "step": 47477 }, { "epoch": 3.7428458809617657, "grad_norm": 0.7482525706291199, "learning_rate": 0.0005, "loss": 1.1289896965026855, "step": 47478 }, { "epoch": 3.742924714229405, "grad_norm": 0.5582603812217712, "learning_rate": 0.0005, "loss": 1.1159358024597168, "step": 47479 }, { "epoch": 3.743003547497044, "grad_norm": 0.4222712218761444, "learning_rate": 0.0005, "loss": 1.1806731224060059, "step": 47480 }, { "epoch": 3.7430823807646827, "grad_norm": 0.40398970246315, "learning_rate": 0.0005, "loss": 1.117964506149292, "step": 47481 }, { "epoch": 3.7431612140323214, "grad_norm": 0.4464205205440521, "learning_rate": 0.0005, "loss": 1.1196389198303223, "step": 47482 }, { "epoch": 3.7432400472999605, "grad_norm": 0.8299656510353088, "learning_rate": 0.0005, "loss": 1.1442952156066895, "step": 47483 }, { "epoch": 3.7433188805675996, "grad_norm": 0.42317909002304077, "learning_rate": 0.0005, "loss": 1.138631820678711, "step": 47484 }, { "epoch": 3.7433977138352383, "grad_norm": 0.5214725732803345, "learning_rate": 0.0005, "loss": 1.1810712814331055, "step": 47485 }, { "epoch": 3.7434765471028775, "grad_norm": 0.4907049834728241, "learning_rate": 0.0005, "loss": 1.1489564180374146, "step": 47486 }, { "epoch": 3.743555380370516, "grad_norm": 0.5029809474945068, "learning_rate": 0.0005, "loss": 1.2533957958221436, "step": 47487 }, { "epoch": 3.7436342136381553, "grad_norm": 0.5487090349197388, "learning_rate": 0.0005, "loss": 1.232132911682129, "step": 47488 }, { "epoch": 3.7437130469057944, "grad_norm": 0.5305987000465393, "learning_rate": 0.0005, "loss": 1.3691166639328003, "step": 47489 }, { "epoch": 3.743791880173433, "grad_norm": 0.6431591510772705, "learning_rate": 0.0005, "loss": 1.2793020009994507, "step": 47490 }, { "epoch": 3.7438707134410723, "grad_norm": 0.5520124435424805, "learning_rate": 0.0005, "loss": 1.2690318822860718, "step": 47491 }, { "epoch": 3.743949546708711, "grad_norm": 0.6447019577026367, "learning_rate": 0.0005, "loss": 1.1159660816192627, "step": 47492 }, { "epoch": 3.74402837997635, "grad_norm": 0.3708040714263916, "learning_rate": 0.0005, "loss": 1.0230746269226074, "step": 47493 }, { "epoch": 3.744107213243989, "grad_norm": 0.550919234752655, "learning_rate": 0.0005, "loss": 1.09621262550354, "step": 47494 }, { "epoch": 3.744186046511628, "grad_norm": 0.4735809862613678, "learning_rate": 0.0005, "loss": 1.1486449241638184, "step": 47495 }, { "epoch": 3.7442648797792666, "grad_norm": 0.46632155776023865, "learning_rate": 0.0005, "loss": 1.0071641206741333, "step": 47496 }, { "epoch": 3.7443437130469057, "grad_norm": 0.43801000714302063, "learning_rate": 0.0005, "loss": 1.0740141868591309, "step": 47497 }, { "epoch": 3.744422546314545, "grad_norm": 0.497147798538208, "learning_rate": 0.0005, "loss": 1.1619442701339722, "step": 47498 }, { "epoch": 3.7445013795821835, "grad_norm": 0.5107156038284302, "learning_rate": 0.0005, "loss": 1.13044273853302, "step": 47499 }, { "epoch": 3.7445802128498227, "grad_norm": 0.4400211572647095, "learning_rate": 0.0005, "loss": 1.1712512969970703, "step": 47500 }, { "epoch": 3.7446590461174614, "grad_norm": 0.6533091068267822, "learning_rate": 0.0005, "loss": 1.2467321157455444, "step": 47501 }, { "epoch": 3.7447378793851005, "grad_norm": 0.5263229012489319, "learning_rate": 0.0005, "loss": 1.158107042312622, "step": 47502 }, { "epoch": 3.7448167126527396, "grad_norm": 0.41593417525291443, "learning_rate": 0.0005, "loss": 1.0163764953613281, "step": 47503 }, { "epoch": 3.7448955459203783, "grad_norm": 0.4661562144756317, "learning_rate": 0.0005, "loss": 1.1327126026153564, "step": 47504 }, { "epoch": 3.7449743791880175, "grad_norm": 0.8468481302261353, "learning_rate": 0.0005, "loss": 1.3405194282531738, "step": 47505 }, { "epoch": 3.745053212455656, "grad_norm": 0.5021398663520813, "learning_rate": 0.0005, "loss": 1.2869598865509033, "step": 47506 }, { "epoch": 3.7451320457232953, "grad_norm": 0.5102289319038391, "learning_rate": 0.0005, "loss": 1.3014271259307861, "step": 47507 }, { "epoch": 3.7452108789909344, "grad_norm": 0.4358724355697632, "learning_rate": 0.0005, "loss": 1.2042102813720703, "step": 47508 }, { "epoch": 3.745289712258573, "grad_norm": 0.6449522972106934, "learning_rate": 0.0005, "loss": 1.3580727577209473, "step": 47509 }, { "epoch": 3.745368545526212, "grad_norm": 0.48779791593551636, "learning_rate": 0.0005, "loss": 1.1106919050216675, "step": 47510 }, { "epoch": 3.745447378793851, "grad_norm": 0.48082566261291504, "learning_rate": 0.0005, "loss": 1.366896629333496, "step": 47511 }, { "epoch": 3.74552621206149, "grad_norm": 0.5388258695602417, "learning_rate": 0.0005, "loss": 1.3081629276275635, "step": 47512 }, { "epoch": 3.7456050453291287, "grad_norm": 0.6072476506233215, "learning_rate": 0.0005, "loss": 1.2019834518432617, "step": 47513 }, { "epoch": 3.745683878596768, "grad_norm": 0.4739116132259369, "learning_rate": 0.0005, "loss": 1.0823452472686768, "step": 47514 }, { "epoch": 3.7457627118644066, "grad_norm": 0.4560260772705078, "learning_rate": 0.0005, "loss": 1.1478010416030884, "step": 47515 }, { "epoch": 3.7458415451320457, "grad_norm": 0.5684524178504944, "learning_rate": 0.0005, "loss": 1.221045732498169, "step": 47516 }, { "epoch": 3.745920378399685, "grad_norm": 0.49027350544929504, "learning_rate": 0.0005, "loss": 1.1942086219787598, "step": 47517 }, { "epoch": 3.7459992116673235, "grad_norm": 0.47970908880233765, "learning_rate": 0.0005, "loss": 1.1205793619155884, "step": 47518 }, { "epoch": 3.7460780449349627, "grad_norm": 0.48143404722213745, "learning_rate": 0.0005, "loss": 1.179453730583191, "step": 47519 }, { "epoch": 3.7461568782026013, "grad_norm": 0.55722576379776, "learning_rate": 0.0005, "loss": 1.2838549613952637, "step": 47520 }, { "epoch": 3.7462357114702405, "grad_norm": 0.5827702283859253, "learning_rate": 0.0005, "loss": 1.1284611225128174, "step": 47521 }, { "epoch": 3.7463145447378796, "grad_norm": 0.479408860206604, "learning_rate": 0.0005, "loss": 1.1280583143234253, "step": 47522 }, { "epoch": 3.7463933780055183, "grad_norm": 0.5612072348594666, "learning_rate": 0.0005, "loss": 1.1778254508972168, "step": 47523 }, { "epoch": 3.7464722112731574, "grad_norm": 0.5146853923797607, "learning_rate": 0.0005, "loss": 1.106473684310913, "step": 47524 }, { "epoch": 3.746551044540796, "grad_norm": 1.4518276453018188, "learning_rate": 0.0005, "loss": 1.1228532791137695, "step": 47525 }, { "epoch": 3.7466298778084353, "grad_norm": 0.4990561902523041, "learning_rate": 0.0005, "loss": 1.011889934539795, "step": 47526 }, { "epoch": 3.746708711076074, "grad_norm": 0.5673588514328003, "learning_rate": 0.0005, "loss": 1.0726593732833862, "step": 47527 }, { "epoch": 3.746787544343713, "grad_norm": 0.4228697121143341, "learning_rate": 0.0005, "loss": 1.1827512979507446, "step": 47528 }, { "epoch": 3.7468663776113518, "grad_norm": 0.49582698941230774, "learning_rate": 0.0005, "loss": 1.1816624402999878, "step": 47529 }, { "epoch": 3.746945210878991, "grad_norm": 0.7378407120704651, "learning_rate": 0.0005, "loss": 1.081579327583313, "step": 47530 }, { "epoch": 3.74702404414663, "grad_norm": 0.524076521396637, "learning_rate": 0.0005, "loss": 1.0637943744659424, "step": 47531 }, { "epoch": 3.7471028774142687, "grad_norm": 0.4908207058906555, "learning_rate": 0.0005, "loss": 1.0352083444595337, "step": 47532 }, { "epoch": 3.747181710681908, "grad_norm": 0.44605886936187744, "learning_rate": 0.0005, "loss": 1.0674415826797485, "step": 47533 }, { "epoch": 3.7472605439495466, "grad_norm": 0.4579826295375824, "learning_rate": 0.0005, "loss": 1.3712221384048462, "step": 47534 }, { "epoch": 3.7473393772171857, "grad_norm": 0.41835033893585205, "learning_rate": 0.0005, "loss": 1.0246777534484863, "step": 47535 }, { "epoch": 3.747418210484825, "grad_norm": 0.48808538913726807, "learning_rate": 0.0005, "loss": 1.160584568977356, "step": 47536 }, { "epoch": 3.7474970437524635, "grad_norm": 0.4862174093723297, "learning_rate": 0.0005, "loss": 1.3792812824249268, "step": 47537 }, { "epoch": 3.7475758770201026, "grad_norm": 0.5534409880638123, "learning_rate": 0.0005, "loss": 1.1499403715133667, "step": 47538 }, { "epoch": 3.7476547102877413, "grad_norm": 0.5931738615036011, "learning_rate": 0.0005, "loss": 1.2547416687011719, "step": 47539 }, { "epoch": 3.7477335435553805, "grad_norm": 0.3947208821773529, "learning_rate": 0.0005, "loss": 0.9718997478485107, "step": 47540 }, { "epoch": 3.747812376823019, "grad_norm": 0.5238825082778931, "learning_rate": 0.0005, "loss": 1.2184617519378662, "step": 47541 }, { "epoch": 3.7478912100906583, "grad_norm": 0.5103442072868347, "learning_rate": 0.0005, "loss": 1.305619716644287, "step": 47542 }, { "epoch": 3.747970043358297, "grad_norm": 0.42178982496261597, "learning_rate": 0.0005, "loss": 1.156572699546814, "step": 47543 }, { "epoch": 3.748048876625936, "grad_norm": 0.5179155468940735, "learning_rate": 0.0005, "loss": 1.1578673124313354, "step": 47544 }, { "epoch": 3.7481277098935752, "grad_norm": 0.4337467849254608, "learning_rate": 0.0005, "loss": 1.0416404008865356, "step": 47545 }, { "epoch": 3.748206543161214, "grad_norm": 0.4911755919456482, "learning_rate": 0.0005, "loss": 1.3572380542755127, "step": 47546 }, { "epoch": 3.748285376428853, "grad_norm": 0.6853912472724915, "learning_rate": 0.0005, "loss": 1.0979641675949097, "step": 47547 }, { "epoch": 3.7483642096964918, "grad_norm": 0.468976765871048, "learning_rate": 0.0005, "loss": 1.0687520503997803, "step": 47548 }, { "epoch": 3.748443042964131, "grad_norm": 0.4746130108833313, "learning_rate": 0.0005, "loss": 1.1812670230865479, "step": 47549 }, { "epoch": 3.74852187623177, "grad_norm": 0.4655328094959259, "learning_rate": 0.0005, "loss": 1.1871036291122437, "step": 47550 }, { "epoch": 3.7486007094994087, "grad_norm": 0.558415412902832, "learning_rate": 0.0005, "loss": 1.1080557107925415, "step": 47551 }, { "epoch": 3.748679542767048, "grad_norm": 0.48089590668678284, "learning_rate": 0.0005, "loss": 1.1047377586364746, "step": 47552 }, { "epoch": 3.7487583760346865, "grad_norm": 0.5929124355316162, "learning_rate": 0.0005, "loss": 1.2622753381729126, "step": 47553 }, { "epoch": 3.7488372093023257, "grad_norm": 0.4177364706993103, "learning_rate": 0.0005, "loss": 1.0737375020980835, "step": 47554 }, { "epoch": 3.748916042569965, "grad_norm": 0.4894011914730072, "learning_rate": 0.0005, "loss": 1.1066148281097412, "step": 47555 }, { "epoch": 3.7489948758376035, "grad_norm": 0.557633638381958, "learning_rate": 0.0005, "loss": 1.2815818786621094, "step": 47556 }, { "epoch": 3.749073709105242, "grad_norm": 0.4706251323223114, "learning_rate": 0.0005, "loss": 1.2977640628814697, "step": 47557 }, { "epoch": 3.7491525423728813, "grad_norm": 0.4710349142551422, "learning_rate": 0.0005, "loss": 1.2358002662658691, "step": 47558 }, { "epoch": 3.7492313756405204, "grad_norm": 0.536386251449585, "learning_rate": 0.0005, "loss": 1.1764863729476929, "step": 47559 }, { "epoch": 3.749310208908159, "grad_norm": 0.45549923181533813, "learning_rate": 0.0005, "loss": 1.0820444822311401, "step": 47560 }, { "epoch": 3.7493890421757983, "grad_norm": 0.5397809147834778, "learning_rate": 0.0005, "loss": 1.0632262229919434, "step": 47561 }, { "epoch": 3.749467875443437, "grad_norm": 0.6874281167984009, "learning_rate": 0.0005, "loss": 1.332545518875122, "step": 47562 }, { "epoch": 3.749546708711076, "grad_norm": 0.43990039825439453, "learning_rate": 0.0005, "loss": 1.0949081182479858, "step": 47563 }, { "epoch": 3.7496255419787152, "grad_norm": 0.491886705160141, "learning_rate": 0.0005, "loss": 1.1413074731826782, "step": 47564 }, { "epoch": 3.749704375246354, "grad_norm": 0.4576275050640106, "learning_rate": 0.0005, "loss": 1.118096113204956, "step": 47565 }, { "epoch": 3.749783208513993, "grad_norm": 0.5896823406219482, "learning_rate": 0.0005, "loss": 1.1640973091125488, "step": 47566 }, { "epoch": 3.7498620417816317, "grad_norm": 0.5050496459007263, "learning_rate": 0.0005, "loss": 1.163546085357666, "step": 47567 }, { "epoch": 3.749940875049271, "grad_norm": 0.5075345039367676, "learning_rate": 0.0005, "loss": 1.207072377204895, "step": 47568 }, { "epoch": 3.75001970831691, "grad_norm": 0.39072370529174805, "learning_rate": 0.0005, "loss": 1.123434066772461, "step": 47569 }, { "epoch": 3.7500985415845487, "grad_norm": 0.5096482038497925, "learning_rate": 0.0005, "loss": 1.386098861694336, "step": 47570 }, { "epoch": 3.7501773748521874, "grad_norm": 0.4438577890396118, "learning_rate": 0.0005, "loss": 1.215500831604004, "step": 47571 }, { "epoch": 3.7502562081198265, "grad_norm": 0.5027350187301636, "learning_rate": 0.0005, "loss": 1.2599380016326904, "step": 47572 }, { "epoch": 3.7503350413874657, "grad_norm": 0.4785446226596832, "learning_rate": 0.0005, "loss": 1.0651668310165405, "step": 47573 }, { "epoch": 3.7504138746551043, "grad_norm": 0.5649831891059875, "learning_rate": 0.0005, "loss": 1.0355833768844604, "step": 47574 }, { "epoch": 3.7504927079227435, "grad_norm": 0.6115285754203796, "learning_rate": 0.0005, "loss": 1.3460612297058105, "step": 47575 }, { "epoch": 3.750571541190382, "grad_norm": 0.39722728729248047, "learning_rate": 0.0005, "loss": 0.9444555044174194, "step": 47576 }, { "epoch": 3.7506503744580213, "grad_norm": 0.49530282616615295, "learning_rate": 0.0005, "loss": 1.1860342025756836, "step": 47577 }, { "epoch": 3.7507292077256604, "grad_norm": 0.5203745365142822, "learning_rate": 0.0005, "loss": 1.2409861087799072, "step": 47578 }, { "epoch": 3.750808040993299, "grad_norm": 0.5551074743270874, "learning_rate": 0.0005, "loss": 1.0755655765533447, "step": 47579 }, { "epoch": 3.7508868742609383, "grad_norm": 0.6227825284004211, "learning_rate": 0.0005, "loss": 1.2628216743469238, "step": 47580 }, { "epoch": 3.750965707528577, "grad_norm": 0.48612621426582336, "learning_rate": 0.0005, "loss": 1.1331262588500977, "step": 47581 }, { "epoch": 3.751044540796216, "grad_norm": 0.4715222418308258, "learning_rate": 0.0005, "loss": 1.1925333738327026, "step": 47582 }, { "epoch": 3.751123374063855, "grad_norm": 0.5452656149864197, "learning_rate": 0.0005, "loss": 1.33870267868042, "step": 47583 }, { "epoch": 3.751202207331494, "grad_norm": 0.5586121678352356, "learning_rate": 0.0005, "loss": 1.193307638168335, "step": 47584 }, { "epoch": 3.7512810405991326, "grad_norm": 0.5948988199234009, "learning_rate": 0.0005, "loss": 1.1931688785552979, "step": 47585 }, { "epoch": 3.7513598738667717, "grad_norm": 0.5060198307037354, "learning_rate": 0.0005, "loss": 1.2851736545562744, "step": 47586 }, { "epoch": 3.751438707134411, "grad_norm": 0.4601762294769287, "learning_rate": 0.0005, "loss": 1.0814521312713623, "step": 47587 }, { "epoch": 3.7515175404020495, "grad_norm": 0.664840042591095, "learning_rate": 0.0005, "loss": 1.1785252094268799, "step": 47588 }, { "epoch": 3.7515963736696887, "grad_norm": 0.4945669174194336, "learning_rate": 0.0005, "loss": 1.3805726766586304, "step": 47589 }, { "epoch": 3.7516752069373274, "grad_norm": 0.5843384265899658, "learning_rate": 0.0005, "loss": 1.188965082168579, "step": 47590 }, { "epoch": 3.7517540402049665, "grad_norm": 0.6464354395866394, "learning_rate": 0.0005, "loss": 1.2914965152740479, "step": 47591 }, { "epoch": 3.7518328734726056, "grad_norm": 0.5324420928955078, "learning_rate": 0.0005, "loss": 1.1397361755371094, "step": 47592 }, { "epoch": 3.7519117067402443, "grad_norm": 0.48601511120796204, "learning_rate": 0.0005, "loss": 1.1323113441467285, "step": 47593 }, { "epoch": 3.7519905400078835, "grad_norm": 0.49505677819252014, "learning_rate": 0.0005, "loss": 1.1111050844192505, "step": 47594 }, { "epoch": 3.752069373275522, "grad_norm": 0.4979972541332245, "learning_rate": 0.0005, "loss": 1.1493016481399536, "step": 47595 }, { "epoch": 3.7521482065431613, "grad_norm": 0.5813664793968201, "learning_rate": 0.0005, "loss": 1.1550649404525757, "step": 47596 }, { "epoch": 3.7522270398108004, "grad_norm": 0.5086292028427124, "learning_rate": 0.0005, "loss": 1.1017825603485107, "step": 47597 }, { "epoch": 3.752305873078439, "grad_norm": 0.5689747929573059, "learning_rate": 0.0005, "loss": 1.1776291131973267, "step": 47598 }, { "epoch": 3.752384706346078, "grad_norm": 0.47401660680770874, "learning_rate": 0.0005, "loss": 1.202873706817627, "step": 47599 }, { "epoch": 3.752463539613717, "grad_norm": 0.5449643731117249, "learning_rate": 0.0005, "loss": 1.1991024017333984, "step": 47600 }, { "epoch": 3.752542372881356, "grad_norm": 0.5367671847343445, "learning_rate": 0.0005, "loss": 1.0618596076965332, "step": 47601 }, { "epoch": 3.7526212061489947, "grad_norm": 0.5329158902168274, "learning_rate": 0.0005, "loss": 1.4182926416397095, "step": 47602 }, { "epoch": 3.752700039416634, "grad_norm": 0.40361541509628296, "learning_rate": 0.0005, "loss": 1.183098316192627, "step": 47603 }, { "epoch": 3.7527788726842726, "grad_norm": 0.7961126565933228, "learning_rate": 0.0005, "loss": 1.345154047012329, "step": 47604 }, { "epoch": 3.7528577059519117, "grad_norm": 0.4366465210914612, "learning_rate": 0.0005, "loss": 1.2270870208740234, "step": 47605 }, { "epoch": 3.752936539219551, "grad_norm": 0.501723051071167, "learning_rate": 0.0005, "loss": 1.0816590785980225, "step": 47606 }, { "epoch": 3.7530153724871895, "grad_norm": 0.45620930194854736, "learning_rate": 0.0005, "loss": 1.0757606029510498, "step": 47607 }, { "epoch": 3.7530942057548287, "grad_norm": 0.595420777797699, "learning_rate": 0.0005, "loss": 1.0284698009490967, "step": 47608 }, { "epoch": 3.7531730390224673, "grad_norm": 0.5203373432159424, "learning_rate": 0.0005, "loss": 1.1795578002929688, "step": 47609 }, { "epoch": 3.7532518722901065, "grad_norm": 0.5175011157989502, "learning_rate": 0.0005, "loss": 1.1469697952270508, "step": 47610 }, { "epoch": 3.7533307055577456, "grad_norm": 0.6436375379562378, "learning_rate": 0.0005, "loss": 1.1399619579315186, "step": 47611 }, { "epoch": 3.7534095388253843, "grad_norm": 0.5281885266304016, "learning_rate": 0.0005, "loss": 1.262667179107666, "step": 47612 }, { "epoch": 3.753488372093023, "grad_norm": 0.49638471007347107, "learning_rate": 0.0005, "loss": 1.206039309501648, "step": 47613 }, { "epoch": 3.753567205360662, "grad_norm": 0.486500084400177, "learning_rate": 0.0005, "loss": 1.1124320030212402, "step": 47614 }, { "epoch": 3.7536460386283013, "grad_norm": 0.3931795656681061, "learning_rate": 0.0005, "loss": 0.99029541015625, "step": 47615 }, { "epoch": 3.75372487189594, "grad_norm": 1.3824870586395264, "learning_rate": 0.0005, "loss": 1.2434383630752563, "step": 47616 }, { "epoch": 3.753803705163579, "grad_norm": 0.5662159323692322, "learning_rate": 0.0005, "loss": 1.1799280643463135, "step": 47617 }, { "epoch": 3.7538825384312178, "grad_norm": 0.5826675295829773, "learning_rate": 0.0005, "loss": 1.200887680053711, "step": 47618 }, { "epoch": 3.753961371698857, "grad_norm": 0.5648913979530334, "learning_rate": 0.0005, "loss": 1.2164809703826904, "step": 47619 }, { "epoch": 3.754040204966496, "grad_norm": 0.4673388600349426, "learning_rate": 0.0005, "loss": 1.1355805397033691, "step": 47620 }, { "epoch": 3.7541190382341347, "grad_norm": 0.5032762289047241, "learning_rate": 0.0005, "loss": 1.263030767440796, "step": 47621 }, { "epoch": 3.754197871501774, "grad_norm": 0.4514443278312683, "learning_rate": 0.0005, "loss": 1.1616935729980469, "step": 47622 }, { "epoch": 3.7542767047694126, "grad_norm": 0.7455868721008301, "learning_rate": 0.0005, "loss": 1.4227333068847656, "step": 47623 }, { "epoch": 3.7543555380370517, "grad_norm": 0.4210895299911499, "learning_rate": 0.0005, "loss": 1.1069289445877075, "step": 47624 }, { "epoch": 3.754434371304691, "grad_norm": 0.6921221017837524, "learning_rate": 0.0005, "loss": 1.227684497833252, "step": 47625 }, { "epoch": 3.7545132045723295, "grad_norm": 0.5885264873504639, "learning_rate": 0.0005, "loss": 1.1918747425079346, "step": 47626 }, { "epoch": 3.754592037839968, "grad_norm": 0.4538912773132324, "learning_rate": 0.0005, "loss": 0.9712061285972595, "step": 47627 }, { "epoch": 3.7546708711076073, "grad_norm": 0.44172340631484985, "learning_rate": 0.0005, "loss": 1.1612929105758667, "step": 47628 }, { "epoch": 3.7547497043752465, "grad_norm": 0.5659976601600647, "learning_rate": 0.0005, "loss": 1.1498688459396362, "step": 47629 }, { "epoch": 3.754828537642885, "grad_norm": 0.4783715605735779, "learning_rate": 0.0005, "loss": 1.1828093528747559, "step": 47630 }, { "epoch": 3.7549073709105243, "grad_norm": 0.46950578689575195, "learning_rate": 0.0005, "loss": 1.0761184692382812, "step": 47631 }, { "epoch": 3.754986204178163, "grad_norm": 0.6029329299926758, "learning_rate": 0.0005, "loss": 1.084134817123413, "step": 47632 }, { "epoch": 3.755065037445802, "grad_norm": 0.4913729727268219, "learning_rate": 0.0005, "loss": 1.1918110847473145, "step": 47633 }, { "epoch": 3.7551438707134412, "grad_norm": 0.5131284594535828, "learning_rate": 0.0005, "loss": 1.2001363039016724, "step": 47634 }, { "epoch": 3.75522270398108, "grad_norm": 0.4907362461090088, "learning_rate": 0.0005, "loss": 1.2618827819824219, "step": 47635 }, { "epoch": 3.755301537248719, "grad_norm": 0.5800991654396057, "learning_rate": 0.0005, "loss": 1.1843641996383667, "step": 47636 }, { "epoch": 3.7553803705163578, "grad_norm": 0.5149943828582764, "learning_rate": 0.0005, "loss": 1.3556252717971802, "step": 47637 }, { "epoch": 3.755459203783997, "grad_norm": 0.43687763810157776, "learning_rate": 0.0005, "loss": 1.218307375907898, "step": 47638 }, { "epoch": 3.755538037051636, "grad_norm": 0.46905943751335144, "learning_rate": 0.0005, "loss": 1.1218864917755127, "step": 47639 }, { "epoch": 3.7556168703192747, "grad_norm": 0.4704025983810425, "learning_rate": 0.0005, "loss": 1.2867562770843506, "step": 47640 }, { "epoch": 3.7556957035869134, "grad_norm": 0.5515212416648865, "learning_rate": 0.0005, "loss": 1.3434655666351318, "step": 47641 }, { "epoch": 3.7557745368545525, "grad_norm": 0.43559515476226807, "learning_rate": 0.0005, "loss": 1.000876545906067, "step": 47642 }, { "epoch": 3.7558533701221917, "grad_norm": 0.4332634210586548, "learning_rate": 0.0005, "loss": 1.0292893648147583, "step": 47643 }, { "epoch": 3.7559322033898304, "grad_norm": 0.4934975802898407, "learning_rate": 0.0005, "loss": 0.9754161834716797, "step": 47644 }, { "epoch": 3.7560110366574695, "grad_norm": 0.4649258255958557, "learning_rate": 0.0005, "loss": 1.0842772722244263, "step": 47645 }, { "epoch": 3.756089869925108, "grad_norm": 0.6135025024414062, "learning_rate": 0.0005, "loss": 1.2871184349060059, "step": 47646 }, { "epoch": 3.7561687031927473, "grad_norm": 0.6806021928787231, "learning_rate": 0.0005, "loss": 1.2668014764785767, "step": 47647 }, { "epoch": 3.7562475364603864, "grad_norm": 0.486529141664505, "learning_rate": 0.0005, "loss": 1.043879508972168, "step": 47648 }, { "epoch": 3.756326369728025, "grad_norm": 0.6463394165039062, "learning_rate": 0.0005, "loss": 1.131166696548462, "step": 47649 }, { "epoch": 3.7564052029956643, "grad_norm": 0.5456267595291138, "learning_rate": 0.0005, "loss": 1.2368700504302979, "step": 47650 }, { "epoch": 3.756484036263303, "grad_norm": 0.47847622632980347, "learning_rate": 0.0005, "loss": 1.0812091827392578, "step": 47651 }, { "epoch": 3.756562869530942, "grad_norm": 0.5569092631340027, "learning_rate": 0.0005, "loss": 1.027113914489746, "step": 47652 }, { "epoch": 3.7566417027985812, "grad_norm": 0.5198986530303955, "learning_rate": 0.0005, "loss": 1.109748363494873, "step": 47653 }, { "epoch": 3.75672053606622, "grad_norm": 0.5171032547950745, "learning_rate": 0.0005, "loss": 1.1966720819473267, "step": 47654 }, { "epoch": 3.756799369333859, "grad_norm": 0.5575889348983765, "learning_rate": 0.0005, "loss": 1.1598515510559082, "step": 47655 }, { "epoch": 3.7568782026014977, "grad_norm": 0.5354279279708862, "learning_rate": 0.0005, "loss": 1.3395522832870483, "step": 47656 }, { "epoch": 3.756957035869137, "grad_norm": 0.4262271225452423, "learning_rate": 0.0005, "loss": 1.1366314888000488, "step": 47657 }, { "epoch": 3.7570358691367756, "grad_norm": 0.47248417139053345, "learning_rate": 0.0005, "loss": 1.1850996017456055, "step": 47658 }, { "epoch": 3.7571147024044147, "grad_norm": 0.9029517769813538, "learning_rate": 0.0005, "loss": 1.3771432638168335, "step": 47659 }, { "epoch": 3.7571935356720534, "grad_norm": 0.5351218581199646, "learning_rate": 0.0005, "loss": 1.377631425857544, "step": 47660 }, { "epoch": 3.7572723689396925, "grad_norm": 0.6152691841125488, "learning_rate": 0.0005, "loss": 1.2262909412384033, "step": 47661 }, { "epoch": 3.7573512022073317, "grad_norm": 0.4818786084651947, "learning_rate": 0.0005, "loss": 1.1577708721160889, "step": 47662 }, { "epoch": 3.7574300354749703, "grad_norm": 0.7700439691543579, "learning_rate": 0.0005, "loss": 1.2226536273956299, "step": 47663 }, { "epoch": 3.7575088687426095, "grad_norm": 0.5330567359924316, "learning_rate": 0.0005, "loss": 1.1208381652832031, "step": 47664 }, { "epoch": 3.757587702010248, "grad_norm": 0.5801302790641785, "learning_rate": 0.0005, "loss": 1.1871685981750488, "step": 47665 }, { "epoch": 3.7576665352778873, "grad_norm": 0.6841099858283997, "learning_rate": 0.0005, "loss": 1.088749647140503, "step": 47666 }, { "epoch": 3.7577453685455264, "grad_norm": 0.4398294687271118, "learning_rate": 0.0005, "loss": 1.0665314197540283, "step": 47667 }, { "epoch": 3.757824201813165, "grad_norm": 0.5992855429649353, "learning_rate": 0.0005, "loss": 1.1078215837478638, "step": 47668 }, { "epoch": 3.7579030350808043, "grad_norm": 0.48794692754745483, "learning_rate": 0.0005, "loss": 1.1868926286697388, "step": 47669 }, { "epoch": 3.757981868348443, "grad_norm": 0.44033122062683105, "learning_rate": 0.0005, "loss": 1.0614906549453735, "step": 47670 }, { "epoch": 3.758060701616082, "grad_norm": 0.46518954634666443, "learning_rate": 0.0005, "loss": 1.1440956592559814, "step": 47671 }, { "epoch": 3.7581395348837208, "grad_norm": 0.5211796164512634, "learning_rate": 0.0005, "loss": 1.2272028923034668, "step": 47672 }, { "epoch": 3.75821836815136, "grad_norm": 0.4008989930152893, "learning_rate": 0.0005, "loss": 1.0880272388458252, "step": 47673 }, { "epoch": 3.7582972014189986, "grad_norm": 0.4653196632862091, "learning_rate": 0.0005, "loss": 1.2042899131774902, "step": 47674 }, { "epoch": 3.7583760346866377, "grad_norm": 0.44472262263298035, "learning_rate": 0.0005, "loss": 1.1386744976043701, "step": 47675 }, { "epoch": 3.758454867954277, "grad_norm": 0.6518172025680542, "learning_rate": 0.0005, "loss": 1.2207974195480347, "step": 47676 }, { "epoch": 3.7585337012219155, "grad_norm": 0.5126938223838806, "learning_rate": 0.0005, "loss": 1.207007884979248, "step": 47677 }, { "epoch": 3.7586125344895547, "grad_norm": 0.5532264113426208, "learning_rate": 0.0005, "loss": 1.059507131576538, "step": 47678 }, { "epoch": 3.7586913677571934, "grad_norm": 0.577275812625885, "learning_rate": 0.0005, "loss": 1.3637109994888306, "step": 47679 }, { "epoch": 3.7587702010248325, "grad_norm": 0.610150933265686, "learning_rate": 0.0005, "loss": 1.399498462677002, "step": 47680 }, { "epoch": 3.7588490342924716, "grad_norm": 0.4222254455089569, "learning_rate": 0.0005, "loss": 1.1317120790481567, "step": 47681 }, { "epoch": 3.7589278675601103, "grad_norm": 0.6098822355270386, "learning_rate": 0.0005, "loss": 1.2248979806900024, "step": 47682 }, { "epoch": 3.7590067008277495, "grad_norm": 0.9516089558601379, "learning_rate": 0.0005, "loss": 1.100293517112732, "step": 47683 }, { "epoch": 3.759085534095388, "grad_norm": 0.5476502776145935, "learning_rate": 0.0005, "loss": 1.276792287826538, "step": 47684 }, { "epoch": 3.7591643673630273, "grad_norm": 0.5479835867881775, "learning_rate": 0.0005, "loss": 1.3438304662704468, "step": 47685 }, { "epoch": 3.7592432006306664, "grad_norm": 0.4862269461154938, "learning_rate": 0.0005, "loss": 1.0364727973937988, "step": 47686 }, { "epoch": 3.759322033898305, "grad_norm": 0.5309627056121826, "learning_rate": 0.0005, "loss": 1.2611101865768433, "step": 47687 }, { "epoch": 3.759400867165944, "grad_norm": 0.47509804368019104, "learning_rate": 0.0005, "loss": 1.1779541969299316, "step": 47688 }, { "epoch": 3.759479700433583, "grad_norm": 0.5215068459510803, "learning_rate": 0.0005, "loss": 1.2981741428375244, "step": 47689 }, { "epoch": 3.759558533701222, "grad_norm": 0.49352961778640747, "learning_rate": 0.0005, "loss": 1.1525349617004395, "step": 47690 }, { "epoch": 3.7596373669688607, "grad_norm": 0.47797733545303345, "learning_rate": 0.0005, "loss": 1.2669599056243896, "step": 47691 }, { "epoch": 3.7597162002365, "grad_norm": 0.6452721953392029, "learning_rate": 0.0005, "loss": 1.0709978342056274, "step": 47692 }, { "epoch": 3.7597950335041386, "grad_norm": 0.5245134830474854, "learning_rate": 0.0005, "loss": 1.1277124881744385, "step": 47693 }, { "epoch": 3.7598738667717777, "grad_norm": 0.43841874599456787, "learning_rate": 0.0005, "loss": 1.047663927078247, "step": 47694 }, { "epoch": 3.759952700039417, "grad_norm": 0.5161498188972473, "learning_rate": 0.0005, "loss": 1.2125236988067627, "step": 47695 }, { "epoch": 3.7600315333070555, "grad_norm": 0.6018449068069458, "learning_rate": 0.0005, "loss": 1.1298340559005737, "step": 47696 }, { "epoch": 3.7601103665746947, "grad_norm": 0.5003300309181213, "learning_rate": 0.0005, "loss": 1.1578946113586426, "step": 47697 }, { "epoch": 3.7601891998423334, "grad_norm": 0.4181133508682251, "learning_rate": 0.0005, "loss": 1.0776792764663696, "step": 47698 }, { "epoch": 3.7602680331099725, "grad_norm": 0.5508769154548645, "learning_rate": 0.0005, "loss": 1.180513620376587, "step": 47699 }, { "epoch": 3.7603468663776116, "grad_norm": 0.4010925590991974, "learning_rate": 0.0005, "loss": 1.2919986248016357, "step": 47700 }, { "epoch": 3.7604256996452503, "grad_norm": 0.5539713501930237, "learning_rate": 0.0005, "loss": 1.1431143283843994, "step": 47701 }, { "epoch": 3.760504532912889, "grad_norm": 0.4764992594718933, "learning_rate": 0.0005, "loss": 1.2016217708587646, "step": 47702 }, { "epoch": 3.760583366180528, "grad_norm": 0.4960702955722809, "learning_rate": 0.0005, "loss": 1.1844546794891357, "step": 47703 }, { "epoch": 3.7606621994481673, "grad_norm": 3.7119059562683105, "learning_rate": 0.0005, "loss": 1.1311602592468262, "step": 47704 }, { "epoch": 3.760741032715806, "grad_norm": 0.4239588677883148, "learning_rate": 0.0005, "loss": 1.1669073104858398, "step": 47705 }, { "epoch": 3.760819865983445, "grad_norm": 0.44190874695777893, "learning_rate": 0.0005, "loss": 1.0630416870117188, "step": 47706 }, { "epoch": 3.7608986992510838, "grad_norm": 0.47784748673439026, "learning_rate": 0.0005, "loss": 1.2922048568725586, "step": 47707 }, { "epoch": 3.760977532518723, "grad_norm": 0.4868296682834625, "learning_rate": 0.0005, "loss": 1.1608244180679321, "step": 47708 }, { "epoch": 3.761056365786362, "grad_norm": 0.4519006609916687, "learning_rate": 0.0005, "loss": 1.2551549673080444, "step": 47709 }, { "epoch": 3.7611351990540007, "grad_norm": 0.5360779762268066, "learning_rate": 0.0005, "loss": 1.2869242429733276, "step": 47710 }, { "epoch": 3.76121403232164, "grad_norm": 0.4670504331588745, "learning_rate": 0.0005, "loss": 1.1135680675506592, "step": 47711 }, { "epoch": 3.7612928655892786, "grad_norm": 0.4896929860115051, "learning_rate": 0.0005, "loss": 1.0239686965942383, "step": 47712 }, { "epoch": 3.7613716988569177, "grad_norm": 0.4003189504146576, "learning_rate": 0.0005, "loss": 0.886850893497467, "step": 47713 }, { "epoch": 3.761450532124557, "grad_norm": 0.4865116477012634, "learning_rate": 0.0005, "loss": 1.229118824005127, "step": 47714 }, { "epoch": 3.7615293653921955, "grad_norm": 0.4653203785419464, "learning_rate": 0.0005, "loss": 1.1324775218963623, "step": 47715 }, { "epoch": 3.761608198659834, "grad_norm": 0.44379815459251404, "learning_rate": 0.0005, "loss": 1.147873878479004, "step": 47716 }, { "epoch": 3.7616870319274733, "grad_norm": 0.4468235671520233, "learning_rate": 0.0005, "loss": 1.2245635986328125, "step": 47717 }, { "epoch": 3.7617658651951125, "grad_norm": 0.44528818130493164, "learning_rate": 0.0005, "loss": 1.086312174797058, "step": 47718 }, { "epoch": 3.761844698462751, "grad_norm": 0.6537942886352539, "learning_rate": 0.0005, "loss": 1.1934930086135864, "step": 47719 }, { "epoch": 3.7619235317303903, "grad_norm": 0.4909402132034302, "learning_rate": 0.0005, "loss": 1.1014984846115112, "step": 47720 }, { "epoch": 3.762002364998029, "grad_norm": 0.47741031646728516, "learning_rate": 0.0005, "loss": 1.216583013534546, "step": 47721 }, { "epoch": 3.762081198265668, "grad_norm": 0.4633890688419342, "learning_rate": 0.0005, "loss": 1.0178511142730713, "step": 47722 }, { "epoch": 3.7621600315333072, "grad_norm": 0.49556005001068115, "learning_rate": 0.0005, "loss": 1.1532347202301025, "step": 47723 }, { "epoch": 3.762238864800946, "grad_norm": 0.46863749623298645, "learning_rate": 0.0005, "loss": 1.1887333393096924, "step": 47724 }, { "epoch": 3.762317698068585, "grad_norm": 0.6361228823661804, "learning_rate": 0.0005, "loss": 1.3364348411560059, "step": 47725 }, { "epoch": 3.7623965313362238, "grad_norm": 0.42674434185028076, "learning_rate": 0.0005, "loss": 1.0037180185317993, "step": 47726 }, { "epoch": 3.762475364603863, "grad_norm": 0.4797123968601227, "learning_rate": 0.0005, "loss": 1.1542062759399414, "step": 47727 }, { "epoch": 3.762554197871502, "grad_norm": 0.5653451681137085, "learning_rate": 0.0005, "loss": 1.2948435544967651, "step": 47728 }, { "epoch": 3.7626330311391407, "grad_norm": 0.5801979303359985, "learning_rate": 0.0005, "loss": 1.327214002609253, "step": 47729 }, { "epoch": 3.7627118644067794, "grad_norm": 0.4811076521873474, "learning_rate": 0.0005, "loss": 1.3554528951644897, "step": 47730 }, { "epoch": 3.7627906976744185, "grad_norm": 0.5360563397407532, "learning_rate": 0.0005, "loss": 1.1010639667510986, "step": 47731 }, { "epoch": 3.7628695309420577, "grad_norm": 0.48893463611602783, "learning_rate": 0.0005, "loss": 1.1409748792648315, "step": 47732 }, { "epoch": 3.7629483642096964, "grad_norm": 0.5254149436950684, "learning_rate": 0.0005, "loss": 1.134168267250061, "step": 47733 }, { "epoch": 3.7630271974773355, "grad_norm": 0.5699630379676819, "learning_rate": 0.0005, "loss": 1.2554396390914917, "step": 47734 }, { "epoch": 3.763106030744974, "grad_norm": 0.5735316276550293, "learning_rate": 0.0005, "loss": 1.2208280563354492, "step": 47735 }, { "epoch": 3.7631848640126133, "grad_norm": 0.44584664702415466, "learning_rate": 0.0005, "loss": 1.1532089710235596, "step": 47736 }, { "epoch": 3.7632636972802525, "grad_norm": 0.5439935922622681, "learning_rate": 0.0005, "loss": 1.13620126247406, "step": 47737 }, { "epoch": 3.763342530547891, "grad_norm": 0.509874701499939, "learning_rate": 0.0005, "loss": 1.1957252025604248, "step": 47738 }, { "epoch": 3.7634213638155303, "grad_norm": 0.46454235911369324, "learning_rate": 0.0005, "loss": 1.157400369644165, "step": 47739 }, { "epoch": 3.763500197083169, "grad_norm": 0.4255654513835907, "learning_rate": 0.0005, "loss": 1.1905711889266968, "step": 47740 }, { "epoch": 3.763579030350808, "grad_norm": 0.40827780961990356, "learning_rate": 0.0005, "loss": 0.9939552545547485, "step": 47741 }, { "epoch": 3.7636578636184472, "grad_norm": 0.5553629398345947, "learning_rate": 0.0005, "loss": 1.0942249298095703, "step": 47742 }, { "epoch": 3.763736696886086, "grad_norm": 0.736721932888031, "learning_rate": 0.0005, "loss": 1.242834448814392, "step": 47743 }, { "epoch": 3.7638155301537246, "grad_norm": 0.5075082778930664, "learning_rate": 0.0005, "loss": 1.3557355403900146, "step": 47744 }, { "epoch": 3.7638943634213637, "grad_norm": 0.42811474204063416, "learning_rate": 0.0005, "loss": 1.066518783569336, "step": 47745 }, { "epoch": 3.763973196689003, "grad_norm": 0.48368480801582336, "learning_rate": 0.0005, "loss": 1.1287415027618408, "step": 47746 }, { "epoch": 3.7640520299566416, "grad_norm": 0.44832324981689453, "learning_rate": 0.0005, "loss": 1.03669273853302, "step": 47747 }, { "epoch": 3.7641308632242807, "grad_norm": 0.5345262289047241, "learning_rate": 0.0005, "loss": 1.2135753631591797, "step": 47748 }, { "epoch": 3.7642096964919194, "grad_norm": 0.46969231963157654, "learning_rate": 0.0005, "loss": 1.185306429862976, "step": 47749 }, { "epoch": 3.7642885297595585, "grad_norm": 0.4909327030181885, "learning_rate": 0.0005, "loss": 0.9374779462814331, "step": 47750 }, { "epoch": 3.7643673630271977, "grad_norm": 0.5695599913597107, "learning_rate": 0.0005, "loss": 1.1368515491485596, "step": 47751 }, { "epoch": 3.7644461962948363, "grad_norm": 0.48044782876968384, "learning_rate": 0.0005, "loss": 1.2431952953338623, "step": 47752 }, { "epoch": 3.7645250295624755, "grad_norm": 0.47228723764419556, "learning_rate": 0.0005, "loss": 1.038672924041748, "step": 47753 }, { "epoch": 3.764603862830114, "grad_norm": 0.7170445322990417, "learning_rate": 0.0005, "loss": 1.1692893505096436, "step": 47754 }, { "epoch": 3.7646826960977533, "grad_norm": 0.43312105536460876, "learning_rate": 0.0005, "loss": 1.0266320705413818, "step": 47755 }, { "epoch": 3.7647615293653924, "grad_norm": 0.47066816687583923, "learning_rate": 0.0005, "loss": 1.1950058937072754, "step": 47756 }, { "epoch": 3.764840362633031, "grad_norm": 0.41976839303970337, "learning_rate": 0.0005, "loss": 1.130739688873291, "step": 47757 }, { "epoch": 3.76491919590067, "grad_norm": 0.4135515093803406, "learning_rate": 0.0005, "loss": 1.121386170387268, "step": 47758 }, { "epoch": 3.764998029168309, "grad_norm": 0.42876937985420227, "learning_rate": 0.0005, "loss": 1.1887896060943604, "step": 47759 }, { "epoch": 3.765076862435948, "grad_norm": 0.47003525495529175, "learning_rate": 0.0005, "loss": 1.2745022773742676, "step": 47760 }, { "epoch": 3.7651556957035868, "grad_norm": 0.7162364721298218, "learning_rate": 0.0005, "loss": 1.2720706462860107, "step": 47761 }, { "epoch": 3.765234528971226, "grad_norm": 0.4542120397090912, "learning_rate": 0.0005, "loss": 1.2326515913009644, "step": 47762 }, { "epoch": 3.7653133622388646, "grad_norm": 0.5518892407417297, "learning_rate": 0.0005, "loss": 1.2506777048110962, "step": 47763 }, { "epoch": 3.7653921955065037, "grad_norm": 0.4980738162994385, "learning_rate": 0.0005, "loss": 1.2401297092437744, "step": 47764 }, { "epoch": 3.765471028774143, "grad_norm": 0.5499925017356873, "learning_rate": 0.0005, "loss": 1.1176711320877075, "step": 47765 }, { "epoch": 3.7655498620417815, "grad_norm": 0.4590449929237366, "learning_rate": 0.0005, "loss": 1.211517572402954, "step": 47766 }, { "epoch": 3.7656286953094207, "grad_norm": 0.4451529383659363, "learning_rate": 0.0005, "loss": 1.1206517219543457, "step": 47767 }, { "epoch": 3.7657075285770594, "grad_norm": 0.37559744715690613, "learning_rate": 0.0005, "loss": 0.9324386119842529, "step": 47768 }, { "epoch": 3.7657863618446985, "grad_norm": 0.34422409534454346, "learning_rate": 0.0005, "loss": 0.9755288362503052, "step": 47769 }, { "epoch": 3.7658651951123376, "grad_norm": 0.4109373986721039, "learning_rate": 0.0005, "loss": 1.1520159244537354, "step": 47770 }, { "epoch": 3.7659440283799763, "grad_norm": 0.5084488987922668, "learning_rate": 0.0005, "loss": 1.1702065467834473, "step": 47771 }, { "epoch": 3.766022861647615, "grad_norm": 0.586485743522644, "learning_rate": 0.0005, "loss": 1.4121193885803223, "step": 47772 }, { "epoch": 3.766101694915254, "grad_norm": 0.4385489523410797, "learning_rate": 0.0005, "loss": 1.0388202667236328, "step": 47773 }, { "epoch": 3.7661805281828933, "grad_norm": 0.46056264638900757, "learning_rate": 0.0005, "loss": 1.2722015380859375, "step": 47774 }, { "epoch": 3.766259361450532, "grad_norm": 0.43104153871536255, "learning_rate": 0.0005, "loss": 1.1169121265411377, "step": 47775 }, { "epoch": 3.766338194718171, "grad_norm": 0.4977409243583679, "learning_rate": 0.0005, "loss": 1.429433822631836, "step": 47776 }, { "epoch": 3.76641702798581, "grad_norm": 0.4056161046028137, "learning_rate": 0.0005, "loss": 1.0243626832962036, "step": 47777 }, { "epoch": 3.766495861253449, "grad_norm": 0.466445654630661, "learning_rate": 0.0005, "loss": 1.0669811964035034, "step": 47778 }, { "epoch": 3.766574694521088, "grad_norm": 0.47638583183288574, "learning_rate": 0.0005, "loss": 1.0473252534866333, "step": 47779 }, { "epoch": 3.7666535277887268, "grad_norm": 0.4690982401371002, "learning_rate": 0.0005, "loss": 1.0928752422332764, "step": 47780 }, { "epoch": 3.766732361056366, "grad_norm": 0.4697532057762146, "learning_rate": 0.0005, "loss": 1.1764590740203857, "step": 47781 }, { "epoch": 3.7668111943240046, "grad_norm": 0.6319376230239868, "learning_rate": 0.0005, "loss": 1.232750415802002, "step": 47782 }, { "epoch": 3.7668900275916437, "grad_norm": 0.4694626033306122, "learning_rate": 0.0005, "loss": 1.252299189567566, "step": 47783 }, { "epoch": 3.766968860859283, "grad_norm": 0.4190211892127991, "learning_rate": 0.0005, "loss": 1.1964738368988037, "step": 47784 }, { "epoch": 3.7670476941269215, "grad_norm": 0.4411506950855255, "learning_rate": 0.0005, "loss": 1.0989539623260498, "step": 47785 }, { "epoch": 3.7671265273945607, "grad_norm": 0.5097970962524414, "learning_rate": 0.0005, "loss": 1.2827932834625244, "step": 47786 }, { "epoch": 3.7672053606621994, "grad_norm": 0.5508875250816345, "learning_rate": 0.0005, "loss": 1.30096435546875, "step": 47787 }, { "epoch": 3.7672841939298385, "grad_norm": 0.5781799554824829, "learning_rate": 0.0005, "loss": 1.2341406345367432, "step": 47788 }, { "epoch": 3.767363027197477, "grad_norm": 0.4748062193393707, "learning_rate": 0.0005, "loss": 1.2179381847381592, "step": 47789 }, { "epoch": 3.7674418604651163, "grad_norm": 0.740987241268158, "learning_rate": 0.0005, "loss": 1.311237096786499, "step": 47790 }, { "epoch": 3.767520693732755, "grad_norm": 0.5480278134346008, "learning_rate": 0.0005, "loss": 1.2896883487701416, "step": 47791 }, { "epoch": 3.767599527000394, "grad_norm": 0.4603583812713623, "learning_rate": 0.0005, "loss": 0.9759777188301086, "step": 47792 }, { "epoch": 3.7676783602680333, "grad_norm": 0.5446767807006836, "learning_rate": 0.0005, "loss": 1.438234567642212, "step": 47793 }, { "epoch": 3.767757193535672, "grad_norm": 0.5580354332923889, "learning_rate": 0.0005, "loss": 1.1532435417175293, "step": 47794 }, { "epoch": 3.767836026803311, "grad_norm": 0.5106849074363708, "learning_rate": 0.0005, "loss": 1.2544641494750977, "step": 47795 }, { "epoch": 3.76791486007095, "grad_norm": 0.549268364906311, "learning_rate": 0.0005, "loss": 1.1078561544418335, "step": 47796 }, { "epoch": 3.767993693338589, "grad_norm": 0.49245983362197876, "learning_rate": 0.0005, "loss": 1.0666850805282593, "step": 47797 }, { "epoch": 3.768072526606228, "grad_norm": 0.5558907389640808, "learning_rate": 0.0005, "loss": 1.2340846061706543, "step": 47798 }, { "epoch": 3.7681513598738667, "grad_norm": 0.46185341477394104, "learning_rate": 0.0005, "loss": 1.211582899093628, "step": 47799 }, { "epoch": 3.768230193141506, "grad_norm": 0.6651843190193176, "learning_rate": 0.0005, "loss": 1.2943050861358643, "step": 47800 }, { "epoch": 3.7683090264091446, "grad_norm": 0.4730473458766937, "learning_rate": 0.0005, "loss": 1.139979600906372, "step": 47801 }, { "epoch": 3.7683878596767837, "grad_norm": 0.3924810290336609, "learning_rate": 0.0005, "loss": 0.9025909900665283, "step": 47802 }, { "epoch": 3.7684666929444224, "grad_norm": 0.5674402117729187, "learning_rate": 0.0005, "loss": 1.2115321159362793, "step": 47803 }, { "epoch": 3.7685455262120615, "grad_norm": 0.3950088322162628, "learning_rate": 0.0005, "loss": 0.9850706458091736, "step": 47804 }, { "epoch": 3.7686243594797, "grad_norm": 0.5174887180328369, "learning_rate": 0.0005, "loss": 1.120173454284668, "step": 47805 }, { "epoch": 3.7687031927473393, "grad_norm": 0.564967930316925, "learning_rate": 0.0005, "loss": 1.269067645072937, "step": 47806 }, { "epoch": 3.7687820260149785, "grad_norm": 0.5557183027267456, "learning_rate": 0.0005, "loss": 1.2629342079162598, "step": 47807 }, { "epoch": 3.768860859282617, "grad_norm": 0.4197372496128082, "learning_rate": 0.0005, "loss": 1.0229063034057617, "step": 47808 }, { "epoch": 3.7689396925502563, "grad_norm": 0.5000292658805847, "learning_rate": 0.0005, "loss": 1.2845500707626343, "step": 47809 }, { "epoch": 3.769018525817895, "grad_norm": 0.486565500497818, "learning_rate": 0.0005, "loss": 1.114579677581787, "step": 47810 }, { "epoch": 3.769097359085534, "grad_norm": 0.5249280333518982, "learning_rate": 0.0005, "loss": 1.2372276782989502, "step": 47811 }, { "epoch": 3.7691761923531732, "grad_norm": 0.4831638038158417, "learning_rate": 0.0005, "loss": 1.2663079500198364, "step": 47812 }, { "epoch": 3.769255025620812, "grad_norm": 0.5191178917884827, "learning_rate": 0.0005, "loss": 1.1812573671340942, "step": 47813 }, { "epoch": 3.769333858888451, "grad_norm": 0.4965883493423462, "learning_rate": 0.0005, "loss": 1.5191452503204346, "step": 47814 }, { "epoch": 3.7694126921560898, "grad_norm": 0.6569348573684692, "learning_rate": 0.0005, "loss": 1.1353886127471924, "step": 47815 }, { "epoch": 3.769491525423729, "grad_norm": 0.4945670962333679, "learning_rate": 0.0005, "loss": 1.3156919479370117, "step": 47816 }, { "epoch": 3.769570358691368, "grad_norm": 0.5964895486831665, "learning_rate": 0.0005, "loss": 1.3612236976623535, "step": 47817 }, { "epoch": 3.7696491919590067, "grad_norm": 0.35069239139556885, "learning_rate": 0.0005, "loss": 0.9840314388275146, "step": 47818 }, { "epoch": 3.7697280252266454, "grad_norm": 0.4743313789367676, "learning_rate": 0.0005, "loss": 1.3143041133880615, "step": 47819 }, { "epoch": 3.7698068584942845, "grad_norm": 0.6410946846008301, "learning_rate": 0.0005, "loss": 1.2739076614379883, "step": 47820 }, { "epoch": 3.7698856917619237, "grad_norm": 0.49361419677734375, "learning_rate": 0.0005, "loss": 1.2556769847869873, "step": 47821 }, { "epoch": 3.7699645250295624, "grad_norm": 0.4352644383907318, "learning_rate": 0.0005, "loss": 1.236344337463379, "step": 47822 }, { "epoch": 3.7700433582972015, "grad_norm": 0.5758061408996582, "learning_rate": 0.0005, "loss": 1.277816653251648, "step": 47823 }, { "epoch": 3.77012219156484, "grad_norm": 0.5204950571060181, "learning_rate": 0.0005, "loss": 0.9999688863754272, "step": 47824 }, { "epoch": 3.7702010248324793, "grad_norm": 0.6048299670219421, "learning_rate": 0.0005, "loss": 1.1370875835418701, "step": 47825 }, { "epoch": 3.7702798581001185, "grad_norm": 0.4607203006744385, "learning_rate": 0.0005, "loss": 1.3111997842788696, "step": 47826 }, { "epoch": 3.770358691367757, "grad_norm": 0.6949854493141174, "learning_rate": 0.0005, "loss": 1.343198537826538, "step": 47827 }, { "epoch": 3.7704375246353963, "grad_norm": 0.5132368206977844, "learning_rate": 0.0005, "loss": 1.1594350337982178, "step": 47828 }, { "epoch": 3.770516357903035, "grad_norm": 0.4477895200252533, "learning_rate": 0.0005, "loss": 1.1532351970672607, "step": 47829 }, { "epoch": 3.770595191170674, "grad_norm": 0.46723097562789917, "learning_rate": 0.0005, "loss": 1.099536418914795, "step": 47830 }, { "epoch": 3.7706740244383132, "grad_norm": 0.46101075410842896, "learning_rate": 0.0005, "loss": 1.0586020946502686, "step": 47831 }, { "epoch": 3.770752857705952, "grad_norm": 0.5792752504348755, "learning_rate": 0.0005, "loss": 1.1947569847106934, "step": 47832 }, { "epoch": 3.7708316909735906, "grad_norm": 0.4382857084274292, "learning_rate": 0.0005, "loss": 1.070811152458191, "step": 47833 }, { "epoch": 3.7709105242412297, "grad_norm": 0.38498160243034363, "learning_rate": 0.0005, "loss": 1.1108014583587646, "step": 47834 }, { "epoch": 3.770989357508869, "grad_norm": 0.46567803621292114, "learning_rate": 0.0005, "loss": 1.2295105457305908, "step": 47835 }, { "epoch": 3.7710681907765076, "grad_norm": 0.3913422226905823, "learning_rate": 0.0005, "loss": 1.084498405456543, "step": 47836 }, { "epoch": 3.7711470240441467, "grad_norm": 0.538828432559967, "learning_rate": 0.0005, "loss": 1.2616113424301147, "step": 47837 }, { "epoch": 3.7712258573117854, "grad_norm": 0.4902646839618683, "learning_rate": 0.0005, "loss": 1.193641185760498, "step": 47838 }, { "epoch": 3.7713046905794245, "grad_norm": 0.5319668054580688, "learning_rate": 0.0005, "loss": 1.1967060565948486, "step": 47839 }, { "epoch": 3.7713835238470637, "grad_norm": 0.5344166159629822, "learning_rate": 0.0005, "loss": 1.2617921829223633, "step": 47840 }, { "epoch": 3.7714623571147023, "grad_norm": 0.47251078486442566, "learning_rate": 0.0005, "loss": 1.1919537782669067, "step": 47841 }, { "epoch": 3.7715411903823415, "grad_norm": 0.5174733996391296, "learning_rate": 0.0005, "loss": 1.0995733737945557, "step": 47842 }, { "epoch": 3.77162002364998, "grad_norm": 0.4430244565010071, "learning_rate": 0.0005, "loss": 1.0332858562469482, "step": 47843 }, { "epoch": 3.7716988569176193, "grad_norm": 0.550708532333374, "learning_rate": 0.0005, "loss": 1.1891521215438843, "step": 47844 }, { "epoch": 3.7717776901852584, "grad_norm": 0.46046462655067444, "learning_rate": 0.0005, "loss": 1.3454697132110596, "step": 47845 }, { "epoch": 3.771856523452897, "grad_norm": 0.5000185370445251, "learning_rate": 0.0005, "loss": 1.0845348834991455, "step": 47846 }, { "epoch": 3.771935356720536, "grad_norm": 0.5772377252578735, "learning_rate": 0.0005, "loss": 1.2099049091339111, "step": 47847 }, { "epoch": 3.772014189988175, "grad_norm": 0.577976644039154, "learning_rate": 0.0005, "loss": 1.1213582754135132, "step": 47848 }, { "epoch": 3.772093023255814, "grad_norm": 0.49039483070373535, "learning_rate": 0.0005, "loss": 1.2173805236816406, "step": 47849 }, { "epoch": 3.7721718565234528, "grad_norm": 0.5087729096412659, "learning_rate": 0.0005, "loss": 1.188234567642212, "step": 47850 }, { "epoch": 3.772250689791092, "grad_norm": 0.7633214592933655, "learning_rate": 0.0005, "loss": 1.3254843950271606, "step": 47851 }, { "epoch": 3.7723295230587306, "grad_norm": 0.4219520688056946, "learning_rate": 0.0005, "loss": 1.1930433511734009, "step": 47852 }, { "epoch": 3.7724083563263697, "grad_norm": 0.5175580382347107, "learning_rate": 0.0005, "loss": 1.2620326280593872, "step": 47853 }, { "epoch": 3.772487189594009, "grad_norm": 0.5451998710632324, "learning_rate": 0.0005, "loss": 1.093217372894287, "step": 47854 }, { "epoch": 3.7725660228616476, "grad_norm": 0.472059428691864, "learning_rate": 0.0005, "loss": 1.098639726638794, "step": 47855 }, { "epoch": 3.7726448561292867, "grad_norm": 0.505832314491272, "learning_rate": 0.0005, "loss": 1.158998966217041, "step": 47856 }, { "epoch": 3.7727236893969254, "grad_norm": 0.5640198588371277, "learning_rate": 0.0005, "loss": 1.3254576921463013, "step": 47857 }, { "epoch": 3.7728025226645645, "grad_norm": 0.3861919343471527, "learning_rate": 0.0005, "loss": 0.8988702297210693, "step": 47858 }, { "epoch": 3.7728813559322036, "grad_norm": 0.5821352601051331, "learning_rate": 0.0005, "loss": 1.2186322212219238, "step": 47859 }, { "epoch": 3.7729601891998423, "grad_norm": 0.5172241926193237, "learning_rate": 0.0005, "loss": 1.1467838287353516, "step": 47860 }, { "epoch": 3.773039022467481, "grad_norm": 0.5426329970359802, "learning_rate": 0.0005, "loss": 1.3061351776123047, "step": 47861 }, { "epoch": 3.77311785573512, "grad_norm": 0.5397965908050537, "learning_rate": 0.0005, "loss": 1.1525323390960693, "step": 47862 }, { "epoch": 3.7731966890027593, "grad_norm": 0.4409922659397125, "learning_rate": 0.0005, "loss": 1.0657655000686646, "step": 47863 }, { "epoch": 3.773275522270398, "grad_norm": 0.5609351992607117, "learning_rate": 0.0005, "loss": 0.9942573308944702, "step": 47864 }, { "epoch": 3.773354355538037, "grad_norm": 0.5058344006538391, "learning_rate": 0.0005, "loss": 1.1020116806030273, "step": 47865 }, { "epoch": 3.773433188805676, "grad_norm": 0.5679904222488403, "learning_rate": 0.0005, "loss": 1.0192813873291016, "step": 47866 }, { "epoch": 3.773512022073315, "grad_norm": 0.45911553502082825, "learning_rate": 0.0005, "loss": 1.1203831434249878, "step": 47867 }, { "epoch": 3.773590855340954, "grad_norm": 0.4234350919723511, "learning_rate": 0.0005, "loss": 1.0758914947509766, "step": 47868 }, { "epoch": 3.7736696886085928, "grad_norm": 0.4544023871421814, "learning_rate": 0.0005, "loss": 1.0679185390472412, "step": 47869 }, { "epoch": 3.773748521876232, "grad_norm": 0.42590582370758057, "learning_rate": 0.0005, "loss": 1.0896828174591064, "step": 47870 }, { "epoch": 3.7738273551438706, "grad_norm": 0.4684154987335205, "learning_rate": 0.0005, "loss": 1.0759146213531494, "step": 47871 }, { "epoch": 3.7739061884115097, "grad_norm": 0.6176900863647461, "learning_rate": 0.0005, "loss": 1.4817553758621216, "step": 47872 }, { "epoch": 3.773985021679149, "grad_norm": 0.5506210327148438, "learning_rate": 0.0005, "loss": 0.9832216501235962, "step": 47873 }, { "epoch": 3.7740638549467875, "grad_norm": 0.5789762139320374, "learning_rate": 0.0005, "loss": 1.265237808227539, "step": 47874 }, { "epoch": 3.7741426882144262, "grad_norm": 0.6622952222824097, "learning_rate": 0.0005, "loss": 1.5288002490997314, "step": 47875 }, { "epoch": 3.7742215214820654, "grad_norm": 0.6347617506980896, "learning_rate": 0.0005, "loss": 0.9378191232681274, "step": 47876 }, { "epoch": 3.7743003547497045, "grad_norm": 0.5134606957435608, "learning_rate": 0.0005, "loss": 1.0794298648834229, "step": 47877 }, { "epoch": 3.774379188017343, "grad_norm": 0.5561254024505615, "learning_rate": 0.0005, "loss": 1.1302392482757568, "step": 47878 }, { "epoch": 3.7744580212849823, "grad_norm": 0.5516802072525024, "learning_rate": 0.0005, "loss": 1.313467264175415, "step": 47879 }, { "epoch": 3.774536854552621, "grad_norm": 0.6055036187171936, "learning_rate": 0.0005, "loss": 1.2115527391433716, "step": 47880 }, { "epoch": 3.77461568782026, "grad_norm": 0.5516918897628784, "learning_rate": 0.0005, "loss": 1.1332359313964844, "step": 47881 }, { "epoch": 3.7746945210878993, "grad_norm": 0.4587523937225342, "learning_rate": 0.0005, "loss": 1.1933109760284424, "step": 47882 }, { "epoch": 3.774773354355538, "grad_norm": 0.5449714064598083, "learning_rate": 0.0005, "loss": 1.1687670946121216, "step": 47883 }, { "epoch": 3.774852187623177, "grad_norm": 0.5740235447883606, "learning_rate": 0.0005, "loss": 1.2493364810943604, "step": 47884 }, { "epoch": 3.774931020890816, "grad_norm": 0.6027452945709229, "learning_rate": 0.0005, "loss": 1.175884485244751, "step": 47885 }, { "epoch": 3.775009854158455, "grad_norm": 0.4803236126899719, "learning_rate": 0.0005, "loss": 1.2504987716674805, "step": 47886 }, { "epoch": 3.775088687426094, "grad_norm": 0.5839945077896118, "learning_rate": 0.0005, "loss": 1.219482660293579, "step": 47887 }, { "epoch": 3.7751675206937327, "grad_norm": 0.5840188264846802, "learning_rate": 0.0005, "loss": 1.2060682773590088, "step": 47888 }, { "epoch": 3.7752463539613714, "grad_norm": 0.5037707686424255, "learning_rate": 0.0005, "loss": 1.0837860107421875, "step": 47889 }, { "epoch": 3.7753251872290106, "grad_norm": 0.5920798778533936, "learning_rate": 0.0005, "loss": 1.1771366596221924, "step": 47890 }, { "epoch": 3.7754040204966497, "grad_norm": 0.5680883526802063, "learning_rate": 0.0005, "loss": 1.1760239601135254, "step": 47891 }, { "epoch": 3.7754828537642884, "grad_norm": 0.5529114603996277, "learning_rate": 0.0005, "loss": 1.1300904750823975, "step": 47892 }, { "epoch": 3.7755616870319275, "grad_norm": 0.532650351524353, "learning_rate": 0.0005, "loss": 1.2165942192077637, "step": 47893 }, { "epoch": 3.775640520299566, "grad_norm": 0.5766903162002563, "learning_rate": 0.0005, "loss": 1.270459532737732, "step": 47894 }, { "epoch": 3.7757193535672053, "grad_norm": 0.4964175224304199, "learning_rate": 0.0005, "loss": 1.130427598953247, "step": 47895 }, { "epoch": 3.7757981868348445, "grad_norm": 0.5887556076049805, "learning_rate": 0.0005, "loss": 1.0775063037872314, "step": 47896 }, { "epoch": 3.775877020102483, "grad_norm": 0.5860036611557007, "learning_rate": 0.0005, "loss": 1.134909749031067, "step": 47897 }, { "epoch": 3.7759558533701223, "grad_norm": 0.510387659072876, "learning_rate": 0.0005, "loss": 1.4010050296783447, "step": 47898 }, { "epoch": 3.776034686637761, "grad_norm": 0.4874645471572876, "learning_rate": 0.0005, "loss": 1.333354115486145, "step": 47899 }, { "epoch": 3.7761135199054, "grad_norm": 0.602013111114502, "learning_rate": 0.0005, "loss": 1.2037297487258911, "step": 47900 }, { "epoch": 3.7761923531730393, "grad_norm": 0.5664818286895752, "learning_rate": 0.0005, "loss": 1.1190685033798218, "step": 47901 }, { "epoch": 3.776271186440678, "grad_norm": 0.42976951599121094, "learning_rate": 0.0005, "loss": 1.0893943309783936, "step": 47902 }, { "epoch": 3.776350019708317, "grad_norm": 0.5737363696098328, "learning_rate": 0.0005, "loss": 1.1803157329559326, "step": 47903 }, { "epoch": 3.7764288529759558, "grad_norm": 0.5285780429840088, "learning_rate": 0.0005, "loss": 1.1509888172149658, "step": 47904 }, { "epoch": 3.776507686243595, "grad_norm": 0.5385838150978088, "learning_rate": 0.0005, "loss": 1.2766406536102295, "step": 47905 }, { "epoch": 3.7765865195112336, "grad_norm": 0.584601640701294, "learning_rate": 0.0005, "loss": 1.208263874053955, "step": 47906 }, { "epoch": 3.7766653527788727, "grad_norm": 0.4839927852153778, "learning_rate": 0.0005, "loss": 1.245710849761963, "step": 47907 }, { "epoch": 3.7767441860465114, "grad_norm": 0.5235174894332886, "learning_rate": 0.0005, "loss": 1.182257890701294, "step": 47908 }, { "epoch": 3.7768230193141505, "grad_norm": 0.46361401677131653, "learning_rate": 0.0005, "loss": 1.1875661611557007, "step": 47909 }, { "epoch": 3.7769018525817897, "grad_norm": 0.47933080792427063, "learning_rate": 0.0005, "loss": 1.2697718143463135, "step": 47910 }, { "epoch": 3.7769806858494284, "grad_norm": 0.7146627306938171, "learning_rate": 0.0005, "loss": 1.4354124069213867, "step": 47911 }, { "epoch": 3.7770595191170675, "grad_norm": 0.48580488562583923, "learning_rate": 0.0005, "loss": 1.2809526920318604, "step": 47912 }, { "epoch": 3.777138352384706, "grad_norm": 0.41411691904067993, "learning_rate": 0.0005, "loss": 1.1872682571411133, "step": 47913 }, { "epoch": 3.7772171856523453, "grad_norm": 0.4773769974708557, "learning_rate": 0.0005, "loss": 1.0273783206939697, "step": 47914 }, { "epoch": 3.7772960189199845, "grad_norm": 0.4434937536716461, "learning_rate": 0.0005, "loss": 1.0269695520401, "step": 47915 }, { "epoch": 3.777374852187623, "grad_norm": 0.4468095302581787, "learning_rate": 0.0005, "loss": 1.1461830139160156, "step": 47916 }, { "epoch": 3.7774536854552623, "grad_norm": 0.4843105375766754, "learning_rate": 0.0005, "loss": 1.0317304134368896, "step": 47917 }, { "epoch": 3.777532518722901, "grad_norm": 0.5305312275886536, "learning_rate": 0.0005, "loss": 1.1181918382644653, "step": 47918 }, { "epoch": 3.77761135199054, "grad_norm": 0.4725796580314636, "learning_rate": 0.0005, "loss": 1.2513153553009033, "step": 47919 }, { "epoch": 3.777690185258179, "grad_norm": 2.978856086730957, "learning_rate": 0.0005, "loss": 1.2498186826705933, "step": 47920 }, { "epoch": 3.777769018525818, "grad_norm": 0.4695146381855011, "learning_rate": 0.0005, "loss": 1.1884825229644775, "step": 47921 }, { "epoch": 3.7778478517934566, "grad_norm": 0.48655226826667786, "learning_rate": 0.0005, "loss": 1.005533218383789, "step": 47922 }, { "epoch": 3.7779266850610957, "grad_norm": 0.5576333999633789, "learning_rate": 0.0005, "loss": 1.466069221496582, "step": 47923 }, { "epoch": 3.778005518328735, "grad_norm": 0.4165763556957245, "learning_rate": 0.0005, "loss": 1.1661453247070312, "step": 47924 }, { "epoch": 3.7780843515963736, "grad_norm": 0.4819973111152649, "learning_rate": 0.0005, "loss": 1.2031075954437256, "step": 47925 }, { "epoch": 3.7781631848640127, "grad_norm": 0.49225035309791565, "learning_rate": 0.0005, "loss": 1.0590758323669434, "step": 47926 }, { "epoch": 3.7782420181316514, "grad_norm": 0.5010620355606079, "learning_rate": 0.0005, "loss": 1.2961511611938477, "step": 47927 }, { "epoch": 3.7783208513992905, "grad_norm": 0.5302522778511047, "learning_rate": 0.0005, "loss": 1.268477439880371, "step": 47928 }, { "epoch": 3.7783996846669297, "grad_norm": 0.4780617952346802, "learning_rate": 0.0005, "loss": 1.130056381225586, "step": 47929 }, { "epoch": 3.7784785179345683, "grad_norm": 0.52369225025177, "learning_rate": 0.0005, "loss": 1.2046730518341064, "step": 47930 }, { "epoch": 3.7785573512022075, "grad_norm": 0.513227105140686, "learning_rate": 0.0005, "loss": 1.1152055263519287, "step": 47931 }, { "epoch": 3.778636184469846, "grad_norm": 0.6165761947631836, "learning_rate": 0.0005, "loss": 1.3312491178512573, "step": 47932 }, { "epoch": 3.7787150177374853, "grad_norm": 0.44285696744918823, "learning_rate": 0.0005, "loss": 1.1449230909347534, "step": 47933 }, { "epoch": 3.7787938510051244, "grad_norm": 0.666943371295929, "learning_rate": 0.0005, "loss": 1.1495962142944336, "step": 47934 }, { "epoch": 3.778872684272763, "grad_norm": 0.4937085211277008, "learning_rate": 0.0005, "loss": 1.0732133388519287, "step": 47935 }, { "epoch": 3.778951517540402, "grad_norm": 0.4905657470226288, "learning_rate": 0.0005, "loss": 1.0766267776489258, "step": 47936 }, { "epoch": 3.779030350808041, "grad_norm": 0.5420469641685486, "learning_rate": 0.0005, "loss": 1.232544183731079, "step": 47937 }, { "epoch": 3.77910918407568, "grad_norm": 0.5032942891120911, "learning_rate": 0.0005, "loss": 1.3064284324645996, "step": 47938 }, { "epoch": 3.7791880173433188, "grad_norm": 0.5386740565299988, "learning_rate": 0.0005, "loss": 1.0792495012283325, "step": 47939 }, { "epoch": 3.779266850610958, "grad_norm": 0.507344126701355, "learning_rate": 0.0005, "loss": 1.2991724014282227, "step": 47940 }, { "epoch": 3.7793456838785966, "grad_norm": 0.459547221660614, "learning_rate": 0.0005, "loss": 1.2296161651611328, "step": 47941 }, { "epoch": 3.7794245171462357, "grad_norm": 0.5134155750274658, "learning_rate": 0.0005, "loss": 1.1404478549957275, "step": 47942 }, { "epoch": 3.779503350413875, "grad_norm": 0.44043925404548645, "learning_rate": 0.0005, "loss": 1.274457335472107, "step": 47943 }, { "epoch": 3.7795821836815136, "grad_norm": 0.4632313847541809, "learning_rate": 0.0005, "loss": 1.1775178909301758, "step": 47944 }, { "epoch": 3.7796610169491527, "grad_norm": 0.7128867506980896, "learning_rate": 0.0005, "loss": 1.2095316648483276, "step": 47945 }, { "epoch": 3.7797398502167914, "grad_norm": 0.49480870366096497, "learning_rate": 0.0005, "loss": 1.1475803852081299, "step": 47946 }, { "epoch": 3.7798186834844305, "grad_norm": 0.4467458128929138, "learning_rate": 0.0005, "loss": 1.126661777496338, "step": 47947 }, { "epoch": 3.7798975167520696, "grad_norm": 0.46251180768013, "learning_rate": 0.0005, "loss": 1.2022337913513184, "step": 47948 }, { "epoch": 3.7799763500197083, "grad_norm": 0.49035733938217163, "learning_rate": 0.0005, "loss": 1.1352771520614624, "step": 47949 }, { "epoch": 3.780055183287347, "grad_norm": 0.7010133266448975, "learning_rate": 0.0005, "loss": 1.3055298328399658, "step": 47950 }, { "epoch": 3.780134016554986, "grad_norm": 0.6339365839958191, "learning_rate": 0.0005, "loss": 1.3825573921203613, "step": 47951 }, { "epoch": 3.7802128498226253, "grad_norm": 0.49171122908592224, "learning_rate": 0.0005, "loss": 1.237862467765808, "step": 47952 }, { "epoch": 3.780291683090264, "grad_norm": 0.46570873260498047, "learning_rate": 0.0005, "loss": 1.155527114868164, "step": 47953 }, { "epoch": 3.780370516357903, "grad_norm": 0.46216365694999695, "learning_rate": 0.0005, "loss": 1.13516366481781, "step": 47954 }, { "epoch": 3.780449349625542, "grad_norm": 0.49518847465515137, "learning_rate": 0.0005, "loss": 1.121401309967041, "step": 47955 }, { "epoch": 3.780528182893181, "grad_norm": 0.5139673948287964, "learning_rate": 0.0005, "loss": 1.161257028579712, "step": 47956 }, { "epoch": 3.78060701616082, "grad_norm": 0.5014046430587769, "learning_rate": 0.0005, "loss": 1.2481197118759155, "step": 47957 }, { "epoch": 3.7806858494284588, "grad_norm": 0.3994830846786499, "learning_rate": 0.0005, "loss": 1.0273301601409912, "step": 47958 }, { "epoch": 3.780764682696098, "grad_norm": 0.5249672532081604, "learning_rate": 0.0005, "loss": 1.0888943672180176, "step": 47959 }, { "epoch": 3.7808435159637366, "grad_norm": 0.5284304618835449, "learning_rate": 0.0005, "loss": 1.2742149829864502, "step": 47960 }, { "epoch": 3.7809223492313757, "grad_norm": 0.4578939974308014, "learning_rate": 0.0005, "loss": 1.235205888748169, "step": 47961 }, { "epoch": 3.781001182499015, "grad_norm": 0.4694662392139435, "learning_rate": 0.0005, "loss": 1.2935839891433716, "step": 47962 }, { "epoch": 3.7810800157666535, "grad_norm": 0.5445913076400757, "learning_rate": 0.0005, "loss": 1.340175986289978, "step": 47963 }, { "epoch": 3.7811588490342922, "grad_norm": 0.5778818130493164, "learning_rate": 0.0005, "loss": 1.42106032371521, "step": 47964 }, { "epoch": 3.7812376823019314, "grad_norm": 0.4727672040462494, "learning_rate": 0.0005, "loss": 1.0843178033828735, "step": 47965 }, { "epoch": 3.7813165155695705, "grad_norm": 0.5272374153137207, "learning_rate": 0.0005, "loss": 1.2152618169784546, "step": 47966 }, { "epoch": 3.781395348837209, "grad_norm": 0.5049267411231995, "learning_rate": 0.0005, "loss": 1.2319265604019165, "step": 47967 }, { "epoch": 3.7814741821048483, "grad_norm": 0.5266788005828857, "learning_rate": 0.0005, "loss": 1.154928207397461, "step": 47968 }, { "epoch": 3.781553015372487, "grad_norm": 0.5972846150398254, "learning_rate": 0.0005, "loss": 1.1860251426696777, "step": 47969 }, { "epoch": 3.781631848640126, "grad_norm": 0.4728131890296936, "learning_rate": 0.0005, "loss": 1.1099542379379272, "step": 47970 }, { "epoch": 3.7817106819077653, "grad_norm": 0.8176379799842834, "learning_rate": 0.0005, "loss": 1.3199729919433594, "step": 47971 }, { "epoch": 3.781789515175404, "grad_norm": 0.45374754071235657, "learning_rate": 0.0005, "loss": 1.1940228939056396, "step": 47972 }, { "epoch": 3.781868348443043, "grad_norm": 0.5606823563575745, "learning_rate": 0.0005, "loss": 1.2807260751724243, "step": 47973 }, { "epoch": 3.781947181710682, "grad_norm": 1.7731040716171265, "learning_rate": 0.0005, "loss": 1.3071036338806152, "step": 47974 }, { "epoch": 3.782026014978321, "grad_norm": 0.5689516663551331, "learning_rate": 0.0005, "loss": 1.2379333972930908, "step": 47975 }, { "epoch": 3.78210484824596, "grad_norm": 0.4666524827480316, "learning_rate": 0.0005, "loss": 1.178056240081787, "step": 47976 }, { "epoch": 3.7821836815135987, "grad_norm": 0.7089977264404297, "learning_rate": 0.0005, "loss": 1.2062759399414062, "step": 47977 }, { "epoch": 3.7822625147812374, "grad_norm": 0.44197607040405273, "learning_rate": 0.0005, "loss": 1.2017929553985596, "step": 47978 }, { "epoch": 3.7823413480488766, "grad_norm": 0.6303586363792419, "learning_rate": 0.0005, "loss": 1.1383036375045776, "step": 47979 }, { "epoch": 3.7824201813165157, "grad_norm": 0.4376752972602844, "learning_rate": 0.0005, "loss": 1.1651577949523926, "step": 47980 }, { "epoch": 3.7824990145841544, "grad_norm": 0.6379732489585876, "learning_rate": 0.0005, "loss": 1.2983747720718384, "step": 47981 }, { "epoch": 3.7825778478517935, "grad_norm": 0.6106238961219788, "learning_rate": 0.0005, "loss": 1.2323124408721924, "step": 47982 }, { "epoch": 3.782656681119432, "grad_norm": 0.5866982936859131, "learning_rate": 0.0005, "loss": 1.2473335266113281, "step": 47983 }, { "epoch": 3.7827355143870713, "grad_norm": 0.6033420562744141, "learning_rate": 0.0005, "loss": 1.1598937511444092, "step": 47984 }, { "epoch": 3.7828143476547105, "grad_norm": 0.44823193550109863, "learning_rate": 0.0005, "loss": 0.9441664218902588, "step": 47985 }, { "epoch": 3.782893180922349, "grad_norm": 0.7444949150085449, "learning_rate": 0.0005, "loss": 1.2687472105026245, "step": 47986 }, { "epoch": 3.7829720141899883, "grad_norm": 0.4269304573535919, "learning_rate": 0.0005, "loss": 0.9834905862808228, "step": 47987 }, { "epoch": 3.783050847457627, "grad_norm": 0.6661505699157715, "learning_rate": 0.0005, "loss": 1.287712574005127, "step": 47988 }, { "epoch": 3.783129680725266, "grad_norm": 0.4990527033805847, "learning_rate": 0.0005, "loss": 1.1841472387313843, "step": 47989 }, { "epoch": 3.7832085139929053, "grad_norm": 0.37913283705711365, "learning_rate": 0.0005, "loss": 1.0132497549057007, "step": 47990 }, { "epoch": 3.783287347260544, "grad_norm": 0.46230369806289673, "learning_rate": 0.0005, "loss": 1.1144211292266846, "step": 47991 }, { "epoch": 3.7833661805281826, "grad_norm": 0.4714263677597046, "learning_rate": 0.0005, "loss": 1.128989815711975, "step": 47992 }, { "epoch": 3.7834450137958218, "grad_norm": 0.6227521896362305, "learning_rate": 0.0005, "loss": 1.2784314155578613, "step": 47993 }, { "epoch": 3.783523847063461, "grad_norm": 0.5933314561843872, "learning_rate": 0.0005, "loss": 1.2814502716064453, "step": 47994 }, { "epoch": 3.7836026803310996, "grad_norm": 0.49521616101264954, "learning_rate": 0.0005, "loss": 1.1685736179351807, "step": 47995 }, { "epoch": 3.7836815135987387, "grad_norm": 0.4597022235393524, "learning_rate": 0.0005, "loss": 1.2508490085601807, "step": 47996 }, { "epoch": 3.7837603468663774, "grad_norm": 0.489608496427536, "learning_rate": 0.0005, "loss": 1.1356377601623535, "step": 47997 }, { "epoch": 3.7838391801340165, "grad_norm": 0.625023603439331, "learning_rate": 0.0005, "loss": 1.1066977977752686, "step": 47998 }, { "epoch": 3.7839180134016557, "grad_norm": 0.6848870515823364, "learning_rate": 0.0005, "loss": 1.1660730838775635, "step": 47999 }, { "epoch": 3.7839968466692944, "grad_norm": 0.45953109860420227, "learning_rate": 0.0005, "loss": 1.3540453910827637, "step": 48000 }, { "epoch": 3.7840756799369335, "grad_norm": 0.6293575763702393, "learning_rate": 0.0005, "loss": 1.2364833354949951, "step": 48001 }, { "epoch": 3.784154513204572, "grad_norm": 0.45765620470046997, "learning_rate": 0.0005, "loss": 1.0772838592529297, "step": 48002 }, { "epoch": 3.7842333464722113, "grad_norm": 0.5366926789283752, "learning_rate": 0.0005, "loss": 1.2710323333740234, "step": 48003 }, { "epoch": 3.7843121797398505, "grad_norm": 0.561880350112915, "learning_rate": 0.0005, "loss": 1.2184488773345947, "step": 48004 }, { "epoch": 3.784391013007489, "grad_norm": 0.46334800124168396, "learning_rate": 0.0005, "loss": 1.154334306716919, "step": 48005 }, { "epoch": 3.784469846275128, "grad_norm": 0.45647767186164856, "learning_rate": 0.0005, "loss": 1.0969841480255127, "step": 48006 }, { "epoch": 3.784548679542767, "grad_norm": 0.45634952187538147, "learning_rate": 0.0005, "loss": 1.0676791667938232, "step": 48007 }, { "epoch": 3.784627512810406, "grad_norm": 0.4784339964389801, "learning_rate": 0.0005, "loss": 1.0090327262878418, "step": 48008 }, { "epoch": 3.784706346078045, "grad_norm": 0.4414282739162445, "learning_rate": 0.0005, "loss": 1.230820655822754, "step": 48009 }, { "epoch": 3.784785179345684, "grad_norm": 0.4781297743320465, "learning_rate": 0.0005, "loss": 1.2939765453338623, "step": 48010 }, { "epoch": 3.7848640126133226, "grad_norm": 0.4499812424182892, "learning_rate": 0.0005, "loss": 1.06680428981781, "step": 48011 }, { "epoch": 3.7849428458809617, "grad_norm": 0.4668176472187042, "learning_rate": 0.0005, "loss": 1.0124824047088623, "step": 48012 }, { "epoch": 3.785021679148601, "grad_norm": 0.5297838449478149, "learning_rate": 0.0005, "loss": 1.3031291961669922, "step": 48013 }, { "epoch": 3.7851005124162396, "grad_norm": 0.5059143900871277, "learning_rate": 0.0005, "loss": 1.1744225025177002, "step": 48014 }, { "epoch": 3.7851793456838787, "grad_norm": 0.435882568359375, "learning_rate": 0.0005, "loss": 1.07487952709198, "step": 48015 }, { "epoch": 3.7852581789515174, "grad_norm": 0.5774083733558655, "learning_rate": 0.0005, "loss": 1.2949312925338745, "step": 48016 }, { "epoch": 3.7853370122191565, "grad_norm": 0.4634556472301483, "learning_rate": 0.0005, "loss": 0.9970288276672363, "step": 48017 }, { "epoch": 3.7854158454867957, "grad_norm": 0.4546945095062256, "learning_rate": 0.0005, "loss": 1.086142659187317, "step": 48018 }, { "epoch": 3.7854946787544344, "grad_norm": 0.6352147459983826, "learning_rate": 0.0005, "loss": 1.2751566171646118, "step": 48019 }, { "epoch": 3.785573512022073, "grad_norm": 0.5448227524757385, "learning_rate": 0.0005, "loss": 1.089491605758667, "step": 48020 }, { "epoch": 3.785652345289712, "grad_norm": 0.6736223697662354, "learning_rate": 0.0005, "loss": 1.2002923488616943, "step": 48021 }, { "epoch": 3.7857311785573513, "grad_norm": 0.5358911752700806, "learning_rate": 0.0005, "loss": 1.161902904510498, "step": 48022 }, { "epoch": 3.78581001182499, "grad_norm": 0.5005374550819397, "learning_rate": 0.0005, "loss": 1.2947161197662354, "step": 48023 }, { "epoch": 3.785888845092629, "grad_norm": 0.5371055603027344, "learning_rate": 0.0005, "loss": 1.252436637878418, "step": 48024 }, { "epoch": 3.785967678360268, "grad_norm": 0.4506056606769562, "learning_rate": 0.0005, "loss": 1.3371436595916748, "step": 48025 }, { "epoch": 3.786046511627907, "grad_norm": 0.5435937643051147, "learning_rate": 0.0005, "loss": 1.2680559158325195, "step": 48026 }, { "epoch": 3.786125344895546, "grad_norm": 0.47385406494140625, "learning_rate": 0.0005, "loss": 1.0446085929870605, "step": 48027 }, { "epoch": 3.7862041781631848, "grad_norm": 0.49411216378211975, "learning_rate": 0.0005, "loss": 1.2435749769210815, "step": 48028 }, { "epoch": 3.786283011430824, "grad_norm": 0.4860209822654724, "learning_rate": 0.0005, "loss": 1.3912327289581299, "step": 48029 }, { "epoch": 3.7863618446984626, "grad_norm": 0.5704952478408813, "learning_rate": 0.0005, "loss": 1.1106228828430176, "step": 48030 }, { "epoch": 3.7864406779661017, "grad_norm": 0.4280773103237152, "learning_rate": 0.0005, "loss": 1.1526449918746948, "step": 48031 }, { "epoch": 3.786519511233741, "grad_norm": 0.5863944292068481, "learning_rate": 0.0005, "loss": 1.1972911357879639, "step": 48032 }, { "epoch": 3.7865983445013796, "grad_norm": 0.5601000785827637, "learning_rate": 0.0005, "loss": 1.092392921447754, "step": 48033 }, { "epoch": 3.7866771777690187, "grad_norm": 0.4768664240837097, "learning_rate": 0.0005, "loss": 1.1315350532531738, "step": 48034 }, { "epoch": 3.7867560110366574, "grad_norm": 0.560394823551178, "learning_rate": 0.0005, "loss": 1.1772959232330322, "step": 48035 }, { "epoch": 3.7868348443042965, "grad_norm": 0.46618548035621643, "learning_rate": 0.0005, "loss": 1.1465479135513306, "step": 48036 }, { "epoch": 3.786913677571935, "grad_norm": 0.5216024518013, "learning_rate": 0.0005, "loss": 1.3675581216812134, "step": 48037 }, { "epoch": 3.7869925108395743, "grad_norm": 0.42729559540748596, "learning_rate": 0.0005, "loss": 1.0947175025939941, "step": 48038 }, { "epoch": 3.787071344107213, "grad_norm": 0.45026272535324097, "learning_rate": 0.0005, "loss": 1.1174111366271973, "step": 48039 }, { "epoch": 3.787150177374852, "grad_norm": 0.5996105670928955, "learning_rate": 0.0005, "loss": 1.2488986253738403, "step": 48040 }, { "epoch": 3.7872290106424913, "grad_norm": 0.4450187385082245, "learning_rate": 0.0005, "loss": 1.272249698638916, "step": 48041 }, { "epoch": 3.78730784391013, "grad_norm": 0.423418253660202, "learning_rate": 0.0005, "loss": 1.0643491744995117, "step": 48042 }, { "epoch": 3.787386677177769, "grad_norm": 0.4485737383365631, "learning_rate": 0.0005, "loss": 1.1046841144561768, "step": 48043 }, { "epoch": 3.787465510445408, "grad_norm": 0.4100914001464844, "learning_rate": 0.0005, "loss": 1.141964316368103, "step": 48044 }, { "epoch": 3.787544343713047, "grad_norm": 0.3822161555290222, "learning_rate": 0.0005, "loss": 1.0671491622924805, "step": 48045 }, { "epoch": 3.787623176980686, "grad_norm": 0.5110607743263245, "learning_rate": 0.0005, "loss": 1.224370002746582, "step": 48046 }, { "epoch": 3.7877020102483248, "grad_norm": 0.5798700451850891, "learning_rate": 0.0005, "loss": 1.2055103778839111, "step": 48047 }, { "epoch": 3.787780843515964, "grad_norm": 0.5411360859870911, "learning_rate": 0.0005, "loss": 1.1058745384216309, "step": 48048 }, { "epoch": 3.7878596767836026, "grad_norm": 0.5361284017562866, "learning_rate": 0.0005, "loss": 1.193656086921692, "step": 48049 }, { "epoch": 3.7879385100512417, "grad_norm": 0.5032201409339905, "learning_rate": 0.0005, "loss": 1.1631176471710205, "step": 48050 }, { "epoch": 3.7880173433188804, "grad_norm": 0.5527970790863037, "learning_rate": 0.0005, "loss": 1.1221582889556885, "step": 48051 }, { "epoch": 3.7880961765865195, "grad_norm": 0.5154818892478943, "learning_rate": 0.0005, "loss": 1.1088647842407227, "step": 48052 }, { "epoch": 3.7881750098541582, "grad_norm": 0.5639486908912659, "learning_rate": 0.0005, "loss": 0.9244296550750732, "step": 48053 }, { "epoch": 3.7882538431217974, "grad_norm": 0.5515919923782349, "learning_rate": 0.0005, "loss": 1.0967836380004883, "step": 48054 }, { "epoch": 3.7883326763894365, "grad_norm": 0.5647596716880798, "learning_rate": 0.0005, "loss": 1.2088723182678223, "step": 48055 }, { "epoch": 3.788411509657075, "grad_norm": 0.49118801951408386, "learning_rate": 0.0005, "loss": 1.1416325569152832, "step": 48056 }, { "epoch": 3.7884903429247143, "grad_norm": 0.4460252523422241, "learning_rate": 0.0005, "loss": 1.1123499870300293, "step": 48057 }, { "epoch": 3.788569176192353, "grad_norm": 0.5304508805274963, "learning_rate": 0.0005, "loss": 1.092970609664917, "step": 48058 }, { "epoch": 3.788648009459992, "grad_norm": 0.5242109894752502, "learning_rate": 0.0005, "loss": 1.2458174228668213, "step": 48059 }, { "epoch": 3.7887268427276313, "grad_norm": 0.4390992820262909, "learning_rate": 0.0005, "loss": 1.1029670238494873, "step": 48060 }, { "epoch": 3.78880567599527, "grad_norm": 0.5628998279571533, "learning_rate": 0.0005, "loss": 1.2201571464538574, "step": 48061 }, { "epoch": 3.788884509262909, "grad_norm": 0.5099627375602722, "learning_rate": 0.0005, "loss": 1.3455959558486938, "step": 48062 }, { "epoch": 3.788963342530548, "grad_norm": 0.4422186613082886, "learning_rate": 0.0005, "loss": 1.1508598327636719, "step": 48063 }, { "epoch": 3.789042175798187, "grad_norm": 0.4523790776729584, "learning_rate": 0.0005, "loss": 0.9804548025131226, "step": 48064 }, { "epoch": 3.789121009065826, "grad_norm": 0.3989700675010681, "learning_rate": 0.0005, "loss": 1.1226147413253784, "step": 48065 }, { "epoch": 3.7891998423334647, "grad_norm": 0.3938102722167969, "learning_rate": 0.0005, "loss": 1.0648561716079712, "step": 48066 }, { "epoch": 3.7892786756011034, "grad_norm": 0.4491158425807953, "learning_rate": 0.0005, "loss": 1.1726033687591553, "step": 48067 }, { "epoch": 3.7893575088687426, "grad_norm": 0.5772749781608582, "learning_rate": 0.0005, "loss": 1.2183631658554077, "step": 48068 }, { "epoch": 3.7894363421363817, "grad_norm": 0.7250996232032776, "learning_rate": 0.0005, "loss": 1.2212516069412231, "step": 48069 }, { "epoch": 3.7895151754040204, "grad_norm": 0.47472190856933594, "learning_rate": 0.0005, "loss": 1.303611159324646, "step": 48070 }, { "epoch": 3.7895940086716595, "grad_norm": 0.47821465134620667, "learning_rate": 0.0005, "loss": 1.0463857650756836, "step": 48071 }, { "epoch": 3.789672841939298, "grad_norm": 0.4674939215183258, "learning_rate": 0.0005, "loss": 0.9856300354003906, "step": 48072 }, { "epoch": 3.7897516752069373, "grad_norm": 0.48385536670684814, "learning_rate": 0.0005, "loss": 1.1879653930664062, "step": 48073 }, { "epoch": 3.7898305084745765, "grad_norm": 0.5170552134513855, "learning_rate": 0.0005, "loss": 0.9862648844718933, "step": 48074 }, { "epoch": 3.789909341742215, "grad_norm": 0.46179813146591187, "learning_rate": 0.0005, "loss": 1.2193971872329712, "step": 48075 }, { "epoch": 3.7899881750098543, "grad_norm": 0.4093474745750427, "learning_rate": 0.0005, "loss": 1.0607552528381348, "step": 48076 }, { "epoch": 3.790067008277493, "grad_norm": 0.6886314749717712, "learning_rate": 0.0005, "loss": 1.1277518272399902, "step": 48077 }, { "epoch": 3.790145841545132, "grad_norm": 0.48423656821250916, "learning_rate": 0.0005, "loss": 1.2469968795776367, "step": 48078 }, { "epoch": 3.7902246748127713, "grad_norm": 0.49903231859207153, "learning_rate": 0.0005, "loss": 1.1457679271697998, "step": 48079 }, { "epoch": 3.79030350808041, "grad_norm": 0.4622763395309448, "learning_rate": 0.0005, "loss": 1.2810688018798828, "step": 48080 }, { "epoch": 3.7903823413480486, "grad_norm": 0.5834469199180603, "learning_rate": 0.0005, "loss": 1.2405332326889038, "step": 48081 }, { "epoch": 3.7904611746156878, "grad_norm": 0.42474544048309326, "learning_rate": 0.0005, "loss": 1.0534491539001465, "step": 48082 }, { "epoch": 3.790540007883327, "grad_norm": 0.45542070269584656, "learning_rate": 0.0005, "loss": 1.1149415969848633, "step": 48083 }, { "epoch": 3.7906188411509656, "grad_norm": 0.4616243541240692, "learning_rate": 0.0005, "loss": 1.0961761474609375, "step": 48084 }, { "epoch": 3.7906976744186047, "grad_norm": 0.682536244392395, "learning_rate": 0.0005, "loss": 1.2269461154937744, "step": 48085 }, { "epoch": 3.7907765076862434, "grad_norm": 0.5373389720916748, "learning_rate": 0.0005, "loss": 1.1806256771087646, "step": 48086 }, { "epoch": 3.7908553409538825, "grad_norm": 0.5267702341079712, "learning_rate": 0.0005, "loss": 1.0755736827850342, "step": 48087 }, { "epoch": 3.7909341742215217, "grad_norm": 0.6932588219642639, "learning_rate": 0.0005, "loss": 1.3453253507614136, "step": 48088 }, { "epoch": 3.7910130074891604, "grad_norm": 0.5202694535255432, "learning_rate": 0.0005, "loss": 1.273418664932251, "step": 48089 }, { "epoch": 3.7910918407567995, "grad_norm": 0.5301429629325867, "learning_rate": 0.0005, "loss": 0.9879845976829529, "step": 48090 }, { "epoch": 3.791170674024438, "grad_norm": 0.5318194031715393, "learning_rate": 0.0005, "loss": 1.0579805374145508, "step": 48091 }, { "epoch": 3.7912495072920773, "grad_norm": 0.679032027721405, "learning_rate": 0.0005, "loss": 1.0921127796173096, "step": 48092 }, { "epoch": 3.7913283405597165, "grad_norm": 0.5008782148361206, "learning_rate": 0.0005, "loss": 1.206362009048462, "step": 48093 }, { "epoch": 3.791407173827355, "grad_norm": 0.4772677719593048, "learning_rate": 0.0005, "loss": 1.20951247215271, "step": 48094 }, { "epoch": 3.791486007094994, "grad_norm": 0.5662356019020081, "learning_rate": 0.0005, "loss": 1.4501376152038574, "step": 48095 }, { "epoch": 3.791564840362633, "grad_norm": 0.608423113822937, "learning_rate": 0.0005, "loss": 1.275266408920288, "step": 48096 }, { "epoch": 3.791643673630272, "grad_norm": 0.5041843056678772, "learning_rate": 0.0005, "loss": 1.1306848526000977, "step": 48097 }, { "epoch": 3.791722506897911, "grad_norm": 0.42158395051956177, "learning_rate": 0.0005, "loss": 1.0940815210342407, "step": 48098 }, { "epoch": 3.79180134016555, "grad_norm": 0.5862001180648804, "learning_rate": 0.0005, "loss": 1.2484362125396729, "step": 48099 }, { "epoch": 3.7918801734331886, "grad_norm": 0.4307454228401184, "learning_rate": 0.0005, "loss": 0.9859830737113953, "step": 48100 }, { "epoch": 3.7919590067008278, "grad_norm": 0.592951774597168, "learning_rate": 0.0005, "loss": 1.1763098239898682, "step": 48101 }, { "epoch": 3.792037839968467, "grad_norm": 0.5511530041694641, "learning_rate": 0.0005, "loss": 1.3286540508270264, "step": 48102 }, { "epoch": 3.7921166732361056, "grad_norm": 0.5622077584266663, "learning_rate": 0.0005, "loss": 1.1323293447494507, "step": 48103 }, { "epoch": 3.7921955065037447, "grad_norm": 0.5930764079093933, "learning_rate": 0.0005, "loss": 1.131956934928894, "step": 48104 }, { "epoch": 3.7922743397713834, "grad_norm": 0.5788667798042297, "learning_rate": 0.0005, "loss": 1.2739739418029785, "step": 48105 }, { "epoch": 3.7923531730390225, "grad_norm": 0.9895085692405701, "learning_rate": 0.0005, "loss": 1.3030874729156494, "step": 48106 }, { "epoch": 3.7924320063066617, "grad_norm": 0.3984651267528534, "learning_rate": 0.0005, "loss": 1.0380706787109375, "step": 48107 }, { "epoch": 3.7925108395743004, "grad_norm": 0.43448296189308167, "learning_rate": 0.0005, "loss": 1.135172963142395, "step": 48108 }, { "epoch": 3.792589672841939, "grad_norm": 0.46136966347694397, "learning_rate": 0.0005, "loss": 1.0579288005828857, "step": 48109 }, { "epoch": 3.792668506109578, "grad_norm": 0.5601168870925903, "learning_rate": 0.0005, "loss": 1.2860724925994873, "step": 48110 }, { "epoch": 3.7927473393772173, "grad_norm": 0.4910474419593811, "learning_rate": 0.0005, "loss": 1.263227939605713, "step": 48111 }, { "epoch": 3.792826172644856, "grad_norm": 0.5588251352310181, "learning_rate": 0.0005, "loss": 1.2014950513839722, "step": 48112 }, { "epoch": 3.792905005912495, "grad_norm": 0.5296682119369507, "learning_rate": 0.0005, "loss": 1.4166818857192993, "step": 48113 }, { "epoch": 3.792983839180134, "grad_norm": 0.5392492413520813, "learning_rate": 0.0005, "loss": 1.100244402885437, "step": 48114 }, { "epoch": 3.793062672447773, "grad_norm": 0.4983564019203186, "learning_rate": 0.0005, "loss": 1.1188536882400513, "step": 48115 }, { "epoch": 3.793141505715412, "grad_norm": 0.4638676345348358, "learning_rate": 0.0005, "loss": 1.0759943723678589, "step": 48116 }, { "epoch": 3.7932203389830508, "grad_norm": 0.47884654998779297, "learning_rate": 0.0005, "loss": 0.9766256809234619, "step": 48117 }, { "epoch": 3.79329917225069, "grad_norm": 0.5049699544906616, "learning_rate": 0.0005, "loss": 1.1510000228881836, "step": 48118 }, { "epoch": 3.7933780055183286, "grad_norm": 0.5833833813667297, "learning_rate": 0.0005, "loss": 1.1129579544067383, "step": 48119 }, { "epoch": 3.7934568387859677, "grad_norm": 0.4500453770160675, "learning_rate": 0.0005, "loss": 1.0714585781097412, "step": 48120 }, { "epoch": 3.793535672053607, "grad_norm": 0.4662206768989563, "learning_rate": 0.0005, "loss": 1.1231600046157837, "step": 48121 }, { "epoch": 3.7936145053212456, "grad_norm": 0.46378907561302185, "learning_rate": 0.0005, "loss": 1.1902579069137573, "step": 48122 }, { "epoch": 3.7936933385888842, "grad_norm": 0.572266161441803, "learning_rate": 0.0005, "loss": 1.3328869342803955, "step": 48123 }, { "epoch": 3.7937721718565234, "grad_norm": 0.5004040598869324, "learning_rate": 0.0005, "loss": 1.010933518409729, "step": 48124 }, { "epoch": 3.7938510051241625, "grad_norm": 0.42127782106399536, "learning_rate": 0.0005, "loss": 1.0797683000564575, "step": 48125 }, { "epoch": 3.793929838391801, "grad_norm": 0.4931640028953552, "learning_rate": 0.0005, "loss": 1.4170844554901123, "step": 48126 }, { "epoch": 3.7940086716594403, "grad_norm": 0.4772448241710663, "learning_rate": 0.0005, "loss": 1.3648635149002075, "step": 48127 }, { "epoch": 3.794087504927079, "grad_norm": 0.5267611742019653, "learning_rate": 0.0005, "loss": 1.309072494506836, "step": 48128 }, { "epoch": 3.794166338194718, "grad_norm": 0.43278244137763977, "learning_rate": 0.0005, "loss": 1.1288251876831055, "step": 48129 }, { "epoch": 3.7942451714623573, "grad_norm": 0.4618726074695587, "learning_rate": 0.0005, "loss": 1.2349247932434082, "step": 48130 }, { "epoch": 3.794324004729996, "grad_norm": 0.4660140573978424, "learning_rate": 0.0005, "loss": 1.3175954818725586, "step": 48131 }, { "epoch": 3.794402837997635, "grad_norm": 0.48324063420295715, "learning_rate": 0.0005, "loss": 1.1579880714416504, "step": 48132 }, { "epoch": 3.794481671265274, "grad_norm": 0.5396285057067871, "learning_rate": 0.0005, "loss": 1.2689235210418701, "step": 48133 }, { "epoch": 3.794560504532913, "grad_norm": 0.6228976845741272, "learning_rate": 0.0005, "loss": 1.0055493116378784, "step": 48134 }, { "epoch": 3.794639337800552, "grad_norm": 0.41149479150772095, "learning_rate": 0.0005, "loss": 1.1098153591156006, "step": 48135 }, { "epoch": 3.7947181710681908, "grad_norm": 0.45190709829330444, "learning_rate": 0.0005, "loss": 1.1623295545578003, "step": 48136 }, { "epoch": 3.7947970043358294, "grad_norm": 0.5941452980041504, "learning_rate": 0.0005, "loss": 1.2449241876602173, "step": 48137 }, { "epoch": 3.7948758376034686, "grad_norm": 0.4627785384654999, "learning_rate": 0.0005, "loss": 1.1851733922958374, "step": 48138 }, { "epoch": 3.7949546708711077, "grad_norm": 0.471890926361084, "learning_rate": 0.0005, "loss": 1.3570117950439453, "step": 48139 }, { "epoch": 3.7950335041387464, "grad_norm": 0.5065486431121826, "learning_rate": 0.0005, "loss": 1.1190412044525146, "step": 48140 }, { "epoch": 3.7951123374063855, "grad_norm": 0.7122114896774292, "learning_rate": 0.0005, "loss": 0.9834836721420288, "step": 48141 }, { "epoch": 3.7951911706740242, "grad_norm": 0.51723313331604, "learning_rate": 0.0005, "loss": 1.087639331817627, "step": 48142 }, { "epoch": 3.7952700039416634, "grad_norm": 0.5890893936157227, "learning_rate": 0.0005, "loss": 0.9701299071311951, "step": 48143 }, { "epoch": 3.7953488372093025, "grad_norm": 0.5241338014602661, "learning_rate": 0.0005, "loss": 1.2966618537902832, "step": 48144 }, { "epoch": 3.795427670476941, "grad_norm": 0.5112950801849365, "learning_rate": 0.0005, "loss": 1.1287096738815308, "step": 48145 }, { "epoch": 3.7955065037445803, "grad_norm": 0.528531551361084, "learning_rate": 0.0005, "loss": 1.1898332834243774, "step": 48146 }, { "epoch": 3.795585337012219, "grad_norm": 0.5493085384368896, "learning_rate": 0.0005, "loss": 1.14274001121521, "step": 48147 }, { "epoch": 3.795664170279858, "grad_norm": 0.4717847406864166, "learning_rate": 0.0005, "loss": 1.1553750038146973, "step": 48148 }, { "epoch": 3.7957430035474973, "grad_norm": 0.5430933237075806, "learning_rate": 0.0005, "loss": 1.1062204837799072, "step": 48149 }, { "epoch": 3.795821836815136, "grad_norm": 0.4650270640850067, "learning_rate": 0.0005, "loss": 1.2674169540405273, "step": 48150 }, { "epoch": 3.795900670082775, "grad_norm": 0.5275084972381592, "learning_rate": 0.0005, "loss": 1.401361107826233, "step": 48151 }, { "epoch": 3.795979503350414, "grad_norm": 0.5341649651527405, "learning_rate": 0.0005, "loss": 1.0086443424224854, "step": 48152 }, { "epoch": 3.796058336618053, "grad_norm": 0.6713458299636841, "learning_rate": 0.0005, "loss": 1.1598708629608154, "step": 48153 }, { "epoch": 3.7961371698856916, "grad_norm": 0.46454405784606934, "learning_rate": 0.0005, "loss": 1.161699891090393, "step": 48154 }, { "epoch": 3.7962160031533307, "grad_norm": 0.5006502270698547, "learning_rate": 0.0005, "loss": 1.1354758739471436, "step": 48155 }, { "epoch": 3.7962948364209694, "grad_norm": 0.46783334016799927, "learning_rate": 0.0005, "loss": 1.2369601726531982, "step": 48156 }, { "epoch": 3.7963736696886086, "grad_norm": 0.5562689900398254, "learning_rate": 0.0005, "loss": 1.1076704263687134, "step": 48157 }, { "epoch": 3.7964525029562477, "grad_norm": 0.4666917026042938, "learning_rate": 0.0005, "loss": 1.112009882926941, "step": 48158 }, { "epoch": 3.7965313362238864, "grad_norm": 0.4936472773551941, "learning_rate": 0.0005, "loss": 1.298978328704834, "step": 48159 }, { "epoch": 3.7966101694915255, "grad_norm": 0.5436891317367554, "learning_rate": 0.0005, "loss": 1.3581476211547852, "step": 48160 }, { "epoch": 3.796689002759164, "grad_norm": 0.49651211500167847, "learning_rate": 0.0005, "loss": 1.047602653503418, "step": 48161 }, { "epoch": 3.7967678360268033, "grad_norm": 0.5068763494491577, "learning_rate": 0.0005, "loss": 1.219965934753418, "step": 48162 }, { "epoch": 3.7968466692944425, "grad_norm": 0.530794084072113, "learning_rate": 0.0005, "loss": 1.3739768266677856, "step": 48163 }, { "epoch": 3.796925502562081, "grad_norm": 0.503206193447113, "learning_rate": 0.0005, "loss": 1.122581958770752, "step": 48164 }, { "epoch": 3.7970043358297203, "grad_norm": 0.4358533024787903, "learning_rate": 0.0005, "loss": 1.1203052997589111, "step": 48165 }, { "epoch": 3.797083169097359, "grad_norm": 0.5716954469680786, "learning_rate": 0.0005, "loss": 1.1751883029937744, "step": 48166 }, { "epoch": 3.797162002364998, "grad_norm": 0.5890143513679504, "learning_rate": 0.0005, "loss": 1.1264125108718872, "step": 48167 }, { "epoch": 3.797240835632637, "grad_norm": 0.7523354291915894, "learning_rate": 0.0005, "loss": 1.3131343126296997, "step": 48168 }, { "epoch": 3.797319668900276, "grad_norm": 0.5230936408042908, "learning_rate": 0.0005, "loss": 1.2297325134277344, "step": 48169 }, { "epoch": 3.7973985021679146, "grad_norm": 0.5572518110275269, "learning_rate": 0.0005, "loss": 1.156288743019104, "step": 48170 }, { "epoch": 3.7974773354355538, "grad_norm": 0.5406946539878845, "learning_rate": 0.0005, "loss": 1.1209592819213867, "step": 48171 }, { "epoch": 3.797556168703193, "grad_norm": 0.6164455413818359, "learning_rate": 0.0005, "loss": 1.2184616327285767, "step": 48172 }, { "epoch": 3.7976350019708316, "grad_norm": 0.45452311635017395, "learning_rate": 0.0005, "loss": 1.250224232673645, "step": 48173 }, { "epoch": 3.7977138352384707, "grad_norm": 0.5262152552604675, "learning_rate": 0.0005, "loss": 1.1954600811004639, "step": 48174 }, { "epoch": 3.7977926685061094, "grad_norm": 0.5741869211196899, "learning_rate": 0.0005, "loss": 1.3091455698013306, "step": 48175 }, { "epoch": 3.7978715017737485, "grad_norm": 0.561183750629425, "learning_rate": 0.0005, "loss": 1.2027356624603271, "step": 48176 }, { "epoch": 3.7979503350413877, "grad_norm": 0.5762547254562378, "learning_rate": 0.0005, "loss": 1.2446585893630981, "step": 48177 }, { "epoch": 3.7980291683090264, "grad_norm": 0.5676718354225159, "learning_rate": 0.0005, "loss": 1.3692432641983032, "step": 48178 }, { "epoch": 3.7981080015766655, "grad_norm": 0.5894756317138672, "learning_rate": 0.0005, "loss": 1.3874857425689697, "step": 48179 }, { "epoch": 3.798186834844304, "grad_norm": 0.4708903133869171, "learning_rate": 0.0005, "loss": 1.157715916633606, "step": 48180 }, { "epoch": 3.7982656681119433, "grad_norm": 0.6029177904129028, "learning_rate": 0.0005, "loss": 1.3325988054275513, "step": 48181 }, { "epoch": 3.7983445013795825, "grad_norm": 0.49235764145851135, "learning_rate": 0.0005, "loss": 1.2052538394927979, "step": 48182 }, { "epoch": 3.798423334647221, "grad_norm": 0.5182831883430481, "learning_rate": 0.0005, "loss": 1.2157707214355469, "step": 48183 }, { "epoch": 3.79850216791486, "grad_norm": 0.5407344102859497, "learning_rate": 0.0005, "loss": 1.2847363948822021, "step": 48184 }, { "epoch": 3.798581001182499, "grad_norm": 0.4910661280155182, "learning_rate": 0.0005, "loss": 0.9752987027168274, "step": 48185 }, { "epoch": 3.798659834450138, "grad_norm": 0.47062256932258606, "learning_rate": 0.0005, "loss": 1.1505205631256104, "step": 48186 }, { "epoch": 3.798738667717777, "grad_norm": 0.5103561878204346, "learning_rate": 0.0005, "loss": 1.3293242454528809, "step": 48187 }, { "epoch": 3.798817500985416, "grad_norm": 0.4701986610889435, "learning_rate": 0.0005, "loss": 1.2859063148498535, "step": 48188 }, { "epoch": 3.7988963342530546, "grad_norm": 0.6820566058158875, "learning_rate": 0.0005, "loss": 1.2666378021240234, "step": 48189 }, { "epoch": 3.7989751675206938, "grad_norm": 0.4840141832828522, "learning_rate": 0.0005, "loss": 1.1919350624084473, "step": 48190 }, { "epoch": 3.799054000788333, "grad_norm": 0.492905855178833, "learning_rate": 0.0005, "loss": 1.2991259098052979, "step": 48191 }, { "epoch": 3.7991328340559716, "grad_norm": 0.6170927286148071, "learning_rate": 0.0005, "loss": 1.226872444152832, "step": 48192 }, { "epoch": 3.7992116673236107, "grad_norm": 0.5169700384140015, "learning_rate": 0.0005, "loss": 1.2562668323516846, "step": 48193 }, { "epoch": 3.7992905005912494, "grad_norm": 0.505968451499939, "learning_rate": 0.0005, "loss": 1.024864912033081, "step": 48194 }, { "epoch": 3.7993693338588885, "grad_norm": 0.4250519275665283, "learning_rate": 0.0005, "loss": 1.1245067119598389, "step": 48195 }, { "epoch": 3.7994481671265277, "grad_norm": 0.47449201345443726, "learning_rate": 0.0005, "loss": 1.0969791412353516, "step": 48196 }, { "epoch": 3.7995270003941664, "grad_norm": 0.5261832475662231, "learning_rate": 0.0005, "loss": 1.174096703529358, "step": 48197 }, { "epoch": 3.799605833661805, "grad_norm": 0.3711145222187042, "learning_rate": 0.0005, "loss": 0.9315657615661621, "step": 48198 }, { "epoch": 3.799684666929444, "grad_norm": 0.4849676191806793, "learning_rate": 0.0005, "loss": 1.1884315013885498, "step": 48199 }, { "epoch": 3.7997635001970833, "grad_norm": 0.6046687364578247, "learning_rate": 0.0005, "loss": 1.1301922798156738, "step": 48200 }, { "epoch": 3.799842333464722, "grad_norm": 0.5785557627677917, "learning_rate": 0.0005, "loss": 1.136318325996399, "step": 48201 }, { "epoch": 3.799921166732361, "grad_norm": 0.5279108285903931, "learning_rate": 0.0005, "loss": 1.28700852394104, "step": 48202 }, { "epoch": 3.8, "grad_norm": 0.45503556728363037, "learning_rate": 0.0005, "loss": 1.0802340507507324, "step": 48203 }, { "epoch": 3.800078833267639, "grad_norm": 0.707545816898346, "learning_rate": 0.0005, "loss": 1.4412715435028076, "step": 48204 }, { "epoch": 3.800157666535278, "grad_norm": 0.4364061951637268, "learning_rate": 0.0005, "loss": 1.1663659811019897, "step": 48205 }, { "epoch": 3.800236499802917, "grad_norm": 0.5201056003570557, "learning_rate": 0.0005, "loss": 1.3321294784545898, "step": 48206 }, { "epoch": 3.800315333070556, "grad_norm": 0.889646589756012, "learning_rate": 0.0005, "loss": 1.1625089645385742, "step": 48207 }, { "epoch": 3.8003941663381946, "grad_norm": 0.46002206206321716, "learning_rate": 0.0005, "loss": 1.1602699756622314, "step": 48208 }, { "epoch": 3.8004729996058337, "grad_norm": 0.5639060139656067, "learning_rate": 0.0005, "loss": 1.1963868141174316, "step": 48209 }, { "epoch": 3.800551832873473, "grad_norm": 0.43117859959602356, "learning_rate": 0.0005, "loss": 1.1727817058563232, "step": 48210 }, { "epoch": 3.8006306661411116, "grad_norm": 0.4526638388633728, "learning_rate": 0.0005, "loss": 1.128347635269165, "step": 48211 }, { "epoch": 3.8007094994087502, "grad_norm": 0.4242195785045624, "learning_rate": 0.0005, "loss": 1.195796251296997, "step": 48212 }, { "epoch": 3.8007883326763894, "grad_norm": 0.5412580370903015, "learning_rate": 0.0005, "loss": 1.1160874366760254, "step": 48213 }, { "epoch": 3.8008671659440285, "grad_norm": 0.4577104449272156, "learning_rate": 0.0005, "loss": 1.1374870538711548, "step": 48214 }, { "epoch": 3.800945999211667, "grad_norm": 0.42322486639022827, "learning_rate": 0.0005, "loss": 1.0315184593200684, "step": 48215 }, { "epoch": 3.8010248324793063, "grad_norm": 0.5557050704956055, "learning_rate": 0.0005, "loss": 1.2398158311843872, "step": 48216 }, { "epoch": 3.801103665746945, "grad_norm": 0.5395053029060364, "learning_rate": 0.0005, "loss": 1.1787772178649902, "step": 48217 }, { "epoch": 3.801182499014584, "grad_norm": 0.4561609625816345, "learning_rate": 0.0005, "loss": 1.100127935409546, "step": 48218 }, { "epoch": 3.8012613322822233, "grad_norm": 0.504352867603302, "learning_rate": 0.0005, "loss": 1.1817691326141357, "step": 48219 }, { "epoch": 3.801340165549862, "grad_norm": 0.3781316578388214, "learning_rate": 0.0005, "loss": 1.2156641483306885, "step": 48220 }, { "epoch": 3.801418998817501, "grad_norm": 0.5218794941902161, "learning_rate": 0.0005, "loss": 1.2084118127822876, "step": 48221 }, { "epoch": 3.80149783208514, "grad_norm": 0.484943687915802, "learning_rate": 0.0005, "loss": 1.0281476974487305, "step": 48222 }, { "epoch": 3.801576665352779, "grad_norm": 0.48457667231559753, "learning_rate": 0.0005, "loss": 1.0870001316070557, "step": 48223 }, { "epoch": 3.801655498620418, "grad_norm": 0.5669206380844116, "learning_rate": 0.0005, "loss": 1.4825239181518555, "step": 48224 }, { "epoch": 3.8017343318880568, "grad_norm": 0.4467225968837738, "learning_rate": 0.0005, "loss": 1.1505192518234253, "step": 48225 }, { "epoch": 3.8018131651556955, "grad_norm": 0.4992655813694, "learning_rate": 0.0005, "loss": 1.2153544425964355, "step": 48226 }, { "epoch": 3.8018919984233346, "grad_norm": 0.5145863890647888, "learning_rate": 0.0005, "loss": 1.1385608911514282, "step": 48227 }, { "epoch": 3.8019708316909737, "grad_norm": 0.5585551857948303, "learning_rate": 0.0005, "loss": 1.2562098503112793, "step": 48228 }, { "epoch": 3.8020496649586124, "grad_norm": 0.4606369435787201, "learning_rate": 0.0005, "loss": 1.0863555669784546, "step": 48229 }, { "epoch": 3.8021284982262515, "grad_norm": 0.4985605478286743, "learning_rate": 0.0005, "loss": 1.0786209106445312, "step": 48230 }, { "epoch": 3.8022073314938902, "grad_norm": 0.5855233073234558, "learning_rate": 0.0005, "loss": 1.171114206314087, "step": 48231 }, { "epoch": 3.8022861647615294, "grad_norm": 0.5308122634887695, "learning_rate": 0.0005, "loss": 1.218687891960144, "step": 48232 }, { "epoch": 3.8023649980291685, "grad_norm": 0.6827623248100281, "learning_rate": 0.0005, "loss": 1.259967565536499, "step": 48233 }, { "epoch": 3.802443831296807, "grad_norm": 0.5326621532440186, "learning_rate": 0.0005, "loss": 1.242088794708252, "step": 48234 }, { "epoch": 3.8025226645644463, "grad_norm": 0.5408157110214233, "learning_rate": 0.0005, "loss": 1.144223690032959, "step": 48235 }, { "epoch": 3.802601497832085, "grad_norm": 0.49854207038879395, "learning_rate": 0.0005, "loss": 1.1336150169372559, "step": 48236 }, { "epoch": 3.802680331099724, "grad_norm": 0.5157985687255859, "learning_rate": 0.0005, "loss": 1.1381913423538208, "step": 48237 }, { "epoch": 3.8027591643673633, "grad_norm": 0.6176673173904419, "learning_rate": 0.0005, "loss": 1.2208806276321411, "step": 48238 }, { "epoch": 3.802837997635002, "grad_norm": 0.6823315620422363, "learning_rate": 0.0005, "loss": 1.3072025775909424, "step": 48239 }, { "epoch": 3.8029168309026407, "grad_norm": 0.6233267784118652, "learning_rate": 0.0005, "loss": 1.2005395889282227, "step": 48240 }, { "epoch": 3.80299566417028, "grad_norm": 0.4662904441356659, "learning_rate": 0.0005, "loss": 1.1579792499542236, "step": 48241 }, { "epoch": 3.803074497437919, "grad_norm": 0.5662466883659363, "learning_rate": 0.0005, "loss": 1.2045371532440186, "step": 48242 }, { "epoch": 3.8031533307055576, "grad_norm": 0.4461383521556854, "learning_rate": 0.0005, "loss": 1.0100386142730713, "step": 48243 }, { "epoch": 3.8032321639731967, "grad_norm": 0.4720412492752075, "learning_rate": 0.0005, "loss": 1.0819023847579956, "step": 48244 }, { "epoch": 3.8033109972408354, "grad_norm": 0.4618379473686218, "learning_rate": 0.0005, "loss": 1.2615538835525513, "step": 48245 }, { "epoch": 3.8033898305084746, "grad_norm": 0.5373113751411438, "learning_rate": 0.0005, "loss": 1.0732377767562866, "step": 48246 }, { "epoch": 3.8034686637761137, "grad_norm": 0.5418069958686829, "learning_rate": 0.0005, "loss": 1.2862720489501953, "step": 48247 }, { "epoch": 3.8035474970437524, "grad_norm": 0.4769953191280365, "learning_rate": 0.0005, "loss": 1.1549614667892456, "step": 48248 }, { "epoch": 3.8036263303113915, "grad_norm": 0.5795477628707886, "learning_rate": 0.0005, "loss": 1.2022123336791992, "step": 48249 }, { "epoch": 3.80370516357903, "grad_norm": 0.6088001728057861, "learning_rate": 0.0005, "loss": 1.2282705307006836, "step": 48250 }, { "epoch": 3.8037839968466693, "grad_norm": 0.5112094879150391, "learning_rate": 0.0005, "loss": 1.1856203079223633, "step": 48251 }, { "epoch": 3.8038628301143085, "grad_norm": 0.4472254514694214, "learning_rate": 0.0005, "loss": 1.1055219173431396, "step": 48252 }, { "epoch": 3.803941663381947, "grad_norm": 0.5655097365379333, "learning_rate": 0.0005, "loss": 1.4009157419204712, "step": 48253 }, { "epoch": 3.804020496649586, "grad_norm": 0.4745621085166931, "learning_rate": 0.0005, "loss": 1.0587553977966309, "step": 48254 }, { "epoch": 3.804099329917225, "grad_norm": 0.4977920651435852, "learning_rate": 0.0005, "loss": 1.1458549499511719, "step": 48255 }, { "epoch": 3.804178163184864, "grad_norm": 0.5617035031318665, "learning_rate": 0.0005, "loss": 1.2984859943389893, "step": 48256 }, { "epoch": 3.804256996452503, "grad_norm": 0.6569923162460327, "learning_rate": 0.0005, "loss": 1.2238962650299072, "step": 48257 }, { "epoch": 3.804335829720142, "grad_norm": 0.5694116353988647, "learning_rate": 0.0005, "loss": 1.127071499824524, "step": 48258 }, { "epoch": 3.8044146629877806, "grad_norm": 0.5418955087661743, "learning_rate": 0.0005, "loss": 1.2352443933486938, "step": 48259 }, { "epoch": 3.8044934962554198, "grad_norm": 0.5160761475563049, "learning_rate": 0.0005, "loss": 1.1646732091903687, "step": 48260 }, { "epoch": 3.804572329523059, "grad_norm": 0.3996511995792389, "learning_rate": 0.0005, "loss": 1.1300753355026245, "step": 48261 }, { "epoch": 3.8046511627906976, "grad_norm": 0.5165314674377441, "learning_rate": 0.0005, "loss": 1.1106010675430298, "step": 48262 }, { "epoch": 3.8047299960583367, "grad_norm": 0.5870198011398315, "learning_rate": 0.0005, "loss": 1.1915900707244873, "step": 48263 }, { "epoch": 3.8048088293259754, "grad_norm": 0.41770946979522705, "learning_rate": 0.0005, "loss": 1.146667718887329, "step": 48264 }, { "epoch": 3.8048876625936146, "grad_norm": 0.4785807132720947, "learning_rate": 0.0005, "loss": 1.1319167613983154, "step": 48265 }, { "epoch": 3.8049664958612537, "grad_norm": 0.6000905632972717, "learning_rate": 0.0005, "loss": 1.3002424240112305, "step": 48266 }, { "epoch": 3.8050453291288924, "grad_norm": 0.4784858822822571, "learning_rate": 0.0005, "loss": 1.3959420919418335, "step": 48267 }, { "epoch": 3.805124162396531, "grad_norm": 0.48463907837867737, "learning_rate": 0.0005, "loss": 1.1610639095306396, "step": 48268 }, { "epoch": 3.80520299566417, "grad_norm": 0.488793283700943, "learning_rate": 0.0005, "loss": 1.1624295711517334, "step": 48269 }, { "epoch": 3.8052818289318093, "grad_norm": 0.5400626063346863, "learning_rate": 0.0005, "loss": 1.2287702560424805, "step": 48270 }, { "epoch": 3.805360662199448, "grad_norm": 0.4457959830760956, "learning_rate": 0.0005, "loss": 1.2150487899780273, "step": 48271 }, { "epoch": 3.805439495467087, "grad_norm": 0.5294069051742554, "learning_rate": 0.0005, "loss": 1.3065752983093262, "step": 48272 }, { "epoch": 3.805518328734726, "grad_norm": 0.4215773046016693, "learning_rate": 0.0005, "loss": 1.041910171508789, "step": 48273 }, { "epoch": 3.805597162002365, "grad_norm": 0.464714378118515, "learning_rate": 0.0005, "loss": 1.1754539012908936, "step": 48274 }, { "epoch": 3.805675995270004, "grad_norm": 0.6503930687904358, "learning_rate": 0.0005, "loss": 1.287030577659607, "step": 48275 }, { "epoch": 3.805754828537643, "grad_norm": 0.42852911353111267, "learning_rate": 0.0005, "loss": 1.162360668182373, "step": 48276 }, { "epoch": 3.805833661805282, "grad_norm": 0.48609480261802673, "learning_rate": 0.0005, "loss": 1.1250065565109253, "step": 48277 }, { "epoch": 3.8059124950729206, "grad_norm": 0.5510438084602356, "learning_rate": 0.0005, "loss": 1.094117283821106, "step": 48278 }, { "epoch": 3.8059913283405598, "grad_norm": 0.4620376527309418, "learning_rate": 0.0005, "loss": 1.2037053108215332, "step": 48279 }, { "epoch": 3.806070161608199, "grad_norm": 0.42913687229156494, "learning_rate": 0.0005, "loss": 1.230032205581665, "step": 48280 }, { "epoch": 3.8061489948758376, "grad_norm": 0.502528190612793, "learning_rate": 0.0005, "loss": 1.0887420177459717, "step": 48281 }, { "epoch": 3.8062278281434767, "grad_norm": 0.43571510910987854, "learning_rate": 0.0005, "loss": 1.2887215614318848, "step": 48282 }, { "epoch": 3.8063066614111154, "grad_norm": 0.5737662315368652, "learning_rate": 0.0005, "loss": 1.050877332687378, "step": 48283 }, { "epoch": 3.8063854946787545, "grad_norm": 0.3646126985549927, "learning_rate": 0.0005, "loss": 1.0150771141052246, "step": 48284 }, { "epoch": 3.8064643279463932, "grad_norm": 0.5063933730125427, "learning_rate": 0.0005, "loss": 1.1032190322875977, "step": 48285 }, { "epoch": 3.8065431612140324, "grad_norm": 0.5706116557121277, "learning_rate": 0.0005, "loss": 1.3020758628845215, "step": 48286 }, { "epoch": 3.806621994481671, "grad_norm": 0.5269619822502136, "learning_rate": 0.0005, "loss": 1.1251492500305176, "step": 48287 }, { "epoch": 3.80670082774931, "grad_norm": 0.496642529964447, "learning_rate": 0.0005, "loss": 1.1831684112548828, "step": 48288 }, { "epoch": 3.8067796610169493, "grad_norm": 0.45459499955177307, "learning_rate": 0.0005, "loss": 1.101609230041504, "step": 48289 }, { "epoch": 3.806858494284588, "grad_norm": 0.4282739758491516, "learning_rate": 0.0005, "loss": 1.0930652618408203, "step": 48290 }, { "epoch": 3.806937327552227, "grad_norm": 0.5336503386497498, "learning_rate": 0.0005, "loss": 1.2388041019439697, "step": 48291 }, { "epoch": 3.807016160819866, "grad_norm": 0.4710195064544678, "learning_rate": 0.0005, "loss": 1.1402900218963623, "step": 48292 }, { "epoch": 3.807094994087505, "grad_norm": 0.46026065945625305, "learning_rate": 0.0005, "loss": 1.1918277740478516, "step": 48293 }, { "epoch": 3.807173827355144, "grad_norm": 0.3996748626232147, "learning_rate": 0.0005, "loss": 1.189692735671997, "step": 48294 }, { "epoch": 3.807252660622783, "grad_norm": 0.4521358609199524, "learning_rate": 0.0005, "loss": 1.126220941543579, "step": 48295 }, { "epoch": 3.807331493890422, "grad_norm": 0.46938586235046387, "learning_rate": 0.0005, "loss": 1.116713047027588, "step": 48296 }, { "epoch": 3.8074103271580606, "grad_norm": 0.49404704570770264, "learning_rate": 0.0005, "loss": 1.1215415000915527, "step": 48297 }, { "epoch": 3.8074891604256997, "grad_norm": 0.40434807538986206, "learning_rate": 0.0005, "loss": 0.9953259229660034, "step": 48298 }, { "epoch": 3.8075679936933384, "grad_norm": 0.48137813806533813, "learning_rate": 0.0005, "loss": 1.2626246213912964, "step": 48299 }, { "epoch": 3.8076468269609776, "grad_norm": 0.47066012024879456, "learning_rate": 0.0005, "loss": 1.2840020656585693, "step": 48300 }, { "epoch": 3.8077256602286162, "grad_norm": 0.5560045838356018, "learning_rate": 0.0005, "loss": 1.3522604703903198, "step": 48301 }, { "epoch": 3.8078044934962554, "grad_norm": 0.5504416227340698, "learning_rate": 0.0005, "loss": 1.113052248954773, "step": 48302 }, { "epoch": 3.8078833267638945, "grad_norm": 0.4805985391139984, "learning_rate": 0.0005, "loss": 1.2794272899627686, "step": 48303 }, { "epoch": 3.807962160031533, "grad_norm": 0.4518366754055023, "learning_rate": 0.0005, "loss": 1.1764861345291138, "step": 48304 }, { "epoch": 3.8080409932991723, "grad_norm": 0.4109339714050293, "learning_rate": 0.0005, "loss": 1.0709328651428223, "step": 48305 }, { "epoch": 3.808119826566811, "grad_norm": 0.6207947134971619, "learning_rate": 0.0005, "loss": 1.2827509641647339, "step": 48306 }, { "epoch": 3.80819865983445, "grad_norm": 0.5165005326271057, "learning_rate": 0.0005, "loss": 1.2894985675811768, "step": 48307 }, { "epoch": 3.8082774931020893, "grad_norm": 0.5298314690589905, "learning_rate": 0.0005, "loss": 1.0850074291229248, "step": 48308 }, { "epoch": 3.808356326369728, "grad_norm": 0.4686996042728424, "learning_rate": 0.0005, "loss": 1.0578932762145996, "step": 48309 }, { "epoch": 3.808435159637367, "grad_norm": 0.39864763617515564, "learning_rate": 0.0005, "loss": 0.8617202043533325, "step": 48310 }, { "epoch": 3.808513992905006, "grad_norm": 0.5011027455329895, "learning_rate": 0.0005, "loss": 1.1445136070251465, "step": 48311 }, { "epoch": 3.808592826172645, "grad_norm": 0.5475748777389526, "learning_rate": 0.0005, "loss": 1.3495757579803467, "step": 48312 }, { "epoch": 3.808671659440284, "grad_norm": 0.501920759677887, "learning_rate": 0.0005, "loss": 1.056274652481079, "step": 48313 }, { "epoch": 3.8087504927079228, "grad_norm": 0.5326864719390869, "learning_rate": 0.0005, "loss": 1.1948161125183105, "step": 48314 }, { "epoch": 3.8088293259755615, "grad_norm": 0.39883357286453247, "learning_rate": 0.0005, "loss": 1.0017778873443604, "step": 48315 }, { "epoch": 3.8089081592432006, "grad_norm": 0.545522928237915, "learning_rate": 0.0005, "loss": 1.3045258522033691, "step": 48316 }, { "epoch": 3.8089869925108397, "grad_norm": 0.5631654262542725, "learning_rate": 0.0005, "loss": 1.2864773273468018, "step": 48317 }, { "epoch": 3.8090658257784784, "grad_norm": 0.5227222442626953, "learning_rate": 0.0005, "loss": 1.1694735288619995, "step": 48318 }, { "epoch": 3.8091446590461175, "grad_norm": 0.5059702396392822, "learning_rate": 0.0005, "loss": 1.2637732028961182, "step": 48319 }, { "epoch": 3.8092234923137562, "grad_norm": 0.4751337766647339, "learning_rate": 0.0005, "loss": 1.159255027770996, "step": 48320 }, { "epoch": 3.8093023255813954, "grad_norm": 0.530777096748352, "learning_rate": 0.0005, "loss": 1.375377893447876, "step": 48321 }, { "epoch": 3.8093811588490345, "grad_norm": 0.455549031496048, "learning_rate": 0.0005, "loss": 1.1015833616256714, "step": 48322 }, { "epoch": 3.809459992116673, "grad_norm": 0.526992678642273, "learning_rate": 0.0005, "loss": 1.2466933727264404, "step": 48323 }, { "epoch": 3.8095388253843123, "grad_norm": 0.5403574109077454, "learning_rate": 0.0005, "loss": 1.2922472953796387, "step": 48324 }, { "epoch": 3.809617658651951, "grad_norm": 0.47802630066871643, "learning_rate": 0.0005, "loss": 1.0537428855895996, "step": 48325 }, { "epoch": 3.80969649191959, "grad_norm": 0.4573720693588257, "learning_rate": 0.0005, "loss": 1.1564182043075562, "step": 48326 }, { "epoch": 3.8097753251872293, "grad_norm": 0.3953785002231598, "learning_rate": 0.0005, "loss": 1.006166934967041, "step": 48327 }, { "epoch": 3.809854158454868, "grad_norm": 0.5705884695053101, "learning_rate": 0.0005, "loss": 1.1224513053894043, "step": 48328 }, { "epoch": 3.8099329917225067, "grad_norm": 0.4274257719516754, "learning_rate": 0.0005, "loss": 1.1043016910552979, "step": 48329 }, { "epoch": 3.810011824990146, "grad_norm": 0.3987131714820862, "learning_rate": 0.0005, "loss": 0.9853307604789734, "step": 48330 }, { "epoch": 3.810090658257785, "grad_norm": 0.5758455395698547, "learning_rate": 0.0005, "loss": 1.299314022064209, "step": 48331 }, { "epoch": 3.8101694915254236, "grad_norm": 0.4577903747558594, "learning_rate": 0.0005, "loss": 1.3082740306854248, "step": 48332 }, { "epoch": 3.8102483247930627, "grad_norm": 0.45013123750686646, "learning_rate": 0.0005, "loss": 1.2174843549728394, "step": 48333 }, { "epoch": 3.8103271580607014, "grad_norm": 0.4446905255317688, "learning_rate": 0.0005, "loss": 0.9984090924263, "step": 48334 }, { "epoch": 3.8104059913283406, "grad_norm": 0.4915333092212677, "learning_rate": 0.0005, "loss": 1.2451610565185547, "step": 48335 }, { "epoch": 3.8104848245959797, "grad_norm": 0.4387122690677643, "learning_rate": 0.0005, "loss": 1.082368016242981, "step": 48336 }, { "epoch": 3.8105636578636184, "grad_norm": 0.49064746499061584, "learning_rate": 0.0005, "loss": 1.1318020820617676, "step": 48337 }, { "epoch": 3.8106424911312575, "grad_norm": 0.40256810188293457, "learning_rate": 0.0005, "loss": 1.1112319231033325, "step": 48338 }, { "epoch": 3.810721324398896, "grad_norm": 0.4202171266078949, "learning_rate": 0.0005, "loss": 1.2153714895248413, "step": 48339 }, { "epoch": 3.8108001576665353, "grad_norm": 0.5173393487930298, "learning_rate": 0.0005, "loss": 1.3164546489715576, "step": 48340 }, { "epoch": 3.8108789909341745, "grad_norm": 0.42079654335975647, "learning_rate": 0.0005, "loss": 1.0731430053710938, "step": 48341 }, { "epoch": 3.810957824201813, "grad_norm": 0.42956623435020447, "learning_rate": 0.0005, "loss": 0.8995916843414307, "step": 48342 }, { "epoch": 3.811036657469452, "grad_norm": 0.5562496185302734, "learning_rate": 0.0005, "loss": 1.2041847705841064, "step": 48343 }, { "epoch": 3.811115490737091, "grad_norm": 0.4415176510810852, "learning_rate": 0.0005, "loss": 1.2713823318481445, "step": 48344 }, { "epoch": 3.81119432400473, "grad_norm": 0.4654236137866974, "learning_rate": 0.0005, "loss": 1.1186797618865967, "step": 48345 }, { "epoch": 3.811273157272369, "grad_norm": 0.5478847026824951, "learning_rate": 0.0005, "loss": 1.138662338256836, "step": 48346 }, { "epoch": 3.811351990540008, "grad_norm": 0.5674982070922852, "learning_rate": 0.0005, "loss": 1.2525527477264404, "step": 48347 }, { "epoch": 3.8114308238076466, "grad_norm": 0.45680803060531616, "learning_rate": 0.0005, "loss": 1.1331888437271118, "step": 48348 }, { "epoch": 3.8115096570752858, "grad_norm": 0.44347190856933594, "learning_rate": 0.0005, "loss": 1.0851713418960571, "step": 48349 }, { "epoch": 3.811588490342925, "grad_norm": 0.4799504578113556, "learning_rate": 0.0005, "loss": 1.2091577053070068, "step": 48350 }, { "epoch": 3.8116673236105636, "grad_norm": 0.46515703201293945, "learning_rate": 0.0005, "loss": 1.1428717374801636, "step": 48351 }, { "epoch": 3.8117461568782027, "grad_norm": 0.4443461000919342, "learning_rate": 0.0005, "loss": 1.0688326358795166, "step": 48352 }, { "epoch": 3.8118249901458414, "grad_norm": 0.4940222203731537, "learning_rate": 0.0005, "loss": 1.3026459217071533, "step": 48353 }, { "epoch": 3.8119038234134806, "grad_norm": 0.5414372086524963, "learning_rate": 0.0005, "loss": 1.1573854684829712, "step": 48354 }, { "epoch": 3.8119826566811197, "grad_norm": 0.49843576550483704, "learning_rate": 0.0005, "loss": 1.2388994693756104, "step": 48355 }, { "epoch": 3.8120614899487584, "grad_norm": 0.48221513628959656, "learning_rate": 0.0005, "loss": 0.9561880826950073, "step": 48356 }, { "epoch": 3.812140323216397, "grad_norm": 0.5892772078514099, "learning_rate": 0.0005, "loss": 1.1201022863388062, "step": 48357 }, { "epoch": 3.812219156484036, "grad_norm": 0.4831302762031555, "learning_rate": 0.0005, "loss": 1.2422865629196167, "step": 48358 }, { "epoch": 3.8122979897516753, "grad_norm": 0.5863611102104187, "learning_rate": 0.0005, "loss": 1.2017364501953125, "step": 48359 }, { "epoch": 3.812376823019314, "grad_norm": 0.45729130506515503, "learning_rate": 0.0005, "loss": 1.0857967138290405, "step": 48360 }, { "epoch": 3.812455656286953, "grad_norm": 0.45039281249046326, "learning_rate": 0.0005, "loss": 1.0509657859802246, "step": 48361 }, { "epoch": 3.812534489554592, "grad_norm": 0.5215976238250732, "learning_rate": 0.0005, "loss": 1.1618083715438843, "step": 48362 }, { "epoch": 3.812613322822231, "grad_norm": 0.41429704427719116, "learning_rate": 0.0005, "loss": 1.2473645210266113, "step": 48363 }, { "epoch": 3.81269215608987, "grad_norm": 0.42763254046440125, "learning_rate": 0.0005, "loss": 1.212516188621521, "step": 48364 }, { "epoch": 3.812770989357509, "grad_norm": 0.8881706595420837, "learning_rate": 0.0005, "loss": 1.1927958726882935, "step": 48365 }, { "epoch": 3.812849822625148, "grad_norm": 0.5955455303192139, "learning_rate": 0.0005, "loss": 1.3353257179260254, "step": 48366 }, { "epoch": 3.8129286558927866, "grad_norm": 0.5289831161499023, "learning_rate": 0.0005, "loss": 1.0910964012145996, "step": 48367 }, { "epoch": 3.8130074891604258, "grad_norm": 0.4116501212120056, "learning_rate": 0.0005, "loss": 1.1034305095672607, "step": 48368 }, { "epoch": 3.813086322428065, "grad_norm": 0.4304262697696686, "learning_rate": 0.0005, "loss": 1.1398738622665405, "step": 48369 }, { "epoch": 3.8131651556957036, "grad_norm": 0.5924549698829651, "learning_rate": 0.0005, "loss": 1.1030454635620117, "step": 48370 }, { "epoch": 3.8132439889633423, "grad_norm": 0.4440430700778961, "learning_rate": 0.0005, "loss": 1.0375418663024902, "step": 48371 }, { "epoch": 3.8133228222309814, "grad_norm": 0.5590898394584656, "learning_rate": 0.0005, "loss": 1.2955350875854492, "step": 48372 }, { "epoch": 3.8134016554986205, "grad_norm": 0.46958568692207336, "learning_rate": 0.0005, "loss": 1.13480806350708, "step": 48373 }, { "epoch": 3.8134804887662592, "grad_norm": 0.6315056085586548, "learning_rate": 0.0005, "loss": 1.2037464380264282, "step": 48374 }, { "epoch": 3.8135593220338984, "grad_norm": 0.48370879888534546, "learning_rate": 0.0005, "loss": 1.1680097579956055, "step": 48375 }, { "epoch": 3.813638155301537, "grad_norm": 0.5645207762718201, "learning_rate": 0.0005, "loss": 1.109627366065979, "step": 48376 }, { "epoch": 3.813716988569176, "grad_norm": 0.5229706764221191, "learning_rate": 0.0005, "loss": 1.1238926649093628, "step": 48377 }, { "epoch": 3.8137958218368153, "grad_norm": 0.4807070195674896, "learning_rate": 0.0005, "loss": 1.1711980104446411, "step": 48378 }, { "epoch": 3.813874655104454, "grad_norm": 0.49013474583625793, "learning_rate": 0.0005, "loss": 1.0493547916412354, "step": 48379 }, { "epoch": 3.813953488372093, "grad_norm": 0.4030826985836029, "learning_rate": 0.0005, "loss": 1.1538392305374146, "step": 48380 }, { "epoch": 3.814032321639732, "grad_norm": 0.6293391585350037, "learning_rate": 0.0005, "loss": 1.3025195598602295, "step": 48381 }, { "epoch": 3.814111154907371, "grad_norm": 0.5311192274093628, "learning_rate": 0.0005, "loss": 1.2220816612243652, "step": 48382 }, { "epoch": 3.81418998817501, "grad_norm": 0.48138654232025146, "learning_rate": 0.0005, "loss": 1.2667615413665771, "step": 48383 }, { "epoch": 3.814268821442649, "grad_norm": 0.510028064250946, "learning_rate": 0.0005, "loss": 1.3574023246765137, "step": 48384 }, { "epoch": 3.8143476547102875, "grad_norm": 0.5181958079338074, "learning_rate": 0.0005, "loss": 1.0860304832458496, "step": 48385 }, { "epoch": 3.8144264879779266, "grad_norm": 0.38174906373023987, "learning_rate": 0.0005, "loss": 1.0159282684326172, "step": 48386 }, { "epoch": 3.8145053212455657, "grad_norm": 0.4261828064918518, "learning_rate": 0.0005, "loss": 1.1023313999176025, "step": 48387 }, { "epoch": 3.8145841545132044, "grad_norm": 0.37480980157852173, "learning_rate": 0.0005, "loss": 0.8570699095726013, "step": 48388 }, { "epoch": 3.8146629877808436, "grad_norm": 0.5212305784225464, "learning_rate": 0.0005, "loss": 1.1097077131271362, "step": 48389 }, { "epoch": 3.8147418210484823, "grad_norm": 0.43034103512763977, "learning_rate": 0.0005, "loss": 1.0392117500305176, "step": 48390 }, { "epoch": 3.8148206543161214, "grad_norm": 0.52857506275177, "learning_rate": 0.0005, "loss": 1.055481195449829, "step": 48391 }, { "epoch": 3.8148994875837605, "grad_norm": 0.3365309536457062, "learning_rate": 0.0005, "loss": 0.9886059165000916, "step": 48392 }, { "epoch": 3.814978320851399, "grad_norm": 0.793039083480835, "learning_rate": 0.0005, "loss": 1.18949556350708, "step": 48393 }, { "epoch": 3.8150571541190383, "grad_norm": 0.40212345123291016, "learning_rate": 0.0005, "loss": 0.9228388071060181, "step": 48394 }, { "epoch": 3.815135987386677, "grad_norm": 0.5509620308876038, "learning_rate": 0.0005, "loss": 1.1129355430603027, "step": 48395 }, { "epoch": 3.815214820654316, "grad_norm": 0.4824274778366089, "learning_rate": 0.0005, "loss": 1.0666731595993042, "step": 48396 }, { "epoch": 3.8152936539219553, "grad_norm": 0.49520111083984375, "learning_rate": 0.0005, "loss": 1.1771323680877686, "step": 48397 }, { "epoch": 3.815372487189594, "grad_norm": 0.5060831904411316, "learning_rate": 0.0005, "loss": 1.2028847932815552, "step": 48398 }, { "epoch": 3.8154513204572327, "grad_norm": 0.6543577313423157, "learning_rate": 0.0005, "loss": 1.2854337692260742, "step": 48399 }, { "epoch": 3.815530153724872, "grad_norm": 0.44222986698150635, "learning_rate": 0.0005, "loss": 1.1951888799667358, "step": 48400 }, { "epoch": 3.815608986992511, "grad_norm": 0.45957276225090027, "learning_rate": 0.0005, "loss": 1.116538166999817, "step": 48401 }, { "epoch": 3.8156878202601496, "grad_norm": 0.5937703847885132, "learning_rate": 0.0005, "loss": 1.2615077495574951, "step": 48402 }, { "epoch": 3.8157666535277888, "grad_norm": 0.4267517328262329, "learning_rate": 0.0005, "loss": 1.0889067649841309, "step": 48403 }, { "epoch": 3.8158454867954275, "grad_norm": 0.4433586895465851, "learning_rate": 0.0005, "loss": 0.991268515586853, "step": 48404 }, { "epoch": 3.8159243200630666, "grad_norm": 0.4667574465274811, "learning_rate": 0.0005, "loss": 1.0449845790863037, "step": 48405 }, { "epoch": 3.8160031533307057, "grad_norm": 0.6742110252380371, "learning_rate": 0.0005, "loss": 1.1451528072357178, "step": 48406 }, { "epoch": 3.8160819865983444, "grad_norm": 0.5294402241706848, "learning_rate": 0.0005, "loss": 1.206202745437622, "step": 48407 }, { "epoch": 3.8161608198659835, "grad_norm": 0.5483182072639465, "learning_rate": 0.0005, "loss": 1.209549903869629, "step": 48408 }, { "epoch": 3.8162396531336222, "grad_norm": 0.7199029326438904, "learning_rate": 0.0005, "loss": 1.2242672443389893, "step": 48409 }, { "epoch": 3.8163184864012614, "grad_norm": 0.5415412783622742, "learning_rate": 0.0005, "loss": 1.1884121894836426, "step": 48410 }, { "epoch": 3.8163973196689005, "grad_norm": 0.5088854432106018, "learning_rate": 0.0005, "loss": 1.319305658340454, "step": 48411 }, { "epoch": 3.816476152936539, "grad_norm": 0.4620816707611084, "learning_rate": 0.0005, "loss": 1.244924545288086, "step": 48412 }, { "epoch": 3.8165549862041783, "grad_norm": 0.49456045031547546, "learning_rate": 0.0005, "loss": 1.2759099006652832, "step": 48413 }, { "epoch": 3.816633819471817, "grad_norm": 0.5240573287010193, "learning_rate": 0.0005, "loss": 1.3928357362747192, "step": 48414 }, { "epoch": 3.816712652739456, "grad_norm": 0.4928331673145294, "learning_rate": 0.0005, "loss": 1.1363394260406494, "step": 48415 }, { "epoch": 3.816791486007095, "grad_norm": 0.5401036739349365, "learning_rate": 0.0005, "loss": 1.411325216293335, "step": 48416 }, { "epoch": 3.816870319274734, "grad_norm": 0.4906030297279358, "learning_rate": 0.0005, "loss": 1.1856732368469238, "step": 48417 }, { "epoch": 3.8169491525423727, "grad_norm": 0.4543890357017517, "learning_rate": 0.0005, "loss": 1.1215815544128418, "step": 48418 }, { "epoch": 3.817027985810012, "grad_norm": 0.44562795758247375, "learning_rate": 0.0005, "loss": 1.2046622037887573, "step": 48419 }, { "epoch": 3.817106819077651, "grad_norm": 0.4880322813987732, "learning_rate": 0.0005, "loss": 1.2172818183898926, "step": 48420 }, { "epoch": 3.8171856523452896, "grad_norm": 0.48649707436561584, "learning_rate": 0.0005, "loss": 1.1478025913238525, "step": 48421 }, { "epoch": 3.8172644856129287, "grad_norm": 0.512232780456543, "learning_rate": 0.0005, "loss": 1.227333903312683, "step": 48422 }, { "epoch": 3.8173433188805674, "grad_norm": 0.5279853343963623, "learning_rate": 0.0005, "loss": 1.3160438537597656, "step": 48423 }, { "epoch": 3.8174221521482066, "grad_norm": 0.4876013994216919, "learning_rate": 0.0005, "loss": 1.060448169708252, "step": 48424 }, { "epoch": 3.8175009854158457, "grad_norm": 0.5864822268486023, "learning_rate": 0.0005, "loss": 1.2202765941619873, "step": 48425 }, { "epoch": 3.8175798186834844, "grad_norm": 0.5233162045478821, "learning_rate": 0.0005, "loss": 1.1107836961746216, "step": 48426 }, { "epoch": 3.8176586519511235, "grad_norm": 0.47105643153190613, "learning_rate": 0.0005, "loss": 1.1663494110107422, "step": 48427 }, { "epoch": 3.817737485218762, "grad_norm": 0.45312702655792236, "learning_rate": 0.0005, "loss": 1.3559879064559937, "step": 48428 }, { "epoch": 3.8178163184864014, "grad_norm": 0.5127418041229248, "learning_rate": 0.0005, "loss": 1.2613399028778076, "step": 48429 }, { "epoch": 3.81789515175404, "grad_norm": 0.4797591269016266, "learning_rate": 0.0005, "loss": 1.0810940265655518, "step": 48430 }, { "epoch": 3.817973985021679, "grad_norm": 0.5641543865203857, "learning_rate": 0.0005, "loss": 1.2007226943969727, "step": 48431 }, { "epoch": 3.818052818289318, "grad_norm": 0.5028535723686218, "learning_rate": 0.0005, "loss": 1.0559911727905273, "step": 48432 }, { "epoch": 3.818131651556957, "grad_norm": 0.4589076340198517, "learning_rate": 0.0005, "loss": 1.1690459251403809, "step": 48433 }, { "epoch": 3.818210484824596, "grad_norm": 0.3972175717353821, "learning_rate": 0.0005, "loss": 1.0722160339355469, "step": 48434 }, { "epoch": 3.818289318092235, "grad_norm": 0.510235607624054, "learning_rate": 0.0005, "loss": 1.1801658868789673, "step": 48435 }, { "epoch": 3.818368151359874, "grad_norm": 0.42228928208351135, "learning_rate": 0.0005, "loss": 0.9917668104171753, "step": 48436 }, { "epoch": 3.8184469846275126, "grad_norm": 0.4808368682861328, "learning_rate": 0.0005, "loss": 1.1828066110610962, "step": 48437 }, { "epoch": 3.8185258178951518, "grad_norm": 0.5334661602973938, "learning_rate": 0.0005, "loss": 1.1640292406082153, "step": 48438 }, { "epoch": 3.818604651162791, "grad_norm": 0.5458549857139587, "learning_rate": 0.0005, "loss": 1.305019497871399, "step": 48439 }, { "epoch": 3.8186834844304296, "grad_norm": 1.3973565101623535, "learning_rate": 0.0005, "loss": 1.3097245693206787, "step": 48440 }, { "epoch": 3.8187623176980687, "grad_norm": 0.46009519696235657, "learning_rate": 0.0005, "loss": 1.1452515125274658, "step": 48441 }, { "epoch": 3.8188411509657074, "grad_norm": 0.42944690585136414, "learning_rate": 0.0005, "loss": 1.0839447975158691, "step": 48442 }, { "epoch": 3.8189199842333466, "grad_norm": 0.5301482677459717, "learning_rate": 0.0005, "loss": 1.276839256286621, "step": 48443 }, { "epoch": 3.8189988175009857, "grad_norm": 0.4825010299682617, "learning_rate": 0.0005, "loss": 1.294464111328125, "step": 48444 }, { "epoch": 3.8190776507686244, "grad_norm": 0.43683144450187683, "learning_rate": 0.0005, "loss": 1.1936225891113281, "step": 48445 }, { "epoch": 3.819156484036263, "grad_norm": 0.5200119018554688, "learning_rate": 0.0005, "loss": 0.9725151062011719, "step": 48446 }, { "epoch": 3.819235317303902, "grad_norm": 0.5782474279403687, "learning_rate": 0.0005, "loss": 1.2597193717956543, "step": 48447 }, { "epoch": 3.8193141505715413, "grad_norm": 0.5387783050537109, "learning_rate": 0.0005, "loss": 1.1842269897460938, "step": 48448 }, { "epoch": 3.81939298383918, "grad_norm": 0.5437667369842529, "learning_rate": 0.0005, "loss": 1.2352299690246582, "step": 48449 }, { "epoch": 3.819471817106819, "grad_norm": 0.4601678252220154, "learning_rate": 0.0005, "loss": 1.1108087301254272, "step": 48450 }, { "epoch": 3.819550650374458, "grad_norm": 0.500610888004303, "learning_rate": 0.0005, "loss": 1.1434142589569092, "step": 48451 }, { "epoch": 3.819629483642097, "grad_norm": 0.4365220367908478, "learning_rate": 0.0005, "loss": 1.0410408973693848, "step": 48452 }, { "epoch": 3.819708316909736, "grad_norm": 0.47006431221961975, "learning_rate": 0.0005, "loss": 1.2736296653747559, "step": 48453 }, { "epoch": 3.819787150177375, "grad_norm": 0.4483494460582733, "learning_rate": 0.0005, "loss": 1.1492118835449219, "step": 48454 }, { "epoch": 3.819865983445014, "grad_norm": 0.5285339951515198, "learning_rate": 0.0005, "loss": 1.1856186389923096, "step": 48455 }, { "epoch": 3.8199448167126526, "grad_norm": 0.44742313027381897, "learning_rate": 0.0005, "loss": 1.189335823059082, "step": 48456 }, { "epoch": 3.8200236499802918, "grad_norm": 0.43297111988067627, "learning_rate": 0.0005, "loss": 1.1821191310882568, "step": 48457 }, { "epoch": 3.820102483247931, "grad_norm": 0.5730781555175781, "learning_rate": 0.0005, "loss": 1.2755281925201416, "step": 48458 }, { "epoch": 3.8201813165155696, "grad_norm": 0.4544593393802643, "learning_rate": 0.0005, "loss": 1.2385163307189941, "step": 48459 }, { "epoch": 3.8202601497832083, "grad_norm": 0.4557507634162903, "learning_rate": 0.0005, "loss": 1.0790510177612305, "step": 48460 }, { "epoch": 3.8203389830508474, "grad_norm": 0.4649627208709717, "learning_rate": 0.0005, "loss": 1.0411033630371094, "step": 48461 }, { "epoch": 3.8204178163184865, "grad_norm": 0.6385750770568848, "learning_rate": 0.0005, "loss": 1.2547861337661743, "step": 48462 }, { "epoch": 3.8204966495861252, "grad_norm": 0.548703134059906, "learning_rate": 0.0005, "loss": 1.2142856121063232, "step": 48463 }, { "epoch": 3.8205754828537644, "grad_norm": 0.4476529359817505, "learning_rate": 0.0005, "loss": 1.0420433282852173, "step": 48464 }, { "epoch": 3.820654316121403, "grad_norm": 0.4407881796360016, "learning_rate": 0.0005, "loss": 1.1769908666610718, "step": 48465 }, { "epoch": 3.820733149389042, "grad_norm": 0.507752537727356, "learning_rate": 0.0005, "loss": 1.1916298866271973, "step": 48466 }, { "epoch": 3.8208119826566813, "grad_norm": 0.4203520119190216, "learning_rate": 0.0005, "loss": 1.2339341640472412, "step": 48467 }, { "epoch": 3.82089081592432, "grad_norm": 0.5109068155288696, "learning_rate": 0.0005, "loss": 1.2085316181182861, "step": 48468 }, { "epoch": 3.820969649191959, "grad_norm": 0.7074394226074219, "learning_rate": 0.0005, "loss": 1.3981462717056274, "step": 48469 }, { "epoch": 3.821048482459598, "grad_norm": 0.48300251364707947, "learning_rate": 0.0005, "loss": 1.202918291091919, "step": 48470 }, { "epoch": 3.821127315727237, "grad_norm": 0.4461917579174042, "learning_rate": 0.0005, "loss": 1.062117576599121, "step": 48471 }, { "epoch": 3.821206148994876, "grad_norm": 0.42989182472229004, "learning_rate": 0.0005, "loss": 1.2308028936386108, "step": 48472 }, { "epoch": 3.821284982262515, "grad_norm": 0.5110786557197571, "learning_rate": 0.0005, "loss": 1.256415605545044, "step": 48473 }, { "epoch": 3.8213638155301535, "grad_norm": 0.4883534610271454, "learning_rate": 0.0005, "loss": 1.1117970943450928, "step": 48474 }, { "epoch": 3.8214426487977926, "grad_norm": 0.7954611778259277, "learning_rate": 0.0005, "loss": 1.3434834480285645, "step": 48475 }, { "epoch": 3.8215214820654317, "grad_norm": 0.5576748251914978, "learning_rate": 0.0005, "loss": 1.0831925868988037, "step": 48476 }, { "epoch": 3.8216003153330704, "grad_norm": 0.4781573712825775, "learning_rate": 0.0005, "loss": 1.333249807357788, "step": 48477 }, { "epoch": 3.8216791486007096, "grad_norm": 0.6102185249328613, "learning_rate": 0.0005, "loss": 1.2817153930664062, "step": 48478 }, { "epoch": 3.8217579818683483, "grad_norm": 0.552986741065979, "learning_rate": 0.0005, "loss": 1.219759464263916, "step": 48479 }, { "epoch": 3.8218368151359874, "grad_norm": 0.45766451954841614, "learning_rate": 0.0005, "loss": 1.148921012878418, "step": 48480 }, { "epoch": 3.8219156484036265, "grad_norm": 0.4908432066440582, "learning_rate": 0.0005, "loss": 1.1268526315689087, "step": 48481 }, { "epoch": 3.821994481671265, "grad_norm": 0.5622887015342712, "learning_rate": 0.0005, "loss": 1.073012113571167, "step": 48482 }, { "epoch": 3.8220733149389043, "grad_norm": 0.4782186448574066, "learning_rate": 0.0005, "loss": 1.2212929725646973, "step": 48483 }, { "epoch": 3.822152148206543, "grad_norm": 0.4763399660587311, "learning_rate": 0.0005, "loss": 1.0787326097488403, "step": 48484 }, { "epoch": 3.822230981474182, "grad_norm": 0.3626549243927002, "learning_rate": 0.0005, "loss": 0.9232149720191956, "step": 48485 }, { "epoch": 3.8223098147418213, "grad_norm": 0.4743334949016571, "learning_rate": 0.0005, "loss": 1.2275958061218262, "step": 48486 }, { "epoch": 3.82238864800946, "grad_norm": 0.3875938057899475, "learning_rate": 0.0005, "loss": 1.1863089799880981, "step": 48487 }, { "epoch": 3.8224674812770987, "grad_norm": 0.506980836391449, "learning_rate": 0.0005, "loss": 1.2813720703125, "step": 48488 }, { "epoch": 3.822546314544738, "grad_norm": 0.5070393681526184, "learning_rate": 0.0005, "loss": 1.3345544338226318, "step": 48489 }, { "epoch": 3.822625147812377, "grad_norm": 0.4991891384124756, "learning_rate": 0.0005, "loss": 1.1592140197753906, "step": 48490 }, { "epoch": 3.8227039810800156, "grad_norm": 0.4848884344100952, "learning_rate": 0.0005, "loss": 1.1140142679214478, "step": 48491 }, { "epoch": 3.8227828143476548, "grad_norm": 0.6405090689659119, "learning_rate": 0.0005, "loss": 1.269834041595459, "step": 48492 }, { "epoch": 3.8228616476152935, "grad_norm": 0.5260899066925049, "learning_rate": 0.0005, "loss": 1.3238247632980347, "step": 48493 }, { "epoch": 3.8229404808829326, "grad_norm": 0.5169250965118408, "learning_rate": 0.0005, "loss": 1.1199326515197754, "step": 48494 }, { "epoch": 3.8230193141505717, "grad_norm": 0.5551010370254517, "learning_rate": 0.0005, "loss": 1.1423814296722412, "step": 48495 }, { "epoch": 3.8230981474182104, "grad_norm": 0.8904977440834045, "learning_rate": 0.0005, "loss": 1.2035402059555054, "step": 48496 }, { "epoch": 3.8231769806858495, "grad_norm": 0.6196509599685669, "learning_rate": 0.0005, "loss": 1.0314005613327026, "step": 48497 }, { "epoch": 3.8232558139534882, "grad_norm": 0.4358806014060974, "learning_rate": 0.0005, "loss": 1.119626522064209, "step": 48498 }, { "epoch": 3.8233346472211274, "grad_norm": 0.4188164174556732, "learning_rate": 0.0005, "loss": 1.0844547748565674, "step": 48499 }, { "epoch": 3.8234134804887665, "grad_norm": 0.5127619504928589, "learning_rate": 0.0005, "loss": 1.165038824081421, "step": 48500 }, { "epoch": 3.823492313756405, "grad_norm": 0.4817585349082947, "learning_rate": 0.0005, "loss": 1.1517189741134644, "step": 48501 }, { "epoch": 3.823571147024044, "grad_norm": 0.46562492847442627, "learning_rate": 0.0005, "loss": 1.1193755865097046, "step": 48502 }, { "epoch": 3.823649980291683, "grad_norm": 0.47103801369667053, "learning_rate": 0.0005, "loss": 0.9744885563850403, "step": 48503 }, { "epoch": 3.823728813559322, "grad_norm": 0.48113715648651123, "learning_rate": 0.0005, "loss": 1.1716389656066895, "step": 48504 }, { "epoch": 3.823807646826961, "grad_norm": 0.6417760252952576, "learning_rate": 0.0005, "loss": 1.256329894065857, "step": 48505 }, { "epoch": 3.8238864800946, "grad_norm": 0.4938456118106842, "learning_rate": 0.0005, "loss": 1.1977856159210205, "step": 48506 }, { "epoch": 3.8239653133622387, "grad_norm": 0.5248764753341675, "learning_rate": 0.0005, "loss": 0.926306962966919, "step": 48507 }, { "epoch": 3.824044146629878, "grad_norm": 0.5175483226776123, "learning_rate": 0.0005, "loss": 1.3068346977233887, "step": 48508 }, { "epoch": 3.824122979897517, "grad_norm": 0.5682626962661743, "learning_rate": 0.0005, "loss": 1.1679673194885254, "step": 48509 }, { "epoch": 3.8242018131651556, "grad_norm": 0.5741828083992004, "learning_rate": 0.0005, "loss": 1.3775217533111572, "step": 48510 }, { "epoch": 3.8242806464327948, "grad_norm": 0.5330941081047058, "learning_rate": 0.0005, "loss": 1.2529124021530151, "step": 48511 }, { "epoch": 3.8243594797004334, "grad_norm": 0.503518283367157, "learning_rate": 0.0005, "loss": 1.1241823434829712, "step": 48512 }, { "epoch": 3.8244383129680726, "grad_norm": 0.5092682242393494, "learning_rate": 0.0005, "loss": 1.2230899333953857, "step": 48513 }, { "epoch": 3.8245171462357117, "grad_norm": 0.560883641242981, "learning_rate": 0.0005, "loss": 1.169294834136963, "step": 48514 }, { "epoch": 3.8245959795033504, "grad_norm": 0.49082186818122864, "learning_rate": 0.0005, "loss": 1.151329755783081, "step": 48515 }, { "epoch": 3.824674812770989, "grad_norm": 0.5276619791984558, "learning_rate": 0.0005, "loss": 1.144549012184143, "step": 48516 }, { "epoch": 3.824753646038628, "grad_norm": 0.5067009329795837, "learning_rate": 0.0005, "loss": 1.1197301149368286, "step": 48517 }, { "epoch": 3.8248324793062674, "grad_norm": 0.44043922424316406, "learning_rate": 0.0005, "loss": 1.0234441757202148, "step": 48518 }, { "epoch": 3.824911312573906, "grad_norm": 0.4365745782852173, "learning_rate": 0.0005, "loss": 0.9980889558792114, "step": 48519 }, { "epoch": 3.824990145841545, "grad_norm": 0.5516452789306641, "learning_rate": 0.0005, "loss": 1.2563872337341309, "step": 48520 }, { "epoch": 3.825068979109184, "grad_norm": 0.45180991291999817, "learning_rate": 0.0005, "loss": 1.088883876800537, "step": 48521 }, { "epoch": 3.825147812376823, "grad_norm": 0.5639746189117432, "learning_rate": 0.0005, "loss": 1.2782658338546753, "step": 48522 }, { "epoch": 3.825226645644462, "grad_norm": 0.4865327477455139, "learning_rate": 0.0005, "loss": 1.1035070419311523, "step": 48523 }, { "epoch": 3.825305478912101, "grad_norm": 0.5208885669708252, "learning_rate": 0.0005, "loss": 1.254030704498291, "step": 48524 }, { "epoch": 3.82538431217974, "grad_norm": 0.4665059447288513, "learning_rate": 0.0005, "loss": 1.0715959072113037, "step": 48525 }, { "epoch": 3.8254631454473786, "grad_norm": 0.5119485855102539, "learning_rate": 0.0005, "loss": 1.210417628288269, "step": 48526 }, { "epoch": 3.8255419787150178, "grad_norm": 0.5055830478668213, "learning_rate": 0.0005, "loss": 1.1304211616516113, "step": 48527 }, { "epoch": 3.825620811982657, "grad_norm": 0.4467078447341919, "learning_rate": 0.0005, "loss": 1.2005424499511719, "step": 48528 }, { "epoch": 3.8256996452502956, "grad_norm": 0.4473892152309418, "learning_rate": 0.0005, "loss": 1.0640830993652344, "step": 48529 }, { "epoch": 3.8257784785179347, "grad_norm": 0.5436131954193115, "learning_rate": 0.0005, "loss": 1.2450512647628784, "step": 48530 }, { "epoch": 3.8258573117855734, "grad_norm": 0.5116367936134338, "learning_rate": 0.0005, "loss": 1.3500232696533203, "step": 48531 }, { "epoch": 3.8259361450532126, "grad_norm": 0.49627068638801575, "learning_rate": 0.0005, "loss": 1.2254009246826172, "step": 48532 }, { "epoch": 3.8260149783208512, "grad_norm": 0.4914596676826477, "learning_rate": 0.0005, "loss": 1.178100824356079, "step": 48533 }, { "epoch": 3.8260938115884904, "grad_norm": 0.4074377119541168, "learning_rate": 0.0005, "loss": 0.9914257526397705, "step": 48534 }, { "epoch": 3.826172644856129, "grad_norm": 0.5238226652145386, "learning_rate": 0.0005, "loss": 1.037185549736023, "step": 48535 }, { "epoch": 3.826251478123768, "grad_norm": 0.6827869415283203, "learning_rate": 0.0005, "loss": 1.001424789428711, "step": 48536 }, { "epoch": 3.8263303113914073, "grad_norm": 0.6008344292640686, "learning_rate": 0.0005, "loss": 1.015804648399353, "step": 48537 }, { "epoch": 3.826409144659046, "grad_norm": 0.515608549118042, "learning_rate": 0.0005, "loss": 1.432037353515625, "step": 48538 }, { "epoch": 3.826487977926685, "grad_norm": 0.4620170295238495, "learning_rate": 0.0005, "loss": 1.1724786758422852, "step": 48539 }, { "epoch": 3.826566811194324, "grad_norm": 0.45299941301345825, "learning_rate": 0.0005, "loss": 1.0949742794036865, "step": 48540 }, { "epoch": 3.826645644461963, "grad_norm": 0.5659868717193604, "learning_rate": 0.0005, "loss": 1.2130892276763916, "step": 48541 }, { "epoch": 3.826724477729602, "grad_norm": 0.47137153148651123, "learning_rate": 0.0005, "loss": 1.1643460988998413, "step": 48542 }, { "epoch": 3.826803310997241, "grad_norm": 0.49146726727485657, "learning_rate": 0.0005, "loss": 1.172602653503418, "step": 48543 }, { "epoch": 3.82688214426488, "grad_norm": 0.4921407401561737, "learning_rate": 0.0005, "loss": 1.1859365701675415, "step": 48544 }, { "epoch": 3.8269609775325186, "grad_norm": 0.5498971343040466, "learning_rate": 0.0005, "loss": 1.3952224254608154, "step": 48545 }, { "epoch": 3.8270398108001578, "grad_norm": 0.4505774676799774, "learning_rate": 0.0005, "loss": 1.047384262084961, "step": 48546 }, { "epoch": 3.8271186440677964, "grad_norm": 0.5413532853126526, "learning_rate": 0.0005, "loss": 1.156947135925293, "step": 48547 }, { "epoch": 3.8271974773354356, "grad_norm": 0.45104143023490906, "learning_rate": 0.0005, "loss": 1.2505743503570557, "step": 48548 }, { "epoch": 3.8272763106030743, "grad_norm": 0.4773530066013336, "learning_rate": 0.0005, "loss": 1.1763941049575806, "step": 48549 }, { "epoch": 3.8273551438707134, "grad_norm": 0.5239595174789429, "learning_rate": 0.0005, "loss": 1.0763448476791382, "step": 48550 }, { "epoch": 3.8274339771383525, "grad_norm": 0.4805929660797119, "learning_rate": 0.0005, "loss": 1.1306387186050415, "step": 48551 }, { "epoch": 3.8275128104059912, "grad_norm": 0.46505939960479736, "learning_rate": 0.0005, "loss": 1.2019388675689697, "step": 48552 }, { "epoch": 3.8275916436736304, "grad_norm": 0.5875998139381409, "learning_rate": 0.0005, "loss": 1.3333433866500854, "step": 48553 }, { "epoch": 3.827670476941269, "grad_norm": 0.5379155874252319, "learning_rate": 0.0005, "loss": 1.104836106300354, "step": 48554 }, { "epoch": 3.827749310208908, "grad_norm": 0.40897366404533386, "learning_rate": 0.0005, "loss": 1.029996633529663, "step": 48555 }, { "epoch": 3.8278281434765473, "grad_norm": 0.4925456941127777, "learning_rate": 0.0005, "loss": 1.341009259223938, "step": 48556 }, { "epoch": 3.827906976744186, "grad_norm": 0.44468510150909424, "learning_rate": 0.0005, "loss": 1.0281246900558472, "step": 48557 }, { "epoch": 3.827985810011825, "grad_norm": 0.8066408634185791, "learning_rate": 0.0005, "loss": 1.194654941558838, "step": 48558 }, { "epoch": 3.828064643279464, "grad_norm": 0.4566972255706787, "learning_rate": 0.0005, "loss": 1.0209890604019165, "step": 48559 }, { "epoch": 3.828143476547103, "grad_norm": 0.42962440848350525, "learning_rate": 0.0005, "loss": 1.1322441101074219, "step": 48560 }, { "epoch": 3.828222309814742, "grad_norm": 0.4287842810153961, "learning_rate": 0.0005, "loss": 1.12813138961792, "step": 48561 }, { "epoch": 3.828301143082381, "grad_norm": 0.46202656626701355, "learning_rate": 0.0005, "loss": 1.1567063331604004, "step": 48562 }, { "epoch": 3.8283799763500195, "grad_norm": 0.6844921112060547, "learning_rate": 0.0005, "loss": 1.2021963596343994, "step": 48563 }, { "epoch": 3.8284588096176586, "grad_norm": 0.43231630325317383, "learning_rate": 0.0005, "loss": 1.1078190803527832, "step": 48564 }, { "epoch": 3.8285376428852977, "grad_norm": 0.4910513460636139, "learning_rate": 0.0005, "loss": 1.096450924873352, "step": 48565 }, { "epoch": 3.8286164761529364, "grad_norm": 0.5089382529258728, "learning_rate": 0.0005, "loss": 1.2939982414245605, "step": 48566 }, { "epoch": 3.8286953094205756, "grad_norm": 0.49718713760375977, "learning_rate": 0.0005, "loss": 1.1422498226165771, "step": 48567 }, { "epoch": 3.8287741426882143, "grad_norm": 0.4347870945930481, "learning_rate": 0.0005, "loss": 1.1456828117370605, "step": 48568 }, { "epoch": 3.8288529759558534, "grad_norm": 0.47854623198509216, "learning_rate": 0.0005, "loss": 1.1087448596954346, "step": 48569 }, { "epoch": 3.8289318092234925, "grad_norm": 0.4952503442764282, "learning_rate": 0.0005, "loss": 1.146543025970459, "step": 48570 }, { "epoch": 3.829010642491131, "grad_norm": 0.413372665643692, "learning_rate": 0.0005, "loss": 0.9896951913833618, "step": 48571 }, { "epoch": 3.8290894757587703, "grad_norm": 0.48827797174453735, "learning_rate": 0.0005, "loss": 1.3221404552459717, "step": 48572 }, { "epoch": 3.829168309026409, "grad_norm": 0.4275316894054413, "learning_rate": 0.0005, "loss": 1.2318880558013916, "step": 48573 }, { "epoch": 3.829247142294048, "grad_norm": 0.40196576714515686, "learning_rate": 0.0005, "loss": 1.1790351867675781, "step": 48574 }, { "epoch": 3.8293259755616873, "grad_norm": 0.4727404713630676, "learning_rate": 0.0005, "loss": 1.0521175861358643, "step": 48575 }, { "epoch": 3.829404808829326, "grad_norm": 0.5854132175445557, "learning_rate": 0.0005, "loss": 1.1725504398345947, "step": 48576 }, { "epoch": 3.8294836420969647, "grad_norm": 0.4423256516456604, "learning_rate": 0.0005, "loss": 1.0145272016525269, "step": 48577 }, { "epoch": 3.829562475364604, "grad_norm": 0.44462668895721436, "learning_rate": 0.0005, "loss": 1.1576604843139648, "step": 48578 }, { "epoch": 3.829641308632243, "grad_norm": 0.4401976764202118, "learning_rate": 0.0005, "loss": 1.094616174697876, "step": 48579 }, { "epoch": 3.8297201418998816, "grad_norm": 0.5172896385192871, "learning_rate": 0.0005, "loss": 1.179957628250122, "step": 48580 }, { "epoch": 3.8297989751675208, "grad_norm": 0.5441446304321289, "learning_rate": 0.0005, "loss": 1.4695031642913818, "step": 48581 }, { "epoch": 3.8298778084351595, "grad_norm": 0.5650084614753723, "learning_rate": 0.0005, "loss": 1.244240641593933, "step": 48582 }, { "epoch": 3.8299566417027986, "grad_norm": 0.5890842080116272, "learning_rate": 0.0005, "loss": 1.1963828802108765, "step": 48583 }, { "epoch": 3.8300354749704377, "grad_norm": 0.5323278307914734, "learning_rate": 0.0005, "loss": 1.09944748878479, "step": 48584 }, { "epoch": 3.8301143082380764, "grad_norm": 0.47711512446403503, "learning_rate": 0.0005, "loss": 1.18503737449646, "step": 48585 }, { "epoch": 3.8301931415057155, "grad_norm": 0.5455014705657959, "learning_rate": 0.0005, "loss": 1.3226308822631836, "step": 48586 }, { "epoch": 3.8302719747733542, "grad_norm": 0.47641721367836, "learning_rate": 0.0005, "loss": 1.2154018878936768, "step": 48587 }, { "epoch": 3.8303508080409934, "grad_norm": 0.6518959999084473, "learning_rate": 0.0005, "loss": 1.1721630096435547, "step": 48588 }, { "epoch": 3.8304296413086325, "grad_norm": 0.4938261806964874, "learning_rate": 0.0005, "loss": 1.2359858751296997, "step": 48589 }, { "epoch": 3.830508474576271, "grad_norm": 0.506418764591217, "learning_rate": 0.0005, "loss": 1.1789036989212036, "step": 48590 }, { "epoch": 3.83058730784391, "grad_norm": 0.39849403500556946, "learning_rate": 0.0005, "loss": 1.0994023084640503, "step": 48591 }, { "epoch": 3.830666141111549, "grad_norm": 1.4679664373397827, "learning_rate": 0.0005, "loss": 1.3314549922943115, "step": 48592 }, { "epoch": 3.830744974379188, "grad_norm": 0.4661485552787781, "learning_rate": 0.0005, "loss": 1.2249956130981445, "step": 48593 }, { "epoch": 3.830823807646827, "grad_norm": 0.5419355630874634, "learning_rate": 0.0005, "loss": 1.0871589183807373, "step": 48594 }, { "epoch": 3.830902640914466, "grad_norm": 0.5109383463859558, "learning_rate": 0.0005, "loss": 1.24093496799469, "step": 48595 }, { "epoch": 3.8309814741821047, "grad_norm": 0.4235706627368927, "learning_rate": 0.0005, "loss": 1.0550636053085327, "step": 48596 }, { "epoch": 3.831060307449744, "grad_norm": 0.5290881395339966, "learning_rate": 0.0005, "loss": 1.0604257583618164, "step": 48597 }, { "epoch": 3.831139140717383, "grad_norm": 0.5412684082984924, "learning_rate": 0.0005, "loss": 1.3376306295394897, "step": 48598 }, { "epoch": 3.8312179739850216, "grad_norm": 0.44914382696151733, "learning_rate": 0.0005, "loss": 1.0286732912063599, "step": 48599 }, { "epoch": 3.8312968072526608, "grad_norm": 0.43975958228111267, "learning_rate": 0.0005, "loss": 1.2323133945465088, "step": 48600 }, { "epoch": 3.8313756405202994, "grad_norm": 0.4870457351207733, "learning_rate": 0.0005, "loss": 1.2599217891693115, "step": 48601 }, { "epoch": 3.8314544737879386, "grad_norm": 0.5028870105743408, "learning_rate": 0.0005, "loss": 1.1232988834381104, "step": 48602 }, { "epoch": 3.8315333070555777, "grad_norm": 0.5603160858154297, "learning_rate": 0.0005, "loss": 1.2005126476287842, "step": 48603 }, { "epoch": 3.8316121403232164, "grad_norm": 0.4958431124687195, "learning_rate": 0.0005, "loss": 1.1059768199920654, "step": 48604 }, { "epoch": 3.831690973590855, "grad_norm": 0.5078713297843933, "learning_rate": 0.0005, "loss": 1.065908670425415, "step": 48605 }, { "epoch": 3.831769806858494, "grad_norm": 0.5214213132858276, "learning_rate": 0.0005, "loss": 1.2605019807815552, "step": 48606 }, { "epoch": 3.8318486401261334, "grad_norm": 0.39960744976997375, "learning_rate": 0.0005, "loss": 1.1610000133514404, "step": 48607 }, { "epoch": 3.831927473393772, "grad_norm": 0.5452695488929749, "learning_rate": 0.0005, "loss": 1.2444517612457275, "step": 48608 }, { "epoch": 3.832006306661411, "grad_norm": 0.6118186712265015, "learning_rate": 0.0005, "loss": 0.8892862796783447, "step": 48609 }, { "epoch": 3.83208513992905, "grad_norm": 0.5460726022720337, "learning_rate": 0.0005, "loss": 0.9907097816467285, "step": 48610 }, { "epoch": 3.832163973196689, "grad_norm": 0.6514857411384583, "learning_rate": 0.0005, "loss": 1.070796251296997, "step": 48611 }, { "epoch": 3.832242806464328, "grad_norm": 0.512192964553833, "learning_rate": 0.0005, "loss": 1.335695505142212, "step": 48612 }, { "epoch": 3.832321639731967, "grad_norm": 0.5143149495124817, "learning_rate": 0.0005, "loss": 1.1681594848632812, "step": 48613 }, { "epoch": 3.832400472999606, "grad_norm": 0.35001230239868164, "learning_rate": 0.0005, "loss": 0.9311084151268005, "step": 48614 }, { "epoch": 3.8324793062672446, "grad_norm": 0.4120543599128723, "learning_rate": 0.0005, "loss": 1.107251763343811, "step": 48615 }, { "epoch": 3.832558139534884, "grad_norm": 0.48036593198776245, "learning_rate": 0.0005, "loss": 1.1960639953613281, "step": 48616 }, { "epoch": 3.832636972802523, "grad_norm": 0.4985853135585785, "learning_rate": 0.0005, "loss": 1.1500568389892578, "step": 48617 }, { "epoch": 3.8327158060701616, "grad_norm": 0.552362859249115, "learning_rate": 0.0005, "loss": 1.12396240234375, "step": 48618 }, { "epoch": 3.8327946393378003, "grad_norm": 0.5143753290176392, "learning_rate": 0.0005, "loss": 0.9862208366394043, "step": 48619 }, { "epoch": 3.8328734726054394, "grad_norm": 0.5131964683532715, "learning_rate": 0.0005, "loss": 1.1262773275375366, "step": 48620 }, { "epoch": 3.8329523058730786, "grad_norm": 0.47704991698265076, "learning_rate": 0.0005, "loss": 1.2676050662994385, "step": 48621 }, { "epoch": 3.8330311391407172, "grad_norm": 0.521378219127655, "learning_rate": 0.0005, "loss": 1.272376537322998, "step": 48622 }, { "epoch": 3.8331099724083564, "grad_norm": 0.5277775526046753, "learning_rate": 0.0005, "loss": 1.3904412984848022, "step": 48623 }, { "epoch": 3.833188805675995, "grad_norm": 0.5031408071517944, "learning_rate": 0.0005, "loss": 1.0796074867248535, "step": 48624 }, { "epoch": 3.833267638943634, "grad_norm": 0.43894851207733154, "learning_rate": 0.0005, "loss": 1.1315361261367798, "step": 48625 }, { "epoch": 3.8333464722112733, "grad_norm": 0.5312449336051941, "learning_rate": 0.0005, "loss": 1.2559889554977417, "step": 48626 }, { "epoch": 3.833425305478912, "grad_norm": 0.42037078738212585, "learning_rate": 0.0005, "loss": 1.0772433280944824, "step": 48627 }, { "epoch": 3.833504138746551, "grad_norm": 0.5288019180297852, "learning_rate": 0.0005, "loss": 1.248652458190918, "step": 48628 }, { "epoch": 3.83358297201419, "grad_norm": 0.4728938043117523, "learning_rate": 0.0005, "loss": 1.211656093597412, "step": 48629 }, { "epoch": 3.833661805281829, "grad_norm": 0.5195821523666382, "learning_rate": 0.0005, "loss": 1.080663800239563, "step": 48630 }, { "epoch": 3.833740638549468, "grad_norm": 0.44677430391311646, "learning_rate": 0.0005, "loss": 1.1312050819396973, "step": 48631 }, { "epoch": 3.833819471817107, "grad_norm": 0.4623095989227295, "learning_rate": 0.0005, "loss": 1.2182810306549072, "step": 48632 }, { "epoch": 3.8338983050847455, "grad_norm": 0.45869192481040955, "learning_rate": 0.0005, "loss": 1.2450816631317139, "step": 48633 }, { "epoch": 3.8339771383523846, "grad_norm": 0.5290290117263794, "learning_rate": 0.0005, "loss": 1.3772929906845093, "step": 48634 }, { "epoch": 3.8340559716200238, "grad_norm": 0.7704567313194275, "learning_rate": 0.0005, "loss": 1.3088335990905762, "step": 48635 }, { "epoch": 3.8341348048876625, "grad_norm": 0.4427422285079956, "learning_rate": 0.0005, "loss": 1.1044762134552002, "step": 48636 }, { "epoch": 3.8342136381553016, "grad_norm": 0.5724158883094788, "learning_rate": 0.0005, "loss": 1.0690125226974487, "step": 48637 }, { "epoch": 3.8342924714229403, "grad_norm": 0.5970041751861572, "learning_rate": 0.0005, "loss": 1.0670225620269775, "step": 48638 }, { "epoch": 3.8343713046905794, "grad_norm": 0.548446536064148, "learning_rate": 0.0005, "loss": 1.2216932773590088, "step": 48639 }, { "epoch": 3.8344501379582185, "grad_norm": 0.8652608394622803, "learning_rate": 0.0005, "loss": 1.20481276512146, "step": 48640 }, { "epoch": 3.8345289712258572, "grad_norm": 0.6101423501968384, "learning_rate": 0.0005, "loss": 1.3323948383331299, "step": 48641 }, { "epoch": 3.8346078044934964, "grad_norm": 0.4928542375564575, "learning_rate": 0.0005, "loss": 1.292923927307129, "step": 48642 }, { "epoch": 3.834686637761135, "grad_norm": 0.6104946136474609, "learning_rate": 0.0005, "loss": 1.0178065299987793, "step": 48643 }, { "epoch": 3.834765471028774, "grad_norm": 0.5288615226745605, "learning_rate": 0.0005, "loss": 1.144397258758545, "step": 48644 }, { "epoch": 3.8348443042964133, "grad_norm": 0.5725725889205933, "learning_rate": 0.0005, "loss": 1.3489506244659424, "step": 48645 }, { "epoch": 3.834923137564052, "grad_norm": 0.4887237548828125, "learning_rate": 0.0005, "loss": 1.1123526096343994, "step": 48646 }, { "epoch": 3.8350019708316907, "grad_norm": 0.5865539312362671, "learning_rate": 0.0005, "loss": 1.1238858699798584, "step": 48647 }, { "epoch": 3.83508080409933, "grad_norm": 0.5606343746185303, "learning_rate": 0.0005, "loss": 1.188392996788025, "step": 48648 }, { "epoch": 3.835159637366969, "grad_norm": 0.5056546926498413, "learning_rate": 0.0005, "loss": 1.0630300045013428, "step": 48649 }, { "epoch": 3.8352384706346077, "grad_norm": 0.4838566780090332, "learning_rate": 0.0005, "loss": 1.200184941291809, "step": 48650 }, { "epoch": 3.835317303902247, "grad_norm": 0.4373907148838043, "learning_rate": 0.0005, "loss": 1.0037633180618286, "step": 48651 }, { "epoch": 3.8353961371698855, "grad_norm": 0.4095279276371002, "learning_rate": 0.0005, "loss": 1.003872275352478, "step": 48652 }, { "epoch": 3.8354749704375246, "grad_norm": 0.5383383631706238, "learning_rate": 0.0005, "loss": 1.324619174003601, "step": 48653 }, { "epoch": 3.8355538037051637, "grad_norm": 0.655447781085968, "learning_rate": 0.0005, "loss": 1.245038628578186, "step": 48654 }, { "epoch": 3.8356326369728024, "grad_norm": 0.484247088432312, "learning_rate": 0.0005, "loss": 1.3618049621582031, "step": 48655 }, { "epoch": 3.8357114702404416, "grad_norm": 0.4744182229042053, "learning_rate": 0.0005, "loss": 1.2766516208648682, "step": 48656 }, { "epoch": 3.8357903035080803, "grad_norm": 0.48424413800239563, "learning_rate": 0.0005, "loss": 1.0449512004852295, "step": 48657 }, { "epoch": 3.8358691367757194, "grad_norm": 0.4839197099208832, "learning_rate": 0.0005, "loss": 1.091262698173523, "step": 48658 }, { "epoch": 3.8359479700433585, "grad_norm": 0.5472300052642822, "learning_rate": 0.0005, "loss": 1.160869836807251, "step": 48659 }, { "epoch": 3.836026803310997, "grad_norm": 0.5410195589065552, "learning_rate": 0.0005, "loss": 1.095501184463501, "step": 48660 }, { "epoch": 3.8361056365786363, "grad_norm": 0.4378131031990051, "learning_rate": 0.0005, "loss": 1.2051054239273071, "step": 48661 }, { "epoch": 3.836184469846275, "grad_norm": 0.42952650785446167, "learning_rate": 0.0005, "loss": 1.094763994216919, "step": 48662 }, { "epoch": 3.836263303113914, "grad_norm": 0.4349987804889679, "learning_rate": 0.0005, "loss": 1.0804479122161865, "step": 48663 }, { "epoch": 3.836342136381553, "grad_norm": 0.6967437863349915, "learning_rate": 0.0005, "loss": 1.292391061782837, "step": 48664 }, { "epoch": 3.836420969649192, "grad_norm": 0.6118800044059753, "learning_rate": 0.0005, "loss": 1.066311001777649, "step": 48665 }, { "epoch": 3.8364998029168307, "grad_norm": 0.4290802776813507, "learning_rate": 0.0005, "loss": 1.1582835912704468, "step": 48666 }, { "epoch": 3.83657863618447, "grad_norm": 0.4707483649253845, "learning_rate": 0.0005, "loss": 1.1976691484451294, "step": 48667 }, { "epoch": 3.836657469452109, "grad_norm": 0.4771321713924408, "learning_rate": 0.0005, "loss": 1.1504888534545898, "step": 48668 }, { "epoch": 3.8367363027197476, "grad_norm": 0.5659695863723755, "learning_rate": 0.0005, "loss": 1.326479196548462, "step": 48669 }, { "epoch": 3.8368151359873868, "grad_norm": 0.46644675731658936, "learning_rate": 0.0005, "loss": 1.1125798225402832, "step": 48670 }, { "epoch": 3.8368939692550255, "grad_norm": 0.4321141541004181, "learning_rate": 0.0005, "loss": 1.1652777194976807, "step": 48671 }, { "epoch": 3.8369728025226646, "grad_norm": 0.491184800863266, "learning_rate": 0.0005, "loss": 1.1759014129638672, "step": 48672 }, { "epoch": 3.8370516357903037, "grad_norm": 0.5033468008041382, "learning_rate": 0.0005, "loss": 1.195768117904663, "step": 48673 }, { "epoch": 3.8371304690579424, "grad_norm": 0.5245769023895264, "learning_rate": 0.0005, "loss": 1.2560495138168335, "step": 48674 }, { "epoch": 3.8372093023255816, "grad_norm": 0.6534573435783386, "learning_rate": 0.0005, "loss": 1.2489590644836426, "step": 48675 }, { "epoch": 3.8372881355932202, "grad_norm": 0.5031471848487854, "learning_rate": 0.0005, "loss": 1.3014702796936035, "step": 48676 }, { "epoch": 3.8373669688608594, "grad_norm": 0.5068495869636536, "learning_rate": 0.0005, "loss": 1.090297818183899, "step": 48677 }, { "epoch": 3.837445802128498, "grad_norm": 0.424050897359848, "learning_rate": 0.0005, "loss": 1.0598304271697998, "step": 48678 }, { "epoch": 3.837524635396137, "grad_norm": 0.5377062559127808, "learning_rate": 0.0005, "loss": 1.1471807956695557, "step": 48679 }, { "epoch": 3.837603468663776, "grad_norm": 0.59617018699646, "learning_rate": 0.0005, "loss": 1.1488139629364014, "step": 48680 }, { "epoch": 3.837682301931415, "grad_norm": 0.4791497588157654, "learning_rate": 0.0005, "loss": 1.1871569156646729, "step": 48681 }, { "epoch": 3.837761135199054, "grad_norm": 0.4670782685279846, "learning_rate": 0.0005, "loss": 1.2535337209701538, "step": 48682 }, { "epoch": 3.837839968466693, "grad_norm": 0.48234114050865173, "learning_rate": 0.0005, "loss": 1.2201400995254517, "step": 48683 }, { "epoch": 3.837918801734332, "grad_norm": 0.5201331973075867, "learning_rate": 0.0005, "loss": 1.2598137855529785, "step": 48684 }, { "epoch": 3.8379976350019707, "grad_norm": 0.533051609992981, "learning_rate": 0.0005, "loss": 1.235390543937683, "step": 48685 }, { "epoch": 3.83807646826961, "grad_norm": 0.6017440557479858, "learning_rate": 0.0005, "loss": 1.174152135848999, "step": 48686 }, { "epoch": 3.838155301537249, "grad_norm": 0.45860686898231506, "learning_rate": 0.0005, "loss": 1.3668415546417236, "step": 48687 }, { "epoch": 3.8382341348048876, "grad_norm": 0.4949546158313751, "learning_rate": 0.0005, "loss": 1.1674014329910278, "step": 48688 }, { "epoch": 3.8383129680725268, "grad_norm": 0.4710431694984436, "learning_rate": 0.0005, "loss": 1.1393896341323853, "step": 48689 }, { "epoch": 3.8383918013401654, "grad_norm": 0.4658232033252716, "learning_rate": 0.0005, "loss": 1.3001493215560913, "step": 48690 }, { "epoch": 3.8384706346078046, "grad_norm": 0.522002637386322, "learning_rate": 0.0005, "loss": 1.075950026512146, "step": 48691 }, { "epoch": 3.8385494678754437, "grad_norm": 0.46279993653297424, "learning_rate": 0.0005, "loss": 1.1894104480743408, "step": 48692 }, { "epoch": 3.8386283011430824, "grad_norm": 0.42286840081214905, "learning_rate": 0.0005, "loss": 1.0980114936828613, "step": 48693 }, { "epoch": 3.838707134410721, "grad_norm": 0.5702160000801086, "learning_rate": 0.0005, "loss": 1.4361916780471802, "step": 48694 }, { "epoch": 3.8387859676783602, "grad_norm": 0.46939194202423096, "learning_rate": 0.0005, "loss": 1.2504559755325317, "step": 48695 }, { "epoch": 3.8388648009459994, "grad_norm": 0.5408172607421875, "learning_rate": 0.0005, "loss": 1.3413516283035278, "step": 48696 }, { "epoch": 3.838943634213638, "grad_norm": 0.4396863281726837, "learning_rate": 0.0005, "loss": 1.0041725635528564, "step": 48697 }, { "epoch": 3.839022467481277, "grad_norm": 0.43035733699798584, "learning_rate": 0.0005, "loss": 1.0858328342437744, "step": 48698 }, { "epoch": 3.839101300748916, "grad_norm": 0.49244099855422974, "learning_rate": 0.0005, "loss": 1.0534000396728516, "step": 48699 }, { "epoch": 3.839180134016555, "grad_norm": 0.5064775347709656, "learning_rate": 0.0005, "loss": 1.1771245002746582, "step": 48700 }, { "epoch": 3.839258967284194, "grad_norm": 0.498022198677063, "learning_rate": 0.0005, "loss": 1.0558645725250244, "step": 48701 }, { "epoch": 3.839337800551833, "grad_norm": 0.4600456953048706, "learning_rate": 0.0005, "loss": 1.027095079421997, "step": 48702 }, { "epoch": 3.839416633819472, "grad_norm": 0.7024398446083069, "learning_rate": 0.0005, "loss": 1.2313458919525146, "step": 48703 }, { "epoch": 3.8394954670871106, "grad_norm": 0.45606178045272827, "learning_rate": 0.0005, "loss": 1.1881859302520752, "step": 48704 }, { "epoch": 3.83957430035475, "grad_norm": 0.532250165939331, "learning_rate": 0.0005, "loss": 1.140562653541565, "step": 48705 }, { "epoch": 3.839653133622389, "grad_norm": 0.5074239373207092, "learning_rate": 0.0005, "loss": 1.1222361326217651, "step": 48706 }, { "epoch": 3.8397319668900276, "grad_norm": 0.5447521805763245, "learning_rate": 0.0005, "loss": 1.1507564783096313, "step": 48707 }, { "epoch": 3.8398108001576663, "grad_norm": 0.5699273347854614, "learning_rate": 0.0005, "loss": 1.3633842468261719, "step": 48708 }, { "epoch": 3.8398896334253054, "grad_norm": 0.49519282579421997, "learning_rate": 0.0005, "loss": 1.230663537979126, "step": 48709 }, { "epoch": 3.8399684666929446, "grad_norm": 0.6104264855384827, "learning_rate": 0.0005, "loss": 1.2168645858764648, "step": 48710 }, { "epoch": 3.8400472999605832, "grad_norm": 0.7063037753105164, "learning_rate": 0.0005, "loss": 1.161485195159912, "step": 48711 }, { "epoch": 3.8401261332282224, "grad_norm": 0.34551069140434265, "learning_rate": 0.0005, "loss": 0.9528970718383789, "step": 48712 }, { "epoch": 3.840204966495861, "grad_norm": 0.48543745279312134, "learning_rate": 0.0005, "loss": 1.2454923391342163, "step": 48713 }, { "epoch": 3.8402837997635, "grad_norm": 0.6021450161933899, "learning_rate": 0.0005, "loss": 1.2255089282989502, "step": 48714 }, { "epoch": 3.8403626330311393, "grad_norm": 0.434023916721344, "learning_rate": 0.0005, "loss": 1.0383988618850708, "step": 48715 }, { "epoch": 3.840441466298778, "grad_norm": 0.4303532838821411, "learning_rate": 0.0005, "loss": 1.203869342803955, "step": 48716 }, { "epoch": 3.840520299566417, "grad_norm": 0.5257019996643066, "learning_rate": 0.0005, "loss": 1.3526322841644287, "step": 48717 }, { "epoch": 3.840599132834056, "grad_norm": 0.4544902443885803, "learning_rate": 0.0005, "loss": 1.2570390701293945, "step": 48718 }, { "epoch": 3.840677966101695, "grad_norm": 0.41727763414382935, "learning_rate": 0.0005, "loss": 0.9997936487197876, "step": 48719 }, { "epoch": 3.840756799369334, "grad_norm": 0.5005797147750854, "learning_rate": 0.0005, "loss": 1.1322081089019775, "step": 48720 }, { "epoch": 3.840835632636973, "grad_norm": 0.545414924621582, "learning_rate": 0.0005, "loss": 1.363480567932129, "step": 48721 }, { "epoch": 3.8409144659046115, "grad_norm": 0.46159806847572327, "learning_rate": 0.0005, "loss": 1.12717604637146, "step": 48722 }, { "epoch": 3.8409932991722506, "grad_norm": 0.4989556670188904, "learning_rate": 0.0005, "loss": 1.3251471519470215, "step": 48723 }, { "epoch": 3.8410721324398898, "grad_norm": 0.48140764236450195, "learning_rate": 0.0005, "loss": 1.1831483840942383, "step": 48724 }, { "epoch": 3.8411509657075285, "grad_norm": 0.537610650062561, "learning_rate": 0.0005, "loss": 1.2236982583999634, "step": 48725 }, { "epoch": 3.8412297989751676, "grad_norm": 0.5193648338317871, "learning_rate": 0.0005, "loss": 1.275230884552002, "step": 48726 }, { "epoch": 3.8413086322428063, "grad_norm": 0.5850437879562378, "learning_rate": 0.0005, "loss": 1.2454173564910889, "step": 48727 }, { "epoch": 3.8413874655104454, "grad_norm": 0.4938422739505768, "learning_rate": 0.0005, "loss": 1.26255202293396, "step": 48728 }, { "epoch": 3.8414662987780845, "grad_norm": 0.5022099018096924, "learning_rate": 0.0005, "loss": 1.2462437152862549, "step": 48729 }, { "epoch": 3.8415451320457232, "grad_norm": 0.554889976978302, "learning_rate": 0.0005, "loss": 1.194400429725647, "step": 48730 }, { "epoch": 3.8416239653133624, "grad_norm": 0.4597931504249573, "learning_rate": 0.0005, "loss": 1.2369301319122314, "step": 48731 }, { "epoch": 3.841702798581001, "grad_norm": 0.5534363985061646, "learning_rate": 0.0005, "loss": 1.024537205696106, "step": 48732 }, { "epoch": 3.84178163184864, "grad_norm": 0.4538573622703552, "learning_rate": 0.0005, "loss": 1.1755367517471313, "step": 48733 }, { "epoch": 3.8418604651162793, "grad_norm": 0.5862858295440674, "learning_rate": 0.0005, "loss": 1.3162493705749512, "step": 48734 }, { "epoch": 3.841939298383918, "grad_norm": 0.6863496899604797, "learning_rate": 0.0005, "loss": 1.059556007385254, "step": 48735 }, { "epoch": 3.8420181316515567, "grad_norm": 0.5080035924911499, "learning_rate": 0.0005, "loss": 1.0493710041046143, "step": 48736 }, { "epoch": 3.842096964919196, "grad_norm": 0.5233174562454224, "learning_rate": 0.0005, "loss": 0.9967912435531616, "step": 48737 }, { "epoch": 3.842175798186835, "grad_norm": 0.5281704068183899, "learning_rate": 0.0005, "loss": 1.1469738483428955, "step": 48738 }, { "epoch": 3.8422546314544737, "grad_norm": 0.5553197860717773, "learning_rate": 0.0005, "loss": 1.2029306888580322, "step": 48739 }, { "epoch": 3.842333464722113, "grad_norm": 0.5010113716125488, "learning_rate": 0.0005, "loss": 1.2049744129180908, "step": 48740 }, { "epoch": 3.8424122979897515, "grad_norm": 0.3975684940814972, "learning_rate": 0.0005, "loss": 1.0193657875061035, "step": 48741 }, { "epoch": 3.8424911312573906, "grad_norm": 0.5163360238075256, "learning_rate": 0.0005, "loss": 1.2352004051208496, "step": 48742 }, { "epoch": 3.8425699645250297, "grad_norm": 0.4122385084629059, "learning_rate": 0.0005, "loss": 1.0676497220993042, "step": 48743 }, { "epoch": 3.8426487977926684, "grad_norm": 0.5096925497055054, "learning_rate": 0.0005, "loss": 1.2033421993255615, "step": 48744 }, { "epoch": 3.8427276310603076, "grad_norm": 0.4530298709869385, "learning_rate": 0.0005, "loss": 1.0358282327651978, "step": 48745 }, { "epoch": 3.8428064643279463, "grad_norm": 0.45262324810028076, "learning_rate": 0.0005, "loss": 1.2217785120010376, "step": 48746 }, { "epoch": 3.8428852975955854, "grad_norm": 0.8228419423103333, "learning_rate": 0.0005, "loss": 1.091526985168457, "step": 48747 }, { "epoch": 3.8429641308632245, "grad_norm": 0.44294703006744385, "learning_rate": 0.0005, "loss": 1.2461458444595337, "step": 48748 }, { "epoch": 3.843042964130863, "grad_norm": 0.47753843665122986, "learning_rate": 0.0005, "loss": 1.0210882425308228, "step": 48749 }, { "epoch": 3.843121797398502, "grad_norm": 0.5207666158676147, "learning_rate": 0.0005, "loss": 1.1698213815689087, "step": 48750 }, { "epoch": 3.843200630666141, "grad_norm": 0.4287883937358856, "learning_rate": 0.0005, "loss": 1.0711452960968018, "step": 48751 }, { "epoch": 3.84327946393378, "grad_norm": 0.8655155301094055, "learning_rate": 0.0005, "loss": 1.2599537372589111, "step": 48752 }, { "epoch": 3.843358297201419, "grad_norm": 0.4495101273059845, "learning_rate": 0.0005, "loss": 1.2192003726959229, "step": 48753 }, { "epoch": 3.843437130469058, "grad_norm": 0.5878040790557861, "learning_rate": 0.0005, "loss": 1.147704839706421, "step": 48754 }, { "epoch": 3.8435159637366967, "grad_norm": 0.5908733010292053, "learning_rate": 0.0005, "loss": 1.2754437923431396, "step": 48755 }, { "epoch": 3.843594797004336, "grad_norm": 0.5260767340660095, "learning_rate": 0.0005, "loss": 1.0362507104873657, "step": 48756 }, { "epoch": 3.843673630271975, "grad_norm": 0.45743346214294434, "learning_rate": 0.0005, "loss": 1.2864038944244385, "step": 48757 }, { "epoch": 3.8437524635396136, "grad_norm": 0.5502171516418457, "learning_rate": 0.0005, "loss": 1.3185440301895142, "step": 48758 }, { "epoch": 3.8438312968072528, "grad_norm": 0.5378175973892212, "learning_rate": 0.0005, "loss": 1.3425731658935547, "step": 48759 }, { "epoch": 3.8439101300748915, "grad_norm": 0.5004507899284363, "learning_rate": 0.0005, "loss": 1.13912034034729, "step": 48760 }, { "epoch": 3.8439889633425306, "grad_norm": 0.5356011390686035, "learning_rate": 0.0005, "loss": 1.2485358715057373, "step": 48761 }, { "epoch": 3.8440677966101697, "grad_norm": 0.4990798532962799, "learning_rate": 0.0005, "loss": 1.143815279006958, "step": 48762 }, { "epoch": 3.8441466298778084, "grad_norm": 0.402849406003952, "learning_rate": 0.0005, "loss": 0.9899091720581055, "step": 48763 }, { "epoch": 3.844225463145447, "grad_norm": 0.4805634021759033, "learning_rate": 0.0005, "loss": 1.0646090507507324, "step": 48764 }, { "epoch": 3.8443042964130862, "grad_norm": 0.5869913101196289, "learning_rate": 0.0005, "loss": 1.1320507526397705, "step": 48765 }, { "epoch": 3.8443831296807254, "grad_norm": 0.46364614367485046, "learning_rate": 0.0005, "loss": 1.3760948181152344, "step": 48766 }, { "epoch": 3.844461962948364, "grad_norm": 0.5200822949409485, "learning_rate": 0.0005, "loss": 1.3663325309753418, "step": 48767 }, { "epoch": 3.844540796216003, "grad_norm": 0.5247843265533447, "learning_rate": 0.0005, "loss": 1.2729771137237549, "step": 48768 }, { "epoch": 3.844619629483642, "grad_norm": 0.6885277032852173, "learning_rate": 0.0005, "loss": 1.3301739692687988, "step": 48769 }, { "epoch": 3.844698462751281, "grad_norm": 0.4559118449687958, "learning_rate": 0.0005, "loss": 1.141876220703125, "step": 48770 }, { "epoch": 3.84477729601892, "grad_norm": 0.6164612770080566, "learning_rate": 0.0005, "loss": 1.1397762298583984, "step": 48771 }, { "epoch": 3.844856129286559, "grad_norm": 0.5437923669815063, "learning_rate": 0.0005, "loss": 1.2377740144729614, "step": 48772 }, { "epoch": 3.844934962554198, "grad_norm": 0.5126107335090637, "learning_rate": 0.0005, "loss": 1.0653076171875, "step": 48773 }, { "epoch": 3.8450137958218367, "grad_norm": 0.4295850396156311, "learning_rate": 0.0005, "loss": 1.1626224517822266, "step": 48774 }, { "epoch": 3.845092629089476, "grad_norm": 0.5276649594306946, "learning_rate": 0.0005, "loss": 1.146883487701416, "step": 48775 }, { "epoch": 3.845171462357115, "grad_norm": 0.48199209570884705, "learning_rate": 0.0005, "loss": 1.278214693069458, "step": 48776 }, { "epoch": 3.8452502956247536, "grad_norm": 0.42892682552337646, "learning_rate": 0.0005, "loss": 0.9983780980110168, "step": 48777 }, { "epoch": 3.8453291288923928, "grad_norm": 0.6077394485473633, "learning_rate": 0.0005, "loss": 1.0242466926574707, "step": 48778 }, { "epoch": 3.8454079621600314, "grad_norm": 0.5316911339759827, "learning_rate": 0.0005, "loss": 1.220151662826538, "step": 48779 }, { "epoch": 3.8454867954276706, "grad_norm": 0.4723118841648102, "learning_rate": 0.0005, "loss": 1.2030036449432373, "step": 48780 }, { "epoch": 3.8455656286953093, "grad_norm": 0.4353686571121216, "learning_rate": 0.0005, "loss": 1.1413557529449463, "step": 48781 }, { "epoch": 3.8456444619629484, "grad_norm": 0.4610616862773895, "learning_rate": 0.0005, "loss": 1.1584781408309937, "step": 48782 }, { "epoch": 3.845723295230587, "grad_norm": 0.4463038146495819, "learning_rate": 0.0005, "loss": 1.2051290273666382, "step": 48783 }, { "epoch": 3.8458021284982262, "grad_norm": 0.5513209700584412, "learning_rate": 0.0005, "loss": 1.181958556175232, "step": 48784 }, { "epoch": 3.8458809617658654, "grad_norm": 0.4830755591392517, "learning_rate": 0.0005, "loss": 0.9553583264350891, "step": 48785 }, { "epoch": 3.845959795033504, "grad_norm": 0.7346814274787903, "learning_rate": 0.0005, "loss": 1.2726449966430664, "step": 48786 }, { "epoch": 3.846038628301143, "grad_norm": 0.534544825553894, "learning_rate": 0.0005, "loss": 1.2522013187408447, "step": 48787 }, { "epoch": 3.846117461568782, "grad_norm": 0.49168843030929565, "learning_rate": 0.0005, "loss": 1.2761070728302002, "step": 48788 }, { "epoch": 3.846196294836421, "grad_norm": 0.4356499910354614, "learning_rate": 0.0005, "loss": 1.1789026260375977, "step": 48789 }, { "epoch": 3.84627512810406, "grad_norm": 0.4945264756679535, "learning_rate": 0.0005, "loss": 1.1966828107833862, "step": 48790 }, { "epoch": 3.846353961371699, "grad_norm": 0.47113871574401855, "learning_rate": 0.0005, "loss": 1.134157419204712, "step": 48791 }, { "epoch": 3.846432794639338, "grad_norm": 0.7870200872421265, "learning_rate": 0.0005, "loss": 1.2940102815628052, "step": 48792 }, { "epoch": 3.8465116279069766, "grad_norm": 0.46415677666664124, "learning_rate": 0.0005, "loss": 1.2016682624816895, "step": 48793 }, { "epoch": 3.846590461174616, "grad_norm": 0.45724380016326904, "learning_rate": 0.0005, "loss": 1.2313357591629028, "step": 48794 }, { "epoch": 3.8466692944422545, "grad_norm": 0.5594639182090759, "learning_rate": 0.0005, "loss": 1.0997370481491089, "step": 48795 }, { "epoch": 3.8467481277098936, "grad_norm": 0.5071156024932861, "learning_rate": 0.0005, "loss": 1.0536216497421265, "step": 48796 }, { "epoch": 3.8468269609775323, "grad_norm": 0.44681620597839355, "learning_rate": 0.0005, "loss": 1.085934042930603, "step": 48797 }, { "epoch": 3.8469057942451714, "grad_norm": 0.47743406891822815, "learning_rate": 0.0005, "loss": 1.0826069116592407, "step": 48798 }, { "epoch": 3.8469846275128106, "grad_norm": 0.47918224334716797, "learning_rate": 0.0005, "loss": 1.1888682842254639, "step": 48799 }, { "epoch": 3.8470634607804493, "grad_norm": 0.46477407217025757, "learning_rate": 0.0005, "loss": 1.0674352645874023, "step": 48800 }, { "epoch": 3.8471422940480884, "grad_norm": 0.4426232576370239, "learning_rate": 0.0005, "loss": 1.113877534866333, "step": 48801 }, { "epoch": 3.847221127315727, "grad_norm": 0.4587891697883606, "learning_rate": 0.0005, "loss": 1.1181180477142334, "step": 48802 }, { "epoch": 3.847299960583366, "grad_norm": 0.629951536655426, "learning_rate": 0.0005, "loss": 1.446319341659546, "step": 48803 }, { "epoch": 3.8473787938510053, "grad_norm": 0.562563419342041, "learning_rate": 0.0005, "loss": 1.192321538925171, "step": 48804 }, { "epoch": 3.847457627118644, "grad_norm": 0.5114551186561584, "learning_rate": 0.0005, "loss": 1.293954849243164, "step": 48805 }, { "epoch": 3.847536460386283, "grad_norm": 0.5928202271461487, "learning_rate": 0.0005, "loss": 1.4358983039855957, "step": 48806 }, { "epoch": 3.847615293653922, "grad_norm": 0.4561229348182678, "learning_rate": 0.0005, "loss": 1.1143083572387695, "step": 48807 }, { "epoch": 3.847694126921561, "grad_norm": 0.4523024260997772, "learning_rate": 0.0005, "loss": 1.1353135108947754, "step": 48808 }, { "epoch": 3.8477729601892, "grad_norm": 0.4903380870819092, "learning_rate": 0.0005, "loss": 1.2505019903182983, "step": 48809 }, { "epoch": 3.847851793456839, "grad_norm": 0.4953899681568146, "learning_rate": 0.0005, "loss": 1.130507230758667, "step": 48810 }, { "epoch": 3.8479306267244775, "grad_norm": 0.4525277018547058, "learning_rate": 0.0005, "loss": 1.1014816761016846, "step": 48811 }, { "epoch": 3.8480094599921166, "grad_norm": 0.4512292146682739, "learning_rate": 0.0005, "loss": 1.0818166732788086, "step": 48812 }, { "epoch": 3.8480882932597558, "grad_norm": 0.4943208694458008, "learning_rate": 0.0005, "loss": 1.1408905982971191, "step": 48813 }, { "epoch": 3.8481671265273945, "grad_norm": 0.4825241267681122, "learning_rate": 0.0005, "loss": 1.1057287454605103, "step": 48814 }, { "epoch": 3.8482459597950336, "grad_norm": 0.5190248489379883, "learning_rate": 0.0005, "loss": 1.2154545783996582, "step": 48815 }, { "epoch": 3.8483247930626723, "grad_norm": 0.41249221563339233, "learning_rate": 0.0005, "loss": 0.956923246383667, "step": 48816 }, { "epoch": 3.8484036263303114, "grad_norm": 0.5099682211875916, "learning_rate": 0.0005, "loss": 1.141812801361084, "step": 48817 }, { "epoch": 3.8484824595979505, "grad_norm": 0.6519246101379395, "learning_rate": 0.0005, "loss": 1.258775234222412, "step": 48818 }, { "epoch": 3.8485612928655892, "grad_norm": 0.572439432144165, "learning_rate": 0.0005, "loss": 1.2701268196105957, "step": 48819 }, { "epoch": 3.8486401261332284, "grad_norm": 0.4996885657310486, "learning_rate": 0.0005, "loss": 1.1140965223312378, "step": 48820 }, { "epoch": 3.848718959400867, "grad_norm": 0.5261220932006836, "learning_rate": 0.0005, "loss": 1.2461192607879639, "step": 48821 }, { "epoch": 3.848797792668506, "grad_norm": 0.43665584921836853, "learning_rate": 0.0005, "loss": 1.0815050601959229, "step": 48822 }, { "epoch": 3.8488766259361453, "grad_norm": 0.4439815282821655, "learning_rate": 0.0005, "loss": 1.0492116212844849, "step": 48823 }, { "epoch": 3.848955459203784, "grad_norm": 0.4285992681980133, "learning_rate": 0.0005, "loss": 1.1107709407806396, "step": 48824 }, { "epoch": 3.8490342924714227, "grad_norm": 0.529842734336853, "learning_rate": 0.0005, "loss": 1.2633881568908691, "step": 48825 }, { "epoch": 3.849113125739062, "grad_norm": 0.38489463925361633, "learning_rate": 0.0005, "loss": 1.1508582830429077, "step": 48826 }, { "epoch": 3.849191959006701, "grad_norm": 0.4808204770088196, "learning_rate": 0.0005, "loss": 1.212255835533142, "step": 48827 }, { "epoch": 3.8492707922743397, "grad_norm": 0.40958547592163086, "learning_rate": 0.0005, "loss": 1.0141915082931519, "step": 48828 }, { "epoch": 3.849349625541979, "grad_norm": 0.527841329574585, "learning_rate": 0.0005, "loss": 1.1222624778747559, "step": 48829 }, { "epoch": 3.8494284588096175, "grad_norm": 0.43267959356307983, "learning_rate": 0.0005, "loss": 1.263289213180542, "step": 48830 }, { "epoch": 3.8495072920772566, "grad_norm": 0.42438340187072754, "learning_rate": 0.0005, "loss": 1.1099772453308105, "step": 48831 }, { "epoch": 3.8495861253448957, "grad_norm": 0.5112963914871216, "learning_rate": 0.0005, "loss": 1.2918553352355957, "step": 48832 }, { "epoch": 3.8496649586125344, "grad_norm": 0.5269396901130676, "learning_rate": 0.0005, "loss": 1.1045382022857666, "step": 48833 }, { "epoch": 3.8497437918801736, "grad_norm": 0.4094468355178833, "learning_rate": 0.0005, "loss": 1.063862919807434, "step": 48834 }, { "epoch": 3.8498226251478123, "grad_norm": 0.45933547616004944, "learning_rate": 0.0005, "loss": 1.215681552886963, "step": 48835 }, { "epoch": 3.8499014584154514, "grad_norm": 0.48693573474884033, "learning_rate": 0.0005, "loss": 1.1197255849838257, "step": 48836 }, { "epoch": 3.8499802916830905, "grad_norm": 0.6836509704589844, "learning_rate": 0.0005, "loss": 1.157791018486023, "step": 48837 }, { "epoch": 3.850059124950729, "grad_norm": 0.5211443901062012, "learning_rate": 0.0005, "loss": 1.3484506607055664, "step": 48838 }, { "epoch": 3.850137958218368, "grad_norm": 0.4055321216583252, "learning_rate": 0.0005, "loss": 1.1314479112625122, "step": 48839 }, { "epoch": 3.850216791486007, "grad_norm": 1.306140661239624, "learning_rate": 0.0005, "loss": 1.269657850265503, "step": 48840 }, { "epoch": 3.850295624753646, "grad_norm": 0.4544272720813751, "learning_rate": 0.0005, "loss": 1.028778076171875, "step": 48841 }, { "epoch": 3.850374458021285, "grad_norm": 0.6271017789840698, "learning_rate": 0.0005, "loss": 1.217729091644287, "step": 48842 }, { "epoch": 3.850453291288924, "grad_norm": 0.4467664361000061, "learning_rate": 0.0005, "loss": 1.2278718948364258, "step": 48843 }, { "epoch": 3.8505321245565627, "grad_norm": 0.483797550201416, "learning_rate": 0.0005, "loss": 1.0731830596923828, "step": 48844 }, { "epoch": 3.850610957824202, "grad_norm": 0.5231785774230957, "learning_rate": 0.0005, "loss": 1.024672508239746, "step": 48845 }, { "epoch": 3.850689791091841, "grad_norm": 0.4884592890739441, "learning_rate": 0.0005, "loss": 1.2849735021591187, "step": 48846 }, { "epoch": 3.8507686243594796, "grad_norm": 0.5389895439147949, "learning_rate": 0.0005, "loss": 1.0430092811584473, "step": 48847 }, { "epoch": 3.8508474576271188, "grad_norm": 0.519702136516571, "learning_rate": 0.0005, "loss": 1.169351577758789, "step": 48848 }, { "epoch": 3.8509262908947575, "grad_norm": 0.8165713548660278, "learning_rate": 0.0005, "loss": 1.316246509552002, "step": 48849 }, { "epoch": 3.8510051241623966, "grad_norm": 0.39793428778648376, "learning_rate": 0.0005, "loss": 1.1768490076065063, "step": 48850 }, { "epoch": 3.8510839574300357, "grad_norm": 0.46258631348609924, "learning_rate": 0.0005, "loss": 1.1133801937103271, "step": 48851 }, { "epoch": 3.8511627906976744, "grad_norm": 0.49349483847618103, "learning_rate": 0.0005, "loss": 1.1864042282104492, "step": 48852 }, { "epoch": 3.851241623965313, "grad_norm": 0.5124980807304382, "learning_rate": 0.0005, "loss": 1.076442003250122, "step": 48853 }, { "epoch": 3.8513204572329522, "grad_norm": 0.45930495858192444, "learning_rate": 0.0005, "loss": 1.13046133518219, "step": 48854 }, { "epoch": 3.8513992905005914, "grad_norm": 0.46449384093284607, "learning_rate": 0.0005, "loss": 1.1224794387817383, "step": 48855 }, { "epoch": 3.85147812376823, "grad_norm": 0.5717353224754333, "learning_rate": 0.0005, "loss": 1.2203036546707153, "step": 48856 }, { "epoch": 3.851556957035869, "grad_norm": 0.5530368685722351, "learning_rate": 0.0005, "loss": 1.3048796653747559, "step": 48857 }, { "epoch": 3.851635790303508, "grad_norm": 0.43588894605636597, "learning_rate": 0.0005, "loss": 1.1024937629699707, "step": 48858 }, { "epoch": 3.851714623571147, "grad_norm": 0.7570782899856567, "learning_rate": 0.0005, "loss": 1.3117148876190186, "step": 48859 }, { "epoch": 3.851793456838786, "grad_norm": 0.5013585090637207, "learning_rate": 0.0005, "loss": 1.0865590572357178, "step": 48860 }, { "epoch": 3.851872290106425, "grad_norm": 0.45746079087257385, "learning_rate": 0.0005, "loss": 1.2261216640472412, "step": 48861 }, { "epoch": 3.851951123374064, "grad_norm": 0.48406580090522766, "learning_rate": 0.0005, "loss": 1.1293838024139404, "step": 48862 }, { "epoch": 3.8520299566417027, "grad_norm": 0.6266694664955139, "learning_rate": 0.0005, "loss": 1.2556051015853882, "step": 48863 }, { "epoch": 3.852108789909342, "grad_norm": 0.4559711217880249, "learning_rate": 0.0005, "loss": 1.0890823602676392, "step": 48864 }, { "epoch": 3.852187623176981, "grad_norm": 0.4121765196323395, "learning_rate": 0.0005, "loss": 0.9515732526779175, "step": 48865 }, { "epoch": 3.8522664564446196, "grad_norm": 0.49564406275749207, "learning_rate": 0.0005, "loss": 1.211906909942627, "step": 48866 }, { "epoch": 3.8523452897122583, "grad_norm": 0.39336997270584106, "learning_rate": 0.0005, "loss": 1.0731899738311768, "step": 48867 }, { "epoch": 3.8524241229798974, "grad_norm": 0.46796658635139465, "learning_rate": 0.0005, "loss": 1.2261775732040405, "step": 48868 }, { "epoch": 3.8525029562475366, "grad_norm": 0.5085837841033936, "learning_rate": 0.0005, "loss": 1.305495262145996, "step": 48869 }, { "epoch": 3.8525817895151753, "grad_norm": 0.5296556949615479, "learning_rate": 0.0005, "loss": 1.3207157850265503, "step": 48870 }, { "epoch": 3.8526606227828144, "grad_norm": 0.4773525893688202, "learning_rate": 0.0005, "loss": 1.1554492712020874, "step": 48871 }, { "epoch": 3.852739456050453, "grad_norm": 0.5242877006530762, "learning_rate": 0.0005, "loss": 1.2475578784942627, "step": 48872 }, { "epoch": 3.8528182893180922, "grad_norm": 0.4770953357219696, "learning_rate": 0.0005, "loss": 1.127037763595581, "step": 48873 }, { "epoch": 3.8528971225857314, "grad_norm": 0.4655989408493042, "learning_rate": 0.0005, "loss": 1.1182705163955688, "step": 48874 }, { "epoch": 3.85297595585337, "grad_norm": 0.4744219183921814, "learning_rate": 0.0005, "loss": 1.250996708869934, "step": 48875 }, { "epoch": 3.853054789121009, "grad_norm": 0.4753183424472809, "learning_rate": 0.0005, "loss": 1.227938175201416, "step": 48876 }, { "epoch": 3.853133622388648, "grad_norm": 0.42770877480506897, "learning_rate": 0.0005, "loss": 1.1715478897094727, "step": 48877 }, { "epoch": 3.853212455656287, "grad_norm": 0.5274155735969543, "learning_rate": 0.0005, "loss": 1.126147985458374, "step": 48878 }, { "epoch": 3.853291288923926, "grad_norm": 0.5792955160140991, "learning_rate": 0.0005, "loss": 1.2143925428390503, "step": 48879 }, { "epoch": 3.853370122191565, "grad_norm": 0.5013130307197571, "learning_rate": 0.0005, "loss": 1.227226734161377, "step": 48880 }, { "epoch": 3.8534489554592035, "grad_norm": 0.45775389671325684, "learning_rate": 0.0005, "loss": 1.2080402374267578, "step": 48881 }, { "epoch": 3.8535277887268427, "grad_norm": 0.5232504606246948, "learning_rate": 0.0005, "loss": 1.2076416015625, "step": 48882 }, { "epoch": 3.853606621994482, "grad_norm": 0.6578025817871094, "learning_rate": 0.0005, "loss": 1.345609188079834, "step": 48883 }, { "epoch": 3.8536854552621205, "grad_norm": 0.4522051215171814, "learning_rate": 0.0005, "loss": 1.2378466129302979, "step": 48884 }, { "epoch": 3.8537642885297596, "grad_norm": 0.49192559719085693, "learning_rate": 0.0005, "loss": 1.240129828453064, "step": 48885 }, { "epoch": 3.8538431217973983, "grad_norm": 0.4599175751209259, "learning_rate": 0.0005, "loss": 1.0076050758361816, "step": 48886 }, { "epoch": 3.8539219550650374, "grad_norm": 0.41780886054039, "learning_rate": 0.0005, "loss": 1.2249577045440674, "step": 48887 }, { "epoch": 3.8540007883326766, "grad_norm": 0.44460782408714294, "learning_rate": 0.0005, "loss": 1.0492405891418457, "step": 48888 }, { "epoch": 3.8540796216003153, "grad_norm": 0.44489941000938416, "learning_rate": 0.0005, "loss": 1.122649908065796, "step": 48889 }, { "epoch": 3.8541584548679544, "grad_norm": 0.4618558883666992, "learning_rate": 0.0005, "loss": 1.0422149896621704, "step": 48890 }, { "epoch": 3.854237288135593, "grad_norm": 0.46993380784988403, "learning_rate": 0.0005, "loss": 1.2379722595214844, "step": 48891 }, { "epoch": 3.854316121403232, "grad_norm": 0.42531269788742065, "learning_rate": 0.0005, "loss": 1.0925488471984863, "step": 48892 }, { "epoch": 3.8543949546708713, "grad_norm": 0.4245533049106598, "learning_rate": 0.0005, "loss": 1.1560447216033936, "step": 48893 }, { "epoch": 3.85447378793851, "grad_norm": 0.5201257467269897, "learning_rate": 0.0005, "loss": 1.1096280813217163, "step": 48894 }, { "epoch": 3.8545526212061487, "grad_norm": 0.5112675428390503, "learning_rate": 0.0005, "loss": 1.1647425889968872, "step": 48895 }, { "epoch": 3.854631454473788, "grad_norm": 0.554954469203949, "learning_rate": 0.0005, "loss": 1.3645248413085938, "step": 48896 }, { "epoch": 3.854710287741427, "grad_norm": 0.5384453535079956, "learning_rate": 0.0005, "loss": 1.135662317276001, "step": 48897 }, { "epoch": 3.8547891210090657, "grad_norm": 0.4529140889644623, "learning_rate": 0.0005, "loss": 1.1358741521835327, "step": 48898 }, { "epoch": 3.854867954276705, "grad_norm": 9.965371131896973, "learning_rate": 0.0005, "loss": 1.1130731105804443, "step": 48899 }, { "epoch": 3.8549467875443435, "grad_norm": 0.4666580855846405, "learning_rate": 0.0005, "loss": 1.0701549053192139, "step": 48900 }, { "epoch": 3.8550256208119826, "grad_norm": 0.44715529680252075, "learning_rate": 0.0005, "loss": 1.1325125694274902, "step": 48901 }, { "epoch": 3.8551044540796218, "grad_norm": 0.7118784189224243, "learning_rate": 0.0005, "loss": 1.2028114795684814, "step": 48902 }, { "epoch": 3.8551832873472605, "grad_norm": 0.5146132111549377, "learning_rate": 0.0005, "loss": 1.1697144508361816, "step": 48903 }, { "epoch": 3.8552621206148996, "grad_norm": 0.5305711627006531, "learning_rate": 0.0005, "loss": 1.293639898300171, "step": 48904 }, { "epoch": 3.8553409538825383, "grad_norm": 0.6687135100364685, "learning_rate": 0.0005, "loss": 1.2707443237304688, "step": 48905 }, { "epoch": 3.8554197871501774, "grad_norm": 0.4997338354587555, "learning_rate": 0.0005, "loss": 1.0483195781707764, "step": 48906 }, { "epoch": 3.8554986204178165, "grad_norm": 0.6284534335136414, "learning_rate": 0.0005, "loss": 1.2811596393585205, "step": 48907 }, { "epoch": 3.8555774536854552, "grad_norm": 0.49295198917388916, "learning_rate": 0.0005, "loss": 1.1542952060699463, "step": 48908 }, { "epoch": 3.8556562869530944, "grad_norm": 0.4903924763202667, "learning_rate": 0.0005, "loss": 1.273808479309082, "step": 48909 }, { "epoch": 3.855735120220733, "grad_norm": 0.47812560200691223, "learning_rate": 0.0005, "loss": 1.1698408126831055, "step": 48910 }, { "epoch": 3.855813953488372, "grad_norm": 0.5771945118904114, "learning_rate": 0.0005, "loss": 1.2125968933105469, "step": 48911 }, { "epoch": 3.855892786756011, "grad_norm": 0.5138512253761292, "learning_rate": 0.0005, "loss": 1.0462418794631958, "step": 48912 }, { "epoch": 3.85597162002365, "grad_norm": 0.42427560687065125, "learning_rate": 0.0005, "loss": 1.1147346496582031, "step": 48913 }, { "epoch": 3.8560504532912887, "grad_norm": 0.6067019701004028, "learning_rate": 0.0005, "loss": 1.1061408519744873, "step": 48914 }, { "epoch": 3.856129286558928, "grad_norm": 0.4252006411552429, "learning_rate": 0.0005, "loss": 1.120041847229004, "step": 48915 }, { "epoch": 3.856208119826567, "grad_norm": 0.5209546685218811, "learning_rate": 0.0005, "loss": 1.1900323629379272, "step": 48916 }, { "epoch": 3.8562869530942057, "grad_norm": 0.4568149447441101, "learning_rate": 0.0005, "loss": 1.1791603565216064, "step": 48917 }, { "epoch": 3.856365786361845, "grad_norm": 0.5092986226081848, "learning_rate": 0.0005, "loss": 1.2906131744384766, "step": 48918 }, { "epoch": 3.8564446196294835, "grad_norm": 0.6104934215545654, "learning_rate": 0.0005, "loss": 1.247849464416504, "step": 48919 }, { "epoch": 3.8565234528971226, "grad_norm": 0.4867924153804779, "learning_rate": 0.0005, "loss": 1.0479508638381958, "step": 48920 }, { "epoch": 3.8566022861647618, "grad_norm": 0.47169029712677, "learning_rate": 0.0005, "loss": 1.1372548341751099, "step": 48921 }, { "epoch": 3.8566811194324004, "grad_norm": 0.5951189994812012, "learning_rate": 0.0005, "loss": 1.2342004776000977, "step": 48922 }, { "epoch": 3.8567599527000396, "grad_norm": 0.4774148464202881, "learning_rate": 0.0005, "loss": 1.198257327079773, "step": 48923 }, { "epoch": 3.8568387859676783, "grad_norm": 0.755143404006958, "learning_rate": 0.0005, "loss": 1.2365821599960327, "step": 48924 }, { "epoch": 3.8569176192353174, "grad_norm": 0.467957466840744, "learning_rate": 0.0005, "loss": 1.0671101808547974, "step": 48925 }, { "epoch": 3.856996452502956, "grad_norm": 0.5875280499458313, "learning_rate": 0.0005, "loss": 1.3433899879455566, "step": 48926 }, { "epoch": 3.857075285770595, "grad_norm": 0.6248909831047058, "learning_rate": 0.0005, "loss": 1.2501667737960815, "step": 48927 }, { "epoch": 3.857154119038234, "grad_norm": 0.4605233371257782, "learning_rate": 0.0005, "loss": 1.1132433414459229, "step": 48928 }, { "epoch": 3.857232952305873, "grad_norm": 0.4972335398197174, "learning_rate": 0.0005, "loss": 1.051397681236267, "step": 48929 }, { "epoch": 3.857311785573512, "grad_norm": 0.48973456025123596, "learning_rate": 0.0005, "loss": 1.0331029891967773, "step": 48930 }, { "epoch": 3.857390618841151, "grad_norm": 0.4822016954421997, "learning_rate": 0.0005, "loss": 1.0419516563415527, "step": 48931 }, { "epoch": 3.85746945210879, "grad_norm": 0.6052600145339966, "learning_rate": 0.0005, "loss": 1.2177098989486694, "step": 48932 }, { "epoch": 3.8575482853764287, "grad_norm": 0.5590615272521973, "learning_rate": 0.0005, "loss": 1.1933726072311401, "step": 48933 }, { "epoch": 3.857627118644068, "grad_norm": 0.5846412181854248, "learning_rate": 0.0005, "loss": 1.0093114376068115, "step": 48934 }, { "epoch": 3.857705951911707, "grad_norm": 0.6064522862434387, "learning_rate": 0.0005, "loss": 1.4997451305389404, "step": 48935 }, { "epoch": 3.8577847851793456, "grad_norm": 0.37950122356414795, "learning_rate": 0.0005, "loss": 1.2159682512283325, "step": 48936 }, { "epoch": 3.8578636184469848, "grad_norm": 0.42557626962661743, "learning_rate": 0.0005, "loss": 1.2346243858337402, "step": 48937 }, { "epoch": 3.8579424517146235, "grad_norm": 0.561112105846405, "learning_rate": 0.0005, "loss": 1.1953259706497192, "step": 48938 }, { "epoch": 3.8580212849822626, "grad_norm": 0.7177362442016602, "learning_rate": 0.0005, "loss": 1.2886205911636353, "step": 48939 }, { "epoch": 3.8581001182499017, "grad_norm": 0.4610152542591095, "learning_rate": 0.0005, "loss": 1.0358963012695312, "step": 48940 }, { "epoch": 3.8581789515175404, "grad_norm": 0.5291688442230225, "learning_rate": 0.0005, "loss": 1.0822337865829468, "step": 48941 }, { "epoch": 3.858257784785179, "grad_norm": 0.44519656896591187, "learning_rate": 0.0005, "loss": 1.2517127990722656, "step": 48942 }, { "epoch": 3.8583366180528182, "grad_norm": 0.5774677991867065, "learning_rate": 0.0005, "loss": 1.3882335424423218, "step": 48943 }, { "epoch": 3.8584154513204574, "grad_norm": 0.4890795052051544, "learning_rate": 0.0005, "loss": 1.0700386762619019, "step": 48944 }, { "epoch": 3.858494284588096, "grad_norm": 0.49491479992866516, "learning_rate": 0.0005, "loss": 1.143766164779663, "step": 48945 }, { "epoch": 3.858573117855735, "grad_norm": 0.517657458782196, "learning_rate": 0.0005, "loss": 1.0984030961990356, "step": 48946 }, { "epoch": 3.858651951123374, "grad_norm": 0.5017257928848267, "learning_rate": 0.0005, "loss": 1.163799524307251, "step": 48947 }, { "epoch": 3.858730784391013, "grad_norm": 0.48572850227355957, "learning_rate": 0.0005, "loss": 1.1862246990203857, "step": 48948 }, { "epoch": 3.858809617658652, "grad_norm": 0.4741574823856354, "learning_rate": 0.0005, "loss": 1.1023257970809937, "step": 48949 }, { "epoch": 3.858888450926291, "grad_norm": 0.43219664692878723, "learning_rate": 0.0005, "loss": 1.0695523023605347, "step": 48950 }, { "epoch": 3.85896728419393, "grad_norm": 0.616679310798645, "learning_rate": 0.0005, "loss": 1.2307775020599365, "step": 48951 }, { "epoch": 3.8590461174615687, "grad_norm": 0.48092377185821533, "learning_rate": 0.0005, "loss": 1.1699965000152588, "step": 48952 }, { "epoch": 3.859124950729208, "grad_norm": 0.47829440236091614, "learning_rate": 0.0005, "loss": 1.2391154766082764, "step": 48953 }, { "epoch": 3.859203783996847, "grad_norm": 0.5583628416061401, "learning_rate": 0.0005, "loss": 1.2543871402740479, "step": 48954 }, { "epoch": 3.8592826172644856, "grad_norm": 0.4914373457431793, "learning_rate": 0.0005, "loss": 1.3593469858169556, "step": 48955 }, { "epoch": 3.8593614505321243, "grad_norm": 0.7557297348976135, "learning_rate": 0.0005, "loss": 1.3603806495666504, "step": 48956 }, { "epoch": 3.8594402837997634, "grad_norm": 0.6131967902183533, "learning_rate": 0.0005, "loss": 1.3299424648284912, "step": 48957 }, { "epoch": 3.8595191170674026, "grad_norm": 0.4454793334007263, "learning_rate": 0.0005, "loss": 1.1608006954193115, "step": 48958 }, { "epoch": 3.8595979503350413, "grad_norm": 0.5314342379570007, "learning_rate": 0.0005, "loss": 1.3029875755310059, "step": 48959 }, { "epoch": 3.8596767836026804, "grad_norm": 0.5846149325370789, "learning_rate": 0.0005, "loss": 1.1706609725952148, "step": 48960 }, { "epoch": 3.859755616870319, "grad_norm": 0.44269880652427673, "learning_rate": 0.0005, "loss": 1.1741665601730347, "step": 48961 }, { "epoch": 3.8598344501379582, "grad_norm": 0.5869947671890259, "learning_rate": 0.0005, "loss": 1.0993502140045166, "step": 48962 }, { "epoch": 3.8599132834055974, "grad_norm": 0.4601731598377228, "learning_rate": 0.0005, "loss": 1.0500373840332031, "step": 48963 }, { "epoch": 3.859992116673236, "grad_norm": 0.4326440393924713, "learning_rate": 0.0005, "loss": 1.0822737216949463, "step": 48964 }, { "epoch": 3.860070949940875, "grad_norm": 0.5032644867897034, "learning_rate": 0.0005, "loss": 1.1440048217773438, "step": 48965 }, { "epoch": 3.860149783208514, "grad_norm": 0.609575629234314, "learning_rate": 0.0005, "loss": 1.3008208274841309, "step": 48966 }, { "epoch": 3.860228616476153, "grad_norm": 0.4463267922401428, "learning_rate": 0.0005, "loss": 0.9787129759788513, "step": 48967 }, { "epoch": 3.860307449743792, "grad_norm": 0.513572096824646, "learning_rate": 0.0005, "loss": 1.1756598949432373, "step": 48968 }, { "epoch": 3.860386283011431, "grad_norm": 0.500594973564148, "learning_rate": 0.0005, "loss": 1.1516857147216797, "step": 48969 }, { "epoch": 3.8604651162790695, "grad_norm": 0.5491337180137634, "learning_rate": 0.0005, "loss": 1.2455745935440063, "step": 48970 }, { "epoch": 3.8605439495467087, "grad_norm": 0.46624627709388733, "learning_rate": 0.0005, "loss": 1.1888846158981323, "step": 48971 }, { "epoch": 3.860622782814348, "grad_norm": 0.4461190700531006, "learning_rate": 0.0005, "loss": 1.169682502746582, "step": 48972 }, { "epoch": 3.8607016160819865, "grad_norm": 0.47375068068504333, "learning_rate": 0.0005, "loss": 1.007927417755127, "step": 48973 }, { "epoch": 3.8607804493496256, "grad_norm": 0.4563691318035126, "learning_rate": 0.0005, "loss": 1.1296699047088623, "step": 48974 }, { "epoch": 3.8608592826172643, "grad_norm": 0.48709604144096375, "learning_rate": 0.0005, "loss": 1.0515446662902832, "step": 48975 }, { "epoch": 3.8609381158849034, "grad_norm": 0.5933085083961487, "learning_rate": 0.0005, "loss": 1.2582910060882568, "step": 48976 }, { "epoch": 3.8610169491525426, "grad_norm": 0.4676189720630646, "learning_rate": 0.0005, "loss": 1.137603759765625, "step": 48977 }, { "epoch": 3.8610957824201813, "grad_norm": 0.4724039137363434, "learning_rate": 0.0005, "loss": 1.1249074935913086, "step": 48978 }, { "epoch": 3.8611746156878204, "grad_norm": 0.46408089995384216, "learning_rate": 0.0005, "loss": 1.172020435333252, "step": 48979 }, { "epoch": 3.861253448955459, "grad_norm": 0.5535109639167786, "learning_rate": 0.0005, "loss": 1.0655543804168701, "step": 48980 }, { "epoch": 3.861332282223098, "grad_norm": 1.1593173742294312, "learning_rate": 0.0005, "loss": 1.255990982055664, "step": 48981 }, { "epoch": 3.8614111154907373, "grad_norm": 0.3952431380748749, "learning_rate": 0.0005, "loss": 1.1226415634155273, "step": 48982 }, { "epoch": 3.861489948758376, "grad_norm": 0.47377872467041016, "learning_rate": 0.0005, "loss": 1.148470163345337, "step": 48983 }, { "epoch": 3.8615687820260147, "grad_norm": 0.466827929019928, "learning_rate": 0.0005, "loss": 1.168105125427246, "step": 48984 }, { "epoch": 3.861647615293654, "grad_norm": 0.40498507022857666, "learning_rate": 0.0005, "loss": 1.1360580921173096, "step": 48985 }, { "epoch": 3.861726448561293, "grad_norm": 0.5834020972251892, "learning_rate": 0.0005, "loss": 1.326773762702942, "step": 48986 }, { "epoch": 3.8618052818289317, "grad_norm": 0.5584735870361328, "learning_rate": 0.0005, "loss": 1.1240119934082031, "step": 48987 }, { "epoch": 3.861884115096571, "grad_norm": 0.5228121876716614, "learning_rate": 0.0005, "loss": 1.0313653945922852, "step": 48988 }, { "epoch": 3.8619629483642095, "grad_norm": 0.49347126483917236, "learning_rate": 0.0005, "loss": 1.2476707696914673, "step": 48989 }, { "epoch": 3.8620417816318486, "grad_norm": 0.3810988664627075, "learning_rate": 0.0005, "loss": 1.181886911392212, "step": 48990 }, { "epoch": 3.8621206148994878, "grad_norm": 0.7541388273239136, "learning_rate": 0.0005, "loss": 1.4081840515136719, "step": 48991 }, { "epoch": 3.8621994481671265, "grad_norm": 0.5156072974205017, "learning_rate": 0.0005, "loss": 1.3151488304138184, "step": 48992 }, { "epoch": 3.8622782814347656, "grad_norm": 0.4991208612918854, "learning_rate": 0.0005, "loss": 1.3314402103424072, "step": 48993 }, { "epoch": 3.8623571147024043, "grad_norm": 0.5816220045089722, "learning_rate": 0.0005, "loss": 1.2147705554962158, "step": 48994 }, { "epoch": 3.8624359479700434, "grad_norm": 0.6341874003410339, "learning_rate": 0.0005, "loss": 1.0557794570922852, "step": 48995 }, { "epoch": 3.8625147812376825, "grad_norm": 0.5418493151664734, "learning_rate": 0.0005, "loss": 1.2341184616088867, "step": 48996 }, { "epoch": 3.8625936145053212, "grad_norm": 0.5150587558746338, "learning_rate": 0.0005, "loss": 1.331997036933899, "step": 48997 }, { "epoch": 3.86267244777296, "grad_norm": 0.4370201528072357, "learning_rate": 0.0005, "loss": 1.1288275718688965, "step": 48998 }, { "epoch": 3.862751281040599, "grad_norm": 0.37121251225471497, "learning_rate": 0.0005, "loss": 1.0489178895950317, "step": 48999 }, { "epoch": 3.862830114308238, "grad_norm": 0.568379819393158, "learning_rate": 0.0005, "loss": 1.2897045612335205, "step": 49000 }, { "epoch": 3.862908947575877, "grad_norm": 0.7069948315620422, "learning_rate": 0.0005, "loss": 1.0564661026000977, "step": 49001 }, { "epoch": 3.862987780843516, "grad_norm": 0.475650817155838, "learning_rate": 0.0005, "loss": 1.0155754089355469, "step": 49002 }, { "epoch": 3.8630666141111547, "grad_norm": 0.5011535286903381, "learning_rate": 0.0005, "loss": 1.2955609560012817, "step": 49003 }, { "epoch": 3.863145447378794, "grad_norm": 0.5403472781181335, "learning_rate": 0.0005, "loss": 1.191301703453064, "step": 49004 }, { "epoch": 3.863224280646433, "grad_norm": 0.6262356042861938, "learning_rate": 0.0005, "loss": 1.1270465850830078, "step": 49005 }, { "epoch": 3.8633031139140717, "grad_norm": 0.6097092032432556, "learning_rate": 0.0005, "loss": 1.2522711753845215, "step": 49006 }, { "epoch": 3.863381947181711, "grad_norm": 0.4866379499435425, "learning_rate": 0.0005, "loss": 1.2050251960754395, "step": 49007 }, { "epoch": 3.8634607804493495, "grad_norm": 0.46300479769706726, "learning_rate": 0.0005, "loss": 0.9753434658050537, "step": 49008 }, { "epoch": 3.8635396137169886, "grad_norm": 0.5442374348640442, "learning_rate": 0.0005, "loss": 1.338417649269104, "step": 49009 }, { "epoch": 3.8636184469846278, "grad_norm": 0.5054799318313599, "learning_rate": 0.0005, "loss": 1.0682388544082642, "step": 49010 }, { "epoch": 3.8636972802522664, "grad_norm": 0.570510983467102, "learning_rate": 0.0005, "loss": 1.2878669500350952, "step": 49011 }, { "epoch": 3.863776113519905, "grad_norm": 0.4487936496734619, "learning_rate": 0.0005, "loss": 0.9481328725814819, "step": 49012 }, { "epoch": 3.8638549467875443, "grad_norm": 0.5321305990219116, "learning_rate": 0.0005, "loss": 1.1654958724975586, "step": 49013 }, { "epoch": 3.8639337800551834, "grad_norm": 0.5128561854362488, "learning_rate": 0.0005, "loss": 0.9918113350868225, "step": 49014 }, { "epoch": 3.864012613322822, "grad_norm": 0.4588567614555359, "learning_rate": 0.0005, "loss": 1.1363348960876465, "step": 49015 }, { "epoch": 3.864091446590461, "grad_norm": 0.4300624132156372, "learning_rate": 0.0005, "loss": 1.0678273439407349, "step": 49016 }, { "epoch": 3.8641702798581, "grad_norm": 0.4659428298473358, "learning_rate": 0.0005, "loss": 1.1925532817840576, "step": 49017 }, { "epoch": 3.864249113125739, "grad_norm": 0.5379688143730164, "learning_rate": 0.0005, "loss": 1.25077486038208, "step": 49018 }, { "epoch": 3.864327946393378, "grad_norm": 0.7515661120414734, "learning_rate": 0.0005, "loss": 1.408033847808838, "step": 49019 }, { "epoch": 3.864406779661017, "grad_norm": 0.5493453741073608, "learning_rate": 0.0005, "loss": 1.3461706638336182, "step": 49020 }, { "epoch": 3.864485612928656, "grad_norm": 0.4318113923072815, "learning_rate": 0.0005, "loss": 1.001774549484253, "step": 49021 }, { "epoch": 3.8645644461962947, "grad_norm": 0.48719507455825806, "learning_rate": 0.0005, "loss": 1.1736845970153809, "step": 49022 }, { "epoch": 3.864643279463934, "grad_norm": 0.4965319335460663, "learning_rate": 0.0005, "loss": 1.207169771194458, "step": 49023 }, { "epoch": 3.864722112731573, "grad_norm": 0.4809896647930145, "learning_rate": 0.0005, "loss": 1.1275746822357178, "step": 49024 }, { "epoch": 3.8648009459992116, "grad_norm": 0.593672513961792, "learning_rate": 0.0005, "loss": 1.2694876194000244, "step": 49025 }, { "epoch": 3.8648797792668503, "grad_norm": 0.4962514638900757, "learning_rate": 0.0005, "loss": 1.1126859188079834, "step": 49026 }, { "epoch": 3.8649586125344895, "grad_norm": 0.5124097466468811, "learning_rate": 0.0005, "loss": 1.2806708812713623, "step": 49027 }, { "epoch": 3.8650374458021286, "grad_norm": 0.40873733162879944, "learning_rate": 0.0005, "loss": 1.1753242015838623, "step": 49028 }, { "epoch": 3.8651162790697673, "grad_norm": 0.43821895122528076, "learning_rate": 0.0005, "loss": 1.1087257862091064, "step": 49029 }, { "epoch": 3.8651951123374064, "grad_norm": 0.4905940294265747, "learning_rate": 0.0005, "loss": 1.3936653137207031, "step": 49030 }, { "epoch": 3.865273945605045, "grad_norm": 0.4511089622974396, "learning_rate": 0.0005, "loss": 1.038545846939087, "step": 49031 }, { "epoch": 3.8653527788726842, "grad_norm": 0.47408437728881836, "learning_rate": 0.0005, "loss": 1.1658496856689453, "step": 49032 }, { "epoch": 3.8654316121403234, "grad_norm": 0.470483660697937, "learning_rate": 0.0005, "loss": 1.0059008598327637, "step": 49033 }, { "epoch": 3.865510445407962, "grad_norm": 0.4833495616912842, "learning_rate": 0.0005, "loss": 1.0591689348220825, "step": 49034 }, { "epoch": 3.865589278675601, "grad_norm": 0.396707683801651, "learning_rate": 0.0005, "loss": 0.9200644493103027, "step": 49035 }, { "epoch": 3.86566811194324, "grad_norm": 0.4000829756259918, "learning_rate": 0.0005, "loss": 0.9147355556488037, "step": 49036 }, { "epoch": 3.865746945210879, "grad_norm": 0.6213212609291077, "learning_rate": 0.0005, "loss": 1.2570081949234009, "step": 49037 }, { "epoch": 3.865825778478518, "grad_norm": 0.5572183728218079, "learning_rate": 0.0005, "loss": 1.1774399280548096, "step": 49038 }, { "epoch": 3.865904611746157, "grad_norm": 0.40382686257362366, "learning_rate": 0.0005, "loss": 1.0361210107803345, "step": 49039 }, { "epoch": 3.865983445013796, "grad_norm": 0.4426659643650055, "learning_rate": 0.0005, "loss": 1.3704109191894531, "step": 49040 }, { "epoch": 3.8660622782814347, "grad_norm": 0.4762035608291626, "learning_rate": 0.0005, "loss": 1.2668702602386475, "step": 49041 }, { "epoch": 3.866141111549074, "grad_norm": 0.44358327984809875, "learning_rate": 0.0005, "loss": 1.134110450744629, "step": 49042 }, { "epoch": 3.8662199448167125, "grad_norm": 0.3914196789264679, "learning_rate": 0.0005, "loss": 1.0436877012252808, "step": 49043 }, { "epoch": 3.8662987780843516, "grad_norm": 0.4333880543708801, "learning_rate": 0.0005, "loss": 1.0572429895401, "step": 49044 }, { "epoch": 3.8663776113519903, "grad_norm": 0.43899503350257874, "learning_rate": 0.0005, "loss": 1.120603322982788, "step": 49045 }, { "epoch": 3.8664564446196295, "grad_norm": 0.4676665961742401, "learning_rate": 0.0005, "loss": 1.401261806488037, "step": 49046 }, { "epoch": 3.8665352778872686, "grad_norm": 0.42220744490623474, "learning_rate": 0.0005, "loss": 1.0994219779968262, "step": 49047 }, { "epoch": 3.8666141111549073, "grad_norm": 0.4809243381023407, "learning_rate": 0.0005, "loss": 1.1861921548843384, "step": 49048 }, { "epoch": 3.8666929444225464, "grad_norm": 0.46150487661361694, "learning_rate": 0.0005, "loss": 1.3206671476364136, "step": 49049 }, { "epoch": 3.866771777690185, "grad_norm": 0.43936774134635925, "learning_rate": 0.0005, "loss": 1.2111475467681885, "step": 49050 }, { "epoch": 3.8668506109578242, "grad_norm": 0.4355926513671875, "learning_rate": 0.0005, "loss": 1.1638153791427612, "step": 49051 }, { "epoch": 3.8669294442254634, "grad_norm": 0.5315237641334534, "learning_rate": 0.0005, "loss": 1.160048484802246, "step": 49052 }, { "epoch": 3.867008277493102, "grad_norm": 0.4620930552482605, "learning_rate": 0.0005, "loss": 0.9752481579780579, "step": 49053 }, { "epoch": 3.867087110760741, "grad_norm": 0.5641688108444214, "learning_rate": 0.0005, "loss": 1.1844747066497803, "step": 49054 }, { "epoch": 3.86716594402838, "grad_norm": 0.4501539468765259, "learning_rate": 0.0005, "loss": 1.1886565685272217, "step": 49055 }, { "epoch": 3.867244777296019, "grad_norm": 0.5350692272186279, "learning_rate": 0.0005, "loss": 1.1301037073135376, "step": 49056 }, { "epoch": 3.8673236105636577, "grad_norm": 0.6875318884849548, "learning_rate": 0.0005, "loss": 1.2390319108963013, "step": 49057 }, { "epoch": 3.867402443831297, "grad_norm": 0.4962215721607208, "learning_rate": 0.0005, "loss": 1.2695136070251465, "step": 49058 }, { "epoch": 3.8674812770989355, "grad_norm": 0.49549341201782227, "learning_rate": 0.0005, "loss": 1.1684449911117554, "step": 49059 }, { "epoch": 3.8675601103665747, "grad_norm": 0.620424211025238, "learning_rate": 0.0005, "loss": 1.394182562828064, "step": 49060 }, { "epoch": 3.867638943634214, "grad_norm": 0.6934415698051453, "learning_rate": 0.0005, "loss": 1.1675348281860352, "step": 49061 }, { "epoch": 3.8677177769018525, "grad_norm": 0.7021735310554504, "learning_rate": 0.0005, "loss": 1.2799237966537476, "step": 49062 }, { "epoch": 3.8677966101694916, "grad_norm": 0.6844563484191895, "learning_rate": 0.0005, "loss": 1.3647910356521606, "step": 49063 }, { "epoch": 3.8678754434371303, "grad_norm": 0.5872755646705627, "learning_rate": 0.0005, "loss": 1.295969009399414, "step": 49064 }, { "epoch": 3.8679542767047694, "grad_norm": 0.5008273124694824, "learning_rate": 0.0005, "loss": 1.1625628471374512, "step": 49065 }, { "epoch": 3.8680331099724086, "grad_norm": 0.5506746768951416, "learning_rate": 0.0005, "loss": 1.2101466655731201, "step": 49066 }, { "epoch": 3.8681119432400473, "grad_norm": 0.5231062173843384, "learning_rate": 0.0005, "loss": 1.333156943321228, "step": 49067 }, { "epoch": 3.8681907765076864, "grad_norm": 0.5249173045158386, "learning_rate": 0.0005, "loss": 1.1339943408966064, "step": 49068 }, { "epoch": 3.868269609775325, "grad_norm": 0.5077009201049805, "learning_rate": 0.0005, "loss": 1.0708298683166504, "step": 49069 }, { "epoch": 3.868348443042964, "grad_norm": 0.6016553044319153, "learning_rate": 0.0005, "loss": 1.0589540004730225, "step": 49070 }, { "epoch": 3.8684272763106033, "grad_norm": 0.7162766456604004, "learning_rate": 0.0005, "loss": 1.330744981765747, "step": 49071 }, { "epoch": 3.868506109578242, "grad_norm": 0.46769872307777405, "learning_rate": 0.0005, "loss": 1.1574993133544922, "step": 49072 }, { "epoch": 3.8685849428458807, "grad_norm": 0.4885071814060211, "learning_rate": 0.0005, "loss": 1.2215925455093384, "step": 49073 }, { "epoch": 3.86866377611352, "grad_norm": 0.7063132524490356, "learning_rate": 0.0005, "loss": 1.3050148487091064, "step": 49074 }, { "epoch": 3.868742609381159, "grad_norm": 0.4571639895439148, "learning_rate": 0.0005, "loss": 1.1323504447937012, "step": 49075 }, { "epoch": 3.8688214426487977, "grad_norm": 0.4563203752040863, "learning_rate": 0.0005, "loss": 1.1678162813186646, "step": 49076 }, { "epoch": 3.868900275916437, "grad_norm": 0.38996806740760803, "learning_rate": 0.0005, "loss": 1.0190131664276123, "step": 49077 }, { "epoch": 3.8689791091840755, "grad_norm": 0.5398305654525757, "learning_rate": 0.0005, "loss": 1.422743797302246, "step": 49078 }, { "epoch": 3.8690579424517146, "grad_norm": 0.4503932595252991, "learning_rate": 0.0005, "loss": 1.2565749883651733, "step": 49079 }, { "epoch": 3.8691367757193538, "grad_norm": 0.4100567400455475, "learning_rate": 0.0005, "loss": 1.1484060287475586, "step": 49080 }, { "epoch": 3.8692156089869925, "grad_norm": 0.43038442730903625, "learning_rate": 0.0005, "loss": 1.218080759048462, "step": 49081 }, { "epoch": 3.8692944422546316, "grad_norm": 0.43712660670280457, "learning_rate": 0.0005, "loss": 1.3205599784851074, "step": 49082 }, { "epoch": 3.8693732755222703, "grad_norm": 0.4917515516281128, "learning_rate": 0.0005, "loss": 1.2223730087280273, "step": 49083 }, { "epoch": 3.8694521087899094, "grad_norm": 1.0883584022521973, "learning_rate": 0.0005, "loss": 1.3067067861557007, "step": 49084 }, { "epoch": 3.8695309420575486, "grad_norm": 0.5005645751953125, "learning_rate": 0.0005, "loss": 1.1962087154388428, "step": 49085 }, { "epoch": 3.8696097753251872, "grad_norm": 0.5436649322509766, "learning_rate": 0.0005, "loss": 1.364806890487671, "step": 49086 }, { "epoch": 3.869688608592826, "grad_norm": 0.6468459367752075, "learning_rate": 0.0005, "loss": 1.1578118801116943, "step": 49087 }, { "epoch": 3.869767441860465, "grad_norm": 0.4638352692127228, "learning_rate": 0.0005, "loss": 1.2026878595352173, "step": 49088 }, { "epoch": 3.869846275128104, "grad_norm": 0.4807175397872925, "learning_rate": 0.0005, "loss": 1.1529676914215088, "step": 49089 }, { "epoch": 3.869925108395743, "grad_norm": 0.5529327988624573, "learning_rate": 0.0005, "loss": 1.1019065380096436, "step": 49090 }, { "epoch": 3.870003941663382, "grad_norm": 0.6094685792922974, "learning_rate": 0.0005, "loss": 1.1870769262313843, "step": 49091 }, { "epoch": 3.8700827749310207, "grad_norm": 0.6371666193008423, "learning_rate": 0.0005, "loss": 1.402640461921692, "step": 49092 }, { "epoch": 3.87016160819866, "grad_norm": 0.6415761709213257, "learning_rate": 0.0005, "loss": 1.2116234302520752, "step": 49093 }, { "epoch": 3.870240441466299, "grad_norm": 0.511803388595581, "learning_rate": 0.0005, "loss": 1.1674295663833618, "step": 49094 }, { "epoch": 3.8703192747339377, "grad_norm": 0.7049905061721802, "learning_rate": 0.0005, "loss": 1.0884106159210205, "step": 49095 }, { "epoch": 3.870398108001577, "grad_norm": 0.4407569468021393, "learning_rate": 0.0005, "loss": 1.1768629550933838, "step": 49096 }, { "epoch": 3.8704769412692155, "grad_norm": 0.4818969666957855, "learning_rate": 0.0005, "loss": 1.1992747783660889, "step": 49097 }, { "epoch": 3.8705557745368546, "grad_norm": 0.5309127569198608, "learning_rate": 0.0005, "loss": 1.2084537744522095, "step": 49098 }, { "epoch": 3.8706346078044938, "grad_norm": 0.5514340400695801, "learning_rate": 0.0005, "loss": 1.258065938949585, "step": 49099 }, { "epoch": 3.8707134410721324, "grad_norm": 0.4173037111759186, "learning_rate": 0.0005, "loss": 0.9392254948616028, "step": 49100 }, { "epoch": 3.870792274339771, "grad_norm": 0.48932063579559326, "learning_rate": 0.0005, "loss": 1.128188133239746, "step": 49101 }, { "epoch": 3.8708711076074103, "grad_norm": 0.42744311690330505, "learning_rate": 0.0005, "loss": 1.2619287967681885, "step": 49102 }, { "epoch": 3.8709499408750494, "grad_norm": 0.4599105715751648, "learning_rate": 0.0005, "loss": 1.1376889944076538, "step": 49103 }, { "epoch": 3.871028774142688, "grad_norm": 0.47966188192367554, "learning_rate": 0.0005, "loss": 1.2134630680084229, "step": 49104 }, { "epoch": 3.8711076074103272, "grad_norm": 0.4444736838340759, "learning_rate": 0.0005, "loss": 1.1865071058273315, "step": 49105 }, { "epoch": 3.871186440677966, "grad_norm": 0.4905194640159607, "learning_rate": 0.0005, "loss": 1.1908347606658936, "step": 49106 }, { "epoch": 3.871265273945605, "grad_norm": 0.5653639435768127, "learning_rate": 0.0005, "loss": 1.1328599452972412, "step": 49107 }, { "epoch": 3.871344107213244, "grad_norm": 0.4480845630168915, "learning_rate": 0.0005, "loss": 1.22825288772583, "step": 49108 }, { "epoch": 3.871422940480883, "grad_norm": 0.484454482793808, "learning_rate": 0.0005, "loss": 1.1882628202438354, "step": 49109 }, { "epoch": 3.871501773748522, "grad_norm": 0.4731118679046631, "learning_rate": 0.0005, "loss": 1.1813406944274902, "step": 49110 }, { "epoch": 3.8715806070161607, "grad_norm": 0.5251443982124329, "learning_rate": 0.0005, "loss": 1.2767820358276367, "step": 49111 }, { "epoch": 3.8716594402838, "grad_norm": 0.4926331341266632, "learning_rate": 0.0005, "loss": 1.2428470849990845, "step": 49112 }, { "epoch": 3.871738273551439, "grad_norm": 0.449374794960022, "learning_rate": 0.0005, "loss": 1.1915299892425537, "step": 49113 }, { "epoch": 3.8718171068190776, "grad_norm": 0.46921786665916443, "learning_rate": 0.0005, "loss": 1.1749317646026611, "step": 49114 }, { "epoch": 3.8718959400867163, "grad_norm": 0.5218474864959717, "learning_rate": 0.0005, "loss": 1.2928709983825684, "step": 49115 }, { "epoch": 3.8719747733543555, "grad_norm": 0.47719353437423706, "learning_rate": 0.0005, "loss": 1.1482573747634888, "step": 49116 }, { "epoch": 3.8720536066219946, "grad_norm": 0.45156827569007874, "learning_rate": 0.0005, "loss": 1.112246036529541, "step": 49117 }, { "epoch": 3.8721324398896333, "grad_norm": 0.4912039637565613, "learning_rate": 0.0005, "loss": 1.1754823923110962, "step": 49118 }, { "epoch": 3.8722112731572724, "grad_norm": 0.6100274920463562, "learning_rate": 0.0005, "loss": 1.2197778224945068, "step": 49119 }, { "epoch": 3.872290106424911, "grad_norm": 0.3980157971382141, "learning_rate": 0.0005, "loss": 0.9652126431465149, "step": 49120 }, { "epoch": 3.8723689396925502, "grad_norm": 0.6014973521232605, "learning_rate": 0.0005, "loss": 1.126695156097412, "step": 49121 }, { "epoch": 3.8724477729601894, "grad_norm": 0.5267083644866943, "learning_rate": 0.0005, "loss": 1.3824087381362915, "step": 49122 }, { "epoch": 3.872526606227828, "grad_norm": 0.5332196354866028, "learning_rate": 0.0005, "loss": 1.1993950605392456, "step": 49123 }, { "epoch": 3.872605439495467, "grad_norm": 0.4590006470680237, "learning_rate": 0.0005, "loss": 1.1274524927139282, "step": 49124 }, { "epoch": 3.872684272763106, "grad_norm": 0.648785412311554, "learning_rate": 0.0005, "loss": 1.0723793506622314, "step": 49125 }, { "epoch": 3.872763106030745, "grad_norm": 0.5740257501602173, "learning_rate": 0.0005, "loss": 1.2324551343917847, "step": 49126 }, { "epoch": 3.872841939298384, "grad_norm": 0.44098663330078125, "learning_rate": 0.0005, "loss": 1.1522938013076782, "step": 49127 }, { "epoch": 3.872920772566023, "grad_norm": 0.5033954977989197, "learning_rate": 0.0005, "loss": 1.1834520101547241, "step": 49128 }, { "epoch": 3.8729996058336615, "grad_norm": 0.4619589149951935, "learning_rate": 0.0005, "loss": 1.2808289527893066, "step": 49129 }, { "epoch": 3.8730784391013007, "grad_norm": 0.6819103360176086, "learning_rate": 0.0005, "loss": 1.4415068626403809, "step": 49130 }, { "epoch": 3.87315727236894, "grad_norm": 0.5451444983482361, "learning_rate": 0.0005, "loss": 1.199291706085205, "step": 49131 }, { "epoch": 3.8732361056365785, "grad_norm": 0.4416132867336273, "learning_rate": 0.0005, "loss": 1.2855708599090576, "step": 49132 }, { "epoch": 3.8733149389042176, "grad_norm": 0.49218377470970154, "learning_rate": 0.0005, "loss": 1.175544023513794, "step": 49133 }, { "epoch": 3.8733937721718563, "grad_norm": 0.5292564034461975, "learning_rate": 0.0005, "loss": 1.2615529298782349, "step": 49134 }, { "epoch": 3.8734726054394955, "grad_norm": 0.48665523529052734, "learning_rate": 0.0005, "loss": 1.167581558227539, "step": 49135 }, { "epoch": 3.8735514387071346, "grad_norm": 0.47005781531333923, "learning_rate": 0.0005, "loss": 1.2282962799072266, "step": 49136 }, { "epoch": 3.8736302719747733, "grad_norm": 0.525200366973877, "learning_rate": 0.0005, "loss": 1.3198142051696777, "step": 49137 }, { "epoch": 3.8737091052424124, "grad_norm": 0.5158941149711609, "learning_rate": 0.0005, "loss": 1.122539758682251, "step": 49138 }, { "epoch": 3.873787938510051, "grad_norm": 0.4969651997089386, "learning_rate": 0.0005, "loss": 1.3897813558578491, "step": 49139 }, { "epoch": 3.8738667717776902, "grad_norm": 0.4252444803714752, "learning_rate": 0.0005, "loss": 1.0219991207122803, "step": 49140 }, { "epoch": 3.8739456050453294, "grad_norm": 0.46985921263694763, "learning_rate": 0.0005, "loss": 0.9546250700950623, "step": 49141 }, { "epoch": 3.874024438312968, "grad_norm": 0.5444537401199341, "learning_rate": 0.0005, "loss": 1.3379452228546143, "step": 49142 }, { "epoch": 3.8741032715806067, "grad_norm": 0.49757614731788635, "learning_rate": 0.0005, "loss": 1.1514527797698975, "step": 49143 }, { "epoch": 3.874182104848246, "grad_norm": 0.4726252257823944, "learning_rate": 0.0005, "loss": 1.207849144935608, "step": 49144 }, { "epoch": 3.874260938115885, "grad_norm": 0.4710249602794647, "learning_rate": 0.0005, "loss": 1.1916122436523438, "step": 49145 }, { "epoch": 3.8743397713835237, "grad_norm": 0.5011462569236755, "learning_rate": 0.0005, "loss": 1.3156726360321045, "step": 49146 }, { "epoch": 3.874418604651163, "grad_norm": 1.3751546144485474, "learning_rate": 0.0005, "loss": 1.408381462097168, "step": 49147 }, { "epoch": 3.8744974379188015, "grad_norm": 0.5738447904586792, "learning_rate": 0.0005, "loss": 1.198456048965454, "step": 49148 }, { "epoch": 3.8745762711864407, "grad_norm": 0.7452558875083923, "learning_rate": 0.0005, "loss": 1.090287685394287, "step": 49149 }, { "epoch": 3.87465510445408, "grad_norm": 0.470369815826416, "learning_rate": 0.0005, "loss": 0.9322616457939148, "step": 49150 }, { "epoch": 3.8747339377217185, "grad_norm": 0.6532109379768372, "learning_rate": 0.0005, "loss": 1.1332131624221802, "step": 49151 }, { "epoch": 3.8748127709893576, "grad_norm": 0.6150819063186646, "learning_rate": 0.0005, "loss": 1.4743331670761108, "step": 49152 }, { "epoch": 3.8748916042569963, "grad_norm": 0.45942649245262146, "learning_rate": 0.0005, "loss": 1.1394010782241821, "step": 49153 }, { "epoch": 3.8749704375246354, "grad_norm": 0.5954152345657349, "learning_rate": 0.0005, "loss": 1.2963190078735352, "step": 49154 }, { "epoch": 3.8750492707922746, "grad_norm": 0.46055907011032104, "learning_rate": 0.0005, "loss": 1.194716453552246, "step": 49155 }, { "epoch": 3.8751281040599133, "grad_norm": 0.48512548208236694, "learning_rate": 0.0005, "loss": 1.200141429901123, "step": 49156 }, { "epoch": 3.8752069373275524, "grad_norm": 0.4205940067768097, "learning_rate": 0.0005, "loss": 1.1028988361358643, "step": 49157 }, { "epoch": 3.875285770595191, "grad_norm": 0.5187051296234131, "learning_rate": 0.0005, "loss": 1.3223862648010254, "step": 49158 }, { "epoch": 3.87536460386283, "grad_norm": 0.5887518525123596, "learning_rate": 0.0005, "loss": 1.2160730361938477, "step": 49159 }, { "epoch": 3.875443437130469, "grad_norm": 0.47248977422714233, "learning_rate": 0.0005, "loss": 1.0491389036178589, "step": 49160 }, { "epoch": 3.875522270398108, "grad_norm": 0.47601521015167236, "learning_rate": 0.0005, "loss": 1.2097835540771484, "step": 49161 }, { "epoch": 3.8756011036657467, "grad_norm": 0.443164199590683, "learning_rate": 0.0005, "loss": 1.113044261932373, "step": 49162 }, { "epoch": 3.875679936933386, "grad_norm": 0.5604333281517029, "learning_rate": 0.0005, "loss": 1.176777958869934, "step": 49163 }, { "epoch": 3.875758770201025, "grad_norm": 0.559425413608551, "learning_rate": 0.0005, "loss": 1.237046241760254, "step": 49164 }, { "epoch": 3.8758376034686637, "grad_norm": 0.4162238836288452, "learning_rate": 0.0005, "loss": 1.036219835281372, "step": 49165 }, { "epoch": 3.875916436736303, "grad_norm": 0.455840140581131, "learning_rate": 0.0005, "loss": 1.037286400794983, "step": 49166 }, { "epoch": 3.8759952700039415, "grad_norm": 0.5317772030830383, "learning_rate": 0.0005, "loss": 1.2611920833587646, "step": 49167 }, { "epoch": 3.8760741032715806, "grad_norm": 0.46740490198135376, "learning_rate": 0.0005, "loss": 1.0153133869171143, "step": 49168 }, { "epoch": 3.8761529365392198, "grad_norm": 0.48630693554878235, "learning_rate": 0.0005, "loss": 1.16740882396698, "step": 49169 }, { "epoch": 3.8762317698068585, "grad_norm": 0.47382357716560364, "learning_rate": 0.0005, "loss": 1.360897183418274, "step": 49170 }, { "epoch": 3.8763106030744976, "grad_norm": 0.6110496520996094, "learning_rate": 0.0005, "loss": 1.3252921104431152, "step": 49171 }, { "epoch": 3.8763894363421363, "grad_norm": 0.4832821190357208, "learning_rate": 0.0005, "loss": 1.210655689239502, "step": 49172 }, { "epoch": 3.8764682696097754, "grad_norm": 0.5105940699577332, "learning_rate": 0.0005, "loss": 1.1618143320083618, "step": 49173 }, { "epoch": 3.876547102877414, "grad_norm": 0.4445106089115143, "learning_rate": 0.0005, "loss": 1.0823590755462646, "step": 49174 }, { "epoch": 3.8766259361450532, "grad_norm": 0.4794858992099762, "learning_rate": 0.0005, "loss": 1.1113100051879883, "step": 49175 }, { "epoch": 3.876704769412692, "grad_norm": 0.3707466125488281, "learning_rate": 0.0005, "loss": 0.9613898396492004, "step": 49176 }, { "epoch": 3.876783602680331, "grad_norm": 0.6875566840171814, "learning_rate": 0.0005, "loss": 1.2167760133743286, "step": 49177 }, { "epoch": 3.87686243594797, "grad_norm": 0.5109007358551025, "learning_rate": 0.0005, "loss": 1.1515141725540161, "step": 49178 }, { "epoch": 3.876941269215609, "grad_norm": 0.5258751511573792, "learning_rate": 0.0005, "loss": 1.1947710514068604, "step": 49179 }, { "epoch": 3.877020102483248, "grad_norm": 0.48549771308898926, "learning_rate": 0.0005, "loss": 1.1530755758285522, "step": 49180 }, { "epoch": 3.8770989357508867, "grad_norm": 0.41464415192604065, "learning_rate": 0.0005, "loss": 1.1665512323379517, "step": 49181 }, { "epoch": 3.877177769018526, "grad_norm": 0.5547399520874023, "learning_rate": 0.0005, "loss": 1.1571226119995117, "step": 49182 }, { "epoch": 3.877256602286165, "grad_norm": 0.6489788293838501, "learning_rate": 0.0005, "loss": 1.2708568572998047, "step": 49183 }, { "epoch": 3.8773354355538037, "grad_norm": 0.48417967557907104, "learning_rate": 0.0005, "loss": 1.2974042892456055, "step": 49184 }, { "epoch": 3.877414268821443, "grad_norm": 0.5685430765151978, "learning_rate": 0.0005, "loss": 1.1407060623168945, "step": 49185 }, { "epoch": 3.8774931020890815, "grad_norm": 0.38328883051872253, "learning_rate": 0.0005, "loss": 1.0110183954238892, "step": 49186 }, { "epoch": 3.8775719353567206, "grad_norm": 0.5303494930267334, "learning_rate": 0.0005, "loss": 1.2663300037384033, "step": 49187 }, { "epoch": 3.8776507686243598, "grad_norm": 0.7369206547737122, "learning_rate": 0.0005, "loss": 1.2960972785949707, "step": 49188 }, { "epoch": 3.8777296018919984, "grad_norm": 0.5097728967666626, "learning_rate": 0.0005, "loss": 1.2399483919143677, "step": 49189 }, { "epoch": 3.877808435159637, "grad_norm": 0.48565566539764404, "learning_rate": 0.0005, "loss": 1.0542521476745605, "step": 49190 }, { "epoch": 3.8778872684272763, "grad_norm": 0.4156557619571686, "learning_rate": 0.0005, "loss": 1.0503509044647217, "step": 49191 }, { "epoch": 3.8779661016949154, "grad_norm": 0.5430694818496704, "learning_rate": 0.0005, "loss": 1.2572760581970215, "step": 49192 }, { "epoch": 3.878044934962554, "grad_norm": 0.5250149965286255, "learning_rate": 0.0005, "loss": 1.352095127105713, "step": 49193 }, { "epoch": 3.8781237682301932, "grad_norm": 0.738828182220459, "learning_rate": 0.0005, "loss": 1.2101202011108398, "step": 49194 }, { "epoch": 3.878202601497832, "grad_norm": 0.6433080434799194, "learning_rate": 0.0005, "loss": 1.3713388442993164, "step": 49195 }, { "epoch": 3.878281434765471, "grad_norm": 0.6548343300819397, "learning_rate": 0.0005, "loss": 1.4089252948760986, "step": 49196 }, { "epoch": 3.87836026803311, "grad_norm": 0.46887630224227905, "learning_rate": 0.0005, "loss": 1.1024905443191528, "step": 49197 }, { "epoch": 3.878439101300749, "grad_norm": 0.6088200807571411, "learning_rate": 0.0005, "loss": 1.3485642671585083, "step": 49198 }, { "epoch": 3.878517934568388, "grad_norm": 0.5451374650001526, "learning_rate": 0.0005, "loss": 1.2027430534362793, "step": 49199 }, { "epoch": 3.8785967678360267, "grad_norm": 0.5372198224067688, "learning_rate": 0.0005, "loss": 1.1927820444107056, "step": 49200 }, { "epoch": 3.878675601103666, "grad_norm": 0.4766033887863159, "learning_rate": 0.0005, "loss": 1.2155205011367798, "step": 49201 }, { "epoch": 3.878754434371305, "grad_norm": 0.6550123691558838, "learning_rate": 0.0005, "loss": 1.1572061777114868, "step": 49202 }, { "epoch": 3.8788332676389436, "grad_norm": 0.4716870188713074, "learning_rate": 0.0005, "loss": 1.0692694187164307, "step": 49203 }, { "epoch": 3.8789121009065823, "grad_norm": 0.4018840491771698, "learning_rate": 0.0005, "loss": 1.173501968383789, "step": 49204 }, { "epoch": 3.8789909341742215, "grad_norm": 0.4867672324180603, "learning_rate": 0.0005, "loss": 1.2219446897506714, "step": 49205 }, { "epoch": 3.8790697674418606, "grad_norm": 0.4362412393093109, "learning_rate": 0.0005, "loss": 1.1901901960372925, "step": 49206 }, { "epoch": 3.8791486007094993, "grad_norm": 0.6135614514350891, "learning_rate": 0.0005, "loss": 1.4212164878845215, "step": 49207 }, { "epoch": 3.8792274339771384, "grad_norm": 0.5005155801773071, "learning_rate": 0.0005, "loss": 1.1609270572662354, "step": 49208 }, { "epoch": 3.879306267244777, "grad_norm": 0.42473188042640686, "learning_rate": 0.0005, "loss": 1.1690136194229126, "step": 49209 }, { "epoch": 3.8793851005124163, "grad_norm": 0.8446664214134216, "learning_rate": 0.0005, "loss": 1.3274563550949097, "step": 49210 }, { "epoch": 3.8794639337800554, "grad_norm": 0.45166102051734924, "learning_rate": 0.0005, "loss": 1.0861494541168213, "step": 49211 }, { "epoch": 3.879542767047694, "grad_norm": 0.5114765167236328, "learning_rate": 0.0005, "loss": 1.2299140691757202, "step": 49212 }, { "epoch": 3.879621600315333, "grad_norm": 0.8516186475753784, "learning_rate": 0.0005, "loss": 1.241959810256958, "step": 49213 }, { "epoch": 3.879700433582972, "grad_norm": 0.5219546556472778, "learning_rate": 0.0005, "loss": 1.3098413944244385, "step": 49214 }, { "epoch": 3.879779266850611, "grad_norm": 0.46293044090270996, "learning_rate": 0.0005, "loss": 1.2088634967803955, "step": 49215 }, { "epoch": 3.87985810011825, "grad_norm": 0.47788941860198975, "learning_rate": 0.0005, "loss": 1.264362096786499, "step": 49216 }, { "epoch": 3.879936933385889, "grad_norm": 0.5023148655891418, "learning_rate": 0.0005, "loss": 1.301296591758728, "step": 49217 }, { "epoch": 3.8800157666535275, "grad_norm": 0.7567309141159058, "learning_rate": 0.0005, "loss": 1.4778666496276855, "step": 49218 }, { "epoch": 3.8800945999211667, "grad_norm": 0.6055938601493835, "learning_rate": 0.0005, "loss": 1.0623037815093994, "step": 49219 }, { "epoch": 3.880173433188806, "grad_norm": 0.46094733476638794, "learning_rate": 0.0005, "loss": 1.076164722442627, "step": 49220 }, { "epoch": 3.8802522664564445, "grad_norm": 0.4651840329170227, "learning_rate": 0.0005, "loss": 1.0946316719055176, "step": 49221 }, { "epoch": 3.8803310997240836, "grad_norm": 0.541435956954956, "learning_rate": 0.0005, "loss": 1.153059959411621, "step": 49222 }, { "epoch": 3.8804099329917223, "grad_norm": 0.5724714398384094, "learning_rate": 0.0005, "loss": 1.38277268409729, "step": 49223 }, { "epoch": 3.8804887662593615, "grad_norm": 0.5746411681175232, "learning_rate": 0.0005, "loss": 1.3150229454040527, "step": 49224 }, { "epoch": 3.8805675995270006, "grad_norm": 0.5698065161705017, "learning_rate": 0.0005, "loss": 1.272042989730835, "step": 49225 }, { "epoch": 3.8806464327946393, "grad_norm": 0.5246397256851196, "learning_rate": 0.0005, "loss": 1.298730731010437, "step": 49226 }, { "epoch": 3.8807252660622784, "grad_norm": 0.6443597674369812, "learning_rate": 0.0005, "loss": 1.2276501655578613, "step": 49227 }, { "epoch": 3.880804099329917, "grad_norm": 0.42957913875579834, "learning_rate": 0.0005, "loss": 0.9030238389968872, "step": 49228 }, { "epoch": 3.8808829325975562, "grad_norm": 0.39795494079589844, "learning_rate": 0.0005, "loss": 1.1390504837036133, "step": 49229 }, { "epoch": 3.8809617658651954, "grad_norm": 0.4808584153652191, "learning_rate": 0.0005, "loss": 1.2436740398406982, "step": 49230 }, { "epoch": 3.881040599132834, "grad_norm": 0.5288257598876953, "learning_rate": 0.0005, "loss": 1.1506154537200928, "step": 49231 }, { "epoch": 3.8811194324004727, "grad_norm": 0.5358414649963379, "learning_rate": 0.0005, "loss": 1.1639771461486816, "step": 49232 }, { "epoch": 3.881198265668112, "grad_norm": 0.4781700670719147, "learning_rate": 0.0005, "loss": 1.1914395093917847, "step": 49233 }, { "epoch": 3.881277098935751, "grad_norm": 0.5033531188964844, "learning_rate": 0.0005, "loss": 1.238973617553711, "step": 49234 }, { "epoch": 3.8813559322033897, "grad_norm": 0.5484330058097839, "learning_rate": 0.0005, "loss": 1.2273316383361816, "step": 49235 }, { "epoch": 3.881434765471029, "grad_norm": 0.5128229856491089, "learning_rate": 0.0005, "loss": 1.3170900344848633, "step": 49236 }, { "epoch": 3.8815135987386675, "grad_norm": 0.45312950015068054, "learning_rate": 0.0005, "loss": 1.015452265739441, "step": 49237 }, { "epoch": 3.8815924320063067, "grad_norm": 0.5121158957481384, "learning_rate": 0.0005, "loss": 1.1970235109329224, "step": 49238 }, { "epoch": 3.881671265273946, "grad_norm": 1.0159366130828857, "learning_rate": 0.0005, "loss": 1.0827131271362305, "step": 49239 }, { "epoch": 3.8817500985415845, "grad_norm": 0.5029914975166321, "learning_rate": 0.0005, "loss": 1.2837624549865723, "step": 49240 }, { "epoch": 3.8818289318092236, "grad_norm": 0.49667149782180786, "learning_rate": 0.0005, "loss": 1.2754237651824951, "step": 49241 }, { "epoch": 3.8819077650768623, "grad_norm": 0.49415358901023865, "learning_rate": 0.0005, "loss": 1.1992666721343994, "step": 49242 }, { "epoch": 3.8819865983445014, "grad_norm": 0.5334591269493103, "learning_rate": 0.0005, "loss": 1.202364206314087, "step": 49243 }, { "epoch": 3.8820654316121406, "grad_norm": 0.5051606297492981, "learning_rate": 0.0005, "loss": 1.1310410499572754, "step": 49244 }, { "epoch": 3.8821442648797793, "grad_norm": 0.4392324686050415, "learning_rate": 0.0005, "loss": 1.0098364353179932, "step": 49245 }, { "epoch": 3.882223098147418, "grad_norm": 0.4552743136882782, "learning_rate": 0.0005, "loss": 1.1766433715820312, "step": 49246 }, { "epoch": 3.882301931415057, "grad_norm": 0.5184508562088013, "learning_rate": 0.0005, "loss": 1.229875087738037, "step": 49247 }, { "epoch": 3.882380764682696, "grad_norm": 0.3557051718235016, "learning_rate": 0.0005, "loss": 0.9910604953765869, "step": 49248 }, { "epoch": 3.882459597950335, "grad_norm": 0.4167317748069763, "learning_rate": 0.0005, "loss": 0.947106122970581, "step": 49249 }, { "epoch": 3.882538431217974, "grad_norm": 0.4794887602329254, "learning_rate": 0.0005, "loss": 1.0488836765289307, "step": 49250 }, { "epoch": 3.8826172644856127, "grad_norm": 0.4239712953567505, "learning_rate": 0.0005, "loss": 1.1428064107894897, "step": 49251 }, { "epoch": 3.882696097753252, "grad_norm": 0.42250242829322815, "learning_rate": 0.0005, "loss": 0.9642443060874939, "step": 49252 }, { "epoch": 3.882774931020891, "grad_norm": 0.4326145350933075, "learning_rate": 0.0005, "loss": 1.1216206550598145, "step": 49253 }, { "epoch": 3.8828537642885297, "grad_norm": 0.5813809633255005, "learning_rate": 0.0005, "loss": 1.1966239213943481, "step": 49254 }, { "epoch": 3.882932597556169, "grad_norm": 0.5360020995140076, "learning_rate": 0.0005, "loss": 1.2734839916229248, "step": 49255 }, { "epoch": 3.8830114308238075, "grad_norm": 0.39943838119506836, "learning_rate": 0.0005, "loss": 1.0609022378921509, "step": 49256 }, { "epoch": 3.8830902640914466, "grad_norm": 0.4718134105205536, "learning_rate": 0.0005, "loss": 1.1807129383087158, "step": 49257 }, { "epoch": 3.8831690973590858, "grad_norm": 0.5590074062347412, "learning_rate": 0.0005, "loss": 1.1457594633102417, "step": 49258 }, { "epoch": 3.8832479306267245, "grad_norm": 0.4892580807209015, "learning_rate": 0.0005, "loss": 1.2974469661712646, "step": 49259 }, { "epoch": 3.883326763894363, "grad_norm": 0.5403457880020142, "learning_rate": 0.0005, "loss": 1.1330474615097046, "step": 49260 }, { "epoch": 3.8834055971620023, "grad_norm": 0.4531523585319519, "learning_rate": 0.0005, "loss": 1.080918312072754, "step": 49261 }, { "epoch": 3.8834844304296414, "grad_norm": 0.4450799226760864, "learning_rate": 0.0005, "loss": 0.9707087278366089, "step": 49262 }, { "epoch": 3.88356326369728, "grad_norm": 0.7463597059249878, "learning_rate": 0.0005, "loss": 1.1185188293457031, "step": 49263 }, { "epoch": 3.8836420969649192, "grad_norm": 0.4684482514858246, "learning_rate": 0.0005, "loss": 1.163725733757019, "step": 49264 }, { "epoch": 3.883720930232558, "grad_norm": 0.56647127866745, "learning_rate": 0.0005, "loss": 1.3239471912384033, "step": 49265 }, { "epoch": 3.883799763500197, "grad_norm": 0.5550776720046997, "learning_rate": 0.0005, "loss": 1.0644540786743164, "step": 49266 }, { "epoch": 3.883878596767836, "grad_norm": 0.5277795195579529, "learning_rate": 0.0005, "loss": 1.316292405128479, "step": 49267 }, { "epoch": 3.883957430035475, "grad_norm": 0.491155743598938, "learning_rate": 0.0005, "loss": 1.2001161575317383, "step": 49268 }, { "epoch": 3.884036263303114, "grad_norm": 0.407974511384964, "learning_rate": 0.0005, "loss": 1.1928348541259766, "step": 49269 }, { "epoch": 3.8841150965707527, "grad_norm": 0.402982234954834, "learning_rate": 0.0005, "loss": 1.0243300199508667, "step": 49270 }, { "epoch": 3.884193929838392, "grad_norm": 0.4242526590824127, "learning_rate": 0.0005, "loss": 1.0920307636260986, "step": 49271 }, { "epoch": 3.884272763106031, "grad_norm": 0.4827366769313812, "learning_rate": 0.0005, "loss": 1.1335362195968628, "step": 49272 }, { "epoch": 3.8843515963736697, "grad_norm": 0.5299469828605652, "learning_rate": 0.0005, "loss": 1.3613015413284302, "step": 49273 }, { "epoch": 3.8844304296413084, "grad_norm": 0.5556995868682861, "learning_rate": 0.0005, "loss": 1.094852328300476, "step": 49274 }, { "epoch": 3.8845092629089475, "grad_norm": 0.4181642234325409, "learning_rate": 0.0005, "loss": 1.1870336532592773, "step": 49275 }, { "epoch": 3.8845880961765866, "grad_norm": 0.4736802577972412, "learning_rate": 0.0005, "loss": 1.3266527652740479, "step": 49276 }, { "epoch": 3.8846669294442253, "grad_norm": 0.6778950691223145, "learning_rate": 0.0005, "loss": 1.3363475799560547, "step": 49277 }, { "epoch": 3.8847457627118644, "grad_norm": 0.5070237517356873, "learning_rate": 0.0005, "loss": 1.0976786613464355, "step": 49278 }, { "epoch": 3.884824595979503, "grad_norm": 0.5400218367576599, "learning_rate": 0.0005, "loss": 1.2263484001159668, "step": 49279 }, { "epoch": 3.8849034292471423, "grad_norm": 0.4814684987068176, "learning_rate": 0.0005, "loss": 1.173636555671692, "step": 49280 }, { "epoch": 3.8849822625147814, "grad_norm": 0.5252808928489685, "learning_rate": 0.0005, "loss": 1.086832046508789, "step": 49281 }, { "epoch": 3.88506109578242, "grad_norm": 0.492112934589386, "learning_rate": 0.0005, "loss": 1.107235312461853, "step": 49282 }, { "epoch": 3.8851399290500592, "grad_norm": 0.5627098679542542, "learning_rate": 0.0005, "loss": 1.399404525756836, "step": 49283 }, { "epoch": 3.885218762317698, "grad_norm": 0.560072660446167, "learning_rate": 0.0005, "loss": 1.3116307258605957, "step": 49284 }, { "epoch": 3.885297595585337, "grad_norm": 0.42960697412490845, "learning_rate": 0.0005, "loss": 1.176609992980957, "step": 49285 }, { "epoch": 3.885376428852976, "grad_norm": 0.6048541069030762, "learning_rate": 0.0005, "loss": 1.3410837650299072, "step": 49286 }, { "epoch": 3.885455262120615, "grad_norm": 0.41729336977005005, "learning_rate": 0.0005, "loss": 1.0096583366394043, "step": 49287 }, { "epoch": 3.885534095388254, "grad_norm": 0.48655247688293457, "learning_rate": 0.0005, "loss": 1.1060442924499512, "step": 49288 }, { "epoch": 3.8856129286558927, "grad_norm": 0.642717182636261, "learning_rate": 0.0005, "loss": 1.2408519983291626, "step": 49289 }, { "epoch": 3.885691761923532, "grad_norm": 0.4138238728046417, "learning_rate": 0.0005, "loss": 1.0969561338424683, "step": 49290 }, { "epoch": 3.8857705951911705, "grad_norm": 0.5020610690116882, "learning_rate": 0.0005, "loss": 0.9943819046020508, "step": 49291 }, { "epoch": 3.8858494284588097, "grad_norm": 0.38970446586608887, "learning_rate": 0.0005, "loss": 1.119651198387146, "step": 49292 }, { "epoch": 3.8859282617264483, "grad_norm": 0.47630271315574646, "learning_rate": 0.0005, "loss": 1.0262888669967651, "step": 49293 }, { "epoch": 3.8860070949940875, "grad_norm": 0.5255093574523926, "learning_rate": 0.0005, "loss": 1.2355122566223145, "step": 49294 }, { "epoch": 3.8860859282617266, "grad_norm": 0.3927256762981415, "learning_rate": 0.0005, "loss": 1.0821846723556519, "step": 49295 }, { "epoch": 3.8861647615293653, "grad_norm": 0.4612007737159729, "learning_rate": 0.0005, "loss": 1.146199107170105, "step": 49296 }, { "epoch": 3.8862435947970044, "grad_norm": 0.49653542041778564, "learning_rate": 0.0005, "loss": 1.1099646091461182, "step": 49297 }, { "epoch": 3.886322428064643, "grad_norm": 0.5041859745979309, "learning_rate": 0.0005, "loss": 1.1682510375976562, "step": 49298 }, { "epoch": 3.8864012613322823, "grad_norm": 0.5541287660598755, "learning_rate": 0.0005, "loss": 1.0262813568115234, "step": 49299 }, { "epoch": 3.8864800945999214, "grad_norm": 0.4396231472492218, "learning_rate": 0.0005, "loss": 1.14828360080719, "step": 49300 }, { "epoch": 3.88655892786756, "grad_norm": 0.434792697429657, "learning_rate": 0.0005, "loss": 1.0820550918579102, "step": 49301 }, { "epoch": 3.886637761135199, "grad_norm": 0.4567432701587677, "learning_rate": 0.0005, "loss": 1.085025429725647, "step": 49302 }, { "epoch": 3.886716594402838, "grad_norm": 0.37957271933555603, "learning_rate": 0.0005, "loss": 1.0032153129577637, "step": 49303 }, { "epoch": 3.886795427670477, "grad_norm": 0.8159806728363037, "learning_rate": 0.0005, "loss": 1.1859326362609863, "step": 49304 }, { "epoch": 3.8868742609381157, "grad_norm": 0.4734281599521637, "learning_rate": 0.0005, "loss": 1.2760801315307617, "step": 49305 }, { "epoch": 3.886953094205755, "grad_norm": 0.562208354473114, "learning_rate": 0.0005, "loss": 1.26972496509552, "step": 49306 }, { "epoch": 3.8870319274733935, "grad_norm": 0.5364094972610474, "learning_rate": 0.0005, "loss": 1.1739356517791748, "step": 49307 }, { "epoch": 3.8871107607410327, "grad_norm": 0.40979495644569397, "learning_rate": 0.0005, "loss": 1.1353261470794678, "step": 49308 }, { "epoch": 3.887189594008672, "grad_norm": 0.5197831988334656, "learning_rate": 0.0005, "loss": 1.1941829919815063, "step": 49309 }, { "epoch": 3.8872684272763105, "grad_norm": 0.47388094663619995, "learning_rate": 0.0005, "loss": 1.1041512489318848, "step": 49310 }, { "epoch": 3.8873472605439496, "grad_norm": 0.5562032461166382, "learning_rate": 0.0005, "loss": 1.2105300426483154, "step": 49311 }, { "epoch": 3.8874260938115883, "grad_norm": 0.42003288865089417, "learning_rate": 0.0005, "loss": 1.0072609186172485, "step": 49312 }, { "epoch": 3.8875049270792275, "grad_norm": 0.5137392282485962, "learning_rate": 0.0005, "loss": 1.2477107048034668, "step": 49313 }, { "epoch": 3.8875837603468666, "grad_norm": 0.49055546522140503, "learning_rate": 0.0005, "loss": 1.1493921279907227, "step": 49314 }, { "epoch": 3.8876625936145053, "grad_norm": 0.5177108645439148, "learning_rate": 0.0005, "loss": 1.2251722812652588, "step": 49315 }, { "epoch": 3.8877414268821444, "grad_norm": 0.38137394189834595, "learning_rate": 0.0005, "loss": 1.0354796648025513, "step": 49316 }, { "epoch": 3.887820260149783, "grad_norm": 0.5625471472740173, "learning_rate": 0.0005, "loss": 1.3258991241455078, "step": 49317 }, { "epoch": 3.8878990934174222, "grad_norm": 0.5149039030075073, "learning_rate": 0.0005, "loss": 1.0953350067138672, "step": 49318 }, { "epoch": 3.8879779266850614, "grad_norm": 0.4749113917350769, "learning_rate": 0.0005, "loss": 1.1856331825256348, "step": 49319 }, { "epoch": 3.8880567599527, "grad_norm": 0.41556093096733093, "learning_rate": 0.0005, "loss": 1.2084789276123047, "step": 49320 }, { "epoch": 3.8881355932203387, "grad_norm": 0.5849409699440002, "learning_rate": 0.0005, "loss": 1.3089337348937988, "step": 49321 }, { "epoch": 3.888214426487978, "grad_norm": 0.5699473023414612, "learning_rate": 0.0005, "loss": 1.193542242050171, "step": 49322 }, { "epoch": 3.888293259755617, "grad_norm": 0.448353111743927, "learning_rate": 0.0005, "loss": 1.158663272857666, "step": 49323 }, { "epoch": 3.8883720930232557, "grad_norm": 0.4354042112827301, "learning_rate": 0.0005, "loss": 1.2253897190093994, "step": 49324 }, { "epoch": 3.888450926290895, "grad_norm": 0.505364716053009, "learning_rate": 0.0005, "loss": 1.0875251293182373, "step": 49325 }, { "epoch": 3.8885297595585335, "grad_norm": 0.5438798069953918, "learning_rate": 0.0005, "loss": 1.202711582183838, "step": 49326 }, { "epoch": 3.8886085928261727, "grad_norm": 0.4901660978794098, "learning_rate": 0.0005, "loss": 1.008493185043335, "step": 49327 }, { "epoch": 3.888687426093812, "grad_norm": 0.5541176199913025, "learning_rate": 0.0005, "loss": 1.2382363080978394, "step": 49328 }, { "epoch": 3.8887662593614505, "grad_norm": 0.5368250012397766, "learning_rate": 0.0005, "loss": 1.1379287242889404, "step": 49329 }, { "epoch": 3.8888450926290896, "grad_norm": 0.4423763155937195, "learning_rate": 0.0005, "loss": 1.1720983982086182, "step": 49330 }, { "epoch": 3.8889239258967283, "grad_norm": 0.6127657890319824, "learning_rate": 0.0005, "loss": 1.2843676805496216, "step": 49331 }, { "epoch": 3.8890027591643674, "grad_norm": 0.4197123944759369, "learning_rate": 0.0005, "loss": 1.1176577806472778, "step": 49332 }, { "epoch": 3.8890815924320066, "grad_norm": 0.429837703704834, "learning_rate": 0.0005, "loss": 0.9597353935241699, "step": 49333 }, { "epoch": 3.8891604256996453, "grad_norm": 0.4103068709373474, "learning_rate": 0.0005, "loss": 1.0262538194656372, "step": 49334 }, { "epoch": 3.889239258967284, "grad_norm": 0.5654336214065552, "learning_rate": 0.0005, "loss": 1.2906959056854248, "step": 49335 }, { "epoch": 3.889318092234923, "grad_norm": 0.5152225494384766, "learning_rate": 0.0005, "loss": 1.0385706424713135, "step": 49336 }, { "epoch": 3.889396925502562, "grad_norm": 0.4840763509273529, "learning_rate": 0.0005, "loss": 1.097915768623352, "step": 49337 }, { "epoch": 3.889475758770201, "grad_norm": 0.44114091992378235, "learning_rate": 0.0005, "loss": 1.2170281410217285, "step": 49338 }, { "epoch": 3.88955459203784, "grad_norm": 0.47079598903656006, "learning_rate": 0.0005, "loss": 1.1360020637512207, "step": 49339 }, { "epoch": 3.8896334253054787, "grad_norm": 0.5493354797363281, "learning_rate": 0.0005, "loss": 1.1739821434020996, "step": 49340 }, { "epoch": 3.889712258573118, "grad_norm": 0.5413395762443542, "learning_rate": 0.0005, "loss": 1.0491855144500732, "step": 49341 }, { "epoch": 3.889791091840757, "grad_norm": 0.48976898193359375, "learning_rate": 0.0005, "loss": 1.2285957336425781, "step": 49342 }, { "epoch": 3.8898699251083957, "grad_norm": 0.6103103756904602, "learning_rate": 0.0005, "loss": 1.0606061220169067, "step": 49343 }, { "epoch": 3.889948758376035, "grad_norm": 0.6807882785797119, "learning_rate": 0.0005, "loss": 1.0968475341796875, "step": 49344 }, { "epoch": 3.8900275916436735, "grad_norm": 0.4501572847366333, "learning_rate": 0.0005, "loss": 1.2060062885284424, "step": 49345 }, { "epoch": 3.8901064249113126, "grad_norm": 0.4063645601272583, "learning_rate": 0.0005, "loss": 1.1304621696472168, "step": 49346 }, { "epoch": 3.8901852581789518, "grad_norm": 0.5344386696815491, "learning_rate": 0.0005, "loss": 1.0575697422027588, "step": 49347 }, { "epoch": 3.8902640914465905, "grad_norm": 0.4419337213039398, "learning_rate": 0.0005, "loss": 1.221787452697754, "step": 49348 }, { "epoch": 3.890342924714229, "grad_norm": 0.5043985843658447, "learning_rate": 0.0005, "loss": 1.3172703981399536, "step": 49349 }, { "epoch": 3.8904217579818683, "grad_norm": 0.4983404874801636, "learning_rate": 0.0005, "loss": 1.280588984489441, "step": 49350 }, { "epoch": 3.8905005912495074, "grad_norm": 0.4391705095767975, "learning_rate": 0.0005, "loss": 1.1424040794372559, "step": 49351 }, { "epoch": 3.890579424517146, "grad_norm": 0.5144033432006836, "learning_rate": 0.0005, "loss": 1.3249945640563965, "step": 49352 }, { "epoch": 3.8906582577847852, "grad_norm": 0.542587161064148, "learning_rate": 0.0005, "loss": 1.1492018699645996, "step": 49353 }, { "epoch": 3.890737091052424, "grad_norm": 0.750032901763916, "learning_rate": 0.0005, "loss": 1.345735788345337, "step": 49354 }, { "epoch": 3.890815924320063, "grad_norm": 0.48182427883148193, "learning_rate": 0.0005, "loss": 1.006353735923767, "step": 49355 }, { "epoch": 3.890894757587702, "grad_norm": 0.5676474571228027, "learning_rate": 0.0005, "loss": 1.1979650259017944, "step": 49356 }, { "epoch": 3.890973590855341, "grad_norm": 0.5141467452049255, "learning_rate": 0.0005, "loss": 1.325840950012207, "step": 49357 }, { "epoch": 3.89105242412298, "grad_norm": 0.545072078704834, "learning_rate": 0.0005, "loss": 1.100395679473877, "step": 49358 }, { "epoch": 3.8911312573906187, "grad_norm": 0.40925779938697815, "learning_rate": 0.0005, "loss": 1.1269972324371338, "step": 49359 }, { "epoch": 3.891210090658258, "grad_norm": 0.575493335723877, "learning_rate": 0.0005, "loss": 1.1655151844024658, "step": 49360 }, { "epoch": 3.891288923925897, "grad_norm": 0.4668245315551758, "learning_rate": 0.0005, "loss": 1.082369327545166, "step": 49361 }, { "epoch": 3.8913677571935357, "grad_norm": 0.4969545602798462, "learning_rate": 0.0005, "loss": 1.1619212627410889, "step": 49362 }, { "epoch": 3.8914465904611744, "grad_norm": 0.40974870324134827, "learning_rate": 0.0005, "loss": 1.0297635793685913, "step": 49363 }, { "epoch": 3.8915254237288135, "grad_norm": 0.5047357678413391, "learning_rate": 0.0005, "loss": 1.1718969345092773, "step": 49364 }, { "epoch": 3.8916042569964526, "grad_norm": 0.5885558724403381, "learning_rate": 0.0005, "loss": 1.3113751411437988, "step": 49365 }, { "epoch": 3.8916830902640913, "grad_norm": 0.5524122714996338, "learning_rate": 0.0005, "loss": 1.124859094619751, "step": 49366 }, { "epoch": 3.8917619235317304, "grad_norm": 0.4856691062450409, "learning_rate": 0.0005, "loss": 1.1635586023330688, "step": 49367 }, { "epoch": 3.891840756799369, "grad_norm": 0.42170292139053345, "learning_rate": 0.0005, "loss": 1.1253842115402222, "step": 49368 }, { "epoch": 3.8919195900670083, "grad_norm": 0.4421348571777344, "learning_rate": 0.0005, "loss": 1.211122751235962, "step": 49369 }, { "epoch": 3.8919984233346474, "grad_norm": 0.44189780950546265, "learning_rate": 0.0005, "loss": 1.1663519144058228, "step": 49370 }, { "epoch": 3.892077256602286, "grad_norm": 0.4791089594364166, "learning_rate": 0.0005, "loss": 1.0392882823944092, "step": 49371 }, { "epoch": 3.8921560898699252, "grad_norm": 0.4585617482662201, "learning_rate": 0.0005, "loss": 1.1977081298828125, "step": 49372 }, { "epoch": 3.892234923137564, "grad_norm": 0.4659459590911865, "learning_rate": 0.0005, "loss": 1.2001662254333496, "step": 49373 }, { "epoch": 3.892313756405203, "grad_norm": 0.4160805940628052, "learning_rate": 0.0005, "loss": 1.039840817451477, "step": 49374 }, { "epoch": 3.892392589672842, "grad_norm": 0.5069677829742432, "learning_rate": 0.0005, "loss": 1.0300852060317993, "step": 49375 }, { "epoch": 3.892471422940481, "grad_norm": 0.6103114485740662, "learning_rate": 0.0005, "loss": 1.2799406051635742, "step": 49376 }, { "epoch": 3.8925502562081196, "grad_norm": 0.41300952434539795, "learning_rate": 0.0005, "loss": 1.0647101402282715, "step": 49377 }, { "epoch": 3.8926290894757587, "grad_norm": 0.5274397134780884, "learning_rate": 0.0005, "loss": 1.3592495918273926, "step": 49378 }, { "epoch": 3.892707922743398, "grad_norm": 0.5424314141273499, "learning_rate": 0.0005, "loss": 1.1313985586166382, "step": 49379 }, { "epoch": 3.8927867560110365, "grad_norm": 0.42687368392944336, "learning_rate": 0.0005, "loss": 0.9804795384407043, "step": 49380 }, { "epoch": 3.8928655892786757, "grad_norm": 0.4691566228866577, "learning_rate": 0.0005, "loss": 1.1365725994110107, "step": 49381 }, { "epoch": 3.8929444225463143, "grad_norm": 0.40346387028694153, "learning_rate": 0.0005, "loss": 0.980027437210083, "step": 49382 }, { "epoch": 3.8930232558139535, "grad_norm": 0.49474626779556274, "learning_rate": 0.0005, "loss": 1.2585937976837158, "step": 49383 }, { "epoch": 3.8931020890815926, "grad_norm": 0.633063018321991, "learning_rate": 0.0005, "loss": 1.3202159404754639, "step": 49384 }, { "epoch": 3.8931809223492313, "grad_norm": 0.4237140119075775, "learning_rate": 0.0005, "loss": 1.177003264427185, "step": 49385 }, { "epoch": 3.8932597556168704, "grad_norm": 0.5674064755439758, "learning_rate": 0.0005, "loss": 1.1322708129882812, "step": 49386 }, { "epoch": 3.893338588884509, "grad_norm": 0.44441673159599304, "learning_rate": 0.0005, "loss": 1.1756932735443115, "step": 49387 }, { "epoch": 3.8934174221521483, "grad_norm": 0.517814040184021, "learning_rate": 0.0005, "loss": 1.3375868797302246, "step": 49388 }, { "epoch": 3.8934962554197874, "grad_norm": 0.39612773060798645, "learning_rate": 0.0005, "loss": 1.0785768032073975, "step": 49389 }, { "epoch": 3.893575088687426, "grad_norm": 0.43781620264053345, "learning_rate": 0.0005, "loss": 1.2240439653396606, "step": 49390 }, { "epoch": 3.8936539219550648, "grad_norm": 0.5565556883811951, "learning_rate": 0.0005, "loss": 1.2953490018844604, "step": 49391 }, { "epoch": 3.893732755222704, "grad_norm": 0.3975510597229004, "learning_rate": 0.0005, "loss": 1.077207088470459, "step": 49392 }, { "epoch": 3.893811588490343, "grad_norm": 0.5169236063957214, "learning_rate": 0.0005, "loss": 1.2178118228912354, "step": 49393 }, { "epoch": 3.8938904217579817, "grad_norm": 0.5172910690307617, "learning_rate": 0.0005, "loss": 1.0466489791870117, "step": 49394 }, { "epoch": 3.893969255025621, "grad_norm": 0.402081161737442, "learning_rate": 0.0005, "loss": 1.027215838432312, "step": 49395 }, { "epoch": 3.8940480882932595, "grad_norm": 0.46913576126098633, "learning_rate": 0.0005, "loss": 1.0577232837677002, "step": 49396 }, { "epoch": 3.8941269215608987, "grad_norm": 0.6129892468452454, "learning_rate": 0.0005, "loss": 1.2300877571105957, "step": 49397 }, { "epoch": 3.894205754828538, "grad_norm": 0.46043694019317627, "learning_rate": 0.0005, "loss": 1.09684419631958, "step": 49398 }, { "epoch": 3.8942845880961765, "grad_norm": 0.515225350856781, "learning_rate": 0.0005, "loss": 1.1799530982971191, "step": 49399 }, { "epoch": 3.8943634213638156, "grad_norm": 0.448617547750473, "learning_rate": 0.0005, "loss": 1.1466866731643677, "step": 49400 }, { "epoch": 3.8944422546314543, "grad_norm": 0.5516320466995239, "learning_rate": 0.0005, "loss": 1.2554734945297241, "step": 49401 }, { "epoch": 3.8945210878990935, "grad_norm": 0.43959522247314453, "learning_rate": 0.0005, "loss": 1.1878770589828491, "step": 49402 }, { "epoch": 3.8945999211667326, "grad_norm": 0.41902193427085876, "learning_rate": 0.0005, "loss": 1.0860460996627808, "step": 49403 }, { "epoch": 3.8946787544343713, "grad_norm": 0.4784112274646759, "learning_rate": 0.0005, "loss": 1.231330156326294, "step": 49404 }, { "epoch": 3.89475758770201, "grad_norm": 0.5054407715797424, "learning_rate": 0.0005, "loss": 1.0865445137023926, "step": 49405 }, { "epoch": 3.894836420969649, "grad_norm": 0.47052711248397827, "learning_rate": 0.0005, "loss": 1.3322029113769531, "step": 49406 }, { "epoch": 3.8949152542372882, "grad_norm": 0.4967077076435089, "learning_rate": 0.0005, "loss": 1.312099575996399, "step": 49407 }, { "epoch": 3.894994087504927, "grad_norm": 0.6592952609062195, "learning_rate": 0.0005, "loss": 1.3601043224334717, "step": 49408 }, { "epoch": 3.895072920772566, "grad_norm": 0.4522116184234619, "learning_rate": 0.0005, "loss": 1.1850945949554443, "step": 49409 }, { "epoch": 3.8951517540402048, "grad_norm": 0.8883266448974609, "learning_rate": 0.0005, "loss": 1.3704473972320557, "step": 49410 }, { "epoch": 3.895230587307844, "grad_norm": 0.42983314394950867, "learning_rate": 0.0005, "loss": 1.1702711582183838, "step": 49411 }, { "epoch": 3.895309420575483, "grad_norm": 0.45499640703201294, "learning_rate": 0.0005, "loss": 1.2100154161453247, "step": 49412 }, { "epoch": 3.8953882538431217, "grad_norm": 0.45738548040390015, "learning_rate": 0.0005, "loss": 1.0281128883361816, "step": 49413 }, { "epoch": 3.895467087110761, "grad_norm": 0.5046026110649109, "learning_rate": 0.0005, "loss": 1.2656221389770508, "step": 49414 }, { "epoch": 3.8955459203783995, "grad_norm": 0.44474178552627563, "learning_rate": 0.0005, "loss": 1.0514607429504395, "step": 49415 }, { "epoch": 3.8956247536460387, "grad_norm": 0.4864739775657654, "learning_rate": 0.0005, "loss": 1.235632061958313, "step": 49416 }, { "epoch": 3.895703586913678, "grad_norm": 0.48530659079551697, "learning_rate": 0.0005, "loss": 1.0716791152954102, "step": 49417 }, { "epoch": 3.8957824201813165, "grad_norm": 0.5112563371658325, "learning_rate": 0.0005, "loss": 1.1422874927520752, "step": 49418 }, { "epoch": 3.8958612534489556, "grad_norm": 0.4697422981262207, "learning_rate": 0.0005, "loss": 1.255074143409729, "step": 49419 }, { "epoch": 3.8959400867165943, "grad_norm": 0.6585204601287842, "learning_rate": 0.0005, "loss": 1.2508134841918945, "step": 49420 }, { "epoch": 3.8960189199842334, "grad_norm": 0.4434475004673004, "learning_rate": 0.0005, "loss": 1.1513850688934326, "step": 49421 }, { "epoch": 3.896097753251872, "grad_norm": 0.6869479417800903, "learning_rate": 0.0005, "loss": 1.3045666217803955, "step": 49422 }, { "epoch": 3.8961765865195113, "grad_norm": 0.5413837432861328, "learning_rate": 0.0005, "loss": 1.4196058511734009, "step": 49423 }, { "epoch": 3.89625541978715, "grad_norm": 0.45840173959732056, "learning_rate": 0.0005, "loss": 1.106673002243042, "step": 49424 }, { "epoch": 3.896334253054789, "grad_norm": 0.4233884811401367, "learning_rate": 0.0005, "loss": 1.0724895000457764, "step": 49425 }, { "epoch": 3.896413086322428, "grad_norm": 0.40588200092315674, "learning_rate": 0.0005, "loss": 1.1724586486816406, "step": 49426 }, { "epoch": 3.896491919590067, "grad_norm": 0.49540674686431885, "learning_rate": 0.0005, "loss": 1.2378567457199097, "step": 49427 }, { "epoch": 3.896570752857706, "grad_norm": 0.4795337915420532, "learning_rate": 0.0005, "loss": 1.16562819480896, "step": 49428 }, { "epoch": 3.8966495861253447, "grad_norm": 0.7330682277679443, "learning_rate": 0.0005, "loss": 1.1303322315216064, "step": 49429 }, { "epoch": 3.896728419392984, "grad_norm": 0.45381641387939453, "learning_rate": 0.0005, "loss": 1.1061158180236816, "step": 49430 }, { "epoch": 3.896807252660623, "grad_norm": 0.5029628872871399, "learning_rate": 0.0005, "loss": 1.1493031978607178, "step": 49431 }, { "epoch": 3.8968860859282617, "grad_norm": 0.4274950325489044, "learning_rate": 0.0005, "loss": 1.177727222442627, "step": 49432 }, { "epoch": 3.896964919195901, "grad_norm": 0.39681190252304077, "learning_rate": 0.0005, "loss": 0.9270937442779541, "step": 49433 }, { "epoch": 3.8970437524635395, "grad_norm": 0.447888046503067, "learning_rate": 0.0005, "loss": 0.9050706624984741, "step": 49434 }, { "epoch": 3.8971225857311786, "grad_norm": 0.4416697025299072, "learning_rate": 0.0005, "loss": 1.156470775604248, "step": 49435 }, { "epoch": 3.8972014189988173, "grad_norm": 0.5195814967155457, "learning_rate": 0.0005, "loss": 1.1412863731384277, "step": 49436 }, { "epoch": 3.8972802522664565, "grad_norm": 0.4788767397403717, "learning_rate": 0.0005, "loss": 1.1622340679168701, "step": 49437 }, { "epoch": 3.897359085534095, "grad_norm": 0.415745347738266, "learning_rate": 0.0005, "loss": 1.0709187984466553, "step": 49438 }, { "epoch": 3.8974379188017343, "grad_norm": 0.5382188558578491, "learning_rate": 0.0005, "loss": 1.1770747900009155, "step": 49439 }, { "epoch": 3.8975167520693734, "grad_norm": 0.5402147173881531, "learning_rate": 0.0005, "loss": 1.127898097038269, "step": 49440 }, { "epoch": 3.897595585337012, "grad_norm": 0.4551009237766266, "learning_rate": 0.0005, "loss": 1.1225663423538208, "step": 49441 }, { "epoch": 3.8976744186046512, "grad_norm": 0.48390644788742065, "learning_rate": 0.0005, "loss": 1.4081242084503174, "step": 49442 }, { "epoch": 3.89775325187229, "grad_norm": 0.5546343326568604, "learning_rate": 0.0005, "loss": 0.9336374998092651, "step": 49443 }, { "epoch": 3.897832085139929, "grad_norm": 0.6053643822669983, "learning_rate": 0.0005, "loss": 1.1801857948303223, "step": 49444 }, { "epoch": 3.897910918407568, "grad_norm": 0.5920146107673645, "learning_rate": 0.0005, "loss": 1.226867437362671, "step": 49445 }, { "epoch": 3.897989751675207, "grad_norm": 0.5105870366096497, "learning_rate": 0.0005, "loss": 1.1589374542236328, "step": 49446 }, { "epoch": 3.898068584942846, "grad_norm": 0.49537643790245056, "learning_rate": 0.0005, "loss": 1.2268199920654297, "step": 49447 }, { "epoch": 3.8981474182104847, "grad_norm": 0.5108227729797363, "learning_rate": 0.0005, "loss": 1.194100260734558, "step": 49448 }, { "epoch": 3.898226251478124, "grad_norm": 0.45127567648887634, "learning_rate": 0.0005, "loss": 1.2071049213409424, "step": 49449 }, { "epoch": 3.898305084745763, "grad_norm": 0.5093688368797302, "learning_rate": 0.0005, "loss": 1.285677433013916, "step": 49450 }, { "epoch": 3.8983839180134017, "grad_norm": 0.5162752270698547, "learning_rate": 0.0005, "loss": 1.0949105024337769, "step": 49451 }, { "epoch": 3.8984627512810404, "grad_norm": 0.416765034198761, "learning_rate": 0.0005, "loss": 1.1094586849212646, "step": 49452 }, { "epoch": 3.8985415845486795, "grad_norm": 0.4925716817378998, "learning_rate": 0.0005, "loss": 1.0264403820037842, "step": 49453 }, { "epoch": 3.8986204178163186, "grad_norm": 0.5237082242965698, "learning_rate": 0.0005, "loss": 1.1649352312088013, "step": 49454 }, { "epoch": 3.8986992510839573, "grad_norm": 0.507968008518219, "learning_rate": 0.0005, "loss": 1.143320083618164, "step": 49455 }, { "epoch": 3.8987780843515965, "grad_norm": 0.6653861403465271, "learning_rate": 0.0005, "loss": 1.2334494590759277, "step": 49456 }, { "epoch": 3.898856917619235, "grad_norm": 0.479846715927124, "learning_rate": 0.0005, "loss": 1.1182334423065186, "step": 49457 }, { "epoch": 3.8989357508868743, "grad_norm": 0.483642578125, "learning_rate": 0.0005, "loss": 1.321321725845337, "step": 49458 }, { "epoch": 3.8990145841545134, "grad_norm": 0.7359700202941895, "learning_rate": 0.0005, "loss": 1.161161184310913, "step": 49459 }, { "epoch": 3.899093417422152, "grad_norm": 0.3760005533695221, "learning_rate": 0.0005, "loss": 1.0185492038726807, "step": 49460 }, { "epoch": 3.8991722506897912, "grad_norm": 0.5510782599449158, "learning_rate": 0.0005, "loss": 1.0678939819335938, "step": 49461 }, { "epoch": 3.89925108395743, "grad_norm": 0.5651661157608032, "learning_rate": 0.0005, "loss": 1.0970165729522705, "step": 49462 }, { "epoch": 3.899329917225069, "grad_norm": 0.5067861080169678, "learning_rate": 0.0005, "loss": 1.1809895038604736, "step": 49463 }, { "epoch": 3.899408750492708, "grad_norm": 0.4490450918674469, "learning_rate": 0.0005, "loss": 1.0258736610412598, "step": 49464 }, { "epoch": 3.899487583760347, "grad_norm": 0.516274094581604, "learning_rate": 0.0005, "loss": 1.3104636669158936, "step": 49465 }, { "epoch": 3.8995664170279856, "grad_norm": 0.45745325088500977, "learning_rate": 0.0005, "loss": 1.2489724159240723, "step": 49466 }, { "epoch": 3.8996452502956247, "grad_norm": 0.5437428951263428, "learning_rate": 0.0005, "loss": 1.2783015966415405, "step": 49467 }, { "epoch": 3.899724083563264, "grad_norm": 0.46671658754348755, "learning_rate": 0.0005, "loss": 1.065721035003662, "step": 49468 }, { "epoch": 3.8998029168309025, "grad_norm": 0.40363308787345886, "learning_rate": 0.0005, "loss": 1.1011326313018799, "step": 49469 }, { "epoch": 3.8998817500985417, "grad_norm": 0.6285935044288635, "learning_rate": 0.0005, "loss": 1.3026721477508545, "step": 49470 }, { "epoch": 3.8999605833661803, "grad_norm": 0.6393488049507141, "learning_rate": 0.0005, "loss": 1.2642605304718018, "step": 49471 }, { "epoch": 3.9000394166338195, "grad_norm": 0.4438067078590393, "learning_rate": 0.0005, "loss": 1.2409956455230713, "step": 49472 }, { "epoch": 3.9001182499014586, "grad_norm": 0.5807003974914551, "learning_rate": 0.0005, "loss": 1.300724744796753, "step": 49473 }, { "epoch": 3.9001970831690973, "grad_norm": 0.4693230092525482, "learning_rate": 0.0005, "loss": 1.1098566055297852, "step": 49474 }, { "epoch": 3.9002759164367364, "grad_norm": 0.5372792482376099, "learning_rate": 0.0005, "loss": 1.1751768589019775, "step": 49475 }, { "epoch": 3.900354749704375, "grad_norm": 0.4476514458656311, "learning_rate": 0.0005, "loss": 1.1882132291793823, "step": 49476 }, { "epoch": 3.9004335829720143, "grad_norm": 0.7158970236778259, "learning_rate": 0.0005, "loss": 1.3285067081451416, "step": 49477 }, { "epoch": 3.9005124162396534, "grad_norm": 0.4333728551864624, "learning_rate": 0.0005, "loss": 1.071666955947876, "step": 49478 }, { "epoch": 3.900591249507292, "grad_norm": 0.5479240417480469, "learning_rate": 0.0005, "loss": 1.262817144393921, "step": 49479 }, { "epoch": 3.9006700827749308, "grad_norm": 0.5013125538825989, "learning_rate": 0.0005, "loss": 1.1816444396972656, "step": 49480 }, { "epoch": 3.90074891604257, "grad_norm": 0.4108890891075134, "learning_rate": 0.0005, "loss": 1.1228885650634766, "step": 49481 }, { "epoch": 3.900827749310209, "grad_norm": 0.4734092056751251, "learning_rate": 0.0005, "loss": 1.0709033012390137, "step": 49482 }, { "epoch": 3.9009065825778477, "grad_norm": 0.5433441996574402, "learning_rate": 0.0005, "loss": 1.2267221212387085, "step": 49483 }, { "epoch": 3.900985415845487, "grad_norm": 0.580985963344574, "learning_rate": 0.0005, "loss": 1.4866275787353516, "step": 49484 }, { "epoch": 3.9010642491131255, "grad_norm": 0.4497859477996826, "learning_rate": 0.0005, "loss": 1.3354039192199707, "step": 49485 }, { "epoch": 3.9011430823807647, "grad_norm": 0.437688410282135, "learning_rate": 0.0005, "loss": 0.90346360206604, "step": 49486 }, { "epoch": 3.901221915648404, "grad_norm": 0.6193516850471497, "learning_rate": 0.0005, "loss": 1.2652990818023682, "step": 49487 }, { "epoch": 3.9013007489160425, "grad_norm": 0.49156883358955383, "learning_rate": 0.0005, "loss": 1.1488041877746582, "step": 49488 }, { "epoch": 3.9013795821836816, "grad_norm": 0.4665830135345459, "learning_rate": 0.0005, "loss": 1.3813631534576416, "step": 49489 }, { "epoch": 3.9014584154513203, "grad_norm": 0.4845462739467621, "learning_rate": 0.0005, "loss": 1.140498161315918, "step": 49490 }, { "epoch": 3.9015372487189595, "grad_norm": 0.6328228116035461, "learning_rate": 0.0005, "loss": 1.4094364643096924, "step": 49491 }, { "epoch": 3.9016160819865986, "grad_norm": 0.5226905345916748, "learning_rate": 0.0005, "loss": 1.1549944877624512, "step": 49492 }, { "epoch": 3.9016949152542373, "grad_norm": 0.5660383701324463, "learning_rate": 0.0005, "loss": 1.3019518852233887, "step": 49493 }, { "epoch": 3.901773748521876, "grad_norm": 0.4959631860256195, "learning_rate": 0.0005, "loss": 1.1834676265716553, "step": 49494 }, { "epoch": 3.901852581789515, "grad_norm": 0.4959162771701813, "learning_rate": 0.0005, "loss": 1.0914020538330078, "step": 49495 }, { "epoch": 3.9019314150571542, "grad_norm": 0.45800331234931946, "learning_rate": 0.0005, "loss": 1.1577086448669434, "step": 49496 }, { "epoch": 3.902010248324793, "grad_norm": 0.5267906785011292, "learning_rate": 0.0005, "loss": 1.1048914194107056, "step": 49497 }, { "epoch": 3.902089081592432, "grad_norm": 0.4102444052696228, "learning_rate": 0.0005, "loss": 1.0959681272506714, "step": 49498 }, { "epoch": 3.9021679148600708, "grad_norm": 0.48450878262519836, "learning_rate": 0.0005, "loss": 1.0500160455703735, "step": 49499 }, { "epoch": 3.90224674812771, "grad_norm": 0.5878028869628906, "learning_rate": 0.0005, "loss": 1.195881962776184, "step": 49500 }, { "epoch": 3.902325581395349, "grad_norm": 0.5804275274276733, "learning_rate": 0.0005, "loss": 1.1850976943969727, "step": 49501 }, { "epoch": 3.9024044146629877, "grad_norm": 0.5346814393997192, "learning_rate": 0.0005, "loss": 1.104170560836792, "step": 49502 }, { "epoch": 3.902483247930627, "grad_norm": 0.5589644908905029, "learning_rate": 0.0005, "loss": 1.1032953262329102, "step": 49503 }, { "epoch": 3.9025620811982655, "grad_norm": 0.5508483648300171, "learning_rate": 0.0005, "loss": 1.0875308513641357, "step": 49504 }, { "epoch": 3.9026409144659047, "grad_norm": 0.5673021674156189, "learning_rate": 0.0005, "loss": 1.2394534349441528, "step": 49505 }, { "epoch": 3.902719747733544, "grad_norm": 0.6767434477806091, "learning_rate": 0.0005, "loss": 1.2792375087738037, "step": 49506 }, { "epoch": 3.9027985810011825, "grad_norm": 0.5104861259460449, "learning_rate": 0.0005, "loss": 1.2245135307312012, "step": 49507 }, { "epoch": 3.902877414268821, "grad_norm": 0.4886687994003296, "learning_rate": 0.0005, "loss": 1.2934134006500244, "step": 49508 }, { "epoch": 3.9029562475364603, "grad_norm": 0.6402227878570557, "learning_rate": 0.0005, "loss": 1.0590637922286987, "step": 49509 }, { "epoch": 3.9030350808040994, "grad_norm": 0.4922228157520294, "learning_rate": 0.0005, "loss": 1.0830819606781006, "step": 49510 }, { "epoch": 3.903113914071738, "grad_norm": 0.4218890070915222, "learning_rate": 0.0005, "loss": 1.19717538356781, "step": 49511 }, { "epoch": 3.9031927473393773, "grad_norm": 0.4249442517757416, "learning_rate": 0.0005, "loss": 1.0765726566314697, "step": 49512 }, { "epoch": 3.903271580607016, "grad_norm": 0.48245570063591003, "learning_rate": 0.0005, "loss": 1.1216833591461182, "step": 49513 }, { "epoch": 3.903350413874655, "grad_norm": 0.4288600981235504, "learning_rate": 0.0005, "loss": 1.1254308223724365, "step": 49514 }, { "epoch": 3.9034292471422942, "grad_norm": 0.5074785351753235, "learning_rate": 0.0005, "loss": 1.218856692314148, "step": 49515 }, { "epoch": 3.903508080409933, "grad_norm": 0.4741005301475525, "learning_rate": 0.0005, "loss": 1.1800256967544556, "step": 49516 }, { "epoch": 3.903586913677572, "grad_norm": 0.45679500699043274, "learning_rate": 0.0005, "loss": 1.4225622415542603, "step": 49517 }, { "epoch": 3.9036657469452107, "grad_norm": 0.5479820966720581, "learning_rate": 0.0005, "loss": 1.1114000082015991, "step": 49518 }, { "epoch": 3.90374458021285, "grad_norm": 0.500896155834198, "learning_rate": 0.0005, "loss": 1.1733207702636719, "step": 49519 }, { "epoch": 3.903823413480489, "grad_norm": 0.5105498433113098, "learning_rate": 0.0005, "loss": 1.2175085544586182, "step": 49520 }, { "epoch": 3.9039022467481277, "grad_norm": 0.5189735293388367, "learning_rate": 0.0005, "loss": 1.2179149389266968, "step": 49521 }, { "epoch": 3.9039810800157664, "grad_norm": 0.4914778470993042, "learning_rate": 0.0005, "loss": 1.2628681659698486, "step": 49522 }, { "epoch": 3.9040599132834055, "grad_norm": 0.569753885269165, "learning_rate": 0.0005, "loss": 1.1312675476074219, "step": 49523 }, { "epoch": 3.9041387465510446, "grad_norm": 0.46449774503707886, "learning_rate": 0.0005, "loss": 1.1719000339508057, "step": 49524 }, { "epoch": 3.9042175798186833, "grad_norm": 0.5008120536804199, "learning_rate": 0.0005, "loss": 1.2059580087661743, "step": 49525 }, { "epoch": 3.9042964130863225, "grad_norm": 0.5675634741783142, "learning_rate": 0.0005, "loss": 1.2574926614761353, "step": 49526 }, { "epoch": 3.904375246353961, "grad_norm": 0.5157154202461243, "learning_rate": 0.0005, "loss": 1.2966160774230957, "step": 49527 }, { "epoch": 3.9044540796216003, "grad_norm": 0.6906276941299438, "learning_rate": 0.0005, "loss": 1.2982091903686523, "step": 49528 }, { "epoch": 3.9045329128892394, "grad_norm": 0.4881483316421509, "learning_rate": 0.0005, "loss": 1.15928316116333, "step": 49529 }, { "epoch": 3.904611746156878, "grad_norm": 0.4684086740016937, "learning_rate": 0.0005, "loss": 1.1460286378860474, "step": 49530 }, { "epoch": 3.9046905794245172, "grad_norm": 0.5026145577430725, "learning_rate": 0.0005, "loss": 1.1390635967254639, "step": 49531 }, { "epoch": 3.904769412692156, "grad_norm": 0.4731047451496124, "learning_rate": 0.0005, "loss": 1.1168711185455322, "step": 49532 }, { "epoch": 3.904848245959795, "grad_norm": 0.5319685339927673, "learning_rate": 0.0005, "loss": 1.081575632095337, "step": 49533 }, { "epoch": 3.904927079227434, "grad_norm": 0.47743719816207886, "learning_rate": 0.0005, "loss": 1.2209023237228394, "step": 49534 }, { "epoch": 3.905005912495073, "grad_norm": 0.4947021007537842, "learning_rate": 0.0005, "loss": 1.2248601913452148, "step": 49535 }, { "epoch": 3.905084745762712, "grad_norm": 0.5403580069541931, "learning_rate": 0.0005, "loss": 1.3814857006072998, "step": 49536 }, { "epoch": 3.9051635790303507, "grad_norm": 0.4986571669578552, "learning_rate": 0.0005, "loss": 1.294921636581421, "step": 49537 }, { "epoch": 3.90524241229799, "grad_norm": 0.4673689305782318, "learning_rate": 0.0005, "loss": 1.1742548942565918, "step": 49538 }, { "epoch": 3.9053212455656285, "grad_norm": 0.5761693120002747, "learning_rate": 0.0005, "loss": 1.1170589923858643, "step": 49539 }, { "epoch": 3.9054000788332677, "grad_norm": 0.651799201965332, "learning_rate": 0.0005, "loss": 1.2906982898712158, "step": 49540 }, { "epoch": 3.9054789121009064, "grad_norm": 0.43642887473106384, "learning_rate": 0.0005, "loss": 1.0611751079559326, "step": 49541 }, { "epoch": 3.9055577453685455, "grad_norm": 0.6038126945495605, "learning_rate": 0.0005, "loss": 1.253082275390625, "step": 49542 }, { "epoch": 3.9056365786361846, "grad_norm": 0.49705079197883606, "learning_rate": 0.0005, "loss": 1.1079765558242798, "step": 49543 }, { "epoch": 3.9057154119038233, "grad_norm": 0.5135762691497803, "learning_rate": 0.0005, "loss": 1.1149165630340576, "step": 49544 }, { "epoch": 3.9057942451714625, "grad_norm": 0.43365639448165894, "learning_rate": 0.0005, "loss": 1.0480149984359741, "step": 49545 }, { "epoch": 3.905873078439101, "grad_norm": 0.4907433092594147, "learning_rate": 0.0005, "loss": 1.1909031867980957, "step": 49546 }, { "epoch": 3.9059519117067403, "grad_norm": 0.4854936897754669, "learning_rate": 0.0005, "loss": 1.0361731052398682, "step": 49547 }, { "epoch": 3.9060307449743794, "grad_norm": 0.5620569586753845, "learning_rate": 0.0005, "loss": 1.195403814315796, "step": 49548 }, { "epoch": 3.906109578242018, "grad_norm": 0.426799476146698, "learning_rate": 0.0005, "loss": 0.9946718215942383, "step": 49549 }, { "epoch": 3.9061884115096572, "grad_norm": 0.5286784768104553, "learning_rate": 0.0005, "loss": 1.2426031827926636, "step": 49550 }, { "epoch": 3.906267244777296, "grad_norm": 0.52138352394104, "learning_rate": 0.0005, "loss": 1.3213441371917725, "step": 49551 }, { "epoch": 3.906346078044935, "grad_norm": 0.5708504915237427, "learning_rate": 0.0005, "loss": 1.2878367900848389, "step": 49552 }, { "epoch": 3.9064249113125737, "grad_norm": 0.5660890936851501, "learning_rate": 0.0005, "loss": 1.0803639888763428, "step": 49553 }, { "epoch": 3.906503744580213, "grad_norm": 0.4871198236942291, "learning_rate": 0.0005, "loss": 1.327394723892212, "step": 49554 }, { "epoch": 3.9065825778478516, "grad_norm": 0.5117250084877014, "learning_rate": 0.0005, "loss": 1.0495072603225708, "step": 49555 }, { "epoch": 3.9066614111154907, "grad_norm": 0.5200002193450928, "learning_rate": 0.0005, "loss": 1.3286197185516357, "step": 49556 }, { "epoch": 3.90674024438313, "grad_norm": 0.4119006097316742, "learning_rate": 0.0005, "loss": 1.1279065608978271, "step": 49557 }, { "epoch": 3.9068190776507685, "grad_norm": 0.41637909412384033, "learning_rate": 0.0005, "loss": 1.0609874725341797, "step": 49558 }, { "epoch": 3.9068979109184077, "grad_norm": 0.5323359966278076, "learning_rate": 0.0005, "loss": 1.3474445343017578, "step": 49559 }, { "epoch": 3.9069767441860463, "grad_norm": 0.7171778082847595, "learning_rate": 0.0005, "loss": 1.1529467105865479, "step": 49560 }, { "epoch": 3.9070555774536855, "grad_norm": 0.4585348069667816, "learning_rate": 0.0005, "loss": 0.9813234806060791, "step": 49561 }, { "epoch": 3.9071344107213246, "grad_norm": 0.4356503486633301, "learning_rate": 0.0005, "loss": 1.270941138267517, "step": 49562 }, { "epoch": 3.9072132439889633, "grad_norm": 0.4803386330604553, "learning_rate": 0.0005, "loss": 1.1008503437042236, "step": 49563 }, { "epoch": 3.9072920772566024, "grad_norm": 0.5143392086029053, "learning_rate": 0.0005, "loss": 1.0644539594650269, "step": 49564 }, { "epoch": 3.907370910524241, "grad_norm": 0.5128523707389832, "learning_rate": 0.0005, "loss": 1.3363711833953857, "step": 49565 }, { "epoch": 3.9074497437918803, "grad_norm": 0.5184471011161804, "learning_rate": 0.0005, "loss": 1.095991611480713, "step": 49566 }, { "epoch": 3.9075285770595194, "grad_norm": 0.4393369257450104, "learning_rate": 0.0005, "loss": 1.0731251239776611, "step": 49567 }, { "epoch": 3.907607410327158, "grad_norm": 0.5362618565559387, "learning_rate": 0.0005, "loss": 1.217151403427124, "step": 49568 }, { "epoch": 3.9076862435947968, "grad_norm": 0.4748358428478241, "learning_rate": 0.0005, "loss": 1.216507077217102, "step": 49569 }, { "epoch": 3.907765076862436, "grad_norm": 0.48213475942611694, "learning_rate": 0.0005, "loss": 1.1347026824951172, "step": 49570 }, { "epoch": 3.907843910130075, "grad_norm": 0.5229143500328064, "learning_rate": 0.0005, "loss": 1.2200136184692383, "step": 49571 }, { "epoch": 3.9079227433977137, "grad_norm": 0.5061835646629333, "learning_rate": 0.0005, "loss": 1.2604155540466309, "step": 49572 }, { "epoch": 3.908001576665353, "grad_norm": 0.46027126908302307, "learning_rate": 0.0005, "loss": 1.195544958114624, "step": 49573 }, { "epoch": 3.9080804099329916, "grad_norm": 0.5113865733146667, "learning_rate": 0.0005, "loss": 1.3336896896362305, "step": 49574 }, { "epoch": 3.9081592432006307, "grad_norm": 0.4471617043018341, "learning_rate": 0.0005, "loss": 1.1009533405303955, "step": 49575 }, { "epoch": 3.90823807646827, "grad_norm": 0.5322472453117371, "learning_rate": 0.0005, "loss": 1.0706998109817505, "step": 49576 }, { "epoch": 3.9083169097359085, "grad_norm": 0.781248152256012, "learning_rate": 0.0005, "loss": 1.3522567749023438, "step": 49577 }, { "epoch": 3.9083957430035476, "grad_norm": 0.565371036529541, "learning_rate": 0.0005, "loss": 1.056588888168335, "step": 49578 }, { "epoch": 3.9084745762711863, "grad_norm": 0.48066624999046326, "learning_rate": 0.0005, "loss": 1.1825692653656006, "step": 49579 }, { "epoch": 3.9085534095388255, "grad_norm": 0.5967612266540527, "learning_rate": 0.0005, "loss": 1.3314307928085327, "step": 49580 }, { "epoch": 3.9086322428064646, "grad_norm": 0.5371701121330261, "learning_rate": 0.0005, "loss": 1.1969366073608398, "step": 49581 }, { "epoch": 3.9087110760741033, "grad_norm": 0.4593691825866699, "learning_rate": 0.0005, "loss": 1.243675708770752, "step": 49582 }, { "epoch": 3.908789909341742, "grad_norm": 0.5693627595901489, "learning_rate": 0.0005, "loss": 1.0532488822937012, "step": 49583 }, { "epoch": 3.908868742609381, "grad_norm": 0.4876370131969452, "learning_rate": 0.0005, "loss": 1.2372565269470215, "step": 49584 }, { "epoch": 3.9089475758770202, "grad_norm": 0.6614982485771179, "learning_rate": 0.0005, "loss": 1.2297134399414062, "step": 49585 }, { "epoch": 3.909026409144659, "grad_norm": 0.5102225542068481, "learning_rate": 0.0005, "loss": 1.14984130859375, "step": 49586 }, { "epoch": 3.909105242412298, "grad_norm": 0.41344907879829407, "learning_rate": 0.0005, "loss": 0.9208613634109497, "step": 49587 }, { "epoch": 3.9091840756799368, "grad_norm": 0.5698060393333435, "learning_rate": 0.0005, "loss": 1.0846564769744873, "step": 49588 }, { "epoch": 3.909262908947576, "grad_norm": 0.4703182578086853, "learning_rate": 0.0005, "loss": 1.2224555015563965, "step": 49589 }, { "epoch": 3.909341742215215, "grad_norm": 0.5579872727394104, "learning_rate": 0.0005, "loss": 1.1009935140609741, "step": 49590 }, { "epoch": 3.9094205754828537, "grad_norm": 0.508410632610321, "learning_rate": 0.0005, "loss": 1.0209254026412964, "step": 49591 }, { "epoch": 3.909499408750493, "grad_norm": 0.6670307517051697, "learning_rate": 0.0005, "loss": 1.1617738008499146, "step": 49592 }, { "epoch": 3.9095782420181315, "grad_norm": 0.6073035597801208, "learning_rate": 0.0005, "loss": 1.1552002429962158, "step": 49593 }, { "epoch": 3.9096570752857707, "grad_norm": 0.46487024426460266, "learning_rate": 0.0005, "loss": 1.258664846420288, "step": 49594 }, { "epoch": 3.90973590855341, "grad_norm": 0.46904444694519043, "learning_rate": 0.0005, "loss": 1.046753168106079, "step": 49595 }, { "epoch": 3.9098147418210485, "grad_norm": 0.5635169148445129, "learning_rate": 0.0005, "loss": 1.2636456489562988, "step": 49596 }, { "epoch": 3.909893575088687, "grad_norm": 0.6438485383987427, "learning_rate": 0.0005, "loss": 1.2173607349395752, "step": 49597 }, { "epoch": 3.9099724083563263, "grad_norm": 0.3869214653968811, "learning_rate": 0.0005, "loss": 1.1572537422180176, "step": 49598 }, { "epoch": 3.9100512416239654, "grad_norm": 0.692266047000885, "learning_rate": 0.0005, "loss": 1.2765662670135498, "step": 49599 }, { "epoch": 3.910130074891604, "grad_norm": 0.46450987458229065, "learning_rate": 0.0005, "loss": 1.2112640142440796, "step": 49600 }, { "epoch": 3.9102089081592433, "grad_norm": 0.7450687289237976, "learning_rate": 0.0005, "loss": 1.3122906684875488, "step": 49601 }, { "epoch": 3.910287741426882, "grad_norm": 0.5221045613288879, "learning_rate": 0.0005, "loss": 1.138869047164917, "step": 49602 }, { "epoch": 3.910366574694521, "grad_norm": 0.6175450682640076, "learning_rate": 0.0005, "loss": 1.4433566331863403, "step": 49603 }, { "epoch": 3.9104454079621602, "grad_norm": 0.45855823159217834, "learning_rate": 0.0005, "loss": 1.0692870616912842, "step": 49604 }, { "epoch": 3.910524241229799, "grad_norm": 0.5660306215286255, "learning_rate": 0.0005, "loss": 1.2569814920425415, "step": 49605 }, { "epoch": 3.910603074497438, "grad_norm": 0.47958698868751526, "learning_rate": 0.0005, "loss": 1.139607548713684, "step": 49606 }, { "epoch": 3.9106819077650767, "grad_norm": 0.5254287719726562, "learning_rate": 0.0005, "loss": 1.4532849788665771, "step": 49607 }, { "epoch": 3.910760741032716, "grad_norm": 0.46391621232032776, "learning_rate": 0.0005, "loss": 1.1952576637268066, "step": 49608 }, { "epoch": 3.910839574300355, "grad_norm": 0.5093062520027161, "learning_rate": 0.0005, "loss": 1.24216628074646, "step": 49609 }, { "epoch": 3.9109184075679937, "grad_norm": 0.4677135944366455, "learning_rate": 0.0005, "loss": 1.0936585664749146, "step": 49610 }, { "epoch": 3.9109972408356324, "grad_norm": 0.5627292990684509, "learning_rate": 0.0005, "loss": 1.239073395729065, "step": 49611 }, { "epoch": 3.9110760741032715, "grad_norm": 0.5617097616195679, "learning_rate": 0.0005, "loss": 1.1850965023040771, "step": 49612 }, { "epoch": 3.9111549073709106, "grad_norm": 0.5007815957069397, "learning_rate": 0.0005, "loss": 1.2400461435317993, "step": 49613 }, { "epoch": 3.9112337406385493, "grad_norm": 0.42659348249435425, "learning_rate": 0.0005, "loss": 1.1556473970413208, "step": 49614 }, { "epoch": 3.9113125739061885, "grad_norm": 0.4882086217403412, "learning_rate": 0.0005, "loss": 1.216238021850586, "step": 49615 }, { "epoch": 3.911391407173827, "grad_norm": 0.48555460572242737, "learning_rate": 0.0005, "loss": 1.3658595085144043, "step": 49616 }, { "epoch": 3.9114702404414663, "grad_norm": 0.5571931004524231, "learning_rate": 0.0005, "loss": 1.398155927658081, "step": 49617 }, { "epoch": 3.9115490737091054, "grad_norm": 0.43583598732948303, "learning_rate": 0.0005, "loss": 1.0377702713012695, "step": 49618 }, { "epoch": 3.911627906976744, "grad_norm": 0.5390387773513794, "learning_rate": 0.0005, "loss": 1.253709316253662, "step": 49619 }, { "epoch": 3.9117067402443833, "grad_norm": 0.5543522238731384, "learning_rate": 0.0005, "loss": 1.1846725940704346, "step": 49620 }, { "epoch": 3.911785573512022, "grad_norm": 0.4756042957305908, "learning_rate": 0.0005, "loss": 1.1858019828796387, "step": 49621 }, { "epoch": 3.911864406779661, "grad_norm": 0.514176070690155, "learning_rate": 0.0005, "loss": 1.2167344093322754, "step": 49622 }, { "epoch": 3.9119432400473, "grad_norm": 0.5905640721321106, "learning_rate": 0.0005, "loss": 1.0538606643676758, "step": 49623 }, { "epoch": 3.912022073314939, "grad_norm": 0.582714855670929, "learning_rate": 0.0005, "loss": 1.072200059890747, "step": 49624 }, { "epoch": 3.9121009065825776, "grad_norm": 0.475674569606781, "learning_rate": 0.0005, "loss": 1.1824042797088623, "step": 49625 }, { "epoch": 3.9121797398502167, "grad_norm": 0.5217474102973938, "learning_rate": 0.0005, "loss": 1.30629563331604, "step": 49626 }, { "epoch": 3.912258573117856, "grad_norm": 0.5519298911094666, "learning_rate": 0.0005, "loss": 1.2250876426696777, "step": 49627 }, { "epoch": 3.9123374063854945, "grad_norm": 0.5028573870658875, "learning_rate": 0.0005, "loss": 1.2431650161743164, "step": 49628 }, { "epoch": 3.9124162396531337, "grad_norm": 0.4917716383934021, "learning_rate": 0.0005, "loss": 1.1560760736465454, "step": 49629 }, { "epoch": 3.9124950729207724, "grad_norm": 0.5690214037895203, "learning_rate": 0.0005, "loss": 1.2281312942504883, "step": 49630 }, { "epoch": 3.9125739061884115, "grad_norm": 0.4763772189617157, "learning_rate": 0.0005, "loss": 1.322544813156128, "step": 49631 }, { "epoch": 3.9126527394560506, "grad_norm": 0.5439927577972412, "learning_rate": 0.0005, "loss": 1.3978931903839111, "step": 49632 }, { "epoch": 3.9127315727236893, "grad_norm": 0.45264267921447754, "learning_rate": 0.0005, "loss": 1.0801159143447876, "step": 49633 }, { "epoch": 3.9128104059913285, "grad_norm": 0.45324957370758057, "learning_rate": 0.0005, "loss": 1.1897050142288208, "step": 49634 }, { "epoch": 3.912889239258967, "grad_norm": 0.4266337752342224, "learning_rate": 0.0005, "loss": 1.142883062362671, "step": 49635 }, { "epoch": 3.9129680725266063, "grad_norm": 0.44809532165527344, "learning_rate": 0.0005, "loss": 1.2196848392486572, "step": 49636 }, { "epoch": 3.9130469057942454, "grad_norm": 0.6927857995033264, "learning_rate": 0.0005, "loss": 1.157059669494629, "step": 49637 }, { "epoch": 3.913125739061884, "grad_norm": 0.46325361728668213, "learning_rate": 0.0005, "loss": 1.241647720336914, "step": 49638 }, { "epoch": 3.913204572329523, "grad_norm": 0.40637052059173584, "learning_rate": 0.0005, "loss": 1.1903436183929443, "step": 49639 }, { "epoch": 3.913283405597162, "grad_norm": 0.6144275665283203, "learning_rate": 0.0005, "loss": 1.2945566177368164, "step": 49640 }, { "epoch": 3.913362238864801, "grad_norm": 0.49721449613571167, "learning_rate": 0.0005, "loss": 1.1437681913375854, "step": 49641 }, { "epoch": 3.9134410721324397, "grad_norm": 0.43838217854499817, "learning_rate": 0.0005, "loss": 1.1059558391571045, "step": 49642 }, { "epoch": 3.913519905400079, "grad_norm": 0.491914302110672, "learning_rate": 0.0005, "loss": 1.1124048233032227, "step": 49643 }, { "epoch": 3.9135987386677176, "grad_norm": 0.5271086096763611, "learning_rate": 0.0005, "loss": 1.2227716445922852, "step": 49644 }, { "epoch": 3.9136775719353567, "grad_norm": 0.6695560216903687, "learning_rate": 0.0005, "loss": 1.1250838041305542, "step": 49645 }, { "epoch": 3.913756405202996, "grad_norm": 0.48578524589538574, "learning_rate": 0.0005, "loss": 1.295484185218811, "step": 49646 }, { "epoch": 3.9138352384706345, "grad_norm": 0.47636139392852783, "learning_rate": 0.0005, "loss": 1.0717958211898804, "step": 49647 }, { "epoch": 3.9139140717382737, "grad_norm": 0.554027795791626, "learning_rate": 0.0005, "loss": 1.121101975440979, "step": 49648 }, { "epoch": 3.9139929050059123, "grad_norm": 0.5090941190719604, "learning_rate": 0.0005, "loss": 1.1685068607330322, "step": 49649 }, { "epoch": 3.9140717382735515, "grad_norm": 0.43868452310562134, "learning_rate": 0.0005, "loss": 1.1535316705703735, "step": 49650 }, { "epoch": 3.9141505715411906, "grad_norm": 0.4789111018180847, "learning_rate": 0.0005, "loss": 1.1554169654846191, "step": 49651 }, { "epoch": 3.9142294048088293, "grad_norm": 0.45595237612724304, "learning_rate": 0.0005, "loss": 1.1772812604904175, "step": 49652 }, { "epoch": 3.914308238076468, "grad_norm": 0.4699055254459381, "learning_rate": 0.0005, "loss": 1.174846887588501, "step": 49653 }, { "epoch": 3.914387071344107, "grad_norm": 0.45775601267814636, "learning_rate": 0.0005, "loss": 1.1850764751434326, "step": 49654 }, { "epoch": 3.9144659046117463, "grad_norm": 0.5724921822547913, "learning_rate": 0.0005, "loss": 1.194471001625061, "step": 49655 }, { "epoch": 3.914544737879385, "grad_norm": 0.4135594367980957, "learning_rate": 0.0005, "loss": 1.1057531833648682, "step": 49656 }, { "epoch": 3.914623571147024, "grad_norm": 0.5876252055168152, "learning_rate": 0.0005, "loss": 1.2114626169204712, "step": 49657 }, { "epoch": 3.9147024044146628, "grad_norm": 0.4676445722579956, "learning_rate": 0.0005, "loss": 1.1200363636016846, "step": 49658 }, { "epoch": 3.914781237682302, "grad_norm": 0.5966374278068542, "learning_rate": 0.0005, "loss": 1.0321604013442993, "step": 49659 }, { "epoch": 3.914860070949941, "grad_norm": 0.5547874569892883, "learning_rate": 0.0005, "loss": 1.1856861114501953, "step": 49660 }, { "epoch": 3.9149389042175797, "grad_norm": 0.6093747019767761, "learning_rate": 0.0005, "loss": 1.3209242820739746, "step": 49661 }, { "epoch": 3.915017737485219, "grad_norm": 0.46079403162002563, "learning_rate": 0.0005, "loss": 1.214345932006836, "step": 49662 }, { "epoch": 3.9150965707528576, "grad_norm": 0.539190948009491, "learning_rate": 0.0005, "loss": 1.2703078985214233, "step": 49663 }, { "epoch": 3.9151754040204967, "grad_norm": 0.4794480800628662, "learning_rate": 0.0005, "loss": 1.1276785135269165, "step": 49664 }, { "epoch": 3.915254237288136, "grad_norm": 0.46264901757240295, "learning_rate": 0.0005, "loss": 1.1276063919067383, "step": 49665 }, { "epoch": 3.9153330705557745, "grad_norm": 0.45757409930229187, "learning_rate": 0.0005, "loss": 1.0050562620162964, "step": 49666 }, { "epoch": 3.9154119038234136, "grad_norm": 0.4275573790073395, "learning_rate": 0.0005, "loss": 1.166208267211914, "step": 49667 }, { "epoch": 3.9154907370910523, "grad_norm": 0.4864262342453003, "learning_rate": 0.0005, "loss": 1.060666561126709, "step": 49668 }, { "epoch": 3.9155695703586915, "grad_norm": 0.5819381475448608, "learning_rate": 0.0005, "loss": 1.1881039142608643, "step": 49669 }, { "epoch": 3.91564840362633, "grad_norm": 0.47495681047439575, "learning_rate": 0.0005, "loss": 1.0667232275009155, "step": 49670 }, { "epoch": 3.9157272368939693, "grad_norm": 0.44337087869644165, "learning_rate": 0.0005, "loss": 1.211177945137024, "step": 49671 }, { "epoch": 3.915806070161608, "grad_norm": 0.4914259910583496, "learning_rate": 0.0005, "loss": 1.2246079444885254, "step": 49672 }, { "epoch": 3.915884903429247, "grad_norm": 0.48230409622192383, "learning_rate": 0.0005, "loss": 1.150721549987793, "step": 49673 }, { "epoch": 3.9159637366968862, "grad_norm": 0.47929731011390686, "learning_rate": 0.0005, "loss": 1.1018162965774536, "step": 49674 }, { "epoch": 3.916042569964525, "grad_norm": 0.42285045981407166, "learning_rate": 0.0005, "loss": 1.0981905460357666, "step": 49675 }, { "epoch": 3.916121403232164, "grad_norm": 0.5443219542503357, "learning_rate": 0.0005, "loss": 1.1877902746200562, "step": 49676 }, { "epoch": 3.9162002364998028, "grad_norm": 0.45817384123802185, "learning_rate": 0.0005, "loss": 1.0791728496551514, "step": 49677 }, { "epoch": 3.916279069767442, "grad_norm": 0.5503106117248535, "learning_rate": 0.0005, "loss": 1.1713919639587402, "step": 49678 }, { "epoch": 3.916357903035081, "grad_norm": 0.4541924297809601, "learning_rate": 0.0005, "loss": 1.1280031204223633, "step": 49679 }, { "epoch": 3.9164367363027197, "grad_norm": 0.5489476919174194, "learning_rate": 0.0005, "loss": 1.2504489421844482, "step": 49680 }, { "epoch": 3.916515569570359, "grad_norm": 0.48483923077583313, "learning_rate": 0.0005, "loss": 1.283385157585144, "step": 49681 }, { "epoch": 3.9165944028379975, "grad_norm": 0.5878080129623413, "learning_rate": 0.0005, "loss": 1.308062195777893, "step": 49682 }, { "epoch": 3.9166732361056367, "grad_norm": 0.5255385637283325, "learning_rate": 0.0005, "loss": 1.1173865795135498, "step": 49683 }, { "epoch": 3.9167520693732754, "grad_norm": 0.49822553992271423, "learning_rate": 0.0005, "loss": 1.1539216041564941, "step": 49684 }, { "epoch": 3.9168309026409145, "grad_norm": 0.5789158940315247, "learning_rate": 0.0005, "loss": 1.3845293521881104, "step": 49685 }, { "epoch": 3.916909735908553, "grad_norm": 0.5290094017982483, "learning_rate": 0.0005, "loss": 1.193335771560669, "step": 49686 }, { "epoch": 3.9169885691761923, "grad_norm": 0.617546796798706, "learning_rate": 0.0005, "loss": 1.2132256031036377, "step": 49687 }, { "epoch": 3.9170674024438314, "grad_norm": 0.5231606960296631, "learning_rate": 0.0005, "loss": 1.0930880308151245, "step": 49688 }, { "epoch": 3.91714623571147, "grad_norm": 0.5344551801681519, "learning_rate": 0.0005, "loss": 1.1200451850891113, "step": 49689 }, { "epoch": 3.9172250689791093, "grad_norm": 0.46344372630119324, "learning_rate": 0.0005, "loss": 1.0388028621673584, "step": 49690 }, { "epoch": 3.917303902246748, "grad_norm": 0.5442262887954712, "learning_rate": 0.0005, "loss": 1.050456166267395, "step": 49691 }, { "epoch": 3.917382735514387, "grad_norm": 0.5094702839851379, "learning_rate": 0.0005, "loss": 1.1586023569107056, "step": 49692 }, { "epoch": 3.9174615687820262, "grad_norm": 0.4364018142223358, "learning_rate": 0.0005, "loss": 1.1476078033447266, "step": 49693 }, { "epoch": 3.917540402049665, "grad_norm": 0.4082731604576111, "learning_rate": 0.0005, "loss": 1.0066853761672974, "step": 49694 }, { "epoch": 3.917619235317304, "grad_norm": 0.5171632766723633, "learning_rate": 0.0005, "loss": 1.2153820991516113, "step": 49695 }, { "epoch": 3.9176980685849427, "grad_norm": 0.6337493658065796, "learning_rate": 0.0005, "loss": 0.9741626977920532, "step": 49696 }, { "epoch": 3.917776901852582, "grad_norm": 0.4415329694747925, "learning_rate": 0.0005, "loss": 1.1628053188323975, "step": 49697 }, { "epoch": 3.917855735120221, "grad_norm": 0.5018892288208008, "learning_rate": 0.0005, "loss": 1.1509803533554077, "step": 49698 }, { "epoch": 3.9179345683878597, "grad_norm": 0.45635372400283813, "learning_rate": 0.0005, "loss": 1.2621684074401855, "step": 49699 }, { "epoch": 3.9180134016554984, "grad_norm": 0.6335890889167786, "learning_rate": 0.0005, "loss": 1.3126250505447388, "step": 49700 }, { "epoch": 3.9180922349231375, "grad_norm": 0.5347139239311218, "learning_rate": 0.0005, "loss": 1.1177515983581543, "step": 49701 }, { "epoch": 3.9181710681907767, "grad_norm": 0.553152322769165, "learning_rate": 0.0005, "loss": 1.2313443422317505, "step": 49702 }, { "epoch": 3.9182499014584153, "grad_norm": 0.5439869165420532, "learning_rate": 0.0005, "loss": 1.1669161319732666, "step": 49703 }, { "epoch": 3.9183287347260545, "grad_norm": 0.4729943871498108, "learning_rate": 0.0005, "loss": 1.108825445175171, "step": 49704 }, { "epoch": 3.918407567993693, "grad_norm": 0.5440915822982788, "learning_rate": 0.0005, "loss": 1.2013428211212158, "step": 49705 }, { "epoch": 3.9184864012613323, "grad_norm": 0.536350667476654, "learning_rate": 0.0005, "loss": 1.1458982229232788, "step": 49706 }, { "epoch": 3.9185652345289714, "grad_norm": 0.4246540069580078, "learning_rate": 0.0005, "loss": 1.0633044242858887, "step": 49707 }, { "epoch": 3.91864406779661, "grad_norm": 0.5002719759941101, "learning_rate": 0.0005, "loss": 1.2259256839752197, "step": 49708 }, { "epoch": 3.9187229010642493, "grad_norm": 0.6618209481239319, "learning_rate": 0.0005, "loss": 1.498785376548767, "step": 49709 }, { "epoch": 3.918801734331888, "grad_norm": 0.6071614623069763, "learning_rate": 0.0005, "loss": 1.2867836952209473, "step": 49710 }, { "epoch": 3.918880567599527, "grad_norm": 0.4888388514518738, "learning_rate": 0.0005, "loss": 1.3046185970306396, "step": 49711 }, { "epoch": 3.918959400867166, "grad_norm": 0.4594624936580658, "learning_rate": 0.0005, "loss": 1.201864242553711, "step": 49712 }, { "epoch": 3.919038234134805, "grad_norm": 0.49327290058135986, "learning_rate": 0.0005, "loss": 1.2069566249847412, "step": 49713 }, { "epoch": 3.9191170674024436, "grad_norm": 0.5309906005859375, "learning_rate": 0.0005, "loss": 1.1466598510742188, "step": 49714 }, { "epoch": 3.9191959006700827, "grad_norm": 0.458575040102005, "learning_rate": 0.0005, "loss": 1.1129616498947144, "step": 49715 }, { "epoch": 3.919274733937722, "grad_norm": 0.5388519167900085, "learning_rate": 0.0005, "loss": 1.245128870010376, "step": 49716 }, { "epoch": 3.9193535672053605, "grad_norm": 0.4515550434589386, "learning_rate": 0.0005, "loss": 0.9472509026527405, "step": 49717 }, { "epoch": 3.9194324004729997, "grad_norm": 0.4408688545227051, "learning_rate": 0.0005, "loss": 1.2663205862045288, "step": 49718 }, { "epoch": 3.9195112337406384, "grad_norm": 0.42595112323760986, "learning_rate": 0.0005, "loss": 0.975180983543396, "step": 49719 }, { "epoch": 3.9195900670082775, "grad_norm": 0.5051700472831726, "learning_rate": 0.0005, "loss": 1.1075654029846191, "step": 49720 }, { "epoch": 3.9196689002759166, "grad_norm": 0.4413289725780487, "learning_rate": 0.0005, "loss": 1.055222749710083, "step": 49721 }, { "epoch": 3.9197477335435553, "grad_norm": 0.5290493965148926, "learning_rate": 0.0005, "loss": 0.9737914800643921, "step": 49722 }, { "epoch": 3.9198265668111945, "grad_norm": 0.4667208194732666, "learning_rate": 0.0005, "loss": 1.118807077407837, "step": 49723 }, { "epoch": 3.919905400078833, "grad_norm": 0.7425576448440552, "learning_rate": 0.0005, "loss": 1.4064439535140991, "step": 49724 }, { "epoch": 3.9199842333464723, "grad_norm": 0.44098708033561707, "learning_rate": 0.0005, "loss": 0.9011296033859253, "step": 49725 }, { "epoch": 3.9200630666141114, "grad_norm": 0.4421657621860504, "learning_rate": 0.0005, "loss": 1.1114568710327148, "step": 49726 }, { "epoch": 3.92014189988175, "grad_norm": 0.5117632150650024, "learning_rate": 0.0005, "loss": 1.4007806777954102, "step": 49727 }, { "epoch": 3.920220733149389, "grad_norm": 0.5036799311637878, "learning_rate": 0.0005, "loss": 1.1733050346374512, "step": 49728 }, { "epoch": 3.920299566417028, "grad_norm": 0.420674592256546, "learning_rate": 0.0005, "loss": 1.2084026336669922, "step": 49729 }, { "epoch": 3.920378399684667, "grad_norm": 0.468475878238678, "learning_rate": 0.0005, "loss": 1.1968247890472412, "step": 49730 }, { "epoch": 3.9204572329523057, "grad_norm": 0.40754809975624084, "learning_rate": 0.0005, "loss": 1.04423189163208, "step": 49731 }, { "epoch": 3.920536066219945, "grad_norm": 0.5853050351142883, "learning_rate": 0.0005, "loss": 1.1275838613510132, "step": 49732 }, { "epoch": 3.9206148994875836, "grad_norm": 0.630950391292572, "learning_rate": 0.0005, "loss": 1.342531681060791, "step": 49733 }, { "epoch": 3.9206937327552227, "grad_norm": 0.44332355260849, "learning_rate": 0.0005, "loss": 1.1819795370101929, "step": 49734 }, { "epoch": 3.920772566022862, "grad_norm": 0.4722554087638855, "learning_rate": 0.0005, "loss": 1.1600489616394043, "step": 49735 }, { "epoch": 3.9208513992905005, "grad_norm": 0.37545737624168396, "learning_rate": 0.0005, "loss": 1.0311812162399292, "step": 49736 }, { "epoch": 3.9209302325581397, "grad_norm": 0.46292367577552795, "learning_rate": 0.0005, "loss": 1.027687430381775, "step": 49737 }, { "epoch": 3.9210090658257784, "grad_norm": 0.5297295451164246, "learning_rate": 0.0005, "loss": 1.1530091762542725, "step": 49738 }, { "epoch": 3.9210878990934175, "grad_norm": 0.4368147850036621, "learning_rate": 0.0005, "loss": 1.0645668506622314, "step": 49739 }, { "epoch": 3.9211667323610566, "grad_norm": 0.38271331787109375, "learning_rate": 0.0005, "loss": 1.0080937147140503, "step": 49740 }, { "epoch": 3.9212455656286953, "grad_norm": 0.5678432583808899, "learning_rate": 0.0005, "loss": 1.102187156677246, "step": 49741 }, { "epoch": 3.921324398896334, "grad_norm": 0.4945966601371765, "learning_rate": 0.0005, "loss": 1.055158019065857, "step": 49742 }, { "epoch": 3.921403232163973, "grad_norm": 0.4112599492073059, "learning_rate": 0.0005, "loss": 1.0938187837600708, "step": 49743 }, { "epoch": 3.9214820654316123, "grad_norm": 0.5732801556587219, "learning_rate": 0.0005, "loss": 1.2800490856170654, "step": 49744 }, { "epoch": 3.921560898699251, "grad_norm": 0.43743637204170227, "learning_rate": 0.0005, "loss": 1.1759053468704224, "step": 49745 }, { "epoch": 3.92163973196689, "grad_norm": 0.45241814851760864, "learning_rate": 0.0005, "loss": 1.0296266078948975, "step": 49746 }, { "epoch": 3.9217185652345288, "grad_norm": 0.5005107522010803, "learning_rate": 0.0005, "loss": 1.2204339504241943, "step": 49747 }, { "epoch": 3.921797398502168, "grad_norm": 0.49561432003974915, "learning_rate": 0.0005, "loss": 1.177525281906128, "step": 49748 }, { "epoch": 3.921876231769807, "grad_norm": 0.5813734531402588, "learning_rate": 0.0005, "loss": 1.1532479524612427, "step": 49749 }, { "epoch": 3.9219550650374457, "grad_norm": 0.5859814286231995, "learning_rate": 0.0005, "loss": 1.1666315793991089, "step": 49750 }, { "epoch": 3.922033898305085, "grad_norm": 0.37013116478919983, "learning_rate": 0.0005, "loss": 1.1127532720565796, "step": 49751 }, { "epoch": 3.9221127315727236, "grad_norm": 0.47895219922065735, "learning_rate": 0.0005, "loss": 1.0377910137176514, "step": 49752 }, { "epoch": 3.9221915648403627, "grad_norm": 0.5267103314399719, "learning_rate": 0.0005, "loss": 1.3113024234771729, "step": 49753 }, { "epoch": 3.922270398108002, "grad_norm": 0.4879063069820404, "learning_rate": 0.0005, "loss": 1.2535426616668701, "step": 49754 }, { "epoch": 3.9223492313756405, "grad_norm": 0.4591069221496582, "learning_rate": 0.0005, "loss": 1.1763195991516113, "step": 49755 }, { "epoch": 3.922428064643279, "grad_norm": 0.4958735704421997, "learning_rate": 0.0005, "loss": 1.1821742057800293, "step": 49756 }, { "epoch": 3.9225068979109183, "grad_norm": 0.40384095907211304, "learning_rate": 0.0005, "loss": 1.0899415016174316, "step": 49757 }, { "epoch": 3.9225857311785575, "grad_norm": 0.5577086806297302, "learning_rate": 0.0005, "loss": 1.214095115661621, "step": 49758 }, { "epoch": 3.922664564446196, "grad_norm": 0.5452356934547424, "learning_rate": 0.0005, "loss": 1.3271617889404297, "step": 49759 }, { "epoch": 3.9227433977138353, "grad_norm": 0.4792366921901703, "learning_rate": 0.0005, "loss": 1.1899878978729248, "step": 49760 }, { "epoch": 3.922822230981474, "grad_norm": 0.5120933055877686, "learning_rate": 0.0005, "loss": 1.194370985031128, "step": 49761 }, { "epoch": 3.922901064249113, "grad_norm": 0.5434490442276001, "learning_rate": 0.0005, "loss": 1.1746453046798706, "step": 49762 }, { "epoch": 3.9229798975167522, "grad_norm": 0.5031340718269348, "learning_rate": 0.0005, "loss": 1.1582679748535156, "step": 49763 }, { "epoch": 3.923058730784391, "grad_norm": 0.5113598108291626, "learning_rate": 0.0005, "loss": 1.2051429748535156, "step": 49764 }, { "epoch": 3.92313756405203, "grad_norm": 0.5403612852096558, "learning_rate": 0.0005, "loss": 1.2313323020935059, "step": 49765 }, { "epoch": 3.9232163973196688, "grad_norm": 0.6384851336479187, "learning_rate": 0.0005, "loss": 1.2161846160888672, "step": 49766 }, { "epoch": 3.923295230587308, "grad_norm": 0.5209177136421204, "learning_rate": 0.0005, "loss": 1.0600290298461914, "step": 49767 }, { "epoch": 3.923374063854947, "grad_norm": 0.43052938580513, "learning_rate": 0.0005, "loss": 1.098586082458496, "step": 49768 }, { "epoch": 3.9234528971225857, "grad_norm": 0.49743178486824036, "learning_rate": 0.0005, "loss": 1.3019742965698242, "step": 49769 }, { "epoch": 3.9235317303902244, "grad_norm": 0.44137632846832275, "learning_rate": 0.0005, "loss": 0.9942070245742798, "step": 49770 }, { "epoch": 3.9236105636578635, "grad_norm": 0.5256338715553284, "learning_rate": 0.0005, "loss": 1.12158203125, "step": 49771 }, { "epoch": 3.9236893969255027, "grad_norm": 0.49257996678352356, "learning_rate": 0.0005, "loss": 1.1161723136901855, "step": 49772 }, { "epoch": 3.9237682301931414, "grad_norm": 0.4003743529319763, "learning_rate": 0.0005, "loss": 0.9840723872184753, "step": 49773 }, { "epoch": 3.9238470634607805, "grad_norm": 0.514631450176239, "learning_rate": 0.0005, "loss": 1.1757488250732422, "step": 49774 }, { "epoch": 3.923925896728419, "grad_norm": 0.45348837971687317, "learning_rate": 0.0005, "loss": 1.2413383722305298, "step": 49775 }, { "epoch": 3.9240047299960583, "grad_norm": 0.46866708993911743, "learning_rate": 0.0005, "loss": 1.0765775442123413, "step": 49776 }, { "epoch": 3.9240835632636974, "grad_norm": 0.5394616723060608, "learning_rate": 0.0005, "loss": 1.2443153858184814, "step": 49777 }, { "epoch": 3.924162396531336, "grad_norm": 0.6672347187995911, "learning_rate": 0.0005, "loss": 1.3055793046951294, "step": 49778 }, { "epoch": 3.9242412297989753, "grad_norm": 0.5674892663955688, "learning_rate": 0.0005, "loss": 0.9409857392311096, "step": 49779 }, { "epoch": 3.924320063066614, "grad_norm": 0.5190253257751465, "learning_rate": 0.0005, "loss": 1.3297433853149414, "step": 49780 }, { "epoch": 3.924398896334253, "grad_norm": 0.5574389696121216, "learning_rate": 0.0005, "loss": 1.2249464988708496, "step": 49781 }, { "epoch": 3.9244777296018922, "grad_norm": 0.48177671432495117, "learning_rate": 0.0005, "loss": 1.2157490253448486, "step": 49782 }, { "epoch": 3.924556562869531, "grad_norm": 0.472232848405838, "learning_rate": 0.0005, "loss": 1.1253671646118164, "step": 49783 }, { "epoch": 3.92463539613717, "grad_norm": 0.46465790271759033, "learning_rate": 0.0005, "loss": 1.1310081481933594, "step": 49784 }, { "epoch": 3.9247142294048087, "grad_norm": 0.7575322389602661, "learning_rate": 0.0005, "loss": 1.302632451057434, "step": 49785 }, { "epoch": 3.924793062672448, "grad_norm": 0.5149134397506714, "learning_rate": 0.0005, "loss": 1.3215558528900146, "step": 49786 }, { "epoch": 3.9248718959400866, "grad_norm": 0.4460139870643616, "learning_rate": 0.0005, "loss": 1.10981285572052, "step": 49787 }, { "epoch": 3.9249507292077257, "grad_norm": 0.5038290619850159, "learning_rate": 0.0005, "loss": 1.1617931127548218, "step": 49788 }, { "epoch": 3.9250295624753644, "grad_norm": 0.4262704849243164, "learning_rate": 0.0005, "loss": 1.0718574523925781, "step": 49789 }, { "epoch": 3.9251083957430035, "grad_norm": 0.48880964517593384, "learning_rate": 0.0005, "loss": 1.1493470668792725, "step": 49790 }, { "epoch": 3.9251872290106427, "grad_norm": 0.48393553495407104, "learning_rate": 0.0005, "loss": 1.3189318180084229, "step": 49791 }, { "epoch": 3.9252660622782813, "grad_norm": 0.45575886964797974, "learning_rate": 0.0005, "loss": 1.1421151161193848, "step": 49792 }, { "epoch": 3.9253448955459205, "grad_norm": 0.4307006299495697, "learning_rate": 0.0005, "loss": 1.0950682163238525, "step": 49793 }, { "epoch": 3.925423728813559, "grad_norm": 0.46854138374328613, "learning_rate": 0.0005, "loss": 1.1265214681625366, "step": 49794 }, { "epoch": 3.9255025620811983, "grad_norm": 0.5443025827407837, "learning_rate": 0.0005, "loss": 1.133630394935608, "step": 49795 }, { "epoch": 3.9255813953488374, "grad_norm": 0.42136579751968384, "learning_rate": 0.0005, "loss": 1.045332670211792, "step": 49796 }, { "epoch": 3.925660228616476, "grad_norm": 0.5126659274101257, "learning_rate": 0.0005, "loss": 1.090408444404602, "step": 49797 }, { "epoch": 3.9257390618841153, "grad_norm": 0.8573492169380188, "learning_rate": 0.0005, "loss": 1.3146562576293945, "step": 49798 }, { "epoch": 3.925817895151754, "grad_norm": 0.5261873602867126, "learning_rate": 0.0005, "loss": 1.1089996099472046, "step": 49799 }, { "epoch": 3.925896728419393, "grad_norm": 0.45415475964546204, "learning_rate": 0.0005, "loss": 1.1637070178985596, "step": 49800 }, { "epoch": 3.9259755616870318, "grad_norm": 0.4402151107788086, "learning_rate": 0.0005, "loss": 0.9859649538993835, "step": 49801 }, { "epoch": 3.926054394954671, "grad_norm": 0.5815607309341431, "learning_rate": 0.0005, "loss": 1.2443597316741943, "step": 49802 }, { "epoch": 3.9261332282223096, "grad_norm": 0.4768889248371124, "learning_rate": 0.0005, "loss": 1.1242215633392334, "step": 49803 }, { "epoch": 3.9262120614899487, "grad_norm": 0.5179239511489868, "learning_rate": 0.0005, "loss": 1.0060744285583496, "step": 49804 }, { "epoch": 3.926290894757588, "grad_norm": 0.6059345006942749, "learning_rate": 0.0005, "loss": 1.2716273069381714, "step": 49805 }, { "epoch": 3.9263697280252265, "grad_norm": 0.4332592785358429, "learning_rate": 0.0005, "loss": 1.2507051229476929, "step": 49806 }, { "epoch": 3.9264485612928657, "grad_norm": 0.5959031581878662, "learning_rate": 0.0005, "loss": 1.1558029651641846, "step": 49807 }, { "epoch": 3.9265273945605044, "grad_norm": 0.594727635383606, "learning_rate": 0.0005, "loss": 1.1860005855560303, "step": 49808 }, { "epoch": 3.9266062278281435, "grad_norm": 0.6281999945640564, "learning_rate": 0.0005, "loss": 1.2665741443634033, "step": 49809 }, { "epoch": 3.9266850610957826, "grad_norm": 0.5305668711662292, "learning_rate": 0.0005, "loss": 1.214585781097412, "step": 49810 }, { "epoch": 3.9267638943634213, "grad_norm": 0.3898271918296814, "learning_rate": 0.0005, "loss": 1.1095460653305054, "step": 49811 }, { "epoch": 3.9268427276310605, "grad_norm": 0.5165947079658508, "learning_rate": 0.0005, "loss": 1.1535040140151978, "step": 49812 }, { "epoch": 3.926921560898699, "grad_norm": 0.49638837575912476, "learning_rate": 0.0005, "loss": 0.9969784021377563, "step": 49813 }, { "epoch": 3.9270003941663383, "grad_norm": 0.7121842503547668, "learning_rate": 0.0005, "loss": 1.4029573202133179, "step": 49814 }, { "epoch": 3.9270792274339774, "grad_norm": 2.1339707374572754, "learning_rate": 0.0005, "loss": 1.1107416152954102, "step": 49815 }, { "epoch": 3.927158060701616, "grad_norm": 0.5142391920089722, "learning_rate": 0.0005, "loss": 1.2160431146621704, "step": 49816 }, { "epoch": 3.927236893969255, "grad_norm": 0.5287588238716125, "learning_rate": 0.0005, "loss": 1.275665521621704, "step": 49817 }, { "epoch": 3.927315727236894, "grad_norm": 0.5088074207305908, "learning_rate": 0.0005, "loss": 1.1536282300949097, "step": 49818 }, { "epoch": 3.927394560504533, "grad_norm": 0.5097270607948303, "learning_rate": 0.0005, "loss": 1.2035462856292725, "step": 49819 }, { "epoch": 3.9274733937721718, "grad_norm": 0.4422144889831543, "learning_rate": 0.0005, "loss": 1.3013317584991455, "step": 49820 }, { "epoch": 3.927552227039811, "grad_norm": 0.49147874116897583, "learning_rate": 0.0005, "loss": 1.272665023803711, "step": 49821 }, { "epoch": 3.9276310603074496, "grad_norm": 0.4697980284690857, "learning_rate": 0.0005, "loss": 1.0236644744873047, "step": 49822 }, { "epoch": 3.9277098935750887, "grad_norm": 0.42082834243774414, "learning_rate": 0.0005, "loss": 1.113119125366211, "step": 49823 }, { "epoch": 3.927788726842728, "grad_norm": 0.4281153380870819, "learning_rate": 0.0005, "loss": 1.1255018711090088, "step": 49824 }, { "epoch": 3.9278675601103665, "grad_norm": 0.4820931553840637, "learning_rate": 0.0005, "loss": 1.0764429569244385, "step": 49825 }, { "epoch": 3.9279463933780057, "grad_norm": 0.7553108334541321, "learning_rate": 0.0005, "loss": 1.1536329984664917, "step": 49826 }, { "epoch": 3.9280252266456444, "grad_norm": 0.4844265878200531, "learning_rate": 0.0005, "loss": 1.0923246145248413, "step": 49827 }, { "epoch": 3.9281040599132835, "grad_norm": 0.4685244858264923, "learning_rate": 0.0005, "loss": 1.200730323791504, "step": 49828 }, { "epoch": 3.9281828931809226, "grad_norm": 0.47642064094543457, "learning_rate": 0.0005, "loss": 1.0249933004379272, "step": 49829 }, { "epoch": 3.9282617264485613, "grad_norm": 0.5509851574897766, "learning_rate": 0.0005, "loss": 1.1482174396514893, "step": 49830 }, { "epoch": 3.9283405597162, "grad_norm": 0.5189322233200073, "learning_rate": 0.0005, "loss": 1.158476710319519, "step": 49831 }, { "epoch": 3.928419392983839, "grad_norm": 0.50173020362854, "learning_rate": 0.0005, "loss": 1.3842377662658691, "step": 49832 }, { "epoch": 3.9284982262514783, "grad_norm": 0.551264762878418, "learning_rate": 0.0005, "loss": 1.1189557313919067, "step": 49833 }, { "epoch": 3.928577059519117, "grad_norm": 0.5164138078689575, "learning_rate": 0.0005, "loss": 1.3003287315368652, "step": 49834 }, { "epoch": 3.928655892786756, "grad_norm": 0.5844800472259521, "learning_rate": 0.0005, "loss": 1.218604564666748, "step": 49835 }, { "epoch": 3.9287347260543948, "grad_norm": 0.5834782123565674, "learning_rate": 0.0005, "loss": 1.184608817100525, "step": 49836 }, { "epoch": 3.928813559322034, "grad_norm": 0.4242132902145386, "learning_rate": 0.0005, "loss": 1.0891411304473877, "step": 49837 }, { "epoch": 3.928892392589673, "grad_norm": 0.4970017969608307, "learning_rate": 0.0005, "loss": 1.261182188987732, "step": 49838 }, { "epoch": 3.9289712258573117, "grad_norm": 0.4710235595703125, "learning_rate": 0.0005, "loss": 1.0655853748321533, "step": 49839 }, { "epoch": 3.929050059124951, "grad_norm": 0.4789080023765564, "learning_rate": 0.0005, "loss": 1.14124596118927, "step": 49840 }, { "epoch": 3.9291288923925896, "grad_norm": 0.6135559678077698, "learning_rate": 0.0005, "loss": 1.2360014915466309, "step": 49841 }, { "epoch": 3.9292077256602287, "grad_norm": 0.561242938041687, "learning_rate": 0.0005, "loss": 1.0827043056488037, "step": 49842 }, { "epoch": 3.929286558927868, "grad_norm": 0.35598140954971313, "learning_rate": 0.0005, "loss": 0.9891533851623535, "step": 49843 }, { "epoch": 3.9293653921955065, "grad_norm": 0.48786649107933044, "learning_rate": 0.0005, "loss": 1.292758822441101, "step": 49844 }, { "epoch": 3.929444225463145, "grad_norm": 0.587094783782959, "learning_rate": 0.0005, "loss": 1.3381567001342773, "step": 49845 }, { "epoch": 3.9295230587307843, "grad_norm": 0.475506067276001, "learning_rate": 0.0005, "loss": 1.1879135370254517, "step": 49846 }, { "epoch": 3.9296018919984235, "grad_norm": 0.3979916572570801, "learning_rate": 0.0005, "loss": 1.064760684967041, "step": 49847 }, { "epoch": 3.929680725266062, "grad_norm": 0.519018828868866, "learning_rate": 0.0005, "loss": 1.133285403251648, "step": 49848 }, { "epoch": 3.9297595585337013, "grad_norm": 0.46074753999710083, "learning_rate": 0.0005, "loss": 1.0371525287628174, "step": 49849 }, { "epoch": 3.92983839180134, "grad_norm": 0.5581446886062622, "learning_rate": 0.0005, "loss": 1.4145445823669434, "step": 49850 }, { "epoch": 3.929917225068979, "grad_norm": 0.4503275156021118, "learning_rate": 0.0005, "loss": 1.131792426109314, "step": 49851 }, { "epoch": 3.9299960583366182, "grad_norm": 0.4610750079154968, "learning_rate": 0.0005, "loss": 1.0426969528198242, "step": 49852 }, { "epoch": 3.930074891604257, "grad_norm": 0.4577113687992096, "learning_rate": 0.0005, "loss": 1.07224702835083, "step": 49853 }, { "epoch": 3.930153724871896, "grad_norm": 0.48114684224128723, "learning_rate": 0.0005, "loss": 1.3683366775512695, "step": 49854 }, { "epoch": 3.9302325581395348, "grad_norm": 0.39475131034851074, "learning_rate": 0.0005, "loss": 0.9394752979278564, "step": 49855 }, { "epoch": 3.930311391407174, "grad_norm": 0.48755577206611633, "learning_rate": 0.0005, "loss": 1.2665927410125732, "step": 49856 }, { "epoch": 3.930390224674813, "grad_norm": 0.46257689595222473, "learning_rate": 0.0005, "loss": 1.053498387336731, "step": 49857 }, { "epoch": 3.9304690579424517, "grad_norm": 0.4993780255317688, "learning_rate": 0.0005, "loss": 1.1436759233474731, "step": 49858 }, { "epoch": 3.9305478912100904, "grad_norm": 0.4678974449634552, "learning_rate": 0.0005, "loss": 1.1929566860198975, "step": 49859 }, { "epoch": 3.9306267244777295, "grad_norm": 0.5739867091178894, "learning_rate": 0.0005, "loss": 1.2398717403411865, "step": 49860 }, { "epoch": 3.9307055577453687, "grad_norm": 0.4807068109512329, "learning_rate": 0.0005, "loss": 1.1803394556045532, "step": 49861 }, { "epoch": 3.9307843910130074, "grad_norm": 0.6170065999031067, "learning_rate": 0.0005, "loss": 1.2306199073791504, "step": 49862 }, { "epoch": 3.9308632242806465, "grad_norm": 0.5370935201644897, "learning_rate": 0.0005, "loss": 1.1080209016799927, "step": 49863 }, { "epoch": 3.930942057548285, "grad_norm": 0.4865916669368744, "learning_rate": 0.0005, "loss": 0.9138547778129578, "step": 49864 }, { "epoch": 3.9310208908159243, "grad_norm": 0.4102383553981781, "learning_rate": 0.0005, "loss": 0.959913432598114, "step": 49865 }, { "epoch": 3.9310997240835635, "grad_norm": 0.474551796913147, "learning_rate": 0.0005, "loss": 1.2714178562164307, "step": 49866 }, { "epoch": 3.931178557351202, "grad_norm": 0.4003038704395294, "learning_rate": 0.0005, "loss": 1.048898458480835, "step": 49867 }, { "epoch": 3.9312573906188413, "grad_norm": 0.49287790060043335, "learning_rate": 0.0005, "loss": 1.2270221710205078, "step": 49868 }, { "epoch": 3.93133622388648, "grad_norm": 0.6606847643852234, "learning_rate": 0.0005, "loss": 1.13124680519104, "step": 49869 }, { "epoch": 3.931415057154119, "grad_norm": 0.45667755603790283, "learning_rate": 0.0005, "loss": 0.9898114800453186, "step": 49870 }, { "epoch": 3.9314938904217582, "grad_norm": 0.5399168729782104, "learning_rate": 0.0005, "loss": 1.197721004486084, "step": 49871 }, { "epoch": 3.931572723689397, "grad_norm": 0.49413400888442993, "learning_rate": 0.0005, "loss": 1.2763895988464355, "step": 49872 }, { "epoch": 3.9316515569570356, "grad_norm": 0.4585643708705902, "learning_rate": 0.0005, "loss": 1.1051702499389648, "step": 49873 }, { "epoch": 3.9317303902246747, "grad_norm": 0.43557673692703247, "learning_rate": 0.0005, "loss": 1.066493034362793, "step": 49874 }, { "epoch": 3.931809223492314, "grad_norm": 0.43268656730651855, "learning_rate": 0.0005, "loss": 1.038367509841919, "step": 49875 }, { "epoch": 3.9318880567599526, "grad_norm": 0.5510062575340271, "learning_rate": 0.0005, "loss": 1.150813102722168, "step": 49876 }, { "epoch": 3.9319668900275917, "grad_norm": 0.5363422632217407, "learning_rate": 0.0005, "loss": 1.1110128164291382, "step": 49877 }, { "epoch": 3.9320457232952304, "grad_norm": 0.47237393260002136, "learning_rate": 0.0005, "loss": 1.200165033340454, "step": 49878 }, { "epoch": 3.9321245565628695, "grad_norm": 0.4418925642967224, "learning_rate": 0.0005, "loss": 1.0577867031097412, "step": 49879 }, { "epoch": 3.9322033898305087, "grad_norm": 0.42809319496154785, "learning_rate": 0.0005, "loss": 1.3541655540466309, "step": 49880 }, { "epoch": 3.9322822230981473, "grad_norm": 0.6573085188865662, "learning_rate": 0.0005, "loss": 1.2049176692962646, "step": 49881 }, { "epoch": 3.9323610563657865, "grad_norm": 0.4791502058506012, "learning_rate": 0.0005, "loss": 1.1620275974273682, "step": 49882 }, { "epoch": 3.932439889633425, "grad_norm": 0.38227495551109314, "learning_rate": 0.0005, "loss": 0.9751812219619751, "step": 49883 }, { "epoch": 3.9325187229010643, "grad_norm": 1.6879236698150635, "learning_rate": 0.0005, "loss": 1.3141878843307495, "step": 49884 }, { "epoch": 3.9325975561687034, "grad_norm": 0.4110287129878998, "learning_rate": 0.0005, "loss": 1.1496593952178955, "step": 49885 }, { "epoch": 3.932676389436342, "grad_norm": 0.5758227705955505, "learning_rate": 0.0005, "loss": 1.1824042797088623, "step": 49886 }, { "epoch": 3.932755222703981, "grad_norm": 0.4133371114730835, "learning_rate": 0.0005, "loss": 1.066412091255188, "step": 49887 }, { "epoch": 3.93283405597162, "grad_norm": 0.5050268769264221, "learning_rate": 0.0005, "loss": 1.2602458000183105, "step": 49888 }, { "epoch": 3.932912889239259, "grad_norm": 0.5362100005149841, "learning_rate": 0.0005, "loss": 1.1443612575531006, "step": 49889 }, { "epoch": 3.9329917225068978, "grad_norm": 0.4413721263408661, "learning_rate": 0.0005, "loss": 1.1478183269500732, "step": 49890 }, { "epoch": 3.933070555774537, "grad_norm": 0.44997087121009827, "learning_rate": 0.0005, "loss": 1.0744571685791016, "step": 49891 }, { "epoch": 3.9331493890421756, "grad_norm": 0.5307081937789917, "learning_rate": 0.0005, "loss": 1.1549999713897705, "step": 49892 }, { "epoch": 3.9332282223098147, "grad_norm": 0.4897637367248535, "learning_rate": 0.0005, "loss": 1.3154358863830566, "step": 49893 }, { "epoch": 3.933307055577454, "grad_norm": 0.47043752670288086, "learning_rate": 0.0005, "loss": 1.254673719406128, "step": 49894 }, { "epoch": 3.9333858888450925, "grad_norm": 0.5338598489761353, "learning_rate": 0.0005, "loss": 1.1686599254608154, "step": 49895 }, { "epoch": 3.9334647221127317, "grad_norm": 0.4944215714931488, "learning_rate": 0.0005, "loss": 1.1925034523010254, "step": 49896 }, { "epoch": 3.9335435553803704, "grad_norm": 0.4447678327560425, "learning_rate": 0.0005, "loss": 1.2149590253829956, "step": 49897 }, { "epoch": 3.9336223886480095, "grad_norm": 0.4300987124443054, "learning_rate": 0.0005, "loss": 1.28481125831604, "step": 49898 }, { "epoch": 3.9337012219156486, "grad_norm": 0.5523361563682556, "learning_rate": 0.0005, "loss": 1.209773063659668, "step": 49899 }, { "epoch": 3.9337800551832873, "grad_norm": 0.4699699580669403, "learning_rate": 0.0005, "loss": 1.3430795669555664, "step": 49900 }, { "epoch": 3.933858888450926, "grad_norm": 0.45566269755363464, "learning_rate": 0.0005, "loss": 1.164304494857788, "step": 49901 }, { "epoch": 3.933937721718565, "grad_norm": 0.45706847310066223, "learning_rate": 0.0005, "loss": 1.0496532917022705, "step": 49902 }, { "epoch": 3.9340165549862043, "grad_norm": 0.5533158779144287, "learning_rate": 0.0005, "loss": 0.946955680847168, "step": 49903 }, { "epoch": 3.934095388253843, "grad_norm": 0.4450160562992096, "learning_rate": 0.0005, "loss": 1.065354824066162, "step": 49904 }, { "epoch": 3.934174221521482, "grad_norm": 0.4809654653072357, "learning_rate": 0.0005, "loss": 1.2000224590301514, "step": 49905 }, { "epoch": 3.934253054789121, "grad_norm": 0.4405476450920105, "learning_rate": 0.0005, "loss": 1.0369771718978882, "step": 49906 }, { "epoch": 3.93433188805676, "grad_norm": 0.4314987361431122, "learning_rate": 0.0005, "loss": 1.108295202255249, "step": 49907 }, { "epoch": 3.934410721324399, "grad_norm": 0.4821612238883972, "learning_rate": 0.0005, "loss": 1.2123355865478516, "step": 49908 }, { "epoch": 3.9344895545920378, "grad_norm": 0.673361599445343, "learning_rate": 0.0005, "loss": 1.2935492992401123, "step": 49909 }, { "epoch": 3.934568387859677, "grad_norm": 0.519070029258728, "learning_rate": 0.0005, "loss": 1.2097041606903076, "step": 49910 }, { "epoch": 3.9346472211273156, "grad_norm": 0.49797677993774414, "learning_rate": 0.0005, "loss": 1.2900471687316895, "step": 49911 }, { "epoch": 3.9347260543949547, "grad_norm": 0.38666868209838867, "learning_rate": 0.0005, "loss": 1.039987325668335, "step": 49912 }, { "epoch": 3.934804887662594, "grad_norm": 0.4933875501155853, "learning_rate": 0.0005, "loss": 1.191955804824829, "step": 49913 }, { "epoch": 3.9348837209302325, "grad_norm": 0.5804265737533569, "learning_rate": 0.0005, "loss": 1.1956450939178467, "step": 49914 }, { "epoch": 3.9349625541978717, "grad_norm": 0.5136538147926331, "learning_rate": 0.0005, "loss": 1.2489107847213745, "step": 49915 }, { "epoch": 3.9350413874655104, "grad_norm": 0.42916810512542725, "learning_rate": 0.0005, "loss": 1.1371424198150635, "step": 49916 }, { "epoch": 3.9351202207331495, "grad_norm": 0.5661870241165161, "learning_rate": 0.0005, "loss": 1.2016184329986572, "step": 49917 }, { "epoch": 3.935199054000788, "grad_norm": 0.5518785715103149, "learning_rate": 0.0005, "loss": 1.4407062530517578, "step": 49918 }, { "epoch": 3.9352778872684273, "grad_norm": 0.6537548303604126, "learning_rate": 0.0005, "loss": 1.4928245544433594, "step": 49919 }, { "epoch": 3.935356720536066, "grad_norm": 0.5890014171600342, "learning_rate": 0.0005, "loss": 1.3224186897277832, "step": 49920 }, { "epoch": 3.935435553803705, "grad_norm": 0.4369664490222931, "learning_rate": 0.0005, "loss": 0.9425946474075317, "step": 49921 }, { "epoch": 3.9355143870713443, "grad_norm": 0.4529244601726532, "learning_rate": 0.0005, "loss": 1.1806939840316772, "step": 49922 }, { "epoch": 3.935593220338983, "grad_norm": 0.5997067093849182, "learning_rate": 0.0005, "loss": 1.2693812847137451, "step": 49923 }, { "epoch": 3.935672053606622, "grad_norm": 0.4899592995643616, "learning_rate": 0.0005, "loss": 1.2526183128356934, "step": 49924 }, { "epoch": 3.935750886874261, "grad_norm": 0.4778444170951843, "learning_rate": 0.0005, "loss": 1.1963180303573608, "step": 49925 }, { "epoch": 3.9358297201419, "grad_norm": 0.5666396617889404, "learning_rate": 0.0005, "loss": 1.0740902423858643, "step": 49926 }, { "epoch": 3.935908553409539, "grad_norm": 0.5464386940002441, "learning_rate": 0.0005, "loss": 1.0474363565444946, "step": 49927 }, { "epoch": 3.9359873866771777, "grad_norm": 0.40740326046943665, "learning_rate": 0.0005, "loss": 1.2144134044647217, "step": 49928 }, { "epoch": 3.936066219944817, "grad_norm": 0.42235279083251953, "learning_rate": 0.0005, "loss": 1.0303369760513306, "step": 49929 }, { "epoch": 3.9361450532124556, "grad_norm": 0.5464560389518738, "learning_rate": 0.0005, "loss": 1.2710225582122803, "step": 49930 }, { "epoch": 3.9362238864800947, "grad_norm": 0.5098451375961304, "learning_rate": 0.0005, "loss": 1.1328496932983398, "step": 49931 }, { "epoch": 3.9363027197477334, "grad_norm": 0.4227020740509033, "learning_rate": 0.0005, "loss": 1.1149221658706665, "step": 49932 }, { "epoch": 3.9363815530153725, "grad_norm": 0.776983916759491, "learning_rate": 0.0005, "loss": 1.0343137979507446, "step": 49933 }, { "epoch": 3.936460386283011, "grad_norm": 0.42628756165504456, "learning_rate": 0.0005, "loss": 1.106335163116455, "step": 49934 }, { "epoch": 3.9365392195506503, "grad_norm": 0.5461846590042114, "learning_rate": 0.0005, "loss": 1.2711349725723267, "step": 49935 }, { "epoch": 3.9366180528182895, "grad_norm": 0.4982336163520813, "learning_rate": 0.0005, "loss": 1.1959024667739868, "step": 49936 }, { "epoch": 3.936696886085928, "grad_norm": 0.7148152589797974, "learning_rate": 0.0005, "loss": 1.2683753967285156, "step": 49937 }, { "epoch": 3.9367757193535673, "grad_norm": 0.5524523854255676, "learning_rate": 0.0005, "loss": 1.1090087890625, "step": 49938 }, { "epoch": 3.936854552621206, "grad_norm": 0.4682822525501251, "learning_rate": 0.0005, "loss": 1.013923168182373, "step": 49939 }, { "epoch": 3.936933385888845, "grad_norm": 0.42687949538230896, "learning_rate": 0.0005, "loss": 1.2607300281524658, "step": 49940 }, { "epoch": 3.9370122191564842, "grad_norm": 0.5909714102745056, "learning_rate": 0.0005, "loss": 1.221729040145874, "step": 49941 }, { "epoch": 3.937091052424123, "grad_norm": 0.45599931478500366, "learning_rate": 0.0005, "loss": 1.130634069442749, "step": 49942 }, { "epoch": 3.937169885691762, "grad_norm": 0.4654589593410492, "learning_rate": 0.0005, "loss": 1.1316215991973877, "step": 49943 }, { "epoch": 3.9372487189594008, "grad_norm": 0.47768500447273254, "learning_rate": 0.0005, "loss": 1.0581066608428955, "step": 49944 }, { "epoch": 3.93732755222704, "grad_norm": 0.48337388038635254, "learning_rate": 0.0005, "loss": 1.2270300388336182, "step": 49945 }, { "epoch": 3.937406385494679, "grad_norm": 0.4891377389431, "learning_rate": 0.0005, "loss": 1.359055757522583, "step": 49946 }, { "epoch": 3.9374852187623177, "grad_norm": 0.5994025468826294, "learning_rate": 0.0005, "loss": 1.2466533184051514, "step": 49947 }, { "epoch": 3.9375640520299564, "grad_norm": 0.44222593307495117, "learning_rate": 0.0005, "loss": 1.263749122619629, "step": 49948 }, { "epoch": 3.9376428852975955, "grad_norm": 0.5024328231811523, "learning_rate": 0.0005, "loss": 1.1084001064300537, "step": 49949 }, { "epoch": 3.9377217185652347, "grad_norm": 0.5911855697631836, "learning_rate": 0.0005, "loss": 1.203526496887207, "step": 49950 }, { "epoch": 3.9378005518328734, "grad_norm": 0.6237953901290894, "learning_rate": 0.0005, "loss": 1.0483235120773315, "step": 49951 }, { "epoch": 3.9378793851005125, "grad_norm": 0.6098304986953735, "learning_rate": 0.0005, "loss": 1.1120387315750122, "step": 49952 }, { "epoch": 3.937958218368151, "grad_norm": 0.4244369864463806, "learning_rate": 0.0005, "loss": 1.0993142127990723, "step": 49953 }, { "epoch": 3.9380370516357903, "grad_norm": 0.5419446229934692, "learning_rate": 0.0005, "loss": 1.2783303260803223, "step": 49954 }, { "epoch": 3.9381158849034295, "grad_norm": 0.528712809085846, "learning_rate": 0.0005, "loss": 1.2391831874847412, "step": 49955 }, { "epoch": 3.938194718171068, "grad_norm": 0.4260290563106537, "learning_rate": 0.0005, "loss": 1.127312421798706, "step": 49956 }, { "epoch": 3.9382735514387073, "grad_norm": 0.5200026035308838, "learning_rate": 0.0005, "loss": 1.053628921508789, "step": 49957 }, { "epoch": 3.938352384706346, "grad_norm": 0.6668843030929565, "learning_rate": 0.0005, "loss": 1.2459821701049805, "step": 49958 }, { "epoch": 3.938431217973985, "grad_norm": 0.4580566883087158, "learning_rate": 0.0005, "loss": 1.258825659751892, "step": 49959 }, { "epoch": 3.9385100512416242, "grad_norm": 0.7011830806732178, "learning_rate": 0.0005, "loss": 1.2397139072418213, "step": 49960 }, { "epoch": 3.938588884509263, "grad_norm": 0.437233567237854, "learning_rate": 0.0005, "loss": 1.0049047470092773, "step": 49961 }, { "epoch": 3.9386677177769016, "grad_norm": 0.4702107012271881, "learning_rate": 0.0005, "loss": 1.1990082263946533, "step": 49962 }, { "epoch": 3.9387465510445407, "grad_norm": 0.4854709208011627, "learning_rate": 0.0005, "loss": 1.1502511501312256, "step": 49963 }, { "epoch": 3.93882538431218, "grad_norm": 0.6035057306289673, "learning_rate": 0.0005, "loss": 1.0597224235534668, "step": 49964 }, { "epoch": 3.9389042175798186, "grad_norm": 0.6807461380958557, "learning_rate": 0.0005, "loss": 1.0689549446105957, "step": 49965 }, { "epoch": 3.9389830508474577, "grad_norm": 0.4228886067867279, "learning_rate": 0.0005, "loss": 1.0866564512252808, "step": 49966 }, { "epoch": 3.9390618841150964, "grad_norm": 0.5691691637039185, "learning_rate": 0.0005, "loss": 1.2896579504013062, "step": 49967 }, { "epoch": 3.9391407173827355, "grad_norm": 0.5549725890159607, "learning_rate": 0.0005, "loss": 1.3163357973098755, "step": 49968 }, { "epoch": 3.9392195506503747, "grad_norm": 0.45076537132263184, "learning_rate": 0.0005, "loss": 1.2412655353546143, "step": 49969 }, { "epoch": 3.9392983839180133, "grad_norm": 0.5321987867355347, "learning_rate": 0.0005, "loss": 1.146767258644104, "step": 49970 }, { "epoch": 3.9393772171856525, "grad_norm": 0.4628450274467468, "learning_rate": 0.0005, "loss": 1.312424659729004, "step": 49971 }, { "epoch": 3.939456050453291, "grad_norm": 0.4750869572162628, "learning_rate": 0.0005, "loss": 1.0769233703613281, "step": 49972 }, { "epoch": 3.9395348837209303, "grad_norm": 0.4857912063598633, "learning_rate": 0.0005, "loss": 1.3254590034484863, "step": 49973 }, { "epoch": 3.9396137169885694, "grad_norm": 0.4747438132762909, "learning_rate": 0.0005, "loss": 1.1837258338928223, "step": 49974 }, { "epoch": 3.939692550256208, "grad_norm": 0.473788321018219, "learning_rate": 0.0005, "loss": 1.3263664245605469, "step": 49975 }, { "epoch": 3.939771383523847, "grad_norm": 0.5306981205940247, "learning_rate": 0.0005, "loss": 1.2549006938934326, "step": 49976 }, { "epoch": 3.939850216791486, "grad_norm": 0.5188583731651306, "learning_rate": 0.0005, "loss": 1.2570054531097412, "step": 49977 }, { "epoch": 3.939929050059125, "grad_norm": 0.4957718551158905, "learning_rate": 0.0005, "loss": 1.1191561222076416, "step": 49978 }, { "epoch": 3.9400078833267638, "grad_norm": 0.5018157362937927, "learning_rate": 0.0005, "loss": 1.1214447021484375, "step": 49979 }, { "epoch": 3.940086716594403, "grad_norm": 0.3808740973472595, "learning_rate": 0.0005, "loss": 1.0361403226852417, "step": 49980 }, { "epoch": 3.9401655498620416, "grad_norm": 0.5810708999633789, "learning_rate": 0.0005, "loss": 1.197019100189209, "step": 49981 }, { "epoch": 3.9402443831296807, "grad_norm": 0.7800164818763733, "learning_rate": 0.0005, "loss": 0.9871290922164917, "step": 49982 }, { "epoch": 3.94032321639732, "grad_norm": 0.5801786184310913, "learning_rate": 0.0005, "loss": 1.3048217296600342, "step": 49983 }, { "epoch": 3.9404020496649586, "grad_norm": 0.6347190737724304, "learning_rate": 0.0005, "loss": 1.692422866821289, "step": 49984 }, { "epoch": 3.9404808829325977, "grad_norm": 0.6455024480819702, "learning_rate": 0.0005, "loss": 1.0915956497192383, "step": 49985 }, { "epoch": 3.9405597162002364, "grad_norm": 0.47900286316871643, "learning_rate": 0.0005, "loss": 1.1598353385925293, "step": 49986 }, { "epoch": 3.9406385494678755, "grad_norm": 0.4292586147785187, "learning_rate": 0.0005, "loss": 1.234762191772461, "step": 49987 }, { "epoch": 3.9407173827355146, "grad_norm": 0.5384387373924255, "learning_rate": 0.0005, "loss": 1.0858666896820068, "step": 49988 }, { "epoch": 3.9407962160031533, "grad_norm": 0.6310807466506958, "learning_rate": 0.0005, "loss": 1.1852974891662598, "step": 49989 }, { "epoch": 3.940875049270792, "grad_norm": 0.5403271913528442, "learning_rate": 0.0005, "loss": 1.1528916358947754, "step": 49990 }, { "epoch": 3.940953882538431, "grad_norm": 0.5603945255279541, "learning_rate": 0.0005, "loss": 1.3038464784622192, "step": 49991 }, { "epoch": 3.9410327158060703, "grad_norm": 0.634440541267395, "learning_rate": 0.0005, "loss": 1.3197157382965088, "step": 49992 }, { "epoch": 3.941111549073709, "grad_norm": 0.45799899101257324, "learning_rate": 0.0005, "loss": 0.9986799359321594, "step": 49993 }, { "epoch": 3.941190382341348, "grad_norm": 0.5305697917938232, "learning_rate": 0.0005, "loss": 1.1785966157913208, "step": 49994 }, { "epoch": 3.941269215608987, "grad_norm": 0.6268765330314636, "learning_rate": 0.0005, "loss": 1.1333626508712769, "step": 49995 }, { "epoch": 3.941348048876626, "grad_norm": 0.4893893003463745, "learning_rate": 0.0005, "loss": 1.1004201173782349, "step": 49996 }, { "epoch": 3.941426882144265, "grad_norm": 0.5556302070617676, "learning_rate": 0.0005, "loss": 1.0345661640167236, "step": 49997 }, { "epoch": 3.9415057154119038, "grad_norm": 0.5829690098762512, "learning_rate": 0.0005, "loss": 1.2062643766403198, "step": 49998 }, { "epoch": 3.941584548679543, "grad_norm": 0.4455246329307556, "learning_rate": 0.0005, "loss": 1.0746493339538574, "step": 49999 }, { "epoch": 3.9416633819471816, "grad_norm": 0.48975032567977905, "learning_rate": 0.0005, "loss": 1.0994415283203125, "step": 50000 }, { "epoch": 3.9417422152148207, "grad_norm": 0.42792850732803345, "learning_rate": 0.0005, "loss": 1.1481528282165527, "step": 50001 }, { "epoch": 3.94182104848246, "grad_norm": 0.4980546534061432, "learning_rate": 0.0005, "loss": 1.1579980850219727, "step": 50002 }, { "epoch": 3.9418998817500985, "grad_norm": 0.5021523237228394, "learning_rate": 0.0005, "loss": 1.1613457202911377, "step": 50003 }, { "epoch": 3.9419787150177372, "grad_norm": 0.43633538484573364, "learning_rate": 0.0005, "loss": 1.1674444675445557, "step": 50004 }, { "epoch": 3.9420575482853764, "grad_norm": 0.4854860007762909, "learning_rate": 0.0005, "loss": 1.1321977376937866, "step": 50005 }, { "epoch": 3.9421363815530155, "grad_norm": 0.46836912631988525, "learning_rate": 0.0005, "loss": 1.0804575681686401, "step": 50006 }, { "epoch": 3.942215214820654, "grad_norm": 0.5635683536529541, "learning_rate": 0.0005, "loss": 1.2655593156814575, "step": 50007 }, { "epoch": 3.9422940480882933, "grad_norm": 0.5160862803459167, "learning_rate": 0.0005, "loss": 1.3522464036941528, "step": 50008 }, { "epoch": 3.942372881355932, "grad_norm": 0.4593159258365631, "learning_rate": 0.0005, "loss": 1.1668108701705933, "step": 50009 }, { "epoch": 3.942451714623571, "grad_norm": 0.42220208048820496, "learning_rate": 0.0005, "loss": 1.1689562797546387, "step": 50010 }, { "epoch": 3.9425305478912103, "grad_norm": 0.46768704056739807, "learning_rate": 0.0005, "loss": 1.1867518424987793, "step": 50011 }, { "epoch": 3.942609381158849, "grad_norm": 0.49435731768608093, "learning_rate": 0.0005, "loss": 1.1158809661865234, "step": 50012 }, { "epoch": 3.942688214426488, "grad_norm": 0.4679872989654541, "learning_rate": 0.0005, "loss": 1.1626665592193604, "step": 50013 }, { "epoch": 3.942767047694127, "grad_norm": 0.5239260792732239, "learning_rate": 0.0005, "loss": 1.2332422733306885, "step": 50014 }, { "epoch": 3.942845880961766, "grad_norm": 0.457791805267334, "learning_rate": 0.0005, "loss": 1.1079773902893066, "step": 50015 }, { "epoch": 3.942924714229405, "grad_norm": 0.4611644744873047, "learning_rate": 0.0005, "loss": 1.1779911518096924, "step": 50016 }, { "epoch": 3.9430035474970437, "grad_norm": 0.5233858227729797, "learning_rate": 0.0005, "loss": 1.1871864795684814, "step": 50017 }, { "epoch": 3.9430823807646824, "grad_norm": 0.46202343702316284, "learning_rate": 0.0005, "loss": 1.0988011360168457, "step": 50018 }, { "epoch": 3.9431612140323216, "grad_norm": 0.6104763746261597, "learning_rate": 0.0005, "loss": 1.3303909301757812, "step": 50019 }, { "epoch": 3.9432400472999607, "grad_norm": 0.5787997841835022, "learning_rate": 0.0005, "loss": 1.201046109199524, "step": 50020 }, { "epoch": 3.9433188805675994, "grad_norm": 0.531609296798706, "learning_rate": 0.0005, "loss": 1.1662907600402832, "step": 50021 }, { "epoch": 3.9433977138352385, "grad_norm": 0.45696330070495605, "learning_rate": 0.0005, "loss": 1.0524144172668457, "step": 50022 }, { "epoch": 3.943476547102877, "grad_norm": 0.4859796464443207, "learning_rate": 0.0005, "loss": 1.2153809070587158, "step": 50023 }, { "epoch": 3.9435553803705163, "grad_norm": 0.5433919429779053, "learning_rate": 0.0005, "loss": 1.1082782745361328, "step": 50024 }, { "epoch": 3.9436342136381555, "grad_norm": 0.5631271600723267, "learning_rate": 0.0005, "loss": 1.2344458103179932, "step": 50025 }, { "epoch": 3.943713046905794, "grad_norm": 0.5779350996017456, "learning_rate": 0.0005, "loss": 1.2719242572784424, "step": 50026 }, { "epoch": 3.9437918801734333, "grad_norm": 0.4961092472076416, "learning_rate": 0.0005, "loss": 1.0850889682769775, "step": 50027 }, { "epoch": 3.943870713441072, "grad_norm": 0.6080313920974731, "learning_rate": 0.0005, "loss": 1.232370138168335, "step": 50028 }, { "epoch": 3.943949546708711, "grad_norm": 0.5612693428993225, "learning_rate": 0.0005, "loss": 1.184863805770874, "step": 50029 }, { "epoch": 3.9440283799763503, "grad_norm": 0.48430052399635315, "learning_rate": 0.0005, "loss": 1.1404751539230347, "step": 50030 }, { "epoch": 3.944107213243989, "grad_norm": 0.37917232513427734, "learning_rate": 0.0005, "loss": 1.0137029886245728, "step": 50031 }, { "epoch": 3.9441860465116276, "grad_norm": 0.6646978855133057, "learning_rate": 0.0005, "loss": 1.2644836902618408, "step": 50032 }, { "epoch": 3.9442648797792668, "grad_norm": 0.4573323130607605, "learning_rate": 0.0005, "loss": 1.201408863067627, "step": 50033 }, { "epoch": 3.944343713046906, "grad_norm": 0.5573046207427979, "learning_rate": 0.0005, "loss": 1.348801612854004, "step": 50034 }, { "epoch": 3.9444225463145446, "grad_norm": 0.5220791101455688, "learning_rate": 0.0005, "loss": 1.1886094808578491, "step": 50035 }, { "epoch": 3.9445013795821837, "grad_norm": 0.5763565897941589, "learning_rate": 0.0005, "loss": 1.238224983215332, "step": 50036 }, { "epoch": 3.9445802128498224, "grad_norm": 0.5141699314117432, "learning_rate": 0.0005, "loss": 0.9974133372306824, "step": 50037 }, { "epoch": 3.9446590461174615, "grad_norm": 0.5811238884925842, "learning_rate": 0.0005, "loss": 1.338566780090332, "step": 50038 }, { "epoch": 3.9447378793851007, "grad_norm": 0.6136621236801147, "learning_rate": 0.0005, "loss": 1.1147739887237549, "step": 50039 }, { "epoch": 3.9448167126527394, "grad_norm": 0.49976399540901184, "learning_rate": 0.0005, "loss": 1.1909044981002808, "step": 50040 }, { "epoch": 3.9448955459203785, "grad_norm": 0.48903927206993103, "learning_rate": 0.0005, "loss": 1.2201787233352661, "step": 50041 }, { "epoch": 3.944974379188017, "grad_norm": 0.5193541049957275, "learning_rate": 0.0005, "loss": 1.0535736083984375, "step": 50042 }, { "epoch": 3.9450532124556563, "grad_norm": 0.48707592487335205, "learning_rate": 0.0005, "loss": 1.0200772285461426, "step": 50043 }, { "epoch": 3.9451320457232955, "grad_norm": 0.457156777381897, "learning_rate": 0.0005, "loss": 1.151647686958313, "step": 50044 }, { "epoch": 3.945210878990934, "grad_norm": 0.445987343788147, "learning_rate": 0.0005, "loss": 1.1118695735931396, "step": 50045 }, { "epoch": 3.9452897122585733, "grad_norm": 0.4827037453651428, "learning_rate": 0.0005, "loss": 1.0699748992919922, "step": 50046 }, { "epoch": 3.945368545526212, "grad_norm": 0.42604944109916687, "learning_rate": 0.0005, "loss": 1.0914270877838135, "step": 50047 }, { "epoch": 3.945447378793851, "grad_norm": 0.5968212485313416, "learning_rate": 0.0005, "loss": 1.2835171222686768, "step": 50048 }, { "epoch": 3.94552621206149, "grad_norm": 0.45330265164375305, "learning_rate": 0.0005, "loss": 1.2082103490829468, "step": 50049 }, { "epoch": 3.945605045329129, "grad_norm": 0.47500017285346985, "learning_rate": 0.0005, "loss": 1.3233429193496704, "step": 50050 }, { "epoch": 3.9456838785967676, "grad_norm": 0.5549580454826355, "learning_rate": 0.0005, "loss": 1.14182448387146, "step": 50051 }, { "epoch": 3.9457627118644067, "grad_norm": 0.49382615089416504, "learning_rate": 0.0005, "loss": 1.0815174579620361, "step": 50052 }, { "epoch": 3.945841545132046, "grad_norm": 0.4445064663887024, "learning_rate": 0.0005, "loss": 1.238290548324585, "step": 50053 }, { "epoch": 3.9459203783996846, "grad_norm": 0.5799030065536499, "learning_rate": 0.0005, "loss": 1.2636523246765137, "step": 50054 }, { "epoch": 3.9459992116673237, "grad_norm": 0.4599449634552002, "learning_rate": 0.0005, "loss": 1.1770018339157104, "step": 50055 }, { "epoch": 3.9460780449349624, "grad_norm": 0.4897741973400116, "learning_rate": 0.0005, "loss": 1.12008535861969, "step": 50056 }, { "epoch": 3.9461568782026015, "grad_norm": 0.5294898152351379, "learning_rate": 0.0005, "loss": 1.236608624458313, "step": 50057 }, { "epoch": 3.9462357114702407, "grad_norm": 0.5701397657394409, "learning_rate": 0.0005, "loss": 1.3392258882522583, "step": 50058 }, { "epoch": 3.9463145447378793, "grad_norm": 0.4689931869506836, "learning_rate": 0.0005, "loss": 1.0808196067810059, "step": 50059 }, { "epoch": 3.9463933780055185, "grad_norm": 0.46726930141448975, "learning_rate": 0.0005, "loss": 1.1038455963134766, "step": 50060 }, { "epoch": 3.946472211273157, "grad_norm": 0.4280598759651184, "learning_rate": 0.0005, "loss": 1.0503042936325073, "step": 50061 }, { "epoch": 3.9465510445407963, "grad_norm": 1.962032675743103, "learning_rate": 0.0005, "loss": 1.1652727127075195, "step": 50062 }, { "epoch": 3.946629877808435, "grad_norm": 0.5137327313423157, "learning_rate": 0.0005, "loss": 1.1576941013336182, "step": 50063 }, { "epoch": 3.946708711076074, "grad_norm": 0.5294355154037476, "learning_rate": 0.0005, "loss": 1.1944869756698608, "step": 50064 }, { "epoch": 3.946787544343713, "grad_norm": 0.4795278310775757, "learning_rate": 0.0005, "loss": 0.9053469300270081, "step": 50065 }, { "epoch": 3.946866377611352, "grad_norm": 0.44467276334762573, "learning_rate": 0.0005, "loss": 0.9939923286437988, "step": 50066 }, { "epoch": 3.946945210878991, "grad_norm": 0.3299805819988251, "learning_rate": 0.0005, "loss": 0.9562699198722839, "step": 50067 }, { "epoch": 3.9470240441466298, "grad_norm": 0.4681989252567291, "learning_rate": 0.0005, "loss": 1.0858485698699951, "step": 50068 }, { "epoch": 3.947102877414269, "grad_norm": 0.43025821447372437, "learning_rate": 0.0005, "loss": 1.1619293689727783, "step": 50069 }, { "epoch": 3.9471817106819076, "grad_norm": 0.5289658308029175, "learning_rate": 0.0005, "loss": 1.0850255489349365, "step": 50070 }, { "epoch": 3.9472605439495467, "grad_norm": 0.5914838314056396, "learning_rate": 0.0005, "loss": 1.18534255027771, "step": 50071 }, { "epoch": 3.947339377217186, "grad_norm": 0.4478667080402374, "learning_rate": 0.0005, "loss": 1.000443935394287, "step": 50072 }, { "epoch": 3.9474182104848246, "grad_norm": 0.5075362324714661, "learning_rate": 0.0005, "loss": 1.212800145149231, "step": 50073 }, { "epoch": 3.9474970437524637, "grad_norm": 0.5626605153083801, "learning_rate": 0.0005, "loss": 1.2554292678833008, "step": 50074 }, { "epoch": 3.9475758770201024, "grad_norm": 0.40789467096328735, "learning_rate": 0.0005, "loss": 1.160506248474121, "step": 50075 }, { "epoch": 3.9476547102877415, "grad_norm": 0.4541630446910858, "learning_rate": 0.0005, "loss": 1.0350807905197144, "step": 50076 }, { "epoch": 3.9477335435553806, "grad_norm": 0.4279896020889282, "learning_rate": 0.0005, "loss": 0.9994364976882935, "step": 50077 }, { "epoch": 3.9478123768230193, "grad_norm": 0.41956108808517456, "learning_rate": 0.0005, "loss": 1.0314815044403076, "step": 50078 }, { "epoch": 3.947891210090658, "grad_norm": 0.47605493664741516, "learning_rate": 0.0005, "loss": 1.0534178018569946, "step": 50079 }, { "epoch": 3.947970043358297, "grad_norm": 0.5709447860717773, "learning_rate": 0.0005, "loss": 1.2254314422607422, "step": 50080 }, { "epoch": 3.9480488766259363, "grad_norm": 0.4697982370853424, "learning_rate": 0.0005, "loss": 1.1712605953216553, "step": 50081 }, { "epoch": 3.948127709893575, "grad_norm": 0.5726985931396484, "learning_rate": 0.0005, "loss": 1.3626987934112549, "step": 50082 }, { "epoch": 3.948206543161214, "grad_norm": 0.5273540019989014, "learning_rate": 0.0005, "loss": 1.2204766273498535, "step": 50083 }, { "epoch": 3.948285376428853, "grad_norm": 0.5161325335502625, "learning_rate": 0.0005, "loss": 1.1161246299743652, "step": 50084 }, { "epoch": 3.948364209696492, "grad_norm": 0.4931360185146332, "learning_rate": 0.0005, "loss": 1.2471346855163574, "step": 50085 }, { "epoch": 3.948443042964131, "grad_norm": 0.49851641058921814, "learning_rate": 0.0005, "loss": 1.2843537330627441, "step": 50086 }, { "epoch": 3.9485218762317698, "grad_norm": 0.4988909065723419, "learning_rate": 0.0005, "loss": 1.102156639099121, "step": 50087 }, { "epoch": 3.948600709499409, "grad_norm": 0.40923237800598145, "learning_rate": 0.0005, "loss": 1.1382886171340942, "step": 50088 }, { "epoch": 3.9486795427670476, "grad_norm": 0.8079124689102173, "learning_rate": 0.0005, "loss": 1.3712825775146484, "step": 50089 }, { "epoch": 3.9487583760346867, "grad_norm": 0.5336219072341919, "learning_rate": 0.0005, "loss": 1.1420698165893555, "step": 50090 }, { "epoch": 3.948837209302326, "grad_norm": 0.7258614301681519, "learning_rate": 0.0005, "loss": 1.1361324787139893, "step": 50091 }, { "epoch": 3.9489160425699645, "grad_norm": 0.4624781012535095, "learning_rate": 0.0005, "loss": 1.1844074726104736, "step": 50092 }, { "epoch": 3.9489948758376032, "grad_norm": 0.48358145356178284, "learning_rate": 0.0005, "loss": 1.2023627758026123, "step": 50093 }, { "epoch": 3.9490737091052424, "grad_norm": 0.5033486485481262, "learning_rate": 0.0005, "loss": 1.0434919595718384, "step": 50094 }, { "epoch": 3.9491525423728815, "grad_norm": 0.5106939673423767, "learning_rate": 0.0005, "loss": 1.2343995571136475, "step": 50095 }, { "epoch": 3.94923137564052, "grad_norm": 0.4637133777141571, "learning_rate": 0.0005, "loss": 1.1506459712982178, "step": 50096 }, { "epoch": 3.9493102089081593, "grad_norm": 0.5949987769126892, "learning_rate": 0.0005, "loss": 1.1442174911499023, "step": 50097 }, { "epoch": 3.949389042175798, "grad_norm": 0.4047180712223053, "learning_rate": 0.0005, "loss": 1.0929620265960693, "step": 50098 }, { "epoch": 3.949467875443437, "grad_norm": 0.5775229930877686, "learning_rate": 0.0005, "loss": 1.2041579484939575, "step": 50099 }, { "epoch": 3.9495467087110763, "grad_norm": 0.5532652139663696, "learning_rate": 0.0005, "loss": 1.1657874584197998, "step": 50100 }, { "epoch": 3.949625541978715, "grad_norm": 0.4720781445503235, "learning_rate": 0.0005, "loss": 1.1147855520248413, "step": 50101 }, { "epoch": 3.949704375246354, "grad_norm": 0.5552063584327698, "learning_rate": 0.0005, "loss": 1.1203107833862305, "step": 50102 }, { "epoch": 3.949783208513993, "grad_norm": 0.4725680649280548, "learning_rate": 0.0005, "loss": 1.3196613788604736, "step": 50103 }, { "epoch": 3.949862041781632, "grad_norm": 0.4535403847694397, "learning_rate": 0.0005, "loss": 1.116806149482727, "step": 50104 }, { "epoch": 3.949940875049271, "grad_norm": 0.4164332449436188, "learning_rate": 0.0005, "loss": 1.2739989757537842, "step": 50105 }, { "epoch": 3.9500197083169097, "grad_norm": 0.5683567523956299, "learning_rate": 0.0005, "loss": 1.3141615390777588, "step": 50106 }, { "epoch": 3.9500985415845484, "grad_norm": 0.5511785745620728, "learning_rate": 0.0005, "loss": 1.0253095626831055, "step": 50107 }, { "epoch": 3.9501773748521876, "grad_norm": 0.4279139041900635, "learning_rate": 0.0005, "loss": 1.2005290985107422, "step": 50108 }, { "epoch": 3.9502562081198267, "grad_norm": 0.5051820874214172, "learning_rate": 0.0005, "loss": 1.0548045635223389, "step": 50109 }, { "epoch": 3.9503350413874654, "grad_norm": 0.5359583497047424, "learning_rate": 0.0005, "loss": 1.1494444608688354, "step": 50110 }, { "epoch": 3.9504138746551045, "grad_norm": 0.46443629264831543, "learning_rate": 0.0005, "loss": 1.1102244853973389, "step": 50111 }, { "epoch": 3.950492707922743, "grad_norm": 0.43458330631256104, "learning_rate": 0.0005, "loss": 1.111867070198059, "step": 50112 }, { "epoch": 3.9505715411903823, "grad_norm": 0.5778449773788452, "learning_rate": 0.0005, "loss": 1.2697769403457642, "step": 50113 }, { "epoch": 3.9506503744580215, "grad_norm": 0.6305368542671204, "learning_rate": 0.0005, "loss": 1.4966179132461548, "step": 50114 }, { "epoch": 3.95072920772566, "grad_norm": 0.49593937397003174, "learning_rate": 0.0005, "loss": 1.420586109161377, "step": 50115 }, { "epoch": 3.9508080409932993, "grad_norm": 0.37607046961784363, "learning_rate": 0.0005, "loss": 1.148485541343689, "step": 50116 }, { "epoch": 3.950886874260938, "grad_norm": 0.4794626533985138, "learning_rate": 0.0005, "loss": 1.0617995262145996, "step": 50117 }, { "epoch": 3.950965707528577, "grad_norm": 0.4856070578098297, "learning_rate": 0.0005, "loss": 1.2270071506500244, "step": 50118 }, { "epoch": 3.9510445407962163, "grad_norm": 0.46050378680229187, "learning_rate": 0.0005, "loss": 1.2123596668243408, "step": 50119 }, { "epoch": 3.951123374063855, "grad_norm": 0.4795452058315277, "learning_rate": 0.0005, "loss": 0.9999077916145325, "step": 50120 }, { "epoch": 3.9512022073314936, "grad_norm": 2.087188959121704, "learning_rate": 0.0005, "loss": 1.217543125152588, "step": 50121 }, { "epoch": 3.9512810405991328, "grad_norm": 0.5299537777900696, "learning_rate": 0.0005, "loss": 1.0665112733840942, "step": 50122 }, { "epoch": 3.951359873866772, "grad_norm": 0.4547550082206726, "learning_rate": 0.0005, "loss": 1.1054627895355225, "step": 50123 }, { "epoch": 3.9514387071344106, "grad_norm": 0.4626448154449463, "learning_rate": 0.0005, "loss": 1.146260142326355, "step": 50124 }, { "epoch": 3.9515175404020497, "grad_norm": 0.4656234681606293, "learning_rate": 0.0005, "loss": 1.1779818534851074, "step": 50125 }, { "epoch": 3.9515963736696884, "grad_norm": 0.41527771949768066, "learning_rate": 0.0005, "loss": 1.2292510271072388, "step": 50126 }, { "epoch": 3.9516752069373275, "grad_norm": 0.6670358777046204, "learning_rate": 0.0005, "loss": 1.298696517944336, "step": 50127 }, { "epoch": 3.9517540402049667, "grad_norm": 0.478729248046875, "learning_rate": 0.0005, "loss": 1.188758134841919, "step": 50128 }, { "epoch": 3.9518328734726054, "grad_norm": 0.5701565146446228, "learning_rate": 0.0005, "loss": 1.1538805961608887, "step": 50129 }, { "epoch": 3.9519117067402445, "grad_norm": 0.372613787651062, "learning_rate": 0.0005, "loss": 0.8912619352340698, "step": 50130 }, { "epoch": 3.951990540007883, "grad_norm": 0.5059596300125122, "learning_rate": 0.0005, "loss": 1.147887945175171, "step": 50131 }, { "epoch": 3.9520693732755223, "grad_norm": 0.5702459216117859, "learning_rate": 0.0005, "loss": 1.3508422374725342, "step": 50132 }, { "epoch": 3.9521482065431615, "grad_norm": 0.562793493270874, "learning_rate": 0.0005, "loss": 1.0420165061950684, "step": 50133 }, { "epoch": 3.9522270398108, "grad_norm": 0.5010920166969299, "learning_rate": 0.0005, "loss": 1.1516393423080444, "step": 50134 }, { "epoch": 3.952305873078439, "grad_norm": 0.46005019545555115, "learning_rate": 0.0005, "loss": 1.1527628898620605, "step": 50135 }, { "epoch": 3.952384706346078, "grad_norm": 0.436516672372818, "learning_rate": 0.0005, "loss": 1.3183009624481201, "step": 50136 }, { "epoch": 3.952463539613717, "grad_norm": 0.5639651417732239, "learning_rate": 0.0005, "loss": 1.1062018871307373, "step": 50137 }, { "epoch": 3.952542372881356, "grad_norm": 0.4616072475910187, "learning_rate": 0.0005, "loss": 1.198107361793518, "step": 50138 }, { "epoch": 3.952621206148995, "grad_norm": 0.4740850329399109, "learning_rate": 0.0005, "loss": 1.1788936853408813, "step": 50139 }, { "epoch": 3.9527000394166336, "grad_norm": 0.42669665813446045, "learning_rate": 0.0005, "loss": 1.0938224792480469, "step": 50140 }, { "epoch": 3.9527788726842727, "grad_norm": 0.3666323125362396, "learning_rate": 0.0005, "loss": 1.0368647575378418, "step": 50141 }, { "epoch": 3.952857705951912, "grad_norm": 0.6183136701583862, "learning_rate": 0.0005, "loss": 1.132863163948059, "step": 50142 }, { "epoch": 3.9529365392195506, "grad_norm": 0.3811993896961212, "learning_rate": 0.0005, "loss": 1.1069519519805908, "step": 50143 }, { "epoch": 3.9530153724871897, "grad_norm": 0.5103868842124939, "learning_rate": 0.0005, "loss": 1.1648950576782227, "step": 50144 }, { "epoch": 3.9530942057548284, "grad_norm": 0.42088034749031067, "learning_rate": 0.0005, "loss": 1.0465643405914307, "step": 50145 }, { "epoch": 3.9531730390224675, "grad_norm": 0.43072980642318726, "learning_rate": 0.0005, "loss": 1.1607658863067627, "step": 50146 }, { "epoch": 3.9532518722901067, "grad_norm": 0.5874446034431458, "learning_rate": 0.0005, "loss": 1.1278045177459717, "step": 50147 }, { "epoch": 3.9533307055577454, "grad_norm": 0.46901530027389526, "learning_rate": 0.0005, "loss": 1.1441231966018677, "step": 50148 }, { "epoch": 3.953409538825384, "grad_norm": 0.4783500134944916, "learning_rate": 0.0005, "loss": 1.1987316608428955, "step": 50149 }, { "epoch": 3.953488372093023, "grad_norm": 0.47441795468330383, "learning_rate": 0.0005, "loss": 1.1620652675628662, "step": 50150 }, { "epoch": 3.9535672053606623, "grad_norm": 0.45812514424324036, "learning_rate": 0.0005, "loss": 1.1566691398620605, "step": 50151 }, { "epoch": 3.953646038628301, "grad_norm": 0.5858235955238342, "learning_rate": 0.0005, "loss": 1.1509827375411987, "step": 50152 }, { "epoch": 3.95372487189594, "grad_norm": 0.5673890113830566, "learning_rate": 0.0005, "loss": 1.3313788175582886, "step": 50153 }, { "epoch": 3.953803705163579, "grad_norm": 0.704380214214325, "learning_rate": 0.0005, "loss": 1.193439245223999, "step": 50154 }, { "epoch": 3.953882538431218, "grad_norm": 0.4227246344089508, "learning_rate": 0.0005, "loss": 1.1036064624786377, "step": 50155 }, { "epoch": 3.953961371698857, "grad_norm": 0.6159459352493286, "learning_rate": 0.0005, "loss": 1.04649019241333, "step": 50156 }, { "epoch": 3.9540402049664958, "grad_norm": 0.4763123691082001, "learning_rate": 0.0005, "loss": 1.1187512874603271, "step": 50157 }, { "epoch": 3.954119038234135, "grad_norm": 0.6021109223365784, "learning_rate": 0.0005, "loss": 1.2103490829467773, "step": 50158 }, { "epoch": 3.9541978715017736, "grad_norm": 0.483637273311615, "learning_rate": 0.0005, "loss": 1.2209413051605225, "step": 50159 }, { "epoch": 3.9542767047694127, "grad_norm": 0.45386382937431335, "learning_rate": 0.0005, "loss": 0.972588062286377, "step": 50160 }, { "epoch": 3.954355538037052, "grad_norm": 0.5072377324104309, "learning_rate": 0.0005, "loss": 1.2444546222686768, "step": 50161 }, { "epoch": 3.9544343713046906, "grad_norm": 0.6082572340965271, "learning_rate": 0.0005, "loss": 1.2532085180282593, "step": 50162 }, { "epoch": 3.9545132045723297, "grad_norm": 0.5173163414001465, "learning_rate": 0.0005, "loss": 1.1131114959716797, "step": 50163 }, { "epoch": 3.9545920378399684, "grad_norm": 0.7489020228385925, "learning_rate": 0.0005, "loss": 1.3801486492156982, "step": 50164 }, { "epoch": 3.9546708711076075, "grad_norm": 0.6144259572029114, "learning_rate": 0.0005, "loss": 1.1525651216506958, "step": 50165 }, { "epoch": 3.954749704375246, "grad_norm": 0.4511832892894745, "learning_rate": 0.0005, "loss": 1.1291680335998535, "step": 50166 }, { "epoch": 3.9548285376428853, "grad_norm": 0.5512756109237671, "learning_rate": 0.0005, "loss": 1.181536078453064, "step": 50167 }, { "epoch": 3.954907370910524, "grad_norm": 0.3948555588722229, "learning_rate": 0.0005, "loss": 1.1056678295135498, "step": 50168 }, { "epoch": 3.954986204178163, "grad_norm": 0.6670334935188293, "learning_rate": 0.0005, "loss": 1.257979154586792, "step": 50169 }, { "epoch": 3.9550650374458023, "grad_norm": 0.4361971318721771, "learning_rate": 0.0005, "loss": 1.0148558616638184, "step": 50170 }, { "epoch": 3.955143870713441, "grad_norm": 0.5461517572402954, "learning_rate": 0.0005, "loss": 1.2307946681976318, "step": 50171 }, { "epoch": 3.95522270398108, "grad_norm": 0.5203462839126587, "learning_rate": 0.0005, "loss": 1.1250426769256592, "step": 50172 }, { "epoch": 3.955301537248719, "grad_norm": 0.5139025449752808, "learning_rate": 0.0005, "loss": 1.1893224716186523, "step": 50173 }, { "epoch": 3.955380370516358, "grad_norm": 0.5520864129066467, "learning_rate": 0.0005, "loss": 1.3219146728515625, "step": 50174 }, { "epoch": 3.955459203783997, "grad_norm": 0.47261756658554077, "learning_rate": 0.0005, "loss": 0.9379690289497375, "step": 50175 }, { "epoch": 3.9555380370516358, "grad_norm": 0.5814278721809387, "learning_rate": 0.0005, "loss": 1.3133118152618408, "step": 50176 }, { "epoch": 3.955616870319275, "grad_norm": 0.7420157790184021, "learning_rate": 0.0005, "loss": 1.2290605306625366, "step": 50177 }, { "epoch": 3.9556957035869136, "grad_norm": 0.4532468616962433, "learning_rate": 0.0005, "loss": 1.3309223651885986, "step": 50178 }, { "epoch": 3.9557745368545527, "grad_norm": 0.44838178157806396, "learning_rate": 0.0005, "loss": 1.001967191696167, "step": 50179 }, { "epoch": 3.9558533701221914, "grad_norm": 0.5470463633537292, "learning_rate": 0.0005, "loss": 1.1748313903808594, "step": 50180 }, { "epoch": 3.9559322033898305, "grad_norm": 0.4196792542934418, "learning_rate": 0.0005, "loss": 0.9666571617126465, "step": 50181 }, { "epoch": 3.9560110366574692, "grad_norm": 0.45947664976119995, "learning_rate": 0.0005, "loss": 1.260530948638916, "step": 50182 }, { "epoch": 3.9560898699251084, "grad_norm": 0.5372182130813599, "learning_rate": 0.0005, "loss": 1.2153072357177734, "step": 50183 }, { "epoch": 3.9561687031927475, "grad_norm": 0.5354455709457397, "learning_rate": 0.0005, "loss": 1.058273196220398, "step": 50184 }, { "epoch": 3.956247536460386, "grad_norm": 0.4289713203907013, "learning_rate": 0.0005, "loss": 1.1113438606262207, "step": 50185 }, { "epoch": 3.9563263697280253, "grad_norm": 0.5172604322433472, "learning_rate": 0.0005, "loss": 1.127540111541748, "step": 50186 }, { "epoch": 3.956405202995664, "grad_norm": 0.4087635278701782, "learning_rate": 0.0005, "loss": 1.093595266342163, "step": 50187 }, { "epoch": 3.956484036263303, "grad_norm": 0.5082496404647827, "learning_rate": 0.0005, "loss": 1.464233636856079, "step": 50188 }, { "epoch": 3.9565628695309423, "grad_norm": 0.5129929184913635, "learning_rate": 0.0005, "loss": 1.2307192087173462, "step": 50189 }, { "epoch": 3.956641702798581, "grad_norm": 0.48094844818115234, "learning_rate": 0.0005, "loss": 1.1687617301940918, "step": 50190 }, { "epoch": 3.95672053606622, "grad_norm": 0.5392653942108154, "learning_rate": 0.0005, "loss": 1.2591464519500732, "step": 50191 }, { "epoch": 3.956799369333859, "grad_norm": 0.44937971234321594, "learning_rate": 0.0005, "loss": 1.2349019050598145, "step": 50192 }, { "epoch": 3.956878202601498, "grad_norm": 0.5830679535865784, "learning_rate": 0.0005, "loss": 1.0650819540023804, "step": 50193 }, { "epoch": 3.956957035869137, "grad_norm": 0.6142694354057312, "learning_rate": 0.0005, "loss": 1.2522598505020142, "step": 50194 }, { "epoch": 3.9570358691367757, "grad_norm": 0.4924059808254242, "learning_rate": 0.0005, "loss": 1.2290763854980469, "step": 50195 }, { "epoch": 3.9571147024044144, "grad_norm": 0.6344268321990967, "learning_rate": 0.0005, "loss": 1.2981395721435547, "step": 50196 }, { "epoch": 3.9571935356720536, "grad_norm": 0.4750433564186096, "learning_rate": 0.0005, "loss": 1.2204835414886475, "step": 50197 }, { "epoch": 3.9572723689396927, "grad_norm": 0.5990437269210815, "learning_rate": 0.0005, "loss": 1.427823543548584, "step": 50198 }, { "epoch": 3.9573512022073314, "grad_norm": 0.48880359530448914, "learning_rate": 0.0005, "loss": 1.2084660530090332, "step": 50199 }, { "epoch": 3.9574300354749705, "grad_norm": 1.2176529169082642, "learning_rate": 0.0005, "loss": 1.1594351530075073, "step": 50200 }, { "epoch": 3.957508868742609, "grad_norm": 0.4452684819698334, "learning_rate": 0.0005, "loss": 1.1052559614181519, "step": 50201 }, { "epoch": 3.9575877020102483, "grad_norm": 0.5248427391052246, "learning_rate": 0.0005, "loss": 1.2228224277496338, "step": 50202 }, { "epoch": 3.9576665352778875, "grad_norm": 0.7173711657524109, "learning_rate": 0.0005, "loss": 1.3107744455337524, "step": 50203 }, { "epoch": 3.957745368545526, "grad_norm": 0.5517319440841675, "learning_rate": 0.0005, "loss": 1.1983634233474731, "step": 50204 }, { "epoch": 3.9578242018131653, "grad_norm": 0.4558318257331848, "learning_rate": 0.0005, "loss": 1.097678780555725, "step": 50205 }, { "epoch": 3.957903035080804, "grad_norm": 0.6041706204414368, "learning_rate": 0.0005, "loss": 1.3419272899627686, "step": 50206 }, { "epoch": 3.957981868348443, "grad_norm": 0.5398641228675842, "learning_rate": 0.0005, "loss": 1.1966967582702637, "step": 50207 }, { "epoch": 3.9580607016160823, "grad_norm": 0.6044514179229736, "learning_rate": 0.0005, "loss": 1.3158302307128906, "step": 50208 }, { "epoch": 3.958139534883721, "grad_norm": 0.4355448782444, "learning_rate": 0.0005, "loss": 1.05183744430542, "step": 50209 }, { "epoch": 3.9582183681513596, "grad_norm": 0.5249333381652832, "learning_rate": 0.0005, "loss": 1.0283528566360474, "step": 50210 }, { "epoch": 3.9582972014189988, "grad_norm": 0.5624526739120483, "learning_rate": 0.0005, "loss": 1.1519814729690552, "step": 50211 }, { "epoch": 3.958376034686638, "grad_norm": 0.5480531454086304, "learning_rate": 0.0005, "loss": 1.4631526470184326, "step": 50212 }, { "epoch": 3.9584548679542766, "grad_norm": 0.47638094425201416, "learning_rate": 0.0005, "loss": 1.1118690967559814, "step": 50213 }, { "epoch": 3.9585337012219157, "grad_norm": 0.5138869285583496, "learning_rate": 0.0005, "loss": 1.3633641004562378, "step": 50214 }, { "epoch": 3.9586125344895544, "grad_norm": 0.4918670952320099, "learning_rate": 0.0005, "loss": 1.135424017906189, "step": 50215 }, { "epoch": 3.9586913677571935, "grad_norm": 0.5069631338119507, "learning_rate": 0.0005, "loss": 1.1046652793884277, "step": 50216 }, { "epoch": 3.9587702010248327, "grad_norm": 0.4615762233734131, "learning_rate": 0.0005, "loss": 1.1951179504394531, "step": 50217 }, { "epoch": 3.9588490342924714, "grad_norm": 0.529606819152832, "learning_rate": 0.0005, "loss": 1.233083963394165, "step": 50218 }, { "epoch": 3.9589278675601105, "grad_norm": 0.6223921775817871, "learning_rate": 0.0005, "loss": 1.206930160522461, "step": 50219 }, { "epoch": 3.959006700827749, "grad_norm": 0.47683894634246826, "learning_rate": 0.0005, "loss": 1.1471889019012451, "step": 50220 }, { "epoch": 3.9590855340953883, "grad_norm": 0.6569611430168152, "learning_rate": 0.0005, "loss": 1.509981393814087, "step": 50221 }, { "epoch": 3.9591643673630275, "grad_norm": 0.45990946888923645, "learning_rate": 0.0005, "loss": 0.9655932784080505, "step": 50222 }, { "epoch": 3.959243200630666, "grad_norm": 0.48677176237106323, "learning_rate": 0.0005, "loss": 1.0783841609954834, "step": 50223 }, { "epoch": 3.959322033898305, "grad_norm": 0.45937177538871765, "learning_rate": 0.0005, "loss": 1.123725175857544, "step": 50224 }, { "epoch": 3.959400867165944, "grad_norm": 0.7945189476013184, "learning_rate": 0.0005, "loss": 1.3117210865020752, "step": 50225 }, { "epoch": 3.959479700433583, "grad_norm": 0.5160386562347412, "learning_rate": 0.0005, "loss": 1.2801079750061035, "step": 50226 }, { "epoch": 3.959558533701222, "grad_norm": 0.46157747507095337, "learning_rate": 0.0005, "loss": 1.1062132120132446, "step": 50227 }, { "epoch": 3.959637366968861, "grad_norm": 0.549268364906311, "learning_rate": 0.0005, "loss": 1.0885581970214844, "step": 50228 }, { "epoch": 3.9597162002364996, "grad_norm": 0.42408788204193115, "learning_rate": 0.0005, "loss": 1.1149249076843262, "step": 50229 }, { "epoch": 3.9597950335041388, "grad_norm": 0.47447669506073, "learning_rate": 0.0005, "loss": 1.113013505935669, "step": 50230 }, { "epoch": 3.959873866771778, "grad_norm": 0.4267350733280182, "learning_rate": 0.0005, "loss": 1.0798474550247192, "step": 50231 }, { "epoch": 3.9599527000394166, "grad_norm": 0.4877256155014038, "learning_rate": 0.0005, "loss": 1.171411395072937, "step": 50232 }, { "epoch": 3.9600315333070557, "grad_norm": 0.5155436396598816, "learning_rate": 0.0005, "loss": 1.2271603345870972, "step": 50233 }, { "epoch": 3.9601103665746944, "grad_norm": 0.5214717388153076, "learning_rate": 0.0005, "loss": 1.091660499572754, "step": 50234 }, { "epoch": 3.9601891998423335, "grad_norm": 0.47185030579566956, "learning_rate": 0.0005, "loss": 1.161354899406433, "step": 50235 }, { "epoch": 3.9602680331099727, "grad_norm": 0.47738543152809143, "learning_rate": 0.0005, "loss": 1.0322365760803223, "step": 50236 }, { "epoch": 3.9603468663776114, "grad_norm": 0.5838833451271057, "learning_rate": 0.0005, "loss": 1.2726311683654785, "step": 50237 }, { "epoch": 3.96042569964525, "grad_norm": 0.5908620357513428, "learning_rate": 0.0005, "loss": 1.4667282104492188, "step": 50238 }, { "epoch": 3.960504532912889, "grad_norm": 0.48791760206222534, "learning_rate": 0.0005, "loss": 1.2331764698028564, "step": 50239 }, { "epoch": 3.9605833661805283, "grad_norm": 0.4362402856349945, "learning_rate": 0.0005, "loss": 1.0849844217300415, "step": 50240 }, { "epoch": 3.960662199448167, "grad_norm": 0.4465467631816864, "learning_rate": 0.0005, "loss": 1.1820825338363647, "step": 50241 }, { "epoch": 3.960741032715806, "grad_norm": 0.43884381651878357, "learning_rate": 0.0005, "loss": 1.1141152381896973, "step": 50242 }, { "epoch": 3.960819865983445, "grad_norm": 0.44306278228759766, "learning_rate": 0.0005, "loss": 1.0732206106185913, "step": 50243 }, { "epoch": 3.960898699251084, "grad_norm": 0.5824589133262634, "learning_rate": 0.0005, "loss": 1.1836249828338623, "step": 50244 }, { "epoch": 3.960977532518723, "grad_norm": 0.45281219482421875, "learning_rate": 0.0005, "loss": 1.0273433923721313, "step": 50245 }, { "epoch": 3.9610563657863618, "grad_norm": 0.46241381764411926, "learning_rate": 0.0005, "loss": 1.0279464721679688, "step": 50246 }, { "epoch": 3.961135199054001, "grad_norm": 0.5196055769920349, "learning_rate": 0.0005, "loss": 1.1320476531982422, "step": 50247 }, { "epoch": 3.9612140323216396, "grad_norm": 0.5599371790885925, "learning_rate": 0.0005, "loss": 1.222923994064331, "step": 50248 }, { "epoch": 3.9612928655892787, "grad_norm": 0.5623029470443726, "learning_rate": 0.0005, "loss": 1.25925874710083, "step": 50249 }, { "epoch": 3.961371698856918, "grad_norm": 0.5300774574279785, "learning_rate": 0.0005, "loss": 1.2434977293014526, "step": 50250 }, { "epoch": 3.9614505321245566, "grad_norm": 0.43838390707969666, "learning_rate": 0.0005, "loss": 1.1465035676956177, "step": 50251 }, { "epoch": 3.9615293653921952, "grad_norm": 0.48090660572052, "learning_rate": 0.0005, "loss": 1.110054612159729, "step": 50252 }, { "epoch": 3.9616081986598344, "grad_norm": 0.492124080657959, "learning_rate": 0.0005, "loss": 1.1637637615203857, "step": 50253 }, { "epoch": 3.9616870319274735, "grad_norm": 0.4595007002353668, "learning_rate": 0.0005, "loss": 1.053180456161499, "step": 50254 }, { "epoch": 3.961765865195112, "grad_norm": 0.4631767272949219, "learning_rate": 0.0005, "loss": 1.135481595993042, "step": 50255 }, { "epoch": 3.9618446984627513, "grad_norm": 0.4176309108734131, "learning_rate": 0.0005, "loss": 0.9228665828704834, "step": 50256 }, { "epoch": 3.96192353173039, "grad_norm": 0.5114122033119202, "learning_rate": 0.0005, "loss": 1.2558104991912842, "step": 50257 }, { "epoch": 3.962002364998029, "grad_norm": 0.5576130747795105, "learning_rate": 0.0005, "loss": 1.239181637763977, "step": 50258 }, { "epoch": 3.9620811982656683, "grad_norm": 0.5226593613624573, "learning_rate": 0.0005, "loss": 1.1726858615875244, "step": 50259 }, { "epoch": 3.962160031533307, "grad_norm": 0.5573690533638, "learning_rate": 0.0005, "loss": 1.3976961374282837, "step": 50260 }, { "epoch": 3.962238864800946, "grad_norm": 0.4295525550842285, "learning_rate": 0.0005, "loss": 0.9496406316757202, "step": 50261 }, { "epoch": 3.962317698068585, "grad_norm": 0.4705224931240082, "learning_rate": 0.0005, "loss": 1.2054905891418457, "step": 50262 }, { "epoch": 3.962396531336224, "grad_norm": 0.4102146029472351, "learning_rate": 0.0005, "loss": 0.9673647284507751, "step": 50263 }, { "epoch": 3.962475364603863, "grad_norm": 0.4400535821914673, "learning_rate": 0.0005, "loss": 1.1640830039978027, "step": 50264 }, { "epoch": 3.9625541978715018, "grad_norm": 0.5722967386245728, "learning_rate": 0.0005, "loss": 1.0524743795394897, "step": 50265 }, { "epoch": 3.9626330311391404, "grad_norm": 0.49235987663269043, "learning_rate": 0.0005, "loss": 1.3185579776763916, "step": 50266 }, { "epoch": 3.9627118644067796, "grad_norm": 0.5996416211128235, "learning_rate": 0.0005, "loss": 1.2757670879364014, "step": 50267 }, { "epoch": 3.9627906976744187, "grad_norm": 0.5246992707252502, "learning_rate": 0.0005, "loss": 1.032794713973999, "step": 50268 }, { "epoch": 3.9628695309420574, "grad_norm": 0.595892071723938, "learning_rate": 0.0005, "loss": 1.1027119159698486, "step": 50269 }, { "epoch": 3.9629483642096965, "grad_norm": 0.442043662071228, "learning_rate": 0.0005, "loss": 1.01560640335083, "step": 50270 }, { "epoch": 3.9630271974773352, "grad_norm": 0.7108312845230103, "learning_rate": 0.0005, "loss": 1.1489837169647217, "step": 50271 }, { "epoch": 3.9631060307449744, "grad_norm": 0.5870974063873291, "learning_rate": 0.0005, "loss": 1.2698471546173096, "step": 50272 }, { "epoch": 3.9631848640126135, "grad_norm": 0.9246363639831543, "learning_rate": 0.0005, "loss": 1.251799464225769, "step": 50273 }, { "epoch": 3.963263697280252, "grad_norm": 0.5752649903297424, "learning_rate": 0.0005, "loss": 1.0619683265686035, "step": 50274 }, { "epoch": 3.9633425305478913, "grad_norm": 0.5940147638320923, "learning_rate": 0.0005, "loss": 1.3395440578460693, "step": 50275 }, { "epoch": 3.96342136381553, "grad_norm": 0.5813341736793518, "learning_rate": 0.0005, "loss": 1.2226743698120117, "step": 50276 }, { "epoch": 3.963500197083169, "grad_norm": 0.5612893104553223, "learning_rate": 0.0005, "loss": 1.216578722000122, "step": 50277 }, { "epoch": 3.9635790303508083, "grad_norm": 0.6045160293579102, "learning_rate": 0.0005, "loss": 1.080417513847351, "step": 50278 }, { "epoch": 3.963657863618447, "grad_norm": 0.463683545589447, "learning_rate": 0.0005, "loss": 1.2081146240234375, "step": 50279 }, { "epoch": 3.9637366968860857, "grad_norm": 0.4734458327293396, "learning_rate": 0.0005, "loss": 1.0864644050598145, "step": 50280 }, { "epoch": 3.963815530153725, "grad_norm": 0.4811301529407501, "learning_rate": 0.0005, "loss": 1.1072461605072021, "step": 50281 }, { "epoch": 3.963894363421364, "grad_norm": 0.6567708253860474, "learning_rate": 0.0005, "loss": 1.235920786857605, "step": 50282 }, { "epoch": 3.9639731966890026, "grad_norm": 0.5652869939804077, "learning_rate": 0.0005, "loss": 1.313368320465088, "step": 50283 }, { "epoch": 3.9640520299566417, "grad_norm": 0.46214956045150757, "learning_rate": 0.0005, "loss": 1.213728904724121, "step": 50284 }, { "epoch": 3.9641308632242804, "grad_norm": 0.6390186548233032, "learning_rate": 0.0005, "loss": 1.1773669719696045, "step": 50285 }, { "epoch": 3.9642096964919196, "grad_norm": 0.4623703360557556, "learning_rate": 0.0005, "loss": 1.0021003484725952, "step": 50286 }, { "epoch": 3.9642885297595587, "grad_norm": 0.6922083497047424, "learning_rate": 0.0005, "loss": 1.3687760829925537, "step": 50287 }, { "epoch": 3.9643673630271974, "grad_norm": 0.5015870332717896, "learning_rate": 0.0005, "loss": 1.1027055978775024, "step": 50288 }, { "epoch": 3.9644461962948365, "grad_norm": 0.5657891631126404, "learning_rate": 0.0005, "loss": 1.0834673643112183, "step": 50289 }, { "epoch": 3.964525029562475, "grad_norm": 0.8038915395736694, "learning_rate": 0.0005, "loss": 1.1591709852218628, "step": 50290 }, { "epoch": 3.9646038628301143, "grad_norm": 0.5260077714920044, "learning_rate": 0.0005, "loss": 1.231061577796936, "step": 50291 }, { "epoch": 3.9646826960977535, "grad_norm": 0.4981551766395569, "learning_rate": 0.0005, "loss": 1.1003468036651611, "step": 50292 }, { "epoch": 3.964761529365392, "grad_norm": 0.5030493140220642, "learning_rate": 0.0005, "loss": 1.1695642471313477, "step": 50293 }, { "epoch": 3.9648403626330313, "grad_norm": 0.49647951126098633, "learning_rate": 0.0005, "loss": 1.2344224452972412, "step": 50294 }, { "epoch": 3.96491919590067, "grad_norm": 0.35951268672943115, "learning_rate": 0.0005, "loss": 0.9876766204833984, "step": 50295 }, { "epoch": 3.964998029168309, "grad_norm": 0.5562493801116943, "learning_rate": 0.0005, "loss": 1.249507188796997, "step": 50296 }, { "epoch": 3.965076862435948, "grad_norm": 0.48681479692459106, "learning_rate": 0.0005, "loss": 1.3634371757507324, "step": 50297 }, { "epoch": 3.965155695703587, "grad_norm": 0.6147134304046631, "learning_rate": 0.0005, "loss": 1.2474137544631958, "step": 50298 }, { "epoch": 3.9652345289712256, "grad_norm": 0.45828041434288025, "learning_rate": 0.0005, "loss": 1.131360650062561, "step": 50299 }, { "epoch": 3.9653133622388648, "grad_norm": 0.5465063452720642, "learning_rate": 0.0005, "loss": 1.0933818817138672, "step": 50300 }, { "epoch": 3.965392195506504, "grad_norm": 0.516889214515686, "learning_rate": 0.0005, "loss": 1.0839545726776123, "step": 50301 }, { "epoch": 3.9654710287741426, "grad_norm": 0.7672098278999329, "learning_rate": 0.0005, "loss": 1.4551997184753418, "step": 50302 }, { "epoch": 3.9655498620417817, "grad_norm": 0.471020370721817, "learning_rate": 0.0005, "loss": 1.0726958513259888, "step": 50303 }, { "epoch": 3.9656286953094204, "grad_norm": 0.51308673620224, "learning_rate": 0.0005, "loss": 1.131263017654419, "step": 50304 }, { "epoch": 3.9657075285770595, "grad_norm": 0.48009929060935974, "learning_rate": 0.0005, "loss": 1.1414778232574463, "step": 50305 }, { "epoch": 3.9657863618446987, "grad_norm": 0.5565695762634277, "learning_rate": 0.0005, "loss": 1.4381365776062012, "step": 50306 }, { "epoch": 3.9658651951123374, "grad_norm": 0.5231583118438721, "learning_rate": 0.0005, "loss": 1.2070759534835815, "step": 50307 }, { "epoch": 3.9659440283799765, "grad_norm": 0.4622359871864319, "learning_rate": 0.0005, "loss": 1.1002750396728516, "step": 50308 }, { "epoch": 3.966022861647615, "grad_norm": 0.49009713530540466, "learning_rate": 0.0005, "loss": 1.2754677534103394, "step": 50309 }, { "epoch": 3.9661016949152543, "grad_norm": 0.49757611751556396, "learning_rate": 0.0005, "loss": 1.1465415954589844, "step": 50310 }, { "epoch": 3.966180528182893, "grad_norm": 0.49814289808273315, "learning_rate": 0.0005, "loss": 1.1538727283477783, "step": 50311 }, { "epoch": 3.966259361450532, "grad_norm": 0.4190712869167328, "learning_rate": 0.0005, "loss": 1.0443036556243896, "step": 50312 }, { "epoch": 3.966338194718171, "grad_norm": 0.5334557890892029, "learning_rate": 0.0005, "loss": 1.1049413681030273, "step": 50313 }, { "epoch": 3.96641702798581, "grad_norm": 0.4790773391723633, "learning_rate": 0.0005, "loss": 1.1394318342208862, "step": 50314 }, { "epoch": 3.966495861253449, "grad_norm": 0.47532930970191956, "learning_rate": 0.0005, "loss": 1.1745474338531494, "step": 50315 }, { "epoch": 3.966574694521088, "grad_norm": 0.4703425168991089, "learning_rate": 0.0005, "loss": 1.1642634868621826, "step": 50316 }, { "epoch": 3.966653527788727, "grad_norm": 0.5817615985870361, "learning_rate": 0.0005, "loss": 1.1233352422714233, "step": 50317 }, { "epoch": 3.9667323610563656, "grad_norm": 0.6453964710235596, "learning_rate": 0.0005, "loss": 1.175074815750122, "step": 50318 }, { "epoch": 3.9668111943240048, "grad_norm": 0.5242921113967896, "learning_rate": 0.0005, "loss": 1.2686001062393188, "step": 50319 }, { "epoch": 3.966890027591644, "grad_norm": 0.5669052004814148, "learning_rate": 0.0005, "loss": 1.1877559423446655, "step": 50320 }, { "epoch": 3.9669688608592826, "grad_norm": 0.4464530050754547, "learning_rate": 0.0005, "loss": 1.0699849128723145, "step": 50321 }, { "epoch": 3.9670476941269217, "grad_norm": 0.4290454685688019, "learning_rate": 0.0005, "loss": 1.0845941305160522, "step": 50322 }, { "epoch": 3.9671265273945604, "grad_norm": 0.4299280047416687, "learning_rate": 0.0005, "loss": 1.1794865131378174, "step": 50323 }, { "epoch": 3.9672053606621995, "grad_norm": 0.42764192819595337, "learning_rate": 0.0005, "loss": 1.0152171850204468, "step": 50324 }, { "epoch": 3.9672841939298387, "grad_norm": 0.501339852809906, "learning_rate": 0.0005, "loss": 1.221116542816162, "step": 50325 }, { "epoch": 3.9673630271974774, "grad_norm": 0.5331674218177795, "learning_rate": 0.0005, "loss": 1.2084972858428955, "step": 50326 }, { "epoch": 3.967441860465116, "grad_norm": 0.38926368951797485, "learning_rate": 0.0005, "loss": 0.9339737892150879, "step": 50327 }, { "epoch": 3.967520693732755, "grad_norm": 0.43728315830230713, "learning_rate": 0.0005, "loss": 1.189435601234436, "step": 50328 }, { "epoch": 3.9675995270003943, "grad_norm": 0.41472652554512024, "learning_rate": 0.0005, "loss": 1.1436153650283813, "step": 50329 }, { "epoch": 3.967678360268033, "grad_norm": 0.4676056504249573, "learning_rate": 0.0005, "loss": 1.132929801940918, "step": 50330 }, { "epoch": 3.967757193535672, "grad_norm": 0.47822749614715576, "learning_rate": 0.0005, "loss": 1.144834280014038, "step": 50331 }, { "epoch": 3.967836026803311, "grad_norm": 0.4781560003757477, "learning_rate": 0.0005, "loss": 1.2054775953292847, "step": 50332 }, { "epoch": 3.96791486007095, "grad_norm": 0.4448983371257782, "learning_rate": 0.0005, "loss": 1.2580928802490234, "step": 50333 }, { "epoch": 3.967993693338589, "grad_norm": 0.6120510101318359, "learning_rate": 0.0005, "loss": 1.2177571058273315, "step": 50334 }, { "epoch": 3.968072526606228, "grad_norm": 0.39169371128082275, "learning_rate": 0.0005, "loss": 0.9528244733810425, "step": 50335 }, { "epoch": 3.968151359873867, "grad_norm": 0.42024436593055725, "learning_rate": 0.0005, "loss": 1.0045233964920044, "step": 50336 }, { "epoch": 3.9682301931415056, "grad_norm": 0.5611579418182373, "learning_rate": 0.0005, "loss": 1.212740421295166, "step": 50337 }, { "epoch": 3.9683090264091447, "grad_norm": 0.46236294507980347, "learning_rate": 0.0005, "loss": 1.214735746383667, "step": 50338 }, { "epoch": 3.968387859676784, "grad_norm": 0.5251564979553223, "learning_rate": 0.0005, "loss": 1.206740379333496, "step": 50339 }, { "epoch": 3.9684666929444226, "grad_norm": 0.41340142488479614, "learning_rate": 0.0005, "loss": 1.0399584770202637, "step": 50340 }, { "epoch": 3.9685455262120612, "grad_norm": 0.5177390575408936, "learning_rate": 0.0005, "loss": 1.1743638515472412, "step": 50341 }, { "epoch": 3.9686243594797004, "grad_norm": 0.40412506461143494, "learning_rate": 0.0005, "loss": 1.168145775794983, "step": 50342 }, { "epoch": 3.9687031927473395, "grad_norm": 0.5203897356987, "learning_rate": 0.0005, "loss": 1.2950549125671387, "step": 50343 }, { "epoch": 3.968782026014978, "grad_norm": 0.4740593135356903, "learning_rate": 0.0005, "loss": 1.1219675540924072, "step": 50344 }, { "epoch": 3.9688608592826173, "grad_norm": 0.6260193586349487, "learning_rate": 0.0005, "loss": 1.2987416982650757, "step": 50345 }, { "epoch": 3.968939692550256, "grad_norm": 0.4777824282646179, "learning_rate": 0.0005, "loss": 1.2067501544952393, "step": 50346 }, { "epoch": 3.969018525817895, "grad_norm": 0.5406056046485901, "learning_rate": 0.0005, "loss": 1.1274323463439941, "step": 50347 }, { "epoch": 3.9690973590855343, "grad_norm": 0.560703694820404, "learning_rate": 0.0005, "loss": 1.2464931011199951, "step": 50348 }, { "epoch": 3.969176192353173, "grad_norm": 0.5316333174705505, "learning_rate": 0.0005, "loss": 1.2796663045883179, "step": 50349 }, { "epoch": 3.969255025620812, "grad_norm": 0.5425503253936768, "learning_rate": 0.0005, "loss": 1.239827275276184, "step": 50350 }, { "epoch": 3.969333858888451, "grad_norm": 0.6113425493240356, "learning_rate": 0.0005, "loss": 1.2276430130004883, "step": 50351 }, { "epoch": 3.96941269215609, "grad_norm": 0.5108301639556885, "learning_rate": 0.0005, "loss": 1.3797802925109863, "step": 50352 }, { "epoch": 3.969491525423729, "grad_norm": 0.4695115089416504, "learning_rate": 0.0005, "loss": 1.0666565895080566, "step": 50353 }, { "epoch": 3.9695703586913678, "grad_norm": 0.5108916759490967, "learning_rate": 0.0005, "loss": 1.2051749229431152, "step": 50354 }, { "epoch": 3.9696491919590065, "grad_norm": 0.433805912733078, "learning_rate": 0.0005, "loss": 1.1916447877883911, "step": 50355 }, { "epoch": 3.9697280252266456, "grad_norm": 0.4188903868198395, "learning_rate": 0.0005, "loss": 0.983535885810852, "step": 50356 }, { "epoch": 3.9698068584942847, "grad_norm": 0.4685709476470947, "learning_rate": 0.0005, "loss": 1.2718689441680908, "step": 50357 }, { "epoch": 3.9698856917619234, "grad_norm": 0.620184063911438, "learning_rate": 0.0005, "loss": 1.3097450733184814, "step": 50358 }, { "epoch": 3.9699645250295625, "grad_norm": 0.4964202642440796, "learning_rate": 0.0005, "loss": 1.2947511672973633, "step": 50359 }, { "epoch": 3.9700433582972012, "grad_norm": 0.4682161509990692, "learning_rate": 0.0005, "loss": 1.230562686920166, "step": 50360 }, { "epoch": 3.9701221915648404, "grad_norm": 0.511364758014679, "learning_rate": 0.0005, "loss": 1.194621205329895, "step": 50361 }, { "epoch": 3.9702010248324795, "grad_norm": 0.4482881724834442, "learning_rate": 0.0005, "loss": 1.0227131843566895, "step": 50362 }, { "epoch": 3.970279858100118, "grad_norm": 0.45877766609191895, "learning_rate": 0.0005, "loss": 1.255075454711914, "step": 50363 }, { "epoch": 3.9703586913677573, "grad_norm": 0.5389971137046814, "learning_rate": 0.0005, "loss": 1.2504832744598389, "step": 50364 }, { "epoch": 3.970437524635396, "grad_norm": 0.40652135014533997, "learning_rate": 0.0005, "loss": 1.1435078382492065, "step": 50365 }, { "epoch": 3.970516357903035, "grad_norm": 0.5410950183868408, "learning_rate": 0.0005, "loss": 1.2413487434387207, "step": 50366 }, { "epoch": 3.9705951911706743, "grad_norm": 0.44346797466278076, "learning_rate": 0.0005, "loss": 1.2731778621673584, "step": 50367 }, { "epoch": 3.970674024438313, "grad_norm": 0.3789750933647156, "learning_rate": 0.0005, "loss": 0.9920392036437988, "step": 50368 }, { "epoch": 3.9707528577059517, "grad_norm": 0.4072570204734802, "learning_rate": 0.0005, "loss": 0.954650342464447, "step": 50369 }, { "epoch": 3.970831690973591, "grad_norm": 0.39727795124053955, "learning_rate": 0.0005, "loss": 1.1252846717834473, "step": 50370 }, { "epoch": 3.97091052424123, "grad_norm": 0.36219385266304016, "learning_rate": 0.0005, "loss": 0.8614943027496338, "step": 50371 }, { "epoch": 3.9709893575088686, "grad_norm": 0.4179133474826813, "learning_rate": 0.0005, "loss": 1.3049960136413574, "step": 50372 }, { "epoch": 3.9710681907765077, "grad_norm": 0.5230509638786316, "learning_rate": 0.0005, "loss": 1.2056570053100586, "step": 50373 }, { "epoch": 3.9711470240441464, "grad_norm": 0.40842458605766296, "learning_rate": 0.0005, "loss": 1.1351075172424316, "step": 50374 }, { "epoch": 3.9712258573117856, "grad_norm": 0.5448755621910095, "learning_rate": 0.0005, "loss": 1.1834511756896973, "step": 50375 }, { "epoch": 3.9713046905794247, "grad_norm": 0.5035710334777832, "learning_rate": 0.0005, "loss": 1.13639497756958, "step": 50376 }, { "epoch": 3.9713835238470634, "grad_norm": 0.5097520351409912, "learning_rate": 0.0005, "loss": 1.2765494585037231, "step": 50377 }, { "epoch": 3.9714623571147025, "grad_norm": 0.4245436191558838, "learning_rate": 0.0005, "loss": 1.0808908939361572, "step": 50378 }, { "epoch": 3.971541190382341, "grad_norm": 0.4630460739135742, "learning_rate": 0.0005, "loss": 1.0832833051681519, "step": 50379 }, { "epoch": 3.9716200236499803, "grad_norm": 0.46538597345352173, "learning_rate": 0.0005, "loss": 1.1380391120910645, "step": 50380 }, { "epoch": 3.9716988569176195, "grad_norm": 0.4732646942138672, "learning_rate": 0.0005, "loss": 1.2195442914962769, "step": 50381 }, { "epoch": 3.971777690185258, "grad_norm": 0.42125874757766724, "learning_rate": 0.0005, "loss": 0.9539234638214111, "step": 50382 }, { "epoch": 3.971856523452897, "grad_norm": 0.4767555296421051, "learning_rate": 0.0005, "loss": 1.1067856550216675, "step": 50383 }, { "epoch": 3.971935356720536, "grad_norm": 0.5007661581039429, "learning_rate": 0.0005, "loss": 1.3066668510437012, "step": 50384 }, { "epoch": 3.972014189988175, "grad_norm": 0.6820459961891174, "learning_rate": 0.0005, "loss": 1.457604169845581, "step": 50385 }, { "epoch": 3.972093023255814, "grad_norm": 0.4277774393558502, "learning_rate": 0.0005, "loss": 0.9633263349533081, "step": 50386 }, { "epoch": 3.972171856523453, "grad_norm": 0.5258113741874695, "learning_rate": 0.0005, "loss": 1.1592308282852173, "step": 50387 }, { "epoch": 3.9722506897910916, "grad_norm": 0.7698420882225037, "learning_rate": 0.0005, "loss": 1.2038202285766602, "step": 50388 }, { "epoch": 3.9723295230587308, "grad_norm": 0.5037769079208374, "learning_rate": 0.0005, "loss": 1.3333419561386108, "step": 50389 }, { "epoch": 3.97240835632637, "grad_norm": 0.41174712777137756, "learning_rate": 0.0005, "loss": 1.0706957578659058, "step": 50390 }, { "epoch": 3.9724871895940086, "grad_norm": 0.454721599817276, "learning_rate": 0.0005, "loss": 1.1237988471984863, "step": 50391 }, { "epoch": 3.9725660228616477, "grad_norm": 0.8437808156013489, "learning_rate": 0.0005, "loss": 1.3027966022491455, "step": 50392 }, { "epoch": 3.9726448561292864, "grad_norm": 0.40724053978919983, "learning_rate": 0.0005, "loss": 1.139877438545227, "step": 50393 }, { "epoch": 3.9727236893969256, "grad_norm": 0.4626915156841278, "learning_rate": 0.0005, "loss": 1.216148853302002, "step": 50394 }, { "epoch": 3.9728025226645647, "grad_norm": 0.4585820138454437, "learning_rate": 0.0005, "loss": 1.0838121175765991, "step": 50395 }, { "epoch": 3.9728813559322034, "grad_norm": 0.5551760792732239, "learning_rate": 0.0005, "loss": 1.260764479637146, "step": 50396 }, { "epoch": 3.972960189199842, "grad_norm": 0.4566999673843384, "learning_rate": 0.0005, "loss": 1.0858559608459473, "step": 50397 }, { "epoch": 3.973039022467481, "grad_norm": 0.3899252414703369, "learning_rate": 0.0005, "loss": 1.0145394802093506, "step": 50398 }, { "epoch": 3.9731178557351203, "grad_norm": 0.4625943899154663, "learning_rate": 0.0005, "loss": 1.2084558010101318, "step": 50399 }, { "epoch": 3.973196689002759, "grad_norm": 0.5260791182518005, "learning_rate": 0.0005, "loss": 1.3068687915802002, "step": 50400 }, { "epoch": 3.973275522270398, "grad_norm": 0.4627703130245209, "learning_rate": 0.0005, "loss": 1.2449657917022705, "step": 50401 }, { "epoch": 3.973354355538037, "grad_norm": 0.4574100375175476, "learning_rate": 0.0005, "loss": 1.1762735843658447, "step": 50402 }, { "epoch": 3.973433188805676, "grad_norm": 0.4804086983203888, "learning_rate": 0.0005, "loss": 1.2192012071609497, "step": 50403 }, { "epoch": 3.973512022073315, "grad_norm": 0.4733754098415375, "learning_rate": 0.0005, "loss": 1.1136809587478638, "step": 50404 }, { "epoch": 3.973590855340954, "grad_norm": 0.3636533319950104, "learning_rate": 0.0005, "loss": 0.9626036882400513, "step": 50405 }, { "epoch": 3.973669688608593, "grad_norm": 0.6347132921218872, "learning_rate": 0.0005, "loss": 1.2307209968566895, "step": 50406 }, { "epoch": 3.9737485218762316, "grad_norm": 0.48279228806495667, "learning_rate": 0.0005, "loss": 1.111835241317749, "step": 50407 }, { "epoch": 3.9738273551438708, "grad_norm": 0.565523624420166, "learning_rate": 0.0005, "loss": 1.1281893253326416, "step": 50408 }, { "epoch": 3.97390618841151, "grad_norm": 0.5018952488899231, "learning_rate": 0.0005, "loss": 1.241145372390747, "step": 50409 }, { "epoch": 3.9739850216791486, "grad_norm": 0.6059261560440063, "learning_rate": 0.0005, "loss": 1.2670649290084839, "step": 50410 }, { "epoch": 3.9740638549467877, "grad_norm": 0.43337586522102356, "learning_rate": 0.0005, "loss": 1.1131446361541748, "step": 50411 }, { "epoch": 3.9741426882144264, "grad_norm": 0.6053874492645264, "learning_rate": 0.0005, "loss": 1.449979305267334, "step": 50412 }, { "epoch": 3.9742215214820655, "grad_norm": 0.6746489405632019, "learning_rate": 0.0005, "loss": 1.2916016578674316, "step": 50413 }, { "epoch": 3.9743003547497042, "grad_norm": 0.5870860815048218, "learning_rate": 0.0005, "loss": 1.3524603843688965, "step": 50414 }, { "epoch": 3.9743791880173434, "grad_norm": 0.4344803988933563, "learning_rate": 0.0005, "loss": 1.193099021911621, "step": 50415 }, { "epoch": 3.974458021284982, "grad_norm": 0.45425429940223694, "learning_rate": 0.0005, "loss": 1.1623420715332031, "step": 50416 }, { "epoch": 3.974536854552621, "grad_norm": 0.5218825340270996, "learning_rate": 0.0005, "loss": 1.108646273612976, "step": 50417 }, { "epoch": 3.9746156878202603, "grad_norm": 0.5271381139755249, "learning_rate": 0.0005, "loss": 1.2670018672943115, "step": 50418 }, { "epoch": 3.974694521087899, "grad_norm": 0.5166348218917847, "learning_rate": 0.0005, "loss": 1.3290146589279175, "step": 50419 }, { "epoch": 3.974773354355538, "grad_norm": 0.5934843420982361, "learning_rate": 0.0005, "loss": 1.2374601364135742, "step": 50420 }, { "epoch": 3.974852187623177, "grad_norm": 0.40782779455184937, "learning_rate": 0.0005, "loss": 1.0271897315979004, "step": 50421 }, { "epoch": 3.974931020890816, "grad_norm": 0.6189144253730774, "learning_rate": 0.0005, "loss": 1.0873076915740967, "step": 50422 }, { "epoch": 3.975009854158455, "grad_norm": 0.4262882471084595, "learning_rate": 0.0005, "loss": 1.1639878749847412, "step": 50423 }, { "epoch": 3.975088687426094, "grad_norm": 0.5681056976318359, "learning_rate": 0.0005, "loss": 1.2012946605682373, "step": 50424 }, { "epoch": 3.975167520693733, "grad_norm": 0.4828145503997803, "learning_rate": 0.0005, "loss": 1.3462932109832764, "step": 50425 }, { "epoch": 3.9752463539613716, "grad_norm": 0.48880842328071594, "learning_rate": 0.0005, "loss": 1.1989024877548218, "step": 50426 }, { "epoch": 3.9753251872290107, "grad_norm": 0.4432394504547119, "learning_rate": 0.0005, "loss": 1.199973702430725, "step": 50427 }, { "epoch": 3.9754040204966494, "grad_norm": 0.45133376121520996, "learning_rate": 0.0005, "loss": 1.2571961879730225, "step": 50428 }, { "epoch": 3.9754828537642886, "grad_norm": 0.5422849655151367, "learning_rate": 0.0005, "loss": 1.249060869216919, "step": 50429 }, { "epoch": 3.9755616870319272, "grad_norm": 0.5602874755859375, "learning_rate": 0.0005, "loss": 1.310579538345337, "step": 50430 }, { "epoch": 3.9756405202995664, "grad_norm": 0.45257291197776794, "learning_rate": 0.0005, "loss": 1.0622014999389648, "step": 50431 }, { "epoch": 3.9757193535672055, "grad_norm": 0.524416446685791, "learning_rate": 0.0005, "loss": 1.2715082168579102, "step": 50432 }, { "epoch": 3.975798186834844, "grad_norm": 0.5258030295372009, "learning_rate": 0.0005, "loss": 1.205428123474121, "step": 50433 }, { "epoch": 3.9758770201024833, "grad_norm": 0.5632043480873108, "learning_rate": 0.0005, "loss": 1.2744646072387695, "step": 50434 }, { "epoch": 3.975955853370122, "grad_norm": 0.4688258171081543, "learning_rate": 0.0005, "loss": 1.1981148719787598, "step": 50435 }, { "epoch": 3.976034686637761, "grad_norm": 0.608957827091217, "learning_rate": 0.0005, "loss": 1.1499193906784058, "step": 50436 }, { "epoch": 3.9761135199054003, "grad_norm": 0.49940481781959534, "learning_rate": 0.0005, "loss": 1.0885933637619019, "step": 50437 }, { "epoch": 3.976192353173039, "grad_norm": 0.5710262656211853, "learning_rate": 0.0005, "loss": 1.155134677886963, "step": 50438 }, { "epoch": 3.976271186440678, "grad_norm": 0.5556796789169312, "learning_rate": 0.0005, "loss": 1.2009299993515015, "step": 50439 }, { "epoch": 3.976350019708317, "grad_norm": 0.5596215724945068, "learning_rate": 0.0005, "loss": 1.0907132625579834, "step": 50440 }, { "epoch": 3.976428852975956, "grad_norm": 0.43276920914649963, "learning_rate": 0.0005, "loss": 1.1631731986999512, "step": 50441 }, { "epoch": 3.976507686243595, "grad_norm": 0.5412521958351135, "learning_rate": 0.0005, "loss": 1.24163818359375, "step": 50442 }, { "epoch": 3.9765865195112338, "grad_norm": 0.528292179107666, "learning_rate": 0.0005, "loss": 1.205627679824829, "step": 50443 }, { "epoch": 3.9766653527788725, "grad_norm": 0.4355660378932953, "learning_rate": 0.0005, "loss": 1.1313323974609375, "step": 50444 }, { "epoch": 3.9767441860465116, "grad_norm": 0.5821951031684875, "learning_rate": 0.0005, "loss": 1.3087835311889648, "step": 50445 }, { "epoch": 3.9768230193141507, "grad_norm": 0.522916853427887, "learning_rate": 0.0005, "loss": 1.3025848865509033, "step": 50446 }, { "epoch": 3.9769018525817894, "grad_norm": 0.45329374074935913, "learning_rate": 0.0005, "loss": 1.2143498659133911, "step": 50447 }, { "epoch": 3.9769806858494285, "grad_norm": 0.48147886991500854, "learning_rate": 0.0005, "loss": 1.3052895069122314, "step": 50448 }, { "epoch": 3.9770595191170672, "grad_norm": 0.5635235905647278, "learning_rate": 0.0005, "loss": 1.3307020664215088, "step": 50449 }, { "epoch": 3.9771383523847064, "grad_norm": 0.46070611476898193, "learning_rate": 0.0005, "loss": 1.1003127098083496, "step": 50450 }, { "epoch": 3.9772171856523455, "grad_norm": 0.5088222622871399, "learning_rate": 0.0005, "loss": 1.190485954284668, "step": 50451 }, { "epoch": 3.977296018919984, "grad_norm": 0.44939011335372925, "learning_rate": 0.0005, "loss": 1.00669527053833, "step": 50452 }, { "epoch": 3.9773748521876233, "grad_norm": 0.5397419333457947, "learning_rate": 0.0005, "loss": 1.287780523300171, "step": 50453 }, { "epoch": 3.977453685455262, "grad_norm": 0.4087185263633728, "learning_rate": 0.0005, "loss": 1.1477398872375488, "step": 50454 }, { "epoch": 3.977532518722901, "grad_norm": 0.7378901839256287, "learning_rate": 0.0005, "loss": 1.2739355564117432, "step": 50455 }, { "epoch": 3.9776113519905403, "grad_norm": 0.48296934366226196, "learning_rate": 0.0005, "loss": 1.1807324886322021, "step": 50456 }, { "epoch": 3.977690185258179, "grad_norm": 0.4397543668746948, "learning_rate": 0.0005, "loss": 1.167378306388855, "step": 50457 }, { "epoch": 3.9777690185258177, "grad_norm": 0.4888089895248413, "learning_rate": 0.0005, "loss": 1.2054927349090576, "step": 50458 }, { "epoch": 3.977847851793457, "grad_norm": 0.4558197259902954, "learning_rate": 0.0005, "loss": 1.2875795364379883, "step": 50459 }, { "epoch": 3.977926685061096, "grad_norm": 0.4042118787765503, "learning_rate": 0.0005, "loss": 1.2024824619293213, "step": 50460 }, { "epoch": 3.9780055183287346, "grad_norm": 0.5659412145614624, "learning_rate": 0.0005, "loss": 1.2755126953125, "step": 50461 }, { "epoch": 3.9780843515963737, "grad_norm": 0.4547971189022064, "learning_rate": 0.0005, "loss": 1.0667411088943481, "step": 50462 }, { "epoch": 3.9781631848640124, "grad_norm": 0.5067888498306274, "learning_rate": 0.0005, "loss": 1.3367187976837158, "step": 50463 }, { "epoch": 3.9782420181316516, "grad_norm": 0.4019898772239685, "learning_rate": 0.0005, "loss": 0.9481989145278931, "step": 50464 }, { "epoch": 3.9783208513992907, "grad_norm": 0.3974064588546753, "learning_rate": 0.0005, "loss": 0.9695289731025696, "step": 50465 }, { "epoch": 3.9783996846669294, "grad_norm": 0.44066694378852844, "learning_rate": 0.0005, "loss": 0.98296058177948, "step": 50466 }, { "epoch": 3.9784785179345685, "grad_norm": 0.5324797034263611, "learning_rate": 0.0005, "loss": 1.2358365058898926, "step": 50467 }, { "epoch": 3.978557351202207, "grad_norm": 0.4907209873199463, "learning_rate": 0.0005, "loss": 1.1393821239471436, "step": 50468 }, { "epoch": 3.9786361844698463, "grad_norm": 0.5580056309700012, "learning_rate": 0.0005, "loss": 1.1240592002868652, "step": 50469 }, { "epoch": 3.9787150177374855, "grad_norm": 0.47270557284355164, "learning_rate": 0.0005, "loss": 1.0899913311004639, "step": 50470 }, { "epoch": 3.978793851005124, "grad_norm": 0.45571818947792053, "learning_rate": 0.0005, "loss": 1.0759799480438232, "step": 50471 }, { "epoch": 3.978872684272763, "grad_norm": 0.5389555096626282, "learning_rate": 0.0005, "loss": 1.053868055343628, "step": 50472 }, { "epoch": 3.978951517540402, "grad_norm": 0.47484633326530457, "learning_rate": 0.0005, "loss": 1.0861520767211914, "step": 50473 }, { "epoch": 3.979030350808041, "grad_norm": 0.5239099264144897, "learning_rate": 0.0005, "loss": 1.1516838073730469, "step": 50474 }, { "epoch": 3.97910918407568, "grad_norm": 0.440428763628006, "learning_rate": 0.0005, "loss": 1.251525640487671, "step": 50475 }, { "epoch": 3.979188017343319, "grad_norm": 0.565095067024231, "learning_rate": 0.0005, "loss": 1.167557716369629, "step": 50476 }, { "epoch": 3.9792668506109576, "grad_norm": 0.39646032452583313, "learning_rate": 0.0005, "loss": 1.1196317672729492, "step": 50477 }, { "epoch": 3.9793456838785968, "grad_norm": 0.4741465747356415, "learning_rate": 0.0005, "loss": 1.1068906784057617, "step": 50478 }, { "epoch": 3.979424517146236, "grad_norm": 0.5585353374481201, "learning_rate": 0.0005, "loss": 1.0542844533920288, "step": 50479 }, { "epoch": 3.9795033504138746, "grad_norm": 0.47516709566116333, "learning_rate": 0.0005, "loss": 1.0626091957092285, "step": 50480 }, { "epoch": 3.9795821836815137, "grad_norm": 0.47199341654777527, "learning_rate": 0.0005, "loss": 1.1068511009216309, "step": 50481 }, { "epoch": 3.9796610169491524, "grad_norm": 0.45817121863365173, "learning_rate": 0.0005, "loss": 1.164758563041687, "step": 50482 }, { "epoch": 3.9797398502167916, "grad_norm": 0.7626316547393799, "learning_rate": 0.0005, "loss": 1.3767023086547852, "step": 50483 }, { "epoch": 3.9798186834844307, "grad_norm": 0.43525275588035583, "learning_rate": 0.0005, "loss": 1.1425609588623047, "step": 50484 }, { "epoch": 3.9798975167520694, "grad_norm": 0.5039048194885254, "learning_rate": 0.0005, "loss": 1.1377863883972168, "step": 50485 }, { "epoch": 3.979976350019708, "grad_norm": 0.6108113527297974, "learning_rate": 0.0005, "loss": 1.2356680631637573, "step": 50486 }, { "epoch": 3.980055183287347, "grad_norm": 0.6677166819572449, "learning_rate": 0.0005, "loss": 1.1475660800933838, "step": 50487 }, { "epoch": 3.9801340165549863, "grad_norm": 0.7613092064857483, "learning_rate": 0.0005, "loss": 1.162617802619934, "step": 50488 }, { "epoch": 3.980212849822625, "grad_norm": 0.46953585743904114, "learning_rate": 0.0005, "loss": 1.1305238008499146, "step": 50489 }, { "epoch": 3.980291683090264, "grad_norm": 0.4562774896621704, "learning_rate": 0.0005, "loss": 1.0719728469848633, "step": 50490 }, { "epoch": 3.980370516357903, "grad_norm": 0.6115090847015381, "learning_rate": 0.0005, "loss": 1.1054744720458984, "step": 50491 }, { "epoch": 3.980449349625542, "grad_norm": 0.5855679512023926, "learning_rate": 0.0005, "loss": 1.3046953678131104, "step": 50492 }, { "epoch": 3.980528182893181, "grad_norm": 0.4188099503517151, "learning_rate": 0.0005, "loss": 1.048436164855957, "step": 50493 }, { "epoch": 3.98060701616082, "grad_norm": 0.6050002574920654, "learning_rate": 0.0005, "loss": 1.0990238189697266, "step": 50494 }, { "epoch": 3.980685849428459, "grad_norm": 0.42492619156837463, "learning_rate": 0.0005, "loss": 1.0764964818954468, "step": 50495 }, { "epoch": 3.9807646826960976, "grad_norm": 0.5826347470283508, "learning_rate": 0.0005, "loss": 1.2900824546813965, "step": 50496 }, { "epoch": 3.9808435159637368, "grad_norm": 0.6541959643363953, "learning_rate": 0.0005, "loss": 1.5424896478652954, "step": 50497 }, { "epoch": 3.980922349231376, "grad_norm": 0.4972035884857178, "learning_rate": 0.0005, "loss": 1.1015079021453857, "step": 50498 }, { "epoch": 3.9810011824990146, "grad_norm": 0.45844539999961853, "learning_rate": 0.0005, "loss": 0.9503037929534912, "step": 50499 }, { "epoch": 3.9810800157666533, "grad_norm": 0.5670531392097473, "learning_rate": 0.0005, "loss": 1.265866994857788, "step": 50500 }, { "epoch": 3.9811588490342924, "grad_norm": 0.4806537628173828, "learning_rate": 0.0005, "loss": 1.180219292640686, "step": 50501 }, { "epoch": 3.9812376823019315, "grad_norm": 0.5710173845291138, "learning_rate": 0.0005, "loss": 1.1793066263198853, "step": 50502 }, { "epoch": 3.9813165155695702, "grad_norm": 0.4998824894428253, "learning_rate": 0.0005, "loss": 1.2135484218597412, "step": 50503 }, { "epoch": 3.9813953488372094, "grad_norm": 0.506648063659668, "learning_rate": 0.0005, "loss": 1.1688103675842285, "step": 50504 }, { "epoch": 3.981474182104848, "grad_norm": 0.5923233032226562, "learning_rate": 0.0005, "loss": 1.1612412929534912, "step": 50505 }, { "epoch": 3.981553015372487, "grad_norm": 0.5387129783630371, "learning_rate": 0.0005, "loss": 1.3104376792907715, "step": 50506 }, { "epoch": 3.9816318486401263, "grad_norm": 0.4674799144268036, "learning_rate": 0.0005, "loss": 1.1415820121765137, "step": 50507 }, { "epoch": 3.981710681907765, "grad_norm": 0.4409120976924896, "learning_rate": 0.0005, "loss": 0.9285910129547119, "step": 50508 }, { "epoch": 3.981789515175404, "grad_norm": 0.4511386752128601, "learning_rate": 0.0005, "loss": 1.07065749168396, "step": 50509 }, { "epoch": 3.981868348443043, "grad_norm": 0.48616549372673035, "learning_rate": 0.0005, "loss": 1.1812164783477783, "step": 50510 }, { "epoch": 3.981947181710682, "grad_norm": 0.45372360944747925, "learning_rate": 0.0005, "loss": 1.188374400138855, "step": 50511 }, { "epoch": 3.982026014978321, "grad_norm": 0.4186094105243683, "learning_rate": 0.0005, "loss": 1.1547101736068726, "step": 50512 }, { "epoch": 3.98210484824596, "grad_norm": 0.5271056294441223, "learning_rate": 0.0005, "loss": 1.2765657901763916, "step": 50513 }, { "epoch": 3.9821836815135985, "grad_norm": 0.508169949054718, "learning_rate": 0.0005, "loss": 1.183534860610962, "step": 50514 }, { "epoch": 3.9822625147812376, "grad_norm": 0.49474936723709106, "learning_rate": 0.0005, "loss": 1.1929097175598145, "step": 50515 }, { "epoch": 3.9823413480488767, "grad_norm": 0.45377641916275024, "learning_rate": 0.0005, "loss": 1.2647396326065063, "step": 50516 }, { "epoch": 3.9824201813165154, "grad_norm": 0.5464780330657959, "learning_rate": 0.0005, "loss": 1.177361249923706, "step": 50517 }, { "epoch": 3.9824990145841546, "grad_norm": 0.4613819420337677, "learning_rate": 0.0005, "loss": 1.2052013874053955, "step": 50518 }, { "epoch": 3.9825778478517933, "grad_norm": 0.46427229046821594, "learning_rate": 0.0005, "loss": 1.0483622550964355, "step": 50519 }, { "epoch": 3.9826566811194324, "grad_norm": 0.4430712163448334, "learning_rate": 0.0005, "loss": 1.172403335571289, "step": 50520 }, { "epoch": 3.9827355143870715, "grad_norm": 0.5816677808761597, "learning_rate": 0.0005, "loss": 1.121800184249878, "step": 50521 }, { "epoch": 3.98281434765471, "grad_norm": 0.5965324640274048, "learning_rate": 0.0005, "loss": 1.1521350145339966, "step": 50522 }, { "epoch": 3.9828931809223493, "grad_norm": 0.5175036787986755, "learning_rate": 0.0005, "loss": 1.17948579788208, "step": 50523 }, { "epoch": 3.982972014189988, "grad_norm": 0.3977592885494232, "learning_rate": 0.0005, "loss": 1.125476598739624, "step": 50524 }, { "epoch": 3.983050847457627, "grad_norm": 0.4814138412475586, "learning_rate": 0.0005, "loss": 1.108025312423706, "step": 50525 }, { "epoch": 3.9831296807252663, "grad_norm": 0.5359646677970886, "learning_rate": 0.0005, "loss": 1.267255187034607, "step": 50526 }, { "epoch": 3.983208513992905, "grad_norm": 0.5157613158226013, "learning_rate": 0.0005, "loss": 1.2694718837738037, "step": 50527 }, { "epoch": 3.9832873472605437, "grad_norm": 0.47950926423072815, "learning_rate": 0.0005, "loss": 1.1429133415222168, "step": 50528 }, { "epoch": 3.983366180528183, "grad_norm": 0.5255470871925354, "learning_rate": 0.0005, "loss": 1.0852487087249756, "step": 50529 }, { "epoch": 3.983445013795822, "grad_norm": 0.3986667990684509, "learning_rate": 0.0005, "loss": 0.9577810764312744, "step": 50530 }, { "epoch": 3.9835238470634606, "grad_norm": 0.5320826172828674, "learning_rate": 0.0005, "loss": 1.406965732574463, "step": 50531 }, { "epoch": 3.9836026803310998, "grad_norm": 0.4518488347530365, "learning_rate": 0.0005, "loss": 1.3022308349609375, "step": 50532 }, { "epoch": 3.9836815135987385, "grad_norm": 0.48621904850006104, "learning_rate": 0.0005, "loss": 1.2381620407104492, "step": 50533 }, { "epoch": 3.9837603468663776, "grad_norm": 0.45096728205680847, "learning_rate": 0.0005, "loss": 1.0537502765655518, "step": 50534 }, { "epoch": 3.9838391801340167, "grad_norm": 0.49918869137763977, "learning_rate": 0.0005, "loss": 1.2358059883117676, "step": 50535 }, { "epoch": 3.9839180134016554, "grad_norm": 0.5556896328926086, "learning_rate": 0.0005, "loss": 1.4146429300308228, "step": 50536 }, { "epoch": 3.9839968466692945, "grad_norm": 0.4493999183177948, "learning_rate": 0.0005, "loss": 1.2429295778274536, "step": 50537 }, { "epoch": 3.9840756799369332, "grad_norm": 0.44554346799850464, "learning_rate": 0.0005, "loss": 1.0509834289550781, "step": 50538 }, { "epoch": 3.9841545132045724, "grad_norm": 0.35030949115753174, "learning_rate": 0.0005, "loss": 0.9297353029251099, "step": 50539 }, { "epoch": 3.9842333464722115, "grad_norm": 0.4972453713417053, "learning_rate": 0.0005, "loss": 1.1514149904251099, "step": 50540 }, { "epoch": 3.98431217973985, "grad_norm": 0.5008050799369812, "learning_rate": 0.0005, "loss": 1.3629066944122314, "step": 50541 }, { "epoch": 3.9843910130074893, "grad_norm": 0.3746371865272522, "learning_rate": 0.0005, "loss": 1.1015124320983887, "step": 50542 }, { "epoch": 3.984469846275128, "grad_norm": 0.4888054430484772, "learning_rate": 0.0005, "loss": 1.202206015586853, "step": 50543 }, { "epoch": 3.984548679542767, "grad_norm": 0.5131155252456665, "learning_rate": 0.0005, "loss": 1.2029626369476318, "step": 50544 }, { "epoch": 3.984627512810406, "grad_norm": 0.4631279706954956, "learning_rate": 0.0005, "loss": 1.3030444383621216, "step": 50545 }, { "epoch": 3.984706346078045, "grad_norm": 0.7538176774978638, "learning_rate": 0.0005, "loss": 1.316483736038208, "step": 50546 }, { "epoch": 3.9847851793456837, "grad_norm": 0.6198901534080505, "learning_rate": 0.0005, "loss": 1.2065829038619995, "step": 50547 }, { "epoch": 3.984864012613323, "grad_norm": 0.5703253149986267, "learning_rate": 0.0005, "loss": 1.1462892293930054, "step": 50548 }, { "epoch": 3.984942845880962, "grad_norm": 0.5770854353904724, "learning_rate": 0.0005, "loss": 1.1980940103530884, "step": 50549 }, { "epoch": 3.9850216791486006, "grad_norm": 0.5060098767280579, "learning_rate": 0.0005, "loss": 1.221269130706787, "step": 50550 }, { "epoch": 3.9851005124162397, "grad_norm": 0.5021098256111145, "learning_rate": 0.0005, "loss": 1.1194860935211182, "step": 50551 }, { "epoch": 3.9851793456838784, "grad_norm": 0.46940746903419495, "learning_rate": 0.0005, "loss": 1.1978119611740112, "step": 50552 }, { "epoch": 3.9852581789515176, "grad_norm": 0.5143791437149048, "learning_rate": 0.0005, "loss": 1.346796989440918, "step": 50553 }, { "epoch": 3.9853370122191567, "grad_norm": 0.5524760484695435, "learning_rate": 0.0005, "loss": 1.227504014968872, "step": 50554 }, { "epoch": 3.9854158454867954, "grad_norm": 0.3947479724884033, "learning_rate": 0.0005, "loss": 1.107348918914795, "step": 50555 }, { "epoch": 3.9854946787544345, "grad_norm": 0.5570516586303711, "learning_rate": 0.0005, "loss": 1.1683769226074219, "step": 50556 }, { "epoch": 3.985573512022073, "grad_norm": 0.48874855041503906, "learning_rate": 0.0005, "loss": 0.9671899080276489, "step": 50557 }, { "epoch": 3.9856523452897124, "grad_norm": 0.6178058385848999, "learning_rate": 0.0005, "loss": 1.2649742364883423, "step": 50558 }, { "epoch": 3.985731178557351, "grad_norm": 0.43040230870246887, "learning_rate": 0.0005, "loss": 1.0205967426300049, "step": 50559 }, { "epoch": 3.98581001182499, "grad_norm": 0.45081084966659546, "learning_rate": 0.0005, "loss": 1.0592153072357178, "step": 50560 }, { "epoch": 3.985888845092629, "grad_norm": 0.766563355922699, "learning_rate": 0.0005, "loss": 1.0651719570159912, "step": 50561 }, { "epoch": 3.985967678360268, "grad_norm": 0.47780415415763855, "learning_rate": 0.0005, "loss": 1.1016448736190796, "step": 50562 }, { "epoch": 3.986046511627907, "grad_norm": 0.6635661721229553, "learning_rate": 0.0005, "loss": 1.169906735420227, "step": 50563 }, { "epoch": 3.986125344895546, "grad_norm": 0.46941590309143066, "learning_rate": 0.0005, "loss": 1.1966546773910522, "step": 50564 }, { "epoch": 3.986204178163185, "grad_norm": 0.49433451890945435, "learning_rate": 0.0005, "loss": 1.1616215705871582, "step": 50565 }, { "epoch": 3.9862830114308236, "grad_norm": 0.5322380661964417, "learning_rate": 0.0005, "loss": 1.141575813293457, "step": 50566 }, { "epoch": 3.9863618446984628, "grad_norm": 0.4759889245033264, "learning_rate": 0.0005, "loss": 1.2604756355285645, "step": 50567 }, { "epoch": 3.986440677966102, "grad_norm": 0.49858540296554565, "learning_rate": 0.0005, "loss": 1.1804968118667603, "step": 50568 }, { "epoch": 3.9865195112337406, "grad_norm": 0.5200333595275879, "learning_rate": 0.0005, "loss": 1.1651421785354614, "step": 50569 }, { "epoch": 3.9865983445013797, "grad_norm": 0.4427843391895294, "learning_rate": 0.0005, "loss": 1.1788256168365479, "step": 50570 }, { "epoch": 3.9866771777690184, "grad_norm": 0.5920681953430176, "learning_rate": 0.0005, "loss": 1.1457116603851318, "step": 50571 }, { "epoch": 3.9867560110366576, "grad_norm": 0.6322240829467773, "learning_rate": 0.0005, "loss": 1.2440117597579956, "step": 50572 }, { "epoch": 3.9868348443042967, "grad_norm": 0.5038393139839172, "learning_rate": 0.0005, "loss": 1.1877059936523438, "step": 50573 }, { "epoch": 3.9869136775719354, "grad_norm": 0.5362566113471985, "learning_rate": 0.0005, "loss": 1.2425962686538696, "step": 50574 }, { "epoch": 3.986992510839574, "grad_norm": 0.5395392775535583, "learning_rate": 0.0005, "loss": 1.3224637508392334, "step": 50575 }, { "epoch": 3.987071344107213, "grad_norm": 0.41899019479751587, "learning_rate": 0.0005, "loss": 1.083848476409912, "step": 50576 }, { "epoch": 3.9871501773748523, "grad_norm": 0.48188504576683044, "learning_rate": 0.0005, "loss": 1.070265531539917, "step": 50577 }, { "epoch": 3.987229010642491, "grad_norm": 0.4826946556568146, "learning_rate": 0.0005, "loss": 1.1150810718536377, "step": 50578 }, { "epoch": 3.98730784391013, "grad_norm": 0.5396974086761475, "learning_rate": 0.0005, "loss": 1.381617546081543, "step": 50579 }, { "epoch": 3.987386677177769, "grad_norm": 0.6050934791564941, "learning_rate": 0.0005, "loss": 1.257439136505127, "step": 50580 }, { "epoch": 3.987465510445408, "grad_norm": 0.7073428630828857, "learning_rate": 0.0005, "loss": 1.1373178958892822, "step": 50581 }, { "epoch": 3.987544343713047, "grad_norm": 0.500889241695404, "learning_rate": 0.0005, "loss": 1.0754852294921875, "step": 50582 }, { "epoch": 3.987623176980686, "grad_norm": 0.5295131802558899, "learning_rate": 0.0005, "loss": 1.1204776763916016, "step": 50583 }, { "epoch": 3.987702010248325, "grad_norm": 0.5709922909736633, "learning_rate": 0.0005, "loss": 1.2952978610992432, "step": 50584 }, { "epoch": 3.9877808435159636, "grad_norm": 0.5185350179672241, "learning_rate": 0.0005, "loss": 1.0391579866409302, "step": 50585 }, { "epoch": 3.9878596767836028, "grad_norm": 0.4849326014518738, "learning_rate": 0.0005, "loss": 1.259850263595581, "step": 50586 }, { "epoch": 3.987938510051242, "grad_norm": 0.4435361623764038, "learning_rate": 0.0005, "loss": 1.14247465133667, "step": 50587 }, { "epoch": 3.9880173433188806, "grad_norm": 0.4672374129295349, "learning_rate": 0.0005, "loss": 1.1362031698226929, "step": 50588 }, { "epoch": 3.9880961765865193, "grad_norm": 0.6208578944206238, "learning_rate": 0.0005, "loss": 1.4043103456497192, "step": 50589 }, { "epoch": 3.9881750098541584, "grad_norm": 0.5228379368782043, "learning_rate": 0.0005, "loss": 1.20127272605896, "step": 50590 }, { "epoch": 3.9882538431217975, "grad_norm": 0.5828454494476318, "learning_rate": 0.0005, "loss": 1.361497402191162, "step": 50591 }, { "epoch": 3.9883326763894362, "grad_norm": 0.543778121471405, "learning_rate": 0.0005, "loss": 1.2908244132995605, "step": 50592 }, { "epoch": 3.9884115096570754, "grad_norm": 0.48824459314346313, "learning_rate": 0.0005, "loss": 1.15104341506958, "step": 50593 }, { "epoch": 3.988490342924714, "grad_norm": 0.5969262719154358, "learning_rate": 0.0005, "loss": 1.2369168996810913, "step": 50594 }, { "epoch": 3.988569176192353, "grad_norm": 0.5113339424133301, "learning_rate": 0.0005, "loss": 1.4592931270599365, "step": 50595 }, { "epoch": 3.9886480094599923, "grad_norm": 0.6148682236671448, "learning_rate": 0.0005, "loss": 1.3132853507995605, "step": 50596 }, { "epoch": 3.988726842727631, "grad_norm": 0.4852999150753021, "learning_rate": 0.0005, "loss": 1.194551944732666, "step": 50597 }, { "epoch": 3.98880567599527, "grad_norm": 0.5037170648574829, "learning_rate": 0.0005, "loss": 1.1480525732040405, "step": 50598 }, { "epoch": 3.988884509262909, "grad_norm": 0.615357518196106, "learning_rate": 0.0005, "loss": 1.3174903392791748, "step": 50599 }, { "epoch": 3.988963342530548, "grad_norm": 0.5534144043922424, "learning_rate": 0.0005, "loss": 1.180940866470337, "step": 50600 }, { "epoch": 3.989042175798187, "grad_norm": 0.45134544372558594, "learning_rate": 0.0005, "loss": 1.102956771850586, "step": 50601 }, { "epoch": 3.989121009065826, "grad_norm": 0.5365650057792664, "learning_rate": 0.0005, "loss": 1.2990078926086426, "step": 50602 }, { "epoch": 3.9891998423334645, "grad_norm": 0.5836442112922668, "learning_rate": 0.0005, "loss": 0.9909828305244446, "step": 50603 }, { "epoch": 3.9892786756011036, "grad_norm": 0.46079492568969727, "learning_rate": 0.0005, "loss": 1.097447395324707, "step": 50604 }, { "epoch": 3.9893575088687427, "grad_norm": 0.4403589367866516, "learning_rate": 0.0005, "loss": 1.0779311656951904, "step": 50605 }, { "epoch": 3.9894363421363814, "grad_norm": 0.43300560116767883, "learning_rate": 0.0005, "loss": 1.1423933506011963, "step": 50606 }, { "epoch": 3.9895151754040206, "grad_norm": 0.5267976522445679, "learning_rate": 0.0005, "loss": 1.2230615615844727, "step": 50607 }, { "epoch": 3.9895940086716593, "grad_norm": 0.574910044670105, "learning_rate": 0.0005, "loss": 1.1377822160720825, "step": 50608 }, { "epoch": 3.9896728419392984, "grad_norm": 0.5580029487609863, "learning_rate": 0.0005, "loss": 1.268986701965332, "step": 50609 }, { "epoch": 3.9897516752069375, "grad_norm": 0.442064493894577, "learning_rate": 0.0005, "loss": 1.038741946220398, "step": 50610 }, { "epoch": 3.989830508474576, "grad_norm": 0.5227027535438538, "learning_rate": 0.0005, "loss": 1.2753599882125854, "step": 50611 }, { "epoch": 3.9899093417422153, "grad_norm": 0.4611131548881531, "learning_rate": 0.0005, "loss": 1.274179458618164, "step": 50612 }, { "epoch": 3.989988175009854, "grad_norm": 0.42778247594833374, "learning_rate": 0.0005, "loss": 1.0646905899047852, "step": 50613 }, { "epoch": 3.990067008277493, "grad_norm": 0.4638828933238983, "learning_rate": 0.0005, "loss": 1.3425520658493042, "step": 50614 }, { "epoch": 3.9901458415451323, "grad_norm": 0.5040997862815857, "learning_rate": 0.0005, "loss": 1.0620980262756348, "step": 50615 }, { "epoch": 3.990224674812771, "grad_norm": 0.5169385671615601, "learning_rate": 0.0005, "loss": 1.1721091270446777, "step": 50616 }, { "epoch": 3.9903035080804097, "grad_norm": 0.5038003325462341, "learning_rate": 0.0005, "loss": 1.1576194763183594, "step": 50617 }, { "epoch": 3.990382341348049, "grad_norm": 0.6011791825294495, "learning_rate": 0.0005, "loss": 1.133338451385498, "step": 50618 }, { "epoch": 3.990461174615688, "grad_norm": 0.5266001224517822, "learning_rate": 0.0005, "loss": 1.0686657428741455, "step": 50619 }, { "epoch": 3.9905400078833266, "grad_norm": 0.44366490840911865, "learning_rate": 0.0005, "loss": 1.1170549392700195, "step": 50620 }, { "epoch": 3.9906188411509658, "grad_norm": 0.4609532058238983, "learning_rate": 0.0005, "loss": 1.140931487083435, "step": 50621 }, { "epoch": 3.9906976744186045, "grad_norm": 0.5806451439857483, "learning_rate": 0.0005, "loss": 1.2648859024047852, "step": 50622 }, { "epoch": 3.9907765076862436, "grad_norm": 0.4981992542743683, "learning_rate": 0.0005, "loss": 1.1395254135131836, "step": 50623 }, { "epoch": 3.9908553409538827, "grad_norm": 0.4007399082183838, "learning_rate": 0.0005, "loss": 1.099879264831543, "step": 50624 }, { "epoch": 3.9909341742215214, "grad_norm": 0.425821989774704, "learning_rate": 0.0005, "loss": 1.036026954650879, "step": 50625 }, { "epoch": 3.9910130074891605, "grad_norm": 0.4698353707790375, "learning_rate": 0.0005, "loss": 1.2379984855651855, "step": 50626 }, { "epoch": 3.9910918407567992, "grad_norm": 0.5032891631126404, "learning_rate": 0.0005, "loss": 1.2500040531158447, "step": 50627 }, { "epoch": 3.9911706740244384, "grad_norm": 0.38542357087135315, "learning_rate": 0.0005, "loss": 1.0058848857879639, "step": 50628 }, { "epoch": 3.9912495072920775, "grad_norm": 0.5637354254722595, "learning_rate": 0.0005, "loss": 1.4537171125411987, "step": 50629 }, { "epoch": 3.991328340559716, "grad_norm": 0.5773968696594238, "learning_rate": 0.0005, "loss": 1.1834073066711426, "step": 50630 }, { "epoch": 3.991407173827355, "grad_norm": 0.4894477128982544, "learning_rate": 0.0005, "loss": 1.0100904703140259, "step": 50631 }, { "epoch": 3.991486007094994, "grad_norm": 0.5374226570129395, "learning_rate": 0.0005, "loss": 1.0765752792358398, "step": 50632 }, { "epoch": 3.991564840362633, "grad_norm": 0.5332844257354736, "learning_rate": 0.0005, "loss": 1.1693360805511475, "step": 50633 }, { "epoch": 3.991643673630272, "grad_norm": 0.42513424158096313, "learning_rate": 0.0005, "loss": 0.9334670305252075, "step": 50634 }, { "epoch": 3.991722506897911, "grad_norm": 0.47677624225616455, "learning_rate": 0.0005, "loss": 1.237961769104004, "step": 50635 }, { "epoch": 3.9918013401655497, "grad_norm": 0.5183591246604919, "learning_rate": 0.0005, "loss": 1.1980528831481934, "step": 50636 }, { "epoch": 3.991880173433189, "grad_norm": 0.43061137199401855, "learning_rate": 0.0005, "loss": 1.1675567626953125, "step": 50637 }, { "epoch": 3.991959006700828, "grad_norm": 0.49135714769363403, "learning_rate": 0.0005, "loss": 1.0911301374435425, "step": 50638 }, { "epoch": 3.9920378399684666, "grad_norm": 0.5158641934394836, "learning_rate": 0.0005, "loss": 1.198371410369873, "step": 50639 }, { "epoch": 3.9921166732361058, "grad_norm": 0.46888962388038635, "learning_rate": 0.0005, "loss": 1.158841848373413, "step": 50640 }, { "epoch": 3.9921955065037444, "grad_norm": 0.5271925926208496, "learning_rate": 0.0005, "loss": 1.1849342584609985, "step": 50641 }, { "epoch": 3.9922743397713836, "grad_norm": 0.48060059547424316, "learning_rate": 0.0005, "loss": 1.0926944017410278, "step": 50642 }, { "epoch": 3.9923531730390227, "grad_norm": 0.8251094222068787, "learning_rate": 0.0005, "loss": 1.3586432933807373, "step": 50643 }, { "epoch": 3.9924320063066614, "grad_norm": 0.45856574177742004, "learning_rate": 0.0005, "loss": 1.1561967134475708, "step": 50644 }, { "epoch": 3.9925108395743, "grad_norm": 0.4690731465816498, "learning_rate": 0.0005, "loss": 1.132745385169983, "step": 50645 }, { "epoch": 3.992589672841939, "grad_norm": 0.515347421169281, "learning_rate": 0.0005, "loss": 1.1879265308380127, "step": 50646 }, { "epoch": 3.9926685061095784, "grad_norm": 0.4656452536582947, "learning_rate": 0.0005, "loss": 1.0425372123718262, "step": 50647 }, { "epoch": 3.992747339377217, "grad_norm": 0.4667644500732422, "learning_rate": 0.0005, "loss": 1.0263457298278809, "step": 50648 }, { "epoch": 3.992826172644856, "grad_norm": 0.5202766060829163, "learning_rate": 0.0005, "loss": 1.329916000366211, "step": 50649 }, { "epoch": 3.992905005912495, "grad_norm": 0.5790290832519531, "learning_rate": 0.0005, "loss": 1.179093599319458, "step": 50650 }, { "epoch": 3.992983839180134, "grad_norm": 0.460600882768631, "learning_rate": 0.0005, "loss": 0.9583672285079956, "step": 50651 }, { "epoch": 3.993062672447773, "grad_norm": 0.4991827607154846, "learning_rate": 0.0005, "loss": 1.150116205215454, "step": 50652 }, { "epoch": 3.993141505715412, "grad_norm": 0.4877985417842865, "learning_rate": 0.0005, "loss": 1.1382795572280884, "step": 50653 }, { "epoch": 3.993220338983051, "grad_norm": 0.44232332706451416, "learning_rate": 0.0005, "loss": 1.1839380264282227, "step": 50654 }, { "epoch": 3.9932991722506896, "grad_norm": 0.4557858407497406, "learning_rate": 0.0005, "loss": 1.1551566123962402, "step": 50655 }, { "epoch": 3.9933780055183288, "grad_norm": 0.4281545579433441, "learning_rate": 0.0005, "loss": 1.0767130851745605, "step": 50656 }, { "epoch": 3.993456838785968, "grad_norm": 0.6088842749595642, "learning_rate": 0.0005, "loss": 1.1031855344772339, "step": 50657 }, { "epoch": 3.9935356720536066, "grad_norm": 0.4920385479927063, "learning_rate": 0.0005, "loss": 1.28411865234375, "step": 50658 }, { "epoch": 3.9936145053212453, "grad_norm": 0.596341073513031, "learning_rate": 0.0005, "loss": 1.4865751266479492, "step": 50659 }, { "epoch": 3.9936933385888844, "grad_norm": 0.4578835070133209, "learning_rate": 0.0005, "loss": 1.1935021877288818, "step": 50660 }, { "epoch": 3.9937721718565236, "grad_norm": 0.5688040256500244, "learning_rate": 0.0005, "loss": 1.0573887825012207, "step": 50661 }, { "epoch": 3.9938510051241622, "grad_norm": 0.40783971548080444, "learning_rate": 0.0005, "loss": 1.1710283756256104, "step": 50662 }, { "epoch": 3.9939298383918014, "grad_norm": 0.5167146921157837, "learning_rate": 0.0005, "loss": 1.1848499774932861, "step": 50663 }, { "epoch": 3.99400867165944, "grad_norm": 0.4330528676509857, "learning_rate": 0.0005, "loss": 1.0509577989578247, "step": 50664 }, { "epoch": 3.994087504927079, "grad_norm": 0.48315373063087463, "learning_rate": 0.0005, "loss": 1.2154691219329834, "step": 50665 }, { "epoch": 3.9941663381947183, "grad_norm": 0.4554368555545807, "learning_rate": 0.0005, "loss": 1.0729563236236572, "step": 50666 }, { "epoch": 3.994245171462357, "grad_norm": 0.4314340353012085, "learning_rate": 0.0005, "loss": 1.2278261184692383, "step": 50667 }, { "epoch": 3.994324004729996, "grad_norm": 0.47989967465400696, "learning_rate": 0.0005, "loss": 1.1324387788772583, "step": 50668 }, { "epoch": 3.994402837997635, "grad_norm": 0.4594244360923767, "learning_rate": 0.0005, "loss": 1.2439407110214233, "step": 50669 }, { "epoch": 3.994481671265274, "grad_norm": 0.4713256061077118, "learning_rate": 0.0005, "loss": 1.1959893703460693, "step": 50670 }, { "epoch": 3.994560504532913, "grad_norm": 0.4258359372615814, "learning_rate": 0.0005, "loss": 1.0378068685531616, "step": 50671 }, { "epoch": 3.994639337800552, "grad_norm": 0.5170056223869324, "learning_rate": 0.0005, "loss": 1.0552436113357544, "step": 50672 }, { "epoch": 3.994718171068191, "grad_norm": 0.5023990273475647, "learning_rate": 0.0005, "loss": 1.2099483013153076, "step": 50673 }, { "epoch": 3.9947970043358296, "grad_norm": 0.5425027012825012, "learning_rate": 0.0005, "loss": 1.2505742311477661, "step": 50674 }, { "epoch": 3.9948758376034688, "grad_norm": 0.630924642086029, "learning_rate": 0.0005, "loss": 1.3737266063690186, "step": 50675 }, { "epoch": 3.9949546708711074, "grad_norm": 0.4434175491333008, "learning_rate": 0.0005, "loss": 1.1301321983337402, "step": 50676 }, { "epoch": 3.9950335041387466, "grad_norm": 0.7416625022888184, "learning_rate": 0.0005, "loss": 1.2351188659667969, "step": 50677 }, { "epoch": 3.9951123374063853, "grad_norm": 0.49423733353614807, "learning_rate": 0.0005, "loss": 1.342975378036499, "step": 50678 }, { "epoch": 3.9951911706740244, "grad_norm": 0.48748525977134705, "learning_rate": 0.0005, "loss": 1.1905218362808228, "step": 50679 }, { "epoch": 3.9952700039416635, "grad_norm": 0.45049118995666504, "learning_rate": 0.0005, "loss": 1.1381282806396484, "step": 50680 }, { "epoch": 3.9953488372093022, "grad_norm": 0.49068814516067505, "learning_rate": 0.0005, "loss": 1.1929829120635986, "step": 50681 }, { "epoch": 3.9954276704769414, "grad_norm": 0.4577780067920685, "learning_rate": 0.0005, "loss": 1.0587847232818604, "step": 50682 }, { "epoch": 3.99550650374458, "grad_norm": 0.5907950401306152, "learning_rate": 0.0005, "loss": 1.1649805307388306, "step": 50683 }, { "epoch": 3.995585337012219, "grad_norm": 0.4573965072631836, "learning_rate": 0.0005, "loss": 1.0307601690292358, "step": 50684 }, { "epoch": 3.9956641702798583, "grad_norm": 0.6133755445480347, "learning_rate": 0.0005, "loss": 1.317050576210022, "step": 50685 }, { "epoch": 3.995743003547497, "grad_norm": 0.5656437277793884, "learning_rate": 0.0005, "loss": 1.189211368560791, "step": 50686 }, { "epoch": 3.995821836815136, "grad_norm": 0.5222828388214111, "learning_rate": 0.0005, "loss": 1.2410240173339844, "step": 50687 }, { "epoch": 3.995900670082775, "grad_norm": 0.509426474571228, "learning_rate": 0.0005, "loss": 1.2086718082427979, "step": 50688 }, { "epoch": 3.995979503350414, "grad_norm": 0.4567604959011078, "learning_rate": 0.0005, "loss": 1.278509259223938, "step": 50689 }, { "epoch": 3.9960583366180527, "grad_norm": 0.504445493221283, "learning_rate": 0.0005, "loss": 1.2185646295547485, "step": 50690 }, { "epoch": 3.996137169885692, "grad_norm": 0.5204206705093384, "learning_rate": 0.0005, "loss": 1.1063426733016968, "step": 50691 }, { "epoch": 3.9962160031533305, "grad_norm": 0.50423663854599, "learning_rate": 0.0005, "loss": 1.2901828289031982, "step": 50692 }, { "epoch": 3.9962948364209696, "grad_norm": 0.5048447847366333, "learning_rate": 0.0005, "loss": 1.1594035625457764, "step": 50693 }, { "epoch": 3.9963736696886087, "grad_norm": 0.4305611252784729, "learning_rate": 0.0005, "loss": 1.0189262628555298, "step": 50694 }, { "epoch": 3.9964525029562474, "grad_norm": 0.4524226188659668, "learning_rate": 0.0005, "loss": 1.24806809425354, "step": 50695 }, { "epoch": 3.9965313362238866, "grad_norm": 0.458902508020401, "learning_rate": 0.0005, "loss": 1.1121500730514526, "step": 50696 }, { "epoch": 3.9966101694915253, "grad_norm": 0.4978131949901581, "learning_rate": 0.0005, "loss": 1.240335464477539, "step": 50697 }, { "epoch": 3.9966890027591644, "grad_norm": 0.5386502742767334, "learning_rate": 0.0005, "loss": 0.9558416604995728, "step": 50698 }, { "epoch": 3.9967678360268035, "grad_norm": 0.45536893606185913, "learning_rate": 0.0005, "loss": 1.0557752847671509, "step": 50699 }, { "epoch": 3.996846669294442, "grad_norm": 0.5350596308708191, "learning_rate": 0.0005, "loss": 1.1622660160064697, "step": 50700 }, { "epoch": 3.9969255025620813, "grad_norm": 0.47325730323791504, "learning_rate": 0.0005, "loss": 1.2964410781860352, "step": 50701 }, { "epoch": 3.99700433582972, "grad_norm": 0.4523453414440155, "learning_rate": 0.0005, "loss": 1.36971914768219, "step": 50702 }, { "epoch": 3.997083169097359, "grad_norm": 0.5173957347869873, "learning_rate": 0.0005, "loss": 1.1664278507232666, "step": 50703 }, { "epoch": 3.9971620023649983, "grad_norm": 0.48493027687072754, "learning_rate": 0.0005, "loss": 1.1561967134475708, "step": 50704 }, { "epoch": 3.997240835632637, "grad_norm": 0.36069729924201965, "learning_rate": 0.0005, "loss": 0.968174159526825, "step": 50705 }, { "epoch": 3.9973196689002757, "grad_norm": 0.6627659201622009, "learning_rate": 0.0005, "loss": 1.4371328353881836, "step": 50706 }, { "epoch": 3.997398502167915, "grad_norm": 0.47164711356163025, "learning_rate": 0.0005, "loss": 1.1223704814910889, "step": 50707 }, { "epoch": 3.997477335435554, "grad_norm": 0.5365921258926392, "learning_rate": 0.0005, "loss": 1.2574166059494019, "step": 50708 }, { "epoch": 3.9975561687031926, "grad_norm": 0.4520478844642639, "learning_rate": 0.0005, "loss": 1.11610746383667, "step": 50709 }, { "epoch": 3.9976350019708318, "grad_norm": 0.4840428829193115, "learning_rate": 0.0005, "loss": 1.2842662334442139, "step": 50710 }, { "epoch": 3.9977138352384705, "grad_norm": 0.7531878352165222, "learning_rate": 0.0005, "loss": 1.2580335140228271, "step": 50711 }, { "epoch": 3.9977926685061096, "grad_norm": 0.5317961573600769, "learning_rate": 0.0005, "loss": 1.2534949779510498, "step": 50712 }, { "epoch": 3.9978715017737487, "grad_norm": 0.5396760106086731, "learning_rate": 0.0005, "loss": 1.2613362073898315, "step": 50713 }, { "epoch": 3.9979503350413874, "grad_norm": 0.45702657103538513, "learning_rate": 0.0005, "loss": 1.171004295349121, "step": 50714 }, { "epoch": 3.9980291683090265, "grad_norm": 0.5576256513595581, "learning_rate": 0.0005, "loss": 1.1946744918823242, "step": 50715 }, { "epoch": 3.9981080015766652, "grad_norm": 0.5407398343086243, "learning_rate": 0.0005, "loss": 1.1648204326629639, "step": 50716 }, { "epoch": 3.9981868348443044, "grad_norm": 0.5284155011177063, "learning_rate": 0.0005, "loss": 1.138458490371704, "step": 50717 }, { "epoch": 3.9982656681119435, "grad_norm": 0.4027840495109558, "learning_rate": 0.0005, "loss": 1.0577222108840942, "step": 50718 }, { "epoch": 3.998344501379582, "grad_norm": 0.395550012588501, "learning_rate": 0.0005, "loss": 1.0121302604675293, "step": 50719 }, { "epoch": 3.998423334647221, "grad_norm": 0.4261651039123535, "learning_rate": 0.0005, "loss": 1.2271041870117188, "step": 50720 }, { "epoch": 3.99850216791486, "grad_norm": 0.4878729283809662, "learning_rate": 0.0005, "loss": 1.3123207092285156, "step": 50721 }, { "epoch": 3.998581001182499, "grad_norm": 0.6159989833831787, "learning_rate": 0.0005, "loss": 1.109231948852539, "step": 50722 }, { "epoch": 3.998659834450138, "grad_norm": 0.4185584485530853, "learning_rate": 0.0005, "loss": 0.9680060148239136, "step": 50723 }, { "epoch": 3.998738667717777, "grad_norm": 0.37213319540023804, "learning_rate": 0.0005, "loss": 0.8856374025344849, "step": 50724 }, { "epoch": 3.9988175009854157, "grad_norm": 0.5462074875831604, "learning_rate": 0.0005, "loss": 1.1927690505981445, "step": 50725 }, { "epoch": 3.998896334253055, "grad_norm": 0.609721839427948, "learning_rate": 0.0005, "loss": 1.2960946559906006, "step": 50726 }, { "epoch": 3.998975167520694, "grad_norm": 0.5533513426780701, "learning_rate": 0.0005, "loss": 1.140859842300415, "step": 50727 }, { "epoch": 3.9990540007883326, "grad_norm": 0.5709088444709778, "learning_rate": 0.0005, "loss": 1.148172378540039, "step": 50728 }, { "epoch": 3.9991328340559718, "grad_norm": 0.8454774022102356, "learning_rate": 0.0005, "loss": 1.1998237371444702, "step": 50729 }, { "epoch": 3.9992116673236104, "grad_norm": 0.4775142967700958, "learning_rate": 0.0005, "loss": 1.0072327852249146, "step": 50730 }, { "epoch": 3.9992905005912496, "grad_norm": 0.44874560832977295, "learning_rate": 0.0005, "loss": 1.134577989578247, "step": 50731 }, { "epoch": 3.9993693338588887, "grad_norm": 0.5123984813690186, "learning_rate": 0.0005, "loss": 1.1640260219573975, "step": 50732 }, { "epoch": 3.9994481671265274, "grad_norm": 0.5367391109466553, "learning_rate": 0.0005, "loss": 1.1569068431854248, "step": 50733 }, { "epoch": 3.999527000394166, "grad_norm": 0.5606088042259216, "learning_rate": 0.0005, "loss": 1.1891562938690186, "step": 50734 }, { "epoch": 3.999605833661805, "grad_norm": 0.428684800863266, "learning_rate": 0.0005, "loss": 1.0039479732513428, "step": 50735 }, { "epoch": 3.9996846669294444, "grad_norm": 0.48548755049705505, "learning_rate": 0.0005, "loss": 1.2882518768310547, "step": 50736 } ], "logging_steps": 1, "max_steps": 63425, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 6342, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 8.335491449759151e+19, "train_batch_size": 8, "trial_name": null, "trial_params": null }